The Oxford Guide to the Transeurasian Languages (Oxford Guides to the World's Languages) 0198804628, 9780198804628

Citation preview

OUP UNCORRECTED PROOF – , 26/05/20, SPi

THE OXFOR D GUIDE TO THE

TR A NSEUR A SI A N L A NGUAGES

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Oxfor d Guides to the Wor ld’s L a nguages

general editors

Adam Ledgeway, University of Cambridge, and Martin Maiden, University of Oxford

advisory editors

Alexandra Y. Aikhenvald  James Cook University Edith Aldridge  University of Washington Stephen R. Anderson  Yale University Bernard Comrie  University of California, Santa Barbara Jan Terje Faarlund  University of Oslo Alice Harris  University of Massachusetts, Amherst Bernd Heine  University of Cologne

Paul Hopper  Carnegie-Mellon University Geoffrey Khan  University of Cambridge Lutz Marten  SOAS, London Marianne Mithun  University of California, Santa Barbara Irina Nikolaeva  SOAS, London Chris Reintges  CNRS, Paris Masayoshi Shibatani  Rice University David Willis  University of Cambridge

published

The Oxford Guide to the Romance Languages Edited by Adam Ledgeway and Martin Maiden The Oxford Guide to the Transeurasian Languages Edited by Martine Robbeets and Alexander Savelyev

in preparation

The Oxford Guide to the Austronesian Languages: Malayo-Polynesian Languages of Asia and Madagascar Edited by Alexander Adelaar and Antoinette Schapper The Oxford Guide to the Uralic Languages Edited by Marianne Bakró-Nagy, Johanna Laakso, and Elena Skribnik The Oxford Guide to the Afroasiatic Languages Edited by Sabrina Bendjaballah and Chris Reintges The Oxford Guide to the Languages of Australia Edited by Claire Bowern The Oxford Guide to the Papuan Languages Edited by Nicholas Evans and Sebastian Fedden The Oxford Guide to the Slavonic Languages Edited by Jan Fellerer and Neil Bermel The Oxford Guide to the Bantu Languages Edited by Ellen Hurst, Nancy Kula, Lutz Marten, and Jochen Zeller The Oxford Guide to the Atlantic Languages of West Africa Edited by Friederike Lüpke The Oxford Guide to the Languages of the Central Andes Edited by Matthias Urban

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

the oxfor d guide to the

Transeurasian Languages EDITED BY

Martine Robbeets & Alexander Savelyev

1

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

1 Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © editorial matter and organization Martine Robbeets 2020 © the chapters their several authors 2020 The moral rights of the authors have been asserted First Edition published in 2020 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2019948973 ISBN 978–0–19–880462–8 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Short Contents Detailed Contents  Series Preface  Abbreviations  The Contributors  Romanization Conventions  Introduction  M a r t i n e R o bbe e t s

and

ix xxxvi xxxvii xliv lii 1

A l e x a n d e r S av e ly e v

5

PART I:  Sources and Classification 

A.  Historical Sources and Periodization 1. Historical sources and periodization of the Japonic and Koreanic languages  M a rc M i ya k e 2. The Altaic languages: Tungusic, Mongolic, Turkic  Vo l k e r R y b at z k i

9 22

B.  Genealogical Classification  3. The classification of the Transeurasian languages  M a r t i n e R o bbe e t s

31

4. The classification of the Japonic languages  E l i s a be t h d e B o e r

40

5. The classification of the Korean language and its dialects  K you -D o n g A h n a n d J a e h o o n Ye o n

59

6. The classification of the Tungusic languages  L i n d s ay  J. W h a l e y a n d S o f i a O s ko l s k aya

81

7. The classification of the Mongolic languages  H a n s N ug t e r e n

92

8. The classification of the Turkic languages  L a r s Joh a nson

105

9. A Bayesian approach to the classification of the Turkic languages  A l e x a n d e r S av e ly e v

115

C. Typology  10. The typological heritage of the Transeurasian languages  M a r t i n e R o bbe e t s 11. Typological profile of the Transeurasian languages from a quantitative perspective  N ata l i i a H ü bl e r

127

145

v

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Short Contents PART II:  Individual Structural Overviews 

161

12. Japanese and the mainland dialects  M a s ayo s h i S h i b ata n i

163

13. Amami and Okinawa, the Northern Ryukyuan languages  Yu t o N i i n ag a

196

14. Miyako, Ishigaki, and Yonaguni, the Southern Ryūkyūan languages  J o h n  R. B e n t l e y

225

15. Korean and the Korean dialects  H o -M i n S o h n

241

16. Jejudo Korean  Ub ong Sh i n, Ji e u n K i a e r ,

258

and

J i you n g S h i n

17. Xibe and the Manchuric languages  Ta e h o J a n g

269

18. Even and the Northern Tungusic languages  B r ig i t t e Pa k e n d o r f a n d N ata l i a A r a l ova

288

19. Nanai and the Southern Tungusic languages  S o f i a O s ko l s k aya

305

20. Dagur  Yo h e i Ya m a da

321

21. Khalkha Mongolian  J a n -O l o f S va n t e s s o n

334

22. Oirat and Kalmyk, the Western Mongolic languages  Á g n e s B i r ta l a n

350

23. The Northwestern Turkic (Kipchak) languages  É va Á. C s at ó a n d L a r s J o h a n s o n

370

24. Turkish and the Southwestern Turkic (Oghuz) languages  J a k l i n K o r n f i lt

392

25. Uyghur and Uzbek, the Southeastern Turkic languages  A bdu r i s h i d Ya k u p

411

26. Sakha and Dolgan, the North Siberian Turkic languages  B r ig i t t e Pa k e n d o r f a n d E ug é n i e S ta p e r t

430

27. Chuvash and the Bulgharic languages  A l e x a n d e r S av e ly e v

446

PART III:  Comparative Overviews 

465

A. Phonology  28.  A comparative approach to the consonant inventory of the Transeurasian languages  A l l a n  R. B om h a r d

vi

Dictionary:

469

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Short Contents 29. A comparative approach to the vowel systems and harmonies in the Transeurasian languages and beyond  A n d r e w J o s e p h , S e o n g y e o n K o, a n d J o h n Wh i t m a n

486

B. Morphology  30. A comparative approach to verbal morphology in Transeurasian  M a r t i n e R o bbe e t s 31. A comparative approach to nominal morphology in Transeurasian: Case and plurality  I lya G ru n t ov a n d O l g a M a z o 32. A comparative approach to the pronominal system in Transeurasian  M i c h a l S c h wa r z , O n d ř e j S r b a , a n d Vác l av B l a ž e k

511

522 554

C. Syntax  33. The nominal group, possessive agreement, and nominal sentences in the Transeurasian languages  I r i n a N e vs k aya a n d L i n a A m a l

587

34. Verbal categories in the Transeurasian languages  A n d r e j M a l c h u kov a n d Pat ry k C z e rw i n s k i

604

35. Complex constructions in the Transeurasian languages  A n d r e j M a l c h u kov a n d Pat ry k C z e rw i n s k i

625

D.  Lexicon and Semantics  36. Basic vocabulary in the Transeurasian languages  M a r t i n e R o bbe e t s

645

37. Numerals in the Transeurasian languages  Vác l av B l a ž e k

660

38. Kinship-term paradigms in the Transeurasian languages  M i l a n va n B e r l o

691

PART IV:  Areal versus Inherited Connections 

703

39. Contact between genealogically related languages: The case of Old Korean and Old Japanese  A l e x a n d e r  T. F r a n c i s -R at t e a n d J. M a r s h a l l U n g e r

705

40. Form and pattern borrowing across Siberian Turkic, Mongolic, and Tungusic languages  G r e g o ry  D. S. A n d e r s o n

715

41. Transeurasian as a continuum of diffusion  E dwa r d Vaj da

726

42. Beck-Wichmann-Brown evaluation of lexical comparisons for the Transeurasian proposal  C e c i l  H. B row n

735

vii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Short Contents PART V: Interdisciplinary Perspectives on the Identity of Transeurasian 

751

43. The homelands of the individual Transeurasian proto-languages  M a r t i n e R o bbe e t s , J u h a J a n h u n e n , A l e x a n d e r S av e ly e v, a n d E vg e n i ya K o rov i n a

753

44. The Transeurasian homeland: Where, what, and when?  M a r t i n e R o bbe e t s

772

45. Transeurasian unity from a population-genetic perspective  C h o o n g wo n J e o n g , C h ua n -C h ao Wa n g , a n d C h ao N i n g

784

46. Transeurasian unity from an archaeological perspective  T ao L i

794

47. Language dispersals and the “Secondary Peoples’ Revolution”: A historical anthropology of the Transeurasian unity  M a r k Ja m e s Hu ds on

806

References  Index 

815 881

viii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents Series Preface  Abbreviations  The Contributors  Romanization Conventions  Introduction  M a r t i n e R o bbe e t s

and

xxxvi xxxvii xliv lii 1

A l e x a n d e r S av e ly e v

5

PART I:  Sources and Classification 

A.  Historical Sources and Periodization  1.  Historical sources and periodization of the Japonic and Koreanic languages  9 M a rc M i ya k e 1.1 Introduction  9 1.1.1 Transcriptions, sources, periodization 9 1.1.2 Sources 10 1.1.3 Periodization 10 1.2 Japonic languages  10 1.2.1 Peninsular Japonic 10 1.2.2 Mainland Japonic 11 1.2.2.1 Pre-Old Japanese (c. 200–late 600s) 11 1.2.2.2 Old Japanese (jōdai nihongo; c. late 600s–800) 12 1.2.2.3 Early Middle Japanese (chūko nihongo; c. 800–1200) 13 1.2.2.4 Late Middle Japanese (chūsei nihongo; c. 1200–1600) 14 1.2.2.5 Early Modern Japanese (kinsei nihongo; c. 1600–late 1800s) 15 1.2.2.6 Contemporary Japanese (gendai nihongo; late 1800s–present) 16 1.2.3 Ryūkyūan languages 16 1.2.3.1 Old Okinawan (late 1500s–early 1700s) 16 1.2.3.2 Middle Okinawan (mid-1700s–late 1800s) 17

1.2.3.3 Modern Okinawan and other Ryūkyūan languages (late 1800s–present) 17

1.3 Koreanic languages 

17

1.3.1 Early Koreanic (c. 200–600) 17 1.3.2 Old Korean (kotay hankwuke; late 500s–mid-900s) 18 1.3.3 Early Middle Korean (EMK; choki cwungsey hankwuke; late 900s–c. 1400) 19 1.3.4 Late Middle Korean (huki cwungsey hankwuke; c. 1400–1600) 20 1.3.5 Early Modern Korean (kuntay hankwuke; c. 1600–late 1800s) 21 1.3.6 Contemporary Korean (hyentay hankwuke; late 1800s–present) 21

2.  The Altaic languages: Tungusic, Mongolic, Turkic  22 Vo l k e r R y b at z k i 2.1 Introduction  22 2.2 Tungusic languages  22 2.2.1 Distribution and number of speakers 22 2.2.2 Scripts and sources 23

ix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents 2.3 Mongolic languages 

24

2.4 Turkic languages 

26



2.3.1 Distribution and number of speakers 24 2.3.2 Scripts and sources 24



2.4.1 Distribution and number of speakers 26 2.4.2 Scripts and sources 27

B.  Genealogical Classification  3.  The classification of the Transeurasian languages  31 M a r t i n e R o bbe e t s 3.1 Introduction  31 3.2 Previous classifications  32 3.3 Classification on the basis of the historical-comparative linguistic method  34 3.3.1 Shared innovations in Japano-Koreanic 34 3.3.2 Shared innovations in Altaic 35 3.3.3 Shared innovations between Turkic and Mongolic 35 3.3.4 Shared innovations between Mongolic and Tungusic 35 3.3.5 Shared innovations between Turkic and Tungusic 35 3.3.6 Shared innovations in Turkic 36 3.3.7 Shared innovations in Tungusic 36 3.3.8 Resulting classification 36 3.4 Classification on the basis of Bayesian inference  37 3.5 Conclusion  37 4.  The classification of the Japonic languages  40 E l i s a be t h   d e B o e r 4.1 Classification of Japonic on the basis of the historical comparative method  40 4.1.1 Introduction 40 4.1.1.1 Geographical distribution of the Japonic language family 40 4.1.1.2 Methodology of the classification 40 4.1.1.3 Areal connections that complicate the classification 40

4.1.2 History of dialect research and main classifications proposed in the previous literature 41 4.1.2.1 The classification of Mainland Japanese 42 4.1.2.1.1 A division into dialect areas 42 4.1.2.1.2 A division into “front of Japan” and “back of Japan” dialects 44 4.1.2.1.3 A division of the dialects in concentric circles 44 4.1.2.2 The classification of Ryūkyūan 45 4.1.2.3 The standard classification of Japanese and Ryūkyūan 46 4.1.3 Segmental phonology 46 4.1.4 Suprasegmental phonology 49 4.1.4.1 Tone classes 49 4.1.4.2 The tone systems of mainland Japan 50 4.1.4.3 The tone systems of the Ryūkyūs 51 4.1.4.4 Tonal features 51 4.1.5 Morphology 53 4.1.5.1 Morphological features that go back to the difference between Eastern Old Japanese and Central Old Japanese53 4.1.5.2 Variations in morphology of more recent date 54

x

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

4.1.6 Lexicon 55 4.1.6.1 Shared lexicon in Kyūshū and the Ryūkyūs 55 4.1.7 Tentative classification 55

5.  The classification of the Korean language and its dialects  K you -D o n g A h n a n d J a e h o o n Ye o n 5.1 Introduction  5.2 Seven dialectal zones in Korean  5.3 Major diagnostic criteria for the dialectal division of Korean 

59

59 59 60 5.3.1 Palatalization: t-palatalization, k-palatalization, and h-palatalization 60 5.3.2 Umlaut 61 5.3.3 Relics of Middle Korean 61 5.3.4 Tone and vowel length 61 5.3.5 Segment insertion and alternation 62 5.4 Characteristics of seven major dialects in Korean  62 5.4.1 Hamgyŏng dialect 62 5.4.1.1 Phonological features 62 5.4.1.2 Grammatical features 63 5.4.1.3 Lexical features 64 5.4.2 P’yŏngan dialect 65 5.4.2.1 Phonological features 65 5.4.2.2 Grammatical features 65 5.4.2.3 Lexical features 65 5.4.3 Central dialects 66 5.4.3.1 Phonological features 66 5.4.3.2 Grammatical features 67 5.4.3.3 Lexical features 67 5.4.4 Ch’ungch’ŏng dialect 68 5.4.4.1 Phonological features 68 5.4.4.2 Grammatical features 69 5.4.4.3 Lexical features 69 5.4.5 Kyŏngsang dialects 70 5.4.5.1 Phonological features 70 5.4.5.2 Grammatical features 71 5.4.5.3 Lexical features 71 5.4.6 Chŏlla dialect 71 5.4.6.1 Phonological features 71 5.4.6.2 Grammatical features 72 5.4.6.3 Lexical features 72 5.4.7 Jeju dialects 73 5.4.7.1 Phonological features 73 5.4.7.2 Grammatical features 74 5.4.7.3 Lexical idiosyncrasies 75 5.5 The genealogical relationship among the Korean dialects  75 5.5.1 Previous analyses of the relationship among the Korean dialects 75 5.5.2 Shared innovations and groupings of the Korean dialects 78 5.5.2.1 First dialectal grouping: The Jeju dialect 78 5.5.2.2 Second dialectal grouping: The Central dialects 79 5.5.2.3 Third dialectal grouping: The Ch’ungch’ŏng dialect 79 5.5.2.4 Fourth dialectal grouping: The Kyŏngsang and Chŏlla dialects 79 5.5.2.5 Fifth dialectal grouping: The Hamgyŏng and P’yŏngan dialects 80 5.6 Summary  80

xi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents 6.  The classification of the Tungusic languages  81 L i n d s ay  J. W h a l e y a n d S o f i a O s ko l s k aya 6.1 Introduction  81 6.2 Approaches to Tungusic classification  81 6.3 Proto-Tungusic segmental inventories  84 6.4 A binary-branching classification of the Tungusic languages  84 6.4.1 *t- 84 6.4.2 *-b- 85 6.4.3 *-k- 85 6.4.4 *-g- 85 6.4.5 *p- 86 6.4.6 *-i/-u 86 6.4.7 *ü1 86 6.4.8 *u2 86 6.4.9 Summary 87 6.5 Interpretation  87 6.6 A Bayesian phylogenetic approach to Tungusic classification  88 6.6.1 The Bayesian method 88 6.6.2 Tungusic data 88 6.6.3 Analysis 89 6.7 Discussion and summary  91 7.  The classification of the Mongolic languages  92 H a n s N ug t e r e n 7.1 Introduction  92 7.2 Challenges  93 7.3 Phonetic tendencies  94 7.3.1 Irregular phonetic developments 97 7.4 Morphology  100 7.5 Lexicon  103 7.6 Semantics  103 7.7 Closing remarks  104 8.  The classification of the Turkic languages  L a r s Joh a nson 8.1 Introduction  8.2 Areas of distribution  8.3 “Languages” and “dialects”  8.4 Proto-Turkic  8.5 Divergence  8.6 Convergence  8.7 Genealogical relatedness  8.8 Contact-induced “take-over” copying  8.9 “Carry-over” copying  8.10 Proposed classifications  8.11 Branches  8.12 Problematic cases  8.13 Classificatory criteria  8.14 Further main issues  8.15 Further phonetic features  8.16 Lexical features  8.17 Morphological features 

xii

Dictionary:

105 105 105 105 106 106 106 106 107 107 108 108 109 110 110 111 112 113

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

8.18 Case markers  8.19 Other morphological features 

113 114

9.  A Bayesian approach to the classification of the Turkic languages  115 A l e x a n d e r S av e ly e v 9.1 Introduction  115 9.2 Data and method  116 9.2.1 Languages 116 9.2.2 Basic vocabulary list 117 9.2.3 Sources and data selection 117 9.2.4 Cognate coding 118 9.2.5 Bayesian analysis 118 9.3 Results  118 9.4 Discussion  119 9.4.1 Topology 119 9.4.2 Time depth 121 9.5 Conclusion  122

C. Typology  10.  The typological heritage of the Transeurasian languages  127 M a r t i n e R o bbe e t s 10.1 Introduction  127 10.2 Data set and methods  128 10.3 Typological profile of the Transeurasian languages  129 10.3.1 Phonology 129 10.3.1.1 F1 Presence of tongue-root vowel harmony 129 10.3.1.2 F2 Absence of r- in initial position in native words 130 10.3.1.3 F3 Absence of velar nasal restricted to initial position 130 10.3.1.4 F4 Presence of voicing distinction for stops 131 10.3.2 Lexicon and semantics 131

10.3.2.1 F5 Preference for non-verbal strategy for (extra-family) verbal copies 131 10.3.2.2 F6 Presence of a two-way proximal–distal distinction in demonstrative pronouns 131 10.3.2.3 F7 Property words are verbally and nominally encoded such that some property words exhibit switched encoding 132 10.3.3 Morphology 132 10.3.3.1 F8 Inflectional morphology is predominantly suffixing 132 10.3.3.2 F9 The imperative is expressed by a bare verb stem 133 10.3.3.3 F10 Absence of obligatory numeral classifiers 133 10.3.3.4 F11 Presence of mi-Ti opposition in first vs. second singular personal pronouns 134 10.3.3.5 F12 Formation of a secondary nasal oblique stem in personal pronouns 134 10.3.4 Syntax 135 10.3.4.1 F13 Dependent-marking of clause arguments 135 10.3.4.2 F14 Dependent-marking in possessive noun phrases 135 10.3.4.3 F15 Extensive use of converb 136 10.3.4.4 F16 Use of locative existential constructions to encode predicative possession 136 10.3.4.5 F17 Use of the ablative case form to encode predicative comparison 137

xiii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents



10.3.5 Grammaticalization patterns 137 10.3.5.1 F18 Direct insubordination 137 10.3.5.2 F19 Grammaticalization from negative verb to verbal negator over a construction comprising an inflected negative auxiliary and an invariant lexical verb 138 10.3.5.3 F20 Grammaticalization of plural/collective markers to express inclusive/exclusive distinction on first-person pronoun 139 10.3.6 Overview 140



10.4.1 Typological features specific to Transeurasian 140 10.4.2 Structural heritage vs. linguistic area 142







10.4 Discussion 

140



10.5 Conclusion 

143

11. Typological profile of the Transeurasian languages from a quantitative perspective  145 N ata l i i a H ü bl e r 11.1 Introduction  145 11.2 Language sample  146 11.3 Typological overview of the Transeurasian languages  146 11.3.1 Phonology 146 11.3.1.1 Vowels 146 11.3.1.2 Positional constraints 147 11.3.1.3 Phoneme inventories 147 11.3.2 Agglutination and position of bound morphemes 147 11.3.3 Noun 147 11.3.4 Pronoun 148 11.3.5 Demonstrative 149 11.3.6 Article 149 11.3.7 Adjective 149 11.3.8 Numeral system 150 11.3.9 Verb 150 11.3.9.1 TAME 150 11.3.9.2 Valency-changing operations 150 11.3.9.3 Verb morphology in subordinate clauses 151 11.3.9.4 Reduplication 151 11.3.10 Attributive possession 151 11.3.11 Predicative possession 151 11.3.12 Alignment 152 11.3.13 Negation 152 11.3.14 Word order 153 11.3.15 Interrogation 153 11.3.16 Comparative construction 153 11.3.17 Coordination and conjunction 153 11.3.18 Obligatoriness of S/A argument 154 11.3.19 Derivation of adpositions 154 11.3.20 Classifiers 154 11.4 Phylogenetic analysis  154 11.4.1 Coding procedure 154 11.4.2 Stability of structural features 155



11.4.3 Bayesian approach to the classification of the Transeurasian languages 156

11.5 Conclusion 

xiv

Dictionary:

158

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents PART II:  Individual Structural Overviews 

161

12.  Japanese and the mainland dialects  M a s ayo s h i S h i b ata n i 12.1 Introduction  12.2 Mainland dialects 

163











163 164

12.2.1 Old Japanese, Eastern Old Japanese, and modern mainland Japanese dialects 164

12.3 Standard Japanese  165 12.3.1 Word order 165 12.3.2 Subject and topic 166 12.3.3 Nominal structures 167 12.3.3.1 Particles 167 12.3.3.2 Nominalization 169 12.3.4 Verbal structures 172 12.3.4.1 Conclusive and adnominal forms 172 12.3.4.2 Auxiliaries and helping verbs 173 12.3.4.3 Converbs 174 12.3.4.4 Voice 174 12.3.4.5 Insubordination 175 12.4 Dialect features  177 12.4.1 Particles 177 12.4.1.1 Dialects lacking nominative and accusative markers 177 12.4.1.2 Dative lumpers and splitters 177 12.4.1.3 Genitive markers178 12.4.1.4 Subject markers 179 12.4.1.5 Particle coalescence 179 12.4.1.6 Nominalization particle 180 12.4.2 Conclusive and adnominal forms 183 12.4.2.1 Conclusive and adnominal merger 183 12.4.2.2 Restoration of the conclusive/adnominal contrast 184 12.5 Hachijō  184 12.5.1 Hachijō and Ryūkyūan 185 12.5.2 Hachijō and Eastern Old Japanese 186 12.5.3 Hachijō, Central Old Japanese, and Ryūkyūan 188 12.5.3.1 Genitive markers 188 12.5.3.2 Subject markers 189 12.5.3.3 Kakari-musubi 190 12.5.3.4 Grammaticalization of existential verbs 191 12.5.3.5 Voice 193 12.5.3.6 Insubordination 194 12.6 Conclusion  195

13.  Amami and Okinawa, the Northern Ryukyuan languages  196 Yu t o N i i n ag a 13.1 Introduction  196 13.2 Historical connections: Genealogy and contact  197 13.3 Phonology  198 13.3.1 Consonants 198 13.3.2 Vowels 201 13.3.3 Syllable structure 202

xv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents







13.3.4 Phonological rules 202 13.3.4.1 Sonorant deletion in Yuwan (Amami) 203 13.3.4.2 Epenthetic vowel in Yuwan (Amami) and Shuri (Okinawa) 203 13.3.4.3 Vowel shortening in Yuwan (Amami) 204 13.3.4.4 Non-existence of vowel harmony in Yuwan (Amami) and Shuri (Okinawa) 204 13.3.5 Morphophonological rule 204 13.3.5.1 Sequential voicing (Rendaku) 204 13.3.5.2 Fusion of topic marker 204 13.3.6 Suprasegmentals 205

13.4 Word morphology and phrasal constituents  206 13.4.1 Pronouns 206 13.4.1.1 Personal pronouns 206 13.4.1.2 Demonstrative pronouns 207 13.4.1.3 Interrogative pronouns 208 13.4.2 Nominals 209 13.4.2.1 Common nouns 210 13.4.2.2 Address nouns 210 13.4.2.3 Numerals 211 13.4.2.4 Diminutive affix 211 13.4.2.5 Plural markers and the animacy hierarchy 211 13.4.2.6 Case markers and the animacy hierarchy 212 13.4.3 Adnominals 216 13.4.4 Verbs 216 13.4.4.1 Structure of verbs and derivational affixes 216 13.4.4.2 Inflectional affixes of verbs 219 13.4.5 Adjectives 222 13.4.6 Compounds 223 13.5 Clauses  224 13.6 Lexicon  224

14.  Miyako, Ishigaki, and Yonaguni, the Southern Ryūkyūan languages  225 J o h n  R. B e n t l e y 14.1 Introduction  225 14.2 Historical connections  226 14.3 Phonology  228 14.3.1 Consonants 228 14.3.2 Vowels 228 14.3.3 Syllable structure 229 14.3.4 Morphophonology 229 14.3.5 Suprasegmentals 229 14.4 Morphology  230 14.4.1 Nouns 230 14.4.2 Pronouns 231 14.4.3 Numerals and classifiers 231 14.4.4 Verbal adjectives 232 14.4.5 Verbs 233 14.4.5.1 Negation 233 14.4.5.2 Aspect/tense 233 14.4.5.3 Mood 234 14.4.5.4 Voice 234 14.4.5.5 Predication 235

xvi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

14.5 Syntax 

235



14.6 Lexicon  14.7 Conclusion 

239 240



14.5.1 The clause 235 14.5.2 The nominal group 236 14.5.3 The predicate 237

15.  Korean and the Korean dialects  241 H o -M i n S o h n 15.1 Introduction  241 15.1.1 Speakers 241 15.1.2 Writing systems 241 15.1.2.1 Chinese characters 241 15.1.2.2 Ch’acha (借字) ‘loan-character writing’ 242 15.1.2.3 The Korean alphabet, han’gŭl 242 15.1.2.4 Romanization 243 15.1.3 Previous scholarship on Korean 243 15.2 Historical connections  243 15.2.1 Korean-Japanese hypothesis 243 15.2.2 The Altaic hypothesis 243 15.2.3 Dialectal situation and standardization 244 15.2.4 Language contact 244 15.2.4.1 Contact with Chinese 244 15.2.4.2 Contact with Japanese 244 15.2.4.3 Contact with English and other Western languages 245 15.2.4.4 Structural impact 245 15.3 Phonology  245 15.3.1 Consonants 245 15.3.2 Vowels and semivowels 246 15.3.3 Syllable structure 246 15.3.4 Morphophonology 246 15.3.5 Suprasegmentals 247 15.3.5.1 Vowel length (:) 247 15.3.5.2 Non-phonemic lexical stress 247 15.3.5.3 Lexical pitches 247 15.3.5.4 Intonation 247 15.4 Morphology  247 15.4.1 Classification of lexical items 247 15.4.1.1 Nominals 247 15.4.1.2 Predicates 247 15.4.1.3 Modifiers 248 15.4.1.4 Particles 248 15.4.1.5 Affixes 248 15.4.2 Inflectional morphology 248 15.4.2.1 Inflection of nominals 248 15.4.2.2 Inflection of predicates 249 15.4.3 Derivational morphology 250 15.4.3.1 Affixation 250 15.4.3.2 Sound gradation 250 15.4.3.3 Compounding 250 15.5 Syntax  251 15.5.1 Word order 251

xvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





15.5.2 15.5.3 15.5.4 15.5.5

Complex sentences 251 Noun phrases 252 Verb phrases 252 Honorifics 253

15.6 Features of Korean vocabulary  253 15.7 Dialects  254 15.7.1 Dialectal division 254 15.7.2 Causes for the formation of dialectal zones 254 15.7.3 Dialectal features 255 15.7.4 Inflectional patterns 256 15.7.5 Linguistic divergence in South and North Korea 256

16.  Jejudo Korean  Ub ong Sh i n, Ji e u n K i a e r , 16.1 Introduction 









and

J i you n g S h i n

258 258

16.1.1 Jejudo Korean: Dialect or language? 258 16.1.2 Location and population 258

16.2 Historical connections: Genealogy and contact  258 16.2.1 The current state of Jejudo Korean 258 16.2.2 Dialect divisions 259 16.3 Phonology  259 16.3.1 Consonants 259 16.3.2 Vowels 259 16.3.2.1 Monophthongs 259 16.3.2.2 Diphthongs 260 16.3.3 Phonological phenomena 260 16.3.4 Suprasegmentals 261 16.4 Morphology  261 16.4.1 Nouns 261 16.4.2 Pronouns 261 16.4.3 Numerals 262 16.4.4 Particles 262 16.4.4.1 Case particles 262 16.4.4.2 Conjunctive particles 263 16.4.4.3 Additional meaning particles 264 16.4.4.4 Pragmatic particles 264 16.4.5 Prefinal endings 265 16.4.6 Quotation suffix 265 16.5 Syntax  265 16.5.1 Honorifics 265 16.5.2 Tense 266 16.5.2.1 Past tense 266 16.5.2.2 Present tense 267 16.5.2.3 Future tense 267 16.5.3 Aspect 267 16.6 Vocabulary  267

17.  Xibe and the Manchuric languages  269 Ta e h o J a n g 17.1 Introduction  269 17.1.1 Alternative names 269 17.1.2 Location and number of speakers 269

xviii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents









17.1.3 Writing systems 269 17.1.4 Previous scholarship 269

17.2 Historical connections: Genealogy and contact  270 17.2.1 Prototypically Transeurasian features 270 17.2.2 Language family and dialects 270 17.2.3 Relationship between Xibe and Manchu 270 17.2.4 Contacts 271 17.3 Phonology  271 17.3.1 Consonants 271 17.3.1.1 Spoken Xibe consonant phonemes 271 17.3.1.2 Written Xibe consonant phonemes 271 17.3.2 Vowels 272 17.3.2.1 Spoken Xibe vowel phonemes 272 17.3.2.2 Written Xibe vowel phonemes 273 17.3.3 Syllable structure 273 17.3.4 Morphophonology 274 17.3.4.1 Processes in which vowels condition consonants 274 17.3.4.1.1 Palatalization 274 17.3.4.1.2 Uvularization 274 17.3.4.1.3 Voicing assimilation (spoken) 274 17.3.4.1.4 Spirantization (spoken) 274 17.3.4.1.5 Labialization (spoken) 274 17.3.4.2 Processes in which consonants condition vowels 274 17.3.4.3 Processes conditioned by syllable structure 274 17.3.4.3.1 Aspiration 274 17.3.4.3.2 Fortis 274 17.3.4.3.3 Devoicing 274 17.3.4.3.4 Fricativization 274 17.3.4.3.5 Lateral retroflexation 274 17.3.4.3.6 Vowel [ɘ] deletion 274 17.3.4.3.7 Vowel [ɘ] epenthesis 274 17.3.4.3.8 Lateral [l] epenthesis 275 17.3.4.4 Vowel harmony 275 17.3.4.4.1 Labial harmony 275 17.3.4.4.2 RTR harmony 275 17.3.5 Stress patterns 275 17.4 Morphology  276 17.4.1 Nouns 276 17.4.1.1 Case 276 17.4.1.2 Number 276 17.4.1.3 Definite marker and possessive marker 277 17.4.2 Pronouns 277 17.4.3 Numerals 277 17.4.3.1 Numbers 277 17.4.3.2 Numerical classifiers 278 17.4.4 Property words 278 17.4.4.1 Adjective-forming suffixes 278 17.4.4.2 Degree 278 17.4.4.3 Intensity by reduplication 278 17.4.5 Verbs 278 17.4.5.1 Finite 278 17.4.5.1.1 Aspect 279 17.4.5.1.2 Negation 279

xix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





17.4.5.2 Nonfinite 279 17.4.5.2.1 Infinitives 279 17.4.5.2.2 Participles 280 17.4.5.2.3 Converbs 280 17.4.5.3 Use of nonfinite suffix as finite 280 17.4.6 Derivational morphology 280 17.4.6.1 Noun-forming derivational suffixes 280 17.4.6.2 Verb-forming derivational suffixes 280

17.5 Syntax 

















280

17.5.1 The clause 280 17.5.1.1 Basic word order 280 17.5.1.2 Subordination, cosubordination, and coordination 281 17.5.1.2.1 Subordinate 281 17.5.1.2.2 Cosubordinate 281 17.5.1.2.3 Coordinate 282 17.5.1.3 Expressions of ‘being’ and ‘having’ 282 17.5.1.3.1 Existential constructions (no LOC) 282 17.5.1.3.2 Possessive constructions 282 17.5.2 The nominal group 282 17.5.2.1 Case system 282 17.5.2.2 Topic marking 283 17.5.3 The verbal group 283 17.5.3.1 Voice and valency 283 17.5.3.1.1 Dative/accusative alternation 283 17.5.3.1.2 Causative and passive constructions 283 17.5.3.1.3 Causative/passive isomorphism 284 17.5.3.1.4 Reflexives and reciprocals 284 17.5.3.2 Tense, aspect, mood 284 17.5.3.2.1 Tense 284 17.5.3.2.2 Aspect and mood 284 17.5.3.3 Conjunct/disjunct 285 17.5.3.4 Negation 286

17.6 Lexicon 

286

18.  Even and the Northern Tungusic languages  288 B r ig i t t e Pa k e n d o r f a n d N ata l i a A r a l ova 18.1 Introduction  288 18.2 Historical connections: Genealogy and contact  290 18.3 Phonology  290 18.3.1 Consonants 290 18.3.2 Vowels 290 18.3.3 Syllable structure 291 18.3.4 Morphophonology 291 18.4 Morphology  292 18.4.1 Inflectional morphology of nouns 292 18.4.2 Pronouns 293 18.4.3 Numerals 294 18.4.3.1 Cardinal numerals 294 18.4.3.2 Numeral derivation 294 18.4.4 Property words 295 18.4.5 Inflectional morphology of verbs 295 18.4.6 Derivational morphology 299

xx

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents 18.4.6.1 Verb > Verb 299 18.4.6.2 Verb > Noun and Noun > Verb 300 18.4.6.3 Noun > Noun 300





18.5 Syntax 

301



18.6 Lexicon 

304



18.5.1 18.5.2 18.5.3 18.5.4

The clause 301 The nominal group 301 The verbal group 302 Complex sentences 303 18.5.4.1 Coordination 303 18.5.4.2 Subordination 303

19.  Nanai and the Southern Tungusic languages  305 S o f i a O s ko l s k aya 19.1 Introduction  305 19.2 Historical connections: Genealogy and contact  307 19.2.1 Phonological features 309 19.2.2 Morphological features 309 19.2.3 Syntactic features 309 19.2.4 Grammaticalization features 310 19.3 Phonology  311 19.3.1 Consonants 311 19.3.2 Vowels 312 19.3.3 Syllable structure 312 19.3.4 Morphophonology 312 19.3.5 Suprasegmentals 313 19.4 Morphology  313 19.4.1 Inflectional morphology of nouns 313 19.4.2 Inflectional morphology of pronouns 313 19.4.3 Numerals 315 19.4.4 Property words 315 19.4.5 Inflectional morphology of verbs 315 19.4.6 Derivational morphology 317 19.5 Syntax  318 19.5.1 The clause 318 19.5.2 The nominal group 318 19.5.3 The verbal group 319 19.6 Lexicon  320 20.  Dagur  Yo h e i Ya m a da 20.1 Introduction 





321 321

20.1.1 Profile of Dagur 321 20.1.2 The writing systems for Dagur 321 20.1.3 Language name 322 20.1.4 Sources 322

20.2 Historical connections  322 20.3 Phonology  322 20.3.1 Consonants 322 20.3.2 Vowels 323 20.3.3 Syllable structure 324 20.3.4 Morphophonology 324

xxi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

20.3.4.1 Vowel harmony 324 20.3.4.2 Inserted consonants 325



20.4 Morphology 

325



20.5 Syntax 

328



20.6 Lexicon 

332



20.4.1 Nominals 325 20.4.1.1 Nouns 325 20.4.1.2 Pronouns 326 20.4.1.3 Numerals 327 20.4.1.4 Adjectival words 327 20.4.2 Verbs 327 20.4.2.1 Finite form 327 20.4.2.2 Nonfinite: Participles, converbs 328 20.4.3 Derivational morphology 328



20.5.1 Clausal syntax 328 20.5.1.1 Basic word order 328 20.5.1.2 Main clause 329 20.5.1.3 Complex sentence 329 20.5.2 The nominal syntax 330 20.5.2.1 Case systems 330 20.5.2.2 Possessive agreement 331 20.5.2.3 Topic and focus marking 331 20.5.2.4 Nominal predicate 331 20.5.3 The verbal syntax 331 20.5.3.1 Voice 331 20.5.3.2 Tense 332 20.5.3.3 Aspect 332 20.5.3.4 Verbal negation 332

21.  Khalkha Mongolian  J a n -O l o f S va n t e s s o n 21.1 Introduction 





334 334

21.1.1 Writing systems 334 21.1.2 Previous scholarship 334

21.2 Historical connections  335 21.3 Phonology  335 21.3.1 Consonants 335 21.3.2 Vowels 336 21.3.3 Syllable and word structure 337 21.3.4 Morphophonology 338 21.3.4.1 Vowel harmony 338 21.3.4.2 Schwa ~ zero alternation 338 21.3.4.3 Other morphophonological processes 339 21.3.5 Suprasegmentals 339 21.4 Morphology  340 21.4.1 Nouns 340 21.4.2 Pronouns 340 21.4.3 Numerals 341 21.4.4 Property words 341 21.4.5 Verbs 342 21.4.5.1 Negation 343 21.4.6 Derivational morphology 343

xxii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

21.5 Syntax 

343



21.6 Lexicon 

349



21.5.1 The clause 343 21.5.1.1 Questions 344 21.5.1.2 Subordinate clauses 344 21.5.1.3 Being and having 345 21.5.2 The nominal group 345 21.5.3 The verbal group 346 21.5.3.1 Voice 346 21.5.3.2 Tense 346 21.5.3.3 Aspect 347 21.5.3.4 Mood 348

22.  Oirat and Kalmyk, the Western Mongolic languages  Á g n e s B i r ta l a n 22.1 Introduction 







350 350

22.1.1 Oirat—the linguistic, historical, and cultural term 350 22.1.2 Geography (location) 350 22.1.3 Demography 351 22.1.4 Orthography (writing systems) 352 22.1.5 Sources and previous scholarship on Oirat and Kalmyk 352

22.2 Historical contexts, influences, dialects  354 22.2.1 Historical Oirat 354 22.2.2 Oirat and Kalmyk dialects 354 22.3 Phonology and phonetics of Oirat  355 22.3.1 The consonant system of Oirat and Kalmyk 356 22.3.2 The vowel system 356 22.3.3 Morphemic (syllable) structure 357 22.3.4 Oirat and Kalmyk morphophonology 358 22.3.4.1 Vowel harmony 358 22.3.4.2 Accent 358 22.4 Morphology  358 22.4.1 Parts of speech 358 22.4.2 Nouns (number and case) 358 22.4.2.1 Plurality 358 22.4.2.2 Cases 359 22.4.3 Pronouns 360 22.4.3.1 Personal pronouns and possessive declension 360 22.4.3.2 Demonstrative pronouns 361 22.4.3.3 Reflexive pronoun 361 22.4.3.4 Interrogative pronouns 361 22.4.3.5 Indefinite pronouns 362 22.4.4 Numerals 362 22.4.5 Verbal morphology 363 22.4.5.1 Imperatives 363 22.4.5.2 Participles 363 22.4.5.3 Converbs 364 22.4.5.4 Tense and aspect markers 366 22.4.5.5 Person 366 22.4.6 Derivational morphology 366 22.4.6.1 Nominal derivation 367 22.4.6.2 Verbal derivation 367

xxiii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

22.5 Syntax 

367



22.6 Lexicon 

369



22.5.1 Syntactic structure, basic word order 367 22.5.2 Topic and focus marking 368 22.5.3 Negation 368 22.5.4 Reported speech 368 22.5.5 Levels of speech 368

23.  The Northwestern Turkic (Kipchak) languages  370 É va Á. C s at ó a n d L a r s J o h a n s o n 23.1 Introduction  370 23.2 The Northwestern branch of Turkic  370 23.2.1 Kazakh 371 23.2.2 Karaim 371 23.3 Typical Turkic features in Northwestern Turkic  373 23.3.1 Sound systems 373 23.3.1.1 Vowels 373 23.3.1.2 Consonants 374 23.3.1.3 Sound harmony 376 23.3.1.4 Some typical morphophonological rules 376 23.3.1.5 Syllable structure 377 23.3.1.6 Word accent 377 23.3.2 Word structure 377 23.3.2.1 Nominal categories: Number, possessive, and case 378 23.3.2.2 Pronouns 379 23.3.2.3 Postpositions 380







23.3.2.4 Verbal categories: Viewpoint aspect, mood, modality, actionality 380 23.3.2.4.1 Intraterminals 380 23.3.2.4.2 Terminals 381 23.3.2.4.3 Postterminals. Evidentials 382 23.3.2.4.4 Aorist 382 23.3.2.4.5 Imperative mood 382 23.3.2.4.6 Voluntative modality 383 23.3.2.4.7 Optative modality 383 23.3.2.4.8 Hypothetical modality 383 23.3.2.4.9 Necessitative modality 384 23.3.2.4.10 Postverbial constructions 384 23.3.3 Copula 385 23.3.4 Nonfinite verb forms 385 23.3.4.1 Action nominals 385 23.3.4.2 Participant nominals 386 23.3.4.3 Converbs 386 23.3.5 Clause structure 386 23.3.5.1 Basic word order 387 23.3.5.2 Nominal phrase compounds 387 23.3.5.3 Clause embedding 387 23.3.5.4 Relative clauses 388 23.3.5.5 Predicative possession and comparison 388 23.3.5.6 Interrogation 389

23.4 Grammaticalization  23.5 Lexicon 

xxiv

Dictionary:

389 389

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

23.6 Notations 

391

24.  Turkish and the Southwestern Turkic (Oghuz) languages  392 J a k l i n K o r n f i lt 24.1 Introduction  392 24.2 Historical connections: Genealogy and contact  393 24.3 Phonology  394 24.3.1 Consonants 394 24.3.2 Vowels 394 24.3.3 Syllable structure 395 24.3.4 Morphophonology 396 24.3.4.1 Syllable-final oral stop devoicing 396 24.3.4.2 The k/∅ alternation 396 24.3.4.3 Word-final liquid devoicing 396 24.3.4.4 Morpheme-initial voicing assimilation 396 24.3.4.5 Labial attraction 397 24.3.5 Suprasegmentals 397 24.4 Morphology  397 24.4.1 Inflectional morphology of nouns 398 24.4.2 Pronouns and “possessive” suffixes 398 24.4.3 Numerals 400 24.4.4 Property words 401 24.4.5 Verb 401 24.4.6 Derivational morphology 403 24.5 Syntax  403 24.5.1 The clause 403 24.5.2 The nominal group 407 24.5.3 The verbal group 408 24.6 Lexicon  409 25.  Uyghur and Uzbek, the Southeastern Turkic languages  411 A bdu r i s h i d Ya k u p 25.1 Introduction  411 25.2 Historical connections: Genealogy and contact  412 25.3 Phonology  413 25.3.1 Consonants 413 25.3.2 Vowels 414 25.3.3 Morphophonology 415 25.3.4 Umlauting 415 25.3.5 Other sound changes 415 25.4 Morphology  416 25.4.1 Nouns 416 25.4.2 Pronouns 417 25.4.3 Numerals and classifiers 419 25.4.4 Verbals: Imperatives 419 25.4.5 Tense and aspect 420 25.4.6 Nonfinite forms 422 25.4.7 Modal forms 424 25.4.8 Derivational morphology 425 25.5 Syntax  425 25.5.1 Compounding 425 25.5.2 Nominal phrases 426

xxv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





25.5.3 Postpositional phrases 426 25.5.4 Relative clauses 426 25.5.5 Voice and valency 426 25.5.6 Negation 427 25.5.7 Verbal phrases coding actionality 427 25.5.8 The sentence: predicate nominals 427

25.6 Lexicon 

429

26.  Sakha and Dolgan, the North Siberian Turkic languages  430 B r ig i t t e Pa k e n d o r f a n d E ug é n i e S ta p e r t 26.1 Introduction  430 26.2 Historical connections: Genealogy and contact  430 26.3 Phonology  432 26.3.1 Consonants 432 26.3.2 Vowels 433 26.3.3 Phonotactics 433 26.4 Morphology  433 26.4.1 Nouns 433 26.4.2 Pronouns and possessive suffixes 434 26.4.3 Numerals 436 26.4.4 Property words 437 26.4.5 Verbs 437 26.4.5.1 Tense 438 26.4.5.2 Mood 438 26.4.5.3 Aspect/Aktionsart 440 26.4.5.4 Nonfinite verbal morphology 440 26.4.6 Derivational morphology 441 26.4.6.1 Verbal derivation 441 26.4.6.2 Nominal derivation 442 26.4.6.3 Other derivational suffixes 442 26.5 Syntax  442 26.5.1 The clause 442 26.5.2 The nominal group 443 26.5.3 The verbal group 444 26.5.4 Complex sentences 444 26.6 Lexicon  445 27.  Chuvash and the Bulgharic languages  446 A l e x a n d e r S av e ly e v 27.1 Introduction  446 27.2 Historical connections: Genealogy and contact  447 27.3 Phonology  449 27.3.1 Consonants 449 27.3.2 Vowels 450 27.3.3 Syllable and word structure 451 27.3.4 Morphophonology 451 27.3.5 Suprasegmentals 451 27.4 Morphology  452 27.4.1 Nouns 452 27.4.1.1 Number 452 27.4.1.2 Case 452 27.4.1.3 Possessive suffixes 453 27.4.2 Pronouns 453

xxvi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





27.4.3 Numerals 454 27.4.4 Property words 455 27.4.5 Verbs 455 27.4.5.1 Person and number 455 27.4.5.2 Tense and mood 456 27.4.5.2.1 Indicative 456 27.4.5.2.2 Imperative 457 27.4.5.2.3 Conditional 458 27.4.5.2.4 Concessive and optative 458 27.4.5.3 Negation 458 27.4.5.4 Potential form 458 27.4.5.5 Aspect 459 27.4.5.6 Nonfinite verbs 459



27.5.1 The nominal group 459 27.5.2 The verbal group 460 27.5.3 The clause 461





27.5 Syntax 

459



27.6 Lexicon 

463

465

PART III:  Comparative Overviews 

A. Phonology  28. A comparative approach to the consonant inventory of the Transeurasian languages  A l l a n  R. B o m h a r d 28.1 Introduction  28.2 The reconstruction of the Proto-Transeurasian consonant inventory  28.3 Correspondences  28.4 Turkic  28.5 Mongolic  28.6 Tungusic  28.7 Korean  28.8 Japonic  28.9 Root structure patterning in Proto-Transeurasian 

469 469 470 472 476 478 480 482 482 485

29. A comparative approach to the vowel systems and harmonies in the Transeurasian languages and beyond  486 A n d r e w J o s e p h , S e o n g y e o n K o, a n d J o h n Wh i t m a n 29.1 Introduction  486 29.2 Vowel system and harmony in the Transeurasian languages  486 29.2.1 Backness (palatal) vs. tongue-root vowel systems 486 29.2.1.1 What is a [±back] system? 487 29.2.1.2 What is a [±rtr] system? 487

29.2.1.3 TRH systems in the Korean, Mongolic, and Tungusic languages 487 29.2.1.4 Evidence in favor of [±rtr] over [±atr] as the harmonic feature 488 29.2.2 Reconstructing [±rtr] harmony in Korean, Mongolic, and Tungusic 489 29.2.2.1 Basic vowel correspondences in the KMT languages 489 29.2.2.2 PH-to-RTRH Shifts in Mongolic and Korean? 490

xxvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

29.3 Reconstruction of classical Proto-Altaic as [±rtr]  490 29.3.1 Ramstedt’s Proto-Altaic as a [±rtr] system 491







29.3.1.1 Proto-Altaic vowel inventory according to Ramstedt (1952–66) 491 29.3.1.2 Vowel quantity according to Ramstedt (1952–66) 492 29.3.1.3 Reflexes in descendent languages 492 29.3.1.4 Reinterpreting the Ramstedt system 493 29.3.1.5 Ramstedt’s Proto-Altaic forms under the tongue-root hypothesis 493 29.3.2 Poppe’s Proto-Altaic as a [±rtr] system 495 29.3.2.1 Proto-Altaic vowel inventory according to Poppe (1960b) 495 29.3.2.2 Vowel quantity according to Poppe (1960b) 496 29.3.2.3 Reflexes in descendent languages 496 29.3.2.3.1 Short vowels 496 29.3.2.3.2 Long vowels 496 29.3.2.4 Poppe’s Proto-Altaic forms under the tongue-root hypothesis 496

29.4 Genetic versus areal accounts of TRH  497 29.4.1 Micro-Altaic 497 29.4.2 Macro-Altaic 502 29.4.3 TRH and tongue-root vowel inventories in Northeast Asia 504 29.4.4 TRH and tongue-root vowel inventories beyond Eurasia? 505 29.4.4.1 TRH in North America 506 29.4.4.2 Other tongue-root systems in North America 506 29.4.5 Two types of [±rtr] system: “balanced” vs. “unbalanced”? 507

B. Morphology  30.  A comparative approach to verbal morphology in Transeurasian  511 M a r t i n e R o bbe e t s 30.1 Introduction  511 30.2 Overview of the shared verbal morphology of Transeurasian  511 30.3 Correlations in form  514 30.3.1 Regular sound correspondences 514 30.3.2 Shared allomorphy 515 30.4 Correlations in function  515 30.5 Paradigmaticity  517 30.6 How likely is coincidence?  519 30.7 Conclusion  520 31. A comparative approach to nominal morphology in Transeurasian: Case and plurality  522 I lya G ru n t ov a n d O l g a M a z o 31.1 Introduction  522 31.2 Number  522 31.2.1 Mongolic 522 31.2.1.1 Singularity in Mongolic 522 31.2.1.2 Paucalis and Dualis in Mongolic 523 31.2.1.3 Plurality in Mongolic 523 31.2.1.3.1 Lexical expression of plurality 526 31.2.1.3.2 Associative plurality 526 31.2.1.4 Summary 526

xxviii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





31.2.2 Turkic 527 31.2.2.1 Chuvash -sem 527 31.2.2.2 Common Turkic -lAr 527 31.2.2.3 The marker -t 527 31.2.2.4 The marker -an 527 31.2.2.5 Fossilized plural markers 528 31.2.3 Tungusic 528 31.2.4 Korean 530 31.2.5 Japonic 530 31.2.6 Number in Transeurasian 532



31.3.1 Mongolic 533 31.3.1.1 Genitive vs. accusative 538 31.3.1.2 Local cases 539 31.3.2 Turkic 539 31.3.3 Tungusic 541 31.3.3.1 The Tungusic case system 544 31.3.4 Korean 545 31.3.5 Japonic 550 31.3.6 Patterns on the proto-level 552

31.3 Case 

533

32.  A comparative approach to the pronominal system in Transeurasian  554 M i c h a l S c h wa r z , O n d ř e j S r b a , a n d Vac l av B l a ž e k 32.1 Typological introduction  554 32.2 Personal pronouns in the Transeurasian languages  555 32.2.1 Pronominal declension in the Turkic languages 563 32.2.2 Pronominal declension in the Mongolic languages 563 32.2.3 Pronominal declension in the Tungusic languages 568 32.2.4 Pronominal declension in Korean 568





32.2.5 Pronominal declension in Old Japanese and Ryukyu dialects (without the number distinction) 572 32.2.6 Proto-Transeurasian system of personal pronouns 573



32.3.1 Demonstrative pronouns 575 32.3.2 Possessive pronouns, enclitics, pronominal n 576 32.3.3 Reflexive or reflexive-possessive pronouns 577 32.3.3.1 Turkic 577 32.3.3.2 Mongolic 578 32.3.3.3 Tungusic 578 32.3.3.4 Korean 579 32.3.3.5 Japanese 579 32.3.4 Interrogative pronouns 579 32.3.5 Indefinite and negative pronouns 582 32.3.6 Nouns in the pronominal function 583 32.3.7 Pronominal verbs 584 32.3.8 Borrowed elements 584

32.3 Other types of pronouns 

575

C. Syntax  33. The nominal group, possessive agreement, and nominal sentences in the Transeurasian languages  I r i n a N e vs k aya a n d L i n a A m a l

587

xxix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

33.1 Introductory remarks on relevant Transeurasian typological features  587 33.2 Nominal groups  588 33.2.1 Definiteness, specificity, topicality 588 33.2.1.1 Indefinite article functions of bir in Turkic 588



33.2.1.2 Specificity and definiteness marking in Turkic, Mongolic, and Tungusic 588 33.2.1.3 Japanese and Korean topic particles 588 33.2.2 Agreement within a noun phrase 589 33.2.2.1 Adjective + noun: Agreement between head nouns and their adjectival attributes 589 33.2.2.2 Numeral + noun: Number agreement in phrases with numerals and quantifiers589 33.2.2.3 Adnominal possession and agreement in attributive possessive constructions590 33.2.2.4 Possessive noun phrases with accompanying affixes on the modifiers 593

33.3 Simple sentences and syntactic functions of case forms  593 33.3.1 Position of sentence members 594

33.3.2 Number and person agreement between the subject and the predicate 594 33.3.3 Syntactic functions of case forms: an overview 595



33.4.1 Existential constructions 596 33.4.2 Predicative possession 597 33.4.2.1 External possession constructions 597 33.4.3 Nonexistential nominal sentences 598



33.4 Nominal sentences 

596



33.5 Some conclusions  Appendix 

598 599

34.  Verbal categories in the Transeurasian languages  A n d r e j M a l c h u kov a n d Pat ry k C z e rw i n s k i 34.1 Introduction  34.2 General morphological pattern  34.3 Valency and voice 

604

35.  Complex constructions in the Transeurasian languages  A n d r e j M a l c h u kov a n d Pat ry k C z e rw i n s k i 35.1 Introduction  35.2 Complement clauses  35.3 Adverbial clauses  35.4 Relative clauses (prenominal) 

625

604 604 606 34.3.1 Causatives, passives, and adversative passives 607 34.4 Aspect, Aktionsart, and related categories  610 34.5 Negation  611 34.6 Modality  613 34.7 Tense  615 34.8 Mood  616 34.9 Evidentiality  616 34.10 Verbalization of participles and renewal of finite forms  617 34.11 Agreement  621 34.12 Politeness and register  623 34.13 Conclusions  623

xxx

Dictionary:

625 625 629 634

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

35.5 Internally headed relative clauses  35.6 Paratactic constructions and adverbial conjunctions  35.7 Concluding remarks 

637 640 641

D.  Lexicon and Semantics  36.  Basic vocabulary in the Transeurasian languages  645 M a r t i n e R o bbe e t s 36.1 Introduction  645 36.2 Overview of the basic vocabulary of Transeurasian  646 36.3 Arguments against borrowing  654 36.3.1 Basic vocabulary 654 36.3.2 Borrowing hierarchy 654 36.3.3 Typology of verbal borrowing 654 36.3.4 Regularity and complexity of sound correspondence 655 36.3.5 Broken contact chain 655 36.3.6 Multiple setting 655 36.3.7 Absence of prototypical loan characteristics of form and meaning 656 36.3.8 Well-spread distribution 656 36.4 Arguments against coincidence  656 36.5 Conclusion  658 37.  Numerals in the Transeurasian languages  660 Vác l av B l a ž e k 37.1 Methodological approach  660 37.2 Turkic numerals  660 37.2.1 Survey of forms 660 37.2.2 Comments 660 37.3 Mongolic numerals  667 37.3.1 Survey of forms 667 37.3.2 Comments 667 37.4 Tungusic numerals  672 37.4.1 Survey of forms 672 37.4.2 Comments 672 37.5 Korean numerals  672 37.5.1 Survey of forms 672 37.5.2 Comments 672 37.6 Japonic numerals  678 37.6.1 Survey of forms 678 37.7 Numerals in the Transeurasian daughter proto-languages and their probable cognates  678 37.8 Conclusion  678 37.9 Systems of cardinal numerals in ‘Paleo-Siberian languages’  678 38.  Kinship-term paradigms in the Transeurasian languages  691 M i l a n va n B e r l o 38.1 Introduction  691 38.2 Methodology  692 38.3 Transeurasian sibling terms  693 38.3.1 Datasets 693 38.3.2 Japonic 694 38.3.3 Korean 696

xxxi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





38.3.4 Tungusic 697 38.3.5 Mongolic 697 38.3.6 Turkic 698

38.4 Discussion and conclusion 

PART IV:  Areal versus Inherited Connections 

699

703

39. Contact between genealogically related languages: The case of Old Korean and Old Japanese  705 A l e x a n d e r  T. F r a n c i s -R at t e a n d J. M a r s h a l l U n g e r 39.1 Introduction  705 39.2 The Proto-Japano-Koreanic reconstruction  706 39.2.1 Phonemic correspondences 706 39.2.2 Morphology 706 39.3 Borrowings from Old Korean into Old Japanese  707 39.3.1 Hypothesized loanwords from Korean 708 39.3.2 Vowel correspondences of borrowings 709 39.3.3 Consonant correspondences 710 39.3.4 Typological and semantic features of borrowings 710 39.3.5 Morphological borrowing (or lack thereof) 711 39.3.6 So-called Paekchean glosses in Nihon shoki 711 39.3.7 Conclusions: Borrowings vs. cognates 713 39.4 Implications for the Transeurasian hypothesis  713 39.4.1 The place of Proto-Japano-Koreanic in Transeurasian 713 40. Form and pattern borrowing across Siberian Turkic, Mongolic, and Tungusic languages  715 G r e g o ry  D. S. A n d e r s o n 40.1 Introduction  715 40.2 Lexical and phonological lateral feature transfer between Turkic, Mongolic, and Tungusic languages of Siberia  716 40.2.1 Lexical copy or transfer 716 40.2.2 Phonological copy or transfer 719 40.3 Morphological and morphosyntactic lateral feature transfer between Turkic, Mongolic, and Tungusic languages of Siberia  720 40.3.1 Derivational morphology copy or transfer 720 40.3.2 Copying of inflectional paradigms and inflectional markers 720



40.3.3 Pattern copying and metatypic shift: Function neutralization and maintenance721

40.4 Lateral transfer of syntactic features between Turkic, Mongolic, and Tungusic languages of Siberia  40.5 Summary 

724 725

41.  Transeurasian as a continuum of diffusion  726 E dwa r d Vaj da 41.1 Introduction  726 41.2 Inner and Northern Eurasia as a language area  726 41.3 Pastoral Eurasia and the “Paleo-Asiatic” languages  728 41.3.1 Yeniseian 728 41.3.2 Yukaghir 729 41.3.3 Chukchi-Kamchatkan 730

xxxii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents





41.3.4 Eskaleut 730 41.3.5 Nivkh 731 41.3.6 Ainu 732

41.4 Korean and Japanese  41.5 Future prospects for Transeurasian studies 

732 733

42. Beck-Wichmann-Brown evaluation of lexical comparisons for the Transeurasian proposal  735 C e c i l  H. B row n 42.1 Introduction  735 42.2 Evidence for Transeurasian  735 42.3 The Beck-Wichmann-Brown system  736 42.3.1 Core Beck-Wichmann-Brown 736 42.3.1.1 Beck-Wichmann-Brown point-assignment 736 42.3.2 Beck-Wichmann-Brown enhancement 737 42.3.3 Beck-Wichmann-Brown results and interpretation 738 42.3.3.1 Beck-Wichmann-Brown metrics 738 42.3.3.2 Interpreting Beck-Wichmann-Brown results 740 42.4 Beck-Wichmann-Brown results for 96 language comparisons including 10 for Transeurasian  740 42.5 Transeurasian comparisons  741 42.6 Conclusion  749

PART V:  Interdisciplinary Perspectives on the Identity of Transeurasian 

751

43.  The homelands of the individual Transeurasian proto-languages  753 M a r t i n e R o bbe e t s , J u h a J a n h u n e n , A l e x a n d e r S av e ly e v, a n d E vg e n i ya K o rov i n a 43.1 Introduction  753 43.2 The Turkic homeland  754 43.2.1 Location 754 43.2.2 Time depth 755 43.2.3 Agricultural vocabulary 756 43.3 The Mongolic homeland  758 43.3.1 Location 758 43.3.2 Time depth 759 43.3.3 Agriculture 760 43.4 The Tungusic homeland  761 43.4.1 Location 761 43.4.2 Time depth 762 43.4.3 Agriculture 763 43.5 The Koreanic homeland  764 43.5.1 Location 764 43.5.2 Time depth 765 43.5.3 Agriculture 766 43.6 The Japonic homeland  767 43.6.1 Location 767 43.6.2 Time depth 768 43.6.3 Agriculture 768 43.7 Conclusion  770

xxxiii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents 44.  The Transeurasian homeland: Where, what, and when?  772 M a r t i n e R o bbe e t s 44.1 Introduction  772 44.2 Linguistic inferences about the Transeurasian past  772 44.2.1 Dating of the ancestral speech communities 772 44.2.1.1 The Proto-Transeurasian speech community 772 44.2.1.2 The Proto-Altaic speech community 773 44.2.1.3 The Mongolo-Turkic speech community 773 44.2.1.4 The Japono-Koreanic speech community 773 44.2.2 Location of the ancestral homelands 773 44.2.2.1 The Proto-Transeurasian homeland 773 44.2.2.2 The Proto-Altaic homeland 774 44.2.2.3 The Proto-Turko-Mongolic homeland 774 44.2.2.4 The Proto-Japano-Koreanic homeland 774 44.2.3 Reconstruction of agricultural vocabulary 775 44.2.3.1 Proto-Transeurasian 775 44.2.3.2 Proto-Altaic 775 44.2.3.3 Proto-Turko-Mongolic 777 44.2.3.4 Proto-Japano-Koreanic 777 44.3 Triangulation  780 44.3.1 Genetics 780 44.3.2 Archaeology 781





44.3.2.1 The origins of millet agriculture in the West Liao River Basin 781 44.3.2.2 The eastward spread of millet agriculture 781 44.3.2.3 The integration of rice and millet agriculture 781 44.3.3 Mapping the three lines of evidence 781

44.4 Conclusion 

783

45.  Transeurasian unity from a population-genetic perspective  784 C h o o n g wo n J e o n g , C h ua n -C h ao Wa n g , a n d C h ao N i n g 45.1 Introduction  784 45.2 An overview of the genetic structure within Transeurasians  786 45.3 A shared genetic substratum among the Altaic populations  787 45.3.1 Western Eurasian admixture in Turkic populations 787



45.3.2 Genetic evidence for the shared pre-admixture substratum among the Altaic populations 789 45.3.3 Paleogenomic evidence for the shared ancestry of the Altaic populations 790

45.4 Genetic connection between Japonic and Koreanic populations  45.5 Genetic connection between the Altaic and the Japonic/Koreanic populations  45.6 Discussion 

46.  Transeurasian unity from an archaeological perspective  T ao L i 46.1 Introduction  46.2 Millet farming in the West Liao River valley of Northeast China  46.3 Millet farming in the Primorye region of the Russian Far East  46.4 Millet- and rice-farming on the Korean Peninsula  46.5 Millet- and rice-farming on the Japanese Archipelago 

xxxiv

Dictionary:

791 791 792 794

794 794 797 799 800

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Detailed Contents

46.6 Assumptions underlying the TEA model examined in archaeological evidence  46.7 Concluding remarks 

802 804

47. Language dispersals and the “Secondary Peoples’ Revolution”: A historical anthropology of the Transeurasian unity  806 M a r k Ja m e s Hu ds on 47.1 Introduction  806 47.2 The Neolithic, agrarian states, and language dispersals  807 47.3 Some caveats to the “secondary peoples’ revolution”  808 47.3.1 States and porous borderlands 808 47.3.2 Agriculture 808 47.3.3 Barbarian states 809 47.4 Some ‘barbarian’ sketches  810 47.4.1 Nomadic pastoralism 810 47.4.2 Sakha sub-arctic horse and cattle husbandry 810 47.4.3 Transeurasian hunter-gatherers 811 47.4.4 The Ryukyu Islands and farming expansions beyond the state 812 47.5 Conclusions  813 References  Index 

815 881

xxxv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Series Preface We know that close study of individual language families and linguistic areas is vital to both the synchronic and the diachronic study of language and to cognitive science more widely. Comparative investigations of this type stimulate exciting synergies between different subdisciplines of linguistics, such as language change, contact linguistics, sociolinguistics, linguistic typology, textual philology, and microvariation in grammar, sound, and meaning within and across languages. Besides reflecting and encouraging the links between these subdomains, the fundamental goal of the series is to publish high-quality, substantial reference works which represent a set of theoretically informed and systematic guides to what is known about the world’s languages. Each Guide focuses on a particular language family, subfamily, or areal grouping, and is edited by leading authorities, who bring together contributions from the best international scholars in the field. The Guides aim to show the more general theoretical significance of the languages’ history, linguistic and sociolinguistic characteristics, and overall to provide an indispensable reference tool both to specialist scholars and students and to professional linguists. The approach adopted in all the Guides is systematic and comparative, informed by the latest research and theoretical and methodological perspectives, and, where appropriate, the authors draw on relevant work in such fields as anthropology, archaeology, and cognitive science. Adam Ledgeway and Martin Maiden University of Cambridge and University of Oxford February 2015

xxxvi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Abbreviations a) Languages AA Afroasiatic Agar. Agarinakasone Ain. Ainu Alt. Altaic, Altay Az. Azerbaijanian Balk. Balkar Bao. Bao’an Bash. Bashkir Bulg. (Old) Bulghar Bur. Buryat Bys. Bystraja dialect of Even CENTR The Korean central dialects Ch. (Mandarin) Chinese ChK Chukcho-Kamchatkan Chu. Chuvash CHUNG Ch’ungch’ŏng dialect of Korean CJ Classical Japanese CJ.Ch Classical Japanese (Chinese set) COJ Central Old Japanese CoMo. Common Mongolian Crim. Tatar Crimean Tatar CTk. Common Turkic Dag. Dagur Dgx. Dongxian Dr. Dravidian EBur. East Buryat EIr. East Iranian EMCh. Early Middle Chinese EMJ Early Middle Japanese EMK Early Middle Korean Eng English EOJ Eastern Old Japanese EOT Eastern Old Turkic Evk. Evenki Evn. Even EYu. Eastern Yughur FN Fulan nura dialect of Mongghul language FU Fenno-Ugric G Karaim dialect from Galicia

Gag. Gagauz GANG Kangwŏn dialect of Korean GYEONGG Kyŏnggi dialect of Korean GYEONGS Kyŏngsang dialect of Korean H Halchi gol dialect of Mongghul language HAM Hamgyŏng dialect of Korean Hater. Hateruma HR Hirara HWANG Hwanghae dialect HY Hua-Yi Yiyu Middle Mongolian IE Indo-European IG Ishigaki J (Contemporary/Standard) Japanese JE Jeju dialect of Korean JEOL Chŏlla dialect of Korean Jur. Jurchen K (Contemporary/Standard) Korean Kalm. Kalmyk Karach. Karachai Karakh. Karakhanid Kaz. Kazakh KBalk. Karachay-Balkar Kgj. Kangjia Khak. Khakas Khal. Khalkha Khamn. Khamnigan Khan. Khanty Kirg. Kirghiz KKalp. Karakalpak KMT Korean-Mongolic-Tungusic KP Kalyānamkara ve Papāmkara Old Turkic Kumkang 1464 Kumkang panya phalamil kyeng   enhay; Middle Korean KZ Komi Zyrian Lam. Lamunkhin dialect of Even LH Late Han Chinese LMCh. Late Middle Chinese LMJ Late Middle Japanese LMK Late Middle Korean

xxxvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

A bbr eviations LOJ Late Old Japanese LOK Late Old Korean MA Muqaddimat al-Adab Middle Mongolian Ma. Manchu Manghr. Mangghuer MCh. Middle Chinese Mgr. Monguor MIr Middle Iranian MJ Middle Japanese MK Middle Korean MMo. Middle Mongolian Mo. Mongolian MoJ Modern Japanese Mog. Moghol MoMo. Modern Mongolian Monghl. Mongghul Mord. Mordva MTk. Middle Turkic Na. Nanai Narin. Naringhol NE Northeastern or Siberian branch of Turkic Neg. Negidal Niv Nivkh NK North Korean Nog. Noghai NW Northewestern or Kipchak branch   of Turkic Oir. Oirat OJ Old Japanese OK Old Korean Okin. Okinawan OOkin. Old Okinawan OOsman Old Osman OR Old Ryukyuan Orkh. Orkhonic Oroq. Oroqen OTk Old Turkic PA Proto-Altaic PAN Proto-Austronesian PB Proto-Basque Pc Paekche PCK Proto-Chukotko-Kamchatkan PIE Proto-Indo-European PJ Proto-Japonic PJK Proto-Japano-Koreanic PJR Proto-Japanese-Ryukyan

PK Proto-Koreanic PKJ Proto-Koreano-Japonic PMo Proto-Mongolic pOJ Pre-Old Japanese PR Proto- Ryukyan PRJ Proto-Ryukyu-Japanese PTEA Proto-Transeurasian languages PTg Proto-Tungusic PTk Proto-Turkic PUr Proto-Uralic PY Proto-Yakut PYONG P’yŏngan dialect of Korean Ru. Russian Ruk. Mantauran Rukai Sal. Salar (Uighur) SE Southeastern or Karluk branch of Turkic SH Secret History Middle Mongolian Skt. Sanskrit Sol. Solon STg. South Tungusic SW Southwestern or Oghuz branch of Turkic SY Shira Yughur, Eastern Yughur (Mongolic) S-Yug. Sary Uyghur, Yellow Uighur (Turkic) Tat. Tatar TEA Transeurasian languages Tg. Tungusic Tib. Tibetan Tk Turkish Tkm. Turkmenian Tofa. Tofalar Tr Karaim dialect from Trakai Turk. Turkic Tuv. Tuvan Ud. Udehe Udm. Udmurt Uig. Uighur Ur. Uralic Uzb. Uzbek VBulg. Volga Bulghar WMo. Written Mongolian WOir. Written Oirat (written in Oirat   or Clear script) WOJ Western Old Japanese XUAR   Xinjiang Uygur Autonomous Region YG Yonaguni Yuk. Yukaghir

xxxviii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Abbr eviations b) Linguistic forms * reconstructed form 1, 2, 3 person A the most agent-like argument of   a transitive verb A.INTRA finite verb form based on an   A-converb form ABIL abilitative ABL ablative ABS absolutive ACC accusative ACOM anticomissive ACQUIS acquisition ACT actional ADD additive ADJ adjective, adjectival inflection ADMIR admirative ADNZ adnominal(izer) ADV adverbial(izer) ADVRS adversative AF actor focus AFF affirmative AG agentivizer AH addressee honorific AI ablative-instrumental ALL allative ALLIT alliterative ALN alienable (possession) AM associated motion AN action nominal ANT anterior AOR aorist APPROX approximative ASP aspect ASS assertive ASSOC associative ATIN.AN action nominal in {-Y/ATIN} ATIN.PN participant nominal in {-Y/ATIN} ATT attenuative ATTR attributive AUG augmentative AUGM augmented AUX auxiliary BEN benefactive

BV basic vocabulary BVi basic-vocabulary index BWB Beck-Wichmann-Brown CAP capability CAR caritive CASa semantic case particle CASb syntactic case particle CAUS causative CL clause CLF classifier CMP complement(izer) CNT numeral counter COLL collective COM comitative COMi composite index (CORi x +BVi). COMP comparative CON construction CONC concessive CONCL conclusive COND conditional CONF confirmative CONJ conjecture CONN connective, conjunctive CONNEG connegative CONT continuous CONTR contrastive COOP cooperative COORD coordinator COP copula CORi correspondence index (number of BWB   correspondence series) CRT correlational CSL causal CVB converb D dependent DAT dative DAY day classifier DEADJ deadjectival DEB debitive DEC decimal, ten DECAUS decausative DECL declarative DEF definite

xxxix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

A bbr eviations DEFER deferential DEL delimiter particle  DELAT delative DEM demonstrative DEONT deontic DES desiderative DESG designative DEST destinative case DET determiner DIAL dialectal DIM diminutive DIR directional, directive DISJ disjunct DIST distal DISTR distributive DP discourse particle DS different subject converb DU dual DUB dubitative DUR durative E.COP copula ‘to be’ EGRE egressive ELAT elative EMPH emphatic ENCL enclitic END ending EP epenthetic vowel EPIST epistemic EPST evidential past EQU equative ERG ergative EVAL evaluative EVID evidential EVK Evenki copy EX exalted (honorific) EXCL exclusive EXIST existential EXP experiencer EXT extended stem F feminine FAC factitive FIEN fientive FIN finite FNOM factive nominal FOC focus particle

FP sentence final particle FREQ frequentative FUT future GA genitive-accusative GAN.PN participant nominal in {GAN} GAN.POST.AN action nominal in {GAN} GEN genitive GER gerund GNR generic HAB habitual HON honorific HONT honorific title HORT hortative HUM human HUMB humble HYP hypothetical ICVB imperfective converb IFQ information question IHRC internally headed relative clause IMM immediate IMP imperative IMP.DIST distant imperative IMPS impersonal INCEP inceptive INCH inchoative INCL inclusive IND indicative INDEF indefinite INDET indeterminative INDP independent INF infinitive INFR inferential INS instrumental INT intentional INTER interrogative INTJ interjection INTR intransitive INTRA intraterminal INTS intensive INV invitation IPFV imperfective IPRF imperfect participle IPST inferential past IRR irrealis ITER iterative

xl

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Abbr eviations JUNC free junctor JUSS jussive LAT lative LD linkage disequilibrium LEVEL speech level LMT limitative LOC locative LST listing M masculine MD mood MDL modality MES mesial MID middle voice MIN minimal MULT multiplicative N noun, nominal NEC necessitive NEG negative/negation NF nonfinite NFUT non-future NHON non-honorific NHUM non-human NMDL negative modal NMLZ nominalizer/nominalization NOM nominative NP noun phrase NPM NP-use marker (Juntai joshi) NPST non-past NPSTP nonpast participle NSIM non-simultaneous converb NUM numeral º stem O object OBJ objective OBL oblique OCMP object of comparison OPT optative ORD ordinal numeral ORIG originative P the most patient-like argument of   a transitive verb PART partitive PASS passive PAUC paucal PCA principal component analysis

PCVB perfective converb PERM permissive PFC predicate of focus construction PFV perfective PH palatal harmony PL plural PLQ polar question PLUPRF pluperfect PN proper name PNM participant nominal POL polite speech level POS posterior POSS possessive POSSES possessor POST postterminal PostP postpositional (phrase) POT potential PRED predicative PRES presumptive PREV preventive PRF perfect PRFL reflexive possessive PRIV privative PROB probabilitive PROC processive PROF professional PROG progressive PROH prohibitive PROL prolative PRON pronominal PROP propositive PROPR proprietive PROSEC prosecutive PROX proximal PRPR preparative PRS present (tense) PRV privative PST past (tense) PSTP past participle PTCL particle PTCP participle PURP purposive Q question (particle) QTF quantifier QUOT quotative, citation

xli

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

A bbr eviations R Russian copy RC relative clause REAL realis RECP reciprocal REDUP reduplication REFL reflexive REL relativizer (ender) REM remote REP repetitive (refactive) RES resultative RESTR restrictive RETR retrospective RQ requestive RTR retracted tongue root RTRH retracted tongue-root harmony S the sole argument of an intransitive verb SBJV subjunctive SEQ sequential SETi Number of BWB sets SFP sentence-final particle SG singular SH subject honorific SIM simultaneous SIMUL simulative SML similative  SMLF semelfactive SOC sociative (societative) SOL solidarity SOV subject-object-verb SPON spontaneous SS same-subject converb SUB subordinate gerund

SUBJ subjective SUBST substantive SUGS suggestive SUFF suffix SUPER superlative SUPP suppositive T chain terminal marker TAM tense-aspect-modality (1 and 2 refer to   different morphemes) TAME tense, aspect, mood, evidentiality TE translation equivalence TEMP temporal TERM terminative TOP topic TR transitive TRH tongue-root harmony UMRK unmarked UNI classifier of the universal quantifier V verb VAL valency change VBLZ verbalizer VCV vocal-consonant-vocal VEN venitive VERB verb-formant VH vowel harmony VN verbal noun VOC vocative VOL volition(al), voluntative VP verb phrase VS variable subject converb X he/she/it Y Yakut copy

c) Other conventions and categories ad anno Domini ASJP Automatic Similarity Judgment Program bc before Christ bp before the present CS Computer Science CTMC Continuous Time Markov Chain ERC European Research Council FLDH Farming Language Dispersal Hypothesis HGDP Human Genome Diversity Project

ICTL International Conference of Turkic Linguistics mtDNA mitochondrial DNA PR China SNP single nucleotide polymorphism TDK Türk Dil Kurumu (Turkish Language  Association) TMRCA time to the most recent common ancestor WAFL Workshop on Altaic Formal Linguistics yBP years before present

xlii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Abbr eviations d) Sources (A)LD  Linguistic atlas of Dolomitic Ladinian and   neighbouring dialects AP Klaproth 1823 B Böhtlingk 1851 C Castrén 1857

Hae Haenisch 1961 Möl Möllendorf 1892 UTIL  Uyγur tiliniŋ izahliq luγiti ‘An explanatory   dictionary of the Uyghur  language’

xliii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors Kyou-Dong Ahn is Lecturer in the Department of English Linguistics at HanKuk University of Foreign Studies (HUFS) in South Korea. Currently, he is collaborating with the Eurasia3angle research group at Max Planck Institute for the Science of Human History in Jena. His publications include From Space to Grammar (2008, VDM Verlag), “Constructions of Viewpoint Aspect and Grammaticalization of Negative Conditionals in Korean” (2017, Concentric: Studies in Linguistics), “From Honor to disparagement: The pragmaticalization of -tapsiko in Korean” (in press, Journal of Pragmatics). Lina Amal has studied Empirical Linguistics at the Goethe University in Frankfurt (Germany) and investigated possessive constructions in Turkish and Korean as part of her Bachelor’s thesis. She was an exchange student at Ewha Womans University in Seoul in 2016 and participated in field research in Kazakhstan and Kirgizia, organized in the framework of the international cooperative project “Interaction of Turkic languages and cultures in postSoviet Kazakhstan”, in 2015. Her mother tongues are German, French, and Arabic. She speaks fluently English, Korean, and Turkish, and has a good command of Russian, Indonesian, and Swahili. Gregory D. S. Anderson is Director of Living Tongues Institute for Endangered Languages and Research Fellow, University of South Africa. He specializes in typology, historical/ comparative linguistics, and language contact of Native Siberian languages. He has ­published the monographs Auxiliary Verb Constructions in Altai-Sayan Turkic and Language Contact in South Central Siberia, as well as grammar sketches of Xakas and Tuvan, and articles on topics in Turkic linguistics. He has also worked extensively on the Munda language family, Tibeto-Burman languages of northeastern India, Burushaski, various languages of Western/Central Africa, and the South Pacific region in Papua New Guinea and Vanuatu. Natalia Aralova obtained a PhD in Linguistics in 2015 and currently works as a postdoctoral researcher in the CNRS lab “Dynamique du Langage” in Lyon. Between 2009 and 2013 she was involved in a project documenting several Even dialects and now she works on the documentation of Negidal. Her research interests include phonetics and phonology, language change and language contact, and documentation of endangered languages, primarily Northern Tungusic. John R. Bentley is Professor of Japanese at Northern Illinois University. His research interests are centered on historical linguistics, Japanese historiography, and Japanese Native Studies (Kokugaku). He received his doctorate at the University of Hawai’i at Mānoa. He is the author of A Linguistic History of the Forgotten Islands: a Reconstruction of the Protolanguage of the Southern Ryūkyūs (Brill), ABC Dictionary of Old Japanese Phonograms (University of Hawai’i Press), and An Anthology of Kokugaku Scholars: 1690–1868 (EAP, Cornell). He is currently finishing a manuscript on the birth of historiography in early Japan. Ágnes Birtalan is Professor of Mongolian studies at the Department of Mongolian and Inner Asian Studies, ELTE (Eötvös Loránd University) Budapest. She majored in Mongolian Studies, Russian Studies, and History. She teaches numerous subjects at her university to many Hungarian and foreign PhD students, and gives lectures abroad. Her main academic interest is in Mongolian dialects and oral tradition, including shamanism and Buddhicized folk religion. She has carried out field research for over two decades. She is author of

xliv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors several books and many articles (elte.academia.edu). She actively participates in organizing international Mongolian studies events and is the president of the International Association of Mongolian Studies. Václav Blažek studied mathematics and physics (1978–83) in addition to Indo-European and African linguistics (1981–3), both at Charles University in Prague. He accomplished his doctoral study of comparative Indo-European linguistics at Masaryk University, Brno (1990–3) with a dissertation “Zoological terminology of Indo-European languages (mammals).” In 1993–4 and again in 1998 he obtained Humboldt fellowships at the University of Cologne. Since 1995 he has been employed at Masaryk University, in the Department of Linguistics and Baltic Studies of the Faculty of Arts. In 1998 he became an associate professor of comparative Indo-European linguistics with a Habilitationschrift “Numerals,” and in 2003 a professor of comparative Indo-European linguistics at Masaryk University. Allan R. Bomhard is a retired linguist living in Florence, South Carolina, USA. His main areas of interest are distant linguistic relationship and Indo-European comparative linguistics, though he has also made meaningful contributions to Afroasiatic and Transeurasian comparative linguistics. He has published over 80 articles and 14 books, with a concentration on Nostratic comparative linguistics—his most recent work on this subject (2018) is the third edition of A Comprehensive Introduction to Nostratic Comparative Linguistics, with ­special reference to Indo-European (4 volumes, 2,755 pages). Cecil H. Brown is Distinguished Research Professor Emeritus at Northern Illinois University, where he taught anthropology and linguistics for 32 years. His research interests include cognitive anthropology, ethnobiology, historical linguistics, and lexical typ­ ology. Brown has published more than 100 articles and chapters as well as three books, Wittgensteinian Linguistics (1974), Language and Living Things (1984), and Lexical Acculturation in Native American Languages (1999). He is co-founder of the Automated Similarity Judgment Program (ASJP), an online database of lexicons for more than 5,000 distinct languages (https://asjp.clld.org/). Éva Á. Csató is Professor Emeritus in Turkic languages at the Department of Linguistics and Philology, Uppsala University. She studied linguistics and Turcology at the University of Oslo. Her research interests include Turkic linguistics, syntactic typology, contact linguistics, documentation and revitalization of the endagered Karaim language, and the study of the Turkic minority language Kashkay in Iran. She has published over 100 articles and edited several volumes, for example, with Lars Johanson The Turkic Languages (Routledge 1998, 2006). She is on the editorial board of the journal Turkic Languages. Patryk Czerwinski studied Japanese at the University of Warsaw and Kyūshū University. He is currently a PhD student at the Department of General and Comparative Linguistics at the University of Mainz. He is writing a descriptive grammar of Orok (Uilta), a critically endangered Tungusic language, as his dissertation. His primary interests are documentation and description of endangered languages, morphosyntactic typology, and areal typ­ology of North and East Asian languages. Elisabeth M. de Boer is Professor of Japanese Linguistics at the Ruhr Universität Bochum and Principal Investigator in a research project funded by the European Research Council on the possible links between dialect diversification in Japanese and prehistoric migration routes in Japan. Her research concentrates on historical changes in the Japanese tone systems, on which she has published various articles as well as the monograph The historical development of Japanese tone: Part 1. From Proto-Japanese to the modern dialects. Part 2. The introduction and adaptation of the Middle Chinese tones in Japan (2010, Harrassowitz).

xlv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors Alexander T. Francis-Ratte is Assistant Professor of Asian Studies at Furman University in Greenville, South Carolina. His research focuses on the history of the Japanese and Korean languages, exploring the question of Japano-Koreanic common origin. Ilya Gruntov is Senior Researcher in the Mongolian group of the Uralic and Altaic studies department, Institute of Linguistics, Russian Academy of Sciences (Moscow). In 1999 he graduated from the Theoretical and Applied Linguistics faculty of the Russian State University for Humanities. His PhD dissertation (2002) was devoted to the reconstruction of the Proto-Altaic case system. He participated in multiple folklore and linguistic ex­ped­ itions to Mongolian and Turkic peoples and is the author of the website http://altaica.ru, devoted to research on the Mongolic, Turkic, Manchu-Tungusic, Korean, and Japonic languages. His main field of research includes the history of Mongolic languages, semantic typology, and historical morphology. Nataliia Hübler obtained an MA in Language and Variation and German linguistics at the University of Kiel, Germany, in 2016. After working for the project Grambank at the MPI for the Science of Human History in Jena as a student assistant, she is now a PhD student within the project Eurasia3angle at the MPI SHH in Jena, where she is exploring the history of Transeurasian languages on the basis of their structural features by applying phylogenetic methods. Her main domain of interest is the dynamics of change of typo­ logic­al features. Mark James Hudson is a researcher in the Eurasia3angle project based at the Max Planck Institute for the Science of Human History in Jena. He is also a Research Associate of the Institut d’Asie Orientale, ENS de Lyon. He is the author of Ruins of Identity: Ethnogenesis in the Japanese Islands (Hawai‘i University Press, 1999) and has co-edited four books, including Multicultural Japan: Palaeolithic to Postmodern (Cambridge University Press, 1996) and Volume 1 of the forthcoming Cambridge World History of Violence. His research focuses on the archeology of the Japanese Islands and questions relating to the ethnic, linguistic, and environmental history of early Eurasia. Taeho Jang, a Korean scholar, is a faculty member in the Linguistics Department at Payap University, Thailand. Taeho completed his PhD at the Central University for Nationalities, China, his MA at the University of Texas at Arlington, and his undergraduate studies at Seoul National University, Korea. He was also a visiting scholar in the Department of Linguistics at the University of Oregon. He has studied spoken Xibe through his fieldwork in Xinjiang, China (1998–2010), including his PhD work, and has published a Xibe grammar (2008). He has also, since 2008, presented papers in LENCA and SLE conferences. Juha Janhunen is Professor of East Asian Languages and Cultures at the University of Helsinki, Finland, and Honorary Professor at the Inner Mongolia University, China. His research is focused on the languages and ethnic history of northern, central, and eastern Eurasia. He has carried out linguistic and ethnological field work in Siberia, Mongolia, Manchuria, and the Amdo Kuku Nor region. He is the author of Manchuria: An ethnic history (1996, Finno-Ugrian Society), a co-author of New Materials on the Khitan Small Script (2011, Global Oriental), and the editor of several periodicals and collective volumes. Choongwon Jeong is a collaborator in the Eurasia3angle research group at the Max Planck Institute for the Science of Human History in Jena, funded by the European Research Council. He also leads a Research Group, funded by the Max-Planck-Harvard Research Center for the Archaeoscience of the Ancient Mediterranean (MHAAM). He obtained his PhD at the University of Chicago, with a thesis titled The genetic history and adaptations of high altitude East Asians in the Tibetan plateau (2016). His publications include Bronze Age population dynamics and the rise of dairy pastoralism on the eastern Eurasian steppe (2018, PNAS),

xlvi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors “Pleistocene North African genomes link Near Eastern and sub-Saharan African human populations” (2018, Science) and many peer-reviewed journal articles. Lars Johanson, born and educated in Sweden, is Professor of Turcology at the University of Mainz, Germany. He has published widely on descriptive and historical linguistics, especially in the domains of aspect-mood-tense and language contact, focusing mostly on the Turkic language family. He has been instrumental in transforming the field of Turcology, which was traditionally more philologically oriented, into a linguistic discipline. He has made a number of pioneering contributions to general linguistics and language typology. He edits the journal Turkic Languages and the monograph series “Turcologica.” Andrew Joseph (Cornell University) works primarily on Manchu, Jurchen, and the wider Tungusic family, with an emphasis on issues of historical reconstruction. Particular research interests include Manchu and Jurchen philology; the phonological organization of Tungusic languages; diachronic aspects of vowel harmony in Tungusic and beyond; and the investigation of the genetic and areal relationships of the Tungusic languages. Jieun Kiaer is Associate Professor in Korean Language and Linguistics at the University of Oxford. She is interested in pragmatic motivations in syntactic architecture in Korean and other OV languages. Her publications include Pragmatic Syntax (2014, Bloomsbury, UK), Jeju Language: Tales from the Edge of the Korean Peninsula (2014, Lincom Europa, Germany), and The Sounds of Korean (2013, Cambridge University Press), co-authored with Jiyoung Shin (first author) and Jaeun Cha. Seongyeon Ko (PhD in Linguistics, Cornell University, 2012) is Assistant Professor of Korean Linguistics and Director of the Korean Studies Program in the Department of Classical, Middle Eastern, and Asian Languages and Cultures at Queens College of the City University of New York. He is the author of Tongue Root Harmony and Vowel Contrast in Northeast Asian Languages and has published articles on the phonetics, phonology, historical linguistics, typology, and variation of Korean, Mongolic, and Tungusic languages. Jaklin Kornfilt is Professor of Linguistics at Syracuse University (NY, USA). She studied at Heidelberg University as an undergraduate student, on a DAAD (German Academic Exchange) fellowship, and she received her PhD in theoretical linguistics from Harvard University. Her research interests are theoretical syntax and its interface with morph­ ology, theoretically oriented linguistic typology, and Turkish and Turkic. She received a Humboldt research prize in 2010, which enabled her to spend an academic year at the University of Stuttgart. She has published a Turkish Grammar in 1997 (Routledge), two coedited books, two co-edited special journal issues, and numerous chapters in anthologies, and articles in journals, anthologies, and conference proceedings. Evgeniya Korovina is Research Fellow of the Department of Ural-Altaic languages, Institute of Linguistics, Russian Academy of Sciences. She is a member of the project “Digital Dialectologic Atlas of Turkic languages in Russia” led by Anna Dybo. She is also working in the field of Tungusic, Mayan, and Polynesian languages. Tao Li holds a Doctor of Philosophy in Anthropological Archeology (University of Pittsburgh, 2016) and a Doctor of Science in Archeometry (University of Chinese Academy of Sciences, 2010). Between September 2016 and February 2019, he worked as a researcher of the Eurasia3angle Project at the Max Planck Institute for the Science of Human History (Jena, Germany). He is now an Associate Professor at the Department of Archaeology at Wuhan University (Hubei Province, China). His research interests include craft production, household archaeology, and social complexity Andrej Malchukov is Senior Researcher at the St. Petersburg Institute for Linguistic Research (Russian Academy of Sciences), and Invited Professor at the University of Mainz. Apart from descriptive work on Siberian (in particular Tungusic) languages, his main

xlvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors research interests lie in the domain of language typology. He has published extensively on the issues of morphosyntactic typology; in particular, he edited The Oxford Handbook of Case (with Andrew Spencer), Studies in Ditransitive Constructions (with Bernard Comrie and Martin Haspelmath), Competing Motivations in Grammar and Usage (with Brian MacWhinney and Edith Moravcsik), and Valency Classes in the World's Languages (with Bernard Comrie). Olga Mazo is Assistant Professor at the Institute for Oriental and Classical Studies (Russian State University for Humanities) and at the Institute for Oriental and Classical Studies (National Research University Higher School of Economics (HSE), Russia). She gradu­ated from the Russian State University for the Humanities, Theoretical and Applied Linguistics faculty in 1997 and received her PhD in Sino-Tibetan studies (“The Comparative Morphology of Sino-Tibetan languages”) in 2004. Since 2013, she has participated in annual folklore and linguistics expeditions to Mongolia and China. Her main field of research includes Old Chinese studies, the history of Mongolic languages, Chinese folklore and religion studies. Marc Miyake is Research Assistant at the British Museum for Beyond Boundaries: Religion, Region, Language and the State, an interdisciplinary research project funded by the European Research Council. His specialty is the phonological reconstruction of ancient Asian languages. Currently he is writing a monograph covering the script, phonology, morph­ology, syntax, and lexicon of the extinct Pyu language from what is now Myanmar. His publications include Old Japanese: A Phonetic Reconstruction (2003, Routledge) and various articles on Pyu, Chinese, Japanese, Korean, Vietnamese, Tangut, Jurchen, and Khitan. He has been a visiting professor at the University of Oregon and the University of Hawaii following a postdoctoral position at Leiden University. Irina Nevskaya is Researcher and Lecturer at Frankfurt University and Honorary Professor at the Freie Universität Berlin in Germany. She is chief researcher and a member of the Scientific Council at the Institute of Philology, Novosibirsk, Russia. Irina Nevskaya supports several research foundations as an expert consultant and initiator of numerous research projects, such as “The Prospective as a grammatical category” (2013) and “Interaction of Turkic languages and cultures in post-Soviet Kazakhstan” (2013–18). Her publications include two dissertations, four monographs, 15 edited books, and numerous chapters in collective monographs. Her research interests are syntax, morphology, history, and typology of Altaic languages, electronic corpora, language documentation, and sociolinguistics. Yuto Niinaga is Assistant Professor at the National Institute for Japanese Language and Linguistics (NINJAL). He has been researching Yuwan (a dialect of Amami) since 2006 and Kudaka (a dialect of Okinawa) since 2012. He wrote a reference grammar of Yuwan as his PhD dissertation at the University of Tokyo in 2014. His current interest is in writing a reference grammar of Kudaka and making a concise Boasian trilogy (grammar, word list, and text) of 14 dialects of Uken village in Amami-Oshima, the biggest island of Amami. Chao Ning is Postdoctoral Researcher of the Eurasia3angle research group at the Max Planck Institute for the Science of Human History in Jena. He is currently contributing his expertise in genetics to this interdisciplinary project. He received his PhD degree from Jilin University, China. His publications include “Refined phylogenetic structure of an abundant East Asian Y-chromosomal haplogroup O*-M134” (Eur J Hum Genet, 2016), “Ancient mitochondrial genome reveals trace of prehistoric migration in the east Pamir by pastoralists” (  Journal of Human Genetics, 2016) and various peer-reviewed articles. Hans Nugteren is a graduate of the now-defunct comparative Altaic linguistics curriculum of Leiden university. He has worked in language documentation and lexicography projects and in Turcology departments in Leiden, Frankfurt am Main, and Göttingen. Sofia Oskolskaya is Junior Research Fellow at the Department of the Languages of Russia at the Institute for Linguistic Studies, Russian Academy of Sciences (St. Petersburg).

xlviii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors Previously, she had a postdoc position within the context of the Eurasia3angle project at the Max Planck Institute for the Science of Human History in Jena. She obtained her PhD in 2017 from the Institute for Linguistic Studies, RAS, with the thesis “Aspect in Nanai”. Her research interests have a special focus on Tungusic languages, in particular Nanai. Brigitte Pakendorf obtained a PhD in Molecular Anthropology in 2001 and a PhD in Linguistics in 2007. After leading a Max Planck Research Group at the MPI for Evolutionary Anthropology in Leipzig, Germany, from 2007 through 2011, she is now a senior scientist at the CNRS research unit “Dynamique du Langage” in Lyon, France. Her research focuses on the documentation of the Tungusic languages Even and Negidal, on language contact in Siberia (in particular involving the Turkic language Sakha), as well as on the inter­dis­cip­lin­ ary investigation of population history. Martine Robbeets is Research Group Leader at the Max Planck Institute for the Science of Human History in Jena and Honorary Professor in Transeurasian Linguistics at the University of Mainz. Currently, she is leading an interdisciplinary research project on the dispersal of the Transeurasian languages, funded by the European Research Council. Previous affiliations include an Associate Professorship of Japanese Linguistics at Leiden University and visiting scholarships at the Universities of Tokyo, Leuven, and Mainz. Her publications include Is Japanese related to Korean, Tungusic, Mongolic and Turkic? (2005, Harrassowitz), Diachrony of Verb Morphology (2015, de Gruyter) and various edited volumes. Volker Rybatzki is Researcher at the Institute for Asian and African Studies at the University of Helsinki in Finland. He specializes in Middle Mongolian as well as in Moghol and other languages spoken in Afghanistan. He recently contributed a series of lexical studies, “Vocabularies from the middle of the 20th century from Afghanistan,” to Acta Orientalia (2017, 2013). In collaboration with Igor de Rachewiltz, he authored Introduction to Altaic Philology (Brill 2010). Alexander Savelyev obtained his PhD from the Institute of Linguistics, Russian Academy of Sciences (Moscow), in 2015 and is now Research Fellow at the same institution. In 2016– 2019, he had a postdoc position within the Eurasia3angle research project at the Max Planck Institute for the Science of Human History (Jena). His research focuses at the MPISHH were first, cultural reconstruction of the Proto-Turkic language and its Transeurasian connections, and second, verifying the internal structure of the Turkic language family. His current research interests include the historical grammar and dialectology of Chuvash, language contact in the Volga-Kama Basin, and documentation of Siberian Turkic languages. Michal Schwarz graduated at Masaryk University, where he established an Asian Studies program. He annually visits Mongolia, Korea, and Vietnam with projects focused on the history of Asian countries, development of language families, and their contacts in Central and Eastern Asia. Masayoshi “Matt” Shibatani is Deedee McMurtry Professor of Humanities and Linguistics at Rice University in Houston, USA. He is Professor Emeritus at Kobe University, Japan. He was Fellow at the Centre for Linguistic Typology at the Australian National University and the Center for Advanced Study in the Behavioral Sciences in Stanford, California. He served as President of the Linguistic Society of Japan. His publications include The Languages of Japan (1990) and edited books Approaches to Language Typology (1995), and Grammatical Constructions: Form and Meaning (1996). He is now co-editor-in-chief for the 12-volume Handbooks of Japanese Language and Linguistics by de Gruyter. Jiyoung Shin is Professor in Korean Linguistics at Korea University. Her main research interests are Phonetics and Phonology, Spoken Grammar, and the role of prosody in subareas of Linguistics in Korean. Her publications include Understanding Speech Sounds (2014,

xlix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors Hankookmunhwasa, Korea) and The Sounds of Korean (2013, Cambridge University Press), co-authored with Jieun Kiear and Jaeun Cha. Ubong Shin is Assistant Professor in Korean Language and Linguistics at the Jeju University in South Korea. His research focuses on the phonetics and phonology of Jejudo Korean, and he is aiming to build a corpus of Jejudo Korean. His publications include “Production and Perception of Vowels in Jeju dialect: Focussed on the age group 50s” (2016, Journal of The Society of Korean Language and Literature, 77(0), 147–76) and Acoustic properties of vowels and word-initial obstruent in the Jejudo Korean (2015, PhD, Korea University). Ho-min Sohn, BA and MA in linguistics (Seoul National University) and PhD in linguistics (University of Hawaii), is Professor Emeritus of Korean linguistics at the University of Hawaii at Manoa (UHM) and President of the Korean Language Education & Research Center. He was director of the UHM Center for Korean Studies and president of both the American Association of Teachers of Korean and the International Circle of Korean Linguistics. His publications include Topics in Korean language and linguistics (2013, Korea University Press), Korean language in culture and society (2005, University of Hawaii Press), Integrated Korean series, The Korean language (1999, Cambridge University Press), Korean: Descriptive grammar (1994, Routledge), Linguistic expeditions (1986, Hanshin Publishing Company), and Woleaian reference grammar (1975, University of Hawaii Press). Ondřej Srba graduated at Charles University and Masaryk University, where he coestablished the Asian Studies program. He annually visits Mongolia and China with regard to projects focusing on oral history, Western Mongols, and editions of rare Mongolian manuscripts. Eugénie Stapert received her PhD in Linguistics in 2013 from the Max Planck Institute for Evolutionary Anthropology and Leiden University. Currently, she is assistant professor at the Leiden University Centre of Linguistics, where she teaches in the programs of Russian and Eurasian Studies, General Linguistics, and International Studies. Her research concentrates on contact-induced change in Dolgan, language and identity in modern Dolgan society, and she continues the documentation of the Dolgan language. Jan-Olof Svantesson is Professor Emeritus in general linguistics at Lund University, Sweden. In his PhD dissertation he analyzed the phonology and morphology of Kammu, an Austroasiatic language spoken in northern Laos, and he is the main author of an encyclopedic dictionary of this language. He also worked with Mongolian, in particular its phon­ ology, but also with other aspects of the language. His main publications include “Kammu phonology and morphology” (PhD dissertation, Lund 1983), The phonology of Mongolian (Oxford University Press, 2005), and Dictionary of Kammu Yùan language and culture (NIAS Press, Copenhagen, 2014). J. Marshall Unger is Emeritus Professor of Japanese at the Ohio State University, where he chaired the Department of East Asian Languages and Literatures from 1996 to 2004, having previously chaired departments at the University of Hawai’i and the University of Maryland. His research has focused on the history of the Japanese language, the teaching of Japanese as a second language, the computerization of diverse writing systems, script reform in Japan, and Japanese mathematics of the 18th and 19th centuries. He has been a fellow of the Ford Foundation, the Japan Foundation (twice), and the John Simon Guggenheim Foundation. Edward Vajda is Professor of Morphological Theory, Historical Linguistics, Russian Language, and Inner and North Asian studies at the Western Washington University in Bellingham, Washington State. He has directed Western’s Linguistics Program and East Asian Studies Program, and edited the New York-based journal Word. His ongoing research

l

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

The Contributors involves fieldwork with the Ket language. While affiliated with MPI-EVA (Leipzig, Germany), in August 2006 he proposed evidence of a genetic connection between the Yeniseian family and the Na-Dene family of North America. Vajda received the Excellence of Teaching Award (1992) and the Paul J. Olscamp Distinguished Research Award (2011). Milan van Berlo is a lecturer at Leiden University. He is currently conducting research on the history of Japanese and the Transeurasian languages by investigating kinship-term paradigms in order to shed light on the connections between the languages. He is also involved in teaching Japanese language and linguistics courses at Leiden University. Chuan-Chao Wang is a former collaborator of the Eurasia3angle project at the Max Planck Institute for the Science of Human History in Jena. Currently, he is full Professor and Director at the Institute of Anthropology, Xiamen University. He obtained his PhD at Fudan University with a thesis titled “Genetic History of East Asian Populations” (2015). His publications include “Ancient human genome-wide data from a 3000-year interval in the Caucasus corresponds with eco-geographic regions” (2018, Nature Communications) and more than 20 first- or corresponding-authored peer-reviewed papers in Science, Am J Phys Anthropol, J Genet Genomics, Scientific Reports, PLoS One, J Hum Genet, etc. Lindsay J. Whaley is Professor of Classics and Linguistics at Dartmouth College. He specializes in language typology, Tungusic linguistics, and language endangerment and revitalization. His publications include An Introduction to Language Typology (1997, Sage) and, together with Lenore Grenoble, Saving Languages (2006, Cambridge University Press). Recent coedited volumes include (with Pirkko Sukhoinen) On Diversity and Complexity of Languages Spoken in Europe and North and Central Asia (2014, Benjamins), and (with Andrej Malchukov) Recent Advances in Tungusic Linguistics (Harrassowitz). John Whitman is Professor and Chair of the Cornell University Department of Linguistics. His specializations are East Asian linguistics, comparative syntax, language typology, and historical linguistics. Recent publications include Korean: A Linguistic Introduction (2019, Cambridge, with Sungdai Cho), “Topic Prominence” in the Blackwell Companion to Syntax (2017, with Waltraud Paul), and the co-edited volume Ryūkyū shogo to Kodai Nihongo: Nichiryū sogo no saiken ni mukete (Ryukyuan and Premodern Japanese: Toward the Reconstruction of Proto-Japanese-Ryukyuan, 2016, Kurosio). Abdurishid Yakup received his PhD from Minzu University of China in 1996 and Doctorate of Literature from Kyoto University in 2000. He specializes in Central Asian philology, Turkic linguistics, linguistic typology of Altaic languages, field linguistics, and language documentation. Currently he is Head of the Turfan Research Group of Berlin Brandenburg Academy of Sciences and Humanities and Changjiang Scholar Distinguished Professor of Minzu University of China. He also acts as Dean of the School of Minority Languages and Literature of Minzu University of China and Director of the Academy for Research on Minority Languages in China. Yohei Yamada has been since April 2014 a Doctoral program student in the Graduate School of Global Studies, Tokyo University of Foreign Studies, in Tokyo, Japan. In July 2013, he obtained an MA in Literature from the Inner Mongolia University, Inner Mongolia, China. Jaehoon Yeon is Professor of Korean Language and Linguistics at SOAS University of London. He was President of EAKLE (European Association of Korean Language Education) and is currently President of ICKL (International Circle of Koran Linguistics) and ISKS (International Society of Korean Studies). Currently he is leading a research project on “Varieties of Korean: Global, Local, and Individual” funded by the Academy of Korean Studies. His publications include Korean Grammatical Constructions: Their Form and Meaning (2003, Saffron Books, London), Korean: A Comprehensive Grammar (2011/2019, Routledge), and The Handbook of Korean Linguistics (2015, Wiley Blackwell).

li

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions 1  Romanization of Japonic forms Contemporary Japanese The transliteration of Japanese expressions adopted in this work basically follows the Yale system according to Martin (1989) for cited linguistic forms such as word and text ex­amples and the Hepburn system (e.g. Iwasaki 2013) for proper nouns, such as personal names, titles of books and articles, geographical names, and historical periods. Proper nouns used in the descriptive portion of the text are spelt according to the normal convention, thus “toukyou” is spelt as Tokyo in the text.

The essential difference between the Hepburn and Yale systems is that the syllables that are written as tsu, shi, chi, fu, and ji in the Hepburn system appear as tu, ti, hu, zi in the Yale system and that palatalized consonants sha, shu, sho, ja, ju, jo, cha, chu, cho in the Hepburn system appear as sya, syu, sho, zya, zyu, zyo, tya, tyu, tyo in the Yale system. Japanese verbs and verbal adjectives can be distinguished according to two prosodic classes, indicated as A and B. Type A corresponds to a high initial tone, type B to a low initial tone. Following nouns and nominal adjectives accent patterns are indicated by number combinations (1.1, 1.2, 2.1 to 2.5, etc.).

Table 1  Basic moras as found in the kana symbols in Yale romanization Vowels a

i

u

e

o

Consonant + Vowels ka

ki

ku

ke

ko

ga

gi

gu

ge

go

sa

si

su

se

so

za

zi

zu

ze

zo

ta

ti

tu

te

to

da

-

-

de

do

na

ni

nu

ne

no

ha

hi

hu

he

ho

ba

bi

bu

be

bo

pa

pi

pu

pe

po

ma

mi

mu

me

mo

ya

-

yu

-

yo

ra

ri

ru

re

ro

wa

lii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Palatalized consonants + vowels kya

kyu

kyo

gya

gyu

gyo

sya

syu

syo

zya

zyu

zyo

tya

tyu

tyo

nya

nyu

nyo

hya

hyu

hyo

bya

byu

byo

pya

pyu

pyo

mya

myu

myo

rya

ryu

ryo

Moraic nasal n Geminate consonants are represented by a sequence of two identical consonants (e.g. kitte ‘stamp’, matti ‘match’). Diphthongs and long vowels are represented as a sequence of two vowels: aa, ai, au, ae, ao, ia, ii, iu, io, ei, ou, etc. (e.g. sensei ‘teacher’). Syllable boundaries remain unmarked unless marking is necessary for the sake of the argument.

Table 2a  Vowel notations for Old Japanese

Table 2b  Consonant notations for Contemporary and Old Japanese

Old Japanese

Old Japanese

Index

Yale

Neutral

a, e, i, o, u

a, e, i, o, u

A-type

e 1, i 1, o1

ye, yi, wo

B-type

e 2, i 2, o2

iy, ey, o

Unvoiced Voiced Nasal

p, t, k, s b, d, g, z, w, y, r m, n

2  Romanization of Korean forms Contemporary Korean

Old Japanese In Old Japanese, the use of Chinese characters for phonetic value indicated two values for later e, i, o in certain syllables, the so-called kōrui-otsurui ‘A type–B type’ distinctions. Rather than using the Yale notation, which hints at preglided or postglided origins underlying these types, the vowel distinctions are indexed with more neutral subscripts, namely i1 versus i2, e1 versus e2, and o1 versus o2 (e.g. pi1 ‘sun’, pi2 ‘fire’, ti ‘blood’, me1 ‘woman’, me2 ‘eye’, te ‘hand’, ko1 ‘child’, ko2 ‘this’, po ‘ear (of grain/rice)’.

