Linguistic Areas Convergence in Historical and Typological Perspective
Edited by
Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Linguistic Areas
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
This page intentionally left blank
Linguistic Areas
Edited by
Yaron Matras University of Manchester
April McMahon University of Edinburgh
and
Nigel Vincent University of Manchester
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Convergence in Historical and Typological Perspective
Editorial matter and selection © Yaron Matras, April McMahon and Nigel Vincent 2006 Individual chapters © contributors 2006
No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP. Any person who does any unauthorized act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The authors have asserted their rights to be identified as the authors of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2006 by PALGRAVE MACMILLAN Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN-13: 978–1–4039–9657–2 ISBN-10: 1–4039–9657–1 This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Linguistic areas : convergence in historical and typological perspective / edited by Yaron Matras, April McMahon, Nigel Vincent. p. cm. Includes bibliographical references and index. ISBN 1–4039–9657–1 (cloth) 1. Areal linguistics. 2. Historical linguistics. 3. Typology (Linguistics) 4. Languages in contact. I. Matras, Yaron, 1963– II. McMahon, April M. S. III. Vincent, Nigel. P130.L56 2006 417—dc22 2005056583 10 15
9 14
8 7 6 5 13 12 11 10
4 3 2 1 09 08 07 06
Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham and Eastbourne
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission.
List of Figures and Maps
vii
List of Tables
viii
List of Abbreviations
ix
Notes on the Contributors
xii
Introduction
xv
1 Areal Linguistics: A Closer Scrutiny Lyle Campbell 2 All or Nothing Thomas Stolz
1
32
3 Keeping Contact in the Family: Approaches to Language Classification and Contact-induced Change April McMahon and Robert McMahon
51
4 Linguistic Areas, Language Contact and Typology: Some Implications from the Case of Ethiopia as a Linguistic Area Walter Bisang
75
5 Structural Isoglosses between Khoekhoe and Tuu: The Cape as a Linguistic Area Tom Güldemann
99
6 The Sri Lanka Sprachbund: The Newcomers Portuguese and Malay Peter Bakker 7 On the Roles of Turkic in the Caucasus Area Lars Johanson 8 The Circle That Won’t Come Full: Two Potential Isoglosses in the Circum-Baltic Area Maria Koptjevskaja-Tamm v 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
135
160
182
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Contents
Contents
9 Complex Emotion Predicates in Eastern Indonesia: Evidence for Language Contact? Simon Musgrave
227
10 Another Look at Australia as a Linguistic Area Claire Bowern
244
11 Towards a Typology of the Siberian Linguistic Area Gregory D. S. Anderson
266
Index of Authors
301
Index of Language Families, Languages and Dialects
306
Index of Subjects
311
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
vi
Figures 2.1 Degrees of similarity 2.2 Languages with overt DET-marking versus languages without overt DET-marking 2.3 Competing isoglosses 3.1 Output of Network for most conservative sublist, Romance and Germanic 3.2 Output of Network for least conservative sublist, Romance and Germanic 3.3 Network for Germanic, ‘wing’ coded as cognate English–North Germanic 3.4 Network for Germanic, ‘wing’ coded as loan into English 3.5 NeighbourNet for Indo-Iranian hihi sublist 3.6 NeighbourNet (Bryant and Moulton, 2004) for Indo-Iranian lolo data 3.7 Neighbour-joining tree of Australian data, drawn using Splitstree 3.8 Splitstree graph for 26 Australian languages 3.9 NeighbourNet graph for 26 Australian languages 5.1 Genealogical classification of Khoe-Kwadi 5.2 Genealogical classification of Tuu 8.1 Circum-Baltic languages 11.1 Russia and the languages of Siberia
34 41 43 61 62 63 64 66 67 69 70 71 102 102 183 267
Maps 5.1 Distribution of Khoisan lineages in the early colonial period 8.1 The Circum-Baltic languages 9.1 Indonesia, showing the location of the languages discussed
vii 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
101 183 230
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
List of Figures and Maps
2.1 Representatives of DET-types in Europe 4.1 Non-phonological features relevant for the Ethiopian linguistic area 4.2 Number of reconstructed glottalized consonants for later stages of Proto-Afro-Asiatic 5.1 The Khoekhoe consonant system 5.2 The ⏐Xam consonant system (preliminary) 5.3 Consonant inventories across the Khoe family 5.4 Consonant inventories across the Tuu family 5.5 Probable lexical borrowings of Khoekhoe from Tuu 5.6 Free, morphologically complex pronouns of !Ora 5.7 Basic time marking in selected Khoekhoe and Tuu languages 6.1 Nominal case marking in Tamil, Sri Lanka Malay and SL Portuguese 6.2 Particles in SL Portuguese, SL Malay and SL Tamil 6.3 Verbal categories in SL Portuguese, SL Malay and SL Tamil 6.4 Pronouns in three Malay varieties 6.5 Sinhala and Vedda nominal endings 8.1 Some areal phenomena in the CB area 8.2 Word accents in Baltic varieties 8.3 Tone distribution in Swedish, Dalecarlian and Norwegian (Swedish examples) 8.4 Quantity in Estonian 9.1 Relative frequency of complex and simple predicate constructions for emotional states 9.2 Languages ordered by proportion of complex predicates (low to high) and geographic position 9.3 Profile-based uniformity measures 10.1 Basic and non-basic vocabulary counts between Bardi, Yawuru and Karajarri 10.2 Samples of Proto-Pama-Nyungan and Proto-Karnic reconstructed lexical items 10.3 Karnic suppletive pronominal forms 10.4 Hypothesized stages in the split of Karnic languages 11.1 List of morphological features, grouped by family 11.2 List of select phonological and syntactic features
viii 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
40 79 80 107 107 108 108 110 112 117 141 143 144 147 150 186 195 196 197 238 239 240 254 258 259 261 293 294
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
List of Tables
1, 2, 3 ABL ABS ACC ACTR AD ADJ AL ALL ANA AOR APPL ARTCL ASS AUGM AUX BEN C CAUS CLASS CNC COMP COND CONN CONV CONV:MAN COP CSY CV D DAT DECL DEI DIM DIR DL DS DSBJ
first, second, third person ablative absolutive accusative actor addressee adjective alienable allative anaphor aorist applicative article associative augmentative auxiliary benefactive common causative classifier concord complementizer conditional clause connector converb converb of manner copula Central Siberian Yupik converb dual dative declarative deictic diminutive directional dual different subject deposed subjectn ix
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
List of Abbreviations
E EMPH EVID F FEM FIN FUT GEN GIV ILLAT IMP IMPV INCH INF, INFIN INS, INSTR INTR IPFV IRR LOC M MPO NEG NEGPOT NFUT NM NOM NP OBJ OBL OBLQ P, PL P/F PAS, PASS PAST PAT PERF PL POSS PP PPL PQ PRART PRED PRES
exclusive emphatic evidential feminine feminine finite future genitive given illative imperative imperfective inchoative infinitive instrumental intransitive imperfective irrealis locative masculine multipurpose oblique negative negative potential non-future noun marker nominative noun phrase object obligation oblique plural present-future passive past tense patient perfect plural possessive postposition participle polar question prepositional article predicative present
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
x List of Abbreviations
PRF PRO PROB PROP PROSEC PRTCPL PST PSYE PURP Q QUOT RECIPR RED RED, REDUP REDPL REF REL REL RELV REP RFLXV S, SG SBEN SBJ SF SP STAT SUBJ SUBJ SUBORD TAG TEMP TR TRANS UNACCOMPL VOL XXVI
perfective pronoun probabilative proper name prosecutive participle past Proto-Siberian Yupik Eskimo purposive question quotative reciprocal reduplication reduplication reduplication referential relative relativizer relevance repetitive reflexive singular self-benefactive subject stem-formant speaker stative subject subjunctive subordinator tag question temporal transitive transitive unaccomplished volition nominal class XXVI
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
List of Abbreviations xi
Gregory D. S. Anderson is a Research Associate at MPI-EVA/Leipzig, Germany and the University of Oregon Department of Linguistics, Eugene, USA. He is the author of grammars on Xakas and Tuvan, two Tuvan–English dictionaries, and a volume on auxiliary verb constructions in Altai-Sayan Turkic languages. Forthcoming volumes are on language contact in South-Central Siberia, and a general–typological study of auxiliary verb constructions. Research interests include languages of Siberia, Munda and Austroasiatic languages, Burushaski and Eleme (Nigeria). Peter Bakker is a Lecturer in the Department of Linguistics at Aarhus University, Denmark. He is the author of monographs on twins languages (Autonomous Languages, 1987) and the genesis of Michif (A Language of Our Own, 1997), co-editor of two books on mixed languages (Mixed Languages, with M. Mous, 1994, and The Mixed Language Debate, with Y. Matras, 2003). He also compiled a linguistic bibliography of Romani with Y. Matras (2003). His main research interests are language contact, language genesis and language change. Walter Bisang is Professor of General and Comparative Linguistics at the University of Mainz, Germany. Since 1999 he has been the spokesman for a Collaborative Research Centre on ‘Cultural and linguistic contacts: Processes of change in Northeast Africa and Western Asia’ financed by the German Science Foundation. He is the author of Hmong-Texte, eine Auswahl mit Interlinearübersetzung (Zürich, 1988) and Das Verb im Chinesischen, Hmong, Vietnamesischen, Thai und Khmer (Gunter Narr, 1992) and numerous articles. His main interests are language typology and language universals, formal versus functional linguistics, grammaticalization, and language contact–areal typology. Claire Bowern is Assistant Professor in the Department of Linguistics at Rice University, Honston, Texas, USA. She is the editor (with Harold Koch) of Australian Languages: Classification and the Comparative Method. Her main research interests involve historical syntax and comparative reconstruction of Australian languages. Lyle Campbell is Professor of Linguistics and Director of the Center for American Indian Languages at the University of Utah, Salt Lake City, Utah, USA. He is the author of fifteen books and 160 articles. He has twice won xii 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Notes on the Contributors
xiii
the Linguistic Society of America’s prestigious Leonard Bloomfield Book Award, first for jointly with Alice Harris Historical Syntax in Cross-linguistics Perspective (1995, Cambridge University Press), then for American Indian Languages: The Historical Linguistics of Native America (1997, Oxford University Press). His research specializations are Native American languages, endangered languages, historical linguistics, typology, sociolinguistics, linguistic anthropology, Uralic and Hispanic linguistics. Tom Güldemann is a Research Associate in the Linguistics Section of the Max Planck Institute for Evolutionary Anthropology, Leipzig, and teaches at the Department of African Studies of the University of Leipzig, Germany. His main research interests are typology, historical linguistics, and the study of Khoisan and Bantu languages. Lars Johanson is Professor of Turcology in the Seminar für Orientkunde at the Johannes Gutenberg-Universität in Mainz, Germany. He is the author of Aspekt im Türkischen (1971), Alttürkisch als ‘dissimilierende Sprache’ (1979), Linguistische Beiträge zur Gesamtturkologie (1991), Strukturelle Faktoren in türkischen Sprachkontakten (1992), Discoveries on the Turkic Linguistic Map (2001), Structural Factors in Turkic Language Contacts (2002), and other books. His main research interests are general and comparative Turkic studies, historical linguistics, code-copying, and aspectology. Maria Koptjevskaja-Tamm is Professor in the Department of General Linguistics at Stockholm University, Sweden. She is the author of Nominalizations (1993, Routledge) and co-editor (with Östen Dahl) of Circum-Baltic Languages (2001, Amsterdam: Benjamins, 2 vols) and contributed to the World Atlas of Language Structures (2005, Oxford University Press). Her research interests are various morpho-syntactic phenomena (including nominalizations, possessive constructions, finiteness) in cross-linguistic and typological perspectives, and cross-linguistic research on lexical semantics, areal typology and traditional historical linguistics. Yaron Matras is Professor of Linguistics in the School of Languages, Lingustics and Cultures at the University of Manchester, UK. His books include Romani: A Linguistic Introduction (2002), The Mixed Language Debate (co-edited with Peter Bakker (2003)), and Markedness and Language Change (with Viktor Elxík (2006)). April McMahon is Forbes Professor of English Language at the University of Edinburgh, UK. She worked previously in the Department of Linguistics at the University of Cambridge and held a Chair in English Language and Linguistics at the University of Sheffield, both UK. Her research interests involve the interaction between phonological theory and historical evidence,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Notes on the Contributors
Notes on the Contributors
as well as issues of language comparison and classification. Her books include Understanding Language Change (Cambridge University Press, 1994), Lexical Phonology and the History of English (Cambridge University Press, 2000), Change, Chance, and Optimality (Oxford University Press, 2000), and Language Classification by Numbers (with Robert McMahon, Oxford University Press forthcoming 2005). She was formerly the President of the Linguistics Association of Great Britain. Robert McMahon is a clinical molecular geneticist at the Western General Hospital in Edinburgh, UK. His work involves tracing inherited conditions through families, and in particular has researched and provided genetic services for cystic fibrosis, muscular dystrophy, inherited cancer and osteogenesis imperfecta (brittle bone disease). He has published a range of articles in professional and scientific journals, and maintains a research interest in issues of human genetics and evolution, and their relationship with language. He is co-author of Language Classification by Numbers (forthcoming, OUP). Simon Musgrave is a post-doctoral Fellow in the Department of Cultures and Linguistics at Monash University, Victoria, Australia. His research interests include languages of Maluku, Indonesia; the syntax and typology of Austronesian languages; non-derivational theories of syntax; and computational tools for linguists. He is co-editor (with Peter Austin) of the volume Voice and Grammatical Relations in Austronesian Languages (forthcoming). Thomas Stolz is Professor of Linguistics at the University of Bremen, Germany. He is the author of Gibt es das kreolische Sprachwandelmodell? (1986), Sekundäre Flexionsbildung (1992) and Die Numeralklassifikation im Klassischen Aztekischen (2001). His research interests are language contact, areal linguistics, typology, language change, morphology, grammaticalization, and the history of linguistic thought. Nigel Vincent is Associate Dean and Mont Follick Professor of Comparative Philology in the Department of Linguistics at the University of Manchester, UK. He has published Widely a Romance languages and has recently translated into English Tore Janson’s book A National History of Latin (2004). He is a former president of the Philological Society and for ten years was editor of the Journal of Linguistics.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
xiv
Introduction
Editing a volume on linguistic areas is both a fascinating and a worrisome undertaking, for essentially the same reason: ‘linguistic area’ is a rather ill-defined and amorphous notion, for which ‘The number of definitions is almost coextensive with the number of linguists working in the field’ (Stolz, this volume, p. 33). Working with a notion which is yet to be defined conclusively can be exhilarating, since by definition there must be work to be done; but it can also be frustrating, since it can sometimes feel as though metatheoretical discussions are impeding progress on the all too necessary description and analysis of particular languages and linguistic areas. It is precisely because of the potential collision between the need for clear and detailed description on the one hand, and theoretical and definitional considerations on the other, that we feel this volume is so timely. For exactly the same reason, the volume does not seek only to describe particular situations, or only to discuss the theoretical issues in a more general and discursive way, but to consider both perspectives, and indeed also the influence of each on the other. Textbook accounts of linguistic areas, perhaps typically, begin by giving apparently straightforward and uncontentious definitions. Campbell (1998: 299) tells us that ‘Areal linguistics, related to borrowing . . . is concerned with the diffusion of structural features across language boundaries within a geographic area’, while Trask (1996: 315) suggests that: centuries of contact between languages can lead to a particularly striking result: several neighbouring but unrelated languages can come to share a number of structural properties with one another, properties which they do not share with their closest genetic relatives elsewhere. A group of languages in which this situation obtains is called a LINGUISTIC AREA, or, using the German term, a SPRACHBUND. Both Trask and Campbell also introduce the term ‘convergence’ for the process of progressive structural assimilation that is characteristic of linguistic areas. However, this appearance of definitional harmony is short-lived, and Campbell in particular goes on to raise a number of salient and persistent difficulties with the notion of linguistic area. Although we have seen that both he and Trask characterize linguistic areas as involving structural borrowing, Campbell (1998: 300) proceeds almost immediately to invoke xv 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon, Yaron Matras and Nigel Vincent
Introduction
‘not only borrowed words, but also shared elements of phonological, morphological or syntactic structure’. So, is lexical as well as structural convergence necessary in a linguistic area? How is that ‘or’ to be interpreted – would we characterize a contact relationship as a linguistic area if lexical and phonological borrowing were involved, but not morphology or syntax, or do we need evidence of interpenetration of features across the board? Does bidirectionality of borrowing relationships count more than the number of features borrowed? Where do we place the cut-off point between a ‘normal’ case of borrowing, and something severe and profound enough to merit the label of a linguistic area? Is the number of languages involved one of the more salient features, as implied by Thomason’s (2001) characterization of linguistic areas as necessarily involving three or more languages? Or is the definition of linguistic areas more a negative one? Thomason (2001: 99) further suggests that ‘perhaps the most prominent similarity . . . is the lack of an explanation for many or most of the areal features’. It is notable also that Trask, Campbell and Thomason all work mainly by illustration rather than by definition in their outlines of linguistic areas: in other words, all work towards what a linguistic area is by giving examples of what a linguistic area has been said to be, rather than by giving a fairly watertight definition and some examples of situations the definition excludes. Even Heine and Kuteva (2005: 174), who do provide a set of characteristics they see as indicative of linguistic areas, note that ‘This characterization is fairly general, it is not meant to be a definition; rather, we use it as a convenient heuristic for identifying possible instances of sprachbunds.’ In turn, the illustrations given tend to show how the concept is elastic and problematic, since it is debatable how far the candidate examples share unifying characteristics. In short, we move rapidly from an apparently clear and helpful definition to the admission that ‘what we understand about linguistic areas is depressingly meager, compared to what we don’t understand about them’ (Thomason 2001: 99). We do not only have a confusion over definitions, or perhaps a proliferation of partially overlapping definitions, but also a proliferation of partially overlapping terms. Campbell (1998) talks not only about linguistic areas, but also about Sprachbunds, diffusion areas, adstratum relationships and convergence areas. While these are seen to a large extent as being intertranslatable terms, there are attempts elsewhere to subdivide linguistic areas, so that Heine and Kuteva (2005: 172), for example, distinguish Sprachbunds, which are ‘defined by the presence of a limited set of linguistic properties’; metatypy, where ‘the languages concerned exhibit a high degree of mutual intertranslatability’; and grammaticalization areas, which ‘are the result of one and the same historical process, more specifically, of the same process of grammaticalization, even if there may be other properties in addition’ (though Heine and Kuteva also concede that these types may not be mutually exclusive). This foregrounding of grammaticalization echoes Campbell’s
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
xvi
(1998: 306) suggestion that different features may be weighted differently in the establishment of a linguistic area, though he does not single out a particular feature or feature type. Again, however, we return to the paradoxes that seem inherent in ideas of convergence, since Heine and Kuteva (2005: 182) suggest that ‘The contribution that the study of grammaticalization can make to defining areal relationship is a modest one’ (whereas, on the other hand, they see cases of metatypy, like Kupwar in India or the East Tucanoan-Tariana contact region in north-west Amazonia, as almost prototypical cases of convergence). These definitional questions, of course, are not peculiar to discussions of linguistic areas. In the literature on many linguistic topics, there have been lengthy discussions of what particular constellation of features we need to identify in order to ‘approve’ a given label. On the other hand, we can ask whether the very existence of those contributory features means there is no need to recognize the superordinate category of grammaticalization or convergence in their own right. Both these rather mechanical, feature-counting and checking approaches can be challenged. Accepting that components exist does not necessarily rule out the recognition of the higher-order category they compose: hydrogen and oxygen are real, but so is water. On the other hand, checking off the number and type of features in any specific case before we agree to identify, say, a linguistic area can be equally side-tracking. The conclusion might be that, since we already know that so many linguistic categories and phenomena are graded, with prototypical and less clear cases, we should not be so reluctant to recognize that there might be more central and more peripheral linguistic areas. Of course, if there were a clear and absolute ruling on how we define linguistic areas, there would be far less need for this volume. The papers included here fall into two categories, although it is perhaps fitting that there is no absolute dividing line between the two: while some focus on how, and indeed whether, linguistic areas can be defined, others deal with the characteristics of individual cases, more or less clear, of convergence. Of the chapters that focus on individual convergence areas, it is in keeping with the theme of the volume that few deal with completely uncontentious or generally accepted examples: either new examples are proposed (as in Anderson); or there is a concentration on individual, less clear aspects of better-known areas (Bakker); or the focus is on a muchdiscussed area, but convergence is not accepted at face value (as in Bowern’s discussion of Australia). Anderson’s outline of Siberia as a linguistic area provides a sound descriptive basis for a discussion of a relatively unfamiliar situation, setting the scene for the assessment of a range of features that appear not to be typologically, logically or genetically related, and which are therefore plausibly indicative of contact. Similarly, Güldemann’s account of possible influence of Tuu on Khoekhoe develops arguments for a Cape linguistic area, and provides a salutary counterpoint to the Greenbergian
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon, Yaron Matras and Nigel Vincent xvii
Introduction
assumption of relatedness among all Khoisan languages. Johanson and Bakker focus on the contribution of individual languages to larger, putative linguistic areas, the former on Turkic in the Caucasus, and the latter on Portuguese and Malay, two newly ‘Indianized’ immigrant languages introduced relatively recently to the well-known Sri Lankan Sprachbund. This in turn is connected to the larger South Asian linguistic area; Heine and Kuteva (2005: 177) suggest that such ‘macro-areas’, also including the Balkans, and Meso-America, ‘are notoriously messy’, and can better be characterized by regarding them as being composed of several smaller ‘micro-areas’, which may be more amenable to investigation. Bisang takes a different approach to a further macro-area, namely Ethiopia, arguing that we should not seek to be too stringent in our application of absolute criteria for the identification of linguistic areas, but that equally we should not reject the category altogether (as Campbell suggests); rather, we need to relax the concept, recognizing a more clinal, graduated concept of zones of contact-induced structural convergence. Stolz, also on the basis of general and theoretical discussion rather than concentration on a specific example, tends to agree with Campbell that historical linguists can become too preoccupied with exactly how we define linguistic areas, concluding (p. 46) that we would do better to ‘either strip the term of its unwelcome and much too suggestive connotations or abolish it for good (but it should be kept in the virtual museum of linguistic thought as an example of how difficulties and misunderstandings can be created by terminology)’. Other contributions, however, problematize the individual examples, but without necessarily seeking to do away with the general concept. Thus, Koptjevskaja-Tamm assesses the contribution of two completely independent sets of features, one syntactic and the other prosodic, to the identification of the Circum-Baltic linguistic area; but although finding a whole series of indubitably contact-induced similarities, she is unable to identify any similarity that unites all the languages spoken around the Baltic. The case for the existence of a Circum-Baltic convergence area thus remains unproved: but this is an excellent example of how we must check and interrogate our evidence if we are to tell convincing cases of convergence from less convincing ones, or properly assess the extent and type of contact that must be at issue for a linguistic area to be identified. In Chapter 10, Bowern also challenges a claimed linguistic area, though on somewhat different grounds. She does not take issue with the geographical distribution of the linguistic features themselves, but rather with the historical mechanisms by which they have come to populate their existing territory. Bowern assesses Dixon’s punctuated equilibrium approach to Australian languages, and proposes an alternative account involving divergence in situ of neighbouring dialectal speech communities, which none the less remained in contact. Genetically related languages would in this case be difficult to subgroup and place on a family tree, since in some cases
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
xviii
isoglosses would overlap, leading to conflicting signals of subgrouping; without seeking to argue against the concept of linguistic areas in a global sense, Bowern none the less challenges it as the only plausible historical account of the contemporary Australian linguistic situation. Bowern also raises the question of how we might in practical terms determine whether specific shared features do reflect contact, or whether they may reflect some other historical explanation, either genetic inheritance or parallel development perhaps. This is absolutely crucial to our development of general theories of convergence, and to our testing of specific cases, and two final chapters address this question in particular. In Chapter 3, McMahon and McMahon consider how quantitative methods might allow us to identify contact-induced features, and how these might behave in simulated and real language data; Musgrave, in Chapter 9 though also raising these general issues in outline, focuses specifically on a particular test case, involving shared features between Austronesian and non-Austronesian languages of Eastern Indonesia. Although these might be seen as contact-induced, and might therefore encourage the proposal of Eastern Indonesia as a linguistic area, Musgrave argues that the data do not support this interpretation. Future discussions might certainly benefit from both the application of quantitative or statistical methods, and the careful analysis of the particular dataset at issue. The chapters collected here, then, represent a whole range of approaches to the idea of linguistic areas: some are positive about the general approach, though they may not find evidence for convergence in a specific situation; others argue that the concept is unhelpful; and yet others argue for its modification, or the application to new data. Together, they provide an appropriately multi-faceted picture of a complex and fascinating topic, which while not easily characterized, is none the less interesting, not least because of its applicability to many language contact situations, regardless of geography and typology. Bringing this range of viewpoints together depends above all on the authors, and we thank them for their prompt delivery of their manuscripts, and their full engagement with readers’ comments. The idea for this volume originated in a North-West Centre for Linguistics conference in 2002 (although not all the papers from the conference are included in the volume, and some papers here were not presented there), and we thank Francesco Goglia, Diane Blakemore and Kevin Watson for their assistance with the organization of that meeting, and the Mont Follick Fund of the University of Manchester for its contribution to costs. We also thank an anonymous reader who, along with the three editors, read and commented on all papers. Finally, we are grateful to the School of Languages, Linguistics and Cultures of the University of Manchester for a contribution to the costs of preparing this manuscript. We hope that this volume will contribute to further debate and discussion of the notion of linguistic areas.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon, Yaron Matras and Nigel Vincent xix
xx
Introduction
Campbell, L. (1998) Historical Linguistics: An Introduction (Edinburgh: Edinburgh University Press). (1999 American rights edition, Cambridge, Mass.: MIT Press.) Heine, B. and T. Kuteva (2005) Language Contact and Grammatical Change, Cambridge Approaches to Language Contact, 3 (Cambridge: Cambridge University). Thomason, S. G. (2001) Language Contact: An Introduction (Edinburgh: Edinburgh University Press). Trask, L. (1996) Historical Linguistics (London: Arnold).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
References
1 Lyle Campbell
‘It is not down in any map; true places never are.’ (Herman Melville, Moby Dick)
1
Introduction
The goal of this chapter is to re-examine areal linguistics and in doing so to arrive at a clearer understanding of the notion of ‘linguistic area’. The conclusion reached is that it is individual historical events of diffusion that count, not the post hoc attempts to impose geographical order on varied conglomerations of these borrowings. It is generally acknowledged that linguistic areas are ‘notoriously messy’, ‘notoriously fuzzy’ things (Thomason and Kaufman, 1988: 95; Tosco, 2000: 332; Heine and Kuteva, 2001: 396), and that ‘what we understand about linguistic areas is depressingly meager’ (Thomason, 2001: 99). I argue that the reason for this is because there is no meaningful distinction between borrowing and areal linguistics. Since we understand a great deal about borrowing, we do, as a consequence, understand linguistic areas – or better stated, we understand their limitations. The review (in Sections 2 and 3) of proposed definitions of ‘linguistic area’ reveals the difficulties.
2
Definitions of ‘linguistic area’
A common perception is that the term ‘linguistic area’ is difficult to define (see Heine and Kuteva, 2001: 409). As Thomason (2001: 99) observes, ‘linguistics has struggled to define the concept [linguistic area] ever since [Trubetzkoy, 1928], mainly because it isn’t always easy to decide whether a particular region constitutes a linguistic area or not’. In spite of prolonged efforts to define ‘linguistic area’, there is no general agreement as to its 1 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Areal Linguistics: A Closer Scrutiny
Areal Linguistics: A Closer Scrutiny
definition, and even for the most widely accepted linguistic areas, such as the Balkans, scholars do not agree wholly on which languages belong to the area, which linguistic traits characterize the area, and even its precise geographical extent. In Stolz’s (2002: 260) words, ‘these terms [Sprachbund, linguistic area, and areal type] seem to invite as many meanings and readings as there are linguistic minds to contemplate them’. I argue that too much effort has been wasted on trying to define the concept, that little progress has been made, and that it would be more productive just to investigate the facts of linguistic diffusion without the concern for defining linguistic areas. The principal definitions that have been given of the notion ‘linguistic area’ (or of related and more or less synonymous terms, Sprachbund, diffusion area, convergence area and so on) follow, presented in chronological order. These various definitions both differ from one other and support my conclusion that linguistic areas boil down merely to a study of local linguistic borrowing and its history, and little else.1 I list with the definitions the key elements in them that relate to the questions (discussed in Section 3) of how linguistic areas are defined. 2.1 Early efforts Areal linguistic-like notions have long existed. There were discussions of borrowed structural traits and disputes about ‘mixed languages’ from the beginning of comparative linguistics (see Girard, 1747; Schleicher, 1850: 143; Ebel 1856 [cited by Kuhn, 1861: 75 and Haarmann, 1976: 20]; Müller, 1861: 90; Schuchart, 1866–68; Whitney, 1868: 197 and 1979[1875]: 119; Schmidt, 1872; Powell 1891: 216–77; Meillet, 1921[1921]: 82; 1967: 102; Bloomfield, 1933: 468; Weinreich, 1953; Vendryes, 1968: 308, 319–20). Before explicit definitions of ‘linguistic area’ were sought, numerous traits were identified of the Balkan area (Kopitar, 1829 and 1857; Schleicher, 1850; Miklosich, 1861; Sandfeld, 1902, 1912, 1930, 1934 and 1938; for others see Schaller, 1975: 37–48) and the South Asian (or Indian) linguistic area (Konow, 1906; Bloch, 1919, 1925, 1930 and 1934; Vendryes 1968: 305). The origins of modern areal linguistics are traced to Franz Boas’s (1917, 1920, 1929) work with American Indian languages. Boas identified examples of shared structural traits which did not seem to fit the genetic classifications (see Boas, 1920: 211). He spoke of ‘acculturation’ and ‘absorption’, and raised the question of the difficulty in some instances of distinguishing what was inherited from what was diffused. Boas’ ‘areal-typological’ approach was influential (see Campbell, 1997a: 62–6); he compared the structural traits of languages in a particular region with their neighbours to determine whether they might be due to diffusion or be inherited, representing genetic relationships. Boas’s thinking influenced the Prague school (Trubetzkoy, 1939; Jakobson, 1931, 1938: 354, 1944; Darnell and Sherzer, 1971; Campbell and Mithun, 1979; Emeneau, 1956: 107).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
2
Lyle Campbell 3
2.2 Definitions
It happens that several languages in a region defined in terms of geography and cultural history acquire features of a particular congruence, irrespective of whether this congruence is determined by common origin or only by a prolonged proximity in time and parallel development. We propose the term language union (jazykovoj sojuz) for such groups which are not based on the genetic principle. (Trubetzkoy, 1923: 116, quoted in Toman, 1995: 204) Key elements are geographical region; and any shared features (whether from common origin or parallel development). Not stated: anything about borrowing or diffusion (though perhaps implied in ‘prolonged proximity’). Trubetzkoy (1928) is better known; his ‘Proposition 16’ in the first International Congress of Linguists is generally cited as the origin of the concept. He spoke of the need for the notion ‘language group’ [Sprachgruppe] – a collection of languages bound to one another by a number of systematic agreements (Trubetzkoy, 1928: 18); he divided Sprachgruppe into two types, families of genetically related languages and Sprachbünde. The latter were defined as: Groups composed of languages which show a high degree of similarity with respect to syntax, a similarity in the principles of morphological construction, and which offer a large number of common culture words, sometimes also an outward similarity in the phonological inventories, – but which possess neither systematic sound correspondences, nor has any correspondences in the phonological make up of the morphological units nor any common basic lexical items – such languages groups we call Sprachbünde. (Trubetzkoy, 1928: 18)2 Key elements are a ‘language group’ with syntactic, morphological and often phonological similarities, lacking systematic sound correspondences, no common basic vocabulary. Not mentioned: borrowing, possibly inferred from ‘lacking systematic sound correspondences, and no common basic vocabulary’. There is nothing especially ‘areal’ about this view, apart from the fact that languages that share such traits tend, by inference, to be near one another. The name ‘linguistic area’ in English comes from Velten’s (1943) translation of Sprachbund (literally ‘language union’), made widely known by Emeneau (1956).3 Trubetzkoy (1931: 233–4; 1931: 350–1), in a paper largely about dialect geography, associated phonological areal traits with isoglosses of dialect geography, but which extend beyond language boundaries into other languages. It has
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
It has been argued that the concept of linguistic area or Sprachbund was first presented in Trubetzkoy (1923), as:
Areal Linguistics: A Closer Scrutiny
become common to associate areal traits with dialect isoglosses (Jakobson, 1931, 1938; Weinreich, 1953; Jacobs, 1954; Martinet, 1956; Emeneau, 1980[1956]: 111, 1971, 1974 and 1980; Ramanujan and Masica, 1969; Katz, 1975; Masica, 1976, 1992: 111). Trubetzkoy credited Jakobson (see below) for examples and pointed out others of his own (for example, glottalized consonants in the Caucasus region shared by North and South Caucasian, Indo-European and Turkic languages, Trubetzkoy, 1931, p. 233). Jakobson (1931) seconded Trubetzkoy’s ‘Proposition 16’, though mentioning ‘die Fragen nach gemeinsamen Erschein gungen . . . die in der Struktur benachbarter Sprachen vorkommen und nicht durch gemeinsamen Ursprung bedingt sind’ (Jakobson, 1931: 234). Key elements are structural traits in common, neighbouring languages, not from a common origin. Here, more so than in Trubetzkoy’s renditions, it is clear that the structures in question are shared by ‘neighbouring languages’ and are not a result of a common origin. Jakobson’s main example was the ‘Polytonie’ (tonal contrasts) of the ‘Baltic Sprachbund’ (see Schaller, 1975: 53; Koptjevskaja-Tamm, 2002: 210). As is often pointed out, the Baltic area was thus postulated on the basis of a single defining trait, a matter to which we shall return.4 Becker’s (1948: 5) characterization of a Sprachbund differs from almost all others, seemingly calling for structural convergence among the languages involved: Under a Sprachbund we understand a group of languages which through common fate in the same culture area and through reciprocal influence have approximated one another so strongly that in any of them roughly the same thing can be said in roughly the same way.5 Key elements are a group of languages, in the same cultural area, mutual influence, converge to say the same thing in the same way. Emeneau (1956) brought areal linguistics back to the attention of scholars, particularly in America, where it had largely been abandoned because of Sapir’s view that grammatical traits are rarely borrowed, having gained precedence over Boas’s. Emeneau spoke of ‘diffusion of linguistic traits across genetic boundaries’ (1956: 105). His definition is: This term ‘linguistic area’ may be defined as meaning an area which includes languages belonging to more than one family but showing traits in common which are found not to belong to the other members of (at least) one of the families. (Emeneau 1956: 124) Key elements are: area, two or more language families, traits not found in other members of the same family. Sherzer’s (1973: 760) definition is often cited: A linguistic area is defined here as an area in which several linguistic traits are shared by the languages of the area and furthermore, there is
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4
evidence (linguistic and non-linguistic) that contact between the speakers of the languages contributed to the spread and/or retention of these traits and thereby to a certain degree of linguistic uniformity within the area. It is important to remember that languages which are unrelated or distantly related may very well and probably do disagree with regard to many traits and yet still in the same linguistic area [sic] according to the above definition, since they share several traits (which one might want to call diagnostic traits). What is significant, then, is that linguistic structure, usually impervious to influences coming from outside its own internal mechanisms, has been affected by linguistic contact. Key elements are several linguistic traits, a geographical area, contactinduced spread of structural traits. Katz’s (1975: 16) definition is precise: One can speak of a Sprachbund if: (a) (b) (c) (d)
at a given time a continuous geographical region, that is intersected by at least one language boundary, is encompassed by at least one isogloss.6
Key elements are geographical region, at least one language boundary, at least one isogloss. Note that by this definition a single trait crossing a single language boundary can be sufficient to define a linguistic area. Schaller’s (1975: 58) definition contrasts with Katz’s: With respect to the members of a Sprachbund, it is concerned with at least a part of the languages, that do not belong to a single language, that are geographically neighboring and, because of mutual influence, show a series of common traits which relate the languages found in phonological, morphological or syntactic domains. A Sprachbund shows at least two common traits which extend to least three languages not belonging to the same family, excluding genetically determined origin or unilateral influence in the range of definition of the Sprachbund.7 Key elements are geographical neighbouring languages, not just a single family, several shared traits (phonological, morphological or syntactic) because of mutual influence; at least two common traits which extend to least three languages not belonging to the same family. Bright and Sherzer (1978: 228) differ slightly but significantly from Sherzer (1973) in that they specify more clearly that borrowing is behind
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell 5
6
Areal Linguistics: A Closer Scrutiny
The term ‘linguistic area’ generally refers to a geographical area in which, due to borrowing, languages of different genetic origins have come to share certain borrowed features – not only vocabulary . . . but also elements of phonological, grammatical, or syntactic structure, which are less liable to be diffused in this way. Key elements are geographical area, languages of different families, shared borrowed traits. In Campbell (1985: 25), I presented what I took to be the common understanding of the notion and added my own view: Areal linguistics, as broadly conceived, deals with the results of diffusion of structural features across linguistic boundaries. As commonly viewed, linguistic areas are characterized by a number of linguistic features shared by various languages (some of which are unrelated or are from different subgroups within a family) in a geographically contiguous area . . . linguistic diffusion and AL [areal linguistics] are to be equated and cannot profitably be separated; i.e. I will argue that there is no sharp boundary between the two, that all areal linguistic phenomena involve diffusion and all structural diffusion involving more than two languages is areal. Key elements are structural diffusion, more than two languages; commonly held to include also a number of shared features, various not closely related languages, geographical area. I emphasized the lack of any significant boundary between individual acts of borrowing and areal linguistics in general. Today, I would amend this definition to abandon the ‘more than two languages’ requirement and would emphasize more fully that any structural borrowing is areal in nature (see Campbell, 1994: 1471). Thomason’s (2001: 99) definition is: A linguistic area is a geographical region containing a group of three or more languages that share some structural features as a result of contact rather than as a result of accident or inheritance from a common ancestor. Key elements are geographical area, three or more languages, shared structural features, from contact (not as a result of accident or inheritance). Aikhenvald and Dixon (2001: 11) offer their view that: A linguistic area (or Sprachbund) is generally taken to be a geographically delimited area including languages from two or more language families,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
the shared traits, and that different language families should be involved:
Lyle Campbell 7
Key elements are geographical area, different language families, fair number of shared distinctive traits. In Campbell (2002: 729) I attempted to give a definition that would represent the field generally (but not necessarily include my own doubts): A linguistic area is a geographical area in which, due to language contact and borrowing, languages of a region come to share certain structural features . . . Central to a linguistic area [are] . . . structural similarities shared among languages of a geographical area (where usually some of the languages are unrelated or at least no all close relatives). It is assumed that the reason the languages of the area share these traits is because they have borrowed from one another. (See Campbell et al., 1986: 530 for a similar definition.) Key elements are geographical area, shared structural features, languages not closely related, borrowing. For some other definitions, not appreciably different from those listed here, see Voegelin (1945, 1961); Wolff (1959); Zeps (1962); Birnbaum (1965: 12); Seidel (1965); Décsy (1973: 29); Aoki (1975); Holt and Bright (1976); Haas (1978); Hill (1978); Lehiste (1988: 59–61); Campbell (1994: 1471; 1996a; 1997b, 1997c); Matthews (1997: 351); Trask (2000: 196–7); and Ramat (2002).
3
Questions about criteria
I turn now to a number of important questions about the criteria deemed necessary, or at least useful, for establishing linguistic areas which emerge above from the survey of definitions. 3.1 Number of languages The question of the number of languages required to constitute a linguistic area frequently comes up. The most common answer assumes that several languages are needed, and that in every case there should be three or more. Thomason (2001: 99) says, ‘the reasons for requiring three or more languages is that calling two-language contact situations linguistic areas would trivialize the notion of a linguistic area, which would then include all of the world’s contact situations’ (see Schaller, 1975: 54, 58). However, there is no inherent linguistic, geographical or other tangible difference between a situation with only two languages which borrow from one another, and one with three or more. If there is no significant difference between borrowing in general and areal linguistics in particular, as I argue, then the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
sharing significant traits (which are not found in languages from these families spoken outside the area). There must be a fair number of common traits and they should be reasonably distinctive.
Areal Linguistics: A Closer Scrutiny
requirement that there must be three or more languages to constitute a legitimate linguistic area is superfluous. For this to make sense, it would be necessary to show that there is a significant difference between diffusion involving two languages and diffusion among three or more, but there is no such difference. The kinds of changes that take place and the mechanisms by which they happen are the same whether two, or more than two, languages are involved (see Katz, 1975: 16). 3.2 Number of language families Some require that two or more language families be involved to define a linguistic area – see Emeneau 1956: 124; 1965: 127; 1978: 1; also Schaller, 1975: 58; van der Auwera, 1998a: 260; Tosco, 2000; Aikhenvald and Dixon, 2001: 11). However, this is at best a desideratum, not a requirement, since the Balkan Linguistic Area, universally accepted, has only Indo-European languages among its members (though some include Turkish) (see Masica, 1992: 110). Some scholars argue that at least some of the languages of an area, even if members of the same family, should not be closely related (see Campbell, 1985: 25; 1994: 1471; Matthews, 1997: 351; Stolz, 2002: 261). In any event, for most scholars, a number of structural traits borrowed across unrelated or only slightly related languages would be considered as stronger evidence of a linguistic area, though the idea of some minimum level of required linguistic diversity among the languages of a linguistic area has not really been a focus of attention. Any attempt to establish one would surely turn out to be arbitrary.8 If we focus on the facts of linguistic diffusion instead of seeking some diagnostic minimum amount of genetic distance as being necessary for the definition of linguistic area, the question of the number of language families needed disappears. 3.3 Number of traits A common question is, ‘How many features must be shared before a linguistically diverse region can reasonably be called a linguistic area?’ (Thomason, 2001: 100; see also Campbell, 2002: 732; Koptjevskaja-Tamm, 2002: 211; Stolz, 2002). Two competing answers have frequently been given: (1) one trait is enough; (2) several traits are necessary. However, again, if there is no significant cut-off between borrowing in general and areal linguistics in particular, then the requirement that there must be several shared traits to constitute a linguistic area becomes superfluous. Any attempt to impose a lower limit proves arbitrary and unjustified by anything other than terminological convenience. Let us examine more closely some views in favour of each of these positions. 3.3.1 The single-trait view Though less common, the notion that a legitimate linguistic area might be defined on as few as one shared trait has had a number of proponents (see
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
8
Jakobson, 1931: 139; Trubetzkoy, 1931: 345; Weinreich, 1953: 378–9; Winter, 1973: 140; Katz, 1975: 16; Masica, 1976: 172; Bright and Sherzer, 1978: 236; Campbell, 1985: 29, 48; Dryer, 1989: 266; Hickey, 1999: 36; Koptjevskaja-Tamm and Wälchli, 2001: 624). As mentioned above, Jakobson (1931) relied on a single trait, ‘Polytonie’, to define the Baltic Sprachbund (see Koptjevskaja-Tamm and Wälchli, 2001: 640–6). Katz (1975: 16) also made the single-trait minimum requirement diagnostic in his definition. In connection with the argument for the sufficiency of a single trait, I proposed: In principle there is no meaningful way of distinguishing LAs [linguistic areas] defined on the basis of several features from those based on but a single shared trait. Nevertheless, the question can be posed, not in the form, does or does not some entity qualify as a LA?, but rather as, how strong or weak is a particular LA? (Campbell, 1985: 29) Single-trait linguistic areas were to be considered the very weakest (see also Masica (1976: 172, 1992: 111). Stolz (2002: 262) adds to this: Specialists with a background in quantitative linguistics have demonstrated convincingly that, no matter how hard you try, there is simply no way to identify a universally valid statistical minimum of similarities necessary for the constitution of a linguistic area except through the absolutely arbitrary decisions of the linguists themselves. 3.3.2 The several-traits view For many scholars, the idea that a linguistic area should exhibit a number of shared traits, ‘isoglosses’, was their principal and in some cases only diagnostic criterion for defining linguistic areas (see Jakobson, 1931: 139; Trubetzkoy, 1931: 345; Becker, 1948; Wagner, 1964; Birnbaum, 1965; Henderson, 1965; Sherzer, 1973, 1976; Schaller, 1975: 54; Bright and Sherzer, 1978: 233; Emeneau, 1978: 1; Sarhimaa, 1991; Aikhenvald and Dixon, 2001: 11, among others).9 Thomason (2001: 101) argues explicitly against the sufficiency of a single trait for defining a linguistic area: ‘if a contact situation is intense enough to produce one shared structural interference feature in a group of geographically close languages, it is intense enough to produce other shared features too’. But this is not really the case. Masica (1992: 111) points out that ‘often a single trait appears to diffuse over a wide area, without other features necessarily being affected’. When diffusion begins within an area, it is logically possible – and probable – that some single trait will be the first to spread; until others develop, it will be the only trait shared in this way. That is, a contact situation may be precisely only intense enough to have produced (so far) a single ‘structural interference feature’ – that is how a linguistic area might start to develop.10
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell 9
Areal Linguistics: A Closer Scrutiny
Thomason (2001: 101) acknowledges that the answer to the question of how many features are needed ‘has to be a judgment call’, and that some areas are considered stronger or weaker, based on how many features they share. However, she asserts that ‘the absence of a clear dividing line between a region that is a linguistic area and a region that is not does not justify adopting the historically implausible position that one shared feature is in principle enough’. Nothwithstanding, it does not follow that anything ‘historically implausible’ is at stake. Rather, there is nothing historically implausible about a single borrowed trait being shared by some languages. Since there is no legitimate boundary between borrowing and areal linguistics, the nagging question of how many traits are required is answered: a single trait is sufficient, albeit any linguistic area so designated would be a very weak one. 3.4 A question of boundaries A major question is, how does one establish the boundaries of a linguistic area – in particular, do areal isoglosses need to bundle? Opinions fall on both sides of the issue, though more often on the side of non-bundling. Emeneau (1978: 2) at times seemed to favour bundling: ‘once several features have been established as having the same boundaries, so that there is an approximation of a “bundling of isoglosses”, the linguistic area can be considered to be typologically established’. Aikhenvald and Dixon (1998: 244) assert that ‘the distribution of each “bundle” of areal features . . . is crucial for determining the boundaries of linguistic areas’, (see Henderson, 1965: 140; Winter, 1973: 140; Haarmann 1976: 24; Masica, 1976: 6, 170, 179; Campbell, 1985: 28). Many, however, disagree. For example, Emeneau (1965: 128) also noted that ‘prima facie, one might expect that a linguistic area in the present sense might be delimited in the same way [by ‘thick bundles of isoglosses’]. Unfortunately, I know of no demonstration of such a bundling of isoglosses’. ‘In linguistic area studies it is doubtful if there will every emerge isoglossbundles’ (Emeneau 1965: 136) (see also Trubetzkoy, 1931: 345; Jakobson, 1944: 193; Becker, 1948: 23; Emeneau 1956: 120; Henderson, 1965: 431; Ramanujan and Masica, 1969: 550; Sherzer, 1973: 132–3; Winter, 1973: 140; Katz, 1975: 12, 16; 1992: 111; Haarmann, 1976: 24; Holt and Bright, 1976; Masica, 1976: 5, 179–80; Campbell, 1985: 27–8; Wintschalek, 1993: 6–7; Dimmendaal, 2001: 387; Koptjevskaja-Tamm and Wälchli, 2001: 624, 728; Campbell, 2002: 732; Koptjevskaja-Tamm, 2002: 215; Stolz, 2002: 264). Matisoff’s (2001: 300) question, ‘does not every “linguistic area” arise from an accumulation of individual cases of “localized diffusion”?’, implies non-bundling (see Koptjevskaja-Tamm, 2002). Clearly, if we shift the focus to individual localized borrowing events, regardless of how they come to be distributed, the issue of isogloss bundling to define the borders of a linguistic area becomes irrelevant. Whatever the distributions of borrowed
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
10
Lyle Campbell
11
traits, the areas they cover become clear as we unravel the history of the changes related to borrowing in the languages involved.
Many see areal linguistics as being akin to dialect geography, where often the isoglosses do not bundle at borders, but rather are more intensely concentrated around some core zone. In Thomason’s (2001: 101) words: In the majority of cases, the boundaries are fuzzy; often there is a central group of languages that share a large proportion of the characteristic features, and scattered peripheral languages or groups that share a considerably smaller number of the features. Frequently, too, there are shared features that are found only in a small subset of the area’s languages. Thomason and Kaufman, 1988: 96–7; Tosco, 2000; Compare Dahl, 2001: 1458. Often, several isoglosses radiate outwards from a centre of influence, resulting in a pattern of more shared traits at the core, with fewer shared features as one moves towards the periphery; some extend further from the core, others extend for less of a distance. In this vein, Becker (1948: 23) spoke of ‘Kernsprachen’ (core, nuclear languages) and ‘Randsprachen’ (peripheral languages) in a Sprachbund. As Thomason (2001: 104) says: it must be acknowledged that deciding where the boundaries are can be a difficult task, that some cases will be truly indeterminate, and that . . . an ancient linguistic area can be overlaid by a more recently emerged linguistic area with different boundaries . . . overlapping and interlocking isoglosses. It has not gone unnoticed that this core–periphery pattern (and the lack of isogloss bundling generally) creates problems with attempts to define linguistic areas. Dahl sees in this core–periphery pattern a major challenge to areal linguistics: An area with the typical center–periphery structure will often exhibit the greatest genetic diversity in the peripheral parts . . . This means that an areal sample that aims at genetic representativeness will over-represent those parts and under-represents the more homogeneous center. It also means that minor adjustments in the way the borders of an area are defined may have rather dramatic consequences for such a sample . . . This again raises the question about the reality of linguistic areas (Dahl, 2001: 1463) (Emphasis added)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
3.5 Core versus periphery
12
Areal Linguistics: A Closer Scrutiny
This variation [fuzzy boundaries, non-bundling] is hardly surprising, since the way a linguistic area arises is through contact-induced changes that occur over a long period of time and spread widely through the region – but always from language to language in a series of events, not in some single mystical area-wide process that affects many languages at once. With the focus on the history of diffusion and not on defining the boundaries, there ceases to be a problem. This lends support to my conclusion that defining the areas is of little importance, and it is the history of diffusion that counts. Indeed, it is the individual borrowing events involving specific individual instances of language contact that produce these isogloss patternings, and the investigation of the history of these individual borrowings should be our primary concern. As Masica (1992: 110) says, ‘the real locus of language “contact” is the mind of the bilingual individual’, echoed also in Giannini and Scaglione’s (2002: 152) words, ‘the process . . . is . . . borrowing, and it is the bilingual speaker who is the material agent in this process.’ In this context, the question of numbers arises again. Consider a situation in which two particular languages in a linguistic area share significantly more traits between the two of them than do any other languages of the area. Do these two, then, constitute the ‘core’ of some particular area? Or, is this to be considered just a case of local borrowing that is not ‘areal’, under the assumption that two languages alone is not enough to constitute an area, and that there must be three or more? Put differently, if all borrowings, hence all areal phenomena, ultimately come down to individual local borrowings, how could it be possible to rule out a situation where only two languages are engaged in borrowing? And, if all known linguistic areas are just the aftermath and build-up of instances of such individual borrowings among pairs of languages, why should we attempt to adorn the concept of ‘linguistic area’ with anything beyond a simple accumulation of individual borrowings that result from their own individual contingent histories, a history of borrowings? In this view, a core where many traits affect only two languages would be as ‘areal’ as a core in which more languages were involved. 3.6 Different kinds of ‘linguistic area’? It is generally recognized that what have been called linguistic areas include things that have widely divergent characters and historical backgrounds, depending on the social, cultural, political, geographical, attitudinal and other factors that correlate with diffusion of linguistic features
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
The ‘core’ and fragmentary periphery are artefacts of the history of borrowings, nothing more. Thomason’s (2001: 102) take on this appears to support this conclusion:
13
in different regions (see Kuteva, 1998: 308–9; Aikhenvald and Dixon, 2001: 11, 13; Dahl, 2001: 1458). As Thomason (2001: 104) explains, ‘[linguistic areas] arise in any of several ways – through social networks established by such interactions as trade and exogamy, through the shift by indigenous peoples in a region to the language(s) of invaders, through repeated instances of movement by small groups to different places within the area’. One finds mentioned in the literature such different sorts of linguistic areas as: incipient ones, only beginning to form and with as yet few shared traits; moribund and decaying ones, where as a result of many changes after the area was actively formed, fewer traits are currently recognizable among the languages; layered ones (with new layers and old layers), and overlapping ones, where different areas formed on top of one another or overlapping one another at different times for different reasons; multilateral versus unilateral areas;11 areas resulting from rapid conquest, to population spread and migration (traits moving with movement of speakers), others through home-grown, stayin-place contact (movement of traits but not of peoples); disrupted areas with ‘latecomers, earlier drop-outs, and temporary passers-by’ (Stolz, 2002: 265) and so on.12 ‘In short, “linguistic area” is not a uniform phenomenon, either socially or linguistically’ (Thomason, 2001: 115). This array of different kinds of linguistic area raises questions about whether the notion of ‘linguistic area’ is justified. Do all these different ‘objects’ legitimately qualify as ‘linguistic areas’, given their very different natures and composition, and given the very different circumstances of their formation (and decay)? The notion of a ‘linguistic area’ offers little on which these different sorts of linguistic areas can be united, other than the fact that they all involve borrowing in some way, but borrowings of different sorts, for different reasons, in different settings and at different times. Thus Dahl asks: In the end, we are led to the following more far-going question about the notion of area: to what extent do areas . . . have a reality of their own and to what extent are they just convenient ways of summarizing certain phenomena? At the most basic level, linguistic contact relationships are binary: one language influences another. An area is then simply the sum of many such binary relationships. (Dahl, 2001: 1458) A linguistic area, to the extent that it may have a legitimate existence at all, is merely the sum of borrowings in individual languages in contact situations. If we abandon the search for an adequate definition of this concept and focus rather on understanding borrowings, those contingent historical events, the difficulty of determining what qualifies as a legitimate linguistic area ceases to be a problem.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
14
Areal Linguistics: A Closer Scrutiny
There is a distinction between what I called ‘historicist’ and ‘circumstantialist’ approaches to areal linguistics (Campbell, 1985, 1997a: 330–1; Campbell et al., 1986). The historicists call for historical evidence that the traits used to define linguistic areas really were borrowed, while circumstantialists tend to amass a number of shared traits among the languages of a region and allow the circumstances to imply the probability of diffusion, but do not require proof of this. A number of scholars hold that only linguistic areas supported by a significant amount of historical evidence documenting the diffusion of the traits in question are fully legitimate (see Dahl, 2001: 1457; Thomason, 2001: 102–3; several chapters in Ramat and Stolz, 2002). It will be noticed, however, that such a requirement, of actual historical evidence of diffusion, highlights the lack of distinction between areal linguistics and borrowing in general. It supports my argument that structural borrowings deserve attention first and foremost, and that linguistic areas are afterthe-fact constructs based on the residue and accumulation of borrowed traits, regardless of how and when they came to be shared among the languages involved. 3.8 What about ‘geography’? Is ‘geography’ required in areal linguistics? It would seem that the answer should obviously be ‘yes’, but in fact not everyone agrees – the answer to the question is more complicated than it seems. 3.8.1 ‘Geography’ and ‘linguistic areas’ It is generally assumed that a linguistic area must be ‘a geographically delimited area’ (Aikhenvald and Dixon, 2001: 11), though some definitions seem to concentrate only on the sharing of features and ignore region. Thomason (2001: 99) explains that ‘the reason for specifying a single geographical region [in the definition of ‘linguistic area’] is obvious: no direct contact among speakers, no linguistic area’. Nevertheless, language contact does not inescapably require geography. As Dahl (2001: 1460) points out, ‘the whole notion of “areal phenomena” is built on the convenient fiction that each language has a specific location in space, that no more than one language is spoken in each place, and that language contact takes place between adjacent languages. However, language contacts typically occur in densely populated places where speakers of many languages live together and bi- or multilingualism is common. In addition, many languages have a widely scattered distribution.’ Dahl (2001: 1458) argues against the ‘many current definitions which make [geographical] contiguity a necessary condition’. Some contact is not infact ‘geographical’ but could be said to be vertical, as in the wellknown instances of influence from older, ‘dead’ ancestor languages (as in
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
3.7 Is diachronic evidence required?
15
the case of the significant impact of Latin on French and Spanish, or of Sanskrit on modern Indic languages). Contact can be vertical, or at least not geographical, in another sense also, where two or more languages are spoken in precisely the same location. Thomason’s (2001: 1) definition suggests this: ‘language contact is the use of more than one language in the same place at the same time’ (emphasis added). That is, by implication, that language contact does not necessarily involve languages in different but adjacent regions; both languages can occupy the same location.13 As some see it, the primary venue for the transfer of features from one language to another is inside a single head – that of the bilingual – making geography unimportant. On the other hand, some language contact, even when geography is involved, is not limited just to adjacent neighbours, but can be longdistance, through trade and travel, conquest and migration.14 For example, we can point to the well-known pattern of the spread of innovations through long-distance contact and linguistic diffusion, first among major cities at some distance from one another, with the innovations reaching the rural areas later, and less distant from the urban elite who changed first – the spread of uvular ‘r’ in several European languages is a ready example of this pattern (see Trudgill, 1983: 56–9). There are also cases, such as the influence of Latin on the structure of various European languages, English in particular, with which it shared no geographical boundary (compare the influence on Sicilian, not from neighbouring Italian, but from languages further afield). Stolz (2002: 265–6) highlights the geography problem in the emigration of speech communities: The emigrants take their language and with it the areally defined features to a place that is located outside the original linguistic area. Does this mean that the language is no longer a member of the linguistic area/ Sprachbund? Or rather does this mean that the linguistic area/Sprachbund automatically expands through migration and may include languages which strictly speaking, are not co-territorial with the rest of the members of the linguistics area? If so, the notion of area would lose its geographical implications because discontinuous, non-contiguous constellations would count as areas as well. The considerations in this section indicate that geographical proximity is not absolutely necessary for borrowing of structural traits across languages. If the limiting case for having a linguistic area is structural borrowing across language boundaries, then in some instances it may well be that such borrowing, while not common, takes place among nonadjacent or even distant languages. What is crucial is the contact, not the geography.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
16
Areal Linguistics: A Closer Scrutiny
The relevance of the preceding discussion about geography and borrowing becomes more apparent when considered in light of the claim made here that it is the diffusion that is of prime importance, and that the geographical aspect of putative ‘linguistic areas’ is derivative. The shared linguistic traits are not brought into existence by, nor somehow explained by, the geographical region, in spite of the fact that the notion of ‘linguistic area’ is often presented, at least implicitly, as some entity where the geography is prime and the linguistic traits themselves are just reflections of some sort of vague geographical determinism. There is no geographical determinism; the linguistic borrowings are prime, and the geographical areas are only a reflection of these, with no significant causal force of their own. Koptjevskaja-Tamm’s (2002: 209) reading of the Baltic linguistic area leads her essentially to the same conclusion: Intensive micro-contacts superimposed on each other sometimes create the impression of an overall macro-contact among the languages in an area, which has not necessarily been there. Therefore the notion of Sprachbund is not satisfactory for describing the linguistic situation in the CB [Circum-Baltic] area. I do not wish to imply that the geographical patterns that can arise as byproducts of the borrowings, those contingent historical events, cannot contribute to historical understanding – they can and do. However, it is necessary to combat the notion that the geography is prime and the borrowings are in some way secondary to and determined by the geography. There is nothing about the geography itself that forces the linguistic behaviour, that in some way causes languages of a region to become more alike. Rather, there are simply a number of individual events of diffusion involving in the main local dyads of languages (in fact, of speakers of the languages); some traits once borrowed may then be borrowed further in other dyadic interactions involving other languages, in this way giving the trait a larger geographic trajectory. Koptjevskaja-Tamm’s (2002: 219) finding for the Circum-Baltic languages is true generally: ‘Convergence that comprises more than two or three languages, it seems, is always the result of the overlapping and superposition of different language contacts’; ‘intensive micro-contacts superimposed on each other sometimes create an impression of an overall macro-contact among the language, which has not necessarily been there’. 3.9 Other questions There are a number of other issues involved in attempts to define linguistic areas, which come up with some frequency in the literature, but which I do not discuss here, in the interest of space, though they deserve mention. Some follow (see Stolz, 2002: 263–4 for others).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
3.8.2 Geographical (non-)determinism
17
(1) The nature of the areal traits For traits to qualify as areal or to carry much weight in defining an area it has been proposed that they should be ‘reasonably distinctive’, marked, and not too natural, unique or unusual in the region, low on hierarchies of borrowability (that is, hard to borrow), or typologically not commonplace. (2) Trait weights Some borrowed traits are deemed to carry more weight for defining a linguistic area than others (in particular typologically complex features, traits that are more difficult to borrow), and various scales, hierarchies and rankings have been proposed to account for the different roles that different kinds of diffused traits are assumed to play. Weight, borrowability and some notion of typological compatibility between borrowed traits and the existing structures of the borrowing language are often mentioned. (For a discussion, see Katz, 1975; Heath, 1978: 104–7; Campbell, 1985, 1996a, 1996b, 2002: 732; Campbell et al., 1986: 535–6; van der Auwera, 1998a; Aikhenvald and Dixon, 2001; Curnow, 2001; Haig, 2001: 218–22; Giannini and Scaglione, 2002; Stolz, 2002: 264–5 and so on). In particular, the fact that different traits count differently for defining a linguistic area because of their different weights/ranks makes isopleth maps (such as van der Auwera’s (1998a)) less useful and less revealing. The isopleth marked on such maps represents languages sharing the same number of traits, but this is misleading. It appears to give as much areal credit to languages sharing features that diffuse easily as it does to languages sharing traits that are much more difficult to borrow, so long as the numbers (though not the weights) are relatively similar. (3) Age The matter of the time depth involved has impinged on some scholars’ notions of how linguistic areas are defined (see Aikhenvald and Dixon 2001: 12). Some believe considerable time is required for a linguistic area to emerge – on the order of millennia – while for others it may be as little as a couple of hundred years (see Haspelmath, 1998; Aikhenvald and Dixon, 2001: 9–10, 13; Thomason, 2001: 102; Watkins, 2001: 49, 55). (4) Parallel innovation, accidental similarity How can traits that are the result of independent but parallel development be distinguished from traits that involve areal diffusion? How can we distinguish traits found in an area that are only accidentally similar from those that have diffused? (Several of the chapters in Ramat and Stolz (2002), for example, grapple with these questions.)
3.10 Conclusion on definitions This survey leads to a similar conclusion to that made by Stolz (2002: 259) – that ‘the search for clearcut definitions [of ‘Sprachbund, linguistics area and areal type’] has been largely futile and will probably never come to a really
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
satisfying conclusion’.15 Every ‘linguistic area’, to the extent that the notion has any meaning at all, arises from an accumulation of individual cases of ‘localized diffusion’; it is the investigation of these specific instances of diffusion, and not the pursuit of defining properties for linguistic areas, that will increase our understanding and will explain the historical facts. With the focus rather on specific instances of borrowing, many of the unresolved issues and indeterminacies that have dogged areal linguistics from the outset cease to be relevant questions.
4
Areal excesses and the attack on family trees
While it is not possible to deal with them in detail here, it should be mentioned that areal linguistic notions figure in some recent proposals that see themselves as going beyond basic historical linguistic methods (see Nichols, 1992, 1995, 1997, 1998; Aikhenvald and Dixon, 2001; Dixon, 1997). I have argued that the use of areal-linguistic-like concepts in these works are mistaken and misleading (for details, see Campbell 2003, forthcoming). However, one aspect of this work does bear closer examination here: the recent attacks on the concept of the family tree in linguistics. Some scholars have recently taken a sceptical view of the validity of family tree diagrams, stemming from beliefs about the degree of convergence or confounding possible in language contact situations. It is important to clarify this debate – the scepticism towards family trees is misplaced. Jakobson (1938) offered a solution to the old debate about the possibility of multiple origins for a single language – that is, to the question about the utility of the family tree model in situations of areal diffusion: ‘La similitude de structure ne s’oppose donc pas, mais se superpose a la “parenté originaire” des langues’ (Jakobson, 1938: 353). He called for adequate description of shared traits without premature generalizations about whether they owe their explanation to a genetic relationship, a mixture or to diffusion (Jakobson, 1938: 365). This remains sound advice. Mainstream historical linguists realize that it is not possible to understand diffusion fully without knowing the genetic affiliation of the languages involved, and vice versa, it is not possible to account fully for what is inherited without proper attention to what is diffused. That is, it is not two distinct, opposed and antagonistic points of view that are involved, but rather both are needed and they work in concert: ‘both the comparative method and areal linguistics are historical disciplines – twin faces of diachronic linguistics’ (Hamp, 1977: 27). Both are necessary if we are to answer the question, ‘What happened?’, the historical linguist’s goal. The following are some citations that question the family tree model and favour areal linguistic alternatives:
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
18
Lyle Campbell
19
The original motivation of both [areal linguistics and language typology] was the insufficiency of the genetic Stammbaum model for the study of relationships among languages. (Dahl, 2001: 1456) A family-tree-like diagram does not adequately demonstrate the many kinds of historical and current relationships between [among] languages. (Aikhenvald and Dixon, 2001: 6) There are a number of misconceptions in these citations. The goal of the historical linguist is to determine the history of the languages involved, whether that has to do with inheritance, diffusion or a combination of both. Indeed, both the inherited and the diffused are necessary at the same time. The matter of the burden of proof requires this. To test any hypothesis of genetic inheritance, it is necessary to demonstrate that it fits the facts better than alternative possible explanations, borrowing being principal among alternatives (though accident, universals and others must also be considered). Similarly, for any hypothesis of borrowing, it is necessary to demonstrate that other possible explanations do not provide a better answer, and the possibility of inheritance from a common ancestor is crucial among those that must be eliminated for the hypothesis of diffusion to stand. That is, it is not a question of driving the genetic explanations as far as possible and then (and only then) turning to areal diffusion as a last resort. Rather, it is a matter of seeking the whole history and testing any hypothesis against other possible explanations. In fact, many of the errors and excesses seen today in both proposals of distant genetic relationships and in proposals of diffusion stem from not considering other possible explanations sufficiently before reaching conclusions in particular cases. Some, in the zeal for areal explanations as presumed challenges to the comparative method, call for alternative models and methods: A main thesis of this essay [Dixon, 1997] is that the family tree model, while appropriate and useful in many circumstances, is not applicable everywhere and cannot explain every type of relationship between languages. We need a more inclusive model, which integrates together the ideas of the family tree and of diffusion area. (Dixon, 1997: 28) To reconstruct the history of a language adequately, a model is needed which is significantly more sophisticated than the family tree based on
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Areal linguistics was originally inspired by the insufficiency of genetic relationships as an explanation for similarities between languages, in particular, by the recognition of grammatical and phonological similarities which were due to language contact. (Dahl, 2001: 1457)
20
Areal Linguistics: A Closer Scrutiny
Though these citations suggest otherwise, mainstream historical linguists agree that the family tree is not everything and does not explain all the kinds of historical relationships that can affect languages. They agree – insist – that attention must be paid to diffusion. Historical linguistics has never been limited to only the family tree – borrowing, wave theory, and later areal linguistics, are all taken into account (see Garrett, 1999). Moreover, a consequence of Dixon’s (1997: 11) ‘assumption 4’, that ‘in the normal course of linguistic evolution, each language has a single parent’, is that the family tree model is always relevant, regardless of whether the application of methods to determine the family tree in given instances is complicated by changes – for example, of an areal linguistic nature – that require the use of other historical linguistic techniques for a full understanding. Therefore, most historical linguists would say that we do not need the more inclusive integrative model that Dixon, Aikhenvald and Chappell have called for – we already have one. As Watkins explains, ‘the resilience and the power of the comparative method lies in its sensitivity to similarity due both to genetic filiation and areal diffusion alike. Both are historical models, and the goal of comparison is history’. Hübschmann (1875) demonstrated this ‘when he proved that Armenian was a separate branch of Indo-European, and not a dialect of Iranian as previously thought’ (Watkins, 2001: 59). Armenian exhibits a huge influence from Iranian, but it was the application of the comparative method which revealed this as diffusion and not inheritance (see Campbell and Poser, forthcoming, for details). Scholars claiming that diffusion has called the comparative method into question seem to have lost sight of this. For example, Aikhenvald and Dixon (2001: 4, 6) claim the family tree is merely a bad ‘metaphor’ that others assume to be a reality: The family-tree metaphor [developed for Indo-European] has been taken over for other parts of the world in stark form, often as the sole model for relationships between [among] languages . . . Rather than asking whether a form of family tree is appropriate to the language situation in some newly studied region, it has often been simply assumed that it is. What began as a metaphor has been ascribed reality, and has acted to constrain enquiry along narrow lines. This can lead at best to a partial and at worst to a mistaken statement of language relationships. (Aikhenvald and Dixon, 2001: 6–7)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
the use of the comparative method. It needs to incorporate the diffusion and layering process as well as other language-contact phenomena such as convergence, metatypy and hybridization. The desideratum is a synthesis of all the processes that affect language formation and development. (Chappell, 2001: 354)
21
Aikhenvald and Dixon appear to hold the view that traditional historical linguists believe that a mere diagram, used to reflect linguistic lines of descent, is the whole story, and they do not address the rest. This rhetoric about ‘metaphor’ is misleading, as there are identifiable historical facts – the objective reality that in language families languages can indeed be related to one another as a result of descent from a common ancestor – this is not merely a metaphor. But this was never considered to be the whole story. As Sebeok (1950: 101) makes clear, if some scholars limit their vision to only what is inherited, too bad for them, but this is not an accurate characterization of what historical linguists generally do, nor of the history of the field, as the Armenian case and many others show. The diagram, which attempts to depict the family tree, is just one part of the larger story, and the handbooks on the history of languages always give attention to borrowing. Family trees are not the targets, not the bad guys. It is never a question of diffusion or convergence versus the family tree; rather it is always a question of both. We want to answer the question, ‘What happened?’, and for that we need both inheritance and diffusion. True, there are cases where it is difficult or even impossible to figure out whether shared traits are a result of inheritance, diffusion, independent parallel development or accident. The difficulty or impossibility of distinguishing what is inherited from what is diffused in some cases is readily acknowledged by mainstream historical linguists; however, this is routinely cited by those who wish to place the comparative method or genetic relationships among languages in a bad light (see Aikhenvald, 2001: 190–1; Aikhenvald and Dixon, 2001: 1; Chappell, 2001: 335, 353–4; Dahl, 2001: 1456; LaPolla, 2001). All retrospective sciences are faced with the same problem: we do our best to recover the past from the evidence on hand, which sometimes is insufficiently well preserved to allow clear answers. However, fortunately in linguistics our methods have proved successful repeatedly in distinguishing specific instances of inheritance from borrowing. Because the methods have been successful in so many cases, we do not abandon them just because the extant evidence in some specific instance is insufficient, just as we do not conclude that an vehicle can never take us anywhere just because on one occasion the petrol ran out. It is definitely not a question of exclusive domains – areal versus genetic – but rather of both working in concert to determine the full history.
5
Conclusions
Some of the conclusions in this chapter may appear dreary: (1) we should abandon the search for a definitive definition of ‘linguistic area’; (2) areal linguistics is not distinct from borrowing/diffusion in general; and (3) the concept ‘linguistic area’ is not significant in itself. Instead of pursuing definitions of linguistic areas, we should attempt to account for the history of
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
individual borrowings and diffusion, together with language change in general, in order to answer the question, ‘What happened?’ Still, I believe the overall conclusion is a positive one. There is something liberating and satisfying about being able to abandon the fruitless search for an acceptable definition of ‘linguistic area’ and the attempts to establish specific linguistic areas around the world, and to be able to get on with the task of trying to answer the question, ‘What happened?’ If we succeed in determining what changes have taken place, and how, when and why they took place, we will have succeeded in providing all the information underlying traditional notions of linguistic areas. We will know which changes are a result of borrowing and which to inheritance, and will know their distribution across languages. The geographical patterning to instances of diffusion will be a natural consequence of this fuller historical account, read directly off the accumulation of history changes in the languages involved.
Notes 1 The terms ‘language contact’, ‘borrowing’, ‘diffusion’, ‘interference’ and ‘contactinduced change’ are widely used and generally understood, though some confusion occasionally occurs, and finer-grained definitions have at times been offered (see Thomason and Kaufman, 1988, Thomason, 2001). I use ‘borrowing’ to mean broadly anything that was not formerly in a language but comes to be there because of contact with another language. I use ‘interference’ to mean things that were already native in the language but have come to be modified in some way through contact with another language. I use ‘diffusion’ as a cover term for the borrowing of any sort of linguistic feature. While it is useful, I do not employ Thomason’s (2001: 129) distinction between shift-induced interference with imperfect learning and native speakers borrowing from another language. In a language contact situation, even within a linguistic area, some speakers can be involved in shifting languages (possibly with imperfect learning) while at the same time other speakers are borrowing, as native speakers; both can influence jointly change in the languages involved. Since in many situations it is, on the one hand, difficult to separate out the effects of the two, and on the other, the necessary information for making the distinction is often not available, I do not make use of the distinction here. Thomason (2001: 129) sees the ‘distinction in their typical linguistic results’, which she believes to be ‘less transferred vocabulary and more structure in shift-induced interference vs. more transferred vocabulary and less structure in borrowing’. In many linguistic areas, of course, there will be a certain amount of both. Thus, for example, I do not see, as Thomason (2001: 111) does, that the Ethiopian highlands linguistic area is solely ‘a result of shiftinduced interference from Cushitic speakers who adopted the languages of the more recently arrived Semitic speakers’. Native speakers of Semitic languages probably also accepted the incoming traits that owe their origins to Cushitic languages – that is, acts of borrowing. Thomason (2001: 130) acknowledges that the dichotomy between these two does not fit mechanisms of change very precisely. 2 My translation; original: ‘Gruppen, bestehend aus Sprachen, die eine grosse Ähnlichkeit in syntaktischer Hinsicht, eine Ähnlichkeit in der Grundsätzen des morphologischen Baus aufweisen, und eine grosse Anzahl gemeinsamer Kulturwörter
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
22
3
4
5
6
23
bieten, manchmal auch äussere Ähnlichkeit im Bestande der Lautsysteme, – dabei aber keine systematische Lautentsprechungen, keine Übereinstimmung in der lautlichen Gestalt der morphologischen Elemente und keine gemeinsamen Elementarwörter besitzen, – solche Sprachgruppen nennen wir Sprachbünde’ (Trubetzkoy, 1928: 18). (See also Stolz, 2002: 260.) Emeneau (1956: 124) mentions his use of Velten’s translation, but points to other sources which brought it to his attention (for example, Voegelin, 1945; see Sebeok, 1950: 101). Jakobson (1938: 353) mainly just repeated Trubetzkoy’s Proposition 16: ‘Les “alliance” (Sprachbünde) possédant des resemblances remarquables dans leur structure syntaxique, morphologique ou phonologoque et les “familles” (Sprachfamilien) caractérisées avant tout par un fond commun de morphemes grammaticaux et de mot usuels.’ Unter einem Sprachbund verstehen wir eine Gruppe von Sprachen, die durch gemeinsame Schicksale im gleichen Kulturraum und durch wechselseitige Beeinflussung einander so stark angenähert wurden, daß man in jeder von ihnen ungefähr das gleiche auf ungefähr die gleiche Art sagen kann. Von einem Sprachbund kann man sprechen, wenn: (a) zu einer gegebenen Zeit (b) ein zusammenhängendes geographisches Gebiet, das (c) von mindestens einer Sprachgrenze durchzogen ist, (d) von mindenstens einer Isoglosse umspannt wird.
7 Bei den Mitgliedern eines Sprachbundes handelt es sich zumindest bei einem Teil der Sprachen um solche, die nicht zu einer Familie gehören, die geographisch benachbart sind und auf grund gegenseitiger Beeinflussung eine Reihe von gemeinsamen Merkmalen aufweisen, die sich auf den lautlichen, morphologischen oder syntaktischen Bereich der betreffenden Sprachen beziehen. Ein Sprachbund weist mindestens zwei gemeinsame Merkmale auf, die sich auf mindestens drei nicht zur gleichen Familie gehörende Sprachen erstrecken, um gennetisch bedingten Ursprung oder einseitige Beeinflussung im Definitionsbereich des Sprachbundes auszuschließen. 8 Alan Dench points out (personal communication) that perhaps to a certain degree this ‘requirement’ is simply an artefact of the discovery procedure. Since it is much easier to discover borrowing against a background of typological or genetic difference than where languages are more alike, the presence of distinct language families may simply make it easier to recognize diffusion, and thus easier to defend, thereby giving the genetic distance an assumed special importance in studies of linguistic areas. 9 As Stolz (2002: 261) points out, some have argued for ‘two’ as the appropriate minimum number of shared traits to define a linguistic area (see Haarmann, 1976: 23; Schaller, 1975: 58; Wintschalek, 1993: 6). 10 A single-trait area may also enter the picture via another route. As Thomason (2001: 101) notes, it is possible that in a former linguistic area evidence of earlier contact-induced changes could have eroded, leaving only one still visible. 11 Directionality is sometimes made an issue in defining linguistic areas (see Stolz, 2002: 264). It has to do, for example, with whether the diffusion is largely unidirectional, say from some dominant language to many of its neighbours, or multilateral, crossing several language boundaries, with uncertain origins (see Thomason and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
Kaufman, 1988: 96; Aikhenvald and Dixon, 2001: 11). The distinction, however, as pointed out by Matisoff (2001: 300) becomes ‘quite artificial’ in many contact situations (see also arguments against this in Dahl, 2001; Stolz, 2002). 12 There are several other proposed areal notions that I do not take up here for lack of time; these include, for example, Bellwood’s (2001) friction zones, and upwelling or starburst zones; Renfrew’s (2000: 27) mosaic-zones; contact superposition zones (Koptjevskaja-Tamm and Wälchli, 2001: 624–6, etc.; Koptjevskaja-Tamm, 2002; see Campbell (2003) for a discussion of some of these). 13 Matisoff (2001: 300), in a separate vein, talks of different but what we might call vertical linguistic areas in the same region characterized by social differences: I would even claim that South-East Asia comprises two linguistic areas at once: one ‘vertical’, distinguishing the languages of the hard-scrabble minority populations of the hills from those of the major languages of the plains (one important difference is the lack of elaborate honorific language or status-based pronominal systems in the languages of the humble hill-dwellers); and one ‘horizontal’, cutting across the entire region. 14 Here, I follow up what I argued in Campbell (1985: 25): ‘linguistic diffusion and AL [areal linguistics] are to be equated and cannot profitably be separated; i.e. I will argue that there is no sharp boundary between the two’. However, I now see this lack of distinction a greater challenge to the notion of linguistic area generally. In recent years some others have also taken similar stances which either challenge the concept ‘linguistic area’ or advocate abandoning it (see Dahl, 2001; Koptjevskaja-Tamm, 2002; Koptjevskaja-Tamm and Wälchli, 2001; Stolz, 2002: 266; Reiter, 1991; van der Auwera, 1998b).
References Aikhenvald, Alexandra Y. (2001) ‘Areal Diffusion, Genetic Inheritance, and Problems of Subgrouping: A North Arawak Case Study’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 167–94. Aikhenvald, Alexandra Y. and R. M. W. Dixon (1998) ‘Evidentials and Areal Typology: A Case Study from Amazonia’, Language Sciences, vol. 20, pp. 241–57. Aikhenvald, Alexandra Y. and R. M. W. Dixon (2001) ‘Introduction’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 1–26. Aoki, Haruo (1975) ‘East Plateau Linguistic Diffusion Area’, International Journal of American Linguistics, vol. 41, pp. 183–99. Becker, Henrick (1948) Der Sprachbund. Leipzig: Humboldt Bücherei Gerhard Mindt. Bellwood, Peter (2001) ‘Early Agriculturalist Population Diasporas? Farming, Languages and Genes’, Annual Review of Anthropology, vol. 30, pp. 181–207. Birnbaum, Henrik (1965) ‘Balkanslawisch Und Südslawisch’, Zeitschrift für Balkanologie, vol. 3, pp. 13–63. Bloch, Jules (1919) ‘La formation de la langue marathe’, Bibliothèque de l’École des Haute Études. Sciences Historiques et Philologiques, 215 (Paris: Champion). Bloch, Jules (1925) ‘Sanskrit et dravidien’, Bulltetin de la Société Linguistique de Paris, vol. 25, pp. 1–21.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
24
25
Bloch, Jules (1930) ‘Some Problems of Indo-Aryan Philology’, Bulletin of the School of Oriental Studies, vol. 5, pp. 719–56. Bloch, Jules (1934) L’Indo-aryen du Véda aux temps modernes; revised English trans. by Alfred Master (1965) Indo-Aryan from the Vedas to Modern Times (Paris: Maissonneuve). Bloomfield, Leonard (1933) Language (New York: Holt, Rinehart & Winston). Boas, Franz (1917) Introduction. International Journal of American Linguistics. (Reprinted: Franz Boas (1940) Race, Language, and Culture (New York: The Free Press), pp. 199–210.) Boas, Franz (1920) ‘The Classification of American Languages’, American Anthropologist vol. 22, pp. 367–76. (Reprinted: Franz Boas (1940) Race, Language, and Culture (New York: The Free Press), pp. 211–8.) Boas, Franz (1929) ‘The Classification of American Indian Languages’, Language, vol. 5, pp. 1–7. Bright, William and Joel Sherzer (1978) ‘Areal Features in North American Indian Languages’, Variation and Change in Language: Essays by William Bright (Stanford: Stanford University Press), pp. 228–68. Campbell, Lyle (1985) ‘Areal Linguistics and Its Implications for Historical Linguistic Theory’, in Jacek Fisiak (ed.), Proceedings of the Sixth International Conference of Historical Linguistics (Amsterdam: John Benjamins), pp. 25–56. Campbell, Lyle (1992) ‘The Meso-American Language Area’, in William Bright and Bernard Comrie (eds), International Encyclopedia of Linguistics, vol. 2 (Oxford: Oxford University Press), pp. 415–17. Campbell, Lyle (1994) ‘Grammar: Typological and Areal Issues’, in R. E. Asher and J. M. Y. Simpson Encyclopedia of Language and Linguistics, vol. 3 (London: Pergamon Press), pp. 1471–4. Campbell, Lyle (1996a) ‘Typological and Areal Issues’, in Keith Brown and Jim Miller (eds), Concise Encyclopedia of Syntactic Theories (Oxford: Pergamon Press), pp. 339–43. [Reprint of 1994, in Encyclopedia of language and linguistics]. Campbell, Lyle (1996b) ‘Phonetics and Phonology’, in Hans Goebl, Peter H. Nelde, Zdenek Stary, and Wolgang Wölck (eds), Contact Linguistics, International Handbook of Contemporary Research, 14 (Berlin: Walter de Gruyter), pp. 98–103. Campbell, Lyle (1997a) American Indian Languages: The Historical Linguistics of Native America. (Oxford: Oxford University Press). Campbell, Lyle (1997b) ‘Typological and Areal Issues in Reconstruction’, in Jacek Fisiak (ed.), Linguistic Reconstruction and Typology (Berlin: Mouton de Gruyter), pp. 49–72. Campbell, Lyle (1997c) ‘Genetic Classification, Typology, Areal Linguistics, Language Endangerment, and Languages of the North Pacific Rim’, in Osahito Miyaoka and Minoru Oshima (eds), Languages of the North Pacific Rim, vol. 2 (Kyoto: Kyoto University), pp. 179–242. Campbell, Lyle (1998) Historical Linguistics: An Introduction (Edinburgh: Edinburgh University Press). (1999 American rights edition, Cambridge, Mass.: MIT Press). Campbell, Lyle (2001) ‘Beyond the Comparative Method?’, in Barry Blake and Kate Burridge (eds), Historical Linguistics 2001 (Amsterdam: John Benjamins), pp. 33–57 Campbell, Lyle (2002) ‘Areal Linguistics’, in Bernard Comrie (ed.), International Encyclopedia of Social and Behavioral Sciences (Oxford: Pergamon), pp. 729–33. Campbell, Lyle (2003) ‘What Drives Linguistic Diversity?’, in Colin Renfrew and Peter Bellwood (eds), Language-Farming Dispersals. (Cambridge: McDonald Institute for Archaeological Research). Campbell, Lyle, Terrence Kaufman, and Thomas Smith-Stark (1986). ‘Mesoamerica as a Linguistic Area’, Language, vol. 62, pp. 530–70.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
Campbell, Lyle and Marianne Mithun (1979) ‘North American Indian Historical Linguistics in Current Perspective’, in L. Campbell and Marianne Mithun (eds), The Languages of Native America: An Historical and Comparative Assessment (Austin, TX: University of Texas Press), pp. 3–69. Campbell, Lyle and William Poser (forthcoming) How to Show That Languages Are Related. Chappell, Hilary (2001) ‘Language Contact and Areal Diffusion in Sinitic Languages’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 328–57. Curnow, Timothy Jowan (2001) ‘What Language Features Can Be “borrowed”?’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 412–36. Dahl, Östen (2001) ‘Principles of Areal Typology’, in Martin Haspelmath, Ekkehard König, Wulf Oesterreicher, and Wolfgang Raible (eds), Language Typology and Language Universals: An International Handbook, vol. 2 (Berlin: Mouton de Gruyter), pp. 1456–70. Darnell, Regna and Joel Sherzer (1971) ‘Areal Linguistic Studies in North America: A Historical Perspective’, International Journal of American Linguistics, vol. 37, pp. 20–8. Décsy, G (1973) Die linguistische Struktur Europas: Verganenheit, Gegenwart, Zukunft. Wiesbaden: Harrassowitz. Diakonof, I. M. (1990). ‘Language Contacts in the Caucasus and the Near East’, in Thomas Markey and John Greppin (eds), When Worlds Collide (Ann Arbor, MI: Karoma), pp. 53–65. Dimmendaal, Gerrit J. (2001) ‘Areal Diffusion Versus Genetic Inheritance: An African Perspective’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 358–92. Dixon, R. M. W. (1997). The Rise and Fall of Languages (Cambridge: Cambridge University Press). Dryer, Matthew (1989) ‘Large Linguistic Areas and Language Sampling’, Studies in Language, vol. 13, pp. 257–92. Ebel, Hermann (1856) Über die Lehnwörter der deutsche Sprache (Berlin: Programm Lehr- und Erziehungs-Institut auf Ostrowo bei Filehre). Emeneau, Murray B. (1956) ‘India as a Linguistic Area’, Language, vol. 32, pp. 3–16; reprinted in Language and Linguistic Area, Essays by Murray B. Emeneau, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press), pp. 38–65. Emeneau, Murray B. (1965) ‘India and Linguistic Areas’, India and Historical Grammar. Ammamalai University, Department of Linguistics, Pub. No. 5, 25–75; reprinted: Language and Linguistic Area, Essays by Murray B. Emeneau, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press), pp. 126–66. Emeneau, Murray B. (1971) ‘Dravidian and Indo-Aryan: The Indian Linguistic Area’, in André F. Sjoberg (ed.), Symposium on Dravidian civilization (New York: Jenkins Publishing Co, Pemberton Press); reprinted in Language and Linguistic Area, Essays by Murray B. Emeneau, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press), pp. 167–96. Emeneau, Murray B. (1974) ‘The Indian Linguistic Area Revisited’, International Journal of Dravidian Linguistics, vol. 3, pp. 92–134; reprinted in Language and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
26
27
linguistic area, essays by Murray B. Emeneau, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press), pp. 197–249. Emeneau, Murray B. (1978). ‘Review of Defining a Linguistic Area, by Colin P. Masica’. Language, vol. 54, pp. 201–10; reprinted with additions in Language and Linguistic Area, Essays by Murray B. Emeneau, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press), pp. 1–18. Friedrich, Paul (1975) Proto-Indo-European Syntax: The Order of Meaningful Elements. Journal of Indo-European Studies, Memoir 1 (Butte, Montana: College of Mineral Science). Garrett, Andrew (1999) ‘A New Model of Indo-European Subgrouping and Dispersal’, Berkeley Linguistics Society, vol. 25, pp. 146–56. Giannini, Stefania and Stefania Scaglione (2002) ‘On Defining the Notion of Areality Coefficient: The Diagnostic Value of Quantitative Criteria’, in Paolo Ramat and Thomas Stolz (eds), Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer), pp. 151–70. Girard, (abbé) Gabriel (1747) Les vrais principes de la langue française ou la parole réduite en méthode conformément aux loix de l’usage en seize discours (Paris: Le Breton). Gumperz, John J. and Robert Wilson (1971). ‘Convergence and Creolization: A Case from the Indo-Aryan/Dravidian Border in India’, in Dell Hymes (ed.), Pidginization and Creolization (Cambridge: Cambridge University Press), pp. 151–67. Haarmann, Harald (1976) Aspekte der Areal-typologie: die Problematik der europäischen Sprachbünde (Tübingen: Gunter Narr). Haas, Mary R. (1978) Language, Culture, and History, Essays by Mary R. Haas, selected and introduced by Anwar S. Dil (Stanford: Stanford University Press). Haig, Jeffrey (2001) ‘Linguistic Diffusion in Present-Day East Anatolia: From Top to Bottom’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 195–224. Hamp, Eric (1977) ‘On Some Questions of Areal Linguistics’, Berkeley Linguistics Society, vol. 3, pp. 279–82. Haspelmath, Martin (1998) ‘How Young Is Standard Average European’, Language Sciences, vol. 20, pp. 271–87. Heath, Jeffrey (1978) Linguistic Diffusion in Arnhem Land (Canberra: Australian Institute of Aboriginal Studies). Heath, Jeffrey (1997) ‘Lost Wax: Abrupt Replacement of Key Morphemes in Australian Agreement Complexes’, Diachronica, vol. 14, pp. 197–232. Heine, Bernd (1994) ‘Areal Influence on Grammaticalization’, in M. Pütz (ed.), Language Contact and Language Conflict (Amsterdam: John Benjamins), pp. 55–68. Heine, Bernd and Tania Kuteva (2001) ‘Convergence and Divergence in the Development of African Languages’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 393–411. Henderson, Eugénie J. A. (1965) ‘The Topography of Certain Phonetic and Morphological Characteristics of South East Asian Language’, Lingua, vol. 15, pp. 400–34. Hickey, Raymond (1999) ‘Ireland as a Linguistic Area’, in J. P. Mallory (ed.), Language in Ulster/Ulster Folklife (Ulster Folklife, vol. 45) (Hollywood, Co. Down, Northern Ireland: Ulster Folklife), pp. 36–53. Hill, Jane (1978) ‘Language Contact Systems and Human Adaptations’, Journal of Anthropological Research, vol. 34, pp. 1–26.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
Holt, Dennis and William Bright (1976) ‘La Lengua Paya Y Las Fronteras Lingüísticas de Mesoamérica’, Fronteras de Mesoamérica, 14a Mesa Redonda, 1, Sociedad Mexicana de Antropología, México, pp. 149–56. Hübschmann, Heinrich (1875) ‘Über die Stellung des Armenischen im Kreise der indogermanischen Sprachen’, Zeitschrift für vergliechende Sprachforschung, vol. 23, pp. 5–49. Jacobs, Melville (1954) ‘The Areal Spread of Sound Features in the Languages North of California’, Papers from the Symposium on American Indian Linguistics Held At Berkeley, July 7, 1951, pp. 46–56. University of California Publications in Linguistics, vol. 10, pp. 1–68. (Berkeley). Jakobson, Roman (1931) ‘Über die Phonologischen Sprachbünde’, Travaux du Cercle Linguistique de Prague, vol. 4, pp. 234–40 (Réunion phonologique internationale tenue a Prague, 18–21/XII, 1930); reprinted 1971 in Roman Jakobson. Selected Writings, vol. 1 (The Hague: Mouton), pp. 137–43. Jakobson, Roman (1938) ‘Sur la théorie des affinités phonologiques entre les langues’, Actes du quatrieme congres international de linguists (tenu a Copenhague du 27 août 1er septembre, 1936), 48–58; reprinted, 1949, as an appendix to Principes de phonologie, by N. S. Troubetzkoy (Paris: Klincksieck), pp. 351–65. Jakobson, Roman (1944) ‘Franz Boas’ Approach to Language’, International Journal of American Linguistics, vol. 10, pp. 188–95. Johanson, Lars (1998) ‘Code-Copying in Irano-Turkic’, Language Sciences, vol. 20, pp. 325–37. Katz, Harmut (1975) Generative Phonologie und phonologische Sprachbünde des Ostjakischen un Samojedischen (Munich: Wilhelm Fink). Konow, Sten (ed.) (1906) Linguistic Survey of India. Calcutta: Superintendent, Government Printing. (Reprint 1967, 1973 (New Delhi: Motilal Banarsidass)) Kopitar, Jeernej (1829)[1857] ‘Albanische, walachische und bulgarische Sprache’, Jahrbüchern der Literatur, vol. 46, pp. 59–106. Vienna. (Reprinted 1857: Kleinere Schriften sprachwissenschaftlichen, geshichtlichen, ethnograpischen, und rechtshistorischen Inhalts, in Fr. Miklosich, Theil 1 (Vienna: F. Beck).) Koptjevskaja-Tamm, Maria (2002) ‘The Circum-Baltic Languages – a Contact-Superposition Zone in the European Periphery’, in Paolo Ramat and Thomas Stolz (ed.), Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer), pp. 209–22. Koptjevskaja-Tamm, Maria and Bernhard Wälchli (2001) ‘The Circum-Baltic Languages: An Areal-Typological Approach’, in Östen Dahl and Maria KoptjevskajaTamm (eds), Circum-Baltic Languages (Amsterdam: John Benjamins), pp. 615–750. Kroskrity, Paul V. (1993) Language, History, and Identity: Ethnolinguistic Studies of the Arizona Tewa (Tucson: University of Arizona Press). Kuhn, Adelbert (1861) Beiträge zur vergleichendem Sprachforschung auf dem Gebiete der arischen, celtischen und slawischen Sprachen (Berlin: Dummler). Kuteva, Tania (1998) ‘Large Linguistic Areas in Grammaticalization: Auxiliation in Europe’, Language Sciences, vol. 20, pp. 289–311. LaPolla, Randy (2001) ‘The Role of Migration and Language Contact in the Development of the Sino-Tibetan Language Family’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics, (Oxford: Oxford University Press), pp. 225–4. Lehiste, Ilse (1988) Lectures on Language Contact (Cambridge, Mass.: MIT Press). Martinet, André (1956) ‘Diffusion of Languages and Structural Linguistics’, Romance Philology, vol. 6, pp. 5–13.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
28
29
Masica, Colin P. (1976) Defining a Linguistic Area: South Asia (Chicago: University of Chicago Press). Masica, Colin P. (1992) ‘Areal Linguistics’, in William Bright (ed.), International Encyclopedia of Linguistics (Oxford: Oxford University Press), pp. 108–12. Matisoff, James A. (2001) ‘Genetic Versus Contact Relationship: Prosodic Diffusibility in South-Easte Asian Languages’, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 291–327. Matthews, Peter (1997) The Concise Oxford Dictionary of Linguistics (Oxford: Oxford University Press). Meillet, Antoine (1921) Linguistique historique et linguistique générale, Collection linguistique publiée par la Société de Linguistique de Paris, 8 (Paris: Champion) (new edition 1975). Meillet, Antoine (1967) The Comparative Method in Historical Linguistics (Paris: Champion) (English translatin 1925: La méthode comparative en linguistique historique, Institutet for Sammenlingende Kulturforskning, Publikationer, A, 2, Oslo). Miklosich, Franz von (1861) ‘Die slavischen Elemente im Rumanischen’, Denkschriften der philosophisch-historische Classe Kaiserliche Akademie der Wissenschaft, Wien, Commission, vol. 20, pp. 1–88 (Vienna: K. Gerolds Sohn). Müller, Max (1861) Lectures on the Science of Language (London: Longmans, Green, and Co). Nadkarni, M (1975) ‘Bilingualism and Syntactic Change in Konkani’, Language, vol. 51, pp. 672–83. Nichols, Johanna (1992) Linguistic Diversity in Time and Space (Chicago: University of Chicago Press). Nichols, Johanna (1995) ‘Diachronically Stable Structural Features’, in Henning Andersen (ed.), Historical Linguistics 1993: Selected Papers from the 11th International Conference on Historical Linguistics (Amsterdam: John Benjamins), pp. 337–56. Nichols, Johanna (1997) ‘Modeling Ancient Population Structures and Movement in Linguistics’, Annual Review of Anthropology, vol. 26, pp. 359–84. Nichols, Johanna (1998) ‘The Eurasian Spread Zone and the Indo-European Dispersal’, in Roger Blench and Matthew Spriggs (eds), Archaeology and Language II: Archaeological Data and Linguistic Hypotheses (London: Routledge), pp. 220–66. Powell, John Wesley (1891) Indian Linguistic Families of America North of Mexico. Seventh annual report, Bureau of American Ethnology, 1–142 (Washington, DC: Government Printing Office); reprinted 1966 in Franz Boas: Introduction to Handbook of American Indian Languages; J. W. Powell: Indian Linguistic Families of America North of Mexico, in Preston Holder (Lincoln, Neb.: University of Nebraska Press.) Ramanujan, A. K. and Colin Masica (1969). ‘Toward a Phonological Typology of the Indian Linguistic Area’, in Thomas Sebeok (ed.), Current Trends in Linguistics, vol. 5 (The Hague: Mouton), pp. 543–77. Ramat, Paolo (2002) ‘Introduction’, in Paolo Ramat and Thomas Stolz (eds.), Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer), pp. ix–xv. Ramat, Paolo and Thomas Stolz (eds) (2002) Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer). Reiter, Norbert (1991) ‘Ist der Sprachbund ein Werk des Satans?’, Zeitschrift für Balkanologie, vol. 26, pp. 43–62.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
Areal Linguistics: A Closer Scrutiny
Renfrew, Colin (2000) ‘At the Edge of Knowability: Towards a Prehistory of Languages’, Cambridge Archaeological Journal, vol. 10, no. 1, pp. 7–34. Sandfeld, Kristian (1902) ‘Der Schwand des Infinitivs im Rumänischen und den Balkansprachen’, Jahresbericht des Instituts für rumänische Sprache zu Leipzig, vol. 9, pp. 75–131. Sandfeld, Kristian (1912) ‘Notes sur les calques linguistiques’, Festschrift Vilhelm Thomsen (Leipzig), pp. 166–73. Sandfeld, Kristian (1930) Linguistique balkanique: Problemes et resultants, Collection Linguistique, Publication par la Société de Linguistique de Paris, 31 (Paris: Champion). Sandfeld, Kristian (1934) ‘Notes de syntaxe comparée des langues balkaniques’, Revue Internationale des Études Balkaniques, vol. 1, pp. 100–9. Sandfeld, Kristian (1938) ‘Les interférences linguistiques’, Actes du quatrieme congres international de linguists (held in Copenhagen 27 August–1 September 1936) (Copenhagen), pp. 60–5. Sarhimaa, Anneli (1991) ‘Karelian Sprachbund? Theoretical Basis of the Study of Russian/Baltic-Finnic contacts’, Finnisch-Ugrische Forschungen, vol. 50, pp. 209–19. Schaller, Helmut Wilhelm (1975) Die Balkansprachen: Eine Einführung in die Balkanphilologie (Heidelberg: Winter). Schleicher, August (1850) Die Sprachen Europas in systematischer Übersicht: linguistische Untersuchungen (Bonn: H. B. König). Schmidt, Johannes (1872) Die Verwandtschaftverhältnisse der indogermanischen Sprachen (Weimar). Schuchart, Hugo Ernst Mario (1866–8) Der Vokalismus des Vulgärlateins. vols. 1–2, 1866–77; vol. 3, 1868 (Leipzig: Teubner). Sebeok, Thomas (1950) ‘The Importance of Areal Linguistics in Uralic Studies’, Memoires de la Société Finno-Ougrienne, vol. 98, pp. 99–106. Seidel, Eugen (1965) ‘Zur Problematik des Sprachbundes’, Beiträge zur Sprachwissenschaft, Volkskunde und Literaturforschung, W. Steinitz zum 60. Geburtstag (Berlin), pp. 372–81. Sherzer, Joel (1973) ‘Areal linguistics in North America’, in Thomas Sebeok (ed.), Current Trends in Linguistics, vol. 10 (The Hague: Mouton), pp. 749–95. Sherzer, Joel (1976) An Areal-Typological Study of American Indian Languages North of Mexico (Amsterdam: North Holland Publishing Co.) Stanford, C. B. (1998) ‘The Social Behavior of Chimpanzees and Bonobos’, Current Anthropology, vol. 39, pp. 399–407. Stolz, Thomas (2002) ‘No Sprachbund Beyond this Line! On the Age-Old Discussion of How to Define a Linguistic Area’, in Paolo Ramat and Thomas Stolz (eds), Mediterranean Languages: Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer), pp. 259–81. Thomason, Sarah G. (2001) Language Contact: An Introduction (Edinburgh: Edinburgh University Press). Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics (Berkeley: University of California Press). Toman, Jindrich (1995) The Magic of a Common Language (Cambridge, Mass.: MIT Press). Tosco, M (2000) ‘Is There An Ethiopian Language Area?’, Anthropological Linguistics, vol. 42, pp. 329–65. Trask, R. L. (2000) The Dictionary of Historical and Comparative Linguistics (Edinburgh: Edinburgh University Press). Trubetzkoy, Nikolai Sergeevich (1923) ‘Vavilonskaja bashnja I smeshenie jazykov’ [The Tower of Babel and the confusion of languages], Evrazijskij vremennik, vol. 3, pp. 107–24. [Cited in Toman 1995.]
