CONTACT LANGUAGES
CREOLE LANGUAGE LIBRARY (CLL) A companion series to the "JOURNAL OF PIDGIN & CREOLE LANGUAGES"
Editors Pieter Muysken (Amsterdam) John Victor Singler (New York)
Editorial Advisory Board Mervyn Alleyne (Kingston, Jamaica) Norbert Boretzky (Bochum) Lawrence Carrington (Trinidad) Chris Corne (Auckland) Glenn Gilbert (Carbondale, Illinois) John Holm (New York)
George Huttar (Dallas) Salikoko Mufwene (Chicago) Peter Mühlhäusler (Adelaide) Pieter Seuren (Nijmegen) Norval Smith (Amsterdam)
Volume 17
Sarah G. Thomason (ed.) Contact Languages: A wider perspective
CONTACT LANGUAGES A WIDER PERSPECTIVE
Edited by
SARAH G. THOMASON
JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984.
Library of Congress Cataloging-in-Publication Data Contact languages : a wider perspective / edited by Sarah G. Thomason. p. cm. -- (Creole language library, ISSN 0920-9026 ; v. 17) Includes bibliographical references and indexes. 1. Languages, Mixed. I. Thomason, Sarah Grey. II. Series. PM7802.C66 1996 417'.22-dc20 ISBN 90 272 5239 4 (Eur.) / 1-55619-172-3 (US) (alk. paper)
96-43552 CIP
© Copyright 1997 - John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA
Table of Contents Acknowledgments Contributors List of maps Introduction Hiri Motu Tom Dutton Pidgin Delaware Ives Goddard Ndyuka-Trio Pidgin George L. Huttar & Frank J. Velantie Arabic-based Pidgins and Creoles Jonathan Owens Kitúba Salikoko S. Mußvene Sango Helma Pasch Prior Pidginization and Creolization in Swahili? Derek Nurse Michif: A Mixed Language Based on Cree and French Peter Bakker & Robert A. Papen Media Lengua Pieter Muysken Callahuaya Pieter Muysken Mednyj Aleut Sarah G. Thomason Ma'a (Mbugu) Sarah G. Thomason Language Index Names Index Subject Index
vii ix xi 1 9 43 99 125 173 209 271 295 365 427 449 469 489 495 503
Acknowledgments The contributors to this volume have been so patient, during years of delay in the editing process, that I owe them an unusually emphatic expres sion of gratitude — not only for their lack of overt impatience, but for their amiable reaction to repeated editorial queries and requests for revision. The series editors, Pieter Muysken and John Singler, have provided valuable advice and helpful nudges throughout this period. My most enthusiastic thanks got to Anna Fenyvesi, whose extreme diligence and word-processing skill produced the final manuscript and whose moral support heartened me throughout the last stages of the process.
Contributors Peter Bakker Institute for General Linguistics University of Amsterdam Spuistraat 210 1012 VT Amsterdam The Netherlands Tom Dutton Department of Linguistics Research School of Pacific Studies The Australian National University P.O.Box 4 Canberra Australia Ives Goddard NHB Rm. 85, MRC 100 Smithsonian Institution Washington, D. C. 20560 USA George L. Huttar International Linguistics Center 7500 W. Camp Wisdom Rd. Dallas, TX 75236 USA Salikoko Mufwene Department of Linguistics University of Chicago 1010 East 59th Street Chicago, IL 60637
x
Pieter Muysken Instituut voor Algemene Taalwetenschap Universiteit can Amsterdam Spuistraat 210 1012 VT Amsterdam The Netherlands Derek Nurse Department of Linguistics Memorial University of Newfoundland St. John's Newfoundland Canada A1C 3X9 Jonathan Owens Universität Bayreuth Sprach- und Literaturwissenschaftliche Fakultät Arabistik Postfach 101251 D-8580 Bayreuth Germany Robert A. Papen Département de Linguistique Université du Québec à Montréal . .8888 Succursale A Montréal, Québec H3C 3P8 Canada Helma Pasch Institut für Afrikanistik der Universität zu Köln Meister-Ekkehart-Strasse 7 5000-Köln 14 Germany
Contributors
Contributors Sarah G. Thomason Department of Linguistics University of Pittsburgh Pittsburgh, PA 15260 USA Frank J. Velantie Afd. Alfabetisering (Volwassenen Edukatie) MINOV Commewijnestraat 31 Paramaribo Suriname
XI
List of Maps Map of Hiri Motu area Approximate current residential areas of Trios, Wayanas, and Bushnegro groups of Suriname Map of the Sudan Languages of the Southern Sudan and adjoining regions Position of the Kongo Kingdom in the 17th-18th centuries Map of Sango area Map of Media Lengua area
10 100 128 137 174 210 370
Introduction Sarah G. Thomason University of Pittsburgh
The best-known contact languages are pidgins and creoles with European lexicons, scattered along the major routes that were followed by European powers engaged in trade and colonization, starting in the Age of Exploration. These languages arose as a direct result of contact with Europeans, as re flected in the fact that their vocabularies are drawn primarily from the European languages of visiting traders, colonizers, or resident slavemasters. But it is becoming increasingly obvious that languages such as the Caribbean and Indian Ocean creoles, the various varieties of Pidgin English, and the several Portuguese-based creoles in South Asia and elsewhere owe their fame (and their prominence in the linguistic literature) to their links with European history rather than to any special linguistic status among contact languages. Pidgins and creoles with non-European lexicons are now being studied in many places around the world, and a third type of contact language — bilingual mixtures that (unlike pidgins and creoles) must have been created by bilinguals — has been reported reliably from widely separated locations. The purpose of this book is to present linguistic and historical sketches of some lesser-known contact languages, in an effort to provide some balance in the worldwide picture of these most dramatic results of language contact. An added goal is to facilitate the investigation of linguistic and historical similari ties and differences among contact languages of all three major types. The twelve case studies collected here range over all three types of contact languages, and several authors also deal explicitly with the problem of finding the boundary between contact language and dialect of the lexifier language. Three chapters focus on particular pidgins (Hiri Motu, Pidgin Delaware, and Ndyuka-Trio Pidgin), two on creoles (Kitúba, Sango), one on a
2
Sarah G. Thomason
specific set of pidgins and creoles (Arabic-based languages), one on the question of early pidginization and/or creolization in Swahili, and five on bilingual mixtures (Michif, Media Lengua, Callahuaya, Mednyj Aleut, and Ma'a). The chapters are arranged according to the type of language: pidgins first, then the chapter covering both pidgins and creoles, then creoles, then the question-mark chapter on Swahili, and finally the two-language mixtures. Geographically, the Americas and Africa are best represented (five chapters each); there is one chapter each from the Pacific and northern Eurasia. This geographical skewing was not a deliberate choice. To some extent it resulted from a historical accident — several scholars who were asked to contribute were unable to do so in the end — but the main reason for the imbalance is that more contact languages that meet the basic criterion for inclusion are known in Africa and the Americas than in other parts of the world. This picture may change dramatically soon, as more information becomes available about contact languages elsewhere, notably among nonAustronesian languages of New Guinea (Foley 1986:30-31, Williams 1993). All the authors were asked to address a series of questions designed to make the case studies easy to compare with each other and with descriptions of other contact languages. The first set of questions concerned the circum stances of the language's use: location, number of speakers, domains of usage, other languages spoken by the users. The next set focused on demon strating the "languageness" of the language under consideration: is it crystal lized, so that it must be learned as a language rather than simply being produced by ad-hoc simplification (or other distortion) from any given speaker's native language? And is it mutually intelligible with any other language, specifically the main lexical-source language? The third set of questions was historical. In discussing the historical circumstances under which the contact language arose, authors were asked to distinguish carefully between documented social conditions and inferences drawn from inevitably incomplete documentation, and to lay out the bases for inferences drawn from linguistic facts. (For instance, if the language is a creole, is there direct evidence of a fully crystallized pidgin stage, or could creolization have been abrupt in a new contact situation?) Finally, authors were asked to describe the language briefly and to consider which of the three obvious candidates for sources of contact-language lexicon and grammar — the lexifier language(s), other language(s) spoken by the creators of the contact language, and univer sal structural tendencies — might be responsible, alone or in combination, for particular features found in the language.
Introduction
3
These questions could not be answered fully for all the languages, because of limitations in the available information or, as in the Swahili chapter, a scope that included non-contact varieties as well as contact lan guages. The chapters also differ in depth of coverage. In some cases the contact languages, and often their lexical and grammatical source languages as well, are still too little known to permit a detailed analysis; in other cases, such as Hiri Motu, more detailed discussions of certain topics are available elsewhere, so that readers are referred to other sources for details. The authors approach their topics in different ways as a result of the kinds of information that are available. So, for instance, Goddard's discussion of Pidgin Delaware is based on philological analysis of the documentation on this long-dead language, while Dutton and Muysken, to name just two of several authors, draw on their own fieldwork data in describing Hiri Motu and Media Lengua. An introduction to a book about contact languages must define the object of study: what is a contact language? The book's title presupposes that contact languages are a well-defined linguistic phenomenon, but — as with so much else in language-contact studies — the point is controversial. Probably no answer will satisfy everyone, but the following definition is at least consistent with the usage of all the contributors to this volume: a contact language is a language that arises as a direct result of language contact and that comprises linguistic material which cannot be traced back primarily to a single source language.1 Because the historical linguist's technical concept of genetic rela tionship requires that members of a language family descend primarily, as whole systems, from a single parent language, contact languages do not belong to any language family: by definition, their genesis was not a matter of descent with modification from a single parent. This definition is thus funda mentally historical; it is based on diversity in the sources of the linguistic structures rather than on (say) typological characteristics of the language. The reason for insisting on a historical definition is that synchronic definitions don't work; there is, for instance, no such thing as a master list of linguistic features that are universally shared by and exclusive to contact languages, or even pidgins and creoles as a set (leaving bilingual mixtures aside). There are, in my view, just three types of contact languages — namely, the types represented in this book. There are prototypical contact languages in all three categories, and there are also various kinds and degrees of deviations from the prototype. All these deviations (e.g. "semi-creoles" and "koinés") are best analyzed in relation to the prototype, as on a continuum, and not as
4
Sarah G. Thomason
separate types of contact languages. Treating pidgins, creoles, and bilingual mixtures as discrete language types of course raises problems of identifica tion. There are many borderline cases: languages that have some pidgin characteristics but are analyzable as simplified versions of their lexifiers; speech forms that have some systematicity but are not clearly stable enough to be classed as languages; and so forth. But there are also many clear cases of contact languages. And since wrestling with fuzzy boundaries is a standard part of the historical linguist's job — consider, for instance, the impossibility of drawing a sharp dividing line between dialects of the same language and separate languages, during a process of language split — the discovery that fuzzy boundaries also exist in the study of contact languages is no surprise. Some languages are clearly contact languages; for other languages there is conflicting evidence, so that no firm identification is possible; and still other languages clearly are not contact languages. Of the languages covered in this book, most are definite contact lan guages by my definition: Hiri Motu, Pidgin Delaware, Ndyuka-Trio Pidgin, various Arabic-based pidgins and creoles, Kitúba, Sango, Michif, Media Lengua, Callahuaya, and Mednyj Aleut. Nurse argues that early Swahili was not a contact language, however, and Owens discusses (critically) the view that Arabic went through a pidgin stage. Ma'a is a contact language if it is a separate language, but its status as an independent language is debatable. Contact languages are not necessarily lingua francas,2 though many of them are. Of the ones discussed here, Hiri Motu, Pidgin Delaware, NdyukaTrio Pidgin, the Arabic-based pidgins and creoles, Kitúba, and Sango are (or were) also lingua francas; Swahili too has long served as a lingua franca, but it did not arise through pidginization or creolization. In sharp contrast, none of the bilingual mixtures fulfills (or fulfilled) such a function. This split is typical: bilingual mixtures usually or always serve as salient markers of ethnic-group identity (in fact, that's why they come into being), while pidgins and creoles arise as lingua francas. As noted above, a major purpose of this book is to make information about lesser-known contact languages more readily available. This in turn should enable specialists who are interested in generalizing over the whole range of contact languages to avoid the European-lexifier bias (especially for pidgins and creoles) of the majority of theoretical proposals in this area. The chapters provide two particularly striking illustrations of the usefulness of a broader data-base.
Introduction
5
The first concerns the issue of how many languages must be present in a new contact situation in order for a stable pidgin to develop. Virtually all generalizations about pidgin and creole languages cite the presence of more than two languages as a typical, or in some cases an absolute, condition for the emergence of a pidgin or creole. Probably the most widely cited formulation of this view is Whinnom's argument that "no pidgin has ever consolidated itself in other than a multilingual situation" (1971:104); numerous other authors (e.g. Mühlhäusler 1986:147) have accepted Whinnom's view. But two authors in this volume describe pidgins that apparently arose in twolanguage contact situations. Goddard suggests that Pidgin Delaware origi nated in communication between Delawares and Dutchmen and then spread to other European groups; Huttar & Velantie say that Ndyuka-Trio pidgin developed for communication between Ndyuka and Trio speakers only. To gether with other cases that have been described recently — most notably Kouwenberg's analysis of Berbice Dutch Creole as a two-language creole that arose out of contact between Dutch and the West African language Eastern Ịjọ (1994) — these examples show that the standard view about numbers of languages in pidgin genesis is simply mistaken. The second striking example is, in effect, the explanation for the first. Whinnom's common-sense reasoning behind his claim that more than two languages are needed for pidgin genesis is that, in a two-language situation, "there cannot be any rèally effective withdrawal of the target language" (1971:104). But here common sense is misleading. All three of the pidgins described in this book — Hiri Motu, Pidgin Delaware, and Ndyuka-Trio Pidgin — emerged in the first instance because the lexifier-language speakers deliberately withheld their full language from outsiders, instead using a simplified foreigner-talk variety that formed the basis for the eventual pidgin. Comparable cases are reported from other parts of the world, e.g. Halb deutsch (Lehiste 1965), but few of these turned into stable pidgins. The three examples side by side in this book provide solid evidence for the crucial importance of speakers' attitudes in determining the linguistic results of language contact. Another important feature of these papers is the thoroughness of the authors' descriptions of the historical circumstances in which the contact languages arose. These descriptions reveal a number of shared historical features, but there are also significant differences. Among the pidgins, Pidgin Delaware, Ndyuka-Trio Pidgin, and some Arabic-based pidgins arose as
6
Sarah G. Thomason
trade pidgins, as did many other pidgins around the world. The setting for the development of Hiri Motu was more complex: in Dutton's account, the Motu used their foreigner-talk variety with all foreigners, but a stable pidgin crys tallized only when many foreigners arrived — including the first police force, whose members came from other islands and were not Motu speakers — to provide a multilingual setting in which a single lingua franca was needed. The three creoles described in the book followed three different routes of development. East African Nubi, though connected with Arabic-based trade pidgins, was creolized when ethnically mixed soldiers established their own permanent settlements. Kitúba arose out of contact between West Africans and Bantu speakers, on the one hand, and among the Bantu speakers them selves in new multilingual settings, on the other; and Sango emerged as a work-group language. The five bilingual mixed languages fall into three general categories, historically speaking. Michif, Media Lengua, Mednyj Aleut, and Callahuaya all arose abruptly, by sudden creation, rather than gradually; but there were apparently two quite different motives for their development. The first three serve as the special languages of new ethnic groups or subgroups — the mixed-blood Métis (French, Cree, Ojibwe), acculturated Quechuas who belong neither to the Quechua world nor to the Spanish world, and mixed-blood residents of Mednyj Island (Russian, Aleut), respectively. The fourth, Callahuaya, functions as a secret language, used by male healers during rituals. The fifth bilingual mixture, Ma'a (Mbugu), is not the result of sudden creation but rather of long-term linguistic persistence in the face of intense cultural pressure from Bantu; far from being a new ethnic group, Ma'a speakers are an old community that has stubbornly resisted total cultural assimilation. At first glance, then, these five languages seem quite different in nature. But they share a characteristic social feature: all of them are in-group languages, used within the community as a sign of community solidarity and not understood by outsiders. They also share a salient linguistic feature: they are all comprised of two components, one from each of two source languages, and the linguistic material from each source language is adopted wholesale, without the kind of distortion that would occur in the absence of bilingualism. Finally, the structural descriptions in the first six case studies provide a strong antidote to the still common view that all pidgins and creoles have similar and simple structures. Features like the systematic OSV and SOV word order patterns of Hiri Motu, the noun class system of Kitúba, and the
Introduction
1
/kp/ and /gb/ phonemes of Sango will surely help to eradicate the idea that pidgins and creoles have maximally simple and more or less identical gram matical structures. The set of case studies in this book is a very small sample of the rich variety of contact languages around the world. We hope, in presenting these studies, to encourage other work along the same lines. Only with the accumulation of many solid case studies can our data-base become substantial enough to support robust generalizations about the nature and development of all types of contact language.
Notes 1.
At least one hedge must be added immediately: non-contact languages may, and often do, contain a great many loanwords, but few loanwords appear in the BASIC vocabulary. English is the most famous example of a loanword-heavy language, with up to 75% of its lexicon borrowed from French and/or Latin. But it is nevertheless easy to prove that English is a Germanic language, because the vast majority of the items in its basic vocabulary, together with most of its grammar, are of Germanic origin. An added complication is that "basic vocabulary" is not a precise concept. The idea is that basic vocabulary items are those that occur in every language, and are therefore (among other things) less likely to be borrowed than culture-specific words. In spite of its vagueness, this notion is of considerable methodological importance in historical linguistics.
2.
And of course not all lingua francas are contact languages. English, for instance, is the world's most-used lingua franca, and English itself is not a contact language.
References Foley, William A. 1986. The Papuan languages of New Guinea. Cambridge: Cambridge University Press. Kouwenberg, Silvia. 1994. A grammar of Berbice Dutch Creole. Berlin & New York: Mouton de Gruyter. Lehiste, Ilse. 1965. "A poem in Halbdeutsch and some questions concerning substratum". Word 21.55-69. Mühlhäusler, Peter. 1986. Pidgin & creole linguistics. Oxford: Basil Blackwell. Whinnom, Keith. 1971. "Linguistic hybridization and the 'special case' of pidgins and creoles". Pidginization and creolization of languages, ed. by Dell Hymes, 91-115. Cambridge: Cambridge University Press. Williams, Jeffrey P. 1993. "Documenting the Papuan-based pidgins of insular New Guinea". Atlantic meets Pacific: A global view of pidginization and creolization, ed. by Francis Byrne & John Holm, 355-367. Amsterdam: John Benjamins.
Hiri Motu Tom Dutton The Australian National
1.
University
Introduction1
Hiri Motu is the name now used officially to refer to the language that used to be known as Police Motu. This is a pidginized form of Motu, the native language spoken around Port Moresby, capital of Papua New Guinea, and for some fifty kilometers or so east and west of it along the coast (Dutton 1969) — see map.2 Until independence in 1975 Hiri Motu (as Police Motu) was the principal lingua franca between peoples speaking mutually unintelligible languages (including Europeans) and the unofficial language of administra tion of much of the southern half of the country that was then known as the Territory of Papua. In the years immediately preceding and following inde pendence Hiri Motu assumed an important role in national politics. Today not so much is heard of it in this role although it still has the potential to become so should social conditions change to favor it. It is still, however, an important language in Papua New Guinea and one of two recognized unofficial national languages of the country, even though it is not being learned and used by the younger generation to the same extent as previously, as Tok Pisin and/or English become more widely known.3 Historically the origin of this language is uncertain as there are no relevant written records. Until recently the popular view was that (and this is the view that underlies the name change from Police Motu to Hiri Motu) it is a continuation of a trade language used by the Motu on annual trading voyages, or hiri, to the Gulf of Papua where the Motu traded with linguistically unrelated peoples.4 However, recent research shows that this cannot be the case. Instead the language most probably developed out of a special variety or register of Motu used by the Motu in talking to those who came to visit or trade with them
Tom Dutton
10
Hiri Motu
11
in their own area, whether from linguistically unrelated areas or not. This variety was a simplified form of Motu, a kind of Motu Foreigner Talk, that was, it will be claimed, taken up, used, and spread in a revised form by members of the first police force and others in British New Guinea, as Papua was called then, from the late 1880s onwards. As a result it became associated with the police force and soon became known as Police Motu.
2.
Distribution and varieties
A survey of the distribution of Police Motu in 1961 (Brett et al. 1962) showed that it was at that time spoken throughout most of Papua except for those areas which had had little contact with the Administration (as, for example, in the Southern Highlands District and in distant parts of the Western, Gulf, and Milne Bay Districts) or where there were competing church languages (as, for example, in the Milne Bay and Western Districts). The Central District contained the highest number of speakers and the number of speakers dimin ished roughly in relation to the distance traveled away from Port Moresby in any direction. The only exception to this was in the coastal area of the Purari River delta where the language was adopted as the language of the Tommy Kabu Movement5 in the 1950s and was spoken "as a conscious preference" to the mother tongue of the members (Hitchcock & Oram 1967:11). In 1962 the number of speakers was estimated to be approximately 65,000, although this did not include 12,000 Motu and Koita villagers around Port Moresby who spoke Motu as their first or second language and who generally also know some kind of Hiri Motu. In the 1966 census, however, something like 110,000 persons over the age of ten years claimed to be able to speak "simple Police Motu", or at least could answer census questions in it, and in the 1971 census upwards of 150,000 persons living in Papua New Guinea are said to have spoken it. Not all of these were Papua New Guineans but the majority of those that were came from the six administrative districts that used to make up the old political unit of Papua. The present situation is unknown as there are no more recent census statistics available. There are some indications also, as noted above, that younger speakers are not learning the language but are learning and using Tok Pisin and/or English instead. Because of its wide distribution it comes as no surprise to find that there is no such thing as standard Hiri Motu. Instead it exists as a series of varieties
12
Tom Dutton
(distinguishable chiefly by their sound systems and vocabulary) representing varying degrees of difference within two dialects — a Central one and a nonCentral one. Thus, for example, whereas speakers of the Central dialect generally maintain phonological distinctions made by the Motu (because their languages are closely related to Motu), speakers of the non-Central dialect do not. Consequently speakers of the non-Central dialect will be heard to say lau for both "I" and "go", whereas Central dialect speakers will say lau for "I" and lao for "go". Similarly, speakers of the Central dialect use more Motu words than speakers of the non-Central dialect do, the latter using words taken from different varieties of English or other languages of Papua. Thus, for example, speakers of the Central dialect will be heard to say ginigunana for "first" while those of the non-Central dialect will be heard to say namba wan, a form borrowed from the variety of pidgin English formerly spoken in Papua and the adjacent Torres Straits.6 Or again, while Central dialect speak ers will use rata for both "breast" and "milk", non-Central dialect speakers will use rata for "breast" and susu for "milk"; this latter word comes from some form of pidgin English or from one or more of the languages of Milne Bay, most probably Suau. Finally, whereas Central dialect speakers will say tamagu for "my father" and tamamu for "your father", non-Central dialect speakers will say lauegu tamana and oiemu tamana respectively. The Central dialect of Hiri Motu is that used mostly by the Motu and speakers of other closely related languages in the neighboring area who generally know some Motu. These languages are those generally referred to in the literature as Austronesian, as distinct from most others in Papua New Guinea, which are non-Austronesian or Papuan — see map. The other dialect is that used by speakers of all other languages, which are mostly not related to Motu, except in the Milne Bay Province and adjacent areas. These are all Papuan languages. This latter dialect is much more widespread than the former and several commentators have advocated that it, or selected parts of it, should be taken as the standard dialect for mass communication purposes. In fact the Central dialect is the one most used for these purposes.7
3.
The pidgin nature of Hiri Motu
As already indicated, Hiri Motu is a pidgin language. That is, it shares a number of features with pidgins observed elsewhere. Thus, for example, it is
Hiri Motu
13
reduced in structure compared with its major lexical-source language, Motu. It lacks, for example, most of the inflectional and derivational morphology of Motu, has no irregular verbs, makes no distinction between alienably and inalienably possessed nouns, has restricted adjective agreement, and has generalized dual forms and a universal postposition dekena or dekenai which is very much like long in English-based pidgins of the neighboring region. Its vocabulary is, moreover, restricted and not based solely on Motu. Hiri Motu can also be said to be like other pidgins in being nonnative to most speakers — there are reputed to be a number of speakers who speak it as their first language, but these have never been surveyed or studied — and in being used for communication between speakers of mutually unintelligible languages across Papua. Thus, even though, as will be argued here, Hiri Motu is most probably a continuation of a simplified form of Motu that was used for communicating with visitors to the Motu area, it has developed its own structure which has to be learned just as does that of other well-recognized pidgins. Thus Motu speakers cannot produce Hiri Motu by ad-hoc simplifica tion, just as English speakers cannot produce Tok Pisin, for example, by adhoc simplification of English (however much most nonspeakers think they can).8 It is, moreover, not immediately intelligible to Motu speakers who have not been previously exposed to it. In other words, Hiri Motu is a language in its own right and is not some debased or broken form of Motu.
4.
The origin and development of Hiri Motu
At the time of first European settlement in 1874 the coastal area around Port Moresby was occupied by two completely different and linguistically unre lated groups of people, the Motu and the Koita (or Koitabu as the Motu call them). The Motu lived in maritime villages between Kapakapa in the east and Galley Reach in the west and were divided into two groups or tribes, the Eastern Motu and the Western Motu, who lived east and west of Bootless Bay respectively. Of these two groups the Western Motu played the most impor tant role in the origin and development of Hiri Motu (as Police Motu), for it was they who were drawn into closest and increasing contact with foreigners after the arrival of the first Europeans. They were, however, distinguished culturally from their eastern counterparts by the fact that at the time of European contact (and for an unknown number of years before) they were
14
Tom Dutton
involved in a complex network of trading relationships with linguistically related and unrelated groups east, west, and inland of their present position. The most spectacular and important part of this trade was the hiri, or annual trading voyage to the Gulf of Papua several hundred kilometers away to the west. During these voyages the Motu visited such groups as the Elema (the general name given to a group of coastal peoples speaking eight closely related languages of the Eleman Family between Cape Possession in the east and the Purari River delta in the west) and their neighbors immediately westwards, the Kikori, who inhabit the delta of the Purari River. These groups speak languages unrelated to Motu and only very distantly related to each other (if they are related at all). On these visits and on return visits made by some of these groups, the Motu and their trading partners communicated with each other in one or more of at least two different trade languages which were pidginized forms of the languages spoken by the Gulf traders. One of these languages, the Hiri Trading Language, Eleman variety, hereafter HTL(E), was based largely on the component languages of the Eleman Family, and the other, the Hiri Trading Language, Koriki variety, or HTL(K), was based on the single language, Koriki. Both of these languages had a number of features in common, although they were not mutually intelligible. The examples in Table 1 show what these languages were like, and most importantly, that they were quite different from, and should not be confused with, Hiri Motu, as has been the case in the past, although they were obviously similar in structure to it in many respects.
Table 1. The Hiri trading languages and Hiri Motu compared. HTL(K)
HTL(E)
Hiri Motu
English
Koa(nu) vapeo? who(his) canoe
A, neia enane eh, who his vevara? canoe Ara enane vevara. I my canoe
(Inai be ) daika (this focus) who ena lakatoi? his canoe (Inai be ) lau(this focus ) I egu lakatoi. my canoe
Whose canoe is this?
Na vapeo. my canoe
(It's) mine.
Hiri Motu
15 Eme enane rare you your name neia? who Ara enane rare I my name maro Moi. focus Moi
Oi- emu ladana you-your name (be ) daika? (focus) who Lau-egu ladana I- my name be Moi. focus Moi
E, uapekai. yes, stay Moi, anene pei Moi, come food navai. eat Na okuai! me give
Ene enane pamora you your friend rare maro neia? name focus who (Ara enane pamora (I my friend rare maro) Elamo. name focus) Elamo Elamo abuviti? Elamo stay E, abuv iti. yes, stay Moi, abusi ma Moi, come water siahu abulari. hot eat Ara porohalaia! me give
Enane pu miai go sago get anea! come
Abuari pai go sago avaia abusi! get come
Pu peo. sago not
Pai penepene sago some lasi. not
Oi- emu pamora you- your friend (be ) daika? (focus) who (Lau-egu pamora (I- my friend be ) Elamo. focus) Elamo Elamo ia noho? Elamo he stay Io, ia noho. yes, he stay Moi, oi mai Moi, you come aniani oi ania. food you eat Lau oi henia/ oi me you give/ you henia lau dekena! give me to Oi lao rabia you go sago oi mailaia! you bring Rabia (ia noho) sago (it stay) lasi. not
Ni noe koana? you name who
Na noe Moi. my name Moi
Ni vake(nu) your friend noe koana? name who (Na vake(nu) (my friend(his) noe) Elamo. name) Elamo Elamo uapekai? Elamo stay
What's your name?
My name is Moi.
What's your friend (or trade partner)'s name? (My friend (or trade partner)'s name is) Elamo. Is Elamo here? Yes, he's here. Moi, come and eat (some food).
Give it to me!
Go and bring some sago!
There's no sago.
16
Tom Dutton
But the hiri was only part of the complex network of trade and exchange that the (Western) Motu were engaged in at the time of European contact, albeit the most spectacular part. Other parts of this network involved contact with speakers of other languages that lived in their immediate neighborhood and a little farther afield. Some of these trading groups spoke (and still speak) languages that are closely related to Motu, while the remainder spoke (and most still speak) languages which, as already indicated, are not related to Motu and are only distantly related to each other in groups or families, if they are related at all. The closest to the (Western) Motu were the Koita, who, at the time of first European contact, lived amongst the Motu as minority sections in Motu villages or in separate villages close by. Inland of the Koita and closely related to them linguistically lived the Koiari and the Mountain Koiari (also spelt Koiali). Because of their position, however, these two groups had little direct contact with the Western Motu, although there was some trade between them. They apparently knew very little if any Motu. Beyond the Koita the Motu also traded directly with linguistically related groups on either side. In the west they visited the Doura and Gabadi of the Galley Reach area and to the east they had developed rather special relation ships with the Vulaa, a tribe inhabiting Hula and associated villages around Hood Point. Motu villagers who did not go on the hiri depended on these people for fish while the hiri traders were away. In return the Vulaa received sago and other gifts from the returning hiri canoes. It is not known for certain what language was used by the Motu and their trade partners in these more local contacts, but it can be assumed on the basis of the available evidence that this was a simplified form of Motu, hereafter referred to as Simplified Motu. Thus, for example, when the Rev. W. G. Lawes of the London Missionary Society settled in Port Moresby in 1874 as the first European missionary amongst the Motu, he attempted to learn Motu. However, it was not until some time later that his son, Frank, who played with the boys in the village and learned Motu from them, drew his father's attention to the fact that he did not speak "true" Motu but only a simplified version of it,9 which he also used in making his first translations into Motu (Taylor 1978). Later Lawes noted that this simplified version of the language that he had been taught was characterized by "a good many colloquialisms" or "instances of pidgin Motuan" in use amongst the Motu which were "not correct grammatically" but were "sanctioned by usage" (Lawes 1896:30).
Hiri Motu
17
This Simplified Motu was, moreover, used by the Motu "in speaking to foreigners", although they themselves would "never do so amongst them selves". Although Lawes did not ever describe this foreigner talk in more detail we have been able to get some further insight into its nature and use in various ways. As a result it is possible to say that this Simplified Motu was not a separate language from Motu (in the sense that it was unintelligible to native Motu speakers and had to be learned by them as a second language), but was merely a special variety or register of Motu used for communicating with a particular class of people, notably those seeking entry to the Motu world. This variety was furthermore most probably not a stable one but probably varied from speaker to speaker over time, and depending on whether the person being spoken to was a complete stranger to the Motu or not. As a variety it was distinguishable from "true" Motu by a number of grammatical features of the sort already noted and to be discussed in more detail below, some of which are not found in Motu. Thereafter nothing is heard of this Simplified Motu until more than twenty years later when a distinct language with many of the same features appears that was to become known as Police Motu and later Hiri Motu. Although there is again no linguistic evidence on the point it is presumed, on the basis of other evidence to be presented, that the two are connected and that Police Motu is a continuation of this Motu Foreigner Talk or Simplified Motu (in a modified form) and not a separate development. There are two reasons for this. The first has to do with the fact that the received oral tradition claims that Police Motu is a continuation of a hiri trade language (as already indicated) and the second with social developments in the Port Moresby area following the arrival of London Missionary Society missionary Lawes in 1874. With respect to the received oral tradition it is to be noted that the principal reason for suggesting the name change Police Motu to Hiri Motu was that this language was supposed to be a "lineal descendant of the language of the Hiri" (Chatterton 1971:2). But it has already been shown that this could not have been the case, because not only did the Motu use at least two such languages (the HTLs) in hiri trading, but also these were markedly different from Police Motu, especially in vocabulary. So either the tradition has to be rejected as nonsense, or it has to be seen as having been misunder stood or misinterpreted at some point. Alternatively Police Motu has to be seen as a "lineal descendant" in a sense different from that normally accepted in linguistics. Given, however, that the Motu had another contact "language",
18
Tom Dutton
Simplified Motu, which shares most grammatical features and vocabulary with Police Motu, a point which is demonstrated and discussed further below, the simplest explanation would appear to be that the tradition has indeed been misunderstood or misinterpreted at some time. This position is attractive, moreover, because it saves both the tradition, in a modified form, and the linguistic definition of continuation. Thus we may suggest that what the tradition really said was simply that Police Motu was the continuation of a trade language, one of several used, and not THE hiri trade language. Thus, both the received oral tradition, once reconstructed, and the linguistic evi dence are consistent with an hypothesis that Police Motu is a lineal descen dant of the contact "language" Simplified Motu. However, for this to be true particular social forces must have been active in and around Port Moresby to keep this simplified form of Motu alive and to extend its use into a general lingua franca before 1890, the year that the first police force was established in Port Moresby. One of these forces, notably the need to trade or converse with non-Motu coming into the Motu area, has already been referred to. A second force, probably not unrelated to the first, was that the Motu were apparently not keen for non-Motu to learn Motu; other forces have to do with social developments in the Port Moresby area following the arrival of mis sionary Lawes in 1874. Thus when news of the discovery of Port Moresby began to filter down through the Australian colonies and overseas, people of all sorts began making their way towards this new terra incognita as if drawn to it by some strange magnet. The first to do so were missionaries, who were soon followed by "scientists", entrepreneurs, explorers, adventurers, traders, and others, who in turn were followed by Government officials. Each of these groups had their effect on the use and transmission of different languages in the area, but it would appear to be the unofficial "visitors" who were the most important as far as the development of Police Motu was concerned. These "visitors" began coming to the area soon after Lawes arrived there in 1874 and continued to do so unchecked for the next sixteen years before the governor of the newly proclaimed colony of British New Guinea was in a position to control them. They included a wide range of ethnic types such as Chinese, "Malays", "South Sea Islanders", persons of mixed race, Maltese, Ceylonese, Indians, Filipinos, Europeans, Americans, and a large number of British-Australian origin. Many of these stayed and "married" Papuan women from the Port Moresby area and their descendants are still to be found there
Hiri Motu
19
today. Apart from a few extraneous cases these unofficial visitors generally fall into two classes: "South Sea Islanders" (who came mainly from what used to be called the New Hebrides but is now called Vanuatu) and "Malays" (who came from around the Singapore-West Indonesian area). Because these "visitors" were foreigners and were dependent on the Motu for food and other services, they would presumably have been in a situation similar to that that missionary Lawes was in when he first arrived. In other words they would have been treated like other visitors and addressed in Motu Foreigner Talk or Simplified Motu and not "true" Motu. And because these foreigners came from so many different sources (although collectively many of them may be labeled "South Sea Islanders" or "Malays"), and because they presumably had no language in common if they did not know some form of "broken" (or pidgin) English,10 the lingua franca of the Torres Straits and Queensland from where most, if not all, are known (or presumed) to have come to Port Moresby, there would have been increased pressure on them to learn and use some form of Motu as a lingua franca. Besides, they were in the minority and therefore in a weak social position. Consequently they would have been forced to accommodate to the Motu and not vice versa. Finally, because many of these foreigners were traders and traveled from one language area to another where some Motu was already known (and no English was) at the time of European contact,11 some form of Motu was bound to become the natural lingua franca between foreigner and non-Motu. In short, the whole sociolinguistic context of the Port Moresby area in the period preceding government intervention predetermined Motu in some form or other to becoming the established lingua franca of the area. But this was not just a change in frequency of use of the language; rather, it was a fundamental one in kind where its function was widened, or changed, from being a unidirectional, or vertical, one (that is, Motu-to-foreigner and vice versa) to a multidirectional, or horizontal, one (that is, foreigner-to-foreigner — includ ing non-Motu Papuans — and vice versa). Just what form of Motu developed out of this contact is difficult to say without written records. However, given (a) that present-day descendants say that some of their forefathers spoke "pidgin Motu", and (b) that, as will be indicated below, a subset of these foreigners was employed by the incoming Government from 1884 onwards and that thereafter Police Motu appears amongst those who were under their charge, it seems reasonable to assume that something similar was the lingua franca of this foreigner community.
20
Tom Dutton
That is, it seems reasonable to assume that a form of Simplified Motu similar to that taught to Lawes, but not necessarily identical with it — in fact, most likely not identical to it, given the different origins, social standing, and presumably different language learning abilities of these foreigners12 — was the lingua franca of this community. In any case, it would seem to be highly likely that this community would play a significant role in establishing a form of simplified Motu as the general lingua franca of the Port Moresby area. But there were other important social changes in and around Port Moresby at the time which also had their effect on the language. These changes were intro duced by the establishment of a Government presence in Port Moresby. When the four-year-old Protectorate of British New Guinea was con verted into the colony of British New Guinea in 1888, Dr. (later Sir) William MacGregor was appointed first Governor. He arrived in September, 1888, proclaimed the colony, and immediately set about the two main tasks of getting to know as much of the country as possible and of building up a suitable administrative structure with the limited resources available. At that time the colony was divided into three divisions — Eastern, Central, and Western — with headquarters at Port Moresby and Samarai and outstations at Rigo and in the Louisiade Archipelago where gold mining was developing. Port Moresby had been surveyed as a town in 1886 but had not yet developed into anything resembling one. At this time much of the country was still unknown and not yet under government influence or control. Those areas that were under some sort of control were those more or less coextensive with mission influence or areas such as the following where commercial activity had been going on for some time: the Port Moresby-Aroma coast and immediate hinterland in the Central Division, the Kiwai area in the Western Division, and the coastal parts of the mainland tip around Milne Bay and islands in the Louisiade Archipelago in the Eastern Division. With MacGregor's arrival conditions changed dramatically for the na tive population. Hitherto, the Protectorate Government had been merely conducting a holding operation, attempting as far as possible to avoid distur bances and to protect the local population from undesirable outside influences and from itself. It had few legal powers to make laws and to enforce them. But a colony was a different matter and it was the change in the nature and methods of the colonial Government that had such an impact on the local population and, in turn, on the linguistic situation. Indeed, the impact was so
Hiri Motu
21
great that by the time that MacGregor left British New Guinea in 1898, what was later to become known as Police Motu had become the principal, al though not the sole, unofficial language of administration in many areas and the scene was set for its further expansion into other areas as similar policies and methods continued bringing new areas under control. The principal agents in this development were the three instrumentalities of law and order: the police force, the village constable system, and the prison system. These three systems were closely related and integrated and fed on, and into, each other in the following way: the prisons provided recruits for the police force and the village constable system, which in turn provided recruits for each other as well as fresh prisoners. When Sir William MacGregor arrived to proclaim the new colony of British New Guinea, one of the most pressing requirements was for a suitable police force with which to help extend government influence over an increas ing area and to enforce law and order over those areas. Hitherto administra tors of the Protectorate had to depend on "commodore justice" and on an unofficial police force composed of a heterogeneous collection of foreigners and local Papuans who acted in various capacities but who had never been trained in any way for the police work they could be called on to do. Consequently MacGregor soon set about organizing an official force which he called the Armed Native Constabulary. But as no trained personnel were available in British New Guinea to form the nucleus of this unit, MacGregor appealed to the Governor of Fiji for assistance, with the result that two Fijians and twelve Solomon Islanders from the island of Malaita were recruited to go to British New Guinea for periods between one and three years — the two Fijians, a sergeant and a corporal, for one year and the remainder, constables, for three years. The members of this force all spoke mother tongues which are Austronesian languages distantly related to Motu. In addition the Solomon Islanders probably spoke some form of Fijian as a result of having spent between six and twelve years working on plantations and/or for the Govern ment in Fiji before being recruited, and all are presumed to have also spoken some form of English as a necessary prerequisite to being chosen to serve under English-speaking commandants in British New Guinea. This nuclear force arrived in British New Guinea in 1890 and was gradually expanded by the addition of Papuan recruits mostly from the Kiwai area of the Western Division, until, by the end of MacGregor's term as Governor in 1898, the force consisted of one hundred and ten Papuan non-commissioned officers and constables drawn from most parts of the country then under control.
22
Tom Dutton
The way in which the police force was formed and developed has a number of implications for the history of Police Motu, whose name is so closely associated with it. The most important of these are: (i) Because the founding members of the force were drawn from areas outside the Central Division, where "true" Motu was (and still is) spoken natively, they did not know any Motu on arrival. Consequently they must have been forced to communicate with one another and with their superiors in some form of English, which the majority, if not everyone, must have known. For the founding Fijians and Solomon Islanders this English was most prob ably what is generally referred to as "broken" English but was in reality a form of pidgin English similar to that found in Vanuatu and the Solomon Islands today. Similarly the founding Papuan members from the Kiwai area of western Papua spoke a form of "broken" or pidgin English, a creolized version of which is the primary language in the Torres Straits today and whose official name is Broken (Shnukal 1988). Thus even though Police Motu eventually became associated with the force as THE language of the force, it was not the sole language of the force; "broken" or pidgin English was equally important in the formative years. In fact both languages were required for a long time as each was used in different parts of the country and a member of the force could be transferred from one part of it to another on duty at any time. But to understand why only Police Motu became associated with the force (as the name indicates) and what it was like, we have to turn to the second main implication of the way the force was formed mentioned above. (ii) Because the force was formed in the way it was and was housed and trained in Port Moresby, and because the language situation in that area was the way it was (viz. Motu was the most widely known language while at the same time very little English of any kind was known), the police could not avoid learning some kind of Motu if they were to carry out their duties in the Port Moresby area. Thus, it will be remembered that the initial nucleus of the force consisted of imported Fijian and Solomon Islanders and that Papuan members were only added "gradually" (to use MacGregor's words). When these imported men arrived they would immediately have recognized a number of what would today be called wantoks (or persons from the same area and/or cultural background) in Papua New Guinea amongst the foreigner population of Port Moresby. As these wantoks had been in Port Moresby for some time and were living in or near Motu villages, it is presumed that they
Hiri Motu
23
could speak some sort of Motu. In addition it is presumed that they would have very soon introduced their newly arrived "friends" into the Port Moresby scene, in particular to its local politics and the utility and necessity of knowing some kind of Motu for survival purposes in the Port Moresby area. Even if this had not taken place, the new arrivals would surely have quickly come to the same conclusion themselves, for they were very soon put to work after their arrival in the Rigo and Mekeo areas east and west of Port Moresby. Here they would have been working with Government officials and other employees, some of whom were the wantoks already referred to and all of whom it is presumed (for reasons given above) spoke some kind of Motu. Consequently, they would have been exposed to Motu being put to use in the field. Presumably they would also have been told (even if they had not needed to know) that in the Port Moresby area "Motu" was the most widespread language and that one had to be able to speak it if one wanted to communicate with the local people. At the same time, they were probably given some elementary advice about its nature and possibly even how easy it was to learn. However, they would hardly have needed much of that kind of instruction since all of these men, as already noted, spoke languages related to Motu. Indeed, they would themselves have easily recognized many of its basic words and structural elements as being similar to their own. This can be seen by looking at Table 2, in which Motu words are compared with some of the most common words from some of those languages that are known or are suspected as having been spoken by various members of the force. For similar reasons, the form of Motu adopted by these first police is likely to have been similar to that used by the Government officers, other employees, and unofficial policemen with whom they worked and/or whom they were replacing — in other words, some form of Simplified Motu. The formation of the police force was thus a new and important element in the development and spread of this variety of language. It provided a particular social environment in which men from different parts of the coun try (and overseas initially) were brought together to work. In this situation, which is akin to those on plantations elsewhere, some common language of communication was soon needed. Initially, for reasons already given, this language must have been "broken" (or pidgin) English. However, the use of "broken" (or pidgin) English in the Port Moresby area was limited by the fact that "Motu" was the established lingua franca there. Consequently, although "broken" English never died out, its use in the Central Division was more
24
Tom Dutton
Table 2. Some basic vocabulary and structural elements of Motu, Fijian, and some Solomon Islands languages compared. Fijian13 English belly bird bone butterfly come die eye five fly (v.) foot go he his hornbill house I in life lobster louse mother my name paddle road salt, sea sugarcane taro two what your (sg.)
Solomon Islands Languages
Motu boga manu turia bebe mai mase mata ima roho ae lao, laka ia -na bina ruma lau lalonai mauri ura tumu sina -gu lada hode dala tasi tohu taro rua dahaka -mu
Bau
Kadavu
Lau
Kwara'ae Kwai Is
kete manumanu sui beebee lako mai mate mata lima vuka yava lako koya -na
kete manumanu tua bebe lako mai mate mata lima vuka laga lako kia -na
oga manu sili bebe mai mae maa lima lofo 'ae lea nia -na bina luma ñau lalo mouri ura 'uu tee -gu sata fote tala asi ofu alo rua ta -mu
sira hai'no' sulia beb mai' mae maa lima loh a'e leka nia -n(a) bina lum naua sae maori deng 'uu tea' -ku sata fote tal asi uuh alo rua tae -mu
14
vale yau e lomani bula urau kutu tina -qu yava voce sala maasima dovu dalo rua cava -mu
vale yau i lomani bula urau kutu tina -qu ila voce saa levu maasima tovu suli rua yava -mu
oga no suli bebe mai mae maa nima lofo 'ae leka nia ? bina luma nau laona mauo uragou 'uu tee ? rata fotee tala asi ofu alo rua taa ?
Hiri Motu
25
restricted than elsewhere. The fact that recruits were taken to Port Moresby for initial training before being distributed around the various Government stations meant that the language traditions, once established, were perpetu ated and, moreover, that because, as MacGregor noted in his Annual Report for 1892/3 (p. xxviii), "thirty or forty men leave the force each year for their own villages", Police Motu went with them to distant parts at a constant rate. This situation was reinforced by the two other ancillary law enforcement agencies, the village constable system and the prisons, both of which were considered by Governor MacGregor as important agents of change. These systems contributed recruits to the police force from an increasingly wide area and returned Police Motu speakers with them. Thus by the time MacGregor's period of governorship came to an end in 1898 there had been significant developments administratively and socially which had their effects on the linguistic situation. At first, MacGregor, like his predecessors in the Protectorate, had to rely on the mission and a number of locally appointed "servants" to maintain some sort of law and order. How ever, after 1890, Government influence and control rapidly outstripped mis sion influence and became the decisive factor in linguistic developments. By this time "Motu" had become the unofficial language of administra tion. This "Motu" was that that subsequently became known as Police Motu (and now Hiri Motu) because of its association with the police, notwithstand ing the fact that it was just as much part of the prison and village constable systems as of the police force. Thereafter this language spread (along with "English") with expanding contact with Europeans, so that today the language is spoken throughout most of what used to be called Papua. In 1971, in response to changing political circumstances in Papua New Guinea, a Study Conference on Police Motu was organized by the Government in Port Moresby to consider various aspects of the language's future. One of the results of that conference was the adoption of a new name for Police Motu. It was felt that because the term Police Motu had "become an anachronism" (Chatterton 1971:1) the name ought to be changed to something more in keeping with the origin of the language. As a result it was changed to Hiri Motu because, as has already been pointed out, it was thought that the language was a continuation of the hiri trading language. This new name was subsequently adopted as the official name of the language despite the misunderstood oral tradition that lies behind it.
26 5.
Tom Dutton Sources of the structure and vocabulary of Hiri Motu
Sourcing the structure and vocabulary of Hiri Motu is complicated by several factors — the generally poor quality and amount of relevant early data available, the complex nature of the external history of the language, and the number and nature of possible source languages available. Assuming, how ever, that Hiri Motu developed out of Simplified Motu as the social evidence outlined above strongly suggests, all discussion of the source of non-Motu features in Hiri Motu must begin with Simplified Motu. Given further that Simplified Motu was a register of Motu and not a separate pidgin language, then it is also reasonable to assume, especially when there is no evidence to the contrary, that this variety was composed solely of Motu elements even though some of these were apparently not used in strictly Motu ways and even though the variety probably varied from speaker to speaker and from context to context. Consequently all non-Motu features found in Simplified Motu, and later Hiri Motu, must have come from some external source or sources if they were not the result of some universal or internally generated tendencies. There are three languages most suspect of being probable source languages for such non-Motu features: Koita, the Papuan language spoken in and around Western Motu villages; Kiwai, the Papuan language spoken by MacGregor's first Papuan police recruits; and "broken" English, the initial lingua franca assumed to have been spoken by the founding members of MacGregor's police force. Each of the three aspects of language, phonology, morphosyntax, and vocabulary, are involved and will be considered separately. 5.1. Phonology In the beginning, as has just been suggested above, Hiri Motu, as Simplified Motu, is most likely to have had a phonology similar to that of Motu. As this variety expanded into a pidgin, however, it became increasingly diversified and the pronunciation in particular began reflecting the features of the mother tongues spoken by the individual users.15 Even so, and despite still observable variation within it, the language developed a number of features which are fairly consistent across the non-Central dialect (and which therefore help define it) irrespective of the nature of the background phonologies of the individual speakers. These features are:
Hiri Motu (i) (ii)
(iii) (iv) (v)
27
the reduction of the Motu kw/gw + au sequence to ko, as in koraia "inceptive aspect marker" (< Motu gwauraia "to talk about") ; the absence of a distinction between Motu /gl and /y/. Thus whereas Motu distinguishes between guria "to bury" and yuria "to pray", there is no contrast between these words in the non-Central dialect of Hiri Motu — they are homophonous; the contrast between Motu /r/ and /l/ is generally neutralized so that words like lau "I" and rau "leaf are both pronounced lau; /h/ is often lost, especially word-medially; the Motu contrast between /ao/ and /au/, between /ae/ and /ai/ and between /oe/ and /oi/ is lost so that each pair is realized as /au/, /ai/, and /oil respectively.16
Distinctive as these features are, it is not possible to trace them to any particular source. On social and historical grounds Koita and Kiwai are the most suspect, if these features were not progressively acquired from other Papuan languages as the dialect developed or if they did not develop indepen dently. Both Koita and Kiwai have a number of these features in common with each other and Motu — e.g., Koita has a contrast between /g/ and /y/, both Koita and Kiwai have /h/ and an open syllable structure and /a+u/ sequences, and Kiwai has a contrast between /r/ and /l/ (Wurm 1973:226). Consequently it is hard to see how they could have been sources. Likewise the languages spoken by the Solomon Islander and Fijian policemen are not likely to have been sources as they are related to Motu and are similar to it in general structure. 5.2. Morphosyntax This is the area of principal difference between Simplified Motu, Hiri Motu, and Motu. Compared with Motu, Simplified Motu and Hiri Motu are gener ally much simpler in structure. They also include a number of features not found in Motu. The features which distinguish Simplified Motu and Hiri Motu from Motu are those set out in the Appendix together with comparative notes for a selection of other languages that are relevant to the problem of sourcing.17 Comparing the Simplified Motu and Hiri Motu features, it is to be noted that, except for four features for which there is no evidence in Simplified Motu (viz. 11, 13, 17, 18), the only features that distinguish Hiri Motu from Simplified Motu are:
28
Tom Dutton 2 3 8 20
"Have" and "have not" Subject and focus markers ese and be Dekena(i) as a generalized locative postposition Word order flexibility
Such a correspondence between the two languages implies that many of the features found in Hiri Motu were acquired from Simplified Motu, and that Hiri Motu is indeed a continuation of Simplified Motu, as the social evidence presented above suggests.18 Where these features came from in Simplified Motu and Hiri Motu is, however, another question and one the linguistic evidence cannot answer unambiguously. On the one hand, most of the fea tures that distinguish Simplified Motu and Hiri Motu from Motu have paral lels in one or more of the languages the Motu were in contact with at the time of first European contact. On the other hand, all except perhaps feature 14 (gwauraia) can be accounted for as simplifications of Motu by the application of universal simplifying principles. But without further evidence it is not possible to make out a convincing case for the primacy of either of these two sources over the other. Indeed, it is highly likely that contact-induced changes and simplifying principles conspired with each other to give the observed results. Where Hiri Motu differs from Simplified Motu, however, it must be the case that Hiri Motu developed those features independently after it diverged from Simplified Motu, that is, after Simplified Motu began to be used in a wider context. As already noted there are four features involved (2, 3, 8, and 20). However, as two of these (3, 8) have similar structures in one or more of the languages that the Motu were in contact with at the time of the foundation of the first police force, or that the police themselves spoke, and as they are, at the same time, derivable from Motu by universal simplifying principles, it is clear that they are in very much the same category as the similarities dis cussed above, i.e. they most probably result from the interaction of simplify ing principles and contact-induced change. The remaining two features are of a different kind because they are not found in Motu and cannot be said to be simplifications of Motu. In fact feature 2 is as complex in Hiri Motu as in Motu and feature 20 is more complex in Hiri Motu than in Motu. Taking each of these in turn: (a)
Feature 2: "have" and "have not"
Hiri Motu
29
In Simplified Motu the method of indicating "have" and "have not" appears to have been similar to the Motu method in which the prepositions mai "with" and asi "without" are used without a verb, as in examples (1) and (2): (1)
Lau na mai egu ira. I be with my axe "I have an axe."
(2)
Lau na asi egu ira. I be without my axe "I have no axe."
In Hiri Motu "have" and "have not" are expressed either by using dekena(i) "at" with noho "stay" or noho lasi "not stay",19 as in example (3), or by using the possessive pronoun with noho, as in (4): (3)
Lau dekena(i) be ira ia noho (lasi). I at focus axe it stay (not) "I have (no) axe."
(4)
Lauegu ira ia noho (lasi). my axe it stay (not) "I have (no) axe."
Interesting as these differences are, however, it is not possible to source them unambiguously. The reason is that there are many competing possibili ties. First, they could represent independent developments. Second, they could represent changes induced by contact with the HTL(E), which has a (partly) similar construction. A third possibility is that they could represent developments induced by contact with the form of 'broken" English assumed to have been spoken by unofficial "visitors" and the founding members of MacGregor's police force recruited in Fiji. This is so because one way of expressing "have" was probably (judging by the fact that it is currently used in Bislama and other modern forms of the "broken" English spoken in the South Pacific in earlier times) to use a possessive construction with stap "to be", as in the following example from Bislama:20 (5)
Akis bilong yu i stap? axe of you be "Do you have an axe?"
30
Tom Dutton
A final possibility is that the constructions could have resulted from changes induced by HTL(E) and "broken" English constructions. Yet despite the uncertainty surrounding the origin of these "have" constructions, one thing is clear and that is that the constructions could not have come from the native languages spoken by MacGregor's policemen and those joining later, as these were Papuan languages which do not use similar constructions. Koita is excluded for similar reasons. (b) Feature 20: word order Word order in Hiri Motu is quite varied, depending on whether the subject (S) and object (O) are nouns or pronouns. There are four possibilities: (i) When both S and are nouns the order is S ese O V, as in example (6): (6)
Sisia ese boroma ia itaia. dog pig he see "The dog saw the pig."
(ii) When S is a noun and is a pronoun the order is be S ia V, or less commonly SVO, as in (7): (7)
Lau be sisia ia itaia. ~ Sisia ia itaia lau. I dog it see dog it see I "The dog saw me."
(iii) When S is a pronoun and is a noun the order is OSV, as illustrated in (8): (8)
Sisia lau itaia. dog I see "I saw the dog."
(iv) When S and are pronouns the order is OSV, or less commonly SVO, as in (9): (9)
Oi lau itaia. ~ Lau itaia oi. you I see ~ I see you "I saw you."
In Simplified Motu, however, word order was apparently invariably SOV (although there is unfortunately no evidence available for pronominal ob jects), as in (10):
Hiri Motu
31
(10) Ia mero itaia. he boy see "He saw the boy." This means that sometime between the documentation of Simplified Motu and that of Hiri Motu word order developed considerable flexibility, a flexibility not found in Motu, which has the basic order SOV for nominal Ss and Os but uses prefixes and suffixes on the verb for pronominal subjects and objects. Excluding ese and be, which are Motu features, either this develop ment must have been an independent one or it derives from the pidgin Englishes and/or other languages spoken by unofficial "visitors" or the first policemen. A further possibility is that it derives from Koita. Of these the most likely source is Koita, if any one particular source was responsible for this development in Hiri Motu, as it has some of the same flexibility as exhibited by Hiri Motu. For example both SOV and OSV word orders occur depending on the nature of the sentence (Dutton 1975). Whether or not Koita was the only or principal source, it would seem that the other possible sources named above could hardly have been sources as (i) the "broken" Englishes presumed to have been spoken by "visitors" and the early policemen did not have this kind of variation (judging by modern forms of them) and (ii) the majority of native languages spoken by the "visitors" and early policemen were Austronesian (including Indonesian, Solomon Islands, Vanuatuan, and Fijian languages) and Papuan ones which did not have this kind of variation. Clearly then it is not possible to identify the source or sources of word order flexibility in Hiri Motu with any confidence. 5.3. Vocabulary The vocabulary of modern Hiri Motu comes from several different sources. The largest percentage of words comes from Motu itself, as the name sug gests. This percentage includes the language's most basic vocabulary (that is, its pronouns, its names for common objects, body parts, kinship terms, adjectives, and simple action verbs) as well as a number of other words introduced by the South Sea Island pastor-teachers of the London Missionary Society who spearheaded the introduction of Christianity into Papua in the 1870s. These words include such items as pakosi "scissors", tamaka "shoes", pavapavana "king", and mamoe "sheep". The remaining Hiri Motu vocabu lary is made up of words that come from other languages of Papua as well as
32
Tom Dutton
from "broken" or other varieties of English that have been spoken in Papua throughout its history. Thus Hiri Motu contains such words as namba wan "first", bulamakau "beef, rais "rice", ti "tea", kesikesi or bisikesi "biscuit", traim "to try", and mikisim "to mix", which were part of the contact vocabu lary used by foreigners speaking "broken" or other forms of English, who came to the Port Moresby area in increasing numbers in the 1870s and 1880s. Subsequently other English words were added as new ideas and goods flowed into the country. The introduction of some of these can be dated quite precisely and their nature has changed over time as changes in technology and in English usage have occurred. The best examples of these are words for such introduced vehicles as aeroplanes, motor vehicles, and boats. Thus when aeroplanes were first introduced into Papua in the 1930s Papuans learned to refer to them as plaimasini or plaimasi "flying machines" in Hiri Motu. However, as aeroplanes became more popular and English speakers referred to them more generally as "aeroplanes" and eventually "planes", so the Hiri Motu word changed from plaimasini or plaimasi to elopleni and pleni. Simi larly trucks {traka in Hiri Motu today) were apparently first referred to as lori (from English "lorry"), and ships {boti in Hiri Motu today) as sisima or sitima (from English "steamer"). A small part of Hiri Motu vocabulary is also made up of words that come from languages both related and unrelated to Motu in Papua, and it is sus pected that many more such words were once part of the language.21 Some, such as kamkam "fowl, bush fowl", o kapore "oh sorry", dimdim "white man", and possibly susu "milk", can be traced to Suau, the Austronesian language related to Motu along the southern coast of the mainland just west of Milne Bay — see map. Others have come from Koriki (nakimi "brother-inlaw"), the substratum language of the HTL(K) trade language referred to above, from Binandere on the north coast (e.g. kiki "to yarn, tell a story"), 22 and from languages of the Torres Straits (mamoos or mamus "chief, police man"), all of which are unrelated to Motu. Thus linguistic features of Hiri Motu support the view suggested by social evidence that Hiri Motu is a descendant of SM, although it is not possible to identify the source or sources of most features of Hiri Motu that distinguish it from Motu and its predecessor Simplified Motu. Many of these were common to other languages that the Motu were in contact with at the time of first European contact but could equally well, and most probably do, represent the results of several processes — the application of universal
Hiri Motu
33
simplifying processes to Motu, changes induced by contact with one or more other languages which have similar structures, and independent develop ments.
6.
Conclusion
To recapitulate, in this paper I have argued, mainly on social grounds sup ported by comparative linguistic evidence, that Hiri Motu is not, as its name is meant to suggest, a continuation of one of the hiri trading languages used by the Motu in trading with linguistically unrelated peoples of the Gulf of Papua. Rather it is a descendant of a simplified form of Motu that was used by the Motu as a contact language with anyone coming to visit them in their own area. This language, unlike the hiri trading languages used by them, which were simplified forms of languages spoken by their Gulf trade partners, was a simplified form of their own language. This Simplified Motu was not related to those spoken on the hiri and was quite separate from them. It existed before the arrival of Europeans and other foreigners but was used by the Motu to communicate with the first European missionary in the area, Dr. Lawes, and taught to him as their language. The origin of this Simplified Motu is un known, but it most probably developed initially out of the contact between the Motu and their closest neighbors, the Koita, although others may have been involved. After the 1870s this Simplified Motu was extended in use by the arrival of a relatively large number of unofficial "visitors" who came to the area before a strong Government presence was established there. Some of these foreigners were later employed by the Government as interpreters, guides, boatmen, and unofficial policemen and were probably largely respon sible for a form of this language being adopted by members of the first official police force, from which the language took its early name, Police Motu. Thereafter the language became associated with the Government and the enforcement of law and order and spread with expanding Government control over the country. From that time on the history of the language became one of increasing competition with the other main lingua franca spoken in Papua before the Second World War, "broken" English. During this time also the language came into contact with other languages besides "broken" English as recruits were drawn into the police force from different areas and as the language expanded into different geographical areas. As a result the language
34
Tom Dutton
developed into two main dialects which varied in the sounds used and in some grammatical features (e.g. word order, object marking on verbs, possessive case) but especially in vocabulary (e.g. in non-Motu items obtained from other languages of Papua, "broken" English, and English). With the coming of the Second World War Police Motu was recorded and described for the first time and given official recognition. At the same time it was spread farther afield, was regularized, and was used increasingly for mass communication. After the war the language increased in status and "broken" English lost its identity as standard English was promoted officially with great vigor. In 1971 the name Police Motu was changed to Hiri Motu and the language began to acquire new ranges of vocabulary in keeping with increased contact with English and Tok Pisin, the other major lingua franca of Papua New Guinea since the Second World War, and in keeping with the changing social condi tions in an increasingly complex society. In 1975 it acquired equal status with Tok Pisin as one of two unofficial national languages.
Hiri Motu
35
Appendix Pidgin Features of Hiri Motu Compared with Those of Simplified Motu and Other Languages In the chart below, HM = Hiri Motu, SM = Simplified Motu, HTL(E) = Hiri Trading Language, Eleman variety, HTL(K) Hiri Trading Language, Koriki variety, and PPE
=
=
Papuan Pidgin English. Feature of HM
SM
HTL(E)
HTL(K)
PPE
l. No irregular verbs
Yes7 - most common ones like "corne, go" appear in simple form as in HM
Yes, no irregular verbs
Yes, no irregular verbs
Yes, no irregular verbs
2. "Have" and "have not" expressed by dekenai + noho orby POSS + N + noho
U ses both full form and short form ofMotu construetion with postposition mai "with"
Sometimes uses No verb used ita "with", sometimes amumuti "stay", but neither in negative sentenees
7- no evidence but presumably gat and stap as in Torres Straits Broken (of whieh PPE was an extension) and Bislama
3. Uses ese and be as subjeet mrkrs or foeus elements
No subjeet or foeus mrkrs (although evidenee very limited)
No subjeet mrkr No subjeet or mrkrs but foeus mrkr maro most eommonly used after subjeet
No subjeet or foeus marking elements
4. Restrieted dual pronoun form
Yes, same as in
HM
7 (free pronoun + oraore "two")
7 (free pronoun + 7 - no evidence rearea "two")
5. No bound
Yes, no bound forms; uses free pronouns taken from Eleman languages
Yes, no bound forms; uses free pronouns taken from Koriki
Yes, same as HM, exeept uses free pronouns derived from English
6. Transitive verbs Yes, same as in are marked by HM final-a
No, transiti ve verbs are not eonsistently marked
No, there is no transitive marker on verbs
No, although -im is often used in similar way
7. No noun classes Yes, same as in HM based on pos session (e.g. alienable vs. inalienable)
Yes, no noun classes but uses a relator enane derived from Eleman languages
Yes, no noun classes and no relators
Yes, same as in HM
Yes, same as in HM pronoun subjeets or objeets on verbs (exeept in Central dialeet); uses free pronouns derived from "true" Motu
Tom Dutton
36 8. Dekena(i) a generalized postposition
No, uses Motu postpositions
? - evidence patchy No, generalized No, uses along/long postposition used as generalized locabut when case is tive preposition marked there is for directions variation between suffix -ai « Motu) and preposed forms ta, ba and la « Eleman languages)
? - no evidence 9. Generalized forms for "like this, like that" and "how"
? - no evidence No, uses Eleman forms although these are similar in structure to HM ones, e.g. ma-feare (lit. this-like) "like this"
?- available 10. Restricted forms for "all" evidence suggests and "the whole" same as in HM
?- no evidence
? - no evidence
No, uses all and altugeta
? - no evidence
11. Restricted adjective agreement
? - no evidence
No, no adjective agreement at all
No, no adjective agreement at all
No, no adjective agreement at all
12. Simple imperative
Yes, same as in HM
Yes, but using Eleman material
Yes, but using Koriki material
Yes
13.Dohore and variants used before verb to indicate future
? - no evidence of ? - no tense No, but Koriki No, but baimbai such use in distinctions in material can be used in similar way available materials verbs, these being used in similar way indicated by although future tense is mostly indicated by adverbs of time such as aire "later, -varia on verbs afterwards", or vevere "tomorrow" for future; mamaro "now" for present. There is no evidence for past tense. If there is no adverb of time in sentence the tense can be read as past, present or future
14. Uses vadaeni, noho and gwauraia after verb to indicate completed, continuous, and intentive/inceptive aspect
No evidence of use of noho, but vada is used before verbs to indicate completed action (as in Motu) and gwauraia after verb (as in HM) to indicate intention
No, uses kaolaia ? - but -varia may after verbs for may be used for completed action intention and sinanaia (which is translation equivalent of gwauraia in HM) or mamaro after verbs for intention (which is translation equivalent of vadaeni in HM)
No? - but some evidence thatfinis is used in similar way for completive; evidence lacking for continuous .and inceptive aspects
Hiri Motu
1
15. Uses lasi as Yes, same as in generalized neg-HM: (a) lasi after ative: (a) after verbs; (b) lasi after verbs; (b) after adjectives adjectives 16.Reflexivity Yes, same as in expressed by HM sibona without special verb forms 17. Temporal ? - no evidence in clauses marked available materials. by negana(i) Juxtaposition the main relating device 18. Uses bema as conditional clause marker
? - no evidence in available materials. Juxta position the main relating device
19. Uses vadaeni Some use of "enough, okay' vadaeni similar to HM as sentence connective 20. Flexible word order depend ing on whether subjects and objects are
Yes, same as in HM: (a) lasi after verbs; (b) lasi after adjectives - no evidence
No,butpea is No, uses: (a) no be used in similar fore verbs; (b) no way: (a) pea after before adjectives verbs; (b) pea after adjectives ? - no evidence
■ no evidence
? - no evidence, Juxtaposition the main relating device
? - V + ane "and, No, uses juxtaposiwhen, if' or tion only juxtaposition used in available materials
- no evidence
No, uses V + ane No, although uses "and, when, if' sapos " i f in similar or juxtaposition way in available materials
No, although form tora "enough, okay" used in similar way
? invariable SOV SOV like Motu although no evidence for pronouns available
No, although form oaio "enough, okay" used in similar way
No, although finis "finished" and orait "okay" used in similar way sometimes
SOV
SVO
nouns or pronouns
Notes 1.
This paper is a condensed version of my book Police Motu: iena sivarai (1985).
2.
For descriptions of Motu see Lister-Turner & Clark (1930) and Taylor (1970). The standard variety is taken to be that described by Lister-Turner & Clark.
3.
Tok Pisin is the fastest growing language in Papua New Guinea. Until independence it was the principal lingua franca and unofficial language of administration of the northern half of the country, or that part of it that was known as the Territory of New Guinea. For detailed descriptions of it see Wurm & Mühlhäusler (1985) and references therein.
4.
For example, Chatterton (1950:5) says, "its origin dates back to before European penetra tion of Papua, when it was the trading language between the Motuans and their customers along the shores of the Gulf of Papua, to whom they bartered pots for sago."
38
Tom Dutton
5.
This was a movement led by Tommy Kabu. As a result of his wartime experiences Tommy Kabu felt that the only way to improve the living standard of his and other linguistically related peoples in the area was to adopt a single unifying language, Hiri Motu. At the time the population of the delta was estimated to be 6000, most, if not all, of whom belonged to the movement (Hitchcock & Oram 1967:5, 18).
6.
As will be indicated below, a form of Pidgin English now often referred to as Papuan Pidgin English following Mühlhäusler (1978) — but actually only an extension of the pidgin English formerly spoken in the Torres Straits and now creolized and referred to as broken (Shnukal 1988) — was spoken in some parts of Papua before Police Motu began to be disseminated.
7.
This is so mainly (a) because of its historical association with its parent Motu; (b) because this was the form of the language most similar to that propagated in early primers of the language; (c) because Central dialect speakers were the best educated (having been the first to be brought under mission and government control); and (d) because Central dialect speakers were readily available to act as translators and interpreters in Govern ment offices (on account of their distribution around Port Moresby, the admininstrative centre). Hence the impression usually given is that the Central dialect is the standard dialect, whereas there is no official standard dialect.
8.
The evidence for this claim is to be found in the various handbooks and language learning courses available on the language (e.g. Chatterton 1946 and others, Dutton & Voorhoeve 1974, Wurm & Harris 1963).
9.
Even so it was only with difficulty that Lawes was able to learn the true language, because many of the villagers were still opposed to imparting this knowledge to strangers, a position some of the older men maintained until the 1920s (Chatterton 1970:95).
10.
The use of the term "broken" English dates back to the very early days in Queensland. However, despite its name this language was really a pidgin language with features similar to those found in pidgin Englishes in the South-West Pacific and elsewhere today. In fact it is still referred to as Broken in the Torres Straits, where it has become creolized, as already noted.
11.
For example, Lawes notes in his journal on 4 April 1876 at Mailu (some 250 or more kilometers east of Port Moresby) that "as most of them know a little of the Port Moresby dialect I find I can communicate with them pretty freely", and again in June of the same year at Hula (about 100 kilometers east of Port Moresby) he notes, "most of the Hula natives understand more or less of the Port Moresby dialect".
12.
I say "presumably" here because without any evidence of the language learning abilities of these foreigners it is not possible to be more precise. Yet we know that individuals in general do display different language learning capabilities, and that level of education is not necessarily an indication of that ability. Thus, for example, J.H.P. Murray, the renowned Lieutenant-Governor of Papua for thirty-three years, admitted to attempting to learn "true" Motu but did not get very far with it. "It is said to be easy," he wrote in a letter to his brother George on 14th March 1904, "but for some reason or other I cannot get the hang of it" (West 1970:35). This lack of success annoyed him, the more so because of his knowledge of Latin, Greek, French and German — he was a judge and classical scholar. By way of contrast, most Papua New Guineans are excellent language learners and many are multilingual without ever having been to school.
Hiri Motu
39
13.
Bau and Kadavu are dialects of Eastern Fijian, which is generally distinguished from Western Fijian or Wayan.
14.
Hornbills are not native to Fiji.
15.
In general these languages are similar to Motu in having open syllables, five-vowel systems, and consonant systems that differ from it mainly in the number of distinctions made at the alveolar point of articulation. Probably the most extreme cases are to be found in some languages of the Gulf of Papua which do not have contrasts between stops, liquids, and/or nasals. As a result common Motu words like lau "I" or lao "go", dala "road", vanagi "canoe", and lasi "no, not" which contain some of these sounds are generally pronounced as , dana, maragi, and nasi respectively.
16.
These tendencies are apparently what has led some observers to claim that "it seems that there is some canon of "correctness" to which good speakers try to conform even when this involves distinctions which are not part of their native linguistic habits" (Wurm & Harris 1963:1). It has long been suggested by some that Hiri Motu, unlike Motu, has an /s/ phoneme. This latter claim must, however, be rejected as not substantiated by the data. Traditionally Motu had no /s/ phoneme — [s] was merely a distributional allophone of /t/ — although s was written by Lawes in reducing the language to writing and has become established. However, with the increasing familiarity with spoken and written English, from which many words are now borrowed into Motu, /s/ has become an added phoneme — e.g., sobea "to survey", so "saw, show", Sabati "Sabbath". The same has happened in Hiri Motu.
17.
The Simplified Motu features are based on those enumerated and discussed in Taylor (1978) and Dutton (1986) and those for Hiri Motu in Dutton (1985:8-16), which, for space limitation reasons, cannot be included here.
18.
Any other assumption is much more complex and therefore less preferable. For example, an obvious alternative is that Hiri Motu acquired its features independently of Simplified Motu and that both accidentally ended up with similar features because of similar social and linguistic forces acting on them.
19.
oth dekena(i) and noho come from Motu although dekena(i) in Hiri Motu is a generalized form of Motu deke + pronoun + ai, SL combination used to indicate nearness to persons, e.g. lau deke-gu-ai (lit. I towards-me-at) "near me" or "beside me". In Hiri Motu dekenaii) does not vary for persons and is used to express a wide range of locational meanings (such as "to, at, from") as well as the instrumental meaning "with".
20.
The most common way of expressing "have" in Bislama and Broken is by using the verb gat (< English got), but the following construction using stap is still very common, especially when the speaker is referring to an object that is on or about his/her person at the time of speaking.
21.
Thus, for example, Lock (n.d. : 3) noted that "nearly all the different tribes have their own formation of Police Motu" and (p.l) "the language takes its name from the Native Police Force whose members picked up more Motu words and added them to the Bastard language".
22.
See my 1980 article (fn. 5, pp. 194-95) for a discussion of the origin of nakimi and my 1987 article for an account of the origin of kiki.
40
Tom Dutton
References Brett, R., et al. 1962. A survey of Motu and Police Motu. Port Moresby: Department of Information and Extension Services. Chatterton, Percy. 1946. A primer of Police Motu. Port Moresby: The Department of Education. Chatterton, Percy. 1950. A Primer of Police Motu. Cairns: Cairns Post. Chatterton, Percy. n.d. (but 1968). A primer of Police Motu. Sydney: Pacific Publica tions. Chatterton, Percy. 1970. "The origin and development of Police Motu". Kivung 3.95-98. Chatterton, Percy. 1971. Hiri Motu {Police Motu). Port Moresby: UFM Press. Chatterton, Percy. 1975. Say it in Motu: An introduction to the common language of Papua. Sydney: Pacific Publications. Dutton, .. 1969. The peopling of Central Papua: Some preliminary observations. Canberra: Pacific Linguistics B-9. Dutton, .. 1975. "A Koita grammar sketch and vocabulary". Studies in languages of Central and South-East Papua, ed. by T. E. Dutton, 281-412. Canberra: Pacific Linguistics C-29. Dutton, Tom. 1980. "A Note on the Origin of the Motu term hirï". Kivung 12/2: 189-206. Dutton, Tom. 1985. Police Motu: iena sivarai [= its story]. Port Moresby: University of Papua New Guinea Press. Dutton, Tom. 1986. "More on Simplified Motu". Language in global perspective: Papers in honour of the 50th anniversary of the Summer Institute of Linguistics, 1935-1985, ed. by Benjamin F. Elson, 525-48. Dallas, Texas: The Summer Institute of Linguis tics. Dutton, Tom. 1987. "The Origin of kiki 'to tell a story, yarn' in Police (now Hiri) Motu". Language and Linguistics in Melanesia 15(1-2). 112-24. Dutton, .., and C.L.Voorhoeve. 1974. Beginning Hiri Motu. Canberra: Pacific Linguis tics D-24. Hitchcock, N.E., and N.D. Oram. 1967. Rabia Camp: A Port Moresby migrant settlement {= New Guinea Research Bulletin, No. 14). Canberra: The Australian National University. Lawes, W.G. 1896. Grammar and vocabulary of the language spoken by the Motu tribe, New Guinea. (Third and enlarged edition.) Sydney: Government Printer. Lister-Turner, R., and J.B. Clark. 1930. Revised Motu grammar and vocabulary. Port Moresby: Government Printer. Lock, Maynard. n.d.(but 1942). Police Motu. Unpublished MS held in the Australian War Memorial, Canberra. Mühlhäusler, Peter. 1978. "Papuan Pidgin English rediscovered". Second International Conference on Austronesian Linguistics: Proceedings, ed. by S.A.Wurm & Lois Carrington, 1377-1446. Canberra: Pacific Linguistics C-61. Shnukal, Anna. 1988. Broken: An introduction to the creole language of the Torres Strait. Canberra: Pacific Linguistics C-107. Taylor, A.J. 1970. Syntax and phonology of Motu {Papua): A transformational approach. Canberra: The Australian National University dissertation.
Hiri Motu
41
Taylor, A.J. 1978. "Evidence of a Pidgin Motu in the earliest written Motu materials". Second International Conference on Austronesian Linguistics: Proceedings, ed. by S.A.Wurm & Lois Carrington, 1325-60. Canberra: Pacific Linguistics C-40. West, Francis. 1970. Selected letters of Hubert Murray. Melbourne: Oxford University Press. Wurm, Stephen A. 1973. "The Kiwaian language family". The linguistic situation in the Gulf District and adjacent areas, Papua New Guinea, ed. by Karl Franklin, 219260.Canberra: Pacific Linguistics C-26. Wurm, S.A., and J.B. Harris. 1963. Police Motu: An introduction to the trade language of Papua (New Guinea) for anthropologists and other fieldworkers. Canberra: Pacific Linguistics B-l. Wurm, Stephen A., and P. Mühlhäusler, eds. 1985. Handbook of Tok Pisin (New Guinea Pidgin). Canberra: Pacific Linguistics C-70.
Pidgin Delaware Ives Goddard Smithsonian
1.
Institution
Introduction
Pidgin Delaware was used as a contact language between Indians and Euro peans on the Middle Atlantic Coast of North America during the Colonial Period. The Indians involved in its use were, in all known cases, speakers of varieties of Delaware, a grouping of two contiguous and closely related Eastern Algonquian languages, Munsee and Unami, each of which comprised a dialect continuum. Munsee was the language of the central and lower Hudson River Valley, westernmost Long Island, the upper Delaware River Valley, and the northern third of New Jersey; Unami was spoken to the south of Munsee in the Delaware River Valley and New Jersey.1 The European users were Dutch, Swedes, and English of the colonies of New Netherland, New Sweden, Pennsylvania, and New Jersey.2 Pidgin Delaware developed in the 1620s, and almost all documentation of it dates from the seventeenth century. It seems unlikely that, even at the height of its use, it would have been known by more than a few thousand people. Pidgin Delaware must have gradually gone out of use during the eighteenth century, although there is little documentation from this period. Presumably it was functionally displaced by American Indian Pidgin English, English, in some local situations other European languages, and, for a few non-Indians, Delaware. As late as 1785, however, the Delaware learned by Whites on the frontier contained Pidgin Delaware features. Pidgin Delaware was a pidginized form of Unami. Nearly all of its attested vocabulary has clear origins in Unami words and expressions. The dialects of Unami were most noticeably differentiated by phonological inno vations in Southern Unami that, under various conditions, dropped w, synco-
44
Ives Goddard
pated vowels, and deleted initial syllables. The Pidgin Delaware materials exhibit phonological features of both Northern and Southern Unami, but some words have only a Southern Unami shape or only a Northern Unami shape in all attestations. There are two major sources for Pidgin Delaware. The most extensive is from the Swedish Lutheran minister Johannes (or Johan) Campanius (b. 1601, d. 1683), who compiled a vocabulary of words and phrases in Pidgin Dela ware during his service in New Sweden, 1642-1648, and translated into it an adapted version of Martin Luther's Little Catechism, which he completed in Sweden in 1656. A small book containing the Catechism and vocabulary was published in 1696 for a Swedish Lutheran mission sent out the following year to the former Swedish colony, where it is reported to have been of some use in preaching to local Indians (Campanius 1696; Collijn 1937; Campanius 1938). The second major source is an anonymous vocabulary titled "The Indian Interpreter" compiled in West New Jersey in the late seventeenth century and surviving in a book of land records from Salem County, New Jersey (Anony mous 1684).3 Other sources for Pidgin Delaware include a few words and phrases in Lindeström's Geographia Americœ (Lindeström 1925:203; written in 1691 from notes made 1654-1656) and in a 1683 report by William Penn (Myers 1937:26-28), a few lines of dialogue and some numbers published by Gabriel Thomas (1698, 1:47, 2:7-13), and scattered words in various early records. It will be argued below (§8) that two other sources contain Pidgin Delaware and are in fact respectively the first and last documentations of it: an ostensible Northern Unami vocabulary from below the Falls of the Delaware at Trenton (de Laet 1633:75) and a Southern Unami vocabulary recorded among Dela ware emigrants in western Pennsylvania in 1785 (Denny 1860:478-481).4 This paper first surveys selected features of the phonology (§2) and morphosyntax (§3) of Unami and then describes the corresponding features of Pidgin Delaware — the phonology, as reflected in the different sources (§4), and the grammatical categories and word order (§5). Readers may wish to read the two phonology sections (§2 and §4) and the two sections on grammar (§3 and §5) in sequence. The sources of the Pidgin Delaware lexicon are examined in two sections, one treating more or less straightfor ward adaptations from Unami Delaware (§6) and one treating words that have idiosyncratic and miscellaneous derivations (§7). Agreement on these words in the various sources is considered as part of the evidence for the status of
Pidgin Delaware
45
Pidgin Delaware as a language, together with some aspects of Pidgin Dela ware vocabulary that shed light on the cultural context of its development. A final section summarizes the external history of this language (§8).
2.
Delaware phonology
The phonemes of Southern Unami (SU), as exemplified by twentieth-century Oklahoma Delaware, are displayed in Table l. 5 Some of the differences between Southern Unami and Northern Unami are relevant to understanding the sources of Pidgin Delaware. The Southern Unami vowels ɔ and ɔ reflect earlier wa and wa-, attested in Northern Unami (NU) and Munsee (Mu);6 in the corresponding sequences in Northern Unami the w is retained but the vowels are apparently rounded to [ɔ] and [ɔ] (generally written (o) and (oa) by the Moravians [e.g. Luckenbach 1838: 33]).7 (1)
SU -k 'and'
(2)
SU təlsi-né-ɔ 'they did it' ( CD *wətələsi-ne-wa8. NU (wtelsinewo) [wtəlsi-né-wɔ]. Mu wtəlnəməné-wa 'id.' (same inflection on a different stem).
Table 1. Phonemes of Twentieth-century Oklahoma Delaware (Southern Unami). Consonants p p-
m
t ts sn l
c čš š-
Vowels k x x-
l
h
ie e-
ə ə. -
-
46
Ives Goddard
In Southern Unami, short vowels (CD *a or *ə) have been lost before clusters beginning with s, š, or x, except after a rhythmically weak9 short-vowel syllable. In Northern Unami these vowels were retained in initial syllables (in known examples) but syncopated word-medially. (3)
SU xkó-k 'snake' ( CD *axko-k. NU (achgook) [áxko۔] (Luckenbach 1838:15). Mu axko-k.
(4)
SU skôntay 'doorway' < CD *əskwa-nte-w.10 NU (esquandey) [əskwantay] (Luckenbach 1838:39).
(5)
SU kší•k•an 'knife' ( CD *paxkdši-kan (cf. SU kší• 'my knife'). NU (bachkschigan) (Luckenbach 1838:43), (pachkschícan) (Zeis berger 1887:108), both [paxksí-k-an]. Mu paxkší-kan.
(6)
SU aləmské- 'they went' ( CD *aləməske-wak. NU (allumsgewak) [aləmské-wak] (Luckenbach 1838:42, 129).
Other rhythmically weak, short-vowel initial syllables were lost in Southern Unami and some varieties of Northern Unami. (7)
SU téme 'coyote, wolf' ( CD *rri9t9me-w (cf. mətme- 'he takes the trail'). NU (tímmeu) [táme-w] and (mĕtümmeu) [mətme-wl 'wolf (Zeisberger 1887:234). n
(8)
SU tá-x-an 'piece of firewood' ( CD *mdta-xan (cf. SU əətá-- 'my wood (pl.)')NU (tachan) [tá-x-an] and (metáchan) [mətá-x-an] (Zeisberger 1887:234).12
(9)
SU náni tnta-ləphitého-n 'there he killed him' 13 (ləp- ( CD *ə1ə-', cf. SU mpəlpi-téhɔ 'I killed him'). NU (nan wtenda luphittehan) [nán wtnta-təphitéha-n] (Lucken bach 1838:24); (plipitehan) [pələphitéhan] 'for one to strike him dead' (Brinton & Anthony 1889:118).14
Southern Unami retains some rhythmically weak, initial short-vowel syl lables, especially in shorter words.
Pidgin Delaware
47
(10) SU ahsn 'stone' ( CD *asən; cf. SU sənémhɔ-n 'metal spoon', ntasəéhɔ 'my spoon'. NU (achsün) [ahsn] 'stone', (achsünĕmhoan) [ahsəné-mh(w)ɔ-n] 'pewter spoon' (Zeisberger 1887:183, 142).
3.
Delaware morphology
Delaware words are typically morphologically complex. Primary noun and verb stems are composed of three types of constituent elements and may comprise an initial (in the case of a few basic nouns and intransitive verbs), an initial and a final, or an initial, medial, and final. Secondary finals derive stems from other stems, generally of a different category. The inflectional morphology includes a set of pronominal prefixes, an ablauting process on the first vowel of verbs in certain categories (called initial change), and numerous suffixes, which on verbs are arranged in seven position classes. Nouns are of either animate or inanimate gender; animate nouns include all logically animate beings and creatures and some non-animate objects (e.g. hítukw 'tree', ahp-n 'bread', hup-k-an 'tobacco pipe', ní-skɔn 'my elbow'). There are two sets of demonstrative pronouns, proximal and distal. Demon stratives are inflected or otherwise differentiated for gender and all other nominal categories and agree with the nouns they are in construction with or to which they refer. (11) SU yú 'this (inanimate)', wá 'this (animate)'; ni 'that (inan.)' ná 'that (anim.)'. Nouns are inflected for plural number (animate l-akl, inanimate l-al ( CD *-al). (12) SU ná hiló-s-ds 'that old man', pi. né-k hilo-s-ês-ak. SU ni ahsn 'that stone', pl. nél ahsna. Animate nouns may take an obviative ending, undifferentiated for number (l-al < CD *-al)\ the obviative is a secondary third-person category whose most salient function is to differentiate third-person animates. (13) SU ná lnu 'that man', né-l lênuwa 'that man (obv.), those men (obv.)'.
48
Ives Goddard
Nouns may be inflected for the absentative to indicate absence, loss, or death (anim. sg. l-al, inan. sg. l-el, other nominal categories 1-ənk()1). (14) SU á Inuwa 'that (absent or deceased) man', nikáhke ləúnka, nikáhke ləúnkahke 'that (abs.) man (obv.), those (abs.) men (nonobv. or obv.)'. Inflection for nominal possessor is by pronominal prefixes and possessorpluralizing suffixes. (15) SU nsí't 'my foot', ksít 'your (sg.) foot', wsí-t 'his foot'.
(16) SU nóx 'my father',ko-x 'your father',óxɔ'his father',15 nóxə 'our (exclusive) father', -ə() 'our (inclusive) fathe kó'Xəwa 'your (pl.) father', oxəwaɔ 'their father, their fathers'. Nouns also form a general locative with the suffix (1-ənkl). (17) SU hakihákan
'field', hakihákanink
'in the field'.
In addition to demonstrative pronouns there is a set of personal pronouns used for emphasis or contrast.16 (18) SUni'I',kl'you (sg.)', né'ka'he', nilóna 'we (exc.)', kilóna 'we(inc.)', kilówa 'you (pl.)', neká-ɔ 'they (anim.)'. Another pronoun set is used for reflexive objects and first- and second-person secondary objects (Goddard 1979:45-47). (19) SU nhákay'myself; me', khákay 'yourself; you (sg.)', hkaya 'himself. Verb stems typically are in derivationally related pairs, differentiated for gender. Intransitives are either animate intransitives or inanimate intransitives. (20) SU máxksu 'he is red' (animate intransitive stem lmaxkosi-1), maxke- 'it is red' (inanimate intransitive stem lmaxke-1). Transitive stems are specialized for objects of one gender as either transitive animates or transitive inanimates. (21) SU əáə 'I found him' (transitive animate lmaxkaw-1), əá 'I found it' (transitive inanimate lmaxk-1).
Pidgin Delaware
49
Intransitive verbs are inflected for subject, and transitive verbs for sub ject and (primary) object. Ditransitive verbs may also be inflected for a definite secondary object. The two major orders, independent and conjunct, use distinct sets of affixes for pronominal inflection;17 the second-person inflections in the non-negative imperative are also distinct. Each order has a number of submodes (Goddard 1979). (22) Independent order, indicative mode: SU kəníhəla 'you (sg.) killed him' (transitive animate stem lnəhl 'kill' + inflection lkə—a-(-w)l, ənihəláɔ 'you (sg.) killed them' (lkə—a-w-akl),wənihəláɔ'he killed him, them'(lw9-a-w-al), kənihəláwwa 'you (pl.) killed him' (Ikə-a-w-9wa-(w)l). (23) Conjunct order, subjunctive mode (suffix l-e-l): nhiláte 'if you (sg.) kill him, them' (l-at-e-l), nhiláte 'if he kills him, them' (l-at-el), nhilé'kwe 'if you (pl.) kill him, them' (l-ekw-el). (24) Imperative order, ordinary mode: nhil 'kill him, them (you sg.)' (1-Øl), nhílo- 'kill him, them (you pl.)' (l-o.l). The independent subordinative is used in focus constructions, including those with deictics (9) and particles (25) that function predicatively, and in senten tial complements (Goddard 1983:355-59). (25) SU tá háč təlkí-lən? 'how big is he?' (animate intransitive stem I9l9ki.1-l 'be (so) big' + lwə(t)-ənl 3 subd.).18 There is a marked preterite that implies an absence of current relevance (suffix l-əp, -əpan-l, etc.). (26) SU ahpúwak 'they are' (animate intransitive stem lapi-l 'sit' + l-w-akl), ahpópanik 'they used to be' (l-w-əpan-ikl). Negative verbs have both a negative particle and a negative morpheme as part of the inflectional ending complex (independent and conjunct l-(o)w(i)-l; imperative l-h-l, l-əyek-l, l-k9h-l). (27) SU nnéə'Isaw it', tá kó- nnemó wən T did not see it'. (28) SU ntakəót'Itold about it', máta ntak.dno-t-dmó-wdn 'I did not tell aboutit'.
50
Ives Goddard (29) SU pənáw 'look at him', káci pənawiyé-k-ac 'do not look at him'.
Of these negative particles, ta kó- and káci occur clause-initially and mata occurs preverbally.19 Word order is not used in Unami to differentiate grammatical functions. Subject, object, and verb may occur in any order. Position before the verb tends to be used for emphasis or new information, and hence indefinites. The word orders in the following examples are typical: (30) SU tá nhíle- ná l 'the man killed a bear, bears' (bear. OBV PRED he.kills.[him] that man);wənihəláɔnélmáxkɔ'he killed the bear, bears' (he.kills.him that.OBV bear.OBv). In equational sentences the new precedes the given: (31) SU éli khíkay méčini'because I'm an old woman now' (because old-woman now l pers./sg.).
4.
Pidgin Delaware phonology
The rendering of the sounds of Unami in the different sources on Pidgin Delaware varies considerably. There are three factors involved: the way the Unami sounds were pronounced by various speakers of Pidgin, the way the pronunciations of Pidgin were written down, and the way the written Pidgin words were transmitted in copying and printing. If the different written sources for Pidgin Delaware are considered together, evidence can be found for all the phonemic contrasts of Unami.20 A general idea of the various transcriptions may be gained from the way the numbers from 'one' to 'ten' are written in the four sources that record them (Table 2). The Munsee numbers illustrate the contrast between the two Delaware languages; from the different shapes of the numbers for 'five', 'nine', and 'ten' it is evident that the Pidgin numbers are specifically from Unami. The orthography of the de Laet vocabulary has a Dutch cast, but with many inconsistencies; these are caused in part by apparent attempts by the writer to use some spellings derived from a knowledge of other languages. The French-inspired (é) in 'one' (which seems to have spread to 'two'; Table 2) may reflect an attempt to capture a word-final short [I], a sound alien to
Pidgin Delaware
51
Table 2. The numbers from 'one' to 'ten'.
'one' 'two' 'three' 'four' 'five' 'six' 'seven' 'eight' 'nine' 'ten'
Munsee
Southern Unami
De Laet (1633)
Campanius (ca. 1645)
"Indian Thomas Interpreter" (1698) (1684?)
nkwti níša nxáh néwa nálan nkwəta-s nisas nxá-se nó.;li.' w i-mp at
kwêt-i nisa naxá né-wa palé-naxk kwtas ní.šas xá-s pé.škunk télənh
cotté nyssé nacha wyweb parenagh cottash nyssas gechas pescon terren
ciútte nissa nachhaa naéuwoc parénachd ciúttas níssas haas paeschung thæræn1
Cutte Nisha Necca Neuwa Palenah Cuttas Neshas Haas Pescunk Tellen
Notes: a b c d e f g h i
Kooty nisha nacha neo pelenach Kootash nishash choeshf peskonk telen
Also (nána). Misprint for *(nywe). Also (næuwo), (næwo). Also (pareenah), (pareenach). Most speakers have Mu xá-s. Probably the vowel sequence is misprinted. Also (pæschun), (paéskon). NU (metellen) [mətélən], (tellen) (Brinton & Anthony 1889:83, 140). Also (thææræn), (tææræn).
Dutch. The phoneme x is written (ch), (g), and (gh), but also (ck) (37), and Dutch recordings of Pidgin Delaware use these spellings for velar stops as well (37, 59, 243). The digraph (ch) is also used in its common French or English values, though probably not precisely enough to distinguish these from each other (45, 46), and it must be the English value of (sh) that is intended in 'six' (Table 2). In one case (sy) appears for š (37); perhaps (sj) was originally written. Other words in de Laet's vocabulary have (s) (or (ss)) for both č (58, 59) and š (Table 2: 'two', 'seven', 'eight', 'nine'). Many forms can be cited from Campanius that show a highly precise and consistent use of Swedish spelling conventions to render Unami phonetics. For example, in conformity with the Swedish of the time, long vowels and consonants are often indicated by doubling (Table 2: 'one', 'eight'; 32, 34, 37); also in conformity with Swedish, however, Campanius tends not to write long vowels before long consonants (Table 2: 'two'; 34, 36, 37), and he may
52
Ives Goddard
write a short stressed vowel as long before a short consonant (Table 2, note i; (tijrue) in 171; 192, 194).21 Campanius often uses accents to mark the main stress.22 The phoneme is rendered (k), (ck), and (ch) (Table 2; 'nine' and note g), in general conformity with Swedish spelling patterns; Campanius almost always intends (k) and (ch) before front vowels to have non-affricated values (contrast 43, 44). The phoneme x (a sound not found in Swedish) is sometimes written like (Table 2: 'five'; 37) or as (h) (Table 2: 'eight', notes a, d), but sometimes made distinct by adding a following (h), to give (ckh) (40, 229) or (chh> (Table 2: 'three'). 23 One word consistently has
(32; 35, 36), (tz) (33; 37, 38), and (in some prevocalic occurrences) ((s)si) (33; 39).24 Even š before low or back vowels, which might be expected to be readily transcribed with Swedish (si), Campanius usually fails to write distinctly (Table 2: 'two', 'seven'; 37). (32) (142) (sijs) 'more'; cf. SU čí.č. (33) (137) (dakótze) 'cold'; cf. SU ntákɔhči 'I am cold'. 25 (34) (140) (nissiaanus) 'child'; cf. SU *nničá.əs 'my child'.26 (35) (3) (suck)'but'; cf. SU šúkw. (36) (136, 137) (chíssogh) 'sun'; cf. SU kíšox. (37) (146) (tzackamææs) 'eel', L (syackameck); (sehacameck);27 cf. SU šɔxamekw. (38) (142) (étzkans) 'needle'; cf. SU éškans,
éskanš.
(39) (5-6) (mosiuttamen) 'bring down' (cf. 120, 199, note 107); cf. SU nəməš.ó'tamən 'I hit it with a shot'. In a few words in which Campanius desired to write prevocalic š he eschewed Swedish (si) and used German (sch) (40, 41, 42, 262), a spelling that can also represent sk or sk (Table 2: 'nine').
Pidgin Delaware
53
(40) (136) (scháckhan), (144) (sáckhang) 'wind'; cf. SU kSàx.ə 'the wind blows'. (41) (150) (scháta), (siaata) 'tobacco', II (shauta); cf. SU kwšá.tay. (42) (149) (schameera), (schaméra) 'run', (shamahala); cf. SU ksa.méhəle. 'he runs'. 28 In two cases (ch) before (i) is used corresponding to č (43, 246); this spelling probably reflects the affricated pronunciation of Swedish (k) before front vowels, which must have been familiar to Campanius even if it was not his own (Haugen 1976:268-70). (43) (143) (chirῶkha) 'leather strap'; cf. SU či . ló . he . s. 29 One example seems to have (ch) before (e) for š: (44) (37, 144) (tackcheene) 'open, open the door'; cf. SU tunksé.ni 'open it'. The nearly complete overlap in the spellings of č and š in the Campanius and de Laet vocabularies suggests that even when they were distinguished from s they were not clearly distinguished from each other. (45) (139) (siææt) 'tendons', L (cheet); cf. SU wčé .t 'sinew'. (46) (138) (siættæn) 'lips', L (chettoen); cf. SU wsé.t.o.n. There is some variation in the forms recorded by Campanius that sug gests that he drew on information from different sources. He gives the variants of some words side by side in the same entry (41, 47, 48, 49), and some variants differ more than seems likely for renderings of the same pronunciation. (47) (138) (alænskan), (laénskan) 'hand', 30 L (rinskan) 'fingers'; cf. NU (linschgan) [-lənčkan] 'finger' (Brinton & Anthony 1889: 64). (48) (147) (homijmi), (mijmi) 'passenger pigeons', 31 L (mymy); cf. SU amí.mi, Mu mí.mi.w.32 (49) (150) (jaésskung), (jésquem) 'corn'; cf. SU xáskwi-m.33 In at least one case dialect variants (cf. 7-10) have been borrowed with differentiated meanings.
Ives Goddard
54
(50) (150) (assísko) 'black or soft stone', 34 (145) (sísko) 'clay'; cf. NU (assisku) [ahsí-sku] 'mud, clay' (Brinton & Anthony 1889: 23), SU sí-sku 'mud' (SU ɔpahsí.sku 'clay', lit. 'white mud'). Initials beginning with SU ɔ. and NU [wo-] (1) appear in Pidgin Delaware with or without the initial w; stems made with Iwa-p-l 'white' are attested in both forms. (51) (142) (wopæk) 'white'; 35 cf. SU . p . e . k '(inan.) which is white', Mu wá.pe.k. (52) (142) (opijmi) 'chestnuts'; cf. SU 5.p.i.m, Mu wa.pi.m. Corresponding to modem Unami / Campanius usually writes (r) (Table 2: 'five', 'ten'), but he has (1) consistently in some words (47) and writes a few words both ways. (53) (146) (tulpa), (turpa) 'tortoise'; cf. SU tó.lpe 'water turtle (sp.)', NU (tulpe) 'sea turtle' (Zeisberger 1887:206). (54) (138) (hwijl), (wijr) 'head', II (wheel) "the head"; cf. SU wíl '(his) head'. Some words have an (h) added before an initial vowel (48, 55), though in most cases initial vowels are kept (52) and initial (h) represents Unami h (56) or x (Table 2: 'eight'). 36 (55) (145) (hóckus) 'fox'; cf. SU .k.wds, some speakers (cf. 63).
h.k.wəs
(56) (8) (hirússus) 'old man'; cf. SU hiló.s.ds. Certainly, different speakers seem to be represented in these variants, and very likely, in some cases, different dialects of Unami. The transcription in "The Indian Interpreter" has both English and nonEnglish features; in fact, some words appear as doublets, with both kinds of transcription (57, 60, 67). (57) (dogwatcha), (taquatse) 'cold' (cf. 33). Some of the non-English features could have been acquired orally by an English speaker learning Pidgin from a Swede or a Dutchman. For example,
Pidgin Delaware
55
some words have (s) for Unami š or č, a highly unlikely substitution for an English speaker but a common one in the Dutch and Swedish sources. (58) (sinques) 'wildcat', (145) (síngwæs), (singues), L (synquoy); cf. NU (uchtschìnque) [wcínkwe.(w)], (tschínque) [čínkwe.(w)] (Zeisberger 1887:33,231),°SU čínkwe. 37 (59) II (sickenom) 'turkey', (147) (síckenem), L (sickenum), T (chekenip);38 cf. SU čí.k.əə, NU (tschìkenum) (Zeisberger 1887:206). For some of these there are also spellings using English conventions that render these sounds correctly. (60) II (singa) 'when', (152) (singa), II (chingo) 'when'; cf. SU cinke. Assuming that the numbers in "The Indian Interpreter" (Table 2) were learned orally from a Swede or Dutchman would explain why Unami š appears as II (sh) before vowels, where [š] would be possible in Swedish or Dutch, but as II (s) word-finally or before a consonant, where it does not occur. The assumption of learning from Swedish speakers would also explain why the Pidgin reflex of SU ní.s.a 'two' has a short (i) in the English sources; the shortening would reflect the Swedish syllable-balance rule (note 21), which typically results in the shortening of long vowels before long conso nants in Campanius's recordings of Pidgin. Other variations in the way Unami sounds are reflected in 'The Indian Interpreter" include, as in Campanius, the use of both (r) and (1) and the sporadic appearance of (h) before initial vowels. (61) II (renus) 'man', (19) (rhénus), L(renoes), II(leno) 'man'; cf. SU l. 39 (62) II (hapitcha) "by and by", (4) (apítzi) 'right away', T (a peechi); cf. NU (apitschi) 'after a little while, by and by' (Brinton & Anthony 1889:22). (63) II (hoccus) 'fox'; cf. (55). The phoneme x is written like h (Table 2: 'five', 'eight') or (Table 2: 'three').
56
Ives Goddard
Some words in "The Indian Interpreter" exhibit distinctively Swedish spelling conventions, in some cases strikingly similar to spellings used by Campanius (cf. also 55 and 63, 60). (64) II (hwissamech) 'catfish', (146) (hwíssamick); cf. SU wí.s.ame.kw, NU (wísameek) (Zeisberger 1887:34). (65) (hwissase) 'afraid', (23) (hwisásse); cf. SU wi.s.á.s.u afraid'.40
'he is
(66) II (twn) 'mouth', (138) (tῶn); cf. SU wtó.n '(his) mouth'. (67) (mwes) 'elk', (6) (moos) 'animal', T (moos) 'cows'; cf. SU mos 'elk', NU (moos) 'elk; cow' (Zeisberger 1887:48, 65); 41 cf. II (muse) 'cow'. (68) (iwse) 'meat, flesh', (48, 138) (ῶs) 'meat, flesh, body', T (youse) 'venison'; cf. SU wiyo.s 'meat'. It seems improbable that it is only by chance that these spellings match so closely ones used by Campanius. In particular, the use of (hw-) as a lexically determined orthographic equivalent of (w-) is a spelling convention only in Swedish (and perhaps cf. 54). 42 Since "The Indian Interpreter" was entered into the Salem land records before Campanius's vocabulary appeared in print, it is likely that a written Swedish wordlist with some relation to Campanius's work was one of its ultimate sources. The extent to which other non-English features in "The Indian Interpreter" (58, 59, 60) may derive from written rather than oral Swedish sources is inherently difficult to estimate, however. In contrast to the earlier sources, Gabriel Thomas's recordings of Pidgin Delaware are internally more consistent and seem to be original. English spellings for the vowels, with their inherent ambiguity, appear to be universal. The digraph (ch) is used not only for č but also to render x in most words (Table 2: 'three', 'five', 'eight'). (69) T (chase) 'skin(s)', (145) (hææs), II (hayes) 'an undressed skin'; cf. SU xé.s 'skin'. (70) T (marchkec) 'red', (142) (mæckhhchæck) 'red' (cf. note 35). This use of (ch) must reflect a knowledge of Dutch, German, or Welsh. The spelling (ch) is also used for preconsonantal h.
Pidgin Delaware
57
(71) T (ochqueon) 'matchcoats', (142) (äquijvan), (quijvan) 'cloth, article of cloth clothing', II (aquewan) 'coat, woolen clothing'; cf. SU ahkwí.'blanket', NU (achquíwannes) [ahkwí.wanəs] (Zeisberger 1887:24). In other words Thomas writes (h) for x and omits preconsonantal h. (72) T (hayly) 'much', 43 (ahalea); cf. SU xahé.li,44 NU (mcheli) (Brinton & Anthony 1889:49). (73) T (andogowa), (undogwa) 'yonder', (149) (úndaque), II (undoque); cf. SU wntahkwi 'direction' (used with deictics). Thomas writes (1) in five words (e.g. 74) and two personal names and (r) in two words (e.g. 75) and the placename Arwaymouse. Initial (h) is added in two words (74, 75), one occurring as a doublet without it. (74) T (halopa) 'tomorrow', II (alloppau), (aloppau); cf. SU (some speakers haláp.a).
aláp.a
(75) T (orit), (horit) 'good', (13) (urítt), (21) (hurít), (1) (hωrítt), II (olet), (Denny 1860:479, 480, 481) (wellett), (willet) (268, 269); cf. SU wəlt 'it is good'.
5.
Pidgin Delaware grammatical categories and word order
Pidgin Delaware makes essentially no use of the extensive inflectional mor phology of Unami. Instead, what few grammatical categories it has are indicated lexically and syntactically. To be sure, there are traces of many Unami inflectional morphemes frozen in invariant Pidgin Delaware words. But even in the few cases in which Pidgin Delaware attests more than one inflectional form made on the same Unami stem, it appears likely that the Pidgin forms are descriptively unanalyzable units. The pronominal categories are reduced to first, second, and third person, undifferentiated for gender or number, and are marked syntactically by independent words. There appears to be no indication of the contrast between singular and plural. The three most generally used pronouns reflect the first and second singular of the Unami emphatic pronoun set and an emphatic form of the inanimate singular proximal deictic.
58
Ives Goddard (76) First person: (2) (nijr), (2) (nijre); II
Campanius's first- and second-person forms in (-r) and (-re) are from inher ited dialectal variants directly reflecting PEA *ni.r, *ki'.r.47 The three common pronouns (76, 77, 78) are used for subject (e.g. 79, 80, 85, 96), object (83, 96), indirect object (82, 89), and possessor (84, 86, 94, 101). They are all attested for both singulars and plurals: e.g. first plural (80, 81, 95, 101), second plural (88, 89), third plural (80, 96). The third-person pronoun, which'is both deictic and anaphoric, was used by Campanius for animates (80, 83, 95, 96) as well as inanimates (93, 94). His forms with (-e) are mostly attested for possessors (84, 91, 100, 101, 102) and for objects (89, 96), but see also (106). First person: (79) II (Nee hatta), T (Nee hata) T have'. (80) (8) (chéko nijr paéntor jῶni tahoáttamen) 'what we understand they want' (INDEF lpers. understand 3pers. like). 48 (81) II (Ne olocko toon) 'we run into holes' (lpers. hole go). (82) T (Chingo kee peto nee chase ...) "When wilt thou bring me skins ...?" (when 2pers. bring lpers. skin). (83) (27) (jῶni, hícka oraaton nijr) 'he keeps me safe' (3pers. away[?] keep lpers.). 49 (84) (93) (nijre Nῶk) 'my father' (lpers. father). Second person: (85) (15) (chijr rῶe) 'you (sg.) say'. 50 (86) T (Keco kee hatah kee weekin?) 'what do you (sg.) have in your house?' (what 2pers. have 2pers. house). 51
Pidgin Delaware
59
(87) (110) (chijr paétton mítzi suvvijvan mvvs) 'you (sg.) give food to all the animals' (2pers. give eat all animal).52 (88) (37) (chijr matta sijs cavínn cattúnga) 'do not lie sleeping any longer (you pl.)' (2pers. not more lie sleep).53 (89) (94) (nijr paétton chijre) 'I give to you (pl.)' (lpers. bring you). (90) (12) (chijr Nitérvus) 'your wife'.54 (91) (12) (chijre Rheenus) 'your husband' (2pers. man). Third person: (92) (144, 149) (jῶni ) 'this way, that road'. (93) II (Kecko ke lunse une) 'what do you (sg.) call this?' (what 2pers. name 3pers.). (94) (94) (ock paétton jomi, mochijrick Nitáppi) 'and gave it to his disciples' (and give 3pers., great friend). (95) (15) (nijr rῶe, jῶni matta manúnckus) 'we should say, he is not bad' (lpers. say, 3pers. not bad).55 (96) (21) (matta jῶmí tahóttamen nijre) 'they do not love me' (not 3pers. love lpers.). 56 (97) (43) (jῶni Rῶaénse) 'his name', or perhaps 'that name'. 57 There are also three-syllable forms of the first- and second-person pronouns. These are derived from Unami plurals but used for both numbers in Pidgin Delaware.The longer second-person form is from the Unami first plural inclusive rather than the second plural; this reinterpretation in Pidgin reflects the fact that in Unami the inclusive pronoun has the same formal relationship to the second singular that the exclusive has to the first singular (18). (98) (1) (nircῶna) 'my, our', II (Ne rune) 'we' (107); cf. SU ni.ló.na 'we (exe.)'. (99) C ( l ) (chirῶna) 'your (sg.,pl.)', II(Ke runa) 'thou' (108) (ke rune) 'thee' (109); cf. SU ki.lóma 'we (inc.)' (cf. SU ki.ló.wa 'you (pl.)').
60
Ives Goddard
Campanius uses these forms only for possessors, a function also performed by the variants (nijre) (in a few cases) and (chijre). (100) (37) (nijre Skinck) 'my eye', (nirῶna Skinck) 'my eyes'. 58 (101) (8) (nijre Saccheeman) 'our superiors', (6) (nirῶna móchijrick Saccheeman) 'our great Lord'. 59 (102) (7) (chijre Nῶk ock chijre Anna) 'your (sg.) father and your (sg.) mother', (21, 122) (chirῶna Nῶk ock Anna) 'your (sg.) father and mother' .60 (103) (1) (chirœna Nissiaanus) 'your (pl.) children'. 61 Thanks to the use of the longer forms of the second-person pronoun, possessednoun phrases like (102) and (103) are distinct from appositional phrases like (104) and (105). (104) (121) (chijr Nῶk ock Anna) 'you fathers and mothers'. 62 (105) (122) (chijr Nissiaanus) 'you children'.63 Only one case has been noted of (chijre) used appositionally before a noun (106). (106) (41) (chijre renáppi) 'you people'. In "The Indian Interpreter" what appear to be the same three-syllable variants used by Campanius are written as two words. They are used as either subject or object; the first-person example is translated as a plural (107; cf. 81), and the two second-person examples are translated as singulars (108, 109). (107) (Ne rune huscu hwissase) "we are afraid" (lpers. pl.[?] very afraid). (108) II (Ke runa matauka) "thou wilt fight." (109) II (Ne taulle ke rune) "I will tel thee." 64 Campanius also attests a third-person pronoun that reflects the Unami thirdperson animate singular emphatic. (110) (30) (naéchkomæ) "his"; cf. NU (nekama) [né.k.əm] "he, she, it" (Brinton & Anthony 1889:93).
Pidgin Delaware
61
The two examples noted of this pronoun, which are within three lines of each other, are each used for a third singular animate possessor. Perhaps this pronoun was used optionally to avoid the potential ambiguity of (jῶi), which before a noun could be a demonstrative (92) as well as, apparently, a pos sessor (97). Pidgin Delaware has an interrogative-indefinite pronoun derived from the Unami inanimate interrogative-indefinite (111). In Pidgin it used not only as an interrogative and indefinite (112, 113, 117, 118), but also as the head of indirect questions (114, 116) and as a relative pronoun (115). (111) (1, 7) (chéko, kéko) 'what?; someone, something', II (Kecko) 'what?', (cacko) 'something', T (Keco) 'what?'; cf. SU kéku 'what?; something, thing' (contrast SU awé.n 'who?; someone, person, living thing'). (112) (151) (Matta chéko maramen) 'no one will buy' (not buy). 65 (113) (153) (bakanta chéko) 'to hit someone' (hit
INDEF
INDEF). 66
(114) (Kec-loe Keckoe kee Wingenum) 'say what you (sg.) want' (2pers. say INDEF 2pers. want).67 (115) (21) (thaan jῶni chéko tahóttamen nijr) 'to those that love me' (to 3pers. INDEF love lpers.). 68 In many cases a pronominal subject or object is not used where its identity can be understood from the linguistic or pragmatic context. (116) (21) (Kónna nijr nirῶna Nitáppi Zaéband, ock chéko pijri hatte mátta kommvvta) 'that we do not steal our neighbor's money or whatever else he has' (that lpers. lpers./poss. friend money, and what other have[,] not steal).69 (117) (153) (Nitappi, chékῶ pijri hatte?) 'Friend, what other (tradegoods) do you have?' (friend, what other have).70 (118) II (Kacko pata) 'What have you brought?' (what bring). (119) II (Match poh) 'he has come' (already come).71 (120) II (Matcha pauluppa shuta) "I have catcht a Buck" (already buck successfully-shoot).72
62
Ives Goddard (121) II (Nee Meele) 'I will give you (sg.)' (lpers. give). (122) (Singa ke petta) 'When will you (sg.) bring it?' (when 2pers. bring). (123) T (Mata namen megis) 'I saw no sheep' (not see sheep). (124) T (Kacha hatah?) 'How many have you?' (how.many have).
The negative is indicated in Pidgin Delaware by a particle, derived from one of the Unami negative particles. (125) (1) (matta) 'not', II (matta), T (mata); cf. SU mata (28). The negative most commonly appears clause-initially, preceding the subject noun phrase, if there is one (also 96, 112, 123). (126) (152) (matta nijr tappin) T am not going to sit' (not lpers. sit).73 (127) II (Matta Ne kamuta) T did not steal it' (not lpers. steal). (128) II (Matta ne hatah) T have nothing' (not lpers. have), Penn (mattá ne hattá) T have not' (in Myers 1937:27). It also occurs immediately preceding the verb (129; see also note 45). 74 (129) (15) (jῶni matta manúnckus) 'he is not bad' (3pers. not angry/ bad; 95); cf. (13) (mátta nijr manunckus) 'we are not angry' (not lpers. angry/bad). Although the second-person pronoun appears before the negative particle in many examples in Campanius having a prohibitive use, this usage is not completely consistent. (130) (13) (Chijr mátta KommWta) 'thou shalt not steal' (2pers. not steal; cf. 127).75 (131) (37) (chijr mátta sijs cavínn cattúnga) 'do not lie sleeping any longer' (88); cf. (142) (matta chijr sijs cattunga) 'do not sleep any longer' (not 2pers. more sleep).76 The most common orders of the basic elements in transitive sentences in Pidgin are SVO (93, 96) and SOV (120, 132, 133; note 53).77
Pidgin Delaware
63
(132) (8) (nijr ock jῶni móchórick tahóttamen) 'and we should love them a lot' (lpers. and 3pers. greatly love).78 (133) T (Kee squa og enychan hatah?) 'Have you (sg.) a wife and children?' (2pers. woman and child have). An indirect object seems to precede (82) or follow (87) the direct object. Pronominal possessors always precede the possessed noun (84, 90, 91, 100103). Nominal possessors, attested only by Campanius, precede or follow. (134) (15) (Renáppi Rῶénse) 'a person's name' (person name). (135) (19) (pijri Rheenus Nitervus) 'another man's wife' (other man wife). (136) (17) (vickchῶmen pijri Renáppi) 'another person's house' (house other person). 79 (137) (18-19) (Nitérvus pijri Rhenus) 'another man's wife' (wife other man).80 Attributives sometimes precede and sometimes follow their noun and some phrasal verbs are used in either order (207 [also note 130], 219). Campanius often adds (-s) or (-z) to preposed possessor or attributive noun phrases in imitation of Swedish; it seems unlikely that this was an established feature of Pidgin. (138) (4) (jœni móchijrick Saccheemans, hῶrítt Manettos Roaénse) 'the name of the great Lord God' (that great chief-'s, good spirit 's name).81 (139) (15) (Hocquaéssungz ock Hackingz hwrítt Manétto) 'the God of heaven and earth' (heaven-'s and earth-'s good spirit).82 In equational sentences the given noun phrase may precede or follow the new: (140) (17) (jœni nirῶna) 'this is mine' (3pers. lpers./poss.). (141) Lindeström (1925:203) (nittappe kire) 'you (sg.) are my good friend' (friend 2pers.).
64 6.
Ives Goddard Pidgin Delaware words from Ordinary Unami
Although the vocabulary of Pidgin Delaware is overwhelmingly derived from Unami, it has undergone many changes in the details of forms and meanings. Many primary lexical items in Pidgin appear to be identical to their Unami source, while others show cluster simplification or other types of shortening. Some Pidgin words derive from Unami but in deviant shapes. In some cases this deviation can be attributed to phonological simplification, one typical pattern being the loss of initial unstressed short-vowel syllables (also 68): (142) (143) (gῶs) 'nail'; cf. SU əó.s. (143) (141) (pack) 'weep'; cf. SU ləpákw 'he weeps'. Where there are deviations from what is attested in Unami, the explanation is not always clear. For example, the word for 'duck' (144) is onomatopoetic and shows dialectal variation even within Unami. (144) L (comconcke) 'duck', (147) (hvvijquinck) 'ducks', (hvviquinck), (quing-quing) 'duck', (quingquing) 'ducks'; cf. SU kwi'kwínkdm, NU (quiquingus) 'large wild duck' (Brinton & Anthony 1889:122). The full-syllable reduplication in the forms in de Laet and "The Indian Interpreter" is probably a contact-language innovation, but obviously the possibility of additional Unami dialectal variants as a source cannot be ruled out. A few Pidgin words are the result of segmenting complex Unami forms, not always at the morpheme boundary (221). Unami dependent nouns, which obligatorily appear with a possessive prefix, would ordinarily lose any trace of the prefix as a result of the phonetic adjustments that generally result in reduced initial clusters and loss of initial unstressed short-vowel syllables in Pidgin (66). Some body-part nouns incor porate the third-person prefix lwə-1, which is used in Unami with forms having an indefinite possessor (54): (145) (138) (wippit) 'tooth', II (wippet) "the teeth"; cf. SU 'his tooth; a tooth', níp.i.t 'my tooth'.
wí.p.i.t
(146) (138) (wikijwan) 'nose' (cf. 153); cf. SU hwikí. 'his nose, a nose'.
Pidgin Delaware
65
(147) (139) (hwisquan) 'elbow'; cf. SU wí.skɔn 'an elbow', wískɔna 'his elbow(s)', NU (wisquon) 'elbow' (Brinton & Anthony 1889:163).83 (148) (139) (onáck) 'forearm'; cf. SU 'his arm; an arm', Mu
wəáxk.u (149) (139) (hutucki) 'shoulder'; cf. SU wtk.i 'his shoulder; a shoul der'. Other body-part nouns probably reflect first-person-possessed Unami forms, with phonetic loss of the prefix n-: (150) (138) (híttaock) 'ear'; cf. SU nhítaɔk 'my ear'. (151) (139) (hícchas) 'fingernail'; cf. SU nhíhas 'my fingernail'. (152) II (hickott) "the legs"; cf. SU nhíka-t 'my leg'. (153) II (hickywat) "the nose" (cf. 146);85 cf. SU nhikí. 'my nose'. At least one noun seems to incorporate the second-person prefix lkə-1: (154) (139) (kitte) 'heart'; cf. SU kté. 'your heart'. 86 The few dependent nouns of relationship reflected in Pidgin either lose any prefix phonetically or incorporate the first-person prefix: (155) (140) (hissimus) 'brother', 87 (issimus) (also Penn, in Myers 1937:27); cf. SU naxvs.əməs 'my younger sibling'. (156) (139) (nῶk) 'father' (see 84, 102, 104), II (noeck); cf. SU nó-x 'my father' (16). (157) (140) (nijlum) 'sister', (nijlon) 'lover, mistress', Denny (1860: 480) (neelum) 'sister'; cf. SU ní.ləm 'my sister-in-law (man speaking)' (see §7 and note 148). One Pidgin kinship term is from the Unami vocative: (158) (8) (anna) 'mother' (cf. 102, 104; also Penn, in Myers 1937: 27), II (aana) "a Mother"; cf. SU ána. 'mother!'. 88 Pidgin nouns are typically taken from the uninflected singular of Unami (36, 37, 38, etc.), and the same form is used for both singular and plural (45,
66
Ives Goddard
48, 67, 71, 87, 100, 101, 103, 104, 105, 106, 145, 152, 175, 176).89 Some Pidgin nouns incorporate the Unami locative ending, but without a locative meaning; these are all words that would often occur in the locative: (159) (26, 149) hácking 'earth', II (hocking) "the ground"; cf. SU hábí 'earth', hák-ink 'on the ground'. (160) (144) (sippússung) 'tidal stream, inlet', II creeke"; cf. SU *si.p.ó.s.ink.90
(sepussing) "a
Some Pidgin nouns are attested with and without the Unami locative ending. (161) (50, 140) (nῶttaas), (nῶttas), (139, 140) (nωttæssungh), (nωttæssung) 'belly; purse, small bag' (both meanings are at tested for both the short and the long forms), II (nutas) "a bag or basket"; cf. SU .t.e.s 'bag, suitcase', NU (menutes) 'sack, bag' (Brinton & Anthony 1889:81). (162) (27, 33) (tutæænung) 'real property; kingdom', II (nee tuttona) "my country"; cf. SUo . t . é . nay 'town^o.t.é.nink 'to, in the town'. 91 (163) (nacking) "the hand"; cf. SU nnáxkink 'in my hand' (cf. 148 and note 30). In the case of most of these doublets there was probably no difference of meaning, though Campanius suggests one in at least one case. (164) (143) (hachi) 'dust' (cf. 159). Unami verbs are taken over by Pidgin Delaware in various inflected forms, but always singular. Inanimate intransitives generally appear in the independent indicative (75, 165). (165) (150) (aritt) 'weak'; cf. SU alt 'it is rotten'. One set, with common adjectival meanings, is from Unami conjunct parti ciples (51; note 35). Verbs commonly used in time expressions appear in various Unami inflectional forms: (166) (136, 152) (nissochkenácka) 'the day after tomorrow'; cf. SU ni.šo.k.wənák.a 'in two days' (changed conjunct mode, referring to past time).
