Subordination and Coordination Strategies in North Asian Languages (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Subordination and Coordination Strategies in North Asian Languages AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LIN...

Author: Edward J. Vajda

379 downloads 1298 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Subordination and Coordination Strategies in North Asian Languages

AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E.F.K. KOERNER (Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin) Series IV – CURRENT ISSUES IN LINGUISTIC THEORY Advisory Editorial Board Lyle Campbell (Salt Lake City) Sheila Embleton (Toronto) Elly van Gelderen (Tempe, Ariz.) Brian D. Joseph (Columbus, Ohio) John E. Joseph (Edinburgh) Manfred Krifka (Berlin) Martin Maiden (Oxford) E. Wyn Roberts (Vancouver, B.C.) Joseph C. Salmons (Madison, Wis.)

Volume 300

Edward J. Vajda (ed.) Subordination and Coordination Strategies in North Asian Languages

Subordination and Coordination Strategies in North Asian Languages

Edited by

Edward J. Vajda

Western Washington University & Max Planck Insitute for Evolutionary Anthropology, Leipzig

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA

4-

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984.

Library of Congress Cataloging-in-Publication Data Subordination and coordination strategies in North Asian languages / edited by Edward J. Vajda. p. cm. -- (Amsterdam studies in the theory and history of linguistic science. Series IV, Current issues in linguistic theory, ISSN 0304-0763 ; v. 300) Includes bibliographical references and index. 1. Russia, Northern--Languages--Subordinate constructions. 2. Russia, Northern--Languages--Coordinate constructions. 3. Russia, Northern--Languages--Syntax. I. Vajda, Edward J. P381.R8S83 2008 494'.6--dc22 2008030304 ISBN 978 90 272 4816 9 (Hb; alk. paper) © 2008 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. • P.O.Box 36224 • 1020 ME Amsterdam • The Netherlands John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA

CONTENTS Editor’s foreword Edward J. Vajda

vii

I. Introduction Subordination, Coordination: Form, semantics, pragmatics Bernard Comrie

1

II. Analytic patterns of subordination and coordination Speech Report Constructions in Ainu Anna Bugaeva

17

The Syntax and Pragmatics of Adverbial Clauses in Eastern Khanty Andrei Filtchenko

31

Null Arguments in Kumyk Adverbial Clauses Linda Humnick

47

Finite Structures in Forest Enets Subordination: A case study of language change under strong Russian influence Olesya Khanina & Andrey Shluinsky

63

Grammaticization and Relative Clauses in Eastern Khanty Olga Potanina

77

Toward a semantic typology of coordination Elena Rudnitskaya & Elena Uryson

85

Question particles or what? Open alternative questions in Udeghe Maria Tolskaya & Inna Tolskaya

97

III. Suffixation as a Technique of Syntactic Subordination The Development of Deconverbal Prepositions: Reanalysis or grammaticalization? Sandra Birzer

109

151

vi

CONTENTS

Imperatives in Conditional and Concessive Subordinate Clauses Nina Dobrushina Morphological Strategies for ‘Complex Sentences’ and Polysynthesis in Central Alaskan Yupik (Eskimo) Osahito Miyaoka

123

143

Converbs in Northern Selkup Riita-Liisa Valijärvi

167

Head-Negating Enclitics in Ket Edward J. Vajda

179

Infinitive Constructions in Ket Marina Zinn

203

Index

215

FOREWORD This volume contains 14 chapters, each reflecting a paper originally presented at the “Third International Symposium on the Languages Spoken in Europe and North and Central Asia” (‘LENCA-3’) held at Tomsk State Pedagogical University in Tomsk, a city in south-central Siberia, Russian Federation, during June 27-30, 2006. The symposium was organized to investigate a broad range of issues involving systems of coordination and subordination in complex sentences in the languages of Eurasia. Most of the papers selected for the present volume deal with complex sentence structures characteristic of North Asia, with some of the studies dealing with languages traditionally spoken adjacent regions of the Pacific Rim, notably Central Alaskan Yupik and Ainu, as well as Russian. From an areal perspective, complex sentence morphosyntax across North Asia and the Pacific Rim reveals an unusually high prevalence of suffixation patterns used to signal syntactic subordination. This prevalence for suffixal subordinators involves a variety of genetic groupings, most notably Samoyedic, Turkic, and Tungusic. Less well known is the fact that similar traits play a role in complex sentence formation in some of the region’s language isolates, such as Ket and Ainu, as well. No general study has surveyed the syntax of complex sentences across this area, which is noted both for its complex web of language contact phenomena as well as for its long-established genetic divisions. Much of the data presented in the individual chapters reflect original fieldwork, and several of the contributions focus on critically endangered languages. Nearly all the genetic groupings in the region under consideration are somehow represented, as are all major formal and functional types of complex sentence formation. The chapters themselves are grouped into three sections. Part One consists of a single chapter that serves as an introduction to the types of issues discussed by the book’s individual case studies. Subordination, Coordination: Form, semantics, pragmatics — setting the scene, by Bernard Comrie (Max Planck Institute for Evolutionary Anthropology, Leipzig), examines instances where the distinction between subordination and coordination is not clear-cut, and takes into account the formal, semantic and pragmatic diversity of such complex sentence types. This chapter also serves as a general introduction to the problem of typologizing complex sentence patterns.

viii

FOREWORD

Part Two contains seven chapters devoted to analytic patterns of subordination and coordination. The individual chapters in this section deal primarily with patterns of complex sentence formation based on conjunctions, clause constituency, and word order rather than on the use of special synthetic verb forms, a topic addressed by the articles in Part Three. Speech Report Constructions in Ainu, by Anna Bugaeva (Chiba University, Japan), is based on extensive examples taken from Hokkaido Ainu narratives. It demonstrates that Ainu speech report constructions cannot be analyzed in terms of a sharp dichotomy between reported and direct speech. Adopting a continuum approach to the description of this type of complex sentence, the author provides a novel analysis of constructions intermediate between direct and indirect speech reporting. The Syntax and Pragmatics of Adverbial Clauses in Eastern Khanty, by Andrei Filtchenko (Rice University & Tomsk State Pedagogical University), discusses a wide variety of adverbial clauses in one of the least studied and most highly endangered forms of Khanty (Ostyak), an Ob-Ugrian language of Western Siberia. The syntactic complexity of adverbial clause structure is considered from a diachronic and synchronic perspective in order to identify both past and ongoing patterns of grammaticalization. Grammaticization and Relative Clauses in Eastern Khanty, by Olga Potanina (Tomsk Polytechnic University), demonstrates how Eastern Khanty tends to avoid using postposed relative clauses by preposing the modifying material before its head noun. This has led to the grammaticalization of certain nouns in the role of bleached semantic heads in what originally were relative clause constructions with preposed modifiers. Textual examples deriving from the author’s original fieldwork provide a basis for concluding that most sentences in Eastern Khanty resemble a string of clauses, loosely linked, often without any overt syntactic indication of subordination. Null Arguments in Kumyk Adverbial Clauses, by Linda Humnick (University of Minnesota), explores the syntax and pragmatics of referring expressions in Kumyk, a Turkic language spoken primarily in the Dagestan region of the Russian Federation. An examination of constraints on the use of null subjects in non-finite clauses within a text corpus provides evidence that adverbial clauses in Kumyk have null subjects that are syntactically unrestricted yet pragmatically constrained in the same way as null subjects of matrix clauses. Though Kumyk is spoken near the Caspian Sea and technically falls outside the geographic designation of "North Asia", it represents a language type common across much of Inner Asia. Finite Structures in Forest Enets Subordination: A case study of language change under strong Russian influence, by Olesya Khanina & Andrey Shluinsky (Max Planck Institute for Evolutionary Anthropology, Leipzig,

FOREWORD

ix

Germany & Institute for Linguistics RAS, Moscow), explores recent contactinduced change in Forest Enets, a Samoyedic language of north-central Siberia with only a few dozen remaining speakers. By comparing subordination techniques used by the last speakers of Forest Enets with texts recorded a few generations back, the study is able to document contact-induced change from a system based originally on the extensive use of non-finite verb forms in subordinate clauses to a system making more extensive use of the pattern of subordinating conjunction and finite verb form usage typical for European languages. The Development of Deconverbal Prepositions: Reanalysis or grammaticalization? by Sandra Birzer (University of Regensburg), compares the historical emergence of Turkic and Russian deconverbal adpositions belonging to the semantic categories “due to”, “although”, and “since” in light of the differing typological profiles of Inner Asian and European languages. An examination of semantic bleaching, univerbation, and argument structure change demonstrates that the rise of these subordinators most closely aligns the trajectories of their emergence with definitions of grammaticalization rather than reanalysis. By juxtaposing its analysis of Asian patterns to earlier research on grammaticalization and reanalysis in European languages, the study makes a contribution to the general linguistic understanding of grammaticalization. The last study in this section, Question Particles or What? Open alternative questions in Udeghe, by Maria Tolskaya (Davis Center for Russian and Eurasian Studies, Harvard University) & Inna Tolskaya (University of Tromsö), analyzes the syntax and pragmatics of interrogative devices in the highly endangered Udeghe language spoken in Russia’s Maritime Province and, more broadly, in other Tungus-Manchu languages of Eastern Siberia and the Far East, such as Evenki and Oroqen. Contextual examples include original material from Maria Tolskaya’s fieldwork on Udeghe, as well as examples cited from previous publication. This is the first analysis of this particular variety of interrogative sentence across Tungusic. Its results permit a new typological interpretation of question sentence formation in one of the most widespread North Asian families. Part Three turns to synthetic techniques of marking subordination and coordination. The six chapters in this section deal primarily with the usage of special suffixal forms to convey syntactic subordination or coordination. Such patterns appear to represent a special areal feature of North and Inner Asia, a fact that renders their study of great potential value for helping understand the typological and genetic profile of Eurasian languages in general. Toward a Semantic Typology of Coordination, by Elena Rudnitskaya (Institute of Oriental Studies, Moscow) & Elena Uryson (Vinogradov Russian Language Institute, Moscow), explicitly compares coordination patterns

x

FOREWORD

expressed by synthetic means and those expressed analytically by finite verbs in clauses linked using subordinating conjunctions. The former technique, exemplified by Korean, represents the type of complex sentence formation found widely across North and Inner Asia. Parallel translations of the Korean examples into Russian and English illustrate how various semantic categories of coordination can be expressed synthetically, as well. By examining the contrast between confirmed vs. cancelled expectations in such utterances, the authors provide a new vantage from which to typologize the expression of such meanings using two structurally distinct types of complex sentence formations. Imperatives in Conditional and Concessive Subordinate Clauses, by Nina Dobrushina (State University Higher School of Economics, Moscow), highlights a rare example of what essentially represents a synthetic form of subordination in a European language: the colloquial usage of pseudoimperative forms in Modern Russian to express conditional and concessive subordination. By including comparative data from Turkish, Bulgarian, Hungarian, Estonian and Aghul (a Nakh-Dagestanian language spoken in the Dagestan, Russia, and also Azerbaijan), the author provides compelling textual evidence that the forms in question cannot be regarded as imperatives but rather are distinct grammatical means of conveying certain types of subordination asyndetically. Because Russian is now the most widely spoken language across the entire geographic region under consideration, this study has obvious implications for our understanding of the areal patterning of syntax in contemporary North Asia. Morphological Strategies for ‘Complex Sentences’ and Polysynthesis in Central Alaskan Yupik (Eskimo), by Osahito Miyaoka (Osaka Gakuin University), provides analysis of the extensive use of cyclical suffixation used for building up complex morphological structures that are the syntactic equivalent of complex sentences in most other languages of the world. Though based on examples from Central Alaskan Yupik, the patterns discussed appear to be likewise present, in their general outlines, in forms of Yupik traditionally spoken in the Russian Far East and thus exemplify an extreme example of the type of syntactic subordination through the build-up of suffixes that can be seen as prevalent across North Asia. Converbs in Northern Selkup, by Riitta-Liisa Valijärvi (Uppsala University), provides a morphological analysis of non-finite verb forms used exclusively in subordinate clauses in the Taz dialect of Selkup, a Samoyedic language of central Siberia. Using examples from field elicitation and earlier recordings of folkloric texts, the author examines a wide range of converbial suffixes and semantic types of subordination. Her conclusions hold relevance for the study of Samoyedic syntax in general.

xi

FOREWORD

Head-Negating Enclitics in Ket, by Edward J. Vajda (Western Washington University), provides an analysis of postposed relational morphemes in Ket, an endangered isolate spoken by fewer than 200 people in Central Siberia. It reveals that most semantic types of subordination use enclitics displaying clear typological and semantic parallels to those found in non-finite verb constructions in the adjacent Turkic and Samoyedic languages. Morphological and phonological data are considered to demonstrate that the morphemes in question are actually clitics rather than converbial suffixes or case endings, suggesting a relatively recent development under the influence of language contact. The last chapter, Infinitive Constructions in Ket, Marina Zinn (Tomsk State Pedagogical University), surveys the diverse semantic functions expressed by non-finite forms of the Ket verb in a variety of types of subordinate clauses. All of the examples derive from the author’s recent fieldwork with some of the language’s last remaining speakers and include hitherto undocumented syntactic types of infinitival constructions. The use of non-finite verb forms in such Ket constructions fit very much into prevalent North Asian patterns of complex sentence formation. This volume should be of value to anyone interested in the syntactic typology of Eurasian languages, particularly linguists studying the morphosyntax of complex sentences. The large number of endangered languages surveyed, as well as the significant amount of original field data cited in the articles make this collection important in the documentation of the world’s disappearing languages. Bellingham, Washington June 15, 2008

Edward J. Vajda

Map 1. Asian languages analyzed in the present volume

SUBORDINATION, COORDINATION FORM, SEMANTICS, PRAGMATICS — SETTING THE SCENE

BERNARD COMRIE Max Planck Institute for Evolutionary Anthropology & University of California, Santa Barbara 1.

Introduction My aim in this article, as in the corresponding presentation at LENCA-3, is to “set the scene” with regard to the distinction between subordination and coordination, and in particular to present and discuss problem cases that do not fit readily into the one or the other category. I emphasize that my aim is to identify problems, rather than to posit solutions (although sometimes my remarks will go some distance in this direction); several of the other articles in this collection do propose solutions to particular problems, and I do not want to repeat their arguments. 2.

Clear cases of subordination versus coordination in Western European languages At the outset, it should be noted that there are, of course, many instances of clear distinctions between subordination and coordination, including but not restricted to major Western European languages. In German [deu], for instance, the word order in subordinate clauses is distinct from that in main clauses, including coordinated main clauses.1 In subordinate clauses, the finite verb stands clause-finally, as in the second clause of (1), whereas in main clauses it stands in clause-second position, as in both clauses of (2).

1

Language names are followed by the corresponding ISO 639-3 code in square brackets at first mention. Note that morpheme glossing is partial, to the extent needed for the points being made. Example sentences with no preceding punctuation are presented as grammatical, those with a preceding asterisk as ungrammatical, following the usual convention; beyond this, an exclamation mark indicates a sentence that is syntactically well-formed but semantically or pragmatically anomalous, and a percentage sign a sentence whose grammaticality is up for discussion.

2

BERNARD COMRIE

(1) Das Auto begann zu schleudern, weil die Straße glatt war. the car began to skid because the road slippery was “The car began to skid, because the road was slippery”. (2) Die Straße war glatt und das Auto begann zu schleudern. the road was slippery and the car began to skid “The street was slippery and the car began to skid”. Interestingly, the distinction seems clearly syntactic, rather than semantic, as one can see by comparing (1), with the subordinating conjunction weil “because”, with (3), with the coordinating conjunction denn “for”; although both express cause, weil requires a clause-final finite verb, denn a clause-second one. (3) Das Auto begann zu schleudern, denn die Straße war glatt. the car began to skid for the road was slippery “The car began to skid, for the road was slippery”. Although English [eng] does not have quite such a blatant distinction between subordination and coordination, there are nonetheless several criteria, developed both in traditional grammar and in generative approaches to syntax from the mid-1950s, that provide reasonably robust criteria for distinguishing between subordination and coordination. If one compares example (4) on the one hand with (5) and (6) on the other, then a generalization emerges whereby English allows relativization of a constituent of a main clause that is accompanied by a subordinate clause, as in (4), but not of a constituent of a clause that is conjoined with another clause; this is the so-called Coordinate Structure Constraint of Ross (1986: 97-120).2 Note that just as in German, the distinction is not predictable on semantic grounds, since the causal conjunction for is coordinating, and thus behaves like and, and unlike the subordinating conjunction because. (4) the car [which — began to skid because the road was slippery] (5) *the car [which — began to skid, for the road was slippery] (6) *the street [which — was slippery and the car began to skid]

2

Here and below, a horizontal line is used to indicate the position in the relative clause that is relativized. Following standard scholarly practice, reference is made to the published version of Ross (1986); the work was originally an MIT doctoral dissertation from 1967 entitled “Constraints on variables in syntax” and is also widely cited as such.

SUBORDINATION AND COORDINATION

3

Another test that distinguishes subordination and coordination in English (and many other languages) concerns the order of clauses within the sentence and the occurrence of pronouns and their antecedents, in particular in subject position, in different clauses. In English, a subordinate clause can precede its main clause — by contrast, one cannot have coordination examples like *and the street was slippery, the car began to skid, corresponding to because the street was slippery, the car began to skid. When a subordinate clause precedes its main clause, as in (9), the pronoun can stand in the subordinate clause and thus linearly precede its antecedent in the main clause; this is not possible when the subordinate clause follows its main clause, as in (10). (When the pronoun follows its antecedent, both clause orders are possible, as in (7–8), and of course this order is also possible with coordination, as in Johni left the room and hei was feeling unwell.) (7) Johni left the room because hei was feeling unwell. (8) Because Johni was feeling unwell, hei left the room. (9) Because hei was feeling unwell, Johni left the room. (10) *Hei left the room because Johni was feeling unwell. 3.

Constructions that appear to be coordinate, but behave as if subordinate In this section, we examine examples that present difficulties because they look like instances of coordination, but nonetheless behave as if subordinate. As noted already in §1, the Coordinate Structure Constraint prevents relativization on a constituent of a single conjunct of a coordinate construction, as shown in English examples (11–12): Sentence (11) consists of the coordination of two clauses, and it is impossible to relativize on the direct object of the second conjunct, as seen in (12). (11) [John plays the flute] and [Mary sings madrigals]. (12) *the madrigals [that [John plays the flute] and [Mary sings —]] However, there are examples that seem to have the same structure as (11) but where relativization of the type disallowed in (12) is possible. Consider example (13), the English go and V construction. Provided (13) is interpreted as a tightly knit sequence of actions (my buying the book followed directly from my going to the store), relativization on the object of the second conjunct is possible, as in (14). (13) I went to the store and bought a book.

4

BERNARD COMRIE

(14) the book [that I went to the store and bought —] In colloquial English, even more extended possibilities exist, as in (15), from colloquial British English, which involves relativization on the object of the last of three conjuncts (cf. Our Ethel’s been and gone and bought that dress). (15) Look at the new dress [that our Ethel’s been and gone and bought —]! One might argue that since (14) is not interpreted as simple coordination — it cannot, for instance, be used where my going to the store and my buying the book are completely distinct events — this extra meaning component overrides the apparently formal coordination structure in order to allow relativization. Of course, such an approach would have to be fleshed out in much more detail, since we noted with respect to German example (3) and English example (5) that semantically more specific coordinating conjunctions — denn, for — do not in themselves permit violations of constraints on coordinate constructions. One might go further and note a more specific similarity between the English construction in (13) and the purpose construction in (16) using an infinitive, in which relativization of the object of the infinitive is perfectly possible, as in (17), since the Coordinate Structure Constraint is inapplicable. (16) I went to the store [to buy a book]. (17) the book [that I went to the store [to buy —]] However, in this connection it is also important to note that the constructions in (13) and (16) are not simple alternants with the same meaning. In particular, (16) does not entail that I succeeded in buying the book, as can be seen in the logically consistent (18), while (13) does entail that I bought the book, as can be seen from the logical contradiction of (19). (18) I went to the store to buy a book, but they didn’t have what I wanted, so didn’t buy anything. (19) !I went to the store and bought a book, but they didn’t have what I wanted, so I didn’t buy anything. An interesting set of relevant examples where a seemingly coordinate structure behaves as if non-coordinate is presented for Afrikaans [afr] by Donaldson (1993: 220-221, 226, 385-387). Before considering the Afrikaans construction, it is perhaps worth noting that Dutch [nld] has a functionally analogous, but structurally distinct, construction using verbs of posture (e.g. Dutch zit, Afrikaans sit “sit”) to express progressive aspect, as in (20).

SUBORDINATION AND COORDINATION

5

However, the Dutch construction, whereby the auxiliary is followed by te “to” and an infinitive, has no formal properties of coordination, instead involving subordination of an infinitival construction to the auxiliary verb. (20) Ik zit te lez-en. I sit to read-INF “I am (sitting) reading”. In Afrikaans, however, the relevant construction apparently coordinates the verb sit “sit” and another verb, in (21) lees “read”. Indeed, the simple sequence of words presented in (21) is ambiguous, since in addition to being an expression of progressive aspect, it can also have the interpretation “I sit and read”, i.e. semantic coordination. We will be concerned primarily with the first, semantically non-coordinate interpretation, although contrast with the second, semantically coordinate interpretation will also be relevant. (21) Ek sit en lees. I sit and read “I am (sitting) reading”. In order to explore the relevant details of the construction, we will use a slightly longer example, namely (22), which has, in addition to the progressive aspect interpretation given as its translation, also the interpretation “He sits on the wall and drinks a Windhoek lager”. As we shall see, under many slight variations, the two distinct interpretations lead to formal distinctions, i.e. the progressive construction does not follow the general pattern of coordinate constructions. (22) Ek sit op die muur en drink ’n Windhoek-lager. I sit on the wall and drink a Windhoek-lager “I am (sitting) on the wall drinking a Windhoek lager”. First, we need to ask how the past tense is formed in Afrikaans. Except in the case of a few anomalous verbs, Afrikaans has only a compound past tense, formed with the present tense of the auxiliary het “have, has” and the past participle, which is formed by prefixing ge- to the verb stem. The past participle stands clause-finally. Compare the present tense in (23) and the past tense in (24). (23) Ek drink ’n Windhoek-lager. I drink a Windhoek-lager “I drink a Windhoek lager”.

6

BERNARD COMRIE

(24) Ek het ’n Windhoek-lager I have a Windhoek-lager “I drank a Windhoek lager”.

ge-drink. PSTPTCP-drink

How, then, does one form the past tense of (22)? It turns out that the answer depends on whether the construction is interpreted as a progressive or as two coordinated clauses. In the progressive interpretation, sit en drink is treated as a single unit, moved to clause-final position, with the past participle prefix optionally attached to the first verb (sit) only, as in (25). (25) Hy het op die muur ’n Windhoek-lager (ge-)sit en drink. he has on the wall a Windhoek-lager PSTPTCP-sit and drink “He was (sitting) drinking a Windhoek lager on the wall”. If the interpretation is as two coordinated clauses, then in each clause separately its verb is moved to clause-final position and obligatorily takes the geprefix, as in (26). (26) Hy het op die muur ge-sit en ’n Windhoek-lager ge-drink. he has on the wall PSTPTCP-sit and a Windhoek-lager PSTPTCP-drink “He sat on the wall and drank a Windhoek lager”. Further evidence that the progressive construction is not syntactically an instance of coordination comes from behavior relevant to the Coordinate Structure Constraint. In the progressive construction, there is no problem relativizing on the object of drink “drink”, as in (27), i.e. this does not constitute relativizing a constituent of a single conjunct of a coordinate structure. In (28), by contrast, with the interpretation of two coordinate clauses, relativizing a constituent of one conjunct is ungrammatical, a result of violating the Coordinate Structure Constraint. (27) die Windhoek-lager, wat hy op die muur (ge-)sit en drink the Windhoek-lager REL he on the wall PSTPTCP-sit and drink “the Windhoek lager that he was (sitting) drinking on the wall” (28) *die Windhoek-lager, wat hy op die muur ge-sit en ge-drink the Windhoek-lager REL he on the wall PSTPTCP-sit and PSTPTCP-drink “the Windhoek lager that he was sitting on the wall and drinking” To sum up this section: There are constructions that appear to be coordinate, indeed they are sometimes homophonous with coordinate constructions, but which nonetheless behave syntactically as if subordinate, with one verb form (with its arguments and adjuncts) dependent on another.

SUBORDINATION AND COORDINATION

7

4.

Constructions that appear to be subordinate, but behave as if main clauses Just as one finds instances of clauses that appear to be coordinate, but behave as if they were subordinate, so too one finds examples of clauses that appear to be subordinate, but have at least some properties according to which they do not behave as subordinate, but rather as main clauses. Consider the behavior of tag questions in English — to simplify the discussion, only examples of opposite-polarity tags will be considered. In general, English allows tags to be attached to main clauses, as transparently in (29), but not to subordinate clauses. This means that in an example like (30), one can have a tag question relating to the main clause, but not one relating to the subordinate clause. (29) The Earth is round, isn’t it? (30) Columbus thought that the Earth was round, didn’t he/*wasn’t it? However, a number of clauses that appear to be subordinate nonetheless behave as if they were main clauses with respect to tag questions, in particular clauses that, despite apparent subordinate form, are interpreted as if they were main clauses, this relating to the fact that they express the essential propositional content of a statement. In (31), for instance, the relative clause is technically subordinate, but it is a nonrestrictive relative, and as such actually adds a new piece of information, in much the same way as a main clause would. And as (31) shows, it is possible to have a tag question relating to this nonrestrictive relative clause. (31) Quito — which is the capital of Ecuador, isn’t it? — lies on the equator. Example (32) seems to have the same kind of sentential complement structure as (30), but its most probable use in discourse is very different. While (30) basically tells about Columbus’ set of beliefs, (32) is basically about whether John is leaving tomorrow or not, and it is perfectly natural to attach a tag question relating to this aspect of the sentence’s overall meaning. (Attaching the tag question don’t I, relating to I think, is pragmatically anomalous, since it is odd for me to ask somewhat else for confirmation of what I think.) In this sense, the function of (32) is closer to that of (33), where John is leaving tomorrow is unequivocally a main clause, than to that of the formally more similar (30). (32) I think (that) John’s leaving tomorrow, isn’t he/!don’t I?

8

BERNARD COMRIE

(33) John, I think, is leaving tomorrow, isn’t he? An interesting further extension is provided by the examples in (34) and (35). Example (34) is a straightforward main clause, and therefore takes an opposite-polarity tag as shown. Example (35a) receives the same interpretation, with the addition of the hedge “I think”, although this is not immediately obvious from the structure, which has an affirmative clause John’s leaving tomorrow and a negative main clause I don’t think. But the interpretation of (35a) is as in (35b), i.e. I do think something, namely that John isn’t leaving tomorrow. The tag question in (35a) not only applies to the apparent subordinate rather than to the apparent main clause, but takes its opposite-polarity form according to the negative interpretation of the apparent subordinate clause, not its affirmative form. In other words, the tag question follows meaning rather than form. (34) John’s not leaving tomorrow, is he? (35) a. I don’t think John’s leaving tomorrow, is he? b. John, I think, isn’t leaving tomorrow, is he? This brief section has thus presented a few examples, all from English, of apparently subordinate clauses that behave as if they were not. 5.