The transliteration conventions for transcribing Korean forms adopted in this work follow Martin (1992: 6–23). Yale romanization is used for cited linguistic forms such as word and text examples, whereas the classical McCune Reishauer system is employed for proper nouns, such as personal names, titles of books and articles, geographical names, and historical periods. For personal and place names used in the descriptive portion of the text, we respect the individualized conventions, and thus the author name Son is spelt as Sohn in the text and the place name Sewul as Seoul. Since the vowels wu and u do not contrast after a bilabial stop (p, ph, pp, m), pwu, phwu, ppwu, and mwu are abbreviated in spelling as pu, phu, ppu, and mu, respectively. Syllable boundaries remain unmarked unless marking is necessary for the sake of the argument.

liii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 3  Hangŭl letters in Yale and McCune Reishauer system romanization Hangŭl

Yale

McCune Reishauer

Consonants p

p, b

ㅃ

ph

p'

pp

pp

t

t, d

th

t'

tt

tt

s

s

ss

ss

c

ch, j

ch

ch'

cc

tch

k

k, g

kh

k'

kk

kk

m

m

n

n

ng

ng

l

l, r

h

h

ㄷ ㅌ ㄸ ㅅ ㅆ ㅈ ㅊ ㅉ ㄱ ㅋ ㄲ ㅁ ㄴ ㅇ ㄹ ㅎ

McCune Reishauer

ㅣ

i

i

ㅔ

wi

wi

ey

e

yey

ye

wey

we

oy

oe

ay

ae

yay

yae

way

wae

u

ŭ

e

ǒ

ye

yǒ

we

wǒ

a

a

ya

ya

wa

wa

wu

u

y(w)u

yu

o

o

yo

yo

uy

ŭi

Vowels

ㅂ ㅍ

Yale

Hangŭl

ㅟ ㅖ ㅞ ㅚ ㅐ ㅒ ㅙ ㅡ ㅓ ㅕ ㅝ ㅏ ㅑ ㅘ ㅜ ㅠ ㅗ ㅛ ㅢ

Middle Korean For Middle Korean the Yale romanization is modified to allow for the representation of unrounded vowels [ə] and [ɨ] by o and u and of rounded [o] and [u] by wo and wu re­spect­ive­ly. The notation ž is used to represent the now-obsolete Middle Korean triangle grapheme ∆. The capitals W and G are used for two other obsolete consonants, for which the phonological interpretations are probably fricative [β]

and [ɣ], respectively. In contrast to modern Hangŭl, the original Middle Korean alphabet used a special symbol to indicate non-automatic reinforcement following /l/, for instance in Korean hal kes ‘things to do’, pronounced with reinforced [k] as opposed to [g] in Korean phal-kwo ‘is selling and’. Reinforcement is represented by a special mor­pho­phon­emic symbol q, e.g. Middle Korean halq kes ‘things to do’. The dots in the Middle Korean words represent the distinctive pitch of the following syllable: one dot (·) for high, two dots (  ̈) for rising, and unmarked syllables are treated as low.

liv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

3  Romanization of Tungusic forms Table 4  Romanization system for Even Novikova 1960

Aralova 2015

Standard orthography

Romanization

Novikova 1960

Aralova 2015

Standard orthography

Romanization

и

i

и, ы

i

т

t

т

t

ӣ и̇

и̇̄

э, ъ э̄ а

и, ы

i:

и, ы

ị

и, ы

ị:

э

e

э

e:

а̄

a:

и

е

ie

æ

ịa

ч

ʧ͡

i:

ǯ

ʤ͡

e

к

k

г

қ

a

а

в

a:

я

ia

j

j g

ɵ

o

ɵ

o

ɣ

ɵ̄

o:

ɵ

o:

h

–

о

ọ

о

ọ

h'

–

ọ:

м

m

о̄ у ӯ

у̇

у̇̄

п б

о

ọ:

у

u

у

u:

л'

p

б

b

л

u:

п

p

ӈ

u

у

ụ:

н'

u:

у

ụ

н

u

р

b

ʤ

к

k

г

g

к

k

в

w s

č

д

k

с

d

ч

g

ie

е

д

d

i

e:

а

a

д

i:

w

с

s

й

j

г

g

х

h

х

h

м

m

н

n

n

н

ɲ

ń

ӈ

ŋ

ŋ

л

l

l

л

l

l

р

r

r

Table 5  Romanization system for Evenki (Khamnigan dialect) Tsumagari 1992

Romanization

Tsumagari 1992

Romanization

Tsumagari 1992

Romanization

i

i

p

p

ɲ

ń

e

e

b

b

ŋ

ŋ

a

a

t

t

l

l

ə

ə

d

d

r

r

o

o

k

k

s

s

u

u

g

g

ʃ

š

ii

i:

c

c

x

h

ee

e:

z

ʣ

w

w

aa

a:

m

m

j

j

əə

ə:

n

n

oo

o:

uu

u:

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 6  Romanization system for Evenki (Eastern dialect) Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

а

a

к

k

ц

c

а̄ б в г

д е

е̄ ё

ё̄

ж з

и ӣ

й

л

a:

м

b

d, ʤ ə, jə ə:, jə: o, jo o:, jo:

о

о̄

п р

э

r

с

ю

s

ю̄

t

я

u

я̄

u:

č s šč y y: ə u, ju u:, ju: a, ja a:, ja:

f

х

j

ь

p

ф

i:

ы̄

o:

ӯ

i

ы

o

у

z

ъ

ŋ

т

ž

щ

n

нг

g

ш

m

н

w

ч

l

h

Table 7  Romanization system for Evenki (Southern dialect) Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

а

a

к

k

ц

c

а̄ б в г

д е

е̄ ё

ё̄

ж з

и ӣ

й

a: b w g d, ʤ ə, jə ə:, jə: o, jo o:, jo: ž z i i: j

л

м н

нг о

о̄

п р с

т

у

ӯ

ф х

l m n ŋ o o: p r s t u u: f h

lvi

Dictionary:

ч

ш

щ ъ

ы ы̄ ь

э

э̄

ю ю̄ я

я̄

č s šč y y: ə ə: u, ju u:, ju: a, ja a:, ja:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 8  Romanization system for Evenki (Northern dialect) Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

Vasilevič 1958

Romanization

а

a

к

k

ц

c

а̄

a:

б

w

г

д

d, ʤ

е

ə, jə

е̄

ə:, jə:

ё

o, jo

ё̄

o:, jo:

ж

с

y

ы̄

y: ə

э̄

s

ə:

ю

t

u, ju

ю̄

u u:

u:, ju:

я

a, ja

я̄

f

х

j

-

э

r

ф

i:

й

р

šč

ь

p

ӯ

i

ӣ

o:

п

s

ы

o

о̄

ш ъ

ŋ

о

č

щ

n

у

z

и

н

т

ž

з

m

нг

g

ч

l

м

b

в

л

a:, ja:

h

Table 9  Romanization system for Hezhe An 1986

Zhang et al. 1989

Romanization

An 1986

Zhang et al. 1989

Romanization

An 1986

Zhang et al. 1989

Romanization

i

i

i

d

d

d

dʐ

ʥ

ʤ

u

u

u

t

t

t

tʂ

ʨ

č

y

y

ü

n

n

n

ʂ

ɕ

č

o

ɔ

o

l

l

l

ʐ

ʥ

ʤ

œ

ə

ə

r

r

r

g

g

g

a

a

a

dz

s

s

k

k

k

ə

ə

ə

ts

ʨ

č

x

x

h

b

b

b

s

s

s

ɢ

g

g

p

p

p

ʥ

ʥ

ʤ

q

k

k

m

m

m

ʨ

ʨ

č

χ

x

h

f

f

f

ɕ

ɕ

č

ŋ

ŋ

ŋ

w

w

w

j

j

j

lvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 10  Romanization system for Jurchen Pevnov 1997

Kane 1989

Romanization

Pevnov 1997

Kane 1989

Romanization

Pevnov 1997

Kane 1989

Romanization

а

a

a

в

w

w

к

k

k

e

ə

m

m

k

k

i

i

m

m

h

h

o

o

t

t

х'

h

h

u

u

d

d

ɣ

g

g

-

ai

s

s

г

g

g

-

əi

š

š

ӈ

ng

ŋ

-

oi

n

n

-

ʾ

-

-

ui

n

n

-

z

z

au

au

l

l

-

ž

ž

oo

ou

l

l

-

c

c

p, f

p, f

r

r

-

dz

ʣ

p, f

p, f

č

č

b

b

ч ǯ

ǰ

ʤ

b

b

j

y

j

э

и о

у

аи эи

ои

уи ау

оу п

п' б

б'

м

м' т

д с

ш н

н' л

л' р

к' х

Table 11  Romanization system for Kur-Urmi Sunik 1958

Romanization

Sunik 1958

Romanization

Sunik 1958

Romanization

и

i

б

b

ӷ

ɣ

ӣ

i:

i

ị

ī

ị:

е

e

е̄

у ӯ

у̇ о

о̄ э

э̄ а

а̄

e: u u: ụ o o: ə

п д

ть ть

дь к г

қ

ф в с

p d

ҳ

ч ӡ

t

м

t

j

d

j

k g q f v s

ə:

з

z

a

j

j

a:

х

h

lviii

Dictionary:

н

нь ӈ л р

χ č ʤ m n ń ŋ l r

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Table 12  Romanization system for Manchu Avrorin 2000

Romanization

Avrorin 2000

Romanization

Avrorin 2000

Romanization

и

i

к

k

м

m

у э

у̇, о̇ о а

о̄

п

п' б

б'

тт т

т' д

д'

к'

u

k'

г

ə

g'

ф

o

šš

ш

p

h'

t

в

w

t'

j

j

d

ч

č

d'

ǯ

ʤ

m' n ń ŋ l l' r c ʣ

қ

h

х'

tt

ӡ

s'

х

b'

ц

s

с'

b

л' р

š

с

p'

л

f'

шш

o:

н' ӈ

f

ф'

a

н

g

г'

u

м'

ӷ

k g

ҳ

h

Table 13  Romanization system for Nanai Standard orthography

Onenko 1980

Romanization

Standard orthography

Onenko 1980

Romanization

и

и

i

б

б

b

ӣ

у

ӯ

э

э̄

о

о̄

а

а̄

е

е̄

ё

ё̄

ю

ю̄

я

я̄

ӣ

у

ӯ э

э̄

о

о̄

а

а̄

е

е̄

ё

ё̄

ю

ю̄

я

я̄

i: u u: ə ə: o o:

п

д т

г (ӷ)

к (қ) х (ҳ)

с

a

в

a:

j

ə / jə

ч

ə: / jə: o / jo o: / jo: u / ju u: / ju: a / ja

м

н

нʾ ӈ

л р

a: / ja:

Dictionary:

п

д т

г

к х

с

в

й ч

м

н н

нг

л р

p d, ʤ t g k h s w j č m n ń ŋ l r

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 14  Romanization system for Nanai (Bikin dialect) Sem 1976

Romanization

Sem 1976

Romanization

Sem 1976

Romanization

а

a

б

b

j

j

d

ц (ч)

c

и̇

о ~ у̇ э

и у

д

ị

т

o

г (ӷ)

ə

к (қ)

i

ф

u

а̄

a:

ǣ

ä:

и̇̄

о̄

ѳ̄ э̄

е̄

ӣ ӯ

с

х (ҳ) в

ị:

ӡ (д')

t

м

g

н

k

н'

f

ӈ

s

л

h w

р

ʣ m n ń ŋ l r

o: ö: ə: e: i: u:

Table 15  Romanization system for Negidal Tsintsius 1982

Romanization

Tsintsius 1982

Romanization

Tsintsius 1982

Romanization

и

i

п

p

м

m

ӣ е

и̇ е̄

э ~ эо а̣̄ ~ о̣̄ а

а̄ о

о̄ у ӯ

i: ə (e) e: ə, ọ ọ: a

б т

д ч з̌

к г

b t d č ʤ k g

н

н' ӈ в с

й х

л

a:

р

o o: u u:

lx

Dictionary:

n ń ŋ w s j h l r

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Table 16  Romanization system for Oroch Avrorin and Lebedeva 1978

Romanization

Avrorin and Lebedeva 1978

Romanization

Avrorin and Lebedeva 1978

Romanization

и

i

п

p

ɣ

ɣ

b

ч

č

t

ǯ

ʤ

d

м

m

ӣ у ӯ

у̇

у̇̄ о

i: u u: ụ ụ: o

б т

д к

k

г

g

в

w

о̄

o:

с

s

ə

ə

j

j

ə̄

ə:

х

h

а

a

а̄

a:

æ

ä:

н

н' ӈ л

n ń ŋ l

р

r

Table 17  Romanization system for Orok Ozolinja and Fedjaeva 2003

Ikegami 1997

Romanization

Ozolinja and Fedjaeva 2003

Ikegami 1997

Romanization

а

a

a

б

b

b

e

e

p

p

i

i

d

d

o

o

t

t

u

u

g

g

ə

ə

k

k

ɵ

ọ

g

g

u, ju

u, ju

k

k

a, ja

a, ja

w

w

aa

a:

s

s

ee

e:

s

s

ii

i:

x

h

oo

o:

ҳ

x

h

uu

u:

j

j

j

əə

ə:

ч

č

č

ɵɵ

ọ:

ǯ

ǰ

ʤ

uu, juu

u:, ju:

м

m

m

aa, jaa

a:, ja:

n

n

ɲ

ń

ŋ

ŋ

l

l

r

r

е

и о у э э

ю я

а̄ е̄

ӣ о̄ ӯ э̄ э̄

ю̄ я̄

п д т

г (ɣ) к ӷ

қ в с

с'

х (h)

н

н' ӈ л р

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Table 18  Romanization system for Oroqen Hu and Ke 1986

Romanization

Hu and Ke 1986

Romanization

Hu and Ke 1986

Romanization

i

i

oo

ọ:

ʧ

č

ii

i:

ͻ

o

ф

f

ɪ

e

ͻͻ

o:

ʃ

š

ɪɪ

e:

əə

ə:

x

h

y

ü

a

a

m

m

u

u

aa

a:

n

n

uu

u:

b

b

ȵ

ń

ʊ

ụ

p

p

ŋ

ŋ

ʊʊ

ụ:

d

d

l

l

ee

ə:

t

t

r

r

ᴇᴇ

e:

g

g

w

w

ə

ə:

k

k

j

j

o

ọ

ʤ

ʤ

Table 19  Romanization system for Solon Hu and Ke 1986

Tsintsius 1997

Romanization

Hu and Ke 1986

Tsintsius 1997

Romanization

i

и~е

i

b

б

b

i:

p

e

d

e:

t

y

g

y:

k

u

ʤ

u:

ʧ

ə:

s

e:

ʃ

e

x

e:

w

ə

j

ə:

m

ọ

n

ọ:

ŋ

o

l

o:

r

ii ɪ ɪɪ u uu ʊ ʊʊ ee ᴇᴇ ε εε ə əə o oo ͻ

ͻͻ a aa

ӣ

и~е е̄

э~ö э̄ ~ ȫ у

ӯ е̄ е̄

и~е е̄

о̇

о̇о̇ о̇

о̇о̇ у̇

у̇у̇ а

аа

a a:

Dictionary:

п д т г

к з̌

ч' с

ш'

х, χ в

й

м н ӈ л р

p d t g h ʤ č s š h w j m n ŋ l r

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 20  Romanization system for Udihe Nikolaeva and Tolskaya 2001

Romanization

Nikolaeva and Tolskaya 2001

Romanization

Nikolaeva and Tolskaya 2001

Romanization

e

ə

b

b

m

m

e:

ə:

p

p

n

n

a

a

d

d

ñ

ń

a:

a:

t

t

ŋ

ŋ

ä

ä

g

g

l

l

ä:

ä:

k

k

w

w

ie

ie

s

s

j

j

ʾo

ʾo

x

h

r

r

ʾe

ʾə

z

ʤ

l

lj

ʾa

ʾa

c

c

ʾä

ʾä

j

Table 21  Romanization system for Olcha Sunik 1985

Romanization

Sunik 1985

Romanization

Sunik 1985

Romanization

и

i

б

b

ҳ

χ

ӣ у ӯ

и̇ (е)

и̇̄ (е̄) э

э̄

у̇ (о) у̇̄ (о̄) а

а̄

i: u u: e e: ə ə: o o: a a:

п д т г

к в

ф с

х ӷ

қ

p

ч

č

d

ǯ

ʤ

t

м

m

g k w f

н

н' ӈ л

n ń ŋ l

s

р

r

h

w

w

ɣ

й

j

q

lxiii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Table 22  Romanization system for Xibe Jang 2008

Romanization

Jang 2008

Romanization

Jang 2008

Romanization

i

i

p

p

f

f

y

ü

b

b

s

s

ɘ

ə

t

t

x

x

u

u

d

d

v

v, w

ɛ

e

k

k

m

m

œ

ö

g

g

n

n

o

o

q

q

ŋ

ŋ

a

a

ɢ

ɢ

r

r

t͡ʂ

č

l

l

d͡ʐ

ʤ

j

j

4  Romanization of Mongolic forms Table 23  Romanization system for Baoan, Bonan (following Hugjiltu 2003) a

e

o

u

i

p

t

ʦ

tr

c

k

b

d

dz

dr

j

g

f

lh

s

sr

sh

x

w

l

z

r

zh

gh

m

n

ny

ng

y

h

Table 24  Romanization system for Buryat Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А  а

a

Л  л

l

X x

x

m

Һ  һ

h

Б  б В  в Г  г

Д  д Е  е

Ё  ё

Ж  ж З  з

И  и

Й  й К  к

b

М  м

w

Н  н

n

g

O o

o

d

Ө  ө

ö

ye/e yo zh z i y k

П  п Р  р С  с

Т  т

У  у

Ү  ү

Ф  ф

p r s t u ü f

Dictionary:

Ц  ц Ч  ч

Ш  ш

Щ  щ Ъ  ъ

Ы  ы Ь  ь

Э  э

Ю  ю Я  я

c ch sh shch " ï ' ê yu/yü ya

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 25  Romanization system for Dagur Daur a

e

o

u

i

aa

ee

oo

uu

ii

p

t

b

d

(f)

š

Table 26  Romanization system for Dongxiang, Santa ­(following Kim 2003) a

e

o

u

i

k

p

t

c

ch

q

k

kh

g

b

d

z

zh

j

g

gh

s

sh

x

h

hh

f

č

l

ǰ s m

m

h

r

n

ng

w

n

y

r l w

y

Table 27  Romanization system for Phags-Pa script Tumurtogoo 2010

Romanization

Tumurtogoo 2010

Romanization

Tumurtogoo 2010

Romanization

a

a

m

m

u

u

b

b

n

n

ŭ

w

č

c

ń

ny

ü

ü

č‘

ch

ŋ

ng

w

w

d

d

o

o

x

x

e

e

ö

ö

y

y

g

g

p

p

z

z

γ

gh

q

q

ž

zh

i

i

r

r

ż

zy

ȷ̌

j

s

s

’

’

k

k

š

sh

l

l

t

t

lxv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 28  Romanization system for Kalmyk Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А  а

a

К  к

k

X x

x

l

Ц  ц

c

Ә  ә

Б  б В  в Г  г

Һ  һ Д  д Е  е Ё  ё

Ж  ж

Җ  җ З  з

И  и

Й  й

ä b w

Л  л

М  м

m

Н  н

n

g

Ң  ң

ŋ

gh

O o

o

d

Ө  ө

ö

e/ye yo zh z'h' z i j

П  п

p

Р  р

r

С  с

s

Т  т

t

У  у

Ч  ч

Ш  ш

Щ  щ Ъ  ъ

Ы  ы Ь  ь

Э  э

Ю  ю

ch sh shch " ï ' ê yu

Я  я

ya

u

Ү  ү

ü

Ф  ф

f

Table 29  Romanization system for Khalkha Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А  а

a

Л  л

l

X x

x

m

Ц  ц

c

Б  б В  в Г  г

Д  д Е  е

Ё  ё

Ж  ж З  з

И  и

Й  й К  к

b

М  м

w

Н  н

n

g

O o

o

d

Ө  ө

ö

ye/yö/e yo zh z i y k

П  п Р  р С  с

Т  т

p r s t

У  у

u

Ф  ф

f

Ү  ү

ü

lxvi

Dictionary:

Ч  ч

Ш  ш

Щ  щ Ъ  ъ

Ы  ы Ь  ь

Э  э

Ю  ю Я  я

ch sh shch " ï ' ê yu/yü ya

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 30  Romanization system for Khamnigan Mongol ­(following Janhunen 2003a)

Table 31  Romanization system for Mangghuer (following Slater 2003a)

u

i

ů

e

u

i

p

t

c

k

p

t

k

kh

b

d

j

g

b

d

g

gh

o

a

s m

n

ng

ch

q

z

zh

j

sh

x

f

s

r

m

n

y

a

h ng

l

h

e

c

l w

o

r

w

y

Table 32  Romanization system for Moghol (following Weiers 2003)

Table 33  Romanization system for Mongghul (following Georg 2003a)

u

i

o

e

å

u

i

o

e

a

p

t

c

k

q

p

t

c

ch

q

k

b

d

j

g

b

d

z

zh

j

g

f

s

sh

x

s

sh

x

h

w

z

zh

gh

m

n

m

l

w

a

h

f l

gh

r

n

ng y

r y

Table 34  Romanization system for Ordos ­(following Georg 2003b)

Table 35  Romanization system for Shira Yugur (following Nugteren 2003)

u

u

ü

i

o

ö

p

t

c

k

q

j

g

gh

b m

ü

i

o

t

c

k

d

j

g

s

sh

n

ö

e

a

b

d dz

l y

a

ts

ng

r

e

s

sh

w

z

zh

m

n

x ng

l r y

h

lxvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 36  Romanization system for Written Mongolian (following Janhunen 2003e)

Table 37  Romanization system for Written Oirat (following Birtalan 2003)

u

i

o

e

v

u

ü

i

p

t

c

k

q

p

t

ty

x

tz b

d

o

g

qh

b

d

m

ä

dy

g j

s

sh

z

zh

h

s

sh

w

z

m

n

ny

l

l

ly

r

r

ry

n

w

a

c

dz f

e

k

ts cz

ö

x gh ng

y

y

5  Romanization of Turkic forms The romanization rules adopted in this volume for the Turkic languages basically follow those used in Johanson and Csató (1998). For the languages that have a standard written form, romanization is provided in comparison to

the standard orthography. Romanization for non-standardized languages is matched with the transcription systems as found in major descriptive works on these languages. In the following tables, Cyrillic letters that are rarely used in recent non-adapted loans from Russian only, such as or , are mostly omitted.

Table 38  Romanization system for Altay (South) Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А  а

a

Й   й

y

 р

r

 с

s

Г 

 г

b g, ɣ

К  к

Р 

t

д

d

М  м

Т   т

Б  б Д 

Дь  дь

d́

Л  л

Н   н

k, ḳ l m n

 е

e

ŋ

Ж  ж

ҥ

ž

O  o

o

z

Ö  ö

ö

i

П  п

p

Е 

З 

з

И  и

lxviii

Dictionary:

С 

У  у Ӱ  ӱ

Ч  ч

Ш  ш

Ы   ы Э  э

u ü č š ï e-

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 39  Romanization system for Azeri Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

A a

А  а

a

L 

Л  л

l

b

M m

ǰ

N 

n

č

O 

o

B b C c Ç  ç

Б  б

Ҹ  ҹ Ч  ч

l

D d

Д  д

d

Ö 

ö

E  e

E 

 e

e

P 

p

Ə  ə

Ә   ә

ä

R  r

Ф  ф

f

S  s Ş  ş

Ғ   ғ

ǵ ɣ

U  u

Х  х

h χ ï

V 

v

i

Y 

y

Ж  ж

ž

Z 

z

ya-

Г 

k ġ

Ff G g Ğ  ğ H h X x I 

ı

İ 

i

J 

j

K k Q q

Ҝ  ҝ

Һ  һ Ы   ы

И  и К  к  г

T  t Ü  ü

М  м Н  н О  о

Ө  ө

П  п Р  р С  с

Ш  ш

Т  т

У  у

Ү  ү

m n o ö p r s š t u ü

В  в

v

 ј

y

З  з

z

Ј 

Я  я

ya-

yu-

Ю  ю

yu-

Table 40  Romanization system for Bashkir Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А а

a

Ҡ ҡ

ḳ

У

u

Б б

В в

Г г

Ғ ғ

Д д

Ҙ ҙ

Е е

З з

И и

Й й

К к

b w, v g ɣ d δ ye, -ez i y k

Л л

М м Н н

Ң ң О о Ө ө

П п Р

С Ҫ

Т

р с ҫ

т

l m n ŋ o ö p r s θ t

Ү

у

ү

ü

Ф ф Һ

х

f χ h

Х Ч

һ ч

Ш ш Ы ы Э Ә

эә

Ю ю Я

я

č š ï eä yuya-

lxix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 41  Romanization system for Chuvash Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А а

a

Н н

n

Х

χ

Ă ă

ə̑o, ə̑

В

v

Е Ĕ

в е ĕ

И и Й й К к

Л л

М м

П п

p

Р р

ye-, -eəo, ə i

r

Т т

t

Ь

o (Viryal dialect),

Э

k

u (Anatri dialect);

l

ụ Ӳ ӳ

m

č š

ъ

s ś

ч

Ш ш

С с Ç ç

У у

y

Ч

х

y

Ы ы э

Ю ю Я

я

ï ́

eyuya-

ü

Table 42  Romanization system for Crimean Tatar Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

А

А

a

N n

Н

n

b

Ñ ñ

ǰ

O o

č

Ö ö

d

P p

e

Q q

ф

f

R r S s

гъ

g ɣ χ

T t

ï

U u

i

Ü ü

ž

V v

k

Y y

l

Z z

m

ya-

а

B b C c Ç ç D d E e F f G g Ğ

ğ

H h I

ı

İ

i

J j K k L l M m

Б

а

б

Дж дж Ч

Д

Э- Ф Г

Гъ Х

Ы И

Ж К

Л

М

ч

д

э-, -ег

х

ы и

ж к

л

м

Ş ş

lxx

Dictionary:

н

Нъ нъ О

о

о-, -ё-

П

п

Къ къ Р

С

Ш Т

У

р с

ш т

у

у-, -юВ

Й З

Я

в

й з

я

ŋ o ö p ḳ r s š t u ü v y z ya-

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 43  Romanization system for Dolgan Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К к

k

С

s

Б В Г

Д Е

а

б в г

д е

Ж ж З

И Й

з

и й

Һ һ

b

Л л

v

Р

y

ү х

ч

Ы ы

p

р

у

Ш ш

ö

П п

i

Ч

o

Ө ө

z

Х

ŋ

О о

ž

т

Ф ф

n

Ӈ ӈ

e

Ү

m

Н н

d

У

l

М м

g

Т

h

с

r

Э

э

t u ü f χ č š ï e-

Table 44  Romanization system for Gagauz Standard orthography (Cyrillic)

Standard orthography (Latin)

Romanization

Standard orthography (Cyrillic)

Standard orthography (Latin)

Romanization

А

A a

a

О

O o

o

Ä ä

ä

Ö ö

ö

B b

b

P p

p

V v

v

R r

r

G g

g

S s

s

D d

d

T t

t

E e, Ye ye

ye-, -e-

U u

u

J j

ž

Ü ü

ü

C c

ǰ

F f

f

Z z

z

H h

h

İ

i

i

Ç ç

č

Y y

y

Ş  ş

š

K k

k, ḳ

I  ı

ï

L l

l

Yu yu

yu-

M m

m

Ya ya

ya-

N n

n

а

Аь аь (Ӓ ӓ) Б В Г

Д Е

б в г

д е

Ж ж

Дж дж (Ӂ ӂ) З

И Й К

Л

М Н

з

и й к

л

м н

о

Оь оь (Ӧ ӧ) П Р

С

Т

У

п р с

т

у

Уь уь (Ӱ ӱ) Ф Х Ч

Ш Ы

ф х

ч

ш ы

Ю ю Я

я

lxxi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 45  Romanization system for Kazakh Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А а

a

К к

k

У

w

Ә ә

ä

Б б

b

В в

v

Г г

g

Ғ ғ

ɣ

Д д

d

Е е

ye-, -e-

Ж ж

ž

З з

z

И и

i

Й й

y

Қ қ

ḳ

Л л

l

М м

m

Н н

n

Ң ң

ŋ

О о

o

Ө ө

ö

П п Р

p

р

r

С с

s

Т т

Ұ Ү

у

ұ

ŭ

ү

ü

Ф ф Һ

х

f χ h

Х Ч

һ ч

Ш ш Ы ы І

Э

t

Я

і

э

я

č š ï ĭ eya-

Table 46  Romanization system for Karachay-Balkar Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

Й

y

Т

t

Б В Г

Гъ Д Е Ё

Ж

а

б в г

гъ д е ё

ж

Дж дж З

И

з

и

b v g ɣ d ye-, -eö ž ǰ z i

К

Къ Л

М Н

Нг

й к

k

къ

ḳ

л

l

м

m

н

нг

n ŋ

(Нъ нъ) О

П Р

С

о

п р с

У У

т

у

у,

w

х

χ, h

(Ў ў, ý) Х Ч

ч

Ш ш Ы ы

o p r s

lxxii

Dictionary:

u

Э

э

Ю ю

č š ï eü

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 47  Romanization system for Karaim Latin orthography (Baskakov et al. 1974)

Cyrillic orthography (Baskakov et al. 1974)

a

а

b c č d

б

ц ч

д

Romanization

Latin orthography (Baskakov et al. 1974)

Cyrillic orthography (Baskakov et al. 1974)

Romanization

a

m

м

m

b

n

ʦ

ṇ

č

o

d

ӧ p

ʒ (Trakai dialect), dz (Halych dialect)

дз

ʣ

ǯ

дж

ǰ

e f

э, -еф

e

s

нъ

n ŋ

о

o

ӧ

ӧ

п

p

р с

r s

f

š (Trakai, Crimean), sz (Halych)

ш

š

t

т

t

ɣ (Trakai, Crimean dialect), h (Halych)

гъ

ɣ

g

г

g

x (Trakai, Crimean), ch (Halych)

х

χ

i

и

i

i ̯ (Trakai), j (Halych, Crimean)

r

н

й

y

k (Trakai, Crimean, Halych), q (Crimean)

к [къ]

k (ḳ)

l

л

l

u

у

u

ü

ÿ

ü

v (Trakai, Crimean), w (Halych)

в

v

y

ы

ï

z ž

з

ж

z ž

lxxiii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 48  Romanization system for Karakalpak Standard orthography (Cyrillic)

Standard orthography (Latin)

Romanization

Standard orthography (Cyrillic)

Standard orthography (Latin)

А

A a

a

О

о

o

Ä ä = A’ a’

ä

O o  (Wo- wo-)

B b

b

Ө

ө

Ö ö = O’ o’

öv

V v

v

G g Ğ ğ = G’ g’

g ɣ

D d

d

ye-, -e-

ye-, -e-

J j

ž

Z z

z

İi=Ii

i

Y y

y

K k

k

Q q

ḳ

L l

l

M m

m

N n

n

Ň ň = N’ n’

ŋ

Ә

Б

В Г

Ғ

Д Е

а ә

б в г

ғ

д е

Ж ж З

з

И и

Й й К

Қ

Л

к

қ

л

М м Н н

Ң ң

Romanization

(Wo’- wo’-) П п Р

С

Т

У Ү

Ў

р с

т

у

ү ў

Ф ф Х

Ҳ Ч

х

ҳ

ч

Ш ш Ы ы Э

э

P p

p

R r

r

S s

s

T t

t

U u

u

Ü ü = U’ u’

ü

W w

w

F f X x

f χ

H h

h

Ç ç = Ch ch

č

Ş ş = Sh sh

š

I ı = I’ i’

ï

E   e

e-

Table 49  Romanization system for Khakas Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

Й

y

Т

t

Б В Г Ғ

Д Е

а

b в

v

ғ

g ɣ

г

д е

Ж ж З

з

И и I i

d -ež z i ĭ

К

Л

М

Н

й

к

л

м

н

Нъ = Ң ң О

Ӧ

П

Р

С

о

ӧ

п

р

с

k, ḳ l m n ŋ o ö p r s

lxxiv

Dictionary:

У Ӱ

Ф Х Ч Ӌ

Ш Ы Э

Я

т

у

u

ӱ

ü

ф х

f χ

ч

č

ӌ

ш ы э

я

ǰ š ï eya-

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 50  Romanization system for Khalaj Doerfer and Tezcan 1980

Romanization

Doerfer and Tezcan 1980

Romanization

Doerfer and Tezcan 1980

Romanization

b

b

ṣ

ṣ

a

a

p

p

ż

ż

e

e

t

t

ṭ

ṭ

ï

ï

s

s

ẓ

ẓ

i

i

ǰ

ǰ

ɣ

ɣ

o

o

č

č

f

f

ö

ö

ḥ

q

ḳ

u

u

x

ḥ χ

k

k

ü

ü

d

d

g

g

ā

a:

z

z

l

l

ī

i:

r

r

m

m

ū

u:

z

z

n

n

ē

e:

ž

ž

w

w

ō

o:

s

s

h

h

š

š

y

y

Table 51  Romanization system for Kirghiz Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К к

k, ḳ

Т

t

Б В Г

Д Е

а

б в г

д е

Ж ж З

з

И и Й й

b v g, ɣ d ye-, -eǰ z i y

Л л

М м Н н

Ң ң О о Ө ө

П п Р

р

С с

l m n ŋ o ö p r s

У Ү

т

у

ү

Ф ф Х Ч

х

ч

Ш ш Ы ы Э

Я

э

я

u ü f χ č š ï e ya-

lxxv

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 52  Romanization system for Kumyk Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К

k

У

u

Б

В

Г

а

б

в

г

Гъ гъ

Гь гь Д

Е

д

е

Ж ж

З

И

Й

з

и

й

b w g ɣ h d e ǰ z i y

к

Къ къ Л

М Н

ḳ

л

l

м

m

н

n

Нг нг О

ŋ

о

o

Оь оь П Р

С

Т

ö

п

p

р

у

Уь уь

ü

Х

ф

f χ

ч

č

Ф

Ч

Ш

Ы

Э

х

ш

ы

э

š ï e-

Ю ю

yu-, yü-

r

с

s

т

t

Table 53  Romanization system for Nogai Standard orthography

Romanization

Standard orthography

Romanization

Standard orthography

Romanization

А

a

К

k, ḳ

У

u

а

(Аь аь) Б В Г

Д Е

Ж З

И Й

б в г

д е

ж з

и й

ä b v, w g d ye-, -ež z i y

Л

М

Н

к

л

м

н

Нъ нъ О

о

Оь оь

П

Р

С

Т

п

р

с

т

l m n ŋ o ö p r s t

lxxvi

Dictionary:

у

Уь уь

ü

Х

ф

f χ

ч

č

Ф

Ч

Ш

Ы

Э

х

ш

ы

э

Ю ю

Я

я

š ï eyu-, yüya-

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 54  Romanization system for Old Turkic Erdal 2004

Romanization

Erdal 2004

Romanization

Erdal 2004

Romanization

p

p

l

l

a

a

b

b

č

č

ï

ï

v

v

y

y

o

o

m

m

ñ

ń

u

u

t

t

š

š

ä

ä

d

d

k

k, ḳ

e

e

n

n

g

g, ɣ

i

i

s

s

ŋ

ŋ

ö

ö

z

z

r

r

ü

ü

: (long vowels)

:

Romanization

Tenišev 1976

Romanization

th

3

č̣

Table 55  Romanization system for Salar Tenišev 1976

Romanization

i

i

Tenišev 1976 t‛

ü

ü

n

n

3‛

č̣h

e

e

j

y

š

š

ö

ö

l

l’

č

č

a

a

ł

l

č‛

čh

o

o

k

s

s

u

u

k k‛

k

s‛

sh

y

ï

ŋ

ŋ

r

r

p p‘

p ph

x

x

c

ʦ

ḳ

c‛

ʦh

m

m

q q‛

ç

ś

v

v

f

f

x̧

ḳh χ

h

h

ʒ ʒ‛

ǰ ǰh

t

t

h

lxxvii

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 56  Romanization system for Saryg-Yugur Tenišev and Todaeva 1966

Romanization

Tenišev and Todaeva 1966

Romanization

Tenišev and Todaeva 1966

Romanization

а

a

’т‛

th

š

e

к

ш

k

ï

’к‛

kh

е

ы

ш"

šh

тш

č̣

’т ‛ ш

č̣h

i

i

k

ḳ

o

o

’k‛

ḳh

ö

ö

(ф)

f

у

u

в

м

ÿ

ü

(х)

v χ

п

p

й

y

’п‛

p

s

т

с

ң

t

’с‛

s

р

r

h

тш‛‛

č

’тш‛‛‛

h

čh m

н

n ŋ

л

l

Table 57  Romanization system for Shor Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К к

k

Т

t

Б В Г Ғ

Д Е

а

б в г ғ

д е

Ж ж З

з

И и Й й

b v g ɣ d ye-, -ež z

Қ қ

Л л

М м Н н

Ң ң О о Ӧ ӧ

П п

ḳ l m n ŋ o ö p

i

Р

р

r

y

C c

s

lxxviii

Dictionary:

У Ӱ

Х Ч

т

у

u

ӱ х

ü χ

ч

č

Ш ш Ы ы Э

Я

э

я

š ï eya-

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 58  Romanization system for Tatar (Kazan) Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К

k, ḳ

У

u, w

Ә

Б В Г

Д Е

а ә

б в г

д е

Ж ж

Җ җ З

з

И и Й й

ä b v, w g, ɣ d ye-, -ež ǰ z i

Л

к

л

l

М м

m

Н н

n

Ң ң О Ө

ŋ

о

o

ө

ö

П п Р

С

Т

p

р

r

с

s

т

Ү

у

ү

ü, w

Ф ф Һ

х

f χ h

Х Ч

һ ч

Ш ш Ы ы Э

Я

š ï

э

e-

я

yu-, yüya-, yä-

Ю ю

t

č

y

Table 59  Romanization system for Tofa Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

Ӄ

ḳ

Ү

ü

Б В Г Ғ

Д Е

а

б в г ғ

д е

Ж ж З

з

b v g ɣ d e ž z

Һ

Л

ӈ

О Ө Р

j

Т

k

У

С

h l m n ŋ

о ө

П п

Й й к

л

Н н

i

К

һ

М м

И и i

ĭ

ӄ

р с

o ö p r s

т

t

у

u

ү

Ф ф Х Ч Ҷ

х

ч

ҷ

Ш ш Ы ы Э ә

э

ъ

χ č ǰ š ï e ä

Ю ю Я

f

я

yuyaˁ

lxxix

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 60  Romanization system for Turkish Standard orthography (Latin)

Romanization

Standard orthography (Latin)

Romanization

Standard orthography (Latin)

Romanization

A a

a

I

ı

ï

R r

r

B b

b

İ

i

i

S s

s

C c

ǰ

J j

ž

Ş ş

š

Ç ç

č

K k

k

T t

t

D d

d

L l

l

U u

u

E e

e

M m

m

Ü ü

ü

F f

f

N n

n

V v

v

G g

g

O o

o

Y y

y

Ğ ğ

ɣ

Ö ö

ö

Z z

z

H h

h

P p

p

Table 61  Romanization system for Turkmen Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

A a

А

a, a:

N n

Н н

n

b

Ň ň

č

O o

d

Ö ö

e

P p

ä, ä:

R r

Ф ф

f

S s Ş

Х

g, ɣ χ i, i:

U u

ǰ

Ü ü

ž

W w

k, ḳ

Y y

l

Ý ý

m

Z z

B b Ç ç D d Е

е

Ä ä F f G g H h I i J j Ž

ž

K k L l M m

Б

Ч

Д

а

б

ч

д

ЕеЭэ Ә Г

ә г

х

И и

Җ җ Ж ж К

Л

к

л

М м

ş

T t

lxxx

Dictionary:

Ң ң О

Ө

о

ө

П п Р

С

р с

Ш ш

Т

У Ү В

т

у

ү в

Ы ы Й й З

з

ŋ o, o: ö, ö: p r s š t u, u: ü, ü: w ï, ï: y z

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 62  Romanization system for Tuvan Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

Л

l

У

u

Б В Г

Д Е

а

б в г

д е

Ж ж З

з

И и Й й К

к

л

М м

b

m

Н н

v

n

Ң ң

g, ɣ

О

d

Ө

e

ŋ

о

o

ө

ö

П п

ž

Р

z

С

i

Т

y

p

р

r

с

s

т

Ү

у

ү

Ф ф Х Ч

х

ч

Ш ш Ы ы Э

э

Ю ю

t

Я

я

ü f χ č š ï e yuya-

k, ḳ

Table 63  Romanization system for Uzbek Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

A a

А

a, ä

R r

Р

r

b

S s

d

T t

e

U u

f

V v

g

X x

h

Y y

i

Z z

ǰ, ž

Oʻ oʻ

k

Gʻ gʻ

l

Sh sh

m

Ch ch

n

ng

ɔ

ʼ

p

ye-

B b D d E e F f G g H h I i J j K k L l M m N n O o P p Q q

Б

Д Э

а

б

д

э-, -e-

Ф ф Г

Ҳ

г

ҳ

И и

Ж ж К

Л

к

л

М м Н н О

о

П п Қ

қ

С

Т

У В

Х

Й З

Ў Ғ

Ш Ч

р с

т

у в

х

й з

ў ғ

ш ч

Нг нг Ъ е-

ъ

s t u v, w χ y z ọ ɣ š č ŋ ʔ ye-

ḳ

lxxxi

Dictionary:

OUP UNCORRECTED PROOF – , 26/05/20, SPi

Romanization Conventions Table 64  Romanization system for Uyghur (Modern) Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

Standard orthography (Latin)

Standard orthography (Cyrillic)

Romanization

A a

A a

a

Q q

ḳ

E e

Ə

Қ қ

ə

ä

K k

б

b

B b

Б

П п

-ng

p

G g

t

L l

Җ җ

ǰ

M m

Ch ch

Ч

N n

X x

X x

č χ

D d

Д

d

O o

r

U u

z

Ö ö

ž

Ü ü

s

W w

š

P p T t J j

R r Z z J j/Zh zh S s Sh sh Gh gh F f

Т

Р З

т

ч

д

р з

Ж ж С

с

Ш ш Ғ

ғ

Ф ф

H h

К к

Ң ң Г

г

Л л

М м Н н Һ һ О о

У у

Ө ө

Ү ү

k ŋ g l m n h o u ö ü v, w

É é

В в E e

ɣ

I i

И и

i

f

Y y

Й й

y

e

Table 65  Romanization system for Yakut Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

Standard orthography (Cyrillic)

Romanization

А

a

К

k

Һ

h

Б В Г

Ҕ

Д

а

б в г

ҕ

д

Дь дь Е

Ж И Й

е

ж и й

b v g ɣ d ʤ ye-, -ež i y, ỹ

Л

М Н

Ҥ

к

л

м н

ҥ

Нь нь О Ө

П Р

С

о ө

п р с

l m n ŋ ń o ö p r s

lxxxii

Dictionary:

Т

У Ү

һ т

у

ү

Ф ф Х Ч

х

ч

Ш ш Ы ы Э

э

t u ü f χ č š ï e

OUP CORRECTED PROOF – FINAL, 28/05/20, SPi

'OJ

(YHQ sakyo(o), sakyu(u), as well as sakee and its variants in various areas across Japan, e.g. sakyoo in the vast Chūgoku region and some parts of Shikoku, as well as Shizuoka and Saitama and sakee in parts of Saitama and Chiba. The most extreme case of the coalescence of the ac­cusa­ tive particle is found in Hachijō, where any of a variety of the final element of the host noun phrase combines with wo, resulting in varied object forms, as summarized by Kaneda (2001: 39). Below are a few examples from his summary.19 (46)  Hachijō object forms . . . e: wo > . . . e:yo (mae wo > me:yo ‘the front.acc’) . . . i wo > . . . yo (umi wo > umyo ‘ocean.acc’) . . . e wo > . . . ei (sake wo > sakei ‘wine.acc’) . . . a wo > . . . o: (na wo > no: ‘name.acc’) . . . o wo > . . . ou (mono wo> monou ‘thing.acc’) . . . u wo > . . . u: (mizu: ‘water.acc’) As in the rough Tōkyō speech form koryaa ‘this.TOP’ (< kore wa), the topic wa merges with the final sound of the host noun phrase, yielding forms such as sakyaa ‘wine.TOP’ (< sake wa) all over the Chūgoku region, in parts of Shikoku, the central region of Honshū, including Shizuoka, Kanagawa, Yamanashi, Nagano, Saitama, and Tochigi. Along with wa, the nominative ga also undergoes a coalescence, resulting in forms such as amya(a) and amɛ(ɛ); e.g. amya(a) in the Hokuriku region of Toyama and Ishikawa Prefectures as well as Awaji Island in  the Inland Sea and the neighboring parts of Ōsaka and 19   Kaneda (2001: 38) entertains the possibility that wo may have changed to yo first and that the coalescence patterns in (46) result from the merger of the final vowels with yo.

179

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani Wakayama Prefectures. In some areas the ga and the wa coalescences converge, obliterating all the precious wa/ga contrasts over which Japanese specialists, Japanese and non-Japanese alike, have racked their brains. Observe the data from Awaji Island in (47). (47) Awajishima dialect (Mihara district; courtesy of Miho Mano) a. Amya   hunyoru. (Standard Japanese: Ame ga hutte iru.) rain.nom fall.prog ‘Rain is falling.’ b. Amya   hurehendzo. (Standard Japanese: Ame wa huranai zo.) rain.top fall.neg.fp ‘As for rain, (it’s) not falling.’ The Kesennuma dialect in Miyagi Prefecture is unique in that a single set of particles a and g̃a marks either the ­subject or the topic, rendering the following sentence ambiguous between the subject and the topic interpretation: Omai g̃a kadaru (you NOM/TOP speak) ‘You speak’ (Yamaura 1986: 78, 79).

12.4.1.6  Nominalization particle Nominalization particles, known as Juntai joshi, are highly interesting both descriptively and theoretically for their dialectal variation and what they tell us about the nature of grammatical nominalizations. Contrary to widely held belief (Horie (2008), Serafim and Shinzato (2009), Frellesvig (2010), Yap, Grunow-Hårsta, and Wrona (2011)), a nominalization particle such as the Standard Japanese form no, as seen in (48b) below, does not nominalize. (48)  Standard Japanese a. Kaasan ga sakana mother nom fish ‘Mother grilled fish.’

o acc

yaita. grilled

b. [Kaasan ga   Ø yaita]NMLZ no  wa kono mother  nom grilled    npm  top this sakana  da. fish   cop ‘What Mother grilled is this fish.’ c. [Kaasan ga  Ø yaita]NMLZ  sakana wa mother  nom    grilled    fish    top kore  da. this  cop ‘The fish that Mother grilled is this.’ What is nominalized here is marked by the adnominal form, yaita in (48b) and (48c), which is identical to the conclusive

180

form in Modern Japanese, as ­discussed earlier. In Classical Japanese prior to the conclusive/adnominal merger, the adnominal (nominalized) form yaita would have been yakitaru contrasting with the conclusive yakitari, and the adnominal would stand as a nom­in­al­iza­tion by itself without a nominalization particle in a context such as the one for (48b). Notice that the relevant adnominal nom­in­al­iza­tion structure is not marked by no when it performs a modification function, as in (48c). The Standard Japanese form no and its counterparts in other modern mainland dialects (as well as Ryūkyūan; see Shibatani and Shigeno 2013) are a later historical development, arising as a marker of an NP use of adnominal grammatical nom­in­al­iza­tions—indicating that the nominalization structures heading an NP are referential, i.e. they refer to a discourse entity.20 It is easy to see the difference between a nominalizer (e.g. a  nominalizing particle) and an NP-use marker when the Japanese pattern such as (48) (or the comparable one in Korean involving the marker kes) is compared with languages having a true nominalizing par­ticle such as the Mandarin Chinese de. A nominalizer, just like the ad­nom­ inal endings in Classical Japanese (and Korean), occurs in all the contexts in which nominalized structures are used, whether they head an NP or function as a modifier, as the distribution of the nominalizing particle de shows in (49). (49)  Mandarin Chinese a. Māmā kǎo yú. mother grill fish ‘Mother grilled fish.’ b. [Māmā kǎo Ø de]NMLZ shì nmlz cop mother grill ‘What Mother grilled is fish.’

yú. fish

c. [[Māmā kǎo Ø de]NMLZ yú] shì zhège. nmlz fish cop this.clf mother grill ‘The fish that Mother cooked is this.’ The true function of the Juntai joshi as a maker of the NP use of adnominal grammatical nominalizations is clearly seen when Standard Japanese forms in (48) are compared with those in other dialects in which a comparable nom­in­al­iza­ tion particle has not (fully) developed. One of the latter is 20   This is also true of the Korean nominalization particle kes, widely believed to be a noun or a nominalizer in the Korean literature—kes, as opposed to Japanese noun mono ‘thing’, never functions as a noun in Modern Korean, and there is no evidence that it was ever a noun in an earlier form of Korean. Shibatani’s work on nominalization distinguishes “denotation” and “reference.” The former refers to the relationship between grammatical units and concepts, while the latter to the relationship between gram­mat­ ical units and entities in the world. In a modification context, nouns and grammatical nominalizations (e.g. compounds and relative clauses) have denotations but not references, whereas those heading an NP have both denotation and reference.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects the Izumo dialect in Shimane Prefecture in western Japan, which has undergone the conclusive/adnominal merger of verbal forms but has not developed a nominalization particle.21 (50)  Izumo dialect (field notes, May 22 2011; July 24 2011) a. Okaka ga sakana  yaita. mother nom fish    grilled ‘Mother grilled fish.’ b. [Okaka ga mother nom da.

Ø yaita]NMLZ wa grilled top

kono sakana this fish

cop

‘What Mother grilled is this fish.’ c. [[Okaka ga Ø yaita]NMLZ sakana]  wa grilled fish    top mother nom   kore da. cop this ‘The fish that Mother grilled is this.’ Comparison of (48b) and (50b) makes it clear that the par­ ticle no in the Standard Japanese form has nothing directly to do with nominalization per se. That yaita ‘grilled one’ in the Izumo dialect is a nominal by itself can be shown by the fact that it can be modified like any ordinary noun by a copu­la­tive genitive modifier. Compare (51). (51) a. [Kono [[sakana no] hurai]] o gosinahai. gen fry acc give.me.hon.imp this fish ‘Give me this fish fry.’ b. [Kono [[sakana no] yaita]]   o gen grilled.one  acc this fish gosinahai. give.me.hon.omp (Lit.) ‘Give me this grilled one of fish.’ Cf. Standard Japanese [Kono [[sakana no] yaita]] no o kudasai. Map 16 of GAJD shows that besides the Izumo area, Aichi, Nagano, and Yamanashi Prefectures have spots lacking a nominalization particle. Recall that Toshima in the Izu Archipelago is one of those preserving some archaic Eastern Old Japanese traits, including the adnominal ending -o (as opposed to COJ -u). The same map shows that Toshima lacks a nominalization particle. Hirayama (1965), however, points out that younger speakers have developed the marker 21   Many speakers naturally could use the Standard Japanese no in (50b), but they would point out that comparable forms without no are perfect Izumo expressions.

nga [ŋga] (a combination of n (< no) and ga) used optionally. Hachijō is similar except that it has developed an optional NP-use marker me (see Section 12.5.2).22 The NP-use marker no of Standard Japanese and of many modern central dialects is related to the COJ genitive no.23 The genitive source is confirmed by the fact that some other dialects have chosen the other COJ genitive ga as an NP-use marker. This use of ga and its variants (e.g. nga [ŋa], a) is a trait of Ishikawa, Toyama, and Niigata Prefectures of the Hokuriku region. Both no and ga dialects may combine these particles with another, typically the shortened form of no, -n, giving doubly marked forms in the sequence of no-n in Ōsaka and Kyōto, and ga-n in Niigata (and the above-mentioned Toshima form n-ga), as below. (52)  Ōsaka dialect a. Wasi wa [[Ø wakai] no/n/non] ga een top young npm nom good I ya. cop

‘I am good with a young one./I like a young one (woman).’ b. [Anta  iku]  no/n/non  ka? you  go  npm    q ‘Are you going?’ (53) Minami-Uonuma dialect (Niigata; field notes May 24 2013) a. [[Ø ako  de  asonderu] ga/gan] ga  ora     there  loc playing  nmp   nom  I ko   date. child  cop ‘The one playing there is my child.’ b. [[Kodomo ga  miti  de   asonderu] ga/gan] child    nom  street  loc playing   npm   mita  te. saw  fp ‘(I) saw the child play in the street.’ Interestingly, ga, but not gan, is also used in the Kōchi Prefecture (as well as some parts of Ehime Prefecture) in Shikoku facing the Pacific Ocean, far from the Hokuriku region facing the Japan Sea. The following examples in (54) are from the Nakatosa town in Kōchi. 22   Akiyamagō seems to have borrowed one of the NP-use markers -n from the neighboring Niigata dialects. The Miyako language Irabu Ryūkyūan also allows unmarked grammatical nominalizations in NP-use (Masanori Shimoji pers.comm.). 23   Shibatani (2017) shows that this connection is indirect in that the NP-use marker no for verbal-based nominalizations developed out of the NP-use marker no for nominal-based nominalizations.

181

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani (54)  Kōchi dialect (field notes December 19, 20, 2013) a. [[Ø  asoko  ni  tacchuu]  ga]  ga  uchi   there loc standing npm  nom we   no  musuko  yo. gen son    cop ‘The one who is standing in the street is our son.’ b. [[Kodomo ga  michi  de  asobiyuu] ga] nom street loc playing   npm child      o   mita. acc saw ‘I saw a child playing in the street.’ The Kyūshū dialects are characterized by their use of to and tu [tsu], which may be related to the archaic genitive tu [twu], seen in ma-tu-ge (eye gen hair) ‘eyelash’, oki tu sima ­(offing gen island) ‘the island in the offing’. The following examples in (55) and (56) are Yatsushiro dialect forms, in Kumamoto Prefecture. (55) Yatsushiro dialect (Nisshin-chō, Yatsushiro-city; field notes October 3 2010) Argument nominalization in NP-use a. [Toochan  ga  mai    asa     yomu] Daddy    nom every morning read.prs to/*tsu  wa  sinbun     tai. nmp   top newspaper cop ‘What Daddy reads every morning is a newspaper.’ b. [Toochan ga  yonda]  ?to/tsu  wa  sinbun.tai. Daddy  nom read.pst  nmp   top  newspaper ‘What Daddy read was a newspaper. (56)  Event nominalization in NP-use a. [Ano ko    ga   naku]  to/*tsu   wa that child nom cry.prs  npm/npm  top hara   hetteru   ken    tai. stomach decrease  because cop ‘(The reason) why the child is crying is because he is hungry.’ b. [Ano ko   ga  naita]    ?to  wa/tsaa (< tsu wa) that  child  nom  cry/pst  npm  top/npm.top hara      ga  hetta     ken    bai. stomach  nom decreased because cop ‘(The reason) why the child cried was because he was hungry.’ Serafim and Shinzato (2009) entertain the possibility that to and tu are related to the Ryūkyūan nominalization ­particle si, but perhaps much closer relatives are so and ho in Yamaguchi Prefecture, the latter of which is more prevalent in and around the western city of Shimonoseki,

182

across the Kanmon Strait from Kyūshū, as in the following examples in (57). (57)  Yamaguchi dialect (courtesy of Takanori Hirano) Yamaguchi City, Miho a. [Hanako ga  Ø nottyoru]   so  wa huri Hanako  nom  riding   npm  top old   kee,    kaesan     yo. because change.do.pol  fp ‘Because the one (the car) Hanako is riding in is old, change (it). Shinonoseki City b. Moti   wa [anko ga   Ø haittyoru] ho rice.cake  top paste nom   included  npm   ga   ee. nom good ‘As for rice cake, the one in which sweet bean pase is included is good.’ Lastly, there is an NP-use marker unrelated to a genitive par­ ticle. The case in point is yatu and its variants (yazu, azu, ezu, etc.) widely spread throughout Akita and Yamagata Prefec­ tures, as well as some northwestern areas in Aomori Prefecture. Yatu is likely related to the pronoun yatu, a derogative form denoting a male person. As a matter of fact, its grammaticalized version as an NP-use marker is observed in Standard Japanese as well, where it may also mark an argument nom­ in­al­iza­tion denoting a non-human, as in example (58). (58)  Standard Japanese a. kore ga  [kinoo  Ø  katta]   no/yatu  da. this  nom yesterday bought npm   cop ‘This is the one that I bought yesterday.’ b. [Taroo ga  kita]  no/*yatu o   wasureteta. nom  came  npm     acc forgotten Taro   ‘I have been forgetting that Taro had come.’ The ungrammatical form (58b) indicates that yatu cannot mark an event nominalization in Standard Japanese, showing that a developing NP-use marker (  Juntai joshi) flags an argument nominalization in NP use first. In the case of the dialects in Akita, yatu has become a full-fledged NP-use marker marking both argument and event nominalizations, as observed in (59). (59)  Akita dialect (Tsunodate; field notes June 4 2014) a. [Ø kokoni aru]   yazi nanda  ka?    here   exist  nmp what.cop  q ‘What is the one that is here?’ b. [Taroo kuru]  yazi matte   ra. Taro  come  npm  waiting  exist ‘I am waiting for Taro’s arrival.’

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects Owing to the derogative nature of the etymological form yatu, speakers tend to avoid using the particle yazi in reference to people. Note, again, that yazi and all the other Juntai joshi discussed above are NP-use markers and that they do not appear in the modification context; cf. (59a) and [[Ø kokoni aru] hako] nanda ka? ‘What is the box which is here?’.

12.4.2  Conclusive and adnominal forms In this subsection, I shall examine the patterns of the conclusive/adnominal merger and split.

12.4.2.1  Conclusive and adnominal merger The historical merger of the conclusive ending with the adnominal ending has had a profound effect on grammarians dealing with Modern Japanese, who have typically treated what I have called adnominal grammatical nominalizations as clauses or even sentences (largely because they contain verbal forms identical with conclusive forms). Our position is that adnominal grammatical nominalizations have been nominal (rather than clausal/sentential) all through the history of Japanese, and their forms have not substantially changed since the Old Japanese period; it is the conclusive form that has changed, being replaced by the adnominal form. Essentially this is the position of Japanese school grammar, which provides two distinct columns of Shūshikei (conclusive form) and Rentaikei (adnominal form) in the conjugation chart, even though the forms are identical.24 When a merger of this type occurs, it is always the ad­nom­ inal/nominalized form that wins because new sentences arise from the insubordination of adnominal event nom­in­ al­iza­tions, as described earlier. This seems to have happened in all Japonic languages, including Ryūkyūan. When the conclusive/adnominal merger occurs in some other ways, the winner can be either form. The conclusive form of the Standard Japanese copula da that supports predicative nominal adjectives (Keiyō dōshi) derives from the verbal complex de aru (< ni=te ari (dat=con exist)). Its adnominal/nominalized form na, on the other hand, derives from naru (< ni aru (dat exist)). In Standard Japanese and many modern central dialects, the copula is the only verbal form that maintains the formal difference between the conclusive form (Hana wa kirei da (flower top 24   The most likely motivation for school grammar to posit two separate conclusive and adnominal columns for Modern Japanese is to make clear the correspondence patterns between the Modern Japanese conjugation chart and the Old Japanese counterpart, which showed a formal distinction between the two. The biggest issue with the school grammar conjugation paradigm is, of course, the term Rentaikei, which is a misnomer in view of the fact that the Rentai (adnominal) forms are used without a modified head nominal (in our NP-use of grammatical nominalizations).

beautiful cop) ‘The flower is beautiful’) and the adnominal/ nominalized form (kirei na hana (beautiful cop.adnz flower) ‘a  beautiful flower’, kirei na (no) (beautiful cop.adnz (npm) ‘a beautiful one/one that is beautiful’). Some dialects, however, have merged the two, and two directions of merger are seen. In a wide area stretching across the Chūgoku region, from Okayama to Yamaguchi Prefectures, and many areas in Shikoku Island, the adnominal na is used as a conclusive form. It may vary from one dialect to another how natural it is to end a sentence with na without some final particle, but the Hiroshima dialect displays the following pattern (see 60). (60)  Hiroshima dialect (courtesy of Kumiko Sakoda) a. Ano  hito   wa genki   na     nee/noo. that person top vigorous cop.concl  fp/fp ‘That person is vigorous, isn’t he?’ b. [genki   na]   hito vigorous  cop.adnz person ‘a vigorous person’ Apparently, it sounds odd for (60a) to end with na without a final particle. Nevertheless, notice that the Standard Japanese counterpart is genki da nee (vigorous cop.concl fp) ‘(he) is ­rigorous, isn’t he?’ with the conclusive form. In contradistinction to such a western Japanese pattern of the da/na merger, the winner is da in the dialects of Akita, Aomori, and some northern areas in Yamagata, where some areas may use either na or da in the modification environment, as in the Honjō dialect in southwestern Akita. (61)  Honjō dialect (Yurihonjō City, Akita; Hidaka 2014) a. Kono  heya  naba  sizuga da. this  room  top  quiet  cop.concl ‘This room is quiet.’ b. [Sizuga da/na]     heya   sa quiet   cop.adnz/cop.adnz room to igo.25 go.hort ‘Let’s go to a quiet room.’ In the case of the copula da supporting a nominal predicate, the adnominal form is da in Yurihonjō City, where Standard Japanese would have the copulative genitive no, as in gakusei no Taroo ‘Taro, who is a student’, while the possessive geni­ tive is no, as in Standard Japanese. 25   In Akita City both conclusive and adnominal forms appear to be the prenasalized nda. E.g. Kogo ndaba sizuga nda ‘This place is quiet’; [sizuga nda] basyo ‘a quiet place’. (http://www.akitaben.com/category4/entry110.html).

183

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani (62) a. Taroo naba  gakusei  da. Taro  top  student  cop.concl ‘Taro is a student.’ b. [gakusei da]    Taroo student  cop.adnz   Taro ‘Taro, who is a student’ c. [bego no]  cow  gen  ‘cow’s tail’

sippo tail

12.4.2.2  Restoration of the conclusive/adnominal contrast The school grammar conjugation chart recognizing separate conclusive and adnominal verbal forms, despite their formal identity, may be justified by the fact that some dialects restore a formal distinction between the two, overtly expressing the important functional difference between the two. One such dialect is Hachijō, to be discussed fully in the next section. The conclusive form of Modern Hachijō is based on the adnominal form and a final particle, either wa or dya. (63)  Hachijō (field notes August 19–22, 2010) a. *Sinbun   o   yomo. newspaper  acc  read ‘(I) read a newspaper.’ b. Sinbun   o   yomo.wa. newspaper  acc   read.concl ‘I read a newspaper.’ c. [yomo]  sinbun read  newspaper ‘a newspaper that (I) read’ As indicated by (63a), the adnominal form alone cannot conclude a sentence in Modern Hachijō. The likelihood that this dialect once had identical conclusive and adnominal forms and that an adnominal form such as the one in (63a) was able to conclude a sentence is indicated by the fact about the other two locales where the EOJ adnominal -o ending is preserved. Both Akiyamagō in the border area of Niigata and Nagono, and Toshima Island north of Hachijō Island in the Izu Archipelago, allow the -o adnominal form (e.g. the Toshima form Ame ga huro (rain fall.adnz/concl ‘Rain falls/It rains’, Hirayama 1965: 56) to conclude a sentence, along with the Akiyamagō -(r)u ending (e.g. yomu ‘read’), which may be either the earlier conclusive form from EOC or a new borrowing from Standard Japanese (see, for example, Map 67 of GAJD). Similar restorations of the formal contrast between conclusive and adnominal forms are seen among Ryūkyūan languages, especially in Northern and Southern Yaeyama Ryūkyūans, as exemplified in (64).

184

(64)  Yoron Northern Ryūkyūan (field notes, July 2–4 2013) a. Sinbun yumyun. newspaper  read.concl ‘(I) read a newspaper.’ b. [Ø  yumyu(u)ru]  sinbun newspaper   read.adnz  ‘a newspaper that (I) read’ (65) Ishigaki Yaeyama Ryūkyūan (field notes, October 23 2012) a. baa ya   sinbun   wa   yumun. I  top  newspaper  acc  read.concl ‘I read a newspaper.’ b. [Baa nu   Ø yumu]    sinbun   ya I   gen    read.adnz  newspaper  top kunu  sinbun. this  newspaper ‘The news paper that I read is this newspaper.’ The likelihood that these languages also once had merged conclusive/adnominal forms, as have many other contemporary Ryūkyūan languages/dialects, is indicated by the fact that their subject markers are related to the genitive nu or ga. Finally, Uwano (2006) reports an interesting prosodic differentiation of the conclusive/adnominal contrast in the Iwayaguchi dialect of Sadogashima Island in Niigata, whose development he reconstructs as in (66). (66) Iwayaguchi dialect (Sadogashima, Niigata; Uwano 2006) kiru ‘wear’ Conclusive *[ki!ru HM>  *[ki]ru HL =  *[ki]ru HL  = [ki]ru  HL  (modern form) Adnominal *[ki!ru HM=  *[ki!ru HM >  *[kiru HH  > ki[ru  LH  (modern form) ageru ‘give’ Conclusive *[age!ru HHM > *[age]ru HHL = *[age]ru HHL > a[ge]ru LH (modern form) Adnominal *[age!ru HHM = *[age!ru HHM > *[ageru HHH > a[geru LHH (modern form)

12.5  Hachijō The study of the Hachijō dialect has a rather long history, starting with F.V. Dickins and Ernest Satow’s “Notes of a visit to Hachijo” (1878), in which they pointed out, early on, similarities between Hachijō and the languages used in Man’yōshū, especially Eastern Old Japanese, such as the adnominal -ke

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects Nara-period EOJ Pre-protoJapanese

Hachijō

Nara-period COJ

Kyōto dialect, etc.

Proto-Japanese Ryūkyūan

Figure 12.5  Hattori’s (1976) positioning of Hachijō

form for adjectives, corresponding to -ki in Central Old Japanese.26 Most subsequent major works on this dialect, such as Hoshina (1900), the first systematic treatment by a Japanese scholar, and Tōjō (1934), mention this connection between Hachijō and Eastern Old Japanese, regardless of whether they refer to the work of Dickins and Satow. Hōjō (1948) and subsequent works, by himself and by others, expand the search for the Hachijō-Eastern Old Japanese (as well as the Hachijō-Central Old Japanese) connections beyond the adnominal form of adjectives, such as the adnominal -o ending for verbs and the kakari-musubi concordance phenomena (see Section  12.5.3.3). Morphological studies of Hachijō from earlier periods culminate in Itoyo (1959), according to Kaneda (2001). This work by Akihiro Kaneda is clearly the most advanced systematic treatment of this dialect and stands out not only as a synchronic description but also as a comparative study vis-à-vis Eastern Old Japanese, Central Old Japanese, and Standard Japanese. However, it has little to say about the historical position of Hachijō in the Japonic family tree. Kaneda (2001: 12–13) does, however, mention past speculations about this question, such as Kindaichi (1955), who divided eastern Japan into the three dialect areas of eastern, northern, and Hachiijō, and Hirayama (1958), who proposed recognizing Hachiijōjima (perhaps including Aogashima) as one of the four major dialect areas, along with eastern Japan, western Japan, and Kyūshū. Hattori (1976), recognizing some Hachijō features that he believed went back to Pre-Proto-Japanese, proposed the most radical, hence ­ contro­versial, ­positioning of Eastern Old Japanese as a sister language of Proto-Japanese. In this scenario, the former survives as Hachijō and the latter splits into Central Old Japanese and Ryūkyūan, as in Figure 12.5. Besides the Aogashima variety (with fewer than 50 surviving native speakers), the Hachijō island varieties are principally spoken in the five communities of Mitsune, Ōkago, Kashitate, Nakanogō, and Sueyoshi, by several thousand speakers.27 My cursory observations confirm Kaneda’s 26   This and the next paragraph are based on Kaneda’s (2001: 5ff.) his­ torio­graph­ic sketch of the scholarship on Hachijō. See his book for the references cited in this section. 27   An evolved form of Hachijō is spoken in Daitōjima Island situated southeast of Okinawa, where it is recorded that 23 people from Hachijō Island emigrated in 1900.

(2001: 2) view that dialect variations among these communities are mostly phonological, with little difference in the lexicon and grammar. Perhaps the most interesting and important characteristic of Hachijō is its possible preservation of features from at least three strands of Japonic languages, namely pre-Old Japanese (Proto-Japonic), Central Old Japanese, and Eastern Old Japanese.

12.5.1  Hachijō and Ryūkyūan The oldest Hachijō features that may go back to Proto-Japonic are those shared by Ryūkyūan languages. Before looking at some of them, a cautionary remark is in order. The Ryūkyūan and the Izu Archipelago, where the Hachijō Island is located, lie in the course of a warm sea-current known as the Kuroshio Current (Black Stream), which starts out off the east coast of Luzon and flows north to Taiwan and the Ryūkyū islands. The main current, which changes its course year to year, then flows northeast along the southern coast of the Japanese main island and cuts across the Izu Archipelago before dissipating in the north Pacific (see Figure 12.1). A major characteristic of the “Kuroshio culture” is the high level of seafaring knowledge that facilitated not only the introduction of the Southeast Asian culture to Japan but also contact between the remote islands and the main islands of Japan. Many point out prehistoric cultural similarities between the Izu and Ryūkyūan islands, and much more recent and frequent contact between these two archipelagos could have been maintained (Hayashi 2013). Thus, the possibility cannot be ruled out that linguistic similarities between Hachijō and Ryūkyūan are due to the past contact. While the field still awaits a systematic comparison between Hachijō and Ryūkyūan, Hirako and Pellard (2013) point out a few lexical items shared by the two languages. Their first observation is about a distinction in the form of the numeral ‘one’ in numeral classifier expressions. Whereas in Modern Japanese the numeral ‘one’ is pronounced /hito-/ regardless of whether it is followed by the native general classifier -tu (hito-tu ‘one THING’) or other classifiers, such as the Sino-Japanese classifier for people -ri (hito-ri ‘one PERSON’). In the Mitsune dialect of Hachijō, according to Hirako and Pellard, they are distinguished as tetsɯ ‘one

185

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani Table 12.1  Hachijō–Ryūkyūan correspondence: Numeral ‘one’ ‘one-THING’

‘one-CLF’

Hachijō (Mitsune)

te-tsɯ

to-

Amami

tˀɨː-tsɨ

tɕˀu-

Okinawan

tiː-tsi

tɕˀu-

Miyakoo

çiti-tsɨ

çitu-

Yaeyama

pˢɨtiː-tsɨ

pˢɨtu-

THING’ and toɾi ‘one PERSON’, and they note similarities to the Ryūkyūan counterparts, as in the abbreviated reproduction of their comparison in Table 12.1. Hirako and Pellard (2013) also point out a few additional words shared by the two languages, such as Hachijō miʑa, miʥa ‘soil, ground’ and Ryūkyūan mitɕa (Amami), ntɕa (Okinawan), mta (Ogami Miyako), nta (Yaeyama); Hachijō meme zͩ u-me, nenezɯ̈-me, mimizu-me ‘earth worm’ and Ryūkyūan mɨmɨdzɨ (Amami), mimidʑi (Okinawan), mimikɯ (Ogami Miyako), mimidzɨ (Yaeyama); Hachijō jo, ijo ‘fish’ and Ryūkyūan ʔjuː (Amami), ʔiju (Okinawan), zzu (Miyako), idzu (Yaeyama); Hachijō tsubuɾi head’ and Ryūkyūan tsɨbuɾu (Amami), tsibuɾu (Okinawan), tsɨbuɾɨ (Yaeyama); Hachijō tommetei, tommete ‘morning’ and Ryūkyūan ɕitɨmuːtɨ (Tokunoshima Amami), sutumiti (Okinawan), sɨtumuti (Miyako), sɨtumudi (Yaeyama); Hachijō distal demonstrative ure, uno ‘that’ and Ryūkyūan medial demonstrative ʔuɾɨː (Tokunoshima Amami), ʔuɾi (Okinawan), ui (Miyako), uɾi (Yaeyama).

12.5.2  Hachijō and Eastern Old Japanese Of the Eastern Old Japanese features that are often pointed out as being preserved in isolated dialect areas in eastern Japan, the least controversial in Hachijō is the adnominal/ nominalization endings for verbs and adjectives, which are respectively -o and -ke, as opposed to COJ -(r)u and -ki. The eastern songs of Man’yōshū include the following containing the -o and -ke endings. (67)  OEJ verbal adnominal/nominalization form a. [pur-o] yoki fall-adnz  snow ‘falling snow’ (Man’yōshū 3423; azuma uta: COJ pur-u) b. [tuger-ar-o]     a  ga inform-hon-adnz   I   nom sena husband ‘my husband who has informed’ (Man’yōshū 3469; azuma uta: COJ tuger-aru-ru)

186

(68)  OEJ adjectival adnominal/nominalization form a. [naga-ke] kono  yo   long-adnz  this night ‘this long night’ (Man’yōshū 4393; sakimori uta: COJ naga-ki) b. [kanasi-ke]   mako beloved-adnz  family ‘beloved family’ (Man’yōshū 4414; sakimori uta: COJ kanasi-ki) The eastern songs in Man’yōshū are not entirely consistent in their rendition, probably because the scribes in charge were speakers of Central Old Japanese, and many songs are written in the then standard Central Old Japanese forms. For example, song 4413 has the standard Central Old Japanese adnominal ending, as in [makanasi-ki] sero ‘beloved husband’. Although modern Hachijō also includes many forms of modern central dialects of the mainland, it uses V-(r)o and A-e adnominal endings in a quite regular manner, as in (69). (69)  Hachijō verbal adnominal form28 a. [Ø  yom-o] sinbun read-adnz newspaper ‘a newspaper that (I) read’ b. [Ø

tattar-o]     ko stand.exist-adn   child ‘a child who is standing’

c. [Ø

not=te   ku-ro]    kuruma ride=conn  come-adnz   car ‘a car that someone comes riding in’

(70)  Hachijō adjectival adnominal form a. [Ø ake-ke] ringo red-adnz apple ‘a red apple’ b. [Ø

taka-ke]   yama high-adnz  mountain ‘a high mountain’

c. [Ø

era-ke]      hito important-adnz  person ‘an important person’

Hachijō adnominal forms play an important role in understanding the nature of grammatical nominalizations and their role in grammar, as pointed out earlier (see Section 12.4.1.6). First, they clearly show that adnominal forms 28 The Hachijō data without source identification are from the author’s field notes dated August 19-22 2010.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects are really grammatical nominalizations, whose foremost function is to denote things and thing-like entities like ordinary nouns. Like Old and Middle but unlike most modern dialects, Hachijō adnominal forms can head an NP as they are, without any additional marker such as the Standard Japanese NP-use marker (  Juntai joshi) no. Observe (71). (71)  NP-use of grammatical nominalizations a. [Yamadasan  ga  mainiti   tonmeteni  Ø Yamada.Mr  nom every.day morning.at   yom-o]   wa  Asahi  sinbun      daa   dyo. read-adnz  top  Asahi  newspaper  cop  fp ‘What Mr. Yamada reads every day is the Asahi Newspaper.’ b. [Sinbun    yo  mainiti    yom-o]   wa newspaper  acc every.day read-adnz  top   yoke   koto  dara. good thing cop ‘To read a newspaper every day is a good thing.’ c. [Ø  wakak-e]   ga   ukuni  aro.dya.   young-adnz  nom there exist.concl ‘A young one is over there.’ The argument nominalizations in (71a) and (71c) denote entities that function as object and subject arguments of the events that the nominalization structures represent, whereas the event nominalization in (71b) denotes an event itself. The term “adnominal” for these nominalizations comes from their use as nominal modifiers, as below, which actually are one and the same structures that head an NP, as in (71). (72) a. [Yamada-san  ga mainiti tonmeteni  Ø Yamada-Mr nom  every.day  morning.at yom-o]   sinbun read-adnz newspaper ‘the newspaper that Mr. Yamada reads every day’ b. [sinbun     yo  mainiti   Ø  yom-o]   kuse newspaper  acc  every.day   read-adnz habit ‘the habit of reading a newspaper every day’ c. [Ø  waka-ke]   gakusei   young-adnz student ‘a young student’ While forms such as those in (71) are perfectly grammatical and are widely used, the NP use of grammatical nom­in­al­iza­ tions may also be marked by -me, signaling the referential function such nominalizations play by virtue of their heading an NP. The innovative marking of nominalizations in NP

use by -me parallels the development and function of the so-called Juntai joshi no, ga, and others in modern central dialects. Another theoretically interesting point that Hachijō adnominal/nominalization forms show is that the conclusive/adnominal merger has been undone with innovations leading to new verbal and adjectival finite, conclusive forms. That Hachijō once showed a formal distinction between conclusive and adnominal forms, even more clearly than Central Old Japanese, is seen from the fact that old conclusive forms have survived in certain morphosyntactic contexts. For example, the COJ conjectural/intentional/ obligation auxiliary -besi called for -(r)u-ending conclusive verb forms, such as nomu-besi ‘(you) should drink’. Hachijō likewise uses the (old) conclusive verb form in this context, as in nomu-beki dara (drink-conj.adnz cop) ‘(you) should drink’, as opposed to the adnominal form nomo. The hearsay evidential form -teiya derives from -te yuu (that say) ‘say that’, and in this context too the old finite verb form is seen, as in nomu-teiya ‘(they say that they) drink’ (Kaneda 2001: 113ff.). These old finite verbal forms are no longer usable in concluding declarative sentences. Instead, the contemporary Hachijō conclusive forms are a combination of an ad­nom­ inal form and a sentence-final particle as in (73). (73) a. Yamadasan wa mainiti tonmeteni Yamada.Mr.  top  every.day  morning.at   sinbun    yo  yom.o.wa. newspaper  acc read.adnz.concl ‘Mr. Yamada reads a newspaper every morning.’ b. Wa  ga  ko  wa  sokoni  tatte I   gen child top there  stand.conn   ir.o.dya. exist.adnz.concl ‘My child is standing there.’ The sentence-final particles above still retain the nuances of the final particles used, similar to Tōkyō Japanese final par­ ticle yo. However, these declarative sentences cannot end without them, unlike Tōkyō declarative sentences with an adnominal/conclusive verb form, which can end without a final particle. Both -wa and -dya, in other words, have come to function as conclusive verbal endings in contrast to the Toshima dialect, which still can end a sentence in the ad­nom­ inal/nominalized form (see Section 12.4.2.2). The grammaticalization of these final particles as conclusive endings is more clearly seen when adjectives are involved. The predicative form of adjectives can be either morphologically transparent (74a) or phonologically integrated (74b).

187

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani (74) a. Sensii no kaban  wa kurok.e.dya. top  black.adnz.concl teacher  gen  bag ‘The teacher’s bag is black.’ b. Sensii   no  kaban  wa  kurokya(a). teacher  gen  bag   top black.concl (kurokya(a) < kuroke.wa) ‘The teacher’s bag is black.’

12.5.3  Hachijō, Central Old Japanese, and Ryūkyūan As mentioned above, Kindaichi’s (1976) speculation is that modern eastern dialects are descendants of Central Old Japanese that spread eastward, wiping out the indigenous Eastern Old Japanese features from the local dialects. Hachijō came into more organized contact with mainland dialects after the late Muromachi period (1336–1573), when the island began to be increasingly controlled by the mainland governments, culminating in its status as a territory of the Tokugawa government (1603–1868), which used the island as one of the penal colonies in the Izu Archipelago, shipping out nearly 2,000 prisoners during this period. Thus, Hachijō has received continuous influence from the mainland dialects, which, of course, continues today. I shall discuss in Section  12.5.3.1 below one Hachijō phenomenon showing a connection between Hachijō and Central Old Japanese, which represents a more remote connection with Ryūkyūan and Proto-Japanese, or from the time when Hachijō (or Proto-EOC), Proto-Central Old Japanese, and Proto-Ryūkyūan were not separate languages. I shall first take up two topics related to the preceding discussions on the adnominal/nominalization forms, namely genitive markers and the kakari-musubi focusing constructions.

12.5.3.1  Genitive markers As we have seen, Hachijō uses the particles no and ga as genitive markers. This is a property shared by Old Japanese and Ryūkyūan, and the usage pattern of the two particles is also similar in the three languages. On the other hand, these particles took different paths of development. In the central Japanese dialects, no became an exclusive genitive maker, with ga taking on a newly developed subject-marking function. In contrast, certain peripheral mainland dialects, Ryūkyūan, and Hachijō continue to use both no and ga as genitive markers, in addition to their newly acquired ­subject-marking function. As mentioned in Section 12.4.1.3, some dialects attach a deferential value to these particles, with no being used for a respectable/distant referent.

188

Nohara (1998: 50) shows that the Naha Okinawan pattern is similar to Old Japanese, with pronouns, proper nouns, and numerals taking ga, and other nouns nu (Japanese no). He specifically notes that there is no deferential value associated with these particles in this dialect. However, my field notes on the Tsuken Island dialect of Okinawan show the use of ga for nouns denoting people close to or familiar with the speaker and the general nu for other types of referent, including respectable people. Observe (75). (75)  Tsuken Okinawan (field notes August 23 2012) a. waa  (ga/*nu)  kutu ‘my shoes’ I    gen/gen  ‘shoes’ b. ari  ga/*nu kutu  ‘his shoes’ c. suutaa   ga/?nu kutu  ‘Daddy’s shoes’ ammaa  ga/?nu kutu  ‘Mommy’s shoes’ d. mmee  ga/nu kutu  ‘grandfather’s shoes’ paapa  ga/nu kutu  ‘grandmother’s shoes’ e. sinsii  *ga/nu kutu  ‘the teacher’s shoes’ f. in/ingwa  *ga/nu sip  ‘the dog’s tail’ Turning now to Hachijō, Kaneda (2001: 35) describes the situ­ation as follows: ordinary nouns appear with no, but nouns for people and living things may appear with ga. Proper nouns and pronouns take ga as a basic rule. Even if people’s names (proper nouns) are involved, remote ancestors and historical ­people, whom the speaker does not know directly, may appear with no. Nouns for living things appear with both ga and no, but if they are individuated, such as a specific cat, it seems that ga more likely appears.

Karimata’s (2015: 138) assessment for Ryūkyūan seems to generally apply to Hachijō that the rule requiring ga for pronouns and proper nouns and nu for others, is no longer strictly maintained even if a dialect uses both particles. In Hachijō, too, I find a great deal of variability in particle choice. (76) a. (w)a  ga    kutu I   gen shoe ‘my shoes’ b. toto/totoodo/tootyan    ga  kutu Daddy/Daddy/Daddy  gen shoe ‘Daddy’s shoes’ c. inume  ga  sippo dog   gen tail ‘a dog’s tail’

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects d. sinsii    ga/no   kaban teacher  gen/gen bag ‘the teacher’s bag’ Most speakers prefer ga for (76a)–(76c), with a smaller number of people allowing both ga and no, whereas a far larger number of speakers use no for (76d). The observed variability here must be carefully ascertained in relation to the referential status of the noun referent, whether or not it is familiar to the speaker and whether or not it is individuated, as Kaneda’s description has it.

12.5.3.2  Subject markers As I have mentioned repeatedly, the original genitive function of no and ga has been extended into a subject-marking function. Naha Okinawan appears to maintain the distribution pattern of the genitive ga and nu, whereby pronominal subjects and those denoting people’s names are marked by ga and those denoting ordinary nouns nu. Observe Nohara’s (1998: 49–55) data in (77), where some variability is seen for kin terms. (77)  Naha Okinawan (Nohara 1998) a. ʔuree mucikasanu ‘waa gaa naran. nom.top can.not that difficult.because I ‘Because it is difficult, I cannot do it.’ b. ʔanma   ga  ʔnzi  kuunri  ʔiitan. mother  nom go  come  said ‘Mother said to go and come back.’ c. Tuzi  nu   cuun. wife  nom come ‘My wife comes.’ d. ʔamankai   mayaa  nu  ‘un. over.there cat   nom exist ‘There is a cat over there.’ My Ūyama data from Ginowan City near Naha City appear more progressive, showing that ga is used more widely than the Naha data collected by Nohara. Observe (78). (78)  Ūyama Okinawan (field notes August 22 2012) a. Mii kutsu  ja, waa ga/*nu kootan. nom/nom  bought new  shoe top  I ‘As for the new shoes, I bought (them).’ b. Okaa   ga/*nu  kin    arattan. Mom  nom/nom clothes washed ‘Mom washed the clothes.’ c. Anu   inagu  ga/*nu  unu  kin    arain. that woman nom/nom this clothes wash ‘That woman washes this clothes.’

d. In   ga/nu  acchoon. dog  nom/nom walking ‘A dog is walking.’ e. Kazi   ga/nu  hukuun. wind  nom/nom blow ‘Wind is blowing.’ f. Kii   ga/nu   tooritoon. tree  nom/nom fallen ‘A tree is fallen.’ Subject marking in Hachijō is similar to the Okinawan situ­ ation examined above; the dialect of the remoter island Aogashima seems more conservative than the Hachijōjima dialects in that no is still used fairly widely in marking or­din­ ary noun subjects in the former, while in the latter such subjects are marked by ga, just like pronominal subjects, as in modern mainland Japanese dialects. At least this was the situation with a speaker of the Aogashima dialect born in 1902 and the speakers of Hachijōjima dialects born between 1891 and 1938, from whom Kaneda (2001) obtained the data. The narrative Kaneda collected from his Aogashima speaker contains the following forms, where ordinary noun subjects are still marked no, though some of them are marked by ga as in (79e). (79)  Aogashima Hachijō (Kaneda 2001) a. sinbun no hutagattaro.wa. newspaper  nom  have.covered.prs.concl ‘The newspaper has covered (it).’ b. kora  sini     no   nuketoro.wa. this   bottom  nom fall.off.prs.concl ‘As for this (plastic can), the bottom has fallen off.’ c. zyu:taido:   byaonin    no  arodo.dya. critically.ill sick.person nom exist.pst.concl (Lit.) ‘A critically ill person existed.’ ‘There was a critically ill person.’ d. Azi   no   cigao.dya. taste  nom differ.prs.concl (Lit.) ‘The taste differs.’ ‘The taste is different.’ e. Nekkome  ga  nemero.wa. cat   nom lick.prs.concl ‘The cat licks (it).’ f. Iro    ga  nikuke-dya. color  nom bad-prs.concl ‘The color is bad.’ Compare the above with the following Hachijō-jima forms from Kaneda (2001), where ordinary noun subjects (and an expected pronominal subject as well) are consistently marked by ga.

189

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani (80)  Hachijōjima dialects (Kaneda 2001) a. a ga huko.wa. I nom wipe.prs.concl ‘I wipe (it).’ b. Hara     ga  tato.wa stomach  nom stand.prs.concl (Lit.) ‘The stomach stands.’ ‘I get mad.’ c. Mi    ga  narara. berry  nom ripen.pst.concl ‘The berries have ripened.’ d. Tricubosa  ga  tenneiyo  makimitte bird    nom  sky    flying aro.wa exist.prs.concl ‘A bird is flying in the sky.’ e. hanzume    kaze   ga  hutte a.while.ago wind nom blowing arara. exist.pst.concl ‘The wind was blowing a while ago.’ f. kono  takami  ga  yokkya. this   height  nom good.prs.concl ‘This height is good.’

12.5.3.3  Kakari-musubi One syntactic phenomenon that appears to be quite unique to the Japonic family is that known as kakari-musubi in Japanese linguistics. This phenomenon, once flourishing especially in Early Middle Japanese, died out in the Late Middle Japanese period and is not seen in most modern Japanese dialects; but its remnants are still observed in both Hachijō and Ryūkyūan. The fact that Ryūkyūan maintains this phenomenon indicates that it is a Japonic property and hence a possibility exists that the Hachijō constructions below come directly from Proto-Japonic rather than via Central Old Japanese. Kakari-musubi is a concordance phenomenon observed between certain adverbial particles, known as Kakari joshi, and the sentence ending. Particles zo, namu, ya, ka called for the adnominal (nominalized) form of a predicate, koso a realis form, and wa/mo a conclusive form.29 These constructions opposed plain conclusive-ending declaratives as interrogative or exclamatory sentences (particles ya and ka), as emphatic sentences (zo, namu, koso), or as topical universal statements (wa and mo). The realis (izen)/irealis (mizen) opposition, which has been lost in modern Japanese, is most clearly seen in conditionals marked by -ba; hana sak-aba. . . ‘because/since the flowers have bloomed. . .’ vs. hana sak-e-ba. . . ‘if the flowers bloom. . .’. 29

190

Below is an authentic COJ kakari-musubi form from Man’yōshū: (81) kagurwoki  kami  ni tuyu  so black   hair  dat   dew  ptcl oki-ni-kyeru fall-pfv-mst.adnz/nmlz ‘It is dew that has fallen on my black hair.’ (Lit.) ‘Dew is what has fallen on my black hair.’ (Frellesvig 2010: 248) Kaneda (2001: 184ff.) discusses the Hachijō kakari-musubi under the heading of the “emphatic predication” involving the particles ka and ko: and a realis predicate form. The par­ ticles involved correspond to OJ koso, which called for a realis predicate form. The emphatic particles may attach to both case-marked noun phrases and nominalizations, as in (82). (82)  Hachijō kakari-musubi (Kaneda 2001) a. ara  sake-i-ka nomar-e. I sake-acc-emph  drank-real ‘I drank SAKE.’ or ‘It is sake that I drank.’ b. [ara  sake-i   nomi]-ka    sitar-e. I   sake-acc drinking-emph did-real ‘My drinking of sake is what I did.’ More detailed descriptions of kakari-musubi are available in Ryūkyūan. The most widely discussed pattern across different Ryūkyūan languages involves the particles du/ru and do/ to, normally identified as “focusing particles”, which cor­re­ spond to OJ zo/so. As with these particles in Old Japanese and Early Middle Japanese, the Ryūkyūan counterparts typically call for an adnominal/nominalized predicate form, as seen in (83). (83)  Naha Okinawan (Karimata 2015: 138) Tigame: wa-ga-du katɕ-uru. letter.top I-nom-foc write-adnz/nmlz ‘As for the letter, it is I who am going to write.’ (Lit.) ‘As for the letter, I am the one who will write it.’ (84)  Miyara Yaeyama Ryūkyūan (Davis 2013: 2) Taa=du suba tsukur-ee-ru? who=foc noodles make-res-adnz/nmlz ‘Who made soba noodles?’ (Lit.) ‘Who is the one who made soba noodles?’ The kakari-musubi constructions involving the nominalized predicate forms as seen above are reminiscent of similar focus/cleft constructions involving nominalizations in many other languages, see (85).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects (85)  Sasak (Malayo-Polynesian; Shibatani 2019) a. Cleft Loq Ali [si Ø  mbace buku=ne] art  Ali  nmlz  af.read  book=this ‘It is Mr. Ali who read this book.’/ (lit.) ‘Mr. Ali is the one who read this book.’ b. Wh-question   Sai    [si   Ø  mbace  buku=ne] who  nmlz    af.read book=this ‘Who read this book?’/(lit.) ‘Who is the one who read this book?’ (86)  Yaqui (Uto-Aztecan; Shibatani 2019) a. Cleft Joan [Ø  wa-me yabe-m tea-ka-me] dem-pl  key-pl  find-pfv-nmlz John  ‘It is John who found keys.’/(lit.) ‘John is the the one who found those keys.’ b. Wh-question   Jabesa  [Ø  wa  jiosam  noktua-me] who    dem book  read-nmlz ‘Who is reading the book?’/ (lit.) ‘Who is the one that is reading the book?’ (87)  Turkish (Erguvanlı 1984) Chomsky-nin  teori-si-dir      bugün Chomsky-gen theory-3poss-pred.m today siz-e    anlat-mal  iste-diǧ-im you-dat tell-inf    want-nmlz-1sg.poss ‘It is Chomsky’s theory that I want to talk to you about today.’ Thus, kakari-musubi constructions find fairly straightforward analogs in other languages once the so-called ad­nom­ inal form is properly recognized as a nominalization. What is unique to kakari-musubi, then, is its involvement of focusing adverbial particles (Kakari joshi) that mark the constituent that is placed under focus. Like Early Middle Japanese and Late Middle Japanese, both Hachijō and Ryūkyūan show deterioration of the kakari-musubi rules such that adverbial focus marking may occur without the expected correlative form of a predicate or conversely, the correlative form of a predicate may occur without the relevant focusing particle.

12.5.3.4  Grammaticalization of existential verbs As in many other languages, existential verbs have played a very important role in the development of the aspectual/ tense/evidential system throughout the history of Japonic

languages. OJ verb ari ‘exist’, for example, was recruited in the periphrastic stative expression V=te ari (V=conn exist) ‘is in the state of V-ing/is in the state of having V’. This form gave rise to the EMJ stative auxiliary -tari, which, in turn, resulted in the past-tense suffix -ta in central modern dialects including Tōkyō Japanese. Other grammaticalized uses of ari from the Early Middle Japanese period include -kyeri for modal past (< ki-ari ‘come exist’), -yeri stative (< *-i ari ‘stem.nmlz-exist’), nari ‘hear’ evidential (< *ne ari ‘sound exist’), meri ‘see’ evidential (< *mi ari/*mie ari ‘see exist’). The descendant of OJ wiru ‘sit down, stay still’ iru has a main-verb use with the meaning of ‘exist, be’ as well as the grammaticalized version forming the periphrastic imperfective/stative aspect expression V=te (i)ru ‘is V-ing/is in the state of having V-en’ in central modern dialects. Modern western Japanese dialects have recruited the descendant form of OJ woru ‘exist’ for similar purposes. Exploiting the two possible periphrastic forms V.i oru (V.stem.nmlz exist) and V.i=te oru (V.stem.nmlz=conn exist), the Kumamoto dialect and some others have developed an imperfective/ perfect opposition vis-à-vis the perfective aspect; e.g. nomu ‘drink’ (perfective); nomi-yoru (< nom.i oru) ‘is drinking’ (imperfective), non-doru ( numun (drink.prs.concl); num.i utan > numutan (read.pst.concl) ‘drank’. According to Kaneda (2001: 106), there are three types of verbal tense paradigm going on in Hachijō. The native ­pattern involves the adnominal verbal stem ending in -o and the existential verb aro for the past tense forms (Table 12.2a). A second pattern (Table 12.2b) adopts the mainland centraldialect conclusive ending -(r)u for both present- and pasttense conclusive forms instead of the Hachijō conclusive form aro.wa. A third type (Table 12.2c) adopts the past form seen in certain modern eastern dialects, which involves the ending -datta, whose ultimate source involves the ex­ist­ ential verb ari twice, as in -datta < *=te/de ar.i tari. A major characteristic of the Hachijō tense/aspect system that distinguishes itself from modern central dialects and that shows its affinity to Classical Japanese lies in the semantics of the present-tense form. While in modern central dialects, the basic present-tense form denotes generic tense or future, as in (88a). (88) a. Kono  ko wa sakana this child top  fish ‘This child eats fish.’

o acc

tabe-ru. eat-prs

191

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani Table 12.2  Hachijō tense/aspect forms (nomo- ‘drink’) Present

Past 1

Past 2

a. Native

nomo.wa nomo.dya

nomara/nomo: (< *nom.i aro.wa)

nomarara/nomaro: (< *nom.i ar.i-aro.wa)

b. Mainland endings

nomu

nomaru (< *nom.i aru)

nomararu (< *nom.i ar.i-aru)

c. Modern eastern endings

nomu

nonda (< *nom.i-ta)

nondatta (< *-te/de ar.i-ta)

b. Kono  ko    wa  sakana  o  tabe=te this   child  top  fish   acc eat=conn i-ru. exist.prs ‘This child is eating fish.’ Specifically, the simple present-tense form in (88a) cannot predicate over an ongoing activity, which calls for the periphrastic imperfective form as in (88b). The situation was different in Classical Japanese, as it is in Modern Korean, whose present-tense verb forms predicate over ongoing activities. The following famous passage from Ise monogatari (The Tales of Ise) (10th–11th centuries ad) contains the present tense form kupu ‘eat’. As indicated in the English translation, this verb is predicating over an imperfective, ongoing activity of the bird’s feeding on fish, which the modern central dialect counterpart cannot do (see also example 31a from Kojiki). (89) siroki tori no sigi no opokisa  naru cop.adnz white  bird  gen  snipe  gen  size midu   no  upe  ni  asobitutu  iwo   wo water  gen  top  dat frolic.cont  fish  acc kupu. eat.prs.concl ‘One that is of the size of a snipe, a white bird, is ­eating fish while frolicking on the water.’ Hachijō present-tense forms, e.g. nomo.wa and nomo.dya for ‘drink’, appear similar to those of Classical Japanese in that these forms may predicate over imperfective, ongoing activities. Between these two present forms, the ones with the -dya conclusive ending predicate much more freely over ongoing events. Observe the following examples from Kaneda (2001: 212) in (90).

192

(90) a. Ao-yo syo:i?  Mesi-yo kamo.dya. Meal-acc  eat.prs.concl what-acc  do? ‘What (are you) doing?’  ‘(I) am eating a meal.’ b. Yohukasi    an-yo   syo? late.at.night what-acc do? ‘What (are you) doing (this) late at night?’   Sinbun-yo    yomo.dya. newspaper-acc read.prs.concl ‘(I) am reading newspaper.’ The imperfective use of the -wa ending for verbs is rather limited in present-day Hachijō. According to Kaneda (2001: 210), the imperfective reading obtains most readily with middle/reflexive situations, as in the following examples from his work in (91). (91) a. Manyo  an-yo si=te aro? now what-acc  do=cont  exist ‘What are (you) doing now?’ Karado:  hukero.wa. body.acc wipe.mid.prs.concl ‘(I) am wiping my body.’ b. Dora    a  ga  huko.wa. let.me.see  I   nom wipe.prs.concl ‘Let me see, I will wipe (there).’ Sentence (91b) contains a simple transitive, as opposed to the middle form in (91a), and it denotes a future activity only, rather than the present ongoing activity predicated by (91a). Past 1 forms (Table 12.2), nomara/nomo:(< *nom.i aro-wa), predicate over both resultative states and past events. In comparison to Past 2 forms with two grammaticalized ex­ist­ ential verbs, nomarara/nomaro: (< *nom.i ar.i aro-wa), Past 1 forms predicate over relatively recent past events and Past 2

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects forms remoter past events with less relevance to the present time. Kaneda’s examples include (92). (92)  Kaneda (2001: 214–16) a. mi ga narara. (Past 1: Resultative) berry nom ripen.pst1 ‘The berries have ripened.’ b. wara  kinei    terebyo  mitara. (Past 1: Past) I   yesterday TV.acc watch.pst1 ‘I watched TV yesterday.’ c. kyonen  wa  sikkai   mi    ga  nararara. last.year top  firmly  berry  nom ripen.pst2 (Past 2: Remote past) ‘As for last year, the berries ripened a lot.’ Kaneda (2001: 208) suggests that the Hachijō tense/aspect system has been undergoing change such that the present form, nomo.wa, is shifting toward the expression of a future event, as in modern central dialects, with the periphrastic imperfective form, nonde aro.wa, being used for present, ongoing events, and Past 1 pushing out Past 2 as the past form.

12.5.3.5 Voice Like modern central dialects, Hachijō has morphological voice categories of causative, passive, spontaneous, and potential. As in other dialects, the latter three share morph­ ology to a large extent. The Hachijō voice morphology retains some usage pattern of the Classical Japanese counterpart. Causatives, e.g. nom.a-se-ro.wa ‘make drink’, express all types of causative situations as in mainland dialects, such as the inducive (‘make’), the permissive (‘let’), and the handsoff (‘let doing’) types. Similar to Classical Japanese causatives involving -su, -sasu, -simu, Hachijō causatives may be used in combination with an honorific verb for the expression of deference, as in (93). (93)  Honorific causative (Kaneda 2001: 275) Omee-ni kogondoo koto made you.hon-dat this.adnz  npm  up.to s-ase-iteete. do-caus-do.hon ‘Making you do even things like this.’ As in Classical and Modern Japanese, the suffix -rare/-re-, as in nom.a-re-ro.wa, represents multiple voice categories of spontaneous (get to drink willy nilly), passive (is drunk),

and potential (can drink/potable). The Classical and Modern central dialect counterparts, in addition, can be used in an honorific function. This function in mainland dialects was, however, a later development in history. In Hachijō, -rare/-re has no honorific use; instead, the honorific suffix -yaro is used as in nom.i-yaro.wa ‘drinks (hon)’. It is not known whether -re has never developed the honorific function or has lost it; the former is likely in view of its late development in Classical Japanese. Like modern central dialects, the passive -re may be used for both direct and adversative/indirect passives, as in (94). (94)  Hachijō passives (Kaneda 2001) a. Uren kucik.a-are-tara. (Direct passive) that.one.dat cheat-pass-pst ‘(I) was cheated by that guy.’ b. Atode  kamoosyaate  okoo     yo later  eating    set.aside.adnz  acc derenka   kam.a-re-tara. (Indirect/Adversative) someone.dat eat-pass-pst ‘I was adversely affected by someone eating the food that I set aside with an intention of eating it later.’ The potential -rare/-re denote different types of enabling situations, such as ability (osie-rare-ro.wa ‘can teach’), property (nom.a-re-ro.wa ‘potable’), and circumstantial (aw.are-ro.wa ‘can (meet because of…)’). Noticed here is that the full -re form can denote these potentials with a consonant base verb, as in the last two examples here, reflecting the older pattern; in Standard Japanese they use the short -e form, as nom-e-ru ‘can drink and a-e-ru ‘can meet’. The same is true with the spontaneous -rare/-re forms such as nak.a-re-ro.wa ‘feel like crying’, which would be nak-e-ru in modern central dialects, though they still generally retain long forms such as omow.a-re-ru ‘get to think willy nilly’ and sinob.a-re-ru ‘get to long for willy nilly’. The Hachijō form such as ki ga s.a-rero.wa (feeling NOM do-pot-concl) ‘get to feel like’ indicates that a larger class of verbs may combine with the potential suffix, similarly to Classical Japanese and differently from modern central dialects, in which the spontaneous suffix is limited to a handful of verbs of cognition. Hachijō has an interesting middle-marking use of the short spontaneous/potential suffix -e, at least for some verbs. This is a unique development not seen in modern central dialects. Observe (95). (95)  Hachijō middle-marking (Kaneda (2001: 275) a. Houte  houte  aor-e-ro.wa. hot hot fan-mid-prs.concl ‘It being very hot, (I) am fanning myself.’

193

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Masayoshi Shibatani b. Karado:  huk-e-ro.wa. body.acc wipe-mid-prs.concl ‘(I) am wiping my body.’ The simple transitive verbs aoro.wa ‘fan’ and huko.wa ‘wipe’ would be used in non-middle/reflexive situations. The extent to which the middle marking applies to different verbs is not clear. Other languages also vary in the size of middle-marking verbs, since middle-marking competes with unmarked self-affecting intransitive verbs.

12.5.3.6  Insubordination As studied in Section 12.3.4.5, in modern central dialects stem nominalizations are often realized as sentences. The context of conversation hints at the intended illocutionary act, supporting stand-alone stem nominalizations as sentences. In Hachijō stem nominalizations ending in -i/ Ø can also function as sentences. The first such case is likely a further development of the chain construction that has been stretched out beyond a single sentential unit. First, observe the following, where the stem nominalizations form a chain construction (96). (96) Ui-mo kam.i, wai-mo  kam.i chiitoni that.one-also eating I-also   eating small narara. became ‘That one was eating, I too was eating, and (it) has gotten small.’ (Kaneda 2001: 279) The discourse data collected by Kaneda contain a large number of sentences ending in stem nominalizations themselves, as well as those marked by the conjunctive particle te/de. Observe (97). (97) Discussing the summer work on the Hachijō island (Kaneda (2001: 466)) Speaker S: Daaga  tengusaate    itemo    are but   agar.weed.quot say.even that   ni  rokugatu  kara  wa  kugatu dat June   from top September made   wa  zuutto    uren  natte up.to  top all.the.way that become arodaazyan. exist.neg ‘But even talking about agar weeds, it’s not the case that we are concerned about them all the time from June to September.’

194

Speaker K: Adani adani   tengusa   ni cannot.help.but agar.weed dat kakar.i. getting.involved ‘Cannot help but be preoccupied about agar weeds.’ S: Itu  nagiru-ka    to   omot=te. when calm.down-q  quot think=conn ‘Wondering when (the sea) calms down.’ K: nnn. hmm ‘Hmm.’ The second case of the sentential use of Hachijō stem nom­ in­al­iza­tions is where they are supported by an accompanying illocutionary act of conveying such emotions as surprise, happiness, disgust, and lamentation, just as when adnominalending nominalizations in Classical Japanese were used as sentences (see Section 12.3.4.1). (98)  Kaneda (2001: 244, 245) a. Soide zenbu heer.i! and all going.in ‘And all (the balls) went in (to the goal) (hurray)!’ b. Waa  ikisug.i! oh    going.over ‘Oh, (the ball) went too far (shucks)!’ c. Kita   to   mou   to  hiker.i. came  quot think quot returning ‘(He) goes back as soon as he has come (hmm).’ d. Nonda  to   mooneeya  sugu     hinne.Ø. drank  quot think   right.away sleep ‘(He) sleeps right after (I thought) he drank (hmm)’. Examples (98c–d) are typical soliloquies, according to Kaneda, which are accompanied by an evaluation, either positive or negative, of the situation denoted. While some of these sentential uses of stem nominaliza­ tions, especially those that express lamentation like (98b) above, appear possible in Tōkyō speech as well, the Hachijō discourse, in comparison to that of modern central dialects, is characterized by a very high token frequency of nom­in­al­iza­tionbased sentences of this type. Kaneda (2001: 243) speculates, correctly it seems, that a prehistoric form of Japanese, when stem nominalizations were flourishing as grammatical nominalizations, may have had a wider usage pattern

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

japanese and the mainl and dialects including their sentential use, and that the Hachijō pattern above is a residue of the older usage ­practice. Similar chain constructions using stem nominalizations are observed among wide (if not all) varieties of Ryūkyūan. The following are from two southern Yaeyama varieties (99). (99) Ishigaki Shika dialect (field notes October 23 2012, February 2 2013) Gusi yu num.i(=te), yuku.i(=te) wine  acc  drinking (=conn)  resting(=conn) ikonnaara. go.hon.imp ‘Drink wine, rest, and go along.’ (100)  Hateruma (field notes July 9 2013) Gaku gara kaer.i(=sita), si school  from  returning(=conn)  hand  arahe.Ø(=sita),   munu  hyatan. washing(=conn) meal  ate ‘Having returned from school, (and) having washed my hands, I ate a meal.’ An interesting development in Ryūkyūan is that stem nom­ in­al­iza­tions may conclude declarative sentences, as in the Miyako Ōgami form below, where the nominalizer -i has developed a tense-marking function of indicating a past event. This past-tense form is known as a “first past form” (第一過去) in the Ryūkyūan literature (see 101). (101) Ōgami first past tense (Pellard 2012: 108; glosses regularized) kii=ia munu=u fa.i=tu today=top thing=acc eat.nmlz=foc kss.i. come.nmlz/1.pst (Lit.) ‘Today, I ate and then came.’ ‘I ate before ­coming.’

Undoubtedly the first past form has developed from ­adverbial (converb) use of stem nominalizations in chain constructions, where they denote an anterior event, as in all the Ryūkyūan examples above.

12.6 Conclusion This chapter has attempted a description of the structure of Japanese from an unconventional perspective, namely that based on the phenomenon of nominalization, which is believed to play central roles in both nominal and verbal structures not only in Japanese but also in other languages, as demonstrated by a series of recent papers by the present author (Shibatani 2009, 2017, 2018a, 2018b, 2019). At the same time, it has shown that a great deal of insight can be garnered from comparisons of a synchronic structure with both historical and dialectal varieties. Japanese is perhaps unique among Transeurasian languages for being blessed with both rich historical data going back to the eighth century and numerous dialects showing unique patterns of development. Past studies on Japanese would have profited more if these advantages, especially dialectal variations, had been fully made use of. I have demonstrated here that what is hidden or what are lumped together in mainstream dialects are illuminated and reveal their underlying principles of organization once they are set side by side with the data from other, often peripheral and marginalized dialects. My goal has been to demonstrate that rich patterns of structural variations across different dialects and sister languages help us understand how language responds to two opposing forces shaping grammar; namely, one that mo­tiv­ ates a uniform expression of the conceptual unity under­ lying different manifestations of a structure and the other that drives differentiation in form in accordance with the function that a given structure plays in different linguistic and communicative contexts.

195

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 13

Amami and Okinawa, the Northern Ryukyuan languages Yu to N i i naga

13.1 Introduction Ryukyuan (or Luchuan) languages are spoken in the Ryukyu archipelago, which is located in the southwest region in Japan.1 The Ryukyuan languages are sister languages of Japanese, and can be divided into the Northern Ryukyuan languages and the Southern Ryukyuan languages (Pellard 2009: 263; Pellard 2015: 15). The Northern Ryukyuan languages are spoken in about 20 islands (Figure 13.12): Amami-Oshima (the northernmost island in the Ryukyu archipelago), Kikai, Kakeroma, Uke, Yoro, Tokunoshima, Okinoerabu, Yoron, Okinawa, Ie, Iheya, Izena, Sesoko, Kume, Kerama, Kudaka, and so forth (Uemura 1992: 771–82). Among these islands, the first eight islands (Amami-Oshima, Kikai, Kakeroma, Uke, Yoro, Tokunoshima, Okinoerabu, and Yoron) are called Amami islands, and they belong to Kagoshima prefecture. The other islands are called Okinawa islands, and they belong to Okinawa prefecture. All these islands were directly ruled by the Ryukyu kingdom until 1609 (Kinjo and Hattori 1955: 399). The language spoken by the people in the Amami islands is called Amami, and the language spoken by the people in the Okinawa islands is called Okinawa (or Okinawan) (Lawrence 2006: 104–6, 115; Pellard 2009: 263; Pellard 2015: 16).3 The Northern Ryukyuan languages are seriously endangered, and only the old generation can speak them fluently. The younger generation speaks Standard Japanese or a local variety of Japanese. From my fieldwork experience, people older than 65 years of age speak their traditional languages fluently. Since the population of the Amami islands stands at about 110,000 and the proportion of people older than 65 years old is about 31% (Kagoshima Prefecture 2017: 49), the 1   In these days, “Ryukyuan” is predominant as the label to indicate the trad­ ition­al languages spoken in the Ryukyu archipelago (Serafim  2008; Shimoji and Pellard 2010; Shimoji 2017). On the other hand, the label “Luchuan” appears in relatively old papers (Chamberlain 1895b; Loveless 1963). 2   Figure 13.1 is a blank map from the Geospatial Information Authority of Japan, which is partially adjusted by the present author. 3  The dialects in Kikai belong to the Northern Ryukyuan languages; ­however, it is not clear yet whether they belong to Amami or Okinawa (Lawrence 2011: 120; Shirata 2016: 1–3).

number of speakers of Amami is considered to be around 34,500 people. On the other hand, the population of the Okinawa islands stands at about 1,380,000 and the proportion of people older than 65 years old is about 19.6% (Okinawa Prefecture 2017: 1–2), so the number of speakers of Okinawa is considered to be around 270,800 people. Therefore, the total estimated number of people who can speak the Northern Ryukyuan languages is around 305,000 people. Most of the people in Amami and Okinawa do not write in their languages. They write in Standard Japanese with kana (Japanese traditional characters) and kanji (Chinese characters). Each letter of kana expresses sounds of Japanese, and each letter of kanji expresses morphemes of Japanese (Coulmas 2003: 40–1, 78–81). People in the Ryukyu archipelago, as well as those in the Japanese main islands, are accustomed to using these characters. Neither of the character sets, however, can express all phonemes in the Northern Ryukyuan languages, which is one of the reasons why the people who speak the Northern Ryukyuan languages do not use those characters in order to write in their languages. In fact, there have been several attempts to write in the Northern Ryukyuan languages (and the other Ryukyuan languages), but there is little agreement about how to write in those languages (Uemura 1992: 774; Ogawa 2015: i–v). There has been much research about the Northern Ryukyuan languages, and most of it is written in Japanese and written about partial phenomena (e.g. phonemes, accents, verbal morphology, or case markers) of the dialects. I introduce here some research which explains the grammatical outline of the languages, i.e. grammar sketches and reference grammars.4 Another important study will be  discussed in the following sections. The first research ­project to have made a sketch grammar and a dictionary of a Ryukyuan language is Chamberlain (1895b). This paper 4   In the present chapter, a grammar sketch or reference grammar includes some phonological descriptions (vowels, consonants, and pitch patterns), morphological descriptions (nominal morphology, verbal morphology, and adjectival morphology), syntactic descriptions (case marking, some special particles, and clause/sentence types), and descriptions about pronouns (personal pronouns, demonstrative pronouns, and interrogative pronouns).

Yuto Niinaga, Amami and Okinawa, the Northern Ryukyuan languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Yuto Niinaga. DOI: 10.1093/oso/9780198804628.003.0014

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Amami-Oshima Kikai

Kakeroma Uke

Yoro Tokunoshima

Okinoerabu Iheya

Yoron

Izena Ie Kume

Sesoko

Kerama 20km

Okinawa Kudaka

Figure 13.1  Islands where the Northern Ryukyuan languages are spoken

Table 13.1  Grammar sketches and reference grammars of the Northern Ryukyuan languages

research and other grammar sketches are summarized in Table 13.1.

Amami  Naze

*Uemura and Suyama (1993)

 Yuwan

*Niinaga (2010), Niinaga (2014), *Niinaga (2015a)

Kamikatetsu Shirata (2016) Okinawa  Shuri

*Chamberlain (1895b), *Kinjo and Hattori (1955),a Loveless (1963), Yushiya (1999)b

 Naha

*Kinjo (1944)

a   In fact, the sections of phonology and grammar are written by Shiro Hattori (Kinjo and Hattori 1955: 308). b   Yushiya (1999) deviates from the definition of reference grammar (tentatively) decided in note 4 in this chapter, since it does not include phono­logic­al description. It should, however, be recognized as a valuable reference grammar, considering its comprehensive coverage (in total 544 pages) of morphosyntax, and translated texts from Shuri.

described the Shuri dialect of Okinawa (the dialect spoken by the people in Shuri, which used to be the capital of the Ryukyu kingdom). Loveless (1963) is the first comprehensive grammar of a language spoken in the Ryukyu archipelago written using technical linguistic terms, and it described the Shuri dialect of Okinawa. Niinaga (2014) is the first reference grammar of a dialect spoken in Amami, describing the Yuwan dialect of Amami-Oshima. Shirata (2016) is a reference grammar of the Kamikatetsu dialect of Kikai.5 This 5   The genetic relationship of the Kamikatetsu dialect of Kikai is not clear yet, but it is grouped into Amami in this list (cf. Shirata 2016: 1–3).

13.2  Historical connections: Genealogy and contact The Ryukyuan languages are sister languages of Japanese, and all of these languages belong to the Japonic family (Chamberlain 1895b: 2–3; Pellard 2009: 263; Pellard 2015: 15 among others).6 The overall relatedness of the Ryukyuan languages (or Japonic languages) to the other Transeurasian languages is beyond the scope of my present research. Therefore, I will present two types of classification of the Northern Ryukyuan languages: regional classification and genetic relationships. First, I will present a regional classification of the Northern Ryukyuan languages suggested by Uemura (1992: 780–2) in Table 13.2. This classification mainly depends on phonological similarities, but it does not distinguish similarities inherited from a common ancestor from those due to chance resemblance or those due to borrowing.7 This classification will be very useful when I introduce the phono­ logic­ al characteristics among the dialects in the Ryukyu archipelago (see Section 13.3). Asterisks (*) in 6   Chamberlain (1895b: 2–3) is the first paper which referred to the family relationship between Ryukyuan and Japanese. 7   Uemura (1992: 780) also said that his subclassification of the Ryukyuan languages does not reflect the genetic relationships among the dialects in the Ryukyu archipelago.

197

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Ryukyu

Table 13.2  Regional classification of dialects of the Northern Ryukyuan languages a.   b.   c.   d.   e.   f.  

?

*Kikai - Kamikatetsu, Onotsu, Shitooke, Sadeku, etc.