Pidgin Delaware
67
(167) (137) (haas kucknuckhácki) 'a week'; cf. SU xá.š txo.kwənakháke 'in eight days' (conjunct subjunctive mode, referring to future time). (168) (137) (rhoongor) 'winter', (152) (rωongor) 'in winter'; cf. SU énta-ló-wank 'when it is winter'.92 Unami animate intransitive verbs most often appear in Pidgin Delaware in their first-person or third-person independent indicative form. The firstperson forms are usually reflected without the prefix, but some Unami vowelinitial stems appear in Pidgin with an initial (t) that reflects the Itl intercalated after prefixes before stems of this shape (see note 91). Some stems are reflected in more than one variant, corresponding to different Unami inflec tional forms. (169) (33, 36) (entaami) 'rise up' (used with second-person and thirdperson subjects); cf. SU ntá.mi 'I got up (from lying)' (cf. á-mi'he got up'). (170) (153) (pæwo) 'come (imperative)', (45) (pæævo), (paévvo), (136) (pæwo) '(it) comes', II (payo) "to come," T (peo); cf. SU -w93 'he comes'. (152)(æmpaa) 'come' ((singa æmpaa) 'when will you come back?'), (45) (phaa) '(let it) come', II (match poh) "he is come or coming" (119); cf. SU mpá T come'. (171) (2, 4, 15, 6) (ῶ) 'say (imperative), (I) say, (you) say, (he) says', (loe) '(you) say' (see 114); cf. SU luwe. 'he says' (underlying lətəwe-l). (6, 13, 23) (tírue), (20) (tijrue), (150) (téroah) '(he) teaches', '(we) saying', '(he) says', '(he) com mands', '(I) say', II (taulle) '(I) tell (you sg.)' (see 109);94 SU ntluwe 'I say', ktluwe 'you (sg.) say'. (172) (143) (darrasymi), (ratísimo) 'dream'; cf. SU ntəla.s.i.mwi dream', la.š.í.mu 'he dreams'.
'I
Pidgin verbs that resemble the bare stem of Unami (e.g. 33, 57, 42, 65, (phaa) in 170) have more than one possible source. Formally, they could be firstperson (or second-person) forms with the prefix lost phonetically; stem variants occurring after preverbs, which would carry the prefix; or impera tives (assuming that the intransitive imperative singular ending -/ had already been lost in some dialects, as optionally in modern Southern Unami; Goddard
68
Ives Goddard
1979:141-42).95 The appearance of initial (t) or (d) on some stems that have an underlying initial vowel in Unami (171, 172) suggests that the Unami sources of some apparent bare stems may also have been prefixed, but some Unami vowel-initial stems are reflected in Pidgin as a bare stems without (173, note 98). (173) (3) (rωaénse) 'name (noun); be named', II (looense) 'be named', (lunse) 'to name'; 96 cf. SU ntəluwé.nsi 'my name is (so)', luwé.nsu 'his name is (so)'. At least one Unami animate intransitive verb is reflected in Pidgin by its conjunct participle; it is always used attributively: (174) (1) (mâchijrick), (2) (móchijrick) 'big', II (mockerick); cf. SU me.xkí.lək '(animate) which is big'. Pidgin verbs derived from Unami inanimate intransitives are freely used with animate nouns (175, 176), and the Pidgin word for 'big' (174), derived from a Unami verb used only with animates, appears freely with inanimates (177). (175) (8) (hῶrítt Manétto) 'good God', (135) (hωrit Manétto) 'good angel', T (huska orit chekenip) 'very good turkeys' (see 75). (176) T (neskec kabay og marchkec moos) 'black horses and red cows' (see note 35). (177) (Makerick Kitton) 'Delaware River, lit. big river' (1682 deed to William Penn, in Myers 1937:77; see 174).97 Some Unami animate intransitive verbs are reflected in Pidgin by subordinative forms (25) or, equivalently, forms with an inanimate secondary object; the two Unami forms are identical in shape and have similar or overlapping functions (Goddard 1983:356-58). The attested examples are verbs that have an inherent locative valence and are often in focus construc tions with a deictic. (178) (41, 141) (tappin) 'sit, be (there)' (cf. 126); cf. SU ntáp.i.n 'I sit (there), I am (there)'. 98 (179) (142) (cavínn) 'lie down', 99 II (Cowin) 'sleep'; cf. SU nkáwi-n 'I sleep (there).', nkát.a-kawín 'I am going to sleep (there)', kawín 'one sleeps (there)'.
Pidgin Delaware (180) T (weekin) 'house' (cf. 182); cf. SU nəwí.hi.n wwí'k'i.n 'he lives (there)'.
69 'I live (there)',
(181) (150) (nῶpon) '[smoke tobacco]', 100 Denny (1860:480) (nopone) 'smoke'; cf. SU ó.ɔ. 'I smoke it' (cf. húpwe. 'he smokes'). A few Pidgin forms in (-ng) probably reflect the indefinite-subject inflec tion of the conjunct order: (182) (17) (wícking) 'dwelling, home'; cf. SU wí.k.ink 'one's home, the house where one lives' (a participle with the locative valence as head; cf. 180). One Pidgin verb has variant forms reflecting the Unami preterite in l-p(an-)l (cf. 26): (183) (4, 7, 10) (ánger), (anckarópp), (ankaróppan) 'die, dead'; cf. SU ánkəl 'he dies, is dying' and (preterite) ankəlo.p 'he died', ankəló.p.ana 'he (absentative) died'. Some Pidgin verbs reflect Unami intransitive forms from more than one mode: (184) (141) (mattapping) 'sit, take a seat',101 II (matapi); cf. SU éntaləmátahpink 'where one sat', nləmátahpi 'I sat down'. Some Pidgin verbs, which presumably originated in Unami questions, occur in second-person form (also 201). (185) (150) (kῶmæn) 'come from'; 102 cf. SU kó- 'you come from (there)' (tá hác kó.ə? 'where did you (sg.) come from?'). Some Pidgin words incorporate the Unami question particle tá with a firstperson or second-person verb form. These seem to have been used in Pidgin as frozen expressions, but Campanius's usage points also to a more general interrogative use for two of them (187, 188). (186) II (Tackomen) "whence carnest thou?", T (takoman?) "from whence com'st?" (cf. 185, note 102). (187) (151, 152) (tacktaan) 'where are you going?; whither?', II (tacktaugh) "whether goest thou?"; cf. SU tá hác ktá.n? 'where did you (sg.) go?', tá hác ktá? 'where are you (sg.) going?'. 103
70
Ives Goddard (188) (152) (tanundin) 'where can I get it?', (43) (tanúndin) 'how?'; cf. SU tá háč núntdn? 'where do I get it from?'.
Unami transitive verbs are usually reflected in Pidgin by an inflected form of the transitive inanimate stem (21). One common type matches the objective (definite-object) forms of the independent indicative, which have the ending complex SU -ə (Class la), - (Class lb), -o-n (Class 2), or -ən (Class 3): (189) (2) (tahóttamen) 'love, like'; cf. SU ntah.t- 'I love it'. (190) (145) (képton) 'tie up'; cf. SU nkáxpto.n 'I tied it'. (191) (8) (nææmæn) 'see': cf. SU nné.ə 'I see it'. As in the case of animate intransitive stems, the Pidgin continuations of some transitive inanimate stems must reflect unprefixed Unami forms. (192) (145) (kræænamen) 'hold on to'; cf. SU át.-ələə 'I am going to hold it', ə1əəə 'one holds it, it is held' (contrast ə 'I hold it'). 104 (193) (26) (oraaton) 'keep'; cf. SU nkát.a-wəlá.to.n 'I am going to keep it', wəlá.to.n 'it is kept' (contrast no.lháto.n 'I keep it'). Some Pidgin verbs reflect Unami absolute (indefinite-object) forms, which lack the final -() of the objective forms.105 (194) (150) (baarham) 'miss'; cf. SU mpálham 'I miss (it)'. (195) (3) (maranijto) 'do, make, work'; cf. NU (mallenito) [maləní . to .] 'he makes (it)' (Zeisberger 1887:119).106 Some Pidgin verbs appear to reflect the imperatives of Unami transitive inanimates, ending in -a (Class la) or -/(Class lb): (196) (3) (bakánta) 'hit'; cf. SU pahkánta 'hit it (imperative)'. (197) (152) (rωksi) 'cut into pieces'; cf. SU ló.ksi 'cut it up (impera tive)'. A few Pidgin verbs are from transitive animate stems and incorporate various inflectional elements.
Pidgin Delaware
71
(198) (2) (pæntor, paéntor) 'understand', (37) (pæntῶre) 'under standing', II (pentor) "the ear or hearing"; cf. SU kpnto.l 'I hear you, I understand you' (contrast ntamə 'I hear it'). Some transitive verbs are attested in Pidgin in more than one form: (199) (5-6) (mosiúttamen) 'bring down' (39),107 (shuta) 'catch' (120); cf. SU nəməš.ó.t. 'I hit it with a shot', məsó.t.a 'hit it with a shot (imperative)'. (200) (17) (pætton) 'bring, give', (144, 152) (pææt) 'bring here, give me', (pata, petta) 'bring', T (beto) 'bring (me)'; cf. SU mpé.t.o.n 'I brought it', mpé.t.u 'I brought (it)'. (201) (2) (quinkijnamen) 'desire', II (wingenum) 'have a mind to (have)', (wingenun) 'like, care for'; cf. SU əwinkí.ə 'you (sg.) like it', winkínam 'he likes (it)', NU (winginamen) [winkí.nmən] 'to delight in, to be pleased with' (Brinton & Anthony 1889:160). (202) (151) (kommijre) 'give in return', (152) (mijre) 'give in return', (150) (mijrikon) 'trade', (151) (miraana) 'give away', (152) (skomerijne) 'will give', (Meele) 'give', (SickeMelee) 'will give', (maleema) 'give (me)'; 108 cf. SU əí.li 'you give to me', mí.li(.l) 'give to me (imperative)', nəmí.lkwən 'he gave it to me', nəmi.lá.na 'I gave him or them (animate or inanimate) to him or them', (na)=č kəíli. '(then) you (sg.) will give it to me', mi.li.me 'give (it) to me (future imperative)'. 109 Pidgin words derived from Unami verbs are often used attributively (e.g. 174, 206) or as nouns (173, 198, 203, 205), and in some cases Unami nouns appear as Pidgin attributives (203) or verbs (204): (203) (2) (nissiaanus hwisásse) 'childlike fear' (child afraid/fear; cf. 34, 65). (204) (27, 52) (hackiháckan) 'field', (149) (hacchiháckan) 'to plant'; cf. SU haki.ha.han 'field', haki.há.k.. 'he plants'. (205) (1) (chíntika) 'holy (1), prayer (43)'; cf. SU nkntka 'I dance'.
72
Ives Goddard (206) (50) (chísbo) 'full', 'completely)', (1) (chísbo simoáckan) 'commandment' (full statement); cf. SU kí.spu 'he is full, has eaten his fill'. 110
The compounding of nouns and verbs serves to build up the vocabulary of Pidgin Delaware: (207) (145) (Manúnckus mochijrick Síngwæs) 'mountain lion' (an gry big wildcat), L (Synquoy Mackyrggh) (wildcat big). 111 (208) (149) (Mamantíckan takhan) 'peachtree, plumtree' (peach/ plum wood).112 (209) (140) (hââs táckquin) 'skin clothing' (skin wear).113 (210) II (Hockung Tappin) 'God' (above sit).114 Many Pidgin expressions are made with (hatte) 'have' 115 and (maranijto) 'do, make' (195): (211) (23) (hwisásse hatte) 'be afraid' (afraid/fear have).116 (212) (137) (nijr uránda hatte) T am hot' (lpers. hot/heat have). (213) (3) (maranijto manúnckus) 'to abuse, do wrong to' (make/do bad[n. 111]).117 (214) (141) (sevarænda hatte) 'be troubled', (10) (sevaraénda ma ranijto) 'offend'.118 (215) (Dutch rendering, ca. 1628) (collatis[-]marrenitten) '(she) ob serves the isolation taboos at first menstruation' ("she bewails or laments over her virginity"; Isaack de Rasieres, in Jameson 1909: 107). (216) (Dutch rendering, 1655 [recalling earlier events]) (rancontyn marenit) (1632), (Rancontyn Marinit) (1643), 'make peace, make a firm peace' (David de Vries, in Myers 1912:16 and Jameson 1909:230).119 Campanius used (chíntika) 'prayer, holy' (205) to form a number of expres sions pertaining to the Christian religion: (217) (4) (maranijto chintika) 'to worship, to pray' (make/do prayer/ holy).
Pidgin Delaware
13
(218) (37) (chíntika Manétto) 'Holy Spirit' (prayer/holy spirit).120 (219) (6) (chíntika saccheeman), (140) (sacchéman chintika) 'priest' (prayer/holy chief, and vice versa).121 (220) (1) (chíntika mamaræckhickan) 'holy scriptures' (prayer/holy writing).122 Although many of Campanius's coinages are inventive, they generally seem to conform to the structure and spirit of Pidgin. One set of striking neologisms that ring true uses 'fathom', the unit of measure for strings of wampum (221), in the meaning 'honor' to differentiate two ranks in the church hierarchy below bishops (222): (221) (139) (wíckan) 'one fathom', II (Cutte wickan cake) "one fathom of Wampum"; cf. SU na.fk.an 'my spine', ni.s.a.í.k.ane 'two yards of wampum', NU (guttawican) 'one fathom' (Zeisberger 1887:70), (tchaikane) '(so many) cubits' (Luckenbach 1838:27).123 (222) (116) (nissa Víckan) 'double honor', (114) (Kiútte Vickan Saccheeman chintika) 'priest' (one fathom/honor chief prayer/ holy; cf. [219]), (114) (passeera Víckan Saccheeman chíntika) 'deacon' (half fathom/honor chief prayer/holy). There are function words, particles, and the like in Pidgin that appear to reflect collocations of Unami elements and other major shifts of category: (223) (4) (taan), (2) (thaan) 'to, of, from, until' (115, n. 102); from Pidgin (taan) 'go' (. 103).124 (224) (49) (tarássi) 'end', (57) (Tharássi) 'death'; cf. SU ntala-=č ... 'I will stop ...' (preverb ala 'stop' with pronominal prefix and future enclitic).125 At least one Pidgin noun is also used prepositionally: (225) (144) (aana) 'road, way, route', (ana mochijrick bij) 'by sea' (road big water); cf. NU (aney) 'road, way' (Zeisberger 1887: 160). 126
74
Ives Goddard
Some Pidgin Delaware words appear to exhibit patterns of word-final free variation. These patterns are reminiscent of, and sometimes identical to, some of the common patterns of variation that stem from the reflection in Pidgin of more than one inflectional form of some Unami words. Several of these patterns involve the addition of a nasal (including (-ng)),127 the replace ment of a nasal by a different nasal, or the unexpected presence or absence of a final vowel (see also 49, 40, 111, 200). In some cases only a form with the non-Unami innovation is found. (226) (136)(chischvickhe), (6)(chischvíckingh), (137)
Pidgin Delaware
75
(234) (145) (arum) 'dog'; cf. NU (allúm) (Zeisberger 1887:59), from PA *a6emwa (cf. SU mwé.k.ane). (235) (45, 51) (aéppe) 'also, with'; survives only in prefixed form in the pronominal set SU né.pe 'I also', etc. (236) (146, 151) (aétticke) 'perhaps, but, rather', T (etka) 'or; with (in lists)'; cf. SU ét 'maybe, I guess, I would conclude', Mu é.t (both shortened forms).133
7.
Pidgin Delaware words not from Ordinary Unami
Although there is variation exhibited within and among the different sources that document Pidgin Delaware, the agreements among these sources are sufficient to show that a Unami-based pidgin language was in use in the Middle Atlantic region during, at the least, the early Colonial period. The most telling of these agreements involve Pidgin words that deviate from Unami in some way and are thus unlikely to have been learned more than once directly from Indians speaking true Unami: (237) L (renoes) 'man', (19) (rhénus), (renus), (leno) (61); it is unlikely that the dialectal form in (-s) was in widespread use in Unami (the last source shows optional replacement by the usual form; see note 26). (238) (25) (síxi) 'quickly', II (mesickecy) 'make haste'; cf. perhaps SU məsí 'all over, here and there' + ksi 'quickly, hard' (not likely to have been a frequent combination). (239) (chirῶna) 'your', II (Ke runa) 'you' (see 99); though the exact usage differs, both Pidgin sources show a shift from first person inclusive to second person. (240) (1) (suhwijvan) 'all', once (26) (suhwij), <sawwe) 'all'; cf. SU čúwi 'full (of)', but wé.mi 'all' (unlikely to have been a common misunderstanding). (241) (153) (hiῶ mátzi) 'I am going away', II (Iough Matcha) "get thee gon"; cf. SU əá.č.i 'I go home', má.c.l 'he goes home', but the first word is uncertain.
76
Ives Goddard (242) (Dutch rendering, 1640s) (materiotty) 'cowards' (NYCD 1:182), (13)(mattarútti) 'good for nothing, useless', 135 II (Matta ruti) "its good for nothing"; no obvious source. (243) (115) (kaccheeus) 'drunk, drunkard', (Dutch rendering, 1658) (cacheus) "crazy, mad or drunk" (NYCD 13:84), II (cakeus) 'be drunk'; cf. SU nkak.í.wsi 'I am drunk', kí.wsu 'he is drunk'. The Pidgin sources consistently reflect the prefixed form of the stem (minus the prefix; cf. 171, 172) without the final vowel. (244) (Dutch rendering, 1640) (keene, keene, keene) 'thank you' (New York Historical Collections 1913:125), (4) (kéne), (27) (keene, keene) 'thank you', (English rendering, 1675) (kenon, kenon) (NYCD 13:524); no obvious source.136 (245) (27) (ítta) (a word used for salutation and address), Lindeström (1925:203) (ita),137 T (hita), (hitah) 'friend (vocative)'; no obvi ous source. (246) (2) (móchij) 'yes', 138 Lindeström (1925:203) (måkij) 'yes, in deed', II (mochee) "ay or yes," T (mogy) 'yes'; perhaps cf. SU nəmáwi=ĉ 'I will go and' (prefixed preverb and future enclitic; cf. 224 and note 125). (247) (6) (lamaéæs) 'fish', II (lamiss), T (lames) (a personal name); cf. rare dialectal SU lamé.s139 beside usual names (from PA * name-? sa).
The diversity in the Pidgin Delaware materials in part reflects the dialec tal diversity of Unami and is thus an indication that Pidgin was used over a dialectally diverse area. Since, however, there is notable variation of this kind within each of the major sources for Pidgin, it was not simply the case that there were regional variants of Pidgin based on local dialects. A further indication that Pidgin Delaware was learned by Delaware speakers is the fact that it was used in the Munsee-speaking areas of New Netherland. For example, the Dutch heard the Pidgin word (cacheus) 'drunk' used by Esopus Indians, near the northern edge of Munsee-speaking territory, in 1658 (243); de Vries recorded a Pidgin expression used by Indians on western Long Island, near the eastern edge of Munsee Speech, in 1643 (216); and Cornelius Melyn was addressed in Pidgin Delaware by Staten Island Indians in 1649 (244, 253; New York Historical Collections 1913:125).
Pidgin Delaware
77
Some variation in the Pidgin materials must be a result of the renewal of vocabulary and phonological shapes by Unami speakers whose usage re flected changes taking place in the Unami dialects. For example, the replace ment of (renoes) 'man' by (leno) (237) reflects both the fact that SU lə (without the -s suffix), which was probably always the most common form, eventually became the only Delaware word for 'man', 140 and also the fact that seventeenth-century Delaware r came to be completely replaced by l by at least the middle of the eighteenth century.141 Pidgin sources show a clear drift towards increasing use of / as opposed to r (Table 2; 53, 54, 74, 75). In contrast, variants showing the addition of h before initial vowels (48, 55, 62, 63, 74, 75) seem to persist in all major Pidgin sources, corresponding to the fact that this dialectal feature was still present in the pronunciation of some Unami speakers in the twentieth century. Similarly, the Pidgin materials show both the Northern Unami retention of w and the Southern Unami loss of w initially and intervocalically (1, 2, 51, 52, 71). As noted, some of the variation in Campanius and "The Indian Inter preter" seems clearly to indicate that these compilations were based on more than one source, which must have included other written materials (41, 47, 48, 49, 57, 60, cf. 64-68). This type of variation thus points to the active transmis sion of knowledge of Pidgin Delaware among Europeans, for which there is also internal evidence (see Table 2; 58, 59, and the discussion following 60). Only a few vocabulary items came into Pidgin Delaware from languages other than Unami. In most cases, however, it is likely, or at least quite possible, that these contributions to Pidgin by its European users were not introduced directly from their ultimate language of origin. (248) (13) (zaéband), (151) (zææband), (27) (sijband), (85) (zéban) 'money', II (sewan) 'wampum' (see also 221); New Netherland Dutch zeewant, sewan, etc. (Friederici 1960:675), a borrowing from Munsee *sé.wan 'loose wampum', lit. '(the) scattered (item)'. (249) (copy) 'horse', T (kabay); from Spanish caballo, but presum ably introduced by Dutch or English speakers who felt their own words were insufficiently exotic. (250) II (Cutte gull) "one guilder or 6 pence"; (gull) is ultimately from Dutch gulden 'guilder', but perhaps by way of Munsee or North ern Unami, one or both of which had this word (Brinton & Anthony 1889:45).
78
Ives Goddard (251) II (Cutte steepa) "one stiver or farthing"; (steepa) is not attested in Delaware but presumably had the same history as (gull) (250), ultimately from Dutch stuiver. (252) (brandywyne) "Rum, &c"; an English borrowing of Dutch brandewijn. (253) (Dutch rendering, 1640) (orit nietap) 'good friend' (New York Historical Collections 1913:125), (10)(nitáppi), (nitáppe) 'good friend',142 Lindeström (1925:203) (nittappe) 'good friend', II (netap) 'good friend', Penn (Myers 1937:27) (netap) 'friend'; this word seems not to be idiomatic in any variety of Delaware.143 (254) (squaw) "a wife," T (squa); from the southern New England pidgin (cf. Massachusett (ussqua), (squa) 'young woman' [Goddard & Bragdon 1988:725]; Narragansett (squàw) 'woman' [Williams 1936:134]).144 (255) II (papouse) "a sucking Child"; also from southern New England (cf. Narragansett (papoòs) 'child' [Williams 1936:28]). (256) (57, 85) (hulijken) 'good', only in (matta hulijken) 'wrong, evil (thing)' (also [21, 27] (hulijk), conforming to the pattern of final nasal loss); not idiomatic in Delaware (cf. (wuliken) 'it grows well' [Brinton & Anthony 1889:174]) but the usual word in New England (with dialectal variation) for 'be good' (cf. Western Abenaki (wligen) 'it is good' [Laurent 1884:66]; Southern New England Pidgin (wunnegin), (waureegun) [Goddard 1977:40]). (257) (148) (sállsenbras) 'sassafras tree'; ultimately from Spanish sasafrás.145
A number of place-names and personal names in early records appear to be Pidgin Delaware (see also 177), and some of these have the flavor of coinages by Europeans: (258) (Sewanhacky), (Sewanhacking) 'Long Island' (Tooker 1911: 232); obviously a jargon expression for 'wampum land' (cf. 248, 159). (259) (Hobocan Hackingh) 'Hoboken' (NYCD 13:1); evidently a jar gon expression meaning 'pipe place' (cf. [143] (hopóckan) '(tobacco) pipe'; 159).146
Pidgin Delaware
79
These two place-names seem to reflect, respectively, European interests and European metaphorical concepts, and their shape shows that they cannot have been formed in Unami. If they were single words, Unami morphophonemics (Goddard 1979:xiii-xiv) would exclude the h-initial forms of 'earth, land' (159) after the rhythmically weak (see note 9) final syllables of Munsee *sé.wan (248) and Unami hup..k .an 'pipe' (cf. Pidgin [143] (hopóckan)); if they were compound stems the first members would require the prenounforming suffix SU -í.i, Mu -í.(wi). (260) (Dutch renderings, 1649) (Kickeeu-sickenom), (Kickee Sickenom) (a chief with authority over lands on the Delaware River near New Sweden; NYCD 12:48-49), lit. 'Live Turkey' (1656; NYCD 1:598); cf. (141) (chícke) 'life, to live' (SU kí.k.e 'he heals, recovers, lives after peril of death'), (59). (261) (Dutch renderings, 1651) (Wappanghzewan), (Wappingzewan) (a chief with authority over land on both sides of the lower Delaware River above New Sweden; NYCDl:596-597), lit. 'white (?) wam pum'; (51?), (248). The contributions of Unami-speakers to the formation of Pidgin Dela ware appear not only in the overwhelmingly Unami vocabulary and grammar but also in some aspects of the way Unami forms were in effect selected. This is particularly evident in the reduction of gender and number categories. To European traders and settlers the wealth of Delaware inflectional morphology would have been an impenetrable mass of seemingly unmotivated and effec tively unlearnable variation (11, 12, 20, 21). The gender system would have been especially difficult to acquire, since many common objects (such as skins, kettles, and pipes) would have had to have been referred to with animate forms of demonstrative pronouns and verbs, which included the most common forms expressing adjectival meanings. In any attempt to use real Delaware for everyday purposes like trading, Europeans would constantly have been using inanimate forms for grammatically animate objects. The solution to this problem in Pidgin was to generalize the forms that referred to inanimate singular. This collapsing and generalization affected the demon strative pronouns (78), the interrogative-indefinite pronoun (111), and most intransitive verbs with adjectival meanings and transitive verbs. The only exceptions are the extremely common adjectival form for 'big' (174) and a few transitive verb forms that centrally (or only) refer to actions between
80
Ives Goddard
people, especially between speaker and addressee (198, 202). No collapsing was possible in intransitive verbs referring to common human states and activities (such as 'fear' [65] or 'sleep' [179]), for which only animate forms exist in Delaware, and these were retained as such. It seems evident that this systematic selection of the inanimate singular as the sole third-person cat egory could only have been accomplished by speakers of Delaware, who must have generalized this pattern from the erroneous use of Unami gender by Europeans. Particularly in cases like the words for 'shoot', which are most commonly used, and most useful, with animate objects (e.g. 199),147 it seems unlikely that Europeans would have consistently learned and adopted the inanimate forms in Pidgin. What is especially striking about this generaliza tion is that it resulted in the collapsing of some distinctions that were common to all the native languages of those who used Pidgin Delaware, such as the distinction between singular and plural and the distinction between animate and inanimate interrogative-indefinite pronouns. Another feature of Pidgin Delaware that helps shed light on the context in which it was developed is the presence of what might be called low words. In Delaware more-or-less serious negative opinions can be expressed by adding to otherwise ordinary verbs and nouns morphemes that refer to inti mate body parts. These expressions are regarded as off-color and are avoided by most speakers, but some men of earlier generations were recalled in the 1960s as having been especially adept at creating them. Some formations of this type are used in Pidgin Delaware as ordinary words: (262) (141) (chalebackschaétti) 'poor man, beggar' (used by Campanius for 'servant'); cf. perhaps SU *kələp.akšé.t.i 'you are weeping (expletive)' (SU kəlp.akw 'you weep' with the incor poration of SU -sé't.i 'anus'). (263) (4) (manúm charakítte) 'be angry' ; cf. SU *manunka.lahkí't-i 'an angry so-and-so' (manunk- 'angry' [cf. note 111] + -a.lahki.t.i 'rectum'; or from the corresponding verb). (264) (115, 141) (nisketij) 'stingy' (only in (mátta nisketij) 'gener ous'); cf. SU *ní-skti 'dirty-butt' (ni.sk- 'dirty' +-ti 'hind parts'; or from the corresponding verb). It seems unlikely that Campanius appreciated the literal meaning of these undignified expressions or the racy import that they would have had in
Pidgin Delaware
81
Unami. Their use as ordinary words in Pidgin Delaware must reflect the somewhat rough-and-tumble, exclusively masculine social context in which Pidgin would have taken shape as a traders' jargon. The same social context accounts for some aspects of the Pidgin vocabulary referring to women: (265) II (Kins Kiste) "a Maid ripe for Mariage"; Pidgin (Kins) 'now' (C [21, 137] (kinz), (kintz) 'now, becoming, almost'; cf. SU č 'just (now)') and (Kiste) 'ripe' (cf. SU kí.ste. 'it is cooked, it is ripe'). The existence of the Pidgin expression for 'observe taboos at first menstrua tion' (215) also effectively documents the use of Pidgin to discuss the sexual availability of young women. In the same vein, the fact that the word given in different variants as 'sister' and 'lover' derives from the Unami for '(man's) sister-in-law' (157, 266) must reflect its use in negotiating sexual access to Indian women. In traditional Algonquian society the women of a man's generation were classified by the kinship system into sexually taboo "sisters" and sexually available "sisters-in-law"; any woman a man called sister-in-law (or the equivalent) was a potential sexual partner or wife.148 A European would be a classificatory "brother" of his Delaware trading partner and would seek or be offered sexual partners from among the women his brother called sister-in-law, who would often be 'sisters' to each other.
8.
History of Pidgin Delaware
The earliest reference to the existence of Pidgin Delaware was by Jonas Michaëlius, who in August, 1628, clearly and somewhat acerbically de scribed the language used between the New Netherland Dutch and the Indians as a "made-up, childish language", greatly simplified, by the conscious design of the Indians, from what the Indians spoke among themselves (Jameson 1909:128). The first permanent Dutch settlements in the territory of any of the Delaware groups had been not in Munsee-speaking territory on Manhattan but in Unami-speaking territory on the Delaware River, specifi cally on Burlington Island (at Burlington, New Jersey, some distance below Trenton) and at Ft. Nassau (on the site of Gloucester). The first settlers had arrived there in 1624, and they were all removed to Manhattan between November, 1626, and October, 1628 (Jameson 1909:84, 88; Weslager &
82
Ives Goddard
Dunlap 1961).149 When de Laet brought out the Latin edition of his book The New World in 1633 he added a section on the Indians of New Netherland that had not been present in the Dutch edition of 1630 (de Laet 1633: Chapter 11; Jameson 1909:57-60). In this new section is a vocabulary that he specifically states to be from the Sankhikan Indians living on the upper Delaware River (meaning the upper part of the navigable river, which ended at the falls at Trenton). Most of the words de Laet gives could be either true Unami or Pidgin Delaware, but there is a complete lack of plural endings where the glosses call for them, which is a Pidgin feature. Distinctively Pidgin is the form of the word (Renoes) 'man' (237) and the Pidgin phrase (SynquoyMackyrggh) '(mountain) lion' (207). The phonological divergences from Unami also parallel those in Pidgin sources (Table 2; 37, 45, 46, 47, 48, 58, 59, 144, 207). A Dutch settler from the Delaware River must have been the ultimate source of de Laet's vocabulary, and it must have been settlers and traders who had been dealing with Unami-speaking Indians living on the Delaware River who introduced Pidgin Delaware to Manhattan after the settlement of New Amsterdam in 1626. The spread of the knowledge of Pidgin Delaware from Dutchmen to Swedes and from Swedes to Englishmen is effectively attested to by the documented use of Dutch interpreters by the Swedish settlers of New Sweden and of Swedish interpreters by the English in Pennsylvania (Myers 1912:86, 88; Weslager & Dunlap 1961:151, 181; Collijn 1937:8-9; Myers 1937:82; Goddard 1995). The last echoes of Pidgin Delaware are in the Delaware vocabulary recorded in 1785 by Ebenezer Denny. Most of this vocabulary is ordinary Southern Unami, but a few expressions that are not correct Unami match the Pidgin of the seventeenth century (Denny 1860:478-81): (266) (neelum) 'sister' (157). (267) (nopone) 'smoke (tobacco)' (181). (268) (keigh willet lenew) "you good men." (269)
Pidgin Delaware
83
The historical and linguistic data indicate that Pidgin Delaware origi nated as a contact language used between Dutch traders and speakers of Unami on the Delaware River in the 1620s and came to be used over the entire Delaware-speaking area for communication between Dutch, Swedes, and Britons and Unami-speaking and Munsee-speaking Delaware Indians. No direct or indirect evidence has been found to support the hypothesis of Thomason (1980:182-86; Thomason & Kaufman 1988:175) that Pidgin Dela ware was already in use in pre-contact times between Delawares and Iroquoian-speakers. Against this hypothesis is evidence that the Iroquoian neighbors of the Delaware did not speak Pidgin Delaware in the seventeenth century. There are direct statements about the difficulty of communicating with Iroquoian speakers and about the inability of interpreters for the Dela wares to understand them (Megapolensis, in Jameson 1909:172; NYCD 13:497). To communicate with Iroquoian speakers, Europeans struggled as best they could to learn their languages, perhaps acquiring pidginized ver sions of them (Thomason 1980:183). This effort is reflected in the vocabular ies that the Dutch recorded of Mohawk (van den Bogaert 1988:51-63) and the Swedes recorded of Susquehannock (Campanius 1696:157-160). It is also significant in this regard that early accounts of Iroquois Indians are scattered with Iroquoian words (chiefly Mohawk), while accounts of Munsee-speakers and Unami-speakers use Pidgin Delaware words in describing and quoting the Indians. Thomason discusses the fact that there is no evidence that Pidgin Dela ware was used by any Indians other than Delawares, and hence no evidence that would directly support pre-European use. 151 She further amply docu ments that there is essentially nothing in the grammar or lexicon of Pidgin that must come from Germanic languages, but a good deal that can only come from Delaware. Especially telling are certain features of word order, such as object optionally before verb (132, 133), optional (but predominant) sen tence-initial negative particle (96, 112, 123, 127, 128), and the possibility of given information following new (141); these features are characteristic of Unami, absent or rare in Germanic, and unmotivated by putative universal patterns of unmarked word order. The typological parallels that suggested to Thomason (1980:180-81, 184) that a pidgin arising between Europeans and Indians would contain substantial contributions from the language of the Europeans cannot be accepted as decisive. The typology must be based on the full range of occurring types, and Pidgin Delaware provides a surprisingly
84
Ives Goddard
well documented case of a type she considered unlikely or impossible: a pidgin used between Europeans and non-Europeans that was based exclu sively on one of the languages of its non-European users. The Pidgin Delaware case also suggests the circumstances under which such a pidgin with a non-European base can arise. Whether or not, to use the traditional terminology of pidgin studies, the Delawares were dominant and the Dutch subordinate, it is entirely to be expected that the tiny population of the first Dutch settlers and traders, especially on the Delaware River, would have tried to deal with the local Indians in the Indian language. Michaëlius's early and contemporary description of the conscious role played by Delaware speakers in simplifying their language to create Pidgin Delaware can be taken at face value and is confirmed by the systematic pattern of the collapsing of Unami categories found in Pidgin. Thus the particular (and typologically atypical) facts of its formation are indeed accounted for by particular features of its sociolinguistic setting, which may well also be atypical among docu mented cases (Thomason 1980:181). The example of Pidgin Delaware shows that in the formation of a pidgin the particular circumstances of the contact situation can completely overwhelm any effect from general typological characteristics observed to be significant elsewhere.
Notes 1.
General information on the Delaware languages and dialects is in Goddard (1978a:7273); their names are discussed in Goddard (1971:14, 22); historical and ethnographic background and maps of locations are in Goddard (1978b). There is evidence for three major Unami dialects, Southern Unami (which became the dominant form of speech among the emigrant Delawares that ended up in Oklahoma), Northern Unami (the dialect used by the Moravian missionaries in their linguistic materials and translations from the eighteenth and nineteenth centuries), and the poorly known Unalachtigo.
2.
From 1676 to 1702 New Jersey was two separate English colonies, West New Jersey and East New Jersey (Barber & Howe 1844:20, 29).
3.
I am indebted to Lorraine Williams for a facsimile copy of the manuscript. It has been published by Nelson (1894:133-139) and, somewhat less reliably, by Prince (1912). Nelson gives the title of the volume in which it is found as "Salem Town Records, Liber B", while Prince gives it as "Salem Surveys, No. 2". As it is apparently upside down within the volume as later used, the modern archival pagination runs in reverse. Nelson describes it as "recorded" (i.e. copied into the records), a description supported by the fact that the first page appears to have been written on the verso of another document; Prince refers to it as "inserted," as though it were a separate manuscript. Both Nelson and
Pidgin Delaware
85
Prince ascribe to it the date of 1684, Prince reporting that it is "next to" a deed bearing this date; the facsimile of the first page, however, clearly shows the presence of the number 1686 on the other side of the leaf. An autopsy of the manuscript, which I have not yet had the opportunity to undertake, should resolve these minor discrepancies. Besides Prince, Pidgin Delaware has been discussed by Thomason (1980), focussing on "The Indian Interpreter," and Holmer (1946), on the basis of Campanius; the transcrip tions of Delaware words given by Holmer are not reliable. 4.
Cited Pidgin Delaware words and phrases are identified below by initial, as follows: = Campanius (1696); D = Denny (1860); II = Indian Interpreter (Anonymous 1684); L = de Laet (1633); T = Thomas (1698). Page references are added for Campanius. Other sources are given in full.
5.
Each word has a single main stress, indicated by an acute accent. There are also geminates: pp, tt, čč, kk, hh, mm, nn, 11, ww.
6.
Historically CD *wa(-) ) Southern Unami ɔ(-) except after CD *; there are a few exceptions or restorations.
7.
Phonemic transcriptions of Southern Unami and Munsee are given in italics; these are based on my field recordings (1965-1970). Letter-for-letter transliterations of Northern Unami (in the Moravians' German-based orthography) and of Pidgin Delaware are in shallowpointed brackets; approximate phonetic interpretations of these transcriptions are in square brackets (some decisions about how similar the details of these phonetic interpretations should be to modern Southern Unami are unavoidably arbitrary). Vertical bars set off segments, morphemes, and morpheme complexes written in morphophonemic (underly ing) transcription; they correspond to the slashes (virgules) in Goddard (1979). Capitali zation in the sources has generally been preserved only in cited phrases. Asterisked preforms are Common Delaware (CD) or, where there is no Munsee cognate, pseudoCommon Delaware. Other protoforms are Proto-Algonquian (PA) or Proto-Eastern Algonquian (PEA).
8.
Some vowel sequences arising from the loss of w in Southern Unami are retained, while others insert y, as in this case.
9.
Rhythmically weak short vowels are the odd-numbered vowels in underlying or Common Delaware sequences of one or more non-final short-vowel syllables ending in a single consonant or consonant + w (Goddard 1979:ix, 21); as a result of analogical leveling, vowel weakening is no longer a completely automatic process synchronically.
10.
Cf. the particle Mu əskwá .ntĕ 'at the door'. In Unami, uninflected inanimate nouns replace a stem-final CD *-e.w with -ay.
11.
Cf. (misprinted) (metimnewak) [mətemé-wak] 'wolves' (Zeisberger 1821:57); L (metumnu) (also misprinted). The same entry in Zeisberger (1887:234) also has (tumme), which is probably the Southern Unami form; Southern Unami has undergone an innova tion whereby animate nouns in -V.w drop the -w and shorten the vowel unless a suffix follows. Munsee has wí.xče.w 'wolf'.
12.
Cf. (tachanall) (3x) and (metachenall), (metachanall) (Luckenbach 1838:43, 247, 248). Munsee has xws 'piece of firewood'.
13.
In Delaware compound verb stems the hyphen marks the phonological word boundary after a preverb and indexes the component parts if they are discontinuous. Forms of
86
Ives Goddard English he are the conventional glosses for the Algonquian third-person animate singular and third-person animates undifferentiated for number; equivalent English translations might be any of the forms of he, she, it, or they, depending on context. Forms of English it conventionally gloss Algonquian inanimates.
14.
Glossed in the source: "to knock somebody on the head"; the stem can be glossed 'kill, strike dead, kill by a blow to the head with a blunt instrument'. The ɔ. in the last syllable of the Southern Unami example is an analogical innovation.
15.
Animate nouns are always obviative when possessed by a third-person animate.
16.
Compare the uses of the emphatic pronouns in Fox (Dahlstrom 1988:166-177).
17.
With a corresponding noun phrase overtly present in the clause, the third-person pro nominal inflection on nouns and verbs functions as agreement.
18.
The particles are tá 'how? where?' (interrogative deictic) and háč (interrogative; dialectally héč).
19.
The most commonly used negative particle in the independent order is tá kó. (akó-, kó-); máta is used with subordinative and conjunct verbs and occasionally with the indepen dent indicative, notably in fixed expressions; káči is the prohibitive 'don't'; there are other specialized and idiomatic negative expressions.
20.
In the present paper it is not possible to comment on all the philological details of even the cited Pidgin forms, and many problems in the interpretation of the attested Pidgin materials are not discussed.
21.
Phonetically Swedish has short vowels before long consonants and long vowels before short consonants in accented syllables.
22.
Generally the acute accent is used; the accented form of (w) is (cb). (aé) is the accented form of (æ) (Table 2: 'four', 'nine', and notes and g); probably this was a printer's substitution. The use of what is printed as a Greek omega (ω, ῶ) (rarely (w), (W), or (vv)) for a vowel corresponding to Unami ɔ. (or sometimes u) was probably an adaptation of the medieval and early modern Swedish use of (w), in the value (uu), for long (u) (e.g. Old Swedish (tw) for (tu) 'thou' [Haugen 1976:304]; Amandus Johnson, in Lindeström 1925:xli); (ω,ῶ) generally contrasts with sometimels ), the usual writings of the vowel corresponding to o. In the examples in this paper, accented spellings of words in Campanius have been cited in preference to unaccented variants.
23.
The spelling (ckh) can also represent kh (167, 122).
24.
The phone [c] was not found in Swedish; [š] only occurred prevocalically, as a pronuncia tion of (sj) (then written (si)). The pronunciation of Swedish (sj) as [š] is supposed to have been already established by the seventeenth century, but the pronunciations of (sj) in modem Swedish cover a considerable range and Campanius's inconsistency in spelling Unami š and his hesitancy to use (si) suggest that there were phonetic differences between the Swedish and Delaware sounds that caused him to fail to identify Unami š with his pronunciation of Swedish (si) in most cases. (I am grateful to Börje Westlund and Einar Haugen for help with the Swedish facts.)
25.
Cf. tahkč-u 'he is cold'.
Pidgin Delaware
87
26.
SU nní.č.a.n 'my child' (now especially 'my daughter'; cf. Tenychan'child' [133]) + non-productive diminutive l-əsl (note 140; cf. 58, 160); such formal diminutives are common as dialect variants in Algonquian.
27.
The latter part of Campanius's form has apparently been influenced by (6) lamaéæs 'fish' (247); II (seh) must be a copying error for *(sch).
28.
I take the variant (4) (skameera) 'run' as an error, perhaps merely a misprint, and not an indication that the Pidgin Delaware words with (sch-) for Unami k(w)š- were pronounced with [sk] (cf. 262).
29.
Pidgin (chirῶkha) points to a dialectal form *či.ló.k.ay, without the non-productive diminutive l-əsl of the attested Unami form {-ay is regularly rendered or reflected in Pidgin by (-a); cf. 41, 162, 225); the reverse relationship is more common (see 34 and note 26).
30.
Campanius (1696:30, 102) uses this word for 'hand' and 'hands', and lists the two variants as synonyms of (nach) (cf. SU nnáxk 'my hand').
31.
Given with (kaak), but this is elsewhere 'goose'.
32.
The Moravians have NU amimi (Luckenbach 1838:28), (amémi) (Zeisberger 1887: 143), but the de Laet vocabulary suggests that there was a Northern Unami dialect (perhaps Unalachtigo) in which the form of this word matched Munsee.
33.
The consistent (j) for x in this word is difficult to explain; conceivably it reflects a Swedish pronunciation of Dutch (g) [y].
34.
The gloss is "swart eller blööt steen", but perhaps swart is a misprint for swagt 'weak, fragile', influenced by the gloss in the preceding entry, gråsteen 'granite, lit. grey stone'.
35.
There is also a variant (147) (waépæck), (wepeck), that probably has its first vowel by contamination with (145) (neskæck) 'black' (cf. Mu né.ske.k '(irían.) which is black') and (142) (mæckhhchæck) 'red' (Mu mé.. '(inan.) which is red'). These Munsee forms are participles and as such have initial change, the ablauting of initialsyllable lal or ləl to le.1; the corresponding Unami participles must also have earlier had initial change, but in twentieth-century Southern Unami the change has been leveled out in stems of certain shapes, including these, and the forms are SU sšk.. and máxke.k.
36.
The addition of h- before initial vowels was a characteristic of the speech of some Unami speakers in the twentieth century; there were also a few cases of dialectal disagreement on the presence of h- in the speech of those that consistently distinguished h-initial and vowel-initial words: SU halpánkəl, alpánkəl 'barrel' (a loan from Dutch half anker [Goddard 1974:154]).
37.
The Pidgin forms with (-s) reflect dialectal variants with the non-productive diminutive -(ə)s (cf. note 26).
38.
Utterance-final devoicing in Unami probably accounts for the occasional appearance of final stops in a few dialectal Pidgin variants where Unami has nasals (see also ex. 153).
39.
The longer form contains the non-productive diminutive -(ə)s (231; cf. 34, 71, and notes 26 and 37).
Ives Goddard
88 40.
Cf. SU nəwi.š.á.s.i
41.
Campanius uses this word for game animals (C [5-6] (mosiuttamen MWS) 'hunt', lit. 'bring down animals' (see n. 107); [52] mochijrick mws 'elk', lit. 'large animal') and domesticated animals (C [19] (móchijrick mws), here used for 'ox or ass'). There is good evidence that the Delaware word means 'elk' (genus Cervus; e.g. Zeisberger's gloss "Eur[opäischer] Hirsch") and not 'moose' {Alces); in Delaware, as in North American English and in the Swedish of New Sweden (Campanius 1696:52), the inherited word for 'moose' {Alces) was applied to 'elk' {Cervus).
42.
An English writer's knowledge of the vocalic (w) of Welsh might account for (66-68), but it seems unlikely that such a writer would have used (w) only in three words that, after all, hardly present problems for English spelling.
43.
In huska hayly, Chetena koon "very great hard frosts," lit. 'very much strong snow/ice'.
44.
In SU xahé-li the initial xah- is a renewal of under-characterized x- 'big, etc.'; cf. SU xé-lo-k 'there are many of them', beside newer xahé-lo-k.
45.
I assume that the two occurrences of II(Me) T' are copyist's errors: II(Me Matta wingenun) T do not care for it', (Me mauholume) 'I will buy it' (cf. II (Nee mauholum) 'I will buy', (Ne Maugholame) 'I bought it'). The entries withMeT are three lines apart and in each case precede a word beginning withm orMand follow an entry beginning with(M). The same error is found in II (Munockon) 'm[il]k or a woman's [breast]' (the entry after (Papouse) 'a sucking child') beside II (Nonackon) 'milk'; cf. SU no.ná..an 'milk; breast'. Prince (1912:515) and Thomason (1980:175, 189) concluded that Pidgin Delaware had (Me) T , taken from English, but two examples written by a copyist who evidently knew Pidgin poorly, if at all, are insufficient to support a conclusion so inconsistent with the otherwise complete lack of English words in Pidgin.
46.
Also (C) (chijk); used in three instances (7, 7, 10) for second singular subjects (with Swedish skal, skalt 'shall' in the corresponding translations) and uncertain in the other occurrence (13) (the Swedish translation has 'we').
47.
These longer forms survived in modern Delaware only as dialectal Munsee sandhi variants used before enclitics (marked with preceding =): Mu ní.l=ha né.ka ó.əh1 'I'm his grandmother' (I.EMPH=PRED he.EMPH his.grandmother). There is every likelihood, however, that they existed in at least some seventeenth-century Unami dialects.
48.
Swedish back-translation alt thet wij förstå, them wara till godt behag 'everything that we understand to be pleasing to them'. Campanius (1696) includes the Swedish text of the catechism and other materials (referred to here as the Swedish model), his Pidgin Delaware rendering, and a more or less literal translation back into Swedish of the Pidgin Delaware rendition (referred to here as the Swedish back-translation).
49.
Swedish back-translation Gud ... be skiärmer och bewarar mig 'God ... protects and preserves me'. hícka'away (?)', attested in Pidgin Delaware only in the occurrences of this phrasal verb, is probably SU (ka 'over there (indefinite)'.
50.
Swedish back-translation tu talar.
51.
Thomas: "What hast got in thy house?"
'I am afraid'.
Pidgin Delaware
89
52.
Swedish model: han gifwer diuren sina födo 'he gives the animals their food'. The (vv) in the last two words (and wherever it appears in Campanius) is a printer's substitution for (w) or ().
53.
The preceding context is: (jῶni hωrítt chíntika Manétto, Kintz póho thaan suhwijvan anckaroppan renáppi, nijr robe: ...) 'The Holy Spirit will right away call to all dead people and say to us: ...' (3pers. good prayer/holy spirit, now call to all dead person, lpers. say).
54.
This may be the only example in Campanius of a form in (-r) used for a possessor.
55.
Swedish model Wij ... skole ursaka honom 'we should forgive (excuse, pardon) him'; back-translation wij skolom säija: then Menniskian är intet så ond 'we should say, this person is not so bad'.
56.
Swedish back-translation the intet älska mig.
57.
Swedish back-translation: thet namnet 'that name', referring back to Gudz Namn 'God's name'.
58.
Swedish back-translations: mitt öga, min ögon.
59.
Swedish back-translations: (8) wåra herrar och förmän; (6) wår stora HErre.
60.
Swedish back-translations: (7) tin Fader och tina Moder, (21) tin Fader och Moder.
61.
Swedish back-translation: edre barn.
62.
Swedish back-translation: I Fäder 'you parents'.
63.
Swedish back-translation: I Barn.
64.
In "The Indian Interpreter" and in Thomas all the second-person pronouns in the English translations are singulars {thou, thee, thy); very likely both authors were Quakers.
65.
Swedish: Thet är ingen som wil kiöpa.
66.
Swedish gloss: Slå någon.
67.
(Kec-loe) is miscopied for *(Kee loe). The translation in the original is: "Say thou what thou hast a mind to."
68.
Swedish back-translation: på them som mig älska.
69.
Swedish back-translation: så at wi wår nästa hans penningar/ eller hwad mehra han äger/ icke borttaga; the word order in Campanius's Pidgin is influenced by Swedish, but this does not account for the omission of the subject pronoun beforehátte'have'.
70.
Swedish gloss: Gode wän/hwad har tu mehr för wahrur.
71.
(match), (matcha) (120), (137, 153) (mætzi) 'now, already'; cf. SU mé.či.
72.
For (pauluppa) cf. SU pal.ae 'buck', now usually 'dollar' (beside xinkɔlp.ae 'buck'); for (shuta) see (39, 199, note 107).
73.
Campanius's situational translation is: T have no time to sit'.
74.
The clause-initial position of Pidgin (mátta) 'not' matches that of modern SU tá kó., and the pre-verbal position is that of modern SU máta (27, 28). The pidgin may accurately
90
Ives Goddard attest an earlier stage of Unami, since the tá of SU tá kó. (also used as a negative in other particle phrases) is evidently historically a reduction of mata .
75.
Swedish model and back-translation: Tu skalt icke stiäla.
76.
Swedish back-translation to p. 37: liggen icke nu längre och sofoven; Swedish gloss to p. 142: Soof intet mehr nu.
77.
Thomason (1980:177-78); other orders in Campanius's translation appear to reflect the word order of the Swedish model.
78.
Swedish: wij skole myckit hålla af them; the appearance of (ock) 'and' in second position mimics the location the synonymous Swedish (och) often has when used as a clause connective.
79.
Swedish model (given also as the back-translation): tin nästas hus 'your neighbor's house'.
80.
Swedish model: tins nästas hustru 'your neighbor's wife'; back-translation: ens annars mans hustru 'another man's wife'.
81.
Swedish back-translation: then stora HErrans Gudz Namn.
82.
Swedish back-translation: himmelens och jordenes gode Gud.
83.
'Elbow' is animate and so takes the obviative ending l-al in the third-person-possessed form (see 15); since, however, the indefinite-possessor form has no third-person pos sessor, it does not take the obviative ending. SU wí-skɔn 'an elbow' is apparently unrecorded, but it is constructed on the model of other animate dependent nouns with indefinite possessor, like wté. 'a heart' (cf. wté.ha 'his heart').
84.
Campanius differentiates (nach) 'hand' (note 30) and (onáck) 'forearm'; these are both from the same Delaware word, though conceivably from different dialects.
85.
For dialectal Pidgin (t) corresponding to Unami n, see note 38.
86.
Campanius also uses (27) (chítto) and (54) (chítta); the position of the accent and the variants of the second vowel are unexplained but presumably secondary.
87.
Also used with extended meaning: (21) híssemis 'generation'.
88.
This is the word Campanius uses for 'mother' throughout his translation, but in his vocabulary he also gives (139) (kahæss); cf. SU nkáhe.s 'my mother', NU (Gáhowees) (Zeisberger 1887:126).
89.
The interpretation of II (ponacka) "two hands" and II (olocko) "holes" (81) is uncertain.
90.
The locative of unattested Unami *sí.p.o.s, equivalent to SU sí.p.u 'river, stream' + non productive diminutive l-əsl (cf. note 26).
91.
The initial (t-) in the Pidgin forms shows that these were taken from the Delaware possessed stem; the pronominal prefixes insert a following ltl before underlying (or historical) vowel-initial stems that are not dependent nouns (contrast 16). The use of the possessed forms of 'town' to mean '(one's) territory' was presumably earlier found in Unami, but it is attested in twentieth-century materials only for the Munsee cognate: Mu nto.té-noy 'my allotment (on the reservation)'.
Pidgin Delaware
91
92.
Campanius's words for 'winter', 'spring', and 'morning' have an ending (-or) that may reflect an iterative ('whenever it is...') unattested in later varieties of Delaware; since other Algonquian languages point to an iterative ending PEA *-i.r, however, the vowel (o) would not be expected in Delaware.
93.
Almost always pé. in modern Southern Unami (Goddard 1979:viii, xx).
94.
II (taulle) could be from SU ktli 'you (sg.) say to me'.
95.
Cf. SU mpá 'I come', kpá 'you come', nkáí-a-pá 'I am going to come', pá, pá-l 'come (imperative sg.)'.
96.
(151) (chékco rcoænse) 'what is (that) called' (what be.named; Swedish Hwad heter thet); II (Kecko Looense) 'what is thy name[?]' (what be.named), (Kecko ke lunse une) 'What dost thou call this [?]' (what 2pers. name 3pers.).
97.
For the variation between (Kitton) and II (kitthanning) 'river' compare (161-163); cf. NU (kittan) [khíthan] 'great river' (Brinton & Anthony 1889:54), NU (kidhannünk) 'in the main R[iver]' (Zeisberger 1887:160).
98.
This form would be used in sentences like: náni ntáp.i.n 'I stayed there; that is where I stayed' (Goddard 1983:356-57). Campanius has a second Pidgin variant from this stem: (20) (háppi) 'stay'; this has the accent and stem shape of the non-third-person forms (SU ntáp.i 'I sit, stay'; cf. SU ahpú 'he sits, stays') but lacks any trace of the prefix or intercalated ltl (cf. 173; contrast 171, 172; for the added (h-, cf. 48, 55).
99.
Also 'bed'; cf. SU kéhkawink 'bed', lit. 'where one sleeps'.
100.
The gloss is omitted, but the entry is followed by the word for 'tobacco' (41).
101.
Campanius: 'to sit down and rest'.
102.
(150) (taan kῶmæn) 'where are you coming from?', (37) (Kῶmen taan Hácking!) 'come out of the earth (imperative)' (come.from PREP earth); cf. Lindeström (1925:203) (tanckomen kire) 'whence comest thou?'.
103.
From the same Unami stem there is also: (4) (taan) 'go', II (toon) (81); cf. SU ntá.n T go (there)'.
104.
This Unami Class lb stem (in -əə) has apparently assimilated to the Class la type (with -amən) in Pidgin. Indefinite-subject transitive inanimates like the second form given here are rarely used in modern Southern Unami but were very likely more common earlier.
105.
By convention, in the glosses of transitive absolute forms the object is parenthesized.
106.
Cf. SU maní . to . 'he makes (it)', backformed from, e.g., əní.to.n T make it' (nn ( *ln); some dialects of Northern Unami also had this assimilation: NU (manitol) 'make it (imperative)' (Luckenbach 1838:27).
107.
(5-6) (mosiuttamen MWS) 'hunt, lit. bring down animals', (57) (mosiuttamen) 'make fall', (105) (mosiúttamen) 'have power over'.
108.
(151) (kékω kommijre) 'what will you give in return?' ; (54) (matta ock nijr hátte chéko kommijre) 'and we do not have anything to give (you) in return' (renders: 'nor have we deserved it'; for (ock) see note 78); (152) (chéko nijr mijre) 'what shall I give you in return?'; (150) (mijrikon) 'trade', (151) 'we want to trade'; (151) (kotz mátta miraana)
Ives Goddard
92
'why don't you give (that) away'; (152) (chææk skomerijne) 'I'll give you money'; II (Nee Meele) "I will give thee," (B.C SickeMelee) "B.C. will give me so much for it," (Maleema cacko) "give me something". II (Si-) may reflect *=či, a variant of =č 'future' (cf. note 125). 109.
For the future imperative see Goddard (1979:58, 145, 191).
110.
(50) (chísbo ncDttaas) 'full belly', (101) (Singa chir chísbo hátte Kavvín Kattúnga) 'when you have fully slept' (when 2pers. full have lie sleep).
111.
(Manúnckus) 'bad-tempered, bad' (Swedish ond); cf. SU manúnksu 'he is angry'; for the other words see (174) and (58).
112.
(149) (Mamantíckan) 'peach, plum', (143) (Mamantickan) 'game', II (mamantuikan) ma.mantí.an; (and variants) "a peach or cherry"; cf. SU ma.mantí.k.anak 'dice' (sg. Indian dice were made from plum pits). (144) (táckhan) (and variants) 'wood, tree'; cf. SU tá.x.an 'a piece of firewood' (8).
113.
(140) (hââs) cf. (69). (140) (táckquin) 'wear, clothing'; cf. SU nták.. 'I wear it'.
114.
(142) (hóckung) 'up, heaven', II (hockung) 'up, up above', (hockcung) "a chamber" (i.e. an upper room, a sleeping loft); cf. NU (hokunk) 'above, above the earth' (Brinton & Anthony 1889:49), SU hukwé.yunk 'up, up above' (with added syllable). II (tappin): see (178).
115.
(1) (hátte), (13) (hattéu), II (hatta), T (hatah) (79, 86); cf. SU hát . e . 'it is (there)'.
116.
(23) (nijr Renáppi, móchæærick hwisásse hatte) 'we people are very much afraid' (lpers. person very afraid/fear have), (móchæærick) 'very, very much' modifies the phrasal verb; it is not the same word as (móchijrick) 'big' (174), as in the example in note 118, but reflects a Unami participle *me.x.é.lhik from *(m)xé.lət 'there's a lot of it', NU (mechelit) "much," (chelit) "a great deal" (Brinton & Anthony 1889:27, 77; other Pidgin sources [see 72] reflect the corresponding particle).
117.
(3) (chijr matta maranijto manúnckus, chirῶna Saccheemans, horít Manéttos Rwaénse) 'you are not to abuse God's name' (2pers. not make/do bad 2pers./poss. chief-' s good spirit's name).
118.
(21, 141) (sevaraénda) 'sadness, sad' (Swedish bedröfwelse, bedröfwat), (141) (mochijrick sevarænda hatte) 'be greatly troubled' (Swedish wara myckit bekymrad); cf. SU ši.ye.ləntam 'he is sorry', NU (schiwelentam) 'he is sorry, grieved' (Zeisberger 1887:179).
119.
Cf. (54, 151) (rankúnti), used in expressions for giving to mean 'freely and in a friendly fashion (but with expectation of a return gift)'; (8) (ranckúnti) does not form a phrasal verb with the preceding (maranijto), whose object is the following clause, but is used to mean 'according to their will' (Swedish til wiljes). A Delaware tradition about the first coming of the Europeans contains the expression SU kəmanni . tó .nen=č ktəli-=č -ɔ-wəlankuntí.ne.n 'we and you will make it so that we will have a lasting peace' ; the first verb contains the stem -manni.t- (from *-malni.t-), corresponding to Pidgin (marenit), and the second verb has wəlankunti- 'be at peace with each other', corresponding to Pidgin (rancontyn). This súggests that Unami expressions may have been the model for some phrasal-verb constructions in Pidgin Delaware.
Pidgin Delaware
93
120.
(135) (manétto) 'God, spirit, angel', Lindeström (1925:203) (manitto) 'the evil one', Danckaerts (1913:76) (maneto) 'the Devil', II (manitto) 'the Devil' (in a second hand: "God"); cf. SU 'god, manitou'.
121.
(140) (sacchéman) 'lord, superior, ruler' (shape influenced by Swedish man 'man'), (Dutch rendering, 1640s) (sackema) (NYCD 1:180); cf. SU sa.k. 'chief'.
122.
(1) (mamaræckhíckan), (143) (malæckhíckan), (143) (aamaræckhickan) 'book, letter, writing, paper', (mamolehickon) 'book, paper'; cf. NU (mamalekhikan) [ma . ma . le . khí . kan] 'writing, letter, book' (Brinton & Anthony 1889:72).
123.
(cake) 'wampum' reflects SU ké.kw 'wampum (collective)'. Twentieth-century Dela wares gave the unit of measure for strings of wampum (SU -a . i . k . an; cf. NU (-awican), (-aikan)) as a yard (three feet), measured from the nose to the fingertips of the laterally extended hand; this seems to agree with the etymological meaning 'spine length' (cf. SU na.í-k.an 'my spine'). Earlier sources from both New England and the Middle Atlantic always refer to measuring wampum strings by fathoms (six feet); possibly the difference has something to do with the fact that a fathom of white wampum was worth the same as a yard of black wampum (Williams 1936:155).
124.
Other examples: (57) (chéko paéwo taan jῶni) 'who comes to him'; (57) (nirῶna chíntika taan chijr) 'our prayer to you'. Campanius uses an apparent variant (4, 25, 43, 135) (taani), (7) (thaani) for 'on, in'.
125.
(49) (thaan nircbna chíckhæ tarássi) 'until the end of our life' (to/until lpers./poss. life end); (94) (singa Boquíckan mítzi tarássi) 'after supper' (when night eat end); (57) (hωrísso Tharássi) 'a peaceful (?) death'. The syllable (-ssi) may reflect a dialect variant *=či of the future enclitic =č (see also 246).
126.
Cf. Mu á.nay 'road, path'; NU (anenk) 'on the road' (Luckenbach 1838:64).
127.
Pidgin (-ng) (or (-ngh)) reflects Unami -nk [-ŋg] but was presumably pronounced [-ŋ] by Germanic speakers.
128.
Most probably II (kishquecon) also means 'day', and the copyist has either skipped to the gloss on the line below or omitted the word for 'eight' that stood before it.
129.
(13) (chéko Renáppi hátte máramen) 'what someone has to trade' (what person have trade), (32) (chijr, hátte máramæn nijr) 'you have bought (= redeemed) us', (151) (maramæn) T want to buy from you', (151) (maramæn) T want to sell', Lindeström (1925:203) (kecko hatte marrama) 'what hast thou to sell?', II (Nee Mauholum) "I will buy," (Ne maugholame) "I bought it," (Kee-mauholume) "wilt thou buy."
130.
Cf. (137) (suckra mochijrick pævo) 'there is a lingering rain' (rain big come); Swedish thet kommer longsamt rägn. (suckolan) 'rain' (used attributively); cf. SU só.k.əla.n 'rain; it is raining'.
131.
(15) (singa Renáppi Rῶbénse simáckot) 'when a person's name is mentioned' (or the like); cf. (1) (chísbo Simoáckan) 'commandment' (full talk/statement), (63) (simockan) 'command', (152)(Pijri Simockan) 'talk, conversation' (other [= various] talk/ statement).
132.
E.g. (1) (renáckát), (6) (renáckot), 'thus, so'; cf. SU li.ná.k.ɔt 'it appears (thus)'.
Ives Goddard
94 133.
The Pidgin Delaware word points to the existence of dialectal Unami *étke 'perhaps', a direct reflex of PA *ye.t(w)eke (reflected by Fox ye.toke, Ottawa i-tik, Ojibwa i.tok); all these Algonquian words are sentence enclitics (occurring after the first word of a clause).
134. With (153) (hiῶ) compare: (36) (hiῶ qui síxi) 'shall soon' (?), II (lough undoque) "goe yonder" (cf. 73), (Hyoah netap) "thou good friend, or good be to thee friend." 135.
Campanius (1696:53, 54, 128) uses this also for 'sin, sinner'.
136.
Cf. SU wans-i 'thank you', Mu anši.k.
137.
(27) (Keene, Keene ítta) 'thanks', (153) (itta ikale) 'go before' (?; (ikale) is SU íka lí '(to) over there'); Lindeström (1925:203) (ita kire) 'God's peace, welcome, how are you?', (kicksj ita) 'that's so' ((kicksj) = (kítzi) 'true', from SU khičí.i).
138.
(2) (móchij) 'yes', (152) (mochij apitzi) 'you will get it right away' (for (apitzi) see 62), (153) (mochij matzi) 'go in peace' (for (matzi) see 241).
139.
Recalled by one speaker as having been used by one man and his immediate family; also attested in the vocabulary recorded from the scout Black Beaver in 1853 (Whipple et al. 1855:58). SU lamé.s must have been a back-formation from the diminutive namé.t.ət 'little fish' in a dialect in which the shift of / ([1] or [r]) to n was an active process of diminutive consonant symbolism. Modern Unami has fossilized traces of this consonant symbolism (hló .s.ds 'old man', hinó.t.ət 'wren', as if 'little old man'; the same metaphor in Shawnee pasito 'old man', pasito.θ 'wren'); a Pidgin example is II (peray) "a Lad" (cf. SU pí.lae 'youth' [Speck 1931:170]), diminutive (penaetit) "a boy" (cf. SU pinaé.tət, the name of Col. Jackson, the last traditional chief of the Turkey phratry; 'little boy' would be a typical nickname, and nicknames often show shortening of vowels and consonants, as here with the i and t).
140.
An exactly parallel innovation replaces a dialect word for 'child' having the -s suffix with a form without this suffix that matches the modern word (34, note 26).
141.
The de Laet vocabulary shows only r (de Laet 1633:75-76), while in the second half of the eighteenth century recordings of Unami by the Moravians and Denny (1860) have only /.
142.
Swedish gode wän (153); used by Campanius for 'neighbor' in the Biblical sense, i.e. 'fellow man'.
143.
Although the Swedish sources show a Delaware shape for this word, the others, both earlier and later, have the two-syllable shape of the term of greeting and address used between Indians and Europeans from southern New England to Virginia in the Colonial period (Goddard 1977:39); cf. (nétop) 'friend', "the general salutation of all English toward them," from Rhode Island (Williams 1936:2), and (netab) 'a friend', "the principall word of kindness," (netap) 'my dear friend', from Virginia (Strachey 1953: 184, 194).
144.
Cf. (139) (ãquæo) 'woman' (cf. SUxkwé . (w)), diminutive (1) (ãquættæt) 'girl'; "The Indian Interpreter" has replaced the first of these while retaining (aquittit) "a little girl," but Thomas has (squatid) 'girl', an altered form of the diminutive that conforms to the new word for 'woman'.
145.
Peter Kalm recorded (saltenbras) and (salssenfras) in use among the Swedes on the lower Delaware River in the eighteenth century (Friederici 1960:561).
Pidgin Delaware
95
146. It is significant that this first mention of Hoboken, in 1630, refers to it explicitly as "the land by us called Hobocan Hackingh"; the later spelling Hoboken conforms to that of a town in Belgium. The name cannot be from Munsee, the local variety of Delaware, which has Mu matásən 'pipe'. 147. Also (151) (pomúttamen) 'shoot'; cf. SU mpəó.t. 'I shot it (with an arrow)'; only animate objects are attested with these words in Pidgin. 148. Ojibwa has ni.nim 'my sister-in-law (man speaking); my brother-in-law (woman speak ing)' (exactly matching SU ní.hm [157]) and its diminutive ni.nimošše.n? 'my crosscousin of opposite sex; my lover, sweetheart'. 149. Some information on this move that actually refers to events of 1626 is dated to 1628 by Weslager & Dunlap (1961:75). 150. The use of American Indian Pidgin English in some of Denny's glosses is another example of the side-by-side existence of Pidgin English and Pidgin Algonquian on the East Coast (Goddard 1977:41). A further illustration of this situation is the phrase (Tatapa — you; all one this) used by a judge in Connecticut in 1704; the judge spoke these words while patting his head in order to convey the idea 'head' to the Indian defendant (Goddard 1977:37-38). The first phrase is probably Pidgin Massachusett (Massachusett (tatuppe) Tike', (yeu) 'this') and the second is the equivalent in Pidgin English. 151. Lindeström's (1925:203-204) mention of Delaware warpath language provides no evi dence for Iroquoian knowledge of Pidgin Delaware, as Thomason (1980:185-86) argues, since all evidence and direct testimony agree that a knowledge of Pidgin Delaware did not confer the ability to understand real Delaware (Michaëlius, in Jameson 1909:128).
References Anonymous. 1684. "The Indian Interpreter". Trenton: Salem Town Records book B.6468, in the State Archives of New Jersey, MS. Barber, John W., and Henry Howe. 1844. Historical collections of the State of New Jersey. Newark: Justus H. Bradley. Brinton, Daniel G., and Albert Seqaqkind Anthony, eds. 1889. A Lenâpé-English dictio nary (= The Pennsylvania Students' Series, 1). Philadelphia: The Historical Society of Pennsylvania. Campanius, Johannes. 1696. Lutheri Catechismus Öfwersatt på American-Virginiske Språket. Stockholm: Burchardi. (Facsimile edition issued with the title: Martin Luther's little catechism translated into Algonquian Indian, bound with Collijn 1937, Stockholm: Ivar Hæggström and Uppsala: Almqvist & Wiksell, 1937.) Campanius, Johannes. 1938. Luther's Catechism translated into the American-Virginian language. Translated by Daniel Nystrom and E.W. Olson. New York: Swedish American Tercentenary Association. Collijn, Isak. 1937. The Swedish-Indian Catechism: Some notes. Issued with the facsimile reprint of Campanius (1696). Uppsala: Almqvist & Wiksell.
96
Ives Goddard
Dahlstrom, Amy. 1988. "Independent pronouns in Fox". In honor of Mary Haas, ed. by William Shipley, 165-94. Berlin: Mouton de Gruyter. Danckaerts, Jasper. 1913. Journal of Jasper Danckaerts 1679-1680, ed. by Bartlett Burleigh James and J. Franklin Jameson. New York: Charles Scribner's Sons. (Re printed: New York: Barnes & Noble, 1959.) de Laet, Johannes. 1633. Novvs Orbis: Seu descriptionis indiæ occidentalis libri XVIII. Leiden: Elzeviers. Denny, Ebenezer. 1860 [1785]. "A military journal kept by Major Ebenezer Denny". Memoirs of the Historical Society of Pennsylvania 7.478-81. Philadelphia: The Historical Society of Pennsylvania. Friederici, Georg. 1960. Amerikanistisches Wörterbuch und Hilfswörterbuch für den Amerikanisten. (= Universität Hamburg, Abhandlungen aus dem Gebiet der Auslands kunde, 53; Reihe B. Völkerkunde, Kulturgeschichte und Sprachen, 29.) Second edi tion. Hamburg: Cram, de Gruyter & Co. Goddard, Ives. 1971. "The ethnohistorical implications of early Delaware linguistic materials". Man in the Northeast 1.14-26. Goddard, Ives. 1974. "Dutch loanwords in Delaware". A Delaware Indian Symposium, ed. by Herbert Kraft, 153-60. (= Pennsylvania Historical and Museum Commission Anthropological Series 4, general editor Barry Kent.) Harrisburg: Commonwealth of Pennsylvania, The Pennsylvania Historical and Museum Commission. Goddard, Ives. 1977. "Some early examples of American Indian Pidgin English from New England". UAL 43:37-41. Goddard, Ives. 1978a. "Eastern Algonquian languages". Northeast, ed. by Bruce G. Trigger, 70-77. (= Handbook of North American Indians 15, general editor William C. Sturtevant.) Washington, DC: Smithsonian Institution. Goddard, Ives. 1978b. "Delaware". Northeast, ed. by Bruce G. Trigger, 213-39. (= Handbook of North American Indians 15, general editor William C. Sturtevant.) Washington, DC: Smithsonian Institution. Goddard, Ives. 1979. Delaware verbal morphology: A descriptive and comparative study. New York: Garland. Goddard, Ives. 1983. "The Eastern Algonquian subordinative mode and the importance of morphology". UAL 49.351-87. Goddard, Ives. 1995. "The Delaware Jargon". New Sweden in America, ed. by Carol E. Hoffecker, Richard Waldron, Lorraine E. Williams, and Barbara E. Benson, 137-49. Newark: University of Delaware Press. Goddard, Ives, and Kathleen Bragdon. 1988. Native writings in Massachusett. 2 vols. (= Memoirs of the American Philosophical Society, 185.) Philadelphia: The American Philosophical Society. Haugen, Einar. 1976. The Scandinavian languages: An introduction to their history. Cambridge, MA: Harvard University Press. Holmer, Nils. 1946. John Campanius' Lutheran Catechism in the Delaware language. (= Essays and Studies on American Language and Literature, 3.) Uppsala: The American Institute in the University of Uppsala. Jameson, J. Franklin, ed. 1909. Narratives of New Netherland 1609-1664. New York: Charles Scribner's Sons. (Reprinted: New York: Barnes & Noble, 1967.)
Pidgin Delaware
97
Laurent, Joseph. 1884. New Familiar Abenakis and English Dialogues. Quebec: Leger Brousseau. Lindeström, Peter. 1925 [1691]. Geographia Americae: With an account of the Delaware Indians. Translated by Amandus Johnson. Philadelphia: The Swedish Colonial Soci ety. Luckenbach, Abraham, trans. 1838. Forty-six select scripture narratives from the Old Testament.... Translated into Delaware Indian. New York: Daniel Fanshaw. Myers, Albert Cook. 1912. Narratives of early Pennsylvania, West New Jersey, and Delaware 1630-1707. New York: Charles Scribner's Sons. Myers, Albert Cook, ed. 1937. William Penn his own account of the Lenni Lenape or Delaware Indians, 683. Moylan, PA: Albert Cook Meyers. Nelson, William. 1894. The Indians of New Jersey. Paterson, New Jersey: The Press. NYCD = Edmund B. O'Callaghan, ed. 1853-1887. Documents relative to the colonial history of the state of New York, 15 vols. Albany: Weed, Parsons. New York Historical Collections. 1913. New York: New-York Historical Society. Prince, J. Dyneley. 1912. "An ancient New Jersey Indian Jargon". American Anthropolo gist, n.s. 14.508-24. Speck, Frank G. 1931. A study of the Delaware Indian Big House Ceremony. (= Publica tions of the Pennsylvania Historical Commission, 2.) Harrisburg: The Pennsylvania Historical Commission. Strachey, William. 1953 [1612]. The historie of traveil into Virginia Britania, ed. by Louis B. Wright and Virginia Freund. (= Works Issued by the Hakluyt Society, second series, 103.) London: The Hakluyt Society. Thomas, Gabriel. 1698. An historical and geographical account of the Province and Country of Pensylvania; and of West-New-Jersey in America London: A. Baldwin. (Facsimile reprint: New York: for Henry Austin Brady by Francis Michelin, 1848.) Thomason, Sarah Grey. 1980. 'On interpreting "The Indian Interpreter"'. Language in Society 9.167-93. Thomason, Sarah Grey, and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Tooker, William Wallace. 1911. The Indian place-names on Long Island and islands adjacent, with their probable significations, ed. by Alexander F. Chamberlain. New York: G.P. Putnam's Sons. van den Bogaert, Harmen Meyndertsz. 1988. A journey into Mohawk and Oneida country, 1634-1635, transl, and ed. by Charles T. Gehring and William A. Starna, with wordlist and linguistic notes by Gunther Michelson. Syracuse: Syracuse University Press. Weslager, CA., in collaboration with A.R. Dunlap. 1961. Dutch explorers, traders and settlers in the Delaware Valley, 1609-1664. Philadelphia: University of Pennsylvania Press. Whipple, Amiel Weeks, Thomas Ewbank, and William Wadden Turner. 1855. Explora tions and surveys for a railroad route from the Mississippi River to the Pacific Ocean: Report upon the Indian tribes. Washington, DC: War Department. Williams, Roger. 1936. A key into the language of America [1643]. Fifth edition. Providence: The Rhode Island and Providence Plantations Tercentenary Committee, Inc.
98
Ives Goddard
Zeisberger, David, trans. 1821. The history of Our Lord and Saviour Jesus Christ.... Translated into the Delaware Indian language. New York: Daniel Fanshaw. Zeisberger, David. 1887. Zeisberger's Indian dictionary. Cambridge: John Wilson and Son, University Press.
Ndyuka-Trio Pidgin George L. Huttar University of Texas at Arlington Summer Institute of Linguistics and Frank J. Velantie Suriname Ministry of Education
1.
Introduction
The name we have given in our title to the language under discussion here is intended to summarize the social and linguistic circumstances of its origin: it arose for purposes of general communication, especially for trade, when speakers of Ndyuka and speakers of Trio came in contact, and reflects lexical and structural features from the languages of both groups. The Ndyuka society and its language were apparently formed in the early eighteenth century when large numbers of slaves escaped from plantations of the Suriname coastal area and made their way to east-central Suriname. Later two other "Bushnegro" societies, as they are called, formed in eastern Suriname: the Paramaccans and the Alukus (also called "Bonis"). These three groups are politically distinct; they refer to themselves, as well as to their speech varieties, by three distinct names. Nevertheless, there is a high degree of linguistic similarity among the three varieties, a similarity which on the one hand unites them linguistically and on the other hand sets them apart from the Bushnegro varieties of central Suriname (Saramaccan and Matawai, and
100
George L. Huttar and Frank J.
Velantie
Approximate current residential areas of Trios, Wayanas, and Bushnegro groups of Suriname
Ndyuka-Trio Pidgin
101
Kwinti)1 and from Sranan. All of these varieties are creoles. Besides the distinctions just mentioned, Ndyukas recognize a number of geographical dialects within what they refer to as Ndyuka tongo "Ndyuka language". The number of Ndyukas today is estimated at around 16,000. The other party in the rise of the pidgin were Trio Indians, living today in southeastern Suriname along the upper Tapanahoni River and its tributary the Palumeu, and in southwestern Suriname along the Curuni. Their language belongs to the Cariban family, as do that of the Wayanas, the other major Indian group of southern and southeastern Suriname, and Kalinya (Carib) of coastal Suriname. As seen in §2 below, the Trios earlier lived lower on the Tapanahoni, where many Ndyukas live today. Their current population is estimated at around 1200, mostly living in two villages (indicated by "T" on the map). 2 This paper summarizes what is known of the history of the language (§2) and describes its phonology (§3), syntax (§4), and lexicon (§5), with a brief comment about its semantic organization (§6). It builds on, but does not presuppose acquaintance with, material from Huttar (1982), supplementing and correcting that material through documentary research and fieldwork carried out in Paramaribo in 1989.3 In order to address the question of the sources of features of a pidgin, the language is viewed throughout the descrip tive sections (§§3-6) in relation to Ndyuka and Trio (and occasionally Sranan and Wayana). A map is included to clarify some points about relations between Bushnegroes and Indians mentioned in the historical section, while an Appendix provides a small amount of information on a trade jargon used between Wayanas and Alukus.
2.
Historical and cultural context
Ndyuka-Trio Pidgin is one of at least three languages reported to have been used between blacks and American Indians of Suriname and French Guiana. It is one of two "trade languages" (handelstalen) referred to by de Goeje (1939:10) as in use between Bushnegroes (Ndyukas and Alukus) and Indians (Trios and Wayanas) in southern Suriname, while Coudreau 1895 mentions what is apparently another variety.4 Known written reference to this language begins with reports in the early twentieth century (de Goeje in Franssen Herderschee 1905b, with details first appearing in de Goeje 1906, 1908a).
102
George L. Huttar and Frank J. Velantie
Trade between either Ndyukas or Alukus and Indians is reported in several nineteenth-century sources (von Sack 1821, Crevaux 1883, Kappler 1887, Coudreau 1895). De Goeje's observations suggest that the language must have arisen and become fairly stabilized lexically before the late nineteenth century:5 The Negroes who established themselves on the Djuka Creek two centuries ago found Trio Indians living on the Tapanahoni. They maintained continu ing relations with them; only later did they come in contact with the Wayanas, and the settling of these Indians on the Tapanahoni dates only from the second half of the nineteenth century. The trade dialect shows clear traces of these circumstances. It consists almost entirely of words borrowed from Trio or from Negro English. The words which are derived from Carib have likewise been introduced by the Negroes; in modern day Negro English one still finds many of these words, mainly names of animals or plants, which the Negroes at some time had to ask from the original population (1908a:215).
At one point ( 1908a:31) he suggests that the Ndyukas and the Trios were trading with each other since not long after the former established themselves on the Tapanahoni, that is, shortly after the treaty of 1760 (Mosis & Scholtens 1988:3; Thoden van Velzen & van Wetering 1988:11,403). Although he does not give reasons for this surmise, it is confirmed by Ndyuka oral history as reported by older Ndyuka men today. De Goeje and others make frequent mention of the Ndyukas' efforts to prevent others from interfering with their lucrative trade with the Indians, a phenomenon observed at least as early as an expedition in 1861 (van Coll 1903:605-606). Franssen Herderschee (1905a: 90, 115-116) reports that the Alukus went to war against the Oyampis in the early 1840s in order to get control of trade with the Europeans of the Oyapock River (on the eastern border of French Guiana) and hence with the Indians of the interior. Von Sack's account of his travels in Suriname in 1810-1812 refers to hunting dogs that the Ndyukas obtained by exchange with "Acouri" Indians6 (1821:109). Neumann states (1967:91) that the Ndyukas had replaced the Saramaccans as the main trading partner of the "Akuris" by around 1800. Although the language apparently arose in a trade situation, or a general contact situation leading to a trade relationship, between Ndyukas and Trios, its use has not been limited to that context. Crevaux (1883:276) reports that he was addressed in this language by a Trio woman, and gives a transcription of what she said that is in general agreement with the form of the language as it is spoken today.7
Ndyuka-Trio Pidgin
103
Ndyuka trade with other groups, at least in the past, has been reported. For example, von Coll (1903:609-610) refers to their trade with the Akurios, whom he identifies as the Oyampis, a group living in extreme southeastern Suriname and northern Brazil. A number of flora and fauna terms in Ndyuka do indicate the likelihood of contact with the Oyampis (Huttar 1989). De Goeje (in Franssen Herderschee 1905b: 130) reports that the Ndyukas would travel for trade purposes to the Yari River in northern Brazil just south of southeastern Suriname, as well as to the Curuni River in far southwestern Suriname. Unfortunately, these reports do not indicate what language was used in any of these trade situations. Wayanas lived on the Yari and Paru Rivers in Brazil in the eighteenth century, while the Curuni has been Trio territory (de Bruijning & Voorhoeve 1977:616, 659). Today, Ndyukas report that the language is now used for their trade and general conversation with the Wayana Indians as well as with the Trios, although contact with the latter group is still more frequent.8 Some of the Wayanas moved to the Tapanahoni and the Palumeu at the urging of the Ndyukas in around 1865 (Neumann 1967:92). De Goeje's observation (1906: 5) that "Wayanas frequently act as middlemen between Trios and Bushnegroes" indicates some of the importance of the relations between Ndyukas and Wayanas early in this century. By de Goeje's time the Trios had already withdrawn as far as possible from the neighborhood of both Bushnegroes and whites (de Goeje 1908b:22). Personal observation by the second author9 indicates that Ndyuka trade with the Trios and Wayanas continued to flourish into the 1960s. Increasing travel of the Indians to the coastal area beginning around this time has not only cut back on such trade, but has also given some Indians the option of using Sranan in their relations with the Ndyukas. Both of these trends have contributed to the pidgin's decline in use, so that today many Ndyuka men in their 30s (perhaps even older) and younger do not know it. Ndyukas refer to the language as Alukuyana, the name they also give to the Wayanas and their language, and Ingii "Indian". Wayanas call it Mekolo "Negro", a term they also use to refer to Ndyukas and their language. There is, however, no hint of any misconception by either group that in speaking this language they are speaking the language of the other group. The language must be specially learned, usually by observation; it cannot be produced on the basis of a knowledge of Trio or Ndyuka alone.
104 3.
George L. Huttar and Frank J. Velantie Phonology10
While Ndyuka (N) has only five vowel qualities in contrast — /i e a o u/ — Trio (T), like Wayana (Jackson 1972:48), has seven: /i e a o u ï /. The pidgin has only the five of N, although apparently free variation includes at least the high central [ï]: the first vowel in punyeke "white-lipped peccary" can be either [u] or [ï], for example. More often T /ï/ corresponds to pidgin /i/, as in T wïw'i, pidgin wiwi "ax". N also has contrast of nasalization, vowel length, and tone. There is free variation in the first of these in some forms in the pidgin, so that one hears, for example, both silon11 and silo for "be, have". Some interjections are pro nounced with exaggerated length and high pitch. The forms nóno "No!" from N and nonó "hole, ground" from T differ by pitch only. There may be a few other pairs of this sort, such as dili (dílil) "expensive" (cf. N díi, Sranan 'diri) and dili (dilî?) "three" (cf. N dií, Sranan dri). Voiceless and voiced stops contrast in N, but not in T, which does not appear even to have voiced variants of the stops. They do not contrast in the pidgin, although voiced allophones do occur (after nasals, for example, and apparently in free variation in some words elsewhere). N has /1/, not /r/, while T has [r] and lateralized r in free variation. Our own work on the pidgin shows [1] almost exclusively, although de Goeje and van Tuinen (as presented in de Goeje 1908a) list several instances of [r]. As pointed out by de Goeje (1906:111), most intervocalic /d/'s in N and Sranan correspond to /1/ in the pidgin: compare N and Sranan dede with pidgin alele "die" and Sranan frede with pidgin pelele (and N feele) " afraid", for example. The pidgin has [p] and [f] in free variation. A somewhat parallel phe nomenon is found in the speech of the Sipaliwini River Trios, for whom Jones (1972:43) reports free variation between [p] and [] in unstressed syllables. T has no consonant sequences except for nasal-obstruent sequences across syllable boundaries. N also has these sequences, plus syllable-initial clusters where the second is /w/ or /y/. The pidgin has the latter feature — e.g., dyampo ~ yampo "collared peccary" — as well as word-initial /tl/ [ti] ~ [tr]: tron ~ don ~ tolon "turn".12 On intonation, see §4, Syntax.
Ndyuka-Trio Pidgin 4.
105
Syntax
The syntax of the pidgin closely follows that of T, resembling that of N only when N and T are alike.13 Basic clause-constituent order is SOV (although transitive sentences with both nominal S and nominal are unusual), as in T, while N has SVO order: (1)
S O V Tano mati ondoo kolu so teke. Tano partner 100 guilder ? take 'Tano's partner took 100 guilders."
When S is pronominal, the order is OSV, occasionally SOV (Ndyuka is SVO, while a pronominal S in Trio is usually marked only in the verb, giving simply OV): (2)
(3)
O S V Disi pumi luku. 3rd 1st see "I see him." S V Kaikushi pumi wani wa. dog 1st want NEG "I don't want dogs."
Although our sample of sentences with indirect object is very small, IO appears to immediately follow S: (4)
S IO V Pumi tata mati so apukuta teke. 1st father friend ? paddle take "My father gave his friend a paddle."
Other sentence constituents occur either initially or after S: (5)
Time Loc Tide pumi apataya today 1st village "Tonight you'll stay in
S V piyu silipi ye. 2nd sleep EXCLAM my village!"
106
George L. Huttar and Frank J. Velantie
(6)
(7)
S Time Accomp O V Mékolo so grandwei Tlio akoloni brudu miang. ? long.ago Trio ACCOMP blood eat/drink negro "Long ago the blacks drank blood with the Trios [to take an oath]." S Degree O Pumi so trángamee oto 1st ? hard game/meat "I want meat very much."
V wani. want
(See also the Locative examples below.) Predicate complements follow S:14 (8)
(9)
Time S PredComp Tide tuna munu. big today/now water/river 'The river is high now." S Piyu
PredComp moime
ye!
2nd
good/beautiful
EXCLAM
"You're great!" Vocatives occur either initially or finally, response particles initially, and the exclamatory particle ye finally: Voc (10) Mati, partner "Friend,
S V piyu kon? 2nd come have you come? [greeting]"
S V Voc (11) Pumi kon pai. 1st come friend "Friend, I have come." S PredComp Time (12) Mekolo mati fosi ye. negro friend first EXCLAM "Long ago the blacks became our friends indeed!"
Ndyuka-Trio Pidgin
107
Response V (13) lya silon. yes be/have "Yes, there are." Yes/no questions are distinguished from statements (and from com mands) by final rising intonation, as in both T and N — indicated here by a question mark: (14) Walapa sale? bow bring "Did you bring any bows?" (15) Walapa sale. bow bring "I brought bows." WH-questions frequently begin with the questioned constituent (a ques tion mark is used in the following examples merely to indicate interrogative status, not to indicate final rising of intonation): S Accomp V (16) Sama piyu akoloni kon? who/what/which? 2nd with come "With whom did you come?/Who came with you?" Time S (17) Onten piyu when? 2nd "What day did
V kon? come you come here?"
S V (18) Omeni Wayana kon? how many? Wayana come "How many Wayanas came?" If is questioned, it may occur either in first position or in the usual position after S: O (19) San what? "What
S V piyu sale? 2nd bring did you bring [to sell]?"
108
George L. Huttar and Frank J. Velantie s (20) Piy 2nd "What
o v sama so sale? who/what/which? ? bring did you bring [to sell]?"
The question word for "where" derives from N "Where is it/she/he?", but is a single morpheme in the pidgin:15 Loc S (21) Ndyuka: Pe a where? 3rd sg. "Where is it?"
V de? be.at
Loc S V (22) Peade so piyu kalala silon? where? ? 2nd bead be/have "Where are your beads?" The locative/existential/possessive verb silon may be omitted in such ques tions: Loc (23) Peade where? "Where
S pumi kaikushi? 1st dog are my dogs?"
Tense distinctions are indicated by adverbs: S Time PredComp Time Means (24) Panakiri granwee mati wa, tidei yenulu whiteman long.ago friend NEG now eye "Long ago the whites were not our friends, now before you as friends]."
V luku. see [you] see [them
Semantic relations between propositions are expressed by juxtaposition of sentences, each followed by a pause (indicated here with commas):16 (25) Panakiri tinigenee so kon, nono, whiteman only ? come no pelele, hesime busi kibri, sontem fear quick jungle hide maybe
moime wa, good NEG alele, nono, die no
ala Tlio all Trio takrume pai! bad friend
Ndyuka-Trio Pidgin
109
"If only whites came, no, that would not be good, all the Trios would be afraid, and would quickly hide in the jungle, where perhaps they would die — no, that would be bad, friend!" Within the noun phrase, possessors and other modifiers always precede the head, as appears to be the case in T. In N, possessors either precede or follow the head, while other modifiers precede the head. (26) purni sisa 1st sister "my sister" (27) Mayoli pataya Mayoli village "Mayoli's village" (28) kili kaikushi man dog "male dog" (29) sineki nono snake hole "snake hole" In the following sentence, piyu sale can be taken as a postnominal relative clause modifying kaikushi, but another interpretation seems equally possible, with "dog" as the object of "bring": Loc
S S V (30) Peade pumi kaikushi piyu sale? where? 1st dog you bring "Where is my dog that you brought/were to bring?" Or: "Where did you bring my dog?" There are a few noun-verb sequences that function semantically (and presumably syntactically) as nouns (31) or as verbs (32-33), with OV or Adjunct-V order of their constituents: V (31) siliba kisi fish catch "fishhook"
George L. Huttar and Frank J. Velantie
110
(32)
Loc V nonó trowe ground throw "fall"
O? V (33) mato waram fire warm "burn" Intensive adverbs follow the adjective or verb they modify, as in T. In N they precede attributive adjectives, but follow predicates, whether adjectives or verbs. Adj Int (34) mumpi fulu greedy much "very greedy" S (35) Disi 3rd "It's
V Int dili mama. expensive very too expensive."
V (36) Woto game/hunt "[This dog]
Int mama. much is an excellent hunter."
Negation is marked by a form at the end of the sentence, thus immedi ately after the verb. Sentences can be negated in T by adding a suffix {wa for many verbs) to the verb, while in N the negative occurs immediately after S. S V Neg (37) Kaikushi silon wa. dog be/have NEG "There are no dogs here." O V Neg (38) Oli wani wa. woman want NEG "[They did] not want women."
Ndyuka-Trio Pidgin
111
The pidgin, like T, has postpositions; N has prepositions, but in locative/ directional phrases it frequently also has a postposition in addition to the preposition (Huttar & Huttar 1994). By far the most common postposition in the pidgin is akoloni, "with (accompaniment)", but locatives baka "behind" and ondoo "under" have also been heard: Accomp O V (39) Tlio akoloni, Mekolo akoloni kasaba miang. Trio with negro with cassava eat "[They] eat cassava with the Trios and the blacks." Loc V (40) Bidon baka silon. metal.barrel behind be/have "[They] are behind the barrel." Loc V (41) Manyan ondoo silon. mango under be/have "[It] is under the mango tree." Location and direction are, however, frequently given with no adposition: Manner Loc V (42) Hesime busi kibri. fast jungle hide "[They] quickly hid in the jungle." Loc S V (43) Tepu pumi komoto. Tepu 1st leave "I've come from Tepu." S Loc V (44) Ala kiyo pataya waka. all young.man village go "All the young men went to the village." The postposition akoloni in (39) above is also used for "two", and by some speakers as part of "three", etc.:17
George L. Huttar and Frank J. Velantie
112 (45) dili akoloni three with "three"
Causation or reason is expressed by dati meke; this may be taken as an unanalyzed whole from N, where it has a similar function as well as sentenceinitial position, since dati hardly occurs in the pidgin otherwise: S PredComp (46) Dati meke mekolo wayana mati. that make negro Wayana partner "That's why Ndyukas are friends of Wayanas." Although some forms occur both reduplicated and unreduplicated (e.g. siki, sikisiki "sick"), there is no clear syntactic difference between the two. There are three forms, so (as in examples 1,4, 6,7, 20, 22, and 25 above), -me, and a-, whose functions are not at all clear. A very cursory discussion of the first two of these is given in Huttar (1982), while de Goeje goes into a bit more detail. His hypothesis about the origin of so is worth repeating in full here (1908a:215-16): Often one encounters "so" before the verb. How one gets this word is not easy to say. An explanation which is at least somewhat satisfactory is the following: Trio: (toena) so wai = "I want to have (water)," in which so = the verb "want". This verb is always inflected with the auxiliary verb form wai, which = 1st person sing. pres. tense indic, mood of the auxiliary verb. Negro English: (mi) wanni = "(I) want to have." Now since "I want" would be one of the first expressions that one would have needed to use, it is possible that the blacks considered "waf as equivalent to "wann/", had to maintain "so" to be understandable even though it was inexplicable to them, and so said, "toena so wannf'; by analogy they also placed "so" before other verbs, and this was regarded by the Indians as a form corresponding to their te, ti, t, which is often placed before the stem of a verb or adjective in Cariban languages.
The form -me occurs frequently, but not obligatorily, after stative predicates, as in exx. (9) and (25) (moime, takrume), and after the same forms function ing adverbially as (25) (= 42) (hesime). Some of these forms, such as moi "good", are also always given with -me (i.e. moime) in isolation as transla tions of Ndyuka words, for which reason we have followed de Goeje in writing the entire form as one word. Some speakers include the -me in translating numerals higher than two, as in dilime "three".
Ndyuka-Trio Pidgin
113
Occasionally - occurs after nonstative predicates, as in the following explanatory comments from de Goeje (1908a:216): In Trio and other Cariban languages mee (or pee) is used to express the socalled translative or essive; exactly the same function is performed by mee in: matti-mee wanni = (I) want (to become your) friend. In addition, mee is used after some words borrowed from Negro English, which can be used in the same way, such as mooi-mee, pretty, pikien-mee, small, a little, etc.
It appears that for some forms, such as moime, the -me is now an inseparable part of the form. For some others, the -me sometimes occurs, and sometimes does not. It is not clear whether the occurrence of /a/ before some forms is a phonological or a syntactic phenomenon. As it often occurs immediately before a verb (or as a part of a verb the etymon of which does not begin with a), it may be a sort of pronominal reprise of the subject, deriving from Ndyuka or Sranan 3rd singular pronoun a — as de Goeje (1908a:216) suggests. Yet it also occurs with forms other than verbs, chiefly proper names:18 (47) aplatti "share" Sranan prati "share" (cf. N paati "share") (48) alele "die" Sranan and/or N dede "die" (49) Alitaní "Litani River"; Alawa "Lawa River"
5.
Lexicon
The lexicon is derived almost entirely from N (or Sranan) and T, with a significant contribution from both sources, but somewhat larger from the former. Other sources include other local Indian languages (Kalinha and Wayana in particular) and French. Personal pronouns and most adverbs and verbs are from N/Sranan, while nouns derive nearly equally from both sources. Many forms presumably contributed by the Ndyukas show a closer phonological resemblance to Sranan than to modern (late twentieth century) N itself. One possible explanation for this is that N actually was phonologically closer to Sranan at the time the pidgin first arose than it is now. If that is the reason for the resemblance of some of the words in the pidgin to Sranan, we have here another piece of evidence for the relatively early origin of the pidgin, especially if we can get some idea of when Sranan and N started
114
George L. Huttar and Frank J. Velantie
diverging phonologically (assuming that they ever were more alike than they are now). 19 Another very likely reason for the appearance of Sranan-derived rather than N-derived words in the pidgin, however, is the propensity of Ndyukas to use Sranan rather than their own language in dealing with outsiders, appar ently as a means of protecting their way of life from outside curiosity and interference. Such a practice seems quite clearly to have been the case for N (and other Bushnegro — see Green 1974:15) interaction with people from coastal Suriname (see discussion in Huttar 1987), and it is not unreasonable to assume a similar practice in their relations with the Indians with whom they came in contact. De Goeje (1908a:204-14) gives a list of words and sentences compiled by a Bishop van Tuinen. The following summary of origins of lexemes largely follows the organization (by de Goeje?) of van Tuinen's list, but includes additional forms that have turned up in our own documentary re search and fieldwork. The list is surely not complete, but it is as far as we know representative. Many meanings for which van Tuinen (or de Goeje 1906 before him) gives only a form derived from N are shown below as having an additional form from T or some other source. Our work has been almost entirely with native speakers of N, while van Tuinen and especially de Goeje in their travels had significant contact with both Bushnegroes and Indians (Franssen Herderschee 1905a, 1905b, de Goeje 1908a, 1908b). Where a T-related and an N-related form are in use for the same meaning, Ndyukas refer to the use of the former as talking more dipi "deep" — i.e. further removed from common knowledge and everyday experience.20 In the following lists, N = Ndyuka/Sranan, T = Trio, = Kalinha/Carib, W = Wayana, and F = French. Where two sources are given as "X and Y", the language has two forms, one from each source; "X and/or Y" means that one form has two possible sources. If a form could have come from either T or Kalinha, or from either T or Wayana, only T is given, as the more likely source. Some of the N forms originally derived from K. BODY PARTS, etc.:
N : blood, mouth, beard,21 neck, foot/leg, feces, fever T: eye, ear, stomach, penis, cough (?)
Ndyuka-Trio Pidgin
115
FAMILY, SOCIAL RELATIONSHIPS, ETC.:
N:
sister, brother, friend/trade partner, friend (term of address), language, paper, payment T: man, woman, child, Bushnegro,22 shaman, medicine/magic, merchan dise23 K: white man, old man W: headman T and N: mother, father, child24 MATERIAL CULTURE:25
N: jungle shelter, mosquito net, boat pole,26 arrow,27 club, pin, needle, garden, soup, salt, barbecue rack, fan, pot, calabash, large metal barrel, storage drum/chest, storage basket/trunk, basket, can/plate, bowl, string/ rope, lock, cigar, comb, loincloth, ornamental band worn on calf, pipe, shirt, cloth(es), ring T: house/hut, canoe, gunwale, bow, iron arrowhead, nail, machete, knife, woven backpack, feathered headpiece K: paddle, gun (?), ax, kashiri, hot Indian drinks W: cotton belt F: bottle (via W?) ?: village, former garden,28 necklace of brown seeds N and T: scissors, hammer, bead, mirror N and F: strong drink N and W: button T and F: hat N and ?: hammock29 N and/or T: cassava sieve, woman's apron N and T and W: fishhook NATURE:
N:
night, star, mountain/hill, jungle, path, savanna, hinterland, creek, river mouth (= body part), rapids, swamp, island, gasoline, smoke, ash, char coal, sand T: sun, moon/month, rain, ground, rock/stone, water/river, sea, fireAight/ candle/tinderbox/matches, resin (for torch) N and/or T: morning30
116
George L. Huttar and Frank J. Velantie
ANIMALS:
N:
capuchin monkey, howler monkey, great anteater, jaguar/various cats, fish, bird, black curassow, parrot, trumpet bird, bee/wasp T: spider monkey, white-lipped peccary, deer K: tapir, turtle, dog31 Saramaccan: collared peccary ?: game/meat32 PLANTS:
N: pineapple, cassava, papaya T: tree, banana/plantain, napi (tuber), sweet potato, reed for arrows, sugar cane, sailgrass N and ?: tobacco, rice PRONOUNS:
N: I, you, he/she/this, all, other, who/what/which33 N and ?: who? ADVERBS, ETC.:
N: much, few/a little, all, here (= this), far, behind, under, upstream, down stream, beyond, finished, today/now, fast, slow, where?, when?/why? (= who/what/which), perhaps, yes, various greetings T: yesterday, with N and T: only, no N and ?: very/too, long ago N and/or T: tomorrow (= morning) ADJECTIVES:
afraid, angry, big, small/young, dead, drunk,34 good, high, short, long/ tall, delicious, lazy, skinny, beautiful (= good), wet , ripe, bad, strong, heavy, sick, three, four, five, six T: thick/fat (= T-derived form for "big"), one, two (= with) K: old(?) (= old man) ?: red (two forms) N and T: big, black and/or W: fat N and ?: white N:
Ndyuka-Trio Pidgin
117
VERBS:
arrive,35 leave behind/remain behind, burst/split, understand/know, break/open, dance, drink/eat,36 go/walk, cut (off)/injure, laugh, lie/de ceive (two forms), loose, make, take, name, dig up, pluck/remove, speak, talk,37 hurt, roll, shoot, hit, close, sharpen, steal, return, blow out/rest, fasten, fight, fight/be angry, distribute/share, forget, sell, hide, wait, want, throw away/divorce/abandon, exist T: perform magic (= shaman) K: (?)paddle N and T: sleep T and/or W: carry/bring
N:
6.
Semantics
We have not studied the semantic organization of either N or T extensively enough to make significant observations about the pidgin in relation to those languages. We do note that the use of one form (nyan) for both "eat" and "drink" is a feature of the pidgin not found in either N (or Sranan) or T.
7.
Conclusion
The trade language here called Ndyuka-Trio Pidgin appears to have been in use for 150-200 years, and to have been stable in structure during the period (100 years) for which we have any written record of utterances in it. Its phonology for the most part includes only those contrasts found in both sources. Its lexicon derives from both Ndyuka/Sranan and Trio, with the former predominating in the pronouns, adverbs, and verbs but the two sources fairly equally balanced in the nouns. Its syntax, by contrast, is essentially Trio, not Ndyuka, being consistent with what is known about that of Trio for every feature which we have examined. Ndyuka-Trio Pidgin is somewhat unusual among pidgins in that it is consistently SOV, and in that one of its two main sources is a creole language.
118
George L. Huttar and Frank J. Velantie
Appendix: Wayana-Aluku trade jargon As we know of nothing else published about the trade jargon ("jargon de traite") used by Alukus and Wayanas, we here present a translation of de Goeje's material on the language (1946:100-101), followed by a brief commentary of our own (starting immedi ately after the word list): Trade jargon 2. In their conversation with the black maroons of Suriname, the Indians of the interior make use of a trade language. In the Tapanahoni and the surrounding regions it is the creole language of Suriname, mixed with Trio words. In the Lawa and Litani [near the Alukus, not the Ndyukas — tr.] it is a very simplified Wayana mixed with Kalinha (Galibi) words. The mistakenly used prefixes of these Kalinha words bear testimony that they are the remains of a trade jargon formerly used by the French in their speech with the Kalinhas of the coastal region. Here is its [vocabulary] list: stomach parika hammock hamaka okiri powdered mixture man Wayana Kalina of pepper and salt pome peíto tomorrow koropo subject friend y-akere-nu creek polili i-pawana-li earth, ground nono friend village pati garden, plantation maina test of mar you (sg.) amole inele maroke he riageability a awini katuli backpack merchandise y-eneka-li other, others amu omatau everything, -one papuru ring beautiful kuranu kusi, kusi-kuši bring, carry s-alei quick, soon good irupa buy s-apui quick t-aka-ne drink s-eneli strong tope hear š-etai s-ikapui awempo do, make finished lie (deceive) ekali give s-iri, s-ili throw, throw away s-imai attach, tie kulua-pok sad, angry kutamke (Negro eat s-one fell word?) š-eketei n-amai bathe oneself s-ukumiti fall n-atati s-ukuti die understand, know n-ikluti see s-ene burn n-isan not see s-ene-wa go arrive n-ompui not give s-ili-wa (etc.) hello! arikito! not want ise-pa (Arawak word?)
Ndyuka-Trio
Pidgin
119
Crevaux (1882a) has entries in his brief vocabulary of Wayana (Roucouyenne) for 44 of the meanings given above. In 37 of these 44, the form given by de Goeje is the same, or nearly the same, as one given by Crevaux. In 4 cases, Crevaux gives a form that differs in a phonologically significant way, leaving 3 for which the forms given by Crevaux are entirely different from what de Goeje reports for the trade jargon. Such correspondence would support de Goeje's characterization of the jargon as basically Wayana with some Kalinha words, but for two considerations. First, since the trade jargon was apparently already in use with outsiders, there is a definite possibility that Crevaux was given words in that jargon rather than in Wayana. Unfortunately, he does not comment on how he came by his wordlist, so it is difficult to judge on that basis to what extent this may have been true. Forms like parika and hamaką (cf. Portuguese, Spanish barriga and Arawak, Spanish hamaca) strengthen the impression that he has recorded a trade jargon here. Second, a check of most of the forms in the jargon against de Goeje's Wayana vocabulary in the same volume (1946:137-238) suggests that very few of the forms given for the jargon are in fact traceable to Wayana. Two or three times as many correspon dences, on the other hand, can be found with Trio forms as given in de Goeje (1906:7187) and Leavitt (1971). Since even a cursory check of the above list against Hoff (1968) and Aloema et al. (1987) shows even more correspondences with Kalinha, it appears that de Goeje's list shows more of the "mixing" with Kalinha words than of the "very simplified Wayana" he mentions. The last three entries provide some grammatical information, namely that negative is expressed by wa or pa placed after the verb, sentence final, in agreement with Crevaux (1882a) on Wayana, with regard to both the position of the particle and the alternation of the forms. De Goeje (1946:110, 120, 124), however, describes Wayana wa as coming before a noun that it negates, or being used as a negative reply, "No!", while other forms are used as negative suffixes on verbs. In Trio, however, wa is a common negative suffix on verbs (Leavitt 1971:9). Thus here, too, the above list gives more support for Trio than for Wayana sources for features of the trade jargon. De Goeje's list is difficult to interpret for phonological information, given that he appears to be representing some noncontrastive differences, especially in the vowels. However, his use of ë as one of the vowel symbols suggests that the jargon had at least six vowels, thus more than the i, e, a, o, u of Aluku.
Notes 1.
See Huttar (1988) for the place of Kwinti with respect to the other five Bushnegro creole varieties.
2.
Ivan Schoen, personal communication, 1989.
3.
Documentary research was carried out by the first author, who alone is responsible for the form of this paper (including any English translations of quotes from sources in other
120
George L. Huttar and Frank J. Velantie languages), while the second author, a native speaker of Ndyuka, interviewed and/or collected word lists from three Ndyuka men, one Ndyuka woman, and one Wayana man. None of our material comes from spontaneous language use. From our interpretation of the records, presumably some at least of the documentary material was spontaneously produced. Special thanks are due to the library staff of the Stichting Surinaams Museum for their patience and expertise in facilitating the documentary research, and to the National Science Foundation for their support under Grant No. BSN8811211. Our thanks to William J. Samarin for helpful comments on an earlier version of the paper, and to Carolyn Dyk for producing the map.
4.
De Goeje (1939:10) enumerates the languages heard in the Indian village Koemakapán, and includes "four creole languages, Negro English, Cayenne patois, Aluku-Wayana trade language, and Ndyuka-Trio trade language" ("vier creoolsche talen, Negerengelsch, patois van Cayenne, Boni-Oajana handelstaal, Djoeka-Trio handelstaal"). Ear lier (1906: 6), however, he says that the Aluku and Wayanas conversed in "Indian" ("het Indiaanscn"), in contrast to the Ndyukas and Trios. But he later (1946:100-101) confirms his 1939 view of the language as other than Wayana (see Appendix to this paper). Ndyukas confirm that there is a language spoken between Alukus and Wayanas that differs from Wayana and from the Ndyuka-Trio pidgin. Coudreau (1895:602) refers to a language used by the "créoles" of French Guiana in addressing Émerillon Indians, and includes this sample, without translation: Mo même, créole, bon, mo même; to même, Émerillon, pas bon, to même.
5.
Neumann (1967:93) draws a similar conclusion from de Goeje's writings: "Die Handels sprache muss...schon vor der Mitte des 19. Jh. existiert haben".
6.
De Goeje (1906:6) identifies this group with the Trios.
7.
'"Panakiri ouani oua, blancs besoin pas. A la pikininialele, là enfants morts. Nono poti, terre trou. Echimeu ouaca, vite pars. Cassava mia oua, cassave manger pas.'" Instead of "Whites not needed...there children dead...earth hole...quickly leave...no cassave to eat", we take these utterances to have meant "Whites not wanted...all children dead...(they have been) put in (this) hole...quickly leave...(there is) no cassava to eat" (or the last part can be taken as "leave right away, don't stay around to eat"). De Goeje (Franssen Herderschee 1905b: 130) interprets the individual words in a way consistent with our interpretation.
8.
There are about 1200 Wayanas living in Suriname today (Ivan Schoen, personal commu nication, 1989).
9.
The second author lived in Drietabbetje, residence of the paramount chief of the Ndyukas and regular stopping place for Indians headed downriver, until the mid-1970s.
10.
For Ndyuka phonology, we use Huttar & Huttar (1994:544-94); for Trio, Jones 1972.
11.
Syllable-final V+N represents a phonologically nasalized V, phonetically realized some times as V followed by nasal, sometimes as nasalized V.
12.
On the relation between /1/ and vowel length in the pidgin and vowel length in N, see Huttar (1982:3).
13.
Comparison of the pidgin syntax with T syntax is hindered by the unavailability of a systematic description of Trio. Our comments on T syntax in this paper are based on
Ndyuka-Trio Pidgin
121
examples scattered through the vocabularies and morphological descriptions of de Goeje (1909) and Leavitt (1971). For Ndyuka syntax, we follow Huttar & Huttar 1994. De Goeje (1908a:215) pointed out that the pidgin sentence structure was "in general the Indian" one, citing among other features the occurrence of object before verb, adverbial adjunct before verb, and negative wa after verb. All examples in this paper are from the pidgin, except as otherwise identified. 14.
Pace Huttar (1982:4-5).
15.
Neither pe nor a is a form in the pidgin, while de is used (by some speakers?) instead of the more usual silon "be, have".
16.
The example is from de Goeje (1908a:214), where nothing is reported about pauses. Our statement about pauses is based on our own fieldwork.
17.
De Goeje (1946:77) gives T akoolon for "un autre". For the pidgin, he glosses akoloni, okoloni as "beide, te zamen" ("both, together").
18.
The addition of initial a to etyma beginning with a consonant (particularly r) is also found in Ndyuka itself: e.g. rain > alen "rain", (English or Dutch) rat > alata "rat", rice (or Dutch rijst) > alisi "rice". Likewise, some Ndyuka personal names appear to alternate between forms with and without initial a, e.g. Puyenke ~ Apuyenke.
19.
Smith (1987:150) suggests that Ndyuka and Sranan split around the first decade of the eighteenth century.
20.
Dipi is a term that is commonly used in Suriname, including in Sranan as spoken by various ethnic groups, to refer to language phenomena that are relatively difficult to understand, and hence all the more mysterious. Hindi of India, for example, is dipi compared to Sarnami Hindi spoken in Suriname; Sranan speakers consider Saramaccan more dipi than Ndyuka, with which it shares a much higher number of cognates.
21.
The corresponding form in modern N, baiba, means "chin", while Sranan barba means "beard".
22.
De Goeje (1908a:205) suggests that this form, mekolo, comes from negro, but "by what route cannot be determined". He gives mékolo as the Wayana form for (Bush)negro (1946:147), and mékolo as one of three T forms for a Ndyuka (1906:74).
23.
De Goeje (1908a:205) suggests Trio ekelama "buy" as a possible source of pidgin atemete. Crevaux (1882b:39) gives arépété for "bagages, objets d'échanges".
24.
The N-derived form is pikinini; the other form, kami, is presumably from T kami "term of address for younger sibling or grandchild" (Leavitt 1971:61-62). But cf. T sikami "boy" (de Goeje 1906:73) and W kahmi "child, term of address" (de Goeje 1946:145).
25.
Called "Ethnographie" in the original list presented by de Goeje (1908a:205).
26.
De Goeje (1980a:206) considers this form, kula, to have come into the pidgin from "Negerengelsch" (which can be either N or Sranan in his scheme), which in turn borrowed it from Kalinha. But since there is a likely kiKongo source for this word in N (see Huttar 1985:59), we assume that Kalinha was the borrowing language rather than the source language. With regard to the pidgin, we assume for simplicity's sake that the pidgin got the form directly from N.
122
George L. Ruttar and Frank J. Velantie
27.
De Goeje says (1908a:206) that this form, kwama, is from Kalinha; but it is also found in N.
28.
The form is kawalimbo. Cf. kaipïrïmbo, from ka:pïrï "make" and -mbo "former" (Hoff 1968:137). Kawembo also occurs alongside gron as an alternative form for "garden".
29.
The form from N, , may in turn derive from Arawak directly or via Portuguese.
30.
"Morning/tomorrow" is toko-toko, which de Goeje (1908a:207) suggests derives from T or N tokro, a quail (Odontophorus guianensis) whose call is heard shortly before daybreak. The modern N word is tokoo; Ndyukas also consider this bird an announcer of daybreak.
31.
Kaikushi; cf. kaiku:si, "jaguar", T kaikui "jaguar, dog". For W, de Goeje (1946:160) gives kaikui for both meanings, but de Goeje (1909:293) gives both kaikúsi and kaikui for "jaguar". De Goeje (1946:248) gives kaikushi for "jaguar" in Triometesem (Wayaricule), a Cariban language of southeastern Suriname and southwestern French Guiana.
32.
De Goeje (1908a:209) gives T joti "game" and W woto "fish" as possible sources of pidgin oto "game, meat".
33.
Sama is presumably from N sáma "who?"; cf. Ndyuka sama "person" (with low tone on both syllables, rather than the high-low tone sequence of sáma), Sranan suma "who?".
34.
Sungu means "sink" in N, but de Goeje gives "dronken" ("drunk"), not "verdronken" ("drowned"), as the meaning of the form in the pidgin.
35.
Komopo, from Sranan komopo "come from, come out of' (cf. N komoto).
36.
De Goeje (1906) gives ñami, de Goeje (1908a) miang or miáng, and our own fieldwork nyan; cf. N nyan (with N alternant nyami).
37.
Krutu, Sranan krutu "hold council, criticize" (cf. Ndyuka kuutu).
References Aloema, Nardo, M. J. Pierre, and . N. van der Ziel. 1987. Kalinha — Nederlands Woordenboek. Paramaribo: Instituut voor Taalwetenschap. de Bruijning, C. F. A., and J. Voorhoeve. 1977. Encyclopedie van Suriname. Amsterdam: Elsevier. van Coll, C. 1903. "Gegevens over Land en Volk van Suriname". Bijdragen tot de Taal-, Land- en Volkenkunde van Nederlandsch-Indie, 55.451-650. Coudreau, Henri. 1895. Chez Nos Indiens: Quatre Années dans la Guayane Française (1887-1891). Paris: Librairie Hachette et Cie. Crevaux, J. 1882a. "Vocabulaire Français-Roucouyenne". Grammaires et vocabulaires roucouyenne, arrouague, piapoco et d'autres langues de la région des Guyanes, ed. by J. Crevaux, P. Sagot, and L. Adam, 1-20. Paris: Maisonneuve et Cie, Libraireséditeurs. Crevaux, J. 1882b. "Quelques Mots de la Langue des Indiens Trios". Grammaires et vocabulaires roucouyenne, arrouague, piapoco et d'autres langues de la région des
Ndyuka-Trio Pidgin
123
Guyanes, ed. by J. Crevaux, P. Sagot, and L. Adam, 39-40. Paris: Maisonneuve et Cic, Libraires-éditeurs. Crevaux, J. 1883. Voyages dans l'Amérique du Sud. Paris: Librairie Hachette et Cie. Franssen Herderschee, A. 1905a. Verslag van de Gonini-Expeditie. Leiden: E. J. Brill. (Originally in Tijdschrift van het Koninklijk Nederlandsch Aardrijkskundig Genoot schap, 1905.1-174.) Franssen Herderschee, A. 1905b. Verslag der Tapanahoni-Expeditie. Leiden: E. J. Brill. (Originally in Tijdschrift van het Koninklijk Nederlandsch Aardrijkskundig Genoot schap, 1905.847-1032.) de Goeje, C. H. 1906. Bijdrage tot de Ethnographie der Surinaamsche Indianen. Leiden: E. J. Brill. de Goeje, C. H. 1908a. Verslag der Toemoekhoemak-expeditie. Leiden: E. J. Brill. de Goeje, C. H. 1908b. Beiträge zur Völkerkunde von Surinam. Leiden: E. J. Brill. de Goeje, C. H. 1909. Etudes Linguistiques Caraïbes. Amsterdam: Johannes Müller. de Goeje, C. H. 1939. "Bij primitieve volken". De Gids, pp. 1-31. de Goeje, C. H. 1946. Etudes linguistiques Caribes, vol. 2. (= Verhandelingen der Koninklijke Nederlandsche Akademie van Wetenschappen, Afd. Letterkunde, New Series, Part 49, No. 2.) Amsterdam: N. V. Noord-Hollandsche Uitgevers Maat schappij . Green, Edward Crocker. 1974. The Matawai Maroons: An acculturating Afro-American society. Washington, DC: Catholic University of America dissertation. Grimes, Joseph E., ed. 1972. Languages of the Guianas. Norman, OK: Summer Institute of Linguistics. Hoff, B. J. 1968. The Carib language. The Hague: Martinus Nijhoff. Huttar, George L. 1982. A Creole-Amerindian pidgin of Suriname. Society for Caribbean Linguistics Occasional Paper No. 15. Huttar, George L. 1985. "Sources of Ndyuka African vocabulary". De Nieuwe WestIndische Gids 59.45-71. Huttar, George L. 1987. "The Afaka script: An indigenous creole syllabary". The Thirteenth LACUS forum 1986, ed. by Ilah Fleming, 167-77. Lake Bluff, IL: Linguis tic Association of Canada and the United States. Huttar, George L. 1988. Notes on Kwinti. Society for Caribbean Linguistics Occasional Paper, No. 19. Huttar, George L. 1989. "The Portuguese contribution to the Ndyuka lexicon". Estudios sobre Español de América y lingüística Afroamericana: ponencias presentadas en el 45 Congreso Internacional de Americanistas {Bogotá, julio de 1985), 263-79. Bogotá: Instituto Caro y Cuervo. Huttar, George L., and Mary L. Huttar. 1972. "Notes on Djuka phonology". Grimes 1972.1-11. Huttar, George L., and Mary L. Huttar. 1994. Ndyuka. London & New York: Routledge. Jackson, Walter S. 1972. "A Wayana grammar". Grimes 1972.47-77. Jones, Morgan W. 1972. "Trio phonology". Grimes 1972.42-46. Kappler, August. 1887. Surinam: Sein Land, seine Natur, Bevölkerung und seine KulturVerhältnisse mit Bezug auf Kolonisation. Stuttgart: J. G. Cotta'sche Verlagsbuch handlung.