Degrees of subordination The examples in previous sections have concerned primarily a mismatch between subordinate or coordinate form and interpretation, with implications for syntactic constructions that are sensitive to the subordination/coordination opposition. In this section, we consider examples that seem to point to degrees of subordination, with some constructions being more subordinate than others, although the distinction is not simply one of form versus interpretation. First, we may note that in English there are in general rather robust judgments across native speakers relating to constraints on extraction, more specifically on relativization, along the lines initially discussed by Ross (1986). For instance, as (36–37) show, there is in principle no constraint in English against relativizing constituents of subordinate clauses. However, (38) shows that there are some constraints, in particular it is not possible to relativize the subject of a finite subordinate clause — although the comparison with (39) shows that things are not quite so simple, since relativizing the subject of a finite subordinate clause is possible if that clause lacks its complementizer (as in (39)), but not if that complementizer is present (as in (38). (36) the person [that I think [that you saw — ]]

9

SUBORDINATION AND COORDINATION

(37) the person [that I think [you saw — ]] (38) *the person [that I think [that — saw you]] (39) the person [that I think [ — saw you]] Given the subtlety of some of the formal distinctions involved, the general robustness of judgment across English speakers is striking. With some other language, one encounters much more variation. In the remainder of this section, I will concentrate on Slavic languages, in particular East Slavic languages (illustrated by Russian [rus]) and West Slavic languages (illustrated by Slovak [slk] and, especially, Polish [pol]). In these languages, there is much more inter-speaker variation, but this turns out to be interesting in that it shows some finite subordinate clauses behaving more canonically with regard to extraction than others. The assignment of no preceding punctuation, asterisks, and percentage signs probably encompasses the judgments of most native speakers of the two languages considered in detail, namely Russian and Polish, though perhaps not all. A useful starting point would be the generalization that extraction – whether for relativization, content question formation, or topicalization – is impossible out of finite subordinate clauses. (In what follows, the discussion of Russian follows Comrie (1973) and references cited there, that of Polish Rothstein (1993: 740).) Thus, in Russian (40) and (42) are impossible, even though their literal English translations are impeccable. There are alternative ways of expressing the ideas intended by (40) and (42), but these avoid relativizing a constituent of a finite subordinate clause. In (41), for instance, relativization is literally on the main clause prepositional phrase of “I know concerning this person that Masha sees him”, while (43) actually involves two content questions each formed from a simple sentence, namely “you think thus” and “someone sees Masha”. (40) *čelovek, [kotorogo ja skazal, [čto person REL.ACC I said that “the person that I said that Masha sees”

Maša Masha

vidit sees

— ]] —

(41) čelovek, [o kotorom ja skazal —, [čto Maša ego vidit]] person about REL.LOC I said — that Masha him sees “the person concerning whom I said that Masha sees him” (42) *Kto vy dumaete, [čto — who you think that — “Who do you think sees Masha?”

vidit sees

Mašu]? Masha

10

BERNARD COMRIE

(43) Kak vy dumaete, kto — vidit Mašu? how you think who — sees Masha “What (lit. how) do you think, who sees Masha?” There is, incidentally, no such general constraint against extracting from infinitival constructions, as illustrated by the grammatical example (44), in contrast with the ungrammatical corresponding finite subordinate clause construction in (45). (44) Vot ogurcy, [kotorye ja obeščal [prinesti —]]. here cucumbers REL I promised bring.INF — “Here are the cucumbers that I promised to bring”. (45) *Vot ogurcy, [kotorye ja obeščal, [čto prinesu —]]. here cucumbers REL I promised that bring.1SG — “Here are the cucumbers that I promised that I would bring”. However, at least some speakers make finer distinctions. For instance, while all or nearly all speakers disallow extraction from an indicative finite subordinate clause, some will tolerate extraction out of a subjunctive finite subordinate clause, as in (46). One interpretation of this is that subjunctive subordinate clauses are in some sense less subordinate than finite subordinate clauses, i.e. that there are different degrees of subordination. There are, however, other possible interpretations in this particular case, e.g. that the Russian subjunctive, which shows gender–number agreement with the subject but no tense distinctions, is less finite than the indicative, which shows not only gender–number or person–number (depending on tense) agreement with the subject, but also tense distinctions. (46) %Vot kniga, [kotoruju ja prikazal, [čtoby on pročital —]]. here book REL.ACC I ordered that.SBJV he read — “Here is the book that I ordered that he read”. Example (47), incidentally, shows that the kind of subject–object asymmetry illustrated by English (36) versus (38) also holds for Russian-speakers who are prepared to accept sentences like (46), where relativization is of the object of the subordinate clause, since they do not accept sentences like (47), where relativization is of the subject of such a clause. (47) *Vot čelovek, [kotoryj nevozmožno, [čtoby — here person REL.NOM impossible that.SBJV — “Here is the person that it is impossible that (he) come”.

prišel]]. came

11

SUBORDINATION AND COORDINATION

In spoken educated Russian, one does incidentally find at least examples of topicalization that violate the constraint against extraction out of a finite subordinate clause, such as the attested (48), which even extracts the topic out of subject position in an indicative subordinate clause. (48) Petrov stranno, [čto — nam pomogal]. Petrov strange, that — us helped “Petrov it’s strange that [he] helped us”. For Slovak, Short (1993: 576-577) notes that constraints against extraction out of a finite subordinate clause are strong, and that while such examples are occasionally heard, they are viewed as “nonce-forms and distortions”. He goes on to say that such examples are rare in print, but does cite example (49), which was apparently judged acceptable not only by the author — or rather, the translator, since it is, perhaps not entirely irrelevantly, a translation from English — but also by the relevant editor.3 (49) %veršovaná tragédia Oulanen, [ktorú Marx považoval, versified tragedy Oulanen REL.ACC Marx thought [že — sa stane Faustom jeho doby]]. that — REFL become Faust his age.GEN “...the verse tragedy Oulanen, which Marx thought would become the Faust of his age” Rothstein (1993: 740) reports succinctly but with interesting nuances on extraction in Polish, with examples involving content questions. Extraction of the object of a subjunctive subordinate clause is apparently unobjectionable, as in (50) (cf. Russian (46), while extraction of the subject of such a clause is rejected (example (51), although some speakers are more tolerant of this latter possibility if the main clause has no overt subject separating the interrogative pronoun from its verb, as in (52). (Polish is a “pro-drop/null-subject” language.) (50) Co Janek chce, [żeby Maria what Janek wants that.SBJV Maria “What does Janek want that Maria buy?” (51) *Kto 3

Janek

chce, [żeby

—

kupiła bought

kupił

—]? — gazetę]?

The accusative case of the relative pronoun in (49) is perhaps surprising, given that the position relativized is subject of the finite subordinate clause. However, in a number of languages this use of the accusative is attested, both in languages like Slovak where the overall construction is marginal, and in languages like English where it is not (e.g. %the man whom I think left for the man who I think left).

12

BERNARD COMRIE

who Janek wants that.SBJV — bought “Who does Janek want to buy the paper?” (52) %Kto chcesz, [żeby — kupił who want.2SG that.SBJV — bought “Who do you want to buy the paper?”

paper

gazetę]? paper

Although in Polish extraction out of an indicative finite subordinate clause is in general impossible, some speakers do allow such extraction with particular main-clause verbs. Thus, (53) and (54) are acceptable to some speakers, though (54) to fewer than (53). In recent work on Polish, main clause verbs that allow such extraction, such as “say” in (53) and “think” in (54), have come to be called ‘bridge verbs’. Note that for speakers who accept (53) but not (54), (53) is arguably less subordinate than (54). (53) %Co Janek mówi, [że Maria kupiła —]? what Janek says that Maria bought — “What does Janek say that Maria bought?” (54) %Co Janek myśli, [że Maria kupiła —]? what Janek thinks that Maria bought — “What does Janek think that Maria bought?” In section 5, we have tried to show that languages that have tests for distinguishing between subordinate clauses and other constructions may end up not with a clear-cut dichotomy, but rather with a cline of more and less subordinate constructions. In the Slavic languages cited, finite (or perhaps: more finite) constructions are more subordinate than nonfinite ones; in Polish, some main-clause verbs define more subordinate constructions than others. 6.

Subordination versus coordination according to interpretation? An interesting question that arises is whether a language can have a subordinate/coordinate opposition corresponding to differences of meaning or interpretation even in the apparent absence of a formal distinction. In the present section, I wish to examine some possible implications of this idea against data from Haruai [tmd].4 As an introduction to the Haruai material, one might consider a distinction found in English, and certainly many other European languages, where subordinate clauses can rather freely occur internal to a main clause, as in (55), while coordinate clauses cannot, as in (56) — the latter example is at best stylistically marked, and may not be acceptable to all 4

Haruai data are from my own fieldwork, conducted in Papua New Guinea in 1985–1986 and funded by the National Science Foundation under grant BNS-8504293.

SUBORDINATION AND COORDINATION

13

speakers. Can one then use the possibility of internal (as opposed to peripheral) occurrence as a criterion for subordinate status? (55) John, when you see him tomorrow, will be wearing a green shirt. (56) %John — and you will see him tomorrow — will be wearing a green shirt. Haruai has one instance of a clausal construction that is clearly subordinate, namely the relative clause construction, as illustrated in (57). The type illustrated here is the right-headed relative clause construction (Haruai also has internally headed relative clauses), in which the head is immediately preceded by a clause, which latter has the same basic structure as a Haruai main clause, including the same kinds of finite verb forms as are found in main clauses, except that this finite verb form, necessarily clause-final in a relative clause, lacks any sentence particle such as declarative -a. As (57) shows, relative clauses can be internal to the main clause. (57) n [nagö nwgw-l-ö] hön pay-n-a. I you see-PRS-2SG pig hit-FUT.1SG-DECL “I will hit the pig [that you saw]”. Haruai has a number of constructions that are translation equivalents to English subordinate clauses, sometimes with a broader range then just subordinate clauses in English. For instance, simple juxtaposition of two main clauses, as in (58), gives a temporal or conditional clause interpretation to the first clause. (58) rwö watk h-ön-a, an environment evening come-FUT.3SG-DECL we “When evening falls, we will sleep”.

hölm-n-N-a. sleep-FUT-1PL-DECL

This relation can be made more explicit by ending the first clause with the neutral demonstrative akw, as in (59); note that, like all post-verbal elements, a demonstrative in this position excludes the possibility of sentence particles — otherwise, affirmative indicative main clauses require a sentence-final particle, the neutral one being declarative -a. (59) rwö watk h-ön akw, an hölm-n-N-a. environment evening come-FUT.3SG-DECL that we sleep-FUT-1PL-DECL “When evening falls, we will sleep”. Haruai also has a number of dependent verb forms that express semantically rather unspecific relations between the dependent and independent clauses, sometimes corresponding to subordinate temporal clauses in English, some-

14

BERNARD COMRIE

times to coordination. One is the conjunctive verb form in -aN (singular; there is a corresponding plural in -N), as in (60). (60) ap.mag yön-aN, an nm-n-N-a. food cook-CNJV.SG we eat-FUT-1PL-DECL “The food will cook, and we will eat”. “When the food cooks, we will eat”. Finally, Haruai has switch-reference verb forms, same-subject -ön and different-subject -mön, as in (61–62), differentiated as to whether the subjects of the two clauses are the same or different, both usually corresponding to coordination in English, though sometimes with a more felicitous translation involving an explicit temporal or causal conjunction. (61) mö hön rg ng ng-ön, köp-a dw-öN-a. woman pig stone put put-SS leaf-SUFF go-PST.3SG-DECL “A woman built a pig house, and went for some leaves”. (62) nöbö mörö wök p-g-mön, glñN gyo.gyö r-öN-a. man garden clear get-PFV-DS bushfowl gyo.gyö do-PST.3SG-DECL “A man cleared his garden, and a bushfowl went ‘gyo gyö’”. I have encountered no reliable instances of the constructions in (58–59) occurring clause-internally, i.e. it is always the case that the clause interpreted as temporal/conditional precedes the other clause, whether or not that first clause has a clause-final demonstrative. For clauses with a conjunctive verb form, there are examples that look prima facie like clause-internal occurrences, such as (63); note that it is more usual than not for Haruai imperative clauses to have an overt second person subject pronoun, although the initial pronoun in (63) could perhaps also be interpreted as a vocative. (63) ñgö [hön röpn-aN] hön-a py. you.PL pig emerge-CNJV.SG pig-SUFF shoot(-IMP.2PL) “You, [when the pigs emerge,] shoot some pigs”. Some instances of what are at least historically conjunctive and switch-reference verb forms do occur clause-internally, but it is arguable that these have been lexicalized. For instance, Haruai has a verb ölöw “[to] dawn”, and like any other verb this can give rise to a different-subject form as in (64a), or a conjunctive form as in (64c). However, these are also the usual translation equivalents of “tomorrow”. In their lexicalized use they readily occur clause-internally, as in (65). But given that at least some lexicalization has

15

SUBORDINATION AND COORDINATION

taken place, it is not clear that the clause-internal adverbial in (65) is to be equated with a productive different-subject or conjunctive verb form. (64) a) ölöw-mön dawn-DS

b) ölöwmön tomorrow

(65) n ölöwmön/ölöwaN I tomorrow “I will go tomorrow”.

c) ölöw-aN dawn-CNJV

d) ölöwaN tomorrow

dy-n-a. go-FUT.1SG-DECL

Nonetheless, I do have at least one clear example in my corpus of a clause-internal occurrence of a different-subject verb form, namely (66). This was said fluently by a speaker at the beginning of an orally narrated story, in order to indicate in advance the topic of the story. The story concerns a man who is trapped in a cave by a sorcerer, but who escapes by using his own magic power in order to persuade birds to peck a hole in the cave through which he leaves. The structure of the relevant part of (66) is the head noun haywö “story” and the preceding text in square brackets. The outer square brackets define a finite clause, formally indistinguishable from a relative clause, which states the content of the story, i.e. that “the birds bored [through] the stone”. The inner square brackets define a different-subject clause that is interpreted as a temporal clause, i.e. the birds bored through the stone “when the man climbed into the cave”. This different-subject clause is clearly internal to the clause on which it is dependent, being located with the subject of that clause to the left and the object and verb to the right. (66) m=akw=m yöwr=y [yöwr [nöbö mgan wölw-mön] well.now bird=INDF bird man hole climb-DS rg ybdw-N] haywö wöp=y=k wc-n-a. stone bore-PST.3PL story now tell-FUT.1SG-DECL “Well now, now I will tell the story of some birds, of how [the birds, [when the man went into the cave,] bored through the stone]”. In sum, one has to look hard to find clear instances of internal occurrences of dependent clauses in Haruai, but they do exist, and suggest that at least when such clauses are interpreted as, for instance, temporal, rather than just as coordination, they are treated as subordinate. Clearly, more work is needed in order to justify this analysis more fully.

16

BERNARD COMRIE

7.

Conclusions In this article, I have examined a number of problems at the subordination/coordination interface, in particular clauses that look like instance of coordination but behave as if subordinate, and of subordinate clauses that behave as if main clauses; clauses that seem to suggest that the opposition is a question of degree rather than a strict dichotomy; and instances where the same formal structure seems to behave now as subordinate, now as coordinate, depending on its interpretation. As indicated in the Introduction, I have concentrated more on identifying problems than on suggesting solutions. But I hope at least to have shown that clear-cut cases of subordination versus coordination as in (1–10) are far from the whole story. Abbreviations – accusative, CNJV – conjunctive, DECL – declarative, DS – different subject, FUT – future, GEN – genitive, IMP – imperative, INDF – indefinite, INF – infinitive, LOC – locative, NOM – nominative, PFV – perfective, PL – plural, PRS – present, PST – past, PSTPTCP – past participle, REFL – reflexive, REL – relative, SBJV – subjunctive, SG – singular, SS – same subject, SUFF – suffix ACC

REFERENCES Comrie, Bernard. 1973. “Clause Structure and Movement Constraints in Russian”. You Take the High Node and I’ll Take the Low Node: Papers from the Comparative Syntax Festival ed. by Claudia Corum, T. Cedric SmithStark & Ann Weiser, 291-304. Chicago: Chicago Linguistic Society. Comrie, Bernard & Greville G. Corbett, eds. 1993. The Slavonic Languages. London & New York: Routledge. Donaldson, Bruce C. 1993. A Grammar of Afrikaans. Berlin & New York: Mouton de Gruyter. Short, David. 1993. “Slovak”. The Slavonic Languages ed. by Bernard Comrie & Greville G. Corbett, 533-592. London & New York: Routledge. Ross, John Robert. 1986. Infinite Syntax! Norwood, New Jersey: Ablex. Rothstein, Robert A. 1993. “Polish”. The Slavonic Languages ed. by Bernard Comrie & Greville G. Corbett, 686-748. London & New York: Routledge.

SPEECH REPORT CONSTRUCTIONS IN AINU1 ANNA BUGAEVA Chiba University (ELDP) & La Trobe University, RCLT 1.

Introduction This paper focuses on types of speech report constructions in Hokkaido Ainu that have been previously analyzed in terms of the dichotomy between direct and indirect speech. Following Aikhenvald (forthcoming) and other authors (Güldemann & von Roncador 2002), I adopt a continuum approach, which implies that a language may have so-called semi-direct speech constructions occupying a middle ground between direct and indirect speech. 1.1 Genetic, dialectal and typological profile of Ainu Ainu is a critically endangered language of unknown genetic affiliation, which shows considerable dialectal variation. The three primary divisions are geographically based, and distinguish between the dialects once spoken on Hokkaido, Sakhalin, and the Kurile Islands. Sakhalin and the Kuriles form part of the Russian Federation today, with Hokkaido being the last autochthonous location of native speakers. The Hokkaido dialects can be roughly divided into Northeastern (Northern, Eastern, and Central) and Southwestern (Southern and Southwestern) groups, which are further subdivided into local sub-dialectal forms (see Hattori 1964: 18). Ainu is polysynthetic and agglutinating, with SV/AOV constituent order. It is predominantly head marking with a few elements of dependent marking. Morphological expression of arguments is tripartite: A, S and O are distinct in 1PL verbal agreement, viz. the prefix ci= marks A, the suffix =as S, and un= O for the first person exclusive, and a= A, =an S, and i= O for inclusive. The personal pronouns in A/S and O position are often omitted because the person and number of both subject and object are marked on predicates. 1

I would like to thank A. Aikhenvald, E. Geniušiene, T. Güldemann, M. Daniel, S. Overall, H. Nakagawa, S. Tida, B. Hellwig and J. Cleary-Kemp for valuable comments on earlier versions of this paper, and the LENCA 3 participants who have discussed with me the problems of “logophoricity” (B. Comrie, O. Miyaoka, and V. Plungjan). None of them necessarily shares my views expressed here or is responsible for any errors. My deepest heartfelt thanks go to my late Ainu consultant Mrs. I. Oda and to my former supervisor T. Satoo.

18

ANNA BUGAEVA

1.2 Preliminaries for analyzing Ainu speech report constructions All languages have means to report words spoken directly by the reporter or by someone else. Reported speech is always “speech within speech”, as characterized by Jakobson (1990: 130; quoted after Aikhenvald 2005: 2). It always implies at least two speech situations. First is the actual external speech situation (often implicit) with its speaker (reporter), addressee, the time of speaking, and the location of the speaker. Second is the prior internal speech situation with its own speaker (author), addressee, etc. Every speech situation has a reference point. Prototypical direct speech is reported from the point of view of the author of speech, just as it was originally told, including all the references to the internal speech situation: (1) Jane said, “I shall bring you here tomorrow”. Prototypical indirect speech is reported from the point of view of the reporter; hence we observe the corresponding adjustment of personal, temporal and spatial deixis of the internal speech situation to match the perspective of the reporter in the external speech situation: (2) Jane said that she would bring him/me there on the next day. Both cases involve a complex sentence containing a matrix clause, referred to below as the reporting clause, or RC. The predicate of the RC normally contains a reporting verb (RV) that encodes the author as subject. Finally, there is the subordinate or embedded clause, i.e. the quote that conveys direct or indirect speech (DS or IS), and the reporting marker (RM). It is well known that a shift in person is the most prominent feature distinguishing direct from indirect speech since some languages may lack the adjustment of categories other than person. “The category of person is clearly definable with reference to the notion of participant-roles: the “first” person is used by the speaker to refer to himself as subject of the discourse; the “second” person is used to refer to the hearer [addressee]; and the third person is used to refer to persons or things other than the speaker or hearer [addressee]” (Lyons 1968: 276). In the case of reported speech there are two speakers, thus two possible “I”s. When reporting, a language can use “I” either with reference to the author (that is DS) or with reference to the reporter (that is, IS). However, in some contexts, the use of the first person for the author may be ambiguous with the person of the reporter. To avoid this kind of ambiguity the language may employ a special indirect strategy for reference to the author, while other references occur as they would have been used in the DS, and that is regarded here as a semi-direct speech (SS).

19

SPEECH REPORT CONSTRUCTIONS IN AINU

2. Speech report constructions in Ainu with a focus on person marking 2.1 Direct speech According to Haiman (1989: 131) and Güldemann (forthcoming: 9), direct speech can be expected to exist in every language and is thus a universal phenomenon, while in a number of languages reported speech categories incorporating partly or completely the perspective of the reporter are missing or extremely restricted. However, Tamura’s (1988: 74) description of Ainu (SH) goes against the above-mentioned generalization as the following implies that prototypical direct speech (3c) is ruled out. When other people’s [the third or second persons’] words are quoted, [the author’s] words spoken in the first person are changed to the indefinite person in quoted speech [(3a), cf. (3c)]. The first person indicates the person who is actually speaking [=reporter], and the first person is not used in quoting the speech of others, [i.e. the first person is used only in so-called self-reporting contexts, as in (3b)].

(3a) “asinuma

arpa=an kusu ne” sekor Ø=hawean. go.SG=IND.S intention COP QUOT 3.S=say.SG “(She) said, ‘I [literally, someone] will go’.” (T4, 74; COL); Saru, SH. IND.SG

(3b) “káni k=arpa wa eci=kor-e kusu 1SG 1SG.S=go.SG and 1SG.A+2SG.O=have-CAUS intention ne na” sekor ku=hawean… COP FIN QUOT 1SG.S=say.SG “(I) said, ‘I’ll go for your sake’.” (T1, 377; COL); Saru, SH. (3c) ?“káni k=arpa kusu ne” sekor 1SG 1SG.S=go.SG intention COP QUOT “(She) said, ‘I’ll go’.” (constructed example: A.B.)

Ø=hawean. 3.S=say.SG

The question here is how ungrammatical is genuine direct speech in Ainu? When one looks for analyses in the published literature, one finds that instances of direct speech have been identified as grammatically possible options (Tamura 1972), though the same author has not treated similar examples in her latest description of Ainu (Tamura 1988). (4) “k=arpa wa eci=kor-e 1SG.S=go and 1SG.S+2SG.O=have-CAUS kusu ne na” sekor Ø=hawean… intention COP FIN QUOT 3.S=say.SG “(S/he) said, ‘I’ll go for your sake’.” (T1, 377; COL); Saru, SH.

20

ANNA BUGAEVA

(5) “paye=as wa arki=as na”. sekor go.PL=1PL.S and come.PL=1PL.S FIN QUOT eci=haweoka a p 2PL.S=say.PL PERF but “You said, ‘I’ll be gone’, but…” (T1, 378; COL); Saru, SH. Present research revealed that relatively short DS reports may occasionally appear in typical semi-direct style prosaic folktales (namely in uwepeker) and refer to the speech of non-protagonists, as in (6), or in conversational turns of different characters in stories narrated in the third person (7). Such use of DS makes the folklore texts polyphonous and the description more vivid. (6) “eci=ekanok kus ne na”, sekor Ø=hawean 1SG.A+2PL.O=meet going.to COP FIN QUOT 3.S=say.SG “(The uncle) said, ‘I’ll come and meet you’.” (B 124; U); Chitose, SH. (7) “mak ku=iki wa eci=tura hawe?” sekor saru how 1SG.S=do and 1SG.A+2SG.O=follow REP.EV QUOT monkey Ø=hawean akusu “…en=ka ta a yak eci=tura 3.S=say.SG then 1SG.O=above at sit if 2SG.A+1SG.O=follow kusu ne na” sekor Ø=hawean intention COP FIN QUOT 3.S=say.SG “The monkey said, ‘How can I follow you?’ (The tortoise) said, ‘If (you) sit on me, you will follow me’.” (T3, 28; U); Saru, SH. In (4–7), no single pronominal affix shows a reporter-oriented shift, and none of the other properties do, which proves that Ainu does have a DSconstruction. 2.2 Semi-direct speech As was mentioned in §2.1, it is the so-called indefinite person that is often used to refer to a participant in the quote in the case of his/her coreference with the third/second person singular of the main clause. “Indefinite” is a general label employed by Tamura for a set of verbal markers, viz. a= for SG/PL.A, =an for SG/PL.S, and i= for SG/PL.O, which are used not only in their proper function to refer to a generalized or indefinite subject and object, but also have a number of special usages: “1SG/PL in quotes” under consideration, 1PL inclusive, and 2SG/PL honorific. As to the function “1SG/PL in quotes”, in Eastern and Central Hokkaido dialects, the above-mentioned affixes crossreference on verbs the originally non-singular pronoun anokay (PL), which is employed to refer to both plural and singular referents. Verbs occur in plural forms (8) and it is only the context that helps to resolve the ambiguity of

SPEECH REPORT CONSTRUCTIONS IN AINU

21

SG/PL. Southern Hokkaido dialects are peculiar in that they have additionally developed a special singular pronoun asinuma “one, someone” (Tamura 1988: 22-24; etymologized as a= and sinuma <3.SG.PR>, for the sake of convenience asinuma is translated as “I”) which is cross-referenced on verbs by the above-mentioned originally non-singular indefinite affixes, and verbs occur in singular forms (9). (8) acapo Ø=cinita a wa ne eper ene Ø=itak i, uncle 3.S=dream PERF and this bear like.this 3.S=say NR “anokay …kamuy mosir ine paye=an…” sekor Ø=ne IND.PL god land to go.PL=IND.S QUOT 3.A=COP “The uncle had a dream. The bear said, ‘I [someone] will go to the land of gods (=die)’.” (SK 57-58; COL); Asahikawa, CH. (9) néa pon menoko “… a=Ø=kor katkemat, asinuma that young woman IND.A=3.O=have lady IND.SG ka, a=Ø=tura kusu ne…” sekor Ø=hawean also IND.A=3.O=follow intention COP QUOT 3.S=say.SG “That young woman said, ‘(If you say so), my lady, I [someone] will also follow you’.” (T2, 20-22; U); Saru, SH. It is worth mentioning that, in the case of coreference of the third/second person author with a participant in the quote, it is not only the first person singular but plural as well that is changed to the indefinite person. (10) “[aoka2] oya-pa suy arki=an kusu ne na!”sekor IND.PL next-year again come.PL=IND.S intention COP FIN QUOT Ø=haweoka kor Ø=paye wa orano k=Ø=okaramotte-pa 3.S=say.PL when 3.S=go.PLand then 1SG.A=3.O=be.reluctant.to.part-PL “They left saying, ‘we [some people] will come again next year’, and I felt reluctant to part with them”. (T4, 75); Saru, SH. In this paper, I propose that Tamura’s “first person in quotes” usage of the indefinite person in Southern Hokkaido dialects, viz. asinuma SG, aoka PL, and the corresponding verbal affixes a= for SG/PL.A; SG/PL.POSS, =an for SG/PL.S, and i= for SG/PL.O, and anokay SG/PL and the verbal affixes in Eastern and Central dialects should be regarded as special logophoric markers (henceforth [self] in translations and LOG in glossing, cf. Tamura’s interpretation) existing in addition to the normal set of personal markers, since they fit with Hagège’s (1974: 287) definition of logophoricity “turning to the 2

The same forms appear, with slight differences, in different dialect groups, cf. anokay in (8).

22

ANNA BUGAEVA

discourse” which he proposed in his discussion of African languages: The term ‘logophoric’ is here proposed to designate a particular category of substitution elements (substitutes), personal and possessive, which refer to the author of the discourse or to a participant whose thoughts are reported.