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
30
31
Trubetzkoy, Nikolai Sergeevich (1928) [Proposition 16]. Acts of the First International Congress of Linguists, 17–18. (Leiden). Trubetzkoy, Nikolai Sergeevich (1931) ‘Phonologie und Sprachgeographie’, Travaux de Circle Linguistique de Prague, vol. 4, pp. 228–34. (French translation, 1949, reprinted 1970: appendix to Principes de Phonologie (Paris: Klincksieck), pp. 343–50.) Trubetzkoy, Nikolai Sergeevich (1939) ‘Gedanken über das Indogermanenproblem’, Acta Linguistica, vol. 1, pp. 81–9. Trudgill, Peter (1983) On Dialect: Social and Geographic Perspectives (New York: New York University Press). van der Auwera, Johan (1998a) ‘Revisiting the Balkan and Meso-American Linguistic Areas’, Language Sciences, vol. 20, pp. 259–70. van der Auwera, Johan (1998b) ‘Conclusions’, in Johan van der Auwera (ed.), Adverbial Constructions in the Languages of Europe (Berlin: Mouton de Gruyter), pp. 813–36. Velten, H. V. (1943) ‘The Nez Perce Verb’, Pacific Northwest Quarterly, vol. 34, p. 271. Vendryes, Joseph (1968)[1921] Le langage: Introduction linguistique à l’histoire (Paris: Albin Michel). Voegelin, Carl F. (1945) ‘Influence of Area in American Indian Linguistics’, Word, vol. 1, pp. 54–8. Voegelin, Carl F. (1961) ‘Culture Area: Parallel With Typological Homogeneity and Heterogeneity to North American Language Families’, Kroeber Anthropological Society Papers, vol. 25, pp. 163–80. Wagner, H. (1964) ‘Nordeuropäische Lautgeographie’, Zeitschrift für celtische Philologie, vol. 29, pp. 225–98. Watkins, Calvert (2001) ‘An Indo-European Linguistic Area and Its Characteristics: Ancient Anatolia. Areal Diffusion as a Challenge to the Comparative Method?, in Alexandra Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 44–63. Weinreich, Uriel (1953) Languages in Contact: Findings and Problems (New York: Linguistic Circle of New York). Whitney, William Dwight (1868) Language and the Study of Language (New York: Scribner & Co.). Whitney, William Dwight (1875) The Life and Growth of Language (New York: D. Appleton & Co.) (Reprinted 1979: New York: Dover). Winter, Werner (1973) ‘Areal Linguistics: Some General Considerations’, in Thomas Sebeok (ed.), Current Trends in Linguistics, vol. 11 (The Hague: Mouton), pp. 135–47. Wintschalek, Walter (1993) Die Areallinguistik am Beispiel syntaktischer Übereinstimmungen im Wolga-Kama-Areal, Studia Uralica, 7 (Wiesbaden: Harrassowitz). Wolff, Hans (1959) ‘Subsystem Typologies and Areal Linguistics’, Anthropological Linguistics, vol. 1, no. 7, pp. 1–8. Zeps, Valdis (1962) Latvian and Finnic Linguistic Convergences, Uralic and Altaic series, 9 (Bloomington: Indiana University Press).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lyle Campbell
2 Thomas Stolz
1
Introduction
We are currently experiencing a boom in all kinds of areally-minded linguistic studies. It will suffice to mention the international project EUROTYP and its many spin-offs, which scrutinize the linguistic geography of various regions worldwide, such as, for example, the Mediterranean (MEDTYP) and so on (Ramat and Stolz 2002). Perhaps less well-known is the recent genesis of another research paradigm that has only partly been inspired by EUROTYP, namely Eurolinguistik, whose proponents aim at establishing some kind of pan-European transnational philology (Reiter, 1999). What all these approaches have in common is their interest in the interface-like character of areal linguistics, although this may not be unique to this linguistic subdiscipline. If one studies the linguistic properties of languages located in the same region, the expertise of various disciplines is called for. Areal linguists must be versed not only in contact linguistics, but also in linguistic typology and universals research; cultural history (of the particular region under scrutiny); descriptive grammar and national philology of the individual languages involved; diachronic grammar, both general and language-specific; and, last but not least, the principles of dialectology/linguistic geography. This is, of course, a rather challenging and demanding combination for the individual researcher, who is therefore well advised to associate with a team of like-minded fellow researchers. On a teamwork basis, one may tackle the frequently asked questions of contact linguistics. Among these questions, there is one that has been keeping experts of language contact busy for decades. How do we define a Sprachbund or a linguistic area? Even before we try to answer this question, a second one comes immediately to mind, namely: Are the two terms synonymous or do they invoke two distinct notions? In what follows, I survey very briefly a small selection of proposals that have been made concerning the search for a watertight definition of the two terms mentioned above. I argue that 32 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
All or Nothing1
Thomas Stolz 33
2
A plethora of definitions
Ever since the first official proposal by Trubetskoy (1928/30), literally dozens of often only partially compatible definitions of what exactly counts as a Sprachbund (and what does not) have been made. The number of definitions is almost coextensive with the number of linguists working in the field of areal linguistics. With a view to solving the problems of finding a widely acceptable definition, other terms have been coined whose exact relationship to the original term is by no means uncontroversial: A dictionary of partial synonyms – • • • • •
Sprachbund (Trubetskoy, 1928/30) Linguistic area (Emeneau, 1956) Sprechbund (Campbell et al., 1986) Sprachliche Konvergenzlandschaft (Stolz, 1991) Contact super-position zone (Koptjevskaja-Tamm and Wälchli, 2001)
However, none of the terminological alternatives has ousted the traditional term Sprachbund. Most of them were meant to tone down the reading the German-derived term invites. For native speakers of German, Sprachbund is suggestive of an especially intensive kind of convergence, because the final constituent of the compound, the noun Bund ‘union, federation, confederacy’ presupposes a particularly close relationship among its component parts. Members of a Bund can hardly be associated with anything outside the Bund, in a manner of speaking. Thus, not just any haphazard similarity of two or more languages may qualify as a feature capable of defining a Bund. This potential exclusiveness was felt to be too strong, as languages may freely enter and leave contact constellations with various other languages – and these contacts are not necessarily transitive in the sense that A is automatically in contact with C if both happen to be in contact with B. The alternative notions try to make do without the idea that the languages under scrutiny are tied to one another, for better or for worse. Notwithstanding these good reasons for terminological alternatives, I stick to the established term Sprachbund (and occasionally use the German plural Sprachbünde) throughout this chapter – although I shall make my own suggestions regarding areal-linguistic terminology in the final section.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
there is no universally acceptable definition of Sprachbund or linguistic area because of their ontological status. My ideas largely concur with the ones expressed by Campbell (Chapter 1 in this volume) and Masica (2001), although they have been developed independently over the course of many years of theoretical and practical work in the domain of areal linguistics.2
All or Nothing
If we abolish exclusiveness as a criterion for the definition of a Sprachbund, what can be said nevertheless is that this term and its competitors cover a wide variety of concepts with which linguists try to capture the phenomenon of linguistic convergence. They all presuppose a qualified form of linguistic similarity, namely a similarity that has developed in a multilingual setting via copying of form, function, categorial distinctions or frequency of use (Johanson, 2000). Moreover, similarity may consist of common innovations or common retentions. It may come in two shapes: either the languages are similar because they display similar concrete or abstract properties (=positive) or they lack these (=negative). Since these observations apply to all kinds of (dis)similarity, there is nothing remarkable about them. What makes them special are areal linguists’ repeated attempts at quantification. In fact, the debate about Sprachbünde and related concepts has been dominated largely by disagreement about quantitative matters (Stolz, 2002: 263). How many shared features are the minimal requirement for a Sprachbund? How many languages are needed for a Sprachbund to develop? The proposals made range from the minimum of one feature common to two languages to rather vague statements such as, for example, ‘many features in a number of languages’. Admittedly, areal linguistics is not unique as to the seductiveness of quantification. In genealogical linguistics, the crucial problem remains of how many features must necessarily be shared to justify the claim that two languages are of the same stock. Figure 2.1 is intended to give the reader an idea of the range of variation of constellations discussed under one and the same heading (the numerical values indicate the degree to which the two languages are similar: 0 = absolute dissimilar versus 100 = identical).
0
Single feature (Haspelmath, 2001)
Metatypy (Ross, 2001)
Language B Language A
100 ------------------ Increasing similarity -----------------Figure 2.1
Degrees of similarity
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
34
The majority of the hypotheses put forward so far can be located somewhere between the two extremes. It is clear that if a single isogloss is enough, the number of Sprachbünde is almost unlimited, because any two languages might have something in common, be it via historical diffusion or incidentally. Taken literally, this interpretation renders the term Sprachbund meaningless, because it may be replaced by ‘similarity’ without any loss of information (but see below). If metatypy is to serve as the minimal requirement, the number of potential candidates for the status of Sprachbund is reduced drastically – and even the paradigm case of Sprachbund phenomena, the Balkan Sprachbund, would not pass the test successfully, as [m]etatypy is . . . the process whereby the language of a group of bi- or multilingual speakers is restructured on the model of a language they use to communicate with people outside their group. In its fullest manifestation, the process includes: (a) the reorganization of the language’s semantic patterns and ‘ways of saying things’; (b) the restructuring of its syntax, i.e. the patterns in which morphemes are concatenated to form (i) sentences and clauses, (ii) phrases, and (iii) words. (Ross, 2001: 145–6). It is clear that this definition describes an extreme and thus relatively rare kind of convergence. Just to name one criterion of metatypy the languages of the Balkan Sprachbund usually fail to meet: native speakers of a given Balkanic language are not necessarily bilingual in any other memberlanguage of the Sprachbund. Note, however, that Ross (2001) does not claim that Sprachbünde are bound in any way to metatypy. As Figure 1.2 suggests, there is ample space between the extremes for an almost endless number of potential definitions. Just like the single-feature approach and the notion of metatypy, none of these potential intermediate solutions can provide us with a solid and convincing definition. Why is this so?
3
Ontology
For a start, a clarification is in order that is not intended to satisfy philosophers but merely states the linguistically obvious. Nevertheless, it is necessary to understand the often ignored reason of the difficulties that arise when one attempts to define Sprachbund. Sprachbund is not a natural (physical) phenomenon of Popper’s World I (Popper and Eccles, 1977). Its ‘existence’ requires typical World III activities: on the one hand, speakers of languages whose communicative practices unintentionally trigger processes of convergence (=invisible hand processes; Keller, 1990) and, on the other, linguists
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 35
All or Nothing
who, whatever their motives might be, recognize a given constellation of facts as evidence of the existence of a Sprachbund. Sprachbünde are not simply ‘there’ – they are constantly created anew by professionals. If the professional linguist feels that a certain number of shared features is necessary for the identification of a Sprachbund, this is largely a personal decision. Another linguist may come forward with a different quantitative requirement and so on. The magic number is, of course, fictitious and depends on the ideas of the individual proponents. However, this does not mean that anything goes. As a matter of fact, the competing models of the Sprachbund have in common the quest for the one characteristic property that keeps Sprachbünde apart from other phenomena of language contact. According to Campbell (Chapter 1 in this volume), there can be no unified definition of the Sprachbund as such – especially if one opts for fixed quantities. What can be said nevertheless is that, in the absence of a universally valid numerical value of shared isoglosses, language contact situations lend themselves to a classification as instances of a Sprachbund if the absolute number of not-genetically-based shared isoglosses among the members of the Sprachbund significantly exceeds the number of the not-genetically-based isoglosses they have in common with languages outside the Sprachbund. What counts as numerically significant, however, depends on the particulars of a given contact scenario: in one case, the members of the Sprachbund may have more properties in common than in another case. There is absolutely no way of making, say, five shared isoglosses the independent yardstick for our decision regarding the existence or absence of a Sprachbund. In addition to such quantitative matters, there is another problem that contributes to the impossibility of defining Sprachbund once and for all. There are different ways to approach the phenomenon, and one encounters examples of all of them in the extant literature: • Geography-first approach: natural topography determines the outer boundaries of a presumed Sprachbund. • History-/culture-first approach: speech-communities are known to have a common cultural and/or socio-political history; their territories (‘habitat’) delimit the extension of the Sprachbund. • Communication-first approach: a situation of multilingual practice identifies a potential centre of diffusion for traits characteristic of a Sprachbund. • Isogloss-first approach: the geolinguistic distribution of a given feature (or zone of overlap of several such features) gives shape to a Sprachbund. There are, of course, advantages and disadvantages with every approach. Going by geographical categories, for example, is an arbitrary solution, as prominent landmarks such as, for example, mountain ranges, rivers, deserts,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
36
swamps and so on do not necessarily constitute insurmountable barriers preventing humans from communicating with each other, even under the technological conditions of pre-modern times.3 If the criterion is applied strictly, say, to a continent such as Europe, whose exact boundaries are still a matter of dispute, one faces the problem of deciding whether or not to include the British Isles, the Faroe Islands and Iceland, the islands of the Mediterranean Sea and the Trans-Caucasian region, because these potential parts of Europe are located beyond a natural geographical boundary, namely the coastline and the Caucasus. If one does not want to exclude these, the geographical criterion might become absolutely abstract, because the boundaries are no longer natural topographic phenomena but longitudes and latitudes. Alternatively, geography has to be complemented by cultural history – and thus the approach is turned into a mixed one (as is done in Décsy, 2000). Moreover, there is no guarantee at all that the languages of a geographicallydefined segment of the world map converge across the board to any noticeable extent. Chances are that only some of the languages under scrutiny have traits in common in such a way that the application of the problematic term Sprachbund seems appropriate, whereas others do not qualify for membership in this or any Sprachbund – that is, the geographically-defined region may prove to be linguistically much too diverse and fragmented. To justify the geography-first approach as the methodologically superior one, it would be necessary to show that it facilitates considerably the identification of linguistic areas – which it does not. Geographical criteria might induce us to draw the line exactly between communities that have been in close contact with each other perhaps for aeons. This is the case, for instance, for the Mediterranean region or Mesoamerica, where cultural traits (religious practices, traditions, myths, calendaric systems and so on) have been diffused since time immemorial, no matter how deep the forests, how high the mountains or how dangerous the sea (Campbell et al., 1986; Stolz and Stolz, 2001). Taken literally, geography would impel us to treat the northern, southern and eastern shores of the Mediterranean separately. The major problem with the culture-/history-first approach is the necessity to prove beforehand the existence of a long-standing Kulturbund (Haarmann, 1976). Even where it is possible to identify Kulturbünde relatively easily, a problem similar to the one mentioned in connection with the previous approach arises: a Kulturbund may not be co-extensive with a Sprachbund. On the one hand, there is nothing that precludes the diffusion of crucial linguistic features beyond the boundaries of the Kulturbund, and on the other, the linguistic diversity within the Kulturbund itself is in principle unlimited – two points that Becker (1948) chose to ignore when he proposed his own culture-first model of the Sprachbund. Cultural traits may be itinerant phenomena largely independent of linguistic correlates, just as linguistic phenomena can cross any kind of border without prior or subsequent far-reaching
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 37
All or Nothing
cultural implications. Common culture, common history and a common market are all factors that create the environment in which the convergence of languages is an option, but not necessarily a must (Sherzer, 1976). Just as is the case with a geographically defined region, a culturallydefined region may be composed linguistically of several constellations that deserve the designation Sprachbund. The two previous approaches share, among other things, the feature they take for granted, that in the areas they postulate (according to non-linguistic assumptions) linguistic interaction of the communities involved not only takes place but also is of a quality such that (more than just haphazard) convergence phenomena are triggered. The proof of whether or not people have in fact been interacting is often relegated to the inventory of shared features, if at all. However, this is not a negligible issue, as, at least theoretically, there is always the chance that similarities among languages of the same geographically or culturally-defined area are not the product of diffusion but rather reflect independent parallel developments that can be explained as being determined by the restrictions that govern the range of structural variability of human language(s) in general (Stolz, 2002: 268–72). If it is in any way crucial for our understanding of the notion of Sprachbund that there are historical ties between the various instances of similar features, then incidental homologies must be counted out as evidence for the existence of a Sprachbund. Thus it is crucial to identify multilingual situations in which members of different speech-communities in fact communicate on a more or less daily basis, but this may turn out to be difficult for times gone by. Probably the best known example of a synchronic multilingual constellation in a restricted geographical area is the Kupwar case, described by Gumperz and Wilson (1971), who observe that Indo-Aryan and Dravidian languages spoken in the small village of Kupwar in India display very strong convergence phenomena, which lead these linguists to assume a common morphosyntax for all languages involved. This is especially remarkable as the speech-communities are socially and in a way also geographically segregated because of religious and economic differences, and a tendency to live in separate neighbourhoods of the village. In spite of these differences, these people interact and communicate daily. The striking morphosyntactic parallels thus find a convincing explanation that could not be provided by the culture-first approach. The geography-first approach fails as well, because it cannot predict why at one point on the map convergence is successful, whereas it does not make itself felt elsewhere in the same area. Normally, communication is not restricted in such a way that it disallows communicants from interacting with different partners in different settings. To continue the last-mentioned example, people in Kupwar do not just have dealings with their neighbours in the same village but also with the outside world.4 Disregarding that there may indeed be differences in the intensity
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
38
with which the different contacts are upheld, the possibility of more than one direction of a speech community’s communicative interaction is a prerequisite for the spread of features beyond a given point on the geolinguistic map. The communication-first approach identifies the potential focal points of a linguistic area, but the fact that people communicate throughout the whole area is, of course, hardly sufficient to identify a Sprachbund. For a Sprachbund to be present, the chain of communication that links the speech-communities to one another must result ultimately in the diffusion of a linguistic property beyond its genetically and/or typologically expected domain.5 Thus, it makes sense to start research on potential Sprachbünde from the areal distribution of features (Bechert, 1981; Nau, 1996), meaning: the most promising way to do the job of areal linguists properly is to follow the lead of dialectology by way of identifying isoglosses first. In doing so, one has to abstract from geographical and culture-historical preconceptions, because linguistic isoglosses very often do not map on to the boundaries of non-linguistic areas (but see Auer, 2004). Having identified one isogloss, the process must be repeated with other phenomena in order to determine whether or not isoglosses cluster anywhere (Masica, 1976). Furthermore, the findings have to be checked against what is known from the cultural, social, economic and political history of the speech-communities whose languages happen to partake in the various isoglosses. In Section 4, I look into the potential of the isogloss-first approach. With a view to keeping the discussion within reasonable limits, I restrict the case-study to Europe and thus adopt, strictly speaking, a mixed approach, in which geographical ideas interfere with purely phenomenological ones.6 Furthermore, I assume that a Sprachbund forms a chain of geographical neighbours that is not interrupted by other languages not belonging to the same Bund, although this issue is far from being settled.
4
On being definite in Europe . . .
For obvious reasons, the following case-study of the distribution of definitenessmarking patterns in Europe can only touch upon this issue, which deserves a more detailed treatment.7 I have chosen definiteness-marking because it counts among the criteria that Haspelmath (2001) uses to characterize the prototypical Standard Average European languages. In his study of the areal distribution of infinitive-loss on the Balkans, Joseph (1983) reaches conclusions similar to the ones I suggest in what follows – that is, the methodological consequences are the same regardless of the phenomenon to be scrutinized. Consider Table 2.1, where, for the sake of brevity, I include one language as being representative of each type of language. The types themselves are distinguished from one another on the basis of the morphosyntactic patterns they employ to encode definiteness within a noun
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 39
40
All or Nothing
Language
+Definite
−Definite
Patterns
Type
German Italian Swedish Icelandic Maltese Basque Latvian Turkish Georgian
der Mann l’uomo mann-en maður-inn ir-ragel gizon-a (vec-ais) vors adam k’aci
ein Mann un uomo en man maður ragel gizon (bat-a) (vecs) vors bir adam −DET
DET[αdef] N DET[+def] N; DET[−def] N N-DET/[+def]; DET N/[−def] N-DET/[+def] DET-N/[+def] N-DET/[+spec] ADJ-DET N DET N/[−def] H
AI A II B C D E F G
phrase. Those morphemes that serve as definite and indefinite articles or (in)definiteness markers in the sample languages are marked in boldtype. The noun is the translation equivalent of English man, the Latvian adjective translates English old. There are in total eight patterns that differ as to the distribution class and/ or position of the (in)definiteness markers. It is tempting to equate the patterns with the types to which the languages of Europe belong. However, this is a much too simplistic interpretation. First, seven of the patterns have one thing in common and thus lend themselves to being subsumed under one super-type, although there is no principle in areal linguistics that imposes a hierarchy that would prefer more general isoglosses over more particular ones. What the seven types have in common is the fact that (in)definiteness is marked overtly by grammatical means, whereas this property is not shared by type H, which happens to dominate in the eastern parts of the continent; see Figure 2.2. The feature ‘overt DET-marking’ forms an isogloss that cuts Europe into two parts: there is the entirely DET-marking West and the largely DET-marking-free East, with its small number of DET-marking islands (Mordvin and Armenian). As this distribution is already geographically significant, one could stop at this point and claim that there are two Sprachbünde, each defined by one feature, namely the presence or absence of overt DET-marking. However, the morphosyntactic differences that characterize the various patterns in Table 2.1 could still be used to refine our areal arrangement. Types A–F differ from G principally because the latter only marks indefiniteness by overt means, while definiteness remains unexpressed. In the vast majority of types, definiteness is the category that is marked obligatorily, whereas overt marking does not always apply to indefiniteness. Because Turkish is the only representative of type G in my sample,8 the new isogloss is almost coextensive with the previous one and thus it is not necessary to draw a new map.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 2.1 Representatives of DET-types in Europe
Saa Norw
Type AI
Type D
Type AII
Type E
Type B
Type F
Type C
Type G
Sw
Far
Finn
ScG
Est
Dan
Mord Bashkir
Fris
Ir
Lat
Du
Chuvash
Ukr Slok
Cz
Fr Rhaet
Slov It
Bas
Tatar
Hung
Occ
Cat
BR
Ger
Bret
Sp
Lith
Pol
Eng
W
Russ
Rom
SCr Alb
Mac
Arm Gr
Sard
Georg
Bulg Turk
Port Malt
Figure 2.2
Languages with overt DET-marking versus languages without overt DET-marking (adapted from Stolz, 2005) 41
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Ice
42
All or Nothing
• Category-based isogloss: types A and B have to be distinguished from types C, D and F because the former have overt markers of both definiteness and indefiniteness, whereas the latter make do with overt definitenessmarking. E is special because the system is based on the distinction of specific versus unspecific and allows for the co-occurrence of the indefiniteness marker and the specificity marker.10 • Hosting word-class isogloss: types A to D have to be distinguished from type F because the latter marks definiteness exclusively on the attribute in a complex NP and never on the head noun itself, whereas in the former, definiteness-marking is obligatory even in simple NPs. E is again special, as the specificity marker is a phrasal affix/clitic. • Morpheme-class membership isogloss: in all types that allow for indefiniteness marking, this is achieved via a free morpheme. However, the languages differ as to the morpheme status of their definiteness markers: many type-A languages in which the definiteness marker is also a freestanding article contrast with types B, C, D and F, whose definiteness marker is a (more or less) bound morpheme (clitic or suffix). Other members of type A behave like the last-mentioned types. • Linearization isogloss: last, but not least, the relative position of the determiner in comparison to the word it associates with morphosyntactically divides the languages into several groups: types A and D display prenominal DETS when it comes to encoding definiteness, whereas B, C, E and F show the inverse order. When indefiniteness is to be expressed, types A and B go together, as they prefer prenominal DETS. Only type E consistently employs DETS located to the right of their hosts. This catalogue of possible beginnings for further analysis of geographical distribution of features corroborates what was said in the previous section. It is up to the choice of the individual researcher to prefer one isogloss over the other(s) and thus create a Sprachbund that could have had a different shape if only a different isogloss had been chosen. Figure 2.3 tries to capture this competition of isoglosses by way of integrating them all in one go. In the subsequent discussion, I skip Basque for the most part because of its rather idiosyncratic behaviour. For the sake of argument, in Figure 2.3, only a selection of the isoglosses is given. The category-based isogloss, on the one hand, groups the Celtic languages (without Breton), Icelandic, Latvian, Lithuanian, Serbo-Croatian and Maltese together. The Baltic languages Latvian and Lithuanian are, of course, closely related and there is thus no need to resort to language contact to explain their similarity; for Serbo-Croatian, see below. Because of considerations of geography and historical contact, only Celtic and Icelandic qualify for a common Sprachbund – leaving to Maltese
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
What can happen next? There are several options, whose potential will only be reviewed for the western part of Europe:9
Category-based
Saa
Hosting word-class
Norw
Morpheme-class membership
Sw
Far
Linearization
Finn
ScG
Est
Dan
Mord Bashkir
Fris
Ir
Lat
Du W
Chuvash
Rhaet Occ
Tatar
Hung Slov
It
Bas
Ukr
Slok
Cz
Fr
Cat
BR
Ger
Bret
Sp
Lith
Pol
Eng
Russ
Rom
SCr Alb
Mac
Georg
Bulg
Arm
Gr
Sard
Turk
Port Malt
Figure 2.3
Competing isoglosses (adapted from Stolz, 2005) 43
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Ice
All or Nothing
the role of an outsider characterized by independent parallel development. On the other hand, the three focal areas where indefiniteness marking is lacking are all located at the periphery of the large area identified for the western part of Europe. The presence of a formal distinction between definiteness and indefiniteness is thus a positive feature of the bulk of the languages spoken in the European West. The hosting-word-class isogloss identifies again the two Baltic languages and Serbo-Croatian as the ones that do not conform to the majority of the other languages with definiteness marking. Because of the close genetic relationship among the Baltic languages, it does not make much sense to invoke contact-induced convergence for definiteness marking exclusively on the attribute. Note, however, that definiteness/indefiniteness also determines the weak or strong inflection of adjectives in Germanic languages (apart from English). Moreover, definiteness marking on the adjectival attribute used to be a commonality in the Slavic phylum a millennium ago but was subsequently lost from all individual languages apart from Serbo-Croatian (and unproductive residues in Slovenian). For Maltese and Greek, it must additionally be conceded that, under certain conditions, definiteness may be marked twice, namely on the head noun and on the adjectival attribute. Be that as it may, this isogloss corresponds largely, but not entirely, to the previous one. The morpheme-class-membership isogloss isolates type A from the other types. In fact, it cuts across type A because, in Table 2.1, the distinction between affix, clitic and free morpheme is applied only summarily. Admittedly, this distinction is tricky in so far as it is difficult to make a categorical distinction between free-standing prenominal articles and proclitics. Given the continuumlike character of linguistic phenomena in general, we are facing a gradient property, and any solution applied to such a property must retain an element of arbitrariness. According to my analysis, Hungarian, Greek, German, Dutch, Frisian and English go together, whereas the remainder of type A (Romance languages outside the Balkans) side with the bulk of the other types. The linearization isogloss shows that the North Germanic branch and the Baltic languages converge with Basque and the Balkan languages Albanian, Romanian, Macedonian and Bulgarian because of their preference for postdetermination – independent of the fact that in complex NPs, mainland Scandinavian languages and Romanian also allow for NP-initial position of DET.11 It is worth noting that, in Old Georgian, similar rules were in operation: the erstwhile definite article normally occurred in post-nominal position but moved to the left if the NP was further determined by quantifiers and other such things (Fähnrich, 1994: 65). In earlier periods of the language, northern dialects of Russian used post-nominal DETS, which, in contradistinction to the Old Georgian ones, were more closely bound elements with suffix-like qualities (Kiparsky, 1967). All other languages in the western part of Europe, without exception, give precedence to predetermination.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
44
The isoglosses overlap in a relatively small section: the northern branch of type AI partakes in all isoglosses in Figure 2.2. This cluster of isoglosses, however, is clearly genetically-induced, as the four languages Dutch, English, Frisian and German all belong to the West-Germanic sub-phylum. If we go by the criterion according to which one shared feature is enough to give rise to a Sprachbund, there are three, probably unwanted, effects: first, the number of Sprachbünde, even within Europe, rockets; second, one and the same language may partake in several Sprachbünde at the same time for the same phenomenon; and third, the Sprachbünde depend on isoglosses based on qualities which, in turn, depend on the relatively (that is, not absolutely) free choice of the individual observer. Other or additional criteria would automatically yield different Sprachbünde and more fragmentation. Because there is no naturally determined number of shared features that define a Sprachbund, not much would be gained by increasing the number of cases of overlap. It is true that the more shared isoglosses there are, the likelier it becomes that we are dealing with a zone of especially intense language contacts. However, how intense language contact and how numerous the evidence of it must be to make a linguist employ the label Sprachbund – are questions whose answers are more of a socio-psychological rather than a linguistic nature.
5
Conclusions
In conclusion, let me summarize the major lines of the foregoing sections. I have argued that, because Sprachbünde are not objects of the physical world but rather projections from the minds of linguists, truth-conditions cannot be applied to the various competing alternative notions. None of these is wrong, nor can any of them claim exclusively to be right. The only criteria of evaluation to which the competing definitions are subject are viability and heuristic power. The coexistence of only partially compatible definitions can perhaps be reinterpreted as the paradigm necessary for the typology of convergence. Each definition put forward so far focuses on a subsection on the continuum of convergence processes. Each definition highlights a particular constellation of facts and conditions typical of a given degree of convergence. The higher the number of features to be shared by convergent languages, the more chances there are that we are dealing with a Sprachbund in the sense of the communication-first approach. The larger the area in which a single feature is attested, the more chances there are that we are dealing with an ‘itinerant’ feature in the sense of the isogloss-first approach. Geography and culture have been shown to be insufficient criteria from which to start areal-linguistic research. The brief remarks I made as to earlier stages of some of the languages under scrutiny already suffice to show that isoglosses themselves are subject to change. This is, of course, largely trivial, as areal diffusion of features also presupposes a chronology – that is, a diachronic expansion of the territory
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 45
All or Nothing
in which a given feature is attested. However, expansion does not tell the whole story. Loss of features is also a common diachronic phenomenon: the disappearance of the definiteness inflection on adjectives in all Slavic languages apart from Serbo-Croatian, the loss of the definite article in Georgian, and the ephemeral existence of suffixed definite articles in northern dialects of Russian suggest that, several centuries back, Europe displayed a different fragmentation into areas and sub-areas. Since articles and definiteness markers have not always been present but have a pre-history of their own in the various languages, the map changes once again if we go back in time another five hundred years. No doubt the genesis and loss of DET-marking strategies can be understood as a bundle of similar processes reinforced by language contact. However, the internal diversity and diachronic dynamics of the areas that result from the isoglosses clearly demonstrate that a Sprachbund in the sense of a special category neatly distinct from other phenomena cannot be justified in this way. Thus, one should either strip the term of its unwelcome and much too suggestive connotations or abolish it for good (but it should be kept in the virtual museum of linguistic thought as an example of how difficulties and misunderstandings can be created via terminology).
Abbreviations Adj=adjective, Alb=Albanian, Arm=Armenian, Bas=Basque, BR=Bielo-Russian, Bret = Breton, Bulg = Bulgarian, Cat = Catalan, Cz = Czech, Dan = Danish, def = definite, Det = determiner, Du = Dutch, Eng = English, Est = Estonian, Far = Faroese, Finn = Finnish, Fr = French, Fris = Frisian, Georg = Georgian, Ger = German, Gr = Greek, Hung = Hungarian, Ice = Icelandic, Ir = Irish, Lat= Latvian, Lith = Lithuanian, Mac = Macedonian, Malt = Maltese, Mord = Mordvin, N = noun, Norw = Norwegian, Occ = Occitan, Pol = Polish, Por = Portuguese, Rhaet = Rhaeto-Romance, Rom = Romanian, Russ = Russian, Saa = Saami, Sard = Sardinian, ScG = Scots-Gaelic, SCr = Serbo-Croatian, Slok = Slovak, Slov = Slovenian, Sp = Spanish, spec = specific, Sw = Swedish, Turk = Turkish, Ukr = Ukrainian, W = Welsh
Notes 1 This chapter is based on the talk I gave on the occasion of the Workshop, ‘Contact, borrowability and typology’ at the annual meeting of the Linguistic Association of Great Britain (University of Surrey at Roehampton, 27 August 2004). Because of a subsequent change of emphasis in the contents, I no longer deem the original title ‘ “Sprachbund” versus “Sprechbund”: The Typology of Linguistic Areas’ to be appropriate. I am grateful to April MacMahon, Yaron Matras and Marianne Mithun for inviting me to the Workshop. The three reviewers of the draft version of this chapter have given me valuable advice. Alan Aydelott deserves a word of thanks too for brushing up my non-native English.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
46
2
3
4
5
6
7
8
9
Tamari Khizanishvili, Aina Urdze and Sonja Kettler have kindly provided technical help where I needed it most. As Campbell’s contribution to this volume gives a full account of the (pre-)history of the Sprachbund debate and reviews the many contemporary competing approaches, I take the liberty of sparing the reader too much repetition and thus focus on a restricted set of aspects relevant to my topic. For further bibliographical references and the presentation of other hotly debated issues connected to the Sprachbund problem, the reader may consult Campbell (Chapter 1 in this volume), Dahl (2001), and/or Stolz (2002). In Stolz (1991), I deliberately looked at the convergence phenomena that happened in the region where Latvian and Estonian meet. Nau (1996) has shown that a number of the features I identified as proof of convergence of the two languages have a much wider distribution and thus their ultimate source cannot be determined. The case of ‘long-distance convergence’ of geographically distant languages via contact with a common superstrate/adstrate is similar in some ways. In the erstwhile Spanish colonial empire, Amerindian and Austronesian languages, which belong to different genetic and typological groups and have never been spoken in the vicinity of one other, have developed common features by borrowing and calquing from the prestige language Spanish (Stolz and Stolz, 1997). This does not mean that only spectacular deviations from the usual genetic or typological picture pass as evidence for convergence (Stolz, 2002). A deviation may just consist in a higher or lower text frequency of a given phenomenon and so on (Stolz, 2003; Stolz and Sansò, forthcoming). As a matter of fact, the property chosen for discussion, definiteness-marking, is also attested in Arabic, Hebrew and Cushitic languages such as, for example, Somali. The isoglosses postulated for the European languages would thus need to be extended into the Middle East and North-East Africa. For practical reasons, my sample is made up exclusively of standard varieties. I therefore cannot discuss the cases of definite articles, which are reported as being developed currently in colloquial varieties of Finnish (Juvonen, 2000), Estonian (Pajusalu, 1997) and Slavic languages of the Western branch (Lötzsch, 1996). Nevertheless, these additional cases also reflect language contact, as the ‘category’ article seems to be spreading from the languages of the western part of Europe to their immediate neighbours on the dividing line that separates languages with grammatical definiteness marking from those without definiteness marking. The presentation is strictly synchronic, although I mention some diachronic points in passing. Moreover, I cannot go into the dialectal differences of article systems studied by Dahl (2004) for continental North Germanic. Dahl’s description suggests that substandard varieties of one and the same standard language may belong to different types. This holds under the proviso that my interpretation of the facts in Tatar, Chuvash and Bashkir can stand the test. If these Turkic languages also happen to have indefinite articles, this does not invalidate the general hypothesis, because type G would still be confined to a genetically-defined group of languages outside the western half of Europe. There is, of course, a variety of other isoglosses in the realm of definiteness marking, namely, co-occurrence of possessive pronoun and definite article (Haspelmath, 1999), multiple definiteness marking (Dahl, 2004), co-occurrence of demonstratives and definite articles (Plank, 2003), coalescence of article and adjacent words
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 47
All or Nothing
(Stolz, 2005) and so on. Since this chapter is not meant primarily as an in-depth study of the areal linguistics of the definite article in Europe, I skip these interesting issue in order to save space for the discussion of more general problems of the Sprachbund debate. 10 The properties of the Basque system have occasioned a controversy between Haspelmath (2001), who only accepts a definite article for Basque to the detriment of bat, and Heine and Kuteva (forthcoming), who take issue with his analysis and argue that, strictly speaking, there is no such thing as a definite article in Basque, although they do accept the indefinite article. Hualde and Urbina (2003: 118–22) show that neither label is fully appropriate for the Basque system. 11 The classification of the Scandinavian and Balkanic articles as two of a kind is exclusively based on the default position of the DET to the right of its morphological host. On closer inspection, it turns out that there are considerable structural differences between North Germanic and Romanian, Bulgarian and Albanian (Börjars, 1994). Suffice it to say that, in the languages of the Balkans, DET retains a number of features characteristic of cliticization whereas, in Danish, Swedish and Norwegian, the suffixed articles do not lend themselves to being analysed as clitics. Thus, one may divide type B into two distinct types, B1 and B2. For the present purpose, however, I can make do with type B as the exact number of types is not relevant for my line of argument.
References Auer, P. (2004) ‘Sprache, Grenze, Raum’, Zeitschrift für Sprachwissenschaft, vol. 23, no. 2, pp. 149–7. Bechert, J. (1981) ‘Notiz über eine Möglichkeit, die historisch-vergleichende Sprachwissenschaft zu vervollständigen, oder: Lesefrüchte zur Verbesserung Mitteleuropas und anderer Weltgegenden’, Papiere zur Linguistik, vol. 25, pp. 85–9. Becker, H. (1948) Der Sprachbund (Leipzig: Gerhard Mindt). Börjars, K. (1994) ‘Swedish Double Determination in a European Typological Perspective’, Nordic Journal of Linguistics, vol. 17, pp. 219–52. Börjars, K., T. Kaufman, and T. Smith-Stark (1986) ‘Mesoamerica as a Linguistic Area’, Language, vol. 62, no. 3, pp. 530–70. Dahl, Ö (2001) ‘Principles of Areal Typology’, in Haspelmath et al., 1456–70. Dahl, Ö. (2004) ‘Definite Articles in Scandinavian: Competing Grammaticalization Processes in Standard and Non-Standard Varieties’, in B. Kortmann (ed.), Dialectology Meets Typology. Dialect Grammar from a Cross-linguistic Prespective (Berlin: Mouton de Gruyter), pp. 147–80. Décsy, G. (2000) The Linguistic Identity of Europe, 2 vols (Bloomington: Eurolingua). Emeneau, M. B. (1956) ‘India as a Linguistic Area’, Language, vol. 32, no. 1, pp. 3–16. Fähnrich, H. (1994) Grammatik der altgeorgischen Sprache (Hamburg: Buske). Gumperz, J. J. and R. Wilson (1971) ‘Convergence and Creolization: A Case from the Indo-Aryan/Dravidian Border’, in D. Hymes (ed.), Pidginization and Creolization of Languages (Cambridge: Cambridge University Press), pp. 151–68. Haarmann, H. (1976) Aspekte der Arealtypologie. Die Problematik der europäischen Sprachbünde (Tübingen: Narr). Haspelmath, M. (1999) ‘Explaining Article-possessor Complementarity: Economic Motivation in Noun Phrase Syntax’, Language, vol. 75, no. 2, pp. 227–43. Haspelmath, M. (2000) ‘The European Linguistic Area: Standard Average European’, in Haspelmath et al., 2001, pp. 1492–510.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
48
Haspelmath, M., Ekkehard König, Wulf Österreicher and Wolfgang Reible (eds) (2001) Language Typology and Language Universals, 2 vols (Berlin: Mouton de Gruyter). Heine, B. and T. Kuteva (forthcoming) Language Contact and Grammatical Replication (Oxford: Oxford University Press). Hualde, J. I. and J. O. de Urbina (2003) A Grammar of Basque (Berlin: Mouton de Gruyter). Joseph, B. D. (1983) The Synchrony and Diachrony of the Balkan Infinitive. A Study in Areal, General, and Historical Linguistics (Cambridge: Cambridge University Press). Johanson, L. (2000) Structural Factors in Turkic Language Contacts (London: Curzon). Juvonen, P. (2000) Grammaticalizing the Definite Article. A Study of Definite Adnominal Determiners in a Genre of Spoken Finnish (Stockholm: Department of Linguistics, Stockholm University). Keller, R. (1990) Sprachwandel: von der unsichtbaren Hand in der Sprache (Tübingen: Francke). Kiparsky, V. (1967) Russische historische Grammatik. Band 2: Die Entwicklung des Formensystems (Heidelberg: Carl Winter). Koptjevskaja-Tamm, M. and B. Wälchli (2001) ‘The Circum-Baltic Languages: An Areal-typological Approach’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), CircumBaltic Languages, 2 vols (Amsterdam: John Benjamins), pp. 615–750. Lötzsch, R. (1996) ‘Interferenzbedingte grammatische Konvergenzen und Divergenzen zwischen Sorbisch und Jiddisch’, Sprachtypologie und Universalienforschung, vol. 49, no. 1, pp. 50–9. Masica, C. P. (1976) Defining a Linguistic Area: South Asia (Chicago: University of Chicago Press). Masica, C. P. (2001) ‘The Definition and Significance of Linguistic Areas: Methods, Pitfalls, and Possibilities (with Special Reference to the Validity of South Asia as a Linguistic Area)’, in P. Bhaskararao and K. V. Subarao (eds), The Yearbook of South Asian Languages and Linguistics (London: Sage), pp. 205–67. Matras, Y. (2002) Romani: A Linguistic Introduction (Cambridge: Cambridge University Press) Nau, N. (1996) ‘Ein Beitrag zur Arealtypologie der Ostseeanrainersprachen’, in N. Boretzky, Werner/enninger and Thomas Stolz (eds), Areale, Kontakte, Dialekte. Sprache und ihre Dynamik in mehrsprachigen Situationen (Bochum: Brockmeyer), pp. 51–67. Pajusalu, R. (1997) ‘Is There an Article in (Spoken) Estonian?’, in M. Erelt (ed.), Estonian: Typological Studies II (Tartu: Department of Estonian of the University of Tartu), pp. 146–77. Plank, F. (2003) ‘Double Articulation’, in F. Plank (ed.), Noun Phrase Structure in the Languages of Europe (Berlin: Mouton de Gruyter), pp. 337–96. Popper, K. R. and J. C. Eccles (1977) The Self and Its Brain – An Argument for Interactionism (Heidelberg: Springer). Reiter, N. (ed.) (1999) Eurolinguistik. Ein Schritt in die Zukunft (Wiesbaden: Harrassowitz). Ross, M. (2001) ‘Contact-Induced Change in Oceanic Languages in North-West Melanesia’, in A. Y. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance. Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 134–66. Ramat, P. and T. Stolz (eds.) (2002) Mediterranean Languages. Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Brockmeyer). Sherzer, J. (1976) An Areal-Typological Study of American Indian Languages North of Mexico (Amsterdam: North Holland).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Thomas Stolz 49
All or Nothing
Stolz, T. (1991) Sprachbund im Baltikum? Estnisch und Lettisch im Zentrum einer sprachlichen Konvergenzlandschaft (Bochum: Brockmeyer). Stolz, T. (2002) ‘No Sprachbund Beyond this Line! On the Age-Old Discussion of How to Define a Linguistic Area’, in Ramat and Stolz (2002), pp. 259–81. Stolz, T. (2003) ‘A New Mediterraneanism: Word Iteration in An Areal Perspective. A Pilot-Study’, Mediterranean Language Review, vol. 15, pp. 1–47. Stolz, T. (2005) ‘Sardinian in Typological Perspective: Morphonological Interaction of Definite Articles and Adjacent Words’, Sprachtypologie und Universalienforschung, vol. 58, nos 2, 3, pp.147–77. Stolz, T. and A. Sansò (forthcoming) ‘The Mediterranean Area Revisited. Word-Iteration as a Potential Mediterraneanism’, Orbis. Stolz, C. and T. Stolz (1997) ‘Universelle Hispanismen? Von Manila über Lima bis Mexiko und zurück: Muster bei der Entlehnung spanischer Funktionswörter in die indigenen Sprachen Amerikas und Austronesiens’. Orbis, vol. 39, no. 1, pp. 1–77. Stolz, C. and T. Stolz (2001) ‘Mesoamerica as a Linguistic Area’, in Haspelmath et al. 2001, 1539–53. Trubetskoy, N. S. (1928/1930) ‘Proposition 16’, in Actes du premier congrès international de linguistes à la Haye, du 10–15 avril 1928 (Leiden: Sijthoff), pp. 17–18.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
50
Keeping Contact in the Family: Approaches to Language Classification and Contact-induced Change April McMahon and Robert McMahon
1
Back to the future: trees and waves
One of the cornerstones of nineteenth-century historical-comparative linguistics is the regularity hypothesis (see Morpurgo Davies, 1998). This idea that regular correspondences, of the kind observed by Grimm, Bopp and their contemporaries, reflect regular, exceptionless sound changes, underlies much of the progress made by the Neogrammarians and in the subsequent development of historical linguistics. Furthermore, it is a very good example of a kind of thinking that has been vital to linguistics more generally – that is, the notion that we can make progress by adopting strong methodological hypotheses. These may subsequently require modification; but adopting them in the first place can have unforeseen positive consequences in helping us to understand the way language works. The regularity hypothesis is also encapsulated in the family tree model (Schleicher, 1863; Koerner, 1983) which maps the consequences of just such idealized, regular changes. If change is regular, we can also reverse it; and this assumption is critical to the operation of the comparative method and the diagnosis of linguistic family relationships. However, this highlights one case where a strong methodological hypothesis does overgeneralize, since, of course, resemblances between languages do not only arise from common ancestry. They can also reflect a rather different kind of history, with contact and borrowing between speakers and their language systems, as encapsulated in turn in the rival wave model (Schmidt, 1872). Both lexical borrowing and convergence have very serious implications for our attempts to classify languages into families, since these can give the appearance of regularity in the relationships between languages, but can be impossible to reverse. 51 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
3
Language Classification and Contact-induced Change
These problems are not of interest only to historians of linguistics: on the contrary, they are brought into sharp relief in the context of recent attempts to introduce quantitative methods into historical linguistics, as we shall see below. It follows that we should be concerned about how historical linguists have reacted to contact; and here, we shall argue that the methodological assumptions involved are not so positive, and require urgent correction if we are to make further progress in comparative work. We shall also argue that, although linguists are very good at formulating bold hypotheses, the discipline has not yet developed a fully scientific attitude to their testability. We shall be suggesting some ways of making our historical hypotheses more testable.
2
Excluding contact
It appears that the general policy on contact-induced change in language classification has so far been one of exclusion, with the mechanism of exclusion itself taking one of two possible forms. On the one hand, it might be analytical: either it is assumed that borrowings are simply not an issue, provided that we select the data properly, or that they can be removed by ‘cleaning up’ the data retrospectively. Alternatively (and as we shall see, this is particularly true for cases of non-lexical borrowing), there are attempts at what we might call philosophical exclusion: that is, we are exhorted to regard contact-induced changes as being so marginal and insignificant that they can comfortably be ignored completely. Our view is that neither type of exclusion can be the right policy. Borrowing and other contact-induced changes have now been established as facts of linguistic life; languages may fit into more-or-less tree-like patterns, but their histories are just as interesting and legitimate even if they don’t. In addition, a policy of excluding contact leads to a profound and disturbing mismatch between the concerns of historical linguists working on language classification and reconstruction, and those of colleagues engaged in research on language change. One of the current major areas of concern in language change is precisely contact and its consequences: we see this amply illustrated in work on pidginization and creolization; mixed languages; language endangerment, attrition and death; and, of course, linguistic areas (for a recent summary, see Thomason, 2001). It is hardly helpful for the unity of the discipline that linguists tracing changes down family trees are so strongly involved with contact-induced developments, while those reversing changes up through family trees, or plotting the shapes of the trees themselves, are excluding just these changes. We shall be arguing that we must reverse these current attitudes to contact in comparative linguistics, for two reasons. We shall show that we cannot in any case rely on screening-out borrowings using traditional historical methods. But, in addition, there are positive benefits that follow from recognizing and using loans, though accessing those benefits also requires the development
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
52
April McMahon and Robert McMahon
53
of new, quantitative, computational techniques.1 We turn initially, however, to the difficulties inherent in both the analytical and philosophical approaches to the exclusion of contact-induced change.
The first difficulties confronting any attempt to exclude loans analytically are purely methodological ones: we must ask whether factoring-out borrowings is practically possible, for example, and whether algorithmic corrections can operate sufficiently generally to be useful, or are simply situation-specific, post-hoc fixes that really just build our existing knowledge back into the analysis. There is certainly a long history in comparative linguistics of assuming that lexical borrowing can essentially be disregarded so long as we pre-select our data for comparison carefully. The ‘industry standard’ for meaning-list comparisons is the Swadesh basic vocabulary list (Swadesh, 1952), of 100 or 200 meanings, which are intended to be culturally neutral, and resistant to borrowing. The problem, of course, is that everything is relative: we can assume that these meanings are more universal and less borrowable than a randomly selected list of the same length; but there are going to be exceptions. In other words, acknowledging that there is less likelihood of borrowing in a basic vocabulary list does not mean this is an absolute impossibility, and it is therefore essential for us to be able to detect loans in Swadesh lists. Embleton (1986) approaches this problem by proposing an algorithmic correction for the effects of borrowing. She points out that the typical reaction on discovering a loan in a basic vocabulary list is to exclude that meaning; the problem is that this will inevitably result in incremental reduction in the length of the list, with results equally becoming less and less robust. For this reason, Embleton turns instead to the development of borrowing parameters. Embleton adds to r, the usual lexicostatistical replacement rate, a further adjustment of b/kX, or the borrowings into language X from each of its kX neighbours. Although ‘ideally of course b is supposed to be zero for the Swadesh-lists’ (Embleton 1986: 79), we know there are cases of loans in the basic vocabulary; the question is how b is to be calculated. Initially, Embleton provides an analysis for Germanic, and the first step is to list all the identifiable borrowings. These are added together to give a value of b for each language pair, and that value is then fed back into the calculations, effectively correcting for those identified loans (and, Embleton argues, providing more secure and realistic results in terms of glottochronological dating, though her lexicostatistical arguments can be accepted and evaluated without reference to any further potential glottochronological application). The problem is that Embleton’s calculations are specific to each pair of languages; they are not generalizable. Worse, when she turns to Romance, she finds that the borrowing rates she calculates still do not give the right
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
2.1 Analytical exclusion
Language Classification and Contact-induced Change
dates, so she must build in what she accepts is a fudge factor, ‘with the borrowing rates all arbitrarily increased by half’ (1986: 141), presumably as a result of undetected borrowings in the Romance lists. Perhaps the greatest difficulty is that this method relies absolutely on finding the loans first; in cases where we are unsure whether borrowing has taken place or not, Embleton’s approach simply cannot help us. None of this would be so problematic if we could demonstrate that the number or proportion of borrowings in the average Swadesh list comparison is small. However, Embleton herself (1986: 100–1) identifies, in a standard 200-meaning list, twelve borrowings from French into English, sixteen from North Germanic into English, nineteen from Danish into Faroese, and fifteen from Dutch into Frisian. This is clearly a relatively heavy concentration, and in our current state of knowledge, we cannot show that it is unrepresentative: on the contrary, these figures are generally comparable to those obtained by Kessler (2001) for languages from a wider range of families. In turn, this raises the further question of whether we can realistically expect to have identified all the actual loans by using conventional techniques. If there are, in fact, further, undiagnosed loans lurking in our lists, this could be having an unpredictable effect on our analyses. We return to this issue in Section 3 below. 2.2 Philosophical exclusion It would seem, then, that there are serious problems in existing attempts to exclude borrowings analytically, by pre-selecting the data or applying a correction after the fact. The alternative is to exclude contact-induced change philosophically, or perhaps definitionally; one possible example comes from Ringe et al. (2002). Ringe and his colleagues, in their Computational Cladistics project, develop a character-based approach, which operates with a pre-selected set of characters, or features. These can be phonological, morphological or lexical, and will have different values across the range of languages being compared; this is comparable to a sociolinguistic approach based on variables, but in a historical dimension. Ringe et al. (2002) focus on the first-order splitting of IndoEuropean, and therefore choose characters that identify particular subfamilies, either individually, such as Tocharian, or as groups, such as Italo-Celtic. The features they choose, of course, are those that have already emerged from prior philological work, so that this approach can be seen as a means of computerizing aspects of the comparative method. In total, Ringe et al. (2002) work with 22 phonological characters, 15 morphological characters, and 333 lexical characters: their software then searches for the ‘perfect phylogeny’, or the tree that is consistent with all the characters considered. Failing that, the best tree will be the one consistent with the most data. We shall not consider Ringe et al.’s method itself in any detail here, though it is undoubtedly promising, and an extremely helpful addition to
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
54
55
the developing range of quantitative approaches: further discussion can be found in McMahon and McMahon (2004). However, Ringe et al.’s treatment of contact is of direct relevance here. The perfect phylogeny approach they adopt is clearly, by its very name, reliant on the family tree model, and Ringe et al. (2002: 65) argue that ‘the tree model of linguistic speciation is normally appropriate, if the loss of contact between diverging dialects has been relatively abrupt and no discontinuities of transmission can be demonstrated for any of the languages in question’. In other words, prioritizing a method that is (at least currently) heavily reliant on trees, necessarily means downplaying the effects of contact. Ringe et al. accept that lexical borrowing will happen; but they assume that ‘most words borrowed from foreign languages can be identified as such in a language’s basic vocabulary’ (2002: 78). The few that are not will cause peculiarities in the tree, and will tend to stop a perfect phylogeny from being generated: this is the case for Ringe et al.’s (2002) own results, which are compromised by the involvement of Germanic in 16 out of 18 problematic characters, a peculiarity that Ringe et al. themselves attribute to contact. However, consequences of contact other than lexical borrowing are relegated to the periphery of linguistics. That is, languages with ‘mixed grammars’ are taken to result from discontinuities in transmission, and are either creoles, or ‘descended from an imperfectly learned second language which became the community norm’ (Ringe et al., 2002: 64). The crux of this argument seems to be their assertion that ‘mixed grammars are not known to result from native-language acquisition’ (2002: 63), so that discontinuous transmission alone can result in mixing. Furthermore, ‘to judge from the aggregate of languages whose histories are actually documented for at least a few centuries, such discontinuities appear to be infrequent’ (2002: 63). However, this approach raises two problems. First, whether the mixing is a result of influence from an L2 on an L1, or from imperfect learning of an L2 with influence from the L1 and hence in the opposite direction, surely does not affect the overall outcome, in the sense of a language with more than one ancestor. Subsequent generations of speakers will indeed have to learn that mixed system; they may not be doing the mixing on-line, as it were, and they may not see the system they are learning as anything out of the ordinary, but from the perspective of a historical linguist, surely that system is still undeniably a mixed one, and therefore strictly incompatible with the family tree? Furthermore, we shall have to be confident that we can always identify such cases. Ideally, of course, we would have a series of diagnostic features that would tell us without question when we are dealing with a language with a history of contact-induced changes, or with creole ancestry, or with discontinuities in transmission. We do not currently have such a set of features. In addition, as Ringe et al. (2002: 65) note, ‘there remains a tiny handful of languages that exhibit unarguably mixed grammars but do not seem to be typical creoles’, for which they cannot account. Given the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon and Robert McMahon
Language Classification and Contact-induced Change
relatively small amount of research to date on mixed languages, it seems unsafe to assume that the category is necessarily insignificant. Thomason (2001: 218), for example, argues that ‘the study of bilingual mixed languages is still in its infancy’, and that ‘the wide range of variation already evident in the languages currently available for study is likely to be the proverbial tip of an iceberg . . . anyone who believes that sweeping generalizations or strong predictions about these languages are possible in our current state of knowledge is at best overoptimistic’. At present, then, excluding any such cases a priori would seem to be premature, providing another reason for not attempting to remove contact-induced changes from our analyses.
3
Identifying and using lexical borrowing
3.1 ‘Historical connectedness’ It seems clear, then, that we cannot guarantee that we can in fact identify and exclude borrowings by conventional historical linguistic means; at the same time, contact-induced change cannot simply be marginalized or assumed to have no effect – it is too common and too far-reaching for this to be a reasonable strategy. One might argue that this demonstrates only one thing: namely, that distinguishing the effects of contact from those of descent with differentiation is impossible, and that we therefore should not seek to make such a distinction at all. This seems to be the approach advocated by Kessler (2001), in his book on word lists. Kessler uses the term ‘historical connection’, and argues that two languages will be connected historically if they are similar to one another, regardless of the source of those similarities: ‘whether language elements share certain properties because they are inherited from a common ancestor language, or whether they share them through borrowing, the languages and the elements in question can be said to be historically connected’ (2001: 5). Kessler argues that he is ‘concerned with distinguishing languages that are historically connected from those that are not . . . I am not proposing new methods for distinguishing the different types of historical connection from each other’ (2001: 5). For this reason, his figures often include measures of traditional cognates as well as loans put together to give a composite similarity score. Though this is a pragmatic approach and has clear benefits in identifying cases where there is no similarity from whatever source, none the less Kessler is amalgamating two different contributions to history that we might want to keep separate. We would argue that a recognition of the importance of both borrowing and common ancestry does not necessitate the extreme step of collapsing the distinction. Kessler himself continues to recognize, though not always to analyse, the difference between the two; ideally, we should be able to see this difference reflected in our quantitative methods.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
56
April McMahon and Robert McMahon
57
How, then, can we use a paradigm involving the formulation and testing of hypotheses by quantitative means, to identify and use borrowings in meaning-list data? The first step is to ask whether traditional Swadesh lists are appropriate in this kind of work at all, and whether computational approaches can be used to get more mileage out of rather basic comparative linguistic data of this type. The interesting prospect that follows from adopting approaches such as this is that we need not simply accept the Swadesh list wholesale; we can test to see which meaning lists might be the most appropriate. We accept, of course, that using basic vocabulary lists will not be trouble-free: there are always going to be problems of translatability, culture specificity, determining procedures for choosing between alternative possibilities and so on (Heggarty et al., 2003; Heggarty, forthcoming); but at least we can hope to establish that we are working with the best meaning lists available, within the envelope created by these problems. One initial hypothesis we might test is the common assumption that the 100-meaning Swadesh list is more universal, more basic and more resistant to borrowing than the 200-meaning list. We tested these two standard lists, to see whether they contained the same proportions of loans, working with data from Kessler (2001), who provides a great deal of helpful information on this matter, although operating with rather different metrics from those developed here. We selected from Kessler’s data five languages with fairly high levels of borrowing: Albanian, French, English, German and Turkish. We then calculated the number of loans (which Kessler marks very carefully in his data) for both the 100- and 200-meaning lists. Across these five languages, we found an overall proportion of 12.3 per cent borrowed items. This in itself is quite high, but in addition it is not evenly distributed: there is more borrowing in the items in the 200-meaning list that are not included in the 100-meaning list (roughly 15.7 per cent versus 8.6 per cent), and this difference is highly statistically significant (χ2 = 10.7; p < 0.001, with one degree of freedom). Note that these results are congruent with the work of Wang and his associates (Wang and Wang, forthcoming), who argue that the Swadesh 100-meaning list is preferable to the 200-meaning list for classifying Chinese dialects. Wang and Wang (2004: 1) base this argument on the assumption that the shorter list contains more ‘high rank words’, and is less susceptible to borrowing, though they do not undertake statistical testing to validate their hypothesis. Our test therefore goes some way towards supporting their approach, though we shall argue below that subdividing the lists offers additional analytical advantages over using any list in its totality. It follows that, if we were interested above all in ‘purifying’ the data, we would be constrained to using the 100-meaning list. None the less, even the 100-meaning list contains a certain proportion of loans; and there is also another argument, well set out by Embleton (1986), to the effect that it is
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
3.2 Testing and subdividing Swadesh lists
Language Classification and Contact-induced Change
better to use more data points than fewer. In particular, Embleton shows that there are clear statistical advantages in using 200 items rather than 100; a list with 500 items brings problems of its own, which arguably outweigh the benefits. We have therefore opted for the 200-meaning list, even with its greater propensity to borrowing – and in fact we can use this tendency to our advantage, because this longer list can be subdivided in helpful ways that might allow us to diagnose even unsuspected loans. In recent work (McMahon and McMahon, 2003, 2004), we have reported the application of a series of quantitative methods to the substantial database of Dyen et al. (1992), which provides lists and cognacy judgements for a standard 200-meaning list, in 95 Indo-European languages and dialects. We deliberately selected an existing database, and in the early stages did not check the data in any way; by accepting the dataset intact, complete with any possible errors and incompatibilities, we left the possibility open that these very errors might be uncovered by our methods. Rather than using the 200-meaning list as a whole, however, we have argued that subdividing the list according to the retentiveness and universality of the meanings provides a method that can identify many known loans. The idea of isolating a very conservative sublist is not a new one, and has been proposed by both Lohr (1999) and by Starostin (see Baxter and Manaster-Ramer, 2000), using rather different techniques. However, although Starostin and Lohr both established very conservative sublists, they then discarded the remainder of the list. We have used two contrasting sublists, derived on the basis of Lohr’s (1999) methodology, which itself relies on assessments of relative universality and retentiveness. These sublists are a highly conservative list of thirty items, and a much more changeable list of twenty-three items from the other end of the spectrum, and are shown below: (1)
(2)
Thirty most conservative meanings: four name three foot to give long sun other to sleep day to eat not five I ear new night one to stand star thou tooth wind
two salt to come thin mother to spit tongue
Twenty-three least conservative meanings: grass mouth stone year bird near wing man neck to walk back to flow to pull to push river straight to think to throw
heavy smooth tail left (hand) rope
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
58
59
We have combined this approach with the use of tree-drawing programs (Felsenstein, 2001), which generate all possible trees, and select the best. In cases where there is no borrowing, trees drawn on the basis of the most conservative and least conservative sublists will give highly similar results; but if there is borrowing, the results tend to be strikingly different. For example, using the least conservative data, which are more prone to change and therefore more likely to be affected by borrowing, Frisian gravitates towards Dutch relative to its more distant position in the more conservative tree. In the same way, English falls within West Germanic in the more conservative tree, but lies outside West Germanic and closer to North Germanic in trees derived from the more changeable data. However, there is an apparent paradox here. We do know that Frisian has borrowed extensively from Dutch, and that English equally has a range of loans from North Germanic. On the other hand, these loans are well-known: as we have already seen, Embleton (1986) and Kessler (2001) provide lists of loans in Indo-European. Dyen et al. (1992) generally follow the practice of marking loans with a special code in their database: there will be a particular numerical code to mark cognates, and a different number in cases of known loans, and the loans will then typically be excluded automatically from the data used in drawing our computational trees. Yet we still find that certain languages, such as Frisian and English, appear in different positions in the trees produced on the basis of the most conservative and the least conservative sublists. It follows that, if these shifts are being caused by loans, those loans must be undiagnosed, rather than being marked as loans in the underlying database. We have returned to the Dyen et al. (1992) data, and checked their codings for all the Germanic forms listed as loans by Embleton (1986). At least six of the items identified by Embleton as involving borrowing in Germanic occur in our set of twenty-three least conservative meanings: these are ‘wing’, ‘left (hand)’, ‘to pull’, ‘to push’, ‘river’, and ‘to throw’; but none at all are in the sublist of thirty highly retentive, highly reconstructible meanings. Second, and potentially much more important, Dyen et al. (1992) have wrongly coded a small number of loans as being cognates, so that these are treated as cases of common inheritance and not excluded from the data used for tree drawing. For example, four loans from Dutch into Frisian are wrongly coded as cognates, including ‘left (hand)’ and ‘river’ from the least conservative sublist; and similarly, nine loans from Danish into English are wrongly coded as cognates, including ‘wing’ from the same sublist. Very small numbers of loans can therefore produce differences between trees of this kind, and – if detected – provide signals of borrowing. Our methods therefore appear to allow us to diagnose loans; and we are fortunate in this case to have independent assessments (from Embleton (1986) and Kessler (2001), for example) of which items are in fact borrowed, and of the source of those loans. This allows us to confirm our findings, and to have some
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon and Robert McMahon
60
Language Classification and Contact-induced Change
3.3 Network representations Contrasting trees drawn using more and less conservative meanings, then, provides a promising way of identifying loans in comparative lexical data. This does help to distinguish the two components of Kessler’s (2001) ‘historical connection’: but it cannot be the whole story, because the method as described so far relies entirely on programs that draw family trees. We are therefore demonstrating borrowing using a notation which is crucially not designed to show borrowing, but rather to represent similarities arising from common ancestry and differential descent. In recent work, we have consequently begun to explore alternative computer programs, again from biology: the first program of this kind is called Network (Bandelt et al., 1995; Bandelt et al., 1999; Forster et al., 2001), and was designed to show complex relationships between molecular haplotypes. The particularly innovative aspect of Network, and the aspect that makes it strikingly appropriate for work on ‘historically connected’ languages, is that the program draws the best available tree in cases where relationships between populations or languages are clear and tree-like, but a more complex network when the connections are more complex and show more interaction. The output will resemble a star diagram, or unrooted tree, but the more network-like examples will also include boxes, or reticulations, linking groups. In genetics, these reticulations will show either that a similar mutation has arisen in two groups by chance, or that there has been gene-flow between the populations; one plausible linguistic interpretation would be that borrowing has taken place. One network drawn using the most conservative sublist shown in Section 3.2 above, for Romance and Germanic, is shown in Figure 3.1 (some Slavic languages are included at the far left, but will not be discussed here). These data clearly generate a standard tree, albeit with some reticulations at the root: these can be interpreted as indicating some lack of certainty over the relative order of splitting among the first-order branches. There is a single reticulation within Romance, suggesting that Provençal is connected with both French and Italian. There are no reticulations within Germanic, where the North and West subgroups are differentiated. Within West Germanic, the large node labelled German in fact collapses German and English, while the node marked Flemish also includes Dutch and Afrikaans, since the highly conservative items in this sublist are all cognate for the members of these groups, and therefore do not allow differentiation.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
confidence in extending the analysis in future to cases where we do not know that loans are involved. We may also in future be able to show exactly which meanings are being affected, by removing individual meanings from the sublists and re-running the tree-drawing programs, to check the effect on the relative position of the languages in question.
April McMahon and Robert McMahon
61
Riksmal
Danish Swedish
Icelandic Pennsylvania German German Flemish
Brazilian Walloon
Portuguese Spanish Italian Romanian
Provençal Catalan
French
Ladin Sardinian
Vlach
Figure 3.1 Output of Network for most conservative sublist, Romance and Germanic
For comparison, Figure 3.2 shows the output of Network for the least conservative sublist shown in Section 3.2 above. Here, Romance is the group at the bottom right, and Germanic appears at the bottom left. Clearly, Network, when run on the least conservative sublist, produces a series of reticulations at the root for Romance, showing connections between the languages: this might be interpreted as suggesting a dialect continuum situation early in the history of the group. Although there are no reticulations for Germanic, we do see in Figure 3.2 a very clear indication of contact, since English has shifted altogether into North Germanic, though it was located squarely in West Germanic in Figure 3.1. Network consequently appears to be even more sensitive to borrowing than the tree-drawing programs presented earlier. In part, this greater sensitivity results from the operation of the programs themselves. The individual items in each sublist are averaged to give a composite distance between languages in the tree-drawing programs, all of which are distance-based. For Network, however, each meaning in each sublist is treated as a separate system with multiple states, so that the effect of a borrowed datapoint will be much more obvious, since it is not smoothed out by the general pattern in the rest of the sublist. However, there is a further
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Frisian
62
Language Classification and Contact-induced Change
Sardinian Swedish Sardinian Catalan Vlach Pro Romanian ven Ladin çal Fren ch Portuguese Walloon Brazilian
English Danish
Italian
German
Riksmal Dutch Pennsylvania German
Flemish Afrikaans
Spanish
French Creole
Frisian
Figure 3.2 Output of Network for least conservative sublist, Romance and Germanic
advantage to using Network, since the program does not only provide a diagram, but also a list of the problems it has encountered in producing that output. In other words, Network constructs a diagram that may not quite be a tree, but also identifies the items for which the data are not consistent with a tree-like structure. It is therefore possible to examine that list and thereby identify the items that are responsible for reticulations or differences between outputs of Network for different datasets (like our more and less conservative sublists); these lists are likely to contain borrowings. This is a considerably less labour-intensive approach, since it uses by-products of the operation of the program itself, than the alternative method of removing single meanings sequentially and re-running tree-drawing programs to see whether a difference in tree structure arises. As noted above, we know from previous work by Embleton (1986) and Kessler (2001), among others, that there are more borrowings in our least conservative sublist than in our most conservative meanings. We have also demonstrated that in the Dyen et al. (1992) database, a small number of these borrowings are erroneously coded as cognates, and are therefore included in the analysis rather than being factored out as unique states. To give a preliminary indication that Network does indeed identify these
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Icelandic Faroese
63
erroneously coded items, which we know independently to be loans, let us consider the single example of ‘wing’. Dyen et al. (1992) have coded this item as a cognate between English and North Germanic. If we retain this coding, and run Network, we find the output shown in Figure 3.3, which expands the relevant portion of Figure 3.2, showing English as a North Germanic language. This run of Network, as usual, generates a list of problem items which cannot easily be reconciled with a single tree representation, and this list contains ‘wing’. However, if we alter the Dyen et al. coding and mark ‘wing’ as a loan into English, the result is as shown in Figure 3.4. Here,
Icelandic Faroese
Swedish
English
German Danish Riksmal
Dutch
Pennsylvania German
Flemish Afrikaans
Frisian
Figure 3.3 Network for Germanic, ‘wing’ coded as cognate English–North Germanic
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon and Robert McMahon
64
Language Classification and Contact-induced Change
Pennsylvania German
Icelandic
Riksmal
Swedish German
Dutch Flemish
Afrikaans
Frisian
English
Figure 3.4 Network for Germanic, ‘wing’ coded as loan into English
English no longer appears inside the North Germanic group. It is true that it is not within West Germanic either, but we ascribe this to two factors. The first is that there may well be other loans in the material, since we have removed only ‘wing’ by recoding here, for the purposes of illustration. The impact of this single change in coding may well be strong enough to remove English from North Germanic, but at the same time too weak to relocate it definitively in West Germanic. The second, and perhaps more important, factor is that recoding ‘wing’ as a unique state for English alone, a choice that, again, we made here for the purposes of illustration, serves to distance English from all the other members of the Germanic group. A full analysis would, of course, require a unified and consistent account of all borrowings among all Germanic languages; our purpose in treating this item and this language individually here is simply to demonstrate that recoding of a single meaning does have an effect on the output of Network, and that Network itself identifies that particular problem feature in the course of its computations.
4
Networks and convergence
The cases discussed above indicate the power and applicability of these methods; but they have been limited in that we have considered only
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Faroese Danish
65
unidirectional, lexical borrowing. However, this volume focuses on situations of convergence, where contact-induced change is often more prominent, and more complex, in two ways. First, and most obviously, convergence typically has its main effects outside the lexicon; and second, whether the borrowing involves lexical items or non-lexical structures, the direction of borrowing may be unclear and is generally not unidirectional. The question is whether network-type methods can be applied to situations of this kind. The programs we are exploring are relatively new and have only recently been applied to linguistic data, so that the illustrations we shall give are only initial indications of the contributions that new methods of comparison and representation might make in the future. Furthermore, methods for comparison of data outside standard lexical lists are in the initial stages of development. A number of methods for phonetic comparison have been proposed (see Kessler, forthcoming; McMahon et al., forthcoming), but all are still being tested and refined. The situation is even more difficult for morphosyntax, the area of the grammar often claimed to be affected in cases of convergence; while in phonetics we do at least have universal frameworks of articulation and transcription in which to embed our comparisons, it is by no means clear how we should construct grammatical comparisons, or even what we should be comparing. In the rest of this chapter, then, we shall focus on the directionality problem rather than extending our approach to non-lexical contact. In view of the hazy boundary between convergence and other types of contact-induced change, it is certainly relevant to consider general approaches to contact that will overlap with convergence situations, rather than designing methods specifically for convergence – especially given current questions over whether convergence should be recognized as a separate category of change at all (see Campbell, Chapter 1 in this volume). The examples of lexical borrowing considered above were essentially unidirectional. True enough, there have been, and are, loans both from English into French, and from French into English, though our identified loans above were only from the latter category. However, borrowings from English into French are typically not found, at least to the same extent, in the basic vocabulary of a standard Swadesh list; and it is generally possible, given the plentiful early attestations of both languages and the detail we have about their histories of sound changes, to tell in which direction a given lexical item has travelled. However, in other language groupings, the direction of lexical borrowing is nothing like so clear-cut, with productive interborrowing being the norm. This is true even for some groups within Indo-European: notably the Indo-Iranian languages have borrowed extensively from one another, typically bilaterally. Again, this kind of contact-induced change lends itself best to analysis using network-type programs, which can accommodate the effects of contact as well as differentiation from a common ancestor, and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
April McMahon and Robert McMahon
Language Classification and Contact-induced Change
which do not impose a tree on non-tree-like data. In this case, we have used NeighbourNet (Bryant and Moulton, 2004). Whereas Network itself is character-based, NeighbourNet is designed to work with distance data, and therefore calculates connections between languages on the basis of the composite distance between them for all elements compared taken together. This allows greater flexibility, and avoids the problem that Network can effectively be sidetracked by the presence of only one or two minor connections, which can take up a prominence in diagrams out of all proportion to their real historical and linguistic relevance. The resolution at which NeighbourNet operates seems well suited to language data (see also McMahon and McMahon, 2006), neither missing important links nor producing so many reticulations that the signal becomes lost in the noise. Two NeighbourNet diagrams including the Indo-Iranian languages from the Dyen et al. (1992) database are shown in Figures 3.5 and 3.6. Figure 3.5 is generated from the hihi sublist, and therefore includes only the most conservative meanings. In contrast, Figure 3.6 shows a NeighbourNet for the same languages, but this time based only on the more changeable, lolo dataset. These figures show a rather unusual pattern, at odds with the fairly
GreekMod IrishA
EnglishST Vlach
GermanST Italian
Waziri
SwedishUp Russian Czech
Afghan
Baluchi Tadzik PersianLt
Kashmiri
Ossetic
Wakhi Khaskura Punjabi GypsyGk
Bengali Hindi Singhalese
Lahnda
NepaliLst Gujarati Marathi
Figure 3.5 NeighbourNet for Indo-Iranian hihi sublist
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
66
April McMahon and Robert McMahon
GreekMod Vlach
IrishA
67
Italian EnglishST SwedishUp
Ossetic Waziri Afghan
Czech PersianLt
Russian
Tadzik Singhalese
Baluchi Wakhi Khaskura
GypsyGk Kashmiri
Nepali Lahnda PunjabiST Hindi
Gujarati
Marathi Bengali
Figure 3.6 NeighbourNet (Bryant and Moulton, 2004) for Indo-Iranian lolo data
consistent representations we have generated for the complete IndoEuropean dataset; for simulated data; and indeed also for hihi versus lolo sublists for a group of Andean languages (McMahon et al., forthcoming). All analyses so far show that, when we are comparing languages we know to be related, the hihi data produces a more tree-like signal. For the same languages, networks constructed from the lolo data have more reticulations and signs of inconsistency. Certainly, in the lolo network in Figure 3.6, we do find signals of probable contact between, for example, Punjabi, Hindi, Bengali and Marathi; but overall, and very unusually, the lolo diagram is rather more tree-like in outline than is the equivalent hihi diagram in Figure 3.5. There are two possibilities here for future investigation. It may be that this is the kind of signal we shall find in other cases where there has been multidirectional lexical borrowing among geographically close languages. On the other hand, these patterns may be a result of the speed of the initial split for some of these languages: the more conservative dataset, by its nature, takes longer to diverge, and hence to produce a structured outcome, than the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
GermanST
Language Classification and Contact-induced Change
more changeable one, so that swift diversification along with typical patterns of borrowing might produce such signals. For the moment, however, this must remain speculative. Lexical borrowing, then, seems to create predictable patterns in network diagrams for related languages; but we certainly have to refine and test these expectations, confronting them with situations where borrowing is more extensive, or is not unidirectional, or both. Furthermore, we must consider language groups and linguistic areas beyond Indo-European. In particular, historical linguists typically agree that contact has played a highly significant, and perhaps unusually prominent role in the development of Australian languages, even if they often fail to agree about much else (Dixon, 1980, 2001; Bowern and Koch, 2004; Evans, 2004). Our case study here is a very limited one, but serves the purpose of showing just how important it is to understand the implications and limitations of different computational methods, and to compare the analyses they produce for the same data. Nash (2002) includes very limited lexical lists (based on Alpher and Nash’s (1999) modified Swadesh list, tailored for use in Australia) for twenty-six languages of south-east Western Australia. In Nash’s paper, these languages are often labelled using the name of the speaker who contributed the partial list, or a geographical location. In Figures 3.7–3.9, the twenty-six varieties are referred to as languages ‘a–z’. Linguistic affiliations in this area are particularly unclear; some of these languages may in fact be related dialects, and in the absence of much prior comparative work, Nash’s judgements of whether items are plausibly cognate rely inevitably on recurrent similarity. The first and simplest option is to apply a tree-drawing and tree-selection program to these data. Such programs generate a range of possible trees, and select from these the tree most consistent with the data. In this case, we have used a Neighbour Joining algorithm, which operates by finding the two closest, most similar candidates (here, the two languages that share most plausible cognates), and bringing these together into a node; this process is then repeated until all languages have been incorporated into the tree. Figure 3.7 shows the best tree of this kind for Nash’s Australian data; the tree is drawn and displayed using Splitstree Version 4.0 (Huson, 1998). At first sight, the diagram in Figure 3.7 seems to show a fundamentally tree-like structure, and we might be tempted to draw phylogenetic conclusions from it: so, languages ‘n’ and ‘o’ might be dialects of the same language, or at least closely related, while ‘u’ is a sister of a branch including ‘g’, ‘i’ and ‘j’. However, the problem is that tree-drawing programs are designed to do exactly that: they are constrained to draw trees, regardless of how tree-like the data in fact are. Just because a tree is the best one that can be generated from a particular program does not mean it is necessarily any good. The next step would have to be further statistical checking of the tree, in particular to assess whether the branches are strongly supported; this
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
68
April McMahon and Robert McMahon
0.1
69
d
k o
r
z
y
x p
w
s m
b
t
c
a u
n e
f h
q
g
v i
j
Figure 3.7 Neighbour-joining tree of Australian data, drawn using Splitstree
would usually involve bootstrapping, or stepwise resampling of the data and redrawing of the tree, to calculate how many times per hundred, or per thousand, runs we found a specific branch or language grouping replicated. Arguably, a better option is to apply network programs to Nash’s data, since such programs have options other than straightforward tree drawing, and can therefore indicate from the outset how tree-like the data and relationships are. Figure 3.8 has again been drawn using Splitstree, but this time with a network rather than a tree option. This time, therefore, Splitstree is looking for any phylogenetic signals, but will not draw a tree in cases where it finds inconsistent data that support different connections between the languages. In such cases, Splitstree will group languages together rather than insert splits. It is clear from Figure 3.8 that phylogenetic signals are weak in the Australian data: twenty of the twenty-six languages are grouped together as a single node. One interpretation of this figure would be that there is a fundamentally tree-like pattern here, but that this has been obscured by long-term contact. Again, however, Splitstree has particular shortcomings and is not the optimal program for analysis of such complex data. On the one hand, Splitstree tends to miss tree-like signals when it is dealing with highly complex datasets, meaning its resolution is not ideal for language data. But on the other, it seeks to maximize a tree-like structure and therefore cannot be relied on to indicate all the specific cases of inconsistency, or in our terms contact, for example, by incorporating reticulations.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
l
70
Language Classification and Contact-induced Change
0.01 a, b, y, z, w, v, u, t, r, q, n, l, k, j, i, h, g, f, e, c
p, s, x
m
o Figure 3.8
Splitstree graph for 26 Australian languages
As we saw above, NeighbourNet seems well suited to being applied to language data, and has been used to construct the graph in Figure 3.9. This constitutes a step forward: some vestiges of a tree-like signal seem to be emerging, though the volume of reticulations is still considerable. A program which depicts any underlying tree-like structure while simultaneously noting connections incompatible with the tree, which are likely to signal contact, is the ideal compromise, since it should be applicable both to ‘clean’, phylogenetically informative data, and to contact languages. However, NeighbourNet is flexible, and can detect signals of both common ancestry and contact, purely because it is designed to find and represent affinities of all sorts. What it cannot do directly is to tell us what these different affinities mean. We cannot assume that a diagram of a particular shape corresponds in every case to a history of a particular type, without considerable further testing and interpretation. In part, this involves application of these methods to as many different languages and situations as possible; but, in addition, we must ask linguists with a particular interest in specific languages to assess the features linking languages in NeighbourNet graphs, to determine whether these are good candidates for family membership or for contact. In other words, programs of this kind are not short-cuts that replace work for linguists; but they might offer an opportunity to assess or propose particular hypotheses of relatedness, and to consider the possible meaning and value of specific features on which these hypotheses are built. In this short chapter, we have only been able to address one aspect of convergence, namely the issue of directionality. Our initial investigations
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
d
April McMahon and Robert McMahon
71
d
o z
p
k
I
y
m
f
e n c q
h a
t r v i
g
b
w u
j
Figure 3.9 NeighbourNet graph for 26 Australian languages
suggest that computational approaches may reveal distinctive patterns arising from multidirectional borrowing of a type commonly found in linguistic areas: if this finding is robust for other parallel cases, we may be able to use NeighbourNet diagrams in future to suggest whether lexical borrowing has taken place in the histories of particular languages and language groups, and to indicate whether it has been unidirectional or multidirectional. We have also shown that it is particularly important to use the right programs, with tree-drawing programs being intrinsically limited in what they can achieve, since they will draw trees regardless of whether these are the best representation of the data. Network approaches are much better suited to the complexities of language data; and of these, NeighbourNet currently seems a strong candidate for future development. One particular characteristic of NeighbourNet means that it should also be appropriate for the work on non-lexical comparison that is now beginning.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
s x
Language Classification and Contact-induced Change
NeighbourNet is distance-based rather than character-based; that is, diagrams are drawn on the basis of the entire set of data under comparison, which are put together to produce a summative distance matrix. This also makes NeighbourNet a phenetic, rather than a cladistic approach (McMahon and McMahon, forthcoming) – that is, it works with measures of distance, or of similarity, whatever these may signify; it does not prioritize similarities that indicate common ancestry over those resulting from contact, parallel development or just plain chance. When we turn to phonetic or morphosyntactic comparisons, we are also dealing with similarities, without prejudice to their historical origin or meaning; indeed, the frequently directional nature of sound change makes it much more likely that parallel changes will take place in unrelated languages here than in the case of the lexicon, where we have the further possibility of converting our similarity calculations into a cladistic, historical dimension by introducing judgements of whether we have plausible cognates across languages. Phenetic, resemblance-driven methods such as NeighbourNet are therefore ideally suited to the next phase of development of these quantitative approaches, when we turn to areas of the grammar beyond the lexicon, and therefore to other aspects of the problem of convergence. To return briefly to the developments in nineteenth-century historical linguistics with which we began this chapter, Morpurgo Davies (1998: 141) discusses Grimm’s work and his various preoccupations, and notes that ‘For Grimm the main task is that of exploiting the new achievements of linguistics to look at the old problem.’ She makes this point in the context of Grimm’s work on the origin of language, not in connection with any of his betterknown historical findings, but we hope it is not too presumptuous to borrow this description for our own attempts to exploit new achievements in linguistics (and indeed, in other disciplines), to look at the old problems of contact and language classification. It is important to stress, however, that this is still crucially part of linguistics: in other words, these programs are not intended to be a substitute for linguistic expertise, but rather a guide to its more efficient and directed application. There is a great deal of work here for the future, and we have only tried in this chapter to provide an initial indication of the advances we might expect if linguists and colleagues from other disciplines can work together in developing quantitative methods. For these future hopes to be realized, however, we must above all take the issues raised by contact-induced change very seriously. Contact cannot be ignored: it is both too pervasive and too unpredictable for that to be an option. But more surprisingly, and more intriguingly, it can benefit us considerably to include contact in our computations. In this way, we achieve a stronger and fuller understanding of the whole histories of the languages concerned by identifying loans and their sources computationally, but can also use cases where borrowings have been misidentified in the past to test and demonstrate our methods. If such
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
72
April McMahon and Robert McMahon
73
Note 1 The work reported here is funded by Arts and Humanities Research Board (AHRB) grant AN6720/APN 12536 to the project ‘Quantitative Methods in Language Classification’. We record here our sincere thanks to the AHRB for making this collaborative work possible, and to our co-workers on the project, Paul Heggarty and Natalia Slaska. We also thank reviewers of this chapter for their helpful suggestions which have greatly improved expression and coverage.
References Alpher, B. and D. Nash (1999) ‘Lexical Replacement and Cognate Equilibrium in Australia’, Australian Journal of Linguistics, vol. 19, pp. 5–56. Bandelt, Hans-Jürgen, Peter Forster, Bryan Sykes and Martin Richards (1995) ‘Mitochondrial Portraits of Human Populations Using Median Networks’, Genetics, vol. 141, pp. 743–53. Bandelt, Hans-Jürgen, Peter Forster and A. Röhl (1999) ‘Median-Joining Networks for Inferring Intraspecific Phylogenies’, Molecular Biology and Evolution, vol. 16, pp. 37–48. Baxter, William and Alexis Manaster-Ramer (2000) ‘Beyond Lumping and Splitting: Probabilistic Issues in Historical Linguistics’, in Colin Renfrew, April McMahon and Larry Trask (eds), Time Depth in Historical Linguistics (Cambridge: McDonald Institute for Archaeological Research), pp. 167–88. Bowern, Claire and Harold Koch (eds) (2004) Australian Languages: Classification and the Comparative Method (Amsterdam: John Benjamins). Bryant, D. and V. Moulton (2004) ‘Neighor-Net: An Agglomerative Method for the Construction of Phylogenetic Networks’, Molecular Biology and Evolution, vol. 21, pp. 255–65. Dixon, R. M. W. (1980) The Languages of Australia (Cambridge: Cambridge University Press). Dixon, R. M. W. (2001) ‘The Australian Linguistic Area’, in A. Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Linguistics (Cambridge: Cambridge University Press), pp. 64–103. Dyen, Isidore, Joseph B. Kruskal and Paul Black (1992) ‘An Indo-European Classification: A Lexicostatistical Experiment’, Transactions of the American Philosophical Society 82, Part 5. Data available at http://www.ldc.upenn.edu. Embleton, Sheila (1986) Statistics in Historical Linguistics (Bochum: Brockmeyer). Evans, Nicholas (ed.) (2004) The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Canberra: Pacific Linguistics). Felsenstein, J. (2001) PHYLIP: Phylogeny Inference Package. Version 3.6 (Seattle: Department of Genetics, University of Washington).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
demonstrations are found to be convincing, we can hope that linguists will come to accept these quantitative methods also in resolving those cases where linguistic opinion differs persistently, and the data are either sparse or unclear.
Language Classification and Contact-induced Change
Forster, Peter, Antonio Torroni, Colin Renfrew and A. Röhl (2001) ‘Phylogenetic Star Construction Applied to Asian and Papuan mtDNA Evolution’, Molecular Biology and Evolution, vol. 18, pp. 1864–81. Heggarty, Paul, April McMahon, Robert McMahon and Natalia Slaska (2003) ‘Lexicostatistics: The Flaws and the Fixes?’, Paper presented at the 16th International Conference on Historical Linguistics, Copenhagen, 11–15 August 2003. Huson, D. (1998) ‘SplitsTree: A Program for Analyzing and Visualizing Evolutionary Data’, Bioinformatics, vol. 14, pp. 68–73. Kessler, Brett (2001) The Significance of Word Lists (Stanford, Calif.: CSLI Publications). Kessler, Brett (forthcoming) ‘Phonetic Comparison Algorithms’, Transactions of the Philological Society. Koerner, Konrad (ed.) (1983) Linguistics and Evolutionary Theory: Three essays by August Schleicher, Ernst Haeckel, and Wilhelm Bleek, with Introduction by J. Peter Maher (Amsterdam: John Benjamins). Lohr, Marisa (1999) Methods for the Genetic Classification of Languages. Unpublished Ph.D. thesis, University of Cambridge. McMahon, April and Robert McMahon (2003) ‘Finding Families: Quantitative Methods in Language Classification’, Transactions of the Philological Society, vol. 101, pp. 7–55. McMahon, April and Robert McMahon (2004) ‘Family Values’, in Christian Kay, Simon Horobin and Jeremy Smith (eds), New Perspectives on English Historical Linguistics, vol. 1: Syntax and Morphology (Amsterdam: Benjamins), pp. 103–32. McMahon, April and Robert McMahon (forthcoming) Language Classification by Numbers (Oxford: Oxford University Press). McMahon, April and Robert McMahon (in press) ‘Cladistics’, in Keith Brown (ed.), The Encyclopedia of Language and Linguistics, 2nd edn (Oxford: Elsevier). McMahon, April, Paul Heggarty, Robert McMahon and Natalia Slaska (forthcoming) ‘Swadesh Sublists and the Benefits of Borrowing: An Andean Case Study’, Transactions of the Philological Society, 2005. Morpurgo Davies, Anna (1998) Nineteenth Century Linguistics. London: Longman. Nash, David (2002) ‘Historical Linguistic Geography of South-East Western Australia’, in John Henderson and David Nash (eds) Language in Native Title (Canberra: Aboriginal Studies Press), pp. 205–30. Ringe, Don, Tandy Warnow and Ann Taylor (2002) ‘Indo-European and computational cladistics’, Transactions of the Philological Society, vol. 100, pp. 59–129. Schleicher, August (1863) Die Darwinische Theorie und die Sprachwissenschaft. Offenes Sendschreiben an Herrn Dr. Ernst Haeckel, o. Professor der Zoologie und Direktor des zoologischen Museums an der Universität Jena (Weimar: Böhlau). Schmidt, Johannes (1872) Die Verwantschaftsverhältnisse der indogermanischen Sprachen (Weimar: Böhlau). Swadesh, Morris (1952) ‘Lexico-Statistical Dating of Prehistoric Ethnic Contacts’, in Proceedings of the American Philosophical Society, vol. 96, pp. 452–63. Swadesh, Morris (1955) ‘Towards Greater Accuracy in Lexicostatistical Dating’, International Journal of American Linguistics, vol. 21, pp. 121–37. Thomason, Sarah Grey (2001) Language Contact: An Introduction (Edinburgh: Edinburgh University Press). Wang, Feng and William S.-Y. Wang (2004) ‘Basic Words and Language Evolution’, in Language and Linguistics, vol. 5, no. 3, pp. 643–62.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
74
Linguistic Areas, Language Contact and Typology: Some Implications from the Case of Ethiopia as a Linguistic Area* Walter Bisang
1
Introduction
The concept of linguistic area or Sprachbund has triggered many discussions among linguists dealing with language contact. Even though various suggestions for its exact definition and numerous papers dealing with the question of whether a certain geographic area really is a linguistic area have been published in the years since the introduction of that term by Trubetzkoy (1930) it still remains unclear what is in fact a linguistic area. As I would like to show in this chapter, this does not come as a surprise if one starts looking more closely at the problems inherent in this concept, which is based on an idealization that takes for granted too much of structural and social homogeneity across potential linguistic areas. In fact, almost all of the criteria discussed in the literature for a more clear-cut and homogeneous definition of what makes a linguistic area turn out to be either arbitrary or difficult to apply. In addition, the correlations between structural changes and their potential social and historical background seem to be of such a general and rather abstract nature that too much rigour in defining linguistic areas may obstruct the finding of potential regularities from the outset. It is for reasons like these that I shall introduce the concept of zones of contact-induced structural convergence (abbrevated to zones of convergence) as a more open concept comparable to geographic dialect continua (see section 3.3). This chapter thus shares Campbell’s (Chapter 1 in this volume) scepticism about definitions of linguistic areas, but it proceeds on the assumption that a less rigid concept in terms of zones of convergence is needed for typological reasons (see below) and for integrating sociohistorical facts with the geographic diffusion of certain features (see section 3.3 in this chapter). Of course, borrowing is always part of what is going on in zones of convergence, 75 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4
Ethiopia as a Linguistic Area
but borrowing alone does not account for clusters of certain features across a particular geographical area. The point of departure for this chapter is the Ethiopian linguistic area1 and the controversial discussion of whether it exists (see Greenberg, 1959; Ferguson, 1970; Zaborski, 1991; Tosco, 1994a, 1994b; and Tosco, 2000; versus Thomason, 2001). It will be shown in Section 2 that no matter how controversially one may discuss the question of whether there is a linguistic area at the Horn of Africa, there can be no doubt about the highly complex multilayered texture of contact situations that make this area into a zone of contact-induced structural convergence. Since it will turn out from Section 3 that it is systematically impossible to provide a clear-cut definition of a linguistic area, the question of whether the Ethiopian zone of convergence is a linguistic area is no longer of any relevance. What matters is a detailed analysis of processes and criteria operating in concrete situations of language contact, and a better understanding of how their consequences – that is, linguistic features – diffuse into larger geographic areas in terms of zones of convergence. Some of these features and criteria are introduced in subsection 2.2 for the Ethiopian zone of convergence. As will be shown, some of them are often excluded explicitly from the discussion on linguistic areas. Zones of convergence are not only important for understanding mechanisms of contact-induced change; they are also relevant for typological research. On the one hand, typologists need to know about them in order to avoid language samples that are statistically biased because of contact-induced convergence among the languages selected (geographical closeness; on problems with this approach see Maslova, 2000; Bisang, 2004). On the other hand, zones of structural convergence may give rise to the emergence of new or rarely attested linguistic types. The latter topic will be discussed in Section 4, on the basis of two examples from Ethiopia. Examples like these are vital if it comes to the status of generalizations on the linguistic potential of humans. If language contact can generate new or rarely attested types, and if zones of convergence can propagate them over an extended geographic area, language contact must be seen as a rather hard explanatory criterion for the structures that are cross-linguistically attested. Under these circumstances, language contact must be considered in conclusions concerning universals in functional linguistics as well as in assumptions concerning universal grammar in Chomsky’s formal approach (see outlook and conclusion in Section 5).