Okinawa

*Northern Amami-Oshima - the Kasari Peninsula (Sani, etc.), Nase, Sumiyo, Yamato, etc. *Southern Amami-Oshima - Setouchi (*Uke, *Yoro, *Kakeroma, etc.) and Ukena (Yuwan, Taken, etc.) *Tokunoshima - Asama, Isen, etc. *Okinoerabu - Wadomari, etc. *Yoron - Mugiyanishi, etc.

Amami Yoron

Northern Okinawa Southern Okinawa Okinoerabu

Amami-Oshima Tokunoshima

Figure 13.2  Genetic relationships among the Northern Ryukyuan languages Pellard 2009

The genetic subgroupings in Figure 13.2 are based on 70 shared innovations (Pellard  2009: 264–5). For example, all the Northern Ryukyuan languages have cognate forms of *weke-ga ‘man’ (Pellard 2009: 264, 267, 270). On the other hand, all the dialects in Amami have cognate forms with *kakazu ‘jaw’, but all the dialects in Okinawa have *kakuzu (Lawrence 2006: 115; Pellard 2009: 264, 266, 270).

*Okinawa islands  g.  

h.  

*Northern Okinawa - Central Yambaru (Bise, the Haneji river, Kushi, Motobu, Nago City and its surrounds, the western part of Nakijin, Onna, *Sesoko), *Ie, *Iheya, *Izena, Northern Yambaru (north of the Shioya Gulf), Southern Yambaru (Kin and its surroundings, etc.), etc. *Central-Southern Okinawa - Itoman, *the islands around the Katsuren Peninsula, *Kerama islands, *Kudaka, *Kume, Naha, *Ou, Shuri, etc.

Uemura 1992: 780–1, 793 a   Uemura (1992: 780) said that some dialects in Uken village have the characteristics of the Northern Amami-Oshima. Indeed, Yuwan (and Taken) have a characteristic of Northern Amami-Oshima, i.e. glottalized nasals. See also Table 13.7.

Table 13.2 indicate names of islands (or regional names which include names of islands) here. Names without asterisks indicate administrative districts, although some of them are no longer used. I quoted here all the names of islands and administrative districts appearing in Uemura (1992: 780–1, 793). In fact, there are more administrative districts, which were not mentioned clearly in Uemura (1992), and they are represented by “and so forth” in Table 13.2. Furthermore, the names of other administrative districts which appear in this chapter are also added to this list. Second, I will present genetic relationships among the Northern Ryukyuan languages, in Figure 13.2.

198

13.3 Phonology There are so many dialects in the Northern Ryukyuan languages that it is difficult to give all of the information about them. Thus, I will choose the Yuwan dialect as a representative of Amami (since Yuwan has been the main subject of my research for over ten years and I know the details of its grammar), and the Shuri dialect as a representative of Okinawa (since there is more research about Shuri than other dialects in Okinawa). Information about the other dialects will be briefly summarized in each section. The following items will be discussed: consonants (Section 13.3.1), vowels (Section 13.3.2), syllable structure (Section 13.3.3), phonology (13.3.4), morphophonology (Section 13.3.5), and suprasegmentals (Section 13.3.6).

13.3.1 Consonants I will present the consonant inventories in the Northern Ryukyuan languages in this section. It is well-known that the consonant inventories in the Northern Ryukyuan languages include so-called “glottalized consonants,”8 which will be explained later. First, I will present the consonant inventories in the Yuwan dialect (Amami), and then I will present those of the Shuri dialect (Okinawa). 8   Yonaguni, one of the Southern Ryukyuan languages, however, also has glottalized consonants. See Table 13.7.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Yuwan (Amami) has 22 consonants, as represented in Table 13.3.9 Some components of phonological oppositions in Yuwan can be analyzed as combinations of a (non-approximant) consonant and an approximant, i.e. /Cj/ or /Cw/, where “C” means any (non-approximant) consonant, as in Table 13.4. Shuri (Okinawa) has 18 consonants as in Table 13.5 (Kokuritsu Kokugo Kenkyuujo 1963: 48–9).10 Some components of phonological oppositions in Shuri can be analyzed as com­

binations of a (non-approximant) consonant and an approxi­ mant, i.e. /Cj/ or /Cw/, where “C” means any (non-approximant) consonant, as in Table 13.6. Three phonemes /ʔ/, /’/, and /ɴ/ in Kokuritsu Kokugo Kenkyuujo (1963) are not adopted in this chapter (see notes in tables 13.3, 13.5, and 13.7). Yuwan has more consonants than Shuri, because the former has more “glottalized consonants” than the latter. Glottalized consonants are indicated by a small upper-right

Table 13.3  Consonants in Yuwan (Amami)   Stop

Bilabial

Alveolar

Alveolo-palatal

Palatal

Velar

Glottal

/p, b/

/t, tˀ, d/

 

 

/k, kˀ, g/

 

a

Affricate

 

 

/c/ [ʨ~ʦ] /cˀ/ [ʨˀ]

 

 

 

Fricativeb

 

/s/ [s~ɕ] /z/ [(d)z~(d)ʑ]

 

 

 

/h/ [h~ç~x~ɸ]

Nasal

/m, mˀ/

/n/c, /nˀ/d

 

 

 

 

Tap

 

/r/ [ɾ]

 

 

 

 

Approximant

/w, wˀ/

 

 

/j, jˀ/

 

 

a   The consonant /c/ is [ʨ] when it precedes /i/; elsewhere, it is [ʦ]; /c/ is optionally glottalized as [ʨˀi] and [ʦˀɨ]. /cˀ/ is always followed by /j/, and /cˀj/ is always [ʨˀ]. b   The consonant /s/ is [ɕ] when it precedes /i/; elsewhere, it is [s]. The consonant /z/ is [(d)ʑ] when it precedes /i/; elsewhere, it is [(d)z]. The consonant /h/ is [ç] when it precedes /i/, it is [x] when it precedes /ɨ/, and it is [ɸ] when it precedes /u/; elsewhere, it is [h]. c   The consonant /n/ is assimilated to the place of articulation of the following consonant, and becomes [ɴ] when it stands in word-final position. The analysis of word-final [ɴ] as /n/ is supported by a phonological rule in Section 13.3.4.2. d   The consonant /nˀ/ is always followed by /j/ (see Table 13.4).

Table 13.4  Consonants followed by approximants in Yuwan (Amami)  

Bilabial

Alveolar

Alveolo-palatal

Palatal

Velar

Glottal

Stop

/pj, bj/

 

 

 

/kj, kˀj, gj/ /kw, kˀw/

 

Affricate

 

 

/cj/ [ʨ] /cˀj/ [ʨˀ]

 

 

 

Fricative

/hw/ [ɸ]

/sj/ [ɕ] /zj/ [(d)ʑ]

 

/hj/ [ç]

 

 

Nasal

/mj/

 

 

/nj/ [ɲ] /nˀj/ [ɲˀ]

 

 

9   All the Yuwan data in the present chapter are taken from my research data. 10  The glottal phonemes /ʔ/ and /’/ in Kokuritsu Kokugo Kenkyuujo (1963: 48) are deleted from table 13.5, since (1) the glottal stop [ʔ] can be analyzed as a part of the phonetic realization of vowels standing in wordinitial position, and (2) the glottal phoneme /’/ can be analyzed as /j/ (when it precedes /i/ or /e/) and /w/ (when it precedes /u/ or /o/) (cf. Nishioka and Nakahara 2006: 192–3). In addition, an archiphoneme /q/ in Kokuritsu Kokugo Kenkyuujo (1963: 46–7) is deleted from this table, since it can be analyzed as a geminate (of the following consonant). Most of the words with /q/ have it in word-medial position, for example “ʔaqkaɴ” [ʔakkaɴ] ‘(I)

don’t walk’ (Kokuritsu Kokugo Kenkyuujo 1963: 47). (I do not use [ː] for consonants in this chapter.) A very restricted number of words, however, have /q/ in word-initial position (Kokuritsu Kokugo Kenkyuujo (1963: 47). Wordinitial /q/ is supposed to glottalize its following consonant. I do not analyze them as glottalized consonants /kˀ, cˀ, sˀ/, since (1) there are few words that have these glottalized onsets, and (2) Uemura (1992: 793) also did not assume that there are glottalized onsets in obstruents (especially in stops) in Shuri. They are analyzed as a geminate (of the following consonant) in this chapter. An exhaustive list of such examples is—kkwa ‘child’, ccju ‘person’, cci (come.seq), ss-a (do-int), ss-i (do-imp), and ssi (do.seq) (Kokuritsu Kokugo Kenkyuujo 1963: 47).

199

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.5  Consonants in Shuri (Okinawa) a  

Bilabial

Alveolar

Alveolo-palatal

Palatal

Velar

Glottal

Stop

/p, b/

/t, d/

 

 

/k, g/

 

Affricate

 

 

/c/ [ʨ] /z/ [(d)ʑ]

 

 

 

Fricativeb

 

/s/ [s~ɕ]

 

 

 

/h/ [h~ç~ɸ]

Nasal

/m/

/n/ , /nˀ/

 

 

 

 

Tap

 

/r/ [ɾ]

 

 

 

 

Approximant

/w, wˀ/

 

 

/j, jˀ/

 

 

c

d

  The so-called “male gentry speech” (Kinjo and Hattori 1955: 319; Loveless 1963: 1) has a different phonetic realization of three phonemes /s/, /c/, and /z/. That is, /s/ is always [s], /c/ is always [ʦ], and /z/ is always [ʣ] (Kinjo and Hattori 1955: 319–22). In addition, /sj/ [ɕ] can precede /i/ and /e/, and there arise /cj/ [ʨ] and /zj/ [(d)ʑ] in the speech style (Kinjo and Hattori 1955: 319–322). Now I avoided the analysis by Kinjo and Hattori (1955: 319–22), which assumes the existence of three new phonemes /ş/ [s], /ç/ [ʦ], and /z̧/ [ʣ]. Instead, I reduced them into phonetic realization and phonotactics. b   The consonant /s/ is [ɕ] when it precedes /i/ or /e/; elsewhere, it is [s] (Kokuritsu Kokugo Kenkyuujo 1963: 49–50). The consonant /h/ is [ç] when it precedes /i/, and it is [ɸ] when it precedes /u/; elsewhere, it is [h]. c   The consonant /n/ is assimilated to the place of articulation of the following consonant, and becomes [ɴ] when it stands in word-final pos­ ition. The analysis of word-final [ɴ] as /n/ is supported by a phonological rule in Section 13.3.4.2. d   The nasal /nˀ/ (/ʔɴ/ in the original text) occupies only the pre-syllable in the word-initial position (see Section 13.3.3), and is assimilated to the articulatory place of the following consonant. a

Table 13.6  Consonants followed by approximants in Shuri (Okinawa)  

Bilabial

Alveolar

Alveolo-palatal

Palatal

Velar

Glottal

Stop

/pj, bj/

 

 

 

/kw, gw/

 

Fricative

/hw/ [ɸ]

 

 

/hj/ [ç]

 

 

Nasal

/mj/

 

 

/nj/ [ɲ]

 

 

glottal stop (ˀ), i.e. /tˀ/, /wˀ/, and /mˀ/.11 At least in Yuwan, all the glottalized consonants appear in word-initial pos­ ition only (or the stem-initial position in the compounds): ˀwaa ‘pig’ or kˀuru+ˀwaa ‘black pig’. Glottalized consonants are said to have an initial glottal stop in the case of sonorants such as /wˀ/ [ʔ͡w] or /mˀ/ [ʔ͡m] (Niinaga, Aoi, and Nakagawa 2011: 294), or non-aspiration or very short aspiration in the case of obstruents (Uemura and Suyama 1993: 392). Some of the glottalized sonorants resulted from vowel deletion in word-initial syllables: *[ʔuma] ‘horse’ > [ʔ͡ma]; *[ʔinoʨi] ‘life’ > *[ʔiɲuʨi] > [ʔɲuʨi].12 Some glottalized obstruents resulted from the deletion of devoiced initial ­vowels and the following double articulation of voiceless stops: *[putari] ‘two people’ > *[pu̥tai] > *[p͡tai] > [tˀai]; *[pitu] ‘human’ > *[pi̥ʨu] > *[p͡ʨu] > [ʨˀu] (cf. Uemura 1992: 794). The existence of glottalized onsets in most Ryukyuan languages is summarized in Uemura (1992: 793). The cor­re­ 11   The most recent IPA chart (revised in 2015) does not have this diacritic, but I will use it for the sake of convenience. 12   There is also palatalization of [n] into [ɲ] (because of the preceding [i]) and vowel raising of [o] into [u] (which occurs in most of the lexicons in the Ryukyuan languages).

200

spond­ence between the dialects and the kinds of glottalized onsets can be summarized in Table 13.7, where the capital “A” represents Amami, and “O” represents Okinawa. See Table 13.2 for the regional classification. The plus sign “+” in Table 13.7 means that the dialects have the opposition between glottal onsets and non-glottal onsets in the specified environment. From Table  13.7, we can extract an implicational relationship as follows (1). (1) Implicational relationship of the glottalized onsets in the Ryukyuan languages: If a dialect has a glottalized onset in nasals, then it has a glottalized onset in stops and approximants.13 From this implicational relationship, we can suppose that glottalized onsets in nasals historically appeared later than those in stops and approximants (only if it is not the case that the glottalized onsets in nasals once appeared and then disappeared in Type II of Table 13.7). 13 The contraposition is also true. If a dialect does not have a glottalized onset in either approximants or stops, then it does not have a glottalized onset in nasals.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.7  Glottalized onsets Type Regional classification

Stops

Approximants

Nasals

I

*Northern Amami-Oshima (A), Yuwan (A),

+

+

+

 

*Tokunoshima (A), *Okinoerabu (A),

 

Central Yambaru (O), *Ie (O), *Ou (O)

II

*Southern Amami-Oshima (A)

+

+

-

 

(excluding part of Uken)

 

 

 

III

*Kikai (A), *Kudaka (O)

+

-

-

IV

*Yoron (A), *Central-Southern Okinawa (O) (excluding *Kume, part of Itoman, *Ou)

-

+

-

V

*Kume (O), part of Itoman (O),

-

-

-

 

the Southern Ryukyuan languagesb

 

 

 

a

Asterisks (*) indicate names of islands (or regional names which include names of islands) here. See also Table 13.2. a   Shuri is regionally classified into Central-Southern Okinawa, and it has a glottalized nasal /nˀ/ as in Table 13.5. The nasal /nˀ/, however, only fills the pre-syllable (see also Section 13.3.3), and cannot be an onset (i.e. cannot be followed by a vowel), e.g. /nˀma/ [ʔ͡mma] ‘horse’ (Kokuritsu Kokugo Kenkyuujo 1963: 44). Thus, the column of “Nasals” in Type 4 is checked as minus “-,” because Table 13.7 is summarized about the glottalized “onsets.” b   Yonaguni (a Southern Ryukyuan language), however, belongs to type III.

13.3.2 Vowels Yuwan (Amami) has mainly six short vowels and six long vowels. Both kinds of vowels are given in Tables 13.8 and 13.9. Long vowels are interpreted as a sequence of short vowels in this chapter, e.g. /ii/ [iː], since they are articulated in the same position as short vowels, and they have double morae of short vowels (see Section 13.3.6). The vowel /e/ is rarely used in Yuwan, e.g. sinsjei [ɕinɕe̞i] ‘teacher’ or ude [ʔude̞] ‘hey.’ All the diphthongs in Yuwan are composed of a vowel plus /i/, e.g. mai ‘hip’ or jui ‘lily’. Shuri (Okinawa) has mainly three short vowels and five long vowels (Kokuritsu Kokugo Kenkyuujo 1963: 27). Both vowels are given in Tables 13.10 and 13.11. The short ­vowels /e/ and /o/ are mainly used before /n/, e.g. kenken [ke̞ŋke̞ɴ] ‘a sound of gong’, tontonmii [to̞nto̞mmiː] ‘ducks and drakes’, and coccongwaa [ʨo̞tʨo̞ŋɡʷɑ̟ː] ‘winter song of a bush warbler’; otherwise, they are rarely used in Shuri, e.g. haberu [χɑ̟be̞ɾu] ‘butterfly’ and horohoro [ho̞ɾo̞ho̞ɾo̞] ‘the rustle of clothing’ (Kokuritsu Kokugo Kenkyuujo 1963:

27, 319, 522). The long vowels /ee/ and /oo/ are very common in Shuri, e.g. kuneeda [kune̞ːdɑ̟] ‘the other day’ and hoocaa [ho̞ːʨɑ̟ː] ‘kitchen knife’ (Kokuritsu Kokugo Kenkyuujo 1963: 213, 333). Table 13.9  Long vowels in Yuwan (Amami)  

Front

Central

Back

High

/ii/ [iː]

/ɨɨ/ [ɨː]

/uu/ [uː]

Mid

(/ee/ [e̞ː])

/əə/ [ɜː]

/oo/ [o̞ː]

Low

 

 

/aa/ [ɑ̟ː]

Table 13.10  Short vowels in Shuri (Okinawa)  

Front

Central

Back

High

/i/ [i]

 

/u/ [u]

Mid

(/e/ [e̞])

 

(/o/ [o̞])

Low

 

 

/a/ [ɑ̟]

Table 13.11  Long vowels in Shuri (Okinawa)

Table 13.8  Short vowels in Yuwan (Amami)  

Front

Central

Back

 

Front

Central

Back

High

/i/ [i]

/ɨ/ [ɨ]

/u/ [u]

High

/ii/ [iː]

 

/uu/ [uː]

Mid

(/e/ [e̞])

/ə/ [ɜ]

/o/ [o̞]

Mid

/ee/ [e̞ː]

 

/oo/ [o̞ː]

Low

 

 

/a/ [ɑ̟]

Low

 

 

/aa/ [ɑ̟ː]

201

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Most of the dialects in Amami, i.e. those in AmamiOshima, Tokunoshima, and Kikai (Onotsu and Shitooke), have the same vowel inventories as Yuwan (Uemura 1992: 788–9; Kibe 2011: 25). Other dialects in Amami, i.e. those in Kikai (except for Onotsu and Shitooke), Okinoerabu, and Yoron, and all the dialects in Okinawa have the same vowel inventories as Shuri (Uemura 1992: 788–9; Kibe 2011: 25).

13.3.3  Syllable structure The syllable structures in Yuwan (Amami) are given in Table 13.12. The slots in parentheses are optional in the syllables. The Greek letter μ indicates that the slot above it has one mora. A syllable in Yuwan has at most two morae (see also Section 13.3.4.3). Phonemes able to fill the syllable slots, i.e. onset, nucleus, and coda, are given in Table 13.13. Table 13.12  Syllable structure in Yuwan (Amami) Syllable slots (Onset-1 (Onset-2)) Nucleus (Coda) Morae

 

 

μ

μ

Examples

 

 

 

 

‘horse’

mˀ

 

a

 

‘I’

w

 

a

n

‘my’

w

 

a

a

‘pig’

wˀ

 

a

a

‘cat’

m

j

a

a

‘child’

kˀ

w

a

 

‘father’

cˀ

j

a

n

Table 13.13  Phonemes able to fill the syllable slots Syllable slots Onset-1

Onset-2 Nucleus Coda

202

glottalized consonants in word-initial position non-glottalized consonants /w/ and /j/ all vowels the same vowel as that in nucleus (i.e. long vowels) /i/ /n/ in word-final position /m/ at a verbal morpheme boundary: jum-gadɨ (read.inf-until) voiceless non-glottalized obstruents (except for /h/), i.e. /p, t, k, c, s/, in non-word-final position: at.taa ‘those people’

Briefly speaking, the onset slots are filled by consonants, the nucleus slot is filled by a vowel, and the coda slot is filled by either a vowel or a consonant. The consonants /n/ and /m/ are homorganic nasals in the coda slot, e.g. nintəə [nintɜː] ‘group,’ anmaa [ʔɑ̟mmɑ̟ː] ‘mother’, jinga [jiŋɡɑ̟] ‘man’, jumba [jumbɑ̟] (read.csl), jumna [junnɑ̟] (read.proh), and jumgadɨ [juŋɡɑ̟dɨ] (read.inf.until). The syllable structure in Shuri is given in Table 13.14.14 The slots in parentheses are optional in the syllables. This syllable structure is different from that of Yuwan since the Shuri dialect has the “pre-syllable.”15 The pre-syllable is a syllable that only appears in the word-initial position and is filled by /n/, /nˀ/, /k/, /c/, or /s/, e.g. n.na [nnɑ̟] ‘everyone’, nˀ.na [ʔ͡nnɑ̟] ‘dung’, k.kwa [kˀʷa] ‘child’, c.cju [ʨˀu] ‘person’, and s.sa [sˀa] (do.int) (Kokuritsu Kokugo Kenkyuujo 1963: 47, 432, 436). The pre-syllable can be filled by two nasals, e.g. nn.zun [nːʑuɴ] ‘look at’, but such examples are very rare in Shuri.16 Phonemes able to occupy positions in the pre-syllable and regular syllables are given in Table 13.15. Briefly speaking, the pre-syllable is filled by nasals or obstruents (/k/, /c/, or /s/); in regular syllables, onset slots are filled by consonants, the nucleus slot is filled by a vowel, and the coda slot is filled by either vowel or consonant. A pre-syllable has at most two morae, and a regular syllable also has at most two morae.17

13.3.4  Phonological rules In this chapter, the difference between phonological rules (Section 13.3.4) and morphophonological rules (Section 13.3.5) is that the former are applied to all of the lexicon in a dialect, but that the latter are applied only to a restricted group (such as compounds or verbs) of a dialect. It is difficult to overview the (morpho-)phonological rules of the dialects in the Northern Ryukyuan languages, since there are few grammars or grammar sketches of  them (see Table  13.1). Thus, I will briefly introduce the (morpho-)phono­logic­al rules in Yuwan (Amami) and Shuri (Okinawa).

  The syllable structure in Shuri is adjusted to that in Yuwan.   The term “pre-syllable” is taken from the analysis of the Irabu dialect (Southern Ryukyuan language) in Shimoji (2017: 18). 16   There are only eight items that have two nasals in the pre-syllable in Kokuritsu Kokugo Kenkyuujo (1963: 430–8). 17   Shuri has, for example, the word jaan ‘last year’, and whether it consists of a regular syllable or two regular syllables is not clear so far. If it is made of one regular syllable, then that means the regular syllable is made of three morae, and that is different from the analysis in Table 13.14. 14 15

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.14  Syllable structure in Shuri (Okinawa) Syllables

[Pre-syllable]

[Regular syllable]

Syllable slots

[Nucleus

(Coda)]

[(Onset-1

(Onset-2))

Nucleus

(Coda)]

Morae

μ

μ

 

 

μ

μ

Examples

 

 

 

 

 

 

‘turnip’

nˀ

n

d

 

i

i

‘look at’

n

n

z

 

u

n

‘everyone’

n

 

n

 

a

 

‘dung’

nˀ

 

n

 

a

 

‘elder sister’

nˀ

 

m

 

i

i

‘child’

k

 

k

w

a

 

‘sky’

 

 

t

 

i

n

‘my’

 

 

w

 

a

a

‘pig’

 

 

wˀ

 

a

a

‘time-out’

 

 

m

 

a

i

‘strange’

 

 

m

j

u

u

‘volume’

 

 

k

w

a

n

‘exist’

 

 

 

 

a

n

All the examples of Shuri in this table are taken from Loveless (1963: 17, 25) and Kokuritsu Kokugo Kenkyuujo (1963: 47) with some phonemes adjusted to the interpretation of this chapter.

Table 13.15  Phonemes able to fill the syllable slots in Shuri (Okinawa) Syllable slots Phonemes Pre-syllable  Nucleus

/n/, /nˀ/, or the same obstruent as that in the onset-1 (/k/, /c/, or /s/)

Coda

/n/

Regular syllable  Onset-1

all consonants

Onset-2

/w/ and /j/

Nucleus

all vowels

Coda

the same vowel as that in the nucleus /i/a; /n/ in word-final position; voiceless obstruents (except for /h/), i.e. /p, t, k, c, s/, in non-word-final position: ap.pii ‘elder brother’

  As far as I know, there is no research available in linguistic litera­ture which deals with diphthongs in Shuri. There are, h ­ owever, some examples of /Vi/ (“V” means any vowel here), e.g. mai ‘time-out’ or kai (all) (Kokuritsu Kokugo Kenkyuujo 1963: 301, 359).

13.3.4.1  Sonorant deletion in Yuwan (Amami) There are no /wi/ or /ri/ sequences in Yuwan (Amami), except for three items, i.e. piri ‘tail end’, rikkoo ‘(by) foot’, and kiri ‘fog’. If this type of sequence occurs at a morpheme boundary, the bilabial approximant /w/ or the tap /r/ is deleted. For example, if the verbal root koow- ‘buy’ is followed by the verbal affix -i (inf), then they become ko-i (< koo-i < koow-i).18 If the verbal root ar- ‘exist’ is followed by the verbal affix -i (inf), then they become a-i (< ar-i). In Shuri (Okinawa), the sequence /wi/ does not exist, but there is the sequence of /ri/, e.g. nur-i (ride-imp) ‘Ride!’ (Kinjo and Hattori 1955: 338).19 Shuri does not have any verbal root which ends with /w/ (Kinjo and Hattori 1955: 332)

13.3.4.2  Epenthetic vowel in Yuwan (Amami) and Shuri (Okinawa) Every syllable in Yuwan (Amami) should have a nucleus filled by a vowel. In other words, Yuwan does not have a syllable without a vowel. For example, if the noun kin

a

  There is also a vowel shortening rule in Section 13.3.4.3 /ooi/ > /oi/.   The imperative affix in Yuwan is -ɨ, and then there is no sequence of /ri/, e.g. nur-ɨ (ride-imp) ‘Ride!’ 18 19

203

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga ‘clothes’ is followed by the particle =n ‘also’, then they become kin=un (not kin=n), where an epenthetic vowel /u/ is inserted. Such vowel insertion does not occur if there is a nucleus filled by a vowel, e.g. hana=n (flower=also), usi=n (cow=also) or siju=n (soup=also). Similarly, every “regular” syllable in Shuri (Okinawa) should have a nucleus filled by a vowel (see also Table 13.14). For example, if the noun cin ‘clothes’ is followed by the particle =n ‘also,’ then it becomes cin=un (not cin=n) ‘clothes too’ (Loveless 1963: 148), where an epenthetic vowel /u/ is inserted. Such vowel insertion does not occur if there is  a  nucleus filled by a vowel, e.g. hana=n (flower=also), usi=n (cow=also), or siru=n (soup=also) (Kinjo and Hattori 1955: 330). It is important to note that epenthetic vowels occur only in regular syllables in Shuri. In a pre-syllable, i.e. word-initial syllabic nasals (see Section 13.3.3), a nucleus can be filled by a non-vowel, and there is no epenthetic vowel, e.g. nnzun (not nunzun) ‘look at’ (Kinjo and Hattori 1955: 334).

13.3.4.3  Vowel shortening in Yuwan (Amami) A syllable in Yuwan has at most three morae (see Section 13.3.3), but a syllable having three morae tends to drop to two morae with vowel shortening. For example, if the verbal root koow- ‘buy’ is followed by the verbal affix -i (inf), then it becomes ko-i via koo-i (< koow-i) (about the deletion of /w/, see Section 13.3.4.1, where the long vowel /oo/ becomes /o/. If the demonstrative word attaa ‘those people’ is followed by the particle =n ‘also’, then it becomes atta=n (< attaa=n), where the long vowel /aa/ becomes /a/. A regular syllable in Shuri has at most two morae. However, we can find many examples where a syllable seems to have three morae. For example, judeen (read.res), judoon (read.prog), and haa=n (tooth=also) (Kinjo and Hattori 1955: 330, 340). We need to clarify whether they consist of a regular syllable or two regular syllables in the future research (see also note 17).

13.3.4.4  Non-existence of vowel harmony in Yuwan (Amami) and Shuri (Okinawa) For those interested in the relationship between the Northern Ryukyuan languages and the Transeurasian languages, I will mention the non-existence of vowel harmony in the Northern Ryukyuan languages. For example, the word siju ‘soup’ in Yuwan, where a front vowel /i/ and a back vowel /u/ can occur in the same morpheme. Shuri has also siru ‘soup’. Other examples are ita ‘board’ and uta ‘song’ in Yuwan, where the high vowels /i/ and /u/ can occur with a low vowel /a/ in the same morpheme. Shuri has also ica ‘board’ and uta ‘song’. In conclusion, neither vowel backness nor vowel height induces vowel harmony in Yuwan and Shuri.

204

13.3.5  Morphophonological rule 13.3.5.1  Sequential voicing (Rendaku) The initial obstruent of the non-initial stem of a compound may be voiced if it is originally voiceless. This voicing is called sequential voicing (or “Rendaku” in Japanese linguistics). Let us see an example in Yuwan. A compound made of kˀuru ‘black’ and sataa ‘sugar’ becomes kˀuru+zataa ‘black sugar’, where the initial consonant of sataa is voiced (i.e. /s/ > /z/). The consonant /h/ in Yuwan does not have a voiced counterpart, and it becomes /b/ when it undergoes sequential voicing: koo ‘river’ + hunɨ ‘boat’ > koo+bunɨ ‘river boat’. There is also sequential voicing in Shuri. For example, a compound made of maru ‘circle’ and tama ‘jewel’ becomes maru+dama ‘circular jewel’, where the initial consonant of tama ‘jewel’ is voiced (i.e. /t/ > /d/) (Loveless 1963: 140).20 Another example is a compound made of jamatu ‘Japan’ and huni ‘ship’, which becomes jamatu+buni ‘Japanese ship’.

13.3.5.2  Fusion of topic marker The Northern Ryukyuan languages have the same topic marker =ja, and it is fused with the preceding word. In Yuwan, the topic marker =ja undergoes the following morphophonemic changes. The topic marker retains its form when it follows a vowel sequence (i.e. long vowels or diphthongs). If it follows a single vowel, it is fused with the word-final single vowel of the preceding words. If it follows a word-final nasal, it becomes =na. The capital “C” represents consonants and “V” represents vowels in the following examples: Fusion of topic marker in Yuwan (Amami) a. /ci/, /si/, or /zi/ + =ja (top) > cjəə, sjəə, or zjəə (e.g. kuci ‘mouth’ + =ja > kucjəə; nusi (rfl) + =ja > nusjəə; tuzi ‘wife’ + =ja > tuzjəə) b. /Ci/ (C is not /c, s, z/) + =ja (top) > Cəə (e.g. kˀubi ‘neck’ + =ja > kˀubəə; tuki ‘time’ + =ja > tukəə) c. /Cɨ/ + =ja (top) > Cəə (e.g. kurɨ ‘this’ + =ja > kurəə; amɨ ‘rain’ + =ja > aməə) d. /Cu/, /Co/, or /Ca/ + =ja (top) > Coo (e.g. wunagu ‘woman’ + =ja > wunagoo; juuto ‘Yuto (personal name)’ + =ja > juutoo; ura ‘you’ + =ja > uroo)

20   In the original text, maru+dama was translated into ‘jewel circle’ (Loveless 1963: 140). However, maru means ‘circle’ and tama means ‘jewel’ (Kokuritsu Kokugo Kenkyuujo 1963: 363, 508), and most of the examples in the original page show that the semantic head of a compound is on the right component in Shuri. Thus, I modify the translation into ‘circular jewels’ in this chapter.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages e. /VV/ + =ja (top) > VVja (e.g. jaa ‘house’ + =ja > jaaja; mai ‘hip’ + =ja > maija)



f. /Vn/ + =ja (top) > Vnna (e.g. ɨn ‘dog’ + =ja > ɨnna; wan ‘I’ + =ja > wanna) ,

j. /Vn/ + =ja (top) > Vnoo (e.g. sansin ‘sanshin (Okinawan musical instrument)’ + =ja > sansinoo) (Nishioka and Nakahara 2006: 31–2; Loveless 1963: 111)

Morphophonemic rules of the topic marker =ja in Shuri are a little simpler than those of Yuwan.

13.3.6 Suprasegmentals

Topic marker in Shuri (Okinawa)

The Northern Ryukyuan languages, as well as most of the Japonic languages, have lexical prosody. For example, Yuwan (Amami) has three accent patterns (Table 13.16). The unit of accent is a word (or a word with enclitics). For example, the word haa ‘leaf’ belongs to the accent pattern in Table 13.16 (a). If it is pronounced in isolation, i.e. without enclitics, it is realized as /háà/, but if it is pronounced with =nu (nom), it is realized as /háá=nù/, and with =gadɨ (lmt), it is realized as /háá=gád�̀/. In all of these cases, the fall comes after the penultimate mora. On the other hand, the word haa ‘tooth’ belongs to the accent pattern in Table 13.16 (b). If it is pronounced with =gadɨ (lmt), it is realized as /háá=gàdɨ̀/, where the fall comes after the syllable including the second mora. Thus, haa ‘leaf’ and haa ‘tooth’ can be distinguished by the accent patterns. When haa ‘tooth’ is pronounced in isolation, or with =nu (nom), however, it is realized as /háá=nù/ or /háà/,21 and it cannot be distinguished from haa ‘leaf’ in those cases.

g. /Ca/ + =ja (top) > Caa (e.g. sima ‘island’ + =ja > simaa) h. / Ci/ or /Ce/ + =ja (top) > Cee (e.g. ami ‘rain’ + =ja > amee; korokke ‘croquette’ + =ja > korokkee) i. /Cu/ or /Co/ + =ja (top) > Coo (e.g. kumu ‘cloud’ + =ja > kumoo; razio ‘radio’ + =ja > razioo)



Table 13.16  Three accent patterns in Yuwan (Amami) a.

Falling after the penultimate mora

b.

Falling after the syllable containing the second mora

c.

Rising after the penultimate mora

Table 13.17  Examples of three accent patterns in Yuwan (Amami) Patterns

Glosses

Examples

 

 

Isolation

X=nu

X=na

X=gadɨ

 

 

 

(NOM)

‘also’

(LMT)

a.

‘leaf’

háà

háá=nù

há=ǹ

háá=gád�̀

 

‘sandal’

sábà

sábá=nù

sábá=ǹ

sábá=gád�̀

 

‘saliva’

júdáì

júdáí=nù

júdáí=ǹ

júdáí=gád�̀

b.

‘tooth’

háà

háá=nù

há=ǹ

háá=gàd�̀

 

‘last year’

kúdù

kúdú=nù

kúdú=ǹ

kúdú=gàd�̀

 

‘left’

hízjáì

hízjáí=nù

hízjáì=ǹ

hízjáí=gàd�̀

 

‘lady’

mə́ə́ràb�̀

mə́ə́ràb�̀=nù

mə́ə́ràb�̀=ǹ

mə́ə́ràb�̀=gad�̀

c.

‘needle’

hàí

hàì=nú

hàì=ń

hàì=gàd�̀

 

‘pan’

nàb�̀

nàb�̀=nú

nàb�̀=ń

nàb�̀=gàd�̀

 

‘sword’

kàtàná

kàtànà=nú

kàtànà=ń

kàtànà=gàd�̀

 When haa ‘leaf ’ and haa ‘teeth’ are followed by =n ‘also’, they undergo vowel shortening (see Section 13.3.4.3).

a

21

  If the fall is located word-finally, then the fall is realized after the penultimate mora.

205

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Furthermore, the word hai ‘needle’ belongs to the accent pattern in Table 13.16 (c). If it is pronounced in isolation, with =nu (nom), or with =gadɨ (lmt), it is realized as /hàí/, / hàì=nú/, or /hàì=gàd�̀/ respectively. In all these examples, the rise comes after the penultimate mora. All nominals in Yuwan can be classified into these accent patterns. I will present some examples in Table 13.17. It is important to note the prosody of hizjai ‘left’. When it is pronounced with =gadɨ (lmt), it is realized as /hízjáí=gàd�̀/. I did not define the pattern in Table 13.16 (b) as “falling after the first syllable” or “falling after the second mora” because neither of these characterizations can explain the lexical prosody of / hízjáí=gàd�̀/. Furthermore, the combination of hizjai ‘left’ with =n ‘also’ is realized as /hízjáì=ǹ/, which shows that the word is being divided into three syllables as /hí.zjá.ìǹ/ (not /hí.zjáìǹ/). This fact shows that Yuwan does not have a syllable composed of cvvc, because /zjain/ (of hizjai=n) is divided into /zja.in/ (see also Section 13.3.3 and Niinaga and Ogawa 2011: 242–3). Shuri (Okinawa) has two accent patterns (Table 13.18). The unit of accent is a word (or a word with enclitics) (“bunsetsu” in Kinjo and Hattori 1955: 53). All nominals in Shuri can be classified into these accent patterns. I will present some examples in Table  13.19, which are cognates of the examples in Yuwan. Yuwan has three accent patterns, but Shuri has two. If you look at the examples in each dialect, you will notice that the words of the accent pattern in Table 13.16 (a) in Yuwan are cognate with the words of the accent pattern in Table 13.18 (a) in Shuri. Similarly, the words of the accent pattern in Table 13.16 (b–c) in Yuwan are cognate with the Table 13.18  Two accent patterns in Shuri (Okinawa) a. If a word has more than two morae, the pitch falls after the second mora; if a word has just two morae, the pitch falls after the first mora.a b. Flat pitch Kokuritsu Kokugo Kenkyuujo 1963: 53 a   Shuri does not have any free words composed of just one mora (Kokuritsu Kokugo Kenkyuujo 1963: 35). If the first syllable (of a word composed of more than two morae) has a long vowel, or if the second mora is /n/, the pitch of the second mora tends to be low, e.g. /náàsààtì/ ‘two days later’ (Kokuritsu Kokugo Kenkyuujo 1963: 53).

Table 13.19  Examples of two accent patterns in Shuri (Okinawa) a. hwáà ‘leaf’, sábà ‘sandal’, júdáì ‘saliva’ b. hāā ‘tooth’, kūzū ‘last year’, hwīzāī ‘left’, mījārābī ‘lady’, hāāī ‘needle’, nāābī ‘pan’, kātānā ‘sword’ Kokuritsu Kokugo Kenkyuujo 1963

206

words of the accent pattern in Table 13.18 (b) in Shuri. The phonetic realization of the accent patterns in Yuwan is different from that for Shuri, but the grouping of words sharing the same accent patterns in Yuwan corresponds to that in Shuri. Diachronically speaking, it is said that Proto-Ryukyuan had three accent patterns (Matsumori 1999: 32–3). The number of accent patterns has been retained in Yuwan, but has been reduced from three to two in Shuri. The correspondence of cognates in accent patterns is seen among all the dialects in Ryukyuan languages, as well as the dialects in Japanese (Hattori 1978–1979; Igarashi 2016).

13.4  Word morphology and phrasal constituents In this section, I will present the morphology of pronouns (Section 13.4.1), nominals (Section 13.4.2), adnominals (Section 13.4.3), verbs (Section 13.4.4), adjectives (Section 13.4.5), and compounds (Section 13.4.6).

13.4.1 Pronouns I will present three kinds of pronouns, i.e. personal pronouns (Section 13.4.1.1), demonstrative pronouns (Section 13.4.1.2), and interrogative pronouns (Section 13.4.1.3). These pronouns cross over three word classes, i.e. nominals, adnominals, and adverbs (although personal pronouns do not become adverbs). In addition, there is another pronoun, i.e. the reflexive pronoun. The reflexive pronoun in Yuwan (Amami) is nusi22 and that in Shuri (Okinawa) is duu (Kokuritsu Kokugo Kenkyuujo 1963: 179–80).

13.4.1.1  Personal pronouns Yuwan (Amami) has two types of personal pronouns, one of which belongs to nominals (personal pronominal nominals), and the other to adnominals (personal pronominal adnominals). The existence of the dual forms, which is marked by the affix -ttəə (du) in Yuwan, is sporadically reported in the description of the dialects in Amami (Yamada 1981: 38–42; Niinaga 2010: 50; Shigeno and Shirata 2016: 114, 119, 125). Every form (all personal pronominal nominals) in Table 13.20 can be the head of an NP. All the case markers can follow these forms, e.g. wa-n=ga (1sg=nom), wa-n=ba (1sg=acc), wa-n=tu (1sg=com), wa-n=sjɨ (1sg=ins), and wa-n=kara (1sg=abl). Genitive markers, however, cannot follow their 22 The main consultant of the present author in Yuwan uses nusi as a reflexive pronoun, but a few other people in Yuwan use duu.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.20  Personal pronominal nominals in Yuwan (Amami) Number Person

Politeness

Singular Dual

Plural

1st

 

wa-n

wa-ttəə

waa-kja

2nd

Honorific

na-n

na-ttəə

naa-kja

 

Non-honorific

ura

ura-ttəə

ura-kja

3rd

 

-

na-ttəəa

-

  The third-person dual pronoun na-ttəə means ‘those two people’. It is possible that its root na- (or maybe nat- at an earlier time) has a historical relationship with the third-person reflexive pronoun nat(Miyara 1995) in Ishigaki.

a

Table 13.21  Personal pronominal adnominals in Yuwan (Amami) Number Person

Politeness

Singular

Plural

1st

 

wa-a

waa-kja-a

2nd

Honorific

na-a

naa-kja-a

 

Non-honorific

ura-a

ura-kja-a

singular and plural forms.23 When the singular and plural forms modify another noun, they take the forms (personal pronominal adnominals) in Table 13.21, e.g. wa-a inja-sa-n kuzɨ (1-adnz small-adj-ptcp shoe) ‘my little shoe(s)’, where wa-a ‘my’ modifies kuzɨ ‘shoe(s)’. The form to mean ‘my’ should be wa-a (1-adnz), not *wa-n=ga (1-sg=gen). The form wa-a is a free word and can modify another noun without an overt genitive case. I assume that words of this kind belong to the word class “adnominals” in this chapter. I avoid the term “possessive pronoun,” since it has a vague meaning, and it cannot denote their word classes (i.e. belong to nominals, adnominals, verbs, adjectives, or adverbs). Third-person singular and plural forms are denoted by demonstrative pronouns (see Section 13.4.1.2). The personal pronouns of Shuri (Okinawa) are given in Table 13.22.24 As far as I know, there are no dual forms in the dialects of Okinawa. The 1sg has three forms, i.e. wan, waa, and wanni. The choice of variant forms depends on the following particles (see Table 13.23). The form wan is the 23  Dual forms can be followed by a genitive marker: wa-ttəə=ga mun (2-du=gen thing) ‘something belonging to the two of us’. 24   All the morpheme boundaries and glossing of Shuri in this chapter depend on the present author’s analyses. I recognize a boundary preceding -taa (pl), since -taa (pl) can also be used with other nominals, e.g. niisee-taa ‘young men’ and harusjaa-taa ‘farmers’ (Loveless 1963: 114; Kokuritsu Kokugo Kenkyuujo 1963: 208).

Table 13.22  Personal pronouns in Shuri (Okinawa) Number Person Politeness

Singular

Plural

1st

 

wan / waa / wanni

wat-taa

2nd

Honorific

unzu

unzu-naa

 

Non-honorific

jˀaa

it-taa

Loveless 1963: 105; Nishioka and Nakahara 2006: 61, 122

basic form (Nishioka and Nakahara 2006: 122; Yushiya 1999: 27) and can be used with many particles, but in some cases other variants are preferred to wan, which is shown by the parentheses in Table 13.23. The form waa is used only when it precedes the nominative case marker, i.e. waa=ga (1sg=nom), or when it modifies another noun, e.g. waa=Ø mun (1sg=nom thing) ‘mine [lit. my thing]’ (Yushiya 1999: 460–1). The variant form wanni is used only when it precedes =nkai (all), =ja (top), and =n (add). When wanni precedes =ja (top), it undergoes a morphophonological change (see Section 13.3.5.2) and becomes wanne=e (1sg=top).  Second-person pronouns do not have any variant form. They can precede all of the case markers, and they can also modify another noun without a genitive case marker, e.g. jˀaa=ga (2.nhon.sg=nom) and jˀaa keena (2.nhon.sg arm) ‘your arm(s)’ (Yushiya 1999: 461; Kokuritsu Kokugo Kenkyuujo 1963: 318). In Shuri, the form that can modify another noun without a genitive case is the same as the form that can head an NP. Thus, I do not recognize personal pronominal “adnominals” in Shuri.

13.4.1.2  Demonstrative pronouns The demonstrative pronouns in Yuwan (Amami) are given in Table 13.24. Yuwan distinguishes three deictic dimensions: proximal (near the speaker), mesial (near the hearer), and distal (apart from both the speaker and the hearer). The three deictic dimensions are expressed by the two sets of roots: ku- (prox), u- (mes), and a- (dist); ka- (prox), ga- (mes), and aga- (dist). Nominals which end with -rɨ (sg) or -taa (pl) can indicate both humans and non-humans, but all the word forms (except for at-t-aa) have a rude implication when they indicate humans. If we want to deictically indicate a person, we can use demonstrative adnominals modifying common nouns, e.g. ku-n cˀju ‘this person’ (prox), u-n cˀju ‘that person’ (mes), and a-n cˀju ‘that person’ (dist), where cˀju means ‘person.’ The demonstrative pronouns in Shuri (Okinawa) are given in Table 13.25. Nominals which end with -ri (sg) or -taa (pl) can indicate both humans and non-humans (Loveless 1963: 105; Nishioka and Nakahara 2006: 14). Shuri distinguishes

207

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.23  Variant forms of first-person singular pronoun in Shuri (Okinawa) waa

wan

wanni

Following particles

waa=ga

N/A

N/A

=ga

(nom)

waa=Ø

(wan=ga)

N/A

=ga/=Ø

(gen)

N/A

wan=Ø

N/A

=Ø

(acc)

N/A

(wan=nkai)

wanni=nkai

=nkai

(all)

N/A

wan=tu

N/A

=tu

(com)

N/A

(wano=o) (wan=ja [waɴ.ja])

wanne=e

=ja

(top)

N/A

(wan=un [wa.nuɴ])

wanni=n

=nb

(add)

a

Loveless 1963: 99, 105, 209; Yushiya 1999: 49, 50, 89, 145, 157, 194, 284–5, 383, 460–1   Yushiya (1999: 157) gave the form wan=nkai [waŋ.ŋ.kai] (1sg=all) ‘to me’ and Loveless (1963: 105, fn. 1) also mentioned the possibility of the form. However, according to the epenthetic rule in Section 13.3.4.2 (Loveless 1963: 147–8), the sequence of /nnk/ is prohibited, and it should become /nunk/. That is, wan=nkai should become wan=unkai [wa.nuŋ.kai]. I have not found data to enable me to decide whether this is an exception to the epenthetic rule or not. b   Yushiya (1999: 361, 397, 474) also gave the form wannun [wan.nuɴ] to mean ‘I also’, but Loveless (1963) and Nishioka and Nakahara (2006) did not give any referential data about the form. a

Table 13.24  Demonstrative pronouns in Yuwan (Amami) Word classes

Meanings

Deixis

 

 

 

 

Proximal

Mesial

Distal

Nominals

Substance (sg)

ku-rɨ

u-rɨ

a-rɨ

 

Substance (pl)

ku-t-taa

u-t-taa

a-t-taa

 

Place

ku-ma

u-ma

a-ma

 

Amount

ka-ssa

ga-ssa

aga-ssa

 

Small amount

ka-hɨdubəi

ga-hɨdubəi

aga-hɨdubəi

Adnominals

Neutral

ku-n

u-n

a-n

 

Derogative

ka-raa

ga-raa

aga-raa

 

Large size

ka-hɨdon

ga-hɨdon

aga-hɨdon

Adverbs

Way

ka-n

ga-n

aga-n

three deictic dimensions, as does Yuwan, but it loses the distinction between mesial and distal in adverbs, i.e. ka-n ‘in this way’ (proximal) and a-n ‘in that way’ (non-proximal).

13.4.1.3  Interrogative pronouns The interrogative pronouns in Yuwan (Amami) are given in Table 13.26. Example (2a) shows that an interrogative pronoun, ta-ru ‘who’ can head an NP. The genitive case marker =ga/=nu

208

cannot follow the pronoun ta-ru ‘who’. If we want to express an interrogative ‘whose’ in Yuwan, we have to use ta-a (see the example (2b)). Example (2c) shows that interrogative pronouns followed by =ka (indef) function as indefinite pronouns. (2)  Interrogative pronouns in Yuwan (Amami) a. a-n    cˀjo=o    ta-ru?   dist-adnz  person=top who-nmlz   ‘Who is that person?’

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.25  Demonstrative pronouns in Shuri (Okinawa) Word classes

Meanings

Deixis

 

 

 

 

Proximal

Mesial

Distal

Nominals

Substance (sg)

ku-ri

u-ri

a-ri

 

Substance (pl)

ku-t-taa

u-t-taa

a-t-taa

 

Place

ku-ma

nˀ-ma

a-ma

Adnominals

Neutral

ku-nu

u-nu

a-nu

 

 

Proximal

Non-proximal

Adverbs

Way

ka-n

a-n

Loveless 1963: 105; Kokuritsu Kokugo Kenkyuujo 1963: 115, 307; Nishioka and Nakahara 2006: 14

Table 13.26  Interrogative pronouns in Yuwan (Amami) Word classes

Forms

Meanings

Word classes

Forms

Meanings

Nominals

nuu

‘what’

Adverbs

ikja-sjɨ

‘how’

 

daa

‘where’

 

ikja-saa

‘how much; how old’

 

ɨcɨɨ

‘when’

 

nuusjattu

‘why’

 

ta-ru

‘who’ (sg)

Adnominals

dɨ-n

‘which’

 

ta-t-taa

‘who’ (pl)

 

ta-a

‘whose’

 

dɨ-ru

‘which one’

 

ikja-sjan

‘what kind of ’

b.  u-rə=ə ta-a mun?   mes-nmlz=top  who-adnz  thing   ‘Whose thing is that one?’ c. ta-ru=ka    umoo-ju-mɨ?   who-sg=indef  exist.hon-umrk-plq   ‘Is there anyone?’ The interrogative pronouns of Shuri (Okinawa) are represented in Table 13.27. In example (3a), the interrogative pronoun taa, ‘who’ heads an NP, and in (3b) it modifies a noun without a genitive case marker. If interrogative pronouns are followed by -gana (indef), the pronouns function as indefinite pronouns (Kokuritsu Kokugo Kenkyuujo 1963: 188). See the example in (3c), where taa ‘who’ followed by -gana (indef) means ‘somebody’. (3)  Interrogative pronouns in Shuri (Okinawa) a. a-re=e taa ja=ga?   dist-nmlz=top  who  cop=ifq   ‘Who is that person?’ (Kokuritsu Kokugo Kenkyuujo 1963: 186)

b. taa   mun=ga?   who  thing=ifq   ‘Whose thing (is this)?’ (Kokuritsu Kokugo Kenkyuujo 1963: 501) c. taa-gana   jub-ee!   who-indef call-imp   ‘Call somebody!’ (Loveless 1963: 196)

13.4.2 Nominals In this chapter, the term “nominals” means words that can head an NP. Nominals include common nouns, address nouns, and numerals. All of them will be introduced in the following subsections (Sections 13.4.2.1–.3). I have already presented the subgroups of nominals in pronouns, i.e. personal pronominal nominals, demonstrative pronominal nominals, and interrogative pronominal nominals (Sections 13.4.1.1.3). The diminutive affix will be discussed in Section 13.4.2.4. The relationship between plural markers and the so-called “animacy hierarchy” will be discussed in Section 13.4.2.5.

209

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.27  Interrogative pronouns in Shuri (Okinawa) Word classes

Forms

Meanings

Word classes

Forms

Meanings

Nominals

nuu

‘what’

Adverbs

ca-a

‘how’

 

maa

‘where’

 

ca-ssa

‘how much; how old’

 

ici

‘when’

 

nuun(di)ci

‘why’a

 

taa

‘who’ (sg)

Adnominals

ca-nu

‘which’

 

tat-taa

‘who’ (pl)

 

 

 

 

ziru

‘which one’

 

 

 

Loveless 1963: 102; Kokuritsu Kokugo Kenkyuujo 1963: 141, 427, 501; Nishioka and Nakahara 2006: 41   The word meaning ‘why’ is nuunci in Loveless (1963: 102) but nuundici in Kokuritsu Kokugo Kenkyuujo (1963: 427).

a

13.4.2.1  Common nouns I will present some examples of common nouns in Yuwan (Amami) and Shuri (Okinawa) in Table 13.28. In Yuwan, common nouns can express both singularity and plurality without any affix. In other words, we cannot decide the number values of common nouns in bare forms without context. This phenomenon is called “general number” by Corbett (2000: 9–19).25 If we want to unambiguously indicate the plurality of common nouns, we can use the enclitic =nkja, e.g. cˀju=nkja ‘people’ and maga=nkja ‘grandchildren’. Common nouns in Yuwan take =nu (gen) to modify another nominal (see Section 13.4.2.6). The plural markers of common nouns in Shuri will be mentioned in Section 13.4.2.5. Common nouns in Shuri take  =nu (gen) to modify another nominal (Nishioka and Nakahara 2006: 95–6).

Table 13.28  Common nouns in Yuwan (Amami) and Shuri (Okinawa) Categories

Meanings

Forms in Yuwan

Forms in Shuri

Human

‘person’

cˀju

cˀju

 

‘child’

warabɨ

warabi

 

‘younger sibling’

ututu

uttu

 

‘grandchild’

maga

nˀmaga

Animal

‘dog’

ɨn

in

 

‘pig’

wˀaa

wˀaa

Plant

‘tree’

kɨɨ

kii

 

‘orange’

nɨkan

mikan

Inanimate

‘rain’

amɨ

ami

13.4.2.2  Address nouns

 

‘stone’

isi

isi

Address nouns are words that can be used to call somebody. For example, akira (personal name), zjuu ‘father’ (elder kinship),26 and soncjoo ‘village mayor’ (official position) are all address nouns in Yuwan. Address nouns can be distinguished from common nouns in Yuwan in that address nouns cannot take =nu (gen) when they modify another nominal (see also Section 13.4.2.6). If we want to have address nouns modify another nominal, we have to juxtapose address nouns with common nouns, e.g. akira tɨɨ (Akira hand) ‘Akira’s hand(s)’, zjuu tɨɨ (father hand) ‘father’s hand(s)’, and soncjoo tɨɨ (vil-

Kokuritsu Kokugo Kenkyuujo 1963

25   It is plausible that common nouns in Shuri also express general numbers, but there is no description of the phenomenon. 26   Most elder kinship terms in Yuwan are address nouns, but there are some words that cannot be used to call referents. For examples, jinga-nəə (man-parent) ‘father’, wunagu-nəə (woman-parent) ‘mother’, and uja ‘parent(s)’ are not address nouns.

210

lage.mayor hand) ‘village mayor’s hand(s)’. In addition, address nouns cannot take =nkja but instead take -taa to indicate plurality, e.g. akira-taa ‘Akira and his companion(s)’ (or ‘people whose names are all Akira’), zjuu-taa ‘father and his companion(s)’ (or ‘fathers’), and soncjoo-taa ‘village mayor and his/her companion(s)’ (or ‘village mayors’). The plural marker -taa can indicate associative plural (X and X’s companion(s)) as well as additive plural (Xs). In Shuri, address nouns can be distinguished from common nouns in that an address noun cannot take =nu (gen) when it modifies another nominal. If we want to have address nouns modify another nominal, we have to juxtapose address nouns with common nouns, e.g. taraa sansin

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages (Tar kind.of.guitar) ‘Tara’s sansin (a kind of guitar)’, where taraa is a personal name (Yushiya 1999: 461). In add­ition, address nouns can take the genitive case marker =ga, e.g. taraa=ga mun (Tara=gen thing) ‘Tara’s thing’ (Yushiya 1999: 460). The plural affix -taa can follow address nouns in Shuri, as well as other nominals, e.g. ziruu-taa (Jiro-pl) ‘the house of Jiro’, where ziruu is a personal name (Kokuritsu Kokugo Kenkyuujo 1963: 501).27

13.4.2.3  Numerals I will present the numerals in Yuwan (Amami) and Shuri (Okinawa) in Table 13.29. A numeral is a word composed of a numeral root and a classifier. For example, the numeral root mii- ‘three’ and the inanimate classifier -cɨ ‘thing’ in Yuwan become the numeral mii-cɨ ‘three things’. In Shuri, the in­ani­ mate classifier is -ci ‘thing’, e.g. mii-ci ‘three things’. There are other classifiers to indicate human, i.e. -i, -cjai, and -tai in Yuwan (e.g. mi-cjai ‘three people’), and -i, -ccai, -ttai, and -tai in Shuri (e.g. mi-ccai ‘three people’). Yuwan does not have native numerals for counting more than four people, and Shuri does not have native numerals for counting more than five people. If we want to count more than four people in Yuwan or more than five people in Shuri, we can use loanwords from Japanese composed of a numeral root and Table 13.29  Numerals in Yuwan (Amami) and Shuri (Okinawa)  

Yuwan (Amami)

Shuri (Okinawa)

 

Inanimate

Human

Inanimate

Human

1

tˀɨɨ

cˀju-i

tii-ci

cu-i

2

tˀaa-cɨ

tˀa-i

taa-ci

ta-i

3

mii-cɨ

mi-cjai

mii-ci

mi-ccai

4

juu-cɨ

ju-tai

juu-ci

yu-ttai

5

ɨcɨ-cɨ

-

ici-ci

icu-tai

6

muu-cɨ

-

muu-ci

-

7

nana-cɨ

-

nana-ci

-

8

jaa-cɨ

-

jaa-ci

-

9

kˀuunu-cɨ

-

kukunu-ci

-

10

tuu

-

tuu

-

Nishioka and Nakahara 2006: 190

27 In fact, the plural form ziruu-taa (Jiro-pl) does not mean any additive (or associative) plurality in this example. Whether or not it can mean ‘Jiro and his companion(s)’ (or ‘people whose names are all Jiro’) should be examined in the future.

the classifier -nin, e.g. zjuu-nin ‘ten people’ in Yuwan and zuu-nin in Shuri (Nishioka and Nakahara 2006: 190). In Yuwan, numerals modify common nouns semantically, but are modified by common nouns syntactically. In (4a) the common noun nasi ‘pear’ followed by the genitive case marker =nu syntactically modifies the numeral mii-cɨ ‘three’ (the curly braces express an NP). Likewise, in (4b) the common noun maga ‘grandchild’ followed by the genitive case marker =nu syntactically modifies the numeral cˀju-i ‘a person’.28 (4) Numerals syntactically modified by common nouns in Yuwan a. wa-n=na {nasi=nu mii-cɨ}NP mura-ta.   1-sg=top  {pear=gen  three-clf.thing}  receive-pst   ‘I received three pears.’ b. wa-n=na {maga=nu    cˀju-i}NP   1-sg=top  {grandchild=gen one-clf.human}   sic-cju-i.   know-prog-npst   ‘I know a grandchild.’

13.4.2.4  Diminutive affix The diminutive affix, which can indicate the smallness of the referent, is -kkwa in Yuwan (Amami) and -gwaa in Shuri (Okinawa). Both of them seem to be related to the common noun kˀwa ‘child’ (Niinaga 2014: 250–5; Kokuritsu Kokugo Kenkyuujo 1963: 196). In Yuwan, the diminutive affix can follow common nouns, e.g. maga-kkwa (grandchild-dim) and usi-kkwa (cow-dim), but it cannot follow address nouns, e.g. *akira-kkwa (Akira-dim), where akira is a personal name. On the other hand, in Shuri, the diminutive affix can ­follow both common nouns and address nouns, e.g. tui-gwaa (bird-dim) and taruu-gwaa (Taro-dim), where taruu is a ­personal name (Kokuritsu Kokugo Kenkyuujo 1963: 196).

13.4.2.5  Plural markers and the animacy hierarchy Yuwan (Amami) has three kinds of plural markers, i.e. -kja, -taa, and =nkja. Personal pronouns use -kja, interrogative pronouns, demonstrative pronouns, and address nouns use -taa, and the other nominals use =nkja, which is summarized in Table 13.30. The differences within the subclass of nominals also exists in the methods of modifying nominals (see Section 13.4.2.6 in detail). The hierarchy given in Table 13.30 is partially similar to the so-called “animacy hierarchy” 28   Both of the examples, i.e. nasi=nu mii-cɨ ‘three pears’ and maga=nu cˀju-i ‘one grandchild’, are objects of sentences. Therefore, they cannot be analyzed as so-called “quantifier float,” since each common noun takes the genitive case marker =nu (not the accusative case marker =ba).

211

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.30  Plural markers and the animacy hierarchy in Yuwan (Amami)  

Personal pronouns

Plural markers -kja

Interrogative pronouns

Demonstrative pronounsa

Address nouns

Other nominals

-taa

-taa

-taa

=nkja

Table 13.31  Plural markers and the animacy hierarchy in Shuri (Okinawa)  

Personal pronouns

Interrogative pronouns

Demonstrative pronouns

Address nouns

Other nominals

 

1st, 2nd.NHON

2nd.HON

 

 

 

 

-naa

-taa

-taa

-taa

-taa, -n(u)caa

Plural markers -taa

  In this table, I give demonstrative pronouns that indicate humans.

a

Table 13.32  Case markers in Yuwan (Amami) Names

Forms

Prototypical functions

Nominative

ga/nu

S, A

Genitive

ga/nu

Modifier in an NP

Accusative

ba

P

Dative 1

n

Beneficiary

Dative 2

nkatɨ

Recipient of information

Allative

kaci

Goal of locomotion

Locative 1

nan/nən

Place of contact

Locative 2

nantɨ/nəntɨ Location

Locative 3

zjɨ

Location distant from the speaker

Instrumental sjɨ

Instrument

Ablative

kara

Source

Comitative

tu

Participant of association

Limitative

gadɨ

Limit

Comparative jukkuma

Standard of comparison

­ roposed in earlier research (Silverstein 1976; Comrie 1989; p Dixon 1994, among others). Shuri (Okinawa) has three plural markers, i.e. -taa, -naa, and -n(u)caa. All nominals can precede -taa except for unzu (the honorific second-person pronoun), and the plural marker -naa is preceded by unzu only (see Section 13.4.1.1). Loveless (1963: 113–14) argues that -nucaa (or -ncaa) is used after a light syllable and -taa is used after a heavy syllable, e.g. dusi-nucaa

212

(friend=pl) ‘friends’, warabi-ncaa (child=pl) ­‘children’, and warabaa-taa (child-pl) ‘children’. The above different plural markers are summarized in Table 13.31.

13.4.2.6  Case markers and the animacy hierarchy The case markers in Yuwan (Amami) are given in Table 13.32. All the case markers in Yuwan are enclitics. I will present examples of all case markers in (5), although examples of genitive case markers will be given later. (5)  Case markers in Yuwan (Amami) a.     b.     c.     d.     e.     f.     g.    

wan=ga  wu-i. 1sg=nom  exist-npst ‘I am (here).’ (Nominative) ɨn=nu wu-i. dog=nom  exist-npst ‘A dog is (here).’ (Nominative) nasi=ba mu-tu-i. pear=acc  pick.up-prog-npst ‘(He) is picking up pears.’ (Accusative) an   cˀju=n   kurɨr-oo. that person.add give-int ‘(I) will give (it) to that person.’ (Dative1) wan=na  arɨ=nkatɨ      jˀi-cja. 1sg=top  that.person=dat2 say-pst ‘I said (it) to that person.’ (Dative2) janməə=kaci  nagɨr-ɨ! garden=all  throw-imp ‘Throw (it) to the garden!’ (Allative) un   sja=nan   cɨbo=nu  a-ta. that  under=loc1  pot=nom  exist-pst ‘There was a pot under there.’ (Locative1)

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages h.         i.     j.     k.     l.    

uzii=ga    daiban+gɨɨ=nantɨ  nasi old.man=nom  big+tree=loc2  pear mu-tu-i. pick.up-prog-npst ‘An old man is picking pears off a big tree.’ (Locative2) usjə=ə  kusabutuu=zjɨ   cɨna-zja. ox=top  thick.grass=loc3 hitch-pst ‘(I) hitched the ox to the thick grass (tied up in a bundle).’ (Locative3) kusa=sjɨ usaw-oo. grass=ins  cover-int ‘Let’s cover (it) by the grass.’ (Instrumental) kuma=kara  mɨzɨ=nkja  cɨnku-da. here=abl   water=pl  load-pst ‘(They) loaded the water from here.’ (Ablative) nan=to=o     asɨb-an-ta=jaa. 2sg.hon=com=top play-neg-pst=sol ‘(I) did not play with you, you know.’ (Comitative)

m.        

urə=ə   mj-an-tɨ=n,     jonban=gadɨ that=top look.at-neg-seq=add  fourth=lmt sic-cju-tat=too. know-prog-pst=ass ‘(They) knew (the song from the first verse) to the fourth, even if (they) did not see it [i.e. a card with the lyrics].’ (Limitative) n. kurə=ə    urɨ=jukkumo=o  hɨɨ-sa-i.   this=top  that=comp=top  big-adj-npst   ‘This one is bigger than that one.’ (Comparative)

On the one hand, the nominative case marker has the two variant forms, =ga and =nu. The difference between the two case markers correlates with the animacy hierarchy (see also Section 13.4.2.5). Pronouns and address nouns use =ga, but the other nominals use =nu (sometimes =ga) as the nominative marker (see 5a–b). On the other hand, the ­genitive case marker has the two variant forms =ga and =nu, too. Not all the nominals use the genitive case markers

Table 13.33  How to modify another nominal in Yuwan (Amami) Modifiers

Examples

 

 

 

Forms

Glosses

Meaning

Personal pro

waa tɨgan

(my letter)

‘my letter’

Interrogative pro

taa tɨgan

(whose letter)

‘whose letter (is this)?’

Demonstrative pro

a-g=ga tɨgan

(dist-nmlz=gen letter)

‘that person’s letter’

Address nouns

akira tɨgan

(Akira letter)

‘Akira (personal name)’s letter’

Other nouns

warabɨ=nu tɨgan

(child=gen letter)

‘a child’s letter’

 

ɨn=nu hagi

(dog=gen foot)

‘dog’s feet’

 

jaa=nu məə

(house=gen front)

‘the space in front of a house’

Table 13.34  Nominative cases compared with modifiers in NPs in Yuwan (Amami)  

Personal pronouns

Interrogative pronouns

Demonstrative adnominals

Address nouns

Other nominals

Nominative case

ga

N/Aa

ga

ga

nu / ga

Modifier in an NP

Adnominals

Adnominals

ga

Juxtaposition

nu

b

  The interrogative pronoun ta-ru ‘who’ takes the focus marker =ga which is homophonous with =ga (nom), in the subject position of a sentence. Thus, it is difficult to decide whether the marker =ga following ta-ru ‘who’ is the nominative marker or the focus marker. b   Demonstrative pronouns in Yuwan have adnominal forms, e.g. a-n (prox-adnz) and u-n (mes-adnz), but these words express the act of pointing only, and they cannot express referents. Let us look at the difference between a-n nɨkan (prox-adnz orange) ‘that orange’ and a-g=ga nɨkan (< a-rɨ=ga nɨkan) (prox-nmlz.sg=gen orange) ‘that person’s orange’. The former demonstrative adnominal a-n ‘that’ does not express any referent by itself, but the latter demonstrative nominal (followed by a genitive case) a-g (< a-rɨ) ‘that person’ expresses a referent. Therefore, the cell at the intersection of “modifier in an NP” and “demonstrative pronouns” is occupied by =ga (gen) since “modifier in an NP” means that both the modifier and the head in an NP have referents. In addition, the cells of “modifier in an NP” indicate cases where the modifier is in the singular form. a

213

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.35  Case markers in Shuri (Okinawa) Names

Forms

Prototypical functions

Nominative

ga/nu

S, A

Genitive

ga/nu

Modifier in an NP

Accusative

Ø

P

Allative

(n)kai

Goal of locomotion

Locative 1

naarii

Location

Locative 2

uti

Location, time

Locative 3

utooti

Location, time

Locative 4

nzi

Location

Instrumental1

si

Instrument

Instrumental2

nakai

Instrument, cause

Instrumental3

saani

Instrument

Ablative

kara

Source

Comitative

tu

Participant of association

Limitative

madi

Limit

Comparative

jaka

Standard of comparison

Yushiya 1999: 195, 254, 458, 465

when they modify another nominal. I will give examples in Table 13.33. Personal and interrogative pronouns are adnominals when they modify other nominals. Demonstrative pronouns take the genitive case marker =ga, whereas address nouns just precede the head nominal without any marker. Other nominals take the genitive case marker =nu (gen). The differences between the nominative case markers, i.e. (5a–b), and how to modify another nominal in an NP, i.e. Table 13.33, will be summarized in Table 13.34. Because of the above distributional differences, the nom­ ina­ tive case markers and the genitive case markers in Yuwan can be distinguished from each other.29 The case markers in Shuri (Okinawa) are given in Table 13.35. All the case markers in Shuri are enclitics. I will present examples of all case markers in (6), although examples of genitive case markers will be given later. 29   The difference between “formal case” and “distributional case” was discussed in Comrie (1991). Distributional cases are those “distinguished formally by at least one noun phrase in the language in question” (Comrie 1991: 45). For example, the definite articles der (nominative) and den (ac­cusa­tive) are formally distinguished when they modify masculine nouns. Considering this fact, the definite articles die (nominative) and die (accusative) can be distinguished when they modify feminine nouns even though they have the same form. Similarly, the case markers =ga (nominative) and =ga (genitive) can be distinguished, since =ga (nominative) can follow both demonstrative

214

(6)  Case markers in Shuri (Okinawa) a.     b.     c.    

waa=ga  abi-ju-n. 1sg=nom bark-aff-ass ‘I will bark.’ (Nominative) (Yushiya 1999: 460)30 in=nu  abi-ju-n. 1sg=nom bark-aff-ass ‘A dog barks.’ (Nominative) (Yushiya 1999: 459) sikuci=Ø  juku-too-ibi-i-n. work=acc lay.aside-prog-pol-aff-ass ‘(He) is laying aside the work.’ (Accusative) (Yushiya 1999: 350) d. jaa=ja     maa=kai   ic-u-ga?   2sg.nhon=top  where=all  go-aff-ifq   ‘Where will you go?’ (Allative) (Yushiya 1999: 463) e. jaa=nu   mee=naarii  kamee-ta-siga,   house=gen  front=loc1 search-pst-advrs   neer-an-ta-n    exist-neg-pst-ass   ‘(I) searched the space in front of the house, but (I) couldn’t find (it).’ (Locative1) (Yushiya 1999: 464) f. wanne=e  hitimiti=uti   c-a-n=doo.   1sg=top   morning=loc2 come-pst-ass=ass   ‘I came (here) in the morning.’ (Locative2) (Yushiya 1999: 465) g. mici=utooti  haaee    s-u-n.   road=loc3  running do-aff-ass   ‘(I) will run on the road.’ (Locative3) (Yushiya 1999: 464) h. tui=nu  sii=nzi   kuga=nu  sidi-ju-n.   bird=gen  nest=loc4  egg=nom hatch-aff-ass   ‘Eggs have hatched in the bird’s nest.’ (Locative4) (Yushiya 1999: 465) i. aka+gaara=si   jaa   huc-ju-n.   red+tile=ins1 house roof-aff-ass   ‘(I) roof the house with red tiles.’ (Instrumental1) (Yushiya 1999: 461) j. warabi=nu  kutu=nakai  sindoo=nu  uhu-sa-n.   child=gen  thing=ins2   worry=nom many-adj-ass   ‘(I) worried about many (things) because of a child.’ (Instrumental2) (Yushiya 1999: 461) k. irana=saani  kusa  ka-ta-n.   sickle=ins3  grass cut-pst-ass   ‘(I) cut the grass by the sickle.’ (Instrumental3) (Yushiya 1999: 462) pronouns and address nouns, but =ga (genitive) cannot ­follow address nouns. The marker =nu does not show that kind of distributional difference. However, it can take another allomorph =n only when it modifies another noun, i.e. in genitive function. Thus, =nu (nominative) can be distinguished from =nu (genitive) even though they have the same form. 30  Yushiya (1999: 460) translates abir- in Shuri into sjaber- ‘speak’ in Standard Japanese. Kokuritsu Kokugo Kenkyuujo (1963: 100) says that abir- means ‘shout; cry or call loudly; bark.’ In this case, I chose the latter translation.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages l. ama=kara  kuma=nkai  cu=nu   there=abl  here=all    person=nom   c-uu-n.   come-aff-ass   ‘A person comes to here from there.’ (Ablative) (Yushiya 1999: 462) m. suu=ja   ari=to=o       niibici   father=top  that.person=com=top marriage   simir-an=i?   do.caus-neg=plq   ‘Does not (your) father let you marry him?’ (Comitative) (Yushiya 1999: 159) n. suu=ja   cinuu=madi   kwaisja   father=top  yesterday=lmt company    juku-too-ta-n.   rest-prog-pst-ass   ‘My husband had been taking off from his work until yesterday.’ (Limitative) (Yushiya 1999: 195) o. are=e   kuri=jaka  taka-sa-tee-gajaa.   that=top  this=comp high-adj-ass.pst-dub   ‘(I) wonder if that one was higher (or more expensive) than this one.’ (Comparative) (Yushiya 1999: 254)

On the one hand, the nominative case marker has the two variant forms =ga and =nu. The difference between the two case markers correlates with the animacy hierarchy (see also Section 13.4.2.5). Pronouns and address nouns use =ga, but the other nominals use =nu (sometimes =ga) as the nom­ ina­tive marker (see 6a–b). On the other hand, the genitive case marker has the two variant forms =ga and =nu too. Not all the nominals use the genitive case markers when they modify another nominal. I will give examples in Table 13.36. Personal and interrogative pronouns are juxtaposed only when they modify other nominals (except for the honorific second-person pronoun unzu, which takes the genitive case marker =ga when it modifies another nominal). Demonstrative pronouns take the genitive case =ga. Address nouns are either juxtaposed with head nominals or followed by the genitive case marker =ga. Other nominals take the genitive case marker =nu when they modify other nominals. The differences between the nominative case markers, i.e. (6a–b), and how to modify another nominal in an NP, i.e. Table 13.36, will be summarized in Table 13.37. Because of the above distributional differences, the nom­ ina­tive case markers and the genitive case markers in Shuri can be distinguished from each other.

Table 13.36  How to modify another nominal in Shuri (Okinawa) Modifiers

Examples

 

 

 

Forms

Glosses

Meaning

Personal pro

waa mun

(1sg thing)

‘mine [lit. my thing]’

 

unzu=ga nasaki

(2.hon.sg=gen mercy)

‘your mercy’

Interrogative pro

taa mun

(who thing)

‘Whose thing (is this)?’

Demonstrative pro

a-ri=ga mun

(dist-nmlz=gen thing)

‘that person’s thing’

Address nouns

taraa sansin

(Tara kind.of.guitar)

‘Tara’s sansin (a kind of guitar)’

 

taraa=ga mun

(Tara=gen thing)

‘Tara’s thing’

Other nouns

tui=nu kusu

(bird=gen droppings)

‘bird droppings’

 

husi=nu naa

(star=gen name)

‘the name of a star’

Kokuritsu Kokugo Kenkyuujo 1963: 186, 501; Yushiya 1999: 460–1; Nishioka and Nakahara 2006: 10, 95–6

Table 13.37  Nominative cases compared with modifiers in NPs in Shuri (Okinawa)  

Personal pronouns

Interrogative pronouns

Demonstrative pronouns

Address nouns

Other nominals

 

1st, 2nd.NHON

2nd.HON

 

 

 

 

Nominative case

ga

(ga ?)

(ga ?)

ga

ga

nu / ga

Modifier in an NP

Juxtaposition

ga

Juxtaposition

ga

Juxtaposition, ga nu

215

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga

13.4.3 Adnominals

Table 13.38  tur- ‘take’ followed by verbal derivational affixes in Yuwan (Amami)

Adnominals are a word class that modifies a noun without any genitive case marker, and cannot head an NP by itself. Adnominals in Yuwan (Amami) were given in Table 13.21 (personal pronominal adnominals), Table 13.24 (demonstrative pronominal adnominals), and Table 13.26 (interrogative pronominal adnominals). Adnominals in Shuri (Okinawa) were given in Table 13.25 (demonstrative pronominal adnominals) and Table 13.27 (interrogative pronominal adnominals).

a. tur-as-ju-i (take-caus-umrk-npst) ‘(I) make (someone) take (something)’ b. tur-arɨ-i (take-pass-npst) ‘(I) suffer from having (something) taken (by someone)’ c. tu-tuk-ju-i (take-prpr-umrk-npst) ‘(I) take (something) for preparation’ d. tur-arɨ-i (take-cap-npst) ‘(I) can take (something)’ e. tu-tu-i (take-prog-npst) ‘(I) am taking (something)’ f.

13.4.4 Verbs

tur-an (take-neg) ‘(I) do not take (something)’

g. tu-ju-i (take-umrk-npst) ‘(I) take (something)’ h. tu-jawu-i (take-pol-npst) ‘(I) take (something)’ [polite]

13.4.4.1  Structure of verbs and derivational affixes I will present the structure of verbs in Yuwan (Amami) in Figure 13.3. Braces mean that -arɨr (cap), -tur (prog), -jawur (pol), and -an (neg) cannot co-occur with -jur (umrk). All the affixes standing between “Root” and “Inflectional affix” in Figure  13.3 are derivational affixes. These derivational affixes do not change word classes. They are used to indicate voice, aspect, mood, polarity, tense, and politeness.31 These derivational affixes cannot appear in verb-final pos­ ition, and they are necessarily followed by an inflectional affix (see Section 13.4.4.2). The two affixes -an (neg) and -tar (pst), however, can appear in verb-final position, and can also be followed by another inflectional affix. The following combinations are impossible: *-arɨr-arɨr (pass-cap), *-arɨr-jur (pass-umrk), *-tuk-tur (prpr-prog), *-tuk-tar (prpr-pst), *-jawur-təər (pol-res). It should be mentioned that -təər (res) can precede another -təər, e.g. mˀarɨ-tə-təəp-pa (be.born-res-res-csl) ‘because (they) had already born’. In Niinaga (2014: 362), I analyzed the second affix (as well as the first affix) as “the resultative affix.” However, the latter -təər may be analyzed as the assertive past affix just like -teer (ass.pst) in Shuri (Okinawa). In Table  13.38 I will present examples of the verbal root  tur- ‘take’ followed by the above derivational affixes with the word-final inflectional affix -i (npst) (although -an (pst) and -tar (pst) do not need the inflectional affix). The gloss umrk (unmarked) of the derivational affix -jur means Root

-as

-arɨr

-tuk

-arɨr

-tur

CAUS

PASS

PRPR

CAP

PROG

i.

tu-təə-i (take-res-npst) ‘(I) have taken (something)’

j.

tu-ta (take-pst) ‘(I) took (something)’

The derivational affixes -as (caus) and -tuk (prpr) cannot precede the inflectional affix -i (npst) directly.

that it always expresses active voice, affirmative polarity, non-progressive aspect, and is unmarked for politeness. There is no grammatical agreement between arguments and verbal affixes in Yuwan, so ‘(I)’ (first-person) in free translation in Table 13.38 is a tentative translation. Derivational affixes are underlined in Table 13.38. The verbal morphology in Yuwan is rich in mor­pho­phono­ logic­al alternations. In Table  13.39, all the verbal affixes (derivational and inflectional affixes) are classified into five groups. The main criterion to classify the five is affix-initial phonemes. Strictly speaking, however, there are some exceptions, and it is the set of morphophonological rules (applying to the verbal affixes) that distinguishes the five affixal groups. The same affix type induces the same morphophonological alternations. For example, all the affixes in type B induce voicing of affix-initial stops and deletion of the root-final consonants when they follow /m/, e.g. kam- ‘eat’ + -ta (pst) > ka-da ‘ate,’ kam- ‘eat’ + -tur (prog) + -i (npst) > ka-du-i ‘is eating’ (about the deletion of /r/, see Section 13.3.4.1.

-jawur

-an

-tǝǝr

-tar

POL

NEG

RES

PST

-jur UMRK

Figure 13.3  Structure of the verbs in Yuwan (Amami) 31

  Backhouse (2004: 52) also called these affixes in Japanese grammatical derivations.

216

-Inflectional affix

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.39  Affix types classified by morphophonological characteristics in Yuwan (Amami) Types

Affix-initial phonemes

A.

Vowels

-an (neg), -arɨr (pass), -arɨr (cap), -as (caus), -azɨi (neg.plq), -ɨ (imp), -ɨba (sugs), -oo (int), -oo (infr)

B.

Alveolar stopsa

-tar (pst), -tuk (prpr), -tur (prog), -təər (res), -tɨ (seq), -tai (lst), -təəra (pos)

C.

Nasals, palatal approximants, and alveolar fricative

-jawur (pol), -jaa (nmlz), -jur (umrk), -jagacinaa (sim), -mɨ (plq), -n (ptcp), -sɨ (nmlz)

D.

N/A

-i/-Ø (inf)

E.

Obstruents

See Niinaga (2014) and Niinaga (2015b) for more details of the morphophonological alternations. I will give partial Yuwan verbal paradigms in Table 13.40, and you can induce some morphophonological rules from the data. Verbal roots can be classified into 17 types of regular verbs and 8 types of irregular verbs. Regular verbal roots are classified by their root-final phonemes. The differences in root-final phon­ emes correlate with the differences of mor­pho­phono­logic­al rules which they undergo. In Table 13.40, “V(non-back)r” means that the verbal-root-final /r/ is preceded by a nonback vowel, “V(back)r” means that the verbal-root-final /r/ is preceded by a back vowel, “V(non-i)” means vowels other than /i/, “C” means any consonant, “G” means a glide (i.e. an approximant), “V” means any vowel, and “$” marks word-initial position. The structure of verbs in Shuri (Okinawa) is given in Figure 13.4 (Kokuritsu Kokugo Kenkyuujo 1963: 72–3, 75–8, 81). Affixes distinguished by a slash cannot co-occur. For example, the assertive past affix -teer cannot co-occur with the past affix -tar (Kokuritsu Kokugo Kenkyuujo 1963: 75).32 All the affixes standing between “Root” and “Inflectional affix” in Figure 13.4 are derivational affixes. The causative affix –as becomes its allomorph –imi when it follows verbal roots that have /s/ in their final position

Verbal affixes

-ba (csl), -boo (cond), -gadɨ (lmt), -sa (pol), -sɨga (pol), -too (csl), -tu (csl), -na (proh)

b

32   The assertive past affix -teer has the same form as the resultative affix -teer, and the latter can co-occur with the past affix -tar, e.g. judee-ta-n (read-res-pst-ass) ‘(I) have already read (it)’ (Kokuritsu Kokugo Kenkyuujo 1963: 74–5).

This is a necessary condition since there are affixes of type E that begin with alveolar stops, i.e. -too (csl) and -tu (csl). b There is one exception; -na (PROH) begins with a nasal. a

Table 13.40  Verbal paradigms in Yuwan (Amami)  

 

 

Affix types

 

 

 

A.

B.

C.

Regular verbs

Examples

-an (NEG)

-tar (PST)

-jur-i (UMRK-NPST) -i/-Ø (INF) -na (PROH)

1

V(non-back)r

abɨr- ‘call’

abɨr-an

abɨ-ta

abɨ-ju-i

abɨ-Ø

abɨn-na

2

V(back)r

tur- ‘take’

tur-an

tu-ta

tu-ju-i

tu-i

tun-na

3

pp

app- ‘play’

app-an

at-ta

app-ju-i

app-i

app-una

4

b

narab- ‘line up’

narab-an

nara-da

narab-ju-i

narab-i

narab-una

5

Vm

jum- ‘read’

jum-an

ju-da

jum-ju-i

jum-Ø/-i

jum-na

6

nm

tanm- ‘ask’

tanm-an

tan-da

tanm-ju-i

tanm-i

tanm-una

7

V(non-i)k

kak- ‘write’

kak-an

ka-cja

kak-ju-i

kak-i

kak-una

8

V(non-i)kk

sukk- ‘draw’

sukk-kan

suc-cja

sukk-ju-i

sukk-i

sukk-una

9

Vs

us- ‘push’

us-an

u-cja

us-ju-i

us-i

us-ɨna

10

ss

kuss- ‘kill’

kuss-an

kuc-cja

kuss-ju-i

kuss-i

kuss-ɨna

11

t

ut- ‘hit’

ut-an

uc-cja

uc-ju-i

uc-i

uc-ɨna

12

$C(G)

mj- ‘look at’

mj-an

mji-cja

m-ju-i

m-ii

mj-uuna

D.

E.

(continued)

217

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.40  Continued  

 

 

Affix types

 

 

 

A.

B.

C.

Regular verbs

Examples

-an (NEG)

-tar (PST)

-jur-i (UMRK-NPST) -i/-Ø (INF) -na (PROH)

13

Cij

kij- ‘cut’

kij-an

ki-cja

ki-ju-i

ki-i

kin-na

14

V(non-i)g

tug- ‘sharpen’

tug-an

tu-zja

tug-ju-i

tug-i

tug-una

15

ik

kik- ‘listen to’

kik-jan

ki-cja

kik-ju-i

kik-i

kik-una

16

i(n)g

uig- ‘swim’

uig-jan

ui-zja

uig-ju-i

uig-i

uig-una

17

in

sin- ‘die’

sin-jan

si-zja

sin-ju-i

sin-Ø/-i

sin-na

Irregular verbs

Examples

-an (NEG)

-ta (PST)

-jur-i (UMRK-NPST) -i/-Ø (INF) -na (PROH)

18

 

k- ‘come’

k-on

cˀja

k-ju-i

k-ii

kˀ-uuna

19

 

hijaw- ‘pick up’

hijaw-an

hija-ta

hija-ju-i

hijə-ə

hijəə-na

20

 

sɨr- ‘do’

sɨr-an

sja

s-ju-i

s-ii

sɨn-na

21

 

hənk- ‘enter’

hənk-jan

hən-cja

hənk-ju-i

hənk-i

hənk-una

22

 

ik- ‘go’

ik-jan

i-zja

ik-ju-i

ik-i

ik-una

23

 

umoor- (exist.hon)

umoor-an

umoo-cja

umoo-ju-i

umoo-i

umoon-na

24

 

sij- ‘know’

sij-an

sic-cja

?

si-i

?

25

 

jurukub- ‘have pleasure’

jurukub-an

juruku-da

jurukub-ju-i

jurukub

jurukun-na

D.

E.

The past affix -tar becomes -ta when it stands in word-final position. If it is followed by another affix, e.g. -oo (infr), the /r/ is retained, e.g. abɨ-tar-oo (call-pst-infr) ‘(Someone) seems to have called (someone)’.

a

Root

-as

-arir

-took

-toor

-teer

CAUS

PASS

PRPR PROG

RES

-jabir

-jur/an

-teer/tar POL AFF/NEG ASS.PST /PST

-Inflectional affix

Figure 13.4  Structure of the verbs in Shuri (Okinawa)

(except for the irregular verbal root s-‘do’) (Kokuritsu Kokugo Kenkyuujo  1963: 81). The polite affix –jabir is relatively old pronunciation, and it is frequently replaced by –ibir in normal pronunciation (Kokuritsu Kokugo Kenkyuujo 1963: 77). The affirmative affix –jur becomes its allomorph –ir when it follows the politeness affix -jabir, e.g. tu-jabi-i-n (take-pol-aff-ass) ‘(I) would take (it)’ (Kokuritsu Kokugo Kenkyuujo 1963: 100) or the passive/capable affix -arir, e.g. karamir-ari-i-n (catchpass-aff-ass) ‘(I) am caught’ and jum-ari-i-n (read-cap-aff-ass) ‘(I) can read’ (Yushiya 1999: 167, 175). Table 13.41 presents examples of the verbal root jum- ‘read’ followed by the derivational affixes above with the word-final inflectional affix -n (ass) (although -an (pst) and -tar (pst) do not need the inflectional affix). There is no grammatical agreement between arguments and verbal affixes in Shuri, so ‘(I)’ (first-person) in free translation in Table 13.41 is a tentative translation. Derivational affixes are underlined in Table 13.41.

218

Shuri is rich in the verbal morphophonological alternations (Kinjo and Hattori 1955: 332–46; Kokuritsu Kokugo Kenkyuujo 1963: 58–81). In Table 13.42, verbal affixes (der­iv­ ation­al and inflectional affixes) described in Kokuritsu Kokugo Kenkyuujo (1963: 58–81, 297, 399) are classified into five groups. The main criterion to classify the five is affixinitial phonemes. Strictly speaking, however, it is the mor­ pho­phono­logic­al rules (applying to the verbal affixes) that distinguish the five affixal groups. The same affix type induces the same morphophonological alternations. For example, all the affixes in type B induce voicing of affixinitial stops and deletion of the root-final consonants when they follow /m/, e.g. kam- ‘eat’ + -ta (pst) + -n (ass) > ka-da-n ‘ate’, kam- ‘eat’ + -toor (prog) + -n (ass) > ka-doo-n ‘is eating’. Now I cannot go into the details of the mor­pho­phono­ logic­al alternations. See Kinjo and Hattori (1955: 332–46) and Kokuritsu Kokugo Kenkyuujo (1963: 58–81) for more details.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.41  jum- ‘read’ followed by verbal derivational affixes in Shuri (Okinawa) a. jum-as-ju-n (read-caus-aff-ass) ‘(I) make (someone) read (something)’ b. jum-ari-ju-n (read-pass-aff-ass) ‘(I) suffer from having (something) read (by someone)’ c. ju-dooc-u-n (read-prpr-aff-ass) ‘(I) read (something) for preparation’ d. ju-doo-n (read-prog-ass) ‘(I) am taking (something)’ e. jum-an (read-neg) ‘(I) do not read (something)’ f.

jun-u-n (read-aff-ass) ‘(I) read (something)’

g. ju-dəə-n (read-res-ass) ‘(I) have read (something)’ h. jun-abi-i-n (read-pol-aff-ass) ‘(I) read (something)’ (Polite) i.

ju-dəə-n (read-ass.pst-ass) ‘(I) read (something)’ (Past tense)

j.

ju-da-n (take-pst-ass) ‘(I) read (something)’ (Past tense)

(Kokuritsu Kokugo Kenkyuujo 1963: 74–7, 81) The derivational affixes -as (caus), -arir (pass), -arir (cap), -took (prpr), and -jabir (POL) cannot precede the inflectional affix -n (ass) ­directly.

Table 13.42  Affix types classified by morphophonological characteristics in Shuri (Okinawa) Types

Affix-initial phonemes

A.

Vowels

-an (neg), -arir (pass), -arir (cap), -as (caus), -a (int), -a (dub), -aii (inv), -ana (opt), -aa (cond1), -ee (cond2), -iwadu (cond.foc), -i (imp1), -ee (imp2), -uka (crt)

B.

Alveolar stops

-tar (pst), -took (prpr), -toor (prog), -teer (res), -teer (ass.pst), -ti (seq), -tai (lst), -taraa (cond3), -taree (cond4)

C.

Palatal approximants, nasals, velar stops, and alveolar fricatives

-jabir (pol), -jagijur (inch), -jur (aff), -jagacii (sim), -mi (plq), -n (ass), -kutu (csl), -ga (ifq), -gajaa (dub), -sa (pol), -si (nmlz), -siga (advrs)

D.

N/A

-i (inf)

E.

Alveolar nasal

-na (proh)

Verbal affixes

I will give partial Shuri verbal paradigms in Table 13.43, and you can induce some morphophonological rules from the data.33 Verbal roots can be classified into 13 types of regular verbs and 7 types of irregular verbs. Regular verbal roots are classified by their root-final phonemes. The differences in root-final phonemes correlate with the differences in the morphophonological rules which they undergo. An example of the passive affix -arir is included in the irregular types since it undergoes a morphophonological alternation different from the regular verbs that have root-final /r/. In Table 13.43, “V(non-i)r” means that the verbal-root-final /r/ is preceded by a vowel other than /i/. Verbal roots that have /ir/ in their final positions can be divided into three types, i.e. “ir1, ir2, and ir3.” Differences among them appear when they take the affixes in type B. If a verbal root of “ir1” is followed by an affix in type B, the initial /t/ of the affix becomes /c/. For example, if cir- ‘put on’ is followed by -tar (pst), it becomes ci-ca-n (put.on-pst-ass) ‘(I) put on (it),’ where -tar became -ca and the final /r/s of the root and affix are omitted. However, if a verbal root of “ir2” is followed by -tar (pst), the final /r/ of the root is not omitted, but it assimilates to the following consonant, e.g. cir- ‘cut’ + -ta-n (pst-ass) > cic-ca-n. If a verbal root of “ir3” is followed by -tar (pst), the affix does not undergo any change and the root-final affix is omitted, e.g. ukir- ‘receive’ + -ta-n (pst-ass) > uki-ta-n. The difference between “d1” and “d2” is like that between ir1 and ir3. The former undergoes the affrication of affix-initial consonant (from /t/ to /c/), but the latter does not.

13.4.4.2  Inflectional affixes of verbs In the previous section, verbal affixes in the Northern Ryukyuan languages, i.e. Yuwan (Amami) and Shuri (Okinawa), were classified in terms of their mor­pho­phono­ logic­al characteristics. Furthermore, verbal inflectional affixes can be classified in terms of their syntactic functions. They are classified into four types, i.e. finite affixes (or mainclause affixes), participial affixes (or clause-adnom­inalizer affixes), converbal affixes (or clause-adverbializer affixes), and masdar affixes (or clause-nominalizer affixes). All the inflectional affixes and their syntactic functions in Yuwan are given in Table 13.44. Their differences and sentential examples will be given later. Affixes marked with an ­asterisk (*) cannot directly follow a verbal root (except for copula, existential, and stative verbal roots). In other words, they need to be preceded by a derivational affix shown in Figure 13.3 (other than -as (caus) and -tuk (prpr)). For example, -n (ptcp) cannot follow the verbal root tur- ‘take’ (*tur-n or *tu-n is ungrammatical), and it needs a derivational affix such 33   I have not given examples of wur- ‘exist’ from Kinjo and Hattori (1955: 334), since it is the same as nur- ‘ride’ except for not being able to take -jur (aff).

219

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Table 13.43  Verbal paradigms in Shuri (Okinawa)  

 

 

Affix types

 

 

 

A.

B.

C.

D.

E.

Regular verbs

Examples

-an (NEG)

-ta-n (PST-IND)

-ju-n (AFF-IND)

-i (INF)

-na (PROH)

1

V(non-i)r

nur- ‘ride’

nur-an

nu-ta-n

nu-ju-n

nu-i

nun-na

2

ir1

cir- ‘put on’

cir-an

ci-ca-n

ci-ju-n

ci-i

cin-na

3

ir2

cir- ‘cut’

cir-an

cic-ca-n

ci-ju-n

ci-i

cin-na

4

ir3

ukir- ‘receive’

ukir-an

uki-ta-n

uki-ju-n

uki-i

ukin-na

5

b

tub- ‘fly’

tub-an

tu-da-n

tub-u-n

tub-i

tub-una

6

m

jum- ‘read’

jum-an

ju-da-n

jun-u-n

jum-i

jum-una

7

d1

kund- ‘tie’

kund-an

kun-ca-n

kunz-u-n

kunz-i

?

8

d2

nind- ‘sleep’

nind-an

nin-ta-n

ninz-u-n

ninz-i

nind-una

9

s

kees- ‘return’

kees-an

kee-ca-n

kees-ju-n

kees-i

kees-una

10

t

tat- ‘stand’

tat-an

tac-ca-n

tac-u-n

tac-i

mat-una

11

k

kak- ‘write’

kak-an

ka-ca-n

kac-u-n

kac-i

kak-una

12

g

kuug- ‘row’

kuug-an

kuu-za-n

kuuz-u-n

kuuz-i

kuug-una

13

in

sin- ‘die’

sin-an

si-za-n

sin-u-n

sin-i

sin-una

Irregular verbs

Examples

-an (NEG)

-ta-n (PST-IND)

-ju-n (AFF-IND)

-i (INF)

-na (PROH)b

14

 

k- ‘come’

k-uun

c-a-n

c-uu-n

c-ii

k-uuna/k-uunna

15

 

nnd- ‘look at’

nnd-an

nn-ca-n

nnz-u-n / nu-u-n

nnz-i / mi-i

nnd-una

16

 

s- ‘do’

s-an

sj-a-n

s-ju-n

s-ii

s-una

17

 

jˀ- ‘say’

jˀ-an

i-ca-n

jˀ-u-n

i-i

jˀ-una

18

 

ik- ‘go’

ik-an

nˀ-za-n

ic-u-n

ic-i

?

19

 

kir- ‘kick’

kir-an

kic-ca-n / ki-ta-n

ki-ju-n

ki-i

?

20

 

-arir (pass)

-ar-an

-at-ta-n

-ari-ju-n

-ar-ii

?

c

 

a

Kinjo and Hattori 1955: 332–4, 338; Loveless 1963: 83–4   Kinjo and Hattori (1955: 332–7) and Kokuritsu Kokugo Kenkyuujo (1963: 58–63) have a variant form of jun- ‘read’, but Yushiya (1999: 124) and Nishioka and Nakahara (2006: 32) do not have the variant form, and they use jum-, i.e. jum-u-n (read-umrk-ind). It is probable that the verbal paradigm of ‘read’ in Shuri has become simpler as a result of diachronic change. b   Some verbal forms containing the negative affix -na are not clear because of the insufficiency of the relevant data. c   The pronunciation of s- ‘do’ followed by type-B and type-C affixes is not palatalized in the modern Shuri, e.g. s-a-n [saɴ] (do-pst-ass) ‘(I) did’ and s-u-n [suɴ] (do-aff-ass) ‘(I) do’ (Yushiya 1999: 13). a

as -jur (umrk), i.e. tu-ju-n (take-umrk-ptcp). Root-final and affix-final /r/s are deleted by a mor­pho­phono­logic­al rule (cf. Table 13.40). Verbs followed by finite affixes can occupy the predicate of a main clause. In (7a), the verb followed by the finite affix -i (npst), i.e. umoo-ju-i ‘exist’ (honorific), occupies the predicate of the main clause. Verbs followed by participial affixes can occupy the predicate of adnominal clauses. In (7b), the verb followed by the participial affix -n (ptcp), i.e. suc-cju-n

220

‘who was pulling’, occupies the predicate of the adnominal clause, whose head NP is cˀju ‘person’. A verb followed by a converbal affix can occupy the predicate of an adverbial clause. In (7c), the verb followed by the converbal affix -pa (caus), i.e. ap-pa ‘there is (something), so. . .’, occupies the predicate of the adverbial clause.34 A verb followed by a 34 The converbal affix -ba (csl) undergoes assimilation with preceding root-final consonant (ar- > ab-) and devoicing of voiced geminate (ab-ba > ap-pa), and finally becomes -pa.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Table 13.44  Inflectional affixes classified by syntactic characteristics in Yuwan (Amami) Types

Functions of verbs Verbal affixes

Finite affixes

Predicate of a main clause

-an (neg),a *-i (npst), -tar (pst), oo (int), *-oo (infr), *-sa (pol), *-sɨga (pol), *-mɨ (plq), -azɨi (neg.plq), -ɨ (imp), -na (proh), -ɨba (sugs), *-u (pfc)

Participial Predicate of an affixes adnominal clause

*-n (ptcp)

Converbal Predicate of an affixes adverbial clause

-ba (csl), *-tu (csl), *-too (csl), -boo (cond), -tai (lst), -gadɨ (lmt), -jagacinaa (sim), -təəra ‘after,’ -tɨ (seq), *-nən (seq)

Masdar affixes

Predicate of a nominal clause

b.  Participial affix -n (ptcp)   hinzjaa=ba   suc-cju-n     cˀju=nu    cˀja.   goat=acc  pull-prog-ptcp  person=nom come.pst   ‘A person who was pulling a goat came.’ c.  Converbal affix -pa (csl)   nɨɨzin=nu    ap-pa,    ar-an=daroo.   carrot=nom  exist-csl  cop-neg=infr   ‘There are (pieces of) a carrot, so maybe (the pickles) are not (mine).’

-i /-Ø (inf), *-sɨ (nmlz)

a   Verbs whose ending is the negative affix -an can fill the predicate of an adnominal (i.e. relative) clause without any relativizer, e.g. urɨ=ba tur-an cˀju (that=acc take-neg person) ‘a person who does not take (it)’.

masdar affix can occupy the predicate of a nominal clause. In (7d), the verb followed by the masdar affix -Ø (zero morpheme), i.e. cɨkɨ-Ø ‘attaching’, occupies the predicate of the nominal clause, which is followed by the copula verb ja-tattu ‘because (it) was . . ..’ In (5a–d), all the verbs preserve their arguments, i.e. the locative NP a-ma=nan (dist-place=loc1) in (7a), the accusative NP hinzjaa=ba (goat=acc) in (7b), the nominative NP nɨɨzin=nu (carrot=nom) in (7c), and the locative NP u-ma=nantɨ (mes-place=loc2) in (7d). In other words, their “internal syntax” (Haspelmath 1996: 52–3) is preserved consistently. Their “external syntax” (Haspelmath 1996: 52–3), however, is different in each case since the verbs occupy different syntactic types of clauses (i.e. main clauses, adnominal clauses, adverbial clauses, and nominal clauses) depending on the verbal affixes. Thus, the verbal inflectional affixes in Yuwan are instances of “word-class-changing inflection” (Haspelmath 1996: 52–3). Regarding the details of verbal inflectional affixes in Yuwan, see Niinaga (2014) or Niinaga (2015b). (7) Finite, participial, converbal, and masdar affixes in Yuwan (Amami) a.  Finite affix -i (npst)   a-ma=nan umoo-ju-i.   dist-place=loc1 exist.hon-umrk-npst   ‘(He) is there.’

d.  Masdar affix -Ø (inf)   u-ma=nantɨ    cɨkɨ-Ø    ja-tat-tu.   mes-place=loc2 attach-inf  cop-pst-csl   ‘(The ship) came alongside there [lit. (The ship) was to dock there].’

Table 13.45  Inflectional affixes classified by syntactic characteristics in Shuri (Okinawa) Types

Functions of verbs Verbal affixes

Finite affixes

Predicate of a main clause

-an (neg),a -a (int), *-a (dub), -aii (inv), -ana (opt), -i (imp1), -ee (imp2), -tar (pst), *-mi (plq), *-n (ass), *-ga (ifq), -gajaa (dub),b *-sa (pol), -na (proh)

Participial Predicate of an affixes adnominal clause

*-u (ptcp)

Converbal Predicate of an affixes adverbial clause

-aa (cond1), -ee (cond2), -iwadu (cond.foc), -uka (crt), -ti (seq), -tai (lst), -taraa (cond3), -taree (cond4), -jagacii (sim), *-kutu (csl), *-siga (advrs)

Masdar affix

*-si (nmlz)

Predicate of a nominal clause

Verbs whose ending is the negative affix -an can fill the predicate of an adnominal (i.e. relative) clause without any relativizer, e.g. wakaran tukuru (understand-neg part) ‘a part (which I) cannot understand’ (Yushiya 1999: 83). b Kokuritsu Kokugo Kenkyuujo (1963: 186, 270) analyzes this form into an affix followed by a particle, i.e. -ga=jaa (ifq=sol), but Yushiya (1999: 251–4) analyzes this form into a single particle =gajaa (inf). The former said that the information-question affix -ga does not need interrogative words if it co-occurs with =jaa (sol). This form /gajaa/ follows bound verbal stems, so I analyze it as a single affix which indicates dubitation, i.e. -gajaa (dub). a

221

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga Infinitives (i.e. verbal word forms containing the infinitival affix -i/-Ø) in Yuwan can be used to make a compound (see Section 13.4.6). Inflectional affixes and their syntactic functions in Shuri (Okinawa) are given in Table 13.45. Affixes marked with an asterisk (*) cannot directly follow a verbal root (except for copula, existential, and stative verbal roots). In other words, they need to be preceded by a derivational affix from those in Figure 13.4 (other than -as (caus), -arir (pass), -arir (cap), -took (prpr), and -jabir (pol)). In (8a), the verb followed by the finite affix -n (ass), i.e. hu-jun ‘(It) falls’, occupies the predicate of the main clause. In (8b), the verb followed by the participial affix -u (ptcp), i.e. jˀ-ur-u ‘which (I) say’, occupies the predicate of the adnominal (i.e. relative) clause. In (8c), the verb followed by the converbal affix -kutu (csl), i.e. juku-ta-kutu ‘since (I) took a rest’, occupies the predicate of the adverbial clause. In (8d), the verb followed by the masdar affix -si (nmlz), i.e. tac-coo-si ‘(someone) standing’, occupies the predicate of the nominal clause. (8) Finite, participial, converbal, and masdar affixes in Shuri (Okinawa) a.  Finite affix -n (ass)   ami=nu    hu-ju-n.   rain=nom  fall-aff-ass   ‘(It) rains.’ (Kokuritsu Kokugo Kenkyuujo 1963: 113) b.  Participial affix -u (ptcp)   jˀ-ur-u    kutu   cik-an=i!   say-aff-ptcp  thing listen.to-neg=plq   ‘Listen to what I say!’ (Kokuritsu Kokugo Kenkyuujo 1963: 274) c.  Converbal affix -kutu (csl)   ihwi   juku-ta-kutu,     noo-ta-n.   a.little take.a.rest-pst-csl  get.well-pst-ass   ‘(I) got well, since (I) took a rest.’ (Kokuritsu Kokugo Kenkyuujo 1963: 136) d.  Masdar affix -si (nmlz)   cˀju=nu    tac-coo-si     umujoo=ja   person=nom stand-prog-nmlz  subtly=top waka-ju-n. feel-aff-ass   ‘(I) subtly feel that someone is standing (there).’ (Kokuritsu Kokugo Kenkyuujo 1963: 554–5) For more details about inflectional affixes in Shuri, see Kinjo and Hattori (1955: 332–46) and Kokuritsu Kokugo Kenkyuujo (1963: 58–81).

13.4.5 Adjectives Adjectival phrases in Yuwan (Amami) are composed of an adjective (or an adjective plus a stative verb). Adjectives in Yuwan are morphologically composed of an adjectival root

222

followed by the adjectival affixes -sa/-soo. In (9a), the adjectival root taa- ‘tall’ is followed by the adjectival affixes -sa/soo. Adjectives in Yuwan cannot express tense or other grammatical information by themselves. To express that information, the stative verbs ar- and nə- are used. The stative verbs can take most of the inflectional affixes in Table 13.44, although semantically impossible combination is prohibited. For example, the imperative affix cannot follow the stative verb *taa-sa ar-ɨ (high-adj stv-imp). Basically, the adjective containing -sa is followed by the stative verb ar- in affirmative polarity, and the adjective containing -soo is followed by the stative verb nə- in negative polarity. In (9b), the adjective taa-sa (tall-adj) is followed by the stative verb a-, which takes the past inflection -tar and becomes a-ta in the surface form (see Table 13.40 for mor­pho­phono­logic­al alternations). In (9c), the adjective taa-soo (tall-adj) is followed by the stative verb nə-, which takes the negative inflection -an and becomes nən in the surface form; the ­negative affix -an omits its initial vowel /a/ after nə-.35 In (9d), the adjective taa-sa (tall-adj) is followed by the stative verb ar-i (stv-npst). They are, however, contracted into a single word form, i.e. taa-sa-i (tall-adj. stv-npst). This contraction also occurs when stative verbs take the participial affix (i.e. clause-adnominal affix) -n. In (9e), the adjective taa-sa (tall-adj) is followed by the stative verb a-n (stv-ptcp), but they are contracted into taa-sa-n (talladj.stv-ptcp). Contraction of adjectives and the stative verb ar- does not occur in the cases of other verbal affixes (see 9b). In (9a–e), the underlined parts are adjectival phrases. (9)  Adjectival phrases in Yuwan (Amami) a. an kɨɨ=ja    taa-sa/taa-soo.   that  tree=top tall-adj/tall-adj   ‘That tree is tall.’ [Positive; non-past] b. an kɨɨ=ja taa-sa a-ta.   that tree=top  tall-adj  stv-pst   ‘That tree was tall.’ [Positive; past] c. an kɨɨ=ja taa-soo nə-n.   that tree=top  tall-adj stv-neg   ‘That tree is not tall.’ [Negative; non-past] d.      

an kɨɨ=ja taa-sa-i. an kɨɨ=ja taa-sa+ar-i [Underlying forms] that tree=top  tall-adj+stv-npst ‘That tree is tall.’ [Positive; non-past]

e. an    taa-sa-n     kɨɨ=ba   an  taa-sa+ar-n    kɨɨ=ba   that tall-adj+stv-ptcp  tree=acc mj-ɨ! mj-ɨ [Underlying forms] look.at-imp   ‘Look at that tall tree!’ [Positive; non-past; ad­nom­ inal (relative) clause] 35   The stative verb nə- necessarily takes the negative affix -an or the nega­tive polar question affix -azɨi (the surface form is nə-əzɨi).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

amami, okinawa, norther n ryukyuan l anguages Adjectival phrases in Shuri (Okinawa) are composed of an adjective (or an adjective plus a stative verb). Example (10a) shows that Shuri has an adjectival form composed of an adjectival root and the adjectival affix -sa, i.e. taka-sa (high-adj). Adjectives in affirmative polarity in Shuri, however, usually contain inflectional affixes from Table  13.45, which are the same inflections as verbs have; thus, adjectives in affirmative polarity may be analyzed as a subclass of verbs. In (10b) and (10c), the adjectives take inflectional affixes, i.e.  the assertive affix -n of taka-sa-n (high-adj-ass) in (10b) and  the dubitative affix -gajaa (plus the derivational affix -tee) in taka-sa-tee-gajaa (high-adj-ass.pst-dub) in (10c). In nega­tive polarity, Shuri uses the stative verb ne-en (stv-neg),36 which follows adjectives containing the adjectival affix -ku (Kokuritsu Kokugo Kenkyuujo 1963: 82–3). The adjectival affix -ku is normally followed by the topic marker =ja in nega­tive polarity, and it becomes -ko=o (adj=top) by a mor­ pho­phono­logic­al rule (Kokuritsu Kokugo Kenkyuujo 1963: 83). (10)  Adjectival phrases in Shuri (Okinawa) a.  anu  mui=nu  taka-sa.   that  hill=nom high-adj   ‘That hill is high.’ (Kokuritsu Kokugo Kenkyuujo 1963: 84)

verb ar- have finished the contraction, and all the inflectional affixes (except for semantically impossible (in Shuri) combinations such as the imperative affix) can follow adjectival stems directly. In other words, the adjectives in Shuri are more grammaticalized than those in Yuwan from a morphophonological point of view.

13.4.6 Compounds In this section, I will present some examples of compounds in Northern Ryukyuan. In Yuwan (Amami) and Shuri (Okinawa), nominal stems (N) can follow other nominal stems (N), verbal stems (V), and adjectival stems (A) in compounds. In addition, verbal stems can follow other verbal stems. A special type of adjectival roots (which cannot occupy word-initial positions) can follow verbal stems, e.g. cja- (opt), jass- ‘easy’, and cjagɨ- ‘seem’ in Yuwan and bu- (opt), jas- ‘easy’, and gi-‘seem’ in Shuri (Loveless 1963: 196, 173). All these examples will be given in Tables 13.46 and 13.47. Table 13.46  Compounds in Yuwan (Amami) Stems Examples

 

 

b. gaka=ga    ka-cjeer-u   jii=ja     deezina   painter=nom draw-res-ptcp  picture=top  very taka-sa-n. high-adj-ass   ‘Pictures drawn by painters are very expensive.’ (Yushiya 1999: 411)

 

Forms

Glosses

Meaning

N+N

sima+hinzjaa

(island+goat)

‘goat in Amami’

V+N

hing-i+madu

(escape-inf+time)

‘time to escape’

c. are=e    kuri=jaka   taka-sa-tee-gajaa.   that=top  this=comp high-adj-ass.pst-dub   ‘(I) wonder if that one was higher (or more expensive) than this one.’ (Yushiya 1999: 254)

A+N

siju+gin

(white+clothes)

‘white clothes’

V+V

izjas-i+kij-an

(let.go-inf+cap-neg)

‘unable to let (it) go’

V+A

jum-Ø+cja-sa

(read-inf+opt-adj)

‘want to read’

d. taki=du     taka-ko=o   ne-en-tar-u=i?   height=foc high-adj=top  stv-neg-pst-ptcp=plq   ‘Was (it) tall?’ (Yushiya 1999: 243) Comparing the data between Yuwan and Shuri, it is possible that a word (i.e. inflected adjectives) in affirmative polarity in Shuri, e.g. (10b–c), was formerly composed of two words (i.e. adjectives whose ending are -sa and the stative verb ar-) such as (9b) in Yuwan. In Yuwan, very restricted affixes, i.e. -i (npst) and -n (ptcp), induce the contraction of the adjective and the stative verb, such as (9d–e). In Shuri, however, all the combinations of the adjective and the former stative   It is possible that the stative verb has two root forms, i.e. ne- and nee-. The former is always followed by the negative affix -an, and becomes ne-en (stv-neg) (Yushiya 1999: 66). The latter is also followed by the negative affix -an, and becomes neer-an (stv-neg) (Yushiya 1999: 272). In addition, the latter can be followed by the politeness affix -jabi, and becomes nee-jabir-an (stvpol-neg) (Yushiya 1999: 66).