Tamura’s characterization of the indefinite person as “first person in quotes” is considered inappropriate because this label is hardly intelligible for nonAinologists, and may be easily mistaken for a kind of “genuine” first person that the speaker uses to refer to himself. Although it is clear from Tamura’s (2000: 74) description that the use of the indefinite person in the case of coreference of the third/second person author with a participant in the quote reflects the reporter’s perspective (as that of IS), she notes that it “is not a so-called indirect narration”, since there separately exists a genuine IS-construction (cf. §2.3). According to Tamura’s (2000: 74-75) analysis which is implicitly dichotomous, SS is to be classified as DS; note that there is no shift in second person reference (11) and all other properties are also the same as those of DS. (11) “iwan pa Ø=ek yak, a=e=ekanok kus ne six year 3.S=come if LOG.A=2SG.O=meet intention COP sekor, Ø=hawean QUOT 3.S=say.SG “’I [lit. self] will meet you in six years’, said (my husband)”. (B 145; KY); Chitose, SH.

na” FIN

However, if we accept Tamura’s view that reports with ‘indefinite person’ are DS, we would either end up having in Ainu two DS constructions with distinct person marking (cf. §2.1), or having to demote the genuine DS-construction, as in examples (4–7), to some marginal status (a kind of slip of the tongue). The latter decision is implicitly accepted in Tamura’s Ainu grammar (1988), since examples of genuine DS have not been included in the discussion. My claim is that it is only a continuum approach to reported speech that can account for these two speech report constructions. That is, the speech report constructions in question should be identified as DS and SS, which differ in degrees of directness and have different discourse functions, which will be discussed in §3. Syntactically, DS and SS in Ainu are encoded by an oblique, non-core construction different from any clausal constituents (italicized terms are from Aikhenvald 2005: 22), which has some resemblance to the subordinate adverbial clause of manner. The border between DS and RC is marked by the RM sekor (Southern and Central Hokkaido), which is an adverbial conjunction.

SPEECH REPORT CONSTRUCTIONS IN AINU

23

The present research has shown that DS (and SS) may be introduced not only by intransitive RVs, as was argued in the previous research (Chiri 1942: 487), but by transitive RVs as well (cf. Table 1). However, these DS/SS-taking transitive RVs are licensed only for Addressee objects and thus have no license for Theme objects to which DS/SS would correspond (e.g. ko-caranke “to scold someone”, since *“to scold for something” is ruled out); hence the above mentioned interpretation of the syntactic function of DS/SS. 2.3 Indirect speech There is also a straightforward indirect speech construction in Ainu with a typical shift in personal deixis to fit in with the perspective of the reporter: the words spoken in the first person in the original speech are accordingly changed to the third person in indirect speech (12a), except for the cases of coreferentiality between the author and reporter, cf. (12b); the words spoken in the second person in original speech are changed to the first person, if the reporter is coreferential with the original addressee, as in (13), and to the third person, if not coreferential. (12a) [Ø=arpa kusu ne yak]o Ø=Ø=ye 3.S=go.SG intention COP COMP 3.A=3.O=say “Hei said that hei/j would go”. (T1, 377; COL); Saru, SH. (12b) [iosi ku=oman kun-i]o ku=Ø=ye afterwards 1SG.S=go.SG going.to-COMP 1SG.S=3.O=say “I said I would go later”. (SK 57-58; COL); Asahikawa, CH. (13)

muka un katkemat Ø=ek akusu, Mukawa from woman 3.S=come then [ku=Ø=kor ka… Ø=Ø=hok, yak]o Ø=Ø=ye 1SG.A=3.O=have thread 3.A=3.O=buy COMP 3.A=3.O=say “A womani from Mukawa came and said that shei/j would buy my threads”. (S 72; COL); Saru, SH.

Unlike other languages, Ainu does not employ logophoric markers to mark the coreferentiality of the author with a participant in the quote in IS, although it does so in SS, as discussed in the preceding section. As a consequence, any third person within the quote may, in principle, refer either to the author, as in (12a) and (13), or to a third party, and it is only the context that can help to resolve the ambiguity. Syntactically, IS is interpreted as a typical embedded complement clause in the O function since it may be followed by transitive RVs only, see Table 1. There are three complementizers: yak, kun-ak and kun-i to mark the syntactic

24

ANNA BUGAEVA

link between IS and RC. Only yak is a semantically neutral complementizer; kun-ak and kun-i are not pure complementizers, because each of them consists of two morphemes: the latter (-(y)ak, -(h)i) are proper complementizers and the former (kun-) may render a number of modal meanings. (14) [kamuy Ø=ne kun-i]O a=Ø=ramu god 3.A=COP surely-COMP LOG.A=3.O=think “I [= self] thought that was surely a god”. (B 139; KY); Chitose, SH. It appears that the respective sets of RVs for DS/SS and IS are mutually exclusive (cf. Table 1). Interestingly, the set of DS/SS-taking verbs appears to be much more extensive than the set of IS-taking verbs, contrary to Aikhenvald’s (2005: 29) generalization that “if direct and indirect speech are distinguished at all, the set of indirect-speech-taking verbs tends to be more extensive”. The reason why Ainu goes against Aikhenvald’s generalization is probably that direct speech here also includes semi-direct speech, since both are encoded by the same sekor-marked construction. 3. Functions of speech report constructions in Ainu In this paper, I argue that DS, SS, and IS in Ainu should be regarded as stylistic and discourse-organizing options; some of them, however, have been conventionalized as primary styles in certain folklore genres. The famous Ainu folklore genres, namely yukar and uwepeker, are traditionally narrated in SS style, i.e. the protagonist is referred to from the reporter’s perspective with the respective logophoric markers because these folktales commonly have the structure of the RC-construction in question: the whole story is in fact a single quote. In colloquial Ainu, SS seems to be a preferable option in those contexts where the use of the first person pronouns and affixes may be ambiguous with the person of the reporter, as in (3a). Some colloquial Ainu texts exhibit interplay between DS and SS, which may be regarded as a discourse-organizing device. I suggest that in contexts such as that shown in (15), DS is used for foregrounding a discourse event and SS for backgrounding. (15) “ne ukuran acapo Ø=sirepa, Ø=ikopepka this evening uncle 3.S=arrive 3.S=talk.about.one’s.experience hawe ene an i, “a=yup-iSS tura REP.EV like.this be.SG NR LOG.POSS=elder.brother-POSS with kim un kuca ot ta okay=anSS a wa mountains be hunting.hut place at be.PL=LOG.S PERF and kunnano ku=yup-iDS ene Ø=itak i,

SPEECH REPORT CONSTRUCTIONS IN AINU

25

in.the.morning 1SG.POSS=elder.brother-POSS like.this 3.S=say NR ukuran utar-pa patek kunne kosonte Ø=Ø=mi wa Ø=rimse, evening people-PL only black dress 3.A=3.O=wear and 3.S=dance ne i ta hure sarampe ku=Ø=miDS wa aynu kut this time at red silk.dress 1SG.A=3.O=wear and Ainu belt ku=Ø=e-kut-korDS wa so-us-i ta 1SG.A=3.O=with.APPL-belt-have and seating.mat-adhere.to-place at sinen ne ku=anDS” sekor ku=wen-tarapDS” alone as 1SG.S=be QUOT 1SG.S=be.bad-dream sekor Ø=itak kusu… an=Ø=nukarSS ka eaykap QUOT 3.S=say because LOG.A=3.O=look even be.unable kusu… sap=anSS ru ne” sekor Ø=itak because return.PL=LOG.S INFR.EV COP QUOT 3.S=say kor Ø=ciskoyoyse and 3.S=break.into.tears “In the evening, the uncle arrived and talked about his experience as follows: ‘I [self] stayed at the hunting hut in the mountains together with my [self’s] brother (= the reporter’s father). In the morning, my brother said as follows: ‘I had a bad dream: In the evening, only the chiefs were wearing black dresses and dancing. At that time, I was wearing a red dress and had a red Ainu belt. I was sitting alone in the corner’, he said. …(At the bottom of the stream there was something that used to be a face and now looked like a mess.) ‘I [self] could not look at (it) and returned (without having looked well at it)’. Said (the uncle) and broke into tears”. (SK 49; COL); Asahikawa, CH. The reporter is an Ainu lady who is recalling a tragic episode from her childhood, i.e. the death of her father who went hunting with his friend and was killed by a bear. In (15), the lady is reporting the speech of her father’s friend (acapo “uncle”), which also contains the embedded speech of her father. I assume that the father’s speech is cast as DS (the use of first person markers: “I”) because it is of more importance for the reporter, and the friend’s speech is cast as SS (the use of logophoric markers meaning “self”) because it is of less importance. There is a strong preference for SS and DS over IS in all kinds of texts. Due to the lack of data, it is hard to give a precise answer to the question of how to use IS in Ainu. I assume that IS might have a stylistic and discourseorganizing backgrounding function, since it is found only in the speech of nonprotagonist participants who always have less discourse salience than the protagonist. I assume that IS proper has emerged rather recently as an

26

ANNA BUGAEVA

extension of the complementation strategy which is commonly employed for embedding expressions of thought, knowledge, emotion, and intention. 4. Summary This article has demonstrated that Ainu possesses three main types of reported speech constructions, with person deixis being a core principle in determining the orientation of reported speech: direct speech (DS) with the author’s perspective, indirect speech (IS) with the reporter’s perspective, and semi-direct speech (SS) with the combined perspective of the two. I have shown that genuine DS reports exist in colloquial and folklore Ainu, and therefore that DS is a full-fledged reported speech category. It has been emphasized that SS cannot be analyzed as a kind of DS, contrary to Tamura (1988), because the use of so-called “indefinite person” in the case of coreference of the third/second person author with a participant in the quote indicates a consistent reporter-oriented shift in pronominal reference, while other references occur as they would have been used in the original speech. I proposed to identify Tamura’s “first person in quotes” usage of the indefinite person in Southern Hokkaido dialects, viz. asinuma SG, aoka PL, and the corresponding verbal affixes a= for SG/PL.A, =an for SG/PL.S, and i= for SG/PL.O, as a special set of logophoric markers which exists in addition to the normal set of personal markers and is employed to refer to the person whose words or thoughts are being reported in a stretch of speech, since they fit in with Hagège’s (1974: 287) definition of logophoricity. My research shows that DS (and SS) may be introduced not only by intransitive reporting verbs, as has been argued in previous research (Chiri 1942: 487), but by transitive reporting verbs as well. It was shown that, in Ainu, the respective sets of reporting verbs for DS/SS and IS are mutually exclusive. The set of DS/SS-taking verbs appears to be much more extensive than the set of IS-taking verbs (see Table 1) which goes against Aikhenvald’s generalization (2005: 29) that “if direct and indirect speech are distinguished at all, the set of IS-taking verbs tends to be more extensive”. This is probably due to the fact that direct speech here also includes semi-direct speech, since both are encoded by the same sekor-marked construction. To summarize, I have argued that DS, SS, and IS should be regarded as stylistic and discourse-organizing options; some of them, however, have been conventionalized as primary styles in certain folklore genres. SS and DS reports are frequent in colloquial and folklore Ainu. IS reports referring to speech proper are rare anywhere, but those referring to thought, knowledge, emotion, and intention are rather common.

27

SPEECH REPORT CONSTRUCTIONS IN AINU

DS/SS-taking RVs

IS-taking RVs

vi

vt

speech RVs proper

cis “cry”; hawean ‘say”; isoytak “tell stories”; itak “speak, say”; rek “sing (of birds)”; sonkoanpa “deliver message”; ukoisoytak(pa)”tell stories to e.o.”; upaskuma “tell old stories, preaching”; uwepekennu “ask sb about the reason”; uwepeker “tell folktales”; wentarap “dream”; yayapapu “apologize”; yayeisoytak “tell the story about oneself”; yayeinonnoitak “pray to gods for oneself”

koyki “scold sb”; hotuyekar “call sb”; itakamkire “promise to sb”; kaspaotte “order sb”; kocaranke “argue with sb”; koitak “speak with sb”; koitakmuye “leave a message for sb”; kosakayokar “shout at sb”; kouepekennu(pa) “ask sb about the reason”; koyki “scold sb”; reko “give a name to sb”; ye “say sth/to sb”

substitutes of speech RVs

an “exist”

ne “be sb”

thought RVs

other

Total

ramuan/raman “think; feel”; yaynu “think”

vt/vd ye “say sth/to sb”; takar “dream of sth”; ekaspaotte (vd) “order sb to do sth”; epakasnu (vd) “teach sth to sb”; erek “sing (of a bird) about sth”; eukoitak “speak to e. o. about sth”; eukoramkor “discuss sth with e.o.”

esanniyo “consider sth”; ramu “think of sth”

inu “hear”

siekimatekka “threaten sb”

18 items

14 items

ekimatek “be frightened by sth”; eraman “know”; nu “hear sth”; oyra “forget sth”; tere “wait for sth/sb” 14 items

Table 1. Sets of DS/SS-taking verbs and IS-taking verbs in Ainu (Southern Hokkaido)

28

ANNA BUGAEVA

Abbreviations 1 / 2 / 3 – 1 / 2 / 3 person, Ø – zero-marked 3rd person, = – inflectional boundary, - – morphological boundary, [ ] – embedded clauses, my comments, A – transitive subject, A.B. – Anna Bugaeva, APPL – applicative, CAUS – causative, CH – Central Hokkaido dialects, COL – colloquial, COMP – complementizer, COP – copula, DS – direct speech, e.o. – each other, EMP – emphatic, FIN – final particle, IND – indefinite, INFR.EV – inferential evidential, IS – indirect speech, KY – kamuy yukar “songs of gods”, LOG = logophoric, NR – nominalizer, O – object, PERF – perfect, PL – plural, POSS – possessive, PR – pronoun, QUOT – quotation marker, RC – reporting clause, REP.EV – reportive evidential, RM – reporting marker, RV – reporting verb, S – intransitive subject, sb – somebody, SG – singular, SH – Southern Hokkaido dialects, SS – semidirect speech, sth – something, TOP – topic, U – uwepeker “prosaic folktales”, vd – verb ditransitive, vi – verb intransitive, VIS.EV – visual evidential, vt – verb transitive. REFERENCES A. Sources B

S

SK T1

T2 T3 T4

Bugaeva, Anna. 2004. Grammar and Folklore Texts of the Chitose Dialect of Ainu (Idiolect of Ito Oda). (ELPR Publication Series A-045). Suita: Osaka Gakuin University. Satoo, Tomomi, ed. 2002. Ainugo Shohoogen Choosa Hookoku 1 [A Fieldwork Report on Ainu Dialects 1]. (ELPR Publication Series A2-014). Suita: Osaka Gakuin University. Sunazawa, Kazuo. 1983. Watashi no Ichidai no Omoide. Ku Sukup Opuspe [My Life Story]. Sapporo: Miyama Shoboo. Tamura, Suzuko. 1972. “Ainugo Saru Hoogen no Ninshoo no Syurui [Types of Grammatical Person in the Saru Dialect of Ainu]”. Gengo Kenkyuu 61.17-39. (Reprint: Ainugo-koo, Bunpoo I, Vol 4, 370-392. Tokyo: Yumani Shoboo, 2001.) Tamura, Suzuko. 1985. Ainugo Onseishiryoo 2 [Ainu Audio Materials 2]. Tokyo: Waseda Daigaku Gogaku Kyooiku Kenkyuujo. Tamura, Suzuko. 1986. Ainugo Onseishiryoo 3 [Ainu Audio Materials 3]. Tokyo: Waseda Daigaku Gogaku Kyooiku Kenkyuujo. Tamura, Suzuko. 1988. “Ainugo [The Ainu Language]”. Gengogaku daijiten, ed. by Takashi Kamei, Rokuuro Koono & Eiichi Chino, 6-94. Tokyo: Sanseidoo. (English version: The Ainu Language. (ICHEL Linguistic Studies v. 2). Tokyo: Sanseidoo, 2000.)

SPEECH REPORT CONSTRUCTIONS IN AINU

29

B. Secondary literature Aikhenvald, Alexandra. 2005. Direct and Indirect Speech in Typological Perspective. Position Paper for RCLT Local Workshop (manuscript). Aikhenvald, Alexandra (forthcoming) “Semi-direct Speech: Manambu and Beyond”. Language Sciences 30. Chiri, Mashiro. 1942. “Ainu Gohoo Kenkyuu – Karafuto Hoogen o Chuushin to Shite [Studies in Ainu Grammar – with an emphasis on the Sakhalin dialect]”. Chiri, Mashiho Chosakushuu, vol. 3, 455-586. Tokyo: Heibonsha. Güldemann, Tom & Manfred von Roncador, eds. 2002. Reported Discourse: a Meeting Ground for Different Linguistic Domains. (TSL 52). Amsterdam & Philadelphia: John Benjamins. Güldemann, Tom. (forthcoming). “Quotative Indexes in African Languages: a Synchronic and Diachronic Survey”. Empirical Approaches to Language Typology 34. Berlin & New York: Mouton de Gruyter. Hagège, Claude. 1974. “Les Pronoms Logophoriques”. Bulletin de la Société de Linguistique de Paris 69.287-310. Haiman, John. 1989. “Alienation in Grammar”. Studies in Language 13:1.129170. Hattori, Shiro, ed. 1964. An Ainu Dialect Dictionary. Tokyo: Iwanami Syoten. Jakobson, Roman. 1990. “Shifters, categories, and the Russian verb”. Selected Writings: Word and language, 130-153. The Hague; Paris: Mouton. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University Press.

THE SYNTAX AND PRAGMATICS OF ADVERBIAL CLAUSES IN EASTERN KHANTY ANDREI FILTCHENKO Rice University & Tomsk State Pedagogical University 1.

Introduction This discussion of complex clauses in Eastern Khanty stems from work on a reference grammar of the endangered and under-described Eastern Khanty dialects. The present article reviews the mechanisms underlying the association between the morphosyntactic structural inventories used to code clause juncture relations vis-a-vis underlying conceptual situations (Cristofaro 2003). The complex clause in Khanty has been the object of studies varying in depth and theoretical framework. Early studies include Steinitz (1980), Zhivotikov (1942), Tereshkin (1961), and Gulya (1965). More recent reviews of clause embedding have been published by Honti (1984), Csepregi (1998), and Nikolaeva (1999). The analysis of polypredicative constructions has been undertaken by Koshkareva (1991), Kovgan (1991) and Skribnik (1991). Most of these studies focused primarily on the better-described and less endangered Western Khanty dialects and set mainly in a formal-structural perspective. The focus here is on the Eastern Khanty adverbial relations coded by the syntactic units of clause complexity, leaving aside other productive adverbial expressions such as lexical adverbs, case-inflected nominals, and postpositional phrases. The structural, semantic and pragmatic features of the constructions coding the conceptually dependent adverbial events are compared to the system’s canonical independent declarative clause as it occurs in isolation. 2.

The simple Eastern Khanty declarative clause The typical Eastern Khanty declarative active-direct clause is generally verb-final with grammatical functions differentiated by a combination of such features as word order, case marking, and verbal coreferential inflection; cf. (1) vs. (2). Agreement is obligatory between the grammatical relation of subject/ agent (S/A) and an intransitive or transitive predicate V. Transitive verbs may also ‘agree’ with the grammatical relation of object (2b,c), expressing pragmatic identifiability, accessibility and a high degree of activation of the referent in the interlocutors’ discourse universe; the latter feature has generally

32

ANDREI FILTCHENKO

been referred to as definiteness (cf. Tereskin 1961 and Gulya 1966). (1)

Agent ma_ tem puƒol-na jo_N al 1SG this village-LOC ten year “I lived in this village for ten years”.

Predicate w´l-s-´m live-PAST2-1SG

Agent Target Predicate (2a) ma_ sart w´l-s-´m. 1sg pike kill-PAST2-1SG “I caught a pike (fish)”. (2b) ´ll´ sart ma_n-n´ big pike 1SG-LOC “I got the big pike ready”.

lo_ƒo_li-s-im cut-PAST2-SG/1SG

(2c) terka_-s-im iw´s-n´ fry-PAST2-SG/1SG stick-INSTR/COM “I fried (it) on sticks”. Verb agreement is key for the elision of clause constituents, with the S/A frequently coded overtly by nothing more than the predicate agreement suffix (2b, c). In the Eastern Khanty narratives, a new referent in discourse is normally introduced by a full NP or a free pronoun in the clause-initial S/A relation and by S/A-V agreement. Thereafter, being identifiable as topical, it is coded by elision and S/A-V agreement. We can provisionally reaffirm a universal correlation (Lambrecht 1994) of the pragmatic status of referents to their formal complexity. In other words, the continuum between a zero and maximal morphological explicitness is counterproportionate to the continuum between topicality on one hand and unidentifiability and inactiveness on the other. This relation appears in Table 1: NP (+agreement) -> (-)

morphological coding pronoun (+agreement) -> pragmatic salience

zero argument (+agreement) (+)

Table 1. Formal complexity and the pragmatic salience of referents

3.

Conceptual subordination It has being noted increasingly that traditional definitions of subordination based solely on morphosyntactic criteria are not universally valid in light of cross-linguistic variation in formal structural means (Comrie 1981, Van Valin & La Polla 1997, Givon 2001, Cristofaro 2003). Rather, a functional definition

ADVERBIAL CLAUSES IN EASTERN KHANTY

33

of subordination must be adopted, implying a universal way of construal of a relation between events. This definition takes as its basis the notions of profiling vs. base, or foregrounding vs. backgrounding (Langacker 1991). An asymmetry in the construal of states-of-affairs’ (SOA) is anticipated, where the dependent SOA lacks autonomous profile, construed in the perspective of the matrix. This study reviews semantic features of the relations between SOA, and the degree of semantic integration and predetermination of information in the dependent SOA (Cristofaro 2003). The semantic integration of SOA iconically reflects the morphosyntactic integration between clauses; i.e., the less independent two conceptualized SOA, the less independent the expressions coding them (Haiman 1985, Newmeyer 1992). Concerning the relation between the linked clauses, the hypothesis is verified that while the foreground information represents the essence of the text and is temporally ordered, the background represents supportive information and is typically out of timeline (Hopper & Thompson 1980, Reinhart 1984, Tomlin 1985). Within this framework, the distinction between prototypical subordination and coordination displays a basic information structure that associates conceptual subordination with reduced pragmatic assertion, while coordination is associated with increased assertion (Lambrecht 1994). 4.

Adverbial clauses Under the functional definition, typical adverbial relation implies that one of the linked SOAs corresponds to the circumstances under which another SOA takes place (Koptjevskaja-Tamm 1993, Haspelmath 1995). In the available Eastern Khanty narrative data, 13% of the adverbial meanings are coded by perfective participial clauses, 5% by imperfective participial clauses, 5% by converbial clauses, 5% by infinitival clauses, and 3% by finite clauses. Example (3) illustrates the expression of a purpose relation, which codes the envisioned result of an independent event. (3) Iwa_n juƒ lalkˆt´l-w´l, to_ ƒ t -a_, Ivan wood chop-3SG fire-ILLAT “Ivan is chopping wood for the fire”.

to_ ƒ o_ t fire

wer -ta_ make-INF

In (3) the first adverbial is an NP inflected for the illative case, the second is an adverbial dependent SOA with the infinitive predicate1 and elided coreferential S argument (Sd=S). Example (4) illustrates an infinitive-supine dependent SOA and elided coreferential S (Sd=S): 1

It should be noted that the Khanty infinite is etymologically connected with the lative suffix, as is true for many other Finno-Ugric languages.

34

ANDREI FILTCHENKO

(4) ra_ƒ ruƒu-l-´m, n'an' flour mix-PRES-1sg bread “I mix flour to make bread”.

wer -tati do-INF/SUP

Another frequent way of coding purpose relation is by the participial clause with the imperfective predicate inflected for possession and case and elided coreferential Sd (Sd=S): (5) wa_sk-a_ qatant´-l-´m, jo_ ƒ o_ -t -a_ m -a_ duck-ILLAT sneak-PRES-1SG shoot-IMPP-1SG-ILLAT “I am sneaking up on a duck to shoot it”. The S/A of the dependent participial clause may be referential with the matrix O, as in (6), where the imperfective predicate is inflected for possession and case and the elided Sd≠S, Sd=O: (6a) n'an' pan-l-ˆm ´ll´wtaki olˆntaƒ-n´ toƒor-l-´m, bread put-PRES-1SG top canvas-LOC close-PRES-1SG na_ m -´ki j´n -t -a_ l -a_ soft-PRED become-IMPP-3SG-ILLAT “I put the bread into the sack, so that (its crust) got softer”. Example (6b) shows the finite coding of this purposive SOA, with elided Sd=O: (6b) n'an' pan-l-ˆm ´ll´wtaki olˆntaƒ-n´ toƒor-l-´m, bread put-PRES-1SG top canvas-LOC close-PRES-1SG na_ m -´ki j´s soft-PRED become.PAST2.3SG “I put the bread into the sack, so (its crust) got softer”. Some purpose SOA clauses (cf. 7) prefer encoding by a finite predicate. Here the elided Sd=O: (7) n'an' jˆƒata-s-ˆm, qotS -aƒˆ ili ´nt´ qotS -aƒˆ . bread look-PAST2-1SG/SG burn-PAST.3SG or NEG burn-PAST.3SG “I checked the bread, (to see) if it had burned or not”. Scalar analysis can be posited, implying a gradation of structural and semantic features such as finiteness, participant coding, and a combination of verbal and nominal categories in the conceptually dependent SOA. Scalar reduction in verbal categorial distinctions and an increase in nominal is associated with rising conceptualization of the dependent SOA as an entity. This is particularly true of situations involving motion towards a landmark,

ADVERBIAL CLAUSES IN EASTERN KHANTY

35

which can be equated to the realization of the projected dependent SOA (Cristofaro 2003). Consistent with this is the identity of spatial and purposive directional markers, the illative case marker /-a/ (5, 6a), or spatial-purposive marker -ati (4). The scale of formal variation of the Khanty conceptually dependent purpose clauses is seen as shown in Table 2: (+) finite (independent)

(-) finite (dependent)

finite predicates with prototypical TAM distinctions with near full illocutionary force finite predicates with reduced AM distinctions and illocutionary force and S≠Sd non-finite predicates with no nominal features, S=Sd non-finite (IMPP) predicates with some aspect distinctions and some nominal features, S=Sd non-finite predicates with no TAM distinctions and numerous nominal features (S≠Sd)

Table 2. Formal variation in conceptually dependent purpose clauses

5.

