Circum-Baltic Languages Volume 1: Past and Present 9027230579, 1588110206, 9027230595, 1588110427, 9789027230577

The area around the Baltic Sea has for millennia been a meeting-place for people of different origins. Among the circum-

144 94 3MB

English Pages 382 [405] Year 2001

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
The Circum-Baltic Languages......Page 1
Editorial page......Page 2
Title page......Page 3
LCC data......Page 4
Table of contents......Page 5
List of abbreviations......Page 7
List of contributors......Page 10
Introduction......Page 13
The Circum-Baltic Languages......Page 15
Part 1: Survey of selected Circum-Baltic languages and language varieties......Page 21
The Latvian language and its dialects......Page 23
The Lithuanian language and its dialects......Page 61
Russian varieties in the southeastern Baltic area......Page 101
Swedish dialects around the Baltic Sea......Page 157
The Finnic languages......Page 199
Part 2: Early history f the Circum-Baltic languages......Page 233
The origin of the Scandinavian languages......Page 235
Baltic influence on Finnic languages......Page 257
Part 3 Contact phenomena in minor Circum-Baltic languages......Page 275
The role of language contact in the formation of Karelian, past and present......Page 277
Syntactic code-copying in Karaim......Page 291
Yiddish in the Baltic region......Page 305
The North Russian Romani dialect......Page 333
On some Circum-Baltic features of the Pskov-Novgorod (Northwestern Central Russian) dialect......Page 359
Name index......Page 381
Language index......Page 389
Subject index......Page 395
Recommend Papers

Circum-Baltic Languages Volume 1: Past and Present
 9027230579, 1588110206, 9027230595, 1588110427, 9789027230577

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

The Circum-Baltic Languages

Studies in Language Companion Series (SLCS) The SLCS series has been established as a companion series to Studies in Language, International Journal, sponsored by the Foundation “Foundations of language”.

Series Editors Werner Abraham

Michael Noonan

Universities of Groningen, Berkeley and Vienna

University of Wisconsin-Milwaukee USA

Editorial Board

Joan Bybee

Christian Lehmann

University of New Mexvico

University of Erfurt

Ulrike Claudi

Robert Longacre

University of Cologne

University of Texas, Arlington

Bernard Comrie

Brian MacWhinney

Max Planck Institute, Leipzig

Carnegie-Mellon University

Marianne Mithun William Croft

University of California, Santa Barbara

University Manchester

Edith Moravcsik

Östen Dahl

University of Wisconsin, Milwaukee

University of Stockholm

Masayoshi Shibatani

Gerrit Dimmendaal

Kobe University

University of Leiden

Russell Tomlin

Martin Haspelmath

University of Oregon

Max Planck Institute, Leipzig

John Verhaar

Ekkehard König

The Hague

Free University of Berlin

Volume 54 The Circum-Baltic Languages: Typology and Contact Edited by Östen Dahl and Maria Koptjevskaja-Tamm

The Circum-Baltic Languages Typology and Contact

Edited by Östen Dahl Maria Koptjevskaja-Tamm Stockholm University

John Benjamins Publishing Company Amsterdam/Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data Circum-Baltic languages : Typology and contact / edited by Östen Dahl and Maria Koptjevskaja-Tamm. p. cm. (Studies in Language Companion Series, issn 0165–7763 ; v. 54–55) Includes bibliographical references and index. Contents: v. 1. Past and present -- v. 2. Grammar and typology. 1. Baltic Sea Region--Languages. I. Dahl, Östen. II. Koptjevskaja-Tamm, Maria. III. Series. P381.B33 C57 2001 491’.9--dc21 isbn 90 272 3057 9 (Eur.) / 1 58811 020 6 (US) (Hb, v. 1; alk. paper) isbn 90 272 3059 5 (Eur.) / 1 58811 042 7 (US) (Hb, v. 2; alk. paper)

2001025046

© 2001 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa

"adr"> "intro"> "bal1"> "bal2"> "cek1"> "cek2"> "ren"> "laa"> "dah"> "lar">

TITLE "Table of contents"

SUBJECT "Studies in Language Companion Series, Volume 54"

KEYWORDS ""

SIZE HEIGHT "220"

WIDTH "150"

VOFFSET "4">

Table of contents

List of abbreviations List of contributors

vii x

Introduction The Circum-Baltic Languages: Introduction to the volume Östen Dahl and Maria Koptjevskaja-Tamm

xv

Part 1 Survey of selected Circum-Baltic languages and language varieties The Latvian language and its dialects Laimute Balode and Axel Holvoet The Lithuanian language and its dialects Laimute Balode and Axel Holvoet Russian varieties in the southeastern Baltic area: Urban Russian of the 19th century Valeriy Cˇekmonas

3

41

81

Russian varieties in the southeastern Baltic area: Rural dialects Valeriy Cˇekmonas

101

Swedish dialects around the Baltic Sea Anne-Charlotte Rendahl

137

The Finnic languages Johanna Laakso

179

Part 2 Early history of the Circum-Baltic languages The origin of the Scandinavian languages Östen Dahl

215

Baltic influence on Finnic languages Lars-Gunnar Larsson

237

< R / /RT REREARGET EFEFFFF

"pug"> "csa"> "jac"> "rus"> "cek3"> "ni"> "li"> "si"> "toc">

vi

Table of contents

Part 3 Contact phenomena in minor Circum-Baltic languages The role of language contact in the formation of Karelian, past and present Stefan M. Pugh

257

Syntactic code-copying in Karaim Éva Ágnes Csató

271

Yiddish in the Baltic region Neil G. Jacobs

285

The North Russian Romani dialect: Interference and Code Switching Aleksandr Yu. Rusakov

313

On some Circum-Baltic features of the Pskov-Novgorod (Northwestern Central Russian) dialect Valeriy Cˇekmonas

339

Name index

i 1

Language index

i 9

Subject index

i 15



List of abbreviations

Abbreviations used in glosses abess abl abs acc acnnr act adess adj adjr adv advr all aux com comm comp cond conneg conv cop dat deb def dem det dim dir du dur elat

abessive ablative absolutive accusative action nominalizer active adessive adjective adjectiv(al)izer adverb adverbializer allative auxiliary comitative common (gender) complementizer conditional form of the verb used with negation converb copula dative debitive definite (article) demonstrative determiner diminutive direct dual durative elative

emph erg ess fem freq fut gen ger hab hon ill imp impf impr indef indir iness inf inst instruc int intr ipfv ips loc masc minf n neg nfin nom

emphatic ergative essive feminine frequentative future genitive gerund habitual honorific illative imperative imperfect impersonal indefinite indirective inessive infinitive instrumental instructive interrogative intransitive imperfective impersonal passive locative masculine Estonian ma-infinitive noun negative nonfinite nominative

viii List of abbreviations

nonpast nonpast nr nominalizer neut neuter obj object obl oblique opt optative part participle pass passive pf perfect pfv perfective pl plural po partial object poss possessive pot potential pp perfective particle ppa present participle active ppp past/preterit passive participle

pref prep pret pres prtv past refl sub sg subj suff sup term tinf to trnsl zerik

prefix preposition preterite present partitive past reflexive subject singular subjunctive suffix superlative terminative Estonian ta-infinitive total object translative Basque “zerik”-case