124
George L. Ruttar and Frank J.
Velantie
Leavitt, Claude W. 1971. Trio language analysis. Alalapadoe, Surinam: Stencil. Mosis, André, and Ben Scholtens. 1988. "Bosnegers en Overheid: Ontwikkeling van de politieke verhoudingen, 1650-1988 (4): Vernieuwing van de vredesverdragen in de 19e eeuw". De Ware Tijd, 21 October 1988, 3. Neumann, Peter. 1967. Wirtschaft und materielle Kultur der Buschneger Surinames: Ein Beitrag zur Erforschung afroamerikanischer Probleme. (= Abhandlungen und Berich te des Staatlichen Museums für Völkerkunde Dresden, Band 26.) Berlin: AkademieVerlag. Sack, Albert von. 1821. Beschreibung einer Reise nach Surinam und des Aufenthaltes daselbst in den Jahren 1810, 1811, 1812, so wie von des Verfassers Rückkehr nach Europa über Nord-Amerika, Second Part. Berlin: Haude & Spener. Smith, Norval S. H. 1987. The genesis of the creole languages of Surinam. Amsterdam: University of Amsterdam dissertation. Thoden van Velzen, H. U. E., and W. van Wetering. 1988. The great father and the danger: Religious cults, material forces, and collective fantasies in the world of the Surinamese Maroons. Dordrecht: Foris Publications.
Arabic-based pidgins and creoles Jonathan Owens Universität
1.
Bayreuth
Introduction: Social and historical background
In this paper I will discuss all the varieties of pidgin and creole Arabic known to date.* In §1 I will give the historical background to the development of Arabic-based /'s in the two periods and areas where they are documented, the early Islamic period (ca. 700-1000 A.D.) and the 19th-century southern Sudan and Chad. In §2 I survey briefly the / varieties, and in §3 I explore in some detail the social and demographic factors which led to the emergence of the Arabic-based /'s in the southern Sudan, Chad, and East Africa. Section 4 provides a linguistic sketch of one creole Arabic variety, Kenyan Nubi, and a discussion of the linguistic interrelations among the different varieties of Sudanic / Arabic, Sudanic colloquial Arabic, and substratai languages. 1.1. Classical Arabic and Arabic dialects It is well known that Classical Arabic (CA, attested between ca. 550-800) differs considerably from the modern Arabic dialects. That varieties sepa rated by over 1000 years should differ is hardly cause for discussion. What is not so clear is how the differences arose. There is agreement on some basic points. First, modern dialects differ considerably among themselves and in very many respects cannot be historically derived from CA via traditional comparative methods. Second, the modern dialects are phonologically and morphologically less complex than CA. And third, CA was but one variety of Old Arabic existing alongside Old Arabic dialects (Rabin 1951).
126
Jonathan Owens
At this point opinion splits. One school holds that CA was a variety spoken by Bedouin tribes (Fiick 1950). A second sees CA as an intertribal koine of some sort (Spitaler 1953, Corriente 1976), which perhaps was originally a functionally-restricted variety used in poetry (Zwettler 1978); it was not the language of everyday use. If one adopts the second viewpoint (which I personally favor), the modern dialects can be seen to a large degree as a continuation of old Arabic dialects, whose character is inferrable in part by the comparative method. If one adopts the first viewpoint, that CA was the spoken variety of the Bedouin Arabs, the question to be answered is how the dialects came to diverge from CA. One answer invokes the role of a / variety. Versteegh (1984) proposes that pidginization/creolization played a key role in the development of early Arabic dialects. He notes that as the conquer ing Arabs spread over a large area of the Mediterranean and the Near East, they came into contact with non-Arabs — Persians, Copts in Egypt, Berbers in North Africa, etc. — and that these peoples had to learn Arabic. Since, however, Arabs were generally a minority in the newly-conquered territories, the subordinate populations did not have adequate opportunity to properly learn Arabic, and hence used language-learning strategies that produced intermediate pidginized varieties. The modern dialects derive ultimately from this pidginization process, and in the end the Arabs themselves, at least'the city dwellers, took over this variety as their own (Versteegh 1984:69). There are at least three problems with Versteegh's thesis taken in its strongest form (cf. below), that the modern dialects at some point have pidgin ancestors. First, although Arabic dialects are structurally simpler than CA, the differences between them and CA are far less than what one finds between modern-day Arabic creoles and modern dialects (see §4.4). This is also true, apparently, for earlier times (ca. 1050). A brief fragment of an early Pidgin Arabic variety discussed by Thomason & Elgibali (1986) deviates almost as much from a modern Arabic dialect as it does from CA (see §2.1). If pidginization/creolization played a significant role, then one might expect to find a greater divergence between CA and Arabic dialects.1 Secondly, /'s generally arise in areas with a high degree of linguistic heterogeneity; but in most conquered areas there were only one or two languages other than Arabic (Aramaic and Greek in the Levant, Coptic in Egypt). Moreover, as Goodman (1986) points out, the Arabic language often penetrated into lands, such as the Levant, well before the areas had been
Arabic-based Pidgins and Creoles
127
brought under Arabic political control. These factors would have favored the growth of bilingualism more than pidginization/creolization. Thirdly, Versteegh's thesis, entailing that the Arabs gave up their own "high" variety in favor of the low one which the conquered peoples used, and made it their own, is not what one usually finds in pidginization/creolization situations, though Versteegh suggests a similar process in the spread of Vulgar Latin in the Roman Empire. One could, perhaps, ask for well-docu mented parallels to substantiate his suggestions. It should be noted that by pidginization Versteegh understands not neces sarily an end point represented by a stable variety, but rather a process of untutored second language learning (1984:40). In these terms Versteegh is certainly correct in pointing out the relevance of pidginization to the develop ment of Arabic in the eighth and ninth centuries, for at this time Arabic acquired large groups of native speakers. I think his thesis that pidginization played a CRUCIAL role awaits a more precise examination of the linguistic and demographic situation in the early Islamic period. 1.2. Sudanic / Arabic For over 1000 years, starting from about 700 A.D., the southern boundary of Arabic speakers lay in the Sudanic region, first around Aswan (see §2.1 below) and then in a region stretching from central Sudan across central Chad to Lake Chad. A major trade and pilgrimage route followed an East-West direction over the area, from West Africa to Egypt and/or Mecca, and in the 16th century the first of a number of centralized kingdoms,2 the Funj empire, developed in east-central Sudan. It is virtually certain that Arabic served as a trade language for the entire region throughout the period, as the data dis cussed in §2.1 below confirms. Unfortunately, before the 19th century no other documentation for the use of nonstandard Arabic in the region has been reported. Since it is in the Sudanic region that the best-attested Arabic-based /'s are found, it is necessary to summarize the historical background of the region. Demographics are discussed in §3. The Sudan underwent drastic political and economic changes in the course of the nineteenth century, changes which left their linguistic mark as well, most particularly in the southern third of the country, approximately south of the Sobat and Bahr el-Ghazal rivers. It is in this region that the present-day Sudanic / Arabic varieties developed in the second half of the
128
Jonathan Owens
Arabic-based Pidgins and Creoles
129
19th century. A brief historical sketch of this region in the 19th and 20th centuries is thus in order.3 In 1820 Muhammad Ali, ruler of Egypt, turned his expansionist plans southwards and established suzerainty in the northern and central Sudan in an area centered roughly around Khartoum. Trade contacts with the northern most of the southern tribes, the Shilluk, were established, and within a few years, 1839-1841, Muhammad Ali dispatched a series of three expeditions under a Major Selim to explore the Nile south of Khartoum. These expedi tions reached Gondokoro near modern-day Juba, and instituted the first trade contacts between the North and the southern hinterland. Initially the main object of export from the South was ivory, and this was effected on a seasonal basis: every year during the Nile high water, traders sailed to the South and returned before the dry season rendered the Sud swamp impassable. The ivory trade was initially a government monopoly, but the increasinglypowerful trading community, a mixed group of Europeans and Arabs of different nationalities, was eventually granted the right to trade directly in the South, and after 1854 the first permanent trading settlements were established there.4 These proliferated quickly, and by 1869 the German explorer Schweinfurth observed traders' camps or 'zeribas' connected by a transport system (Junker 1888:33) at a distance of every 18-21 miles in the Bahr el-Ghazal region and across into what is today Zaire. Zeribas were also established east and north of Gondokoro, and south into northern Uganda. In addition, itinerant Sudanic colloquial Arabic-speaking traders or 'jallabi' (pl. jallaba) began coming in increasing numbers from Darfur into the South, some of them returning annually to the North, others settling permanently in the South. Frequent intercamp contact continued into the period when the South was ruled directly by the Egyptian government (Junker 1891,111:378). The trading settlements were initially manned largely by northern Sudanese Arabs and Nile Nubians (not to be confused with the Nubi of East Africa), many of whom married local women and settled permanently in the South. Relations between the camps and the local tribes were, allowing for local variations, bad. As early as 1863, for instance, Grant (1864:357) reports that the Bari used every opportunity they could to steal from the "Turks", and later Gessi (1892:211) observed that, although the Egyptian government had been established at Rumbek for some years, the Dinka-inhabited area around the outpost was unsafe for travel. The militarily superior northern traders freely commandeered cattle and provisions for themselves from among the
130
Jonathan Owens
local groups, and because the traders' agents were paid partly in slaves (Gray 1961:46 ff.), and because it provided a lucrative export commodity, slave taking became common. Many of the slaves themselves became integrated into the military structure of the camps, so that soon the southern population outnumbered the northerners (see (2) below). The trading camps thus became entities unto themselves, supporting populations of up to 10,000, cut off from the indigenous population except to the extent that the camps had to requisi tion their supplies and labor in one way or another from the local population. In response to the increased trafficking in slaves, which by 1870 (Schweinfurth 1918:479) had outstripped ivory as the premier export from the South, the Egyptian government sought to place the South directly under its control, a mission entrusted successively to Baker (1870-1873), Gordon (18731876), and Emin Pasha (1878-1888). Because of their lack of resources and manpower, the government often simply commissioned the standing army of one trader or another to serve as a government force, and in many respects, e.g. the practice of plundering local villages for provisions, relations between the camps and the native populations remained unchanged (see e.g. Junker 1891, 1:279, 322, 355, 340, 438, 512, Gessi 1892:430). The government established a measure of control along the Nile; in the western areas and in Bahr el-Ghazal province its control was far more tenuous (Junker 1888: 49). During the period of Egyptian rule, a significant number of Egyptian functionaries entered the southern Sudan (Kaye 1988:47), their number reaching perhaps 300-400. However, the greater number of government soldiers were locally recruited (see e.g. Stanley 1890:136). The anti-slavery efforts, and even more the encroachment of the Egyp tian government, which from 1878 made a concerted effort to keep northern traders out of the South, were enough to antagonize many of the permanent traders, particularly those in the western areas, and some rebelled. The most serious revolt was that of Sulaiman, son of Zubeir Rahman, in 1877-79, and although the revolt was put down, one of Zubeir's lieutenants, Rabeh, took a sizable group with him into the Chari-Logone river basin, eventually captur ing the kingdom of Bornu. It is this group which Zeltner (1988:265) quite plausibly suggests was responsible for bringing Turku (see §§2.2, 4.2) to the Chari river basin.5 The Egyptian government never established wholly effective control over the South, and during the Mahdist uprising, with Khartoum falling in 1885, the South was invaded as well (in 1884); most of it was lost either to the
Arabic-based Pidgins and Creoles
131
Mahdi's troops or to indigenous groups. Soon only a few government settle ments remained along the Nile south of Gondokoro. Stanley (1890:246) reports that at this time the governor, Emin Pasha, had about 1300 regular soldiers more or less under his command, along with 3,000 irregulars and camp followers. Jephson (1890:222) gives a higher figure for Emin's Equatoria followers, putting the figure at 8,000. In 1888 Emin was forced to abandon the last settlements, and he withdrew to northern Uganda, from where in 1889 most of his Egyptian administrators and officers returned first to the East African coast and then to Egypt or Europe.6 A number of Emin's soldiers and camp followers (700 according to Saghayroon 1981:28) stayed in Uganda, however, and these were quickly co-opted first by the British East Africa Company as early as 1893, then by the British government to help establish British control in Uganda and Kenya. Some of his former troops served the Germans in Tangyanika as well (Furley 1959:326; see §3.2). These soldiers eventually settled in Uganda and Kenya, where their descendants, the (East African) Nubi, still live. In the southern Sudan former soldiers and camp followers of the traders, as well as some former government troops (Junker 1891, III:405, 454, 479), stayed on. Some of these joined the Mahdists, while others formed themselves into private armies with local spheres of influence (Collins 1962:71), and through these groups the / Arabic was kept alive. The British established control of the Sudan in 1898, though the Mahdists had been driven out of parts of the South by Belgian forces as early as 1896. After numerous local rebellions were suppressed in the South, that area enjoyed a measure of peace until around independence in 1956. At that time a rebellion against the central government broke out, continuing intermittently until the present day.
2.
Varieties of modern Arabic-based p/c's
There are four varieties of modern Arabic-based p/c's. Two of these are still spoken today. One is known only through a work written around 1930, and the other, Maridi Arabic, has been termed the oldest pidgin on record (Thomason & Elgibali 1986). Beyond summarizing them in this section, I will have nothing to say about them — for Maridi because there is no new data to add, and for Turku because a detailed treatment can be found in Tosco & Owens (1993).
132
Jonathan Owens
2.1. Maridi Arabic The existence of early Arabic pidgins, or pidgin-like varieties, is confirmed by an eleventh-century text discussed by Thomason & Elgibali (1986). It appears in an 11th-century work on geography written by the Andalusian (Spanish) Al-Bakri, a geographer who himself never left his native land. It is a short text, a little over 50 words, which is said to represent a very bad form of Arabic spoken by people in the region of Maridi. Mainly on the basis of plausible historical and textual evidence, Thomason & Elgibali conclude that the pidgin was spoken in Mauritania. There are other indications, however, pointing to a textual source in Upper Egypt or northern Sudan. In particular, the text contains linguistic features which today are found either uniquely or largely in Egyptian Arabic, e.g. CA q = ', fw' < fwq "on", and CA j = d (dwmaa < jwmaa "Juma"); the last is a correspondence which as a perfectly regular feature is attested today only in Upper Egypt (Behnstedt & Woidich 1985:1 Off.; see Kaye 1985). The text, moreover, was recounted to Al-Bakri by a trader from Aswan who had heard it in the course of his travels. It is, then, quite plausible to assume that the text represents a variety spoken somewhere in Egypt or the Sudan but which Al-Bakri or his source transposed to Mauritania to conventionally represent a bad form of Arabic spoken there. The text itself is likely a caricature which is partly based on an actual spoken pidgin variety, but which is also partly taken from local Arabic dialects and partly perhaps deliberate distortion in order to produce a "bad" Arabic (e.g. incongruous use of the dual; see Owens 1989 for discussion). The preponderance of evidence thus points to a source for the Maridi text in the upper Nile valley. If this is the case, it would indicate the possible presence of Arabic-based pidgins in the region for nearly a thousand years. 2.2. Turku Turku, as mentioned in §1.2 above, was brought to Chad and NE Nigeria by Rabeh's soldiers ca. 1900. Writing of the colloquial Arabic spoken in the Lake Chad area, Lethem (1920:xiii) reports that words and phrases intro duced by the followers of Rabeh "are generally known as 'turuk', this being the term applied in Bornu to Rabeh's non-Arab Sudanese troops". Even today the Kotoko in northern Cameroon refer to Rabeh's troops as 'turks' (Zeltner & Tourneux 1986:11). Turku was thus a nonnative Arabic which had its
Arabic-based Pidgins and Creoles
133
origins in the same social milieu as did the / varieties of the Sudan (see §1.2). I have no information about the number of Turku speakers, though Maigret (1931, cited in Samarin 1955:259) reported that Turku was one of the three trade languages (along with Sango and Bangala) used in French Equato rial Africa. It was used primarily in Chad and northern Cameroon, though I believe it never played an important role in Nigeria (see Tosco & Owens 1993 for discussion as to why). Precise linguistic details of Turku are known through a lexicon (about 1500 entries), a very inadequate five-page grammar, and collection of short phrases/texts by Gaston Muraz (ca. 1929), a French colonial medical officer. Though not the work of a professional linguist, the practically-oriented gram mar, the oldest detailed grammar of a Sudanic Arabic / available, contains enough data to construct a basic grammar and provide comparative material. It can, for example, be ascertained that Turku was an expanded and probably stable pidgin (see Prokosch 1986, Tosco & Owens 1993). (I have no informa tion about its status as a creole, and suspect it was not one.) In addition, Derendiger (1912), describing the variety of Arabic spoken in Bagirmi, notes that a continuum existed from a colloquial Arabic dialect to a pidginized form. Derendiger offers only fragmentary evidence of this continuum, con centrating on the standard dialect in the area. What examples he does give of the pidginized form resemble Turku and other Sudanic p/c's. So far as is known, there is no variety in present-day Chad or the Central African Republic known as Turku. Turku was, however, a vehicular language which the French colonial government found useful (hence Muraz's book), and in present-day Chad Arabic continues to survive in various pidginized forms (alongside the colloquial Chadian Arabic spoken by ca. 300,000 people), none of which, unfortunately, have yet been described. Future research will thus have to determine whether modern Arabic pidgins continue Turku or arose independently.7 2.3. Juba Arabic The / of the southern Sudan is often conventionally termed "Juba Arabic" (Watson 1984, Miller 1983, 1985). This term unfortunately appears to restrict the variety to the area of Juba, or to suggest some special historical link between it and the area of Juba. The language in fact is the lingua franca of the
134
Jonathan Owens
entire southern Sudan, and is spoken by southerners living outside the south (e.g. in the Khartoum area). Other terms however, like "southern Sudanese Arabic" or "lingua franca Arabic of the Southern Sudan", are either too unwieldy (second alternative) or misleading in their own way (colloquial Arabic dialects are also spoken in the South). Although it is mainly a second language, Mahmud (1983) reports that in Juba (and no doubt elsewhere) the language has gained native speakers; Miller (1985:155) reports that as many as 40% of the population of Juba (total ca. 200,000) now speak Juba Arabic as a mother tongue. According to Mahmud (1983:109), interethnic group marriage is the most important factor favoring the use of Arabic in southern households, with 40% of Mahmud's respondents from households of mixed marriages reporting Arabic as their first language, against 15% of those from households where the parents spoke the same language. The southern Sudanese languages spoken alongside Juba Arabic include Bari, Lotuxo, Moru, Madi, Zande, Dinka, and Nuer, as well as a host of smaller languages. Juba Arabic is structurally quite distinct from and mutually unintelligible with any variety of colloquial Arabic. This structural distinctness will be briefly illustrated in §4.4. below. The structural differences are reflected in the social status of the languages. Miller (1985, 1988:304) says that southern Sudanese consider colloquial Arabic and Juba Arabic different languages and that those southerners with a knowledge of both mix and manipulate the two according to the exigencies of the situation. The difference between the two can take on strong symbolic significance. Bell (1974:4), for instance, notes that one objective propagated by the Azania Liberation Front, the southern Sudanese movement which fought the central government up to 1972, was furthering of a common southern Sudanese nationality through the develop ment of the "pidgin Arabic spoken in the South". No important structural differences between pidginized and creolized varieties of Juba Arabic have been reported, although detailed linguistic and sociolinguistic data is lacking. Under influence from colloquial Sudanese Arabic a continuum has developed (Mahmud 1979, Miller 1985, 1986) rang ing from pidginized varieties on to colloquial dialects, a situation which renders an adequate characterization of Juba Arabic all the more difficult. Watson (1984, also 1989:108), for example, makes valuable attempts to render a description of the basilectal Juba Arabic variety; he presents a number of constructions with variant realizations, one closer to Sudanic
Arabic-based Pidgins and Creoles
135
colloquial Arabic than the other(s) (e.g. on pp. 17 and 39 on the passive), without any comment on the sociolinguistic implications of using one form or the other. No dialect surveys of Juba Arabic exist, though there are clearly dialect differences, and grammatical sketches and sociolinguistic surveys are limited mainly to the Juba variety (Nhial 1975, Mahmud 1979, 1983, Miller 1983, 1985, 1986, 1988, Watson 1984, 1989, Vincent 1986). 2.4. East African Nubi East African Nubi, also known as KiNubi (Kaye 1991:5), is spoken as a first language by 10,000-15,000 Kenyans and an unknown number of Ugandans. The number of Ugandan speakers was for many years considerably larger than those in Kenya, though because of their close association with the Amin regime (he himself is a Kakwa) many fled Uganda to Kenya at his downfall. There were once also East African Nubi speakers in Tanzania and Somalia; whether some still remain is unknown. In East Africa Nubi is almost exclu sively a first language, except perhaps in northern Uganda, where, because of proximity to the southern Sudan, Nubi functions to some degree as a lingua franca. Juba Arabic and Nubi are mutually intelligible, and Nubi, like Juba Arabic, is not mutually intelligible with colloquial Arabic. Almost all Nubi are bilingual in Swahili and Nubi, and it is not uncommon for them to speak other languages as well (Heine 1982:16). The Nubi are mainly urban dwell ers, the largest concentration in Kenya being in the Kibera area of Nairobi; smaller settlements are found in Eidama Ravine, Kisumu, Nakuru, and Mom basa. In Uganda the largest settlements were at Bombo and Kampala/Mengo, as well as smaller settlements in most other cities.
3.
The genesis of Sudanic / Arabic
Discounting possible, though presently unverifiable, influence from pidgin/ trade languages prior to 1820, the formative period of Sudanic / Arabic has to be placed between 1854 and 1888, the period between the opening of the South to large-scale northern influence and the expulsion of the Egyptian government from the South. I take 1888 to be the endpoint because at about this time East African Nubi became isolated from the Arabic of the South, so present-day similarities between East African Nubi and Juba Arabic (see §4.2
136
Jonathan Owens
below) must be accounted for by their common pre-1888 origin. That is, the Arabic of the South had crystallized in a stable pidgin or creole variety by 1888. Two important characteristics of Sudanic / Arabic are thus of general interest for creole studies (Carrington 1987:78): it stabilized (a) within little more than one generation and (b) within a social milieu where native speakers of Arabic constituted no more than 10-25% of the total population. To gain a closer insight into the genesis of Sudanic / Arabic in the 19th century (§3.4), it is necessary to look more closely at the language situation of the 19th-century South (§§3.1, 3.5), its demography (§3.2), and the social structure of the population (§3.3). 3.1. The languages of the South The southern Sudan is (and in the 19th century was) a linguistically heterogeneous area. Its languages fall mainly into two large clusters, Eastern and Central Sudanic (Greenberg 1966:85ff.). Niger-Congo is represented by Zande (Adamawa-Eastern branch). (1) Language of the southern Sudan (partial list):
While there is some mutual intelligibility among these languages — I have placed brackets around those which are to a degree mutually intelligible8 — the diversity is considerable.
Arabic-based Pidgins and Creoles
Languages of the Southern Sudan and adjoining regions
137
138
Jonathan Owens
3.2. The 19th-century population Thanks to the observations of the German explorer Schweinfurth (1918:50508), who traveled around the Bahr el-Ghazal area during 1868-1871, we have some rough estimates of the population in the region. The picture he paints is of a great upheaval. He divides the population into the following categories; the list in (2) is supplemented by the observations of other writers. (2)
A. Indigenous populations, not connected with trading camps: 190,000. B. Camp populations: (i) The traders' soldiers, officials, and translators: 5,000-10,000. Particularly important here are the Nile Nubians, who consti tuted the largest single northern group.9 Translators were mainly southern Sudanese from various ethnic groups born and brought up in the North, though there were also southerners, particularly in the early years of southern penetration, who were sent by chiefs of various southern groups to the camps of traders in the South for the express purpose of learning Arabic (Petherick & Petherick 1869, I:141,274). (ii) Permanently settled traders (jallaba): 2,000. (iii) Seasonally migrant traders: 2,000. These moved back and forth between Darfur and the South. They were mainly from western Sudanese Arabic tribes, like the Baggara. Gessi (1892:386), who was largely responsible for excluding northerners from the South after 1878, estimated that up to 20,000 jallaba had been active in the South. Being itinerant, this group had the least intense contact with the southern Sudanese. (iv) Slave soldiers (called faruch, basingers, or nyarakik): 5,000, from various southern ethnic groups. These were mainly offspring of the soldiers. They worked as gun carriers and comprised the main fighting force of the zeribas (traders' camps). During the period of government rule some of these troops became regular government soldiers, some continued to serve local traders, and others constituted a floating element owing allegiance variously to the govern ment/traders/Mahdists/local groups, according to the prevail ing political climate (see below).
Arabic-based Pidgins and Creoles
139
(v) Field and household slaves from southern Sudanese ethnic groups: 40,000. Every soldier had at least one slave; Gessi (1892:222), who admittedly was prone to some exaggeration, gives a ratio of 10 or 20 to one (slaves to soldiers). In this category belong concu bines. Writing during the period of Egyptian rule, Gessi reports in a letter (1892:433), "I must not forget to mention that 100 regular troops have been sent from Khartoum to Lado, another 100 will follow shortly. These men will have no women with them. As soon as they arrive in the interior they will each lay claim to 3 wives, 3 concubines, and many servants." (vi) Other camp followers: 1,000.10 Total camp population: 55,000-60,000. Of the ca. 250,000 inhabitants of Bahr el-Ghazal in 1870, 60,000, or about one quarter, belonged to the trading camps. Of those in the camps, only 9,00014,000 would have been native or near-native Arabic speakers (see n. 9 for Nile Nubians) — that is, (2) B.iii. Schweinfurth's estimates, it should be noted, describe the situation in 1870, sixteen years after the establishment of the first trading camp in the South. Two factors would have favored a gradual decrease in the percentage of native Arabic speakers. First, given the population growth through birth within the camps (see §3.4), the percentage of native Arabic speakers would have tended to decline. Secondly, the government policy (after 1878) of prohibiting northern traders in the South and of favoring the recruitment of native southerners to the army would have led to a relative decrease in the number of native Arabic speakers. Reports by writers after Schweinfurth tend to confirm this development. Junker (1891, III:526, III:572, I:462), for ex ample, writes that in the government post at Bor the Egyptian officer in charge had a household of 32 persons, including 6 concubines; a clerk had 28 in his household, a sergeant 12, and so on. The bulk of these household followers were from the South. Casati, an Italian who served the Egyptian government for ten years in the South, reported that he had 3,000 troops under him, two-thirds of whom were volunteers from the area or freed slaves (1891, I:49). Jephson, who accompanied the Stanley/Emin Pasha relief expedition and spent a number of months alone with Emin and his followers, reports (1890:48) that most of Emin's soldiers in the South did not want to accom-
140
Jonathan Owens
Emin to the coast because, though they were called 'Sudanese', they were really recruited from southern tribes — Dinka, Madi, Bari, Azande, Bongo, Makaraka, Mangbetu, Moru and others. The situation among Emin's followers in Uganda was similar. Furley (1959:316), for instance, reports that in their camps in the 1890s there were 10 camp followers — one can assume not mother-tongue colloquial Arabic speakers — for every soldier, and the soldiers themselves would not all have been native colloquial Arabic speakers. The ratios can be judged with some accuracy for the Sudanese serving the Germans in Tanganyika, from figures found in the German Reichstag Report of 1897-98 {Stenographische Berichte, pp. 248-49), which contains basic statistics on the colonial police and army forces; these are divided between Bantu and Sudanese troops, and the Sudanese are divided in turn into various subgroups.11 As of April 1898 there were 762 Sudanese troops, 466 soldiers, and 296 police. The soldiers were divided into 14 companies and the police were stationed around 13 different regions. In the preceding year 126 Sudanese had died, deserted, or (the large majority) been discharged, ostensibly because they were weary of service, though more likely because the German administration feared that a rebellion by the Sudanese in Uganda, begun in 1897, would spread into their own territory. A breakdown is given of the 762 Sudanese troops according to 50 different groups, only some of which clearly correspond to known ethnic or linguistic entities, and some of which would allow overlap with other groups in the count. I could identify about 35 of the groups, though the largest, termed "Sudanese" (= 134), is too general to ascertain what language they spoke. Subtracting this from the total, one is left with 628. Of these, 11 groups can be positively identified as possibly having a good command of colloquial Arabic. It is in fact unlikely, however, that all would have known Arabic well. Nubi clans are reckoned patrilineally according to tribal origin, so to be called a "Jaali" (an Arabic tribe), for instance, one would have needed only to have a "Jaali" father or grandfather. Nothing can be directly inferred about the language competence of the individual. Of the appellations used in the report, the following are all groups from northern Sudan or from Arabic countries: Nubians (= 107), Jaali (Ja'aliin, a northern Sudan Arab tribe, 19), Dagalani (?Dongola Nubians, 5), Barbarani (?from the city of Berber in n. Sudan, 7), Dongolani (Dongola Nubians, 3), Abbadi (Abbaabda, an Arabicized Beja group, 1), Chami (Shami, from Syria/Damascus, 1), Arabs (18), Turks (6), Kurds (2), and Syrians (1): The subtotal of these is 170, or 27% of the total (of
Arabic-based
Pidgins and
Creoles
141
628). This means that only 27% of the troops at most are likely to have had a good command of Arabic. When one recalls that each soldier had a household of 5-10 followers, most of whom would have been from the southern Sudan, with some perhaps from East Africa, the percentage of fluent Arabic speakers falls well below 10%. Allowing, conservatively, 5 followers for every soldier, 628 x 5 = 3140 + 628 = 3768, the percentage of potential Arabic speakers falls to 4.5%. If the Tanganyika situation is representative of that in Britishcontrolled East Africa, one can understand that post-creolizing processes would have been virtually nonexistent. 3.3. Social structure in the southern Sudan Junker presents evidence for an emerging tripartite social structure in the South by 1885. His best evidence pertains to the better-documented Egyptian government-controlled areas, though there is some indication that the situa tion was mirrored in other areas as well. At the bottom were local groups who had only transitory contact with the government camps and traders — at best brief trading exchanges or employment as porters, at worst victimization at the hands of a plundering expedition. At the top were the northerners, Sudanese, Egyptian, and Nubian officers and the northern traders who had managed to remain in the South. In the middle were the southern soldiers, whether born in the camps or recruited from the indigenous population. On the one hand they distanced themselves from the southern villagers, a process observed by Baker as early as 1864: All the best slave hunters, and the boldest and most energetic scoundrels were the negroes who had at one time themselves been kidnapped. These fellows aped a great and ridiculous importance. On the march they would seldom condescend to carry their own gun; a little slave boy invariably attended to his master, keeping close to his heels, and trotting along on foot during a long march, carrying a musket much longer than himself; a woman generally carried a basket with a cooking pot and a gourd of water and provisions, while a hired native carried the soldier's change of clothes and ox-hide upon which he slept. Thus the man who had been kidnapped became the kidnapper and the slave became the master... (Baker, quoted in Bradnum 1969:236)
On the other hand, according to Junker there was a tension between the northerners in the South and the southern black troops. He contends that the northerners tended to eschew the company of southerners, even if they were
142
Jonathan Owens
their own soldiers, and that the southerners in turn got along most easily with officers of black origin (1891, I:543), for instance those from the Nuba Hills (I:340, 356). This point is further substantiated by Gessi (1892:429), who observed that "during the war [against Sulaiman; see §1.2] we had clear proofs that it was the Arabs and not the natives who revolted... the negroes helped the Egyptian government." Although Gessi betrays an open anti-northern bias, his observation does bear witness to an Arab-southern Sudanese division. Tensions were greatest between the southern troops and lower ranking Nubians (Junker 1891, I:543, III:427, Emin/J. 180, 182, 190), particularly after the closing of the South to northerners and the repression of the slave trade (III:136; see Jephson 1890:110 for further evidence). In non-government-controlled areas Junker reports (1888:33) that, early on in the development of the network of trading camps, southerners living along the transport network joined, or were forced to join, the camps, forming an army (see (2) B.iv) which became the core of Sulaiman's troops (see n. 5 above), as well as participating in the Mahdist invasion of the South (Junker 1888:33, Emin/J. 1888:196, 199, 208). Junker observes (1888:33) that these were a "terror to the free indigenous groups" (cf. the Baker quote above). But Emin also reports (185, 208) that southerners in the army of the Mahdist leader in the South, Karamallah, found him excessively strict and fled his service. In 1884-85 there were at least two mutinies by southern troops in Mahdist employ in the South (Collins 1962:47). Here too evidence suggests a class of southerners distinct from both the local groups and the ruling elite. Junker's three-class analysis of southern society finds confirmation in the cleavages that became evident during the Mahdist invasion of the South. The indigenous population rose against the government forces, though they generally did not work in close alliance with the Mahdists (Junker 1888:145, 165 n. 2, 1891, III:458), and frequently fought them as well (Collins 1962:73, 75, 83, 114, 133). Aside from those who were officers of the Egyptian government, though even here there were exceptions, the Nubians and Arabs supported the Mahdists (Junker 1891, III:459, Jephson 1890:283). The south ern troops in government employ strongly opposed the Mahdists (Junker 1891, III:427, 478) and frequently had to fight the indigenous population as well. Irregular southern soldiers tended to change their loyalty according to the exigencies of the situation, sometimes supporting the government (Junker 1888:73, Emin/J. 214) or the Mahdists (Emin/J. 199, 205), sometimes joining
Arabic-based Pidgins and Creoles
143
the local groups against either (Emin/J. 185, 204, 220), and sometimes estab lishing their own local areas of control (Collins 1962:71). The three-cornered fight among indigenous groups, Mahdists/Northerners, and southern govern ment troops reflected three basic divisions in southern society. 3.4. Emergence of a / Four factors are frequently, if not always, relevant to the development of a /: speakers of diverse languages must be (i) mixed together (Whinnom 1971) and (ii) in daily contact, and (iii) either this mixture occurs in the presence of a dominant people using another language (the European-based creoles) or at least none of the languages of the various groups has a 'prestige' status and another language is available as the prestige model (Lubumbashi Swahili — Polomé 1968:14, Tok Pisin — Stoller 1979:75). Finally, (iv) the prestige target either cannot be learned by the subordinate groups, for what ever reason (when the prestige language is the language of the superordinate group), or, with the acquiescence of the superordinate group, a simplified version of a more complex variety becomes established.12 I now examine these points relative to the emergence of Sudanic / Arabic. These four factors obtained in the traders'/government camps that facili tated the emergence of a Sudanic Arabic-based /. First, the soldiers in the southern camps, as noted above, quickly gathered a slave and soldier popula tion taken from various southern ethnic groups. Many intermarriages in the camps would have been intertribal, a factor reported to favor creolization (Mahmud 1983; Polomé 1983:130 on Lubumbashi Swahili). Secondly, the contact among the different groups within the camps was regular. It had to be, for in daily life the camps were largely cut off from the local populations. The traders were universally hated by the southern tribes, even if alliances with local groups were regularly formed to further the activities of the traders and the local chiefs. Intensity of contact is further indicated by the presence of female slaves — male slaves more often being sent north for further exportation — who often became the wives of the northern soldiers. Nearly all writers remark on the large number of wives among the camp soldiers. Grant, for example, describing a seasonal trading camp at Faloro in northern Uganda (1864:332), reports that female slaves far outnumbered males, and Stanley, describing the situation over 25 years later (1890:136), writes that some of Emin's officers
144
Jonathan Owens
had between 18 and 100 women, children, and servants working in their homes. The women slaves had an important domestic function. Casati (1891, I:267) observes that they were the effective household heads, a situation Saghayroon reports as continuing among the East African Nubi (1981:86). When traveling, the government troops of the South never went without their wives and servants (Junker 1891, II:216, 254). The male offspring of these marriages were integrated into the socioeconomic milieu as soldiers, soon numbering as many as the northerners themselves (see §3.2 above). The slaves, concubines, and their offspring would undoubtedly have continued to speak a southern Sudanese language, though at the same time a language developed, the / Arabic, which would have been used between the different groups. Thirdly, the language of the dominant group was Arabic (§3.2). Finally, the question has to be posed why a / Arabic rather than a colloquial Arabic dialect developed in the South. This brings us to the fourth point, whether or not it was possible for colloquial Arabic to be learned. The answer is no, for demographic and social reasons. In the camps (traders', later government) a mixed population emerged which by 1880 is reported in many places to be roughly on the order of 10:1 in favor of native southerners vs. northerners, a large preponderance of non-colloquial-Arabic speakers. Fur ther hampering the acquisition of a colloquial Arabic dialect was the exist ence of a social stratification separating southerners from the northerners (see §3.3 above). Moreover, the important domestic role of southern women would have meant that the first language learned by children was either an indigenous language or a nonstandard form of Arabic spoken by the women and their offspring, or both. The reason for the emergence and stabilization of Sudanic / Arabic is to be sought in the crystallization of a class of southern ers who no longer belonged to indigenous southern groups, but who were also not part of the ruling northern elite. Parallels with the early development of the Atlantic creoles on the West African coast (Hancock 1986) are important, perhaps crucial. In both places the emergence of a stable / coincided with the formation of a social class that stood between an indigenous population and a dominant class. On the West African coast these were the creoles, who had a position between the whites and the African majority. In the 19thcentury southern Sudan the new class comprised the soldiers of the traders and the Egyptian government who had a station between the northern Arabs and Nile Nubians, on the one hand, and the surrounding African groups on the
Arabic-based Pidgins and Creoles
145
other. What I would hypothesize, then, is that for both the creoles and the southern Sudanese soldiers, the / was an expression of social class. 3.5. Language use in the 19th-century South Amidst all the figures and descriptions of life in 19th-century southern Sudan, it is difficult to determine exactly what languages/dialects were spoken by whom. Outside of the trading camps it appears that what little Arabic was used was restricted to some of the chiefs (Saghayroon 1981:4). In describing his travels in Equatoria, Casati, for example, reports on numerous occasions (1891, I:102, 144, 149, 196) conversations with local inhabitants in African languages, but none in Arabic, and he emphasizes the language and cultural differences between southerners and Arabs (1891, I:263). Both Schweinfurth and Junker emphasize the role of translators (Arabic-southern Sudanese languages) in their travels. Junker never traveled without translators for the local languages. As an extreme example, Jephson (1890:406), traveling south of Equatoria province, describes one chain of translation he needed to speak to a Munyoro chief that went from Swahili to Arabic to Alur to Lunyoro (and back). Within the camps both Arabic and local languages were used. Casati (1891, I:58) notes that among the southern soldiers divisions ran along ethnic lines, and Junker (1891, I:303) reports that no officer of the Egyptian govern ment would have marched against 'rebellious' southerners with a contingent composed solely of troops from one ethnic group, the danger of insubordina tion from a homogeneous group being too great. Junker further writes (1891, I:315) of large government caravans divided into discrete groups of Bari, Moru, Kakwa, Nubians, and others, and their wives and servants arranged "partly according to tribe, partly according to where their men were in the caravan". Ethnically-based divisions and solidarities such as the ones repre sented in these reports were no doubt symbolized inter alia by language differences (see also Jephson, cited in Stanley 1890:134). Crabtree (1913: 155) notes that among the East African Nubi there could always be found a speaker of any of the languages of the southern Sudan.13 Arabic was of course also important. Gordon, observing the traders' camps in Equatoria in 1875, wrote, "no nation could uproot Egypt from these lands even if they possessed them. Arabic [what variety is not stated] must be the language of these countries" (letter quoted in Gray 1961:113), and
146
Jonathan Owens
Mahmud reports (1983:13) that among Emin's southern soldiers who entered Uganda with him were those who spoke only Arabic. One reason for a mutiny by the Sudanese troops in Uganda in 1897 was that none of the British officers could communicate with their men in Swahili or Arabic (Furley 1959:321). It would appear, then, that in the camps Arabic was an important lan guage, though it had not displaced the local languages. What forms of Arabic were used is not entirely known. Colloquial Arabic was certainly used by Arabs and many Nile Nubians, and their use of dialect features is reflected today in the many traces of frozen morphological material found in East African Nubi (Owens 1985a:250 ff.). Among the southerners, knowledge of Arabic varied from place to place. Junker (1891, I:283), for example, writing about his 1876 travels, notes that around the newly-founded (1874) government station of Lado few of the Bari had learned Arabic, whereas just across the Nile near the longer-established Gondokoro they had "acquired the vehicularly necessary Arabic words". When describing the actual Arabic used by southerners, Junker is inconsis tent. In places he speaks of "vehicular" or "indigenized" ("negrisiertes" — 1891, I:172) Sudan Arabic, and elsewhere of two Azande princesses who spoke "some Arabic" (I:505). There were also southerners who had learned fluent colloquial Arabic, both in Khartoum (I:375) and perhaps in the South as well (III:150), though reading between the lines it would appear that refer ences to "Arabic" in the South by Junker and other European observers often refers to something other than a colloquial Sudanic Arabic dialect. One report explicitly confirms this. Stanley (1890:168) relates the complaints of his Zanzabari porters, who object to being called "slaves" by Emin Pasha's troops: "...we know enough Arabic to know what they mean, bad as their slang Arabic is...". However, direct comments on the Arabic used became available only in the 20th century, these mainly from colonial officers and missionaries. Crabtree (1913:154), writing about the languages of Uganda, speaks of a "modified form of Arabic" called Nubian, noting that it forms a lingua franca of sorts along the Nile valley. Other than Muraz (see §2.2), the only actual early textual data comes from Thorburn (1924).
Arabic-based Pidgins and Creoles 4.
147
Structural features
In this section I will first (§4.1) give a skeletal outline of Kenyan Nubi (KN), which will serve as a reference point for a comparative discussion of other Sudanic / Arabic varieties (§4.2) and the relationships between the / varieties, substratai languages (§4.3), and colloquial Arabic (§4.4). 4.1. sketch of Kenyan Nubi In this section I will give a skeletal outline of Kenyan Nubi (my own data). The choice of Kenyan Nubi (KN) is one of convenience: it is the bestdescribed Arabic-based / in print (Heine 1982) and the one I am most familiar with (Owens 1977). 'UN' marks examples from Uganda Nubi. 4.1.1. Phonology The phonemic inventory is given in (3). (3)
p b f v m w
t d s z n l,r
c j sh
k g
i e h
u o ,
ny y
Stress is distinctive, both lexically and grammatically; compare ína "we, us" and iná "there", amsúku "catching" and amsukú "caught". The main syllable types are CV and CVC. Sandhi phenomena are important in Nubi. East African Nubi words undergo a fair degree of alternation in fast speech. I briefly mention (nonexhaustively of course) three important phenomena here. First, VOWEL HAR MONY is a lexically-defined phenomenon. Leaving out of account a word-final vowel, a word can have only front vowels i, e or back vowels o, u; the low vowel a occurs in both contexts: tegíli "heavy", shunú "what", and áshrubul ásurubu "drink", but kábitu "write" and boyí "far", where word-final vowel does not fall under the rule. Vowel harmony is also a morphological rule. Two sets of morphemes can be distinguished. The morphemes gi- "progressive", bi- "future", and fi "at" have the form before back vowels (4a), while ta
148
Jonathan Owens
"of, ma "with", and na/la "to, for" have the form Ce before front vowels, Ca before a, and Co before back vowels (4b): (4)
a. fi bé "at home", fu sú "at the market" bi-jére "will run", bu-fútu "will pass" b. te bé "of home", ta rági "of a man", to sunú "of what"
This is more typical of UN than of KN. In close juncture a high vowel i, u of a morpheme-final CV syllable tends to be lost: (5)
gi ásadu→ gásadu "is asking", bi ágderi→ bágderi "will be able", bu-dugu íta → budugíta "you will be hit"
Before pronoun objects, final high vowels of verbs tend to be dropped, with the stress of the pronoun (always initial) remaining on the "new" syllable. In disyllabic verb stems the lexical stress of the verb is lost; in tri- or more syllabic stems the stem stress becomes secondary. (6) bu-dugú íta → bu-dugíta "will hit you" ásadu úmwon → ásadúmwon "asked them" gi-nadi ítakum → gi-nadítakum "are calling you (pl.)" Similar rules apply to further combinations with pronouns (e.g. preposition + pronoun, as in na íta → néyta "to you"). 4.1.2. Morphology In the verbal noun, the penultimate syllable is stressed, as in áshrubu "drink": ashrúbu "drinking", jére "run" : jére "running". The most significant deriva tional process is compounding. Compounds are syntactically similar to in alienable related nouns, though ta " o f can never occur between the nouns in a compound, as it can in a sequence of nouns, and in some compounds the first noun may lose its secondary stress; examples are mamá arús "bride", waja rás "headache", and gálam híbiri "pen". A predicative adjective (used only as a finite predicate) is formed from a transitive verb by the addition of ma-, e.g. séretu "tear" : maséretu "torn". Reduplication in Nubi has an intensive or distributive meaning; many, though not all, verbs can be reduplicated, and other word classes allow reduplication on a less regular basis. Examples are given in (7):
Arabic-based Pidgins and Creoles (7)
149
gátagáta "cut in pieces", kubu "spill" : kubkub "spill carelessly", jérejére "run all over", kurukum "yellow" : kúrukumkúrukum "dark yellow", guwóm guwóm "very fast" (UN).
The verb has the following structure: (8)
ANT
FUT
PROG (stem)
(kán) (bi) (gi) úwo áshrubu "he drank", úwo ma bi-gi-ashrubu "he won't be drinking", kán úwo bi-áshrubu "he would have drunk". The sequence is as stated, bi and gi are prefixes, while kan can occur before the subject. All combinations of prefixes + stem occur. Subject + bare stem = past, kán + bi = counterfactual. The stem has a single segmental form. The imperative is the bare stem (+ kum pl.): rúwa "go!", rúwa- "go (pl.)!". The negative imperative consists of máta(-kum) + stem, as in mata(kum) rúwa "don't (pl.) run!", or mata + V-k in pl., as in máta ágilib-kum "don't turn (pl.)!". 4.1.3. Clause structure The canonical clause structure is Adv S V O Adv. The basic typology is SVO; adverbial elements (time, place, manner) occur freely initially or, more com monly, at the end of the clause. A very few verbs allow SVOO (wedí "give, send", nadí "call, name"), and a larger number allow SVOAdj (e.g. so "make into something"). Generally an IO is marked by lalna. (9)
a. Kásulu gumásh dé nedíf. wash clothes the clean "Wash those clothes clean." b. Úwo wedí ána ája dé or Wedí he gave me thing the gave "He gave me that thing."
aja thing
dé na ána. the to me
The optative is formed with ke(de) "let, should" (UN also kelé): (10) a. Ké ítakum rúwa. let you/pl go "You (pl.) should go." b. Kelé gusú. let looked "Let it be looked for." (UN)
150
Jonathan Owens
An unspecified subject is indicated by means of a verb stressed on the final syllable with the subject deleted. This is generally translated as a passive, though intransitives occur as well: (11) a.
Úmwon gi-jere "they are running" : gi-jeré "someone is run ning". b. Adurb ína iná kamán ína sídu géliba ténna. shot us there again we closed hearts our "We were attacked, but we dug ourselves in." (UN)
Equational sentences have a zero verb in the present and kan in the past: (12) a. b.
úwo tájirilmálim. "he is rich/a teacher." úwo kán tájirilmálim. "He was rich/a teacher."
Existential sentences are marked by fi. This is used with all locative predi cates, and with a VN marked by the preposition fi it signals an action occurring at the time of speaking: (13) a.
Úwo fi fi bé/iná. he exist at home/there "He is at home/there." b. Bagará fí na ana. cows exist at me "I have cattle." c. Uwofi fi ashrúbu. he exist at drinking "He is drinking (right now)."
There also exists a "contingent be", which either represents a state of affairs as temporary or conveys an element of speaker subjectivity to the statement. (14) Úwo tájiri. "He is rich." vs. Úwo kún tájiri. "He is temporarily rich/ appears (to me) to be rich." Besides the use of existential sentences to indicate possession (second ex ample), there is a morphological verb "have", éndiléndis(i): (15) Úwo bi-éndi mótoka. he FUT-have car "He will have a car."
Arabic-based Pidgins and Creoles
151
The most important FOCUS construction consists of a left-topicalized constituent followed by the morpheme yd. This morpheme can in principle mark any constituent in any part of the sentence (one per major S), and used alone it serves as a presentative locative (in variation with the common yadál yadé). (16) a. Nyerekú fi fi skul yadá. child exist at school "The child is at the school right there." b. Úw wósul yá/yadá. he arrived "Here he has arrived." (UN) Ana biyo yd tinén. I
bought FOCUS
two
"I sold/bought TWO." As a focus marker the ya-marked constituent is normally S-initial, and any major constituent, including the verb, can be topicalized with yd. S-initial Q markers are generally marked by yd (focused constituent is in boldface): (17) a. Kátifu wáraga yá mdriya dé so. write letter FOCUS wife the did "Write a letter is what my wife did." b. Mára táki dé bará yá ána áj. wife your the only FOCUS I want "It is only your wife whom I want." Munú yd gi-já? who
FOCUS
PROG-come
"Who is coming?" Some DEPENDENT CLAUSE MARKERS are adverbial — báda "after", bakán "when", láman/náman/ládi "up to", and mishán/misén "because" — and one is conditional: kán "if. (18) a. Fi bakán dé ína dúrubu ládi ína ámurugu járman. at place the we shot until we removed Germans "At that place we attacked until we removed the Germans." (UN)
152
Jonathan Owens b. Bakán when úmwon they "When
úmwon ásuma kúra gi-kóre bumára dé they heard feet PROG-run much DEF jére. ran they heard our footsteps coming loud, they fled." (UN)
RELATIVE CLAUSES consist of allab + S. The markers allab are optional (though they usually occur). The DO extraction site is optionally marked by a resumptive pro, the locative site by fógo,14 and a definite relative clause is usually closed by dé.
(19) a.
Kéli (al) gi-gén jám séderi dé tayí. dog (that) PROG-stay near tree DEF mine "The dog which is sitting next to the tree is mine." (UN) b. bagara al íta áyinu (úwo) dé cow which you saw (it) this "the cow which you saw" c. ká ø ána kán kéytu gumás tayí fógo yá iná. had sewn clothes mine in-it FOCUS there place ø I "The place where I had sewn my clothes is over there."
The NOUN CLAUSE MARKERS are ø, zalja (gal), gal, kélkedé (after verbs of wishing and kútu "make"). (20) a. Ana árufu za gal kalám táki batál. I know that word your bad "I know that what you say is bad." b. Ana asuma min sabí tayí za (gal) azól nade I heard from friend mine that man that lísa má já. still not come "I heard from my friend that that man didn't come yet." c. Kalám al kútu anás náde ké jére bára áwaja. word which put people those that run outside wrong "The thing that made those people run outside is stupid." (UN) Verbs of perception and logó/ligó "find" take a verbal complement marked by
Arabic-based Pidgins and Creoles
153
(21) Ána shúf/ligó úmwon gi-álabu mpíra. I saw/found them PROG-play football. "I saw/found them playing football." 4.1.4. Negation The negative marker is ma in KN and ma or mafi in UN. It occurs immedi ately before the verb prefixes, starting with bi-, in KN, and before a preverb prefix or S-finally in UN. The existential negative is mafi (see §4.1.2 for the negative imperative). (22) Uwo má gi-já. he not PROG-came "He is not coming." = úwo gi-ja mafi (UN) 4.1.5. NP and Nominals Sequence: N Poss Adj Num RC Dem "all": (23) bagará máriya táki kubar-ín náde kúlu cattle woman your big-pl. those all "all those big cows of yours" In genitives, a possessor N is marked by ta. In inalienable relations the ta is optional. (24) silá ta rági dé "the weapon of this man", ída (ta) mulódo "hoe handle" Pronoun paradigms are given in (25): (25) Subject/Object SG
1st: ána 2nd: íta 3rd: úwo
Possessive
PLU
SG
PLU
ína/nína ítakum úmwon
tayí táki tó
téyna/ténna tákum tómwon
Nouns, adjectives, and demonstratives vary for number, though the only consistent agreement occurs between N and demonstrative. Moreover, only a few adjectives and not all nouns have separate sg/pl forms. The most common plural marker is -a; singular nouns ending in -a shift to final stress. Examples are bagara "cow" : bagará "cows", shéder "tree" : shederá "trees", ída "arm(s)", kabír "big", kubár or kubará "big (pl)".
154
Jonathan Owens
4.1.6. Prepositions, comparative, and question words The Nubi prepositions are gerí ma "near to", boyí ma "far from",fi "at", la/na "at/to" (human objects), bén "between", gidám "in front", wára "behind", jam "near", min "from", ládi "up to", té/téte "under", báda "after", kábla ta "before", fi rás "on top", fu ústu "in the middle", fijo (
Bírisi dé towíl min (or fútu) birisi táan. mat this long than (pass) mat other "This mat is longer than the other." (UN) b. Twaha ágara fútu anás dé kúlu. twaha read pass people DEF all "Twaha read more than all the others." (UN)
The Nubi question words are munu "who", shunu "what", mitén "when", kefin "how", atú/yatú "which" (nonhuman), wen "where", wonu "where" (predicative), kam "how many", le "why", malu "why" (mainly predicative), and músu "tag marker". 4.1.7. Coördination The coördination morphemes are aw "or", wála "or" (alternative Q), ma "and", "and", and Ø "and". S coördination is the same as for an NP, except that lakin "but" is used and ma is not generally used. (27) a.
Bágara tayi sakár bágara táan dé kebir. cow my small cow other DEF big "My cow is small and that one is big." (UN) b. Gumú masí. got up went "They got up and went." (UN) c. Asede sá tinén lakin úwo lisa já á. now hour 2 but he yet come not "It's two o'clock, but he hasn't come yet." (UN)
Arabic-based Pidgins and Creoles
155
4.2. The Sudanic / varieties There is linguistic evidence for postulating a common origin for the three varieties of the Arabic-based Sudanic /'s — Turku, Juba Arabic, and East African Nubi. The family tree appears in (28). (28)
Common Sudanic Pidgin/Creole Western Turku
Modern Chadian dialects (?)
Eastern Sudan (Juba Arabic)
East Africa (Nubi)
The basis for postulating a common Sudanic / is the existence of linguistic features shared among Turku, Juba Arabic, and East African Nubi, features which are either unique to these varieties or shared only with certain Sudanic Arabic dialects. In this section I will present an inventory of common traits which distinguish the three different varieties of Sudanic / as a group, and I will also note some of the differences among the varieties. I am concerned here to establish that there are general features which identify Sudanic / as a unitary entity, with local differentiation. In §§4.3-4.4 below I will discuss the structural relations among Sudanic / Arabic, Sudanic colloquial Arabic and southern Sudanese languages. The unique features set Sudanic / Arabic apart from Sudanic colloquial Arabic, which itself sub sumes a considerable degree of dialect variation. In the examples, M = Muraz 1929 for Turku and W = Watson 1984 for Juba Arabic. Here and below, "colloquial Sudanic Arabic" refers to the colloquial dialects spoken between Lake Chad and the Red Sea. 4.2.1. Features shared by all Sudanic / varieties Phonological features include Sudanic / Arabic k, as in muk "brain" (corre sponding to colloquial Arabic x, as in mux "brain"); lack of emphatics in Sudanic / Arabic, e.g. tíni "mud", vs. emphatics in colloquial Arabic, e.g. tiin "mud"; lack of pharyngeal consonants in Sudanic / Arabic, e.g. gata "cut" (cf. colloquial Arabic gaŧa'); s ~ sh variation in Sudanic / Arabic in certain words, e.g. shems ~ sems "sun", vs. colloquial Arabic sh (shams "sun"); lack of geminates in Sudanic / Arabic, e.g. kutu "put" (cf. colloquial Arabic xuŧŧu "put [IMP]"); e next to alveolars in Sudanic / Arabic, as in tehe "down", vs. a in colloquial Arabic (tahat "down"); and a qualitative five-
156
Jonathan Owens
vowel system throughout Sudanic / Arabic, with vowel length nondistinctive except for a few instances of an a/aa contrast (compare the Sudanese colloquial Arabic system, with three short vowels — a, i, and — and five long vowels: aa, , ii, oo, and uu). Sudanic / Arabic, as opposed to Sudanic colloquial Arabic, has a very reduced derivational system. Its only productive nominal affixal category is number (e.g. bagara "cow" : pl. bagará), shown generally by stressed final a (see §4.1.5). In Sudanic colloquial Arabic, by contrast, a noun is marked morphologically for number (sg., dual, pl.) and gender (masculine, feminine), and it takes possessive pronominal suffixes. In all of the Sudanic / Arabic varieties, subject and object pronouns are the same, but the possessive pro noun is distinct, e.g. Nubi ana "I, me" vs. tayí "mine" (see e.g. ex. 25 above). Furthermore, in all the Sudanic / Arabic varieties the possessive is formed from what is etymologically a relational marker + pro suffix: Nubi táki "your (sg.)" < bita'-ki "your-fem.sg.", Turku anaki < Chad Arabic hanaa-ki "yourfem.sg.". In all the varieties there is a single segmental verb form (vs. Sudanic colloquial Arabic, which has two basic stems, perfect and imperfect); and finally, a verbal prefix bi- ~ be- serves as a future marker in all Sudanic / varieties (see e.g. ex. 8 above). In the syntax, the existential particle fi occurs before all locative predi cates in Sudanic / Arabic; see ex. (13) above and (29): (29) a.
b.
Juba Arabic:
Turku:
Úwo fi le ana. (W 9) he exist at me "He is at my house." Káshim sandúk fí tít. (M 272) mouth box exist below "The opening of the box is underneath."
Compare Sudanic colloquial Arabic, as in (30), where the existential particle does not occur before a locative predicate: (30) Hu hinaak. he there "He is there." Moreover, fi = "at" or "to", vs. Sudanic colloquial Arabic, in which "at" but "to" = le:
-fi
Arabic-based Pidgins and Creoles (31) a.
b.
157
Nubi:
Hu masha fi be. he went to home "He went home." Sudanic colloquial Arabic: Masha le l-béet. went to home "He went home."
There are a number of vocabulary items unique to the Sudanic / varieties. In the examples below, the Sudanic / Arabic forms are from East African Nubi unless otherwise marked, and the Sudanic colloquial Arabic etymology is given in parentheses: (32) a. b. d. e. f.
ma "and" or "with" (W 18, M 271) (< ma' "with") filél "night" (M 136) (< fi l-leel "in the night") sabí "friend" (M 177, W 130) (< saahib-í "my friend") kore/kor/kori "shout, cry" (W 127, M 127) (< koorak) sú "feather, hair" (M 177, T 177) (< suuf "wool, feather") bíyo (East African Nubi), biyú (Juba Arabic), bí(h) (T) "buy or sell" (W 134, M 175) (< bii' "sell")
4.2.2. Dialect differentiation The diagram in (28) indicates that East African Nubi and Juba Arabic in the east form a branch, as opposed to western Sudanic / Arabic. A number of features separate the two groups. Moreover, I have found that East African Nubi and — at least in its basilectal form — Juba Arabic are mutually intelligible, though neither variety is immediately intelligible to a native speaker of a colloquial Arabic dialect. In this section the relevant examples for the eastern varieties can be found in the structural sketch of Nubi, and a reference is given to Watson (1984) when he gives the corresponding ex ample for Juba Arabic. The Eastern branch comprises Juba Arabic and East African Nubi, and the Western branch is represented by Turku. The Eastern varieties have ta as the associative particle, followed by a suffix, as in táki "yours", whereas in Turku it is ana, as in anáki "yours" (and see 24 above). In the Eastern varieties the numerals 11-19 are synthetic, as in talatásar "13", whereas in Turku they consist of two words, the word for "10" + "unit": asar taláta "13". The Eastern tense/aspect/mood (TMA) system consists of FUTURE bi- + PROGRESSIVE gi- + verb stem (see 8 above); in Turku the markers for future {be) and progressive (gaid) do not co-occur: Eastern
158
Jonathan Owens
bi-gi-jére "will be running", but Turku be jéri "will run" and gáid jéri "is running". The Eastern varieties have an unspecified subject construction, described in §4.1.3 above and illustrated in (11) (see W 39 for Juba Arabic). This construction is lacking in the Turku data. In the Eastern varieties, yá, yaú (Juba Arabic) serves as a focus particle (see 16 above and Watson 49). The reflex of this form is attested in Turku only in the presentative yaoda "here it is". The Eastern varieties have a morphological verb for "have" which is lacking in Turku (see 15 above and W 73). Examples of vocabulary differences are given in (33): (33) East African Nubi/Juba Arabic rúwa wedí "give" munú kefín mátakum + Verb
Turku masini "go" wadi "send" yatu "who" kikef "how" verb + mafi "don't (pl.)"
There are also a few features which link East African Nubi to Turku, against Juba Arabic. Both UN and Turku, for example, have S-final negatives (see §4.1.4). 4.2.3. The linguistic origins of the east/west division A number of the differences between the Eastern and Western varieties are attributable to the different colloquial Arabic dialects which contributed to their formation. The two main Sudanic Arabic dialects in the area are a western one (=WSA), spoken in NE Nigeria, northern Cameroon, Chad, and the western Sudan (Owens 1985b), and an eastern one (=ESA), the colloquial Arabic of the central, northern and eastern Sudan (Kaye 1976, Persson & Persson 1979, Reichmuth 1983). The difference in associative particles (ta/ ana, described above) stems from ESA bitaa' vs. WSA hana, as does the distinction between the analytic and synthetic "teen" numerals: ESA talatashar "13" vs. WSA 'ashara talaata. Similarly, many of the vocabulary differences parallel vocabulary differences between ESA and WSA, e.g. ESA wadda "send" vs. WSA wadda "give", ESA minú "who" vs. WSA yatú/munú "who", and ESA kefin "how" vs. WSA kekkeef/kefkef "how". A further source of difference stems from the stronger colloquial Egyp tian Arabic influence on the Eastern varieties. This influence goes back to the
Arabic-based Pidgins and Creoles
159
era of Egyptian rule in the South (ca. 1870-88) and is reflected, for instance, in the northern Egyptian pronunciation of g for Old Arabic j in certain words (cf. East African Nubi rági "man" and rujai "men", with Egyptian g in the singular and Sudanic j in the plural.) Egyptianisms are almost totally lacking in Turku.15 To summarize this section so far, three points favor the assumption of some sort of historically unitary origin for Sudanic / Arabic. First, as shown in §§1.2 and 3, there is a good deal of evidence indicating that there was a high degree of intercommunication between the traders' and government camps where Sudanic / Arabic developed in the 19th century. The demo graphic and social basis for the spread of features was present. Secondly, the number of shared features is sufficiently large that assuming an independent development for all of them in two or more Sudanic / Arabic dialects means postulating a very high degree of parallel development within a temporally and geographically limited period/area. Thirdly, there are some features, for instance the lexical correspondences and some of the shared morphological features, which I think can only be explained by direct transmission from a common parent. At the same time, there is a clear cleft between the Eastern (Juba Arabic, East African Nubi) and Western (Turku) varieties of Sudanic / Arabic. I would explain the differences in the following way. In the late 19th century the common ancestor of Turku, Juba Arabic, and East African Nubi devel oped in the camps of the southern Sudan. It had a high degree of uniformity, though already in this period dialect differences would have appeared, based on substratai influence (see §4.3) and the different Arabic dialects spoken in the area. In 1879 a large group of speakers of this / variety, perhaps over 10,000 people (see n. 5), left the southern Sudan with Rabeh for the Chari/ Logone River basin. They introduced Turku into this region, and Turku in turn was heavily influenced by the local Chadian Arabic, increasing the differences between the Western and Eastern Sudanic / varieties. 4.2.4. Variation in the Eastern varieties Within the Eastern varieties there is further variation. This can be broken down into two main categories, areal variation and variation due to decreolization. The former is, of course, a universal feature of Sudanic / Arabic, while the latter plays a role only in the Sudan. There is little information on dialect variation within Juba Arabic, though Bell (1974:10) shows that substratal influence plays a significant role.
160
Jonathan Owens
There are differences between Ugandan and Kenyan Nubi — or, at least, there were differences in 1972 and 1975 when I did my research. Since the fall of Amin, many Uganda Nubi have taken up residence in Kenya, as mentioned above; the Uganda/Kenya dialect differentiation is therefore less sharp than it formerly was. Here I will list some of the major differences; explanations for them remain to be worked out.16 In these examples, the Uganda dialect is represented by Kampala and Bombo, while the Kenya dialect is represented by Nairobi and Edlama Ravine. In Uganda, all words with sh have variants with s, as in shár ~ sár "month, moon"; in Kenya, sh alternates with s only in some words, e.g. shunú ~ sunú "what", but only shár "month, moon". Moreover, in some Uganda words with original sh only s now appears, e.g. súlu "carry", while Kenya retains sh (with or without an s variant), e.g. shúlu, the only Kenya form for "carry". The Uganda dialect tends toward open syllables, as in ásurubu "drink", tegtti "heavy", and fó "up, above", while Kenya retains more closed syllables, as in áshrubu, tegíl, and fok. Uganda has sentence-final negation, but Kenya has preverbal negation; see §4.1.5 above for examples. There are some lexical differences, e.g. Uganda áyinu vs. Kenya shúfu "see", Uganda bumára vs. Kenya zayd "very", and Uganda sókol vs. Kenya hája "thing", and some semantic differentiation as well, as in Uganda rásulu "arrive, reach" vs. Kenya rásulu "send someone". In addition to areal dialect variation, Sudanic / Arabic in the Sudan is subject to varying degrees of decreolizing influence.17 The limited amount of structural analysis of this process that has been carried out is based on the variety spoken in Juba (Mahmud 1979, Miller 1983, 1985). Mahmud (1979), for instance, describes a process by which the basilectal tense/mode verb markers bi/gi- are gradually reinterpreted as and/or replaced by the indicative marker b(i)- and the person/gender prefixes 'a-, ta-, ya-. 4.3. Substratai influences The genetic assignment of creole languages to their lexical source languages is not without its problems. Boretzky (1983) and Holm (1988), for example, emphasize the contribution of other languages (West African in the case of the Atlantic creoles), and in broader terms Thomason & Kaufman (1988) argue that genetic affinity can be claimed only when languages can be shown to be related at all linguistic levels. In light of these general considerations, as
Arabic-based Pidgins and Creoles
161
well as the historical genesis of Sudanic / Arabic (see §1.2 and §3 above), the relative contribution of non-Arabic languages needs to be ascertained. This question takes on general importance in the light of Givón's contention (1979:31) that endogenous p/c's, those whose speakers continued to live in the area where the substratai languages were spoken (Chaudenson 1977:264), show a greater degree of substrate influence than do exogenous ones. The extent and sense in which Sudanic / Arabic speakers are to be considered endogenous is problematic (see §§1.2 and 3.2), though it can be safely said that they underwent a smaller geographical displacement than did those of most oceanic creoles. In this section I explore some aspects of substratai influence, though the provisional nature of the analysis must be emphasized. An important practical problem can be noted. There are a large number of southern Sudanese languages which potentially contribute(d) to Sudanic / Arabic, and one should really survey all of them; however, detailed studies of most of these languages are lacking. I have therefore compared Sudanic / Arabic to Bari, for which Spagnolo's (1933) grammar is a valuable source of data. The choice of Bari is also relevant for historical reasons. Bari speakers, living around Gondokoro, the earliest settlement in the South, very likely would have played an important role in the formation of Sudanic / Arabic. Given the fact that a large number of slaves, porters, etc., were taken from centralSudanic-speaking peoples (e.g. Bongo and Mamvu), comparison with one of these languages would also be of interest. Given the large number of Nile Nubians in the southern Sudan in the 19th century, I looked at this language as well (Armbruster 1960). Nile Nubian is typologically different from Sudanic / Arabic, however (e.g. SOV rather than SVO, a pro-drop language, a suffixal case and number system, etc.), and I found very few common features. Among the phonological features found in varieties of Sudanic / Arabic and Bari but generally absent in colloquial Arabic are the sounds ny and gw (Spagnolo 1933:1, e.g. East African Nubi gworoko "cheeks"; gw is a rare sound in East African Nubi); ny is found in Sudanic colloquial Arabic, though its lexical reflexes in Sudanic / Arabic do not come from Arabic — e.g. nyerekú "boy" < Bari nyerku. The common East African Nubi alternation z ~ j, as in azu ~ aju "want" < Col Ar aayiz, is due perhaps to Bari substratai influence, since Bari has only j , not z; and similarly, the alternation s ~ sh attested to one degree or another in all Sudanic / Arabic varieties is related
162
Jonathan Owens
to the lack of sh in Bari. For the latter two features, note that the lack of a sibilant voicing opposition s ~ z or place contrast s ~ sh is common to many southern Sudanese languages (e.g. Shilluk, Dinka, and Lwo — Tucker & Bryan 1966:402). It is interesting to observe in this regard that the same types of sound substitution characteristic of Sudanic / Arabic are mirrored in Arabic loanwords in southern Sudanese languages (e.g. Bari said "witness" ← Ar shaahid, misan "because" ← Ar mishaan (cf. East African Nubi misén) — Spagnolo 1933:15, 16, and see Jakobi & Braiman 1990). The Sudanic / Arabic reflex of colloquial Arabic x is similarly to be understood in terms of substratai influence: few of the southern Sudanese languages have x. Bari substratai influence could also have played a role in the lack of geminate consonants, as Bari (and most southern Sudanese languages) lack these in all but a few morphophonologically-determined contexts. Morphologically Bari is, if anything,, further removed from Sudanic / Arabic than is colloquial Arabic. Syntactically there are some interesting points of convergence with Bari, where colloquial Arabic and Sudanic / Arabic diverge completely or partly, though these points of similarity do not appear to be systematic. They include the following. In colloquial Arabic, pronoun objects are indicated by bound suffixes which are distinctive from subject pronouns. In Sudanic / Arabic, subject and object pronouns are the same, but they differ from possessive pronouns (see ex. (25) above). A similar pattern is attested in Bari (e.g. nan "I, me" vs. lio "mine" — Spagnolo 1933:80-82). The use of (what in colloquial Arabic is) an independent pronoun for object function in Sudanic / Arabic, seen as a consequence of the general loss of bound affixes, is not unexpected. There is nothing in principle, however, which would have prevented extending the independent pronoun function to possessive contexts as well (cf. many Atlan tic creoles), and so the use of a different set of possessive forms may indicate Bari influence. In Sudanic / Arabic, male/female species are rendered by "name of species" + man/woman", e.g. bagara rágilmára "bull/cow", literally "cow man/cow-woman". A parallel is found in Bari noun + lalεt/narakwan "male/ female x" (Spagnolo 1933:28; also see Tucker 1940:132 for a similar con struction in central Sudanic languages). Bari ko = both "and", a nominal conjunct, and "with", an accompaniment preposition. One can compare here the Sudanic / Arabic morpheme ma, which serves as a nominal conjunct "and" as well as an instrumental/accompaniment preposition "with, by means of".18
Arabic-based Pidgins and Creoles
163
Kaye (1988:45) points out that the Sudanic / Arabic use of the preposi tion// to mark both locative "at" and motion towards "to" (see ex. 31 above) is paralleled by Bari i/ko "at, to" (Spagnolo 1933:56, 57). Finally, the East African Nubi noun-clause subordinator gal < "say" is paralleled by Bari adi < "say", which marks direct quotes (e.g. mukungu akulya adi "the sub-chief spoke, saying" — Spagnolo 1933:250). However, East African Nubi gal (or za gal) is a general noun-clause subordinator, whereas Bari adi is used only to mark direct or indirect speech. 4.4. Relations to colloquial Arabic The relation of Sudanic / Arabic to colloquial Arabic can be considered in terms of inheritance, convergence between colloquial Arabic and southern Sudanese languages, and differences. 4.4.1. Inheritance While Bari substratai influence is observable at a number of places in the structure of East African Nubi and Juba Arabic, the major source of Sudanic / Arabic structure is Arabic. This is evident in the lexicon, up to 90% of which is colloquial Arabic, and, in consequence of this, in the phonology, which is generally derivable via regular sound laws from colloquial Arabic sources (Owens 1985a, and see examples in §4.2.1 above). Elsewhere (Owens 1990, 1991) I have compared 24 grammatical features in East African Nubi, Sudanic colloquial Arabic, and two southern Sudanese languages — Bari and Mamvu (Vorbichler 1971) —and have shown that Nubi and Sudanic colloquial Arabic are structurally closer to each other than East African Nubi is to either of the southern Sudanese languages (Mamvu was least similar). The greater similarity of Nubi to colloquial Arabic can be illustrated with one example comparing Nubi, Bari, and Sudanic colloquial Arabic relative clause structure (Owens 1990:237). In East African Nubi the relative clause is optionally introduced by a morpheme al- or ab- and is optionally closed by the demonstrative dé. Objects and locative extraction sites are optionally marked by a pro-copy (for object) or by fogo (for locative). (34) Maria al-áana bi-áainu úo saabá dé éendis rági má. woman who-I FUT-see her tomorrow DET have man not "The woman whom I'm going to see tomorrow has no husband." (UN, Heine 1982:34 [his transcription])
164
Jonathan Owens
In Bari (Spagnolo 1933:89, 252), relative clauses are introduced by a near demonstrative, agreeing in number and gender with the antecedent. A morpheme gwon optionally follows the demonstrative, and a definite relative clause is closed by an agreeing proximal demonstrative. No pro-copies are left behind. (35) Ngutu lo (gwon) angödin do lo nye a b'urönit. man DEM (gwon) accused you DEM he is liar "The man who accused you is a liar." (Spagnolo 252) In Sudanic colloquial Arabic the relative clause is optionally introduced by al- and optionally closed by the proximal demonstrative dá. An object extraction site is generally marked by a resumptive pronoun agreeing with the antecedent noun. (36) Ar-raajil al-t-itham-a (da) kaddaab. the-man al-you-accuse-him (DEM) liar "The man who you are accusing is a liar." All three languages close the relative clause with a demonstrative, and all of them can introduce the relative clause with a discrete morpheme; in both Nubi and colloquial Sudanic Arabic, this morpheme is al-. Only Nubi and Arabic mark the extraction site with a resumptive pronoun. Thus, while there are similarities among all three languages, the closest parallels are found between East African Nubi and Sudanic colloquial Arabic. 4.4.2. Convergence There are also many Sudanic / Arabic constructions whose correspondence to Bari and Sudanic colloquial Arabic is more or less the same. Parallelisms might be due to chance convergence or they may be evidence for areal features. Where the Sudanic Arabic features are different from those in other Arabic dialects, borrowing from African languages into Sudanic colloquial Arabic may be indicated. Sudanic / Arabic, Bari, and colloquial Arabic all have basically a three-valued height contrast in vowels (high-mid-low). The placement of a question word in the sentence is also relevant here. In Sudanic / Arabic, Bari (Spagnolo 1933:92, 93), and Central Sudanic (Tucker 1940:148), ques tion words take the normal position of the constituent they mark: they are not moved to a sentence-initial position. In most Arabic dialects question words are S-initial. However, in the one Sudanese Arabic dialect for which I have
Arabic-based Pidgins and Creoles
165
detailed data, Nigerian Arabic, question words are usually not fronted (and similarly in Roth's examples for Abbeche Arabic — 1979:225). The presentative "here is, voilà" in Bari is formed from ma(s) + demon strative (Spagnolo 1933:86). This can be compared to Turku yao-da, Juba Arabic yau-da, and East African Nubi ya-dé/ya-dá (see 17 above) "here is". The morpheme may derive from the colloquial Arabic vocative particle yaa + the demonstrative da/de "this". The presentative in Nigerian and western Chadian Arabic is similarly formed from demonstrative + suffix, -wa/ya, dáwa/dí-ya "here it is (m./f.)". Finally, in some features Bari and Sudanic / Arabic converge, but this convergence might result from independent development. For instance, collo quial Arabic — unlike both Bari and Sudanic / Arabic — is a pro-drop language. In the case of Sudanic / Arabic, the necessity of expressing a subject independently of the verb most likely derives from the fact that most inflectional processes have disappeared, including (vis-à-vis colloquial Ara bic) the expression of pronominal elements in the verb. 4.4.3. Differences between Sudanic /'s and colloquial Arabic Thus far I have emphasized the relative similarity between Sudanic /'s and Sudanic colloquial Arabic. The significant structural differences between them should not be overlooked, however. In some cases the / varieties lack features present in Arabic and in others the opposite holds, as I will briefly illustrate here. A number of basic structural features are shared by all Arabic varieties (in this section simply "Arabic"), both dialects and the classical language, which are lacking in the / varieties. For instance, in Arabic the consonantally-based root plays a central role in the morphology, the verb is marked for person, number, and gender of the subject, nouns have masculine or feminine gender, there is a contrast between long and short vowels, and so on (see Owens 1989 for details). All of these features are missing in the / varieties. On the other hand, the / varieties have features absent in Arabic. Arabic does not have a morphological verb for "have" or contingent "be" (see exx. 14 and 15 above), or a tense/mood/aspect system like the /'s (see 8 above), and it does not have a passive like the p/c's (11). This last feature is particularly interesting. Recall that in Juba Arabic and East African Nubi a passive is formed by shifting stress to the final syllable. Speaking of the formation of passive verb forms, Keenan (1985:251) writes, "...we know of
166
Jonathan Owens
no language which forms passives by merely reduplicating or changing tonal marking on active transitive stems". Stress change on the verb stem, however, is precisely the formal mechanism used in the formation of the Sudanic / Arabic passive.
5.
Conclusion
Contact varieties of Arabic arose as soon as Arabs expanded out of Arabia in the 7th century. So far as our sources go, however, it was only in the 19thcentury Sudan that conditions were conducive to the development and con solidation of a stable pidgin and creole Arabic. This process has gone furthest, in the sense that creolized varieties have become well-established, in East African Nubi and Juba Arabic. Structurally, Sudanic / Arabic has inherited most of its lexicon and phonology and some of its syntax from Sudanic colloquial Arabic; for some features substratai influence, for instance from Bari, is probable. Because of the lack of mutual intelligibility with and the structural differences from colloquial Arabic, particularly in morphology, the Sudanic / varieties of Arabic are clearly to be considered different lan guages from colloquial Arabic.
Notes *
The following abbreviations are used: / = pidgin/creole, CA = Classical Arabic, KN = Kenyan Nubi, UN = Uganda Nubi, ESA = Eastern Sudanic Arabic, and WSA = Western Sudanic Arabic; "Sudanic" refers to the entire Sudanic area, from the Red Sea to West Africa, the Sudan to the contemporary political state; V = stressed vowel, ' = ' = sh = ƒ, and € = an emphatic or pharyngeal
1.
Versteegh might reply that decreolization could have narrowed original differences. While this is not a factor to be overlooked, it remains speculative; moreover, the difference between CA and a peripheral dialect such as Nigerian Arabic is not qualita tively different from that between CA and Cairene Arabic, although Nigerian Arabic has been far less subject to decreolization influence from standard CA. Sarah Thomason (p.c.) points out a possible parallel with Afrikaans, though many problems remain to be cleared up regarding the genesis of Afrikaans.
2.
Later kingdoms were established in Darfur and Waddai; in the far west of the Arabicspeaking region the kingdom of Kanem-Bornu traces its history back at least to the 9th century (Lange 1977). This kingdom was never ruled by Arabs, however.
Arabic-based Pidgins and Creoles
167
3.
This summary is based on Holt & Daly 1979, Gray 1961, and Collins 1971; data on language use in the South (see §3.5) has to be gleaned from the writings of 19th-century explorers and administrators. The following were used (with dates on which their writings on the South are mainly based): Schweinfurth (explorer, scientist, 1868-1871), Grant (explorer, 1863), Junker (explorer, scientist, 1877-1886), Gessi (administrator 18751879), Casati (administrator 1879-1888), Stanley (explorer, publicist 1888-89), Emin Pasha (= Edward Schnitzer, administrator, letters to Junker 1883-85; in Junker 1888, designated Emin/J.) and Jephson (explorer 1888-1889). Jephson's book is officially coauthored by Stanley, though the entire manuscript was written by Jephson.
4.
Austrian missionaries made the first attempt, in 1851, to found a permanent base near Gondokoro, but by 1858 it had to be withdrawn. The English vice-consul in Khartoum, Petherick, founded the first southern trading camp at Jur (near modern Wau) in 1854 (Junker 1888:31).
5.
The population around Sulaiman's (and Rabeh's) troops had a similar composition to the camps described by Schweinfurth (see §3.2). Junker (1888:74) reported that Sulaiman had 10,000 southerners among his troops. After Sulaiman's defeat, Gessi (1892:328) estimated that 3,000 jallaba, 5,000 women and slaves, and 2,500 southern soldiers and their followers (not less than 5,000) fled north with Sulaiman. A few loanwords from the 19th-century southern Sudan in present-day Nigerian Arabic confirm the presence of soldiers from the 19th-century Province of Bahr el-Ghazal in Rabeh's army. Examples are bazingir "foot soldier" (< bazingir), bánda kréj "term for Rabeh's soldiers" (Kresh = a central-Sudanic-speaking group located in the 19th century, as well as today, west of Wau on the southern Sudan-Central African Republic border, banda = an Adamawa-East-speaking group of central Central African Republic).
6.
Stanley (1890:353) gives the following figures: 555 of Emin's followers accompanied Stanley to the coast, including 44 officers, 90 wives, 107 children, 223 soldiers/orderlies, and 91 followers. Stanley does not say how many of these were northerners and how many southerners, though he does make clear that not all were Egyptians or northern Sudanese. Casati (1891, II:165) writes that in Wadelai, just before Emin left Equatoria, there were gathered a large number of administrators, most of them Egyptian.
7.
Jean-Claude Zeltner, who has worked in Chad since 1948, suggests (personal communi cation) that modern-day Chadian pidgins do continue Turku; however, they tend to be more influenced by standard dialects than was Turku.
8.
No systematic studies of mutual intelligibility are available; I have used Tucker (1940) for the Central Sudanic group, and I would like to thank Rainer Vossen for discussion of the language situation in the southern Sudan.
9.
Nile Nubians themselves spoke different dialects, Kenzi, Mahas, and Dongolese. How good their Arabic was we have no way of gauging. Even today there are no detailed studies of bilingualism among the Nile Nubians of northern Sudan. Both Armbruster (1960:15) and Rouchdy (1980:336) report that Nubians who travel outside their home land speak better Arabic than those who stay, and Lepsius (1880:Foreword) found Arabic useful for communicating with his Nubian consultants. Armbruster, who served as a colonial official among the Nubians, found that after a while his Nubian was better than the Arabic of the Nubians (1960:8). Junker, who was a seasoned traveler in the South,
168
Jonathan Owens refers to northern Sudanese as "Nuboarabs" (e.g. 1891, III:111, 520), which would suggest that Nubians had a good command of Arabic. Schweinfurth (1918:456) reports their Arabic to have been good.
10.
It is unclear in which category the porters belong. On an irregular basis they constituted a sizable portion of the camp populations. Schweinfurth, for example, speaks of a column of 250 soldiers and 3,000 porters from the Bongo tribe (1918:207), and elsewhere of 120 soldiers with 500 carriers (218). Grant (1864:345) speaks of the difficulty the traders had in recruiting porters who protested that they were not slaves. Nonetheless, the local populations often had no choice but to serve as commanded (Gray 1961:115).
11.
I am indebted to Dr. Angelike Jacobi for drawing my attention to this document.
12.
This latter situation does not apply in the case of the Sudanic p/c's, but it is, I suspect, relevant to the creoles in the Zaire River basin. Fabian (1986), for example, argues that Lubumbashi Swahili became established through the efforts of the white authorities in the Belgian Congo to encourage a lingua franca in the area.
13.
While investigating Nubi in Uganda in 1972 and in Kenya in 1975, I found no Nubi who still spoke any of the southern Sudanese languages.
14.
The functions oi fógo are multifarious: partitive, goal, place. In its relativizer function it can perhaps be seen as a lexical substitute for fi úwo "at/to him/her/it".
15.
Spagnolo (1933:15, 16) gives two Arabic loanwords in Bari which derive from Cairene Arabic, garama "pain" and sigin "prison".
16.
Kaye (1988:30) notes that in certain respects (e.g. the tendency towards open syllables) UN reflects basilectal Juba Arabic forms more than does KN. This merits further investi gation. But the source of the differences may also be sought in the varying substratai influences found among the original Nubi settlers of Uganda and Kenya, or alternatively in dialect differences already established in 19th-century Sudanic / Arabic (where substratai influence may have been partly responsible). Watson (1989:101), for instance, reports that Zande speakers are more inclined to add a word-final vowel in speaking Juba Arabic than are speakers of Juba Arabic in Juba. This difference mirrors that between UN (more V-final forms) and KN (more C-final). Similarly, Watson (1989:99) says that certain Jur Dinka speakers of Juba Arabic tend to drop final g, d, k, where these are retained in Juba. One example he gives, fok (Juba) vs. fo (Jur) "above" is identical to a KN/UN isogloss.
17.
I see no basis for Watson's claim (1989:97) that Juba Arabic derives much of its vocabulary from Classical Arabic. His only example, Juba Arabic ainu "see" < 'aayan, is most likely from Sudanese colloquial 'aayan "see, look at", attested in Hillelson 1925 (Roth-Laly 1969:328). The common CA word for "see" is ra'aa.
18.
Cf. also Dinka ke/kek = "and/with" (Nebel 1978:20).
Arabic-based Pidgins and Creoles
169
References Armbruster, Charles. 1960. Dongolese Nubian grammar. Cambridge: Cambridge Univer sity Press. Behnstedt, Peter, and Manfred Woidich. 1985. Die ägyptisch-arabischen Dialekte. Wies baden: Ludwig Reichert. Bell, Herman. 1974. "Pidgin Arabic and the language survey of the Sudan". Ms. Boretzky, Norbert. 1983. Kreolsprachen, Substrate, und Sprachwandel. Wiesbaden: Harrassowitz. Bradnum, F. 1969. The long walks: Journeys to the sources of the White Nile. London: Gollancz. Carrington, Lawrence. 1987. "The substance of creole studies: A reappraisal". Pidgin and creole linguistics: Essays in memory of John E. Reinecke, ed. by Glenn Gilbert, 77-92. Honolulu: University of Hawaii Press. Casati, Gaetano. 1891. Zehn Jahre in Äquatoria und die Rückkehr mit Emin Pasha. Bamberg: Buchner. Chaudenson, Robert. 1977. "Towards the reconstruction of the social matrix of creole languages". Valdman 1977.259-76. Collins, Robert. 1962. The Southern Sudan. New Haven: Yale University Press. Collins, Robert. 1971. Land beyond the rivers: The Southern Sudan 1898-1918. New Haven: Yale University Press. Corriente, Frederico. 1976. "From Old Arabic to Classical Arabic through the PreIslamic Koine". Journal of Semitic Studies 21.62-98. Crabtree, O. 1913. "The languages of the Uganda Protectorate". African Affairs 13.15266. Derendiger, R. 1912. "Notes sur le dialecte Arabe du Tchad". Revue Africaine 56.341-58. Emin Pasha (Edward Schnitzer). 1888. "Correspondence with Junker". Junker 1888 (cited as Emin/J.). Fabian, Johannes. 1986. Language and colonial power. Cambridge: Cambridge Univer sity Press. Fück, Johann. 1950. Al- Arabiyya. Berlin: Akademie Verlag. Furley, . 1959. "The Sudanese troops in Uganda". African Affairs 58.311-28. Gessi, Romolo. 1892. Seven years in the Sudan. London: Sampson, Low and Marston. Givón, Talmy. 1979. "Prolegomena to any sane creology". Hancock 1979.3-36. Goodman, Morris. 1986. Review of Versteegh 1984. Journal of Pidgin and Creole Linguistics 1.165-70. Grant, James. 1864. A walk across Africa. London: Blackwood. Gray, Richard. 1961. A history of the Southern Sudan 1839-1889. London: Oxford University Press. Greenberg, Joseph. 1966. The languages of Africa. Bloomington: Indiana University Press. Hancock, Ian, ed. 1979. Readings in creole studies. Ghent: Story Scientia. Hancock, Ian. 1986. "Domestic hypothesis, diffusion and componentiality". Substrata versus universals in creole genesis, ed. by Pieter Muysken and Norval Smith, 71-102. Amsterdam: John Benjamins.
170
Jonathan Owens
Heine, Bernd. 1982. The Nubi language of Kibera: An Arabic creole. Berlin: Dietrich Reimer. Hillelson, S. 1925. Sudan Arabic: English-Arabic vocabulary. London: Sudan Govern ment. Holm, John. 1988. Pidgins and creoles, vol. I. Cambridge: Cambridge University Press. Holt, Peter, and Martin Daly. 1979. The history of the Sudan. London: Wiedenfelt and Nicholson. Jakobi, Angelika, and Ali Braiman. 1990. "Sudan Arabic loanwords in Kaduru, a Kordofanian Nubian dialect". Tradition, Migration Notstand: Themen heutiger Sudan ethnographie, ed. by . Streck, 55-68. Göttingen: Edition Re. Jephson, A. Mounteney, and Henry Stanley. 1890. Emin Pasha und die Meuterei in Äquatoria. Leipzig: Brockhaus. Junker, Wilhelm. 1888. Der Sudan unter ägyptischer Herrschaft. Leipzig: Brockhaus. Junker, Wilhelm. 1891. Reisen in Afrika. Vienna: Hölzel. Kaye, Alan. 1976. Chadian and Sudanese Arabic in the light of comparative Arabic dialectology. The Hague: Mouton. Kaye, Alan. 1985. "On the importance of pidgins and creoles for historical linguistics". Diachronica 2.201-30. Kaye, Alan. 1988. "The history and development of Juba and Ki-Nubi Arabic". Ms. Kaye, Alan. 1991. "Peripheral Arabic dialectology and Arabic pidgins and creoles". Languages of the World 2.4-16. Keenan, Edward. 1985. "Passives in the world's languages". Language typology and syntactic description, vol. I, ed. by Timothy Shopen, 243-81. Cambridge: Cambridge University Press. Lange, Dierk. 1977. Le Diwân des Sultans du (Kânem)-Bornô: Chronologie et Histoire d'un Royaume Africain. Wiesbaden: Steiner. Lepsius, Richard. 1880. Nubische Grammatik. Berlin: Hertz. Lethem, G. 1920. Colloquial Arabic, Shuwa Arabic dialect of Bornu, Nigeria and of the region of Lake Chad. London: Crown Agent for the Colonies. Mahmud, Ushari. 1979. Variation in the aspectual system of Juba Arabic. Washington, DC: Georgetown University dissertation. Mahmud, Ushari. 1983. Arabic in the Southern Sudan: History and spread of a pidgincreole. Khartoum: Khartoum University Press. Maigret, Julien. 1931. Afrique Equatorial Français. Paris: Société d' Edition Géo graphique, Maritimes et Coloniales. Miller, Catherine. 1983. "La Juba Arabie, une lingua franca du Soudan Meridional". Materiaux Arabes et Sudarabiques 105-18. Paris: Paul Geuthner. Miller, Catherine. 1985. "Un exemple d'évolution linguistique: Le cas de la particule ge en Juba Arabie". Matériaux Arabes et Sudarabiques 156-66. Paris: Paul Geuthner. Miller, Catherine. 1986. "Analyse des usages de l'arabe dans une communauté non arabophone: Le cas Tribunal Coutumier de Juba". Actes du TV Colloque International sur l'Acquisition du Langage, ed. by H. Blanc, M. Douaron and D. Véronique, 296306. Paris: Didier. Miller, Catherine. 1988. "Langues et Tribunaux en Equatoria". Materiaux Arabes et Sudarabiques 23-58. Paris: Paul Geuthner.
Arabic-based Pidgins and Creoles
171
Muraz, Gaston. 1929(?). Vocabulaire du Patois Arabe Tchadien ou 'Tourkou' et les Dialectes Sara Madjinngaye et Sara Mbaye. Nancy: Charles Lavauzelle. Nebel, Arthur. 1978. Grammatica e Dizionario Dinka. Bologna: Editrice Missionaria. Nhial, Abdon. 1975. "Kinubi and Juba Arabic: A comparative study". Directions in Sudanese linguistics and folklore, ed. by S. Hurreiz and H. Bell, 81-94. Khartoum: Institute of African and Asian Studies. Owens, Jonathan. 1977. Aspects of Nubi syntax. London: London University dissertation. Owens, Jonathan. 1985a. "The origins of East African Nubi". Anthropological Linguis tics 27.229-71. Owens, Jonathan. 1985b. "Arabic dialects of Chad and Nigeria". Zeitschrift für arabische Linguistik 14.45-61. Owens, Jonathan. 1989. "Zur Piginisierung und Kreolisierung im Arabischen". Afrika und Übersee 72.91-107. Owens, Jonathan. 1990. "East African Nubi: Bioprogram vs. inheritance". Diachronica 7.217-50. Owens, Jonathan. 1991. "Nubi, genetic linguistics, and language classification". Anthro pological Linguistics 33.1-30. Persson, Andrew, and Janet Persson. 1979. Sudan Colloquial Arabic for beginners. Horsley's Green, England: Summer Institute of Linguistics. Petherick, John, and Katherine Petherick. 1869. Travels in Central Africa. London: Tinsley Brothers. Polomé, Edgar. 1968. "Lubumbashi Swahili". Journal of African Languages 7.14-28. Polomé, Edgar. 1983. "Creolization and language change". The social context of creolization, ed. by Ellen Woolford and William Washabaugh, 126-36. Ann Arbor: Karoma. Prokosch, Erich. 1986. Arabische Kontaktsprachen. Graz: Institut für Sprachwissen schaft. Rabin, Chaim. 1951. Ancient West Arabian. London: Taylor's Foreign Press. Reichmuth, Stefan. 1983. Der arabische Dialekt der Shukriyya im Ostsudan. Hildesheim: Georg Olms. Roth-Laly, Arlette. 1969. Lexique des Parlers Arabes Tschado-Soudanais. Paris: CNRS. Roth-Laly, Arlette. 1979. Esquisse Grammaticale du Parler Arabe d'Abbeché. Paris: Paul Geuthner. Rouchdy, Aleya. 1980. "Languages in contact: Arabic-Nubian". Anthropological Lin guistics 22.334-44. Saghayroon, Ibrahim. 1981. The Sudanese Muslim factor in Uganda. Khartoum: Khartoum University Press. Samarin, William. 1955. "Sango, an African lingua franca". Word 11.254-67. Schweinfurth, Georg. 1918. Im Herzen von Afrika. Leipzig: Brockhaus. Spagnolo, A. 1933. Bari grammar. Verona: Mission Africane. Spitaler, Anton. 1953. Review of Fück 1950. Bibliotheca Orientalis 3.144-48. Stanley, Henry. 1890. In darkest Africa, II. New York: Scribners. Stenographische Berichte über die Verhandlung des deutschen Reichstages, 1897/1898. 10 Legislaturperiode, 1 Session, 1 Anlagen-Band. Stoller, Paul. 1979. "Social interaction and the development of stabilized pidgins". Hancock 1979. 69-79.
172
Jonathan Owens
Thomason, Sarah, and Alaa Elgibali. 1986. "Before the lingua franca: Pidginized Arabic in the eleventh century A. D." Lingua 68.317-49. Thomason, Sarah, and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thorburn, D. 1924. "Sudanese soldiers' songs". Journal of the African Society 24.314-21. Tosco, Mauro, and Jonathan Owens. 1993. "Turku: A descriptive and comparative study". Sprache und Geschichte in Af rika 14.177-267. Tucker, Archibald. 1940. The Eastern Sudanic languages. London: Oxford University Press. Tucker, Archibald, and Margaret Bryan. 1966. The Non-Bantu languages of Northeastern Africa. London: Oxford University Press. Valdman, Albert, ed. 1977. Pidgin and creole Linguistics. Bloomington: Indiana Univer sity Press. Versteegh, Kees. 1984. Pidginization and creolization: The case of Arabic. Amsterdam: John Benjamins. Vincent, George. 1986. "Juba Arabic from a Bari perspective". Current approaches to African linguistics, ed. by Gerrit Dimmendaal, 71-78. Dordrecht: Foris. Vorbichler, Anton. 1971. Die Sprache der Mamvu. Glückstadt: Augustin. Watson, Richard. 1984. Juba Arabic for beginners. Juba: Summer Institute of Linguis tics. Watson, Richard. 1989. "An introduction to Juba Arabic". Occasional Papers in the Study of Sudanese Languages 6.95-117. Whinnom, Keith. 1971. "Linguistic hybridization and the 'special case' of pidgins and creoles". Pidginization and creolization of languages, ed. by Dell Hymes, 91-115. Cambridge: Cambridge University Press. Zeltner, Jean-Claude. 1988. Les Pays du Tchad dans la Tourmente, 1880-1903. Paris: Harmattan. Zeltner, Jean-Claude, and Henri Tourneux. 1986. L'Arabe dans le Bassin du Tchad. Paris: Karthala. Zwettler, M. 1978. The oral tradition of Classical Arabic poetry. Columbus: The Ohio State University Press.
Kitúba* Salikoko S. Mufwene University
1.
of
Chicago
Introduction
This paper makes a number of observations on Kitúba,1 a creole of Central Africa spoken in the geographical area extending from as far east as the Kwilu River (in the western part of Zaire) to as far west as the Atlantic Ocean, thus including the southern part of the Republic of the Congo (see map). It is not clear how far into Angola Kitúba is spoken. The paper is intended to complement previous writings on the identity, ethnographic status, and gen esis of the language, notably Fehderau (1966), Samarin (1982, 1990), and Mufwene (1988a). Some points are then made on its normalization (i.e. development of norms) and divergence into distinct dialects, taking issue with the currently prevalent characterization of creolization as nativization of a pidgin. The last part of the paper formulates several research questions for the future which might shed light not only on the development of Kitúba but also on creole genesis in general, especially regarding factors which determine how creoles select their structural features from among the language varieties out of whose contact they have been formed. Accordingly, the paper is organized into five main parts which correspond to the above questions.
2.
Names and identity
Kitúba, short for Kikóngo-Kitúba, has also been identified by several other names, including Kikóngo ya letá (shortened to Kiletá) "the public adminis-
174
Position of the Kongo Kingdom in the 17th-18th
Salikoko S. Mufwene
centuries
Kitúba
175
tration's Kikongo"; Kikóngo ya bula-matádi or bula-matári (shortened to Kibula-matádi or Kibula-matári) "the colonial agent's Kikongo" (literally, "the stone-breaker's Kikongo" — see below); Mόno kutúba (whence Kitúba) "I speak/say";2 and Ikéle vé "be not; it isn't true". All these names are quite descriptive of the language's identity and/or origin in relation to its lexifier, ethnic Kikóngo.3 Kikóngo ya letá is due to the fact that the language was adopted by the Belgian colonial administration in the "Leopoldville province", west of the then-Belgian Congo (now called Zaire; see map). Kikóngo ya bula-matádi alludes to the time when the railroad connecting Matadi (Zaire's main harbor) to Kinshasa was built (1891-1898). The tracks run across mountains, which required blasting rocks (matádi) during the construction. The language of the labor force, part of which was brought from West Africa (Senegal, Sierra Leone, and the Gold Coast) and from east of the Kwango River (outside the Kongo territory), was Kitúba. In the Belgian Congo, it was used also by the colonial administration (called both letá — from French l'état "the state" — and bula-matádi4), which recruited the local and other national and foreign labor for the railroad and other public develop ment projects. Kikóngo ya letá and Kikóngo ya bula-matádi then became alternative names for the new language variety based originally on the Kimanyánga dialect of ethnic Kikóngo and used by the colonial agents and their helpers. The terms basically mean "the administrators' way of speak ing". 5 Both Mono kutúba and Ikéle vé have to do with the less agglutinating nature of Kitúba's verbal morphosyntax in comparison with the canonical Bantu system (Mufwene 1988a).6 Most obvious are the absence of a subjectagreement prefix on the verb, the lack of object pronominal prefixes, and the absolute reliance on independent and morphologically invariant pronouns as well as on their syntactic positions to determine the subject and objects in a sentence. Aside from derivational extensions, the verb form varies only according to tense, as illustrated in (1): (1)
Móno Ø-mon-áka yándi ye yándi 0-mon-áka móno. me AGR-see-ANTER him/her and him/her AGR-see-ANTER me "I saw him/her and he/she saw me."
Other names of interest are Kikwángo and Kikóngo. The first has been used especially in association with the Christian missions in the KwangoKwilu district, thus identifying missionaries as important agents in the spread
176
Salikoko S. Mufwene
of Kitúba. The second name has been commonly used in the same area (where ethnic Kikóngo is not spoken) as the neutral term without any metalinguistic function. The term Kitúba, adopted in this paper (after Fehderau 1966) instead of any of the other names, is mostly academic. 7 In sum, the above names suggest changes in the structure of the lexifier, or they refer to the kinds of contact situations in which Kitúba either origi nated or spread, or they identify a geographical area where Kitúba serves at least as a lingua franca. As shown in Mufwene (1988a, 1989a, 1994, after Fehderau 1966), Kitúba's structures are often different from those of the varieties globally referred to as ethnic Kikóngo. The following selected comparisons illustrate this point, comparing features of Kitúba with those of the composite variety of Kikóngo that Lumwamu (1973) calls "diasystem." To start with, Kitúba has a predominantly phonological tone or accent system, instead of the lexical and/or grammatical tone system attested in ethnic Kikóngo and in most Bantu languages. To be sure, as observed by Mufwene (1989a) and Ngalasso (1991), Kitúba has an important proportion of polysyllabic words either with only low tones (e.g. muntu "person", munoko "mouth, opening", dikulu "leg, foot", mbala "time" as in "five times") or with a high tone on the last syllable or on both the penultimate and last syllables (e.g. nzilá "way, road", mbalá "yam", dilálá "citrus fruit", mabélé "breasts, milk"). However, the majority of the words have only one high tone borne by the penultimate syllable (e.g. dísu "eye", kwísa "come. IMPERATIVE", kapíta "foreman", bába "mute"). Derived words and conju gated verbs are especially subject to this tone placement rule, e.g. ku-sál-a "to work" ~ sal-á(k)a "worked.ANTERIOR", ku-sad-ís-a "help" ~ n-sád-i "helper", ku-pés-a "give" ~ ku-pes-íl-a "give on the behalf, or for the benefit, of".8 Moreover, unlike in ethnic Kikóngo, tone alone may not be used for tense/mood/aspect distinctions. Regarding segmental phonemes, some pho netically complex segments that appear in Lumwamu's (1973) diasystem of ethnic Kikóngo are not attested in Kitúba, e.g. /pf, ts, dz/. Compared to Lumwamu's diasystem, Kitúba's verb forms are simple; Kikóngo's verb forms, such as the one in (2a) are expressed as (2b) in Kitúba: (2)
a.
(Yándi) ka-ku-zól-elé. /She AGR-you-like-NEAR.PERFECT "/She likes you."
b.
Yándi/*Ø zól-a /She like-ENDING
ngé. you
Kitúba
177
Note especially that Kikóngo's diasystem allows null subjects, whereas Kitúba does not allow them. In (2a) yándi is used only for emphasis or contrast, whereas in Kitúba it is a required marker of person and number. A correlate of this phenomenon is the loss of subject-verb agreement in Kitúba, which was mentioned above. In losing the pronominal object prefix (compare 2a with 2b), Kitúba has also lost the reflexive, illustrated in (3); as in (2), (3a) is Kikóngo and (3b) Kitúba: (3)
a.
Pételo
ka-ki-zól-elé.
Peter
AGR-Self-like-NEAR.PERFECT
"Peter likes himself." b. Pételo Peter
zól-a yándi mosi. like him one
Instead of the reflexive Kitúba uses, in (3b), an emphatic pronominal con struction consisting of the invariant third person singular (3sg.) pronoun yándi and the quantifier mósi. The same combination is also used in contexts which are not reflexive, as below: (4)
a.
Yándi kwis-á(k)a yándi mósi. /She come-ANTER[IOR] him/her one "He/She came alone/by him-/herself."
b.
Yándi mósi muntu kwis-á(k)a. He one person come-ANTER "He himself came."
Yándi muntu kwis-á(k)a. "He/She [is] the one [that] came."
A concomitant of the loss of subject-verb agreement in Kitúba is the loss of the head-modifier agreement marker in noun phrases. Unlike in Kikóngo (per Lumwamu's diasystem), an invariant connective ya occurs between the head noun and the modifying phrase, including relative clauses, as illustrated in (5). Likewise, the form of the demonstrative yáyi remains invariant, unlike in Kikóngo. The Kitúba examples are second in each pair (a', b', and so on); class markers are identified with "cl. 1", "cl. 2", etc.
178
Salikoko S. Mufwene (5)
a.
mw-ána ú-mbote/w-á Cl. l-Child AGRl-good.NOUN/AGRl-CONN "good child"
mbote good.NOUN
a'. mw-ána ya mbóte cl.l-child CON(NECTIVE) good.NOUN b.
b-ána bá-mbote cl.2-child AGR2-good.NOUN "good children"
b'. b-ána ya mbóte. c.
Kí-íma ki-a cl.7-thing AGR7-this "this thing"
'. K-íma yáyi cl.7-thing this "this thing" d. Bí-íma bi-a cl.8-thing AGR8-this "these things" d'. B-íma yáyi cl.8-thing this e.
Kí-íma ki-a ka-ta-sál-a Cl.7-thing AGR7-this 3sg.-DUR-do-ENDING ka-ki-zéy-i ko. 3sg.-AGR7-know-PERF NEG "The thing that he is doing, he does not know."
e'. K-íma (yáyi) ya yándi ké(l)e cl.7-thing (this) CON he/she be zába yó vé. know it NEG
sál-a yándi do-ENDING he/she
Although the noun-class system is preserved, several nouns that formerly varied their prefixes in classes 9 and 10 have been reassigned to classes la (with a null prefix) and 2. Thus the nouns ndeke "bird", mbúta "older, elder",
Kitúba
179
and 'nzo "house, building" form their plurals as ba-ndeke, ba-mbúta, and ba'nzo, with the nasal prefix now interpreted as part of the stem. Although this is part of a more general process discussed in Mufwene (1979) — see also Bokamba (1977) for Lingála — the process is partly due to loss of the agreement prefixes, which in the specific case of classes 9 and 10 (i-/z-), helped distinguished them. As observed in Mufwene (1988a, 1990a), Kitúba uses basically the same tense/aspect distinctions as ethnic Kikóngo. The only difference lies in how it expresses them. Kitúba relies more on periphrasis, using only two tense suffixes, the ANTERIOR-á(k)a and the NARRATIVE-Ø, whereas Kikóngo relies more heavily on agglutination and grammatical tone, although the Kikóngo system is not exclusively agglutinative. The examples in (6) illustrate some of the ways in which Kitúba has simplified Kikóngo's tense/aspect morphologi cal system, for instance in regularizing discontinuous tense/aspect marking in the following way: aspectual markers precede the verb stem, whereas tense markers follow it. Note that tense is suffixed in Kitúba but prefixed in Kikóngo, when it is expressed segmentally. Again, the second member of each pair is Kitúba. (6)
a.
Tu-di-idí. we-eat-PERFECT "We have eaten."
a'. Béto we b.
Tu-á-di-idí. We-ANTERIOR-eat-PERFECT "We have eaten."
b'. Béto we
mé(n)e 'di-a. finish eat-ENDING
mé(n)e di-á(k)a. finish eat-ANTER
Tu-di-ánga. we-eat-HABITUAL "We eat."
c'. Béto we
ké(l)e
di-á(k)a.
be
eat-?ANTER
Salikoko S. Mufwene
180 d.
Tu-á-di-ánga. We-ANTER-eat-HABITUAL "We used to eat."
d'. Béto we e.
vand-á(k)a sit-ANTER
di-á(k)a eat-ANTER
Tu-tá-di-á. we-DURATIVE-eat- ? "We are eating."
e'. Béto ké(l)e 'di-a. f.
Tu-di-á. we-eat-FUTURE "We will eat."
f.
Beto ata we
'di-a.
FUTURE eat
As shown in Mufwene (1988a, 1991c), the periphrastic constructions themselves have been selected from Kikóngo, most likely from Kimanyánga, whose role in the formation of Kitúba is discussed in §4 below. The free aspectual morphemes mé(n)e "finish", ké(l)e "be", and vand-a "sit" are all attested in Kimanyánga, with more or less the same syntactic distributions. What is particularly significant here is morphosyntactic regularity of the tense/ aspect system, compared to the lexifier's. As we examine Kikóngo's verb forms, it is not so evident that temporal affixes are clearly distinguished for tense and aspect functions. For instance, one might argue that the suffix -idí alone stands for both, although it is conceivable to posit a Ø tense prefix to the verb, by analogy to a-V-idí in (6b). However, note that the aspectual marker is not always a suffix in Kikóngo, as may be inferred from the still incomplete analysis of (6e). And, if we treat FUTURE as tense, the tense marker may not always be a prefix. If we treat it as a mood, then it shares the same slot as some aspect suffixes. Kitúba regularizes things by making aspect periphrastic, treating FUTURE as modal (a conceivable though not the only analysis!), and showing tense consistently with a suffix (including 0 for NARRATIVE). Both Kikóngo and Kitúba have serial-like verb combinations; the former (7a) uses them only in the historical present (HP) (Mufwene 1988a), while the latter (7b) uses them only in the narrative tense.
Kitúba (7)
181 a. María ú-bák-a m-beele ú-lwek-a Mary AGRl-take-HP cl.9-machete AGRl-wound-HP bákála di-ándi. cl.5a-man AGR5-her "Mary took a machete and wounded her husband." b. María baka mbelé lwad-ís-a yakála na Mary take machete have.a.cut-CAUS-ENDING cl.la-man yándi. her
Overall, Kitúba's morphosyntax appears to be simpler than that of ethnic Kikóngo. However, this is not necessarily true of its semantics, as shown in Mufwene (1988a, 1990a) regarding time reference. Whether or not its structural features amount to a common denominator of various dialects of Kikóngo, which are not all mutually intelligible, is open to debate. The fact that it is different from Lumwamu's (1973) diasystem, which is supposed to represent features shared by the different Kikóngo dialects, suggests that it probably is not. Kitúba is not automatically intelligible to Bakongo.9 They must learn its novel tone system and its modified grammatical rules, as well as the part of the vocabulary contributed by dialects or languages other than their own. Fehderau (1966) has related several of Kitúba's structural features to the Kimanyánga dialect only, which he has proposed, convincingly so far — thanks to independent historical evidence — as Kitúba's lexifier. I return to the development question in §4.
3.
Ethnographic status
Since its beginning, Kitúba has functioned as a lingua franca, especially in the centres extra-coutumiers ("non-traditional [urban] centers") formed by the colonial administration, in the Christian missions, and in the factory towns created by large concession and exploitation companies such as the Kasai Company and the Lever Brothers. Africans from different ethnolinguistic groups were brought to live next to, or with, each other in these new localities. The new ethnographic conditions created by these new settings usually led to a division of labor between the ethnic languages and Kitúba, with the ethnic languages restricted to home or intimate situations and most of the public life
182
Salikoko S. Mufwene
conducted in the lingua franca. Thus, Kitúba has served more and more as a vernacular for many, i.e. as their everyday language, hence as their primary means of communication.10 With the vernacularization (i.e. usage as a ver nacular) also began its expansion and normalization as an autonomous system (with the regional dialects briefly discussed below), hence its status as what most creolists would call a creole.11 To date, Kitúba still serves in both capacities: (i) as a major vernacular for most of the urban population in the Bandundu and Lower-Zaire regions of Zaire (the former Leopoldville Province in the Belgian Congo, extending from west of the Kasai River to the Atlantic Ocean) and in the southern part of the Popular Republic of the Congo; and (ii) as a lingua franca for the rural population in the same geographic area. While the younger urban population generally speaks it natively, a good proportion of the urban adult population uses Kitúba as a second or third language, although the overwhelming major ity of them are fluent speakers. One reason for this differential state of affairs is the continuous rural exodus, which brings to the city a population of nonnative speakers in search of jobs. This new population has contributed to variation in city speech, although at any time the less fluent speakers have been a minority compared to fluent or native speakers, and their influence on the overall system may be considered rather marginal and minimal, even if possibly adding up little by little. This ethnographic state of affairs certainly accounts for words or phrases that are associated with some of the local Bantu languages but are not attested across the board in Kitúba territory. In any case, the variation says nothing special about the state of development of Kitúba, which expanded and normalized soon after the urban centers, the Christian missions, and the factory towns started, and then vernacularized. Vernacular ization certainly does not entail elimination of variation.12 The nonnative varieties of Kitúba spoken in rural areas call for more discussion. Country folk do not normally use these varieties for communica tion among themselves, unless there is among the participants in a speech event a stranger who cannot speak the local vernacular, usually the local ethnic language.13 They thus use it only occasionally as a courtesy to strang ers. Among the most common occasions are market days, visits to the city to see relatives or buy goods, visits to the regional health centers, and visits by regional administrators and politicians. Despite the higher proportion of second-language features in the rural varieties of Kitúba, one should resist the temptation to characterize these
Kitúba
183
deviations from the city vernacular norm as repidginization of the creole. The normal conditions for pidginization are not met, especially as in most cases Kitúba is used between country folk and fluent Kitúba speakers, who make available the city vernacular norm. The rural speakers only fail to speak it fluently, and usually they can improve their varieties if they interact regularly with fluent speakers. On the other hand, it seems legitimate to characterize the spectrum of varieties of Kitúba in both the city and rural areas as a continuum ranging from the native city norm to the most deviating rural speech. In addition to this variable spectrum is the variety of Kitúba spoken by the educated, who often use French not only as a more prestigious lingua franca but also as a second or third language. As observed in Mufwene (1979), it is not unusual for these speakers to transfer French structural features into Kitúba, as in reported speech. The normal Bantu reported speech style is quotative, as in (8); however, due to French influence, sentences such as (9) are also common with the same meaning:14 (8)
Pételo tub-áka nde: "móno ata kwísa. " Peter say-ANTER COMP me FUTURE come A. "Peter said [C0MP Ø], 'I will come."' B. "Peter said that he will/would come."
(9)
Pételo tub-áka nde yándi ata kwísa. "Peter said that he will/would come."
Although many French words have been borrowed by Kitúba, the edu cated variety often transfers others which are not really part of the system. For instance, biló "office, desk" (< French bureau), vwatíl/vwatír "car" (< French voiture), and ku-luwé "to rent" (< French louer) are established borrowings; but the status of le dernier "the last", often heard instead of ya nsuka (literally "CONNECTIVE end"), is not so clear. Less clear is the status of le premier "the first" and le deuxième "the second"; they alternate with the indigenous phrases ya ntéte (lit. "CONNECTIVE first") and ya zóle (lit. "CONNECTIVE two"), respectively.15 In noneducated speech le premier and le deuxième, but not le dernier, are commonly used to rank-order students in a class by their grades, although none of them is normally used to rank people by, for instance, order of arrival. However, in educated speech these terms may be used both ways. Despite the prestige of French in Zaire, it must be noted that this colonial language does not participate in the spectrum of Kitúba varieties. Although
184
Salikoko S. Mufwene
this could be characterized as a continuum, it would be a continuum without the Anglophone Caribbean kind of basilect-to-acrolect gradation. Certainly the variety of Kitúba spoken by the educated is not a mesolect. As it indicates no particular change in the direction of either French or Kikóngo, no notion of decreolization can be associated with it either. All we learn is simply that, subject to various influences, a spectrum or continuum of speech varieties obtains which can be associated with change, contrary to what has been assumed — too hastily, it seems — of the Caribbean situation. The way Kitúba is used makes it difficult to estimate accurately the number of its speakers, although we can assume that everybody in Kitúba territory speaks it as a first, second, or third language, regardless of the degree of fluency. In Zaire alone, it may be estimated that about 5-6 million people speak it. Ethnographically, its prestige ranks between French, the official language, and the ethnic languages, e.g. Kiyómbe, Kiládi, Kintándu, Kifióti, Kimanyánga, Kiteké, Kiyánsi, Kimpendé, Kimbala, Kingóngo, Kihungán, Kiyaka, Kisúku, and Kiboma. The ethnic languages belong to groups and H of Bantu, according to Guthrie (1953), and they differ in a number of their structural features, such as the number of their segmental phonemes, their tonal patterns, and whether or not they have subject-verb agreement.16 It is not obvious to what extent the structural differences among these ethnic lan guages are responsible for the three major dialects that Fehderau (1966) has identified for Kitúba, viz. the western dialect (WD on the map), spoken between the Kwango River and the Atlantic Ocean, south of the Zaire River; the eastern dialect (ED), spoken between the Kwango and Kwilu Rivers, south of the Kasai River; and the northern dialect (ND), spoken in the Popular Republic of the Congo.17 Correlation between dialectal variation in Kitúba and the ethnic languages of the corresponding geographical areas is quite likely but remains to be shown. The perspective presented in this essay is, as in Mufwene (1988a, 1989a, 1990a), primarily that of a native speaker of the eastern variety.18 As shown in §5 below, the structural differences between the eastern and the western varieties are not so strong as to affect the validity of the conclusions reached in §4.
Kitúba 4.
185
Genesis and development
There is more consensus on the status of Kitúba as a creole (see note 11) than on when and how it developed. According to Fehderau (1966), Kitúba started in the sixteenth century as a koine formed out of the contact of "dialects" of Kikóngo which in many cases are not mutually intelligible, especially those that are noncontiguous. Its development is associated with the trade of ivory and rubber, later on of slaves, against European goods. For about three centuries after the Portuguese discovered the mouth of the Zaire/Congo River in 1482, Euro-African contacts were restricted to the coastal area. Africans, in this case mostly the Bakóngo, served as middlemen in the trade with the hinterland. Fehderau claims that Kitúba originated in the hinterland's trade centers where speakers of different "dialects" of Kikóngo came into contact with one another.19 Manyánga, a town on the Zaire River, mid-course be tween the coast and Kinshasa (see map), seems to have been (one of) the most important trade center(s). According to Fehderau (1966), Kimanyánga, the local Kikóngo "dialect", appears to have had the most influence on Kitúba. It shows the highest rate of correspondences in the vocabulary and grammatical morphemes; even many of the periphrastic constructions now typical of Kitúba's aspectual delimitation, e.g. mé(n)e + verb stem for PERFECT and ké(l)e + verb stem for DURATIVE, may be traced back to options available in Kimanyánga (see also Mufwene 1988a, 1989a). On the basis of lexical and grammatical evidence, Fehderau hypothesized that Manyanga may have been the birthplace of Kitúba. An alternative to the above position is the view that speakers of Kimanyánga must have played a central role in the hinterland trade, which made their language particularly important as a lingua franca. These two hypotheses are not mutually exclusive; future historical research may help determine whether one of them is more plausible and whether there are other likely alternatives. According to Fehderau, the Belgian colonial administration played no role in the genesis of Kitúba, except perhaps in prompting a second wave of pidginization/creolization by bringing non-Kikóngo-speaking labor to the Lower Zaire region in the late nineteenth century and by exporting the original contact language east of the Kwango River (see map) as more and more colonial posts were created in the hinterland. Needless to say, mission aries, concession companies, and merchants were often ahead of the colonial
186
Salikoko S. Mufwene
administration in this role. That second wave of pidginization/creolization could be what produced today's Kitúba. Fehderau characterizes the original lingua franca of the first wave as a koine, although, as Samarin (1990) notes, there is no historical documentation of it.20 By the same token, we are unable to determine how much restructuring was involved in the original lingua franca, if indeed it developed out of contact primarily among the Bakóngo themselves. As the new language got established outside the former Kongo Kingdom, the regional impact of the local, non-Kikóngo language varieties must have been unavoidable. The divergence of Kitúba into distinct regional dialects must have started then. Its use as a vernacular in the new centres extra-coutumiers, the factory towns, and the missions marks what most creolists may call creolization. Although Samarin will certainly agree with the scenario about how Kitúba "creolized", spread, and split into regional dialects, he believes that it did not actually start until after the appropriation of the then-Congo by King Leopold II of Belgium in 1885, when the Berlin Treaty apportioned Africa to the European colonial powers (1982, 1990). As noted above, Samarin (1990) questions the existence of the putative koine stage of Kitúba, arguing that, like other linguistic facts, this variety would have been reported in historical documents if it had existed. He notes that observations abound in these documents on language diversity in Central Africa. Further, he attributes the beginning and formation of Kitúba (as a pidgin without a koine ancestor) PRIMARILY to the West African labor force and escort of interpreters that the colonial agents brought with them on their way southward (1982, 1990). The indigenous Africans putatively spoke this language to accommodate them and the colonial administration that they represented, of course not without affecting it with their own Bantu features; in the process the population of the new colonial centers vernacularized and helped spread Kitúba. According to Samarin, the West Africans included speakers of Kru, Mande, and Kwa languages.21 Were it not for the presence of speakers of Kru and probably of WestAtlantic languages22 (which have a complex morphosyntax) among these West Africans, Samarin's hypothesis would raise very few questions, perhaps no significant ones. It is tempting to assume, for instance, that the partly isolating morphosyntax of Kitúba is due (perhaps in part) to the influence of the varyingly isolating morphosyntax of the Kwa and Mande languages spoken by some of the West Africans. One would easily be led to assume that
Kitúba
187
a differential contribution of these languages to the formation of Kitúba, as opposed to that of Lingála, might account for morphosyntactic differences between the two language varieties.23 There are, however, no historical records to verify these speculations. Nonetheless, the assumption of the determinative influence of West African languages on the structure of Kitúba is only one of the possible explanations; it may be disputed by some linguistic and sociohistorical con siderations. Starting with linguistic evidence, the periphrastic aspectual de limitations selected by Kitúba were, as noted above, already available in Kimanyánga. As shown in Mufwene (1988a), the restructuring of Kikóngo's time reference was mostly morphological (in fact a matter of reducing the agglutinating and other complex morphological options); the semantic system itself has not changed significantly, if at all. Hazel Carter (personal communi cation, 1988) has also observed that, in some Kikóngo dialects, delimiting tense-aspect periphrastically rather than morphologically is either a common or the only option, especially with regard to aspect. As for the concord system, it is noteworthy that some of the Bantu languages involved in the development of Kitúba do not have the subject-verb concord rule that is typical of Bantu languages. For instance, in Kiyánsi, the initial verbal prefix is invariant with regard to person and number and covaries with tense and mood markers, as illustrated below.24 (A bar above a vowel indicates mid tone.) (10) Sample Verb Conjugation in Kiyánsi: me mā 'dia
nzé mā 'dia
me PERF eat
2sg. PERF eat
ndí
mā
'dia
bí
ā
he/she PERF eat
'dia
we PERF eat
"I have eaten" "You have eaten" "He/She has eaten" "We have eaten" me e-dí "I ate"
nzé e-dí "You ate"
ndi ē -dí "He/She ate"
me āyí'dia me FUT eat "I will eat"
nzé āyí 'dia 2sg. FUT eat "You will eat"
ndi āyi 'dia bí he/she FUT eat "He/She will eat"
āyí 'dia we FUT eat "We will eat"
ndi á-di-áná "He/She eats"
bi a-di-áná "We eat"
me a-di-áná nzé a-di-áná "I eat [HABIT]" "YOU eat"
bí
e-dí "We ate"
At present it is not clear whether, or to what extent, lack of subject-verb agreement is a feature of other Bantu languages of the region, especially those of group B, which differ from the Bantu languages of other groups in a number of other ways (e.g. a larger number of vowels, including diphthongs).