Temporal relations Temporal relations imply occurrence of one SOA in reference to another (normally matrix) and may be classified based on the available Eastern Khanty data into the relation of temporal posteriority and temporal simultaneity. 5.1 Temporal posteriority The relation of temporal posteriority implies the dependent SOA occurring prior to the matrix SOA, and the matrix event time being a reference for the dependent event. In Eastern Khanty, this relation may be coded by converbial clauses (5.1.1), imperfective participial clauses (5.1.2), perfective participial clauses (5.1.3), and finite clauses (5.1.4). 5.1.1 Converb. Converbial clauses are a typical means of coding the dependent SOA. Eastern Khanty converbial clauses code the semantic relation of posteriority. (8) tSˆml -ali amˆs -min -n´ , ni m´na_-ƒ´n juƒa-t´ little-DIM sit-CONV-LOC woman go-PAST.3SG gather.wood-PAST.3SG “After sitting awhile, the woman went off to gather wood”. The dependent SOA is a converb predicate inflected for locative, and Sd=S. 5.1.2 Imperfective participle. Eastern Khanty imperfective participles with the marker /-t_/ attached to the base stem most frequently have the function of the participial predicate of the embedded dependent clause with a temporal relation

36

ANDREI FILTCHENKO

implying the dependent SOA occurring prior to the matrix SOA. In (9), the dependent SOA is the imperfective predicate inflected for possession and case, and Sd=S. (9) pˆr -pˆ t-t -al -n´ mo_ro_k-k´ j´ƒa_-ƒ´n, back-become-IMPP-3SG-LOC health-PRED become-PAST.3SG “Finally, the eagle got better”. (Kalinina 1970) 5.1.3 Perfective participle. The Eastern Khanty perfective participles with the marker /-_m/ most frequently have the function of a temporal adverbial for the finite matrix predicate and code the adverbial temporal relation of precedence (posteriority). In these cases perfective participial predicates are normally inflected for possession. (10) qunto qat to_ ƒ o_ -n´ i -m -´l , ra_tS uw´ƒt´-t´ … when house fire-LOC eat-PS.3SG old.man see-PAST-3SG “When the house burned down, the old man saw …” The dependent SOA is the perfective predicate inflected for possession, and Sd≠S. 5.1.4 Finite clauses. Some examples of Eastern Khanty complex clauses may be categorized as falling at the far periphery of the domain of subordination. Subordinate SOAs here are coded by the clauses with finite verbal predicates, and subordination is naturally less of a structural notion, but rather a cognitivesemantic one. These finite complex sentences represent strings of clauses, loosely linked, often without any overt syntactic indication of subordination, or overt clause-linkage markers, hardly demonstrating any grammar specific for subordination. However, a degree of semantic integration between these statesof-affairs is evident in recurrent elision of the core semantic roles, marked word order, and reduced illocutionary force of these linked finite clauses. These features align conceptually with at least some degree of subordination. (11a) waƒ money

tuntil -l -i , pay-PRES-PS.3SG

ma_n-n´ 1SG-LOC

t´ƒr´ƒ´t´-l-i, throw-PRES-PS.3SG

(11b) tuƒta ja-w´-w´l-t fire eat-CAUS-PRES-3PL “When (as soon as) I’m paid and gone, they’ll burn everything”. In (11), while the Target roles are coreferential, the Agent roles are evidently non-coreferential. The Present-Future form of the finite passive predicates in the clauses casts these propositions within the irrealis scope of the hypoth-

ADVERBIAL CLAUSES IN EASTERN KHANTY

37

esized future. The SOA of the passive clause (11a) is conceptually dependent, while that of the next finite active clause (11b) is the independent matrix SOA. It could be argued that the temporal relation in (11) is merely inferred from the order and context, that is, from the paratactic character of the complex clause. Alternatively, a degree of conceptual subordination can be posited for the preceding (11a) setting temporal specification for the matrix Action clause (11b). The passive clauses in their parenthetic character may be viewed as functionally similar to temporal adverbials. Like nonfinite constructions, Eastern Khanty finite clauses with a temporal relation are typically preposed to the matrix clause. They they do not code the relation of temporal anteriority; rather, they imply that the dependent SOA occurred prior to the matrix SOA. 5.2 Temporal simultaneity As regards to ordering relative to the event time of the matrix SOA, adverbial clauses may code the relation of temporal simultaneity (overlap), implying the dependent SOA occurred in the time frame of the matrix SOA. 5.2.1 Converb. Apart from coding the dependent SOA in the semantic relation of posteriority, Eastern Khanty converbial clauses may code very proximal posteriority bordering temporal overlap. (12) loqapa -ja am´s -m´n jis -m´n illa_n´ woqˆ ju-w´l garbage-ILLAT sit-CONV cry-CONV once fox come-PRES.3SG “While he was sitting by the trash pile weeping, there passed a fox”. The dependent SOA is the converb predicate, and S≠Sd. 5.2.2 Imperfective participle. Eastern Khanty imperfective participles may function as participial predicates in an embedded dependent clause with a temporal relation implying that the dependent SOA occurred in the time frame of the matrix SOA. (13) qatn -t´ pit -t´ , pa_ni puƒol-pa_ sick-IMPP become-IMPP and village-LAT a_rki pers´ƒ je-s-i many strange become-PAST2-PS.3SG “I’m getting sick, and there are more and more strangers in the village”. The dependent SOA is the imperfective predicate and Sd S. 5.2.3 Perfective participle. Eastern Khanty perfective participial constructions are also used, though much less frequently, to code the semantic relation of

38

ANDREI FILTCHENKO

temporal overlap of the dependent and the matrix SOAs, when functioning as the temporal adverbial for the finite matrix predicate. (14a) ´Nk -a_ l suj -a_ mother-3SG voice-ILLAT

m´n -m -a_ l -n´ , go-PP-3SG-LOC

(14b) uw´ƒt-´t´ see-PAST.3SG (14c) ´Nk -a_ l j´Nq -ja_ l qan´N -n´ jal -w´l mother-3SG water-PROLAT edge-LOC stand-PRES.3SG “Going to his mother’s voice, he saw her standing by the water’s edge”. Each of the three portions in (14) contains a different relation. In (14a) we have a dependent SOA with PP predicate inflected for possession and case, and Sd=S. In (14b) we have a matrix clause. And in (14c) we have a dependent perception SOA with finite predicate Sd S. Scalar reduction in verbal categories and the increase of nominal categorial distinctions is associated with an increase in assertiveness and a decrease in illocutionary force of dependent clauses. This is manifested in the ability to conceptualize the dependent state-of-affairs as a noun-like entity rather than a process. In the dependent predicates with reduced verbal categorial distinctions, the use of spatial case markers (locative or prolative in (5, 14)) to locate the dependent state-of-affairs with regard to the time expressed in the matrix clause is evidence of the gradual semantic integration of the dependent state-of-affairs into the matrix. The use of nominal possessive inflections to code the agent referent in the S/A relation of the dependent stateof-affairs, also contributes to general reduction of finiteness. (+) finite (independent)

(-) finite (dependent)

full finite predicates with prototypical TAM distinctions with near full illocutionary force S Sd, S=Sd non-finite predicates with no TAM distinctions and no nominal features S=Sd non-finite predicates with no TAM distinctions and some nominal features S=Sd, S Sd

Table 3. Formal variation in predicates of dependent temporal adverbials

6.

Reality condition relations Reality condition relations encode the speaker’s commitment to the truth value of the utterance (Bybee 1985: 169) as expressed between the two SOAs, where the realization of the dependent SOA represents a condition for the realization of the matrix.

ADVERBIAL CLAUSES IN EASTERN KHANTY

39

The most frequent type of conditionals in Eastern Khanty contain the affix /-N-/ in the position normally occupied by the tense marker in finite predicates of the adverbial subordinate clause. Parallel to the established typological tendency (Givon 2001: 300), in Eastern Khanty the modality of the proposition does not affect the propositional frame of the clause (the pragmatic functions, semantic roles and grammatical relations of the referents, predication types and verbs’ transitivity), and does not appear to impose lexical restrictions. Conditional clauses show regular expression of transitivity, voice, aspect, negation, S/A-V agreement (but not O-V), and pragmatic-semanticgrammatical mappings. (15) ´nt´

m´n-N-a_n wo_ƒ-na_ tu-l-uj-´n go-COND-2SG force-COM take-PRES-PS-2SG “If you don’t go, you’ll be taken by force”. (Gulya 1966: 121) NEG

A dependent conditional if-clause has the irrealis negative modal scope; the agreement between the S/A and the predicate in both if- and then-clause is only the S/A-V. The passive matrix is in PRES; (Sd=S); but varying semantic roles: (Agentd=S) (Agent=non S) In communicative pragmatic terms, Eastern Khanty conditionals cast the proposition as the weakest assertion of possible or desired truth; or as a false assertion contradicting explicit or assumed belief, as these conditional propositions express temporal relation of either simultaneity or precedence of the speech-time to the event-time (15). The formal conditional affix /-N-/ occurs on the verbal predicate of the if-adverbial clause, which is consistent with the widely attested typological tendency (Bybee 1985). Another Eastern Khanty conditional is marked by the free morpheme to_N, which follows the predicate with overall perfectivity sense contributing to the sense of negative assertion, explicitly contradicting the fact. These mood constructions also have a certain deontic sense, expressing a desire, or expectation. (16) ma_ 1SG to_ N

nipik taja-m-am-n´, book have-PP-1SG-LOC no_N-a_ m´-ƒa_s-i COND 2SG-ILLAT give-PAST1-PS.3SG “If I had a book, I would have given it to you”. (Gulya 1966: 122)

Dependent conditional if-clause has the perfective participial adverbial clause and the Sd coreferential with S representing the subsequent counterfactual finite conditional that codes an event that never actually occurred. Thus the matrix has an implication “but X did not”.

40

ANDREI FILTCHENKO

(17) qunt´-n´ to_ N , ma_ m´n-ƒa_l-´m, when-LOC COND 1SG go-PAST3-1SG tut-n´ jo_ƒ-´n ma_-nt ka_s-ka_l that-LOC 3SG-LOC 1SG-ACC find-PAST3.3SG “If I had gone, he would have found me”. (Gulya 1966: 122) The conditional if-clause is coded by the past finite predicate with the overt if/when conjunction in locative case and followed by to_N, rendering the clause conceptually dependent, reduced in illocutionary force, with the implied matrix “but X did not”. The subsequent “ergative” finite then-clause has full illocutionary force and Sd S. The implicit negative matrix is more explicit in (18), where the counterfact reason is stated in the subsequent but-clause: (18) ma_ n -n´ tu -ƒas -ˆ to_ N 2, 1SG-LOC bring-PAST1-PS.3SG COND jo_ƒ-´n ma_n-a_ ´nt´ m´-ƒa_s-i 3SG-LOC 1SG-ILLAT NEG give-PAST1-PS.3SG “I should’ve brought it, but he didn’t give it to me”. (Gulya 1966: 122) The dependent clause expresses a counterfactual assertion, contains a finite passive predicate followed by to_N, and has reduced illocutionary force. It is followed by the reason but-clause with Sd S, implying the counter-fact that “X did not”. Most conditional meanings are coded by finite conceptually dependent predicates, except for rare PP conditionals (16). Imperfective N-conditionals require the same subject in both clauses, whereas the counterfactual perfective to_N-conditionals are more found in sentences with different subjects in each clause (70%). The scale of structural features such as finiteness and participant coding in the conceptually dependent conditional clause has a marked cline towards finiteness. Reduction in verbal categorial distinctions is manifest in the form of absent tense distinctions in irrealis imperfective N-conditionals and in reduced marking of the pragmatic status of the O-argument via conjugation. Both types of conditionals lack in illocutionary force without at least some contracted form of the matrix. Nominal categories are not expressed, correlating to weaker ability to conceptualize the dependent SOA as entities:

The conditional marker to_N could be related to the attributive nominal to_N “straight, correct, true, honest” and its adverbial derivation to_Nq´ “straightly, truly, honestly”. There may likewise be an etymological connection with the bound imperfective conditional marker /N-/ and the non-bound perfective/completed conditional marker to_N. 2

ADVERBIAL CLAUSES IN EASTERN KHANTY

(+) finite (independent)

(-) finite (dependent)

41

finite predicates with full TAM distinctions and near full illocutionary force (S Sd) finite predicates with full TAM distinctions and reduced illocutionary force (S=Sd) finite predicates with reduced aspect & tense distinctions and lower illocutionary force (S=Sd) non-finite predicates with reduced TAM distinctions and some nominal features (S=Sd)

Table 4. Formal variation in predicates of dependent condition adverbials

7.

Reason relations Reason relations involve situations where the realization of the dependent SOA is the reason behind the occurrence of the event expressed in the matrix SOA. The Eastern Khanty conceptually dependent purposive SOA is encoded by a finite clause and is non-factual. (19) ma_ qolˆsla-l-ˆm, patˆ -l -´m 1SG cough-PRES-1SG get.cold-PRES-1SG “I cough (because) I caught a cold”. The dependent finite clause has the present predicate and (S=Sd). (20) Iwa_n naƒˆnt´-w´l rut' saƒˆ, qul ˆl ru_ ƒ -w´lt put -oƒ Ivan swear-PRES.3SG Russian way fish down jump-PRES.3PL pot-PROLAT “Ivan swears in Russian (because) the fish is jumping out of the pot”. Note that the dependent finite clause contains a present predicate form and (S Sd). Differentiating types of subordination among reason constructions is problematic. The reason for that is due to the preferred coding of these conceptually dependent reason SOAs by finite clauses, albeit with reduced illocutionary force. The scale has a greater cline towards finiteness. Higher clausal independence and verbal categorial distinctions are associated with retained independent profile, assertiveness, high illocutionary force, and conceptualization of reason SOAs as processes rather than entities, as shown in Table 5: (+) finite (independent)

finite clauses with prototypical TAM distinctions with full illocutionary force (S Sd)

(-) finite (dependent)

finite predicates with prototypical TAM distinctions and near full illocutionary force (S=Sd)

Table 5. Formal variation in dependent reason adverbial clauses

42 8.

ANDREI FILTCHENKO

Conclusions The deviation from the independent simple clause is measured with regard to two main parameters: the verb form, and the participant coding: i.e. the combination in the dependent predicate of prototypical verbal categorial distinctions typical for independent predicates. Since time reference, aspect and mode typically code the occurrence of a SOA through time, i.e., properties of the process, these distinctions often lack in the more entity-like dependent SOAs (Cristofaro 2003: 177). In Eastern Khanty, temporal and purposive adverbial relations are more construable as entities and thus have less regular TAM distinctions. On the other hand, relations such as reason and reality condition remain construable as processes and retain comparably more prototypical verbal distinctive features. Instead of a discrete opposition between coordination and subordination clause-linkage strategies, there appears to be a continuum of combinable features, whose presence or absence and combination locates each clause in relation to either of the prototypes (Haiman & Thompson 1984, Lehmann 1988, Matthiessen & Thompson 1988, Van Valin & La Polla 1997, Givon 2001). These features manifest the degree of lexical-grammatical elaboration in a clause, resulting in a continuum of downgrading the conceptually subordinate clause from the prototypical finite matrix-like clause to the prototypical embedded subordinate one with low degree of predicate grammaticalization. The connection between morphosyntactic and semantic features of adverbial relations is functionally motivated by cross-linguistic principles including economy and iconicity (Givon 2001, Haiman 1985). Eastern Khanty actively applies the economy principle via co- and switch-reference devices, i.e., via elision and use of possessive markers on the participial dependent predicates in purpose and temporal adverbials, when only the information minimally required for adequate processing of the utterance is coded explicitly. Once the TAM values, participant information, etc. are determined in the matrix SOA, they are not re-specified in the conceptually dependent SOA. Various elisions are more frequent in purpose relation due to their inherent typical discourse motivated tendency towards co-reference. Where sharing of participants with the matrix SOA is less typical, elision is less frequent, explicit expression of participants of the conceptually dependent SOA occurs for signaling a change, or to prevent ambiguity (cf. Table 6). A key principle at play in the reduced explicitness of nominal arguments and of verbal morphosyntax in the dependent clauses is the iconicity of independence (Givon 2001, Cristofaro 2003), manifested in Eastern Khanty in the syntactic integration of the linked SOAs, where much, or all of the relevant features are expressed in the matrix.

ADVERBIAL CLAUSES IN EASTERN KHANTY

(+) independence low semantic integration

43

Finite DS reason, finite DS purpose, finite DS conditional, finite DS temporal [S Sd] Finite SS reason, finite SS conditional, finite SS temporal [S=Sd]

high semantic integration (-) independence

Non-finite temporal and DS purpose [S Sd] Non-finite SS temporal, SS purpose, SS conditional [S=Sd]

Table 6. Formal variation in conceptually dependent adverbial clauses

Presence of nominal features in non-finite temporal or purpose adverbial clauses, are manifestations of these predicates’ usage in prototypical noun discourse-pragmatic function of, but not verbal processes (Hopper & Thompson 1985, Givon 2001, Croft 2001, Cristofaro 2003). In CG terms, the change in profile of the dependent predicate from sequential scanning to summary scanning allows for the presence of nominal formal properties (case, possession) on the dependent predicate and lack of verbal distinctions (TAM) (Cristofaro 2003). In Eastern Khanty, over 70% of the participial adverbial dependent predicates appear inflected for nominal categories: possessive expression of the person/number of the S/A argument and case. Consequently, at least in some instances (5%), prototypical verbal properties such as aspectual (including Aktionsart) distinctions co-occur with the prototypical nominal properties (possession, case). verbal categories (TAM)

nominal categories (case, poss)

Finite reason adverbials – [+TAM] Finite purpose & temporal adverbials – [+TA] Finite to_N-conditional adverbials – [+TA] Finite N-conditional adverbials – [+A] Non-finite purpose and temporal Non-finite temporal – [+poss] Non-finite temporal – [+A], [+poss], [+case] Non-finite conditional, purpose, temporal – [+poss], [+case]

Table 7. Categorial variation in predicates of dependent adverbial clauses

Dominance of the matrix SOA’s profile over that of the dependent one correlates to the dependent SOA’s interconnectedness with the matrix to the extent that it cannot be conceptualized as completely distinct, i.e. there is a certain pragmatic sentential deficiency of the dependent clauses (Cristofaro 2003: 264). Eastern Khanty conceptually dependent clauses are characterized by decreased autonomous profile, which is associated with pragmatic presupposition (Langacker 1991) correlating to low assertiveness in the utterance and lower illocutionary force.

44

ANDREI FILTCHENKO

(+) full /near full illocutionary force

Finite reason adverbials Finite purpose & temporal adverbials Finite conditional adverbials Finite purpose adverbials Non-finite conditional, purpose and temporal adverbials

(-) low illocutionary force Table 8. Variation in the pragmatics of dependent adverbial clauses

A multi-parametrical scale of Eastern Khanty adverbial clauses thus illustrates the degrees of delexicalization (diversion from the lexical prototype), decategorization (switch in typical categorial distinctions) of the conceptually dependent verbal clauses, reflecting the functional-cognitive variation in the degree of semantic/conceptual independence of the adverbial SOAs: (+) independent linked clauses, coordinatelike interclausal relation

(-) dependent linked clauses, subordinatelike interclausal relation

full finite predicates with prototypical TAM distinctions & full/near full illocutionary force finite predicates with reduced TAM distinctions and reduced illocutionary force non-finite predicates with some TAM distinctions, no nominal features, low illocutionary force non-finite predicates with no TAM distinctions and no nominal features non-finite predicates with no TAM distinctions and some nominal features non-finite predicates with no TAM distinctions and multiple nominal features finite purpose adverbials non-finite conditional, purpose and temporal adverbials

Table 9. Dependence continuum of predicates of adverbial clauses

Abbreviations – agent, ACC – accusative, CAUS – causative, COM – comitative, CONV – con-verb, COND – conditional, ILLAT – illative, IMPP – imperfective participle, INF – infinitive, INSTR – instrumental, LAT – lative, LOC – locative, NEG – negative, O – object, PL – plural, PRED – predicate, PRES – present tense, PROLAT – pro-lative, PS – perfective participle, S – subject, SOA – state of affairs, SG – singular, V – verb. A

REFERENCES Cheremisina M. I. & N. I. Koshkareva. 1991. Slozhnoe i oslozhnennoe predlozhenie v khantyjskom jazyke [Complex and extended sentences in Khanty]. Novosibirsk: Izd. “Nauka”. Comrie, Bernard. 1981. Language Universals and Linguistic Typology.

ADVERBIAL CLAUSES IN EASTERN KHANTY

45

Chicago: University of Chicago Press. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Croft, William. 2001. Radical Construction Grammar. Cambridge: Cambridge University Press. Givon, Talmy. 2001. Syntax. Amsterdam & Philadelphia: John Benjamins. Givon, Talmy. 1991. “The Evolution of Dependent Clause Morphosyntax in Biblical Hebrew”. Approaches to Grammaticalization ed. by Elizabeth Closs Traugott & Bernd Heine. Amsterdam & Philadelphia: John Benjamins. Gulya, Janos. 1966. Eastern Ostyak Chrestomathy. (Indiana University Publications. Uralic and Altaic Series, Vol. 51). The Hague, Netherlands: Mouton de Gruyter. Gulya, Janos. 1970. “Aktiv, Ergative und Passiv im Vakh-Ostjakischen”. Simposium über Syntax der uralischen Sprachen ed. by Wolfgang Schlachter, 81-90. Göttingen: Vandenhoeck & Ruprecht. Haiman, John. 1985. Natural Syntax. Cambridge: Cambridge University Press. Haspelmath, Martin. 1995. “The Converb as a Cross-Linguistically Valid Category”. Converbs in Cross-Linguistic Perspective ed. by Martin Haspelmath & Ekkehard König, 1-56. Berlin & New York: Mouton de Gruyter. Honti, László. 1984. Chrestomathia Ostiacica. Budapest: Tankönyvkiadó. Hopper, Paul & Sandra A. Thompson. 1980. “Transitivity in Grammar and Discourse”. Language 56.251-299. Koptjevskaja-Tamm, Maria. 1993. “Finiteness”. Encyclopedia of Language and Linguistics ed. by Ronald Asher & Seumas Simpson, 1245-1248. Oxford: Pergamon Press; Aberdeen: Aberdeen University Press. Kovgan, E. 1991. Opredelitel’nye konstruktsii v zapadnykh dialektakh khantyjskogo jazyka. [Attributive clauses in Western Khanty dialects.] Unpublished Candidate Degree dissertation. Novosibirsk. Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University Press. Langacker, Robert. 1991. Foundations of Cognitive Grammar. Stanford: Stanford University Press. Lehmann, Christian. 1988. “Towards a Typology of Clause Linkage”. Clause Combining in Grammar and Discourse ed. by John Haiman & Sandra A. Thompson, 193-224. Amsterdam & Philadelphia: John Benjamins. Li, Charles N., Thompson, Sandra A. 1976. “Subject and Topic: A new typology of languages”. Subject and Topic ed. by Charles Li, 457-490. New York, San Francisco & London: Academic Press. Matthiessen, Christian & Sandra A. Thompson. 1988. “The Structure of Discourse and Subordination”. Clause Combining in Grammar and Discourse ed. by John Haiman & Sandra A. Thompson, 275-327. Amsterdam & Philadelphia: John Benjamins.

46

ANDREI FILTCHENKO

Nikolaeva, Irina. 1999. Ostyak. Munich: Lincom Europa. Newmeyer, Frederick. 1992. “Iconicity and Generative Grammar”. Language 68.756-96. Reinhart, Tanya. 1984. “Principles of Gestalt Perception in the Temporal Organization of Narrative Texts”. Linguistics 22.779-809. Skribnik E. & E. Kovgan. 1991. “Sistema prichastnykh opredelitel’nykh konstruktsij v obsko-ugorskikh jazykakh [The system of attributive participial constructions in Ob-Ugric languages]”. Jazyki narodov Sibiri: grammaticheskie issledovanija [Languages of Siberia: grammatical studies] ed. by M. I. Cheremisina, 84-108. Novosibirsk: Izd. “Nauka”. Tereshkin, N. 1961. Ocherki dialektov khantijskogo jazyka. Vakhovskij dialect [Overview of Khanty dialects: Vakh dialect]. Leningrad: Izd. “Nauka”. Thompson, Sandra A. 1987. “Subordination and Narrative Event Structure”. Coherence and Grounding in Discourse ed. by Robert Tomlin, 435-454. Amsterdam & Philadelphia: John Benjamins. Thompson, Sandra A. 1985. “Grammar and Written Discourse: Initial and final purpose clauses in English”. Quantified Studies in Discourse ed. by Talmy Givon. Text 5:1/2.435-454. Tomlin, Robert 1985. “Foreground-Background Information and the Syntax of Subordination”. Text 5:1/2.85-122. Van Valin, Robert & Randy Lapolla. 1997. Syntax: Structure, meaning and function. Cambridge: Cambridge University Press. Zhivotikov, P. K. 1942. Ocherk khantijskoj grammatiki [A sketch of Khanty grammar]. Khanti-Mansijsk: Izdanie Okruzhnoj komissii po razrabotke literaturnogo khantyjskogo jazyka [District Committee for Khanty Literacy Development].

NULL ARGUMENTS IN KUMYK ADVERBIAL CLAUSES 1 LINDA HUMNICK University of Minnesota 1.

Introduction As part of broader research on the syntax and pragmatics of referring expressions in Kumyk, a Turkic language spoken primarily in the Dagestan region of the Russian Federation, this study explores constraints on the use of null subjects in non-finite clauses within a text corpus. Kumyk allows null subjects in both finite and non-finite clauses. The null subjects of finite clauses are pronouns in the binding theory sense that they typically have an antecedent in the discourse but are not syntactically bound. In these cases, the reference of a null subject is partially restricted by the agreement in person and number marked in the verb form. For example, in (1), the form of the past tense suffix, -dï, indicates that the subject of the verb aldï “bought” is third person singular, restricting reference to entities with those properties. (1) Murat bazarga gete. Onda Ø alma Murat market.DAT go.3S.PR there apple “Murat went to the market. (He) bought apples there”.

aldï.2 buy.3S.PST

Beyond the restrictions imposed by subject agreement, the identification of the referent of a null pronoun is governed by pragmatic, rather than syntactic constraints. These pragmatic constraints allow the hearer to select the likely referent of the null form from among appropriate entities available from the discourse context.

1

Research for this study was partially supported by NSF grant # 0519890 to Jeanette Gundel. I also wish to acknowledge Jeanette Gundel, Paul Humnick, Agaragim Sultanmuradov, and Ruhaniyat Musaeva for their assistance in various aspects of this project. 2 The Cyrillic orthography is transliterated here using a modified IPA system. The grapheme “g” represents both the voiced velar stop and its allophone, the voiced uvular fricative, though this distinction is represented by two different graphemes in the Cyrillic orthography. Capital letters in suffix forms, such as -mE represent vowel harmony alternations represented in the orthography.

48

LINDA HUMNICK

In contrast to the subject agreement found in finite clauses, most non-finite clause types allow null subjects yet exhibit no morphological agreement in the verb form that might constrain the reference of the null argument. This characteristic is illustrated by the non-finite verb alïp “buy” in the second sentence of (2), as the suffix -ïp is unmarked for person or number. (2) Murat Murat

bazarga market.DAT

gete. go.3S.PR

Onda there

Ø

alma apple

alïp, buy.CONV

o bizge geldi. 3S 1PL.DAT come.3S.PST “Murat went to the market. Having bought apples there, he came to us”. The lack of subject agreement in non-finite clauses raises the question of whether the reference of their null subjects is constrained by any other syntactic factors or is constrained only by pragmatic factors, suggesting an inference system of reference tracking (Foley & Van Valin 1984). A survey of non-finite clauses in Kumyk indicates that reference of null arguments in some types of non-finite clauses is syntactically restricted to a particular argument of the matrix clause either by the category of the verbal suffix or by the way in which the non-finite clause adjoins to the matrix clause. This phenomenon is typical of certain types of complement clauses and adnominal subordination. In other types of non-finite clauses — primarily adverbial clauses — the reference of a null subject is not syntactically restricted by such means. In these cases, reference is assumed to be only pragmatically constrained. It is useful, then, to talk about how to capture the pragmatic constraints that assist the hearer in identifying the referent of the null form and/or govern the speaker’s choice between a reduced or null form and a more explicit form. This study looks at pragmatic criteria for null forms within a cognitively based model of referential choice primarily based on the Givenness Hierarchy model of Gundel, Hedberg & Zacharski (1993). The next section of this paper describes the distinction between types of non-finite clauses which exhibit syntactically restricted null arguments and types which have null arguments with syntactically unrestricted reference, setting the stage for a discussion of how the use of null arguments may be restricted by pragmatic considerations. Section 3 describes the theoretical model within which such pragmatic considerations are explored and the method of application to Kumyk data from a text corpus. Finally, Section 4 summarizes the results of applying the Givenness Hierarchy model to the data, and Section 5 discusses the implications of the study.

KUMYK ADVERBIAL CLAUSES

49

2.