Table 13.47  Compounds in Shuri (Okinawa) Stems Examples

 

 

 

Forms

Glosses

Meaning

N+N

zii+maami

(earth+beans)

‘peanuts’

V+N

kam-i+mun

(eat-inf+thing)

‘food’

A+N

taka+dukuru

(high+place)

‘a high place’

V+V

jum-i+uwa-ju-n

(read-inf+finish-

‘finish reading.’

V+A

ic-i+bu-sa-n

(go-inf+opt-adjass)

‘want to go’

36

aff-ass)

Loveless 1963: 176–8, 185, 196

223

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Yuto Niinaga When verbal stems precede other stems in compounds, the preceding verbal stems take the infinitival affix -i (or -Ø).

13.5 Clauses Northern Ryukyuan languages have verb-final word order in the clauses. Clausal differences (i.e. between main clauses, adnominal clauses, adverbial clauses, and nominal clauses) can be marked by the verb-final affixes. Sentence types which correspond to three major speech-act distinctions, i.e. assertion, question, and command, are also marked by the verb-final affixes. See Section 13.4.4.2 and examples in Section 13.4.2.6 in detail. The Northern Ryukyuan languages have many particles other than case markers discussed in

224

Section  13.4.2.6. For example, topic marker =ja and focus marker =du and =ga. For more details, see Niinaga (2014: 433–512) for Yuwan (Amami) and Yushiya (1999: 467–78) and Loveless (1963: 98–100, 123–7, 133) for Shuri.

13.6 Lexicon In the Northern Ryukyuan languages, there are many loanwords from Chinese, but most of them are said to be borrowed through Japanese (Ishizaki 2004: 1). For example, it seems that dooi ‘logic’ and kasjəə ‘help’ in Yuwan (Amami) correspond to doori (道理) and kasee (加勢) in Japanese. In Shuri, sumuci ‘books’ and siwa ‘anxiety’ correspond to shomocu (書物) and sewa (世話) in Japanese (Kokuritsu Kokugo Kenkyuujo 1963: 40).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 14

Miyako, Ishigaki, and Yonaguni, the Southern Ryūkyūan Languages J oh n  R. B e n t l e y

14.1 Introduction Okinawa Prefecture in southern Japan is divided into two linguistic spheres: northern Ryūkyūan (see Niinaga, this volume: Chapter 13) and southern Ryūkyūan. In Japanese the southern Ryūkyūs, including Yonaguni, are commonly referred to as the Sakishima Islands, and I will collectively refer to these languages accordingly. Sakishima can be fur­ ther subdivided into two groups, Miyako and Yaeyama. For many years these various languages have been referred to as hōgen “dialects” in Japanese, but based on mutual unintel­ ligibility, it is clear that these “dialects” can actually be cat­ egorized as a number of languages. The major languages of Sakishima are Miyako (represented by Hirara), Ishigaki, and Yonaguni (represented by Sonai). In their native pro­nun­ci­ ation, these languages are known as mya:ku (or me:ku) ‘Miyako’, sïka ‘the languages of four villages (on Ishigaki Island)’, and dunan ‘Yonaguni’. There is a large expanse of ocean between the main island of Okinawa and Miyako, cre­ ating a natural barrier that has allowed for the development of innovations not found in the northern Ryūkyūs. Sakishima consists of 18 inhabited islands, with a current population of 106,068.1 According to the latest Okinawa ­prefectural demographic data (from 2013), the number of Sakishima residents over 65 years of age is 22,435, meaning that 21% of the population is elderly. Based on a statistical survey by Anderson (2015: 488—9), one can roughly con­ clude that fewer than 20% of the speakers are “full ­speakers” of their respective languages. Until 500 years ago the islands of Sakishima were pol­it­ic­ al­ly distinct from the northern Ryūkyūs, with Ishigaki as the regional capital of Sakishima. Records indicate that Miyako sent tribute to the Chūzan court in Shuri, the capital of Okinawa, in 1390.2 Under the protection of Chūzan, Miyako ships began to raid the islands in Yaeyama. Around the year  Census data from May 2017, Okinawa Prefectural statistics (http:// www.pref.okinawa.jp/toukeika/so/so_index.html). 2   This historical account is based on Asato et al. (2012: 90–6). 1

1500 Akahachi, a leader in Ishigaki, rebelled3 and an armada of 100 ships, with warriors from Okinawa and Miyako invaded and subjugated Ishigaki Island. With this an embassy from Shuri was based in Ishigaki to oversee the island, bring­ ing Sakishima under the umbrella of the Chūzan court. The languages of Sakishima are oral, and the current writing system is based on mainland Japanese. There have been recent attempts to record a number of these languages in the Japanese syllabary known as katakana, as opposed to Romanization. As two examples, the Ishigaki dictionary by Miyagi (2003) represents the language of sïka on Ishigaki Island in katakana, but it also provides a phonetic transcrip­ tion in IPA. There is also a language dictionary of the Miyako language by Yonaha (2003), completely written in katakana. Owing to differences in the phonology of the languages of the southern Ryūkyūs from that of standard Japanese, some orthographic innovations were required. Miyagi represents the Ishigaki vowel written as ï in our system in katakana with an attending smaller ウ, for instance シゥ sï or ミゥ mï, a transcription device that ultimately goes back to Miyara (1930). The phonology of the dialects of Miyako is vastly dif­ ferent from standard Japanese, so Yonaha has used several innovative symbols to represent a number of distinctive sounds in Miyako. For example, in Miyako the vowel /ï/ is acoustically so high that it is auditorily a fricative; thus, pïtu ‘person’ sounds as pstu, which Yonaha transcribes as ピストゥ. When words have this high, central vowel in word-initial pos­ ition, Yonaha has used katakana graphs such as サ˚ or ス˚; thus, ïzamma ‘parents’ is transcribed as ス˚サ˚ンマ. As Japanese society has finally come to the understanding that these languages are endangered—and must be protected and propagated—there has been greater interest in the lan­ guages of Okinawa, especially the southern Ryūkyūs. Maner Thorpe (1983) wrote a dissertation on the languages of the Ryūkyūan islands, and attempted to reconstruct a 3   Much of this historical information comes from stories written down by the conquering groups. It is plausible that Akahachi simply attacked the marauding pirates from Miyako, and Chūzan responded in kind, and this was later justified as “a rebellion.”

John R. Bentley, Miyako, Ishigaki, and Yonaguni, the Southern Ryūkyūan Languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © John R. Bentley. DOI: 10.1093/oso/9780198804628.003.0015

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley Ikema Irabu Minna

Ōgami

Miyako Kurima

Tarama

Hatoma

Yonaguni

Ishigaki

Kohama Taketomi Iriomote

Kuroshima Aragusuku

Hateruma

0

15

30

MILES

Figure 14.1  East China Sea

­ roto-­language. Nakamoto (1976) has investigated the phon­ p ology of the various islands, and his work is important for stu­ dents wishing to better understand the languages. Wayne Lawrence has written extensively on various linguistic issues regarding the languages of Sakishima, and has an important description in The Languages of Japan and Korea (2012). This same volume also has an important chapter on Yonaguni by Izuyama (2012). The most comprehensive work on these languages to date appears in Handbook of the Ryukyuan Languages (Heinrich et al. 2015).

14.2  Historical connections Even simple comparative work will make it clear that the lan­ guages of the Ryūkyūs are genetically related to mainland Japanese. This fact was pointed out as early as 1895, when Basil Hall Chamberlain (1895a: 460) argued that Luchuan (our Ryūkyūan) and Japanese were sister languages. Many s­ cholars have accepted this theory somewhat uncritically; Unger (2009: 96–106) has put forth a robust argument that Ryūkyūan should not be viewed exclusively as a sister language, but should be considered as a type of aberration, because it was formed through successive waves of linguistic contact. My own research (as well as that of a number of linguists) concludes that the ancestor of Ryūkyūan can be linked to an

226

earlier state of Japanese, and should be viewed as a sister language to Japanese. Based on evidence from verb classes and assibilation of the nominalizing particle, Serafim (2003) argues that Ryūkyūan split from a branch of Japanese in northeastern Kyūshū. Regarding the date of that split, Serafim (2008: 98) later argued, “This means that P(roto) J(apanese), the ancestor of later varieties of Japanese, can­ not be the O(ld) J(apanese) language of the Nara period, but must predate it” (my emphasis). Thus, according to Serafim, Ryūkyūan and Japanese had to have separated before the Nara era (710–94). Using innovations exclusively found in Ryūkyūan, Pellard (2015: 15–23) argues that Ryūkyūan is a sister language of Japanese, descending from Japonic. He also contends that the two languages split earlier than the seventh century, before the extant records in Japan were written. Building on the foundation laid by Pellard, we find there are a number of innovations that occur only in Ryūkyūan, and then drilling further down there are innovations found only in Sakishima: thus, Ryūkyūan is a sister language of Japanese, and Sakishima Ryūkyūan is a sister language of northern Ryūkyūan. First, consider three words spread throughout the Ryūkyūs that are absent in mainland Japanese: ‘pig’ PR *uwa: (Kasari Ɂwa, Shuri Ɂwa:, HR wa: Tarama wa:ku, IG o:, Hateruma uwa, YG wa:), ‘earth’ PR *mita: (Kasari mica, Shuri mica:, HR mta, Tarama mta, IG nta, Hateruma nta, YG Nta), and ‘male’ PR *weke-: (Kasari Ɂiŋga,

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages Shuri ikiga, HR bikidum, Tarama bikidum, IG bigiduŋ, Hateruma bidumu, YG biNga).4 There is also a stratum of vocabulary that appears to be innovative within Sakishima. Consider these three words: ‘sand’ Proto-Sakishima *enago: (HR mnagu, Tarama mnagu, IG ino:, Hateruma iʃo:ŋ, YG c’inaN), but Okinawan ʃina; ‘one person’ Proto-Sakishima *tapuka: (HR tavkya, Tarama tauke:, IG taNga:), but Okinawan tʃui; ‘autumn’ Proto-Sakishima *sisanasi: (HR ssansï, Tarama ssaŋsï, IG sïsanasï, Hateruma sïsancï), but Okinawan aki. Finally, Table 14.1 outlines time words as found in Old Japanese, Early Middle Japanese (EMJ), Proto-Sakishima, and Proto-Okinawan. Here is evidence that the languages of Okinawa follow mainland Japanese, but Sakishima has innovated a number of terms, where the term “three days” or “four days” is suffixed with *nari ‘before’ (YG nadi, Hatoma nati, Hateruma narï) or *ima ‘present’ (HR nna, YG nma), “today, the present day” being counted as the first day. The only reconstruction ­worthy of note is “today.” Almost every attestation through the archi­ pelago has kyu:, which goes back to mainland Japanese kyou. In the earliest surviving document with transcriptions, the 15thcentury manuscript “Translation of the Ryūkyūan Language” transcribes ‘today’ as 交哇, Late Middle Chinese kja:w-Ɂwa, or Early Mandarin kjaw-wa.5 Regardless of the reconstruction, it is apparent the Shuri informant was saying kyou wa, ‘As for today’. Nakamoto (1983: 236) over-reconstructs *kepu, Table 14.1 Time words in Old Japanese, Early Middle Japanese (EMJ), Proto-Sakishima, and Proto-Okinawan

which should go back to a pre-Ryūkyūan state. All existing forms can be traced back to kyou, suggesting this may have been a mainland word that replaced an earlier term. Aside from Yonaguni, a singular language situated on a tiny island only 70 miles (112 km) from Taiwan, both Miyako and Yaeyama have a variety of dialects. Depending on how granular a scholar is with regard to differences, the Miyako group of islands has roughly 20 dialects, and there are about 15 dialects in Yaeyama. As far as internal phonological distinctions, in Miyako word final *ri has become a lateral in the dialects of Irabu Island, but in Hirara the tap has lenited and the front vowel backed to ï. On Irabu ‘who’ is taru, but in several dialects on the island of Miyako the tap has lenited, and au has under­ gone monophthongization, to:. Finally in the old area of Irabu Village there has been a replacement of most of the personal pronouns (Table 14.2). The islands of Minna and Tarama are located about 43 miles (70 km) west of the closely clustered islands of Miyako, Irabu, Kurima, and Ōgami. The dialect of Tarama often pre­ serves a velar g before a high mid-vowel, where Nagahama in Irabu has a bilabial fricative: ‘scrape’ tuv (Nagahama) ­versus tugï (Tarama), ‘row’ kuv (Nagahama) versus kugï (Tarama), and ‘peel off’ piv (Nagahama) versus pigï (Tarama). The dialects in Yaeyama demonstrate a number of related changes. One change between the dialects is deaffrication of ci > cï > si in Hateruma, Taketomi, and Kuroshima, while Ishigaki and Iriomote preserve the affricate. Written docu­ ments and some older speakers preserve words with pa- or pu-, such as pana ‘flower’ and pusuŋ ‘decrease’, but many speakers now articulate these words with a bilabial fricative ɸ: fana and fusuŋ. Interestingly, where most *pu have len­ ited to fu, the dialect on the tiny island of Hatoma preserves the older pu. There are also some dialects in Yaeyama that monophthongize vowels, while others preserve the ­diphthong: ‘frog’ IG auda versus Hateruma ota, and ‘burn’ IG muiŋ, Sonai mo:siŋ, and Hateruma më:ruŋ.

OJ/ EMJ

ProtoSakishima

ProtoOkinawan

Two days before

---

*yo:kanari

*wototopi no asita

Day before yesterday

wototopi *mi:kanari

*wototopi

Yesterday

kinopu

*kinou

*kinopu

Today

kepu

*kyou

*kyou

Table 14.2  Personal pronouns of Irabu and Hirara

Tomorrow

asu

*asita

*asita

Pronoun

Irabua

Hirara

Day after tomorrow

asate

*imami:ka

*asate

I

a(n)

ba:

Two days after

---

*imayo:ka

*asate no asita

We

anti

banta

Thou (vulgar)

ya(n)

vva

You (vulgar)

iti

vvata

Thou (elevated)

---

unzyu

You (elevated)

---

unzyuta

Time

  Unless specified Shuri data are from Kokuritsu Kokugo Kenkyūjo (1963); Miyako data are from Hirayama (1983), Ishigaki data are from Miyagi (2003), while Yonaguni data are either from my own fieldwork or Miyara (1930). Other Sakishima data are from Miyara. Language abbreviations: PR Proto-Ryūkyūan, HR Hirara, IG Ishigaki, YG Yonaguni. 5   Late Middle Chinese and Early Mandarin are from Pulleyblank (1991). 4

  Old Irabu data is from Shimoji (1979).

a

227

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley

14.3 Phonology 14.3.1 Consonants The consonantal inventory of the languages of the southern Ryūkyūs is represented in Table 14.3. The letter c represents [ts~tʃ] in all three languages, while the letter z often represents a voiced affricate [dƷ], as well as a fricative [z] in rare cases. In trying to capture the phono­ logic­al features of the various consonants across the spec­ trum of dialects in the Sakishima Islands, it should be understood that not all the symbols above appear in all ­languages. For example, the voiced fricative /v/ appears only in the dialects of Miyako. The following description exemplifies the phonology across the three languages.

14.3.2 Vowels The three languages under consideration have between three and six vowels, as noted in Table 14.4. The vowel ï represents [ɨ], though some scholars have argued for [ɿ] in Miyako (cf. Karimata 2005). If we restrict our analysis to vowels within morphemes, and exclude those that are morphophonemically created, then Yonaguni has the fewest contrasts, preserving only three vowels (i, a, u), while most dialects of Miyako and Yaeyama have six. Table 14.3  The consonantal inventory of the languages of the southern Ryūkyūs

Hateruma has seven vowels, with a contrast between /e/ and /ë/, but as I have argued earlier (2008a: 136–8), the vowel /ë/ is the result of monophthongization. The follow­ ing list (Table 14.5) provides a comparison of nouns from the three representative languages (Hirara, Ishigaki, and Yonaguni). A variety of changes have taken place in the three lan­ guages. One change seen across the languages is medial vowel loss, resulting in syllabic compression, creating gemin­ates and velar or syllabic nasals. Examples include *kura ‘child’ (HR ffa:, IG fa:, and YG ha), *siru ‘white’ (HR ssu, IG sïsu, and YG c’u:), and *emi ‘ocean, sea’ (HR im, IG iŋ, and YG uNnaga). In Yonaguni syllabic compression has been quite extreme, where k’uN ‘to use’ with its glottalized velar is evidence that contraction has occurred: *tukauN > *su̥kḁuN > *skuN > k’uN. Another change is plosivization of approximants, where *w becomes b, and *y becomes d: *wata ‘belly’ (HR bata, IG bata, and YG bata) and *yama ‘mountain’ (HR yama, IG yama, and YG dama). Like mainland Japanese, some Sakishima phon­emes have also undergone palatalization: *ti ‘blood’ (HR cïsï, IG cï:, and YG c’i:) and *s- ‘do’ (HR ʃi [infinitive], IG ʃi: [imperative]). The verb ‘to do’ in Yonaguni has been replaced by the verb iruN or kiruN, which are a transitive and in­transi­ tive pair. Affrication also appears in a number of languages. In the Miyako dialects of Ikema and Nagahama the verb kak‘to write’ shows affrication of the second velar when in the adnominal form: kaki ‘writing’ versus kacï ‘the written X’. Finally, nasalization occurs in very limited environments. In this context nasalization refers to a non-nasal mora losing its vowel through devoicing, and then the consonant losing

 

Labial

Coronal

Dorsal

Laryngeal

Plosives

p b

t d

k g

 

Glottalized

pʹ

tʹ

kʹ

 

Table 14.5  Comparison of nouns in Hirara, Ishigaki, and Yonaguni

Aspirated

 

t

h

k

 

HR

pïti-

Affricates

 

c z

 

 

IG

pïti-

taru

kai

Glottalized

 

cʹ

 

 

YG

t’u

ta:

kaNgi

Fricatives

f v

s (z)

 

h

HR

icï

Nasals

m

n

N

 

IG

icï

sïsu

 

Liquid

 

r

 

 

YG

ici

c’u:

 

Approximates

w

 

y

 

HR

midum ‘woman’

nu:

h

‘one’

‘when’

taru

ssu

‘who’

‘white’

‘field’

kagi ‘shade’

 

num ‘flea’

IG

mi:duŋ

nu:

nuŋ

Table 14.4  Vowel system of Hirara, Ishigaki, and Yonaguni

YG

minuNga

nu:

nuoŋ

High

i

 

ï

 

u

HR

pïruma ‘daytime’

uwari

‘to end’ yuï

Mid

 

e

 

o

 

IG

pïro:ma

uwarï

yuŋ

Low

 

 

a

 

 

YG

cu:ma

uwaruŋ

duyu

228

‘lily’

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages its place of articulation and assimilating to the neighboring nasal. Ikema (Miyako) and Yonaguni demonstrate a similar development: ‘cloud’ *kumo is Ikema m̥ mu and Yonaguni Nmu, while *pune ‘boat’ becomes Nni in Yonaguni.

14.3.3  Syllable structure The general syllabic structure in Sakishima is C1YV1V2C2, where Y is a glide. In Yonaguni the final consonant is N, a mora nasal. Miyako has cases of double consonants in the C1 position, such as ssu ‘white’, vvyamasïkaï ‘envious’, or ffu ‘black’, due to the devoicing of vowels. All three languages have the same syllabic constraint: the minimal length of a word is two morae. Thus, monomoraic morphemes in main­ land Japanese are lengthened in Sakishima (as well as Okinawa): pa: ‘tooth’, ti: ‘hand’, and pu: (YG hu:) ‘sail’.

14.3.4 Morphophonology Because these languages are agglutinative, a variety of changes occur when case markers, particles, or other suffixes are attached. In general, three main types of changes are observed: monophthongization, contraction, and gemination. In Hirara monophthongization occurs when the topic marker (y)a is affixed to nouns, for example pïto: (from pïtu+a). In Tarama, when the topic marker is affixed to a word ending in -i, monophthongization occurs, resulting in e: kadi ‘wind’ versus kade: ‘as for the wind’ (from kadi+a). This same type of phenomenon appears in the same en­vir­ on­ment in Ishigaki: kagi ‘key’ becomes kage: ‘as for the key’ (from kagi + ya). In an environment where the proceeding mora is palatalized, monophthongization occurs, but the palatalized consonant is preserved: kiʃi ‘scrap’ but kiʃe: ‘as for the scrap’. In Hateruma a secondary vowel, ë, with vowel length, appears when a counter is affixed to a numeral, as example (1) demonstrates. (1) (i)  pïtu + ïra > pïtë:ra ‘one flat object’ (ii)  futa + ïra > futë:ra ‘two flat objects’ (iii)  mi: + ïra > mï:ra ‘three flat objects’ Contraction has occurred predominantly in verbal forms, where complex forms have been compressed into simpler forms. Consider examples (2) and (3), one each from Hirara and Ishigaki. (2) Hirara ami nu ff-ada-kara rain nom fall-neg-cond ‘If it does not rain….’

In example (2), the verbal form (underlined) is a contraction of an earlier *pur-ada-kara rain-neg-cond. (3) Ishigaki para-ru-na:-tta soŋga kïno: yesterday-top go-deb-neg (inf)-pst but ‘I was not able to go yesterday, but….’ (Miyagi 2003) Example (3) is actually more complex. ‘Today’ kïno: is a ­contraction of kïnu ya, where the glide lenites, and the two vowels undergo monophthongization (kïnu ya > kïnu ya > kïnua > kïno:). The earlier form of the verb is *para-ro-nak-u ar-ita (go-deb-neg-inf to be-inf-pst), with /ku/ leniting to fu then to zero, and the medial /ri/ completely disappears, creating the geminate and vowel length. Finally, gemination occurs frequently in the dialects of Miyako when the topic marker or the object marker (u) ­follows a noun that is consonant-final: HR pavva ‘snake (pav) + top’, HR inna ‘dog (in) + top’, or HR kwa:ssu ‘candy (kwa:sï) + acc’, HR yuvvu ‘rice gruel (yuv) + acc’. This same phenomenon occurs to a lesser extent in Ishigaki with the topic marker: IG innya ‘dog (iŋ) + top’.

14.3.5 Suprasegmentals There has been some divergence and innovation in the accent systems of the three languages under discussion. In many cases each dialect has a different type of accent ­system. In Miyako a number of dialects have lost all tonal distinctions, while they are still preserved in Tarama (cf. Bentley 2008a: 85). Matsumori (2008: 106–9, 2010: 494–502) argues for at least three distinctions as preserved in data gathered from the dialect on Tarama island. As Matsumori (2010: 494, fn. 4) argues, this three-way distinction is appar­ ent only when a particle is affixed to the noun. This work is promising, but still preliminary, so for the time being I pro­ vide the following simple overview of the accent system in Tarama in Table 14.6. Somewhat as in Tarama, most dialects in Ishigaki have a two-way tonal system, marked and unmarked, as the data in Table 14.7 demonstrate. Table 14.6  Accent system in Tarama Unmarked

Marked

pàà ‘leaf’ f ùcï ̀ ‘mouth’

pàá ‘tooth’

tàkàrà ‘treasure’

gàráʃà ‘crow’

fúsï̀ ‘comb’

229

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley Table 14.7  Two-way tonal system of Ishigaki dialects Unmarked

Marked

pàà ‘tooth’ fùcï ̀ ‘comb’

páà ‘leaf ’

tàgàrà ‘treasure’

pàkámà ‘trousers’

fùcḯ ‘mouth’

Table 14.8  Three-way tonal distinction of longer words in Yonaguni Low

High

Falling

hàà ‘tooth’

háá ‘leaf’

---

kùcì ‘comb’

kùdí ‘nail’

kúi〵 ‘voice’

tàgàrà ‘treasure’

kùNgání ‘gold’

tùgára〵 ‘snake’

Yonaguni has a three-way tonal distinction, low, high, and falling. With monomoraic words, high versus low is the only distinction. The three-way distinction appears in longer words, as noted in Table 14.8.

myu:ni ‘the august boat’, miumaituga ‘the august path (to Okinawa) before us’, and miya ‘the august house (of the deity)’. The third prefix is u-, found in polite usages such as ukagi ‘thanks due to a superior for their assistance’, uturïmucï ‘a reception, receive a guest’, and utaki ‘a peak respected as a place where spirits dwell’. A suffix can also be attached to nouns to show respect: mai denotes respect, much as sama does in modern Japanese: wa:mai ‘your honor’. Levels of respect are also demonstrated by specific personal pronouns, explained in Section 14.4.2. The languages of Sakishima have a rich variety of clitics (case markers, particles, and other markers), listed al­pha­bet­ ic­al­ly: (y)a topic marker, du a kakari musubi particle, ga nom­ ina­tive, kara ablative marker, n/ŋ a focus marker, ni locative marker, nu nominative or genitive marker, tu ­associative marker, and (y)u accusative case marker. Table 14.9 is a list of clitics found in the three languages (data respectively from Miyagi 2003, Hirayama 1983, and Hirayama 1964). An emerging trend in the languages of Sakishima is to leave unmarked both the nominative and accusative cases as word order has begun to mark case, as in English (Aso 2010: 200–1). When marking is deemed necessary to avoid Table 14.9  Important clitics in Ishigaki, Hirara, and Yonaguni

14.4 Morphology

Function

Marker

IG

HR

YG

NOM

ga

☓

○

○

14.4.1 Nouns

NOM

nu

○

○

○

TOP

(y)a

○

○

○

ACC

(y)u

○

○

☓

GEN

ga

☓

○

☓

GEN

nu

○

○

○

LOC

ni~N

☓

○

○

LOC

kai

○

☓

○

ABL

kara

○

○

○

INS

ʃi

○

○

○

INS

kara

○

○

○

ASSOC

tu

○

○

○

ALL

kai~ŋkai~ŋki

○

○

○

COMP

yuri

☓

○

☓

COMP

yaka

○

☓

○

QUOT

de~ti

○

○

○

FOC

du

○

○

○

FOC

mai

☓

○

☓

FOC

ŋ~N

○

○

○

In the languages of Sakishima nouns are manipulated by the addition of prefixes or suffixes. Two of the most productive types are plurals and markers of respect. In Ishigaki there are two plural markers: nume: and ta:~da:. The first attaches to animate and inanimate nouns: yarabinume: ‘the children’, dusïnume: ‘friends’, and mayanume: ‘cats’. The second suffix attaches to all objects: kuritta: ‘these things’, piʃida: ‘the daughters’, and bagada: ‘we’. There are several ways to use nominals to denote respect. The two most productive ways are (i) to add a prefix or suf­ fix, and (ii) to use a noun reserved for a specific, honorific function. In Miyako and Ishigaki there are three prefixes introduced from the prestige language of Shuri, terms which his­tor­ic­ al­ly came from mainland Japan. Of highest respect is gu-, as in guburi: ‘denoting that what a subordinate does brings dishonor’, guhairo: ‘receive something (as a gift) from a superior’, and guju: ‘on official business’. These words are remnants from a bygone era of feudal rule. The second prefix is m- or mi-, which is clearly related to the honorific mi in mainland Japanese. In Miyako folklore recorded by the Russian scholar Nikolai Nevskii (1892–1937), Karimata et al. (1998: 32, 35, 102), we have examples such as

230

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages confusion, the three language groups in Sakishima mark the nominative case with either ga or nu. Yonaguni does not mark the accusative, akin to the languages of the ­northern Ryūkyūs, but Miyako and Ishigaki mark the ac­cusa­tive with (y)u; however, marking is not compulsory. In the following examples, (4) and (5), the (i) nominative case marker (4) and accusative case marker (5) are dropped (Ø) in Ishigaki. (4) du: cu:sa ara-ba body.Ø strong.inf be-if ‘If your body is strong….’

du

to the second-person pronoun to elevate the listener: wa: ‘you’ versus wa:mai ‘your honor’. This same suffix is affixed to proximal or distal pronouns to show respect to people in the third-person: kunu pïtunumai ‘this honorable person’, or unu pïtunumai ‘that honorable person’. On the other hand, adding nza to a pronoun creates a derogatory term: wanza ‘you (scoundrel)’, kunza ‘this (fool)’, or kanza ‘that (chump)’.

14.4.3  Numerals and classifiers

foc

(5) fa:ya tamu nuba tur-i katan child.top firewood.Ø  ptcl take-inf shoulder u   ka:   nu   pagiruŋ  keŋ   tamu     nuba gen skin nom  peel.off  until  firewood.Ø  ptcl katamidir-i      du carry.completely-inf  foc ‘My children would carry firewood on their shoulders until the skin completely peeled off!’

While Sino-Japanese numerals from the mainland have spread throughout the islands, the native, cardinal nu­mer­ als in the languages of Sakishima are still preserved in the memory of elderly speakers. These numerals are essentially the same among the three languages, aside from predictable sound changes. Table 14.12 outlines the numerals in Ishigaki, which is representative of the entire area.6 These languages use Sino-Japanese for numerals above 1000. Ten thousand is maŋ. Miyagi (2003: 2–13) gives the following

14.4.2 Pronouns

Table 14.10  Personal pronouns

Like mainland Japanese, the languages of Sakishima have two types of pronouns: personal and demonstrative. Table 14.10 and Table 14.11 outline these two groups in the three languages.  As is true of mainland Japanese, different personal pronouns are reserved to demonstrate respect to superiors. In Miyako the second-person pronoun vva is for equals or subordinates, while unzyu is for superiors. In Ishigaki a suffix -mai is added

Gloss

Miyako

Ishigaki

Yonaguni

1st person

ba(n)

banu

anu/banu

2nd person

vva

wa:

Nda

3rd person

kai

kari

kari

interrogative

taru/to:

taru/ta:

ta:

reflexive

du:

du:

du:

Table 14.11  Demonstrative pronouns  

Miyako

Ishigaki

Yonaguni

Gloss

Form

Place

Form

Place

Form

Place

proximal

kui

kuma

kuri

kuma

ku:

kuma

mesial

ui

uma

uri

uma

u:

uma

distal

kai

kama

kari

kama

kari

kama

interrogative-where

nzi

nza

ziri

zïma

Ndi

Nma

interrogative-which

ifucï

 

ifucï

 

iguci

 

interrogative-when

icï

 

icï

 

ici

 

interrogative-what

no:

 

no:

 

nu:

 

6

The section on numerals is heavily indebted to Miyagi (2003).

231

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley Table 14.12  Ishigaki numerals  

Ones

Tens

Hundreds

Thousands

1

pïtu-

tu:

tu:nu tu: / mumu ʃin

2

futa-

tu:nu futa:cï

pya:kunu futa:cï

niʃin

3

mi:-

tu:nu mi:cï

pya:kunu mi:cï

sanzin

4

yu:-

tu:nu yu:cï

pya:kunu yu:cï

yu:ʃin

5

icï-

tu:nu icïcï

pya:kunu icïcï

guʃin

6

mu:-

tu:nu mu:cï

pya:kunu mu:cï

rukuʃin

7

nana-

tu:nu nanacï pya:kunu nanacï

nanaʃin

8

ya:-

tu:nu ya:cï

pya:kunu ya:cï

haciʃin

9

kukunu- tu:nu kukunucï

pya:kunu kukunucï

ku:ʃin

reading for the number 876: pya:kunu ya:cï tu tu:nu nanacï tu mu:cï, literally ‘one hundred of (times) eight and ten of (times) seven and six’. The following classifiers are found in Ishigaki: (i) Animate objects: -rï ‘people (up to five people)’, -pïtu ‘people (when counting six or more people)’, -kara ‘living things (animals, birds, fish, insects)’, -dai ‘generations’. (ii) Measurement of size, time, or weight: -ka ‘days’, -ke:ra ‘times’ (as in pïtuge:ra ‘one time’), -naŋka ‘one week (of mourning after a death)’, -nicï ‘days’, -gwadcï ‘months’, -ti ‘years’, -suŋ ‘inch’, -iru ‘fathom’, -busï ‘inch’, -ʃaku ‘one shaku (about one foot)’, -kwaŋ ‘one kan (roughly 8.25 pounds)’, -kiŋ ‘one kin (about 1.3 pounds)’, -ʃu ‘one shō (about 1.5 quarts)’, -go: ‘one gō (about 1/3 of a pint)’, -tu ‘one to (roughly four gallons)’, -tabarï ‘a bundle (literally “one handful”)’, -maragï ‘ten bundles’, -mutasï ‘a mass, a lump’, -ci: ‘a pair (of some object)’. (iii) Shape: -mutu ‘cylindrical objects’, -sicï ‘very slender, pliable objects’, -mai ‘flat objects’, -saja ‘a bunch’, -ku ‘round objects’, -ira ‘flat, pliable objects’, -fubï ‘cloth­ ing (things that go around one’s neck)’. (iv) Miscellaneous objects: -pïsa ‘shoes’, -su ‘ships’, -kiburï ‘households’, -ki: ‘serving of food’, -cïŋ ‘food’, -cïkumi ‘handful’.

14.4.4  Verbal adjectives What is usually labeled keiyōshi ‘adjectives’ in Japanese is here termed ‘verbal adjectives’. In Ryūkyūan these appear historically to have been formed by affixing a nominative suffix -sa to the stem and then adding the existential verb

232

ar- ‘to be.’ Thus taka-sa ‘height’ becomes takasa:ŋ ‘to be high’. Or kai-sa ‘beauty’ becomes kaiʃa:ŋ ‘beautiful’. Consider three examples (6a–c) from the language of Ishigaki: (6) Ishigaki a.         

kunu yama nu du kanu this.gen mountain nom foc that.gen yama    yakan  takasa:-rï mountain  comp  high-attr ‘It is this mountain that is higher than that mountain.’

b. o:cïkï kaiʃa:-ŋ   weather  beautiful-concl   ‘The weather is beautiful (fine weather).’ c. appa  ŋ   iza-r-i      su    nu   du   mom  dat scold-pass-attr thing nom  foc nugurisa:-da afraid-pst   ‘I was afraid of being scolded by my mom.’ When a verbal adjective modifies a noun or noun phrase, the verbal adjective changes to an attributive form, as the following example (7) demonstrates: (7) a. umuʃir-i munu (< umussa:ŋ)   interesting-attr thing   ‘an interesting thing’ b.  nugur-i:   ʃinʃi (< nuguriʃa:n)   scary-attr teacher   ‘a scary teacher’ Negating a verbal adjective is done by attaching a negative suffix (-ne:nu) to the stem in (8): (8) a. takasa-ne:nu   high-neg   ‘It is not high.’ b. kunu timpura: mittʃa-ne:nu   this.gen tempura.top poor (tasting)-neg   ‘This tempura is not bad.’ The past tense of verbal adjectives is formed by adding the past suffix (-da) to the stem in (9): (9) a. kïno:      attsa:-da   yesterday.top hot-pst   ‘It was hot yesterday.’ b.  iku-da   cïkë:     kï-su       nu   go-pst  month top come-attr-thing  nom paiʃa:-da  yo: early-pst  emph   ‘You came early last month!’ (lit. Last month your coming was early!)

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages

14.4.5 Verbs Verbs in Sakishima demonstrate a system that gradually becomes more complex as the speaker travels southwest from Miyako toward Yonaguni. Miyako has a simple twoclass system, with consonant-stem and vowel-stem verbs. There are two irregular verbs, ‘to do’ and ‘to come’. The ­citation form in Miyako is generally the infinitive plus -m. Ishigaki also has a two-class verb system with two irregu­ lar verbs. However, because of several phonological changes a number of verbs have fluctuating stems. For example, the verb i:ŋ ‘receive’ is yu:nu ‘not receive’ in the negative, but the attributive form is i:. Clearly, the negative form is from iu, where the initial vowel becomes a glide: iu > yu. The citation form in Ishigaki is generally the attributive form plus -ŋ. Yonaguni has the most complex verbal structure of any language in Sakishima. Owing to heavy phonological changes which have induced intense syllabic compression verbal classes have superficially become more complex. I have ­earlier argued (Bentley 2008: 189–91) that this was not always the case, and it appears that Yonaguni originally had a verbal system more akin to that found in Ishigaki.

14.4.5.1  Negation Negation is formed by affixing a negative suffix (HR -aN, IG -anu, YG -anu’N): HR ka:N, IG ka:nu, YG kanuN ‘(I) won’t buy (it)’. In Hirara, the negative past tense is formed by adding the nega­ tive suffix (n/m) to the past suffix. In Ishigaki and Yonaguni the negative suffix attaches to the stem, and the past-tense suffix attaches to the imperfect of the negative, as in (10–12). (10) Hirara kazi n fuk-ai    du ku:-rai-tta-m wind dat blow-pass  foc come-pass-pst-neg ‘I was blown about by the wind and could not come (to your house).’ (11) Ishigaki kari du ba: na: kak-ana:-tta foc my name write-neg-pst he ‘He did not write down my name.’ (12) Yonaguni u ya kag-anu-taN it top write-neg-pst ‘(I) did not write it.’

14.4.5.2  Aspect/tense The languages in question have a past/non-past binary ­system. The past tense is formed by affixing the following

verbal auxiliary to the verbal stem: HR -ïta’ï, IG -uda, YG -taN. Examples are HR tu’ïta’ï, IG turuda, YG tutaN ‘(He) took it’. The non-past (that can also be analyzed as the predicative form) is formed in Miyako by attaching -m to the infinitive of the verb, while in Ishigaki -ŋ attaches to the attributive form. In Yonaguni the non-past is formed like that in Ishigaki, with -N attaching to the attributive form. In Miyako aspect is communicated by adding an aspectual auxiliary verb (u’ï) to the conditional form of a verb (stem+i), as in cïzïki u’ï ‘continuing’. Ishigaki has the same type of con­ struction, but the aspectual auxiliary verb attaches to the infinitive: cïzïgidu uda ‘It continued’, while Yonaguni again mirrors Ishigaki: cidiki buN ‘(He) continued (doing it)’. Ishigaki also has an aspect marker, -itta, which denotes that an action has been completed: kak-itta write-pst ‘I have fin­ ished writing it’. This same aspect appears in Yonaguni as kat-yaN write-pst ‘I have written (it)’. Hirara has a similar system, but the auxiliary is the same as the past marker. The difference is that while the past auxiliary attaches to the infinitive of the verb, the aspect marker attaches to the con­ clusive, as in (i) uk-ita’ï ‘(I) received it’, versus (ii) uk-i’ïta’ï ‘(I) have received it’. Questions in Sakishima are formed by adding a question particle at the end of the sentence. In Miyako a yes-or-no question is formed by placing the question particle na at the end of the sentence, as in (13). (13) uya nu a-ï munuï nu du father  gen say-attr words nom  foc ssai-n      na understand-neg  q ‘Don’t you understand what your father is saying?’ Hirara also inserts a question particle, ga, either within a sentence (a) or at the end (b) to ask a question that requires specific information, as in (14). (14) a. uma n u-taï so: to: ga ya-ta’ï   there  loc  be-pst  thing  top  who  be-pst   ‘Who was (the person) that was over there?’ b.  uma   n   u-tai   sïtu:     taru  du   there  loc be-pst person-top  who   foc   a-tai   gai:   be-pst  q.emph   ‘Who was that person who was over there (I won­ der)?’ (from Ikema) In Ishigaki yu: is attached either medially or at the end of a sentence to form a question. The particle wa: is also used, but this gives the sentence a sense of irony. A question with the particle ba is emphatic. The following examples (15a–c) reflect each case.

233

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley (15) a.  nara     ŋ   haru-ŋ  yu:   that.person  foc go-concl  q sïki-mi:     rya hear.inf-try.inf  q   ‘Go ask if that person also wants to go.’ (or ‘Go ask if they are going.’) b. nusïta  kare:  nu  anzirï  kutu   why   he  top  nom  that.thing s-ï     wa: do-concl  q   ‘Why would he have done such a thing? (He wouldn’t have!).’ c. wada:  m   ma:zoŋ   iz-am   ba   you.pl  foc together sing-neg  q   ‘Won’t you sing together with us?’ (i. e. ‘Please sing with us!’) Yonaguni has two particles to pose questions: ka, which also displays doubt on the part of the speaker, and na to pose a simple question, similar to Hirara (16). (16) a. anu N hir-u ka ya foc go-concl q ptcl   I   ‘Maybe I will go also.’ (lit. ‘Should I go also?’) b. Nda hir-anu  na   you go-neg  q   ‘Are you not going to go?’

14.4.5.3  Mood Here I limit my description to three moods: the imperative, prohibitive, and the intentional. In Miyako the imperative is formed by attaching the vowel -i to the stem: kak-i write-imp ‘Write!” ʃinʃi:n nar-i teacher ptcl become-imp ‘Become a teacher!’ The same construction is found in Ishigaki and Yonaguni: IG kak-i, YG kag-i ‘Write!’ A particle ya can also be added to intensify the command, IG kak-ya and YG kag-ya ‘Write (dammit)!’ In Miyako and Ishigaki the prohibitive is formed with na affixed to the attributive form of the verb, while in Yonaguni it affixes to a compressed indicative form: HR pama n ikï na, IG pama ŋ iku na, YG hama Nki hiNna ‘Do not go to the beach!’ In Miyako the intentional mood has two forms, one from just the imperfect stem of the verb, or one with -di affixed to the imperfect stem. Compare examples (17a) and (17b). (17) a. atu   icizikan   sï-cïka:   ika   after one.hour do-cond go ‘Let’s go in (after) one hour.’ b. acï karya: bo:ʃu: kavva di   hot because hat.acc wear ptcl   ‘I will wear a hat because it is hot.’

234

The intentional mood in Ishigaki is the same as the first form in Miyako, based on the imperative: guzo: kaka ‘I will write a letter’. The same mood in Yonaguni resembles the two forms found in Miyako, where a verb in the attributive can reflect a desire to do something, while an invitation to do something is communicated with the verb in the infini­ tive plus daNgi, as in example (18). (18) a. nni   gara  hir-u:   boat  ins    go-vol   ‘Let’s go by boat.’ b. mituda nai-bi: raku raku ni   man and wife become.pol  easy.adv kurasi  i-daNgi live   do-vol   ‘Let’s become man and wife and live a life of ease.’ (Iwase 1983: 56)

14.4.5.4  Voice There are two voices, the causative and passive. The three languages under discussion all share the same feature in relation to the passive, where it can represent both a passive and a potential (debitive), and context makes the distinction clear. In Miyako the passive (-rai’ï) attaches to the stem of the verb. The liquid and its vowel (ra) are suppressed when attaching to a consonant stem, see (19). (19) Miyako a. kai  n   na  ba:  ya  abi-rai-n   he  dat  top  I   top call-pass-neg   ‘I am never called by him.’ (‘He never calls me.’) b. ba nu mai yum-ai   I nom foc read-deb   ‘Even I can read (that).’

gami until

du foc

sï do

In Ishigaki we have: (20) Ishigaki a. ba  nu   kai  ŋ   kaka-riŋ   I   nom  dat also write-pass   ‘I can also write that.’ b. uyav nu kutu nu umo:-riŋ   parent gen thing nom think-pass   ‘I think about my parents.’ (lit. ‘It is about my par­ ents that I think.’) As Shimoji (2008: 493–4) points out, this sentence may have been created in the passive because of some adverse reaction, such as sadness. Here are two examples (21) from Yonaguni:

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages (21) Yonaguni a. nu ba kaga-riru   what ptcl write-pass   ‘What can you write?’

Nga q

b. kaga-ritti-waru-N   write-pass-(ger)-ex-concl   ‘(The boss) was writing.’

(26) a. bana: mainicï hatai kai paru-ŋ every.day field loc go-concl   I.top   ‘I go to the field every day.’

There are two causative suffixes in the languages of Miyako and Ishigaki, HR -asï and -asïmi’ï, and IG -asïŋ and -asïmi(ru)ŋ. Yonaguni has only one suffix, -amir, which is a compressed version of Ishigaki’s second suffix. An example from each language is below (22–24): (22) Hirara ffa n funn u yuma-sïmi’ï child dat book acc read-caus ‘I had (him) read a book to the child.’ (23) Ishigaki uri kai guzo: kaka-ʃa he dat letter.Ø write-caus ‘I had him write a letter.’

b. kuma naŋga  u-ŋ   here dat   be-concl   ‘I’m right here.’ Ishigaki has two different conditionals, (i) a tentative condi­ tional, or a conditional based on a set of assumptions, and (ii) a factual conditional, based on a set of facts. (27) takes the particle kka:, while (28) takes ya: (27) paiʃ-a uki kka: maniau-ŋ early-adv rise ptcl be.on.time-concl ‘If you get up early you will make it (on time).’ (28) fa: yu iz-a: barasa child acc scold-cond bad ‘Is it wrong if I scold the child?’

(24) Yonaguni u nu sa Nki minaga hug-amir-i sweep-caus-imp he gen self ptcl yard ‘Have him sweep the yard, himself!’

14.4.5.5  Predication I specifically address two forms of predication—the final form and the conditional. As noted above, the predication form, also called the conclusive or final form, has a nasal at the end. In Miyako there are two different verbal forms, usually labeled the infinitive and the predicative. Interestingly, one form of the infinitive is the same phono­logic­ally as one of the predicative forms. Thus, we have kakï ‘write’ versus kakïm ‘write’. The first form can take the prohibitive na, as in kai yu kakï na ‘Don’t read that!’ The other form with -m, is a strong indicative form, as it takes emphatic particles, such as kakïm do: ‘(I will) write this!’ The suffix -gagara ssaiN ‘maybe’ also follows this form (25). (25) kyu: mai kïsi-m today-top also come-concl ‘(He) may come today also.’

the particle seems to alter the meaning, there are five basic meanings: (i) incomplete aspect (future), (ii) intention, (iii) habitual action, (iv) statement of fact, and (v) present aspect. Examples of (iii) and (iv) follow in (26).

gagara ssaiN maybe

There is only one predicative form in Ishigaki, but it has ten different meanings or usages, according to Miyagi (2003: 34–5). If we simplify his list, stripping out examples where

du foc

ar-ï be-inf

In example (28), ‘scold’ is iza+ya, which is simplified to iza:, preserving vowel length.

14.5 Syntax 14.5.1  The clause The languages of Sakishima overall share a common sen­ tence structure. Nouns are affixed with case markers and particles, thus allowing word order to be manipulated in order to emphasize certain elements, while avoiding ambi­ guity. The basic word order of these languages is sov, with the verb at the end of a clause or sentence, where much of the “conjugating” takes place at the end of the sentence. Consider a typical sentence from each language group (the following three examples (29a–c) are from Miyara 1930: 133): (29) a. Hirara   pai   ŋkai  ni   san   cyo:   iki-ba  pama beach   south  dir  two three blocks go-if   ŋkai  idi’ï-m    yo:  dat  reach-concl  emph   ‘If you go south two or three blocks you will reach the beach.’

235

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley b. Ishigaki   paita  kai     ni  san  cyo:  south direction dir two three.blocks    paru-kka:  pama  kai  uri-ŋ   go-if   beach dat go.down-concl   ‘If you go two or three blocks in a southern direction, you will go down to (reach) the beach.’ c.          

Yonaguni hai nu kata Nki ni san cyo: south gen direction dir two three blocks hita-ya  hamabata  nki  tundiru-N go-if  beach   dat reach-concl ‘If you go two or three blocks in a southern direction, you will reach the beach.’

Each language represents nominal coordination in exactly the same way, by inserting tu ‘and’ between nouns (A tu B), as in kuri tu uri tu ‘this and that and….’. Another coordinat­ ing particle used between nouns represents a choice, analo­ gous to modern Japanese demo, are demo kore demo ii ‘That or this, either one is fine’. However, in Sakishima each of the three languages uses a different particle for this function. For example, Hirara inserts mai ‘too’: kari mai kuri mai zyo:kaï ‘This too or that too, either is fine’. Ishigaki uses yarabaŋ ‘either or’: kari yarabaŋ kuri yarabaŋ misaŋ, ‘Either that or this, either (one) is fine’. Finally Yonaguni employs attaNttiN ‘either or’: ku: attaNttiN kari attaNttiN NsaN ‘This or that, either is fine’. The representation of “existence” in these languages is closely related to the usage in mainland Japanese, employ­ ing a form either iru or oru ‘to be, exist, to have’. The verbs in question are: HR u’ï, IG uŋ, while YG has buN. Consider example (30): (30) a.       b.      

Hirara pïtu   nu   du  u’ï  saika person  nom  foc be  ptcl ‘There is someone home, isn’t there?’ Ishigaki icïmadi-n   na:i  un-na   yo: always-cond  not  be-proh  ptcl ‘Don’t you dare remain as you are now!’

c. Yonaguni   ara:gu  katta:-ru   duci   Ntaintu   very  close-attr friend three.people  bu-taN be-pst   ‘There were three friends who were very close.’ (Iwase 1983: 197)

236

14.5.2  The nominal group The general rule in Sakishima is that nu marks the subject, while (y)u marks the direct object. Miyako differs from Ishigaki and Yonaguni in that ga marks the human agent of an action, while nu marks all other subjects that are not used pronominally. Thus “the alternation between =ga and =nu implies that indefinite and newly introduced referents, which cannot be replaced by pronouns, are always marked by =nu even if the referents are kinship terms, social status terms, etc.” (Shimoji 2008: 194–5). In Ishigaki ŋa is not used as a marker in isolation. In Yonaguni the subject marker is ŋa, while nu is an attribu­ tive marker. All three languages mark case with a case marker after the noun, but zero case marking is widespread in Ishigaki and Yonaguni. Contrast Hirara with Ishigaki in (31): (31) a. Hirara   ba  ga   aï-ta’ï   kutu  u   ba:  I   nom say-pst thing acc  ptcl  ubuinu-u’ï remember-prog   ‘Do you remember what I had said?’ b. Ishigaki   wa:    unu    kutu   sikattu    you-top this.gen thing.Ø clearly  u:-ya:N7 remember-prog   ‘Do you clearly remember this thing (event)?’ Topic marking in Sakishima employs the marker (y)a, which can coalesce into vowel length when following a vowel-final noun. (32) a. Hirara fun   ya:  kuma  kara  du  idi’-ï   boat  top  here  from  foc depart-concl   ‘A boat is departing from here.’ b. Ishigaki ai ya yaima ŋga ŋ ibe:-da   Indigo top Yaeyama loc foc grow-pst   ‘(I know) Indigo plants used to grow in Yaeyama also.’

7   The speaker of this sentence, Mr. Kuroshima (b. 1895), appears to have used a lenited form of ubuiruŋ ‘to remember’, where medial bu­ has weak­ ened: bu > fu > u (cf. Nihon hōsō kyōkai 1972, 2: 236).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages c. Yonaguni siNsi   ya  gu-zi     mai   ni  teacher  top  five-o’clock  before  ptcl  ugi-taN get.up-pst   ‘The teacher got up before five o’clock.’ There are two ways to mark focus in the languages of Sakishima: an inclusive focus marker (mai in Hirara and m, n, ŋ in Ishigaki, and N in Yonaguni), and kakari musubi, which is a syntactic agreement of a certain particle and the predi­ cate ending of a verb. In Sakishima a kakari musubi marker forces the verb into the attributive case. Below is an example of the inclusive particle (33). (33) a.      

Hirara bam mai iki:-mi:-di foc go-try-vol I ‘I will also go have a look, OK?’

b.      

Ishigaki kuri ŋ kari m ba: munu this foc that foc I.poss thing ‘Both this and that are mine.’

c.      

Yonaguni anu N hiru kaya foc go q I ‘Would it be OK if I went, also?’

Another type of focus is known as kakari musubi. In relation to kakari musubi, Miyako and Yonaguni have a reduced ­system, where only the particle du survives. Below is an example of each (34). (34) a. Hirara pya:pya:ti ik-aba du zyo:kaï   quickly go-cond foc be.good   ‘It is better if you go quickly.’ b.      

Yonaguni u   du  tur-i this  foc take-imp ‘Take this!’

According to Miyagi there are two kakari particles in Ishigaki: du and yu (2003: 72–3). He argues that du cannot be used in an imperative sentence, so yu replaces it. Also, yu appears in questions; however, it appears that the ac­cusa­ tive marker simply replaces du. Consider that when a verb contains an invitation to the listener to do something or

when the speaker expresses a desire, du cannot be used, but is replaced with yu. This appears to be a constraint on the usage of the focus particle. Below are four examples in (35), illustrating du and yu. (35) a. saki  du  num-u,    cya:  ya    sake  foc drink-attr tea  top  num-an-u drink-neg-attr   ‘I will drink sake; I will not drink tea.’ b. kuri yu tur-i   this acc take-imp   ‘Take this one!’ c. cya: yu num-a   tea acc drink-vol   ‘Let’s drink tea.’ d.        

wa: ya kuri yu tur-ya, bana: kari you top this acc take-imp I top y   u   tur-a that  acc take-want ‘You take this one, and I will take that one.’

I argue that we analyze yu as a simple accusative marker instead of as a kakari marker, because I have not been able to find any examples of yu marking the subject of a nominal phrase. Further work is required, because of the dearth of good examples. I have only found one example (36) where yu appears to be used as a focus marker and not an accusative: (36) kyu: ya har-anu, bana: acca yu har-a I.top tomorrow foc go-vol today top go-neg ‘I won’t go today; I will go tomorrow.’ (lit. ‘It is tomor­ row when I will go’)

14.5.3  The predicate Sentences are negated by attaching a negative suffix to a verb or verbal adjective. Consider the examples in (37). (37) a.      

Hirara sïdain du kak-aN-fu nari-u’ï gradually foc write-neg-av become-prog ‘I’m gradually losing my ability to write.’ (lit. ‘I am becoming so that I do not write.’) b. Ishigaki   bana:  kanzi   kak-anu   I.top  kanji.Ø write-neg   ‘I don’t write kanji.’

237

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley c.      

Yonaguni nuNdi  kag-anu   Nga why   write-neg  q ‘Why don’t you write?’

When a verb is placed before a noun or noun phrase, the verb modifies the constituent it precedes. Consider example (38): (38) Ishigaki kuma  ga   dzï      kak-u    pïtu  loc character.Ø write-attr person  here   ur-an-u be-neg-concl  ‘There is nobody (person) here who writes (Chinese) characters.’ As noted above, the passive auxiliary does double duty, act­ ing as a passive as well as a potential (debitive). However, Ishigaki and Yonaguni also have a separate potential suffix: IG -ibusïŋ and -iʃʃiŋ, YG -icuN. The following examples are illustrative (39). (39) a. banu  ŋ   kakï-busï-ŋ   I   foc write-deb-concl   ‘I can also write it.’ b. wa:  ya  sï-ss-anu   You  top do-deb-neg   ‘You cannot do it.’

(41) uta iz-ïttasa:-ŋ song.Ø sing-want-concl ‘(I) want to sing a song.’ Yonaguni also has one auxiliary to show desire, related to the first auxiliary in Miyako: -ibusaN, see (42). (42) u du kat-ibusar-u this foc write-want-attr ‘I want to write this.’ The languages of Miyako and Ishigaki have a rich system of respect language, where age and social status require the speaker to elevate his or her speech. Because of spatial limi­ ta­tions, I will only describe the respect systems of the lan­ guages of Miyako and Ishigaki. Yonaguni’s respect system appears to be a simplified version of that found in Ishigaki. Miyako’s honorific system is unique in Sakishima as it preserves both elevated and humble forms. In Hirara respect is demonstrated through the use of a special verb, either a  suppletive verb, or a verb with a suffix added that then ­elevates the verb and by extension the sentence. Examples of re­placing a regular verb with an honorific verb appear in (43). (43) a. ʃinʃa: kui: yu Nkigi’-ï   teacher-top this acc eat (ex)-concl   ‘Teacher, will you taste (eat) this?’ Here Nkigi’ï replaces both fo: ‘eat’ and num ‘drink’. b. sïgu       ʃu:      nkai    immediately grandfather dir  ssai’-ï (< ssai’ï replaces a’ï ‘to say’) inform (ex)-concl   ‘Tell grandfather right away.’

c. siNsi ya kati-ca-nuN   teacher top write-deb-neg   ‘The teacher cannot write it.’

c. ʃinʃa:     nnama    teacher.top now  mmya:-ï (< mmya:ï replaces u’ï ‘to be’) be (ex)-concl   ‘Is the teacher home now?’ (lit. ‘Is the teacher there now?’)

As in mainland Japanese, there are two different auxiliaries to demonstrate a desire to do something in Miyako: -ïbussa and -ïtaka’ï, see examples in (40). (40) a. unnagi   nu  kutu:    umu:  cïka:     that time  gen thing.acc think when  no:ti:  nya:n  du  nak-ïbusï-ffu  na’-ï what  like   foc  cry-want-av become-concl   ‘When I remember (what happened) at that time, I want to cry, as it was like nothing else.’ b. kunu zï za kak-ïtaff-anya:n   this.gen character top write-want-neg   ‘I don’t want to write this character.’ Ishigaki has only one auxiliary to show desire, -ïttasa:ŋ in (41):

238

Related to example (43c), in conversation with the elderly on Miyako, the phrase for ‘Good morning’ is ukyu: ura:mma ‘Are you up yet?’8 Here ura:mma appears to be the verb u’ï ‘to be’ in the imperfect, where the liquid is preserved, plus an honorific suffix -amma’ï. Another way to create an honorific verb is to attach an honorific auxiliary (-isama’ï) to the stem of the preceding verb, which elevates the sentence. Examples include kak-isama’ï 8

  This example is taken from Nihon hōsō kyōkai (1972, 2: 187).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the souther n ryūkyūan l anguages ‘writing (what a superior is doing)’, and mac-isama’ï (‘Forgive me, as) I have kept you waiting.’ As noted above, the honor­ ific replacement for ‘eat’ is Nkigi’ï, but there is also another form, sïkiraisama’ï ‘to eat (highest form of respect)’,9 which appears to be based on the Okinawan verb ʃiki-in ‘to prepare food to cook’. Thus, this complex exalted form can be ­analyzed as (44): (44) *sukerar-e- sam-i-owarset (food before someone)-  ex-inf-ex-inf-ex10 The humble form in Miyako is primarily constructed around suppletive verbs, as the following example (45) demonstrates: (45) ʃu:n     kui:  yu grandfather  dir  this u’isi’-ï (< u’isi’ï replaces agi’ï ‘give’) acc   give (humb)-concl ‘I will give this to grandfather.’ In Ishigaki the verb o:ruŋ (from *owar-) is honorific, and replaces the verbs ‘go’, ‘come’, and ‘be’. This verb can also be added to the infinitive of other verbs to elevate the sentence in relation to a superior. Two examples are given in (46): (46) a. ʃinʃi:    taŋga  o:ru-ŋ   teacher.Ø  only   go (ex)-concl   ‘Only the teacher will (respectfully) go.’ b. zin ya taka:niŋ ar-o:ru-ŋ   money top much have-ex-concl   ‘You (sir) have a lot of money.’

14.6 Lexicon The native vocabulary of Sakishima has a number of layers of loans mixed in, most originating owing to successive waves from the prestigious language of Shuri (Okinawan). The key is to tease out what is original and what is a loan. While there are many clear cognates with mainland Japanese, there are also many native words that do not have clear cognates with mainland Japanese. As Serafim (2003) pointed out, the ancestor of the language of the Ryūkyūs appears to have originated from somewhere in the north­ eastern region of Kyūshū. It is not my intention to support or reject this argument, but it is possible, however, that some 9   Nishioka (2010: 198–9) notes that his informant once said his mother mentioned that this verb was highly honorific. 10   Thanks to Leon Serafim for this analysis.

of the words particular to the islands are simply altered forms of mainland, Western Japanese words. The well-known word for ‘snake’, pabu, may simply be a voiced form of OJ papu ‘to slither’. Based on work by Serafim, I earlier argued that the word for ‘child’ in the Ryūkyūs preserves an earlier stage of the Proto-Japonic word *kura ‘child’, where -r loss resulted in kwo in Old Japanese, but in Ryūkyūan the tap sur­ vived long enough for the initial velar to lenite (*kura > fura), giving us Hirara ffa:, Ishigaki fa:, and Yonaguni ha (Bentley 2008b: 17). The native vocabulary of Sakishima also preserves inter­ esting words that likely either were replaced by other words from mainland Japanese, or are internal innovations in the Ryūkyūs. One example is the pair of nouns for the male and female. The proto-forms are *weke- ‘male’ and *wome‘female’. The term for female survives in classical Japanese as womina, but the male term does not appear in the ancient Japanese corpus. Using kinship terms, the following division was observed and recorded by Miyara (1930) in the sïka language of Ishigaki. I have also provided data from Miyako. Yonaguni does not appear to have made a distinction and thus is ignored (Table 14.13). The lexicon of the Sakishima islands is fairly diverse. Whereas Japanese is a mixture of native and Sino-Japanese words, in Sakishima we find a layer of native words overlaid with several successively newer layers of loans from Shuri or mainland Japanese (via Okinawa in many cases). Because the Kingdom of Ryūkyū was a tributary state of China, there are also a number of Chinese loans.11 Table 14.14 lists a num­ ber of loans. Lawrence (2012: 408) believes the word ‘fist’ ko:ʃa: ori­gin­ ated from Fuzhou k‘au213tsɔ53, but he acknowledges that this loan (and most others) came via Okinawa. Lawrence also Table 14.13  Kinship terms in Ishigaki and Miyako  

Ishigaki

Miyako

Term

Nobility Commoner Nobility Commoner

Father

ʃu:

aja

uja

ïza

Mother

appa

abu

pa:mma

anna

Grandfather

uʃumai

abuzi

ʃu:

annauja

Grandmother mmi

a:pa

You (sing.)

ura

unzu

mma vva

unzyu

11   A number of scholars have also proposed an Austronesian influence in the languages of the area. See Hudson (1999: 92–5, 101–2), and Hudson (2012).

239

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

John R. Bentley Table 14.14  Chinese loans Original

Miyako

Ishigaki

Yonaguni

Ch. ‘fist’

ko:sa

ko:ʃa:

---

Ch. ‘emperor’

pundai

fundai

huNdai

Ch. ‘earthen mortar’

naipa

daiba: / daipa:

daiba

J kō ‘incense’

ko:

ko:

ku:

J kōkō ‘filial piety’

ko:ku:

ko:, ko:ko:

ku:

J sinpai ‘anxiety’

ʃiwa

siNpai

siba

J meriken-ko ‘flour’

mirukuŋ-gu: mirikiŋ-ku: ---

J kuruma ‘car’

kuruma

kuruma

kurima

J bentō ‘lunch box’

binto:

binto:

binto:

J nenbutsu ‘pray to Buddha’

ninbucï

nimbuzï

niNbuci

Skt. jhāpita ­‘cremation’

dabï

dabï

dabi

Eng. ‘anchor’

---

anka:

aNka

240

notes that ‘earthen mortar’ appears to be a loan to Sakishima that did not come via Okinawa, as the word does not seem to exist in any northern Ryūkyūan language. The loans from mainland Japanese are quite apparent as sound changes typ­ical of these languages either have not been completed, or have not occurred. Notice that a word like ‘filial piety’ is ko:ko: according to Miyara from the 1920s, but today most of the elderly say just ko:. Needless to say, this word preserves the vowel and vowel length as found in mainland Japanese, koukou.

14.7 Conclusion This chapter has described the phonology, morphology, syntax, and lexicon of an important language group in the Ryūkyūs that has not received much interest outside of Japan. These languages are critical for our understanding of the language history of the entire region, as well as Japonic, especially since the diversity of the languages in Sakishima has preserved important phonological and morphosyntac­ tic data to help scholars unravel the centuries of changes that have created the languages as they exist today.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 15

Korean and the Korean dialects H o -M i n S oh n

15.1 Introduction

Vietnam (109,000), and Kazakhstan (108,000). Thus, Korean is a native or ethnic language of approximately 83 million people at present.

15.1.1 Speakers The Korean language or Korean, called han’guk-ŏ or han’gukmal in South Korea and chosŏn-ŏ or chosŏn-mal in North Korea, is the official, national, and native language of the Korean people living in North and South Korea. It is also one of the two official languages in the Yanbien Korean Autonomous Prefecture and Changbai Korean Autonomous County of China. It is also an ethnic language of the Korean diaspora scattered around the world. Korean is also spoken by many thousands of non-Koreans as a second or foreign language. The current population of South Korea is approximately 51 million and that of North Korea around 25 million. The total population of the Korean diaspora at present is estimated at approximately 7 million, the major countries with a large Korean population being China (2,584,000), the United States (2,239,000), Japan (856,000), Canada (224,000), Uzbekistan (186,000), Russia (167,000), Australia (164,000),

15.1.2  Writing systems The Korean alphabet, called han’gŭl (lit. ‘great script’), is the main writing system used to represent native and SinoKorean vocabulary, and loanwords, while Chinese characters are optionally used to represent only Sino-Korean words.

15.1.2.1  Chinese characters Up to the 19th century, when Western cultures began to permeate East Asia, China had long been the center of East Asian civilization. Chinese civilization was propagated to neighboring countries mainly through written Chinese. Thus, the Chinese script has long been an integral part of the writing systems of Koreans.

NORTH KOREA PYONGYANG SEOUL SOUTH KOREA

Figure 15.1  Korean peninsula

Ho-Min Sohn, Korean and the Korean dialects In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Ho-Min Sohn. DOI: 10.1093/oso/9780198804628.003.0016

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ho-min Sohn It is generally assumed that the pronunciation of Chinese characters used in the northern part of China during the Su and Tang dynasties, around the seventh and eighth cen­tur­ ies, constituted the basis of Sino-Korean (SK) sounds.

15.1.2.2  Ch’acha (借字) ‘loan-character writing’ Early scholars devised the ch’acha writing system (i.e. idu, hyangch’al, and kugyŏl) using Chinese characters or their abbreviations for the transcription of native Korean affixes and words. The ch’acha scripts were used to record Korean expressions by means of (a) Chinese characters borrowed in  their Chinese meaning but read as the corresponding Korean sounds (glossograms), or (b) Chinese characters ­borrowed in their Chinese sounds only (phonograms). For instance, in 折叱可kesk-sk-e ‘(I) shall pick (a flower) and’ (Samguk yusa, 1285), the underlined element is a glossogram indicating the meaning of the verb ‘pluck, break’ and the rest are phonograms representing the epenthetic -sk and the conjunctive suffix (-e ‘and’). However, ch’acha scripts were extremely inadequate for written communication and difficult to learn. It was under these circumstances that han’gŭl was created, a phonetic writing system that was completely disengaged from the Chinese script.

15.1.2.3  The Korean alphabet, han’gŭl This indigenous phonetic alphabet is one of the most remarkable writing systems ever devised. King Sejong, the fourth king of the Chosŏn dynasty, promulgated han’gŭl to the public on October 9, 1446 in the name of Hunmin jŏngŭm ‘The Correct Sounds to Educate People’ (written by Sejong himself). This document was accompanied by Hunmin jŏngŭm haerye ‘Explanations and Examples of the Hunmin jŏngŭm’, compiled by a group of scholars commissioned by Sejong. The core of the former is the device of twentyeight han’gŭl letters (17 consonants and 11 vowels and diphthongs): ㄱ k, ㅋ kh,ｏng, ㄷ t, ㅌ th, ㄴ n, ㅂ p, ㅍ ph, ㅁ m, ㅈ c, ㅊ ch, ㅅ s, ㆆɂ, ㅎ h, ㅇɦ, ㄹ l, ㅿ z, • ɔ, ㅡ u, ㅣ i, ㅗ o, ㅏ a, ㅜ wu, ㅓ e, ㅛ yo, ㅑ ya, ㅠ yu, and ㅕ ye. These han’gŭl letters are spelled in syllable blocks, as in 한글 han-kul. Hunmin jŏngŭm haerye explains and illustrates the contents of Hunmin jŏngŭm in detail. Here is the gist. (a) The seventeen consonant letters were designed to depict the shapes of the speech organs. The symbol ㄱ k depicts the shape of the root of the tongue blocking the throat; ㄴ n, the shape of the tongue touching the alveo-dental area; ㅁ m, the shape of the mouth (i.e. lips); ㅅ s, the shape of an incisor; and ㅇɦ, the shape of the throat. Other consonants were made by adding strokes to the above basic symbols.

242

(b) Eleven vowel and diphthong letters make up the syllable-medial sounds: three cardinal • ɔ, ㅡ u, and ㅣ i; four derived ㅗ o, ㅏ a, ㅜ wu, and ㅓ e; and four yodized ㅛ yo, ㅑ ya, ㅠ yu, and ㅕ ye. For •, the tongue retracts and its sound is deep; the roundness of its shape is a depiction of Heaven. For ㅡ, the tongue retracts a little, and its sound is neither deep nor shallow; the flatness of its shape is a depiction of Earth. For ㅣ, the tongue does not retract, and its sound is shallow; the uprightness of its shape is a depiction of Man. Thus, the cardinal vowel letters symbolize the trinity of Heaven, Earth, and Man. The other letters are combinations of these cardinal letters. (c) In ㅗ o, ㅏ a, ㅑ ya, and ㅛ yo, the round dot (Heaven), i.e. a short stroke in the contemporary alphabet, is located above and on the outside, hence they are called yang ‘bright’ sounds. In ㅜ u, ㅓ e, ㅠ yu, and ㅕ ye, the round dot is located below and on the inside, hence they are called yin ‘dark’ sounds. Since 1446, han’gŭl letters have undergone modifications, due partly to the sound changes that have since taken place and partly to the research conducted by many han’gŭl ­scholars. Thus, the consonant letters ㅿ z, ᅘ hh, and ㆆɂ have been abolished. The consonant letter ｏ, which formerly represented the nasal sound ng, is no longer in use. Instead, ㅇ, which has lost the original voiced glottal friction, takes dual functions, assuming null sound quality in the syllable-initial position and representing ng in the syllablefinal position, as in 앙 ang. The vowel sound that the basic letter • represented has disappeared from the language, except in Chejuan. This letter is now used only in combination with other letter(s) to represent other vowel or diphthong sounds. The letter has been changed to a short vertical or horizontal stroke, as in ㅏ a and ㅗ o. When han’gŭl was created, the letters were combined into blocks consisting of spoken (phonetic) syllables. The current practice, since Korean linguists’ enactment of the Unified Spelling System in 1933, is binding letters into meaning-based (morpho-phonemic) syllables. For instance, the previous spelling 기픈 ki.phun ‘deep’ has been changed to 깊은 kiph.un to represent the meanings of the adjective 깊 kiph- ‘be deep’ and the relativizer은 -un. As a result, the present system consists of 24 letters that include 10 vowels and diphthongs and 14 consonants, as in (1). (1)  a.  Vowel and diphthong letters:  ㅏ a, ㅑ ya, ㅓ e, ㅕye, ㅗ o, ㅛ yo, ㅜ wu, ㅠ yu, ㅡ u, ㅣ i

b.  Consonant letters: ㄱ k, ㄴ n, ㄷ t, ㄹ l, ㅁ m, ㅂ p, ㅅ s, ㅇ ng, ㅈ c, ㅊ ch, ㅋ kh, ㅌ th, ㅍ ph, ㅎ h

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

kor ean and the kor ean dialects 15.1.2.4 Romanization Three Romanization systems are commonly used to transcribe Korean expressions: (a) the South Korean government system, (b) the McCune-Reischauer system, and (c) the Yale system. (a) is the standard system used in South Korea; (b) is widely used by the non-Korean public and scholars in all fields other than linguistics; and (c) is used by linguists. In this chapter, the Yale system is followed for cited linguistic forms (with han’gŭl spellings as  needed), while the McCune Reischauer system is employed for non-linguistic terms. Individualized or conventionalized spellings are followed for personal names and place names.

15.1.3  Previous scholarship on Korean Innumerable studies on structural aspects of Korean are available in Korean and English. Grammar books began to be published around the beginning of the 20th century. Classical ones include Gale (1894), Chu (1910), Ch’oe (1926), and Ramstedt (1939). Selected recent ones include H.  Kim (1972); M. Kim (1983); Huh (1984); Nam and Ko (1985); H. Lee (1989); Martin (1992); Kwon (1992); Seo (1996); Sohn (1994, 1999); I.  Lee and Ramsey (2000); Song (2005); Ko and Koo (2010); Tranter (2012); Yeon and Brown (2011); and Brown and Yeon (2015). Selected works on historical linguistics include Hong (1966); Martin (1966); Huh (1975); B. Kim (1984); Whitman (1985); Lyu and Kim (2005); Ko (2007); D.  Kim (2007); Vovin (2010); K. Lee and Ramsey (2011); and P. Nam (2009, 2012). Works on Korean dialects include H.  Kim (1982); Hyun (1985); Chon and Choy (1989); King (2006); and Yeon (2012). This chapter is based essentially on my own earlier studies, notably Sohn (1999), while extensively referring to other relevant works.

15.2  Historical connections Many scholars have attempted to compare Korean with neighboring languages in search for possible genetic af­fili­ ation. There were some efforts to relate Korean to the Dravidian Tamil language (e.g. Hulbert 1905; Clippinger 1984; Kang 1990) in view of the extensive typological similarities and numerous lexical lookalikes. This hypothesis is regarded as controversial by mainstream historical linguists, however. The major languages Korean has been compared with are Japanese and Altaic languages.

15.2.1  Korean-Japanese hypothesis Since the 18th century, despite the skepticism expressed by  some scholars, numerous scholars have proposed that Korean and Japanese are each other’s closest sister language (for a brief survey, see Sohn  1999: 29–36). In particular, Martin (1966) compares 320 sets of word pairs, reconstructing their hypothetical original forms and establishing phono­logic­al correspondence rules. Although the research is criticized as too mechanical and unrealistic, Martin’s work is a linguistic milestone for subsequent research such as Whitman (1985, 2012). K.  Lee and Ramsey (2011: 29–30) judge that at most 200 lexical pairs in the two languages look convincing and no more than fifteen possible comparisons are found in the sets of inflectional morphemes. Vovin (2010), on the other hand, claims that the majority of the cognates established by Martin and Whitman have turned out to be borrowings, mainly from Korean to Japanese, viewing the Korean–Japanese relationship as a contact relationship, rather than of a common origin. These opposing arguments among experts suggest that the Korean–Japanese ‘close relationship’ hypothesis is still far from being proven either positively or negatively.

15.2.2  The Altaic hypothesis Ramstedt (1928, 1949) is known as the first to assign Korean to the Altaic family in a systematic way. K. Lee and Ramsey (2011: 15–26) are in support of the Korean–Altaic hy­poth­ esis, displaying vowel and consonant correspondences between Middle Korean and Poppe’s (1960b) Proto-Altaic and claiming that earlier Korean had a vowel harmony system similar to the Proto-Altaic front-back harmony system. They present morphological similarities as well, claiming that the three Middle Korean verbal endings, i.e. the nominalizer -m, future modifier ending -lq, and past modifier ending -n, were almost identical to the Proto-Altaic endings -m, -r, and -n. K. Lee and Ramsey (2011: 23–6) indicate that the Altaic language closest to Korean is Manchu, a southern Tungusic language, and claim that at least 250 basic Manchu lexical items correspond to Korean words. Robbeets (2005, 2015, 2017e) supports the Japanese–Korean–Transeurasian hypothesis based on shared morphological innovations. Some historical linguists, including Vovin (2009b), object to the proposition of a genetic relationship of Japanese and Korean to Altaic languages. Further rigorous investigation, refinement, and verification of the hypothesis are called for in order to resolve the issue. Although typological traits are not to be included in establishing a genetic grouping of languages, the shared

243

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ho-min Sohn grammatical patterns and functions between Korean and Japanese on the one hand and between Korean and the Altaic on the other are too extensive and striking to be coincidental. The fact that Korean has hardly borrowed any morphosyntactic features from Chinese despite more than two thousand years of close contact suggests that the shared typological traits among Korean, Japanese, and the Altaic are not contact-induced but of a common origin.

15.2.3  Dialectal situation and standardization Although there are distinct dialectal zones, speakers from different zones have little difficulty in communication. The only exception is the critically endangered speech of the Cheju Island. This speech has such a low comprehensibility to the speakers of the other dialects that it has recently been proposed that it be given language status: Chejuan. Then, Korean and Chejuan may constitute a linguistic subgroup, Koreanic. When the unified spelling system was legislated by the Korean Language Association in 1933, the Seoul speech spoken by educated middle-class citizens was defined as the “standard language.” After the post-1945 division, however, two standard languages emerged. In South Korea, the notion of the “standard language” has been followed in all walks of life. North Korea has originated and used the so-called “cultured language” based on the speech of their capital, P’yongyang. Despite dialectal variations, the structural essentials discussed in this chapter are shared by all dialects, including Chejuan, warranting Koreanic as a subgroup as against the other Transeurasian subgroups. Chejuan is on the verge of extinction because only those born before 1950 can understand and speak this language today. Even the dialectal ­differences among mainland Koreans tend to gradually diminish owing to language standardization. Recently, language revitalization efforts have been launched to revive Chejuan (e.g. C. Yang, O’Grady, and S. Yang 2017).

15.2.4  Language contact Korean historians agree that it was in the period of Three Ancient Kingdoms—Koguryŏ (37 bc–ad 668), Paekche (18 bc –ad 660), and Silla (57 bc–ad 935)—that the first sophisticated states took shape on the Korean peninsula. The Silla kingdom unified the three kingdoms in 668, although a greater portion of the Koguryŏ territory was lost, including the whole Manchurian region. Our information about the Three Kingdoms period comes from Korea’s oldest extant history, titled Samguk sagi ‘Chronicle of the Three Kingdoms’, compiled by Kim Pusik, a Koryŏ dynasty historian, in 1145. A  glimpse of the available Silla language fragments in Samguk sagi and elsewhere leads us to regard Contemporary Korean as the descendant of the language of Unified Silla.

244

Through the historic period, Korean vocabulary has been enormously enriched, owing mainly to constant language contact with other countries. Contacts with Chinese, Japanese, and English have produced the most powerful and extensive influences on Korean, especially on the Korean vocabulary.

15.2.4.1  Contact with Chinese The oldest and most powerful contact is with Chinese, as Korea has spent its entire history next to China, a giant of world culture. Korean borrowings from China are mostly written Chinese, including Buddhist texts, Confucian classics, historical and literary works, and various cultural and institutional terms. Practical knowledge of Chinese in Korea is assumed to date back to 194 bc, when Wiman, from Yen in China, founded a primitive Korean state in the northwestern part of the Korean peninsula. According to Samguk sagi, all three ancient dynasties used written Chinese. The use of Sino-Korean (SK) words in Korea gained more popularity after Silla’s unification was achieved with Tang China’s military support. Accordingly, native place names were Sinicized as two Chinese-character words and all official titles were made Sino-Korean words. The number of Sino-Korean words began to overtake that of native words in the Koryŏ dynasty (918–1392) and Sino-Korean words pervaded the spoken language. The Chosŏn dynasty (1392–1910) saw the allout infiltration of Sino-Korean words into every facet of Korean culture and society, chiefly because of the dynasty’s adoption of Confucianism as a national ideology.

15.2.4.2  Contact with Japanese Although Korea has been in contact with Japan since the Three Kingdoms period, the major linguistic impact of Japanese on Korean was made during the colonial period (1910–45). First, all Korean students and the educated public became KoreanJapanese bilinguals. Second, Japanese words encroached extensively on the Korean lexicon as loanwords. Upon Korea’s liberation from Japan in 1945, the nationwide movement to “eradicate Japanese remnants” put most native Japanese words into disuse. Yet, quite a few Japanese words are still used in colloquial speech, such as kwutwu ‘leather shoes’, kapang ‘bag’, ippai ‘fully’, yudori ‘flexibility’, and muteyppo ‘carelessness’. Third, Korea has imported thousands of Sino-Japanese (SJ) words that the Japanese coined by Sinicizing Western words and concepts, as in ene-hak 언어학 (SJ gengo-gaku) ‘linguistics’ and hyen-kum 현금 (SJ gen-kin) ‘cash’. The influx of Sino-Japanese words is still in progress. Fourth, many native Japanese words written in Chinese characters as glossograms came into Korea as disguised Sino-Japanese loans. While such glossograms are pronounced as native words in Japanese, they are pronounced like any other Sino-Korean words in Korea, as in ip-kwu 입구 ( J. iri-kuchi) ‘entrance’, kyen-sup 견습 ( J. mi-narai) ‘internship’,

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

kor ean and the kor ean dialects and hal-in 할인 ( J.  wari-biki) ‘discount’. Such disguised Japanese loans are exempt from the post-1945 expulsion.

15.2.4.3  Contact with English and other Western languages Since the beginning of the 20th century, Western words have been coming into Korean as loanwords. Until 1945, loanwords were introduced almost exclusively through Japanese. Since 1945, South Koreans have been in direct contact with many foreign languages, notably with English. As a result, Korean has accommodated thousands of English words as loanwords for both need- and prestige-based motivation. North Korea, in a powerful language purification process, has banned, as much as possible, the use of Western loanwords, including English.

15.2.4.4  Structural impact Extensive language contact has made some structural impact on Korean. Middle Korean lexical tones are due to the influx of tone-carrying Chinese words. A massive influx of l-initial loanwords (e.g. latio 라디오 ‘radio’, lopi 로비 ‘lobby’) has disrupted the sound pattern in which no word allows a word-initial li­quid. Sino-Korean words are partly responsible for the disruption of the vowel harmony in many native words. Sino-Korean compounds reflect Chinese verb-object order in transitivity, as in sal-kyun 살균 (kill-germs) ‘sterilization’. Chinese and other foreign predicates are introduced into Korean as verbal or adjectival nouns and never as predicates. To function as predicates, they must be compounded with a native predicate (mostly with ha- ‘do, be’), as in SK poksa-hata 복사하다 and

loan khophi-hata 코피하다 ‘to copy’, SK chongmyeng-hata 총명하다 and loan sumathu-hata 스마트하다 ‘be smart’. Many Sino-Korean roots function as derivational affixes (e.g. negative prefixes pi-, mu-, mi-, mol-). The most widely used SK suffix, -cek ‘-tic’ came from SJ -teki, as in SK mincwucek 민주적 (SJ minshu-teki) ‘democratic’, which was created in Japan as a phonogram in view of its phonetic similarity to English ‘-tic’. Many Sino-Korean roots are also used as function words, as in caki 자기 ‘oneself, you’, tangsin 당신 ‘you, him/herself’, amu ( mn

m + l > mn r + n > rd s + l > st

Karachay-Balkar käs-im-nän ‹кесимден› 〈self-poss1sg-abl〉 ‘from myself’ Kumyk salam-na 〈straw-loc〉 ‘in the straw’ Noghay ädem-nen 〈man-abl〉 ‘from the man’ Tatar uram-nar 〈street-pl〉 ‘streets’ Noghay ädem-ner 〈man-pl〉 ‘men’.

Noghay ḳar-dị̈ŋ 〈snow-gen〉 ‘of the snow’ Kazakh tas-tar 〈stone-pl〉 ‘stones’

Table 23.10  Examples of total progressive assimilation l + d > lː n + d > nː

n + l > nː

Noghay tas-tïŋ̣ 〈stone-gen〉 ‘of the stone’

m + n > mː

t + m > tp

Kazakh batpan ‘unit of weight’ < batman

ḳ + ġ > ḳː

z + l > zd

Noghay ḳïz-dị̈ŋ 〈girl-gen〉 ‘of the girl’

s + n > st s + m > sp

z + n > zd

Kazakh kes-pe- 〈cut-neg〉 ‘not to cut’ Kazakh ḳị̈z-dar 〈girl-pl〉 ‘girls’

23.3.1.5  Syllable structure

Kumyk awul-laš ‘fellow villager’ ← awul ‘village’ and ‑daš

Karachay-Balkar kälinːän ⟨bride.abl〉 ‘from the bride’ Tatar urmanːan ⟨forest.abl〉 ‘from the forest’

Crimean Tatar čobanːar 〈shepherd.pl〉 ‘shepherds’ Noghay zamanːar 〈time.pl〉 ‘times’ Tatar tun-nar 〈furcoat.pl〉 ‘fur coats’

Karachay-Balkar ana-mːï 〈mother-poss1sg.gen〉 ‘of my mother’

Karachay-Balkar čïḳːan 〈go out.an〉 ‘having gone out’

Table 23.11  Examples of partial regressive assimilation n + b > mb

Karachay-Balkar om‑bir ‹онбир› Tatar um-bịr ‹унбер› ‘eleven’ Kazakh sön-be- 〈extinguish-neg〉 ‘not to extinguish’

Turkic syllables contain a vowel with usually one preceding and one following consonant. Vowel hiatus and con­ son­ant clusters are avoided, particularly word-initially. The restrictions are less strong in high-copying varieties, owing to the abundance of lexical items of non-Turkic origin, e.g. Lithuanian Karaim Troχ from Troki, the Polish name of the town of Trakai. Sequences of vowels are avoided.

['jɑlbɑrmɑ] ‘to pray’, ḱöǵu̇ t ́ ['kjøgjʏtj] ‘grass’, and in some verbs including a causative suffix, e.g. śeḱịrmä ['sjekjirmjæ] ‘to jump’, ‘to dance’. Many adverbs display an irregular accentuation pattern, e.g. Lithuanian Karaim ɣaĺe ['γɑːlje] ‘now’, ančeχ ['ɑnčɛχ] ‘only’.

23.3.1.6  Word accent

23.3.2  Word structure

The rules of word accent vary considerably. Syllables are accentable, i.e. capable of carrying high pitch, or nonaccentable. A high-pitch accent falls on the last accentable syllable of native words. High-pitch accent interacts in complex ways with patterns of dynamic stress. The distribution of accentable and non-accentable syllables is conservative also in high-copying languages such as Karaim, and as a rule not influenced by contact. However, the phonetic realization of the accent in Karaim is different from that in most other Turkic languages, because dynamic stress and high pitch always fall on the same syllable. When a suffix is non-accentable, as for example the neg­ ation suffix {‑MA}, the preceding syllable is accentuated, e.g. Lithuanian Karaim anla-ma-dị̈-m [ɑn'lɑmɑdɯm] ⟨understandneg-di.pst-1sg⟩ ‘I didn’t understand’. Compounds are accentuated on their first element, e.g. in the Lithuanian Karaim ́ ̣ń ['bjʏgjʏnj] ‘today’, which consists of bu ‘this’ compound büǵu̇ and ḱüń ‘day’. This pattern is also valid in compounds that are no longer analyzable, e.g. Lithuanian Karaim yalbarma

Turkic word structure is characterized by the Transeurasian core feature CS 9 = suffixing inflectional morphology (Robbeets 2017h and this volume: Chapter 10). Turkic morphology is, moreover, agglutinative, in the sense that morpheme boundaries are usually transparent and that there is typ­ ical­ly a one-to-one correspondence between morphemes and their functions. This juxtaposing technique makes word forms easily segmentable and the semantics of morpheme combinations highly predictable. There are very few cases of fusion, i.e. merging of grammatical functions into portmanteau morphemes. The regularity is enhanced by a strong tendency of bound morphemes to show phono­lo­gic­ al­ly predictable allomorphs. They are relatively unchangeable and their realizations mostly depend on harmonic and other phonotactic properties according to a limited number of assimilation rules. Another characteristic is a high degree of synthesis, a parameter based on the number of morphemes per word.

377

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson Word classes include nouns, verbs, adjectives, adverbs, pronouns, determiners, numerals, copulas, conjunctions, postpositions, interjections, and particles. Grammatical gender is lacking, and no traces of Proto-Turkic gender distinctions are found. A few languages have introduced feminine lexical forms and use them obligatorily, e.g. Lithuanian Karaim dost ‘friend’ vs. dost-ča ‘girlfriend’. Lithuanian Karaim uses copied non-Turkic derivational suffixes to mark feminine lexical forms, e.g. ɣaver ‘friend’ ́ ‘the master of the house’ vs. ɣaver-ka ‘female friend’, yubiy ́ vs. yubiy-čä́ ‘housewife’. Gender is sometimes marked with copied adjectival morphology, e.g. Lithuanian Karaim Ol e-dị́ inteĺịgent-na ⟨X e.cop-di.pst3sg intelligent-F⟩ ‘She was intelligent’. Definite articles do not occur. Indefinite articles are formally identical with the numeral ‘one’, e.g. Noghay bir ‘one, a/an’. In Karaim Bible translations, the personal and demonstrative pronoun ol ‘he, she, it, that’ renders the Hebrew definite article. Specific-nonspecific distinctions are grammaticalized in most Turkic languages. Turkic languages display numerous bound morphemes serving word formation and grammatical marking, nonclitic and clitic suffixes, and bound particles. Prefixes do not normally occur, the few exceptions being copied from contact languages, e.g. the Lithuanian Karaim superlative Table 23.12  Case suffixes in Kazakh Nominative

-Ø

Accusative

‑nị, ‑nï,̣ ‑nụ̈ ‑nụ, ‑dị, ‑dï,̣ ‑dụ̈ , ‑dụ, ‑tị, ‑tï,̣ ‑tụ̈ , ‑tụ

Genitive

‑nịŋ, ‑nïŋ̣ , ‑nụ̈ ŋ, ‑nụŋ, ‑dịŋ, ‑dïŋ̣ , ‑dụ̈ ŋ, -dụŋ, -tịŋ, -tïŋ̣ , -tụ̈ ŋ, -tụŋ

Dative

-ɣa, -ge, -ḳa, -ke

Locative

-da, -de, -ta, -te

Ablative

-dan, -den, -tan, -ten

prefix, copied from Slavic, in expressions such as nay-yaχšị̈-raχ ⟨super-good-comp⟩ ‘best’. There are rich possibilities for forming morphologically complex words. Nominal stems and verb stems are sharply distinguished with denominal and deverbal suffixes forming separate classes. For many grammatical notions, a synthetic expression is the only available option, e.g. Kazakh passive tab-ị̈l- ⟨find-pass⟩ ‘to be found’ from tap- ‘to find’, causative ụmụt-ḳị̈z ~ ụmụt-tị̈r- ‘to make forget’ from ụmụt‘to forget’. For more information about word formation in Lithuanian Karaim see Csató (2016). Turkic languages prefer non-verbal strategies for copying verbs. Copied nominal forms of verbs are combined with light verbs such as et- and ḳïl- ‘to do’, e.g. Armeno‑Kipchak skazat et- ‘to tell to say’ from Russian, Karachay-Balkar razï et- ‘to satisfy’, Bashkir χịδmät it- ‘to work’, and Kazakh ụ̈ mịt ḳïl- or ụ̈ mịt yet- ‘to hope’ from Arabic-Persian. Partial emphatic reduplication of nominal property words is productive in Turkic and occurs with a limited set of adjectives mostly describing form, color, and other visible qualities, e.g. Tatar ˈap-aḳ ‘pure white’ (with accentuation of the first element). The absence of obligatory numeral classifiers is shared by all Northwestern Turkic languages.

23.3.2.1  Nominal categories: Number, possessive, and case The nominal paradigm in Turkic includes singular and p ­ lural forms, case suffixes, and possessive suffixes. The ­plural suffix is {+LAr}, e.g. Kazakh bala-lar ‘children’, kịm-der ‘who (plural)’, at-tar ‘horses’. Compare to Lithuanian Karaim bala-lar, ḱiḿ-ĺär, at-lar. The category of case in Turkic typically includes six cases: nominative, accusative, genitive, dative, locative, and ablative. Table 23.12 shows the Kazakh case suffixes. The distribution of the variants is governed by the morphophonological rules of sound harmony and consonant assimilation.

Table 23.13  Number and case in Lithuanian Karaim  

Singular ‘horse’

Plural ‘horses’

Singular ‘dog’

Plural ‘dogs’

Nominative

at

at-lar

it ́

it-́ ĺär

Accusative

at-nï ̣

at-lar-nï ̣

it-́ ńị

it-́ ĺäŕ-ńị

Genitive

at-nïṇ

at-lar-nïṇ

it-́ ńịń

it-́ ĺäŕ-ńịń

Dative

at-χa

at-lar-ɣa

it-́ ḱä

Locative

at-ta

at-lar-da

it-́ tä́

it-́ ĺäŕ-ǵä it-́ ĺäŕ-dä́

Ablative

at-tan

at-lar-dan

Instrumental

at-ba

at-lar-ba

it-́ tä́ ń it-́ bä́

378

́ it-́ ĺäŕ-däń it-́ ĺäŕ-bä

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages In Lithuanian Karaim, an instrumental case in {+BA}, the suffixed form of the postposition bïla ‘with’, has become grammaticalized through copying the combinational properties of the Slavic instrumental. For instance, the instrumental case is assigned to the predicative element in Ol vaχt-ta e-d́ị üŕät ú ̣̇ v́č ú ̣̇ -b ä́ ⟨that time-loc e.cop-di.pst3sg teacher-ins⟩ ‘At that time X was a teacher’. The instrumental marker is not accentable. The Lithuanian Karaim nominal paradigm is presented in Table 23.13. Table  23.14 shows the possessive suffixes in Kazakh. The Kazakh suffixes are compared to the Karaim ones in Table 23.15.  In constructions of similarity, the similative suffix {+D2A2y} marks the standard in Kazakh, e.g. bụlbụl-day ‘like a nightingale’. Lithuanian Karaim uses the postposition ḱibịk in similative constructions, e.g. arslan ḱibịk ‘like a lion’. Suffixes of the type {+(Ị)rAːK} ~ {+rAK} are used as comparative markers in Middle Kipchak yaχšï-raḳ ‘better’, yaman-raḳ ‘worse’, yïraḳ-raḳ ‘farther’, Lithuanian Karaim yaχšị̈-raχ ‘better’, Halich Karaim ḱiʦ́i-räk ‘smaller’, KarachayBalkar uzun-uraḳ, Tatar ụzụn-raḳ ‹озынрак› ‘longer’, Bashkir Table 23.14  Possessive suffixes in Kazakh  

Singular

Plural

1

-m, -ịm, -ị̈m

-mịz, -mị̈z, -mụ̈ z, -mụz, -ụ̈ mụ̈ z, -ụmụz, -ịmịz, -ị̈mị̈z,

2

-ŋ, -ịŋ, -ị̈ŋ

-ŋịz, -ŋị̈z, -ŋụ̈ z, -ŋụz, -ịŋịz, -ị̈ŋị̈z, -ụ̈ ŋụ̈ z, -ụŋụz

3

-ị, -ï,̣ -sị, -sị̈

-larị̈, -lerị, -darï,̣ -derị, -tarï,̣ -terị

Table 23.15  The possessive paradigm in Lithuanian Karaim  

Singular

Plural Singular

Plural

1

at-ị̈m ‘my horse’

́ at-ị̈mị̈z it-ịm ‘my dog’

it-́ ịḿịź

2

at-ị̈y

at-ị̈yị̈z

3

at-ị̈

it-́ ịy

it-́ ịyịź

it-́ ị

δur-ị̈raḳ ‘bigger’, küb-äräk ‘more’, ḳị̈śḳa-raḳ ‘shorter’, Noghay üyken-irek ‘bigger’. The superlative is expressed by a particle in Middle Kipchak äm burun ‘first of all’, äŋ artïḳ-raḳ ‘biggest’, ‘most’, Karachay-Balkar em ullu ‘biggest’, em alamat ‘prettiest’, Tatar iŋ aḳị̈l-lị̈ ‘cleverest’, Bashkir iŋ δur ‘biggest’, iŋ küp ‘most’, Noghay äŋ üykän ‘biggest’, Kirghiz eŋ ǰaḳšï ‘best’. In Lithuanian Karaim, the comparative suffix is added, e.g. eŋk yaχšị̈-raχ ⟨super good-comp⟩ ‘best’. The Turkic pattern of comparison, i.e. marking the standard NP with the ablative, is used in Kazakh Ȧdịl Mȧryȧm-nan aḳïldị̈ ⟨Adil-abl Maryam intelligent-comp⟩ ‘Adil is more intelligent than Maryam’, ́ yaχšị̈-raχ ⟨I-abl good-comp⟩ ‘betLithuanian Karaim ḿeń-däń ter than I’. Moreover, non-Turkic syntactic patterns based on a comparative junctor have been copied, e.g. Lithuanian Karaim yaχšị̈-raχ ńečị́ k ḿeń ⟨good-comp as I⟩ ‘better than I’.

23.3.2.2  Pronouns The pronominal system in Northwestern Turkic is not characterized by the presence of the mi-Ti opposition in first- vs. second-singular personal pronouns as in core Transeurasian (Robbeets 2017h: CS 11). The Kazakh pronouns are men ‘I’, sen ‘you’ from East Old Turkic bän ‘I’ and sän ‘you’. The formation of a secondary nasal oblique stem in personal pronouns is observed in pronouns of the third-person (Robbeets 2017h: CS 12). In Kazakh the oblique stem has an -n, e.g. ol ‘he/she/it’, but on-da ⟨he/she/it-loc⟩. The Kazakh and the Lithuanian Karaim personal pronouns are compared in Tables 23.16 and 23.17. The instrumental suffix is not accentuable but places the accent on the preceding syllable (marked by ˈ in Table 23.17). A two-way proximal-distal distinction is typical in demonstrative pronouns (Robbeets 2017h: CS6). Several languages possess expanded demonstratives, simple demonstratives combined with deictic particles, e.g. Middle Kipchak oš-bu, oš-ol, Lithuanian Karaim uš-pu, Galician Karaim us-pu, Kirghiz ušul, Tatar šušï ̣, Tatar and Bashkir šul ‹шул›, Bashkir ụšụ ‹ошо› ‘this here’. Emphatic combinations with deictic par­ ticles include Bashkir bị̈naw, anaw, Noghay munaw, anaw,

Table 23.16  Kazakh personal pronouns  

I

you

he, she, it

we

you (plural)

they

Nominative

men

sen

ol

bi̩z

si̩z

olar

Genitive

meni̩ŋ

seni̩ŋ

onị̈ŋ

bi̩z-di̩ŋ

si̩z-di̩ŋ

olar-dị̈ŋ

Accusative

meni̩

seni̩

onị̈

bi̩z-di̩

si̩z-di̩

olar-dị̈

Dative

maɣan

saɣan

oɣan

bi̩z-ge

si̩z-ge

olar-ɣa

Locative

men-de

sen-de

on-da

bi̩z-de

si̩z-de

olar-da

Ablative

men-den

sen-den

on-dan

bi̩z-den

si̩z-den

olar-dan

379

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson Table 23.17  Lithuanian Karaim personal pronouns  

I

you

he, she, it

we

you (plural)

they

Nominative

ḿeń

śeń

ol

śiź

alar

Genitive

ḿeńi̩m

śeńi̩ń

anïṇ

́ biź ́ biź-ńi̩ ń

śiź-ńi̩ń

alar-nïṇ

Accusative

ḿeńi̩

śeńi̩

anï ̣

śiź-ńi̩

alar-nï ̣

Dative

maya

saya

anar

Locative

śeń-dä́ ́ śeń-däń

an-da

śiź-ǵä śiź- dä́

alar-ɣa

ḿeń-dä́ ́ ḿeń-däń

́ śiź-däń

alar-dan

Ablative Instrumental

ḿeˈńi̩m-bä́

śe-ˈńi̩ń-bä́

an-dan aˈnïn-ba

Table 23.18  Reflexive pronouns in Kazakh and Karaim  

Kazakh

Karaim

‘myself ’

öz-ịm

öź-ụ̇ m

‘yourself ’

öz-ịŋ

öź-ụ̇ y

‘himself ’, ‘herself ’, ‘itself ’

öz-ị

öź-ụ̇

‘ourselves’

öz-ịmịz

öź-ụ̇ mụ̇ ź

‘yourselves’

öz-ịŋịz, öz-derịŋịz

öź-ụ̇ yụ̇ ź

‘themselves’

öz-derị

öź-ĺärị

 

⟨self-poss1sg⟩

⟨self-poss2sg⟩ ⟨self-poss3sg⟩ ⟨self-poss1pl⟩

⟨self-poss2pl⟩ ⟨self-poss3pl⟩

Kazakh mị̈naw, anaw, sonaw, Noghay mine-munaw ‘this here’, äne-anaw ‘that there’, and Kazakh mị̈na(y), ana(y). Some of the most frequent interrogative pronouns are Kazakh ne ‘what’, ḳalay ‘how’, kịm ‘who’, ḳašan ‘when’, ḳayda ‘where’, neše ‘how many’, and Lithuanian Karaim ḱim ‘who’, ńe ‘what’, ńečị́ k ‘how’, ńińdị́ ‘what (like)’, kaysï ̣ ‘which’, ḱeyŕe ‘whereto’, ńek ‘why’, ńeǵä ‘for what’, ńe üčú ̣̇ ń ‘for what’, kačan ‘when’, ńečä́ ‘how many’. Some indefinite pronouns are Lithuanian Karaim bar ‘all’, ɣar ‘each’, ńe-eś  ‘something’, ḱim-eś  ‘somebody’, kačań-eś  ‘sometimes’, ḱišị́ -dä́ ‘nobody’, ńeḿät ́ ‘nothing’, ɣ́eč ‘no’, öńǵä ‘other’. The reflexive pronouns are based on öź ‘self’ and possessive suffixes (Table 23.18). Examples of adjectival pronominal forms are Lithuanian Karaim aley ‘so’, buley ‘this way’. In her study on clusivity in Turkic, Nevskaya (2005a) states that an inclusive/exclusive distinction in first-person pronouns expressed by plural/collective markers is not typical of Turkic (compare Robbeets 2017h: CS 20).

23.3.2.3  Postpositions Northwestern Turkic languages, like other Turkic languages, use postpositions. Frequently used postpositions are Kazakh

380

́ biź-ńi̩ ́ biź-ǵä

́ ä́ biź-d ́ äń ́ biź-d

́ biź-ˈńi̩ ń-bä́

śiź-ˈńi̩ń-bä́

alar-da aˈlar-ba

ụ̈ šịn ‘for’ vs. Lithuanian Karaim üčú ̣̇ ń ‘for, about’, Kazakh ́ ‘as, like’ (governing the sịyaḳtï ̣ vs. Lithuanian Karaim ḱibịk nominative of nouns and the genitive of pronouns), Kazakh soŋ ‘after’ vs. Lithuanian Karaim son (governing the ablative), Kazakh ḳaray ‘according to’ vs. Lithuanian Karaim ḱöŕä (governing the dative). Secondary postpositions based on nouns denoting body parts, directions, and locations combined with a possessive and a case suffix include Lithuanian Karaim üśtú ̣̇ ńä ‘on’ or ́ ‘from the top of’, tǘ bu̇́ ̣ ńä ‘under’ or ‘to ‘onto’, üśtú ̣̇ ń-däń ́ ́ ⟨bottom-poss3-abl⟩ ‘from below’, aln-ị̈n-a ́ under’, tüb-ụ̇ ń-däń ⟨front-poss3-dat⟩ ‘in front of’ or ‘to the front of’, aln-ị̈n-dan ‘from the front of’, art-ïṇ -a ⟨back-poss3-dat⟩ ‘behind’ or ‘to behind’, art-ïṇ -dan ⟨back-poss3-abl⟩ ‘from behind’, ara-sïṇ -a ⟨interval-poss3-dat⟩ ‘between among’, ara-sïṇ -dan ⟨intervalposs3-abl⟩ ‘from between’, kat-ị̈n-a ‘by’, utru ‘opposite to’. The use of the dative case to denote both direction and location is an archaic feature.

23.3.2.4  Verbal categories: Viewpoint aspect, mood, modality, actionality Turkic has a wide variety of simple and compound forms expressing viewpoint aspect, mood, and tense. The verbal morph­ology comprises numerous categories expressing grammatical notions of actionality (Aktionsart), voice/ di­ath­esis (passive, middle-reflexive, causative, cooperativereciprocal), deontic modality (possibility, impossibility, necessity), epistemic modality, evidentiality (indirectivity), negation, viewpoint aspect (intraterminal, postterminal), mood (imperative), modality (voluntative, optative, hypothetical, prospective, etc.), tense (past), interrogation, and subject person-number agreement. For a description of the finite categories in Noghay see Karakoç (2005). 23.3.2.4.1 Intraterminals Intraterminals in Northwestern Turkic go back to constructions of the type Yaz-a tur-ur ⟨write-a.cvb stand.aux-aor.3sg⟩, literally ‘X stands writing’, in which the lexical verb is a

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages converb in {‑y/A} followed by the auxiliary ‘to stand’ in the aorist. The markers later occur in reduced shapes, e.g. Yaza-dï, Yaz-a-t,Yaz-a ⟨write-a.intra.prs-(cop3sg)⟩ ‘X writes’. Forms of the type {‑A-dỊ} are found in Karachay-Balkar Bar-a-dï-la ⟨go-a.intra.prs-cop-3pl⟩ ‘They go’, Noghay Bar-a-dị̈ ‘X goes’, Kele-dị́ ‘X comes’, Tur-a-dï ̣ ⟨stand-a.intra.prs-cop3sg⟩ ‘X stands’, Kazakh Ber-e-dị ⟨give-a.intra.prs-cop3sg⟩ ‘X gives’, Ịš-e-dị ⟨drink-a.intra.prs-cop3sg⟩ ‘X drinks’, Kel-e-dị ⟨come-a.intra. prs-cop3sg⟩ ‘X comes’, Ụš-a-dï ̣ ⟨fly-a.intra.prs-cop3sg⟩ ‘X flies’. Kirghiz and Altay exhibit {‑A-t}, e.g. Kirghiz Bar-a-t ⟨go-a. intra.prs-cop3sg⟩ ‘X comes/will come’, Kör-ä-t ⟨see-a.intra.prscop3sg⟩ ‘X sees/will see’. The auxiliary has often been reduced, particularly in the first- and second-persons; see the Lithuanian Karaim paradigm. Negation is marked with the suffix {‑MA}. An example of the Kazakh and the Lithuanian Karaim present-tense forms is given in Table 23.19. The negated Lithuanian Karaim forms are shown in Table 23.20. In Lithuanian Karaim, the suffix -t/-t ́ in the 3sg is an obli­ gatory marker of the third-person. The present-tense forms of sana- ‘to count’ are sane-y-m ⟨count-a.intra.prs-1sg⟩ ‘I count’, saney-s ⟨count-a.intra.prs-2sg⟩ ‘you count’, etc. Observe the morphophonological change of the stem vowel a > e preceding y. This change does not affect the backness of the syllable; thus the 3pl form is sane-y-d-lar ⟨count-a.intra.prs-cop-3pl⟩ ‘they count’. Kazakh intraterminal constructions are based on a converb form of the lexical verb and an inflected form of a postural or motion verb used as auxiliary verb, e.g. -p žatïṛ Table 23.19  The present tense in Kazakh and Lithuanian Karaim bar‘to go’

Kazakh Singular

1

bar-a-mị̈n ‘I go’

bar-a-mị̈z bar-a-m

2

bar-a-sị̈ŋ

bar-asị̈ŋdar

bar-a-s

bar-a-sïẓ

3

bar-a-dị̈

bar-a-dị̈

bar-a-t

bar-a-dlar

Plural

Lithuanian Karaim Singular Plural

Singular negated

23.3.2.4.2 Terminals Terminals formed by {‑DỊ} are mostly used as simple eventoriented past items representing events as anterior to the primary deictic orientation point. They do not exhibit any specific features (Table 23.22). The meaning can also be rendered as a perfect. For instance, Al-dị̈-m ⟨take-di.pst-1sg⟩ can be translated both as ‘I took’ and ‘I have taken’. Table 23.21  Lithuanian Karaim intraterminal in the past al‘to take’

Singular affirmative

Plural affirmative

1

́ al-ị̈r e-dị-k

2

́ al-ị̈r e-dị-m ⟨take-aor e.cop-di.pst-1sg⟩ ‘I was taking’, ‘I used to take’ ́ al-ị̈r e-dị-y

3

al-ị̈r e-dị́

bar-a-bị̈z

Table 23.20 Negated present-tense forms in Lithuanian Karaim  

⟨b.cvb lie.aux-aor⟩, -p wotị̈r ⟨b.cvb sit.aux-aor⟩, -p žụ̈ r ⟨b.cvb move. aux-aor⟩, -p tụr ⟨b.cvb stand.aux-aor⟩. An intraterminal past tense in Kazakh is, for instance, Ber-ịp žat-ị̈r ye-dị ⟨give-b.cvb lie.aux-aor e.cop-di.pst⟩ ‘X was giving’. The Lithuanian Karaim intraterminal past tense is formed with the aorist plus the past tense of the copula e- ‘to be’, see Table 23.21. Intraterminals may get habitual readings. Moreover, several languages possess special habitual intraterminals: Karachay-Balkar has a form consisting of the {b} converb, see Section 23.3.4.3, plus {+čAn}, e.g. Jˇaaz-ïw ̣ -čan-ma ⟨write-b. cvb-hab-1sg⟩ ‘I often write’. A frequently used Kirghiz habitual past is formed with {‑čỤ} + agreement markers, e.g. Kel-čụ̈ -bụ̈ z ⟨come-hab-1pl⟩ ‘We used to come’, Oḳuːčụ-mụn ⟨read-hab-1sg⟩ ‘I used to read’. The Lithuanian Karaim forms ́ in Table 23.21 can get a habitual reading, e.g. Al-ị̈r e-dị-m ⟨take-aor e.cop-di.pst-1sg⟩ ‘I used to take’.