2 Structural convergence in Ethiopia – is there an Ethiopian linguistic area? 2.1 The languages/language families involved and the controversy of whether there is an Ethiopian linguistic area The Ethiopian linguistic area or Sprachbund as it is discussed in the literature (Greenberg, 1959; Ferguson, 1970, 1976; Zaborski, 1991; Crass and Bisang,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
76
2004) consists of some eighty languages spoken in Ethiopia and Eritrea, plus some languages from Sudan (Beja), Djibouti (Afar, Somali) and Somalia (Somali). If one looks at the languages involved from their geographical distribution a description such as ‘the linguistic area at the Horn of Africa’ would seem to be more accurate. In spite of this, I will follow the general practice of referring to this area as the ‘Ethiopian linguistic area’ or the ‘Ethiopian Sprachbund’ when I discuss Sprachbund phenomena at the Horn of Africa. The language families involved in this area are Afro-Asiatic and Nilo-Saharan. The branches of Afro-Asiatic relevant to the Ethiopian linguistic area are Ethio-Semitic (Amharic, Gurage, Tigre, Tigrinya and so on; for a more detailed classification see Hetzron, 1975, 1997); Cushitic (primarily Agaw, Highland East Cushitic; some regional influence from Afar-Saho and Oromo); and Omotic, whose status in Afro-Asiatic still seems to be controversial.2 The Nilo-Saharan languages involved belong to several branches depending on the geographic area to be analysed. As was pointed out by Zaborski (1991) that the rather extensive area covered by the Ethiopian linguistic area must be divided into several smaller zones of contact of different intensity between speakers of different languages. He suggests six sub-areas that may again be subdivided into even smaller contact zones. The first to postulate the existence of an Ethiopian linguistic area was Greenberg (1959), who describes this area as ‘marked by relatively complex consonantal systems, including glottalized sounds, absence of tone, word order of determined followed by determiner [sic], closed syllables, and some characteristic idioms’. In two earlier papers, Leslau (1945, 1952) already pointed out a considerable number of structural similarities between Ethio-Semitic and Cushitic languages in Ethiopia, with the aim of showing that the common features are related to the influence of Cushitic on Ethio-Semitic. Since these publications, the vast majority of researchers dealing with languages from this area in some way adopted the idea of a linguistic area. One of its recent supporters is Thomason (2001) (see also Thomason and Kaufman, 1988). Thomason tentatively combines the concept of a linguistic area with her concept of language shift. While the concept of a linguistic area is based on findings from Greenberg (1959), Ferguson (1976) and Zaborski (1991), Thomason’s evidence for shift-induced change is based mainly on two rather old papers by Leslau (1945, 1952), who postulates unilateral Cushitic influence on a number of structural properties of Ethio-Semitic. The following quotation provides a good summary of Thomason’s view: The Ethiopian linguistic area follows the pattern . . . in the Balkans and the Baltic: areal features with differential spread and different detailed linguistic patterns in the various languages. But unlike the Balkans and the Baltic (as far as we can tell about those cases), the Ethiopian area arose primarily through imperfect learning, during a process of group language shift from Cushitic to Semitic. Of course this picture might change once
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 77
78
Ethiopia as a Linguistic Area
Before Tosco (2000), the existence of the Ethiopian linguistic area was a matter of general agreement. While Tosco himself subscribed to the Ethiopian linguistic area in some earlier papers (1994a, 1994b),3 his new look at this phenomenon from the perspective of Aikhenvald and Dixon’s (2001) approach leads him to the conclusion that such a phenomenon does not exist in Ethiopia. In terms of Aikhenvald and Dixon (2001), linguistic areas are the result of a rare type of contact with no social dominance involved: linguistic areas [arose] as the result of equilibrium situations [and] involve long-term language contact with multilateral diffusion and without any relationship of dominance. In contrast, areas which became established as a result of sudden migrations or other punctuations tend to involve dominance of one group over other(s), and the diffusion is often unilateral. (Aikhenvald and Dixon, 2001; quoted as Aikhenvald and Dixon, forthcoming, pp. 17–18 in Tosco, 2000). Tosco (2000) presents three main arguments against the existence of an Ethiopian linguistic area: • If there is a unilateral direction of diffusion from Cushitic to Ethio-Semitic (Amharic) as suggested by Leslau (1945, 1952) and Thomason (2001), the social situation responsible for such an asymmetry cannot be one of equilibrium and thus there is no language area in terms of Aikhenvald and Dixon (2001). • If the languages involved in a language area are supposed to be genetically unrelated, the status of the Ethiopian language area as a whole is rather dubious, because the large majority of its languages belong either to the Semitic or the Cushitic branch of one and the same family – that is, Afro-Asiatic. • Finally, the features of a linguistic area should not occur in the languages that belong to the same language families but are outside that linguistic area. Since this is not always the case, this is a third argument against the existence of an Ethiopian language area. As I pointed out (together with Crass) in Crass and Bisang (2004), Tosco’s (2000) arguments are not as strong as they might seem at first glance. The diffusion of structural features is not necessarily unilateral, as postulated in
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
more is known of the occurrence or nonoccurrence of some of these features in Omotic and especially in non-Afro-Asiatic languages of the Ethiopian highlands – and, for that matter, once more is known about the linguistic makeup of Proto-Afro-Asiatic. But for the features listed here, judging by what we know now, the process of spread was uniform and unidirectional. (Thomason, 2001: 113)
the first of the above arguments – that is, there are cases in which Ethio-Semitic influenced Cushitic. The scenario of an overall asymmetric shift-induced interference as suggested by Thomason (2001) is rather unlikely if one looks at the social and linguistic complexity attested at the Horn of Africa. In fact, the language situation in that area seems to be more complex than in the Balkans or the Baltic (pace the above quotation from Thomason, 2001). A more detailed look at Ethiopia also reveals that genetic relatedness as the second of the above arguments turns out to be no a priori argument against a linguistic area. If genetic relatedness is not of absolute relevance, the third argument concerning the presence of areal features outside the linguistic area is also weakened. Finally, the case of Ethiopia also shows that even the absence of a certain feature can be taken as an indicator of contact-induced change. In Section 2.2, I shall discuss some of the features mentioned in the discussion of Ethiopia as a linguistic area. Section 2.3 will present this region as a highly complex area of multi-layered contact situations responsible for structural convergence. 2.2 Contact-induced features 2.2.1 Survey of the features involved Ferguson (1970, 1976) further elaborates on Greenberg (1959) and his claim of an Ethiopian linguistic area (see section 2.1 above). Ferguson (1976) discusses the Ethiopian linguistic area in terms of eighteen features. Zaborski (1991) rightly criticizes these features, accepting only the phonological feature of glottalization (plus palatalization to some extent) plus the syntactic features set out in Table 4.1. For the purpose of this chapter, I shall briefly discuss the features of glottalized consonants (section 2.2.2), pharyngeal fricatives (section 2.2.3), verb-final word order (point 1; section 2.2.4), adpositions (point 6; section 2.2.5) and converbs (point 5; section 2.2.6). The feature of glottalized consonants illustrates how a genetically determined feature can be relevant
Table 4.1 Non-phonological features relevant for the Ethiopian linguistic area (Zaborski, 1991) 1 2 3 4 5 6 7 8
Basic word order: SOV Subordinated clauses occur in front of the matrix clause Adjectives precede their head noun (AdjN) Main verbs precede auxiliaries Use of converbs for expressing co-ordination Postpositions ‘quoting clauses’, i.e. quoted speech marked by a verb of saying (see also Crass et al., 2001) Compound verbs with verbs of saying, living or being
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 79
Ethiopia as a Linguistic Area
for describing contact-induced structural convergence. Pharyngeal fricatives can be used to show that the absence of a certain feature is areally significant. A closer look at verb-final word order and adpositions reveals that the diffusion of features not only took place from Cushitic to Ethio-Semitic, but also in reverse, from Ethio-Semitic to Cushitic, and thus disproves claims of unilateral influence (Leslau, 1945, 1952; Tosco, 2000; Thomason, 2001). Finally, the case of converbs is instructive because it shows that language contact may increase the need to mark a certain function without determining the concrete morphosyntactic realization of that marker. Different strategies may be used for the same purpose. Finally, from a cross-linguistic phenomenology of converbs, the converbs attested in the Ethiopian area may have some properties characteristic of that area. 2.2.2 Glottalized consonants Leslau (1945: 61) explains the existence of glottalized consonants in EthioSemitic as a consequence of Cushitic influence (see also Thomason and Kaufman, 1988, who adopt this view). However, from a more general perspective, the glottalized consonants of Ethio-Semitic languages correspond to pharyngalized or velarized consonants in other Semitic languages (Ehret, 1995; Orel and Stolbova, 1995).4 To Tosco (2000: 342), emphatic consonants are ‘maybe the Afro-Asiatic feature par excellence’. If this is true, one may assume that glottalization does not qualify as a feature for postulating a linguistic area. Nevertheless, Crass (2002) presented good evidence of the areal relevance of this feature in spite of its genetic character. Thus, even though the feature of glottalized consonants seems to be based on a distinctive feature common to the whole family of Afro-Asiatic, it can be used in the context of linguistic areas. The argument of Crass (2002) is based on the fact that in many reconstructions of later stages of Proto-Afro-Asiatic the number of glottalized consonants has become reduced in comparison to Proto-Afro-Asiatic. The reconstructions in Table 4.2 show only three glottalized consonants, one or none, respectively. Through language contact, many of the languages belonging to the subfamilies presented in Table 4.2 reintegrated glottalized consonants during their later development. Thus, one can conclude that some genetically related languages may lose a genetic feature and later reintegrate it through contact with genetically related languages that have preserved this feature (for more details see Crass, 2002). Table 4.2 Number of reconstructed glottalized consonants for later stages of Proto-Afro-Asiatic Proto-East-Cushitic (Sasse, 1979; p. 5): Highland East Cushitic (Hudson 1989; p. 11) Agaw (Appleyard 1984; p. 58)
3 glottalized consonants 1 glottalized consonant No glottalized consonant
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
80
Walter Bisang 81
Pharyngeal fricatives are reconstructed for Proto-Afro-Asiatic but they do not occur in most languages of Ethiopia. The loss of these consonants seems to be a common development of the Ethio-Semitic and the Cushitic languages of this area (for more details, see Crass, 2002; Crass and Bisang, 2004: 180–1). Most languages having glottalized consonants do not have pharyngeal fricatives. It is for this reason that Tosco (2000: 343) suggested a smaller ‘central Ethiopian area’. One type of exception to this pattern are languages with glottalized consonants and pharyngeal fricatives (Tigre, Tigrinya, Harari, Saho, languages of the Dullay Cluster – Amborn et al., 1980). With the exception of the Cushitic languages belonging to the Dullay Cluster that is spoken in the south-west of Ethiopia, the languages that kept pharyngeal fricatives are situated at the Eastern and Northern fringes of the postulated Ethiopian linguistic area. A look at the South-Semitic languages of Yemen and of Tanzania shows that they still have pharyngeal fricatives. While these languages preserved this consonant type, a large number of languages within the Ethiopic zone of convergence no longer have it. In this sense, the non-existence of a genetically determined feature can become relevant for discussing contactinduced convergence. 2.2.4 Verb-final word order (OV) Since Leslau (1945: 73) it is generally assumed that OV in Ethio-Semitic languages is a result of the influence of Cushitic (Thomason and Kaufman, 1988: 131; Campbell, 1994: 1471; Tosco, 2000: 344). This assumption is further corroborated by the fact that the South-Semitic languages in Yemen, which are not in contact with Cushitic, still have VO, whereas the South Cushitic languages in Tanzania, which are again not in contact with Ethio-Semitic, have OV. In spite of this, the above scenario reflects a macro-perspective: on the micro-level of concrete individual contact situations, it seems very likely that Ethio-Semitic languages have also been responsible for the diffusion of OV (see, for example, Bender’s 1979 hypothesis on OV in the Gumuz languages, Nilo-Saharan). 2.2.5 Adpositions The covariation of prepositions with VO and postpositions with OV has been proved statistically (Dryer, 1992). In an area in which OV languages and VO languages are in mutual contact, such as the Ethiopian zone of convergence, we also find the phenomenon of circumpositions as well as OV & Prep and VO & Postp: • Amharic: OV & Prep / Circumpositions • Oromo: VO & Postp / Circumpositions
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
2.2.3 Pharyngeal fricatives
82
Ethiopia as a Linguistic Area
2.2.6 Converbs Converbs are non-finite verbal forms used in adverbial subordination and in sequentialization (event E1 and then event E2). A thorough typological account of converbs is provided in Haspelmath and König (1995) (see also Haspelmath, 1995; Nedjalkov, 1995; Bisang, 1995, 2001b). The following two examples from Amharic illustrate the adverbial function (1) and the sequential function (2) of the converbal form: (1)
Amharic: Converb in adverbial function: Kasa gänzäb särq-o bä-polis täyazä. Kasa money steal-CONV.3.SG.M PREP-police be.arrested.PF.3.SG.M ‘Having stolen money, Kasa was arrested by the police.’
(2)
Amharic (Kapeliuk, 1988: 149) Zäbäññ-off-u wätet-äw eskä afaf zälq-äw guardian-PL-DEF come.out-CONV: 3.PL up.to peak climb-CONV.3.PL tämälles-äw ‘menemyallä dähna näw’ come.back-CONV.3.SG anything there.is.not good COP.PRS.3.SG bel-äw näggär-u-t say-CONV.3.PL say/tell-PF.3.PL-OBJ.3.SG ‘Les gardes sortirent, grimpèrent au sommet, revinrent et lui dirent: “Il n’y a rien, tout va bien”.’
Converbs in Ethio-Semitic languages are generally described as the result of Cushitic influence. Hetzron (1975: 113) assumes that this process of convergence took place very early, at a time before Ethio-Semitic split into different branches. The converbal forms of Amharic are historically based on a verbal noun plus possessive suffixes. Many other Ethio-Semitic languages also have other structures to express finite forms which differ from the Amharic solution. Thus, in many languages of the Gurage Cluster, the suffix-m, which seems to be identical to the focus marker in some languages, is added to finite verb forms. In Chaha, another Gurage language, a verbal form based on the imperative is used in converbal function (Leslau, 1969). From looking at these patterns, it seems that there is a contact-induced need to have an equivalent to converbs in a large number of Ethio-Semitic languages, but the means to express this category vary even within a comparatively
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
The co-occurrence of VO & Postp in Oromo has already been described by Praetorius (1893).5 Nevertheless, textbooks on language contact and on linguistic areas concentrate only on Amharic – that is, on the part of contactinduced change that goes from Cushitic to Ethio-Semitic. The case of Oromo, which proves the relevance of the opposite direction from Ethio-Semitic to Cushitic, has gone largely unnoticed.6
small cluster of dialects/languages, as is the case with Gurage. The fact that contact-induced change is of such prominence in clause combining falls in line with the observation of Matras (1998) that the elements copied in language contact are first those from discourse, and second those from clause combining. In addition to their formal variability, the converb forms of the Ethiopian zone of convergence have some characteristic properties that may distinguish them from converb forms in a cross-linguistic perspective. They only have one or two converbs whose meaning is semantically rather open (there is, for example, no conditional converb). As in Amharic, person marking on the converb is also attested cross-linguistically, but it is by no means necessary. These two characteristics may turn out to be another specific property of the Ethiopian zone of convergence. 2.3 On the characteristics of the Ethiopian zone of convergence What characterizes the Ethiopian zone of convergence is its complexity. There is a large number of small geographic zones of contact in which Ethio-Semitic, Cushitic, Omotic and Nilo-Saharan languages influence each other. These small contact zones further condense into a large zone of convergence which may well be called a linguistic area (but see section 3 below). The dynamics within this large zone is only partially understood. The usual social factors such as social networks (Milroy and Milroy, 1985, on the impact of language use from a network perspective in Ethiopia; see Meyer and Richter, 2003), maxims of linguistic behaviour in terms of the invisible hand (Keller, 1990, 1994) or leaders of language change (Labov, 2001) are certainly involved in the actual geographic diffusion of the relevant language structures. One may also safely assume the existence of continua as we know them from dialectology – that is, continua in which not all the speakers of the different varieties are in fact in contact. The following extract from Appleyard (1989) is an excellent characterization of the Ethiopian zone of convergence: It is perhaps too simplistic a statement to say . . . that the presence of parallel systems in Ethiopian Semitic and Cushitic means that the former must have borrowed from the latter. What has to be taken into account here is that the languages of the horn of Africa, Semitic and Cushitic, share a common development and an infinitely complex network of mutual influence even where it cannot be assumed that their speakers have ever been in direct contact with one another. (Appleyard, 1989: 304) From this perspective, unilateral accounts for the Ethiopian zone of convergence or the scenario of shift-induced change do not look very plausible. The social situation and its association with individual language structures is just too complex for such a view to be adequate. Thomason’s (2001: 113) statement that ‘the Ethiopian area arose primarily through imperfect learning, during a process of group language shift from Cushitic to Semitic’ (see also the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 83
Ethiopia as a Linguistic Area
quotation in Section 2.1) is extremely global. In a situation of an ‘infintely complex network of mutual influence’ as Appleyard (1989) puts it, no overall generalization concerning the impact of imperfect learning can be made. There are a considerable number of individual areas and sub-areas of contact, each with its own social background and its own structural properties. Of course, it makes sense to assume that imperfect learning is involved with different degrees of intensity in many of these sub-areas. But there are many other factors that can be involved. Thomason (2001: 128–56) herself lists a number of factors, such as code-switching, passive familiarity, bilingual first-language acquisition, deliberate decision and so on. In addition, there are Amharic and Oromo, with their general status of languages of wider communication. But again, no general statement can be made about their local prestige and about how they influence the more local varieties. A better understanding of what really makes the Ethiopian zone of convergence and its dynamics is only possible if there are many more studies of concrete local contact situations plus a detailed description of the individual languages involved. At the time of writing, this means that a lot more research is needed.
3
On the problematic status of the notion of linguistic area
3.1 Defining a linguistic area: setting the stage for a problematic field Since Trubetzkoy (1930) introduced the concept of Sprachbund (linguistic area, linguistic area in Section 2 above) this term has been the object of a plethora of controversial definitions. Trubetzkoy’s own programmatic outline is based on strong syntactic and morphological similarities, on a great number of common cultural words, and sometimes on a certain similarity of the phonological inventory in the structures of the languages involved. Explicitly exluded from this characterization of the Sprachbund are systematic sound laws, correspondences in the phonological material of morphological elements and common basic vocabulary because these factors indicate genetic relatedness.7 Trubetzkoy’s outline is rather general in various ways and it leaves a lot of room for further specifications. It does not say anything about how to quantify the syntactic, morphological and so on factors in order to come up with a more precise definition of what makes a Sprachbund, and it lacks overt reference to the geographic factor and the question of how to quantify spatial closeness. It only refers to ‘groups consisting of languages’ (Gruppen bestehend aus Sprachen), an expression that may at best trigger a geographic interpretation by pragmatic inference. Given the leeway intrinsic to Trubetzkoy’s (1930) outline, linguists of later generations tried to make it more precise from a number of perspectives. But, contrary to expectation, this striving for precision led to a situation in which the terms of Sprachbund/linguistic area ‘invite as many meanings and readings as there are linguistic minds to contemplate them’ (Stolz, 2002: 260).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
84
As I would like to show in this chapter, most criteria suggested in the literature for making the definition of a Sprachbund more clear-cut or discrete (for a good overview, see Stolz, 2002) do not work. If individual linguists use them for their definitions, most of them do this from the perspective of the contact zone they are dealing with. If this procedure leads to the terminological confusion described above, this may well be because of the differences between individual zones of contact and our lack of understanding of how, and to what extent, these differences can be integrated on a higher level. In Section 3.2, I shall discuss the way in which more clear-cut criteria are problematic. Section 3.3 will sketch the options for further research on contact-induced structural convergence. 3.2 Problems with more restricted criteria The criteria for making the definition of a Sprachbund more clear-cut are problematic in the following two ways. They are either difficult to apply – that is, they do not work in practice and may lead easily to the conclusion that there is no linguistic area; or they are arbitrary – that is, there is no motivation to stipulate them. There are at least two criteria that are difficult to apply. One is the claim that the features relevant for a linguistic area have to be identical to grant a maximum degree of similarity. The other criterion has to do with symmetry, and postulates that linguistic areas emerge exclusively out of contact with multilateral diffusion without any relationship of dominance (Aikhenvald and Dixon, 2001). The claim of exact identity does not only raise practical problems, it also gets right to the heart of formal as well as functional theories and the question of how far languages can be compared. In my view, neither formal nor functional approaches can provide any ultimately safe ground from which to argue about the comparability of languages. I would like to show this by confronting two diametrically opposed approaches – namely, Chomsky’s formal theory and Croft’s radical construction grammar. Formal approaches of the Chomskyan type deductively stipulate universal categories and relations that are reflected in the patterns of syntactic distribution across the world’s languages. From such a perspective, researchers dealing with linguistic areas should take much more seriously the formalist caveat that structures that appear to be identical at first glance may turn out to be considerably different in terms of universal grammar. The problem with the assumption of stipulated universal categories is, to put it in the terms of Croft (2001: 34), that ‘[p]ositing universal categories would imply identical behaviour across languages, which is empirically false’. If categories and relations are only language particular, as pointed out by Croft (2001), crosslinguistic comparability is only possible across constructions and their properties. Thus a researcher who strives for exact identity would have to
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 85
Ethiopia as a Linguistic Area
find a way of crossing the unbridgeable gulf between stiplulated universal categories (claimed to be false by Croft’s radical construction grammar) and the comparison of markedness patterns within constructions (claimed to be only surface phenomena by representatives of Chomsky’s formal theory). In addition to the theoretical problems associated with exact identity, there are some practical problems from observing what happens in concrete situations of language contact. If linguistic areas are a result of contact, exact identity implies that languages copy linguistic structures as a whole, with all their morphosyntactic, semantic and so on properties. But this is not the case. Johanson (1992, 2002) rightly points out that copying of this type (Globalkopieren or global copying, in his terms) is extremely rare. This is also confirmed from the perspective of bilingualism, an important factor for the emergence of contact-induced change. As was pointed out by Silva-Corvalán (1994), bilinguals tend to reduce the cognitive load of dealing with two languages and thus create structures that cannot meet the criterion of absolute identity. The second criterion, concerning symmetry and multilateral diffusion without social dominance, has to cope with the problem pointed out by Dahl (2001) that almost every case of language contact is in some way asymmetric. This also seems to apply to contact zones described in terms of linguistic areas. If diffusion of linguistic structures within so-called linguistic areas ultimately does not differ from diffusion elsewhere – that is, if it proceeds in terms of social networks or leaders of language change (see section 2.3), the presence of asymmetric social constellations is almost inevitable. Thus we may end up with no linguistic area if the criterion of symmetry is applied too strictly. The list of arbitrary criteria to be discussed here is longer than the one of criteria that are difficult to apply. Three of these criteria are connected to quantification. Thus we find restrictions concerning the minimal number of languages involved in a linguistic area (Haarmann, 1976: 23 postulates at least two languages; Schaller’s (1975) and Thomason’s (2001): 99)8 definitions need at least three languages); the minimal number of language families involved (two families in Emeneau’s (1956) definition;9 three families in Schaller’s (1975: 58) definition); and the minimal number of features involved (for Katz (1975: 16), one isogloss is enough; Wintschalek (1993: 6) needs more than one isogloss; Schaller (1975: 58) and Haarmann (1976: 23) need at least two features). Any quantification, irrespective of whether it concerns the domains of the number of languages, language families or linguistic features involved, is arbitrary as long as nothing else follows from it. It remains meaningless, as long as one cannot prove that there is a significant correlation between the presence of an exact number of items in any of the above domains of quantification and either a certain linguistic structure attested in a number of areas or a certain social pattern common to a number of areas. In my view, the existence of such a correlation is implausible and very unlikely.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
86
Exclusivity of features is another arbitrary criterion. It determines that a feature associated with a certain linguistic area must not be present in the languages that do not belong to that linguistic area but are from the same families (see, for example, Haarmann, 1976: 36; Kristophson, 1993: 2). The criterion of exclusivity becomes arbitrary in at least the following two scenarios. The first is related to the accidental adjacency of features. What happens if a feature that occurs in a certain zone of convergence by some accident also occurs in an adjacent language? In such a case, it is arbitrary to integrate that language into a postulated linguistic area because such an integration does not reflect the motivations responsible for its occurrence (for example independent-language internal development or migration of a group of speakers with that feature into an area adjacent to the postulated linguistic area). The second scenario is related to the overall geographic distribution of linguistic features. If linguistic areas are like dialect continua, with clusters of features whose composition changes from one set of features (language) to another, it may well happen that a certain feature is also present in a neighbouring language which for some reason is not part of a given linguistic area. The above question of whether a given language is part of a linguistic area itself leads to a further set of problems represented by the criterion of clear boundedness. This criterion requires internal homogeneity and discreteness in the sense that it is clearly determined which are the languages to be included into a given linguistic area and which are to be excluded (Sherzer, 1976: 221). As will be seen from the following two scenarios, it is extremely hard to apply this criterion neatly to real situations of language contact. If one takes again the situation of the dialect continuum as the first criterion, drawing a clear-cut borderline between the languages belonging to the linguistic area and the ones that do not is most likely to be impossible unless one has recourse to choosing arbitrarily particular features or a particular number of features. In the second scenario with a diffusion pattern comparable to a prototype with a core area scoring for the highest number of n shared features followed by adjacent areas with n – 1, n – 2 and so on features, the stipulation that only the core languages (or any weaker stipulation including, n – 1, n – 2 and so on features) are part of the linguistic area is likely to be arbitrary. In addition, the areas with n – 1 features and below (but not n = 1) are not homogeneous in the sense that they may contain different sets of features – a situation that calls for another decision with a high potential of arbitrarity. Even if it might be possible to suggest a non-arbitrary division in a certain area with a core-type diffusion pattern, it is unlikely that the criteria that apply to that area are of general relevance to a larger number of other areas in a way that allows them to be part of a more general definition of a linguistic area. The last criterion to be discussed in this section is the criterion of typological independence of the features involved (discussed by Campbell, 1994: 1471). This criterion is related to the question of whether one looks only at absolute universal implications (that is, implications attested in 100 per cent of a large
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 87
88
Ethiopia as a Linguistic Area
3.3 Options and conclusions The discussion of more restrictive criteria in section 3.2 shows clearly that it is extremely hard, if not systematically impossible, to produce a clear-cut definition of a linguistic area. This situation leaves basically two options: Option A: to abandon the concept of linguistic area; or Option B: to try to find a definition which is loose enough on the one hand but tight enough on the other (i)
to integrate sociohistorical facts and the geographic diffusion of certain features or clusters of features and (ii) to allow for the possibility of drawing further typological conclusions from the existence of a linguistic area. I tend to prefer the second conclusion (Option B), but this is more a project for extensive and detailed empirical research in the future than a result. Even though the concept of a linguistic area is associated with an idealization that is beyond a clear-cut definition, this does not imply that it is necessary to deny the existence of zones of contact-induced structural convergence (such as the Ethiopian zone of convergence discussed earlier in this chapter, the zones of convergence in the Balkan (Sandfeld, 1930), the Baltic area (Stolz, 1991), the Circum-Baltic area (Koptjevskaja-Tamm and Wälchli, 2001), South Asia (Emeneau, 1956, 1965; Masica, 1976), or East and mainland South East Asia (Bisang, 1996)). Such an approach appears to be both more open and more modest. It is basically analogous to geographical dialect continua plus sociolinguistic/sociohistorical and cognitive motivations (for example, economy versus iconicity, parsing and so on) and it is free from any associations with a rigid, scientifically unfounded definition of the type involved with the concept of linguistic area. To conclude this section, I would like to look at the motivations that may justify the wish to have a more clear-cut definition – that is, at the explanatory potential expected by those linguists who try to search for it. I can think of the following two reasons why finding such a definition may look attractive: • the prospect of drawing further typological conclusions from the existence of a linguistic area: ‘If x is a linguistic area it must have certain structural properties’; and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
sample; see Hawkins, 1983, 1994) or whether one also includes statistical universal correlations (that is correlations attested only in a statistically significant number of languages). In the latter case, there is the problem of how to calculate the degree of independence of individual features, and of where to draw a borderline without being arbitrary.
Walter Bisang 89
From a closer look at language contact and what we know about its results, neither of these prospects is very promising. The individual structural properties discussed for each linguistic area are extremely divergent from each other. We find only more general structural similarities such as perhaps a certain preference for periphrastic expression formats instead of morphological ones, or we can find that there is more structural similarity at higher levels of structural organization, such as discourse and clause combining, than at lower levels, such as morphology (based on findings from Matras, 1998). These similarities seem to be too general to correspond to the homogeneity associated with the concept of linguistic area. In addition, the similarities across different zones of convergence are too general to be discussed in terms of a more clear-cut definition. Rather, they call for a more open definition keeping the balance between looseness and tightness in the sense of Option B above. That it seems reasonable to assume certain social and historical facts behind the observation of structural convergence among languages seems to be straightforward. But given the vague correlation between phenomena of structural convergence and their concrete historical or social background, too clear-cut a definition does not look promising for finding potentially existing correlations. What is needed is, again, a balanced definition in the sense of Option B. Since a clear-cut definition of a linguistic area turns out to be highly problematical and since, in addition, such a definition is unlikely to contribute to drawing typological conclusions and conclusions concerning the interaction of social/historical facts with properties of linguistic structures, the more general concept of zones of contact-induced convergence is inevitable (if one does not want to exclude the existence of contact-induced convergence in general). This concept should replace the term ‘Sprachbund/linguistic area’ to avoid associations with unredeemable prospects of definitorial and explanatory precision, or with problematic idealizations of homogeneity across different areas.
4 On the typological relevance of zones of contact-induced convergence One of the reasons why zones of convergence are typologically relevant is their potential to create new or rarely attested linguistic types (see section 1). If new structures can arise within zones of convergence this means that the variety of linguistic structures as we find them attested in the world’s languages may not reflect the whole linguistic potential of humans. Given a
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
• the prospect of finding some concrete conclusions concerning social and/ or historical properties of a speech community: ‘If x is a linguistic area, this is because of the social or historical facts x, y and z’.
Ethiopia as a Linguistic Area
contact situation between languages with the ‘right’ linguistic types there is no reason why new types may not arise. As a consequence, it is ultimately impossible to draw any absolutely safe generalizations from the set of crosslinguistically attested types (for some further discussion, see section 5; also Maslova, 2000; Bisang, 2001a, 2004). It is not the purpose of this section to provide a large list of contactinduced rare or new types and to analyse them from a more general perspective. This can only be the topic of a special paper based on a great deal more research. Thus the next two sections present only two sketches of interesting case studies on rare or perhaps new types that emerged in the Ethiopian zone of convergence. Some more general consequences of the structural potential of contact-induced convergence beyond functional typology will be discussed in the next section – that is, in the outlook and conclusion. 4.1 The case of Bayso Bayso is a Cushitic language (East Cushitic: South Lowland East Cushitic: Omo-Tana) with OV word order plus AdjN (adjective in front of the noun) and NGen (genitive following its head noun). This combination of wordorder types is remarkable, in fact, the covariation of OV with (NGen & AdjN) is excluded by Greenberg’s Universal 5: (3) Greenberg’s Universal 5 (1966: 110; see also Hawkins, 1994: 320):10 If a language has dominant SOV order and the genitive follows the governing noun, then the adjective likewise follows the noun. If we look at the word-order patterns of other languages around Bayso, we can distinguish two groups of languages, called Highland East Cushitic (HEC) and Lowland East Cushitic (LEC) by Tosco (1994a: 416–17): • Highland East Cushitic (HEC): OV with AdjN and GenN; and • Lowland East Cushitic (LEC): OV with NAdj and NGen. Given the above two word-order patterns, the extraordinary pattern of Bayso most probably resulted from its position in a contact zone between Highland East Cushitic (AdjN & GenN) and Lowland East Cushitic (NAdj & NGen): It seems safe to assume that Bayso is currently moving from the N Mod order of its closest relatives (the other Omo-Tana languages) to Mod N – very possibly under pressure from its neighbors, the premodifying HEC languages. Influence from HEC on the Bayso Adj N order is suggested by Banti (1988: 243). (Tosco, 1994a: 432) The case of Bayso thus shows that rare types of covariation can occur in situations of language contact.11
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
90
Walter Bisang 91
Dimmendaal (1998; see also Bisang (2001a) for a more detailed summary) describes the contact area between Omotic (Afro-Asiatic; see Note 1) and two different branches of East Sudanic (Nilo-Saharan) – that is, Surmic and Nilotic. Southwest Surmic has verb-first word order plus postnominal location. Postnominal location is illustrated by urut-a ‘end’12 in example (4) from the Southwest Surmic language of Didinga: (4)
Didinga (Southwest Surmic, South Surmic; Dimmendaal, 1998: 46): ai othori erag-a urut-a is dog granary-LOC end-LOC ‘A dog is under the granary.’
The Nilotic languages, like the Surmic languages, are also verb-initial and are dominantly head marking. Although verb-initial word order is a notable exception within Nilo-Saharan languages, it seems quite plausible to assume that Proto-Nilotic also had verb-initial structure (see Dimmendaal, 1998: 73–5). In contrast to Surmic and Nilotic, the Omotic languages are characterised by verb-final word order plus postpositions. The marker of postnominal location in (4) is analysed as a ‘(post)modifier comparable to English “underneath” ’ by Dimmendaal (1998: 46), not as a postposition. Thus, Surmic languages such as Didinga do not violate Greenberg’s (1966: 78) Universal 3, according to which ‘languages with dominant VSO order are always prepositional’.13 Postpositions are typical of OV-languages, to which belongs the majority of Omotic languages. It is thus very likely that the emergence of a postnominal location marker in Didinga is the result of Omotic influence. However, the Omotic influence was not strong enough to make Didinga into a language of the type verb-first plus postpositions. A very interesting question in this context is whether contact between languages of the type VO&Prep and OV&Postp can lead to typologically more remarkable structures. This seems indeed to be the case in some Omotic languages, where location markers are not limited to the postnominal position. If the location-marking phrase is understood as being given or presupposed, location markers occur in front of the noun that is to be located; otherwise they occur in the postnominal position as one would expect of an OV language (see Dimmendaal (1998: 78) on the basis of data from Azeb Amha on Maale). The emergence of such a completely new type seems to be plausible if one starts from the idea that contact between languages of radically different word-order types such as VO versus OV or Prep versus Postp generates more freedom of word order, which can then be used for pragmatic purposes.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4.2 The case of some Omotic languages in contact with Surmic and Nilotic
Ethiopia as a Linguistic Area
5
Outlook and conclusion
One of the central questions in linguistics concerns the reason why languages have the structural properties they do rather than other logically possible structures. As I have argued elsewhere (Bisang, 2001a, 2004), theories dealing with this question have to consider the following three factors responsible for the propagation and selection of linguistic structures: • universal grammar; • human cognitive equipment; and • social factors. In terms of Chomskyan-type formal linguistics, the only factor of interest to explain the structures of the world’s languages is universal grammar. The existence of an innate universal grammar is a very strong claim which has to be seen against the functional claim that linguistic structures can be accounted for without assuming universal grammar. The burden of proving the existence of universal grammar is clearly with those who postulate it. In my view, such a proof is problematic, because social and cognitive factors may disfavour structural options that would be fully compatible with universal grammar. Even if Chomsky (2002: 124) does not accept falsification as ‘a meaningful proposal to get rid of a theory’ on an abstract theoretical level of argumentation, in practice, the architecture of the language faculty is developed on the basis of empirical evidence. This evidence is not only the result of universal grammar – it is also influenced by cognitive and social factors. The latter two factors may in fact obstruct the linguist’s access to the full potential of universal grammar (if it exists at all). If zones of structural convergence show that social and cognitive factors have to be taken at least as seriously as universal grammar in order to explain the structural variance found in the world’s languages, they, together with their social dynamics, get a rather strong explanatory potential for the structural range of variation as we find it in the world’s languages. For functional typological conclusions, the situation is not much better. As was pointed out by Maslova (2000: 307; see also Bisang, 2004), there is no way to ‘distinguish between general distributional universals and accidental statistical properties of the current language population’. This is because it is impossible to state whether and to what extent the distribution of linguistic types is statistically neutral and, consequently, to draw watertight conclusions concerning universal generalizations. Needless to say, an important factor for assessing the problem of statistical neutrality is language contact and the impact of zones of convergence.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
92
Walter Bisang 93
Notes
1 Specialists dealing with the linguistic area of Ethiopia use the term ‘language area’, a term also used by Emeneau (1956; see also Note 10). Since the term ‘linguistic area’ is more widely used, I prefer to use this here. 2 According to Fleming and Bender (1976), Omotic represents an independent branch of Afro-Asiatic. Ehret (1978) interprets Omotic as the first split of AfroAsiatic. In terms of Lamberti (1991), Omotic is nothing more than a subfamily of Cushitic. 3 ‘The status of Ethiopia as a language area has since long been well established, and Emeneau’s definition of the Indian case well applies to Ethiopia’ (Tosco, 1994a: 415). ‘The status of Ethiopia as a typical language area, along the lines of Emeneau’s classical definition . . . is since long well established’ (Tosco, 1994b: 225). 4 To Dolgopolsky (1977) and many others before him, glottalization reflects the initial articulation of emphatic consonants in Proto-Semitic. 5 ‘All the nouns which bleached into governing relational words in the earlier period have to take the function of prepositions or clause-initial conjunctions by nature. But there are also some postpositions from the old period which also followed the new word order, even if they did so with some reluctance. Apart from the old predominant postpositional position they every now and then are already used prepositionally’ (Praetorius, 1893: 286–7; my translation) (‘Alle Nomina, welche sich in der jüngeren Periode zu regirenden Verhältnisswörtchen abgeblasst haben, müssen naturgemäss als Präpositionen und vorgestellte Conjunktionen auftreten. Aber auch einige Postpositionen der alten Zeit haben sich, freilich mehr oder weniger zögernd, der neuen Wortfolge angeschlossen und zeigen neben der vorherrschenden alten postpositionalen Stellung hie und da schon präpositionellen Gebrauch’). 6 The postnominal position of the relative clause in Oromo provides similar evidence of Ethio-Semitic influence on Cushitic (see also Praetorius, 1893: 81–2, 84). 7 Trubetzkoy’s original definition of Sprachbund in German runs as follows: Gruppen bestehend aus Sprachen, die eine grosse ähnlichkeit in syntaktischer Hinsicht, eine ähnlichkeit in den Grundsätzen des morphologischen Baus aufweisen, und eine grosse Anzahl gemeinsamer Kulturwörter bieten, manchmal auch äussere ähnlichkeit im Bestande der Lautsysteme, – dabei aber keine systematische Lautentsprechungen, keine übereinstimung in der lautlichen Gestalt der morphologischen Elemente und keine gemeinsamen Elementarwörter besitzen. – solche Sprachgruppen nennen wir Sprachbunde [sic]. (Trubetzkoy, 1930: 17) 8 Thomason’s (2001) definition is as follows: ‘a linguistic area is a geographical region containing a group of three or more languages that share some structural features as a result of contact rather than as a result of accident or inheritance from a common ancestor’.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
* This chapter is part of ongoing research in the Collaborative Research Centre (Sonderforschungsbereich) ‘Cultural and Linguistic Contact: Processes of Change in North-Eastern Africa and in West Asia’ supported by the German Science Foundation (Deutsche Forschungsgemeinschaft) since 1997. I would like to thank the German Science Foundation for generously supporting this research.
Ethiopia as a Linguistic Area
9 Since Emeneau’s (1956) definition is referred to quite often in the literature, I mention it here: ‘A language area is an area which includes languages belonging to more than one family but showing traits in common which are found not to belong to the other members of (at least) one of the families’ (Emeneau, 1956: 16, n. 28). 10 See also Hawkins (1994: 320): ‘if NGen then NAdj in postpositional languages (equivalently: if AdjN then GenN). This generalization holds exceptionlessly for the Expanded Sample of Hawkins (1983: 67–8)’. 11 In the meantime, it turned out that OV & (NGen & AdjN) can also be the product of a language-internal development, as is the case in Tigre (Tosco, 1998). 12 From a typological perspective, this construction is remarkable, because it is double-marking – that is, the locative relation is marked at the head of the relational-noun construction (urut-a) as well as at the dependent (erag-a). 13 From the point of view of later publications on word-order typology, violations of Greenberg’s Universal 3 are attested even though such cases are highly marked. According to Dryer (1992: 83), there are twelve genera in which the type VO&Postp is attested against seventy genera with the type VO&Prep.
References Aikhenvald, A. Y. and R. M. W. Dixon (eds) (2001) ‘Introduction’, Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press). Amborn, H., G. Minker, and H.-J. Sasse (1980) Das Dullay: Materialien zu einer ostkuschitischen Sprachgruppe (Berlin: Reimer). Appleyard, D. (1984) ‘The Internal Classification of the Agaw Languages: A Comparative and Historical Phonology’, in J. Bynon (ed.), Current Progress in Afro-Asiatic Linguistics: Papers of the Third International Hamito-Semitic Congress (Amsterdam/ Philadelphia: John Benjamins), pp. 33–67. Appleyard, D. (1989) ‘The Relative Verb in Focus Constructions: An Ethiopian Areal Feature’, Journal of Semitic Studies, vol. 34, pp. 291–305. Banti, G. (1988) ‘Adjectives in East Cushitic’, in M. Bechhaus-Gerst and F. Serzisko (eds) Papers from the International Symposium on Cushitic and Omotic Languages. Cologne, 6–9 January 1986 (Hamburg: Buske), pp. 205–59. Bender, M. L. (1979) ‘Gumuz: A Sketch of Grammar and Lexicon’, Afrika und Übersee, vol. 62, pp. 38–69. Bisang, W. (1995) ‘Verb Serialization and Converbs: Differences and Similarities’, M. Haspelmath and E. König (eds), Converbs in Cross-linguistic Perspective (Berlin: Mouton de Gruyter), pp. 137–88. Bisang, W. (1996) ‘Areal Typology and Grammaticalization: Processes of Grammaticalization Based on Nouns and Verbs in East and Mainland South East Asian Languages’, Studies in Language, vol. 20, no. 3, pp. 519–97. Bisang, W. (2001a) ‘Areality, Grammaticalization and Language Typology. On the Explanatory Power of Functional Criteria and the Status of Universal Grammar’, W. Bisang (ed.), Language Typology and Universals (Berlin: Akademie-Verlag). Bisang, W. (2001b) ‘Finite vs. Non Finite Languages’, M. Haspelmath, E. König, W. Österreicher and W. Raible (eds), Language Typology and Language Universals, Vol. 2 (Berlin and New York: Mouton de Gruyter), pp. 1400–13. Bisang, W. (2004) ‘Dialectology and Typology – An Integrative Perspective’, B. Kortmann (ed.), Dialectology Meets Typology. Dialect Grammar from a Cross-linguistic Perspective (Berlin: Mouton de Gruyter), pp. 11–45.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
94
Campbell, L. (1994) ‘Grammar: Typological and Areal Issues’, in R. E. Asher and J. M. Y. Simpson (eds), The Encyclopedia of Language and Linguistics, vol. 3 (Oxford: Pergamon Press), pp. 1471–4. Chomsky, N. (2002) On Nature and Language (Cambridge: Cambridge University Press). Crass, J. (2002) ‘Ejectives and Pharyngeal Fricatives: Two Features of the Ethiopian Language Area’, Baye Yimam, Richard Pankhurst, David Chapple, Yonas Admasu, Alula Pankhurst, and Birhanu Teferra (eds), Ethiopian Studies At the End of the Second Millennium, vol. III. Proceedings of the 14th International Conference of Ethiopian Studies, 6–11 November 2000, Addis Ababa (Addis Ababa: University Press), pp. 1679–91. Crass, J. and W. Bisang (2004) ‘Einige Bemerkungen zum äthiopischen Sprachbund und ihre Relevanz für die Areallinguistik’, in W. Bisang, Th. Bierschenk, D. Kreikenbom and U. Verhoeven (eds), Kultur, Sprache, Kontakt (Würzburg: Ergon). Crass, J., B. Dehnhard, R. Meyer and A. Wetter (2001) ‘Von “sagen” zum Verbbildungsmorphem: Die Grammatikalisierung des Verbs “sagen” einmal anders’, Afrikanistische Arbeitspapiere, vol. 65, pp. 129–41. Croft, W. A. (2001) Radical Construction Grammar. Syntactic Theory in Typological Perspective (Oxford: Oxford University Press). Dahl, Ö. (2001) ‘Principles of Areal Typology’, M. Haspelmath, E. König, W. Österreicher and W. Raible, (eds), Language Typology and Language Universals, vol. 2 (Berlin and New York: Mouton de Gruyter). Dimmendaal, G. J. (1998) ‘A Syntactic Typology of the Surmic Family from an Areal and Historical-Comparative Point of View’, in G. J. Dimmendaal, and Marco Last (eds), Surmic Languages and Cultures (Cologne: Köppe), pp. 35–81. Dolgopolsky, A. (1977) ‘Emphatic Consonants in Semitic’, Israel Oriental Studies, vol. 7, pp. 1–13. Dryer, M. (1992) ‘The Greenbergian Word Order Correlations’, Language, vol. 68, pp. 81–138. Ehret, Ch. (1978) ‘Omotic and the Subgrouping of the Afroasiatic Language Family’, R. L. Hess (ed.), Proceedings of the Fifth International Conference on Ethiopian Studies, Session B, 13–16 April 1978 (Chicago: University of Iblinois), pp. 51–62. Ehret, Ch. (1995) Reconstructing Proto-Afroasiatic (Proto-Afrasian): Vowels, Tone, Consonants, and Vocabulary (Berkeley, Calif., Calif.: University of California Press). Emeneau, M. (1956) ‘India as a Linguistic Area’; republished in A. S. Dil (ed.), Language and Linguistic Area: Essays by Murray B. Emeneau (Stanford, Calif.: Stanford University Press, 1980), pp. 105–25. Emeneau, (1965) M. ‘India and Linguistic Areas’; reprinted in A. S. Dil (ed.), Language and Linguistic Area: Essays by Murray B. Emeneau (Stanford, Calif.: Stanford University Press, 1980), pp. 126–66. Ferguson, C. A. (1970) ‘The Ethiopian Language Area’, Journal of Ethiopian Studies, vol. 8, no. 2, pp. 67–80. Ferguson, C. A. (1976) ‘The Ethiopian Language Area’, M. L. Bender, J. D. Bowen, R. L. Cooper and C. A. Ferguson (eds), Language in Ethiopia (London: Oxford University Press), pp. 63–76. Fleming, H. and M. L. Bender (1976) ‘Non-Semitic Languages: Cushitic and Omotic’, J. D. Bowen, R. L. Cooper and C. A. Ferguson (eds), Language in Ethiopia (London: Oxford University Press), pp. 34–53. Greenberg, J. H. (1966) ‘Some Universals of Grammar With Particular Reference to the Order of Meaningful Elements’, J. H. Greenberg (ed.), Universals of Language (Cambridge, Mass.: MIT Press), pp. 73–113.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 95
Ethiopia as a Linguistic Area
Greenberg, J. H. (1959) ‘Africa as a Linguistic Area’, W. Bascom and R. Herskovits (eds), Continuity and Change in African Languages (Chicago: University of Chicago Press), pp. 15–27. Haarmann, H. (1976) Aspekte der Arealtypologie: Die Problematik der europäischen Sprachbünde (Tübingen: Narr). Haspelmath, Martin (1995) ‘The Converb as a Cross-linguistically Valid Category’, in M. Haspelmath and E. Konig (eds.), Converbs in Cross-linguistic Perspective. Structure and Meaning of Adverbial Verb Forms, Empirical Approaches to Language Typology, 13 (Berlin: Mouton de Gruyter), pp. 1–56. Haspelmath, M. and E. König (eds.) (1995) Converbs in Cross-linguistic Perspective. Structure and Meaning of Adverbial Verb Forms, Emperical Approaches to Language Typology, 13 (Berlin: Mouton de Gruyter). Hawkins, J. A. (1983) Word Order Universals (New York: Academic Press). Hawkins, J. A. (1994) A Performance Theory of Order and Constituency (Cambridge: Cambridge University Press). Hetzron, R. (1975) ‘Genetic classification and Ethiopian semitic’, in J. Bynon and Th. Bynon (eds), Hamito-Semitica: Proceedings of a Colloquium Held by the Historical Section of the Linguistics Association (Great Britain) At the School of Oriental and African Studies, University of London, on the 18th, 19th and 20th of March 1970 (The Hague, Paris: Mouton), pp. 103–27. Hetzron, R. (ed.), (1997) The Semitic Languages (London, New York: Routledge). Hudson, G. (1989) Highland East Cushitic Dictionary, Kuschitische Sprachstudien 7 (Hamburg: Buske). Johanson, L. (1992) Strukturelle Faktoren in türkischen Sprachkontakten (Proceedings of the Wissenschaftlichen Gesellschaft an der J. W. Goethe-Universität Frankfurt a.M.). (Stuttgart: Steiner). Johanson, L. (2002) Structural factors in Turkic Language Contacts (London: Curzon). Kapeliuk, O. (1988) ‘Amharique’, in D. Cohen (ed.), Les langues chamito-sémitiques. Texte réuni par D. Cohen, Les langues dans le monde ancien et moderne, 3 (Paris), pp. 146–57. Katz, H. (1975) Generative Phonologie und phonologische Sprachbünde des Ostjakischen und Samojedischen (Munich: Fink). Keller, R. (1990) Sprachwandel (Tübingen: Francke). Keller, R. (1994) On Language Change: The Invisible Hand in Language (London: Routledge). (Translation and expansion of Sprachwandel). Koptjevskaja-Tamm, M. and B. Wälchli (2001) ‘The Circum-Baltic Languages: An Areal-Typological Approach’, Ö. Dahl and M. Koptjevskaja-Tamm (eds), Circum-Baltic Languages. Typology and Contact (Amsterdam/Philadelphia: John Benjamins), pp. 615–750. Kristophson, J. (1993) ‘Ein neuer Beitrag zur Sprachbunddiskussion’, Zeitschrift für Balkanologie, vol. 29, no. 1, pp. 1–11. Labov, W. (2001) Principles of Linguistic Change. Social Factors (Oxford: Blackwell). Lamberti, M. (1991) ‘Cushitic and its Classifications’, Anthropos, vol. 86, pp. 552–61. Leslau, W. (1945) ‘The Influence of Cushitic on the Semitic Languages of Ethiopia. A Problem of Substratum’, Word, vol. 1, no. 1, pp. 59–82. Leslau, W. (1952) ‘The Influence of Sidamo on the Ethiopic Languages of Gurage’, Language, vol. 28, no. 1, pp. 63–81. Leslau, W. (1969) ‘The Pseudo-Gerundive in Caha’, Rassegna di studi etiopici, vol. 23, pp. 27–42.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
96
Masica, C. P. (1976) Defining a Linguistic Area (Chicago: University of Chicago Press). Maslova, E. (2000) ‘A Dynamic Approach to the Verification of Distributional Universals’, Linguistic Typology, vol. 4, no. 3, pp. 307–33. Matras, Y. (1998) ‘Utterance Modifiers and Universals of Grammatical Borrowing’, Linguistics, vol. 36, pp. 281–331. Meyer, R. and R. Richter (2003) Language Use in Ethiopia from a Network Perspective (Frankfurt: Lang). Milroy, J. and L. Milroy (1985) ‘Linguistic Change, Social Network and Speaker Innovation’, Journal of Linguistics, vol. 21, pp. 339–84. Nedjalkov, V. (1995) ‘Some Typological Parameters of Converbs’, M. Haspelmath and E. König. (eds), Converbs in Cross-linguistic Perspective (Berlin: Mouton de Gruyter), pp. 615–750. Orel, V. E. and O. V. Stolbova (1995) Hamito-Semitic Etymological Dictionary: Materials for a Reconstruction (Leiden: Brill). Praetorius, F. (1893) Zur Grammatik der Gallasprache (Berlin: Peiser); reprinted 1978 Hildesheim, New York: Olms. Sandfeld, Kr. (1930) Linguistique Balkanique, problèmes et résultats (Paris: Klincksieck). Sasse, H-J. (1973) ‘Spuren von Pharyngalen im Galab’, Afrika und Übersee, vol. 56, pp. 266–75. Sasse, H-J. (1979) ‘The Consonant Phonemes of Proto-East-Cushitic (PEC): A First Approximation’, Afroasiatic Linguistics, vol. 7, no. 1, pp. 1–67. Schaller, H. (1975) Die Balkansprachen (Heidelberg: Universitätsverlag Winter). Sherzer, J. (1976) An Areal-Typological Study of American Indian Languages North of Mexico. (Amsterdam: North-Holland). Silva-Corvalán, C. (1994) Language Contact and Change: Spanish in Los Angeles (Oxford: Clarendon Press). Stolz, T. (1991) Sprachbund im Baltikum? Estnisch und Lettisch im Zentrum einer sprachlichen Konvergenzlandschaft (Bochum: Universitätsverlag Dr. Norbert Brockmeyer). Stolz, T. (2002) ‘No Sprachbund Beyond This Line! On the Age-Old Discussion of How to Define a Linguistic Area’, Ramat, P. and Stolz, T. (eds), Mediterranean Languages. Papers from the MEDTYP Workshop, Tirrenia, June 2000 (Bochum: Universitätsverlag Dr. Norbert Brockmeyer). Thomason, S. G. (2001) Language Contact: An Introduction (Edinburgh: Edinburgh University Press). Thomason, S. G. and T. Kaufman (1988) Language Contact, Creolization and Genetic Linguistics (Berkeley, Calif.: University of California Press). Tosco, M. (1994a) ‘The Historical Syntax of East Cushitic: A First Sketch’, in T. Bearth, W. J. G. Möhlig, B. Sottas and E. Suter (eds), Perspektiven afrikanistischer Forschung: Beiträge zur Linguistik, Ethnologie, Geschichte, Philosophie und Literatur, X. Afrikanistentag (Cologne: Köppe), pp. 415–40. Tosco, M. (1994b) ‘On Case Marking in the Ethiopian Language Area’, in V. Brugnatelli, (ed.), Sem, Cam, Iafet (Milano: Centro studi camito-semitici), pp. 225–44. Tosco, M. (1998) ‘A Parsing View on Inconsistent Word Order: Articles in Tigre and its Relatives’, Linguistic Typology, vol. 2, pp. 355–80. Tosco, M. (2000) ‘Is There an ‘Ethiopian Language Area?’, Anthropological Linguistics, vol. 42, pp. 329–65. Trubetzkoy, N. S. (1930), ‘Proposition 16’, Actes du premier congrès international de linguistes à la Haye, du 10–15 avril 1928 (Leiden: A. W. Sijthoff), pp. 17–18.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Walter Bisang 97
98
Ethiopia as a Linguistic Area
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Wintschalek, W. (1993) Die Areallinguistik am Beispiel syntaktischer Übereinstimmungen im Wolga-Kama-Areal (Wiesbaden: Harrassowitz). Zaborski, A. (1991) ‘Ethiopian Language Subareas’, in S. Pilaszewicz and E. Rzewuski (ed.), Unwritten Testimonies of the African Past, Proceedings of the International Symposium held in Ojrzanów n. Warsaw on 7–8 November 1989 (Warsaw: Wydawnictwa Uniwersytetu Warszawskiego), pp. 123–34.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Structural Isoglosses between Khoekhoe and Tuu: The Cape as a Linguistic Area1 Tom Güldemann
In historical times, the wider Cape region, including the Orange River area, hosted languages of two very different Khoisan language families, namely of Tuu (specifically its !Ui branch) and Khoe (specifically its Khoekhoe branch). Khoekhoe displays a number of linguistic features that do not exist in the languages of its genealogical sister, the Kalahari branch of Khoe. A comparison beyond the limits of this family shows that the innovative structures in Khoekhoe often have a great similarity to properties of the Tuu family, particularly its !Ui branch. This observation leads to the hypothesis that the genuine linguistic character of Khoekhoe vis-à-vis Kalahari Khoe is to a considerable extent the result of contact with Tuu languages, which have been in the relevant area for a longer time. This chapter will (a) outline briefly the historical context of the contact situation; (b) identify commonalities of the two groups, with a particular focus on the assumed Tuu substrate interference in the morphosyntax of Khoekhoe; and (c) discuss a few implications of the data for the population history in southern Africa and for historical and contact linguistics in general.
1
Khoisan-internal language contact in the Cape
Khoisan research started to focus on areal and contact linguistics only recently (see, for example, Güldemann, 1998; Traill and Nakagawa, 2000); this mainly because Greenberg’s (1963) hypothesis of a genealogical MacroKhoisan unit has overshadowed the discipline as far as historical aspects of the languages were concerned, and has thus diverted the interests of scholars from non-genealogical approaches. While a ‘Khoisan language family’ is generally accepted among non-specialists, most Khoisan linguists (including me) do not consider the proposed evidence to support this view (see Güldemann and Vossen, 2000; Güldemann, forthcoming d);2 I pursue instead the idea that commonalities across certain click language groups in 99 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
5
100
The Cape as a Linguistic Area
southern Africa are mediated by language contact rather than by inheritance from a common ancestor. Embedded in this general approach, this chapter will focus on a subpart of southern Africa, namely the Cape, including the region along the Orange River.
In historical times, the Cape hosted languages from two genealogically unrelated groups: Tuu (alias Southern Khoisan; see Güldemann (2004b) for the new terminology) and Khoe (alias Central Khoisan). Khoe is the core of a higher-order family Khoe-Kwadi, proposed by Güldemann (2004a) and Güldemann and Elderkin (forthcoming). The Tuu family, together with the language complex Ju (alias Northern Khoisan) and the still isolated language c Hõa, is subsumed under the non-genealogical, typological grouping called ‘Non-Khoe’ because of considerable morphosyntactic similarities between the three units (see, for example, Güldemann and Vossen, 2000). Map 5.1 shows the geographical distribution of all southern African Khoisan lineages. The member languages of Khoe-Kwadi and Tuu and their subgroupings, as far as they are known, are shown in Figures 5.1 and 5.2, respectively (DC = dialect cluster; † = extinct); groups and languages that are of higher relevance for the present discussion, and for which sufficient data are available are in bold type. Unfortunately, the general access to linguistic data on the languages of the Cape is severely restricted in various respects. Most importantly, the majority of relevant varieties are extinct today without having been documented at all, or at least not sufficiently and according to modern standards. From some regions, especially the southern- and westernmost parts of the Cape, the available material comprises at best the names of ethnic groups and short word lists in highly defective transcriptions. This also means that the picture regarding language and dialect geography is very scanty. It cannot even be stated with confidence that San languages spoken along the coast belonged to the Tuu family. Of the extinct languages and dialect clusters, there are only three on which a more extensive or even sizeable amount of linguistic data are available; namely !Ora on the Khoekhoe side and c Ungkue and ⏐Xam on the !Ui side, all recorded in the nineteenth and early twentieth centuries. A drawback of these materials is that they display considerable lacunae and are not accompanied by a full linguistic analysis. Only two of the languages at issue are still spoken at the time of writing within the area: the Richtersveld in the far north-western corner of South Africa still hosts a few thousand speakers of Nama (Khoekhoe), and fewer than twenty old people scattered over the Northern Cape north of the Orange River retain a knowledge of one or more N || ng varieties (!Ui). Both Nama and N || ng in South Africa are not documented sufficiently, so the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
1.1 Languages, classification and data
101
Map 5.1 Distribution of Khoisan lineages in the early colonial period
linguistic material used in this study is also defective. Information on Nama only comes from varieties north of the Orange River, the use of which is problematic, for reasons given below. The available N || ng data, figuring in the literature under such language names as c Khomani and N⏐huki, are limited in terms of size and quality.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
102
The Cape as a Linguistic Area
Shua: Tshwa:
Cara, Deti†, ⏐Xaise, Danisi, Ts’ixa, etc. Kua, Cua, Tsua, etc.
Kxoe: G || ana: Naro:
Kxoe, || Ani, Buga, G⏐anda, etc. G || ana, G⏐ui, c Haba, etc. Naro, etc.
West
Khoekhoe North: Eini (DC)†, Nama-Damara (= Standard Khoekhoe), Hai || ’om (DC) South: Cape varieties (DC)†, !Ora (DC)† Figure 5.1 Genealogical classification of Khoe-Kwadi
TUU !Ui: ⏐Xam (DC)†; N || ng (DC including c Khomani, N⏐huki); c Ungkue†; || Xegwi† Taa: !Xõo (DC including N⏐amani†, N⏐u || ’en†, Kakia†); Lower Nosop (⏐’Auni, ⏐Haasi)† Figure 5.2 Genealogical classification of Tuu
A final problem of the accessible data concerns the complex processes involving some speech communities in their more recent history. In order to ascertain the result of assumed linguistic contacts between Khoekhoe and Tuu varieties, they should ideally not have undergone considerable linguistic changes and secondary interferences later. However, the situation is, at least for Khoekhoe, far more complicated in that we have no direct access to a variety that is unambiguously either South or North Khoekhoe. The data on North Khoekhoe is largely confined to modern Standard Namibian Khoekhoe and its dialects; this is also known as Nama-Damara and was codified to a considerable extent through missionary activities. At the core of this language are the varieties of the pastoral Nama on both sides of the Lower Orange River, who expanded from the seventeenth to the eighteenth century into the large region north of the river and west of the Kalahari, into what today is known as Namibia (Vedder, 1934; Budack, 1986). However, it also received input from at least two other sources: (a) the unidentified language(s) spoken by the Damara before their linguistic assimilation to the pastoral Nama;3 and (b) the Khoekhoe varieties of the Orlam groups who were displaced from South Africa, and incorporated a considerable number of South Khoekhoe speakers from the Cape. Since the Cape varieties have not been documented sufficiently, the closest we can get to genuine South Khoekhoe is the !Ora dialect chain. It is certain
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
KHOE-KWADI Kwadi single language Khoe Kalahari East
103
that an essential component of the !Ora, who finally settled around the confluence of the Vaal and Orange Rivers, were Cape Khoekhoe who from the seventeenth century onwards tried to escape the European colonization emanating from the south. It is equally clear, however, that these mixed with other people on their way to and within their final abodes; of particular relevance here is the attested contact with North Khoekhoe-speaking pastoralists such as the Nama of the Lower Orange, and upriver the Eini between Augrabies and Upington (these are commonly, but inappropriately subsumed under the !Ora). 1.2 The language contact situation According to everything we know from non-linguistic evidence, the San, with a foraging mode of life, are the oldest attested ethno-historical layer in southern Africa. In the Cape, this cultural classificatory entity correlates with the linguistic unit !Ui. From about 2,000 to 2,500 years ago onwards, archaeological findings registered a new cultural type in southern Africa based to a large extent on animal husbandry. In the Cape this correlates with a distinct linguistic group – the Khoekhoe. The Cape is thus characterized by a fairly neat correspondence of a linguistic dichotomy of Khoekhoe versus !Ui with a cultural dichotomy of larger pastoralist tribes versus small bands of hunter-gatherers similar to San. Therefore, it has been assumed for a long time that the new pastoral culture was introduced by an immigrating population, along with a new linguistic tradition – an idea I follow here. The trajectory of the pastoral expansion in the archaeological record, and the fact that the closest linguistic relatives of Khoekhoe are found today in the north, in Botswana, Namibia and Angola, would suggest that the pre-Khoekhoe entered the Cape from a more northern, rather than an eastern, direction. Pastoralism did not cover the entire area under consideration: in having to skip dry, ecologically unsuitable zones like the Kalahari, Karoo and so on it remained restricted to the neighbourhood of the coast and beside great rivers. In these areas we can assume a cohabitation of !Ui-speaking San and Khoekhoe-speaking pastoralists for about two millennia, with regional differences (see Yates et al., 1994; Webley, 1998). While a population distinction involving subsistence, social organization and language (epitomized in the old ‘Bushman’–‘Hottentot’ dichotomy) is synchronically irrelevant on a larger geographical scale, it does apply to the Cape from early times up to recent historical times; despite the possible inter-group mobility of individuals and/or small social units. Compared to other cases of language contact between different Khoisan language groups, the situation in the Cape is unique. Usually, contact across marked linguistic boundaries is relevant at the periphery of language territories, and subsequent diffusion of borrowed features into neighbouring areas would be relatively slow. The territory of Khoekhoe, however, was originally
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
entirely within the confines of the Tuu-speaking area (comparable, for example, to the case of Ethio-Semitic within the Cushitic-speaking area), so that language contact between the two populations in the Cape was intensive and sustained over a long time, and Khoekhoe was affected as a whole. The intensity and character of social interaction between the more territorial San bands and the larger and more mobile pastoral groups varied over space and time, so that different sociolinguistic patterns are relevant for the linguistic contact. As a generalization, two idealized patterns should be distinguished. When Khoekhoe newly entered an area populated by San and/or settled in small numbers, the relationship between the two groups was more or less equal and the language contact characterized by mutual bilingualism and borrowing. While this is especially relevant for the historically remote, initial stage of Khoekhoe colonization in the Cape, it is still attested in some areas for the colonial period.4 With the permanent settlement of pastoral groups in certain zones, the development of an asymmetric contact pattern, often in the form of a client relationship of San towards pastoralists, must be assumed. This could be followed by the incorporation of San individuals, or even whole social units, into Khoekhoe-speaking ethnic groups. In the final stage, Khoekhoe would have been the target of a complete language shift on the part of the San. This is attested for many places in the wider Cape area, and such cases are still evident in the twentieth century – for example, with speakers of N || ng varieties in the southern Kalahari. The linguistic effect of such a contact pattern would have been interference of the !Ui substrate in the respective Khoekhoe variety. In the following sections, I will try to show that Khoekhoe is untypical of its family Khoe in a number of respects, and that the Khoekhoe-specific features have counterparts in the Tuu family, in particular its !Ui branch. The observation that many isoglosses are properties of the entire Tuu family (or larger parts thereof), but not of Khoe clearly suggests that Tuu languages have remained relatively stable, but Khoekhoe has innovated vis-à-vis the Kalahari branch. This leads to the central hypothesis of this chapter, namely that Khoekhoe – in its distribution area south of the Kalahari – was subject to linguistic interference from Tuu which, as a family, has been present there for a much longer period. 1.3 Identifying isoglosses The above hypothesis stresses that Khoekhoe has changed under Tuu influence. Such innovations are identified in particular in the domain of morphosyntax and constitute the main body of this chapter (see section 4). This is not to say that Tuu languages were not affected by the contact; rather, the influences were bilateral and the entire range of isoglosses, be they lexical, phonetic-phonological or morphosyntactic, most probably
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
104
105
define a linguistic sub-area in the wider geographical context of Khoisan southern Africa. The focus on the assumed Tuu substrate in Khoekhoe morphosyntax is in part because of insufficient comparative data regarding phonetics-phonology and lexicon, so that it is still difficult to determine the direction of borrowing and interference. Regarding the profile and evaluation of the isoglosses, several remarks are in order. Ideally, a linguistic feature proposed for the Cape should be found in !Ui and Khoekhoe, but should exclude languages outside this area, in particular those of the Kalahari branch of Khoe outside the Tuu influence. Being able to present such a clear-cut picture here, however, is the exception rather than the norm. A general problem is the present lack of a full comparative picture for southern Africa as a whole because of the overall poor documentation of Khoisan languages. I have already pointed out the serious gaps in the languages directly concerned; but such lacunae are also relevant for Kalahari Khoe as the essential tertium comparation is for Khoekhoe as well as for Tuu and other Non-Khoe languages outside the Cape. A number of supposedly transferred features concern only a structural principle, with no linguistic substance. A few of these also seem to affect certain Kalahari Khoe languages, sometimes in a more incipient form. Such a situation does not necessarily exclude the Tuu substrate hypothesis for Khoekhoe. The groups for which this is relevant so far, namely Naro and possibly G || ana, are located in the south-western periphery of Kalahari Khoe and are in direct contact with Non-Khoe languages (Naro is, in fact, sandwiched geographically between Ju in the north and Tuu in the south). With such a picture, it is possible to argue that the relevant feature in Khoe displays a north–south cline of increasing change towards the Non-Khoe pattern, corresponding to the historical encroachment of Khoe into NonKhoe speaking areas. The particular salience of the feature in Khoekhoe could then still be attributed to its longest and/or most intimate contact with Tuu. A related point is the fact that some features proposed as Khoekhoe innovations through contact with Tuu have in fact a greater areal distribution in southern Africa in that they are more widespread in Non-Khoe as a whole. In such cases it is mainly for the greater plausibility in historical and geographical terms that the influence is ascribed to Tuu. However, this does not exclude the possibility of yet earlier influences on pre-Khoekhoe by some Non-Khoe language(s) other than from the Tuu family (see, for example, Güldemann, forthcoming e). In general, in cases where no transfer of linguistic substance is involved, it is hard to tell whether Khoekhoe (partially) acquired them before or after it entered the Cape. That is, the substrate in Khoekhoe cannot always be pinned down to the !Ui languages it was in contact with last. Also, some isoglosses seem to hold only for subparts of Khoekhoe and/or !Ui; in particular,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
linguistic similarities appear to be especially strong between languages of the Lower Orange River area – that is, North Khoekhoe and ⏐Xam. While such a picture might well reflect the real situation, it cannot (yet) be excluded on the basis that it is in fact an artefact of the better availability of data in this subregion. In general, the list of isoglosses discussed here is not intended to be exhaustive or final, but rather represents work in progress that will have to be modified when more comparative data becomes available. It might well be possible that the list can be extended, but also that a feature to be given below turns out later not to single out Khoekhoe against Kalahari Khoe and/or not to be related to a structure in Tuu. A final point relates to the degree of similarity between the compared features of the two groups. My analysis here will be guided by the assumption that a feature transfer from a source to a target language should not be expected to yield a complete identity of the associated linguistic structures (see, for example, Boretzky (1983) for a similar view regarding African substrates in Atlantic creoles). Instead of requiring that the transferred feature turns up in Khoekhoe as a neat copy of the Tuu original, it may be sufficient to show that it has been integrated as a basic structural pattern/principle, provided this identifies Tuu as the most probable donor in the genealogical and geographical context. While this may sound vague, the idea will, I hope, become clearer when the individual isoglosses are discussed.
2
Phonological isoglosses between !Ui and Khoekhoe
To begin with, I will discuss structural similarities in the phonological inventories of !Ui and Khoekhoe. As indicated above, there are serious gaps in the documentation of most languages of the Cape and this concerns in particular the sound systems of the extinct languages recorded in the nineteenth and early twentieth century. However, the data on the consonant systems of at least ⏐Xam5 and !Ora seem to allow an approximate comparison between parts of the two language groups. Tables 5.1 and 5.2 display the consonants of Khoekhoe (conflating Namibian Standard Khoekhoe and !Ora) and ⏐Xam, respectively. For a better comparison, the phonemes are given in a unitary transliteration. The phonological features are abbreviated as: Al alveolar, As aspiration, Dt dental, E egressive (= non-click), Gl glottal(ization), I ingressive (= click), Lb labial, Lt lateral, Pl palatal, Vl velar. The consonant systems of Khoekhoe and ⏐Xam have an overall similar organization. The major difference concerns the inventory size, which, however, can be explained by just three salient distinctions: early Khoekhoe (like !Ora) lacks the egressive non-homorganic cluster t(s)x, the labial click influx ◎; and the click accompaniment kh.6 There are still other similarities between the attested !Ui and Khoekhoe varieties of the Cape which single them out against the rest of their respective
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
106
Tom Güldemann
107
I Pl
E Vl
E Gl
x
h ‘
Table 5.1 The Khoekhoe consonant system
Non-nasal sonorants (1) Plain Fricatives (3) Plain Simple stops (7 + 8) Plain Voiced Complex stops (4 + 8) Plain + Gl Plain + As Stop clusters (0 + 8) Plain + /x/ Plain + /k’/~[kx’] Nasals (2 + 4) Plain
E Lb
E Al
I Lt
I Dt
I Al
r~l s p (b)
t (d)
||
(g || )
⏐ (g⏐)
! (g!)
c (g c)
k (g)
(ts’) || h
|| ’ ⏐h
⏐’ !h
!’ ch
c’ (kh)~kx
(kx’)
(th)~ts
!x (!kx’)
cx (c kx’)
n⏐
n!
|| x ⏐x (|| kx’) (⏐kx’) m
n
n ||
nc
Note: Phonemes in parentheses restricted to !Ora. See text on page 106 for list of abbreviations used.
Table 5.2 The ⏐ Xam consonant system (preliminary) (21 + 37) Non-nasal sonorants (3) Plain Fricatives (3) Plain Simple stops (7 + 10) Plain Voiced Complex stops (4 + 10) Plain + Gl Plain + As Stop clusters (1 + 12) Plain + /x/ Plain + /k’/~[kx’] Plain + /kh/~[kx] Nasals (3 + 5) Plain Note:
E Lb
E Al
I Lt
w
r~l
y
I Dt
I Al
I Pl
I Lb
s (p) b
m
E Vl
E Gl
x
h ‘
t d
g ||
||
⏐ g⏐
! g!
c gc
◎ g◎
k g
t’~ts’ ts(h)
|| ‘ || h
⏐’ ⏐h
!’ !h
c’ ch
◎’ ◎h
kx’ kx
tx~tsx
|| x || kx’ || kh
⏐x ⏐kx’ ⏐kh
!x !kx’ !kh
cx c kx’ c kh
n
n ||
n⏐
n!
nc
n◎
ng
See text on page 106 for list of abbrevations used.
families. In order to demonstrate this, I present a comparison of the size of the consonant inventory and the proportion of non-clicks versus clicks across Khoe (see Table 5.3) and Tuu (see Table 5.4). Comparing Tables 5.3 and 5.4, a first commonality between !Ui and Khoekhoe in the Cape
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(17 + 28)
108
The Cape as a Linguistic Area
Table 5.3 Consonant inventories across the Khoe family
1 Non-clicks 2 Clicks 3 Total 4 Proportion 1/2
Khoekhoe
North Kua
Kxoe
Naro
!Ora
Nama
36 30 66 1.20
33 36 69 0.90
24 32 56 0.75
17 28 45 0.61
12 20 32 0.60
Table 5.4 Consonant inventories across the Tuu family
1 Non-clicks 2 Clicks 3 Total 4 Proportion 1/2
Taa East !Xõo
⏐⏐ Xegwi
!Ui
43 83 126 0.52
48 22 70 2.18
c Khomani 27 41 68 0.66
⏐ Xam
21 37 58 0.57
(in italics) is that the size of their consonant inventories is smaller vis-à-vis their respective genealogical relatives (see line 3 in the tables), whereby the lowest figures are found in the Lower Orange area, with Nama on the Khoe side and ⏐Xam on the Tuu side. This phenomenon is in particular a result of a restricted system of egressive (= non-click) consonants (see line 1 in the tables). Despite the relatively small consonant inventory, clicks as a segment type have a high phonological load in languages of the Cape, which can inter alia be seen by the proportion of non-click versus click phonemes (see line 4, low figure = high click load and vice versa). Table 5.3 shows that Khoekhoe has the lowest figures within the Khoe family and approaches the generally low figures of the Tuu family in Table 5.47 – an observation already made by Traill (1980: 170–1). A possible historical explanation for these two shared features is that !Ui languages of the area had a more limited consonant inventory, especially of egressive non-clicks vis-à-vis ingressive clicks. In contact with these languages, Khoekhoe developed in a similar direction: it decreased the number of phonemes, but more so in the domain of egressives, and thereby increased the importance of click phonemes. A possible Khoekhoe–!Ui isogloss relating to the observation that aspirated alveolar /th/ and velar /kh/ are virtually universal across Khoisan might exist in the sub-area of the Cape around the Lower and Middle Orange River. Nama (like Standard Khoekhoe and in part Eini) lacks clearly
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Kalahari
109
aspirated egressives. However, Beach (1938: 218–21) has shown that these sounds are present from a historical perspective in that the plosives /th/ and /kh/ attested in !Ora have undergone lenition to the affricates /ts/ and /kx/ in North Khoekhoe. This seems to have a partial parallel in !Ui varieties of the ⏐Xam and N || ng clusters on both sides of the Orange River (Güldemann, forthcoming c). The lack of aspirates has been observed by Traill (1997: 7) as an anomaly in c Khomani. For ⏐Xam, it can be assumed that at least some affricates go back to earlier aspirates, because there are cognate pairs such as ⏐Xam tsaa versus East !Xõo thaa ‘thing’. The change of /th/ > /ts/ (via /tsh/) and /kh/ > /kx/ (via /kxh/) is not the only case of the fricativizing lenition of certain egressive stops. Ejective egressive consonants in !Ora and ⏐Xam are also affricates (the presence of / kx’/ instead of /k’/ is in fact the normal case in Khoisan as a whole). Ejectives and aspirates can be seen within Khoisan as a natural class (called in Tables 5.1 and 5.2 ‘complex’ stops) in that they are in terms of phonetic elaboration between ‘simple’ and ‘cluster’ segments. It seems, then, that languages along the Lower Orange River have extended fricative lenition to all complex plosives, irrespective of their place of articulation (alveolar and velar) and elaboration gesture (aspiration and glottalization), thereby replacing in a more or less complete fashion ejective and aspirated plosives by ejective and aspirated affricates. A final trait shared by Nama and ⏐Xam but not yet reported in other languages relates to the sound system outside the normal language register. Both languages use typical speech styles to represent some animal characters in their oral literature; in these special registers, certain phonemes are replaced in a regular way by other sounds, some of them not occurring in the normal phoneme system (see Schultze (1907: 390–1) for Nama; and Bleek (1936) for ⏐Xam).
3
Lexical influences of Tuu in Khoekhoe
Affinities in vocabulary between certain varieties of Khoekhoe and Tuu have been observed since the earliest research on Khoisan; at that time, these were predominantly viewed as the result of borrowing from Khoekhoe into San languages. However, I have remarked above that lexical comparisons in the area at issue suffer seriously from the lack of representative and reliable data and, as a result of this, from the difficulty of determining conclusively the direction of transfer. That is, the above interpretation may be relevant for certain lexical isoglosses, but is not stipulated by the data by default. I shall not attempt an extensive and detailed analysis of lexical data here, but confine myself to presenting evidence that the opposite borrowing direction is also relevant. In other words, pace Köhler (1973/4: 192), there are good indications that a certain portion of the Khoekhoe lexicon can be explained plausibly in line with the present Tuu substrate hypothesis.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
Table 5.5 gives a selection of such lexical items. There are different reasons for assuming that the items in Table 5.5 were borrowed from Tuu into Khoekhoe and not the other way around. More generally, these lexemes have not been reconstructed for Kalahari Khoe (see Vossen, 1997), while their forms in Khoekhoe are largely homogeneous; these observations suggest that the items are Proto-Khoekhoe innovations. East !Xõo from the Tuu family is geographically far from the Cape and Namibia and is thus unlikely to have been subject to considerable Khoekhoe influence. Hence, all lexical series involving this language (see lines 1, 2, 4, 6, 7, 10) favour a reconstruction of the word for Proto-Tuu, making Khoekhoe more of a receiver than a donor. In cases where a reflex of a given word is found in more than one !Ui language (see lines 2–6, 8, 10), it is also more plausible to reconstruct a Proto-!Ui form rather than assuming that the relevant languages borrowed the same item independently from Khoekhoe. There are also structural considerations in support of the Tuu substrate hypothesis. Sometimes the Tuu forms have a more complex sound shape than the Khoekhoe ones, which could be motivated by phonetic simplification during the nativization of loans in Khoekhoe (see lines 3 and 8, where an original click cluster ⏐⏐kx’, as attested in !Ui, has been simplified to ⏐⏐’ or ⏐⏐x in Khoekhoe – a common process in some varieties). In other cases, the Tuu forms are morphologically more complex, which would be difficult to explain, if they had a shallow history in the respective languages (see line 1, where the Tuu forms have endings that are semantically not transparent, or line 4, where across the whole family the Tuu forms participate in a regular Table 5.5 Probable lexical borrowings of Khoekhoe from Tuu Khoekhoe North Lexeme 1 again, also 2 beard 3 cheetah 4 come, go to 5 do, make 6 help 7 knock 8 monkey, small 9 mosquito 10 sense, feel
Tuu
South Nama/Eini
|| xa-
!Ui
Taa
!Ora
N ⏐⏐ ng
⏐ Xam
|| xa-
–
|| xa(m) (vb.) –
|| xale (vb.)
n⏐um !kx’auru si, saa
– – see, saa
n⏐um – sii, sa-
tii uie !hun’a hit || hore
– uhi !huu-!huu –
– thã
– tã~ a
n⏐om (vb.) n⏐um- n⏐um !’aru-(ru)- !’aru-ru- !kx’aru sii sii si, sa dii hui !huu pound || ’ore-
dii hui !huu || xori-
– – –
di nhu‘i – || kx’ore –
ts’uru-rutsã
ts’uru-ru- – thã tjha’n
ts’utu tãa
c Ungkue
East !Xõo
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
110
111
stem alternation). Another even clearer indication for the validity of the explanation proposed here exists when Khoekhoe forms display a suffix -RV (see lines 3, 9); this element attested so far in Eini and !Ora identifies borrowings in general (that is, also Bantu loans) and has clearly been added in the above cases to a more simple Tuu form. These few examples will suffice to show that lexical evidence does support the general hypothesis of a Tuu substrate in Khoekhoe.
4 Morphosyntactic Khoekhoe innovations and their explanation in terms of substrate interference from Tuu As mentioned above, the morphosyntax of Khoekhoe is viewed here as the domain where linguistic influences by Tuu languages can be felt most strongly. The following section presents a number of isoglosses in support of this hypothesis. They pertain to such different areas as pronominal, nominal and verbal morphology; phrase and clause syntax; and clause linkage. Their sheer number, the fact that many of them are independent of each other, and the observation that several involve similarity in linguistic substance make it unlikely that the present substrate hypothesis is spurious. 4.1 Pronominal system A first domain where linguistic interference from Tuu into Khoekhoe can be diagnosed has been treated already by Güldemann (2002) and concerns the pronoun system. Pronominal marking in Khoekhoe is fairly complex in that there are three paradigms of markers, each of which occurs in its particular array of morphosyntactic contexts: (a) an elaborate system of persongender-number enclitics (henceforth just PGNs), inherited from Proto-Khoe; (b) a smaller set of four pronoun bases, two of which can be traced back to Proto-Khoe; and (c) a paradigm of free pronouns, formed by the combination of an initial pronoun base and a final PGN and behave syntactically like nouns (see inter alia Haacke (1977); Vossen (1997); Güldemann (2004a) for more information on synchronic and diachronic aspects of all the markers). Table 5.6 gives the free pronouns of !Ora (the overall situation in North Khoekhoe is identical). Such an elaborate set of complex forms cannot be reconstructed for Proto-Khoe; it can be observed instead that the relevance of this pronoun type increases within Khoe according to a north–south cline, with the climax found in the Khoekhoe branch. This seems to be related to the widespread presence in Non-Khoe of complex pronominals that are achieved by so-called ‘pronoun modification’ (see Güldemann, forthcoming a). The way Khoekhoe may have been influenced by this phenomenon will not be discussed here, however. Of relevance for the present discussion are (a) the pronoun bases for first-person exclusive and third-person forms (highlighted in bold); and (b) the way the exclusive– inclusive distinction is expressed (see the first and second columns).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
112
The Cape as a Linguistic Area
Table 5.6 Free, morphologically complex pronouns of !Ora (Meinhof, 1930, p. 43) Gender
Person 1st person exclusive
Singular
Plural
Dual
Common Feminine Masculine Common Feminine Masculine Common Feminine Masculine
//’ãi-’i ti-ta ti-re si-da si-sj si-tjj si-m si-sam si-kham
1st person inclusive
sa-da sa-sj sa-tjj sa-m sa-sam sa-kham
2nd person
3rd person
sa-s sa-ts sa-du sa-sao sa-kao sa-khao sa-saro sa-kharo
//’ãi-s //’ãi-b //’ãi-n //’ãi-dj //’ãi-ku //’ãi-kha //’ãi-sara //’ãi-khara
Firm Proto-Khoe reconstructions only exist for two of the four pronoun bases in Khoekhoe, *ti encoding speaker and *sa encoding addressee. The first-person exclusive *si is virtually restricted to Khoekhoe and the thirdperson *⏐⏐’ãi absolutely so; and both occur only in the complex free pronouns given in Table 5.6. Güldemann (2002), to which the reader is referred for more details, argues that both *⏐⏐’ãi and *si were borrowed by Proto-Khoekhoe from a Tuu language, most probably of the !Ui branch. The third-person pronoun base *⏐⏐’ãi, which today functions mainly as an unmarked anaphoric pronominal, can be shown within Khoekhoe to derive from a reflexive and discourse-anaphoric intensifier ‘same, self’. Precisely such an element is attested in the !Ui language ⏐Xam. That the borrowing went from !Ui into Khoekhoe can inter alia be discerned from the fact that the intensifier in !Ui can in turn be traced back to a verb ‘have, own’ in Proto-Tuu. The Khoekhoe pronoun base *sii for first-person exclusive is viewed as a borrowing of the Proto-Tuu first-person exclusive pronoun *si. As can be seen in Table 5.6, *sii is intimately associated with the exclusive–inclusive opposition, because this arises (and only there) between two complex pronoun series sharing the first-person non-singular PGN enclitics: exclusive [*sii + PGN] versus inclusive [*saa + PGN]. Since *saa comes from Proto-Khoe *sa encoding the addressee, the relevant complexes are literally inclusive ‘you + we’; the Tuu borrowing *sii as pronoun base, on the other hand, renders complexes that are literally ‘exclusive we + we’. In general, formal and systemic properties of free complex pronouns suggest strongly that Khoekhoe not only innovated the categorial opposition exclusive–inclusive, but also borrowed linguistic substance from Tuu, integrating it creatively in its inherited system.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Number
Tom Güldemann
113
The Kalahari and Khoekhoe branches of Khoe differ in their functional and formal properties of grammatical gender. In Kalahari Khoe, gender can be expressed on nouns by means of third-person PGNs, but this is far from obligatory; in some languages, PGNs on nouns are in fact rare and the association between a noun and a gender is exploited for derivational processes (see Vossen (1986) on Naro; Kilian-Hatz and Heine (1998) and Heine (2000) on Kxoe). While the derivational use of gender has been retained in Khoekhoe, there is a clear tendency for a lexeme to have a default gender and the PGN marking on nouns as such has become virtually obligatory (with the exception of a restricted and motivated set of contexts). I propose that the increased fixation of gender with noun lexemes vis-à-vis Kalahari Khoe relates to contact with Tuu. These languages, as far as they possess a gender system, have lexically fixed gender. Khoekhoe would have taken over this principle, thereby also grammaticalizing the regular PGN marking on nouns. There is also a concrete semantic detail in Khoekhoe that seems to corroborate this hypothesis. Cross-linguistically as well as in some Kalahari Khoe languages (see Vossen (1986) for Naro), feminine gender is associated with small size – a feature exploited for deriving diminutive nouns by means of feminine gender assignment. The situation is more complicated in Khoekhoe, as discussed by Güldemann (1999): the mere switch of the default gender can denote that the referent of a given noun lexeme has some unusual property (for example, in size/shape), and feminine gender can thus also come to express an entity that is large or voluminous, as shown in (1) below: (1)
Standard Khoekhoe pén-i versus pén-s pen-M.S pen-F.S ‘the pen’ ‘the unusually fat pen’ (Hagman, 1977: 23)
The possible semantic connection between feminine and great size seems to be influenced by a phenomenon attested at least in the Tuu language East !Xõo, but which might well be of more general importance in the area. Traill (1994: 177) writes regarding the feminine marker qáe: ‘When suffixed to plant names it signifies a broader-leaved more substantial variety; with certain other objects it signifies more substantial size, weight.’ The presence of this linguistic property in contact languages can be assumed to have facilitated the acceptance in Khoekhoe of a typologically highly unusual association between feminine gender and greater size. 4.3 Nominal derivation Another feature distinguishing at least North Khoekhoe from Kalahari Khoe is an extended use of derivation usually reserved for nouns. Example (2)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4.2 Nominal gender
114
The Cape as a Linguistic Area
(2)
Standard Khoekhoe (a) !ó+-ró tào-p difficult-DIM path-M.S ‘the slightly difficult road’ (b)
!hóá-ró c. tií-ro-ta converse-DIM SP-DIM-1SG ‘converse a bit little me’ (Hagman, 1977: 33, 74, 45)
Nominal derivation markers are also found in !Ui to apply to hosts other than nouns. The diminutive suffix -◎ua (derived from a Proto-Tuu noun for ‘child’; see Bleek (1928–30: 97; 1956: 243)) is used in ⏐Xam on attributive adnominal predicates and in the form tam◎ua also with sentential predicates, as shown in (3) below: (3)
⏐Xam (a) !wãa aa c eni-◎ua child.1 1REL be.small-DIM ‘ein kleines Kind’ [a child which is smallish] (Meriggi, 1928/9: 146) (b)
⏐5a-/5a-ten tam◎ua c xii-a P-star-P DIM shine-STAT ‘the Stars shine a little’ (Bleek and Lloyd, 1911: 74–5)
The phenomenon of pronoun modification in Non-Khoe has been mentioned above. It reflects the similar treatment of nouns and pronouns, including derivational suffixing. The example in (4) from a northern N || ng variety (!Ui), where a feminine marker xae modifies the first-person singular pronoun, suggests that pronoun derivation also applies to some Tuu languages. This would mean that the whole range of non-canonical uses of the diminutive suffix -ro in North Khoekhoe have partial precedents in Tuu. (4) N⏐huki n xae ke dy4‘an 1SG F DECL walk ‘I (feminine) go.’ (Westphal field notes)
4.4 Nominal co-ordination Nominal co-ordination in Khoekhoe is characterized by the following morphosyntactic structure: [Nominal-(PGN) tsii Nominal-(PGN) tsii-PGN] in
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
below demonstrates that the diminutive suffix -ro also occurs on adjectival, verbal and pronominal hosts.8
Tom Güldemann
115
(5)
Standard Khoekhoe (a) tií-ta ts6i saá-ts ts6i-m SP-1SG and AD-2M.Sand-1C.D ‘I and you’ (masc., female speaker) (Hagman, 1977: 49) (b)
‘áo-p ts6i tará-s ts6i-rà man-M.S and woman-F.S and-3C.D ‘the man and the woman’ (Hagman, 1977: 48)
This fairly complex strategy is apparently a combination of two features that are in principle independent of each other. The first phenomenon is the possibility of plural pronominals to appear in a so-called ‘inclusory’ use (Lichtenberk, 2000). This seems to be a more widespread trait in the area, especially in dual conjunctions (see Dickens (n.d.: 33) for the Ju language Ju⏐’hoan, and Heine (1999: 68) for the Kalahari Khoe language || Ani). It also applies to Khoekhoe, as shown in (6) below: (6)
!Ora ti tara-s tsi-kham 1SG woman-F.S and-1M.D ‘meine Frau und ich’ [my wife and I, not: my wife and we two] (Meinhof, 1930: 45)
This use of inclusory pronouns seems to have been generalized in Khoekhoe to all types of nominal co-ordination. This process can be motivated by interference from a co-ordination pattern that is described repeatedly from !Ui languages. Bleek (1928–30: 172) reports for ⏐Xam: ‘Where there are several subjects they are first enumerated, then repeated by a pronoun.’ Meinhof (1929: 169) mentions the same pattern in c Ungkue and gives the example in (7) (the pronoun referring to the entire conjunct is n) below: (7)
c Ungkue !hoeti nan° koro nan tuj n a //’a lion and jackal and ostrich 3PL ?PST go ‘Der Löwe und der Schakal und der Strauß sie gingen.’ [Lion, jackal and ostrich they went] (Meinhof, 1929: 169)
For the complex Khoekhoe structure, it can be assumed that the inclusory pronouns, presumably established before, were extended to non-dual
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
which the final PGN refers to the referential sum of the conjunct (see Hagman (1977: 48–50) and Haacke (1992) for Standard Khoekhoe, and Meinhof (1930: 44–5) for !Ora). The examples below involve the co-ordination of both pronouns and nouns (the crucial PGN is highlighted in bold).
116
The Cape as a Linguistic Area
co-ordination in line with the final ‘summarizing’ pronouns found in nominal conjuncts of !Ui languages.