Types of non-finite clauses exhibiting null subjects Non-finite clause types in Kumyk may be categorized by differences in the form of the non-finite suffix and differences in the way in which the non-finite clause is conjoined to the matrix clause — for example whether it is conjoined as a noun modifier or as an adverbial element. The purpose of this section is to categorize non-finite clauses in terms of a single parameter: the degree to which the referents of their null arguments are restricted or unrestricted. 2.1 Syntactically restricted null arguments The reference of null arguments in Kumyk non-finite clauses may be restricted syntactically in two different ways. The first way is restriction by means of a verb suffix form that signals control by the matrix subject or object. The second type of restricted reference is associated with clauses which function as modifiers of a matrix clause argument. In the latter case, the reference of the null argument of the non-finite clause can only be interpreted as co-referent with the matrix argument that the clause modifies. When a non-finite clause verb form ends with the infinitive suffix -mE or -mEgE, its null subject must be interpreted as coreferent with either the matrix subject or object, depending on the semantic type of the verb.3 In example 3, both non-finite clause null subjects are coreferent with the matrix subject qajnana “mother-in-law”. (The non-finite clause is enclosed in brackets, and the subjects of both clauses are in bold.) (3) Qajnanai [Øi adamlanï chaqïr-ma] mother.in.law man.3PL.ACC invite-INF [Øi bajram et-me] sjue. feast do-INF want.3S.PR “The mother-in-law wants to invite people and celebrate”. Example (4) illustrates the ungrammaticality of the use of the suffix -mE with the subject-control verb sjue “wants” in a case in which the overt non-finite clause subject gelin “bride” is not co-referent with the matrix subject qajnana “mother-in-law”. If the subject of the clausal complement of sjue is not coreferent with the subject of sjue, then the verb of the complement clause must end in a nominal suffix, such as -mak in (4).

3

The choice between subject control versus object control appears to be related to the semantic type of the verb rather than to formal differences in the form of the suffix. For example a verb such as sjue “wants” would always be a subject control verb, while a phrasal verb like ixtijar bere “gives the right to” is inherently an object control verb.

50

LINDA HUMNICK

(4) *Qajnana [gelin adamlanï bajramga chaqïr-ma] sjue. mother.in.law bride man.3PL.ACC feast.DAT invite-INF want.3S.PR “The mother-in-law wants the bride to invite people to the feast”. (5) Qajnana [gelin adamlanï bajramga chaqïr-mak-nï] sjue. mother.in.law bride man.3PL.ACC feast.DAT invite-VN-ACC want.3S.PR “The mother-in-law wants the bride to invite people to the feast”. The second category of non-finite clause in which the reference of null arguments is restricted consists of participial clauses with the suffixes -EgEn (present) or -gEn (past). These clauses can modify any argument of the matrix clause, and the null argument of the participial clause is always interpreted as coreferent with the argument it modifies. For example, in (6), the participial clause (in brackets) modifies qatïnï “his wife”; thus the null subject of the nonfinite verb etegen “making” is interpreted as having the same referent as qatïnï. (6) Arigi üjde [Øi ash etegen] qatïnïi further.MOD room.LOC food make.PR.PRT wife.3.POSS bir xarbuz alïp gele. one watermelon take.CONV come.3S.PR “His wife, who was making food in the back room, takes one watermelon and comes”. According to Foley & Van Valin’s (1984) typological study of complex clause structures, restricted-reference null arguments of the type found in (5) and (6) are commonly found cross-linguistically in clause combinations that are characterized by tight semantic linkage or that involve structural features such as embedding or subordination. They analyze such types of clause combinations as having a ‘core juncture,’ which means that at least one argument is “shared” — in other words, an argument of one clause is an empty category controlled by an argument of another clause. This type of clause combination contrasts with clause combinations in which the arguments of each clause are independent of the arguments of the other, a feature which is typical of clause linkage at the lower end of the interclausal semantic relations hierarchy (1984: 266, 188). 2.2 Unrestricted null arguments A second group of non-finite clause forms is characterized by both lack of coreferentiality restrictions and alternation between implicit and overt subjects. This group consists of two types: 1) clauses ending with the participial suffixes -EgEn or -gEn followed either by a locative suffix or an adverbial postposition and 2) clauses ending with the -Ip converb form. All of these types fall within

51

KUMYK ADVERBIAL CLAUSES

Haspelmath’s (1995) category of ‘adverbial clause’, i.e., a clause that serves as a modifier of a verb, clause, or sentence. Examples (7) and (8) exhibit adverbial clauses with overt subjects, showing the grammaticality of overt subjects that are coreferent or noncoreferent with the subject of the main clause. In (7) the subject of the adverbial clause is the overt form, Naima, which is coreferent with the null subject of the matrix clause. In (8) atasï, “her father”, which is the overt subject of the adverbial clause, is not coreferent with Naima, the overt subject of the matrix clause. (The non-finite clause is marked in brackets, and the subjects of the non-finite and matrix clauses are in bold.) (7) [Naimai getgende] Naima go.PST.PRT.LOC “After leaving, Naima cried”.

Øi

jiladï. cry.3S.PST

(8) [Atasïj getgende,] Naimai father.3S.POSS go.PST.PRT.LOC Naima “When her father left, Naima cried”.

jiladï. cry.3S.PST

When an adverbial clause and a matrix clause have coreferent subjects, an overt subject may occur in the adverbial clause, as seen previously in (8) or in the matrix clause, as in the corresponding example (9) below. It is also possible for coreferent subjects of both the adverbial clause and the matrix clause to be implicit, as in (10). (9)

[Øi

getgende] Naimai go.PST.PRT.LOC Naima “After leaving, Naima cried”.

(10) [Øi

getgende] Øi go.PST.PRT.LOC “After leaving, (he/she) cried”.

jiladï. cry.3S.PST jiladï. cry.3S.PST

When the subjects of both clauses are coreferential, there cannot be more than one overt subject, except in cases when one occurs as an emphatic reflexive4. In fact, in a construction in which an adverbial clause with an overt 3rd-person 4

While the general use of a reflexive pronoun with a coreferent overt subject appears to be infelicitous, cases that are clearly emphatic usages are acceptable. Example (i) occurs in the extended text corpus. Since the reflexive here is marked with a 1st person suffix, there is no question that it is coreferential with the matrix subject, men “I”. (i) Özjum de bilmej, men bir qïznï betine tiklenip self.RFLX.1SG EMPH not.knowing, I one girl’s face.at staring “Not even realizing it myself, I kept staring at one girl’s face”.

turganman. remained.

52

LINDA HUMNICK

subject is followed by a matrix 3rd person pronominal subject, as in (11), the pronoun is interpreted as not coreferent with the matrix subject. (11) [Naimai getgende,] o*i/j jiladï. Naima PST.PRT.LOC 3S cry.3S.PST “When Naima left, hej/shej cried”. While the constraint against multiple overt coreferent subjects in adverbial-matrix clause combinations shares some similarities with cases of null arguments of non-finite clauses that are controlled by the matrix subject, there are two significant differences. The first difference is that, as explained above, the overt subject may occur in the adverbial clause as well as the matrix clause. Secondly, unlike many languages, Kumyk allows null subjects of adverbial clauses that are not coreferent with the subject of the matrix clause, as shown in example (12). In the second sentence of this example, the referent of the null subject of the adverbial clause is Naima’s father, while Naima is the matrix subject. (12) Aksham Atasïi sumkanï hazir evening father.3S.POSS suitcase.ACC ready

boldu. become.3S.PST

[Erten Øi getgende] Naimaj jiladï. morning go.PST.PRT.LOC Naima cry.3S.PST “In the evening, her father got the suitcase ready. In the morning, when [he] left, Naima cried”. Given the above evidence against subject control in adverbial clauses, one may assume that the use of null arguments of adverbial clauses is subject to pragmatic constraints—as is the case for null arguments of matrix clauses. 3. Exploring the pragmatic parameters of null arguments 3.1 Theoretical model A number of theories have linked the use of various forms of referring expressions with pragmatic factors such as the referent’s degree of givenness in the discourse or the mental accessibility or cognitive status of the referent (Chafe 1976, Prince 1981, Givón 1983, Ariel 1988, 1990, Gundel et al. 1993, Grosz et al. 1983, 1995, Walker, Joshi & Prince 1998). In order to discuss the parameters constraining the use of null arguments, this study follows a cognitive-based approach to referential choice informed primarily by the Givenness Hierarchy model of Gundel et al. (1993). In this model, the form of referring expression that a speaker chooses from among multiple possible categories (e.g., null argument, pronoun, or full noun phrase) is assumed to signal the speaker’s assumptions about the cognitive status of the referent –

53

KUMYK ADVERBIAL CLAUSES

that is, its memory or attention status in the hearer’s mind (1993: 274-275). The cognitive statuses are assumed to be part of the conventional meaning of lexical items like pronouns and determiners. For example, a null argument signals that the referent is at the center of the hearer’s attention, while, in many languages, a demonstrative noun phrase signals that the referent of the expression is at least familiar. In the Givenness Hierarchy model, the set of cognitive statuses is implicationally related and hierarchically ordered from most restricted (in focus) to least restricted (type identifiable), as shown in Figure 1, along with the corresponding referring expressions in English (1993: 274-275). Status

In > Focus

Activated >

Familiar >

Uniquely > Identifiable

Referential >

Type > Identifiable

Form

it

this, that, this N

that N

the N

indefinite this N

aN

Figure 1: The Givenness Hierarchy

Correlating a type of referring expression with a cognitive status in the hierarchy means that its referents must always have at least that status—though a given form may be used to represent referents of a higher status. For example, in English, the demonstrative pronoun that may be used to represent a referent in focus, but the pronoun it may not be used to represent a referent that is at most activated. Cross-linguistically, null arguments correlate with referents of the highest cognitive status, those the speaker assumes to be in focus, or at the center of the hearer’s attention. It follows that the reference of a null argument is constrained to referents that are in focus in a given context. This constraint is pragmatic in the sense that the meaning of the referring expression correlates with the speaker’s assumptions about the hearer’s mental state rather than inherent properties of the referent. 3.2 Methodology In order to determine the cognitive status of the referent of a referring expression, the Givenness Hierarchy model uses a coding protocol that describes sufficient conditions for assigning a particular cognitive status (Gundel et al. 1993, Gundel 2004). In this study, the relevant distinction is between the cognitive status “in focus” and any status that is lower than in focus. According to Gundel et al., a referent is in focus if it is “at the center” of an individual’s attention (2003: 284). Specific criteria for coding this status are listed in Figure 2.

54

LINDA HUMNICK

A referent is IN FOCUS if it meets at least one of the following criteria: 1. It is mentioned in subject position in the immediately preceding sentence/clause. 2. It is mentioned earlier in the same sentence. 3. It is mentioned in syntactic focus position of the immediately preceding clause (postcopular position of a cleft or existential sentence in English). 4. It is a higher-level topic that is part of the interpretation of the preceding clause (whether it is overtly mentioned there or not). 5. It is mentioned in the two immediately preceding clauses. 6. It is the event denoted by the immediately preceding sentence. Figure 2: Coding protocol for in focus status (Gundel 2004)

While the criteria of the coding protocol have been tested crosslinguistically, the theoretical model does not exclude the possibility of certain language-specific adjustments to the criteria that are in character with their purpose. For example, in this study, I include non-subject arguments that are the experiencers of psychological verbs as belonging to the same category as grammatical subjects in criterion 1. This decision is based on evidence from Turkish in Turan (1998: 153) that the experiencer object, particularly in the case of a psychological verb, is of a similar level of salience as the grammatical subject.5 As an example of how the criteria of the coding protocol are applied in order to determine whether or not the referent of a referring expression is in focus, consider the text excerpt in (13), consisting of the first three sentences (13a–c) of a narrative text. In this illustration, brackets enclose the referring expressions selected for coding, while other mentions of the same referents (which are relevant to the application of the coding criteria) appear in bold type. (13a) Bir pajxammarnï bir qïzï bir esheki one prophet.GEN one girl.3.POSS one donkey.3.POSS bir mishiki bir de iti bolgan. one cat.3.POSS one and dog.3.POSS be.3S.PST “A prophet had one daughter, one donkey, one cat and one dog”.

5

An example of an experiencer object in Kumyk is illustrated in the expression pajxammarnï esine gele, which translates literally, ”it came to the prophet’s mind (that)…”

KUMYK ADVERBIAL CLAUSES

(13b) [Qïznï] erge girl.GEN husband.DAT vaqtisi gelip time.3.POSS come.CONV “When it comes time for comes to her”.

55

berme jarajgan give.INF possible.PST.PRT [bugar] gelechi gele. 3.1.DAT suitor come.3S.PR the daughter to be given in marriage, a suitor

(13c) [Qïznï girl.GEN [Ø]

atasï] da razi bolup, father.3.POSS and agreed be.CONV qïzïn berezhek bola. girl.3.POSS.ACC give.FUT be.3S.PR “The girl’s father agreeing, (he) will give the girl”.

The referent of the expression qïznï in (13b) is the daughter of the prophet. The same referent is mentioned in the previous sentence in (13a), but no criterion for in focus is satisfied. Note that, while qïznï is part of the complex subject of the previous sentence, it is not the unique grammatical subject and, therefore, does not satisfy criterion 1. The referent of the pronoun bugar in (13b) is also the daughter. In this case, the referent is evaluated as in focus by criterion 2, “mentioned earlier in the same sentence”. In (13c), qïznï atasï “girl’s father” refers to the prophet, who was mentioned in (13a) but not in (13b), thus the referent fails to satisfy a criterion for in focus. The prophet is also the referent for the second bracketed expression in (13c) which is a null argument. Since the prophet is mentioned earlier in the same sentence, this referent is in focus by criterion 2. The data for this study is part of ongoing research on the cognitive status of referring expressions in Kumyk discourse. The primary data consists of two texts (one oral, one written) in which I have coded the cognitive status for referents of all referring expressions, with the exclusion of null referents that are syntactically restricted (cf. §2.1) and referring expressions within quoted speech. Supporting data is gathered from an extended text corpus. In this study, null subjects of matrix clauses are assumed to be pragmatically constrained; thus, the coding of their referents serves as a control group with which to compare the coding of null subjects of adverbial clauses. There are 29 instances of this type in the primary corpus. The target group consists of 23 null subjects of adverbial clauses: 19 adverbial clause subjects which are coreferent with the subject of the matrix clause (henceforth, same subject or SS reference) and 4 null adverbial clause subjects which are not coreferent with the matrix subject (henceforth, different subject or DS reference). The distinction between SS and DS subjects is only made in the discussion of the fact that the use of DS null subjects in adverbial clauses is

56

LINDA HUMNICK

more restricted than that of SS reference. Finally, I compare adverbial clauses with null subjects to 26 adverbial clauses of the same type with overt subjects without distinguishing between SS and DS subjects. Since null subjects alternate with overt subjects in adverbial clauses, it is useful to consider whether or not the cognitive status of referents of overt subjects of adverbial clauses is restricted to the same degree as that of null subjects. In summary, within the Givenness Hierarchy model, the cognitive status of referents associated with a particular form of referring expression provides a constraint on the use of that form. The purpose of the study is to test the hypothesis that syntactically unrestricted null arguments in matrix and adverbial clauses in Kumyk are constrained to referents that are in focus, thus enabling hearers to identify the referents of null arguments in these contexts. A further goal of this study is to support the analysis that null subjects of SS adverbial clauses are more similar to matrix null subjects than to the syntactically controlled null arguments of other types of non-finite clauses. In cases of subject control, the null subject of the non-finite clause is always interpreted as coreferent with the matrix subject, and the use of the null form is obligatory — not licensed by the pragmatic status of the referent, which is often lower than in focus.6 In adverbial clauses, on the other hand, null subjects are not obligatory, but alternate with overt subjects. Evidence that the choice between a null form versus an overt form is motivated by constraints on the cognitive status of the referent would support the analysis that adverbial null subjects are not syntactically controlled. To provide such evidence, it is necessary to demonstrate both the similarity in constraints between matrix and adverbial clause null subjects and the fact that the cognitive status of referents of overt subjects of adverbial clauses is not restricted to the same degree as that of null subjects. 4. Data analysis 4.1 Null subjects constrained to referents in focus The Kumyk data fits the expected pattern of correlation between null arguments and the highest category of salience that is demonstrated in crosslinguistic research related to the Givenness Hierarchy and other models. As illustrated in Figure 3, with the exception of two forms (one matrix and one adverbial), both matrix and adverbial clause null subjects have referents that are coded to be in focus by the criteria of the coding protocol.

6

While in some languages it is possible for referents of syntactically controlled arguments to be consistently in focus, this is not true of infinitive and participial clauses in Kumyk, in which the null subject of the non-finite clause often precedes the matrix subject.

57

KUMYK ADVERBIAL CLAUSES

MATRIX CLAUSE

ADVERBIAL CLAUSE NULL SUBJECTS

ADV

NULL SUBJECTS

SS In Focus 18

OVERT SUBJECTS

In Focus 28

Not in Focus 1

Not in Focus 1

DS In Focus 4

Not in Focus 0

In Focus 5

CLAUSE

Not in Focus 21

Figure 3: Data overview

In the two exceptional cases, an argument can be given for why the status of the referent should be treated as comparable to in focus. In both cases, the use of the null form appears to be justified by a factor I call script repetition, which denotes the repetition of a particular series of events in the narrative. In this text, the primary participant, the prophet, promises his only daughter to four different suitors. The first suitor’s representatives come and take the daughter. When the second suitor’s representatives come, the household cat turns into a girl, and they take her away. When the third and fourth suitors’ representatives come, the dog and the donkey, respectively, turn into girls and are taken away. Each group of representatives is a different entity, yet the third and fourth groups are represented with a null form, even though they are not mentioned in the immediately preceding sentences. The fact that the same type of group is mentioned in the same order in the same series of events, however, appears to bring this type into focus for the hearer, thus justifying the use of null forms in reference to the third and fourth groups. In summary, these two apparent counterexamples are not out of character with referents classified as in focus by the criteria of the coding protocol, which means that null arguments have a strong correlation with referents that are in focus. The data support the hypothesis that the use of null subjects, whether in matrix or adverbial clauses, is constrained to referents that have the cognitive status ‘in focus’. This finding is significant, as it shows adverbial null subjects to be following the same pattern as syntactically free matrix null arguments. One might note here that, of the complex constructions in which the subject of the adverbial clause is a null form, the cases are rather evenly divided between cases in which the matrix subject is also represented as a null form and cases in which the matrix subject is overt. Given the methodology applied by this study, all matrix subjects that are coreferent with the preceding adverbial clause subject would be assumed to represent referents that are in focus (cf. criterion 1 of the coding protocol). The Givenness Hierarchy model allows for referents that are in focus to be represented by forms that signal a lower status (e.g., activated, familiar, etc.). This follows from the fact that a particular status on the hierarchy is understood to entail all lower statuses. Since the use of a weaker form could implicate that the status associated with

58

LINDA HUMNICK

the stronger form is not met, it is important to consider contextual motivations for such usage. As an example, an adverbial clause with one participant may be followed by a matrix clause with multiple participants, one of whom is the referent brought into focus via its mention in the adverbial clause. Since the presence of multiple participants increases the likelihood of ambiguity in the interpretation of null arguments, an overt argument may be used for the referent which is in focus in order to reduce the processing effort. In this type of case, as well as others, the use of an overt form — which signals a lower status — for a referent which is in focus is understood to be governed by principles such as Grice’s Maxim of Quantity (1975) or the maximization of relevance (Sperber & Wilson 1986). Finally, the contrast between the cognitive status of referents of null subjects of adverbial clauses and the cognitive status of referents of overt subjects of adverbial clauses provides further evidence that the form of referring expression used in the subject position of an adverbial clause is pragmatically constrained. While the referents of null arguments are almost without exception in focus, the referents of overt subjects of adverbial clauses are predominantly of a lower status than in focus, though up to 25% do refer to referents that are in focus. The fact that a significant number of referents of overt subjects are in focus does not contradict the hypothesis of this study, since, as mentioned above, the implicational nature of the Givenness Hierarchy allows for referring expressions which overtly signal a lower status to be used to represent referents of a higher status. 4.2 Additional observations about DS null subjects Null subjects of adverbial clauses that are not coreferent with the matrix subject occur infrequently in the extended data, and only four times in the primary corpus. In one case, the DS null subject is a non-referential expletive subject, leaving three cases of referential null DS subjects which are pertinent to this study. The rare use of null DS subjects contrasts with the common use of overt DS subjects, which occur with relatively equal frequency to SS subjects in adverbial clauses. Despite the small numbers, a characterization of the use of DS nulls provides insight into the types of criteria that constrain reference enough to overcome a possible bias towards SS interpretation of null arguments. Beyond the fact that all referents of DS nulls are at least in focus by the coding protocol, this section describes three notable features of their use. First, all three referents of DS null subjects are coreferent with the subject of the matrix clause of the previous sentence. The fact that the DS null subject is coreferent with the highest ranked grammatical argument of the previous sentence means that they are also the backwards center as defined by Centering

KUMYK ADVERBIAL CLAUSES

59

Theory (Grosz et al. 1995, Walker et al. 1998), a status which is often associated with deletion or pronominalization of arguments. Whether or not the constraint is framed in the terms of Centering Theory, this is more restrictive than the case of matrix null subjects, which may represent referents that are in focus without being mentioned in the subject position of the previous matrix clause (cf. criteria 3, 4, and 5 of the coding protocol). Not only are the referents of DS null subjects in focus and coreferent with subjects of the previous matrix clause, but in each case in this corpus, the referent may be considered the VIP participant or the protagonist of the narrative, or a plural referent which includes the protagonist. In a cognitive study of referential choice by Grüning and Kibrik (2002), the factor of protagonisthood is demonstrated to contribute to the overall salience of a referent, as is apparently the case here. A third factor that characterizes cases of DS null subjects in this corpus is occurrence within a structure known as ‘tail-head link’. This term refers to cases in which a non-finite clause joined to the left of a matrix clause repeats the information of the previous matrix clause, as illustrated in 14. In this example, the verb geldi “came” (in bold) of the matrix clause in (a) is repeated in the non-finite clause in (b) via the verb form gelgen “having come”. The null subject of gelgen in the non-finite clause (in brackets) is not coreferent with the matrix subject of (b), qïznï qajnanasï “the girl’s mother-in-law”, but, rather, is coreferent with bular in the previous clause (a). (14a) Øi Jashnï gijindirip bulari üjune geldi. child.ACC dress.CONV 3.1.PL home.DAT come.3S.PST “After dressing the child, they (the girl and her sisters-in-law) came home”. (14b) [Øi Üjune gelgen song] qïznï qajnanasïj home.DAT come.PST.PRT after girl.GEN mother.in.law.3S.POSS jashga qaramaga gelinine kömek ete. child.DAT look.INF.DAT daughter.in.law.3.POSS.DAT help do.3S.PR “After coming home, the girl’s mother-in-law helps her daughter-in-law look after the child”. The phenomenon of tail-head link is extended here to cases in which the verb is not actually repeated but there is a strong semantic link between the verb of the previous clause and the verb of the non-finite clause, as with the pair jata “lies down” and turganda “gets up”. In general, tail-head link may be viewed as a form of parallelism that creates a strong expectation that the subjects will be coreferential.

60

LINDA HUMNICK

In summary, the DS null subjects in this study are restricted to use with highly salient referents in contexts with little possibility for ambiguity. They represent the highest ranked grammatical argument of the previous clause, the highest ranked participant in the discourse context, and are found in contexts in which the semantic link between two verbs and the assumed parallelism between their subjects also constrain reference. 5.

Conclusion This study provides evidence that adverbial clauses in Kumyk have null subjects that are syntactically unrestricted yet pragmatically constrained in the same way as null subjects of matrix clauses. Using the cognitive statuses of the Givenness Hierarchy model, the results show that null subjects are constrained to use with referents that have the cognitive status ‘in focus’. Overt subjects of adverbial clauses, on the other hand, are not constrained to referents in focus, which indicates that the choice between null and overt forms is motivated by the status of the referent. The possible evidence for tighter pragmatic constraints on null subjects of adverbial clauses that are not coreferent with the matrix subject (DS nulls) provides an interesting puzzle. On one hand, the evidence that the reference of null subjects of adverbial clauses is syntactically unrestricted together with the common occurrence of overt DS subjects in complex clauses might lead one to expect DS null subjects of adverbial clauses to be more common in the data and to follow the same pragmatic constraints as SS null subjects. This study suggests, however that DS null subjects are much more restricted than their SS counterparts, occurring only with highly salient referents in contexts in which they are coreferent with the subject of the previous clause and in which ambiguity is minimized through the use of tail-head link constructions. If this turns out to be the case, it may indicate a strong tendency to interpret null arguments of non-finite clauses as coreferent with co-sentential subjects rather than subjects of previous clauses. The source of such a tendency in non-finite clauses with syntactically unrestricted reference remains an open question, but may be related to a cross-linguistic tendency towards subject control of implicit non-finite subjects (Haspelmath 1995: 29). Finally, it is worth noting that syntactically unrestricted reference of null arguments in the non-finite clauses of this corpus is limited to cases in which clauses are linked via chronological coherence, which Foley & Van Valin consider the weakest type of semantic linkage (1984: 269). Further study is needed to determine if clausal arguments exhibit syntactically unrestricted null reference in types clause combinations that are higher in the interclausal semantic relations hierarchy, or if this is merely an effect of the limits of the data.

KUMYK ADVERBIAL CLAUSES

61

Abbreviations – ablative, ACC – accusative, CONV – converb, DAT – dative, EMPH – emphatic, FUT – future, GEN – genitive, INF – infinitive, LOC – locative, MOD – modifier, PL – plural, POSS – possessive, PR – present, PRT – participle, PST – past, RFLX – reflexive, S – singular, VN – verbal noun. ABL

REFERENCES Ariel, Mira. 1988. “Referring and Accessibility”. Journal of Linguistics 24.6587. Ariel, Mira. 1990. Accessing Noun-Phrase Antecedents. London: Routledge. Chafe, Wallace. 1976. “Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View”. Subject and Topic ed. by Charles Li, 25-55. New York: Academic Press. Foley, William & Robert Van Valin. 1984. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Givón, Talmy. 1983. Topic Continuity in Discourse: A quantitative crosslanguage study. Amsterdam & Philadelphia: John Benjamins. Grice, Herbert Paul. 1975. “Logic and Conversation”. Syntax and Semantics 3: Speech acts ed. by Peter Cole & Jerry L. Morgan, 41-58. New York: Academic Press. Grosz, Barbara J., Joshi K. Aravind & Scott Weinstein. 1983. “Providing a Unified Account of Definite Noun Phrases in Discourse”. Proceedings for the 21st Annual Meeting of the Association for Computational Linguistics ed. by Donald E. Walker, 44-50. Morristown, NJ: Association for Computational Linguistics. Grosz, Barbara, Aravind K. Joshi & Scott Weinstein. 1995. “Centering: A framework for modeling the local coherence of discourse”. Computational Linguistics 21.203-225. Grüning, André & Andrej Kibrik. 2005. “Modeling Referential Choice in Discourse: A cognitive calculative approach and a neural network approach”. Anaphora Processing: Linguistic, cognitive and computational modeling ed. by António Branco, Tom McEnery & Ruslan Mitkov, 163-198. Amsterdam & Philadelphia: John Benjamins. Gundel, Jeanette, Nancy Hedberg, & Ron Zacharski. 1993. “Cognitive Status and the Form of Referring Expressions in Discourse”. Language 69.274-307. Gundel, Jeanette, Michael Hegarty & Kaja Borthen. 2003. “Cognitive Status, Information Structure, and Pronominal Reference to Clausally Introduced Entities”. Journal of Logic, Language and Information 12.281-299.