́ al-ị̈r e-dị-yịź ́ al-ị̈r e-d-ĺär

Table 23.22  The Kazakh and Lithuanian Karaim past-tense paradigm in {‑DỊ}

Plural negated

ber‘to give’

Kazakh Singular

Plural

1

al-mị̈-m ⟨write-neg.a.intra.prs-1sg⟩ ‘I do not take’

al-mị̈-bị̈z

1

ber-dị-m ber-dị-k ‘I gave/ have given’

2

al-mị̈-s

al-mị̈-sị̈z

2

ber-dị-ŋ

3

al-mị̈-t

al-mị̈-d-lar

3

ber-dị

ber-dị-ŋịz

Lithuanian Karaim Singular Plural ́ ị-m ́ beŕ-d

́ ị-k ́ beŕ-d

́ ị-y ́ beŕ-d

́ ị-yịź ́ beŕ-d

́ ị́ beŕ-d

́ ị-ĺär ́ beŕ-d

381

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson 23.3.2.4.3  Postterminals. Evidentials Turkic postterminals do not envisage an event directly, but relate it to a later orientation point at which it is in some sense still valid. In this indirect view, the result may be more in focus than the event itself. This indirect way of pres­en­ta­tion has frequently led to evidential or indirective connotations. A typical Turkic category of evidentiality is indirectivity, e.g. Turkish {‑mỊş}. In Northwestern Turkic evidentiality is expressed by postterminal items or evidential copulas. Finite postterminals are mostly evidential, signaling awareness of revealed facts indicating that the narrated event is stated by reference to its reception by a conscious recipient. For Turkic evidential postterminals the source of information is (1) hearsay (report), (2) inference (logical conclusion), or (3) perception, interpretable as ‘evidently’, ‘obviously’, ‘reportedly’, ‘as it turns out’, ‘as I see/hear/smell’, etc. Northwestern Turkic postterminals in {‑GAn} and {‑(V) p-tỊ(r)} tend to get evidential readings, e.g. Bashkir Yaz-ɣan ⟨write-gan.post⟩ ‘X has evidently written’. They often vacillate between evidential and non-evidential meanings. Evidential particles are stable markers of evidentiality. The {‑GAn} postterminals in Tatar and Bashkir express results with evidential connotations: ‘evidently’, ‘as it turns out’, e.g. Bashkir Al-ɣan-dar ⟨take-gan.post-3pl⟩ ‘They have obviously taken it’, Ḳar yaw-ɣan ⟨snow snow-gan.post.3sg⟩ ‘It has snowed (as I see)’, Aḳsa-m-dï ̣ ụnụt-ḳan-mïṇ ⟨moneyposs1sg-acc forget-gan.post-1sg⟩ ‘I have forgotten my money (as I realize)’. Examples of postterminal paradigms are given in Tables 23.23 and 23.24. In West Kipchak {‑GAn} expresses past events of current relevance often on the basis of results or indirect evidence, e.g. Crimean Tatar Al-ɣan-mïẓ ⟨take-gan.post-1pl⟩ ‘We have taken it’, Kumyk Bar-ɣan-man ⟨go-gan.post-1sg⟩ ‘I have gone’, Karachay-Balkar Ayt-ɣan-sa ⟨say-gan.post-2sg⟩ ‘You have

Table 23.23  The Tatar postterminal inflection  

Singular

Plural

1

{-GAn-mỊn}

{-GAn-bỊz}

2

{-GAn-sỊŋ}

{-GAn-sỊz}

3

{-GAn}

{-GAn-nAr}

Table 23.24  The Bashkir postterminal inflection

said’, ǰaz-ɣan-ma ⟨write-gan.post-1sg⟩ ‘I have written’. The Karachay-Balkar postterminal paradigm is shown in Table 23.25. Evidential copular particles of the types är-kän ⟨e.cop-gan. post⟩ regularly render indirective meanings. They are ­temporally indifferent, referring to non-past and past events. The indirective copula particle är-kän combines with intraterminal forms, e.g. Kazakh Bịl-e-dị yeken ⟨know-intra-cop3sg evid.ptcl⟩ ‘X obviously knows/knew’, Žị̈la-p otị̈r ye-ken ⟨cry-b.cvb sit.aux-aor evid.ptcl⟩ ‘X is/was obviously crying’, Kel-e žat-ị̈r ye-ken ⟨come-e. cvb lie.aux-aor evid.ptcl⟩ ‘X is/was obviously coming’. Postterminals of the type {b} tur-ụr ⟨b.cvb stand.aux-aor⟩, e.g. Yaz-ị̈p tur-ụr ‘X is in the state of having written’ are peri­phras­ tic forms. These constructions may also render indirective meanings, e.g. Noghay Yaz-ị̈p-tị̈ ⟨write-b.cvb-cop3sg⟩ ‘X has apparently written’, Kazakh Kel-ịp-tị ⟨come-b.cvb-cop3sg⟩ ‘X has evidently come’, Sen özger-me-p-sịŋ ⟨you change-neg-b.cvb2sg⟩ ‘You have not changed (as I see)’ (perceptive use), Kirghiz Ber-be-p-tịr ⟨give-neg-b.cvb-cop3sg⟩ ‘X has apparently not given it’, Altay Bar-ị̈p-tụr ⟨go-b.cvb-cop3sg⟩ ‘X has apparently left’. A common characteristic of high-copying peripheral Turkic languages such as Karaim is that they lack this typ­ ical Turkic category of indirectivity (Csató 2000b). 23.3.2.4.4 Aorist The oldest intraterminal, the so-called aorist in {-(V)r}, has almost lost its original aspectual value and just represents some last residues of an originally indicative category. In modern languages the aorist expresses modal concepts of prospectivity, uncertainty, and wish, e.g. Noghay Ber-er ⟨give-aor3sg⟩ ‘X will give’, ‘X is inclined to give’, ‘X is likely to give’. In Lithuanian Karaim the aorist renders a non-modal prospective aspect (Table 23.26). 23.3.2.4.5  Imperative mood As in many other languages of the world, the imperative singular in Turkic languages is homonymous with the bare verb stem and thus seems to be ontogenetically prior to the indicative. The marker of the thematic base is {-Ø}, e.g. Tatar Uḳị̈! ‹Укы!› ‘Read!’, Ịšlä! ‹Эшлә!› ‘Work!’. The simplest im­pera­ tive plural marker is {‑(Ị)ŋ}, e.g. Kazakh Tụr-ụŋ! ‘Stand up!’, Kör-ụ̈ ŋ! ‘See!’, Oḳu-ŋ! ‘Read!’. In Lithuanian Karaim the plural Table 23.25  Karachay-Balkar postterminal paradigm bar‘to go’

Singular

Plural bar-ɣan-bị̈z

 

Singular

Plural

1

{-GAn-mỊn}

{-GAn-bỊδ}

2

{-GAn-hỊŋ}

{-GAn-hỊGỊδ}

2

bar-ɣan-ma ⟨go-gan.post-1sg⟩ bar-ɣan-sa

bar-ɣan-sïẓ

3

{-GAn}

{-GAn-dAr}

3

bar-ɣan-dï ̣

bar-ɣan-dï-̣ la

382

1

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages Table 23.26  The aorist paradigm in Lithuanian Karaim at- ‘to throw’

Affirmative singular

Affirmative plural

Negated singular

Negated plural

1

at-ar-m ⟨throw-aor-1sg⟩ ‘I will throw’

at-ar-bïẓ

at-ma-m

at-ma-bïẓ

2

at-ar-s

at-ar-sïẓ

at-ma-s

at-ma-sïẓ

3

at-ar

at-ar-lar

at-mas-t

at-mas-t-lar

Table 23.27  The imperative paradigm in Lithuanian Karaim

Table 23.28  Kazakh optative markers

Second-person singular

Second-person plural

 

Singular

Plural

Al! ‘Take!’

Al-γị̈n! ‘Take!’

Al-ị̈yị̈z! ‘Take!’

1

{-GAy-mỊn}

{-GAy-mỊz}

Ḱeĺ! ‘Come!’

Ḱeĺ-ǵịń! ‘Come!’

Ḱel-ịyịź! ‘Come!’

2

Oχu! ‘Read!’

Oχu-γụn! ‘Read!’

Oχu-yụz! ‘Read!’

{-GAy-sỊŋ} polite {-GAy-sỊz}

{-GAy-sỊŋ-dAr} polite {-GAy-sỊz-dAr}

At! ‘Throw!’

At-χï ̣n! ‘Throw!’

At- ïỵ ị̈z! ‘Throw!’

3

{-GAy}

Korχ! ‘Fear!’

Korχ-kụn! ‘Fear!’

Korχ-ụyụz! ‘Fear!’

forms are marked by {‑(Ị)yỊz}. In the singular, the suffix {‑GỊn} can be added in formal style (Table 23.27). An emphatic or softening element {‑GỊl} or{‑GỊn} can be added to imperative forms in other Turkic languages as well. In some languages, enclitic non-accentable particles of the type {-čỤ} or {-čỊ} may be added to imperatives to implore, entreat, beg, or encourage the addressee, thus softening the imperative request, e.g. Tatar Al-čị̈! ‘Please take it!’, Kirghiz Ḳoy-čụ! ‘Let it be!’, Kazakh Kel-šị! ‘Please come!’/‘come, I beg you!’, Noghay Bar-šï!̣ ‘Please go!’, Kel-šị! ‘Please come!’. 23.3.2.4.6  Voluntative modality Turkic voluntatives are used in non-factual sentences indicating that a given action is desirable. The voluntative paradigm includes first- and third-person forms. The marker {‑(A)yỊn} of the 1sg is found in many modern languages, e.g. Kirghiz Ḳara-yïṇ ! ⟨look-vol1sg⟩, Kör-öyün! ⟨see-vol1sg⟩ ‘Let me see!’. The markers {‑(y)AyỊ-m} and {‑(A)yỊ-m} occur in the western parts of the Turkic family, e.g. Lithuanian Karaim Al-ayị̈m! ⟨take-vol1sg⟩ ‘Let me take!’ The marker of the 1pl is {‑(A)yỊ-K} in almost all Kipchak languages, e.g. KarachayBalkar Ač-ayï-̣ ḳ! ⟨open-vol-1pl⟩ ‘Let us open!’, Crimean Tatar Al-ayï-ḳ! ⟨take-vol-1pl⟩ ‘Let us take!’, Lithuanian Karaim Barayị̈-χ! ⟨go-vol-1pl⟩ ‘Let us go!’, Bashkir Al-ayị̈-ḳ! ‹Aлaйык› ⟨take-vol-1pl⟩ ‘Let us take!’, Ịšlä-yị-k! ‹Эшләйик› ⟨work-vol1pl⟩ ‘Let us work!’, Kazakh Al-ayị̈-ḳ! ⟨take-vol-1pl⟩ ‘Let us take!’, Söyle-yị-k! ⟨speak-vol1pl⟩ ‘Let us speak!’. In the third-person Kipchak languages display the marker {‑sỤn}, e.g. Karachay-Balkar Oḳu-sụn! ⟨read-vol3sg⟩ ‘Let X read’, Lithuanian Karaim Bol-sụn! ⟨become/be-vol3sg⟩ ‘Let it be’, Kazakh Söyle-sịn! ⟨speak-vol3sg⟩ ‘Let X speak’. Bashkir

Table 23.29  The optative paradigm in Lithuanian Karaim bar‘to go’

Singular

Plural

1

bar-ɣey-m ‘may I go’

bar-ɣey-bị̈z

2

bar-ɣey-s

bar-ɣey-sïẓ

3

bar-ɣey, often bar-γey-t

bar-ɣey-lar, often bar-ɣey-d-lar

shows {‑hỊn}, e.g. Ịšlä-hịn! ‹Эшләhен› ⟨work-vol3sg⟩ ‘Let X work!’ In the plural the plural suffix is added, e.g. Tatar Al-sị̈n-nar! ⟨take-vol-3pl⟩ ‘Let them take!’, Bashkir Ịšlä-hịn-där ‹Эшләhендәр› ⟨work-vol-3pl⟩ ‘Let them work!’ Kirghiz and Kazakh use {‑sỊn} for both single- and multiple-subject referents, and Kirghiz can also form the plural {‑(Ị)š-sỊn}. 23.3.2.4.7  Optative modality The optative is an irrealis modality fulfilling a wide range of functions developed along various diachronic paths. It ba­sic­ ally conveys volitional meanings of will, wish, desire, hope, expectation, aspiration, purpose, incitement, request, inducement, invocation, promise, advice, etc. The optative is formed with suffixes of the type {‑GAy}. The Kazakh and Karaim optative inflections are shown in Tables 23.28 and 23.29.  The past tense is formed with the addition of the past ́ ⟨want-opt-e.cop. copula, e.g. Lithuanian Karaim ḱl'ä-ǵäy-dị-m di.pst-1sg⟩ ‘I would like to’.

23.3.2.4.8  Hypothetical modality The hypothetical modality is an imaginative modality presenting conceivable situations as seen by the mind’s eye in the sense of ‘Imagine it is the case!’, ‘What if it is the case?’

383

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson Table 23.30  The Kazakh hypothetical paradigm tap‘to find’

Singular

Plural

1

tap-sa-m ⟨find-hyp-1sg⟩ ‘if/when I find’

2

tap-sa-ŋ

tap-saŋ-dar

3

tap-sa

tap-sa

tap-sa-ḳ

Table 23.31  The Lithuanian Karaim hypothetical paradigm ayt‘to say’

Singular

Plural

1

ayt-sa-m ⟨say-hyp-1sg⟩ ‘if/when I say’

ayt-sa-χ

2

ayt-se-y

ayt-sa-yïẓ

3

ayt-sa

ayt-sa-lar

As such it can convey volitional meanings of desire, wish, and hope. It is an old volitional mood whose functions largely overlap those of the voluntative and the optative. A hypothetical (conditional) in {-SA} is inflected in the ­regular way. The Kazakh and Karaim hypothetical paradigms are shown in Tables 23.30 and 23.31.  23.3.2.4.9  Necessitative modality Turkic necessitative modality expresses the need for a given action. It imposes, proposes, compels, incites, or encourages the action, signaling that the subject referent is obliged, bound, due, destined, expected, or likely to perform it. The obligation may be compulsive (‘X has to/must’) or debitive or advisory (‘X ought to/should’). Several Kipchak languages use a marker of this kind, e.g. Tatar {‑AsỊ//‑ysỊ} as in Bar-asị̈-bị̈z ⟨go-nec-1pl⟩ ‘We must go’.

23.3.2.4.10  Postverbial constructions A- and B-converbs are used to form postverbial constructions together with grammaticalized auxiliaries, e.g. Kazakh y er-ịp ket- ⟨melt-b.cvb go.away.aux⟩ ‘to melt away’. Postverbials are actional modifiers that operate on various types of verbs and actional phrases, describing the manner of action in qualitative and quantitative terms more accurately than is possible with single lexical verbs. One function is to express properties that affect the internal phase structure of the actional phrase, i.e. to specify particular phases of its content. The constructions express transformativity, nontransformativity, durativity, iterativity, habituality, possible or attempted action, subjective vs. objective version, etc. Postverbial constructions are based on limited sets of auxiliary verbs with strongly generalized grammatical meanings. The lexical source verbs are motion verbs meaning ‘to go’, ‘to come’, ‘to go away’, ‘to go out’, ‘to proceed’, postural verbs

384

meaning ‘to stand’, ‘to remain’, ‘to sit’, ‘to lie’, phasal verbs such as ‘to begin’, ‘to finish’, and verbs denoting various ­activities such as ‘to do’, ‘to put’, ‘to place’, ‘to hold’, ‘to throw’, ‘to send’, ‘to give’, ‘to take’, ‘to see’. For details see Johanson (2000c), Csató and Johanson (2018). Most Northwestern Turkic languages have rich systems of postverbial constructions. Certain postverbial constructions are transformativizers operating on actionally ambiguous actional phrases to highlight a dynamic initial or final phase of the action, thus blocking non-transformative readings, e.g. ‘to write down’ rather than ‘to write’, ‘to sit down’ rather than ‘to sit’, ‘to catch sight’ rather than ‘to look’, ‘to catch fire’ or ‘to burn down’ rather than ‘to burn’. Connotations of thoroughnesss, resoluteness, unexpectedness, suddenness, or quickness are frequent side effects. Highlighting the sudden transgression of the initial limit of the action yields ingressive readings. Only a few examples can be given here: Karachay-Balkar ayt-ïḅ ḳoy‑ ⟨say-b.cvb put.aux⟩ ‘to blurt out’, Tatar ḳurḳị̈t-ị̈p ḳuy‑ ⟨frighten-b.cvb put.aux⟩ ‘to frighten’. Verbs meaning ‘to throw (away)’ are used in constructions of the type B-converb + taslawhich may express fast energetic resolute action but also casual careless performance without special concern or attention, e.g. Tatar ịšlä-p tašla‑ ⟨do-b.cvb throw.away.aux⟩ ‘to get done (efficiently)’, ịč‑ịp tašla- ⟨drink-b.cvb throw.away. aux⟩ ‘to drink fast’. Verbs meaning ‘to go (away)’ are used in similar constructions, mostly derived from intransitive verbs, sometimes with directional connotations, e.g. Tatar awị̈r‑ị̈p kit- ⟨hurt-b.cvb go.away.aux⟩ ‘to begin to hurt’, Kazakh w öl-ịp ket‑ ⟨die-b.cvb go.away.aux⟩ ‘to die/to pass away’. Verbs meaning ‘to take’ are used in postverbial constructions such as Tatar aŋla‑p al‑ ⟨understand-b.cvb take.aux⟩ ‘to realize’ rather than ‘to understand’. Other constructions are non-transformativizing, highlighting the non-dynamic phase of an action. They operate on transformative and actionally ambiguous actional phrases, turning them into non-transformatives. They block dynamic limit-oriented readings and specify a non-dynamic phase, e.g. ‘to be ill’ rather than ‘to fall ill’, ‘to look’ rather than ‘to catch sight’, ‘to eat’ rather than ‘to eat up’, or ‘to read’ rather than ‘to read and finish reading’ (Johanson 2000c). Actional phrases of all types are combinable with constructions based on B-converbs when interpreted serially, i.e. ‘to act repeatedly, continuously, frequently, several times, on several occasions, regularly, usually, habitually’, e.g. Kumyk oχu-p tur- ⟨read-b. cvb stand.aux⟩ ‘to read repeatedly, habitually’, Karachay tig-ịb tur- ⟨sew-b.cvb stand.aux⟩ ‘to sew repeatedly’. Constructions based on source verbs meaning ‘to sit down/to sit’ are used in similar ways, e.g. Bashkir uyla-p ultị̈r- ⟨consider-b.cvb sit.aux⟩ ‘to consider continuously’, Tatar tụr-ị̈p utị̈r- ⟨stand.up-b.cvb sit.aux⟩ ‘to stand up repeatedly’, Kirghiz ište-p otur- ⟨work-b.cvb sit.aux⟩ ‘to work continuously’. Many constructions are based on source verbs meaning ‘to move’, ‘to walk’, ‘to proceed’, e.g. Bashkir 〈b〉 ‑yụ̈ rụ̈ -, Noghay

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages 〈b〉 yür‑, Kirghiz 〈b〉 ǰür(ü)-, Kazakh 〈b〉 žụ̈r- have similar functions, e.g. Tatar uḳị̈-p yụ̈rị- ⟨read-b.cvb move.aux⟩ ‘to read’. Constructions with A-converbs + bär‑ ‘to give’ express continued or uninterrupted action in some languages, e.g. Kazakh tamaḳ že-y ber- ⟨food eat-a.cvb give.aux⟩ ‘to keep eating’. Postverbial constructions based on motion verbs meaning ‘to come’ and ‘to go away’ may express spatial orientation, specifying whether an action is directed toward a deictic center, e.g. toward the speaker or the addressee, or away from it. Some examples: Noghay uš-ị̈p kel‑ ⟨fly-b.cvb come.aux⟩ ‘to fly here’, Kazakh žụ̈r-ịp ket- ⟨move-b.cvb go.away.aux⟩ ‘to move away’. Constructions based on verbs meaning ‘to give’ and ‘to take’ may express so-called version, which indicates whether a given action is performed to the benefit or affliction (advantage or disadvantage) of the performer or some other entity. This is mostly a question of beneficence, i.e., to whose benefit or in whose interest the action is carried out: ‘to act for one’s own sake’ vs. ‘to act for the sake of someone else’, e.g. Tatar aŋlat-ị̈p bir- ⟨explain-b.cvb give.aux⟩ ‘to explain to somebody’. Potentiality, the physical or mental ability or inability to perform actions, is expressed by various postverbial constructions. In Northwestern Turkic, A-converb + al- is used, e.g. Karachay-Balkar bar-al- < bar-a al- ⟨go-(a.cvb) take.aux⟩ ‘to be able to go’, Tatar yaz-a al- ⟨write-a.cvb take.aux⟩ ‘to be able to write’. In Crimean Tatar and Bashkir, the auxiliary bil- ‘to know’ expresses ability, e.g. Crimean Tatar at-a bil‑ ⟨throw-a.cvb know.aux⟩ ‘to be able to throw’, Bashkir yüδ-ä bịl- ⟨swim-a.cvb know.aux⟩ ‘to be able to swim’. In Lithuanian Karaim, potential forms can be built synthetically with the postverbial suffix {‑(y)Al} or analytically by combining the potential form of the copula bol- with the infinitive in {-MA} of the lexical verb, e.g. Bol-al-a-m śöźĺä-ḿä karayče ‘I can speak Karaim’. The Lithuanian Karaim synthetic forms are shown in Table 23.32. In some peripheral Turkic languages, including Lithuanian Karaim, the number of postverbial constructions is reduced. There are a few lexicalized frequentatives formed with the auxiliary ‘to come’, e.g. ḱet-́ ḱäĺä- ⟨travel-freq⟩ ‘to travel’, ‘to drive around’ from ḱet-́ ‘to travel’, ‘to drive’. On the ambiguity of postverbial constructions see Csató et al. eds. 2019. Table 23.32  The potential paradigm in Lithuanian Karaim bar‘to go’

Singular

Plural

1

bar-al-a-m ⟨go-pot-a.intra.prs-1sg⟩ ‘I can go’

bar-al-a-bị̈z

2

bar-al-a-s

bar-al-a-sïẓ

3

bar-al-a-t

bar-al-a-dlar

23.3.3 Copula Turkic languages make use of various copular devices including copular verbs, subject person-number agreement suffixes, and copular particles. The copular verbs in Northwestern Turkic include residues of the old copula verb är- ‘to be’, e.g. ́ ⟨e.cop. the Lithuanian Karaim past-tense copula forms e-dị-m ́ di.pst-1sg⟩ ‘I was’, e-dị-y ⟨e.cop.di.pst-2sg⟩ ‘you were’, etc. Other copular forms are derived from bol- ‘to become/be’, e.g. bol-ɣan ⟨become/be-gan.post⟩ ‘X has been/was’. The copulas can be combined to form intraterminal past or habitual pasts such as bol-ɣan e-dị́ ⟨be-gan.post e.cop.di.pst-1sg⟩ ‘X was (being)’, ‘X used to be’. The copula devices marking agreement are non-past and non-modal, and they attach to nominal stems such as Kirghiz Men ḳïrɣïz-mïṇ ⟨I Kirghiz-1sg⟩ ‘I am a Kirghiz’. Agreement markers are often omitted in the third-person, e.g. Kazakh Dos‑ị̈m žaḳsï ̣ adam ⟨friend-poss1sg good man⟩ ‘My friend is a good man’. The auxiliary verb tur- ‘to stand’ is used in Lithuanian Karaim as a marker of the third-person, e.g. bar-a-t ⟨go-a. intra.pres-cop3sg⟩ ‘X goes’. Existential copulas are used in existential and possessive expressions, e.g. Lithuanian Karaim bar-(t) ⟨existing-cop3sg⟩ ‘is existing’ and its negation yoχ-tụr ⟨non-existing-cop3sg⟩ ‘is non-existing’ (Csató 2014). Old written Turkic languages display the negative copular verb form Är-mäz ⟨e.cop-neg.aor3sg⟩ ‘It is not’. This type is retained as a negative copular particle in many modern languages, e.g. Kazakh Men mụɣalịm ye-mes-pịn ⟨I teacher be-neg. aor-1sg⟩ ‘I am not a teacher’. Some Kipchak languages employ negative particles of the type tägül ‘not’, e.g. Crimean Tatar dägil, Karachay-Balkar tüːl, Kumyk tu̇ gu̇ l, Lithuanian Karaim ́ iĺ, Tatar and Bashkir tügịl. tǘ v́ụ̇ĺ, Halich Karaim div́ Copula particles are grammaticalized markers of evidentiality and serve different modal and discourse functions.

23.3.4  Nonfinite verb forms 23.3.4.1  Action nominals Action nominals display morphological properties typical of nouns. They take on possessive suffixes referring to the owner of the action, e.g. Kazakh kel-gen-ịm ⟨come-gan.post.an-poss1sg⟩ ‘my coming’, ‘that I come/came/have come’, Kirghiz kör-gönụ̈m ⟨see-gan.post.an-poss1sg⟩ ‘that I see/saw/have seen’, and take part in possessor–possessee constructions. They can take on plural suffixes referring to actions or kinds of action and case markers indicating their syntactic function. They can express various actional, aspectual, and modal contents. Infinitive markers are Kazakh {-Ụw}, e.g. wotị̈r-ụw ← wotị̈r‘to sit’, kel‑ụ̈ w ← kel- ‘to come’, Karachay-Balkar and Kumyk

385

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson {‑(Ị)w}, e.g. Karachay-Balkar ǰaz-ị̈w ‘to write’, Kumyk get-ịw ‹гетив› ‘to go’, Tatar al-ị̈w ‘to take’. Other Kipchak equivalents are formed with {‑mAḲ}, e.g. Kumyk oχu-maḳ ‘to read’. Lithuanian Karaim oχu-maχ is only used as a lexical noun meaning ‘reading’. The Karachay-Balkar infinitive suffix is {‑mAK-lỊG}, e.g. al-maḳ-lị̈ḳ ‘capturing’. The marker {‑mA} is used in Lithuanian Karaim oχu-ma ‘to read’, ḱeĺ-ḿä ‘to come’. Some forms convey purposive meanings, e.g. Kumyk gät-mä ‘in order to go’, which suggest that they originate in dative forms such as {‑mA-GA}. Action nominals in {‑(Ị)š} ~ {‑(y)Ịš} mostly express the manner of performance, e.g. Kirghiz ayda-š ‘way of driving’. A number of languages possess infinitives in {-(V)r-GA} consisting of aorist + dative, thus obviously of purposive ­origin. Tatar and Bashkir use {‑(Ị)r-GA}, e.g. Bashkir al-ị̈r-ɣa ‘to take’, ịšlä-r-gä ‘to work’. The negated forms are Tatar {‑mAs-KA}, Bashkir {‑mAθ-KA}. Karachay-Balkar employs {‑(A)r-GA}, e.g. töl-er-ge ‘to pay’, čab-ar-ɣa ‘to run’, tur-ụr-ɣa ‘to stand’, ḳara-r-ɣa ‘to look’. Action nominals in {‑GỤ} and {‑GỊ}̣ + possessive suffixes express desire, e.g. Bashkir Säy ịs-kị-m kil-ä ⟨tea drink-gu.anposs1sg come-prs⟩ ‘I want to drink tea’, Noghay Bar-ɣï-̣ m kele-dị ⟨go-gu.an-poss1sg come-prs-cop3sg⟩ ‘I would like to go’.

23.3.4.2  Participant nominals

Participant nominals refer to entities that act or undergo actions and can serve as attributes in nominal phrases. Headless participant nominals can take on inflectional suffixes, e.g. Karachay-Balkar ḳal-an-nar ⟨remain-gan.pn-pl⟩ ‘remnants’, ket-gen-ler-ịgịz ⟨go-gan.pn-pl-poss2pl⟩ ‘those of you who have gone away’. Participant nominals can refer to first arguments, non-first arguments, and circumstants (optional modi­fiers expressing time, place, manner). Many participant nominals are diathetically versatile, offering different readings according to their actancy structure. When derived from intransitive verbs they refer to first arguments. When derived from transitive verbs they refer to non-first arguments and circumstants, e.g. Kirghiz ayt-ḳan-ị̈m ⟨say-gan.pn-poss1sg⟩ ‘what ́ I have said’, Lithuanian Karaim beŕ-ǵäń ⟨give-gan.pn⟩ ‘the one ́ ⟨bring-gan.pn⟩ ‘the one which is brought’. which is given’, eĺtḱäń The reference to non-first arguments leads to “impersonal” readings, e.g. Bashkir uḳï-ɣan kitap ⟨read-gan.pn book⟩ ‘a/the book that one has read’, Lithuanian Karaim oburla-ɣan sanduχ ⟨enchant-gan.pn chest⟩ ‘the enchanted chest’. Participant nominals of this type have therefore often been mistaken for passives or referred to as “quasi-passives”.

23.3.4.3  Converbs The most frequently used converbs are the so-called A-converbs and B-converbs. In many Northwestern languages, A-converbs have two variants: a postconsonantal {-A} and a postvocalic {‑y},

386

e.g. Karachay-Balkar kül-ä ⟨laugh-a.cvb⟩, sana-y ⟨con­sider-a.cvb⟩, Lithuanian Karaim ayt-a ⟨say-a.cvb⟩, anle-y ⟨understand-a.cvb⟩. In modern languages, the use of A-converbs is restricted to certain complex verb forms, e.g. the intraterminal present in Lithuanian Karaim bar-a-m ⟨go-a.cvb-1sg⟩ ‘I go’ and postverbial constructions. The A-converbs are mostly confined to reduplications of identical or antonymic verbs expressing repeated or intensive action, e.g. Kumyk išlä-y išlä-y ⟨work-a. cvb work-a.cvb⟩ ‘permanently working’, Bashkir ḳara-y-ḳara-y ⟨look-a.cvb look-a.cvb⟩ ‘incessantly looking’. B-converbs include Karachay-Balkar ayt-ị̈b ⟨say-b.cvb⟩, išle-b ⟨work-b.cvb⟩, Kumyk oχu-p ⟨read-b.cvb⟩. Stem-final labial consonants may be dropped and therewith cause contraction in length of the vowel, e.g. Kirghiz taːp ⟨find-b.cvb⟩ of the verb tap- ‘to find’. The final ‑l of monosyllabic stems is sometimes lost, e.g. Kazakh bo-p from the verb root bol- ‘to become/be’. B-converbs may fulfill various and sometimes vague semantic functions. They can modify the content of a superordinate verb or combine with it to form a compound whose members maintain their lexical meanings, e.g. Bashkir ḳayt-ïp̣ kil- ⟨return-b.cvb come-⟩ ‘to come back’. The system of converbs is highly elaborated in the typical Northwestern languages. Karakoç (forthcoming) gives the following description of the frequently used converbs in Noghay: {-(Ị)p} has a vague semantic content, while {‑A} is mainly found in reduplicated forms, e.g. kele kele ‛by coming’. The negated form {-MAy} often conveys the meaning ‛without doing’. Some of the complex converbs are: {‑GAn-dA} ‛when X does/ did/ has done’, {-(A)yat-ḳan-da} ‛when X is/ was doing’, {‑(A)yatïr-γan-da} ‛when X is/was doing’, {‑MAG-A tur-γan-da} ‛when X is/was about to do’, {‑(A)yAK-tA} ‛when X is/was going to do’, {‑(A)yat-ï ̣p} ‛while X is/was doing’, {-A  tur-ï ̣p} ‛while X is/was doing’, {‑GAn-šA (deri)/(deyim)} ‛until X’, {‑GAn-lAy} ‛as soon as X’, {‑GAn-nAn) soŋ} ‛after X’, {-GAn-lỊ (beri)} ‛since X’, {-GAn sayïn} ‛each time when X’, {GAn-Ịn-nAn/{-GAn-nAn) sebep}, {-GAn üšin} ‛because of X’, {‑GAn-dAy} ‛as if X’, {-GAn-GA köre} ‛in view of’. The com­ bin­ation {‑(A)r . . . -MAs} conveys the meaning ‛as soon as X’.

23.3.5  Clause structure The typical Turkic syntactic features are similar to those of other Transeurasian languages. With respect to relational typology based on the expression of grammatical relations Turkic adheres to the nominative-accusative pattern (rather than the absolutive-ergative pattern). Thus in main clauses the subject stands in the nominative case and the specific direct object in the accusative case. The syntax is head-marking with markers showing relations between constituents placed on the head of the respective phrases. Turkic has a left-branching syntax with modifiers

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages and dependents preceding their heads, the so-called “rectumregens principle.” Lithuanian Karaim syntax is strongly accommodated to the dominating languages of the Circum-Baltic area and manifests some non-Turkic features that will be briefly presented here (Csató 2000a, 2002a).

23.3.5.1  Basic word order The unmarked order of clause constituents is subject + object + predicate (verbal or nominal). See example (1). (1) Kazakh Mụγalịm  bala-lar    ụ̣̈ čụ̈ n   aŋgịme-ler teacher  child-pl  for.postp  story-pl subject   adverbial       object ǰaz-a-dï.̣ write-a.intra.prs-3sg verbal predicate

‘The teacher writes stories for the children.’ Deviations, e.g. cases of “scrambling,” occur for discoursepragmatic and stylistic reasons. The constituent order is less variable in non-main clauses. Under contact influence, Galician and Lithuanian Karaim have developed an SVO basic word order; see (2). (2)  Lithuanian Karaim Uŋlu   ɣazːan   yaz-ar   e-dị́   yomaχ-lar great   hazzan   write-aor   cop-di.pst3sg   story-pl subject      verbal predicate       object ulan-lar  üčú ̇ ń. child-pl for.postp adverbial

‘The great hazzan used to write stories for the children.’ In some recent Lithuanian Karaim publications, verb-final sentences are preferred, a purely normative effort to Turkicize a word order that is alien to the spoken language.

23.3.5.2  Nominal phrase compounds In noun phrases determiners such as demonstratives, at­tribu­tive adjectives, and relative clauses precede the head noun. Possessive noun phrases exhibit both ­dependent- and head-marking (Robbeets 2017h: CS 14). Thus, attributes expressing possessors stand in the genitive, whereas their head, indicating the possessee, carries a possessive suffix, e.g. Kazakh bala-nị̈ŋ ana-sï ̣ ⟨child-gen mother-poss3sg⟩ ‘the child’s mother’. The main type of nominal compounds follows the pattern noun + noun + possessive suffix, e.g. Bashkir Ụ̈ fụ̈ ḳala-hï ̣ ⟨Ufa city-poss3⟩ ‘the city of Ufa’, baškụrt tịl-ị ⟨Bashkir language-poss3⟩ ‘Bashkir language’, Kazakh köz žas-ï ̣

⟨eye moistness-poss3sg⟩ ‘tear’, Kirghiz ḳïrɣïz til‑ị ⟨Kirghiz language-poss3⟩ ‘the Kirghiz language’, ǰekšembi kün-ụ̈ ⟨Sunday day-poss3⟩ ‘Sunday’. In Lithuanian Karaim genitive constructions follow both the ngen + n(poss) and the reverse n(poss) + ngen patterns, e.g. T́eńrị́ ́ biyĺịǵńịń ad-ï ̣ ⟨God-gen name-poss3sg⟩ or ad-ï ̣ T́eńrị-ńịń, biź-nịń ́ ́ ịḿịź ⟨we-gen country-poss1pl⟩ or biyĺịǵ-iḿiź biź-ńịń (Csató 2011a). The possessive suffix may be missing, e.g. b́iź-ńịń ́ ́ ́ biź-ńịń. This structure is no biyĺịk ⟨we-gen country⟩ and biyĺịk innovation in Karaim. It also occurs in other Turkic languages as a non-canonical possessive construction (Csató 2018). The genitive attribute may be preceded by another ́ ́ attribute of the head noun, e.g. uŋlu biź-ńịń biyĺịǵ-ịḿịź ⟨great we-gen country-poss1pl⟩ ‘our great country’.

23.3.5.3  Clause embedding Typical Turkic embedded clauses differ syntactically and morphologically from main clauses. They are based on nonfinite verb forms with units of the infinite verbal morph­ ology functioning as bound junctors. Complement clauses are based on action nominals referring to actions. Participant nominals refer to entities that act or undergo actions, and thus form relative clauses. Certain verbal nominals are used for both participant nominals and action nominals, e.g. Kazakh {‑GAn}. Adverbial clauses are based on converbs. Some typical Turkic complement, relative, and adverbial clauses are given in examples (3–5). (3)  Kazakh complement clause Men   saɣan   ụ̈ y-de   Ḳazaḳ-ša I  you.dat  home-loc  Kazakh söyle-ytịn-ịmịz-dị    ayt-ḳï-̣ m  speak-atin.an-poss1pl-acc tell-gu.an-poss1sg kel-e-dị. come.postv-a.intra.prs-3sg ‘I want to tell you that we speak Kazakh at home.’ (Abish 2016) (4)  Kazakh relative clause Men-ịŋ  ǰụwị̈rda  alpị̈s  ǰas-ḳa   kel-etịn I-gen   soon  sixty  year-dat  come-atin.pn dos-ị̈m    bar. friend-poss1sg existing ‘I have a friend who will soon be sixty years old.’ (Abish 2016) (5)  Kazakh adverbial clause Bịz  bịr bala  tụw-ị̈l-ɣan-da  ḳattï ̣ we  a child  give.birth-pass-gan.pn-loc  very  ḳụwan-a-mïẓ . glad-a.intra.prs-1pl ‘We are very glad when a child is born.’ (Abish 2016)

387

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson Lithuanian Karaim embedded clauses are similar to the Standard European type. They are right-branching structures introduced by free junctors and based on finite verb forms. See the following examples (6–8). (6)  Lithuanian Karaim complement clause Ḱĺä-y-m saya ayt-ma ḱi yu̇ v́-ďä want-a.intra.prs-1sg you.dat say-inf that home-loc śöźÍe-y-b́ịź  karayče. speak-a.intra.prs-1pl  Karaim ‘I want to tell you that we speak Karaim at home’. (7)  Lithuanian Karaim relative clause Bar-t kolega kaysï-̣ nị̈n té rk altmïṣ̌ existing-3sg friend which-gen soon sixty yïl-ị̈      bol-ụr. year-poss3sg become- aor3sg ‘I have a friend who will soon be sixty’. (8)  Lithuanian Karaim adverbial clause ́ ulan tuv-a-t Ńečí k bir astrï ̣ when a child is.born-a.intra.prs-3sg very ́ ́ biyäń-ä-b ịź. be.happy-a.intra.prs-1pl ‘We are very happy when a child is born’. Most Turkic languages are characterized by extensive use of converbs (Robbeets 2017h: CS 15). In Karaim the use of nonfinite verb forms is reduced.

23.3.5.4  Relative clauses Turkic relative clauses are based on participant nominals carrying markers such as Kazakh {‑GAn}. See example (9). (9) Kazakh w ụš-ḳan ḳụs fall-gan.pn bird ‘the bird that fell’ Participant nominals lack adjectival features. When used attributively, they display structures similar to identity compounds of the type dost-um Ali ⟨friend-poss1sg Ali⟩ ‘my friend Ali’. Kirghiz ǰaz-ɣan ḳat ⟨write-gan.pn letter⟩ ‘letter which is written’ thus consists of two juxtaposed elements ‘what is written’ + ‘letter’. Participant nominals and constructions can also be used without a head (see 10). (10) Kazakh kel-etịn-der come-atin.pn-pl ‘those who will come’ In typical Northwestern languages such as Kazakh the head of the relative construction can bear a possessive

388

suffix agreeing with the first argument of the participant nominal (11). (11) Kazakh men-ịŋ ḳorḳ-atị̈n mụɣaḷïm ̣ -ị̈m I-gen afraid-atin.pn teacher-poss1sg ‘the teacher whom I am afraid of ’ The semantic relation between the head noun and a participant nominal is underspecified. This typical Turkic feature has been changed in Karaim. It employs the Standard European type of relative clauses, based on finite verb forms and introduced by a relative pronoun specifying the semantic relation (Csató 2018). See examples (12–14). (12)  Lithuanian Karaim ́ ́ ị-m ́ bu ḱišị́ ḱiḿ-ǵä beŕńäk beŕ-d give-di.pst-1sg this person rel.pron-dat gift ‘this person to whom I gave a gift’ (13)  Lithuanian Karaim bu ḱišị́ ḱiḿ-ńị tǘ ńäǵụ̇ ń ḱöŕ-du̇́ ̣ -m this person rel.pron-acc yesterday see-di.pst-1sg ‘this person whom I saw yesterday’ (14)  Lithuanian Karaim bu ḱišị́ ḱiḿ-bä́ kïz-ị̈m this person rel.pron-with daughter-poss1sg bar-a-t walk-a.intra.prs-3sg ‘this person with whom my daughter is walking’

23.3.5.5  Predicative possession and comparison In Turkic predicative possession is typically rendered by existence clauses (Robbeets 2017h: CS 16), e.g. Kazakh Ahmet-tịŋ kịtab-ị̈ bar ⟨Ahmet-gen book-poss3sg existing⟩ ‘A. has a book/books’. In typical Turkic predicative comparisons, the ablative case is used (Robbeets 2017h: CS 17). Northwestern languages including Kazakh also have a comparative suffix. See example (15). (15) Kazakh Ȧdịl Mȧryȧm-nan kịšị-rek. np np-abl small-comp ‘Adil is smaller than Meryem.’ (Csató and Abish 2015) In comparisons of similarity, the similative suffix {‑DAy}, mentioned earlier, marks the noun phrase that functions as the standard (16). (16)  Kazakh comparison of similarity Ȧdịl bụlbụl-day sayra-y-dï.̣ Adil nightingale-sim sing-a.intra.prs-cop3sg ‘Adil sings like a nightingale.’

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages Some Northwestern languages have developed alternative strategies. Predications indicating possession are formed in Karaim as in other Turkic languages, e.g. Bar-t aχča-m ⟨existing-3sg money-poss1sg⟩ ‘I have money’. An alternative and equally frequent construction with the same meaning is Ḿeń-dä́ bar-t aχča ⟨I-loc existing-3sg money⟩ ‘I have money’.

23.5 Lexicon

23.3.5.6  Interrogation

Table 23.33  Verbs denoting basic activities

In Turkic, polar questions are typically marked by an interrogative particle. Karaim employs the interrogative particle {-MA} (mo, me, mä), though less frequently than most Turkic languages. The particle is often used in a non-Turkic way: it  follows the first word of the clause even when it questions the whole clause, e.g. B́iĺ-mịm ḿeń mä ḱibịt-́ ḱä bar-ị̈m ⟨know-neg.a.intra.1sg I q shop-dat go-aorist1sg⟩ ‘I don’t know whether I will go to the shop’. The Polish interrogative particle czy has been copied and is frequently used clauseinitially (Csató 1999).

23.4 Grammaticalization Robbeets (2017h) suggests three cases of core grammaticalization in Transeurasian languages. According to her, direct insubordination (CS 18) implies that the Transeurasian languages “display a recurrent tendency to reanalyze non-finite suffixes as finite ones without the omission of a specific matrix predicate,” a tendency called “direct in­sub­or­din­ ation.” She continues: “Comparative evidence indicates that these markers originated as deverbal noun suffixes marking a derivational process at the lexical level which were then extended to function as (ad)nominalizers in dependent clauses at the syntactic level and eventually—through a pragmatic role in discourse—were extended still further to mark finite forms in independent clauses.” The grammaticalization path from verb forms used nonfinitely, to forms used nonfinitely and finitely, and forms used only finitely is valid for all Turkic languages. In Northwestern Turkic, certain verbal markers have different functions according to their syntactic posi­ tions. Kazakh {‑GAn} is thus used as a marker of both finite and nonfinite verb forms; compare the finite use Ket-ken ‘X has evidently gone’ and the nonfinite use ket-ken ‘gone’. Grammaticalization from negative verb to verbal negator over a construction comprising an inflected negative auxiliary and an invariant lexical verb (Robbeets 2017h: CS 19) is plausible for Turkic. The preaccentuation pattern of the negative marker {-MA}, e.g. in Karaim ḱeĺ-ḿị-t ́ [ˈkjeljmjitj] ⟨come-neg-a.intra3sg⟩ ‘X does not come’ reveals its relatively recent development from an enclitic particle, maybe ori­gin­ al­ly a form of a negative verb (Johanson 1981: 12).

The native lexicon of the Northwestern languages exhibits significant correspondences in the basic vocabulary. Tables 23.33–23.39 show some cognate lexical items in the NWS language Kazakh and the NWW language Lithuanian Karaim in comparison with the SWS language Turkish.      

 

NWS Kazakh

NWW Lithuanian Karaim

SWS Turkish

‘to see’

kör-

ḱöŕ-

gör-

‘to hear’ / ‘to listen’

y

estị-

ešịt-́

išit-

‘to come’

kel-

ḱeĺ-

gel-

‘to lie’

žat-

yat-

yat-

‘to sit’

w

oltur-

otur-

‘to stand’ / ‘to stop’

tụr-

tur-

dur-

‘to eat’

že-

aša-

ye-

‘to drink’

ịš-

ič-́

ič-

‘to go’ / ‘to leave’

ket-

ḱet-́

git-

‘to take’

al-

al-

al-

‘to say’

söyle-

söylä-

‘to give’

ber-

śöźĺä́ beŕ-

‘to hold’

ụsta-, tụt-

tut-

tut-

‘to sleep’

ụyị̈ḳ-

yukla-

uyu-

‘to live’

tịrị-

tí ŕịĺ-

yaša-

‘to die’

w

öl-

öĺ-

öl-

‘to kill’

w

öltịr-

öĺtú ̣̇ ŕ-

öldür-

‘to suck’

y

em-

öḿ-

em-

‘to spit’

tụ̈ kịr-

tǘ ḱụ̇ ŕ-

tükür-

‘to laugh’

kụ̈ l-

ḱüĺ-

gül-

‘to fear’

ḳorḳ-

korχ-

ḳorḳ-

‘to hunt’

awla-

avla-

avla-

‘to cut’

kes-

ḱeś-

kes-

‘to split’

žar-

yar-

yar-

‘to dig’

ḳaz-

kaz-

ḳaz-

‘to swim’

žụ̈ z-

čom-

yüz-

otị̈r-

ver-

(continued )

389

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Éva Á. Csató and L ars Johanson Table 23.33  Continued

Table 23.35  Adjectives denoting colors

 

NWS Kazakh

NWW Lithuanian Karaim

SWS Turkish

‘to fly’

ụš-

uč-

‘to fall’

tụ̈ s-

‘to wash’ ‘to pull’

 

NWS Kazakh

NWW Lithuanian Karaim

SW Turkish

uč-

‘blue/green’

žasïḷ

yešị́ ĺ

yešil

tǘ š-́

düš-

‘yellow’

sarï ̣

sarï ̣

sarï

žụv-

yuv-

yïḳa-

‘white’

aḳ

aχ

aḳ

tart-

tart-, čé ḱ-

ček-

‘black’

ḳara

kara

ḳara

‘to sew’

tịk-

tí ḱ-

dik-

‘to play’

w

oyna-

oyna-

‘to burn’

žan-

yan-

yan-

‘to flow’

aḳ-

aχ-/aү-

aḳ-

‘to rain’

žaw-

yav-

yağ-

oyna-

Table 23.34  Adjectives denoting basic properties  

NW Kazakh S

NWW Lithuanian Karaim

SW Turkish S

‘big’

ụlị̈

uŋlu

ulu

‘long’

ụzị̈n

uzun

uzun

‘wide’

keŋ

ḱeń

geniš

‘thick’

ḳalïŋ

kalị̈n

ḳalïŋ

‘heavy’

awị̈r

avur

a(γ)ïr

‘small’

kịšị

ḱičị́

küčük

‘short’

ḳị̈sḳa

kïsχa

kïsa

‘narrow’

tar

tar

dar

‘fat’

semịz

śeḿịź

semiz

‘cold’

sụwị̈ḳ

suvuχ

so(γ)uḳ

‘warm’

ị̈stị̈ḳ

iśśị

sïǰaḳ

‘full’

tolị̈

tolu

dolu

‘new’

žaŋa

yanγï

yeni

‘old’

kart, kärị

kart

yašlï (ḳarï- prov. ‘to grow old’; ḳarï ‘woman, old woman’)

Table 23.36  Nouns denoting kinship  

NWS Kazakh

NWW Lithuanian Karaim

SW Turkish

‘father’

ata-baba

ata

ata, baba

‘mother’

ana

ana

ana

‘daughter girl’

ḳị̈z

Kïz

ḳïz

‘son’

ụl

Uvul

o(γ)ul

‘husband’

kụ̈ yew ( er ‘male, man’)

er

koca (er ‘male, man’)

y

Table 23.37  Nouns denoting body parts  

NWS Kazakh

NWW Lithuanian Karaim

SW Turkish

‘leg’

ayaḳ

Ayaχ

ayaḳ

‘mouth’

awị̈z

Avaz

a(γ)ïz

‘stomach’ ‘belly’

ḳarị̈n

karïnw kursaχ

ḳarïn

‘ear’

ḳụlaḳ

Kulaχ

ḳụlaḳ

‘eye’

köz

ḱöź

göz

‘nose’

mụrị̈n

Burun

burun

‘tooth’

tịs

T́iš ́

diš

‘tongue’

tịl

tí Í

dil

‘fingernail’

tï ̣rnaḳ

Tïrnaχ

tïrnaḳ

‘foot/leg’

ayaḳ

Ayaχ

ayaḳ diz

‘good’

žaksị̈

yaχšị̈

iyi

‘straight’

tụ̈ zụ̈ w

tǘ ź

düz

‘knee’

tịze

T́iź

‘dry’

ḳụrγaḳ

kuru

kuru

‘head’

bas

Baš

baš

‘right’

tụwra

duγru

do(γ)ru

‘hand/arm’

ḳol

Kol

kol

‘near’

žaḳị̈n

yuvuχ

yaḳïn

‘wing’

ḳanat

kanat

ḳanat

‘far’

ụzaḳ

uzaχ

uzaḳ

390

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the northwester n tur kic ( kipchak ) l anguages Table 23.37  Continued  

NWS Kazakh

NWW Lithuanian Karaim

SW Turkish

‘neck’

moyïn

boyụn

boyun

‘back’

arḳa

Arka

arḳa

‘heart’

žụ̈ rek

üŕäk ḱöńụ̈ ĺ

yürek gönül

‘liver’

bawị̈r

Bavur

ba(γ)ïr ‘bosom breast heart’ (ḳara) ǰi(γ)är ‘liver’

‘tail’

ḳụyrị̈ḳ

Kuyruχ

ḳuyruḳ

‘bone’

sụ̈ yek

śüv́äk

kemik

‘skin’

terị

té ŕi

deri

‘hair’

šȧš

čač

sač

Table 23.38  Nouns denoting environmental entities  

NWS Kazakh

NWW Lithuanian Karaim

SW Turkish

‘cloud’

bụlt

bulụt

bulut

‘moon’

ay

ay

ay

‘snow’

ḳar

kar

ḳar

‘star’

žụldị̈z

yulduz

yïldïz

‘sky’

kök, aspan

ḱöḱ

gök

Table 23.39  Nouns denoting periods of time  

NWS Kazakh 

NWW Lithuanian Karaim

SW Turkish 

‘day’

kụ̈ n

ḱüń

gün

‘year’

žị̈l

yïl

yïl

The West Kipchak vocabularies have been strongly influenced by Russian, some of them also by Caucasian languages. In Karachay and Balkar, old Alan and more recent Ossetic

influences can be traced. Many loans mirror close contacts with West Old Turkic. The multilingualism of small Turkic peoples such as the Karaim has led to extensive borrowing. Lexical copying has been necessary for communication with speakers of the dominant languages of the respective regions of residence, and it has compensated for insufficiencies of the inherited Turkic lexicon. Profound effects on the Karaim vocabulary have been exerted by Hebrew, Polish, Ukrainian, Russian, and Lithuanian. Many lexical items pertaining to modern life have been copied from Russian, Polish, and Lithuanian. While Hebrew words are a basic layer in the religious vocabulary, part of the religious terminology is also of Arabic-Persian origin. The names of the days in Kazakh are dụysenbị ‘Monday’, seysenbị ‘Tuesday’, sȧrsenbị ‘Wednesday’, beysenbị ‘Thursday’, žụma ‘Friday’, senbị ‘Saturday’, žeksenbị ‘Sunday’; in Lithuanian Karaim are yeχḱụ̇ ń ‘Sunday’, yeχbašḱụ̇ ń ‘Monday’, ortaḱụ̇ ń ‘Tuesday’, χanḱu̇ṇ́ ‘Wednesday’, ḱičị́ barasḱị ‘Thursday’, barasḱị ‘Friday’, šabbatḱụ̇ ń ‘Saturday’. The names of the months in Kazakh are ḳaŋtar ‘January’, aḳpan ‘February’, nawrị̈z ‘March’, sȧwịr ‘April’, mamïṛ ‘May’, mawsïm ̣ ‘June’, šịlde ‘July’, tamïẓ ‘August’, ḳị̈rkụ̈ yek ‘September’, ḳazan ‘October’, ḳaraša ‘November’, želtoḳsan ‘December’; in the Karaim lunar calendar the names are artarị̈χ-ay ‘March–April’, kural-ay ‘April–May’, baškusχan-ay ‘May–June’, yaz-ay ‘June–July’, ulaγ-ay ‘July–August’, čí ŕịk-ay ‘August–September’, ayrïχ̣ sị̈-ay ‘September–October’, ḱüź-ɑy ‘October–November’, soγụm-ay ‘November–December’, kïš-ay ‘December–January’, karakị̈š-ay ‘January–February’, śüv́ụ̇nč-ay ‘February–March’, and artị̈χ-ay ‘March–April (in leap years)’.

23.6 Notations In order to make phonological features more transparent, Turkic words and expressions are given here in three different notations: a Turcological one written in italics; an orthographic form between angle quotation marks ‹ ›; and in some cases a broad IPA transcription between square brackets [ ]. Morpheme segmentation is marked with a hyphen. Word accent is denoted by ' in front of the accented syllable. Brackets of the type ⟨ ⟩ are used for glosses. X stands for ‘he/she/it’ in translations of examples. Curly brackets of the type { } are used for morphophonemic transcriptions summarizing suffix variants.

391

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Ch a pter 24

Turkish and the Southwestern Turkic (Oghuz) languages J a k l i n K or n f i lt

24.1 Introduction Turkish is one of the Oghuz languages, which form the southwestern branch of the Turkic language family. Other languages belonging to that branch are as follows: (i) SWW, a western subbranch (West Oghuz), consisting of Turkish, Gagauz, and Azeri/Azerbaijani; (ii) SWS, a southern subbranch (South Oghuz), consisting of non-standard dialects of Southeast Anatolia, Southern Iran, Iraq, and Afghanistan; (iii) SWE, an eastern subbranch (East Oghuz), consisting of Turkmen and Khorasan Oghuz. (cf. Johanson 1998a: 82, and Johanson, this volume: Chapter 8) Turkish, with approximately 71 million speakers, is the language with the largest representation within this branch by far. Azerbaijani has approximately 26 million speakers, and Gagauz around 700,000 speakers. Turkish and Gagauz belong to the western subgroup of the Oghuz languages, while Turkmen, which has approximately 7.5 million s­peakers, belongs to the eastern branch. Turkish is the official language of the Turkish Republic. In addition to Turkey, it is spoken in Cyprus, where it is the official language of Northern Cyprus, the Turkish part of the island, and is a co-official language in the Greek Republic of Cyprus. Furthermore, the language is represented in Northwestern Europe through the many “guest workers,” as well as ­speakers in whitecollar professions, especially in Germany. (Currently there are about two generations of native Turkish speakers in Northwestern Europe who were born and raised there.) Azerbaijani is the official language of the Republic of Azerbaijan. While it is also spoken in Iranian Azerbaijan (and is referred to as “Azeri” in that region), it has no official status in Iran. It is also spoken in Dagestan, in Georgia, and in Turkey, as well as in diaspora communities in North America and Europe. Gagauz is the official language of the autonomous region of Gagauzia in Moldova; it is also spoken in Ukraine, Russia, and Turkey. Turkmen is the official

language of Turkmenistan, and it is also spoken in northeastern Iran and in northwestern Afghanistan. Turkish is currently written in a modified Latin script, which was adopted in the context of a series of reforms after the founding of the Turkish Republic. The details of the modifications will be pointed out in Section  24.3, when the consonant and vowel inventories will be discussed. Azerbaijani is written in a modified Latin script in the Republic of Azerbaijan, in Cyrillic script in Dagestan, and in the Persian adaptation of Arabic script in Iran. Gagauz is written in a modified Latin script, modeled after the modern Turkish orthography, with some additional symbols. The official writing system for Turkmen is a modified Latin alphabet; however, Cyrillic is still being used widely. The Oghuz languages, and especially those of the western subgroup, are, at least in part, mutually intelligible. This is true especially for Turkish and Azerbaijani, whose dialects are almost identical in adjacent geographical regions; as a reviewer points out, Turkish dialects in Eastern Turkey are closer to standard Azeri/Azerbaijani than to standard Turkish. In Section 24.5, on structural properties, only Turkish will be addressed, as the largest and best-studied representative of Oghuz, i.e. of the NW branch of Turkic languages. For detailed references to historical scholarship in Turkic, see the references in Johanson (1998a), as well as in Johanson’s article on the genealogical classification of Turkic (this volume: Chapter 8). A brief discussion of some relevant work follows. For the classification of the Turkic languages, useful references are the contributions in Deny et al. (1959), especially Menges (1959), and also Johanson (1998a); generally differing views are offered in Schönig (1997). The following works are useful for the historical background to Turkish: Karamanlıoğlu (1972), von Gabain (1963), and Mansuroğlu (1954). Of these, von Gabain is especially helpful with regard to the Oghuz languages. Lewis (1999) is an interesting and somewhat controversial account of the language reform in the early years of the Turkish Republic.

Jaklin Kornfilt, Turkish and the Southwestern Turkic (Oghuz) languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Jaklin Kornfilt. DOI: 10.1093/oso/9780198804628.003.0025

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages Lewis (2000) is a revised and enlarged edition of the author’s original grammar published in 1967. It is a comprehensive and detailed treatment of Turkish grammar with useful quotations from contemporary literature and the press. Göksel and Kerslake (2005) is more extensive, with a more pedagogical focus, and with attention to intonation; Underhill (1976) is a semi-pedagogical grammar, written in an informal early generative framework. Kornfilt (1997) is a reference grammar published in a series of descriptive grammars, under the general editorship of Bernard Comrie, that uses a questionnaire common to the entire series and is thus intended primarily for use by typological and the­or­et­ic­al linguists. Lees’ (1961) pioneering treatment of Turkish phonology is still referred to extensively. An early account of the pragmatic functions of Turkish word order is given in Erguvanlı (1984). Useful overview articles on basic aspects of Turkish, i.e. on its syntax, semantics, pragmatics, and phonology can be found in Boeschoten and Verhoeven (1991). Of those, Kornfilt (1991) offers discussion of generative studies on syntactic aspects of Turkish, in addition to a description of Turkish syntax. Underhill (1986) consists of brief capsules on studies in English on numerous aspects of Turkish. Johanson and Csató (1998) consists of useful overview art­icles on Turkic as a language family as well as on individual Turkic languages.

script was adopted in the course of the “writing reform” of 1928 (put into force in 1929), one of the various reforms introduced after the founding of the Turkish Republic with the aim of westernizing the country. However, the Uyghur script was also employed by the Anatolian Turks up to the 15th century, which might explain some features of the Arabic script as used by the Turks of that period which differ from standard Arabic usage, e.g. vowels are written out in Turkish words. The dialect of the earliest Anatolian texts has various features in common with the Oghuz dialect as documented for the 11th century, before the migration to Anatolia, and with Turkmen. Some of these are listed below:

24.2  Historical connections: genealogy and contact

(i) The future-tense suffix -ecek, -acaq appears for the first time in the 13th century (but is used as a participle marker and not yet as a finite verb suffix, as is also possible in Modern Turkish). (ii) The suffix -iser, -ïsar is the most widely used suffix for the future tense in Anatolian Turkish between the 13th and 15th centuries and is seen only very infrequently in some Turkmen and Qïpchaq works.

The question of the ancestor language of the Oghuz languages is not settled. The language of the oldest documents (i.e. the Orkhon inscriptions and the Old Uyghur manuscripts) is the ancestor of another group, namely of the Central Asiatic Turkic languages; the Oghuz languages are presumably descendants of the language of the “Western Türküt” mentioned in the Chinese Annals. The ancient languages of this group would be Old Anatolian and Old Ottoman, which are Middle Turkic written languages. The first Anatolian Turkish documents date from the 13th century and show that the literary tradition of Central Asia was only very tenuously carried over by the Turkish people (who had been converted to Islam earlier) after invading Anatolia from the east in the late 11th century. It is clear that these tribes were influenced heavily by both Persian and Arabic from the very beginnings of their settling down in Anatolia, given the higher prestige and development of the culture and literature of these neighboring Muslim nations. From the very beginning of its Anatolian period, Turkish was written in the Arabic script, until the Latin

(i) d for t in Old Turkic. (A number of these ds became devoiced again through assimilation in the 15th century.) (ii) Initial b changes to v: bar- → var- ‘to go; to arrive’; ber- → ver- ‘to give’ (iii) Suffix-initial γ, g disappears (in most instances; there are some surviving suffixes such as -gil, -gen, but those aren’t productive). (iv) Word-final γ, g disappears in polysyllabic words. (v) Instead of the 2pl imperative ending -ler, -lar in Old Turkic, the forms -nüz, -nuz, -niz, -nız are found (and remain today). Forms which are limited to Anatolian Turkish are the following:

However, the differences between Old Turkic and early Anatolian Turkish must not have been great and their phonologies are essentially identical (although some studies claim that Old Turkic had phonemically long vowels—a property which seems to have disappeared by the time of early Anatolian Turkish). The vocabularies are also similar to a large extent, although obviously many borrowings from Islamic sources are seen in the area of religious-mystical concepts. See also Johanson’s chapter on the genealogical classification of Turkic for detailed information on vo­cabu­ lary differences among branches of Turkic (this volume: Chapter 8). In the works of the 14th century and afterward, pe­cu­li­ar­ ities of Eastern Turkic, which had crept into Anatolian Turkish because of the Eastern origins of some authors,

393

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt disappeared almost completely, while the component of Arabic and Persian words and forms increased; such Eastern Turkic features include: initial m instead of b in words containing a nasal: men instead of ben ‘I’; min- instead of bin- ‘to ride’; initial b, which, as mentioned above, changed to v in Anatolia and neighboring areas, remained unchanged in Eastern Turkic and is also sometimes found in early Anatolian works: ber- instead of ver- ‘to give’; and, as another example for a different feature, bol instead of ol ‘to be’. In the literature written for scholarly, administrative, and literary purposes, the Persian and Arabic components became so prevalent that “Ottoman” became a mixed language, having lost some of its characteristic Turkic properties to the point of not being usable as a medium of communication common to all social classes. During the same period, however, there was also considerable production of mystical literature and folk poetry which was written for the less educated classes, in the language used by those segments of the population, namely Anatolian Turkish, as influenced very little by Persian and Arabic. These works are very close to the “Republican Turkish” of today and can essentially be understood by contemporary speakers without too much difficulty. Among the authors of the court literature there were also some who called for a purification of the language and, starting in the 18th century, there was a general movement toward a language with local (rather than foreign) features. This movement became particularly strong after the turn of the century. Systematic efforts to establish criteria for maintaining the vocabulary as well as the structural properties of Turkish were con­tinued into the founding of the Republic and the reform movements. The language reform, which started with writing reform, should therefore be viewed within a tradition of a search for a national identity, combined with a general movement toward westernization. A Turkish Language Academy (Türk Dil Kurumu: TDK) was founded in Ankara, with the tasks of etymological research and creation of new words, the latter in accordance with the Turkish rules of word formation and using Turkic roots, where the purification of the language from Arabic and Persian vocabulary had created gaps which could not be filled with current synonyms. The work of the Academy can be viewed as having been essentially successful in creating a widely understood language with a transparent morphological component and its own, typologically consistent syntax. As pointed out by Johanson in his chapter on the genealogical classification of Turkic (this volume: Chapter 8), Turkic dialectology has a long tradition, starting with Maḥmūd al-Kāšɣarī’s compendium Dīvān luɣāt at-Turk (1072–1094), which gives detailed accounts of varieties ­spoken by Turkic tribes in the 11th century; see Dankoff and Kelly (1982–5). However, dialects spoken within Turkey have been studied

394

only based on a recently evolving interest in them; the chapter on Turkish dialects by Brendemoen (1998), in Johanson and Csató’s The Turkic Languages, points out the dearth of genuine dialectological studies on Turkish dialects, while including some interesting information on some dialects in the region of the Black Sea, based on his own studies of those dialects. Fortunately, workshops and conferences about Turkish dialects have come into existence recently, so that one can be hopeful for important results. The presentations on mainly generative syntactic studies of some Turkish dialects at international conferences on Altaic (e.g. WAFL: Workshop on Altaic Formal Linguistics, and ICTL: International Conference on Turkish Linguistics) offer support for such hope.

24.3 Phonology 24.3.1  Consonants An inventory of the phonemic consonants in Turkish is given in Table 24.1. The consonants k, g, and l have two phonetic mani­fest­ ations: palatal and velar. The distribution of these forms is determined by the backness or frontness of the vowel in the same syllable; e.g. both the [k] and the [l] are velar in kol ‘arm’, while both the [k] and the [l] are palatal in kil ‘clay’. There are exceptions to this generalization in some borrowings; e.g. in kalp ‘heart’, while [k] is velar, the [l] is palatal. Where the palatal [k] precedes a back vowel, a palatal glide is inserted; e.g. kase ‘bowl’ is pronounced [kyase]. This glide is signaled in the orthography by placing a circumflex on the back vowel, e.g. kâse, although this orthographic convention is not standardized. The orthography does not distinguish between the velar and palatal versions of these three consonants, which are written just like their phonetic representations in Table 24.1. [č] is written as ç, [ ǰ] is written as c, [š] is written as ş, and [ž] is written as j. Otherwise, all of the consonant symbols in Table 24.1 are rendered in the same way in Turkish orthography.

24.3.2  Vowels An inventory of Turkish phonemic vowels is given in Table 24.2. All vowels of the native vocabulary are phonemically short. There is vocalic length on the surface, having two sources: (i) borrowings with unpredictably long vowels: e.g. ka:fiye ‘rhyme’; ara:zi: ‘(piece of) land’; this unpredictable length in vowels is not expressed in the orthography; (ii) compensatory

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages Table 24.1  Turkish consonants  

Bilabial

Labiodental

Dental, Alveolar

Palato-alveolar

Palatal

Velar

Glottal

  voiceless

p

 

t

č

 

k

 

 

b

 

d

ǰ

 

g

 

Stop, affricate voiced

Fricative  

voiceless

 

f

s

š

 

 

 

 

voiced

 

v

z

ž

 

 

 

m

 

n

 

 

 

 

 

 

1

 

 

 

 

liquid

 

 

r

 

 

 

 

Glides

 

 

 

 

y

 

h

Nasal Lateral liquid Non-lateral

Table 24.2  Turkish vowels [−back] [−round]

[+round]

[+back] [−round]

[+round]

[+high]

i

ü

ɨ

u

[−high]

e

ö

a

o

 

lengthening in words of Turkic origin (and in some loanwords, as well), where an original voiced velar fricative (which is no longer part of the surface inventory of segments in modern standard Turkish) used to follow a vowel and was in syllable-final position: e.g. dağ [da:] ‘mountain’, a word of Turkic origin, originally [daɣ], and bağ [ba:] ‘vineyard’, a word of Persian origin, originally [baɣ]. The voiced velar fricative is still found in many Turkish dialects, and it is represented in standard orthography with the symbol ğ, although it is not pronounced. The orthographic representations of the vowels in Table 24.2 are identical in Turkish orthography, with the exception of [ɨ], which is rendered as a dotless i; ı. Because of this, a capital i with a dot (İ), e.g. the first letter in the name of the city of İstanbul, represents a front, high, unrounded vowel, while a capital i without a dot (I), e.g. the first letter in the name of the city Isparta, represents a back, high, unrounded vowel. Vowel length is independent from stress: e.g. ara:zí ‘(piece of) land’, where the final syllable with its short vowel bears stress, rather than the long vowel; or ba:da ‘vineyard, locative’, where, once again, the final syllable with its short vowel bears stress, rather than the long vowel.

A characteristic property of Turkish vowels is that nonhigh vowels cannot be round, unless they are in a wordinitial syllable. While many borrowed stems are exceptional in this respect (e.g. futbol ‘soccer’), there is only one suffix that is exceptional: the progressive suffix -(I)yor. One other characteristic property of vowels, namely vowel harmony, will be discussed in Section 24.3.4.

24.3.3  Syllable structure The canonical shape of a Turkish syllable is (C)V(C). Turkish does not in general allow consonant clusters in the syllable onset, other than in learned, borrowed vocabulary, and even these limited instances have consonant clusters only in careful speech, despite exhibiting clusters in the orthography, e.g. Fransa ‘France’, stres ‘stress’. In casual speech, complex onsets (including items exhibiting word-initial consonant clusters as in the examples given) are broken up by an epenthesized high vowel, e.g. [fɨransa], [sɨtres] or [sitres]. In most, but not all, instances, the epenthesized vowel takes on the backness value of the first stem vowel, i.e. of the vowel following the epenthesized vowel; in add­ition, if that first stem vowel is round, the epenthesized vowel gets rounded, too: e.g. klüp [kulüp] ‘club’, where for most speakers, the epenthesized vowel does not harmonize with the stem vowel in backness, but does harmonize in rounding. Some word-initial consonant clusters in borrowed vo­cabu­ lary have been treated with the prothesis of a vowel preceding the cluster, and are usually written with that vowel; e.g. iskelet ‘skeleton’, ɨspanak (orthographically: ıspanak) ‘spinach’.

395

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt Turkish allows certain kinds of syllable-final consonant clusters. Three types of clusters are allowed as a coda: (i) sonorant+obstruent: kalp ‘heart’, kart ‘card’; (ii) voiceless fricative+oral stop: zift ‘pitch’; (iii) k+s: boks ‘boxing’. Where a stem has a consonant cluster in syllable-final position that does not fall under any of these three types, again a high vowel is epenthesized which often, but not always, undergoes harmony with the preceding vowel, e.g. ‘abdomen’ accusative karn-ɨ (orthographically: karnı), nominative karɨn (ortho­graph­ ic­al­ly: karın); ‘time’ accusative vakt-i, nominative vakit. (Note that in the first example, the epenthesized vowel harmonizes with the stem vowel, while in the second example, it does not.) A subcase of syllable-final consonant clusters are gemin­ ate consonants. While Turkish does tolerate geminate con­ son­ant sequences when their members are heterosyllabic (e.g. at-ta ‘horse-locative’), it does not allow them to occupy syllable-final position. Rather than breaking up such clusters by epenthesis, however, the language has a rule of degemination, e.g. ‘feeling’ accusative hiss-i, nominative his; ‘line’ accusative hatt-ɨ (orthographically: hattı), nominative hat. Additional phonological generalizations will be addressed in the next section.

24.3.4  Morphophonology Some phonological rules concerning consonants are briefly discussed here.

24.3.4.1  Syllable-final oral stop devoicing Turkish has a rule that devoices regular oral stops as well as affricates in syllable-final position, e.g. šarap (ortho­ graph­ic­al­ly: şarap) ‘wine’, accusative šarab-ɨ (ortho­graph­ic­ al­ly: şarabı), locative šarap-ta (orthographically: şarapta); pirinč (orthographically: pirinç) ‘rice’, accusative pirinc-i (orthographically: pirinci), locative pirinč-te (ortho­graph­ic­ al­ly: pirinçte).

24.3.4.2  The k/∅ alternation The final k of a polysyllabic word is deleted phonetically in intervocalic position, where the preceding vowel is short. This alternation is orthographically rendered as a k/ğ alternation, e.g. kütük ‘trunk’, accusative kütüü (ortho­graph­ic­al­ly: kütüğü). It is possible to view this phenomenon as a subcase of the voiced/voiceless alternation discussed above for oral stops and affricates. If it is assumed that the alternating ks are derived from underlying gs as a result of syl­lable-final stop devoicing, and if a rule of intervocalic fricativization is posited for the voiced velar stop, the data are essentially covered.

396

24.3.4.3  Word-final liquid devoicing Liquids devoice word-finally, a phenomenon especially common in the Istanbul dialect and in the speech of educated speakers in the other big cities: vaṛ ‘there is’, beḷ ‘waist’. It should be noted, however, that this is not a completely unified phenomenon; some speakers devoice only the pal­atal l, while other speakers do not make a distinction between the two variants of the lateral. (The r is devoiced by all speakers who observe the liquid devoicing rule.)

24.3.4.4  Morpheme-initial voicing assimilation A morpheme-initial obstruent assimilates in voicing to the preceding segment within the word. This rule has to apply after syllable-final stop devoicing has taken place. Note, for example, the voicing alternation with respect to the morpheme-initial affricate in the “profession”-morpheme -CI: (i) sinema - ǰɨ ‘film maker’ (cf. sinema ‘movie, movie theater’), av - ǰɨ (orthographically: avcı) ‘hunter’ (cf. av ‘hunting’); (ii) šarap-čɨ (orthographically: şarapçı) ‘wine maker, wine seller’ (cf. şarap ‘wine’, underlyingly /şarab/). Probably the most striking property of Turkish phon­ology is the fact that the distribution of vowels within a word is governed by vowel harmony (VH), i.e. vowels share the specification for the feature [back] and, if they are high, they also share the specification for [round]: gül-ümüz-ün rose-1pl-gen, ‘belonging to our rose’; gül-ler-imiz-in rose-pl1pl-gen, ‘belonging to our roses’; pul-umuz-un stamp-1pl-gen, ‘belonging to our stamp’; pul-lar-ɨmɨz-ɨn stamp-pl-1pl-gen, (ortho­graph­ic­al­ly: pullarımızın) ‘belonging to our stamps’. Note that the [−high] vowel of the plural morpheme, while undergoing vowel harmony for backness, does not undergo rounding harmony. Moreover, since there is a condition (mentioned earlier in the discussion on vowels) on [−high] vowels to the effect that they have to be [−round] if they are in a non-initial syllable, the negative specification of this vowel for rounding is fully determined. Note also that once a non-round vowel follows a round vowel, all vowels to the right of that non-round vowel will be non-round as well, irrespective of their height. Another way of stating the rules for VH would be as follows: Where a vowel does not share the specification for a harmony feature with preceding vowels, it will create its own harmony domain, in the sense that it will determine the specification with respect to that particular feature for the following vowels. This description also characterizes the application of vowel harmony where an exceptional vowel occurs. Many stems have exceptional vowels that violate either backness or rounding harmony or both at once; as mentioned earlier, the second vowel of the progressive marker -(I)yor is also exceptional in this respect and never

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages alternates. (Capital letters denote archiphonemes whose missing feature values are predictable by rule. In the case of vowels, I stands for a [+high], A for a [−high] vowel before application of vowel harmony. In the case of consonants, a capital letter stands for a segment which may undergo syllable-final stop devoicing, morpheme-initial voicing assimi­ lation, or intervocalic k-deletion. Symbols in parentheses denote affix allomorphy in those instances where the segment in question is deleted after a ‘like’ segment (i.e. a vowel after a vowel, a consonant after a consonant).) In such cases, it is the exceptional vowel (or, if there is more than one, the last exceptional vowel) that determines what kind of vowel harmony the following vowels will undergo. This is illustrated by the following examples: ešofman-ɨn-ɨ (ortho­graph­ic­al­ly: eşofmanını) track.suit-3sg-acc, ‘his track suit, accusative’, imanɨn-ɨ (orthographically: imanını) faith-3sg-acc, ‘his faith, accusative’; afgan-istan-ɨn-ɨ (orthographically: Afganistanını) Afghan-country-3sg-acc, ‘his/her Afghanistan, accusative’.1 Sometimes, however, a consonant rather than a vowel can determine (backness) harmony. This happens when a p ­ alatal consonant unpredictably follows a back vowel in the same ­syllable and where that consonant is in stem-final position (or a member of a stem-final consonant cluster). In such cases the following vowels will exhibit front harmony; i.e. the ‘trigger’ of vowel harmony will be the exceptional consonant rather than the regular vowel, e.g. kalp ‘heart’, a stem which we discussed in Section  24.3.1, as having a pal­atal lateral liquid, rather than a velar one, despite the [+back] stem vowel. This exceptional palatal consonant triggers vowel harmony for [−back] vowels that follow the stem; e.g. accusative: kalb-i, with a front vowel, rather than the expected kalb-ɨ.

24.3.4.5  Labial attraction There are a number of stems with a vowel sequence of a . . . u and an intervening labial consonant (the latter can also be part of a consonant cluster). Since the second vowel, being high, should undergo rounding harmony, it should surface as an ɨ. Its rounding has traditionally been ascribed to the preceding labial consonant. Some examples are avlu ‘courtyard’, havlu ‘towel’. However, in modern standard Turkish, there are a number of examples where the sequence a . . . ɨ shows up despite an intervening labial consonant, e.g. çarmıh ‘cross’, sabır ‘patience’. Moreover, an even larger number of stems exhibit a . . . u sequences without any intervening labial consonant; e.g. ka:nun ‘law’, arzu ‘desire’, fasulye ‘bean’. 1 Now that the reader is familiar with the consonants and vowels of Turkish as well as with the orthographic symbols corresponding to them, I shall use Turkish orthographic symbols for the Turkish examples in the rest of the chapter.

24.3.5  Suprasegmentals Turkish has in general word-final stress which is quantityinsensitive: odá ‘room’; okyanús ‘ocean’; oku-yabil-ecek-lerin-í read-abil-fut-3pl-acc ‘that they will be able to read’. Some suffixes are exceptional, however, in two main respects: (i) They reject stress when they are in word-final position: okuyacák-tɨ read-fut-pst ‘he was going to read’; (ii) They divide the word into stress domains when they are not word-final: ver-é-me–yecek-lerin-ì give-abil-neg-fut-3pl-acc ‘that they will not be able to give’. In both circumstances, the vowel preceding the exceptional vowel (within the exceptional morpheme) receives primary stress. There is a rule that applies within a phrase or a compound to reduce stresses left-to-right: kúš kafes-ì bird cage-compound marker ‘bird cage’. This rule can be used to account for the stress in words like verémeyeceklerinì which consist of more than one stress domain and exhibit word-final non-primary stress, whereby the stress in the first domain trumps the stress in the second domain, with the latter becoming a secondary stress, and the stress in the first domain becoming the primary stress in the word. Exceptionality with respect to stress is also exhibited by some unsuffixed stems. Such items do not fall into one clearly and independently defined set. Many (but not all) borrowed stems and almost all place names fall into this group, within which there are subregularities: they are stressed on the antepenultimate syllable, if it is the first non-final closed syllable; otherwise, the penultimate syl­ lable is stressed. Thus, in the lexemes that fall into this group, the stress is quantity-sensitive. Some illustrative examples ­follow: Érzurum, İstánbul, Adána, fasúlye ‘bean’, lokánta ‘restaurant’. Another property of Turkish concerning stress was briefly mentioned in the discussion of vowels, namely that long vowels that are not word-final are not stressed, when the word-level stress is regular and thus falls on the final syllable, whose vowel is typically short; one example mentioned there was ara:zí ‘(piece of) land’, where the final ­syl­lable with its short vowel bears stress, rather than the long vowel.

24.4 Morphology Turkish is a prime example of a language with agglutinative morphology. For most morphemes, it is possible to establish a one-to-one relationship between form and function. Most bound morphemes are suffixes. There is only one prefixing morphological process: a reduplication process that intensifies the meaning of an adjective or adverb, and which involves

397

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt reduplicating the first (C)V sequence of the root, followed by a “linking” consonant. What consonant will show up in that capacity is not always predictable, but the selection of that consonant appears to involve a dissimilation process; e.g. açık ‘obvious’, apaçık ‘completely obvious, crystal clear’; beyaz ‘white’, bembeyaz ‘completely, intensively white’; sarı ‘yellow’, sapsarı ‘intensively and completely yellow’.

24.4.1  Inflectional morphology of nouns Turkish has no morphological topic markers; topics are typically found in clause-initial position, as will be mentioned in the section on syntax. Gender is not morphologically marked, either. Case and number are marked on the head noun of a noun phrase; however, there is no concord in Turkish noun phrases, and thus case and number are not marked on modi­ fiers such as adjectives, nor on determiners; e.g. bu eski ­ev-ler-de this old house-pl-loc ‘in these old houses’. Note that neither the adjective nor the demonstrative bear any morphological marking for either plural or locative case. Another noteworthy property of noun phrases might be called “anti-concord”: when a numeral or certain quan­ti­ fiers are present as modifiers in a noun phrase, the head noun cannot bear the plural morpheme as in (1). (1)  a. bardak-lar   glass-pl   ‘glasses’ b.  beş   bardak (*-lar)   five  glass (*-pl)   ‘five glasses’ c. birçok bardak(*-lar)   many  glass(*-pl)   ‘many glasses’ Turning to the ordering among nominal inflectional morphemes, the first member of a sequence of nominal inflectional suffixes is the plural marker -lAr, illustrated above: ev-ler house-pl ‘houses’. Next come nominal agreement suffixes. These are often referred to as “possessive suffixes” in traditional literature, the reason being that the nominal stem they attach to is often, if not always, interpreted as possessed by a noun phrase within the clause or phrase. The reason they are referred to as “agreement suffixes” here is that they express the person and number features of their “possessors.” These suffixes will be listed in the next section. Case morphemes occur last, e.g. ev-ler-imiz-i house-pl-1placc ‘our houses (accusative)’. Not more than one case mor-

398

pheme can occur within an immediate sequence of suffixes. The question of how many cases Turkish has is a matter of debate. Traditional Turkish grammars usually assume five cases: nominative: not marked overtly; accusative: -(y)I; dative: -(y)A; locative: -DA; ablative: -DAn. It is a fairly wellknown phenomenon that the accusative and the genitive don’t show up when the constituent in question (i.e. the direct object or the subject of a nominalized clause, re­spect­ ive­ly) is non-specific (non-referential), see (2). (2)  a. Oya kedi-yi yakala-dı.   Oya cat-acc  catch-pst   ‘Oya caught the cat.’ b. Oya bir  kedi-yi   yakala-dı.   Oya a   cat-acc  catch-pst   ‘Oya caught a (specific) cat.’ c. Oya bir  kedi  yakala-dı.   Oya  a  cat  catch-pst   ‘Oya caught a (non-specific, non-referential) cat.’ d. Oya kedi  yakala-dı.   Oya cat  catch-pst   ‘Oya caught cats (Oya cat-caught).’ As we shall see in the examples of possessive phrases, the possessor is marked with the genitive; thus, the genitive has to be recognized as an additional case in Turkish; its shape is -(n)I(n). (The reason why the genitive hasn’t been recognized as a regular case in traditional Turkish grammars is probably because it is not assigned by a verb.) Furthermore, Turkish has an instrumental: -(y)lA, which is a cliticized form of a formerly unbound morpheme. From a synchronic point of view, there are two criteria that argue against viewing it as a genu­ine case morpheme: (i) it is exceptional from the point of view of stress (as are all other cliticized morphemes), while all other case morphemes (including the genitive) are regular; (ii) it follows the genitive when it is suffixed to a personal pronoun and hence behaves like a postposition that governs a case—namely the genitive in this instance—and not like a regular case morpheme, which can never immediately follow another case suffix, as was mentioned earlier. This art­icle will not take a stand on this controversial issue. We shall turn to nominal, “possessive,” suffixes in the next subsection.

24.4.2  Pronouns and “possessive” suffixes Personal pronouns express person and number, but not gender. They are marked with case suffixes in the same way as full noun phrases; there are a few instances of case-marking

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages by a postposition, where personal pronouns bear a different case; this will be pointed out later. A list of personal pronouns follows in Table 24.3. These are all forms in the nominative, i.e. without an overt case suffix. A list of these pronouns as marked with the six basic cases follows in Table 24.4. Note that the stem vowel of the first- and second-person singular pronoun changes from /e/ to /a/ in the dative form. Furthermore, the suffix-final consonant of the genitive suffix changes from /n/ to /m/ when that suffix attaches to the 1sg as well as 1pl pronoun. The 2pl forms can also be used when the addressee is singular, as an expression of respect. As mentioned earlier, with respect to some postpositions, while noun phrases with full nouns appear in the nom­ina­ tive in those contexts, pronouns have to be marked with the genitive: erkek gibi ‘like a man’, Oya kadar ‘as much as Oya’; but: sen-in gibi you [sg]-gen ‘like you’, sen-in kadar you [sg]-gen ‘as much as you’. Demonstrative pronouns simply consist of the demonstrative determiners without a head noun; in both usages, i.e. as demonstrative determiners, as well as when functioning as demonstrative pronouns, demonstratives express three degrees of distance, with proximal bu ‘this’, medial şu ‘that’, and distal o ‘yonder’. Interrogative pronouns are kim ‘who’, ne ‘what’, hangi-si which-3sg; ‘which (of them)’. They can be inflected for number and case, e.g. kim-ler who-pl (nominative), kim-ler-i whopl-acc, kim-ler-e who-pl-dat, kim-ler-de who-pl-loc, kim-ler-den who-pl-abl, ne-ler what-pl (nominative); ne-ler-i what-pl-acc, Table 24.3  Personal pronouns Person

Singular

Plural

1

ben

biz

2

sen

siz

3

o

onlar

ne-ler-e what-pl-dat, ne-ler-de what-pl-loc, ne-ler-den what-plhangi-ler-i which-pl-3 (nominative) ‘which ones’, hangiler-in-i which-pl-3-acc, hangi-ler-in-e which-pl-3-dat ‘to which ones’, hangi-ler-in-de which-pl-3-loc ‘at which ones’, hangiler-in-den which-pl-3-abl ‘from which ones’. Reflexive pronouns are forms of the reflexive morpheme kendi- which are inflected for the nominal agreement (“possessive”) forms; we shall turn to those after having looked at those agreement suffixes (Table 24.5). This nominal agreement paradigm is used on the head nouns of possessive noun phrases as well as on the ­nominalized verbs of gerundive and participial complements. Examples for this usage in possessive noun phrases are given in (3). abl;

(3) a.      b.    

(biz-im)  ev-imiz we-gen house-1pl ‘our house’ Ali-nin numara-sı Ali-gen  number-3sg ‘Ali’s number’

We shall see examples of nominal agreement suffixes when used on gerundive and participial predicates in Section 24.5. Having seen the nominal agreement forms, we can now look at reflexive pronouns. As just mentioned, the reflexive form kendi is inflected for person and number, by virtue of getting suffixed with the nominal agreement morphemes we just saw; Table 24.6 includes reflexive pronouns for all six persons, in the nominative and accusative. These reflexive pronouns are quite similar to their English counterparts, in that they typically have to find an ante­cedent in their own clause. The 3sg has a variant: kendisi(n); this item finds its antecedent preferably outside the clause, and is thus more similar to a personal pronoun than to a genuine reflexive pronoun.

Table 24.4  Pronouns with six basic cases Nominative

Accusative

Dative

Locative

Ablative

Genitive

ben

ben-i

ban-a

ben-de

ben-den

ben-im

sen

sen-i

san-a

sen-de

sen-den

sen-in

o

on-u

on-a

on-da

on-dan

on-un

biz

biz-i

biz-e

biz-de

biz-den

biz-im

siz

siz-i

siz-e

siz-de

siz-den

siz-in

onlar

onlar-ı

onlar-a

onlar-da

onlar-dan

onlar-ın

399

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt Table 24.5  Turkish nominal “possessive” agreement markers 1sg

-(I)ma

2sg

-(I)n

3sg

-(s)I(n)

1pl

-(I)mIz

2pl

-(I)nIz

3pl

-lArI(n)

a The suffix-initial vowels in parentheses are deleted after a stemfinal vowel; the suffix-initial consonant in parentheses is deleted after a stem-final consonant. The suffix-final consonant in par­en­ theses is deleted in word-final position.

24.4.3  Numerals The Turkish numeral system is decimal. The numerals from one to ten, decimals from twenty to fifty, and names for hundred and thousand are non-derived (see Tables 24.7–24.8).

Numerals are typically additive: 1913: bin-dokuz-yüz-on-üç ‘thousand-nine-hundred-ten-three’. Ordinal numbers are derived from cardinal numbers by adding the suffix -(I)ncI, as in Table 24.8. Restrictive collective numerals are formed by suffixing cardinal numbers with the nominal agreement suffixes; e.g. üç-ümüz three-our ‘the three of us’; dörd-ünüz four-your [pl] ‘the four of you’. Distributive numerals are formed by attaching the suffix -şAr to cardinal numbers; e.g. yedi-şer ‘seven for each’, as in  çocuk-lar-a yedi-şer elma ver-di-k child-pl-dat seven-şAr apple give-pst-1pl ‘we gave the children seven apples each’. When these forms are fully duplicated, they signal partitioning: yedi-şer yedi-şer seven-şAr seven-şAr ‘in groups of seven’. There is no classifier system in Turkish; however, there is one lexeme, tane, which originally meant ‘grain’, and which is used for countable inanimate entities; one might translate it as ‘item’. While prescriptive grammar requires this “classifier” to be used with inanimates, it is often also used in colloquial speech for animates (and even humans), as well, as illustrated in example (4).

Table 24.6  Reflexive pronouns in nominative and accusative Nominative

Accusative

Gloss

 

1sg

kendi-m

kendi-m-i

self-1sg (-acc)

‘myself’

2sg

kendi-n

kendi-n-i

self-2sg (-acc)

‘yourself’

3sg

kendi(n)

kendin-∅-i

self-3sg (-acc)

‘him-/her-/itself’

1pl

kendi-miz

kendi-miz-i

self-1pl (-acc)

‘ourselves’

2pl

kendi-niz

kendi-niz-i

self-2pl (-acc)

‘yourselves’

3pl

kendi-leri(n)

kendi-lerin-i

self-3pl (-acc)

‘themselves’

 

Table 24.7  Turkish numeral system

Table 24.8  Turkish ordinal numbers

1

bir

20

yirmi

1

bir-inci

‘first’

2

iki

30

otuz

2

iki-nci

‘second’

3

üç

40

kırk

3

üç-üncü

‘third’

4

dört

50

elli

4

dörd-üncü

‘fourth’

5

beş

100

yüz

5

beş-inci

‘fifth’

6

altı

 

 

6

altı-ncı

‘sixth’

7

yedi

 

 

7

yedi-nci

‘seventh’

8

sekiz

 

 

8

sekiz-inci

‘eighth’

9

dokuz

 

 

9

dokuz-uncu

‘ninth’

on

 

 

10

on-uncu

‘tenth’

10

400

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages (4) Dün iki tane balık  yakala-dı-m yesterday  two  TANE  fish catch-pst-1sg ‘Yesterday, I caught two fish.’

(5) Oku-n-a-ma-yacak-tı. read-pass-abil-neg-fut-pst ‘(It) was not going to be able to be read.’

24.4.4  Property words

This section starts by observing the order among verbal suffixes as well as some co-occurrence restrictions. The leftmost productive class in the string of verbal suffixes is the category of voice. This group consists of the middle/reflexive -(I)n, the reciprocal -(I)ş, the passive -Il/n, and the causative -DIr/t. (The -Il allomorph of the passive follows consonants, the -n allomorph follows vowels. -DIr is the basic allomorph of the causative; -t occurs after polysyllabic stems which end in a vowel or in the liquids r and l.) The middle/reflexive and the reciprocal cannot co-occur; in those instances where the passive co-occurs with either one, it has to follow them. In the very few examples where the causative can co-occur with the middle/reflexive and the reciprocal it has to follow them, and, while it can co-occur with the passive, it has to precede it; e.g. gör-üş-tür-ül-dü-ler see-recp-caus-pass-pst-3pl ‘they were caused to see/meet each other; they were made to/allowed to meet with each other’. Suffixes of this group can be followed by the verbal neg­ ation marker -mA, which is one of the suffixes that are exceptional from the point of view of word stress in rejecting word-final stress and causing the preceding vowel to be stressed. This suffix is followed either by one of the various mood markers or by purely verbal or gerundive/participial forms, the latter expressing tense in varying degrees of differentiation. The mood markers are: the desiderative -sA, the necessitative –mAll, and the optative -(y)a; see example (6).

Property words in Turkish are adjectives and adverbs. As in many other Turkic languages, many adjectives can also be used as nouns; e.g. büyük ‘big, old’, büyük-ler-im old-pl-1sg ‘my elders’; hasta ‘sick’, hasta-lar-ı ziyaret et-ti-m patient-placc visit do-pst-1sg ‘I visited the patients’; genç ‘young’, pist-te sadece genç-ler dansed-iyor dance floor-loc only young (­people)-pl dance-prog ‘only young people are dancing on the dance floor.’ A productive suffix deriving denominal adjectives is -lI; e.g. kuvvet ‘strength’; kuvvet-li ‘strong’. Both the comparative and the superlative are analytic; the basis of comparison is suffixed with the ablative suffix -DAn and precedes the adjective; e.g. su-dan hafif water-abl light ‘lighter than water’. The morpheme daha ‘more’ can be inserted for emphasis (e.g. su-dan daha hafif water-abl more light ‘lighter than water’), but can be left out without any ill effect, as in the first of these examples for comparatives. ‘Less than. . .’ is typically expressed by inserting az ‘less’ after the ablative expression and the adjective; e.g. su-dan az ağır water-abl less heavy ‘less heavy than water’; daha ‘more’ can be used in such a context, as well: su-dan daha az ağır water-abl more less heavy ‘less heavy than water’. The morpheme daha ‘more’ is obligatory, however, when the ablative expression is missing; e.g. bu kitap daha yeni this book more new ‘this book is newer’; bu bisiklet daha ucuz this bi­cycle more cheap ‘this bicycle is cheaper’; bu bisiklet o bisiklet-ten (daha) ucuz this bicycle that bicycle-abl more cheap ‘this bicycle is cheaper than that bicycle’. The superlative is formed by placing en ‘most’ before the adjective: Turgay en meşhur kaleci-miz-di Turgay most famous goalie-1pl-pst ‘Turgay was our most famous goalie’. Many adverbs have the same shapes as adjectives; e.g. iyi ‘good; well’. However, some adverbs are derived morphologically from adjectives, via the suffix -CA; e.g. dikkat-li attention-lI, attention-‘with’ ‘careful, attentive’; dikkat-li-ce ‘in a careful manner; carefully’.

24.4.5  Verb The verbal root can occur without any suffixation, as the imperative: Oku! ‘read!’ When suffixed, the verbal root can be the core of complex and long predicates, as in (5).

(6) Gör-üş-me-meli-yiz see-recp-neg-nec-1pl ‘We shouldn’t/mustn’t see each other.’ The suffixes of the mood category are mutually exclusive. The tenses are: definite past: -DI; reported past: -mIş; aorist: -(A)r; future: -(y)AcAK; present progressive: -(I)yor. These forms have also aspectual connotations: the past tenses denote accomplished actions and the aorist refers to actions that are either extended or repeated over a period of time. The present progressive is similar to its English equiva­lent in denoting an action that, roughly speaking, takes place at the time of the utterance. One difference is that stative verbs, unlike those in standard English, can take the progressive in Turkish (7). (7) Chomsky-yi dinle-mek isti-yor-um. Chomsky-acc  listen.to-inf  want-prog-1sg ‘I want (*am wanting) to listen to Chomsky.’

401

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt The main participial forms are those used in relative clauses: -(y)An and -DIK, and they will be discussed in Section 24.5. Also in this group (from the point of view of positional slots within the morphological word) are so-called verbal nouns and converbs (these are terms often used in traditional literature). The “verbal nouns” consist mainly of the infinitive suffix -mAK and the nominalized subjunctive marker ‑mA and will also be listed among the derivational morphemes that convert verbs into nominals. -DIK is a general indicative nominalizer that corresponds rather closely to the English gerundive -ing. It has non-future temporal value; for future, the suffix -(y)AcAK is used, which doubles as a finite future marker. Converbs (or gerundives, as they are also called, especially in traditional literature) are suffixes that yield adverbial forms. Some ex­­amples are the manner suffix -(y) ArAk, the conjunction adverbial -(y)Ip, which denotes close successions of actions, and the time adverb suffix -(y)IncA. In general, only one of the suffixes in this group can occur at a time. In other words, within the morphological sequence, the various gerundive, participial, and nominal markers take the place of the tense or mood markers, whether they have tense connotations themselves or not. However, two tense markers (as well as a tense and a mood marker) can co-occur in immediate succession to form complex tenses; in such examples, it is appropriate to view the second marker as a copula carrying the main tense or mood; the preceding sequence (see 8) would be analyzed as a participial: (8) Sınav-ım-a başlı-yor-du-m. exam-1sg-dat  start-prog-pst-1sg ‘I was starting my exam (when . . . )’ Note that in such sequences, the present progressive marker -(I)yor retains its aspectual meaning. The reported past marker -mIş is used as a perfective aspect marker in example (9) (i.e. when it is the first member of the sequence). (9) Sınav-ım-a başla-mış-tı-m. exam-1sg-dat  start-pfv-pst-1sg ‘I had started my exam (when . . .)’ All tense and some mood markers can occur as the first members in these sequences; however, only the two past tense markers and the mood marker for the desiderative (the latter as a conditional in these contexts) can occur as the second member, i.e. as the main tense or modality marker. However, all the tenses can be used as a main tense or modality within a periphrastic construction with an auxiliary verb. The most widely used auxiliary is the verb ol- ‘be, become’; see example (10).

402

Table 24.9  Verbal agreement markers 1sg

-Im

2sg

-sIn

3sg

-∅

1pl

-Iz

2pl

-sInIz

3pl

-lAr

(10) a. Sınav-ım-a başlı-yor ol- acağ-ım   exam-1sg-dat start-prog  be-fut-1sg   ‘I shall be starting my exam.’ b. Sınav-ım-a başla-mış ol-acağ-ım   exam-1sg-dat  start-pfv be-fut-1sg   ‘I shall have started my exam.’ This mixed positional group is followed by agreement markers, wherever such markers are possible. Among the suffixes that cannot be followed by agreement markers are the infinitive marker -mAK, the participial marker -(y)An (unless it functions as a verbal noun), and the gerundive marker -(y)Ip. Verbal agreement markers can be seen in Table 24.9. The verbal paradigm appears with the predicates of main clauses and of “direct complements” (for discussion of the latter, see Section 24.5). Verbal agreement is used with a main clause predicate verb, as seen in (11). (11) a. (Ben) bu başvuru-yu yarın   I this  application-acc  tomorrow    bitir-eceğ-im. finish-fut-1sg ‘I shall finish this application tomorrow.’ b.  (Biz)  her   sabah   erken  uyan-ır-ız.   We  every morning early  wake.up-aor-1pl   ‘We wake up early every morning.’ Verbal agreement used with a main clause predicate adjective is illustrated in (12). (12) a. bugün çok hasta-yım.   today very sick-1sg   ‘I am very sick today.’ b. (Siz) çok tembel-siniz.   you[pl]  very  lazy-2pl   ‘You are very lazy.’

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages

24.4.6  Derivational morphology Among the morphemes that derive nominals, those that attach to verbal stems precede those that attach to nominal ones, where the two types co-occur. -Im derives (mostly abstract) nouns from verbs, -lI derives adjectives from nouns, -lIK derives abstract nouns from adjectives. These derivational suffixes and their interactions are illustrated by the following examples: ver-im give-abstract noun, ‘profit’; ver-im-li give-abstract noun-with (adjective) ‘profitable’; ver-im-li-lik give-abstract noun-with (adjective)-abstract noun, ‘profitability’. The suffixes exemplified in the last two examples can attach to underived nominals, as well: bahçe-li garden-with ‘with a garden’; dürüst-lük honest-abstract noun, ‘honesty’. All of these derivational suffixes are productive; two other productive derivational suffixes are the action/manner suffix -(y)Iş (as in ara-yış search-abstract noun; ‘quest’), and the result/action morpheme -mA (as in kavur-ma roast-resultative; ‘roasted meat’). Among der­iv­ation­al suffixes which can attach to underived nominals, we find -CI, deriving nouns meaning ‘professional’ (dondurma-cı ‘ice cream-­professional’, ‘ice-cream maker/seller’), and -sIz, deriving adjectives meaning ‘without’ (e.g. para-sız money-without ‘without money’, or ‘free’, as in a free event that does not charge entrance fees; metelik-siz small coin-without ‘broke’). There is only one completely productive morpheme that derives verbs from nominals: -lA, which has a meaning related to the causative; e.g. karşı-la-mak opposite-lA-infinitive ‘to go to meet; to respond; to reply to’; kara-la-mak black-lAinfinitive ‘to blacken’.

24.5 Syntax The syntax of Turkish is very similar to that of other Turkic languages, as well as to that of Mongolian. The main properties are right-headedness across category types (i.e. clauses, verb phrases, noun phrases, postpositional phrases), and subordination which heavily (but not exclusively) relies on nom­ in­alized embedded clauses. Other general features include very flexible word order, productive omission of the subject, the somewhat different omission of other arguments, as well, and lack of a designated movement of question elem­ents to the front of the clause (i.e. Turkish is a “wh-insitu” language).

24.5.1  The clause The unmarked word order in sentences is SOV; if there is more than one object, and if one of them is a direct object,

the order with the indirect object closer to the verb is judged by many native speakers to be less marked than others, at least with certain verbs, as in (13). (13)  Oya hediye-yi kız-ın-a ver-di. Oya  present-acc  daughter-3sg-dat  give-pst ‘Oya gave the present to her daughter.’ However, other orders are possible, as well. Most of the constituent orders which differ from the unmarked order have a pragmatic, discourse-oriented function, in that the pos­ ition immediately preceding the verb is the focus position and the sentence-initial position is the topic position. New information and material stressed for emphasis appear in focus position and, in addition to being syntactically marked in this way, also receive intonational stress. The topic, i.e. the ma­ter­ial that the sentence is about, is placed at the beginning of the sentence and is often separated from it—orthographically by a comma and by a slight pause in speech. In contrast with a number of other SOV languages (e.g. Japanese), Turkish very productively permits nonverb-final constructions. Such sentences arise when material is added as an afterthought or when the speaker assumes the hearer knows about it: Oya kız-ın-a ver-di hediye-yi, where the direct object hediyeyi ‘present-acc’ is post-verbal: compare with (13). For this example to be felicitous, it must be clear within the discourse that there existed a present, and that something happened to that present, or even that Oya gave the present to somebody. Some recent literature claims that such post-verbal, backgrounded constituents have the same pragmatic as well as structural properties as topicalized constituents. This claim has been successfully contested on structural grounds, showing that post-verbal constituents have different syntactic properties from those of pre-verbal ones. On the other hand, the pragmatic aspect of this issue is still being debated in the literature. An embedded sentence which is nominalized takes up the same position that the corresponding noun phrase with the same grammatical relation would and can move around within the main clause with the same ease as a regular noun phrase; see (14). (14) a.  Oya ban-a  [yarış-ı kazan-dığ-ın]-ı söyle-di.   Oya  I-dat race-acc  win-ind.nmlz-3sg-acc  say-pst   ‘Oya told me that she won the race.’ b. Oya ban-a — söyle-di  [yarış-ı  kazan-dığ-ın]-ı.   (The original site of the “scrambled” embedded clause is marked with a —.) The constituents of the embedded sentence are somewhat less free in their word order. While they can also successfully

403

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt vary from the canonical SOV order within their own clause, such constituents have to move to the right of the highest sentence when they cross the boundary of their own clause and cannot move into higher material; thus, compare the following examples in (15) with the last set of examples in (14).

‘why’, niçin ‘why’, niye ‘why’, hangi ‘which’, kim ‘who’. These elements, which are inherently focused, can be found in various positions, as focused constituents can in general. However, again like focused elements in general, they are preferred when they immediately precede the verb as in (17).

(15) Oya ban-a [— kazan-dığ-ın]-ı  söyle-di  yarış-ı. *Oya  ban-a  [— kazan-dığ-ın]-ı yarış-ı söyle-di (The original site of the “scrambled” constituent is marked with a —.)

(17) Oya-ya hediye-yi kim ver-di? Oya-dat  present-acc  who  give-pst ‘Who gave the present to Oya?’

As mentioned earlier, an important property of Turkish is that a subject can be left unexpressed in finite clauses (i.e. those exhibiting some type of subject-predicate agreement) as well as in possessive noun phrases, as in (16). The sites of the “missing” subjects are marked with a —. (16) a. —  yarış-ı   kazan-acağ-ım.       race-acc  win-fut-1sg   ‘I will win the race.’ b. Oya [ —  ev-in]-i sat-tı.   Oya    house-3sg-acc  sell-pst   ‘Oya sold her house.’ The ability to freely omit the subject in clauses that exhibit predicate-subject agreement, and to also omit the possessor in noun phrases has been traditionally linked to the rich agreement morphology of Turkish, i.e. to the fact that agreement suffixes will identify the person and number of the subject (or possessor) which is unexpressed. The subject and the possessor can be overtly realized, i.e. they are not in complementary distribution with the agreement morph­ology on the predicate or on the possessee; this redundancy is usually motivated by discourse factors, such as emphasis. Although Turkish has no agreement markers for nonsubjects, it is possible to omit such constituents, as well; e.g. kaybet-ti-m lose-pst-1sg ‘I lost (it)’. Such examples are more restricted, however, than examples with omitted subjects. Such utterances are felicitous only if the antecedent of the omitted constituent has been mentioned in the discourse or has somehow been made clear by a pragmatic act. The formation of content questions in Turkish does not involve a dedicated movement rule to a clause-initial pos­ ition, as in many Indo-European languages. In other words, just like the majority of Turkic languages (and of other Altaic languages), Turkish is “wh-in-situ.” The language has various wh-question particles, most of which are morphologically derived from the particle ne ‘what’: ne ‘what’, neden

404

Note that the subject here is the wh-word kim ‘who’; the word order here is not the canonical SOV, but rather (IO) OSV. Although a subject-initial order would also be well-formed with the wh-element in the canonical subject pos­ition, the example as given illustrates the preference for the wh-constituent to be placed left-adjacent to the verb. In the answer, a corresponding order, with the focused constituent corresponding to the wh- element, would likewise have the focused constituent to the immediate left of the verb as in (18). (18) Oya-ya hediye-yi kız-ı ver-di. Oya-dat  present-acc  daughter-3sg  give-pst ‘It was her daughter who gave the present to Oya.’ Yes-no questions are formed by suffixing the particle -mI to the constituent questioned; if the whole sentence is questioned, the particle is attached to the verb, following the subject agreement markers in simple tense/aspect forms and preceding the copula (see the discussion of complex verbal forms in the morphology section) and its tense and agreement markers in complex forms, as in example (19). (19) a. (Sen)  Oya-ya hediye-yi ver-di-n=mi?2   you Oya-dat  present-acc  give-pst-2sg=q   ‘Did you give the present to Oya?’ b. Sen   hediye-yi    Oya-ya=mı   ver-di-n?   you present-acc  Oya-dat=q  give-pst-2sg   ‘Was it Oya whom you gave the present to?’ c. Sen   Oya-ya   hediye-yi   ver-ecek=mi-y-di-n?   you Oya-dat  present-acc  give-fut=q-cop-pst-2sg   ‘Were you going to give the present to Oya?’ 2 Standard Turkish orthography dictates that the Yes/No question par­ ticle be written separately from the word it cliticizes on. The reason for this is presumably the fact that, as a clitic, this particle does not carry wordlevel final stress; instead, it is pre-stressing. However, from a mor­pho­ phono­logic­al point of view, the particle is clearly part of the word, because it undergoes vowel harmony. I therefore include it in the word in the representations above.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages In both the first and third examples, the entire sentence is questioned, and the question particle has attached to the predicate. However, there is a difference: while the predicate of the first example is a simple one, the predicate of the third example is complex, exhibiting both future and past suffixes. The question particle attaches to the inflected main verb in both instances; in the first example, this means that the particle is the last morpheme in the word (and in the sentence), while in the third example, the inflected main verb is a participle, followed by the inflected copula. The question particle, by virtue of attaching to the main verb participle, is now followed by the inflected copula. In the second example, only the indirect object is questioned; the question particle, by attaching to the IO, makes it into the focus of the sentence, and thus causes it to be placed to the immediate left of the verb—as we saw earlier, the preferred option for any focused constituent. (As is the case with other focused constituents, this placement is not ob­liga­tory, but only preferred in general.) Embedded clauses in Turkish lack complementizers that introduce (or terminate) clauses. The most widespread strategy of subordination involves nominalized clauses, whose predicates are “nominalized” by using various morphemes (as we saw in Section 24.4). The subject agreement markers on these “nominalized” predicates come from the nominal (i.e. “possessive”) rather than the verbal paradigm; more­ over, these nominalized clauses carry overt case markers, and their subject is in the genitive, rather than in the nom­ ina­tive as in a fully tensed clause, see (20). (20)  a.         b.        

[Oya-nın yarış-ı kazan-dığ-ın]-ı race-acc win-ind.nmlz-3sg-acc Oya-gen duy-du-m. hear-pst-1sg ‘I heard that Oya won the race.’ [Oya-nın  yarış-ı kazan-ma -sın]-ı race-acc win-sbjv.nmlz-3sg-acc Oya-gen isti-yor-um. want-prog-1sg ‘I want Oya to win the race.’

The two ‘nominalization’ morphemes illustrated in (20) (i.e. -DIK, the “indicative nominalization,” and -mA, the “sub­junct­ive nominalization”) are the main nominalization forms exhibited by embedded clauses that function as arguments of the verbs of the higher clause. The semantics of that higher verb and the propositional properties of the clause determine which one of the two morphemes will be chosen, as illustrated by the examples in (20) and their translations.

A subset of the verbs that take clauses with the sub­junct­ ive nominalization marker also take clauses that are marked with the infinitive suffix -mAK. These infinitival clauses are comparable to English infinitivals in that they necessarily lack overt subjects; note also that they do not carry agreement morphology, see example (21). (21) Ben  [beğen-il-mek]  isti-yor-um. want-prog-1sg  I like-pass-inf ‘I want to be liked.’ Infinitivals can take case markers, too, and are thus shown to be genuine nominalized clauses, as well, as illustrated in (22). (22) Mustafa [tevkif ed-il-mek]-ten kork-uyor. Mustafa  arrest  do-pass-inf-abl  fear-prog ‘Mustafa is afraid of getting arrested.’ Clauses that are postpositional objects and adverbial clauses are either nominalized, or, in the case of adverbial clauses, exhibit predicates that have special “adverbial” forms; those are typically referred to as converbs. The nominalized clauses of this type exhibit similarities to the argument clauses just discussed, but they are also different, in that when their predicate is an indicative, the subject is in the nominative rather than in the genitive. When the nominalization is a subjunctive, the subject is in the genitive, as expected. To discuss these details would go beyond the scope of this chapter; there is some literature that addresses these issues. The most typical adverbial clauses are those with a converb predicate. The subject of such clauses is typically (and, in manner adverb clauses, obligatorily) omitted; where the subject does show up, it is in the nominative (rather than in the genitive, as might be expected in a nonfinite clause), as in (23c). (23) a.  Oya sahne-ye [___  kork-arak ] çık-tı. fear-manner.cvb  go.up-pst   Oya  stage-dat      ‘Oya went on stage while being afraid (in a fearful manner).’   (The “silent” subject is represented with an underlined space.) b. Akgül   [___  yorul -unca ]  ev–e    dön-dü.   Akgül     tire-temp.cvb  home-dat  return-pst   ‘Akgül returned home when (she) got tired.’ (The “silent” subject is represented with an underlined space.) c.        

Akgül    [oğl-u      yorul-unca ] Akgül    son-3sg[nom] tire-temp.cvb ev-e   dön-dü. home-dat  return-pst ‘Akgül returned home when her son got tired.’

405

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt As mentioned above, adverbial clauses can also bear the indicative and the subjunctive nominalization markers. A small subset of embedded clauses exhibits verbal morph­ ology and syntax identical to that of main sentences. Such clauses occur with verbs of belief and are, essentially, interchangeable with corresponding -DIK clauses (i.e. in­di­ca­tive nominalizations), which can also be taken by verbs of belief. In some of the few instances in the literature where these constructions have been noted, they have been called “direct complements.” They are of two types: (i) The embedded subject is marked nominative; the embedded verb exhibits regular verbal subject agreement marking, as in (24). (24) Arkadaş-lar-ım  [ben Almanya-ya gid-eceğ-im] I[nom]  Germany-dat  go-fut-1sg friend-pl-1sg san-ıyor. believe-prog ‘My friends believe that I will go to Germany.’ (ii) The embedded subject is marked accusative; the embedded verb exhibits only tense/aspect marking, but no agreement marking, as in (25). (25) Arkadaş-lar-ım  [ben-i  Almanya-ya gid-ecek]  friend-pl-1sg I-acc Germany-dat  go-fut san-ıyor. believe-prog ‘My friends believe that I will go to Germany.’ (lit.: ‘My friends believe me to be going to Germany.’) In addition, there are speakers who also accept a hybrid form where the embedded subject is accusative, but where the embedded verb exhibits regular verbal agreement markers, as in (26). (26) Arkadaş-lar-ım  [ben-i  Almanya-ya gid-eceğ-im] friend-pl-1sg I-acc Germany-dat  go-fut-1sg san-ıyor. believe-prog ‘My friends believe that I will go to Germany.’ (lit.: ‘My friends believe me to be going to Germany.’) No speakers accept such tensed direct complement clauses, when the embedded verb does not exhibit subject-agreement marking and when the embedded subject is in the nom­ina­ tive, as in (27). (27) *Arkadaş-lar-ım [ben Almanya-ya gid-ecek] friend-pl-1sg [nom]  I[nom]  Germany-dat  go-fut san-ıyor. believe-prog Intended: ‘My friends believe that I will go to Germany.’ (lit.: ‘My friends believe I will go to Germany.’)

406

Like all phrases in the language, relative clauses in Turkish are head-final, i.e. the modifying clause precedes the clause-external head. The verbs of such modifying clauses are nominalized, i.e., traditionally speaking, they are participles. Also, just as is the case with all nominalized embedded clauses we have seen so far, they lack complementizers. There is a gap in the position of the constituent within the clause that corresponds to the head. The indicative nominal marker -DIK is the basic type of nominalizing morpheme in these constructions; -mA, the subjunctive nominalizer, never occurs in relative clauses, and neither does the infinitive. -DIK is replaced by the morpheme -(y)An where the constituent which is the target of the relativization is a subject, part of a subject or a non-subject of a clause that lacks a subject (e.g. as in an intransitive passive construction as in the last example in 28); yet another difference between relative clauses with -DIK and those with -(y)An follows from this last property: -DIK is, as usual, followed by nominal agreement morphology; -(y)An never is. These descriptions are illustrated below in (28); the “gapped” target of relativization is marked with an underline, (28) a.  [Akgül-ün ___  git-tiğ-i] üniversite go-dik-3sg  university   Akgül-gen   ‘the university that Akgül goes to’ b. [ ___  üniversite-ye   gid-en]   öğrenci      university-dat  go-(y)An  student   ‘the student who goes to the university’ c. [[___  oğl-u]   üniversite-ye   gid-en]   kadın       son-3sg  university-dat  go-(y)An  woman    ‘the woman whose son goes to school’ d. [___  gid-il-en]    üniversite       go-pass-(y)An  university   ‘the university that is gone to’ Embedded wh-questions are similar to their affirmative counterparts, i.e. they are nominalized; however, only the indicative clauses with -DIK and -(y)AcAK can host embedded wh-questions; subjunctive nominalized clauses with -mA cannot; in this respect, embedded wh-questions and relative clauses are similar. Coordination of clauses offers the choice among the options of (i) a coordinating conjunction, ve ‘and’, borrowed from Arabic, and showing up between the conjuncts; (ii) the native morphemes -DA ‘also’ and -(y)sA ‘in contrast, on the other hand’, which cliticize to the first constituent of the second conjunct, and (iii) under certain circumstances, the simple sequencing of clauses next to each other, without any coordination marker. These options are illustrated in (29), in this order.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages (29) a.        