FSwd Gal Grg Grk Grm HGrm Hng Ice Ing Ir It Jat Kar Kom Krm Ksh Lat LGrm Lith

Finland Swedish Galindian Georgian Greek German High German Hungarian Icelandic Ingrian Irish Italian Jatvingian Karelian Komi Karaim Kashubian Latin Low German Lithuanian

Additional abbreviations Abx Arm Balt Blg BY Bylr CourlY Cur CY Cz Dal Dan Dut Eng ErzaMrd Est EY Fin Fr

Abkhaz Armenian Baltic Bulgarian Baltic Yiddish Belarusian Curonian Yiddish Curonian Central Yiddish Czech Dalecarlian Danish Dutch English Erza Mordvin Estonian: Eastern Yiddish Finnish French



List of abbreviations

LivK LivSal

Livonian in Curonia Salis-Livonian, Livonian in Vidzeme (near Salis). LRmn Latvian Romani Ltg LatgalianLtv Latvian Lud Ludian Mar Mari MarEast Eastern Mari Mrd Mordvin NEst Northern Estonian NEY Northeastern Yiddish NRRD North Russian Romani dialects Nsam North Sami OHGrm Old High German OLith Old Lithuanian Olo Olonetsian ONrs Old Norse OPrs Old Prussian Oss Ossete PC partitive nominal construction Plb Polabian

Pol PPC Rmn Rus SAE Sam SCr SEst SEY Slve Spn StY Swd SY Ttr Udm Ukr Vot Vps WY Yid ZY

Polish pseudo-partitive nominal construction Romani Russian Standard Average European Sami Serbian/Croatian South Estonian, Southeastern Yiddish Slovenian Spanish Standard Yiddish Swedish Southern Yiddish Tatar Udmurt Ukrainian Votian Veps Western Yiddish Yiddish zameter yidish, Samogitian Yiddish

ix



List of contributors Östen Dahl, [email protected] Dept. of Linguistics, Stockholm University, SE-106 91, Stockholm, Sweden Maria Koptjevskaja-Tamm, [email protected] Dept. of Linguistics, Stockholm University, SE-106 91, Stockholm, Sweden Laimute Balode, laimute.balode@helsinki.fi Dept. of Slavic and Baltic Languages and Literatures, P. O.Box 4, Vuorikatu 5B, FIN-00014 Helsinki, Finland Axel Holvoet, [email protected] Lietuviu kalbos institutas, Antakalnio 6, 2055 Vilnius, Lithuania Valeriy Cˇekmonas, [email protected] Dept. of Slavic languages and literatures, Vilnius University, Krokuvos 1–24, LT-2005 Vilnius 5; Lithuania Anne-Charlotte Rendahl, [email protected] Dept. of Linguistics, Stockholm University, SE-106 91, Stockholm, Sweden Johanna Laakso, [email protected] Institut für Finno-Ugristik der Universität Wien, Spitalgasse 2–4, Hof 7, A-1090 Wien, Austria Lars-Gunnar Larsson, Lars-Gunnar.Larsson@finugr.uu.se Dept. of Finno-Ugrian languages, University of Uppsala, Box 256, SE-751 05, Uppsala, Sweden Stefan Pugh, [email protected] Dept. of Slavic Languages and Literatures, Duke University, Durham, NC 27 706, UK Éva Á. Csató Johanson, [email protected] Department of Asian and African Languages, Uppsala University, Box 527, SE-751 20 Uppsala, Sweden



List of contributors

Neil G. Jacobs, [email protected] Department of Germanic Languages and Literatures. The Ohio State University, 314 Cunz Hall, 1841 Millikin Road, Columbus, OH 43210, USA Alexandr Rusakov, [email protected] Dept. of General Linguistics, St. Petersburg University, Lenin str. 11, apt. 4, 197136, St.Petersburg Russia Vytautas Ambrazas, [email protected] Kraziu˛ g. 7–3, 2001 Vilnius, Lithuania Bernhard Wälchli, [email protected] Dept of Linguistics, Stockholm University; and Inst. für Allgemeine Sprachwissenschaft, Universität Bern, Länggassstr. 11, CH-3012 Bern Helle Metslang, [email protected] Tallinn Pedagogical University, Chair of Estonian, Narva Road 29, 10120 Tallinn, Estonia Baiba Metuzale-Kangere, [email protected] Dept. of Baltic Studies, Stockholm University, SE-106 91, Stockholm, Sweden Simon Christen, [email protected] Stadtbachstr. 42a, CH-3012 Bern, Switzerland Leon Stassen, [email protected] Instituut voor Algemene Taalwetenschap, Erasmusplein 1 (k. 9–12), NL-6525 GG Nijmegen, The Netherlands Thomas Stolz, [email protected] Universität Bremen, Fachbereich 10: Linguistik Postfach 33 04 40, D-28334, Bremen, Germany Kersti Boiko, [email protected] Faculty of Modern Languages, Finno-Ugric Programme, University of Latvia, Visvalza 4a, Riga, LV-1050, Latvia

xi



Introduction



The Circum-Baltic Languages* Introduction to the volume Östen Dahl and Maria Koptjevskaja-Tamm

Areal linguistics and typology have multiple connections. The use of typology has long been recognized in areal linguistics: typological considerations are an important tool in evaluating suggested isoglosses, i.e. if typologically marked (crosslinguistically infrequent, idiosyncratic) features are attested in several neighbouring languages, this similarity can hardly be attributed to an accident. Typologists, on the other hand, have shown an increasing interest in areal generalizations: while some typologists try to minimize the risk for a potential areal (and genetic) bias in a sample by various sophisticated sampling procedures, others find uneven areal distributions of features a fascinating object of study: the most important recent contributions to areal typology include Dryer’s work on word order (1989), Nichols’ work on correlations among marking type and various other language parameters (1992), Dahl’s work on tense and aspect (1995), Stassen’s book on intransitive predication (1997). In practice, however, the contacts between areal linguistics and typology, as well as communication between experts in these two fields, often are weak. One basic problem is, of course, that both types of research require hard work with a huge amount of data, but with completely different focuses. For areal linguists, the desideratum is a complete documentation — both synchronically and diachronically — of linguistic properties in a restricted area, including minimal variation among closely related language varieties. Typologists, who sometimes find this preoccupation with details boring, dispense with a large portion of them — much to the irritation of areal linguists, who find this attitude superficial and suspicious. Practical difficulties are also an obstacle in contacts between areal linguistics and typology. Areal studies require enormous knowledge — of the languages spoken in the area and of the linguistic literature concerning their synchronic state and history. One can hardly expect areal linguists to have a good orientation in linguistic typology. On the other hand, one cannot expect that typologists will be able to have a good orientation in descriptions of various phenomena which appear or have appeared in small publications at various “obscure” places, are written in “smaller” languages and are directed towards a narrow circle of specialists.