188
Salikoko S. Mufwene
In any case, it may be premature, if not arbitrary, to attribute Kitúba's tendency toward an isolating morphosyntax primarily to West African influ ence, even if such influence may have helped establish the feature in the developing system. Samarin's hypothesis might also appear to be weakened by absence of motivation. Why, one might be inclined to ask, would the West Africans have formed a Bantu-based lingua franca to communicate among themselves? Why would they not have used one of the West African pidgins, which were already well established along the West African coast by the second half of the nineteenth century? If the purpose of the lingua franca was to communi cate with the Europeans, why not use French or English (or perhaps any of the West African pidgin English varieties), which, unlike Kitúba or any of the local Bantu languages, would have guaranteed them a higher social status? Note that a correlate of the Belgian colonial administration's commitment to using local lingua francas to communicate with the indigenes was the associa tion of the European languages (French and Flemish) with power. The Bel gian colonial linguistic policy was designed to keep the vast majority of the indigenes ignorant of metropolitan languages. It is in part for this reason that fluent command of these languages became an implement of status and/or power in colonial and early postcolonial days.25 Thus, it would certainly have been in the interest of the West Africans not to be confused with the local Bantu population. Being identified immediately by the kind of European-like language they spoke should have given them a number of colonial advan tages. However, saying that the West Africans developed, or helped develop, Kitúba does not entail that they used it for communication among themselves. We simply have no information about the language(s) they used. It is assumed that, in order to serve as interpreters for Europeans, they must have been able to speak some European language, since, according to Samarin, the Europeans generally did poorly in learning African languages. What Samarin argues is that they must have used the emerging lingua franca, probably what they took to be a major indigenous language, to communicate with the indigenes. The contribution of West Africans to the development of Kitúba is thus plausible under the following scenario not elaborated by Samarin: in their attempts to communicate with the local Bantu populations, the West Africans might have misshaped one of the local lingua francas (i.e. some local language variety used for interethnic communication, probably Kikóngo
Kitúba
189
Kimanyánga); the indigenes in turn accommodated the powerful newcomers by speaking this variety, which they may have interpreted as the colonial way of speaking this language variety. This alternative, which is consistent with the above-stated Belgian colonial linguistic policy of communicating with the indigenes in their own languages, is also consistent with the practice among the Bantu, accurately reported by Samarin, to accommodate their neighbors in communicating with them in their own language. One of the reasons why almost no precolonial pidgin lexified by Bantu languages has been reported is apparently this practice: mostly those who were multilingual ventured in long distance interethnic trade (Samarin 1986). Some languages must have pre vailed on some trade routes. Like Bobángi (the lexifier of Lingála, according to Hulstaert 1989 and Samarin) on the northern, mid course of the Zaire River, Kimanyánga may have prevailed on the trade, and later, porterage route from the Kóngo-Teké border to the Atlantic Ocean (see also Samarin 1990). In reducing the motivation for Bantu speakers to develop a pidgin but assuming that they probably helped one of the local languages emerge as the lingua franca, Samarin's hypothesis becomes more plausible than Fehderau's. Support for Samarin's position is, however, still indirect and may be adduced from the following incident reported by Hulstaert (1989:101) about the mis taken identity of Lingála by early missionaries on the midcourse of the Zaire River. I think what we learn from it may well apply to Ktuba. 26 In 1895 the Catholic missionaries settled in the Equateur Region (Northwest of Zaire), on the big bend of the Zaire River toward the east (see map). However, they did not discover until 1898 that the language they spoke with the indigenes was not their native language variety, Lomóngo-Lonkundó; it was a nonlocal variety (based on Bobángi) which the indigenes associated with the Europeans and used to communicate with them and their African associates. It apparently took a fire in the kitchen of one of the boarding schools for a nun to discover this, as she heard a girl yell in Lonkundó, rather than in Lingála-Bobángi, that the building was afire. When the nun asked the girl to repeat what she had said, the girl translated it automatically into Lingála-Bobángi. A discussion that ensued between them clarified the linguistic confusion, after which, according to Hulstaert, the missionaries decided to learn the local indigenous language. It must be noted that Samarin's hypothesis may still prevail over a third alternative, which I will present anonymously, according to which Ktuba started among the indigenes in the colonial posts and the factory towns (both of which developed into the centres extra-coutumiers) and the missions. In
190
Salikoko S. Mufwene
these new settings, Africans of various ethnolinguistic backgrounds were brought into contact with one another and would of necessity have developed a new lingua franca. The following factors would support the hypothesis that the Bantu population formed Kitúba: (i) they communicated more among themselves than with the foreign population; (ii) their languages share more typological and lexical kinship with each other than with the European or West African languages, which explains why, despite some extensive restruc turing, Kitúba remains Bantu; and (iii) the West Africans did not really need a Bantu language to communicate either among themselves or with the Europe ans. These three reasons are all rather weak in the absence of other Bantubased pidgins to support the hypothesis that such circumstances encourage the development of pidgins. After all, there were many other attempts in the twentieth century by missionaries to use Bantu languages for communication not only with the local populations but with others that moved to the missions. While these attempts contributed to making these languages widely spoken, they did not result in Kitúba-like phenomena. Recall that the only other case, Lingála, remains disputable as an exclusively Bantu creation, and there is no compelling evidence for treating Zaire Swahili as an erstwhile pidgin (see Mufwene 1989a, Kapanga 1991). The development of Fanakalo (Heine 1970) in South Africa seems similar to those of Kitúba and Lingála, being associated with mining and foreign labor, reminiscent even by its alternative names (e.g. Isikula "coolie/Indian language/speech", Isilungu "European language/speech", and Kitchen-Kafir, by association with the South African whites' kitchens) of the history suggested by some of Kitúba's names, such as Kikóngo ya letá and Kibulamatári, discussed above in §2. Fanakalo has spread further north to places like Zimbabwe and Zambia among miners but not as a generalized lingua franca like town Bemba. As noted above (contrary to Mufwene 1989a), it was not absolutely necessary for the Bantu to communicate with each other in a pidgin in the colonial posts, the factory towns, and the missions, particularly in the ethnic Kikongo territory. Even if they may have resorted to some jargons in the initial stages, they probably switched to the local Bantu languages that func tioned as lingua francas. This is not an unusual solution with migrants in Central Afica. In the city and the Christian missions, newcomers to socially multilingual neighborhoods have not abandoned their ethnic languages but have simply learned the varieties that would enable them to interact with and
Kitúba
191
be integrated in the local populations. Because of its association with colonial power, the variety probably initiated by the West Africans and carried over to the different posts may have been adopted, for reasons of practicality, espe cially if the locals themselves used it to communicate not only with the foreigners but also with the other Bantu speakers associated with the spread of colonization. However, this emerging language variety was likely enriched with some Bantu features, e.g. in the noun class prefixes and some tense/aspect affixes (if these had been done away with, as in Atlantic creoles). The Bantu speakers may have done this easily, and therefore participated in the development of Kitúba, once they recognized that the variety perhaps originally produced by the West Africans in speaking with them was based on a Bantu language. And in reply to objection (iii), the West Africans needed a Bantu language not to communicate among themselves but to serve as interpreters between the Bantu populations and their colonial employers. Thus reinterpreted, Samarin's hypothesis is not as implausible as it may seem. In fact it is very similar to the third position on the development of Kitúba, except that it involves the West Africans as those who apparently initiated the process. The colonial posts, the factory towns, and the missions with which Kitúba's development was associated provide the sociohistorical setting. The West Africans as interpreters between the Europeans and the Bantu populations were instrumental in the development of the colonial posts and factory towns in particular. Their association with colonial power must have made the contact language they may have initiated a viable alternative for communication, once it was identified by the indigenes as a lingua franca. Its use in all the colonial posts where the West Africans and others carried it must have given it widespread usefulness that later on relegated the Bantu languages that hitherto had functioned as lingua francas to a lower ethno graphic status, especially in the colonial posts. It may be surmised that, despite the similarity on at least the lexical level of the initial-stage Kitúba — as used by the West Africans — to Kikóngo and to some of the languages around the area, the Bantu populations, including the Bakóngo, just assumed that it was the colonial way of speaking its lexifier and largely accommodated the powerful foreigners. On the other hand, they must not have been passive language learners in this kind of setting. While accom modating the colonial, European and West African, foreigners, they must have brought the new language closer back to the Bantu patterns. In turn, the
192
Salikoko S. Mufwene
West Africans must have adjusted back to the variety spoken by the Bantu speakers. Whether or not things worked out this way remains a speculation difficult to verify beyond doubt in the absence of a sample of West Africans' Kitúba and, generally, of early linguistic records. One particular factor en couraging the above division of labor between the West Africans as perhaps the initiators of the restructuring of Kitúba's lexifier and the Bantu as those who shaped Kitúba as we know it is that, despite its development toward an isolating morphosyntax (Mufwene 1988a), Kitúba is still in several ways more different from Kwa, or West Atlantic, or Mande languages than it is from Kimanyánga or any of the surrounding Bantu language varieties, e.g. in the noun-class and time reference systems (Mufwene 1988a, 1990a). One more thing must be explained further here. The selection and further development by the Bantu speakers of Kitúba, a language which may have been initiated by West Africans, as a lingua franca and subsequently a vernacular of the colonial posts, the factory towns, and the missions simply has to do with practicality. This selection made it possible for them to communicate not necessarily with each other but primarily with the colonial agents, often through their interpreters, as we learn from Samarin (1982, 1989). Those who aspired to command positions must have had to speak it. Note that as colonization progressed, the Belgian colonial system in particular required its agents, especially the Africans, to learn the local lingua franca in order to communicate with the indigenes directly. In any case, we must remember that a different solution had been found for intra-Bantu communi cation in precolonial times, and there was no compelling reason to abandon it, at least not totally.27 It is also significant ethnographically that the Bantu languages have not been displaced in the cities that developed from the colonial posts and in the missions; they have simply been assigned more intimate functions. Multilingualism among the Bantu speakers has continued to date, despite the increasing ethnographic status of Kitúba and Lingála as languages of wider communication. It is likewise important to explain, as Samarin generally attempts to do, why the European colonial agents are not invoked as central actors in the present account of Kitúba's development. In the colonial posts which urban ized into the centres extra-coutumiers, the Europeans (at least the Belgians) usually buffered themselves from the Bantu populations by a zone where the foreign Africans and the police or army were stationed. This topological structure, which continued up to independence, guaranteed that the Bantu
Kitúba
193
population not associated with power interacted primarily among themselves and with the inhabitants of the buffer zones. The Europeans thus played no significant part other than providing the settings. Given the above adaptation of Samarin's hypothesis, it seems justified to look for determinative influence in the formation of Kitúba in at least some of the indigenous Bantu languages. Whereas the West Africans, on this hypoth esis, would presumably have restructured its Bantu lexifier in some way, e.g., perhaps, in assigning Kitúba a more isolating morphosyntax than it has today, the indigenous Bantu-speakers would then have restructured it in turn, espe cially as they used it more and more frequently and among themselves. In recognizing its kinship to Kimanyánga, as corroborated by linguistic analysis (Fehderau 1966), there must have been more and more motivation for reBantuizing its system, though this time not according to Kimanyánga, as overwhelming as its influence remains. Before closing this part of the essay, I should explain an important, differential European linguistic element in Kitúba. There are both Portuguese and French words in this "creole". Interestingly, they seem to fall into different domains. As in other Bantu languages, most of the Portuguese words have to do with the goods that the Portuguese traded with the Africans, e.g. lóso "rice" < arroz, mesa "table" < mesa, and sabátu/sampatu "shoe" < zapato?. Common words of French origin in Kitúba are associated with colonial administration, e.g. hiló "office" < bureau, palató "official's mes senger" < planton "messenger soldier", and komí "secretary, registrar" < commis. Since Portuguese trade started earlier, in the sixteenth century, one might be inclined to conclude that Kitúba developed before colonization and that the Portuguese words got integrated in it the same way they did in other Bantu languages, as in fact did some English words (e.g. búku < book and katekisímu < catechism), which attest to early contacts with Anglophone missionaries before contacts with the Francophone ones.28 However, this conclusion is easily refuted. The Portuguese and English words may have entered Kitúba through the indigenous languages. Note that, until the colonization period, the Portuguese presence in Central Africa was on the coast; they traded with the hinterland indirectly through African middlemen. It is likely that during that time the Africans simply resorted to multilingualism, especially in learning one of the ethnic languages emerging as a lingua franca. Kimanyánga might very well have emerged then as a lingua franca. The very fact that no phenomenon similar to Kitúba developed
194
Salikoko S. Mufwene
in Angola and Mozambique, former Portuguese colonies, is probably evi dence of a different way of interacting with the indigenes, which continued through the colonial period. As for the missionaries, they were more inter ested in the ethnic vernaculars, as evidenced in part by the above story from Hulstaert on the misidentification of Lingála-Bobángi. Other authors, such as Polomé (1963) and Samarin (1986), support this view. In fact, an important proportion of European students of African languages have been missionar ies, from William Bentley at the turn of the century to our contemporary Ian Daeleman, in the case of Kikóngo. All in all, while it is possible that Kitúba started before the colonization of the region by both the Belgians and the French, the evidence for Samarin's hypothesis that it started with colonization itself seems more convincing. If I were forced to take a position, I would reinterpret Samarin's position as follows. By the time the Belgians colonized Zaire, then the Congo, Kimanyánga would already have emerged from previous trade of ivory and other commodi ties as an important lingua franca. When the Belgians and their West African escorts and labor arrived, they would have identified it as such; the West African interpreters would have tried to speak it with the indigenes but would probably have distorted it in the process.29 As the indigenes accommodated them, they further restructured it bringing it closer to the Bantu system but not as close as Lingála. Kitúba was then born from the original attempts of the West Africans to communicate with the indigenes, but as the latter adopted and vernacularized it in the colonial posts and later the missions, they contributed significantly to its current shape. The question of why Kitúba is more isolating, and less close to the Bantu system, than Lingála remains unanswered and will be addressed in a separate paper. I suspect that the clue to this mystery lies in Kimanyánga itself and some of the Bantu languages such as Kiteké and Kiyánsi, which do not seem to be as fully agglutinating as the standard Bantu canon. A comparison of their features with those of the languages that participated in the vernacularization and normalization of Lingála is of course needed to make the case convincing.
5.
Creolization and divergence
If, like Chaudenson (1979, 1988), Valdman (1983), and myself in this essay and elsewhere (e.g. 1989a, 1991a, 1991b), we interpret "creolization" as the
Kitúba
195
vernacularization and normalization of a pidgin, it seems that Kitúba outside the former Kongo Kingdom must have "creolized" soon after the centres extra-coutumiers, the factory towns, and the Christian missions were estab lished at the turn of the century, especially when the economic exploitation of the then-Congo and efforts to christianize and "civilize" it intensified around World War I. However, if "creolization" is interpreted as nativization of an antecedent pidgin, the probable date of this process is difficult to determine for Kitúba. Independently of reasons for questioning the significance of nativization in creolization (see e.g. Mufwene 1991a, 1991b), part of the difficulty in inferring the date of the process derives from the fact that it is not clear what proportion of native speakers relative to the total population of a contact situation would bring it about. Moreover, even if we knew that critical proportion of native speakers to "pidgin-speakers", there are problems with inferring the probable time of creolization qua nativization. The permanent population of the Christian missions has always been very small, with most of the population being renewed in overlapping periods, hardly permitting an important community of native speakers to emerge locally. The colonial centres extra-coutumiers and the factory towns have had more stable popula tions; it should therefore be helpful to check their census records, especially during the colonial days, and to determine the period when the native-speaker population reached that critical mass. To date I know of no such study, and I have so far been unable to conduct one myself. The question of the time of Kitúba's "creolization" will thus have to be settled elsewhere, along with that of whether or not the association of the process with nativization is valid. One particular problem with Kitúba is finding old(er) texts that may allow conclu sive comparisons with today's primarily spoken texts. Whenever Kitúba "creolized", the regional variation noted by Fehderau (see §3) is real and is supported by various kinds of linguistic evidence. However, the correlation between the evidence and the structure of the local languages still remains to be determined.30 For instance, the eastern dialect (marked ED on the map) has a word-initial floating high tone on a number of monosyllabic words, such as 'nti "tree, stick", 'nda "long, tall", and ' "hat"; this feature is consistent with the stereotypical observation that Kitúba is an accent language, with the accent (or high tone) on the penult. In the western dialect (WD on the map), an augmentative vowel i is usually prefixed to such words, producing ínti, índa, and ímpu and thus regularizing the accent-on-the-penult pattern.
196
Salikoko S. Mufwene
Another noteworthy difference between the eastern and western dialects lies in the details of durative aspect delimitation. The pattern in the eastern dialect is to combine ké(l)e "be" with a verb that need not be in the infinitive, as shown in (11a). In the western dialect the verb must be in the infinitive, which functions in Bantu morphosyntax as a verbal noun; the overall con struction is locative, with the infinitive preceded by the connective na, as in (11b): (11) a.
ED:
Móno/Mú ké(l)e (kú-)dia. me be (INF-)eat "I am eating." b. WD: Mono kéle na *(kú-)dia.
Some common words are also pronounced differently. For instance, in the following pairs the eastern dialect uses a voiceless fricative where the western dialect uses a voiced one: mása/máza "water", mbísi/mbízi "fish, animal", and mandefu/(ma)ndévu "beard".31 As noted by Fehderau (1966), there are words used in one or two dialects but not in the other(s). Examples are the words bilongo "medicine" and mfúlu "bed", used in the western dialect (per Fehderau), which are strange to me as a speaker of the eastern dialect, which has nkísi and mbéto, respectively. There is internal variation even within the broad regional dialects identi fied by Fehderau. Within the eastern dialect, for instance, there are some differences, perhaps minor for some, between the variety spoken in the Bandundu city area (at the confluence of the Kwilu and Kwango Rivers — see map) and that spoken in the Kikwit area, up the Kwilu River (southeast of the Bandundu region). For instance, the first person singular pronoun mú is avoided in the Kikwit variety; the full variants mono and múnu are used instead. Also, the Kikwit variety has a polite imperative form, with the suffix -éte (illustrated in 12a), which is not used in the Bandundu city variety. Here the bare verbal stem, similar to the imperative in Lingála, is the only option:32 (12) a. Kikwit: b. Bandundu:
Kwis-éte. "Come, please." Kwísa. "Come."
The Bandundu city variety uses a number of words which have cognates in, or have been borrowed from, Lingála, such as kubimisa "to let out", mpási "pain (physical or moral)", kusidíka "to get angry", kukauka "to dry" (transi tive or intransitive), mbísi ya kukaúka "smoked or dry fish", niama "animal,
Kitúba
197
meat", and kusukúla "wash" (transitive or intransitive). Instead, the Kikwit variety uses, respectively, kubasísa (a causative derivative from the same verbal base as kubasíka "get out", which is also used in the Bandundu city variety); kiádi "moral pain" opposed to mpási "physical pain, hardship, difficulty"; kúwa makási (literally, "perceive anger") "to get angry"; kuyúma "dry" (intransitive) vs. kuyumísa "dry" (transitive); mbísi ya kuyúma "dry or smoked fish/meat"; mbísi "meat, animal, fish"; and kuyobísa "to wash" (transitive) vs. kusukúla "to wash or bathe" (intransitive). The Bandundu variety also uses a comitative conjunction na where the Kikwit variety uses tí (also used in possessive constructions), as in 13 and 14: (13) a.
Bandundu: Yándi kwis-áka na bô. he/she come-ANTER with them "He/She came with them." b. Kikwit: Yándi kwisáka tí bô.
(14) a.
Bandundu: Yándi kéle he/she be "He/She has b. Kikwit: Yándi kéle tí
na with it." yô.
yô. it
There is no evidence that any of this variation indicates restructuring of Kitúba grammar. If Fehderau is right in assuming that Kitúba "creolized" after World War II (perhaps because he might associate "creolization" with nativization by a yet undetermined proportion of the speakers), then it is worth noting that I have not detected any structural changes in my dialect of Kitúba over the last thirty years. Although isolated forms seem to suggest a lot of variation, some of the combinations are not considered normal and indeed help to distinguish fluent speakers from less fluent speakers. For instance, in the Bandundu city variety, the first person singular pronoun is móno, múnu, or raw; the durative marker is the copula ké(l)e in the present, and the verb that is delimited in the durative might, or might not, be inflected in the infinitive. Thus, in principle, several combinations of the above alternative forms are possible, but only some of them are used by native and other fluent speakers. In (15) the odd combinations that fluent speakers would not produce are marked with "!": (15) a.
móno/múnu kéle kú-dia "I am eating." me be INF-eat
Salikoko S. Mufwene
198 b. ?!móno kéle 'dia. c. móno/múnu kéle 'dia. d. mú kée 'dia. e. ?!móno/múnu kée kúdia. f. ! mú kée kúdia.
Of all the combinations, (15a) and (15d) are the best; (15c) is acceptable. The others are not so well formed. One could also see here some justification for Ngalasso's (1984) assumption that Kitúba might be going back to an agglutinating system, although I doubt very much that this variation reflects a diachronic process (Mufwene 1988a). Whatever way individual linguists choose to interpret such variation in Kitúba, its existence may not be denied and researchers should be aware of it.
6.
Conclusions
In this essay I have tried to situate Kitúba geographically and to explain why it is called so many different names, all of which are justified either by its history or by the contrasts popularly perceived between its structural features and those of the Bantu languages spoken in the part of Africa, mostly Zaire, where it is spoken. Having presented some of its structural features discussed in the literature, especially in Fehderau (1966) and Mufwene (1988, 1989a, 1990a), I described its ethnographic status in Zaire, compared to French, the official language, and to the local Bantu vernaculars. I then reviewed Fehderau's (1966) and Samarin's (1990) positions on the genesis of Kitúba, adducing several reasons for preferring Samarin's hypothesis. I find it plau sible to assume that Kitúba developed in the late nineteenth century with the colonization of the then Congo and with the development of the centres extra-coutumiers, the factory towns, and the Christian (mostly Catholic) missions. Though I acknowledge the trigger role of the West African colonial escorts, I am more inclined to assume that, as the indigenous Africans accommodated the powerful newcomers in learning the colonial language variety, they contributed to shaping it to what it is like today.33 Evidence for Bantu contribution may be sought in the regional variation in Kitúba, though the specific correlation between each linguistic region and its variety of Kitúba remains a desideratum for future research.
Kitúba
199
On the other hand, the data covered in this essay prompt some general observations and questions regarding creole genesis that I cannot resist men tioning here. Tone, with which I began the discussion of Kitúba's structural features, is a convenient starting point. The predominance of a phonological tone system in Kitúba suggests that tone leveling may be one of the options for a creole that develops out of contact among tonal languages, which would include both the Bantu and West African languages in this case. However, because Kitúba differs from Lingála in respect to tone, which also developed out of contact among Bantu and West African speakers of equally tonal languages, the following question arises: why did Lingála turn out to have lexical and grammatical tones, instead? What arouses more curiosity about the difference between Kitúba and Lingála is the fact that speakers of more or less the same kinds of West African languages and in some cases the same Bantu languages (e.g. Kiyánsi and Kiteké) were involved. Future research will have to address this question if we wish to understand better how creoles select their structural features. In the case of Bantu languages, it will be necessary to determine whether or not the same dialects were involved, as dialects of the same language may differ typologically. It will probably also require understanding similar developments in coastal Swahili and Comorean (Jouannet 1988) to determine what conditions lead to loss (partial or com plete) of lexical and/or grammatical tone systems. Another structural feature of Kitúba that I have found difficult to account for (see e.g. Mufwene 1989a) is the use of independent pronouns instead of the Bantu pronominal prefixes, as discussed in §2. I argued in Mufwene (1989b, 1991c) that, being tonic (with their contrastive and emphatic func tions in Kikóngo and most other Bantu languages), the independent pronouns were more salient, which granted them some sort of unmarked value in the language contact situation. As the agreement pronominal prefixes vary in part according to the class of the subject noun or any other noun that governs agreement, a system using them may be considered more complex than a system that does not. Complexity, perhaps combined with the fact the prefixes are atonic, must have made the agreement pronouns a dis- or less favored choice, especially since the mostly structural information they provide is not crucial. Whether or not the above conjecture is correct, a question similar to that raised above in respect to the tone system arises: why has Lingála kept the canonical Bantu system with subject-verb agreement, which allows the omis-
200
Salikoko S. Mufwene
sion of the independent subject in anaphoric and deictic contexts? To be sure, since some of the languages in the contact situation, such as Kiyánsi and apparently Kiteké, do not seem to have the subject-verb agreement system or to allow pro-drop, the scale must have tipped in favor of the independent pronouns. However, as observed above, Kiyánsi and Kiteké were also in volved in the genesis of Lingála. Thus another question bears on the initial one: were there more Kiyánsi and Kiteké-like languages or were there more speakers of such languages involved in the development of Kitúba than in the formation of Lingála? Future research will also have to address these ques tions. I note in Matuka (1990) that subject-verb agreement in Kimanyánga is also less regular than in the Bantu canonical system. There are cases where no agreement occurs (marked by 0): 3 4 (16) a.
Mbuta muuntu Old man Ø-mwen-i
w-e-zieet-a kuku, AGR-ANTER-travel-ENDING here ki-ina ki-ndumba-ndumba.
AGR-See-NEAR.PERFECT
AGR7-that
cl.7-girl-girl
"My old man traveled this way and saw that young woman over there." b. Kiedika Ø-mwen-i m-buungu y-oyo. truth AGR-see-NEAR.PERFECT cl.9-cup AGR9-that "Truly, you have seen that cup [of wine]." Although at present I have no information about how regular the subjectverb agreement system is in Bobángi, Lingála's lexifier (at least according to Samarin and Hulstaert), the question arises as to whether the irregularity of subject-verb agreement in Kimanyánga played a significant role in determin ing the total absence of this canonical Bantu feature in Kitúba. While looking into the above questions, we should also examine closely the pattern of agreement between the head noun and its modifiers in Kimanyánga, because Kitúba lacks this morphosyntactic feature too. There is yet another open question about a structural feature of Kitúba. In examining relative clauses, I note that at least in my dialect of Kiyánsi they are introduced by an invariant complementizer kó (specialized for relative clauses) rather than by a general connector for all nominal modifiers, as in Kitúba. In Kikóngo, relative clauses are often introduced by a demonstrative which agrees in class and number with the head noun, as illustrated in §2. A second Kikóngo alternative consists of contact relative clauses, where no
Kitúba
201
connector or complementizer introduces the relative clause and its verb starts with the subject prefix only. A third alternative consists of bounding the relative clause to the right with a demonstrative marker which agrees in class and number with the head noun (Mufwene 1988a, 1990b). Could the selection of the general connective in Kitúba simply have been determined by the fact that Kikóngo itself does not seem to have any complementizer introducing relative clauses? The choice in Kitúba of relative clauses with an invariant connective ya, the form of which may be related to some of the demonstra tives that introduce relative clauses in Kikóngo (e.g. dia, mia, kia, and bia), raises the question of whether introducing relative clauses with a connective is less marked than introducing them with a complementizer or a relative pronoun. If Kitúba's general connective is related etymologically to Kikóngo's demonstratives, in which only the initial consonant varies for agreement purposes, it will help in future research to examine the extent to which the presence of a particular parametric option in the lexifier plays a role in determining the selections made by the new language variety. What we should not discount by fiat is the possibility of partial convergent influence with the connective ya originating partly in Kikóngo and invariance deter mined by Kiyansi-like languages. The question of the role of the lexifier itself in determining the structural options selected by the new language variety becomes more significant if we pay attention to details of Kitúba's time reference system, as presented more fully in Mufwene (1988a, 1990a). As noted in §2 above, the aspectual markers were selected from Kimanyánga. However, the source(s) of the tense markers, -a(k)a for "ANTERIOR" and the null suffix "NARRATIVE", is/are not obvious. To be sure, it is conceivable that the null tense marker may have developed from Kikóngo's historical present tense, helped by the loss of subject-verb agreement. However, the source of the ANTERIOR marker re mains obscure. As argued in Mufwene (1991c), salience may account for the selection of periphrasis over agglutination for aspect marking: it seems to have made the periphrastic option unmarked in the contact situation. Reason ing the same way, it may be reasonable to associate the selection of á(k)a with the high tone (or accent) on its penult, which made it not only consistent with the dominant penultimate tone placement system but also more salient than the tense prefix alternative, which most likely would have been assigned a low tone instead. Given the possibility that á(k)a may have been selected from a language or dialect other than the lexifier, the following question is
202
Salikoko S. Mufwene
relevant: under what circumstances may grammatical morphemes be selected from outside the lexifier? To conclude, after examining several incomplete facts of Kitúba I have raised several questions that cannot be answered without further investigation of the languages, especially the Bantu ones, out of whose contact it devel oped. More demographic facts about ethnolinguistic contacts and dates there of are also badly needed, but we may never be able to lay our eyes on such information. This study also suggests that a closer comparison of the develop ments and structures of Kitúba and Lingála might shed light on some of the questions. These are but a handful of the several issues that must be addressed if we wish to understand the general phenomenon of the formation of contactinduced language varieties.
Notes *
I am grateful to Morris Goodman, William Samarin, and Sally Thomason for very helpful comments on a previous version of this essay. I alone assume responsibility for the remaining shortcomings, especially where I failed to follow their advice.
1.
Throughout this paper I will adopt the unconventional practice of indicating tones on language names, as there are a few names either with no high tone at all or with a high tone that is not borne by the penultimate syllable. As in most of the African linguistic literature on languages with only two tone levels, only one tone level will be indicated, the one with the smaller distribution, here the high tone, which is indicated by an acute accent on the vowel. The low tone is not shown. Where data or language names are hyphenated, the convention followed is that of African languages, in which a homorganic nasal goes with the consonant that follows it, e.g. Li$ngá$la and Ki$kó$ngo.
2.
Móno kutúba is a foreignism for what native speakers say as móno túba. In the narrative tense (Dahl 1985), verbs are normally used without any prefixes. In the phrase móno kutúba, ku- has been prefixed to the stem túba "say/speak" based on the wrong analysis that in the narrative tense verbs are used invariably in the infinitive. Ki- in Kitúba is the normal Bantu nominal prefix for instruments (Class 7), which applies also to languages, as in Kikóngo, Kiyómbe, Kimanyánga, Kifióti, Kiládi, and Kiteké.
3.
Reference to Kikóngo (spelled Kikóóngo by some linguists, e.g. Daeleman 1972, 1982) as one language is somewhat simplistic. First, there is not always guaranteed mutual intelligibility among the putative "dialects" of Kikóngo, viz. Kiyómbe, Kimanyánga, Kintándu, Kifióti, Kiládi, etc., all spoken in the area of the Kongo Kingdom (see map). Second, speakers of these language varieties do not consider themselves to be speaking the same language. They do, however, consider themselves as forming one group compared to other Bantu groups associated with languages such as Kiteké, Kimbala, Kimpendé, and Kiyánsi (my own ethnic language) in the area extending from the eastern and northeastern borders of the Kongo Kingdom to the western side of the Kasai River
Kitúba
203
(see map), where Kitúba is also spoken. The distinction "ethnic Kikóngo" vs. "KikóngoKitúba" is significant insofar as the latter is also referred by its users, especially outside the former Kongo Kingdom, as Kikóngo. Here and elsewhere in my work, I have used "ethnic Kikóngo" for the linguists' construct denoting the cluster of language varieties (Hinnebusch 1989) enumerated in the second sentence. 4.
According to William Samarin (personal communication, 1992), the term bula-matádi was first used in reference to the British explorer Sir Henry Morton Stanley "because of some dynamiting he did to get his boat-parts (boilers?) to the [Stanley] Pool" (the present location of Kinshasa) from the Atlantic coast in 1879. This was during his second exploration of the Congo, before the Berlin Treaty in 1885 and the appropriation of the Congo by King Leopold II of Belgium.
5.
The geographical location of the Kongo Kingdom (see map), as the gateway of colonial expeditions into the hinterland, must have been an important factor in the adoption of Kitúba (based on ethnic Kikóngo) as a lingua franca. The need for the imported labor force to communicate with the local population in what was taken to be their language (see below) is another relevant factor. Another significant factor is the importance of Manyánga (see map) as a trade post, hence a major location of cross-language intercourse, dating from before colonization. This last factor explains why features of Kimanyánga appear to have prevailed.
6.
I speak of "canonical system" or Bantu "canon" in this essay to suggest that although some features, especially morphosyntactic ones, have generally been associated with membership in the Bantu group, not all members of this family have them (Mufwene 1994).
7.
Because of the facts described in n. 2, Ngalasso (1991) argues that Kikóngo as a language name applies more adequately to Kitúba than to ethnic Kikóngo.
8.
One exception to this generalization comprises some deverbal derivatives that also involve partial reduplication, such as bi-lómba-lomba "habit of asking for anything" and bi-túba-tuba "habit of talking too much and becoming indiscreet". These contrast with reduplications such as poló-poló "characteristic of a person that talks too much, indis creet", malémbe-malémbe "very slowly", and mbángu-mbángu "very fast". For more on reduplication in Kitúba see Ngalasso (1993).
9.
Ba- in Bakóngo is the plural of the singular agentive prefix mu-; they are attached to language name stems to designate speakers.
10.
I assume that usage of a language variety as a primary means of communication in a particular locality or region accounts for its being subsequently associated with the locality or region as indigenous to it, i.e. its being called the locality's or region's vernacular, at least in the etymological sense of vernacular as "homeborn".
11.
I am one of those who do not consider nativization, in the sense of usage as a mother tongue, as a necessary condition for the creolization of a pidgin. There are to my knowledge no structural features of creoles which are due to children and which distin guish them from pidgins. I also think that in most creole communities the expansion and normalization of the system was due to its vernacularization, as it was used more and more as a primary, everyday means of communication and in a wider variety of contexts. I have some reservations about calling Kitúba a creole (Mufwene in press), although this
204
Salikoko S. Mufwene position does not affect the rest of the essay. I just fear that the term creole has been overextended unnecessarily beyond its social historical context of the European settle ment colonies of the Americas and the Indian Ocean.
12.
Generally, the longer the newcomers live in the city, the closer their second-language variety gets to the local city norm, in the same way that, for instance, today's foreigner speech in the USA adjusts progressively to the local norm but is not normally expected to change the shape of American English.
13.
In villages with more than one ethnic group the groups live in separate parts of the village, using their respective ethnic languages as their vernaculars. Interethnic communication is usually in the language of the dominant group, though sometimes Kitúba is used as well.
14.
Since, due to colonial circumstances, the educated way is more tolerated than corrected or ridiculed (Mufwene 1988b), the coexistence of the two rules for reported speech creates confusion. Note that the subordinate clause's tense in (9) is not reoriented, unlike the pronoun. Spoken, either sentence may be misinterpreted, depending on whether the French-derived rule or the Bantu rule is applied.
15.
The plural is formed both by changing the article, as in French, and by attaching the class 2 prefix ba- to the phrase, as to Bantu nouns of class 1. This yields ba+les premiers; my intuitions are, however, less clear on whether or not ba+les deuxièmes is equally accept able.
16.
At least Kibwaal, my dialect of Kiyánsi, lacks subject-verb agreement, and uses an invariant connective in Head+Modifier constructions.
17.
Fehderau observes that the division is made particularly for convenience. It is, however, noteworthy that the Kwango River is near the eastern border of the former Kongo Kingdom. Most of the languages spoken east of the river belong to Group of the Bantu family and differ more significantly from the Kikóngo "dialects" (Group H) than the latter do among themselves (see n. 3). The usage of the term Kikwángo for a particular variety of Kitúba coinciding roughly with Fehderau's "eastern dialect" suggests that at least part of the division is real to speakers of Kitúba and is not simply academic.
18.
I grew up a bilingual speaking both Kitúba and Kiyánsi natively.
19.
It is not clear whether the role of Portuguese in the development of Kitúba (as suggested by Holm 1989) went beyond that of simply motivating the contacts. In other territories that later became Portuguese exploitation colonies, especially Mozambique and Angola, no linguistic developments similar to Kitúba have been reported, in contrast to settlement colonies such as the Bight of Biafra islands, where Portuguese creoles developed.
20.
I mention this matter only in order to report Fehderau's position in full. I do not find the koine/pidgin distinction useful, except sociologically or politically (Mufwene 1988a, 1989a, in press). Similarly, sociological and political, but not formal linguistic, criteria justify the dialect/language distinction, on which the koiné/pidgin contrast is based. Note that the extent of structural differences between the lexifier and its koiné or pidgin is determined in part by the extent of typological kinship between the language varieties in contact (Thomason 1983, Mufwene 1986, Keesing 1988, Thomason & Kaufman 1988). As revealed by the comparison of Lingála and Kitúba in Mufwene (1989a), the dialect/ language distinction does not appear to be relevant to the differences. Lingála, which from the beginning seems to have developed out of contact of different languages, has in
Kitúba
205
some ways remained closer to the Bantu morphosyntactic stereotype than has Kitúba, despite the development of Kitúba, according to Fehderau, from contact among Kikóngo "dialects." For instance, Lingála has preserved a subject-verb agreement system in which subject prefixes support pro-drop, a head-modifier concord system, and a reflexive prefix, although it has lost the object prefix. Kitúba has none of these features, even in the variety used in the Lower Zaire region, the Bakóngo territory. 21.
Fehderau (1966) acknowledges the presence of West Africans in the Lower Zaire region during this period, especially during the construction of the railroad, but he assigns them no role in the formation of Kitúba. Given his position on the genesis of this language variety, he probably assumed that the West Africans just learned it well enough to communicate with the local Africans.
22.
I infer the presence of West-Atlantic languages in the Lower Zaire from Fehderau's (1966) mention of Senegal as one of the places from which West African labor was imported. Samarin (1986) also mentions them.
23.
Lingála developed on the northern course of the Zaire River (see map). Unlike Kitúba, it has preserved much of the Bantu agglutinating system in the verb form, including a rich suffixal tense-aspect system (Mufwene 1989a). According to most Zairean linguists, Lingála arose out of contact among primarily Bantu-speakers. However, as for Kitúba, Samarin, and also Hulstaert (1989), argue fairly plausibly for the conclusion that Lingála, too, developed thanks to the West African colonial auxiliaries.
24.
Regarding the morphosyntactic selections made by Kitúba, compared to Lingála, the question remains as to why Kiyánsi did not have the same impact on Lingála, since it was also involved in the riverine language contacts out of which Lingála developed. The answer depends perhaps on whether or not other Bantu, non-Kikóngo languages involved in the contacts that resulted in today's Kitúba are similar to Kiyánsi in lacking subjectverb agreement. Research underway suggests that at least some dialects of Kiteké have morphosyntactic features similar to those of Kiyánsi. It is possible that differences in the timing of the contacts and/or in the proportion of such languages in the contacts account for these formal disparities between Lingála and Kitúba.
25.
Nowadays French, alone or combined with higher education, will not earn any native a high social status without money. French with money gives more prestige than one of the national lingua francas with money but without French (i.e. without much schooling).
26.
Like Samarin, Hulstaert thinks that Lingála did not exist before the European coloniza tion of Central Africa, and that the principal agents of its formation are foreigners who were accommodated by the indigenes (1989:102-3). He thinks that the natives helped the new language variety normalize by reproducing the foreigners' speech, which they dubbed "missionary talk" (my free translation, based on how the indigenes usually reacted to the way most missionaries spoke Kitúba or Lingála).
27.
It may be argued that because of the multitude of languages in contact, and the power advantage it offered, the emergence of Kitúba as a lingua franca must have made it less necessary for the Bantu speakers to learn other traditional Bantu languages. Yet, as the acquisition of any of the latter must have guaranteed a more intimate and trustworthy rapport with its speakers, especially those native to the region surrounding the colonial post or factory town, some of the Bantu labor must have found it helpful to learn at least one other indigenous language in addition to Kitúba.
206
Salikoko S. Mufwene
28.
After Morton Stanley's expedition in the last quarter of the nineteenth century, Western Zaire's contact with English speakers was almost only with the Protestant missionaries. Businesses were generally run by Belgians. The Portuguese, whose presence remained significant in the region, were only merchants.
29.
To be sure, there were also several Zanzibari in the colonial labor force, who had been recruited on the example set by Henry Morton Stanley on his first trip to the mouth of the Congo River, as he started from Zanzibar on the Indian Ocean coast. Most of these must have spoken Swahili. The fact that this lingua franca was not adopted in the western part of today's Zaire reflects the Belgian colonial policy to work with the indigenes in one of their local major languages, which is not necessarily in conflict with the modified Samarin hypothesis endorsed in this essay. The Zanzibari have been ignored in this discussion because several of the issues I make concern primarily the role of non-Bantu speakers in the formation of Kitúba.
30.
The differences do not, however, hinder mutual intelligibility, at least not significantly.
31.
At least in the Bandundu subvariety of the eastern dialect, mandef is said with three lov/ tones.
32.
Generally, the Bandundu city variety is believed to have been much influenced by Lingála, which is the lingua franca a few miles north and west on the Kasai River. Despite the prestige of Lingála, this popular belief is yet to be demonstrated.
33.
William Samarin (personal communication, 1992) subscribes to this position. He adds: "It is quite likely that the foreign African workers of the French and the E[tat] I[ndépendent du] C[ongo], who moved about so frequently and in such numbers between the Lower Congo [River] and the Middle Congo, did not distinguish between two languages. Following Thomason and Kaufman, I believe that they might not have had a single model they were aiming at. Of course, they noticed differences, but the local languages were all Bantu, and one used one's ingenuity to make oneself understood. If this is reasonable, it is another reason for believing that it was local people who took the jargon and stabilized it/'
34. Matuka does not provide the tones. I have adjusted the glosses.
References Bokamba, Eyamba. 1977. "The impact of multilingualism on language structures: The case of Central Africa". Anthropological Linguistics 19.181-202. Chaudenson, Robert. 1979. Les créoles français. Evreux, France: Editions Fernand Nathan. Chaudenson, Robert. 1988. Créoles et enseignement du français. Paris: L'Harmattan. Daeleman, Jan. 1972. "Kongo elements in Saramacca Tongo". Journal of African Lan guages 11.11-44. Daeleman, Jan. 1982. African origins of Brazilian black slaves: Linguistic criteria. The Mankind Quarterly 23.89-117. Dahl, Östen. 1985. Tense and aspect systems. Oxford: Basil Blackwell. Fehderau, Harold. 1966. The origin and development of Kituba. Ithaca, NY: Cornell University dissertation.
Kitúba
207
Guthrie, Malcolm. 1953. The Bantu languages of Western Equatorial Africa. London: Oxford University Press. Heine, Bernd. 1970. Status and use of African lingua francas. München: Weltforum Verlag. Hinnebusch, Thomas J. 1989. "Bantu". The Niger-Congo languages: A classification and description of Africa's largest language family, ed. by John Bendor-Samuel, 450-73. Lanham: University Press of America. Holm, John. 1989. Pidgins and creoles, vol. 2: Reference survey. Cambridge: Cambridge University Press. Hulstaert, Gustaaf. 1989. "L'origine du lingala". Afrikanistische Arbeitspapiere 17.81114. Jouannet, Francis. 1988. Des tons à l'accent: essai sur l'accentuation du comorien. Aixen-Provence: Service des Publications, Université de Provence. Kapanga, Mwamba Tshishiku. 1991. Language variation and change: A case study of Shaba Swahili. Urbana: University of Illinois dissertation. Keesing, Roger M. 1988. Melanesian Pidgin and the Oceanic substrate. Stanford: Stanford University Press. Lumwamu, François. 1973. Essai de syntaxe systématique des parlers kongo. Paris: Klincksieck. Matuka, Yeno Mansoni. 1990. Three Kongo palavers: Wedding, bereavement and recon ciliation. Ms. Mufwene, Salikoko S. 1979. "Some grammatical changes and variations in KikongoKitúba and their implications for language planning". Paper presented at the Tenth Annual Conference on African Linguistics, Urbana: University of Illinois. Mufwene, Salikoko S. 1986. "The universalist and substrate hypotheses complement one another". Substrata versus universals in creole genesis, ed. by Pieter Muysken and Norval Smith, 129-62. Amsterdam: John Benjamins. Mufwene, Salikoko S. 1988a. "Formal evidence of pidginization/creolization in Kituba". Journal of African Languages and Linguistics 10.33-51. Mufwene, Salikoko S. 1988b. "Why study pidgins and creoles?" Column, Journal of Pidgin and Creole Languages 3.265-76. Mufwene, Salikoko S. 1989a. "La créolisation en bantou: les cas du kituba, du lingala, et du swahili du Shaba". Etudes Créoles 12.74-106. Mufwene, Salikoko S. 1989b. "Some explanations that strike me as incomplete". Journal of Pidgin and Creole Languages 4.117-28. Mufwene, Salikoko S. 1990a. "Time reference in Kituba". Tense-modality-aspect sys tems in pidgins and creoles, ed. by John Singler, 97-117. Amsterdam: John Ben jamins. Mufwene, Salikoko S. 1990b. "Creoles and Universal Grammar". Linguistics 18.783807. Mufwene, Salikoko S. 1991a. Review of John Holm's Pidgins and Creoles, vols. 1 & 2. Language 67.380-87. Mufwene, Salikoko S. 1991b. "Language genesis and human evolution". Diachronica 8.239-54. Mufwene, Salikoko S. 1991c. "Pidgins, creoles, typology, and markedness". Develop-
208
Salikoko S.
Mufwene
ment and structures of creole languages: Essays in honor of Derek Bickerton, ed. by Francis Byrne and Thorn Huebner, 123-43. Amsterdam: John Benjamins. Mufwene, Salikoko S. 1994. "Restructuring, feature selection, and markedness: From Kimanyanga to Kituba". Proceedings of the Twentieth Annual Meeting of the Berke ley Linguistics Society: Special session on historical issues in African linguistics, ed. by Kevin E. Moore, David A. Peterson, and Comfort Wentum, 67-90. Berkeley: Berkeley Linguistics Society. Mufwene, Salikoko S. In press. "Jargons, pidgins, creoles, and koinés: What are they?" Pidgins and creoles: Structure and status, ed. by Arthur Spears and Donald Winford. Amsterdam: John Benjamins. Ngalasso, Mwatha M. 1984. "Pidgins, créoles ou koinès? A propos de quelques langues véhiculaires africaines". Cahiers de l'Institut de Linguistique de Louvain 9.135-61. Ngalasso, Mwatha M. 1991. "Tons ou accents? Analyse des schèmes intonatifs du kikongo véhiculaire parlé dans la région de Bandundu (Zaire)". Etudes Créoles 14.147-62. Ngalasso, Mwatha M. 1993. "Les procédés repétitifs in kikongo: le redoublement et la reduplication". Topics in African Linguistics, ed. by Salikoko Mufwene and Lioba Moshi, 45-66. Amsterdam: John Benjamins. Polomé, Edgar. 1963. "Cultural languages and contact vernaculars in the Republic of the Congo". Texas Studies in Literature and Language 4.499-511. [Reprinted in Lan guage, society, and paleoculture: Essays by Edgar . Polomé, ed. by Anwar S. Dil, 116. Stanford: Stanford University Press.] Samarin, William J. 1982. "Colonization and pidginization on the Ubangi River". Journal of African Languages and Linguistics 4.1-42. Samarin, William J. 1986. "Protestant missions and the history of colonialism". Journal of Religion in Africa 16.138-63. Samarin, William J. 1989. "Language in the colonization of Central Africa". Canadian Journal of African Studies 23.232-49. Samarin, William J. 1990. "The origins of Kitúba and Lingala". Journal of African Languages and Linguistics 12.47-77. Thomason, Sarah G. 1983. "Chinook jargon in areal and historical context". Language 59. 820-70. Thomason, Sarah G., and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Valdman, Albert. 1983. "Creolization and second language acquisition". Pidginization and creolization as language acquisition, ed. by Roger Andersen, 212-34. Rowley, MA: Newbury.
Sango* Helma Pasch Universität zu Köln
1. Introduction Sango is the national and (next to French) the second official language of the Central African Republic (CAR),1 one of the few African nations where a single language is understood practically everywhere throughout the country. (See map of the Sango area.) Its speakers number about three million. Despite its official status, Sango is used primarily for oral communication; French is the main medium for written communication. Most inhabitants of CAR speak Sango as a second language. Their mother tongues are the ethnic languages of the country, most of which — Banda, Gbaya, Zande, Ngbaka-Ma'bo, etc. — belong to the Ubangian or Eastern branch of the Niger-Congo family. In the northwestern part of the country some Mbum languages of the Adamawa branch of Niger-Congo are spoken (Pana, Kare, etc.), and in the north and the east are several languages belonging to the Central Sudanic branch of the Nilo-Saharan family (Sara, Runga, Yulu, Kara, etc.). In the southwest and the southeast small groups of speakers of Bantu languages (Kako, Aka, Mbimu, etc.) can be found.2 The four main ethnic groups of Muslim immigrants, who live scattered throughout the country, speak Arabic, Ful, Hausa, and Kanuri, respectively. Only in Bangui, the capital, where since independence (1960) many inhabitants have no longer used the ethnic languages or taught them to their children, has Sango generally acquired the status of a first language (Pasch 1992b). This Bangui variant of Sango is also generally accepted as the standard form, although it was never officially declared as such. This is especially true for a subvariant used in radio broadcasting which has only a few words of
—
■
"
Pasch
'
Helma
210
Sango
211
French origin, words which are commonly heard in everyday speech. Sango emerged out of contact among several languages. Its development is the result of the activities of the French and Belgian colonial powers in the Ubangi River bend during the last two decades of the 19th century.3 The lexifier language is Ngbandi, an Ubangian language — specifically, Yakoma and Riverine Sango, two riverine Ngbandi dialects which are practically identical (Pascal Boyeldieu, personal communication, 1987). It has been claimed that Sango is a creole which developed out of a pidgin (Samarin 1971, 1982a,b, 1984-85, 1986b, 1989b), and it has also been suggested that it is merely a variant of Ngbandi, albeit one which is "issue du Ngbandi, langue parlée par les Sango et les Yakoma de Haut-Oubangui" (Diki-Kidiri 1978: 111). In this paper I will try to demonstrate that Sango originally arose as a pidgin which later underwent creolization, and that this development shows clear features of both pidginization and creolization. The first step is to ask what sociohistorical and sociolinguistic conditions must exist for pidgins and creoles to develop, and the next step is to determine whether or not such conditions that prevailed at the time Sango emerged. I will then investigate the linguistic processes that were involved in the development of Sango and attempt to determine whether they reflect pidginization and creolization. In the course of this paper I will look at the development of the phonol ogy, tonology, morphology, syntax, and lexicon of Sango. Ideally, for this purpose the structural features of Pidgin Sango should be compared with those of its particular lexifier dialects. Unfortunately, however, there are almost no data available on Pidgin Sango and not many on either Riverine Sango or Yakoma. Forms and structures that modern Sango shares with Ngbandi must therefore be taken as remnants of Pidgin Sango. Lexical terms of Riverine Sango and Yakoma are scattered unsystematically throughout the 1978 dictionary of Bouquiaux et al. Yakoma was investigated by Boyeldieu (1975, 1982), but his linguistic description is limited to the conjugational paradigms of verbs. Consequently, Ngbandi, as described by Lekens (1923, 1952, 1955, 1958) and by Toronzoni (1989), will represent the lexifier language in my comparison. This should result in little distortion, as both Yakoma and Riverine Sango are very similar to the variant of Ngbandi described by Lekens (Tanghe 1940:110f.). Ideally, too, the specific features of languages that were in contact with the lexifier at the relevant period should be compared with Pidgin Sango in
212
Helma Pasch
order to search for substrate contributions to the pidgin. But here too there are difficulties. Most of the other vernaculars spoken in the Ubangi bend and neighboring areas are also not well described. A notable exception is Banda, for which there is a detailed grammar by Cloarec-Heiss (1986), although this is difficult to use for comparative purposes. There are also vocabularies and short grammars for Zande (Gore 1931, Lagae 1921, Tucker & Bryan 1966) and for Ngombe (Rood 1958). Because there is so little direct attestation of early vehicular (or Pidgin) Sango, the structural comparison in this paper is based primarily on modern Sango. It seems safe to assume that Sango had already stabilized as a lan guage when the first publications on it appeared — Brachiel's vocabulary (1909; cf. Samarin 1982a:4) and the grammars of Giraud (1908), Calloc'h (1911), Clerc (1911), Eboue (1918), and, later, Tisserant (1950). These works are neither rich in data nor detailed in description, but the material they do contain is consistent with modern data.4
2.
The sociohistorical and sociolinguistic conditions for the development of Sango
In order to investigate the origin and development of Sango, it is necessary first to analyze the historical and social circumstances for the relevant period, since it is these, rather than linguistic factors, which are in the first instance responsible for linguistic interference (cf. Hellinger 1985:93, Stolz 1986:248, Thomason 1986, Thomason & Kaufman 1988). This is true both with regard to the direction of interference and to its extent, and, by extension, to the development of pidgins and creoles as well. Thomason (1986:264) and Thomason & Kaufman (1988:35), for example, acknowledge the relevance of linguistic factors in contact-induced language change, but at the same time clearly state that these are of secondary importance. According to Foley's definition (1988:162), a pidgin is a contact lan guage consisting of an amalgam of linguistic elements of two or more languages. It results from extended contact between socially highly unequal groups of people with no language in common, and it evolves when these groups create a new language in order to fulfill the needs of verbal communi cation. Holm (1988:4-5) claims that the languages of the different groups involved should not be closely related, because otherwise koineization or
Sango
213
dialect-leveling rather than pidginization results. Mufwene (1988:33) consid ers the difference between koineization and creolization to be a gradual one and allows for the source languages to be closely related. So do Thomason & Kaufman (1988:169), who even allow interintelligibility between pidgin and creole languages and their lexifiers, at least with respect to Kitúba and Chinook Jargon. An important sociolinguistic criterion for a pidgin is that it is not the mother tongue of any of the groups using it (Hymes 1971:43), although there may be some individuals who learn it as a first language (Thomason & Kaufman 1988:169f.). Sango developed at the end of the last century in the Ubangi bend as a result of the Belgian and French conquests along the Zaire and Ubangi rivers (Samarin 1982a,b, 1984-85, 1989a,b, Pasch 1992a,b). The historical condi tions for the development of Sango, as Samarin describes them, tended to favor the development of a pidgin language. The French and Belgians had brought along non-local African workers, many of whom were of West African origin; these were often generally called Senegalese or Hausa, and their languages included Ful, Wolof, Bambara, Soninke, Susu, Kru, Vai, Khasonke, etc. (Samarin 1982b:417). Other workers came from the lower part of the Zaire River, e.g. Bakongo (Samarin 1986a:214f.), and there were possibly also Swahili, who had crossed the continent with slave traders (cf. Samarin 1984-85:345f., 1989a: 173), and Bantu speakers from along the banks of the Zaire and Ubangi rivers. Samarin claims that these foreign African workers, in addition to their other tasks, also served as linguistic interpreters, and that they were mainly responsible for the communication between the Europeans and the local population groups (1982b:411-12, 198485:334, 1989a:234). Some of these workers, i.e. those who had worked for the Europeans from the beginning of the conquest of Central Africa, used the Bantu-based contact language Kitúba for in-group communication purposes.5 The West Africans among them would have learnt it at the lower part of the Zaire River, where they had started working for the Europeans and where they were responsible for the emergence of Kitúba (Mufwene, this volume). This as sumption is supported by Mufwene's hypothesis that Kitúba was the language of the labor force from West Africa and other areas outside the Congo territory. Their West African languages, which differed widely from each other, played a very insignificant role in in-group communication.6 These languages could not compete with Kitúba, which was the lingua franca (or
214
Helma Pasch
vehicular language) of all or most of the workers. The other foreign workers, especially those hired farther upstream, also spoke Bantu languages such as Lingala and Ngombe. The Ubangi River up to Bangui/Zongo, which the French and Belgians, respectively, founded within a few days at the end of June 1889, could be navigated with steamboats. But from that point onwards, transport had to be carried out on canoes borrowed from the riverine population (Samarin 198485:312). This meant that the Europeans depended on the help of the local population not only for food and labor but also for canoes, and on the willingness of the owners of these canoes to cooperate with them (Samarin 1984-85:338-41, 1989a:117). The peoples around Bangui — the Ngbaka-Ma'bo, the Bondjo, and the Buzeru — showed little readiness to cooperate with the Europeans, but the Gbanziri, who lived farther upstream, were willing to do so. Soon, at least from 1890 onwards, they turned out to be the one group that cooperated the most with the French and probably also with the Belgians (Samarin 198485:338-41, 1989a:157).7 But very soon, in 1890-91, the Belgians began to employ the Yakoma as their local helpers. After some time the Yakoma began to work in increasing numbers for the French as well, and they soon became the most important ethnic group to cooperate with the Europeans (Samarin 1984-85:343-46, 357). Even before the arrival of the Europeans, some variant of the Yakoma language is likely to have served as a vehicular language throughout the Ubangi bend as far down as Bangui/Zongo and possibly even farther down stream. This lingua franca had probably not yet become a pidgin (Diki-Kidiri 1982, Pasch 1992a:20-l). The linguistic competence of the peoples in the Ubangi bend by the end of the 19th century was as follows. The autochthonous population used their ethnic Ubangian languages, which are not mutually intelligible, for in-group communication, and they used Yakoma as a lingua franca. The foreign African workers brought in by the Europeans used Kitùba as their in-group medium; other Bantu languages, such as Lingala or Ngombe, were also in use. The fact that some of these foreign workers served as interpreters for the Europeans means that they also knew French. Since these two groups, i.e. the autochthonous population and the foreign African workers brought in by the Europeans, did not have a language in common, the latter tried to learn the local lingua franca, i.e. Yakoma, but due to insufficient access to this lan-
Sango
215
guage they could not acquire it completely. As a result, a new language, vehicular Sango,8 was created, with Yakoma as the target language. It is quite probable that the foreign African workers were mainly respon sible for the creation of the new language, just as the West Africans had been responsible for the creation of Kitúba. In communicating with the local population they tried to use the local lingua franca, but since they did not have a complete command of it they made mistakes. According to Bouquiaux et al. (1978:21), the Yakoma deliberately used a simplified form of their own vernacular with the newly arrived groups as well as with the non-riverine peoples, and it may therefore be assumed that the other riverine peoples using vehicular Yakoma did the same. I assume that the Ubangians went even further in their attempts to accommodate to the newcomers, and that they adopted the latters' variant of the language, which then stabilized. This assumption is in agreement with Thomason & Kaufman's view (1988:174f.) that pidgin genesis results from mutual linguistic accommodation. In the beginning the developing lingua franca, Sango, was mainly a medium of communication between the foreign African workers of the Euro peans and the local population. According to Samarin (1982b:412) the Euro peans hardly used it at all, and most communication with the local population was carried out not by the Europeans but by their nonnative African workers who served as interpreters. The French government did, however, favor the diffusion of Sango throughout the Oubangui-Chari territory. If the interpreters regularly used the emerging Sango for mediating communication between the colonial military forces and administration, on the one hand, and the native population on the other, then this lingua franca must have been spread quickly throughout the territory, primarily by nonnative speakers. French, the language of greatest socioeconomic and political prestige, had no chance of becoming the target language of the new lingua franca. The interpreters among the African workers used French as a means to communi cate with their European masters, but the majority of the Africans probably did not have sufficient access to it to learn it. It therefore follows that Sango played a role similar to that of Lingala and Kitúba, both of which spread as lingua francas in their respective territories during the conquest and under the rule of French-speaking Europeans. In all three cases, the French language was very much limited in use. It is an exaggeration to claim that the Europeans normally did not communicate with the local population directly, but only through their inter-
216
Helma Pasch
preters (cf. Samarin 1982b:412ff; Pasch 1992a:22). They did communicate directly, and the main medium of communication was not French but Yakoma. Contrary to Samarin (1982b:414), I assume that the French, the Belgians, and the other Europeans who followed them as merchants and missionaries began to learn Yakoma because they were often not happy with the translations of their interpreters (cf. Samarin 1982b:412). This assumption is confirmed by Godart & Zoubé (1986:171), who evaluated the archives of the first Catholic Mission of Bangui, St. Paul. They claim that the appearance in Bangui on 2 June 1911 of Calloc'h's grammar and vocabulary, which was written explicitly for administrators and merchants, was much appreciated by the Europeans. The new language that was created by speakers of various ethnic lan guages in less than ten years was used only in a few specific domains: military and administration, and the provision of food and labor. In the beginning it was characterized by the French and Belgian conquerors as a highly reduced form of riverine Sango. This is indicated by the early expressions "patois", "sabir", "volapük" (Samarin 1982a:8), "corrupted Sango" and "Sango tí turugu" ("soldiers' Sango"; Samarin 1984-85:359).9 These names show that the Europeans working in the area of the Ubangi bend realized that the language was highly reduced in form and function. Sango soon became the language of trade as well, especially long distance trade. The traders in the region have traditionally been foreigners, often Muslims from Tchad and Cameroon, who normally do not and did not know the local lingua francas in the territory of the Oubangi-Chari colony. Sango became an important means of communication in the markets and shops in the townships, and its command an indispensible tool for succeeding there.10 After the Europeans had established themselves on the Ubangi River, a kind of intraterritorial mobility began, caused by the needs of conquest and administration. When the French headed north to fight the slave hunters Rabah and Senoussi (cf. Gentil 1902), they needed large numbers of porters. These were recruited mainly among the population of the actual CAR terri tory, and it may be assumed that they were controlled by the foreign African workers of the French, who gave their orders in Sango. Under the guidance of these foreign Africans, Sango was also spoken among the canoers and porters in the conquest expeditions to the east. Administrative problems, too, made it necessary for the French to move some of the foreign African workers to
Sango
217
different posts in the newly conquered territory, because the French them selves did not have enough personnel, and they also had difficulties in recruiting a local labor force (Goutalier 1954:303). Thus many posts were held only by the nonlocal African workers, normally Zanzibari or Senegalese (Samarin 1984-85:314). It follows that non-Yakoma11 were responsible for the spread of pidginized Sango from the very beginning. As time passed the situation changed, and the expatriates doing administrative work in the vari ous posts were gradually replaced by Yakoma, who later also provided most of the militia members in the French Haut Congo (Samarin 1984-85:344f., 357f.). Interference from Kitúba (§3.3 below), which must have fallen out of use shortly after the emergence of Sango (since it is not indicated in any historical source), indicates that the Yakoma also used the emerging Sango rather than their own vernacular or the old Yakoma lingua franca. Furthermore, Sango may have played a role in organizing the communal work in the collecting and planting of rubber plantations, other forced labor, and the construction of the Congo-Océan railway in the territory of what is now the Republic of Congo, because the present-day Central African Repub lic was the main recruitment center for this labor (Godart & Zoubé 1986: 53ff.). Sango must also have been an important means of communication in the "village de liberté", located next to the mission de la St. Famille at the eastern end of the Ubangi bend, which existed from 1897 to 1929 (Godart & Zoubé 1986:101). Families and children of different ethnic origins lived there, most of whom had sought refuge from the slave hunters. And Sango was certainly a major means of communication in the mission stations St. Famille and St. Paul, where the missionaries provided secular and religious education to children of different ethnic backgrounds. Some of these children had been bought by the missionaries from slave traders; others were orphans, and a third group were sent there by their parents (Godart & Zoubé 1986:93). In the course of time Sango spread throughout the country, strongly supported by the missionaries and traders. Its functions and structure ex panded so that it became a language that could meet all the needs of daily communication, that is, a creole. Sango did not, however, become the first language of any group until about the middle of the 20th century, when it became the first language in Bangui. It may be concluded that conditions in the Ubangi bend at the time when Sango developed were appropriate for the development of a pidgin. Several
218
Helma Pasch
socially unequal groups of people were brought together in a short period of time who had no language in common and no chance to learn each other's language. Even the target language, chosen by the group in power, could not be acquired, and a new language had to be created. A comparison with the conditions under which many other pidgins and creoles developed does reveal a number of significant differences. These do not, however, tell against the hypothesis that Sango emerged as a pidgin. First, most of the pidgins in the Atlantic, Indian, and Pacific oceans developed as trade languages or as a result of a slavery and plantation economy. This does not strictly apply to Sango, although the transport system was heavily based on forced labor. In addition, although population movement was involved, in the case of Sango it was not movement of large groups; and, unlike the insular slave societies, in the Ubangi bend it was the militarily stronger group and their workers who moved, i.e. the group having the political power, while the conquered groups remained in their original territory. More important is the fact that the Europeans as a rule deliberately promoted their own languages, Dutch, English, French, or Portuguese. This was not the case with Yakoma, the lexifier of Sango, despite the facts that European colonial activities did cause the emergence of Sango and that the choice of the target language was made by the Belgians and the French, i.e. the groups with political power. It becomes obvious that the choice of a lexifier does not solely depend on its prestige, but on the prestige of those who make the choice. It must be taken into account that Yakoma was the most prestigious among the non-European languages in contact. The Yakoma, remaining in their own territory, were economically independent and had easy access to all means of survival for which the Europeans' foreign African workers had to ask them. Thus they had greater socioeconomic prestige than the foreigners, although these latter had more political and military power (Pasch 1992a:21). The new lingua franca, Sango, which was in the process of developing, became for the Central Africans practically the only medium through which they could gain access to the white society; that is, it took over the role of the vertical medium that Yakoma had initially had. As a result, the new language developed a higher socioeconomic prestige than its lexical base language. Yakoma maintained its status as a horizontal ethnic language.
Sango 3.
219
Linguistic structure
A comparison of Sango with Ngbandi and the other source languages reveals a number of retentions, losses, and innovations in Sango. It must be admitted, though, that in the case of some innovations, especially certain lexical items, it cannot be determined whether they are substrate contributions or borrow ings. The fact that in the case of language shift only a few lexical terms from substrate languages are usually retained (Thomason & Kaufman 1988:114) only apparently suggests that the words entered the language after it had stabilized. If we consider that the vocabulary of areas such as body parts, plants, and tools was probably not present in the early pidgin phase of Sango and that Yakoma lost its role as a target language shortly after the emergence of Sango, then it seems likely that this vocabulary of modern Sango is the result of substrate contributions rather than of borrowing.12 The formation of some function words by way of grammaticalization processes can also not be assigned with certainty to the early period of development, and it is possible that some of them emerged only after Sango became stabilized. But it is certain that grammaticalization processes had already begun in the pidgin phase.13 There are, in fact, no innovations in Sango which are to be explained solely by universal developmental tendencies of creole languages. In most cases either one or more contributor languages can be determined as sources or the development can be explained by language-internal rules and restric tions or grammaticalization processes. 3.1. Phonology Sango has retained almost the entire phoneme inventory of its lexical base language. Ngbandi and Sango are therefore in a small degree mutually intelligible (Pasch 1993:281).14 In this respect Sango differs from most Euro pean-based creoles, which for phonological reasons alone are not mutually intelligible with their lexifier languages. The quasi-complete retention of the phonological system of the lexifier is, however, not unique. Kitùba has also retained enough of the phoneme systeme of Kikongo — all five vowels and most consonants — that there is some mutual intelligibility between the two languages.
220
Helma Pasch
3.1.1. Vowels The vowel phonemes of Ngbandi are the same as those of Sango. In the latter language nasalized vowels are far less frequent than in Ngbandi, and in most cases nasalization is optional. Vowel length is distinctive neither in Ngbandi (Lekens 1923:1) nor in Sango, although in both languages there are a few words which are regularly pronounced with long vowels. NGBANDI
SANGO
(Lekens 1923:1-2, Toronzoni 1989:55)
i
u e
o
ĩ
ü
(Bouquiaux et al. 1978:29)
i e
u o
ĩ
ữ
εɔεɔã aãa Lekens (1923:1), in his presentation of vowel phonemes, distinguishes be tween /a/ and /a/, a distinction which he then disregards in the rest of his grammar. It may or may not be safe to assume that this distinction is not phonemic in Ngbandi, but I omit it here for lack of information. According to Boyeldieu (1975:21), the vowel system of Yakoma differs from those of Sango and Ngbandi in that it lacks /a/, so that it has only four nasalized vowel phonemes. Neither Lingala nor Kitúba has nasalized vowel phonemes, whereas nasalized vowels do occur in most Ubangian languages (Bouquiaux et al. 1978:28f.). French may also have played a role in the retention of nasal vowels in Sango, though their infrequency in Sango may reflect the influence of Lingala and/or Kitúba. 3.1.2. Consonants All the consonant phonemes of Sango are also found in Ngbandi. In addition, Lekens (1923:4) lists five labialized phonemes in Ngbandi: /kpw/,/bw/,/gw/, /kw/ and /nw/. The first four of these occur in Sango as sequences of stop plus semivowel, while the latter has no counterpart in Sango. Toronzoni (1989:90) lists the first four of these but not /nw/ for Ngbandi, and he also lists /pw/, /dw/, /tw/, /fw/, /zw/, /sw/, and /lw/. He notes, however, that the phonemic status of these "séquences consonantiques ... constituées soit de deux pho nèmes ... soit de trois phonèmes" is not unquestioned (Toronzoni 1989:83). It therefore seems likely that these labialized sounds do not represent single phonemes but rather sequences of consonant and semivowel. Lekens also
Sango
221
mentions /kpm/ as a phoneme; the status of this is not quite clear. Since Lekens does not list /kp/ among the phonemes, and since all words with initial /kpm/ in his dictionary (1958) precede a nasalized vowel, it is likely that the nasal component is the result of coarticulation and not a distinctive feature of the stop phoneme, which would then be /kp/. NGBANDI
(after Lekens 1923:3-4 and Toronzoni 1989:74, 77, 90)15 t kp kpw p bw d g gb gbw b ng ngb nd mb f s v z nv nz n nw ny m 1 w h
SANGO
(Bouquiaux et al. 1978:21) P t b d mb f v mv m
k g nd s z nz n l/r
kp gb ng
ny
w
h
In Yakoma, as in Sango, [h], [?] and Ø occur as free variants of one phoneme /h/ (Boyeldieu 1975:21). For Ngbandi, Toronzoni (1989:71) says that /h/ has the facultative allomorph Ø, which according to his examples, however, looks more like a free variant. In Sango, /h/ has the free variants [h], [y], and 0 ; before low vowels except /a/ it has the free variants [h] and [w]. Before /a/, however, [y] and [w] do not occur as variants of /h/ in any of the languages. The phonemic status of [1] and [r] is unclear in Sango. They occur in many cases, even in loanwords, as free variants. This happens regardless of the environment; however, [r] occurs word-initially only in a very few loan words, and if both [1] and [r] occur in one word the first is normally realized as [1] and the second as [r] (Bouquiaux et al. 1978:30). Only a small number of words — personal pronouns and a few basic nouns — are always realized with either [1] or [r] (Diki-Kidiri 1982:90-92, Bouquiaux et al. 1978:30f.). In Ngbandi, the distributional rules are stricter: / does not occur before nasalized vowels, and in intervocalic position it is realized as [r] (Toronzoni 1989:69). In Yakoma, [1] and [r] also occur almost as free variants, though [r] does not appear in word-initial position. According to Boyeldieu (1975:18f.) the distribution follows a complicated set of phonotactic rules; in intervocalic
222
Helma Pasch
position [n] is a third variant of the same phoneme, which he denotes /L/. 16 The phonological resemblances between Sango and Ngbandi are such that shared lexical items, as well as common borrowings, are generally fully identical. Typical examples are given in (1). (1)
NGBANDI
SANGO
dà zíà bí punda kpwi
dà zíà bì púndà kpi/kwi
"house" "to let" "servant" (< English) "donkey" (< Swahili) "to die"
The reason for the almost total retention of the phonemic inventory of the base language in the development of Sango could be due to the fact that the Ngbandi phonemic inventory is shared by Lingala and Kitúba. These two languages also have the phonemes /nt/ and /mf/, and Kitúba also has /pf/ and /ts/ in addition; for speakers of Lingala and Kitúba, who took part in the creation of Sango, the Ngbandi phonemic inventory was thus familiar though smaller than their inventories. The Bantu-based contact language Lingala, like many other northern Bantu languages but unlike the majority of Bantu languages, even has labial-velar stops. Both Lingala and Kitúba have prenasalized stops and fricatives.17 These Bantu languages also have [1] and [r] as free variants, but not [h], [w], or [y]. The phonemes of Sango are also found in the neighboring Ubangian languages, but some of the more complex sounds of these languages are not found in Sango, e.g. the labial-velar nasal /ngm/ of Gbaya and the labiodental flap /V/ of Banda-Linda (cf. Bouquiaux et al. 1978:29). The implosives /ɓ/ and /ɗ/ do occur in Sango, but only as free variants in loanwords; that is, they are not phonemic. The seven-vowel system common to most source languages was retained in Sango. The Sango phone mic inventory is not, however, simply the common denominator of the contributor languages, since the rules for the distribution of [r] and [1] had to be learned or created by speakers of the substrate languages. An exception among the source languages is Kitúba, which has only 5 vowel phonemes and no labial-velar stops. The Sango phonemic inventory is thus more complicated in this respect when compared to that of Kitúba.
Sango
223
3.2. Tones Ngbandi is described by Lekens (1923:7) as a tone language with three tone levels: high, mid and low. It has six tonemes, three register tones and three contour tones (low-high, high-low, low-mid). The contour tones, which Toronzoni (1989:98f) describes as sequences of tonemes ("deux tonèmes en succession"), occur far less frequently in the texts than the register tones. Although tonal variants are frequent, Lekens (1923:5) emphasizes the fact that mistakes of tone would make the language unintelligible, and he assumes that the functional load of tone in Ngbandi is greater than in other Sudanic18 languages. Ngbandi has both lexical and grammatical tone. Lekens (1923:10) char acterizes the former as a "usually" unchangeable feature of lexical items, especially with respect to nouns, adjectives, adverbs, and conjunctions. (2)
a. b. c. d.
L: yà M: ya H: yá HL: yâ
"woman" "plantation" "wind, blow" "to pick up"
Grammatical tone is used to mark deverbal derivations and tense/aspect in verbal conjugation, which in most cases is combined with segmental morphological marking. Lexical tone is retained in Sango. Grammatical tone is retained in Sango only as a secondary marker of deverbal nouns (see §3.3.1). The loss of tonal functions as part of the reduction processes in the development of Sango is comparable to that in Kitúba, where tone is also no longer used to mark tense and aspect (cf. Mufwene 1988:35). The loss of tone is not, however, an inevitable feature in pidgin/creole genesis; many Atlantic creoles have tone systems that certainly do not come from their European lexifier languages (Holm 1988:137-43). 3.3. Morphology Sango has four bound morphemes, the subject marker à-, a suffix -ngo that marks deverbal nouns, the plural marker a-, and the nomina agentis marker wà-, all of which were retained from Ngbandi. However, a number of Ngbandi affixes, especially those marking verbal extensions, were lost.
224
Helma Pasch
3.3.1. Deverbal derivations In Ngbandi and Yakoma there are at least three types of deverbal derivations which are marked tonally and/or by segmental morphology. The first type is marked only by tone. The tone of monosyllabic verbs is generally raised when they are nominalized; in nominalizations of bisyllabic verbs, the tones of both syllables are raised if they are low. If they are high-low they become low-low, mid-mid, or low-mid in the nominalized form (Boyeldieu 1975: 95f.).19 (3)
Ngbandi: a. nzì b. tέ c. d d. tò e. sàrà f. kr
"to "to "to "to "to "to
steal" eat" dance" pierce" itch" cough"
→ nzi → tέ → d → tò → sara → kɔrɔ
"theft" "food" "dance" "spear" "itch" "cough"
The derived nouns usually denote either the instruments with which the actions of the verbs are executed or the processes or results of the respective basic verbs. In the latter case they may appear as the direct object of the verb:20 (4)
Ngbandi: a.
lò kpwε kpwε. 3sg flee flight "S/he flees (a flight)." (Lekens 1923:126)
b.
lò nɔ n. 3sg go course "S/he goes (a course)." (Lekens 1923:126)
c.
h lo tέ a-tέ! give 3sg food 3sg-eat "Give him food to eat!" (Lekens 1958:905)
This derivational pattern was lost in Sango, and it appears that the cognate object construction with deverbal nouns marked by the suffix -ngɔ (see below) partly took over this function. A number of lexicalized forms were, however, retained: (5)
Sango: a.
lò tene tene. 3sg say word "S/he spoke/talked (a word)."
Sango
225 b.
álà fn fɔnɔ. 3pl walk(V) walk(N) "They walked (a walk)."
The second way of marking deverbal nouns in Ngbandi is by means of reduplication. It occurs above all with nouns that function as attributive adjectives, ones which Toronzoni (1989:124) calls "adjectifs déverbatifs". (6)
a. Yakoma: b. Yakoma:
de né
Ngbandi: de
"be cold" "be heavy"
→ →
"be cold"
→
d. Ngbandi: gbî "make warm"→
dèdê "cold" nènê "heavy(ness)" (Boyeldieu 1975:102) dedè "cold, calm(ness)" (Lekens 1958) gbìgbî "warm" (Toronzoni 1989:125)
The loss of this word formation pattern in the development of Sango is another example of morphological reduction, despite the fact that a few derived forms were retained and lexicalized. (7)
Sango: ngu à-dé. water SM-be.cold "The water is cold."
dedê ngú. cold water "cold water"
Samarin (1971:136) sees in the loss of reduplication a characteristic feature of pidginization in Sango, while in many other instances the development of reduplication is considered characteristic of pidginization. In the third pattern of building deverbal nouns in Ngbandi, tone is simply a secondary means of marking nominalization, the suffix -ngɔ being the primary means. This suffix carries a mid tone which assimilates the tone of the last syllable of the base verb. If the verb is bi- or polysyllabic, the tone of the first syllable receives a low tone (Lekens 1923:82, Toronzoni 1984:130). This derivational pattern has been retained in Sango, albeit in a slightly modified form, and is the only way of deriving deverbal nouns in Sango. According to Diki-Kidiri (1977:75), ALL syllables of nominalized verbs bear mid tone, but according to the Sango-English dictionary (1980:vi) they only GENERALLY do so, i.e. not invariably.
226
Helma Pasch (8)
Ngbandi: a. nzi b. bàtà c. là
(9)
Sango:
"to steal" → nzi-ngɔ "stealing" "to keep, herd" → bata-ngɔ "herding" "to sleep" → la-ngɔ "sleep"
a. sùkùlà "to wash" → sukula-ngɔ "washing" b. píkà "to beat" → pika-ngɔ "beating" c. zingɔ "to wake up" → zingɔ-ngɔ "awakening"
These deverbal nouns always refer to the action of their basic verbs, never to the results. Thus, when following the basic verb they only apparently occupy the position of the direct object; their function is instead that of an adverb emphasizing the verb (Pasch 1985:20). (10) Ngbandi: a. álà kpwε gi kpwε-ngɔ. 3pl flee only flee-NOM "They keep on fleeing." (Lekens 1923:148) b. alà gwé gwe-ngɔ. 3pl go go-NOM "They just/simply go." (Lekens 1923:148) (11) Sango:
a.
lò ku ku-ngɔ. 3sg wait wait-NOM "S/he just waited (for no special purpose); s/he kept on waiting."
b. mbi to susu tɔ-ngɔ. lsg fry fish fry-NOM "I fried the fish (and did not boil or grill it)." c. e yé tí híngà hinga-ngɔ. lpl want SUB know know-NOM "We just want to know (for the sake of knowing)." In spite of their morphological similarity, the deverbal nouns are not functionally identical in Sango and Ngbandi. Several of the functions that Lekens describes for Ngbandi do not exist in Sango, and I assume that they were lost during the development of Sango. On the other hand, some of the functions of the deverbal nouns in Sango are not documented for Ngbandi. These are among the innovations which qualify this language as a separate entity, distinct from its base language (cf. Weinreich 1963:139).
Sango
227
One example concerns the marking of the progressive and the habitual. The progressive is marked in Ngbandi by the verb de "go on, continue" + só "so that" + the nominalized form of the verb or by dε + nà "at" + the nominalized form of the verb (Lekens 1923:73). In Sango both the progres sive and the habitual, as well as the future, are expressed by means of the auxiliary yέkέ + the infinitive of the main verb. (12) Ngbandi: a. ló dε só ha-ngɔ ní ko. 3sg go.on so.that say-NOM DEF DEM "He was telling it." (Lekens 1923:73) b. ló de na te-ngɔ yé. 3sg go.on PREP eat-NOM thing "He is/was eating something." (Lekens 1923:73) (13) Sango:
lò
yέkέ tέ kbέ. 3sg AUX eat food "S/He is/was eating."
A second example is the modal + complement construction: in Ngbandi complements of modal verbs are nominalized verbs, whereas in Sango they are infinitives, preceded by the subordinatór tí. (14) Ngbandi: . híngà kĩ-ngɔ da ngá? 2sg know build-NOM house EMPH "Could you build a house?" (Lekens 1958:404) b. lo da tí gwe-ngɔ. 3sg be.about SUB go-NOM "He was about to leave." (Lekens 1923:90) (15) Sango:
a. m híngà tí sárà dà? 2sg know SUB make house "Can you build a house?" b. bàbà tí mbi à-ga tí gwè. father SUB lsg SM-come SUB go "My father was about to leave." (Bouquiaux et al. 1978:107)
In Sango only adjuncts of purpose, but not complements, are expressed
228
Helma Pasch
by deverbal nouns. They are introduced by tí. But even this construction is optional, infinitival forms being the usual alternative. (16) Sango: a. mbi gw-è tí mú kεkε. lsg go SUB take wood b. mbi gwè tí mu-ngɔ kεkε. lsg go SUB take-NOM wood "He went to get a stick." (Samarin 1967:116) 3.3.2. Verbal extensions A series of verbal extensions is documented in Ngbandi and Yakoma. Some of these, besides their tonal marking, are marked by reduplication of the first syllable, others by a suffix, and a third group by both reduplication and a suffix. These extensions serve to mark the frequentative, the intensive, the reciprocal, etc. (Boyeldieu 1975:95-107, Toronzoni 1989:229-36). (17) Yakoma: a. b. d. e. f.
tέ tàrà kpí gbi te sú
"to eat" "to try" "to die" "to burn" "to meet" "to flee"
→ tέ-tέ → tà-tàrà → kpí-rí → gbi-ri → té-ngbi → sú-ngà
"to eat frequently" "to try several times" "to die numerously" "to burn completely" "to meet each other" "to leave behind"
Some fossilized forms have been retained in Sango, but the pattern is no longer productive. (1.8) Sango: a. b. d.
sú "to tear" → sú-rú fa "to cut" → fá-rà ma "to hear" → má-ngbì bì "to throw" —» bì-ngà
"to tear to pieces" "to hack to pieces" "to understand each other" "to disperse" (Bouquiaux 1978:45f.)
Most of the functions that are performed by verbal extensions in Ngbandi are expressed periphrastically in Sango. Only for the reciprocal has Sango developed a new grammatical pattern, which is constructed on the basis of the reflexive. 3.3.3. Reflexive and reciprocal Sango retained its marking of the reflexive in both form and structure from its lexifier language. In both Ngbandi and Sango the reflexive is expressed by a
Sango
229
possessive construction in the object position, with tèré "body" as the head and a pronoun as a modifier coreferential with the subject; the only difference between the two languages is in the possessive marking. In Sango the pronoun is linked to tèré by means of the subordinator tí, while in Ngbandi it directly follows the head, since an alienable relationship obtains. The pattern of marking the reflexive by means of the denotation for "body" is found in a number of Atlantic creoles (Holm 1988:204), but it is of course not exclusive of creoles, but is found in many other languages too — as shown by the Ngbandi example (19a). (19) a.
b.
Ngbandi: à-so tέrέ ló. 3sg-beat body 3sg "He beat himself." (Lekens 1923:89) Sango:
alà sùkúlà tέrέ tí álà 3pl wash body SUB 3pl "They washed themselves."
In Sango the reciprocal is not marked by the suffix -ngbi, as it is in Ngbandi and Yakoma. Instead, Sango has developed a new marking which is based on the reflexive. Syntactically, tèré "body" has again the role of the direct object, but in the reciprocal construction it remains unspecified for person. (20) a.
Ngbandi: àlà s-ngbi. 3pl beat-RECIP
"They beat each other." (Lekens 1923:90) b.
Sango:
álà píkà tέrέ. 3pl beat body "They beat each other."
In the Ubangian language Banda, reflexive and reciprocal are marked by the same markers ("réfléchi réciproque"). These are lexicalized possessive constructions consisting of t ("person") plus a pronoun which is coreferen tial with the subject (Cloarec-Heiss 1986:103-5): (21) Banda:
nje
yí
tnje.
3pl
PERF.love
REFL/RECIP.3pl
"They love themselves/they love each other." (CloarecHeiss 1986:105)
230
Helma Pasch
This construction might be the source for the Sango reciprocal construc tion. If it is, Sango has innovated, in that it has developed a new reciprocal form different from the reflexive. Lingala and Kitúba are not possible sources for the reciprocal construction in Sango, since both mark reciprocal by the suffix -ana (cf. Everbroeck 1969:92f., Swift & Zola 1963:222f.); this same suffix marks the reciprocal in many Bantu languages. 3.4. The verb phrase The subject is marked in Sango by a prefix or by subject pronouns which directly precede the verb, and aspect (or mode) is marked by periphrasis or by tone on both the pronoun and the verb. In Ngbandi, conjugation for subject is likewise marked by pronouns preceding the verb, but here the subject markers include information about aspect and mode and also about number. 3.4.1. Marking of tense, aspect and mode Like the other Ubangian languages (cf. Boyd 1989:206), Ngbandi does not have tense, but it does mark aspect and mode. Perfective/realis21 and imperfective/irrealis are marked on the subject pronoun and on the verb; number (plural) is marked only on the verb. (22)
Ngbandi: a. REALIS:
bi//lo bì "I/you/he threw" elĩlálà bí "we/you/they threw" mbi/mɔ/lo ka "I/you/he gave" ká "we/you/they gave" e/ĩ/álà (Lekens 1923:66) b . IRREALIS:
mbí/mó/ló bì élĩ/álà 'bi álà 'bi mbílmólló ka élĩ ka álà 'ka
"I/you/he will throw" "we/you (pl.) will throw" "they will throw" 'T/you/he will give" "we/you (pl) will give" "they will give" (Lekens 1923:71)
For Yakoma, Boyeldieu (1975:40-48) distinguishes three aspectual cat egories, perfective, imperfective, and irrealis (which he calls "positif, in achevé, volitif").
Sango
231
(23) Yakoma:
bì d mbi d mbi d lò d ló d lò d è d é dɔ é d
"I pull" "I would pull" "that I pull" "s/he pulls" "s/he would pull" "that s/he pull" "we pull" "we would pull" "that we pull" (Boyeldieu 1975:45)
In the verbal conjugation of Sango, only the irrealis may optionally be marked by tone. A high tone precedes the first syllable of the verb (Diki-Kdiri 1978:181) and thus falls on the pronominal subject. It is used to mark the conditional and, according to Diki-Kidiri, the hypothetical. The first of these may also be expressed by the conjunction tongànà " i f without any special tonal marking, and it usually is. The second is expressed by the conjunction ádù introducing the protasis. It is usually combined with ká "(adversative) and, but", introducing the apodosis (Bouquiaux et al. 1978: 139). (24) Sango: . bàtà ngbo nà dà fàde lò 2sg.coND keep snake PREP house then 3sg kírì lò tε mɔ. return 3sg bite 2sg "If you keep a snake in the house, it will turn back and bite you." (Diki-Kidiri 1978:183) b. tongànà mɔ bàtà ngbo nà dà... if 2sg keep snake PREP house molenge á-zingo. child SM.coND-wake "The child might wake up." (Diki-Kidiri 1978:183) d. m/έngε à-zíng... child SM-wake "The child woke up." e. âdù CONT
mbi ká
mbi gwè nà
ngòndà
1sg
1sg
bush
2sg but
go
PREP
"If I were you I would not go into the bush." (Bouquiaux et al. 1978)
pεpέ. NEG
232
Helma Pasch
A second way of marking habitual aspect in Ngbandi is by means of the suffix -ndo, which has high tone in the realis and mid tone in the irrealis. (25) Ngbandi: a. REALIS:
mbí ndó-bì . 1sg HAB.REAL-throw
"I threw." (Lekens 1923:69) b. IRREALIS: lá kwέ mbi
ndo-tέ
mbipa.
day all lsg HAB.IRREAL-eat pigeon "I'll eat pigeon every day." (Lekens 1923:74) This marker has been lost in Sango, in which habitual aspect is marked in the same way as the progressive and the future, by the auxiliary yέkέ. The almost complete loss of morphological marking of aspect and mode in Sango probably occurred during the pidginization phase. In the expansion phase, then, periphrastic constructions took over the functions. 3.4.2. The subject marker àThe prefix à- functions in Sango as a subject marker for nominal subjects and as an indefinite/impersonal/nonhuman subject pronoun. Most of its functions are identical to the ones it has in Ngbandi. In Ngbandi, even in written texts, it has also the function of a personal subject pronoun. In Sango à- is used as a personal subject pronoun in speech but hardly ever in written texts, where the personal pronouns lò and álà are used. My consultants generally rejected the use of the prefix à- as a personal subject pronoun, and it appears that the language is in the process of regularizing the distinction between personal and nonpersonal pronouns in subject position. (26) a. Ngbandi: à-ngba
yàkéré
INDEF-stay little
Sango:
"still a bit" (Lekens 1923:64) à-língbì pepέ. INDEF-SUffice NEG
"It is not sufficient/admissible." (Clerc 1911:319) b.
Ngbandi: bi à-vu. night SM-black "The darkness spread." (Lekens 1923:63)
Sango
233
c.
Sango:
bî à-vu. night SM-dark "The darkness spread." (Bouquiaux et al. 1978)
Ngbandi:
a-tɔ sa yá ní tε 3sg-cook meat COMP LOG eat "He cooked meat so that he would eat." (Lekens 1923:161) lò tô nyàmà títέnέ lò tέ. 3sg cook meat so.that 3sg eat "He cooked meat so that he would eat." (Lekens 1923:161)
Sango:
According to some examples in Eboué (1918) and even in Samarin (1967), the independent pronoun lò "s/he" and the prefix à- can be used interchangeably in subject position in Sango. But when they occur in series, for example when the subject pronoun is modified, lò stands in the first position and à- in the second position (cf. Samarin 1967:140f.). (27) a.
à-mú nà mbéní zò ndé. SM-give PREP INDEF person other "He gave it to a different person." (Samarin 1967:76)
b.
lò yἐkἐ hándà mbi. 3sg AUX deceive lsg "He is deceiving me." (Samarin 1967:74)
lò mvἐní à-píkà lò. 3sg self SM-beat 3sg "He beat him/her himself." (Eboué 1918:13)
d.
lò (à-)yἐkἐ gà. 3sg SM-AUX come "S/He is/was coming." (Calloc'h 1911:17)
e. álà (à-yἐkἐ) gà. 3pl (SM-AUX) come "They came/come." (Calloc'h 1911:17) Since the choice between à- and lò as a subject pronoun for referring to human subjects is already documented in the early grammars of Sango, it
234
Helma Pasch
looks like a retention from Ngbandi. The use of à-, which otherwise functions as a subject marker, as a subject pronoun might, however, also result from substratai influence from Lingala, where for the present ("présent d'état"; cf. Everbroeck 1969:34) an apparently similar construction is used. The agree ment marker a-, used on the verb for human subjects, is also used pronominally. (28) Lingala: a. a-kosala. HUM.sg-work "He works."
b.
tatá a-kosala. father HUM.sg-work "Father works." (Everbroeck 1969:34)
3.4.3. The serial verb construction A Sango clause generally cannot have more than one conjugated verb; thus, for instance, sequences of actions are expressed by sequences of clauses. Pronominal subjects occur before each conjugated verb. For nominal sub jects, while the first verb has the subject marker à-, the subject of the second verb is the independent pronoun lò "s/he/it" or álà "they." (29) a.
b.
lo tε kobe, lò ny5 ngú. 3sg eat food 3sg drink water "S/He ate and drank water." kli ní à-tε kbἐ, lò ɔ ngú/*à-ny. man DEF SM-eat food 3sg drink water "The man ate and drank water."
There is, however, a construction which I call the serial verb construction (SVC). This fits the definition by Jansen et al. (1988:125), according to which the verbs in such constructions contain neither overt conjunctions nor com plementizers. Furthermore, none of the verbs is syntactically marked as an auxiliary of the other, but they nevertheless form together one single predicate. One function of the SVC in Sango is the marking of aspect. The verb kírì "return" as the first verb marks a single repetition of an action, and dἐ "stay", which can only be used with negative polarity, marks actions that have not yet taken place. (30) a.
Kosi à-kírì à-hé bíá. Kosi SM-return SM-sing song "Kosi sang once again."
Sango
235 b. Kosi à-kirî (nà ndo sò) lò hé bíá. 3sg sing song Kosi SM-return (here) "Kosi returned (here) and he sang a song." c.
Kosi à-dε à-mú wáli pε-ἐ. Kosi SM-stay SM-take woman NEG "Kosi is/was not yet married."
Another function of the SVC is to render factive verbs nonfactive. In these sentences tἐnἐ "say" is the second verb. (31) a.
Kosi à-balà-híngà à-tἐnε baba tí là à-kirî Kosi SM-see/know SM-say father SUB 3sg SM-return àwè. already "Kosi thought/was sure that his father had returned."
b. Kosi à-ba mbi ga. Kosi SM-see lsg come "Kosi saw me come." It should be mentioned that the SVC is not very common in the spoken language, and that in speech it is mostly heard from speakers in Bangui. It is, however, frequently used in written texts. The formal marking of the SVC is based on the categorial distinction between the personal pronouns Id (3sg) and àlà (3pl), on the one hand, and the subject marker à- on the other. Speakers who do not make this distinction also do not distinguish serial verb constructions from sequences of clauses. (32) mbéní finí fille ni à-londo, à-ga nà école... INDEF new girl DEF SM-arise SM-come PREP school "A new girl arose and came to school." (Samarin 1967:144) In other words, since the distinction between the personal pronoun ló and the subject marker à- is not made by all speakers, it is clear that the SVC is not a fully stabilized grammatical construction in the language as a whole. The origin of the Sango serial verb construction is difficult to determine. Although it is similar to that of Ngbandi, it differs from the Ngbandi SVC in that third-person subjects are marked both on the first verb and on the second; in addition, the functions of the SVC differ in the two languages. In the first position of the Ngbandi SVC, three verbs are used to mark aspect: ga "come"
236
Helma Pasch
"apporte la notion de rapidité" (Toronzoni 1989:524), gwe "go" indicates duration, and dili "return" indicates repetition. Only the last of these has a functional correspondence in Sango, in which kirí "return" represents a functional retention from Ngbandi. Furthermore, in Ngbandi the SVC may occur in series, a feature not documented for Sango. (33) Ngbandi: a. àlà gà gbð náko. 3pl come seize tortoise "They seized the tortoise." (Toronzoni 1989:524) b. z ga dili ho sa så. leopard come return kill animal two "The leopard again killed two animals." (Toronzoni 1989:527) A SVC with pronominal subjects, however, has the same syntactic structure in Ngbandi as in Sango, the subject preceding each verb. (34) Ngbandi: a. wa dala ání gá ání gwé. then above.all lsg come lsg go "Only then we can leave." (Toronzoni 1989:529) b. dî ne ya ko ni dî ni and 2sg return 2sg say that ? LOG return LOG to mbela. utter call "And you repeat that I must begin anew to make a roll-call." (Toronzoni 1989:529) The syntax of the Sango SVC is not likely to be a contribution from Kitúba, which also has a SVC (Mufwene 1988:41ff.). Neither the syntactic structures of the Kitúba and Sango SVCs nor their grammatical functions correspond; in Kitúba the SVC marks instrumentais: (35) Kitúba: maria haka mbelé bula yakála na Maria take machete hit man of "Mary hit her husband with a machete." (Mufwene 1988:41)
yandi. 3sg.POSS
The SYC in Sango cannot be completely explained, either formally or in its functions, as a result of superstrate or substrate contributions, although the
Sango
237
resemblance to the Ngbandi SVC indicates partial retention from the lexifier. At least some of the syntactic features could also be explained on the basis of language-internal restrictions. Specifically, there is a rule in Sango according to which only subordinate verbs appear in their infinitival form, while verbs constituting predicates are always directly preceded by their pronominal subjects. It appears that this rule is a generalization of the Ngbandi rule that requires only first- and secondperson pronoun subjects to be copied before each verb in a SVC. 3.4.4. The copula yèkè Sango has a copula, yèkè, that also serves as an auxiliary. It was not retained from the lexical base language Ngbandi, which has no copula and in which nonverbal predication is mostly expressed by juxtaposition (Samarin 1971: 125). In Sango this pattern is used optionally only with equational clauses; with locational predicates the use of the copula is mandatory. NGBANDI
(36) a.
è á-zi. lpl pl-people "We are people." (Samarin 1971:125)
SANGO
b.
zò à-yèkè zo. person SM-COP person "All people are equal." (Boganda, founder of CAR)
. nzàpa nzönl d. nzàpa God good God "God is good." (Lekens 1923:113) e.
(a-yèkè) nzoní SM-COP good
là nà tá tí . f. là yèkè nà tá tí màfútà. 3sg with pot SUB oil 3sg COP PREP pot SUB oil "S/he has a pot of oil." (Samarin 1971:125)
As an auxiliary, yèkè is used to mark the imperfective aspect in Sango referring to progressive, habitual or future events. Aside from the optional and rare tonal marking of the conditional, the marking of the imperfective is the only formal marking of aspect or mode in Sango, and it is the only one that is regularly used by all speakers. In Ngbandi, all these categories are marked differently. The habitual is marked by the prefix -ndo (37c), the progressive by the verb de plus the nominalized form of the verb (37e), and the future by a high tone, the marker of irrealis, on the subject pronoun preceding the verb (37a).
238
Helma Pasch (37) a.
Ngbandi: mbí
bi.
lsg.IRREAL t h r o w
"I shall throw." (Lekens 1923:71) b.
Sango:
mbi yέkέ gwe. lsg AUX corne "I am going/I will go" (Tisserant 1950:30)
c.
Ngbandi: zë ndo-tè gi sa. leopard -eat only meat "The leopard eats only meat." (Lekens 1923:98)
d.
Sango:
e.
Ngbandi: ló de na te-ngo yé. 3sg be.busy PREP eat-NOM thing "He is was eating." (Lekens 1923:73)
f.
Sango:
ló yέkέ ny5 sámbá. 3sg AUX drink alcohol "He is drinking alcohol/drinks habitually/is a drinker."
ló yέkέ te kôbέ. 3sg AUX eat food "He eating."
Samarin (1971, 1986a) calls yeke a verb meaning "to exist", but it is not a full verb. It can neither be nominalized by the suffix -ngo nor appear in subordinate position in the inifinitival form. Yέkέ can, however, be used like verbs with regard to past events. This means a generalization of rules in the contributor languages. In Ngbandi, copula-less clauses can refer only to the present tense unless they are general statements. In all other tenses and aspects a copula verb is required, usually du "to be somewhere". In Banda the copula də is also only used in the present, and the same is true for the copula ke(le) of Kituba. (38) a.
Sango:
gírírí mbi yeke ná so. formerly lsg AUX PREP village DEM "Formerly I was in that village." (Clerc 1911:365)
Sango
239 b. Ngbandi: to mbi à-dù àndo gbià tí father lsg SM-cop formerly chief SUB village "My father was the chief of the village." (Lekens 1923:124) c.
Banda:
d. Kitúba:
kowo rrrrr. iron COP red-hot "The iron is red-hot." (Cloarec-Heiss 1986:390) tata ikele? father COP "Is father here (available)?" (Swift & Zola 1963:111)
The Sango use of the copula yèkè in all tenses/aspects is a generalization of rules in the contributor languages, in which copula or copula-less clauses refer only to present events or states. This is an innovation which reflects the general tendency toward regularization that is characteristic of pidginization and creolization. According to Samarin (1986a), έkέ comes from Kitúba, which has an element (i)kclc/kc(lc) that also functions both as a copula and as an auxiliary. This etymology at first sight seems quite convincing since the Kitúba variant ike (Salikoko Mufwene, personal communication, 1992) or ke looks like a direct source form. This interpretation appears all the more likely in view of the fact that the Sango copula yέkέ occurs frequently in a shortened formkέ which is practically identical to the short form of the copula in Kitúba. (39) Kitúba: a. bantu yankaka ikele ve. people other COP NEG "There are no other persons." (Swift & Zola 1963:65) b. mwana ke maladi. child PREP lsg COP sick "My child is sick." (Swift & Zola 1963:29) mu ke
na
lsg COP PREP
kwenda . go
there
"I am going there." (Swift & Zola 1963:31) The problem with this proposal is that, while the short copula forms ke and kέ, are practically identical in the two languages, the long Sango form yèkè has no equivalent in Kitúba. Samarin (1986a:212) sees the missing link
240
Helma Pasch
in the Sango form a-ike (SM-COP) "it is, there is", which was documented in several examples by Villelune (1909) in his travel accounts: (40) Sango: a. kembe a-iké fa. rope SM-COP cut "The rope is cut." (Villelune 1909, cit. in Samarin 1986a:212) b. koli man
so
a-ike
na
sendeli
DEM
SM-COP
PREP
watchman
"That man is among the watchmen." (Villelune 1909, cit. in Samarin 1986a:212) These examples show that the Kitúba inflected form ike was reanalyzed as the stem of the copula, and in order to make it agree with nominal subjects it was given the Sango subject-marker prefix à-. The vowels /i/ and Id occur frequently as variants in Sango, despite their status as separate phonemes, and it is conceivable that ike varied with eke or ekε, which soon replaced the original form. The initial semivowel, finally, is explained by Samarin as the result of diphthongization. 3.4.5. Negation Ngbandi expresses negation by means of two discontinuous negative con structions. In the first an element má carries a high tone and follows the verb; this element is repeated at the end of the clause, where it carries a mid tone (Lekens 1923:83). (41) Ngbandi: mbi
kwi má mbi koi
lsg.IRREAL d i e
NEG l s g
ma.
One NEG
"I don't want to die alone." (Lekens 1958:556) The second Ngbandi negative construction consists of tá, which precedes the verb and which may optionally precede or follow the subject pronoun, plus a second element, ma, which again stands clause-finally (Lekens 1923: 83). (42) Ngbandi: a.
tá mb pa nvέnè ma. NEG 2sg tell lie NEG "You don't lie." (Lekens 1923:83)
Sango
241 b.
lò tá bi kekέ ma. 3sg NEG threw wood NEG "S/He did not throw a stick." (Lekens 1923:83)
Both of these negative constructions were lost in Sango, which has just one negative marker with two variants, pepέ and àpε; these are always clause-final. The form of the first variant is clearly a retention from Ngombe, where the absolute negator is pέpέ (Rood 1958:362). The form of the second variant may be of double origin, the first being the absolute negator ip of Ngbandi, the second the French negative marker pas. (43) Sango:
lò támbúlá kέtέ pepέ. 3sg walk little NEG "Il/elle n'a pas du tout (pas un peu) marché." (Tisserant 1950:216)
The position of pep έl àpe at the end of the clause is surely a contribution from Lingala or Kituba, where the negative marker is always clause-final.22 Additional sources for this position are Ngbandi, where the second element of the negator is always clause-final, and French, where in negative imperatives the negator often appears at the end of the clause. 3.5. The noun phrase The features of the Sango NP to be discussed in this section are plural a-, the agent-noun prefix wá-, determiners, numerals, adjectives, and possessive constructions. 3.5.1. The plural marker áThe Sango prefix á- "plural" is identical to that of Ngbandi, Banda, and Zande. Its syntactic position — it is a clitic preceding the first element of the noun phrase — is the same as in Ngbandi, but differs from the marker's position in the other two languages (cf. Tucker & Bryan 1966:88-90, 145-46). Plural is marked in Sango as it is in the neighboring Ubangian languages, primarily with denotations for humans but normally not for animals or inani mate objects. It may be concluded that the form and function of á- are of multiple origins, while its syntactic behavior is retained from Ngbandi.
242
Helma Pasch NGBANDI
(44) a.
nyi/á-ya child/pl-children "child/children"
SANGO
b. m lέngέ/a-m lέngέ child/pl-child
. bià zòlá-bía zi d. ko ta zòlá-kota zò real person/pl-real persons big person/pl-big person "a real person/real persons" "big person/persons" The few suppletive plural noun stems in Ngbandi were lost; all plural forms in Sango are regular. 3.5.2. The nomen agentis marker wàBoth in Ngbandi and in Sango the prefix wà- marks the nomen agentis, the possessor, or someone else having a specific characteristic with reference to the noun stem or the nominalized verb following it. The pattern is productive in both languages, and in both languages wà- is also found as a noun with the meaning "owner". NGBANDI
SANGO
(45) a.
wa-kwa NA-work "worker"
b. wà-kwa NA-work
wa-gbe NA-hunt "hunter"
d. wa-gi-ng NA-search-NOM
e.
wa-lu-ngo kâwà NA-plant-NOM coffee "coffee planter"
f. wà-lu-ng
nyama animal
káwà
It must be noted, however, that some speakers of Sango do not use wà- very often. They do not consider it good Sango, but rather as a Yakoma feature, and they therefore replace forms such as wàlung kâwà "coffee planter" with possessive constructions, e.g. zo tí lu-ng kâwà (lit. "person SUB plant-NOM coffee"). 3.5.3. Determiners Ngbandi has two demonstratives, ó lo "this here" and (d) ko "that there"; presumably Yakoma has (and had) the same set. Sango has lost these demon-
Sango
243
stratives, replacing them with an innovative form, só, the origin of which is unknown. NGBANDI
(46) a.
SANGO
mbo ó lo dog DEM nearby "this dog here" (Lekens 1958:158)
b. mbò só dog DEM "this/that dog"
. wáli (ó) woman DEM far "this woman here" (Lekens 1958:158)
d. wáli só woman DEM "this/that woman"
The definite marker ni, which in Ngbandi also has the function of introducing relative clauses, has been retained in Sango, although it no longer has the function of introducing relative clauses; that function has been taken over by the demonstrative so. (47) a.
b.
Ngbandi: là ma yé ni à-gbo ma gbekwà. 3sg hear thing DEF SM-roar as thunder "He heard something which roared like thunder." (Lekens 1958:727) Sango:
lò ma tcnc só, mbi tέnέ nà là, pepέ. 3sg hear word DEM lsg say PREP 3sg NEG "He did not listen to the words that I told him."
The Sango indefinite marker mbéní has emerged by reanalysis of the adjective mbá "other" plus the definite marker ni (< Ngbandi mba-ni, lit. "other-DEF", "the other one"). NGBANDI
(48) mbá ge other side "the other side" (Lekens 1958:563)
SANGO
mbéní yé (ndé) INDEF thing different "something (else)"
While the three determiners seem to reflect significantly different etymo logical developments, they have retained the same syntactic positions as the corresponding determiners in Ngbandi; for the definite and the demonstrative
244
Helma Pasch
markers, these positions also match those in Lingala and Kitúba. The pre nommai position of the indefinite marker is, however, also found in most other Ubangian languages (Tucker & Bryan 1966:137, 159). 3.5.4. Numerals In Sango the cardinal numerals follow the noun. The same word order can be found in Ngbandi and in the other Ubangian languages, as well as in Kitúba and Lingala, so that multiple sources for this feature are possible. NGBANDI
(49) a.
zo k i person one "one person"
SANGO
b. wali woman one "one woman"
KITÚBA
. bantu zole men two "two men"
In prenominal position Sango numerals have ordinal function. This is a retention from Ngbandi and other Ubangian languages, where the same pattern prevails (Boyd 1989:208). 3.5.5. Adjectives In Sango, as in Ngbandi and most other Ubangian languages, adjectives regularly precede the head noun. NGBANDI
(50) a.
SANGO
kota ε b. sioní homara big tree bad lion "big tree" "bad lion" (Lekens 1955:290)
ZANDE
. gbégberé mámá bad leopard "bad leopard"
According to Samarin (1967:61-63), adjectives can optionally follow the noun, a word order which is common to Bantu languages and to French and which would indicate free word order for adjectives. But so few examples have been documented, all from the beginning of this century, that the regular pattern appears to be ADJ-N. (51) ouali kété woman small "small woman" (Villelune 1909, cit. in Samarin 1986a:213) There are a number of complex nouns in Sango, composed of a noun followed by an adjective. They are considered lexicalized noun-adjective constructions, and they resemble constructions in Lingala or French.
Sango
245
(52) Sango: a. b. c. d.
b i kota (servant big) b, i kέtέ(servant small) bwa kota (father big) bwa kέtε (father small)
"cook, main servant" "kitchen helper" "missionary, priest, father" "vicar, chaplain"
Lingala: e. mo-to mo-lai 1-man 1-big "a big man" (Everbroeck 1969:27) f. ba-to ba-kuse 2-man 2-small "small men" (Everbroeck 1969:27) It seems to have been especially the French-speaking missionaries who used this construction in order to denote their servants and clergymen of different hierarchies. This becomes more apparent in view of the fact that they used pre- and postposed adjectives in Sango in analogy to French. Complex nouns as in (52a-d) were obviously used frequently enough to permit several authors of Sango grammars and dictionaries to regard the sequence noun-adjective as grammatical (e.g. Calloc'h 1911:4, Clerc 1911:303, and even Samarin 1967: 61-63 and Bouquiaux et al. 1978:152, 161). The following examples, cited from Bouquiaux et al. (1978), reflect very clearly the patterning of French pre- and postnominal adjectives. (53) Sango: a. kota dà big house "big house" b.
vs. dà kota housebig "palace"
kέtέ gàrá vs. gàrá kέtέ small market market small "small market" "market of secondary importance"
It should be noted, however, that this word order, which is documented only for the two adjectives kota "big" and kete "small", has become almost obsolete in Sango. It is used only to denote the younger and elder brothers and sisters of one's parents. Its use early in the developmental period of Sango is an indication of the variability in pidgins; regularization occurred when the language stabilized. (54) a. bàbá kota father big "elder brother of father"
b.
bàbá kέtέ father small "vounger brother of father"
246
Helma Pasch
3.5.6. Possessive constructions Sango retained its possessive construction from Ngbandi, but the marking of different kinds of possessive relationships found in Ngbandi and Yakoma has undergone reduction in Sango. There are three different kinds of possessive constructions in Ngbandi. In the first, used to mark inalienable possession and to build N-N compounds, the head and the modifier are juxtaposed. Alienable possession is marked by the possessive marker té, and prospective possession is marked by tí. (55) Ngbandi: a. to mbi father lsg "my father" b. yé té zò tí zo. thing POSS.AL person POSS.PROS person "A person's thing is for him/her." (Lekens 1958:906) In Sango, only the construction with the marker tí was retained, and it expanded its functions to become the only general possessive marker. This nondifferentiation in possessives, which is a simplification of the possessive constructions of the lexifier, is probably a contribution from the Bantu-based substrates, where analogous possessive constructions exist. (56) Sango:
a. bàbá tí father POSS
mbi lsg
"my father" b. ye so à-yèkè tí mbi. thing DEM SM-cop SUB lsg "This thing is for me/is mine." wáli ni à-Vv bέ tí m lέngέ tí lò. woman DEF SM-buy food SUB child SUB 3sg "The woman bought food for her child." Lingala:
d. buku ya ngáí book POSS lsg "my book" e. mbóka ya
y
village POSS
"your village."
2sg
Sango
247 Kikongo: f. mukanda ya book poss "Soft's book"
sof i Sofi
There are, however, many compound terms, some of which were prob ably retained from the base language and lexicalized. For some speakers, compounding by juxtaposition is no longer regarded as grammatical. Accord ing to them, the use of the subordinator tí is mandatory even in the formation of compound terms. (57) a. yángà dà mouth house "door" b. ngú Nzàpa water God "rain"
or
yángà tí dà mouth SUB house
or
ngú tí Nzàpa water SUB God
As in Lingala and Kitúba, attributes are frequently expressed by means of possessive constructions in Sango, which has only a few adjectives (cf. Pustet 1989). This function is not documented for Ngbandi. (58) a.
Sango: gbògbò tí we bed SUB iron "iron bed"
b. Lingala: mai ya moto water POSS fire "hot water" (Everbroeck 1969:51) The loss of the possessive marker tε in Sango might have been facilitated by its phonetic resemblance to the other possessive marker, tí, a fact which is significant enough to make Lekens (1923:16) ask whether the two forms are not perhaps the result of confusing the vowels ε and /, which in many Ngbandi words in fact occur as free variants. At the same time, however, he traces té back to tere "body" and tí to ti "hand", two etymologies which accord nicely with the different functions of the two possessive markers. A further loss concerns the word order in possessive constructions. In Ngbandi, the word order Head-Poss-Modifier may be inverted, with the head following the modifier. The function of the inversion is not quite clear, but it seems that it is used to emphasize the modifier.
248
Helma Pasch (59) Ngbandi: a. té là ngú mέ lò ndó-gb, là. poss 3sg water REL 3sg -bathe PRED 'This is the water in which he bathes." (Lekens 1958:906) b. ngú té lo... "His water ..."
In Sango this inversion is ungrammatical. It was presumably lost as a result of regularization. 3.6. Function words As in the case of bound morphemes, some Ngbandi function words were retained in Sango, while others were lost. Unlike the bound morphemes, however, some function words are contributions from French and Arabic, and in addition various innovations have resulted from grammaticalization pro cesses. 3.6.1. Prepositions Sango has only one locational and temporal preposition, nà "at, on, in, with", which is identical to the least specific Ngbandi preposition na "at, on, in". Since a preposition na is also found in other Ubangian languages, in Bantu languages, and in the Bantu-based contact language Kitúba, the Sango prepo sition may be interpreted as coming from multiple sources. (60) a.
Ngbandi: álà kárà sàmbîà na gέrέ là. 3pl strew pepper PREP leg 3sg "They put pepper on his/her leg." (Lekens 1958:618)
b.
Sango:
e gwe/dùti nà dà. lpl go/sit PREP house "We went/remained home."
c.
Lingala: to-fandi na mesa. lpl-sit PREP table "We are at the table." (Everbroeck 1969:99)
d.
Kitúba:
P. vand-áka na mbwála. P. BE-past PREP village "P. was in the village."23 (Mufwene 1988:45)
Sango
249 e.
Zande:
na
kumba
PREP
man
"with the man" (Tucker 1959:103) Ngbandi also has a number of other prepositions; these developed from relational nouns. Some examples are gbè "under, below; lower part of some thing", pέ "behind, after; back part of something", ndo "at, on; place", and yá "in; interior, belly". All of these items have retained their original lexical meaning in addition to their grammatical function as prepositions (examples from Lekens 1923:106f.): (61) Ngbandi: a. b. c. d.
pέ dà yá dà ndo mbi gbc kekέ
"behind the house" "in the house" "on me" "under the tree"
One Ngbandi preposition, kd "at, on", did not develop from a relational noun but rather from the adverb ká "there". Like the prepositions just cited, it has also retained its lexical meaning in addition to its grammatical function. (62) Ngbandi: ká tέrέ kεkέ at body tree "against (the body of) the tree" (Lekens 1958:310) With the exception of ná, Ngbandi prepositions lost their grammatical function in Sango, but the corresponding lexical items were retained as relational nouns: gbè "lower part of something, root"; pέ, forming part of the Sango word έ "back (part of something)" (< Ngbandi pε "back part of something" + "back", i.e. the body part); yd "inner part of something, belly"; and ndo "place". They have still retained some of their functions of marking locational and temporal relations, albeit in adverbial constructions introduced by the preposition ná. NGBANDI
SANGO
(63) a. ká (ndo) té gbia nà ndo tí gbia at (place) POSS chief PREP place SUB chief "at the chief's" (Lekens 1958:311) b. ká yá dà nà yá tí dà at in house PREP interior SUB house "in the house" (Lekens 1958:311)
250
Helma Pasch c.
gbe kekέ nà gbè tí keke under tree PREP lower.part SUB tree "under the tree" (Lekens 1958:197)
d. na gbe tànge nà gbè tí gbôgbô at under bed PREP lower.part SUB bed "under the bed" (Lekens 1958:618) e.
έ é nà έ behind lpl PREP back.part "behind us" (Lekens 1923:106)
tí e SUB lpl
f.
na pέ ko ni nà pèkô tí mbi at behind back lsg.LOG PREP back SUB lsg "behind me" (Lekens 1958:618)
g.
yé yá mbendà ne? yé nà yá tí mbénda thing in calabash what thing PREP interior SUB calabash "What is in the calabash?" à-yèkè nye? (Lekens 1958:1008) SM-cop what
Like the noun-derived prepositions, ka also lost its prepositional function in Sango, but it retained its lexical adverbial meaning: (64) Sango: gwè ká go there "go there" (Bouquiaux et al. 1978) It is obvious that the loss of the prepositions and the retention of the corresponding lexical items in Sango is a result of imperfect language acqui sition. The lexical meaning was learned, but not the grammatical function. This degrammaticalization and reanalysis might be due to substrate influence from Lingala, where most prepositional phrases also consist of the very general preposition or na "on, in, at, with" plus a relational noun. LINGALA
(65) a. olna PREP
SANGO
nsé
ya
g lomani
below
POSS
cupboard
"under the cupboard" (Everbroeck 1969:100)
b. nà gbe
tí
table
PREP below SUB table
"under the table"
Sango
251 . na
káti ya ebale middle poss river "in the middle of the river" (Everbroeck 1969:100)
PREP
d. nà popo tí á-zo PREP middle SUB pl-person "among the people"
As shown in examples (63a, b, d, f), prepositions often occur in series in Ngbandi, where the first, usually kà or nà, is comparatively general in meaning and the second is more specific. It is conceivable that, at the time Sango developed, the series of prepositions was reanalyzed by learners as a sequence of a preposition and a relational noun specifying the relation estab lished by the preposition. This reanalysis would have provided the model for locational prepositional phrases in Sango consisting of the general preposition nà and a relational noun. In Sango there are also a number of nonlocational prepositions that are built on the model of possessive constructions. These are teneti, ngbàngàti, and ndálítí, all of which mean "because of'. (66) a.
b.
è ga pepέ teneti kòbélà. lpl come NEG because.of sickness "We did not come because of sickness." (Bouquiaux et al. 1978:338) mbi lango pεpέ ngbàngàti nzàrà. lsg sleep NEG because.of hunger "I don't sleep because of hunger." (Bouquiaux et al. 1978:258)
c. mó ga ndálítí nyέ? 2sg come because.of what "Why did you come?" (Bouquiaux et al. 1978:240) These prepositions are found neither in Ngbandi nor in any of the substrate languages. They are lexicalized and grammaticalized possessive constructions which emerged within Sango as language-internal developments. They consist of the respective lexical items + the subordinator tí, the slot of the modifier being left open: tεnεti < tεnε tí X (lit. "affair SUB X") "affair/problem of X", ngbàngàti
252
Helma Pasch (67) a.
b.
Ngbandi:
á-sà kwé a~lí gi tongà ó ko. pl-animal all SM-do only like DEM DEM "All animals did like that." (Lekens 1958:947)
Sango:
a-nyàmà kwέ à-sárà gi tongànà pl-animal all SM-do only like "All animals just did like him."
là. 3sg
3.6.2. Conjunctions and complementizers In both Ngbandi and Sango, conjunctions are used less frequently than in western European languages. As Lekens points out, however, Ngbandi does possess a number of conjunctions, e.g. kandâ/kâ "when(ever)", ko "but", se "when, if, in case, as soon as", só "since (the long time that), lest", and be "although". Although most of these conjunctions have been retained in Sango, their use seems to be restricted to the written language. In the spoken language they are usually replaced by conjunctions of French origin, such as mε "but", dipt "since", si "if, and parce que "because, so that", and by one conjunction of Arabic origin, wàlà "or". The fact that none of these French conjunctions is listed in Clerc (1911), Calloc'h (1911), or Eboué (1918) might indicate that they are more recent borrowings. It should not be understood as proof of later borrowing, however, since it is quite possible that these authors were anxious to describe a "pure" language, free of foreign — especially French — influence, as are some missionaries and linguists working on Sango today. Tisserant (1950) gives me "but" and depuis "since", and I assume that at least these two French conjunc tions appeared in Sango early in its history. Three conjunctions developed in Sango through the same grammaticalization processes as those which led to the formation of the prepositions of purpose. That is, they are also grammaticalized possessive constructions, but here the modifier may optionally be filled by the demonstrative so. Two of these conjunctions are tenεti(so) "since, because" (< tεnε tí so, lit. "affair SUB DEM"), ngbàngàti (SO) "because, so that" (< ngbàngà tí so, lit. "judgment SUB DEM").