Oya opera-ya   git-ti   ve   Ali  ev –de Oya  opera-dat  go-pst  and Ali home-loc kal-dı. stay-pst ‘Oya went to the opera and Ali stayed at home.’

b.        

Oya opera-ya git-ti,  Ali   de   ev-de  Oya opera-dat  go-pst  Ali and home-loc kal-dı. stay-pst ‘Oya went to the opera and Ali stayed at home.’

c. Oya opera-ya git-ti,   Ali-yse     ev –de    Oya opera-dat  go-pst  Ali-however stay-pst   kal-dı. home-loc ‘Oya went to the opera, Ali however stayed at home.’ d. Oya opera-ya   git-ti,   Ali  ev-de    kal-dı.   Oya opera-dat  go-pst  Ali home-loc  stay-pst   ‘Oya went to the opera (and) Ali stayed at home.’ The meaning of “being” is expressed by a (sometimes abstract, i.e. not overtly realized) copula, which sometimes shows up as the palatal glide /y/, and is sometimes deleted, for phonological reasons, as in (30). (30) Bugün çok hasta-y-ım/yorgun-um. today  very  sick-cop-1sg/tired-1sg ‘I am very sick/tired today.’ While the presence of the palatal glide in examples such as the last one may be due to phonological rules (e.g. it is possible that the glide is inserted to separate two vowels and does not show up after a consonant, as in the second ex­­ ample), the presence of such a glide in other tenses cannot be argued away easily, see (31). (31) Dün çok hasta-y-dı-m/yorgun-du-m. yesterday  very  sick-cop-pst-1.sg/tired-[cop]-pst-1sg ‘Yesterday, I was very sick/tired.’ It is hard to imagine a phonological rule which would insert a glide preceding a consonant (i.e. here, the initial con­son­ ant of the past tense morpheme). There is no such insertion rule elsewhere in the language. It makes much better sense to posit a bound copula morpheme, consisting of the palatal glide, and to delete it after a consonant. Modern standard Turkish does not have a natural, native verb “have.” The notion of “having” is expressed via an ­existential construction whose subject is a possessive, as in (32). (32) a. (Ben-im) garaj-da üç araba-m  var. garage-loc  three  car-1sg there.is   I-gen   ‘I have three cars in the garage.’

b. (Ben-im)  üç   kız-ım     var. three daughter-1sg  there.is   I-gen   ‘I have three daughters.’ Rather like lacking a simple, transitive verb “have,” Turkish does not have a simple transitive verb “need.” However, there now does exist a verb “need,” gerekse-, which is a transitive verb, i.e. it is supposed to take an accusative object; this is a neologism, having come about in the 20th century. There are very few tokens of this transitive verb, and all the native speakers of Turkish I polled informally rejected it. There is now also a related verb with the bound reflexive morpheme -In: gereksin-, which takes a dative complement; in addition, there is a deverbal noun, marked with nominal “possessive” agreement: gereksinim-agr.; this is placed into the existential construction just discussed. Native speakers feel more comfortable using either one of these non-transitive constructions, as in (33). (33) a. Ben san-a gerek-s-in-iyor-um.   I[nom] you-dat  necessary-deadj.verb-refl-prog-1sg   ‘I need you, am in need of you.’ b. Ben-im  san-a    I-gen   you-dat    gerek-s-in-im-im necessary-deadj.verb-refl-deverbal.n-1sg var.   there.is   ‘I need you.’ (lit.: ‘I have (a) need of you; my need of you exists’) These new forms are not being used very widely.

24.5.2  The nominal group Turkish has morphological case, expressed as suffixes placed on the nominal head of noun phrases; case is assigned by verbs and postpositions, and nominative/genitive, i.e. the subject cases, are presumably assigned by verbal and nom­ in­al agreement, respectively. A list of the case morphemes was offered in Section 24.4, along with a discussion. As mentioned in Section 24.4, there are no morphological markers for either focus or topic. Instead, the placement of a constituent in clause-initial position will mark it as a topic, while the placement of a constituent to the immediate left of a verb will mark it as a focused element. A list of nominal “possessive” agreement suffixes was given in Section 24.4, too. The order of the constituents in a noun phrase is as follows. In possessive noun phrases, the possessor precedes the head noun; in “regular,” i.e. non-possessive, noun phrases, modifiers precede the head. Where there is both an adjectival modifier and an article (only the in­def­in­ite

407

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt article is overtly expressed in Turkish), the adjective precedes the article; where there is both a numeral and an adjective, the unmarked order is for the numeral to precede the adjective, as in (34).

In addition to this difference, based on referential semantics and on genitive marking (versus lack thereof), we saw earl­ ier that the possessor in a possessive noun phrase can move around, as in (38).

(34) a.  Ali-nin piyano-su   Ali-gen  piano-3sg   ‘Ali’s piano’

(38) ___  doktor-u kadın-ın   doctor-3sg  woman-gen This kind of movement is impossible in compounds: * ___ doktor-u kadın.

b. yeni   bir  piyano   new a  piano   ‘a new piano’ c. üç    yeni   piyano   three new piano   ‘three new pianos’

24.5.3  The verbal group

Compounds, whose non-heads are nominal, carry a compound marker on the compound head, as shown in (36). This marker is actually the 3sg nominal agreement marker. Thus, we can say that nominal compounds differ from possessive noun phrases only with respect to the non-head: in possessive noun phrases, this is the possessor, which is typically referential, and it is marked with the geni­tive; in nominal compounds, the non-head is generic and thus cannot be marked with the genitive. A possessive noun phrase corresponding to the compound is illustrated in (37).

The various verbal suffixes and the ordering among them were discussed in the morphology section. Here, I will look mainly at types of passive, a syntactic process that reduces the valency of the verb. As shown in the morphology section, the passive in Turkish is marked by the morpheme -Il (with a mor­pho­ phon­emic alternant -(I)n) on the verbal stem. From the ­syntactic point of view, there are two types of passive constructions; they will be referred to as “transitive passive” and “intransitive passive”; the former type is derived from transitive verbs, the latter from intransitive ones. The term “transitive” refers to verbs that take direct objects (noun phrases that are marked accusative when they are specific), and “in­transi­tive” refers to verbs that do not take such objects (i.e. that either lack objects altogether or take only indirect or oblique objects). The two constructions exhibit the following surface differences: the patient of the action (or, in other words, the direct object of the corresponding active sentence) is the subject of the transitive passive ­construction. This claim is supported by the fact that these subjects exhibit syntactic properties typical of subjects in general: they appear in the nominative case in fully tensed clauses, and in sentence-initial position when the word order is unmarked; the verb agrees with these constituents, as it does with subjects in general. These constituents can correspond to the understood subjects of infinitivals; and the agent of the action can appear in an agentive phrase. However, the non-accusative objects that an intransitive verb might co-occur with are not surface subjects in an intransitive passive construction (in the sense that they do not exhibit the criteria just mentioned), and agentive phrases are judged to be awkward at best, if not completely ungrammatical. These generalizations are illustrated in (39).

(37) kadın-ın doktor-u woman-gen  doctor-3sg ‘the woman’s doctor’

(39) a. (Biz)  sev-il-iyor-uz.   we   love-pass-prog-1pl   ‘We are loved.’

A relative clause would, in an unmarked order, precede the numeral and the adjective, but follow the possessor, as in (35). (35) Ali-nin [pencere-den  düş-en] üç yeni Ali-gen  window-abl fall-(y)An  three  new piyano-su piano-3sg ‘Ali’s three new pianos, which fell out from the window’ The genitive-marked possessor can “scramble” in either direction, while the article and numerals cannot. The adjective is not free to move, either, as far as spoken language and written prose are concerned. In poetry, however, an adjective can occur to the right of its head. Space considerations make it impossible to discuss nominal compounds, but I will illustrate this construction briefly in (36). (36) kadın doktor -u woman doctor -compound marker ‘woman doctor; gynecologist’

408

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

tur kish and the southwester n tur kic l anguages b. Biz-e  yardım  ed-il-di.   we-dat  help   do-pass-pst   ‘We were helped.’ c. *biz (-e) yardım  ed-il-di-k. do-pass-pst-1pl   we (dat)  help   Intended reading: ‘We were helped.’ Verbs that do not take any objects at all can also appear in impersonal passive constructions, as in (40). (40) a. Koş-ul-du.   run-pass-pst   ‘It was run (i.e. running took place).’ b. Eğlen-il-di.   amuse-pass-pst   ‘Fun was had.’ Agentive phrases are ill-formed in such objectless constructions. Another valency-related morphosyntactic phenomenon is the causative. The variants of this morpheme as well as its position within the sequence of verbal morphemes were discussed in Section 24.4. From a syntactic point of view, the causative can be viewed as the opposite of the passive, given that the passive reduces the valency of the verb, while the causative adds to it, by adding a causer. For the purposes of this survey, the most important aspect of the Turkish causative is case-marking: when an intransitive verb is causativized, the original agent gets marked with the accusative; when a transitive verb is causativized, the original agent gets marked with the dative, as in (41). (41) a.      

Intransitive verb Ali   koş-uyor. Ali run-prog ‘Ali is running.’

b.      

Causativized form Oya Ali-yi koş-tur-uyor. Oya  Ali-acc  run-caus-prog ‘Oya is making Ali run.’

c.      

Transitive verb Ali   konçerto-yu   çal-dı. Ali concerto-acc  play-pst ‘Ali played the concerto.’

d.      

Causativized form Oya Ali-ye konçerto-yu çal-dır-dı. Oya  Ali-dat  concerto-acc  play-caus-pst ‘Oya made Ali play the concerto.’

Negation is expressed on the verb, via the suffix -mA, which is one of the pre-stressing suffixes, as in (42). (42) Ali konçerto-yu çál-ma-dı. Ali  concerto-acc  play-neg-pst ‘Ali didn’t play the concerto.’ When the predicate is not verbal, but rather is adjectival or nominal, this bound negation marker cannot be used. Instead, the free morpheme değil ‘is not’, a morpheme which can be characterized as a negative copula, must be used, as in (43). (43) Ali dün hasta  değil-di. Ali  yesterday  sick neg.cop-pst ‘Ali was not sick yesterday.’ Tense-aspect and mood markers were described in Section 24.4.

24.6 Lexicon The vocabulary of modern standard Turkish is a mixture of lexemes inherited from its Turkic mother language (i.e. a language closely related to Old Turkic, but not identical to it), borrowings from Arabic and Persian (mainly going back to Ottoman), borrowings from (dialects of) modern Greek, as well as from French and Italian, and, more recently, from English. Some of the Turkic items are made up by the Language Academy, based on Turkic stems, and constructed according to Turkish morphological generalizations, using Turkish or Turkic functional bound morphemes. The core vocabulary is essentially Turkic; some Arabic and Persian loanwords are still being used, despite the efforts of the Language Academy to “purify” the language. Many of those borrowings are in the realm of academic, legal, administrative, and political activities; this is not surprising, given that the nomadic Turkic tribes that invaded Asia Minor starting in the 11th century did not have native vocabulary for these areas which became important to them once they settled in an area roughly corresponding to today’s Turkey (minus the European part of the country, which was conquered later). Many of the inhabitants they encountered once they were settled were Greek, or at least spoke Greek; many culinary items in today’s Turkish are Greek, especially those having to do with seafood; e.g. karides ‘shrimp’, ıstakoz ‘lobster’, barbunya ‘red mullet’. This is not surprising, given that the invading nomads came from landlocked regions and obviously had no native lexemes for such items. Borrowings from French are not surprising, either; many intellectuals

409

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Jak lin Kor nfilt and administrators studied in France, both during Ottoman times and later in Republican times. Moreover, the Ottoman Empire and France maintained close contacts politically. As for Italian, it was spoken in Constantinople, given a Venetian as well as a Genoese presence in the city which continued (albeit not as extensively) during Ottoman and Republican years. Thus, today’s vo­cabu­lary of Turkish has, in addition to these diverse borrowings, some formations which either never existed in any of the donor languages, or at least not in exactly the same shape; e.g. postahane ‘post-place, house’, ‘post office’, which is postxaane or postxuune in Persian; the

410

Turkish posta was presumably borrowed from Italian, and hane was borrowed from Persian; apparently the Persian compound was not borrowed in its entirety. Perhaps more interesting is pastahane ‘cake-location’, ‘pastry shop’; the expressions in Persian for ‘pastry shop’ do not involve xaane ‘house’ at all. At the moment, there is a tug of war between certain French borrowings, which are older, and English borrowings, which are more recent; e.g. büro ‘office’ seems to be giving way to ofis ‘office’, and afiş ‘poster’ seems to be yielding to poster ‘poster’. We will have to wait and see which side will win.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Ch a pter 25

Uyghur and Uzbek, the Southeastern Turkic languages A bdu r ish i d Ya k u p

25.1 Introduction The southeastern branch of Turkic is represented by two Central Asian languages, Uyghur and Uzbek. Uyghur is spoken by about ten million people predominantly in the People’s Republic of China as first language, where the Uyghur population numbers 10,069,346. According to the 2010 census, 8,345,66 Uyghurs live in Xinjiang Uyghur Autonomous Region (XUAR), and make up 99.4% of the Uyghur population and 43.3% of the total population of Xinjiang. Nearly 9,000 Uyghurs are scattered in Taoyuan County of Hunan Province. They are the descendants of the Uyghur soldiers who had been sent by the Hongwu Emperor of the Ming dynasty to the region in 1368. Outside China, approximately 350,000 Uyghurs live in several Central Asian countries, predominantly in Kazakhstan (223,100, census 2009), Uzbekistan (55,220, census 2008), and Kyrgyzstan (49,000, census 2009). Uzbek is the largest Turkic language in Central Asia and the second largest Turkic language next to Republican Turkish with regard to the number of speakers, being spoken and written by 21 million Uzbeks in Uzbekistan alone (census 2011). Outside Uzbekistan it is also spoken by a considerable proportion of the Uzbek population in the northern regions of Afghanistan (34,000,000, census 2012), the Khatlon and Sughd regions of Tajikistan (927,000, census 2012), in the southern parts of Kyrgyzstan (836,065, census 2014), in the Chimken oblast and Zambyl region of Kazakhstan (456,997, census 2009) and in Turkmenistan (293,000, census 2016). The Uzbek population in Russia is estimated at between 300,000 (census 2010) and nearly two million. There are also some Uzbeks in China (10,569, census 2010). The self-referent of Uyghurs is uyγur (in Arabic-based Uyghur script: ‫)ئۇيغۇر‬, and the Uyghurs’ designation of their language is uyγurče or uyγur tili, both meaning ‘the Uyghur language’. Modern Uyghur has been written in a variety of scripts in just under one century: the Arabic-based Chagatai script; Cyrillic (mainly in the former Soviet Union); a modified Arabic Uyghur script called kona yėziq ‘old script’; a

Pinyin-based Latin script which is influenced by Cyrillic and called yėŋi yėziq ‘new script’. Between 2000 and 2001 a Latinbased new Uyghur alphabet was introduced for techno­ logic­al purposes, called Uyghur computer script (Uyghur: uyγur komputėr yėziqi). The ethnic name Uzbek (Uzbek: Oʻzbek, Ўзбек) goes back to the name, Özbeg, of the Uzbek ruler Özbeg-Khan (1312–41/42) of the Golden Horde. Until the 1920s, both Uyghur and Uzbek were written in Arabic script, mainly based on the late Chagatai writing system. In 1923, a modified version of Arabic was introduced in Uzbekistan and it was replaced by a Latin alphabet between 1929 and 1930, which has been replaced by the Cyrillic-based Uzbek script since the 1940s. In 1992, a Latin script was announced as the official writing system of Uzbekistan, though the Cyrillic script is still widely used. Although Latin and Cyrillic have each been introduced once for written Uyghur, the Arabic-based Uyghur script was used to write Uyghur until the 1960s, though some necessary changes have been made. In the 1960s a Chinese Pinyin-based writing system, namely yėŋi yėziq ‘new script’, was introduced. It had been used in parallel to the Arabic-based Uyghur script until the 1980s, and then was gradually replaced with the Arabicbased Uyghur script, which is now called kona yėziq ‘old script’ or uyγur ereb yėziqi ‘Uyghur Arabic script’. The Pinyin-based script is mainly used in the transcription of historical texts and in recording standard pronunciation. In order to fulfill the increasing demands of information science, in 2001 a Latin-based computer script, called uyγur komputėr yėziqi ‘Uyghur computer script’ or uyγur latin yėziqi ‘Uyghur Latin script’ was introduced. The Uyghurs in Kazakhstan and Uzbekistan have used a Cyrillic-based Uyghur script since the 1920s. The so-called Uyghur Arabic script is basically phonemic, having 8 vowel signs and 24 signs for consonants, including signs for f, ž which do not have phonemic value. An early stage of Modern Uyghur and its regional varieties were documented by Shaw (1878), Radloff (1886), Katanov (1893), Raquette (1912–14), Malov (1934, 1961) and Jarring (1946–51). Jarring’s Studien zu einer osttürkischen Lautlehre (1933) represents the first scientific description of the Uyghur

Abdurishid Yakup, Uyghur and Uzbek, the Southeastern Turkic languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Abdurishid Yakup. DOI: 10.1093/oso/9780198804628.003.0026

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Abdurishid Yakup phonology. Nadzhip’s Modern Uigur (1971) and Ahmet’s Hazirqi zam uyγur tili ‘Modern Uyghur’ (1983) present first rather systematic outlines of the Modern Uyghur literary language. Tömür’s Hazirqi zaman uyγur tili grammatikisi, ‘Modern Uyghur grammar morphology’ (1987), Stroj ujgurskogo jazyka edited by Sadvakasov (1989), Spoken Uyghur by Hahn (1991), and Xiandai Weiwueryu cankao yufa ‘A reference grammar of Modern Uyghur’ (2014) edited by Tohti are the most representative works among the large number of publications on Modern Uyghur phonology and grammar. Nadzhip’s Ujgursko-Russkij slovar’ (1968) and Uyγur tiliniŋ izahliq luγiti ‘An explanatory dictionary of the Uyghur language’ (1990–8, henceforth UTIL) edited by Yakup et al. are the best documentations of the lexicon of the Modern Uyghur literary language. Jarring’s An Eastern Turki-English dictionary (1964), Sadvakasov’s Jazyk Ujgurov Ferganskoj doliny (1–2, 1970, 1976), Osmanov’s Hazirqi Zaman uyγur tili dialektliri ‘Dialects of Modern Uyghur’ (1990), Yakup’s The Turfan dialect of Uyghur (2005), and Qasim’s Hazirqi Zaman uyγur tili dialektliri dialektliriniŋ fonetika, grammatika ve leksikisi üstide tetqiqat ‘A comparative study on the phonetics, grammar and lexicon of Modern Uyghur dialects’ (1–2, 2014) are the most representative publications concerning the regional varieties of Uyghur. Besides Polivanov’s work published in 1926, von Gabain (1945), Kononov (1960), and Sjoberg (1963) are three early grammatical overviews of the Uzbek literary language in European languages. Borovkov (1959), Uzbek tiliniŋ izåhli luγåti ‘A comprehensive dictionary of the Uzbek language’ edited by Ma’rufov (1–2, 1981), Waterson (1980), and Mamatov et al. (2008) document the lexicon of Uzbek. Olufsen (1905), Jarring (1937), Wurm (1959), outline some important regional varieties of Uzbek. Boeschoten (1998) and Bodrogligeti’s monographs (2002, 2003) might be considered the newest rather comprehensive description of the Uzbek literary language.

25.2  Historical connections: Genealogy and contact One of the linguistic features shared by Uyghur and Uzbek is the palatalization of the intervocalic consonant in the word for ‘foot’, that is ayaq. It was adaq in Old Turkic which is still preserved in Tuvan, while Yakut shows a devoiced con­son­ ant, namely ataχ in the word, and in Yellow Uyghur and Khakas one finds the alveolar fricative, e.g., azaq, azaχ. Another linguistic feature shared by Uyghur and Uzbek is the preservation of the suffix-initial uvular consonant in -GAn, e.g., barγan ‘having gone’, while in the languages in the southwestern group of Turkic languages it is dropped, e.g.,

412

Turkish varan, Azerbaijanian učan ‘having flown’. Long con­ son­ants observed in some numerals, e.g., ikki ‘two’, sekkiz ‘eight’, and use of verbal nouns in -(X)š might also be con­ sidered further common linguistic features shared by Uyghur and Uzbek. However, Uyghur shows certain unique linguistic features distinct from Uzbek, e.g., the so-called “Uyghur umlauting,” devoicing of high vowels before the voiceless stops, fricatives, and affricatives, and the use of a considerable number of copies from Chinese. Uzbek also shows some linguistic features distinct from Uyghur, e.g., the la­bial­ized back vowel å (IPA: [ɒ]), the derivative suffix -li developed from -lIG, coding of focal intra-terminality by means of -(a)yåtir plus personal markers, auxiliary verbs used to code actionality, which are used with both the converbs ending in -(I)b and the so-called vowel converbs, etc. The Chinese linguists usually identify the following three regional varieties within the Uyghur language spoken in the XUAR: (i) The Central dialect, (ii) the Khotan dialect, and (iii) the Lopnor dialect. The Central dialect is spoken by more than 90% of the Uyghur population in a wide region from Qomul in the east to the border areas of Yarkand in the south, while the Khotan dialect is distributed to the nine counties from Charqaliq to Guma in Khotan District. The Lopnor dialect is mainly spoken in Lopnor County proper, extending to some parts of Miren, Charqaliq, Sohu Rural District of Bügür, and Pohu Rural District of Korla. The basis of this classification is mainly (i) the realization of the so-called presentfuture, which consists of “{Converb in -a + tur-} + personal endings” as in bar-a-dur-men ‘I go’, and (ii) the forms of the so-called non-perfect participle, which goes back to “{Converb in -(y)A + tur- } + -GAn” as seen in bar-a-dur-γan ‘that have to be gone’. The auxiliary verb tur- (< turur) has disappeared in the Central dialect (turur > dur > ø), whilst the Khotan dialect preserves it in the form of -t(ʔ) (turur > dur > t(ʔ)) in both instances and the Lopnor dialect shows -di (turur > dur > di) in the case of (i) and du(r) (< dur < turur) in the case of parameter (ii). The parameters which serve as the basis of this classification are basically identical with the last two points suggested by Tenishev (1963). The result is also basically the same as Tenishev’s, though Chinese scholars’ identification of subgroups within the Central dialect shows some divergences from his. The Uyghur variety spoken in the Kelpin region located between Aksu and Kashghar displays the features of the Khotan dialect with respect to both parameters; meanwhile it shows some features unique to this variety of Uyghur, i.e. tense vowels, preservation of l when it is preceded by nasals. Although the Lopnor dialect is spoken by less than 1% of the Uyghur population only, it occupies a special position in the Turkic family of languages. It exhibits rather strong progressive labial assimilation spread until the fourth, sometimes even the fifth syllable, e.g., ötkörmölör ‘sifters’; strong

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

u yghur a nd uzbek, the southeaster n tur kic l a nguages progressive assimilation of alveolar fricatives, nasals, liquids, and occasionally some plosives is also impressive, e.g., yolvazzi ‘of the tiger’ (genitive of yolvaz ‘tiger’), süttü ‘the milk’ (accusative of süt ‘milk’). The so-called “Uyghur umlauting” is not observable in the traditional Lopnor dialect. Lexemes specific to the Lopnor dialect are numerous, including the preservation of a considerable number of Old Uyghur words. Lexemes copied from Mongolian make it fairly different from other regional varieties of Uyghur. However, in the last fifty years unique features of this language variety have been decreasing in favor of Standard Uyghur, following the popularization of Standard Uyghur through school education and modern media. Basically, the Uyghur varieties spoken in Semirechiye show features of the Ili variety of Uyghur, whereas the Uyghur varieties spoken in Fergana demonstrate A heavy influence of Uzbek spoken in adjacent regions, though their main linguistic features are identical to the Kashghar variety of Uyghur; see Kajdarov (1969: 263–318). Within the Uzbek varieties spoken in Uzbekistan, the southern Uzbek varieties, mainly spoken in Samarkand, Buchara, Tashkent, and Fergana, are heavily influenced by Tajik. However, the influence of Tajik is relatively weak in the Uzbek varieties spoken in the Fergana valley, Andijan, Kokand, and in adjacent villages. The so-called Kipchak Uzbek, or northern Uzbek, spoken in the southern part of Kazakhstan, shows very close linguistic features to Kazakh, therefore some scholars consider it as a variety of Kazakh (Wurm 1959: 489). The Uzbek varieties spoken in Chorasan and in the border area to Turkmenistan demonstrate fairly strong linguistic features of southwestern Turkic. The establishment of the Modern Uyghur and Uzbek languages started in the first decade of the 19th century, mainly based on the literary tradition of the so-called Chagatai, served as a literary medium for various Central Asian Turkic languages from the 15th century to the first half of the 19th century. The regional varieties of Uyghur spoken in the Xinjiang region, particularly the Uyghur varieties spoken in the Ili and Turfan regions, played crucial roles in the formation of the Modern Uyghur literary language, while the Uzbek variety spoken in Tashkent was chosen as the basis of the literary Uzbek language, though the northern varieties have played an important role in its formation. The modern literary Uzbek language incorporated elements of the Uzbek varieties spoken in Fergana, especially in its morphology. Since the establishment of the Republic of Uzbekistan, a considerable number of words inherited from Chagatai and replaced by Russian lexemes in the Soviet period have been replaced by native words, though some of them have clear Arabic and Persian origin. After the establishment of XUAR in 1955, the influence of eastern dialects spoken in Turfan and Qomul on the Uyghur literary language has increased.

Since the 1980s, the spoken Ürümči variety was identified as the basis of the standard pronunciation of Uyghur. In recent years, features of southwestern dialects have also been incorporated. Uyghur is one of the two official languages of XUAR, the other one being Mandarin Chinese, the state language of China. Uyghur also became the first language of the majority of some other ethnic groups in Xinjiang, e.g., Salar, Tatar, and Uzbek. Besides Chinese, Uyghur also functions as a regional lingua franca, thus quite a number of Han-Chinese, Hui, Kazakh, Mongolian, and Shibe residents of Xinjiang have a good command of Uyghur. On the other hand, a considerable number of Uyghurs learn and use Chinese as a first language, and the number of Uyghurs who actively use Chinese as a first and second language has been increasing. The Uyghurs in Central Asian countries are ba­sic­ally trilingual; besides Uyghur, most of them speak Russian and another official language of subsequent countries. Uzbek is the official language of Republic of Uzbekistan. However, it operates in strong competition with Russian in all fields, especially in higher education as well as in inter­ ethnic communication. Chinese is the main contact language of Uyghur, whereas Uzbek shows a strong influence of Tajik and of Russian. Arabic and Persian elements observed in Uyghur and Uzbek reflect their historical influence on the Chagatai literary language. Mongolian copies in Uyghur are also inherited from Old Uyghur and the Chagatai literary language. The Uzbek variety spoken in China shows a large number of copies from Uyghur which cover lexicon, morphology, and syntax, reflecting a heavy code-copying process. A considerable number of Chinese lexemes observed in the China variety of Uzbek has also come via Uyghur as intermediary.

25.3 Phonology 25.3.1 Consonants Uyghur has 22 consonant phonemes, /p/, /b/, /m/, /v/, /t/, /d/, /n/, /l/, /r/, /s/, /z/, /č/, /ǰ/, /š/, /y/, /k/, /g/, /ŋ/, /q/, /γ/, /χ/, /h/. In Table 25.1 the consonants in brackets have graphic representation in the Uyghur writing system but no phonemic value. The labiodental fricative f occurs only in copies from Arabic, Chinese, Persian, and Russian. In the spoken language it may be replaced with the bilabial plosive stop p, e.g., fayuen ‘court’ ( uon-naː- ‘do ten times’; ikkis ‘second’ > ikkis-teː- ‘do for the second time’; or from pronouns, e.g. tuoχ ‘what’ > tuoχ-taː- ‘do what’ (Sakha hesitative), kim ‘who/what’ > kim-neː- ‘do what’ (Dolgan hesi­ta­tive), χanna ‘where’ > χanna-laː- ‘go where’. In Dolgan, most loanverbs from Russian are integrated with the help of this suffix, e.g. zvonit-taː- < zvonit’ ‘to phone’, zamečaj-daː- < zamečat’ ‘notice’.

26.4.6.2  Nominal derivation Among the most productive nominalizing suffixes in Sakha and Dolgan is -SIt, which derives nouns referring to a person who performs an action on the object designated by the base noun, e.g. bul-čut ‘hunter’ < bult ‘prey’, ïnaχ-sït ‘milker’ < ïnaχ ‘cow’, üle-hit ‘worker’ < üle ‘work’. Among the suffixes deriving nouns from verbs both Sakha and Dolgan have -Iː, e.g. χa:y-ï: ‘prison’ < χa:y- ‘lock up’. Furthermore, in Sakha but not Dolgan the suffix -(Aː)hIn derives abstract nouns from verbs e.g. battaː- ‘to push, squeeze’ > battaːhïn ‘pressure’. Lastly, the suffix -(Aː)ččI, which expresses habitual aspect (see Section 26.4.5.3), also functions as an agent nominalizer in Sakha (but not in Dolgan), e.g. kömölöh-öːččü ‘helper’ < kömölös- ‘to help’. Sakha has copied the diminutive suffix -čAːn from Evenki, e.g. uol-čaːn-ïŋ [boy-dim-poss.2sg] ‘your little boy’. In addition, both Dolgan and Sakha have copied the emphatic-diminutive suffix -kAːn from Evenki, e.g. kïrakaːn ‘tiny’ < kïra ‘small’. Dolgan uses a combination of the two with the demonstratives bu ‘this’ and iti ‘that’, turning them into emphatic adverbs, e.g. bu-kaː-čaːn ‘just like this’ < bu ‘this’.

442

26.4.6.3  Other derivational suffixes Adjectives can be derived from nouns with the proprietive -LAːχ, e.g. öj-döːχ ‘clever’ < öj ‘mind’. The suffix -(I)msAχ derives adjectives with a meaning of loving something, having a habit, e.g. et-imseχ ‘loving meat’ < et ‘meat’, while the suffix -TAːɣI derives adjectives referring to a place from nouns and adverbs e.g. tïa-taːɣï ‘someone/something in the taiga’ < tïa ‘taiga’, ïraːχ-taːɣï ‘someone far away’ < ïraːχ ‘far’, i.e. the tsar. The most productive suffix to derive adverbs from adjectives is -LIk, e.g. üčügej-dik ‘well’ < üčügej ‘good’. Finally, the suffix -LIː derives nouns referring to ethnolinguistic groups or their languages, e.g. haχa-lïː ‘in Sakha’.

26.5  Syntax 26.5.1  The clause The basic word order in Sakha is SOV. Deviations are pos­ sible, but carry pragmatic meaning (cf. Stapert 2013: 244–6). Converbs precede the final finite verb or auxiliary, adjectives, numerals, and demonstratives precede nouns, and the basic order within noun phrases is demonstrative–numeral– adjective–noun. This holds for Dolgan as well, but there is significantly more freedom in word order, probably due to contact with Russian (Stapert 2013: 262). The particle duo marks yes/no questions, while the enclitic =Ij attaches to the predicate in content questions (20a, b). (20) a.       b.          

Dolgan (Story_Syn_Holiday_MSA_41) die-tiger    kel-te10      duo house-dat.3sg come-pst2.poss.3sg  q ‘Did he come home?’ Sakha (RaxA_275) onu min Innokentij Vlasevič-tan ïjït-a-bïn ptcl 1sg Innokentij Vlasevich-abl ask-prs-pred.1sg en  Kuoma-nï   χahan  kör-bük-kün=üj 2sg Kuoma-acc when see-pst2-pred.2sg=q ‘So I asked Innokentij Vlasevich, “when did you see Kuoma?” ’

As is common in Turkic languages ( Johanson 1998b: 41), in Sakha and Dolgan ‘being’ is expressed by the addition of predicative suffixes to the predicate nominal in the present 10   Note that -te is the contracted form of -bit-e; the 3sg simple past tense of kel- would be kel-l-e.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the north siberian tur kic l anguages tense (21) or with the inflected copula e- in the past tense. The predicative suffix is zero for 3SG, and 3PL is marked by the simple plural (cf. Table 26.4). (21)  Sakha (RaxA_251) beje-ŋ     em-čik-kin, self-poss.2sg medicine-nmlz-pred.2sg buol-bat      duo aux-neg.prs[pred.3sg]  q ‘. . . you yourself are a healer, aren’t you?’ In contrast to what is found in Turkic (Johanson 1998b: 56–7), the basic means to express ‘having’ in Sakha and Dolgan makes use of the proprietive suffix -LAːχ with the possessor cross-referenced with predicative suffixes (22a). The absence of having is expressed as in other Turkic languages, with the negative existential item suoχ/huoχ and the person who lacks being cross-referenced with possessive suffixes on the lacking entity in Sakha (22b), and on huoχ in Dolgan (22c). The common Turkic means of expressing ‘having’ with possessive-marked possessum and the ex­ist­ en­tial noun baːr is rare, though found in the speech of individual speakers. Dolgan additionally makes use of constructions in which dative-marked pronominal possessors combine with the existential noun; this is rare in Sakha. A further means to express ‘having’ in Sakha is with the suffix -LAn (22d), which arguably grammaticalized out of the common verbalizer -LAː and the reflexive suffix -(I)n. (22) a      

Sakha (Chir_114) oɣonńor-doːχ-χun duo old.man-propr-pred.2sg q ‘Do you have a husband?’

b. Sakha (Chir_116)   huoχ  oɣonńor-um   huoχ   neg   old.man-poss.1sg  neg   ‘No, I don’t have a husband.’ c.      

Dolgan (Story_tundra_Syn_PPK_15) ah-a     huoχ-put food-poss.3sg  neg-1pl ‘We don’t have food.’

d.          

Sakha (RaxA_271) če,   üčügej-dik  olor,   die-t-e, [. . .] ptcl good-adv  sit[imp.sg] say-pst1-poss.3sg elbeχ  oɣo-lon much child–acquis[imp.sg] ‘Well, live well, he said . . . have lots of children,. . .’

26.5.2  The nominal group It is in the case system that Sakha and Dolgan most notably differ from their Turkic relatives (cf. Pakendorf 2007: 94–207). The most salient differences concern the lack of genitive and locative cases and the existence of the partitive and comparative cases as well as a separate comitative case. In contrast to other Turkic languages, not only subjects of main clauses and generic indefinite direct objects, but also subjects of relative clauses and possessors stand in the unmarked nominative case. A remnant of the old Turkic genitive is found, however, in sequences of possessive NPs with a 3sg possessor (Pakendorf 2007: 99). The accusative case marks definite and specific indefinite direct objects, while the partitive marks partially affected or indefinite direct objects in the imperative mood. This case is more widespread in Dolgan than in Sakha, occurring even with verbs in the indicative mood (23). (23)  Dolgan (Story_Syn_Hunt_SEK_07) onton ke atïn istoria-ta keps-iem then contr other story.R-part tell-fut.1sg ‘And then I will tell another story.’ Unlike other contemporary Turkic languages, the dative in Sakha and Dolgan conflates dative, locative, and allative functions and marks recipients, addressees of speech, stative location, and the direction of movement. The ablative marks the source of motion as well as a temporal starting point. In contrast to other “Transeurasian” languages, it does not mark the standard of comparison, since this function is carried by the dedicated comparative case. The instrumental marks instruments, means, adverbials of time, manner, cause, and space and is distinct from the comitative, which in Sakha is used to express a joint action of two animate referents of equal standing as well as the inclusive coordination of two NPs. This case is very rare in Dolgan, which instead makes heavy use of the postposition gïtta ‘with’. There is no plural agreement after numerals, e.g. bies oɣo [five child] ‘five children’, alta kïːs [six girl] ‘six girls’, although there is plural agreement after nouns marked with the associative plural (24). Adjectives do not agree with their head nouns in either case or number. (24)  Dolgan (Story_Syn_Holiday_MSA_23) Nata-laːχ huoχ e-t-iler be-pst1-pred.3pl Natasha-propr neg ‘Natasha and her family weren’t there.’ As is common in the “Altaic” languages, the possessum agrees with the possessor in number and person. These

443

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Brigit te Pak endor f and Eugénie Stapert ­ ossessive suffixes also function to cross-reference the subp ject in certain TAM forms (see Section 26.4.5). There is no formal distinction in the third-person between a single possessor with plural possessum (‘his horses’ at-tar-a [horsepl-poss.3sg]), a plural possessor with a single possessum (‘their horse’ at-tara [horse-poss.3pl]), or a plural possessor with plural possessum (‘their horses’ at-tara [horse-pl.poss.3pl]). There are no morphological means to mark topic or focus in Sakha or Dolgan. Rather, a variety of particles are used, such as buollaɣïna or ke (the latter only in Dolgan, cf. 6a or 23), or the converb of the generic verb of speech dien (cf. Matić and Pakendorf 2013: 383).

26.5.3  The verbal group Predicates agree with their subjects in number and person, but as mentioned above there is no plural agreement after numeral-marked NPs. The overt causee of causatives derived from intransitive verbs stands in the accusative case. Causatives derived from transitive verbs rarely occur with an overt causee in oral speech; but where they do, this mostly carries accusative marking (25), although dative marking also occurs. (25)  Sakha (XatR_126) onu  kenniki   manna  ostuoruja-tïn ptcl afterward here  history.R-acc.3sg ïl-an    balïːha   arχïːba-tïn take-seq.cvb hospital.R archive.R-acc.3sg ïrït-tar-bïp-pït scrutinize-caus-pst2-1pl ‘Afterward in order to take his (medical) history we made the hospital archives scrutinize (everything), . . ..’ The agent of passive constructions is nearly always omitted, but where present it carries dative case-marking (26). (26)  Dolgan (Story_Xet_Family_SNB_09) kogda min oččuguj er-dek-pine kihi-ler-ge be-temp-cond.1sg human-pl-dat when.R 1sg tiny iːt-illi-bit-im raise-pass-pst2-poss.1sg ‘When I was small I was raised by (other) people.’ In the oral corpora, the reflexive has mainly “reflexivepossessive” or “reflexive-benefactive” meanings (cf. V.  Nedjalkov 2003: 39–41) and hardly any simple reflexive meanings. This might be due to the nature of the corpora, which cover a rather restricted set of genres. The reciprocal functions as a true reciprocal, but also has “assistive,” “comitative,” and “sociative” meanings (cf. V. Nedjalkov 2003:

444

30–2). Verbal predicates are mostly negated with a negative suffix (cf. Section 26.4.5); nominal constituents are negated with suoχ (cf. 22b).

26.5.4  Complex sentences There are various means to coordinate NP constituents (cf. Pakendorf 2007: 177–9 and Stapert 2013: 276–93 for examples): with the proprietive suffix -LAːχ; with the postposition kïtta/gïtta ‘with’ (this is the predominant means in Dolgan); with the conjunction uonna ‘and’ (Sakha only) or onton ‘and’ (Dolgan only); with the comitative case if two human subjects are being coordinated; with repetition of the plural suffix when proper nouns are being coordinated, or with repetition of ikki ‘two’—though this construction is archaic. Main clauses (i.e. clauses with independent finite verbs) can be coordinated asyndetically, via mere juxtaposition, or with overt conjunctions, such as uonna in Sakha and onton in Dolgan (cf. Stapert 2013: 274–7). A further frequent means of clausal coordination is by way of converbs—sometimes extending to quite long chains (e.g. Pakendorf 2007: 277, ex.114)—with only one final finite verb (27). (27)  Sakha (Mat1_033) onton üöreχ-pin bïraɣ-an kel-en then studies-acc.1sg throw-seq.cvb come-seq.cvb bar-am-mïn    χattaːn    üören-em-min  seq-seq.cvb-pred.1sg repeat.seq.cvb learn-seq.cvb-pred.1sg vsjö.taki    medučilise-ni after.all.R  medical.institute.R-acc büt-er-bit-im end[intr]-caus-pst2-poss.1sg ‘Then after I gave up studying I came here, but then went back and finished (at) the medical institute after all.’ As is common in the Transeurasian languages, sub­or­din­ ation in Sakha and Dolgan largely makes use of nonfinite predicates. However, temporal clauses can be expressed as finite clauses, with the predicate carrying the temporalconditional mood marker (itself originally a participle) plus dedicated subject agreement suffixes. When the subject of the temporal clause is not coreferential with the subject of the main clause, the subordinate predicate is expressed by dative-marked present or past participles, with possessive suffixes referencing the person and number of the sub­or­ din­ate clause subject. In contrast, when the predicate of the main and the temporal clause are coreferential, the sub­or­ din­ate predicate is expressed by sequential or simultaneous converbs. The sequential converb also expresses the predicate of causal clauses.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

the north siberian tur kic l anguages As in other Transeurasian languages, the predicate of complement clauses in both Sakha and Dolgan is an accusativemarked participle, with possessive suffixes indexing the subject of the complement clause. Purpose constructions in Sakha and Dolgan are expressed by the dedicated purposive converb -AːrI (for coreferential purpose clauses) or accusative- and dative-marked participles (cf. Stapert 2013: 297–9). There are two further purpose constructions, however, in which the languages differ: in Sakha the converb of the generic verb of speech dien is commonly used to mark purpose clauses, as is widespread in Turkic languages (cf. Matić and Pakendorf 2013: 376–7). This construction is entirely lacking in Dolgan; instead, more than half the instances of purpose clauses in the corpus are expressed with the Russian subordinator čtoby in addition to the a­ ccusative-marked participle (Stapert 2013: 300–1). Although all the Transeurasian languages make use of prenominal participial relative clauses, there is substantial variation as to the locus and extent of person and case agreement across languages (cf. Pakendorf 2012 for details). Sakha and Dolgan belong to the type called “head-marked” in Pakendorf (2012): in non-subject relative clauses the subject of the relative clause is indexed on the head noun with possessive suffixes (28). (28)  Dolgan (Story_Syn_Syndassko_PPK_28) onton  tup-put    die-lere    baːr then  build-pst.ptcp house-poss.3pl  exist e-t-e be-pst1-poss.3sg ‘Then there was the house that they built.’ In Sakha, relativization on a wide range of constituents is possible, comprising subject, direct object, oblique, possessor, and locative. In Dolgan, in contrast, non-subject relative clauses occur rarely in spontaneous speech, and possessive person marking does not always agree with the number and person of the subject of the relative clause. Instead,

3SG marking appears to be used as a default (Stapert 2013: 311–15).

26.6  Lexicon Sakha and Dolgan share a large number of cognates with other Turkic languages, especially in the basic lexicon. Nevertheless, both languages have adopted very many lexical copies, mainly from Mongolic and Russian, and to a lesser extent from Evenki (Kałużyński 1962; Pakendorf and Novgorodov 2009a; Romanova et al. 1975; Stapert 2013: 117–63). The Mongolic copies in Sakha are found not only in ­cultural domains such as livestock terminology and terms pertaining to law and order, but even in basic lexicon such as kinship and body-part terminology (Pakendorf and Novgorodov 2009b). Like the other minority languages of Siberia, Sakha has copied extensively from Russian, especially with respect to terms denoting a modern, Europeanized lifestyle. However, even some terms denoting natural phenomena entered the Sakha lexicon from Russian (Pakendorf 2007: 296–7; Pakendorf and Novgorodov 2009b). Interestingly, notwithstanding the evidence for structural influence from Evenki on Sakha (Pakendorf 2007: 303–5), the lexical impact of Evenki on Sakha is rather small, with only 35 items of Evenki origin being found in literary Sakha (Romanova et al. 1975: 163–6). Sakha dialects, however, have adopted more copies from Evenki, especially those spoken in areas where Sakha people have adopted reindeer herding. Compared to Sakha, Dolgan has undergone considerable semantic changes in its lexicon, especially in the domains of bodypart and kinship terminology, as well as having copied more items from Russian and Evenki (Stapert 2013: 128–9). Some of the semantic changes in Dolgan are likely to have occurred under influence from Evenki, such as the extension of the word ulluŋ ‘sole’ to also denote ‘foot’, or the restructuring of kinship terminology (Stapert 2013: 131–42).

445

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 27

Chuvash and the Bulgharic languages A l e x a n de r S av e ly e v

27.1 Introduction Chuvash is the only modern representative of the Bulgharic branch, which is considered the earliest to split from the Proto-Turkic language. The traditional term “Bulghar(ic)” is preferred in this chapter to the label “Oghur(ic)” that has been used in recent decades for the same grouping by some western scholars (Golden 1992: 17; Golden 1998: 18; Clark 1998a: 434; Johanson, this volume: Chapter 8).1 The name “Chuvash” is an adaptation of the endonym čə̂ovaš, and simi­lar designations of the Chuvash people and their language are used by all their neighbors. Another term, vet’ke ‘the Chuvash’, ved’eń ‘Chuvash (adj)’ of unclear but certainly early origin (Napol’skikh 2015: 67) survives in Erzya Mordvin. Chuvash is spoken primarily in the middle Volga River basin (the Volga–Kama region) in east-central European Russia. About half of the 1.5 million Chuvash population (figures for 2010) inhabit the Republic of Chuvashia, where their language has an official status. Smaller Chuvash-speaking groups traditionally reside in neighboring provinces of the Volga–Kama region, particularly republics of Tatarstan and Bashkortostan, as well as Samara, Ulyanovsk, Orenburg, Saratov, and Penza oblasts (see Figure 27.1). The number of Chuvash speakers was estimated as 1.04 million by the 2010 census, which makes it one of the largest minority languages in Russia. That said, it is a highly vulnerable language because of the dominant position of Russian in the region. From 2002 to 2010, Chuvash lost about a quarter of its speakers. Recent sociolinguistic studies (Alòs i Font 2015, 2016) indicate very rapid and extensive language shift in Chuvashia, not to mention Chuvash-populated areas outside the republic (Dolgova et al. 2004: 78–81). The extinct relatives of Chuvash, the Bulghar varieties— referred to as “West Old Turkic” by Róna-Tas and Berta (2011)—were spoken mainly in East and South European steppelands from the mid-first to the early second millennium ad. They are known from very scarce and sometimes uncertain 1   I use the label “Bulgharic” as a classificatory term for one of the two major branches of Turkic, while “Bulghar” is applied only to those extinct Bulgharic varieties that are associated with the historical Bulghar tribes.

written sources, and much better on account of numerous borrowings in the neighboring languages (see Section 27.2). Proper Chuvash has been attested in Latin and Cyrillic scripts since the early 18th century; for overviews of prestandard Chuvash sources, see Jegorov (1949), Sergejev (2004), and Savelyev (2014). A standardized Cyrillic-based writing system was created in the early 1870s by the Christian missionary and educator Ivan Yakovlev; with only minor modifications, it has survived to this day. Chuvash is one of the best-studied Turkic languages, with its first descriptive grammar issued as early as the mid-18th century (Sočinenija 1769). The cornerstone of both synchronic and historical Chuvash studies are the works by Nikolaj Ašmarin, who demonstrated close affinities between modern Chuvash and medieval Bulghar varieties (1902), prepared detailed monographs on Chuvash grammar in general (1898) and syntax in particular (1903, 1923) and, finally, published the colossal 17-volume Thesaurus Linguae Tschuvaschorum (1928–50). General surveys of Chuvash as seen in a broader Turkic and Altaic context were made by Ramstedt (1922), Poppe (1925), Jegorov (1954/1971), and Fedotov (1980). Levitskaja (1976) and Mudrak (1993, 2002a) provided comprehensive studies on the evolution of Chuvash phonology and morphology. There are two ­etymological dictionaries of Chuvash, although both are somewhat outdated (Jegorov 1964; Fedotov 1996). More upto-date etymologies for the inherited vocabulary of Chuvash can be found in Starostin et al. (2003), while Scherner (1977), Rédei and Róna-Tas (1982), Róna-Tas (1982), and Agyagási (2005) provide research into different layers of Chuvash borrowed vocabulary. The studies focused on the loans from Chuvash and extinct Bulgharic varieties into various other languages are, inter alia, Wichmann (1903), Gombocz (1912), Räsänen (1920), Dobrodomov (1974), Fedotov (1990), Dybo (2010), and Róna-Tas and Berta (2011). The system of Chuvash dialects was depicted by Kanjukova (1965), Sergejev (2007), and the group of authors behind the five-volume Materialy po čuvašskoj dialektologii edition (Pavlov et al. ­1960–1997). Krueger (1961) remains the only detailed description of Chuvash grammar available in English. The recently published grammars of Chuvash that need to be mentioned are

Alexander Savelyev, Chuvash and the Bulgharic languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Alexander Savelyev. DOI: 10.1093/oso/9780198804628.003.0028

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages

Figure 27.1  The geographical distribution of Chuvash in the Volga–Kama region

Sergejev et al. (2012) (in Chuvash) and Pavlov (2014; 2017) (in Russian).

27.2  Historical connections: Genealogy and contact Early scholarship from the 18th century associated Chuvash with the Uralic languages, being unable to disentangle complicated areal phenomena in the Volga-Kama region (see, e.g., language groupings in Pallas 1787–89). The Turkic origin of Chuvash was proposed no later than by Klaproth in 1828 and convincingly proved by Schott in 1841. In 1863, Feizkhanov managed to read three grave inscriptions in the Volga Bulghar language based on his knowledge of the con­ tem­por­ary Chuvash. Strong arguments relating Chuvash to

Volga Bulghar were summarized by Ašmarin in 1902; since then, the Volga Bulghar → Chuvash linguistic continuity has gained general acceptance in the field. Together with its extinct relatives, Chuvash forms the separate Bulgharic branch of the Turkic family, which exhibits many differences from the so-called Common Turkic languages. To take the phonological level alone, Bulgharic is characterized by such striking features as rhotacism (PTk *ř > Chu. r vs. CTk *z), lambdacism (PTk *ɬ > Chu. l vs. CTk *š in the ­syllable-final position), palatalization of dental consonants (PTk *t, *d > Chu. č, PTk *s > Chu. š before PTk *i, *ï and diphthongs), and retention of Proto-Turkic falling diphthongs in the word-initial position (PTk *i ̯V > Chu. yV-), see (Dybo 2010: 83). Because of the complicated phonological changes over the history of the Bulgharic branch, it was long assumed that there were, in fact, no regular cor­ respondences between Chuvash and Common Turkic.

447

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev Although contradicting one of the basic assumptions of the historical comparative approach, this view enjoys some popularity even today. For a set of regular, even if rather complex, correspondences between the two major branches of Turkic, see Mudrak (1989, 1993, 2002a: 677–706), Dybo and Mudrak (2006: 9–227), Dybo (2007: 10–64). It is generally agreed that the homeland of the Late ProtoTurkic language, the common ancestor of Proto-Bulgharic and Proto-Common Turkic, was located in the eastern part of the Eurasian steppes (see Robbeets et al., this volume: Chapter 43 for the details). The split of Proto-Bulgharic from Proto-Turkic can be dated approximately to between 2500 and 2000 years ago, with quantitative approaches ­pointing to around 100–0 bc (Dybo 2007: 66; Mudrak 2009: 181; Savelyev, this volume: Chapter  9). The earliest history of Proto-Bulgharic speakers is a matter of long-standing discussion; in particular, the question of whether ProtoBulgharic was a major lexical donor for other languages in the region divides scholars into two camps. The first view stems from the Proto-Turkic reconstruction that essentially favors Common Turkic data as opposed to the evidence from the Bulgharic branch, and it comes along with a ­skeptical eye on the Turko-Mongolic genealogical relationship. The proponents of this viewpoint consider Proto-Bulgharic as a primary source of non-Common Turkic loanwords found in such families as Mongolic and Samoyedic (e.g., Janhunen 2010: 293). The other view is based on paying special attention to Chuvash evidence in reconstructing Proto-Turkic and cor­ relates with a pro-Altaicist position. In this perspective, it is Proto-Turkic rather than Proto-Bulgharic that is regarded as the source for the first layer of Turkic borrowings in ProtoSamoyedic (Dybo 2007: 135–54), while ancient lexical parallels between Turkic and Mongolic are explained by inheritance and—in some interpretations, e.g., Savelyev (2017: 136–7)— partly as Proto-Turkic loanwords in Proto-Mongolic. The coming of the Bulgharic-speaking nomadic tribes to the European steppes is commonly associated with the Hunnic or immediately post-Hunnic period. While “Hunnic,” i.e. the language of the European Huns of the late fourth and fifth centuries ad, is sometimes referred to as a Bulgharic variety, this connection is based mainly on historical rather than proper linguistic evidence. Extremely scarce remnants of “Hunnic”—a number of personal names and only a few common nouns—do not provide any conclusive proof and may actually be linguistically heterogeneous. Soon after the collapse of the Hunnic Empire in the mid-450s, different tribal confederations previously unattested in the European sources came on the scene; although being presumably Bulgharic-speaking, few of them left a real linguistic trace. One of the historical groups for which a Bulgharic affiliation has been widely accepted are the Avars, who arrived in the Ponto-Caspian steppe zone in the 550s, then conquered the

448

Carpathian Basin and stayed there until the Hungarian conquest in 895–902 (Golden 1992: 108–12; Róna-Tas and Berta 2011: 36–8). Avar is known primarily from a few titles and personal names as well as from a short inscription in Greek letters. For its recent reading based on Chuvash data and for an overall interpretation of the Avar language (or at least one of the languages spoken by the Avars) as a Bulgharic variety, see Mudrak (2005a). Similarly, little is known on the language of the Khazars, who dominated the Pontic steppe and the Northern Caucasus region from the late seventh to the tenth century. The list of Khazar titles, proper nouns, and rare common nouns displays, however, some apparently Bulgharic features, as in the classic example of Šar-kel. That is the name of a Khazar fortress in the lower reaches of the Don, referred to as Бѣла Вѣжа ‘White Tower’ in Old Russian sources and inevitably linked to the Chuvash words šurə̂ ‘white’ (< VBulg. *šårї, cf. CTk *sārїɣ ‘yellow’) and kil ‘home’ (< ‘house’; the word is only scarcely attested in Common Turkic). The proper Bulghar dialects have their roots in the shortlived empire established by Kubrat in the 630s. This polity, known as Great Bulgharia to Greek sources, is traditionally localized along the Kuban River in the Northwest Caucasus region and in the Eastern Pontic steppe zone (Golden 1992: 244). Kubrat’s empire dissolved soon after his death c. 650 and the Bulghars split into five groups, each led by one of Kubrat’s sons. Two of these groups are of particular interest from a historical linguistic perspective. The first, led by Asparukh, moved westward and established the First Bulgharian empire in the Danube River basin c. 679. The Turkic-speaking elite ruled over the kingdom for several centuries until they were assimilated by the Slavs. The Danube Bulghar language is poorly attested in titles, proper names, and several short texts written in the Cyrillic and Greek scripts; what is more informative, it is regarded as a major source of early Turkic borrowings in Hungarian and  South Slavic varieties, including Old Church Slavonic. Another group of the Bulghars moved northward and reached the Volga-Kama region in the eighth century (Zimonyi 1990: 180; Kazakov 1997: 36). The Volga Bulghars were first dom­ in­ated by the Khazar Khaganate, but they achieved independence by the mid-tenth century. Their empire gained control over vast territories along the Volga and Kama Rivers, with its metropole in present-day southern Tatarstan. The language of the Volga Bulghars is limitedly attested in epitaph inscriptions in the Arabic script dating from the 13th–14th centuries (Erdal 1993). These records clearly indicate Volga Bulghar as a Chuvash-type Turkic language, cf. VBulg. sekir ‘eight’ > Chu. sakə̂r ‘eight’ vs. CTk *sekiz, VBulg. tüweti- > Chu. tə̂ovatə̂ ‘four’ vs. CTk *dȫrt, etc. Volga Bulghar is also known owing to borrowings in the neighboring Uralic—most notably, Permic—languages (Róna-Tas 1988:

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages 760–6). In addition, it served as a lexical donor for Old Russian (Dybo 2010: 123–125; Agyagási 2016) and as a substratum for later Volga Kipchak dialects (Róna-Tas 1976). A language with a phonological system transitional from Volga Bulghar to the contemporary Chuvash was the source for the main layer of Bulgharic borrowings in Mari (RónaTas 1988: 768–71; Savelyev 2018). Internal relations between the extinct members of the Bulgharic branch are problematic, and the ongoing attempts to classify them (cf. Robbeets 2015: 11) are based primarily on historical sources, not linguistic evidence per se. It should, however, be kept in mind that the names attested in early medieval chronicles, such as “Avar” or “Khazar,” referred to multilingual polities, not to linguistically monolithic groups, and Bulgharic varieties labeled differently because of historical considerations may, in fact, have been very similar to one another. Judging by the scarce data available, there are relatively minor differences between the extinct Bulgharic languages, for example between Danubian Bulghar and Volga Bulghar (Mudrak 2005b; Dybo 2010). The Chuvash language is generally regarded as an immediate—and only—descendant of Volga Bulghar. The replacement of the name of the people and their language was probably caused by the cataclysmic events that befell the Volga-Kama region in the 13th–14th centuries: the Volga Bulgharian empire was destroyed in the Mongol invasion of 1236, after which its former domains were devastated by a series of Russian military campaigns, by the civil war in Golden Horde in 1359–80, and by Timur in the 1390s. This period was marked by extensive migration of Bulgharicspeaking groups to the forest area in present-day Chuvashia, where they got into interaction with the Mari people. It is controversial to what extent the contact with the Uralic speakers, along with the long evolution of the Bulgharic branch independently from the other Turkic languages, may have contributed to the peculiarities distinguishing Chuvash from Common Turkic. For different in­ter­pret­ ations and conclusions, see Agyagási (1998), Johanson (2000a), Mudrak (2002a). The basic dialectal division of Chuvash is between Viryal (Upper) and Anatri (Lower) dialects. The former occupies the northwest of Chuvashia, and the latter is spoken in the southeast. Outside of Chuvashia, the Anatri features are dominant, though there are Viryal-speaking areas as well. The split between the two dialects, which are mutually completely intelligible, can be provisionally dated to the late 16th–early 17th centuries. Overall, Viryal is a more archaic variety. The key innovation that defines the Anatri dialect is the change of Proto-Chu. *o > u (see Section 27.3.2). Other dialectal varieties debated in the literature are Anat Yenči (Middle-Lower) and Poškărt (Northwestern). Anat Yenči is spoken in the northeastern part of Chuvasia and is described

by some Chuvash scholars as a separate dialect. Yet, it is more often interpreted as a group of mixed or transitional varieties between Viryal and Anatri (Kanjukova 1965; Sergejev 2007). The highly idiosyncratic nature of the Poškărt variety, spoken at the northwestern extremity of Chuvashia, has been traditionally attributed to a Mari substrate imposed on Viryal speakers (Ašmarin 1898). However, it was recognized as a separate—and in some respects most archaic—variety of Chuvash in later studies (Mudrak 1993; Savelyev 2018). Over the last thousand years, Volga Bulghar and Chuvash have played a key role in interactions within the VolgaKama Sprachbund, the  term referring to extensive and recurrent contact ­relations primarily between Finno-Ugric (Mari, Permic, and Mordvin) and Turkic (Bulgharic and Volga Kipchak) languages of the region (Bereczki 2005; Agyagási 2012). Chuvash was an important source of superstratum influence for the Finno-Ugric languages, but was itself severely affected by Volga Tatar. Conversely, evidence of substratum influences can be traced from Finno-Ugric to Chuvash and Tatar, and from Chuvash to Tatar (see also ­earlier in this section for contacts of Volga Bulghar and Section 27.6 for borrowed lexicon in Chuvash).

27.3 Phonology 27.3.1 Consonants The native Chuvash consonant inventory includes 17 phon­ emes represented in Table 27.1. All non-palatal consonants are realized as palatalized (to different extents, depending on the specific consonant) in a front-vowel context, cf.: par- [par-] ‘to give’ vs. per- [p’εr’-] ‘to throw, launch’. As can be seen from Table 27.1, there is no opposition between voiced and unvoiced stops, which can be reconstructed to Proto-Turkic (Dybo and Mudrak 2006) and is widespread in the Transeurasian area in general (Robbeets 2017c: 592). However, all non-sonorant consonants have semivoiced vs. voiceless allophones.2 The semi-voiced variants occur intervocalically, which refers to the position between two vowels or after a sonorant before a vowel. The voiceless allophones appear in all other positions, i.e. word-initially, before a sonorant, before or after a non-sonorant consonant, or word-finally. There are also voiceless gemin­ates that are frequent in the intervocalic position, particularly as an emphatic element within a non-verbal stem (e.g., in numerals 2   My transcription does not mark this allophonic alternation, for the sake of simplicity.

449

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev Table 27.1  Consonant phonemes  

Labial Alveodental Postalveolar Palatal Velar

Plosive

p

t

 

t́

k

Fricative

 

s

š

ś

x

Affricate

 

 

č

 

 

Nasal

m

n

 

ń

 

Approximant

v

 

 

y

 

Trill

 

r

 

 

 

Lateral

 

l

 

ĺ

 

in a substantive use, see Section 27.4.3, and in kinship terms as well as onomatopoeia) or at morpheme boundaries in nominal inflection (see Section 27.3.4). A distribution of voiced vs. voiceless obstruents resembling the Chuvash system was historically present in the neighboring Mari language (Kangasmaa-Minn 1998: 222), and it is likely that Chuvash has lost the original voiced/voiceless distinction under the Mari substratum influence. Word-initial sonorants are rare in Turkic languages and can hardly be reconstructed for Proto-Turkic. Chuvash develops /m-/ from PTk *b before an old nasal, e.g., mə̂oyə̂or ‘horn’ < PTk *buyŋuř—a process that is common for all Turkic languages except for most Oghuz varieties (Dybo and Mudrak 2006: 148–9). The initial /v-/ derives from Proto-Turkic long rounded vowels, e.g., var ‘inside’ < PTk *ȫř. The initial /l-/ is still extremely rare in native Chuvash words, though it occurs in a few inherited stems owing to the elision of the preceding vowel, e.g., lar- ‘to sit (down)’ < PTk *olur-. The initial /y-/ comes from the glide element of Proto-Turkic diphthongs, e.g., yal ‘village, society’ < PTk *i ̯el. The sonorants /r/, /n/, /ń/, and /ĺ/ occur word-initially in borrowed lexemes only.

27.3.2 Vowels The native Chuvash vowel system is presented in Table 27.2.

Chuvash distinguishes full and reduced vowels, the latter being described as the unrounded /ə/ and /ə̂/ for the standard language. However, the Viryal dialect and—as I showed earlier in Savelyev (2013) and Savelyev (2015) in contradiction to the previous descriptions—many Anatri varieties maintain the old opposition between rounded and unrounded reduced vowels, cf. kər- ‘to enter’ vs. kəorü ‘son-in-law’, pə̂rax‘to throw’ vs. pə̂orụ ‘calf ’. The contrast between full and reduced vowels is a dis­ tinct­ive feature of the Volga-Kama Sprachbund (Róna-Tas 1986; Tumaševa 1988). It applies across genealogical boundaries, being present in Tatar, Bashkir, Mari, and Mordvin dialects. Although it was assumed that the Chuvash system had been copied from Mari (Agyagási 1998; Johanson 2000a), the direction of influence may well be the opposite (see, e.g., Bereczki 1994; Aikio 2014: 130–131 against Mari as the source of the contrast in the Volga-Kama region). The Viryal dialect has /o/, which corresponds to /u/ in the Anatri dialect and the standard language. Both Viryal and Anatri vowel systems include the specific tense vowel /ụ/, which is rarely mentioned in the literature but clearly contrasts with /o/ and /u/ respectively, cf. Viryal, Anatri utə̂ ‘hay’ vs. Viryal otə̂, Anatri utə̂ ‘island’. The sources of /o/ ~ /u/ correspondence in the modern Chuvash are VBulg. *o (< PTk *o, *ō) and VBulg. *å (< PTk *a, *ā). The latter change illustrates another instance of converged evolution in the Volga-Kama region as the labialization of

Table 27.2  Vowel phonemes  

Front

 

Unrounded

Rounded

Close

i

ü

Mid

e

 

Open

450

Mid-front

 

Mid-back

Unrounded

Rounded

  ə

əo  

Back

Rounded

Rounded

Unrounded

Rounded

 

 

ï

u  ụ

ə̂

ə̂o

 

(o)

 

a

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages the ancestral *a is a common feature in the local languages, cf.: PTk *a, *ā> Kazan Tatar å, PUr *a > Proto-Mari *å (East o, West a) (Aikio 2014: 140). It is likely that this change has been induced by late Volga Bulghar/early Chuvash in both Mari and—owing to substratum interference—Tatar (Čečenov 2002: 225; Dybo and Mudrak 2006: 224). Chuvash exhibits front vs. back vowel harmony. In many cases, owing to the drastic phonological changes, it has back-vowel stems where Proto-Turkic has front-vowel harmony, and vice versa, cf.: Chu. kur- ‘to see’ < PTk *gör-, Chu. pürne ‘finger’ < PTk *bi a̯ rŋak. Vowel harmony applies within stems and across morpheme boundaries. Most inflectional and derivational suffixes have two allomorphs in accordance with the stem, with /a/ – /e/, /ə̂/ – /ə/, and /ụ/ – /ü/ being in contrast (hereinafter represented by the capital A, Ə, and U respectively). The high vowel /i/ is neutral to vowel harmony in the sense that it can occur after either a front or a back vowel in the previous syllable, cf. pəott-i [pregnant-nmlz] ‘the pregnant one’ vs. pə̂ott-i [porridge-poss.3] ‘(his) porridge’. Violation of palatal harmony in a back-vowel environment, caused by an adjacent palatal consonant, is frequent in the Anatri dialect and in the standard language, while Viryal often retains the original phonetic structure, cf. Viryal pot́ak vs. Anatri, standard Chu. puté k ‘lamb’. There are some restrictions applying to the distribution of vowels. The high /ï/ appears in the first syllable only; however, in many dialectal varieties, it can occur in suffixes as the back-vowel counterpart of /i/: Viryal (the speech of Xĕrlĕ Čutaj) varr-ï [middle-poss.3] ‘(its) middle’. Viryal /o/ ~ Anatri /u/ as used in native words also appear in  the first syllable only, though this restriction does not apply to /ụ/. There is no phonological opposition between unrounded and rounded reduced vowels in non-initial syllables; reduced vowels may become labialized after a rounded vowel in the preceding syllable or near a labial consonant.

27.3.3  Syllable and word structure The basic syllable structure is (C)V(C)(C). As is common in Transeurasian languages, Chuvash does not favor con­son­ ant clusters syllable-initially. However, they can appear in spontaneous speech owing to the elision of the first-syllable reduced vowel, e.g., /ə̂śta/ = [əɕta ~ ɕta] ‘where’, /pərre/ = [p’ёr’(:)ɛ ~ pr’ɛ] ‘one’. Sonorant-obstruent sequences are the only consonant cluster type that is allowed syllable-finally; they occur rather frequently in verbal stems (typically at the boundary of a primary verbal root and a non-productive suffix, e.g., vïr-t- ‘to lie (down)’, cf. vïr-ə̂n ‘place’) and quite

rarely in native nouns (e.g., śurt ‘house’, kə̂ovayt ‘firewood; campfire’). By far the majority of verbal stems in Chuvash are monosyllabic, while synchronically underived nouns tend to be disyllabic. Reduced vowels never occur at the end of verbal stems. At the same time, around a quarter of native nonverbal stems have the structure (C)VC(С)Ə, where the final reduced vowel may disappear owing to prosodic or syntactic factors.

27.3.4 Morphophonology Extensive morphophonological changes apply to stem-final vowels occurring before vowel-initial suffixes. In this position, /a/ and /e/ disappear: kala- ‘to speak’ > kal-in [speak-fut.3sg. conc]. Reduced vowels behave the same way, and the preceding consonant undergoes lengthening: sə̂onə̂o ‘spear’ > sə̂onn-i [spear-poss.3]. High /i/, /ụ/, and /ü/, which go back to Proto-Turkic combinations of a vowel and an approximant, prevent a hiatus in this environment by recovering their original shape, being represented as /iy/, /ə̂ov/, and /əov/ respectively, cf. śerśi ‘sparrow’ > śerśiy-əm [sparrow-poss.1sg], sụ- ‘to milk’ > sə̂ov-at-ə̂p [milk-prs-1sg], pü ‘body’ > pəov-ə [body-poss.3]. The stem-final or suffixal /t/ changes to /č/ in a number of cases when it is followed by an inflectional /ə/: śunat ‘wing’ > śunač-ə [wing-poss.3]. As the /t/ > /č/ palatalization is a historical process that has only taken place in a certain phonological context (see Section 27.2), /t/ remains intact in many other forms, e.g., śit-ə-p [achieve-fut-1sg].

27.3.5  Suprasegmentals The accent system represented in standard Chuvash is close to that of the Viryal dialect. As a general rule, the last syl­ lable with a full vowel is stressed: aván ‘good’, sərá ‘beer’, karlánkə̂ ‘throat’. In the words with reduced vowels only, the accent falls on the first syllable: xə́rlə ‘red’, və̂́orə̂om ‘long’. Most Anatri varieties represent a more innovative system where the last syllable is usually stressed, regardless of the quality of the vowel: karlankə̂́ ‘throat’, xərlə́ ‘red’. The accent system based on the opposition between etymo­logic­al­ly low and high vowels, with the latter being often reduced as in Chuvash, is also found in the Uralic languages of the Volga-Kama region, including East Mari as well as Mordvin and Permic dialects. While the ultimate origin of the system in the Volga-Kama Sprachbund is under discussion, it is likely that at least some of the local Uralic languages adopted it from Chuvash (Bereczki 1988).

451

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev

27.4 Morphology

27.4.1.2 Case

As is common in the Transeurasian area, Chuvash is a strongly agglutinative, suffixing language. Prefixal inflection is restricted to a few marginal cases only, such as the partial emphatic reduplication of adjectives and adverbs (see Section 27.4.4).

27.4.1 Nouns Chuvash nouns are inflected for number, case, and possessive suffixes. The general order of suffixes in nominal forms is possessive–number–case: tus-ə-sem-pe [friend-poss.3-pl-ins] ‘with his friends’. This fact contrasts Chuvash with the Common Turkic languages, where the order of possessive and number markers is opposite.

27.4.1.1 Number The number marking system is singular–plural. The ­singular is unmarked, and the plural is expressed with the  suffix -sem/-sen. The former allomorph occurs in the nominative and in the innovative oblique cases, while the  latter appears in the archaic oblique cases (see Section 27.4.1.2). The plural marker is not subject to vowel harmony, which reflects its late formation from an enclitic (< PTk *sayïn ‘all’). The back-vowel variant has evolved secondarily in the Viryal dialect: xola‑sam [city-pl]. The plural marker is also used for marking associative plurality: Timər-sem [‘male.name’-pl] ‘Timĕr and his relatives/friends’. There is no trace of the Common Turkic plural marker -lAr in Chuvash.

According to the local grammar tradition (e.g., N. Andrejev 1961), the Chuvash case system includes eight cases: ­nominative, genitive, dative-accusative, locative, ablative, instrumental-comitative, caritive, and causal-final. The first five are archaic cases that are continuations of Proto-Turkic cases. The new cases—instrumental-comitative, caritive, and causal-final—are formed on the basis of Proto-Turkic postpositions that functioned as enclitics in Chuvash until as late as the 18th century (see, e.g., Savelyev 2014: 254–5) or even later. The “traditional” Chuvash cases are presented and exemplified in Table 27.3. The choice between the case allomorphs is determined by the final sound of the stem. In Table 27.3, the capital C represents any stem-final consonants except for R, which stands for one of the three dental consonants r, l, n. For diachronic reasons, R-final stems take specific locative and ablative allomorphs. The capital V represents stem-final vowels except for Ә, which stands for reduced vowels. Stems ending in Ә, U, and i, as marked in the genitive or the dative-accusastive, behave morphophonologically like consonant-final stems and take a vowel-initial suffix; the rules described in Section 27.3.4 are applied here. The consonant -n- in the dativeaccusative and the locative is epenthetic; some Viryal ­varieties use -y- instead. The instrumental-comitative marker has the free variant form -pAlA. Yet another variant, ‑pAlAn, possibly formed under the influence of Tatar bilän ‘together’, occurs in the Anatri dialect only. In addition to the presented case markers, there are a number of “quasi-case” markers in the sense that they are not usually included in the case system because of their compatibility with stems of certain semantic classes only

Table 27.3  Case markers Function Marker

452

-C

-R

-V

-Ә

nom

-ø

yat xir ‘name’ ‘field’

ura śilə ‘foot, leg’ ‘udder’

gen

(-V)-n, (-C)-Әn

yat-ə̂n

xir-ən

ura-n

śill-ən

dat/acc

(-V)-nA, (-C)-A

yat-a

xir-e

ura-na

śill-e

loc

(-V)-rA, (-R)-tA, otherwise (-C)-rA

yat-ra

xir-te

ura-ra

śilə-re

abl

(-V)-rAn, (-R)-tAn, otherwise (-C)-rAn yat-ran xir-ten ura-ran

śilə-ren

ins/com

-pA

yat-pa

xir-pe

ura-pa

śilə-pe

car

-sӘr

yat-sə̂r

xir-sər

ura-sə̂r

śilə-sər

caus/fin

-šӘn

yat-šə̂n xir-šən ura-šə̂n

śilə-šən

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages or because their status as “true” case markers is otherwise debated. The approximative suffix -(n)AllA is formed on the basis of the dative-accusative marker -(n)A and the adverbial marker -llA (see Section  27.4.4): və̂orman-alla [forest-approx] ‘toward the forest’. The terminative case is expressed by the suffix -ččen: śiččə-ččen [seven-term] ‘up to seven o’clock’. The egressive suffix -RAnpa comes from a combination of the ablative -RAn and the instrumental-comitative -pA: śulla-ranpa [summer-egre] ‘since summer’. The proprietive meaning can be expressed with the suffix -(l)lӘ (see also Section 27.4.4): viśə laša-llə̂ śïn [three horse-propr person] ‘the person having three horses’. There are also a number of fossilized case markers. The suffix -(Ә)n is the marker of the old instrumental case (< PTk *-n) that is no more productive when attached to nouns, but it can still be recognized in some adverbial forms: ukśa-n [money-ins] ‘for money’. The non-productive vocative markers, which are frequent in kinship terms and personal names, include the suffixes -ụk, -ay (~ ‑ey ~ -i), and several more dialectal variants: Naśt-ụk [‘female.name’-voc] ‘Nastya (Anastacia)’, kukam-ay [mother’s.mother-voc] ‘(my) mother’s mother’.

Sources from as late as the end of the 19th century witness distribution of postconsonantal -Ә and postvocalic -U in the 2sg marker (Ašmarin 1898: 133), but the former is not usually mentioned in synchronic descriptions from the 20th century (e.g., N. Andrejev 1961: 608). The 3sg/pl and plural markers are expressed with the same suffixes, both -ə and -i going back to the Proto-Turkic third-person possessive marker *-I and, further, to the Proto-Altaic third-person personal pronoun (Schwarz et al., this volume: Chapter 32). The Proto-Turkic *-sI, which is the postvocalic variant of *-I, developed semantic specialization in Chuvash, where -(Ә)šə is used with kinship terms and names of the young of domestic animals: kukk-ə̂šə [uncle-poss.3] ‘(his) uncle’, pə̂orụ-šə [calfposs.3] ‘(his) calf’. Words for ‘mother’ and ‘father’, as well as kinship terms based on these two, have suppletive possessive forms: pre-modern Chu. *at(t)a + the historical vocative suffix -ay > *at(t)ay > Chu. atté ‘(my) father’, Chu. aśa ‘male’, dial. ‘father’ > aś-ụ [father-poss.2sg] ‘(your) father’, pre-modern Chu. *aś-ə̂šə [father-poss.3] > aššə ‘(his) father’; pre-modern Chu. *an(n)a + the vocative suffix -ay > *an(n)ay > Chu. anńe ‘(my) mother’, anń-ü [mother-poss.2sg] ‘(your) mother’, Chu. ama ‘female’, dial. ‘mother’ > am-ə̂šə [mother-poss.3] ‘(his) mother’.

27.4.1.3  Possessive suffixes The system of Chuvash possessive suffixes is presented in Table 27.4.

Table 27.4  Possessive suffixes  

SG

 

DIR

1

PL OBL

DIR

-(Ә)m

OBL -(Ә)mӘr

2

-U

-Un-

3

(-C)-ə (-V)-i

(-C)-(ə)n(-V)-in- 

-Әr (-C)-ə (-V)-i

(-C)-(ə)n(-V)-in- 

27.4.2 Pronouns The Chuvash personal pronouns are presented in Table 27.5. Etymologically, Chuvash personal pronouns have a particularly archaic shape as compared to Common Turkic languages and are most probably inherited from ProtoAltaic or even Proto-Transeurasian (for this and other Transeurasian connections for Chuvash pronouns, see Schwarz et al., this volume: Chapter  32). There is no ­opposition between exclusive and inclusive first-person pronouns. The first- and second-person pronouns are reinforced by the deictic particle e- of likely Transeurasian origin, although Adamović (1984: 6) pointed to its possible connections with the deictic e- in some languages of

Table 27.5  Personal pronouns Person SG

PL

DIR

OBL

DIR

t

1

e-pə

man-

e-pir

pir-

2

e-sə

san-

e-sir

sir-

3

və̂ol

ə̂on- [dat/acc], otherwise un-

vəosem (< və̂ol-sem [3-pl])

vəosen-

453

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev Eastern Europe, cf. Komi e-tajə ‘this’, Mord. e-te ‘this’. The direct 1pl and 2pl stems occur without the e-element in the Northwestern dialect: pẹr ‘we’, sẹr ‘you (pl)’. The Chuvash 2pl pronoun can be used for singular honorific reference, which can be a recent innovation driven by the contact with Russian. Personal pronouns inflected in the genitive function as possessive pronouns. They are commonly used to reinforce the third-person possessive suffix but rarely occur together with first- and second-person possessive suffixes: un-ə̂n all-i [3sg-gen hand-poss.3] ‘his hand’, (? man-ə̂n) all-ə̂m [1sg-gen hand-poss.1sg] ‘my hand’. The system of demonstrative pronouns as used for ­deictic reference consists of four elements expressing three degrees of distance, with proximal ku and śakə̂, medial śavə̂ and distal leš. The opposition between ku and śakə̂ seems to  reflect direct/emphatic vs. non-direct/non-emphatic deixis respectively (Ašmarin 1898; pace Sergejev 1994, interpreting the relations between the proximal demonstratives in the opposite way). All the deictic demonstratives can be used anaphorically, along with the 3sg personal pronoun və̂ol and the special anaphoric pronoun xayxi. The reflexive pronouns presented in Table  27.6 are derived from the bound stem xə̂o- (ə̂γ- in the Northwestern dialect) corresponding to the Common Turkic emphatic particle *oq ~ *uq (Ramstedt 1922: 16; Räsänen 1957: 39; Mudrak 2009: 75). The basic interrogative pronoun stems are kam ‘who?’, mən ‘what?’, and the bound stem (PTk *qay-), underlying the forms xə̂ô śan ‘when?’, xə̂̂ošə ‘which?’, and ə̂śta ‘where?’ The quantitative interrogative pronoun miśe is derived from mən ‘what?’ (cf. more archaic Northwestern and Viryal forms mənče, minče). The interrogative pronouns serve as a basis for negative and indefinite pronouns. Negative pronouns are derived with the preposing negative marker ni (ni-kam ‘nobody’, ni-mən ‘nothing’, etc.), which is borrowed from Ru. ni- with the same meaning. In indefinite pronouns, the interrogative pronoun part is preceded by the indefiniteness marker tA- (ta-kam ‘somebody’, te-mən ‘something’, etc.) ori­gin­at­ing from the disjunctive conjunction te (cf. te. . . te. . . ‘whether. . . or. . .’).

Table 27.6  Reflexive pronouns

27.4.3 Numerals The Chuvash numeral system is decimal. The numerals from ‘one’ to ‘ten’, decimals from ‘twenty’ to ‘fifty’, and names for ‘hundred’ and ‘thousand’ are synchronically non-derived (see Table 27.7). All these numerals are inherited from Proto-Turkic, except for the numeral for ‘thousand’, which is a Common Turkic borrowing in Chuvash. The numerals for ‘sixty’ and ‘seventy’ are old derivatives of ‘six’ and ‘seven’ formed with the suffix -mӘl (PTk *-mïɬ): Chu. utmə̂l ‘sixty’ (cf. ultə̂ ‘six’), śitməl ‘seventy’ (cf. śičə ‘seven’). The numerals for ‘eighty’ and ‘ninety’ are multiplicative: sakə̂r-vun(n)ə̂ [eightten] ‘80’, tə̂oxə̂or-vun(n)ə̂ [nine-ten] ‘90’. The other numerals are additive: pin=te ik-śəor vun-viśśə [thousand=coord two-hundred ten-three] ‘1213’. The cardinal numerals from ‘one’ to ‘ten’ as well as ‘fifty’ appear in two variants, depending on their syntactic function (Table 27.7). The full, or emphatic, form has a lengthened medial consonant and appears in substantive use: tə̂ovattə̂-ra [four-loc] ‘at four (o’clock)’. The short form has a single consonant instead and is used attributively: tə̂ovatə̂ yulanụt [four horseman] ‘four horsemen’. Short variants can lose their final reduced vowel in accordance with the general tendency (see Section 27.3.3), and are then termed as “clipped forms” by Clark (1998a: 442). The formal difference between attributive and substantive cardinal numerals is unusual for the Turkic languages, although Levitskaja (1976: 45-7) refers to a similar distinction, involving the long vs. short medial consonant op­pos­ ition, in the West Siberian Tatar varieties. In the Volga-Kama region, a system distinguishing between the two syntactic functions can also be found in Mari, cf. Proto-Mari *koktə > kok ‘two’ (through *kokt, attributive), koktə̂(-t) (substantive). Therefore, it cannot be excluded that the opposition in Chuvash has evolved under the Mari substratum influence. Table 27.7  Non-derived numeral lexemes  

Short

Full

 

Short

Full

1

pər

pərre

9

tə̂ xə̂ r

tə̂oxxə̂or

2

ikə, ik

ikkə

10

vunə̂, vun

vunnə̂

3

viśə, viś

viśśə

20

śirəm

4

tə̂ovatə̂, tə̂ovat

tə̂ovattə̂

30

və̂otə̂or xərəx

o

o

 

SG

PL

5

pilək

pillək

40

1

xam

xam-ə̂r

6

ultə̂, ult

ulttə̂

50

2

xụ

xə̂ v-ə̂r

7

śičə, śič

śiččə

100

śəor

3

xə̂oy

xə̂oy-sem

8

sakə̂r

sakkə̂r

1000

pin

454

o

alə̂, al

allə̂

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages Ordinal numerals are derived from the full forms of car­ dinal numerals with the suffix -məš: pillək-məš [five-ord] ‘fifth’. Collective numerals can be divided into two types. General collective numerals are formed from the full forms of cardinal numbers with the suffix -Әn, which is identical to the old instrumental case marker, and function as adverbial depictives: ultt-ə̂n kil- [six-ins come] ‘to come, being six in number’. Restrictive, or elective, collective numerals are derived with the combination of the archaic collective marker -s- (-š- for the third-person) and a plural possessive suffix. The first- and second-person forms of restrictive collective numerals are based on the clipped forms of cardinal numerals: ik-s-əmər [two-poss. 1pl] ‘the two of us’, tə̂ovat-s-ə̂r [four-poss.-2pl] ‘the four of you’. The third-person restrictive collective numerals are derived from the full forms of car­ dinal numerals: śiččə-š-ə [seven-poss. 3pl] ‘the seven of them’. Distributive numerals are derived from the clipped forms of cardinal numerals with the suffix ‑šAr < PTk *ɬar: vun-šar [ten-distr] ‘in tens’. There are no obligatory classifiers in Chuvash. One of the most common counting words is pəorčə ‘grain, granule, little part’ as used in the third-person possessive form pəorč-i [grain-poss.3]. It is associated with things consisting of numerous tiny items, such as hay, beads, peas, etc.: pər śüś pəorč-i [one hair grain-poss.3] ‘one hair’ (*pər śüś would be grammatically incorrect).

27.4.4  Property words Property words in Chuvash are adjectives and adverbs. As in other Turkic languages, many of the descriptive adjectives in Chuvash can function as nouns: nayan ‘lazy; a lazybone’, ïrə̂ ‘good; a good thing, a good spirit, etc.’. Other adjectives require a nominalizer to be used in a nominal function: šurə̂ ‘white’ > šurr-i [white-nmlz] ‘the white one’. Frequent denominal adjectival suffixes are ‑(l)lӘ with a proprietive meaning and -xi with a temporal meaning: və̂y ‘strength’ > və̂y-lə̂ ‘strong’, kaś ‘evening’ > kaś-xi ‘evening (adj)’. The comparative degree is formed with the suffix -rAx; the variant -tArAx is obligatory after r and allowable after l n: śülə-rex [high-comp] ‘higher’, yïvə̂r-tarax [heavy-comp] ‘heavier’, avan‑rax ~ avan-tarax [good-comp] ‘better’. The superlative degree is formed analytically, by adding the word či ‘most’: načar ‘bad’ > či načar ‘the worst’. The intensive forms of adjectives are derived with partial reduplication, a phenomenon that is extremely widespread in the Turkic family and in the Transeurasian languages in general. The most common model is reduplication of the word-initial CV-sequence with an additional p, e.g. xup~xura [ints~black] ‘very black, pitch-black’, təop~təottəom [ints~dark] ‘very dark, pitch-dark’. There are, though, numerous intensive

forms that deviate from the regular model, e.g. śap~śutə̂ [ints~light] ‘very light’, kə̂on~kə̂ovak [ints~blue] ‘very blue’, yəm~yešəl [ints~green] ‘very green’. Certain adjectives retain a different pattern that does not apply reduplication and relies on specific (sometimes unique) intensive markers instead: vəor‑śənə [ints-new] ‘brand new’. A large number of adverbs are formally identical to adjectives, e.g., avan ‘good; well’. The two most frequent adverbial suffixes are ‑(l)lA and -Әn. The suffix ‑(l)lA derives mainly domain adverbs from nouns and adjectives, e.g., mečək-le vï ĺa- [ball-adv play] ‘to play ball’, śənə-lle purə̂n- [new‑adv live] ‘to live in a new way’. The suffix -Әn, the old instrumental case marker (see Section  27.4.1.2), was interpreted by Ašmarin (1903: 143) as referring to a non-permanent property or pointing to subjective perception, cf. (1) and (2). (1) kun sulxə̂n tə̂or-at́ day cool stand-prs.3sg ‘The weather is cool.’ (2) kun sulxə̂n-ə̂n tə̂or-at́ day cool-adv stand-prs.3sg ‘The weather is [perceived by the speaker as] cool.’ (Ašmarin 1903: 144).

27.4.5 Verbs Chuvash verbs are inflected for mood, tense, person and number, negation, and potential form. The basic order of suffixes in a verbal form is potential form–negation–mood (tense)–person and number: kal-ay-ma-r-ə̂m [speak-pot-negpst-1sg] ‘I was not able to say’. The bare verbal stem is identical to the 2sg imperative form. Verbal inflection exhibits almost no irregularities; a rare exception are ten monosyllabic verbs that lose their final -r- in a number of forms, cf. čar- ‘to stop’ > čar-nə̂ [stop-pst.ptcp], but par- ‘to give’ > pa-nə̂ [give-pst.ptcp].

27.4.5.1  Person and number As well as other Turkic languages, Chuvash has two main paradigms of verbal person-number markers. As sum­mar­ ized by Dybo (2017: 128), paradigm I of the Turkic personnumber markers coincides to a large extent with the possessive suffixes (see Section 27.4.1.3) and occurs in the verbal forms of non-participial origin. Paradigm II is based on personal pronouns (see Section  27.4.2) used as postpositional clitics, and appears in the verbal forms that, ultimately, go back to participles. In Chuvash, paradigm I occurs in the past tenses as well as in the conditional mood, while ­paradigm II appears in the future and—with some further

455

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev Table 27.8  Verbal person-number markers Paradigm

I

 

II

 

 

SG

PL

SG

PL

1

-Әm

-ӘmӘr

-Әp

-(Ә)pӘr

2

-Әn

-Әr

-Әn

-Әr

3

-’ə

-’əś

-ə

-əś

development—the present tense as well as the concessive and the optative moods. The two paradigms of verbal personnumber markers in standard Chuvash are presented in Table 27.8. As well as Old Turkic and most Siberian Turkic languages, Chuvash copied the 1pl verbal marker from the possessive paradigm, while most of the other con­tem­por­ary Turkic languages make use of the specific 1pl marker *-UK (Mudrak 2009: 29–30; Dybo 2017: 128–9). The Chuvash development seems to represent an innovation against the background of such branches as Kipchak and Oghuz Turkic. The Chuvash second-person markers as represented in paradigm II are copied from paradigm I. Mudrak (2002a: 710–12) reconstructs the original paradigm-II second-person markers -(Ә)s (singular) and -(Ә)sӘr (plural) based on the archaic concessive forms in the Northwestern dialect. (see Section 27.4.5.2.4) He further attributes the partial merger of the two paradigms in Chuvash to an early influence of Kipchak Turkic languages, which underwent a similar process. The difference between the third-person markers in the two paradigms is that -ə in paradigm I causes palatalization of the preceding *t(> č  ), while its counterpart in paradigm II does not, cf. kur-t-ə̂m [see-pst-1sg], kur-č-ə [see-pst-3sg] vs. kur-at ́ < *kur-at-ə [see-prs-3sg]. For the specific paradigm of person-number markers in the imperative, see Section 27.4.5.2.2.

27.4.5.2  Tense and mood 27.4.5.2.1 Indicative The indicative is the only mood to have tenses, including the present, the future, and several past tenses. All tenses are expressed with finite verbs and take person-number agreement, except for the indefinite past and the indefinite pluperfect. The present tense is marked by the suffix -At- (with the cumulative -at́ / -et for 3sg and -Aśśə for 3pl) and expresses both habitual (3, 4a) and durative meanings (4b). It is sometimes referred to as the non-past tense as the same markers are used to express scheduled future events (4c).

456

tənəl-ə tavra śavrə̂n-at́ (3) śər xə̂oy-ən Earth refl.3sg-gen axis-poss.3 around rotate-prs.3sg ‘The Earth rotates on its axis.’ (4)    

epir ïyt-at-pə̂r 1pl ask-prs-1pl

a. ‘We ask.’ b. ‘We are asking.’ c. ‘We will [definitely] ask.’ The future tense expresses a hypothetical future situation (5). Benzing (1959: 740, 747) and Levitskaja (1976: 63–5) have pointed out correctly that the Chuvash future tense marker is -Ә-. That contradicts a widespread view (e.g., Jegorov 1957: 190) that Chuvash has no overt future tense marker (in this interpretation, the reduced vowel is regarded as a part of the person-number suffix). (5) epir ïyt-ə̂-pə̂r 1pl ask-fut-1pl ‘We will [probably] ask.’ The simple past tense is also referred to as the “categorical” past tense in the Chuvash grammar tradition (e.g., N. Andrejev 1961: 618). It is typically used when the speaker knows for sure that something has taken place, because he witnessed the event or for other reasons. The simple past-tense markers are -t- (-č- for the third-person) after -R-, otherwise -r- (for -R-, see Section 27.4.1.2) (6). The Northwestern dialect also uses the converb marker -sa to express the simple past tense. The simple past in Chuvash contrasts with the so-called “indefinite” past tense, which is used to express the typologically frequent combination of the ­perfect (resultative) and indirect (reported) past meanings (7, 8). The indefinite past (perfect-indirect past) tense is formed by using the past participle marker -nӘ in a finite position. (6) və̂ol kay-r-ə 3sg go-pst-3sg ‘He went (away).’ (7) və̂ol kay-nə̂ 3sg go-pst.ptcp ‘He is [apparently] gone.’ (8) veś-ex tụ-nə̂ everything-emph do-pst.ptcp ‘[Someone] has [apparently] done everything.’ = ‘Everything is (has been) done.’ The imperfect tense is marked by the suffix -att- (-atč- for the third-person): purə̂n-att-ə̂mə̂r [live-iprf-1pl] ‘we used to

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages live’. The imperfect-tense marker is historically analyzable as a combination of -At-, which is identical to the presenttense marker of ultimate participial origin, and the reflex of the Proto-Turkic defective existential verb *e(r)- inflected with the past-tense marker *-t- (Levitskaja 1976: 65). There are several ways to express pluperfect meanings. The pluperfect tense marked by -sAtt- (-sAtč- for the third-person) is referred to as the “long-past categorical” tense by the Chuvash grammarians (e.g., N. Andrejev 1961: 619) and is used in various pluperfect functions for events that were witnessed by the speaker (9). The marker -sAtt- is analyzable as a combination of the converb marker -sa and the reflex of the ProtoTurkic existential verb *e(r)- in the past tense (Levitskaja 1976: 72). In addition, different past-tense forms can be inflected with the clitic -(č)čə, which goes back to PTk *e(r)-t-i [be-pst3sg]) and functions as the so-called retrospective shift marker (Plungian and van der Auwera 2006: 344). The combination of the past participle marker -nӘ with -(č)čə is usually interpreted as expressing another pluperfect tense that indicates reported pluperfect situations and the resultative (perfect) in the past (10). The retrospective shift marker can reinforce the pluperfect forms in -sAtt-, while some dialects also allow the use of -(č)čə after the simple past and imperfect tense forms. Furthermore, the retrospective shift marker is  compatible with non-verbal predicates, such as nouns, adjectives, and the existential-possessive predicates pur ‘there is’ and śuk ‘there is no’ (11). In the modern Chuvash, the synthetic pluperfect forms are gradually replaced by the analytic construction of the main verb and the auxiliary pul- ‘to be’, both inflected with the past-participle marker -nӘ (12). (9) ə̂śta kay-ma tux-satt-ə̂n where go-inf come.out-pluprf-2sg ‘Where were you going?’ (lit. ‘Where had you come out to go?’, canceled prospective) (10) pəltər sad-a mulkač pï-nə̂-ččə come-pst.ptcp-retr last.year garden-dat/acc hare ‘Last year, a hare [apparently] came (lit. ‘had come’) to the garden.’ (reported remote past/resultative in the past) (11) ača   čux-n-e      man  ünerśə child time-poss.3-dat/acc 1sg.gen artist pul-as   əoməot  pur-ččə be-fut.ptcp dream exist-retr ‘As a child, I had a dream of becoming an artist.’ (12) pəčək Atńer škul-a vəoren-me little ‘male.name’ school-dat/acc study-inf ulttə̂-r-ax  kay-nə̂   pul-nə̂ six-loc-emph go-pst.ptcp be-pst.ptcp ‘Little Atner started school when he was as young as six years old.’

27.4.5.2.2 Imperative There are no tenses in the imperative mood. The imperative (functionally, the hortative in the first-person and the jussive in the third-person) forms are inflected with a special set of person-number markers (Table 27.9) that are not identical to the corresponding markers in the indicative and other moods. The imperative forms can be reinforced by various clitic markers. The retrospective shift marker ‑(č)čə as attached to forms in the imperative functions as a politeness marker, and the emphatic imperative markers include -xa (a discourse particle with a large number of functions), ‑xalə (< xalə ‘now’ of Arabic origin), and -sAm (linked to the 2sg conditional form PTk *-sA-ŋ in Levitskaja 1976: 77) (13). (13) kala-sam-ččə ača-m tell-emph-pol child-poss.1sg ‘Please tell me, my child!’ The prohibitive paradigm makes use of free negative markers derived from existential verbs. The first-person prohibitive forms are built on imperative forms followed by mar, a particle that is historically the negative aorist participle form of the defective existential verb PTk *e(r)- (Levitskaja 1976: 64; Mudrak 2002a: 710). For the second- and third-persons, Chuvash uses the prepositional particle an. Its ancestral form, VBulg. *en (~ *eŋ), has been traditionally interpreted as a loan from Permic *e‑n [neg.aux-2], cf. Udm., Komi en (Wichmann 1903: 148; Jegorov 1964: 26–7). The proposal is based on the fact that the Permic word is of Uralic origin (< Proto-Uralic negative auxiliary *e-), while parallels for Chu. an in other Turkic languages are scarce and somewhat unreliable (? Oghuz eŋ ~ aŋ ‘no’ attested in Maḥmūd alKāšġarī’s Dīwān). Another old argument is that preverbal particles are “an absolutely anti-Turkic feature,” as formulated recently by Manzelli (2015: 642–5). However, a borrowing of the Permic prohibitive marker into Chuvash seems to be quite unlikely. First, there are no other grammatical markers in Chuvash reliably borrowed from Permic (on the very limited Permic influence on Chuvash, which is restricted to the lexical domain only, see Section 27.6). Second, despite the lack of reliable cognates in Turkic, Chu. an may still be of Table 27.9  Imperative person-number suffixes  

SG

PL

1

-Am

-Ar

2

-∅

-Әr

3

-tӘr

-ččӘr

457

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev genuine origin as it has plausible parallels in the other Altaic branches, cf. the negative verbs PTg *e-, PMg *e- as reconstructed by Robbeets (2015: 192–202).

27.4.5.2.3 Conditional The conditional mood is typically used to indicate events that are dependent on irreal conditions as well as counterfactual situations (14). The standard Chuvash conditionalmood marker -Әtt- (-əčč- for the third-person) is analyzable as a combination of the future-tense marker -Ә- and the reflex of the Proto-Turkic existential verb *e(r)- inflected with the past-tense marker *-t- (Levitskaja 1976: 80–1). This grammaticalization pattern has been further recycled by some varieties of the Viryal dialect, which use the retrospective shift marker -(č)čə (< PTk *e(r)-t-i [be‑pst-3sg]) as attached to the future-tense forms in order to express conditionality: par-ə̂-p-čə [give-fut-1sg-retr] ‘I would give/have given’. (14) pəśer-se par-ə̂tt-ə̂m=ta śə̂onə̂ox śuk cook-cvb give-cond-1sg=emph flour neg.exist ‘I would have cooked [pancakes] for [you], but there is no flour.’ (G. V. Zotova, “Śuralnă kuna ujavlani”)

27.4.5.2.4  Concessive and optative The concessive mood can indicate subjunctive situations in addition to the proper concessive relations. Ašmarin (1898: 308–9) compared the Chuvash concessive-mood marker -in with the Erzya Mordvin ino ‘so be it’ and explained both as loans from the Russian dialectal particle in, ino ‘so be it’. This proposal requires further investigation. In standard Chuvash, the concessive-mood marker is attached to the future-tense forms only (15), but “present-concessive” and “past-concessive” forms are also attested in early records of the Chuvash dialects: pïr-at-s-ə̂n [go‑prs‑2sg-conc] ‘(even) if you go’, tiv-r-in [touch-pst-3sg.conc] ‘(even) if he had touched’. (15) xam-ə̂n var-ə̂m-ran refl.1sg-gen womb-poss.1sg-abl śural-nə̂-sker-e       par-ə̂p-in     śeś be.born-pst.ptcp-nmlz-dat/acc give-fut.1sg-conc only ‘[Will the Lord] only [be pleased] if I give those born from my own womb?’ (Micah 6: 7) In the 2sg/pl, the concessive displays an archaic feature in that it makes use of the original paradigm-II second-person markers -(Ә)s- and -(Ә)sӘr-. The concessive-mood marker -in is modified as -Әn in the 2sg—most likely, under the influence of the paradigm-I second-person marker -Әn—and is

458

omitted in the plural (but preserved as -iń/-ẹn in the Northwestern dialect) (Mudrak 2002a: 711). The optative mood is based on the retrospective shift marker -(č)čə, cliticized to the concessive-mood form. Forms in the optative are typically used in subordinate clauses to indicate wishful irreal conditions (16). (16) ex

Ilempi man kin-əm ‘female.name’ 1sg.gen daughter.in.law-poss.1sg pul-in-ččə     tə̂ovan  xər-e be-fut.3sg.conc-retr own  daughter-dat/acc  yurat-nə̂    pek   yurat-ə̂tt-ə̂m love-pst.ptcp  alike love-cond-1sg ‘Oh if only Ilempi had been my daughter-in-law, I would have loved [her] as I loved my own daughter!’ (Je. G. Selivanova, “Ilempi”) intj

In contemporary Chuvash, the synthetic concessive and optative forms are extremely rarely used. Similar functions are performed by analytical constructions with different forms of the verb pul- ‘to be’, including pul-in [be-fut.3sg. conc] ‘so be it; although’, pul-san [be-cond.cvb] ‘if’, pul‑san=ta [be-cond.cvb=emph] ‘even if’.

27.4.5.3 Negation Finite verbs make use of the negative marker -mA- in all tenses of the indicative and in all other moods except for the imperative, see Section 27.4.5.2.2 for the details. In the present and imperfect tenses, negation is marked by the suffix -mAs-, which is historically the negative form of the future participle marker -as, cf.: vïĺa-ma-r-ə̂m [play-neg-pst-1sg] ‘I did not play’, vïĺa-mas-t-ə̂p [play-neg-prs-1sg] ‘I do not play’. Non-finite verb forms, see Section 27.4.5.6, make use of non-verbal negators, i.e., the negative particle mar (< PTk *e(r)-m-er [be-neg-aor.ptcp]), the negative existential-possessive predicate śuk, and the caritive case marker -sӘr, cf.: parə̂nmalla mar [give.up-nec.ptcp neg.ptcl] ‘one should not give up’, kur-as śuk [see-fut.ptcp neg.exist] ‘one is not supposed to see’, vaska-ma-sə̂r [hurry-inf-car] ‘without hurry’.

27.4.5.4  Potential form Both affirmative and negative verb forms can be inflected with the specific potential form marker ‑Ay-: pïr-ay-at-pə̂r [come-pot-prs-1pl] ‘we are able to go’, kil-ey-me-r-ə [come‑ pot‑neg‑pst-3sg] ‘he was not able to come’. The affirmative potential form is rarely used in the modern Chuvash, while the negative form is more frequent. In addition, analytical constructions with pul-tar- [be-caus] ‘to be able, be possible’ can be used to mark physical capability to do something or

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages epistemic possibility: kə̂otart-ma pultar‑at-ə̂p [show-inf be.ableprs-1sg] ‘I am able to show’. The synthetic potential marker may occur in such constructions as well: pul-ma pultar-aymas-t́ [be-inf be.possible‑pot‑neg-prs.3sg] ‘that cannot happen, that is impossible’.

27.4.5.5 Aspect Chuvash does not have aspect as a separate grammatical category. Some aspectual concepts are covered by tense marking, e.g., the imperfect and the pluperfect. In rare cases, aspect can be marked by specific suffixes, such as the attenuative-iterative marker -kAlA: vula- ‘to read’ > vulakala‘to read a little from time to time’. As well as in other Turkic languages, in Chuvash aspect is most commonly expressed with analytical, in particular, converb constructions, e.g., üs-se pïr- [grow-cvb go] ‘to be growing’ (gradative), kan-sa il[rest-cvb take] ‘to take a rest for a while’ (delimitative), etc. For a recent comprehensive study of the functional domain of aspect/aktionsart in Chuvash, see Lebedev (2016).

27.4.5.6  Nonfinite verbs The system of nonfinite (nominal) verb forms includes infinitives, participles, and converbs. The infinitive markers are -mA < PTk *-maq and -mAškӘn (< pre-modern Chu. *-mAkšӘn, the form in PTk *-maq inflected with the causal-final case marker). The two infinitive markers are mostly interchangeable, but -mAškƏn has a more specific focus on supine functions. The most common participle forms, all of genuine origin, are the past participle in -nӘ, the present participle in -AkAn, the habitual participle in -An, the future participle in -as, and the necessitive participle in -mAllA (the infinitive form in -mA inflected with the adverbial -llA). The less frequent participle markers include the potential -i, the simulative -Anśi, and the sufficientive -mAlӘx. Participles can be used as both attributes and predicates, cf.: kay-malla śul [go-nec.ptcp road] ‘the road that one has to go along’, un kay-malla [3sg. gen go-nec.ptcp] ‘he has to go’. Most participle forms make use of the nominalizer -i to derive verbal nouns that often function as predicates of subordinate clauses (17). The most common converb in -sA expresses simultaneity (17), sequentiality, or manner of action. Other frequent converb markers include the simultaneous -A (occurs typically in aspectual converb constructions and, as a fossilized ­suffix, in postpositions), the temporal/conditional -sAn ~ -sAssӘn, and the terminative -iččen (the potential participle form in -i inflected with the terminative case marker).

(17) Kuĺuk pər šarla-ma-sə̂r kəneke ‘female.name’ one make.noise-inf-car book vula-n-in-e       itle-se   lar-sa read-pst.ptcp-nmlz-dat/acc listen-cvb sit-cvb čə̂olxa   śïx-at́  stocking knit-prs.3sg ‘Without making a sound, Kuĺuk is sitting and knitting stockings and listening to [a friend of her] read a book.’ (M. D. Trubina, “Xĕllexi kaşsenče”)

27.5 Syntax Most syntactic features of Chuvash are shared with Turkic and Transeurasian languages, which does not exclude some internal and contact-induced innovations. As early as a ­century ago Ašmarin (1903: IV–V) noticed that some of the syntactic patterns that occur in Chuvash, especially in oral spontaneous speech, were affected by or even copied from Russian. Uralic influence on the Chuvash syntax is also more than probable in view of long-term and tight contact ­relations between the Uralic and Turkic languages in the Volga-Kama region, but this question requires further investigation. Possible inter-family influence from Tatar, which has significantly affected Chuvash in general, is difficult to detect because of the original close resemblance between the Chuvash and Tatar syntactic patterns.

27.5.1  The nominal group Chuvash noun phrases are normally head-final, i.e. all types  of modifiers precede their head. Modifiers do not agree in number or case with head nouns. The basic order of  modifiers within noun phrases is case-marked nominal modifier–demonstrative–relative clause–quantifier–classifying adjective–qualitative adjective. Chuvash makes use of several kinds of noun–noun combinations that are prototypical for Turkic languages. Genitive constructions (N-gen N-poss.3) are possessive noun phrases marking the possessor as referential: laša-n xür-i [horse-gen tail-poss.3] ‘the tail of the horse’. In contrast to that, possessive compounds (N-∅ N-poss.3) refer to such a relationship of possession where the possessor is non-referential: laša xür-i [horse tail-poss.3] ‘the horse tail’. Possessive compounds can be also used appositively: Atə̂l šïv-ə [‘river.name’ riverposs.3] ‘the river Volga’. In a number of cases, nominal combinations can be formed by two nouns in juxtaposition (N-∅ N-∅), the first one functioning as an adjective, e.g., when the dependent noun denotes the material the head noun is made of: čul śurt [stone house] ‘stone house’.

459

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev In apposition, the proper name typically precedes the single-word appellative but follows the appellative expressed by a noun phrase: Almuš patša [‘male.name’ king] ‘the king Almuš’, pə̂olxar patš-i Almuš [Bulghar king-poss.3 ‘male.name’] ‘the Bulghar king Almuš’. As in other Turkic languages, the head noun does not bear plural marking when preceded by a quantifying modifier: ultə̂ xəś [six sword] ‘six swords’, numay tə̂ošman [many enemy] ‘many enemies’. Some noun phrases with a quantifying or pronominal modifier in Chuvash deviate from the standard Turkic ­patterns. The determiner pur=te [exist=emph] ‘all’ as used together with a personal pronoun takes the final position: vəosem pur=te [3pl exist=emph] ‘they all’. The reflexive pronoun xə̂o- is often pleonastically preceded by a personal pronoun: və̂ol xə̂j [3sg refl.3sg] ‘he himself’. Both cases can presumably be attributed to late Russian influence, cf. Ru. oni vse [3pl all.pl] ‘they all’, on sam [3sg refl] ‘he himself’. Forms with possessive suffixes inflected for the first- and second-persons (see Section 27.4.1.3) are infrequent in con­ tem­por­ary Chuvash (particularly in the Viryal dialect). Analytical constructions with possessive pronouns are used instead: san-ə̂n laša [2sg-gen horse] ‘your horse’. A similar tendency has been attested in the neighboring Mari language (Galkin 1964: 71). Synchronically, non-third-person possessive suffixes in Chuvash express a sort of inalienable possession relation. In the second-person, they are typically attached to nouns denoting body parts or kinship: all-ụ [hand‑poss.2sg] ‘your (sg) hand’, yə̂mə̂k-ə̂r [younger.sisterposs.2pl] ‘your (pl) younger sister’. Using the first-person possessive suffixes means, in addition, that the speaker exhibits a deep emotional connection with the possessed entity: pičče-m [older.brother-poss.1sg] ‘my (beloved) older brother’, śəršïv-ə̂mə̂r [land-poss.1pl] ‘our (beloved) motherland’, puś-ə̂m [head‑poss.1sg] ‘my (poor) head’. As in other Turkic languages, direct objects with a low referential status remain unmarked, i.e. expressed with the default nominative form, while definite and specific indefinite direct objects are marked by the dative-accusative: kəneke vula- [book read] ‘to read a book’, kəneke-ne vula[book‑dat/acc read] ‘to read the book’. Along with the proper accusative and dative functions, the dative-accusative also marks destination: Muskav-a an=kuś [‘city.name’-dat/acc neg=move.imp.2sg] ‘do not move to Moscow!’ Besides, the dative-accusative occurs in the expressions of purpose and various temporal concepts, and it can also function as dativus ethicus. The genitive not only indicates the possessor in possessive constructions but also marks the subject in some modal expressions (18). The locative and the ablative are used in their proper functions for both location and time. As in many Turkic and Transeurasian languages, the ablative is also used in comparative constructions: yur-tan šurə̂ śeśke

460

[snow-abl white bloom] ‘the bloom that is whiter than snow’. The instrumental-comitative is used not only for means of action and accompaniment but also to indicate reason, among other semantic tasks: xïrə̂m-ə pit vïś-n-i-pe [belly‑poss.3 very be.hungry-pst.ptcp-nmlz-ins/com] ‘because of being very hungry’. The caritive expresses the absence of something: yə̂nə̂š-sə̂r [mistake-car] ‘without mistakes’. In addition to reason and purpose, the causal-final case can also indicate benefactive relations: ača-sem-šən uyav [child-pl-caus/fin ­festival] ‘children’s festival’ (lit. ‘the festival for children’). (18) un-ə̂n śiy-es kil-et 3sg-gen eat-fut.ptcp come-prs.3sg ‘He wants to eat.’ Chuvash makes much use of constructions with postpositions, including in places where other Turkic languages ­prefer synthetic case-marking. The choice between case suffixing or postpositional strategy in Chuvash depends on various factors. For example, postpositions, not case suffixes, are used in personal locational constructions, i.e. when spatial relations refer to an animate referent rather than to a place: Chu. yal-a kil [village-dat/acc come.imp.2sg] ‘come to the village!’ vs. man pat-a kil [1sg.gen vicinity-dat/acc come. imp.2sg] ‘come to me!’, cf. Turkish bana gel [1sg.dat come. imp.2sg]. Ašmarin (1903: 20) questioned whether the extensive use of postpositions in Chuvash might have been driven by contact with the Indo-European languages, which tend to use analytic declension (with prepositions) in order to express case relations. Within a noun phrase, modifiers are typically coordinated by the clitic =tA, which is widely spread across the Transeurasian languages: Chu. xərlə=te šurə̂ čeček-sem [red=coord white flower‑pl] ‘red and white flowers’. The basic way to coordinate two noun phrases is via the instrumental-comitative suffix -pA: vatə̂ śïn-pa viləm [old person-ins/com death] ‘the old man and death’ (the name of a fairy tale). Noun phrases can also be coordinated by =tA, as well as by the free coordinator tata (a reduplication of the former).