The verbal derivation system of Proto-Khoe has been reconstructed to a considerable extent by Vossen (1997: 349–55). For the present discussion it is relevant that verb root reduplication served to derive iterative and causative forms. Causative reduplication, which is also found in the area outside the Khoe family, existed side by side with a suffix *-ka(xu), which is presumably older. The situation in Khoekhoe deviates from this picture in several respects. First, all, verb root reduplication has an overall higher functional load in derivation. For North Khoekhoe, Haacke (1999: 133–9) identifies several types of reduplication distinguished by their suprasegmental patterns; their functions are causative, intransitive inchoative ‘become’, and (in combination with reflexive -sen) pretence ‘make oneself as if’. Moreover, causativization is encoded only by means of reduplication and a more lexicalized device not found in Kalahari Khoe, namely the suffix -(s)i (see Haacke (1999: 144–5) for Standard Khoekhoe, and Meinhof (1930: 48) for !Ora); the Proto-Khoe ka-causative is not attested in Khoekhoe. Although the data are still insufficient to state this with more confidence, it is probable that the formal shift in verbal derivation in Khoekhoe can also be attributed to Tuu influence. First, !Ui varieties, like ⏐Xam, show a ubiquitous use of reduplication, most probably also involving different tonal patterns; functions attested so far are (a) nominal plural; (b) deverbal nominalization; (c) verbal causative; and (d) verbal iterative-intensive. Second, East !Xõo possesses a postverbal and a preverbal sí (sV) which are both transitivizing (Traill, 1994: 30, 185); it is worth investigating whether a cognate of such an element in Tuu was the source of the innovated Khoekhoe causative -(s)i. This new suffix and, even more so, the increased use of reduplication might have ousted in Khoekhoe the older Khoe causative in -ka. 4.6 Predicate operators The predicate marking of time displays considerable differences between Kalahari Khoe and Khoekhoe regarding type and position of grams as well as their basic functions. Old Damara varieties also deviated significantly from the modern Khoekhoe pattern (Vedder, 1923: 159–60). That Damara was different before its contact with the language(s) of the pastoral Khoekhoe suggests that the modern predicate marking reflects the original situation in Nama and other more southern Khoekhoe varieties. Time encoding of the predicate is achieved in Kalahari Khoe mainly by verbal suffixes (involving the so-called ‘juncture’ morphemes) and phrase-final auxiliaries. In contrast, Khoekhoe makes exclusive use of particles which, with one important exception, occur in unmarked clauses before the verb phrase.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4.5 Verbal derivation
Tom Güldemann
117
Table 5.7 Basic time marking in selected Khoekhoe and Tuu languages
Nama
Perfective
Tense Ø Verb phrase !Ora Tense Ø Verb phrase ⏐Xam Tense Ø Verb phrase East !Xõo Tense Ø Verb phrase
Imperfective
Relevance/ stative
Tense
ra
Verb phrase Tense
Tense
na Verb phrase Tense
Tense
n/e Verb phrase Tense
Tense
ba Verb phrase Tense
Verb phrase Verb phrase Verb
hã hã/-a -a
Verb /îi (phrase)
In terms of functional distinctions, only tense, but not aspect, has been reconstructed for Proto-Khoe (Vossen 1997: 360–5; Güldemann and Vossen, 2000: 116), while Khoekhoe clearly displays an aspect distinction with the pattern of unmarked perfective versus marked imperfective.9 The unmarked position of the imperfective gram is after the tense particles, but before the verb phrase. Moreover, there is one consistently phrase-final marker, the stative-relevance particle hã, which goes back to the existential verb ‘stay, be (t)here, exist’. Beach (1938: 192, n. 2) reports that hã directly after the verb was reduced by some !Ora speakers to -a. This profile of time marking in Khoekhoe matches fairly closely the patterns in the better-known Tuu languages East !Xõo and ⏐Xam. Here, one also finds an opposition between a zero perfective and an overt imperfective, the latter placed between tense gram and verb, as well as a postverbal stativeresultative particle derived from a verb meaning ‘be present, be (t)here, exist’. The structural parallels between Khoekhoe and Tuu can be seen in Table 5.7, where the three basic categories and their marking are displayed for the four better-known languages. Compare also the formal proximity between postverbal -a in !Ora on the one hand and in ⏐Xam (and other !Ui varieties) on the other. 4.7 Predicate formation Compared to canonical Kalahari Khoe languages, Khoekhoe is characterized by a heavy reliance on lexically complex predicates. That is, an extensive portion of the verbal lexicon is made up of compound verbs; they are mostly binary (hence Maingard’s (1962: 30) term ‘double verb’), but three or more verb stems can also occur. This phenomenon is a powerful addition to already existing strategies inherited from Khoe for the expression of lexical concepts and the marking of grammatical relations, namely verbal derivation and postpositional phrases.10 The compound verb formation in Khoekhoe can be related directly to ubiquitous verb serialization and verb compounding in Tuu languages
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Language
The Cape as a Linguistic Area
(a general feature of Non-Khoe) in that different predicate types in the Tuu substrate developed in Khoekhoe in a more restricted direction, namely lexicalized compound verbs. The variability of these complex predicates, however, eclipses that of their Tuu model, since compound verbs in Khoekhoe can also incorporate nouns, adpositions, and adverbs (see Haacke, 1995). While it is conceivable to identify calqued patterns and loan translations, examples as in (8) below might be viewed as non-diagnostic for a specifically Khoekhoe-Tuu contact, because they are arguably universal; more crosslinguistic research is needed here. (8)
Khoekhoe !Ui bury khao- c aa dig-put.in (Nama) //au/’ee take uu- c ’ui take-out (Nama) /ii/hing out bring uu-haa take-come (!Ora) ⏐ii tsaa
dig put.in (N⏐huki) take go.out (⏐Xam) take come
(⏐Xam)
However, the parallels also include more marked predicate types. Thus North Khoekhoe and ⏐Xam share a pattern that is thus far unprecedented in the rest of Khoe (including !Ora) and Tuu. Haacke (1995) reports for Namibian Standard Khoekhoe a type of mono-clausal sentence exemplified in (9) below: an intransitive verb (here, !goaxa ‘approach’) is incorporated in a higher verb (mû ‘see’) and the subject of the former (audos ‘car’) becomes the object of the main predicate. The remarkable phenomenon is that the matrix is a perception verb so that this form of subject raising is not causative. (9)
Standard Khoekhoe audo-s-a ra mû !goaxa car-F.S-OBJ IPFV see approach ‘See a car approaching.’ (Haacke, 1995: 357)
The origin of this structure can be explained by calquing from ⏐Xam, where verb serialization is of the ‘nuclear’ or ‘root’ type (as opposed to ‘core’ serialization; see Foley and Van Valin, 1984).11 That is, all verb stems create an uninterrupted chain followed by all nominal non-subject terms (Güldemann, forthcoming b). Thus, a meaning ‘see X doing’ is expressed in a surface sequence [see do X] as in (10) below: (10)
⏐Xam si tang //’a-ng d7a n/ii t0e !k’waa aa /uuk-a 1PL.E ? go-? ? see lie hartebeest.1 1REL die-STAT ‘We did see a dead hartebeest lying there!’ (Bleek and Lloyd, 1911: 10–11)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
118
Tom Güldemann
119
A comparison of the structure schemas in (11) below makes the similarities and differences between Khoekhoe and ⏐Xam more transparent. ⏐Xam [see do] X Khoekhoe X [see do] German see [X do] or [X do] see
The two relevant order patterns available in German are also given in (11) in order to show the distinctness of the Lower Orange languages, namely that the primary constituent is formed by the two verbs, not – as in German – by the lower verb and its subject. The position of this noun phrase is the major difference between the Khoekhoe and ⏐Xam structures. This is not surprising, though: Standard Khoekhoe conforms to the verb-final word order inherited from Khoe, while ⏐Xam is consistently verb-medial, like all Tuu languages. On a more abstract level, then, the structures in the two languages are identical. The pattern in North Khoekhoe, which is typologically rare, is historically the result of reanalysis of a verb serialization type in ⏐Xam towards a fixed compound. 4.8 Clause syntax and sentence-type marking Little is known as yet about the clause syntax of Kalahari Khoe languages. It is certain, however, that Khoekhoe represents a syntactic type of its own. This is particularly clear in North Khoekhoe. Here, a marked configuration has been generalized to the structure of basic clauses: a PGN-enclitic referring to the subject, often in conjunction with following sentence type markers, serves as a syntactic pivot establishing a clause-second position of the ‘Wackernagel’ type (in bold type in the examples). This results in an obligatory bisection of the clause into a prefield and a postfield. One salient function of the prefield is to host pragmatically sensitive constituents such as contrastive or assertive foci, topics and subject topics (see Dempwolff, 1927). The following examples demonstrate several variants of this theme. In (12a) below, the prefield contains a focused object followed by the subject PGN da. In examples (12b, c), the subject PGNs -s and -ta precede the highly grammaticalized declarative marker ge~gye (see Hagman, 1977: 53–4); the prefield fillers are respectively a focused verb and, as a default, a subject topic. Finally, (12d) displays the sequence of a subject PGN -p and the question marker kha [kxa], an emphatic supplement to interrogative intonation and in complementary distribution with ge~gye (Hagman, 1977: 139–44). (12)
Standard Khoekhoe (a) x{-e da do toma thing-C.S: OBJ 1C.P.SBJ do NEG ‘Wir tun nichts.’ [We do nothing] (Dempwolff, 1934/5: 91)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(11)
The Cape as a Linguistic Area
(b)
//oo-s ge go die-3F.S.SBJ DECL PST ‘She died.’ (Haacke and Güldemann, field notes)
(c)
ti-ta gye //s tite 1SG-1SG.SBJ DECL die NEG.FUT ‘Ich werde nicht sterben.’ [I won’t die] (Dempwolff, 1934/5: 53)
(d)
c ’8u-p kxa kè//’6i-p-à //án-’è eat-3M.S.SBJ PQ PST 3–M.S-DSBJ meat-3C.S: OBJ ‘Did he eat the meat?’ (Hagman, 1977: 143)
The structure with a clause-second pivot was not fully developed in ProtoKhoekhoe, but seems, in this rigid form, to be an innovation of North Khoekhoe. In !Ora, it is present but far less salient, because (a) several frequent clause patterns display a deletion of the subject; and (b) the particle kye~tye, the cognate of the Nama declarative marker, did not grammaticalize in the same way; according to Wallmann (1857: 33), the latter fact also applies to the speech varieties of the Orlam, who contained a large portion of South Khoekhoe speakers and merged with the Nama in Namibia. The following examples show, however, that !Ora does display a latent clause-second position (see the subject PGNs in bold) and that the constituents in the prefield have a similar profile to that in North Khoekhoe: a focused object in (13a), an interrogative word in the constituent question of (13b), a verb in the polar question of (13c), and a verb in the obligation clause of (13d) (which is a context where the particle kye~tye is obligatory). (13)
!Ora (a) !xo-bi-r na c ae pipe-M.S-1M.S.SBJ IPFV smoke ‘Eine pfeife rauche ich.’ [I smoke a pipe] (Meinhof, 1930: 60) (b)
hama-ts koko c ’u-b ho hã where-2M.S.SBJ PST eat-M.S find RELV ‘Wo hast du Speise gefunden?’ [Where have you found food?] (Meinhof, 1930: 61)
(c)
!hami !u kao ka na hunt go 2C.D.SBJ want IPFV ‘Wollt ihr jagen gehen?’ [Do you want to go hunting?] (Meinhof, 1930: 61)
(d)
⏐⏐xara-e-b tje ni punish-PASS-3M.S.SBJ DECL OBL ‘Er muß bestraft werden.’ [He has to be punished] (Meinhof, 1930: 53)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
120
121
Vedder (1923: 159, 161) reports that clause bisection and a declarative marker of the Nama type were not present in early Damara varieties of Namibia, which suggests again that modern speech varieties of the Damara are the result of Khoekhoeization. Clause structures reminiscent of the North Khoekhoe pattern can be identified in Tuu, significantly with a geographical peak in !Ui along the Lower Orange, the original distribution area of Nama. According to Güldemann (forthcoming b), sentence-type markers establish in ⏐Xam a clause-second position in unmarked sentences; see the grammaticalized declarative marker ken~gen in (14a) below and the interrogative particle ba in (14b) (see also xa in (15b)). (14)
⏐Xam (a) au
too-gen n/e !ii-ya red.ochre-DECL IPFV be.red-STAT ‘But ochre is red.’ (Bleek and Lloyd, 1911: 346–7) CONN
(b)
a ba /’uru-wa ha !khwãa 2SG PQ forget-STAT 1DEI child.1 ‘Hast thou forgotten this child?’ (Bleek, 1928–30: 167)
⏐Xam also possesses a type of bisected clause – fairly frequent in discourse – in which a special pronominal occurs in clause-second position; this agrees with a pragmatically sensitive constituent in the prefield. This is shown in (15a) and (15b); the second clause demonstrates that the pronoun can co-occur with an adjacent sentence type marker, here interrogative xa. (15) ⏐Xam (a) au
!u-koo
CONN person.
aa se //xam-ki /u-wa ha ãa //hara 1PRO SUBJ do.also-? put-TR 1PRO MPO:1PRO specularite.2
1-other ‘The other onex should in turn put aside for himy hisx specularite.’ (Bleek and Lloyd, 1911: 376–7) (b)
!udi xa aa n/aa !utau who.1 Q 1PRO see sirius ‘Who was it who saw Sirius?’ (Bleek and Lloyd, 1911: 338–9)
Thus, it can be diagnosed that a pronoun is used as the syntactic pivot in bisected clauses with a marked information structure. This can be compared to the occurrence of subject PGNs in the ‘Wackernagel’ position of Khoekhoe, a structural pattern which, looking at the !Ora data, was also used originally for fronting marked constituents.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
122
The Cape as a Linguistic Area
(16)
N⏐huki (a) c ou a ke c hoa n/a n anci man this DECL speak with 1SG father ‘This man speaks with my father.’ (Westphal, field notes) (b)
ki dya xae Dorki n//a where Q PROP stay ‘Where is Dorkie?’ (Westphal, field notes)
I argue here that sentence type markers and pronouns placed after subjects or fronted constituents in !Ui languages served as important structural input for the emergence of the ‘Wackernagel’ phenomena in Khoekhoe. Moreover, its declarative marker is a probable borrowing of !Ui ke(n). For the latter hypothesis, there exists concrete evidence: declarative ke(n) has a plausible grammaticalization history in !Ui which would be untenable if the borrowing direction was from Khoekhoe to !Ui; ke(n) presumably goes back to an identificational copula as shown in the following example (17) from N || ng. (17)
N⏐huki n/u tyu ke diviner COP ‘It is a diviner.’ (Westphal, field notes)
4.9 Marking of nominal participants Another remarkable feature restricted to North Khoekhoe might also be influenced by syntactic properties of Tuu languages. Virtually all noun phrases in the postfield after the subject PGN are marked in Standard Khoekhoe by a suffix -a. This applies irrespective of the semantic role, that is, to objects, obliques, adverbial adjuncts, and even so-called ‘nachlaufende’ / ‘deposed’ nominal subjects. This principle, exemplified in (18) below, can be schematized as [Prefield–Wackernagel slot–NP-a–NP-a– . . . –Verb]. (18) Standard Khoekhoe o-gu gye /gawi-priester-ga mari-te { (ga < gu-a te < ti-a) then-3M. P.SBJ DECL high-priest-M.P: DSBJ money-F.P: OBJ take ‘Da nahmen sie, die Hohepriester, das Geld . . .’ [Then they, the high priests took the money] (Dempwolff, 1934/35: 90)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Similar evidence emerges from !Ui varieties of the N || ng cluster north of the Orange. Westphal’s field notes on N⏐huki reveal that declarative and interrogative particles also occur in clause-second position; thus, ke after the subject topic c ou a ‘this man’ in (16a) below, and xae after the question word ki dya ‘where’ in (16b).
123
This is parallel to the equally stringent, syntax-sensitive marking of grammatical relations in Tuu (and other Non-Khoe for that matter; see Güldemann and Vossen, 2000). Here, the following basic pattern applies to all postverbal nominals (= all non-subjects including adverbials of place and time; subjects are always preverbal): irrespective of the semantic role, the first noun phrase is unmarked, while every following one is preceded by a so-called multipurpose oblique (MPO) marker. This gram in ⏐Xam is au so that the predicate pattern is [Verb–NP–au NP–au NP–. . .], as in (19) below: (19)
⏐Xam hi-ng ⏐ueng-ki /‘ee //xauken au’ /o’a au h /kx’aa 2PRO-DECL do.thus-? enter blood MPO stomach MPO 2PRO hand ‘They put the blood in the stomach with their hands like this.’ (Bleek and Lloyd, 1911: 278–9)
The structures in Khoekhoe and Tuu are certainly not identical; they are, however, parallel in one important respect: the morphological treatment of nominal participants is steered exclusively by their linear, syntactic position instead of their semantic role. 4.10 Proposition type clauses A final case where Khoekhoe seems to have innovated vis-à-vis Kalahari Khoe because of interference from Tuu concerns the grams involved in sentential complementation and reported discourse. Conforming with the general head-final syntax of Khoe, the relevant clause linkers, the complementizer !xai- in (20a) below and the quotative marker ti in (20b), are postposed to the associated lower clause. (20)
Standard Khoekhoe (a) tsií //’ií-p-à-kxv ke kè mií-pa CLAUSE !xáis-a and 3–M.S-OBJ-1M.D.SBJ decl PST say-APPL COMP-OBJ ‘And we told him that . . .’ (Hagman, 1977: 138) (b) CLAUSE ti-b ge go mî-ba-he QUOT-3M.S.SBJ DECL PST say-APPL-PASS ‘ “. . .”, he is told.’ (Haacke and Güldemann, field notes)
The complementizer !xai- is transparently derived from the noun ‘place, matter’ (marked in (20a) by the third-person feminine singular PGN -s). The quotative gram ti is synchronically a similative marker ‘like’ and, when preceded by the demonstrative nee ‘this’, can also occur in this context as a manner deictic ‘this way, thus, so’. While different in their syntax, these grams are related to elements found in Tuu in general and !Ui in particular, because East !Xõo and ⏐Xam have quotatives-complementizers with precisely such an etymological origin (Güldemann 2001, forthcoming b).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
The default introducer of reported discourse in East !Xõo is té’3; this is likely to originate in a deictic té’àã ‘this place, this way, thus’, which consists of the class-2 noun te and the proximal demonstrative ‘VV agreeing with it in class. The parallel between Khoekhoe and ⏐Xam not only concerns the grammaticalization history, but also the very origin of one marker concerned. The quotative-complementizer ti (ee) of ⏐Xam is also a noun ‘place, way, matter’; it is usually followed by the agreeing relative pronoun ee, as shown in (21a) and (21b) below. I assume that !Ui ti was borrowed into Khoekhoe as the similative-quotative marker ti. (21)
⏐Xam (a) hi-ng c aken ti ee CLAUSE 2PRO-DECL say matter.22REL ‘They spoke thus, “. . .” ’ (Bleek and Lloyd, 1911: 4–5) (b)
u g//kx’o’en-a ha ti eeCLAUSE 2PL ook-TR 1PRO matter.22REL ‘. . . you have looked at him whether . . .’ (Bleek and Lloyd, 1911: 46–7)
A piece of evidence in support of the borrowing hypothesis, which is independent of the clause-linkage domain, is that !Ora has a manner deictic heti ‘like this’. This is identical in form and function with ⏐Xam he ti, in which the deictic pronoun he precedes and agrees in class with the noun ti, thus ‘this matter, this way’. Example (22) below shows this phrase as part of a ⏐Xam discourse connector conveying ‘that’s why, therefore, so’: (22)
he ti-ken ee ... 2DEI matter.2-DECL 2REL Lit.: ‘that (matter) is which . . .’ > ‘that’s why, therefore, so . . .’
In summary, I propose to identify two types of interference from Tuu languages in the domain of sentential complementation and reported discourse of Khoekhoe. First, the recruitment as a clause linker of a noun ‘place, way, matter’ with a semantic component of similarity and manner has been calqued from Tuu (note that although this grammaticalization path is attested cross-linguistically, it is not the default pattern in the wider area). Second, Khoekhoe has borrowed directly the !Ui element ti in its more grammaticalized use as a similative and quotative marker.
5
Conclusions
The linguistic data presented in sections 2–4 are considered here as good evidence for the hypothesis stated in section 1 that Khoisan languages from
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
124
125
two distinct families, Tuu and Khoe, form a linguistic area in the Cape and in particular that Khoekhoe as a whole contains a strong linguistic substrate from the indigenous Tuu languages. Even if, after more research, individual isoglosses identified here turned out to be invalid, the hypothesis as a whole is unlikely to be falsified. So the results of this investigation have important implications for the reconstruction of the early linguistic history in southern Africa before the Bantu expansion as well as for contact linguistics in general. The degree to which Khoekhoe has changed because of Tuu influence vis-àvis its closest relatives in the Kalahari is so high that only a special historical and sociolinguistic scenario can explain it. Apart from this purely linguistic consideration, there are other indications concerning non-linguistic characteristics of the Khoekhoe population that support this idea; many of them have been noted and discussed in the past (see, for example, Schapera, 1926). I will confine myself to mentioning two such features. First, it has been observed that the Khoekhoe are biologically close to their San neighbours.12 This is significant, because Khoe-speaking groups of the northern- and easternmost branches (Shua, Tshwa and Kxoe) do not match in terms of genetics with other Khoisan-speaking groups further south and west, but rather with their Bantu neighbours. Second, the pastoral Khoekhoe are in terms of their subsistence similar to San groups in a major aspect: traditionally, they relied heavily in their diet on the gathering of wild food plants and hunting; this strong foraging component made them one of the few pastoral populations in Africa – possibly the only one – that used to be fully independent of agricultural food production. There are not many scenarios that can explain these salient commonalities between the Khoekhoe and their Tuu-speaking San neighbours. My hypothesis is that the widely assumed expansion of a Khoe-speaking pastoralist population over southern Africa took place at its southernmost frontier according to a pattern in which the incorporation of features of the indigenous population played an essential role not only in linguistic terms but also in other respects. I assume that the Khoe movement at its very periphery relied heavily on the interaction with the original inhabitants of the newly colonized areas. As pioneers, these ‘vanguard’ Khoe will have needed local know-how to enable than to use the resources of the new territory. Moreover, considering more recent social relations in pastoralist–San interaction as well as patterns of pastoral food production in southern Africa, it is quite probable that San women in particular were incorporated into Khoe ethnic units; and these would have acted as major mediators of features that the historically attested Khoekhoe and !Ui groups have in common. Hahn’s (1870: 7) metaphorical view of ‘Buschmänner und Hottentotten [als] Geschwister derselben Mutter, aber ethnisch grundverschiedene Charactere’ [Bushmen (alias San) and Hottentots (alias Khoekhoe) as siblings of one mother, but ethnically entirely different characters] would then have a very real background in history.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
One factor will have been the cultural prestige of pastoralists in general, and the frequent status of San as their clients. In attested cases of intermarriage between foraging and food-producing groups, this and other factors lead to a pattern in which San women become espoused in the prestige group and are integrated there together with their children, but not the other way around. As a result, the gene flow is predominantly unilateral, namely from the San population into the population of the other group, in the case at issue, the Pre-Khoekhoe.13 Another factor might have been the important role of stock posts on the periphery of pastoral settlement areas. These would have been occupied predominantly by men tending the stock and looking for new pastures. In such an environment, the role of San women might even have been greater than in the normal contact situation. Clearly, large parts of the historical development briefly sketched above must remain speculative, but some components thereof might be testable in future research (for example, the above prediction regarding biological profiles through population genetics). For the present, this scenario is a good fit of the empirical facts and thus serves as a promising working hypothesis. It can account for the strong biological affinity between the two populations, the important subsistence mode of hunting and gathering in both groups, and, last but not least, the fact that the striking linguistic deviation of Khoekhoe from the rest of the Khoe family is explained in a plausible way as the result of borrowing and interference through shift from Tuu languages.14 If the present hypothesis is not severely weakened in future research, it is of great importance for the evaluation of the linguistic history of the Khoe family and Khoisan in general. Khoekhoe is a very distinctive subgroup within Khoe, which was plausibly interpreted in the family tree model as the result of a primary branching in the group (see Vossen, 1997). Under the above scenario, however, this differentiation is to a large extent a result of Tuu-mediated innovations in Khoekhoe. Hence, it is possible that the splitoff of the Khoekhoe group is a later event in the divergence history of Khoe. That is, a heavy, but historically secondary, linguistic substrate might skew the modern picture in a family to the extent that other, earlier processes of divergence become less salient. This phenomenon may well be relevant for language families in other parts of the world. From a general Khoisan perspective, the present case is a first concrete indicator that even a large amount of linguistic similarity between different lineages in this group can be explained in terms of contact rather than inheritance from a common ancestor. Some conclusions from this study are also relevant for the research on language contact in general. Khoekhoe has retained a wide array of structural devices inherited from Khoe, such as the PGN system (see section 4.1), the inventory of verbal derivation suffixes (see sections 4.5 and 4.7), and verb-final clause order after the subject pivot (see section 4.8), to mention just a few. In fact, there is no clear case as yet that a novel, contact-induced
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
126
127
structure has entirely replaced an older one. As a corollary, Khoekhoe displays a canonical genealogical affiliation with other Khoe languages, which implies that there has been no interrupted language transmission in the history of this group. In many isoglosses identified above, the interference feature was added to and combined with pre-existing structure. In other words, the transfer was not a simple take-over, but rather a creative incorporation of the new feature. For example, this can be observed with the complex pronouns with borrowed exclusive sii, which contrast with an equally innovative inclusive series based on inherited Khoe material (see section 4.1). Also, in North Khoekhoe the inherited verb-final syntax with a considerable flexibility according to pragmatic functions has been reconciled with the rigid constituent order of Tuu, yielding the structure [Prefield s (S) OV] 15 (see section 4.8). Thus, one cannot diagnose a word-order homogenization, which in general is a fairly typical contact-induced change. Another case in point would be the borrowing of the quotative marker ti but its position after the clause as opposed to the source item in !Ui (see section 4.10). Sometimes it appears that a linguistic structure or principle established in Khoekhoe by contact-induced change is ‘optimized’ and becomes more productive than in the source language – a phenomenon also discussed in other studies on language contact (see Thomason and Kaufman, 1988: 129– 46). This means that the subsequent language-internal dynamics are in principle independent of the source structure, and the original linguistic interference need not give an exhaustive account for the modern properties of an innovation. Several examples can be given for this. The grammaticalization of the intensifier //’ãi to a marker of plain anaphoric reference is a purely Khoekhoe-internal development, which in fact obscures the historical relation to the !Ui source (see section 4.1). Compared to Tuu, the formation of complex predicates by compounding has been diversified in Khoekhoe, involving the possible incorporation of elements other than verbs (see section 4.7). The prefield before the clause-second position is exploited in Khoekhoe for all kinds of constituents while it seems to be restricted in !Ui to nominal participants (see section 4.8). Finally, both Tuu and North Khoekhoe possess a default participant marker; however, these have a different scope, applying in Tuu to every valence-external (mostly all but the first) nominal after the verb, but in North Khoekhoe to every nominal after the subject PGN (including even subjects) (see section 4.9). The interrelatedness of different contact-induced phenomena had a cumulative effect for Khoekhoe-internal processes. For example, a subject PGN in the Wackernagel slot combines with a co-referential pronoun base in the prefield as a subject topic to create a central context for complex independent pronouns (see sections 4.1 and 4.8). Also, the subject PGN and the sentence type markers simultaneously enhance the Wackernagel position, while the two marker types in ⏐Xam are largely complementary (see section 4.8).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
The overall result of these mechanisms can be summarized as follows: Khoekhoe displays a marked distinction from its sister languages of the Kalahari branch of Khoe as well as from its contact languages of the Tuu family. That is, the considerable restructuring of Khoekhoe has not led to a notable alignment/convergence with the linguistic type of the substrate.16 Instead, Khoekhoe developed a characteristic structural profile that is different from both input languages before the contact. In some domains, this is accompanied by a considerable increase in complexity – an observation made also in other cases of language contact (see Thomason and Kaufman, 1988: 22–34). It was mentioned in sections 4.6 and 4.8 that the sparse linguistic data point to a significant linguistic deviation of older Damara varieties from the Khoekhoe spoken by this population today and that this suggests its linguistic ‘Khoekhoeization’. This leads to another point worth mentioning, namely that the substrate in Khoekhoe has been transferred to a geographic area where the original contact explanation is not feasible – a phenomenon that tends to obscure the actual historical processes. I mentioned briefly in section 3 that previously linguistic similarities between Khoekhoe and !Ui languages often received a historical explanation different from that entertained here, namely that they are the result of the borrowing of Khoekhoe features into San languages. I also indicated in section 1.2 that this transfer direction is indeed relevant; however, rather for more recent than for early history. As a generalized interpretation, this scenario is unfounded and is clearly influenced by the prejudice that foragers are unlikely to have a major impact on a ‘culturally superior’ pastoral society, linguistic or otherwise. This chapter tries to give evidence that the opposite is the case. Hunter-gatherers can have a profound influence on the language of newcomers, in this case even to the extent that the resulting language group Khoekhoe is classified as a primary unit within its higher order unit Khoe. Since the Khoekhoe expansion into the extreme south of Africa was from all we know the first introduction of food production into this part of the continent, this case is significant from a more general perspective; it is a potential test case for the possible linguistic results of this type of population shift on a global scale. This study indicates that one must not underestimate for the early linguistic history the impact of languages spoken by huntinggathering populations (today often obliterated) on the languages of the first food-producing immigrants of a certain area. The extent to which such historical processes might have shaped the modern distribution of linguistic features in the world can be discerned in yet another way. In the Cape region and areas further east, some substrate features from Tuu seem to have diffused into still later linguistic arrivals – that is, several Bantu languages and Afrikaans. Thus, Güldemann (1999) identifies in Nguni languages the typologically marked semantic association of feminine gender with large size (see section 4.2)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
128
129
and the use of a diminutive suffix on less common hosts like adjectives and predicates (see section 4.3). In the historical study of Afrikaans, several of its features are explained in part by substrate influence from other languages, whereby Khoekhoe has been playing an increasingly important role in the discussion. With the growing knowledge on San languages of the area, the exclusive focus on Khoekhoe turns out to be inappropriate. Just to mention two examples, the existence in Afrikaans of an associative plural form and, in view of its status as a ‘creoloid’, its surprising retention from Dutch of a complex clausesecond syntax has been ascribed partially to substrate interference from Khoekhoe (see Besten, 2001, 2002, respectively). However, both features also have parallels in !Ui languages, which should also be evaluated as possible contributors.17 Due to the enormous gaps of linguistic documentation in the area regarding Khoisan languages as well as their contact with Bantu and DutchAfrikaans, it will never be possible to determine conclusively the exact historical scenario for an innovation in these later linguistic arrivals in southern Africa. Clearly, direct contact with San languages has played a part and must not be underestimated; at the same time, it is possible that a !Ui substrate has, so to speak, ‘seeped up’ into these languages via Khoekhoe, which acquired a number of linguistic features from Tuu in earlier periods.
Notes 1 This chapter was presented as a paper on the following occasions: 14. Deutscher Afrikanistentag in Hamburg (11 October 2000); 23. Jahrestagung der Deutschen Gesellschaft für Sprachwissenschaft in Leipzig (1 March 2001); Substrate Workshop at the Max-Planck-Institut für Evolutionäre Anthropologie Leipzig (20 October 2001); Annual Conference of the North West Centre for Linguistics (NWCL): Linguistic Areas, Convergence and Language Change in Manchester (2 November 2002); and International Symposium on Khoisan Languages and Linguistics in Memory of Jan W. Snyman in Riezlern (6 January 2003). Thanks to W. Haacke for his comments on a draft version. Examples are cited in the original orthography, except for !Ui languages, for which I use a modified transliteration. The abbreviations in glosses are: AD addressee, APPL applicative, C common, COMP complementizer, CONN clause connector, COP copula, D dual, DECL declarative, DEI deictic, DIM diminutive, DSBJ deposed subject, E exclusive, F feminine, FUT future, IPFV imperfective, M masculine, MPO multipurpose oblique, NEG negative, OBJ object, OBL obligation, P plural, PASS passive, PQ polar question, PRO pronoun, PROP proper name, PST past, Q question, QUOT quotative, REL relative, RELV relevance, S singular, SBJ subject, SP speaker, SUBJ subjunctive, STAT stative, TR transitive. 2 Accordingly, the term ‘Khoisan’ is used here in a neutral, non-genealogical sense for click languages that do not belong to such securely established African language groups as Bantu or Cushitic. 3 Today, central Nama and Damara varieties are virtually indistinguishable linguistically. There is evidence, however, that early Damara differed from Nama (see Vedder, 1923). While the affiliation of early Damara remains unclear, on account
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
4 5
6
7
8
9
10
11 12
13
14
15
The Cape as a Linguistic Area of lexical data (Haacke, Eiseb and Namaseb 1997), it is possible that it was a Khoe language; but in my view outside the Khoekhoe branch. For example, the source of Krönlein’s material on a !Ui variety known in the literature as N⏐usa was a Khoekhoe-speaking Orlam. The sound structure of ⏐Xam is not completely recoverable from the available transcriptions. However, this concerns, first of all, vowel features; the consonant inventory is more secure. It is in fact possible that Khoekhoe, too, initially had this accompaniment and, thus, three instead of two click clusters. The aspirated velar plosive /kh/ is phonetically affricated [kx] (see Table 5.1). This segment as a cluster offset might then be hard to distinguish from the fricative /x/, which could have led to the merger of two originally distinct phoneme series, click + /kh/~[kx] and click + /x/. For the record, the orthography of the series click + /x/ in Standard Khoekhoe is in fact //kh, /kh, !kh, c kh. || Xegwi is exceptional within Tuu in that it was subject to extensive click loss, apparently associated with the intimate contact with Bantu languages (see Traill and Vossen, 1997). The diminutive suffix in !Ora is -da. Since Khoekhoe has a noun daro ‘child’, it is worth investigating in the future whether both -da and -ro are grammaticalized forms of this lexeme whereby the stem’s truncation during grammaticalization targeted different segments in South and North Khoekhoe. In any case, the extended use of North Khoekhoe -ro seems to have emerged later than the primary development in the domain of noun phrases from a compound [noundaro] – the probable source structure – to a word form [noun-ro]. The form in North Khoekhoe is ra, changing to re and ro after the past grams ke and ko, respectively. The !Ora form na (with an allomorph ra, see Maingard, 1962: 5, 25) is not, as sometimes stated, a present marker, but is functionally similar to North Khoekhoe ra (in spite of certain grammaticalized uses). The probable reconstruction of the Proto-Khoekhoe imperfective is thus *da. Note that some Kalahari Khoe languages do have this feature, for example Naro (Visser, 2003). This may also be a result of contact with Non-Khoe languages (see section 1.3). This type is also found in other language groups of Non-Khoe, such as the Ju family. This was, in fact, the principal reason that Schultze (1928) subsumed both population groups under one entity and coined the term ‘Khoisan’ (see inter alia Jenkins et al. (1971) for modern genetic data). Only later did ‘Khoisan’ also gain the connotation of being a linguistic unit. The historically more recent case of Nama–Damara relations in Namibia has yielded very similar results. According to Nurse et al. (1976) and Soodyall and Jenkins (1992), there is little genetic influence of the pastoral Nama on their former Damara clients, but a clear genetic contribution from the Damara in the Nama. Note that the historically much later emergence of the Baster groups from the social interaction between indigenous Khoekhoe and colonizing Boer farmers, especially in the frontier zone of the Cape colony, is surprisingly parallel to the present case in its overall scenario and result. Small and capital letter ‘s’ mean pronominal PGN subject and nominal ‘deposed’ subject, respectively. Recall, however, that a nominal subject can also surface when the prefield is filled by a nominal stem coreferential with the subject PGN.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
130
131
16 It is possible that the distinct culture and identity of Khoekhoe pastoralists and San, and the fact that their social relations became over time increasingly asymmetric were favourable circumstances for the maintenance of the linguistic distinctness of the languages, which before the contact were in any case distant (as mentioned in section 1, the languages are not related genealogically and also represent the major typological split in Khoisan Southern Africa of Khoe-Kwadi versus Non-Khoe). 17 See section 4.8 for the clause-second phenomena in !Ui. An associative plural in !Ui is attested with ⏐Xam -gu (Bleek, 1928–30: 92). This is reminiscent of the third-person masculine plural PGN -gu of Khoekhoe, which is also prominent in the associative plural of Namibian Khoekhoe (see Hagman, 1977: 29), so that one might be tempted to view ⏐Xam -gu as a borrowing. However, the history of associative plurals in Khoisan languages of the Cape is far from clear. There ae also !Ui-internal explanations for the ⏐Xam suffix in view of potential grammaticalization sources in related languages such as c Ungkue gu (plural: gu-kn) ‘creature’ and N || ng gau (plural: gu-ke or gon) ‘thing’.
References Beach, Douglas M. (1938) The Phonetics of the Hottentot Language (Cambridge: Heffer). Besten, Hans den (2001) ‘The Complex Ancestry of the Afrikaans Associative Constructions’, in Carstens, Adelia and Heinrich Grebe (eds), Taallandskap: Huldigingsbundel vir Christo van Rensburg (Pretoria: J. L. Van Schaik), pp. 49–58. Besten, Hans den (2002) ‘Khoekhoe Syntax and its Implications for L2 Acquisition of Dutch and Afrikaans’, Journal of Germanic Linguistics, vol. 14, no. 1, pp. 3–56. Bleek, Dorothea F. (1928–30) ‘Bushman Grammar: A Grammatical Sketch of the Language of the ⏐xam-ka-!k’e’, Zeitschrift für Eingeborenen-Sprachen, vol. 19, pp. 81–98; vol. 20, pp. 161–74. Bleek, Dorothea F. (1936) ‘Special Speech of Animals and Moon Used by the ⏐Xam Bushmen from Material Collected by Dr. W. H. I. Bleek and Miss L. C. Lloyd Between 1870 and 1880’, Bantu Studies, vol. 10, pp. 163–99. Bleek, Dorothea F. (1956) A Bushman Dictionary, American Oriental Series, 41 (New Haven, Conn.: American Oriental Society). Bleek, Wilhelm H. I. and Lucy C. Lloyd (1911) Specimens of Bushmen Folklore (London: George Allen). Boretzky, Norbert (1983) Kreolsprachen, Substrate, und Sprachwandel (Wiesbaden: Otto Harrassowitz). Brenzinger, Matthias and Christa König (eds.) (forthcoming) Khoisan Languages and Linguistics: The Riezlern Symposium 2003, Quellen zur Khoisan-Forschung, 17 (Cologne: Rüdiger Köppe). Budack, K. F. R. (1986) ‘Die Klassifikation der Khwe-khwen (Naman) in Südwestafrika’, in Voßen, Rainer and Klaus Keuthmann (eds), Contemporary Studies on Khoisan, vol 1, Quellen zur Khoisan-Forschung, 5 (Hamburg: Helmut Buske), pp. 107–43. Dempwolff, Otto (1927) ‘Die Hervorhebung von Satzteilen als Anlass zur Verwendung besonderer Wortformen’, In Franz Boas, Otto Dempwolff, G. Panconcelli-Calzia, A. Werner and D. Westrmann (eds), Festschrift Meinhof (Glückstadt/Hamburg: J. J. Augustin), pp. 73–9. Dempwolff, Otto (1934/5) ‘Einführung in die Sprache der Nama-Hottentotten’, Zeitschrift für Eingeborenen-Sprachen vol. 25, pp. 30–66, 89–134, 188–229.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
Dickens, Patrick J. (n.d.) Ju/’hoan Grammar. Unpublished ms. (Windhoek: Nyae Nyae Development Foundation). Foley, William A. and Robert D. Van Valin (1984) Functional Syntax and Universal Grammar (Cambridge: Cambridge University Press). Greenberg, Joseph H. (1963) The Languages of Africa (Bloomington, Ind.: Indiana University). Güldemann, Tom (1998) ‘The Kalahari Basin as an Object of Areal Typology: A First Approach’, in Schladt (ed.), 137–69. Güldemann, Tom (1999) ‘Head-initial Meets Head-final: Nominal Suffixes in Eastern and Southern Bantu from a Historical Perspective’, Studies in African Linguistics, vol. 28, no. 1, pp. 49–91. Güldemann, Tom (2001) Quotative Constructions in African Languages: A Synchronic and Diachronic Survey. Habilitation thesis, Institut für Afrikanistik, University of Leipzig. Güldemann, Tom (2002) ‘Die Entlehnung pronominaler Elemente des Khoekhoe aus dem !Ui-Taa’, In Theda Schumann, Mechthild Reh, Roland Kiessling and Ludwig Gerhardt (eds), Aktuelle Forschungen zu afrikanischen Sprachen: Sprachwissenschaftliche Beiträge zum 14. Afrikanistentag, Hamburg, 11–14 October 2000 (Cologne: Rüdiger Köppe), pp. 43–61. Güldemann, Tom (2004a) ‘Reconstruction Through “de-construction”: The Marking of Person, Gender, and Number in the Khoe Family and Kwadi’, Diachronica, vol. 21, no. 2, pp. 251–306. Güldemann, Tom (2004b) ‘ “Tuu”: A New Name for the Southern Khoisan Family’, in Tom Güldemann (ed.), Studies in Tuu (Southern Khoisan). University of Leipzig Papers on Africa, Languages and Literatures, 23 (Leipzig: Institut für Afrikanistik, University of Leipzig), pp. 2–9. Güldemann, Tom (forthcoming a) ‘Complex Pronominals in Tuu and Ju With Special Reference to Their Historical Significance’, Afrika und Übersee. Güldemann, Tom (forthcoming b) ‘Syntax: ⏐Xam of Strandberg’, in Voßen (ed.), The Khoisan Languages. Güldemann, Tom (forthcoming c) ‘Phonology: Other Tuu Languages’, in Vossen (ed.), The Khoisan Languages. Güldemann, Tom (forthcoming d) ‘Greenberg’s “case” for Khoisan: The Morphological Evidence’, in Vossen, Rainer (ed.), Problems of linguistic-historical reconstruction in Africa, Sprache und Geschichte in Afrika, 19 (Cologne: Rüdiger Köppe). Güldemann, Tom (forthcoming e) ‘Person–gender–number Marking from ProtoKhoe-Kwadi to its Descendents: A Rejoinder With Particular Reference to Language Contact’, in König, Christa and Rainer Vossen (eds), Festschrift for Bernd Heine, Routledge African Linguistics Series (London: Routledge). Güldemann, Tom and Rainer Vossen (2000) ‘Khoisan’, in Bernd Heine and Derek Nurse (eds), African Languages: An Introduction (Cambridge: Cambridge University Press), pp. 99–122. Güldemann, Tom and Edward D. Elderkin (forthcoming) ‘On External Genealogical Relationships of the Khoe Family’, in Brenzinger and König (eds), Khoisan Languages. Haacke, Wilfrid H. G. (1977) ‘The So-called “Personal Pronoun” in Nama’, in Anthony Traill (ed.), Khoisan Linguistic Studies 3, Communications 6. (Johannesburg: African Studies Institute, University of the Witwatersrand), pp. 43–62. Haacke, Wilfrid H. G. (1992) ‘Compound Noun Phrases in Nama’, in Derek F. Gowlett (ed.), African Linguistic Contributions (Presented in Honour of Ernst Westphal). (Pretoria: Via Afrika), pp. 189–94.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
132
133
Haacke, Wilfrid H. G. (1995) ‘Instances of Incorporation and Compounding in Khoekhoegowab (Nama/Damara)’, in Anthony Traill, Rainer Vossen and Megan Biesele (eds), The Complete Linguist: Papers in Memory of Patrick J. Dickens (Cologne: Rüdiger Köppe), pp. 339–61. Haacke, W., E. Eiseb, and L. Namaseb (1997) ‘Internal and External Relations of Khoekhoe Dialects: A Preliminary Survey’, in W. Haacke and E. D. Elderkin (eds), Namibian Languages: Reports and Papers (Namibian African Studies, vol. 4) (Cologne: Rüdiger Köppe), pp. 125–209. Haacke, Wilfrid H. G. (1999) The Tonology of Khoekhoe (Nama/Damara). Quellen zur Khoisan-Forschung, 16 (Cologne: Rüdiger Köppe). Hagman, Roy S. (1977) Nama Hottentot Grammar, Language Science Monographs, 15 (Bloomington, Ind.: Indiana University). Hahn, Theophilus (1870) ‘Beiträge zur Kunde der Hottentotten’, Jahresberichte des Vereins für Erdkunde zu Dresden 6/7, Anhang 1: 1–73. Heine, Bernd (1999) The //Ani: Grammatical Notes and Texts. Khoisan Forum, Working Papers 11 (Cologne: Institut für Afrikanistik, Universität zu Köln). Heine, Bernd (2000) ‘Grammaticalization Chains Across Languages: An Example from Khoisan’, In Spike Gildea (ed.), Reconstructing Grammar: Comparative Linguistics and Grammaticalization, Typological Studies in Language, 43 (Amsterdam: John Benjamins), pp. 177–99. Jenkins T., H. C. Harpendi, H. Gordon, M. M. Keraan and S. Johnston (1971) ‘Redcell-enzyme Polymorphisms in the Khoisan Peoples of Southern Africa’, American Journal of Human Genetics, vol. 23, pp. 513–32. Kilian-Hatz, Christa and Bernd Heine (1998) ‘On nominal gender marking in Kxoe’, in Schladt (ed.), Language Identity, 65–93. Köhler, Oswin (1973/4) ‘Neuere Ergebnisse und Hypothesen der Sprachforschung in ihrer Bedeutung für die Geschichte Afrikas’, Paideuma, vol. 19/20, pp. 162–99. Lichtenberk, Frantisek (2000) ‘Inclusory pronominals’, Oceanic Linguistics, vol. 39, no. 1, pp. 1–32. Maingard, Louis F. (1962) Korana Folktales (grammar and Texts) (Johannesburg: Witwatersrand University Press). Meinhof, Carl (1929) ‘Versuch einer grammatischen Skizze einer Buschmannsprache’, Zeitschrift für Eingeborenen-Sprachen, vol. 19, pp. 161–88. Meinhof, Carl (1930) ‘Der Koranadialekt des Hottentottischen’, Zeitschrift für Eingeborenen-Sprachen, vol 12 (Hamburg: Eckardt and Messtorff). Meriggi, Piero (1928/9) ‘Versuch einer Grammatik des ⏐Xam-Buschmännischen’, Zeitschrift für Eingeborenen-Sprachen vol. 19, pp. 117–53, 188–205. Nurse, George T., A. B. Lane and Trefor Jenkins (1976) ‘Sero-genetic Studies on the Dama of South West Africa’, Annals of Human Biology, vol. 3, no. 1, pp. 33–50. Schapera, Isaac (1926) ‘A Preliminary Consideration of the Relationship Between the Hottentots and the Bushmen’, South African Journal of Science, vol. 23, pp. 833–66. Schladt, Mathias (ed.) (1998) ‘Language, Identity, and Conceptualization Among the Khoisan’, Quellen zur Khoisan-Forschung, 15 (Cologne: Rüdiger Köppe). Schultze, Leonhard (1907) Aus Namaland und Kalahari (Jena: Gustav Fischer). Schultze, Leonhard (1928) Zur Kenntnis des Körpers der Hottentotten und Buschmänner (Jena: Gustav Fischer). Soodyall, Himla and Trefor Jenkins (1992) ‘Mitochondrial DNA polymorphisms in Khoisan Populations from Southern Africa’, Annals of Human Genetics, vol. 56:315–24. Thomason, Sarah G. and Terrence Kaufman (1988) Language Contact, Creolization, and Genetic Linguistics (Berkeley, Calif.: University of California Press).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tom Güldemann
The Cape as a Linguistic Area
Traill, Anthony (1980) ‘Phonetic diversity in the Khoisan Languages’, in Jan W. Snyman (ed.), Bushman and Hottentot Linguistic Studies (Papers of Seminar Held on 27 July 1979), Miscellanea Congregalia 16 (Pretoria: University of South Africa), pp. 167–89. Traill, Anthony (1994) A !Xóõ dictionary, Quellen zur Khoisan-Forschung, 9 (Cologne: Rüdiger Köppe). Traill, Anthony (1997) Extinct: South African Khoisan languages (CD and booklet). (Johannesburg: Department of Linguistics, University of the Witwatersrand). Traill, Anthony and Rainer Vossen (1997) ‘Sound Change in the Khoisan Languages: New Data on Click Loss and Click Replacement’, Journal of African Languages and Linguistics, vol. 18, no. 1, pp. 21–56. Traill, Anthony and Hirosi Nakagawa (2000) ‘A Historical !Xóõ-⏐Gui Contact Zone: Linguistic and Other Relations’, in Herman M. Batibo and Joe Tsonope (eds), The State of Khoesan Languages in Botswana (Gaborone: Basarwa Languages Project), pp. 1–17. Vedder, Heinrich (1923) Die Bergdama, 2 vols. Hamburgische Universität, Abhandlungen aus dem Gebiet der Auslandskunde 11/14 (Hamburg: L. Friederichsen). Vedder, Heinrich (1934) Das alte Südwestafrika: Südwestafrikas Geschichte bis zum Tode Mahereros 1890 (Berlin: Martin Warneck). Visser, Hessel (2003) ‘Naro Compounds’, in Brenzinger and König (eds), Khoisan Languages and Linguistics. Vossen, Rainer (1986) ‘Some Observations on Nominal Gender in Naro’, in Franz Rottland (ed.), Festschrift zum 60. Geburtstag von Carl F. Hoffmann, Bayreuther Beiträge zur Sprachwissenschaft, 7 (Hamburg: Helmut Buske), pp. 373–90. Vossen, Rainer (1997) Die Khoe-Sprachen: Ein Beitrag zur Erforschung der Sprachgeschichte Afrikas, Quellen zur Khoisan-Forschung, 12 (Cologne: Rüdiger Köppe). Vossen, Rainer (ed.) (forthcoming) The Khoisan languages (London: Routledge). Wallmann, J. C. (1857) Die Formenlehre der Namaquasprache: Ein Beitrag zur südafrikanischen Linguistik (Berlin: Wilhelm Hertz). Webley, Lita (1998) ‘Tracing the emergence of a pastoralist identity in Namaqualand’, in Andrew Bank (ed.), The proceedings of The Khoisan Identities and Cultural Heritage Conference; Held At the South African Museum, Cape Town, 12–16 July 1997 (Cape Town: The Institute for Historical Research, University of the Western Cape and Infosource), pp. 80–6. Yates, Royden, Anthony Manhire and John Parkington (1994) ‘Rock Painting and History in the South-western Cape’, in Thomas A. Dowson and David LewisWilliams (eds), Contested Images: Diversity in Southern African Rock Art Research (Johannesburg: Witwatersrand University Press), pp. 29–60.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
134
The Sri Lanka Sprachbund: The Newcomers Portuguese and Malay Peter Bakker
1
Introduction
The Sprachbund Sri Lanka has not received a lot of attention in the literature. It is connected to the South Asian Sprachbund, covering mainly Indo-Aryan, Dravidian, Austroasiatic and Tibeto-Burman languages, all of them present in the area for millennia. Many of the traits identified for the South Asian Sprachbund are also found in the languages of Sri Lanka. The two main languages of Sri Lanka are Sinhala (Indo-Aryan) and Tamil (Dravidian). In this chapter I shall focus on the ‘Indianization’ of two immigrant languages that do not belong to any of these families: Malay (Austronesian) and Portuguese (Romance). Their case is rather spectacular, as the sources suggest that the local varieties of Malay and Portuguese changed their typological character radically in a very short time. The creolized and hence relatively isolating languages became agglutinative in probably just a few generations, and both created new forms for almost all the semantic categories expressed in the previously creolized local Indic and Dravidian verbs, nouns and discourse structure. Such processes, called convergence (an ambiguous term) or metatypy, are often assumed to take millennia, but that assumption has had to be revised. Some of the data and argumentation have been discussed elsewhere (Bakker, 1995, 2000a, 2000b, 2003a). I shall first describe (in section 2) the demographic and linguistic situation in Sri Lanka, then focus on local Portuguese (section 3) and Malay (section 4), and compare them in section 5. In section 6 I shall put the languages in a wider context; in section 7 I discuss various implications and end with my conclusions in section 8.
2
Sri Lanka
Sri Lanka is an island of about 65,000 km2, some 50 kilometres from the south-east coast of India. In 2001, almost 19 million people were living there (Nanayakkara, 2001). The numerically and politically most important 135 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
6
The Sri Lanka Sprachbund
languages are Sinhala (Indic, Indo-European; 72 per cent of the population) and Tamil (Dravidian; 27 per cent of the population). These two languages have converged remarkably, sharing most of their basic grammatical traits, with Sinhala converging in the direction of Tamil (De Silva, 1974; Gair, 1976; Robuchon, 2003), and I shall not discuss this further. There are at least three minority languages on this island that are influenced by Sinhala and Tamil and have become part of the same Sprachbund. We have insufficient data on Kuravar or so-called Gypsy Telugu, a Dravidian language spoken by nomadic snake-charmers (Thananjayarajasingham, 1973; Karunatillake, 1982). But as this is a Dravidian language like Tamil, its structural similarities with Tamil would be inherited rather than taken over from another language, and therefore it will not be dealt with here. The other languages of Sri Lanka are Vedda, Malay and (Creole) Portuguese, all spoken by small minorities on the island: Malay by 0.2 per cent of the population, and Portuguese by even fewer. Vedda is the language of the original inhabitants of the island, sometimes assumed to be extinct, but apparently still spoken (Philip Baker, personal communication, 2003). Original Vedda may be unrelated to Indo-Aryan or Dravidian, or related distantly. However, in the twentieth century, the language shared so many forms and traits with Sinhala that it was sometimes classified as a dialect of Sinhala. Later research established that it is a separate language, although it is heavily ‘Singalesized’ (De Silva, 1972; Dharmadasa, 1974). In this chapter I shall focus on the other two languages of Sri Lanka – namely Malay and Portuguese, since we do not know what kind of language Vedda was before contact with Sinhala. I shall discuss Vedda briefly in section 6. In contrast, we have sufficient material on Malay and Portuguese.
3
Sri Lanka Portuguese
The Portuguese came to Sri Lanka in 1517 and left the island in 1658. Portuguese is, of course, a well-described language. The Portuguese language creolized in Sri Lanka. This creole is fairly well documented, and it shows many traits common with other creoles: loss of (most of the) inflection, formation of a particular tense–mood–aspect system, and so on. Examples and analyses of older Sri Lanka Creole Portuguese texts can be found in Jackson (1990) and Jayasuriya (1995/1996, 1996, 1997/1999). In his studies on modern Sri Lanka (SL) Portuguese, Ian Smith (1979a, 1979b: 186) mentions fifteen characteristics shared by SL Portuguese, Tamil and Sinhala and on Sri Lanka. These are, briefly: nominal case marking; similar semantic categories in verb inflection; SOV basic word order; postpositions; pre-nominal adjectives; auxiliaries following the verb; left branching relative clauses without relative pronouns; the use of past participles in connected sentences with the same subject; the use of sentence-final quotative particles; constituent-final markers of conditionals; temporal conjunctions
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
136
at the end of relative clauses; use of the order standard–marker–adjective in comparative constructions; use of nominal and verbal reduplication with the same semantics; the use of certain particles; and the use of datives for certain (English) subjects. The details of this list are perhaps in some cases obscure because of their brevity, but I have mentioned them to show that the local Portuguese must have taken them over from Tamil and Sinhala. This is clear from Smith’s papers (1977, 1979a, 1979b, 1984, 2001) as well as Jayasuriya (1999) and Stolz (1989). Neither earlier Sri Lanka Creole Portuguese nor European Portuguese share these features with the Sri Lankan languages. Smith (1979a, 1979b, 1984) also compared modern Sri Lanka Portuguese with creoles AS A CLASS, by taking typical creole language features shared by many creoles (including older Sri Lanka Creole Portuguese – SLP), which he compared with modern SLP and with standard Portuguese. None of these creole traits are shared with Portuguese, and only two of the fifteen or sixteen supposedly ‘universal’ creole features are present in modern SLP. SLP is no longer creole-like, but very close to Tamil and Sinhala. SL Portuguese has become an agglutinative language under the influence of Tamil. It is clear that Portuguese in Sri Lanka first creolized and then Tamilized. The language makes use of Portuguese elements to express Tamil grammatical categories. Modern Sri Lanka Portuguese uses, for example, morphemes derived from Portuguese prepositions as case suffixes on the model of Tamil. It uses Portuguese verbs and pronouns as sources for the topicalizer and the quotative, reportative, emphatic and other particles, all of them prominent grammatical markers in Tamil. Not the morphemes themselves, but the categories were borrowed, and these are expressed by Portuguese and never by Tamil elements. It is a typical case of metatypy, a term coined by Ross (1996). Only an indefinite marker is borrowed directly from Tamil. The Sri Lanka Portuguese example shows clearly that the language changed from being an analytic, prepositional and SVO language to an agglutinative, postpositional and SOV language, undoubtedly under the pressure of Tamil, and apparently within a short period: early twentieth-century Sri Lanka Portuguese sources still show a creole-like language (but see section 7 below). SLP is not an intertwined language, however. Intertwined languages combine lexical stems from one language with inflectional elements from another. Examples are early Angloromani, Calo, Ma’a and Media Lengua. In this typical Media Lengua sentence, Quechua elements are in italics and Spanish elements are in bold: Mañana pweblo-mun i-na-mi ga-ni ‘Tomorrow I have to go to town’. In SLP, however, no grammatical morphemes (apart from one, apparently) are taken over directly from the dominant language. Instead, the language uses native material to create the same semantic and often the same morphological categories as in the dominant language. The result is a language that has the same semantic categories as Tamil, and is grammatically close to Tamil, but virtually all the roots and all the bound
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Peter Bakker 137
138
The Sri Lanka Sprachbund
(1)
e: w eli-pɘ diñe: ruja: -dá: (SL Portuguese) I him-DAT money PAST-give na: n avan-ukku calli-ya kúTu-tt-an (SL Tamil) I him-DAT money-ACC give-PAS-CNC ‘I gave him the money.’ (Brazilian Portuguese: ‘(Eu) dei o dinheiro para/ a ele)’, Portugal: ‘Dei-lhe o dinheiro’)
(2)
ɘkɘ-ntu fu: lɘ pɘ-bota:
(3)
jeenti-s juuntu naanda papiyaa (SL Portuguese) people-PL ASS VOL.NEG talk aa-kaL -ooTa kade-kka maaTT-aa (SL Tamil) people-PL ASS talk-INF VOL.NEG-3SG.F ‘She refuses to talk with people.’ (Smith, 2001: 402) (Portuguese: ‘Ela recusa falar com pessoas’; the example here uses other stems: gente ‘people’, junto ‘near’, não ‘no(t)’ and the contact Portuguese verb papia ‘speak’, ubiquitous in Portuguese creoles)
(4)
osi: r-su fi: yə-pəportuge: s nuku-səwə (SL Portuguese) avar’-r’a makaL-ukku portyugi: s ter’i-ya (SL Tamil) he-POSS daughter-DAT Portuguese NEG-know-NEG ‘His daughter doesn’t know Portuguese.’ (Portuguese ‘Sua filha não fala português’; the verb sabe ‘knows’, used in SLP, cannot be used here in Portuguese)
na: -poy na: (SL Portuguese) that-LOC flower INF-put NEGPOT-can TAG at-ila pu: po: T-a e: l-a: t e: (SL Tamil) that-LOC flower put-INF can-NEGPOT TAG ‘[You] can’t embroider [litt. put flowers] on that [sewing machine], can you?’ (Portuguese: ‘Naquilo não se pode bordar, não é’)
Even though I give Tamil equivalents, Sinhala has converged towards Tamil to such an extent that both Sri Lankan languages could be a source. In fact, Jayasuriya (1999) argues that not Tamil but Sinhala is the main source of SLP structure. Her arguments are twofold: all constructions attributed to Tamil by Smith (1979) are also normal Sinhala constructions, and in her article she provides Sinhala equivalents to many of Smith’s examples. Her other arguments are historical: the first Portuguese settlements were in areas where Sinhala was the local language, and many others also spoke Sinhala as well as Tamil because it was the main language of the island. The structural similarities with Sinhala, however, are not greater than the similarities
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
morphemes are Portuguese and not Tamil. Some examples (from Smith, 1979a; Smith and Paauw, 2004) are:
Peter Bakker 139
with Tamil. At least some of the communities have traditionally been Tamilspeaking rather than Sinhala-speaking.
Sri Lanka Malay
The Sri Lanka Malay language appears to have undergone the same trajectory as SLP. As with Portuguese, it is fairly well known what Malay looked like when it was imported to the island by the Dutch. The Dutch brought Malay speakers to Sri Lanka from around 1656, both as a work force and as deportees from what is now Indonesia. According to Robuchon, Malay speakers also visited Sri Lanka and settled there before the Dutch era, but these do not seem to have left linguistic traces, except for a few place names. The ancestors of the Sri Lanka Malays were taken to the island between 1656 and 1850, by both the Dutch and the British (the latter after 1795). According to Adelaar and Prentice (1996), there were some 47,000 speakers of Malay scattered over Sri Lanka in the 1990s, most of them in Colombo, Gampaha, Hambantota and Kandy – to which we can add the community of Kirinda. Sri Lanka Malay is a language with postpositions, pre-nominal determiners, pre-nominal adjectives and SOV word order (Hussainmiya, 1987; Bichsler-Stettler, 1989; Adelaar, 1991; Slomanson, 2003), whereas standard Malay has prepositions, post-nominal demonstratives and adjectives, and SVO order. Most of these traits are not shared with the source dialect(s) of Malay. The source dialect of Sri Lanka Malay (SLM) must have been an early lingua franca Malay (see Adelaar and Prentice, 1996; Paauw, 2003), as it shows the pidgin-creole-like properties of this dialect, as well as some of the Hokkien Chinese influences (for example, the pronouns go ‘I’ and lu ‘you’). Here again, we have a development from a rather isolating language to an agglutinative language, constituting in many ways a radical typological change, enough for Adelaar (1991: 33) to consider SLM a ‘language in its own right’, and not merely a dialect of Malay. Hussainmiya (1987: 168) considers it a language ‘of Malay words with a syntactic surface structure of [Sri Lanka Tamil]’. Therefore, it is no longer an Austronesian language, but a language ‘which realigned itself with the Dravidian grammatical forms’. The following examples illustrate this: (5)
dey ruma-na pi-kalu avan uttu-ku pon-al he house-DAT go-if ‘if he goes home’
(6)
se lambat-pa 1SG late-ABS
(SL Malay) (SL Tamil)
rataη-le
(SL Malay)
come-PST
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
4
The Sri Lanka Sprachbund
nan sunaηg ivittu vandjn 1SG late being came ‘I came [being] late.’ (Robuchon, 2003: 9) (7)
(8)
(SL Tamil)
se sigit dalaη-na pi-pa se sbayaη 1SG mosque come-ACC go-ABS 1SG pray paLLikkuLLj poyviTTu, toLugugirjn ‘Once I enter the mosque, I pray.’ (Robuchon, 2003: 9)
(SL Malay)
loram-pe baapa aapa pdkajang kdja 2PL-GEN father what work do Onga-da uappaa enna ueele se 2PL-GEN father what work do ‘What work did your father do?’
(SL Malay)
aaDa
(SL Tamil)
ASP
nj-u-iru-nd-aar PAST-PPL-PERF-PAT-AGR
(SL Tamil)
It is clear that SLM, like SLP, underwent metatypy under Tamil influence. In the next section I shall discuss the developments in the two languages together.
5
Sri Lanka Malay and Sri Lanka Portuguese: a comparison
As Malay and (creole) Portuguese have both converged towards Tamil or Sinhala, the two languages should also show similar structures. We have already shown that they share a number of typological traits with Tamil. In this section I shall discuss first some aspects of morphosyntax, and then phonology. 5.1 Morphology and syntax If we compare the semantic distinctions expressed morphologically in the Tamil noun and verb, we get almost identical features in both SL Portuguese and SL Malay. There is an important distinction between the noun phrase and the verb phrase, however. In the noun phrase, all three languages make use of suffixes for number and case, whereas the semantic distinctions in the verb are expressed by preverbal elements in Malay and Portuguese, and suffixes in Tamil. Sentential structure seems also close to Tamil, with a few possible exceptions. A number of discourse and evidential markers also appear to be calqued from Tamil. This will be shown in Tables 6.1, 6.2 and 6.3 for nominal marking, particles and verbal categories. The tables are based on all the sources together. There are some gaps caused by the succinctness of the sources, and there were some slight differences in the data on local Tamil, perhaps because of dialect variation. SLM data were collected from a variety of speakers by Bichsler-Stettler (1989), Adelaar (1991), Robuchon (2003), Slomanson (2003), Hussainmiya, Smith, and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
140
Peter Bakker 141 Nominal case marking in Tamil, Sri Lanka Malay and SL Portuguese SL Portuguese
SL Malay
SL Tamil
NOM
Ø
Ø
Ø
ACC
Ø/-pɘ
-na (H, A, R) <M. nya ‘the’,‘that’
Ø/-a(y)/-e
DAT
-pɘ
-pe(i) (H, A) <M. punya ‘of’ -na (inan.) (R) -daη (anim.) (R)
-(u)kku
GEN
-su(wɘ)
-pe (H, A, R) <M. punya ‘of’
-Ra -ooTe
LOC
-(u)ntu
-ka (H, A, R) <M. dekat’near’
-(i)la(y)/-ile -(i)TTa(y)
COLLOC
-ntaa
ASSOC
N ju: ntu
-samma (H, R) <M. sama ‘with’
-o: Ta(y)
INSTR
N wɔndɘ, vɔɔndɘ
-dari (H) <M. dari ‘from’
-a: la(y)
ABL
N impə
-ka dupa (R) <M dekat ‘near’ + ??
-ile iruntu, il-irunthu, iruntu
-kiTTe
In SL Malay column: H = Hussainmiya 1987; A = Adelaar 1991; R = Robuchon 2003.
Paauw (2004), and these sources display a number of differences only partly reflected in these tables. Tables 6.1–6.3 illustrate the close semantic (and often syntactic) parallels between SL Portuguese, SL Malay and SL Tamil. By giving etymological sources of the elements in SL Portuguese and SL Malay, it is made clear that practically all these grammatical markers have their source in Portuguese and Malay respectively, apart from a few of unidentified origin. None of them is borrowed from Tamil or Sinhala. The etymologies presented in the tables are almost always as given in the sources, or have been provided by David Gil (Malay) or myself. Whenever there is a gap in the table, I have found no examples in the available data. In the noun phrase (Table 6.1), the same categories are expressed in the same position with regard to the noun (suffixes or postpositions). In all the Sri Lankan languages, the markers are postnominal, in sharp contrast to
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 6.1
The Sri Lanka Sprachbund
both ‘normal’ Malay and Portuguese, including the creolized versions, which only have prepositions. There is a difference in degree of suffixation, however. Even though there is a sliding scale between suffixation, cliticization and (in this case) postpositions, the postnominal markers seem to be connected more loosely to the nouns in SLP and SLM than in Tamil, where they are all suffixes. In all three languages, the accusative, dative, genitive and locative are case markers, whereas the associative and instrumental are cases in Tamil and postpositions in SLP and SLM (there are no data about the collocative for SLM). Finally, Robuchon discusses the ablative case, which is a compound of the locative -ka with an additional element dupa, modelled on the Tamil and/or Sinhala -e ind la. In this respect, the fate of the Sourashtra language is also worth mentioning. The Sourashtra speakers immigrated into the Tamil area in India in the sixteenth century, after moving from Gujarat at some time in the eleventh century, with a period of sojourn in Marathi- and Teluguspeaking areas. Smith (2001: 395) shows a close parallel Dravidianization of both Indo-Aryan Sourashtra and SLP in the nominal realm. There are a few differences too, however. The postpositions do not always govern the same case markers in Tamil as in the other two languages. Whereas all markers are etymologically Portuguese in SLP, Sourashtra also inherited some markers from Gujarati, and borrowed some from Marathi, Telugu and Tamil. Perhaps a closer typological similarity between Sourashtra and the other Indic languages facilitated this direct borrowing. Tamil has a number of discourse particles, which have been calqued in SLP and SLM. The particles are often enclitics in all three languages, and they express the same semantic categories (see Table 6.2), in the same positions in the sentence (sentence-final, or attached to the verb). It is clear from this table that the three Sri Lanka languages under discussion here express the same semantic categories by such particles, and that both SLP and SLM use native material and not Tamil elements. Such discourseorganizing elements tend to be among the first to be borrowed (Matras 1996, 1998b; Stolz and Stolz, 1996), and from that perspective it may be something of a surprise that the forms themselves have not been borrowed. The verb phrase is illustrated in Table 6.3; again, the same categories are expressed in the three languages, but most often by different means: preverbal elements in SL Portuguese, preverbal elements and suffixes in SL Malay, and only suffixes in SL Tamil. The preverbal elements in SLP and SLM are undoubtedly an inheritance of the creole-like structure of the languages before Tamilization. The SLM suffixes probably moved to postverbal position under Tamil influence. It remains unclear, however, why only the NP underwent suffixation and not the VP. Standard Malay also has a number of preverbal TMA markers. The difference between Malay and SLM is that the SLM markers are all reduced in form compared to the Malay
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
142
Peter Bakker 143
Category
SL Portuguese
SL Malay
SL Tamil
Gloss
Reportative
-ski
-kangnag <M. kata nya ‘it is said’
-am
‘it is said’
Quotative
fala:
kata (R)
eNTu, endru
‘(s)he says’
Tag question
na:
-kah <M. kah ‘is it’
e:
‘isn’t it’
Conjunctive
ta: m
-le(y)
-um
‘and’
Emphatic
me:
-jo <juga
-ta: n
‘even, -self’
-a
question marker
(juo, jo) ‘also, even’ Interrogative
-si <siapa ‘who?’ ?
markers. All three languages show the same semantic range of their verbal categories, although the labels do not always seem to be appropriate. The semantic categories were ‘borrowed’ from Tamil, but not the form of the morphemes, nor the position relative to the verb. In this, the verb differs from the noun, where both the functions and the positions were taken over from Tamil. There is no room here to speculate about the possible reason for the asymmetry between the verb phrase and the noun phrase. Table 6.3 is not complete: Slomanson (2003) and Smith and Paauw (2004) report a past tense suffix -ay(ng) or -nya(ng). Smith and Paauw have a few additional forms as well. Therefore, the range of semantic categories is not exhaustive for any of the languages. More detailed data from these languages will probably show even greater parallels. One parallel that does not appear in Table 6.1 is the fact that the case markers and postpositions are added to the postnominal plural element, copying the agglutinative patterns of Tamil. In Tamil the plural element is -kaL, in Malay -pada and in Portuguese -s. The SL Malay plural suffix is derived from a Javanese prenominal plural marker according to Adelaar (1991). Robuchon also mentions the ‘absolutive’ construction in which only the second verb gets person inflection, and the first verb, co-referential with the second, gets an absolutive -pa suffix (see Examples 6 and 7 above). This construction is also present in Tamil and (adopted from Tamil) in Sinhala. SL Portuguese uses the suffix -sərə with the same function (see example in Smith, 1977: 354).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 6.2 Particles in SL Portuguese, SL Malay and SL Tamil
The Sri Lanka Sprachbund
Table 6.3
Verbal categories in SL Portuguese, SL Malay and SL Tamil
Category
SL Portuguese
SL Malay
Imperative
V
V
V
do!
Present
tɘ-V
arɘ-V
V-kk-r-an
I am doing
Past
ja: -V
so-V <M. sudah ‘already’ ay(ng)-, nya (ng)-
V-cc-an
I did
Potential
lo-V
a(n)ti-V < M. nanti ‘soon’
V-pp-an
I will V, I usually V
Perfective
ka: -V
V-ir(u)-
I have done
Descriptive
ki: -V ???
V-(kkr-)a-
I (used to) V
Conditional
V-kalu <M. kalu kan(da)-V
V-al
if I V
Permissive
jesɘ-V
???
SL Tamil
mes(tɘ) V, mesɘ V V-num ‘must’ <M. mesti
Obligational
Gloss
Should I do? May I do? I must do
Negative
nuku-V
ta-V <M. taq, t(a)ra ‘not’
V-(kka)-illa(y) I did no do, I don’t
Volitive-negative
na: -V, naanda
tuma-V <M. taqmau ‘don’t want’
V-(kka)-ma: TT
I won’t do, don’t do
V-a-illa V-(kk)-a: ta
don’t do don’t do
Negative descriptive nikara Negative-imperative numis(tɘ)-V
Sources:
ta-V-kalu taq ‘not’
Hussainmiya (1987), Bichsler-Stettler (1989), Adelaar (1991), Robuchon (2003).
5.2 Phonology Sri Lanka Malay phonology is thoroughly influenced by Tamil. Adelaar (1991) mentions the presence of retroflex consonants, gemination and contrastive vowel length. All three are absent in Malay but prominent in Sinhala and Tamil, which must be the source of these innovations.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
144
Robuchon (2003), however, points to the fact that the accent in SLM is systematically put on the first syllable (Bichsler-Stettler (1989) says the penultimate syllable of the root), which is very different from Sinhala and Tamil. Sri Lanka Portuguese phonology has been described in detail in Smith (1977), who also compares it to Tamil and sixteenth-century Portuguese. This discussion is based on his analysis. The SLP and Tamil phoneme inventories show some significant differences. Tamil has two distinctive /r/ sounds, SLP only one. Tamil has retroflex consonants, SLP has none. SLP has three fricatives /j f s/ that Tamil lacks, and SLP has /æ/ and /ɔ/ vowels that Tamil lacks. SLP has voiced and unvoiced stop series where Tamil only has an unvoiced series. On the phonetic level, however, the stop systems appear to be more similar. SLP displays clear influence from Tamil. Both Tamil and SLP have distinctive vowel length and predictable stress, in contrast to Portuguese. The nasal vowels of Portuguese have been reanalysed as sequences of vowels plus nasal consonants (Portuguese pão, SLP pá: m ‘bread’). Tamil has no nasal vowels. SLP unrounds the /u/ when unstressed, as in Tamil. A rule in SLP retracting the vowels /i: i e: e/ before (retroflex consonants and) /r/ is reminiscent of a rule in Tamil but not identical. In short, the SLP system is partly inherited from Portuguese, partly influenced by Tamil. SLM phonology seems more Tamilized than SLP phonology. SLP and SLM seem to have a phonological system in between the ones of their lexifiers and Tamil.
6
The wider context
In this section I shall discuss briefly some connections of SLP and SLM with mainland South Asia, with the situation of Vedda, and with regional contact varieties of Portuguese and Malay. 6.1 South Asia The common traits of the South Indian languages of Dravidian, Austroasiatic, Munda and Indo-Aryan stock were noticed long ago, and there is a considerable body of literature on this (for example, Masica, 1971; Bhaskararao and Subbarao, 2001). Masica (2001) has argued that the South Asian area is in fact connected to some wider Eurasian areas. For example, co-referential complements (‘I want to go’) show clear clines through Western Eurasia (Masica, 2001: 226–7); SOV word order is common in large parts of Eurasia (Masica, 2001: 242); definite objects are marked in many languages from Turkey to South Asia; and numerical classifiers, so common in South East Asia, are also found in the eastern part of South Asia.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Peter Bakker 145
The Sri Lanka Sprachbund
I have already alluded to the fact that the Sri Lanka Sprachbund is connected to the South Asian linguistic area. Some of the traits shared by the Sri Lanka languages with the languages of mainland South Asia are the so-called experiencer dative, word order, case inventories, agglutination and so on. The question of how Sinhala and Tamil came to share so many traits with South Asian languages, even though separated by a considerable body of water, is an important one, but beyond treatment in this chapter. 6.2 Malay contact varieties Sri Lanka Malay is unique among contact varieties of Malay in that it developed case marking, postnominal modifiers, verb-final sentential word order and related traits. On the other hand, it has a number of features in common with other contact varieties. Adelaar and Prentice (1996) discuss eight features that contact varieties of Malay have in common with each other, and in which they differ from standard Malay. I shall deal with those briefly here: (A) Whereas standard Malay expresses possession by means of juxtaposition possessed–possessor, SLM uses the opposite order with a linker -pe, which is related directly to the contact Malay construction ‘possessor + punya + possessed’. (B) Malay has distinct pronouns for six or seven persons, with a wide variety of forms. The contact varieties, according to Adelaar and Prentice (1996) tend to express plural pronouns by means of a postnominal element orang following the singular pronouns. In SLM we do not only find orang as a plural marker, but also a suffix -pada in the second and third person plural (see Table 6.4). According to Adelaar this suffix derives from a Jakartanese Malay predicative plural marker, into which it was borrowed from Javanese (1991: 32). (C) The number of productive affixes is reduced in the contact varieties. Often tɘr- and bɘr- are the only productive prefixes. Neither of these is reported for the SLM sources. (D) The verb ada ‘to be, to have’ indicates aspect. This is also the case in SLM (see Example 8 above). (E) The demonstratives itu and ini are postnominal in Standard Malay, but prenominal in the contact varieties. Furthermore, they function like determiners rather than demonstratives. In SLM they are indeed prenominal, but here they function like demonstratives, not determiners. (F) In contact varieties, the verb pɘrgi ‘to go’ is also used as a preposition ‘towards’. In Bazaar Malay, it is only used for ‘to go’, in the reduced form pi. It is not used as a preposition in Sri Lanka Malay.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
146
Peter Bakker 147 Pronouns in three Malay varieties
Indonesian Malay (selection)
WW II Bazaar Malay (Bakker, 2003b)
Sri Lanka Malay
1 SG
aku (polite saya)
sa(ha)ya
se <M. saya (R, A) (polite)go < Hokkien gua (A)
2 SG
engkau, kamu
lu, tuan, etc.
lu < Hokkien lu (R, A)
3 SG, dia, ia
dia, ia
de <M. dia
1 PL
kami (EXCL.) kita (INCL.)
kita
ikan <M kami (1PL.EXCL.)kitaη < kita + oraη (A)
2 PL
saudara engkau kamu, kalian
kamu
rompada < ?? + pada (R)loraη
3 PL
mereka, dia, orang itu
dia
dempada <de + pada (R)deraη
Note:
A = Adelaar 1991; R = Robuchon 2003.
(G) One of the verbs ‘to give’ (kasi/beri) or ‘to make’ (bikin/buat) is used as a causative (+V). I have not found examples in the data that shed light on this. According to Bichsler-Stettler (1989), SLM has morphological causatives: the causative suffix -kam/-kan/-kaη/-kiη, which may be derived from Malay bikin. (H) The preposition sama or another word is used as a multifunctional preposition. This word is used as a postposition in SLM, but it does not seem to be very frequent. The locative postposition -ka seems much more frequent. This comparison of SLM with these widespread features of contact varieties of Malay reveals some similarities (A, B, D, E). On the basis of these, both Adelaar and Prentice (1996) and Paauw (2003) assume that SLM derives from the same source as the other contact varieties of Malay. The data are indeed compatible with this assumption, but only if we assume the loss of some of the features in later stages. 6.3 Portuguese contact varieties There are a number of Portuguese creoles in India. SLP and the Indian Portuguese creoles show a number of striking resemblances. Some of these are as a result of the creolization process, whereas others are caused by South Asian influences, and some must result from a common source, perhaps an early version of the creole that spread to other areas of India.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 6.4
The Sri Lanka Sprachbund
Ferraz (1987) is a comparative study of Portuguese creoles of West Africa and Asia. It appears that the Portuguese creoles regionally have traits in common, which are not creole universals, and are not always attributable to substrate languages. Ferraz (1987) shows clear differences between the Portuguese creoles of the Gulf of Guinea and of Upper Guinea, and these differ in turn from the Asian creoles. What is interesting for us in this respect are the common traits of the Asian Portuguese creoles. Schuchardt’s statement (Schuchardt, 1889) that they fall into four groups has been corroborated by later studies (Ferraz, 1987: 348). These groupings are Malayo-Portuguese (Java, Malacca, Singapore); Sino-Portuguese (Hong Kong, Macau); Indo-Portuguese (Diu, Daman, Norteiro, Goa, Mangalore); and Dravido-Portuguese (Cannanore, Mahé, Cochin, Nagappattinam, Sri Lanka) (Schuchardt, 1889: 476). The creoles from these four groups also have traits in common, according to Dalgado, caused by ‘partial reciprocal transfusion’ because of contacts between speakers (cited in Ferraz, 1987: 337). Several authors have pointed to common traits. I shall discuss some of the traits discussed by Ferraz which many Asian creoles have in common, but not the African ones. For these Asian creoles, older, mainly nineteenthcentury sources have been used. The examples are from older, nineteenthcentury Sri Lanka Portuguese (OSLP) and from Portuguese (P). (1) The possessive is expressed as POSSESSOR linker POSSESSED (Portuguese POSSESSED linker POSSESSOR, or POSSESSOR POSSESSED): OSLP eu su vida, P a minha vida ‘my life’. (2) The modifier precedes the noun (Portuguese: noun-modifier): OSLP fɔme soldadu, P soldado com fome ‘hungry soldier’. (3) The suffix -mente is used for manner adverbs, also with some stems where Portuguese does not use it: OSLP tantumente, P tanto ‘so much’. (4) Preverbal lo is used for future, from Portuguese logo ‘later’: OSLP eu lɔ ama ‘I shall love’. (5) Portuguese já ‘already’ is used as a preverbal perfective marker (also used in the Gulf of Guinea). (6) Noun plurality (perhaps actually denoting plurals of diversity) is expressed with reduplicated nouns (see also Jayasuriya, 2003); OSLP fifi, P filhos ‘sons and daughters’. (7) Objects are marked with a preposition derived from Portuguese per ‘for’: OSLP eu lo dar per ti ‘I shall give you’. (8) The Portuguese verb ter ‘to have’ is used as a copula: OSLP alis nõ tiNa muytu bunitu ‘Alice was not very pretty’. (9) The Old Portuguese word laia ‘way, manner’ is used as a function word: este ləi ɔme ‘such a man’. (10) Ferraz also identifies fourteen words that are widespread in Asian creoles and unique to them, some from Portuguese, others from Asian
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
148
Peter Bakker 149
Apart from (7), all these traits are also found Sri Lanka (and the SLP case marker for dative and accusative -pɘ may be derived from this preposition). Even though some may be local and independent developments, the fact that some typologically uncommon traits are shared suggests that there was contact between the communities. Some traits suggest influence from lingua franca Malay, such as the reduplicated nouns and perhaps the possessive construction. In several respects, these creoles differed from the surrounding languages, for example, SVO word order (India SOV), reduplicated nouns for plurality (as in Malay, but not in India). In later stages, these creoles may have converged towards the South Asian type – for example, in developing SOV word order. It is striking that most of the creoles for several centuries did not participate in the convergent developments of the South Asian Sprachbund. On the other hand, these common traits suggest that the Indian Portuguese creoles, including Sri Lanka Portuguese, share part of their past, or have been part of a network. To what extent these shared traits go back to a common origin, perhaps in Goa or in a Malay-speaking area, remains to be established. 6.4 A note on Vedda Vedda is not a subject of this study, but it needs more than mere mention in passing. The close connection and shared traits between Sinhala and Vedda are visible in these sentences, from De Silva (1972: 42, 97, 105), in Vedda with their Sinhala equivalents. (9)
(10)
V: kataragama deyaläätto mangacc-alaa indiya-poorugampojj-en S: kataragama deyyo äv-illaa indiyaa-ven name-of-a-God deity come-PST.3 India-(CLASS)-ABL V: lankaa-pojja-Ta diyagan-äll-en S: lankaava-Ta muud-en Sri.Lanka-(CLASS)-DAT river-INSTR ‘The God Kataragama came to cross by the sea.’
mangacc-anna mangäcc-uvaa. e-nna ää-vaa. become-infin go-PST.3 over to Sri Lanka from India
Word order is almost the same in both languages, and most of the endings are the same, but many roots have different shapes. Vedda, however, has recurrent elements between the nominal roots and endings that show similarities with classifiers elsewhere. It shows some similarities with intertwined languages (Bakker, 1997, 2003a) that have roughly the same grammar (word
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
languages such as Konkani, Malayalam or Malay: OSLP atarde, P tarde ‘afternoon’; OSLP fruidade, P frio ‘coldness’, frialdade, OSLP ade P pato ‘duck’, OSLP sayan, Malay sayang ‘regret, pity’, etc.
The Sri Lanka Sprachbund Table 6.5 endings
NOM ACC DAT GEN ABL IND DEF PLU
Sinhala and Vedda nominal
Sinhala
Vedda
-ɘ? -wɘ -Tɘ -ge -gen -eku-a-in-
-aa /-i ? -Ø -Ta -ga -gen -eku-a- / -i-? -an-
order, grammatical endings) from one language and a lexicon from another. There are differences as well: there are also many Sinhala roots, and the shape of the endings of the two languages are not identical (see Table 6.5). The classifier system seems unique to Vedda. These are only preliminary remarks, awaiting more thorough studies on this language, either on the basis of published material (De Silva, 1972; Dharmadasa, 1974) or on the basis of fieldwork. 6.5 Summary It is clear that SLM and SLP are spoken in a convergent area that also covers Vedda. This area is connected typologically to South Asia, and historically connected to other contact varieties of Malay and Portuguese.
7
Discussion
This chapter has shown the significant changes in the grammatical systems of two Sri Lankan languages. In both cases this seems to have come about in a situation where the ancestral language continued to be spoken alongside dominant Tamil. It must be said that this situation is not unique. There are also examples of similar forms of convergence of isolating languages towards agglutinative languages in other parts of the world, such as Chinese dialects converging towards Mongolian and/or Tibetan. The Sri Lanka situation also sheds light on a number of issues in the study of convergence, sometimes providing counter-examples against some widespread assumptions about this process. I shall deal with these here. 7.1 Speed of development Seen synchronically, the case of the Sri Lanka Sprachbund may be in itself not so uncommon. For example, along the borders between Papuan and Austronesian languages many parallel cases have been reported of local convergence (Ross, 2001; Foley, 1986: 281–2; see also below). Cases in which
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
150
relatively isolating languages adopted some agglutinative structures using native material are, for example, Armenian and Asia Minor Greek under Turkish influence (Thomason, 2001) and, if Donegan and Stampe (2004) are right, Munda. The cases along the Chinese–Mongolian/Turkic/Tibetan borders are most significantly parallel – for example Hezhou and Hui, both agglutinative Mandarin under Uyghur Turkic influence; and Wutun – agglutinative Chinese under the influence of Mongolian and Tibetan (Li, 1983, 1984; Lee-Smith, 1996; Lee-Smith and Wurm, 1996). What makes the Sri Lanka case spectacular, however, is the documented speed with which the development took place. I have argued elsewhere that SLP and SL Malay must have changed their typological make-up rapidly (Bakker, 2000a, 2000b). This statement, however, is not uncontroversial (see, for example, Paauw et al., 2004). Smith (2001) for example, believes that the ‘Dravidianization’ of SLP is one or two centuries old: ‘Tamil influence is not due to language death – comparative evidence allows us to push most of its Dravidian features back 150 to 200 years, to when the Sri Lanka Portuguese was a lingua franca on the island’ (Smith, 2001: 392). The fact that it was a lingua franca was apparently a reason for this Dravidianization: ‘its role as a lingua franca meant Sri Lanka Portuguese continued to feel the influence of mother tongue Tamil and Sinhala speakers’ (p. 393). How fast did the radical changes from SVO and relatively isolational to SOV and agglutinative take place in the two Sri Lanka languages? It is worth repeating some of the arguments in favour and against. 7.1.1 Earlier documents We have rather extensive documentation of Portuguese creole from earlier periods. The British colonial officer, Hugh Nevill, who worked in Sri Lanka between 1869 and 1886, collected manuscripts of Creole Portuguese texts from several locations – among others, Batticaloa. Even though these song texts may already have been archaic at the time they were recorded, they show a language very different from what was recorded by Smith from people born in the first decades of the twentieth century. Therefore it looks as if it took no more than fifty years for the language to change radically. Songs in the archaic style were still used in the late1980s (Jayasuriya, 1996: 255). The fact that these older texts do not display the heavy Tamil influence would be an argument in favour of a recent development – say, between the 1880s and the 1920s. A counter-argument against this would be that there was a situation of diglossia, and these song texts would be in the high variety of Portuguese, whereas the colloquial variety was already spoken alongside the high variety. Tamil and Sinhala are also in a situation of diglossia. In creole studies it is also a well-known situation that early sources of creoles appear to be much closer to the lexifier and further away from
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Peter Bakker 151
The Sri Lanka Sprachbund
substrate or adstrate influence. Early sources of Jamaican, Saramaccan, Sranan, Negerhollands and other creoles are all less creole-like than the later sources. When we have independent contemporary early sources, such as with Negerhollands Creole Dutch, the sources match in details. There is little or no reason to assume that those sources do not display a language actually spoken. I assume it is the same with Sri Lanka Portuguese, and that the nineteenth-century song texts were only slightly archaic when they were recorded in the late nineteenth century. For Malay, a similar argumentation can be made: early sources (from 1806 or 1820 to the 1930s) show a form of Malay with little grammatical influence from Tamil. This can be related to a diglossic situation. However, a collection of pantuns (traditional improvised rhymes) published in 1906 show ‘heavy influence from colloquial SLM’ (Smith and Paauw, 2004), and that may indeed indicate that Tamil influence dates from before the twentieth century, and is the strongest argument for diglossia. In short, the evidence is indecisive here. However, the authors do seem to agree that the process took place rapidly. 7.1.2 Older speakers were/are familiar with the ‘conservative’ variety A more important argument may be that Smith’s SLP informants of the 1970s were still to some extent familiar with a more conservative variety. Smith (1977: 352) wrote: ‘The data of B[aticcaloa] P[ortuguese] is of two types: normal colloquial speech and formal, consciously archaizing speech which my older consultant could produce under stimulus.’ As Smith’s older informant was born in 1906, we can assume that this archaic form could still be heard in his youth, and not only in songs. In a number of cases, Smith was able to elicit a continuum of sentences (1977: 367), some with Portuguese prepositions, others with more creole-like prepositions, and the more colloquial form with a suffix. These elicited sentences had SVO word order, either based on the archaic style, or because English was used as a language of elicitation (as is well known, elicitation tends to trigger word order of the intermediary language). We have no comparable argumentation from Malay. For SLP, the above-mentioned facts suggest that the SLP diglossia situation existed in the early decades of the twentieth century. This, however, is compatible both with a twentieth-century genesis and an earlier genesis (eighteenth–nineteenth century). 7.1.3 In older sources one occasionally encounters suffixes and other ‘modern’ SLP or SLM features Authors on SLP (Smith) and SLM (Hussainmiya) report that the modern features of the two Sri Lanka languages are indeed found occasionally in early sources. One can use these as an argument that the modern languages were already spoken at that time, and claim that these suffixes reveal a
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
152
Peter Bakker 153
7.1.4 SLP was a lingua franca, and therefore there was early influence from Tamil and Sinhala speakers Even if it is true that SLP was a lingua franca, SL Malay was never one, so this argument is invalid for SLM. There have been many lingua francas, but to my knowledge none underwent contact-induced language change to the extent that SLP has. In fact, Smith compares the situation of SLP with that of Sourashtra, a language of India under strong Tamil influence. Sourashtra has never been a lingua franca either. On the contrary, Sourashtra is the language of the home, and Tamil the language of wider communication. It is clear that being a lingua franca does not lead to such convergent influences, and that languages can undergo Dravidianization without being a lingua franca. 7.1.5 Conclusion: when and how fast? After presenting these arguments and counter-arguments, it is time to draw conclusions on the speed of development. We have to split the problem in two: the speed of the process; and the time of the relevant events – first, did convergence take place rapidly or slowly, and, second, did it take place soon after contact, or in the twentieth century? Depending on one’s view of what a rapid or a slow process is, one can view the process as either. Compared, however, to what is generally assumed, even a few centuries would count as rapid. In that sense the Dravidianization of SLP and SLM were rapid processes. The question of when these developments took place, leads one to conclude that it was a matter of one or two generations. The two main opinions on the matter of when the convergence happened both have to accept that it is a very rapid process. If it happened soon after arrival, as Paauw (personal correspondence, 2004) suggests, leading to a situation of persistent diglossia (with the assumption that a more creole-like or more classical variety survived alongside the Tamilized version in a diglossic situation), the convergence or metatypy must have happened within one or a few generations. Alternatively, if it happened in the first half of the twentieth century (assuming that written sources of Malay and Portuguese in Sri Lanka reflected the only language spoken), it must also have happened within one or two generations. In short, the Dravidianization of SLM and SLP was a rapid (less than 200 years) or extremely rapid (say, one or two generations) process. It could have happened soon after arrival, or in the first half of the twentieth century, and in both scenarios extremely rapidly.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
failed inhibition of colloquial language features. On the other hand, one can also claim that these features reveal slips of the tongue or pen in the direction of Tamil, reflecting occasional tendencies in the spoken language of that time.
154
The Sri Lanka Sprachbund
Gumperz and Wilson (1971) have discussed the language situation in the village of Kupwar in India in some detail. Four different languages are spoken there – two Indic (Indo-European): Marathi and Urdu; and two Dravidian: Kannada and Telugu. All the languages have converged to such an extent that they share most of the grammatical system, and translation is largely a matter of replacing the morphemes of one language with others. In that sense Kupwar and Sri Lanka are very reminiscent of one another. In the famous and often cited Kupwar case, the languages in contact converged in the direction of a language with a less complex morphology. In Sri Lanka, two languages with little morphology acquired agglutinative morphology through language contact. The Sri Lanka case shows that a Sprachbund does not necessarily lead to simplification. 7.3 Multilingualism is not necessary in a Sprachbund Not all linguistic areas are characterized by the presence of many languages and widespread multilingualism, as is sometimes claimed. A linguistic area can have as few as two languages, and regional bilingualism rather than multilingualism. In the Sri Lanka situation, Convergence took place between individual pairs of languages (Malay–Tamil; Portuguese Creole– Tamil; elsewhere Portuguese Creole–Sinhala). In some Malay communities, notably in Kirinda, both Sinhala and Tamil are/were second languages. 7.4 Convergence can be unidirectional It has been argued repeatedly that convergent areas come into being in situations of widespread and mutual multilingualism (see, for example, Thomason, 2001), where languages influence each other in a variety of directions. In the case of Sri Lanka, however, it seems to be the case that very few local Tamil or Sinhala speakers learned Portuguese or Malay. The SLP speakers and the SLM speakers are at the time of writing dominant in Tamil, and this may have been the case for a long time. In other words, only the Malays and Portuguese speakers were speakers of other languages as well, and therefore mutual language knowledge is not a necessary condition for the emergence of linguistic areas. The changes in Sri Lanka have clearly beeen unidirectional, from the local languages to the immigrant languages. See also Ross (2001), who considers unidirectionality to be part of the definition of metatypy. 7.5 Identification of source language can be obvious Fourth, historical linguists are often hesitant to point to a particular source language for areal changes – for example, in the Balkans. There are cases where it is very possible to identify a single source language, such as SLM
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
7.2 Linguistic areas are not always characterized by simplification, but in many instances also by increasing complexity
Peter Bakker 155
and to some extent also SLP, where Tamil is the obvious source language in most communities.
Conclusions
The most important conclusion of this study is that convergence or metatypy can take place rapidly. It is often assumed that convergence as described above for Sri Lanka is a slow process, taking centuries, if not millennia. This is what Foley (1986) claims for cases of Austronesian languages converging towards Papuan languages. Austronesian languages of Papua New Guinea generally have SVO word order and prepositions. Papuan languages generally have postpositions and verb-final word order. Motu is genetically Austronesian, but it is spoken in an area surrounded by Papuan languages. Motu shows the typically Papuan patterns of verb-final word order and postpositions. Compare the following Motu sentence with the equivalent in the genetically related Austronesian language Tolai (Foley, 1986, 281–2): (11)
tau ese au-na imea bogarai-na-i vada e hado man ACT tree-the garden middle-its-at PERF 3SG plant
(Motu)
(12)
a tutuna i ga oe ra davai livuan ta ta uma the man 3SG PAST plant a tree middle at the garden ‘The man planted a tree in the middle of the garden.’
(Tolai)
In Motu, sentence word order is S-O-LOC-V, but in Tolai it is S-V-O-LOC. In Motu, all nouns have postnominal modifiers (case markers), but in Tolai, article-like elements and adpositions precede the nouns. There is no doubt that Motu changed its word order to align with patterns of neighbouring languages, which also show such structures. Foley also discusses cases of Papuan languages spoken in areas adjacent to Austronesian areas that have changed their word order and other typological features in the direction of Austronesian languages. Apparently, a change in the opposite direction took place there. His dating, however, may be far too deep, for both types: ‘This structural diffusion is the result of language interference over a couple of millennia’ or ‘over at least a couple of millennia’ (Foley, 1986, 282). As we have seen above, exactly the same type of change took place in one or two generations in the Sri Lanka case discussed above, or at most 200 years if one assumes that Dravidianization is older than witnessed in the documentation. The process does not have to take millennia. It is also absolutely possible that the time depth is just as shallow in the case of Papuan–Austronesian contact – that is, several hundred years or decades rather than millennia. Lack of knowledge of the social and historical situation make a time estimate impossible in the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
8
The Sri Lanka Sprachbund
Papuan–Austronesian cases, but in Sri Lanka we have solid historical facts at our disposal. In this chapter I have outlined the main features of the newer languages in the Sri Lanka convergent area. It is unusual in that we have proof that convergence can take place in a relatively or extremely brief period. The Sri Lanka Sprachbund is also unusual in that only pairs of languages were involved rather than multilingual situations. The implications of these observations will have to lead to a revision of some widespread assumptions about convergent areas.
Acknowledgements I have presented data from the Sri Lanka Sprachbund at a number of conferences since 1995. Discussion with audiences in Lunteren (1995), Cambridge (1999), Groningen (1999) and Leipzig (2001) have contributed to this chapter. I am grateful to the editors of this book for their invitation to write about the Sri Lanka situation. The chapter is an offshoot of work initiated when staying at NIAS in Wassenaar in 1995–6. I would like to thank Véronique Vanhée for her assistance with Portuguese. I am also grateful to Sander Adelaar, Shihan de Silva Jayasuriya, Scott Paauw, Ian Smith and Peter Slomanson for discussion. They know much more about Malay and Sri Lanka Portuguese than I do, and they will probably disagree with some of my interpretations of their work. They are, of course, not responsible for my errors and they can be expected not to share my opinions. While revising this chapter, the 2004 tsunami devastated several Sri Lanka Malay communities, killing thousands of Sri Lanka Malay speakers. I would like to dedicate this chapter to these victims.