(68) a.
alà kpe ngbangati (so) mbi pikà alà. 3pl run because lsg beat 3pl "They fled because I had beaten them." (Bouquiaux et al. 1978:259)
Sango
253 b. mbi sárà yé só ngbàngàti (so) mbi yέkέ zò vuk . lsg do thing DEM because lsg COP person black "I do this because I am a black man." (Samarin 1967:102)
The third conjunction, tongànà "when, if' (< Ngbandi tongà + ná, lit. "like PREP"), also developed through grammaticalization. Its source is the preposition tongànà, which underwent expansion to gain the additional func tion of a conjunction. (69) tongànà m, yé m gwe nà Bangui. if 2sg want 2sg go PREP Bangui "If you want, you may go to Bangui." The conjunctions teneti (or its short form tέti) and tongànà were already documented in 1911 by Clerc and Calloc'h and then later by Eboué (1918), and it may be concluded that they were present in Sango as soon as it emerged. The development of these conjunctions might have been condi tioned by the fact that the Ngbandi conjunctions could not be learned by the majority of second-language speakers in the Ubangi bend. It appears that they also did not retain conjunctions from the other African source languages, so that they had to create new ones. These last three conjunctions occur frequently in written texts. It should be noted, however, that in everyday speech the conjunctions of French origin are used far more often than these three. The conjunction titene "in order that" (< titέnέ, lit. "SUB say", "in order to say") is a lexicalized subordinate construction. Samarin (1967:122) says that he documented its use only from town dwellers, i.e. inhabitants of Bangui, while country folk did not seem to use it unless they had considerable contact with town dwellers. This distribution suggests that the conjunction titene is a relatively recent development. Another conjunction, nà, has emerged as a result of expansion from the preposition nà. Some Sango speakers use nà only in its prepositional function, rejecting any other use as wrong. They mark the cördination of sentences by juxtaposition (70a). Such a usage might have been retained from Ngbandi, where nà occurs solely as a preposition. Other speakers, however, accept nà as a cördinating conjunction (70b), a usage that appears in all the Ubangian substrate languages as well as in Kitúba and Lingala.
254
Helma Pasch (70) a.
là tέ , lò nyö ngú. 3sg eat food 3sg drink water "S/He ate (food) and (s/he) drank water."
b. lò tε kóbέ nà lò nyö ngú. "S/He ate (food) and (s/he) drank water." The Ngbandi complementizer yá "that" (Lekens 1923:152) was lost in Sango, but a new one developed as a result of grammaticalization, again on the model of the verb tènè "to say": àtènè "that" (< àtènè, lit. "iNDEF-say", "they say, it is said") introduces complements of certain verbs of utterance and cognition, rendering the implicative reading of the latter a nonimplicative one (cf. Pasch & Givón 1988). (71) a.
cf.
mbi ma àtènè Pepe à-yèkè ga kékèrékè. lsg hear that Pepe SM-AUX come tomorrow "I heard that Pepe will come tomorrow."
b. mbi ma Pepe a-yèkè ga. "I heard Pepe coming."24
3.7. Word order The word order of Sango, SVO, is a retention from the lexifier Ngbandi and from the substrates. The type of facultative inversion found in French is not possible in Sango. (72) mbi to susu tɔ-ngɔ. lsg fry fish fry-NOM "I fried the fish (and did not boil or grill it)." The prenommai position of adjectives (§3.5.5), the postnominal position of determiners (§3.5.3) and numerals (§3.5.4), and the head-modifier order in possessive constructions (§3.5.6) have likewise been retained from Ngbandi. As noted earlier, the variable position of adjectives was apparently optional early in the twentieth century when the language was still young; the postnominal position was a contribution from French and/or Kituba, Lingala, and Ngombe. The postnominal position of determiners agrees with that of Ngbandi, but also with that of Lingala, Kitúba, and Ngombe, and the other Ubangian languages.
Sango
255
3.8. Lexicon Sango shares the majority of its basic and cultural vocabulary with Ngbandi: 46% of the items on the 100-word Swadesh list (cf. e.g. Lehmann 1968:11213), taken from Bouquiaux et al. (1978), and 76% of a selection of items from Eboué (1918) are fully or almost fully identical with Ngbandi words. The shared vocabulary allows, in certain contexts, a certain low degree of mutual understanding (compare 73a and 73b). (73) a.
b.
Ngbandi:
lo vo (á-) 3sg buy pl-chicken "He bought chickens." (Lekens 1923:91)
Sango:
lô vo (á-) 3sg buy (pl-)chicken "He bought chickens."
This difference in shared vocabulary of 46% and 76%, respectively, has a historical explanation. After Lingala took over the function of the lingua franca in Belgian territory, Sango maintained its lingua franca status only in French terrritory. Here, where the Yakoma constitute only a small riverine population, access to the Yakoma language was difficult for geographical reasons (see §2 above).25 Therefore speakers had to derive the necessary vocabulary from other languages for Sango, which was expanding in func tions and in areal distribution. French, Kitúba, and Lingala were the most important sources for lexical items after Yakoma had lost its place as the target and lexifier language. In this respect Sango differs from many other pidgins and creoles, which coexisted with their lexifiers after stabilization. In such cases the lexifiers remained primary sources from which the creoles could draw lexical items (cf. Holm 1988:74f.). It is true that a large part of the lexicon of the base language was lost in Sango. This applies not only to cultural vocabulary but also to basic vocabu lary. Some items have been subject to reduction by form selection; for instance, two bodily organs which in Ngbandi are denotated separately, gonda "heart" and be "liver", are both named be in Sango. The generic term kwa "hair", in combination with a modifier, is used in Sango to denote any kind of hair on birds, animals, and humans, e.g. kwa ti ndέkέ (lit. "hair POSS bird") "feather". With respect to humans it denotes hair on any part of the
256
Helma Pasch
body, e.g. kwa tí yángá (lit. "hair POSS mouth") "beard"; compare Ngbandi pu "feather" and ngbakwa "beard", respectively. It follows that the terms bέ and kwa have undergone semantic expansion and become less specific; if further specification becomes necessary it is expressed by means of modifiers in possessive constructions or by adjectives, e.g. kota bέ (lit. "big heart") "jealousy, envy", kwa tí gέrέ (lit. "hair POSS foot") "hair of the leg/foot". At the time Sango developed, as we know from the documentation by Calloc'h (1911) and Clerc (1911), some words of Ngbandi origin, e.g. ti "arm, hand" and hu "bird", were in use as variants in addition to words of Lingala origin, màboko "arm, hand" and ndèkέ "bird". In other words, at that time there was more variation in the original lexicon than there is today. When Calloc'h and Clerc wrote their grammars, other Ngbandi items of the basic word list were completely replaced by terms from Lingala and/or Kitúba, which were the most important substrate contributors to the lexicon; for instance, Ngbandi kpwo "wash" was replaced by Kitúba sùkulà. These words, along with a series of other non-Ngbandi terms not be longing to the basic vocabulary, were already documented in Clerc (1911), Calloc'h (1911), and Eboué (1918), and on the basis of such early documenta tion it may be concluded that they are in fact substrate contributions, most likely from Lingala and/or Kitúba. Among these items in Clerc (1911) are the following: sùkúlà "wash",26 kàmátà "seize", fúltà "pay", pikà "beat", sàndukù "box", fùngulà "open", tàmbúlà "walk", "spear", màfutà "oil", màkákò "monkey" (for this word Eboué gives bakuya, which is of Banda origin), màtàbisî "gift", mingi "much, many", nyàmà meat" (for this word Eboué gives mi, which is of Ngbandi origin). Calloc'h lists kùsâlà "work", "wealth", nzàrà "hunger", ndέkέ "bird" (for this term Clerc gives hu, which is a Ngbandi term), ndèkô "friend", and màboko (with ti as a variant, ti being the only term for "hand" listed by Clerc and Eboué). Eboué finally lists fùngulà "to open" (for which Calloc'h gives zi which is of Ngbandi origin) and mosika {màsikà according to Bouquiaux et al. 1978) "unmarried young man/woman". 27 Several of these examples might have originated in both Lingala and Kitúba since, along with many other words, they can be found in both languages. The possibility of double sources exists despite the fact that not all the words have the same tone patterns in both languages. The fact that the high(er) tone appears on the penultimate syllable at first sight makes Kitúba the more probable source for certain trisyllabic words, e.g. sùkúlà and
Sango
257
tàmbúlà, which correspond to súkòlà and tambólà in Lingala (see §3.2). It is also quite possible, however, that the respective items have a double etymol ogy, i.e. that the phonological form and the semantic content are contributions from both languages, but that they received the Kitúba-based tone pattern upon integration into Sango. The assumption that the Kitúba tone pattern has become operative in Sango for trisyllabic words of Bantu origin explains the difference in tone patterns between Sango and Lingala in some items, other wise identical, which are not documented in Kitúba. LINGALA
(74) a. kàmàtà b. màtàbïsl c. lôngôlà (Bouquiaux
SANGO
kàmatà màtàbisî lùngulà et al. 1978)
"to seize" "gift" "to remove"
French is as important a contributor of lexicon as are Kitúba and Lingala. Even in modern Sango, as well as in the Ubangian vernaculars, there are numerous loanwords from French. In Sango, recent borrowings from French are normally not phonologically adapted and can thus be easily distinguished from older contributions, which they are gradually replacing. Instead of "school" and sánzérí "watchman", for instance, many people now say école and sentinelle, respectively. Only a few lexical contributions can be found from other languages, including Banda,28 Arabic, Portuguese, and English: bamàra "lion" and kongbà "toad" (< Banda); bagàrá "cow" and sàndukù "basket" (< Arabic); mézà "table"29 and "rice" (< Portuguese); búku "book" and milki "milk powder" (< English). Most of the numerals have been retained from Ngbandi, but the words for "one", , and "ten", bàlè (lit."ten one"), have not. According to in Ngbandi, comes from Bouquiaux et al. (1978:273), the first, which is Lingala . Bouquiaux et al. (1978:272), however, list a variant of òkò, in riverine Sango, a variant of Yakoma, which might just as well be the source, unless double etymology is assumed for that word. The form bàlè in the word for "ten" is an innovation the origin of which is unclear. The Ngbandi term, sui, is given for "ten" by Clerc (1911:318) and by Calloc'h (1911:10), but Calloc'h also cites bàlè as a variant in addition to sui. Eboué (1918:11) and Tisserant (1950:14), however, give only bàle. The numerals are used when counting persons, animals, and objects other than money. For counting prices and money, French numerals have
258
Helma Pasch
almost completely replaced those of Sango. This is especially true for numer als larger than 10. The words for "arm, hand", "bird", "monkey", and "ten" in the early grammars and dictionaries suggest that variants of Ngbandi and Bantu origin, respectively, were frequent at the time the authors compiled their works. The coexistence of such variants probably did not last long, however, as becomes apparent by the fact that (for instance) sui, for "ten", is not documented after 1911. Relexification, a process which can be observed in some Europeanbased creoles, cannot be documented in Sango. One reason is that here the lexifier and the substrate languages are much closer to each other than in the case of creoles with European lexifiers and African substrates. A possible second reason is that there are no available dictionaries of Sango, its lexifier and its substrate contributors which are detailed enough to allow a systematic investigation of lexical replacements. In his grammar of Ngbandi (1923:99-103) Lekens lists four and a half pages of ideophones which denote the manner in which activities are carried out. All of them are reduplicated and some are tripled. The loss of the ideophones in Sango, like the loss of the verbal extensions (§3.3.2), is an example of loss of reduplication.30 (75) Ngbandi: a. à-tc bɔfú bɔfa 3sg-eat "he ate like a toothless old person" b. à-no véngbè véngbè. 3sg-go.down "he walked with a bent back" a-kpwε pérengu pérengu 3sg-run "he ran on crooked legs" d. - gbu gbu gbu "he rose quickly" e. - d ndu ndu ndu "he strolled/dawdled" Lekens does not provide any other information, but it may be assumed that many of these ideophones can only be used in combination with the verbs cited or with verbs that are semantically close, an assumption that William Samarin (personal communication, 1989) confirms with respect to the Ubangian lan guage Gbeya.
Sango
259
Almost all of these ideophones, as well as of those from the substrate languages, were lost in Sango, and the few that have been retained behave very much like other adjectives or adverbs, and they may even function as such — that is, they may qualify the action of any verb. Some of the exceptions are kpítípkítí, kíríkíri, and mbírímbírí, which emphasize the notion of blackness (Bouquiaux et al. 1978), crookedness, and straightness/straightforwardness, respectively. (76) Sango: a.
lo fá á-kongba tí yá tí dà 3sg cut pl-good SUB interior SUB house kíríkíri in.disorderly, way "He destroyed the things in the house left and right." (Samarin 1967:80)
b.
zo kíríkíri à-lingbï tí mú ɔrɔ pepέ. person any SM-can SUB take medicine NEG "Not just anybody can take the medicine." (Samarin 1967:81)
c.
pindiri à-vukɔ kpítípkítí charcoal SM-black pitch-black "Charcoal is pitch-black." (Bouquiaux et al. 1978:183)
The adverb ngbii "for some time", "which also occurs as ngbi ngbi, with as many repetitions as one desires, and ngbiii, with a vowel as long as one desires" and "which many speakers have identified with the French word jusqu'à" (Samarin 1967:79) appears to be a good example for showing how ideophones that have been retained in Sango have lost their original specific status and acquired that of a "normal" adverb. (77) lô bâ mbi ngbii. 3sg see lsg for.a.long.time "S/he looked at me for a while." (Samarin 1967:79) The loss of so many ideophones can surely be interpreted as the result of imperfect language acquisition and, in the example given, as a result of pidginization.
260 4.
Helma Pasch Conclusions
The comparison of Sango and Ngbandi shows that Sango is the result of a series of significant changes, losses, and innovations which occurred due to imperfect acquisition of Yakoma by second-language speakers. Sango con sists of retentions from its lexifier, on the one hand, and of contributions from various other Ubangian languages and also Kitúba, Lingala, and Ngombe, on the other. There have also been a number of language-specific innovations based on grammaticalization processes. The retentions from the lexifier are most obvious in the domain of the lexicon, although here too losses and substrate contributions from other languages can be found. The word class in which loss is most significant is the ideophones, where only a few examples were retained in Sango. Loss is most evident in specialized vocabulary, i.e. names for animals, plants, body parts, and tools. The degree of retention is, however, lower than in many European-based creoles, some of which have retained more than 90% of their lexicon from the lexifier. This difference can be explained by the fact that Sango soon gained higher prestige than its lexifier Yakoma, which by consequence lost its function as a source of new lexicon. Retentions from Ngbandi can also be found in the syntax, as well as in the phonology and tonology. Ngbandi word order patterns were retained in Sango, but in a reduced form: variations in word order were not retained. The Ngbandi phonemic system was similar enough to that of the substrate lan guages that it could be almost completely retained. Lexical tone was retained, together with the corresponding words from Ngbandi and the substrates. For trisyllabic words of Bantu origin the tone pattern of the Kitúba substrate, which has become operative in Sango, obtains quite generally. Unlike lexical tone, grammatical tone was almost completely lost, thereby resulting in losses in the marking of aspect and mode. Among the bound morphemes, the habitual marker was lost, as well as all the verbal extensions. Four bound morphemes, however, were retained: the plural marker, the marker of nomina agentis, the subject marker, and the suffix -ngɔ that marks deverbal nouns. Only this one means of forming deverbal nouns was retained; two others were lost. The functions of the deverbal nouns, which are not identical to those of Ngbandi, are Sango innovations.
Sango
261
The loss of the possessive marker te and the loss of juxtaposition as a means of marking inalienable possession implies the loss of differentiation in possessive constructions. The Bantu-based substrates have probably served as a model for this nondifferentiation. Regularization has likewise resulted from the loss of the two discontinu ous negative markers, which were replaced by a single new negator with two variants and a single syntactically fixed position. The form of one variant of the new negator is formally a contribution from Ngombe, and its syntactic position comes from Lingala and/or Kitúba. Other substrate contributions in the domain of function words, conjunc tions, and the copula have resulted in restructuring without regularization. Most conjunctions are of French origin, and one comes from Arabic; the copula yèkè can be traced to Kitúba. There have also been a considerable number of language-internal inno vations in Sango; among these are the prepositions of purpose, the comple mentizers and some of the conjunctions, the marking of topics, and the serial verb construction. These result either from the creation of new terms or from innovations in syntax, or both. On the basis of the results presented in this paper, I propose that Sango is a creole language which has emerged out of a pidgin, although restructuring is clearly less drastic than in languages like Tok Pisin and Saramaccan, in that it has affected mainly the tonology, the morphology (which in the lexifier language is not rich), and the lexicon, but little of the phonology and the semantics. According to the definition given in §2 above, which applies to a major ity of cases, a pidgin is a reduced language that results from extended contact between socially highly unequal groups of people with no language in com mon and that evolves when they create a new language to meet the needs of verbal communication. These sociolinguistic prerequisites obtained at the time Sango emerged. The colonial powers had brought together peoples speaking different languages who had no language in common and who had too little access to the target language, Yakoma, to acquire it completely. Therefore, they created a new language for which Yakoma constituted the lexifier. The choice of this target language, which originally had the function of a regional lingua franca, was a deliberate one, made by the Europeans. The readiness of the Europeans to learn a language of socially inferior status is in stark contrast to the behavior of Europeans in plantation societies, where the European languages constituted the only possible targets.
262
Helma Pasch
Since we do not have data on Sango at the time of its emergence, we do not know exactly what it looked like; that is, we cannot determine all the characteristics of a drastically reduced language. The fact that it was called a "corrupted [Riverine] Sango" or a "volapük" by early eyewitnesses (Samarin 1984-85:359) indicates that the language was indeed reduced. The reduction processes that Sango underwent can, however, be partly reconstructed. Reduction is still observable in modern Sango in the domain of the lexicon. There is no specialized vocabulary to denote animals, plants, body parts, or tools,31 for instance, a specific word for "heart", distinct from the word for "liver", is lacking, and kwa refers to all kinds of hair. This lack of a specialized vocabulary does not lead to communication difficulties, since the speakers borrow terms from French or some Ubangian language to denote the respective objects. They are, however, very much aware of the lack. Reduction is also visible in the almost complete loss of grammatical tone, in the loss of differentiation of possessive constructions, in the loss of two discontinuous negators, and in the loss of prepositions and conjunctions. An important sociolinguistic criterion for a pidgin is that it is not the mother tongue of any of the groups using it. The early expressions Sango tí turugu "soldiers' Sango" (cf. Samarin 1984-85:359), langue commerciale (Calloc'h 1911), and langue véhiculaire clearly show that the language was reduced in its functions, which were mainly trade and military activities,- and that it was not the first language of the different groups using it. A creole "has a jargon or pidgin in its ancestry", but it has undergone expansion and is thus a means of communication that can be and is used in all areas of life, not just one single area such as trade or transport (cf. Holm 1988:7). It need not be the native language of a speech community, as Holm (ibid.) claims, but it must fulfill the functions of a vernacular for a given community (cf. Mufwene 1988). As a result of functional expansion, Sango became a language that, without any doubt, is capable of fulfilling any communicative purpose. Struc tural expansion is clearly found in the formation of prepositions, conjunc tions, complementizers, and the serial verb construction. Creolization, i.e. the vernacularization of Sango, began early in the twentieth century, when Sango developed into an in-group language among the people of different ethnic and linguistic origins working for the French military and administration, as well as among those who lived in the missions and in the "village de liberté".32 In the CAR capital Bangui, Sango is today the first language for most speakers, i.e. the language they know best. For the younger generation it has
Sango
263
become the mother tongue, i.e. the language learned first. This means that it also meets the conditions of a nativized creole. French and the ethnic lan guages are used as second languages, but many parents do not teach their children their ethnic languages at all. In rural areas, by contrast, Sango is for the majority of people a second language. Rural Sango is a reduced secondlanguage variant of a creole, not a pidgin. Rural speakers get sufficient Sango input of native-speaker quality from clergy and from military and administra tive officials. Furthermore, we define pidgins and creoles as mixed languages (cf. Whinnom 1971, Agheyisi 1971:16ff, Holm 1988:34-3, Thomason & Kaufman 1988:3), with one major base or lexifier language and one or more secondary source languages.33 Sango does fulfill this criterion, since the major part of the lexicon of Sango was retained from Ngbandi and a considerable part of its grammar was contributed by the substrate languages. The function words of Kitúba, French, and Arabic origin were already documented in the earliest grammars of Sango, that is, only 26 years after the conditions for its emergence were established. The length of this period can be deduced from the facts that the French and the Belgian conquerors, together with their Kitúba-speaking African helpers, reached the Yakomaspeaking area for the first time in 1885 (Samarin 1982a:34) and that the grammars by Calloc'h and Clerc, where these function words were first documented, were published in 1911. The copula from Kitúba and the negator from Ngombe and Kitúba/ Lingala are two instances of substrate interference which, by themselves, imply significant structural changes that could hamper comprehension. Such drastic changes are not to be expected from borrowing in such a short span of time, however; they are only likely in a situation involving a developmental break in the transmission of a language, i.e. one of pidginization or creolization. Finally, there are many features of Sango which are shared by its lexifier and the substrate languages. This explains the retention of the phonemic system, the SVO word order, the preposition nà, the postnominal position of determiners, and other features.
264
Helma Pasch
Notes *
This paper is a preliminary version of a chapter of my habilitation thesis.
1.
Before Sango was declared an official language in March 1991, French was the sole official language of the country.
2.
This classification of languages is from Greenberg 1966.
3.
Portuguese traders and American Protestant missionaries soon followed the French and Belgian conquerors.
4.
The works by Giraud (1908) and Brachiel (1909) were not available to me. Modern Sango examples in this paper for which no references are given were collected partly with the help of Central African informants in Germany and partly during several sojourns in CAR. The last of these, an eight-month field trip in 1991-92, was sponsored by the German Research Foundation, to which I would like to express my deepest gratitude.
5.
In 1982a Samarin assumed that the expatriate workers, i.e. "the black personnel, who had been recruited elsewhere,... arrived in the Ubangi basin speaking Bangala, already a somewhat stabilized pidgin, or interacting with each other in a Bantu-based jargon" (1982a:31). Later (1989b:705) he allows for the possibility that they spoke Kitúba, because they had been among the Bakongo for such a long time. He himself is not fully convinced, however.
6.
In Kitúba, which developed because of the imported West African labor force, substratai influence from West African languages is likewise almost nonexistent (cf. Mufwene, this volume).
7.
According to Samarin (1984-85:341), Gbanziri was the lingua franca in the Ubangi bend before Sango emerged, but I consider this unlikely (Pasch 1992a:21). Nevertheless, the question arises as to why there are no traces of Gbanziri in Sango. This might be due to the brevity of the period during which the Gbanziri were the most important cooperators with the Europeans. Some authors, e.g. Bouquiaux et al. (1978), do not mention the GbanziriEuropean cooperation at all. A better explanation for the apparent lack of traces of Gbanziri in Sango would be that the Gbanziri used vehicular Yakoma rather than their own language to communicate with the Europeans.
8.
It is not known why the name Sango and not Yakoma was given to the new lingua franca (Samarin 1984-85:359).
9.
That lingua francas emerge as soldiers' languages, or acquire that function, is a phenom enon which has been observed on several occasions in Africa (e.g. with Lingala, Bambara, and Nubi; cf. Heine 1978).
10.
It is a commonplace in CAR that in their own territory the Gbaya refuse to speak Sango and that people in the markets refuse to sell goods to customers who address them in Sango. Today this is definitely not true, and it is doubtful that it ever was, though individual refusals may have taken place. Some of my consultants told me that as children they were obliged to learn Sango so that they could sell their goods in the markets. An old man told me that he learnt Sango in Berberati, a Gbaya town, when he was a young man.
Sango
265
11.
Although Pidgin Sango emerged as a result of imperfect acquisition of Yakoma by the foreign Africans, these might have regarded their speech as Yakoma rather than as a new language distinct from Yakoma.
12.
With respect to the vocabulary, many people still consider Sango incomplete, even though it is the mother tongue of the majority of the inhabitants of Bangui. While most speakers make do with borrowings from French, some of the Christian missions, several linguistic institutions, and some journalists of Radio Centrafrique have developed new lexicon on the basis of Sango word-building rules or by borrowing from Ubangi vernacu lars, preferably from Yakoma. The efforts of the different institutions are, however, not coördinated.
13.
Grammaticalization is frequently regarded as a phenomenon which occurs only in fully developed languages. It depends, however, more on human creativity than on the devel opmental status of a given language. Heine (1991) shows grammaticalization processes in Kenya Pidgin Swahili.
14.
Mutual intelligibility is restricted to words and short sentences comprised of Yakoma words. Oral and written texts and conversations are, however, unintelligible to a monolin gual speaker of the other language.
15.
Boyeldieu (1975:21) gives /mv/ for Yakoma instead of Ngbandi /nv/.
16.
There are, however, exceptions to these rules; see Boyeldieu (1982:33).
17.
For labial-velar stops in the northern Bantu languages, see de Boeck (1948).
18.
Lekens' use of the term "Sudanic" agrees with Westermann's (1911, 1927); the group includes languages classified by Greenberg (1966) in the Nilo-Saharan and Niger-Congo families.
19.
Note that the lexical tone of the verb stem affects the resulting forms (cf. Lekens 1923:6571).
20.
The following abbreviations are used in this and other examples: ADJ = adjective, AL = alienable possession, AUX = auxiliary, COMP = complementizer, COND = conditional, CONJ = conjunction, COP = copula, DEF = definite marker, DEM = demonstrative, EMPH = emphasis marker, HAB = habitual, HUM = human, INDEF = indefinite marker, INDSUBJ = indefinite subject, IRREAL = irrealis, LOG = logophoric pronoun, N = noun, NA = nomen agentis, NEG = negator, NOM = nominalizer, PERF = perfective, pi = plural, POSS = possessive marker, PRED = predicate marker, PREP = preposition, PROS = prospective possession, REAL = realis, RECIP = reciprocal, REFL = reflexive, sg = singular, SM = subject marker, SUB = subordinator, SVC = serial verb construction, V = verb.
21.
In a number of idiomatic expressions yèkè is used like a verb, i.e. without a nominal or infinitival predicate after it. Nzàpa à-yèkè "God is" is probably the example most frequently cited in the literature (e.g. Bouquiaux et al. 1978). The most frequent use of yèkè in this sense occurs in greeting formulas, for example, mɔ yèkè? - έ, mbi yèkè (lit. "2sg be - yes, lsg be") "[How] Are you?" - "Yes, I am [fine]", and when asking whether someone is present: baba a-yèkè? - έ, a-yèkè (lit. "father SM-be - yes, SM-be") "Is [your] father at home/present?" - "Yes, he is." The use of yέkέ in greeting formulas was already documented by Eboué (1918:54).
266
Helma Pasch In nonformalized speech this use of occurs mostly with negative polarity, as in yé/ngînzà "There isn't any thing/any money!", which is a reply given to beggars when they are refused money.
22.
In Mongo and Ngombe negation is marked by a prefix on the verb.
23.
Mufwene (1988:45) calls na a connective.
24.
Other verbs of cognition, in their nonimplicative reading, introduce their complements by means of the conjugated verb which is also the second verb in a serial verb Pepe a-ga (lit. "lsg see/know lsg say Pepe SMconstruction: mbi balhínga mbi come") "I thought that Pepe would come."
25.
In the northwestern part of Belgian territory, which corresponds to the northern half of the Equateur province of Zaire, speakers of Ngbandi and Yakoma were the majority of the population. Therefore their language was much more accessible than in the French territory.
26.
The marking of tones is taken from Bouquiaux et al. (1978).
27.
In Lingala this term denotes only young girls, but not men, which means that its meaning has been generalized in Sango to refer to any young unmarried person.
28.
There are even fewer words from Ubangian languages other than Banda. In recent times, however, words from Ubangian languages have been borrowed deliberately by linguists and radio journalists in order to enrich the vocabulary of Sango and replace (for example) the French names of the months in the calendar.
29.
This word, which is found in all grammars and textbooks of Sango edited by Catholic missionaries, has become obsolete. Many speakers have never heard it. The term used currently is table.
30.
The adverb "slowly", the only example of optional reduplication in Sango, is not listed in Lekens' Ngbandi grammar (1923).
31.
I am not referring to terminology for modern technical facilities but to lexicon which is found in the vernacular languages in Central Africa.
32.
In the domain of the lexicon, expansion still continues today, especially with respect to technical and religious terminology. It proceeds by word formation and through borrow ings, most of which come from French.
33.
The definition according to which pidgins and creóles have the lexicon of one language and the grammar of another, however, is an oversimplified characterization which does not apply (cf. Traugott 1977:73).
References Agheyisi, Rebecca. 1971. West African Pidgin English: Simplification and Simplicity. Stanford, CA: Stanford University dissertation. [Published, Ann Arbor, MI: Univer sity Microfilms, 1984.]
Sango
267
Bouquiaux, Luc; Jean-Marie Kobozo; and Marcel Diki-Kidiri. 1978. Dictionnaire SangoFrançais et Lexique Français-Sango. Paris: SELAF. Boyd, Raymond. 1989. "Adamawa-Ubangi". The Niger-Congo languages, ed. by John endor-Samuel, 178-215. Lanham, MD: University Press of America. Boyeldieu, Pascal. 1975. Etudes Yakoma, langue du groupe Oubanguien (RCA), mor phologie - synthématique. Paris: SELAF. Boyeldieu, Pascal. 1982. "Structures sociales et particularismes linguistiques en pays de langue Ngbandi. Eléments pour une étude." Le domaine Ngbandi, ed. by Pascal Boyeldieu and Marcel Diki-Kidiri, 13-80. Paris: SELAF. Brachiel, G. H. 1909. Vocabulaire sangho (Haut-Oubangui). Paris: Boyard. Brachiel, G. H. 1931. Zande and English dictionary. London: The Sheldon Press. Calloc'h, Jean-René. 1911. Vocabulaire Français-Sango et Sango-Français, langue commerciale de l'Oubangui-Chari. Paris: Librairie Paul Geuthner. Clerc, M. 1911. "Vocabulaire Sango". Bulletin de la Société d'Anthropologie de Paris 2.206-19. Cloarec-Heiss, France. 1986. Dynamique et équilibre d'une syntaxe: le banda-linda de Centrafrique. Paris: SELAF. De Boeck. 1948. "Kp en gb en Nord-Bantoe". Zaïre 2/1.57-63. Diki-Kidiri, Marcel. 1977. Le Sango s'écrit aussi: Esquisse linguistique du Sango. Paris: SELAF. Diki-Kidiri, Marcel. 1978. Grammaire Sango, phonologie et syntaxe. Thése de 3ème cycle, INALCO, Université de Paris III. Diki-Kidiri, Marcel. 1982. "Variations libre entre phonèmes en sango, langue nationale de la R.C.A." Le domaine Ngbandi, ed. by Pascal Boyeldieu and Marcel Diki-Kidiri, 81-103. Paris: SELAF. Eboué, A. F. 1918. Langues Sango, Banda, Manja. Paris: Emile Larose, LibrairieEditeur. Everbroeck, R. Van. 1969. Le Lingala parlé et écrit. (No place) Foley, William A. 1988. "Language birth: The processes of pidginization and creolization". Linguistics: The Cambridge Survey, vol. 4, Language: The Socio-cultural Context, ed. by Frederick J. Newmeyer, 162-83. Cambridge: Cambridge University Press. Gentil, Emil. 1902. La chute de l'Empire de Rabah. Paris: Librairie Hachette et Cie. Giraud, G. 1908a. "Vocabulaire sango, manjia, banda, bakongo et azandé". Revue coloniale (Paris) 1908.236-91, 332-54. Giraud, G. 1908b. Vocabulaire des dialectes sango, bakongo, azandé (Congo français). Paris: Challamel. Godart, Louis, and Cyprien Zoubé. 1986. Nos pères dans la foi: Les anciens de la mission Saint Paul. Bangui: Foyer de Charité de Bangui. Gore, the Rev. Canon E. C , and Mrs. E. C. Gore. 1931. Zande and English dictionary. London: The Sheldon Press. Goutalier, Régine. 1954. "Les débuts difficiles de la capitale de la République Centr africaine, Bangui de 1889-1893". Cahiers d'Etudes Africaines 14.299-316. Greenberg, Joseph H. 1966. The languages of Africa. The Hague & Bloomington: Mouton and Indiana University.
268
Helma Pasch
Heine, Bernd. 1978. Sprache, Gesellschaft und Kommunikation in Afrika. Munich: Weltforum. Heine, Bernd. 1991. "The development of Kenya Pidgin Swahili". Kontakte und Simplifikation, ed. by Norbert Boretzky, Werner Enninger and Thomas Stolz, 29-54. Bochum: N. Brockmeyer. Hellinger, Marks. 1985. Englisch-orientierte Pidgin- und Kreolsprachen. Darmstadt: Wissenschaftliche Buchgesellschaft. Holm, John. 1988. Pidgins and creoles, vol. 1: Theory and structure. (Cambridge Language Surveys.) Cambridge: Cambridge University Press. Hymes, Dell, ed. 1971. Pidginization and creolization of languages. Cambridge: Cam bridge University Press. Jansen, Bert; Hilda Koopman; and Pieter Muysken. 1988. "Serial verbs in the creole languages". Amsterdam Creole Studies II, ed. by Pieter Muysken, 125-59. Amsterdam: Universiteit van Amsterdam, Institut voor Allgemeene Taalwetenschap. Lagae, CR. 1921. La langue des Azande, vol. 1: Grammaire, exercices, légendes. Ghent. Lehmann, Winfred. 1968. Historical Linguistics. New York: Holt, Rinehart and Winston. Lekens, Benjamin. 1923. Spraakkunst der Ngbanditaal. Brugge: . Beyaert/C. Houdmont-Cortvriendt. Lekens, Benjamin. 1952. Dictionnaire Ngbandi. Anvers: Editions de Sikkel. Lekens, Benjamin. 1955, 1958. Ngbandi Idioticon, I Nederlands-Frans en Ngbandi; II Ngbandi en Frans-Nederlands. Tervuren: Musée Royal du Congo Belge. Mufwene, Salikoko. 1988. "Formal evidence of pidginization/creolization in Kituba". Journal of African Languages and Linguistics 10.33-51. Mufwene, Salikoko. 1990. "Time reference in Kikongo-Kituba". Pidgin and creole tensemood-aspect systems, ed. by John Victor Singler, 97-117. Amsterdam & Philadelphia: John Benjamins. Mufwene, Salikoko. 1993. "Kitúba". This volume. Pasch, Helma. 1985. "Die deverbale Derivation im Sango". Afrikanistische Arbeits papiere 4.5-29. Pasch, Helma. 1992a. "The emergence of Sango". California Linguistic Notes 23/2.1924. Pasch, Helma. 1992b. "Rapport préliminaire sur une recherche socio-linguistique sur les variantes du Sango". Afrikanistische Arbeitspapiere 29.175-86. Pasch, Helma. 1993. "Phonological similarities between Sango and its base language: Is Sango a pidgin/creole or a koiné?". Topics in African linguistics (Current Issues in Linguistic Theory, 100) ed. by Salikoko Mufwene & Lioba Moshi, 279-92. Amster dam & Philadelphia: John Benjamins. Pasch, Helma, and Talmy Givón. 1988. "Verb complementation in Sango". Afrikanis tische Arbeitspapiere 16.69-96. Pustet, Regina. 1989. Die Morphologie des 'Adjektivs" im Sprachvergleich. Tübingen: Peter Lang. Rood, N. 1958. Ngombe-Nederlands-Frans Woordenboek!Dictionnaire ngombe-néerlandais-français. Tervuren. Samarin, William J. 1967. A grammar of Sango. The Hague: Mouton. Samarin, William J. 1971. "Salient and substantive pidginization". In Hymes, ed., 11740.
Sango
269
S amarin, William J. 1979. "Simplification, pidginization and language change". Read ings in Creole Studies, ed. by Ian F. Hancock, 54-68. Ghent: E. Story-Scientia. Samarin, William J. 1982a. "Colonization and pidginization on the Ubangi River". Journal of African Languages and Linguistics 4.1-42. Samarin, William J. 1982b. "Goals, roles and language skills in colonizing central equatorial Africa". Anthropological Linguistics 24/4.410-22. Samarin, William J. 1984/85. "Communication by Ubangian water and word". Sprache und Geschichte in Africa 6.309-73. Samarin, William J. 1986a. "The sources of Sango's Be". Journal of Pidgin and Creole Languages 1:205-223. Samarin, William J. 1986b. "Chinook Jargon and pidgin historiography". Canadian Journal of Anthropology 5.23-34. Samarin, William J. 1989a. Black Man's Burden: African Colonial Labor on the Congo and Ubangi Rivers, 1880-1900. Boulder, CO: Westview Press. Samarin, William J. 1989b. "The colonial heritage of the Central African Republic: A linguistic perspective". The International Journal of African Historical Studies 22: 697-710. Sango-English Dictionary. 1980. Sibut, Central African Republic: La Presse Biblique Baptiste. Stolz, Thomas. 1986. Gibt es das kreolische Sprachwandelmodell? Frankfurt: Peter Lang. Swift, L.B., and E.W. Zola. 1963. Kitúba Basic Course. Washington, DC: Foreign Service Institute. Tanghe, Basilius Oct. 1940. "La langue Ngbandi". Aequatoria 4.110-1. Thomason, Sarah Grey. 1986. "Contact-induced change: Possibilities and probabilities". Akten des 2. Essener Kolloquiums über "Kreolsprachen und Sprachkontakte ", ed. by Norbert Boretzky, Werner Enninger, and Thomas Stolz, 261-84. Bochum: Studien verlag N. Brockmeyer. Thomason, Sarah Grey, and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley & Los Angeles: University of California Press. Tisserant, R. P. 1950. Sango: Langue véhiculaire de l'Oubangui-Chari. Issy-les Moulineaux: Les Presses Missionaires. Torozoni, Ngama-Nzombio Tra Ndele. 1989. Description du Ngbandi, Langue Oubanguienne du Nord-Ouest du Zaïre. Brussels: Université Libre de Bruxelles dissertation. Traugott, Elizabeth. 1977. "Pidginization, creoHzation, and language change". Pidgin and creole linguistics, ed. by Albert Valdman, 70-98. Bloomington: Indiana Univer sity Press. Tucker, Archibald Norman, with P. E. Hackett. 1959. Le groupe linguistique zande. (Annales du Musée Royal du Congo Belge, Série 8: Sciences de l'Homme, Lin guistique, 22.) Tervuren. Tucker, Archibald Norman, and M.A. Bryan. 1966. Linguistic analyses. The non-Bantu languages of North-Eastern Africa. London: Oxford University Press. Villelune, Baron E. de. 1909. Au coeur de l'Afrique. Paris: Gabriel Beauchesne. Weinreich, Uriel. 1963. Sprachen in Kontakt: Ergebnisse und Probleme der Zwei sprachigkeitsforschung, tr. from English by Jörg Kohlhase. Munich: C. H. Beck.
270
Helma Pasch
Westermann, Dietrich. 1911. Die Sudansprachen: Eine sprachvergleichende Studie. Hamburg: L. Friedrichsen. Westermann, Dietrich. 1927. Die westlichen Sudansprachen und ihre Beziehungen zum Bantu. Berlin: de Gruyter. Whinnom, Keith. 1971. "Linguistic hybridization and the 'special case' of pidgins and creoles". In Hymes, ed., 91-115.
Prior Pidginization and Creolization in Swahili? Derek Nurse Memorial
1.
University
of
Newfoundland
Recent background to Swahili
Until the late eighteenth or early nineteenth century, Swahili was only used by a relatively small population on the coast and offshore islands of East Africa. By the beginning of the nineteenth century, as the result of local political events, the economic power of Zanzibar Town had grown to the point where it expanded off the island, initially to the adjacent islands and mainland and, by mid-century, west across East Africa to Lakes Nyasa, Tanganyika, and Victoria, and beyond (Heine 1970, also for much of §1; Brumfit 1980). Although the trade caravans are often described as being led by Arabs, they must have included Swahili speakers, because communities with a nucleus of native speakers and/or people using Swahili as a trade lingua franca sprouted along the trade routes, especially in Tanganyika and into Uganda, Rwanda, Burundi, and what is now Zaire. In addition, by mid-century, European explorers, missionaries, and traders were active, soon followed by soldiers and administrators. By the end of the nineteenth century Swahili was estab lished in Tanganyika. First the German and then the British colonial govern ment, after some initial hesitations, slowly accepted this fact: the churches, government, civil service, and army came to use Swahili for their own operations. In Kenya, use of Swahili spread more slowly: by the mid-twenti eth century it was used by the army, on farms, in certain branches of govern ment, by many people in eastern Kenya, and by some in central Kenya. Finally, when independence came in the 1960s, new African governments
272
Derek Nurse
also espoused the de facto language situation, with enthusiasm in Tanzania and with reservations in Kenya. Over 50 million people speak Swahili today. It is the official language of Tanzania and a national language in Kenya (total populations, over 50 mil lion). Forms of Swahili are spoken by several million people in Zaire, by perhaps 2 million in Rwanda and Burundi, and by very small numbers in southern Somalia (20,000?), Zambia, Malawi, the northern Mozambique coast, and on northern Madagascar. Despite an early toehold at the court of Buganda, later use by colonial army and police, and a short upsurge of enthusiasm following the overthrow of Amin by the Tanzanian army in 1979, Swahili has very limited use in Uganda. It is spoken on the island of Socotra off the northern tip of Somalia, and by an expatriate community of unknown size in various countries in the Persian Gulf. Swahili was used at the Sultans' courts in the Comoro Islands for an unidentified period up to the end of the nineteenth century, alongside Comorian, which is a closely related language but not a form of Swahili. Swahili had once a wider distribution on the Mozambique coast, and archaeological evidence suggests that Madagascar may have had a larger Swahili community in the first millennium A.D. It is useful to distinguish different levels of ability in Swahili. It is spoken as a first language (a) by traditional Swahili dialect communities along the coast of Somalia, Kenya, and Tanzania, whose ancestors have spoken Swahili for a millennium or more (see §2) and who number perhaps 600,000, of which about 500,000 are on Zanzibar and nearby Pemba Island; (b) by an unknown but smaller number of adults in urban centers in East Africa, whose ancestors spoke some other language a few generations ago but raised children who came to adopt Swahili as an interethnic language; and (c) by an increasing number of young people, especially in Tanzania, who might have spoken an ethnic language in their early years and perhaps still use it on occasion, but who, as a result of schooling and national policy, use Swahili in most daily situations. A larger number of people have Swahili as a second or even a third language. In Tanzania, second-language Swahili speakers are relatively few in number and are mainly older people in the west of the country. In Kenya, Zaire, and probably Rwanda and Burundi, many more people are in this category. Here the status of Swahili is much as described by Heine (1970) for Kenya: knowledge and use of Swahili is higher among males, the young, in towns, and in smaller ethnic groups, and lower among females, the old, in rural areas, and in larger ethnic groups (which have more sense that Swahili is alien).
Prior Pidginization and Creolization in Swahili?
273
Closely correlated with language ability is the type of Swahili used. Today's Standard Swahili (StS) was based on Unguja, the dialect of Zanzibar Town. It is the form spoken, or aimed at, by Tanzanians. Local Tanzanian varieties coexisting with, or replacing, StS are legion (see e.g. Legère 1986). StS seems to have entered Kenya from Tanzania. Away from the coast, although Swahili is the lingua franca and there is a consciousness of StS, there is much more variation in the Swahili used, and wide use of Pidgin Swahili (PiS). In Zaire, after the initial settlement by Swahili speakers from the east coast, the language developed in relative isolation from its place of origin (Fabian 1986, Mufwene 1989, Polomé 1971, 1985). The forms of Swahili spoken there today are certainly restructured and influenced by local lan guages, but there is considerable disagreement about the relative roles of different processes in their development — the possibility of transmission by descendants of the original settlers, the degree of pidginization and creoliza tion by large numbers of migrant workers, and even the possibility of acquisi tion through straight second language learning (Holm 1989:565). Within East Africa itself there are also forms of Swahili generally agreed to be pidginized. Most of the now reduced East Indian population in East Africa speaks kiHindi ("Indian language"); on up-country farms and planta tions, kiSetla ("settlers' language") and kiShamba ("plantation language") were used; and a kiVita ("war language") was said to be dying in the late 1950s, but no good study has ever been done of language use in the army in post-independence East Africa. Wald (1981) describes two variants of PiS used on the northern Kenya coast and in Mombasa, where he estimates that 60% of the urban population spoke PiS in the mid-1970s.1 All these forms of Swahili are more or less well known. What has never been addressed comprehensively is whether mainstream Swahili itself ever underwent pidginization and creolization. It has often been claimed that Swahili initially arose out of intermarriage between (superstrate male) Arabs and (substrate female) Bantu speakers (e.g. Broomfield 1931; Krumm 1940; Reusch 1953; Hancock 1971:519; Holm 1989:564). Heine (1973:169, n. 100), following suggestions by others, allows the possibility that certain features of coastal Swahili support the notion of possible prior pidginization. It is these uncertainties we now address. (See also Bender 1987.)
274 2.
Derek Nurse Historical social background to Swahili
Bantu-speaking Early Iron Age people have been on the East African coast since the first century A.D. (Chami 1988, Schmidt 1988). Swahili communi ties have been on the coast since at least 800 A.D. (Nurse & Spear 1985). It is possible that the language was spoken away from the coast earlier, but we have no evidence for that. The current state of archaeology suggests these earliest coastal communities were in the Lamu Archipelago (northern Kenya), on Zanzibar Island, on Kilwa (southern Tanzania), and possibly near Mukdisho (Somalia, formerly called Mogadishu). Most of the gaps between these isolated settlements were filled in by 1200 A.D., so that the ancestors of today's 15-20 coastal dialect communities were largely in place along a 1000mile more or less continuous strip from Barawa in southern Somalia to at least northern Mozambique. Political power did not reside in one place. At any one point in time there were often competing centers of power, and over the centuries power shifted fron one center to another. The dialects themselves can be divided into northern dialects, spoken on what is today the coast of Somalia and Kenya, and southern dialects, spoken on the Tanzanian coast. Unguja, the basis for today's standard language, is a southern dialect with a northern dialect overlay. In what follows I will try to focus both on Unguja and on the dialect continuum as a whole. There is a dearth of hard facts about the historical social situation. The first reliable account of the coast, the Periplus of the Erythraean Sea , is by an outsider and dates from the second century A.D. Then, after a gap of several centuries, we find a thin string of Arab travelers' accounts that continue until the arrival of the Portuguese at the start of the sixteenth century (see FreemanGrenville 1962). Both in their geographical coverage and in their description of what they saw, these accounts, mainly by merchants, leave much to be desired. They do, however, provide details about the nature and the consider able volume of goods that moved in both directions between the coast and southern Arabia. Some of the local goods originated inland and were brought to the coast for shipment. Findings from archaeology complement the travel ers' descriptions, as do local oral traditions for more recent centuries. Written records in Swahili are of little help, as they do not start till around the late seventeenth century and are mainly concerned with religious topics. Com parative linguistic data from Swahili dialects and closely-related languages is a major source of information — mainly lexis, but also details of phonology and morphology in some cases.
Prior Pidginization and Creolization in Swahili?
275
Almost from the beginning Swahili-speaking settlements, all on offshore islands or a very narrow mainland strip, had trade as a central component. They traded to the west with a range of inland communities, adjacent and nonadjacent; from north to south with each other; and to the north, east, and south with southern Arabia, India, the Comoros, and Madagascar. There are even reports of Chinese boats on the coast. From a short time after their appearance in 800 A.D. until well into the present century they acted as entrepots for goods moving in both directions. They also continually assimi lated people from both directions. Merchants from southern Arabia came at first only to trade, but later they settled there, as did Arabs (and possibly Persians earlier) for religious reasons — initially to escape Islam, later to spread it. At all periods Arabs on the coast seem to have been low in numbers but high in prestige and power. While it is true that in the eighteenth and nineteenth centuries many upcountry slaves were taken, some being kept at the coast and others sent north, there is little evidence for earlier centuries that slaves were a major component of the trade with southern Arabia. Large plantations based on slave labor were a nineteenth-century development, certainly on Zanzibar, Pemba, Mombasa, and Malindi (Cooper 1977) and probably also those on the northern Kenya coast opposite Lamu. Much of today's population of Zanzi bar and Pemba is descended from slaves. The evidence for most of the millennium up to 1800 A.D., however, suggests that people from adjacent local communities were voluntarily attracted by the magnet of the towns, and that over the centuries many assimilated into the Swahili population. Al though evidence from the last two or three centuries suggests that Swahili society was stratified ("Arab", "Swahili", "local African") it is unclear how far back this model can be projected. While there is a lack of detail about the sociohistorical background of Swahili, enough has been said to indicate that the sociohistorical conditions where a pidgin might arise seem to have existed for at least a millennium: use of Swahili as a coastal trading lingua franca, assimilation of large numbers of (mainly local) outsiders into Swahili society, and an adstrate- or superstratesubstrate relationship between Swahili and local languages (typologically very similar along much of the coast).
276 3.
Derek Nurse The linguistic development of Swahili
Linguistic evidence for a distinct Proto-Swahili is slight, but if such a lan guage existed it probably barely predated the diaspora that led to the earliest settlements described at the beginning of §2 above; a century or two before 800 A.D. 2 seems appropriate. Swahili is genetically grouped with five other local languages to form what has been recently called "Sabaki": a date of perhaps 500 A.D. would fit Proto-Sabaki. These six languages in turn form part of a larger grouping called the "North East Coast Bantu" (NECB) languages, a group which has over 20 members, including Sabaki. ProtoNECB can be assigned to around 2000 years ago. It is possible to chronolo gize most of the processes of loss, simplification, replacement, and innovation that characterize Swahili by looking at their distribution within NECB, Sabaki, and Swahili. Below I group together different kinds of processes in ways that are familiar from pidginization elsewhere, attempt to see if they cluster at certain time periods, and trace possible links to external events. Detailed evidence for this analysis is in Nurse & Hinnebusch (1993). 3.1. The lexicon The core lexicon of all Swahili dialects, as evidenced by use of variants of Swadesh's 100- and 200-word list, can be broken down as follows: an average of 82% inherited items, 15% borrowed items, and 3% items of unclear origin. The borrowed items break down into 10-11% from local languages and the other 4-5% from overseas sources (mostly Arabic, but also East Indian or unknown). The majority of words borrowed from local lan guages are from neighboring Bantu languages, with a minority from other Swahili dialects or Cushitic (for this and following, see Nurse & Hinnebusch 1993, Chapters 2 and 3). By looking at the ways in which loanwords have or have not undergone regular, ordered, phonological change in Swahili, we can date their approximate entry into the language. Loans from Cushitic are very early; loans from other Bantu languages are difficult to date because the source languages do not differ greatly from Swahili, but they seem to cover a wide time span; words from other Swahili dialects are also not easy to date, but they were apparently mainly absorbed over the last few centuries; and words from Arabic are recent — most were absorbed in the last few centuries, from Omani and possibly also Mehri in the Hadramaut (in southern Arabia).
Prior Pidginization and Creolization in Swahili?
277
A similar breakdown of larger segments of the Swahili lexicon has not been properly carried out. The only analyses of dictionaries (themselves incomplete) have looked at percentages of foreign, but not local, loans. Analyses of written sources suffer from the same failing, and have varied widely in their assessment of the Arabic component, depending on the type, register, or content of the source. Assessment of the Arabic component (again, mainly Omani and mainly recent) from both types of source ranges from ca. 20% to 50% of the total vocabulary (see Krumm 1940, Bertoncini 1985). Arabic loans cluster in certain semantic areas — religion, law, admin istration, trading activities (here there is also an admixture of terms from East Indian), nautical, measurement, communication, and many imported food plants and fruit trees. The dramatic rise in the Arabic component from core to culture vocabulary, and the semantic areas in which foreign loans cluster, resemble the patterning of French loans in English (Williams 1977). A genetic interpretation of such percentages suggests that Swahili as the recipient language, with most of its vocabulary inherited, has borrowed widely and heavily from a range of donor languages. That is not the only possible interpretation. In the Atlantic creóles, where a straight genetic inter pretation is inappropriate, the bulk of lexis comes from the European lexifier languages, which are not their genetic forebears, and in Africa nonpidgins can have a higher percentage of borrowed items than pidgins (Heine 1979, Holm 1989). The clear break in continuity in language development brought about by pidginization and creolization seems to involve initially a small lexicon, with subsequent relexification. This tends to leave tell-tale traces in the new lexicon: irregular phonological correspondences between items in the new language and those in the lexifier language(s) (Polomé 1980), reanalysis of morpheme boundaries (assuming the languages involved are typologically different and/or mutually unintelligible), a range of meaning in various lexi cal items which is greater than in the source languages, and a tendency towards multifunctionality of lexis (e.g. in prepositions). None of this is found in Swahili. Most Swahili vocabulary can be regularly derived phonologically from its various proto-stages. What little apparent phonological irregularity there is exists only in loanwords traceable to nearby and related languages. Similarly with 'morphological reanalysis': while some loanwords, such as kitabu "book", kilabu "club", and ndama "calf (from Ar. kitaab, Eng. club, and Southern Cushitic dama, respectively), have undergone morphological reanalysis (and phonological adaptation), the whole inherited set of vocabu-
278
Derek Nurse
lary shows no sign of reanalysis by an assimilated population. And while some inherited items have naturally undergone semantic extension and other change, there is no sign of a general tendency to extension of meaning or multifunctionality. 3.2. Phonology, pidginization, and creolization Among phonological features said to occur often in pidginization and/or creolization are a reduction in the phonological inventory; specific kinds of phonological change; a high level of phonological irregularity in "inherited" items; and the clear phonological influence of substrate languages in the resulting creole. Since any putative pidginization in the Swahili dialect spectrum as a whole is many centuries in the past, we would not expect to find a drastic reduction in any of their phonological inventories today. Comparison of contemporary phonological inventories with that of Proto-Bantu shows a slight but regular and explicable reduction in the vowel inventory, and a considerable increase in the consonant inventory. Comparison with ProtoNECB, Proto-Sabaki, Proto-Swahili, contemporary NECB or Sabaki lan guages, and indeed with eastern Bantu languages in general indicates that the Swahili vowel and consonant inventory is comparable in size and structure. Comparison of the phonological changes undergone by Swahili in gen eral over the past millennium and more with the typology of changes consid ered by Heine (1979) to characterize Bantu-based pidgins in general is also not revealing. Most of the changes in Heine's list have not occurred in Swahili, even though the phonological starting points were available. Thus, for instance, voiced fricatives have not devoiced, plosives have not replaced implosives, phonemic aspiration has not been lost, and it is not obvious that a drift to bisyllabic word structure, or the replacement of context-sensitive by context-free rules, has occurred to a greater degree than in other languages or neighboring Bantu languages. Some of Heine's other characteristics are indeed found — reduction to five vowels and the loss of vowel length distinctions.3 For Swahili, reduction to five vowels and loss of distinctive length can be roughly dated to the late first, or early second, millennium A.D. However, since these two features are very widespread in eastern Bantu (Nurse 1987) and result from common processes, they do not appear to be connected to pidginization.
Prior Pidginization and Creolization in Swahili ?
279
Heine's two remaining features are more problematic. One involves the replacement of palatal by alveolar fricatives. It can be shown that replace ment of at least the voiceless member of the pair has occurred in many Swahili dialects: where Unguja and a minority of other dialects have (to give just one of many examples) shingo "neck", the majority of Swahili dialects have singo, as have most other NECB languages. Nurse & Hinnebusch (1993) posit an original palatal for this fricative. Why a handful of Swahili dialects keep the palatal, while all other dialects plus over twenty NECB languages replaced it by the alveolar, is not clear, but if it does result from prior pidginization, it was clearly not restricted to, nor did it necessarily start in, Swahili. The remaining feature in Heine's list is the reduction/loss of tonal contrasts. Here the Swahili evidence is inconclusive. Whereas most other Bantu languages have primarily tone, most Swahili dialects have penultimate stress4 and no tones. Since the (physically isolated) northern dialect Mwiini still has skeletal pitch accent, and can be shown to have been isolated since ca, A.D. 1100, the final disappearance of tone in the rest of Swahili is likely to have resulted from areal diffusion at some point after that date. How to explain the loss of tone in Swahili is a perennial debate. Other languages in East Africa (e.g. Tumbuka and Nyakyusa), spoken far from the coast and not known to have been lingua francas, have also replaced tone by stress. The appearance of the penultimate stress system is unlikely to be directly bor rowed, as neither Arabic nor most adjacent Bantu languages have it. That leaves as explanations (a) lengthy use of Swahili as a lingua franca, as Heine's data would suggest (his suggestion is typological, and the exact mechanism for this shift is unclear), and/or (b) an internal phonological explanation whereby High Tone Displacement and Doubling, attested in neighboring languages, led to prominence of the penultimate syllable and ultimately to penultimate stress. It is not thus clear whether there is or is not any connection in this case between prior pidginization and replacement of tone by penultimate stress. There is no evidence of phonological irregularity in inherited lexical items. Swahili consonants and vowels can be derived by an ordered set of processes from Proto-Bantu through various intermediate stages to today's dialects and standard language. The processes themselves, with the two possible exceptions just mentioned (voiceless palatal to alveolar fricative and replacement of tone by penultimate stress), do not resemble Heine's pidgin
280
Derek Nurse
processes. Exceptions to the rules can be shown to result from interdialect borrowing, which, until this century, was mainly in the direction north to south. There is evidence that Arabic has had some phonological influence on Swahili (Tucker 1946/47). The massive lexical borrowing from Omani Ara bic has brought in its wake a small set of borrowed consonant phonemes: /9, ð, ɤ, r/. The phoneme /r/ has also entered from local languages, either in loans or by assimilation of speakers who treated /1/ as [r], later 1r1. Most coastal speakers use these sounds regularly, others occasionally. In Mwani, far to the south and little affected by the recent Omani wave of influence, most Arabic consonants have been been replaced in loanwords by less marked equiva lents. In Mwiini, in the north, influence from Arabic and also from southern Somali dialects has been more pervasive, and additional consonants are found. Arabic has also brought to Swahili three new vowel sequences ([ei, ai, au]) and certain new consonant clusters. In earlier loans these clusters have been broken up by epenthetic vowels, in later loans they are either maintained or are in process of being broken up. In Mwiini these consonant clusters are strictly maintained, and even inherited words have clusters alien to, and only found in loans in, other dialects (e.g. mafia "oil" vs. other Swahili mafuta). But Swahili had if anything a substrate relationship to Arabic, and all these features are simply borrowed. Because Swahili has been adjacent to languages of similar phonological structure and type along most of its 1000-mile-long area, phonological adstrate influence has been minimal. It is most obvious on local (northern and southern) parts of the coast where Swahili communities have been adjacent to commu nities with different systems. Thus Mwani has undergone considerable phono logical change parallel to that in neighboring Makonde, which indicates a community long bilingual. Certain phonological changes in Mwiini suggest influence from southern Somali. The dialects of the northern Kenya coast have dentalized palatals (*c, nj, nz > t, nd, nd) which, together with a set of loanwords, I attribute (Nurse 1985) to the assimilation of Cushitic speakers, most likely at the end of the first millennium A.D. Bajuni, in far northern Kenya, shows further dentalization (*z > ð) and a new set of loanwords, most likely from a southern Somali source, in the middle of this millennium. In summary, all Swahili shows some signs of phonological borrowing from Arabic relatively recently, and individual parts of the Swahili coast show signs of having undergone earlier influence, via loans, assimilation of speakers, or bilingual-
Prior Pidginization and Creolization in Swahili ?
281
ism, from local sub- or ad-strate languages. No such influence can be detected in early Swahili as a whole. 3.3. Morphosyntax: noun phrase and concord Bantu languages are distinguished by a system whereby all nouns are as signed to one of a set of genders. The general characteristics of the genders are a nominal prefix (usually of the shape CV-), a regular pairing of singular and plural (the pairings are relatively stable across Bantu), and a concord system whereby other members of the NP, and also the verbal subject and object, agree with the gender of the noun. Pronouns reflect the same genders. The relationship between natural and grammatical gender is usually opaque. Genders are primary or derived; that is, a noun has a primary gender but can be recategorized into other derived genders. Derived genders usually have few regular members, tending to indicate categories such as augmentative, diminutive, and locative (others are possible). Many Bantu languages also have a 'preprefix', usually of the shape V-, which, when present, tends to resemble in function the definite article of western European languages. The number of genders in East African Bantu languages ranges between 12/13 and 20, the average being 17/18. Swahili dialects behave almost identically with respect to the number of genders, all being at the low end of this range. Pidginization in Bantu languages is marked by a drastic reduction in the inflectional and semantic inventory, which for nouns means some or all of the following: number and gender distinctions are in general neutralized or lost, redundancy in marking or concord is suppressed, the number of genders is reduced, synthetic forms are replaced by analytic forms where possible, and complex allomorphy is simplified. Some of these have indeed affected Swahili, and in what follows I deal with the changes chronologically. It is likely that pre-Proto-NECB had 18 genders,5 a nominal preprefix, and a primary vs. derived system in which certain genders were largely or exclusively derived (#12-13 "diminutive singular and plural", #16-18 "loca tives") and others could be primary or derived (#5-6 "augmentative", #11 "long thin objects, singular", #14 "abstracts/mass", #15 "verbal noun", and possibly #7-8 "diminutive singular and plural"). The first of these to disap pear was #13 "diminutive plural", which left its singular #12 paired with #8. This loss occurred very early, at or before the proto-NECB stage, since no NECB language shows any sign of #13 today. It is apparently an areal feature,
282
Derek Nurse
as it has also disappeared in a large number of other languages down the eastern side of East Africa. Next to go were the three locative genders ##1618. The pattern of replacement can be seen by comparing prefixal forms (older) in Luguru with suffixal forms (newer) in Swahili: Luguru is a lan guage closely related to Swahili. Thus Luguru: #16 pa-cisima "at/by the well", #17 ku-cisima "to/from the well", #18 mu-cisima "in the well", vs. Swahili kisima-ni "at, by, to, from, in the well". Although the nominal genders themselves were replaced, their concord forms were retained, e.g. Sw. kisimani rn-na "in the well (#18) there is"; and kisimani ku- "at/by the well (#17) there is". This replacement occurred at the Proto-Sabaki stage, as shown by the fact that no Sabaki (see §3) languages have this today, while most other NEC languages do.6 It also appears to be an areal feature, having occurred in a large number of languages down the eastern side of East Africa. This form of replacement is particularly interesting not only because it involves loss of inherited gender, but because it resulted in a single locative morpheme taking over the functions of a range of relational markers, which is often cited as a pidginization feature. What we have seen so far were in fact pre-Swahili replacements. The next inherited feature to be replaced was the preprefix, whose functions were subsumed partly by word order but mainly by the use of demonstratives. Although this must postdate Proto-Sabaki, since it is actively retained in two Sabaki languages and has left frozen traces in forms in some Swahili dialects, it is impossible to date the change more precisely. The preprefix has disap peared in all Swahili dialects, so it might be assignable to Proto-Swahili, the first real Swahili feature. As with the locatives, this also involves replacement of a synthetic form by an analytic one. The next gender to go was #12 "diminutive singular". Early loss of #13, its original plural (above), reduced the two original sg.-pl. pairings #12-13 and #7-8 to #12-8 and #7-8. Not unexpectedly, this redundancy was removed by the exclusive use of #7-8 for diminutives. Gender #12 is retained in a few isolated northern Swahili dialects (Bajuni ngombe "cow", ki-gombe "small cow #7", ka-gombe "very small cow #12"), where it is not likely to be the result of borrowing from other languages. It can also be seen in a very few frozen lexical items in the rest of Swahili. This is clearly post-Proto-Swahili; an informed guess based on the geographical distribution of its loss (most of Swahili and two other, non-adjacent, Sabaki languages) would place this in the middle of the present millennium.
Prior Pidginization and Creolization in Swahili?
283
Demonstrably recent, because it is based on phonological loss visibly under way around 1700 A.D., is the falling together of genders #11 "long thin objects, singular" and #14 "abstracts/mass". At a proto-stage, #14 was marked by wu-, but w was deleted before rounded vowels in most Swahili dialects many centuries ago, leaving #14 u-. Not long before 1700 A.D. / was also lost before rounded vowels, also reducing #11 lu- to -. At this point the two gender markers were homophonous, and Swahili is still grappling with the difficulty of reconciling two genders with very different reference. What we have seen thus far is the reduction, over a period of some two millennia, of an original system of 18 genders to a system with only 12: ##1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11/14, and 15 (this treats the locatives as nonexistent because only concord forms exist). As far as I know, only two other Bantu languages in East Africa (Comorian and some dialects of Chaga) have such a reduced system.7 This reduction has a rather predictable corollary. Many Bantu languages have a small class of adjectives, partly (though not only) because these languages can express a range of semantic modification of the basic reference of a noun in a primary gender by recategorizing it into derived genders. Thus, to take an imaginary but Swahili-like example, from primary m-tu "person, #1", we could get, by recategorizing the noun, ji-tu "big person = giant, #5", ki-ji-tu "small person", #7 (+ #5)", lu-tu "tall, thin person, #11", and ka-m-tu "very small person = dwarf, (#1) + #12". Although the obvious reference of these derived nouns is to physical size, this is associated with other properties. So, for instance, small objects in #12 could also become despised, ugly, or abnormal; or unusually tall, thin people in #11 could become risible and foolish. It is therefore not surprising that in a system such as Swahili's, where the number of genders has been reduced, recategorization plays a lesser role than in other Bantu languages and many of these semantic distinctions are expressed by adjectives, of which Swahili has a larger than average set. Many adjectives are borrowed from Arabic, most of these apparently recently.8 This again involves replacement of a synthetic (inflectional) feature by an analyti cal (lexical) one. 3.4. Morphosyntax: verbs A typical one-word Bantu verb has the following shape: (1)
#SM + TM + + V-stem + derivational morpheme + Suffix#
284
Derek Nurse
This statement is simplified, because it ignores certain structural possibilities (occurring mainly word-initially, word-finally, and after [TM]) which are widespread in Bantu. It also ignores the fact that certain of these slots may not be filled. Examples (St. Swahili) are given in (2): (2)
a.
tu + li + m + pig + a we PAST it beat NEUTRAL "we beat it"
b. a + me + pig + w + a it PERFECT beat PASSIVE NEUTRAL "it has been beaten" SM, the subject marker, agrees with the subject and in most Bantu languages is obligatory except in some imperatives. O, the object marker, is not always obligatory, but when present it agrees with the direct or indirect object. TM represents a slot, rather than a category. In many conservative Bantu lan guages, tense (and to some extent modality9) marking occurs at TM, aspect and modality at Suffix (see the Kikuyu example in (3)). By contrast, in languages such as Swahili, TM contains tense and aspect, the Suffix having withered to a few modal markers. Most Bantu languages also have verbal structures involving two words. One common type consists of an auxiliary verb, with structure as above, followed by the (uninflected) infinitive form of the second verb. The infini tive form is usually ku-, which, however, is also the form of #17 "locative". Another type consists of an auxiliary, usually from a restricted set ("be", "have", etc.), followed by a second verb. In this type of construction, both verbs are generally inflected as above, but tense is indicated at the TM slot of the first verb, aspect at the TM slot of the second verb. Examples are given in (3). (3)
a.
one-word verb: Kikuyu: a-ra-mü-gwat-aga SM-MIDDLE.PAST-OM-seize-IMPERFECTIVE "he was seizing him"
b. AUX + infinitive: Kami: t-o-genda##ku-gula SM-NONPAST-gO##tO-buy "we will buy"
Prior Pidginization and Creolization in Swahili?
285
. AUX + 2nd verb: Pokomo: h-aku-wa##hu-ki-ceza SM-FUTURE-be##we-ING-play "we will play" Finally, many Bantu languages have multiple discrete tense reference (e.g. two futures, three pasts), and it has been assumed that such a system repre sented the earlier eastern Bantu norm. As Holm observes (1989:148), "The verb phrase has been of central importance in creole studies... in distinguishing creole varieties... from noncreole varieties... of the same lexical base", because creoles and noncreoles differ dramatically in their TMA systems. During pidginization, synthetic forms are replaced widely by analytic ones (and since the Bantu verb is characterized by much inflection, it is a particularly rich field for investiga tion). Thus creole verbs in general have little or no inflection, TMA being indicated by verbal markers deriving from auxiliaries. In Bantu-based pidgins and creoles, tense and modality categories are reduced or lost, as are various concords. Complex morphophonemic alternations are replaced by less com plex structures.10 It has often been remarked that Swahili has a much reduced tense/aspect system, compared to languages such as Kikuyu.11 Kikuyu has three discrete past tenses, two futures, and a 'present' — all marked inflectionally at TM — and a range of aspects, all marked at Suffix. Swahili has one past, one future, and a 'present/progressive', all marked at TM, which is, however, also used to mark aspect. Use of suffixal inflection in Swahili is minimal, aspect being marked either at TM or via two-word structures. While this massive restruc turing seems to be consonant with possible earlier pidginization, closer inves tigation makes caution advisable. All NECB languages, with a few exceptions (which can be shown to result from later innovation), also have greatly reduced systems. This reduction is not the result of pidginization in Swahili plus some kind of later areal diffusion, but rather derives from Proto-NECB, which can be credited with just three tenses: two pasts (Near and Far) and a nonpast, which might be better interpreted as a "progressive" (Nurse & Hinnebusch 1993, Chapter 4). This "progressive" later split in some NECB languages, among them Swahili, and was replaced by a "future" and a "progressive/present", but it was retained in others. Proto-NECB can be conservatively dated to the first century A.D., and no one has yet suggested that it was spoken at the coast. So while the reason for this dramatic reduction
286
Derek Nurse
from an earlier system with multiple tense reference is not clear, it is far too early to be attributed to any use of Swahili as a coastal lingua franca, or to early Arabs. Swahili also differs from languages such as Kikuyu in other respects. Many of its tense and aspect markers are demonstrably derived from auxilia ries, and it makes more use of two-word verbs, with both words inflected and with an auxiliary as the first verb. So, for example, li- "past" derives from older a+li- "past+be", which became plain li- by phonological attrition; ta"future" most probably derives from an older verb "to want"; na- "progres sive/present" from older "be" or "have" ("have" = "be with"); me-"perfect" from older "to finish"; sha- "alternative perfect" from another verb "to be finished" (this sha- is commonly used in up-country PiS); and ja- "not yet -ed" from older "to come". In languages closely related to Swahili, "go" and other verbs for "want" have also been grammaticalized as tense markers. But these innovations did not all occur in a short time span, or even over a period of a few centuries. Comparative evidence suggests that na- appeared at the Proto-Sabaki stage (ca. 500 A.D.), li- and ta- at the Proto-Swahili stage (ca. 700 A.D.) or shortly thereafter, and me- after 1700 A.D. The evidence for sha- and ja- is harder to interpret, but their geographical distribution within Sabaki suggests they emerged between 800 and 1700 A.D. It remains unclear why the functions of Suffix were transferred partly to the new two-word verbs and partly to TM. Proto-NECB had inherited a suffix -aga "imperfective", which is still alive and well in most East African Bantu languages. This was lost at the Proto-Sabaki stage or shortly thereafter, as no Sabaki language has it today; only frozen traces remain in two dialects. Its functions have been largely taken over by two-word verbs. Proto-Sabaki expressed "perfect" by a suffix -ile, retained by some dialects today and lost in others. This suffix has had a long lingering death, as already in 1700 A.D. it was slowly being replaced by me-, but it is still in limited use in Swahili dialects today. This replacement process illustrates well the problems involved in inter preting all this material. Older -ile caused a range of mutations of the final stem consonant to which it was suffixed, resulting in a complex set of morphophonemic alternations. Many Bantu languages have lived with the same complications for centuries, if not millennia. Swahili was already dis mantling these alternations in 1700 A.D., by simply affixing -ile to an unmutated stem. The appearance of the new me- prefix provided an alternative
Prior Pidginization and Creolization in Swahili?
287
way of solving what was apparently felt to be a problem. Why did Swahili get rid of a feature that, although complex, is retained in dozens of other lan guages? It is tempting to argue that its demise was hastened by large numbers of non-Ll speakers, using Swahili as a lingua franca, who found simple meeasier to handle. There is some minor negative evidence that this might have been the case: -ile is mainly used today by elderly men (and women?) who are relatively high up in the social scale and conscious of the connection between language and social standing. Most older Swahili writing is poetry, composed in a deliberately stylized register; we have no record of how people spoke until the nineteenth century. So how can we prove that non-Ll speakers actually pushed -ile and its complications over the cliff? A final feature that has been claimed as characterizing PiS is the loss of concord marking, especially of subjects. In the northern isolate Mwiini, certain subject markers delete regularly. In older Swahili verse the same subjects also deleted. Older Swahili poetry is conservative and conventional ized, overtly based on usage held to be even older, so if this deletion had penetrated the stylized poetry, it is likely to have been widely used in speech. This might indicate a degree of pidginization. Use of these markers has been restored in all Swahili (except Mwiini) today. The preceding discussion is complicated and can be summarized as follows. For reasons now unrecoverable, the inherited tense system under went a drastic reduction at an early, Proto-NECB, stage — that is, in the first centuries A.D. At the Proto-Sabaki stage, the older suffixal -aga "imperfective" disappeared. In a sense the subsequent long development of the Swahili system was an ongoing replacement of these early losses. First, in ProtoSabaki, the na- "progressive" appeared (probably related to the demise of -aga, but it is unclear which happened first). Next, at Proto-Swahili, very shortly after the Proto-Sabaki stage, li- "past" and ta- "future" (both initially from auxiliaries) appeared. At some intermediate point, sha- "perfect" and ja"not yet" developed. Finally, around 1700 A.D. or slightly earlier, limited SM deletion was possible and me- started to replace the suffix -ile "perfect". In some of these cases, new grammaticalized auxiliaries appeared long after older inflectional structures and contrasts were lost. In other cases, older inflectional forms only started to fade after the appearance of the newer grammaticalized alternatives. The details of some of these replacements will be familiar to creolists: anterior/past < "be", irrealis/future < "go, come, want", perfect/completive < "finish", and "progressive" used as "present".
288
Derek Nurse
But they will be equally familiar to general linguists as universals or quasiuniversals. Where new auxiliary-based forms demonstrably appeared long after older inflectional structures had gone, we can hardly appeal to pidginization; where newer and older forms demonstrably co-existed for a time, pidginization is a possible explanation. 3.5. Syntax No observer has ever claimed to find substantial syntactic traces in Swahili that might point to pidginization or creolization. Wald (1985) identifies certain discourse features that involve replacement in minor contexts of (Bantu) verbal strategies by (Arabic) nominal strategies. He does not suggest that these result from anything other than borrowing.
4.
Chronology and likely sociohistorical background
I now attempt to chronologize all the preceding and to integrate the changes into a known or inferred social setting. Since little is known of most of the period, except for the last few centuries, the 'integration' in fact consists of looking at clusterings of processes linked elsewhere to pidginization, and hypothesizing about possible social situations. The chronology starts at the Proto-NECB level, several centuries before Swahili had emerged as a discrete speech form, because one of the putative pidgin features that have been mentioned as characterizing Swahili, the reduction of inherited tense inflections, took place then. The loss of gender #13 also occurred at this time, as did a number of other lexical innovations and phonological changes not mentioned above. Some of these latter reflect contact with neighboring Bantu and Cushitic communities. The nature of that contact and the reason for the loss of tense distinctions is not known. The next identifiable stage is that of Proto-Sabaki. Characterizing this stage are the appearance of na- "progressive" and the loss of -aga "imperfective", and the replacement of the three locative prefixai classes by a single allembracing locative suffix. There is also — not mentioned above — a set of loans, both from Cushitic and from other Bantu languages. Since the third identifiable stage, Proto-Swahili, was only a century or two later than Proto-Sabaki, and since both communities are likely to have
Prior Pidginization and Creolization in Swahili?
289
been located on the coast, Proto-Swahili is treated here also. This is distin guished by the innovations (a)li- "past" and ta(ka)- "future", initially auxilia ries, and later grammaticalized TMs; incipient replacement of the preprefix and of #12 "diminutive"; and a set of loanwords, including considerable material from other Bantu languages and from Southern Cushitic as well as a very limited number of items from Persian and Arabic. At these two stages, Proto-Sabaki and Proto-Swahili, we see a clustering of some of the kinds of linguistic features commonly associated with pid ginization. What were the likely social correlates at this point? The ProtoSabaki and Proto-Swahili communities were the first Bantu communities on the part of the coast where we assume they developed, the northern Kenya coast. Southern, and possibly early Eastern, Cushitic communities were cer tainly present. The archaeological evidence shows the earliest Swahili com munities on this part of the coast to have been mixed — farmers, fishermen, and cattle keepers to some extent, and small-scale traders (Horton 1980, 1986) with merchants from southern Arabia. There is no sign here of Islam, extensive Arabic settlement or penetration, or plantations, so the relationship with outsiders was probably an adstrate one; but it was possibly different with Cushitic communities. The Cushitic loanwords and the phonological evi dence (see below) suggest that Cushitic speakers could have been assimilated into this early Swahili society and/or that, adjacent to the Swahili community, the conditions existed at least for Swahili as a lingua franca, and possibly for a form of pidgin Swahili among these Cushites, who provided the Swahili with some of the goods they needed for trading with outsiders. As far as we know, they were the only other local people present in this place at this time. Slightly later, in the closing centuries of the first millennium, speakers of early northern Swahili dialects lived in much the same area. Early northern Swahili is characterized by a large set of loanwords from both early Southern Somali (an Eastern Cushitic language) and Southern Cushitic, and by the dentalization process described above in §3.2, which I ascribed (Nurse 1985) to the influence of a Cushitic community. At or about this point, too, penulti mate stress probably started to replace tones or pitch accent. All this does not suggest a new situation, but rather a continuation of what has been outlined above. By A.D. 1200, Swahili communities had reached new economic heights, now more clearly based on trade. Communities grew larger and more numerous, Islam had appeared and the number of imported goods grew, and to obtain these imported goods, local goods had to be found for exchange. By
290
Derek Nurse
this time, Swahili traders were not only dealing with adjacent hunting com munities, who provided skins, ivory, and other products, but they were also trading farther inland, with other Bantu-speaking communities, and far south down the coast, with the gold-producing region of Zimbabwe. A possible outcome — though not necessarily the only one — of this multilateral trading situation might have been an expanded pidgin, or pidgins. For the next four or five centuries we see the emergence of other features often associated with pidgins: new grammaticalized auxiliaries at TM starting to replace older inflectional markers (e.g. me- for -ile "perfect") and the limited loss of agreement. By the seventeenth and early eighteenth centuries, after the Portu guese exit, Omani Arabic influence had started to reassert itself and we see the familiar Arabic phonological intrusions carried in the wake of an increas ing wave of lexical loans. The nineteenth century saw the establishment of plantations, with huge numbers of non-L1 speakers crowded together, espe cially on Zanzibar and Pemba. We know almost nothing about linguistic conditions on these plantations, but it seems very likely that Swahili-based pidgins were used. We also do not know if they were a continuation of the pidgins that may have existed since the late first millennium A.D., starting in the north.
5.
Conclusions
In principle, to establish prior pidginization and creolization, we need to be able to show adequately that processes widely associated with pidginization and creolization have in fact played a role in the development of the language in question, and we must be able to offer evidence of the conditions under which the "deculturization/acculturation of outsiders" (Polomé 1980:197) took place. For the period in question, 800-1800 A.D., we have had difficulty with both. There is almost no objective external evidence for conditions along the coast. Written accounts are inadequate, oral traditions do not touch on most of the period or most of the issues, and archaeology is silent about languages: pots do not speak. We have no samples of any early pidgin forms of Swahili, and we are not likely to find any. We have therefore had to extrapolate from the linguistic evidence. Although some of the diachronic processes we can show to have taken place in Swahili are elsewhere documented for pidgins
Prior Pidginization and Creolization in Swahili?
291
and creoles, especially in morphology, we cannot show conclusively that these occurred over a short period, as we would expect if they resulted from pidginization, nor can we show that they did not result from processes also known to operate widely in nonpidgins and noncreoles. My main conclusions, then, are these: (i) It is POSSIBLE that certain morphological replacements and losses in early Swahili itself resulted from the assimilation of speakers whose first language was not Swahili (i.e. speakers of Cushitic) into Swahili society. (ii) It is LIKELY that conditions on the northern Kenya coast from 800 A.D. onwards were conducive to the emergence of pidgin forms of Swahili adjacent to the Swahili communities, in that the earliest Swahili communities on islands and the coastal strip depended to a considerable extent on local adjacent communities, and later, as middlemen, they had to trade in both directions with local suppliers and outsider traders. The linguistic evidence suggests heavy early contact with adjacent local peoples. These conditions obtained from Proto-Sabaki (ca. 500 A.D.) through Proto-Swahili to the end of the first millennium, while in Swahili itself we see the emergence of features often associated both with pidginization and with creolization. (iii) The same kinds of linguistic features, though fewer in number, and the same socio-economic-cultural conditions then obtained the length of the coast, from southern Somalia to Mozambique, from the late first millennium into the present century, with local variation. (iv) The structures of early Swahili cannot be shown to result from significant linguistic contact between Arabic speakers and Bantu speakers. (v) Even in later periods it cannot be shown that Arabic speakers played any direct role in restructuring Swahili. The Arabic contribution to Swahili dates mainly from the last four centuries and is primarily lexical, with a limited number of nonlexical features having been introduced via the lexis. Although many of the nineteenth-century plantations were owned by Arabs, and led to conditions in which pidgins could well have existed, Swahili would have been the target language, and direct Arabic presence was probably minimal.
Notes 1.
See also Heine 1973, Scotton 1969, and Whiteley 1974 for descriptions of other East
292
Derek Nurse African pidgin forms.
2.
It should be emphasized that the dates in this paragraph and the preceding section do hot rest on the use of glottochronology, but on the juxtaposition of linguistic and archaeologi cal data. Currently, the earliest archaeological dates for the Swahili coast are ca. 800 A.D., and for the general NECB area are in the first century A.D. If Proto-NECB is assigned to some two millennia ago, then the approximate dates for Proto-Sabaki and Proto-Swahili can be assigned over the following eight centuries on the basis of the relative differences between NECB and Sabaki languages, and between them and Swahili dialects. For more detailed reasoning, see Nurse & Hinnebusch (1993). At least the Swahili dates are accepted by most scholars, e.g. Ohly (1982) and Whiteley (1969).
3.
This has occurred in all NECB languages except the northern dialect Mwiini. Loss of distinctive vowel length appears to have been a diffusional feature.
4.
The exceptions are Mwiini, at the northern end of the Swahili spectrum, and Mwani (a closely-related language), at the southern end of the spectrum. They, together with Comorian, have a pitch-accent system, which might be interpreted as an intermediate step from 'tone' to stress. One Zanzibar dialect has something other than penultimate stress.
5.
The genders in Bantu languages, both synchronic and reconstructed, are assigned con ventional numbers. Each number is assigned to a separate singular or plural category, not to the pairing of these. Thus #1 refers to the mu-, singular, gender, #2 to ba-, the most widespread plural form of mu-, #3 again to mu-, but different from #1 mu- in that it cooccurs with a different set of nouns, and its most widespread plural is mi-, #4, etc. Notations such as #1-2, #3-4, #5-6, #7-8, #9-10, #11-10, #12-13, and #16-18 indicate common pairings. Proto-Bantu has over twenty such genders but most contemporary languages have somewhat less. While this common numbering system may do occasional injustice to the description of individual contemporary languages, it has the advantage of underscoring the similarities between them. I use the term 'gender' for what others call 'noun class'. For a discussion, see Lyons (1968:283-88) and Greenberg (1978:49). 'Gender' here has no reference to biological gender.
6.
Active nominal use of the three locatives occurs in Mwani, where its presence is ambiguous: it could be retained from Proto-Sabaki or, more likely, it could be borrowed from adjacent Mozambique languages, where it is also active.
7.
This is the general Swahili system. A few dialects have one or two more.
8.
Two other features of the Swahili gender system are ignored here because they do not seem relevant to the main issues. One is the fact that, at least recently, two genders, #5-6 and #9-10, have attracted a disproportionate number of new members. This is hard to explain since we do not know all the reasons why new nouns are assigned to genders, and also since such catch-all genders are also a feature of other Bantu languages. The other is "animate recategorization" (Wald 1975), whereby animates, whatever their primary gender, tend to have concord forms in #1/2, the human animate gender. Such recategorization, an innovation of the last two or three centuries, is not obviously due to influence from Arabic or neighboring languages.
9.
Mood does not fit well into this schema and is largely ignored here.
Prior Pidginization and Creolization in Swahili?
293
10.
Heine (1973, 1979) has a long list of other features of pidginized Bantu verbal systems, but they are not particularly relevant to Swahili.
11.
By 'Swahili' here is presumably meant Standard Swahili. Mwiini and Mwani have even more reduced systems than Standard Swahili.
References Bender, M. Lionel. 1987. "Some possible African creoles: a pilot study". Pidgins and creoles: Essays in memory of John E. Reinecke, ed. by Glenn G. Gilbert, 37-60. Honolulu: University of Hawaii Press. Bertoncini, Elena. 1985. "Quantitative analysis of Swahili vocabulary". Studies in Afri can Linguistics, Supplement 7.22-26. Broomfield, G. W. 1931. "The re-Bantuisation of the Swahili language". Africa 4/1.7785. Brumfit, Ann. 1980. "The rise and development of a language policy in German East Africa". Sprache und Geschichte in Afrika 2.219-329. Chami, Felix A. 1988. "Excavation of a coastal early iron age site in Kisarawe District, Tanzania". Nyame Akuma 30.34-35. Cooper, Fred. 1977. Plantation slavery on the east coast of Africa. New Haven: Yale University Press. Fabian, Johannes. 1986. Language and colonial power: The appropriation of Swahili in the former Belgian Congo. Cambridge: Cambridge University Press. Freeman-Grenville, Grenville S. P. 1962. The East African coast: Select documents from the 1st to the earlier nineteenth century. Oxford: Clarendon Press. Greenberg, Joseph H. 1978. Universals of human language, vol. 3: Word structure. Stanford: Stanford University Press. Hancock, Ian F. 1971. "A survey of the pidgins and creoles of the world". Hymes 1971.509-23. Heine, Bernd. 1970. Status and use of African lingua francas. Munich: Weltforum Verlag. Heine, Bernd. 1973. Pidgin-Sprachen im Bantu-Bereich. Berlin: Dietrich Reimer. Heine, Bernd. 1979. "Some linguistic characteristics of African-based pidgins". Readings in creole linguistics, ed. by Ian F. Hancock, Edgar C. Polomé, Morris Goodman, and Bernd Heine, 89-98. Ghent: E. Story-Scientia. Holm, John. 1989. Pidgins and creoles, vol. 1: Theory and structure. Cambridge: Cambridge University Press. Horton, Mark. 1980. Shanga 1980: An interim report. Nairobi: National Museums of Kenya. Horton, Mark. 1986. "Asiatic colonization of the East African coast: The Manda evi dence". Journal of the Royal Asiatic Society 1986/2.202-13. Hymes, Dell, ed. 1971. Pidginization and creolization of languages. Cambridge: Cam bridge University Press. Krumm, Bernhard. 1940. Words of oriental origin in Swahili. London: Sheldon Press.
294
Derek Nurse
Legère, Karsten. 1986. "Kimgao and southern Swahili in mainland Tanzania". Lin guistische Studien, Series A, 148.146-59. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: CUP. Maw, Joan, and David Parkin, eds. 1985. Swahili language and society. Vienna: AfroPub. Mufwene, Salikoko. 1989. "La crèolisation en bantou: les cas du kituba, du lingala urbain, et du swahili du shaba". Etudes créoles XII/1.74-106. Nurse, Derek. 1985. "Dentality, areal features, and phonological change in Northeastern Bantu". Studies in African Linguistics 16/3.243-79. Nurse, Derek. 1987. "Towards a typology of diachronic phonological change in Bantu languages". Journal of the Atlantic Provinces Linguistic Association 9.100-122. Nurse, Derek, and Thomas H. Hinnebusch. 1993. Swahili and Sabaki: A linguistichistory. Berkeley: University of California Press. Nurse, Derek, and Thomas Spear. 1985. The Swahili: Reconstructing the history and language of an African society 800-1500. Philadelphia: University of Pennsylvania Press. Ohly, Rajmund. 1982. Swahili - the diagram of crises. Vienna: Afro-Pub. Polomé, Edgar 1971. "The Katanga (Lubumbashi) Swahili Creole". Hymes 1971. 5761. Polomé, Edgar 1980. "Creolization processes and diachronic linguistics". Theoretical orientations in creole studies, ed. by Albert Valdman and Arnold Highfield, 185-202. New York: Academic Press. Polomé, Edgar 1985."Swahili in the Shaba region of Zaire". In Maw & Parkin 1985, 47-65. Vienna: Afro-Pub. Reusch, Richard. 1953. "How the Swahili people and language came into existence". Tanganyika Notes and Records 34.20-27. Schmidt, Peter R. 1988. "Eastern expressions of the 'Mwitu' tradition: Early iron age industry of the Usambara Mountains, Tanzania". Nyame Akuma 30.36-37. Scotton, Carol Myers. 1969. "A look at the Swahili of two groups of up-country speakers". Swahili 39.101-110. Tucker, Archibald N. 1946/47. "Foreign sounds in Swahili". BSOAS 11/4.854-71; 12/ 1.214-32. Valdman, Albert, and Arnold Highfield, eds. 1980. Theoretical orientations in creole studies. New York: Academic Press. Wald, Benji. 1975. "Animate concord in Northeast Coastal Bantu". Studies in African Linguistics 6.267-314. Wald, Benji. 1981. "Swahili pre-pidgin, pidgin, and depidginization in Coastal Kenya: A systematic discontinuity in non-first varieties of Swahili". Historicity and variation in creole studies, ed. by Arnold Highfield and Albert Valdman, 7-24. Ann Arbor: Karoma. Wald, Benji. 1985. "The influence of Arabic on the syntax of Swahili discourse". Studies in African Linguistics, Supplement 9.315-317. Whiteley, Wilfred H. 1969. Swahili: The rise of a national language. London: Methuen. Whiteley, Wilfred H. 1974. Language in Kenya. Nairobi: Oxford University Press. Williams, M. Joseph. 1977. Origins of the English language. New York: The Free Press.
Michif: A mixed language based on Cree and French* Peter Bakker University
of
Amsterdam
and Robert A. Papen Université
du Québec à
Montréal
1. Introduction Michif1 is a language in which, broadly speaking, the verbs come from one linguistic system and the nouns from another. It is spoken as a first language by several hundred "Métis" — persons of mixed American Indian-European ancestry in various areas of the Canadian prairie provinces as well as in North Dakota and Montana in the United States. With very few exceptions, the verbs in Michif are Cree (an Indian language of the Algonquian family) and the nouns are French. Demonstratives and question words are Cree. The dialect dealt with here is mainly that of the Turtle Mountain area of North Dakota, though we also provide information on Manitoba and Saskatchewan dialects. Dialect differences do not cause difficulties; speakers of all dialects understand one another easily. The examples given below come primarily from a dictionary compiled by two native speakers (Laverdure & Allard 1983) and from our own field work.2 First we briefly describe the sociopolitical history of the Métis Nation (§2). We then discuss the most important languages spoken by the Métis (§3). In §4 we describe the structural features of the Michif language; this will be a relatively long section, as much more has become known about Michif since
296
Peter Bakker and Robert A. Papen
the only other grammatical sketch was published (Rhodes 1977). In the final section we present a number of hypotheses concerning the origin and the nature of this mixed language.
2.
The Métis Nation: history and present situation
The original meaning of the word "métis" or "métif ' is a person of mixed ancestry. In the Canadian situation this meant French and American Indian origin, from the beginning of the colony (Giraud 1945[1986]). In English the Métis, as well as people of mixed British-Indian ancestry, were usually called "halfbreeds". The Métis sometimes use this term for themselves when speak ing English. The word "Métis" (often spelled without the accent) came to cover all persons of mixed Indian and European ancestry. The Métis started to develop a new culture, integrating elements from both societies and developing distinct cultural traits. They used to come together in the Red River Valley in Manitoba seasonally, but they lived a more or less nomadic life following the bison herds in other seasons. Only the descendants of these people will be dealt with in this paper and will be called Métis. These historical Métis are sometimes called the "Red River Métis". The term "Métis" has quite different connotations today than it had in the nineteenth century, when it referred exclusively to the French-speaking halfbreeds of the Red River colony. Today the term (variably pronounced /metí/, /metis/ or /metis/) more commonly refers to anyone who has mixed Indian and White ancestry and who does not have legal Indian status. However, many people who fit that description do not use the term to describe them selves, and those who accept it cover a wide range, from people living in small French-speaking communities to those living on or at the fringes of Indian reserves. In the present paper, however, the term "Métis" refers only to the descendants of the Red River Métis. In the beginning of the colony in the 17th century, unions between French Canadians and Indian women were not at all uncommon. In fact, they were for a brief time officially encouraged by the Church and by the political leaders (e.g. Champlain, Talon, and even Louis XIV, who provided a financial subsidy for mixed marriages). Generally, offspring of such unions were integrated into their mother's tribe or into the Canadian French population. They did not identify themselves as a separate group of "Métis" (Dickason 1985).
Michif: A Mixed Language based on Cree and French
297
Starting in the latter part of the 17th century (1668-1725), the British Hudson's Bay Company (HBC) succeeded in establishing a number of trading posts on the western and eastern shores of Hudson Bay, but the first real and consequential contact between European men and Indian women in the West occurred some fifty years later (1731-1753), when Sieur de la Vérendrye and his sons established a number of fur-trading posts in central Canada, stretching from Fort St. Pierre at Rainy Lake (west of Lake Superior) to Fort La Corne a few miles east of the junction of the northern and southern branches of the Saskatchewan River in what is now central Saskatchewan. These trading forts were manned by groups of twenty to thirty men on extended duty. It thus seems that the existence of such large groups living among the Indians "was sufficient to lead to contacts between the Whites and the Native peoples that were numerous and lasting enough for us to assume that the first fusion of races took place during the preliminary stage of French penetration" (Giraud 1945[1986], 1:215). At the same time, and despite official HBC policy, unions between HBC employees and the local population created a new European-Indian population in the vicinity of western Hudson Bay. These people generally continued to live with their Indian mothers. Most HBC employees came from Scotland and spoke dialects of English and Gaelic. Métis expansion did not really began until after 1784, when a new Montreal-based trading company was founded, the North West Company (NWC), which was to become a most serious and highly successful rival of the HBC. The new company expressly hired French-speaking employees from what was now called Lower Canada (Québec). Even though the company leaders were mostly Scottish, all guides, servants, interpreters, voyageurs (canoe-men), helpers, and a number of traders (called bourgeois) operating major trading forts were French-speaking Canadians. The new company's policy was to continue the standard practice in the West. The employees lived in close contact with the local tribes in order to trade with them. They quickly married Indian or Métis women. As these couples often had many children, the Métis population increased rapidly. In 1810 Lord Selkirk obtained from the HBC 100,000 square miles of territory in the valley of the Red River in order to create a settled colony that he called Assiniboia. The colony, around present-day Winnipeg, was to be settled by mostly Irish and Scots colonists. For obvious reasons, the North West Company and their French-speaking Métis employees were opposed to the colony. The NWC fostered the idea of a Métis nation and the fact that the
298
Peter Bakker and Robert A. Papen
Métis were the "Lords of the Soil". In 1815, the Métis officially proclaimed themselves as "la Nouvelle Nation des Bois Brulés", insisting on their right to the land as native sons. Meanwhile, the two trading companies and indeed the two mixed Euro pean-Indian linguistic groups (speaking English and French, in addition to Indian languages) continued to be rivals until the HBC absorbed the NWC in 1821. This merger forced many employees of the NWC to establish them selves in the colony. The Métis population rose rapidly. In 1870, for instance, there were 12,000 inhabitants in Manitoba, 6000 of them French-speaking Métis, 4000 English-speaking Métis, and only 1500 Whites and 600 Indians (Howard 1952:147). The majority of the French-speaking Métis did not settle down to agricul ture, however. They preferred the less sedentary activities of bison hunting — which had become a major industry, feeding the growing size of the HBC employee roster — and freighting for the company, particularly south to the U.S. and west to present-day Saskatchewan and Alberta. In 1869, two years after confederation, Canada obtained all of the territories in the Northwest which had until that time belonged to the HBC. These included, of course, the Métis settlements in the Red River valley. The Métis were afraid of the everincreasing encroachment of White immigrants on their traditional lands and way of life, and Métis nationalist sentiment was again whipped up. Under the leadership of Louis Riel, the Métis established a provisional government in 1869. This sudden action forced the Canadian government to negotiate with the Métis. These negotiations led to the Manitoba Act, by which Manitoba became part of Canada. Among other things, the Manitoba Act promised that 1,400,000 acres of land would be distributed among the Métis. This would help them hold their own against the flow of new Canadian immigrants. However, Métis rights were soon ignored, and in many cases their lands were expropriated illegally or bought for a pittance by unscrupulous land specula tors. Large numbers of Métis, who were used to hunting bison farther to the west, left the newly created province and settled in the Northwest Territories, the name then given to all the lands beyond Manitoba. They stayed in great numbers on the Saskatchewan River, particularly at traditional wintering places such as St. Laurent and St. Antoine de Padoue, also known as Batoche. However, the same problems concerning land rights arose. To organize themselves into a political collectivity, the Métis of Saint Laurent and Batoche
Michif: A Mixed Language based on Cree and French
299
met in 1873 to form a provisional government, using their laws of the bison hunt as its basis. Despite this, the Canadian government continued to encourage immigrant farmers to settle in the fertile areas claimed by the Métis. Also, huge tracts of land were awarded to the railroad companies which were pushing the rails to the Pacific. The Métis thus quickly found themselves in the same situation that had driven them from the Red River. In 1884 the Métis of St. Laurent sent for their former leader, Louis Riel, who had since become an American citizen and who was teaching school in Montana. Riel and the Métis, aided by several Indian bands (Crees led by Poundmaker and Big Bear), proclaimed them selves to be in open rebellion against the federal government. Unfortunately, hasty actions on the part of some insurgents led to destruction of property, which in turn led to loss of support from the English-speaking "halfbreeds" and the White settlers. From then on the movement was restricted to Frenchspeaking Métis only. After a brief encounter in which the Métis were victorious, the Dominion forces, led by General Middleton, quickly overcame the insurgents at the Battle of Batoche. Riel was taken prisoner and sent to trial for treason in Regina, where he was found guilty and sentenced to death by hanging. It was only in the 1990s that Louis Riel was rehabilitated by the Canadian authori ties. Many of the Métis who had taken part in the insurrection, fearful of reprisals from the government and the Whites of the Northwest, fled the area. Some went to the United States (mostly to the Turtle Mountain area of North Dakota, where there was already a large Métis population), and others drifted back to Manitoba or to the missions on the Qu'Appelle River. Still others trekked farther west to Alberta or north around and even beyond Lake Athabasca.
3.
Métis languages: historical and contemporary
The Métis, with their multiple ancestry, had a long tradition as multilinguals. The most important languages are French, Cree, Saulteaux/Ojibwe,3 Michif,4 and, more recently, English. After a short discussion of Métis language use in the past, sketches of these languages will be briefly presented.
300
Peter Bakker and Robert A. Papen
3.1. Historical sources The Métis are descendants of French voyageurs and fur traders and Indian women. The children born in these marriages à la façon du pays must have grown up in a bilingual and bicultural enviroment. In historical sources one can find many references to Métis languages. Usually these refer to Métis multilingualism, Métis as interpreters, and language mixing on the part of the Métis. Dowse, for example, talks about the Métis as "this large favorable intermediate class, speaking both the Indian and the civilized languages" (1877:11). However, when "mixing of languages" is mentioned, it is usually said to be a mixture of French, Cree, and English, and not just Cree and French as we would expect, as in this account by a European adventurer who had learned Cree (McDougal 1903:271-72): "We met a French half-breed that morning, and I inquired of him as to water on the road between here and the South Branch. He began his reply in Cree, then went into broken English, and was bringing in some French when I quietly interjected an inquiry if he could speak Cree. He laughingly apolo gised and then became intelligible."
Here the author may be referring to Métis code switching, not to a mixed language. He seems to know Cree and French as well. A few lines later, however, McDougal predicts the emergence of something like Michif: "This was how languages had been formed. Here was a people who, if left to themselves long enough, would construct a distinct language out of a fusion of English, French, Cree and Saulteaux."
Code mixing was already common among the French voyageurs before 1850, according to Ross (1856[1962]:79): "They [the voyageurs] are generally great talkers, have long yarns to tell, and are not over scrupulous in their narrations, which are made up of an almost unintelligible jargon of the English, French and Indian languages."
This same Alexander Ross, who had retired to the Red River settlement and who was intimately acquainted with the Métis, mentions the existence of a language that might have been Michif, although it could refer to code mixing too (1856[1962]:200): "the vernacular of both Canadians and halfbreeds" at Red River "is a provincial jargon of French and Indian mixed up together." However, the first unambiguous reference to Michif in print known to us dates from 1947, though it refers to the 1930s (De Mishaegen 1947; our translation):
Michif: A Mixed Language based on Cree and French
301
"[The Métis] usually settle near the borders of the towns of the Whites or they form separate colonies such as Camperville at Lake Winnipegosis, where a language is spoken that is a mixture of French and the language of the Crees."
From these and other references to Métis language use (additional sources can be found in Bakker 1996), we can conclude that Cree and French (among other languages) were both spoken by large parts of the Métis population, and that code mixing was frequent. In 1930 Michif definitely existed. Although there is no written document proving this, Michif must have been the lan guage of daily intercourse among the Métis of Red River at least from the first decades of the 1800s. In any case, French and Cree were also spoken by many of these people, beside the mixture. Historical sources also mention that, for numerous Métis, Cree (or Ojibwe, a related Algonquian language) was the better known of the two languages. In fact, Cree monolingualism was not at all uncommon among uneducated Métis, particularly those living well away from the Fort Garry area (at the site of present-day Winnipeg) (Giraud 1945[1986]). 3.2. Source languages: Métis French and Métis Cree Michif is based on Métis French (Papen 1993) and lingua franca Plains Cree. Both differ in certain ways from Canadian French and Plains Cree. The French language is still spoken by the Métis and also by a few elderly Indians on the reserves in the West. This type of French is readily distinguishable from any other variety of French. Although historically de rived from Québecois French and perhaps Acadian French, it has undergone a different evolution, having been isolated from its Québec roots and coexisting with Cree and Ojibwe and now with English. Phonologically, Métis French stands apart from all other French dialects in that standard French /e/ and /o/ are systematically raised to [i] and [u], thus coalescing with /i/ and /u/ respectively. This sound change must have taken place at least 150 years ago, since George Catlin, travelling in northern Minnesota in the 1830s, painted a Cree chief called Bro-cas-sie, He who has Eyes Behind him (also known as Broken Arm) (Catlin 1844[1973]:55, 57). Apparently Catlin had not recognized the Métis French pronunciation of bras cassé (lit. "arm broken") as French, thinking instead that it was Cree. It should be noted that mid-vowel raising is not totally unknown in Canadian French varieties.
302
Peter Bakker and Robert A. Papen
In addition, in French words containing two sibilants, both must be identical, either both hushing or both hissing. For example, chaise "chair" is pronounced [sεz] as if the word were seize "sixteen"; sauvage "Indian" is pronounced either [savaz] or [šavaž]. This is clearly due to influence from Cree and/or Ojibwe. A further phonological distinction which sets Métis French apart is the palatalization of dental plosives before high front vowels: /t/ and /d/ are pronounced [č] and , respectively. In Québecois French these plosives also become affricates in identical positions, but the affricates are alveolar [ts] and [dz] rather than alveopalatal. Again, alveopalatal affricates are not unknown in Canadian French. Métis French also differs lexically from other Canadian French dialects in that some of its lexical items are not known in Canadian French dialects, e.g. boulet [bule] "ankle" ("fetlock" in standard French) and ponque [p k] "rec tum" (origin unknown). Furthermore, all Métis French dialects use a number of Cree or Ojibwe words differing in numbers and kinds from place to place. The syntactic differences between Métis French and Canadian French dialects are few. Possession is expressed syntactically as l'homme son livre where other dialects have le livre de l'homme or le livre à l'homme. This is a direct copy of the Cree construction na:pe:w o-masinahik, lit. "the.man his-book".5 In most Métis French dialects definite articles6 follow numerals, e.g. deux les chevaux "two horses" for either deux chevaux or les deux chevaux (the order in Cree is also numeral - demonstrative - noun). Further, in all dialects we are familiar with, the third person pronoun, particularly the plural form, is usually ça for referring to persons as well as objects, perhaps to avoid possible confusion between masculine il(s) and feminine elle(s) (see e.g. Douaud 1985). Other differences between Métis French and Canadian French can be found in Papen (1984a,b) and Douaud (1985). As for the Cree part of Michif, this also differs slightly from Plains Cree as described and recorded by Bloomfield (1930), Wolfart (1973), and others. Cree was the lingua franca of the Canadian Plains, spoken as a second language by Blackfoot, Assiniboine, Dene/Chipewyan, Ojibwe, and others. Not much is known about this lingua franca variety, except that it differs only minimally from Plains Cree as a first language, just as in the cases of other Algonquian lingua francas (see Rhodes 1982). Ojibwes were numerous among the ances tors of the Michif speakers. Not surprisingly, then, there are a few clear
Michif: A Mixed Language based on Cree and French
303
phonological, morphological and lexical differences between the Cree compo nent of Michif and ordinary Plains Cree, many of which are reminiscent of Ojibwe. 3.3. Status of Michif Michif is typically a home language. English is spoken with outsiders, and usually by Métis in the presence of outsiders. The Métis consider Michif as their own language, specific to their ethnic group. It can be safely said that few, if any, speakers of Michif are monolingual nowadays. They all speak English too. Approximately one in three Michif speakers also knows Métis French. Few speak Cree in addition to Michif (perhaps one in thirty Michif speakers), and very few speak Michif, French, and Cree in addition to English. In Manitoba many Michif speakers also speak Ojibwe (often better than they speak Michif). In Turtle Mountain, few speakers of Michif know either French or Cree. Their knowledge of these languages is limited to French nouns and adjectives and to Cree verbs, as well as some lexical items belonging to minor categories from either or both languages, even if some of these speakers claim to be able to speak French.7 The speakers of Michif are very much aware of the mixed character of their language. They can easily say, for a given sentence, which part is French and which part is Cree, probably judging from phonological and morphologi cal properties of the words. The language usually has a low status. Crawford (1985a) cites negative comments on Michif by the Turtle Mountain Métis, e.g. "a slang language", "a bunch of jumble", and "a jargon". Nevertheless, it is interesting to note that, at an earlier time, Michif became the dominant language on the Turtle Mountain reservation, even to the extent of replacing Ojibwe, the language of the reservation's official tribal affiliation (Crawford 1976 and elsewhere). Some Métis call the language "Cree", though they quickly comment that it is mixed with French. But when asked to interpret for speakers of Cree, they usually fail: Cree and Michif are not mutually intelligible. It can be doubted whether speakers bilingual in French and Cree can understand Michif, let alone speak the language, but this has not been tested.
304
Peter Bakker and Robert A. Papen
3.4. Distribution of Michif There are unfortunately no statistical data on the location and number of speakers of Michif. Canadian government surveys on the language use of Canada's Native people do not recognize Michif as a separate language. In the 1981 census, 75% of the Métis were reported to have English as their mother tongue, 12.7% had an Algonquian language as their mother tongue, and 8.9% reported French as their first language {Canada's Native People 1984). Obviously, some of these might, and indeed must, be speakers of Michif. The government of Manitoba recently recognized Michif as a lan guage, but no data on its use are presently available. Various Métis organizations in Manitoba and Saskatchewan are cur rently promoting the language. Pemmican Press, a publisher linked to the Manitoba Métis Federation, published the first Michif dictionary in 1983, thus giving the language a wider reputation (Laverdure & Allard 1983). Typically, Michif is not spoken in a continuous area. The area in which Michif-speaking communities are found is vast, but these communities are generally situated far away from each other, with many other ethnic and linguistic groups living in the area between them, and also within the Métis communities. The only way to indicate where the language is spoken is to enumerate the individual communities and show them on a map. A list of the communities is given in our Appendix. 3.5. Other Métis languages: Cree, French, Ojibwe, Bungi, Brayet, and English Several other languages are also associated with Métis communities. In some Métis communities Plains Cree is spoken (a few Métis in Duck Bay, Manitoba, and thousands of Métis in Alberta), and in quite a few others French is spoken, in Manitoba, Saskatchewan, and Alberta. In northern Manitoba, many Métis speak the Swampy Cree dialect. Other Métis languages are Ojibwe (especially in Manitoba, where it is called Saulteaux), Bungi (virtually extinct; formerly spoken in Manitoba), and Brayet (probably extinct). Métis Ojibwe has not been studied or described, to the best of our knowledge; it may have features that distinguish it from other Ojibwe dialects. Bungi is a dialect of Scottish English with some Cree and Ojibwe admixture (Blain 1989). It might be that the Algonquian element has dimin ished in recent decades. The word Bungi is derived from the Ojibwe word
Michif: A Mixed Language based on Cree and French
305
panki: "a piece, a little bit", a name given in the 18th century to a band of Ojibwes who begged for food using this word (Chappell 1817:199; George Barnstorn in 1839, cited in Bishop 1975:203). In more recent times Bungee has been used to refer to an Indian person as well as to the language (Stobie 1968, Walters 1969-1970). The Bray et language is mentioned in Stobie (1971) as being a mixture of French and Ojibwe. A mixture of French and Ojibwe is also mentioned by Peterson (1981:178-179), referring to the Métis of the Great Lakes around 1800. The name of the language is derived from the Canadian French word braguette* "breech clout" or "Indian". No data on this language are known to us. None of our Métis consultants remembers such a language. English should also be mentioned here. As noted above, virtually all speakers of Michif are also fluent in English. In fact, most of them speak only English nowadays in their daily communications. Many Canadian Métis (and Cree) have a distinct accent in their English. The pressure of English is quite great, and it has had a pervasive influence on Métis French. Michif words borrowed from English sometimes retain their original pronunciation and sometimes show phonological adaptation. For example, le siding [sajdirj], un black eye [blækaj], le cutter [kAdsr], but le steam [liscim], la factory [lafæktri]. (See also §4.4.3 for further discussion of English influence in Michif.) Two other mixed languages should be mentioned briefly here. The Ojicree language, also called Severn Ojibwe, spoken in parts of Manitoba and western Ontario, is not a mixture of Cree and a European language, but of Cree and Ojibwe. This language is spoken by people who call themselves Cree, but who speak Ojibwe with Cree admixture. Perhaps a group of Cree mixed with Ojibwes, thus creating this "mixed dialect" (Todd 1970, Wolfart & Shrofel 1977). It is possible that similar mixed languages exist on the CreeOjibwe language border in Eastern Saskatchewan. Also, members of the mixed Cree-Assiniboine tribe (Sharrock 1974) continue to speak a mixture of the two languages (called Nehipwat?) on some of the South Saskatchewan reserves. This mixture must have developed some centuries ago; nothing is known about its structure, which is now nearing extinction. It should also be mentioned that many individual Métis, especially in the nineteenth century, also acquired other European languages — German, Ukrainian, etc., but especially Gaelic. A significant number of the HBC workers were Scottish Highlanders and spoke Gaelic, and it seems that some Natives and Métis learned some Gaelic too (see Stobie 1968).
306 4.
Peter Bakker and Robert A. Papen Structural sketch of Michif
As Michif has such peculiar structural features and a rather complex interac tion between Cree and French elements, it is necessary to give a rather detailed structural survey of the language, indicating what is French, what is Cree, and what is an innovation in Michif. Rhodes' 1977 structural sketch is good, but it is sketchy and not easy to obtain. Furthermore, it contains a number of mistakes, especially for the French part. In this sketch we use a phonological transcription for French and Cree,9 in which some Cree and French vowels of approximately the same value are nevertheless distinguished in writing because they do have different phonetic values, especially the allophones. Cree elements will be italicized in all of the example sentences. 4.1. Variability Michif is a homogeneous language; speakers from all the scattered communi ties can understand one another without any effort. Nevertheless, there is quite a bit of variation within any community as well as from one community to another. Crawford (1985b) lists the following differences within the Turtle Moun tain speech community: /s/ vs. /h/ (e.g. pi:htikwe: vs. pi:stıkwe: "Come in!"); nı- vs. dı- as first person marker; hi: vs. ti: as past tense marker; ši- vs. či- as conjunct verb marker; and varying forms of contraction, e.g. nıwi:- vs. its contracted formmi: (for the latter, we recorded ny- in southern Saskatchewan). Some of these sources of variation are present in all communities. This is not an exhaustive list. Also, the degree to which French and Cree are used in Michif varies from community to community and from person to person. This is elegantly dem onstrated in Laverdure & Allard (1983). In many instances, two different translations are given for a single English example sentence, one with more Cree and the other with more French. Examples are given in (1) and (2): (1)
a.
mıyuw-a:š-m gOOd-be-it
ka:-kıše:wa:t-ısı-hk. COMP-kind-BE-INDEF.ACTOR
b. mıyuw-a:š-m kã ti bũ čœr. good-be-it when you good heart "It's good to be benevolent."
Michif: A Mixed Language based on Cree and French (2)
a.
mıša:hta la m zũ. big.make.IMP the house
b.
ply gru ušta la m zũ. more big make.it.IMP the house "Enlarge the house."
307
There are also differences between communities. In Camperville, for in stance, more Cree (e.g. Cree adpositions) is used than in the other communi ties. However, it must be remembered that passive knowledge of a language is always greater than active knowledge. This makes it possible for people to understand one another in spite of differences in the amounts of French and Cree. Speakers from different communities therefore have no difficulty un derstanding one another. The varieties spoken in Manitoba, southern Saskatchewan, Montana, and North Dakota are so similar that they must have a common origin (for detailed argumentation, see Bakker 1996). 4.2. Phonological system In this section, after discussing the phonology of the French component and the phonology of the Cree component, we will examine the relationships between the two phonological systems in Michif usage. 4.2.1. The French component As mentioned in the introduction, the noun phrase is part of the French domain in Michif. French nouns and adjectives, as well as the determiners (but not the demonstratives), are French. The pronunciation of these forms is typically that of Michif French, as discussed in §3.2. The (surface) phoneme inventory is as follows (see n. 9 for information about phonetic realizations): CONSONANTS
VOWELS ORAL
i e
u
y œ a
a
P b f v m
NASAL
ũ ã
t d s z n r
č
k
J š ž
g
ñ
ŋ
1 w
j
308
Peter Bakker and Robert A. Papen
Some comments are in order. The French upper mid vowels (front unrounded and back rounded) do not occur as phonemes; only the front rounded ones do. Etymologically, mid vowels are almost always raised to the high position. This sometimes occurs also for the front rounded mid vowel /œ/, which can be raised to [y]. But most speakers maintain a distinction between [žy] "juice" and [žœ] "game", [ny] "nude" and [nœ] "knot", etc.10 Mid vowel raising also affects the back rounded nasal /o/, which is typically pronounced [u]; how ever, læ/ is not raised. Further, the French schwa is raised to [i] in Michif, though it obeys approximately the same deletion conditions as in French. The central low vowel /a/ is often pronounced [æ] rather than [a], possibly due to English influence. The mid vowel /ε/ is lowered to [æ] before preconsonantal or prefinal /r/: [færm] "farm" (< ferme), [bjær] "beer" (
Michif: A Mixed Language based on Cree and French
309
"half" (<moitié), etc. Velar stops are also palatalized before mid front vowels: /k/ becomes [č] and /gl becomes [j], as in [piče] "picket" (
310
Peter Bakker and Robert A. Papen
VOWELS
CONSONANTS
ORAL
p i e:
i:
u
a
t
u: m w
a:
š
h
n j
NASAL 1
Ü
Some of the nasal vowels in Michif are in free variation with a V/n/ sequence, as in the 3.inan.obj. suffix: /-e:n/ ~ /æ:n/ ~ /; /). Others seem always to be nasal vowels, apparently in all Michif dialects, such as ũhĩ "those (obviative animate or plural inanimate)"; this nasalization of Cree o:hi: is not caused by adjacency of nasal consonants. Métis French has no nasal /ĩ/, but Michif has. This is probably due to influence from Ojibwe (notably the Saulteaux dialect spoken in Manitoba). The Cree component of Michif also maintains Cree morphophonemic alternations. These include t-insertion between a vowel-final personal prefix and a vowel-initial stem and various vowel coalescence rules (e.g.V1V:2→ V:2, V:1V2 → V:1 and V1V2 → V1) (cf. Wolfart 1973:79ff.), although further research might reveal that the Michif rules differ somewhat from the Cree rules. Both Rhodes (1977) and Evans (1982) identify an innovative phonologi cal process that occurs in the Cree part of Michif but not in Cree: obstruents are voiced after a nasal, as in [ta:nde:] "where" (Cree tamte: or ta:nite:) and [na:ndaw] "anyhow, anything" (Cree na:nitaw or na:ntaw)\ this rule also involves a typical Cree vowel deletion rule and the t-insertion rule mentioned above. Together, these three rules give rise to surface voiced obstruents in Michif, as in (3): (3)
a.
# nı + ıtw + a:n# 4 1 say nı t ıtw n tıtw n dıtw dıtw [dıtwa:n] "I say it"
a:n a:n a:n a:n
b. #nı-še:kı-h-ıku-n # 1sg-fright-CAUS-INV-NON3 n-še:kıhıkun n-že:kıhıkun že:kıhıkun [že:kıhıkun] "it frightens me; I am alarmed"
Michif: A Mixed Language based on Cree and French
311
This also occurs word-internally: (4)
#pımı-pahta:-w# move-run-3sg. pım-pahta:-w [pımbahta:w] "he runs"
The French and Cree components also have slightly different rules for assign ing stress. Standard French has fixed stress on the last syllable of a "breath group" (usually a major syntactic phrase). Cree assigns secondary stress to the final syllable and primary stress to the antepenult; the penult is either extremely long or extremely short. (5)
a. maskisìn b. maskísinà maskisínihkè:w
"shoe" "shoes" "(s)he makes shoes"
Usually the stress systems of Cree and French are compatible, except in cases where French or English words have more than three syllables (which doesn't happen often in Michif). These seem generally to have Cree stress patterns and length patterns in the penult: (6) (7)
(French /kõbinezõ/ combinaison) "long underwear" "taxidriver"
4.2.3. Autonomy or convergence of the two phonological systems in Michif? Both the French and the Cree phonological systems remain independent in Michif. The phonological rules that operate in one component do not apply in the other. However, there does also seem to be some degree of convergence, though perhaps not as much as Evans (1982) would have it. The most commonly discussed area of phonological interaction is mid vowel raising, already mentioned in the section on Métis French (§4.2.1). Another area of possible convergence is sibilant harmony. In Michif, if a word contains two sibilants both must be identical, either both hushing or both hissing, as in [šeš] "dry" (<sèche), [šavaž] "Indian" (<sauvage), and [zezy] "Jesus" (< Jésus). Generally the process is regressive assimilation, though examples of progressive assimilation can also be found, e.g. [rises] "wealth" (
312
Peter Bakker and Robert A. Papen
where [s] and [š] are not distinguished. One finds the same in Cree English, where [su:] might refer either to shoe or to Sioux. A third point of possible convergence is vowel length. Some speakers of Michif seem to reanalyze the phonetic differences between some of the French vowels as length differences. For these speakers, the vowel pairs [i] [ı] and [a] [a] seem to differ not only in phonetic value but also in length, just as in Cree. In Quebec French there are also two types of /a/ and /ε/ that differ in length and quality, and sometimes also two types of /i/. Finally, three nasal vowels — , /ũ/, and — occur in the Cree component of Michif. In some words these reflect an original vowel + /n/ sequence; in other words the sequences are more or less mysterious, e.g. (Cree anihi) "those"and ci: "question particle". This is probably a result of influence from Ojibwe (although the corresponding Ojibwe words do not have nasal vowels here): Ojibwe has all three of these nasal vowels, while French lacks both /ü/ and /i/. Moreover, the French vowel /a/ is very common in the French component of Michif, but nonexistent in the Cree component. A further complicating factor is of course the growing influence of English on Michif, particularly but not only in the nominal lexicon. Some English sounds, e.g. and [ŋ], had achieved phonemic status already in Métis French, but it seems likely that, as more and more English items are borrowed, more and more phonological segments and processes will be adopted, such as diphthongs, stress assignment, and aspiration. The other possible convergent features are undoubtedly due to Cree and/or Ojibwe influence, but all of them are also characteristic of Métis French, the source French dialect of Michif. That is, the influence is not from the Cree compo nent of Michif on the French component, but rather from Cree on Métis French. Many areas of Michif phonology are not yet well understood, and much work remains to be done here. Nevertheless, it is clear that two separate phonological systems must be posited for Michif, systems that have remained separate since the inception of this mixed language. Nothing would be gained (in fact, much would be complicated) by trying to combine them into a single system, as many rules are limited to either the French part or the Cree part of the language. This implies that each lexical item must be marked [± French] or [± Cree] in the (mental) lexicon of Michif speakers in order to ensure that the item undergoes the right set of phonological rules. Similarly, each phono logical rule must be marked [± French] or [± Cree].