27.5.2  The verbal group Most topics related to the verbal group syntax were covered in Section 27.4.5 and other sections of the chapter; in this section, I discuss specifically verbal valencies. The number of arguments controlled by a predicate may vary from zero to three, and the valency of the verb is changed by voice suffixes. The valency can be reduced by the markers -l- and -n-, that have passive, automative, or reflexive functions (19, 20). Clauses with verbs inflected with these markers may include a reason of action or an instrument (both

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages expressed with the instrumental-comitative case), but never an agent (Salo 2013: 233) (21). (19) ača śural-at́ child be.born-prs.3sg ‘The child is being born.’ (20) kalav śə̂omə̂ol vulan-at́ short.story easily be.read-prs.3sg ‘The short story is read with ease’. (21) yə̂və̂ś-sem śil-pe avkalan-aśśə wind-ins/com bend-prs.3pl tree-pl ‘The trees are bending in the wind.’ (Ašmarin 1928–50, I: 36) The highly productive causative marker -(t)tAr and the non-productive causative suffixes ‑(Ә)t, ‑(Ә)r, ‑Ar, -At (all distributed according to the historical morphophonological context, see Mudrak 2014) increase the valency of the verb by adding the causer argument, the causee being expressed in the dative-accusative (22). (22) śak temə̂-pa man-a universsittet-ra this topic-ins 1sg-dat/acc university-loc Iraida.Gennadyevna  soččineni   śïrtar-č-ə ‘female.name’   composition make.write-pst-3sg ‘Iraida Gennadyevna made me write a composition on this topic at the university.’ (“Čăvaš xalăx sajčĕ”)

27.5.3  The clause The basic word order is Subject–Object–Verb (23). The Object– Subject–Verb structure occurs when the subject takes up the focus position (24). The word order Object–Verb–Subject is common when the subject is a non-emphatic pronoun used as clitic (25). Other combinations are also possible in spontaneous speech, folklore, and literature. (23) epə san-a yurat-at-ə̂p 1sg 2sg-dat/acc love-prs-1sg ‘I love you.’ (24) Yakur-a Petər śap-r-ə Yakur-dat/acc Petӗr hit-pst-3sg ‘It is Petӗr who hit Yakur.’ (Ašmarin 1903: 90) (25) ə̂on-a pasar-ta kur-t-ə̂m epə 3sg-dat/acc market-loc see-pst-1sg 1sg ‘I saw him on the market.’ (Ašmarin 1903: 90) Existential and possessive predication constructions make use of nominal predicates, the affirmative pur and the negative

śuk. The former can be omitted in many cases. Both pur and śuk can be inflected with the retrospective shift marker -(č)čə to express existential-possessive relations in the past (11). In addition, all non-present tenses make great use of the verb pul- ‘to be, exist’ in ‘being’- and ‘having’- constructions. The possessor in predicative possession constructions is normally marked by the genitive case marker (11). Temporary possession is expressed with the locative construction (26). The possessor is cross-referenced on the possessee (the ­subject of the predicative possession clause) via possessive suffixes or not, depending on a number of factors such as alienable/inalienable distinction, referentiality, and dialectal variation (Ašmarin 1903: 119–21). (26) un-ə̂n ukś-i aššən-če 3sg-gen money-poss.3 father.poss.3-loc ‘His money is with his father.’ (Ašmarin 1903: 286) Untypically for the Turkic languages, Chuvash does not mark nominal predicates by person-number suffixes: epə čə̂ovaš [1sg Chuvash] ‘I am Chuvash’, čüreče uśə̂ [window open] ‘the window is open’. Clausal negation in Chuvash makes use of both suffixal and analytical negators, see Section 27.4.5.3 for the details. Chuvash does not have special morphological means for marking topic or focus. The topic/focus relations are typically marked by word order. Yes/no and alternative questions are marked with the clitic =i attached to the predicate (27). Some Anatri varieties use the interrogative particle =A (28) instead. The latter has been compared to Udm. =a since Wichmann (1903: 148), but the exact relations between the dialectal Chuvash and the Permic interrogative marker remain unclear. There are several other interrogative markers in Chuvash with more ­specific functions, including the emphatic interrogative =ši and the mirative =im. (27) śakə̂n-t-i śəršïv san-a kə̂omə̂ol-a this-loc-nmlz land 2sg-dat/acc mood-dat/acc kil-et=i     kil-mes-t=i come-prs.3sg=q come-neg-prs.3sg=q ‘Do you like the local area or not?’ (Ašmarin 1928–50, XIII: 100) (28) es šïv-a kə-me kay-r-ə̂n=a 2sg water-dat/acc enter-inf go-pst-2sg=q ‘Did you go swimming?’ (Ašmarin 1928–50, I: 3) As is common in Turkic and other Transeurasian languages (see Malchukov and Czerwinski, this volume: Chapter 35), clause subordination in Chuvash makes great use of subordinate clauses based on nonfinite predicates. According to the general left-branching tendency, subordinate clauses

461

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev precede the independent clause. The order of constituents in embedded clauses remains basically intact as compared to the equivalent finite constructions. Nor is there specific strategy for case-marking on the embedded predicate arguments, i.e. they are marked by the same case as they would be in the equivalent independent clause. In this respect, Chuvash deviates from the standard Turkic model with a Transeurasian background, where the embedded subject is commonly marked with an oblique case. Complement clauses are primarily based on participles, along with the existential nominals pur and śuk. Examples (29), (30), and (31) illustrate complement clauses taking up the position of the subject, the direct object, and the oblique object. Most types of complement clauses require further nominalization of the nonfinite predicate, commonly by means of the suffix -i. The nominalizer originates from the third-person possessive suffix -i, but it is used regardless of the subject deictic reference. Complement clauses marked with the first- or second-person possessive marker occur only in archaic constructions that are of very limited use in modern Chuvash (32). (29) [pir-ən  un-pa    kiləš-mell-i]     śeś 1sg-gen 3sg-ins/com agree-nec.ptcp-nmlz only yul-č-ə remain-pst-3sg ‘It only remained for us to agree with him.’ (I. Andreev 1961: 211) tə̂ompay sə̂omax=ta (30) [čə̂ovaš-sen Chuvash-pl.gen ‘tămpaj’ word=emph purr-in-e]     pəl-m-eśśə=ši exist-nmlz-dat/acc know-neg-prs.3pl=emph.q ‘Don’t they really know that there is the word tămpaj in Chuvash?’ (A. P. Kazanov, “Tămsaj je tămpaj”) (31) [śemye-ren uyrə̂m purə̂n-n-in-če] nim family-abl separate live-pst.ptcp-nmlz-loc nothing layə̂xx-i=te    śuk good-nmlz=emph  neg.exist ‘There is no good in living separately from one’s family.’ (“Xïpar” newspaper, 03.11.2010) (32) [man san-a temən čul sə̂omax 1sg.gen 2sg-dat/acc some that.much word kal-ass-ə̂m]      pur say-fut.ptcp-poss.1sg  exist ‘I have a lot to tell you.’ (I. Andreev 1961: 166) Adverbial clauses are based on converbs or nominalized participles in oblique cases (33) and express modifiers of the main clause predicate. Nonfinite verb forms with different semantics, see Section 27.4.5.6, take up the position of the

462

embedded predicate and form adverbial clauses of manner, purpose, reason, condition, etc. In addition, Chuvash makes extensive use of analytic constructions based on a participial form followed by a postposition (34). (33) [ep  mən-šən    kil-n-i-pe] 1sg what-caus/fin come-pst.ptcp-nmlz-ins/com čuxlan  kə̂štax think  a.bit ‘Think for a bit about why I came here.’ (A. A. Tarasov, “Sutnă pürtri julaški kaś”) (34) Vańuška kaĺĺex [niməś-sem və̂orman tə̂orə̂ox along ‘male.name’ again German-pl forest yïtə̂-sem-pe   śüre-me  pultar-ass-i     śinčen] dog-pl-ins/com walk-inf be.able-fut.ptcp-nmlz about šuxə̂šla-r-ə think-pst-3sg ‘Vańuška thought once again that the Germans might have been walking with dogs around the woods.’ (L. Ja. Agakov, “Ïltăn văčăra”) Relative clauses are based on non-nominalized participles and take up the prenominal slot as modifiers of the main clause argument: numay kalaś-akan śïn [much talk-prs.ptcp person] ‘the person that talks [too] much’. Among several patterns of non-subject relativization found in Turkic and other Transeurasian languages (see Malchukov and Czerwinski, this volume: Chapter 35), Chuvash uses that called “attributive clause” in Comrie (1998: 254) and “unmarked relative clause” in Pakendorf (2012: 274). This pattern implies that the relative clause subject is not cross-referenced on the participle, nor on the head noun. As a result, Chuvash subject and direct object relative clauses are structurally identical in that the complex clause [arslan vəoler-nə] pəl [lion kill‑pst.ptcp elephant] can be translated both as ‘the elephant that was killed by a lion’ and ‘the elephant who killed a lion’. The same pattern is applied in relativization of oblique objects and adjuncts: epə kišər pa-nə̂ śïn [1sg carrot give-pst.ptcp person] ‘the person to whom I gave a carrot’, pičče-m purə̂n-as śurt [older.brother-poss.1sg live-fut. ptcp house] ‘the house in which my older brother will live’. In using the unmarked strategy of non-subject relativization, Chuvash clusters with the Kipchak languages, including Volga Tatar (Pakendorf 2012: 264), so this may well be an areal phenomenon. Relativization of possessors of subjects and objects is another available option in Chuvash: ïvə̂l-ə əner man-a śap-nə̂ śïn [son-poss.3 yesterday 1sg-dat/acc hit-pst.ptcp person] ‘the person whose son hit me yesterday’, ïvə̂l-n-e əner epə śap-nə̂ śïn [son-poss.3-dat/acc yesterday 1sg hit-pst.ptcp person] ‘the person whose son I hit yesterday’.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

chuvash and the bulgharic l anguages Clause linkage based on the so-called “adverbial conjunctions”, see Malchukov and Czerwinski (this volume: Chapter 35), can be illustrated by the construction with the sA-converb form of the citation marker te-. This conjunction can introduce reported direct speech (35) or, owing to further grammaticalization, clauses with a purposive or causal meaning (36). Some of the other Chuvash conjunctions seem to be (half-)calqued from Russian, cf., e.g., Chu. mən-šən tesen [what-caus/fin say‑cond.cvb] and Ru. potomu čto ‘because’. vəosem śə̂ol-sa xə̂ovar-nə̂ (35) pir-ən pə̂orə̂ov-a save-cvb leave-pst.ptcp 1pl-gen calf-dat/acc 3pl te-se   kala-r-ə   vət  anńe say-cvb say-pst-3sg  emph mother ‘Mother said that they had saved our cattle.’ (V. M. Kuz’mina, “Jămra ačisem”) (36) epə  un-ran  və̂ol  Izraiĺ  patš-i    an=pul-tə̂r 1sg 3sg-abl 3sg Israel king-poss.3  neg=be-imp.3sg te-se   pə̂orə̂on-t-ə̂m say-cvb turn.away-pst-1sg ‘I have rejected him as king over Israel.’ (lit. ‘I have turned away from him saying: “Let him not be the king of Israel”!’) (1 Samuel 16: 1) As is common in Transeurasian languages, Chuvash makes use of constructions based on the simultaneous/sequential converb (-sA) in order to coordinate clauses. Long clause chains based on forms in -sA are possible, e.g., in a narrative context. Other frequent means of clause coordination include the coordinating marker =tA cliticized to the first conjunct and simple clause sequencing.  Along with the presented strategies of clause linking, Chuvash, especially in its spoken non-standard form, makes use of clauses based on relative pronouns and of other IndoEuropean-like patterns that are most likely calqued or halfcalqued from Russian.

27.6 Lexicon Most of the Chuvash vocabulary used in daily communication (around two-thirds, according to my estimate) is of ultimate Proto-Turkic origin. In the domain of basic vocabulary, the ratio of inherited lexemes reaches 90%–95%. The major differences between Chuvash and Common Turkic basic vocabularies originate in semantic shifts accumulated over their two thousand years’ evolution in parallel, cf. PTk *yǖř > Chu. śə̂ovar ‘mouth’, CTk *yǖz ‘face’ (CTk *aɣïz ‘mouth’), PTk *si ̯ārïɣ > Chu. šurə̂ ‘white’, CTk *sārïɣ ‘yellow’ (CTk *āq, *ürüŋ ‘white’), etc. The Chuvash basic vocabulary exhibits many of the semantic distinctions that are not universal cross-linguistically

but were characteristic to Proto-Turkic and can also be found in Common Turkic languages, such as the ‘arm’/‘hand’ distinction, the lexical distinction in terms for older and younger siblings (see van Berlo, this volume: Chapter 38), and the 2D/3D distinction in terms for ‘round’ and ‘thin’ (but not for ‘thick’, as in other Turkic languages). Lexical isoglosses connect Chuvash with different subgroups of Common Turkic. For example, an archaic isogloss that is known from Old Turkic and is preserved in Chuvash and South Siberian Turkic as geographically peripheral Turkic languages is the word for ‘wife, woman’: Chu. arə̂m < pre-modern Chu. *av-r-ə̂-m < VBulg. *εv-r-i-m [house-locnmlz-poss.1sg], lit. ‘the she-who-is-in-the-house of mine’, cf. Siberian Turkic ep-či, a derivative of ep ‘house’ with the same meaning (originally ‘housewife’). A remarkable number of lexical items, including those in the domain of basic vocabulary, relate Chuvash specifically to the Oghuz subgroup, cf. Chu. kə̂ovapa and Oghuz *gȫpek ‘navel’ of the same origin vs. reflexes of *kīn-dük in the other Turkic languages. Besides, shared lexical and semantic isoglosses connect Chuvash with the Volga Kipchak languages, which is apparently a result of the extensive co-evolution of Turkic languages in the Volga-Kama region. Borrowed vocabulary in Chuvash goes back to different sources. An early layer of borrowing is represented by some Old Ossetic loans from the Volga Bulgharian period or even prior to it (e.g., Chu. kəle ‘heel’). Contrary to a commonly cited misconception, the number of loanwords of substratum Uralic origin in Proto-Chuvash as revealed so far is not large at all, with a few dozen reliable borrowings from Mari (e.g., kə̂otkə̂o ‘ant’) and a dozen terms of Permic origin (e.g., pə̂ši ‘elk’). The Viryal and particularly Northwestern dialects adopted some more Mari lexemes owing to relatively recent interactions. Since the late 13th century and until the most recent times, Chuvash and particularly its Anatri dialect have been crucially affected by Tatar, with at least several hundred borrowings (e.g., akə̂š ‘swan’). In addition, Tatar served as a mediator for most of the Arabic and Persian as well Middle Mongolian loanwords in Chuvash. Some terms were borrowed into the ancestor of Chuvash directly from Persian (e.g., erńe ‘week’) and Middle Mongolian (e.g., pusaxa ‘threshold’); the presence of direct Arabic loans in Chuvash is yet to be verified. Russian has increasingly influenced Chuvash since the mid-16th century, resulting in an enormous number of borrowings belonging primarily to the semantic domains of administrative rule, society, and particularly modern life.

Acknowledgments The research leading to these results has received funding from the European Research Council (ERC) under the European

463

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A lex ander Savelyev Union’s Horizon 2020 research and innovation programme (grant agreement No 646612) granted to Martine Robbeets. I am grateful to all the colleagues who provided their comments on the first draft of this chapter. Klára Agyagási, the reviewer of the chapter, drew my attention to some less-accessible literature on contact relations of Chuvash. Artem Fedorinchyk

464

and Maria Kholodilova shared their expertise on different aspects of descriptive grammar of the contemporary Chuvash, especially on the syntax part. Ruslan Idrisov offered his feedback on my description of the Chuvash phonology. A special thanks goes to Elena Savelyeva, who created a map of the Chuvash-speaking area especially for this chapter.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

PA RT III

Comparative Overviews

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

A. Phonology

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 28

A comparative approach to the consonant inventory of the Transeurasian languages A l l a n  R. B om h ar d

28.1 Introduction Stretching across northern Eurasia, from as far as the shores of the Black and Mediterranean Seas in the west to the shores of the Pacific Ocean in the east, is a discontinuous band of languages, often assumed to have sprung from some common source—the great Transeurasian (formerly “Altaic”) language family. Current thinking among sup­porters is that the following languages are to be included in this language family (going from west to east): Turkic, Mongolic, Tungusic, Koreanic, and Japonic (Japanese-Ryūkyūan). The assumption that these languages are genetically related, however, is not without controversy, especially concerning the inclusion of Koreanic and Japonic. Indeed, the debate over whether this is a valid language family is both long and contentious. Stefan Georg and Alexander Vovin may be mentioned as two contemporary critics. As noted by Ruhlen (1987: 128): The study of the Altaic family has had a long and stormy history, and even today there is considerable disagreement among specialists over exactly which languages belong to the family. As I mentioned in Section 28.4, one often finds in the history of genetic classification that some close-knit family serves as a nucleus around which increasingly larger families are built. With Altaic, we find an unusual development in the opposite direction: the original conception of Altaic in the 19th century was considerably broader than it is today.

And also by Golden (1992: 16): The Altaic languages consist of the following groupings: Turkic, Mongolic, Manchu-Tungusic, in all likelihood significant elements of Korean and possibly one of the strata that came to constitute Japanese. The whole relationship is problematic. At present, specialists are divided as to whether the Altaic languages are the descendants of a grouping of tongues stemming from a common source (which broke up anywhere from 4000–2000 years ago or considerably earlier, according to some hypotheses) or were initially unrelated

languages that have converged as a consequence of cen­tur­ ies of borrowing and contact, i.e. areal phenomena.

And finally by Unger (1990: 481): In sum, the Altaic panel concluded that perpetuating the term “Altaic” probably does more harm than good, especially in the case of East Asia. We urge researchers to focus instead on the more likely family that, at George Grace’s suggestion, we call Macro-Tungusic, i.e., the generally recognized Tungusic languages plus Korean, Japanese, and the now extinct languages of Korea before the Silla unification. “Macro-Tungusic” may not have the exciting trans­con­tin­ en­tal ring of “Altaic,” but perhaps that is the lesson to be learned from our experience to date: language history is not a big empty canvas to be filled with broad brushstrokes and daring, sweeping lines, but a broken mosaic that must be reassembled methodically, bit by bit.

More recently, however, Unger (2014: 224–5) has taken a more affirmative view, especially concerning the possible genetic relationship between Koreanic and Japonic: The Proto-Korean-Japanese hypothesis is the best working hypothesis available. a)  It is the most parsimonious hypothesis. i) As invalid etymologies are weeded out, new matches are being found to take their place and improve the overall coherence of sound correspondences. ii) Separating out early Korean-to-Japanese loanwords removes complications in sound cor­res­pond­ences. They are few in number, readily identified, and not suggestive of intimate, long-term contact. iii) Grammatical morphemes are poor candidates for borrowing. (1)  Syntactic calquing is not observed. (2) Both languages have focus particles that are similar in function and not found in nearby SOV languages. Though the attested reflexes are not directly cognate, they can be accounted for

Allan R. Bomhard, A comparative approach to the consonant inventory of the Transeurasian languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Allan R. Bomhard. DOI: 10.1093/oso/9780198804628.003.0029

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d etymologically; focus particles thus appear to be a common innovation of Japanese and Korean. b) Tungusic at least is an indisputable language family. Whether or not there was a Transeurasian family including Turkic, Mongolic, and Tungusic, a Macro-Tungusic family, of which Proto-Koreo-Japonic was one branch, seems likely. c) If the Macro-Tungusic home area were the region around the Bóhǎi Gulf from Shāndōng to Liáoníng, the sep­ar­ ation of pre-Korean and pre-Japanese may be explained as follows… i) As Proto-Sinitic speakers moved from the TibetoBurman home area into the Huánghé valley (signaled by the spread of millet), Proto-Tungusic speakers split off and moved north and northeast, ultimately into Eastern Siberia. ii) The next group to split off consisted of pre-Korean speakers, who moved into Southern Manchuria. iii) As Sinitic speakers thrived with the advent of rice agriculture, its techniques became known to the remaining coastal dwellers, who were subsequently assimilated or migrated. Pre-Japanese speakers brought wet-field rice to the peninsula. d) Any claim that Japanese or Korean is a true isolate must be affirmatively reconciled with non-linguistic data from archaeology and human genetics.

Given the controversial nature of the Transeurasian Hypothesis, in general, it is not surprising that there are still many points of controversy regarding the reconstruction of the consonant inventory of the Transeurasian proto-language. Not only do these involve problems within Proto-Transeurasian proper, there are also problems within the individual branches. Moreover, there are currently two competing reconstructions for Proto-Transeurasian. The traditional/earlier reconstruction, advocated by Gustav John Ramstedt, Nicholas Poppe, John Street, and, most recently, Martine Robbeets, among others, posits a two-way contrast in the series of stops and affricates: (i) voiceless (aspirated) and (ii) voiced, while the reconstruction advocated mostly by Russian scholars (Vladislav  M.  Illič-Svityč, Sergej Starostin, Anna Dybo, Oleg Mudrak, Edkhiam R. Tenishev, as well as the Americans Toby D. Griffen and Allan R. Bomhard) posits a three-way contrast: (i) voiceless aspirated; (ii) voiceless unaspirated; and (iii) voiced. The Russian views developed out of Illič-Svityč’s work on Nostratic (see the next section). Finally, different cor­re­ spond­ence sets have been proposed. It should be mentioned that there has been extensive contact, going back millennia, between the various Transeurasian languages (for more information, see Vajda, this volume: Chapter 41, on contact between Old Korean and Old Japanese;

470

see Francis-Ratte and Unger, this volume: Chapter  39; see also Hudson, this volume: Chapter 47; see Robbeets, this volume: Chapter  36 on basic vocabulary, and Li, this volume: Chapter 46). The lexical borrowing that has taken place has inevitably led to the spread of phonological features (for details, see Anderson, this volume: Chapter 40), which adds further complications to the reconstruction of the con­son­ ant inventories of the parent languages of each branch, not to mention Proto-Transeurasian.

28.2  The reconstruction of the Proto-Transeurasian consonant inventory We will begin by looking at Proto-Transeurasian. The views expressed by Poppe (1960b) may be taken as representing the traditional/earlier reconstruction of the Proto-Transeurasian consonant inventory (for information on the vowel systems and vowel harmony in the Transeurasian languages, see Joseph et al., this volume: Chapter 29). Poppe (1960b) reconstructs the Proto-Transeurasian consonant inventory in Table 28.1 (where different from that used in this chapter, Poppe’s transcription is given in parentheses) (see also Ramstedt 1952–7; Robbeets 2017c: 591–2; G. Starostin 2016: 6; Street 1974: 3). Poppe reconstructs just two series of stops and affricates in Proto-Transeurasian: (i) voiceless (aspirated) (*p, *t, *č, *k) and (ii) voiced (*b, *d, *ǯ, *g). Poppe notes (1960b: 9–10) that there is a possibility that the contrast could have been between (i) voiceless aspirated and (ii) voiceless unas­pir­ ated stops and affricates instead. The stops include labial (*p, *b), dental (*t, *d), and velar (*k, *g) points of articulation, together with a series of palato-alveolar (“post-alveolar”) affricates (*č, *ǯ = [tʃ], [dʒ]). According to Poppe, neither the liquids nor the velar nasals were used word-initially, while the voiceless stops and palato-alveolar affricate were Table 28.1  Poppe’s reconstruction of the Proto-Transeurasian consonant inventory p

t

č

k

b

d

ǯ

g

n

nʸ (= ń)

-ŋ-

-r- (= -r¹-)

-rʸ- (= -r²-)

-l- (= -l¹-)

-lʸ- (= -l²-)

s m

y (= j) Poppe (1960b)

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages strongly aspirated. Poppe’s reconstruction relies heavily on Mongolic. In addition to *-l¹- and *-r¹-, which represent plain *-land *-r- respectively, Poppe (1960b: 74–92) reconstructs two additional sounds, which he writes *-l²- and *-r²-, based upon the set of correspondences represented in Tables 28.2 and 28.3. Poppe notes that these may represent palatalized variants, that is, *-lʹ- (*-ĺ-) (= *-lʸ-) and *-rʹ- (*-ŕ-) (= *-rʸ-), re­spect­ive­ly. As can be seen from Tables 28.2 and 28.3, the reconstruction of Proto-Transeurasian *-l²- and *-r²- rests critically on the evidence from (Chuvash-)Turkic, and that evidence is open to different interpretations. For example, according to Róna-Tas (1998: 71–2): In Chuvash and its earlier forms, r is found instead of z in other Turkic languages. This important feature of “rhoticism” is reflected in early Turkic loanwords in Common Samoyed, Mongolian and Hungarian. While some scholars take it to reflect a special Turkic consonant or consonant combination, there is no need for such a reconstruction. Proto-Turkic had

Table 28.2  Correspondences for *-l²Pre-Mongolic

*l² (perhaps = lʹ)

Proto-Mongolic

*l² (perhaps = lʹ)

Witten Mongolian

l

Manchu-Tungus

l

Korean

l/r

Pre-Turkic

*l² (perhaps = lʹ)

Chuvash

l

Proto-Turkic

*š

Old Turkish

š

Cf. Poppe 1960b: 77

Table 28.3  Correspondences for *-r²Proto-Mongolic

*r

Witten Mongolian

r

Manchu-Tungus

r

Korean

-r- / -l

Pre-Turkic

*r²

Chuvash

r

Proto-Turkic

*z

Old Turkish

z

Cf. Poppe 1960b: 77

the opposition z vs. r, which was lost in the ancestor of Chuvash, but preserved in all other Turkic languages. The loss was due to the change of z to r, a development that occurred twice in the history of Chuvash. The second time, it affected Proto-Turkic intervocalic and final d, which became r via an intermediate stage z. As shown by Tekin (1969, 1986) and others, z has developed into r in other Turkic languages as well. The change z > r thus began in ­special phonetic en­vir­on­ments in all Turkic languages at the beginning of late Ancient Turkic. It later ceased except in the ancestor of Chuvash, where it was generalised in all positions. In most Turkic languages, non-initial š corresponds to an l in Chuvash and its ancestors as well as in early Turkic loanwords in Common Samoyed, Mongolian and Hungarian. Scholars such as Ramstedt, Poppe and Doerfer have taken these cases of “lambdaism” to represent a special Proto-Turkic sound or sound combination. The case is similar to that of rhoticism: Chuvash and its ancestors lost the opposition š vs. l. But not even in Chuvash was the change to l complete, since ś is often found here instead of Proto-Turkic š.

In 2005, Robbeets undertook an exhaustive evaluation of the lexical evidence relating to Japonic and Koreanic, spe­cif­ic­ al­ly, and further to Turkic, Mongolic, and Tungusic, in general. She accepted only a small core of the most reliable lexical evidence, which she then used as a basis to establish sound correspondences. Among other things, her analysis supports the traditional/earlier reconstruction of the ProtoTranseurasian consonant inventory established by Ramstedt and Poppe. Of particular interest to the discussion at hand is that Robbeets rejects the reconstruction of ProtoTranseurasian *-lʸ- (= *-l²-) and *-rʸ- (= *-r²-). She maintains that *-lʸ- and *-rʸ- may ultimately go back to original con­ son­ant clusters, which developed into *-lʸ- and *-rʸ- in some cases. Russian scholars, on the other hand, generally support the reconstruction of Proto-Transeurasian *lʸ (= *l²) and *rʸ (= *r²) (cf. Starostin et al. 2003: 63–6; G. Starostin 2016: 6–7; see also Tenishev and Dybo  2001–6.3: 26–32 (for Turkic); Miller 1971: 114–53 (for Japanese). A new approach to the reconstruction of the ProtoTranseurasian consonant inventory began in the mid-1960s and early 1970s with the work of Illič-Svityč on Nostratic. Illič-Svityč replaces the traditional/earlier reconstruction, consisting of a two-way contrast in the Proto-Transeurasian series of stops and affricates, with a three-way contrast (see Table 28.4 for details) (cf. Illič-Svityč 1965 and 1971–84 I: 168–70—these tables were prepared by Vladimir Dybo). Illič-Svityč’s reconstruction is based upon his analysis of the lexical material from the Transeurasian languages as compared with the proposed cognates from the remaining Nostratic languages.

471

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d Table 28.4  Proto-Transeurasian consonant inventory Stops and affricates

pʰ- (= pʽ-)

tʰ (= tʽ)

čʰ (= čʽ)

kʰ (= kʽ)

p

t

č

k

b

d

ǯ

g

s

š

Sibilants

zNasals and liquids

m

n

nʸ (= ń)

-l-

lʸ (= ĺ)

-r-

-rʸ- (= -ŕ-)

Glides

ŋ

-y- (= -j-)

Starostin et al. 2003

Illič-Svityč’s views on Proto-Transeurasian consonantism have subsequently been followed by Russian scholars, including Sergej Starostin, Anna Dybo, and Oleg Mudrak in their massive 2003 book entitled Etymological Dictionary of the Altaic Languages. Starostin et al. (2003: 22–135) reconstruct the Proto-Transeurasian consonant system in Table 28.4 (where different from that used in this chapter, their transcription is given in parentheses) (similar systems are reconstructed for Proto-Transeurasian by Bomhard 2018 vol. 1: 245; Griffen 1994: 42–3; S. Starostin 1991: 5–24; Dybo 1996: 44 and 2007: 13; Tenishev and Dybo 2001–6 vol. 3: 9; G. Starostin 2016: 7). Unlike Poppe, Starostin et al. reconstruct three series of stops and affricates for Proto-Transeurasian, based

upon voicing and aspiration contrasts: (i) voiceless as­pir­ ated (*pʰ, *tʰ, *čʰ, *kʰ); (ii) plain (unaspirated) voiceless (*p,  *t, * č, *k); and (iii) voiced (*b, *d, *ǯ, *g). The stops include labial (*pʰ, *p, *b), dental (*tʰ, *t, *d), and velar (*kʰ, *k, *g) points of articulation, together with a series of palato-alveolar (“post-alveolar”) affricates (*čʰ, *č, *ǯ = [tʃʰ], [tʃ], [dʒ]). These are accompanied by labial, dental, palatalized, and velar nasals (*m, *n, * nʸ, * ŋ), nonpalatalized and palatalized li­quids (*-l-, *-r-; *lʸ, *-rʸ-), and a glide (*-y-). Finally, there are three sibilants (*s, *z-, *š). Starostin et al. note that *z and *y are in complementary distribution: *z occurs only in initial position, while *y is never found at the beginning of a word. Starostin et al.’s reconstruction relies heavily on Tungusic, which they consider to be the most archaic branch. It should be repeated here that Starostin et al. support the reconstruction of Proto-Transeurasian *lʸ and *-rʸ-.

28.3 Correspondences This section contains the tables of sound correspondences for the various consonant series as proposed by (i) Poppe (1960b) (Table  28.5), (ii) Starostin et al. (2003) (see also Griffen 1994: 42–7; Dybo 1996 and 2007: 13–14: 44–5; Tenishev and Dybo  2001–6 vol. 3: 10; Bomhard  2018 vol.1: 250) (Table 28.6), and (iii) Robbeets (2015: 119–20, Table 4 [medial consonants only], and 2016b: 206–7) (Table 28.7).

Table 28.5  Table of sound correspondences proposed by Poppe (1960b) PTEA

Proto-Mongolic

Middle Mongolian

Manchu / Tungusic

Korean

Proto-Turkic

p-

p-

h-

f-

p- ~ ph-

h-

t-

t-

t-

t-

t- ~ th-

t- (d-)

k-

q-/k-

q-/k-

k-/g-

k-

q-/k-

č-

č-

č-

č-

b-

b-

b-

b-

d-

d-/ǯ-

d-/ǯ-

d-/ǯ-

g-

g͔-/g-

g͔-/g-

g-/ŋ-

ǯ-

ǯ-

ǯ-

ǯ-

s-

s-/š-

s-/š-

s-

y- (= j-)

y-

y-/Ø-

y-/č-

m-

m-

m-

m-/nʸ-

m-

b-

n-

n-

n-

n-

n-

y-

472

tp-

bj-

k-

q-/-k ǯ-

s-

sy- (> ǯ-, ž-, ǯ-, sdepending upon the language)

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages PTEA

Proto-Mongolic

Middle Mongolian

Manchu / Tungusic

Korean

Proto-Turkic

nʸ- (= ń-)

ni-

ni-

nʸ-

n-

y-

VpV

b

b

f ~ p/w

-Cp-

-b-

-b-

-p-/-b ~ -p-/-w-

-b

-b-/-w-

-b-/-w-

-b- ~ -b-/-w-

-b-

-b- (strong position)

-b-

-b-

-b- ~ -w-

-b-

-b- (weak position)

-γ-/-w-/-y-

-γ-/-w-/-y-

-b-/-w-

-b-

-p- (weak position)

-γ-/-w-/-y-

-γ-/-w-/-y-

-f- ~ -p-/-w-

-p-

-d, -d-

-t, -t-

-t, -t-

-t, -t-

p -w-/-p

-p-

t (before i)

č

č

č

t

t(V)

t

t

t

t

-d, -d-

-d/-ǯ, -d-/-ǯ-

-d/-ǯ, -d-/-ǯ-

-d/-ǯ, -d-/-ǯ-

-d, -d-

-k

-g͔/-g

-g͔/-g

-q/-k

-q/-k

VkV

q/k

q/k

q/k

k

-g/-g ͔

VgV (strong position)

g/g ͔

g (weak position)

γ/g

g (before i)

y

-g

g

q/k

k (Evenki)

q/k

g/g ͔

χ/x

γ/g

γ/g

χ/x

γ/g

-g/-g ͔

-g/-g ͔

-q/-k

-γ/-g

-č-, -č

-č-, -č

-č-, -č/-š

-č-, ?

-č-, -č

-ǯ-

-ǯ-

-ǯ-

-ǯ-

-y-

-s-, -s

-s-, -s

-s-, -s

-s-

-Ø-, -t

-s-, -s

-š-

-š-

-š-

-y-, -y (= -j-, -j)

-y-/-γ-, -y

-y-/-γ-, -y

-y-

-y-

-y-, -y

-m-, -m

-m-, -m

-m-, -m

-n-, -n

-n-, -n

-n-, -n

-n-, -n

-n-, -n

-nʸ (= -ń)

-n

-ni

-nʸ

-ñ/-y

VŋV, -ŋ

ŋ

ŋ

ŋ

ŋ

-ŋg-

-ŋg- (-ŋg͔-)

-ŋg- (-ŋg͔-)

-ŋg- (-ŋg͔-)

-ŋ-

-ŋ- (weak position)

-γ-

-γ- (-g-)

-ŋ-

-ŋ-

l

l (r)

l/r

l

l/r

š

l lʸ (= l²)

lʸ

l

l

r

r

r

r (l)

rʸ (= r²)

r

r

r

r -r-/-l

z

The above table is a summary—the sound correspondences proposed by Poppe are somewhat more complicated than indicated in the above table. Poppe’s (1960b) book should be consulted for details on the developments in the individual daughter languages as well as examples illustrating those developments. See also Poppe’s (1955) book on Mongolian.

473

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d Table 28.6  Table of sound correspondences proposed by Starostin et al. (2003) PTEA

Proto-Tungusic

Proto-Mongolic

Proto-Turkic

Proto-Koreanic

Proto-Japonic

pʰ-

p-

h-/y-

Ø-/y-

p-

p-

-pʰ-

-p-

-h-/-b-, -b

-p-

-p-

-p-

tʰ-

t-

t-/č(i)-

t- [dV+lʸ rʸ r]

t-

t-

-tʰ-

-t-

-t-/-č(i)-, -d

-t-

-t-

-t-

kʰ-

x-

k-

k-

k-

k-

-kʰ-

-k-/-x-

-k-/-g[Vh]-, -g

-k-

-k-/-h-

-k-

čʰ-

č-

č-

č-

č-

t-

-čʰ-

-č-

-č-

-č-

-č-

-t-

p-

p-

b-/h-

b-

p-

p-

-p-

-b-

-b-

-b-

-p-

-p-

t-

d-/ǯ(i ̯)-

d-/č(i)-

d-

t-

t-/d[i ə]

-t-

-t-

-d-/-č(i)-

-t-

-t-/-r-

-t-

k-

k-

k-

g-

k-

k-

-k-

-k-

-g-, -g

-k-/-g[Vr]-

-Ø-/-h-, -k

-k-

č-

ǯ-

d-/č(i)-

d-

č-

t-

-č-

-s-

-č-

-č-

-č-

-s-

b-

b-

b-

b-

p-

p-/b[a, ə, Vy]

-b-

-b-

-h-/-[R]b-/b[Vg], -b

-b-

-b-, -p

-p-/[iV, y]w

d-

d-

d-/ǯ(i)-

y-

t-

d-/t[V + Cʰ]

-d-

-d-

-d-/-ǯ(i)-

-d-

-t-/-r-

-t-/[iV y]y

g-

g-

g-

g-

k-

k-

-g-

-g-

-h-/-g[Vh]-, -g

-g-

-Ø-/-h-, -k

-k-/[iV]Ø

ǯ-

ǯ-

ǯ-

y-

č-

d-

-ǯ-

-ǯ-

-ǯ-

-y-

-č-

-y-

m

m

m

b-, -m-

m

m

n

n

n

y-, n

n

n

ŋ-

ŋ-

Ø-/y-/g[u]-/n[a o e]

Ø-/y-

n-

Ø-/n-(/m[i̯]-)

-ŋ-

-ŋ-

-ŋ-/-n-/-m-/-h-

-ŋ-

-ŋ-/-Ø-

-n-/-m-

nʸ

nʸ

ǯ-, -y-/-n-

y-, -nʸ-

n-, -nʸ-

m-, -n-/-m-

-r-

-r-

-r-

-r-

-r-

-r-/-t-

-rʸ-

-r-

-r-

-rʸ-

-r-

-r-/-t[i u]-

l

l

l-/n-, -l-

y-, -l-

n-, -r-

n-, -r-

lʸ

l

d-/ǯ(i)-, -l-

y-, -lʸ-

n-, -r-

n, -s-

-y-

-y-

-y-/-h-

-y-

-y-/-Ø-

-y-/-Ø-

s

s

s

s

s-/h-, -s-

s

474

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages PTEA

Proto-Tungusic

Proto-Mongolic

Proto-Turkic

Proto-Koreanic

Proto-Japonic

z-

s-

s-

y-

s-

s-

š

š

s-/č[A]-, -s-

s-/č[A]-, -s-

s

s

As noted above, *z and *y are in complementary distribution: *z occurs only in initial position, while *y is never found at the beginning of a word.

Table 28.7  Table of sound correspondences proposed by Robbeets (2015, 2016b) Proto-Japonic

Proto-Koreanic

Proto-Tungusic

Proto-Mongolic

Proto-Turkic

PTEA

p-

p-

p-

p-

p-

p-

-p-

-p-

-p-

-ɣ-

-p-

-p-

p- /w-

p-

b-

b-

b-

b-

-p-/-w-

-p-

-b-

-b-/-ɣ-

-b-

-b-

-np-

-pC-

-PC-

-PC-

-P(C)-

-m(P)T-

-np-

-Rp-

-RP-

-RP-

-RP-

-Rp-

t-

t-

t-

t-

t-

t-

-t-

-t-

-t-

-t-

-t-

-t-

t-/y-

t- (ci-)

d- (ji-)

d- (ji-)

y-

d-

-t-/-y-

-l-

-d- (-ji-)

-d- (-ji-)

-d-

-d-

-nt-

-c-

-TC-

-TC-

-TC-

-n(T)K-

-nt-

-Rc-

-RT-

-RT-

-RT-

-Rt-

k-

k-

k-

k-

k-

k-

-k-

-k- (-h-)

-k-

-k-

-k-

-k-

k-

k-

g-

g-

k-

g-

-k-

-k- (-h-)

-g-

-g-

-g-

-g-

-nk-

-kC-

-KC-

-KC-

-KC-

-ŋ(K)T-

-nk-

-Rk-

-RK-

-RK-

-RK-

-Rk-

t-

c-

č-

č-

č-

č-

-t-

-c-

-č-

-č-

-č-

-č-

-si

-l(i)

-l(č)

-l(č)

-l(č) ~ -š

-lč

k-

k-, h-

x-

k-

k-

x-

-k-

-k-

-x-

-g- ~ -k-

-g- ~ -k-

-x-

s-

s-

s-

s-

s-

s-

-s-

-s-

-s-

-s-

-s-

-s-

m-

m-

m-

m-

b-

m-

-m-

-m-

-m-

-m-

-m-

-m-

n-

n-

n-

n-

y-

n-

-n-

-n-

-n-

-n-

-n-

-n(continued )

475

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d Table 28.7  Continued Proto-Japonic

Proto-Koreanic

Proto-Tungusic

Proto-Mongolic

Proto-Turkic

PTEA

-r-

-l-

-r-

-r-

-r-

-r-

-r-

-l-

-r-

-r-

2

-r -

-r-

-r-

-l-

-l-

-l-

-l-

-l-

This table contains correspondences for consonant clusters as well. According to Robbeets, these clusters can be divided into homorganic clusters and heterorganic clusters. A homorganic cluster is composed of a sonorant plus a stop (PTEA *-Rp-, *-Rt-, *-Rk-). They developed into nasal clusters in Proto-Japonic, which, in turn, developed into voiced stops in Old Japanese (PTEA *-Rp- > PJ *-mp- > OJ -b-; PTEA *-Rt- > PJ *-nt- > OJ -d-; PTEA *-Rk- > PJ *-ŋk- > OJ *-g-). In heterorganic clusters, on the other hand, the nasal and the stop had a different place of articulation, which resulted in the insertion of a epenthetic stop (PTEA *-m(P)T-, *-n(T)K-, *-ŋ(K)T-). The nasal was lost in Proto-Turkic, ProtoMongolic, and Proto-Tungusic (PTEA *-m(P)T-, *-n(T)K-, *-ŋ(K)T- > *-PT-, *-TK-, *-KT-), while the final stop was lost in Koreanic and Japonic. The resulting clusters developed into voiced stops in Old Japanese, this development being comparable to that of the homorganic clusters mentioned above (PTEA *-m(P)T- > PJ *-mp- > OJ -b-; PTEA *-n(T)K- > PJ *-nt- > OJ -d-; PTEA *-ŋ(K)T- > PJ *-ŋk- > OJ *-g-).

We will now sketch the early evolution of the individual branches, concentrating on how the reconstructed ProtoTranseurasian consonant system developed in each.

28.4 Turkic The Turkic languages constitute the most geographically widespread Transeurasian branch. There are some thirty Turkic languages, as well as numerous dialects, some of which are quite different from the standard/national forms of the languages in question. Chuvash is the most divergent Turkic language. Indeed, it appears likely that Proto-Turkic initially split into two branches: (i) Chuvash and (ii) all of the others. Consequently, this language family is sometimes referred to as Chuvash-Turkic. For information on the ge­nea­logic­al classification of the Turkic languages, see Johanson (this volume: Chapter 8); for information on Chuvash and the Bulgharic languages, see Savelyev (this volume: Chapter 27). The Proto-Turkic consonant system was not particularly complicated—it may be reconstructed in Table 28.8 (cf. Starostin et al. 2003: 136; Johanson 1998a: 95; Robbeets 2005: 75). According to Starostin et al. (2003), the consonant system reconstructed by them for Proto-Transeurasian underwent the following developments in Proto-Turkic. First, the initial voiced labial and velar stops were retained in Proto-Turkic (*b-, *g- > *b-, *g-), while the voiced dental stop *d- and the voiced palato-alveolar affricate *ǯ- became *y- in initial position (*d-, *ǯ- > *y-). All of the medial voiced stops were retained in Proto-Turkic (*-b-, *-d-, *-g- > *-b-, *-d-, *-g-). The medial voiced palato-alveolar affricate *-ǯ- also

476

Table 28.8  Proto-Turkic consonant inventory p

t

č

k

b

d

y (= j)

g

-n-

-nʸ- (= -ń-)

-ŋ-

-r-, -l-

-rʸ-, -lʸ(= -ŕ-, -ĺ-)

s -m-

*p, *t, *k are assumed to have been fortes and *b, *d, *g to have been lenes (cf. Róna-Tas 1998: 71; Johanson 1998a: 95; Robbeets 2005: 75). 2. Robbeets (2005: 75) does not reconstruct *lʸ for Proto-Turkic. She does, however, reconstruct all of the other sounds listed in this table, including *rʸ, which she accepts as a possible Proto-Turkic phon­ emes and which she writes *r2 (cf. Robbeets 2005: 78). 3. Tenishev and Dybo (2001–6 vol. 3: 17) reconstruct a more complicated consonant system for Proto-Turkic. 4. As noted by Robbeets (2005: 76), Proto-Turkic *k and *g had front and back allophones, depending upon the quality of adjacent vowels. These allophones later became phonemic. Cf. Menges (1968b: 81–107) for a discussion of the development of these (and other) sounds in the individual Turkic daughter languages. 1.

became *-y- (*-ǯ- > *-y-). Robbeets, however, assumes that Proto-Transeurasian initial *g- became *k- in Proto-Turkic (see also Johanson 1998a: 95–6), which seems highly probable in view of the evidence presented by her. Next, the initial plain (unaspirated) voiceless stops became voiced stops in Proto-Turkic (*p-, *t-, *k- > *b-, *d, *g-), while the plain (unaspirated) palato-alveolar affricate *č- became *d- (*č- > *d-). Robbeets, on the other hand, assumes that the initial plain (unaspirated) stops and palatoalveolar affricate were retained, except for *p-, which has disappeared in most modern Turkic languages (*p- > *h- > *Ø-).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages According to Starostin et al., the medial plain (unas­pir­ated) stops and palato-alveolar affricate were retained in ProtoTurkic (*-t-, *-k-, *-č- > *-t-, *-k-, *-č-), except for *-p-, which was voiced (*-p- > *-b-), while Robbeets assumes that all of the medial plain (unaspirated) stops and palatoalveolar stops were retained as such (*-p-, *-t-, *-k-, *-č- > *-p-, *-t-, *-k-, *-č-). Starostin et al. further assume that medial *-k- became *-g- when followed by a vowel and *r (*-k[Vr]- > *-g[Vr]-). Finally, the initial aspirated voiceless stops reconstructed by Starostin et al. for Proto-Transeurasian merged with the plain (unaspirated) voiceless stops in Proto-Turkic (*tʰ-, *kʰ-, *čʰ- > *t-, *k-, *č-), except for *pʰ-, which was lost (*pʰ- > *h- > *Ø-) (see also Robbeets  2005: 75). Medial aspirated voiceless stops, including *-pʰ-, underwent the same development (*-pʰ-, *-tʰ-, *-kʰ-, *-čʰ- > *-p-, *-t-, *-k-, *-č-). Furthermore, Starostin et al. maintain that ProtoTranseurasian *š > *čʰ (> *č) before back vowels but *s elsewhere in Proto-Turkic, while initial *nʸ- > *ǯ- (> *y-); *lʸ- > *d- (> *y-); *n-, *l- > *d- (> *y-); *m- > *b-; *ŋ- > *Ø-; and *d-, *z- > *ǯ- (> *y-). Robbeets (2005: 77) notes that the following sounds did not occur in word-initial position in Proto-Turkic: *n, *m, *ŋ, *l, *r. She further notes that it is not certain whether *nʸ (= *ń) also occurred in word-initial position, though it did occur medially. By way of comparison, it may be noted that the reconstructed Proto-Turkic consonant inventory given at the beginning of this section is very close to the consonant system of Early Old Turkic (cf. Erdal  1998: 139–40 and

Table 28.9  Old Turkic consonant inventory according to Erdal Labials Alveolars Palatals Velars Unvoiced orals

p

t

č

k

Voiced orals

v

d

y

g

s, z

š

n

nʸ (= ñ)

Sibilants Nasals

m

Liquids

ŋ

r, l

Erdal 2004: 62–85 According to Erdal, the voiced oral stops had fricative variants β (or  v), δ, γ, but were realized as stops (b, d, g) after r, l, n, and ­(partially) z.

2004: 62–85—Erdal does not include sounds found only in loanwords) (see also Robbeets  2015: 38). Table  28.9 shows Turkic consonant inventory. Menges (1968b: 81), on the other hand, reconstructs a more complex system for Common Turkic and Ancient Turkic (see also Tenishev and Dybo 2001–6 vol. 3: 17). Table 28.10 shows the Common Turkic consonant inventory. For the development of the consonants in the Turkic daughter languages, cf. the table of sound correspondences and accompanying notes (for consonants) in Starostin et al. 2003: 137–46 (see also Dybo  2007: 16–22; Johanson  1998a: 95–106; Robbeets 2005: 75–9; Róna-Tas 1998: 71–2; Tenishev and Dybo 2001–6 vol. 3: 13–16). For details on the phono­logic­al systems of the various modern Turkic languages, cf. Johanson and Csató (1998).

Table 28.10  Common Turkic consonant inventory according to Menges Occlusives Fricatives Sibilants Affricates Nasals Liquids Deep Velar

q

ɫ

(ḫ), γ

Velar Pre-palatal

ŋ k, g

Palato-alveolar

(x) š

Dental

t, d

(ð)

Labial

p, b

(f), v ?

Lingual Semivowels

č, ǯ s, z

ň n

l

m

r

j (asyllabic i ̯) w (asyllabic u̯)

The consonant inventory reconstructed by Menges represents a later stage of development. Menges (1968b: 81) mentions that all of the above phonemes are found in modern Turkic and that a few more have been added.

477

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d

28.5 Mongolic Proto-Mongolic has a relatively shallow time depth. As the ancestor of all modern Mongolian languages, it represents the language that existed at the time of the geographical dispersal of the Mongols in the 13th century ad. Related Mongolic languages/dialects that existed alongside ProtoMongolic as currently reconstructed were replaced around that time. The Proto-Mongolic consonant system is nearly identical with Middle Mongolian (cf. Starostin et al. 2003: 149; Janhunen 2003c: 6; Robbeets 2005: 72–3; Poppe 1960b: 9)—it may be reconstructed in Table 28.11. The Proto-Mongolic consonant inventory reconstructed by Starostin et al. included labial, dental, and velar points of articulation (voiceless: *t, *k; voiced: *b, *d, *g)—the voiceless labial member was missing (cf. Janhunen 2003c: 5). There were also corresponding labial, dental, and velar nasals (*m, *n, *ŋ) as well as voiceless and voiced palatoalveolar affricates (*č, *ǯ) (cf. Janhunen 2003c: 5). There was a sibilant (*s) and a glottal fricative (*h) (Janhunen [2003c: 5–6, 7] reconstructs a voiceless velar fricative *x here). Starostin et al. complete the system with *w, *r, *l, and *y. However, Janhunen does not reconstruct *w for ProtoMongolic. According to Janhunen (2003c: 10), *r and *l did not occur in word-initial position in Proto-Mongolic. Janhunen (2003c: 7) notes that *x was most likely pronounced as a laryngeal [h]. Though medial intervocalic *-x- was lost, initial *x- was regularly preserved in Written Mongolian, though it is not represented as such in the Written Mongolian writing system. Starostin et al. (2003: 80–1) propose a rather complicated series of changes leading from Proto-Transeurasian to Proto-Mongolic—these changes may be summarized as follows:

Table 28.11  Proto-Mongolic consonant inventory b

t

č

k

d

ǯ

g

s m

n

w

-r-

ŋ y (= j)

-lPoppe (1955: 96–8 and 1960b: 10–12) reconstructs *p for ProtoMongolic as does Robbeets (2005: 72—see below), while Janhunen (2003c: 6) does not.

478

First, in initial position, (1) *š- > *čʰ- before back vowels, but *š > *s in other positions; (2) *pʰ- (> *f) > *h-; *č- > *t-; *nʸ- > *ǯ- and *lʸ- > *d-, while *ŋ- > *Ø-, *n-, or *g-, depending upon the following vowel; and (3) *z- > *s-. In medial position, *-rʸ- > *-r-; *-lʸ- > *-l-; and *-nʸ- > *-n- or *-y- (distribution unclear). Next, dentals were palatalized before *i: *tʰ[i]- > *čʰ[i]-; *t[i]- > *č[i]-; and *d[i]- > *ǯ[i]-. Note: this change must have taken place after the merger of the vowels *ï and *i (*ï, *i > *i) in Proto-Mongolic (cf. Janhunen  2003c: 7). At the same time, medial *-b- > *-w- (except in clusters and before *k and *g); *-g- > *-h- (except in clusters and before *g); and *-ŋ- > *-h- (except in clusters). Then, medial plain (unaspirated) stops were voiced: *-p- > *-b-; *-t- > *-d-; and *-k- > *-g-. Note: medial *-č- remained unchanged: *-č- > *-č-. Following that, the medial voiceless aspirated stops and the voiceless aspirated palato-alveolar affricate were deaspirated: *-pʰ- > *-p-; *-tʰ- > *-t-; *-kʰ- > *-k-; *-čʰ- > *-č-. Likewise, the initial dental and velar voiceless aspirated stops and the voiceless aspirated palatoalveolar affricate were deaspirated: *tʰ- > *t-; *kʰ- > *k-; and *čʰ- > *č-. Inasmuch as Robbeets (2005: 72–3) starts with a different reconstruction for the Proto-Transeurasian consonant inventory as well as due to the complicated nature of the Proto-Mongolic developments, it is worth repeating her views on this matter (footnotes are not included here): It is clear that Proto-Mongolic had voicing distinction for its stops and affricates. An important feature at the basis of the classification of the Mongolic languages is the reflex of PMo *p-. The initial labial stop PMo *p- is preserved as f- in most Gansu-Qinghai dialects (e.g. Bao. felaŋ, Mgr fulaan ‘red’) and as x- in Dagur and Dongxiang (e.g. Dong. xulan, Dag. xulaan ‘red’), but it disappeared in the other dialects (e.g. WMo. ulaɣan, Mogh. uloon, Kalm. ulaan, Khal. ulaan ‘red’). It is therefore safe to reconstruct PMo *p- in Proto-Mongolic forms for which Starostin reconstructs PMo *h- (e.g. I reconstruct PMo *pulaɣan ‘red’ instead of PMo *hulaɣan ‘red’). WMo. p occurs only in foreign loanwords in initial and medial position. Poppe also reconstructs a pre-Mongolian *β, which had already converged with *ɣ in Proto-Mongolic. The reason for its reconstruction is purely external, based on comparison with Turkic p, as in WMo. qaɣa- ‘close’ and Tk. qapa- ‘close’. The deep velar consonant *q was a stop in Proto-Mongolic, but is now pronounced as a fricative x. The deep velar con­ son­ant PMo *ɣ is still pronounced as a stop in the modern dialects as it probably was in Proto-Mongolic. The velar con­son­ants are joined with only certain vowels: q and *ɣ can be followed only by a, o, u, and k, g can only be joined with e, ö, ü, i. WMo. b, d, g, ɣ, s, m, n, ŋ, l, r can occur in word-final

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages position, but the final voiced stops go back to final voiceless stops in Proto-Mongolic, thus WMo. čig ‘direction’ < PMo *tik ‘id.’ Most of the contemporary Mongolic languages have a palatal sibilant š resulting from an original PMo *s before *i (< *i or *ï). In Written Mongolian the palatal š occurs also before vowels other than *i. Poppe attributes such occurrences of WMo. š to the secondary origin of the following vowel and he assumes that this vowel developed from PMo *i. Another explanation for WMo. š followed by a vowel other than -i- would be borrowing. Such may well be the case for WMo. šatu ‘stair’ which is following Poppe in borrowing from Tk. šatu ‘stair’. No palatal nasal can be reconstructed in Proto-Mongolic, but Poppe remarks that it is possible to distinguish between PMo *n followed by *i and PMo *n followed by other vowels. He suggests that PMo. *ni reflects an earlier palatal nasal followed by any vowel *ńV. Proto-Mongolic distinguishes between the liquids *l and *r, but both liquids do not appear in word-initial position. Initial l- and r- occur only in a small number of loanwords and onomatopoeia. Consonant clusters can occur in medial position, but there are no native Mongolic words that possess clusters word-initially or word-finally. Proto-Mongolic is particularly rich of sonorant clusters. Obstruent clusters are also reconstructable. The sonorant clusters frequently consist of a sonorant (r, l, n, m) followed by a stop P (b/p), T (t/d) or K (k/g) such as *lp, *lb, *lt, *ld, *lk, *lg; *rp, *rb, *rt, *rd, *rk, *rg; *nb, *nt, *nd, *nk, *ng (or *ŋ(g)), *mb, *mt. The obstruent clusters consist of a stop P (p/b), T (t/d) or K (k/g) followed by a dental (t/d) or velar stop (k/g) or sibilant (s), such as *pk, *bk, *pt, *bt, *ps, *bs, *tk, *dk, *gd, *kt.

Medial clusters are sometimes formed through secondary developments such as vowel syncope (e.g. WMo. aluqan ‘hammer’ > Khal. alxan, Bur. alxa, Kalm alxE) or they can develop on etymological or morphological boundaries (e.g. WMo. ög- ‘give’ + -te- passive > WMo. ögte- ‘be given’). Consonant clusters are prone to metathesis and assimilation (e.g. PMo *todka- ~ *togta ‘stop, fasten’ that is reflected in WMo. todqa- ‘hesitate’, Khal. totgor ‘fasten’, Kalm. totxə- ‘snare, trap’ versus WMo. toɣta- ‘to stop, establish’, Khal. togta- ‘stop’, Kalm. tokto- ‘stand firmly’) and sporadic elision of a liquid (e.g. the Mongolian doublet kelbej- ~ kebüji- ‘lean, be inclined to one side’).

Neither Starostin et al. nor Janhunen reconstruct the postvelars *q and *ɣ (also written *γ) as separate phonemes for Proto-Mongolic—they were exclusively nonphonemic pos­ition­al variants (allophones) of the velars *k and *g, thus: *q and *ɣ could only appear before *a, *o, *u, while *k and *g appeared before *e, *ö, *ü, *i (cf. Robbeets 2005: 73 [quoted above]; see also Poppe  1960b: 9, 16–20, 23–5, 53–62). Poppe (1955: 95 and 1960b: 9) reconstructs a more ­complicated consonant system for Common Mongolic in Table 28.12. According to Poppe (1955: 15), Common Mongolic still had initial *p- (or *φ-), and the sequences *-aγa-, *-aγu-, etc. were still preserved. Moreover, the vowels *ï and *i were differentiated only after *q, *γ and *k, *g. Elsewhere, *ï had already merged with *i (*ï > *i) and had caused the pal­at­al­iza­tion of immediately preceding dental stops (*ti > *či; *di > *ǯi). Poppe (1955: 96) considers the ancient voiceless stops and palatoalveolar affricate to be aspirated consonants in Common Mongolic and the ancient voiced stops and palato-alveolar

Table 28.12  Common Mongolic consonant inventory proposed by Poppe Labial Stops Affricates Fricatives

Palatal

Velar Velar

Postvelar

Voiceless

p

t

k

q

Voiced

b

d

g

γ (g͔)

Voiceless

č

Voiced

ǯ

Voiceless Voiced

Nasals

Dental & Alveolar

s&š w (β) m

y (= j) n

Laterals

l

Vibrants

r

ŋ

479

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d affricate to be unaspirated consonants—they were realized as voiceless in some positions and voiced in other positions. Modern Mongolic languages have reintroduced /š/ through loanwords. Several languages have also added /p/, /f/, and /w/, though their status tends to be rather marginal. New sequences of dentals before /i/ have been introduced (/ti/, /di/), which were not subject to the earlier process of pal­at­ al­iza­tion (see the preceding paragraph). Initial *h- has been mostly lost in the Modern Mongolic languages, though traces are still found in Dagur. Medial *-h- has been completely lost. Let us now look at the consonant system of Written (Literary) Mongolian. It is important to include Written Mongolian here for comparison. The reason for this is that, owing to the relatively shallow time depth commonly assumed for Proto-Mongolic, the Written Mongolian con­son­ant inventory is very close, though not quite identical, to that reconstructed for Proto-Mongolic, even allowing for idiosyncrasies of the Written Mongolian writing system. Unfortunately, relatively little is known about the linguistic situation prior to Proto-Mongolic, though it may be assumed that several (perhaps mutually intelligible) pre-Proto-Mongolic dialects existed. If only we had in-depth knowledge about these pre-Proto-Mongolic dialects, the reconstruction of ProtoMongolic as a whole would undoubtedly be both different and pushed much further back in time. The Written Mongolian consonant inventory is represented in Table 28.13 (cf. Hambis 1945: XII; see also Grønbech and Krueger 1993: 9–10; Janhunen 2003e: 35; Poppe 1974: 17; Robbeets 2015: 38; Rybatzki 2003b: 64 [Middle Mongolian]). For details on the phonological systems of the various modern Mongolic languages, cf. Janhunen (2003); see also Nugteren (2011); Poppe (1955)—see also Yamada (this volume: Chapter 20).

28.6 Tungusic The Tungusic branch contains two subgroups: (i) Manchu, Sibo (also called Sibe, Xibe, Xibo), and Jurchen (extinct—formerly spoken in China) and (ii) all other Tungusic languages (Evenki, Even, Solon, Negidal, Nanai [also called Gold, Goldi], Ulch, Oroch, and Udihe). Consequently, this language family is sometimes referred to as Manchu-Tungus. For information on the Southern Tungusic languages, see Oskolskaya (this volume: Chapter 19); see also Pakendorf and Aralova (this volume: Chapter 18) for information on Even and the North Tungusic languages. Starostin et al. (2003: 156) reconstruct the Proto-Tungusic consonant system in Table 28.14 (see also Tsintsius 1949b; Robbeets 2005: 68). The Proto-Tungusic consonant inventory included labial, dental, and velar points of articulation (voiceless: *p, *t, *k; voiced: *b, *d, *g). There were also corresponding labial, dental, and velar nasals (*m, *n, *ŋ) as well as voiceless and voiced palato-alveolar affricates (*č, *ǯ) and a palatalized nasal (*nʸ). There were two sibilants (*s, *š) and a voiceless velar fricative (*x). Starostin et al. complete the system with *l, *-r-, and ­*-y-. According to Starostin et al. (2003: 78–9), the Tungusic consonant system is the most conservative. They note that the following changes took place and in the exact order listed: (i) voicing of initial *t- and *č- (*t-, *č- > *d-, *ǯ-); (ii) spirantization of *kʰ (*kʰ > *x); (iii) merger of initial voiceless aspirates with their plain (unaspirated) voiceless counterparts (*pʰ-, *tʰ-, *čʰ- > *p-, *t-, *č-); (iv) voicing of medial *-p- (*-p- > *-b-) and development of medial *-č- to *-s-; (v) merger of medial voiceless aspirates with their plain (unaspirated) voiceless counterparts (*-pʰ-, *-tʰ-, *-čʰ- > *-p-, *-t-, *-č-).

Table 28.13  Written Mongolian consonant inventory Occlusives Bilabial

Affricates

Voiceless

Voiced

p

b

Voiceless

Fricatives

Voiced

Voiceless

Semivowel

480

Liquids

Vibrants

l

r

w t

d

Palatal Guttural

Nasals m

Labiodental Dental

Voiced

s č

k

g

q

γ

ǯ

(z)

n

š ŋ y

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages Table 28.14  Proto-Tungusic consonant inventory p

t

č

k

b

d

ǯ

g

s

š

x

l, -r-

-y- (= -j-)

n

nʸ (= -ń-)

m

languages that preserve initial *x-; ii) the distinction between *-k- and *-x- seems to reflect the Altaic distinction *-k- : *-k῾(see above), thus exactly paralleling the distinction *k- : *xin word-initial position.

This is a very important point, inasmuch as it is, in part, the basis for the reconstruction by Starostin et al. of a three-way contrast in the series of stops and affricates in ProtoTranseurasian: (i) voiceless aspirated (*pʰ, *tʰ, *čʰ, *kʰ); (ii)  plain (unaspirated) voiceless (*p, *t, * č, *k); and (iii) voiced (*b, *d, *ǯ, *g). Menges (1968a: 35–6), on the other hand, reconstructs a slightly more complex consonant system for Proto-Tungusic in Table 28.15, showing the velar ~ uvular variants mentioned in note 2. Gorelova (2002: 85) lists (Table 28.16) the con­son­ants for Literary Manchu (see also Robbeets 2015: 39).

ŋ

*-r- and *-y- only occur medially. The distinction between velar and uvular consonants found in modern Tungusic languages represents a later development. They arose as positional variants (allophones) adjacent to front or back vowels (cf. Gorelova 2002: 86 [for Literary Manchu]). 3. Proto-Tungus had an extensive system of medial consonant clusters (cf. Robbeets 2005: 70 for details). 4. *x is lost in the majority of the Tungusic daughter languages. 1. 2.

Finally, initial *z- became *s-, and the palatalized con­son­ ants were depalatalized (*lʸ, *rʸ > *l, *r), except for *nʸ, which was retained. Regarding the reconstruction of Proto-Tungusic inter­ vocal­ic *-x-, Starostin et al. (2003: 160) note:

Table 28.15 Proto-Tungusic consonant inventory proposed by Menges

Intervocalic *-x- is an innovation in PTM reconstruction, first proposed in Dybo (1990). It is based on the distinction between -k- and -x in Ulcha, Orok and Nanai. Northern languages, as well as Oroch, Udehe and Manchu have completely merged the reflexes of *-k- and *-x-. Such a reconstruction seems probable for two reasons: i) the languages that preserve the distinction between *-k- and *-x- are exactly the same

p

t

tʸ (= tj)

q/k

b

d

dʸ (= dj)

ġ/γ//g

s w?

ł/l

y (= j)

w?

m

n

nʸ (= nj)

ŋ

Where different from that used in this chapter, the transcription used by Menges is given in parentheses.

Table 28.16  Literary Manchu consonant inventory Place of articulation Mode of articulation Obstruents Fricatives

Bilabial

Labiodental

Front

Back

p

t

k

Voiced

b

d

g

s, š

h

Voiceless

f v

j*

Voiceless

č (c)

Voiced Nasals

Dorsal

Voiceless

Voiced Affricates

Labial

čž (cz) m

n

Laterals

l

Flapped

r

ŋ

*[j] corresponds to [y] in other systems where [j] is instead used to indicate [ž]. Gorelova (2002: 85) Following the views of Russian scholars, Gorelova (2002: 86) notes that /k/, /g/, /h/ have velar allophones [k], [g], [x] before the vowels e, i, u but uvular allophones [q], [ɢ], [χ] before the ­vowels a, o, ū (the symbol /ū/ is used to indicate two sounds: [ʊ] after uvulars and [o] in borrowings).

481

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d The most comprehensive treatment of Tungusic phon­ ology is Tsintsius (1949b).

Table 28.18  Late Middle Korean consonant inventory Plain

p

t

k

c

Aspirated

ph

th

kh

ch

28.7 Korean

Reinforced

[pp

tt

kk

(cc)]

Written records in Korean begin to appear after 1446, the date when the Korean writing system (Hangul [한글]) was invented. There is only fragmentary evidence available before that date, inasmuch as everything was written in Chinese. Information from the time of the invention of writing, that is, Late Middle Korean, is used to reconstruct earl­ier stages of Korean, such as Early Middle Korean and Old Korean. Old Korean is usually taken to be the form of the language ­spoken in the Kingdom of Silla, which was founded around 57 bc and lasted until ad 935. For information on the classification of the Koreanic languages, see Sohn, this volume: Chapter 15. Starostin et al. (2003: 163) reconstruct the Proto-Korean consonant system in Table 28.17 (see also Robbeets 2005: 60). Old Korean had a series of plain (unaspirated) voiceless con­son­ants (p, t, k, č [= tš͜ ]) as well as a series of voiceless as­pir­ated consonants (pʰ, tʰ, kʰ, čʰ [= tš͜ ʰ]) (cf. Lee and Ramsey 2011: 64)—there was no voicing contrast in Old Korean. The voiced consonants reconstructed by Starostin et al. for Proto-Transeurasian were devoiced and merged with the plain voiceless consonants in Proto-Korean (*b, *d, *g, *ǯ > *p, *t, *k, *č), though medial voiced labial and dental stops remained (Starostin et al. reconstruct medial *-b- and *-d- to account for the alternations *-p-/*-w- and *-t-/*-r- in verbal stems—these were eventually lost as well). Moreover, the voiceless aspirated consonants reconstructed for Proto-Transeurasian by Starostin et al. merged with the plain (unaspirated) consonants in pre-Proto-Korean (*pʰ, *tʰ, *kʰ, *čʰ > *p, *t, *k, *č). Hence, the voiceless as­pir­ ates found in Old Korean must have arisen at a later date (cf. Lee and Ramsey  2011: 64). Their origin is not clear. In Middle Korean, new voiceless aspirates arose from combinations of the plain consonants with /h/ (-ph-, -hp-, etc.) and from clusters that arose through syncope (see also Robbeets 2005: 65–6). The process lasted several centuries.

Voiced

W

Nasal

m

Table 28.17  Proto-Korean consonant inventory p

t

č

k

-b-

-d-

m

n

nʸ (= ń)

ŋ

r

y (= j)

’ (Ø)

s

482

h

Liquid

G n

s

h

ss

hh

z

ŋ

l

The reinforced (“fortified”) consonants were not yet phonemic in Late Middle Korean (cf. Lee and Ramsey  2011: 128–9). Sohn (1999: 153) transcribes them as /p’/, /t’/, /k’/, /c’/, while Starostin et al. (2003) write /p:/, /t:/, /k:/, /c:/.

The tense, unaspirated (“reinforced”) series of Modern Korean arose in Late Middle Korean and does not go back to either Proto-Korean or Proto-Transeurasian (cf. Lee and Ramsey  2011: 128–30; Whitman  2012: 431). According to Robbeets (2005: 64), the reinforced consonants result from Middle Korean consonant clusters (see also Starostin et al. 2003: 164). Robbeets (2005: 60–1) does not reconstruct a voicing contrast for Proto-Korean. She points out, however, that the obstruents /p/, /t/, /k/, /c/ are automatically voiced in certain environments in Modern Korean, and this was probably also the case in Middle Korean. Robbeets (2005: 64) also suggests that there may have been a phonemic distinction between *r and *l in Old Korean, though this is uncertain. In any case, only /l/ is found in Middle Korean. According to Lee and Ramsey (2011: 128), the Late Middle Korean consonant inventory is represented in Table 28.18 (see also Robbeets 2005: 60). For information on Modern Korean phonology, see especially Duck-Young Lee (1998) (see also Song  2005: 22–44; Hansol H. B. Lee 1989: 10–36).

28.8 Japonic The final section deals with Japonic. Questions concerning both the prehistory and early history of Japonic are still being worked out—even the internal subgrouping of the Japonic languages remains unsettled (see de Boer, this volume: Chapter  4). While it seems probable that the Yayoi invaders brought a Transeurasian language with them when they entered Japan, or better, a Transeurasian language that had already come into contact with one or more Austronesian languages (cf. Robbeets 2017a), the language and ethnic identity of the indigenous Jōmon inhabitants of Japan remain unknown. Japonic has a rich history (see Miyake,

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages this volume: Chapter 1), and there is an extensive literature dealing with earlier stages of development as well as with the standard modern language and its dialects. I have only included references to a very small sampling of the relevant literature here. According to Starostin et al. (2003: 167), the consonant system of Proto-Japonic is to be reconstructed in Table 28.19 (see also Robbeets 2005: 52). For comparison, the Proto-Ryūkyūan consonant system has been reconstructed in Table 28.20 (cf. Thorpe 1983; see also Bentley, this volume: Chapter 14). Starostin et al. (2003: 167) note that the pairs *b ~ *w and *d ~ *y were in complementary distribution in Proto-Japonic: *b- and *d- occurred only word-initially, while *-w- and *-yoccurred only medially between vowels. They also reconstruct Proto-Japonic initial *b- and *d- and maintain that, in many cases, these sounds go back to Proto-Transeurasian initial *b- and *d-. As pointed out by Robbeets (2005: 53), “[t]he reconstruction of the Proto-Japanese consonant inventory is relatively uncontroversial.” However, as opposed to Starostin et al., Robbeets argues against the reconstruction of voiced obstruents in Proto-Japonic, though she does include them in the consonant inventory of Old Japanese (cf. Robbeets 2005: 53, 55) (footnotes are not included): The difference between the Proto-Japanese and Old Japanese consonant system listed in table 1 and in table 2 lies in the appearance of a voiced obstruent series b, d, g, z. There is

Table 28.19  Proto-Japonic consonant inventory p

t

b

d

m

n

w

y (= j)

k

r s Table 28.20  Proto-Ryūkyūan consonant inventory p

t

k

b

d

g

m

n

w

y (= j) r s z

good reason to argue that the voiced obstruents were prenasalized in Old Japanese. Old Japanese prenasalization can be described as a nasal onglide preceding a medial voiced obstruent and the prenasalized obstruents are usually transcribed as [nb], [nd], [ng], [nz]. The first indication that prenasalization has developed into voice distinction comes from the observation that Old Japanese voiceless obstruents develop into voiced obstruents in a prenasalized environment. From the ninth century on, a heterogeneous group of phonological changes known collectively as onbin altered the Old Japanese phonological pattern in a way that a syllable-final nasal was allowed. Japanese verb morphology shows the effects of onbin changes clearly. Contemporary Japanese gerunds such as yonde ‘read’, tonde ‘fly’ and sinde ‘die’ have developed from the Old Japanese forms yomite ‘read’, tobite ‘fly’ and sinite ‘die’. In the formation of the verb gerund -mite, -bite, and -nite all become -nde by vowel elision and in the new prenasalized environment, the obstruent -t- becomes voiced -d-. The second indication is that morphological and etymological boundaries sometimes occur on a voiced obstruent, evidencing the contraction of a morpheme with a final nasal syllable and a morpheme with an initial obstruent. Examples can be found in internal derivations such as hude ‘writing brush, painting brush, pen’ from humi ‘writing’ and te ‘hand’; OJ kabi1 ‘sprout, unhusked ear of grain’ from OJ kami1 ‘top, head, upper part’ and OJ ipi1 ‘rice’; kabuto ‘helmet, headpiece’ from kami ‘top, head, upper part’ and huta, OJ puta ‘lid’, nazo ~ nanzo ‘riddle, why’ from nani ‘what’ and -so ‘thing’, nado ‘or something’ from nani ‘what’ and -to ‘place’, uzi ‘clan, lineage’ from umi, the deverbal noun of umu ‘give birth to, bring forth’ and ti ‘blood; line’, yagate ‘before long’ from yami, the deverbal noun of yamu ‘stop’ and the adjective stem kata‘hard’, yugake ‘archer’s glove’ from OJ yumi-kake2 ‘archer’s glove’ and originally from yumi ‘bow’ and the deverbal noun of kakeru ‘cover with (cloth), spread over, veil’. Third, a number of modern dialects, notably those of northern Honshū, southern Shikoku and some Ryūkyū dialects have retained prenasalization to some extent. Sendai in northeast Honshū, for example, pronounces mado ‘window’ as [mãdo] with nasality on the preceding vowel. The Kobama dialect in the Ryūkyūs has kandu for standard kado ‘corner’ and pangun for hagu ‘strip off, tear off’. In many other dialects, including standard Tokyo, prenasalization has been lost. However, in Tokyo dialect there is a residue of the prenasalized velar, like in the pronunciation of kagi ‘key’ as [kaŋi] or the nominative case marker ga as [ŋa], with tendency of [ng] to absorb the nasality and to become [ŋ]. Finally, foreign written sources suggest that the voiced obstruent series in Late Middle Japanese and Modern Japanese were still prenasalized to a certain extent. In the Japanese-Portuguese dictionary of 1604 -nd- and -ng- are

483

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

All a n R. Bomhar d observed for -d- and -g- in the speech of Kyūshū at that time. An 18th-century Korean glossary of Japanese, the Wagoruikai, writes Hangǔl digraphs equivalent of -mp-, -nt- and -ŋk- for the medial voiced stops -b-, -d- and -g-. And the Nihon Kan-yakugo, a 15th century Chinese language guide transcribes Japanese syllables preceding a voiced stop with a final nasal. On the basis of morphological, etymological, dialectal and textual evidence it is safe to assume that the Old Japanese obstruents OJ b, d, g, z resulted from the rephonologization of prenasalized obstruents pJ *np, pJ *nt, pJ *nk, pJ *ns. Reminiscent of how the Altaic languages do not allow for consonant clusters in initial position, Old Japanese did not permit word-initial voiced obstruents except in mimetic adverbs. From the ninth century on, as loans from Chinese began to have a major impact, the restriction was relaxed and initial voiced obstruent began to appear in borrowings and in contracted native forms.

But, Starostin et al. (2003: 167–8) speak in favor of the reconstruction of a series of intervocalic voiced stops in Proto-Japonic: Besides voiceless intervocalic stops, OJ also had voiced -b-, -d-, -g-. The general consensus now is that in most cases these voiced stops reflect PJ clusters *-mp-, *-nt-, *-nkwhich are the only consonant clusters possible in PJ and may have actually been pronounced as prenasalized stops. In some cases these clusters actually reflect original PA clusters; but, as we tried to show above, in many more cases they go back to plain voiceless or voiced consonants in syllables with high pitch. It is therefore also possible to regard the OJ situation as original, or possibly as resulting from a merger of clusters *-mp-, *-nt-, *-nk- and voiced *-b-, *-d-, *-g- (from earlier plain stops).

The evolution of the Proto-Transeurasian consonant system in Japonic is rather complicated. Starostin et al. (2003: 81) propose the following series of changes and in the order listed (see also S. Starostin 1991 and 1997): (i) The laterals *l- and *lʸ- were lost: *l- and *lʸ- > *n; (ii) *-g- > *-ɣ- in the third syllable; (iii) the affricates and sibilants were modified: (A) affricates: initial *č- > *čʰ-; medial *-č- > *-š-; *čʰ > *tʸʰ (= *t́ʽ); and *ǯ > *dʸ; (B) sibilants: *z, *š > *s; (iv) plain (unas­pir­ated) consonants became aspirated consonants when another aspirated consonant immediately preceded or followed in a word: *CVCʰV, *ZVCʰV > *CʰVCʰV; and *CʰVCV (but not *CʰVZV) > *CʰVCʰV (note: the cover symbol Z denotes any voiced consonant); (v) voiced stops were pal­at­al­ized after *i̯-diphthongs (*i̯a, *i̯o, *i̯u) and before *-y-: *b > *bʸ; *d > *dʸ; and *g > *gʸ; (vi) likewise, initial *ŋ- was pal­at­al­ized before *i̯: *ŋ- > *nʸ-; (vii) medial *-Cʰ- and *-Z(but not *-C-) > *-nC- when the syllable had high pitch; (viii)

484

initial *b- > *β- before low vowels; initial *d- > *δ- before any vowel; (xi) (*k > g; *p > *b), *t > *d (but *dʸ before front ­vowels); (x) voiced palatalized stops became fricatives: *bʸ > *β; *dʸ > *δ; and *gʸ > *ɣ; (xi) *Cʰ, *Z > *C; (xii) initial *ŋ- was lost (*ŋ- > *Ø-); (xiii) *rʸ > *t before *-i, *-u; (xiv) *lʸ > *nlʸ before *-rV- and *r > *nr before *-rV-; (xv) *nr > *nt and *r > *t (sporadically); (xvi) *lʸ > *s; *rʸ > r; *l > *r; and *nʸ > *ŋ; (xvii) *ŋ > *m- (initially), *-m-/*-n- (medially); and (xviii) voiced fricatives were lost: *β > *b- (initially), *-w- (medially); *δ > *d- (initially), *-y- (medially); and *-ɣ- > *-Ø- (medially). Starostin et al. (2003: 82) speculate that most of the phono­logic­al changes in the early prehistory of Japanese occurred under the influence of an Austronesian substratum. According to Frellesvig (2010: 34), Old Japanese (c. ad 700– 800) had the inventory of consonant phonemes represented in Table 28.21 (see also Miyake 2003a: 164–97; Robbeets 2005: 53 and 2015: 41). There was considerable phonetic variation in the realization of the obstruents in Old Japanese (cf. Frellesvig 2010: 35–8). Table  28.22 gives the phonetic realization of Old Japanese obstruents. Significant phonological changes took place between Old Japanese (c. ad 700–800) and Early Middle Japanese (c. ad 800–1200), affecting both syllable structure and segmental phonology. Only minor phonological changes have taken place since then. On the other hand, significant gram­mat­ical changes took place in the Late Middle Japanese period (c. ad 1200–1600), affecting both morph­ology and syntax. Old Japanese syllable structure was completely straightforward. According to Frellesvig (2010: 39), all syllables were open and short, conforming to the repetition of a simple (C)(G)V pattern, where C = any consonant, G = a glide (either /w/ or /y/), and V = any vowel. For more information on Japanese phonology, see especially Labrune (2012).

Table 28.21  Old Japanese consonant inventory Labial

Alveo-dental

Tenuis

p

t, s

k

Media

-b-

-d-, -z-

-g-

Nasal

m

n

Liquid Glide 1. 2.

Palatal

-rw

y

/b/, /d/, /g/, /z/, and /r/ did not occur initially. The phonetic realization of /s/ and /z/ is uncertain.

Velar

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

consonant inventory of tra nseurasian l a nguages Table 28.22  Phonetic realization of Old Japanese obstruents Phoneme

Phonetic realization Word-Initial

Word-Medial

/p/

[p ~ ɸ]

[b ~ β]

/t/

[t]

[d]

/k/

[k ~ x]

[g ~ ɣ]

/s/

[s ~ s]

[z ~ dz]

t

/b/

[mb ~ mβ]

/d/

[nd]

/g/

[ŋg ~ ŋɣ]

/z/

[nz ~ ndz]

Adapted from Frellesvig 2010: 35, as are the accompanying notes. 1. Medial /b/, /d/, /g/, /z/ were prenasalized [mb ~ mβ], [nd], [ŋg ~ ŋɣ], [nz ~ ndz] in Old Japanese. This feature was subsequently lost. 2. /p/, /t/, /k/, /s/ were tense (“tenues”), while /b/, /d/, /g/, /z/ were lax (“media”). 3. Medial /p/, /t/, /k/, /s/ had voiced variants [b], [d], [g], [z], re­spect­ive­ly. 4. The sibilants [s-], [-z-], [-nz-] had affricate variants [ts-], [-dz-], [-ndz-], respectively. The distribution seems to have been conditioned by the following vowel. 5. Initial /p/ and /k/ (but not /t/) had fricative variants [ɸ] and [x], respectively.

28.9  Root structure patterning in Proto-Transeurasian The following discussion of root-structure patterning in Proto-Transeurasian is based upon Starostin et al. (2003: 22–4) (see also Bomhard 2018 vol. 1: 247–8; Robbeets 2017c: 588–9). The Transeurasian languages are agglutinating in structure. Pronominal stems and particles were monosyllabic (*(C)V), while nominal and verbal stems were typically disyllabic (*(C)VCV or *(C)VCCV). Polysyllabic stems could be derived from the disyllabic stems by the addition of suffixes. The addition of a suffix caused no changes in the vowel of the stem, but the vowels of suffixes were subject to vowel harmony, which means that their vowels were adjusted to the vowels of the stem. The undifferentiated stems were real forms in themselves and could be used without additional suffixes. The suffixes, both derivational and inflectional, were added mechanically to the stem. According to Starostin et al. (2003: 22–4), the most common root-structure pattern in Proto-Transeurasian was *CVCV, occasionally with a medial consonant cluster—*CVCCV. The final vowel, however, was very unstable: it is best preserved in Tungusic languages (though it is not always easily reconstructable owing to morphological processes), and it is

frequently dropped in Korean, Mongolic, and Turkic (in the latter family, in fact, in the majority of cases). Japanese usually preserves the final vowel, although its quality is normally lost; however, in cases where the final (medial) root consonant is lost, Japanese reflects original disyllables as monosyllables. Japanese also has quite a number of monosyllabic verbal roots of the type *CVC-. These roots were originally disyllabic as well. However, reconstructing them as *CVCa- is certainly incorrect. The Old Japanese verbal conjugation shows explicitly that the verbal stems can be subdivided into three main types: *CVCa- (those having the gerund in -e < *-a-i), *CVCə- (those having the gerund in -i < *-ə-i), and *CVC- (those having the gerund in -ji < *-i). Here, there is a possibility that the latter type reflects original verbal roots *CVCi (occasionally perhaps also *CVCu, though there are reasons to suppose that some of the latter actually merged with the type *CVCə-). The gerund form in *-i may actually reflect the original final root vowel that had earlier disappeared before other verbal suffixes of the type *-V(CV)-. A small number of trisyllabic roots such as *alakʰu ‘to walk’, *kabari ‘oar’, *kʰobani ‘armpit’, etc. can also be reconstructed for Proto-Transeurasian. It cannot be excluded that, in many or most of these cases, the final syllable was originally a suffix, but the deriving stem was not used sep­ar­ ate­ly, and the derivation had already become obscure in the proto-language. The monosyllabic structure *(C)V was typical for pronominal and auxiliary morphemes, but a small number of verbal (and, quite exceptionally, nominal) monosyllabic roots can also be reconstructed. A special case involves a number of verbal roots that appear as monosyllables of the type *CV in some languages but have the structure *CVl(V) or, less frequently, *CVr(V) in others. Starostin et al. (2003) reconstruct disyllables here, but note that the exceptional loss of *r and *l remains unexplained. A possible solution would be to reconstruct those roots as *CVC, with occasional loss of the root-final resonant. However, the number of examples is not large, and the roots in question are frequently used as auxiliary verbs, which by itself could explain the exceptional phonetic development. It is also possible that *-r- and *-l- were originally suffixed and that the roots belonged instead to the rare type *CV. Starostin et al. note that the problem requires further investigation. There were four fundamental stem types in ProtoTranseurasian:

1. Verbal stems 2. Nominal and adjectival stems 3. Pronouns 4. Particles

There was a strict distinction between nominal and verbal stems in Proto-Transeurasian.

485

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Ch a pter 29

A comparative approach to the vowel systems and harmonies in the Transeurasian languages and beyond A n dr e w J ose ph , S eong y eon K o ,

29.1 Introduction Attempts to reckon with the similarity of the vowel systems of the Altaic languages have a long history. Philip Johan von Strahlenberg (1730) had already perceived the profound structural affinities in the vowel domain specifically among the micro-Altaic subset of Turkic(-Chuvash), Mongolic, and Tungusic in the first half of the 18th century.1 The body of literature from the ensuing nearly three centuries of research is of course too vast to encompass here. We instead concentrate on the most influential modern treatments of the vowel correspondences among Transeurasian languages. In our view, it is an observable fact that there is a representative Transeurasian vowel inventory type, in the sense of  a particular configuration of segmental vowels; closely related featural distributions; associated phonological processes such as vowel harmony, etc.; all to be discussed in greater detail in upcoming sections. The classical Altaic reconstructions capture the unity of the vowel inventories of ProtoTurkic, Proto-Mongolic, etc., directly, by assuming a common ancestor with precisely the same Transeurasian inventory properties, such that daughter languages belong to this type as a result of inheritance (retention). Another logical possibility is that the wide distribution of this common Transeurasian vowel inventory type is the result of pervasive contact. It is worth emphasizing that these two explanations are not 1   In this paper “Micro-Altaic” refers to the “core” Altaic families, Turkic, Mongolic, and Tungusic, particularly in the context of a hypothesized genetic relationship. “Altaic” with no prefix refers to the grouping proposed in the loci classici of Altaic comparative studies, Ramstedt (1952−66) and Poppe (1960b), where Korean is included, if cursorily, with the micro-Altaic core. We also use the abbreviation KMT to refer to the subset Korean, Mongolic, and Tungusic. “Macro-Altaic” will be reserved for hypotheses of even larger genetic groupings, particularly those that include Japonic (e.g., Miller 1971; Starostin 1991; Robbeets 2005; Starostin et al. 2003). “Transeurasian” in the sense of Johanson and Robbeets (2010a) and other recent works refers to the same grouping, but with no commitment as to whether it is genetic or areal. We use the term “Northeast Asian” to refer to the explicitly areal grouping of families located entirely or in part in this region.

and

J oh n W h i t m a n

mutually exclusive, but in any case the vowel inventory type can be treated as an autonomous object of inquiry. (We would further argue that it is only after the phonetic and phonological properties of the Transeurasian type are more fully examined within their typological and areal contexts that questions about their origins can be answered satisfactorily.) In this chapter, we review the standard treatments of the Transeurasian vowel correspondences, including their reconstructions of hypothetical proto-inventories, for the purpose of establishing a description of the Transeurasian vowel inventory type. We then offer a phonetic and phonological analysis of this inventory type, and propose interpretations of the main Transeurasian reconstructions, providing phonetic substance and phonological detail according to our analysis. Having established the properties of the Transeurasian vowel inventory type, and having applied our understanding to the interpretation of the classic reconstruction systems, we turn to critical issues raised along the way. In particular, we endeavor to situate the Transeurasian vowel inventory in its typological neighborhood, and in its linguistic geographical setting. Finally, we offer a tentative assessment of the conclusions we believe can be drawn from our analysis as to the historical status of Transeurasian.

29.2  Vowel system and harmony in the Transeurasian languages 29.2.1  Backness (palatal) vs. tongue-root vowel systems Within the Transeurasian domain, there are two principal types of vowel inventory, corresponding to two principal types of vowel harmony.2 2   For the time being we set aside two additional types of harmony: height harmony, involving [±high] or [±low]; and labial harmony, involving [labial]

Andrew Joseph, Seongyeon Ko, and John Whitman, A comparative approach to the vowel systems and harmonies in the Transeurasian languages and beyond In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Andrew Joseph, Seongyeon Ko, and John Whitman. DOI: 10.1093/oso/9780198804628.003.0030

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian Table 29.2  Ewen inventory

Table 29.1  Anatolian Turkish vowel inventory Set A (= [-back])

i

y

e

ø

Set A (= [-rtr])

i

ə

u

o

Set B (= [+back])

ɯ

u

a

o

Set B (= [+rtr])

ɪ

a

ʊ

ɔ

Kornfilt 1997

J. Kim 2011

29.2.1.1  What is a [±back] system?

Table 29.3  Khalkha

In the first type, the decisive phonological dimension is that of vowel backness, formally represented using the dis­tinct­ ive feature [±back] (or [±front]). This type is also frequently referred to as palatal harmony. In languages of this type such as Standard Anatolian Turkish (Table 29.1), vowel ­harmony based on backness requires, in principle, that all vowels in a given phonological word agree for the feature [±back]. Typically, stems obey the requirement and their vowels are non-alternating, but they also spread or otherwise impose their specification for [±back] onto the vowels of any concatenated affixes, resulting in agreement throughout the phonological word.

Neutral

i

Set A (= [-rtr])

e

u

o

Set B (= [+rtr])

a

ʊ

ɔ

Svantesson 1985; Svantesson et al. 2005

Table 29.4  Middle Korean Neutral

i

Set A (= [-rtr])

ə

ɨ

u

Set B (= [+rtr])

a

ʌ

o

S. Ko 2018; J. Kim 1999; cf. K.-M. Lee 1972

29.2.1.2  What is a [±rtr] system? In the second type of vowel inventory/harmony, the crucial featural dimension is not identical to any of the major vowel features of backness, height, and roundness, and thus has historically proven more difficult to characterize with respect to its articulatory basis and acoustic properties. This type of vowel harmony has been called pharyngeal (Ladefoged and Maddieson  1996); horizontal (Jacobson  1942); crossheight (Stewart 1976); relative-height (Ard 1984); open-close (Jungraithmayr  1971; Hattori  1975); diagonal (C.-W.  Kim 1978); apertural (Janhunen  1981b); or tongue-root vowel harmony, among other terms, reflecting disparate views of the salient phonetic and phonological properties of the vowel classes segregated by vowel harmony. Here, we employ the phonological feature [±rtr] (± retracted tongue root) to formally represent this contrast.3 In languages of this or [round]. In many, or perhaps all, purported cases of height harmony in Altaic, an alternative analysis in terms of tongue root harmony is available (see Ko 2018; Ko 2013b on Udihe; see van der Hulst and van de Weijer 1995 for further examples). Labial harmony is widespread in Transeurasian languages, but absent in Korean and Japanese, and more generally, in littoral Northeast Asia. It is typically superimposed on another type of harmony. See Kaun (1995, 2004) for a general discussion. See Section 29.4.5 for some comments on the relationship between labial harmony and vowel inventory structure. 3   Elsewhere (Ko 2011, 2013a, 2018; Ko et al. 2014), we have argued against proposals that [±rtr] is secondary (as claimed for Mongolic, e.g., in Svantesson 1985; Svantesson et al. 2005; Janhunen 2003), or (Joseph and Whitman 2013; Ko et al. 2014; Ko 2018) that specific proto-languages such as Proto-Tungusic should be reconstructed without vowel harmony (as in Starostin et al. 2003). We have also presented arguments (Ko et al. 2014) that [±rtr] is specifically implicated in Transeurasian tongue-root vowel harmony, rather than [±atr] (advanced tongue root) as claimed in Zhang and Dresher 2004 for Written

type in the Transeurasian domain, such as the Tungusic language Ewen (Table 29.2), the Mongolic language Khalkha (Table 29.3), or the oldest fully attested Korean language Late Middle Korean (Table 29.4), vowel harmony based on tongue-root position requires that, in principle, all vowels in a given phonological word agree for the feature [±rtr] (except for neutral vowel(s)). As in the case of backness vowel harmony, stems are typically non-alternating and subject to the vowel-harmony requirement, as well as governing the alternation of affix vowels within the phonological word.

29.2.1.3  TRH systems in the Korean, Mongolic, and Tungusic languages Although many Tungusic and Mongolic languages (as well as Old Korean as reconstructed by K.-M. Lee 1972) have long been held to have palatal harmony (PH), a growing body of research (listed in the remainder of this section) suggests that, at least, the non-Turkic families in the group are best described as having tongue-root harmony (TRH). TR analyses of the vowel systems of the Korean, Mongolic, and Tungusic (KMT) languages: (i) Tungusic (Novikova  1960; Ard  1981, 1984; Hattori 1982; J. Kim 1989, 1993, 2011; de Boer 1996; Zhang 1996; Zhang and Dresher  2004; Dresher and Zhang  2005; Manchu (see the note no.4 below for a brief discussion). We refer the reader to the works cited for a fuller treatment.

487

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman Li  1996; Ko  2013b, 2018; cf. Hayata  1980 for a height harmony analysis) (ii) Proto-Tungusic (Li  1996; Joseph and Whitman  2013; Ko et al. 2014; Ko 2018) (iii) (Eastern) Mongolic (Činggeltei 1959, 1963; Svantesson 1985; Svantesson et al. 2005; Ko 2018) (iv) Proto-Mongolic (Ko 2011, 2013a, 2018) (v) (Middle) Korean (J.-H. Park 1983; B.-G. Lee 1985; J. Kim 1988, 1993, 1999; J.-S. Lee 1992; Y. Lee 1993; M.-H. Cho 1994; D.-Y.  Lee  1994; J.-K.  Kim  2000; Park and Kwon 2009; Ko 2010a, 2010b, 2013a, 2018) (vi) Across languages deemed Altaic (Vaux 2009; Ko et al. 2014; Ko 2018; Robbeets 2015) The tongue-root analysis of modern varieties in the group is supported by the various pieces of phonetic and phonological evidence listed below. Phonetic and phonological evidence in favor of the TRH analysis for the KMT languages: (i) Tongue-root position in X-ray tracings: the Set B ­vowels are produced with more tongue-root retraction (Činggeltei and Sinedke  1959; Buraev  1959; Novikova 1960) (ii) Size of pharyngeal cavity (Möömöö 1977 as cited in Svantesson et al. 2005; Novikova 1960; Li 1996) (iii) Greater muscular effort or tension associated with the active feature (Möömöö 1977) (iv) Impressionistic “voice quality” phenomena (v) Acoustics of vowels: e.g., formant frequencies (Svantes­ son 1985; Svantesson et al. 2005 for Mongolic; Aralova et al. 2011 for Ewen; Kang and Ko 2012 for Ewen and Buriat; Lulich and Whaley 2012 for Oroqen; Yun et al. 2016 for Nanai; Kang et  al. 2017 for Ewen; but see Aralova 2015 for a more recent opposing view) (vi) Distinction between velar and uvular consonants conditioned by the two sets of vowels (Vaux 1999, 2009)

29.2.1.4  Evidence in favor of [±rtr] over [±atr] as the harmonic feature We will assume that [±atr] and [±rtr] are phonologically distinct features, following Goad (1992). Our position is that TRH in KMT is based on the feature [±rtr] rather than [±atr] (see especially Joseph and Whitman  2013; Ko et  al. 2014; Ko 2018). Whether a TRH language exploits [±atr] or [±rtr] as the “active” feature is determined according to the notion of phonological markedness in Table 29.5. For instance, the behavior of neutral vowels with respect to vowel harmony indicates that the retraction of the tongue root is a more marked gesture than the non-retraction (or advancement) of the tongue root, which in turn means

488

Table 29.5  Phonological markedness Marked

Unmarked

subject to neutralization

result of neutralization

unlikely to be epenthetic

likely to be epenthetic

trigger of assimilation

target of assimilation

remains in coalescence

lost in coalescence

retained in deletion

lost in deletion

Rice 2007: 80

that [±rtr], rather than [±atr], is the phonogically active feature.4 Evidence from the behavior of neutral vowels in KMT vowel harmony: (i) Neutral vowels do not trigger harmony: the class of vowels found in suffixes attached to neutral roots— i.e., the default class—does not bear the active feature. (ii) Neutral vowels may block harmony: the feature that fails to propagate over neutral vowels is the active feature. (iii) The inactive feature surfaces when a harmonic contrast is neutralized. An additional piece of evidence in favor of the [±rtr] analysis comes from the contrast between velars and uvulars, which is widespread in Tungusic and Mongolic.5 The general pattern is that velars become uvulars when adjacent to [+rtr] vowels. When adjacent to the neutral vowel /i/ (/i/ does not trigger or block vowel harmony), velars surface as velars. This is as expected under the [±rtr] analysis. However, if we assume that [±atr] is the active feature and the uvulars are the default, we face a contradiction regarding the feature specification of the neutral vowel /i/: it must be specified [+atr] in order to trigger velarization of the uvulars, while it cannot be specified [+atr] if it is to be neutral/transparent in [±atr] harmony; see Figure 29.1. a. /k, x, g/ → [q, χ, G] when adjacent to a tautosyllabic [+RTR] vowel b. /k, x, g/ → [k, x, g] when adjacent to a neutral vowel (e.g., /i/)

Figure 29.1  Velar ~ uvular alternation

4   Written Manchu has been analyzed, e.g. by Zhang and Dresher (2004), as an ATR (rather than RTR) harmony language. However, the direction of the assumed merger between *i and *ɪ and the result of neutralization between /u/ and /ʊ/ together indicate that [rtr] is the active or dominant feature since it is neutralized diachronically and synchronically. 5   Of course, Turkic also shows a similar, very widespread alternation between velars and uvulars seemingly conditioned by vowel backness. We will return to this issue in Section 29.3.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian In Goad’s (1992) system, the feature [±rtr] may be borne by consonants as well as vowels, while [±atr] may not. On this approach, consonant alternations such as the velar ~ uvular alternation in Tungusic and Mongolic are expected in languages where [±rtr] is the active feature, but not in languages with active [±atr]. On different grounds, it has been argued that Middle Korean vowel harmony is better accounted as [±rtr] rather than [±atr] harmony (J.  Kim  1988, 1993, 1999; J.-K.  Kim  2000; Ko  2010a, 2013a, 2018). See Ko (2018) for a comprehensive discussion and [±rtr] analyses of all Altaic languages asserted in the literature to have tongue-root or height harmony.

29.2.2  Reconstructing [±rtr] harmony in Korean, Mongolic, and Tungusic 29.2.2.1  Basic vowel correspondences in the KMT languages The evidence from family-internal vowel correspondences strongly supports the conclusion that the proto-languages of the KMT branches of Altaic should be reconstructed as tongue-root systems. The following tables show the basic vowel correspondences for Tungusic (Table 29.6), Mongolic (Table 29.7), and Korean (Table 29.8), respectively. None of the correspondences in these three families reveals any trace of PH, except for Kalmyk (and Oirat) in Mongolic; outside of Kalmyk/Oirat, all rounded vowels in languages with regular vowel harmony are realized as back vowels. Even the traditional assumption that Kalmyk/Oirat has a PH system has recently been challenged on (articulatory) phonetic grounds. Washington (2016: 27–8) points out that Pavlov’s X-ray tracings of Kalmyk vowels (1983: 35–40, 65–83) show very clear differences in tongue-root position between “front” and “back” vowels, on a par with those of Table 29.6  Vowel correspondences in Tungusic Proto-Tungusic

*i *ɪ *ə *a *u

Benzing (1955a)

*i *ï *ä *a

*ü

*u

*ö

*o

Ewen

i

ɪ

ə

a

i/u

ʊ

u/o

ɔ

Oroqen

i

ɪ

ə

a

i/u

ʊ

u/o

ɔ

Oroch

i

i

ə

a

i/u

ʊ

u

ɔ

Udihe

i

i

ə

a

i/u

u

u

ɔ

Nanai

i

ɪ

ə

a

u

ɔ

u

ɔ

Uilʹta (Orok)

i

ɪ

ə

a

u

ʊ

u/o

ɔ

Manchu

i

i

ə

a

u

Joseph and Whitman 2013; Ko et al. 2014

*ʊ

*o

*ɔ

ʊ/u u (~ ə)

ɔ

Table 29.7  Vowel correspondences in Mongolic Proto-Mongolic

*i

*ə

*a *u

*ʊ

*o *ɔ

Poppe (1955)

*i

*e

*a *ü

*u

*ö *o

i

e

a

u

ʊ

o

ɔ

Chakhar i, ɪ

ə

a

u

ʊ

o

ɔ

Baarin

i

ə

a

u

ʊ

o

ɔ

Kangjia

i

e

a

u

ʊ

o, u

ʊ, ɔ

Monguor

i

i, e

a

u

u, o

o, u

o

Bonan

i, ɯ

ə

a

u

u

o

o

Santa

i

ie, ə

a

u

u

o

o

Moghol

i

e

a, o

u

u

o

o

Buriat

i

e

a

u

ʊ

u

ɔ

Khamnigan

i

e

a

u

ʊ

u

ɔ

Dagur

i

ə

a

u

ɔ, wa

u

ɔ

Kalmyk

i

e

a

y

u

ø

o

Mongolian Proper

a

Khalkha

Ko et al. 2014; Ko 2018, modified from Svantesson et al. 2005: 180 a Poppe (1955) and Janhunen (2003: 5) reconstruct two high unrounded vowels, front *i and back *ï in (pre)-Proto-Mongolic. Under our TRH analysis of Proto-Mongolic, the harmonic counterpart of *i will be *ɪ, not *ï. This vowel is not included in the table for the sake of convenience, as Svantesson et al. (2005) do not reconstruct *ɪ for Old Mongolian.

Buriat. On this interpretation, all vowel-harmony languages in Mongolic are exclusively of the retracted tongue-root harmony (RTRH) type. Were we to reconstruct Proto-Altaic on the basis of ProtoKorean, Proto-Mongolic, and Proto-Tungusic facts—regardless of how we reconstruct Proto-Turkic—a mechanical application of the “majority-rules” principle will favor: (i) reconstruction of vowel harmony and (ii) reconstruction of RTRH. But much more important, there is a clear internal phonological route from RTRH to PH, but none in the op­pos­ite direction. As Vaux (2009) points out, there are no known phonetic principles or attested cases that support the shift from PH to RTRH, whereas the reverse shift is phonetically grounded—a tongue-root movement entails a secondary upward and/or forward movement of the tongue body, not vice versa—and well-attested in a number of languages. (See also Ko 2013a, 2018 for further discussion.)

489

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman Table 29.8  Vowel correspondences in Korean

Table 29.9  Reconstructed Proto-Altaic vowel system

Proto-Korean

*i *e *ə *a *ɨ *ʌ *u *o

[Front]

Old Korean (Whitman 2015)

*i *e *ə *a *ɨ *ʌ *u *o

[−round]

[+round]

[−round]

[+round]

Middle Korean (K.-M. Lee 1972)

i

jə

[Closed]

i

y

ɯ

u

[Middle]

e

ø

Northwest Korean

i

[Open]

ɛ

Northeast Korean

ʌ

u

o

jo oa a

ub a

u

o

i

jə

ə

a

ɨ

a

u

o

Central Korean

i

jə

ə

a

ɨ

a

u

o

Southeast Korean

i

jɨ

ɨc

a

ɨ

a

u

o

Southwest Korean

i

jə

ə

a

ɨ

a

u

o

Jeju Korean

i

jə

ə

a

ɨ

ɔ

u

o

ə

a

ɨ

d

Slightly modified from Ko et al. 2014 and Ko 2018, which is based on Kwak 2003, to include PK *e and its reflexes. This is the result of merger of *ə with *o (Kwak 2003). This is the result of merger of *ɨ with *u (Kwak 2003). c This is the result of merger of *ə with *ɨ (Kwak 2003). d This is viewed as a result of later development. a

b

Nevertheless, the conventional view in the literature maintains that vowel harmony in the proto-languages (or the oldest attested stage of the languages, such as Old Mongolian) operated on a palatal/backness contrast, and that the attested RTRH in later varieties is the result of a shift of harmonic contrast from a palatal to a [±rtr] contrast. Accounts of this type include Svantesson’s Mongolic Vowel Shift hypothesis (1985) and K.-M. Lee’s Korean Vowel Shift hypothesis (1964 et seq.).

29.3  Reconstruction of classical Proto-Altaic as [±rtr] It has long been assumed that Proto-Altaic had a vowel system based on a palatal (or backness) contrast. Poppe (1960b) in particular reconstructs the following nine-vowel system for Proto-Altaic, which is based on a PH analysis of Turkic

490

o a

Poppe 1960b: 92

and Mongolic languages and their respective protosystems6 (see Table 29.9). It was noticed as early as Ard (1981; 1984) that of the four branches of Altaic, only Turkic has PH. Vaux (2009) and Ko (2018) argue that reconstruction of RTRH for KMT supports reconstruction of RTRH for the larger putative protofamily (Figure 29.2). This is evident in Poppe’s system, where Korean is one of the two primary branches; the other is what we can call “Proto-Micro-Altaic” (Turkic-MongolicTungusic). Poppe’s tree in fact strengthens the argument for RTRH for Proto-Altaic, since RTRH is found in both primary branches, whereas PH is found in only one (Figure 29.3). Furthermore, any permutation of the internal structure using the four family-branches leads to the same conclusion. On this view, Turkic innovates with respect to the RTRH-to-PH shift. This idea can be formalized as a series of R Altaic P Turkic

R KMT

R Mongolo-Tungusic

29.2.2.2  PH-to-RTRH Shifts in Mongolic and Korean? Ko (2013a, 2018) argues in detail against specific proposals that Mongolic and Korean have undergone shifts from PH to RTRH. The arguments for a vowel shift in Korean depend crucially on evidence from Mongolian loanwords (K.-M. Lee 1964; K.-M.  Lee  1972; Lee and Ramsey  2011). But once Old Mongolian is understood as an RTRH language, these arguments disappear.

[Back]

R Mongolic

R Korean

R Tungusic

Figure 29.2  RTR analysis of Proto-Altaic Ko et al. 2014 6   Ramstedt’s (1952−66 vol. 1) view was slightly more circumspect and nuanced. He noted that there was a geographical dimension to the distribution of different vowel inventories: the farther east one goes in the Altaic domain, the more retracted the pronunciation of the “front” vowels *ä, *ö, *ü. In other words, in more easterly varieties of Mongolic including Khalkha and Buriat, as well as Tungusic and Korean, these vowels are back vowels. He assumed that the western (essentially Turkic) type was original, but presented virtually no evidence, taking the Altaic vowel harmony system itself as the crucial piece of information: that is, Altaic vowel harmony had an apparently clear phonetic/phonological basis in Turkic, i.e., the presumed vowel backness contrast. At the time, TRH systems like that of Tungusic and Korean were unfamiliar and seemed to lack any such clear basis, so it was natural to assume that they had resulted from secondary developments.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian R Altaic R Micro-Altaic P Turkic

R Korean

R Mongolo-Tungusic R Mongolic

R Tungusic

Figure 29.3  RTR analysis based on Poppe’s (1960b: 8) family tree (simplified) Ko 2018: 247; cf. Vaux 2009 Proto-Altaic: [low] > [coronal] > [RTR] > [labial]

[low] > [coronal] > [dorsal] > [labial]

Proto-Turkic: [low] > [dorsal] > [labial]

(= Proto-Tungusic)

(reanalysis)

(fusion)

Figure 29.4  Turkic shift: A hypothesis Ko 2018

changes in contrastive hierarchy (Ko 2018: 258; cf. “contrast shifts” in Dresher et al. 2014) consisting of “reanalysis” ([α rtr] > [α dorsal]; cf. Kalmyk/Oirat) and “fusion” as shown in Figure 29.4. We emphasize here that we do not intend the hypothesis of a “Turkic shift” as an argument “for” or “against” ProtoAltaic as a phylogenetic hypothesis. Our point here is merely that if Proto-Altaic is a valid genetic unity and if vowel harmony is reconstructed for the proto-family, a shift from RTRH to PH is much better motivated than a shift in the opposite direction. Recent research suggests that the shift to PH in Turkic may be shallower. Washington (2016) observes that the articulatory basis for vowel harmony in Kazakh and Kyrgyz involves retraction and advancement of the tongue root and that vowel harmony in these languages also has the acoustic properties of a tongue-root contrast. As noted in Section 29.2.2.1, Washington also suggests, following Pavlov (1983: 35–40, 65–83) that Kalmyk/Oirat is articulatorily a TRH language rather than a PH language as generally claimed in the literature (e.g., by Svantesson et al. 2005). For Mongolic, this ana­ lysis would eliminate the need for contrast shift by reanalysis in Kalmyk/Oirat, such that Mongolic as a whole can be characterized as an RTRH family along with Tungusic and Korean.

For Turkic, it is noteworthy that Kazakh and Kyrgyz belong to the Northwest/Kipchak branch of Turkic in standard classification schemes (e.g., Johanson 2016a; Johanson, this volume: Chapter 8). On the one hand, the presence of any TRH languages in Turkic strengthens the argument that Proto-Altaic should be reconstructed as a TRH system. On the other hand, the hypothesized distribution of TRH systems within Turkic—according to this admittedly incomplete picture—is logically compatible with either innovation or retention. In other words, it is conceivable that ProtoTurkic was itself also originally a TRH system, and that most subordinate taxa shifted to PH systems.7 However, on balance, the evidence continues to support the “Turkic shift” hypothesis, that Proto-Turkic had a PH system and that some individual languages or perhaps even Northwest/ Kipchak Turkic became a TRH system via contact.

29.3.1  Ramstedt’s Proto-Altaic as a [±rtr] system The seminal modern statement of the Altaic hypothesis is that of Ramstedt, in particular the three-volume Einführung in die altaische Sprachwissenschaft (1952–66). In this section, we present the system of vowel correspondences proposed by Ramstedt (1952–66 vol. 1) to account for proposed lexical comparisons across the core Altaic taxa of Turkic, Mongolic, and Tungusic, as well as Korean.

29.3.1.1  Proto-Altaic vowel inventory according to Ramstedt (1952–66) Ramstedt’s eight-vowel system is given in Table 29.10 in his original notation.8 Ramstedt (1952–66 v. 1: 137) expressed some doubt about the necessity of reconstructing an opposition between front *i and back *y (that is, central [ɨ] or [ɯ] as in Turkic) for the proto-language, given that the contrast seemed to be lost in so many descendent languages. He set up *y provisionally, and suggested that the original phonetic difference with respect to *i must have been less salient than that between Turkic /ɯ/ and /i/. 7   This idea is supported on a theoretical basis by Vaux’s view (1999) that uvulars are conditioned by retraction of the tongue root (i.e., [−atr] specification in his feature theory). Under this view, the presence of uvulars in many Turkish dialects and Turkic languages is readily accounted for if we assume that the proto-language had a TRH system. See Vaux (1999, 2009) for more on this. 8   Ramstedt’s transcription symbols generally have values according to the Uralic phonetic tradition. Where needed, we have supplied approximate IPA equivalents.

491

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman Table 29.10  Proto-Altaic vowel inventory Back

Front

Rounded

Unrounded

Rounded

Unrounded

Narrow (close)

u

y (i̮a)

ü

i

Wide (open)

o

a

ö

e (äb) (e̮c)

Ramstedt 1952–66 < i̮ > = [ɨ ~ ɯ] (high central or back) < ä > = [ɛ ~ æ] (lower front) c < e̮ > = [ɘ ~ ə ~ ɜ] (mid central) a

b

Table 29.11  Proto-Altaic vowel correspondences (first syllable) Proto-Altaic

*a

*o

*u

*y

*e

*ö

*ü

*i

Turkic

a

o

u (o)

y

e, ä

ö

ü (i)

i

Mongolic

a

o

u

i

e

ö

ü

i, e

Tungusic

a

o, u

u (o)

ia͔

e̮ (e͔b ~ ä͔c > ȯd)

u, u̇e

u̇, i

i, e

Korean

a

o

o

i

e̮ (o)

u (e̮?)

i ̮ (u)

i

Ramstedt 1952–66 Parenthesized reflexes in the table above represent important variant outcomes in particular daughter languages or later diachronic stages of development. < i͔ > = [i ̠ ~ ï ~ ɪ] < e͔ > = [e̠ ~ ɛ̠] c < ä͔ > = [ɛ̠ ~ æ̠] d < ȯ > = [ɵ] ~ [ɞ]? e < u̇ > = [ʉ]? a

b

29.3.1.2  Vowel quantity according to Ramstedt (1952–66) Ramstedt (1952–66 v. 1: 164–6) was also skeptical about the reconstruction of primary long vowels in Proto-Altaic, instead emphasizing the quite numerous processes that gave rise to historically secondary long vowels in many Altaic languages or dialects. He notes that Proto-Tungusic and Proto-Korean seem to have the requisite evidence for a primary length contrast, but he does not elaborate on that observation.