References Adelaar, K. A. (1991) ‘Some Notes on Sri Lanka Malay’, in H. Steinhauer (ed.), Papers in Austronesian Linguistics, vol. 1, Pacific Linguistics Series, A-81 (Canberra: Research School of Pacific Studies), pp. 23–37. Adelaar, K. A. and D. J. Prentice (1996) ‘Malay: Its History, Role and Spread’, in Stephen A. Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds), Atlas of Languages of Inter-cultural Communication in the Pacific, Asia and the Americas (Berlin: Mouton de Gruyter), pp. 673–9. Bakker, Peter (1995) ‘Over Convergentie in een Sprachbund’ [On Convergence in a Sprachbund], in Erica Huls and Jetske Klatter-Folmer (eds), Artikelen van de Tweede Sociolinguïstische Conferentie (Delft: Eburon), pp. 37–51. Bakker, Peter (1997) ‘A language of Our Own’. The Genesis of Michif – the Mixed CreeFrench Language of the Canadian Métis (New York: Oxford University Press). Bakker, Peter (2000a) ‘Convergence Intertwining: An Alternative Way Towards the Genesis of Mixed Languages’, in D. Gilbers, J. Nerbonne and J. Schaeken (eds) Languages in Contact (Amsterdam: Rodopi), pp. 29–35.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
156
Bakker, Peter (2000b) ‘Rapid Language Change: Creolization, Intertwining, Convergence’, in C. Renfrew, A. McMahon, R. L. Trask (eds), Time Depth in Historical Linguistics, Papers in the Prehistory of Languages (Cambridge: McDonald Institute for Archaeological Research), pp. 585–620. Bakker, Peter (2003a) ‘Mixed Languages as Autonomous Systems’, in Yaron Matras and Peter Bakker (eds), The Mixed Language Debate (Berlin: Mouton de Gruyter), pp. 107–50. Bakker, Peter (2003b) ‘Bazaar Malay’, Paper presented at the Seventh International Symposium on Malay/Indonesian Linguistics, 27–9 June 2003, Berg en Dal/ Nijmegen. Bhaskararao, Peri and Karumuri Venkata Subbarao (eds) (2001) Tokyo Symposium on South Asian Languages: Contact, Convergence and Typology. Special issue of The yearbook of South Asian languages and linguistics 2001 (Thousand Oaks: Sage). Bichsler-Stettler, A. (1989) ‘Aspects of the Sri Lanka Malay Community and its Language’, M. A. Thesis, University of Bern. De Silva, M. W. Sugathapala (1972) Vedda Language of Ceylon (Texts and Lexicon, Münchener Studien zur Sprachwissenschaft, 7 (Munich: R. Kitzinger). De Silva, M. W. Sugathapala (1974) ‘Convergence in Diglossia: The Sinhalese Situation’, International Journal of Dravidian Linguistics, vol. 3, no. 1, pp. 60–91. Dharmadasa, K. N. O. (1974) ‘The Creolization of an Aboriginal Language: The Case of Vedda in Sri Lanka (Ceylon)’, Anthropological Linguistics, vol. 16, no. 2, pp. 79–106. Donegan, Patricia and Davis Stampe (2004) ‘Rhythm and the Synthetic Drift of Munda’, Yearbook of South East Asian languages and linguistics 2004 (Berlin: Mouton de Gruyter), pp. 3–36. Emenau, Murray B. (1964) ‘India as a Linguistic Area’, in Dell Hymes (ed.), Language in Culture and Society (New York: Harper & Row), pp. 642–53; reprinted from Language (1956), vol. 32, pp. 3–16. Ferraz, Luiz Ivens (1987) ‘Portuguese creoles of West Africa and Asia’, in: Glenn G. Gilbert (ed.), Pidgin and Creole Languages. Essays in Memory of John E. Reinecke (Honolulu: University of Hawaii Press), pp. 337–60. Foley, W. A. (1986) The Papuan Languages of New Guinea (Cambridge: Cambridge University Press). Gair, James W. (1976) ‘The Verb in Sinhala, With Some Preliminary Remarks on Dravidianization’, International Journal of Dravidian Linguistics, vol. 5, no. 2, pp. 259–73. Gumperz, John and Robert Wilson (1971) ‘Convergence and Creolization: A Case from the Indo-Aryan/Dravidian Border in India’, in D. Hymes (ed.), Pidginization and Creolization of Languages (Cambridge: Cambridge University Press), pp. 151–67. Heine, Bernd and Tania Kuteva (2001) ‘Attributive Possession in Creoles’, Paper available on the Pidgins and Creoles Archive, at: http: www.pca.uni-siegen.de/doc/7/ 7.pdf; accessed June 23 2001. Hussainmiya, B. A. (1987) ‘ “Melayu Bahasa”: Some Preliminary Observations on the Malay Creole of Sri Lanka’, in B. A. Hussainmiya (ed.), Lost Cousins: The Malays of Sri Lanka (Malaysia: Universiti Kebangsaan), pp. 153–72. Jackson, Keith (1990) Sing Without Shame. Oral Traditions in Indo-Portuguese Creole Verse, With Transcription and Analysis of a Nineteenth-century Manuscript of Ceylon Portuguese. (Amsterdam: John Benjamins). Jayasuriya, Shihan de Silva (1995) [1996]. ‘Portuguese and English Translations of Some Indo-Portuguese Songs in the Nevill Collection’, Journal of the Royal Asiatic Society of Sri Lanka, New Series, vol. 40, pp. 1–102.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Peter Bakker 157
The Sri Lanka Sprachbund
Jayasuriya, Shihan de Silva (1996) ‘Indo-Portuguese Songs of Sri Lanka: The Nevill Manuscript’, Bulletin of the School of Oriental and African Studies, vol. 59, no. 2, pp. 253–67. Jayasuriya, Shihan de Silva (1997) [1999]. ‘Portuguese and English Translations of Oersaan and Falentine: Hugh Nevill Collection of Indo-Portuguese Verses’, Journal of the Royal Asiatic Society of Sri Lanka, New Series, vol. 40, pp. 1–102. Jayasuriya, Shihan de Silva (1999) ‘Portuguese in Sri Lanka: Influence of Substratum’, Journal of the Royal Asiatic Society, Third Series, vol. 9, part 2, pp. 251–7. Jayasuriya, Shihan de Silva (2003) ‘Reduplication in Indo-Portuguse, MalayoPortuguese and Sino-Portuguese’, in Silvia Kouwenberg (ed.), Twice as Meaningful. Reduplication in Pidgins, Creoles and Other Contact Languages, Westminster Creolistics Series 8 (London: Battlebridge), pp. 185–91. Karunatillake, W. S. (1982) ‘Nominal Inflection in Sri Lanka Gypsy Telugu: An Outline’, International Journal of Dravidian Linguistics, vol. 11, no. 2, pp. 377–82. Kuiper, F. B. J. (1967) ‘The Genesis of a Linguistic Area’, Indo-Iranian Journal, vol. 10, pp. 81–102. Lee-Smith, Mei W. (1996) ‘The Hezhou language’, in Stephen A. Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds), Atlas of Languages of Inter-cultural Communication in the Pacific, Asia and the Americas (Berlin: Mouton de Gruyter), pp. 865–73. Lee-Smith, Mei W. and Stephen A. Wurm (1996) ‘The Wutun language’, in Stephen A. Wurm, Peter Mühlhäusler and Darrell T. Tryon (eds), Atlas of Languages of Intercultural Communication in the Pacific, Asia and the Americas (Berlin: Mouton de Gruyter), pp. 883–97. Li, Charles N. (1983) ‘Languages in Contact in Western China’, Journal of East Asian languages, vol. 1, pp. 31–51. Li, Charles N. (1984) ‘From Verb-medial Analytic Language to Verb-final Synthetic Language: A Case of Typological Change’, Proceedings of the Tenth Meeting of the Berkeley Linguistics Society (Berkeley, Calif.: Berkeley Linguistics Society), pp. 173–85. Li, Charles N. (1986) ‘The Rise and Fall of Tones Through Diffusion’, Proceedings of the Twelfth Meeting of the Berkeley Linguistics Society: 173–85. Masica, Colin P. (1971) Defining a Linguistic Area. South Asia (Chicago: University of Chicago Press). Masica, Colin P. (1998) The Indo-Aryan Languages (Cambridge: Cambridge University Press). Masica, Colin P. (2001) ‘The Definition and Significance of Linguistic Areas: Methods, Pitfalls, and Possibilities (with Special Reference to the Validity of South Asia as a Linguistic Area)’, in Bhaskararao and Subbarao (eds), Tokio Symposium, pp. 205–67. Matras, Yaron (1996) ‘Prozedurale Fusion: Grammatische Interferenzschichten im Romanes’, in Martin Haase and Nicole Nau (eds), Sprachkontakt und Gramatikalisierung, special issue of STUF, vol. 49, no. 1, pp. 60–78. Matras, Yaron (1998a) ‘Utterance modifiers and universals of grammatical borrowing’, Linguistics, vol. 36, no. 2, pp. 281–331. Matras, Yaron (1998b) ‘Convergent development, grammaticalization, and the problem of “mutual isomorphism”, in W. Boeder, Chr. Schroeder, Karl Heinz Wagner and Wolfgang Wildgen (eds), Sprache in Raum und Zeit. In Memoriam Johannes Bechert, Band 2, Beiträge zur Empirischen Sprachwissenschaft (Tübingen: Gunter Narr), pp. 89–103. Nanayakkara, A. G. W. (2001) Census of Population and Housing 2001 (Sri Lanka: Census and Statistics) (www.statistics.gov.lk/census2001/index.html).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
158
Paauw, Scott (2003) ‘What is Bazaar Malay?’ Paper presented at the Seventh International Symposium on Malay/Indonesian Linguistics, 27–9 June 2003, Berg en Dal/ Nijmegen. Paauw, Scott, Ian Smith and B. A. Hussainmiya (2004) ‘Sri Lanka Malay: The State of the art’, Yearbook of South East Asian languages and linguistics 2004 (Berlin: Mouton de Gruyter), pp. 197–215. Robuchon, Gérard (2003) ‘Malayo Language in Sri Lanka’, paper presented at the Seventh International Symposium on Malay/Indonesian Linguistics, 27–9 June 2003, Berg en Dal/Nijmegen. Ross, Malcolm (1996) ‘Contact-induced Change and the Comparative Method: Cases from Papua New Guinea’, in Mark Durie and Malcolm Ross (eds), The Comparative Method Reviewed: Irregularity and Regularity in Linguistic Change (New York: Oxford University Press), 180–217. Ross, Malcolm (2001) ‘Contact-induced Change in Oceanic Languages in North-west Melanesia’, in Alexandra Aikhenvald and R. M. W. Dixon (eds), Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press), pp. 134–66. Schuchardt, Hugo (1889) ‘Beiträge zur Kenntnis des kreolischen Romanisch VI. Allgemeineres über das Indoportugiesische (Asioportugiesische)’, Zeitschrift für Romanische Philologie, vol. 13, pp. 476–516. Slomanson, Peter (2003) ‘The Sri Lankan Malay Verb: Lankan Or Malay?’ Paper presented at the Seventh International Symposium on Malay/Indonesian Linguistics, 27–9 June 2003, Berg en Dal/Nijmegen. Smith, Ian R. (1977) ‘Sri Lanka Creole Portuguese Phonology’, International Journal of Dravidian Linguistics, vol. 7, pp. 248–406. Smith, Ian R. (1979a) ‘Convergence in South Asia: A Creole Example’, Lingua 48: 193–222. Smith, Ian R. (1979b) ‘Substrata Versus Universals in the Formation of Sri Lanka Portuguese’, Papers in Pidgin and Creole linguistics, vol. 2, pp. 183–200. Smith, Ian R. (1984) ‘The Development of Morphosyntax in Sri Lanka Creole Portuguese’, in M. Sebba and L. Todd (eds), Papers from the Creole conference, York University (York: York University, Dept. of Language), pp. 291–301. Smith, Ian R. (2001) ‘Creolization and Convergence in Morphosyntax: Sri Lanka Portuguese and Sourashtra Nominal Marking Typology’, in Bhaskararao, and Subbarao (eds), Tokyo Symposium, pp. 391–409. Smith, Ian R. and Scott Paauw (2004a) ‘Sri Lanka Malay: Creole Or Convert?’ Paper presented at the SPCL meeting at the Linguistic Society of America, January 2004. Stolz, Thomas (1989) ‘On Portuguese-based Creoles: Diachronic Aspects of Sri-Lanka– Portuguese–Creole Morphology. Nominal Inflection’, Paper read at the Utrecht Summer School on Language Change. Stolz, Thomas and Christel Stolz (1996) ‘Funktionswortentlehnung in Mesoamerika. Spanisch-amerindischer Sprachkontakt’ (Hispano-indiana II). STUF, vol. 49, no. 1, pp. 8–123. Thananjayarajasingham, S. (1973) ‘Bilingualism and Acculturation in the Kuravar Community of Ceylon’, Anthropological Linguistics, vol. 15, no. 6, pp. 276–80. Thomason, Sarah Grey (2001) Language Contact. An Introduction (Edinburgh: Edinburgh University Press).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Peter Bakker 159
On the Roles of Turkic in the Caucasus Area Lars Johanson
1
Turkic in the Caucasus area
In this chapter I shall discuss briefly the roles of the Turkic varieties in the Caucasus area in order to illustrate some processes in historical contact linguistics. The term ‘Turkic’ will be used for ‘Turkic-speaking’, without any genetic or cultural implications. The general term ‘code’ refers to a language or a variety of a language. The following kinds of Turkic are found in the area today: Azeri or Azerbaijanian (azærbayjanja) is mainly spoken in the Republic of Azerbaijan and in Iran, by at least 20 million people, and there is a sizeable Azeri-speaking population in the south-eastern part of Georgia. Scattered speaker groups are found in Armenia, Daghestan and Nakhichevan. The linguistic borderline between Azeri and Turkish runs through East Anatolia. For Azeri dialects, see Cafer2lu and Doerfer (1959: 281). Kumyk (qumuqfa) is spoken by about 280,000 people north of the Azeri area, in the lowlands on the north-easternmost fringe of Daghestan. The area extends from Derbend in the south to Achi-Su in the north, close to the lower course of the Terek river. In the south, the area is confined to a narrow strip; the middle part is interrupted by a Dargi-speaking zone. Dialects include Boinaq, Khasavyurt and Khaidaq (Benzing, 1959: 392). Noghay (noγayfa) has fewer than 80,000 speakers. The Black Noghay area extends between the lower reaches of the rivers Terek and Kuma in Northern Daghestan, while the area of Central Noghay is largely situated in the Stavropol’ territory. Small groups of White Noghay are found in Chechnya and Karachay-Cherkessia, on the Kuban’ river and its tributaries. Karachay-Balkar (qarafay-malqarfa) is spoken by about 150,000 Karachay and 85,000 Balkar in the Alpine ravines on the northern flank of the Greater Caucasus range. The Karachay inhabit the uppermost river 160 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
7
Lars Johanson
161
Small groups of Stavropol’ Turkmen (türkpen, Russian truxmen) live near the speakers of Central Noghay of the Stavropol’ territory. Several brands of Turkic meet in the Caucasus area. Azeri belongs, like Turkish, to the western subbranch of the South-western – Oghuz – branch. Stavropol’ Turkmen belongs to the eastern subbranch of the same branch. Kumyk and Karachay-Balkar belong to the western, Kipchak-Kuman, subbranch of the North-western – Kipchak – branch. Noghay belongs to the southern – Aralo-Caspian – subbranch of the same branch. For the purposes of this chapter, the Caucasus area will be divided into a Northern and a Southern sub-area. The Northern sub-area corresponds, from the European point of view, to Ciscaucasia and consists of the northern side of the Greater Caucasus range, the plains north of it, and the northern coastal region. The Southern sub-area consists of the southern coastal region, Transcaucasia, including extensive adjacent zones of Kartvelian, Armenian, Iranian and Neo-Aramaic varieties.
2
Migrations
The Turkic languages are genealogically related to each other in a straightforward way, and exhibit rather well-defined common features. This family’s huge geographical area of distribution extends from the south-west, Turkey and its neighbours, over Western Turkistan to the south-east, to Eastern Turkistan and further into China. From there it stretches to the north-east, via southern and northern Siberia, up to the Arctic Ocean; and to the northwest, across western Siberia and Eastern Europe, – today even to Western Europe. In the past, the area also included compact Turkic-speaking regions in the Ponto-Caspian steppes, in the Crimea and so on. There have been massive displacements of Turkic-speaking groups throughout their known history. Numerous movements away from relatively homogeneous speech communities have caused successive splits into groupings and subgroupings of genealogical descendants. The family has undergone differentiation into a few primary branches, known as Oghuz, Kipchak, Bulgar and so on. Through further differentiation, more specialized kinds of Turkic have emerged as secondary or tertiary branches. From here the family tree branches out further into dialect groups, and regional and local dialects. The continuous migrations have led to the separation of related varieties, so that these do not occur in clear-cut geographic clusters. About twenty Turkic languages, in the political sense, are used today, each consisting of a set of dialects plus a standard variety with a certain area of validity.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
valleys in Karachay-Cherkessia. The Balkar live to the south, east and south-east of the Karachay area, in the higher valleys of KabardinoBalkaria. Dialects include Karachay, Malqar (Cherek), Baksan, Chegem and Khulam-Bezinga (Pritsak, 1959: 342; Appaev, 1960).
3
The Roles of Turkic in the Caucasus Area
Contact processes
The continuous movements and displacements of Turkic groups have led to numerous contacts with other codes, both inside and outside the family. When dealing with traditionally nomadic populations, we may reckon with linguistically relatively homogeneous tribes: basic communities, groups of families living together under the same leaders, and sharing certain customs. Contact influence is already operative at this level, because of exogamy and other kinds of encounters with outside groups. There may be influences from mutually intelligible codes and from foreign codes. Similar phenomena typically occur at higher levels, within tribal confederacies. These alliances for common political purposes are not necessarily kin-based, united by ties of descent from common ancestors. In the contact history of Turkicspeaking groups, such confederacies have usually been multidialectal and even multilingual. The differentiation of Turkic has been caused partly by complicated contact processes, in which Turkic varieties have copied foreign linguistic elements from their neighbours – some more and some less. Intensive contacts with Mongolic, Iranian and Slavic varieties have resulted in convergence of different kinds (Johanson, 2002). Since the spread of Islam among the Turks from the tenth century onwards, most Turkic varieties have been influenced heavily by Arabic, mainly through Persian mediation. Because of contactinduced change, many Turkic codes have become more similar to genetically unrelated codes. Code shift has often taken place. The contacts have also induced numerous changes in non-Turkic varieties, or led to code shift of non-Turkic groups.
4
The introduction of Turkic
When and how did Turkic enter the Caucasus area? The processes of migration have been highly complex, and most answers given so far are partly speculative. There are many open historical questions with respect to the origins of the Turkic populations and their linguistic backgrounds. Turkic does not belong to the indigenous languages of the Caucasus area (Nakh-Daghestanian, Abkhaz-Circassian, Kartvelian), but was introduced relatively late, at least later than the ancestors of Armenian and some Iranian languages. One early main wave of immigration came from the north and included a number of different nomadic complexes. The history of the Eurasian nomads offers many examples of sudden and far-flung political expansion. Numerous tribal complexes with Turkic-speaking elements have passed into the Caucasian area. Turkic-speaking groups probably participated in the campaigns or political organization of the Hunnic tribal groups, which appeared on the northern slopes of the Caucasus in the middle of the fourth century. But it is unknown
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
162
163
to what extent the Huns or their ruling groups can be defined linguistically as Turkic. Nothing is known about a Hunnic language. This case teaches us an important lesson. The old nomadic complexes were linguistically and ethnically heterogeneous, comprising elements of different origin. The known designations refer to the representative groups of the tribal confederacies, but do not tell us which tribes were included. The ethnic or linguistic affiliation of a constituent tribe is not necessarily identical with that of the leading elite group of the complex. Titles are not limited to specific linguistic groups. Given the heterogeneous composition of the nomadic complexes, it is often impossible to determine with which tribes or under which tribes Turkic-speaking groups appeared in the Caucasus area. The same principles apply to conglomerates such as Bulgars, Western Türk, Khazars, Oghuz, Pechenegs, Kumans, Mongols, later Kipchak groups and so on. • From the end of the fifth century, the Bulgars (Onóguroi, Búlgaroi) appeared in the steppes along the Kuban’ river north-west of the Caucasus, and later established the independent state of Great Bulgaria on the eastern Pontic steppes, with its centre on the Kuban’ river. We do not know what languages the confederated tribes spoke. • Under the rule of the large tribal confederacy of the Western Türk, numerous tribes moved westwards and spread out over the steppes as far west as the lower Volga. Again, we know very little about the linguistic and ethnic composition of the tribes dominated by the ruling groups. • When the West Türk Empire was destroyed in 657–9, tribal groups known as Khazars (Xázaroi) established a realm that existed from the eighth to the eleventh centuries. It extended over the Volga – Caspian – North Caucasian – Pontic steppes including the Kuban’ region and the Daghestanian foothills. • The Khazars, whose state religion was Judaism, fought against the Arabs in the Caucasus region – for example, at Derbend and Daryal Gorge (Daryal deresi) in the upper Terek valley. One Bulgar conglomerate remained along the Kuban’ river, in so-called Black Bulgaria, in the southern part of the Khazar realm. Newcomers belonging to Kipchak Turkic confederacies destroyed the Khazar state and finally absorbed its population, including the groups of Black Bulgaria. • The Pechenegs roamed on the steppes up to the eleventh century and were finally defeated in Bulgaria. • The Kumans, who appeared at the middle of the eleventh century, destroyed the Khazar state. They entertained good relations with Georgia. Kuman groups served as auxiliaries in the Georgian army and helped to defeat the Seljuks in 1121. • At the beginning of the thirteenth century, Mongol armies invaded the Pontic steppes, defeating the Kumans. Under the so-called Golden Horde,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Roles of Turkic in the Caucasus Area
Derbend and the river Terek in the Caucasus area became the frontier against the Ilkhans, the Mongol successor-state in Persia. • Under the overlordship of the Western Türk, Oghuz tribes had migrated westwards rather early, from the end of the ninth century onwards; they were later defeated by Pechenegs and Kumans. • Oghuz tribes also migrated south-westwards, some of them having advanced greatly by the second half of the tenth century. An Islamic tribal confederacy known as the Seljuk set out for conquests in Iran, the Arab countries and the Byzantine Empire. They seized Persia, overran Transcaucasia in 1045, moved on to Mesopotamia, and, exerting their dominion over the Caliphate (1055), conquered the greatest parts of its empire, finally taking possession of Anatolia. These confederacies laid the foundations for the south-western branch of Turkic in Turkmenistan, Iran, Transcaucasia, Anatolia and the Balkans. Thus, with the Seljuk conquest in the eleventh century, Oghuz Turkic arrived in Azerbaijan. The linguistic effects of the old migratory movements are difficult to assess. The early Oghuz in the Ponto-Caspian steppe regions, the Pechenegs and so on, are only known from titles and personal and tribal names. Not a single line in Khazar has been found. The only relatively well-attested old language is Kuman, as documented in the Codex Cumanicus (fourteenth century). The political expansion of the old nomad complexes was not necessarily tantamount to the expansion of Turkic. Some of the incoming complexes were led by Turkic elites. Others were led by non-Turkic elites, though they contained Turkic subgroups. Interestingly, the armies under nominally Mongol leadership were predominantly Turkic-speaking; the Mongol expansion meant a breakthrough for the expansion of Kipchak Turkic, rather than for Mongolic.
5
The Turkic varieties
Let us summarize some basic facts about the individual Turkic varieties of the Caucasus area: • Kumyk entered the area in the early Middle Ages. The territory of Kumyk has had vivid contacts with elements from the steppe from the Hunnic era onwards; that is, in the Bulgar, Khazar, Kipchak and Mongol periods. The Kumyks may be partly of local (that is, Daghestanian) origin, including speakers of Nakh-Daghestanian who moved to West Kipchak at different periods. The language clearly goes back to the early kind of Kipchak Turkic that is known from the eleventh century onwards and may have spread into the north Caucasian area at the time of the Khazar empire. It became one of the Turkic languages of the lowland areas that functioned as a lingua franca in the region, serving trade and inter-group
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
164
165
communication. Kumyk was dominant in the eastern part of the north Caucasus area, often used by Chechens, Avars and other tribes. Today, few non-Kumyks learn the language, since Russian has taken over the inter-group communication function. • Noghay arrived relatively late in the area, namely after the break-up of the Golden Horde at the end of the fourteenth century. It is a Kipchak language of a later type, not of the early western type to which Kumyk belongs. The Noghays once formed a nomadic state on the lower Volga under Russian supremacy; they were later expelled from the area by the Kalmyks and scattered all over the Ponto-Caspian steppes. Most groups were absorbed by other Kipchak elements and gave up their language. Only the parts that lived in the north Caucasian area maintained it. The Noghays, originally Kipchakicized Mongols, also comprise Turkicized groups of local origin. Noghay once functioned as a lingua franca in north-western and central Daghestan. • The speakers of Stavropol’ Turkmen are descended from Turkmen tribes brought from the Mangyshlak region to the North Caucasus in the eighteenth century. • The Karachay-Balkar language is an older entrant from the steppes. The ancestors of its speakers must have lived in the north Caucasian plain, from where they were pushed into the mountainous areas by the Mongol invasion of the thirteenth century. Their new habitat was the home of Alan tribes, which had remained in North Caucasia since the Khazar era. The Karachay-Balkar developed a close symbiosis with the Alans and were strongly influenced by them. They have even been called Alans themselves by some of their neighbours. In the late Middle Ages, at the beginning of a period of cooling weather that caused highlanders to move downhill, the Karachay-Balkar were driven by Kabardians and Circassians to poorer locations in the highlands. Today’s Karachay-Balkar speakers seem to be a mixture of elements from Bulgar, Oghuz and Kipchak tribes, with Iranian Alans and assimilated indigenous Abkhaz-Circassian and even Kartvelian (Svan) elements. In particular, groups of Digor Ossetic speakers have shifted to KarachayBalkar. The Kabardians refer to the Karachay-Balkar as Ossetes, whereas the Digor Ossetes call them As or Assi. The Karachay-Balkar differ from other Turkic groups in terms of specific anthropological and cultural features, sharing, for example, the Nart heroic epics with Ossetes, Kabardians, Abkhaz and others. The language of the so-called Black Bulgars settling in the Kuban’ region in the south part of the Khazar empire may be one of the basic elements of Karachay-Balkar. These Bulgars were later assimilated by other Turkic-speaking groups. The ancestors of the Karachay-Balkar were probably Kipchakicized at the end of the eleventh century (Pritsak, 1959: 341). However, the attempts to connect balqar ~ bolqar ~ malqar with the name of the Bulgars seem to rest on a superficial sound resemblance.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Roles of Turkic in the Caucasus Area
• Many different ethnic and linguistic elements have been involved in the history of Azerbaijan. Its northern part was the ancient Albania, inhabited by the Albanói, the linguistic ancestors of the present-day Udi speakers. Numerous peoples from the east passed into the area. The Seljuk conquest in the eleventh century eventually led to its massive Turkicization, with older Iranian and Paleo-Caucasian populations being largely assimilated. The Mongol conquests brought in new waves of Oghuz and other Turkic groups, which additionally advanced Turkicization. Azerbaijan is now an almost exclusively Turkic-speaking country.
6
Contact languages
Turkic has been in contact with most languages of the Caucasus area (Gecadze, 1977). This relatively limited territory displays a remarkable linguistic diversity, with numerous languages spoken under symbiotic conditions for more than 4,000 years, probably with widespread bi- and multilingualism. (For a survey of peoples and languages, see Geiger et al., 1959.) • Karachay-Balkar has had close contacts with the Digor dialects of the Iranian language Ossetic, which goes back to varieties of the nomadic complex of the Alans. It has also had contact with the Abkhaz-Circassian languages Abkhaz, Abaza, Adyghe and Circassian, and the Kartvelian language Svan. • Noghay used to be in close contact with Kalmyk. White Noghay, which formerly interacted with Circassian, Abaza and Karachay, now has little contact with Caucasian languages. Small groups near Mineral’nye vody still live in the neighbourhood of Abkhaz and Circassian groups. Black Noghay in Northern Daghestan is in touch with Kumyk. • Kumyk has been in contact with its immediate Nakh-Daghestanian neighbours: for example, Avar, Lak, Dargi and Khaidaq. • The language of the Stavropol’ Turkmen has been influenced by Noghay and Kumyk. Judging from the Noghay clan name türkpen, some groups of speakers are being absorbed by the Noghays (Golden, 1992: 392). • Azeri has been in contact with Tatic (Tat and Talysh), Persian, Lezghic, Udi, Armenian, Georgian, Mingrelian, Kurdic, Neoaramaic and so on.
7
Linguistic effects
When trying to assess the linguistic effects of these encounters, we face, as always in contact linguistics, the problems of distinguishing inherited features from those that are copied, and determining the source and direction of contact influence. The effects of the old migration waves that reached the Caucasus area from the north are the most difficult ones to interpret. In the following, some ascertained or potential contact-induced phenomena will be mentioned.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
166
167
The linguistic effects will be discussed in terms of the model of codecopying; see, for example, Johanson (2002: 8–19). Copying means the insertion of copies of elements of a model code into a basic code. Global copying concerns morphemes or morpheme sequences as a whole with their material, semantic, combinational and frequential properties. Selective copying concerns individual material, semantic, combinational and frequential properties. Copying never just means the transfer of elements from one code to another code, but always implies creative adaptation. In the case of ‘take-over’ influence (adoption), speakers take over copies from a dominant foreign code in their own primary code. In the case of ‘carry-over’ influence (imposition), speakers carry over copies from a given primary code into their variety of a foreign code. In the case of code shift, copies carried over from a former primary code may remain as substratum features in the new primary code. In contact situations, code maintenance – with adstratum features – and code shift – with substratum features – yield different results. Many contactinduced phenomena found in varieties of the Caucasus area may result from substratum influence following code shift. In this case, the question is: Which groups have shifted, the existing local populations or the incoming groups? 7.1 Lexical copying Lexical copying between Turkic and non-Turkic varieties is common in the Caucasus area. The Turkic influence is considerable in various regions. For centuries, until recent times, Turkic varieties have been prestige codes represented by elite groups ruling large parts of the area (Menges, 1968: 176). In North Caucasus, Turkic varieties of groups occupying the lowland areas used to function as lingua francas for trade and inter-group communication. Up to the period of Russian dominance, mountaineers coming down to the lower regions to trade or to work would learn these Turkic varieties. It was quite common for Caucasians to be at least bilingual. The lowland markets were instrumental in spreading Turkic varieties (Wixman, 1980). As mentioned earlier, north-western and central Daghestan was dominated by Noghay, while north-eastern Daghestan was dominated by Kumyk. Up to the first part of the twentieth century, there was a tendency to extend Turkicization into the North Caucasus area, for political and economic reasons. Even more important was the transregional validity of Azeri, also throughout Persia, a situation that continued at least until the eighteenth century. Azeri was used for inter-group communication in south-eastern Daghestan, and even further north-west. In the Northern subarea, Turkic and non-Turkic varieties have exchanged a quantity of loanwords. The foreign lexical impact on Noghay and Kumyk is, however, relatively moderate. For the Turkic lexical influence on nonTurkic varieties, see Wagirov (1989, esp. ch. 2); Džidalaev (1990); Alekseev and Wejxov (1997: 117–18 and the literature noted there).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Roles of Turkic in the Caucasus Area
In the habitat of the speakers of Karachay-Balkar, a homogeneous cultural area with many similarities between its various linguistic communities, the lexical exchange has been relatively balanced. The Ossetic vocabulary mirrors long-standing and intensive Turkic contacts. The way of the borrowing may be difficult to determine, since many copied Turkic words are also found in other Ponto-Caspian languages. Though code shift to Turkic must have been common, there are no indications of heavy substratum effects on Karachay-Balkar. The basic lexicon is of Turkic origin (Siemeniec-Gopas, 2000). The loanwords from the adstratum languages belong to restricted domains, mainly nouns relating to the local conditions of life, flora and fauna, local cultural terminology and so on. Karachay-Balkar is rich in Ossetic and Kartvelian loanwords. The numerous loans from Digor Ossetic are also present in Karachay, although its speakers have not had any direct contact with this variety during recent centuries. There are certain differences. Thus the Ossetic word cægat ‘north, the northern side of a mountain’ has been copied into Balkar as feget in the same meaning, while it means ‘forest’ in Karachay. The long and strong Alan influence is manifest in numerous ethnonyms and topographic names. Some Alan names of saints and other terms from the Christian sphere have been preserved (Pritsak, 1959: 341). The Ossetic numerals were previously used by the Balkar (Abaev, 1933: 80–1). An influence from the Caucasian environment (Comrie, 1981: 212) is found in the former Karachay-Balkar and Kumyk vigesimal counting system, with numerals based on the concept of twenty – for example, Karachay-Balkar eki jiyirma bla on (‘two twenty with ten’) for ‘fifty’, Kumyk üf yigirma (‘three twenty’) for ‘sixty’.1 The situation in the Southern sub-area is quite different. The vocabulary of Transcaucasian Turkic is heavily influenced by a substratum: groups speaking Iranian and Caucasian varieties have shifted to Turkic, carrying over local terminology to their new primary code. According to Stilo (2005), the substratum is essentially Tatic. Thus words such as tum ‘seed’ and jüt ‘wooden plough; pair’ match Tat tum and jut exhibiting the typical sound changes -xm- > -m- and -ft- > -t-; compare Persian toxm, joft. Persian has for centuries been a culturally dominant adstratum code from which languages of the whole area have copied lexical items relating to a wide range of domains, in particular cultural, erudite vocabulary. Azeri and Kumyk have exerted strong lexical influence on neighbouring Nakh-Daghestanian languages. For example, Lezghian has copied Turkic words extensively, many of which ultimately go back to Persian and Arabic elements; for example, daγ ‘mountain’, saburlu ‘patient’, rangsuz ‘colourless’. It has also adopted Turkic suffixes; for example, -fi in words such as kolxozfi ‘kolkhoz farmer’, and even the plural suffix -lar/-ler in words such as daγlar ‘mountains’ (Haspelmath, 1993: 72, 101, 120). On the Turkic lexical influence on Georgian, see Golden (1994). There is a very extensive pattern of borrowings from later Oghuz Turkic, coming from
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
168
Lars Johanson
169
the Turkish dialects spoken in the Eastern part of the Ottoman Empire and from the Safavid state in Iran (Golden, 1999: 54)
All Turkic varieties of the area have copied phonetic and phonological features including certain common characteristics of the Caucasian languages. Strongly aspirated voiceless stops, for example, [thaph] ‘to find’, occur in various contact zones. In Azeri dialects, strongly aspirated unvoiced stops occur typically in initial, but also in final position, for example, [thoph] ‘cannon’, Nukha dialect [syth] ‘milk’. Kumyk voiceless stops are aspirated, for example, [ph], [th] and [qh], except after other consonants and word-finally. In Karachay-Balkar, phonological copying has created a consonant system whose structure is very similar to that of Ossetic. Voiceless obstruents form pairs based on the feature of aspiration: [ph] versus [p]; [th] versus [t]; [kh] versus [k]; [qh] versus [q]; and [ h] versus [ ]. The aspirated sounds occur word-initially, for example, [khεlthir-] ‘to bring’; [ ha] ‘hair’; [qhuru] ‘dry’; [thana] ‘calf’; in intervocalic position, for example, [qhɯ hɯr-] ‘to shout’; [qhathɯn] ‘woman’; and as the second part of consonant clusters, for example, [øtkhεn] ‘passed’; [ølthyr-] ‘to kill’; [qhorqhup] ‘fearing’ (Pritsak, 1959: 349–50). Karachay-Balkar has also copied the Caucasian articulation of ejectives, obstruents produced with a glottalic egressive airstream mechanism. KarachayBalkar ejectives are mainly suffix-initial voiceless obstruents, for example, [batht’ ɯ] ‘sank’ (bat- ‘to sink’ + past tense), obviously used as partitioners that makes bound morphemes identifiable in spite of their phonetic variability. In Kumyk dialects, certain consonants are articulated with a glottal stop, mainly in loanwords from the Caucasus, but also in native words, for example, [qɯrq] ‘forty’. This feature is typical of the southern Khaidaq dialect spoken in the Derbend region but is also to some extent present in the central Kumyk dialect spoken in Boynaq (Bujnaksk) and neighbouring villages. The northern Azeri dialects of Zaqatala and Gax, spoken in regions bordering on Daghestan and Georgia, exhibit pharyngalized vowels typical of the Lezghic language Tsakhur. In words such as [olka] ‘country’ and [o ] ‘three’ they obviously function as substitutes for the rounded vowels [ø] and [y]; compare Standard Azeri [ølkæ], [y ]. These sound types may be a result of adstratum influence (adoption, ‘take over’ processes) or substratum influence (imposition, ‘carry over’ processes), or both. In the North Caucasus area, adoption seems more plausible in many cases. For example, the Karachay-Balkar glottalic stops are marginal phonemes without a significant functional load. This is also the case in the neighbouring Ossetic dialect, whose Iranian sound system has not undergone any radical changes.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
7.2 Sounds and sound patterns
The Roles of Turkic in the Caucasus Area
The situation seems to be different in the Southern sub-area. The innovations found in Azeri dialects of this area seem to be a result of strong substratum influence. Northern dialects have much in common with some southern dialects spoken in Iran – for example, the Khoy dialect. The front stops [k] and [g] are palatalized, e.g. [kjim] ‘who’, [gjyn] ‘day’. In certain northern and southern dialects they are pronounced as palatal affricates. Stilo (1994) suggests that the obstruents [], [t ], [], [d ], [kj] and [gj] are the result of systemic shifts because of a very early substratum. This could be a result of code shifts to Turkic from Iranian varieties of the Tatic type, which, in turn, may also have had Caucasian substrata. North-western Persian and Armenian exhibit similar consonant systems. According to Stilo, Armenian was probably influenced by Urartian, and the Iranian dialects could have inherited the same system. Another Azeri pecularity that may have emerged in a similar way is the voicing of [q] to [G], e.g. [Gɯz] ‘girl’. Some dialects also display a rather un-Turkic fricativization of labial consonants, e.g. [gjælif] < [gjælip] ‘has come’ in the Genje dialect. Another phonological substratum effect may be the infringements on the systems of sound harmony in various Turkic varieties in the Caucasus area. Deviations of this kind are found in some Karachay-Balkar dialects as a result of contacts with Abkhaz-Circassian languages (Akbaev, 1963: 47); they are also typical of Kumyk dialects and of Azeri dialects spoken in zones of close contact with Nakh-Daghestanian languages such as Dargi, Lak and Tsakhur – for example, Khaidaq [giraman] ‘I enter’ (Dmitriev, 1940, 1962; Kerimov, 1967). In Azeri dialects, the sound harmony is disturbed by fronting tendencies. Many suffixes are invariable – that is, exhibit non-harmonic forms, for example, in the Nukha and Baku dialects. On the other hand, Turkic-like sound harmony tendencies may be found in some non-Turkic varieties. Thus the emergence of a front versus back vowel harmony and a rounded versus unrounded vowel harmony may be observed in the Lezghic languages Udi and Lezghian as well as in some other Caucasian languages. 7.3 Grammar Contacts between Turkic and non-Turkic varieties have led to changes in grammatical structures in the sense of adstratum and substratum effects. The Turkic varieties have, however, essentially retained their basic grammatical properties. Grammatical morphemes are mainly of Turkic origin, though sometimes used in new ways according to foreign models. Even where foreign structures have been copied, native morphological material has normally been employed. The Turkic varieties do not share the common morphosyntactic feature of what has been called the Caucasus Sprachbund, namely the characteristic ergative alignment. On the other hand, it seems that Turkic has exerted considerable grammatical influence on many non-Turkic languages of the area.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
170
Lars Johanson
171
Some North Caucasian areal phenomena may have resulted from, or have been stimulated by, contacts with Turkic. Although Ossetic has essentially preserved its Iranian character, it shows certain signs of foreign grammatical influence. It lacks the complex system of verbal suffixes typical of Turkic, but its nominal inflection tends towards the agglutinative type, for example, with case endings following the plural endings. One North Caucasian areal feature is the syntactic subordination by means of non-finite verbal forms, for example, adverbial clauses with converbs instead of finite verbs. For example, the use of the Ossetic converb in -gæ may have been supported by contact with Turkic in the sense of frequential copying; that is, shift in relative frequency under the influence of an element in a model code. Abkhaz-Circassian and Ossetic converb forms of the verbum dicendi ‘to say’ are used as citation particles to convey reported speech or thought, which matches the typically Turkic use of converb forms such as Karachay-Balkar deb ‘saying’, for example, Ossetic zæ γ gæ. Following Turkic models, Ossetic has also developed postpositions such as kæ s gæ, converb of ‘to look’, meaning ‘according to’, for example, mænmæ gæ s gæ ‘according to me’ (‘to me looking’); compare Karachay-Balkar köre. 7.3.2 The Southern sub-area The situation in the Southern sub-area is, again, rather different. In a geographically contiguous area we are confronted with certain common tendencies of the individual languages, independently of their linguistic relatedness. I shall comment on some cases in which Turkic varieties have played more or less important roles. 7.3.2.1 Intraterminals. Several languages of the Southern sub-area exhibit similarities in their aspect-tense structures (for Kartvelian, see Christophe, 2004). Intraterminals – that is, presents and imperfects – are often formed with inessive constructions based on locative metaphors meaning ‘to be in[side]’, for example, Armenian ertumem ‘I am going’; Talysh kårdæ (< kårde-dæ) ‘is doing’; Lezghian fizwa ‘is going’; Bagvalal ig’iyax ira ‘is doing’; Neo-Aramaic brixšin ‘I am going’. The development of these items may well be interpreted as aspectual convergence in the direction of Turkic, where intraterminality is a stable grammatical category. Items such as Azeri getmækdæ and Turkish gitmekte
‘is going’ are widespread in the Turkic family, for example, Modern Uyghur qilmaqta ‘is doing’, qilmaqta idi ‘was doing’. There are numerous parallels between the verbal systems of Azeri and some adjacent non-Turkic varieties. Nakh-Daghestanian languages such as Andi, Avar, Archi and Lezghian possess old intra-terminal items that have undergone defocalization, such as the Azeri old present in -Ar, for example, yazar ‘will write, writes’; compare the new focal present in -Ir, for example, yazïr ‘is writing, writes’ (Johanson, 2000c: 92–3). The Lezghian -da form has,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
7.3.1 The Northern sub-area
The Roles of Turkic in the Caucasus Area
by virtue of a similar value, general and partly modal functions interpretable as ‘habitual’, ‘future’, and so on, for example, fida ‘goes, will go’, fidaj ‘went, would go’. It has therefore been characterized as a ‘Future’ with a ‘future/ habitual polysemy’ (Haspelmath, 1993: 276). The item kulše ‘throws, will throw’ of Khvarshi (Daghestanian Tsezic group) displays analogous uses. The Iranian languages Talysh and Tat exhibit similar items with habitual and modal (mostly potential) readings, for example, Talysh ahåndi ‘read, used to read, would read’. The relevant similarities between Azeri and Tat are discussed by Grjunberg (1962); see the comparative table pp. 17–18. 7.3.2.2 Evidentials. Evidentials, categories that state the existence of a source of evidence for a propositional content (Aikhenvald, 2004: 1), are easily copiable in areal contacts. Turkic languages, which possess stable evidentials of the indirective type (Johanson, 2000, 2003) are known as sources of their diffusion. Indirectives appear in contact areas such as the Balkans, Anatolia, the Volga region and Central Asia (Johanson, 2000a: 83–4; 2003: 288–9). If they appear in a number of languages of the Caucasian area, one is justified in considering possible Turkic influences. The expression of indirectivity is closely connected with postterminality, a view on events from an orientation point posterior to their inherent relevant limit (Johanson, 1971: 283; 2000: 29, 102–4). Expressions of postterminality vary according to the degree of focality, the relative narrowness of the time interval around the orientation point. With high-focal postterminals, the interval is narrow, which yields static or resultative meanings, for example, English is gone. With low-focal postterminals, the interval is wider, which yields ‘perfect’ meanings, for example, English has gone. The development from postterminality to evidentiality can be accounted for as a process of defocalization. Focal postterminals may lose their focality and finally even their postterminal character, which often leads to the emergence of new high-focal postterminals. The widespread development of postterminals into evidentials in the Caucasus region is not necessarily a result of Turkic influence, but contact with Turkic may have supported it in the sense of frequential copying and internal ‘latent tendencies towards indirectivity’ (Johanson, 1996: 87). In most Kartvelian languages, defocalized postterminals have evidential functions. Modern Standard Georgian does not possess any special morphological category to code evidentiality, but one of the meanings of the so-called perfect is regarded as being evidential, for example, uc’vimia ‘it has obviously rained’. This is thought to be a rather recent phenomenon, since the Old Georgian perfect was purely resultative. Christophe (2005) asssumes that Kartvelian evidentials have evolved from high-focal postterminals through processes of cyclic grammaticalization. The Georgian static item sc’eria ‘it is written’ and the resultative dac’erila (< Old Georgian dac’eril ars) originally differed with respect to the degree of focality of the postterminal view. The
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
172
173
former item had the highest degree, whereas the latter became increasingly defocalized. Mingrelian resultatives are products of the same diachronic process of cyclic grammaticalization by which postterminals turn into evidentials, leaving a gap that is filled by new postterminals. The new resultatives take over the original function of the older ones, but only to become increasingly evidential themselves. Laz, Mingrelian, Svan and some adjacent western dialects of Georgian have developed elaborate Turkic-like evidential systems that also include present, imperfect and future items (Harris, 1985: 296–300). The Georgian particle turme ‘apparently’, for example, turme tovli mosula ‘it has apparently snowed’, also expresses evidentiality with non-past events, which is reminiscent of the use of Turkish and Azeri imiš. Some Georgian dialects, for example, Ingilo, spoken in an Azeri surrounding, use the particle qopil (a), a perfect form (‘it has been’) based on the past participle of ‘to be’. It might be interpreted as a semantic copy of Azeri imiš; compare camodis qopil ‘(s)he is apparently coming’ (Boeder, 2000: 314, nt 17) with Azeri gælir imiš. The difference lies in the fact that qopil is an invariable particle, added to inflected verb forms, whereas imiš is added to non-finite forms and bears the personal endings. Boeder still sees good reasons for considering qopil an areal phenomenon due to code copying (2000: 284). The inflected copula is suffixed to finite verb forms in several Kartvelian varieties, for example, qopil-m in Khevsur, an East Georgian dialect that has been in close contact with Chechen-Ingush. The Laz dialects of the Findikli-Arhavi region in Turkey use the evidential particle doren, a form of ‘to be’, after finite verb forms. In the Hopa region, however, the corresponding element is used as a finite copula verb, with the preceding verb form lacking personal endings; this matches the use of imiš. The use of the East Armenian perfect to express evidential meanings corresponding to those of the Georgian perfect may be because of contacts with Turkic and Iranian (Johanson, 1992: 245, 282; 2002: 99, 146–7; Kozintseva, 2000: 414). West Armenian, formerly spoken in Eastern Turkey and thus a contact language of Turkish, has a distinct evidential particle eγer developed from the inferential participle of the verb ‘to be’; see the above-mentioned Ingilo particle qopil. Like qopil, turme and imiš, eγer is a more general evidential particle that is not only used with the perfect (Donabédian, 1996: 95–7). Tat displays the evidential particle miš, a copy of Azeri imiš. The Tat perfect may have evidential readings itself, but the particle is often added to it (Grjunberg, 1963: 88). Turkic-like evidentiality systems exist in numerous Nakh-Daghestanian languages. Defocalized postterminals display evidential functions in Avar. The Nakh languages Chechen and Ingush use the participle xnnə ‘been’ of xlar ‘to be’, for example, a: ra v-eannə xnn-u: ‘he has apparently gone out’ (Boeder, 2000: 314). This particle is reminiscent of the Khevsur particle qopil, but, like Azeri imiš, it bears the inflection marker.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Roles of Turkic in the Caucasus Area
Abkhaz-Circassian languages possess evidentials that seem to have already existed in Proto-Abkhaz and Proto-Circassian in the eighth and ninth centuries. If this is true, it cannot, of course, be attributed to Turkish influence. Chirikba stresses that there was ‘no discernible Turkish presence in Abkhazia prior to the sixteenth century’ (Chirikba, 2003: 266). This does not, however, exclude Turkic impact in general. The discussion on possible contact-induced effects has so far focused one-sidedly on the contribution of Oghuz Turkic – that is, Turkish and Azeri, and very often exclusively on contacts during the period of Ottoman domination. According to Boeder (2000: 227) the evidentiality system of Mingrelian may partly be a result of contacts with Turkish. Svan, which has always been remote from Ottoman domination, may have modelled its system after that of Mingrelian (see Sumbatova, 1999; Friedman, 2000: 357). Although it is still not clear when the evidential variants of the Georgian perfect arose, Boeder nevertheless concludes: ‘For the time being, the hypothesis that the evidential meaning of the Georgian perfect developed under the influence of Turkish–Georgian bilingualism is a real possibility, as far as we know’ (2000: 297–8). While this may be true, the presence of evidentials is certainly not a convergence phenomenon in which only Oghuz Turkic has been involved. The emergence of Turkic-like evidential categories may result from, or have been supported by, other Turkic varieties in the area. Golden notes that even the Georgians have dealt with virtually every Turkic group that entered the North Caucasus region (1999: 96). We are confronted with the much wider issue of Balkan–Pontic–Caucasus–Caspian contacts. The Bulgarian evidentiality system points clearly to pre-Ottoman contacts with Turkic (Johanson, 1996). Certain Kipchak Turkic languages may use the particle e(r)ken, once widespread in the area (Johanson, 2004; forthcoming), as an utterance-final particle added to main clauses, without carrying personal markers. Karakoç (2005: 31) deals with Noghay sentences in which the personal markers are carried by eken, and sentences in which they are carried by the preceding verb, for example, Nege kelgen ekenler? ‘Why do they seem to have come?’; Nege kelgenler eken? ‘I wonder why they have come’. Turkish and Azeri imiš only represents the first model. As we have seen, however, the second model is found in many evidential constructions in the Caucasus area. 7.3.2.3 Syntax. The Southern sub-area exhibits many common syntactic features. Azeri dialects exhibit strong Iranian influences in the syntax. Stilo (2004) discusses possible substratum structures such as the use of the subjunctive after modals for example, istiræm gedæm ‘I want to go’, copied Iranian function words, loss of the question particle mi (replaced by intonation), confusion of subjunctive and optative forms, for example, istir gedæ ~ istir getsin ‘wants to go’ (also typical of Northern Talyshi), and certain loan translations. The unmarked constituent order in Turkic is verb-final. This is also the case in Georgian, Armenian, Nakh-Daghestanian and Abkhaz-Circassian languages
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
174
175
such as Dargi, Lezghian, Chechen, Ingush and so on. Georgian and Armenian have, however, switched from verb-object to object-verb order since the medieval period. This typological reversal may well be the result of areal contacts with languages of the opposite type. In Old Georgian, adjuncts generally followed their nominal head. Turkic languages normally have preposed relative clauses (Johanson, 1992: 274–5; 2002: 135–7); this is also the case in Nakh-Daghestanian. Ossetic is the only Iranian language of the area that sometimes exhibits preposed relative clauses. Postposed relative clauses of the Iranian kind are normally found in spoken Georgian, Armenian, Circassian and Azeri (Stilo, 2004). Spoken Azeri also makes heavy use of postverbal subordinate clauses and regularly places constituents expressing destination/goal after the verb, for example, gediræm evæ ‘I go home’ (Stilo, 2004). The decline of ergative constructions and the emergence of Turkic-like syntactic structures in some Caucasian varieties are not necessarily contactinduced phenomena, but they may be rooted in internal tendencies that have been supported and generalized through contacts with Turkic in the sense of frequential copying. For the loss of ergativity in one dialect of Udi (Gukasjan, 1973: 43), Azeri influence seems to have been the decisive triggering factor.
8
Similar and different roles
The languages of the Caucasus area as defined in the present chapter certainly do not qualify as a single Sprachbund. The sub-areas, however, rather constitute continuous communication areas exhibiting a good deal of bi- or multilateral convergence. It is often impossible to judge the directions of influence. The spread of areal features cannot always be accounted for by one source of diffusion. If we had access to detailed historical data, however, cases of multilateral convergence would probably turn out to consist of series of unilateral and bilateral code-copying processes (Johanson, 1992: 279–84; 2002: 143–9). Many common features observed in the sub-areas may be because of contact-induced frequential copying that has supported latent internal tendencies. The Turkic languages of the Northern and Southern sub-areas differ considerably from one another with respect to their roles. 8.1 The Northern sub-area In spite of a long, symbiotic coexistence with their non-Turkic neighbours, the Turkic languages of the Northern sub-area display relatively little foreign influence. Kumyk and Karachay-Balkar have essentially maintained the features of Early West Kipchak – ‘Kipchak-Kuman’ – as documented in the Codex Cumanicus (Magomedov, 1966; Xabifev, 1966; Berta, 1998). Noghay exhibits little influence from the Caucasus area beyond loanwords, although
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Roles of Turkic in the Caucasus Area
Menges speaks of a strong influence of Circassian (1959: 436). None of the languages shows any major syntactic peculiarities. The situation of Karachay-Balkar, which has long developed in relative isolation from other Turkic languages, is particularly interesting. In the 1930s, the Marr school of linguistics focused on this language as an exemplary crossbreed of Turkic and Caucasian (‘Japhetic’) elements. According to Marr’s theory, languages arise by processes of interweaving and combining – linguistic development is seen as constant ‘mixing’. The alleged ‘mixed’ status of Karachay-Balkar is, however, by no means supported by the data. The neighbouring languages have left relatively little imprint on it. It has retained the character of its Turkic subgroup, just as Ossetic has largely retained its Iranian character. The influence exerted by Ossetic and the North Caucasian languages on Karachay-Balkar is much more limited than the reverse influence. As already mentioned, while language shift must have been common through the centuries, the overall picture is not dominated by substratum influence. 8.2 The Southern sub-area The situation in the Southern sub-area is quite different. Transcaucasia and northern Iran constitute a multilingual convergence area whose languages share typological traits at all levels. Azeri displays numerous deviations from the inherited type of Turkic, much more than the Turkic varieties of the Northern subarea. Transcaucasia was largely Turkicized during the Seljuk era. Oghuz Turks had begun to move into Azerbaijan in the eleventh century. Turkicization began in the north and continued into the south by the twelfth century (Golden, 1992: 225). At the time of the immigration, numerous mutually unintelligible varieties were spoken in the area, for example, various brands of Tatic, Caucasian Albanian, Kurdic, Armenian, Aramaic and possibly remnants of Parthian. According to Stilo (2004), these varieties had probably already formed an ‘Arax Sprachbund’, as we find here today. The linguistic diversity facilitated a shift to a Turkic lingua franca, which soon became the primary code of large parts of the population. There was probably a rapid shift to Turkic, a process which subsequently expanded into other Iranian-speaking areas. Many Caucasians had probably shifted to Iranian, Armenian and Aramaic before they shifted to Turkic; others may have shifted directly to Turkic. Many features from Iranian and its Caucasian substrata were imposed on the incoming Turkic varieties, which had already been subject to Iranian influence in Central Asia and Khorasan. Stilo points out that Persian was not a spoken language in the area at the time of Oghuz immigration. It has, however, served ever since as a highly influential adstratum language because of its cultural dominance. The number of Turkic newcomers who had moved into Azerbaijan, displacing the old ruling class and causing the code shift, seems to have been
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
176
Lars Johanson
177
9
Reasons for the linguistic impact
What was the reason for the strong linguistic impact of the small Turkicspeaking elite groups? The situation may be compared with that of other areas in which Turkic has played major roles. Many originally non-Turkic groups have abandoned their primary codes in favour of Turkic varieties. In Central Asia, sizeable Iranian-speaking groups shifted to Turkic. In the north of the Turkic world, Tungusic-speaking groups adopted Yakut (Pakendorf, 2001). Southern Siberia offers similar scenarios, with Southern Samoyed, Ob-Ugrian and Yenisei Ostyak (Ket) substrates. Old Bulgar groups went up the Volga River to the Volga-Kama confluence area and established a state there. Some of these Bulgars imposed their language on Finno-Ugric populations, which led to the emergence of Chuvash (Johanson, 2000b). On the other hand, Turkic-speaking groups have shifted to non-Turkic languages. One of the important Bulgar movements, caused by pressure from the Khazars, was the migration of one group to the Danube, the Byzantine frontier. These Balkan Bulgars very soon gave up their primary code in favour of a Southern Slavic variety, today’s Bulgarian. In old Eastern Europe, other Turkic groups of unknown size, Bulgars, Kumans, Pechenegs and others, adopted Slavic varieties. In recent times, Turkic groups have massively replaced their primary codes with Russian. Contacts leading to shift depend on social dominance relations. The size of the incoming groups relative to the local groups may be relevant, but political expansion is not always accompanied by mass population movements. The proportions may vary considerably. Elite minorities may impose their codes on comparatively large local indigenous populations. Shifts of this kind occur without major demographic changes.2 The introduction of Turkic into the Caucasus area has rather been the result of minor movements, but the influential roles of the Turkic varieties is obvious. One reason for the linguistic success of Turkic elite groups was their advanced political organization, which contributed significantly to their dominance. As mentioned, they were prestige languages and served as lingua francas in bilingual intercourse. The situation of the Turkic-speaking newcomers in the Caucasian area differed essentially from that of the Turkic groups on the Balkans. Both Caucasian sub-areas were characterized by an extreme diverseness of mutually unintelligible codes which mainly exhibited rather complex structures.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
small. The shift was not based on a massive migration, and thus took place with little genetic impact. Today’s Turkic-speaking Azeris are related most closely to their nearest non-Turkic neighbours, that is, genetically closer to Armenians and some Caucasian-speaking groups than to other Turkic-speaking populations (Nasidze and Stoneking, 2001). This closeness is certainly caused by shared Caucasian substrata.
178
The Roles of Turkic in the Caucasus Area
Prestige languages with structural properties that made them relatively easy to learn (Johanson, 1992: 199–206; 2002: 41–8) could be used to bridge the gaps between the local speech communities.
Balkan–Pontic–Caucasus–Caspian contacts
What must be kept in mind is that Oghuz Turkic is not the only source of Turkic impact on the non-Turkic varieties, and of Turkic contributions to common areal tendencies. The highly probable influences from various older Turkic tribes that invaded the area should not be ignored: Bulgar elements from the end of the fifth century onwards, Western Türk elements from the sixth century onwards, Khazar elements from the seventh century onwards, and Kipchak Turkic elements of the earlier and the later type from the tenth century onwards. It is not possible to establish, as Gadžieva (1979) has attempted, one single type of early Turkic for the Caucasus area, but it is necessary to deal with the linguistic situations within a more general framework of Balkan–Pontic–Caucasus–Caspian language contacts.3
Acknowledgement Work on this chapter, which is a contribution to the Special Research Project (SFB) 295 of the Deutsche Forschungsgemeinschaft, was carried out while the author was Guest Scientist at the Max Planck Institute for Evolutionary Anthropology, Leipzig. The chapter is based partly on a talk given at a meeting of the Genetics-Linguistics Group (‘geneling’) on 2 April 2002. I profited very much from the paper ‘Languages and Prehistory in the Caucasus’ given by Johanna Nichols on 18 April 2002. I am particularly grateful for the fruitful discussions I had with Helma van den Berg, the ardent investigator of Daghestanian languages, whose life ended only one year later, on 11 November 2003 in Derbend, Daghestan.
Notes 1 As is well known, counting by twenties is also found in some West European languages. Thus Old French quatre vins dis (modern French quatre-vingt-dix) for ‘ninety’ is modelled on Basque laurogeitahamar (< laur-hogei-eta-hamar). 2 A further question is whether the incoming groups have been predominantly male, mating with indigenous females who have shifted their code. New methods of population genetics will finally allow answers to questions of these kinds. 3 It is not possible to deal with linguistic similarities between the Balkan and Caucasus areas in this chapter. It is, however, interesting to note that, according to Nedjalkov (2002), Karachay-Balkar is the only Turkic language in which verbs marked with the reciprocal suffix -(V)š have a competitive meaning ‘to do something competing with each other’. This meaning happens to be similar to that of the Bulgarian reflexive clitic.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
10
Lars Johanson
179
Abaev, V. I. (1933) ‘Obšfie elementy v jazyke osetin, balkarcev i karafaevcev’, Jazyk i myšlenie, vol. 1, pp. 71–89. Aikhenvald, A. Y. (2004) Evidentiality (Oxford: Oxford University Press). Aikhenvald, A. Y. and R. M. W. Dixon (eds) (2003) Studies in Evidentiality, Typological Studies in Language, 54 (Amsterdam: John Benjamins). Akbaev, C. X. (1963) Fonetika Dialektov Karafaevo-Balkarskogo Jazyka (Cerkessk: Karafaevo-Cerkesskoe Knižnoe Izdatel’stvo). Alekseev, M. E. and E. M. Wejxov (1997) Lezginskij jazyk (Moscow: Academia). Appaev, A. M. (1960) Dialekty Balkarskogo Jazyka v ix Otnošenii k Balkarskomu Literaturnomu Jazyku (Nal’fik: Kabardino-Balkarskij Naufno-Issledovatel’skij Institut). Baskakov, N. A., P. A. Azimov, G. I. Donidze, S. K. Kenešbaev, L. A. Pokrovskaja, V. V. Rešetov, E. V. Sevortjan, E. R. Tenišev, E. I. Ubrjatova, M. J. Xamzaev, W. W. Wiraliev, K. K. Judaxin (1966) Jazyki narodov SSSR 2 (Moscow: Nauka). Benzing, J. (1959) ‘Das Kumükische’, in Deny et al., Philologiae, pp. 391–406. Berta, Á. (1998) ‘West Kipchak Languages’, in L. Johanson and É. Á. Csató (eds), The Turkic Languages (London and New York: Routledge), pp. 301–17. Boeder, W. (2000) ‘Evidentiality in Georgian’, in Johanson and Utas (eds), Evidentials, pp. 275–328. Cafer2lu, A. and G. Doerfer (1959) ‘Das Aserbeidschanische’, in Deny et al. (eds), Philologiae, pp. 280–307. Chrikba, V. (2003) ‘Evidential Category and Evidential Strategy in Abkhaz’, in Aikhenvald and Dixon, Studies in Evidentiality, pp. 243–72. Christophe, B. (2004) Studier i de sydkaukasiske sprogs aspektologi, Acta humaniora, 190 (Oslo: University of Oslo, Faculty of Arts). Christophe, B. (2005) ‘Some Remarks on Postterminality and Evidentiality in South Caucasian’, in Dag Haug and Eirik Welo (eds), Haptacahaptaitis: Festschrift for Fridrik Thordarson (Oslo: The Institute for Comparative Research in Human Culture). Comrie, B. (1981) The Languages of the Soviet Union (Cambridge: Cambridge University Press). Deny, J., K. Grønbech, H. Scheel, Z. V. Togan and P. N. Boratav (eds) (1959) Philologiae turcicae fundamenta, vol. 1 (Wiesbaden: Steiner). Dmitriev, N. K. (1940) Grammatika kumykskogo jazyka (Moscow/Leningrad: Akademija nauk SSSR). Dmitriev, N. K. (1962) Stroj tjurkskix jazykov (Moscow: Izdatel’stvo vostofnoj literatury). Donabédian, A. (1996) ‘Pour une interprétation des différentes valeurs du médiatif en arménien occidental’, in Z. Guentchéva (ed.), L’énonciation médiatisée, Bibliothèque de l’infomation grammaticale, 35 (Louvain and Paris: Peeters), pp. 87–108. Džidalaev, N. N. (1990) Tjurkizmy v dagestanskix jazykov (Moscow: Nauka). Friedman, V. A. (2000) ‘Confirmative/non-confirmative in Balkan Slavic, Balkan Romance, and Albanian with Additional Observations on Turkish, Romani, Georgian and Lak’, in Johanson and Utas, Evidentials, pp. 329–66. Gadžieva, N. Z. (1979) Tjurkojazyfnye arealy Kavkaza (Moscow: Nauka). Gecadze, I. O. (1977) K lingvo-geografifeskomu obzory kavkazskogo jazykovogo areala. Areal’nye issledovanija v jazykoznanii i etnografii (Leningrad: Nauka). Geiger, B., T. Halasi-Kun, A. H. Kuipers, and K. H. Menges (1959) Peoples and Languages of the Caucasus. A Synopsis, Janua Linguarum, 6 (’s Gravenhage: Mouton and Co).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
References
The Roles of Turkic in the Caucasus Area
Golden, P. B. (1992) An Introduction to the History of the Turkic Peoples. Ethnogenesis and State-formation in Medieval and Early Modern Eurasia and the Middle East, Turcologica, 9 (Wiesbaden: Harrassowitz). Golden, P. B. (1994) ‘Georgio-Turkica’, Acta Orientalia Academiae Scientiarum Hungaricae, vol. 13, pp. 101–16. Golden, P. B. (1999) ‘The Nomadic Linguistic Impact on Pre-Cinggizid Rus’ and Georgia’, Archivum Eurasiae Medii Aevi, vol. 10, pp. 72–97. Grjunberg, A. (1962) ‘K voprosu o jazykovom vzaimodejstvii (na materiale jazyka severoazerbajdžanskix tatov)’, Kratkie soobšfenija Instituta Narodov Azii, vol. 40, pp. 11–23. Grjunberg, A. (1963) Jazyk severoazerbajžanskix tatov (Leningrad: Izdatel’stvo Akademii nauk SSSR). Gukasjan, V. L. (1973) Vzaimootnošenie Azerbajdžanskogo i Udinskogo Jazykov (Baku: Institut Jazykoznanija Akademii Nauk Azerbajdžanskoj SSR). Haspelmath, M. (1993) A Grammar of Lezgian, Mouton Grammar Library, 9 (Berlin/ New York: Mouton de Gruyter). Harris, A. C. (1985) Diachronic Syntax: The Kartvelian Case, Syntax and Semantics, 18 (Orlando, Fla.: Academic Press). Johanson, L. (1971) Aspekt im Türkischen. Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems, Studia Turcica Upsaliensia, 1 (Uppsala: Almqvist & Wiksell). Johanson, L. (1992) Strukturelle Faktoren in türkischen Sprachkontakten, Sitzungsberichte der Wissenschaftlichen Gesellschaft an der J. W. Goethe-Universität Frankfurt am Main, 29 (Stuttgart: Steiner). Johanson, L. (1996) ‘On Bulgarian and Turkic indirectives’, in N. Boretzky, W. Enninger and Th. Stolz (eds), Areale, Kontakte, Dialekte, Sprache und ihre Dynamik in mehrsprachigen Situationen, Bochum-Essener Beiträge zur Sprachwandelforschung, 24 (Bochum: Brockmeyer). Johanson, L. (2000a) ‘Turkic indirectives’, in Johanson and Utas, Evidentials, pp. 61–87. Johanson, L. (2000b) ‘Linguistic convergence in the Volga area’, in D. Gilbers, J. Nerbonne and J. Schaeken (eds), Languages in Contact, Studies in Slavic and General Linguistics, 28 (Amsterdam and Atlanta: Rodopi), pp. 165–78. Johanson, L. (2000c) ‘Viewpoint operators in European languages’, in Ö. Dahl (ed.), Tense and aspect in the languages of Europe (Berlin/New York: Mouton de Gruyter), pp. 27–187. Johanson, L. (2002) Structural Factors in Turkic Language Contacts (London: Curzon). Johanson, L. (2003) ‘Evidentiality in Turkic’, in Aikhenvald and Dixon, Studies in Evidentiality, pp. 273–91. Johanson, L. (2004) ‘On the Turkic origin of Hungarian igen “yes” ’, Acta Orientalia Academiae Scientiarum Hungaricae, vol. 57, pp. 93–104. Johanson, L. (2005) ‘Participles in Caucasus Turkic’, in Christophe, Festschrift Fridrik Thordarson. Johanson, L. and B. Utas (eds) (2000) Evidentials. Turkic, Iranian and Neighbouring Languages (Berlin/New York: Mouton de Gruyter). Karakoç, B. (2005) Das finite Verbalsystem im Nogaischen, Turcologica, 58 (Wiesbaden: Harrassowitz). Kerimov, I. A. (1967) Oferki kumykskoj dialektologii (Maxafkala: Dagufpedgiz). Kozintseva, N. (2000) ‘Perfect forms as a means of expressing evidentiality in Modern Eastern Armenian’, in Johanson and Utas, Evidentials, pp. 411–17. Magomedov, A. G. (1966) ‘Kumykskij jazyk’, in Baskakov et al., Jazyki, pp. 194–212.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
180
181
Menges, K. H. (1959) ‘Die aralo-kaspische Gruppe’, in Deny et al., Philologiae, pp. 434–88. Menges, K. H. (1968) The Turkic Languages and Peoples. An Introduction to Turkic Studies, Ural-Altaische Bibliothek, 15 (Wiesbaden: Harrassowitz). Nasidze, I. and M. Stoneking (2001) ‘Mitochondrial DNA Variation and Language Replacements in the Caucasus’, Proceedings of the Royal Society, B 268 (London), pp. 1197–206. Nedjalkov, V. P. (2002) ‘Karachay–Balkar reciprocals’, Turkic Languages, vol. 6, pp. 19–80. Pakendorf, B. (2001) ‘Genetic and linguistic perspectives on the prehistory of the Yakuts’, Turkic Languages, vol. 5, pp. 138–43. Pritsak, O. (1959) Das Karatschaische und Balkarische’, in Deny et al., Philologiae, pp. 340–68. Wagirov, A. K. (1989) Zaimstvovannaja leksika abxazo-adygskix jazykov (Moscow: Nauka). Siemeniec-Gopas, E. (2000) Karachay–Balkar Vocabulary of Proto-Turkic Origin, Studia Turcologica Cracoviensia, 7 (Kraków: Ksiegarna Akademicka). Stilo, D. (1994) ‘Phonological Systems in Contact in Iran and Transcaucasia’, in M. Marashi (ed.), Persian Studies in North America. Studies in Honor of Mohammad Ali Jazayery (Bethesda, Md: Iranbooks), pp. 75–94. Stilo, D. (2004) ‘Some Preliminary Thoughts on the Non-Persian Iranian Substratum of Azerbaijan’. Paper read at the conference ‘The Evolution of Turkic in Iran’, Johannes Gutenberg University Mainz, 17–19 December. Stilo, D. (2005) ‘Ambipositions as an Areal Response: The Case-study of the Iranian Zone’, in L. Johanson and C. Bulut (eds) Turkic-Iranian contact areas. Historical and linguistic aspects (Wiesbaden: Harrassowitz), pp. 297–320. Sumbatova, N. (1999) ‘Evidentiality, Transitivity and Split Ergativity’, in W. Abraham and L. Kulikov (eds), Tense–Aspect, Transitivity and Causativity (Amsterdam: John Benjamins), pp. 63–95. Wixmann, R. (1980) Language Aspects of Ethnic Patterns and processes in the North Caucasus. (Chicago: University of Chicago). Xabifev, M. A. (1966) Karafaevo-balkarskij jazyk’, in N. A. Baskakov et al., Jazyki, pp. 213–33.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Lars Johanson
The Circle That Won’t Come Full: Two Potential Isoglosses in the Circum-Baltic Area Maria Koptjevskaja-Tamm
1 The Circum-Baltic languages – a coastal contact-superposition zone in the European periphery 1.1 Introducing the Circum-Baltic languages Although humans have inhabited the region around the Baltic Sea at least since the end of the last glacial era, our knowledge about the languages spoken in the area covers a much shorter time span. In historical times, this area was mainly a meeting-place of languages from two linguistic stocks: Indo-European (Baltic, Germanic and Slavic languages) and Uralic/Finno-Ugric (Finnic and Saami). Archaeologists, geneticists and linguists claim to trace back the two language stocks in the area to at least the second millennium BC, and suggest various competing theories on which one was the first and where. In addition, there are three ‘exotic’ languages that have all been used in the area for a considerable time: the Indo-Aryan language(s) Romani, spoken all over the Circum-Baltic area in different varieties, and the Turkic languages Tatar and Karaim. Which languages should count as Circum-Baltic (CB) languages (the term launched in Dahl and Koptjevskaja-Tamm, 1992, and further developed in Dahl and Koptjevskaja-Tamm, 2001) is, of course, open to discussion, for several reasons, the main one being the geographical delineation of the area. In Dahl and Koptjevskaja-Tamm (2001: xvi–xx), we include as CB languages those in Figure 8.1 (see also Map 8.1). The list in Figure 8.1 is simplified in several respects. First, it does not contain extinct languages – for example, Polabian (Slavic), Old Prussian, Jatvingian, Curonian and Galindian (Baltic). In addition, it more or less ignores dialectal variations, which in some cases make the distinction between languages and dialects particularly troublesome. Thus, Northern and Southern Estonian are sometimes considered to be two different 182 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
8
Maria Koptjevskaja-Tamm
183
Indo-European Germanic West: High German, Low German, Yiddish North: Danish, Swedish, Dalecarlian, Norwegian Baltic: Latvian, Lithuanian Slavic West: Polish, Kashubian East: Belarusian, Russian, Ukrainian Indo-Aryan: Romani Finno-Ugric Finnic: Estonian, Finnish, Ingrian, Karelian, Ludian, Olonetsian, Veps Saami: Southern S., Ume S., Pite S., Lule S., Northern S., Inari S., Skolt S. Turkic: Karaim, Tatar Figure 8.1 Circum-Baltic languages
Northern Saami
Inari Saami
Skolt Saami
Lule Saami Pite Saami Ume Saami
Karelian
Southern Saami
Finnish Olonetsian
Dalecarlian
Norwegian
Swedish Estonian Livonian
Latvian Danish Northern Frisian Low German
Ludian Veps
Ingrian Votian
Russian
Lithuanian Kashubian
Karaim
Belarusian Polish High German
Map 8.1 The Circum-Baltic languages [non-territorial languages (Romani, Yiddish, Tatar) not shown]
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
languages (Estonian language, eesti keel versus Võru language, võro kiil); the High Latvian dialect spoken in the eastern part of Latvia is sometimes referred to as the ‘Latgalian language’; and Norwegian has two written languages, Bokmål (based on Norwegianized Danish) and Nynorsk (based on Norwegian dialects). The list above, however, treats the different Saami
The Circum-Baltic Area
varieties as distinct languages. ‘Dalecarlian’ (egentligt dalmål) refers to the highly conservative Scandinavian vernaculars that are spoken in the Swedish province of Dalarna (Dalecarlia) and are not comprehensible to speakers of Standard Swedish. We have chosen to treat these as a separate language (or, perhaps, even as a language chain), breaking with the tradition of counting them as the ‘highly deviating variants’ of the Eastern Swedish dialects. Both extinct varieties and dialectal variation are, of course, crucial for the study of areal phenomena. The ethnic groups and the languages in the CB area have been involved in various kinds of contact, from more local to those stretching over large territories. Since time immemorial, the area itself has been divided and re-divided constantly among different spheres of influence. Thus, the period AD 800–1000 meant expansive activities of the Scandinavian Vikings and the emergence of the Scandinavian, Polish and Russian states, each with its own sphere of dominance. The period AD 1100–1500 saw Denmark’s expansion, the crusades and the establishment of the Teutonic Order states in Northern Baltikum, dominance of the Hanseatic leagues, and expansion of the Polish and the Lithuanian states, later of the PolishLithuanian Commonwealth. After that the area went on to be shared and re-shared among powers such as Sweden, the Polish–Lithuanian Commonwealth, Prussia (later Germany) and Russia (later the Soviet Union). Each of the dominant powers brought with it a new prestige language (Danish, Low German, the Eastern Slavic variety used in the Grand Duchy of Lithuania, Swedish, Polish, German and Russian), which expanded over a large area and influenced the local languages. Indirectly, the spread of prestige languages entailed convergence among the local vernaculars themselves. The CB zone has also been divided and re-divided among the three main religions in the area – Catholicism, Lutheranism and Greek Orthodoxy. The ‘deviating’ religions, Judaism and the ‘Old Faith’, have played a highly important role in the preservation of Karaim and Yiddish, as well as Russian ‘Old Believers’ dialects in the CB area. More or less local contacts among particular ethnic groups and languages or language groups in the area are numerous and diverse. One contact-intensive zone is, for example, the territory where Latvian, Livonian and Estonian are spoken (Stolz, 1991). Thus, a strong Finnish substratum is generally recognized in Latvian, especially in its northeastern dialects (Tamian), covering the area originally inhabited by Livonians. Livonian itself is at present spoken by only a few dozen speakers and is largely influenced by Latvian. Northern Russian has a number of features generally attributed to the Finnic substratum and, primarily, to contacts with the smaller Finnic languages (Ingrian, Karelian, Ludian, Veps and Votian). These languages themselves are now on the verge of extinction or, in the case of Karelian, are mainly used as one part
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
184
185
of bilingual mixed codes (Sarhimaa, 1999). Swedish in Finland and Estonia is (or was) spoken in the Finnic area; there are various Slavic varieties spoken in the Baltic area. Given all this, it is justified to expect an abundance of contact phenomena in the CB languages. There is a relatively old tradition of studying linguistic contacts around the Baltic Sea by looking at loan words. The area was also among the first to receive the newly-coined label ‘Sprachbund’, applied to it by R. Jakobson (1931). Starting primarily from the 1970s, there has been an intensive hunt for isoglosses in the CB area, with the resulting rich flora of partly overlapping proposed Sprachbünde, primarily in the two above-mentioned main hotbeds of areal phenomena (Koptjevskaja-Tamm and Wälchli, 2001). Thus, the Latvian–Livonian–Estonian zone stretching in different directions over the Baltikum and further constitutes the core in what has been termed the ‘Peipus-Bund’ (Décsy, 1973), the ‘Baltischer Sprachbund’ (Haarmann, 1970, 1976) and the ‘Convergence zone in the Baltikum’ (Stolz, 1991; see also Falkenhahn, 1963). The Eastern Finnic– Northwestern Russian(–Baltic) zone forms the core in Matthiassen’s (1985a, 1985b) ‘Eastern Baltic Sprachbund’ and in Sarhimaa’s (1999) ‘Karelian Sprachbund’. In addition, the CB area partially overlaps with two other suggested convergence zones including Scandinavian–Celtic–Northern Finnic– Saami – the ‘Wikinger-Bund’ (Haarmann, 1976), and Polish–Kashubian– Belarusan–Ukrainian–Lithuanian – the ‘Rokytno-Bund’ (Haarmann, 1976), or the ‘Baltic–Slavic contact area’ (Wiemer, 2004; see also Falkenhahn, 1963). The latest, and probably most ambitious, contribution to the field are the two volumes by Dahl and Koptjevskaja-Tamm (2001) that contain numerous papers on relations among the Circum-Baltic languages. In the concluding chapter in their book, Koptjevskaja-Tamm and Wälchli look at a large number of potential areal features in the CB area, at both the micro-level – by giving a nuanced and detailed analysis of these phenomena – and at the macro-level – by plotting the same phenomena against a general cross-linguistic background. In the latter, we differentiate between global and European perspectives. Several important conclusions follow from this analysis: (1) the highest concentration of typologically unusual areal properties is found in the eastern part of the CB; (2) the CB region forms a border zone between the Central Eurasian languages in the East and the Standard Average European languages in the West; and (3) the isoglosses pick up different subsets of the languages, in many cases also extending outside the CB area proper. In the next section we shall look at some examples illustrating these generalizations (see also Table 8.1).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
186 Table 8.1 Some areal phenomena in the CB area Languages primarily involved
Possible source(s)
In embryo attested in IndoEuropean; common innovation in the CB area with several layers of influence (Baltic → Finnic → Russian) Various hypotheses; in Baltic and Northern Russian; probably a combination of inherited Indo-European models reinforced by contacts with Finnic Various hypotheses
1.
Case alternation for marking total versus partial objects/ subjects
Finnic, Baltic, Eastern and Western Slavic
2.
Nominative object in various constructions (imperatives, infinitives dependent on impersonal verbs, etc.)
Finnic, Baltic, Northern Russian
3.
Case alternation in predicate adjectives and nominals (Stassen, 2001)
Finnic, Saami, Mordvin, Komi, Baltic, Eastern Slavic, Polish
4.
Alternation between casegovernment and agreement within numeral constructions
Baltic, Slavic, Finnic, Sami
Typological status Globally
Expansion of the Indo-European model in Baltic and Slavic Indo-European (most probably Baltic) influence on Finnic and Sami
Europe
Probably unusual, but not unique (cf. Basque, French)
Probably unusual, but not unique (cf. Nenets, Kamassian, Southern Paiute, Yindjibarndi)
Fairly infrequent, but far from unique. Occurs mainly at the fringe of Indo-European and is most probably a non-IndoEuropean characteristic Very unusual, probably not unique (cf. Arabic)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Phenomenon
A shared innovation, probably starting from Finnic. However, also some Indo-European preconditions
5.
Evidential mood
Baltic, Southern Finnic (Estonian, Livonian)
6.
Reflexive postfixes as markers of valence recession (Haspelmath, 1987)
Northern Various hypotheses Germanic, Baltic, Eastern Slavic
An areal property of several wellestablished areas worldwide. ‘Fusing’ evidentiality with tense in the same marker is crosslinguistically rare Fairly unusual
187
Only in the CB languages
1.2 The main isoglosses in the CB area As mentioned earlier, the most striking isoglosses cross-linguistically in the CB area are mainly found in its eastern part – that is, in the Baltic and Finnic languages and in Northern Russian, and include the following (for details see Koptjevskaja-Tamm and Wälchli, 2001): (1) The alternation between the accusative and the genitive (in Baltic and Slavic) or the partitive (Finnic) case1 between high transitivity and low transitivity objects, or ‘total’ versus ‘partial’ objects. The details and the relevance of this distinction is subject to a considerable cross-linguistic variation, where Finnic and Standard Russian show the most versus the least grammaticalized system, with some of the north-western Russian dialects sharing considerable properties with Finnic. The factors underlying this distinction include polarity of the clause (affirmative versus negative), as in example (1) below from Polish, aspect, affectedness of the object and so on. (1)
Polish (a) Przya-p-em tvoj-a accept-PAST-1SG your(SG)-ACC.SG.F ‘I accepted your proposal.’ (b)
Nie
przya-p-em tvoj-ej accept-PAST-1SG your(SG)-GEN.SG.F ‘I did not accept your proposal.’ NEG
propozycj-e proposal-ACC
propozycj-i proposal-GEN
The alternation between accusative-marked ‘total’ and genitive-marked ‘partial’ objects is well-known from some of the older Indo-European
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
languages, but not on a scale even slightly comparable with the modern situation in Baltic, Slavic and Finnic. From a typological point of view, even though differentiation of object marking depending on such factors as polarity, aspect and affectedness of the object are quite spread (Hopper and Thompson, 1980; Næss, 2003), the implementation of these factors in the systems found in the Baltic area seems to be fairly rare, to say the least. (2) Less canonical subjects – for example, subjects in existential clauses, can sometimes be marked with the genitive (in Baltic and Slavic) or the partitive (Finnic) case. Again, the most grammaticalized system is found in Finnic, where the partitive case consistently marks subjects in existential clauses, as in example (2) below from Finnish, and in semantically related clause types. The partitive case is particularly preferred if the clause is negated and the subject refers to a quantatively non-delimited entity. Roughly the same conditions govern the choice between the nominative and the genitive case marking on Baltic and Slavic existential subjects, but on a significantly more restricted scale, with the most grammaticalized distinction again being found in Northwestern Russian. (2)
Finnish (a) Kirj-at o-vat book-NOM.PL be.PRES-3PL ‘The books are on the table.’ (b)
pöydä-lla table-ILLAT
Pöydä-llä o-n kirj-oja table-ILLAT be.PRES-3SG book-PART.PL ‘There are (some) books on the table.’
Typologically it is even more difficult to find counterparts to this kind of alternation. It is also interesting that the case alternations cover both objects and ‘less subject-like’ subjects. One obvious alternative to account for these rules could be the unaccusative hypothesis, according to which intransitive verbs are split among those that take ‘good’ subjects and those that in fact take objects (under certain conditions ‘disguised’ as subjects). However, there are reasons to consider unaccusativity as being not particularly suitable for the situation under consideration. The best parallels to the Finnic–Baltic–Russian case alternations under examples (1) and (2) are provided by the alternation between the absolute and the partitive cases (or ‘zerik-case’) in Basque (for examples and discussion, see Koptjevskaja-Tamm and Wälchli, 2001: 666) and by the rules governing the use of the partitive article in French. (3) In several constructions in Finnic, Baltic and Northern Russian (both Old Northern Russian and modern Northwestern dialects), the object appears in the nominative case and not in the accusative. One common context is provided by an infinitival clause functioning as the subject of a necessitive matrix predicate – see example (3) from Northwestern Russian dialects.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
188
189
The Latvian correspondence to such clauses involves the so-called ‘debitive mood’, also with the object in the nominative. The set of nominative-object constructions differs greatly across the languages. Thus, while nominative objects in Northern Russian and Lithuanian are normally restricted to clauses with clearly non-finite predicates (infinitives and converbs), Finnic requires, in addition, nominative objects in imperative clauses and clauses with impersonal passives. It has been suggested that the common denominator of all these contexts is their systematic lack of an overt personal subject (Timberlake, 1974). (3)
Northwestern Russian dialects (a)
Topim peck-u heat.PRES.1PL oven-ACC ‘We are heating the oven.’
(b)
Nado/Pora topit’ peck-a it-is-necessary/it-is-time heat.INF oven-NOM ‘It is necessary/It is time to heat the oven.’
The alternation between the nominative and the accusative cases for object marking is widely attested in Uralic, but is normally used for semantic reasons – that is, to distinguish between indefinite, mass or otherwise non-individuated objects, and others; this is also quite common cross-linguistically. Some Uralic languages (for example, the two Samoyedic languages, Nenets and Kamassian) have a syntactic rule by which objects to imperatives appear in the nominative case. A similar rule is attested elsewhere – for example, in the Pama-Nyungan Australian language Yindjibarndi. However, on the whole, the nominative-object constructions of the kind(s) found in Finnic, Baltic and Northern Russian seem to be very rare. (4) Finnic, Baltic and Slavic display double (or sometimes multiple) options in the case marking of predicate adjectives and nominals. The choice between the nominative and some oblique case, for example, the instrumental in example (4) below from Lithuanian, can be described as roughly correlating with the distinction between timestable and temporary situations, respectively (Stassen, 2001), but the rules and their implementation differ considerable across languages. (4)
Lithuanian (a) Jis yra mokytoj-as he.NOM is teacher-NOM ‘He is a teacher.’ (b)
Jis buvo mokytoj-u he.NOM was.3SG teacher-INSTR ‘He was (working as) a teacher.’
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
In Indo-European, the predicative instrumental is restricted to Baltic and Slavic; within the latter group its frequency is considerably higher in the northern languages (East-Slavic, Polish) than in the southern ones. Within Uralic, multiple encoding of non-verbal predicates is not restricted to Finnic, but is also found in Saami, Mordvin and Komi, which has been used as an argument for the Uralic origin of the EastSlavic and Polish pattern. Cross-linguistically, as shown in Stassen (2001), double encoding is fairly infrequent, but by no means unique: some variant of it occurs in at least four other areas. (5) In Finnic and most Slavic, most cardinal numerals higher than ‘one’ alternate between case-governing and agreeing with their complements under well-defined – and very similar – syntactic conditions. As example (5) below, from Russian, shows, when the numeral is in one of the direct cases (nominative or accusative), the complement appears in the genitive (or in the partitive in Finnic). Otherwise, both the numeral and the complement are in the same case (here in the dative). In Baltic, the two properties are associated primarily with different sets of numerals, which is more common both cross-linguistically and within Indo-European. (5)
Russian (a) Ja vi}u I.NOM see.PRES.1SG ‘I see five glasses.’ (b)
pjat’ five.NOM/ACC
Ja priš-l-a s I.NOM COME-PAST-F.SG WITH ‘I came with five glasses.’
stakan-ov glass-GEN.PL
pjat’-ju FIVE-DAT
stakan-ami GLASS-DAT.PL
Within Indo-European, the Slavic situation is almost unique. Numerals agreeing with complements and those governing them are widely attested across Indo-European, but these properties are normally associated with different sets of numerals – agreement with lower and government with higher numerals – and this principle is carried out quite consistently in Lithuanian, while Latvian shows a more complicated situation. Within Uralic, the basic numeral construction involves a non-inflected numeral preceding its nominal complement, which carries case inflection of the whole construction. Only Saami shows something remotely reminiscent of the Finnic pattern. Cross-linguistically, we do find both agreement of numerals with their complements, and government of complements by higher numerals in a number of languages, even though those are relatively infrequent phenomena. However, the complex Finnic and Slavic systems seem to lack any counterpart anywhere, and it is thus highly probable that the Finnic system is ‘borrowed’. Since there were no historical preconditions for such an extensive influence on Finnic from Slavic (at far as we know),
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
190
191
as opposed to that from Baltic, we can probably conclude that Baltic and Slavic had similar rules at an earlier stage, which were later simplified in Baltic. (6) Estonian, Livonian, Latvian and Lithuanian use special ‘evidential’ verb forms for marking that the speaker’s factual claims are based on indirect evidence, rather than on direct, or attested, evidence (evidential, quotative, relative or oblique mood). These are basically non-finite verb forms, primarily participles, sometimes ‘frozen’ in an oblique case: thus, -vat in example (6) below, from Estonian, is historically the partitive case form of the present participle. (6)
Estonian Sina rääki-vat saksa keel-t you(SG).NOM speak-QUOT German language-PART ‘You are said to speak German; they say you can speak German.’
Evidentiality is an areal property of several well-established areas worldwide and is known to be susceptible to diffusion through intensive language contact (Aikhenvald, 2004). Although many (or even most) languages express evidentiality distinctions in one or another way, they tend to do this by special verbal affixes/clitics, or by various modals. The situation reported above is quite unusual in that the expression of evidentiality is ‘fused together’ with the expression of tense: in de Haan (2005)’s global sample, such languages constitute about 5 per cent (for example, Turkish and Georgian).2 There are also interesting isoglosses not restricted to the eastern part of the Baltic area. An isogloss connecting Scandinavian, Baltic and East Slavic languages is the expression of certain verbal voice functions (reflexive, reciprocal, anti-causative, passive) by means of verbal postfixes – that is, affixes in the last position of a word, following, for example, tense/aspect and agreement or infinitive markers (Russian obnima-l-i-s’ ‘hug-PAST-PL-RECIPR versus obnima-t’-sja ‘hug-INF-RECIPR’, or Swedish kyss-te-s ‘kiss-PAST-RECIPR’ versus kyss-a-s ‘kiss-INF-RECIPR’). These affixes have all developed because of coalescence between the main verb and permutable reflexive and reciprocal pronouns (-s/-st in Scandinavian, -s in Baltic and -s’/-sja in East Slavic). They manifest an extensive polysemy, largely following the cross-linguistically well-attested grammaticalization paths in the development of reflexive markers to markers of middle voice and, later, possibly, to passives (Geniušien<, 1987; Haspelmath, 1987; Kemmer, 1993), although the sets of meanings differ considerably across the languages. As shown by Haspelmath (1987), postfixes as markers of reflexivity and related meanings are limited cross-linguistically mainly to the languages of the Baltic Sea region. Thus, in this respect, Baltic, East Slavic and Scandinavian manifest a transition from the cross-linguistically second most frequent pattern of many other
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
Indo-European languages, in which marking is accomplished by means of a reflexive pronoun/pronominal clitic, to the cross-linguistically most frequent pattern of the Finno-Ugric languages, which use suffixation for similar purposes. Lithuanian and some Latvian dialects are particularly interesting here, since the position of the reflexive element varies: it is placed between the prefix and the stem in prefixed verbs, otherwise wordfinally, for example, moky-ti-s ‘teach-INF-REFL’ (= ‘to learn’) versus ne-si-moly-ti ‘NEG-REFL-teach-INF’ (= ‘not to learn’). Whatever the proportion between possible contact-induced changes (Ureland, 1982) and the universal grammaticalization may be in this case, the use of postfixes for transitivityreducing verbal alternations is a peculiar CB phenomenon. As mentioned in section 1.1, the CB region forms a border zone between the Central Eurasian languages in the East and the Standard Average European (SAE) languages in the West. The following example will illustrate this point. Stolz (2001) shows that the European, or, primarily, SAE-languages show a cross-linguistically very high predilection to use the same marker for the expression of typical comitative (Peter is eating soup with his friend) and instrumental functions (Peter is eating soup with a spoon). Most languages in Stolz’ (2001) global sample separate the two constructions: in Russian, for example, the instrumental function is normally expressed by the instrumental case, whereas the comitative function requires in addition the preposition s ‘with’ (lo}k-oj ‘spoon-INSTR’ versus s drug-om ‘with friend-INSTR’). In this respect, the CB languages show interesting diversity: some languages use the same marker (Germanic, Latvian, Estonian, Saami and Livonian), and others strictly separate the two (Finnish, Russian, Polish), wheras Lithuanian is a mixture – while it normally separate the two functions, it can optionally use the same preposition for both. Stolz concludes that the complete merger of comitative and instrumental functions in Estonian, Livonian, Sami and Latvian is likely to be the result of the Germanic (superstratal) influence on these languages, even though the Latvian situation could in principle be explained for by internal factors. Some other examples include the expression of comparison, sentential possession, polarity questions (Koptjevskaja-Tamm and Wälchli, 2001). Western CB languages often behave like SAE languages, and eastern CB languages often follow the Central Eurasian pattern. There is, however, no bundle of isoglosses cutting the CB area neatly in two parts. Finally, as should be clear from the examples and from Table 8.1, the isoglosses pick up different subsets of the languages, in many cases also extending outside the CB area proper. The most general conclusion in Koptjevskaja-Tamm and Wälchli (2001) is that convergence in the CB area works primarily on a micro-level, reflecting language contacts of, maximally, two or three languages. One particularly beautiful example illustrating this point is discussed by Nau (1996). She considers verbal prefixes and particles that in various ways modify the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
192
193
meaning expressed by the verb, primarily making it telic (bounders), for example, ‘up’ in ‘eat up’. Prepositional verbs are often quoted as an areal phenomenon in the CB area (Campbell, 1996). Nau analyses the similarities among CB bounders at three different layers – material similarities; semantic, functional and lexical convergence, and morphosyntactic similarities – and demonstrates that languages may group themselves differently at different layers. It seems, therefore, that convergence comprising more than two or three languages in the CB area – and, most probably, in many other parts of the world – is always the result of the overlapping and superposition of different language contacts. This is particularly true for an area such as the CircumBaltic region that has never been economically, politically, culturally or linguistically united. Therefore, in Koptjevskaja-Tamm and Wälchli (2001) we have suggested the term ‘contact superposition zone’ as a better refection of the complex linguistic situation in the CB area (and, most probably, in many other places) than the vaguely defined and often misused term ‘Sprachbund’. Now, the important issue that has not yet been touched upon is whether there are any isoglosses at all covering all the CB languages. In the rest of the chapter, I shall focus on two properties that at first sight do seem to unite languages spoken in both the eastern and western parts of the CB area, and thus appear to be potential ‘Circum-Balticisms’. The first one, polytonicity, has been known for many years; the second one, the word order in possessive NPs, has been suggested as a potential CB feature in recent years, as a result of large-scale typological research on word order. As far as I know, these are the two most promising cases for being Circum-Balticisms, but, somewhat disappointingly, neither of the properties will stand the proof of coming full circle around the Baltic Sea. However, each of them bears witness to extremely interesting and to a certain extent underestimated linguistic contacts.
2
Polytonicity and initial stress
Polytonicity, i.e. the existence of tonal suprasegmental oppositions in a language, was the original impetus for talking about a Sprachbund in the circum-Baltic area. Jakobson (1931 (1971) (a), (b)) suggested that several languages spoken around the Baltic sea together built a ‘phonetic’ Sprachbund. These included Norwegian (except for an area in the west), most Danish dialects, Swedish (apart from most of the dialects in Finland and in the neighbouring areas and in Estonia), some Low German dialects, Northern Kashubian, Lithuanian, Latvian, Livonian and Estonian. Jakobson’s ideas were further elaborated by Lehiste (1983, 1988, 1997) and by Wiik (1995, 1997) who uses the term ‘The Baltic Sea Prosodic Area’.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
194
The Circum-Baltic Area
(1) a clearly archaic opposition of tone contours in long (bimoraic) syllable cores – found in the Baltic languages (see section 2.1); (2) word tones whose relatively recent phonologization goes back to the distinction between former mono- and bisyllabic words, or to that between words with one or two stresses – found in most Swedish, Norwegian and Danish dialects (see section 2.2); and (3) ‘overlength’, a relatively recent compensatory lengthening of erstwhile long initial syllables through loss or reduction of non-initial ones – found in Estonian, Livonian and Low Latvian (see section 2.3). The relevance of Northern Kashubian and Low German for the CircumBaltic prosodic features is disputable and will be discussed in section 2.4 below. 2.1 Baltic languages Baltic polytonicity is part of a wider phenomenon once covering large parts of the Indo-European dialect area. It is related genetically to polytonicity in Slavic languages (which still exists in certain varities of Slovene and Serbo-Croat) and in Classical Greek. This is one of the best-studied areas in Baltic linguistics (for a recent overview, see Dogil, 1999; also Balode and Holvoet 2001a, 2001b). In the Baltic languages tone oppositions apply to bimoraic syllable cores – long vowels, diphthongs and diphthongoid sequences of short vowels and tautosyllabic sonorants. The Baltic varieties differ, however, in how many tones they have, how these are realized phonetically, and in which contexts they apply. Lithuanian has two tones (called acute and circumflex, see Table 8.1) in stressed syllables, both in the Standard variety and in the two main dialect groups – \emaitian, spoken in the north-western part of Lithuania, and Aukštaitian. Acoustically, however, the tones in the \emaitian dialects differ significantly from those in the Aukštaitian dialects and in the Standard language, with the acute tone being replaced by a broken tone (with a glottal closure inside the syllable core), reminiscent of the third Latvian tone (see below). Since Lithuanian has a mobile and free word accent, tone distinctions are not restricted to the same place in a word. Tone differences are most pronounced in the case of diphthongs and diphthongoid sequences, and there is also a general tendency in Standard Lithuanian and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Tonal phenomena in the CB languages are of the ‘word accent’, ‘lexical accent’ or ‘pitch accent’ type, as opposed to word tones in such languages as Mandarin. That is, here the choice between the accents is made only once in each word, whereas in tone languages (almost) every syllable has its own tone. However, there are (at least) three different groups of word-accent phenomena in the CB region:
195
Aukštaitian to restrain tonal oppositions to only these contexts. Interestingly, pitch is not the primary correlate of the modern Lithuanian accent (for details, see Dogil, 1999: 886). The Latvian dialects are traditionally classified into three groups: the most conservative Central dialects, Tamian (spoken in the north-western part of Latvia), and High Latvian (Latgalian, spoken in the eastern part of Latvia); the Central and Tamian dialects are sometimes referred to as ‘Low Latvian’. The traditional Latvian system, retained in some Central dialects, has three tones – see Table 8.2. The third tone, called ‘broken tone’ (Stoßton), has resulted from a retraction of stress to an originally acute syllable – and is thus ultimately connected to the stress fixation on the initial syllable in Latvian (which, in turn, has most probably arisen as a result of contact with Finnic – see section 2.5 below). The broken tone involves a glottal closure as a part of its realization, and has parallels in one of the \emaitian (Lithuanian) tones, in Livonian and in Danish (see sections 2.2, 2.3 and 2.6). The tonal opposition in Latvian is most pronounced in the initial (stressed) syllable, but even in unstressed long syllables there is sometimes an opposition between long vowels and diphtongs with and without glottal closure (see Table 8.2). Most Latvian dialects have reduced the original three-tone
Table 8.2 Word accents in most Baltic varieties Lithuanian: Standard Lithuanian and Aukštaitian laukas Circumflex ‘field’ (rising or even, drawn) tone Acute d< ti ‘put: (falling) tone INF’
Broken tone/ glottal closure within the syllable
–
\emaitian
Latvian: Standard Latvian and some Central dialects Stressed
lãuks ‘field’
dîet0 ‘put: INF’
Unstressed
Tamian and some Central dialects
labas[laba:s] luõks/ ‘good’ ‘onion (NOM/ leaf’ ACC.PL.FEM. DEF) luôks lùoks ‘bow, ‘bow, arch’ arch’ luôgs ‘window’ luôgs labas ‘window’ [laba/as] ‘good’ (LOC.PL.FEM. INDEF) luõks ‘onion leaf’
Latgalian (High Latvian)
lùks* (1) ‘onion leaf’ (2) ‘bow, arch’
lûgs ‘window’
Note: * Neutralization of the circumflex–acute opposition in Latgalian is realized as the falling accent.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
196
The Circum-Baltic Area
system to a two-tone opposition, differing by which two of the three tones have been merged together.