xvi The Circum-Baltic Languages

The area around the Baltic sea provides ample illustration of these difficulties. Not only are there many languages and many dialects spoken here, but there are also several strong local linguistic traditions, some of them with fairly old roots. Making the information accumulated within those traditions more easily accessible to linguists outside them is a highly desirable but not an easily attainable goal. Recent developments in linguistics have brought about more intensive contacts with typologists and specialists in particular languages. Thus, the combined efforts of numerous linguists of different denominations involved in the programme “EUROTYP” resulted in many interesting generalizations about Europe as a linguistic area. The end of the cold war and new media of scientific communication paved a way for new fruitful dialogues among various specialists. There are still seemingly unsurmountable difficulties. General typologists and specialists in Baltic, Finno-Ugric, Slavic and Germanic languages will all perceive the goals of the areal study of CB-languages differently. The present pair of volumes represents the first major attempt to reconcile those differences. Our topic, then, is Circum-Baltic languages — the languages spoken around the Baltic Sea. Obviously, the delimitation of this set of languages will have to remain vague, for several reasons. The first source of vagueness resides in the

Northern Sami

Inari Sami

Skolt Sami

Lule Sami Pite Sami Ume Sami

Karelian

Southern Sami

Finnish Olonetsian

Dalecarlian

Norwegian

Swedish Estonian Livonian

Latvian Danish Northern Frisian Low German

Ludian Veps

Ingrian Votian

Russian

Lithuanian Kashubian

Karaim

Belarusian Polish High German

Map 1.The Circum-Baltic languages. Non-territorial languages not shown: Romani varieties, Yiddish, Tatar.

The Circum-Baltic Languages xvii

preposition “around”. If a language is spoken on the coast of the Baltic, the case is clear, but how far from there should we go? Then, some languages may not have native speakers in the area but are still important for the study of language contacts, for instance, Latin and French. Should they be included? Finally, to make a list of the Circum-Baltic languages we need to draw the borderline between languages and dialects or varieties, a notoriously hopeless task. The list of Circum-Baltic languages given below, and the map of the area (Map 1), are therefore somewhat arbitrary, and should be taken only as a starting-point for the discussion.

Organization of the volumes The first of the two volumes — Circum-Baltic Languages: Volume 1 — Past and Present — surveys important sub-groups in the present-day Circum-Baltic languages, placing them in their geographical, historical and societal setting and discussing specific contact situations. The second volume — Circum-Baltic Languages: Volume 2 — Grammar and typology — focuses on grammatical phenomena in the Circum-Baltic languages, relating them to the larger typological perspective. Each of the volumes contains three sections. The first section of the first volume contains overviews of four subsets of Circum-Baltic languages and language varieties, representing all the major languages families in the area. Latvian and Lithuanian and their dialects are presented in two chapters by Laimute Halmode and Axel Holvoet. Johanna Laakso and AnneCharlotte Rendahl give surveys of the Finnic languages and the CB Swedish dialects, respectively. Finally, Valeriy Cˇekmonas discusses Russian dialects in the CB area in two chapters. The second section is devoted to the early history of the CB languages. Östen Dahl discusses the origin of the Scandinavian languages and Lars-Gunnar Larsson the influence of the Baltic languages on the Baltic Finnic languages. The third and last section of the first volume treats contact phenomena in some of the minor (in terms of number of speakers) CB languages and language varieties. Karaim, a Turkic language spoken by a small group in Lithuania, is treated by Éva Ágnes Csató. The formation of Karelian, a Finnic language spoken in the Karelian Republic (Russia) is discussed by Stefan M. Pugh. Neil Jacobs surveys the varieties of Yiddish in the CB region and Aleksandr Yu. Rusakov discusses interference and code switching in the variety of Romani spoken in Northern Russia, and Valeriy Cˇekmonas looks for contact-induced phenomena in the Pskov-Novgorod dialect of Russian. The first and largest section of the second volume comprises six chapters, which all treat grammatical phenomena in the languages east of the Baltic from the point of view of diachronic development and areal influence. Three of them focus on

xviii The Circum-Baltic Languages

nominal case: Simon Christen discusses different syntactic positions in which the genitive case may appear in the Baltic and Finnic mentioned and Baiba Metuza¯leKangere and Kersti Boiko compare the case systems of Latvian and Estonian. Vytautas Ambrazas concentrates on a more specific diachronic development: how the use of the nominative for object marking arose in the eastern CB area. The contributions of Helle Metslang and Bernhard Wälchli both treat the historical development of the use of verb particles for aspect or Aktionsart marking in Estonian, Latvian and Livonian. The three chapters in the following section also treat grammatical phenomena, but from a more explicitly typological point of view. A shared focal point of the chapters is the role of nominal case in various syntactic constructions: Leon Stassen’s chapter with the role of cases such as the instrumental, essive and translative in nonverbal predication, Maria Koptjevskaja-Tamm’s with partitive and pseudo-partitive constructions, and the chapter by Thomas Stolz with the expression of comitative and instrumental roles. In the concluding chapter of the second volume, Maria Koptjevskaja-Tamm and Bernhard Wälchli survey a number of important features of CB languages, arguing that although the notion of a Sprachbund is not satisfactory for characterizing the linguistic situation in the CB area, the study of the CB languages from an areal-typological point of view reveals a linguistic landscape with many interesting properties of its own.

List of Circum-Baltic languages Germanic West High German (HGrm) Low German (LGrm) Yiddish (Yid) North Danish (Dan) Swedish (Swd) Dalecarlian (Dal) Norwegian (Nrw) Baltic West †Old Prussian (OPrs) ††Curonian (Cur) †Jatvingian (Jat)

The Circum-Baltic Languages xix

Central Lithuanian (Lith) Latvian (Ltv) East ††Galindian (Gal) Slavic West Polish (Pol) Kashubian (Ksh) †Polabian (Plb) East Belarusian (Bylr) Russian (Rus) Ukrainian (Ukr) Indo-Aryan Romani (Rmn) with varieties/sub-languages: Kelderash, Lovari, Kalo, Baltic, North Russian Finno-Ugrian Finnic Veps (Vps) Karelian (Kar) Olonetsian (Olo) Ludian (Lud) Finnish (Fin) Ingrian (Ing) Votian (Vot) Estonian (Est) with varieties/sub-languages: South Estonian, Northern Estonian (NEst) Sami (Sam) with varieties/sub-languages: Southern Sami, Ume Sami, Pite Sami, Lule Sami, Northern Sami, Inari Sami, Skolt Sami Turkic Karaim (Krm) Tatar (Ttr) † = extinct; †† = only onomastic sources and substratum In addition, more or less isolated dialects under strong influence of other languages, e.g. *Leivu (a Hargla Estonian dialect between Aluksne and Gulbene/Latvia), *Krevinian (Votian near Bauska/Latvia), Estonian Swedish, Nehrungskurisch, Latgalian, Russian of the Old Believers in the Baltics, Urban Russian in the Baltics,



xx

The Circum-Baltic Languages

Baltendeutsch, Halbdeutsch.