Michif: A Mixed Language based on Cree and French
313
4.3. The verb phrase in Michif The verb phrase is basically that of Plains Cree (as described in Wolfart 1973), but French verbs also play a minor role. Michif and Plains Cree verb paradigms can be found in Rhodes (1977). 4.3.1. The Cree verb phrase There is little reduction of Plains Cree verb paradigms in Michif. Only very few Michif speakers know the dubitative and preterit verb forms; but these are rare, defective, or nonexistent in most contemporary dialects of Plains Cree as well. Due to lack of space it is impossible to provide a full account of the Cree verb, so we will give a very simplified overview here. Like other Algonquian languages, Plains Cree has two main orders, the independent and the conjunct. The independent order is used in main clauses; the conjunct order is used in embedded clauses and in WH-interrogatives, and rarely in main clauses. They have specific negators: e:ka or e:ka:wija for conjunct verbs, namo:ja for independent verbs. In the conjunct order, person agreement is exclusively suffixing, while in the independent order person agreement is prefixing and suffixing. A complementizer prefix čı-lšı "future", e:- "contemporary nonindividuated", or ka:- "contemporary individuated" is required in the conjunct order. The subjunctive order (in " i f clauses) is marginal and is derived from the conjunct order by the addition of a suffix - i / -u to the conjunct verb form. There are four stem classes: inanimate intransitive (II), animate intransi tive (AI), transitive inanimate (TI), and transitive animate (TA). These have different inflectional markers for subject/object and often different stem forms as well. Tense is marked by prefixes: ka- "future", wi:- "volitional future", ki:"past". Present tense is unmarked. There are two imperatives, the immediate and the delayed imperative; they are negated by ka:ja. All verb forms have person marking; there are no nonfinite verb forms. Furthermore, the Michif verb seems to have the full derivational and compo sitional possibilities of the Plains Cree verb, with preverbs, object incorpora tion, voice and valency marking suffixes, agreement morphemes, etc. A morpheme slot for the Cree (and Michif) verb would look like Table l. 11
314
Peter Bakker and Robert A. Papen
Table 1. Morpheme order slots in the Cree/Michif verb. 1
2
CONJ order
TENSE/
marker or IND order 1/2 prefixes
MOOD
3 preverbs
4 stem
5 voice/ valency/ direction/ aspect?
6 person agreement
7 plural abs
SUBJ
Slots 3 and 4 can have any number of fillers. Slot 4, the stem, is complex (see below). The other slots in Table 1 can contain only a limited number of elements, i.e. only closed-class morphemes. These elements, for slots 1, 2, 5, 6, 7, and 8, are given in Table 2. All the possible morphemes from these slots are listed in the table, with short glosses.
Table 2. The closed class morphemes of the Cree/Michif verb. 1
2
5
6
7
CONJ:
TENSE/MOOD:
DIRECTION:
OBV. AGREEMENT:
pl.ABS:
SUBJ:
e:ka:ČI-FUT
O/ZC¿(NEG)PAST
a: (non-3SUBJ, 3OBJ) e: (3SUBJ, 3'OBJ)
yı OBV. SUBJ
ık/ak waw
-ıl-
ki: (POS)PAST
PERS:
nı-I kı-2
ka/kita FUT wi: VOL
ı (2SUBJ, 1OBJ)
kaki: IRR/POT
ıku (3SUBJ, non-3 OBJ)
ıtı (ISUBJ, 2OBJ)
ım POSSD.OBJ INDEP.AGREEM.: n w/n 3 VOICE/VALENCY: ıte: INAN.PASS?
8
: 1pl, 21 a:waw 2pl
ıka: ANIM.PASS?
ısu REFL Itu RECIP ıwı COP a:n INAN.PASS?
h CAUS st BEN aw EXTRA.ARG am INAN.OBJ ike: DETRANS
maka DEAN me:
ANIM.INDFZR
hka:su PRETEND
CONJ.AGREEMENT: (j)a:n 1 (j)an 2 (j)a:hk 1pl Ų)ahk 21 (j)e:k 2pl tlhtlklhk 3
Michif: A Mixed Language based on Cree and French
315
A list of the preverbs (slot 3) would be quite long; the stem (slot 4) can contain a potentially infinite number of fillers. Preverbs are typically elements such as kakwe:- "try", ma:čı- "start", and su:hkı- "strong(ly)". What is expressed with an adverb or an auxiliary in other languages is often expressed with a preverb in Cree. Cree verb stems are almost always complex and can be analyzed into smaller units — bound morphemes carrying meaning. Stems can contain three components, traditionally called initials, medials, and finals. It is often not easy to draw boundaries among stem morphemes, incorporated elements, derivational elements, and inflection. It would be far beyond the scope of this paper to discuss all the combina torial possibilities of these elements and the phonological processes that go with them. The direction markers (slot 5) might require some explanation. The Cree verb has a person hierarchy, given in (8): (8)
2nd person > 1st person > 3rd person > 3' person (obviative)
This means two things. First, if the agent is lower in the hierarchy than the patient, then the verb receives an "inverse direction marker", which indicates that the action goes against the hierarchy. Otherwise there is a "direct" marker. Second, whenever two or more person affixes are present in one complex verb, the relative order is as stated in (8). Some examples are given in (9): (9)
a.
nı-wa:pam-a:-w 1-see-DIR-3 "I see him"
b. nı-wa:pam-ık[u] 1-see-INV.3 "He sees me"
c.
kı-wa:pam-ıtı-n 2-see-INV-NON3 "I see you"
d. kı-wa:pam-/-n 2-see-DIR-NON3 "You see me"
It should also be mentioned that verbs in Cree agree in gender with their objects (in transitive sentences) and their subjects. Gender in Cree is based on animacy/inanimacy. Verbs often have different stems, e.g. wa:paht- "see it", wa:pam- "see him/her". They also have different suffixes: wa:pam-e:w "he sees him", wa:paht-am "he sees it"; nıwa:pam-a:w "I see him", nıwa:pahtæ:n "I see it"; etc. This is also true for Michif. Because Michif nouns derive from French, they must be marked for regular French gender (but cf. §4.4.1 below) in order to select the correct French determiner; in addition, however,
316
Peter Bakker and Robert A. Papen
they must be marked for Cree gender in order to guarantee the correct verb forms. How this is done is theoretically problematic, particularly as most Michif speakers do not know Cree (see also §4.4.4). Here is an example to illustrate the animacy distinction: (10) a. b.
ki:-mıčımm-e:w aitiht laržã. PAST-hold-.→3' some money ki:-mıčımm-am aitiht la pej. the payment PAST-hold-TI.3→4 "He kept part of the money/payment."
The verb endings differ according to the animacy of the object (in the example the verb stem happens to be the same for animate and inanimate arguments). In (10), French laržã "money" is animate because Cree šu:nıja:w is animate; but la pej is inanimate, presumably following Cree (inanimate) nominalized verbals in -wın, such as tıpahamakowm "payment". Rhodes (1977) gives a full list of Michif verb paradigms. Some forms not elicited by Rhodes are generally known, however, even at Turtle Mountain. He says, for instance, that the exclusive/inclusive distinction for first person plural is lost in the conjunct order in Michif; specifically, he says that exclusive -ja:hk and inclusive -jahk have merged into -jahk. Our data show that this is not the case, at least not for most speakers. Here are two examples of inclusive and exclusive verb forms in both the conjunct order and the independent order: (11) gi:-ša:pu-st-a:na:n lɨ rũ-d port ši-pi:stıkwe:-ja:hk. lPAST-pass-go-lpl.EXCL the circle-of door -enter-lpl.EXCL "We [but not you] walked through the archway to come in." (12) la munisjũ či: kı-t-aja:~na:n šı-papa:-ma:či:-jahk? the ammunition Q 2-r-have-lpl.INCL -about-hunt-lpl.INCL "Do we have enough ammunition to go hunting?" However, it is true that some speakers do not use the inclusive/exclusive forms in all the cases where they would be required in Cree. Another slight difference is the Michif form ki~V-a:na:n (lpl incl.) versus Cree kı-V-a:naw. Rhodes (1977) gives only immediate imperatives. In our data we have some examples of delayed imperatives as well:
Michif: A Mixed Language based on Cree and French
317
(13) te:pwe:-hk-an pari e:-aja:-jan-ı. call-DIMP.AI-2 ready COMP-BE-AI.2-COND "Call when you're ready." (14) wi:pat pe:-ki:wke:-hk-e:k. soon come-visit-DIMP.-TA.2pl→1 "Come visit me soon." Although Michif Cree verbs are in principle identical to Plains Cree verbs, the Michif Cree verb seems to be subject to a tendency to use analytic constructions where synthetic constructions are possible and perhaps somewhat more common in non-Métis Cree. Benefactives, for example, tend to become analytic in Michif Cree. Example (15)a shows the normal Cree synthetic form, while (15)b has the analytic form, with pur nija "for me" (15) a. b.
en sæcyr IAFsg belt
d-ušt-am-a:-šu-n. l-make-BEN-TA-REFL-NON3
en sæcyr d-ust-am-e:n pur nıja. l-make.it-BEN-4sg for me "I am making myself a belt."
4.3.2. Non-Cree verb forms Michif uses two types of French verbs, though rather marginally. First, some French (or English) verbal lexical forms are integrated into Cree verb mor phology; and second, some French verbs are conjugated as they would be in French. Both types will be discussed. Some French and English verbal lexical items can take the full set of Cree verbal inflectional affixes. Many of these verbs seem to refer to nontraditional actions, e.g. bénir "to bless", témoigner "to witness in court", and bousiller "to plaster". Others, however, do not fall into this category: astiner "to argue", broder "to embroider", arranger "to fix", etc. The way in which these verbs are integrated into the Cree verb pattern is rather odd. It is always the infinitive form of the French verb that is used, together with a prefixed dummy element which is homophonous with the French masculine definite article le and the French object clitic le. We gloss it as a masculine article, as it is not used with intransitive verbs and as it varies in gender with similar nominal constructions (exx. 20 - 23). The French verb retains its original phonological structure. These verbs have Cree person and
318
Peter Bakker and Robert A. Papen
can also have Cree tense/mood marking and preverbs. Examples are given in (16) and (17): (16)
pε:r müd. DAMsg priest PAST-DAMsg-bless-INF-TA.3→3' DAMsg people "The priest blessed the people."
(17) kiskejım-e:w su mεtr know.TA-TA.3→3' John his master C0MP-DAMsg-argue-INF-3pl "John knows who's his master when they argue." (French obstiner) Some English verbs (such as to celebrate, to haul, to settle, to deal, to can, to box, and to gamble), many of them denoting nontraditional activities, can be used in a similar way: they retain English phonemes, but require the French masculine definite article as well as the French infinitive ending -er [i], and also Cree prefixes and suffixes. (18)
kat di žyjεt gi:- -sεlɨbre:t-t-na:n. DAMsg four of July l.PAST-DAMsg-celebrate-iNF-AI.lpl "We celebrated the Fourth of July."
(19) nu ki:-kıške:ht-am la bãk NEG PAST-know.TI-TI.3→4 DAFsg bank e:-ki:- -rab-i-hk. COMP-PAST-DAMsg-rob-INF-IMPRS
"He didn't know about the bank being robbed." The combination of Cree verbal affixes and French elements is also found in the formation of Cree verbs from French nouns or adjectives. Some of these are derived by means of the Cree suffix -ıwı "to be"; the appropriate Cree inflectional endings are then added. This is a productive process. The resulting verbs have the approximate meaning "to be X-y". The suffix -ıwı does not seem to be used with Cree stems, however. The French definite article, in the appropriate gender, must always precede the French noun or adjective. (20) ti suji la-bu-rw-an-wa. your.pl shoe DAFsg-mud-BE-4-pl "Your shoes are muddy." (French la boue "the mud")
Michif: A Mixed Language based on Cree and French
319
(21) ki:-te:pIšk-am e:-la-mitrεs-rwI-t. PAST-succeed-TI.3→4 COMP-DAFsg-teacher-- "She succeeded in becoming a teacher." (French la maîtresse "the teacher") (22)
ki:-li-fu-Iw-Iw e:-li-pči-Iw-it. PAST-DAMsg-crazy-BE-3sg.ANIM COMP-DAMsg-little-BE-3sg "He was crazy when he was small." (French le fou "the crazy", le petit "the small")
(23) ma:ma: gi:-It-Iku-na:n e:-ka mother l.PAST-speak-INV-TA.3→lpl NEG la-tεt-ǰyr-iwi-ja:hk. DAFsg-head-hard-BE- 1p1 "Mother told us not to be boneheads." (French la tête dure "the hard head") In another productive construction, the Cree verb suffix -Ihke: "to make" may be added to French adjectives and nouns. Again, the French definite article is required: (24) li-žali-Ihke:-w. DAMsg-pretty-MAKE-TA.3→3' "S/he is decorating him/her." (French joli "pretty") Apart from the integrated French verbs that take Cree affixes, there are French verbs that can be used in bare forms. Probably all Michif speakers use at least some French verbs, often frozen expressions like alũ "let's go", sa prā "one has to" (lit. "it takes"), and the French copula (see §4.3.3 and §4.6.1). Speakers who know French can, in principle, use any French verb, but French verbs are nevertheless relatively infrequent. Speakers sometimes use forms that are not the ones that whould be expected in French, e.g. participles instead of infinitives or vice versa. In (25)a-b below the verb is ambiguous between the French infinitive form in -er and the participial form in -é(e) but in (25)c-d participles are used, where French would have an infinitive: (25) a.
i va pas-i. "It will pass."
320
Peter Bakker and Robert A. Papen b.
i va bɔk-i. "He will buck."
c.
i va ufєr sa plas. "He will offer his seat." (cf. French il va offrir "he will offer (inf.)"/// a offert "he has offered (part.)")
d. i va dispary. "He will disappear." (cf. French il va disparaître (inf.)/// est disparu (part.)) Standard French imperatives can consist of the root (in the singular) or the root plus suffix Id (in the plural). The latter, written -ez in French, is phonologically homophonous with participial and infinitive suffixes (-é(e) and -er, respectively). Michif imperatives can be derived from French singu lar or plural imperatives. There seems to be no plural/singular distinction between these forms, and they are sometimes used in combination. (26) ta:š-i ãǰyr li pwæ. attempt endure the pain "Try to endure the pain." French imperatives can trigger a Cree conjunct clause: (27) dikuraž ka:ja šI-nakat-a:t. discourage.IMP.2sg NEG coMP-leave-TA.3-^3' "Discourage him from leaving her." Our data provide a few examples of verbs with cliticized pronouns, but most of them seem to be more or less fixed expressions: (28) i-m-a-ufer "He offered me." i-1-a-noz-amyzi "He amused us." sa-s-akord pa "They don't get along." ži-1-flεt / ži flεt "I flatter him." i-vœ-s-marji "He wants to get married." The only French verb forms found in Michif are infinitives/participles and analytic forms with past tense and future tense. No other forms (notably the synthetic future and several synthetic past tenses) are used in Michif. There is a special place for the verbs "to be" and "to have"; these are the only French verbs that have rather elaborate French paradigms. Only 2pl forms are lacking in our data.
Michif: A Mixed Language based on Cree and French (29)
321
"to be": INFINITIVE pur jet and: PRESENT
PAST
lsg: žł /žł swi 2sg: tł 3sg: ilł/ sł lpl: ũ-1-ł 3pl: i-sũ:
žł-t-ε t-łt-ε i-l-łt-e / ũ -1-łt-ε i-sũt-ε
FUTURE
žł tł s(ł)t-ε ũ
va jεt va jεt i / sa va jεt va jεt i vũ jεt
(30) "to have" (future not attested) lsg 2sg 3sg lpl 3pl
PRESENT
PAST
ž-a t-a i-l-a, sa-l-a ũ:-l-a i/sa-l-ü
ž-av-ε t-av-ε i-l-av-ε ũ-1-av-ε i/sa-1-ũv-ε
The use of both of these verbs is rather limited. The forms derived from French "to be" are often used as a copula (beside corresponding Cree verbs in equative clauses). The inflected forms of the French verb avoir "to have" are used mostly as a tense auxiliary with French verbs and also in a variety of (fossilized) verbal expressions such as avoir besoin "need" and il y a "there are". It is used as an auxiliary even where French would use a form of être "be", e.g. in i-l-a vny "he came" and i-l-a arivi "he arrived" (French il est venu, il est arrivé). 4.3.3. Reanalyzed French verb forms Modality is the only area in which Michif shows significant innovation compared to French and Cree. It is not clear why the modality system is the locus of so much innovation. In Michif a French adverb or phrasal expression is used to introduce a proposition in the conjunct order (also, rarely, a French infinitive or complement clause), as in (31) and (32). (31) æ sǣ äkur a
dog WISH
e:-ja:w-ak. coMP-have-TA.l-»3
"I wish I had a dog." (32) saprä
ti äjyr li træ. NECESSITY you endure the noise "You have to endure the noise."
322
Peter Bakker and Robert A. Papen
Below is a list, probably not exhaustive, of the modality markers found in our data: (33)
ãkur pa mwajæ magre fu(lε)-bæ sa-prã sa-s-pura-bæ sa-sεr-da-rjæ si-pa-si
(< F. encore "again") "I wish" (< F. pas moyen "no means") "it is impossible" (< F. malgré "in spite of') "oblige, coerce" (< F. fallait/faut bien) "it is/was necessary" (< F. ça prend "it takes") "it is necessary" (< F. ça se pourrait) bien) "it is possible" (< F. ça sert delà rien) "it is useless" (< F. je ne sais pas si) "it is doubtful"
Numerous French (and some English) adjectives and, more rarely, adverbs and nouns occur with the impersonal French expression si (< F. c'est "it is"). These can be in the present or in the past tense, and they can be positive or negative: (34) s-ɨ-d-valœr (< F. c'est de valeur) s-ɨ-tεrib (< F. c'est terrible) s-ɨ-bæ-rar (< F. c'est bien rare) sɨ-b -n-a-swεti(< F. c'est bien à souhaiter) s-ɨtε-pa-nɨsɨsεr (< F. c'était pas nécessaire) (< F./E. c'est pas fair) s-ɨ-pa-fεr s-ɨ-t-izi (< F./E. c'est easy)
"it's too bad" "it's awful" "it's rare" "it's to be hoped" "it was unnecessary" "it's not fair" "it's easy"
Cree has impersonal constructions — also used in Michif — which might have served as a model for the reanalysis of the French forms. These Cree constructions have a conjunct verb form, but this does not seem to be obliga tory; in contrast, the French-based forms always have a complement with a conjunct verb. Some Michif examples: (35) ki:ja:m či-pe:htwa:-hk uta. allright coMP-smoke-IMPRS here "It is all right to smoke here." (36) piku NECESSITY
či-šıpwe:hte:-ja:n. COMP-leave-AI.1
"I have to go." (37) ma:škut ka-ıšpajı-n. apt FUT-happen-II.4 "It will probably happen."
Michif: A Mixed Language based on Cree and French
323
4.4. The noun phrase The Michif noun phrase is fundamentally French, in all its syntactic functions — subject, direct or indirect object, object of an adposition, etc. Nouns are obligatorily accompanied by a French possessive element or article (definite or indefinite, masculine or feminine, singular or plural). Cree demonstratives (animate or inanimate) can be added to the French definite article. In all dialects a few Cree nouns are also used, but they are rare. Except in Camperville, Manitoba, the number of Cree nouns used in Michif now here exceeds fifteen. Cree nouns generally have a French determiner: lɨ we:pınıke:win "the garbage", æ wi:htıkuhka:n "a wolverine", εn wi:nma:kuš "a dowdy woman" (cf. §4.4.2). A great number of English nouns are used. These may be phonologically adapted to French or not (see §4.4.3), and they almost always have a French determiner. All adjectives are French (see §4.4.4); Cree has no distinct class of adjectives. Other categories within the NP can be either French or Cree. These include quantifiers, numerals and degree adverbials (see below). The internal order of the noun phrase is given in (38): (38) NUM - DEM - ART - ADJ - noun - ADJ - S 4.4.1. French nouns Most French nouns in Michif are identical to nouns found in archaic dialects of Canadian French, except for phonological differences (since a number of French nouns have been reanalyzed phonetically in Michif). Some vowelinitial nouns are reanalyzed with an initial consonant, either by agglutination of a consonant segment taken from a French determiner (39)a or by truncation of the initial syllable (39)b. See §4.2.1 for an argument that the initial consonants in (39)a are part of the noun rather than the result of a liaison rule. (39)
a.
AGGLUTINATION:
zafer zavœgl za:br lamu:r nivrɔñ nil
"business" "blind person" "tree" "love" "drunkard" "island"
( < les affaires ) (< les aveugles "the blind ones") (< les arbres "the trees") (< l' amour "(the) love") (< un ivrogne "a drunkard") (< une île "an island")
324
Peter Bakker and Robert A. Papen lɔm/nɔm/zɔm/tɔm "a man" (< l'homme "the man", etc.) lur/zur/nur "bear" (< l'ours "the bear" etc.) b.
TRUNCATION:
spjũ
"spy" "impolite" la lεksjũ "election" la sjεt; en sjεt "plate" la pɨtal, en pɨtal "hospital" früti
(espion) (effronté) (élection) (l'assiette) (l'hôpital)
The gender assigned to the truncated nouns in (39)b seems to be identical to the gender of the original French noun. The gender assigned to the noun in Michif, however, is not necessarily that of European French or even of Canadian French where the two varieties differ. Some gender differences are illustrated in (40) and (41): (40) Masculine in Michif, feminine in French: byt "butte, hill" muš "fly" ševr, šεv "goat" sum "whitewash" grif "claw" (41) Feminine in Michif, masculine in French: batũ "stick" mask "mask" bol "bowl" 4.4.2. Cree nouns There are only a few Cree nouns in Michif. Most of them denote berries, plants, animals, kinship terms, and common household objects. With few exceptions, they behave as if they were French nouns, although they retain a Cree phonological shape. For instance, they are pluralized by French articles instead of Cree suffixes. Only historical Cree borrowings into Canadian French (e.g. lepišu "bobcat" [Canadian French: "woolen slippers"], mocassin "moccasin", and mitasse "leggings") have French phonology. The Cree nouns musum "grandfather" and kuhkum "grandmother" (lit. "your grandmother") are used by Métis and Cree alike, even when speaking English or French. All Michif speakers make ample use of Cree nominalization possibilities. In Cree any verb can be nominalized by means of the nominalizing suffixes -kan ("instrument"), -win (neutral nominalizer), and -šk/-hk "repetitive V-er" or (-hk only) "action by unspecified actor(s)".
Michif: A Mixed Language based on Cree and French (42) lɨ pakamahı-kan
325
"the striker"
(. pakamahw-e:-w "s/he beats him") (43) lɨ "the toy" (. me:taw-e:-w "s/he is play ing") (44) lɨ we:pm-ıke:-wın "the garbage" (C.we:pm-ıke:-w "s/he throws things") (45) en pu:ju:-šk "a (fem.) quitter" (. pu:ju:-w "s/he quits") (46) æ dutama:ke:-hk "a panhandling" (. nıtutama:-ke:-w (AI) "s/he asks people for things") Cree nouns in -wın are always inanimate. All our examples have French masculine gender. Cree relative clauses can also be used as Michif noun phrases. Conjunct verb forms are in themselves relative clauses; these can take French determin ers. (47) (æ) ka:-ata:we:-t IAMsg C0MP-buy-3sg
"one who buys, a buyer"
(48) (æ) ka:-pe:hkısčı-ke:-t "one who cleans, a custodian" IAMsg C0MP-clean-DETRANS-3sg All in all, Cree nouns remain a marginal part of most of the Michif dialects. 4.4.3. English nouns As noted above, English influence on Michif is now pervasive. In fact, any English noun can be used in Michif, and it is hard to distinguish between code mixing and borrowing. English nouns usually have French (masculine or feminine) determiners. They agree in animacy/inanimacy with Cree demonstratives and verbs (compare Hogman 1981 for agreement in a Cree dialect with heavy French borrowing). Stress is often assigned according to Cree stress assignment rules (see §4.2.2.). French gender assignment is not well understood; it seems to be more or less arbitrary. It is often the same as in the equivalent French nouns (49), but sometimes it isn't (50). (49) la light lɨ base
"(electrical) light" (F. la lumière, une ampoule) (F. le but, le coussin)
(50) la fun æ binder
(F. le plaisir) (F. une moissonneuse)
326
Peter Bakker and Robert A. Papen
Plurality is only marked in the French article, not in the English noun: li stepsister "the stepsisters", li drug "drugs", li bεŋ "bengs" (traditional Métis food, from Canadian French beigne < French beignet). It should also be mentioned here that some French words that are plural in English also take a plural French article: li policik "politics" (French la politique). li polis "po lice" (also la polis; French la police). Lilla polis, even with singular article, always triggers plural agreement on the verb. 4.4.4. Noun modifiers Noun modifiers in Michif include articles, possessive elements, demonstra tives, quantifiers, relative clauses, adjectives, and adverbs. We will discuss all of these here. The French definite and indefinite article forms are used in Michif, in both genders. Only the plural indefinite article des is rare or nonexistent; the etymologically definite form li is used instead. The liaison rules of French do not exist in Michif (see §4.2.1). The forms of the article in the Turtle Mountain dialect are given in Table 3. Michif uses the complete French possessive paradigm (see Table 4), but see also §4.2.1 for discussion of the prevocalic forms and liaison. Cree nouns mostly have Cree possessive prefixes and suffixes (see Table 4). There are a few exceptions, such as musum "grandfather" and kuhkum "grandmother", which can have French possessive determiners, and probably also nouns formed by productive nominalizing (mentioned in §4.4.2). Both Cree and Michif (the Cree component as well as the French compo nent) distinguish between alienable and inalienable possession. Cree inalien ably-possessed nouns (mostly kinship and body parts) have an obligatory possession marker; nonpossession is marked by a prefix mi-, which cannot be used with other nouns. In French words, inalienably-possessed nouns always have a possessive element. (See also §4.6.7 below for possessive/genitive syntax.)
Table 3. French articles in Michif. INDEF.Sg MASC:
FEM:
INDEF.pl
DEF.Sg
DEF.pl
(F. un)
li (rarely di) (F. des)
li (F. le)
li (F. les)
en (F. une)
li (rarely di) (F. des)
la (F. la)
li (F. les)
Michif: A Mixed Language based on Cree and French
327
Table 4. French and Cree possessive elements in Michif. FRENCH POSSESSIVES: SINGULAR
masc. 1sg.: mu/mũ 2sg.: tu/tu 3sg.: su/sũ 1pl: nɔt/nut [2pl.: vɔt/vut 3pl: lœ
PLURAL
pre-V mũn tũn sũn nɔt/nut vɔt/vut lœr
fem. ma ta sa nɔt/nut vɔt/vut lœ
pre -C mi ti si nu vu lœ: /ly
pre-V miz tiz siz nuz vuz] 12 lœz/lyz
MICHIF AND CREE POSSESSIVES, with musum "grandfather" as stem:
lsg.:
nı-mušum
2sg.: 3sg.:
kı-mušum u-mušum-a
(51)
CREE:
lpl.excl.: nı-mušum-ına:n lpl.incl.: kı-mušum-ına:n (Cree: -maw) 2pl.: kı-mušum-ıwa:w 3pl.: u-mušum-ıwa:w-a
ALIENABLE POSSESSION:
nı-paskısıkan paskısıkan * mı-paskısıkan
"my gun" "a/the/someone's gun" "someone's gun"
INALIENABLE POSSESSION:
nı-ste:s * ste:s ? mı-ste:s
"my brother" "brother" "someone's brother".
FRENCH INALIENABLE POSSESSION:
mũ frεr * frεr
"my brother" "a brother"
Demonstratives in Michif are virtually always Cree. This produces a number of interesting phenomena, since Cree distinguishes animate and inanimate gender for nouns and demonstratives must agree in gender and number with the noun. Moreover, when a Cree demonstrative is used in Michif, it is accompanied by a form of the French definite article or a possessive. This article must agree in French (masculine/feminine) gender and number with the noun. The Cree verb also agrees in animacy/inanimacy with the (French or Cree) subject and/or the object. This means that nominal
328
Peter Bakker and Robert A. Papen
Table 5. Cree demonstratives in Michif.
ANIM.Sg. ANIM.pl. INAN.Sg. INAN.pl.
PROXIMAL
INTERMEDIATE
DISTANT
awa ũ:kık/u:kık u:ma ũ:hi:/ũ:hi
ana anĩkĩ/anıkık/enekık anıma anıhı/anĩhĩ/enehĩ
naha ne:kık ne:ma ne:hĩ, nεhi
forms must be doubly marked for two types of gender in Michif speakers' mental lexicons; Cree gender must be marked in the Cree verb and demon strative, and French gender must be marked in the French article. Some typical examples are given in (52): (52) awa lɨ garsũ awa la fij u:ma lɨ papji u:ma la bwεt anima la m zũ ne:ma lɨ šã ũ:hĩ li zafεr
"this boy" "this girl" "this paper" "this box" "that house" "that field" "those businesses"
Some speakers also use French demonstratives, but always marginally and in a limited number of contexts — specifically, only in temporal expressions. They do not seem to be very productive. The form of the French demonstra tive varies: [ɨsti, stɨ, st, ɨs, s, ɨst] (from popular French c'te). These may be fossilized forms in many cases. Some examples: (53) la sãdr ki:-mišı-pahkıstı-n is præta. the ash PAST-much-fall-II.3 DEM spring "We had an ash fallout this spring." (54) wi:-wi:wı-w st-atɔn. voL-marry-AI.3 DEM-fall "She'll get married this fall." Alternatively, the French demonstrative may be used to express the notion "such an X": (55) nu čı-ki:-tu:tam-an st-afεr. NEG COMP-PAST-do.TI-TI.2—4Sg DEM-thing "You shouldn't do such a thing."
Michif: A Mixed Language based on Cree and French
329
Quantifiers may be either French or Cree. Most quantifiers, whether French or Cree, require the presence of the French article between the quantifier and the noun, as in (56)-(60): (56)
li žεn d d ten plA young turkey "ten young turkeys"
(57) plys kɨ hen lɨ sɔkl more than one DAMsg plowshare "more than one plowshare" (58) æ pči bræ la sup IAMsg little bit DAFsg soup "a bit of soup" (59) misce:t lɨ mãži many DAMsg food "a lot of food" (60) j k pe:jak norãž dawe:jım-a:w. only one IAMsg orange 1.want-TA.1→3' "I want only one orange." Algonquian languages in general allow the preposing of certain NP elements, often quantifiers, thus creating discontinuous constituents (Dahlstrom 1987). This is also possible in Michif, as illustrated in (61): (61) kat gi:-aja:w-ak li mud la prɔmji ninety l.PAST-have-3pl DAMsg people DAFsg first žurni dɨ lã. day of the.year "I had 90 people over on New Year's day." Most relative clauses modifying nominais are Cree; these are typically conjunct clauses introduced by the complementizer prefix ka:-. (62) bakwat-a:w-ak li mũd ka:-kımutı-čik. l.hate-TA-3pl DAMsg people coMP-steal-3pl "I hate people who steal."
330
Peter Bakker and Robert A. Papen (63) ke:kwaj ka:-itwe:-ja:hk something -say-AI. 1EXCL "something that we said"
A few occurrences of French relative forms, possibly fossilized, have also been noted, with the shapes kɨ, kɨla, kɨlɨ; these are almost always followed by French predicates (adjectives, nouns). (64) æ ku en parson kili vjœ.... DAMsg blow IAFsg person REL.s/he.is old "Once a person is old..." (65) dilu kili bũ water REL.s/he.is good "fresh water" (66) awıjak kili drul someone REL.s/he.is funny "someone funny" There is no separate class of adjectives in Cree. All Michif adjectives are therefore of French origin. In Cree, noun-modifying elements are either noun prefixes or verbal constructions. These verbal constructions can have the form of relative clauses, which can be roughly equivalent in meaning to adjectives; as mentioned in §4.4.2, Cree relative clauses are common in Michif. (67) dilu ka:-wi:hkaši-hk water coMP-taste.good-II.3 "fresh water" (68) æ morsu:-d-vjäd IAMsg piece-of-meat "a big piece of meat"
ka:-mıša:-k C0MP-big-IL3
It also happens (fairly rarely) that Cree "adjective-like" elements are prefixed to French nouns: (69)
napakı-lɨ-pwεsũ flat-DAMsg-fish "flatfish" (Cree napakaw "it is flat"; compare napakı-tâpânask "a flat sled")
Michif: A Mixed Language based on Cree and French
331
French adjectives are typically either pre- or postposed according to the French pattern. When they precede the noun, they agree in gender and number with the noun they modify, as in French (70)-(71), but when they follow the noun they don't (unlike in French), as in (72)-(74): (70)
gru sarpã IAMsg big.MASC snake.MASC "a big snake"
(71) la grus DAFsg big.FEM "the big tent"
tat tent.FEM
(72) la m zũ blã DAFsg house.FEM white.MASC (French: la maison blanche [blãš]) "the white house" (73) en fam IAFsg wife.FEM "a new wife" (74)
nœ new.MASC
(French: une femme neuve?)
lɨ bwa: šεš DAMsg wood.MASC dry.FEM (French: le bois sec) "the dry wood"
French adjectives may be reinforced with vre (< F. vrai(e) "real") or by reduplication. Both these reinforcements are also known in Canadian French dialects, and in Michif: (75) en vre žali rob IAFsg real nice dress "a real(ly) pretty dress" (76) li vjœ-vjœ kı-musum-ma:n DAMsg old-old 2-grandfather-21 "our (incl) very old grandfather" Comparatives and superlatives can be French or Cree or mixtures of both systems. (77) li ste:k s-i la ply mijœr the steak it-is DAFsg most best "Steak is the best meat."
vjãd. meat
(French: la meilleure viande)
332
Peter Bakker and Robert A. Papen (78)
tũ nawat mıšıkıtı-w DAMsg horsefly more big-AI.3 "A horsefly is bigger than a fly."
ašpıcı εn muš. compared.to DAFsg fly
In Cree there are uninfected adverbs, e.g. ki:htwa:m "again" and mıtunı "much". These are used in Michif too, along with French adverbs, such as a pi pre "approximately", di kučym "usually", and surtu "especially". Things that are expressed by verb-modifying adverbs in other languages can be expressed in Cree by preverbs. In Michif this is also common: (79)
mıyu-atuške:-w. well-work-AI.3 "He works well."
(80)
li bɔn mark a likɔl mıšı-mıjustam-ıh-ıku-w. plA good mark at the.school big-be.glad-CAUS-INV-3 "Good marks in school make him very happy."
In French, manner adverbs can be formed from adjectives by addition of the suffix -ment, which is productive in French but does not exist in Michif. Michif often uses the Cree preverbs, but adverbs are also apparently derived from French adjectives by the addition of the preposition ã (< F. en) in front of the adjective: (81)
či:stastam si parɔl ã grus. pick.TI.3→4 POSS-3pl word in big "She speaks in an affected way."
(82) ä rof pi:kıškwe:-w. in rough speak-AI.3 "He talks in a rough way." 4.4.5. French derivational and compositional processes New concepts seem to be expressed in Michif by circumlocution rather than composition or derivation. French compositional rules are severely reduced in Michif, and only a few French derivational suffixes are used productively. These are added to French stems and sometimes to English stems; most of them are nominalizing suffixes that have been used to create words not found in French dialects:
Michif: A Mixed Language based on Cree and French (83) a. lɨ vɔl-až "theft" (F. vol- "steal" + F. -age; cf. F. le vol "theft") b.
sykr-až "a sweet" (F. sucre "sugar" + F. -age; cf. F. bonbon)
c.
trɔst-ab "trustworthy" (Engl, trust + F. able; cf. F. fiable)
d. bεg-œr, bεg-œz "beggar" (Engl, beg + F. eur/euse; cf. F. mendiant(e)) e.
avar-isœ "miser" (F. avare + F. isseux, isseuse ; cf. F. avare)
f.
ažãsri "office" (F. agence + F. -erie; cf. F. agence)
g. kũtãtri "happiness" (F. content "happy" + F. -erie; cf. F. bonheur) h. lãtũ, lãtɔn "slowpoke" (F. lent "slow" + F. -on, -onne; cf. F. lambin{e)) i.
kũfidasjũ "confidence" (F. confid(ence) + F. -ation; cf. F. confidence)
j.
pwεsunjĩ "fisherman" (F. poisson "fish" + F. ier; cf. F. pêcheur)
k. pruvas "proof (F. prouv(er) + F. asse; cf. F. preuve) We have very few examples of composition-like contructions: (84) a. b. c.
fe-lɨ-s "a hypocrite" (lit. F. faire le saint; cf. un(e) hypocrite) fε-lɨ-grɔs "a boaster" (lit. F. faire le gros; cf. un vantard) en fε-la-bɔn "a prude" (lit. F. faire la bonne; cf. une prude)
333
334
Peter Bakker and Robert A. Papen (85) 1-aržã pa-vrε "fake money"
(Lit. F. l'argent pas-vrai)
The pattern most often used for extending the lexicon is circumlocution, creating analytic constructions rather than compositions (usually N P N, often N à N or N de N). (86) a.
en butej ã vitr DAFsg bottle in glass "a carafe" (French: une carafe)
b.
en m zu d log DAFsg house of log "a log house" (French: une cabane)
c.
pči pwesu d-or DAMsg little fish of-gold "a goldfish" (French: poisson rouge)
4.5. Prepositons and postpositions French has prepositions; Cree has postpositions, prepositions and a locative nominal affix -ıhk/-ohk. Michif uses both the French and the Cree forms, but again there is considerable dialectal variation. Some speakers (e.g. many of those who live in Turtle Mountain and southern Manitoba) use French prepositions almost exclusively, others (e.g. those of Camperville, Manitoba) mostly use Cree adpositions, and still others use both systems. Michif preposi tions of French origin often derive from the related French adverbs rather than prepositions — as also happens sometimes in Canadian French (and in older stages of European French). For instance, instead of the French preposition derrière "behind" Michif uses anarjcr "behind", derived from the French adverb en arrière "in the back" (87b). Some French prepositions have been reanalyzed to express typically Cree functions. For example, the Michif preposition da/dã/sa (< F. dans "in") 13 no longer means uniquely "in(side)", as it does in French. Rather, it has taken on the Cree locative meaning: (87) a.
dã lɨ uhčı in the fridge from "out of the fridge" (Cf. Cree tahkascikan-ihk ohci, lit. "fridgeLoc from")
Michif: A Mixed Language based on Cree and French
335
b. ãnarjεr dã lɨ magaz behind in the store "behind the store" (not "in the back of the store") c.
utahk da lɨ ša:r "behind the car" (not "in the back of the car") (Cf. Cree otahk ota:pa:na:sk-ohk, lit. "behind car-LOC")
(88) a:tıht manıša li brãš dã lɨ za:br uhčı. some cut.off.IMP.TI2 plA branch LOC DAMsg tree from "Cut off some of the branches from the tree." In Michif the French-derived preposition da seems to have absorbed the function of the Cree locative suffix, as both are used in combination with certain prepositions and postpositions. Some prepositions merge with articles in French, but they merge differ ently in Michif: (89)
de + le → de + les → à + le → à + les → dans + un → d+un →
FRENCH
MICHIF
du dɨ des au aux dans un d'un
(dy in some fossilized expressions) li (di in some fossilized expressions) - ( in some fossilized expressions) dæ dæ
The preposition di (< F. de) is not always used in Michif as it is in French. The French semantic distinction between partitive and nonpartitive forms is lost in Michif; in addition, it seems to be possible to omit di in a number of cases: (90) la šeyer ili pl li sab. DAFsg pail it.is full the sand "The pail is full of sand." (French: pleine de sable) (91) mi suyi i-sü pl d sab. my.pl shoes are-3pl full of sand "My shoes are full of sand." (92)
žyp di rob pl li pli d akorjijü IAMsg skirt of dress full plA ruck of accordion "a skirt full of accordion rucks" (French: pleine de plis)
336
Peter Bakker and Robert A. Papen
There are some fossilized partitive forms, e.g. dilu "water" (from French de Veau "water"; cf. Veau "the water"), which has indefinite and definite mean ing — as does lu "water", which also occurs in Michif. Finally, preposition stranding occurs with some prepositions in Michif, just as it does in Canadian French: (93) la mɨčɨn e:-nıpe:-h-ıwe:-hk DAFsg drug coMP-sleep-CAUS-INDFZR-IMPRS "The drug people are put to sleep with."
avεk. with
4.6. Michif syntax Cree is a nonconfigurational language with relatively free word order. Stan dard French has a rather rigid word order and a clearly articulated syntactic structure; spoken French, however, often diverges from this pattern and has much freer word order. Michif syntax is basically that of Cree — that is, its word order is relatively free. Of course, the more French elements are used, the closer the syntax conforms to spoken or written French norms. Some examples of different sentential word orders are given in (94)-(99). (94) SVO: la žyma: ki:-aja:w-e:w pči pul . DAFsg mare PAST-have-TA.3→3' IAMsg little foal "The mare had a foal." (95) SOV: kahkıja:w awıjak la pwi dawe:stam-w-ak. all somebody DAFsg rain want.it-TI3pl-4sg "All the people want rain." (96) VSO:
ki:-wanıst-a:w lɨ žwal su liku. PAST-lose.it-3 DAMsg horse his halter "The horse lost its halter."
(97) OVS: la kuron ka-kiskaw-e:w Prince Charles. DAFsg crown.ANIM FUT-wear-TA.3→3' Prince Charles "Prince Charles will wear the crown." (98) VOS: ki:-ucıpıt-am sa tεt la tɔrčy. PAST-pull-TI.3.4sg his head.INAN DAFsg turtle "The turtle pulled back his head."
Michif: A Mixed Language based on Cree and French (99) OSV:
337
be:bi la pr ses ki:-aja:w-e:w. IAMsg baby DAFsg princess PAST-have-.→3' "The princess had a baby."
Sentences which consist only of French elements normally follow colloquial French word order: (100) žɨ rɨspεk mu nuk. "I respect my uncle." (101)
li žard d čwɨzɨn i furni tutli žardinaž. DAMsg garden of kitchen 3SUBJ furnish all plA vegetable "The kitchen-garden furnishes all the vegetables."
If a French verb is used, all nominal subjects require a resumptive pronoun (subject reprise), which is always based on the French masculine form, seemingly always singular: (102) Delia il-a fe sãbiã. Delia he-has made seeming "Delia used to pretend." (Fr.: Delia (elle) a fait semblant) (103) žyl il-a ãfarži lɨ žwal. Jules he-has hobble(d) DAMsg horse "Jules hobbled the horse." English NP syntax sometimes intrudes in an otherwise French NP: (104)
la tεt ruž pik-dɨ-bwa il-ɨ gru. DAFsg head red woodpecker he-is big "The red-headed woodpecker is big." (F.: le pic de bois à tête rouge)
(105)
lɨ pči šavæž bibi DAMsg little Indian baby sauvages)
(F.: le petit bébé sauvage/ des
(106) æ vre bũ balãsi rɨpa IAMsg real good balanced meal "a really well balanced meal" (F.: un très bon repas équilibré or un repas très bien équilibré)
338
Peter Bakker and Robert A. Papen
4.6.1. Copula constructions There are four kinds of copula constructions in Michif: a Cree-based predica tive construction using verbal adjectives or nouns, an innovative equative construction without any verb or copula, a Cree-based identifying construc tion using the Cree demonstrative determiner, and a French-based equative clause construction using the French copula. Cree has no copula as such. Copula constructions are expressed by an NP and a verbal adjective or noun, in any order. The Cree verb can be either a stative verb or a verbalized noun with a copula-like suffix -ıwı. This is the first type of Michif copula construction. The Cree verb agrees in animacy with the noun (object or subject), in both Cree and Michif. Michif examples: (107) papıkwa:-w lɨ 1 ž. be.rough-AI.3 DAMsg cloth "The cloth is rough." (108) la bwet misa:-w. DAFsg box be.big-II.3 'The box is big." (109)
lɨ za:br mıšıkıtı-w. DAMsg tree be.big-AI.3 "The tree is big."
The -IWI suffix does not seem to be used with Cree stems in Michif, strangely enough, although it is quite common in Cree. But in Michif it is used productively only with French nouns and adjectives, always preceded by the French definite article, as in (110): (110) la pusjer-Av-an lɨ pɔrtmãtũ. DAFsg dust-BE-. DAMsg suitcase "The suitcase is dusty." In the second type, which is relatively rare, the noun and its predicate (adjective or noun) may simply be juxtaposed in any order, producing an equative sentence: (111) en krem la žymã. IAFsg cream DAFsg mare "The mare is a buckskin."
Michif: A Mixed Language based on Cree and French
339
The third type is an identifying construction, in which a Cree demonstra tive is used along with a (usually French) noun as a subject. The predicate always precedes the demonstrative. (112)
za:br IAMsg
tree
ana. DEM.ANIM.sg
"That is a tree." (113) ma tãt my
aunt
ana.14 DEM.ANIM.Sg
"That is my aunt." (114)
n-arabjæ ana lɨ IAMsg Arabian DEM.ANIM.Sg DAMsg "That horse is an Arabian."
žwal. horse
Rhodes (1977:13) found an interesting variant of this construction. When the predicate noun is possessed (alienably), the possessor functions as the predi cate, and the logical predicate noun is displaced to the right. This is an emphatic construction. (115) nıja u:ma mu papji. 1sg DEM.INAN.sg my-MASC paper "This is my paper." The final type of copula construction involves a French copula, derived from the inflected forms of the French verb être, with cliticized pronouns (a list of these forms can be found in §4.3.2). Both predicate-subject and subjectpredicate orders occur: (116) e:waku la rob ili kwarεk. that DAFsg dress COP correct "That dress is appropriate." (117)
sɨ žali larãžmã d flœr. COP nice DAFsg.arrangement of flower "The flower arrangement is nice."
4.6.2. Interrogatives In this section we will discuss polar questions (yes-no questions), WH-questions, and embedded questions.
340
Peter Bakker and Robert A. Papen
Polar questions in Michif differ from declarative sentences in having a question element čĩ or či (Crée ci),15 either after the first major constituent or at the end of a sentence. (118)
la munisjü: čĩ kıt-a:ja:-na:n šı-papa:-ma:či:-jahk ? DAFsg ammunition Q 2-have-21 coMP-about-hunt-AI.21 "Do we (incl.) have ammunition for hunting?"
(119) kı:-kı:muci-natustaw-a:w-ak ti: zami: čĩ? 2.PAST-secretly-listen.TA-3-pl your.pl friend Q "Did you eavesdrop on your friends?" There are also cases where we don't find the question element: (120) ki:-kaškıst-e:n kekwaj ka:-wi:-ušt-a:-jan? 2.PAST-be.able-4sg what coMP-voL-do-DIR-2sg "Were you able to do what you wanted to do?" Interrogative pronouns and interrogative adjectives are fundamentally Cree. The latter agree in gender (animate/inanimate) and in number with the noun they modify. The most important question words are listed in (121)-(122) and some examples are given in (123)-(124). (121) awi:na, awe:na ki:kwe:, ke.'kwaj ta:nde ta:ne:hkı, ta:n hkı ta:nšı, ta:šı, ta:šıšı ta:nšpi:
"who" "what" "where" "why" "how" "when"
(122) "which": ta:na ta:nima ta:nıhı
(anim. sg.) (inan. sg.) (inan. pl. /anim. obv.)
(123) tama lɔm ka:-naskwe:ht-ahk? which man C0MP-answer-TI.3 "Which man answered?" (124)
ta:nıhı li livr ka:-utınam-a:n? which.pl plA book coMP-take-TI.l→4 "Which books shall I take?"
Michif: A Mixed Language based on Cree and French
341
A few French question words are also used, but they are rare: (125)
kɨ laž kɨ-t-a? which age that-you-have "How old are you?"
(126) kubjæ di fwe ka:-wa:pastam-an la niž? how.many of time coMP-see.it-2 DAFsg snow "How many times did you see the snow?" (i.e. "how old are you?") (127)
kεl sort di flœr anima? which kind of flower that "What kind of flower is that?"
Embedded polar questions sometimes use the Cree form ki:spm "if, although in Cree itself embedded questions are given in the form of direct questions. The use of these embedded questions in Michif seems to be due to French influence (although Drapeau, personal communication 1993, says these con structions also exist in the closely related Algonquian language Montagnais). (128) gwe:čım-ık ki:spm čı-wı:čıhıh-ıwe:-ja:n. 1 .ask-TA.INV.3 if COMP-help.TA-INDEFZR-1 "He asked me if I would help people." In most cases embedded WH-questions are identical to nonembedded sen tences, with the exception of the use of ıšpi: instead of taspi: and ıta or ıte: instead of ta:nde: as relative question markers (see §4.6.3), the forms that would be required in Cree also. (129) nı-nıšıtut-e:n ta:nšı ıtwe:-jan. l-understand-TI.4sg how speak-AI.2 "I understand the way you speak." 4.6.3. Subordinate clauses Apart from the relative clauses mentioned in §4.4.2. and §4.4.4., there are also local/temporal relatives and adverbial clauses. First, there is a locative rela tive form ıte:/ıta "where", and there is a temporal relative pronoun ıšpi: which is sometimes combined with a French relative clause:
342
Peter Bakker and Robert A. Papen (130) ka-ıtuhte:-h-ıtı-n ıte: e:-wi:-tuhte:-jan. 2.FUT-go-CAUS-TA.INV-NON3 where COMP-VOL-go-AI.2sg "I will take you where you want to go." (131) bakwat-e:n čı-tuhte:-ja:n d-æ nikol ıta 1.hate-TI.3→4 -go-AI.1 in-IAMsg school where strik. k-ɨl-ɨ C0MP-3sg-BE strict "I don't like to go to a school where they are strict."
Most adverbial conjunctions are from French; they control either a Cree conjunct clause or a French complement clause: (132) nu pe:ht-am akuz e:-pa:kı-payın-ıyı-k su NEG hear-TI.3→4 because COMP-thick-become-PS-IL4 his zaraj. ear "S/he doesn't hear it because of an ear infection." (133) pe:hta: zyskatã či-takušımı-ja:n. wait until coMP-arrive-AI.1 "Wait until I arrive." (134) sit brav la fij apre k-ɨl-a su:vi 3.BE.IAMsg brave DAFsg girl after -3-havesave.PART be:bi. lɨ DAMsg baby "She was a heroine after she saved the baby." A very peculiar construction, using French lexical material but nonexistent in French, is used as a manner adverbial: (135) pa mwaj lœz aržã šı-akım-ım-ıht kisũ no means their money coMP-count-PO-IMPRS COMP.3.BE.pl riš. rich "They can't count their money, they're so rich." (136) ki:-tu:hka:pı-w k-ɨl-ɨtε syrpri. PAST-have.eyes.open-AI.3 -3-was surprised "S/he was open-eyed with surprise."
Michif: A Mixed Language based on Cree and French
343
(137) nu mıhče:t meri či-lɨ-bud -hke:-t k-ɨl-ɨ NEG much Mary COMP-DAMsg-pout-make-AI.3 -3-is tužur budœz. always moody "It doesn't take a lot for Mary to pout, she's so moody." 4.6.4. Coördinating elements In Michif two NPs can be connected by pi "and" (from French puis "and then") or, rarely, by avik (F. avec "with"). French et is apparently used rarely or not at all. Sometimes the Cree forms e:kwa and mi:na are used to connect NPs. French ub or ubædũ "or" (from F. ou bien/ou bien donc "or"; these are emphatic forms — the normal French form is ou) is used for "or". Sentences can also be connected by e:kwa "and" and pi, and also by ub dũ. Cree and French verbs are sometimes connected by pi: (138) pi:kupıč-ıka:t-e:-wa pi dɨsk-i. fallow-PASS.INAN-DIR-4pl and disk-INF "They (inan.) are fallowed and disked." 4.6.5. Obviation Cree obviation marking is reduced in Michif. Obviation in Cree, which is marked in the verb as well as in the noun, distinguishes two (or more) third persons within a phrase, a sentence, or a stretch of discourse. The third person not mentioned previously is marked "obviative" (-a after consonants, -wa after vowels), while the known third person remains unmarked and is called "proximate". Obviation is not marked on inanimate nouns in Plains Cree, and number is not marked on obviative nouns. Animate third persons possessed by third persons obligatorily have the obviative suffix in Cree, but in Michif only Cree nouns have obviative markers, not French nouns: (139)
u-ma:ma:-wa his/her-mother-OBV "his/her mother"
(140) *su š -wa his/her dog-OBV "his/her dog"
344
Peter Bakker and Robert A. Papen
When two third persons do not have a relation of possession, but are part of the same discourse, the newly mentioned noun phrase receives obviation marking as "Irene" does in (141): (141) John ki:-wa:pam-e:w Irene-a. John PAST-see-→3' Irene-OBV "John saw Irene." This use of what we could call syntactic obviation is reduced by comparison to Plains Cree, where it is obligatorily used for animates (and never for inanimates). In Michif, obviation is always marked on personal names, often on other nouns referring to humans, sometimes on nouns referring to animals, and never on inanimate entities. (142)
nɔm dã lɨ bãd dɨ mud IAMsg man in DAMsg crowd of people ki:-nipa-h-e:w Bobby Kennedy-wa. PAST-die-CAUS-→3' B.K.-OBV "A man in the crowd killed Bobby Kennedy."
(143) Pitœrli bũ y-wa ki:-pimičišahw-e:w. Peter DAMsg good god-OBV PAST-follow-TA.3-^3' "Peter was a disciple (lit. "followed the Lord")." It must be stressed, though, that even when French forms are not marked for obviation, an obviative agreement marker is nevertheless usually present on the Cree verb, as in (144) (and see 159 for an examples without -ji "obviative"). (144) bæcis su frer nipa:-ji-w. John his brother sleep-OBV-3sg "John's brother is sleeping." If a third person possessed noun is the object of a verb, it has an agreement marker -ım- on the verb (see ex. 135 above). 4.6.6. Negation Both French and Cree negation occur in Michif. Cree has different negators for main clauses (namo:ja and variants), for embedded clauses (e:ka:wija, mostly shortened to e:ka:), and for imperative clauses (ka:ja). Namo:ja is also used to answer polar questions.
Michif: A Mixed Language based on Cree and French
345
Standard French has two main negators: non is used as a negative answer to polar questions and ne...pas is used elsewhere, the two elements separated by the verb and its clitics. Other French negative adverbs are ne...rien "nothing", ne...personne "no one", ne...plus "no longer", etc. In spoken French ne is usually omitted. The word pas is also used to negate nouns, which are given in the partitive form with de "of, e.g. pas de pommes "no (of) apples". By far the most common negative particle in Michif is nu (most likely from French non). This particle is typically used with nonimperative Cree verbs. It usually occurs immediately before the verb in independent or main clauses, and it immediately precedes the complementizer in negative embed ded clauses. In both cases Cree would use e:ka. If there is an adverb, nu immediately precedes it, in an innovative use of the French-origin negative element. (145) nu
ka-pakıtm-ıka:šu-w
NEG FUT-allow-PASS.ANIM-3
čı-ıtuhte:-t. COMP-go-3
"She won't be allowed to go." (146) nu kı-tıpe:jım-a:w lɨ minuš. NEG 2-own-TA.l/2→3 DAMsg cat "The cat doesn't belong to you." (147) li nasjŭ æ treti aja:-w-ak nu plA nation IAMsg treaty haveTI-3-pl NEG šı-nu:tm-itu-čık. COMP-fight-RECIP-3pl
"Nations have treaties not to fight each other." (148) nu tultã aja:-w a lɨkɔl. NEG always be-AI.3 in the.school "S/he is not always at school." In Michif, nouns are usually negated by the French particle pat (< F. pas de), and pa is used to negate adjectives. French verbs are sometimes negated with pa (never accompanied by the element ne, just as in colloquial French), but these may all be fixed expressions. Here, negation follows French syntax, preceding the noun and adjective but following the verb:
346
Peter Bakker and Robert A. Papen (149) lɔm pat barb ana. man NEG beard DEM.ANIM.sg "That's the man without a beard." (150) pat wicıhıwe:wın. NEG help "There is no help." (151) žɨ pãs
pa
šı-ıtuhte:-t.
I think NEG COMP-go.AI-3
"I don't think he will go." The most common Cree negator in Michif is namu(:), which precedes the verb, the noun, or the adverb it negates. It occurs in Michif in the same position as nu, with one exception: nu is not found in embedded sentences. (152) namu NEG
nı-wi:-ituhta:-n. l-VOL-go.AI-NON3
"I won't go/I don't want to go." (153) mısče:t laržã aja:w-e:w ma:ka namu: much the.money have.TA-3→3' but NEG fatas. braggart "He has a lot of money but he is no braggart." (154) namu: wa:hıjaw d-aja:-na:n a Bismarck NEG far 1-BE-AI.lpl to Bismarck "We are not far from Bismarck."
æ IAMsg
unčı. from
The Cree negator e:ka: is also used in Michif. As in Cree, it always modifies embedded clauses and precedes the conjunct verb. In Michif these can be Cree-based or French-based: (155) ahkıkı-n la mus ita lɨ salej e:ka: grow.II-4sg DAFsg moss where DAMsg sun NEG e:-wa:še:n-ike:-t. COMP-shine.AI-DETRANS-3
"Moss grows where the sun doesn't shine." As in Cree, imperative and subjunctive verbs (and rarely conjunct verbs) are negated by the Cree form ka:ja.
Michif: A Mixed Language based on Cree and French
341
(156) ka:ja napač-im-in. NEG bother-with words-IMP.TA.2→l "Don't hassle me." It should also be mentioned that some Cree negative + noun/ adverb construc tions are used in Michif too. In Cree (and Michif) these form a phonological unit, e.g. Cree namo:ja na:nita:w "nowhere, nothing", Michif nama na:nda:w or nu: na:nda:w; Cree namoja ki:kwe:, often nama: ki:kwe: or ma:ki:kwe: "nothing, not at all", Michif namake:kwaj or no ke:kwaj; and Cree namawijak "nobody", Michif nu awijak. These combinations can function as negators too. They seem to have scope only over nouns, not over verbs or sentences. (157) namake:kwaj 1 syk ašte:-w. nothing DAMsg sugar stand-3 "There is no sugar." 4.6.7. Possessive constructions We noted in §4.4 that the Michif noun phrase is fundamentally French; this includes syntactic ordering. The only exception to this is in possessive and genitive constructions, where syntactic ordering is more typical of Cree than of French (although Métis French also has this construction). This is true even if French possessive determiners are used. Here is a typical example: (158)
mữ pči gan Isg.poss little boy "my son's pony"
pči žwal P0ss.3sg little horse
The equivalent French phrase would be le petit cheval de mon petit garçon/ fils. Cree would have ni-kosis o-mistatim-wa (lit. my-son his-horse-OBv). Note that in Cree the possessed noun requires the obviative marker if it is possessed by a third person. In Michif the possessed noun will be marked for obviation only if it happens to be a Cree noun, as in (159): (159) Joe u-ma:ma:-wa ki:-api-w dā la šãbr Joe 3sg.POSS-mother-OBV PAST-sit-3 in DAFsg room ka:-pe:ta:-hk. COMP-listen?-3lNDEF.ACTOR
"Joe's mother was sitting in the parlour."
Peter Bakker and Robert A. Papen
348 5.
Origin of Michif
We turn now to the origin of Michif, discussing first the lexical, phonological, semantic, morphological, and syntactic components and then the historical context in which Michif came into being. 5.1. Lexicon As is clear from the discussion above, the Michif lexicon comes from two main sources: French and Cree. English is now also a major influence. Speakers clearly know which elements originate from which language. It seems that French verbs can (in principle) always be used by speakers who know (some) French and who suppose the listener understands it. The ad dressee plays a major role in the lexical choice of the speaker. People who do not know French also use French verbs, some of them integrated into Cree verb morphology (§4.3.2). The same is true for Cree nouns: the less Cree a speaker knows, the smaller the chance of Cree borrowings in the nominal component. One could argue that these nouns behave as if they were borrow ings into French. The approximate distribution of French and Cree elements in Michif is given in Table 6. The different ranges within some categories refer to ex tremes of 17 individual speakers from five different communities; see Bakker 1996 for details. Table 6. Distribution of French and Cree elements in Michif. The figures are percent ages, drawn from Bakker's corpus of questionnaire data based on 17 Michif speakers in five communities
Nouns: Verbs: Question words: Personal pronouns: Adverbials: Postpositions: Coörd, conj.: Prepositions: Numerals: Demonstratives: Negators:
FRENCH
CREE
OTHER
83-94 <0-4 <1 <1 30 0 40 70-100 99 <1 70
1 88-99 100 100 70 100 55 0-30 1 >99 30
English, Ojibwe Mixed Cree-French
— — — — English
— — — —
Michif: A Mixed Language based on Cree and French
349
As we have seen, lexical categories with verbal properties tend to be Cree and lexical categories with nominal properties tend to be French. The demonstra tives are an exception: these are always Cree, except for some fossilized French demonstratives in fixed expressions. Elements with adjectival or adverbial meaning and complementizers can be Cree or French. In some dialects they coexist; other dialects tend to use more French or more Cree. As mentioned above, all dialects of Michif contain numerous borrowings from English and, to a lesser extent, from other languages. All Michif dialects contain a number of Ojibwe elements, e.g. ušpe:hike:- "to write" (Plains Cree masmahike:; Ojibwe ošipi:ke:) and nawapu: "lunch" (Ojibwe nawappo "take provisions along").16 The future conjunct complementizer či-/ši-, which in Plains Cree has the form ta-lkita-, is identical to the Ojibwe element či-, although Moose Cree dialects have ci- (James 1983) as a shorter variant of kici-. In Turtle Mountain it is said that the language also contains Assiniboine and Gaelic elements, but thus far only pizcn "gopher" has been identified as an Assiniboine loanword (Jan van Eijk, personal communication, 1993). Despite these influences, the lexicon is basically Cree and French, with some important more recent influence from English and minor influence from Ojibwe and other languages. 5.2. Phonology Michif clearly has two separate phonological systems, with phonemes, phonotactics, and phonological rules existing and functioning only in one part and not in the other, including different solutions for things like hiatus and different allophones for similar phonemes in the two components. There is also, however, a significant degree of merger of the two systems: stress assignment in the French part is the same as in Cree; French differences in vowel quality are starting to be reanalyzed as length differences, as in Cree; Cree has borrowed some phonemes from Ojibwe (or French) — albeit some what marginally — such as the nasal vowels and /y/. The loss of nasalization in some French words may also be a result of Cree influence. (See §4.2.3 for discussion of these features.) Rhodes (1985), Evans (1982), and Orser (1984) all agree that there are separate phonological systems for the two parts of the language. It is not clear to what extent the various morphophonological rules of Cree (as dealt with in Wolfart 1973:79-82) are present in the Cree part of
350
Peter Bakker and Robert A. Papen
Michif. Certainly a number of French morphonological rules (e.g. liaison) are lost in Michif, at least partly. Certain phonological developments in each main component of Michif seem at first glance to be due to influence from the other component (such as stress assignment in the French part). But some may be attributed to influence from other languages, notably Ojibwe (such as /s/ and Id palatalization and the presence of nasal vowels), and others are probably due to internal developments within Michif (such as the rule nit→ d; see §4.2.2). The palatalization of /g/ before front vowels is also known in both Québecois and Acadian French. The labialization of velar stops before rounded vowels in French words may be an independent internal develop ment, although similar processes play a role in eastern dialects of Cree. A final rather mysterious phonological aspect of the language is the presence of an opposition /e/ vs. /i/ in the Cree component, but not in the French compo nent. Northern Plains Cree and other Cree dialects spoken in the woodlands lack Id vs. /i/, but these were probably not the dialects with which most of the Métis were in contact. This feature may be a Michif innovation. 5.3. Semantics A number of obligatory semantic distinctions made in French noun phrases are lost in the French component of Michif. Michif noun phrases do not distinguish mass nouns from count nouns (cf. the French partitive expressions with de(s)). In Michif, definite forms of the noun are used instead of the partitive. Similarly, there is no distinction between definite and indefinite plurals in Michif (cf. French les vs. des); both are expressed with li. The French distinction between masculine and feminine nouns is preserved in Michif, as we can see in the French articles, but it is lost in pronouns and verb clitics, where only the masculine form is used, and it is followed only in prenominal adjectives, not in postnominal ones (where the form for either of the genders is used for both masculine and feminine). Obviation is marked on the French noun in a more limited number of contexts than in Cree. Despite the virtual absence of Cree nouns, Cree animate/inanimate dis tinctions are preserved for nouns, as can be seen in the agreement markers of Cree verbs and Cree demonstratives. Some influence from English is evident in the fact that singular nouns referring to semantically plural entities have a plural agreement marker on the Cree verb.
Michif: A Mixed Language based on Cree and French
351
French verb forms in Michif are severely reduced by comparison to standard French or Métis French. The mood system, aspect distinctions, derivation, voice/valency, etc., are lost, due to the disappearance of all nonpresent synthetic verb forms. Only periphrastic past and future tense are (very occasionally) used. Only one stem is used for all verb forms. As for Cree semantic distinctions, these seem to be fairly well preserved in Michif. The marking of tense/mood, stem formation, derivation, voice/ valency, and other categories are all intact. The loss of the dubitative and preterite in Michif seems to parallel developments in Plains Cree. It is true, however, that almost no speakers make full use of all verb forms. Condition als, passive-like constructions, delayed imperatives, synthetic reciprocals/ reflexives, and obviative/possessive agreement markers are among the first categories to be lost. This occurs in Plains Cree too and is perhaps attributable to language death rather than to ordinary language change. It is probable that there is some influence from Ojibwe in some voice/ valency changing possibilities (-ika-, -ikate:-, -iwe:-). In a number of instances French words have taken over functions of Cree elements, such as the French preposition da functioning as a general locative marker like the Cree locative suffix. Another example is the (more or less) obligatory possessive marker for inalienable possession, and the syntax of the French negative element nu. A number of Michif words of French origin have a meaning which differs from that of French dialects, e.g. French taureau "bull" vs. Michif turu "pemmican", French biche "female deer" vs. Michif bis "elk" (in some areas), French cerise "cherry" vs. Michif sriz "[certain] berry, and French grive "thrush" vs. Michif griv "robin" (see Papen 1984b). A significant innovation in Michif is the impersonal mood system, prob ably based on a Cree model, but using French lexical material. In short, it is clear that Michif semantics is different from both source languages, but closer to Cree than to French. 5.4. Morphology and syntax NP derivational morphology is severely reduced in Michif. The NP word order is very close to French, although Michif has the order numeral-deter miner-noun and not, as in French, determiner-numeral-noun. Cree nouns are hardly ever used, but when they occur they have obviative markers according to the Cree rule.
352
Peter Bakker and Robert A. Papen
Sentential word order is based on Cree rather than on French, although word order gets closer to that of colloquial French when more French ele ments are used. The Cree and French components of Michif mostly remain separate, but in a number of instances we find elements from both languages within word boundaries. French nouns denoting animates can take Cree obviative and preterit suffixes and also some Cree adjective-like prefixes. French nouns and adjectives can be verbalized with the Cree suffixes -iwi "to be" and -ihke: "to make". In most instances these French elements are preceded by French (< F. le), and in verbs they must end with a French infinitive marker. Here too, then, we find that Cree plays a more important role than French, and that Cree morphology has been better preserved than French morphology. 5.5. Origin of the mixture It is hard to make any firm statement about the origin of Michif. The facts are, in short, as follows: (i) Métis multilingualism has always been very widespread in all phases of the history of the Métis nation. Almost all of the guides and interpreters in the Canadian West, from the late eighteenth century until the late nineteenth century, were Métis. (ii) Code switching and code mixing (the use of two languages within one phrase or sentence) has always been extensive, but when mentioned it is said to be a mixture of "Indian, French and English" (never "Indian and French"). The earliest mentions date from the early nineteenth century. (iii) The first unambiguous reference to what must have been Michif dates from the 1930s. (iv) Historical sources relating to Métis people do not mention a mixed language as (the only one) spoken by an individual or in a community. (v) The oldest (published) language materials in Michif date from around 1970. (vi) Métis communities can be (Métis) French-speaking, Cree-speaking (northern Plains Cree, southern Plains Cree, Swampy Cree, and possibly Woods Cree), Ojibwe-speaking, Michif-speaking, or a combination of these. Virtually all Métis also speak English, and some Métis also know other European languages.
Michif: A Mixed Language based on Cree and French
353
(vii) Most Michif speakers know neither French nor Cree. (viii) Wherever Michif exists, Cree and French are spoken too, or were spoken until recently. (ix) Wherever Cree and French are mixed and form a new mother tongue, the mixture is very similar: verbs tend to be Cree, nouns tend to be French (cf. Drapeau 1980, 1991). There is one thing we can be very sure about: the language must have arisen among fluent bilinguals (cf. Thomason & Kaufman 1988: 228-33). The Cree and French are too close to the original to justify any other conclusion. The amount of grammatical or lexical reduction that took place is small. The French component is more reduced than the Cree component, though, and it is better to view Michif as an intrusion of French into Cree rather than the other way around. That is, Cree is the base language of Michif. For the origin of Michif, there are some important questions to be raised. In the first place: when did Michif become a native tongue? Most of the present-day speakers know neither French nor Cree, and many grew up as monolingual Michif-speakers until school age. But when did it become a first language? It is not known when Michif became a native tongue. It is a fact that people who were born around the turn of this century say they learned the language as their first language. For that generation, then, it was undoubtedly their native tongue. In a number of cases, their grandparents taught them the language. Most of the Michif speakers that Bakker worked with had been monolinguals until school age. Only a minority spoke French, and one or two of them could also speak Cree. This would mean that Michif as the only language known by some individuals was spoken in any case around the turn of the century, but that it was already spoken alongside Cree and French before that, possibly long before that. In fact, Métis elders could name a few individuals among their ancestors as speakers of Michif who could be identi fied as born between 1825 and 1840, but they could also speak Cree and/or French. Also, some of the communities where Michif is spoken nowadays were already Métis buffalo hunters' communities in the 1840s, so that Michif would have crystallized by then. The language is remarkably homogeneous overall, and Michif speakers are descendants of the poorer buffalo hunters rather than the wealthier French-speaking Métis farmers. The second question we should ask is: do all the dialects of Michif derive from one "Proto-Michif", which was already spoken when the Métis people
354
Peter Bakker and Robert A. Papen
still formed one community, or did the different dialects develop indepen dently in the different Cree-French bilingual communities? Some of the communities where Michif is now spoken were already Métis communities in the 1840s, inhabited by bison hunters. The elite of the Métis farmers spoke French rather than Michif, whereas the bison hunters spoke Michif (cf. Giraud 1945, Bakker 1996). If we take into consideration the close similari ties among the varieties of Michif in these villages, the conclusion is almost inescapable that Michif must have jelled before 1840. The first generation of speakers may have been speakers of the lingua franca second-language Plains Cree rather than native speakers of Michif. The bison hunters were often Ojibwes and Assiniboines (and Crees) who were also fluent in Cree. This explains some of the simplifications of Cree elements as well as the presence of Ojibwe elements in all subsystems of Michif. This means that all varieties of Michif go back to the language spoken by the Métis bison hunters in the early 19th century. The admixture of Cree with French as spoken in northwestern Sas katchewan (see e.g. Hogman 1981 on Buffalo Narrows and Bakker 1996 on Ile-a-la-Crosse) is so different from Michif that it must have arisen indepen dently. In this area, some of the French is "mission" French rather than Métis French, and the Cree is northern Plains Cree rather than southern Plains Cree, as in Michif. Furthermore, the number of French nouns used is "only" a few hundred, and no French verbs at all are used. The input languages and the nature of the mixture are very different. This is also what some of the Métis elders say: there is the true Michif, the Red River language, and there is French-Cree, the latter having emerged from the contact of Cree with the French as spoken in the church and schools run by Oblate missionaries — who were (are) present in virtually all Métis communities. If these French-Cree varieties developed independently, the next ques tion that arises is: why are they so very similar? Why are the nouns borrowed from French? It must be emphasized that the French words are in no way limited to items introduced by the Europeans. They even include items that were new to the French fur traders, such as native tools, plants, and animals. Moreover, the items designated by French nouns also have equivalents in Cree. A "filling of lexical gaps" cannot be the explanation of the use of the French elements. The reason is probably to be sought in the properties of the Cree verb: it is often said that Algonquian languages are "one gigantic verb". The verb is undoubtedly the central element in Cree grammar, often providing
Michif: A Mixed Language based on Cree and French
355
a kind of summary of the complete predication. This might make it much easier for nouns to be borrowed. Nouns could function as clarifications of, or additions to, the verb. In addition, verbs are very rich in compounding and derivation and an important part of linguistic creativity in Algonquian, more so than nouns (cf. Goddard 1990). If Michif is an independent development in these different areas, it must have originated with Cree-French code-mixing. However, there are no cases of code-mixing described in the literature in which the verbs come from language A and the nouns from language B. But this may very well be due to the nature of the languages involved. Perhaps the reason for the peculiar mixture is that Algonquian languages differ typologically in certain important respects from the languages for which code-mixing has been investigated. There is some evidence that this is indeed so. Drapeau (1980) describes the influence of French on a dialect of Montagnais — an Algonquian language closely related to Cree (and some times even called East Cree) — where very extensive borrowing of French nouns, but apparently no verbs, is taking place in a French-Montagnais bilingual community. Drapeau observes (1991) that a language very similar to Michif is now developing in that community among young people. In short, it seems that cases of Algonquian-French code mixing are structurally similar to Michif, the verbs being Algonquian and the nouns French. This might be an argument in support of the view that Michif is nativized code-mixing. Another possibility is proposed in Bakker (1996). There it is stated that the sociohnguistic motivation for the development of Michif was the emer gence of a new ethnic identity, mixed Cree and French (cf. also Thomason & Kaufman 1988: 228-33). A new language was needed to express this identity. The most obvious way to form a new language was the mixing of the two community languages, Cree and French. A common way in which this could have happened would be to combine the grammar of one language (phonol ogy, morphology, syntax) and the lexicon of the other language (see Bakker & Van der Voort 1991 and Bakker 1996, Chapter 7, in which some fifteen cases of so-called "language intertwining" are discussed). This is what hap pened with nations of mixed ancestry who were in very similar sociolinguistic circumstances. Cases have been described in South Africa/Namibia (men tioned in Den Besten 1987), Indonesia (De Gruiter 1990), and elsewhere. Other examples are Media Lengua (Quechua grammar, Spanish lexicon; Muysken, this volume), Callahuaya (Muysken, this volume), and some dia-
356
Peter Bakker and Robert A. Papen
lects of Romani (Bakker & Van Der Voort 1991, Boretzky & Igla 1994). The result of the mixture of French and Cree is different because of the typological properties of French and Cree, more specifically because of the impossibility of separating affixes from stems in the Cree verb: stems can almost always be analyzed into smaller units of bound morphemes. It could be argued that the verb stem in Cree is not an unanalyzable lexical element, but a grammatical part of the language, consisting of bound morphemes (cf. Goddard 1990), traditionally called initials, medials, and finals in the Algonquian literature (see e.g. Bloomfield 1946). This may have triggered the peculiar combination of the two languages (Bakker 1996). One would thus have a Cree grammati cal system (which would include the whole verb) and a French lexicon. One would then expect Cree nominal morphology on French nouns, and this is what we indeed find, such as the obviative marker. This is argued in detail in Bakker 1996, Chapter 9.
6.
Conclusion
Michif was spoken as early as the first decades of the 19th century, although the first mention of the language dates from a century later and the first attestation is from the late 20th century. Few if any Michif speakers in the late 20th century spoke Cree beside Michif, and only one in three spoke French beside Michif. Most of them had grown up as monolingual Michif speakers. The contacts of the Métis with French- and Cree-speaking people necessi tated the preservation of these languages, beside Michif, in Métis communi ties. The reason why Michif has its peculiar and unique dichotomy of French and Cree elements may be due to the fact that Cree verbs can be said to consist of bound grammatical elements only and almost no unanalyzable lexical stem. The real dichotomy would be lexical (French) vs. grammatical (Cree), not noun' vs. verb. Another possibility (not necèssarily contradictory) is an origin in Cree-French code-mixing. It is clear that Michif is not a pidgin, not a creole, not an interlanguage, not foreigner talk, not a fossilized second language, not code-mixing or codeswitching, or anything else known from language contact studies. It is a mixed language in all areas of its grammar and lexicon. It is a system in its own right. In short, a unique language in the world.
Michif: A Mixed Language based on Cree and French
357
Appendix Below is a preliminary list of communities with native speakers of Michif. In this list only locations are given where Michif is or was a community language, not places where Michif speakers moved later. Sections A-G concern Michif; section H gives locations for French-Cree. A. SASKATCHEWAN (sources: Papen fieldwork data 1986, Bakker fieldwork data 1988, 1990, Fredeen 1991): 1) Yorkton area: no data, but possibility of several hundreds of speakers in Kamsack area. 2) Duck Lake area (Duck Lake and surrounding Indian reserves, Batoche, St. Laurent): probably extinct. 3) Domrémy, St. Louis: extinct. 4) Fort Qu'Apelle area (Fort Qu'Apelle, Lebret, Abernethy, Balcarres, Indian Head, Katepwa, and possibly elsewhere, especially on the reserves; see Fredeen 1991): between 50-200 speakers, almost all older than sixty. 5) Willowbunch: virtually extinct. 6) Assiniboine: no data, possible a few speakers. 7) Maple Creek: no data. B. 1) 2) 3)
MANITOBA (sources: Bakker fieldwork data 1990, Crawford 1983): Camperville, Duck Bay: around 50 speakers. St. Lazare, San Clara, Boggy Creek: (many?) fewer than 100 speakers. St. François Xavier, Ste Anne des Chênes, St. Charles, St. Norbert, St. Vital, Ste. Agathe, St. Eustache: no data available.17 4) The Pas area: no hard data. 5) Grand Marais: extinct. 6) Northern Manitoba: no data. C. 1) 2) 3) 4) 5) 6) 7) 8) 9)
ALBERTA (sources: Bakker fieldwork 1990, Canadian Encyclopedia): Lac La Biche: probably no speakers. Peace River area : no data. Beaver River valley: no data. St. Paul area: probably extinct. Big Prairie, Fishing Lake, Gift Lake, East Prairie, Buffalo Lake, Paddle Prairie: no data. Lac Ste. Anne: probably extinct. St. Albert: extinct. Calgary area? possibly also elsewhere in Alberta.
D. NORTHWEST TERRITORIES: no data (cf. Slobodin 1966:12-14).
358
Peter Bakker and Robert A. Papen
E. NORTH DAKOTA (sources: Crawford 1976, 1983, 1985a, 1985b, Rhodes 1977, 1982, fieldwork Papen, Bakker): 1) Turtle Mountain Reservation and surrounding area (St. John, Dunseith, Rolla): probably several hundred speakers. 2) Devil's Lake: no data. 3) scattered speakers all over North Dakota. F. MONTANA (source: Dusenberry 1985): 1) Choteau, Augusta, Dupuyer, Zortman, Lewistown: no data 2) Milk River: Havre, Chinook, Malta, Glasgow: no data. G. MINNESOTA (source: Rhodes 1982): 1) Red Lake: extinct. 2) Pembina: extinct. H. FRENCH CREE: A few thousand speakers in Northwest Saskatchewan — Ile-a-laCrosse, Beauval, Buffalo Narrows; there are also a few speakers in the surrounding communities.
Notes *
We thank the following people for help, data or comments: Paul Chartrand, Murray Hamilton, and our Michif-speaking consultants in Manitoba, Saskatchewan and North Dakota. We also thank Tim van Baar, Lynn Drapeau, Pieter Muysken, David Pentland and Maaike Verrips for useful suggestions. None of them is responsible for any shortcom ings. Bakker's fieldwork was financially supported by a Canadian Studies Graduate Student Award (1987-1988) and between 1988 and 1991 by the Linguistic Research Foundation, which is funded by the Netherlands Organization for Scientific Research, NWO.
1.
The term Michif, pronounced /mičif/, comes from the French adjective métif, métive "of mixed blood", which was used alongside the modern term métis, métisse until the nineteenth century. The latter eventually won out, but not before the former was well established in North America. In the first half of the 18th century it was pronounced /mitif/, witness the spelling mitifs in that period (see Trudel 1960:83). See §4.2 for phonological details on the particular phonetic interpretation of this word.
2.
Papen did fieldwork regularly on Métis French from 1970 onwards in Saskatchewan and Manitoba, and fieldwork on Michif in 1985 and 1986 on the Turtle Mountain Reservation in North Dakota. Bakker did fieldwork on Plains Cree, Michif, and Métis Cree in Saskatchewan, Manitoba, Alberta, and North Dakota in 1987-1988 and in 1990. See Bakker (1996) for a fuller study of Michif.
3.
Ojibwa/Ojibwe, Saulteaux, and Chippewa are the various local names for one and the same language.
4.
Michif speakers rarely know both Cree and French. See below.
Michif: A Mixed Language based on Cree and French
359
5.
The following abbreviations and symbols are used in this paper: → = subject left of arrow, object right, lsg = first person singular, lpl = first person exclusive plural, 2sg = second person singular, 21 = first person inclusive plural, 2pl = second person plural, 3sg = animate third person singular, 3pl = animate third person plural, 3' = animate obviative, 4sg = inanimate singular, 4pl = inanimate plural, ABS = absolutive (transitive object, intransitive subject), AI = animate intransitive verb, ANIM = animate, ARG = argument, BEN = benefactive, CAUS = causative, COMP = complementizer, COND = conditional (called "subjunctive" by Algonquianists), conj = conjunct order verb, DAFsg = definite article feminine singular, DAMsg = definite article masculine singular, DEAN = deanimizer, DEF = definite, DEM = demonstrative, DETRANS = detransitivizer, DIMP = delayed imperative, DIR = direct, EXCL = exclusive, FEM = feminine, FUT = future, IAMsg = indefinite article masculine singular, IAFsg = indefinite article feminine singular, II = inanimate intransitive verb, IMP = imperative order verb, IMPRS = impersonal, INAN = inanimate, INCL = inclusive, ND = independent order verb, INDEF = indefinite, INDFZR = indefinitizer (makes object indefinite), INF = infinitive, INTRANS = intransitive, INv = inverse, IRR = irrealis, LOC = locative, MASC = masculine, NA = animate noun, NEG = negation, NI = inanimate noun, NID = dependent inanimate noun, OBV = obviative, PART = participle, PASS = passive, PAST = past, pl = plural, plA = plural article, PO = possessed object, POS = positive, POSS = possessive, POSSD = possessed, POT = potential, PS = possessed subject, Q = question marker, RECIP = reciprocal, REFL = reflexive, REL = relative, sg = singular, SUBJ = subject, TA = animate transitive verb, TI = inanimate intransitive verb, VOL = volitional.
6.
It can be doubted whether the elements derived from le, la, and les still function as definite articles in Michif.
7.
This contradicts Evans (1982) and particularly Weaver (1982), who interviewed twelve informants on the Métis reservation and who stated that all were fluent in English, eleven fluent in Michif, six spoke "passable or good French", five spoke at least a little Cree, and two spoke a little Ojibwe.
8.
The French word braguette originally meant "codpiece". Now it refers to "trousers fly" in Canadian French. In Canadian French it also referred to the breech clout worn by Indians. In many French Canadian dialects /g/ is palatalized before mid front vowels and braguette is thus pronounced /brajεt/. In Métis French /brajε/ is still used to mean "breech clout" but also a "diaper" and even "Indian".
9.
An explanation of the orthography is in order here. Deciding on an orthographic system for Michif poses a number of problems. One solution would have been to use Cree and French standard orthography, but this would conceal many peculiarities of Michif. The system we use is based on the phonological systems of the Cree and French components, but as the phonetic ranges of the vowels and their allophonic values differ for the French and the Cree components, we list below the phonemic symbols and the approximate phonetic range of the phonemes with phonetic realizations that are not self-evident. VOWELS:
a: a e: i: I
(Cree) (Cree) (Cree) (Cree) (Cree)
long [a:], ranges also over [ɔ:] and [å:] short [a], ranges also over [] and [a], even [e] [e:], sometimes [e:] or [æ:] in closed syllables long [i:] short [i], sometimes sounding like short [i]
360
Peter Bakker and Robert A. Papen u: u a ε æ i
u y
(Cree) [u:], sometimes [:] (Cree) [u], sometimes [ ] and [] (French) usually pronounced [a] but ranges from [æ] to [a] (French) /e/ ranges from [e] (open syllables) to [æ] and [a] (French) [æ]; only nasal vowel; is equivalent to oral /ε/ (French) [i] in open syllables, [I] in closed syllables. In unstressed syllables both [i] and [I] occur. (French) [I] and sometimes [ ] (schwa). (French) [ɔ], also ranges over [] (French) [œ] in closed syllables and [ø] in open syllables, sometimes [y] or even close to [u] (French) ranges from [u] to [v], sometimes even [o] (French) [y], sometimes close to [Y] and [u] or [i]
Ə i I
(English) schwa ~ (English) usually pronounced [i] (English) usually pronounced [I]
ce
CONSONANTS:
hC
Cree aspirated consonants /hp/, /hk/, and /ht/ are realized as [xp], [xk], and [xt] by many speakers. Some speakers have [st] instead of [ht].
10.
This contradicts Evans (1982), who says that /ø/ is always raised to [y] in the Turtle Mountain dialect. She also posits the phoneme Id for French. In our Turtle Mountain data, the only words that contain /e/ are [elefā] "elephant", [he:rü] "heron", and [pe.ši] "sin" (< F. péché), and also some loanwords from English.
11.
This schema differs slightly from the morpheme slots in Wolfart (1973) and Dahlstrom (1986) in that we omit positions for dubitative markers, we separate tense/mood and preverbs from the stem, and we include voice/valency markers.
12.
We do not have 2pl forms in our data. Both Rhodes (1977) and Orser (1984) give second person plural forms. These seem to be rather marginal.
13.
Lynn Drapeau suggests (personal communication, 1992) an Algonquian origin for da. The distribution of dā in Michif would be exactly parallel to Montagnais nta, a preposi tion meaning "in". In Plains Cree, however, this preposition nta does not exist, to the best of our knowledge. Furthermore, the optional nasal character of the vowel of da, plus the variant form sa attested in French dialects and in Michif, make a French origin more likely.
14.
The phrase ana ma tāt would be "this my aunt", with attributive use of the demonstrative.
15.
Pentland (1982) suggests that Plains Cree ci is borrowed from the colloquial French question-element ft. Most other Cree dialects have na. But even if this is correct, the Michif question marker functions like the Cree element rather than like the French verbfinal marker.
16.
Probably nawapu: entered Michif via Métis French, since it is also typical of Métis French.
17.
These are the original Red River colony Métis settlements.
Michif: A Mixed Language based on Cree and French
361
References Bakker, Peter. 1996. "A language of our own". The genesis of Michif - the mixed CreeFrench language of the Canadian Métis. Oxford: Oxford University Press. Bakker, Peter, and Hein van der Voort. 1991. "Para-Romani languages: An overview and some speculations on their genesis". In the margin of Romani: Gypsy languages in contact, ed. by Peter Bakker & Marcel Cortiade, 16-44. Amsterdam: Publications of the Institute for General Linguistics, University of Amsterdam, 58. Besten, Hans Den. 1987. "Die Niederländischen Pidgins der alten Kapkolonie". Beiträge zum 3. Essener Kolloquium über Sprachwandel und seine bestimmenden Faktoren, vom 30.9-2.10 1987 [sic: 1986] an d der Universität Essen, ed. by Norbert Boretzky, Werner Enninger, & Thomas Stolz, 9-40. Bochum: Studienverlag Dr. N. Brockmeyer. Bishop, Charles A. 1975. "The origin of the speakers of the Severn dialect". Papers of the Sixth Algonquian Conference, ed. by William Cowan, 196-208. Ottawa: National Museum of Man, Canadian Ethnology Service Paper 23. Blain, Eleanor. 1989. The Bungee dialect of the Red River settlement. Winnipeg: Univer sity of Winnipeg M.A. Thesis. Bloomfield, Leonard. 1930. Sacred stories of the Sweet Grass Cree. Ottawa: National Museum of Canada. Bloomfield, Leonard. 1946. "Algonquian". Linguistic structures of Native America, by Harry Hoijer et al, 85-129. (Viking Fund Publications in Anthropology 6.) New York: Viking Fund. Boretzky, Norbert, and Birgit Igla. 1994. "Romani mixed dialects". Mixed languages, ed. by Peter Bakker & Maarten Mous, 35-68. Amsterdam: Institute for Functional Research into Language and Language Use (IFOTT). Canada's Native People, in the 1981 Census of Canada. 1984. Ottawa: Ministry of Supply and Services. Catlin, George. 1844. Letters and notes on the manners, customs, and conditions of the North American Indians. 2 vols. Reprinted, New York: Dover Books, 1973. Chappell, Edward. 1817. Narrative of a voyage to Hudson's Bay. London: J. Mawman. Crawford, John 1976. "Michif: A new language". North Dakota English 4/1.3-10. Crawford, John 1983. "Speaking Michif in four Métis communities". Canadian Journal of Native Studies 3.47-55. Crawford, John . 1985. "What is Michif? Language in the Métis tradition". In Peterson & Brown 1985,231-41. Crawford, John 1985b. "Dialects of Michif: A beginning". Proceedings of the Linguis tic Circle of Manitoba and North Dakota 25.14-15. Dahlstrom, Amy. 1986. Plains Cree morphosyntax. New York: Garland. Dahlstrom, Amy. 1987. "Discontinuous constituents in Fox". Native American languages and grammatical typology, ed. by Paul D. Kroeber & Robert E. Moore, 53-73. Bloomington: Indiana University Linguistic Club. Dickason, Olive Patricia. 1985. "From 'One Nation' in the Northeast to 'New Nation' in the Northwest: A look at the emergence of the Métis". In Peterson & Brown 1985, 1936.
362
Peter Bakker and Robert A. Papen
Douaud, Patrick. 1985. Ethnolinguistic profile of the Canadian Métis. Ottawa: National Museum of Man, Mercury Series 99. Dowse, Thomas. 1877. Manitoba and the Canadian Northwest. The Chicago Commercial Advertiser, August 30th, 1877. Drapeau, Lynn. 1980. "Les emprunts au français en montagnais". Inuktitut et langues Amérindiennes au Québec, 29-49. (Cahiers de Linguistique, no. 10.) Québec: Presses de l'Université du Québec. Drapeau, Lynn. 1991. Michif replicated: The emergence of a mixed language in Northern Québec. Paper read at the 10th International Conference on Historical Linguistics, Amsterdam. Dusenberry, Vernon J. 1985. "Waiting for the day that never comes: The dispossessed Métis of Montana". In Peterson & Brown 1985, 119-36. Evans, Donna. 1982. "On coexistence and convergence of two phonological systems in Michif'. Work Papers of the Summer Institute of Linguistics, University of North Dakota Session, 26.158-73. Fredeen, Shirley. 1991. Sociolinguistic survey of indigenous languages in Saskatchewan: On the critical list. Saskatoon: Saskatchewan Indigenous Languages Committee. Giraud, Marcel. 1945. Le Métis dans l'Ouest Canadien. (English translation, The Métis in the Canadian West, Edmonton: University of Alberta Press, 1986.) Goddard, Ives. 1990. "Primary and secondary stem derivation in Algonquian". Interna tional Journal of American Linguistics 56.449-83. Gruiter, V.E. de. 1990. Het Javindo. De verboden taal. The Hague: Moesson, 2nd revised edition. Hogman, Wes. 1981. "Agreement for animacy and gender in the Buffalo Narrows dialect of French/Cree". MASA: Journal of the University of Manitoba Anthropology Stu dents' Association 7.81-94. Howard, Joseph Kinsey. 1952. Strange empire: The story of Louis Riel. New York: William Morrow. James, Deborah. 1983. "Simple versus conjunct verbs in Moose Cree: Some whys and wherefores". Actes du quatorzième Congrès des Algonquinistes, ed. by William Cowan, 345-61. Ottawa: Carleton University. Laverdure, Patline, and Ida Rose Allard. 1983. The Michif dictionary: Turtle Mountain Chippewa Cree. Winnipeg: Pemmican Publications. McDougal, John. 1903. In the days of the Red River Rebellion: Life and adventures in the far West of Canada (1868-1872). Toronto: W. Briggs. Mishaegen, Anne de. 1947. In de Canadeesche wouden. Naarden: Rutgers. Muysken, Pieter. 1996. Callahuaya. This volume. Muysken, Pieter. 1996. Media Lengua. This volume. Muysken, Pieter, and Peter Bakker. 1996. "Relexification". Amsterdam: University of Amsterdam, MS. Orser, Lori L. 1984. Mitchif: A problem in classification. Lawrence: University of Kansas M.A. thesis. Papen, Robert A. 1984a. "Un parler français méconnu de l'Ouest Canadien: le métis. 'Quand même qu'on parle français, ça veut pas dire qu'on est des Canayens!' " La langue, la culture et la société des francophones de I'Ouest, ed. by Pierre-Yves
Michif: A Mixed Language based on Cree and French
363
Mocquais, André Lalonde, and Bernard Wilhelm, 121-36. Regina: Institut de Recher che du Centre d'Etudes Bilingues. Papen, Robert A. 1984b. "Quelques remarques sur un parler français méconnu de l'Ouest canadien: le métis". Revue Québécoise de Linguistique 14/1.113-39. Papen, Robert A. 1993. "La variation dialectale dans le parler français des Métis de l'Ouest canadien." Francophonies d'Amérique 3.25-38. Pentland, David H. 1982. "French loanwords in Cree". Kansas Working Papers in Linguistics 7.105-17. Peterson, Jacqueline. 1981. The people in between: Indian-White marriage and the genesis of a Métis society and culture in the Great Lakes region, 1680-1830. Chicago: University of Illinois at Chicago dissertation. Peterson, Jacqueline, and Jennifer S.H. Brown, eds. 1985. The new peoples: Being and becoming Métis in North America. Winnipeg: University of Manitoba Press. Rhodes, Richard. 1977. "French Cree - a case of borrowing". Actes du huitième Congrès des Algonquinistes, ed. by William Cowan, 6-25. Ottawa: Carleton University. Rhodes, Richard. 1982. "Algonquian trade languages". Papers of the Thirteenth Algonquian Conference, ed. by William Cowan, 1-10. Ottawa: Carleton University. Rhodes, Richard. 1985. "Métchif: A second look". Actes du 17ème Congrès des Algonquinistes, ed. by William Cowan, 287-96. Ottawa: Carleton University. Ross, Alexander. 1856. The Red River Settlement: Its rise, its progress and present state, with some account of the Native races and its general history to the present day. London: Smith, Elder. Reprinted, Minneapolis: Ross & Haines, 1957, 1962. Sharrock, Susan R. 1974. "Crees, Cree-Assiniboines, and Assiniboines: Interethnic social organization on the far Northern Plains". Ethnohistory 21/2.95-122. Slobodin, Richard. 1966. Métis of the McKenzie District. Ottawa: Canadian Research Centre for Anthropology. Stobie, Margaret. 1968. "Background of the dialect called Bungi". Historical and Scien tific Society of Manitoba, Series III, 24.65-75. Stobie, Margaret. 1971. "The dialect called Bungi". Canadian Antiques Collector 6/8.20. Thomason, Sarah Grey, and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Todd, Evelyn M. 1970. A grammar of the Ojibwa language: The Severn dialect. Chapel Hill: University of North Carolina dissertation. Trudel, Marcel. 1960. L'esclavage au Canada français: Histoire et conditions de l'escla vage. Québec: Les Presses de l'Université Laval. Walters, Francis J. 1969-1970. "Bungee as she is spoke". Red River Valley Historian and History News (Quarterly Journal of the Red River Valley Historical Society), fallwinter 1969-1970, 3/4.68-70. Weaver, Deborah. 1982. "Obviation in Michif'. Work Papers of the Summer Institute of Linguistics, University of North Dakota Session, 26.174-262. (Also, Grand Forks: University of North Dakota M.A. thesis.) Wolfart, H. Christoph. 1973. Plains Cree: A grammatical study. Philadelphia: American Philosophical Society. Wolfart, H. Christoph, and Salina M. Shrofel. 1977. "Aspects of Cree interference in Island Lake Ojibwa". Actes du huitième Congrès des Algonquinistes, ed. by William Cowan, 156-67. Ottawa: Carleton University.
Media Lengua Pieter Muysken Universiteit
1.
van
Amsterdam
Introduction
The variety of Media Lengua (ML, literally "half language" or "halfway language") described here is spoken in Central Ecuador (Muysken 1979, 1981b, 1986). Linguistically speaking, it is essentially Quechua (Q) with the vast majority of its stems replaced by Spanish (Sp) forms. This process of replacement is commonly called RELEXIFICATION. Examples of Media Lengua utterances are given in (l)-(3); in each set, (a) is ML, (b) is the regional Quechua equivalent, and (c) is the regional Spanish equivalent.1 (1)
a.
ML: Unu fabur-ta pidi-nga-bu bini-xu-ni. one favor-ACC ask-NOM-BEN come-PROG-1 "I come to ask a favor." b. Q: Shuk fabur-da maña-nga-bu shamu-xu-ni. Sp: Vengo para pedir un favor.
It is clear that (l)a has resulted from putting the phonological shapes of the words in (l)c into the lexical entries in (l)b. Thus shuk is replaced by unu, maña- by pidi-, etc. Several things should be noted. First, we get an emphatic form of the indefinite article in Media Lengua, unu, rather than Spanish unemphatic un. Second, the Spanish irregular verb form vengo appears in a regularized stem form, bini. Third, the Quechua rule voicing the accusative case marker -ta to -da after fabur has not applied in Media Lengua; Quechua dialectological evidence suggests that this is a recent change. Fourth, what is peculiar about Media Lengua is not so much that it contains Spanish words (many dialects of Quechua do as well), but that ALL Quechua words, including all core vocabulary, have been replaced. Fifth, the Spanish forms have been
366
Pieter Muysken
adapted phonologically to Quechua; mid vowels have been replaced by high vowels. Quechua word order and morphology have been retained. A similar example is given in (2): (2)
a.
ML: Kuyi-buk yirba nuwabi-shka. cavia-BEN grass there.is.not-SD "There turns out to be no grass for the cavias." b. Q: Kuyi-buk k'iwa illa-shka. c. Sp: No hay hierba para los cuyes.
Note that the Quechua word kuyi "cavia (guinea pig)" appears in the local Spanish as well. The Media Lengua verb maintains the Quechua-specific "sudden discovery tense" suffix -shka. The Quechua negative existential verb stem illa- has been relexified with a newly formed "frozen" stem nuwabi-, derived from Spanish no "not" and haber "have". The Spanish verb "have" has an impersonal form hay which also has existential meaning. (3)
a.
ML: Yo-ga awa-bi kay-mu-ni. I-TOP water-LOC fall-CIS-1 "I come after falling into the water." b. Q: Ñuka-ga yaku-bi urma-mu-ni. Sp: Vengo despues de caer en el agua.
Examples such as (3) show the extent to which Media Lengua utilizes the possibilities of Quechua verbal affixation. Cislocative -mu can be attached to nonmovement verb stems to indicate that the subject comes after some action; this possibility exists in both Media Lengua and Quechua. What examples (l)-(3) illustrate is that (a) Media Lengua is essentially the product of replacing the phonological shapes of Quechua stems with Spanish forms, maintaining the rest of the Quechua structure; (b) the Spanish forms chosen have undergone regularization and adaptation to Quechua morphophonology; (c) Media Lengua is conservative in sometimes reflecting earlier stages in Quechua pronunciation; (d) Media Lengua is not made up on the spot every time it is spoken; (e) the occurrence of Spanish strong alternants, frozen composites, etc., is an indication that we do not have a simple process of vocabulary replacement here; and (f) the Quechua and Spanish that have contributed to Media Lengua have influenced each other in other ways as well. In this paper I will try to give an account of both sociolinguistic and structural aspects of Media Lengua. In §2 I sketch the recent linguistic history
Media Lengua
367
of Ecuador and the current sociolinguistic situation. In §3 I try to give a detailed description of one particular variety of Media Lengua, referred to in the text simply as "Media Lengua", and trace properties that are due to Spanish, properties due to Quechua, and innovative features. Section 4 con trasts Media Lengua with other contact varieties spoken in Ecuador, and in §5 I discuss other varieties of Media Lengua and try to place ML in the overall context of the contact history of Quechua and Spanish in the Andes. Section 6 provides a brief conclusion and a suggestion for further research. The type of Media Lengua described in §3 below is spoken in semirural communities near the town of San Miguel de Salcedo, in the Cotopaxi province of Ecuador, by Indian peasants, weavers, and construction workers. The area, situated at an altitude of 2,800 meters, is one of the poorest in a poor country; but it is relatively accessible from the capital, Quito. Salcedo — transformed from an Indian town and Inca tambo (stopping-off point and inn) by the Spanish in 1570 — has around 5,000 inhabitants, and is the regional center and market town for a rural area with another 10,000 inhabitants, mostly Indians. Fieldwork on Quechua, rural Spanish, and Media Leng ua in the area was carried out in the period 1974-76 and in 1978 (cf. Muysken 1977, Muysken & Stark 1978, Muysken 1984). About twenty hours of Quechua and thirty hours of local Spanish were recorded and analyzed. Three samples of Media Lengua were gathered, totaling about four hours of conversation of five speakers.
2.
Sociolinguistic history and current status
To understand how Media Lengua emerged it is important to consider the history of both Quechua and Spanish in Ecuador in some detail. Quechua is a language family primarily spoken in the central Andean highlands, from Northern Argentina to Southern Colombia, and in adjacent parts of the Amazonian lowlands. Its origin probably lies in central Peru, where the precursors of the two most important varieties of Quechua may have been spoken around 500 A.D. The spreading of Quechua was largely due to the Incas, who chose Quechua — a widely-used trade language at the time — as the language of empire, and brought it from Peru to Bolivia, Argentina, and Chile.
368
Pieter Muysken
The introduction of Quechua into Ecuador proceeded differently, according to Torero (1974). Traders had brought a variety that he terms Chinchay Standard to Ecuador in the pre-Inca period. During the Inca period, starting in the late 15th century, another variety was introduced, which Torero calls Chinchay Inca, used by administrators sent from Cuzco (Peru) to Ecuador. Present-day Ecuadorian Quechua has Chinchay Standard as its basis, with some features of Chinchay Inca in addition. Most probably, Quechua does not antedate Spanish by much in Ecuador. While Quechua may have been a lingua franca prior to the Incaic conquest, the majority of the population of Highland Indians did not speak Quechua (but some pre-Inca tribal language) at the time of the Spanish conquest around 1540, and the elites knew some Quechua in addition, according to the colonial documentary sources. It is highly unlikely that any Quechua was spoken in the Ecuadorian Amazon basin at that time. The Spanish conquerors did not immediately impose their language upon the Indian serfs and peasants, but rather encour aged the use of Quechua, in the beginning the Indian language of elite intertribal communication, as a general language of the Indian caste. During the colonial period, Quechua became the native language of the conquered populations. This may be due in part to the linguistic policies of the Spanish regime — including the use of Quechua by the Church as the language of evangelization — but it must be explained primarily as a result of detribalization, the imposition of a caste system, and the growing identification of the Indian peasantry with a by now mythicized Inca past. After independence in 1830, the feudal agrarian economy and the caste system were slowly broken down. Some Indian workers started to move to the cities, and bilingualism increased. This brought with it a change in the status of Quechua: it became a low-prestige language, associated with backward ness and ignorance, rather than the language of one of the two clearly identifiable castes. At present Quechua has the status of an oppressed lan guage, used by over a million Highland Indian peasants but held in contempt by most Whites and mestizos. Present-day Ecuadorian Quechua presents a considerable amount of phonological variation, but little syntactic or morpho logical variation. Sources include the dictionary by Muysken & Stark (1978), a pedagogical grammar by Stark et al. (1972), a comparative syntactic de scription of the verb phrase by Muysken (1977), and a comprehensive de scription by Cole (1982). In many ways it resembles the Quechua II dialects of southern and central Peru and of some regions, like Cajamarca, in northern
Media Lengua
369
Peru, but it differs from all these dialects in its considerably simplified morphology. This simplification could be accounted for by assuming that Ecuadorian Quechua has its roots in the pre-Incaic Coastal trade language, termed Chinchay Standard above, itself a simplified version of a Coastal Quechua II dialect. Map I gives an idea of the present-day distribution of Quechua in Ecuador and of the different varieties of Media Lengua discussed below. Spanish became the prestige language in the new colonial situation and is the majority language in contemporary Ecuador, spoken by everybody except some of the Indian peasants in the Highlands (many of the peasants are also bilingual) and tribal groups in the Amazonian and Coastal Lowlands. Presentday Ecuadorian Spanish has been described most completely by Toscano Matteus (1953). There are considerable lexical and phonological differences between Coastal and Highland Spanish, on the one hand, and between edu cated urban Spanish and different rural varieties on the other hand. On the whole Ecuadorian Spanish conforms to the generalization that holds for all varieties of South American Spanish: many archaic Peninsular features have been preserved, and the dominant dialectal influences have been the Andalusian dialects of Southern Spain. When we consider mutual Quechua-Spanish influence, the picture is quite complex. Lexically, the influence of Quechua on Spanish has been slight, limited to the domains of agriculture, food, flora and fauna, the household and family, and exclamatives. Syntactically and semantically, however, Quechua has had considerable influence on the Highland varieties of rural Spanish, even though specific changes can often be shown to have internal motivation as well. I return to this in §4 below. The 400-year domination of Spanish over Quechua has left many traces in the subjugated language. These mainly concern the lexicon, which is crucial to understanding the emergence of Media Lengua as well and will be discussed in some detail in §3.3. Only a few aspects of the grammar have been influenced by Spanish. Morphological and syntactic influence of Span ish on Quechua appears to have been rather slight, although it has not been adequately studied yet. We find Spanish diminutives in Quechua (4), and Spanish agentive markers (5): (4)
wasi-ta "little house" (Sp -itulita, as in pollito "chick") wawki-ku "little brother" (Sp -icu/ica, as in ratico "moment")
Map of Media Lengua area: The distribution in Ecuador of Quechua and of the various contact varieties discussed, PID = attestation of Pidgin Spanish.
Pieter Muyssken
Media Lengua (5)
371
awa-dur "weaver" wachachi-lun "midwife"
(Sp -dor, as in trabajador "worker") (Sp -Ion, as in dormilón "sleepy")
The last example is from the Quechua of Saraguro, Loja. It is possible, of course, that -lun is not derived from Spanish but from a pre-Incaic Indian language. The suffix -ku could also be connected with an Aymara polite address suffix. These borrowed suffixes are in no way regular and are rarely productive. Only -itulita is marked for gender in Quechua, in as yet little understood ways. Quechua -ku is invariant for gender and occurs only with names for humans. Saraguro Quechua -lun does not function as an agentive in Spanish, but rather marks a characteristic of a person. Syntactically, there are several ways in which Spanish has influenced Ecuadorian Quechua. First, we encounter in the speech of most Quechua speakers, bi- or monolingual, the use of Spanish conjunctions such as i "and", and piru "but", in addition to Quechua conjunction suffixes. Second, we find occasional subordinating markers such as porke "because" and prepositions such as sin "without", but only in the speech of bilinguals. Third, we encoun ter instances of SVO or VSO order instead of SOV order for some speakers, although Ecuadorian Quechua as a whole seems to be stricter in maintaining SOV order than most Peruvian Quechua dialects. This is probably because verbal marking for person is less extensive in Ecuadorian Quechua, espe cially in subordinate clauses. Spanish lexical influence may have begun at the earliest stage, in which the Chinchay Standard lingua franca was expanded at the expense of native languages of the peasant population, and it may have increased after indepen dence, when bilingualism became more frequent. This becomes clear when we consider earlier word lists and dictionaries. The evidence found in Velasco's relatively short word list (1780 [1964]) is a case in point. Most of the borrow ings mentioned are clearly related to the Catholic faith, such as those in (6); the Quechua forms are on the left, Velasco's glosses are in the middle column, and English glosses are on the right: (6)
anima bendicishca diospac simi christiano runa
"alma" "bendito" "sermon" "fiel"
"soul" "blessed" "sermon" (lit. "word of God") "Christian (person)"
It is not clear to what extent these words were used by people other than the missionaries themselves. However, the fact that Quechua anima does not
372
Pieter Muysken
correspond to the Spanish word alma given as a gloss suggests that some words were adopted by the Indians themselves. In addition, Velasco mentions a few non-religious borrowings, such as the following: (7)
azuti azutina
"azote" "azotar" "afilar" "piedra aguzadera"
filana afilaría rumi
"whip" "to whip" "to whet" "whetstone"
These words, which can still be found in contemporary varieties, show that borrowings occurred in ordinary vocabulary in the eighteenth century. Notice that the borrowings are not limited to nouns, a point to which I will return below. Another source of evidence for the history of borrowing is developments that have taken place within Spanish itself. These might be lexical changes, as in the case of alpargatis "sandal", a borrowing in Highland Quechua corre sponding to an early colonial word derived from Arabic. This word is now no longer used in Spanish outside the Quechua context, suggesting that it was an earlier borrowing. More systematic evidence comes from sound changes. In colonial Span ish, which had strong Andalusian roots, the h in words like hondo "deep" and hacienda "estate" was pronounced. In contemporary Ecuadorian Spanish it is generally not pronounced, although Toscano Matteus (1953) mentions some attestations of h in modern dialects. A sizable number of borrowings of words which were originally pronounced with an h are pronounced in Quechua with an initial aspirated or fricative velar, and must have been borrowed before the loss of h; examples are given in (8): (8)
[xamaca] [xazinda] [xacha] [xabas] [xundu]
< < < < <
hamaca hacienda hacha haba{s) hondo
"hammock" "estate" "axe" "lima bean(s)" "deep"
In contrast, the verb haber "have" appears in Media Lengua as abi-, not xabi-. Unfortunately, it has been impossible so far to date the loss of h in highland Spanish. Spanish lexical influence on Quechua has been pervasive in many re gions, where we sometimes find up to 40% borrowed Spanish vocabulary
Media Lengua
373
(types) in spoken (not elicited) Quechua. Domains of borrowing include dress, kinship, tools and machines, social status, commerce, social celebra tion, religion, time reference, cosmos and topography, measures, and psycho logical states. Both in Quechua and in Spanish the use of vocabulary from the other language has sociolinguistic and stylistic connotations, but in Quechua it is unavoidable. To get an idea of the extent and the nature of Spanish lexical influence, consider the beginning of a traditional story (which will contain less Spanish influence than many other forms of discourse) in (9). Spanish elements are in boldface: (9)
THE GREEDY SELLER:
Ñuka parlu-wa-da parla-gri-ni ñ Collana-munda awilu-guna I a.story am.going.to.tell my from.Collana grandparents parla-shka-da. Chimborazu-mun-shi primero shuk pobre ri-n have.told. To.Chimborazo.they.say first a man poor goes k'atu-na-un. Chi-munda-ga ri-n, ri-n, ri-n, k'atu-na-un. Mana with.goods, therefore he goes, goes, goes, with.goods. Not kay-bi k'atu-y pudi-sha, chay-mun k'atunga-bu ri-n. Chi-munda here sell capable there with.goods he.goes. In that shuk amu kaballu monta-shka tupa-sha-ga, ni-shka: way a.lord on.horse mounted meeting, said.to.him: -May- ri-xu-ngi? Where are you going? -Shina k'atu-na-un ri-xu-ni, k'atu-na-un rixu-ni. Thus with.goods I.am.going, with.goods I.am.going. -Haku. Shina-ga ñ pweblo-W k'atu-ngi. Let's.go. Thus in my town you.can.sell. Shina paramo-da pusha-n, pusha-n, pusha-n, ri-n-shi. Hence across.the.plain he takes him, takes him, takes him along. Pay ri-n, ri-n, ri-n, ri-n. De repente paxa uku-mun yayku-n, china-ga. He goes, goes goes. Suddenly in.straw' they enter, like that. I will now turn to a consideration of the sociolinguistic profile of the Media Lengua-speaking community. The Indians in the region can be subdi vided into three groups: (i) CARGADORES. This group is transient, coming from a monolingual Quechua-speaking mountain region about 80 kilometers away, and arriving in the market town of Salcedo knowing little or no Spanish. The cargadores
374
Pieter Muysken
establish no local ties, coming and going depending on their earnings, and they generally learn just enough Spanish to be able to do their work. They are culturally and also dialectologically distinguished from the Indians of the surrounding communities. They do not speak Media Lengua, but rather Quechua and some rudimentary Spanish. (ii) CAMPESINOS. This group lives in the communities surrounding Salcedo but has frequent contacts with, and is dependent on, people from the town. Culturally rather distinct, the campesinos make no attempt to become part of urban mestizo society and culture. Many campesinos speak rather fluent but highly nonstandard Spanish, although they speak Quechua at home. (iii) OBREROS. Since 1909, when Salcedo was linked to the capital Quito by train, but particularly since 1967, the date of the land reform, many younger men from the Indian communities have started to work in Quito, in construction or in industry. They return to their communities with some money, speaking fluent urban Spanish, and they speak Quechua only infre quently. This is the group that primarily speaks Media Lengua. Media Lengua is spoken both as a native and as a second language by acculturated Indian peasants, craftsmen, and construction workers. It is the native language of the younger adults and most of the children in the commu nities nearest to Salcedo. Media Lengua-speaking communities contain Me dia Lengua speakers who do not know Spanish. In 1900 these communities must have been almost totally monolingual in Quechua, with a few young adult male (not necessarily fluent) bilinguals who introduced the Spanish lexical shapes. Nowadays some younger children may have Spanish as their first language, and some older people speak Quechua natively, but the middle generation has Media Lengua as its first language. Everyone speaks Media Lengua, however, though only within the community. Given that (a) nearby communities are mostly Quechua-speaking; (b) the language is slightly more conservative phonologically than the local Quechua (recall the absence of voicing in the suffix in fabur-ta in ex. 1 above); and (c) old people were mostly Quechua-speaking in 1975, but the middle generation was Media Lengua-speaking, an origin of the language between 1920 and 1940 is indicated. The villages where Media Lengua is spoken are socially and geographi cally intermediate between the blanco world of the urban centers in the valleys and the neighboring Indian world of the mountain slopes (see Figure
D.
Media Lengua
375
Figure 1. A cross-section of the Inter-Andean valley. The different zones for Spanish, Media Lengua, and Quechua are schematically indicated; elevations are given in meters.
In the communities, Spanish is the language of contacts with the non-Indian world and of the school, Quechua is the language of tradition and of contacts with the Indian campesinos higher up the slopes, and Media Lengua is the language of daily life within the community. Why did Quechua speakers relexify their language and create Media Lengua? It may be useful here to compare Media Lengua with pidgins and creoles. Most theories of pidgin (and hence creole) genesis assume that pidgins emerged through the need for communication among people with different language backgrounds, processes of incomplete target-language learning due to quantitatively restricted second language input, and qualita tive restrictions on the target-language input through the use of foreigner-talk registers by speakers of dominant groups. I will argue here that none of these processes contributed to the emergence of Media Lengua. First, Media Lengua is an intragroup language, not known outside the communities where it is spoken. Because its structure is almost entirely Quechua and the vocabulary taken from Spanish has been both relexified and adapted to Quechua phonology, it is no more intelligible to Spanish speakers than Quechua is. In fact few, if any, Spanish speakers understand Media Lengua, while a substantial number of them have some knowledge of Quechua through contacts with Indians from different communities in the region. And when I played tapes of Media Lengua to Quechua speakers from nearby areas, they could recognize it as some sort of strange Quechua — presumably because of the suffixes — but could not understand it.
376
Pieter Muysken
Second, Media Lengua cannot be considered to represent a stage in learning Spanish as a second language, since many Media Lengua speakers also speak fluent Spanish, and since Media Lengua is very different from QuechuaSpanish interlanguage, as I will show in §4 below. Furthermore, if Media Lengua were the product of an ongoing process of second-language learning, we would expect much more variation in the language than actually occurs. Finally, Media Lengua and Spanish foreigner talk share almost no fea tures. Spanish foreigner talk is characterized by the use of infinitives, frequent diminutives, and a general reduction of Spanish surface structures. Only coincidentally, as in the absence of articles (which are lacking in Quechua as well), do we find any similarity with Media Lengua. If these factors typical of pidgin genesis are not responsible for the emergence of Media Lengua, how then can it be explained? I will suggest that Media Lengua came into existence because acculturated Indians could not identify completely with either the traditional rural Quechua culture or the urban Spanish culture. Thus it was not communicative needs that led to its emergence, but rather expressive needs. It appears that ethnic self-identifica tion is of crucial importance in determining the relation between Quechua, Media Lengua and Spanish in the Ecuadorian Highlands. Media Lengua is not the product of an interlanguage arrested and fixed, resulting from an emer gency contact situation; but rather it is a departure from Quechua through massive relexification, and not at all along the path of Quechua-Spanish interlanguage (cf. also Le Page & Tabouret-Keller's 1985 analysis of the multilingual situation of Belize in terms of "acts of identity"). The Media Lengua-speaking communities studied here are located on the fringe of a Quechua-speaking area, to which these communities histori cally belonged. Due to their geographical situation and to the necessity for and ability of their inhabitants to make frequent trips to the capital to look for work, the communities have come to be culturally differentiated from neigh boring areas, to the extent that their people find it necessary to set themselves apart from their neighbors. The following comments on Media Lengua illustrate attitudes people hold toward the language; (10), (11) and (13), originally in Spanish, are given only in English translation, while (12) is given in the Media Lengua original. (10) "That way the people living [i.e. right after the Spanish conquest] got confused, that's why that Media Lengua has come out, because they understood neither Quechua nor Spanish very well."
Media Lengua
11
This comment illustrates the sense of cultural indeterminacy (although it misses the point that the creators of Media Lengua must have been bilingual to some extent); the other two stress the private communal character of Media Lengua: (11) "We just speak it among ourselves, among people we know; the others always speak Spanish. Only in this province of Cotopaxi we talk like this. The other provinces don't speak it, only Quechua." (12) Media Lengua-ga así Ingichu-munda Castallanu-da - Media Lengua- thus Quechua-from Spanish-c talk-NOM kiri-xu-sha, no abla-naku-ndu-mi asi, chaupi-ga Castellanu want-PROG-suB, not talk-pl-SUB-AFF thus, half- Spanish laya, i chaupi-ga Ingichu laya abla-ri-na ga-n. Isi-ga asi like, and half- Quechua like talk-REFL-NOM be-3. This- thus nustru barrio-ga asi kostumbri-n abla-na. our community- thus accustomed-3 talk-NOM "Media Lengua is thus if you want to talk Spanish from Quechua, but you can't, then you talk half like Spanish, and half like Quechua. In our community we are accustomed to talking this way." The position of Media Lengua halfway between Quechua and Spanish is stressed in (12) and (13). Notice the colloquial name given to Media Lengua: quichua chico or utilla ingiru, both meaning "little Quechua". (13) "Yolanda knows some Quechua, right?" "Big Quechua [i.e. Quechua] she doesn't know, but little Quechua [i.e. Media Lengua] she has learned from her grandmother." If we assume that Media Lengua provides communities and individuals with a way of articulating their sense of cultural identity — an identity which cannot be fitted into the traditional strict equations Quechua = Indian and Spanish = White, which the caste society of the Ecuadorian Andes has provided — then the conservatism apparent in Media Lengua becomes more understandable. A final issue to be considered here is the retention of Quechua stems in Media Lengua. In my Media Lengua data a number of Quechua stems have been retained; these are listed in Appendix 2. Some Quechua items in Media Lengua refer to specific culture items. An example is Quechua shutichiy "baptism"; here it is not semantic complexity that has prevented relexifica-
378
Pieter Muysken
tion, since Spanish bautismo, the term the priest would use anyway, is perfectly equivalent. Another category consists of Quechua words which are also used in regional Spanish, such as wawa "child". For the remaining elements it is difficult to determine a specific reason why they were retained. It may be that (for instance) warmi "wife, woman" was retained because the Spanish alternative mujer both ends in a consonant and has final stress, two features that are alien to Quechua (cf. §3.1. below). In any case, Appendix 1, presenting Swadesh's 200-word core vocabu lary list, shows how different Media Lengua is from the local Quechua. Discounting suffixes, which are all Quechua, we find the following distribu tion: in the Quechua list 181 out of 208 items (= 87%) are native Quechua, while in the Media Lengua list 106 of the 118 items that occur in my corpus (= 90%) are from Spanish. The latter percentage matches that for the overall vocabulary of Spanish origin in the Media Lengua rpus: out of a total of 1798 verb tokens, 1597 (= 89%) are of Spanish origin. Consequently, the main difference between Quechua and Media Lengua is not a gradual one; there is a quantitative leap, involving most of the core vocabulary. By con trast, in Ecuadorian Quechua dialects the percentage of Spanish borrowings varies between 11% and 40% (types). There is no variety attested in between the most hispanicized Quechua (40% Spanish types) and Media Lengua (90% Spanish types). This makes a scenario for Media Lengua in terms of heavy borrowing even less plausible.
3.
Structure
Basically, as noted above, Media Lengua is Quechua with Spanish stems. To see what this means, consider Table 1, in which a schematic structural comparison between the two languages is presented. I will refer to aspects of this table in the following discussion. However, Media Lengua does have structural features that are not found in Quechua. Some of these are due to Spanish and some to independent developments. These will be discussed as part of my detailed survey of Media Lengua structure: I focus in §3.1 on phonological adaptations of Spanish elements, and in §3.2 on morphological properties. Section 3.3 deals with semantic changes in lexically related words, and §3.4 concerns syntactic features of Media Lengua.
Media Lengua
379
Table 1. Principal Structural Features of Quechua and Spanish. QUECHUA
SPANISH
a. Word order: a. XV dominant Adjective/noun Possessor/possessed postpositions/case suffixes b. Argument marking: b. NP-CASE . Verb marking: 1. Main clauses — tense/person 2. Nominalization 3. Gerunds d. Subordination: d. Nominalizations -y infinitival -shka definite -na indefinite -k agentive Gerunds . . Validation (indication of the source of the information given): semiobligatory through enclitic elements
Word order: VX dominant Noun/adjective dominant Possessed/possessor prepositions Argument marking: (P) NP / word order Verb marking: 1. Finite clauses — tense/person 2. -r infinitives 3. -ndo gerunds 4. -do past participles Subordination: 1. que + finite clause 2. -r infinitives 3. Gerunds 4. Participles
Validation periphrastic and optional
Before turning to these different aspects, it is perhaps useful to make my use of the term relexification more precise. Generally relexification is used to refer to a process of lexical borrowing which involves (a) a large part of the vocabulary and (b) the replacement of native items, rather than the mere addition of vocabulary. The way I use the term goes one step further: I assume that in fact the Spanish forms do not enter as full lexical entries, but as phonological shapes which are grafted onto Quechua lexical entries. Thus the original entry is not replaced, but merely altered in outer shape. Below I will adduce evidence for this view of the process. 3.1. Phonology Pending a more detailed analysis of the phonology of Media Lengua and of Central Ecuadorian Quechua dialects, the following remarks give a prelimi-
380
Pieter Muysken
nary sketch of how Media Lengua phonology relates to that of Quechua and Spanish. Table 2 gives the phoneme inventories of Central Ecuadorian Quechua and Spanish. Because we are concerned here with the fate of Spanish words when they are incorporated into the predominantly Quechua phonology of Media Lengua, the most important differences involve elements or combinations of elements present in Spanish but not in Quechua. The voiced stops [b], [d], and [g] occur in Quechua primarily in loans from Spanish and from unidentified American Indian substrate languages. In addition, they can result from rules that voice initial consonants of suffixes. Unlike Peruvian and Bolivian Quechua, where [e] and [o] often occur as the result of the (sometimes lexicalized) lowering of /i/ and /u/ in the context of the postvelar stop /q/, Ecuadorian Quechua, lacking /q/, only has [i] and [u].