Scandinavian polytonicity is considered to be a relatively recent phenomenon found in most dialects of Norwegian, Swedish (apart from the dialects in contact with Finnish, Saami and Estonian), including Dalecarlian, and in Danish – but not in Icelandic or Faroese. Word accents par excellence (that is, tonal accents) are found across Norwegian and Swedish, while Danish has an opposition between syllables with and without a glottal closure, stød. It is widely agreed that the phonemic opposition itself arose in the span around 1000–1200 BC, after the syncope period (ending about 800), but that the pitch differences underlying it had been around for a considerable time. Earlier attempts to relate the Scandinavian word accents to the older Indo-European phenomena have been rejected by subsequent research. The literature on the Scandinavian word accents is extensive (see Riad 1998, 2003, 2005 for recent overview of the different subsystems, and of the various theories regarding their origins). Swedish, Dalecarlian and Norwegian have a tonal opposition in nonmonosyllabic words.4 The exact phonetic realization differs considerably across the different varieties, with the principal division between Accent 2 having one or two pitch peaks. The basic distributional rule normally makes appeal to the etymology of the word (before 1000–1200): • originally (that is, after syncope/apocope) monosyllabic words now have Accent 1; and • originally (that is, after syncope/apocope) bi- and polysyllabic words now have Accent 2. In modern languages, there are a number of polysyllabic words with Accent 1 that have arisen because of a later vowel epenthesis, contain a later suffixation of the definite article, or are later borrowings (see Table 8.3). The Danish correspondence to the Norwegian–Swedish–Dalecarlian distinction between Accents 1 and 2 is not tonal, but involves an opposition between
Table 8.3 Tone distribution in Swedish, Dalecarlian and Norwegian (Swedish examples)
Originally one syllable Originally two syllables
Now one syllable (article not counted)
Now two syllables (article not counted)
1and-en (< and hinn) ‘the mallard’
1segel (< segl) ‘sail’ 2ande-n ‘the spirit’
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
2.2 Mainland Scandinavian languages3
197
syllables with and without a glottal closure, stød respectively (lit. ‘thrust, push’; cf. Stoß). There have been several theories attempting to account for the rise of word accents (for overviews, see Oftedal, 1952; Riad, 1998; Lahiri et al., 1999). The latest – and, in my opinion, the most consistent – is elaborated by Riad (1998), who suggests that the rise of Accent 2 is a result of the resolution of a word internal stress clash. Frequent internal stress clash was, in turn, a result of syncope and other process in late Proto-Nordic, whereby stresses within a single word often occurred on adjacent syllables. Its resolution involved de-stressing the second syllable, but retaining the stress-related pitch contour. ‘Accent 2 becomes distinctive in late PN [Proto-Nordic] as a consequence of the phonological loss of secondary stress in many inflectional morphemes, whereby the tonal behaviour becomes partially divorced from stress’ (Riad, 1998). Most researchers agree that the Danish opposition between syllables with and without stød has replaced the older tonal opposition (see section 2.5). 2.3 Estonian and Livonian Estonian–Livonian polytonicity is, again, a different phenomenon, related to the reduction of non-initial syllables (and ultimately with the fixed initial stress) and a compensatory secondary lengthening of the initial syllable, or overlength. Most Finnic languages have a two-term phonemic quantity system, with both consonants and vowels being either short (Quantity 1) or long (Quantity 2). In Estonian and Livonian this system was knocked off balance some 500 years ago, because of a massive apocope and syncope – the deletion or shortening of non-initial syllables (apocope and syncope in Estonian have tentatively been put in the thirteenth century, and between the fourteenth and sixteenth centuries, respectively). Apocope in Estonian was compensated for by lengthening of the initial syllable, thus creating over-long syllables (Quantity 3) out of the erstwhile long ones (see Table 8.4). Quantity 3 is often (but, not always) accompanied by a characteristic falling contour, and there have been long and hot debates in the recent decades on the status of Quantity 3 (Lehiste, 1983; 1988; 2003; Lehiste and
Table 8.4 Quantity in Estonian Quantity 1
Quantity 2
Quantity 3
Consonants
lina ‘line: NOM.SG’ (cf. Fi liina)
Vowels
sada ‘hundred’
linna ‘town: GEN.SG’ (cf. Fi linnan [Hrt] ‘palace: GEN.SG’) saada ‘send: IMP’
linna ‘town: ILL.SG’ (cf. Fi linnaan ‘palace: GEN.SG’) saada ‘get: INF2’
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
Roos, 1997). The issue at stake is whether its tonal contour should count merely as a concomitant feature of the primarily quantitative distinctions, or as a distinctive tone of its own. Most linguists, however, seem to agree that Estonian might be developing tonal oppositions on the basis of an earlier quantity system. Thus the ongoing changes in the placement of secondary stress in words with Quantity 3 syllables, and in their assignment to inflectional paradigms show that ‘overlength’ no longer has a special quantitative status, compared to Quantity 2. Interestingly, experimental studies have shown that the falling fundamental frequency contour may contain a period of reduced intensity, similar to the Danish, Latvian and Livonian stød (Lehiste, 1988). Lehiste (1988, 2003) suggests that the characteristic pitch contour in Quantity 3 syllables might have arisen because of spontaneous tonogenesis – the overlong syllable retained both the approximate duration and the pitch contour of the original disyllabic sequence (step-down). On the other hand, she does not exclude areal influence completely. Livonian has an even more extended secondary lengthening, which also combines with the opposition between syllables containing a glottal closure and those without it. The interpretation of the Livonian situation differs considerably among different researchers. Finally, Low Latvian also shows compensatory secondary lengthening because of the reduction or loss of non-initial syllables, which combines with the original Baltic tone contour opposition in long syllable cores (for details on Livonian and Low Latvian, see Koptjevskaja-Tamm and Wälchli, 2001: 643–4). 2.4 Other languages The Germanic dialects spoken in the Cologne–Trier area (Western Germany, Eastern Netherlands, Belgium, Luxembourg) have a tonal distinction between Push-tone (Germ Stoßton, Schärfung, Dutch valtoon, stoottoon) and Drag-tone (Schleifton, sleeptoon). These varieties include the Central Franconian (Mosel-Franconian and Ripuarian) dialects of German and the Limburgian dialects of Dutch, and the phenomenon is known as ‘the Rhineland Accentuation (Rheinische Akzentuierung)’. The literature is extensive; for some of the latest overviews, both synchronic and diachronic, see Gussenhoven and Bruce (1999), Lahiri et al. (1999) and Gussenhoven (2000). The relevant German dialects are divided into two broad groups, Rule A dialects and Rule B dialects, which show opposite conditions on the distribution of the tones. The ‘marked’ tone developed in syllables with bimoraic cores that contained Middle High German non-high vowels. The Dutch Limburgian dialects appear to have undergone a development similar to that of the German Rule A dialects. There are also some additional (and different for the different dialects) requirements on schwa-apocope in the following syllable. Rhenish West Germanic varieties differ crucially from Scandinavian in
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
198
199
having tonal oppositions in monosyllables. In any case, they are spoken relatively far away from the Baltic coastline, in an area that is not adjacent to any of the Circum-Baltic word-accent languages. Jakobson’s understanding of Northern Kashubian (its northernmost dialect Slovincian, now defunct) as a language showing tonal oppositions is based on an old description by Lorentz (1903). It is, however, unclear whether the tones were distinctive or whether they simply correlated with vowel length – Stankiewicz (1993: 292) reinterprets Lorentz in favour of the latter option. This is far from being a unique situation. For example, the Limburgian dialect spoken in the Dutch town of Weert in the north-western periphery of the Rhineland Accentuation area has been described previously as having a tonal opposition. Heijmans (2003), however, shows that the only contrastive feature in alleged pairs is vowel length. Significantly, the distribution of the vowel length in Weert corresponds to a large extent to the distribution of the word accents in the neighbouring Baexem dialect, which are also accompanied by certain quantitative differences. The interesting question, asked but not yet answered in Heijmans (2003: 35) is whether ‘the Weert vowel quantity opposition could have arisen by reanalyzing a prior lexical tone contrast (a longer vowel duration has often been claimed to cooccur with Accent II) or by merely imitating these durational differences that come with the word tones’. 2.5 Polytonicity in the CB area and language contacts 2.5.1 The eastern semicircle Now, to what extent can polytonicity in the CB area count as a ‘real’ isogloss for the purposes of areal linguistics and a quest for Sprachbünde or linguistic areas? The question is complicated, since there are no obvious connections among the three groups of polytonicity phenomena found in the CB area and considered in this section. Let us first consider the relevant languages at the eastern coast of the Baltic Sea. There are numerous examples of mutual contact-induced phenomena among the two Finnic languages, Livonian and Estonian, on the one hand, and Latvian on the other, and the similarities in the prosodic systems of all the three languages do not come as a surprise. Thus it is reasonable to believe that the complicated suprasegmental system in Livonian, including the presence of glottal closure, has arisen, at least partly, because of intensive language contact with Latvian. Conversely, the presence of secondary lengthening in Low Latvian, superimposed over the original Baltic tonal system, can also be ascribed to intensive contacts with Livonian. All three languages have initial stress. Remember that in both Latvian and Estonian (as well as in Livonian), reduction in non-initial, unstressed syllables is responsible for the important developments within the suprasegmental systems – for the rise of the Broken tone in Latvian and for the rise of the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
200
The Circum-Baltic Area
(1) Thus it is widely accepted that the initial stress in Latvian is related to Finnic influence. However, partial stress reduction is attested even in some north-western dialects of Lithuanian, where the Finnic influence is doubtful. And certain facts within the Latvian system point to a gradual accent shift in the history of Latvian, which might easily be explained by language-internal factors. (2) Turning to the weakening of non-initial syllables in Latvian, this process was even stronger in Low Latvian – that is, in the area with the strongest Livonian and Estonian influence. Also within the Finnic, apocope is most pronounced in Livonian and, within Estonian, in the dialects of Saaremaa and Hiiumaa, all of which form a contiguous area with Low Latvian. However, apocope is known also in Veps and in southern dialects of Finnish, and might probably also be explained by language-internal factors. Here we have a textbook example of the classical dilemma in historical linguistics – that is, whether there are any reasons for resorting to external (language-contact) factors in order to explain cross-linguistic similarities based on linguistic changes that could be accounted for by languageinternal factors. And, as in quite a few other cases, it seems reasonable to assume that both types of factor could have contributed to the final result. In other words, an incipient, internally motivated linguistic change in a language may be reinforced by contacts with another language that either shows the ‘target’ characteristics of such a change or is moving in the same direction. The former case might have been relevant for the development of the initial stress in Latvian, and the latter for the weakening of non-initial syllables in Latvian, Estonian and Livonian. In the same way, even though spontaneous tonogenesis can explain the specific pitch contour of Quantity 3 syllable in Estonian and the gradual reinterpretation of its status as tonal rather than quantitative, it could have largely profited from the presence of other polytonic languages in its immediate vicinity – primarily, Latvian. Prosodic properties are per se quite contagious, even when they are not necessarily distinctive within a language. And when the people all around you are, in addition, used to distinguishing among different tones, you might gradually learn to do the same thing in your own language. In short, the polytonicity phenomena and, more generally, the suprasegmental systems in Lithuanian, Latvian, Livonian and Estonian, demonstrate important similarities, which to a high degree seem to be based on a complicated network of mutual areal influences.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Quantity 3 in Estonian. Both the initial stress and apocope, in turn, are also plausible candidates for contact phenomena in the region, but, again, not in a straightforward way.
Maria Koptjevskaja-Tamm
201
The Continental Scandinavian word accents, on the other hand, do not appear to be related to the polytonicity phenomena on the eastern coast of the Baltic Sea in an obvious way. They do, however, demonstrate a number of interesting contact-induced phenomena. Thus the lack of word accents in most Swedish dialects spoken in Estonia, Finland and in Överkalix in Norrbotten (the former Finnish territory in North-Eastern Sweden) is considered traditionally as their loss because of the language contacts with the Finnic-speaking people.5 A similar case is assumed for the city of Bergen in western Norway, which has the accent distinction, differs from the surrounding countryside, which lacks any accents. The local Norwegian variety spoken in Bergen was once heavily influenced by Low German because of the presence of a Hanseatic office. It is believed that the accent loss began as an innovation in the city of Bergen, under Low German influence, then spread to the countryside, but was later re-introduced in the city itself (Lorentz, 1995: 42). The history of the spread of word accents across Scandinavian is still very little understood. As Riad (2003: 91) formulates it, there are three famous questions involved in the diachrony of Scandinavian tone accents: (1) the origin of lexical tones; (2) the development of a phonological distinction; and (3) the relationship between the tonal accent system of Swedish and Norwegian dialects, on the one hand, and the stød system of Danish, on the other. Researchers coming from different traditions seem to agree on the existence of different pitch contours in Proto-Scandinavian (or Common Nordic) for quite a long period. These were phonologically predictable from the number of syllables in a word. The crucial point occurred when the association between the different pitch contours and the number of syllables became looser – that is, when the pitch contour normally associated with monosyllables could be used with ‘new’ bisyllabic words, created by vowel epenthesis and by suffigating definite articles to nouns. According to the traditional way of interpreting this change (as well as other pan-Scandinavian innovations in the period AD 600–1200), it ‘applied to a basically uniform language – ‘Proto-Nordic’ – and yield another, which, however, either immediately (partly via those changes themselves) or shortly afterwards split up in two dialects: ‘West Nordic’ (Norwegian, Icelandic) and ‘East Nordic’ (Danish, Swedish) (Dahl, 2001: 226). As Dahl argues, neither the assumption that the Scandinavians preserved a common language over many centuries nor – and even more importantly – that they also changed their language all at the same time and in the same fashion, is particularly plausible. He suggests instead a different scenario. What looks like a uniform language across Scandinavia
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
2.5.2 The western semicircle
202
The Circum-Baltic Area
Thus, the explanation of the uniform language that we find in the runic inscriptions from the eleventh century onwards, in written documents from the thirteenth century onwards and indirectly in later spoken and written Scandinavian dialects is that at least the ruling classes in the central parts of the Scandinavian countries were using a common language which had spread very recently. The final result of this spread was the obliteration of the dialects or languages spoken earlier in the peripheral parts of the area. Consequently, instead of increasing linguistic diversity (by giving rise to the split between West and East Nordic) as suggested by the traditional account, the outcome was a decrease in diversity, a unification of the languages spoken in Scandinavia. (Dahl, 2001: 227) Given the role of Denmark as the dominating power in Scandinavia during the centuries before the Viking Age (at least from the eighth century onwards), it is natural to look for the cradle of the prestigious Scandinavian variety there. Dahl talks about two important commercial and political centres in Scandinavia during the centuries before the Viking Age, both founded in the eighth century AD. These were Hedeby, close to the present-day city of Schleswig on the east coast of Jutland (nowadays Germany, but a traditional Danish-speaking territory), and Birka in the Mälar provinces (Uppland) in Central Sweden. Both were, most certainly, parts of the same network, but the Mälar provinces were most probably a peripheral part of the Danish sphere of influence. Certain linguistic and archaeological facts may suggest that a relatively early prestigious Scandinavian variety originated in Hedeby and spread relatively quickly to Birka. The Hedeby–Birka language then gradually spread over a large area in what are now Sweden and Denmark. Also, Norway had some urban settlements in the ninth century with strong connections to Hedeby, and Denmark ruled over at least some parts of it. The spread of the prestige language was undoubtedly connected to various modifications in its local varieties, and to a development of significant dialect variation among them. There are many historical questions on the relations between the populations of what are now Denmark, Sweden and Norway that have to be worked out. Thus, for example, one possible hypothesis is that there might have been earlier waves of immigration from Denmark to the Mälar provinces in Sweden. However, even later, right after the beginning of the second millennium AD and at the end of the Viking Age, there was an intensive period when Denmark exerted a considerable political and cultural influence over the Mälar provinces in Sweden, with the new centre in Sigtuna. This was the period of common innovations, mainly phonological ones, in what is now East Nordic (Danish, Swedish, Dalecarlian), and, as Dahl (2001: 230) notes, ‘even the traditional accounts describe this change as a spread, starting in Denmark
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
is the result of a gradual language shift, and more specifically, the spread of a ‘prestige dialect’:
203
and later continuing to Sweden’. Denmark was the most expansive power in the Baltic and in Scandinavia during the twelfth and thirteenth centuries, and around AD 1200, with a population twice as large as that of Sweden and Norway combined. It continued to be the leading state in Scandinavia during the whole of the Middle Ages. Much research has been devoted to describing the different versions of polytonicity across the Continental Scandinavian varieties (a much quoted source is Gårding, 1977), and the distribution of these patterns may bear witness to the complex linguistic contacts among them. In a number of publications, Riad (1998, 2000, 2003) addresses a question that has not received much attention, namely ‘the diachronic developments of the tone accent system into a finer typology of tonal grammars’. He makes several concrete suggestions as to how to interpret the relative ‘linguistic’ distance among the Scandinavian tone accent systems in terms of geographic (and temporal) spreads. As noted in section 2.2, the main parameter here is the realization of Accent 2 as involving one versus two pitch peaks, but there are also additional distinctions in each of the main types. Here I shall leave out the details of the typology itself, but will concentrate instead on the emerging diachronic picture. As mentioned in section 2.2, Riad suggests that tonal accents originate from stress clash (which, in turn, resulted from syncope and other prosodic developments) and argues that one of the ‘two-peaked’ realizations of Accent 2 is the most conservative one. In this variety, the same Accent 2 pattern applies both to simple and to complex stems, for example nunna ‘nun’ and mellanmålet ‘snack between meals’ (< mellan ‘between’ and målet ‘the meal’) – the latter ones demonstrate a particular clear case of two stressbearing units combined in a single word. All the other accent patterns can be linked to each other and to the most conservative one(s) by a series of linguistic changes, or minimal steps. The most conservative variety (in several subtypes) is found • in Swedish spoken in central Sweden (around Stockholm); • in the Swedish varieties spoken on the Baltic coast – in Luleå in the North of Sweden, around Halland in the South of Sweden and in Snäppertuna in Nyland (Finland); • in some varieties spoken in the Swedish (formerly Danish) province of Halland on the eastern coast of Kattegatt; • in Dalecarlian (most significantly in its most archaic Elfdalian vernacular); and • and in the south-eastern Norwegian variety spoken in Stavanger, on the north-eastern coast of the North Sea. There is also some indirect evidence that stød in Danish seems to have developed from a similar tonal system, with the centre in Zealand. The linguistic ‘distance’ between a pair of accent types normally reflects their
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
geographic proximity, but not always in a straightforward way. This can be explained by taking into consideration the partition of Scandinavia into different spheres of influence at the end of the Viking Age and in the Middle Ages, with the two important political centres – Zealand in Denmark and Uppland (the Mälar provinces) in Sweden. Thus, for example, the accent patterns in Southern Sweden and in Southern Norway are much closer to the conservative ‘Central Swedish’ accent type than to those used in geographically adjacent areas. This can be explained by the southern (Danish) influence with a starting point in the use of the conservative type in Zealand. During the spread, or after it, there was a tone shift into what is known as the ‘Dala accent type’ in Southern Sweden (in the provinces of Skåne and Blekinge, which both were parts of Denmark until the seventeenth century), while Zealand itself developed stød and lost its tones. Riad (in particular, 2000, and 2003) shows that even the latter development can be linked easily to the Central Swedish accent type. A transitional type is found in the dialects of the Western Mälardalen (Eskilstuna), which do have tonal distinctions, but in addition show certain phonetic similarities with stød (the so-called ‘curl’, ‘creak’ and sometimes stød), as well as similarities with Danish in the distribution of these features. Riad does not discuss the question of where the Scandinavian tone accents, or rather, the lexical distinctions expressed by them, were born. Neither does he ask the obvious question of why Zealand and Uppland used the same accent type to start with. I believe that his analysis is very much in line with Dahl’s picture of the language spread in Scandinavia. In other words, it would be reasonable to assume that the lexical distinctions expressed by tone accents originated somewhere in Denmark and spread from there by means of the prestige language, with one of the first ‘landing sites’ being the political and cultural centre in the Mälar provinces (Uppland) in Sweden. It is not quite clear to me whether it was from Zealand that Uppland got its accent type, or from Jutland. The predominant opinion seems to learn towards Zealand (and Fyn) – that is, Central Danish – rather than Jutish, as the innovation centre for the development of stød (see Riad, 2000: 293). However, this does not exclude the possibility that the accent type itself underlying this development could have originated in Jutland. The modern distributional peculiarities of stød on Jutland might perhaps be accounted for by a later apocope affecting the existing stød system in various ways (this latter explanation is also suggested in Riad, 2000: 293). The presumed date for the phonologization of word accents, 1000–1200, is based partly on the fact that neither Icelandic nor Faroese shows any similar distinctions. The simplest hypothesis is that the word accents were not yet present in the language of the West Norwegians who colonized these islands. This is, however, problematic, given the fact that both languages have suffixed definite articles that played a crucial role in the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
204
205
phonologization of word accents across Scandinavia (see section 2.2.). The situation is, in fact, similar to that in the Swedish varieties spoken in the Finnish and Estonian environment, which also have suffixed articles, but no word accents. The standard explanation, applied there, might perhaps also apply to Icelandic and Faroese: the word accents might have been lost in a contact situation. This sounds somewhat unexpected, given the reputation of Icelandic as a very conservative Scandinavian variety retained in geographic isolation. However, recent genetic research on Icelandic women (based on comparisons of their mitochondrial DNA with other European populations) shows that they are most similar to the Welsh and British, and that as many as 50 per cent of them may have Celtic origins. Other DNA traces suggest that Icelandic women share DNA also with the Saami, Finns, Russians, Germans, Austrians, Turks and others. The results of this research are not final and are open to criticism from various quarters. However, they do prove that the Icelandic population is far from being homogeneously Scandinavian, and imply that the Icelandic language might have been subject to various contact-induced changes. If the word accents were gradually lost in Icelandic, they must have originated before AD 1000, since Iceland was settled mainly in the late ninth and early tenth centuries. An alternative hypothesis is that the varieties spoken by the first settlers in Iceland lacked both word accents and suffixed articles, but that they were later introduced in Icelandic through fairly regular contact with Norwegians. 2.5.3 Two semicircles or one circle? Now back to the main problem: is there any evidence that the polytonicity in the Scandinavian languages is related to any of the polytonicity phenomena on the eastern coast of the Baltic Sea? A simple answer would be – ‘no’ or at least ‘not that we know of’. The distance is too great – in both geographical and historical terms. There is, however, enough room for further speculation. Wiik (1995, 1997) suggests that the prosodic features of the Circum-Baltic languages all result from close contact between the Finno-Ugric (FU) (or Uralic) and Indo-European (IE) peoples. His main thesis, based on the models by the archaeologists Nuñez and Dolukhanov, is that the periglacial zone of the latest glacial period in Europe (roughly equivalent to the Baltic sea basin, but also stretching southwards to the Rhine) was inhabited by a population speaking Uralic. Their languages were later partly replaced by expanding Indo-European, with the linguistic influences going in two opposite directions: from Finno-Ugric to Indo-European, and vice versa. According to Wiik, a shift from Finno-Ugric to Indo-European could have taken place in the vicinity of the language boundary that moved gradually from northern Germany and Poland to northern Scandinavia and the southern border of Estonia, and could have left a Finno-Ugric substratum in Indo-European. A shift from Indo-European to Finno-Ugric could have taken place in the coastal areas of Finland, Estonia and Latvia, where the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
206
The Circum-Baltic Area
According to this contention, many of the dialect boundaries of the Scandinavian, Russian, and Baltic languages reflect the movement of the FU–IE language boundary through Scandinavia, north-western Russia, and southern Balticum during the last millennium. The further north an area is situated, the more recently it was taken over by the IE language and the stronger its FU substratum. The same process took place on the northern and eastern coasts of the Baltic Sea in that the IE settlers when assimilating themselves with the FU population (the majority population of the area) left their IE substratum in the local FU language, in many cases causing the FU language to split into separate dialects (and later, perhaps, independent languages). The general tendency is that the closer a FU dialect is to the coast, the stronger and of younger layer its IE substratum. (Wiik, 1997: 77) Wiik suggests that one of the Finno-Ugric substratum features in Indo– European is the generalised initial stress in Germanic (as well as in Latvian (see section 2.3) and in the north-western Russian dialects, known for a number of Finnic-substratum phenomena). This ‘main event in the split of Proto-Indo-European into Proto-Germanic and the other IE languages’ had dramatical consequences within Germanic, known as Verner’s law, which was later introduced into Finnic in the form of consonant gradation. Wiik also proposes a few other FU-substratum features in Germanic – for example, umlaut as a reflex of vowel harmony. Some of Wiik’s suggestions have met with a considerable amount of scepticism and criticism on the part of historical linguists. The accent shift in the Germanic is probably the most plausible candidate for a contact-induced change. Here, Wiik follows Salmons (1992) who suggests a shared Germanic–Celtic accent shift talking place in prehistoric north-western Europe on the basis of early and profound contact with a Finno-Ugric language. This is based on the generally accepted view that Proto-Finno-Ugric had an initial stress – a view that might be disputed (Viitso, 1997: 224–5). There are also additional considerations that cast some doubt on the Salmons–Wiik suggestion (see Koptjevskaja-Tamm and Wälchli, 2001: 640). Tones, on the other hand, or rather, the existence of the grave tone (a tonal peak occurring outside the stressed syllable) are regarded by Wiik as a typical Proto-Indo-European phenomenon. In his opinion, the Scandinavian tonal opposition derives from Proto-Indo-European, and the Estonian tonal
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Indo-European settlers would have assimilated with the local indigenous Finno-Ugric people and left an Indo-European substratum in Finno-Ugric. In Wiik’s view, the different Indo-European varieties in Northern Europe originated on the basis of the different Finno-Ugric substratum features left in the previously homogeneous Indo-European protolanguage of Northern Europe.6 The same applies, mutatis mutandis, to the various Finno-Ugric languages:
207
opposition between Quantity 2 and Quantity 3 can be associated with the Germanic difference between presence and the absence of the grave tone. As Wiik (1997: 79) puts it, ‘if there is no grave tone in a language, the tonal peak always occurs on the stressed syllable, and the tone and stress go hand in hand’. From this perspective, the Baltic and Scandinavian tones are both relics of Indo-European, which could be characterized by ‘singing’ speech. The assumption is that tones had been around in the Germanic languages for quite a long time, presumably for a longer time than was normally believed in Scandinavian studies before they became phonologized in the Scandinavian languages. This hypothesis has a certain degree of attraction, given that even other Germanic languages have tones. On the other hand, as mentioned above, earlier attempts to relate Scandinavian tones to archaic Indo-European systems have been rejected by later research. So what remains here is territory where things cannot be proved or disproved – recognizing non-distinctive tones (‘singing intonation’) in the distant past of languages is not that simple. Finally, a few words on the prosodic phenomena found in the CB area from a typological perspective. Tone languages – languages, in which more or less each syllable may have a lexically significant pitch – are very frequent. ‘Pitch-accent’ systems, in which stress co-occurs with tone, are not unheard of at other places as well – for example, in the Daghestanian languages, Avar and Godoberi (Kodzasov, 1999); in Serbo-Croat (as a relic of the Indo-European system); and in Japanese. However, it seems that the class of languages that have been identified under this and similar labels is not coherent (Larry Hyman, personal communication). Also, we lack sufficient information about the distribution of similar phenomena across the world. To conclude: it is remarkable that lexical accents are found in three different groups of Circum-Baltic languages – undoubtedly, a very high concentration in Europe, and probably also globally. However, there is no evidence for any real diachronic connections here, primarily among between the languages to the west of the Baltic Sea (Scandinavian) and those to the east (Baltic, as well as Estonian and Livonian). The conclusion is thus quite disappointing. But stress, pitch, vowel quantity and secondary features of the glottal-closure type are intimately connected, and there is reason to believe that the historical developments that have led to the rise (and fall) of lexical accents in the Circum-Baltic and Rhenish Germanic varieties should also have parallels elsewhere.
3 Rigid GenN word order combined with the (flexible) SVO basic order Most of the CB languages combine the (flexible) SVO basic order on clause level with GenN (the possessor preceding the possessee) as the only possible, or by far most frequent, word order in noun phrases. This is true for Baltic,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
208
The Circum-Baltic Area
(7)
(a)
Lithuanian broli-o brother-GEN.SG ‘a/the brother’s chair’
(b)
Estonian stala-s tüdruk-u table-NOM.SG girl-GEN.SG ‘a/the girl’s dog’
koer dog-NOM.SG
(c)
Northern Sami neidda girl:GEN.SG ‘a/the girl’s house’
(d)
Swedish viessu flicka-n-s hund house:NOM.SG girl-DEF.COM.SG-GEN dog ‘the girl’s dog’
Slavic dialects in the Balticum (Russian, Belarusian and Polish; see Cekmonas, 2001: 122) frequently prepose their genitives, as opposed to the more frequent NG order across Slavic (for example, }any otec ‘father-in-law’, lit. ‘wife: GEN.SG. father’ in the Baltic Russian Rural dialects, as opposed to otec }eny in Standard Russian).8 The same is true for Russian in Finland, in particular, for the Kyyrölä Russian – for example, Russian spoken by the former inhabitants of Kyyrölä village on the Karelian Isthmus, evacuated to safer places in Finland because of the Winter War of 1939–40 (Leisiö, 2001: 98–114). Gen corresponds to the possessor in the morphological genitive case for Finnic, Baltic and Slavic, and to the possessor with the s-marker (s-genitives) for the Scandinavian languages. In the global perspective, there is nothing strange in the SVO/GenN combination: both GenN and NGen are equally common among the SVO languages of the world (Dryer, 1992, 1997). Here, however, Europe differs from the world on the whole in that the consistent GenN order in combination with SVO is found only in the Circum-Baltic languages and in the two Finno-Ugric languages, Komi and Mordvin. It also seems to be a general tendency on a worldwide scale that SVO/GenN languages often occur in geographical clusters and tend to be adjacent to OV languages with GenN order (Dryer, 1999, 2005). Thus, if an SVO language is spoken in the geographic proximity to SVO/GenN languages, there is a high chance that it
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Finnic, Northern/Eastern Saami,7 Standard Swedish, Danish and to a certain degree Norwegian; see example (7):
209
will also manifest the GenN order. In other words, if several languages in the same area show the SVO/GenN properties, there is a high chance that this might be a result of language contacts. Thus, against both the European and the global background, the rigid GenN word order combined with the (flexible) SVO basic word order appears as a potential Circum-Baltic isogloss. The ensuing sections will show that this is not so. However, also in this case, there are numerous interesting language contact phenomena, worth being taken seriously. 3.1 The eastern semicircle: Finnic, Baltic and Slavic The GenN order is on the whole characteristic for Uralic languages, which often have SOV as their basic word order (apart from the above-mentioned languages). The Uralic SOV languages are fairly good representatives for the Eurasian language type, with its word order properties of SOV/NumN/DemN/ AdjN/GenN/RelN. Note that, contrary to what had previously been assumed, in the general case neither the order of adjectives with respect to nouns, nor the order of demonstratives correlates significantly with the basic word order. Thus, independently of the basic word order, adjectives normally follow nouns, while demonstratives and numerals normally precede nouns (Dryer, 1992, 1997). Dryer (1997) suggests that the European VO languages (that is, both SOV and VSO languages) form the following hierarchy from most heavily ‘Noun+ Modifier’ to most heavily ‘Modifier + Noun’: Celtic < Albanian, Romance < Greek, Slavic < Baltic, Finnic In Celtic, all nominal modifiers follow their heads, while in Baltic and Finnic9 all nominal modifiers, apart from relative clauses and prepositional phrases, precede the head. Dryer’s point is that the extreme positions of Celtic and Finnic/Baltic cannot be accounted for by appealing to any global cross-linguistic correlations, but might be explained in terms of their relative geographical and chronological distance from the dominant Eurasian OV type. Thus Finnic languages (together with Komi, Mordvin and Northern and Eastern Saami) are apparently recent members of the VO area in Europe and are geographically close to the area of the dominant Eurasian OV type. As mentioned above, most of their eastern relatives conform to the Eurasian type – whereas VO order is the only deviation from it found in Finnic, Northern and Eastern Sami, Komi and Mordvin. The Celtic languages, on the other hand, are both geographically furthest away from the area of the dominant Eurasian OV languages, and have been separated from them for a considerable period of time. All the other Indo-European groups in Europe are either later migrations from the east and/or migrations that have not gone that far. Dryer’s explanation stops here. So what about the other OV/GenN languages in the CB area?
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
Starting with Baltic (which shares with Finnic the same place in Dryer’s hierarchy), it is difficult to justify the difference between its predominant GenN order and the Slavic NGen order in terms of geographical and chronological distance from the Eurasian OV type. Chronologically, they are probably the same, while geographically Slavic, in particular eastern Slavic, is probably closer. A hypothesis compatible with Dryer’s account would thus have to seek at least some reasons for the Baltic GenN order in Finno-Ugric influence (Matthew Dryer, personal communication). Lithuanian and Latvian differ significantly in obligatoriness of GenN order. Modern Lithuanian allows for both GenN and NG word orders, even though the former is used significantly more frequently. In Latvian, GenN is as fixed as in Finnic. Not surprisingly – as in many other cases – Latvian is closer to Finnic than to Lithuanian, and it is therefore reasonable to attribute this difference to a massive Finnic influence, as this is normally assumed. The Lithuanian case is more complicated. In several investigations on word order in selected Old Lithuanian texts, Vasiliauskien< (1994, 2003) shows that the majority of the sixteenth–nineteenth-century texts on the whole show a predominance of NGen occurrences (see also Schmalstieg, 2003: 149–50, for discussion). The frequency of GenN/NG word order varies considerably, however, from text to text. Since Old Indo-European languages such as Latin, Greek and Vedic had both order types GenN and NG, the relative mobility of genitive attributes in Lithuanian seems to be an old feature. Old Lithuanian and some Lithuanian dialects would thus reflect a state of affairs similar to that found in other old Indo-European languages, with both GenN and NGen. Why Lithuanian has moved towards the predominance of GenN structures is not completely clear – as far as I know, no one has ever suggested any Finnic influence here (for one scenario, see Say, 2004). The frequent GenN order in the Slavic dialects spoken in the Baltic and Finnish linguistic surrounding, as opposed to the normal NG order otherwise, is most probably a result of influence from the Baltic languages. 3.2 Scandinavian languages: a western semicircle? 3.2.1 Types of possessive NPs The Germanic languages outside the CB area either have both GenN/NG orders, or simply prefer NG order. Moreover, the rigid GenN word order in possessive NPs with s-genitives in Standard Mainland Scandinavian is a relatively recent phenomenon. The natural question is, thus, whether there is any connection between the GenN in Swedish, Danish and Norwegian, on the one hand, and in Baltic/Finnic? Hardly any, as I will try to show below. First, compared to Finnic, Baltic and Slavic, with their fairly homogeneous possessive NPs, Mainland Scandinavian varieties on the whole show an impressive diversity in their possessive NPs, probably the highest structural diversity found across any other group of genetically closely related varieties
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
210
Maria Koptjevskaja-Tamm
211
(1) ‘Standard’ S-genitive constructions, illustrated in example (7d), and several groups of S-genitive-related constructions that will be considered in the next section. (2) ‘The father his book’-constructions (‘constructions with resumptive possessive pronouns’ in Norde (1997); ‘a pronominal auxiliary construction’ in Delsing (1993); ‘constructions with linking pronouns’ in Koptjevskaja-Tamm 2003(a); ‘garpe-genitiv’ in the Norwegian tradition); see example (8) below. (3) Constructions with prepositional possessors and the order NGen; see example (9) below. (4) Constructions with dative-marked possessors; see example (10) below. (5) ‘The house his Per’-constructions (‘h-genitives’ in Delsing, 2003a); see example (11) below. Examples: (8)
Constructions with resumptive pronouns (a) Norwegian (‘garpe-genitives’) Olav si-n hest Olav REFL.POSS-N.SG horse ‘Olav’s horse’ (b)
(9)
Western Jutish (Danish) æ mand si-n / hans hatt the man REFL.POSS-COM.SG / his hat ‘the man’s hat’ (Delsing, 1993:153)
Prepositional possessors hest-en til Olav horse-DEF.SG.COM to Olav ‘Olav’s horse (Standard Norwegian)
(10)
Dative-marked possessors bo: k-a prest-um book-DEF.F.SG.NOM priest-DEF.M.SG.DAT ‘the priest’s house’ (Västerbotten, Swedish (Larsson, 1929: 125))
(11)
‘H-genitives’ (a) häns PRART.M.GEN/his ‘Viktor’s horse’
Viktor Viktor
hesst horse
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
in Europe. In addition, a large portion of these constructions manifests NGen word order (for details, see Delsing, 1996, 2003a, 2003b); also KoptjevskajaTamm, 2003a). The main construction types include the following:
The Circum-Baltic Area
(b)
ba: t-o häns Lejonflickt boat-DEF.SG PRART.M.GEN/ his Lejonflyckt ‘Lejonflyckt’s boat’ (Överkalix, Swedish, Källskog (1992: 152–3))
The properties and the distribution of s-genitive constructions and their relatives will be considered in section 3.2.2. ‘The father his book’-constructions (with the GenN-order) in various versions are well attested across Germanic (for details, see KoptjevskajaTamm, 2003a). One variant, known as ‘garpe-genitiv’ in the Norwegian tradition, constitutes one of the standard constructions in Nynorsk and is making its way into Bokmål; it is also used in many western and northern varieties of Norwegian. ‘Garpe’ was the pejorative name for Hanseatic merchants used in Bergen, and the ‘garpe-genitive’ construction was itself borrowed from Low German (and re-analysed) in the Hanseatic city of Bergen, from where it spread to other Norwegian varieties (for details, see Torp, 1990). A similar construction is used in the Southern and Western Jutish dialects of Danish. Swedish varieties are normally held to lack a counterpart to ‘the father his book’-construction – a view that might be in need of revaluation, given some of the Dalecarlian facts (see section 3.2.2 below). Constructions with prepositional possessors and the order NG are, of course, well known from all the non-Scandinavian Germanic varieties (and modern Romance), where the standard possessive prepositions have originated from prepositions with the ablative/separative meaning (von in German, of in English, de in French). Standard Norwegian and most Norwegian dialects have standard ‘possessive’ prepositions – curiously, this time their original meaning is most often ‘goal, direction’: compare til ‘to’ in Standard Norwegian and til or åt ‘to’ in dialects. Av ‘of’ is used in a restricted area in Southern Norway (Telemark and Agder; see Torp, 1973). In many cases, such constructions coexist with other possessive constructions in the same variety. In Swedish, the corresponding constructions with at/åt are also attested in some Northern varieties (for example, in Medelpad and Värmland – Delsing, 1996: 54). Constructions with dative-marked possessors in Germanic are found primarily in the northern Swedish dialects (the provinces of Norrbotten, Västerbotten), and in Dalecarlian. At least, in the Norrbotten dialects the word order may vary; in Dalecarlian the dative possessor is always postposed (NGen), and this seems to be the case in Västerbotten too. ‘The house his Per’-constructions (‘h-genitives’ in Delsing, 2003a) are, as far as I can judge, mainly a Scandinavian speciality (for some parallels, see Koptjevskaja-Tamm, 2003a: 629–32). These periphrastic constructions are restricted to proper names and some kin terms as possessors and, thus coexist with other possessive NPs for all the other possessors. Possessive pronouns look very much like preproprial articles (that is, articles pertaining to proper names), typical in many Scandinavian varieties, and this construction
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
212
213
and preproprial articles show a high degree of geographical overlapping. However, as pointed out by Delsing (2003a), there are certain discrepancies here (for details and analysis, see also Koptjevskaja-Tamm, 2003a: 630–2). In Mainland Scandinavian, this construction, with varying properties, is attested in most of Norway (excluding a few areas in the south) and in two non-contiguous northern areas in Sweden – an inland zone comprising parts of Jämtland and Medelpad, Härjedalen, Västerdalarna, and parts of Värmland, and a coastal zone, comprising the provinces of Västerbotten and Norrbotten (but excluding Lapland). Its most elaborate form is found in Colloquial Icelandic. The word order is normally NGen, but some varieties allow for GenN as well. This list already shows that the clustering of SVO/GenN languages in the Circum-Baltic languages is partly an illusion, to a certain extent characteristic of the clash between the methods of large-scale typological research (including areal typology), on the one hand, and those of more fine-grained areal linguistics. Large-scale typology is, of necessity, selective in the number and types of varieties included in the sample. Non-standard varieties are often invisible in large-scale typological research, and what looks like two languages spoken in the immediate vicinity of each other may turn out to be separated by a number of others. In our concrete case, not only is there a number of Scandinavian varieties with the SVO/NG word order, but these are also often spoken in territories relatively close to those inhabited by a FinnoUgric population (Finns and Saami). These considerations cast doubts on any attempts to link the SVO/GenN word order in those Scandinavian varieties where it occurs with a similar phenomena in Finno-Ugric. However, the Scandinavian possessive constructions themselves, and in particular, s-genitives, do bear witness to important and interesting language contacts, similar to the lexical accents considered in section 2 above. 3.2.2 S-genitives and related phenomena: distribution, typology and language contacts The structure of possessive NPs across Standard Mainland Scandinavian (for Norwegian, mainly in Bokmål, see below) is very simple: non-pronominal possessors are marked with the genitive marker -s, while pronouns have a special possessive form. The marker -s does not only have the same form as the English Saxon genitive (’s), but also behaves in roughly the same way or at least a very similar way: (1) it attaches to all kinds of non-pronominal possessors, not necessarily to proper names and to kin terms. Here, the Scandinavian languages are even more consistent than English, where the Norman of-genitive may be preferred for possessors that are very low in animacy; (2) it appears always at the very end of a noun, after all suffixes; compare Swedish en pojke-s ‘a boy-GEN’, pojke-n-s ‘BOY-DEF.COM.SG-GEN’, pojk-ar-s ‘BOY-PL-GEN’, pojk-ar-na-s ‘BOY-PL-DEF-GEN’;
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
(3) it also shows the tendency to attach to the final word of a NP resulting in what is traditionally called ‘group genitives’, as in [mannen på gatans] åsikter – lit. ‘the man in the street’s opinions’ (that is, an ordinary person’s opinion); (4) its form is even more consistent than the form of -’s in English, in that it is always pronounced in the same way; (5) the possessor always precedes the possessee; and (6) preposed possessors (both pronominal and lexical) are incompatible with definite and indefinite articles pertaining to the host nominal, compare examples (12) and (13) below. The description given above is somewhat simplified: thus, some nouns have special possessive forms, group genitives are severely restricted, there are special cases where s-genitives co-occur with indefinite articles and so on (see Koptjevskaja-Tamm, 2003b and, in particular, Börjars, 2003, for details). (12)
(a)
en INDEF:COM
(c)
(13)
hund (b) dog
‘a dog’ den stor-a DEF:COM.SG big-DEF ‘the big dog’
hund-en dog-DEF.COM.SG ‘the dog’ hund-en dog-DEF.COM.SG
(a)
*flicka-n-s en girl-DEF.COM.SG-GEN INDEF:COM flicka-n-s hund girl-DEF.COM. dog SG-GEN ‘*the girl’s a dog/a the girl’s dog’
hund/ en dog INDEF:COM
(b)
hund-en / *den flicka-n-s hund / dog-DEF.COM. girl-DEF.COM.SG-GEN dog/ SG / DEF. COM.SG *flicka-n-s den hund girl-DEF.COM.SG-GEN DEF.COM.SG dog ‘*the girl’s the dog / the girl’s dog’ *flicka-n-s girl-DEF.COM.SG-GEN
The ‘s-genitive’-construction exemplified above is found, first of all, in Standard Danish and in Standard Swedish, which lack any competing standard ‘possessive’ preposition, analogous to the English of. In Standard Norwegian, s-genitives are much more frequent in Bokmål than in Nynorsk and are often considered to be a Danish leftover in the language. As mentioned above, English uses a very similar construction. Turning to dialects (the most comprehensive sources here are Delsing, 1996, 2003b), s-genitives are found in most Danish dialects, apart from
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
214
215
Western Jutish. In Norwegian, they are restricted mainly to the southernmost dialects in Telemark and Agder. In Swedish, s-genitives are quite widespread in southern, western and eastern dialects – that is, in the grammatically least ‘spectacular’ dialects. They are absent from most of the northern dialects (see below) and from Dalecarlian; the Swedish dialects in Finland and Estonia use s-genitives in a different construction, as will be shown below. It has been proposed that the properties (2)–(4) – and, less directly, (1) – of the marker -s can be attributed to its status as a clitic, a right-edge marker (phrasal affix) or as a marker that at least sometimes has right-edge marking properties (the literature is extensive; some examples are Carstairs, 1987; Johannesen, 1989; Delsing, 1993. For a principled account of the conflicting properties of the s-marker, see especially Börjars, 2003). The rigid GenN word order, property (5) in Standard Mainland Scandinavian is a relatively recent phenomenon, dating in Swedish from 1250–53; before that, genitives could both precede and follow their head. Stabilization of GenN order in Mainland Scandinavian is described as going hand-in-hand with the transition from the morphological genitive case to the phrase-final s-genitive and with its reanalysis as a determiner, incompatible with other articles; that is, property (6) (see (Delsing, 1991; Norde, 1997). Curiously, as far as I can judge, the evident parallels both in the synchronic behaviour of s-genitives in English, Danish and Swedish and in their historical development, although often acknowledged, have not yet been taken seriously.10 The development of s-genitives in each of the languages is normally considered as being triggered by language-internal factors, where the key process is taken to be ‘degrammaticalization’ of one of the several morphological genitive endings (-s) and its victory over all the others (see Börjars, 2003, for the arguments against applying this term to the Swedish s-marker). For English and Norwegian, a competing account suggests that the s-genitive arose because of the contamination of two different constructions – constructions with genuine morphological genitives and constructions with resumptive pronouns (‘the father his book’-constructions), in which the weak form of the resumptive pronoun was phonetically similar to the genitive ending -s. For Swedish, Norde (1997: 91) rejects this account, because -s was not homonymous with any possessive pronoun. She also adds that ‘there are no indications that RPP-constructions [resumptive possessive pronouns] were ever relevant in Swedish. In modern standard Swedish, they do not occur and I found no evidence of RPP-constructions in the Old and Middle Swedish texts I examined’. Now, not only are the s-genitives in Standard Mainland Scandinavian (and English) very similar to each other, but the above-mentioned properties make them cross-linguistically very unusual, not to say unique, in several respects. First, it is difficult to find real counterparts to the s-marker with properties (1)–(4); that is, a short uniform marker attaching to the edge of the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
possessor-NP. In the many years I have been involved in cross-linguistic research on possessive NPs, I have not come across any other languages where a possessor is marked with a comparable marker. This does not, of course, rule out such a possibility in principle, but still suggests that it cannot be a frequent option cross-linguistically. One possible parallel would be group-inflecting case markers in languages such as Basque, but these normally build a paradigm that involves more distinctions than the one between unmarked versus genitive case (see Dahl and Koptjevskaja-Tamm, 2005, for some reflections on case hierarchies, in particular in relation to Scandinavian vernaculars). S-genitives share, of course, a number of similarities with postpositions, but then it should be borne in mind that they occur in typically prepositional languages. This problem has a diachronic counterpart. As mentioned above, the emergence of s-genitives is nowadays often described as ‘degrammaticalization’, which, in turn, has been used in some quarters as an important (not to say the chief) argument against the notion of ‘grammaticalization’. While proponents of grammaticalization constantly come up with new examples, its opponents similarly are constantly quoting the same few examples, with s-genitives in the first place. ‘Degrammaticalization’, even in the broad understanding of this term, is thus very unusual. Incompatibility of possessors with articles pertaining to the whole possessive NP is attested in a number of languages and is, on the whole, more typical of possessors preceding possessees (GenN) (for details, see Haspelmath, 1999). However, it is, on the whole, not very frequent cross-linguistically. Thus the s-genitive constructions in Standard Mainland Scandinavian and in English share several very unusual properties cross-linguistically. In addition, it is, of course, striking that the possessive markers in all these languages have exactly the same form. All this contributes to the impression that language contacts must have played an important part here. I assume that a standard explanation for the emergence of these similarities would be to invoke the close genetic relationship and the inherited similarities among the varieties under consideration that would underlie parallel developments springing from one and the same source. In this case, for example, the shared morphological genitive ending -s must have been particularly salient, judging from its later career across Germanic. Possessor–article complementarity might possibly be explained by more general grammaticalization processes that are connected with the chronological relations between the rise of the possessive construction and the rise of obligatory definiteness markers (the former pre-dating the latter), as suggested by Haspelmath (1999). This explanation is not convincing. As we have already seen, the Mainland Scandinavian varieties – and in particular, the less levelled varieties – abound in structurally different possessive NPs that are not related at all to the older Germanic genitive constructions. But even those that are do not necessarily show the same properties as the ‘standard s-genitive construction’, as will be
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
216
Maria Koptjevskaja-Tamm
217
(1) Unrestricted s-genitives with definite possessees: in the Swedish varieties in Finland and Estonia, all non-pronominal possessors attach s-markers and precede head nouns with suffixed definite articles. (2) ‘Generalized’ genitives with proper names and definite possessees are found over large parts of western Norrland in Sweden. The generalized marker is sometimes -s, but not necessarily (for example -a, -ses, -sa are attested). The word order is normally NG, but even GenN occurs. (3) ‘Generalized’ genitives with proper names and bare possessees are reported in a few varieties. (4) Dative-marked possessors with the clitical es marker are found in a few archaic Dalecarlian vernaculars (for example, in Elfdalian). We see that the properties associated with the s-marker do not have to cluster. Thus all these variants have a ‘generalized’ marker associated with the possessor. However, the first three involve a straightforward morphologically bound marker and lack group genitives. The older salient genitive marker is sometimes only generalized to proper names and kin terms – similarly to the situation in German and Dutch; interestingly, the generalized marker is not necessarily based on -s. Possessors do not have to be in complementary distribution with the article. The Dalecarlian variant involves, most probably, a reduced form of the third-person enclitic masculine singular genitive pronoun os (< Old Scand. hans) and may, thus, basically be a construction with a resumptive pronoun (‘the father his book’-construction). The development of this construction may also have been pushed forward by contamination with the old genitive construction, in the same way as this has been suggested for English and Norwegian (for details, see Dahl and Koptjevskaja-Tamm, 2005). If this hypothesis is true, it slightly weakens Norde’s (1997: 91) assumption that constructions with resumptive pronouns have never been relevant for the history of Swedish: even if they are not attested in the written texts, they may have been around in less formal varieties. To summarize this long discussion: several groups of facts make it improbable that the standard s-genitive constructions in Danish, Swedish and Norwegian could have arisen independently of each other. The relevant facts include: • their cross-linguistically unusual (or even unique) properties; • their geographic distribution (standard languages and the strongly levelled varieties); • the enormous diversity in possessive NPs across Scandinavian; among other things, the existence of several groups of related constructions that lack some of the key properties of the standard s-genitives.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
shown here. The s-genitive-related constructions in Mainland Scandinavian (in Swedish and Dalecarlian) include at least the following variants:
The Circum-Baltic Area
Mutatis mutandi, it is difficult to imagine a parallel and completely independent emergence of s-genitives in English, on the one hand, and in Mainland Scandinavian, on the other. Östen Dahl and I are currently considering various scenarios for these developments. Given the dominant role of Denmark in Scandinavia over a long period and in view of the hypothesis that various innovations spread from the Danish territory together with the prestige language, considered in section 2.5, it is natural to suspect that the emergence of s-genitives in both Swedish and Norwegian has been largely influenced by language contact with Danish. The details of this development and of the interaction between language-internal factors and linguistic contacts are yet to be established. Trying to understand when, where and how linguistic contact could have shaped these constructions in Scandinavian, on the one hand, and in English, on the other, is an even greater challenge. It is acknowledged that the Danelaw in England had a certain impact on Old English (which at that time was fairly close to Old Danish); for an important recent contribution to this topic, see McWhorter (2002). However, Danish rule in England had ended by 1066, at a time when s-genitives had not gained ground in any of the languages under consideration. A hypothesis we are working with is that the presence of the Danes in England could have led to contact-induced changes not only in English, but also in Danish. Curiously, this possibility has never been discussed. In the particular case of s-genitives, we hypothesize that the intensive linguistic contact between the English and Danish populations in the Danelaw and in Denmark (primarily in Jutland) could have contributed to the development of one simple possessive construction in both English and Danish. The s-genitive construction is, undoubtedly, much more homogeneous and simple than its predecessors, it is easily acquired (see Hammarberg and Koptjevskaja-Tamm, 2002 on the acquisition of s-genitive constructions by students of Swedish), and seems to be a good choice in communication between speakers of two relatively close linguistic varieties. A good parallel might be the remarkable expansion of the borrowed garpe-genitive construction across Norwegian (see section 3.2.1). There is, of course, a good chance that we shall never get a satisfactory answer to the question of why standard possessive constructions in Danish, Swedish, Norwegian and English are so similar to each other. However, this should not diminish the relevance of the question itself.
4
Conclusions
In their important paper on areal linguistics, Campbell et al. (1986: 533) distinguish two groups of areal studies: • those that simply catalogue observed similarities among neighbouring languages – the ‘circumstantialist’ approach; and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
218
Maria Koptjevskaja-Tamm
219
In doing areal linguistics, we in fact need both steps – collecting observed similarities and checking whether they can be accounted by language contacts. Discovering similarities themselves is a difficult task, since what can be meant by a similarity differs greatly from case to case, so hypotheses about suggested similarities among languages should always be warmly welcomed, and it is all the more important that such hypotheses come from different sources – for example, both from the more traditional contact linguistics and from largescale typological research, as in the two cases considered in this chapter. Neither of the two phenomena that at first sight seem to unite languages around the Baltic Sea – polytonicity and possessive NPs – completes a full circle around the Baltic Sea area, and thus is a true ‘Circum-Balticism’. This confirms, therefore, the general conclusion in Koptjevskaja-Tamm and Wälchli (2001) that there appear to be no ‘real’ isoglosses covering all the CB languages, that the convergence in this area works primarily on a microlevel, and that the whole area hardly lives up to being conceived of as a Sprachbund, or a linguistic area. So far, a negative and perhaps disappointing result, particularly given the general focus of the present volume. Bad luck, undoubtedly. However, to quote a comment by one of the editors, ‘testing hypotheses about contacts and Sprachbund-ness is an essential part of the work to be done in this area’, and this is especially true for those phenomena that look particularly promising. On the other hand, we have seen that each of the phenomena considered in the chapter bears witness to interesting linguistic contacts. Some of these are generally acknowledged – primarily contacts between genetically unrelated (Estonian/Livonian-Latvian, Swedish-Estonian, or Swedish-Finnish) or not immediately related languages (Norwegian–Low German, Russian– Lithuanian). Others, however, are largely underestimated – these are contacts among the closely related Germanic varieties (those among Scandinavian varieties and, possibly, those between Old Danish, on the one hand, and Old English on the other). Contacts of the former type traditionally attract a great deal of attention, in historical linguistics, in areal linguistics and in typology, while contacts of the latter type are often ignored, underestimated or even invisible in such research (see McMahon and McMahon in Chapter 3 in this volume for a beautiful illustration of the impact that ‘underestimated’ borrowings between related languages can have on the evaluation of their genetic relatedness). Thus historical linguistics usually appeals to contact-induced change only when language-internal factors do not work. It also tends to rely exclusively on written sources without paying sufficient attention to the geographical and social stratification of the language varieties that in one
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
• those that attempt to find arguments to explain these similarities. Thus, while some of them have arisen from borrowing or convergence, others may simply be accidental and/or typologically frequent, or even inherited.
The Circum-Baltic Area
or another way might have contributed to the formation of the language under consideration (as is the case, for example, with most of the research on the development of s-genitive constructions in English and Swedish). Areal linguistics looks specifically for shared structural traits that do not seem to fit genetic classifications and may thus easily miss contact-induced changes within closely-related varieties. Also, typological research is mainly interested in shared traits that are independent of genetic relations. This is reflected in its methodology, primarily in the choice of representative samples: in compiling a sample, typologists try to avoid too many genetically related varieties spoken in the vicinity of each other, which, in practice, normally means that the cross-dialectal diversity in an area is reduced to one standard language. Contacts among genetically closely related varieties are definitely worth much more attention than they traditionally receive – I hope that this chapter has shown that.
Acknowledgements In writing this chapter I have received much inspiration, support and help from Östen Dahl, Tomas Riad and Martin Tamm. I would also like to thank Claes-Christian Elert, Olle Engstrand and Larry Hyman for discussing with me tonal accents in Scandinavian languages and elsewhere, and to Anette Rosenbach and Paola Crisma for talking s-genitives. Thanks to Brigitte Pakendorf for providing me with sources of genetic research in Europe, and to the editors for valuable comments on the preliminary version. My general understanding of the CB area has to a great extent been formed thanks to collaboration with Bernhard Wälchli.
Notes 1 As will become clear from the list, the Baltic/Slavic genitive case often corresponds to the Finnic partitive case. The difference is, however, not purely terminological: the Finnic partitive case is not used in adnominal possession, which is the prototypical function of the genitives in Baltic, Slavic and Finnic. 2 de Haan (2005) himself classifies Lithuanian, Latvian and Estonian as having separate evidential affixes – which strikes me as surprising. 3 A note on the traditional dialectal division of the Mainland Scandinavian languages. The dialects of Danish have been divided traditionally into Jutish (in Jutland), Insular Danish (on the islands of Funen and Zealand and the smaller islands to the south of them), and East Danish (Bornholm, and the southern Swedish provinces of Skåne, Halland and Blekinge, which were all under Danish rule until the seventeenth century). The traditional division of the Swedish dialects considers the latter group as Southern Swedish dialects; in addition there are Eastern (sveamål, with the core area in Uppland; these also include Dalecarlian), Western, or Guthnic (north-west of Skåne), Northern (north of the Eastern area along the coast of the Gulf of Bothnia), Gotlandic, or Gutnish (on the island of Gotland), and Trans-Baltic (in Finland and Estonia). The dialects of Norwegian
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
220
4
5
6
7 8
9
10
221
are traditionally divided into five main groups: Western, Central, Eastern, Trønder, and Northern. Some Northern Swedish dialects have a special tonal contour, ‘circumflex’, in overlong monosyllables that have arisen because of apocope in Accent 2 bisyllables – for example, kasta > kâast ‘to throw’. According to some researchers, however, polytonicity had simply never spread to these dialects, or to their ancestors on the Swedish mainland – a view rejected by most of the experts. A note on the recent genetic research on the Circum-Baltic populations. Rosser et al. (2000) and, in particular, Zerjal et al. (2001) show that the frequencies of certain Y-chromosomal markers – haplogroups 1 and 16 – separate speakers on the western and eastern coasts of the Baltic Sea. Haplogroup 1 is strongly represented among the Swedes (including Gotlanders) and Norwegians, but weakly among the Saami, Finns, Estonians, Latvians and Lithuanians. This fits its known pattern of distribution, with the high frequency in West Europe (reaching ‘near-fixation’, 98.5 per cent in the west of Ireland) and a lower frequency in the east. One interpretation is that such chromosomes reflect the population inhabiting (West) Europe before there was a major demographic expansion of agricultural migrants from the Near East (Rosser et al., 2000: 1539). Haplogroup 16 shows the opposite pattern: a high frequency among the Saami, Finns, Estonians, Latvians and Lithuanians, and low among the Swedes (including Gotlanders) and Norwegians. This fits again with its known distribution: a high frequency in Northern Asia, but absent from most of Europe. A further analysis supports the idea that the genetic history of Y-chromosomes among Latvians and Lithuanians is still distinct from that of the Uralic speakers. Thus ‘our interpretation of the Y-chromosomal data is that the major genetic difference in this area is geographical, distinguishing populations living on the western side of the Baltic from those on the eastern side. However, a significant difference was also detectable between Finno-Ugric speakers and Baltic speakers on the eastern side, where the Latvians showed greater genetic similarity to the Lithuanians than to the Estonians, demonstrating that linguistic differences can have a lesser, but still important influence on the distribution of genetic diversity’ (Zerjal et al., 2001: 1086). Finally, Tambets et al. (2004: 661) argue that the ‘the large genetic separation of the Saami from other Europeans is best explained by assuming that the Saami are descendants of a narrow, distinct subset of Europeans’. I am grateful to Brigitte Pakendorf for providing me with these references. Southern Saami has a flexible word order, with SOV as the basic form. The use of the alternative strategy of expressing possessors, by means of the so-called possessive adjectives that precede their heads, has declined considerably down the ages in most varieties of Slavic, and Slavic on the whole seems to have been on a steady move towards the postposition of possessors. Komi and Mordvin are not in Dryer’s sample; note also that Dryer argues in terms of whole language families, which explains why Swedish, Danish and Northern/Eastern Saami are absent from the hierarchy. Jespersen (1912) suggests that the increased use of prenominal genitives in English was a result of the presence of Scandinavian settlers, but the hypothesis was later rejected in Mitchell (1985). However, neither of the two provides an argument to support his position. I am grateful to Paola Crisma for drawing my attention to these sources.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
222
The Circum-Baltic Area
Aikhenvald, A. Y. (2004) Evidentiality (Oxford: Oxford University Press). Balode, L. and A. Holvoet (2001a) ‘The Latvian Language and its Dialects’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, vol. 1, pp. 3–40. Balode, L. and A. Holvoet (2001b) ‘The Lithuanian Language and its Dialects’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, pp. 41–80. Börjars, K. (2003) ‘Morphological Status and (de)grammaticalisation: The Swedish Possessive’, Nordic Journal of Linguistics, vol. 26, no. 2, pp. 133–6. Campbell, L. (1996) ‘Typological and Areal Issues in Grammar’, in K. Brown and J. Miller (eds), Concise Encyclopedia of Syntactic Theories (Oxford: Elsevier Science Ltd.), pp. 339–43. Campbell, L., T. Kaufman and Th. C. Smith-Stark (1986) ‘Meso-America as a Linguistic Area’, Language, vol. 62, no. 3, pp. 530–7. Cairstairs, A. (1987) ‘Diachronic Evidence and the Affix-clitic Distinction’, in A. Giacalone Ramat, O. Carruba and G. Bernini (eds), Papers from the 7th International Conference on Historical Linguistics (Amsterdam/Philadelphia: John Benjamins), pp. 151–62. Cekmonas, V., ‘Russian Varieties in the Southeastern Baltic Area: Rural Dialects’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, vol. 1, pp. 101–36. Dahl, Ö. (2001) ‘The Origin of the Scandinavian Languages’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, pp. 215–3. Dahl, Ö. and M. Koptjevskaja-Tamm (1992) ‘Language Typology Around the Baltic Sea: A Problem Inventory’, Papers from the Institute of Linguistics University of Stockholm (PILUS). Dahl, Ö. and M. Koptjevskaja-Tamm (eds) (2001) The Circum-Baltic Languages: Typology and Contact. Studies in Language Companion Series (SLCS) (Amsterdam/ Philadelphia: John Benjamins). Dahl, Ö. and M. Koptjevskaja-Tamm (2005) ‘The Resilient Dative’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Typological Profile of Swedish (Special issue of Sprachtypologie und Universalienforschung), pp. 58–78. Décsy, G. (1973) Die linguistische Struktur Europas. Vergangenheit–Gegenwart–Zukunft (Wiesbaden: Otto Harrassowitz). de Haan, F. (2005) ‘Coding of Evidentiality’, in M. Haspelmath, M. Dryer, D. Gil and B. Comrie (eds), The World Atlas of Language Structures (Oxford: Oxford University Press), pp. 318–21. Delsing, L.-O. (1991) ‘Om genitivens utveckling i fornsvenskan’, in S.-G. Malmgren and B. Ralph (eds), Studier i svensk språkhistoria 2 (Gothenburg: Göteborg Institutionen för Svenska språket), pp. 12–30. Delsing, L.-O. (1993) ‘The Internal Structure of Noun Phrases in the Scandinavian Languages’, Doct. diss. (Lund: University of Lund, Department of Scandinavian languages). Delsing, L.-O. (1996) ‘Nominalfrassyntax i skandinaviska dialekter’, in Ø. A. Vangsnes (ed.), Syntaktisk variasjon i nordiske målföre. Artiklar frå ein dugnad under MONS 6 i Bergen, 24. november 1995 (Bergen: Nordisk institutt, Universitetet i Bergen), pp. 24–74. Delsing, L.-O. (2003a), ‘Perifrastisk genitiv i nordiska dialekter’, in A. V. Øystein, A. Holmberg and L.-O. Delsing (eds), Dialektsyntaktiska studier av den nordiska nominalfrasen (Oslo: Novus), pp. 65–84.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
References
223
Delsing, L.-O. (2003b), ‘Syntaktisk variation i nordiska nominalfraser’, in A. V. Øystein, A. Holmberg and L.-O. Delsing (eds), Dialektsyntaktiska studier av den nordiska nominalfasen (Oslo: Novus), pp. 11–64. Dogil, G. (1999) ‘Baltic Languages’, in H. van der Hulst (ed.), Word Prosodic Systems in the Languages of Europe (Berlin: Mouton de Gruyter), pp. 877–96. Dryer, M. (1992) ‘The Greenbergian Word Order Correlations’, Language, vol. 68, pp. 81–138. Dryer, M. (1997) ‘Aspects of Word Order in the Languages of Europe’, in A. Siewierska (ed.), Constituent Order in the Languages of Europe (Berlin/New York: Mouton de Gruyter), pp. 283–319. Dryer, M. (1999) Explaining the Order of Genitive and Noun in SVO Languages. Third Biennual Meeting of the Association for Linguistic Typology. Amsterdam. Dryer, M. (2005) ‘Order of Genitive and Noun’, in M. Haspelmath, M. Dryer, D. Gil, and B. Comrie (eds), The World Atlas of Language Structures (Oxford: Oxford University Press), pp. 350–53. Falkenhahn, V. (1963) ‘Die Bedeutung der Verbalrektion für das Problem eines litauisch-polnischen Sprachbundes’, Zeitschrift für Slawistik, vol. 6, pp. 893–907. Fikkert, P. and Jacobs, H. (eds) (2003) Development in Prosodic Systems (Berlin/New York: Mouton de Gruyter) Geniušien<, E. (1987) The Typology of Reflexives (Berlin/New York: Mouton de Gruyter). Gussenhoven, C. (2000) ‘On the origin and development of the Central Franconian tone contrast’, in A. Lahiri (eds), Analogy, Levelling, Markedness: Principle of Change in Phonology and Morphology (Berlin/New York: Mouton de Gruyter). Gussenhoven, C. and G. Bruce (1999) ‘Word Prosody and Intonation’, in H. van der Hulst (ed.), Word Prosodic Systems, pp. 233–71. Gårding, E., The Scandinavian Word Accents (Lund: C. W. K. Gleerup, 1977). Haarmann, H. (1976) Aspekte der Arealtypologie. Die Problematik der europäischen Sprachbünde. (Tübingen: Narr). Haarmann, H. (1970) Die indirekte Erlebnisform als grammatische Kategorie. Eine eurasische Isoglosse (Wiesbaden: Otto Harrassowitz). Hammarberg, B. and M. Koptjevskaja-Tamm (2002) ‘Adnominal Possession: Combining Typological and Second Language Perspectives’, in A. Giacalone Ramat (ed.), Typology and Second Language Acquisition (Berlin/New York: Mouton de Gruyter), pp. 125–79. Haspelmath, M. (1987) Transitivity Alternations of the Anticausative Type, University of Cologne, Department of Linguistics, Working Paper No. 5 (Neue Folge) (Cologne). Haspelmath, M. (1999) ‘Explaining Article–Possessor Complementarity: Economic Motivation in Noun-Phrase Syntax’, Language, vol. 75, pp. 227–43. Heijmans, L., (2003) ‘The Relationship Between Tone and Vowel Length in Two Neighboring Dutch Limburgian Dialects’, in P. Fikkert and H. Jacobs (eds), Devlopment in Prosodic Systems, pp. 7–45. Hopper, P. and S. A. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. van der Hulst, H. (ed.) (1999) Word Prosodic Systems in the Languages of Europe (Berlin: Mouton de Gruyter). Jakobson, R. (1931a), ‘K xarakteristike evrazijskogo jazykovogo sojuza’, in R. Jakobson (ed.), Selected Writings (The Hague: Mouton, 1971), pp. 144–201.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
Jakobson, R. (1931b), ‘Über die phonologischen Sprachbünde’, in R. Jakobson (ed), Selected Writings (The Hague: Mouton, 1971), pp. 137–43. Jespersen, O. (1912) Growth and structure of the English language (Leipzig: Teubner). Johannesen, J. B. (1989) ‘Klitika – en avgrensning’, Norsk linguistisk tidsskrift, vol. 2, pp. 117–4. Kemmer, S. (1993) The Middle Voice (Amsterdam: Benjamins). Kodzasov, S. (1999) ‘Caucasian: Daghestanian Languages’, in H. van der Hulst (ed.), Word Prosodic Systems, pp. 995–1020. Koptjevskaja-Tamm, M. (2003a) ‘Possessive Noun Phrases in the Languages of Europe’, in F. Plank (ed.), Noun Phrase Structure in the Languages of Europe (Berlin/ New York: Mouton de Gruyter), pp. 621–722. Koptjevskaja-Tamm, M. (2003b) ‘A Woman of Sin, a Man of Duty, and a Hell of a Mess: Non-determiner Genitives in Swedish’, in F. Plank (ed.), Noun Phrase Structure in the Languages of Europe (Berlin/New York: Mouton de Gruyter), pp. 515–58. Koptjevskaja-Tamm, M. and B. Wälchli (2001) ‘The Circum-Baltic Languages: An Areal-typological Approach’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, pp. 615–750. Källskog, M. (1992) Attityd, interferens, genitivsyntax. Studier i nutida överkalixmål (Uppsala: Dialekt-och folkminnesarkivet). Lahiri, A., T. Riad and H. Jacobs (1999) ‘Diachronic Prosody’, in H. van der Hulst (ed.), Word Prosodic Systems, pp. 335–422. Larsson, S. (1929) Substantivböjningen i Västerbottens folkmål (Uppsala: Appelberg). Lehiste, I. (1988) Lectures on Language Contact. (Cambridge, Mass.: MIT Press). Lehiste, I. (1983) ‘Prosodic Change in Progress: Evidence from Estonian’, in I. Rauch and G. F. Carr (eds), Language Change (Indiana: Indiana University Press), pp. 10–27. Lehiste, I. (2003) ‘Prosodic Change in Progress: From Quantity Language to Accent Language’, in P. Fikkert and H. Jacobs (eds), Development in Prosodic Systems, pp. 47–66. Lehiste, I. (1997) ‘Search for Phonetic Correlates in Estonian Prosody’, in I. Lehiste and J. Roos (eds), Estonian Prosody, pp. 11–35. Lehiste, I. and Roos, J. (eds) (1997) Estonian Prosody (Tallinn: Institute of the Estonian Language). Leisiö, L., (2001) Morphosyntactic Convergence and Integration in Finland Russian (Tampere: University of Tampere). Lorentz, F. (1903) Slovinzische Grammatik (St Petersburg: Mokslas). Lorentz, O. (1995) ‘Tonal Prominence and alignment’, Phonology at Santa Cruz, vol. 4, pp. 39–56. Mathiassen, T. (1985a) ‘A Discussion of the Notion ‘Sprachbund’ and its Application in the Case of the Languages in the Eastern Baltic Area’, International Journal of Slavic Philology, vol. 21/22, pp. 273–81. Mathiassen, T. (1985b) Slavisk-baltisk-østersjöfinske syntaktiske isoglosser og spørsmålet om et Sprachbund i den østlige del av østersjøområdet. X Nordiska Slavostmötet 13–17 August 1984 (Åbo: Research Institute of the Åbo Akademi Foundation). McWhorter, J. H. (2002) ‘What Happened to English?’, Diachronica, vol. 19, no. 2, pp. 217–72. Mitchell, B. (1985) Old English syntax (Oxford: Clarendon Press). Nau, N. (1996) ‘Ein Beitrag sur Arealtypologie der Osteseeanrainersprachen’, in N. Boretzky (ed.), Areale, Kontakte, Dialekte, Sprachen und ihre Dynamik in mehrsprachigen Situationen (Bochum: Brockmeyer), pp. 51–67. Norde, M. (1997) The History of the Genitive in Swedish. A Case Study in Degrammaticalization (Amsterdam: Amsterdam University).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
224
225
Næss, Å. (2003) Transitivity: From Semantics to Structure (University of Nijmegen). Oftedal, M. (1952) ‘On the Origin of the Scandinavian Tone Distinction’, Norsk tidsskrift for sprogvidenskap, vol. 16, pp. 201–25; reprinted in E. H. Jahr and O. Lorentz (eds), Prosody/Prosody (Oslo: Novus, 1983). Riad, T. (1998) ‘The Origin of Scandinavian Tone Accents’, Diachronica, vol. XV, no. 1, pp. 63–98. Riad, T. (2000) ‘The Origin of Danish Stød’, in L. Aditi (ed.), Analogy, Levelling and Markedness. Principles of Change in Phonology and Morphology (Berlin/New York: Mouton de Gruyter), pp. 261–300. Riad, T. (2003) ‘Diachrony of the Scandinavian Accent Typology’, in P. Fikkert and H. Jacobs (eds), Development of Prosodic Systems, 91–143. Riad, T. (2006) ‘Scandinavian Accent Typology’, in Ö. Dahl and M. KoptjevskajaTamm (eds), The Typological Profile of Swedish (Special Issue of Sprachtypologie und Universalienforschung), pp. 36–57. Rosser, Z., T. Zerjal, et al. (60 names) (2000) ‘Y-chromosomal Diversity in Europe Is Clinal and Influenced Primarily by Geography, Rather Than by Language’, American Journal of Human Genetics, vol. 67, pp. 1526–43. Salmons, J. (1992) Accentual Change and Language Contact. Comparative Survey and a Case Study of Early Northern Europe (London: Routledge). Sarhimaa, A. (1999) Syntactic Transfer, Contact-induced Change, and the Evolution of Bilingual Mixed Codes. Focus on Karelian–Russian Language Alternation (Helsinki: Finnish Literature Society). Say, S. (2004) ‘Grammaticalization of Word Order: Evidence from Lithuanian’, in O. Fischer, M. Norde and H. Perridon (eds), Up and Down the Cline: The Nature of Grammaticalization. (Amsterdam/Philadelphia: John Benjamins), pp. 363–84. Schmalstieg, W. (2003) ‘Review of Dahl, Östen and Maria Koptjevskaja-Tamm “The Circum-Baltic Languages: Typology and Contact” ’, Acta Linguistica Lithuanica, XLIX, pp. 131–52. Stankiewicz, E. (1993) The Accentual Patterns of the Slavic Languages (Stanford, Calif.: Stanford University Press). Stassen, L. ‘Nonverbal Predication in the Circum-Baltic Languages’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, vol. 2, pp. 569–90. Stolz, T. (1991) Sprachbund im Baltikum? Estnisch und Lettisch im Zentrum einer sprachlichen Konvergenzlandschaft (Bochum: Universitätsverlag Dr N. Brackmeyer). Stolz, T. (2001) ‘On Circum-Baltic Instrumentals and Comitatives: To and Fro Coherence’, in Ö. Dahl and M. Koptjevskaja-Tamm (eds), The Circum-Baltic Languages, vol. 2, pp. 591–612. Tambets, K., Rootsi, S., et al. (43 names) (2004) ‘The Western and Eastern Roots of the Saami – the Story of Genetic “outliers” Told by Mitochondrial DNA and Y Chromosomes’, American Journal of Human Genetics, vol. 74, pp. 661–8. Timberlake, A. (1974) The Nominative Object in Slavic, Baltic, and West Finnic (Munich: Sagner). Torp, A. (1990) ‘Det sogenannte “Garpegenitiv”: Ursprung, Alter und Verbreitung im heutigen Norwegisch’, in L. Elmevik and K. E. Schöndorff (eds), Niederdeutsch in Skandinavien, vol. III (Berlin: Erik Schmidt), pp. 151–66. Torp, A. (1973) ‘Om genitivomskrivninger og-s-genitiv i norsk’, Maal og minne. Oslo: Det norske samlaget, pp. 124–50. Ureland, P. S. (1982) ‘Some Contact-linguistic Structures in Scandinavian Languages’, Language contact in Europe. Working groups 12 and 12, XIIIth International congress of linguists, Tokyo (Tübingen: Max Niemeyer Verlag).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Maria Koptjevskaja-Tamm
The Circum-Baltic Area
Vasiliauskien<. V. (1994) ‘On the Placement of the Genitive Attribute in Old Lithuanian’, Papers from Department of General Linguistics, Umeå University (Umeå: Umeå University), pp. 99–115. Vasiliauskien<. V. (2003) ‘Postposition in the Attributive Phrase of Old Lithuanian’, Zeitschrift für Litteraturwissenschaft und Linguistik, vol. 33, no. 129, pp. 96–108. Viitso, T.-R. (1997) ‘The Prosodic System of Estonian in the Finnish Space’, in I. Lehiste and J. Roos (eds), Estonian Prosody. Wiemer, B. (2004) ‘Population Linguistics on a Micro-scale. Lessons to Be Learned from Baltic and Slavic Dialects in Contact’, in B. Kortmann (eds), Dialectology Meets Typology (Berlin/New York: Mouton de Gruyter), pp. 497–526. Wiik, K. (1995) ‘The Baltic Sea Prosodic Area Revisited’, in S. Suhonen (ed.), Itämerensuomalainen kulttuurialue – The Fenno-Baltic cultural area (Helsinki: Department of Finnish), pp. 75–90. Wiik, K. ‘On the Baltic Sea phonetic area’, in I. Lehiste and J. Roos (eds), Estonian Prosody, pp. 235–50. Zerjal, T., L. Beckman, et al. (5 names) (2001) ‘Geographic, Linguistic, and Cultural Influences on Genetic Diversity: Y-chromosomal Distribution in Northern European Populations’, Molecular Biology and Evolution, vol. 18, no. 6, pp. 1077–8.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
226
Complex Emotion Predicates in Eastern Indonesia: Evidence for Language Contact? Simon Musgrave
In Eastern Indonesia, the area extending from the east coast of Sulawesi in the east to Papua in the west, and from Halmahera in the north to the Lesser Sunda islands in the south, many languages from the Austronesian family are spoken, as are many non-Austronesian languages. These varied languages have been in contact over a long period of time, at least several thousand years, and different ethnic and linguistic groups have held power at various times. In addition to the groups that have been long-term residents of the area, other groups have had an influence more recently: traders from the west of the Indonesian archipelago using Malay as a lingua franca, followed by Portuguese and Dutch colonizers. Given this social and historical background, we would certainly expect to observe many contact phenomena in the languages of this area, and previous work has found that this is indeed the case. Various structural linguistic features have been claimed to have diffused in this area, such as word-order properties (especially clause-final negation), possessive categories and constructions, and discourse structures. Some languages of the region have lexical cognates and grammatical features from both Austronesian and non-Austronesian groups, and assigning a genetic affiliation in such cases can be difficult (see van Staden, 2000: 21–4 for a discussion of the North Halmahera languages). Reesink (1998) suggests that the area, in particular the Bird’s Head peninsula of West Papua, can be treated as a Sprachbund. More recent work by Reesink et al. (forthcoming) suggests that diffusion has occurred from Austronesian to non-Austronesian languages and vice versa, and correspondingly in various geographical directions, especially both east to west and west to east. In this chapter, I investigate one particular areal feature: the use of complex predicates to denote emotional states. The morphological and syntactic properties of this construction are discussed in Klamer (2001). Here, I consider whether the presence of this feature in Austronesian languages can be interpreted as a contact phenomenon and therefore as additional evidence for treating Eastern Indonesia as a linguistic area. 227 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
9
Complex Emotion Predicates in Eastern Indonesia
The evidence examined below will show that this feature almost certainly was not borrowed from non-Austronesian into Austronesian languages. Additionally, I shall use a quantitative technique to compare language samples (Speelman et al., 2003) to test the hypothesis that the relative frequency of use of complex predicates in various languages provides evidence of contact between the two language groups. The statistical evidence of the distribution of complex predicates, compared to the distribution of simple predicates in the same semantic area, does not support the hypothesis that non-Austronesian languages have exerted pressure on Austronesian languages with regard to this feature. Although other studies (Reesink, 1998; Reesink, et al., forthcoming) give some grounds for treating Eastern Indonesia as a linguistic area, I conclude that the feature considered here, complex emotion predicates, does not provide additional evidence for the hypothesis.1
1
Emotion predicates – two construction types
Western Malayo-Polynesian languages typically use simple predicates to denote emotional (and cognitive) states. If there is a second argument (the stimulus), this is usually introduced by a preposition:2 (1) Indonesian Marisa sangat takut kepada-nya Marisa very afraid to.3 ‘Marisa was very afraid of him.’ (Mira, 1995: 259) (2)
Acehnese h’an-galak-geuh keu-u-muda NEG-like-3 DAT-coconut-young ‘He doesn’t like young coconut.’ (Durie, 1985: 62)
(3)
Balinese Cerik-cerik jani lebihan sing demen teken jaja roti child-REDUP now more NEG like with cake ‘Children today more and more do not like cakes.’ (Artawa et al., 1997: 4)
Languages from Eastern Indonesia,3 which are Central Malayo-Polynesian languages, often use complex predicates to denote meanings in this semantic field. These complex predicates consist of a body-part noun, which is often possessed, and an attributive word denoting the state of the body-part noun: (4)
Kambera Mbaha-nanya-ka na eti-na na maramba be.wet-3S.SUBJ-PRF ART liver-3SG.POSS ART king ‘The king is pleased.’ (Klamer, 2001: 98)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
228
Simon Musgrave
Alune lalei-le ‘era le au dana (en)i = tuae inside-NM ill because 1SG take 3SG.POSS.Al = palm.wine ‘He’s upset because I took his palm wine.’ (M. Florey, personal communication)
This type of construction is attested in at least the following Austronesian (afterwards AN) languages of this region: Buru (Grimes, 1992, cited in Klamer, 2001), Tetun (van Klinken, 1999, cited in Klamer, 2001), and Muna (van den Berg, 1989). This complex-predicate construction also occurs in non-Austronesian (afterwards non-AN) languages of the region. For example, in the Bird’s Head peninsula, Hatam, Mpur, Sougb, Meyah and Inanwatan have this construction: (6)
Hatam Ni-ngon sibui sut gi po da di-hig pia le pai pia 3SG-heart twist along NOM ANA I 1SG-ask QUOT then say QUOT ‘She was angry about what I asked (her), (I) said that.’ (Reesink, 1999: 195)
(7) Mpur A-wanken bapak 3SG.M-nose big ‘He is angry’ (Odé 2002: 58) [Lit: He has a big nose. Also used with this meaning.] (8) Sougb Em me-doc ecgu an s/he 3SG-front bad concerning ‘S/he hates s.o.’ (Reesink, 2002: 210) (9) Inanwatan Qótoqowar-e mir-i mé-sowate-bi child-M belly 3SG-good-3SG.M.PRES ‘The boy is glad.’(de Vries, forthcoming, ex. 4.8) (10) Meyah Odou okora rot bua liver hurt concerning you ‘S/he loves you.’ (Gravelle, 2002: 141) And the construction occurs in many languages spoken further east in New Guinea, for example Kalam: (11) Kalam mapm nh liver perceive ‘be sorry’ (Foley, 1986: 118)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(5)
229
Indonesia, showing the location of the languages discussed
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
230
Map 9.1
231
Klamer (2002) considers this construction to be a typical feature of the AN languages of Central and Eastern Indonesia. Therefore it is a tempting hypothesis to interpret this feature as a contact phenomenon; that is, the complexpredicate construction in AN languages is a borrowing from the non-AN languages as a result of long-term contact between the two groups. However, there are various types of evidence that all point to the conclusion that this interpretation would be incorrect, and in the next section I shall present and discuss these. In section 3, I consider whether a weaker hypothesis can be defended, that the complex-predicate construction is more common in Eastern Indonesian languages than in other AN languages as a result of contact with non-AN languages.
2
Evidence against simple borrowing
2.1 The complex-predicate construction is widespread in the languages of the world Collocations of a body-part noun, representing the seat of emotion, plus a qualifying attribute occur in many languages that are widely distributed both geographically and in terms of genetic affiliation. English has examples such as heavy-hearted and light-hearted, and similar examples can be found in most European languages. I offer examples from three further languages from three continents: (12) Hausa, Africa, Afro-Asiatic (Jaggar, 2001: 431) yi farin cik do white.of stomach ‘be happy’ (13) yi bakin cik do black.of stomach ‘be sad’ (14) Kayardild, Australia, Tangkic (Evans, 1995: 198) kurndu-rdami chest-blunt ‘be sad’ (15) bardaka mirra-wa-tha stomach good-INCH ‘feel happy, feel glad’ (Evans, 1995: 297) (16) Lakota, N. America, Siouan bluchalwaxtešni øbyufhat- waxte = šni 3OBJ- 1ACTR- BY.HAND- heart- be.good = NEG ‘I made him/her angry.’ (de Reuse, 1994: 214)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Simon Musgrave
232
Complex Emotion Predicates in Eastern Indonesia
2.2 The complex-predicate construction exists in other Austronesian languages
(17) Indonesian Hati-ku jadi besar liver-1SG become big ‘I am much encouraged’ (LIT: ‘My liver becomes big) Echols and Shadily (1989) list also: (18) Indonesian hati beku hati kecut hati buntu
LIT: liver frozen LIT: liver shrunken LIT: liver blocked
cold hearted cowardly, timid afraid, uneasy
The construction is attested in languages right to the western end of the archipelago, for example in Gayo of northern Sumatra (D. Eades, personal communication). Even more significantly, both the simple and complex-predicate constructions are attested in at least one Formosan language, Tsou. Example (19) below shows a complex predicate, while the embedded clause of example (20) has a simple predicate: (19) Tsou os’o cong’eneni koyu ‘e o’oko’u PF.1SG hurt-BF ear NOM children-1.GEN ‘I feel distressed about my children.’ (Huang, 2002, ex. (3b)) (Lit: ‘My ear aches for my children.’) (20) La’u eainca no koyu’u mita sU’uno HAB-1SG say-PF OBL ear-1.GEN AUX-3SG angry ‘I think s/he is angry’ (Huang, 2002, ex. (3c)) (Lit: ‘My ear says s/he is angry.’) It is very implausible that the complex-predicate construction occurs in Tsou as a result of borrowing, however indirect, from a non-AN language of Eastern Indonesia. Of course, this does not rule out the possibility that the construction was borrowed from another non-AN language closer to hand – for example, a Sinitic language. But given the fact that the accepted reconstruction of the AN family has a number of Formosan groups and Malayo-Polynesian (that is, all other AN languages) (see, for example, Blust, 1977) as the first-level split of Proto-AN, the existence of the construction in a Formosan language and in more than one branch of Malayo-Polynesian suggests that the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Malay has a number of collocations of hati ‘liver’ with attributive words. For example:
233
construction could have already been present in Proto-AN and inherited wherever it occurs in the family. The data given in section 2.1 also suggest that it may be possible that the complex-predicate construction arose as an independent innovation in various braches of the AN family. But the spread within the AN family suggests that the default assumption should be that – unless a convincing case can be made for borrowing in every AN language where the construction is attested – this construction is an inheritance from Proto-AN. 2.3 Lack of direct contact For at least one language with the complex-predicate construction, Kambera, there is no evidence at all that there has ever been direct contact between the speakers of this language and speakers of any non-AN language from Papua4 (M. Klamer, Personal Communication). It is possible that the construction could be borrowed by indirect contact; that is, that change was induced in some AN language or languages by contact with non-AN speakers, and that speakers of Kambera were in contact with speakers of those AN languages and borrowed the construction from them. But this is already a significantly more complex scenario, and, in light of the evidence presented in section 2.2, can be considered unlikely. 2.4 Stimulus arguments and complex predicates Most examples of the complex-predicate construction have only a single argument, the experiencer of the emotion. But examples do occur with two arguments, and this is sometimes treated as an oblique, introduced by an adposition: (21) Buru Da bele-k lale-n tu ringe 3SG be-stupid inside-3 with 3SG ‘S/he is confused with him.’ (Klamer, 2001) The following example is from Saliba, an AN language spoken on the island of Saliba in Papua New Guinea. Although this language is from a different subgroup of AN (Oceanic) and is outside the geographical area of immediate interest here, it shows a similar pattern: (22) Saliba Se-nuwa-mode-mode soba unai 3PL-mind-RED-worry dance PP.sg ‘They are disturbed by the dance/are busy with the dance.’ (Margetts, 1999: 237) Both simple and complex predicates can be seen with oblique arguments in some languages: for example, Hatam. Example (6) above shows a complex
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Simon Musgrave
234
Complex Emotion Predicates in Eastern Indonesia
(23) Lene i-ttei i-ttei y-ug ngat di mai su karena then 3pl-carry 3pl-carry 3pl-go see REL die already because digit tut no-nti ri drui lo mai hak ashamed along 3SG-EMPH GIV startled but die just ‘So, they carried, they carried (her), they went to see the dead, because they were ashamed for themselves, they were startled but she just died.’ (Reesink, 1999: 148) Sougb (Reesink, 2002: 209–10) has complex predicates with direct objects and with prepositional objects. Klamer (2002) argues that the absence of prepositional complements with emotion predicates in the languages of Eastern Indonesia is a typical feature of the area. I have argued elsewhere (Musgrave, 2002a; ch. 4), that the wide distribution of the prepositional complement structure in the AN family, including Formosan languages and plausible traces in Oceanic languages, suggests that this is an old feature in the family. If prepositional complements are an old feature in AN languages, and their absence is a typical feature of Eastern Indonesian languages, then it would seem that the loss of the prepositional complement was an innovation in Eastern Indonesian languages (I am deliberately not attributing this to any particular branch of AN at this stage, as further work is necessary before any solid hypothesis can be offered). In turn, this would suggest that if the apparent blending of the old AN construction type with the Eastern Indonesian construction type were to be analysed as a contact phenomenon, it would have to be the result of some process prior to the loss of prepositional complements, at a time when both constructions coexisted. This situation could have arisen as a result of early contact-induced change in Eastern Indonesia but, as with the data discussed previously, the default assumption must be that this situation arose because both possibilities existed in the relevant proto-language as AN inheritances. The preceding discussion assumes that, in the case of complex predicates, the body part and attributive word function as a single unit that can take an argument in the form of an adpositional phrase, illustrated in example (24a) below. However, this assumption may not be justified. An alternative analysis might be that the body part functions as the head of a noun phrase and any adpositional phrase coding the stimulus is within that NP, as shown in (24b): (24) (a) . . . [ [ [ Attribute] [ BodyPartNoun ]NP ] [PP . . .] ] (b) . . . [ [ Attribute ] [ [ BodyPartNoun ] [PP . . .] ]NP ]
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
predicate with a prepositional stimulus; example (23) below from the same language shows a simple predicate with a prepositional stimulus. The glossing and the free translation of this example do not correspond, but the important point in the example is that the simple predicate digit ‘ashamed’ is followed by the prepositional phrase tut nonti:
235
If (24b) was the correct analysis, then the fact that the stimulus is coded as an adpositional phrase would be in accordance with standard assumptions about the argument-taking possibilities of nouns; that is, that nouns cannot license direct arguments. Any influence of the simple predicate construction type on the complex predicate type could then only be the result of a re-analysis. The sources I have consulted do not give sufficient information to decide which analysis is correct,5 but the argument given previously would still apply: any such re-analysis would have to take place when the simple predicate with prepositional complement construction existed in Eastern Indonesia and the complex-predicate construction was also available. Again, the simplest assumption would be that both possibilities existed in the relevant proto-language as a result of inheritance. 2.6 Summary The evidence discussed in this section makes it clear that it is unlikely that the complex-predicate construction for emotional states in AN languages of Eastern Indonesia has appeared there as a result of contact with non-AN languages of the region. The evidence suggests instead that this construction was a feature of much older AN languages that has been continued in AN languages in Eastern Indonesia and elsewhere, or possibly that it was innovated independently in various places.
3
Quantitative evidence for contact-induced change
The previous section presented evidence to show that the complex-predicate construction for emotional states in AN languages of Eastern Indonesia is unlikely to be a borrowed feature in these languages. Nevertheless, scholars who have studied this area intensively regard this construction as a feature typical of the area (Klamer, 2002; G. Reesink, personal communication). Such a position would suggest that the construction is at least more noticeable – that is, more common – in this region than in other parts of the AN area. The fact that many of the non-AN languages of the region, from the Bird’s Head and from Halmahera (Tidore), also have the construction then suggests another hypothesis: that the complex-predicate construction has always been possible in all AN languages, but has become more prominent in AN languages of Eastern Indonesia as a result of contact with non-AN languages in which such constructions are also prominent. In this section, I explore the possibility of testing this hypothesis against quantitative data. 3.1 Profiles Counting occurrences of a linguistic feature is not always a reliable technique for analysing patterns in data. Frequency counts for most features are very dependent on the genre and subject matter of the texts used as data.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Simon Musgrave
Complex Emotion Predicates in Eastern Indonesia
Comparison of samples, either within a language or between languages, is crucially dependent on matched data being available. In the case of languages with no written tradition, such as most of those discussed in this chapter, matched samples are impossible to obtain. Simply having a large enough quantity of text to make statistical testing meaningful is a major problem, as will become clear below. What is needed in this situation is a technique that compares measures which are established internal to the sample for a single language, and which are therefore not liable to the type of bias just discussed. Such a technique is described by Speelman et al. (2003): the use of profiles to compare language varieties or languages. The technique uses linguistic features with multiple possible realizations as basic data. What is compared is the relative frequency of occurrence of the various realizations in the sample, the profile of the feature in that sample, rather than absolute frequency of occurrence of a single item. The technique was originally applied to lexical variation, in particular the case where a single meaning can be expressed in a language by two synonymous words. Thus Speelman et al. (2003) discuss an example from the Dutch and Belgian varieties of the Dutch language, both of which have the two synonyms, spijkerbroek and jeans for the meaning ‘jeans’, but which have very different profiles for the use of the two words. The profiling technique of Speelman et al. was developed initially for the study of formal onomasiological variation – that is, the situation described in the previous paragraph where one concept can be denoted by two different forms. The technique can easily be extended to cover other types of data; indeed, Speelman et al. (2003) mention their more recent work covering morphological and syntactic variation. I would suggest that the data discussed in this chapter lend themselves to this approach rather well. The key assumption in the profile method is that on any given occasion a speaker can choose one of two possible means of expressing a concept, but that over a large body of data, systematic biases in this choice will become apparent. When the method is applied in this way, a synchronic view of the data is implied: speakers are making individual choices between synchronically present alternatives. In the case considered here, all the languages discussed thus far have both simple and complex-predicate constructions available for the expression of emotional states. Some states are expressed by simple predicates, some by complex predicates, and in some cases, a choice between the two is possible, as in the following examples from Hatam: (25) srek pai leu bi-nhye-pa-ro mpe di i-gga yam slide say from INS-voice-ANA-but time REL 3pl-angry RECIP ‘(she) slid down and said in order for the time they (would be) angry with each other.’ (Reesink, 1999: 149)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
236
Simon Musgrave
237
It would be desirable for the purposes of this analysis to see that every possible concept in this semantic field could be expressed either way; that is, that there is a clear choice on every occasion. However, given the small quantity of data available, this is not a practical requirement; the direct parallel exemplified above is a unique pairing in the data. Therefore it is necessary to take a diachronic view of the data and assume that a choice between the two modes of expression in this semantic field was available to speakers of each language at some time in its development, but that such a choice need not be synchronically available. The diachronic distribution of the two constructions in a language represents the lexicalized result of choices made by generations of speakers in the past. Those choices potentially were influenced by contact with speakers of other languages, and if that was indeed the case, a statistical measure of the similarity of the profiles in pairs of languages (uniformity in the terminology of Speelman et al.) might show the effect of contact. 3.2 Data and discussion Table 9.1 gives a summary of the data used in the current analysis. In all cases (apart from Indonesian), the text sample available is that given in the published (or to be published) grammar of the language. While this means that no attempt could be made to make the samples comparable, this is not a serious problem for the profiling technique, as discussed above. The column headed Frequency per 1k6 words gives an indication of how frequent all emotion predicate constructions are in the sample. These figures show considerable variation between the text samples (mean = 4.8, standard deviation = 0.7), with the values for Muna, Taba, Hatam and Inanwatan all lying more than three standard deviations from the mean. This is not unexpected, given that the texts come from a variety of genres such as autobiography, narrative and procedural text, but I assume that this is not a significant problem for the profile method. Table 9.2 shows the languages arranged in ascending order according to the proportion of complex predicate emotion constructions used, and by geographical position from west to east. This table shows that there is some correlation between the proportion of complex predicates and geographic position, with Tukng Besi an anomaly and Taba and Tetun Fehan having a lower position than might be expected. The case of Taba is particularly striking: given the geographical proximity of Tidore, and other non-AN languages of Halmahera, the contact hypothesis would predict that Taba would show a relatively high proportion of complex-predicate constructions.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(26) Gi-nog lene ni-ngon-ə kinei sut yok NOM-like then 3SG-heart bad along 3PL ‘Therefore she got angry with them.’ (Reesink, 1999: 146)
Relative frequency of complex and simple predicate constructions for emotional states
Language
Affiliation
Words
Indonesiana Tukang Besi Kambera Muna Taba Tetun Fehan Tidore Hatam Inanwatan
AN AN AN AN AN AN Non-AN Non-AN Non-AN
4,755 2,995 626 2,474 1,951 691 10,598 5,026 2,407
Simple predicate 15 11 3 13 19 3 27 5 4
Complex predicate 1 0 1 6 0 0 13 4 2
Frequency (per 1 k words) 3 4 6 8 10 4 4 2 2
Proportion simple/complex (%) 94/6 100/0b 75/25 68/32 100/0* 100/0* 68/32 56/44 67/33
Source
(Online) Donohue (1999) Klamer (1998) van den Berg (1989) Bowden (2001) van Klinken (1999) van Staden (2000) Reesink (1998) de Vries (forthc.)