62

LINDA HUMNICK

Gundel, Jeanette. 2004. Coding Protocol for Statuses on the Givenness Hierarchy. Ms. University of Minnesota. Haspelmath, Martin. 1995. “The Converb as a Cross-Linguistically Valid Category”. Converbs in Cross-Linguistic Perspective ed. by Martin Haspelmath & Ekkehard König, 1-56. Berlin: Mouton de Gruyter. Prince, Ellen F. 1981. “Toward a Taxonomy of Given-New Information”. Radical Pragmatics, ed. by Peter Cole, 223-255. New York: Academic Press. Sperber, Dan & Dierdre Wilson. 1986. Relevance: Communication and cognition. Oxford: Blackwell. Turan, Ümit. 1998. “Ranking Forward-Looking Centers in Turkish: Universal and language-specific properties”. Centering Theory in Discourse ed. by Walker, Marilyn, Aravind K. Joshi & Ellen Prince, 139-160. Oxford: Clarendon Press. Walker, Marilyn, Aravind K. Joshi & Ellen Prince. 1998. Centering Theory in Discourse. Oxford: Clarendon Press.

FINITE STRUCTURES IN FOREST ENETS SUBORDINATION A CASE STUDY OF LANGUAGE CHANGE UNDER STRONG RUSSIAN INFLUENCE1

OLESYA KHANINA Max Plank Institute for Evolutionary Anthropology, Leipzig ANDREY SHLUINSKY Institute for Linguistics RAS, Moscow 1.

Introduction Enets is a moribund Northern Samoyedic language traditionally spoken on the right bank of the Lower Yenisei River (Taimyr Peninsula, Western Siberia). Now the use of Enets is basically restricted to two villages of the area, Potapovo in the south (Forest dialect) and Vorontsovo in the north (Tundra dialect). Enets is no longer used on an everyday basis, however, about 30–50 people, all over 45 and all native or near-native speakers of Russian, retain substantial production competence and rather full comprehension competence. Yet, field reports (cf., e.g., Helimski, Ms) show that the Enets spoken by these present-day speakers is structurally different in some respects from the language of their parents and grandparents reported in earlier sources, such as Castrén (1854), Prokofjev (1937), Tereschenko (1966, 1977, 1993), Sorokina (1975, 1981, 1985), Cheremisina et al. (1986), Sorokina & Bolina (2005), Urman-chieva (2006). Enets subordination by means of non-finite forms was analyzed by Irina Sorokina in Cheremisina et al. (1986) and Sorokina (1981, 1985), with detailed attention both to forms used and their functions. Non-finite forms are understood in this paper as verbal forms that can never be a head of an independent clause. While these forms are indeed the most typical way to express a subordinated state of affairs in Uralic and in other languages of the region (Cheremisina et al. 1984, 1986; cf. also Anderson 2004), finite structures may also be used for this purpose, especially with verbs of speech 1

We would like to thank the audience at LENCA-3 for their pertinent comments after our presentation; we are also grateful to Eugen Helimski and Bernard Comrie for their critical remarks expressed in person, and to Elena Skribnik for fruitful discussion of general subordination issues. Olesya Khanina’s work on this paper was in part supported by the Fritz Thyssen Foundation (Cologne). That support is hereby acknowledged.

64

OLESYA KHANINA & ANDREY SHLUINSKY

production. Finite structures are understood here as clauses governed by verbal forms that can be used as a head of an independent clause. The following examples illustrate both non-finite (1–2) and finite structures (3–4) of Enets used to express a semantically subordinate state of affairs. (1) b’i-D bar-xon d’ir’i-da-xu-da [water-GEN bank-LOC live-PART-DAT-3SG.OBL.SG]non-finite clause bu odu m’e. s/he boat make.S:3SG “While he lived on the river bank (= to his living on the river bank), he made a boat”. (Sorokina 1981: 143) (2) sojzan mosra-ba-d ek’i d’er’i noda-ba [well work-NMLZ-2SG.OBL.SG]non-finite clause this day hear-S:1PL “Today we have learnt that you worked well (= your working well)”. (Cheremisina et. al. 1986: 127-128) (3) mora-b’i-za kora [kill-NARR-SO:3SGs.SGo]finite clause male.reindeer kas-ta-go osa-xu-da. get.dry-CAUS-DUR.S:3SG meat-DAT.SG-3SG.OBL.SG “He dries the killed (= that he killed) male reindeer for the meat” (Sorokina & Bolina 2005: 51) (4) myt’ e-b’ s&’i e-n’i-s& ’ I [mother-1SG I.ACC not.IRR-SUBJ-PST mydy-s takr’i-q. see-CONN]finite clause hide-R:1SG “I have hidden so my mother would not see me”. (field notes, elicitation) We are unaware of any studies of finite structures like those used in (3) and (4) for Enets. One of the reasons for possible absence of such studies is the former marginality of this morphosyntactic pattern, which became widespread only in last decades due to the mass Russian-Enets bilingualism of the Enets, and in particular, due to the ongoing language shift to Russian. Another possible reason for the lack of attention to this topic is the problematic theoretical status of the interclausal relations in cases like (3) and (4). Indeed, some factors, like linear order, point to the dependent syntactic status of the clauses like morab’iza “he killed” in (3) and eb’ S’i en’iS’ mydys “let my mother not see me” in (4), while other factors, like lack of any markers of the supposed clausal dependence, point to their independent syntactic status. If the latter factors are to be taken as decisive, one should speak rather of discourse-

SUBORDINATION IN FOREST ENETS

65

level strategies of marking the interclausal relations, than of any morphosyntactic subordination; cf. Cristofaro (2003) for a consistent treatment of syntactic subordination vs. semantically subordinated state of affairs. And there is nothing unexpected that discourse-level phenomena are less studied than morphosyntactic phenomena. In this paper we will give an overview of all Forest Enets finite structures that may be used to encode semantic relations of complementation, adverbial and adnominal modification. Without aiming to provide a final answer, we will comment, for each case, on the morphosyntactic arguments for analyzing these structures as morphosyntactically dependent or independent clauses. Particular emphasis will be placed on comparison of the finite structures and their uses for semantic subordination in Enets as it is spoken today with structures used for the same purpose by Enets spoken in 1970–1990. For more on ‘Modern Enets’, cf. ex. (4); on ‘Pre-Modern Enets’, cf. ex. (1–3). Possible reasons for each case of the micro language change will be discussed. Modern Enets data represent spontaneous Forest Enets narratives collected in 2005, a total of 523 sentences, and a set of ca. 300 elicited sentences. Both narratives and elicited data were collected during our short-term field trip to Potapovo in September 20052. Pre-Modern data consist of ca. 3850 sentences from the text collection Sorokina & Bolina (2005). In both cases we limit ourselves to the Forest dialect of Enets, as our field data comes from this variety only, as well as most texts in Sorokina & Bolina (2005). The orthography was neither unified in Sorokina & Bolina (2005), nor was it in our field data: each word tends to be spelled as it was pronounced by an individual speaker; further study is necessary in order to work out a phonological transcription from which all possible individual pronunciations of Modern Enets could be derived. Pre-Modern Enets data was published in Cyrillic which we have transliterated for the purpose of this paper. The rest of the paper will be organized as follows. Section 2 will discuss finite structures headed by a verb in indicative mood: its subsections will discuss unmarked structures (§2.1), as in (§3) above, structures marked by a conjunction (§2.2), and structures marked by a demonstrative (§2.3). Section 3 will discuss finite structures headed by a verb in subjunctive mood, as in ex. (4) above. Section 4 will draw some conclusions. 2

The fieldtrip was supported by the Max Planck Institute for Evolutionary Anthropology (Leipzig). We express our deep gratitude to the institution and, personally, to Bernard Comrie for giving us this invaluable opportunity to embark on the study of Enets. Our warmest thanks go to our Enets consultants Antonina Pujakovna Bolina, Leonid Dmitrievich Bolin, Nadezhda Konstantinovna Bolina, Zoja Nikolaevna Bolina, Nikolaj Dmitrievich Lyrmin, Viktor Nikolaevich Palchin, Ivan Ivanovich Silkin, and Nikolaj Ivanovich Silkin (in alphabetical order).

66

OLESYA KHANINA & ANDREY SHLUINSKY

2.

Finite structures with a verb in indicative mood A finite structure with a verb in indicative mood appeared in (3). The main criterion to identify this structure is the ability of its verb to be used as a head of an independent declarative clause. For example, morab’iza in (3) can be a head of an independent declarative sentence, meaning thus “He killed”. 2.1 Unmarked finite structures with a verb in indicative mood These structures do not have any further marker that could eventually point to their function as a semantically subordinated clause. In other words, not only can their verb be used as a head of an independent clause, the whole structure can function as an independent sentence. This was actually the case of morab’iza “He killed” in (3). The functional load these structures may take is adnominal modification, as in Pre-Modern (3) and Modern (5), and complementation, as in Pre-Modern (6) and Modern (7). (5) Modern Enets kaTa n’e-n’ n’e-xun dud’inka-xaD man child-1SG.GEN.SG child-LOC.SG [Dudinka-ABL.SG to-Da-D a-s& ad’i-D. come-CAUS-SO:3SGs.SGo-PST] sit-S:1SG “I am staying with my grandson (=my son’s child) that (my son) brought from Dudinka” (field notes, elicitation) (6) Pre-Modern Enets obu d’od’i-gon onej ne-r modea-za: what time-LOC Enets woman-DEF see-SO:3SGs.SGo d’aza, c&ej d’azulaa m’eon tar’i d’aza. [go.S:3SG yesterday way along just go.S:3SG] “Some time later, the Enets woman sees her going along the road she took yesterday”. (Sorokina & Bolina 2005: 97) (7) Modern Enets mana man’: aba aj, obus& c&’i n’i say.S:3SG PTCL.DICT [sister INTERJ why this not.REAL.S:3SG neri-r, c&aj n’i bat-or? get.up-FREQ.CONN tea not.REAL.S:3SG pour-FREQ.CONN] “He said: Sister, why doesn’t she get up (and) pour me some tea?” (field notes, spontaneous narrative) In all these sentences the structure in question can function as an independent clause: dud’inkaxaD toDaDas& from (5) would mean then “S/he

SUBORDINATION IN FOREST ENETS

67

brought him/her from Dudinka”; d’aza, c&ej d’azulaa m’eon tar’i d’aza from (6) would mean “(S/he) goes, (s/he) goes along road she took yesterday (literally, along the yesterday way)”; aba aj, obus& c&’i n’i nerir, c&aj n’i bator? from (7) would mean “Sister, why doesn’t she get up (and) pour me some tea?” 2.1.1 Adnominal modification. In this function, the unmarked indicative finite structures are attested both in Pre-Modern and Modern Enets. However, the use of these structures for adnominal modification seems to have increased from Pre-Modern to Modern Enets. Very few clear cases of the phenomena were attested in Pre-Modern Enets, while in Modern elicited data we have significantly more examples of these structures. However, this difference may actually derive from the difference in data type – elicited vs. natural – and not from the real structural change from Pre-Modern to Modern Enets. From a morphosyntactic point of view, the unmarked finite indicative structures in adjectival function are rather to be analyzed as subordinate clauses per se. The syntactic dependence of these structures on the head noun phrase is evidenced by their linear order: they can easily be embedded into the main clause, as in (5). 2.1.2 Complementation. In complement function, unmarked finite indicative structures are regularly attested both in Pre-Modern and Modern Enets. The complement-taking verbs they usually go with are verbs of emotion, perception, cognition and speech production. Only some of these verbs are occasionally attested with complements encoded by other complementation strategies as well, while all other verbs always encode their propositional arguments with unmarked finite indicative structures. The fact that these structures were extensively used for semantic complementation already in Pre-Modern Enets puts a question whether this was an original situation in Enets or whether the influence of Russian syntax, where finite complement clauses predominate, was that strong already in the 1970s to 1990s. One must study Enets texts of earlier periods – i.e., not later than the 1950s, when the state Russification program started3 – to answer this question. Judging from their semantics, sentences of the kind illustrated in (6–7) have a verb with a propositional valency, and one would expect a clause to fill this valence being thus semantically subordinate. However, from a morphosyntactic point of view, there is actually no strong evidence for the subordinate syntactic status of the structures in question in (6–7). The 3

See Vakhtin (2001), Alpatov (1997), and also Grenoble (2004) for details regarding USSR language policy in the region.

68

OLESYA KHANINA & ANDREY SHLUINSKY

embedding of the finite indicative structures into the clause with the complement-taking verb seems problematic. It was not attested in texts, and in elicitation it was strictly rejected in most cases, while it could also be occasionally allowed by some speakers. It would be useful, thus, to check in future the unmarked finite indicative structures against some further syntactic tests for subordination (see, e.g., Haspelmath 1995, Culicover & Jackendoff 1997). This could be impeded by the moribund status of the language, as most speakers seem to allow unrealistically much if asked about grammaticality during elicitation sessions. 2.2 Finite structures with a verb in indicative mood marked by a conjunction Another type of finite indicative structures are structures whose syntactically subordinate status is clearly marked by a conjunction. However, if the conjunction is eliminated, they can also function as independent clauses, similarly to unmarked finite indicative structures. These structures were attested in complement (8) and adverbial functions (9-10). (8) Modern Enets mut’ b’ije-ep kun’ eTe-j I remember-SO:1SGs.SGo [how father-1SG.NOM.SG te poDuro-go-s& . reindeer harness-DUR.S:3SG-PST] “I remember my father harnessing the reindeer”. (field notes, elicitation) (9) Pre-Modern Enets toz to-jz an’ kasa-da k’ee-ed [as come-R:3SG EMPH brother-3SG.GEN.SG side-DAT.SG] kasa-da ner-ta-gu-s&’ pEa-za. brother-3SG.ACC.SG stand-CAUS-DUR-INF begin-SO:3SGs.SGo “As soon as he reached his brother’s side, he began to stand his brother up” (Sorokina & Bolina 2005:191) (10) Modern Enets toDc& ik ol’ga-eT kuji to-b’i... mana [...] [as Olga-FATHER dead come-NARR.S:3SG] say.S:3SG “As soon as Olga’s late father came, he said […]” (field notes, spontaneous narrative) Complement clauses encoded with the help of these structures were occasionally attested both in Pre-Modern and Modern Enets by verbs of cognition and emotions; the conjunction used the most often here was kun’ “how”.

SUBORDINATION IN FOREST ENETS

69

As for adverbial relations, indicative finite structures with conjunctions were occasionally attested both in Pre-Modern and Modern Enets for two semantic types of relations: temporal relation (9–11) and locative relation (12– 13). In Pre-Modern Enets we also encountered some cases of concessive relations (14)4. 2.2.1 Temporal relations. In the case of temporal relations the finite indicative structure goes immediately before the semantically main clause. Most often the conjunction toD “as” is used, thus encoding immediate anteriority, as in (9); in Modern Enets it is usually attested in the form of toDc&ik5, as in ex. (10). In Pre-Modern data there were also attested three examples of the structures with the conjunction kuna “when”. In this case a clause with kuna can be embedded into the semantically main clause, pronouncing thus an additional argument for the syntactically subordinate status of these structures. (11) Pre-Modern Enets Nobgutun k’iuz-noju kuna kaja-ku-za once morning-ADV [when sun-DIMIN-3SG.NOM.SG/PL soje oz’i-go-s& ’, bu nod-b’i bunyk mozu […] recently appear-DUR.S:3SG-PST] s/he hear-NARR.S:3SG dog barking “One morning when the sun was just rising he heard a dog barking […]” (Sorokina & Bolina 2005: 212) 2.2.2 Locative relations. Somewhat less often the finite indicative structures with conjunctions are used to encode locative relations. In this case the structure in question may go before, as in Pre-Modern (12), or after the semantically main clause, as in Modern (13). Structures with the conjunction ku “(to) where” encode direction, as in (12) while structures with the conjunction kunyn/kunny “where” encode location per se, as in (13). (12) Pre-Modern Enets ku d’azu-ma-d koma-z, tony d’aza-z. [where go-NMLZ-DAT want-S:1SG] there go-S:1SG “I go where I want (to go)” (Sorokina & Bolina 2005: 140)

4

In Pre-Modern Enets there were also attested isolated uses of finite indicative structures with conjunctions for other adverbial relations; we do not discuss them here because of the paucity of the data. 5 Apparently from the collocation toD c&ikoD “and”, see Sorokina & Bolina (2001: 140).

70

OLESYA KHANINA & ANDREY SHLUINSKY

(13) Modern Enets myt’ tony kan’i-D kunny kaTa n’e-j d’ir’i. I there leave-S:1SG [where man child-1SG.NOM.SG live.S:3SG] “I went (there), where my son lives” (field notes, elicitation) 2.2.3 Concessive relations. Data from Pre-Modern Enets also contain a few examples of finite indicative structures with the conjunction kun’r’i “however” encoding concession; in all these examples the semantically dependent clause goes before the semantically main clause. (14) kun’r’i jet pon’i-b’i-za, boa e-b’i. [however PTCL do-NARR-SO:3SGs.SGo] bad be-NARR.S:3SG “However he did it, this was bad”. (Sorokina & Bolina 2005:281) There are no Modern Enets examples of these structures rendering concessive relations, but this may well be connected with a limited nature of our Modern data. We would not expect this structure to disappear in Modern Enets. Apart from this doubtful difference, there are no structural changes from PreModern to Modern Enets in the use of finite indicative structures with conjunctions. 2.3 Finite structures with a verb in the indicative mood, marked by a demonstrative These structures are used only as adnominal modifiers. Demonstratives function as relative pronouns, the rest of the structure being the same as in an independent indicative clause. As in the case of finite indicative structures marked by a conjunction, the syntactically subordinate status of finite indicative structures with a demonstrative is clear, as the demonstrative represents by itself a marker of syntactic subordination. (15) Modern Enets en’c&i, c& iki br’igada-xan moT ara, texe d’aDa. person [this.NOM brigade-LOC.SG work.S:3SG] there go.S:3SG “There goes a man that works in a herder-brigade”. (field notes, elicitation) This strategy of relative clause formation has no restrictions on the syntactic position of the relativized noun phrase: even an adjunct can be relativized this way, see (16).

SUBORDINATION IN FOREST ENETS

71

(16) Modern Enets ugulu-xon boc&ka-j c& i-kun b’i noob’ira-D `mokac&i. corner-LOC.SG butt-1SG.NOM.SG [this-LOC water keep-S:1SG] stand.S:3SG “In the corner there’s a cask where I keep water”. (field notes, elicitation) A relative clause encoded by a finite indicative structure with a demonstrative usually follows its head noun and may, thus, be embedded into the main clause, as was the case both in (15) and (16). This strategy of relative clause formation was not attested in Pre-Modern Enets, and in our Modern Enets data it came up in elicitation only. Therefore, this strategy may be seen as a direct outcome of Russian interference, as in Russian relative clauses are mostly finite clauses containing a relative pronoun6. 3.

Finite structures with a verb in subjunctive mood A finite structure with a verb in subjunctive mood was illustrated in ex. (4) above. The main criterion to identify this structure is its ability to function as an independent non-declarative clause. For example, eb’ s&’i en’is&’ mydys in (4) can be a head of an independent optative sentence, meaning thus “Let my mother not see me!”. Thus, the only difference between indicative finite structures, discussed in the previous section, and subjunctive finite structures, discussed here, is the mood of the verb: zero-marked indicative with proper declarative semantics vs. n’i-..-s&’-marked subjunctive with proper optative semantics.7 Besides, all finite subjunctive structures do not manifest any other marker of dependency, i.e., no conjunction or demonstrative can go with them. These structures were attested in our data only as different-subject adverbial modifiers of goal, as in (4), and as different-subject arguments of the verb koma “want”, as in (17). (17) Modern Enets bu koma onej baDa toxola-n’i-r’i-c& . s/he want.S:3SG [Enets language learn-SUBJ-SO:2DUs.SGo-PST] “He wants you two to learn Enets”. (field data, elicitation) Functioning as an adverbial modifier, such a structure can be embedded into a semantically main clause, as in (4). Therefore, at least when used in this function, it is also a case of syntactic subordination. Unfortunately, the data on 6

Note that it is, however, quite surprising to see demonstrative pronouns, and not interrogative pronouns used in Enets finite relative clause formation: in modern Russian, relative pronouns are the same as interrogative pronouns, and different from demonstrative pronouns. 7 See Tereschenko (1966: 452) and Sorokina (1975, 1987) for more details on subjunctive (‘konjunktiv’) mood semantics in independent sentences.

72

OLESYA KHANINA & ANDREY SHLUINSKY

different-subject wanting is insufficient to make any claims about the syntactic status of the finite subjunctive structures in this case. Interestingly, these structures were attested only in Modern Enets and only in elicitation, while in Pre-Modern data different-subject goal relations were expressed by a postpositional strategy, as in (18), and different-subject wanting was surprisingly not attested at all in the whole corpus of about 3850 sentences in Sorokina & Bolina (2005). (18) Pre-Modern Enets bus’i pez’i-za kamaza… bazutu-da old_man wood-3SG.ACC.PL prepare.S:3SG [fire-3SG.NOM.SG sojzaan lojdu-ma d’eon. well burn-NMLZ for] “The old man prepared some firewood, in order for his fire to burn well”. (Sorokina & Bolina 2005: 64) This mismatch between Pre-Modern and Modern structures might again be explained by the interference from Russian. The contexts of use of the subjunctive finite structures in Modern Enets are exactly the same contexts where the Russian subjunctive mood (conjunction c&toby + past tense) is used. Besides, in its independent uses, the Russian subjunctive mood also conveys optative-like non-declarative situations. 4.

Conclusion In this paper we have analyzed the possible forms of finite structures used in Forest Enets to encode semantic relations of complementation, adverbial modification and adnominal modification, and the particular functional slots these forms occupy in these semantic domains. Unmarked indicative structures are used for expression of adnominal modification and for expression of semantic arguments of verbs of emotion, perception, cognition and speech production. Indicative structures with conjunctions are used for expression of adverbial modification by location in time or in space, and for expression of semantic arguments of verbs of cognition and emotions. Indicative structures with demonstratives are used occasionally in Modern Enets for expression of adnominal modification. Finally, subjunctive structures were shown to be used in Modern Enets for expression of adverbial modification by different-subject goal, and for expression of different-subject arguments of the verb koma “want”. In most cases the structures described above were attested both in PreModern and Modern Enets, however the increase in frequency of their usage is evident; in some cases, though, new structures arose really recently. The main

SUBORDINATION IN FOREST ENETS

73

responsibility for this ever-growing usage of finite structures in the encoding of subordinate state of affairs is to be found in interference from Russian morphosyntax. Today many Enets speakers are more proficient in Russian than in Enets, and the others have fairly equal competence in both languages. Therefore, one could claim that the system of subordination in Enets is undergoing a process of rather fast restructuring where a number of crucial roles are transferred from the non-finite structures to the finite ones. The intermediate syntactic status of many finite structures – manifesting either no syntactic dependence, or weak evidence for syntactic dependence – seems to support this thesis. Indeed, a young means of expressing a certain semantic function is generally not expected to possess a clear syntactic profile differentiating it from other means that could equally be used in this function. However, the frequent association of a language structure with a particular function often leads to its grammaticalization, i.e., to its adherence to specific syntactic restrictions, properties, etc. On the latter process, for example, cf. Bybee (1998), Bybee & Hopper (2001) and Haspelmath (to appear). Therefore, it could be suggested that if Enets survives as a living language, some time later the structures analyzed in this paper may well develop a particular syntactic profile of syntactically subordinate structures (e.g., they could start using an obligatory subordinating conjunction, thus following the Russian pattern of finite subordinate clause formation). Abbreviations 1, 2, 3 – 1, 2, 3 person, ABL – ablative, ACC – accusative, ADV – adver-bial marker, CAUS – causative, SUBJ – subjunctive, CONN – connegative (verb form used with negation marker), DAT – dative, DEF – definite, DIMIN – diminutive, DU – dual, DUs – dual subject in subject-object conjugation type, DUR – durative, EMPH – emphatic, FATHER – suffix marking someone’s father (used with proper names), FREQ – frequentative, GEN – genitive, INF – infinitive, INTERJ – interjection, IRR – irrealis, LOC – locative, NARR – narrative, NMLZ – nominalization, NOM – nominative, OBL – oblique case, PART – participle, PL – plural, PST – past, PTCL – particle, PTCL_DICT – speech particle, R – reflexive conjugation type, REAL – realis, S – subject conjugation type, SG – singular, SGo – singular object in subject-object conjugation type, SGs – singular subject in subject-object conjugation type, SO – subject-object conjugation type. REFERENCES Alpatov, Vladimir A. 1997. 150 jazykov i politika: 1917–1997 [150 languages and politics: 1917–1997]. Moscow: Institut Vostokovedenija RAN.

74

OLESYA KHANINA & ANDREY SHLUINSKY

Anderson, Gregory D. S. 2004. “The Languages of Central Siberia: Introduction and overview”. Languages and Prehistory of Central Siberia ed. by Edward J. Vajda, 1-119. Amsterdam & Philadelphia: John Benjamins. Bybee, Joan. 1998. “A functionalist approach to grammar and its evolution”. Evolution of Communication 2.249-278. Bybee, Joan & Paul Hopper, eds. 2001. Frequency and the Emergence of Linguistic Structure. Amsterdam & Philadelphia: John Benjamins. Castrén, Matthias A. 1854. Grammatik der samojedischen Sprachen. St. Petersburg: Kaiserliche Akademie der Wissenschaften. Cheremisina, Maya I. et al. 1984. Predikativnoe sklonenie prichastij v altajskikh jazykakh [Predicative declension of participles in Altaic languages]. Novosibirsk: Izd. “Nauka. Cheremisina, Maya I. et al. 1986. Strukturnye tipy sinteticheskikh polipredikativnykh konstruktsij v jazykakh raznykh sistem [Structural types of synthetic polypredicative constructions in languages of different typologies]. Novosibirsk: Izd. “Nauka”. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Culicover, Peter W. & Ray Jackendoff. 1997. “Semantic subordination despite syntactic coordination”. Linguistic inquiry 28:2.195-218. Grenoble, Lenore A. 2003. Language Policy in the Soviet Union. Dordrecht: Kluwer. Haspelmath, Martin. 1995. “The Converb as a Cross-Linguistically Valid Category”. Converbs in Cross-Linguistic Perspective ed. by Martin Haspelmath & Ekkehard König, 1-55. Berlin & New York: Mouton de Gruyter. Haspelmath, Martin. To appear. “Creating Economical Morphosyntactic Patterns in Language Change”. Language Universals and Language Change ed. by Jeff Good. Oxford: Oxford University Press. Helimski, Eugen. Ms. Fonetika i morfonologija ènetskogo jazyka v uslovijakh jazykovogo sdviga [Enets phonetics and morphophonology under language shift]. Unpublished manuscript. Prokofjev, Georgij N. 1937. “Ènetskij (enisejsko-samojedskij dialekt) [Enets (Yenisei-Samoyedic dialect)]”. Jazyki i pis’mennost’ narodov Severa [Languages and writing systems of the peoples of the North] ed. by G. N. Prokofjev, 75-90. Moscow & Leningrad: Uchpedgiz. Sorokina, Irina P. 1975. Morfologija glagola ènetskogo jazyka [Verbal morphology in Enets]. Unpublished Ph.D. Dissertation. Leningrad. Sorokina, Irina P. 1981. “Zavisimye predikaty s padezhnymi formantami v ènetskom jazyke [Dependent predicates with case affixes in Enets]”. Padezhi i ikh èkvivalenty v stroe slozhnogo predlozhenija v jazykakh narodov Sibiri [Cases and their equivalents in the structure of complex sentence in the

SUBORDINATION IN FOREST ENETS

75

languages of the peoples of Siberia] ed. by Maya I. Cheremisina et al. 138148. Novosibirsk: IIFF SO AN SSSR. Sorokina, Irina P. 1985. “Polipredikativnye konstruktsii s poslelogami v ènetskom jazyke [Polypredicative constructions with postpositions in Enets]”. Polipredikativnye konstruktsii v jazykakh raznykh sistem [Polypredicative constructions in the languages of different typologies] ed. by Maya I. Cheremisina et al. 139-148. Novosibirsk: IIFF SO AN SSSR. Sorokina, Irina P. 1987. “Kategorija vremeni-naklonenija v ènetskom jazyke”. [Tense and mood in Enets]. Problemy fonetiki i morfologii jazykov narodov Severa [Problems of phonetics and morphology in the languages of the peoples of the North] ed. Ju. A. Sem, 58-68. Leningrad: Leningradskij Gos. Pedinstitut im. Gertsena. Sorokina, Irina P. & Darja S. Bolina. 2001. Ènetsko-russkij i russko-ènetskij slovar’ [Enets-Russian and Russian-Enets dictionary]. St. Petersburg: Prosveshchenie. Sorokina, Irina P. & Darja S. Bolina. 2005. Ènetskie teksty [Enets texts]. St. Petersburg: Izd. “Nauka”. Tereschenko, Natalja M. 1966. “Ènetskij jazyk [Enets]”. Jazyki narodov SSSR [Languages of the peoples of the USSR], vol. 3. Finno-ugorskie i samodijskie jazyki [Finno-ugric and Samoyedic languages] ed. by Vasilij I. Lytkin & Klara E. Majtinskaja. 438–457. Moscow: Izd. “Nauka”. Tereschenko, Natalja M. 1977. Sintaksis samodijskikh jazykov: prostoe predlozhenie [The syntax of Samoyedic languages: simple clause]. Leningrad: Izd. “Nauka”. Tereschenko, Natalja M. 1993. “Enetskij jazyk [Enets]”. Jazyki mira: ural’skie jazyki [Languages of the world: Uralic languages] ed. by K. E. Majtinskaja, 343-348. Moscow: Izd. “Nauka”. Urmanchieva, Anna Ju. 2006. “Vremja, vid ili modal’nost’: glagol’naja sistema ènetskogo jazyka”. [Tense, aspect or modality: the Enets verbal system]. Voprosy jazykoznanija 4.88-100. Vakhtin, Nikolaj B. 2001. Jazyki narodov severa v XX veke: ocherki jazykovogo sdviga [Languages of the people of the North in the 20th century: essays on language shift]. St Petersburg: Bulanin.