Note * In 1991, a six-year research program called “Language Typology around the Baltic Sea” was launched by the Faculty of Humanities at Stockholm University, with Maria Koptjevskaja-Tamm as main researcher. A large part of the work behind these volumes was supported financially within this research program. We want to express our thanks here both to our sponsors and to all the people who have contributed to the volumes.

References Dahl, Östen. 1995. “Areal tendencies in tense-aspect systems”. In: Bertinetto, Pier Marco, Bianchi, Valentina, Dahl, Östen & Squartini, Mario (eds.), Temporal Reference, Aspect and Actionality. Vol. 2: Typological perspectives, 11–28. Torino: Rosenberg & Sellier. Dryer, Matthew. 1989. “Large linguistic areas and language sampling”. Studies in Language 62: 808–45. Nichols, Johanna. 1992. Linguistic Diversity in Space and Time. Chicago and London, The University of Chicago Press. Stassen, Leon. 1997. Intransitive predication. Oxford studies in typology and linguistic theory. Oxford: Clarendon Press.



Part 1

Survey of selected Circum-Baltic languages and language varieties



The Latvian language and its dialects* Laimute Balode and Axel Holvoet

General data 0.1 Number of speakers Latvian is now spoken by about 1,340,000 people in Latvia, and by about 110,000 people outside Latvia (though some sources estimate the number of speakers outside Latvia at about 300,000). No figures are available as concerns the number of representatives of the individual dialects. In 1935, Latvians constituted 77% of the population of Latvia. The largest minorities were: Russians (8.8%), Jews (4.9%), Germans (3.3%), Poles (2.5%) and Belarusians (1.4%). In 2000, 57.6% of the population consisted of Latvians, and the largest minorities were: Russians (29.6%), Belarusians (4.1%), Ukrainians (2.7%), Poles (2.5%) and Lithuanians (1.4%). The Livonians, a Finnic population once inhabiting most of the coastal regions of Latvia, have been largely assimilated. At the close of the 19th century, some 3,000 people spoke Livonian, but their number has now dwindled to less than a hundred. 0.2 Territorial division Latvia is traditionally divided into four regions: – – – –

Kurzeme (Courland proper, the western part of the former Duchy of Courland); Vidzeme (the western part of Livonia, former Swedish Livonia); Latgale (Letgalia, the eastern part of Livonia, former Polish Livonia); Zemgale (Semigalia, the eastern part of the former Duchy of Courland, around Jelgava, Germ. Mitau, and further to the east); the easternmost part of Semigalia, stretching in a narrow strip south of the Daugava, is sometimes singled out and referred to as Upper Courland (Augškurzeme) or Selonia (Se¯lija).

4

Laimute Balode and Axel Holvoet

Tamian dialects Livonian dialects

Central dialects High Latvian dialects

Curonian dialects Semigalian dialects

1.

The Latvian language

1.1 Latvian among the Baltic languages When compared to Lithuanian, with which it constitutes the eastern branch of the Baltic languages, Latvian appears to have gone through a more rapid development. For the common Eastern Baltic ancestor language we may reconstruct sound shapes and forms quite similar to those of modern Lithuanian. The most important features which give Latvian its present-day appearance are the following: 1.1.1 The original mobile stress has been replaced by an immobile stress on the first syllable, cf. Lithuanian metù m«etam(e)

Latvian [mjæÁtu] [«mjæ˜tamjæ]

metu metam

meaning [Ámæt˜u] [Ámætam]

‘throw’ (pres.1sg) ‘throw’ (pres.1pl)

All vowels in final syllables, whether open or closed, have undergone some kind of shortening or reduction; this may involve:

The Latvian language and its dialects

a.

frequent loss of original short vowels, usually retained in Lithuanian:

Lithuanian m«eta m«etat(e) naktìs

Latvian [«mjæ˜ta] [«mjæ˜tatjæ]/[«mjæ˜tat] [nakjÁtjis]

met metat nakts

meaning [mæt] [Ámæt˜at] [naktës]

‘throw’ (pres.3) ‘throw’ (pres.2pl) ‘night’ (nom.sg)

b. shortening of long vowels and diphthongs (Lithuanian has retained part of the old word-final long vowels and diphthongs, though shortening occurs here as well): di«enos a«kys linaı˜ lieta«us

[«djieno˜s] [«alkji:s] [lji«nai] [ljie«taus]

dienas acis lini lietus

[Ádienas] [Áatëslis] [Álini] [Álietus]

‘day’ (nom.pl) ‘eye’ (nom.pl) ‘flax, linen’ (nom.pl) ‘rain’ (gen.sg)

Original sequences of the type /V + N/ (V representing any vowel and N any nasal sonorant) have undergone changes before a consonant or word-finally. They have been replaced with long vowels or diphthongs (probably through an intermediary stage of nasalized vowels): kri«nta penkì

[«kjrjin.ta] [pjænjÁkji]

krı¯t pieci

[kri˜t] [Ápietësi]

‘fall’ (pres.3) ‘5’ (nom.masc)

For consonants, the most striking feature is the change of palatalized velars into dental affricates (probably through the intermediary stage of palato-alveolar affricates): kèpti gérti

[Ákjæpjtji] [´gjæ˜rjtji]

cept dzert

[tësept] [dëzert]

‘bake’ (inf) ‘drink’ (inf)

1.1.2 In morphology, Latvian has carried through a number of innovations (the central dialect is the most innovative in this respect, cf. 3.1.2 below): 1.1.2.1 The neuter forms of adjectives and pronouns have been completely lost, whereas Lithuanian has largely retained them (nouns are either masculine or feminine in both languages). 1.1.2.2 Simplification of nominal declension: virtual loss of the instrumental case (which had certain consequences for the government of prepositions, cf. 1.1.3), partial (for adjectives, complete) loss of distinctions between the inherited declension classes.