Table 2. Phoneme inventories of Quechua and Spanish. QUECHUA CONSONANTS:
unaspirated aspirated (voiced affricates
t th d ts, dz s z n 1 r
p ph b
m
w
ch chh
k kh g)
sh zh ñ 11
X
h
ch
k
u
QUECHUA VOWELS: i
a SPANISH CONSONANTS:
t d s n 1 r, rr
P b f m
w SPANISH VOWELS:
g X
ñ 11
u
i
e
0
a
Media Lengua
381
We find that e and are often, but not always, pronounced as i and u, respectively (with some variation that also occurs in the Quechua pronuncia tion of Spanish loans). The Spanish vowels [e] and [o] are often retained in names and interjections. In stressed position [e] and [o] are more frequently retained than in unstressed position. High-frequency verbs such as dizi"say", azi- "do, make", bini- "come", and pudi- "can, be able" are always pronounced with high vowels. By contrast, the negator no and the singular pronouns yo "I", bos "you", and el "(s)he" very rarely are. Note that these last elements do not fit the Quechua bisyllabic template for separate words. The Spanish cluster ue [we] is sometimes pronounced [u], sometimes [wi], sometimes [i]: (14) llubi- "rain" (Sp llueve [llwébe]) nustru "we, our" (Sp nuestro [nwéstro]) afwira "outside" (Sp afuera [afwéra]) pwirta "door" (Sp puerta [pwérta]) manil "Manuel" (Sp [manwél]) The cluster ui is pronounced [u], as in sanduchi "sandwich" (Sp sanduiche [sandwiche]). The Spanish cluster ie [ye] is pronounced [i]: (15) risin "recently, just" (Sp recién [resyén]) bin "well" (Sp bien [byen]) siti "seven" (Sp siete [syéte]) nadis "nobody" (Sp nadie(s) [nádye]) birnis "Friday" (Sp viernes [byérnes]) In the name Rafael the vowel sequence is pronounced [i], but the word for maiz "corn" is [mais], and Sp traer is [trai-]. The sequence [ai] occurs in native Quechua words as well. With consonants and consonant clusters, a few complications arise. While Spanish ƒ is adapted to aspirated [ph], Spanish b, d, and g are retained (as they are in Spanish loans in Quechua itself). Spanish intervocalic s is pronounced as a voiced [z]: (16) azi- "do" (Sp hacer [aser]) dizi- "say" (Sp decir [desir]) kizu "cheese" (Sp queso [keso]) kaza "house" (Sp casa [kasa]) konozi- "know" (Sp conocer [konoser])
382
Pieter Muysken
It should be noted that this voicing does not apply to Quechua intervocalic [s] : Q wasi "house", Q kasa "frost". Spanish s is palatalized before t, as in ML phishta "feast" (Sp fiesta [fyesta]) and ML prishta- "loan" (Sp prestar). Some consonant clusters are retained: (17) br. tr: pr: kw:
ambri- "be hungry" trai- "bring" prishta- "loan" kwantu "how much", kwatru "four"
The cluster str is sometimes pronounced [rzh], as in ML nurzhu "we, our" (Sp nuestro [nwestro]). As in many non-bilingual varieties of rural Spanish, Spanish initialfw is pronounced [xw]: MLxwirti (Spfuerte [fwerte], [xwerte]). 3.2. Morphology A number of the morphological features of Media Lengua merit special discussion. To begin with, it is important to consider the typological charac teristics of Quechua with respect to borrowing. One of the ways in which languages differ is with respect to the bor rowability of different categories. Earlier studies on lexical borrowing show that nouns are more frequently borrowed than verbs. Nouns are syntactically inert, that is to say, they do not influence the syntactic make-up of the clause in the same way as verbs. In my study of Spanish borrowings in the Quechua of Ecuador I have been able to establish that, in this case too, nouns constitute the largest group of borrowings. But when we look at verbs we see that Quechua — in contrast to other languages — has borrowed relatively large numbers of verbs from Spanish. In comparison with e.g. English borrowing in Canadian French (Poplack et al. 1988), the distribution of Spanish-toQuechua borrowings over the different syntactic categories is more nearly even. How can we explain these high percentages of borrowed verbs? The agglutinating structure of Quechua may provide an answer. The morphology in an agglutinating language is characterized by the fact that suffixes can be attached to word stems without specific morphophonemic restrictions or alternations:
Media Lengua
383
(18) wasi ukhu-bi house inside-LOC "inside the house" (19) wasi-da-mi riku-rka xwan-cha-ga house-ACC-AFF see-PAST Juan-DIM- "It's a house that he, Juan, saw." In this way Spanish verbs can easily be turned into Quechua verbs: (20) trabaja{r) → trabaxa(na) entende(r) → intindi(na) This is a productive process and can be applied to each Spanish verb. The verb behaves completely like a native Quechua verb. The morphology of Quechua thus makes it possible to adopt any Spanish verb. The mid vowels Id and /o/ are collapsed with the high vowels /i/ and /u/, respectively, and the stem-vowel is maintained as part of the new root. Spanish irregular verbs are regularized in Media Lengua. As shown in (21), they derive from inflected third-person singular or infinitive Spanish forms: (21)
MEDIA LENGUA
i(but: bamu-chi da-ldalibiazirisabi-
SP INFINITIVE
SP lsg
b.
3sg
ir voy va (← Sp vámonos "let's go") dar doy da (dale) ver veo ve hacer hago hace reírse rie saber sé sabe
These verbs receive the normal Quechua affixes: (22) a.
SP
ML: no
sabi-ni-chu.
not
know-1-NEG
Q: mana yacha-ni-chu. Sp: no sé. "I don't know" ML: ya i-gri-ni. already go-INC-1 Q: ña ri-gri-ni.
"go" "give" "see" "do" "laugh" "know"
384
Pieter Muysken Sp: ya me voy. "I'm already going." ML: bos-mu da-ni-mi. you-to give-1-AFF Q: kan-mu ku-ni-mi. Sp: te (lo) doy a tilvos. "I give (it) to you."
In addition to verb regularization, we encounter several other processes of lexical adaptation of Spanish vocabulary in Media Lengua, such as "freez ing" (the combination of separate Sp words in a single ML word), reduplica tion, the selection of emphatic, strong forms, morphological regularization to fit the Quechua CVCV pattern, and reduction of bisyllabic verbs to monosyl labic forms: (23)
(24)
"FREEZING": SPANISH
MEDIA LENGUA
no ha habido no hay aún no a mí en qué
nuwabishka nuwábi núway aúnu ami inki
"there has been no ..." "there is no ..." "not yet" "me" "what"
REDUPLICATION:
a.
ML: yo-ga bin-bin tixi-y-da pudi-ni. I-TOP well-well weave-INF-ACC can-1 Sp: yo puedo tejer muy bien. "I can weave very well." b. ML: anda-y brebe-brebe kuzina-ngi. walk-IMP quick-quick cook-2 Sp: anda a cocinar breve. "go cook quickly." In neither case would we encounter reduplication in Quechua. The examples in (25) achieve an effect similar to "freezing": (25)
SELECTION OF MORPHOLOGICALLY STRONG FORMS:
miu (Spanish mio "mine") "my, I" tuyu (Spanish tuyo "yours") "your"
Media Lengua (26)
385
MORPHOLOGICAL REGULARIZATION:
Sp: ML: Q:
reloj relóxo rílux "watch"
sol sol, solo inti "sun"
Media Lengua has several instances of optional reduction of lexical items, e.g. the alternations yuya-ni ~ ya-ni "I think" (< Q yuya- "think") and dizi- ~ zi- "say, want" (< Sp decir "say"). When "I think" is affixed to a statement as a parenthetical, the result is ya-ni', otherwise, it is yuya-ni. Interestingly enough, the alternation is absent in Quechua (even though the verb yuyaoccurs in Quechua), and with yuya- it is limited to the first person. The choice between zi- (73 occurrences) and dizi- (50 occurrences) seems to be deter mined by considerations of emphasis and allegro vs. lento style. The existence of this type of variation in Media Lengua attests to its established character and to its stylistic range, characteristic of a language that serves as the principal language of its speech community. If Media Lengua were a recent development without native speakers, this type of variation would not be present. A further issue to be discussed here concerns Spanish morphology. I said above that only stems were relexified, and that all affixes are Quechua. This is not entirely true. We encounter the Spanish subordinating gerundive affix -ndu (originally -ndo) in Media Lengua, in places where Quechua would have the "same subject" subordinator -sha or the "different subject" subordinator -kpi. Media Lengua shows all three forms. Exx. (27) and (28) illustrate same subject subordination with -ndu and -sha, respectively: (27) alla-bi-ga entonces-ga artu terreno propio tini-ndu-ga there-LOC-TOP then- much land own have-suB-TOP riku-ya-na, no ? rich-become-NOM no "There one could become rich then, having one's own land, no?" (28) Isi-munda-ga asta kolera muri-sha bini-xu-ka-ni. this-from-TOP even anger die-suB come-PROG-PAST-1 "From this I was coming even dying from anger." In (29) and (30) different-subject subordination with -ndu and -kpi are illus trated:
386
Pieter Muysken (29) ahi-da-ga abin, piru tarde-ya-ndu-ga gana-u-nga-y there- there.is but late-become-suB-TOP win-PROG-3FUT-EMP "It is there, but when it becomes late he will be winning." (30) mama- abiza-k i-kpi-ga dizi-shka. mother-to tell-AG go-suB- say-SD "She said it when I went to tell my mother."
In Muysken (1981b) I showed that cases such as (29) are quite rare, but that sentences like (27) are frequent in the samples. It is not clear, however, that -ndu is a counterexample to the claim that no affixes are relexified in Media Lengua. There is some evidence for an earlier Indian-Spanish contact language spoken throughout Ecuador, the most strik ing feature of which was the use of -ndo or -ndu as a generalized verb marker (see Muysken 1980, and §4 below). It may well be that Media Lengua was influenced by this contact vernacular and adopted -ndu from it in certain adverbial contexts, rather than relexifying directly from Quechua in these cases. Two other Spanish affixes also occur in Media Lengua: diminutive -itul -ita and past participle -do. The first occurs in Quechua as well, and can be seen as a borrowing. The second occurs only with Spanish adjectives that were probably taken over as a whole. Therefore we can maintain the generali zation that no affixes were relexified. One might argue that the absence of relexified affixes is not due to structural properties of roots and affixes, but rather to the fact that there are no direct semantic equivalents of Quechua affixes in Spanish. Thus the fact that (for instance) the Quechua causative suffix -chi occurs in Media Lengua would be due to the absence of a similar causative suffix in Spanish. This argument is very plausible at first sight, but it has several deficits. First, not all Quechua affixes lack a Spanish equivalent. So, for instance, Quechua 2sg. -nki could have been replaced by Spanish -s, and Quechua lpl. -nchi by Spanish -mos. In a dialect of Quechua spoken else where (northern Chimborazo) the Quechua agentive suffix - is often re placed by Spanish-derived -dur (from -dor), but no such thing happens in Media Lengua. We could extend this list for other suffixes as well. The point is that semantic considerations may have led to the relexification of some affixes, but not others. They cannot account for the fact that NO affixes were relexified.
Media Lengua
387
Second, relexification has sometimes involved very considerable semantic adjustments. Consider the Quechua impersonal verb yarixa-n "it hungers (someone)". In Spanish YOll say ya tenga hambre "I have hunger". Now, it would appear to be impossible to relexify here; but there is a Media Lengua impersonal verb ambri-naya-n "I feel like hunger", containing the Quechua suffix -naya "having a physical inclination". Similar cases abound. They demonstrate that semantic factors did not prevent relexification of lexical items, even in complicated cases. Analogous arguments will be discussed in the next section, where I analyze the relexification of function words. There is no general semantic correspondence between Quechua and Spanish function words, but they are all relexified. Here there is one exception, however - the Quechua verb ka"be" - and this may weIl be due in part to semantic factors: Spanish has two verbs, ser and estar, with a semantically cQmplex distribution, and this may ac count in part for the fact that Media Lengua has consistently maintained ka-. The possible other reason is morphological: estar is somewhat irregular and ser is suppletive, like English be. On the whole, we can safely conclude that semantic reasons do not explain why Quechua affixes were not relexified. So far I have established that no affixes have been relexified and that this cannot be for semantic reasons. The same holds for clitics. Quechua has a class of phonologically dependent elements which are syntactically independent and have clausal scope, but which do not have affix properties (Muysken 1981a, Lefebvre & Muysken 1988). Affixes are sensitive to the category of the word to which they are attached, they determine the category of the resulting word, and they can only be attached to major lexical categories, such as nouns and verbs not to particles, negation markers, etc. Clitics do not have these properties: they can be attached to any constituent or lexical element and do not change the categorial status of their base. This class includes the elements in (31): (31) a.
b. c.
-mi -shi -cha -chu -pish -tik
AFFIRMATIVE
-rl
EMPHATIC
-ma -ga
EMPHATIC
HEARSAY DUBITATIVE NEGATION, QUESTION
too,
INDEFINITE
EMPHATIC
TOPIC
388
Pieter Muysken
Even though these elements cannot appear as independent words, they have all the syntactic properties of elements separately introduced by the phrase structure rules. In Media Lengua they are not relexified: (32) a.
isti sabi-nga-ma-bish. this know-3FUT-EMP-too "He will certainly know." b. kital-mi uyari-xu-n how-AFF hear-REFL-3 "how it sounds" c. nustru-ga alla-wa-bi-mi sinta-nchi. we-TOP there-DIM-LOC-AFF live-1pl "We live there."
It is hard to establish that this is not for semantic reasons. There is evidence that bilingual rural Spanish has an equivalent for -pish "too, INDEFINITE": we sometimes encounter an enclitic -tan (from Spanish también "too") that is used in the same way as Quechua and Media Lengua -pish: (33) Sp: onde-tan ha ido? where-too has gone "Where then has he/she gone?" (34) Sp: yo-tan quiero. I-too want "I want it too." Examples (33) and (34) suggest that there must be another reason why this form was not adopted in Media Lengua. In the next section we will see that lexical meaning is not what sets stems apart from the rest: a number of relexified stems do not have a lexical meaning. 3.3. Lexicon If we accept the characterization of Media Lengua as a form of Quechua with Spanish phonological shapes, then Media Lengua must be seen as a compro mise between the Quechua lexical system and the Spanish one. We have already seen several instances of this compromise. Here the processes operat ing in it and the consequences it has for the grammar of the language will be examined more systematically.
Media Lengua
389
One of the differences between Quechua and Spanish is that the latter has more separate verbal roots, while the former has fewer roots and tends to form new verbs through a complex system of verbal derivational suffixes. Thus we have examples like the following: (35)
QUECHUA
SPANISH
apa-mu-
llevar traer
"take" "bring"
Here the cislocative suffix -mu is added to the original root to create a cislocative verb with a meaning that is expressed with a separate verb in Spanish (and English). Sometimes the derived meaning is simply compositionally determined, and often the root + affix combination has undergone semantic specialization or drift. We may hypothesize that the lexicalized forms are replaced in Media Lengua by Spanish simplex roots, and the compositional forms by forms either directly patterned on the complex Quechua original or including both a Spanish root with the complex meaning and the Quechua derivational suffix. Muysken (1981b) presents data that illustrate this pattern for the Quechua verb riku- "see" and derived forms. The examples in (36)-(38) are directly related to the forms in (35). Sentences (36) and (37) illustrate the use of the Spanish-derived forms trayi- and lleba-, respectively: (36) intonsi lindu radiyu-da trayi-shka. then nice radio-ACC bring-SD "Then it turned out they'd brought a nice radio." (37) awa-da ahi-munda lleba-nga zin Ambatu-mun. water-ACC there-from take-3FUT they.say Ambato-to "From there they will take the water to Ambato, they say." In (38) we see an example of a redundant form: (38) grabadora-da trayi-mu-ngi. tape.recorder-ACC bring-CIS-2 "Bring the tape recorder." The interaction between relexification and derivational morphology needs to be studied in much more detail, for different groups of verbs. An interesting and complicated case involves the Media Lengua verbs kiri- "wish" and dizi- ~ zi- "want, say". Quechua has two verbs that express wishing and wanting:
390
Pieter Muysken (39) Q: - "wish" ni"want, say"
These verbs occur in constructions such as the following: (40) a. papa-da -ni. potato-ACC wish-1 b. papa-da ni-ni. miku-na -ni. eat-NOM wish-1 d. miku-sha ni-ni. eat-lFUT say-1
"I want potatoes" "I want potatoes" "I want to eat" "I say I'll eat", "I want to eat"
In Quechua both verbs, ni- and -, can take NP complements and infiniti val complements. In the latter case, ni- selects -sha on the infinitive verb, and - selects -na or another marker. In Media Lengua we find that (a) Quechua - has been relexified as Media Lengua kiri- "want" (Spanish querer), and Quechua ni- has been relexified as Media Lengua (di)zi- "wish, say" (Spanish decir); (b) with NP complements only Media Lengua kirioccurs; and (c) with infinitival complements kiri- often gets -na or another nominalizer, and (di)zi- often, but not always, gets -sha complements. Note that the relexification process has been only partial here. Whereas in Quechua the verb ni- can take NP complements, the corresponding Media Lengua item (di)zi- cannot. More details, including quantitative data, are given in Muysken (1981b). Relexification must necessarily take place on the basis of meaning correspondences. A Spanish stem, as close as possible in meaning to the Quechua original, is used to replace it. The question now is how closeness in meaning is determined. It turns out that we have to distinguish here between lexical meaning and grammatical meaning. The former can be determined by reference to some extralinguistic entity, the latter only by reference to the language systems themselves. Relexification is feasible, with all the compli cations mentioned, for lexical items, but it operates in a very incomplete manner for grammatical items. In order to study the differences between the restructuring of lexical meaning and the restructuring of grammatical meaning, we need to consider Quechua grammatical elements in more detail. As is shown in Figure 2, Quechua grammatical morphemes fall into three groups — lexical, suffixal, and clitic.
Media Lengua
391
Figure 2. Classification of the types of Quechua grammatical morphemes.
We encounter reflexes of all Quechua grammatical morpheme classes in Media Lengua. As we saw above, the Quechua clitics are all present, as are the Quechua inflectional suffixes. The Quechua lexical elements have all been relexified, in a way that I will discuss below. In addition, we encounter elements in Media Lengua which are absent in Quechua, such as prepositions and lexical conjunctions. They will be discussed in §3.4. Disregarding these last cases for the moment, it is possible to maintain the generalization that the inventory of Media Lengua categories of gram matical items directly reflects that of Quechua. As it turns out, however, the internal semantic organization of the categories does not; in fact, it is a system which has been restructured under the influence of Spanish. In Muysken (1988) I illustrated this for demonstratives, and in Muysken (1989) I tried to argue the same point for question words. Here I will briefly summarize these results. The semantic organizations of the systems of deictic (demonstrative and locative) pronouns in Quechua and Spanish diverge. In Figure 3, a much simplified version of the systems involved, the feature [± LOC] refers to the possibility that the element will be used as a locative
392
Pieter Muysken
Figure 3. The demonstrative systems of Quechua and Spanish.
deictic {here versus this). The feature [± PROX] refers to the distance from the speaker indicated by the deictic (this/here versus that/there). In short, the Quechua system does not distinguish the locative forms from the non-locative deictics lexically; instead, the distinction is made through case suffixes. The Media Lengua forms, morphologically based on Spanish deictic elements, do not derive directly from Quechua semantic equivalents. Nor do they immediately correspond to the Spanish system. The Quechua form kay "this, here" has been split into Media Lengua isti "this" (← Sp este "this") and aki "here" (← Sp aquí "here"), along the lines of the Spanish system, in which demonstratives and locatives are kept apart. The Media Lengua attributive use of aki "this ... here", as in aki kaza "this house (here)", is impossible in Spanish, however, though it is possible with Quechua kay. The form isti "this" does not occur with locative case -bi, while aki frequently does. Only aki has locative use. Both isti and aki can be marked for case. Of course the adding of
Media Lengua
393
a case marker to aki corresponds to the Quechua system, not to Spanish, where locative pronouns are used without prepositions and case markers do not exist. The fact that aki is used independently without a case marker corresponds to Spanish, not to Quechua usage. A few instances of Quechua kay are maintained in Media Lengua. The schematic examples in (41) and (42) illustrate the options in the Media Lengua system: (41) a.
ML: isti Sp: esta b. ML: aki Sp: *aquí
(42) a.
kaza casa kaza casa
"this house" "this house (here)"
ML: aki-bi sinta-ni here-LOC live-1 Q: kay-bi kawsa-ni
b. ML: aki sinta-ni here live-1 Q: *kay kawsa-ni
"I live here"
"I live here"
A similar pattern of compromise restructuring is found with the [-PROX] forms in Media Lengua, where there is an even greater departure from Quechua semantics in that the distinction between the Spanish locative deictics allí, allá, and ahí (indicating roughly different distances between the location referred to and the speaker) is introduced into Media Lengua. Media Lengua isi "that" does not co-occur with locative -bi, which is the most frequent affix with the forms allá, allí, and ahí. All three forms are also used attributively and optionally occur without case marking, unlike Quechua but like Spanish. This is most striking for ahí. Both the proximate and the non-proximate deictic elements reflect a system that is best interpreted as the result of restructuring the Quechua system under the influence of Spanish. The resulting system matches neither completely. I should also stress that it shows internal variation: almost one third of the independently used deictic locative elements occur without sup porting case-marking. Consider now the question words. In (43) the relevant forms are given for Quechua, Spanish and Media Lengua:
394
Pieter Muysken (43)
QUECHUA
SPANISH
MEDIA LENGUA
pi ima mayxan mashna
ima-munda
quién qué cuál cuánto a cómo cuándo qué horas por qué
ima-shna may+case
cómo (d)ónde
kin kilinki kwal kwantu a komo kwandu ki uras purki (purki-munda) komo undi+case
ima uras
"who" "what" "which" "how many" "for how much" "when" "at what time" "why" "how" "where"
These examples show that Media Lengua forms, with the exception of the form for "where", are modeled on Spanish rather than Quechua patterns. If it had been the reverse, we would have found forms such as *(in)ki-munda "what from" for "why" and *(in)ki-shna "what like" for "how", which do not occur at all. The form purki-munda, which follows both Quechua and Span ish, occurs only once. A third complex case of relexification of grammatical elements involves the Media Lengua personal pronoun system. Here relexification is much stricter. Consider the paradigms in the three languages: (44)
QUECHUA
MEDIA LENGUA
ñ kan
yo/ami+CASE bos
pay
el
ñukunchi kan-guna pay-guna
nustru bos-kuna el-kuna
SPANISH
1sg: yolme/mí 2sg: tu/te/tí vos/te usted/le 3sg: él/le ellalle nosotros/nos lpl: 2pl: ustedes/les 3pl: ellos/les ellas/les
"I" "you (intímate)" "you (familiar)" "you (polite)" "he" "she" "we" "you (pl.)" "they (masc)" "they (fem.)"
The Spanish column includes not only the nominative form but also the object clitic form and the non-nominative form where this differs from the nomina tive. When we compare the three paradigms, we see that the Media Lengua and Quechua systems are very similar, and that bos-kuna "you (pl.)" is a
Media Lengua
395
direct relexification of Quechua kan-guna (where the Spanish familiar sec ond-person pronoun vos is taken as the equivalent of Quechua kan). Simi larly, Media Lengua el- and Quechua pay-guna. The Media Lengua pronouns bos and el are a little more complicated, since either relexification of the Quechua model or target simplification — loss of distinctions existing in Spanish — could be involved here. The only case of non-relexification is the non-nominative Media Lengua personal pronoun ami, which was discussed above as a case of freezing, resulting from Spanish mí"(to) me". This pronoun occurs with the Quechua case markers -da "accusative" and -mu "dative", as in (45): (45) ML: ami-mu da-ngi me-to give-2 ñuka-mu ku-ngi Q: Sp: le darás a mí "give it to me" In the corpus there are some small deviations from the Media Lengua para digm given in (44), such as a few Quechua forms and the occasional use of miu (← Sp mío "mine") for the lsg and tuyu (← Sp tuyo "yours") for the 2sg. 3.4. Syntax What consequences did relexification have for the syntax of Quechua, lead ing to Media Lengua syntax? Remarkably, relexification had relatively few consequences, and most of them can be related to lexical innovation or restructuring. In this section I will discuss four sets of cases: word order (§3.4.1), comparatives (§3.4.2), reflexives (§3.4.3), and embedded WH-questions, complementizers, and conjunctions (§3.4.4). 3.4.1. Word order In some cases the subcategorization features of relexified items were not entirely retained from Quechua, but adopted from Spanish as well, which led to word-order changes in Media Lengua. (This is under the assumption that the position of adjectives, adpositions, and verbs is part of the lexical — subcategorization — features of the items themselves.) A case in point is borrowed prepositions. Prepositions are not very frequent in the sample — there are about 50, as against hundreds and hundreds of Quechua case-markers. All preposi-
396
Pieter Muysken
tions in Media Lengua are regular Spanish-derived prepositions except for the postposition despwesitu "after", which is relexified from the Quechua lexical postposition k'ipa, with the same meaning. The instances of entre in Media Lengua could be regarded as relexification of the Quechua postnominal element -pura, as in (46): (46) ML: Q: Sp:
entre seys-mi ga-nchi. sukta-pura-mi ga-nchi. six-among-AFF be-lpl somos seis. "There are six of us."
In some ways, Quechua -pura is distinct from the postposed elements which mark case in Quechua and which can best be analyzed as case-markers, not as elements of the category P. It may be one of the rare instances of a true postposition (P) in Quechua; if it is, relexification as entre would be expected. All other occurrences of Spanish prepositions, with the exception of despwesitu "after", will have to be considered as cases of code-switching, in exclamations such as pór dyos "please", or as true borrowings, which do not correspond to existing Quechua lexical entries. Another instance of direct relexification is given in (47): (47) sikyera karga-bu-lla-ish, p e r h a p s load-BEN-DIM-IND,
pero komi-nga-bu-lla-ish
da-chun,
but
give-SUB
eat-NOM-BEN-DIM-IND
sikyira para komir. perhaps for eat "Let them then give the food for a load, to eat, perhaps to eat." Here komi-nga-bu, which contains the Quechua nominalizer -nga and the Quechua case marker -bu, is juxtaposed to para komir, which contains the Spanish preposition para and the Spanish infinitive marker -r. In (48) we see a formal representation of the relexification of the Spanish preposition entre: (48) + P N _ sukta-pura-mi six among-AFF
+P _ N entre seys-mi among six-AFF
Ex. (49) shows how a prenominal adjective in Quechua is relexified as a postnominal adjective in Media Lengua:
Media Lengua
397
On the whole, Media Lengua has maintained the predominantly SOV word-order characteristics of Quechua. It is true that 21% of the sentences in the sample which contain both a verb and a complement in the VP (object, adverbial complement) show VX word order, but this result parallels that for local varieties of Quechua. 3.4.2. Comparatives In Quechua, comparatives are formed with the uninflected verb yalli "sur pass", which functions as a serial verb (Muysken 1977). The object with which something is compared receives -da "accusative case": (50) Q: kan Huzi-da yalli puri-ngi. you José-CC pass walk-2 "You walk faster than José." In Media Lengua yalli is relexified as gana- (← Sp ganar "win"), but it cannot appear uninflected: (51) ML:
*bos Huzi-da gana anda-ngi. you José-ACC win walk-2
Instead, gana- must appear either in adverbial subordination with the -sha marker or in a coördinated clause: (52) ML:
Xwan-mi Pedro-da gana-sha John-AFF Peter-ACC win-suB "John is taller than Peter."
grande ga-n. tall be-3
(53) ML:
Takunga-mi riko ga-n Salsedo-da gana-n. Latac.-AFF rich be-3 Salcedo-ACC win-3 "Latacunga is richer than Salcedo."
The impossibility of using uninflected gana- in Media Lengua comparatives suggests that in some cases relexification can involve the simplification of a lexical entry. In Quechua yalli must be marked to indicate that person mark-
398
Pieter Muysken
ing on it is optional (in contrast with all other Quechua verbs). The Media Lengua verb is not marked in this way. 3.4.3. Reflexives Quechua reflexives do not involve a reflexive pronoun, unlike Spanish reflex ives. Instead, the affix combination -lla-di "just, precisely" (lit. "delimitativeemphatic") is added to the subject NP: (54) Q: ñuka-lla-di riku-ni, I-just see-1 "I see myself." When, in the 3rd person, -lla-di is added to the adverb shina "thus", the resulting meaning is reflexive: (55) Q: pay shina-lla-di riku-n. he thus-just see-3 "He sees himself." In Media Lengua, 3rd-person reflexives are formed as in Quechua, with shina being relexified as asi "thus" (from Spanish así), but 1st-person reflexives involve a double-pronoun construction: (56) ML:
Huzi asi-lla-di-mi mata-ka. José thus-just-AFF kill-PAST "José killed himself."
(57) ML:
yo-lla-di bi-xu-ni ami-lla-da-di. I-just see-PROG-1 me-just-ACC "I see myself."
This construction is limited to the 1st person because only the 1st person has a separate non-nominative pronoun, ami. While the resulting Media Lengua double-pronoun construction could of course be interpreted as an adaptation to Spanish syntax (cf. 58 below), this is by no means the necessary conclusion. In Spanish the reflexive pronoun is an unstressed clitic in preverbal position. The Media Lengua reflexive element is a strong form which also occurs in nonreflexive contexts, as in (45). Of course we do encounter stressed postverbal pronouns in Spanish emphatic reflex ives:
Media Lengua
399
(58) Sp: yo me veo a mí. "I see myself." In any case, the difference between Quechua and Media Lengua is caused by a lexical element, ami, which is present in Media Lengua but absent in its source language. 3.4.4. Embedded wh-questions, complementizers, and conjunctions In Quechua, embedded WH-questions are formed by fronting a WH-pronoun, as are nonembedded ones; but in addition they are nominalized, as are other embedded clauses. Compare (59)a with (59)b and (60)a with (60)b: (59) a.
b.
(60) a.
b.
Q: mana yacha-ni-chu [Xwan shamu-shka-da]. not know-1-NEG [John come-NOM-ACc] "I don't know that John has come." Q: mana yacha-ni-chu [pi shamu-shka-da]. know-1-NEG [who come-NOM-ACc] not "I don't know who has come." Q: mana yacha-ni-chu [Xwan shamu-na-da]. not know-1-NEG [John come-NOM-ACC] "I don't know that John will come." Q: mana yacha-ni-chu [pi shamu-na-da]. know-1-NEG [who come-NOM-ACc] not "I don't know who will come."
Whereas the nominalizers in (59) and (60) are part of the verb morphology, they are generally assumed to function as complementizers and are subcategorized by the matrix verb. The WH-pronoun pi "who", in (59)b and (60)b, is fronted, but it does not appear as the complementizer. In Media Lengua the situation is different. We encounter the -na and -shka nominalizers in sentential complements, but in embedded WH-questions there is ordinary verb inflection: (61) a.
ML:
b.
ML:
c.
ML:
no not no not *no not
sabi-ni-chu know-1-NEG sabi-ni-chu know-1-NEG sabi-ni-chu know-1-NEG
[Xwan bini-shka-da]. [John come-NOM-ACC] [kin bini-rka]. [who come-PAST] [kin bini-shka-da]. [who come-NOM-ACC
400
Pieter Muysken
The contrast between (61) and (59) suggests that in Media Lengua (where embedded WH-questions have the ordinary past-tense marker -rka instead of the nominalizer -shka) the WH-element does function as the complementizer. Similarly, the Media Lengua equivalent of (60)b will be (62)a and not (62)b: (62) a.
b.
ML:
no sabi-ni-chu [kin not know-1-NEG [who "I don't know who will ML: *no sabi-ni-chu [kin not know-1-NEG [who
bini-nga]. come-3FUT] come." bini-na-da]. come-NOM-ACC]
It is tempting to interpret this development as the beginning of a typological shift from a -final system (like Quechua) to a -initial one (like Spanish). The problem is that few or no other cases of clause-initial com plementizers occur in Media Lengua. Admittedly, there are some occurrences of clause-initial ki "that", porke "because", and aunke "although". These three cases are somewhat different and need to be analyzed separately. The "that" complementizer ki occurs as an alternative to /-shka-/ and /-na-/ complementation. Consider (63): (63) ML: Sp:
no be ki no i-sha-chu dizi-n? not see that not go-lFUT-NEG say-3 no ve que dice que no irá? "Don't you see he does not want to go?"
Here ki is part of the fixed expression no be ki, directly borrowed from Spanish. Note that be "you see" does not receive Quechua person marking, whereas dizi-n, the verb in the domain of ki, receives main-clause person marking. The next example, (64), contains an example of a nominalized -shka complement: (64) pega-shka dizi-n dizi-ndu ke bos-kuna-ga ke that hit-NOM say-3 say-suB that you-PL- bos-kuna-lla-cha komi-ngichi. you-pl-DIM-DUB eat-2pl "They say he hit them, saying that you, that you just eat." Here again komi-ngichi "you (pl.) eat", the verb in the domain of ke (=ki), receives ordinary main-clause person marking. The same utterance contains an example of -shka complementation. The compound P+COMP porke "be-
Media Lengua
401
cause" occurs both with main-clause person-marked verbs, as in (65), and with clauses marked with an adverbial subordinator, as in (66): (65) todabia no byen aprendi-naku-n porke eskwela-bi still not well learn-pl-3 because school-LOC anda-naku-n. go-pl-3 "They don't learn well yet because they go to school." (66) porke no awa abi-kpi no kosecha-nchi. because not water be-suB not harvest-lpl "Because there is no water we don't harvest." In my Media Lengua sample, aunke only occurs with subordinate verbs, as in (67): (67) aunke duru llubi-xu-kpi-sh sali-gi-xu-ni-mi. though hard rain-PROG-suB-IND go-INC-PROG-1-AFF "Although it is raining hard I am going to go out." Regarding the introduction into Media Lengua of the clause-initial comple mentizer ki and its compound forms, we can conclude that they are relatively infrequent; that they sometimes introduce main-clause person-marked verbs, sometimes subordinate verbs; and that they sometimes appear to be direct borrowings from Spanish, rather than replacements of Quechua categories. On the whole, the Quechua system of nominal and adverbial marking on verbs to indicate subordination is the predominant type of complementation in Media Lengua. Possibly at a later stage the language will develop a regular clause-initial COMP position. With embedded WH-questions and complementizers a large-scale syntac tic change may be going on, triggered by the adoption of the Spanish question system as a complementizer system. The other instances of syntactic differ ences between Quechua and Media Lengua are lexically induced as well. The claim that Media Lengua is a form of Quechua with Spanish vocabulary can be maintained as far as these items are concerned, but the sporadic adoption of certain sets of items has triggered changes. The case of coördinating conjunctions is more problematic for the relexification analysis. Media Lengua has productively adopted both the Spanish and the Quechua system of conjunctions (more details in Muysken 1981b). In Quechua, conjunctions such as -pish "also" and -ti(k) "emphatic" are always
402
Pieter Muysken
cliticized to the element on their left; they are arguably generated as a separate phrase-structure category (unlike case-markers). In Spanish we find conjunctions such as y "and" and "or" as a separate lexical category. Are the Spanish conjunctions that we find in Media Lengua cases of reflexification of Quechua categories, or cases of direct borrowing from Spanish? In Media Lengua the conjunctions are used as in Spanish and coexist with the Quechua cliticized conjunctions; furthermore, in Quechua itself there is frequent bor rowing of Spanish conjunctions. Thus, the introduction of Spanish conjunc tions into Media Lengua is the one exception to the generalization that Media Lengua lexical categories, arising through relexification, correspond strictly to Quechua categories. This exception may be explained by the fact that coördinating clausal conjunctions, which occur at the discourse level, are less closely integrated into the grammar of the language and can be borrowed more easily. The category of complementizers, for instance, is a much more crucial part of sentence grammar. 3.5. Summary To summarize the analysis in this section, we can draw the following conclu sions from the case of Media Lengua: (a) The different components of a lexical entry function so independently of each other that (apparently) a phonological representation can be substi tuted in an entry without affecting the other sets of features (syntactic, subcategorization, semantic, selectional). (b) For relexification to occur, the only requirement is that source- and target-language lexical entries share some semantic features; other shared features are not required, although they will often be present. (c) A language which emerged through relexification has the same lexical categories as its source. The morphosyntactic and syntactic categories, which are expressed in Quechua through affixation, have been maintained in Media Lengua. (d) There is a considerable difference between the relexification of content words and the relexification of function words. In the former we saw straightforward relexification, generally maintaining the Quechua semantic distinctions. In the latter case there was drastic restructuring of the system. How can we explain this difference? Relexification can only take place under conditions of semantic resemblance. When the Quechua verb riku-
Media Lengua
403
"see" is relexified as bi- (from Spanish ver), this is possible because there is a large shared element of meaning, established through reference to some extralinguistic mental representation. Now consider the sets of function words discussed in §3.3. These do not have a meaning outside the linguistic system that they are part of, since their meanings are paradigmatically defined within that linguistic system. So when you relexify a system of paradigmatically organized function words, the semantic organization of the target language automatically comes in, and the result is at best a compromise between source- and target-language systems.
4.
The status of Media Lengua as a contact variety
The status of Media Lengua will be approached from two perspectives. First, to what extent is it different from other linguistic results of contact between the Spanish and Quechua cultures in the Andes? And second, to what extent is Media Lengua really a separate and coherent language system in its own right, a discrete system with its own rules and patterns? I will discuss each of these perspectives in turn. It is at least conceivable that Media Lengua evolved from an earlier contact vernacular that was used between the highlanders and the colonizers, at a time when the highland groups themselves were not completely Quechuaspeaking. We have no record of such a contact vernacular, much less of its linguistic features. What we do have is evidence that there were and are Spanish-based pidgins spoken in the Western Amazon basin and also by the Cayapa on the Pacific coast in Northern Ecuador. Simson (1886), in a travel account, and Gnerre (1975) have provided some data on varieties of this pidgin. Sample sentences are given in (68): (68)
AMAZON PIDGIN:
a.
ese canoa andando Consacunti cuando será llegando that canoe going Consacunti when will.be arriving ese Tonantins tiene ? that Tonantins has. got b. así luna será tiene. thus moon will.be has.got c. ese Consacunti, Carapaná llegando, más lejos será that Consacunti Carapaná arriving farther will.be
404
Pieter Muysken tiene ese Carapaná, Tonantins llegando? has.got that Carapaná Tonantins arriving d. sí, más lejos tiene. has.got yes farther e. ese Consacunti saliendo luna asi tiene, Carapaná that Consacunti leaving moon thus has.got Carapaná llegando luna donde será tiene. arriving moon where will.be has.got f. no será así tiene? not will.be thus has.got g. tuyo no sabiendo leña cortando. yours not knowing wood cutting h. tuyo chacra cortando no sabiendo tiene, mujer no yours land cutting not knowing has.got woman not será cojiendo. will.be taking i. tuyo tabaco fumando no queriendo tiene. you tobacco smoking not wanting has.got"
Features that can be identified on the basis of Gnerre's and Simson's pidgin data are that the verb consists of the gerundive and está "to be, to be characterized by". In Simson's data it is not the verb estar that is overgeneralized in its use, but rather the future of the verb ser, i.e. será lit. "he will be". There is SOV word order, with the main-sentence verb following the subordinate-sentence verb (consistent with SOV basic order). Prepositions tend to be absent. We find the generalized use of Spanish tiene "has got", probably as an emphatic marker. The Spanish demonstrative pronoun ese "that" is used as definite marker. While in Gnerre's data the second person pronoun is vos "you (informal)", Simson has tuyo "yours (intimate)". Media Lengua shares both the word-order patterns and the use of vos and tuyo for the second person with this Amazon pidgin. In addition, there are some occurrences of Spanish gerunds in Media Lengua. There, however, they occur only in adverbial subordinates, while in the Amazon pidgin they are characteristic of main clauses. For the rest, of course, Media Lengua is characterized by abundant Quechua morphology, which is absent in the Amazon pidgin. We can only conclude that, whatever the connection may be between Media Lengua and a possible earlier highland pidgin (assuming that the latter resembled modern Amazon pidgin), linguistically Media Lengua did not derive its features from the earlier pidgin.
Media Lengua
405
A second possibility is that Media Lengua is the result of the process of acquisition of Spanish by Quechua speakers, a type of fossilized interlanguage. This possibility can be ruled out, I think. Consider second-language acquisition data from a young cargador (load bearer) who came to Salcedo from a nearby monolingual Quechua area higher up in the mountains and only recently started learning Spanish: (69)
QUECHUA-SPANISH INTERLANGUAGE:
a. Fragment 1. Yo no, aứno sabemos castellas tan I not not.yet we.know Spanish also ahí-ca barrio chiquito. Así trabajando me voy, there- small village thus working I go no podiendo trabajar, mano todito dentró no not able to.work hand wholly entered just más espino, [de goto] también dentra no.más. thorn... also enters just Por. eso tan no aguantamos mi soplido no.más therefore also not we.stand my advance just cogiendo me voy no.más mi cobrado mi pago también. taking I go just my income my pay also Ahí mismo está pago. right there is pay b. Fragment 2. Aquí-ca buscando carga no.más es. Nada.más here- seeking load just is nothing tan. Buscando carga, encontramos, cargando, ir, no else seeking load we.find carrying go not encontramos, no ir. Sentamos. Así no.más es. we.find not go we.sit like.that just is Segúndo que cuadra indo, caminando. Cuatro cuadra according to block going walking four block juindo, cuatro sucres mismo. Cinco cuadra jué, cinco going four sucres exactly five block went five sucres mismo. Así, así cojo. Así-ca día.lunis saco sucres exactly thus thus I.get thus- Monday I.get veinte sucre situ. Día.martes saco quince sucres. twenty sucres Tuesday I.get fifteen sucres c. Fragment 3. Todo indo a Machachi yo solo quedando all going to Machachi I only staying
406
Pieter Muysken cocinando comemos no.más, no tengo. Papa tan sí cooking we.eat just not I.have Potato also yes sabo pelar. SI? Sí sabe pelar, todo sí sabe. YES? Yes knows peel all yes knows I.know peel Sí sabe. Papa también pelo no.más. Machachi I.peel just Machachi yes knows potato also indo a la feria día.dúmingo quedamos en la casa going to the fair Sunday we.stay in the house en la carpa quedamos cuidando, quedamos-ca lo.que in the tent we.stay watching we.stay- what da la gana come hago. Qué arrosicu, qué papa comes to mind eats I.make what rice what potato frito, qué.cosa, yo solo no.más. fried whatever I alone just
A detailed analysis of the speech of this and other beginning secondlanguage learners yields a number of characteristic interlanguage features. Of the relevant sentences, 68% have SXV word order (as in Quechua), and 100% of the few adjective-noun combinations have noun-adjective order (as in Spanish); we encounter 26 cases of possessed NP de possessor NP, as in vecino de Francisco "neighbor of Francisco". In addition, there is one case of possessor/possessed order (preferred in Quechua): Francisco casa "Fran cisco house". Thirty-six percent (39 out of 109 cases) of the prepositions obligatory in Spanish have been deleted. Most instances concern locative en "in" (16 out of 33 cases) and directional a "to" (17 out of 22 cases). This tendency diminishes quickly for more proficient bilinguals; for one slightly more advanced speaker they are deleted only in 4 out of 60 cases, and for a relatively fluent speaker from the same group of cargadores apreposition is deleted only once out of 123 cases. Plurals are only infrequently marked; note that in Quechua, in contrast to Spanish, plural marking is optional. Definite and indefinite articles are fre quently not present where they would be expected in regional standard Spanish contexts. This feature, while needing much more study, appears to be a general characteristic of vernacular Spanish in the area. We encounter very frequent use of gerunds. Verb marking is highly irregular: of clear lsg contexts, there are 49 cases of -o (the target marker), 22 cases of -al-e (third person in the target), 28 cases of -mos (lpl in the target), and 9 irregular forms. Of all the verbs, 31% have a correct present-tense form,
Media Lengua
407
20% have an incorrect present-tense form, 16% occur in the gerund form, and 12% occur in the infinitive. While other speakers also show some irregular verb marking, this is not a feature of nonstandard Spanish, and it disappears rapidly with more advanced speakers. No Quechua morphology or lexicon is directly evident, with the excep tion of the Quechua topic marker -ca or -ga, as in ahí-ca barrio chiquito "there-TOP village little" and in quedamos-ca lo que da la gana come hago "we.stay whatever comes to mind he.eats I.make". An enclitic tan (from Spanish también "also") occurs as an emphatic negative marker or as an indefinite marker; as in nada más-tan "nothing else". As with the topic marker -ca, this phenomenon is not limited to bilinguals, but it certainly is stigmatized as if it were. Generally there is clause juxtaposition (so that the relations between the propositional contents of the different clauses must be inferred) instead of clause subordination. Memorized chunks — fixed expressions from Spanish — are used, which sometimes creates the false impression of syntactic complexity: según lo que cuadra indo "according to that block going". There is overgeneralized use of familiar vos as a 2sg pronoun. We also find a highly irregular use of clitics, and particularly the absence of thirdperson object clitics such as le. Clearly, Media Lengua is very different from this type of QuechuaSpanish interlanguage. The interlanguage lacks the Quechua morphology and elaborate syntax characteristic of Media Lengua, and it is characterized by a paratactic system of phrase formation and a very rudimentary form of Spanish morphology. Moreover, most Media Lengua speakers are bilingual in Media Lengua and some form of fossilized Quechua-Spanish interlanguage (gener ally more advanced than that of the young cargador illustrated above). We can only conclude that Media Lengua is unlike both the contact pidgin of the Amazon and Quechua-Spanish interlanguage. Regarding the "languageness" of Media Lengua, we must next ask to what extent Media Lengua is a coherent system. As I stressed before, it is an intracommunity language and is not used with outsiders, and therefore it (and other varieties of Media Lengua spoken elsewhere, discussed below in §5) could have escaped notice. Nonetheless, the language itself is clearly per ceived by its speakers as a separate system, related to Quechua. In one recording, a (trilingual) speaker started giving the following equivalences:
408
Pieter Muysken (70) Entonsis ahi-munda-ga bulta asi "Breve bini-ngi" ,en Media Then there-from-TOP again thus quick come-2 in Media Lengua-bi. Entonsis Ingichu-bi-ga "Ukta shamu-ngi", en Lengua-LOC then Quechua-LOC- quick come-2 in castellano-bi-ga "Vendrásbreve". Ahi-munda por ejemplo Spanish-LOC-TOP come.2 quick there-from for instance nosotros zi-nchi: "Mañana pweblo-mun i-na-mi ga-ni. " we say-lpl Tomorrow town-to go-NOM-AFF be-1 Ahi dis ingichu-bi-ga: "Ñuka kaya llakta-mun there say Quechua-LOC- I tomorrow town-to ri-na-mi ga-ni". Castellano-bi-ga "Mañana tengo que go-NOM-AFF be-1 Spanish-LOC-TOP Tomorrow have.l to ir a Salcedo, o a pueblo. " go to Salcedo or to town. "Thus therefore again "Come quickly", in Media Lengua. Then in Quechua ..., and in Spanish .... Then we say for instance "Tomorrow I have to go to town". Then in Quechua say ..., and in Spanish ...."
This type of metalinguistic commentary, which other speakers were also capable of, suggests that Media Lengua is perceived as a clearly separate and discrete system. A final relevant question concerns the amount of variation in the lan guage. I elicited material on Media Lengua with one family in 1975, and then I recorded material with that family and with another household in 1978, for a total of three samples. The sample CF represents data elicited from a 30-yearold couple, both native speakers of Media Lengua; sample CI is conversa tional data from these same speakers and their children, recorded two years later; and sample MI is conversational data from a female 37-year-old native speaker of Media Lengua and two nonnative speakers of Media Lengua. Sample CF clearly represents a conservative norm for Media Lengua, while samples CI and MI show cases of Media Lengua-Sp code-switching, as well as innovations within Media Lengua. Still, the three samples show a remark able degree of uniformity in essential respects, such as the amount of Spanish vocabulary present. Some of the relevant figures are given in Table 3. Most features have been discussed above. The percentage of instances where -chu is absent has to do with the fact that in Quechua and Media Lengua negation is roughly like the French ne pas construction: in addition to mana "not" the
Media Lengua
409
clause contains a negative particle. In Quechua it is sometimes omitted, and in Media Lengua it is omitted more frequently. Particularly the CI and the MI samples show rather similar patterns. The CF sample was elicited, and shows less influence from Spanish (e.g. in the occurrence of -ndu and in the form of the locative deictics), more careful speech (e.g. in the presence of -chu), and more complex morphosyntax (e.g. in the number of derived verb forms). Thus Media Lengua is not only perceived as a separate system, it has its own regularities as well.
Table 3. Comparison of three samples of Media Lengua for a number of variable features. CF = elicitations with family A; CI = informal recordings with family A; MI = informal recordings with family B.
% %
%
%
%
%
of Quechua verb tokens (the number of Quechua retentions) of derived verbal forms (amount of Quechua derivational morphology present) of -ndu compared to total of adverbial subordinated forms (replacement of the Quechua subordinators -kpi and -sha', cf. §3.3.) of locative deictics without case-marking (adaptation to Spanish lexical semantics; cf. §3.3.) of negative main clauses without the negative particle -chu (characteristic of unmonitored speech, and possibly also of Spanish influence) of VX orders in clauses with an object (possibly Spanish influence)
CF 12
CI 10
MI 13
19
9
8
45
67
65
0
26
32
28
69
68
18
24
22
410 5.
Pieter Muysken Other varieties of Media Lengua
It is very difficult to determine how many communities have developed forms of Media Lengua, and whether the phenomenon is limited to Ecuador. In fact, the only published material on varieties of Media Lengua is my own work. Why isn't more known about this very interesting phenomenon? First of all, very little empirical research of any kind on day-to-day speech in Indian communities has been carried out in Ecuador. Most linguists have gathered traditional folk narratives, or they have elicited judgments, or they have trained informants in Bible translation. Consequently, what they elicited was hatun ingichu, big Quechua. Second, Media Lengua is very much an in-group language; it sounds like Quechua when you hear people speak but don't quite hear what they are saying. I discovered it only by accident, even though I had been gathering Quechua data in the Cotopaxi area for months. Third, most linguistic work has focused on Indian groups with strongly developed tribal identities, such as the Otavalo and the Salasaca Indians. It is less likely that these groups would relexify their Quechua, and as far as I know they have not done so. Nonetheless, I discovered two other varieties similar to the type of Media Lengua described above in §3: Saraguro (Loja) and Catalangu, spoken out side of Cañar. I stumbled on both of these while doing dialectological survey work. By that time, of course, I knew what to look for. For both varieties I have only elicited data. I will discuss them separately and will then turn to a comparison of all three varieties and to the issue of whether they all emerged separately. 5.1. Saraguro Media Lengua The Saraguro Media Lengua data were gathered from a group of young adolescents, not from a community near the Spanish-speaking regional cen ter, but fairly high in the mountains. This difference from the Salcedo area may reflect the greater degree of hispanization of the Saraguro communities. Apparently the line between the Spanish-speaking and the Quechua-speaking worlds, where Media Lengua flourishes, lies farther away from the urban center. Examples are given in (71):
Media Lengua
411
(71) a.
ML: donde-manta bini-ku-ngi? Q: may-manta shamu-ku-ngi? where-from come-PROG-2 "Where are you coming from?" b. ML: miu-ta awa-naya-ku-n. Q: -ta yaku-naya-ku-n. I-ACC
c.
d.
e.
f.
g.
Water-DES-PROG-3
"I am thirsty." ML: miu-kuna aki-pi bibi-nchi. kawsa-nchi. Q: ñukanchi kay-pi lpl here-LOC live-lpl "We live here." ML: anda-sha kiri-ni Kwenka-man. Q: ri-sha -ni Kwenka-man. go-1FUT want-1 Cuenca-to "I want to go to Cuenca." ML: undi-pi puni-rka-ngi miu punchu-ta? Q: may-pi chura-rka-ngi ñ punchu-ta? where-LOC put-PAST-2 I poncho-ACC "Where did you put my poncho?" ML: si masiado llubi-kpi no anda-sha-chu. Q: yalli tamia-kpi mana ri-sha-chu. if too.much rain-suB not go-lFUT-NEG "If it rains too much I won't go." ML: Xwan sabi-n uste asi-shka-ta-kari. Q: Xwan yacha-n kan rura-shka-ta(-kari). Johnknow-3 you do-NOM-ACC-EMP "John knows what you have done."
We find a number of ways in which the grammar and lexicon of Saraguro Media Lengua differ from those of Quechua. The differences include wordorder features, possessive marking, conditionals, obligation marking, WHwords, and personal pronouns. In Saraguro Media Lengua a sizable number of sentences show SVX order, while this is less frequent in Quechua. Without detailed quantitative study of both varieties in the Saraguro area, however, it is difficult to say how different they are in word-order patterns.
412
Pieter Muysken
All Quechua possessors, with the exception of lsg and lpl, must be marked with genitive -pa. In Saraguro Media Lengua, by contrast, it is always possible to delete -pa, except with the 3sg and the plural pronouns. (72) ML: Manuel-pa kasa Manuel kasa el-pa kasa * el kasa
"Manuel's house" "his house"
In Quechua conditional clauses are not distinguished from temporal clauses; both are formed with -shpa "proximate subordinator" or -kpi "obviative subordinator". In Saraguro Media Lengua the optional Spanish condi tional si " i f may be added in addition to the Quechua suffix: (73) ML:
si masiado llubi-kpi no
anda-sha-chu.
if tOO.mUCh rain-SUB not gO-lFUT-NEG
"If it rains too much, I won't go." Quechua has an obligation construction in which the main verb appears with the nominalizing suffix -na, followed by the copula: (74) Q:
ri-
ga-ni.
go-NOM be-1
"I have to go." (lit. "I am to go.") In Saraguro Media Lengua, instead of the copula the verb tini- "have" appears in an otherwise similar construction: (75) ML:
miu anda- tini-ni. I go-NOM have-1 "I have to go."
It is quite possible that this construction emerged in Saraguro Media Lengua under the influence of the Spanish construction tener que "have to". In Quechua there is a difference between human and nonhuman, and between independent and attributive, in the inventory of the WH-words. In Saraguro Media Lengua, by contrast, there is only one generalized nonlocative WH-word, kin, derived from Spanish quién "who": (76) a.
kin-ta
asi-ku-ngi?
what-ACC do-PROG-2
"What are you doing?"
Media Lengua b.
c.
413 kin kebra-rka bentana-ta? who break-PAST window-ACC "Who broke the window?" kin kuchillu-wan uste pan-ta korta-rka-ngi? which knife-with you bread-ACC cut-PAST-2s "With which knife did you cut the bread?"
In Quechua, nonemphatic pronouns are often deleted. This does not appear to be the case in Saraguro Media Lengua. Again, a more detailed study is needed, with recorded conversational data. Other features of Saraguro Media Lengua will be discussed in §4.3 below. 5.2. Catalangu Catalangu is in many ways both formally and sociolinguistically similar to Saraguro Media Lengua. It is spoken in the Cañar province by the Indians who live close to the provincial urban centers. The present study, which can be considered as no more than exploratory, was conducted through elicitation from two families and a group of adolescents outside the town of Cañar. The two most likely derivations of the name "Catalangu" are given in (77). (77) castellano → catalangu castilla lengua (Quechua: castilla shimi) → catalangu Both these derivations suggest that Catalangu is a kind of Spanish. It is much closer to Spanish than the two documented varieties of Media Lengua are. Some sample Catalangu sentences are given in (78): (78) a.
a-kin-ta-pi
buska-ri-ndu?
ACC-who-ACC-EMP Seek-INC-GER
"Who are you looking for?" b. ami-ta xwin ambri-naya-ndu. me-ACC very hungry-DES-GER "I am very hungry." c. tuy amigu-ruku. I your friend-EMP "I am your friend."
414
Pieter Muysken d. ellos-kuna Sigsihuaicu-manta es. they-pl Sigsihuaicu-from is "They are from Sigsihuaicu." e. kumida-ka sabruzu sidu. been food- good "The food turns out to be very good." f. kasa-pi-ka nuway nadin-pis. house-LOC-TOP there.is.not nobody-EMP "There is nobody in the house." g. yu-ka nu pudi-gu asi-r isti-ta-ka. I- not can-1 do-INF this-ACC- "I can't do this."
I will discuss the process of verb regularization and the pronominal system in the next section. Here I will briefly mention a few other characteristics of Catalangu. Verb marking is highly variable. Twenty-eight percent of the verbs in my sample received the -ndo gerundive ending, 49% received more or less regular Spanish endings, and 22% had newly-formed endings. The last cat egory is the most interesting. We find four cases of a lsg ending -gu: (79) a. yu-ka Pedro llama-gu. I- Peter call-1 "I am called Peter." b. yu-ka nu kiri-gu. I-TOP not
want-1
"I don't want it." This ending probably derives from the Spanish irregular ending -go in tengo "I have" (inf. tener), traigo "I bring" (inf. traer), caigo "I fall" (inf. caer), and vengo "I come" (inf. venir), but it is used with verbs that have other endings in Spanish. There is some evidence that Catalangu has developed an aspect system with four distinctions: (80) V-ndu ista V-ndu V-du tara W-ndu
unmarked present progressive aspect perfective aspect potential aspect
Media Lengua
415
The progressive aspect appears in sentences such as those in (81): (81) a.
ML: ki ista-s asi-ri-ndu? what be-2 do-INC-GER "What are you doing?" b. ML: ki asi-n-ta-s-pi? what do-ASP-2- "What are you doing?"
The variant in (81)b is more similar to the Quechua equivalent: (82) Q: ima-ta rura-ku-ngi? what-ACC do-ASP-2 "What are you doing?" The perfective aspect is illustrated in (83) and the potential in (84): (83) ML:
Manuel kurta-du-pis la manu. Manuel cut-ASP-EMP the hand "Manuel has cut his hand."
(84) ML:
kin tara bini-ndu? who POT come-GER "Who might be coming?"
Almost half of the nonsubject NPs in my sample have no case marking. The Quechua accusative case suffix -ta is deleted most often, followed by locative case suffix -pi. Only two examples of the use of Spanish prepositions occur in my data, both involving a "to" (ACC): (85) a. Juan-ka parisi-0 a su papa-pi. John- resemble-3 to his father-EMP "John looks like his father." b. miu-pa tiyu isu-pi mata-r a-l wagra. I-GEN uncle had-EMP kill-INF ACC-DET bull "My uncle had the bull killed." In most nonaccusative contexts, Quechua case marking appears. The Quechua plural marker - is also found. Besides -ka, a topic marker which also occurs in Quechua and in rural Spanish, Catalangu has two other enclitic clausal markers, -pi{s) and -ruku. The suffix -ruku is derived from Quechua, where it occurs both as an adjective suffix meaning "old (of men)" and as an augmentative nominal suffix:
416
Pieter Muysken (86) Q: allku-ruku dog-big "big dog"
In Catalangu it occurs as a degree marker and as an emphatic affirmative marker, in both cases translatable as "indeed": (87) a.
llubindu-ruku ista. raining-indeed is "It is really raining." b. yu-ka tu amigu-ruku. I-TOP your friend-indeed "I am really your friend."
The suffix -pi resembles Quechua locative -pi, but it is probably derived instead from Spanish pues "well, then". Its use and placement need to be investigated further; examples are given in (88): (88) a. yu-ka awa-pi kai-pis. I-TOP water-LOC fell.l- "I fell into the water." b. kasa-pi-ka nuway nadin-pis. house-LOC-TOP there.is.not nobody-EMP "There is nobody at home." Luchu-ka asi-pi enseñar-pi bin. Lucho- make-EMP teach-EMP well "Lucho teaches well." We can conclude our discussion of Catalangu by noting its partial resem blance both to Quechua-Spanish interlanguage (e.g. in the reduction of Span ish verbal morphology) and to the varieties of Media Lengua (e.g. in the use of some Quechua particles and suffixes), in addition to its many original features. 5.3. A comparison Did the Media Lengua varieties all emerge separately, or are they the reflexes of a single earlier linguistic system that was once spoken throughout the Ecuadorian Andes and is now fragmented? The three varieties discussed above are many hundreds of kilometers apart. The way they have regularized
Media Lengua
All
Spanish verbs, presented in Table 4, could be evidence for or against a common origin. This regularization, due either to the fact that all verbs are regular in Quechua or to a process of formal simplification accompanying the relexification, has produced some interesting differences among the three varieties. Particularly striking is the substitution of anda- (Sp. "walk") for ir in two varieties. Like Quechua, Media Lengua has a number of impersonal verbs with animate objects. While they are based on Spanish vocabulary, they have no direct Spanish equivalent. Table 5 gives these verbs in the three varieties. This type of modelling on Quechua lexico-syntactic frames, using Spanish vo cabulary, is the dominant process in the formation of all varieties of Media Lengua. However, the way the modelling has been achieved differs in the three cases. Another test case is the pronominal system. For reasons of brevity, I will discuss only the personal pronouns here, presenting the Quechua system together with that of the varieties of Media Lengua, Catalangu, and Spanish (Table 6). Table 4. Regularized Spanish verbs in the three varieties of Media Lengua. ML-SARAGURO
ML-SALCEDO
CATALANGU
SPANISH
andabamuchi dalibiaasiriidentradizibenditu-
ibamuchi dabiaziridintradizibindizi-
andabamuchi da-/dalibiasiriyi-
ir 'go' vámonos "let's go" dar "give" ver "see" hacer "do/make" reir "laugh" entrar "enter" decir "say" bendecir "bless"
Table 5. Reflexes of the Quechua impersonal verbs in the three varieties. QUECHUA: ML-SARAGURO: ML-SALCEDO: ML-CATALANGU: SPANISH:
yarkaambriambri-na(ya)ambri-nayadar hambre "feel hungry"
chirifriachirifriyihacer frío "be cold"
yaku-nayaawa-nayaawa-nayatener sed (de agua) "feel thirsty (for water)"
418
Pieter Muysken
Table 6. Personal pronouns in Quechua, the three varieties, and Spanish.
1sg: 2sg: 3sg: lpl:
QUECHUA
ML-SARAGURO
ML-SALCEDO CATALANGU SPANISH
ñuka kan pay ñukanchik
miu/yo ústi el miukuna
yo bos el nustru
ústikuna elkuna isikuna
boskuna elkuna
2pl: kankuna 3pl: paykuna
yu ústi isti nustru nustrukuna ustidis elloskuna
GLOSS
yo vos/tu/usted el/ella
"I" "you" "he/she"
nosotros ustedes ellos/ellas
"we" "you" "they"
The gender distinctions in the 3sg and 3pl and the politeness distinction in the 2sg, which are characteristic of the Spanish system, have disappeared in the varieties of Media Lengua or Catalangu. The distinctions made in the Quechua system have been preserved. It is difficult, however, to see whether relexification is involved or generalization in terms of an unmarked six-term system (three persons, two numbers), or both. Consider, for instance, the marking of the plural pronouns. Saraguro ML has miu- "we", using the Quechua plural marker -, while in Quechua the form is only indirectly related to the singular pronoun: ñuka "I" — ñukanchik "we". It is striking how many differences there are among the three varieties. The divergences among the forms in the three varieties suggest that relexification occurred several times independently in the Ecuadorian high lands, each time in a similar fashion but with slightly different results. 5.4. Are there varieties of Media Lengua in Peru and Bolivia? There is no indication in any of the published literature, to the best of my knowledge, that varieties of Media Lengua have emerged in Peru or Bolivia — in spite of the fact that varieties of Quechua have been much more thoroughly investigated in these countries, on the whole, than in Ecuador. What we do find is relexification in the WAYNO, a type of bilingual song popular both in Peru and in Bolivia (cf. e.g. Escobar & Escobar 1981, from which the examples below are taken). In this type of song there is frequent code-switching and borrowing and, in addition, occasional relexification in paired verses:
Media Lengua (89)
419
yacha-ra-nki-taq sabi-ra-nki-taq. know-PAST-2-
"you knew." In (89) yacha- is the Quechua word for "know" and sabi- is the equivalent Spanish word. Similarly, in (90) sapa- is the Quechua word and sulu- the Spanish word for "alone". In both cases the Spanish-derived word has the Quechua lexical semantics and morphosyntax. Spanish solo, for instance, would never appear with person inflection in Spanish. (90) sapa-y-ta riku-wa-spa sulu-y-ta riku-wa-spa alone-1 -ACC see-lOBJ-SUB "seeing me alone" Here the Spanish and the Quechua words participate in a kind of "doubling" poetic structure. The evidence for relexification rather than borrowing in these cases consists of the already-mentioned incorporation into the Quechua lexicon and the fact that core lexical items are affected; these are never replaced by borrowings in ordinary discourse. What the wayno examples demonstrate, to my mind, is not so much that there are cases of Media Lengua in Peru and Bolivia, but rather that the linguistic mechanism of relexification is potentially available there as well. If there is a cultural need for it in a specific community, a variety of Media Lengua could emerge. Perhaps a more thorough search will bring them to light.
6.
Conclusions and suggestions for further research
In this paper I have tried to describe the main features of the variety of Media Lengua spoken around Salcedo in central Ecuador. In addition, I have sug gested a possible scenario for its genesis, in the context of the sociolinguistic history of Quechua and Spanish in Ecuador. I have also briefly compared the variety of Media Lengua under study to other varieties and to other contact vernaculars in this part of the Andes. I concluded that Media Lengua does not resemble other types of contact vernaculars in the Andes, as far as we can tell from the available evidence.
420
Pieter Muysken
Further research is especially needed for the following purposes: (a) to determine whether Media Lengua resembles results of language mixture elsewhere, such as Michif in Canada and the U.S. and Ma'a (Mbugu) in Tanzania. There are resemblances as well as differences; perhaps a wider comparative perspective will also tell us something about possible scenarios for the emergence of languages like Media Lengua. (b) to study the interaction between the processes leading to Media Lengua and other contact processes such as code-switching, borrowing, and second language acquisition. Saraguro Media Lengua and Catalangu have only been studied superficially so far; a more detailed analysis of these varieties should yield more solid results, particularly with respect to the question of interaction.
Media
Lengua
All
Appendix 1. The 200-word Swadesh list of core vocabulary for the local Quechua and for Media Lengua (items for which the Media Lengua form is unknown are omitted). SWADESH LIST
1. 2. 3. 5. 7. 9. 11. 13. 19. 20. 22. 23. 26. 27. 30. 31. 32. 34. 36. 37. 38. 40. 46. 47. 50. 53. 54. 57. 58. 59. 60. 61. 62. 63. 66. 67. 69. 72.
all and animal at bad because big bite burn child cold come day die dog drink dry dust earth eat egg fall few fight five flower fly four freeze fruit give good grass green hand he hear here
LOCAL QUECHUA
MEDIA LENGUA
tukuy -un, -pish, i animal -pi mana alli, malu -munda hatun kanirupawawa chiri shamup'uncha wañuashku ubiachakishka pulbu ashpa mikululun urmaashalla, p'itilla makanakupishka sisa bulachusku kasap'ruta kualli k'iwa birdi maki pay uyakay-bi
tuditu -un, -pish, i animal -pi malu -munda grandi murdikimawawa chiri binidia muriashku, pirru tuma-, ubiasiku pulbu tirrinu kumiwibu kayip'iti piliasinku flor bulakwatru ilap'ruta dabwinu yirba birdi manu el uyaaki, aki-bi
422 13. 74. 75. 77. 78. 79. 81. 82. 83. 85. 90. 94. 95. 96. 97. 98. 104. 105. 107. 108. 109. 100. 111. 112. 115. 119. 127. 128. 129. 130. 133. 134. 135. 136. 138. 139. 140. 146. 155. 156. 157. 159. 161.
Pieter Muysken hit hold how husband I ice in kill know laugh live man much meat mother mountain new night no old one other person play rain river say scrape sea see sharp short sing sit sky sleep small some straight suck sun swim that
makachariimashna kusa ñuka iladu -pi wañuchiyachaasikawsak'ari ashka aicha mama urku mushuk tuta mana ruku, mauka, paya shuk shuk, kashuk hinti, runa pukllatamia (hatun) yaku niaspimar rikup'ilu uchilla kantatiyasilu, hawa pacha puñuuchilla shuk riktu chupainti nadachay
pigatinikumu maridu yo, rniu iladu -pi matasabiribibi-, sintaumbri muchu karni mama sirru mushuk, nwibu nuchi no ruku, bixu unu utru hinti xugallubia riu (di)ziaspi-, raspamar bip'ilu chikitu kantasintasilu durmichikitu algunos riktu chupasol, solo nadaisi
Media 162. 163. 166. 167. 168. 169. 170. 174. 175. 176. 177. 178. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 194. 195. 197. 198. 199.
423
Lengua there they think this thou three throw tree turn two vomit walk wash water we wet what when where white who wide wife wind with woman worm ye year
chay-bi pay-guna yuyakay kan kimsa sitayura muyuishki lanzapurimaillayaku ñukunchi shutuk ima ima uras may-bi yuruk pi hatun, kampu warmi wayra -un warmi k'uru kan-guna wata
alli-bi, alli el-kuna (yu)ya-, pinsaisti bos tres butaarbol birados gumitaandalabaawa nustru moxa-shka, moxa-du inki kwandu undi, undi-bi blanku kin anchu warmi bintu -un warmi, muhir gusanu bos-kuna arm
Appendix 2. Quechua vocabulary in Media Lengua. A: Quechua culture items B: Quechua borrowings in local Spanish (CF - elicited data from family A; CI - informal data from family A; MI - informal data from family B) nouns chiri (CF, MI) kancha (CF) p'iti (CF, CI) kuyi (CF)
"cold" "toasted corn" 'a little bit" "guinea pig"
424
Pieter Muysken warmi (CF, CI) kincha (CF) punchu (CF) warangitu (CF) wagra (CF) ingillu (CF) llama (CF, MI) ukucha (CF) wawa (CI) chichi (CI) wambra (CI) chugllu (CI) lancha (CI) runa (CI) masha (CI) wayku (CI, MI) pampa (CI) killa (CI) mishki (MI) ashku (MI) k'uchi (MI) yanga p'uncha (MI) pata (MI)
"wife, woman" (altern, muhir) "yard, corral" "poncho" "cactus beer" "cattle" "Quechua" (← Sp. *inga-illu) "sheep" "mouse" "child" "way of preparing meat" "boy" "corn cob" "drizzle" "indian" (orig. "person") "son-in-law", "brother-in-law" A "ravine" "plain" "month" "cactus juice" "dog" (altern, pirru) "pig" (altern, chanchu) "weekday" "side" (altern, ladu)
chayku- (CF) chiri- (CF) tuku- (CF, CI) ga- (CF, CI) ashna- (CF) lati- (CF) shutichi- (CF) yuyari- (CF) (yu)yakillana- (CF) k'uya- (CF) rikchari- (CF) uya- (CF) mayllachi- (CI) ubia- (CI) aspa- (MI) kayna- (MI) wanllamu- (MI) macha- (MI)
"get tired" "be cold" (altern, chiri azi-, friyu ga-) "end up, do" (altern, azi-), "turn" "be" "stink" "bark" "baptize" A "remember" "think" "be lazy" "like, love" "wake up" (altern, rekorda-) "hear" "sprinkle" "drink" (altern, toma-) "scrape" (altern, raspa-) "pass the day" "bring presents or souvenirs" "get drunk" (altern, chuma-)
verbs
Media Lengua
425
other manchanida (CF) k'ari (CF) kwenta (CF) mushuk (CF) yangapacha (CF) kay-mu (CF) k'willa (CF) millani (CF) chawpi (CI) laya (CI) ruku (CI) washa (CI) ñutu (CI) p'akta (MI)
"terribly" "strong" "like" "new" (altern, nwibu) "like anything, for nothing "to here" (altern, aki-mu) "pretty" "scary" "half "like" "old" "behind" "fine-grained" "enough"
Note 1.
My spelling system for Quechua and Media Lengua is similar to the one commonly accepted now in Peru. I do not use i and everywhere, but sometimes e and for the midhigh vowels, according to the pronunciation. For the consonant system I use the array of characters represented in Table 2 above. For Spanish names I use the Quechua pronuncia tion in the Media Lengua and Quechua examples: xwan, mariya, xusi, pidru, etc. The spelling is roughly phonetic, based on the local dialect variants. Thus the accusative case marker is sometimes represented as -ta and sometimes as -da, depending on the pronun ciation in a particular context. For Spanish I have remained fairly close to standard orthography, except where there was a strongly divergent pronunciation. The following abbreviations are used in the glosses: 1 = first person, IFUT = first person future, IOBJ = first person object, 2 = second person, 3 = third person, 3FUT = third person future, ACC = accusative, AFF = affirmative, AG = agentive nominalizer, ASP = aspect, BEN = benefactive, purposive, CIS = cislocative, DES = desiderative, DIM = diminutive, DUB = dubitative, EMP = emphatic, GEN = genitive, GER = gerund, IMP = imperative, INC = inchoative, IND = independent, INF = infinitive, LOC = locative, NEG = negation, NOM = nominalizer, OBJ = object, PAST = past tense, pl = plural, POT = potential mood, PROG = progressive aspect, PROX = proximate, REFL = reflexive, SD = sudden discovery tense, SUB = adverbial subordination, TOP = topic marker. When several words in the gloss correspond to a single Media Lengua or Quechua item, they are separated by a period. While distinguishable morphemes are separated by a "-" in the gloss, Spanish inflections are often marked "=".
426
Pieter Muysken
References Cole, Peter. 1982. Imbabura Quechua. (Lingua Descriptive Studies.) Amsterdam: NorthHolland. Escobar, Gabriel, and Gloria Escobar. 1981. Huaynos del Cusco. Cusco: Editorial Garcilaso. Gnerre, Maurizio 1975. "A Spanish pidgin of the Shuar". University of Rome MS. Lefebvre, Claire, and Pieter Muysken. 1988. Mixed categories. Dordrecht: Kluwer. Le Page, Robert, and Andrée Tabouret-Keller. 1985. Acts of identity. Cambridge: Cam bridge University Press. Muysken, Pieter. 1977. Syntactic developments in the verb phrase of Ecuadorian Quechua. Dordrecht: Foris. Muysken, Pieter. 1979. "La mezcla entre quechua y castellano". Lima: Lexis 3/1.41-56. Muysken, Pieter 1980. "Sources for the study of Amerindian contact vernaculars in Ecuador". Amsterdam: Amsterdam Creole Studies III: 66-82. Muysken, Pieter. 1981a. "Quechua word structure". Binding and filtering, ed. by Frank Heny, 279-327. London: Croom Helm, Cambridge, MA: MIT Press. Muysken, Pieter. 1981b. "Halfway between Quechua and Spanish: The case for relexification". Historicity and variation in creole studies, ed. by Arnold Highfield and Albert Valdman, 52-78. Ann Arbor: Karoma. Muysken, Pieter. 1984. "The Spanish that Quechua speakers learn: L2 learning as normgoverned behaviour". Second Languages: A cross-linguistic perspective, ed. by Roger W. Andersen, 101-24. Rowley, MA: Newbury House. Muysken, Pieter. 1986. "Contactos entre Quichua y Castellano en el Ecuador". Memorias del Primer Simposio Europeo sobre Antropología del Ecuador, ed. by S.E. Moreno Yanez, 377-451. Quito: Ediciones Abya-Yala. Muysken, Pieter. 1988. "Lexical restructuring and creole genesis". Akte des 4. Kolloqui um über Sprachkontakt, ed. by Norbert Boretzky et al., 193-210. Bochum: Brock meyer. Muysken, Pieter. 1989. "Media Lengua and linguistic theory". Canadian Journal of Linguistics 33/4.409-22. Muysken, Pieter, and Louisa Stark. 1978. Diccionario Español-Quechua, QuechuaEspañol. Quito-Guayaquil: Museo del Banco Central. Poplack, Shana, David Sankoff, and Christopher Miller. 1988. "The social correlates and linguistic processes of lexical borrowing and assimilation". Linguistics 26.47-104. Simson, Alfred J. 1886. Travels in the wilds of Ecuador. London: Sampson Low, Marston, Searle, and Rivington. Stark, Louisa, et al. 1972. El Quichua de Imbabura: Una gramática pedagógica. Otavalo, Ecuador: Instituto Inter Andino de Desarrollo. Torero, Alfredo. 1974. El Quechua en la historia social Andina. Lima: Universidad de Ricardo Palma. Toscano Matteus, Humberto. 1953. "El español en Ecuador". Revista de Filología Española, Anejo 61. Velasco, Juan de S.J. 1780 [1964]. Vocabulario de la lengua índica. Republished as Llacta 20. Quito: Instituto Ecuatoriano de Antropología y Geografía.
Callahuaya Pieter Muysken Universiteit van Amsterdam
1.
Introduction
A case of language mixture which has not been discussed in the general language contact literature so far is Callahuaya (also spelled Kallawaya and Callawaya) or Machaj juyay "language of the people, the family", which is spoken in curing rituals in a region of northwest Bolivia, northeast of Lake Titicaca, by the older members of a group of 2,000 itinerant healers. The language is a form of Quechua with a vocabulary drawn from different sources, but apparently mostly from the now extinct language Puquina (?Maipuran). Other contributing languages may include Tacana (PanoTacanan). This paper is a preliminary summary of the information now available about the language. Much more detailed work is needed, particu larly on the Callahuaya lexicon; in addition, there are still manuscript sources and 19th-century materials that need further study.
2.
Location, present position, and sociolinguistic history
The geographical center of the Callahuaya healers, who also work in the capital La Paz and travel widely in South America, is Charazani, a town in the province of Bautista Saavedra. The farthest the Callahuaya have travelled is Panama (Guirault 1974:63), and there were once fixed routes and zones divided among the healers according to provenance and rank. Stark (1972: 199) suggests a number of 2,000 for the ethnic group of Callahuayas as a whole, while Rösing (1990:39, 46) mentions 500 as the number of inhabitants of Charazani and cites 10,668 inhabitants, mostly peasants, as the official
428
Pieter Muysken
1960 census figure for the entire province. She estimates the number of curers as much higher than earlier estimates of 28 or 37. Girault (1974:71) cites a figure of 2,000 for the group of healers and their families. He stresses that only people from seven particular villages in the province could become healers: Curva, Chajaya or Gral, González, Kamlaya, Wata Wata, Inka, and Chari (1974:61). It is clear from Rösing's work (1990:39) that only a minority of the curers still use the Callahuaya language during the rituals. All of them speak Quechua, many also speak Aymara, and some speak Spanish as well. Ranaboldo (1986:117, 126) confirms that the present-day Callahuayas are all descendants of families in which there has been a famous healer in the past. Nonetheless, only a few older healers still know how to speak the language. Girault (1974:71, 72) underlines the fact that the Callahuaya, rather than being a highly conservative atavism in Bolivian society — the way they are often portrayed — constitute a highly innovative elite. Many have abandoned traditional medicine and are middle-scale entrepreneurs, jewelers, pharma cists, etc. The Callahuaya themselves founded a primary school in the village of Chajaya, as far back as 1928. Even the traditional medicine they practiced had been enriched with herbs from many of the regions to which they travelled, and is constantly adapting itself to new influences and develop ments. The general area of Charazani is a Quechua-speaking island (a total of 16,500 speakers in the provinces of Muñecas and Bautista Saavedra; cf. Albo 1980:92-93) in the midst of a generally Aymara-speaking region, but origi nally the area was Puquina-speaking (Stark 1985, Torero 1987, Saignes 1989). The ancient Tiwanaku empire, associated with the Puquina language and situated in the Lake Titicaca basin, went under before the 12th century and was fragmented into a number of chiefdoms. From then on successive waves of Aymara-speaking invaders caused a progressive Aymarization of the area, and when the invaders allied themselves with the Incas in the 15th century, the Puquina culture was definitively crushed. Only pockets of Puquina survived, and the language finally vanished in this century from its last enclave in the village of Curva, north of Charazani in the Bautista Saavedra province. Quechua was introduced into the area by mitmaqs, forc ibly resettled laborers from Peru who were brought to this strategic region (it is flanked by the slopes of the Amazon basin) by the Incas in accordance with their policy of forced resettlement of entire populations. At the end of the 16th
Callahuaya
429
century the area was trilingual — Puquina/Aymara/Quechua — and now Quechuization is almost complete. A factor contributing to this process was the location of Charazani on the road linking Peruvian Cuzco and the Bolivian mining center Potosi, along which many Quechua-speaking forced laborers were brought south (Albo 1987:376). The Quechua spoken there resembles that of the Cuzco area, together with features from the Ayacucho region, both of which are in southern Peru. This mixed character could be due to the different origins of the mitmaqs. The mixed language Callahuaya probably emerged at some point during the process of shift from Puquina to Quechua. Torero (1987) notes that the curing rituals required a secret language, and that the increased radius of the curers' activities (throughout the Quechua-speaking Andes) made a Quechua-based secret language desirable, because to an eavesdropper it would sound like Quechua. It is less likely but not impossible that the language emerged during the period of the Inca empire; a group called the Callahuayas played a special role at the Inca court, possibly already as curers, and a mixed Puquina/Quechua secret language could have played a role. Against this possibility, however, we should note the fact that Callahuaya has words for elements of Spanish origin where Aymara and Quechua have borrowings (Albo 1989). This suggests at least lexical elaboration during the Colonial or Republican periods. It may even be that Callahuaya emerged in the diaspora, as it were, during the period when Puquina was still spoken regularly in Charazani and when many of the herbal curers were not living in Charazani and its environs but rather in some urban center. If most Quechua morphology were intact, as Stark (1972) claims, while the Puquina lexicon is reduced and several other languages have contributed as well, this would suggest either that original Quechua speakers invented Callahuaya (Stark 1972) or that Quechuization was well advanced when the language emerged. We will see below that there has been more morphological restructuring than Stark claims. Evidence against an early origin for the language is also found in the surprising observation by Girault (1974:60-61) that there are many 19thcentury references to the Callahuaya, but that the earliest colonial reference to Indians from Charazani acting as healers dates from 1776. Given an abun dance of detailed colonial sources about medicine and customs in general, the implication is that the Callahuaya emerged as a group of healers only in the 18th century.
Pieter Muysken
430 3.
The sources and their interpretation
We have several sources for the language Machaj Juyai, or Callahuaya proper. Oblitas Poblete (1968) provides a grammatical sketch in traditional terms and an extensive vocabulary. His claim that the Callahuaya language originated as the Inca language is not well founded, but this does not make his work less valid. He is from the area, and Rösing (1990:64) considers him the most important researcher of Callahuaya culture. Stark (1972) provides a brief sketch and word list, and argues that the language is a somewhat pidginized merger of a Puquina lexicon and a Quechua grammar. She calculates that 70% of the 200-word Swadesh list is from Puquina. Torero (1987) surveys the distribution of four major Indian languages in the Titicaca basin during the Colonial period — Aymara, Puquina, Quechua, and Uru- quilla — and concludes that Callahuaya shares 41 % of its lexicon with (what is known of) colonial Puquina (or 64%, if the calculation is based on the 100-word Swadesh list). Girault (1989) provides a brief grammatical sketch in traditional terms and an extensive vocabulary, including a cassette with recorded verb para digms and phrases in the language. Girault, who died in Bolivia in 1975, had done an enormous amount of work on Callahuaya herbal medicine previ ously, and his work, much of which was published posthumously, is consid ered to be of great value. Rösing (1990) describes Callahuaya ritual curing practices from the perspective of ethno-psychiatry, reproducing many prayers in Quechua. She has worked among the Callahuaya steadily since 1983 and has published a series of detailed studies in German; Rösing 1990 is only the introduction to these studies. Finally, Mondaca has made his fieldwork notes available (a brief introduction with four typed pages of words and phrases, which I'll cite as n.d. 3-6), resulting from ethno-medical research focusing on curing prac tices. Several problems of interpretation arise with all the available sources. First, we know very little about Puquina, a language in which only a few Christian religious texts survive, so that detailed lexical analysis in particular is hazardous. Second, Puquina probably borrowed extensively from Quechua and Aymara starting in the 12th century (just as the latter two languages have been avid borrowers from each other). Ethnolinguistically, therefore, an element
Callahuaya
431
may have been viewed as Puquina by Callahuaya speakers even if it had a different etymological origin. Third, Albo (1989) suggests that perhaps Tacanan or Maipuran lexical elements may be present in Callahuaya. Indeed, my own research indicates that there is some Tacanan vocabulary in the language (Appendix 2), though there is none from neighboring languages such as Apolista (Maipuran) and the isolate Leco (Appendix 1). Fourth, Callahuaya is a ritual curing language, and this may give it pragmatic, lexical, and stylistic characteristics that make it hard to compare with languages of daily use. The grammatical sketches, which treat it as if it were a language of daily use (including all tenses for all persons, etc.), may therefore be less reliable. Fifth, very few actual sentences are given in the sources, which mostly give extensive paradigms and information about phrasal categories. This makes it hard to evaluate Stark's 1972 claim that Callahuaya is somewhat pidginized syntactically. Finally, the sources are not always in agreement, a matter to which I will return below. While Guirault (1974:69) claims that there is dialect variation in Callahuaya between speakers from the zone of Curva and speakers from Chajaya and Kkamlaya, this variation is limited to idioms and phrases, and does not affect the overall grammar.
4.
Basic structure
After these cautionary remarks I should emphasize that the basic features of the language are generally agreed upon. Consider a typical sentence such as (l): 1 (1)
Cchana-chi-rqa-iqui Call-CAU-PAST-1 -2
isna-pu-na-iqui-paq. go-MOV-NOM-2-BEN
"I had you called so that you can go." Here cchana- "call" and isna- "go" are non-Quechua, but all the other morphemes are Quechua and the structure corresponds to Quechua structure. A similar example is given in (2): (2)
mii-qa llalli oja-cu-j-mi acha-n. man- good eat-REF-AG-AFF be-3 "The man is a very greedy eater." (Oblitas Poblete 1968:40)
432
Pieter Muysken
Here mii "man", oja- "eat", and acha- "be" are non-Quechua. Again, the rest of the morphemes and the syntactic structure are Quechua. Taking this to be the basic pattern, let us now consider points where there is no exact correspondence between Quechua and Callahuaya grammar. 4.1. Phonology Stark (1972:200-206) has presented a reasonably detailed analysis of the phonological inventory of Callahuaya, and the information she provides does not conflict with what can be inferred from the other sources. In essence, Callahuaya is a merger of Quechua and Puquina, with a series of aspirated and glottalized stops (lacking in Puquina) from Quechua, and a five-vowel system with distinctive length from Puquina (Quechua has a three-vowel system, and the Southern varieties lack a length distinction). While long vowels only appear in words of Puquina origin, aspirated and glottalized stops appear in words both of Quechua and of Puquina origin. When we consider the distribution of consonants and consonant clusters, the Callahuaya system resembles that of Quechua much more than that of Puquina. 4.2. Plural marking The Spanish plural suffix-s is very common in Quechua, but not in Callahuaya, where the Quechua plural - is prevalent. Compare (3) and (4) (Stark 1972: 216): (3)
Quechua:
q"ari-s, man-pl
warmi-s, alkalde-tah woman-pl mayor-EMP
ri-n-ku. go-3-pl
(4)
Callahuaya: laja-, atasi-kuna, alkalde-tah isna-n-ku. man-pl woman-pl mayor-EMP go-3-pl "The men, the women, and the mayor went."
Girault (1989:143) gives the doubled form --s for Callahuaya. In some varieties of Bolivian Quechua we also have doubled forms, but always in the order -s-kuna.
Callahuaya
433
4.3. Gender marking and compounding Gender for animals can be marked with a preposed china "female", as in Quechua, but for humans the words laja "man" and atasi "woman" follow rather than precede the modified element, unlike the Quechua pattern. In addition, one gets the impression that distinguishing genders in this way is much more frequent than it would be in Quechua. Compounded place names also seem to have a modified-modifier order (Girault 1989:199): (5)
jiri killo usi p'ala jikus laya
"stone ordinary" "house ancient" "place shady"
In Quechua we find the opposite order in place names: (6)
chiri yacu "cold water" pucara"red fortress" yana cocha "black lake"
4.4. Question words, indefinites, and demonstratives The question words and the indefinite and demonstrative elements are given in Table 1. The starred forms are Quechua borrowings; most of the other forms may be derived by means of the Puquina question word qui (Torero 1987). It is not always clear whether a given word is interrogative or deictic. Thus Oblitas Poblete sometimes glosses khii "what" as deictic "today" or "third person singular" (1968:30). Similarly, the form khistu, glossed as a deictic in Table 1, is sometimes used as a question word. 4.5. Nominal si There appears to be a nominal suffix -si, which does not occur in Quechua: (7)
ata-si ni-si uj-si ? qaj-si ? qom-si
"woman" (compare Puquina ata-jo) "I" (compare Puquina ni) "one" (compare Quechua huq) "seven" "dog"
434
Pieter Muysken
Table 1. Question words, demonstratives, and indefinites (G = Girault, OP = Oblitas Poblete, S = Stark, M = Mondaca). Starred forms are clearly Quechua. CALLAHUAYA
khii/OP, S, G; k'ijka/G khii/OP; kii/G khiitaq/OP; kitaq/G; khiru/S khii-pacha/OP, G u-khii-pas/OP; u-khii-paj/G khii-lla-pas/OP khi-chaina/OP khiru/OP; kh'iru/G khiro-ja/OP; khi'iruja/G khiro-jta-taj/OP khiru/OP; kirulG u-khiru-ja-pa/G khiru-caj/OP khiru-pi-ta/G khiru-pi-tu/G khiiri-lla-pas/OP; kh 'iri-paj/G u-khiri-lla-pas/OP khistu/OP, S; kistu/G, M khistu-pi/S, M; kistu-pi/G, M ' urna i-taj/OP; kh ' ichej/G (cf. Tacanan ketsono "when") 'una-lla-pas/OP, kuna-lla-paj/G ku'uta/G * u-jaik'aj/OP; u-jaika/G * jaik'aj/OP * ima-paj-taj/OP
GLOSS
what today who and then no one some how someone which whose that (dem.) none that yonder there there yonder some none this here when something how much never when wherefore
4.6. Copula The auxiliary verb acha- "be", which can occur in composite tense para digms, in predicative expressions, etc., can also be used transitively, as can be seen from the accusative suffix on the infinitival complement: (8)
ujsi-lla khii-ta isna-y-ta acha-waj-chu? one-DEL what-EMP go-INF-ACC be-2P0T- Q "With what in mind would you be walking alone?"
Stark (1972) also glosses acha- as "say", and Oblitas Poblete (1968:43) gives an example in which it means "pass the time":
Callahuaya (9)
435
nisi acha-ku-ni ratu achira-spa ratu juya-spa. I be-REF-1 now sing-SUB now weep-suB "I pass the time now singing now weeping."
4.7. Case marking The Callahuaya case system is largely identical with the Quechua system: about ten affixes are attached to the rightmost head noun in the noun phrase. Girault (1989:148) mentions that the Quechua locative -pi alternates with -pichu. This ending surprisingly appears sometimes on direct objects in his sample sentences: (10) chuinin usi-pichu wanaku-rka-n. he house-LOC adorned-PAST-3 "He adorned the house." (11) iktara-rka-n
nisi-pichu.
cheat-PAST-3 I-LOC
"He cheated me." Mondaca (n.d.:6) gives the compound form k'apa-pi for the locative: (12) ojari-sun-chej Escoma-k'-pi. eat- lFUT-pl Escoma-LOC "We'll eat in Escoma." On the other hand, accusative -ta is generally absent, both in preverbal and in postverbal positions, where it would have been called for in conservative varieties of Quechua. The -chu form recalls Quechua I varieties, where we have locative chaw. Other evidence for Quechua I influence on Callahuaya comes from vowel lengthening in the first person, also a typical Quechua I feature. The mixture of Cuzco Quechua, Ayacucho Quechua, and Quechua I features in Callahuaya may indicate that the language is a frozen form of the originally very mixed mitmaq variety of Quechua spoken in the area; in most places there was homogenization during the colonial period, but Callahuaya may have preserved the mixed character of the language resulting from the Inca policy of transferring groups of speakers all over the empire.
436
Pieter Muysken
4.8. Word order A final issue that merits discussion is word order. The data in Oblitas Poblete (1968) all conform to Quechua OV word order, as may be expected from the discussion so far. Question words are initial, objects and complements tend to precede the verb, predicates precede the copula, and all modifiers precede the head noun. There appears to be a system of nominal postpositions similar to the one in Quechua, but with different lexical shapes: (13) catu paa piti juna pig waqas
"inside" "outside" "above" "below" "front" "back"
(cf. (cf. (cf. (cf. (cf. (cf.
Q. Q. Q. Q. Q. Q.
ukhu) hawa) hawa) ura) chimpa) washa)
In Girault's sample sentences a slightly different picture is suggested. The patterns resembling those of Quechua are present, but in addition there is a rather large number of VO sentences, instances of noun-adjective order, clause-final question words, postverbal adverbs, etc. A striking case is given in (14), which could be interpreted as containing a preposition-like element created with a deictic. Alternatively, one could say that the case marker -wan appears on the leftmost element of the noun phrase rather than the rightmost element. (14) kistu-wan sokena-a-kunas soke-say kamiska. this-with remedy-1-pl cure-1 FUT patient "With these remedies I'll cure the patient." The problem with interpreting the variation in word order is that it might be caused by an elicitation technique that takes a Spanish sample sentence as the point of departure. That this is what Girault did appears from cases such as (15) (1989:150): (15) khoaja-ka-nki kistu-man-pi. look-PAST-2 this-to-LOC "You looked in this direction (Sp. en esta dirección)'' Here the unexpected locative element -pi may well have been triggered by the Spanish preposition -en "in".
Callahuaya 5.
431
Person marking
The largest interpretive problem arises from a discrepancy within Girault's work and between Girault 1989, on the one hand, and Oblitas Poblete 1968 and Stark 1972 on the other. The problem involves the second and third person, which are marked in Quechua with -nkil-yki and -n, respectively: (16) wasi-yki puri-nki
"your house" "you walk"
wasi-n puri-n
"her/his house" "s/he walks"
There is nothing in Oblitas Poblete 1968, Stark 1972, or Mondaca (n.d.) that suggests that this same situation does not hold for Callahuaya. 5.1. Verbal inflection We could draw the same conclusion from the verbal paradigms in Girault (1989:135-42), but from the actual examples he gives (1989:149-53) a differ ent picture emerges. The Quechua second-person verb form -nki is used five times in the present tense for a second person in Girault's Callahuaya sen tences and twice unambiguously for a third person, but in six examples it is ambiguous, because the Spanish translation has a third person — which in Spanish could also be the polite second-person form — and the sentences have no context. The glosses suggest a third-person form, but in some of the examples a second-person form seems indicated (e.g., "Why are you laugh ing?" is more natural than "Why is he laughing?"). (17) yani kkena yuna-nki. much money earn-2 "Ganabas mucho dinero./You made a lot of money." (18) ikili-n acha-pu-nki kitaj? father-2 be-BEN-3 who "Quién es tu padre?/Who is your father?" (19) kiru-ma isna-nki? where-to go-? "Adonde vá?/Where are you/is he going?" The Quechua verbal third-person form -n is used for a third person in Girault's Callahuaya sentences (thrice), and in the past and the plural we have only sample sentences that conform to the Quechua pattern.
438
Pieter Muysken
We can conclude from this that there has been variable overgeneralization of the verbal second-person marker -nki to third-person contexts in Girault's data, while the verbal third-person marker -n is not used in secondperson contexts. 5.2. Nominal inflection For nominal person marking, the picture is still more complicated. Possession is marked in either of two ways (according to Girault): (i) The noun is optionally preceded by a pronoun with a marker that sometimes resembles the Quechua genitive case (in either a nonemphatic or an emphatic form), and followed by a person marker (again, in either an emphatic or a nonemphatic form). The person pronouns with their possessive forms are given in Table 2. The Callahuaya plural pronominal forms are directly modelled on the Quechua forms. The emphatic forms with -j-ta resemble Quechua indepen dently-used possessives (consisting of genitive -j and accusative -ta), which, however, can never be used prenominally in Quechua. The nonemphatic form nisi-p contains an older form of the genitive. The iki in the second-person nonemphatic pronoun resembles the Quechua second-person possessive end ing (see below). The ku in the third-person nonemphatic resembles the Callahuaya second-person emphatic nominal ending (see below). The overall picture is one of great irregularity. It should be noted that, in Girault's account, the third-person pronoun chuinin appears to be derived from second-person chuu. In Oblitas Poblete (1968:39) the third person is given as jiru "that", khii "who", or pichi. The last Table 2. The pronouns of Callahuaya (based on Girault 1989). PRONOUN
NONEMPHATIC
EMPHATIC
QUECHUA
1sg 2sg 3sg lpl
nisi chuu chuinin nisi-nchej
nisi-p chu-niki-j chunin-ku
nisi-j-ta chuu-j-ta chuninku-j nisi-nchij
2pl 3pl
chuu-kunas chuinin-kunas
ñuqa qan pay ñuqa-nchij ñuqa-yku qan-kuna pay-kuna
chuu-kuna-j-ta
Callahuaya
439
Table 3. Nominal inflection in Callahuaya (based on Girault 1989). NONEMPHATIC
EMPHATIC
QUECHUA
... + ku
i/V ni-i / C
... + ku
iki / V ni-iki / C
3sg
longV / V i/n ni I , W n/V in / n nin / C, VV ki
... + chi
lpl
Vl
n/V ni-n / C ex iku / V ni-iku / C in nchis / V ni-nchis / C iki-chis / V ni-iki-chis / C
1sg
2sg
2pl 3pl
ilni + ku
nlinlnin+ku
n-ku / V
ni-n-ku / C
form could be related to Puquina pi "you" (containing perhaps the emphatic particle chi as well; see below). In any case, there may not have been a clear distinction between third and second person in the language that gave rise to Callahuaya, probably Puquina, if we take this as evidence. The nonemphatic endings are given in the first column in Table 3, and the emphatic endings comprise the singular forms + a particle from the second column. This paradigm is remarkable in several ways. First, the second and third persons have been exchanged, in the singular. In parallel with this, the Quechua 2pl marker -chis is used as an emphatic form in the 3sg, and the Quechua 3pl marker is used as a second-person emphatic form. Finally, the morphophonemic alternations in the Callahuaya person forms are patterned on the Quechua use of euphonic ni after consonants, but have incorporated sensitivity to the Puquina feature of vowel length (described in detail by Stark 1972). (ii) To make matters even more complicated, Girault (1989:147) reports that with nominal possession we have genitive -j or benefactive -paj on the prenominal possessor, and -n (-an after consonants) on the possessed ele ment. The -n form is the expected Quechua third-person form (since a nominal possessor is by definition third person), while the alternant -an is not
440
Pieter Muysken
derived from Quechua. Thus nominal and pronominal possession follow different rules. 5.3. Analysis When we look at the sample sentences, -iki is used once for second person and three times for third person, and once it is ambiguous. The form -n is used only once, for third person. There are three examples where third person is marked -n verbally and -iki nominally, as in (20) and (21): (20) jita-n chamatu-ki waspakas-manta. lead-3 donkey-3 field-ABL "He led his donkey through the field." (21) suwi-ki lojminaku-rka-n. hand-3 burn-PAST-3 "He burnt his hand." There are no cases where the reverse situation holds. Thus it seems that the Quechua -nkil-n inversion is more prevalent in the nominal than in the verbal paradigm. The fact that verbal agreement, pronominal possessive agreement, and nominal possessive agreement behave differently in Callahuaya suggests that the Quechua agreement system has been considerably restructured in Callahuaya. Difficult questions remain. First, why only in Girault's data and not in Oblitas Poblete and Stark? The answer may be that the latter two researchers approached Callahuaya from a Quechua background. Oblitas Poblete was presumably a Quechua native speaker, being from the region, and Stark is primarily a Quechuist. Girault is not necessarily a more accurate observer than the others (he is not a trained linguist and the notes are sketchy), but he is less biased towards Quechua. Thus the Callahuaya he recorded may be more accurate in this respect. Recall that Callahuaya is being replaced by Quechua even in the healing practices; one could easily imagine a more Quechua-like Callahuaya co-existing with an originally more heterogeneous linguistic sys tem. A second issue is the variation in Girault's own data. Is it due to influence from the Quechua superstrate, or is it inherent in the Callahuaya system itself? A third issue is the cause of the -nkil-n inversion. A number of possibili ties present themselves. First, as noted above, the substrate system giving rise
Callahuaya
441
to Callahuaya may not have distinguished second and third person grammati cally. Another possibility is that the distinction is not always clearly made in the local Quechua. This is clearly not the case for simple cases, but Rösing (1990:24) notes that in the plural, with a first-person object, second and third person are collapsed. Ni-wa-n-ku means both "they say to me" and "you (pl.) say to me", and ni-wa-nki-chik is not used. Notice, however, that here -n encroaches upon the territory of the second person, while in Callahuaya -nkil -yki encroaches upon the territory of the third person. Another problem with an explanation in terms of the local Quechua is that this variety itself may be the result of a Puquina/Callahuaya substratum. A third option is that pragmatic features of the curer's discourse lead to a confusion of the second and third person. From the (Quechua) prayers in Rösing (1990) it is clear that, while the curer is addressing the patient, he is really addressing spirits through the patient and talking about the patient in the third person (Rösing 1990:278): (22) a.
kay-qa uj salud-ni-n-paj this-TOP one health-EU-3-BEN
b. kay-qa uj vida-n-paj this-TOP one life-3-BEN "This is for her (read: YOUR) health and this for her (YOUR) life..." Again, this could explain why -n would be used for second person, but not why -nkil-yki would be used for third person. A fourth possibility involves the notions of simplification, functional load, and saliency. Note that Quechua second person -nkil-yki is both phonologically more salient than -n and has a higher functional load — it has a separate nominal and a verbal form, and it is not homophonous with another particle. The suffix -n can be either third person or the affirmative validator after a vowel. It is possible that in a simplified form of Quechua the secondperson markers would be more viable than the third-person form. This, perhaps together with the first explanation, could explain the pattern found in Girault's data.
442 6.
Pieter Muysken A comparison with the Media Lengua of Ecuador
When we compare Callahuaya with Media Lengua, the mixed Quechua/ Spanish language of Ecuador that I have studied (Muysken 1981; this vol ume), a number of things come to mind. Both varieties can be considered types of Quechua, structurally: the morphosyntax is dominantly from one language. This is where the similarities stop, however. First of all, the HISTORY of the languages is very different. While Media Lengua probably emerged early in the 20th century, an 18th- or early 19thcentury origin for Callahuaya is more likely. Also, Spanish is the NEW lan guage in Media Lengua, while Puquina is probably the OLD language in Callahuaya. A second difference concerns the integrity of the MORPHOSYNTAX. While Media Lengua grammar is a complete mirror of the local Quechua grammar, the grammar of Callahuaya only roughly resembles the local Quechua. I have noted a number of crucial differences. Similarly, the PHONOLOGY of Media Lengua very closely resembles that of Quechua (with a few voiced conso nants added in Spanish-derived items), while the phonological inventory of Callahuaya is a genuine compromise between Puquina and Quechua. Third, the AMOUNT OF LEXICAL ADMIXTURE from other languages differenti ates the two varieties. The Media Lengua lexicon is mostly Spanish, with a few Quechua items, while the Callahuaya lexicon contains Quechua, Puquina, Aymara, and also a large set of items with an as yet unidentified origin. A final difference concerns the FUNCTION of the languages. Callahuaya is a rather secret language used only by male healers in curing rituals, while Media Lengua is an in-group community language, by no means secret.
443
Callahuaya Appendix 1
The vocabulary of Callahuaya (Girault 1989) compared with what is known of the neighboring languages Apolista, Leco, and Tacana. The Apolista and Leco data are based on Montaño Aragón (1987, 1989); the Tacana data are based on Girard (1971) and Key (1968).
all amulet arm ashes aunt bee big black bone breast breed brother caimán camote canoe chicha cold corn cry dance day devil dog drink ear earth eat eye fall father finger fire fish fist flee
CALLAHUAYA
LECO
APOLISTA
TACANA
kumu sepja kkatu suwi ayaru lumiska kkatu khami kkusi yapaj jalawi/ tutu, kkatu ulla chuku kkatu enke,soko,ttuku/ wachaku laja khami
piwke mullu
anta
erna -bai eti-mo kwaibe wasa-ide
mullt''a
utilu ku'uku tukaku kaman jankku, rekkañito/ kumsi ttajchiku kkalla pakas, tuji ojaku chej chichiku ikili pikkalau jekanan, lumin/ jitiska suwimokko khojma
tsao -tzedo, atso
kiaj dudo towiri eru maropa
kkusi ajos kkaso
mapa iruyani irini chimapi
kati tutas tara/ta
awini machinki
caoba siri rige epani
chuchi skui ezña waq'a nío bukuy -tuni lal k'uriraj piru noteh
-to romi
moah ch"a
-idaja egua
yuho
chuwita
tata -batsu cuati
444
food foot forehead friend genios go, walk god hair hand head here house husband jaguar kill know make man money moon morning mother mouth neck no nose palmera people quickly rain river sepe Ins. sin sit down sky speak star stomach stone sun swim
Pieter Muysken
CALLAHUAYA
LECO
ojana chami methinaja kilki
sokoch riru
isnaku tutu nuki chiwi chiwi suwi ppeke kis tupi usi jatanakujmii jallachiku seka yapaku laja kkena okko, jaislumin kurea kkuna mili asa,kkasa jawillu katanichu, u chuini
APOLISTA
jana* -watsi -mata* epereje
bura ankari iriraj qaqa bueh waru kuy auvón polo wilaraj yatij kiah waju kerecha
TACANA
?meku -chipami -pisi -mbaku
-ina -me -chua
panishi miki wachi
etai
inami ashi
badi
sorami
cuara -cuatsa -nara mabe -lei
?mijis k'ura nda en, en -ape salk'a
mii ureraj inun jaa, karusta
niamari k"atan esera
buja jucha f'erah kauhut uruj pulea kitinu jiri kkatu luminakamañito/ jena wayanajaku
juchay kumaku janawin kalla juyaku khoo keripiti
nay manu
enui
epacuapecha
nuri
atoai
ipa amushiti
ineti joani
445
Callahuaya
swollen teeth then there there is tomorrow tongue tree want water white wind woman work yes yuca
CALLAHUAYA
LECO
pij naska kaa khi'i, jicha kh'irupitu achaku sakaman yaye lapha, sahachata jattaku mimi pokhosti sako atasi rakaku sekan
waju
APOLISTA
TACANA
-tse reta ni ? mijis -asa* atapi notuj nduwa
chami liki beni
awini sejmukuy oy, oi
niyipi hehe achikigloss
Appendix 2 The vocabulary for body parts of Callahuaya (Girault 1989) compared with that of Tacana (Key 1968, Girard 1971). Possible items shared by Callahuaya and Tacana are starred. TACANA (Key + Girard)
arm back
suwi elejan
beard blood body
chiwira, kkasa chiwi kampro kupillu
bone breast breast chin ear elbow excrement eye face
chuku kkatu enke/soko/ttuku * chchapu kaa jittin kkalla mukkle *ajej chej kkasaj
CALLAHUAYA (Girault)
bai G badede, bishi, exataxa , badede G kwei ami K.G -kita , kwita, -ka (der. suff.) G tsao K, tsau G tsedu G, (a)kwa G * atso K, atsu G jawi G -ida-ha K, ija G batso K, ba-chu G dumi G, *ze G -to-a K, a-tuka G -bo K, bu G
446
Pieter
TACANA
fat foot forehead hair hand head heart intestine jaw leg lip liver lungs meat mouth nail neck nerve nose navel penis saliva scar shoulder skin stomach throat
thuii chami *mathi-naja chiwi suwi ppeke sau* chunli, willun khallinaja, kkarie chami upachina, kuwin kichuana chchaju, wakhas pipi *asa, *kkasa silluna jawillu ankku chuini kerikili, pupulu milo phusi sejti-naja * lira-pitij * piti-kara keri mallka
tongue
*yaye
tooth vein waist
kaa siku kampro simi tekkni jumi
Muysken
CALLAHUAYA
tsei watsi , tsaka G, watsi G * -mata eina K -me , me G -choa , tsapa G, yuka G -nido, moesomo -sere K, du(ku) G dabi , G tidada , ti G keke K takwa K, G data, *shaw (cf. heart) rami/yami K, rami G * kwatsa K, kwatsa G tijiG piti/piki K, na G * sa(?)runa G (cf. heart) , wi G tsu?u G kwi kedi K dipi K * yara-pa K, ara G *-biti eto, ede , ? G, tu G tsoi-hani K, kani G, piti/tipi G *yana K, *ana G *asa(MA) (cf. mouth), tse K, tse G sa(?)runa G tima
Note My spelling system for Quechua and Callahuaya is similar to the one now commonly accepted in Bolivia. The following abbreviations are used in the glosses: 1 = first person, 2 = second person, 3 = third person, ACC = accusative, AFF = affirmative, AG = agentive nominalizer, BEN = benefactive, purposive, CAU = causative, DEL = delimi-
Callahuaya
447
tative, EMP = emphatic, EU = euphonic particle, FUT = future, INF = infinitive, LOC = locative, MOV = movement, NOM = nominalizer, PAST = past tense, pl = plural, POT = potential mood, Q = question, REF = reflexive, SUB = adverbial subordination, TOP = topic marker. When several words in a gloss correspond to a single Media Lengua or Quechua item, they are separated by a period. Distinguishable morphemes are separated by a "-" in the gloss.
References Albό, Xavier. 1980. Lengua y sociedad en Bolivia 1976. La Paz, Bolivia: Proyecto INENaciones Unidas. Albo, Xavier. 1987. "Commentary on Torero 1987". Revista Andina 5/2.375-77. Albό, Xavier. 1989. "Introduction". In Girault 1989, 13-17. Girard, Victor. 1971. Proto-Takanan phonology. Berkeley & Los Angeles: University of California Press. Girault, Louis. 1974 [1987]. "La cultura Kallawaya". Dualismo o pluralismo cultural en Bolivia. Mesa redonda sobre expresiones de la cultura boliviana en el lapso 19251974. Included in Louis Girault, Kallawaya. Curanderos itinerantes de los Andes. Investigaciuón sobre prácticas medicinales y mágicas, 23-34. La Paz: UNICEF-OPSOMS. Girault, Louis. 1989. Kallawaya: El idioma secreto de los incas. La Paz, Bolivia: UNICEF-OPS-OMS. Key, Mary Ritchie. 1968. Comparative Tacanan phonology. The Hague: Mouton. Mondaca, Jaime, (n.d.) La lengua callawaya: Apuntes de un cuaderno de campo. Univer sity of St. Andrews, Scotland: Centre for Latin American Linguistic Studies, Working Paper 18. Montaño Aragón, Mario. 1987. Guía etnográfica lingüística de Bolivia {Tribus de la selva) I. La Paz, Bolivia: Editorial Don Bosco. Montaño Aragón, Mario. 1989. Guía etnográfica lingüística de Bolivia {Tribus de la selva) II. La Paz, Bolivia: Editorial Don Bosco. Oblitas Poblete, Enrique. 1968. El idioma secreto de los incas. Cochabamba, Bolivia: Los Amigos del Libro. Rosing, Ina. 1990. Introducción al mundo callawaya: Curación ritual para vencer penas y tristezas. Cochabamba, Bolivia: Los Amigos del Libro. Saignes, Thierry. 1989. "Presentation". In Girault 1989, 9-12. Stark, Louisa R. 1972. "Machaj-Juyai: Secret language of the Callahuayas". Papers in Andean Linguistics 1/2.199-228. Stark, Louisa R. 1985. "The Quechua Language in Bolivia". South American Indian Languages: Retrospect and prospect, ed. by Harriet E. Manelis Klein and Louisa R. Stark, 516-45. Austin, TX: University of Texas Press. Torero, Alfredo. 1987. "Lenguas y pueblos altiplánicos en torno al siglo XVI". Revista Andina 5/2.329-406.
Mednyj Aleut Sarah G. Thomason University of Pittsburgh
1.
Introduction
Mednyj (or Copper Island) Aleut was spoken on Mednyj (Copper) Island, one of the two Commander Islands, which belong to Russia. There are only a few sources available on the language's structure, and all of them rely on data gathered by just three fieldworkers: G. A. Menovščikov (before 1963) and Evgenij V. Golovko & Nikolai B. Vakhtin, who conducted fieldwork in 1982, 1985, and 1987. The major studies of the language, in chronological order, are Menovščikov (1968), Menovščikov (1969), Thomason (1981), Vakhtin (1985), Thomason & Kaufman (1988:Ch. 9.4), Golovko & Vakhtin (1990), Golovko (1994), and Sekerina (1994). The time gap of twenty years (or more) between the two rounds of fieldwork may explain at least in part the occasional discrepancies between Menovščikov's data and Golovko & Vakhtin's data (see below) — that is, the language may have changed significantly during those twenty years as its use declined rapidly. The studies by Menovščikov and by me (with and without Kaufman) are based on Menovščikov's data; the studies by Golovko, Vakhtin, and Sekerina are based on Golovko & Vakhtin's data. The account presented here draws on all these works. The most useful sources of original data and structural analyses are Menovščikov (1969), Vakhtin (1985), and Golovko & Vakhtin (1990); Sekerina (1994) contains useful new analyses of some of Golovko & Vakhtin's data. All the sources listed in this paragraph also provide some discussion of the language's origin and development. Mednyj Aleut is now moribund. According to Golovko (1994:113), there are only 10-12 active speakers left (all of them native speakers), and they no longer live on Mednyj Island: in the late 1960s they were moved to the village of Nikol'skoje on Bering, the other, and much larger, Commander Island.
450
Sarah G. Thomason
Bering Aleut is also moribund, with no more than 7 speakers remaining (ibid.). The situation has been changing fast. According to the census of 192627, the Commanders had an Aleut population numbering "345..., of which 332 were Aleut-speaking" (Levin & Potapov 1956:986, cited in Bergsland 1959:15). Menovščikov reports a population of 300 Aleuts on the Command ers (1968:386), with only 20 or 30 elderly people still speaking Aleut (1969: 133). Comrie (1981:252) says that, according to the 1970 census, 96 of the 441 Aleuts in the USSR spoke Aleut. These figures conflict with Menovščikov's slightly earlier figures. I have no explanation for the discrep ancy, but it's worth noting that census figures on language are often distorted — various social and political factors can skew the results. In any case, it is clear that both Mednyj Aleut and Bering Aleut are rapidly being replaced by Russian; all current speakers of both are also fluent in Russian. When Menovščikov conducted his fieldwork, the native language of the younger generation was already Russian. Mednyj Aleut is a distinct language, with its own grammar. It is not, despite the name, Aleut: it did not arise through a continuous process of unbroken transmission from one generation to the next, but through a more abrupt (nongenetic) process. For over a century, probably continuing after its speakers were moved to Bering Island, it served as the main language of its speech community; and it is mutually unintelligible with real Aleut (including Bering Aleut). Specifically, it is a two-language mixture: most of its struc tures are Aleut, but the entire finite verb inflection, together with pronominal markers that comprise part of the verb complex, are Russian, not Aleut. There are other Russian interference features as well, but they are scattered through out the grammar as ordinary borrowings rather than being wholesale replace ments, as is the case in the finite verb. And some of the other Russian grammatical features also seem to be exclusive to Golovko & Vakhtin's data from the 1980s; the fact that they are not found in Menovščikov's data suggests the possibility that they appeared in Mednyj Aleut between ca. 1960 and the early 1980s, or at least during the lifetimes of Menovščikov's presum ably older consultants — not in the 19th century, when the language emerged. None of Menovščikov's consultants knew that their verbs had Russian end ings, though they did know that their verbal conjugation was different from that of of Bering Island Aleut (1969:133). In the following sections I will first describe the historical background for the emergence of Mednyj Aleut (§2) and then outline the language's
Mednyj Aleut
451
major structural features (§3). In §4, finally, I will argue for a developmental process that involved Aleut/Russian bilinguals, extensive code-switching, and a creative combining of disparate elements into a single language that became the main language of a community.
2.
Historical background
The Aleutian Islands, of which the Commander Islands are (in effect) a western extension, were discovered by Vitus Bering, a Dane sailing under the Russian flag, in 1741. Bering himself drowned when his ship was wrecked on what is now called Bering Island, but survivors from the wreck brought furs back to Russia, thus inspiring others to travel the same route. The first furhunting expedition, led by Emelian S. Basov, set out for the Commanders in 1743, and soon other Russian promyshlenniks (fur trappers and traders) followed, moving from the Commanders eastward into the Aleutians proper (Pavel Golovin, in Dmytryshyn & Crownhart-Vaughan 1979:1). Exclusive rights to exploit the Aleutians and neighboring regions (including the Com manders) were granted to the Russian-American Company some fifty years later, in 1797, and the Aleutians were referred to as the Company's colonies. Earlier promyshlenniks had built settlements on some of the islands, but the Company established its own outposts. By 1819 there were permanent Com pany settlements on a number of islands, including both of the Commanders (ibid., p. 5). The Company administered the islands until it ceased operations in the 1860s; in 1867 the United States bought Alaska and the Aleutians — not including the Commanders — from Russia. The Commanders were uninhabited when they were discovered in 1741, and for some time afterwards. In 1819 their only inhabitants would have been the Company's settlers — probably not all Russians — whose mission was to hunt fur seals. The Company first settled a population of Aleuts on the two islands only in 1826, and Russian employees of the Company kept arriving as well. After some years, according to Menovščikov (who doesn't give an actual date), there were 300 Aleuts and about 30 Russians on Mednyj. Unions between Russian men and Aleut women produced a population called creoles at a very early period; later the creole population grew because, as Russian men completed their fixed terms of service with the Company, many of them married Aleut women and settled permanently in the Commanders and
452
Sarah G. Thomason
elsewhere in the Aleutians (ibid., pp. 70-71). The 300 Aleuts reported by Menovščikov no doubt included both the pure-blood Aleuts and the mixedblood creoles. Overall population estimates for the Aleutians before European contact vary widely. According to one estimate, there were 25,000 Aleuts in the archipelago before 1741 {Encyclopaedia Britannica 1:565); but Golovin says there were nearly 10,000 Aleuts when the Russians arrived (Dmytryshyn & Crownhart-Vaughan 1979:20). Everyone agrees, however, that the number of Aleuts decreased sharply after 1741, both because of diseases introduced by the Russians (venereal diseases and smallpox) and because the Russians killed many Aleuts in battles in the early years. The 1806 census reports a total of 5,234 Aleuts (ibid., p. 20). In 1860, according to Golovin, there were a total of 595 Russians, 1896 creoles, and 4645 Aleuts in the Company's domain (ibid., p. 141). Golovin says that "the creole population increases rapidly" by compari son to the Aleut population (ibid., p. 68). His table provides interesting population figures for the Company's Atkan department, which included the islands of Atka, Amlia, Mednyj, Bering, and Attu, in 1860 (ibid., p. 141): 4 Russians, 197 creoles, and 763 Aleuts — no other groups were represented on these islands — for a total population of 964.1 These numbers are compatible with Menovščikov's figures, but they are difficult to reconcile with Khlebnikov's figures for 1827, as reported by Golovko & Vakhtin (1990: 116). Khlebnikov apparently said that in 1827, just one year after the RussianAmerican Company settled Aleuts on Bering and Mednyj, Bering Island had 17 Russians, 48 creoles, and 45 Aleuts (ibid.);2 the number of creoles looks suspiciously high, even if some of them came from other islands. Golovin's official 1860 figures suggest that it took much longer for creoles to achieve numerical equality with Aleuts in the region. The later figures cited by Golovko & Vakhtin, however, are compatible with the figures given by Golovin for 1860: in 1879 Mednyj Island is said to have had 50% Aleuts and 50% creoles (Benedykt Dybowski,3 in Ljapunova 1982:192, cited in Golovko & Vakhtin 1990:116), and by 1897 the creoles outnumbered the Aleuts, 65% to 35% (S. Patkanov in Ljapunova 1982:192, cited in Golovko & Vakhtin 1990:116). The composition of the Aleut component of the Mednyj Island popula tion during its years as a Company colony is a complicated issue. Bergsland (1959:14) says that the first Aleut settlers were brought to Bering Island from
Mednyj Aleut
453
Atka and to Mednyj Island "mainly from Attu" — both in 1826 — and that, later, eastern Aleuts and other people arrived (Sergejev 1932:129, Jochelson 1933:42f., and Hrdlička 1945:385, all cited by Bergsland). Menovščikov (1968:386) makes the same statement about the first Aleuts on the two Commander Islands. But this situation may have changed rapidly after 1826. The Company's practice of moving Aleuts from one island to another re sulted, in the words of one 19th-century observer, in the following situation: "These people have lost almost entirely their tribal distinctions....[because they] have been transported from island to island...by traders" (Dall 1877:22). Golovko & Vakhtin (1990:98), citing Asinovskij et al. (1983), say that, in the decades after 1826, "several dozens of Aleut and creole families were brought here from the Aleutian Islands, the Kuril Islands, the Pribylof Islands, and...Kamchatka. They formed the core of the Commander Islands popula tion...." In any case, the fact that Bering Aleut reflects Atkan features, while the Aleut component of Mednyj Aleut has Attuan features, presumably means that the early Aleut settlements on the Commanders were split in the manner reported by Bergsland and Menovščikov. The question of significant influ ence from languages other than Aleut and Russian on the development of Mednyj Aleut (and Bering Aleut) can be dismissed, because no such influ ence has been detected or suggested. This does not, of course, mean that nonAleut, non-Russian people were entirely absent from the Commanders; it only means that, if they were there, their languages had no impact on local linguistic developments. Russian promyshlenniks stopped coming to the Commanders during the second half of the 19th century when the fur seal trade fell off, and the Russian population was sharply reduced after 1867, when the sale of the Aleutians to the United States caused the final disappearance of the RussianAmerican Company. The official Russian presence was negligible from then until about the 1940s, when Russian administration (again) brought education and bilingualism in Russian to the Commanders. Since then both Mednyj Aleut and Bering Aleut have lost ground steadily to Russian; as noted above, the process of language shift is now almost complete. The creole population of the colonies was set apart both from the Russian employees of the Company and from the Aleuts during the 19th century. Golovin presents the following picture (Dmytryshyn & Crownhart-Vaughan 1979:17-19). The creoles' legal status was higher than that of the Aleuts: they had no obligations and paid no taxes, and they were completely free to earn
454
Sarah G. Thomason
their own livings (unlike the Aleuts, who were obliged to work for the Company and who earned much lower wages). In the early days of the colonies all the creoles were illegitimate; later on, Russian Orthodox clergy men persuaded Russian men to marry their Aleut or creole mistresses. Rus sians looked down on creoles because of the population's illegitimate origin, so much so that, according to Golovin, the word "creole" was "used as a pejorative" (p. 17). The Aleuts also held the creoles in contempt at the time of Golovin's investigations, and for the same reason: the creoles' mothers were "immoral women" (p. 18). Like the Russians in the colonies, the creoles looked to the Company for their livelihood (many, but by no means all, worked for the Company on contract, as did the Russians). Economically and politically, then, the creoles' position was much better than that of the Aleuts. But they harbored resentments about their social position: they were looked down on by both Russians and Aleuts. This situation had presumably changed by 1920, when Suvorov reported that the inhabitants of the Commander Islands "call themselves creoles and obviously get offended when called Aleuts" (1912:98, cited in Golovko & Vakhtin 1990:116). But Golovin's account indicates that the mid-19th-century position of the creoles, though secure economically, was not enviable socially. After 1867, the legal distinction between creole and Aleut faded when the creoles lost their special privileges, and the social distinction also vanished, though perhaps not immediately (cf. Golovko & Vakhtin 1990: 115). Not surprisingly, there is little detailed information available about the language situation on the Commanders during the 19th century. Bergsland (1994:xxxiii) says that, "From the very beginning, as early as 1745 on Attu (Berg in Jochelson 1933:2), the Russians had Aleuts learn Russian. In Octo ber 1778 Captain James Cook's party found at Unalaska 'about 60 Rus sians...' and reported that Aleut children were taken from their parents and brought up to speak the Russian language (Beaglehole 1967:1141ff.)." It seems reasonable to assume that the Russians' encouraged Aleuts to learn Russian on the Commanders as well as elsewhere. The Russians themselves, obviously, would have spoken Russian, and the Aleuts would have spoken Aleut. Creoles (at least the initial population) would know Aleut, and many of them, especially the ones who were employed by the Company or who lived in households with Russian fathers, would have known Russian as well. All Aleuts and creoles who attended the Company's schools would have learned
Mednyj Aleut
455
Russian there, and those who became Christians (951 of the 964 inhabitants of the Atkan department in 1860, according to Golovin's figures — Dmytryshyn & Crownhart-Vaughan 1979:55) would have heard Russian in church. Golovko (1994:114-15) cites Beklemišev's 1884 statement that "the popula tion [of Mednyj Island] speak a mixed language." Golovko questions whether this mixed language was Mednyj Aleut or merely a high level of codeswitching; in either case the statement provides evidence of language mixing — with or without a stable structure — on the island by 1884. And finally, Ljapunova cites an observation by S. Patkanov, who lived on Mednyj during the 1890s, to the effect that "Aleuts and creoles consituted one group and all spoke Aleut" (1982:192, cited by Golovko & Vakhtin 1990:116). Of course, as Golovko & Vakhtin note, this "Aleut" might well have been Mednyj Aleut.