Notes: a The Indonesian data were collected from a sample of on-line texts generated by a Google search on the term hati. b Both Taba and Tetun Fehan have the complex-predicate construction (J. Bowden, personal communication; Klamer, 2001; van Klinken, 1999), while Tukang Besi does not have the complex-predicate construction (M. Donohue, personal communication).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
238
Table 9.1
Simon Musgrave
239
Proportion of complex predicates
Geographical Position W®E
Tukang Besi Taba Tetun Fehan Indonesian Kambera Muna, Tidore – Inanwatan Hatam
Indonesian Kambera Muna Tukang Besi Tetun Fehan Taba Tidore Inanwatan Hatam
Note: Tukang Besi, Taba and Tetun Fehan all have zero complex predicates in the sample. Tukang Besi was rated lowest, as the language lacks this construction (M. Donohue, personal communication). Taba was rated lower than Tetun Fehan on the basis that the sample for this language was larger.
Speelman et al. (2003) use a descriptive measure of uniformity to compare samples. This measure is based on the log-likelihood test, a statistical test that is appropriate for sparse data with small sample sizes (Dunning, 1993). This measure ranges from 100, indicating that the two profiles are uniform, to zero, indicating that the two profiles have nothing in common. The uniformity figure can therefore be interpreted as a measure of the similarity of the targeted profile in the two samples. In addition to the uniformity measure, it is possible to calculate the probability that the observed similarity is due to chance. As with most statistical tests, the probability is dependent on the sample size. The difference between two very similar samples may be improbable if the measure has been calculated over a large sample. Conversely, and more relevantly in this case, a large difference between two samples (that is, they have a low uniformity score) calculated over a small sample may be attributed to random variation. Table 9.3 presents the results when this measure is calculated for each pair-wise comparison of the languages in the sample (where the result is statistically significant; that is, the probability of the observed uniformity occurring as a result of chance is less than 0.05, the result is in bold face). The crucial data in this table can be seen in the columns for the three nonAN languages. If the frequency of complex predicate emotion constructions in Austronesian languages has been affected by contact with non-AN languages, we would predict that these columns would show increasing values as we read down them. The languages are ordered from most westerly at the top to most easterly at the bottom, and a correlation between geographic proximity to non-AN languages and degree of influence would be predicted. Clearly, this is not the case; the patterns are in fact rather arbitrary,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 9.2 Languages ordered by proportion of complex predicates (low to high) and geographic position
240
Complex Emotion Predicates in Eastern Indonesia
Table 9.3 Profile-based uniformity measures
Indonesian Kambera Muna Tukang Besi Tetun Fehan Taba Tidore Inanwatan Note:
81.25
74.67 93.42
93.75 75.00 68.42
Tetun Fehan
Taba
93.75 93.75 75.00 75.00 68.42 68.42 100 100 100
Tidore Inanwatan
73.75 92.50 99.08 67.50 67.50 99.08
72.92 91.67 66.67 66.67 66.67 66.67 99.17
Hatam
61.81 80.56 87.13 55.56 55.56 55.56 88.06 88.89
Bold type indicates p < 0.05.
although it is important to bear in mind that few of the uniformity results are statistically significant. The alternative hypothesis, that the observed profiles of emotion predicate constructions in the language sample are not the result of contact between AN and non-AN speakers, but rather are the result of independent developments, can also be roughly tested against this data. This can be done by testing the hypothesis that the languages within the AN group are more similar to each other than they are to the non-AN languages. Using only the data points from Table 9.3 that are statistically significant, all of which are among the lower uniformity scores, we can treat the intra-group comparisons (in this case all AN/AN comparisons) as one sample, and the inter-group comparisons (the AN/non-AN comparisons) as a second sample. A t-test applied to this data shows a statistically significant difference between the two; that is, the difference between the two samples is unlikely to be a result of variation within a single population.7 This result is only weak evidence in favour of an independent development scenario, as the same result could be caused by the two language groups having started very far apart for the linguistic feature under consideration, and having moved closer together as a result of contact-induced change without having reached the point where the two groups are indistinguishable for this feature.8
4
Conclusion
This chapter has investigated one particular linguistic feature, constructions used to express emotional states, in AN and non-AN languages from Eastern Indonesia. I have argued that, although there are factors which combine to give an impression that this feature may be evidence for convergence, the data does not in fact support this interpretation. The available evidence is not sufficient to support even the hypothesis that the relative frequency of
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Kambera Muna Tukang Besi
241
use of different constructions has been affected by contact between speakers of languages from the different groups. However, this last conclusion is extremely weak, based as it is on a very small data sample. Statistical techniques have already proved their worth in historical linguistics (see, for example, work on frequency and word-order change in Old English such as Kroch et al., 2000), and their use should be particularly welcomed in the study of contact phenomena. Lexical or syntactic borrowing of completely new items is spectacular and usually clear-cut, but we should expect to find more subtle effects of contact between speakers of different languages which may only manifest themselves in quantitative data. The profile technique of Speelman et al. (2003) is a promising tool for such investigations, but, as well as needing such techniques, research of this type will need access to sufficient quantities of data. At present, this restricts the geographical areas in which statistical techniques can usefully be applied, but a number of current projects offer hope that this situation may improve in the future. These projects include the Spinoza Collection of Areal Linguistic Analyses, a database that will include text samples from a large number of language from several regions, especially the Balkans, Eastern Indonesia and Bolivia/Rondonia (see Musgrave, 2002b, for discussion of an earlier version of the database), and the DOBES project, which is archiving text collections from various languages in different parts of the world (see the website at http: //www.mpi.nl/DOBES/).
Notes 1 The research or this chapter was begun while I was a post-doctoral researcher at the University of Leiden Centre for Linguistics, affiliated with the Spinoza Project: Lexicon and Syntax directed by Professor Pieter Muysken, and the chapter was completed while I was an Australian Research Council Post-Doctoral Fellow at Monash University. An early version of the chapter was presented to the Third Northwest Linguistics Centre International Conference, Manchester, in November 2002. I am grateful to the audience on that occasion for helpful comments, especially David Gil and Claire Bowern. I am also grateful to Margaret Florey and Lourens de Vries for giving me access to unpublished data, to John Bowden and Mark Donohue for answering questions, to Dirk Geeraerts and Dirk Speelman for assistance with statistical analysis, and to Marian Klamer, Anna Margetts, Mark Donohue, Ger Reesink and Yaron Matras for discussion and comments. Remaining errors are my responsibility. 2 For extensive discussion of these predicates in Indonesian, see Musgrave (2002a, ch. 4). 3 See Map 9.1 for the geographical location of the Eastern Indonesian languages discussed here. 4 Mark Donohue (personal communication) suggests that the island of Sumba was probably inhabited prior to the arrival of AN people. Non-AN languages survive on other islands such as Timor and Alor, and it could be a historical accident that no such languages survive on Sumba. If this speculation is correct, there would have been a period in which AN speakers were in contact with non-AN speakers on Sumba.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Simon Musgrave
Complex Emotion Predicates in Eastern Indonesia
5 Reesink (2002: 211) argues that the positioning of verbal morphology in Sougb confirms that the complex predicate is a single unit in that language. See also the discussion in Klamer (2001). 6 1 k = 1,000. 7 The difference is statistically significant for a one-tailed test, which is appropriate in this case as we can predict the direction of the difference between the two samples. 8 Yaron Matras points out that alternative scenarios are possible. Two languages might both have complex predicates as a result of inheritance, and might both acquire a competing structure via internal change. One language might initially replace the inherited construction more radically, while the other language retains a higher proportion of complex predicates. If the profile of the innovative language then converges with that of the conservative language as a result of contact with a third language, then a quantitative measure of uniformity will not reveal much about the actual process of development. This objection is less relevant to the present case, where quantitative analysis supports a negative conclusion, but it raises the more general point that such analysis can only be one technique among many in the study of language contact, and that its results are indicative rather than probative.
References Artawa, Ketut, Put Artini and Barry J. Blake (1997) ‘Balinese grammar and discourse’, Unpublished MS, Udayana University, STKIP Singaraja and LaTrobe University. van den Berg, René (1989) A Grammar of the Muna Language, Verhandeling van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde, 139 (Dordrecht: Foris Publications). Blust, Robert A. (1977) ‘The Proto-Austronesian Pronouns and Austronesian Subgrouping: A Preliminary Report’, Working Papers in Linguistics, University of Hawai’i, vol. 9, no. 2, pp. 1–15. Bowden, John (2001) Taba: Description of a South Halmahera Language, Pacific Linguistics, C-521 (Canberra: The Australian National University). Donohue, Mark (1999) A Grammar of Tukang Besi (Berlin/New York: Mouton de Gruyter). Dunning, T. (1993) ‘Accurate Methods for the Statistics of Surprise and Coincidence’, Computational Linguistics, vol. 19, pp. 61–74 Durie, Mark (1985) A Grammar of Acehnese, Verhandeling van het Koninklijk Institut voor Taal-, Land- en Volkenkunde, 112 (Foris: Dordrecht). Echols, John M. and Hassan Shadily (1989) An Indonesian–English Dictionary (Ithaca, NY: Cornell Univesity Press), 3rd edn. Evans, Nicholas D. (1995) A Grammar of Kayardild (Berlin/New York: Mouton de Gruyter). Foley, William A. (1986) The Papuan Languages of New Guinea (Cambridge: Cambridge University Press). Gravelle, Gilles (2002) ‘Morphosyntactic Properties of Meyah Word Classes.’ In Reesink (ed.), Languages of the Eastern Bird’s Head, pp. 109–80. Huang, Shuanfan (2002) ‘Tsou Is Different: A Cognitive Perspective on Language, Emotion, and Body.’ Cognitive Linguistics, vol. 13, pp. 167–86. Jaggar, Philip J. (2001) Hausa, London Oriental and African Language Library, 7 (Amsterdam/Philadelphia: John Benjamins). Klamer, Marian (1998) A Grammar of Kambera (Berlin/New York: Mouton de Gruyter). Klamer, Marian (2001) ‘Phrasal Emotion Predicates in Three Languages of Eastern Indonesia’, In G. Booij and J. van Marle (eds), Yearbook of Morphology 2000 (Dordrecht: Kluwer), pp. 97–121.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
242
243
Klamer, Marian (2002) ‘Typical Features of Austronesian Languages of Central/ Eastern Indonesia’, Oceanic Linguistics, vol. 41, pp. 363–83. van Klinken, Catharina Lumien (1999) A Grammar of the Fehan Dialect of Tetun, Pacific Linguistics, C-155 (Canberra: The Australian National University). Kroch, Anthony, Ann Taylor and Donald Ringe (2000) ‘The Middle English Verbsecond Constraint: A Case Study in Language Contact and Language Change’, in Susan C. Herring, Pieter van Reenen and Lene Schøsler (eds), Textual Parameters in Older Languages, Current Issues in Linguistic Theory, 195 (Amsterdam: John Benjamins), pp. 353–91. Margetts, Anna (1999) Valence and Transitivity in Saliba, an Oceanic Language of Papua New Guinea, MPI Series in Psycholinguistics, 12). Nijmegen: Max Planck Institute. Mira, W. (1995) Sekelan dendam Marisa (Jakarta: Penerbit PT Gramedia). Musgrave, Simon (2002a) Non-subject Arguments in Indonesian, Ph.D. dissertation, University of Melbourne. Musgrave, Simon (2002b) ‘Inducing Typological Generalizations in a Cross-linguistic Database’, Proceedings of the International Workshop on Resources and Tools in Field Linguistics, LREC 2002, accessed on http://www.mpi.nl/lrec/papers/lrec-pap-21LREC_COL.pdf. Odé, Cecilia (2002) ‘A Sketch of Mpur’, In Reesink (ed.), Languages of the Eastern Bird’s Head, pp. 45–107. Reesink, Ger P. (1998) ‘The Bird’s Head as Sprachbund’, In Jelle Miedema, Cecilia Odé and Rien A. C. Dam (eds), Perspectives on the Bird’s Head of Irian Jaya, Indonesia: Proceedings of the Conference, Leiden, 13–17 October 1997 (Amsterdam: Rodopi), pp. 603–42. Reesink, Ger P. (1999) A Grammar of Hatam, Pacific Linguistics, C-146 (Canberra: The Australian National University). Reesink, Ger P. (2002) ‘A Grammar Sketch of Sougb.’ In Reesink (ed.), Languages of the Eastern Bird’s Head, pp. 181–275. Reesink, Ger P. (ed.) (2002) Languages of the Eastern Bird’s Head, Pacific Linguistics, C-524 (Canberra: The Australian National University). Reesink, Ger P., Marian Klamer and Miriam van Staden (forthcoming) ‘Eastern Indonesia as a Linguistic Area’, in Pieter Muysken (ed.), From Linguistic Areas to Areal Linguistics. de Reuse, W. (1994) ‘Noun Incorporation in Lakota (Siouan).’ International Journal of American Linguistics, vol. 60, pp. 199–260. Speelman, Dirk, Stefan Grondelaers and Dirk Geeraerts (2003) ‘Profile-Based Linguistic Uniformity as a Generic Method for Comparing Language Varieties’, Computers and the Humanities, vol. 37, pp. 317–37. van Staden, Miriam (2000) Tidore: A Linguistic Description of a Language of the North Moluccas Ph.D. dissertation, University of Leiden. Vries, Lourens de (forthcoming) A Short Grammar of Inanwatan, and Endangered Language of the Bird’s Head of Papua (Canberra: The Australian National University).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Simon Musgrave
Another Look at Australia as a Linguistic Area Claire Bowern
1
Introduction1
Australia has been described as an ancient and complex linguistic area, with few recoverable genetic groups (see, for example, Dixon, 1997, 2001, 2002). Evidence cited in support of this areal model includes: the lack of bunching isoglosses for major features such as prefixation, bound pronouns, noun classes and phonological changes; the difficulty of classifying the various Pama-Nyungan daughter languages into discrete subgroups and subfamilies; the apparent time-depth of settlement of the continent; and long-standing, widespread multilingualism. There are, indeed, many isoglosses that criss-cross the continent, and progress has been slow in establishing subgrouping and completing appropriate reconstructions. However, I argue here that contrary to recent high-profile work on Australian languages, Australia is not a ‘special case’ where traditional methods are of no use in reconstruction and classification. Moreover, I hope also to put aside the idea that the use of the terms ‘genetic relationship’ or ‘subgroup’ necessarily implies the ability to model the languages on a neat, preferably binary, branching family tree. Finally, I shall argue that our lack of ability to model a given group of languages on a family tree does not necessarily imply that the group is the result of convergence or is of great antiquity. I begin by giving a brief overview of the assumptions made in the punctuated equilibrium model, and pointing out some problems. In section 3 I summarize the arguments put forward to account for why more traditional models are assumed not to work in Australia, and present some of the major problems in applying punctuated equilibrium. In section 4 I offer an alternative scenario to account for the facts and patterns we see. A case study from Central Australia illustrating the alternative is given in section 5.
2
Overview of punctuated equilibrium
In this section I concentrate on the major features of punctuated equilibrium as a model of language change. I discuss the notions of ‘punctuation’ 244 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
10
245
and ‘equilibrium’ and the problems of applying these notions to language change, and linking language change to non-linguistic events in a systematic way. I address Dixon’s criticisms of the comparative method and the familytree model before moving to the ‘special situation’ of Australian languages. To my knowledge, no detailed critique of punctuated equilibrium specifically applied to Australian languages has yet appeared in print. The reviews of Dixon (1997) were largely favourable, and concerned in the main with the second part of the book (which deals with endangered languages), and in most cases did not address any of the specific claims of the theory. Joseph (2001) was just such a review. The only real exceptions to this are Crowley (1999) and Kuteva (1999), both of which mention instances of mismatches between punctuation and family-tree-like splits, and equilibrium and linguistic diffusion, and Campbell (2003, forthcoming) who provides a general critique.2 The model is also cited favourably by Nettle (1998) and Bellwood (2001), while Schulze (2004) calls Dixon (2002) a ‘thrilling story about the Australian languages’ which ‘will satisfy all the demands for contemporary linguistics when referring to “the” Australian languages’. 2.1 General summary Dixon adapted the term ‘punctuated equilibrium’ from evolutionary biology (and specifically from Eldredge and Gould, 1972) to name his model, in which language change can occur at vastly different rates over different periods of time. In Dixon’s model, language groups are assumed to have coexisted for most of human history without a great deal of disruption. During this time, languages slowly continued to change – largely, Dixon (1997: 71) implies, by borrowing from one another. In this way traits can diffuse across an area. During equilibria, local as well as larger linguistic areas are created, and these cross boundaries of genetic families. The long periods of equilibrium are interspersed with much shorter periods of punctuation. A punctuation might be caused by various different types of event, natural and of human origin. The introduction of new technology, invasion, mass migration, fire or flood are some examples. Dixon (1997: 73) argues that it is only in times of punctuation that recognizable splits – of the kind well modelled by a family tree – occur. It is worth noting that Dixon’s ‘punctuated equilibrium’ does not seem to be a direct adaptation of Eldredge and Gould’s (1972) ideas from biological to linguistic diversification. Eldredge and Gould’s punctuated equilibrium is an alternative, as they write, to ‘phyletic gradualism’. That is, their model (as I understand it) argues for rapid change among small, peripheral populations, some of which may later spread out over a larger geographical area, obliterating other populations and creating what appear to be ‘breaks’ in the fossil record. They are contrasting this to a model in which species mutate gradually over a larger, interconnected population. The rate of speciation is incidental to Eldredge and Gould’s model; moreover, speciation is tied to
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
geographical isolation in a way that it is not for Dixon (1997). In Dixon’s punctuated equilibrium, spread indirectly causes family-tree-like splits (‘phyletic speciation’, in biological terms), or is at least directly correlated with it. Such an idea is completely absent from Eldredge and Gould (1972). Moreover, convergence between different populations/languages, a large factor in Dixon’s punctuated equilibrium, plays no role in Eldredge and Gould’s punctuated equilibrium. The only ‘convergence’ in their theory is the normative influence on mutation of a large, static, interconnected population. With these differences in mind, let us turn now to a more detailed description of Dixon’s (1997) model. First, Dixon claims a direct correlation between patterns of linguistic relationship and non-linguistic, sociohistorical, occurrences; that is, that particular types of human and other activities produce particular types of language split. Second, language change is characterized by long periods of stasis (or ‘equilibrium’) and much shorter periods of punctuation. Equilibrium has been the natural state for most of the world’s languages for most of human history, while the family-tree model is applicable only in the time following a punctuation and language spread. Punctuations are often non-linguistic phenomena; thus (in Dixon’s model) language changes at different rates at different times, but the rate is correlated loosely to whether it is in a state of punctuation or equilibrium. Furthermore, Dixon says that languages in an equilibrium situation will ‘converge to a common prototype’; that is, they will tend to continue to borrow lexical items and grammatical structures until they reach approximately 50 per cent in common (first proposed in Dixon, 1970; and 1972; see also Dixon, 2002: 26–30 et passim). According to Dixon, any lexical or grammatical component of a language can be borrowed, although languages tend to borrow vocabulary more easily than grammatical items. However, he states that there is no universal agreement that basic vocabulary is borrowed less than non-basic (although this may be the case in European languages).3 Therefore, adjacent languages that have little in common are likely to have come into contact with one another relatively recently. Greater weight in the model is given to convergence over divergence (divergence typically happens when languages move apart; convergence happens when they move together); more weight is given to linguistic change through external factors – such as diffusion – than to internally motivated changes. Thus genetic relationship is only one type of linguistic ‘relationship’; areal relationships exist too. Fundamental for punctuated equilibrium is that the family-tree model of linguistic relationship is only applicable in a period following a punctuation; at other times, Dixon argues, there is so much diffusion between adjacent languages that borrowing blurs any recognizable splits. 2.2 General problems with the punctuated equilibrium model The first problem of the notion of ‘punctuation’ seems to me to be that it does not distinguish between the languages of the ‘punctuators’ and those of
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
246
247
the ‘punctuatees’. While all the examples of punctuations and family trees that Dixon (1997) gives are language spreads and therefore punctuators, he does not provide a means of differentiating the two within his model. ‘Punctuator’ languages, for example, can show substratum influences (see, for example, the development of a set of retroflex consonants in Indic languages, or some of the features of Afrikaans that can be attributed to substrate influence from Khoi languages). We also find convergence in cases of punctuation. Examples include the borrowing of conjunctions from Spanish in Native American languages such as Huastec (Thomason and Kaufman, 1988: 80) and Pipil (Campbell, 1987). The languages of the ‘punctuatees’ seem to be assumed to be swallowed up by those of the punctuators, but this does not always happen; not every conquest results in a language shift. Moreover, many cases of convergence are the result of (in Dixon’s terms) short-term punctuation rather than long-term equilibrium. A second, and very important, point regarding the correlation between punctuation, equilibrium and type of language change, is made in Watkins (2001: 55ff) with regard to his discussion of ancient Anatolia as a linguistic area. He points out that while Anatolia is a very good example of a linguistic area, with Indo-European and non-Indo-European languages in contact and each borrowing phonology, morphology and syntax, the languages in question are at the same time undergoing regular (that is, family-tree-modelable) change and speciation. Watkins sees no reason for separating areal and genetic developments, because the two occur concurrently. A similar view is implicit in many other studies which distinguish common descent from parallel innovation and borrowing (such as Lass, 1997; and Sihler, 2000). There is also a considerable literature that deals with diffusion of changes through languages and challenges to regularity in sound change, such as Labov (1981). Third, Dixon associates the creation of language families and rapid splits with non-linguistic punctuation, and much slower change and development with periods of equilibrium, partly with the current number of languages in the world (approximately 6,000); he argues (1997: 29) that a rate of splitting equivalent to that of Indo-European would have given about 1034 languages over the last 100,000 years, and even allowing for language death, this is still many orders of magnitude higher than we find. Therefore, he reasons, Indo-European splitting must be abnormal, and for most of human development over time languages have not been splitting like this, but rather have existed in ‘equilibrium’ with slower rates of change and much more modest splitting. However, Dixon appears to have vastly underestimated the factor of language death in reducing the number of currently spoken languages. He allows for the ‘occasional death along the way’. Take Dixon’s example of a family splitting into two every 6,000 years. If no languages had become extinct
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
over that period, we would expect approximately 217 (or about 132,000) languages to be spoken in the world today, assuming that language has been spoken for approximately 100,000 years. If we assume, however, that the rate of language death is approximately 10 per cent per generation, and use the same rate of splitting, we end up with a figure of approximately 22,000 spoken languages after 100,000 years.4 Just as growth is exponential, so is language death, because in this model when a language ‘dies’ we remove from the calculation any offspring that the node would have had. Assuming a death rate of 20 per cent with that rate of splitting brings us down to a modern figure of just under 3,000 (or 1.617 ≈ 2,950) languages.5 Moreover, language split in the last 10,000 years is not an accurate reflection of the previous 90,000, because of the role of agriculture in boosting population numbers and the social, economic and migrationary possibilities brought about by this. (See the papers in Harris (1996) for examples and discussion of the impact of population figures on language viability; and Nettle (1999) for a more sophisticated mathematical model of language diversity. Pagel (2000) also deals with rates of divergence, though not directly with levels of diversity and the number of languages expected to be extant. Both Pagel and Nettle address the effects of agriculture on population and language numbers.) 2.3 The family-tree model Although Dixon (1997) challenges the applicability of the family-tree model to Australian languages, to my knowledge he does not challenge the principle of the regularity of sound change or the applicability of the comparative method in broad terms, although he claims that the amount of borrowing between Australian languages make its application susceptible to false results. The comparative method is based on the assumption that internal language change is regular, and that regular, systematic correspondences between languages are meaningful and highly unlikely to be a result of chance. What one does with this inference, however, is not strictly part of the comparative method.6 The comparative method relies only on our assumption that languages employ regular sound change, and exceptions to this require special pleading. If Australian languages could be shown to change sporadically, this would mean that the comparative method could not be applied in Australia, although nowhere in Rise and Fall of Languages (1997) (or elsewhere, to my knowledge) does Dixon challenge the idea of regularity of sound change in principle, but he does argue that there are cases where there has been so much borrowing that the ‘correct’ correspondence set will be impossible to determine. Dixon (1997: 149–52) does attack the comparative method as a ‘discovery procedure’, lambasting those (such as Hoenigswald, 1960; and Anttila, 1989) who equate the comparative method with an automatic synchronic phonological analysis. He argues that there are many places where a strict
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
248
249
application of the comparative method will yield the wrong result. He is, of course, right to attack such reconstructions, although one would be hard put to find a thorough, detailed and plausible reconstruction of any family using the comparative method that applied it blindly, without taking into consideration the likelihood of the changes to be reconstructed and the possibility of language contact. In reality, neither synchronic nor diachronic phonology is practised as a mechanical discovery procedure. Problems such as opacity are encountered, recognized and dealt with in both synchronic and diachronic linguistics. Dixon further criticizes the comparative method because it will overgenerate in some cases. For example, if a change occurs in all the languages of a group independently, it may well be reconstructed to the ancestor of the group, giving the false appearance of a common innovation. We cannot reconstruct states for which we have no evidence. This is, of course, a weakness of the comparative method; however, it is a problem with all reconstructions of the past – if evidence has been obliterated, we shall not see it, no matter what method is used, be it carbon-dating, stratigraphy, thermo-luminescence, punctuated equilibrium or glottochronology. This is not a valid criticism of the comparative method per se, it is a fact of prehistory. Dixon’s third criticism of the comparative method is that proto-languages look neat and uniform, whereas ‘natural languages are rather like an old garment that has been patched and mended’ (Dixon, 1997: 45). Dixon implies here (and elsewhere) that the comparative method can only reconstruct regularity, and that we can never reconstruct all aspects of a proto-language. I do not dispute that we shall never be able to reconstruct all aspects of a protolanguage, but it is simply not true that we can only reconstruct regularity. To take some established examples from Indo-European: we can reconstruct certain irregularities in paradigms, such as the presence of full grades in locatives of ablauting root nouns despite the expected zero grade. We can reconstruct a vowel *a distinct from that of the sequence *h2e (that is, the a-colouring laryngeal + e) despite their merger in all daughter languages: compare the behaviour of the Greek root δαμ1ζω ‘I tame’, and its ø-grade passive participle δμ2τ3ς (< *damH-/dmH-tó-), which shows a ‘real’ a which deletes in the zero grade, with the participle of the root *steh2- ‘stand’, that is, στατ3ς, from *sth2-tó-, with vocalization of the laryngeal. In the one case, the laryngeal vocalises (and gives us a vowel /a/), in the other, the original vowel deletes. Another example of reconstructing irregularity involves the remodelling of forms of the nominative case in some paradigms, following Szemerenyi’s Law (involving compensatory lengthening over a resonant as the result of the loss of final *-s). There are numerous other minor ‘rules’ to deal precisely with the irregularities that we can reconstruct. Another ‘reconstruction of irregularity’ that Dixon does not mention is loan-word analysis. It is precisely because these forms are irregular that we can identify them as being worthy of separate explanation (Campbell, 1996,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
makes this point forcefully). It is also possible to reconstruct loans into proto-languages, using with case the same methods we use to identify loans in modern languages. For example, it may be the case that the loan from B to A shows a different reconstructible accent pattern, an odd consonant cluster, a morpheme otherwise unidentifiable, or an odd root shape. Such criteria allow us to infer, for example, that Proto-Germanic *rok(i)ja- (Gothic reikeis, Old High German rihhi, English rich) is a loan from Celtic. Comparison with Latin rex and Sanskrit raj- imply a reconstruction *rjg-, and Celtic exhibits a regular sound change *j > i, but this is not a regular change in Germanic. The irregular correspondence identifies the word as a probable loan. Thus instances where the comparative method fails can be as instructive as those where it succeeds. So, in summary, Dixon’s model would not only have numerous problems in fitting the data as we know it, but his criticisms of the alternative models are not as serious as he would have us believe.
3
Punctuated equilibrium in Australia
3.1 Summary Dixon (1997: 91) regards Australia as a ‘prototypical example of a long-time diffusion area’. Indeed, he states that his punctuated equilibrium model is the only way to account for the current distribution of Australian languages. In this section I outline the main arguments that Dixon presents, applied specifically to Australian languages.7 Dixon’s views of several aspects of Australian languages are at variance with those of most other Australianists (as I understand them), and go some way to explaining his insistence on the lack of applicability of the family-tree model in Australia. The main difference is the claim that Pama-Nyungan languages are archaic in relation to Proto-Australian, and the non-PamaNyungan languages are innovative. This follows in part from the fact that Dixon’s (2002) reconstructions follow general typological principles – that is, that languages tend to go from isolating to agglutinative to inflectional and back again; this is the basis for his view of the non-Pama-Nyungan languages as innovative and the Pama-Nyungan ‘area’ as an archaic diffusion area (see further Dixon, 2002: 696 ff). The Pama-Nyungan type is assumed to be closer to the original type, and the non-Pama-Nyungan languages are assumed to have innovated from that type by cliticizing pronouns or catalysts to verbs to form inflectional bundles, as has also happened sporadically in Pama-Nyungan languages such as the Southern dialects of Baagandji (Hercus, 1986). Dixon also assumes that Australian languages have been evolving fairly much in situ since the initial expansion into Australia (probably around 60,000 years ago). His reasons are partly related to origin myths (see Dixon,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
250
251
1996) and partly because, as he states, if Proto-Australian had been spoken only a few thousand years ago, the split and expansion would have happened rather recently, and a family-tree model should be applicable in that case (Dixon, 1997: 92). Finally, Dixon presents two arguments that Pama-Nyungan languages do not form a genetic family. The first is that if Pama-Nyungan were a genetic group, we should be able to reconstruct innovations between Pama-Nyungan and higher-level groups (see, for example, Dixon, 2001: 93). The paucity of such reconstructions should cause us to question the genetic status of Pama-Nyungan itself. He further argues (in the appendix of Dixon, 2001) that many lexical items taken to be common retentions from earlier stages of Pama-Nyungan (for example, the widespread first-person dual pronoun ngali) are better treated as innovative diffusions across a large area. The widescale diffusion and convergence Dixon argues for is also his explanation for the lack of progress in reconstructing subgroups of Pama-Nyungan and for the non-bunching isoglosses that criss-cross the continent. 3.2 Problems with punctuated equilibrium as applied to Australia Dixon is correct to point out the lack of reliably-established groups and subgroups in Australia; however, there are problems with the arguments he puts forward to claim that traditional methods do not work in Australia. Many previous attempts at subgrouping have been based on lexicostatistics. For example, the O’Grady et al. (1966) classification was based on very little data (a 100-item list) and was never meant to stand as a final classification (Wurm, 1972: 109). It dates from a time when our basic knowledge of Australian languages was still very patchy – indeed, part of its purpose was a ‘stock-take’ of the languages still spoken. It pre-dates almost all full-length reference grammars of Australian languages (see, Koch (2004) for more information). However the O’Grady et al. (1966) classification, in much later Australian comparative work, has either been assumed or rejected out of hand, without a proper evaluation of their groupings by the use of more data. It is sometimes forgotten that it was a provisional schema, with each grouping requiring confirmation or correction by traditional historical-comparative methods. However, until recently, detailed data have not been readily available to allow much accurate reconstruction to take place. Dixon argues against the validity of a reconstructed language Proto-PamaNyungan partly on the grounds that one cannot reconstruct any innovations from a higher proto-language that set it apart from other Australian groups. One must realize, however, that demonstrating genetic relatedness and demonstrating membership of a subgroup are different enterprises; for example, one does not need to believe in and reconstruct Proto-Nostratic in order to define Proto-Indo-European! It is true, of course, that if one views Proto-Pama-Nyungan as a subgroup of ‘Proto-Australian’, to justify this one must provide cognates from Proto-Australian and show innovations to
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
Proto-Pama-Nyungan, but it is not necessary to do this to show the cohesiveness of a Pama-Nyungan family itself. Finally, let us consider the arguments that Dixon uses to claim that Pama-Nyungan is an ancient linguistic area. Part of his argument rests on the use of origin myths and oral history (see, for example, Dixon, 1996) which, if true, would tie certain Pama-Nyungan groups to particular areas for at least the last 10,000 years and further into the Pleistocene epoch. For example, many Indigenous Australian myths speak of times when the sea level was lower than its current level (which last occurred more than 7,000 years ago), and when people could walk between areas now separated by ocean. Dixon writes (2002: 10) ‘all along the east and south-east coasts Aborigines have legends that clearly relate to historical facts (note that not all legends have an historical basis, but a number undoubtedly do).’ It is worth pointing out in general terms some of the problems with using origin myths and other types of myths to date population expansions and other aspects of prehistory. Take as an example the following story about a volcanic eruption, told in Dyirbal to Dixon in the 1960s and quoted from Dixon (1972: 29): It appears that beneath the veneer of fantasy some myths may provide accurate histories of events in the distant past of the people. There is, for instance, a Ngadyan myth that explains the origin of the three volcanic crater lakes Yidyam (Lake Eacham), Barany (Lake Barrine) and Ngimun (Lake Euramoo). It is said that two newly-initiated men broke a taboo and so angered the rainbow serpent, the major spirit of the area . . . As a result, ‘the camping-place began to change, the earth under the camp roaring like thunder. The wind started to blow down, as if a cyclone were coming. The camping-place began to twist and crack. While this was happening there was in the sky a red cloud, of a hue never seen before. The people tried to run from side to side but were swallowed by a crack which opened in the ground . . .’ This is a plausible description of a volcanic eruption. After telling the myth, in 1964, the storyteller remarked that when this happened the country round the lakes was ‘not jungle – just open scrub’. In 1968, a dated pollen diagram from the organic sediments of Lake Euramoo by Peter Kershaw showed, rather surprisingly, that the rain forest in that area is only about 7,600 years old. The formation of the three volcanic lakes took place at least 10,000 years ago. All this points to the story of the volcanic eruptions, and of the spread of the rain forest, having been handed down from generation to generation for something like ten millennia. This is perfectly possible: recent archaeological word suggests that aborigines have been in Australia for at least 25,000 8 years, and the Dyirbal could well have been in more or less their present territory for 10,000 years or more.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
252
253
There are several problems with the assumption that Dixon draws. First, just because people tell a story now and claim it as their own, that should not imply that they are the direct descendants of the people that the story happened to. Dixon gives no dates for population settlements in the areas he talks about, but from Mulvaney and Kamminga (1999: 334) it appears that the area was settled around 5,000 years ago, and extensive exploitation began from around 2,000 years ago, considerably later than the eruption, even allowing for problems in the archaeological record. Furthermore, the events in such stories are often very general and do not need to have been true (for example, seeing an island off the coast, it is very easy to hypothesize that it was once possible to walk to it). The events, even if true, might not refer to the same area. Australia is, after all, a rather large land mass. More important for methodology, for every element in a story that might be true, there is usually a chunk of the story that is discarded (for example, the rainbow serpent in the Dyirbal myth quoted above). This very selective use of data leads us to ignore almost everything that does not fit our ideas of the prehistory of the area and only pick up on the aspects of the myth that fit. To return briefly to archaeology, we do have evidence for settlement beginning well over 40,000 years ago in different parts of the country (Mulvaney and Kamminga, 1999; O’Connor, 1999). However, we also have ample evidence that the current distribution of languages is highly unlikely to represent anything like patterns of original settlement. The archaeological literature is full of descriptions of site abandonment and recolonization over time periods of over 10,000 years (see, for example, Smith et al., 1991). We also have considerable evidence for Holocene epoch expansion, intensification and reoccupation of previously abandoned areas (Veth, 1993; McConvell, 1996, for example). Thus we have no evidence that the distribution of Pama-Nyungan languages reflects the dispersal of the original settlers in Australia, but we do have quite a bit of evidence for a lack of continuity in many parts of the country. Finally, let us examine briefly Dixon’s use of lexicostatistics – that is, the claims that basic and non-basic vocabulary are borrowed equally, and the ‘50 per cent idea’ – that is, that languages tend to borrow back and forth until they have approximately 50 per cent of their lexicon in common. A corollary of this view is that it should not matter what size of list is used for lexicostatistical comparison. In order to test this, I conducted a lexicostatistical experiment using word lists from three north Western Australian languages: Bardi, Yawuru and Karajarri. Bardi and Yawuru are both members of the (non-Pama-Nyungan) Nyulnyulan family, although their territory is not contiguous. Yawuru and Karajarri are contiguous; Karajarri is a PamaNyungan language of the Marrngu subgroup. Speakers of all three languages are in close contact and have been so for a considerable period. The first list contained 100 items of basic vocabulary (the Swadesh list modified for use
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
in Australian languages by O’Grady and Klokeid, 1969). The second list was the first 100 items common to each language in the English-language dictionary, including basic and non-basic vocabulary.9 If basic and non-basic vocabulary are borrowed at the same rate, we should expect few differences in cognate counts between the lists. Furthermore, if basic vocabulary cognacy is an unreliable guide to genetic relatedness (as Dixon, 1997, claims) there should not be a significant difference between related and unrelated languages. The results are given in Table 10.1. In the basic word list, the unrelated languages Bardi and Karajarri share 8 per cent of their vocabulary, while in the non-basic list they share 21 per cent. This is most easily explained by the two languages borrowing culturally specific (non-basic) vocabulary at a higher rate than basic vocabulary. The related languages show the opposite pattern – Bardi and Yawuru share approximately 30 per cent on the 100-word basic list but half of that on the non-basic list. This is again explained most easily by basic vocabulary being retained and non-basic vocabulary being borrowed. In this case, we know that Yawuru has borrowed extensively from Karajarri for terminology for ceremonial items, and flora and fauna; this accounts for the drop between basic and non-basic vocabulary. While a single experiment with so few data points does not prove the point adequately, the findings are suggestive, especially since the Karajarri/Nyulnyulan case is one that Dixon (for example, 2002) uses to support his argument. Let us turn now to the claim that languages borrow back and forth until they exhibit approximately 50 per cent common vocabulary. This figure and the algebra behind it were first published in Dixon (1972: 332–7) as a thought experiment. It is asserted as fact in Dixon (1997: 26) – ‘Australian languages have an interesting property . . .’; and is used as further evidence for an extended period of equilibrium in Australia. The figure of 50 per cent was derived as follows: suppose there are three languages in a row: A, B and C, and each shares 20 per cent of its vocabulary in common with its neighbours. In time T, language B replaces 1 per cent of its vocabulary by borrowing from its neighbours. Supposing that each language borrows about equally frequently from each of its neighbours, about 20 per cent of the 1 per cent of vocabulary lost will be held in common with neighbour A (and similarly for neighbour C), but about half
Table 10.1 Basic and non-basic vocabulary counts between Bardi, Yawuru and Karajarri Basic vocabulary Bardi – Karajarri Bardi – Yawuru
8% 31%
Non-basic vocabulary 21% 15%
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
254
Claire Bowern
255
of the new vocabulary will have been borrowed from A (and similarly for C); thus after time T the new percentage of common vocabulary will be:
Thus the percentage of common vocabulary will slowly rise. Now, assuming that instead of 20 per cent in common the languages had 70 per cent, in this case, using the same reasoning will produce 70 − (0.7 + 0.7) + 0.5 + 0.5 = 69.6% That is, the figure for the common vocabulary will slowly drop; and the figures will converge at approximately 50 per cent.10 The ‘50 per cent figure’ has been discredited comprehensively by Alpher and Nash (1999: 33), who point out that lexical replacement is likely to be largely a factor of internal replacement rather than external replacement by borrowing, and that the equilibrium rate is directly dependent on the amount of lexical replacement that is achieved by mechanisms other than borrowing.
4
An alternative model
I have gone into a discussion of punctuated equilibrium in some detail, in order to describe the model and to summarize the relevant issues in accounting for the current distribution of Australian languages. In this section, I outline an alternative scenario and the accompanying model of linguistic differentiation as it applies to the Australian situation. The ideas on which this model is based have been around for some time, and various authors, including McConvell and Evans (1997), assume something similar to what I am proposing here. The model for differentiation between languages comes almost entirely from Ross (1988, 1996, 1997). However, it is worth spelling out the scenario in detail, and making a number of assumptions explicit, precisely for the reason that this model does seem to be assumed or subsumed under a family-tree model. Many authors (such as Evans, 2004a) still work in terms of a splitting family tree, although they acknowledge that the details of the splits are yet to be fully described. This model is one of language change and classification/genetic relatedness; it is not a one encompassing sociohistorical factors (other than the point that speakers of different linguistic allegiances must remain in contact); thus, while I subscribe to the view (following Thomason and Kaufman, 1988) that the history of a group of languages is a function of the history of its speakers, and does not exist in isolation, I do believe that sociohistorical factors should be modelled independently of their linguistic
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
20 − (0.2 + 0.2) + 0.5 + 0.5 = 20.6%
Australia as a Linguistic Area
consequences; as Kuteva (1999) and Campbell (2002, 2003) have shown convincingly, there is no one-to-one correspondence between non-linguistic events and types of language change. While linguistic history is a function of non-linguistic events, the two can (and should, in my opinion) be modelled separately. I suggest here one possible way in which the languages of Australia, particularly of the Pama-Nyungan group, could be related, and a scenario to account for their current distribution. I am not claiming by any means that it is the only way, but it does fit our current knowledge not only of the Australian prehistoric situation, but also our knowledge of mechanisms of language change elsewhere in the world – in my model, Australia is perhaps an unusual case, but it does not require special pleading. The scenario I propose to account for the lack of clear binary splits in Pama-Nyungan has its basis in dialect geography. The principles of dialect geography are well known from works such as Chambers (1998) and will be repeated here only briefly. The three most important for my purposes are: • speech communities are differentiated by isoglosses; isoglosses tend to bunch along natural barriers to communication but do not necessarily do so; • speech communities in the same geographical area often form chains of mutual intelligibility – that is, adjacent languages may differ minimally but both ends of the continuum may be mutually unintelligible; and • speech varieties at the epicentre of an area tend to be innovative, while isolated conservative pockets exist on the fringes.11 Now consider the relationship between dialects and the family-tree model; it has long been a paradox of genetic linguistics that the relationships between languages may be modelled on a family tree, but the relationships between dialects do not fit a tree so easily. See, for example, Hock (1991: 432ff, esp. 450): The linguistic relationship between neighbouring dialects of the same language very commonly cannot be stated in terms of tree diagrams. This is a consequence of the fact that these speech varieties remain mutually intelligible, stay in close contact, and therefore continue to interact with each other on a day-to-day basis, with shifting realignments as political and social circumstances change. It is therefore unrealistic to expect clear, ‘tree-diagram’ splits in such dialect continua. Next, consider what we would expect to happen if dialectal speech communities continued to diverge, in situ and remaining in linguistic and social contact with one another. I argue that in such a situation we would expect to find a series of dialectal epicentres, with the speech communities around
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
256
257
those epicentres speaking varieties more similar to one another than to varieties of other epicentres; we would expect to find some isoglossic scarps; that is, divisions where isoglosses have accumulated and bunched; we would also expect to find conflicting subgrouping (that is, where isogloss areas overlap). In other words, we would find languages exhibiting traits that would in family-tree terms lead them to be classified in multiple groups. In this scenario, the possibilities for subgrouping would be quite limited, and often conflicting. However, crucially, the languages are still genetically related, in the same way that dialects are without bunching isoglosses. They will still probably exhibit regular correspondences and are still descended from a single parent. There seem to be three major advantages to this model over punctuated equilibrium to account for the distribution of languages in Australia. First, there is a place for both divergence and convergence as processes of language change; punctuated equilibrium stresses convergence as the main mechanism of language change in Australia. Second, it makes Pama-Nyungan look much more similar to other areas of the world. We no longer have to assume that Australia is a special case. Third, and related to this, we do not have to assume in this model that there has been intensive diffusion of many linguistic elements that in other parts of the world are resistant to borrowing (such as shared irregularities). This model does assume that the dispersal of speakers of Pama-Nyungan is relatively recent; not, that is, of the order of 50,000 years or more ago, but perhaps an order of magnitude more recent. This is in accordance with the relevant archaeological evidence, although, of course, linking the spread of Pama-Nyungan with the series of (re)colonizations of abandoned or previously uninhabited areas is thus far an unsupported assumption.
5
Case study: the Lake Eyre Basin
The languages of the Lake Eyre Basin are an excellent case with which to illustrate this model. Most of the languages are quite well described; we have adequate data for synchronic analysis and reconstruction at almost all levels of lexicon, morphology and grammar. Second, we have good archaeological data from a number of sites in the region; we also have extensive data on the trade and mythological networks of the area and their relations with other groups, especially to the west (my data are from McBryde, 1987, and the summary in Mulvaney and Kamminga 1999, as well as unpublished work by Luise Hercus not quoted here). Third, given the lack of consistency in previous classifications, and the fact that Dixon (2001) classifies Karnic as a small linguistic area, the Lake Eyre Basin should be a good area to test the ideas of genetic relatedness in this chapter. The evidence I see in favour of considering Karnic as a genetic subgroup of something (if not Pama-Nyungan in the way it is currently formulated) is
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
258
Australia as a Linguistic Area
5.1 Contact and trade We have solid evidence of well-defined and long-standing contact through trade and ceremony which crosses language boundaries but does follow major isoglosses. Hercus (1980), for example, describes one of the mythological tracks that cuts through the Simpson Desert and goes (among other places) between Arabana country and Lower Southern Aranda. McBryde (1987: 260) gives the major sources of raw materials and trade networks in the Lake Eyre Basin. If Karnic were simply a diffusion area, we would not expect recurrent and regular similarities between languages that are not in frequent direct or semi-direct contact (such as Arabana-Wangkangurru and Wangkumara). We do find them, however. 5.2 Lexical innovation There are innovations in vocabulary between Proto-Pama-Nyungan and Proto-Karnic – that is, there are forms that are widespread in Pama-Nyungan languages (whether genetic or areal terms) that are not shared by Karnic languages: Karnic languages share another term. These include, but are not limited to, the items shown in Table 10.2. From this we can both reconstruct a set of shared innovations between Proto-Pama-Nyungan and Proto-Karnic, and provide evidence independently for the unity of Karnic. 5.3 Shared morphological irregularity Karnic languages share irregularities in certain pronominal paradigms, and these are not shared by other Pama-Nyungan languages in the area. These include suppletive forms in the first person ergative, nominative and dative, stem vowel changes in the third-person masculine singular, and the stem of the second-person dual. The forms are shown in Table 10.3. Another change
Table 10.2 Samples of Proto-Pama-Nyungan and Proto-Karnic reconstructed lexical items Proto-Pama-Nyungan
Proto-Karnic
Gloss
*ngatyi *nguna *ngalirna *kami *kutharra
*kami *parri No category *kanyini *parrkulu
mother’s father lie down 1dl excl. mother’s mother two
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
presented in full in Bowern (1998), and in summary in Bowern (2001). The previous classifications are also summarized in Bowern (2001) and Koch (2004). The evidence is summarized in the following sections. Because of limitations of space, the supporting evidence is not presented in detail here, but has been argued in full in earlier publications.
Claire Bowern
259
Table 10.3 Karnic suppletive pronominal forms
Third-person masculine singular Second person dual
ngathu: Ergative nhulu: Ergative nhula
nganyi: Nominative nhinha Accusative
ngantya Dative
shared by all Karnic languages in the pronouns is the use of a reflex of the Proto-Pama-Nyungan locative in the functions usually associated with the dative. In all languages apart from Arabana-Wangkangurru, this use is also found on nouns (Arabana-Wangkangurru continues the Proto-Pama-Nyungan dative *-ku on nouns). 5.4 Shared phonological aberrancy Karnic languages are the only languages in the area to reflect a voicing contrast in apical stops. This evidence is difficult, since not all languages preserve it clearly. Arabana-Wangkangurru and Pitta-Pitta reflect it as a three-way contrast in rhotics (tap ≠ trill ≠ glide); whereas other Karnic languages, in addition to showing the rhotic distinction also show a voicing contrast in apicals in clusters (nt ≠ nd(r)) as in the Central languages, or at all points of articulation, as in Yandruwandha and Wangkumara. Further details can be found in Bowern (1998: 39–40, 43). 5.5 The diffusion scenario If we assume (as Dixon does) that Karnic is a typical diffusion area with no recognizable genetic subgroups, and most similarities between languages related either to diffusion or inheritance from a common ancestor of the date of Proto-Australian, we have to assume that the following characteristic ‘Karnic’ features have all diffused across areas that were not in close contact but were not borrowed between languages with close contact: • suppletive first person pronouns: ergative ngathu, nominative nganyi, accusative nganha, dative ngantya, with ngantya being found nowhere else in the area; • a shift of the locative case to the meanings usually associated with the dative; and • the change of *d to a tap intervocalically and voicing contrasts in apicals. We should also have to assume that, despite good archaeological evidence to the contrary, trade routes and dreaming tracks have shifted in the near past, bringing Arabana and Lower Southern Aranda speakers closer than in previous periods.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
First person
260
Australia as a Linguistic Area
• the prestopping of sonorants – nasals and laterals in Arabana-Wangkangurru and Adnyamathanha; nasals only in Arandic; laterals and n in Yandruwandha; apical and lamino-dental nasals and laterals in Diyari and Ngamini; • *l > r (a tap) in Arabana-Wangkangurru and Wangkajutjurru (but not its closest relative Pitta-Pitta); • inclusive pronouns have been borrowed from Yandruwandha (only in this language do the phonetic developments make sense) – Diyari borrowed the dual ngaldra and Wangkumara the plural ngandra; • trilled release of apical stops in homorganic nasal clusters in Diyari and Yandruwandha (Yandruwandha also has this phoneme in initial position; cf. rdriya ‘tooth’); and • personal pronouns marked for kin terms (probably from an Arandic or Thura-Yura language).
5.1 Summary Karnic appears to be a fairly typical area within Pama-Nyungan. There is a large amount of recent diffusion identifiable through the usual methods. There are some old dialect areas and some more major isogloss lines that can be used as the basis for genetic classification. There is no need to assume convergence over 50,000 years (as Dixon must do to explain the similarities). It is, in fact, possible to provide a concise explanation for the situation of Karnic by assuming that the modern languages developed out of dialect areas. Such a proposal has been made for the Admiralty Islands by Ross (1988) and for Fiji by Geraghty (1983). In each case the overall patterns appear to be similar; in an examination of the relative chronology of the area, one finds many loan words, some recent irregular phonological borrowings and structural calquing as a result of multilingualism (Ross’s (1998) metatypy), and many old features that, all else being equal, one would reconstruct to a common proto-language, some of which crosses language boundaries. This may well be simply what modern languages that have evolved in situ and in continued contact with their neighbours look like, and there is no need to posit types of changes any different from those familiar from ‘normal’ family trees (although this requires much more investigation). The stages hypothesized are summarized in Table 10.4. A model such as this would result in some major dialectal divisions (a certain number of ‘splits’) but not necessarily intervening proto-languages, and not necessarily clear splits that the family tree implies. This is, indeed, exactly what we find. The difficulties in drawing family trees for dialects are
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
We may contrast this with the patterns of features and general similarities that occur in various different groups in the Lake Eyre Basin. The major ones are listed below (see also Hercus 1972, 1979):
Claire Bowern
261
Stage I Stage II
Stage III Stage IV (Modern attested languages)
Proto-Karnic Isoglosses appeared as speech communities diversified; ‘Common Karnic’ developed dialects. There was movement of speakers, but different dialects maintained in contact More isoglosses built up. Isoglosses show ‘bundling’ but still a great deal of overlapping ‘Dialects’ are mutually unintelligible; certain ‘family-treelike’ splits are found, but there are also a large number of isogloss boundaries which reflect old dialect areas
well attested. Why, then, should it be thought a catastrophe for the method if the family tree does not work well in languages that have evolved from a complex dialect area? Indeed, why should we expect it to work as well in such areas as it does in areas where continued contact has been much less? This casts no doubt on the genetic relationship between the languages, or, ultimately, on our ability to reconstruct the changes in the languages, but it does make it difficult to draw well-defined family trees.
6
Conclusions and further directions
I have shown that the punctuated equilibrium model and the classical familytree models are not the only way to view the history of Pama-Nyungan expansion in Australia. Furthermore, I have concluded that we can assume a date of expansion more recent than initial colonization and still explain the lack of an articulated family tree, and that we can reconcile diffusion areas with recent migration and still talk in terms of a genetic relationship between many of the languages of Australia. The current distribution of Australian languages need not be explained by the heavy diffusion scenario that Dixon argues for, but can be explained within mainstream models of language change. Indeed, I argue that the Pama-Nyungan family can be explained most easily as a relic of a dialect area. This implies that the similarities between Pama-Nyungan languages are a result primarily of common genetic inheritance, and the overlapping isoglosses to a large extent result from either parallel innovation or from old dialect areas. I have also proposed a way to model this relationship. Of course, this model requires intensive testing, both at the level of Pama-Nyungan but also at the level of subgroups of Pama-Nyungan. Work on this is progressing, including the results published in Bowern and Dench (2001) and Koch (2004); other work is in progress. Without detailed, step-by-step reconstruction and plotting of isoglosses, hypotheses of innovations and the like, however, these ideas remain untestable.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Table 10.4 Hypothesized stages in the split of Karnic languages
262
Australia as a Linguistic Area
1 A much shorter version of this chapter appeared as Bowern (2003a). I thank Harold Koch and Barry Alpher for discussion of the material in this chapter, although I alone am responsible for errors of fact or judgement. 2 See also Terrill (2003) for criticisms in relation to the Solomon Islands. 3 Dixon’s evidence for this is lexicostatistical counts on 100-, 200- and 500-item lists for several Australian languages; he reports that it does not matter which word list is used; the results are always the same to within 5 per cent. 4 Assuming a constant rate of death across time/generations and, for the sake of simplicity in calculation, allowing for ‘fractional’ languages, a death rate of 10 per cent gives us a figure of 1.817, or approximately 21,860. 5 There are several reasons why this experiment is not very meaningful. For example, assuming a rate of split of Indo-European as given in Dixon (1997: 29) over the whole world would require a language death rate of 41 per cent per generation to leave us with 6,000 modern languages. That in itself is perhaps not unreasonable, although I have no data on rates of language attrition for the relevant periods, and I do not know how such data would be obtainable. Furthermore, the model in this simplistic form is meaningless. We know, even from looking only at Indo-European, that language families split and diversify at different rates, and as far as I am aware we do not know what causes different rates of speciation; population numbers and density and migration are obviously two factors, as Nettle (1999) shows, but they are not the only factors. 6 Nor, incidentally, is the comparative method necessarily related to the family tree; this point will be taken up further below. The comparative method can still be used, for example, within linguistic areas and in the analysis of loan words (for an Australian example, see Koch, 1997, on loan words between Arandic and Warlpiri), which is what makes it such a powerful tool. 7 One cannot do justice to the full picture in few pages; the reader should consult Dixon (2002) for a much more detailed account. 8 The figure is now 55,000 years+. See, for example, Mulvaney and Kamming (1999), and Roberts et al. (1993). 9 The Bardi data are combined from Aklif (1999) and Bowern (2003b). The Yawuru dictionary is Hosokawa (n.d.). The Karajarri data are from McKelson (1989). 10 Further calculations are made for languages with multiple neighbours, but Dixon (1972) does not model replacement caused by mechanisms other than borrowing. 11 Note that the ‘epicentre’ of a linguistic area need not be the geographical centre of the area.
References Aikhenvald, Alexandra and R. M. W. Dixon. (eds.) (2001) Areal Diffusion and Genetic Inheritance: Problems in Comparative Linguistics (Oxford: Oxford University Press). Aklif, G. (1999) Ardiyooloon Bardi ngaanka: One Arm Point Bardi dictionary. Halls Creek, Western Australia, Kimberley Language Resource Centre. Alpher, B. and D. Nash (1999) ‘Lexical Replacement and Cognate Equilibrium in Australia.’ Australian Journal of Linguistics, vol. 19, pp. 5–56. Anttila, Raimo (1989/1972) Historical and Comparative Linguistics, Current Issues in Linguistic Theory, 6 (Amsterdam: John Benjamins).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Notes
263
Bellwood, P. (2001) ‘Archaeology and the Historical Determinants of Punctuation in Language Family Origins’, in Dixon and Aikhenvald (eds), Areal Diffusion, pp. 27–43. Bowern, C. (1998) ‘The Case of Proto Karnic: Morphological Change and Reconstruction in the Nominal and Pronominal System of Proto Karnic (Lake Eyre Basin)’, BA Hons sub-thesis, Australian National University. Bowern, C. (2001) ‘Karnic Classification Revisited’, in J. Simpson, D. Nash, M. Laughren, P. Austin and B. Alpher (eds), Forty Years On: Ken Hale and Australian Languages (Canberra: Australian National University), pp. 245–61. Bowern, C. (2003a) ‘Another Look at Australia as a Linguistic Area’, Proceedings of the Berkeley Linguistic Society Annual Meeting, 14–17 February 2003 (Berkeley, Calif.: Berkeley Linguistic Society), pp. 55–66. Bowern, C. (2003b) Supplement to Ardiyooloon Bardi Ngaanka: One Arm Point Bardi Dictionary (One Arm Point, Western Australia). Bowern, C. and Harold Koch (eds.) (2004) Australian Languages: Classification and the Comparative Method, Current Issues in Linguistic Theory, 249 (Amsterdam: John Benjamins). Campbell, Lyle (1987) ‘Syntactic Change in Pipil.’ IJAL, vol. 53, pp. 253–80. Campbell, Lyle (1996) ‘On Sound Change and Challenges to Regularity’, in M. Durie and M. Ross (eds), The Comparative Method Reviewed: Regularity and Irregularity in Language Change (Oxford, Oxford University Press), pp. 72–89. Campbell, Lyle (2002) ‘What Drives Linguistic Diversity and Language Spread?’, in Colin Renfrew and Peter Bellwood (eds), Language-Farming Dispersals (Cambridge: McDonald Institute for Archaeological Research), p. 22. Campbell, Lyle (2003) ‘Beyond the Comparative Method?’, in Barry Blake and Kate Burridge (eds.), Selected Papers from the Fifteenth International Conference on Historical Linguistics (Amsterdam: John Benjamins). Chambers, J K. (1998) Dialectology (New York: Cambridge University Press), 2nd edn. Crowley, T. (1999) ‘Review of R. M. W. Dixon 1997, The rise and fall of languages’, Australian Journal of Linguistics, vol. 19, pp. 108–15. Dench, Alan (2001) ‘Descent and Diffusion: The Complexity of the Pilbara Situation’, in Dixon and Aikhenvald (eds), Areal Diffusion, pp. 105–33. Dixon, R. M. W. (1970) ‘Languages of the Cairns Rain Forest Region’, in S. A. Wurm and D. C. Laycock (eds), Pacific Linguistics Studies in Honour of Arthur Capell (Canberra: Pacific Linguistics), pp. 651–87. Dixon, R. M. W. (1972) The Dyirbal Language of North Queensland, Cambridge Studies in Linguistics, 9 (Cambridge: Cambridge University Press). Dixon, R. M. W. (1996) ‘Origin Legends and Linguistic Relationships’, Oceania, vol. 67, pp. 127–39. Dixon, R. M. W. (1997) The Rise and Fall of Languages (Cambridge: Cambridge University Press). Dixon, R. M. W. (2001) ‘The Australian Linguistic Area’, in Dixon and Aikhenvald (eds), Areal Diffusion, pp. 64–104. Dixon, R. M. W. (2002) Australian Languages: Their Nature and Development (Cambridge: Cambridge University Press). Eldredge, Niles and Stephen Jay Gould (1972) ‘Punctuated Equilibria: An Alternative to Phyletic Gradualism’, in T. J. M. Schopf (ed.), Models in Paleobiology (San Francisco: Freeman, Cooper), pp. 82–115. Evans, N. (2004a) ‘Introduction: Comparative Non-Pama-Nyungan and Australian Historical Linguistics’, in N. Evans (ed.), The Non-Pama-Nyungan Languages of Northern
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Australia as a Linguistic Area
Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Canberra, The Australian National University), pp. 2–24. Evans, N. (ed.) (2004b) The Non-Pama-Nyungan Languages of Northern Australia: Comparative Studies of the Continent’s Most Linguistically Complex Region (Canberra: The Australian National University). Geraghty, Paul (1983) The History of Fijian Languages (Honolulu: University of Hawai’i Press). Harris, D. R. (1996) The Origins and Spread of Agriculture and Pastoralism in Eurasia (London: UCL Press). Hercus, Luise (1972) ‘The Pre-stopped Nasal and Lateral Consonants of ArabanaWangganguru’, Anthropological Linguistics, vol. 14, no. 8, pp. 293–305. Hercus, Luise (1979) ‘In the Margins of an Arabana-Wangganguru Dictionary: The Loss of Initial Consonants’, in Stephen A. Wurm (ed.), Australian Linguistic Studies, Pacific Linguistics, C-54 (Canberra: The Australian National University), pp. 621–5. Hercus, Luise (1980) ‘ “How We Danced the Mudlungga”: Memories of 1901 and 1902’, Aboriginal History, vol. 4, nos 1–2, pp. 5–31. Hercus, Luise (1986) The Baagandji Language, Pacific Linguistics, B-67 (Canberra: The Australian National University). Hock, Hans Henrich (1991) Principles of Historical Linguistics, 2nd edn (Berlin/New York: Mouton de Gruyter). Hoenigswald, Henry M. (1960) Language Change and Linguistic Reconstruction (Chicago: University of Chicago Press). Hosokawa, K. (n. d.) ‘Yawuru Dictionary’, AIATSIS (Canberra). Joseph, Brian (2001) ‘Review of Rise and Fall of Languages’, Journal of Linguistics, vol. 37, no. 1, pp. 180–6. Koch, Harold (1997) ‘Comparative Linguistics and Australian Prehistory’, in McConvell and Evans (eds), Archaeology and Linguistics, pp. 27–43. Koch, Harold (2004) ‘A Methodological History of Australian Linguistic Classification and Subgrouping’, in C. Bowern and H. Koch (eds), Australian Languages: Classification and the Comparative Method (Amsterdam/Philadelphia: John Benjamins). Kuteva, Tania (1999) ‘Language and Societies: The “Punctuated Equilibrium” Model of Language Development’, Language and Communication, vol. 19, pp. 213–28. Labov, William (1981) ‘Resolving the Neogrammarian Controversy’. Language, vol. 57, pp. 267–305. Lass, R. (1997) Historical Linguistics and Language Change (Cambridge: Cambridge University Press). McBryde, Isobel (1987) ‘Goods from Another Country: Exchange Networks in the Lake Eyre Basin’, in D. J. Mulvaney and P. J. White (eds), Australians to 1788, vol. III of Australians: A Historical Library (Sydney: Fairfax, Syme and Weldon Associates), pp. 252–73. McConvell, Patrick (1996) ‘Backtracking to Babel: The Chronology of Pama-Nyungan Expansion in Australia’, Archaeology in Oceania, vol. 31, pp. 125–44. McConvell, Patrick and Nicholas Evans (1997) Archaeology and Linguistics: Aboriginal Australia in Global Perspective (Melbourne: Melbourne University Press). McKelson, K. (1989) ‘Karajarri Dictionary’, AIATSIS (Canberra). Mulvaney, John and J. Kamminga (1999) The Prehistory of Australia (London: Smithsonian Institution Press). Nettle, D. (1998) Linguistic Diversity (New York: Oxford University Press). Nettle, D. (1999) ‘Linguistic Diversity of the Americas Can Be Reconciled With a Recent Colonization’, Proceedings of the National Academy of Sciences, USA 96: pp. 3325–29.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
264
265
O’Grady, G. N. and T. Klokeid (1969) ‘Australian Linguistic Classification: A Plea for Coordination of Effort’, Oceania, vol. 39, pp. 298–311. O’Grady, G. N., Stephen Wurm and Kenneth Hale (1966) Aboriginal Languages of Australia (Victoria, BC: University of Victoria). O’Connor, Sue (1999) 30,000 years of Aboriginal occupation. Terra Australis, vol. 14. Pagel, M. D. (2000) ‘The History, Rate and Pattern of World Linguistic Evolution’, in C. Knight, M. Studdert-Kennedy and J. R. Hurford (eds), The Evolutionary Emergence of Language: Social Function and the Origins of Linguistic Form (Cambridge: Cambridge University Press), pp. 391–416. Roberts, Richard, Rhys Jones and Michael Smith (1993) ‘Optical Dating At Deaf Adder Gorge, Northern Territory, Indicates Human Occupation Between 53,000 and 60,000 Years Ago’, Australian archaeology, vol. 37, pp. 58–9. Ross, Malcolm (1996) ‘Contact-induced Change and the Comparative Method: Cases from Papua New Guinea’, In Mark Durie and Malcolm Ross (eds), The Comparative Method Reviewed: Irregularity and Regularity in Linguistic Change (New York: Oxford University Press), pp. 180–217. Ross, Malcolm (1988) Proto-Oceanic and the Austronesian Languages of Western Melanesia Pacific Linguistics, C-98 (Canberra: The Australian National University). Ross, Malcolm (1997) ‘Social Networks and Kinds of Speech-community Event’, in Roger Blench and Matthew Spriggs (eds), Archaeology and Language, vol. 1 (London: Routledge), pp. 209–61. Schulze, Wolfgang (2004) ‘Review of Dixon (2004)’. http://cf.linguistlist.org/cfdocs/ new-website/LL-WorkingDirs/pubs/reviews/get-review.cfm?SubID = 11183 Sihler, A. (1995) New Comparative Grammar of Greek and Latin (Oxford: Oxford University Press). Sihler, A. (2000) Language History: An Introduction (Amsterdam: John Benjamins). Smith, Moya, E. Williams and R. J. Wasson (1991) ‘The Archaeology of the JSN Site: Some Implications for the Dynamics of Human Occupation in the Strzlecki Desert During the Late Pleistocene’, Records of the South Australian Museum, vol. 25, no. II, pp. 175–92. Terrill, Angela. ‘Punctuated Equilibrium in the Solomon Islands’, Terrill: Paper presented at the International Conference on Historical Linguistics, Copenhagen, 10–16 August 2003. Thomason, Sarah Grey and Terrence Kaufman (1988) Language Contact, Creolization and Genetic Linguistics (Berkeley, Calif.: University of California Press). Veth, Peter (1993) Islands in the Interior. International Monographs in Prehistory. Archaeological Series, 3 (Ann Arbor: University of Michigan). Watkins, C. (2001) ‘An Indo-European Linguistic Area and its Characteristics: Ancient Anatolia. Areal Diffusion as a Challenge to the Comparative Method?’, in Dixon and Aikhenvald (eds), Areal Diffusion, pp. 44–63. Wurm, Stephen A. (1972) Languages of Australia and Tasmania. (The Hague: Mouton de Gruyter).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Claire Bowern
Towards a Typology of the Siberian Linguistic Area1 Gregory D. S. Anderson
The Languages of Siberia The roughly forty native languages of Siberia fall into ten small language families and isolates. Despite the great geographic and genetic diversity exhibited among the various indigenous Siberian languages, they nevertheless show a range of features cross-cutting these boundaries. In particular, the languages show a sufficient quantity of logically (and typologically) independent features clustered together macro-areally, coupled with an extensive degree of local bilingualism and language intermixture on the micro-areal level, such that it has become meaningful to discuss a Siberian linguistic area.
(1) Genetic units in Siberia • Tungusic: Evenki, Even, Nanai, Negidal, Orochi, Orok, Udihe, Ul’cha; • Chukotko-Kamchatkan: Al’utor, Chukchi, Itel’men, Kerek†, Koryak; • Eskimo-Aleut: Naukan, Sireniki†, Yup’ik, (+Aleut); • Yeniseic: Arin†, Assan†, Ket, Kott†, Pumpokol†, Yugh†; • Ob-Ugric: Khanty, Mansi (each = several languages/unintelligible dialects); • Samoyedic: Nenets, Enets, Nganasan, Sel’kup, Sayan Samoyed†; • Mongolic: Buryat; • Turkic: Altai, N. (Qumandy, Quu, Tuba), Altai, S. (Altai, Telengit, Teleut), Chulym, Dolgan, Shor, Tofa, Tuvan, Xakas, Yakut [Sakha]; • Nivkh: Amur Nivkh, Sakhalin Nivkh; • Yukaghiric: Chuvan†, Omok†, Wadul [Tundra], Odul [Kolyma]. 266 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
11
267
In the following I present a discussion of various features that make up the constellation of characteristics typifying the languages of the Siberian linguistic area. After offering a brief overview and exemplification of a range of phonological and morphosyntactic features of the languages of the Siberian macro-area I then examine the opposition of both dative and allative case forms, and of instrumental and comitative case forms in the indigenous languages of the region in somewhat more detail. In a series of papers (Anderson 1997, 2001a, 2001b, 2003a, 2003b, 2004b), I have introduced a range of phonological, nominal, verbal, morphosyntactic and syntactic features that characterize the languages of the Siberian area. Space does not permit a full elaboration of all of these, which would require a long, book-length study, but I outline below some of the more salient of these characteristic features.
Figure 11.1
Russia and the languages of Siberia
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
268
The Siberian Linguistic Area
In terms of phonology, Siberian languages exhibit a range of different but commonly shared characteristics. This includes such features as some kind of system of vocalic harmony and the presence of a high back unrounded vowel [m] (sometimes rendered d in certain transcription traditions). In addition, a four-way place contrast of nasals is common across the languages of the macro-area opposing labial, dento-alveolar, palatal and velar places of articulation [m/n/ñ/ŋ] (Anderson 2003a, 2003b). Most Siberian languages have this formal contrast, and in some, it appears to be diachronically stable – that is, it is both an old and a current feature of the languages, as in example (2). This includes languages belonging to the Tungusic, Ob-Ugric and Samoyedic families, as well as the isolates Nivkh and Yukaghir.
(2) Siberian languages with stable contrast of m/n/ñ/ŋ (i) PROTO-TUNGUS *m, *n, *ñ, *ŋ (Cincius, 1949: 180, 183, 192, 203) EVK EVN NEG ORCH ORK UDH ULCH NAN MAN gloss ñuŋun ñuŋen ñuŋun ñuŋun nuŋu ñuŋun ñüŋün ñuŋun ningun ‘6’ moo moo moo moo moo moo moo moo moo ‘tree’ (ii) NIVKH (Panfilov, 1962: 6–7, Gruzdeva, 1998: 24; see Krejnovich, 1937) Amur E. Sakhalingloss Amur E.Sakhalin gloss ñin, ñen ñenŋ ‘one person’ ñdñ ñan ‘one animal’ ŋamk ŋamk ‘seven people’ ŋamk ŋamk ‘seven animals’ (iii) YUKAGHIRIC (Krejnovich, 1958, 1982: 13–14) (Wadul) (Odul) Proto-Yukaghir gloss amun amun *amun ‘bone’ ñaʁa ñaʁa *ñaʁa ‘together’ aŋa-ŋ a ŋa *aŋa ‘mouth’ (iv) SAMOYEDIC (Tereshchenko, 1966a: 371ff) Nenets Enets Nganasan Sel’kup gloss maha maha məku moqal ‘back’ ŋe ŋo ŋuoj – ‘foot, leg’ ñe ne nə (ne) ‘wife, woman’ ñu: ñi ñio/eñi ñóane ñüüñe ‘tadpole’ (v) OB-UGRIC (Tereshkin, 1966: 321ff) Khanty2 yaŋ moni ñoxə sem ‘10’ ‘younger brother’ ‘meat’ ‘eye’
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
1 Select phonological characteristics of the Siberian linguistic area
269
Other languages suggest rather that the opposition is secondary – see example (3) – perhaps because of the long-term presence in Siberia of the languages in question and their subsequent interaction with other indigenous Siberian languages that possessed this opposition. Generally speaking, the most unstable and marked of these sounds in Siberian languages is the palatal nasal ñ; m, n and ŋ are found in all the languages concerned (although with different phonotactic restrictions, see in part below). In the case of the following languages, ñ, and thus the four-way contrast, is either marginal or the result of a relatively recent borrowing or rephonemicization of n in the context of front or high vowels, or through some other, now opaque, cause. Such languages include Buryat, the Yeniseic languages (for example, Ket) and the Chukotko-Kamchatkan languages. (3)
Languages where contrast appears secondary: (i) MONGOLIC (Darbeeva, 1996: 86) Buryat Old Mongol Xalxa ñüden nidün nüd nüdn ñuur niγur nüür nür ñutag nitug nutag nutg ñüsegen nifügün ñücgen nucgn (ii)
(iii)
YENISEIC (Starostin, 1982) Pump Arin Assan Kott ddñe tin tin, tunu ti: ni oñaŋ dña, una kun
Kalmykgloss ‘eye’ ‘face’ ‘homeland’ ‘bare, naked’
Ket dd·ñ ɔʔn ku:ñe
Yugh gloss ddn ‘fir, spruce’ ‘7’ ku:hn ‘wolverine’
CHUKOTKO-KAMCHATKAN (Skorik, 1986: 80, 79, 81) Itel’men Chukchi Al’utor Kerek Koryak gloss ñeñek’efx neneneqey ununuki nananaqi qeyəkmiŋeqey ‘child’
There are also indigenous Siberian languages that seem to have possessed ñ (and thus the four-way nasal contrast) at an earlier period but no longer do so – for example, Eskimo-Aleut and many Turkic languages of Siberia. In Eskimo, *ñ merged with *n in all languages except word-initially in the now extinct Sireniki, where it merged with *y. Common Turkic had *-ñ but this was lost in the history of most modern Turkic languages, where it merged with *-y (or *-n). Some Siberian Turkic languages have had it reintroduced through the means described above or other phonological processes – for example, distant nasal assimilation of or d y to ñ word-initially in North Altai Turkic see example (4ii) below. Some Turkic languages preserve the original Common Turkic *-ñ in non-initial position – for example, Tofa, and, at least in these cases, this language should be included in the group in example (2) above; note that *ñ was lacking word-initially in Turkic.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
(4)
Languages in which contrast is old and/or restructured (i) ESKIMO-ALEUT (Mudrak, 1986: 234–5) PSYE Sireniki Naukan *n n n *ñ y-, -nn PSYE *ñaʁuja *naŋə*ñavə (ii)
Sireniki yaʁə′ja naŋataʁá yavə′tiəx
Naukan naʁújaq naŋáquq na′vəxta: qa:
gloss ‘sea-gull’ ‘come to end’, ‘die’ ‘(ex)change’
TURKIC N. Altai dialects (Baskakov, 1966: 26; 1972: 27) Qumandd-kizi Tuba-kizi Altai-kizi (S. Altai) ñanñand’anñaman ñaman d’aman ñeŋil (ñeŋil) d’eŋil
gloss ‘return’ (intr.) ‘bad’ ‘light, easy’
It is worth at least mentioning that South Siberian Turkic and Eskimo occupy the extreme south-west and north-east peripheral areas of the Siberian linguistic area, and therefore perhaps the loss of this four-way nasal contrast is not all that surprising. With regard to the reintroduction of the four-way nasal contrast in north-eastern Turkic varieties, the influx of ñ-initial words in Yakut and Dolgan has come primarily through loans from Evenki – see example (5) – some with original initial ñ, some shifted in Yakut/Dolgan from n before high or front vowels (whether the latter are found in the loan source or altered secondarily).
(5)
Yakut, Dolgan loans with ññiimde ~ d’iimde ~ ñiimñe ‘reindeer intestinal contents’ (< nimde/nimne)
ñöökü ‘lean-to/shed built in forest’ (< neeku ‘bark-covered lean-to, platform, shed’) ñu(o)rii ñdmaat ‘woman’s decorated riding staff’ ‘old hunters’ tradition of dividing up the spoils from a hunt with one’s neighbours’ (< ñoori) (< nimaat) ñuogu ‘decorated belt for the front reindeer’ (< ñogu ‘first (reindeer)’ A further feature common to indigenous Siberian languages is the presence of the velar nasal sound in onset position. In the languages of northern and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
270
Gregory D. S. Anderson
271
(6)
Languages with word-initial ŋ(i) PROTO- TUNGUS *ŋ- (Cincius, 1949: 239) EVK EVN NEG ORCH ORK UDG ULCH NAN MAN gloss ŋe: le- ŋe: l- ŋe: le- ŋele- ŋele- ŋele- ŋele- ŋele- gele- ‘fear’ (ii) CHUKOTKO-KAMCHATKAN (Skorik, 1986: 79, 85) Chukchi Koryak Al’utor Kerek ŋoyŋən ŋoyŋən ŋoyŋən ŋuyŋən (iii)
gloss ‘tail’
NIVKH (Gruzdeva, 1998: 11–13, 24, 32)
ŋaarla ‘very fat’, ŋaχ ‘soft roe’, ŋdñk ‘face’ (iv)
DOLGAN [Turkic] (Ubrjatova, 1985: 41)
ŋassa ‘pipe’ (7)
Languages with syllable- (and morpheme-), but not word-initial ŋ-: (i) YAKUT [Turkic] (Korkina et al., 1982) silimŋe ‘in the glue; to the glue’ (ii)
KHANTY (Tereshkin, 1966: 224) xotŋon mdsŋdn ‘two houses’ ‘two cows’
For a more detailed analysis of the phonology of nasals in Siberian languages, see Anderson (2003a, 2003b). Certainly, there are a range of other phonological features that characterize the languages of Siberia (or a rather large but geographically definable subset of such languages). One such feature (Anderson, 2001a) is a diachronic process of deaffrication, which changes affricates (usually palatal, sometimes alveo-dental ones) into corresponding (palatalized) stops. This phenomenon may have been operative at the proto-language level – that is, be an old process in the history of a give language or group of languages, as in Northern Yeniseic, Orok and probably also Yukaghiric, or it may be an active idiolectal or dialectally differentiating process that is operating not on the level of diachronic phonology, but rather a quasi-active process of the living language(s) in their current state. Such is the case in various Tungusic languages, local varieties of certain Altai-Sayan Turkic languages,
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
eastern Siberia, one frequently finds syllable-initial ŋ-. This includes both languages that allow ŋ- word initially (Evenki, Chukchi, Nganasan, Nanai and so on) and those that allow this in onset position in medial syllables only – that is, in syllable initial position (and indeed morpheme-initially, sometimes arising through a process of nasal assimilation from g as in Yakut) but not in syllables at word-edge (Yakut, Khanty and so on; see Anderson, 2004a).
272
The Siberian Linguistic Area
2 Select verbal morphology features in the Siberian linguistic area Verb morphology also shares a number of distinctive features across the indigenous languages of Siberia. Among these shared and areally diagnostic features may be included morphological means of encoding desiderative mood – see example (8) – and reciprocal voice example (9)), both of which are widespread in the languages of Siberia. (8) Languages with Desiderative morphology (i) CHUKOTKO-KAMCHATKAN Koryak y(A)- -ŋ ¨ y-afafgañ-n-ək ye-lqəñ-n-ək DESID-laugh-DESID-INF DESID-leave-DESID-INF ‘want to laugh’ ‘want to leave’ (Zhukova, 1972: 202) (ii) TUNGUSIC Evenki Udihe i ə-mu: s aßa-ksa laugh-DESID catch-DESID/ATT ‘want to laugh’ ‘want/try to catch’ (Bulatova and Grenoble, 1999: 10)
xai tukä-mu: i-mi again run-DESID-1 ‘I want to run again’ (Nikolaeva and Tolskaja, 2001: 319)
(iii) ESKIMO-ALEUT -yu- in Yupik məqsyuGaquq ‘S/he wants to drink.’ aglyayuGaqutən ‘You want to go.’ (iv) YUKAGHIRIC Yukaghir (Wadul) met me-yuo-l-bu-d’eŋ I PV-see-DEP-DESID-1 ‘I want to see.’ (Krejnovich, 1958: 125)
Yukaghir (Odul) tet qoŋ-uol’-d’ek you go-DESID-2 ‘You want to go.’ (Krejnovich, 1982: 146)
(v) TURKIC Tuvan ol ulustar fedip keliksep tur(u)gan that people-PL arrive-CV CLOC-DESID-CV AUX-PAST.I ‘They wanted to come.’ (Anderson and Harrison, 1999)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
and Sel’kup. In some instantiations, this deaffrication process has introduced a phonemicization of a palatalized–non-palatalized opposition for alveolar stops into the sound system.
273
(9) Languages with morphological reciprocal: (i) YUKAGHIRIC Yukaghir (Wadul) Yukaghir (Odul) nin-aina: ŋi el’ne-kudede-t∫ eil’i RECIP-shoot-3PLNEG RECIP-kill-[FUT]-1PL ‘They started shooting each other.’ ‘We will not kill each other.’ (Krejnovich, 1958: 120) (Krejnovich, 1958: 120) (ii) MONGOLIC Buryat bari-lda-xa fight-RECIP-FUT.VN/INF ‘to fight each other’ (Darbeeva, 1997: 45) (iii) TUNGUSIC Even ak-nil aw-mat-ta brother-PL wash-RECIP-NONFUT.3PL ‘The brothers washed each other.’ (Mal’chukov, 1995: 14)
Even baka-lda meet-RECIP/SOC ‘meet each other’ (Malchukov, 1995: 14)
(iv) NIVKH v- (Amur) v-/o-/u- (Sakhalin) v-or ‘meet’, v-aγaγ ‘disturb e.o.’ o-smu ‘love e.o.’ (Gruzdeva, 1998; Panfilov, 1962) (v) CHUKOTKO-KAMCHATKAN Chukchi qora-t vetʔatə-velge-rkə-t reindeer-PL butt-RECIP-IMPERF.REALIS-PL ‘The reindeer butt each other.’ (Kämpfe and Volodin, 1995: 99) (vi) TURKIC Tofa karla∫- k9türü∫brush.snow-RECIP lift-RECIP ‘brush snow of e.o.’ ‘lift together’ (Rassadin, 1978: 135)
3
Select syntactic features of the Siberian linguistic area
With regard to the syntax of the indigenous languages of Siberia, the following general statements can be made: in terms of simplex clauses, the basic constituent order Subject Object Verb (SOV) is nearly universal among
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
the languages of the Siberian macro-area. Some are more rigidly verb-final than others, but SOV was the default order attested before the pronounced influence of Russian was made manifest in the colloquial varieties of numerous Native Siberian languages (for representative examples, examine virtually any sentence from any Siberian language offered below). In terms of complex sentence structure, Siberian languages make extensive use of so-called converb forms, or various types of adverbial or nominalized verb formations, as well as the characteristic and common use of case morphology to mark a range of functional types of subordinate clauses. With regard to the latter type of construction, case-marked clausal subordination, there are several basic formal subtypes attested in Siberian languages. In one type, the cases attach to some overtly nominalized form of the verb (participial, infinitival and so on) – see example (10) – while in another, the case may also attach to either a bare stem form, a semiinflected form or even to a finite verb form, as in certain formations in Northern Yeniseic languages. (10) Case-marked clausal subordination with nominalized verb (i) SAMOYEDIC Kamass nere: -bi-ñi nere:-bi-nan be.scared-PRTCPL-1.LOC be.scared-PRTCPL-2.LOC ‘When I am scared.’ ‘When you are scared.’ (Künnap, 1993: 387) (ii)
OB-UGRIC Mansi juw joxt-um-um-t xo:təl-as house arrive-PRTCPL-1-LOC day.dawn-PST ‘When I arrived home, the day had (already) dawned.’ (Rombandeeva, 1993: 299)
(iii)
MONGOLIC Buryat xele-xe-de-m ünenše-xe-güy tell-FUT-LOC/DAT-1 believe-FUT-NEG ‘When I tell (my friends), they won’t believe it.’ (Cheremisina et al. 1984: 156)
(iv)
TUNGUSIC Evenki min-duk pektmre:vun-me ga-na-duk-in bega itten-e-n I-ABL gun-ACC take-PRTCPL-ABL-3 month PASS-NUFT-3 ‘a month had passed since he took my gun from me’ (Nedjalkov, 1997: 51)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
274
Gregory D. S. Anderson
275
(v)
TURKIC Tuvan men kel-gen-im-de azfldaar men I come-PST.PRTCPL-1-LOC work-PRES/FUT 1 ‘When I come (here), I work.’ (Anderson and Harrison, 1999: 73)
(vi)
YUKAGHIRIC Kolyma (Odul) Yukaghir qa:qa:-pe-gi ajli-de-ge ‘el + qon-ŋi-lek’ mon-de-ge grandfather-PL-POSS forbid-3.NF-LOC NEG-go-PL-PROHIB say-3.NF-LOC tamun-gele uørpe-p-ki el + med-o:l-ŋi that-ACC child-PL-POSS NEG-listen-DESID-3PL.INTR ‘Their grandfather forbids (it), saying “Don’t go” but the children do not obey.’ (Maslova, 2003: 372)
(vii)
NIVKH mu aj-ind-ƒtox fχa -imγ-d boat make-MOD-NOMLZR-DAT (=PURP.CONV) money 1-give-FIN ‘He gave me money to make a boat.’ (Gruzdeva, 1998: 51)
(11) Case-marked clausal subordination with (semi-)finite or bare-stem verb: (i) SAMOYEDIC Enets sIraʔ niñ kodia-hað-oñ ŋo-:ñ desumaʔ snow-GEN on sleep-ABL-PX.1sg leg-1SG get.sick-AOR.3SG ‘Since I was sleeping on the snow, my leg got sick.’ (Künnap, 1999a: 35) (ii) CHUKOTKO-KAMCHATKAN Chukchi yəme-γte nelγə-n γəm-nan tə-ttʔə-γʔe-n hang.up-ALL pelt-ABS I-ERG 1SUBJ-knock.over-PERF-3OBJ əweyofγən vessel.ABS ‘When I hung up the pelt I knocked over the vessel.’ (Kämpfe and Volodin, 1995: 106)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
bira dagadun o:-ri-du-v so:t edmni-l-le-n river near become-PRTCPL-DAT-1 very blow.wind-INCH-NFUT-3 ‘When I found myself near the river, a strong wind began to blow.’ (Nedjalkov, 1997: 51)
276
The Siberian Linguistic Area
While one or the other of such case-marked subordination patterns may dominate or be well attested within the structure of a given Siberian language, other subpatterns may also be attested in that grammatical system when viewed as a whole. For example, Chukotko-Kamchatkan, Yukaghiric and Samoyedic languages show not only the more cross-linguistically atypical pattern (cases attached to bare stem or (semi-)inflected form), but they may also show formations of the more common first type with nominalized verb forms, while Turkic, where such nominalized verb forms serve as hosts for a range of case-marked clausal subordination functional subtypes, the second type of construction is also attested to a restricted degree.
4
Secondary ‘areal’ features resulting from Russian contact
As alluded to above, while the diffusion of the features that constitute the diagnostic characteristics of the Siberian linguistic area has occurred over numerous centuries, or even millennia, there has been another, more recent and less lengthy, but nevertheless significant homogenizing force operative among the indigenous languages of Siberia, that in certain speech varieties has given rise to further structural convergence, namely the influence of the socially dominant Russian language, which the vast majority of indigenous Siberians speak fluently. In terms of the syntax of the complex sentence, this has resulted in newly emergent, Russian-type syntactic properties appearing commonly in the languages of the Siberian area. Take, for example, the use of a clause-initial subordinator and a semantically vacuous or scopeless negative operator in the formation of certain kinds of temporally subordinate clauses in a range of modern varieties of various unrelated Native Siberian languages. These mark temporally subordinate clauses of the ‘until’- or ‘before’-type – see example (12). (12) (i) TURKIC Abakan Xakas poka pís par-ba-an-de ib-zer until we go-NEG-PAST-LOC house-ALL ‘Until we came home.’ poka pol-bas-tar soox-tar poka turu-bas-pdn until be-NEG.FUT-PL cold-PL until stand-NEG.FUT-1 ‘Until it gets cold.’ ‘Until I stand.’ (Anderson, 2005)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(iii) YENISEIC Yugh u kidagej ku-daχ-dine :r you here 2-live-ABL ‘Since you lived here.’ (Werner, 1997a: 236)
277
(ii) YENISEIC Baklanixa Ket:(+neg, +subordinator) dìl’get o:l’gə d-a:n-il-de-n as’ka es ben children outside I-play-PST-SF-PL SUBORD sky NEG qonden ban-dina get.dark-INF AUX.N-DAT ‘The children played outside until the sky grew dark.’ (Grišina 1977: 105) Ket [village unspecified] (+neg, +subordinator, loss of dative of auxiliary noun baŋ) as’ka a(t) ben d-ik-si-vis, bu kinil’ bən ɔ-γɔtn SUBORD I NEG 1-PV -PRES-come he from.here NEG I-go ‘Until I come, he won.’t leave from here’ (Kostjakov, 1976: 59) In both of these sets of examples the following observations can be made: there is variation between a mixed structure with a case-marked verb, a clauseinitial subordinator and a scope-less negative operator and a fully finite, case-less form, with the negative and clause initial subordinator. In less Russianized varieties, neither the clause initial subordinator nor a scope-less negative operator is found – see example (13). Note that the Russian-Aleut mixed language Copper Island Aleut (14i) also shows a construction similar to the one of Russian origin, which is given in (14ii). (13)
(i) TURKIC Abakan Xakas min tur-γan-ja I stand-PST-P/E ‘Until I stand.’ (Field notes)
Abakan Xakas soox pol-γan-ja cold become-PST-P/E ‘Before it gets cold.’ (Field notes)
(ii) YENISEIC Sumarokovo Ket: (ostensible original construction) dìl it d-a:n-is-ta-n qon-iy-o-v-γon baŋ-diŋa children I-play PRES-SF-PL get dark STAT-PST-INAN-PST.INCH AUXN-DAT ‘the children played (outside) until it got dark’ (Grišina, 1977: 105) (14)
(i) Copper Island Aleut ya vfera abaa-l poka ni = qaxfakfaa-l I yesterday work-PST until NEG = get.dark-PST ‘Yesterday I worked until it got dark.’ (Golovko and Vaxtin, 1990: 103) (ii) Russian poka my ne priš-l-i domoj until . . .we NEG come-PST-PL homeward ‘Until/before we came (or come) home.’
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
5
The Siberian Linguistic Area
On case morphology in the Siberian linguistic area
Among the most salient features of the systems of nominal inflection attested in the indigenous languages of Siberia is an elaborated system of nominal case. Generally speaking, the total number of cases increases from south and west to north and east, being least developed in various Ob-Ugric and Altai-Sayan Turkic varieties, and reaching their maximal expression in Siberia in Northern Tungusic Even and Evenki, and Chukotko-Kamchatkan Koryak, with well over a dozen case forms. With respect to the categories expressed within the case systems of the indigenous languages of Siberia, in addition to the case oppositions discussed in more details in Sections 5.1 and 5.2 below, one particularly common and typically Siberian case category is a prolative or prosecutive case. This marks motion along or past an object. Prolative case forms are commonly found especially among the languages of the northern and eastern regions of Siberia – see example (15). (15) Languages with prolative case (i) TUNGUSIC Evenki Even oro-r hoktoron-duli: d’uu-li hukti-Yə-tSə: -tin deer-PL path-PROL runIMPF-PST-3PL
house-PROL
d’uu-l-duli house-PL-PROL
‘Deer were running ‘along the house’ ‘along the houses’ along the path.’ (Bulatova and Grenoble, 1999: 10) (Mal’chukov, 1995: 9) (ii) SAMOYEDIC Nenets to-wna lake-PROL ‘along the lake’ (Prokof’ev, 1937a: 26)
Nganasan Enets turku-manu Tau-mone lake-PROL Nganasan-PROL ‘along the lake’ ‘along the Nganasan’ (Prokof’ev, 1937b: 62) (Castrén, 1854: 177)
(iii) CHUKOTKO-KAMCHATKAN Koryak (Skorik, 1986: 95) wayam-gepen ŋawəfŋ-epen yaya-ypen ‘along the river’ ‘past the woman’ ‘past the house’ (iv) ESKIMO-ALEUT Siberian Yupik -kun/-gnəkun/-txun (sg/dl/pl) yuk igləxta-quq sna-kun man go.along-3 bank/coast/edge-PROL ‘The man is going along the edge, bank, coast.’ (Menovshchikov and Vaxtin, 1983: 92)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
278
(v) YENISEIC Ket baŋ-bes ground-PROL
279
Yugh bu lεz-bes ɔ-a-de
Yugh bεiŋ sez-bes ɔŋ-ɔ h: n-de
s/he forest-PROL 3.-PRES.3-go
they river-PROL PL-3. PAST-go
‘along the ground’
‘He is going through ‘They went along the forest.’ the river.’ (Werner, 1997c: 105) (Werner, 1997a: 79)
(vi) TURKIC Xakas kök ds saray üst-ün-je kölefke-le-n-íp par-fatxan blue smoke barn top-3-P/E ring-VSF-RFLXV-CV PRFV.II-IMPERF ‘The pipe gurgled, the blue smoke ringing along the top of the barn.’ (Pritsak, 1959) (vii) YUKAGHIRIC Wadul enu-pul-γan yalγi-pe-γan gorot-qan river-PL-PROL lake-PL-PROL city-PROL ‘along rivers’ ‘along lakes’ ‘about town’ (Krejnovich, 1958: 57)
Odul Omolon-gen Omolon-PROL ‘along the Omolon (river)’ (Krejnovich, 1958: 57)
Note that among non-Samoyedic languages of southern and western Siberia, prolative case is found only in Xakas and Tofa (where it is restricted or vestigial), two languages with known Sayan Samoyedic substrata. In the following sections, I examine two further common features found in the grammars of indigenous Siberian languages. In section 5.1, the characteristic opposition of dative and allative case forms are examined, while section 5.2 presents data on the opposition of comitative and instrumental case forms in the languages of the Siberian linguistic area.