5

6

Laimute Balode and Axel Holvoet

1.1.2.3 Simplification of verbal inflection: the distinction of three types of presents (-a-, -a¯- and -i-stems) is virtually abandoned; in most dialects the distinction of two types of preterite stems is abandoned as well. 1.1.2.4 A special inflectional form of the verb was created to express necessity, corresponding to the modal verbs ‘must’, ‘have to’ of other languages: the so-called debitive. The object of a debitive is usually in the nominative, the subject (agent) in the dative: Man ja¯pl¸auj siens. me:dat mow:deb.pres hay:nom ‘I have to mow the hay.’

The debitive is not conjugated for person, and the tenses are formed by addition of the corresponding tense forms of the auxiliary bu¯t ‘be’. The debitive was originally an agglutinative form arising from the coalescence of the relative pronoun with the verb in a non-finite relative clause added to the subject of a possessive construction (cf. 1.1.4). The construction from which the above example arose would have to be rendered literally as ‘I have some hay (which) to mow’. 1.1.3 In morphosyntax, a striking feature of Latvian is the fact that prepositions can govern different cases in the singular and the plural. This is probably due to the loss of the instrumental, which coalesced with the accusative in the singular and with the dative in the plural. As a result, some prepositions seemed to govern the dative in the plural, but the other case form in the singular. This pattern was generalized, and in most of the dialects all prepositions now govern the dative plural, regardless of the case they govern in the singular (though some dialects retain the genitive plural with at least some prepositions). 1.1.4 A peculiarity of Latvian syntax is its possessive construction, which is not based on a verb meaning ‘have’, but on the verb ‘be’, with the object of possession usually in the nominative (though in some dialects it is always in the accusative), and the possessor in the dative (cf. Latin mihi est ‘I have’). Man ir ma¯ja. me:dat be:pres.3 house:nom.sg ‘I have got a house.’

This construction is rare in Lithuanian, which has the verb ture˙ti ‘have’, but it is reminiscent of the possessive constructions used in the Finnic languages and in Russian. This possessive construction formed the basis for a peculiarly Latvian form expressing necessity, cf. 1.1.2.4.

The Latvian language and its dialects

1.2 The formation of the Latvian language At the end of the 12th century (the earliest period covered by the chronicles relating the German missionaries’ activities and the subsequent conquest of Livonia by the Swordbearers or Livonian Knights), the territory of present-day Latvia was inhabited by five tribes: –



the Finnic Livonians (Livones), who occupied a coastal area reaching much farther inland than nowadays; they were the first with whom the German merchants, missionaries and knights established contact, and so gave their name to the whole of present-day Latvia and Estonia (German Livland, whence the Latinized form Livonia and, in later times, Polish Inflanty); subsequently this denomination was restricted to the Duchy of Livonia, extending to the north-east of the Daugava; four Baltic (Indo-European) tribes: – the Curonians (Curones) in the western part of Latvia (Courland proper); – the Semigalians (Semigalli) east of the Curonians, to the south of the Bay of Riga; – the Selonians (Selones), east of the Semigalians, in Upper Courland; – the Letgalians (Lethgalli), in the central and eastern parts of Latvia, to the north-east of the Daugava.

The territories of all these Baltic tribes, with the exception of the Letgalians, extended into the territory of present-day Lithuania; they were thus partly assimilated by the Lithuanians. It is now generally recognized that Curonian was a separate dialect of Baltic, displaying a series of features which place it in an intermediate position between Lithuanian and Latvian (some would prefer to say: between Eastern and Western Baltic). Its phonetic features can partly be reconstructed on the basis of place names, which, in turn, make it possible to identify a number of Curonian substratum features in present-day Latvian and Lithuanian, occurring in, or spreading from, formerly Curonian areas. Much less is known about Semigalian and Selonian, though a few putative substratum features are ascribed to these dialects as well. The fourth of the Baltic tribes, the Letgalians, must have played a crucial part in the formation of the Latvian nation and the Latvian language. In the 13th century Chronicon Heinrici the terms Let(t)(h)i ‘Latvians’ and Letthigalli ‘Letgalians’ are used interchangeably (cf. Lethi, qui proprie dicuntur Lethigalli ‘the Latvians, properly called Letgalians’), i.e. the Letgalians were the Latvians par excellence. They did not resist the Livonian Order as fiercely as other tribes (such as the Semigalians) did, and even collaborated with it. In recognition of this they were given preferential treatment by the Order, probably also in the form of allotments of land in areas depopulated by war. It is thus conceivable that the Latvian language as we know it

7

8

Laimute Balode and Axel Holvoet

is the result of Letgalian expansion into areas originally inhabited by Curonians, Semigalians, Selonians and, of course, (non-Baltic) Livonians. This theory, advanced by historians and once commonly held by linguists as well, has been disputed in more recent times by linguists advocating the view that the present-day Letgalian dialect is a separate Baltic language, separate from Latvian. These linguists tend to project the distinctive features of Letgalian into a distant past, and to regard this dialect as the only direct continuation of the ancient language of the Letgalian tribe. This Letgalian separatist view raises serious doubts, however, as none of the distinctive features of Letgalian can be considered ancient. In the domain of phonetics, Letgalian stands clearly apart from the rest of Latvian dialects, but none of its phonetic features can be projected into a distant past. In morphology, Letgalian has a number of archaisms, but these are typical peripheral archaisms also found in parts of Courland. Otherwise Latvian is relatively monolithic: there are no isoglosses going back to a dialectal differentiation within Common Baltic (Proto-Latvian). If we do find forms displaying, in their phonetic shape, a treatment deviating from the common Latvian sound laws, these are substratum forms, which is not surprising if we take into account the theory of Letgalian expansion expounded above. In addition, some of the differentiating features used in classifying Latvian dialects are connected with external influence, i.e. with a non-Indo-European substratum. Apart from the theory of the Letgalian origin of Latvian, other expansion theories have been advanced as well. Some scholars are inclined to assign a certain role to a putative Semigalian expansion as a unifying factor contributing to the relative uniformity of the Latvian dialects. Latvian is attested in writing since the end of the 16th century. From the 16th to the early 19th century Latvian was written mainly by German Lutheran pastors, who used the central dialect as the vehicle of their religious and didactic literature. With regard to its dialectal basis, their language was remarkably uniform from the start. Under the influence of the Latvian national revival, the Latvian written language was reformed from the 1860s onwards: German influences in syntax and lexicon were banished (many German borrowings being replaced by neologisms). From the 18th century onwards, there is also a tradition of writing in the High Latvian (Letgalian) vernacular. Its rise was conditioned by the political and religious separation of (Roman Catholic) Polish Livonia from (Lutheran) Swedish Livonia and Courland. The functional scope of the High Latvian written standard is restricted: it is connected mainly with the pastoral activities of the Roman Catholic Church, though, to a limited extent, it has been used for literary purposes as well, and it enjoys some degree of official recognition, e.g., in education.