GRAMMATICIZATION AND RELATIVE CLAUSES IN EASTERN KHANTY OLGA POTANINA Tomsk Polytechnic University 1.

Introduction Relative clauses in Eastern Khanty, an SOV language, are pre-nominal modifiers and clause-initial. Most of the examples below reveal the tendency in Eastern Khanty towards the avoidance of complex sentences: most of the examples represent non-finite (participial) relative clauses and are by far more frequent than finite relative clauses which are more recent and very rare. Most of the sentences resemble a string of clauses, loosely linked, often without any overt syntactic indication of subordination (like overt linkage markers, for instance). But all the clauses seem to perform the main function of subordination to combine several ideas into one linguistic unit. The paratactic character of the Khanty complex sentences reveals the semantic relation between clauses by inference. One of the reasons for considering them relative clauses can be conceptual dependence. There is of course some semantic dependence between clauses. As far as we can tell given the data we had access to, subordination is not well developed in Eastern Khanty. From the point of view of evolution of syntactic constructions, “human languages have had complex sentence structure available throughout recorded history” (Hopper & Traugott 2003: 177) but Khanty preserved the earlier “paratactic” features of clause combining. 2.

Relative clauses in Eastern Khanty To introduce a relative clause in Eastern Khanty interrogative pronouns can be used: kojˆ “who, whose (typically only for persons)”, mo_ƒo_l'i “what” (typically for animals and other non-human animates, as well as for lifeless things), and “what (used possessively)” (Gulya 1966: 82-83). The relative pronoun strategy of relative clause formation in Eastern Khanty is not the main strategy and is infrequent.

78

OLGA POTANINA

Here are some examples of the use of these pronouns: (1a) kojˆ ´nt´ ropˆlt´-w´l, ´nt´ li-w´l who NEG work-PRES.3SG NEG eat-PRES.3SG “Whoever does not work, does not eat”. (Gulya 1966: 86) (1b) mo_ƒo_l'-a_ koj´mp-ˆn, no_N-a_ m´-l-im what-LAT want-PRES.2SG 2SG-LAT give-PRES-1SG/SG “I give you what you want”. (Gulya 1966: 86) (1c) mo_ƒo_l'i ma_n-a_ mas-w´l, t'u m´ji-ƒil´-ƒa_s what 1SG-LAT must-PRES.3SG DEM give-PAST-3SG “He gave me what I need”. (Gulya 1966: 86) Such constructions appeared in Eastern Khanty under the influence of Russian. Example (1a) is the translation of the well-known Russian proverb. This relative clause forming strategy places the relative clause at the beginning of a sentence and introduces it by a relative pronoun, which may be case marked. All clausal arguments tend to precede the verb. These relative clauses are headless and this strategy does not seem to mark the grammatical function of the missing head in the matrix clause. The grammatical function of the missing argument can be easily identified due to a rather fixed word order. The term ‘headless’ is sometimes applied to relatives meaning “the one who”, “whoever”. Examples (1a, b, c) given above demonstrate hypotactic combining of clauses: the matrix clause and the subordinate clause are interdependent. Relative clauses in examples (1a, b) are introduced by relative pronouns that refer to an antecedent, while (1c) contains a resumptive pronoun t'u that marks the return to the matrix clause. Examples (1a, b) are different from (1c): in the former we find no trace of the missing head noun while in the latter this position is occupied by the demonstrative pronoun t'u which is a highly referring alternative and corresponds to the non-referring and indefinite interrogative pronoun. Probably in (1a, b) the missing head has no specific semantic content and sentences are more general (generic statements) and denote some fact (which is one of the uses of the Present Tense). Example (1c) is more specific and the use of the Past Tense requires this argument that is referentially identical to the interrogative pronoun that introduces the relative clause. The use of demonstrative and deictic particles as an argument of one clause to refer to another whole clause is an important device for indicating some relationship among sentences. The use of tense and aspect has the same function (Mithun 1984: 498), when the relationship is that of temporal

RELATIVE CLAUSES IN EASTERN KHANTY

79

sequence. The following examples demonstrate the obligatory presence of the highly referring argument in the main clause – toƒ´pa – that is coreferential with an entire clause: (2a) kol't'´ƒ jo_-s-´n, toƒ´-pa m´n-a_ whence come-PAST-2SG there-ALL go-IMPR.2SG “From where you came, there you go”. (Gulya 1966: 96) (2b) ko&l´-pa kit-l-im, toƒ´-pa m´n-a_ti where-ALL send-PRES-1SG/SG there-ALL go-IMPR.3SG “Where I send him, there he should go”. (Gulya 1966: 142) These examples also express some general idea and the use of the Imperative Mood in the main clause implies, as in (2a), that it does not matter where one came from, one should go back, or, as in (2b), that it is more important that one obeys the order and no matter where one is sent to, one should go. Moreover, toƒ´pa is used to refer to the place that is not visible, not concrete and is more independent of the context. These examples are not typical for the Khanty culture where speaking contributes much to general communication. It is more characteristic of written languages in which reference is often made to entities and events that are not visible to the writer or reader (Perkins 1992: 67). Eastern Khanty makes a robust use of various non-finite constructions for a variety of functions. Non-finite embedded subordinate clauses are the most productive means of making complex sentences in Eastern Khanty. Often syntactically and semantically these clause-size modifiers bear a systematic relation to full-fledged clauses. In Eastern Khanty, which is primarily a spoken language, the more concrete word taƒˆ “place” seems to be a major and obligatory device of forming relative clauses and denoting direction or location: (3a) t'utˆ nomˆƒsˆ-l-´m, taƒa-pa ap-ˆm joƒ´n-a_ k´l-as, thus think-PRES-1SG place-LAT father-POSS.1SG river-LAT die-PAST.3SG lo_ƒ´nta pa_ taƒˆ-n´ joƒ´n t'el´ƒ-´ki surely other place-LOC river not frozen-PRED “So (I) think: probably (my) father drowned where the river is not frozen”. (Tereshkin 1961: 105) (3b) kul wel-ta_-l taƒˆ-j-a wa_n t´ƒˆ kˆt'-´s fish kill-NPP-3SG place-EP-LAT near place exist-PAST.3SG “That place where he was fishing was not far”. (Tereshkin 1961: 106)

80

OLGA POTANINA

(3c) k´lƒos kartovja amt-´m taƒˆ-j-a joƒ-m-am´n, kolkhoz potatoes put-PP place-EP-LAT come-PP-1DU ma_ lo_ƒ-a_, ´t'-ˆm-a, t'u ko_l at-l-´m 1SG 3SG-LAT brother-POSS.1SG-LAT DEM word say-PRES-1SG “When we approached the place where the kolkhoz grows potatoes, I say to him, my elder brother”. (Tereshkin 1961: 106) Examples (3a, b, c) represent a relatively explicit type, i.e., the case when an overt noun phrase can be found in both clauses. The relative clause constructions (3a, b) are somewhat emphatic constructions which foreground some part of the sentence (similar to English cleft sentences). Example (3a) represents a more complex construction of a sentential complement of the verb with an embedded relative clause with a finite verb, while (3b) and (3c) contain participial relative clauses. Moreover, (3a) has another structural peculiarity, as the relativized position is case marked while participial construction in (3b) allows relativization of unmarked arguments. Since the NPs in (3a) already contain case markers, additional use of relative pronouns is not important for maintaining co-reference. As illustrated by examples (3a, b, c), in Eastern Khanty taƒˆ “place” is a frequent device for relative clause formation indicating direction or location. However, in the examples below, taƒˆ “place” is no longer the head of the relative clause construction: (4a) m´N pu_t'k-a_l'i lulp´nˆ-t´ taƒˆ kol´nt´-ƒal-oƒ 1PL bird-DIMIN sing-NPP place listen-PAST-1PL “We listened to the birds singing (lit., “birds’ singing place”)”. (Gulya 1966: 92) (4b) lop´l-t´ j´-min taƒˆ angle-NPP become-CONV place “Fishing” (= “place that has become fishing”) (Gulya 1966: 137) (4c) ma_ ˆl-´lˆnt´-l-´m iƒ-n´ ma_n-t 1SG down-lie-PRES-1SG bear-LOC 1SG-ACC nu-li-t´ taƒˆ ´nt´ wu-t-am-a up-eat-NPP place NEG see-IMP.P-1SG-ILLAT “I lay down (so as) not to see the bear ripping me apart”. (= “the bear’s place ripping me apart”) (Gulya 1966: 129) This process turns a relative clause construction into a noun phrase denoting a more abstract phenomenon that “place”. The analysis of examples (4a, b, c) suggest that the postpositional noun taƒˆ “place” is undergoing

RELATIVE CLAUSES IN EASTERN KHANTY

81

grammaticalization, and now turns into a nominalizer performing the function of linguistic representation of highly abstract notions for which Eastern Khanty lacks nouns. The original locative meaning of taƒˆ “place” limits the use of this nominalizer to the relative clause constructions. But now it has assumed a more general abstract meaning as the original locative meaning “place” has bleached out. The noun is gradually undergoing desemanticization. Moreover, examples (5a, b, c) below demonstrate a case of the postpositional noun taƒˆ “place” taking on a different, and more abstract temporal meaning. The stable syntactic position of this nominalizer, together with its occurrence in certain combinations, contributes to the loss of the referential meaning of this word. The lexeme “place” has acquired some grammatical properties of a functional word, namely, in examples under (4) and (5) functioning as a nominalizer it is not marked for case, it does not have plural and dual forms and is used in non-finite relative clause constructions only. The degree of grammaticization of taƒˆ is high due to its stable and fixed syntactic position. Most of the relative clauses with the nominalizer taƒˆ serve to express the time of an action or other abstract notions: (5a) ajƒ´ wo&l-´m taƒˆ little be-PP NMLZ “The place that was little”. (=childhood) (Csepregi 2005) (5b) ma&ki wo&l-´m taƒˆ past be-PP NMLZ “The place that was in the past”. (Csepregi 2005) (5c) ´jnam to&N´mt´-t´ taƒˆ to&j-´l everything understand-NPP NMLZ have-PRES.3SG “Everything has sense”. (Ajpin 2003: 81) The nominalizer taƒˆ “place” seems to be found mainly in the Eastern dialects of Khanty. Eastern Khanty has most of the cross-linguistically attested nominalizers, including qu “man”, ot “thing”, wer “affair”. Two different ways of forming relative clauses in Khanty (with a finite verb and an interrogative pronoun and with a participial verb and no relative pronoun) can be explained in terms of underlying causal principles of possible change in language type as the result of areal contacts with the speakers of a different language type (in this case SVO Russian language of their neighbors). Hopper & Traugott (2003: 126) give the example of Estonian, historically an OV language with a prenominal participial relative clause, which under the influence of its Germanic and Slavic neighbors has assumed some VO features

82

OLGA POTANINA

and the newer type of relative clause formation — with a finite verb and a relative pronoun. The participial constructions, illustrate a process that is common in Eastern Khanty as well as typologically across languages: the conversion of a finite verb phrase into a component of a noun phrase in another clause, a typically pre-nominal clause-initial attributive nominal modifier. Most Eastern Khanty examples demonstrate use of the gap strategy, where the missing co-referent argument is omitted without a trace. According to Givón, such languages may simply extend their discourse anaphora strategy to the more syntactic environment of relative clauses (Givón 2001: 185). The examples below also represent the result of nominalization, with clauses having less of a verbal character. Perfect participles designate a resultant state, while imperfect participles denote more complex, verb-like relations. (6a) ma_ wer-m-a_m rˆt 1SG do-PP-POSS.1SG boat “The boat I have made”. (Field Notes) (6b) no_N m´n-t-a_n lo_k 2SG go-NPP-POSS.2SG way “The way along which you go”. (Field Notes) The majority of the Eastern Khanty examples demonstrate this nominalization strategy when the matrix clause has fully finite syntax while relative clauses are nominalized. The ‘nominal’ examples in Eastern Khanty function exactly like the examples of adjectives combined with nominal inflections: it refers to an object involved in the action denoted by the verb, not the action itself (Croft 1991: 75). Eastern Khanty permits the relativization of subjects, direct objects, locatives, and instrumentals: (7a) jaƒ m´n-´m-´n qat-´t j´m qat-´w people go-PP-3PL house-PL good house-ATTR “The people who had abandoned the(ir) houses (now) have good houses”. (Field Notes) (7b) ma_ kolente-l-´m mer´m muƒuj jatesw-w´l aNk-im 1SG listen-PRES-1SG tale DET tell-PRES.3SG mother-POSS.1SG “I listen to the tale that my mother is telling”. (Field Notes)

RELATIVE CLAUSES IN EASTERN KHANTY

83

(7c) tS'ul'a_n´ ´j´mkit´m quj-t ´nt´ qoƒ nom´ƒs´k-min then young man-PL NEG long think-CONV noƒ wer-´t m´n-´t toƒ-l-a down do-PRES.3PL go-PAST.3PL DET-3SG-ILLAT qo w´l-´t tS'u jaƒ where live-PAST.3PL DET people “Then the young men, not thinking much, made arrows and went where those people lived”. (Field Note) (7d) ma_ wel-ka_s-im ko_tS´ƒ` ti ni o_ƒo_-w´l n'an' 1SG do-PAST-1SG knife DEM woman cut-PRES.3SG bread “I made the knife which that woman cuts the bread with”. (Field Notes) There are also occasional (7a) that show embedded post-head but prematrix non-finite relative clauses with no relativizer. Increasingly in the recent data there are examples of the relative clauses not only following the head but also coded by the finite dependent clauses. In examples (7a–d) the syntactic and semantic features of the relative clauses show that these clause-size nominal modifiers bear systematic identity to full-fledged independent clauses. 3.

Conclusion The investigation of the relativization strategies in Eastern Khanty demonstrates that the most frequent non-finite relative clause constructions gave rise to the nominalizer taƒˆ “place”. This element is undergoing grammaticalization and serves now to express highly abstract ideas like “childhood”, “the past”, “fishing”, etc. for which Eastern Khanty lacks nouns. The postpositional noun taƒˆ “place” has undergone a semantic shift, resulting in a more abstract temporal meaning. The stable syntactic position of this nominalizer, along with its high frequency in certain combinations, contributes to the loss of the referential meaning of this word. Abbreviations ALL – allative case; ATTR – attributive; CONV – converb; DEM – demonstrative; ILLAT – illative case; IMP – imperative; INF – infinitive; INSTR – instrumental case; LAT – lative case; LOC – locative case; NEG – negative; NMLZ – nominalizer; NPP – non-past participle; O – object; POSS – possessive; PL – plural; PRED – predicate; PRES – present tense; PROLAT – prolative case; PP – past participle; SG – singular.

84

OLGA POTANINA

REFERENCES Ajpin, Eremej. 2003. Ja slushaju zemlju [I listen to the earth]. St. Petersburg: Prosveshenie. Croft, William. 1991. Syntactic categories and grammatical relations. Chicago: University of Chicago Press. Csepregi, Martha. 2005. “Prichastnye oboroty v khantyjskom jazyke”. Paper presented at the 10th Finno-Ugric Congress, August 15–21, 2005. JoshkarOla. Givón, Talmy. 2001. Syntax. Vol. 2. Amsterdam & Philadelphia: John Benjamins. Gulya, Janos. 1966. Eastern Ostyak Chrestomathy. (Indiana University Publications. Uralic and Altaic Series, Vol. 51). The Hague, Netherlands: Mouton de Gruyter. Hopper, Paul & Elizabeth Traugott. 2003. Grammaticalization. Cambridge University Press. Mithun, Marianne. 1984. “How to Avoid Subordination”. Proceedings of the 10th Annual Meeting of the Berkeley Linguistic Society, February 17–20. 493–509. University of California: Berkeley Linguistic Society. Perkins, Revere. 1992. Deixis, grammar and culture. Amsterdam; Philadelphia: John Benjamins. Potanina, Olga. 2005. “Relativization strategies in Eastern Khanty”. Proceedings of the University of Cambridge Third Postgraduate Conference in Language Research, 403-410. Cambridge: CIRL. Tereshkin, Nikolaj. 1961. Ocherki dialektov khantyjskogo jazyka. [Grammar sketches of Khanty dialects]. Moscow & Leningrad: Akademija Nauk SSSR.

TOWARD A SEMANTIC TYPOLOGY OF COORDINATION1 ELENA RUDNITSKAYA Institute of Oriental Studies, Moscow ELENA URYSON Vinogradov Russian Language Institute, Moscow 1.

Introduction The object of our study is the semantics of coordination. The present article investigates two types of linguistic units: 1) Main coordinate conjunctions linking clauses, cf. the Russian i, a, no (cf. Poshel dozhd’, i stalo prokhladnej; Poshel dozhd’, no prokhladnej ne stalo), or the English and, but (cf. It started raining and got cooler; It started raining, but didn’t get cooler); 2) Semantically similar entities (affixes) in a language without Europeanlike conjunctions or syntactic coordination, cf. Korean -ko, -myense, -se, (u)na, -ciman, -(nu)ntey. Coordinate conjunctions/affixes in a given language are a subsystem of linguistic units. We aim to establish semantic oppositions that are typologically relevant for such a subsystem. A typological classification of coordinate constructions has already been created (Haspelmath 2004). However, the account of the semantic features of a conjunction/affix is restricted in this study by the traditional terms ‘conjunctive’ or ‘adversative’. We will use more detailed semantic meta-language and describe the meaning of ‘conjunctive’ (or ‘coordinative’) and ‘adversative’ units with a set of simple semantic elements. In this paper, we will explore two hypotheses: (A) Coordinating conjunctions/affixes in different languages express a unitary set of meanings. So, one can to some degree predict the meaning of coordinate conjunctions/affixes in any given language. We ignore idiomatic semantic components that are irrelevant for typology. (B) The choice of a conjunction in the P-Q coordination construction can be conditioned by two factors: (a) the meaning of P and Q; (b) the strategy of 1

We thank Hwan Sogen for her extensive help in getting and discussing the Korean data. We also thank Maria Polinsky and our LENCA-3 audience for helpful comments. Elena Uryson is grateful for financial support from “RGNF” (project 06-04-00289a), “OIFN RAN”, and the Russian President Science Program (project NSH-5611.2006.6).

86

ELENA RUDNITSKAYA & ELENA URYSON

linking P and Q chosen by the Speaker. The set of such strategies is universal. Languages, however, differ in strategies preferred in this or that context/ situation. The present investigation is the first stage of checking these hypotheses. As a start, we have taken the set of meaning oppositions contained in a sub-system of Russian coordinate conjunctions i, a, no (Uryson 2000, 2004a,b, 2006). In the paper we test whether Korean data meet these oppositions; we also draw English data. 2.

Coordination patterns in Korean Clause coordination constructions in Korean (an Altaic language) can be divided into two types. For the sake of simplicity, we consider only bi-clausal coordination: a) The predicate in each clause P and Q is a full verb; P and Q are linked by a connector (an adverb-like word, not an affix). Such structures can be represented as PFINAL CONN QFINAL. (1) [Swuni-nun Yenge-lul kaluchy-ess-ta] kuliko [Swuni-TOP English-ACC teach-PAST-DECL] KULIKO [Hakswu-nun Ilpon mal-ul pauw-ess-ta] [Hakswu-Top Japanese word-ACC learn-PAST-DECL] “Swuni taught English, and Hakswu studied Japanese”. b) Q’s predicate has a full form, but P’s predicate is a converb, cf. (Haspelmath 1995). In a coordinate sentence, this converb ends with a “coordinate” affix. The scheme of such sentences is: PCONV-AFF QFINAL. In (2), the converb kaluchi-ko has the -ko “conjunctive” affix. (2) [[Swuni-nun Yenge-lul kaluchi] -ko [[Swuni-TOP English-ACC teach] -KO [Hakswu-nun Ilpon mal-ul pauw]] -ess-ta [Hakswu-TOP Japanese word-ACC study]] -PAST-DECL “Swuni taught English, and Hakswu studied Japanese”. There are different ways of looking at the syntactic status of (1) and (2). In (1), it is not entirely clear whether we face one complex sentence or two subsequent sentences in a two-sentence micro-text. An alternative translation of (1) would be “Swuni taught English. And Hakswu studied Japanese”. Cases like (2) can be regarded both as coordinate and subordinate because they meet only some of syntactic coordination tests and criteria, cf. (Yi 1994, Yoon 1994, Rudnitskaya 1998). In this article we do not consider this issue.

87

THE SEMANTICS OF COORDINATION

In this paper we restrict ourselves to affix coordination such as (2). We consider only the affixes -ko, -se, -una, -ciman, leaving out -myense and nuntey for the sake of brevity). This set was obtained via translating Russian sentences with i, a, no into Korean. Russian sentences can also be translated with connectors as in (1), but in our data, connectors occur much more rarely. Russian, English and Korean conjunctive and adversative units are given in the Table 1. Conjunctive Adversative

Russian i a, no

English and but

Korean -ko, -se -ciman, -(u)na

Table 1. Cross-linguistic comparison of conjunction forms and functions

3.

Semantic oppositions in the system of coordinate conjunctions/ affixes The only distinctive features usually employed for the semantic analysis and classification of coordinate conjunctions/ affixes are ‘conjunctive’ and “adversative”. Note that we do not discuss the analysis of conjunctive affixes as contextual converb affixes (Alpatov & Podlesskaya 1995). These two features cannot adequately account for all uses of the coordinate units. In particular, you cannot identify the class of certain uses of a (cf. Kolja bogatyj – Vanja bednyj “Nick is rich, and/ but John is poor”). We use a more detailed set of features that reveals the semantic essence of the ‘conjunctive’ and ‘adversative’ units and therefore accounts for meta-textual functions of the conjunctions/affixes. As a result, we can fix typologically relevant distinctions both in the classes of conjunctive and adversative units. Our description is based on two well-known ideas, cf. for instance (Lakoff 1971, Carlson 1985, Blakemore 1987, Kitis 2000, Levin 1970, Sannikov 1989, Uryson 2000, 2004a,b, 2006). First, an adversative conjunction (such as but, Russian no) is associated with “cancelled expectation”. Cf.: (3) (P) It started raining in the morning, BUT (Q) the children were taken to the beach. Here the first conjunct P induces the expectation “the children were not taken to beach”, and this expectation falls in the second conjunct Q. Second, a conjunction such as and, Russian i, is the unmarked member of the sub-system. In particular, the conjunction and/ i confirms the expectation, as illustrated in (4): (4) (P) It started raining in the morning, AND (Q) the children were not taken to the beach.

88

ELENA RUDNITSKAYA & ELENA URYSON

Here the first conjunct P induces an expectation “the children were not taken to beach”, and the second conjunct Q coincides with this expectation. A serious difficulty for the semantic analysis of coordinate conjunctions is that they are used not only in cases of ‘cancelled’ or ‘confirmed’ expectations, but also in other contexts: (5) (P) The sun hid behind a cloud, BUT (Q) then showed again. According to the analysis above, the expectation in (5) would be something like “the sun is going to be behind a cloud for a long time or forever”. However, P in (5) does not induce any expectations of this sort. Such contexts require additional analysis. Cf. also (6). (6) (P) Nick is rich, BUT (Q) John is poor. In (6), but does not obviously imply any ‘cancelled’ expectations. In Russian contexts such as (6), the conjunction a is used; this use of a is called “a of comparison” by Russian grammarians. A goal of the present article is to describe cases like the ones illustrated in (5) and (6). It is natural to interpret a ‘cancelled’ expectation as contrariness-toexpectation. Therefore, the main feature used in our account is ‘in accordance with vs. contrary to expectation’. A contrariety to expectation is marked by an adversative conjunction/affix; an accordance to expectation — by a conjunctive one. If, however, a language has more than one adversative or conjunctive unit (such as Russian and Korean), the choice of a concrete entity depends on the nature of the expectation. The feature of accordance/contrariety to expectation thus falls into several sub-features. 3.1 Expectation based on prior knowledge of the scenario In the P-Q coordination construction, the first conjunct P induces an expectation based on our knowledge of the reality and its laws, that is, of scenarios. The second conjoint Q denotes a situation that accords to or contradicts this expectation. 3.1.1 Correspondence to the “scenario” expectation. The simplest case involves a situation P, which is the reason for situation Q, or in other words, Q is implied by P; as in (4). Cf. also (7a). (7a) (P) Temno, I (Q) nichego ne vidno “(P) It is dark, AND (Q) you cannot see anything”.

THE SEMANTICS OF COORDINATION

89

Here a situation like P typically induces the situation described in Q. So, P induces an expectation ‘nothing can be seen’, with Q expressing the expected situation. Cf. the Korean equivalent in (7b): (7b) (P) Etwuwe-SE / Etwup-KO, (P) dark-SE /-KO, (Q) amwu kes-to poi-ci anh-nun-ta (Q) any thing-ALSO be-seen-CONV not-PRES-DECL “(P) It is dark, AND (Q) you cannot see anything”. Ex. (8a) illustrates another case: (8a) (P) Temno, I (Q) u nas net fonarja “(P) It is dark, AND (Q) we have no lantern”. Situations like P and Q usually cause the same situation “nothing can be seen”. The Korean equivalent appears in (8b): (8b) (P) Etwup-KO/ *Etwuwe-SE, (P) dark-KO/ *dark-SE (Q) wuli-hanthye-nun tung-i eps-ta (Q) we-LOC-TOP lantern-NOM be.NEG-DECL “(P) It is dark, AND (Q) we have no lantern”. In (8a) and (8b), both P and Q induce the expectation “nothing can be seen”. So in (8a), Q confirms the expectation already induced by P. P and Q are both reasons for a certain situation R. These examples require some commentary. Korean, unlike Russian or English, has two conjunctive units: -se and -ko. The morpheme -se has a semantic component ‘P is the reason for Q’. That is why -se cannot be used in (8a). -Ko does not involve any casual relation between P and Q, neither it marks the absence of this relation. Thus, -ko is possible both in (7a) and (8a). If -ko is used in (7), the causal relation between P and Q is implied from the context. Cf. also (11a), (12b), (13a) and (14a). English and Russian do not differentiate between sentences such as (7a) and (8a). Both types require and, i. 3.1.2 A cancelled scenario expectation. We have distinguished two types of accordance to expectation. Types of contrariety are the same. In the simplest case the second conjoint Q negates the expectation, as in (3). Cf. also: (9a) (P) Temno, A/ NO (Q) my vse vidim “(P) It is dark, BUT (Q) we see everything”.