5.1 Dative versus allative case One of the most salient features of the case systems of indigenous Siberian languages is the opposition of a dative and an allative case form. Generally speaking, the former marks recipient arguments, or other roles typically associated with the syntactic relation of indirect object, while the latter frequently serves to indicate motion towards an entity (or location). Both of these case forms are found throughout the Siberian Tungusic languages, and
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
280
The Siberian Linguistic Area
(16) Proto-Tungus DAT versus ALL (i) Evenki nuŋartin bəyə tkə:n-du: oron-mo ani:-ra they boy-DAT deer-ACC give-AOR ‘They gave the boy a deer.’ (Bulatova and Grenoble, 1999: 9) (iii) Evenki tirgaka:kin bira-tki:ollo-mo:-sina-ß noon river-ALL fish-GO-INCEP-1PLEX ‘At noon we went to the river to fish.’ (Bulatova and Grenoble, 1999: 10) (v) Udihe bi sin-du xeleb-wa bu-o:-mi I you-DAT bread-ACC give-PAST-1 ‘I gave you (some) bread.’ (Nikolaeva and Tolskaja, 2001: 524)
(ii) Nanai ogda-du ogda-t∫ i boat-DAT boat-ALL ‘to the boat’ ‘towards the boat’ (Sem, 1997: 184) (iv) Even d’uu-du d’uu-tki house-DAT house-ALL ‘to the ‘towards house’ the house’ (Mal’chukov, 1995: 9)
(vi) Udihe zaŋä-ziga Moskwa-tigi ŋene-zeŋe-ti boss-PL Moscow-ALL go-FUT-3PL ‘The bosses will go to Moscow.’ (Nikolaeva and Tolskaja, 2001: 517)
(vii) Oroch(i) υ gda-du υ gda-ti boat-DAT boat-ALL ‘to the boat’ ‘towards the boat’ (Lebedeva, 1997: 225)
(viii) Negidal min-du: min-tixi: I-DAT I-ALL ‘to me’ ‘towards me’ (Cincius, 1997: 197)
Two comments need to be made on the history and function of this opposition in Tungusic. First, while the dative case indeed marks the recipient argument with verbs such as ‘give’, the allative is found on the recipient of verbs of speaking in Siberian Tungusic languages. Further, while this opposition can be reconstructed to Proto-Tungusic, it is largely lacking in Manchu and in Jurchen, both influenced not insignificantly by Mandarin Chinese, which lacks such a system of inflection. Among Samoyedic languages, the opposition is primarily limited to the so-called Ket dialect of Sel’kup. While a dative–[al]lative opposition is found in a variety of Uralic languages, this particular opposition in Ket Sel’kup appears to be a recent innovation rather than an archaic retention
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
may be reconstructed relatively straightforwardly to Proto-Tungusic based on these correspondence sets. See example (16):
Gregory D. S. Anderson
281
(17) SAMOYEDIC Sel’kup (Ket dialect) kula-ndn versus crow-DAT ‘to the crow’ (Kuper, 1986)
kula-ni crow-ALL ‘towards the crow’
Certain Koryak varieties, as well as Kerek, show an opposition between dative and allative – see example (18i) below. Note that in Chukchi – example (18ii), dative case is limited to pronouns, so the opposition is not well-developed. In Al’utor, the allative is lacking altogether, with allative functions marked by the dative (see example 18iii). The presence of a fullydeveloped opposition of allative versus dative in both pronouns and nouns in Proto-Chukchi-Koryak (or Proto-Northern Chukotko-Kamchatkan) is doubtful. Note in this regard that the divergent Itel’men also lacks a dative case. Therefore, the dative–allative opposition does not appear to be particularly old in Chukotko-Kamchatkan as a whole. However, the Chukchi pattern, as it is fairly specific and difficult to motivate diachronically, may be the original one in Northern Chukotko-Kamchatkan; that is, the opposition was originally found only in pronouns, and later extended to all nouns in Koryak, possibly under Northern Tungusic influence. (18) CHUKOTKO-KAMCHATCKAN (Stebnitskij, 1937: 144), (Skorik, 1986: 93) (i) Koryak wayam-en añpefe-nanan añpefe-nayten [river-DAT] [father-DAT] [father-ALL] ‘to the river’ ‘to (his) father’ ‘towards (his) father’ < weyem ‘river’ < eñpif ‘father’ (ii)
(iii)
Chukchi
əmək-ə əmək-agtə I-DAT I-ALL ‘to me’ ‘towards me’ (Kämpfe and Volodin, 1995: 86) ‘Palana Koryak’ qlegi ʔənpəqlawol-en rara-n bewweevlin drop.by old.man-DAT house-DAT set.off.for.PST.3 ‘Drop by and visit at the old man’s.’ ‘He set off for home.’ (Zhukova, 1980: 48–9)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
from an earlier system. It is mainly lacking in other Sel’kup varieties. See example (17):
The Siberian Linguistic Area
Note that in Koryak both the dative and the allative are dominant or strong suffixes and trigger dominant/strong vocalism in the stem (for example, mil’ut ‘hare’ > mel’otan [hare-DAT]). Old Turkic also possessed an opposition between dative <-GA> and allative <-GArU> cases, the latter perhaps derived secondarily from the former at the pre-Old Turkic stage. The dative case has been preserved intact in all the modern Siberian Turkic languages. On the other hand, the old allative has lost its case function, being preserved in a lexicalized manner in only a small number of adverbial expressions – for example, Xakas tasxar ‘to the outside’, Isker ‘to the east’ (Anderson, 1998: 13). However, two Turkic languages have reintroduced the formal opposition into the case system, namely Tuvan and Xakas of the Altai-Sayan region of south-central Siberia. In the former, an earlier equative-cum-prolative case has taken on the function of the allative, while the latter innovated a new form completely, representing a grammaticalization and subsequent fusing of an earlier ‘auxiliary’ or relational noun meaning ‘side’. See example (19): (19) TURKIC (i) Tuvan sen klub-ce bar-ba-an-dn-da, men baza bazdn-ga olur-ayn you club-ALL go-NEG-PAST-2-LOC I also house-DAT sit-1.INT ‘If you’re not going to the club, I’m going to stay home too.’ (Shamina, 1993: 65) (< equative case) Tuvan xoy börü-ge fi-dir-t-ken sheep wolf- DAT eat- CAUS- (CAUS)-PAST ‘The sheep was eaten by the wolf.’ (Shcherbak, 1977: 68) (ii)
Xakas ol pís-ke fooxta-an s/he we-DAT say-PAST ‘He told us.’ (Field notes) Xakas min íje-m aγdr-fatxan üfün klub-sar par-bas-pdn I mother-1 be.ill-PRS.PRTCPL for club-ALL go-NEG.FUT-1 ‘Because my mother is sick I’m not going to the club.’ (Cheremisina, 1995: 20) (< ‘side’)
In fact, various Tuvan dialects show another, different new allative form. This appears to have arisen from a relatively recent fusing of a postposition. The suffix takes the shape -DIvA, and examples include daγ-ddva ‘towards the mountain’ and ot-tuva ‘towards the fire’ (Anderson and Harrison, 1999).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
282
Gregory D. S. Anderson
283
(20) NIVKH (1) Nivkh tha dkdn-doχ thaxta-ya NEG e.b.-DAT be.angry-IMP ‘Don’t be angry at (your) elder brother.’ (Gruzdeva, 1998: 20)
(ii) Nivkh in-doχ ph-vo-x thamdid xer-ya we-DAT REFL-village-LOC/ABL what tell-IMP ‘Tell us what is (going on) in your village.’ (Gruzdeva, 1998: 21)
(iii) Nivkh (iv) Nivkh i eri-rχa vi-nd-dj-ra t∫-dtk haimŋaf-toγo hunv-nd-ra I river-ALL/TERM go-FUT-FIN2-father old. live-FIN-PRED PRED age-ALL/TERM ‘I shall go (up) to the river.’ ‘Your father lived (in)to old age.’ (Gruzdeva, 1998: 21) (Gruzdeva, 1998: 21) The following observations can be made regarding the distribution of the opposition of dative and allative case forms among the languages of Native Siberia. This contrast can be reconstructed to Proto-Tungusic but is lacking in the Tungusic languages, which have undergone extensive influence from Chinese, that is, those outside the Siberian area. Further, it appears to have been an archaic feature in Turkic, albeit one in which the allative formally appeared to have been derived from the dative. This old contrast was lost in most Turkic languages but reintroduced in individual Turkic languages later through various independent and different developments. In the isolate language Nivkh, the comparative picture is, unsurprisingly, difficult to assess, as the opposition appears in both main dialects of the language and thus appears to be old, but like Old Turkic, the allative appears to be an augmented form of (or derived from) the dative. Among western and central Siberian languages, apart from the Turkic languages just mentioned, the opposition appears to be restricted to a single dialect of Sel’kup (where Evenki influence is possible). Finally, in the Chukotko-Kamchatkan languages, we find the following situation: in Koryak and the recently extinct Kerek, one finds the opposition in varying degrees throughout the nominal system, while in Chukchi, this is largely lacking except with pronominals, and in Al’utor and Itel’men entirely so. The influence of Northern Tungusic (Even and/or Evenki) is a possible cause of its appearance in ChukotkoKamchatkan languages, in particular Koryak, which would then reflect a secondary, rather than a primary, opposition of the case forms in question.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
In the isolate language Nivkh (Gilyak) of south-eastern Siberia and Sakhalin Island, there is also an active opposition between formal markers of dative and allative case. As in the case of Old Turkic, it appears that the allative form may represent an augmented or modified form of the dative case. See example (20):
284
The Siberian Linguistic Area
Another common feature seen in the grammars of indigenous languages of Siberia is the contrast of an instrumental and a comitative case. The former marks instrumental roles in the sentence, while the latter marks accompaniment. Both of these can be reconstructed for Northern Tungusic. The instrumental is actually found in all Siberian Tungusic languages, however, a comitative case form is found only in the Southern Tungusic language Orok, with a probably non-cognate form – see example (21). Otherwise, the instrumental has comitative functions in the Tungusic languages, e.g. Udihe – example (22). (21) TUNGUSIC (i) Evenki si: tara bər-it-pi: asi: kiŋnə:-l- i ami:n-dula: -βi: you that gun-INS-REFL woman ski-PL-INS father-LOC-POSS garpa-kal is-fa: -n shoot-FUT.IMPER.2SG go-PST-3 ‘Shoot that one with ‘The woman on the skis went up your gun.’ to her father.’ (Bulatova and Grenoble, 1999: 9) (ii)
(iii)
(iv)
Evenki bi: əkin-nu:n-mi: təßlə:-m I sister-COM-REFL.SG collect.berries-1SG ‘I went with my sister to pick berries.’ (Bulatova and Grenoble, 1999: 12) Even d’uu-t∫ house-INS
d’uu-nun house-COMIT (Mal’chukov, 1995: 9)
Orok
ŋInda-ndo dog-COMIT ‘together with the dog’ (Novikova and Sem 1997: 212)
versus
Orok a: pυ n-d i hat-INS ‘with the hat’ (Novikova and Sem 1997: 212)
(22) Udihe Udie tue-ze-zi sita-na-zi-fi bagdi-e-ti Udihe winter-N-INS child-PL-INS-REF.PL live-PST-3PL ‘The Udihe lived in winter huts with their children.’ (Nikolaeva and Tolskaja, 2001: 118)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
5.2 Instrumental versus comitative case
285
Note that a comitative case form is found in Solon (Cincius, 1997a), where it has the clearly non-cognate shape -gili:, but it is lacking in Negidal (Cincius, 1997b), which is geographically in the Southern Tungusic region, but linguistically more Northern-like. While an instrumental–comitative opposition may well thus be an old one in Proto-Tungusic, the particular forms in Evenki and Even (and Solon as well) suggest that the formal realization of this opposition might in fact be a Northern Tungusic innovation. Orok might be an archaic retention – formally or functionally speaking – of an earlier Proto-Tungusic opposition, or an unrelated development. Note that the comitative case is only possible with animates in Orok; however, Northern Tungusic shows no such restriction. Resolving the details of these historical issues awaits further research. In Chukotko-Kamchatkan, the instrumental–comitative opposition is clearly reconstructable in form and function to the proto-language level. In fact, in opposition to the instrumental there are two comitative forms, called variably the first and second comitative, or the sociative and the comitative. While the instrumental consists of a typologically and areally typical case suffix, the comitative/sociative forms on the other hand use a circumfix. In Chukchi, this consists of a ga- prefix followed by the stem, followed by the instrumental suffix for the first-person comitative/sociative, or by the suffix -ma for the (second-person) comitative. In Koryak and Koryak-like varieties, one finds rather a variety of prefixal elements; for example, (g)awun- ~ gA- in the comitative and gA- ~ gAyqə- in the sociative (PsCh-Ko = Proto-Chukchi-Koryak) see example (23): (23) CHUKOTKO-KAMCHATKAN Chavchuven Kamen Palana Chukchi PCh-Ko INSTR -(t)A -(t)a -(t)A ~ -(t)a -(t)A *-(t)A COMIT-1 gA-..-(t)A ga-..-(t)a gA/a-..-(t)A/a gA-...-(t)A *gA-...-(t)A COMIT-2 gawun-..-ma gawun-..-ma awun-..-ma ga-..-ma *ga-..-ma (Stebnitskij 1937: 187)
INS/ERG
SOC COM
Koryak Kerek -te, -nek, yëk -ta, -nak, -yik g(e)- . . -(t)e g(a)- . . -ma
Chukchi Al’utor Itel’men -k, -ne, -rëk- ta, -nak, -l’ -tëk(N=)g(a)-..-(t)a -g(e)-- . . -(t)e g(a)--(t)a k-/x- . . -L’ g(a)- . . -ma -g(a)- . . -ma g(a)- . . - k- . . -famxma . . fom
(Skorik, 1986: 93) When data from Itel’men is considered, it appears that the comitative, sociative and instrumental forms must all be reconstructed to the ProtoChukotko-Kamchatkan level, although the details of the history of the
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
suffixal element in the (second-person) comitative remains to be worked out. Functionally speaking, the sociative is generally used with animate referents, the comitative with inanimates (for example, in Koryak (Zhukova, 1972: 120). Note that, when used with inanimates, the opposition is one of degrees of connectedness or comitativity, with the sociative marking a greater degree of connectedness. Compare the Koryak sentences in example (24iii–iv) below. (24) CHUKOTKO-KAMCHATKAN (i) Koryak ga-kayŋ-a gayqe-kayŋ-a gawen-kayŋd-ma COM-bear-SOC COM-bear-SOC COM-bear-COM ‘with the bear’ (Zhukova, 1972: 107) (ii) Koryak geyqe-miml-e ga-ŋavdqqal’u-ta gawen-meml-ema COM-water-SOC COM-girl-SOC COM-water-COM ‘with water’ ‘with a little girl, daughter’ ‘with water’ (Zhukova, 1972: 120) (iii) Koryak efgi ənno geyqe-milgəʕəy-e kuʕeqevəŋ nota-ytəŋ today he COM-gun-SOC went tundra-ALL ‘Today he went to the tundra with his gun.’ (Zhukova, 1972: 121) (iv) Koryak ənno gawen-melgəʕəy-ma gapəlqalin he COM-gun-COM drowned ‘He drowned with his gun.’ (Zhukova, 1972: 121) (v) Chukchi ənpd-nafg-ərgəna-t qorat old.man-PL reindeer-PL ‘the old men’s reindeers’ (Skorik, 1986: 107)
ga-npənafg-ərgəna-qora-ma COM-old.men-reindeer-COM
‘with the old men’s reindeers’
A few comments need to be made regarding the use of the sociative and comitative in Chukotko-Kamchatkan languages. First, as the second set of Koryak examples show, the comitative and sociative suffixal elements belong to different harmonic classes and therefore trigger different forms of the stem. Also, the (second-person) comitative form functions as a phrasal-type circumfix
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
286
287
in large compound structures, see the Chukchi form in example (24v). The history of these three case forms is not certain, but may be as follows. The instrumental appears to be older in the language (for example, it has innovated a grammatical ergative function) with the comitative forms innovated secondarily. Possibly the sociative is historically an instrumental case form of a derived (nominal) form in ga-. The comitative, on the other hand, may be derived from a similar form followed by the postposition omakaŋ. Note that in Koryak there is generally no comitative form of personal pronouns, and omakaŋ is used instead (with a locative form of the pronoun) – see example (25): (25) Koryak gəm-ək omakaŋ I-LOC together.with ‘with me’
muy-ək omakaŋ we-LOC together.with ‘with us’ (Zhukova, 1972: 45)
There are also predicative functions of the comitative and sociative cases in the formation of certain kinds of adverbial subordinate clauses in ChukotkoKamchatkan languages – see example (26): (26) Chukchi ŋewəsqet ga-gəntow-ma kulil’ərʔu-gʔi woman COM-run.away-COM cry.out-PERF ‘The woman cried out while running away.’ (Kämpfe and Volodin, 1995: 54) These are likely to be secondary formations in the history of ChukotkoKamchatkan, after the development of the case forms with nominals. Also, case marked clausal subordination may reflect an areally diffused property in this family of languages. In Yukaghiric languages, both an instrumental and a comitative case form are attested. These show clear cognacy across both Odul and Wadul, and can be reconstructed easily for Proto-Yukaghiric. Note that, regarding the comitative, the default uses of this in Yukaghir include marking possession and conjoining two nouns – see example (27): (27)
YUKAGHIRIC Yukaghir (Krejnovich, 1982: 45, 46, 47) (i) Yukaghir (Wadul) -lek, -leŋ sa-lek pajduk stick-INS hit ‘hit with a stick’
Yukaghir (Odul) -le, -lek foγoye-le knife-INS ‘with a knife’ (Krejnovich, 1982: 49–50)
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
(ii)
(iii)
Yukaghir (Odul) núme-ñej dwelling-COMIT ‘with a dwelling, he has a dwelling’ (Krejnovich, 1982: 45) Yukaghir (Wadul) ile-ñej la:me me-qaldej-ŋi reindeer-COMIT dog PV-run.off-3PL ‘The dog ran off with the reindeer’, ‘The dog and the reindeer ran off.’ (Krejnovich, 1982: 46)
As is well known, both Khanty and Mansi are not single languages but rather clusters of related dialects, and each probably constitutes three or four separate languages. The dialect/language variation situation is quite complex with regard to the case opposition under consideration. According to Honti (1998: 343), comitative and instrumental were both present in the Proto-Ob-Ugric language, but the opposition was an innovation at this level. Note that, in general, there is considerable variation among the number and type of case forms found among the Ob-Ugric languages. Mansi varieties tend to have six to seven, but Khanty ‘dialects’ range from three to eleven distinct case forms. Generally, we find the comitative–instrumental opposition in eastern Khanty varieties and the (now often extinct) southern varieties of Mansi. Compare, for example, the following situation in eastern (Vakh, Vasyugan) Khanty and Tavda Mansi – see example (28). Note that Sosva Mansi lacks the comitative form. (28)
OB-UGRIC E. Khanty COM -naat/-näät INS -(t)ə/ b
Tavda Mansi -naat/-näät -(t)əl
This correspondence suggests a fairly straightforward reconstruction to Proto-Ob-Ugric of both the functional opposition of instrumental and comitative, as well as formal markers encoding this contrast. In terms of the origin at the Proto-Ob-Ugric level, it is often suggested that the comitative derives from a either a pronominal base or a fused postpositional element. One western Khanty variety, Sherkal, in fact shows a postpositional construction for the comitative with pronouns, for example, maa naataaŋeem (Honti, 1998) where the first syllable of the postpositional element is cognate with the comitative case suffix in eastern Khanty and Mansi. Also, some Mansi varieties use a cognate form of the instrumental case, but use a postposition jot for the comitative function (Riese, 2001: 26).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
288
Gregory D. S. Anderson
289
(29)
SAMOYEDIC (i)
(ii)
(iii)
Sel’kup (Ust’-Ozornoe) golaj ü: da-n bare hand-INS ‘with bare hands’ (Tjuxterevo) tobə-n leg/foot-INS ‘with his foot’ (Karelino) tobe-t leg/foot-INS ‘with his foot’ (Bekker, 1978: 136)
(iv) Sel’kup (Laskino) versus ad i-n-opti father son-GEN-COMIT ‘the father together with this son’
Ust’-Ozornoe tan-opti YOU-COM ‘(together) with you’ (Bekker, 1978: 139)
The historical situation is significantly more complicated than it would at first appear. First, the instrumental has the appearance of the genitive, and it is not clear whether these case forms are to be considered to be historically separate or connected. Second, Nenets has a postposition/adverb ŋsbt ‘together’ (Bekker, 1978: 140) which appears to be cognate with the Sel’kup element. Third, most Sel’kup dialects have innovated a new instrumental/ comitative case form, possibly from a fusing of another, different adverb/ postposition that might historically derive from a non-finite form of ‘be’ – that is, ‘being’ > INS > INS/COM; this may in fact be a Southern Samoyed innovation, as a cognate element seems to have existed in the extinct Kamas (Künnap, 1971). An example of a Sel’kup form with the new instrumental/comitative is (Ust’-Ozornoe) inne-za-t [brother-INS/COM-PL] ‘with the (five) brothers’ (Bekker, 1978: 144).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Various Sel’kup dialects make use of an instrumental–comitative opposition. The comitative may sometimes attach to a genitive, and not a nominative, form of the stem, suggesting a possibly recent origin in a particular postpositional or auxiliary noun formation – see example (29):
The Siberian Linguistic Area
The instrumental–comitative opposition is otherwise generally lacking in the case morphology of Samoyedic languages – for example, Nganasan has a comitative but no instrumental, while Kamas[sian] has an instrumental but no comitative (Simoncsics, 1998; Künnap, 1971). The opposition is indeed also lacking in various dialects of Sel’kup (Helimski, 1998a). According to the latter researcher, the instrumental–comitative opposition outlined above in Sel’kup is mainly found in lexicalized expressions, and the active case systems employ a single instrumental/comitative/ sociative form. A comitative case form is found in numerous attested modern Uralic languages – for example, Saami (Sammallahti, 1998); Estonian (Viitso, 1998); Nganasan (Helimski, 1998b); or Mari (Kangasmaa-Minn, 1998), but instrumental cases per se are not overly common in Uralic. An opposition of instrumental and comitative is found, however, in Komi (Hausenberg, 1998) and Komi-Permyak (Riese, 1998), but Udmurt (Csúcs, 1998) lacks a comitative, while the divergent Yaz’va dialect of Komi lacks an instrumental (Riese, 1998).3 The Ob-Ugric developments were discussed above. Unfortunately the details of the case systems of the various daughter languages of Proto-Uralic in general, and of the Samoyedic languages in particular, is highly complex and space does not permit us to pursue these issues to the degree of specificity that would be necessary here. The Mongolic language Buryat is another Siberian language that exhibits the areally common opposition between an instrumental and a comitative case form.
(30)
MONGOLIC Buryat mal-aar mal-tai cattle-INS cattle-COMIT ‘with the cattle’ ‘together with the cattle’ (Poppe, 1960: 39)
This is an old feature in Buryat, as formally and functionally cognate elements are found in most other Mongolic languages – for example, Classical Mongolian (Orlovskaja, 1997); seventeenth-century literary Oyrat (Yaktonova, 1996); Bao-an (Todaeva, 1997a); and Dagur (Todaeva, 1997b). Within the languages of the Turkic family, only the most north-eastern languages show such an opposition. In Dolgan, there are two variants of the comitative case, in contrast to a single instrumental case form. In the case of the so-called second comitative, this is an element used historically to mark attributive or possessive adjectives (cf. Yukaghir), still used as such in various other Turkic languages of Siberia. Dolgan’s close sister-language
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
290
Gregory D. S. Anderson
291
Yakut (Sakha) has this opposition between comitative and instrumental as well, with formally cognate elements – see example (31): TURKIC (i)
(ii)
Dolgan munu ikki ilii-tinen kusput this.ACC two hand-3.INS grab-PAST.II ‘He grabbed this with both hands.’ (Ubrjatova, 1985: 121) Dolgan oγo-luun beye-liin child-COMIT self-COMIT ooññu-ur play-PRES ‘He himself is playing with the child.’ (Ubrjatova, 1985: 122)
(iii)
Yakut (Sakha) Alasov iye-tiniin kel-le (~-neen) Alasov mother-3.COMIT come-PAST ‘Alasov came with his mother.’ (Ubrjatova, 1985: 124)
kddh-a kihi-leek olor-or daughter-3 person-COMIT.II sit-PRES
‘His daughter is sitting with the person.’
Alasov emeexsin-niin Alasov old.woman-COMIT kel-li-ler come-PAST-3PL ‘Alasov came with the old woman.’ (Ubrjatova, 1985: 123)
(iv) Yakut (Sakha) kini, araaha, eye-nen tönnör ete he PRTCL peace-INS (re)turn-P/F SBJNCTV ‘He, by all appearance, would return peacefully.’ (Petrov, 1984: 49) According to Tenishev et al. (1988) the opposition of instrumental and comitative is an old one in Turkic, to be reconstructed back to Proto-Turkic. According to this belief, the old instrumental -In is preserved as such mainly in lexicalized adverbial expressions throughout the modern Turkic languages (for example, Turkish gündüzün ‘during the day’ Tofa kdhdn ‘during the winter’), while the earlier comitative, a fused version of the postposition ‘(together) with’, has various reflexes in the modern Siberian Turkic languages as an instrumental case (for example, Xakas -naŋ), including Yakut and Dolgan. The opposition of instrumental and comitative is an areal feature of the languages of Siberia. It is found in Northern Tungusic, although here possibly renewing an opposition found as far back as Proto-Tungusic, and in
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
(31)
The Siberian Linguistic Area
additional possibly influenced by Yukaghir, with which Northern Tungusic speakers have had a long interaction. The opposition is a well-developed one in Proto-Chukotko-Kamchatkan. Correspondences among Yukaghiric varieties suggest that this is an old feature in this language group as well. In the Turkic varieties possessing it, it is unclear if this is a partial innovation, renewing an old Turkic opposition based on contact with Northern Tungusic, or whether these peripheral Turkic languages with the opposition are archaic in this regard. In other Turkic languages, the old comitative has become an instrumental case, while the old instrumental case is found lexicalized in various adverbial expressions. Other elements, including ones used to mark possessive adjectives, were seemingly co-opted into a comitative function. In Mongolic, it is clearly an old feature, and likely to be reconstructed back to Proto-Mongolic. The most problematic groups are the Uralic languages of Siberia. It is found only in certain varieties of the Ob-Ugric languages Khanty and Mansi, but the correspondences suggest that it goes back to their common ancestor, Proto-Ob-Ugric. In Samoyedic, it appears only in certain Sel’kup dialects and thus appears to be a relatively recent innovation, and at any rate appears to be marginal in the system. As with the previous opposition, the Sel’kup data may also reflect Tungusic influence in this instance, but Eastern Khanty influence is in fact more likely as a source for this innovation in Sel’kup.
4
Summary
From a macro-areal perspective, despite the obvious fact that the indigenous languages of Siberia exhibit considerable genetic and typological diversity with respect to one another, they nevertheless possess a cluster of features that pattern with one another but are not logically or typologically connected. These include features of the phonology, systems of nominal and verbal morphology, and the syntax of the simple and complex sentence. With regard to nominal morphology, two characteristic features of case systems commonly attested in the languages of Siberia were discussed in some detail above. These include, on the one hand, an opposition between dative and allative case forms, and on the other, a formal contrast between instrumental and comitative case functions. In the first case (the dative–allative opposition), the feature clusters primarily around languages that have had significant and prolonged interaction with Tungusic languages (apart from Turkic, where the opposition is clearly old). In the case of the instrumental–comitative opposition, the directions of influence are more complex. Some groups clearly reflect an old opposition (Yukaghiric, Chukotko-Kamchatkan, Mongolic), while with others (for example, Tungusic, Turkic, Ob-Ugric) the situation is less clear. Northern Tungusic languages might reflect Chukotko-Kamchatkan influence, but Yukaghiric influence is perhaps more likely in this instance (large numbers of Yukaghiric speakers moved to Northern Tungusic). The north-eastern
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
292
293
Turkic varieties, on the other hand, may well reflect secondary and later Northern Tungusic influence, albeit reinforcing a potentially archaic contrast. The situation with the western and central Siberian languages is also not clear at present. Ob-Ugric seems to have innovated this contrast fairly early, at the proto-language level; however, its trigger is currently opaque. Sel’kup is even more confusing as the opposition is quite recent, and Khanty influence is possible as an explanation, but this is far from certain.
Table 11.1 List of morphological features, grouped by family DAT: ALL
NAUKAN SIRENIKI YUPIK AL’UTOR CHUKCHI ITEL’MEN KEREK KORYAK EVEN EVENKI NANAI NEGIDAL OROCHI OROK UDIHE UL’CHA YUKAGHIR NIVKH BURYAT KET ALTAI, N ALTAI, S CHULYM DOLGAN SHOR TATAR, SIB TOFA TUVAN XAKAS YAKUT ENETS NENETS NGANASAN SEL’KUP MANSI KHANTY
− − − − N−/P+ − N−/P+ + + + + + + − + + − −13 − − − − − −17 − − −21 + + − − − −25 +26 − +/−28
INS: COM
PROL
RECIP
DESID
−4 − − + + + + + + + − − − + − − +12 − + − − − − +18 − − − − − + − − − + − +/−1
+ + + + −7 −9 − + + + −11 + + + + + + − − + − − − − −20 − +22 − −23 − + + + + − −29
− − −5 +6 +8 + + + + +10 + + + + + + + + +14 − + + + + + + + + + + − − − − −27 −30
+ + + + + + + + + + + + + + + + + − − − − −15 −16 −19 − − + + −24 − − − + − − −
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
As is probably obvious from the present discussion, the features of the Siberian linguistic macro-area cluster around those of the Northern Tungusic languages and this is not by accident. Indeed, the highly mobile Evenki (and to a lesser degree its sister-language, Even) both have the local bilingualism relationships and widespread distribution necessary to make them likely vectors of diffusion for at least some of these features, whether they are older Tungusic features (the dative–allative contrast) or seemingly later innovations (the instrumental–comitative opposition). However, Tungusic > non-Tungusic is in no sense the only direction of influence apparent in these developments, but rather one in a highly complex mosaic of linguistic interactions operative over centuries and millennia across the languages of the macro-region. Certainly, an understanding and elucidation of the multifaceted dynamics of diffusion and borrowing evidenced by the distribution of these and numerous other potential areal features unfortunately still remain in their infancy. Further insights into the complex histories of the case systems and other features of the languages of the Siberian linguistic area must await future research.
Table 11.2 List of select phonological and syntactic features Four-way nasal contrast [m: n: ñ: ŋ] Opposition is old and stable Opposition is old but lost Tungusic Yupik Nivkh Turkic (most) Yukaghir Samoyed Ob-Ugric I
ŋ- (initial velar nasal) Stable feature
Marginal feature
Tungusic Nivkh Chukotko-Kamchatkan N. Samoyed (some)
N. Samoyed (some) Dolgan Yupik Kott
Opposition is new Chukotko-Kamchatkan Yeniseic Mongolic Turkic (some)
II
Word-internal (morpheme-initial) only Yakut Khanty Selkup Yupik
De-affrication (Anderson, 2001a) Old (proto-language/ Dialectal (idiolectal or shallow historical) historical) process variation Ket, Yugh Sel’kup Orok Ob-Ugric Yukaghiric ?? Altai-Sayan Turkic Northern Samoyed Tungusic III
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
294
Gregory D. S. Anderson
295
Case-marked clausal subordination + nominalized verb form + partially inflected verb stem Turkic N. Samoyed Mongolic Yeniseic Ob-Ugric Nivkh Tungusic Yukaghiric Chukotko-Kamchatkan Samoyedic + bare-stem Chukotko-Kamchatkan Yupik Yukaghiric
+ finite verb form Yeniseic (Ket, Yugh)
Notes 1 An earlier version of this chapter was presented as a paper at the Language Areas conference in Manchester. My thanks to audience members for valuable discussion at that time, and to the helpful comments of three anonymous reviewers. All errors remain the author’s. 2 Note that some Khanty dialects have a retroflex nasal as well, so there is in fact a five-way contrast. 3 Komi is in many ways linguistically a peripheral member of Siberian linguistic area, and there are in fact Komi dialects found in Siberia at the present time. A fuller demonstration of this requires a complete study in its own right and is not pursued further here. 4 INS but no COM case; there are a variety of COM-type formants, however. 5 COM meaning of -lgutke- (de Reuse, 1994, Central Siberian Yupik); the lexicalized postbase -ute seems to have a RECIP/DISTRIB meaning; for example, pellghute ‘part from e.o.; divorce’. 6 RECIP and COM/DISTRIB functions realized by separate markers. 7 Note the noun derivational (/adverbial?) suffix -jIKWI; for example, ret-jek wə ‘along the road’. 8 Analyzed by Bogoraz as a RECIP, but by Skorik as not fully grammaticalized and meaning ‘collide with’. 9 The affix -xʔal/-xol functions as an ABL and a PROL. 10 RECIP and COM/DISTRIB are morphologically distinct maf/t versus -ldd. 11 Preserved in various lexicalized adverbs, (Sem, 1997). 12 Postpositional clitic -ñe/-ñeŋ/-ñej/-nd’e/-ño COM versus INS -lek. 13 Note DAT-ACC -(a)x versus DAT-ALL -Tox. 14 RECIP and COM/DISTRIB may be marked by differing formal means; there are two possibly related affixes -lda- and -lsa-. 15 -ZA in lexicalized stems suuza ‘thirst’ (‘want water’). 16 Optional meaning associated with OPT -Gay. 17 DAT is -GA in impersonal declension and -Gar in ‘personal-possessive’ inflections. There is also an enclitic -dek (cf. diegi ‘side’) with an ALL-type function.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
IV
The Siberian Linguistic Area
18 COM is rare in Dolgan. 19 -IAK baar means ‘wanted to X but didn’t’. 20 -fa is more of an ALL-PROL in function, cf. Tyvan -že/fe ALL. 21 Old ALL preserved in adverbial forms; for example, dnaard ‘to there’ mdnaard ‘to here’ oŋγaard ‘to the right’. 22 Quasi-adverbial -šA has function of Russian po; for example, PROL or ‘according to’, etc. 23 The case/adverbial affix -fA has among other meanings a PROL-type one; for example, folfa ‘along the road’, firfe ‘along the land’. 24 Optional function of -Gay (OPT); also lexicalized forms such as suxsa ‘thirst’ found in almost all Turkic languages. 25 DAT-ALL -tə/-nda versus ALL-TERMINAT -dya. 26 In Ket dialect of Sel’kup -nyŋ DAT versus -ni ALL kulanyŋ ‘crow-DAT’ versus kulani ‘crow-ALL’. 27 In N. Mansi dialects (Rombandeeva, 1973), -xat-/-axt- RFLXV = RECIP too. 28 In Eastern Khanty dialects (Vax-Vasjugan, Surgut), there are two so-called INS and two ALL (-nV/-tV (INSTR) -A/-pA (ALL) and -nat/-at (INS) -a/-nam (ALL), respectively). 29 In Vakh Khanty the ABL and PROL functions are fulfilled by a single form. 30 According to Steinitz (1937, 1950), the RFLXV affixes -ant- and -ës- may also be used in RECIP functions.
References Abondolo, Daniel (ed.) (1998) The Uralic Languages (London: Routledge). Abondolo, Daniel (1998) ‘Khanty’, in D. Abondolo (ed.), The Uralic Languages, pp. 358–86. Alkor, J. A. (ed.) (1937). Jazyki i pis’mennost’ narodov severa, vol. 1 (Moscow/Leningrad: Ufpedgiz). Anderson, G. D. S. (1997) ‘Towards an Areal Linguistic Typology of Native Siberia’, Paper Presented at the NSL-10 Conference, Chicago. Anderson, G. D. S. (1998) Xakas. Languages of the World/Materials, 251 (Munich: LINCOM EUROPA). Anderson, G. D. S. (2001a) ‘Deaffrication in the (Central) Siberian Area’, in Howard I. Aronson (ed.), Non-Slavic Languages 9: Linguistic Studies (Columbus, Oh.: SLAVICA), pp. 1–17. Anderson, G. D. S. (2001b) ‘Language Contact and Macro-Areal Typological Change: Complex Sentence Structure in Siberia/Northern Eurasia’. Paper presented at the 23rd Jahrestagung der Deutschen Gesellschaft der Sprachwissenschaft, Leipzig, February, 2001. Anderson, G. D. S. (2003a) ‘Yeniseic Languages from a Siberian Areal Perspective’, in Edward Vajda and Gregory D. S. Anderson (eds), Studia Yeniseica Sprachtypologie und Universalienforschung, vol. 56, nos 1/2, pp. 12–39. Anderson, G. D. S. (2003b) ‘Towards a Phonological Typology of Native Siberia’ forthcoming in D. A. Holisky and K. Tuite (eds), Current Trends in Caucasian, East European and Inner Asian Linguistics. Papers in Honor of Howard I. Aronson (Amsterdam: John Benjamins). Anderson, G. D. S. (2004a) ‘Areal and Phonotactic Distribution Of ŋ’, in M. van Oosterdorp and J. van de Weijer (eds), The Internal Organzation of Phonological Segments. Studies in Generative Grammar (Berlin/New York: Mouton de Gruyter).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
296
297
Anderson, G. D. S. (2004b) ‘The Languages of Central Siberia: Introduction and Overview’, in Edward J. Vajda (ed.), Languages and Prehistory of Central Siberia (Amsterdam: Benjamins), pp. 1–123. Anderson, G. D. S. (2005) Language Contact in South Central Siberia, Turcologica, 54 (Wiesbaden: Harrassowitz). Anderson, G. D. S. and K. David Harrison (1999) Tyvan. Languages of the World/ Materials, 257 (Munich: LINCOM EUROPA). Bekker, È. G. (1978) Kategorija padezha v sel’kupskom jazyke (Tomsk: Izdatel’stvo Tomskogo universiteta). Benzing, J. (1955) Lamutische Grammatik (Wiesbaden: Franz Steiner Verlag). Benzing, J. (1956) Die tungusische Sprachen: Versuch einer vergleichenden Grammatik. (Wiesbaden: Franz Steiner Verlag). Bogoraz, W. (1922) ‘Chukchee’, in F. Boas (ed.) Handbook of American Indian Languages (Washington, DC: Bureau of American Ethnology), pp. 631–903. Bulatova, N. Ja. (1980) ‘O formakh naklonenij glagola v govorakh evenkov Amurskoj oblasti’, in Voprosy jazyka i fol’klora narodnostej severa (Yakutsk), pp. 36–41. Bulatova, N. Ja. and L. Grenoble (1999) Evenki. Languages of the World/Materials, 141 (Munich: LINCOM EUROPA). Castrén, M. A. (1854) Grammatik der samojedischen Sprachen, reprinted as Uralic and Altaic Series 53, 1966 (Bloomington, Ind.: Indiana University Press). Castrén, M. A. (1857) Versuch einer koibalischen und karagassischen Sprachlehre, A. Schiefner (ed.). (St. Petersburg: Kaiserliche Akademie der Wissenschaften). Castrén, M. A. and T. Lehtisalo (1960) Samojedische Sprachmaterialen. Mémoires de la Société Finno-Ougriènne 122. Cheremisina, M. A. (1995) ‘Osnovnye tipy analitifeskix konstrukcii skazuemogo v tjurkskix jazykax juzhnoj Sibiri’, in M. N. Cheremisina (ed.), Jazyki korennyx narodov Sibiri (Novosibirsk: RAN). Cheremisina, M. I., E. I. Ubrjatova and F. A. Litvin (1984) Predikativnoe sklonenie prifastij v altajskix jazykax (Novosibirsk: Nauka). Cincius, V. I. (1949) Sravnitel’naja fonetika tungusko-mandzhurskix jazykov (Leningrad: Uchpedgiz). Cincius, V. I. (1982) Negidal’skij jazyk. Issledovanija i materialy (Leningrad: Nauka). Cincius, V. I. (1997a) ‘Solonskij jazyk’, in pp. 226–236. Cincius, V. I. (1997b) ‘Negidal’skij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 188–201. Csúcs, Sándor (1998) ‘Udmurt’, in Abondolo (ed.), The Uralic Languages, pp. 276–304. Darbeeva, A. A. (1996) Istoriko-sopostavitel’nye issledovanija po grammatike mongol’skikh jazykov. Fonetika (Moscow: Nauka). Darbeeva, A. A. (1997) ‘Burjatskij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 37–51. de Reuse, W. (1994) Siberian Yupik Eskimo: The Language and its Contacts with Chukchi (Salt Lake City: University of Utah Press). Grišina, N. M. (1977) ‘Upotreblenie slova bang v slozhnom predlozhenii ketskogo jazyka’, Jazyki i toponimija, vol. 4, pp. 102–107. Gruzdeva, Ekaterina (1998) Nivkh. Languages of the World Materials, 111 (Munich: LINCOM EUROPA). Hausenberg, Anu-Reet (1998) ‘Komi’, in Abondolo (ed.), The Uralic Languages, pp. 305–26. Helismki, Eugene (1998a) ‘Sel’kup’, in Abondolo (ed.), The Uralic Languages, pp. 548–79. Helismki, Eugene (1998b) ‘Nganasan’, in Abondolo (ed.), The Uralic Languages, pp. 480–515.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
Honti, László (1998) ‘Ob-Ugrian’, in Abondolo (ed.), The Uralic Languages, pp. 327–57. Janhunen, Juha (1998) ‘Samoyedic’, in Abondolo (ed.), The Uralic Languages, pp. 357–79. Jartseva, V. N., Ju. S. Eliseev, K. E. Majtinskaja and O. I. Romanova (eds) (1993) Jazyki Mira: Ural’skie Jazyki (Moscow: Indrik). Kämpfe, H. R. and A. P. Volodin (1995) Abriss der Tschuktschischen Grammatik (Wiesbaden: Harrassowitz). Kangasmaa-Minn, Eeva (1998) ‘Mari’, in Abondolo (ed.), The Uralic Languages, pp. 219–248. Keresztes, László (1998) ‘Mansi’, in Abondolo (ed.), The Uralic Languages, pp. 387–427. Khelimski, Eugene (1993a) ‘Sel’kupskij jazyk’, in Jartseva et al. (eds), Jazyki Mira, pp. 356–72. Khelimski, Eugene (1993b) ‘Matorsko-Taigisko-Karagasskij jazyk’, in Jartseva et al. (eds), Jazyki Mira, pp. 372–79. Khonti, L. (1993) ‘Xantyjskij jazyk’, in Jartseva et al. (eds), Jazyki Mira, pp. 301–19. Konstantinova, O. A. (1964) Evenkijskij jazyk (Moscow/Leningrad: Nauka). Korkina, E. I. and E. I. Ubrjatova (1982) Grammatika sovremennogo jakutskogo literaturnogo jazyka (Moscow: Nauka). Kostjakov, M. M. (1976) ‘Ketskie sootvetsvija russkomu slozhnopodchinennomu predlozheniju s pridatochnym vremeni’ Jazyki i toponimija, vol. 1, pp. 56–63. Krejnovich, È. A. (1937) Fonetika nivkhskogo jazyka (Moscow/Leningrad: Uchpedgiz). Krejnovich, È. A. (1958) Yukagirskij jazyk (Moscow/Leningrad: AN SSSR). Krejnovich, È. A. (1982) Issledovanija i materialy po jukagirskomu jazyku (Leningrad: Nauka). Künnap, Ago (1971) System und Ursprung der Kamassischen Flexionssuffixe I: Numeruszeichen und Nominalflexion. Mémoires de la Société Finno-Ougrienne 147 (Helsinki). Künnap, Ago (1993) ‘Kamassinskij jazyk’, in Jartseva et al. (eds), Jazyki Mira, pp. 380–88. Künnap, Ago (1999a) Enets. Languages of the World/Materials, 186 (Munich: LINCOM EUROPA). Künnap, Ago (1999b) Kamass. Languages of the World/Materials 185 (Munich: LINCOM EUROPA). Kuper, Sh. C. (1986) ‘Mesto ketskogo dialekta v sisteme sel’kupskogo jazyka’, in V. Nedjalkov (ed.), Fonetika jazykov Sibiri i sopredel’nykh regionov (Novosibirsk: Nauka), pp. 103–6. Lebedeva, E. P. (1997) ‘Orochskij jazyk’, Romanova et al. (eds), Jazyki Mira, pp. 215–26. Mal’chukov, Andrei L. (1995) Even. Languages of the World/Materials, 12 (Munich: LINCOM EUROPA). Maslova, E. (2003) A Grammar of Kolyma Yukaghir (Berlin/New York: Mouton de Gruyter). Menovshchikov, G. A. (1962) Grammatika jazyka aziatskikh eskimosov, vol. I (Moscow/ Leningrad: Nauka). Menovshchikov, G. A. (1964) Jazyk sirenikskikh eskimosov (Moscow/Leningrad: Nauka). Menovshchikov, G. A. (1975) Jazyk naukanskikh eskimosov (Leningrad: Nauka). Menovshchikov, G. A. and N. B. Vakhtin (1983) Eskimosskii iazyk (Leningrad: Prosveshchenie). Mikola, Tibor (1988) ‘Geschichte der samojedischen Sprachen’, in D. Sinor (ed.), pp. 219–63. Mudrak, O. A. (1986) ‘Opyt rekonstruktsii fonetiki prajazyka amerikanskikh i aziatskikh èskimosov’, in P. Ya. Skorik (ed.), Paleoaziatskie jazyki (Leningrad: Nauka), pp. 218–39. Nedjalkov, Igor V. (1997) Evenki (London: Routledge).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
298
299
Nikolaeva, Irina (1999) Ostyak. Languages of the World/Materials, 305 (Munich: LINCOM EUROPA). Nikolaeva, Irina and Maria Tolskaja (2001) A Grammar of Udihe (Berlin: Mouton de Gruyter). Novikova, A. I. and L. I. Sem (1997) ‘Orokskij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 201–15. Novikova, K. A. (1991) Evenskij jazyk (Leningrad: Prosveshchenie). Orlovskaja, M. N. (1997) ‘Staropis’mennyj mongol’skij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 18–29. Panfilov, V. Z. (1962) Grammatika nivkhskogo jazyka, vol. I (Moscow/Leningrad: AN SSSR). Panfilov, V. Z. (1968) ‘Nivxskij jazyk’, in V. V. Vinogradov (ed.), Jazyki Narodov SSSR V: Mongol’skie, tunguso-man’fz˘urskie i paleoaziatskie jazyki (Moscow: Indrik), pp. 408–34. Petrov, N. E. (1984) Modal’nye slova v jakutskom jazyke (Novosibirsk: Nauka). Petrova, T. I. (1967) Jazyk orokov (ul’ta) (Leningrad: Nauka). Pevnov, A. M. (1997) ‘Chzhurchzhen’skij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 260–7. Poppe, N. (1960) Buriat Grammar. Uralic and Altaic Series, 2 (Bloomington, Ind: Indiana University Press). Pritsak, Omeljan (1959) Das Abakan-türkische (Chakassische)’, in J. Deny (ed.), Philologiae Turcicae Fundamenta, vol. I (Wiesbaden: Harrassowitz), pp. 598–622. Prokof’ev, G. N. (1937a) ‘Nenetskij (jurako-samoedskij) jazyk’, in Alkor (ed.), Jazyki i pis’mennost’ narodov severa, pp. 5–52. Prokof’ev, G. N. (1937b) ‘Nganasanskij (tavgijskij) jazyk’, in Alkor (ed.), Jazyki i pis’mennost’ narodov severa, pp. 53–74. Prokof’ev, G. N. (1937c) ‘Enetskij (ostjako-samodijskij) jazyk’, in JAlkor (ed.), Jazyki i pis’mennost’ narodov severa, pp. 75–90. Prokof’ev, G. N. (1937d) ‘Sel’kupskij (ostjako-samoedskij) jazyk’, in Alkor (ed.), Jazyki i pis’mennost’ narodov severa, pp. 91–124. Rassadin, V. I. (1978) Morfologija tofalarskogo jazyka v sravnitel’noj osvešfenii (Moscow: Nauka). Rassadin, V. I. (1997) ‘Tofalarskij jazyk’, in È. R. Rassadin (ed.), Jazyki Mira: Tjurkskie jazyki (Moscow: Indrik), pp. 371–82. Riese, Timothy (1998) ‘Permian’, in Abondolo (ed.), The Uralic Languages, pp. 249–75. Riese, Timothy (2001) ‘Vogul (Mansi)’, Languages of the World Materials, 158 (Munich: Lincom Europa). Romanova, O. I., V. M. Alpatov, I. V. Kormushin and G. Ts. Pürbeev (eds) (1997) Jazyki Mira: Mongol’skie jazyki, Tungusko-man’chzhurskie jayki, japonskij jazyk, korejskij jazyk (Moscow: Indrik). Rombandeeva, E. I. (1973) Mansijskij (vogulskij) jazyk (Moscow: Nauka). Rombandeeva, E. I. (1993) ‘Mansijskij jazyk’, in Jartseva et al. (eds), Jazyki Mira, pp. 283–381. Sammallahti, Pekka (1998) ‘Saamic’, in Abondolo (ed.), The Uralic Languages, pp. 43–95. Sem, L. I. (1997) ‘Nanajskij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 173–88. Shamina, L. A. (1993) ‘Složnye predloženija v tuvinskom jazyke , vyražajušfie otnošenija oboslovennosti meždu sobytijami’, in D. A. Mongush and B. I. Tatarintsev (eds), Voprosy tuvinskogo jazykoznanija (Kyzyl: TNIIJaLI), pp. 63–75. Shcherbak, A. M. (1977) Ocherki po sravnitel’noj morfologii tjurkskix jazykov (Imja) (Leningrad: Nauka). Simoncsics, Péter (1998) ‘Kamassian’, in Abondolo (ed.), The Uralic Languages, pp. 580–601.
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Gregory D. S. Anderson
The Siberian Linguistic Area
Skorik, P. Ya. (1986) ‘Kategorii imeni sushchestvitel’nogo v chukotsko-kamchatskikh jazykakh’, in P. Ya. Skorik (ed.), Paleoaziatskie jazyki (Novosibirsk: AN SSSR). Stebnitskij, S. N. (1937) ‘Osnovnye foneticheskie razlichija dialektov nymylanskogo (korjakskogo) jazyka’, in I. I. Meshchaninov (ed.), Pamjati V. G. Bogoraza (Moscow/ Leningrad: Ufpedgiz). Starostin, S. A. (1982) ‘Praenisejskaja rekonstruktsija i vneshnie svjazi enisejskikh jazykov’, in V. V. Ivanov, V. N. Toporov and B. A. Uspenskij (eds), Ketskij sbornik, Antropologija, ètnografia, mifologija, lingvistika (Leningrad), pp. 144–237. Steinitz, Wolfgang (1937) ‘Xantyjskij (ostjackij) jazyk’, in Jazyki i pis’mennost’ narodov severa. vol. 1 (Moscow/Leningrad: Uchpedgiz), pp. 193–228. Steinitz, Wolfgang (1950) Ostjakische Grammatik und Chrestomathie (Leipzig: Harrassowitz). Sunik, O. P. (1968a) ‘Udegejskij jazyk’, in P. Ya. Skorik (ed.), Jazyki Narodov SSSR, vol. V (Moscow: AN SSSR), pp. 210–32. Sunik, O. P. (1968b) ‘Ul’chskij jazyk’, in P. Ya. Skorik (ed.), Jazyki Narodov SSSR, vol. V (Moscow: AN SSSR), pp. 149–71. Tenishev, È. R., V. D. Arakin and G. F. Blagova (1988) Sravnitel’no-istoricheskaja grammatika tjurkskikh jazykov. Morfologija (Moscow: Nauka). Tereshchenko, N. M. (1966a) ‘Vvedenie (Samodijskie jazyki)’, in Vinogradov (ed.), Jazyki Narodov SSSR, pp. 363–75. Tereshchenko, N. M. (1966b) ‘Nenetskij jazyk’, in Vinogradov (ed.), Jazyki Narodov SSSR, pp. 376–95. Tereshchenko, N. M. (1966c) ‘Nganasanskij jazyk’, in Vinogradov (ed.), Jazyki Narodov SSSR, pp. 416–37. Tereshchenko, N. M. (1966d) ‘Enetskij jazyk’, in Vinogradov (ed.), Jazyki Narodov SSSR, pp. 438–57. Tereshchenko, N. M. (1979) Nganasanskij jazyk (Leningrad. Nauka). Tereshkin, N. I. (1966) ‘Khantyjskij jazyk’, in Vinogradov (ed.), Jazyki Narodov SSSR, pp. 319–42. Todaeva, B. X. (1997a) ‘Baoan’skij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 29–37. Todaeva, B. X. (1997b) ‘Dagurskij jazyk’, in Romanova et al. (eds), Jazyki Mira, pp. 51–60. Ubrjatova, E. I. (1985) Jazyk noril’skikh dolgan (Novosibirsk: Nauka). Viitso, Tiit-Rein (1998) Estonian’, in Abondolo (ed.), The Uralic Languages, pp. 115–48. Vinogradov, V. V. (ed.) (1966) Jazyki Narodov SSSR, vol. III, Finno-ugors’kie i samodijskie jazyki (Moscow: Nauka). Volodin, A. P. (1976) Itel’menskij jazyk (Leningrad: Nauka). Volodin, A. P. (1997) ‘Itel’menskij jazyk’, in Jazyki Mira: Paleoaziatskie jazyki, pp. 60–72. (Moscow: Indrik). Werner, H. K. (1997a) Das Jugische (Sym-Ketische). Veröffentlichungen der Societas Uralo-Altaica, 50 (Wiesbaden: Harrassowitz). Werner, H. K. (1997b) Abriss der kottischen Grammatik (Wiesbaden: Harrassowitz).
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
300
Abaev, V. I., 168 Adelaar, K. A., 139, 140, 143, 144, 146, 147 Aikhenvald, Alexandra Y., 6, 8–10, 13, 14, 17–21, 24, 78, 85, 172, 191 Akbaev, E. X., 170 Aklif, G., 262 Alekseev, M. E., 167 Alpher, B., 68, 255 Amborn, H., 81 Anderson, G. D. S., 267, 268, 271, 272, 275, 276, 282, 283 Anttila, Raimo, 248 Aoki, Haruo, 7 Appaev, A. M., 161 Appleyard, D., 83, 84 Artawa, Ketut, 228 Artini, Put, 228 Auer, P., 39 Bakker, Peter, 135, 149, 151 Balode, L., 194 Bandelt, Hans-Jürgen, 60 Banti, G., 90 Baskakov, N. A., 270 Baxter, William, 58 Beach, Douglas M., 109, 117 Bechert, J., 39 Becker, Henrick, 4, 9–11, 37 Bekker, È. G., 289, 290 Bellwood, Peter, 24, 245 Bender, M. L., 81, 93 Benzing, J., 160 Berta, Á., 175 Besten, Hans den, 129 Bhaskararao, Peri, 145 Bichsler-Stettler, A., 139, 140, 145, 147 Birnbaum, Henrik, 7 Bisang, W., 76, 78, 81, 82, 88, 90–2 Black, Paul, 58, 59, 62, 63, 66, 160, 163, 165, 166 Blake, Barry J., 228 Bleek, Dorothea F., 15, 109, 121, 131 Bleek, Wilhelm H. I., 4, 114, 118, 121 Bloch, Jules, 2 Bloomfield, Leonard, 2 Blust, Robert A., 232 Boas, Franz, 2, 4 Boeder, W., 173, 174
Bogoraz, W., 295 Bolinger, Dwight, 68 Boretzky, Norbert, 106 Börjars, K., 48, 214, 215 Bowern, Claire, 68, 258, 259, 261, 262 Bright, William, 5, 7, 9, 10 Bruce, G., 198 Bryant, D., 66, 67 Budack, K. F. R., 102 Bulatova, N. Ja., 272, 278, 280, 284 Cafer2lu, A., 160 Campbell, Lyle, xv, xvi, xviii, 2, 6–10, 14, 17, 18, 20, 24, 33, 36, 37, 47, 65, 75, 81, 87, 193, 218, 245, 247, 249, 256 Castrén, M. A., 278 Eekmonas, V., 208 Chambers, J. K., 256 Chappell, Hilary, 20, 21 Cheremisina, M. A., 282 Cheremisina, M. I., 274 Chirikba, V., 174 Chomsky, N., 76, 85, 86, 92 Christophe, B., 171, 172 Cincius, V. I., 271, 280, 285 Comrie, B., 168 Crass, J., 76, 78, 80, 81 Croft, W. A., 85, 86 Crowley, T., 245 Csúcs, Sándor, 290 Curnow, Timothy Jowan, 17 Dahl, Östen, 11, 13, 14, 19, 21, 24, 47, 86, 182, 185, 201, 202, 204, 216–18 Darbeeva, A. A., 269, 273 Darnell, Regna, 2 Décsy, G., 7, 37, 185 de Haan, F., 191, 220 Delsing, L.-O., 211–15 Dempwolff, Otto, 119, 120, 122 Dench, Alan, 23, 261 de Reuse, W., 231, 295 De Silva, M. W. Sugathapala, 136, 149–51 de Vries, Lourens, 229 Dharmadasa, K. N. O., 136, 150 Dickens, Patrick J., 115 Dimmendaal, Gerrit J., 10, 91 301
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Authors
Index of Authors
Dixon, R. M. W., xviii, 6, 8–10, 13, 14, 17–21, 24, 68, 78, 85, 244–54, 257, 259–62 Dmitriev, N. K., 170 Doerfer, G., 160 Dogil, G., 194, 195 Dolgopolsky, A., 93 Donabédian, A., 173 Donegan, Patricia, 151 Donohue, Mark, 241 Dryer, Matthew, 9, 81, 94, 208–10, 221 Dunning, T., 239 Durie, Mark, 228 Dyen, Isidore, 58, 59, 62, 63, 66 D}idalaev, N. N., 167 Ebel, Hermann, 2 Eccles, J. C., 35 Echols, John M., 232 Ehret, Ch., 80, 93 Eiseb, E., 130 Elderkin, Edward D., 100 Eldredge, Niles, 245, 246 Embleton, Sheila, 53, 54, 57–9, 62 Evans, Nicholas D., 68, 231, 255 Fähnrich, H., 44 Falkenhahn, V., 185 Felsenstein, J., 59 Ferguson, C. A., 76, 77, 79 Ferraz, Luiz Ivens, 148 Fleming, H., 93 Foley, William A., 118, 150, 155, 229 Forster, Peter, 60 Foster, A., 60 Fraser, Bruce, 58 Friedman, V. A., 174 Gad}ieva, N. Z., 178 Gair, James W., 136 Gårding, E., 203 Garrett, Andrew, 20 Gecadze, I. O., 166 Geeraerts, Dirk, 228, 236 Geiger, B., 166 Geniušien<, E., 191 Geraghty, Paul, 260 Giannini, Stefania, 12, 17 Girard, (abbé) Gabriel, 2 Golden, P. B., 163, 165, 166, 168, 169, 174, 176 Gould, Stephen Jay, 245, 246 Gravelle, Gilles, 229 Greenberg, Joseph H., 76, 77, 79, 90, 91, 94, 99 Grenoble, L., 272, 278, 280, 284
Grišina, N. M., 277 Grjunberg, A., 172, 173 Grondelaers, Stefan, 228, 236 Gruzdeva, Ekaterina, 268, 271, 273, 275, 283 Gukasjan, V. L., 175 Güldemann, Tom, 99, 100, 105, 109, 111–13, 117, 118, 120, 121, 123, 128 Gumperz, John J., 38, 154 Gussenhoven, C., 198 Haacke, Wilfrid H. G., 111, 115, 116, 118, 120, 123, 130 Haarmann, Harald, 2, 10, 23, 37, 86, 87, 185 Haas, Mary R., 7 Hagman, Roy S., 113–15, 119, 120, 123, 131 Hahn, Theophilus, 125 Haig, Jeffrey, 17 Hale, Kenneth, 251 Hammarberg, B., 218 Hamp, Eric, 18 Harris, A. C., 173 Harris, D. R., 248 Harrison, K. David, 272, 275, 283 Haspelmath, Martin, 17, 39, 47, 48, 82, 168, 172, 191, 216 Hausenberg, Anu-Reet, 290 Hawkins, J. A., 88, 90, 94 Heath, Jeffrey, 17 Heggarty, Paul, 57, 65, 67 Heijmans, L., 199 Heine, Bernd, xvi–xviii, 1, 48, 113, 115 Henderson, Eugénie J. A., 9, 10 Hercus, Luise, 250, 257, 258, 260 Hetzron, R., 77, 82 Hickey, Raymond, 9 Hill, Jane, 7 Hock, Hans Henrich, 256 Hoenigswald, Henry M., 248 Holt, Dennis, 7, 10 Holvoet, A., 194 Honti, László., 288 Hopper, P., 188 Hosokawa, K., 262 Hualde, J. I., 48 Huang, Shuanfan, 232 Hübschmann, Heinrich, 20 Huson, D., 68 Hussainmiya, B. A., 139, 140, 151, 152 Jackson, Keith, 136 Jacobs, Melville, 4 Jaggar, Philip J., 231
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
302
Jakobson, Roman, 2, 4, 9, 10, 18, 23, 185, 193, 199 Jayasuriya, Shihan de Silva, 136–8, 148, 151 Jenkins, Trefor, 130 Jespersen, O., 221 Johannesen, J. B., 215 Johanson, Lars, 34, 86, 162, 167, 171–5, 177, 178 Jones, Rhys, 262 Joseph, Brian D., 39, 245 Juvonen, P., 47 Källskog, M., 212 Kamminga, J., 253, 257 Kämpfe, H. R., 273, 275, 281, 287 Kangasmaa-Minn, Eeva, 290 Kapeliuk, O., 82 Karakoç, B., 174 Karunatillake, W. S., 136 Katz, Harmut, 4, 5, 8–10, 17, 86 Kaufman, Terrence, 1, 7, 11, 14, 17, 22, 24, 33, 37, 77, 80, 81, 127, 128, 247, 255 Keller, R., 35, 83 Kemmer, S., 191 Kerimov, I. A., 170 Kessler, Brett, 54, 56, 57, 59, 60, 62, 65 Kilian-Hatz, Christa, 113 Kiparsky, V., 44 Klamer, Marian, 227–9, 231, 233–5, 241 Klokeid, T., 254 Koch, Harold, 68, 251, 258, 261, 262 Kodzasov, S., 207 Koerner, Konrad, 51 Köhler, Oswin, 109 König, Christa, 82 König, Ekkehard, 82 Konow, Sten, 2 Kopitar, Jeernej, 2 Koptjevskaja-Tamm, Maria, 4, 8–10, 16, 24, 33, 88, 182, 185, 187, 188, 192, 193, 198, 206, 211–14, 216–19 Korkina, E. I., 271 Kostjakov, M. M., 277 Kozintseva, N., 173 Krejnovich, È. A., 268, 272, 273, 279, 288 Kristophson, J., 87 Kroch, Anthony, 241 Kruskal, Joseph B., 58, 59, 62, 63, 66 Kuhn, Adelbert, 2 Künnap, Ago, 274, 275, 290 Kuper, Sh, C., 281 Kuteva, Tania, xvi–xviii, 1, 13, 48, 245, 256
Labov, William, 83, 247 Lahiri, A., 197, 198 Lamberti, M., 93 Lane, A. B., 130 LaPolla, Randy, 21 Larsson, S., 211 Lass, R., 247 Lebedeva, E.P., 280 Lee-Smith, Mei W., 151 Lehiste, Ilse, 7, 193, 197, 198 Leisiö, L., 208 Leslau, W., 77, 78, 80–2 Levin, Beth, 58, 59, 62, 63, 66 Li, Charles N., 151 Lichtenberk, Frantisek, 115 Lloyd, Lucy C., 114, 118, 121, 123, 124 Lohr, Marisa, 58 Lorentz, F., 199 Lorentz, O., 201 Lötzsch, R., 47 Magomedov, A. G., 175 Maingard, Louis F., 117, 130 Manaster-Ramer, Alexis, 58 Manhire, Anthony, 103 Margetts, Anna, 233 Martinet, André, 4 Masica, Colin P., 4, 8–10, 12, 33, 39, 88, 145 Maslova, E., 76, 90, 92, 275 Matisoff, James A., 10, 24 Matras, Yaron, 83, 89, 142 Matthews, Peter, 7, 8 McBryde, Isobel, 257, 258 McConvell, Patrick, 253, 255 McKelson, K., 262 McMahon, April, 55, 58, 66, 67, 72 McMahon, Robert, 55, 58, 65–7, 72 McWhorter, J. H., 218 Meillet, Antoine, 2 Meinhof, Carl, 115, 116, 120 Menges, K. H., 167, 176 Menovshchikov, G. A., 278 Meriggi, Piero, 114 Meyer, R., 83 Miklosich, Franz von, 2 Milroy, J., 83 Milroy, L., 83 Minker, G., 81 Mira, W., 228 Mitchell, B., 221 Mithun, Marianne, 2 Morpurgo Davies, Anna, 72 Moulton, V., 66, 67 Mudrak, O. A., 270 Müller, Max, 2
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Authors 303
Index of Authors
Mulvaney, John, 253, 257, 262 Musgrave, Simon, 234, 241 Næss, Å., 188 Nakagawa, Hirosi, 99 Namaseb, L., 130 Nanayakkara, A. G. W., 135 Nash, David, 68, 69, 255 Nasidze, I., 177 Nau, N., 39, 47, 192, 193 Nedjalkov, Igor V., 274 Nedjalkov, V. P., 82, 178 Nettle, D., 245, 248, 262 Nichols, Johanna, 18 Nikolaeva, Irina, 272, 280, 285 Norde, M., 211, 215, 217 Novikova, A. I., 284 Nurse, George T., 130 O’Connor, Sue, 253 Odé, Cecilia, 229 Oftedal, M., 197 O’Grady, G. N., 251, 254 Orel, V. E., 80 Orlovskaja, M. N., 290 Paauw, Scott, 138, 139, 141, 143, 147, 151–3 Pagel, M. D., 248 Pakendorf, B., 177, 221 Panfilov, V. Z., 268, 273 Parkington, John, 103 Petrov, N. E., 291 Plank, F., 47 Poppe, N., 290 Popper, K. R., 35 Poser, William, 20 Powell, John Wesley, 2 Praetorius, F., 82, 93 Prentice, D. J., 139, 146, 147 Pritsak, Omeljan, 161, 165, 168, 169, 279 Prokof’ev, G. N., 278 Quirk, Randolph, 68 Ramanujan, A. K., 4, 10 Ramat, Paolo, 7, 14, 17, 32 Rassadin, V. I., 273 Reesink, Ger P., 227–9, 234–7, 241 Reiter, Norbert, 24, 32 Renfrew, Colin, 24, 60 Riad, T., 196, 197, 201, 203, 204 Richards, Martin, 60 Richter, R., 83 Riese, Timothy, 289, 290 Ringe, Donald, 54, 55, 241
Roberts, Richard, 262 Robuchon, Gérard, 136, 139, 140, 142, 143, 145 Röhl, A., 60 Rombandeeva, E. I., 274, 296 Roos, J., 198 Ross, Malcolm, 35, 137, 150, 154, 255, 260 Rosser, Z., 221 Russian, 65 Šagirov, A. K., 167 Salmons, J., 206 Sammallahti, Pekka, 290 Sandfeld, Kristian, 2, 88 Sansò, A., 47 Sarhimaa, Anneli, 9, 185 Sasse, H.-J., 81 Say, S., 210 Scaglione, Stefania, 12, 17 Schaller, Helmut Wilhelm, 2, 4, 5, 7–9, 23, 86 Schapera, Isaac, 125 Schleicher, August, 2, 51 Schmalstieg, W., 210 Schmidt, Johannes, 2, 51 Schuchart, Hugo Ernst Mario, 2 Schultze, Leonhard, 109, 130 Schulze, Wolfgang, 245 Sebeok, Thomas, 21, 23 Seidel, Eugen, 7 Šejxov, E. M., 167 Sem, L. I., 280, 284 Shadily, Hassan, 232 Shamina, L. A., 282 Shcherbak, A. M., 282 Sherzer, Joel, 2, 4, 5, 9, 10, 38, 87 Sihler, A., 247 Silva-Corvalán, C., 86 Simoncsics, Péter, 290 Skorik, P. Ya, 269, 271, 278, 281, 285, 286, 295 Slaska, Natalia, 65, 67 Slomanson, Peter, 139, 140, 143 Smith, Ian R., 136–8, 140, 142, 143, 145, 151–3, 253, 262 Smith, Michael, 262n.8 Smith, Moya, 253 Smith-Stark, Thomas, 7, 14, 17, 33, 37 Soodyall, Himla, 130 Speelman, Dirk, 228, 236, 237, 239, 241 Staden, Miriam van, 227, 228 Stampe, Davis, 151 Stankiewicz, E., 199 Starostin, S. A., 58, 269 Stassen, L., 189, 190
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
304
Stebnitskij, S. N., 281, 285 Steinitz, Wolfgang, 296 Stilo, D., 168, 170, 174–6 Stolbova, O. V., 80 Stolz, Christel, 37, 47, 84, 85, 88, 148 Stolz, Thomas, 2, 8–10, 13–17, 23, 24, 32–4, 37, 38, 47, 48, 84, 85, 88, 137, 142, 148, 184, 185, 192 Stoneking, M., 177 Subbarao, Karumuri Venkata, 145 Sumbatova, N., 174 Swadesh, Morris, 53, 54, 57, 65, 68, 253 Sykes, Bryan, 60
van den Berg, René, 229 van der Auwera, Johan, 8, 17, 24 van Klinken, Catharina Lumien, 229 Van Valin, Robert D., 118 van Staden, Miriam, 227, 228 Vedder, Heinrich, 102, 116, 121, 129 Velten, H. V., 3, 23 Vendryes, Joseph, 2 Veth, Peter, 253 Viitso, Tiit-Rein, 206, 290 Visser, Hessel, 130 Voegelin, Carl F., 7, 23 Volodin, A. P., 273, 275, 281, 287 Vossen, Rainer, 99, 100, 117, 123
Tambets, K., 221 Taylor, Ann, 54, 55, 241 Tenishev, È. R., 291 Tereshchenko, N. M., 268 Tereshkin, N. I., 268, 271 Terrill, Angela, 262 Thananjayarajasingham, S., 136 Thomason, Sarah Grey, 1, 6–15, 17, 22, 23, 52, 56, 76–81, 83, 84, 86, 93, 127, 128, 151, 154, 247, 255 Thompson, S. A., 188 Timberlake, A., 189 Todaeva, B. X., 290 Tolskaja, Maria, 272, 280, 285 Toman, Jindrich, 3 Torp, A., 212 Torroni, Antonio, 60 Tosco, M, 1, 8, 11, 76, 78, 80, 81, 90, 93, 94 Traill, Anthony, 99, 108, 109, 113, 116, 130 Trask, R. L., xv, xvi, 7 Trubetzkoy, Nikolai Sergeevich, 1–4, 9, 10, 23, 75, 84, 93 Trudgill, Peter, 15
Wagner, H., 9 Wälchli, Bernhard, 24 Wallmann, J. C., 120 Wang, Feng, 57 Wang, William S.-Y., 57 Warnow, Tandy, 54, 55 Watkins, Calvert, 17, 20, 247 Webley, Lita, 103 Weinreich, Uriel, 2, 4, 9 Werner, H. K., 276, 279 Whitney, William Dwight, 2 Wiemer, B., 185 Wiik, K., 7, 193 Wilson, Robert, 38, 154 Winter, Werner, 9, 10, 208 Wintschalek, Walter, 10, 23, 86 Wolff, Hans, 7 Wurm, Stephen A., 151, 251
Ubrjatova, E. I., 271, 291 Ureland, P. S., 192
Xabifev, M. A., 175 Yaktonova, N. S., 290 Yates, Royden, 103 Zaborski, A., 76, 77, 79 Zeps, Valdis, 7 Zerjal, T., 221 Zhukova, A. N., 272, 282, 286, 287
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Authors 305
Abaza, 166 Abkhaz, 165, 166, 170 Proto-Abkhaz, 174 Abkhaz-Circassian, 162, 165, 166, 170, 171, 174 Acehnese, 228 Adnyamathanha, 260 Adyghe, 166 African languages, 100, 106, 129, 148 Afrikaans, 60, 128, 129, 247 Afro-Asiatic, 77, 78, 80, 81, 91, 93, 231 Proto-Afro-Asiatic, 78, 80, 81 Alans, 165, 166 Albanian, 48, 57, 176, 209 Altai, 266, 269–70, 272, 278, 282 Alune, 229 Al’utor, 269, 271, 281, 284, 285 Amerindian, 47 Amharic, 77, 78, 81–4 Andean languages, 67 Angloromani, 137 Arabana-Wangkangurru, 258, 259, 260 Arabic, 47 Aralo-Caspian, 161 Aramaic, 161, 171, 176 Arandic, 260, 262 Arin, 269 Armenian, 20, 21, 40, 151, 161, 162, 166, 170, 171, 173–6 Aryan, 38, 135–6, 142, 145, 182 Asian languages, 135, 145–9 Asiatic, 77–8, 80–1, 91, 93, 231 Assan, 269 Aukštaitian, 194–5 Australian languages, 68, 70, 71, 244, 245, 248, 250, 251, 253–5, 261, 262 Proto-Australian, 250, 251, 259 Austronesian, 47, 135, 139, 150, 155–6, 227–9, 232, 239 Avars, 165, 166, 171, 173, 207 Azerbaijan, 160, 164, 166, 176 Azeri, 160–1, 166–77
Balkan, 8, 35, 39, 44, 48, 77, 79, 88, 154, 164, 172, 174, 177–8, 241 Balkanic, 35, 48 Balkan–Pontic–Caucasus–Caspian, 174, 178 Balkar, 160, 161, 165, 166, 168–71, 175, 176, 178 Baltic, 9, 16, 42, 44, 77, 79, 88, 182, 185, 187–91, 193–4, 198–9, 201, 203, 205–10, 213, 219–21 Bantu, 111, 125, 128–30 Bardi, 253, 254, 262 Basque, 42, 44, 48, 178, 188, 216 Bayso, 90 Bengali, 67 Birka, 202 Bulgar, 161, 163–5, 177, 178 Old Bulgar, 177 Bulgarian, 44, 48, 174, 177, 178 Burman, 135 Buru, 229, 233 Buryat, 269, 273, 274, 290
Baexem, 199 Baku, 170 Balinese, 228
Daghestanian, 162–4, 166, 168, 170–5, 207 Dalecarlian, 184, 196, 202, 203, 212, 215, 217, 220
Calo, 137 Caspian, 161, 163–5, 168, 174, 178 Caucasian, 37, 162–72, 175–7 Celtic, 42, 54, 185, 205, 206, 209, 250 Chechen, 173, 175 Chinese, 57, 150–1, 281, 283 Hokkien Chinese, 139 Chukchi, 269, 271, 273, 275, 281, 284, 285, 286, 287 Proto-Chukchi-Koryak, 281, 285 Chukotko-Kamchatkan, 266, 269, 271–3, 275, 276, 278, 281, 283–7, 292 Proto-Chukotko-Kamchatkan, 286, 292 Circassian, 162, 165, 166, 170, 171, 174–6 Proto-Circassian, 174 Circum-Baltic languages, 16, 88, 182, 194, 199, 207, 209, 213, 221 Copper Island Aleut, 277 Curonian, 182 Cushitic, 22, 47, 77–83, 90, 93, 104, 129
306 10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Language Families, Languages and Dialects
Index of Language Families, Languages and Dialects 307
East !Xõo, 109, 110, 113, 116, 117, 123, 124 Eini, 103, 108, 111 Elfdalian, 203, 217 Enets, 268, 275, 278 English, 3, 15, 40, 44, 45, 54, 57, 59–61, 63–5, 91, 137, 152, 172, 213–21, 231, 241 Old English, 218, 219, 241 Estonian, 47, 182–5, 191–4, 196–200, 204, 206–8, 219–21, 290 Hiiumaa dialect, 200 Saaremaa dialect, 200 Ethio-Semitic, 77–83, 93 Ethiopian, 22, 76–81, 83–4, 88, 90 Eurasian, 145, 162, 185, 192, 209–10 European languages, 4, 8, 20, 32, 39, 44, 47, 54, 58, 59, 65, 67, 68, 103, 136, 137, 154, 161, 178, 182, 185, 187, 190–2, 194, 196, 205–7, 209–10, 221, 231, 246–7, 249, 251, 262 Even, 3, 32, 37, 75, 87, 88, 92, 125, 138, 142, 149, 151, 153, 167, 170, 232, 266, 273, 278, 280, 284, 285, 294 Evenki, 270, 271, 272, 274, 278, 280, 283, 284, 285, 294 Faroese, 54, 196, 204, 205 Finnic, 182, 184–5, 187–90, 195, 197, 199–201, 206, 208–10, 220 Finnish, 47, 184, 188, 192, 196, 200, 201, 204, 210, 219 Finno-Ugric, 177, 182, 192, 205–6, 208, 210, 213, 221 Proto-Finno-Ugric, 206 Flemish, 60 Formosan, 232, 234
French, 15, 54, 57, 60, 65, 178, 188 Old French, 178 Frisian, 44, 45, 54, 59 Galindian, 182 Genje, 170 Georgian, 46, 163, 166, 168, 172–5, 191 Old Georgian, 44, 172, 175 German, 39, 44, 45, 57, 60, 93, 119, 184, 191, 193, 194, 201, 217, 219, 250 Central Franconian dialects, 198 Low German, 184, 193, 194, 201, 212, 219 Germanic, 44, 45, 47, 48, 53–5, 59–64, 182, 192, 198, 206, 207, 210, 212, 216, 219, 250 Proto-Germanic, 206, 250 Gilyak, see Nivkh Greek, 44, 151, 184, 194, 209, 210, 249 Gumuz, 81 Gurage, 77, 82–3 Gypsy Telugu, see Kuravar G || ana, 105 Halmahera, 227, 235, 237 Hatam, 229, 233, 236, 237 Hausa, 231 Hebrew, 47 Hedeby-Birka, 202 Hindi, 67 c Hõa, 100 Hungarian, 44 Hunnic, 162–4 Icelandic, 42, 196, 201, 204–5, 213 Inanwatan, 229, 237 Indian, 93, 135, 145, 147, 149 Indic, 15, 135, 136, 142, 154, 247 Indo-Aryan, 38, 135–6, 142, 145, 182 Indo-European, 4, 8, 20, 54, 58, 59, 65, 67, 68, 136, 154, 182, 187, 190, 192, 194, 196, 205–7, 209, 210, 247, 249, 251, 262 Old Indo-European, 210 Proto-Indo-European, 206, 251 Indo-Iranian, 65–7 Indonesian, 227, 228, 231–2, 234, 237, 241 Iranian, 20, 161, 162, 165, 166, 168–77 Italian, 15, 60 Italo-Celtic, 54 Itel’men, 269, 281, 284, 285, 286
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Damara, 102, 116, 121, 128, 129, 130 Old Damara, 116 Danish, 48, 54, 59, 183, 184, 193–8, 201–3, 208, 210, 214–15, 217–21 Old Danish, 218, 219 Western Jutish, 204, 211, 212, 214, 220 Dargi, 160, 166, 170, 175 Didinga, 91 Digor, 165–6, 168 Diyari, 260 Dolgan, 270, 271, 290, 291 Dravidian, 38, 135–6, 139, 142, 145, 151, 153–5 Dutch, 44, 45, 54, 59, 60, 129, 198, 199, 217, 236 Limburgian dialects, 198–9
Index of Language Families, Languages and Dialects
Jatvingian, 182 Ju, 100, 105, 115, 130 Kabardino-Balkaria, 161 Kalam, 229 Kalmyk, 166, 269 Kamassian, 189 Kambera, 228, 233 Kamchatkan, 266, 269, 271–3, 275, 276, 278, 281, 283–7, 292 Kannada, 154 Karachay-Balkar, 160, 161, 165–6, 168–71, 175–6, 178 Karaim, 182, 184 Karajarri, 253, 254, 262 Karnic, 257–60 Proto-Karnic, 258 Kartvelian, 161, 162, 165, 166, 168, 171–3 Kashubian, 185, 193, 194, 199 Kayardild, 231 Kerek, 269, 271, 281, 284, 285 Ket, 177, 269, 277, 279, 281 Khaidaq, 160, 166, 169, 170 Khanty, 271, 288, 289, 292, 293, 295 Khazars, 163, 177 Khoe, 99, 100, 104, 105, 106, 107, 108, 110, 111, 112, 113, 115, 116, 117, 118, 119, 123, 125, 126, 127, 128, 130, 131 Kalahari Khoe, 99, 105, 106, 110, 113, 115, 116, 117, 119, 123, 130 Proto-Khoe, 111, 112, 116, 117 Khoe-Kwadi, 100, 131 Khoekhoe, 99, 100, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131 Khoi, 247 Khoisan, 99–101, 103, 105, 108–9, 124–6, 129–31 c Khomani, 101, 109 Khoy, 170 Kipchak, 161, 163–5, 174, 175, 178 Komi, 190, 208, 209, 221, 290, 295 Konkani, 149 Koryak, 4, 269, 271, 272, 278, 281, 282, 284, 285, 286, 287 Kott, 269 Kumans, 163, 164, 177 Kumyk, 160, 161, 164–70, 175 Kuravar, 4, 136 Kurdic, 166, 176
Lak, 166, 170 Lakota, 231 Latgalian, 183, 195 Latin, 15, 210, 250 Latvian, 40, 42, 47, 183, 184, 185, 189, 190, 191, 192, 193, 194, 195, 198, 199, 200, 206, 210, 219, 220 Low Latvian, 194, 195, 198, 199, 200 Laz, 173 Lezghian, 168, 170, 171, 175 Lezghic, 166, 169, 170 Lithuanian, 41, 42, 185, 189–95, 200, 208, 210, 219–21 Old Lithuanian, 210 Livonian, 184, 185, 191, 192, 193, 194, 195, 197, 198, 199, 200, 207, 219 Maale, 91 Macedonian, 44 Malay, 135, 136, 139, 140, 141, 142, 143, 144, 145, 146, 147, 149, 150, 151, 152, 153, 154, 227, 232 Sri Lanka Malay, 139, 140, 141, 142, 143, 145, 146, 147, 150, 151, 152, 153, 154 Malayalam, 149 Malayo-Polynesian, 228, 232 Maltese, 42 Mansi, 274, 288, 289, 292 Marathi, 67, 142, 154 Marr, 176 Ma’a, 137 Media Lengua, 137 Meyah, 229 Mingrelian, 166, 173, 174 Mongolic, 162, 164, 266, 269, 273, 274, 290, 292 Old Mongol, 269 Proto-Mongolic, 292 Mordvin, 40, 190, 208, 209, 221 Motu, 155 Mpur, 229 Muna, 229, 237 Nakh-Daghestanian, 160, 162, 164, 166, 168, 170, 171, 173–5 Nakhichevan, 160 Nama, 100, 101, 102, 103, 108, 109, 116, 118, 120, 121, 129, 130
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
308
Nanai, 271, 280 Naro, 105, 113, 130 Naukan, 270 Negidal, 280, 285 Nenets, 189, 268, 278, 289 Ngamini, 260 Nganasan, 268, 271, 278, 290 Nguni, 128 Nilo-Saharan, 77, 81, 83, 91 Nilotic, 91 Proto-Nilotic, 91 Nivkh, 266, 268, 271, 273, 275, 283 Noghay, 160, 161, 165–7, 174, 175 Nordic, Proto-, 197, 201 Norrbotten, 201, 212–13 Norwegian, 48, 193, 194, 196, 201, 203–5, 208, 210–21 Bokmål, 183, 212–14 Nynorsk, 183, 212,214 Nostratic, Proto-, 251 Nukha, 169, 170 Nyungan, 189, 244, 250–3, 256–61 Nhuki, 101, 114, 118, 122 N || ng, 100, 101, 104, 109, 114, 122, 131 Ob-Ugric, 266, 268, 274, 278, 288, 290, 292–3 Proto-Ob-Ugric, 288, 292 Oceanic, 233, 234 Odul, 268, 272, 273, 275, 279, 287, 288 Oghuz, 161, 163–6, 168, 174, 176, 178 Old Prussian, 182 Omo-Tana, 90 Omotic, 77, 78, 83, 91, 93 !Ora, 99–103, 106, 108–9, 111, 115–22, 124, 130 Orok, 271, 284, 285 Ossetic, 165, 166, 168–9, 171, 175–6 Pama-Nyungan, 189, 244, 250–3, 256–61 Proto-Pama-Nyungan, 251, 252, 258, 259 Papuan, 150, 155–6 Parthian, 176 Persian, 162, 166, 168, 170, 176 Pitta-Pitta, 259, 260 Polabian, 184, 185, 187, 190, 192, 208 Polish, 184, 185, 187, 190, 192, 208 Polynesian, 228, 232
Ponto-Caspian, 161, 164, 165, 168 Portuguese, 135–43, 145, 147–54 Asian Portuguese creoles, 148 Baticcaloa Portuguese, 152 Creole Portuguese, 136, 137, 151 Old Portuguese, 148 Sri Lanka Creole Portuguese, 136–43, 145, 147, 149–55 Punjabi, 67 Romance, 44, 53, 54, 60–2, 135, 209, 212 Romani, 182 Rumanian, 44, 48 Russian, 39, 44, 46, 161, 165, 167, 177, 185, 187–92, 205–6, 219, 274, 276–7, 296 Kyyrölä Russian, 208 Northern Russian, 184, 188 Saami, 182, 183, 185, 190, 192, 196, 205, 208, 209, 213, 221, 290 Saharan, 77, 81, 83, 91 Saliba, 233 Samoyedic, 189, 266, 268, 274–6, 278–9, 281, 289–90, 292 Sanskrit, 15, 250 Scandinavian, 44, 48, 184–5, 191, 196, 198, 201–8, 210, 212–13, 215–21 Proto-Scandinavian, 201 Seljuk, 164, 166, 176 Seljuks, 163 Sel’kup, 268, 272, 281, 283, 289, 290, 292, 293 Semitic, 22, 77–83, 93 Proto-Semitic, 93 Serbo-Croatian, 42, 44, 46, 194, 207 Sicilian, 15 Sinhala, 135–46, 149–51, 153, 154 Sinitic, 232 Sireniki, 269, 270 Slavic, 44, 46, 47, 60, 162, 177, 182, 184–5, 187–91, 194, 208–10, 220–1 Slovene, 194 Slovincian, 199 Sougb, 229, 234, 241 Sourashtra, 142, 153 South-Semitic, 81 Spanish, 15, 47, 137, 247 Stavropol’ Turkmen, 161, 165, 166 Surmic, 91 Svan, 165, 166, 173, 174 Swedish, 48, 184–5, 191, 193, 194, 196, 201–4, 208, 210–12, 214–15, 217–21 Överkalix, 201, 212 Västerbotten, 211, 212, 213
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Language Families, Languages and Dialects 309
Index of Language Families, Languages and Dialects
Taba, 237 Talysh, 166, 171, 172 Tamian, 184, 195 Tamil, 135–46, 150–5 Sri Lanka Tamil, 138–42 Tana, 90 Tat, 166, 168, 172, 173 Tatar, 47, 182 Tatic, 166, 168, 170, 176 Tetun, 229, 237 Tetun Fehan, 237 Thura-Yura, 260 Tibeto-Burman, 135 Tocharian, 54 Tofa, 269, 273, 279, 291 Tolai, 155 Tsakhur, 169, 170 Tsou, 232 Tuba, 270 Tukng Besi, 237 Tungusic, 177, 266, 268, 271–4, 278, 280–1, 283–5, 291–4 Proto-Tungus, 268, 280 Proto-Tungusic, 280, 281, 283, 285, 291 Turkic, 47, 151, 160–78, 182, 266, 269–74, 276–9, 282, 283, 290–2, 296 Old Turkic, 282, 283 Proto-Turkic, 291 Turkish, 8, 40, 57, 151, 160, 161, 169, 171, 173, 174, 191, 291 Tuu, 99–100, 102, 104–14, 116–19, 121–30 Proto-Tuu, 110, 112, 114 Tuvan, 272, 275, 282, 283 Udi, 166, 170, 175 Udihe, 272, 280, 284, 285
Ugric, 177, 182, 192, 205–6, 208, 210, 213, 221, 266, 268, 274, 278, 288, 290, 292–3 !Ui, 99–100, 103–10, 112, 114–18, 121–5, 127–31 Proto-!Ui, 110 c Ungkue, 100, 115, 131 Uralic, 182, 189–90, 205, 209, 221, 281, 290, 292 Proto-Uralic, 290 Urdu, 154 Vedda, 136, 145, 149–50 Veps, 184, 200 Võru, 183 Wadul, 268, 272, 273, 279, 287, 288 Western Türk, 163, 164, 178 Xakas, 276, 277, 279, 282, 291 |Xam, 100, 106, 108, 109, 112, 114–19, 121, 123, 124, 127, 130, 131 Yakut, 177, 270, 271, 291 Yandruwandha, 259, 260 Yawuru, 253, 254, 262 Yeniseic, 266, 269, 271, 274, 276–7, 279 Yindjibarndi, 189 Yugh, 269, 276, 279 Yukaghiric, 266, 268, 271–3, 275, 276, 279, 287–8, 292 Proto-Yukaghir, 268 Proto-Yukaghiric, 287 Yura, 260 \emaitia, 194, 195
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
310
acculturation, 2 adoption, 167, 169 adpositions, 79, 80, 118, 155 adstratum, 167–70, 176 archaeological evidence, 257, 259 aspect, 16, 18, 60, 71, 117, 125, 136, 146, 171, 187, 188, 191 aspiration, 106, 108, 109, 130, 169 basic code, 167 basic vocabulary, 3, 53, 55, 57, 65, 84, 246, 253, 254 Black Bulgaria, 163 bootstrapping, 69 borrowing, 1–23, 47, 51–68, 71, 73, 75, 76, 104, 105, 109, 111, 112, 122, 124, 126–8, 131, 142, 168, 196, 219, 231–3, 241, 245–8, 254, 255, 257, 260, 262, 269, 294 borrowed traits, 6, 14, 17 hierarchies of borrowability, 17 Bund, 33, 39 Peipus-Bund, 185 Rokytno-Bund, 185 Wikinger-Bund, 185 calquing, 47, 118, 260 case marking, 136, 137, 143, 146, 171, 174, 175, 184, 199–201, 203, 208, 210, 215, 216, 218–20, 274, 276 ablative, 142 accusative, 188, 189 allative, 267, 279–83, 292, 294 associative, 129, 131, 142 comitative, 192, 267, 279, 284–92 dative, 142, 146, 149, 190, 211, 212, 258, 259, 267, 277, 279–83, 292, 294 ergative, 170, 175, 258, 259, 287 garpe genitive, 212 genitive, 90, 142, 187, 188, 190, 208, 210–18, 220, 289 instrumental, 142, 167, 189, 190, 192, 267, 279, 284–92 locative, 142, 147, 259, 287 nominative, 188, 189 partitive, 187, 188 similative, 123, 124 sociative, 285–7, 290
causative, 116, 118, 147, 191 Chechens, 165 cladistic approach, 72 classification, 36, 48, 52, 72, 77, 100, 102, 244, 251, 255, 260 code shift, 162, 167, 168, 176 code-copying, 167 cognates, 56–63, 68, 109, 116, 120, 254, 284, 285, 289–91 common ancestry, 6, 19, 21, 51, 56, 60, 65, 70, 72, 93, 100, 126, 259, 292 comparative linguistics, 2, 51–3 comparative method, 18–21, 245, 248–50, 262 complementizer, 123, 124 complex predicates, 117, 118, 127, 227, 228, 233, 234, 236, 237, 241 compound verb, 117 consonant inventory, 107, 108, 130 constituent order, see word order contact, 5–7, 12–24, 32, 33, 36, 37, 42–7, 51–6, 61, 65, 67–70, 72, 75–93, 99, 100, 103–5, 108, 113, 116, 118, 125–31, 136, 138, 145–50, 153–5, 160, 162, 166–75, 182–5, 191–3, 196, 199–201, 205, 206, 209, 218–20, 227, 228, 233–42, 246–9, 253, 255–6, 258–61, 276 converbs, 82 convergence, 4, 18, 20, 21, 33–5, 38, 44, 45, 51, 64, 65, 71, 72, 75, 76, 79–85, 87–90, 92, 128, 135, 150, 153, 155, 156, 162, 171, 174–6, 184, 185, 192, 193, 219, 240, 244, 246, 247, 251, 257, 260, 276 long-distance convergence, 47 core–periphery pattern, 11 creole, 55, 106, 136–40, 142, 147–9, 151–3 creolization, 147 Daghestan, 160, 165–7, 169 declarative, 119–22 definiteness, 39, 40, 42, 44, 46, 47, 216 defocalization, 171–3 derivation, 113, 114, 116, 117, 126 diachronic evidence, 14 dialect continuum, 61, 87 311
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Subjects
Index of Subjects
dialects, 55, 58, 68, 83, 150, 161, 187–9, 206, 214, 215, 220, 221, 250, 256, 257, 260, 266, 292, 295, 296 dialect continuum, 61, 75, 87, 88, 102, 256 dialect convergence, 150 dialect epicentre, 256 dialect geography, 3, 11, 100, 256 dialect variation, 140, 184, 202, 220, 288, 290 prestige dialects, 202 diffusion, 1–24, 35–9, 45, 75, 78, 80–8, 103, 155, 172, 175, 227, 245–7, 250, 251, 257–61, 276, 294 diglossia, 151–3 diminutive, 113, 114, 129, 130 directionality, 65, 71 discontinuous transmission, 55 discourse particles, 142 distance, 205, 209, 210 composite, 66, 72 genetic, 8, 23, 47 linguistic, 203 diversity, 8, 11, 37, 46, 148, 166, 176, 192, 202, 210, 217, 220, 221, 248, 266, 292 ejectives, 109, 169 emotion predicates, 228, 234 equilibrium, 78, 244–7, 249–51, 254, 255, 257, 261 evidentiality, 140, 172–4, 220 exclusive, 21, 111, 112, 116, 127, 129 family trees, 18–21, 51, 52, 55, 60, 126, 161, 244–6, 255, 256, 260–2 focus, 8, 10, 12, 13, 18, 47, 54, 65, 82, 99, 100, 105, 129, 135, 136, 193, 219 focal postterminals, 172 frequential copying, 171, 172, 175 gender, 111, 113, 128 gene flow, 126 genetic impact, 177 genetic relatedness, 79, 84, 219, 251, 254, 255, 257 genetic subgroup, 257 geographic closeness, 76 global copying, 86 glottal stop, 169 glottalized consonants, 4, 79–81 glottochronology, 249 Golden Horde, 163, 165 grammaticalization, 122, 124, 127, 130, 131, 172, 173, 191, 192, 216, 282
imperfect learning, 22, 55, 77, 83, 84 imperfective, 117, 130 imposition, 167, 169 inclusive, 19, 20, 111, 112, 127 increase in complexity, 128 indirectivity, 172 interference, 9, 22, 79, 99, 104, 105, 111, 115, 123, 124, 126, 127, 129, 155 interrogative, 119–22 intertwined languages, 137 intraterminal, 171 isogloss, 3, 9–11, 36, 39–43, 45–7, 104–6, 109, 111, 125, 127, 185, 187, 191–3, 219, 244, 251, 256–8, 260, 261 bundles, 11, 192 dialects, 4, 11 iterative, 116 Kabardians, 165 Kulturbund, 37 language death, 151, 247, 248, 262 language family, 99 language shift, 77, 83, 104, 176, 202, 247 lexical, 3, 51–6, 60, 65, 67, 68, 71, 72, 104, 109–11, 117, 130, 137, 167, 168, 193, 194, 199, 201, 204, 207, 213, 214, 227, 236, 246, 251, 255 lexicostatistics, 251, 253 lingua franca, 139, 149, 151, 153, 164, 165, 167, 176, 177, 227 loanwords, 167–9, 175, 185, 260, 262 metatypy, 20, 35, 135, 137, 140, 153–5, 260 mixed languages, 2, 52, 56, 277 model code, 167, 171 morphology, 89, 111, 154, 241, 247, 257, 272, 274, 278, 290, 292 multi-lingualism, 14, 154, 166, 244, 260 nasals, 260, 268, 271 negation, 227 Neighbour Joining, 68 NeighbourNet, 66, 67, 70–2 Neogrammarians, 51 networks, 13, 67, 83, 86, 257, 258 network programs, 69 noun , 123, 124, 141, 148 number, 111, 140, 149 numeral, 190 object, 113, 118–20, 122, 145, 148, 187–90, 234 oblique, 123, 189, 191, 233 origin myths, 250, 252
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
312
parallel innovation, 17, 247, 261 particles, 116, 117, 122, 136, 137, 140, 142, 171, 192 pastoralists, 103, 104, 126, 131 perfect, 172–4 perfective, 117, 148 person, 83, 111–14, 121, 123, 131, 143, 146, 173, 174, 217, 251, 258–60 pharyngalized vowels, 169 pharyngeal fricatives, 79, 81 phonetic comparison, 65 phonology, 105, 140, 144, 145, 247, 249, 268, 271, 292 pidgin, 139 polar question, 120 polytonicity, 193–207, 219, 221 possessive NP, 216 postfield, 119, 122 postpositions, 81, 91, 93, 136, 139, 141–3, 155, 171, 216 postterminality, 172 prefield, 119–21, 127, 130 primary code, 167, 168, 176, 177 profile, 105, 117, 120, 128, 236, 237, 239, 241, 244 prolative, 278, 279, 282 pronoun, 47, 111, 112, 114, 115, 121, 124, 127, 192, 212, 215, 217, 251, 287 proposition, 4, 16, 3, 23 quantitative methods, 52, 56, 58, 72, 73 reconstruction, 52, 110, 125, 130, 232, 244, 249–51, 257, 261, 288 reduplication, 116, 137 regularity hypothesis, 51 relation, 33, 94, 104, 127, 216, 250, 262, 279 relative chronology, 260 relative clauses, 136, 137, 175, 209 relevance, 16, 55, 66, 76, 79, 80, 82, 87, 89, 100, 103, 111, 117, 187, 194, 218 reported discourse, 123, 124 resultative, 117, 172 reticulations, 60–2, 66, 67, 69, 70 sentence type, 119, 121, 122, 127 several-traits view, 9 similarity, 3, 17, 20, 33–5, 42, 56, 68, 72, 84, 85, 89, 99, 106, 111, 124, 126, 142, 219, 221, 237, 239 single-trait view, 8 sound change, 72, 247, 248, 250 sound harmony, 170
313
speech community, 39, 89 Splitstree, 68–70 spread, 9, 12, 13, 15, 39, 77, 78, 111, 147, 162–4, 175, 184, 188, 201, 202, 204, 212, 218, 221, 232, 245, 246, 252, 257, 294 Standard Average European, 39, 185, 191 stative, 117 structural features, 6, 7, 78, 93 subareas, 77, 84, 175, 177 subgroups, 6, 60, 164, 244, 251, 257, 259, 261 subject, 45, 104, 110, 118–22, 126, 127, 130, 136, 149, 176, 187–9, 205, 235 sublists, 58–60, 62, 67 subsistence, 103, 125, 126 substrata, 170, 176, 177, 279 substrate, 99, 104, 105, 109–11, 118, 125, 126, 128, 129, 148, 152, 167–70, 174, 176, 184, 205, 206 substratum, see substrate superstrate, 47 Swadesh list, 54, 57, 65, 68, 253 tense, 117, 136, 143, 169, 171, 191 topic, 47, 76, 90, 119, 122, 127, 218 Transcaucasia, 161, 164, 176 transitivity, 187, 192 tree-drawing programs, 59, 61, 62, 68, 71 Turkicization, 166, 167, 176 typological change, 139 typological independence, 87 typology, 19, 32, 45, 90, 94, 203, 213, 219 unidirectionality, 154 uniformity, 5, 237, 239–41 verb serialization, 117–19 vigesimal counting, 168 Wackernagel position, 127 wave model, 51 word accent, 194 word order, 84, 91–3, 127, 139, 144–52, 155, 192, 207–17, 221, 227, 241, 273 basic word order, 136, 209 comparatives, 137 verb-final, 80–1, 91, 119, 126, 155, 174–5 verb-first, 90–1
10.1057/9780230287617 - Linguistic Areas, Edited by Yaron Matras, April McMahon and Nigel Vincent
Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-04-20
Index of Subjects