The Latvian language and its dialects

1.3 Substrata and adstrata It follows from the above survey of the original population structure of the territory of present-day Latvia that we must reckon with a Finnic substratum in Latvia. In fact, several layers of substratum features may be distinguished. Though archaeological and toponymical data are not unambiguous, it seems probable that the whole of present-day Latvia was once inhabited by Fenno-Ugric tribes, who entered the Baltic area in the 3rd millennium B.C. They were probably pushed back to the north about the 12th century B.C. by Baltic tribes coming from the south-east (the bearers of the so-called Hatched Pottery Culture). A few features distinguishing Latvian as a whole from Lithuanian and Old-Prussian can be ascribed to Finnic influence. The initial stress of Latvian, as opposed to the mobile stress of Lithuanian and Old Prussian, is most often cited as an instance of this (which does not mean that no other explanations for this feature are available). Of course, a stronger admixture of Finnic features may be expected in those areas where the presence of a Finnic population is attested in historic times, i.e. from the 12th century onwards. The so-called ‘Livonian’ (Tamian) dialects of Latvian are defined mainly on the basis of features which can be traced back to a Finnic substratum. On the other hand, there are substratum features in those Baltic dialects which were assimilated by Latvian as a result of the Letgalian (and perhaps also, to a certain extent, the Semigalian) expansion. These are easiest to identify for Curonian, which has had a few distinctive features opposing it to what is now known as Latvian, and partly also to Lithuanian from Common Baltic times onward. Like Lithuanian, Curonian preserves the nasal diphthongs /aN/, /eN/, /iN/, /uN/. These combinations are called diphthongs (or diphthongoid sequences) because, even though the sonorants can no longer be syllabic in Baltic, in combination with vowels, they function as long vocalic segments, i.e. they distinguish tones (see 1.5.5.3 below). In the prehistory of Latvian they changed into nasal vowels which were subsequently denasalized, cf. 1.1.1. Forms which retain the structure /V + N/ occur everywhere in Latvian, perhaps as a result of interdialectal borrowing, but they are particularly frequent in Courlandish dialects, where they are considered Curonianisms. A few other features, recognized as characteristic of Curonian, have left traces only in toponymy. Considerably less is known about Semigalian and Selonian, but the view has been advanced that they had retained the nasal diphthongs in the same way as Curonian had. Some Slavonic languages (Polish, Russian, Belarusian) have evidently played the part of adstratum languages with regard to the High Latvian dialects of Letgalia (Polish Livonia), which show a number of phonetic features strongly reminiscent of Slavonic (apart from strong lexical influences).

9

10

Laimute Balode and Axel Holvoet

1.4 External influences In prehistoric times (i.e. before the 13th century), at least some of the Baltic tribes inhabiting Latvia, mainly the Letgalians, were tributary to the North-Russian principalities of Polock and Pskov, and must have undergone a strong cultural influence from them. The oldest layer of religious terminology connected with Christianity (prior to the activities of German missionaries) is of Slavonic (Old Russian) origin, e.g. baznı¯ca ‘church’ (OR božı˘nica), gre¯ks ‘sin’ (OR greˇchu˘), etc. Slavonic influence was discontinued as a result of the conquest of Livonia by the Swordbearers (the Livonian Order). It was renewed after the incorporation of part of the Livonian Order State (the Duchy of Livonia) into the Polish-Lithuanian Commonwealth, though it had lasting effects only in that part of it where PolishLithuanian ascendancy was maintained after 1660 (so-called Polish Livonia). In this area, Latvian (Letgalian) was exposed to the influence of Polish and East Slavonic (Russian and Belarusian). In Courland and Swedish Livonia, German influence was strong; this is reflected in a large number of loans, the oldest layer of which (up to the 16th century) is Low German. German-Latvian bilingualism was, however, very restricted, as the German gentry tended to isolate itself from the native population rather than to assimilate to it. Structural (e.g., syntactic) influence of German on Latvian should therefore not be overestimated as far as the popular dialects are concerned. In the written language, German syntactic influence was, of course, pervasive, as Latvian was written mainly by German pastors up to the 19th century. After the incorporation of the whole of Latvia into the Russian Empire (Swedish Livonia — 1721, Polish Livonia — 1772, Courland — 1795) Russian influence was initially slight (with the exception of Polish Livonia, incorporated into the Vitebsk Province) owing to the autonomy which the Baltic provinces enjoyed within the Russian Empire. A policy of Russification was pursued more forcibly from the 1880s onward, and it was renewed after the incorporation of Latvia into the Soviet Union in 1940. 1.5 Phonemic inventory A survey of the phonemic inventory is here given for the standard language; divergences from this system will be noted in the survey of the individual dialects. 1.5.1Vowel system Vowels i(˜)

u(˜) e(˜)

o(˜) æ(˜)

a(˜)

The Latvian language and its dialects

In the standard orthography, length is marked by a macron (·a¯Ò, ·e¯Ò etc.). The phonemes /e/ and /æ/, /e˜/ and /æ˜/ are not distinguished in orthography, although the opposition is distinctive, as shown by the minimal pairs cep / tësep/ ‘you bake’ vs. cep /tësæp/ ‘s/he bakes’, /sle˜dëz/ ‘you shut’ vs. /slæ˜dëz/ ‘s/he shuts, they shut’ (both written sle¯dz). In textbooks and dictionaries, the phonemes /æ/ and /æ˜/ are sometimes marked ·eÒ í and ·e¯í Ò. The vowels /o/ and /o˜/ occur in recent loans only, and quantity is not marked here (i.e. the macron is not put over ·oÒ). Rising diphthongs (i.e. with rising degree of sonority) ie

uo

The second component is actually a centralized vowel, which is more fronted in the case of /ie/ than of /uo/. The transcriptions /i6/, /u%/ would describe them best for practical purposes. In standard orthography these diphthongs are written ·ieÒ and ·oÒ respectively; the letter ·oÒ can thus denote either a diphthong (in native words and older borrowings) or the vowel /o/ (in recent borrowings). In historical and comparative literature ·uoÒ is written instead of ·oÒ when the diphthong is meant. Falling diphthongs (i.e. with diminishing degree of sonority) ui

iu ei

(oi) ai

au

/oi/ is restricted to recent borrowings (boikots ‘boycot’). The inclusion of /ui/ and /iu/ is a matter of convention, as both components have a similar degree of sonority here. To the diphthongs occurring in this table we can add the positional diphthongs /eu/ and /ou/, which are variants of the sequences /ev/ and /ov/ before consonants and word-finally, e.g. tev /teu/ ‘you’ (dat) as opposed to tevi /tevi/ ‘you’ (acc). The diphthongs /au/ and /iu/ can also be positional diphthongs, e.g. tavs /taus/ ‘your’ (nom.sg.masc) compared to tava /tava/ ‘your’ (nom.sg.fem) and zivs /zius/ ‘fish’ (nom/gen.sg) as compared to zivis /zivis/ ‘fish’ (nom/acc.pl). 1.5.2Origins of the vowel system Most Latvian vowels come from similar IE vowels, but IE *o has merged with *a into /a/, and its long counterpart *o¯ is represented by the diphthong /uo/. In the 19th century, /o/ and /o˜/ were reintroduced through borrowings. The vowels /e/ and /æ/, and their long counterparts /e˜/ and /æ˜/, are from IE *e and *e¯ respectively; the split was conditioned by the phonetic environment, the closed /e/ occurring when the following syllable contained a front vowel or /j/. Hence metu /mæt˜u/ ‘throw’ (pres.1sg) and met /mæt/ ‘id.’ (pres.3) from an original form *meta, in contrast to met /met/ ‘id.’ (pres.2sg) from an older form meti. The falling diphthongs basically