90

ELENA RUDNITSKAYA & ELENA URYSON

Here, a situation like P usually causes a situation “not-Q”; so, P induces an expectation “nothing can be seen”. Q “we can see everything” cancels this expectation.]. Example (9a) gives the Korean equivalent: (9b) (P) Etwup-CIMAN/ Etwuw-UNA, (P) dark-CIMAN/ dark-UNA (Q) wuli-hanthye-nun motun ke-i (Q) we-LOC-TOP all thing-NOM “(P) It is dark, BUT (Q) we see everything”.

poi-n-ta. be.seen-PRES-DECL

Here P expresses a typical reason for not-Q. Therefore, P induces an expectation of not-Q. The adversative in (9a) and (9b) corresponds to the conjunctive in (7a) and (7b). The Russian sentence and its English equivalent in (10a) illustrate another type of contrariety-to-expectation: (10a) (P) Temno, NO (Q) u nas est’ fonar’ “(P) It is dark, BUT (Q) we have a lantern”. Example (10b) gives the Korean equivalent: (10b) (P) Etwup-CIMAN/ Etwuw-UNA, (P) dark-CIMAN/ dark-UNA (Q) wuli-hanthye-nun tung-i (Q) we-LOC-TOP lantern-NOM “(P) It is dark, BUT (Q) we have a lantern”.

iss-ta exist-DECL

In (10a) and (10b), P induces the same expectation as in (9a) (“nothing can be seen”). But Q (“we have a lantern”) does not directly cancel this expectation, but instead gives an additional sufficient reason for the expectation to be cancelled. So, P and Q are two reasons for a not-R (“something can be seen”). Example (10a), with an adversative conjunction, corresponds to (8a) with a coordinating conjunction. For the sake of brevity, we will not consider other types of adversative contexts here and will limit ourselves to a few comments on the example already discussed. English has only one adversative but; it is used in both cases under discussion. Korean and Russian each have two morphologically distinct adversatives: -una and -ciman; no and a. As far as we could ascertain, Korean adversatives -ciman and -una have no semantic differences. There is only stylistic distinction between them: -una is more literal, but -ciman is neutral and often used in a spoken discourse. We do

THE SEMANTICS OF COORDINATION

91

not consider syntactic distinction between -una and -ciman not directly relevant to the semantic topic of this paper. Russian adversative conjunctions are most interesting. In case (i), both can be used. Subtle semantic distinctions between them are described in (Sannikov 1989, Paducheva 1997, Uryson 2004a). In Russian sentences such as (10), no is normally used. But in a reply a is natural: -Ne khodite tuda! Temno! – A u nas fonar’ est’! “Don’t go there! It is dark! – But we have a lantern!” 3.2 An expectation related to the “inertness” of consciousness The Russian, English and Korean examples in (11a) and (11b) illustrate what can be called the “inertness” of consciousness, as the coordinating conjunctions linking the second clause of these sentences express an expectable elaboration of the situation expressed in the first clause: (11a) (P) Utrom nachalsja dozhd’ I (Q) lil ves’ den’. “(P) It started raining in the morning, AND (Q) (it) poured all day long”. (11b) (P) Achim-ey (P) morning-LOC

pi-ka rain-NOM

o-ki sicakhay-ss-KO, go-NOMIN begin-PAST-KO

(Q) halwu congil ssotacy-ess-ta (Q) day entire pour-PAST-DECL “(P) It started raining in the morning, AND (Q) (it) poured all day long”. One could ask whether there is any ‘confirmed expectation’ in such sentences. In (11a) and (11b), P introduces a certain situation (‘it started raining’). Having fixed P in his consciousness, the subject/ reader/ hearer gets used to it, so to say, gets ‘tuned’ in to it. It is kind of difficult to get tuned to a new situation. In other words, because of the ‘inertness’ of the consciousness, we are ready for the rain to continue rather than to stop. This ‘expectation’ (or, better said, ‘tuning’) holds not because of any scenario (if the rain starts, it may or may not continue for a long time), but because of the inertness of consciousness. Therefore, Q “the rain poured all day” confirms this ‘tuning’ of consciousness; Q accords to ‘expectation’. In this case, a conjunctive is normally used. An ‘adversative’ illustration is provided by the Russian and English examples in (12a) and their Korean equivalent in (12b): (12a) (P) Solnce sprjatalos’ za oblako, NO/A (Q) potom pokazalos’ opjat’. “(P) The sun hid behind a cloud, BUT (Q) then showed again”. (12b) (P) Hay-ka

kwulum

twi-lo

swum-ess-CIMAN/-UNA,

92

ELENA RUDNITSKAYA & ELENA URYSON

(P) sun-NOM

cloud

back-DIR

hide-PAST-CIMAN/-UNA

(Q) taum-ey tasi naw-ass-ta. (Q) later-LOC once.more show-PAST-DECL “(P) The sun hid behind a cloud, BUT (Q) then showed again”. Here, the situation P “the sun is behind a cloud and does not shine” is fixed in the consciousness, and the subject is ready for perceiving it. In other words, this situation is ‘expected’ to keep. The expectation holds not because of any scenario (if the sun hides behind a cloud, it may or may not stay there for a long time), but because of the inertness of consciousness. Therefore, Q “the sun then showed again” cancels this ‘tuning’ of consciousness. Thus, Q contradicts the ‘expectation’; in this case, an adversative is normally used. ‘Expectation’ due to the inertness of consciousness is an important case for us because its analysis allows one to understand the meta-text (discourse) functions of conjunctives and adversatives. These functions are closely related to strategies of linking P and Q chosen by the Speaker. We will have a brief outlook on this issue. 4.

Strategies of linking P and Q In this section, we consider Speaker’s different strategies that determine the choice of conjunctives vs. adversatives. We show that in certain contexts, Russian, English, and Korean differ in strategies preferred. 4.1 Different strategies in a given language Sometimes we can use an adversative or a conjunctive for describing the same situation. Cf. examples (12a) and (12b), which contain adversative conjunctions with the coordinating conjunctions in Russian and English (13a) and Korean (13b): (13a) (P) Solnce sprjatalos’ za oblako I (Q) potom pokazalos’ opjat’. “(P) The sun hid behind a cloud, AND (Q) then showed again”. (13b) (P) Hay-ka (P) sun-NOM

kwulum cloud

twi-lo swum-ess-KO, back-DIR hide-PAST-KO

(Q) taum-ey tasi naw-ass-ta. (Q) later-LOC once.more show-PAST-DECL “(P) The sun hid behind a cloud, AND (Q) then showed again”. In (13a, b), P has a general interpretation “something happens to the sun”. Therefore, the consciousness is ‘tuned’ to changes with the sun going on. In other words, the state of affairs “happenings with the sun” will keep, or

THE SEMANTICS OF COORDINATION

93

something else will happen to the sun. Q “the sun showed again” confirms this expectation. So, a conjunctive marks accordance to ‘expectation’. In (12a, b), P has a more specific interpretation “the sun hid behind a cloud”. The tuning of the consciousness (the ‘expectation’) induced is that the sun will stay behind the cloud for a considerable time. The second conjoint Q “the sun showed again” cancels this expectation; an adversative is used. In this case, the choice of an adversative or a conjunctive depends on the Speaker’s interpretation of the situation P. We believe that in any language with conjunctives and adversatives, the choice between them can depend on this factor. 4.2 Distinctions between languages (strategies in comparison contexts) The languages under discussion differ in strategies used in some contexts. The Russian, English and Korean (14a) examples provide a sample comparison context. (14) [Spoken to compare two friends.] (P) Kolja bogatyj, (Q) Vanja bednyj. “(P) Nick is rich, AND/ BUT (Q) John is poor”. (14a) (P) Kkollya-nun (P) Nick-TOP

pwuca-KO/-CIMAN/-UNA rich-KO/-CIMAN/-UNA

(Q) Panya-nun kananha-ta (Q) John-TOP poor-DECL “(P) Nick is rich, AND/ BUT (Q) John is poor”. Here, Russian allows only the adversative conjunction a, whereas English and Korean often allow both an adversative and a conjunctive (but and and; -ko and -ciman/-una). The situation of comparison presupposes three components: (a) the (two) compared objects; (b) the base of comparison; (c) the result of comparison (in our case, the difference between the objects). The Speaker can specially highlight one of these three components. For the Russian sentence in (13a), it is important that first Nick and then John is in focus. The change of the object in focus is marked by the adversative a. P is about Nick, it induces the expectation (the ‘tuning’) that Q will be about the same object as P (also about Nick). The expectation is due to the inertness of consciousness. However, John is the focus of Q. The expectation induced by P is cancelled; an adversative is used. Russian thus highlights the change of object in focus, as well as the difference between the objects; cf. Kalkova & Podlesskaya (2001), Uryson (2004b).

94

ELENA RUDNITSKAYA & ELENA URYSON

English uses two strategies in such contexts. According to one of these strategies, the adversative but is chosen for linking P and Q. But marks the contrast between the compared objects; this contrast breaks the inertness of the consciousness. That is why the adversative can be used in (13). According to the other strategy, the conjunctive and is chosen in the same case. The conjunctive can be used here because P and Q have a joint component in the situation of comparison. It is the base of comparison (in (13), Nick’s and John’s incomes). This component gets fixed in P and remains in Q. The base of comparison’s not changing in Q accords to “tuning” of the Speaker’s consciousness induced by P; so and can be also used in (14). Korean possesses the same two strategies as English does for expressing these types of meanings. A different situation pertains in the case of dialog or narration shaped as comparison. Example (15a) shows a Russian dialog between persons A and B, the same is reproduced for Korean in (15b). (15a) A: B: “A: B:

Chto delajut deti? Katja risuet, A Petja s koshkoj igraet. What are the children doing? (P) Kate is drawing, AND/BUT (Q) Peter is playing with the cat”.

In this sentence, BUT requires a special continuation, such as “as always”, expressing irony and disapproval. (15b) A: Ai-tul-un mwues-ul ha-ko iss-e yo? child-PLUR-TOP what-ACC do-CONV be-INF POLITE B: Kkacya-nun kulim-ul kuli-KO/-CIMAN/-UNA, Kate-TOP picture-ACC draw-KO/-CIMAN/-UNA Ppeyccya-nun koyangi-lang nol-a yo Peter-TOP cat-COMIT play-INF POLITE Even though (15a, b) resembles a comparison context; the Speaker does not compare Kate and Peter, but instead simply reports what are they doing. It is important that Kate and Peter are doing different things. The Russian sentence in (15a) uses the same ‘adversative’ strategy as for sample comparison. In English, the conjunctive and is preferred in such contexts. And is used here because P and Q have a joint component (‘general topic’ - ‘children’s actions’). P fixes this component; Q keeps it. The base of comparison’s not changing in Q accords to “tuning” of the Speaker’s consciousness in P, so and can be used in (15a). But if the Speaker wants to express irony, sarcasm, etc., s/he can also employ non-standard strategies and use but in cases like this.

THE SEMANTICS OF COORDINATION

95

In the Korean equivalent (15a), one can choose between ‘adversative’ and ‘conjunctive’ strategies, similar to sample comparison in (14a, b). Thus, dealing with the choice of these strategies, Russian is the most rigid language, Korean is the most democratic one, whereas English is somewhere in the middle. 5.

Conclusions The main coordinate conjunctions have often been described in terms of ‘confirmed/cancelled expectations’. We have fixed two psychological types of expectations. The first type is already well described; the expectation is based on our common knowledge (probably shaped as scenarios/frames). The expectation of the second type emerges due to the ‘inertness’ of the mind. ‘Confirmed expectation’ of any type is marked by a conjunctive. ‘Cancelled expectation’ of both types is marked by an adversative. Finally, the languages under discussion differ in preferred strategies of conjunctive/adversative item choice. REFERENCES Alpatov, V. M. & V. I. Podlesskaya. 1995. “Converbs in Japanese”. Converbs in Cross-Linguistic Perspective ed. by Martin Haspelmath & Ekkehard König, 465-486. Berlin & New York: Mouton de Gruyter. Blakemore, Diana. 1987. Semantic Constraints on Relevance. Oxford: Blackwell. Carlson L. 1985. Dialogue Games: An approach to discourse analysis. Dordrecht: Reidel. Haspelmath, Martin. 1995. “The Converb as a Cross-Linguistically Valid Category”. Converbs in cross-linguistic perspective ed. by Martin Haspelmath & Ekkehard König, 1-56. Berlin & New York: Mouton de Gruyter. Haspelmath, Martin 2004. “Coordinating Constructions: An overview”. Coordinating Constructions, Typological Studies in Language ed. by Martin Haspelmath, 3-40. Amsterdam & Philadelphia: John Benjamins. Kalkova, T. & V. I. Podlesskaya. 2001. “The Order of Syntactic Constituents vs. the Order of Discourse Units: The case of Russian adversative constructions”. Item Order: Its variety and linguistic and phonetic consequences ed. by Bohumil Palek & Osamu Fujimura, 377-397. Prague: Karolinum Press. Kitis, Eliza. 2000. “Connectives and frame theory: The case of hypotextual antinomial ‘and’”. Pragmatics and Cognition 8:2.357-409. Lakoff, Robin 1971. “If’s, And’s, and But’s about Conjunction”. Studies in Linguistic Semantics, 114-149. New York: Holt, Reinhart & Williams.

96

ELENA RUDNITSKAYA & ELENA URYSON

Levin, Yu. I. 1970. “Ob odnoj gruppe sojuzov russkogo jazyka” [About one group of Russian conjunctions]”. Mashinnyj perevod i prikladnaja lingvistika [Machine translation and applied linguistics] 13.64-87. Paducheva, E. V. 1997. “Egocentricheskaja semantika sojuzov a i no [Egocentric semantics of Russian conjunctions a and no]”. Slavjanskie sochinitel’nye sojuzy [Slavic coordinative conjunctions] ed. by. T. M. Nikolaeva, V. S. Efimova & I. Fuzheron, 36-47. Moscow: Institut Slavianovedenija i Balkanistiki. Rudnitskaya, Elena. 1998. “Syntactic Properties of the Altaic Coordination Construction in Korean”. STUF 51:2.179-198. Sannikov, V. Z. 1987. Russkie sochinitel’nye konstrukcii [Coordinate constructions in Russian]. Moscow: Izd. “Nauka”. Uryson, E. V. 2000. “Russkij sojuz i chastica i: struktura znachenija [The Russian conjunction and particle i: semantic structure]”. Voprosy jazykoznanija 3.97-121. Uryson, E. V. 2004a. “Sojuzy a i no i figura govorjashchego [The Russian conjunctions a and no and ‘the figure of speaker’]”. Voprosy jazykoznanija 6.64-83. Uryson, E. V. 2004b. “Nekotorye znachenija sojuza a v svete sovremennoj semanticheskoj teorii [Some meanings of the Russian conjunction a in the light of modern semantics theory]”. Russkij jazyk v nauchnom osveshchenii [Russian language in scholarly perspective] 2.64-83. Uryson, E. V. 2006. “Semantika sojuza no: dannye o dejatel’nosti soznanija [The semantics of the conjunction no “but”: data on the functioning of consciousness]”. Voprosy jazykoznanija 5.22-42. Yi, E.-Y. 1994. Adjunction, Coordination and their Theoretical Consequences. Cornell University. [Unpublished Ms.] Yoon, J. H.-S. 1994. “Korean Verbal Inflection and Checking Theory”. Working Papers in Linguistics, ed. by Colin Phillips & Heidi Harley, 251270. Cambridge, MA: Department of Linguistic and Philosophy, MIT.

QUESTION PARTICLES OR WHAT? OPEN ALTERNATIVE QUESTIONS IN UDEGHE

MARIA TOLSKAYA Davis Center for Russian and Eurasian Studies (Harvard University) INNA TOLSKAYA University of Tromsø 1.

Introduction Depending on the logical structure, two main types of alternative questions (Alt-Qs) are traditionally described: (i) A or B, asking for a choice between two distinct propositions, and (ii) A or not A, asking for a choice between two alternatives that differ only in truth value. There seems to be one more type of Alt-Qs, ‘open alternative questions’. In this case the answer does not involve picking one out of two (or more) mutually exclusive and jointly exhaustive values for a variable, as is it happens in (i) and (ii). In the first part of open AltQs the speaker is wondering about a proposition, and the second part, containing an interrogative word, signals uncertainty or incompletion. Typical examples are Are you kidding, or what?; Is the Universe Expanding, Contracting, or What?. Since open Alt-Qs sound less categorical than yes-no questions and A or not A Qs, and leave room for discussion, they have made their way into media news headlines, imitating informal style of speech. Consider the following Russian and German examples from the Internet: Pugajut ili chto? (“Are they frightening or what?’); (4) Leben oder leiden oder was sonst? (“To live, to suffer, or what else?”). However, open Alt-Qs have not received focused attention in the literature, with the exception of the passing example in (Bolinger 1978: 89): Is today the seventeenth, the sixteenth, the eighteenth, or what? In a more generalized form, the interpretation of open Alt-Q could be represented as follows: Is proposition A true; and if not, then what subset of not-A is true? Then open Alt-Qs are naturally analyzed as coordination of two underlying questions: one, corresponding to a polar (yes-no) question and the other one, corresponding to a reduced wh-Q. Open Alt-Qs cannot be answered by choosing one of two distinct propositions, as expected for other types of alternative Qs. The two underlying questions taken separately may demand two different replies: ‘yes’ or ‘no’ for the polar question, and a portion of relevant

98

MARIA TOLSKAYA & INNA TOLSKAYA

information for the wh-Q. Thus, despite not being a recognized question type, open Alt-Qs undoubtedly form a separate class. They seem a widespread property of world languages, although there is a lot of cross-linguistic variation in their formal structure. Remarkably, in Tungusic languages one of the disjuncts can be expressed by a pro-verb, as illustrated by the examples below: (1) Si min-tigi eme-zeNe-i you me-LAT come-FUT-2SG “Will you come to me or what?”

ja-zaNa-i?1 PROV-FUT-2SG

(2) Negidal KoskoN-dula˘-s eme-ča˘-v bu˘-ze˘-s=ku e˘-za-s=ku? cat-LOC-2SG come-PAST-1SG give-FUT-2SG=DIS PROV-FUT-2SG=DIS “I came to take your cat; will you give it (or not)?” (Tsintsius 1982: 85) (3) Oroch Agduči-za-i=nu, ja˘-za-i=nu? tell-FUT-1SG=DIS PROV-FUT-1SG=DIS “Should I tell or what? (Avrorin 2001: 40) (4) Oroqen TarI g´n´-r´-n yE:-ra-n that go-AOR-3SG PROV-AOR-3SG “Is he going or not (lit.: or what)?” (Whaley 2005) We will examine several possibilities of the morphological and pragmatic analyses of open Alt-Qs in Tungusic languages. Section 1 of this paper outlines the basic information about the Tungusic languages and sources of the data. The second section is for the most part descriptive: it presents the two main strategies of forming Alt-Qs in Udeghe. Section 3 approaches the problem from the information structure perspective, comparing intonation of various types of Alt-Qs and discussing their conventionalization. Section 4 examines open Alt-Qs, from the speech production viewpoint. Section 5 addresses this construction as an example of possible grammaticalization in progress. Section 6 contains a summary and conclusions. 2.

Tungusic languages Tungusic (Manchu-Tungus) languages, considered to form a branch of Altaic, are spoken in Siberia, North China and Mongolia. They are generally subdivided into three groups: the Northern group (Evenki, Even, Oroqen, 1

The language and the source are indicated for each example, except where the language is Udeghe and the example from our materials. Glosses are ours, except for Evenki and Oroqen.

OPEN ALTERNATIVE QUESTIONS IN UDIHE

99

Solon and Negidal), the Southern (Nanai, Orok, Ulcha, Oroch and Udeghe) and the Manchu group (Manchu, Xibo (Sibe) and the extinct Jurchen). The dividing line between these groups of Tungusic languages cannot be drawn sharply. The main techniques of inter-clausal coherence are parataxis (juxtaposition without overt linkage), and hypotaxis (the use of participial and converbial clauses). The use of sentence-final particles, head-finality and presence of postpositions are the most striking SOV features of Tungusic languages. An essential property of these languages is the fully conjugated negative auxiliary. Another noteworthy feature is the inflected pro-verb, derived from the same root as interrogative pronouns. According to published descriptions, the proverb is attested in Udeghe, Oroch, and Oroqen (je-/ja-yE- respectively); Nanai, Orok and Ulch (xai-); Sibe (ana-); and in Evenki and Negidal (e:-). We have no information on pro-verbs in other Tungusic languages. This paper is based mainly on the data from Udeghe; other Tungusic languages are used occasionally for comparison. Data comes from a variety of published sources, as indicated following each example. Examples not identified by source come from our own field materials2. 2.1 Udeghe Udeghe (Udihe) is spoken in the Maritime and Khabarovsk districts of the Russian Far East. It belongs to the southern branch of Tungusic languages and has been affected by Chinese and Russian. Udeghe is only used by the older generation in everyday oral communication. Fewer than 100 speakers remain. Udeghe, like Tungusic languages in general, is an agglutinating language, using suffixation. Pronominal arguments (subjects, objects and possessors) are easily omitted, when they are predictable due to agreement affixes or context. The word order is largely dictated by the information structure, but, crucially, the focused element must be preverbal. In Udeghe, the negative auxiliary verb inflects for voice, mood, aspect, tense, and person/number, and is immediately followed by the unchangeable present stem of the matrix (content) verb, which can be deleted under ellipsis, as in A or not (A) Alt-Qs: (5) Guline-ze-fi=es e-ze-fi=es [guline]? leave-SUBJ-1PL.IN=DIS NEG-SUBJ-1PL.IN=DIS leave “Shall we leave or not?”

2

The Udeghe material was collected during fieldwork in 1989–2006, supported by the Cultural Initiative Foundation sponsored by George Soros and by the Endangered Languages Fund.

100

MARIA TOLSKAYA & INNA TOLSKAYA

3. Alternative questions 3.1 The pro-verb jeThough typologically rare, pro-verbs occur in Tungusic and other languages of Altaic family, such as Buryat, Tuvan, and Altai. The Udeghe proverb is derived from the root je-/ja-/i- that crosscuts the boundaries of several word classes:

j’eu je-we je-du je-le je-digi je-li je-uxi j’e-fe jeu-ze je-ze-Nezi je-me ja-Nnaja-mu˘ije-ne-mi

(NOM) (ACC) (DAT) (LOC) (ABL) (PROL) <what-LAT.ADV> <what-ACC.ADV> <what-N> <what-N-ABL.ADV> <what-ADJ>

Simultaneity in Signed Languages: Form and Function (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Saami Linguistics (Amsterdam Studies in the Theory and History of Linguistic Science Series IV: Current Issues in Linguistics Theory)

Emotion in Dialogic Interaction: Advances in the Complex (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Linguistics and Philosophy: Essays in Honor of Rulon S. Wells (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Grammar from the Human Perspective: Case, Space and Person in Finnish (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Proto-Japanese; Issues and Prospects (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Verb Constructions in German and Dutch (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Phonetics and Phonology: Interactions and interrelations (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Morphological Analysis in Comparison (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Papers in theoretical linguistics (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Romance Languages and Linguistic Theory 2004: Selected Papers from 'Going Romance', Leiden, 911 December 2004 (Amsterdam Studies in the Theory and History ... IV: Current Issues in Linguistic Theory)

The Laws of Indo-European (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

The Paradox of Grammatical Change: Perspectives from Romance (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, CILT 293)

Language Typology: A Functional Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Typological Models in Reconstruction (Amsterdam Studies in the Theory & History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Postvelar Harmony (Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Simultaneity in Signed Languages: Form and Function (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Saami Linguistics (Amsterdam Studies in the Theory and History of Linguistic Science Series IV: Current Issues in Linguistics Theory)

Emotion in Dialogic Interaction: Advances in the Complex (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Linguistics and Philosophy: Essays in Honor of Rulon S. Wells (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Grammar from the Human Perspective: Case, Space and Person in Finnish (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Proto-Japanese; Issues and Prospects (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Verb Constructions in German and Dutch (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Phonetics and Phonology: Interactions and interrelations (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Morphological Analysis in Comparison (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Papers in theoretical linguistics (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Romance Languages and Linguistic Theory 2004: Selected Papers from 'Going Romance', Leiden, 911 December 2004 (Amsterdam Studies in the Theory and History ... IV: Current Issues in Linguistic Theory)

The Laws of Indo-European (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

The Paradox of Grammatical Change: Perspectives from Romance (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, CILT 293)

Language Typology: A Functional Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Language Typology: A Functional Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Typological Models in Reconstruction (Amsterdam Studies in the Theory & History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Postvelar Harmony (Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Loan Phonology (Current Issues in Linguistic Theory)

Anaphora Processing (Current Issues in Linguistic Theory)

Loan Phonology (Current Issues in Linguistic Theory)

Loan Phonology (Current Issues in Linguistic Theory)

Connectives in the History of English (Amsterdam Studies in the Theory and History of Linguistic Science)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

The Transformational-Generative Paradigm and Modern Linguistic Theory (Current Issues in Linguistic Theory)

Germanic linguistics: syntactic and diachronic (Current Issues in Linguistic Theory)

Subordination and Coordination Strategies in North Asian Languages (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Subordination and Coordination Strategies in North Asian Languages (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Simultaneity in Signed Languages: Form and Function (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Saami Linguistics (Amsterdam Studies in the Theory and History of Linguistic Science Series IV: Current Issues in Linguistics Theory)

Emotion in Dialogic Interaction: Advances in the Complex (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Linguistics and Philosophy: Essays in Honor of Rulon S. Wells (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Grammar from the Human Perspective: Case, Space and Person in Finnish (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Proto-Japanese; Issues and Prospects (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Verb Constructions in German and Dutch (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Phonetics and Phonology: Interactions and interrelations (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Morphological Analysis in Comparison (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Papers in theoretical linguistics (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Romance Languages and Linguistic Theory 2004: Selected Papers from 'Going Romance', Leiden, 911 December 2004 (Amsterdam Studies in the Theory and History ... IV: Current Issues in Linguistic Theory)

The Laws of Indo-European (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

The Paradox of Grammatical Change: Perspectives from Romance (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, CILT 293)

Language Typology: A Functional Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Language Typology: A Functional Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

Typological Models in Reconstruction (Amsterdam Studies in the Theory & History of Linguistic Science, Series IV: Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Postvelar Harmony (Current Issues in Linguistic Theory)

English Adjective Comparison: A Historical Perspective (Amsterdam Studies in the Theory and History of Linguistic Science, Series IV: Current Issues in Linguistic Theory, Volume 299)

Loan Phonology (Current Issues in Linguistic Theory)

Anaphora Processing (Current Issues in Linguistic Theory)

Loan Phonology (Current Issues in Linguistic Theory)

Loan Phonology (Current Issues in Linguistic Theory)

Connectives in the History of English (Amsterdam Studies in the Theory and History of Linguistic Science)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

Subordination (Oxford Studies in Typology and Linguistic Theory)

The Transformational-Generative Paradigm and Modern Linguistic Theory (Current Issues in Linguistic Theory)

Germanic linguistics: syntactic and diachronic (Current Issues in Linguistic Theory)

Recommend Documents