3.
Structural features
The sketch of Mednyj Aleut structure in this section is based primarily on the accounts in Menovščikov (1969; see also examples and analyses in Thomason 1981 and Thomason & Kaufman 1988), Golovko & Vakhtin (1990), and Sekerina (1994). Because examples are given in those sources, I will not provide many here. Mednyj Aleut phonological features are primarily those of ordinary Aleut, including the small set of non-Aleut Russian phonemes that occur in morphemes of Russian origin. Table 1 displays the three phonemic invento ries: Mednyj Aleut (from Sekerina 1994:20-21), Aleut (from Bergsland 1994: xvii),4 and Russian. In the Mednyj Aleut and Aleut inventories, phonemes in parentheses appear only in Russian morphemes; /v/ is presumably a native Aleut phoneme — it occurs in native Aleut words in the Attuan dialect of Aleut, though not in native words in other dialects. A comparison of the Mednyj Aleut and Aleut arrays in Table 1 shows that only three native Aleut phonemes are missing entirely from Mednyj Aleut: /ð/, /hŋ)/, and /hw/. The phoneme /g/, found only in Russian loanwords in ordinary Aleut, is also missing from Mednyj Aleut, and the Mednyj Aleut phoneme /r/ is found in loanwords in ordinary Aleut. Otherwise the invento ries are identical, including the distinctive vowel length and the two marked phoneme types — the "aspirated" nonobstruents, which are either preaspirated or fully voiceless phonetically, and the uvular obstruents, which
456
Sarah G. Thomason
Table 1. Phonemic inventories of Mednyj Aleut, Aleut, and Russian. (p) (b)
t (d)
č
k
q
(p) (b) (f)
(f)
S
(Š)
X
X
v m hm
z
(ž)
y r)
y
w
n hn 1 hl r
i i: (e)
j hj
t (d)
č
k (g)
q
S
X
X
v ð z m n hm hn 1 hl w (r) hw
h
u u: (o)
y y r) hr)
j hj
i i: (e)
a a:
h
u u: (o) a a:
MEDNYJ ALEUT
p b f v m
p' b' f' v' m'
t d s z n r 1
ALEUT
t' d' s' z' n' r'
c
č
k g x
š ž
i e
u o a
j RUSSIAN
contrast with velars. Russian lacks all three of these features, and the charac teristic Aleut palatal pronunciation of /s/ and /z/ is also quite different from Russian. Sekerina says that the velar/uvular distinction is being lost in Mednyj Aleut (1994:21; cf. Golovko & Vakhtin 1990:101), so that uvulars alternate with velars in Aleut words that originally had uvulars. She also reports the preservation of distinctive palatalization in Mednyj Aleut words of Russian origin, but she does not say whether any of these Russian loanwords are demonstrably old in Mednyj Aleut, as opposed to being recent borrowings (i.e. since the 1940s). This question is crucial because, as Menovščikov observes (1969:122), early Russian loanwords in Aleut were nativized into Aleut phonology, while recent loanwords keep their Russian phonemic and
Mednyj Aleut
457
phonotactic structure. The loss of the velar/uvular distinction at least must be recent, since the alternations still occur. In phonology, then, there is little evidence of Russian influence in the Aleut component of Mednyj Aleut. And because (as in ordinary Aleut) borrowed Russian phonemes are confined to Russian loanwords, the Aleut and Russian components of the mixed language are phonologically distinct. The morphology is a different story. The basic split, as mentioned in §1 above, is between the finite verb morphology and its attendant pronouns, which are Russian, and everything else, which is Aleut. Specifically, noun inflection, derivational morphology in all lexical classes, and — at least according to Menovščikov — nonfinite verb inflection are Aleut. Nouns preserve all the Aleut categories and distinctions: the two noun cases (absolutive vs. relative), the pronominal possessive suffixes, and the singular/dual/plural number distinctions. Russian also has all these catego ries, but there are different distinctions within the Russian categories: six cases instead of two, preposed pronouns instead of pronominal possessive suffixes, and singular vs. plural, but no dual number; in addition, Russian has grammatical gender (masculine/feminine/neuter, animate/inanimate), which both Aleut and Mednyj Aleut lack — though Mednyj Aleut does have gender distinctions in the Russian pronouns that constitute part of its verb complex. Both Aleut and Russian personal pronouns are used in Mednyj Aleut, though usually in different constructions. The Aleut pronouns are used as direct objects when they would be used as direct objects in ordinary Aleut; Russian pronouns (but never Aleut pronouns) are used as subjects in the past tense, where Russian lacks person marking (Golovko & Vakhtin 1990:104). Russian and Aleut pronouns do sometimes appear in the same constructions, however, as in examples (1) and (2). In (1) the benefactive object is expressed by an Aleut pronoun, while in (2) it is expressed by a Russian pronoun (both examples are from Golovko & Vakhtin 1990:105):5 (1)
àsu-x ŋun aynu:-j! nail-ABS lsg.OBJ drive.in-IMP "Drive in a nail for me!"
(2)
ya tibe cíbu-x ukayla:ya: sa:-l 1sg.SUBJ2sg.OBJparcel-ABS bring-PAST "I brought you a parcel."
458
Sarah G. Thomason
As we will see below, the past tense suffix -/ is of Russian origin. Both Aleut and Russian demonstratives are used in Mednyj Aleut, but to judge by Sekerina's examples, e.g. (3) and (4) (her examples 5a and 2b, p. 24), they do not appear in the same constructions: (3)
híŋa tayágu-x sisaxta:-l. this man-ABS lose-PAST "This man is lost."
(4)
eta moj asxinu-ŋ. this lsg.poss daughter-lsg.poss "This is my daughter."
Ex. (4) displays two especially striking features. First, it has double marking for lsg.poss — both the Russian possessive pronoun moj and the Aleut lsg.poss suffix -ŋ. And second, the Russian possessive pronoun lacks the feminine gender suffix -a that is required in the corresponding Russian sentence, eto moja doč' "This is my daughter". Aleut and Mednyj Aleut lack a lexical category of adjectives, expressing adjectival notions instead by means of possessed nouns or verbs (Sekerina 1994:24). Mednyj Aleut has a complete Aleut system of nominal and verbal deriva tional morphology, with suffixes to express such categories as agent, instru ment, location, and diminutive in nouns and transitive, detransitive, causative, resultative, inchoative, distributive plural, and desiderative in verbs. There are also derivational deverbative nominal constructions and deverbative ver bal constructions (see Golovko & Vakhtin 1990:102 and Vakhtin 1985:38-40 for details and examples). There are no Russian derivational patterns. The entire system of finite verb inflection in Mednyj Aleut is comprised of Russian grammatical elements, at least in Menovščikov's data; Golovko & Vakhtin found more variation, with Aleut pronouns used in unaccusative constructions. The Russian conjugational system is largely intact in Mednyj Aleut, though the Russian feminine gender suffix -a is used only sporadically in past-tense forms (Golovko & Vakhtin 1990:109). In present-tense verb forms, each tense/person/number combinaton is expressed by a single Russian suffix (see Menovščikov 1969:132 and Thomason & Kaufman 1988:234-35 for examples). This Russian feature is flexional in nature, quite different typologically from the heavily agglutina tive Aleut system, in which tense and person/number are always expressed by
Mednyj Aleut
459
separate suffixes. Other features of both the Russian system and the ordinary Aleut system are agglutinative, and in these cases the Russian forms have simply replaced the Aleut forms. Here too, however, some of the replace ments have effected typological change. So, for instance, the two Aleut negative verb suffixes ("actual negation" and "non-actual negation" — Golovko & Vakhtin 1990:109) have been replaced in Mednyj Aleut by a single prefix ni-, whose source is the Russian negative particle ne (which precedes the verb in Russian). The Aleut negative suffixes are still used systematically in Mednyj Aleut — but only on nouns, not on verbs. The analytic future tense of Aleut has been replaced by the analytic future of Russian. The Russian imperative in -j is also used in Mednyj Aleut; Golovko & Vakhtin (1990:111) note that Attuan Aleut has an identical imperative marker, -j{a) (compare Bering Aleut -ð), so that this Mednyj Aleut suffix may have two sources. Because Russian has no dual number, Mednyj Aleut has none in finite verb inflection — though nouns and pronouns preserve the Aleut dual as well as the singular and plural distinctions. Menovščikov reports only one means of past-tense formation: the verb takes the Russian suffix -/ and is obligatorily preceded by Russian subject pronouns (which are needed because the Russian past-tense verb is not marked for person). Golovko & Vakhtin, however, found two different past tense constructions. Both have the suffix -l; but while one construction is the one Menovščikov described (except for the use of Aleut pronouns instead of Russian ones in unaccusatives), the other is agglutinative, with the Russian present-tense person/number suffixes following the past-tense suffix -/. In plural forms the Russian past-tense plural suffix -i intervenes between the -/ and the person/number suffix (Golovko & Vakhtin 1990:108). The second of these constructions matches the tense + person/number suffix sequences of ordinary Aleut. Menovščikov draws a sharp contrast between finite verb forms (and pronominal attachments), on the one hand, and nonfinite verb forms on the other. In nonfinite forms the pronominal and other orphological features are Aleut, not Russian, as in the gerund construction tiŋ ayjaxčayu-ŋ "if I go" (lit. lsg. go.GERUND-lsg.POSs). Of course these forms are morphologically nomi nal, with the addition of the preposed pronoun; but they function as verbs syntactically, so the contrast with finite verb forms is noteworthy. Sekerina calls Mednyj Aleut syntax "very Russianized" (1994:28), but this may be an exaggeration. The main syntactic features that can safely be
460
Sarah G. Thomason
attributed to Russian are a relatively free word order (as opposed to more rigid Aleut SOV word order), the lack of a copula in the present tense (like Russian, unlike Aleut), complex sentence constructions in which Russian complemen tizers and conjunctions appear (ibid., 29), and modal constructions in which Russian modals have replaced suffixal Aleut forms (Golovko & Vakhtin 1990:110). But the use of complementizers and conjunctions is merely lexical borrowing, unless the constructions differ typologically from corresponding Aleut constructions.6 In any case, there are also numerous Aleut syntactic features in Mednyj Aleut. Perhaps the most striking one is an agreement pattern in which a verb agrees with the subject in person but with the topic in number, as in (5), in which the plural noun "snot" is the subject but the topic is lsg., so that the verb has 3sg. agreement (ibid., p. 106): (5)
kanú-ni-ŋ Snot-pl-lsg.POSS
yu:-it. run-3sg.PRES
"My nose is running." Golovko & Vakhtin point out that there are also examples in which number agreement is with the subject rather than with the topic, as in Russian (ibid., p. 107). Nevertheless, as they say, the unusual Aleut pattern is preserved at least as one of two alternative patterns of number agreement. Other Aleut syntactic features in Mednyj Aleut mentioned by Golovko & Vakhtin are the comitative construction and the use of postpositions (though Russian prepositions are also used) (1990:105). Finally, the Mednyj Aleut lexicon contains many Russian words, includ ing nouns, verbs, adverbs, and function words (conjunctions, prepositions, etc.), as well as numerous calques. Of course, this is also true of ordinary Aleut; in the Aleut of Alaska (i.e. not including the Commander Islands), according to Bergsland, more than 800 Russian loanwords have been re corded (1994:xxxiii). Some Russian words might well have come to Mednyj Island either with the first Aleuts in 1826 or with Aleuts who came soon afterward, and thus before Mednyj Aleut existed. An added complication is the need to distinguish Russian words that entered Mednyj Aleut during its formative period, or soon afterward, from Russian words that came in during the second wave of bilingualism in Russian, which started in the 1940s. As far as I know, no one has studied the Mednyj Aleut lexicon in an effort to make this distinction. In any case, the basic vocabulary of the language is mainly of Aleut origin. Sekerina says that, for the corpus she analyzed, 94% of the verbs and 61.5% of the nouns are from Aleut (1994:29).
Mednyj Aleut
461
To sum up, then, Mednyj Aleut grammar and lexicon are primarily of Aleut origin, with the sole exception of the finite verb morphology, which has been lifted wholesale — both categories and morphemes — from Russian. There are also many Russian words in the language. There are some Russian phonemes, but they are confined to Russian loanwords; and there are a number of syntactic patterns which show Russian influence. There are also some Aleut features even in the finite verb complex, notably the special topicnumber agreement pattern, the preposed Aleut pronouns with unaccusatives, and the Aleut agglutinative tense + number + person/number pattern in one of the two alternative past-tense formations. (In this last construction, the suf fixes themselves are Russian; only the pattern is Aleut.) There seem to be a few differences between Menovščikov's data and Golovko & Vakhtin's, though all these authors' analyses are too sketchy to be sure about this; these differences might reflect changes in the language between the time of Menovščikov's fieldwork on Mednyj and the time of Golovko & Vakhtin's on Bering. In general, although the more recent fieldwork has provided valuable new data and more extensive data than Menovščikov published, Golovko & Vakhtin's results do not change the picture significantly: Mednyj Aleut is a mixed language, comprised of largely intact systematic components from two different languages that are typologically and genetically unrelated. In my final section I will consider the origin of this mixture.
4.
How and why did Mednyj Aleut arise?
As with most contact languages, the precise period during which Mednyj Aleut emerged is unknown and unknowable. The terminus a quo is 1741, when Russians first encountered Aleuts; the terminus ad quem is the first publication of Mednyj Aleut data, a glossary containing about 1500 items (Menovščikov 1963, cited by Bergsland 1994:xliii). Since Menovščikov's consultants were elderly in about 1960, we can safely push the terminus ad quem back to ca. 1900. And since there is no evidence of any kind to suggest that Mednyj Aleut developed on some island other than Mednyj, the simplest (and therefore preferable) assumption is that it arose on Mednyj Island, and thus not earlier than 1826. This still leaves a range of over 70 years for its emergence, but there is no solid evidence that would permit us to narrow this range.
462
Sarah G. Thomason
We can, however, draw some inferences from the documented historical setting and from the language's structure. The discussion in this section parallels the analysis in Thomason (1981) and Thomason & Kaufman (1988: 233-38); just as the more detailed information about the structural features leaves the older picture intact, so the more detailed information about the historical background (in §2 above) leaves my origin theory largely intact. First, the implications of the language's structure: because both the Aleut parts and the Russian parts of Mednyj Aleut are fully elaborated Aleut and Russian — not simplified overall, and not seriously distorted except in relatively minor details — the people who created the language must have been bilingual in Aleut and Russian. That is, imperfect learning played no role in the development of Mednyj Aleut. This fact immediately rules out the possibility that Mednyj Aleut arose as a pidgin, as Golovko & Vakhtin (1990) have suggested (cf. Golovko 1994). The reason is that processes of pidgin genesis are rather closely allied to second-language acquisition processes, and in both cases imperfect learning is a major factor in determining the structure of the lexifier language or target language. The linguistic evidence does not permit any determination of whether the Aleut/Russian bilinguals who created Mednyj Aleut had one of their two languages — Aleut or Russian — as their dominant language; the only certain point is that they must have been truly bilingual. Moreover, the structure of Mednyj Aleut shows clearly that the language is basically Aleut, with Russian features incorporated into an Aleut base. The fact that the most surprising Russian features comprise a complete system of inflectional verb affixes doesn't alter the nature of the mixture. The language is mainly Aleut, both lexically and grammatically, with Russian contributing one entire grammatical subsystem and numerous other more typical borrow ings, including such features as loanwords, mostly in the nonbasic vocabulary; superficial word-order features; function words; and phonological features, notably the loss of the velar/uvular distinction. Second, the implications of the historical setting: although there is very little information about language use in the early period of Aleut settlement on Mednyj Island, it seems safe to assume that Aleuts and especially creoles had enough exposure to spoken Russian and, in some cases, to written Russian to become at least partially bilingual. Given their Russian fathers, their privileged legal status, and their access to Company officials, the creoles were probably skilled bilinguals from the beginning. It is harder to estimate
Mednyj Aleut
463
the probability that Russians — in particular those Russians who stayed on Mednyj after the end of their contract service to the Russian-American Company — became fluent in Aleut. But it is not unlikely that at least some Russians did learn Aleut, perhaps a sizable percentage of the Russians who settled permanently on the island. Significantly, as the historical sketch in §2 showed, the creole population emerged on the Aleutians and the Commanders early in the 19th century as a separate caste, distinct both from the Russians and from the Aleuts. Looked down on (according to Golovin) by both of the other groups, but with special legal and economic privileges, they occupied an uncomfortable position in their society. Other mixed-blood populations, such as the Métis who devel oped the mixed language Michif (see Bakker & Papen, this volume), and other groups set apart from their milieu, such as the Quechua speakers who developed Media Lengua (see Muysken, this volume), were also faced with the problem of locating themselves within the larger society. Like the Métis creators of Michif and the Quechua-speaking creators of Media Lengua, the mixed-blood Aleut/Russian creoles were surely the creators of the mixed language Mednyj Aleut. Of all the people on Mednyj Island, they were the most likely to be fully bilingual as a group, and also the most likely to feel separate enough from the other groups to want a new language of their own. I therefore agree in part with Golovko, who proposes that the reason for the language's emergence was "the aspiration of a group of people for a separate identity" (1994:118) — but only in part, because in my view the people already had a separate identity; what they aspired to was a salient social symbol of that identity, namely, a special language.7 Any proposal about why such a language arose on that particular island is at best highly tentative. Not only is there no direct evidence to support any specific explanation, but there are a great many other contact situations with similar social conditions and no resulting contact language. In particular, all of the Company's other islands, including Bering Island, also had populations comprised of Russians, creoles, and Aleuts (and often other groups as well); so why did a mixed language arise on Mednyj and not elsewhere, say on Bering? There is no answer to this question, although the numerical differ ence noted by Menovščikov (1969) — more Aleuts on Bering, proportionally, than on Mednyj — is likely to be one factor, because with fewer. Russians present it would have been harder for creoles and Aleuts to become fluent in Russian. In general, however, trying to predict the emergence of mixed
464
Sarah G. Thomason
languages (including pidgins, creoles, and two-language mixtures like Mednyj Aleut) is like trying to predict language change: the enterprise is risky to the point of rashness. There are too many crucial factors that can't be known and therefore can't be taken into account. One crucial question that can be usefully addressed is this: by what mechanism, or mechanisms, did Mednyj Aleut arise? I have argued elsewhere (1996) that there are more mechanisms of interference than have generally been recognized — specifically, that not all contact-induced language change comes about through code-switching. Other mechanisms are code alternation (the use by the same speaker of two different languages, but in different settings), passive knowledge of another dialect or language, "negotiation", second-language acquisition strategies, and changes brought about by speak ers' deliberate decisions. In the development of Mednyj Aleut, it seems likely that both code-switching and deliberate decisions by speakers — individual creativity — were operative. Code-switching, including switching between two languages within the same sentence, would certainly have been available to a population of fluent Aleut/Russian bilinguals. In principle, such speakers could incorporate Rus sian into their Aleut or Aleut into their Russian. The particular type of incorporation that needs to be accounted for is adding Russian finite verb endings to Aleut verb stems. Given the relative ease with which lexical items are code-switched (and borrowed), by comparison to inflectional affixes, one might expect the process to involve bringing Aleut stems into a Russian matrix rather than bringing Russian affixes into an Aleut matrix. But if this is what happened, we then need to explain why the rest of the language's structure — including many grammatical morphemes and the vast majority of the basic lexical items — is Aleut, not Russian. And if the code-switching went in the other direction, taking only the finite verb inflection from Russian, then we need to explain that surprising choice. So a second step is needed, one in which code-switching speakers fix one part of the grammar in a mixed form and then put that grammatical system as a whole together with structures from just one of the two languages, Aleut. There is no way to tell whether interference features from Russian other than the finite verb morphology entered Mednyj Aleut before, during, or after the Russian morphological system was grafted onto an Aleut foundation; but since all the other borrowings from Russian can be matched in ordinary contact situations all over the world, the finite verb morphology is the only
Mednyj Aleut
465
grammatical subsystem that requires a particular explanation. This is where I would appeal to linguistic creativity, to decisions made by the bilingual creators of the mixed language: creativity was needed to establish Mednyj Aleut as a permanent coherent linguistic system. This may in fact be a universal of two-language mixtures whose primary function seems to be to serve as an identifying symbol of a new ethnic (sub)group; certainly it applies also to Michif and Media Lengua, for instance. A final question concerns the nature of the mixture. Why should only the finite verb morphology be so deeply changed through borrowing, while other grammatical subsystems remain Aleut? One possible answer, suggested in Thomason (1981) and Thomason & Kaufman (1988:237), is that the Russian and Aleut systems of noun inflection fit together fairly well: they have similar inflectional categories (although Aleut has no grammatical gender), and often even similar items within each category, e.g. singular vs. plural number (although Aleut also has a dual). The verb systems, by contrast, differed greatly. There was some overlap in inflectional categories, but the basic structures of the two systems are dissimilar, and the number of Aleut verb forms is immense. As a result, Russians who knew some Aleut would be likely to find the noun more manageable than the verb; and that in turn makes it more likely that the speakers who developed the mixed language might have used the Russian endings as one means of lessening the learning burden for Russian (semi-)bilinguals who wanted to use it (notably the Russians who intermarried with Aleuts, and then, later, the Russians who intermarried with creoles). If this inference is valid, it has implications for the period of origin for Mednyj Aleut. The language is most likely to have arisen soon after a creole population became established on Mednyj Island, and certainly before most Russians left the island after 1867: after 1867 there would have been few Russians to learn the mixed language. An early origin is also favored by the likelihood that the creoles would have felt most separated from the rest of the society while there were both Russians and Aleuts on the island — especially because, as noted above, the distinction between creoles and Aleuts largely or entirely vanished once the creoles lost their privileged legal status as a result of the Company's departure. I should emphasize that my hypothesis about the development of Mednyj Aleut cannot be tested; its persuasiveness must rest on the degree to which the various pieces of the proposal fit with the known facts about processes of
466
Sarah G. Thomason
interference and about the particular historical setting in which this contact language arose. Some aspects of the proposal are solidly established, such as the claim that only fluent bilinguals could have created the language; other aspects are more speculative and therefore more tentative. The hypothesis does, however, account for the facts, and I have seen no other hypothesis that does as good a job in accounting for all the historical and linguistic facts.
Notes 1.
Ljapunova reports figures of 90 people on Mednyj and 300 on Bering (1982:191, cited by Golovko & Vakhtin 1990:113), but either she does not break the numbers down into ethnic groups or Golovko & Vakhtin do not provide the breakdown.
2.
Kiril T. Khlebnikov's description of Russian America, in a series of reports from 1817 to 1832, is described by Dmytryshyn & Crownhart-Vaughan as "classic" (1979:ix; see Khlebnikov 1976). See also two references cited in Golovko & Vakhtin (1990) — Khlebnikov (1979) and Khlebnikov (1985).
3.
Benedykt Dybowski (1885) provides 33 Bering Island words and two Mednyj Island words (Bergsland 1994:xli).
4.
Bergsland's table of Aleut phonemes displays both Atkan and Eastern variants, and he also mentions a specific Attuan feature (the phoneme /v/). My Table 1 gives only the Atkan phonemes, together with the Attuan /v/, including — in parentheses — phonemes found only in Russian loanwords. Bergsland's phonemic array does not completely match Menovščikov's (1968:387), but that may be because Menovščikov's phonemic analysis differs from Bergsland's rather than because of any phonetic differences between general Atkan and the Aleut of the Commander Islands. Menovščikov's list of phonemes has neither aspirated nonobstruents nor long vowels (he says that these features lack phone mic status — ibid., pp. 387, 388). He includes a phoneme transcribed as / followed by a hard sign, and identifies it as a tongue-back affricate; it isn't clear what he has in mind here, but perhaps this is connected with Bergsland's aspirated phoneme /h1/.
5.
The following abbreviations are used in examples: 1 = 1st person, 2 = 2nd person, 3 = 3rd person, ABS = absolutive, IMP = imperative, OBJ = object, pl = plural, POSS = possessive, PRES = present tense, sg = singular, and SUBJ = subject.
6.
Golovko & Vakhtin (1990:110) also comment that Mednyj Aleut no longer has Aleut sentence syntax in compound sentences, but — aside from the modal construction — they give no examples that clearly show a nonlexical difference between the Mednyj Aleut constructions with Russian lexemes and ordinary Aleut constructions.
7.
I also disagree with much of the rest of Golovko's reasoning on this point. For instance, I see no evidence to support his claim that "only people with a good sense of humour could launch a new mixed language into the world" (1994:119).
Mednyj Aleut
467
References Asinovskij, A.S.; Nikolai . Vakhtin; and Evgenij V. Golovko. 1983. "Etnolingvističeskoje opisanije Komandorskix Aleutov". Voprosy Jazykoznanija 6.108-16. Beaglehole, John Cawte, ed. 1967. The voyage of the Resolution and discovery 17761780, part 2. Cambridge. Beklemišev, O. 1884. Komandorskix ostrovax i kotikovom promyšle. St. Petersburg. Bergsland, Knut. 1959. Aleut dialects of Atka and Attu. (Transactions of the American Philosophical Society, n.s. 49.) Philadelphia: The American Philosophical Society. Bergsland, Knut. 1994. Aleut dictionary/Unangam tunudgusii: An unabridged lexicon of the Aleutian, Pribilof and Commander Islands Aleut language. Fairbanks: Alaska Native Language Center, University of Alaska. Comrie, Bernard. 1981. The languages of the Soviet Union. Cambridge: Cambridge University Press. Dall, William H. 1877. Tribes of the extreme Northwest — the Aleutians & adjacent territories [with articles by William H. Dall, J. Furuheim, and George Gibbs]. (Contributions to North American ethnology, vol. I.) Washington, DC: Government Printing Office. [Facsimile reproduction by The Shorey Book Store, Seattle, 1970.] Dmytryshyn, Basil, & E.A.P. Crownhart-Vaughan, eds. and translators. 1979. The end of Russian America: Captain P. N. Golovin's last report, 1862. Portland: Oregon Historical Society. [The Russian original was published in Morskoj Sbornik in 1862.] Dybowski, Benedykt. 1885. Wyspy Komandorskije. Lwow. Encyclopaedia Britannica. 1966. Chicago & London: Encyclopaedia Britannica, Inc. Golovko, Evgenij V. 1994. "Copper Island Aleut". Mixed languages, ed. by Peter Bakker & Maarten Mous, 113-21. Amsterdam: Institute for Functional Research into Lan guage and Language Use (IFOTT), University of Amsterdam. Golovko, Evgenij V., & Nikolai B. Vakhtin. 1990. "Aleut in contact: The CIA enigma". Acta Linguistica Hafniensia 72.97-125. Hrdlička, Aleš. 1945. The Aleutian and Commander Islands and their inhabitants. Philadelphia: Wistar Institute of Anatomy and Biology. Jochelson, Waldemar. 1933. History, ethnology and anthropology of the Aleut. Washing ton, DC: Carnegie Institution of Washington Publication 432. Khlebnikov, Kiril T. 1976. Colonial Russian America: Kyrill T. Khlebnikov's reports, 1817-1832. Tr. by Basil Dmytryshyn & E.A.P. Crownhart-Vaughan from Materialy dlja istorii russkix zaselenij po beregam vostocnago okeaną, supplement to Morskoi Sbornik (the official journal of the Naval Ministry), St. Petersburg, 1861. Portland: Oregon Historical Society. Khlebnikov, Kiril T. 1979. Russkaja Amerika v neopublikovannix zapiskax K. T Khlebnikova. Compiled by R. G. Ljapunova & S. G. Feodorova. Leningrad: Nauka. Khlebnikov, Kiril Timofejevich. 1985. Russkaja Amerika v "Zapiskax" Kirila Khlebnikova: Novo-Arxangel'sk. Compiled by S. G. Feodorova. Moscow: Nauka. Levin, Maksim G., and Leonid P. Potapov, eds. 1956. Narody Sibiri. Moscow: Izdatel'stvo Akademii Nauk SSSR. Ljapunova, Roza Gavrilovna. 1982. "Ethnohistoire des aléoutes des îles du Commandeur". Inter-Nord 16.189-203.
468
Sarah G.
Thomason
Menovščikov, Georgij Aleksejevič. 1963. "AleutRussian glossary, Atkan of Bering Island and Attuan of Copper Island". Published in Jazyki i Toponomija 1977.167-98. Tomsk. Menovščikov, Georgij Aleksejevič. 1964. "K voprosu o promtsajemosti grammatičeskogo stroja jazyka". Voprosy Jazykoznanija 5.100-106. Menovščikov, Georgij Aleksejevič. 1968. "Aleutskij jazyk". Jazyki narodov SSSR, vol. 5: Mongol' skie, Tunguso-Man'čzurskie i Pleoaziatskie jazyki, 386-406. Leningrad: Nauka. Menovščikov, Georgij Aleksejevič. 1969. "O nekotoryx social'nyx aspektax èvoljucii jazyka". Voprosy sociaVnoj lingvistiki 110-34. Leningrad: Nauka. Sekerina, Irina A. 1994. Copper Island (Mednyj) Aleut (CIA): A mixed language. Languages of the World 8.14-31. Sergejev. 1932. Yazyk i Pis'm 3. [Title of article not available to me.] Suvorov, E.K. 1912. Komandorskije ostrova i pusnoj promyšel na nix. St. Petersburg. Thomason, Sarah G. 1981. "Are there linguistic prerequisites for contact-induced lan guage change?" Paper presented at the Tenth Annual University of WisconsinMilwaukee Linguistics Symposium: Language Contact. [Published in revised form as "Contact-induced language change: Possibilities and probabilities", Akten des 2. Essener Kolloquiums über 'Kreolsprachen und Sprachkontakte', ed. by Norbert Boretzky, Werner Enninger, & Thomas Stolz, 261-84. Bochum: Studienverlag Dr. N. Brockmeyer, 1986.] Thomason, Sarah G. 1996. "On mechanisms of interference". Studies for Einar Haugen, ed. by Stig Eliasson & Ernst Håkon Jahr. Berlin: Mouton de Gruyter, to appear. Thomason, Sarah G., and Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Vakhtin, Nikolai B. 1985. "Nekotoryje osobennosti russko-aleutskogo dvujazyčija na Komandorskix ostrovax". Voprosy Jazykoznanija 5.35-45.
M A ' A (MBUGU) Sarah G. Thomason University
1.
of
Pittsburgh
Introduction
Ma'a (also called Mbugu) is a mixed language spoken in several separated areas of the Usambara Mountains of northeastern Tanzania. Its grammar is almost entirely Bantu; its lexicon comes variously from Bantu languages (mainly Pare, but also Shambaa and Swahili) and non-Bantu languages (mainly Cushitic). Most of the basic vocabulary is of Cushitic origin. No region is exclusively populated by Ma'a speakers (Mous 1994:177). They live in close proximity to speakers of two Bantu languages, the local language Shambaa and the language of Pare immigrants from the Pare Mountains (also in Tanzania, northwest of the Usambaras). Ma'a is the group's self-name; Mbugu is the name given to them by others. Both names are used in the literature, though Mbugu is more common than Ma'a. The name problem is further complicated by the fact that all members of the group speak both a Bantu language — which everyone, group members included, calls Mbugu — and the mixed language Ma'a. The Bantu language Mbugu is very similar to Pare. When talking to outsiders the Mbugu often refer to the mixed language as Mbugu, usually "real Mbugu" or "old Mbugu" to distinguish it from the Bantu language Mbugu; but when talking among themselves they always call it Ma'a (Brenzinger 1987:6). Following the lead of Brenzinger (1987) and Mous (1994), I will refer to the people as Mbugu. I will call their Bantu language Mbugu too, but will use Ma'a as the name of their other language — like Tucker & Bryan (1974) but unlike Mous, who calls them Normal Mbugu (= Mbugu) and Inner Mbugu (= Ma'a), respectively.
470
Sarah G. Thomason
Brenzinger (1987:13) cites the following census figures for the Mbugu people: 2,000 in 1931, 13,454 in 1948, 11,682 in 1957, and 29,871 in 1967. But as Brenzinger points out, interpreting these figures is very difficult, both for cultural reasons and because the Mbugu were lumped together in the 1967 census with another ethnic group. Moreover, the number of speakers of Ma'a may not be identical to the number of members of the ethnic group; in the 1948 census, for instance, the number of speakers is given as "about 11,000" (Whiteley 1960:96) or 12,000 (other authors). According to Mous, all its speakers know Shambaa as well as Ma'a, and most know Swahili as well; in addition, they all know Mbugu, and both Ma'a and Mbugu are learned first by children, possibly at the same time. Information about Ma'a can be found in a variety of sources ranging from 1885 (Farler) to 1994 (Mous). Mous (1994) provides the most complete description of the language, and his discussion of the circumstances of its use is also especially valuable. Most authors who have discussed Ma'a have tried to account for its unusual mixed nature. Among the more recent authors who have considered the question of its origin are Goodman (1971), Thomason (1983), Brenzinger (1987), Thomason & Kaufman (1988), Sasse (1992), and Mous (1994); see Thomason (1983) and Brenzinger (1987) for discussion of earlier authors' views. Because of the influential work of Greenberg (1955, 1963), standard classifications of Ma'a place it in the Cushitic branch of the Afro-Asiatic language family. Other authors, notably Mous (1994) and Möhlig (1983), argue that, far from being a Cushitic language, Ma'a is a Bantu language with some added lexicon (mainly of Cushitic origin). In Thomason (1983) I argued that Ma'a has too little Cushitic grammar to be considered a Cushitic language, in spite of its largely Cushitic lexicon (see also Thomason & Kaufman 1988:223-28). Here I will argue that Ma'a is not a Bantu language either, and that its mixed structure is in part the result of several centuries of more or less gradual Bantuization and in part — as Mous (1994:199) has proposed — the creation, by deliberate decision, of its speak ers. In the following sections I will survey the historical background for the emergence of present-day Ma'a (§2) and outline the language's salient lin guistic structures (§3). I will then consider the probable historical route by which Ma'a reached its present state (§4).
Ma'a (Mbugu) 2.
471
Historical background
According to Brenzinger, the first known European contact with the Mbugu was in 1852 (Krapf 1964 [1858]:295, cited in Brenzinger 1987:19). Actual historical documentation on the Ma'a language is scanty until well into the 20th century, much too late to establish a firm time line for the language's development. Most of what we know comes from oral histories of the Pare and the Shambaa, the Bantu neighbors of the Mbugu people, and of the Mbugu themselves. According to the main oral tradition, the original Mbugu homeland was in Lukipya, which is Laikipya in present-day Kenya, now home to speakers of Masai, a Nilo-Saharan language (Mous 1994:177). From there, harassed by the Masai (and possibly by other groups), the Mbugu moved to the northern Pare Mountains. They were said to have had problems with the Pare too, and, according to the traditions of the Pare (Kimambo 1969) and the Shambaa (Feierman 1974), the Mbugu people came to the South Pare mountains about 350 years ago, settling in Vudee in the northwestern region of South Pare. Later, in "an attempt to resist encroachment on their ways of living", many Mbugu left the Pares and moved to the Usambara mountains (Kimambo 1969:62). The Mbugu clans that remained in South Pare shifted to Pare, though apparently not right away: an elderly Mbugu, Mzee Muse Njarita Msemo, who was born in 1894, told Brenzinger that his grandfather spoke "another language of his own in addition to Pare" but refused to teach it to his children — because, said Mzee Msemo, the Mbugu wanted to fit into their neighborhood and therefore to adopt the local language (Brenzinger 1987: 23). Overall, the oral history gathered by various fieldworkers highlights repeated problems with intruders and neighbors. The Mbugu were cattleherders; the Masai stole their cattle, and the agriculturalist Pare harassed them. Various Mbugu clans moved from place to place in an effort to remove themselves from trouble spots, and the routes they followed were complex, though a general movement from northwest to southeast can be traced (from Laikipya to the north Pares to the south Pares to the western Usambaras). In more recent times their problems with their neighbors can sometimes be identified rather precisely. Late in the 19th century, for instance, some Shambaa chiefs sought to enrich themselves by selling Mbugu people into slavery; the Mbugu reacted by barricading themselves behind palisades "where they could live in peace and not be enslaved" (Feierman 1974:172).
472
Sarah G. Thomason
Most Mbugu clans that moved to the Usambaras did not at first shift either to Pare or to Shambaa — with the exception of one small group, the Nango, who moved to the Usambaras before the main migrations and appar ently shifted to Shambaa shortly after arriving there. According to Feierman (1974:77), "the Nango chose to live the Shambaa way," but the clans that retained Ma'a did not. Both Kimambo and Feierman portray the Mbugu people as determined to resist cultural assimilation, and this view is supported by other authors' comments as well. For instance, Copland says that "the Mbugu were not disposed to sink their individuality as the Nango had done" (1933-34:242). Green observes that "the Wambugu are a reserved and uncommunicative people regarding their past history and present customs.... They keep them selves apart from the indigenous Wasambaa [Shambaa]" (1963:175); and he refers to their "present independent attitude" and to their "extreme conserva tism" as seen in their resistance to cultural influence both from the Shambaa and from Europeans (p. 177). Kimambo and Feierman also noted this feature of the Mbugu community (see their comments quoted above), as did other observers. The crucial points about Mbugu history, then, are the forced wanderings and the desire, on the part of those clans that have maintained Ma'a, to keep themselves separate from their neighbors. (The episodes of harassment are of course likely to be connected with the long-standing resistance of the Mbugu to total cultural assimilation.) Maintaining Ma'a is both a symbol and a mechanism of their cultural conservatism, and it works: as Mous points out, Ma'a "is considered incomprehensible and very difficult by the neighbours" (1994:176).
3.
Structural features
Grammatical sketches of Ma'a appear in three works: Green (1963), Tucker & Bryan (1974), and Mous (1994). Tucker & Bryan and Mous also provide systematic comparison with corresponding Mbugu structures. Less system atic grammatical notes can be found in other sources; of all the sources, only Copland offers a Ma'a text (1933-34:243-245). I will not provide a grammati cal description here, because Mous's 1994 survey is readily available. My goals in this section are to summarize the grammatical features of Ma'a and to
Ma'a (Mbugu)
473
compare them, as sets of features, to Mbugu and also to Cushitic, which is the source of most (not all) of the non-Bantu elements in Ma'a. 1 This comparison will help to determine the extent to which Ma'a can reasonably be considered a separate and independent language. Mous (1994) argues that Ma'a is not an independent language, but merely a lexical register of Mbugu. In his view there is just one language, Mbugu; Mbugu and Ma'a differ only in the phonological shapes of a subset of lexical and grammatical morphemes, including most of the basic vocabulary. This linguistic characterization appears to fit most of his data quite well, and it is also consistent with most of the data reported by earlier fieldworkers. Whether it justifies analyzing the two forms as variants of a single language is debatable (see below), though Mous is surely right as far as the current sociolinguistic status of Ma'a is concerned. More significantly for the histori cal issue, Mous also apparently believes that Ma'a never was a separate language; I will argue against this view in §4. The Ma'a and Mbugu phonemic inventories are alike in most respects. The major exceptions are the presence in Ma'a, primarily in words of Cushitic origin, of three phonemes unknown in the relevant Bantu languages: /ł/, which is a highly marked phoneme type in the world's languages, and the more ordinary /x/ and /?/. Occasionally /ł/ appears in words of Bantu origin. Mous argues convincingly that such occurrences are due to a more or less deliberate Ma'aization of the lexicon of Ma'a — that is, /ł/ has become a symbol of the differentness of Ma'a. It is, he says, the sound that is pointed to as an example of why Ma'a is so difficult (1993:10). The Ma'a lexicon, including non-Bantu words, contains marked sounds that occur in Bantu but not in Cushitic, notably prenasalized voiced stops and allophonic implosion. Aside from /ł/, however, all marked Cushitic phoneme types are absent from Ma'a, including ejectives, labialized dorsals, retroflex stops, and phonemic vowel length. Most other Ma'a phones and phonemes are common in both Bantu and Cushitic languages. All the productive inflectional and derivational Ma'a morphology com prises fully elaborated Mbugu patterns, including the grammatical mor phemes that express them. The most striking morphological features are the typical Bantu noun-class system, including the usual agreement prefixes on verbs and adjectives, and a complete Bantu system of verb inflection with such complexities as negative marking that varies according to person and tense. In all these respects Ma'a therefore matches Mbugu and other Bantu
474
Sarah G. Thomason
languages exactly, and differs sharply from Cushitic: unlike Cushitic, Ma'a has no masculine/feminine gender distinction, no optionality in singular/ plural marking, no singulative, few inflectional suffixes, prominent tense (in Cushitic aspect is more prominent), no productive prefixed reduplication, etc. It is worth noting that a sizable number of attested Ma'a nouns of Cushitic origin lack the noun-class prefix that would be expected in a corre sponding Bantu noun. A count of the nouns in Ehret (1980) turned up 239 nouns with class prefixes and 134 nouns without. Some nouns of the latter group would be prefixless in Bantu too, but many would have an overt prefix. Various authors have noted the inconsistent usage of the class prefixes on Ma'a nouns of non-Bantu origin, so this feature is not peculiar to Ehret's data. Ma'a personal pronouns differ in form but not in grammatical categories from those of Mbugu and other Bantu languages. Most notably, they lack the Cushitic gender distinctions in the second and/or third person singular. Simi larly, Ma'a demonstratives differ in form from those of Mbugu, but the threeway Ma'a distinction, near/far/farther, is identical to the Mbugu system and typical of Bantu rather than of Cushitic languages, which tend to have either a two-member system or a four-member system (Mous 1994:191). There is one syntactic difference between Ma'a and Mbugu demonstratives, however: the Ma'a demonstratives do not agree with the head noun, while Mbugu (and other Bantu) demonstratives do carry agreement markers (Mous 1994:190). Ma'a pronominal possessors, too, differ in form from Mbugu posses sives. In both Cushitic and Bantu the possessive follows the noun, so it is not surprising to find Ma'a and Mbugu in agreement here. But the Ma'a posses sive system differs syntactically from Mbugu and other Bantu languages in a way that parallels the difference in the demonstratives: in attributive usage the possessives take no noun-class agreement markers, although in predicative usage they do agree with a head noun (Mous 1994:192). Bantu possessives, by contrast, always agree with the head noun. As Mous notes, the Ma'a possessives also differ from the Cushitic pattern, which would show mascu line/feminine gender distinctions in the second and/or third person, and not all the forms come from Cushitic. There is very little information available about Ma'a syntax — not surprisingly, given the near-total lack of textual material. Agreement patterns, as noted above, are Mbugu (and typically Bantu), except for the demonstra tives and the attributive pronominal possessives. Ma'a and Mbugu have SVO word order and prepositions, while Cushitic has SOV word order and post-
Ma'a (Mbugu)
475
positions. According to Ehret (personal communication, 1982), the Ma'a copula is obligatory, as in Cushitic; in Bantu the copula is optional.2 Another syntactic difference between Ma'a and Mbugu concerns the expression of possession: Ma'a uses an ordinary transitive verb -lo "have", like Cushitic; Bantu, in contrast, has no such verb, employing a construction "be with" instead. One lexical semantic feature should be mentioned here. According to Ehret (personal communication, 1982), Ma'a has "a five-part color division — black, white, red, yellow, green — with each expressed by a simple adjective", while Bantu languages "express only black, white, and red with simplex terms". Mous (personal communication, 1993) has not yet studied this aspect of the lexicon. If Ehret is right, it would be worth investigating other well-defined lexical semantic domains, such as kinship terms and folk taxonomies for animals and plants, to see whether other non-Bantu patterns exist. The most salient difference between Ma'a and Mbugu is of course the lexicon. Ehret estimates that about 50% of the Ma'a vocabulary is Cushitic (personal communication, 1982), and all authors agree that Ma'a has many words of non-Bantu origin. Mous's investigation of the non-Bantu portion of the Ma'a lexicon, however, has revealed a more complex picture than the twoway split between one Bantu language and one Cushitic language that is assumed by other authors. (The information that follows comes from Mous 1993.) First, not all the Ma'a words of Cushitic origin can be from the same Cushitic language, because some of them show innovations that place them in different Cushitic branches. Of the Southern Cushitic words, some must come from Iraqw, others must come from West Rift Southern Cushitic, and still others could come from either Iraqw or West Rift. Another set of Cushitic-origin words in Ma'a can be shown to have cognates in the Eastern Cushitic language Oromo but not in any Southern Cushitic language — that is, it looks as if these words are not of Southern Cushitic origin at all, though they are Cushitic. And finally, a sizable number of the non-Bantu words of Ma'a can be traced to Masai. Although some of the words listed by Mous for all these sources would be considered nonbasic vocabulary, many of them look quite basic: "person", "cheek", "woman, wife", "go", "child" (from Iraqw); "milk", "ashes", "dance", "noise" (from West Rift Southern Cushitic); "arm", "tooth", "liver", "sleep", "girl", "boy", "father", "uncle", "skin", "fat", "throat", "run", "ten", "egg", "get up", "put", "fat" (from Cushitic, but not Southern Cushitic); and
476
Sarah G. Thomason
"dog", "heart", "know", "body", "friend", "bite", "small", "grandmother", "buttocks", "come", "dance", "work", "jump, fly", "return", "rain", "borrow", "go", "pour", "scrape", "speak", "forget" (from Masai). Many of these words could have entered Ma'a as loanwords when the Mbugu encountered other peoples in their wanderings, but the diffusion of the words in all parts of the lexicon suggests very intimate contact, and is puzzling (see §4 for further discussion). Mous (1994:196-197 et passim) emphasizes that Ma'a lexical items stand in a one-to-one correspondence with Mbugu lexical items, and that the same is true of the few non-Bantu Ma'a suffixes and Mbugu suffixes. As he puts it, "many lexical entries have two forms...with one meaning and one morphological classification".3 This fact, together with the paucity of nonBantu grammatical features in Ma'a, is the reason Mous considers Ma'a and Mbugu to be different registers of the same language, rather than two differ ent languages. As noted above, Mous's view fits the current sociolinguistic status of Ma'a very well. In many respects Ma'a resembles languages like AngloRomani, in which a group of people who have shifted to a dominant language (English, in the Anglo-Romani case) nevertheless retain some lexicon of their original ethnic language (Romani) for use, with dominant-language grammar, as a secret language. There are also some differences between Ma'a and Anglo-Romani. First, Anglo-Romani is now apparently learned only as a second language, while children still learn Ma'a (beside Mbugu) as a first language. Second, there are a few grammatical differences between Ma'a and Mbugu, while there are none between Anglo-Romani and English. Still, just as it is hard to view Anglo-Romani as a completely independent language, given its grammatical identity to English and the fact that all its speakers are also fully bilingual in English, so it is difficult to view Ma'a as a completely separate language, independent of Mbugu. The few grammatical features in which Ma'a differs from Mbugu are hardly enough to set Ma'a apart structur ally. It seems to me that in this case the decision about "languageness" is a matter of definition, and reasonable people might well disagree about the most appropriate definition. My own view is that Ma'a is linguistically a separate language but sociolinguistically a register of Mbugu. The crucial linguistic observation is that the two languages are not mutually intelligible, any more than AngloRomani and English are mutually intelligible. A simple thought experiment
Ma'a (Mbugu)
477
shows why the claim that Anglo-Romani is English and Ma'a is Bantu is problematic. Suppose that the Mbugu were to decide that their existence among Bantu speakers is insupportable, and that they moved (yet again) to a new location, where they settled among non-Bantu speakers. Suppose further that, finding their Mbugu language not useful for communicating with their new neighbors, they gave it up and spoke only Ma'a among themselves. Finally, suppose that all Pare and Shambaa speakers shifted to Swahili. At this point Mbugu, Pare, and Shambaa have all disappeared, so that the intricate social and linguistic links between Ma'a and Bantu have been severed for ever. Now: would Ma'a be classified as a Bantu language? Surely not, especially if the scholars doing the classifying lacked all knowledge of the speakers' sociolinguistic history. My point is that the temptation to classify of Ma'a as a register of Mbugu arises solely from the social circumstances of its use, not from its linguistic make-up: given its mixed nature — Bantu gram mar, largely non-Bantu basic vocabulary — it would surely be called a separate language if it were the only language of a speech community. In any case, using social factors as a criterion for linguistic classification is not a procedure that recommends itself to historical linguists.
4.
How and why did Ma'a arise?
A successful answer to this question must take into account the attested linguistic features of Ma'a and also what is known and what can be inferred of the history of the Mbugu people. As we saw in §2, the Mbugu moved often, sometimes (perhaps) as an entire group and sometimes (certainly) in smaller groupings of clans, typically as the result of pressure from belligerent or intrusive neighbors.4 We also saw that the Mbugu have a well-deserved reputation for being culturally conservative, stubbornly resisting total assimi lation to their close neighbors. The survey of Ma'a structural characteristics in §3 showed that almost all the grammatical features, including all the productive inflectional and deriva tional morphology, are Bantu, identical to those of Mbugu. We saw a few non-Bantu features outside the lexicon: the phonemes /ł/? x/, the inconsistent usage of class prefixes on nouns, the agreement patterns in demonstratives and pronominal possessives, the semantics of the color terms, and the expres sion of possession by means of a transitive verb meaning "to have", as opposed to the Bantu "be with" construction.
478
Sarah G. Thomason
The other major linguistic feature of Ma'a that must be accounted for is the very mixed lexicon — Bantu words (not all from Mbugu, Pare, or Shambaa), Southern Cushitic words, apparently Eastern Cushitic words, and Masai words. I argued in 1983 that Ma'a was originally a Cushitic language but that its original grammar, together with about half of its lexicon, was replaced by Bantu grammar and lexicon over several centuries as bilingual Ma'a speakers incorporated more and more Bantu features into their speech. The idea was that intense cultural pressure from their Bantu neighbors had a profound effect on the Mbugu, but that their resistance to assimilation led them to maintain at least vestiges of their language — by now, only some of the vocabulary. On this view, present-day Ma'a is a drastically Bantuized relic of an earlier Cushitic language. This view has been challenged by several scholars since 1983. In the remainder of this section I will argue that the linguistic evidence favors my historical scenario over the other major propos als. Brenzinger (1987) and Sasse (1992) propose that Ma'a arose through partial relexification from Ma'a — the original language of the Mbugu people — into the Bantu language to which they had already shifted. As Sasse explains it (1992:27), this relexification would have been effected by younger, essen tially monolingual, speakers who adopted some of the (Cushitic) vocabulary of their elders and used the mixture as an in-group language. Brenzinger cites Boretzky (1985) as the source of the idea that mixtures like Ma'a and AngloRomani result from shift with subsequent relexification from the original ethnic-group language to the new native language. But, though Brenzinger says that Boretzky's proposal is "more probable than the heretofore assumed borrowing of the entire grammar from Pare" (1987:129), he does not say WHY it is more probable, and the only reason Boretzky himself gives is that so much borrowing would lead quickly to complete shift (1985:68). A problem with this assertion is that Boretzky does not provide any criteria for predicting when language shift will occur, and making such predictions is risky in the extreme. In particular, attitudinal factors play a major role in encouraging or discourag ing language shift, and there is much too little information available about people's reasons for shift to permit any firm predictions in this domain. This problem is especially acute when the group in question — like the Mbugu — has a long and well-documented history of resisting cultural assimilation. Sasse also has another reason for rejecting the hypothesis of gradual grammatical replacement through borrowing as the source of current Ma'a
Ma'a (Mbugu)
479
structure. He says that "the model is extremely problematic because borrow ing of morphology to the extent that entire systems are replaced has never been attested in an observable case" (1992:27). I have two responses to this objection. One is that there are apparently other cases of intensive language contact that resemble what I have proposed for Ma'a. One such case is Laha, as reported by Collins (1980:14, cited in Grimes 1994:263): Laha has maintained its indigenous language in the face of increasing pressure from Ambonese Malay but only at the expense of drastic revision of its grammar....Bit by bit the grammar of Laha has become nearly interchangeable with AM grammar.
Other cases, though less sweeping in their effects, are relevant in this context because they also involve significant morphological borrowing. Here are just two of many examples. In Asia Minor Greek, Turkish inflectional suffixes are added to verbs of Greek origin (Dawkins 1916:144; see Thomason & Kaufman 1988:219 for discussion). And in the Pacific, Thurston (1987:55) shows parallel possessive pronominal paradigms for two languages of New Britain, Lusi (an Austronesian language) and Anêm (a non-Austronesian language), with identical affixes for several person/number combinations. According to Thurston, Anêm has borrowed this paradigm (and the affixes) from Lusi. Sasse does not accept such examples as evidence that entire morphologi cal systems may be replaced through borrowing, and this leads me to my second response to his argument: if the first steps in morphological replace ment have been taken — namely, systematic paradigmatic borrowing of inflectional affixes, as in the Asia Minor Greek and Anêm cases — what reason is there to assume that total replacement cannot be the end result of such a process? That is, what constraint would permit the replacement of some forms in a paradigm but rule out the replacement of all the forms in the paradigm, or permit the replacement of one inflectional paradigm in a lan guage while ruling out the replacement of all the inflectional paradigms in the language? Sasse suggests no constraint that would serve as a stop sign for such changes — "you may replace this many affixes but no more" — and it is hard to imagine what a viable constraint would look like. It seems to me that the burden of proof must be on scholars who believe in such limits on borrowability. The question of whether a given phenomenon is linguistically possible is settled as soon as it occurs for the first time anywhere, and the borrowing of single inflectional affixes and partial para-
480
Sarah G. Thomason
digms is certainly very well attested.5 The next question is, how likely is it that such borrowing will continue until morphological replacement is complete? In most parts of the world (at least) it isn't very likely — but for social reasons, not linguistic ones. And since the most likely social reasons have to do with a short-circuiting of the borrowing process by shift away from the borrowing language, then the impossibility of predicting language shift will make social constraints on morphological replacement as hard to support as linguistic constraints.6 It is not clear from Brenzinger's and Sasse's discussions whether or not they believe that a period of Bantuization might have preceded the shift-withrelexification they propose as the origin of Ma'a. It is clear, however, that they would allow only limited borrowing of Bantu morphology into Ma'a before the shift. Similarly, the origin hypothesis developed by Mous (1993, 1994) does not allow for a period of extensive gradual Bantuization of an originally Cushitic language. In fact, in arguing that Ma'a "is a lexical register that was created by speakers of Normal Mbugu...consciously and on purpose" (1994:199), Mous seems to be claiming that there was no actual pre-Ma'a language at all — not a Cushitic language, and not any other kind of language either. Mous provides no detailed explanation of where Mbugu speakers might have acquired the Ma'a words of Cushitic and Masai origin. One possibility is that the words might have been adopted in different places* at different times, from people the Mbugu encountered during the wanderings that eventually led them to the Usambaras. Another possibility is that there was a single language that provided the non-Bantu lexicon of Ma'a, but that it was itself a mixed language, developed perhaps by a mixed group of people who served the Masai as forced labor (Mous, personal communication, 1993). The latter scenario has the virtue that it would account for the diverse nature of the non-Bantu portion of the Ma'a lexicon — something that none of the other hypotheses (including mine) do. In any case, one feature shared both by the shift-with-relexification hypothesis and by Mous's hypothesis is a lack of any significant degree of gradual Bantuization of a previously non-Bantu language. This feature is in sharp contrast to my hypothesis of more or less gradual Bantuization, accord ing to which evidence of incremental grammatical borrowing of Bantu struc ture is to be expected.7 Evidence of gradualness in the appearance of Bantu features will therefore support my hypothesis over theirs. Two kinds of evidence are possible: first, there might be fossilized forms that preserve no-
Ma'a (Mbugu)
481
longer-active non-Bantu structural features in Ma'a; and second, changes might be ordered in such a way as to prove that productive Bantu and nonBantu structures coexisted in the same chronological layers. Both kinds of evidence can be found in Ma'a. First, Ma'a has a number of forms that contain fossilized non-Bantu structural features. There are a few apparently unproductive non-Bantu gram matical suffixes: -eno (Cushitic; Mous 1994:188) and the causative suffixes -ti and -ri and perhaps other "frozen causative derivational suffixes" (Mous 1994:186).8 There are also several pairs of forms that reflect Cushitic pat terns, though they are clearly fossils in Ma'a (the following examples are all from Ehret, personal communication, 1982). Compare singular nihi "animal" and plural nihena (with a Cushitic plural suffix), singulative tambala "a snail" and plural tamba, and relics of Cushitic gender marking: im' "brother" vs. irjinta "sister" and i'alú "sheep" vs. i"alè "ram". These last features in particular, though of little importance for a synchronic analysis of Ma'a, are potentially of great significance for a diachronic analysis. If Ma'a did not at one time have Cushitic grammar, as Mous seems to be suggesting, where did these Cushitic grammatical features come from? And if all original Ma'a speakers shifted to Bantu and then partially relexified their Bantu language, they might have preserved a few gender-specific forms like "brother" vs. "sister" and "sheep" vs. "ram", but it seems relatively unlikely that they would have preserved a few relics of Cushitic number marking beside their regular and productive Bantu number marking. Moreover, given the preva lence of regular Bantu causatives in Ma'a, the appearance of frozen Cushitic causative suffixes on Ma'a verbs suggests that Cushitic causative formations remained productive until fairly recently in Ma'a. In addition, some nouns show competition between older Cushitic singu lars and collectives, on the one hand, and newer Bantu singular and plurals, on the other: compare, for instance, iłare "cloud" (with a Bantu class prefix; Tucker & Bryan 1966:207) vs. the unmarked singular tare (Green 1963:185); plural matare "clouds" (with a Bantu class prefix; Tucker & Bryan 1966: 207) vs. iareno "clouds" (with a Cushitic collective suffix; Green 1963:185) vs. matareno (with both the Bantu prefix and the Cushitic suffix; Green 1963: 185, Tucker & Bryan 1966:207); singular i'alú "a sheep" with plurals máalú (with a Bantu prefix; Tucker & Bryan 1966:207) vs. aleno (with a Cushitic suffix; from Meinhof 1906, cited by Green 1963:185). The existence of apparently competing forms suggests the usual historical process by which
482
Sarah G. Thomason
both internally- and externally-induced replacements occur, with a transi tional period during which both old and innovative forms are used (see Thomason 1996 for further discussion of this competition). These variant forms therefore also point to a period during which Cushitic structures were being replaced gradually by Bantu structures. The other type of evidence supporting a hypothesis of gradual Ban tuization comes from chronologically ordered changes. These are difficult to work out, because the massive amount of material in Ehret (1980) requires careful interpretation; many of Ehret's cognate sets and reconstructions are not very convincing. (Hetzron & Talos [1982] also express some skepticism about details of Ehret's analyses, though they give proper credit to Ehret's impressive achievement in gathering and analyzing so much valuable data.) Still, some points seem clear. The best examples are suffixes added before and after the regular Bantu-induced sound change that opened syllables by deleting syllable-final consonants in non-Bantu lexical items. First, Cushitic suffixes (e.g. -ti CAUSATIVE) are added to Cushitic stems after this sound change, proving that some non-Bantu structural features were productive after Bantu interference began to affect Ma'a structure. An example is -sahéti "collect" (cf. the base verb -sahe "meet" < *t'aHem). Second, Bantu suffixes can be found on Cushitic stems that have not yet undergone the sound change, proving that Bantu structural interference features appeared in Ma'a at different times — not all together, as would be expected in the Brenzinger/ Sasse and Mous scenarios. An example is -mur-ija "to spread (something)"; -ija, a Bantu causative suffix, also appears on Cushitic stems that have undergone the sound change (e.g. -fi-ija "to complete", with a root that originally ended in k). And third, Cushitic suffixes are added to Bantu loanwords, e.g. -eto in -eto "fig tree" (cf. Pare m- "fig tree"). 9 Taken together, the evidence of incremental Bantuization of Ma'a gram mar shows conclusively that a pre-Ma'a language had productive non-Bantu grammar, including inflectional morphology, and that the loss of its nonBantu grammatical features took place over a considerable period of time, as borrowed Bantu structures and grammatical morphemes replaced competing non-Bantu structures and grammatical morphemes. This evidence is consis tent with the non-Bantu features in present-day Ma'a that we noted above — several phonemes, omission of some noun-class prefixes on non-Bantu nouns, a few non-Bantu agreement patterns, etc. All this evidence is insuffi cient to prove that the entire process of Bantuization was incremental. But
Ma'a
(Mbugu)
483
since some grammatical replacements demonstrably proceeded gradually, the simplest hypothesis is that most of the other changes also proceeded gradu ally. There are no known barriers to such a sequence of changes: the first steps are attested in numerous languages, and no one has defended a con straint against further changes of the same type. At the least, then, the burden of proof must be on those who argue that increasingly pervasive grammatical borrowing was NOT the route by which Ma'a reached its present state. In addition, at least one published comment suggests gradual Bantuization, in the form of a significant difference between Ma'a as spoken by younger people and as spoken by elders:10 Here it is certainly true that the affix-system is better established among younger people than among the older generations, and it may well be that if they do not become wholly assimilated by the Shambala, they will in time not only operate the affix-structure of a Bantu language but will also acquire a sufficient number of Bantu stems and roots for them to be considered speakers of a Bantu language (Whiteley 1960:96).
But if total grammatical replacement through borrowing is possible, what explains its rarity? The reasons are social: few known ethnic groups have come under such intense cultural pressure and nevertheless retained (in a drastically altered form) their ethnic-group language — although Laha and perhaps also some languages of New Guinea fit into this same category. In this case, the reason surely lies in the well-documented long-term resistance by the Mbugu to total cultural assimilation by their Bantu-speaking neighbors. Of course the language is not the only aspect of the group's older culture that has changed out of recognition even while being "preserved"; Copland, for instance, reports that he "experienced great difficulty in obtaining texts and indeed [he is] inclined to believe the statements made by the Mbugu them selves that they no longer possess songs and folk-tales in their own language" (1933-34:243). But language is a salient group marker, powerful enough that some new (sub-)ethnic groups have invented their own (e.g. Media Lengua, Michif, and Mednyj Aleut, all described in this volume), so it is not surprising that a group clinging to their separate identity would also cling as much as possible to their ethnic language. A significant role still remains for the creativity that Mous (1993, 1994) has documented in current Ma'a usage, for instance the insertion of the "difficult" Ma'a phoneme /ł/ into new words, to emphasize the differentness of the language. Although, as I have argued, I believe that the linguistic and
484
Sarah G. Thomason
social evidence points to a long developmental process for Ma'a, at some point the speakers clearly came to view their non-Mbugu language as a, or the, symbol of their differentness, and to emphasize the distinctive features by introducing deliberate alterations in the language. In other words, there is no inherent incompatibility between a gradual developmental process and cre ative manipulation of non-Bantu features during part of the developmental period. Finally, there is also no inherent incompatibility between a gradual developmental process and subsequent shift-with-relexification of the type proposed by Brenzinger (1987) and Sasse (1992): the shift could have taken place after, rather than instead of, Bantuization of Ma'a. Mous has shown that present-day Ma'a is sociolinguistically dependent on Mbugu, so it is reason able to claim that Ma'a speakers have now shifted to Mbugu, in spite of the fact that both Ma'a and Mbugu are learned by Mbugu children as first languages. In sum, what makes Ma'a such a striking case is not the linguistic processes that led to its current state; all the processes I have discussed are well known from other contact situations, though not, for the most part, in the same degree. What is truly unusual about the history of Ma'a is the combina tion of social factors that caused its speakers to carry out total restructuring of their language instead of simply shifting to Mbugu.
Notes 1.
See Thomason (1983) for a more detailed typological comparison. Mous has now provided further information and some new typological comparisons; but the overlap in coverage in Mous (1993, 1994) and Thomason (1983) is by no means complete.
2.
Mous (1994) does not mention this feature, so I don't know whether it occurs in his data or not.
3.
Significantly, lexical entries with only one phonological shape are all Mbugu. There are apparently no Ma'a words that lack precise synonyms in Mbugu (unless the semantics of color terms is such a case).
4.
But they may sometimes have had other motives for moving as well: the text in Copland (1933-34) begins, "A long time ago the Mbugu came to Pare. They said, 'The cattle now have no grass to eat, let us go to the plains.' They fought with the powerful Masai."
5.
The wholesale incorporation into one language of an inflectional system from another language is also well established, namely by the case of Mednyj Aleut, in which Russian finite verb inflection has been grafted onto an Aleut base (see the previous chapter). But
Ma'a (Mbugu)
485
in considering the development of Ma'a, both Sasse and I are focusing on incremental morphological replacement, not on the kind of sudden change — or burst of creativity — that must have given rise to Mednyj Aleut. So the Mednyj Aleut case is not directly relevant here. It IS relevant, however, if the question of linguistic possibility is considered independently of the question of developmental process. For further discussion of the relationship between linguistic possibility and social probability, see Thomason 1996. 6.
Some non-Austronesian languages of New Guinea may constitute a significant exception to the generalization that sweeping morphological replacement is unlikely, because nonAustronesian New Guinea languages are notorious for being lexically dissimilar but grammatically similar. As Foley points out, both grammatical elements and basic vocabu lary — including whole pronoun systems — can be shown to have been borrowed among these languages (1986:Ch. 7). But it is difficult to assess the level of morphological borrowing, because very little historical work has been done on these languages.
7.
As far as I can tell, gradual Bantuization is also not expected under two other origin hypotheses which will not be discussed in detail here. This is certainly true of Möhlig's view (1983:258-59, cited in Mous 1994:198) that Ma'a must be Bantu, on the ground that the Bantu grammar can't be borrowed because the lexicon is Cushitic. As noted above, and as Mous (1994:198) points out, there ARE many Bantu words in Ma'a, and extensive structural borrowing is consistent with a borrowing scale like the one in Thomason & Kaufman (1988:74-76). The main problem with Möhlig's view, however, is that the nonBantu words in Ma'a are especially prominent in the basic vocabulary, which is not true of ordinary lexical borrowing, and which would in particular not be expected if Ma'a had arisen through Cushiticization of a Bantu language (see Thomason & Kaufman 1988:227228 for a more detailed discussion of such ą scenario). The other hypothesis is Carol Myers-Scotton's code-switching model, according to which Ma'a arose through codeswitching that led to an eventual "turnover" in the matrix language (1993:220-23). Myers-Scotton's suggestion that code-switching was a mechanism in the development of Ma'a is certainly plausible, though other mechanisms are also possible (see Thomason 1996 for discussion of other mechanisms of intereference). To the extent that I under stand her specific proposals, they seem to follow the Brenzinger/Sasse approach in rejecting incremental incorporation of Bantu features, including inflectional features, into a non-Bantu language; if this interpretation is correct, evidence of incremental borrowing is as serious a problem for her scenario as for theirs.
8.
Here Mous disagrees with Ehret, who says that the causative suffix -ti is productive in Ma'a (personal communication, 1982).
9.
This third type of layered change is less significant than the other two, because a nonBantuized pre-Ma'a could certainly have already had some Bantu loanwords, to which Cushitic affixes would naturally be added if the grammar of the language was Cushitic. It should also be noted that Mous (personal communication, 1993) has not identified a Cushitic source for the suffix -eto, though he believes that it is likely to be of Cushitic origin; it is in any case apparently not of Bantu origin.
10.
Hetzron & Tàlos make a comment with a similar import: "Ma'a...is on its way [to] becoming Bantu through assimilation of more and more Bantu lexical and grammatical elements, an interesting fact noted by specialists" (1982:240). But since they give no data or specific sources for this picture of gradual Bantuization, I don't know what evidence it is based on.
486
Sarah G. Thomason
References Boretzky, Norbert. 1985. "Sind Zigeunersprachen Kreols?" Akten des 1. Essener Kollo quiums uber "Kreolsprachen und Sprachkontakte", ed. by Norbert Boretzky, Werner Enninger, and Thomas Stolz, 43-70. Bochum: Studienverlag Dr. N. Brockmeyer. Brenzinger, Matthias. 1987. Die sprachliche und kulturelle Stellung der Mbugu (Ma?a). Cologne: University of Cologne M.A. thesis. Collins, James T. 1980. "Lana, a language of the Central Moluccas". Indonesia Circle 23.3-19. Copland, B.D. 1933-34. "A note on the origin of the Mbugu with a text". Zeitschrift für Eingeborenen-Sprachen 24.241-45. Dawkins, R.M. 1916. Modern Greek in Asia Minor: A study of the dialects of Silli Cappadocia and Phárasa with grammars, texts, translations, and glossary. Cam bridge: Cambridge University Press. Ehret, Christopher. 1980. The historical reconstruction of Southern Cushitic phonology and vocabulary. (Kölner Beiträge zur Afrikanistik, 5.) Berlin: Dietrich Reimer. Farler, [Archdeacon]. 1885. "A collection of words from the Wa-Mbugu." A pocket vocabulary of the Ki-Swahili, Kiy-Nyika, Ki-Taita, and Ki-Kamba languages...Also a brief vocabulary of the Kibwyo dialect, collected by Archdeacon Farler, by Archibald Downes Shaw, 198-204. London: Society for Promoting Christian Knowledge. Feierman, Steven. 1974. The Shambaa kingdom: A history. Madison: University of Wisconsin Press. Foley, William A. 1986. The Papuan languages of New Guinea. Cambridge: Cambridge University Press. Goodman, Morris. 1971. "The strange case of Mbugu". Pidginization and creolization of languages, ed. by Dell H. Hymes, 243-54. Cambridge: Cambridge University Press. Green, E. Clive. 1963. "The Wambugu of Usambara (with notes on Kimbugu)". Tanganyika Notes and Records 60-61.175-89. Greenberg, Joseph H. 1955. Studies in African linguistic classification. New Haven. Greenberg, Joseph H. 1963. The languages of Africa. Bloomington: Indiana University Press. Grimes, Barbara Dix. 1994. "Cloves and nutmeg, traders and wars: Language contact in the Spice Islands". Language contact and change in the Austronesian world, ed. by Tom Dutton & Darrell T. Tryon, 251-74. Berlin & New York: Mouton de Gruyter. Hetzron, Robert, & Endre P. Talos. 1982. "Christopher Ehret's The historical reconstruc tion of Southern Cushitic phonology and vocabulary: A review article". Sprache und Geschichte in Afrika 4.239-50. Kimambo, Isaria N. 1969. A political history of the Pare of Tanzania, 1500-1900. Nairobi: East African Publishing House. Krapf, Johann Ludwig. 1964. Reisen in Ostafrika ausgeführt in den Jahren 1837-1855 von Johann Ludwig Krapf Stuttgart: F.A. Brockhaus. [Reprinted from the 1858 original, Stuttgart: In Commission bei W. Stroh.] Meinhof, Carl. 1906. "Linguistische Studien in Ostafrika, X: Mbugu". Mittheilungen des Seminars für Orientalischen Sprachen 9/3.293-323.
Ma'a (Mbugu)
487
Möhlig, Wilhelm J.G. 1983. "Mbugu". Lexikon der Afrikanistik, ed. by Herrmann Jungraithmayr & Wilhelm J.G. Möhlig, 158-59. Berlin: Dietrich Reimer. Mous, Maarten. 1993. "The making of Ma'a". Paper presented at the Colloquium on Synchronic and Diachronic Sociolinguistic Methods and Interpretations, Universität Bayreuth. Mous, Maarten. 1994. "Ma'a or Mbugu". Mixed languages, ed. by Peter Bakker & Maarten Mous, 175-200. Amsterdam: Institute for Functional Research into Lan guage and Language Use (IFOTT), University of Amsterdam. Myers-Scotton, Carol. 1993. Duelling languages: Grammatical structure in codeswitching. Oxford: Oxford University Press. Sasse, Hans-Jürgen. 1992. "Theory of language death". Language death: Factual and theoretical explorations with special reference to East Africa, ed. by Matthias Brenzinger, 7-30. Berlin & New York: Mouton de Gruyter. Thomason, Sarah G. 1983. "Genetic relationship and the case of Ma'a (Mbugu)". Studies in African Linguistics 14.195-231. Thomason, Sarah G. 1996. "On mechanisms of interference". Studies for Einar Haugen, ed. by Stig Eliasson & Ernst Håkon Jahr. Berlin & New York: Mouton de Gruyter, to appear. Thomason, Sarah G., & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thurston, William R. 1987. Processes of change in the languages of north-western New Britain. (Pacific Linguistics, Series B, 99.) Canberra: Department of Linguistics, Research School of Pacific Studies, The Australian National University. Tucker, Archibald Norman, & Margaret A. Bryan. 1974. "The 'Mbugu' anomaly". Bulletin of the School of Oriental and African Studies 37.188-207. Whiteley, Wilfred Howell. 1960. "Linguistic hybrids". African Studies 19/2.95-97.
Language Index A Adamawa (-Eastern) branch of NigerCongo: 136, 209; see also Ubangian (or Eastern) branch of Niger-Congo Afrikaans: 166 Afro-Asiatic language family: 470 Aka: 209 Aleut: 6, 450-468, 484; dialects: 452453. See also Atkan Aleut, Attuan Aleut, Bering Aleut, Mednyj Aleut Algonquian languages: 91, 94, 295, 303, 313, 329, 355, 356 Aluku-Wayana trade jargon: 118-119, 120 Alukuyana (the Ndyukas' name for Ndyuka-Trio Pidgin and Wayana): 103 Alur: 145 Ambonese Malay: 479 American Indian Pidgin English: 43, 95 Aném: 479 Anglo-Romani: 476, 477, 478 Anuak: 136 Apolista: 443-445 Arabic: 125-172, 209, 248, 252, 257, 261, 263, 276, 277, 279, 280, 283, 288, 289, 290, 291, 372; Arabic dialects via pidginization (hypothesis): 127; Arabic as a lingua franca: 127; see also Chadian Colloquial Arabic, Classical Arabic, Egyptian Arabic, Old Arabic, Sudanic Colloquial Arabic Arabic-based pidgins and creoles: 2, 4, 5, 6, 125-172; lacking features present in Arabic: 165; with features absent in Arabic: 165-166; see also Maridi Arabic, East African Nubi, Kenyan
Nubi, Sudanic pidgin/creole Arabic, Turku Aramaic: 126 Arawak: 118, 119, 122 Asia Minor Greek: 479 Assiniboine: 305, 349 Atkan Aleut: 453, 466 Atlantic creoles: 144, 277 Attuan Aleut: 453, 459, 466 Austronesian languages: 21, 31; see also Motu, Fijian, Suau Aymara: 371, 428, 429, 430, 442 Bajuni (dialect of Swahili): 280 Baka: 136 Bambara: 213, 264 Banda: 209, 212, 229, 238, 239, 256, 257, 266 Banda-Linda: 222 Bangala: 133, 264 Bantu-based pidgins and creoles: 278, 281, 285 Bantu languages: 6, 140, 175, 176, 182, 183, 184, 186, 187, 188, 189, 191193, 194, 198, 199, 200, 202, 203, 204, 205, 206, 209, 213, 214, 230, 244, 248, 257, 265, 276, 278, 279, 281, 283, 284, 285, 286, 288, 289, 290, 292, 469-487; see also North East Coast Bantu, Proto-Bantu Bari: 134, 136, 161-165 Bemba, Town: 190 Berbice Dutch Creole: 5 Bering (Island) Aleut: 450, 453, 459, 466 Bislama: 29, 39 Bobángi: 189, 194,200
490 Bongo: 136 Brayet: 304, 305 Broken: 22, 38, 39; see also Pidgin English Bungi: 304 Bushnegro creoles: 99-101, 119; see also Kwinti, Matawai, and Saramaccan Callahuaya: 2, 4, 6, 355, 427-447; alternate names: 427; dialects: 431 Carib: see Kalinha/Kalinya Cariban languages: 101, 102, 113, 122; see also Kalinha/Kalinya, Trio, Triometesem (Wayaricule), and Wayana Caribbean creoles: 1 Catalangu: 413-418, 420 Cayenne patois: 120 Central Sudanic languages (NiloSaharan family): 136, 167, 209 Chadian Colloquial Arabic: 133, 155, 156, 159 Chaga: 283 Chinook Jargon: 213 Chippewa: 358 Classical Arabic: 125-127, 168 Comorian (Comorean): 199, 272, 283, 292 Copper Island Aleut: see Mednyj Aleut Coptic: 126 Cree: 6, 295-363; see also Moose Cree, Plains Cree, Swampy Cree, Woods Cree Cushitic languages: 276, 277, 280, 288, 289, 291, 469-487 D Delaware: 43-98; phonology: 45-47; morphology: 47-50 Didinga: 136 Dinka: 134, 136, 162, 168 Dodoth: 136 Dongolese (dialect of Nile Nubian): 167
Language
Index
Dutch: 5, 79, 218; spelling conventions: 50-51, 55, 56, 87; see also Flemish E East African Nubi: 6, 125, 135, 146, 155-160, 161-166, 264; see also Kenyan Nubi, Uganda Nubi Eastern branch of Niger-Congo: see Ubangian languages Eastern Cushitic: 289, 475, 478 Eastern Sudanic languages: 136 Egyptian Arabic: 132, 158-159, 166 Eleman languages: 14, 35 English: 7, 11, 12, 13, 19, 21, 22, 32, 34, 39, 43, 88, 188, 193, 204, 206, 218, 257, 277, 296, 297, 298, 299, 300, 303, 304, 305, 306, 308, 311, 312, 317, 318, 322, 323, 324, 325326, 332, 337, 348, 350, 352, 359, 360, 476, 477; spelling conventions: 51,54,55,56,88 English-based creoles: 102; see also Broken, Ndyuka, Tok Pisin F Fanakalo: 190 Fijian: 21, 27, 39 Flemish: 188 Fox: 86, 94 French: 6, 7, 38, 114, 115, 183-184, 188, 193, 204, 205, 209, 211, 214, 215, 218, 241, 244, 245, 248, 252, 253, 254, 255, 257, 259, 261, 262, 263, 264, 265, 266, 277, 295-363; Acadian French: 301, 350; Québecois French: 301, 308, 312, 350; spelling conventions: 50, 51 Ful: 209, 213 G Gaelic: 297, 305, 349 Gbanziri: 264 Gbaya: 209, 222 Gbeya: 258 German: 38, 305; spelling conventions:
Language Index 56,85 Germanic languages: 7, 83, 93 Greek: 38, 126, 479 H Halbdeutsch: 5 Hausa: 209 Hindi: 121; Sarnami Hindi (Hindi spoken in Suriname): 121 Hin Motu (formerly called Police Motu): 1, 3, 4, 5, 6, 9-41; dialects: 1112, 34, 38; official name: 25, 34; phonology: 26-27; morphosyntax: 2730; word order: 30-31; lexicon: 31-32 Hiri Trading Languages: 14-15, 17, 2930, 32, 33, 35-37 I Ijo: 5
Indian Ocean creoles: 1 Ingii ("Indian"; the Ndyukas' name for Ndyuka-Trio Pidgin): 103 Iraqw: 475 Iroquoian languages: 83 Isikula (alternative name for Fanakalo): 190 Isilunga (alternative name for Fanakalo): 190
J Juba Arabic: 133-135, 155-160, 168 Kako: 209 Kakwa: 136 Kalinha/Kalinya: 101, 113, 114-117, 118-119, 121, 122 Kallawaya: see Callahuaya Kami: 284 Kanuri: 209 Kara: 209 Kare: 209 Kenyan Nubi: 125, 147-154, 160, 168; phonology: 147-148; morphology: 148-149; syntax: 149-154 Kenya Pidgin Swahili: 265
491 Kenzi (dialect of Nile Nubian): 167 Khasonke: 213 Kiboma: 184 Kibwaal (dialect of Kiyánsi): 204 Kifióti: 184,202 KiHindi ("Indian language", a variety of Pidgin Swahili): 273 Kihungán: 184 Kikóngo (Kikóóngo): 121, 173-208, 219; dialects of: 202, 204, 247 Kikóngo Kitúba (another name for Kitúba): 173, 203 Kikuyu: 284, 285, 286 Kiládi: 184, 202 Kimanyánga (dialect of Kikóngo): 180, 181, 184, 185, 187, 192, 193, 200, 201, 202; importance as a lingua franca: 185, 188-189 Kimbala: 184, 202 Kimpendé: 184, 202 Kingóngo: 184 Kintándu: 184, 202 KiSetla ("settlers' language", a variety of Pidgin Swahili): 273 KiShamba ("plantation language", a variety of Pidgin Swahili): 273 Kisúku: 184 Kitchen Kafir (alternative name for Fanakalo): 190 Kiteké: 184, 194, 199, 200, 202, 205 Kitúba: l, 4, 6, 173-208, 213, 214, 215, 217, 219, 220, 222, 223, 230, 236, 238, 239-240, 241, 244, 247, 248, 253, 254, 255, 256, 257, 260, 261, 263, 264; alternate names for: 173176, 198; dialects of: 182-184, 186 KiVita ("war language", a variety of Pidgin Swahili): 273 Kiwai: 27 Kiyaka: 184 Kiyánsi: 184, 187, 194, 199, 200, 201, 202, 204, 205 Kiyómbe: 184, 202 Koita: 11, 13, 16,27,30,31,33 Koriki: 14, 32, 35 Kresh: 136
492 Kru languages: 186, 213 Kwa languages: 186, 192 Kwai Is: 24 Kwara'ae: 24 Kwinti: 101, 119 L Laha: 479, 483 Latin: 7, 38, 127 Lau: 24 Leco: 443-445 Lendu: 136 Lingála: 187, 189, 190, 192, 194, 199, 200, 204-205, 214, 215, 220, 222, 230, 234, 241, 244, 245, 247, 248, 250, 253, 254, 255, 256, 257, 260, 261, 263, 264, 266 Lomóngo-Lonkundó: 189 Lotuxo: 134, 136 Lubumbashi Swahili: 143, 168, 213, 265, 271-294 Lugbara: 136 Luguru: 282 Lunyoro: 145 Lusi: 479 Lwo: 162 M Ma'a: 2, 4, 6, 469-487; alternate names: 469 Madi: 134, 136 Mahas (dialect of Nile Nubian): 167 Maipuran: 427, 431 Makonde: 280 Malay: 479 Mamvu: 136, 163 Mande languages: 186, 192 Mangbetu: 136 Maridi Arabic: 126, 132; oldest pidgin on record: 131 Masai: 471, 475, 478, 480 Massachusett: 95 Matawai: 99 Mbimu: 209 Mbugu: 469-487 Mbum languages: 209
Language
Index
Media Lengua: 2, 3, 4, 6, 355, 365-426, 442, 463, 465, 483; alternate names for: 377; orthography: 425; varieties of: 410-419 Mednyj Aleut: 2, 4, 6, 449-468, 483, 484 Mekolo ("Negro"; the Wayanas' name for Ndyuka-Trio Pidgin, Ndyuka, and the Ndyukas): 103 Michif: 2, 4, 6, 295-363, 420; dialects: 295, 302, 310, 323, 349, 353-354, 463, 465, 483 Mohawk: 83 Mondari: 136 Mongo: 266 Montagnais: 341, 355, 360 Moose Cree: 349 Moru: 134, 136 Motu: 6, 9-41; see also Motu Foreigner Talk, Police Motu, Simplified Motu Motu Foreigner Talk: 11, 17, 19; see also Simplified Motu Munsee: see Delaware Murle: 136 Mwani: 280, 292, 293 Mwiini (dialect of Swahili): 279, 280, 287, 292, 293 N Ndyuka: 5, 99-124 Ndyuka tongo ("Ndyuka language"): 101; see also Ndyuka Ndyuka-Trio Pidgin: 1, 4, 5, 99-124; phonology: 104; syntax: 105-113; lexicon: 113-117; semantics: 117 Nehipwat: 305 New Guinea languages (non-Austronesian): 2, 12, 485 Ngbaka-Ma'bo: 209 Ngbandi: 211-270; see also under its dialects Yakoma, Riverine Sango Ngombe: 212, 214, 241, 254, 260, 261, 263, 266 Niger-Congo languages: 136, 209, 265 Nigerian Arabic: 165, 166, 167 Nilo-Saharan languages: 209, 265, 471
Language Index North East Coast Bantu: 276, 278, 281, 285, 287, 288, 292 Nubi: See East African Nubi, Kenyan Nubi, Uganda Nubi Nubian (Nile Nubian): 161, 167 Nuer: 134, 136 Nyakyusa: 279 Ojibwa/Ojibwe: 6, 94, 95, 299, 301, 302-303, 304, 305, 309, 310, 312, 348, 349, 350, 351, 352, 354, 358, 359 Ojicree (also called Severn Ojibwe): 305 Old Arabic: 125, 159 Oromo: 475 Ottawa: 94 P Pana: 209 Pano-Tacanan languages: 427 Papuan languages: See New Guinea languages (non-Austronesian) Papuan Pidgin English: 35-37, 38; see also Pidgin English Pare: 469, 471, 472, 477, 478 Persian: 289 Pidgin Algonquian: 95 Pidgin Delaware: 1, 3, 4, 5, 43-98; phonology: 50-57, 82; grammatical categories: 57-62; word order: 62-63; variation in: 75-77 Pidgin English: 1, 12, 13, 19, 22, 23, 2930, 31, 32, 33-34, 35-37, 38, 95; see also American Indian Pidgin English, Broken, Papuan Pidgin English, West African Pidgin English Pidgin Massachusett: 95 Pidgin Sango: 211, 212, 265 Pidgin Spanish: 370, 403-404, 407 Pidgin SwahiU: 273, 287, 289, 290, 293; see also Kenya Pidgin Swahili, KiHindi, KiSetla, KiShamba, KiVita Plains Cree: 301, 302, 304, 309, 313, 317, 344, 349, 350, 351, 352, 354, 360
493 Pokomo: 285 Police Motu (former name of Hiri Motu): 9, 11, 13, 17, 18, 19, 20-25, 33, 34, 37, 38, 39 Portuguese: 119, 122, 193, 204, 218, 257 Portuguese-based creoles: 1, 204 Proto-Algonquian: 85 Proto-Bantu: 278, 279, 292 Proto-Eastern Algonquian: 85 Proto-Michif?: 353-354 Proto-Sabaki: 276, 278, 282, 286, 287, 288,289,291,292 Proto-Swahili: 276, 278, 282, 286, 288289 Puquina: 427-447 Q Quechua: 6, 355, 365-426, 427-447; dialects: 365, 367, 368, 371, 374, 379, 435 R Riverine Sango (dialect of Ngbandi): 211,257 Romani: 355-356; Anglo-Romani: 476, 477, 478 Runga: 209 Russian: 6, 450-468, 484 S Sabaki: 276, 278, 282 Sango: 1, 4, 6, 133, 209-270; old names for: 216; standard Bangui variety: 209, 235, 253, 262-263 Sara: 209 Saraguro Media Lengua: 410-413, 416418,420 Saramaccan: 99, 121,261 Saulteaux (alternate name for Ojibwa/ Ojibwe): 299, 300, 304, 310, 358 Shambaa: 469, 470, 472, 476, 478 Shawnee: 94 Shilluk: 136, 162 Simplified Motu: 16-17, 18, 19, 20, 23, 26-33, 35-37, 39
494 Somali: 280, 289 Soninke: 213 Southern Cushitic: 277, 289, 475, 478 Spanish: 6, 119, 355, 365-426, 428, 429, 432, 436, 437, 442; dialects: 369, 372; see also Pidgin Spanish Sranan: 101, 103, 113-114, 117, 121, 122 Suau: 12,31 Sudanic languages: 223, 265 Sudanic pidgin/creole Arabic: 125, 127131, 133, 135-146, 155-160, 161, 168; phonology: 155-156, 161, 164; morphology: 156; syntax: 156-157, 164-165; lexicon: 157; common origin for all varieties: 155, 159; see also East African Nubi, Juba Arabic, Turku Sudanic Colloquial Arabic: 125, 129, 134-135, 161 Susquehannock: 83 Susu: 213 Swahili: l, 3, 4, 145, 146, 190, 199, 206; dialects: 274, 276, 279, 280, 282, 283, 289, 292, 469, 470; ProtoSwahili: 276; Standard Swahili (based on Unguja, the dialect of Zanzibar Town): 273, 274, 293; see also Bajuni, Kenya Pidgin Swahili, Lubumbashi Swahili, Mwiini, Pidgin Swahili, Unguja Swampy Cree: 352 Swedish: 86, 87, 88, 89, 90, 91, 92, 93, 94; spelling conventions: 51-53, 55, 56, 63, 86 T Tacana: 427, 443-446 Tacanan languages: 431 TokPisin: 11, 13,34,37,143,261 Town Bemba: 190 Trio: 5, 99-124 Triometesem (Wayaricule): 122 Tumbuka: 279
Language
Index
Turkish: 479 Turku: 130, 131, 132-133, 155-159, 167 U Ubangian languages: 209, 211, 214, 222, 229, 230, 244, 248, 253, 254, 257, 262, 266 Uganda Nubi: 147-154, 158, 160, 163, 168 Ukrainian: 305 Unalachtigo: see Delaware Unami: see Delaware Unguja (the dialect of Zanzibar Town, basis of Standard Swahili): 273, 274, 279 Uru-quilla: 430 V Vai: 213 Vulgar Latin: 127 W Wayana: 100, 101, 102, 104, 113, 114117, 118-119, 121 Wayana-Aluku trade jargon: 118-119, 120 Wayaricule (alternate name for Triometesem): 122 Welsh: spelling conventions: 56, 88 West African Pidgin English: 188 West Atlantic languages: 186, 192, 205 Wolof: 213 Woods Cree: 352 Y Yakoma (dialect of Ngbandi): 211, 214, 215, 216, 217, 218, 219, 220, 221, 224, 225, 228, 229, 230, 231, 242, 246, 255, 257, 260, 261, 263, 264, 265, 266 Yulu: 209 Z Zande: 134, 136, 168, 209, 212, 249
Names Index A Agheyisi, Rebecca: 263 AlBakri: 132 Albo, Xavier: 428, 429, 431 Ali, Muhammad: 129 Allard, Ida Rose: 295, 304, 306 Aloema, Nardo: 119 Amin: 135, 160 Anthony, Albert Seqaqkind: 46, 51, 53, 54, 57, 60, 64, 65, 66, 71, 77, 78, 91, 92,93 Armbruster, Charles: 161, 167 Asinovskij, A.S.: 453 Baker: 130, 141, 142 Bakker, Peter: 301, 307, 348, 353, 354, 355-356, 357, 358, 463 Barber, John W.: 84 Barnstorn, George: 305 Basov, Emelian S.: 451 Beaglehole, John Cawte: 454 Behnstedt, Peter: 132 Beklemišev, O.: 455 Bell, Herman: 134, 159 Bender, M. Lionel: 273 Bentley, William: 194 Berg: 454 Bergsland, Knut: 450, 452-453, 454, 455,461,466 Bering, Vitus: 451 Bertoncini, Elena: 277 Besten, Hans Den: 355 Bishop, Charles A.: 305 Black Beaver: 94 Blain, Eleanor: 304 Bloomfield, Leonard: 302, 356
Boganda: 237 Bokamba, Eyamba: 179 Boretzky, Norbert: 160, 355-356, 478 Bouquiaux, Luc: 211, 215, 220, 221, 222, 227, 228, 231, 232, 233, 245, 250, 251, 252, 255, 256, 257, 258, 259, 265, 266 Boyd, Raymond: 230, 244 Boyeldieu, Pascal: 211, 221-222, 224, 225, 228, 230, 231, 265 Brachiel, G.H.: 212, 264 Bradnum, F.: 141 Bragdon, Kathleen: 78 Braiman, Ali: 162 Brenzinger, Matthias: 469, 470, 471, 478, 480, 482, 484, 485 Brett, R.: 11 Brinton, Daniel G.: 46, 51, 53, 54, 57, 60, 64, 65, 66, 71, 77, 78, 91, 92, 93 Broomfield, G.W.: 273 Brumfit, Ann: 271 Bryan, Margaret A.: 162, 212, 241, 244, 469,472,481 Calloc'h, Jean-René: 212, 216, 233, 245, 252, 253, 256, 257, 262, 263 Campanius, Johannes: 44, 51-56, 58, 606.1, 62-63, 66, 69, 72-73, 74, 77, 8081, 83, 85, 86, 87, 88, 89, 90, 91, 93, 94. Carrington, Lawrence: 136 Carter, Hazel: 187 Casati, Gaetano: 139, 144, 145, 167 Catlin, George: 301 Chami, Felix A.: 274 Champlain: 296
496 Chappell, Edward: 305 Chartrand, Paul: 358 Chatterton, Percy: 17, 25, 37, 38 Chaudenson, Robert: 161, 194-195 Clark, J.B.: 37 Clerc, M.: 212, 232, 238, 245, 252, 253, 256, 257, 263 Cloarec-Heiss, France: 212, 229, 239 Cole, Peter: 368 Collijn, Isak: 44, 82 Collins, James T.: 479 Collins, Robert: 131, 142-143, 167 Comrie, Bernard: 450 Cook, Captain James: 454 Cooper, Fred: 275 Copland, B.D.: 472, 483, 484 Corriente, Frederico: 126 Coudreau, Henri: 101, 102, 120 Crabtree, O.: 145, 146 Crawford, John C : 303, 306, 357, 358 Crevaux, J.: 102, 119, 121 Crownhart-Vaughan, E.A.P.: 451-452, 453-454, 455, 466 D Daeleman, Jan: 194, 202 Dahl, Östen: 202 Dahlstrom, Amy: 86, 329, 360 Dall, William H.: 453 Daly, Martin: 167 Danckaerts, Jasper: 93 Dawkins, R.M.: 479 de Boeck: 265 de Bruijning, C.F.A.: 103 de Goeje, C.H.: 101-102, 103, 104, 112113, 114, 118, 119, 120-121, 122 de Laet, Johannes: 44, 50-51, 53, 64, 82, 85,94 de la Vérendrye, Sieur: 297 Denny, Ebenezer: 44, 57, 65, 69, 82, 85, 94,95 de Rasieres, Isaack: 72 Derendiger, R.: 133 de Vries, David: 72, 76 Dickason, Olive Patricia: 296
Names Index Diki-Kidiri, Marcel: 211, 214, 215, 220, 221, 222, 225, 227, 231-232, 245, 250, 251, 252, 253, 255, 257, 258, 259, 260, 265, 266 Dmytryshyn, Basil: 451-452, 453-454, 455, 466 Douaud, Patrick: 302 Dowse, Thomas: 300 Drapeau, Lynn: 341, 353, 355, 358, 360 Dunlap, A.R.: 81-82,95 Dusenberry, Vernon J.: 358 Dutton, Tom: 3,6,9,31,38,39 Dybowski, Benedykt: 452, 466 Dyk, Carolyn: 120 E Eboué, A.F.: 212, 233, 252, 253, 255, 256, 257, 265 Ehret, Christopher: 474, 475, 481, 482, 485 Elgibali, Alaa: 126, 131, 132 Emin Pasha: 130, 131, 139-140, 142, 143-144, 145, 146, 167 Escobar, Gabriel: 418 Escobar, Gloria: 418 Evans, Donna: 310, 311, 349, 359, 360 Everbroeck, R. Van: 230, 234, 245, 247, 248,250,251 Ewbank, Thomas: 94 F Fabian, Johannes: 168, 273 Farler, [Archdeacon]: 470 Fehderau, Harold: 173, 176, 181, 184, 185-186, 189, 193, 195, 196, 197, 198, 204, 205 Feierman, Steven: 471, 472 Foley, William A.: 2, 212, 485 Franssen Herderschee, A.: 101, 102, 103, 114, 120 Fredeen, Shirley: 357 Freeman-Grenville, Grenville S.P.: 274 Friederici Georg: 77, 94 Fück, Johann:126 Furley, O.: 131, 140, 146
Names
Index
G Gentil, Emil: 216 Gessi, Romolo: 129, 130, 138, 139, 142, 167 Girard, Victor: 443-446 Giraud, G.: 212, 264 Giraud, Marcel: 296, 297, 301, 354 Girault, Louis: 427, 428, 429, 430, 431, 432, 433, 434, 435, 436, 437-439, 439-440,441,443-446 Givón, Talmy: 161,254 Gnerre, Maurizio: 403-404 Godart, Louis: 216, 217 Goddard, Ives: 3, 5, 49, 58, 67, 68, 78, 79, 82, 84, 85, 87, 91, 92, 94, 95, 355, 356 Golovin, Pavel: 451-452, 453-454, 455, 463 Golovko, Evgenij V.: 449, 450, 452, 453, 454, 455, 456, 457, 458, 459, 460, 461, 462, 463, 466 Goodman, Morris: 126-127, 202, 470 Gordon: 130, 145 Gore, Mrs. E.C.: 212 Gore, the Rev. Canon E.C.: 212 Goutalier, Regine: 217 Grant, James: 129, 143, 167, 168 Gray, Richard: 130, 145, 167, 168 Green, E.Clive: 472,481 Green, Edward Crocker: 114 Greenberg, Joseph H.: 136, 264, 265, 292, 470 Grimes, Barbara Dix: 479 Gruiter, V.E. de: 355 Guthrie, Malcolm: 184 H Hamilton, Murray: 358 Hancock, Ian F.: 144,273 Harris, J.B.: 38, 39 Haugen, Einar: 53, 86 Heine, Bernd: 135, 147, 163, 190, 264, 265, 271, 273, 277, 278-280, 291, 293 Hellinger, Mariis: 212 Hetzron, Robert: 482, 485
497 Hillelson, S.: 168 Hinnebusch, Thomas J.: 203, 276, 279, 285, 292 Hitchcock, N.E.: 11,38 Hoff, B.J.: 119, 122 Hogman, Wes: 325, 354 Holmer, Nils: 85 Holm, John: 160, 204, 212-213, 223, 229, 255, 262, 263, 273, 277, 285 Holt, Peter: 167 Horton, Mark: 289 Howard, Joseph Kinsey: 298 Howe, Henry: 84 Hrdlička, Ales: 453 Hulstaert, Gustaaf: 189, 194, 200, 205 Huttar, George L.: 5, 101, 103, 111, 112, 114, 119, 120, 121 Huttar, Mary L.: 111, 120, 121 Hymes, Dell: 213 I Igla, Birgit: 355-356
J Jackson, Colonel: 94 Jackson, Walter S.: 104 Jakobi, Angelika: 162, 168 James, Deborah: 349 Jameson, J. Franklin: 72, 81, 82, 83, 95 Jansen, Bert: 234 Jephson, A. Mounteney: 131, 139, 142, 145, 167 Jochelson, Waldemar: 453, 454 Johnson, Amandus: 86 Jones, Morgan W.: 104, 120 Jouannet, Francis: 199 Junker, Wilhelm: 129, 130, 131, 139, 141-142, 144, 145, 146, 167 Kabu, Tommy: 11, 38 Kalm, Peter: 94 Kapanga, Mwamba Tshishiku: 190 Kappler, August: 102 Kaufman, Terrence: 83, 160, 204, 206,
498 212,213,215,219,263,353,355, 449, 455, 458, 462, 465, 470, 479, 485 Kaye, Alan: 130, 132, 135, 158, 163, 168 Keenan, Edward: 165 Keesing, Roger M.: 204 Key, Mary Ritchie: 443-445 Khlebnikov, Kiril Timofejevich: 452, 466 Kimambo, Isaria N.: 471, 472 King Leopold II: 186,203 Kobozo, Jean-Marie: 211, 215, 220, 221, 222, 227, 231, 232, 245, 250, 251, 252, 253, 255, 257, 258, 259, 260, 265, 266 Koopman, Hilda: 234 Kouwenberg, Silvia: 5 Krapf, Johann Ludwig: 471 Krumm, Bernhard: 273, 277 L Lagae, CR.: 212 Lange, Dierk: 166 Laurent, Joseph: 78 Laverdure, Patline: 295, 304, 306 Lawes, Frank: 16 Lawes, W.G.: 16-17, 18, 19, 20, 33, 38 Leavitt, Claude W.: 119, 121 Lefebvre, Claire: 387 Legére, Karsten: 273 Lehiste, Ilse: 5 Lehmann, Winfred: 255 Lekens, Benjamin: 211, 220-221, 223, 224, 225, 226, 227, 229, 230-231, 232, 233, 237, 238, 239, 240-241, 243, 246, 247, 248, 249, 250, 252, 254, 255, 258, 265, 266 Le Page, Robert: 376 Lepsius, Richard: 167 Lethem, G.: 132 Levin, Maksim G.: 450 Lindeström, Peter: 44, 63, 74, 76, 78, 86,91,93,94,95 Lister-Turner, R.: 37 Ljapunova, Roza Gavrilovna: 452, 455,
Names
Index
466 Lock, Maynard: 39 Louis XIV: 296 Luckenbach, Abraham: 45, 46, 73, 91, 93 Lumwamu, François: 176, 177, 181 Luther, Martin: 44 Lyons, John: 292 M MacGregor, Dr./Sir William: 20-21, 22, 25, 26, 29, 30 Mahmud, Ushari: 134, 135, 143, 146, 160 Maigret, Julien: 133 Matuką, Yeno Mansoni: 200, 206 McDougal, John: 300 Meinhof, Carl: 481 Melyn, Cornelius: 76 Menovšcikov, Georgij Aleksėjevič: 449, 450, 451-452, 453, 455, 456-457, 458,459,461,463,466 Michaëlius, Jonas: 81, 84, 95. Miller, Catherine: 133, 134, 135, 160 Miller, Christopher: 382 Mishaegen, Anne de: 300-301 Möhlig, Wilhelm J.G.: 470, 485 Mondaca, Jaime: 430, 434, 435, 437 Montaño Aragón, Mario: 443-445 Mosis, André: 102 Mous, Maarten: 469, 470, 471, 472, 473, 474, 475-476, 480, 481, 482, 483, 484, 485 Msemo, Mzee Muse Njarita: 471 Mufwene, Salikoko S.: 173, 175, 176, 179, 180, 181, 183, 184, 185, 187, 190, 192, 194-195, 198, 199, 200, 201-202, 203-204, 205, 213, 223, 236, 239, 248, 262, 264, 266, 273 Mühlhäusler, Peter: 5, 37, 38 Muraz, Gaston: 133, 146, 155 Murray, J.H.P: 38 Muysken, Pieter: 3, 234, 358, 365, 367, 368, 386, 387, 389, 390, 391, 397, 401,442,463 Myers, Albert Cook: 44, 62, 65, 68, 72,
Names
Index
78,82 Myers-Scotton, Carol: 485 N Nebel, Arthur: 168 Nelson, William: 84-85 Neumann, Peter: 102, 103, 120 Ngalasso, Mwatha M.: 176, 198, 203 Nhial, Abdon: 135 Nurse, Derek: 4, 274, 276, 278, 279, 280, 285, 289, 292 Oblitas Poblete, Enrique: 430, 431, 433, 434-435, 436, 437, 438, 440 Ohly, Rajmund: 292 Oram, N.D.: 11,38 Orser, Lori L.: 349, 360 Owens, Jonathan: 4, 131, 132, 133, 146, 147, 158, 163, 165 P Papen, Robert A.: 301, 302, 351, 357, 358, 463 Pasch, Helma: 209, 213, 214, 215-216, 218,219,226,254,265 Patkanov, S.: 452, 455 Penn, William: 44, 62, 65, 68, 78 Pentland, David H.: 358, 360 Persson, Andrew: 158 Persson, Janet: 158 Peterson, Jacqueline: 305 Petherick, John: 138, 167 Petherick, Katherine: 138 Pierre, M.J.: 119 Polornè, Edgar C.: 143, 194, 273, 277, 290 Poplack, Shana: 382 Potapov, Leonid P.: 450 Prince, J. Dyneley: 84-85, 88 Prokosch, Erich: 133 Pustet, Regina: 247 R Rabah: 216 Rabeh: 130, 132, 159, 167
499 Rabin, Chaim: 125 Ranaboldo: 428 Reichmuth, Stefan: 158 Reusch, Richard: 273 Rhodes, Richard: 295-296, 302, 306, 310,313,316,339,349,358,360 Riel, Louis: 298-299 Rood, N.: 212, 241 Rosing, Ina: 427-428, 430, 441 Ross, Alexander: 300 Roth-Laly, Ariette: 165, 168 Rouchdy, Aleya: 167 S Saghayroon, Ibrahim: 131, 145 Saignes, Thierry: 428 Samarin, William J.: 120, 133, 173, 186, 188-189, 189-190, 191, 192-193, 194, 198, 200, 202, 203, 205, 206, 210, 211,212,213,214,215-216,217, 225, 228, 233, 235, 237, 238, 239, 240, 244, 253, 258, 259, 262, 263-264 Sankoff, David: 382 Sasse, Hans-Jürgen: 470, 478-479, 480, 482, 484-485 Schmidt, Peter R.: 274 Schnitzer, Edward: see Emin Pasha Schoen, Ivan: 119, 120 Scholtens, Ben: 102 Schweinfurth, Georg: 129, 130, 138139, 145, 167, 168 Scotton, Carol Myers: 291 Sekerina, Irina A.: 449, 455, 456, 458, 459, 460 Selkirk, Lord: 297 Senoussi: 216 Sergejev: 453 Sharrock, Susan R.: 305 Shnukal, Anna: 22, 38 Shrofel, Salina M.: 305 Simson, Alfred J.: 403-404 Slobodin, Richard: 357 Smith, Norval S.H.: 121 Spagnolo, A.: 161, 162, 163, 164, 165, 168 Spear, Thomas: 274
500 Speck, Frank G.: 94 Spitaler, Anton: 126 Stanley, Sir Henry Morton: 130, 131, 139, 143-144, 145, 146, 167, 202, 206 Stark, Louisa R.: 367, 368, 427, 428, 429, 430, 431, 432, 434, 437, 439, 440 Stobie, Margaret: 305 Stoller, Paul: 143 Stolz, Thomas: 212 Strachey, William: 94 Sulaiman: 130, 142, 167 Suvorov, E.K.: 454 Swift, L.B.: 230, 239 T Tabouret-Keller, Andrée: 376 Talon: 296 Tàlos, Endre P.: 482, 485 Tanghe, Basilius Oct.: 211 Taylor, A.J.: 16, 37, 39 Thoden van Velzen, H.U.E.: 102 Thomas, Gabriel: 44, 51, 56-57, 85, 88, 89,94 Thomason, Sarah Grey: 83-84, 85, 88, 90, 95, 126, 131, 132, 160, 166, 202, 204, 206, 212, 213, 215, 219, 263, 353, 355, 449, 455, 458, 462, 465, 470, 478, 479, 482, 484, 485 Thorburn, D.: 146 Thurston, William R.: 479 Tisserant, R.P.: 212, 238, 241, 253, 257 Todd, Evelyn M.: 305 Tooker, William Wallace: 78 Torero, Alfredo: 368, 428, 429, 430, 433 Torozoni, Ngama-Nzombio Tra Ndele: 211, 220, 221, 223, 225, 228, 236 Toscano Matteus, Humberto: 369, 372 Tosco, Mauro: 131, 133 Tourneux, Henri: 132 Traugott, Elizabeth: 266 Trudel, Marcel: 358 Tucker, Archibald Norman: 162, 164, 167, 212, 241, 244, 249, 280, 469, 472, 481
Names
Index
Turner, William Wadden: 94 V Vakhtin, Nikolai .: 449, 450, 452, 453, 454, 455, 456, 457, 458, 459, 460, 461,462,466 Valdman, Albert: 194-195 van Baar, Tim: 358 van Coll, C: 102, 103 van den Bogaert, Harmen Meyndertsz: 83 van der Voort, Hein: 355-356 van der Ziel, C.N.: 119 van Eijk, Jan: 349 van Tuinen, Bishop: 104, 114 van Wetering, W.: 102 Velantie, Frank J.: 5 Velasco, Juan de S.J.: 371-372 Verrips, Maaike: 358 Versteegh, Kees: 126, 127, 166 Villelune, Baron E. de: 240, 244 Vincent, George: 135 von Sack, Albert: 102 Voorhoeve, C.L.: 38 Voorhoeve, J.: 103 Vorbichler, Anton: 163 Vossen, Rainer: 167 W Wald, Benji: 273, 288, 292 Walters, Francis J.: 305 Watson, Richard: 133, 134-135, 155, 157, 168 Weaver, Deborah: 359 Weinreich, Uriel: 226 Weslager, CA.: 81-82,95 Westermann, Dietrich: 265 West, Francis: 38 Westlund, Börje: 86 Whinnom, Keith: 5, 143, 263 Whipple, Amiel Weeks: 94 Whiteley, Wilfred Howell: 291, 292, 470, 483 Williams, Jeffrey P.: 2 Williams, Lorraine: 84
Names
Index
Williams, M. Joseph: 277 Williams, Roger: 78, 93, 94 Woidich, Manfred: 132 Wolfart, H. Christoph: 302, 305, 310, 313, 349, 360 Wurm, Stephen A.: 37, 38, 39
501 Z Zeisberger, David: 46, 47, 54, 55, 56, 57,70,73,75,85,88,90,91,92 Zeltner, Jean-Claude: 130, 132, 167 Zola, E.W.: 230, 239 Zoubé, Cyprien: 216, 217 Zubeir Rahman: 130 Zwettler, M.: 126
Subject Index A Attitudinal factors in motivating language shift: 478 Azania Liberation Front (Sudan): 134 Basic vocabulary: 7, 485 Basilectal variety: 134, 157, 160 Bilingual mixed languages: 1, 295-363, 365-426, 427-447, 449-468, 469-487; via sudden creation: 6, 485; as new ethnic languages: 6, 355, 376, 450451, 463, 465, 483; as secret lan guages: 6; via gradual replacement: 6; as in-group languages: 6, 428, 429, 442, 478 Bilingualism rather than pidginization/ creolization: 127, 167, 280, 300, 303 Borrowability: 355-356, 382-383, 386387 Borrowing of morphology: 478-480 Bushnegroes: 99, 103, 114 Code alternation: 464 Code-switching/code-mixing: 300, 301, 325, 352, 355, 356, 396, 408, 418, 420, 451, 464, 485; difficulty of distinguishing code-switching from borrowing: 325, 419 Colonial administration and language contact: 20-25, 33-34, 95, 133, 138, 139-145, 167, 168, 173-175, 181, 185-186, 188-189, 190, 191, 192-193, 203, 214, 215, 216-217, 261, 264, 271, 272, 368, 428-429, 435, 451, 453-454 Compounding as a means of pidgin
vocabuary coinage: 72, 247 Contact language (defined): 3-4; distinguished from variety of lexifier language: 211 Contact languages as marks of ethnic/ social identity: 134, 144, 355, 375, 376-377, 407, 427-428, 442, 450-451, 463, 465, 470, 483, 484 Continuum (especially between contact language and lexifier language): 134, 183-184 Convergence in pidgin/creole genesis: 164-165, 201, 311-312; see also Multiple origins for contact-language features Creole genesis: see Pidgin/creole genesis Creolization: 2, 6, 133, 135, 182, 194198, 217, 262; as nativization of pidgin?: 173, 185, 186, 195, 203-204, 211,261,263 D Decreolization: 160, 166 Dialect differences in contact languages: see Variation in contact languages Dialect leveling: 213 Diasystem: 176-177, 181 Double-articulated stops in contact languages: 7,221,222 Double marking in contact-language morphology: 432, 458, 481 E Endogenous vs. exogenous pidgins, substrate influences in: 161 European trade and colonization: 1, 102, 193, 194, 271, 274, 297, 368, 451 Expansion in creolization: 262
504
Subject
Index
Foreigner talk: 6, 11, 17, 19, 80, 215, 375, 376
Loanwords in basic vocabulary: 7, 239240, 277, 354, 475-476, 485
G Gender: see Noun classes Genetic relationship of languages: 3, 1718, 160 Grammaticalization: 219, 248, 251, 252, 260, 265, 287, 289, 290
M Métis Nation: 295, 296-305 Missionaries and contact languages: 16, 17, 18, 31, 33, 44, 167, 175, 181, 182, 185, 186, 189, 190, 193, 194, 195, 198, 205, 206, 216, 217, 245, 252, 264, 265, 266, 271, 275, 296, 354, 368, 371-372, 454, 455 Mixed marriages: French and Indian: 296, 297, 300, 451-452, 454, 462-463 Morphological replacement through borrowing: 478-480, 482-483, 485 Morphology in contact languages: 13, 57, 74, 79, 82, 148, 156, 159, 165, 175-180, 187, 199-200, 201, 223-230, 232-234, 241, 242, 260, 261, 285, 313-322, 351, 382-388, 404, 429, 432, 433, 435, 437-441, 457-460, 461, 464-465, 473-474; in Swahili: 281-288 Multiple origins for contact-language features: 241, 242, 248, 256; see also Convergence in pidgin/creole genesis Mutual intelligibility: 2, 13, 14, 17, 134, 136, 157, 166, 167, 181, 206, 213, 214, 219, 223, 255, 265, 277, 303, 375, 450, 472, 476
H "Have" in contact languages: 28-30, 35, 39, 72, 108, 121, 150, 158, 320-321, 366, 372, 475, 477 I Ideophones: 258-259, 260 Implosives in contact languages: 222, 473 Interlanguage: 405-407, 416; see also Second language acquisition and use Interpreters: 82, 138, 145, 188, 191, 192, 213, 214, 215-216, 297, 300, 303, 352 Koinés: 3, 185, 186, 204, 212-213; Arabic koiné: 125 Kongo Kingdom: 174, 202, 203 L Language death: of Ndyuka-Trio Pidgin: 103; possible attrition in Plains Cree: 351; see also Simplification Lexifier language: 2, 204, 211, 213, 218, 255, 260, 263, 277 Lingua franca: 4, 6, 7, 9, 13, 19, 20, 37, 43, 83, 127, 133, 135-136, 146, 168, 176, 181-182, 183, 185, 186, 188, 189, 190, 191, 192, 193, 194, 203, 204, 205, 206, 213, 214, 215, 216, 217, 218, 261, 264, 271, 272, 275, 279, 286, 287, 289, 301, 302, 354, 367, 368, 369, 371, 386 Linguistic creativity: 464, 465, 483, 485
N Nasalized vowel phonemes in contact languages: 104, 220, 307, 308, 309310,312 Native speakers of pidgins?: 13 Nonstandard dialects of Arabic: 127 Normalization (development of norms in a contact language): 173, 182, 195, 203, 205; see also Stability of contact varieties Noun classes: in Kitúba: 178-179; in Sango: 232-234; in Bantu and Bantubased pidgins: 281, 292; in Swahili: 281-283, 292; in Michif: 315-316,
Subject
Index
317, 324, 325, 326-328, 331, 337; in Media Lengua: 418; in Callahuaya: 433; in Mednyj Aleut: 457, 458; in Ma'a: 473-474 Official-/national-language status: of contact languages: 9, 209, 264; of Swahili: 272 OSV word order in pidgins: 6, 30, 31, 105, 336-337 P Passive in contact languages: 165-166 Percentage of native speakers of lexifier language: 136, 139, 140-141, 266 Pidgin/creole genesis: 173, 185-194, 199; conditions for: 143, 183, 211, 212, 213, 217-218, 275, 289, 290, 291, 375-376, 462 Pidgins, different, co-existing in the same area: 22, 95 Postpositions in contact languages: 111, 334, 396, 460 Predicting the emergence of mixed languages: 463-464 Prestige: 183, 184, 205, 206; as a factor in pidgin/creole genesis: 143, 218, 260, 275, 303, 368, 369, 454, 463 R Red River Métis: 296; see also Métis Nation Relexification: 258, 277, 365, 366, 375, 376, 377-378, 379, 386-388, 389-390, 394-395, 396, 397, 401-403, 417, 418,419,478,480,481,484 Repidginization: 183
Second-language acquisition (especially as a factor in contact-language genesis) and use: 38, 127, 140, 144, 146, 182-183, 204, 259, 260, 272, 273, 287, 290, 291, 312, 354, 374,
505 375, 376, 405-407, 420, 462, 464, 465, 476 Secret languages: 6, 429, 442, 476 Semi-creoles: 3 Serial verb constructions: in Kitúba: 180-181; in Sango: 234-237, 261 Simplification: in pidgins and creoles: 2, 13, 16,80,81, 143, 179, 181,215, 223, 224, 225, 232, 262, 441; in Quechua: 369; in Swahili: 281-283, 287, 288-290; see also Foreigner talk, Language death Slavery: 99, 130, 138, 139, 141, 142, 143-144, 161, 167, 168, 185,216, 217,218,275,471 Sources of contact-language structure and lexicon: 26-33, 43-44, 59, (4563,) 64-81, 89-90, 104-117, 160-166, 348-352, 455-461, 477-478, et passim SOV word order in contact languages: 6, 30-31, 62-63, 83, 105, 117, 336, 397, 404, 406, 436, 460 Stability of contact varieties: 2, 4, 17, 44-45, 75-76, 102, 103, 127, 133, 136, 144, 166, 173, 212, 215, 219, 235, 245, 255, 264, 403, 407-408, 450, 465 Substrate influences in pidgin/creole genesis: 160-163, 166, 168 T Target language in pidgin/creole genesis?: 206, 215, 218, 261, 375 Tommy Kabu Movement: 11, 38 Tone: in contact languages: 104, 176, 181, 195, 199, 223, 231, 237, 260, 261, 262; in Swahili: 279 Trading relationships: 1, 9, 14, 16, 18, 37, 79, 99, 102-103, 114, 118, 129130, 138, 141, 143, 144, 145, 185, 189, 193, 216, 264, 271, 274, 275, 289-290, 291, 297, 298, 368, 451, 453 Trade jargons: 118-119 Trade languages: see Lingua franca Trade pidgins: 6, 9, 17, 99, 102
506 Translators: see Interpreters Two-language pidgins and creoles: 5 U Universal simplifying principles/ developmental tendencies in pidgin/ creole genesis: 28, 32-33, 83, 219 V Variation in contact languages: 11-12, 34, 38, 75-77, 90, 101, 104, 111, 112, 113, 135, 157-158, 173, 182-184, 186, 195-198, 245, 256, 257, 273,
Subject
Index
295, 302, 306-307, 309, 334, 348, 367, 369, 376, 385, 408-409, 410419, 431, 438, 456-457, 460, 481482; see also Continuum Vernacularization: 203-204 Vulgar words in contact-language lexicon: 80-81 W Wantoks: 22-23 Withholding the "real" language from outsiders: 5, 16, 19,38, 114