11

12

Laimute Balode and Axel Holvoet

also continue similar IE diphthongs, with the exception of /ui/ (occurring mainly in borrowings) and /iu/ (in onomatopoeia only). IE *oi and *ai have, of course, coalesced into /ai/; this diphthong, as well as *ei, is sometimes retained as such, but under certain conditions (the precise nature of which is still unclear) both are changed into the rising diphthong /ie/. 1.5.3Consonant system

Nasal Plosive Affricate Fricative Approximant Lateral Trill

labial

dental & alveolar

m pb

n td tës·cÒ dëz·dzÒ sz

fv

l r

palatoalveolar

palatal

velar

\ ·n¸Ò c ·k¸Ò K ·gˇÒ

(]) kg

tëw·cˇÒ ·džÒ w ·šÒ Š ·žÒ

x ·hÒ j ˆ ·l¸Ò

The characters used in standard orthography are added in angle brackets. The phonemes /f/ and /x/ were originally absent from the Latvian phonemic system; they occur only in recent loans. There is a distinct series of palatal consonants, but palatalized consonants (i.e. with palatalization as a secondary articulation) are now lacking. The only exception used to be palatalized /rj/ (written ·r¸Ò), but this sound is becoming extinct in both dialects and standard language, and in the standard orthography it was abolished in 1946. 1.5.4Origins of the consonant system The voiceless plosives come from similar IE sounds; the voiced plosives may represent IE unaspirated or aspirated voiced plosives (i.e. /d/ may represent IE *d or *dh). The IE palato-velars *k’, *g’ are represented by /s/, /z/ (as opposed to /w/ ·šÒ, /Š/ ·žÒ in Lithuanian), so that Latvian /s/ can be of twofold origin (IE *s or *k’). The Latvian palato-alveolar fricatives /w/ ·šÒ, /Š/ ·žÒ arise from combinations of /s/ and /z/ with /j/ (e.g. šu¯t ‘sew’ as opposed to Lith. siu¯ti). The dental affricates have arisen from original /k/ and /g/ before front vowels, or from the combinations of /k/ and /g/ with /j/ (cf. celt ‘lift, build’ as against Lith. kelti, dzert ‘drink’ as against Lith. gerti). The palato-alveolar affricates /tëw/ ·cˇÒ and /dëŠ/ ·džÒ appear (apart from in loans) as a result of analogical development, having been substituted for original /tës/ and /dëz/ by analogy with the alternation /s/ : /w/ and /z/ : /Š/ appearing in inflectional paradigms, e.g. la¯cis ‘bear’, gen.sg la¯ˇca (instead of *la¯ca) by analogy with the type lu¯sis ‘lynx’, gen.sg lu¯ša. The palatal sonorants /ˆ/ ·l¸Ò and /\/ ·n¸Ò arose from

The Latvian language and its dialects

combinations of /l/ and /n/ with /j/. In native Latvian words, the palatal plosives /c/ ·k¸Ò and /K/ ·g’Ò occur only in the clusters /wc/ ·šk¸Ò and /ŠK/ ·žgˇÒ, arising from /sk/ and /zg/ before front vowels (and from the combinations of these clusters with /j/), e.g. šk¸irt ‘separate, distinguish’ as opposed to Lith. skirti ‘id.’. In isolation, they appear mainly in borrowings, where they represent German /k/ and /g/ before front vowels (e.g. g’ek¸is ‘fool’ from Germ. Geck) and Russian palatalized /tj/, /dj/ (e.g. sog’is ‘judge’ from Old Russ. sudija). 1.5.5Prosodic features 1.5.5.1 Stress. Modern Latvian has a basically non-distinctive, immobile stress on the initial syllable. Exceptions are few; they comprise negative pronouns (ne’kas ‘nothing’), and compounds developed from various types of phrases. The latter group comprises mainly adverbs, such as uzÁreiz ‘at once’ (from the prepositional phrase uz reizi ‘at one time’), and superlatives such as visÁlaba¯kais ‘the best’, originally visu laba¯kais ‘the best of all’. Minimal pairs are extremely rare, e.g. Áa¯re ‘arable land’ vs. a¯Áre ‘look here’. 1.5.5.2 Quantity. Vowels can be short or long. Quantity is independent of stress, but is ultimately connected with tone (see below 1.5.5.3). Quantity is distinctive in any position: saka [Ásak˜a] ‘say’ (pres.3) : sa¯ka [^sa˜ka] ‘begin’ (pret.3) laba [Álaba] ‘good’ (nom.sg.fem.indef) : laba¯ [Álaba˜] ‘id.’ (nom.sg.fem.def)

1.5.5.3 Syllabic tones (syllabic accents). The syllabic tones or accents of Latvian are different modes of realization of word stress. Just as word stress commonly involves (with varying degrees of prominences) three features: expiratory strength, tone height, and quantity (length), so does syllabic accent. The difference consists in that, in the case of long vocalic segments, these features are unevenly spread, and may either be concentrated on its initial part (in the first mora), or be spread over the initial and final parts. An additional feature occurring in combination with tonal features proper is glottalization. As tonal differences (pitch) are not the only feature involved here, the term ‘tones’ is less appropriate than ‘syllabic accent’; still less felicitous is the term ‘intonation’ (still used in Latvian terminology: intona¯cija), which could be confused with sentence intonation. As mentioned, tone is exclusively a feature of long vocalic segments. As such are regarded: 1. long vowels, e.g. ma¯cı¯t [ma˜tësi˜t] ‘teach’ (inf) 2. falling and rising diphthongs with /i/ and /u/ as their second resp. first component, e.g.