29.3.1.3  Reflexes in descendent languages Ramstedt’s Proto-Altaic vowel inventory encompasses the following hypothesized regular correspondences in the first syllable in Table 29.11. Ramstedt notes that the reflex of Tungusic /i͔/ (retracted or centralized /i/) is /e/ in Goldʹ (Nanai). That is, Ramstedt’s /i͔/ is equivalent to the “dim” vowel *ï in Benzing’s (1955a) ProtoTungusic system and *e in Cincius (1949b)’s reconstruction.

492

Tungusic and Korean /e̮/ is a schwa-type mid central vowel. Ramstedt provided a contemporary IPA equivalent [ë] (centralized [e]), but in current notation the value can be represented as mid central unrounded [ɘ ~ ə ~ ɜ]. For both Tungusic and Korean, this vowel is understood to have been retracted initially from the original mid front position of PA *e yielding a “mid-mixed dull ö (as in English turn, girl)” (Ramstedt 1952–66 vol. 1: 137), equivalent to Benzing’s PTg *ä and Cincius’s PTg *э. (The Tungusic reflex notated < ȯ > ought to indicate an upper-mid back round vowel between [o] and [u], but the discussion suggests a mid central round vowel [ɵ] ~ [ɞ].) In Korean as well as some Tungusic dialects, /e̮/ from PA *e “culminated in a true o-sound (as in French homme)” under certain conditions. Ramstedt likewise analyzed PA *ö as having undergone gradual retraction in Tungusic to a “front, narrow” < u̇ > “like the u in Swedish lund, hund, huld” and ultimately approaching < u > “like the u in German gut” in Korean. Significantly, he characterized the putative Korean and

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian Tungusic retraction of PA *e and *ö as an innovation of the eastern Altaic zone. This hypothesis will be examined in greater detail in Sections  29.3.1.4 and 29.3.1.5; we note in passing, however, that Ramstedt included eastern Mongolic, Khalkha, and Buriat reflexes such as < ɯ̬ > (“like Swedish u in du, nu, sju”) as examples of the process affecting PA *ö. Furthermore, Ramstedt (1952–66 vol. 1: 138) states that PA *ü gave Tungusic /u̇/ and /i/. (This corresponds to Benzing’s PTg *ü and Cincius’s PTg *уи.) In Korean, PA *ü gave /i/̮ (a back i-sound equivalent to Turkish /y/, thus [ɨ ~ ɯ]) via retraction and delabialization. Mongolic reflexes are said to vary according to dialect, giving a true /ü/ (IPA [y]) in Kalmyk, but more retracted reflexes in Khalkha /ɯ/ [ÿ ~ y̠ ~ ʏ̝]; Buriat /u̇/ [ʉ]; and eastern Mongolic /ɯ/ [ÿ ~ y̠ ~ ʏ̝] or /u͕/ [u̟ ~ ü]. In addition, Ramstedt notes that Turkic frequently has /i/ [i] as a reflex of PA *ü rather than the regular /ü/ [y]. Ramstedt (1952–66 vol. 1: 139) makes the important claim that the Proto-Altaic front vowels *ä, *ö, *ü thus tend to “retain” their front articulation in daughter languages in the western part of the Altaic zone, but that the farther east one goes, the more retracted (farther back) their articulation becomes. He further claims that vowel-harmony patterns demonstrate that the Turkic values are conservative in comparison to the innovative central and back reflexes of eastern languages, but the argument is not presented.

29.3.1.4  Reinterpreting the Ramstedt system We have argued that Proto-Korean, Proto-Tungusic, and Proto-Mongolic must be reconstructed as TRH systems rather than PH systems, as in Table 29.12. Table 29.12  Vowel correspondences among Proto-Mongolic, Proto-Tungusic, and Proto-Korean (first syllable) PMo

*a

*ɔ

*ʊ *i *e

*o

*u

*i, *e

PTg

*a *ɔ, ʊ *ʊ *ɪ *ə

*o

*u

*i

PK

*a

*o

*o *i *ə *u (~ *ə?) *ɨ (~ *u?)

*i

TRH-based revision of Ramstedt (1952–66)

As noted above, recent phonetic studies have called into question the longstanding assumption that Turkic is uniformly of the PH type. We suspect that further research will reveal that at least some other Turkic languages also employ an articulatory contrast involving tongue-root position, but we provisionally assume that Proto-Turkic was in fact a PH system similar to Turkish. We thus revise the Turkic reflexes to appropriate IPA representations as in Table 29.13, and juxtapose them with the revised Proto-Altaic reconstructions based on the TRH analysis (Table 29.13).

29.3.1.5  Ramstedt’s Proto-Altaic forms under the tongue-root hypothesis This interpretation of the Proto-Altaic vowel inventory can be applied to Ramstedt’s reconstructed Proto-Altaic forms. Here, we offer a few examples of better-known com­pari­ sons. We do not necessarily endorse any of Ramstedt’s cognate sets or reconstructions; the immediate goal is simply to demonstrate the look of Ramstedt’s Proto-Altaic when the “eastern” type inventory is treated as conservative (see Table 29.14). This reinterpretation is not merely a notational modification. Although the abstract phonematic relations among reconstructed vowel phonemes as well as the harmony class division are undisturbed, this revision bears important consequences. The distinctive feature [±back] (and/or [±front], [palatal]) is not the same as [±rtr] (and/or [±atr]). The acoustic and articulatory properties of tongue-root systems make particular synchronic and diachronic phonological processes more or less likely than in palatal/backness systems. Moreover, the hypothesis that Proto-Mongolic and ProtoAltaic, in particular, were both tongue-root systems leads to different conclusions about which languages and dialects are innovative under each of those nodes, and thus to a different diachronic narrative of the subgrouping under each node—at least as viewed from the perspective of the vowel systems. Looking beyond Altaic, the tongue-root analysis also aligns Transeurasian languages with a broader areal grouping in Northeast Asia, and perhaps beyond. (We return to this topic in Section 29.4.)

Table 29.13  Proto-Altaic vowel correspondences (first syllable) PA

*a

*ɔ

PTk

*a

*o

PMo

*a

*ɔ

PTg PK

*ʊ

*ɪ

*ə

*u (o) *ɯ *e, *ɛ

*o

*u

*i

*ø

*y (i)

*i

*ʊ

*i

*e

*o

*u

*i, *e

*a *ɔ, *ʊ

*ʊ

*ɪ

*ə

*o

*u

*i

*a

*o

*i

*ə

*o

*u (~ *ə?) *ɨ (~ *u?)

*i

Ramstedt (1952–66) revised

493

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman Table 29.14  Examples of Proto-Altaic forms under the tongue-root hypothesis *a

PA *ǯapa-

Turkish

jap- ‘to make, to build’

R *ǯapa-

Kyrgyz

ǯap- ‘to make, to build’

WMo

ǯaga-/ǯiga- ‘to assemble, to put together, to combine, to unite, to join’

[PMo *ǯaga-

Dagur

ǯaa- ‘to busy (oneself) with’

~ *ǯaba-]

Khalkha

dzā- ‘to split up, to divide’

Buriat

zā- ‘to assemble, to put together, to combine, to unite, to join’

Kalmyk

zā- ‘to arrange, to divide’

Ewenki

ǯawa- ‘to take, to hold, to grab’

Solon

dʹžao- ‘to take, to hold, to grab’

Manchu

ǯafa- ‘to take, to hold, to grab’

Nanai

ǯapa- ‘to take, to hold, to grab’

Korean

čap- ‘to take, to hold, to grab’

PA *mɔrɪ-

Turkish

mor ‘horse’

R *mory-

Uighur

mor ‘horse’

Uriyangqai

mor ‘horse’

Khalkha

morin ‘horse’

Ewenki

murin ‘horse’

Korean

mor, ma̮r, ma̮l ‘horse’

PA *pʊla-

MTk

ulas ‘reddish’

R *pula-

MMo

hula’an ‘red’

WMo

ulagan ‘red’

Ewenki

hulama, hularin, etc. ‘red’

Manchu

fulgijan ‘red’

Jurchen

fûh-lâh-kiāŋ ‘red’

Korean

pulgi̮n ‘red’

PA *ɪra-

Kazakh

yzaŋ, zaŋ ‘furrow at the border of fields; planted garden bed’

R *yra- (?)

Chaghatay

yzan ‘furrow at the border of fields; planted garden bed’

Chuvash

jə̑ran ‘furrow at the border of fields; planted garden bed’

Mongolian

iraga ‘ripples on the surface of water’

(Nanai) Goldʹ

irun ‘furrow at the border of fields; planted garden bed’

Manchu

irun, iri ‘furrow at the border of fields; planted garden bed’

Korean

iraŋ ‘furrow, boundary’

a

[PTg *ǯapa-]

*ɔ

*ʊ

*ɪ

494

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian *ə

*o

*u

PA *dəgV-

Old Turkish

jeg ‘better, superior’

R *degV-

Chuvash

śi ‘top, surface’

Mongolian

dege-re ‘above, on, over’

Ewenki

dīlē̮ ‘above’

Manchu

dele ‘above, on’

Korean

te̮ ‘more, still more’

PA *mor

Mongolian

mören ‘river’

R *mör (?)

Ewenki

mu̇ ‘water’

Manchu

muke ‘water’

Korean

mul ‘water’

Turkish

tüš- ‘to fall, drop, sink, set, descend’

PA *tuR *tü- (?) [PTg *ti-,

tüšür- ‘to let drop, to fell’ Ewenki

tigde̮ ‘rain’

Goldʹ

tugde̮ ‘rain’

*tu- ‘to fall’]

tulsi- ‘to fall, drop (frequently)’ tugbu- ‘to let fall, to let drop’

*i

a

Korean

či- ‘to fall, to drop, to sink’

PA *sil(ə)

Turkish

sil, silä ‘broth’

R *sil(e) ?

Mongolian

šilü ‘broth’

Kalmyk

šüln̥ ‘broth’

Ewenki

sile̮ ‘broth’

Korean

sul ~ sjūl ‘(distilled) liquor, spirits’

Forms marked “R” are in Ramstedt’s system. All daughter-language reflexes are given in his notation, with his glosses.

29.3.2  Poppe’s Proto-Altaic as a [±rtr] system

29.3.2.1  Proto-Altaic vowel inventory according to Poppe (1960b)

The system of Altaic reconstruction of Nikolaus Poppe, particularly the version contained in his Vergleichende Grammatik der altaischen Sprachen (1960b), agrees with the Ramstedt system in many details. In this section, we ­present the system of vowel correspondences proposed by Poppe (1960b) to account for proposed lexical comparisons across the core Altaic taxa of Turkic, Mongolic, and Tungusic,9 as well as Korean—the same taxa as in Ramstedt (1952–66).

Poppe’s nine-vowel system is given in Table  29.15 in his original notation.

  Poppe also uses the term “Manchu-Tungus(ic) [Mandschu-tungusisch]”, apparently interchangeably with “Tungus(ic) [Tungusisch]” in some cases. Elsewhere, it is clear that “Tungus [Tungusisch]” refers to northern Tungusic languages such as Ewenki and Ewen. 9

Table 29.15 Proto-Altaic vowel inventory according to Poppe (1960b) Front Unrounded Close

*i, *iˉ

Mid

*ė, *ėˉ

Open

*ä, *ǟ

Back Rounded *ü, *ǖ

Unrounded *ï, *ïˉ

*ö, *ȫ

Rounded *u, *ū *o, *ō

*a, *ā

495

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman Table 29.16  Proto-Altaic vowel correspondences (short vowels, first syllable) according to Poppe (1960b) PA

*a

*o

*u

*ï

*ä (= *e)

*ö

*ü

*i

Turkmen

a

o

u

ï (i)

ä

ö

ü

i

Yakut

a

o

u

ï

ä

ö

ü

i

u

ï

ä

ü

i

Azeri

a

i- < *ai̮-b < *aCu- < *Ca-

Ci-̮ < *Co-

ʋǝ̑- < *uCǝ̑- < *Cu-

Cǝ̑- < *Cïšu- < *sï-

a

ʋa- < *öCǝ̑- < *Cö-

i, ə

PMo

*a

*o

*u

*ï

*e

*ö

*ü

*i

WMo

a

o

u

i

e

ö

ü

i

Kalmyk

a

o

u

i

e

ö

ü

i

u

u

o

u

Chuvash

Monguor “Tungus” (= Ewenki / Ewen)

a

o

u

i̮

e̮c

u, e̮

u, i

i

Manchu

a

o

u (~ o < *u)

i

e

u

u

i

Korean

a

o, u

u (~ a̮/o)

i, je̮

e̮

u, i̮, e̮

i ̮, u

i

< ï > = [ɨ ~ ɯ] < i̮ > = [i ̠ ~ ï ~ ɪ] (NB: < i ̮ > is also used for Korean /ɨ/.) c < e̮ > = [ɘ ~ ə ~ ɜ] a

b

Poppe’s (1960b) system has an additional vowel quality in comparison to Ramstedt’s (1952–66) system. He tentatively reconstructed this additional front vowel quality *ė (a higher [e] vs. the lower *ä [ɛ ~ æ]) to account for certain otherwise unexplained correspondences between Turkic and Mongolic. (The lower Proto-Altaic vowel is transcribed *ä here, but is also rendered *e throughout in Poppe’s system (1960b)). The higher vowel *ė is also reconstructed for Proto-Turkic and Proto-Mongolic but not Proto-Tungusic or Proto-Korean, and the contrast between *ė and *ä is said to be present in Middle Mongolian and some other spoken Mongolic languages (Poppe 1960b: 92). On the other hand, Poppe suggests that *ė may be merely a secondary development within Mongolic, and in­de­pen­ dent­ly also a secondary development within Turkic. According to Poppe (1960b: 102–5), in both branches, *ė is at least partly derived from *ä (= *e) in the neighborhood of palatal consonants. Note also that the putative *ė lacks a “[back]” counterpart in Poppe’s system.

29.3.2.2  Vowel quantity according to Poppe (1960b) Poppe reconstructed a full set of primary long vowels for Proto-Altaic, as he does for Proto-Turkic, Proto-Mongolic, and Proto-Tungusic.

496

29.3.2.3  Reflexes in descendent languages 29.3.2.3.1  Short vowels Poppe’s Proto-Altaic inventory of short vowels encompasses the following hypothesized regular correspondences in the first syllable, see Table 29.16.10 29.3.2.3.2  Long vowels Poppe’s Proto-Altaic inventory of long vowels encompasses the following hypothesized regular correspondences in the first syllable in Table 29.17.

29.3.2.4  Poppe’s Proto-Altaic forms under the tongue-root hypothesis The following are a few examples of better-known com­pari­ sons, applying the tongue-root analysis of the Proto-Altaic vowel inventory to the reading of Poppe’s reconstructed Proto-Altaic forms. As with the examples for Ramstedt’s 10   Poppe does not give a complete set of reflexes for Azeri, Chuvash, Monguor, or Manchu, and there is no formal statement of Korean vowel reflexes at all. In this table, we have added reflexes observed in Poppe’s lexical comparisons in shaded cells. Furthermore, we have ignored the tentatively reconstructed ninth vowel, PA *ė. Note that only Proto-Mongolic is a reconstructed system; the other inventories are based on the modern languages.

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian Table 29.17  Proto-Altaic vowel correspondences (long vowels, first syllable) according to Poppe (1960b) PA

*ā

*ō

*ū

*ïˉ

*ǟ (= *ē)

*ȫ

*ǖ

*iˉ

Turkmen

ā

ō

ū

ˉï ( ī )

ǟ

ȫ

üj

ī (ij)

Yakut

ā

uo

ū

ˉï

iä

üö

ǖ

ī

ï

ä (~ e?)

ö

a

ʋa- < *ȫCǝ̑ʋa- tuku-no ‘moon-gen’ *pə-nə > po-no ‘fire-gen’ *me-na > mi-na ‘water-gen’ *ta-na > ta-na ‘hand-gen’

(ii)

Verb stem + honorific *kik-əs > kik-os ‘hear-hon’ *əp-əs > op-os ‘bear-hon’ *i-tor-as-i-te > i-twor-as-i-te ‘hon-take-hon-inf-ger’

Table 29.23  Nivkh alternations for possessive prefixes ñi-řlə

1sg-sky  ‘my sky’

ñe-řla

1sg-harpoon

‘my harpoon’

ț‘i-řlə

2sg-sky  ‘your sky’

ț‘e-řla

2sg-harpoon

‘your harpoon’

ji-řlə

3sg-sky  ‘his sky’

je-řla

3sg-harpoon

‘his harpoon’

p‘i-řlə

refl-sky 

p‘e-řla

refl-harpoon

‘self’s harpoon’

‘self’s sky’

vowel in certain Amami varieties reflects PJ *e. We are grateful to Elisabeth de Boer for the corrected form *ta-na for ‘hand-gen’. Whitman (2016) reconstructs the same vowel inventory for Proto-Japonic, pointing out that the Proto-Japonic harmonic opposition is interpretable as a TRH system; it is in fact identical to the six-vowel systems of Nivkh and among Tungusic languages, Nanai, which are interpreted by Gruzdeva (2015) and Ko et al. (2014) respectively as RTR harmony systems. Table 29.23 shows the Nivkh alternations for possessive prefixes plus řle ‘sky’ and řla ‘harpoon’.11 In Old Japanese the [-rtr] allomorph of the genitive *nə > OJ no surfaces in non-harmonizing environments (such as when the genitive attaches as a clitic to free noun stems), suggesting that the language is [+rtr] dominant.12   In Section 29.4.3 we give a fuller discussion of RTRH in Nivkh.   Elisabeth de Boer points out to us that the many compounds formed with the bound noun ama ‘heaven, sky’ selecting genitive ama-no (e.g ama11 12

The second factor limiting the scope of harmonic alternations in Old Japanese is the rise of a new set of verbal endings from stem + auxiliary or infinitive + auxiliary, where the auxiliaries, originating as independent words, were never the targets of harmony. Similar factors seem to have been at work in Nivkh. Nevertheless, the question remains why the verbal inflectional endings, infinitive -i, conclusive -u, and attributive -u do not show harmonic alternations in Old Japanese.

no gawa ‘heaven-gen river, Milky Way’) rather than the expected but ­unattested ama-na constitute a counterexample to this (or any) account of vowel harmony in the genitive forms of bound nouns. We have no definitive explanation for this example, but we suggest that given the large number of genitive compounds involving the free form of the noun, ame (e.g. ame no sita ‘heaven gen below ‘under the heavens’), analogy may be at work here. This seems particularly likely in the instance of ama-no gawa ‘Milky Way’, since rendaku (sequential voicing) on gawa ‘river’ is hard to explain without appealing to some kind of analogy.

503

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman In fact, there is an argument that the last of these endings does show a harmonic alternation in Eastern Old Japanese. It is well known that the EOJ verbal attributive is attested in some texts as [+rtr] *-o > -wo and the adjectival attributive as [+rtr] *-ke > -kye, contrasting with Central OJ -ki. The [+rtr] variants are not attested in a majority of tokens in the Eastern Old Japanese corpus, leading researchers to point to regional differences within Eastern Old Japanese and the likelihood of faulty transcription. But the data suggest that the identity of the stem vowel plays a role in the selection of the [+rtr] variants. Of the 52 tokens of EOJ verbal attributive -(w)o collected by Ikier (2006), the largest single number follow auxiliaries with final stem vowel /a/: future -am- (20 tokens), tentative -unam- (5 tokens), negative -an- (3 tokens), and progressive -ar- (2 tokens); to these may be added perfective -n-, reconstructable as *-na-. When directly affixed to lexical verb stems, the -(w)o allomorph appears 11 times after final stem vowel /a/, 4 times after /i/ or /u/ (both of which may be diachronically derived from a [+rtr] or [-rtr] source), but only once after [-rtr] *ə > /o/. However, this single exception, involving the verb *əməp > omop- ‘think’, involves the syllable /mə/, where [-rtr] *ə > /o/ had merged with [+rtr] *o > /wo/ in Eastern Old Japanese. Previous research has suggested that the Eastern Old Japanese [+rtr] attributive forms preserved the original shape of the attributive suffixes in Proto-Japonic. But this view has trouble explaining the high incidence of the [-rtr] allomorphs in the same data, sometimes in the same poem. The interpretation of the alternation in the Eastern Old Japanese attributives as the product of TRH suggests an alternative explanation: that what the Eastern Old Japanese data preserve is a reflex of vowel harmony. We return to the question of why other verbal inflectional endings show no trace of vowel harmony. As de Boer (1996) points out to us, this is also true of other parts of the Old Japanese adjectival paradigm, adverbial -ku and conclusive -si. Linguists agree that adjectival inflection is relatively shallow in Japonic: adjectival conclusive -si, for example, cannot be reconstructed for Proto-Ryūkyūan. It seems likely the adjectival paradigms emerged in varieties of Japonic after vowel harmony ceased to be productive. The nonfinite forms adverbial -ku and the verbal infinitive -i occurred in nonfinal positions, where they were ­targets for mid vowel raising, which merged *e with *i and *o with *u (this process explains why the OJ reflex of PJ *me- ‘water’ is mi-). As suggested above, mid vowel raising would have had the effect of eliminating vowel harmony in most prosodic positions for non-low vowels. It seems likely that this process played a major part in the loss of pro­duct­ ive vowel harmony across Japonic.

504

29.4.3  TRH and tongue-root vowel inventories in Northeast Asia As already mentioned, Ramstedt (1952–66 vol. 1: 137–8) notices that there is a geographical (east–west) cline in the distribution of the inventory types we have analyzed as TRH and PH within Altaic, as he defined it (namely Turkic, Mongolic, Tungusic, Korean). Specifically, he pointed out that the farther east one goes within Altaic, the more retracted the pronunciation of the reflexes of his PA “front” vowels *ä, *ö, *ü. Although he did not elaborate his reasoning, he stated clearly his assumption that the “eastern” type—i.e., the TRH type—was innovative. Janhunen (1981b) points out that across Northeast Asia more generally, “apertural harmony” (our RTRH) is an eastern feature, while “palato-velar harmony” (our PH) is a western feature. With respect to Altaic languages, we have argued above that the domain of the “eastern” RTRH feature extends farther west, into the center of the region, to encompass (at least) Korean, Mongolic, and Tungusic. Janhunen explains the near-complete predominance of RTRH in modern KMT languages as the result of diachronic “vowel rotation” as in Figure 29.5. The vowel rotation hypothesis is similar to Svantesson’s (1985) “velarization” hypothesis for Mongolic (cf. Section 29.2.2). Janhunen (2003c) likewise adopts a version of this hypothesis to account for “rotated” inventories in Mongolic generally. On the other hand, Janhunen observes that in some other Northeast Asian families, RTRH may be a primary feature. As Janhunen points out, modern Chukotko-Kamchatkan languages generally involve some reduction of an original [±rtr]-dominant system. Fortescue (2005) reconstructs the Proto-Chukotko-Kamchatkan vowel inventory as follows (Bobaljik 2009 reconstructs an identical seven-vowel inventory apart from substituting *ɛ for *æ.): *ü1

*ö3

*ä5

*u2

*o4

*a6

*ü →

*u *ö *o *ä

→

*ŭ3

*u1

*ŏ5

*o2

*a6

*ɔ4

*a

Figure 29.5  “Vowel rotation” in Northeast Asia: the Korean case Janhunen 1981b

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian In modern Chukchi, *æ and *e have merged; in Koryak varieties, *æ merges either with *a or with *e. However, even after vowel mergers, harmony generally survives as a stem property (except where vowel harmony is lost entirely, as in varieties of Alutor and Kerek in which all harmonic pairs have been neutralized). Furthermore, although no Chukotko-Kamchatkan language attests a surface distinction, the vowel *ə is covertly harmonic, in the sense that certain stems consistently have a dominant /ə/, while ­others consistently have a recessive /ə/—and this behavior is stable across harmonic languages in the family. Thus, especially as seen from an areal perspective, a plausible internal reconstruction would suggest a “perfect” eightvowel system with [±rtr]-dominant harmony, in which PCK *ə resulted from an earlier merger of a harmonic pair of ­vowels that can be represented provisionally as *ɨ (recessive/[-rtr]) versus *ə (dominant/[+rtr]), as in Table 29.25. In Nivkh, too, Janhunen (1981b: 139–40) points out the evidence for an apertural (tongue-root) harmonic opposition, based on an alternation in prefix shape between i- before /u, ə/ and e- before /o, a/. Shiraishi and Botma (2013; see also Shiraishi and Botma 2016) survey 335 disyllabic roots in a spoken corpus. They find that /o/ never co-occurs steminternally with /ə/, and that /a/ never co-occurs with /u/ and only once with /ə/, suggesting vestiges of stem-internal co-occurrence restrictions based on a [±rtr] opposition. Another candidate for vestigial RTRH, this time to the north of KMT, is Yukaghir. Nikolaeva (2006: 57ff) reconstructs a PH system for Proto-Yukaghir. But as Maslova (2003a: 35, fn. 8) notes, attributing the observation to Bernard Comrie and Christian Lehmann, the present-day Kolyma Yukaghir harmony “might be more appropriately described as (advanced) tongue root (rather than palatal) harmony.” In the Kolyma Yukaghir system, /e/, /ø/ contrast with /a/, /o/. The high vowels /i/, /u/ are transparent, but stems with /i/, /u/ Table 29.24  Proto-Chukotko-Kamchatkan vowel inventory Recessive

*i

*u

*æ

Dominant

*e

*o

*a

Transparent

*ə

Fortescue 2005

Table 29.25  Pre-Proto-Chukotko-Kamchatkan vowels Recessive/[-rtr]

*i

*ɨ

*u

Dominant/[+rtr]

*e

*ə

*o

Recessive/[-rtr]

*æ

Dominant/[+rtr]

*a

Based on Fortescue 2005

Table 29.26  Proto-Yukaghir vowels (i) Nikolaeva (2006: 57) Front

*i

*e

*ö

(*ü)

Back

*y

*a

*o

*u

(ii) RTRH reinterpretation [-rtr]

*i

*e

*o

*u

[+rtr]

*ɪ

*a

*ɔ

*ʊ

normally belong to the same class as /e/, /ø/, with the majority of exceptions involving /i/. The matter awaits further investigation, but Nikolaeva’s (2006: 57) reconstruction might be reinterpreted as indicated in Table 29.26. Neither Kolyma nor Tundra Yukaghir attests *y ([ɨ]/[ɯ]) or *ü ([y]) (Nikolaeva 2006: 57). On the RTRH interpretation in Table  29.26 (ii), this reflects loss of the tongue-root ­contrast for the high vowels, a typologically and areally common development, as we have seen. Thus, it would seem that there are scarcely any Northeast Asian language families (including isolates) that do not show some evidence of phonological organization of the vowel inventory around a contrast referencing tongue root position.13 In the following section, we briefly examine ­similar patterns in neighboring North America.

29.4.4  TRH and tongue-root vowel inventories beyond Eurasia? Janhunen (pers. comm.) has suggested that the linguistic area relevant for the spread of “apertural” (tongue-root) vowel inventories might extend beyond Eurasia into North America. Our view is that Eskimo-Aleut, Sahaptian, and Salishan in particular all exhibit evidence of tongue-root organization of some kind. 13   Yet another candidate for vestigial RTRH in the region is Ainu. Citing Chiri (1952), Shibatani (1990: 13–16) describes two suffixation processes that trigger what he refers to as “vocalic euphony.” In the first process, transitive verbs are derived from certain bound verb stems by suffixation of a single vowel segment, which may be any of the five phonemic vowels /a, e, i, o, u/. However, the suffix vowels are not interchangeable, and they are subject to certain restrictions on their co-occurrence with stem vowels. Specifically, /a/ and /o/ may not co-occur, and /o/ and /u/ may not co-occur. All other combinations appear to be licit. The second process involves derivation of the “personal form” of nouns, i.e., the form of nouns when they are used “with reference to a possessor.” The suffix of the personal form is again a single vowel segment, obeying exactly the same distributional pattern as the transitive verb suffix. Shibatani suggests that the front vowels /i/ and /e/ can be regarded as neutral, but finds the opposition between /o/ on the one hand and /a, u/ on the other difficult to capture in a phonologically natural way. However, we note that in almost all Tungusic languages, for example, similar co-occurrence restrictions are observed within stems: no **a . . . ɔ, **ɔ . . . a, **ɔ . . . ʊ, **ʊ . . . ɔ, owing to the operation of “labial attraction” (rounding harmony among non-high v­ owels) and general constraints on the distribution of round vowels.

505

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman 29.4.4.1  TRH in North America Of the three families mentioned above, Sahaptian, the family consisting of Sahaptin and Niimi'ipuutímt (aka Nez Percé), spoken in Washington and Oregon (US), shows perhaps the greatest resemblance to TRH systems of the Northeast Asian zone. Rude (2006) reconstructs the ProtoSahaptian short vowel inventory as in Table 29.27. Sahaptian vowel harmony is of the dominance type. The dominant vowels—also called the “strong” vowels—are *o and *ɑ. The corresponding “weak” or recessive vowels are *u and *æ, respectively. The vowels *i and *ɨ are neutral in the sense that they can both occur in “strong” and “weak” vowel contexts. If any morpheme in the phonological word contains either *o or *ɑ, then the entire word is subject to “retraction” (backing and/or lowering) of any underlying *u and *æ to *o and *ɑ, respectively. However, the so-called “neutral” vowels, Proto-Sahaptian *i and *ɨ, are in fact covertly harmonic. Some morphemes with only *i or only *ɨ trigger “strong” harmony in the word. (The rest—the majority—are “weak” or recessive.) The overall Sahaptian system is thus strikingly similar to that of Chukotko-Kamchatkan and, albeit less strikingly, to core Altaic-type [±rtr] systems. The two harmonic pairings of vowels are at different heights, which immediately suggests a tongue-root feature analysis as in Table 29.28. The vowels *i and *ɨ lack [+rtr] counterparts at the surface, but behave as if each had descended from an original harmonic pair, from which the [+rtr] member has been lost by merger with (i.e., neutralization to) its corresponding [-rtr] member (with persistence of harmonic class membership as a stem property). Thus, on internal grounds, it is plausible to reconstruct a pre-Proto-Sahaptian inventory with a “perfect” eight-vowel system as in Table 29.29. Note that this is virtually identical to the vowel inventory we have proposed for pre-Proto-Chukotko-Kamchatkan on precisely parallel internal grounds. In both cases there are only two phonological heights, as also in Proto-Tungusic, and others; and in both cases there is only one pair of round vowels, as also in Middle Korean.14 Table 29.27  Proto-Sahaptian vowel inventory *i

*ɨ

*u *o

*æ

Table 29.28  Proto-Sahaptian vowel inventory (a tongue-root ­proposal) [-rtr]

*i

*ɨ

*u

[+rtr]

—

—

*o

[-rtr]

*æ

[+rtr]

*ɑ

Table 29.29  Pre-Proto-Sahaptian vowel inventory (a tongueroot proposal) [-rtr]

*i

*ɨ

*u

[+rtr]

*e

*ə

*o

[-rtr]

*æ

[+rtr]

*ɑ

29.4.4.2  Other tongue-root systems in North America In Eskimo-Aleut and Salishan, the phonological processes that point to some kind of tongue-root organization of the vowel inventory are different from—and, in an important sense complementary to—those of Northeast Asian TRH languages. In both of these North American families, the evidence comes from processes of vowel-consonant harmony, in which certain classes of consonants trigger allophonic changes in the realization of vowels within domains of varying degrees of locality. Eskimo-Aleut has the widely assumed structure in Figure 29.6. In all three lower branches of Eskimo-Aleut, there is a pervasive process of vowel retraction triggered by uvular consonant phonemes. The common Eskimo-Aleut vowel inventory is /a, i, u/, with the addition of /ə/ in Yupik and certain varieties of Inuit. The peripheral vowels generally also have long counterparts /aː, iː, uː/. The effect of vowel retraction triggered by uvulars in some representative var­ ieties is summarized in Table 29.30. Abstracting away from the subtle differences among the varieties, these facts suggest that a similar allophonic process was present in Proto-Eskimo-Aleut. We assume that just as [+rtr] can condition uvular allophones of underlying velar consonants in Altaic TRH systems, the same feature can

*ɑ

Proto-Eskimo-Aleut

Rude 2006 14  Another intriguing parallel with Middle Korean is the association between [+rtr] vowels and diminutive semantics (as well as between [-rtr] vowels and augmentative semantics) within the sound-symbolic system of Proto-Sahaptian.

506

Eskimo Aleut

Yupik

Figure 29.6  Structure of Eskimo-Aleut

Inuit

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

vowel systems and harmonies in tr anseur asian Table 29.30  Vowel retraction in Eskimo-Aleut Branch

Language

/a/

/i/

/u/

Environment

Aleut

Aleut

[ɑ]

[e]

[o]

adjacent to uvulars /q, χ, ʁ/ Bergsland (1994, 1997)

Yupik

St. Lawrence Island Yupik

[ɑ]

[e]

[o]

adjacent to uvulars /q, χ, ʁ/ Krauss (1975); Badten et al. (2008)

Central Alaskan Yup'ik

no retraction [e]

[o]

adjacent to uvulars /q, χ, ʁ/ Miyaoka (2012) and before /a/

Inuit

Eastern Canadian no retraction [e ~ ɛ] Inuktitut Greenlandic

[ɑ]

[o ~ ɔ] in proximity to uvulars

[e ~ ɛ ~ ɐ] [o ~ ɔ] preceding uvulars /q, ʁ/

also condition retracted (acoustically, primarily ­lowered) allophones of non-retracted vowels (whether by spreading or by some other mode of propagation) in the Eskimo-Aleut languages. The two types of tongue-root system are thus mirror images of each other. The Salishan languages of British Columbia (Canada) and Washington, Oregon, Idaho, and Montana (US) are classified into three primary branches: Nuxalk (aka Bella Coola); Coast Salish (including Halkomelem, Lushootseed, Squamish, etc.); and Interior Salish (Shuswap, Lillooet, Snchitsuʔumshtsn [aka Coeur d’Alene], Spokane-Kalispel-Flathead [aka Montana Salish], etc.). All three primary branches attest a type of vowel–consonant harmony that has been referred to as “post-velar harmony,” “faucal harmony,” “automatic vowel darkening,” or “vowel flattening,” among other terms, but is otherwise quite similar to the “retraction” processes mentioned above for Eskimo-Aleut. For example, Nuxalk (Bella Coola) has a three-vowel inventory /a, i, u/. These are retracted to [ɑ], [ɪ(ː) ~ ɛː], [o̞], respectively, between the post-velar consonants /qʰ, q’, χ/ (Nater  1984).15 The Coast Salish language Halkomelem has a similar process, as do the Interior Salish languages Shuswap, Lillooet, Snchitsuʔumshtsn (Coeur d’Alene), and Spokane-Kalispel-Flathead (Bessell 1998). Thus, it is plausible to assume that Proto-Salishan also possessed a similar allophonic process. A small number of Athabaskan languages also seem to have similar retraction processes. However, in those cases, the distribution has been taken to reflect contact with either Eskimo-Aleut or Salishan, and is not reconstructed for ProtoAthabaskan. For example, the process of “vowel flattening” in Tsilhqot’in (aka Chilcotin) in British Columbia is widely attributed to contact with nearby Interior Salishan languages. 15   Salishan “post-velars” are articulated between the velar and uvular positions. As far as we are aware, they are phonologically non-distinct from uvulars. Within the single phonemic series, there is a tendency for stops to be articulated at the post-velar position, whereas fricatives are articulated at the uvular position.

Source

Denis and Pollard (2008) Fortescue (1990)

So what is the point of this excursion into Northeast Asian and North American vowel systems beyond Altaic/ Transeurasian? It is to demonstrate that, as in the Central Sudanic Zone, tongue-root-type vowel inventories have apparently spread far and wide, across great distances and all manner of genetic taxa. In earlier work (especially Ko et al. 2014), we had already noted that RTRH has to be reconstructed at the proto-family level for Chukotko-Kamchatkan, Nivkh, and possibly Yukaghir, in addition to the KMT languages of the Altaic grouping. If we further extend this assumption to Sahaptian, Eskimo-Aleut, and Salishan, then the conclusion that RTRH in this wide but contiguous region was spread by contact becomes all but inescapable. On the other hand, the fact that RTRH must be reconstructed for so many proto-languages of presumably unrelated families sim­ul­tan­eous­ly indicates that, as a feature, RTRH must be truly ancient, and its spread also correspondingly ancient.

29.4.5  Two types of [±rtr] system: “balanced” vs. “unbalanced”? In the preceding sections, we have presented arguments in favor of reconstructing three of the four Altaic protolanguages as [±rtr] harmony systems and supported the view that this feature should be reconstructed for any putative Altaic (or Transeruasian) proto-language. Among the KMT families, however, Proto-Korean is an outlier in one important respect: it has a [labial] (or [round]) contrast only for the non-low vowels. We might refer to the ProtoMongolic/Proto-Tungusic type as a “balanced” or “fourround” system: the inventories of languages of this type have equal numbers of low and non-low vowels, and each contain two low and two non-low round vowels. Proto-Korean, in contrast, is “unbalanced” in that there are only two low vowels, and round vowels are found only within the non-low height class.

507

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

A ndr ew Joseph, Seongyeon Ko, and John Whitman At the same time, the Proto-Korean non-low central v­ owels *ɨ and *ʌ (the structural analogues of PMo/PTg/PA *o and *ɔ, resp.) are understood by many authors on internal grounds to have a special status. In Late Middle Korean, /ɨ/ and /ʌ/ are restricted in their distribution, occurring not at all (in the case of /ʌ/) or only in a single lexical item (in the case of /ɨ/) in absolute onset position. They are considered to have been the target of syncope (K.-M. Lee 1991; Martin 1996), and are generally characterized as “weak” vowels. Ko (2018) analyzes the Late Middle Korean vowel inventory with the featural specification in Table 29.31. There are two ways to account for the “unbalanced” nature of the Proto-Korean system, from a Transeurasian and broader Northeast Asian perspective. One view would be that the pre-Proto-Korean antecedents of *ɨ and *ʌ were originally round and low. On this view, the system in Table 29.31 would have arisen by restructuring from a pre-Proto-Korean system like the one in Table 29.32. According to this analysis, the pre-Proto-Korean sources of LMK /ɨ/ and /ʌ/ are *o and *ɔ (that is, just like their analogues in Proto-Mongolic/Proto-Tungusic/Proto-Altaic). Delabialization of these vowels, perhaps triggered by weakening, gave rise to a restructuring of the system into one like the Old Korean system in Table  29.8, since loss of the [labial] feature would have eliminated the contrast with *ə and *a. Effectively, delabialization triggered the raising (and centralization) of these vowels. The analysis of the inventory in Table 29.32, motivated by the single change of delabialization, thus brings pre-Proto-Korean into line with Proto-Mongolic/Proto-Tungusic/Proto-Altaic. This is similar in spirit to the vowel-shift hypothesis of K.-M. Lee (1964, 1972), projected back to the pre-Proto-Korean level. Lee also reconstructs the ancestors of LMK /ɨ/ and /ʌ/ as round, *ɔ̈ and *ɔ. Table 29.31 Feature specification of Late Middle Korean vowels i

ə

a

ɨ

ʌ

u

o

[coronal]

+

−

−

−

−

−

−

[labial]

−

−

−

−

−

+

+

[low]

−

+

+

−

−

−

−

[rtr]

−

−

+

−

+

−

+

Ko 2018

508

Table 29.32 Feature specification of pre-Proto-Korean vowels *i

*ə

*a

*o

*ɔ

*u

*ʊ

[coronal]

+

−

−

−

−

−

−

[labial]

−

−

−

+

+

+

+

[low]

−

+

+

+

+

−

−

[rtr]

−

−

+

−

+

−

+

cf. Ko et al. 2014 Here, we temporarily ignore PK *e, which we assume is specified [coronal, +rtr].

A key observation, it seems to us, is that within Transeurasian and Northeast Asia generally, families with “unbalanced” systems like Korean also do not evince labial harmony. Within Transeurasian, this is true of Korean and Japanese. Within Northeast Asia and beyond, it is true of languages to the east as well: Nivkh, Chukotko-Kamchatkan, Eskimo-Aleut, Sahaptian, and Salishan. Rounding harmony is a “western” feature: it is present in Turkic, Mongolic, Tungusic, and Yukaghir. We have seen that the latter three families have both rounding harmony and RTRH, and all three have “balanced” or “four-round” vowel inventories. This leads to the second possible hypothesis about the Proto-Korean ­system: that it represents a primary type within Transeurasian, and the genesis of a “four-round” system is associated with the introduction of labial harmony. At our current state of knowledge it is impossible to choose between these two hypotheses. The next stage of research requires a more thorough investigation of the correlation between RTRH, labial harmony, and vowel inventory configuration.

Acknowledgment Ko and Whitman’s work on this paper was supported by the Laboratory Program for Korean Studies through the Ministry of Education of the Republic of Korea and the Korean Studies Promotion Service of the Academy of Korean Studies (AKS-2016-LAB-2250004).

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

B. Morphology

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Ch a pter 30

A comparative approach to verbal morphology in Transeurasian M a rt i n e R obbe e ts

30.1 Introduction Supporters and critics of Transeurasian relatedness seem to agree on at least this one point, namely that patterned morphology could substantially help settle the controversy. Vovin (2005c: 73) begins his critique of Starostin et al. (2003) with the postulation that “The best way . . . is to prove a suggested genetic relationship on the basis of paradigmatic morph­ology,” whereas Dybo and Starostin (2008: 125) agree that “regular paradigmatic correspondences in morphology are necessarily indicative of genetic relationship.” Since consensus between supporters and critics is a rare commodity in the Transeurasian debate, I will devote this chapter to an overview of the verb morphology shared between the Transeurasian languages and indicate to what extent the correlations may be regarded as “paradigmatic.” Since the beginnings of the historical comparative study of the Transeurasian languages, the emphasis has always been on lexical research. Although the field of contemporary Transeurasian linguistics was practically founded with the posthumous publication of Ramstedt’s Formenlehre in 1952 (Ramstedt 1952–66), few linguists followed in his footsteps, preferring to turn their attention rather to the comparison of sounds, words, or typological structure. The publication of the first part of Poppe’s Vergleichende Grammatik der altaischen Sprachen (1960b, Vergleichende Lautlehre) was intended to precede a second part on comparative morphology, but unfortunately, the comparative phonology was the only volume to appear. Some exceptions to the underrepresentation of comparative verbal morphology include Poppe (1972), Nasilov (1978) and Kormušin (1984), but like Ramstedt’s Formenlehre, none of these contributions took Japanese data into account. Only few comparative studies of Transeurasian verbal morphology include Japanese as well as Korean comparanda. Miller compared deverbal verb suffixes (1981), denominal verb suffixes (1982), negation (1971: 245–84; 1985) and gerunds (1971: 285–91). Menges treated transitivity pairs (1975: 32–5), negation (1975: 96–109; 1984: 262–3) and gerunds (1975: 110–11). Street dealt with denominal verbs (1978: 44,

204, 219), actionality (1978: 74–5, 113, 115, 188, 178–9, 219–20, 247–9, 230–1), diatheses (1978: 53, 55, 71, 75, 179, 181–2, 185–8, 199–200, 208, 230, 239–42, 257–9) and converbs (1978: 200, 251–2). Finch (1987) reconstructed Altaic verb classes. Vovin (1998a) provided a sketch of comparative Altaic verb morphology including Japanese, but in (2001) he restricted the same evidence to a comparison of Japanese, Korean, and the Tungusic languages and later in (2005c) he distanced himself from his former positions by completely rejecting all evidence in support of Transeurasian affiliation. “The Etymological Dictionary of the Altaic Languages”—including Japanese and Korean—published by Starostin et al. (2003: 173–229) provided an overview of morphological elements that look similar on the surface, but being a dictionary rather than a comparative grammar, it did not investigate these look-alikes in more detail. In this chapter, I will present correlations in the verbal morphology between the Transeurasian languages, summarizing my findings on this topic as discussed in more detail in Robbeets (2015). In Section 30.2, I introduce the data set, providing a tabular overview of common verbal morphology. I intend to reserve Section 30.3 for a discussion of the formal correlations while I will focus on functional correlations and indications of paradigmaticity in Sections 30.4 and 30.5, respectively. In Section 30.6, I will assess the likelihood that the correlations in the verb morphology are purely coincidental. Finally, in Section 30.7, I will conclude this chapter.

30.2  Overview of the shared verbal morphology of Transeurasian In Robbeets (2015), I undertook a detailed comparative study of elements of verb morphology shared between the Transeurasian languages. In Table 30.1, I present an overview of the different etymologies for auxiliaries and suffixes. The final column of this table provides numbered

Martine Robbeets, A comparative approach to verbal morphology in Transeurasian In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Martine Robbeets. DOI: 10.1093/oso/9780198804628.003.0031

Table 30.1  Overview of verb morphology shared among the Transeurasian languages PJ

PK

PTg

PMo

01

*ana-negation

*ana-negation

*an-negation

*ana-negation

[*an-] negation

02

*ə-negation

*e-negation

*e-se-negation

03

*-lA-manipulative *-ra-manipulative

*-lA:-manipulative

*-lA-manipulative *-lA-manipulative

31, 32/34

04

*-nA-processive

*-na-processive

*-nO-processive

*-nA-processive

*-nA-processive

*-(X)n-processive

28, 32

05

*(-)ki-‘do, make’ iconic

*-ka-iconic

*-ki-iconic

*-ki-iconic

*(-)ki-‘do, make’ iconic

*ki(-)l- /-kI-‘do make’ iconic

14, 40

06

*-mA-inclination

*-ma-inclination

*-mO-inclination

*-mA-inclination

*-mA-inclination

07

*-gA-inchoative

*-ka-inchoative

*-k(O)-inchoative

*-gA-inchoative

*-gA-inchoative

08

*-ti-causative

*-ta-causative passive

*-ti-causative passive *-ti-causative passive *-ti-causative

*-tI-causative passive 8, 40

09

*-pU-reflexive anticausative

*-pa-reflexive anticausative

*-pO-anticausative

10

*-dA-fientive

*-ya-fientive passive

11

*-rA-anticausative *-ra-anticausative

12

*-gi-creative causative

*-(k)i-creative *-ki-creative causative anticausative causative passive

-gi:-creative causative

13a

*-rA lexical nmlz

*-ra lexical nmlz

*-l lexical nmlz

*-rA lexical nmlz

*-r lexical nmlz

13b

*-wo-ra clausal nmlz

*-wo-l clausal nmlz

*-rA clausal nmlz

*-r clausal nmlz

13c

*-wo-ra relativizer

*-wo-l relativizer

*-rA relativizer

13e

*-wo-ra finite

*-wo-l finite

*-rA finite

*-r finite

*-m lexical nmlz

*-m lexical nmlz

*-mA lexical nmlz

*-mA ~ *-m lexical *-mA ~ *-m lexical

*-mA lexical nmlz

32, 28, 32 *e-negation

34

26, 32/34 *-(X)k-~-(X) g-inchoative

15 32/34

*-p-reflexive anticausative

*-βU-reflexive anticausative

*-U-reflexive anticausative

2, 38/39

*dA:-fientive

*-dA-fientive passive

*-(A)d-fientive anticausative

10, 32/34

*-rA-anticausative

*-rA-anticausative *-rA-anticausative

29, 32/34 15, 40

nmlz

*-rV lexical nmlz

29, 32/34 29, 32/34

*-rV relativizer

29, 32/34

*-rV finite

29, 32/34 26, 32/34

nmlz

14b

*-wo-m clausal nmlz

*-wo-m clausal nmlz

*-mA clausal nmlz

*-mA ~ *-m clausal nmlz

26, 32/34

14c

*-wo-m finite

*-wo-m finite

*-mA finite

*-mA ~ *-m finite

26, 32/34

*-n lexical nmlz

*-n lexical nmlz

*-nA ~ *-n lexical

*-n lexical nmlz

15a

*-n lexical nmlz

nmlz

*-n lexical nmlz

28

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

14a

*-(u)l-anticausative

PTk

Sound Corr. No

PTEA

PTEA 15b

PJ

PK

PTg

PMo

PTk

Sound Corr. No

*-wo-n clausal nmlz

*-wo-n clausal nmlz

*-nA ~ *-n clausal

*-n clausal nmlz

*-n clausal nmlz

28

*-nrelativizer

28

nmlz

*-wo-n relativizer

*-wo-n relativizer

15d

*-wo-n finite

*-wo-n finite

*-nA ~ *-n finite

*-n finite

*-n finite

28

*-ka resultative lexical

*-ka(-)i resultative lexical nmlz

*-xA: ~ *-kA: resultative lexical

*-xA ~ *-kA resultative lexical

*-xA ~ *-kA resultative lexical

22 32/34

nmlz

nmlz

nmlz

*-xA: ~ *-kA: clausal nmlz

*-xA ~ *-kA clausal nmlz

16a

*-xA ~ *-kA resultative lexical

nmlz

nmlz

16b

22 32/34

16c

*-ka relativizer

*-xA: ~ *-kA: relativizer

*-xA ~ *-kA relativizer

*-xA ~ *-kA pfv.fut relativizer

22 32/34

16d

*-ka finite

*-xA: ~ *-kA: past finite

*-xA ~ *-kA past finite

*-xA ~ *-kA future finite

22 32/34

*-sa resultative lexical

*-sA ~ *-si: < *sA-i: resultative lexical

*-sA ~ *-si: < *sA-i: *-sA resultative lexical

nmlz

nmlz

*-sA ~ *-si: clausal nmlz *-sA ~

*-si: clausal nmlz

17

*-sA resultative lexical nmlz

nmlz

*-sA perfective clausal nmlz

*-sA ~ *-si: relativizer *-sa finite

24 32/34

24 32/34 24 32/34

*-sA ~ *-si: finite

*-sA ~ *-si: finite

*-sA past finite

24 32/34

18

*-i ~ ø nominalizer

*-i ~ ø nominalizer infinitive converb

*-i ~ ø nominalizer converb adverb

*-i: ~ ø nominalizer

*-i ~ ø nominalizer converb adverb

*-I ~ ø nominalizer infinitive converb adverb

40

19

*-xU ~ *-kU nominalizer infinitive

*-ku nominalizer infinitive converb adverb

*-k(ʌ) / *-ku nominalizer infinitive converb adverb

*-xu: ~ -ku: nominalizer converb adverb

*-xU ~ -kU nominalizer infinitive converb

*-xU ~ -kU nominalizer infinitive

22 38/39

20

*ø imperative

*ø imperative

*ø imperative

*ø imperative

*ø imperative

*ø imperative

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

15c

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

Martine Robbeets sound correspondences in line with the consonant and vowel correspondences in Tables 36.2 and 36.3. I restrict myself to a small number of examples in this chapter, as space is limited, but I refer to Robbeets (2015) for a detailed description of the underlying forms. The reconstruction of the individual morphemes for ProtoTurkic, Proto-Mongolic, Proto-Tungusic, Proto-Koreanic, and Proto-Japonic relies on material from their earliest unambiguously written stages—Old Japanese, Late Middle Korean, Middle Mongolian, Written Mongolian, and Old Turkic—and is supplemented by the most relevant contemporary var­ ieties. A contemporary variety is considered relatively relevant for reconstruction purposes if it derives from a major node in the family tree for which historical records either are lacking or do not provide satisfactory morphological information. The reconstruction of Proto-Turkic morphemes is based on the main sources of evidence of the earliest split into the western Oghuric branch and the eastern Common Turkic branch, namely Chuvash and Old Turkic. The reconstruction of Proto-Mongolic morphemes is mainly based on Middle Mongolian and Written Mongolian. When supplementary information is needed, however, it is retrieved from contemporary varieties. Moreover, I added evidence of verbal morphology in Khitan texts by Kane (2009: 144–58), where available. For the historical study of Tungusic languages, records are unfortunately rather scarce for most of the languages. The oldest historical records are restricted to the Manchu branch of Tungusic, written in Jurchen, the official language of the Jin Dynasty (1115–1234), and in its immediate descendant, Manchu, the official language of the Qing Dynasty (1644–1911). Owing to the fragmentary attestation of Jurchen, I retrieve morphological information about Tungusic mainly from Manchu as representative of the Manchuric branch, from Even and Evenki as representatives of the eastern and western subbranches of Northern Tungusic, and from Udehe and Nanai as representatives of the eastern and western subbranches of Southern Tungusic, respectively. The reconstruction of Proto-Koreanic morphemes is based on Middle Korean and contemporary Korean. As happened with Mongolic under Chingis Khan, the Silla unification erased all coexisting varieties of Koreanic. In this study, the label “Middle Korean” mainly refers to Late Middle Korean, the language written down after the invention of the Korean script in 1446. The reconstruction of Proto-Japonic morphemes is based on representatives of its two branches, i.e. Mainland Japanese and Ryukyuan. The better-attested earliest source of Mainland Japanese is (Western) Old Japanese, representing the language spoken in Central Japan in the Nara period (710–94). As far as the Ryukyuan languages are concerned, evidence comes from Northern Ryukyuan languages such

514

as Shodon or Shuri or from Southern Ryukyuan languages such as Hirara or Hateruma.

30.3  Correlations in form 30.3.1  Regular sound correspondences The consonants and vowels involved in the comparisons correspond regularly according to the correspondences given in Tables 36.2 and 36.3, respectively. The vowel-harmony archiphonemes in Table 30.1 represent the following alternations: PK *O = *ɨ ~ ʌ ; PTg *A = *a ~ e [RTR *a ~ ə]; PMo *A = *a ~ e [RTR *a ~ ə]; PMo *U = *ü ~ u [RTR *u ~ ʊ]; PTk *A = *a ~ e, PTk *U = ü ~ u, PTk *I = ï ~ i, and PTk *X = ï ~ i ~ u ~ ü. The reinterpretation of the quality of the Koreanic, Tungusic, and Mongolic vowels in light of the Retracted Tongue Root (RTR) interpretation by Joseph et al. (this volume: Chapter 29) leads to the reconstruction of RTR harmony in the original Transeurasian suffixes with archiphonemes PTEA *A = *a ~ ə and PTEA *U = *u ~ ʊ. It is not unlikely that Turkic shifted to a front-back sound harmony system and that the RTR system became distorted in Japonic, due to areal influences at the periphery of the Transeurasian family. However, a number of Japonic verb suffixes in Table  30.1 have left traces of a vowel-harmonic-like alternation, notably PJ *-ra- ~ -rə- denominal verbalizer, PJ *-na- ~ *-nə- processive, PJ *-ma- ~ *-mə- inclination, PJ *-pa- ~ -pə- reflexiveanticausative, and PJ *-ra ~ *-rə nominalizer. This opposition may imply an original RTR based contrast.1 According to the vowel correspondences, the vowel harmony between PTEA *-a- and *-ə- regularly merges into PJ *-a-, while the harmony between PTEA *-u- and -ʊ- regularly merges into PJ *-u-. This merger and the widespread *A-vocalism in the Proto-Transeurasian suffixes in general may explain why the large majority of Japanese suffixes display a-vocalism. The irregular low central vowel in the Japonic iconic suffix *-ka- and causative-anticausative suffix *-ta- may be the result of analogy, seeking resonance with the other suffixes in the verbal paradigm. The consistently reduced vowel reflexes (*-ɨ- / -ʌ-) in the Korean suffixes are probably due to vowel reduction in unstressed position, reminiscent of the frequent weakening 1 Vowel alternations for the denominal verbalizer PJ *-ra- ~ -rə- are reflected in OJ wa-ra-p- ‘to laugh’ and OJ no2-ro2-p- ‘curse’, for the processive *-na- ~ *-nə- in OJ ata-na-p- ‘to harm, injure (tr)’ and OJ udu-no-p- ‘to prize, value (tr)’, for the inclinational *-ma- ~ *-mə- in OJ to2ga-me2- (< *tonka-maCi-) ‘to censure, find fault with (tr)’and OJ to2yo2-mo-s- ‘to make resound’, for the reflexive-anticausative *-pa- ~ -pə- in OJ kanape2- (< *kana-pa-Ci-) ‘to make suitable, possible, make fit, grant (a request)’ and OJ uru-po-p- ‘get damp, get moist, receive profits, get enriched’, and for the nominalizer *-ra ~ *-rə in OJ aka-ra ‘red’ and OJ woso2- ro2 ‘precocious, early ripening’.

OUP CORRECTED PROOF – FINAL, 26/05/20, SPi

compar ative tr anseur asian ver bal mor phology of the final vowel involved in lexical items. Morphological items, especially suffixes, tend to be unstressed and are therefore liable to undergo irregular reductive developments such as the loss of word-final vowels in some reflexes of the lexical nominalizers PTEA *-rA and *-mA. Vowel erosion in word-final suffixes is to be expected especially following sonorants such as /m/ and /r/, because their high sonority allows them to be articulated without a final vowel. This may explain why the vowels are retained in the reflexes of other word-final suffixes such as PTEA *-sA and *-xA ~ *-kA.

30.3.2  Shared allomorphy Two velar suffixes, PTEA*-xA ~ *-kA for the resultative nominalizer in Table  30.1 (16) and PTEA *-xU ~ *-kU for the nominalizer/infinitive in Table 30.1 (19) reveal the original conditioning factor behind the velar fricative cor­res­pond­ ence 22 in Table 36.2. They display a peculiar allomorphy in a phonological environment that often involves con­tinu­ ants. Since the velar involved becomes a voiceless obstruent in this environment, I suggested that it arose through de-fricativation (Robbeets 2015: 402, 406, 416, 483). The allomorphy is illustrated in example (1) for the resultative nominalizer, whereby Japonic and Koreanic merged the original allomorphy in a single voiceless velar suffix *-kA, while the Altaic languages preserved traces of the original allomorphy in continuant environment. (1)  PTEA *-xA resultative nominalizer → *-kA/ r_ or n_ or ß_ a. Proto-Japonic *-kA resultative nominalizer   PJ *ata- ~ atu- ‘to be warm’ in place name Atami ( do > io > obl > gen > ocmp), rela­tiv­iza­tion of possessors aligns with obliques (obl) or even object of com­ parison (ocmp) at the lower end of the hier­archy. In fact, however, in the Transeurasian languages, rela­tiv­iza­tion of possessors follows the strategy available for their heads; in particular, the subject relativization strategy can be extended to the possessor of the subject. This is well-known for Turkic as the Mother Node Principle (Hankamer and Knecht 1976; see also Kornfilt 1997: 58 ff. for a detailed dis­ cussion).5 As the examples below show, rela­tiv­iza­tion of the possessor of the subject follows the same strategy as rela­tiv­ iza­tion of the subject itself (note, in particular, the absence of possessive agreement in 31b), while rela­tiv­iza­tion of the possessor of the object follows the same pattern as relativ­ ization of the object. (31) Turkish a. [ — okul-a   gid-en]  adam school-dat  go-ptcp man ‘the man who goes to school’ b. [[ — kız-ı] okul-a gid-en] daughter-3sg school-dat go-ptcp adam man ‘the man whose daughter goes to school’ (Hankamer and Knecht 1976)

5   “If a subconstituent of the major constituent of the RC is relativized, the participle is chosen which would be appropriate for the relativization of the major constituent itself”’ (Hankamer and Knecht 1976).

635

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Andr ej Malchukov and Patryk Czerwinski (32) Turkish a. [ —  tanı-dığ-ım]  kadın    know-ptcp-1sg  woman ‘the woman whom I know’ b. [[ — koca-sın]-ı tanı-dığ-ım] kadın husband-3sg-acc know-ptcp-1sg woman ‘the woman whose husband I know’ (Malchukov 1998) In accordance with the Mother Node Principle, in Turkish and other Turkic languages, the same relativization strategy is used for the argument and its possessor, as illustrated in (31) and (32). As noted in Malchukov (1998), the same is true for Tungusic; the subject-relativization strategy extends to the possessor of the subject NP; in both cases the participle remains unmarked (33). (33) Even a. Etiken  atika-ŋa-n old man  wife-aln.poss-3sg hagdan-ni grow.old-nfut:3sg ‘The old man’s wife died (lit. grew old).’ b. [ —  atika-ŋa-n    hagdan-ča]    wife-aln.poss-3sg  grow.old-pfv.ptcp etiken old man ‘the old man whose wife has died’ Similarly, relativization of the possessor of the object pat­ terns with relativization of the object itself; the same holds for objects of postpositions (34). (34) Even a. Turaaki hiakita öjde-le-n doo-n. settle-nfut:3sg crow tree top-loc-3sg ‘The crow settled on (the top of) the tree.’ b. [ —  öjde-le-n   turaaki  doo-ča-n]    top-loc-3sg  crow    settle-pfv.ptcp-3sg hiakita tree ‘the tree on top of which the crow settled’ Also in Japanese and Korean relativization of possessors is pos­ sible, although, at least in the case of Japanese, it is restricted to the cases of inalienable possession (Ishizuka 2009). (35) Japanese Naomi-ga kawa-o mui-ta ringo Naomi-nom skin-acc peel-pst apple ‘the apple whose skin Naomi peeled’ (Ishizuka 2009)

636

Apart from the similarities, there are more subtle differences with respect to relativization of various nominal constituents further down Keenan and Comrie’s Accessibility Hierarchy. Thus, there are differences between Transeurasian languages as to whether the object relative clauses can be extended to obliques and adjuncts. In Even, relativization is restricted to  arguments and does not extend to adjuncts (for this ­reason it is used as a test for determining verbal valency in Malchukov  1995; Malchukov and Nedjalkov  2015). Thus, ­relativization of oblique objects effectively patterns in the same way as relativization of direct objects; cf. the rela­tiv­ iza­tion of a locative oblique in (36b). Importantly though, adjuncts cannot be relativized using this strategy. Thus, the following construction in (36c) with an attempted rela­tiv­ iza­tion of a locative adjunct is ungrammatical in Even. (36) Even a. (Bej) d´uu-la man house-loc aanŋat-ta-n/d´ep-te-n. stay.overnight-nfut-3sg/eat-nfut-3sg ‘The man stayed overnight in the house/ate in the house.’ b. (bej) aanŋat-ča-n d´uu. man stay.overnight-pfv.ptcp-3sg house ‘the house where (the man) stayed overnight’ c. *(bej) d´ep-če-n     d´uu man eat-pfv.ptcp-3sg house (intended meaning) ‘the house where (the man) ate’ On the other hand, in Japanese, both arguments and adjuncts can be relativized; rather, relativization is pragmatically constrained by the Relevance Condition (Kuroda  1976). Similarly, in Korean relativization is fairly non-restricted and applies to a very broad range of syntactic roles on Keenan and Comrie’s Hierarchy, including indirect objects, obliques, genitives, and objects of comparison (O’Grady et al. 2000). Relativization can also apply to an adjunct, as illustrated in (37). (37) Korean [nay-ka chayk-ul ilk-un] kyosil i-nom book-acc read-ptcp classroom ‘the classroom in which I read a book’ (O’Grady et al. 2000: 354) Other Transeurasian languages show variation in this regard. Thus, Turkic languages occupy an intermediate position, since relativization is less strictly constrained by valency, as the following Turkish example illustrates: taşın-dığ-ım ev [come-ptcp-1sg.poss house] ‘the house where I came’, or: ‘the

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

complex constructions in tr anseur asian l a nguages house I moved out of’, or: ‘the house I moved into’ (Johanson 1998b: 63). The ambiguities mentioned above may be related to still another type of nonfinite clause (exemplified for Japanese and Korean in 38 and 39), whose status as a relative clause is controversial. (38) Japanese akatyan-ga naku koe cry-npst voice baby-nom ‘the voice of a baby crying’ (Yoon 1993: 200) (39) Korean ai-ka wun-un soli baby-nom cry-ptcp sound ‘the sound of a baby crying’ (Yoon 1993: 200) In this construction the head is an abstract noun, which does not appear as an argument in the restrictive clause at all. For this reason, the participial clause can be seen as a complement clause with a nominal head, rather than a rela­ tive clause. In some recent typological studies (Comrie 1998; Matsumoto 1997; see also Matsumoto et al. 2017 for a stateof-the-art discussion of this phenomenon), the existence of such structures in Asian languages was taken as evidence for the existence of a superordinate pattern of “nounmodifying clause” that is ambiguous between a relative clause and a noun complement clause. It is plausible that languages that allow “noun-modifying constructions,” as illustrated above, impose fewer restrictions on relativization than those that do not. In any case, Northern Tungusic lan­ guages, which disallow noun-modifying constructions of the type illustrated here, are more constrained as to what constituents are eligible for relativization.

35.5  Internally headed relative clauses Internally headed relative clauses (IHRCs for short) are spe­ cial in that the notional (semantic) head (i.e. the relativized noun) does not constitute the syntactic head of the relative clause (taking a restrictive clause as a modifier). This is most obvious in the cases where the semantic head appears inside of the restrictive clause. Lehmann (1984, 1986) labels such a construction “circumnominal relative clause” (“Zirkumnominaler Relativsatz”), as opposed to the more common types of prenominal and postnominal relative clauses. However, as noted above, the internal position of the semantic head (center embedding) is not the defining property of an internally headed clause. Rather, it is the dependent status of the semantic head within the re­strict­ive clause. The internally headed clause is a relatively uncommon

type of relative clause; in most agglutinating languages it is either missing or marginal at best. It is more prominent in polysynthetic languages with elaborate argument indexing (cf. Dryer’s 2013 WALS map). Among the Transeurasian languages, internally headed relative clauses are well-known and have been extensively studied for Japanese (at least since Kuroda  1976, cf. also Shimoyama  1999). Within the internally headed relative clause the “semantic” head (relativized noun) retains its original syntactic position within the restrictive clause, while the clause-external nominalizer no takes the case marker to indicate the syntactic dependency of the whole construction. In example (40), the semantic head (under­ lined) is the object of the embedded clause, which stays in situ within the clause; the clause itself is in the direct-object position embedded under the matrix predicate tabe- ‘to eat’. (40) Japanese Yoko-wa [[Taroo-ga sara-no ue-ni Yoko-top Taroo-nom plate-gen top-loc keeki-o oi-ta]   no]-o    tabe-ta. cake-acc put-pst  nmlz-acc   eat-pst ‘Yoko ate a piece of cake which Taro put on a plate.’ (Shimoyama 1999) It has been pointed out that the internally headed clause in Japanese is structurally similar to a complement clause with the nominalizer no. The interpretation depends on the selectional restriction of the matrix verb and is not signaled morphosyntactically (Kikuta 2001); note the identical struc­ tures of the complement clause in (41a) and the internally headed relative clause in (41b). (41) Japanese a. [[Kinoo ringo-o mora-tta] yesterday apple-acc receive-pst no]-o   oboe-te     i-ta. nmlz-acc  remember-cvb  aux-pst ‘I remembered that I had received some apples yesterday.’ ringo-o morat-ta] b. [[Kinoo Yesterday apple-acc receive-pst no]-o   tonari-ni    osusowakesi-ta. nmlz-acc  neighbour-dat  share-pst ‘I shared with the neighbor the apples I received yesterday.’ (Kikuta 2011) Internally headed clauses in Japanese are clearly marginal compared to the basic prenominal strategy; they are used less frequently, and are subject to additional restrictions in that typically only subjects and objects can be relativized.

637

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Andr ej Malchukov and Patryk Czerwinski Among other Transeurasian languages, internally headed clauses have been, until recently, documented only for Korean, where they make use of the nominalizer kes, which closely mirrors the use of no in Japanese (although se­man­tic­al­ly kes  corresponds rather to Japanese koto6). The following ex­ample shows an internally headed clause with the subject ­relativized, in the direct-object position within the matrix clause (42a). In this case, its status as an internally headed clause is not immediately obvious, as the semantic head (relativized subject) occurs in the initial position within the embedded clause (so that the construction in (42a) can be mistaken for a postnominal relative clause with an external head). This becomes obvious, however, through comparison to the (object) complement clause with kes (42b). (42) Korean a. John-un [totwuk-i kakey-eyse John-top thief-nom shop-abl nao-nun]   kes-ul  cap-ass-ta. exit-prs.ptcp  nmlz-acc  catch-pst-decl ‘John caught the thief who happened to come out of the shop.’ b. John-un [totwuk-i kakey-eyse John-top thief-nom shop-abl nao-nun]   kes-ul  po-ass-ta. exit-prs.ptcp  nmlz-acc  see-pst-decl ‘John saw the thief coming out of the shop.’ (Kim 2009) As in Japanese, interpretation of this structure ultimately depends on the selectional restriction of the matrix verb (Chung and Kim 2003). Given the similarities between the Japanese and Korean structures, the most plausible explanation is that Korean structures calque the no-patterns in Japanese. This scenario is compatible with the observation that internally headed clauses are more restricted and less accepted in Korean (Horie  2011). It should be further noted that in both lan­ guages internally headed clauses are subject to similar ­syntactic restrictions; typically subjects or objects can be relativized using the internal strategy. On the other hand, Korean is more liberal, in allowing relativization of certain oblique positions (e.g. instrument; M. Kim 2007). In (core) Altaic studies, the existence of IHRCs in Japanese and Korean has not attracted much attention, probably because it was seen as a local feature indicative of Japanese– Korean convergence. More recently, IHRCs have been ­discovered for Tungusic (Malchukov  1996). The Evenki 6   In fact, material similarity between kes and koto has been attributed in the specialist literature either to cognacy (Martin 1991; Whitman 1985), or to an ancient borrowing from Korean into Japanese (Vovin 2010: 150).

638

example in (43) exemplifies the most common type of IHRC in the object position with the subject relativized; the next ex­ample, (44), from Udihe shows an IHRC in the subject position with the object relativized. (43) Evenki (Malchukov 1996 citing Brodskaya 1988: 55) (. . .)  baka-ra-n    [eseŋer-du  Muriwul nfut -3 sg   tree-dat   Muriwul     findal-al-di-wi d´awaad´a-ri-wa-n] hand-pl-ins-ref.poss hold-nfut.ptcp-acc-3sg ‘(. . .) he found Muriwul who was holding onto the tree with his hands.’ (44) Udihe (Nikolaeva and Tolskaya 2001: 681) zugdi-ziga-wa wo:-ni] [Bi od´o-i I grandfather-1sg house-pl-acc make.pst-3sg uligdig´a beautiful ‘The houses that my grandfather built are beautiful.’ The first type is particularly common in Tungusic languages; the second type with IHRC in the subject position is used restrictedly (only an object, not a subject, can be relativized; “subject-in-subject constraint” in Malchukov 1996). Certain languages, such as Udihe, show fewer restrictions in terms of positions relativized; it also allows for IHRCs to appear in oblique positions (Nikolaeva and Tolskaya 2001: 684) Note that the most common type of IHRC in the directobject position is again structurally identical to comple­ ment clauses, but also to residual types of adverbial clauses taking the accusative case. Malchukov (1996) suggested that IHRCs actually developed through reanalysis of adverbial temporal clauses headed by the accusative-marked par­ti­ ciple. This would also account for their specific discoursepragmatic function: in contrast to prenominal clauses, IHRCs are often appositive rather than restrictive. When writing his study of IHRCs in Tungusic, Malchukov was unaware of an earlier study by Kubo (1985), who identi­ fied IHRCs in Manchu. In fact, these constructions come in two types; relative clauses in the subject position feature the nominalizer ngge, which can also form complement clauses. This structure has close parallels in Korean and Japanese. On the other hand, relative clauses in the object position do not feature this nominalizer and follow the same pattern as IHRCs in other Tungusic languages. Both types are infrequent and subject to similar restrictions, which have been observed for other Tungusic languages: only sub­ jects and objects can be relativized, and the clause itself is appositive rather than restrictive. The second surprise came when Malchukov (1996, 2006) while studying a peculiar internally headed strategy in Tungusic, discovered a similar construction in Sakha. For

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

complex constructions in tr anseur asian l a nguages Sakha this pattern has been documented in the literature, but misinterpreted. Thus, in the comprehensive treatise on the Sakha syntax by Ubrjatova (1976), this construction appeared under a peculiar heading of “complement clauses related to verbs with concrete semantics” (Ubrjatova 1976: 249–50). Here is one of the few examples cited. (45) Sakha [Taba batta-ga erge bergehe-te, ilberkej nek torn shabby reindeer fell-3sg old hat-3sg son-o   sir-ge   sıt-al.lar-ı-n] coat-3sg  floor-dat  lie-prs.ptcp.pl-3sg-acc xarbaa-ta . . . seize-pst:3sg ‘He seized the old hat of reindeer fell and the torn shabby coat lying on the floor. . .’ (Ubrjatova 1976: 249) While the terminology used by Ubrjatova (“complement clauses related to verbs with concrete semantics”) is selfcontradictory, it reflects the fact that such structures are semantically equivalent to relative clauses, but structurally are identical to complement clauses. In other words, they can, without any structural adjustment, appear as direct objects of verbs taking sentential complements such as kör- ‘to see’. Normally one would explain the convergence between Sakha and Even through structural influence from Sakha on Even, which is pervasive in Western Even dialects (Malchukov 2006). However, this explanation is implausible given that IHRCs are documented for other Tungusic languages, but are not attested for Turkic other than Sakha. For this reason, Malchukov (1996, 2006) attributes this convergence to ­substrate influence from Tungusic on Sakha. What can account for this convergence across several families within of Transeurasian? It should be pointed out that independent development as the only explanation is unlikely. As noted above, IHRCs are more typical of polysyn­ thetic ­languages with extensive head-marking (for subjects and objects), rather than languages of the “Uralo-Altaic” type, where they remain marginal, if found at all. Also, the connection between Korean and Japanese is hardly accidental; it points to calquing of Japanese structures by Korean (the use of kes as a complementizer is a relatively recent in­nov­ ation, also in complement clauses; Rhee 2008). Furthermore, the connection between Sakha and Tungusic is likely to be areal (if the absence of internally headed structures in other Turkic languages is confirmed), and should be due to a Tungusic substrate in Sakha. This leaves parallels between Japanese and Tungusic, which cannot be explained through areal influence. In the latter case, it seems more likely that we are deal­ ing with independent developments. For Japanese, it has

been suggested by Kikuta (2001), that IHRCs most probably derive from complement clauses; in fact this development was repeated cyclically (first in Old Japanese when ­participles doubled in function as direct nominalizations, then in the second cycle after they lost that function and the nominalizer no was introduced). For Tungusic, on the other hand, it has been suggested by Malchukov that IHRCs derive from reanalysis of temporal adverbial clauses (Malchukov  1996). In fact, if the accusative-marked ­temporal clauses were still productive in Tungusic, one could confuse the IHRC with a temporal adverbial clause with a participle in the accusative case. Currently however, accusative-marked temporal clauses are almost extinct in Tungusic, but should have been available at earlier stages (cf. the converbialized participial forms in -d´an-ma- and -ktaa-wa- in Evenki, which contain a petrified accusative marker; cf. I. Nedjalkov 1995: 451). As for the two types of IHRCs in Manchu, one type, which lacks the nominalizer, is also found in other Tungusic lan­ guages, so it does not require a special explanation. As for the type of IHRC featuring nominalizer ngge in the subject position, it has close parallels to Korean and Japanese; here an areal influence cannot be excluded. Which languages may play a role in these convergent tendencies remains to be investigated; neither Mongolian nor Chinese come into question as structural models. Thus, although the Manchu ngge (as well as Japanese no) has a close parallel in the Mongolian nominalizer -Xi ~ -Ki (cf. Manchu: mini-ngge ‘mine’; Mongolian xaan-ii-x ‘the king’s one’), the Mongolian nominalizer does not form IHRCs. The same is true for the nominalizer and relative clause marker de in Chinese. If the scenarios postulated for IHRCs in Tungusic (Malchukov 1996) and Japanese (Kikuta 2001) are confirmed, these scen­ arios are compatible with the view of IHRCs as intermediate structures halfway on the developmental path from adver­ bial into complement clauses or vice versa. This is also com­ patible with typological findings, which explain the origin of IHRCs in other languages either through re­analy­sis of a  complement clause (Lehmann  1984: 383) or through re­analy­sis of adverbial clauses with temporal, contingency, or topic functions (Dik 1997: 85–90). It is also compatible with more theoretical accounts of IHRCs in their relation to other types of subordinate clauses as proposed by Croft (2001), and adapted for Japanese and Korean by Horie (2011). Following Croft (2001), Horie (2011) regards IHRCs as an intermediate structure between complement and relative clauses. As shown in Figure 35.3 (adopted from Horie  2011), on this approach, IHRCs constitute a link between the other two types in the semantic space for complex constructions designed by Croft (2001). The dashed triangle shows an extension to certain types of (concessive) adverbial clauses for Japanese, as shown in Figure 35.3.

639

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Andr ej Malchukov and Patryk Czerwinski cosubordination

coordination

serial verbs paratactic clauses speech complements

adverbial clauses

purpose clauses adjoined relative clause correlative clauses

complements

relative clauses internally headed relative clauses

Figure 35.3  Japanese internally headed relative clauses on the semantic map of complex constructions Adopted from Horie 2011

We find this analysis insightful, but in need of a correction; to the extent that the Transeurasian data is representative, IHRCs should have a direct link on the map with adverbial clauses as well, a connection not mediated through (pre­nom­ inal) relative clauses, as the map in Figure 35.3 would suggest.

35.6  Paratactic constructions and adverbial conjunctions In all Transeurasian languages, in addition to the dominant strategies based on nonfinite verb forms discussed in Sections  35.2–35.5, clause linking can be signaled through conjunction-like adverbial relators. These relators are often referred to as conjunctions, but as Johanson (1998b: 49) points out for Turkic, many forms referred to as conjunctions (such as anın ‘therefore’) are in reality conjunctive adverbs. Indeed, most of them do not qualify as subordinators. An exception to this may be clauses with com­ple­ment­izers introduced by con­ junctions derived from converbial forms of the verb ‘to say’, attested in many Transeurasian languages (cf. Turkish diye, Mongolian ge-j, Northern Tungusic gö-mi, Japanese to it-te, etc.), but also in this case the subordinating function is ques­ tionable. In fact, these forms often grammaticalize rather into modal hearsay particles, as illustrated in the following example from Even, featuring the quotative particle göön (46). (46) Even Gaaran´d´a göönni: “Oŋalgan, göön, hii Gaaran´d´a say-nfut-3sg Oŋalgan quot you Čaamaj  nuŋu  bi-se-nri.   Oon,  göön, most   fool  be-nfut-2sg  how  quot d´eb-d´i-n,   göön,  ia-č   čiki-d´i-n,   göön . . .” eat-fut-3sg  quot  what-ins  cut-fut-3sg  quot ‘Gaaran´d´a (the bird) said: “Oŋalgan, what a fool you are! How will he (the fox) eat you, how will he cut the tree . . .”’ (Malchukov 2000, citing Novikova 1980: 136)

640

Japanese and Korean stand out among the Transeurasian languages in that they feature bona fide complementizers, as  described in Section  35.2. As noted in Section  35.2, in Japanese the use of the analytic strategies is related to the need to restore the distinction between finite and ­nonfinite forms through grammaticalization of a new series of nom­in­alizers. Within relative clauses, analytic strategies are marginal if found at all, and are due to recent contact with other lan­ guages. In Turkish, there is a pattern borrowed from Persian where the relative clause is introduced by a (borrowed) complementizer ki (47). (47) Turkish bir adam [ki çocuk-lar-ın-ı sev-me-z] a man that child -pl-3sg-acc love -neg-aor yalnız  yaşa-malı-dır alone  live -nec-epist.cop ‘A man who does not love his children must live alone.’ (Kornfilt 1997: 60) Furthermore, certain Turkic languages may calque Russian patterns with relative pronouns (cf. Anderson, this volume: Chapter 40). This is also occasionally observed in Tungusic, as the following Evenki example (48) illustrates (Malchukov 2006 citing Brodskaya 1988). (48) Evenki [I-le hurkeken suru-re-n] gorot-tu] go-nfut-3sg town-loc where-all boy umekemet  tungus   achin. single    Tungus  be.absent ‘In the town where the boy is going, there is not a single Tungus.’ (Malchukov 2006 citing Brodskaya 1988: 53) In (48) the interrogative pronoun in the locative case ­obviously mimics the use of relative pronouns in Russian. A more interesting case of interference is the “mixed” ­pattern, which blends the native prenominal strategy with the use of a relative pronoun modeled on Russian (Malchukov 2003). Note that this construction, illustrated in (49), lacks the ­relative pronoun, but the case expected on the relative ­pronoun now appears on the participle (cf. the locative case on the relative pronoun in 48 and on the ­par­ti­ciple in 49). (49) Evenki [[Hurkeken suru-mechin-dule-n] gorot-tu] town-loc boy go-fut.ptcp-all-3sg umekemet tungus   achin. single Tungus  be.absent ‘In the town where the boy is going, there is not a single Tungus.’ (Malchukov 2006 citing Brodskaya 1988: 53)

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

complex constructions in tr anseur asian l a nguages In the domain of adverbial clauses, some Transeurasian ­languages developed adverbial subjunctors that serve the functions comparable to adverbial subordination. The num­ ber of these subjunctors differs across languages. In Northern Tungusic they are less grammaticalized, while in Manchu they are very common. Thus, Gorelova (2002: 356) lists the following items used in the connective (co­ord­in­at­ ing) function: bime (originally the imperfect converb derived from the verb bi-‘to be, to exist’); geli ‘also, again, still’; jai ‘again, still, more, later’; kemuni ‘often, still, yet’. The linkers akūci ‘otherwise’ (< akū ‘there is/are not’ + o-ci conditional converb derived from the verb o-‘to be, to become’), bimbime (originally the imperfect converb derived from the word form bimbi) and bai ‘only’ are used as the adversative con­ junction ‘but’. As is clear from this list, most conjunctional adverbials represent either lexicalized forms of converbial forms based on auxiliaries, lexicalized local cases of demon­ stratives, or adverbs and particles in conjunctional use. Similar patterns recur across the Transeurasian languages. Thus, across different families, we find conjunctions and conjunctional adverbs based on nominalizations/par­ti­ciples (cf. Japanese no-de (nmlz-ins) ‘because’ vs. Even -ri-d’i (ptcp. ipfv-ins.refl.sg)); auxiliaries, including copulas (cf. Japanese de-mo (cop-foc) ‘even though’ vs. Even bi-mi-de (cop-cond.cvbfoc)), and verbs of speech (cf. Japanese to it-te-mo (cmp saycvb-foc) ‘even though’, Manchu se-ci-be (say-conc.cvb from cond.cvb-acc) ‘although’), as well as demonstratives (cf. Japanese sore-kara (that-abl) ‘after that’ vs. Even, Evenki ta-duk (that-abl) ‘then’), and relative clauses based on ‘light nouns’ (cf. Korean -ul ttay-(ey) (fut.ptcp time-loc) ‘when’ vs. Turkish zaman ‘time’ as in çın-tığ-ı zaman (go-ptcp-3sg time) ‘when he goes’; Kornfilt 1997: 69). Also in the domain of adverbial subordination, Japanese shows stronger analytic tendencies, featuring numerous forms qualified as conjunctions (such as node, illustrated in 50), which represent combinations of the nominalizer no and case/postpositional forms. This is illustrated in (50) for node ‘because’ (nmlz-ins); cf. also: noni ‘even though’ (nmlzloc); tokoroga ‘but’ (nmlz-nom); tokorode ‘in these circum­ stances’ (nmlz-ins), etc. (50) Japanese Ame-ga hut-ta-node mizutamari-ga rain-nom fall-pst-because puddle-nom deki-ta. emerge-pst ‘Because it rained, puddles appeared.’ (Kaneko and Bekki 2014) As we have pointed out, such conjunctions, representing combinations of a nominalizer with an oblique case/ postposition, actually constitute a structural parallel to

“quasi-converbs” in other languages. Synchronically, these structures can be seen as “stranded” participial markers which in combination with case markers grammaticalized into bona fide conjunctions. Diachronically, these construc­ tions can be related to “noun-modifying” constructions reminiscent of relative clauses headed by a “light noun,” as exemplified in (38) and (39).

35.7  Concluding remarks As is clear from the presentation given here, Transeurasian languages show numerous similarities in the expression of complex constructions. Undeniably, many similarities are due to independent developments against the background of shared basic structural properties. First and foremost, this is true of the use of nonfinite (deranking) strategies instead of finite clauses in complex constructions, a feature characteristic of many head-final languages of the ag­glu­tin­ at­ing type. Many structural similarities are also likely to be  independent developments, such as the use of similar quasi-converbial strategies across Transeurasian (e.g. with par­ti­ciples in the instrumental or locative case for anterior­ ity), or the use of similar strategies for object vs. subject rela­tiv­iza­tion. Some variation among languages can be straightforwardly attributed to the variation in structural properties (such as  the presence or absence of possessive agreement). For example, as noted a number of times in this chapter, most Tungusic languages (except for Manchu) use possessive agreement on nonfinite forms (participles and converbs) to index the subordinate subject, if different from the matrix subject. Those languages which do not have gram­mat­ical­ized (possessive) agreement (Japanese, Korean, most Mongolic), have to rely on other strategies to avoid ambiguity, for example through the use of specialized nonfinite forms in same-subject and different-subject contexts. Similarly, the more “analytic” profile of Japanese as compared to other languages manifested in several domains (cf. the use of complementizers in complement clauses and conjunctions in adverbial subordination) is a consequence of recent cycles of grammaticalization of nominalizers. On the other hand, some of these convergences are clearly due to language contact. This is most clear when certain markers are borrowed; for example, some Turkic languages borrowed conjunctions from Persian (cf. ki ‘that’), or Arabic (cf. ve ‘and’), and many Turkic, Mongolic, and Tungusic lan­ guages copied Russian conjunctions (Anderson, this volume: Chapter 40). In many cases, however, the influence is more subtle, as it involves structural rather than material influ­ ence (pattern-borrowing rather than matter-borrowing in

641

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Andr ej Malchukov and Patryk Czerwinski terms of Matras and Sakel 2007). Such cases are more diffi­ cult to distinguish from independent developments, as our discussion of internally headed relative clauses above dem­ onstrates. For internally headed relative clauses, we tenta­ tively conclude that both language contact (most obviously between Northern Tungusic and Sakha) and internal devel­ opments were involved (cf. the parallel developments of internally headed structures involving nominalizers in Manchu and Japanese). In other cases, involving more local patterns, we can be relatively certain that language contact was a factor. In addition to the cases discussed above, we can mention that in both Mongolic and Northern Tungusic we find the pattern of a converb combined with a diminu­ tive marker (cf. Mongolian modal converb -ng-xAAN, Even simultaneous converb in -ni-kAn). This convergence is hardly coincidental, as converbs normally do not feature nominal morphology. Similarly, the formal similarity of the Tungusic limitative converb in -dAlA (Manchu -tAlA) with the Mongolian terminative converb -tala is not coincidental and should be interpreted as a borrowing from Mongolian into Tungusic (Benzing  1955a: 1092); the connection with the Korean ­terminative converb in -tolok is less clear. Finally, there are cases involving both structural and material similarity between individual languages, which are

642

indicative of either cognacy or early borrowing between the families of Transeurasian. These cases have been reviewed recently by Robbeets (2015), who argues for a common ori­ gin of some nonfinite forms in Transeurasian, including the participle in -RA, which has developed into the finite “aorist” in different families of (core) Altaic (see Malchukov and Czerwinski, this volume: Chapter 34, for further discussion). These cases are the most difficult to interpret, as distin­ guishing cognates from early borrowings and historical accidents remains a considerable challenge (see Robbeets 2015, this volume: Chapter 30, for further discussion).

Acknowledgments We are grateful to Jaklin Kornfilt and Martine Robbeets for the useful comments on the first draft of this chapter. The usual disclaimers apply. This work was partially supported by the Laboratory Program for Korean Studies through the Ministry of Education of the Republic of Korea and the Korean Studies Promotion Service of the Academy of Korean Studies (AKS-2016-LAB-2250004).

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

D.  Lexicon and Semantics

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Ch a pter 36

Basic vocabulary in the Transeurasian languages M a rt i n e R obbe e ts

36.1 Introduction Here I will address one of the major objections raised against the genealogical relationship of the Transeurasian languages, notably the paucity of basic vocabulary in both quantity and quality. Indeed, critics of Transeurasian affiliation frequently refer to the small number of correlating basic vocabulary items as a whole as well as to the weak regularity of their sound correspondences. About a century after Nicolaas Witsen (1692) and Phillip von Strahlenberg (1730) mooted the contours of the Transeurasian language family, the idea was criticized by the French sinologist and medical doctor Abel-Rémusat (1820), who was the first to explicitly use the lack of basic vocabulary as a counter-argument. His main criticism was that the correlating words between the Turkic, Mongolic, and Tungusic languages seemed to be the result of a long-standing political dominance of Turkic speakers, while the basic vocabulary shared between these languages—in AbelRémusat’s wording “les mots qui désignent des objets de première nécessité”—was radically different.1 Following in Abel-Rémusat’s footsteps but seemingly unaware of the basic vocabulary list published by Swadesh (1955) in the preceding year, Clauson (1956: 182) made his case against the so-called “Altaic theory” arguing that “[t]he basic words, that is the numerals, the basic verbs like ‘to say, to give, to take, to go’ and so on, the basic nouns like ‘food, horse’, and the basic adjectives like ‘good, bad’ are all entirely ­different.” Doerfer (1988: 155–6) rejects Clauson’s concept of basic vocabulary, stating that “Mit vollem Recht haben sich . . . die Vertreter der altaistischen Verwandtschaftsthese durch Sir Gerards Liste nicht erschüttern lassen . . .”, and moves the basic vocabulary argument to a new plane. Focusing on terms for body parts, he argues that the cor­re­ la­tions shared between the Turkic, Mongolic, and Tungusic   Abel-Rémusat (1820: 394): “Les mots de ces . . . langages, particulièrement ceux qui désignent des objets de première nécessité, . . . sont radicalement différents . . .” 1

languages can be attributed to borrowing because these languages share more core than non-core body-part terms. Ramer et al. (1997) reject Doerfer’s conclusion by his own criteria, showing that there are more three-way cognates among the Altaic core than among the non-core body-part terms. The argument that the Altaic languages did not sufficiently share basic vocabulary was further tackled by Sergej Starostin and his colleagues, who added evidence from Japonic and Koreanic languages and created a massive etymo­logic­al corpus, including 144 different etymologies for 100 basic vocabulary items (Starostin et al. 2003: 230–4). However, these new matches were, in their turn, criticized for reason of phonological, morphological, or semantic overpermissiveness, among others by Robbeets (2005), leaving room for a reduced core of reliable etymologies, and by Vovin (2005c), completely rejecting all evidence advanced so far. A basic vocabulary list is a compilation of concepts that are relatively independent of cultural context and available across the languages of the world and, therefore, can be used for testing the stability of historical-comparative linguistic evidence. The strength of the argument lies mainly in the fact that words with basic meanings tend to resist borrowing more successfully than random lexical items. The basic vocabulary list most commonly used in historical linguistics is the Swadesh 100 list (Swadesh 1955). Among the shortcomings of this list, we find the fact that it was put together on the basis of linguistic intuition and that it contains mostly nouns and too few verbs. For languages where verbs are basic to word formation and many nouns are derived from verbs, such as the Transeurasian languages, a standard Swadesh list produces too few useful comparanda.2 Therefore, the Swadesh list has been recently updated by the Leipzig-Jakarta list (Tadmor et al. 2010), which takes a more systematic and empirical approach and partly remedies the imbalance in the vocabulary. Although 62 items on the 2  Compare, for instance, (12) ‘breast’ and (69) ‘suck’, in which the Tungusic and Mongolic basic items for ‘breast’ are derived from the verbs ‘to suck’ by a deverbal noun suffix in *-n: PTg *xökö- ‘to suck’ → *xökö-n ‘breast’ and PMo *kökö- ‘to suck’ → *kökö-n ‘breast’.

Martine Robbeets, Basic vocabulary in the Transeurasian languages In: The Oxford Guide to the Transeurasian Languages. First edition. Edited by: Martine Robbeets and Alexander Savelyev, Oxford University Press (2020). © Martine Robbeets. DOI: 10.1093/oso/9780198804628.003.0037

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Martine Robbeets lists overlap, the difference between the two lists consists mainly in the replacement of a good number of nominal concepts by adjectival and verbal concepts in the LeipzigJakarta list. Moreover, on the basis of a solid cross-linguistic investigation, the Leipzig-Jakarta list takes into account factors other than low borrowability, such as the degree to which the meanings are universal, the degree to which the words are simplex, and the probability of attrition. In this chapter, I will advance Transeurasian etymologies for basic vocabulary items. In Section 36.2, I will present an overview of the evidence, paying attention to the regularity of sound correspondences. In Section 36.3, I will argue why the compared words are unlikely to be the result of borrowing. In Section 36.4, I will exclude chance similarity as a possible explanation for the correlations. Finally, I will conclude this chapter in Section 36.5. In the supplementary information to this volume, available at http://www.oup. co.uk/companion/Transeurasian. I have provided the full data sets underlying my reconstructions.

36.2  Overview of the basic vocabulary of Transeurasian The Transeurasian basic vocabulary, summarized in Table 36.1, consists of lexical etymologies supporting the reconstruction of a Proto-Transeurasian form with a meaning that belongs to the Leipzig-Jakarta 100 basic vocabulary list. There are 93 etymologies spread over 64 distinct concepts of the basic vocabulary list. As there are no reliable cognates in Transeurasian for the following basic items, they yield empty rows and are therefore left out from Table 36.1: 2 nose, 6 tongue, 10 root, 13 rain, 15 name, 17 wing, 18 flesh/ meat, 20 fly, 21 night, 23 neck, 28 bitter, 29 to say, 34 who?, 35 3sg pronoun, 38 horn, 41 yesterday, 42 to drink, 44 navel, 47 back, 48 wind, 49 smoke, 52 egg, 57 good, 59 knee, 60 sand, 61 to laugh, 64 leaf, 66 liver, 71 ant, 72 heavy, 75 to eat, 76 thigh, 83 eye, 85 tail, 89 to see, and 93 bird. The reconstructions in Table 36.1 are based on lexical data from 54 contemporary and 6 historical varieties of the Transeurasian languages, including 23 Turkic languages in addition to Old Turkic, 10 Mongolic languages in addition to Written Mongolian and Middle Mongolian, 10 Tungusic languages in addition to Manchu and Jurchen, Korean in addit­ ion to Middle Korean and Japanese, and 10 Ryukyuan languages in addition to Old Japanese. In the online supplementary information (SI 2) http://www.oup.co.uk/companion/Transeurasian, I have added a detailed documentation of all basic vocabulary comparative sets in support of Transeurasian affinity. The reconstruction of the individual proto-languages is based on the scientific literature in the field of Turkic,

646

Mongolic, Tungusic, Koreanic, and Japonic reconstruction, combined with personal insights. In the online supplementary information (SI 1), I have provided correspondence sets comparing contemporary and historical varieties of these languages, in order to infer the reconstruction of the basic phoneme inventory of the uncontroversial families. In case my personal reconstruction of an individual proto-form deviates from the standard reconstruction proposed in previous literature, I explicitly motivate my choice in the explanatory text added to the relevant etymology in SI 2. The reconstruction of the basic phoneme inventory for Proto-Japonic is based on Martin (1987) with integration of the revision of the vowel system proposed by Frellesvig and Whitman (2008). The Ryukyuan correspondences involved in the reconstruction of Proto-Japonic are based on research by Thorpe (1983) and Pellard (2013, 2021/in press). The phono­logic­al correspondences of Proto-Koreanic to Late Middle Korean and Contemporary Korean are based on Martin (1996) and Lee and Ramsey (2011). For the reconstruction of the Tungusic basic phoneme inventory, I refer to Cincius (1949b), Benzing (1955a) and Starostin et al. (2003). The Mongolic sound correspondences and reconstructions are based on Poppe (1955), Starostin et al. (2003), and Nugteren (2011). For the phonological reconstruction of Turkic, I refer to Starostin et al. (2003). The reconstruction of proto-forms in this chapter does not necessarily confirm all reconstructions proposed elsewhere by other authors in this reference guide. I collected the data set by consulting dictionaries in combination with written sources against the background of etymologies proposed in the past by various linguists and evaluated in Robbeets (2005). In this monograph, I constructed a database of previous etymological proposals, which I submitted to an item-by-item evaluation in order to reach a restricted core of reliable evidence in support of the genealogical relationship between Japanese and the other Transeur­ asian languages. The data set presented here is updated in the sense that I expanded my previous data set with ety­molo­ gies for basic items that do not have a Japanese cognate, consistently added cognates from the Ryukyuan languages, expanded the Tungusic cognates, carried out a detailed morphological analysis to delimit the roots more precisely and to identify petrified suffixes more accurately, and answered to some of the criticism in reviews of earlier etymologies. The evidence in Table 36.1 corresponds regularly in form and function to such an extent that it yields ProtoTranseurasian reconstructions for items that occur on the Leipzig-Jakarta 100 basic vocabulary list. The penultimate column attributes a number to each regular sound cor­re­ spond­ence, which refers to the inventories of consonant cor­res­pond­ences in Table 36.2 and vowel correspondences in Table 36.3. Gray shading marks the cells in which the

Table 36.1  Summary of the basic vocabulary comparative sets in support of Transeurasian affinity LJ item

Proto-Japonic

Proto-Koreanic

1 fire

*pɨ(r)i ‘fire’

*pɨl ‘fire’

3 to go

*na-‘go away’

*na-‘go out’

*-na:-‘go out to’

*mɨl ‘water’

*mö: ~ mu: ‘water’

4 water 5 mouth

*kutɨ-i ‘mouth, hole, opening’

Proto-Tungusic

8 bone

*pəni(C)a ‘bone’

9 2sg pronoun *na ‘2 sg pronoun’ *kə- ‘come’

12 breast

*kɨkɨ-rə ‘heart’

14 1sg pronoun

*wa-n- ‘1sg/pl pronoun’

*peCi ‘bone’

27, 32

56/4

25, 37, 29

65/9

13, 38, 8

41/8 84/5

41, 26, 41

41/8

7, 40

228/4 Poly

1, 34, 28, 40

157/6 NO

27, 33

250/2

*si ‘2 sg pronoun’

23, 40

250/2

*kel- ‘come’

13, 34

161/3

*kökü-r2 ‘breast’

21, 37, 14

34/6 Poly

3, 32

87/4 87/4

*tï:n ‘spirit, breath’

*peni- ‘knee’

*ne ‘2 sg pronoun’

*xökö-n ‘breast’

*kökö-n ‘breast’ *ba-n-‘1pl excl. pronoun’

16 louse

*bi ‘1 sg pronoun’

*bi ‘1 sg pronoun’

3, 40

*sir-ke ‘louse’

*sir-ke ‘nit’

23, 40, 18, 33 205/3

*ta(r)i ‘upper limb, arm, hand’

*tali ‘lower limb, leg’

19, 32, 31, 40 38/8 NO

*sune ‘lower limb, leg’

*son ‘upper limb, arm, hand’

23, 36, 28

38/8 NO

22 ear 24 far

68/7

*ama-n ‘mouth, opening’ *či < *ti ‘blood’

*bi ‘1 sg pronoun’ 19 arm/hand

1, 37, 31 *mören ‘river’

*si ‘2 sg pronoun’ 11 to come

Sem. Corr. no

*mara ‘rare, from afar’

*melu-‘be far’

*gar ‘hand, arm’

*karï ‘arm’

15,32,29

38/8

*kul ‘ear’

*kul-kak ‘ear’

13, 39, 31

186/4

25, 33, 31

17/7 (continued )

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

*ti ‘blood, spirit’

Sound Corr. no

Proto-Turkic

*kut ‘cavity, hole’ *amga < *?ama-g ‘mouth’

7 blood

Proto-Mongolic

Table 36.1  Continued Proto-Japonic

Proto-Koreanic

Proto-Tungusic

Proto-Mongolic

Proto-Turkic

Sound Corr. no

Sem. Corr. no

25 to do/ make

*-ka-iconic

*-ki-iconic

*-ki-iconic

*ki-‘do, make’

*kïl-‘do, make’

13, 40

33/6 Gram

26 house

*(y)ipi ‘house, hut’

*cip(i) ‘house’

*ji:b ‘house’

9, 47, 4, 47

50/8

13, 32b, 10, 32b

162/4

1, 32

41/8

27 stone/rock 30 tooth

*kada: ‘rock, cliff’ *pa ‘tooth’

*kada ‘rock, cliff’

*pal ‘tooth’ *ari-ga ‘molar, canine, fang’

31 hair

41/8

*kama ~ kami ‘hair of *kama ‘whirl of hair the head’ on the head’

13, 32, 26

1/16

*kara ~ ka(r)i ‘hair’

13, 32, 29

1/16

*kïl(k) ‘hair’

13, 40, 31

1/16

*kök ‘big, thick, healthy’

13, 37, 14, 37 188/3

*bạnɨ ‘big’

25, 32, 28

188/3

41, 6, 32, 28

188/3

3, 40, 31

235/2 109/4

*kal ‘hair’ *kilga ‘coarse hair’

32 big

*ar2(-)ïg ‘molar, fang’ 32, 30

*kɨkɨ- ‘be large in amount’

*hɨkɨ-‘big’

*mana-‘be big, many’ 33 one

*mani ‘large group’

*man ‘big, high’

*amban ‘big’

*amban ‘big’

*pilɨ- ~ *pilʌ-‘to begin’

*bir ‘one’

36 to hit/beat

*tuk-‘hit with force’

*t(ʌ)ki-‘hit, strike’

*dug- ‘hit’

9, 39, 16

85/5

37 leg/foot

*panki ‘lower leg, foot’

*pal ‘foot, leg’ PK *pʌlk ‘arm’

*palgan ‘foot’

1, 32, 18

15/10

39 this

*i ‘you’ (derogatory 2sg)

*i ‘this’ (proximal demonstrative)

*i ‘he, she, it’ (3sg)

*ï(-)n-‘that’ (distal demonstrative)

40

Gram

*kɨ-‘this’ (proximal demonstrative)

*kɨ-‘that’ (medial demonstrative)

*kö ‘this’ (proximal demonstrative)

13, 37

Gram

42

Gram

*ə-‘that’ (distal demonstrative)

*e-‘this’

*i ‘he, she, it’ (3sg)

*e-‘this’

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

LJ item

40 fish

*(y)iwə ‘fish’

*diɣa- ‘fish’

43 black 45 to stand

*kara ‘black’ *tata- ‘stand, rise, run high’

*ili-‘stand (up), rise’ *keme-‘bite, gnaw’

*ka wh-interrogative *ka interrogative particle particle

51 child

*wara-pa ‘child’

*(w)ura-‘sell’

*pʌlʌ-kʌ-‘sell’

54 new

*ara- ‘new, pure’

55 to burn (intr)

*tak- ‘burn (tr)’

*tʌkʌ-/*taki-‘be on / set fire’

56 not

*ana-negation

*an-negation

62 to hear

*uka- ‘receive, perceive, hear’

63 soil

*tutɨ-i ‘soil, ground’

65 red

13, 33, 26

233/3

93/6 Gram

*ba:la ‘young animal, 3, 32, 31, 32 child’

6/12

1, 39, 29

6/12

*tama- ‘pay’

7, 32, 26, 32

10/13

*bu:-‘give’

3, 39, 30

10/13

*arï- ‘be(come) pure’

41, 29

6/12 Poly

*yak-‘burn (tr)’

9, 32, 14

104/4

*ana-negation *e-se negation

*an-‘be unbecoming’ 41, 28, 32

253/2 Gram

*e-negation

42

253/2

41, 31

11/7

46, 14, 32

11/7

7, 38, 8

2/9

1, 39b, 31

288/4

*ala-‘make known, know’ *uka- ‘understand’

*tutɨ ‘bank, ridge, ground’ *pɨl-kɨ-‘be red’

32/5

*ka- wh-interrogative 21, 32 pronoun

*pure ‘child, seed’

*ari-‘be pure’

*alʌ- ‘know’

40, 18

25, 37, 10, 37 233/3

*xa- wh-interrogative pronoun

*e-negation 58 to know

*kem-‘bite, chew, gnaw’

*mödö-‘gnaw, bite’

*puril ‘child’ *tama- ‘give’

13, 32, 29, 32 63/5

*pula-‘be red’

*pula-ɣan ‘red’

*uk- ‘understand , hear’

(continued )

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

50 what?

67/4

7, 32b, 8, 32b 35/6 NO

*kam- ‘bite, chew’ *mɨlɨ-‘bite’

53 to give

*kara ‘black’

*tʌtʌ-‘run’ *ilʌ- / *ilɨ-‘come up, rise’

46 to bite

9, 40, 4, 34

Table 36.1  Continued LJ item

Proto-Japonic

67 to hide 68 skin/hide

*kapa ‘skin, bark, shell’

Proto-Koreanic

Proto-Tungusic

*sum- ‘hide, lurk in’

*sume- ‘hide, conceal’

Proto-Mongolic

*kap(ʌ)-k ‘skin, bark outer layer’

69 to suck

*ka:p-ïk ‘bark, shell’ *xökö-‘to suck (breasts)’

*əpə- ‘carry on back’

*ep-‘carry on back’

73 to take

*tɨra- ‘take, hold’

*tɨlɨ-‘hold up, lift, raise’

*kökü-‘to suck (breasts)’

*ebe- ‘carry’

*al-‘take’ 74 old

*muka-si ‘be long ago, ancient’

*al-‘take’

*muk- ‘be(come) old’ *kari-‘weaken’

77 thick

*puta- ‘be thick’

*putʌ- ‘become thick’

78 long

*nanka-‘be long’

*nʌlkʌ-‘be(come) old, long (in time)’

*büdü- ‘large’

*uri ‘long past, former’ *ola-‘last long’ 79 to blow 80 wood

*pulɨ-‘blow’

*ora-‘be late’ *pu:-‘blow’

*karï-‘be(come) old’

Sound Corr. no

Sem. Corr. no

23, 38, 26

222/4

13, 32, 2

1/16

21 37, 14

191/5

34, 4, 34

23/8

7, 37, 29

3/8

41, 31

3/8

25, 38, 14

62/6

13, 32, 29, 40 62/6 NO 3, 38, 10

188/3

27, 32b,18, 32b

17/7

*ur2a-‘be long (time/ 46, 29 space)’

17/7

43, 29, 32

17/7

1, 38

158/5

13, 37

4/13

25, 35

4/13 Poly

*kɨ ‘tree, wood’

*kɨl ‘tree, wood’

*moro ‘woods, wooded hill’

*molo ‘hill, mountain’

81 to run

*pasa-‘run’

*pas- ‘hurry’

1, 32, 24

80/4 NO

82 to fall

*tira-‘fall, scatter’

*ti-‘fall, scatter’

7, 40

141/5

84 ash

*papV ‘ash’

*pap ‘dust, waste’

3, 31, 2

66/8

86 dog

*inu ‘dog’

47, 28

230/6

*mo: ‘wood, tree’

*ina ‘dog’

*mo wood, tree’

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

70 to carry

Proto-Turkic

87 to cry/ weep 88 to tie

*ulɨ-‘cry, howl’ *kuku- ‘tie, wrap’

*uli-‘howl’ *xuku- ‘wrap’

*ma-‘to be tasty, sweet’

*ma-‘to be tasty’ *tʌlʌ-‘to be sweet’

21/9 21, 39, 14, 39 229/3 Poly

*boɣo-‘tie, bind, bundle’ 90 sweet

*u:lï-‘cry, howl’

*bog-‘tie, strangle’

3, 36, 16

229/3

9, 32b, 31

60/6

25, 31 *da:l-‘to be sweet’

*tura ‘rope, string, line’

*cul ‘rope, string, line’

19, 38, 31

55/8

92 shade/ shadow

*kan(a)-ka ‘reflection, shade’

*kʌnʌl-ʌk ‘shadow’

13, 32b, 28, 32b

43/7

94 salt

*tura-‘bitter, unbearable’

7, 39, 30

60/6

95 small

*tipi-‘to be small’

19, 40, 2

99/4

96 wide

*nənpa- ‘become wide and long’

27, 34, 6, 33

188/3

9, 31, 6

188/3

*tu:r2 ‘salt’ *čipi-‘to be small, narrow’ *nelpʌ-‘be wide’

*nepte- ‘become flat and wide’

*nebse-‘be wide and long’ *dalba-‘to be wide and flat’

*yalpa-‘to be wide, flat’

97 star

*pəsi ‘star’

*peli ‘star’

1, 34, 20b, 40 267/2

98 in

*soko ‘depth’

*soko ‘depth, deep inside’

23, 35, 14

36/6

99 hard

*kata-‘be hard’

100 to crush/ grind

*pinta-‘crush’

*kata-‘be hard’

*örü ‘interior’

*ö:r2 ‘interior’

44, 30

36/6

*kata-‘become hard’

*kat-‘be hard’

13, 32, 8, 32

70/6

1, 40, 12, 33

58/7

23, 38, 29

58/7 Poly

*pinče- ‘crush’

*sura- ‘grind, rub, make smooth’

*sür- ‘rub, smear’ *niki- ‘crush, knead’

*niku- ‘crush, knead’ yïk- ‘crush’

27, 40, 22

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

91 rope

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

Martine Robbeets Table 36.2  Consonant correspondences between the Transeurasian languages PJ

PK

PTg

PMo

PTk

PTEA

1.

*p-

*p-

*p-

*p-

*b-

*p-

2.

*-p-

*-p-

*-p-

*-ɣ-

*-p-

*-p-

3.

*p- / *w-

*p-

*b-

*b-

*b-

*b-

4.

*-p-/*-w-

*-p-

*-b-

*-b-/ -ɣ-

*-b-

*-b-

5.

*-np-

*-pC-

*-PC-

*-PC-

*-P(C)-

*-m(P)T-

6.

*-np-

*-Rp-

-RP-

*-RP-

*-RP-

*-Rp-

7.

*t-

*t-

*t-

*t-

*t-

*t-

8.

*-t-

*-t-

*-t-

*-t-

*-t-

*-t-

9.

*t- /*y-

*t- (ci-)

*d- (ji-)

*d- (ji-)

*y-

*d-

10.

*-t-/ *-y-

*-l-

*-d- (-ji-)

*-d- (-ji-)

*-d-

*-d-

11.

*-nt-

*-c-

*-TC-

*-TC-

*-TC-

*-n(T)K-

12.

*-nt-

*-Rc-

*-RT-

*-RT-

*-RT-

*-Rt-

13.

*k-

*k-

*k-

*k-

*k-

*k-

14.

*-k-

*-k- (-h-)

*-k-

*-k-

*-k-

*-k-

15.

*k-

*k-

*g-

*g-

*k-

*g-

16.

*-k-

*-k- (-h-)

*-g-

*-g-

*-g-

*-g-

17.

*-nk-

*-kC-

*-KC-

*-KC-

*-KC-

*-ŋ(K)T-

18.

*-nk-

*-Rk-

*-RK-

*-RK-

*-RK-

*-Rk-

19.

*t-

*c-

*č-

*č-

*č-

*č-

20.

*-t-

*-c-

*-č-

*-č-

*-č-

*-č-

20b.

*-si

*-l(i)

*-l(č)

*-l(č)

*-l(č)~ -š

*-lč

21.

*k-

*k-, h-

*x-

*k-

*k-

*x-

22.

*-k-

*-k-

*-x-

*-g-~-k-

*-g-~-k-

*-x-

23.

*s-

*s-

*s-

*s-

*s-

*s-

24.

*-s-

*-s-

*-s-

*-s-

*-s-

*-s-

25.

*m-

*m-

*m-

*m-

*b-

*m-

26.

*-m-

*-m-

*-m-

*-m-

*-m-

*-m-

27.

*n-

*n-

*n-

*n-

*y-

*n-

28.

*-n-

*-n-

*-n-

*-n-

*-n-

*-n-

29.

*-r-

*-l-

*-r-

*-r-

*-r-

*-r-

30.

*-r-

*-l-

*-r-

*-r-

*-r2-

*-r-

31.

*-r-

*-l-

*-l-

*-l-

*-l-

*-l-

652

OUP CORRECTED PROOF – FINAL, 27/05/20, SPi

basic vocabul ary in the tr anseur asian l anguages Table 36.3  Vowel correspondences between the Transeurasian languages OJ < PJ

MK < PK

PTg

PMo

PTk

PTEA

32.

-a- < *-a-

-a- < *-a-

*-a-

*-a-

*-a-

*-a-

32b.

*CaCa

*CʌCʌ

*CaCa

*CaCa

*CaC

*CaCa

33.

-a- < *-a-

-e- < *-e-

*-e-

*-e-

*-e-

*-ə-

34.

-o- < *-ə-

-e- < *-e-

*-e-

*-e-

*-e-

*-ə-

35.

-o-