13

14

Laimute Balode and Axel Holvoet

teikt ‘say’ (inf) augt ‘grow’ (inf) liepa ‘lime, linden’ (nom.sg) roka ‘hand’ (nom.sg) 3. diphthongoid sequences consisting of a vowel and a sonorant. Though, in principle, sonorants cannot be syllabic in Latvian (they could in Indo-European), their combinations with short vowels yield long (i.e. able to carry tone), vocalic segments, e.g. manta ‘thing, possession’ (nom.sg) varde ‘frog’ (nom.sg) Some of the central dialects (and, theoretically, the standard language as well, see below), have a system of three tones. a. The falling tone (krı¯toša¯ intona¯cija, German fallender Ton). The initial part of the vocalic segment is pronounced with greater intensity and a higher pitch; both intensity and pitch decrease toward the end of the vocalic segment. b. The drawn tone (stiepta¯ intona¯cija, German Dehnton) shows no decrease in either intensity (expiratory strength) or pitch toward the end of the articulation of the vocalic segment: both are maintained at the same level. The total average length of a vocalic segment with drawn tone exceeds that of a segment with falling tone by about 1/3. In the case of diphthongs and diphthongoid sequences, the length of the second component (the half-vowel or sonorant) distinctly exceeds that observed in falling diphthongs. c. The broken tone (lauzta¯ intona¯cija, German Stoßton) is similar to the falling tone in that the initial part is more prominent in both intensity and pitch, but after this initial segment there is a glottal stop, and the remaining part is pronounced voicelessly (whispered). The distinctive nature of the tones is shown by the following near-homonyms, distinguished only by tone (for details on the use of the tone marks cf. 1.5.5.4): lùoks luõks luôgs

‘bow, arch’ ‘onion leaf ’ ‘window’

(falling tone) (drawn tone) (broken tone)

In unstressed position (if the syllable bears neither main nor secondary stress), the opposition between drawn and falling tone is neutralized; only the glottal stop remains: labãs laba¯s

[laba.s] [laba‘as]

‘good’ (nom/acc.pl.fem.def) ‘good’ (loc.pl.fem.indef)

The non-glottalized realization of unstressed long vowels and diphthongs is usually marked with the sign of the drawn tone, though it is more similar to the falling one.

The Latvian language and its dialects

Of course, syllables bearing secondary stress may also display tone oppositions, cf. ap-sèja ‘sow’ (pret.3) ap-s«e ja ‘bind up, bandage’ (pret.3)

However, many speakers of Latvian neutralize the opposition here as well. This means that in unstressed syllables (with the possible exception of syllables bearing secondary stress) tonal oppositions proper are neutralized. There may be an opposition between ‘glottalized’ and ‘non-glottalized’, but this is not, strictly speaking, a tonal feature, though under stress it occurs as a concomitant of tone. In Latvian linguistics, glottalization is considered to be a tonal feature, which means that tone is regarded as independent of word stress. This is reflected in the fact that tone marks are put over all syllables containing long vocalic segments. The system with three tones is variously reduced in most Latvian dialects. In most of the Central dialect, the falling and broken tones have merged into a single, broken tone. In principle, the three-tone system is recommended for the standard language, but the distinction of three tones is hard to elicit from speakers distinguishing only two in their native dialect. For most speakers of the standard language, the opposition should be thus formulated as ‘drawn vs. non-drawn’, further differentiation (with an opposition between ‘falling’ and ‘broken’ instead of one single ‘non-drawn’ tone) being observed only in the pronunciation of those speakers who distinguish three tones in their native dialect. In High Latvian (Letgalian), the falling and drawn tones have merged, which results in an opposition of ‘falling vs. broken’. As the difference is one of glottalization, tone has actually become neutralized in this dialect. There is only an opposition between presence or absence of stød (a situation comparable to that of Danish but not the remaining Scandinavian languages). 1.5.5.4 The use of tone marks. The tones are not marked in the standard orthography, only in linguistic publications. The use of tone marks is as follows: –





The falling tone is marked with ·`Ò; in the case of diphthongs and diphthongoid sequences this mark is put over the letter denoting the first component: kàsis ‘hook’, màlks ‘draught’, bràukt ‘drive’ The drawn tone is marked with ·«Ò; in the case of diphthongs and diphthongoid sequences this mark is put over the letter denoting the second component: mãte ‘mother’, mañta ‘thing, possession’, laı˜ks ‘time’. The broken tone is marked ·ˆÒ; this mark is put over the second component of a diphthong: vâks ‘lid’, da¥rbs ‘work’, laûzt ‘break’.

Under certain conditions, which vary from one dialect group to another, in original diphthongoid sequences of the type /V + r/ (/ar/, /er/, /ir/, /ur/), the first component, which was originally always short, may be lengthened or diphthongized.

15

16

Laimute Balode and Axel Holvoet

Tonal differences are then realized within the vowel (the sonorant ceasing to belong to the vocalic segment which can carry tone), and all three tone marks are put over the first letter, e.g. kãrpa ‘wart’, dârbs2 ‘work’ (in the Curonian dialects, compared to da¥rbs in the standard language. The significance of the index ·2Ò is explained in the next paragraph). If, in such a case, the tone is falling, a macron is added to the tone mark: v×ards ‘word’ (unlike kàrte ‘map’, a recent loan where the vowel did not undergo lengthening). The three-tone system is the original one. The systems with two tones result from the reduction of this system, and this reduction has completely obscured the original distribution of tones. The historical equivalence between the Latvian and Lithuanian tones can therefore be established only on the basis of the three-tone system. In historical-comparative works on Baltic languages, Latvian words are, in principle, cited with the tone attested in the dialects distinguishing three tones. In citing words and forms from dialects with reduced tone systems, an index ·2Ò is added to indicate that the tone marked on the word in question is the regular outcome of neutralization, and therefore provides no clue as to the original tone of the word, e.g. daîkts2 ‘thing’. 1.5.6Note on Latvian orthography Though Latvian has only one official orthography, a special variety of this is used in linguistic (especially comparative and historical) literature on Latvian and the Baltic languages in general, in etymological dictionaries etc. It differs from the standard orthography on three points: –

– –

2.

instead of the macron (length mark), three different tone marks are used, according to the tone of the syllable (see 1.5.5.4 above), and tone marks are added on diphthongs where these are absent from the standard orthography, e.g. rãdı¯t vs. ra¯dı¯t, naûda vs. nauda; where the letter ·oÒ of the standard orthography denotes the diphthong /uo/, ·uoÒ is written, e.g. rùoka vs. roka; where the letters ·eÒ, ·e¯Ò of the standard orthography denote [æ], [æ˜], the signs ·eÒ, í ·e¯í Ò are used, e.g. víesíels vs. vesels.

The Latvian dialects

2.1 Classification Since August Bielenstein (Bielenstein 1863/64), the Latvian dialects have been classified into three groups: –

the central dialects (Latv. vidus dialekts, Germ. der mittlere Dialekt) are spoken in most of Vidzeme (except for a transitional zone bordering on Letgalia in the east)

The Latvian language and its dialects





as well as in Semigalia and Courland proper (with the exception of its northern part); they have formed the basis of the standard language since the earliest writings (in the 16th century). the Tamian dialects (Latv. ta¯mnieku or lı¯biskais dialekts, Germ. tahmisch or livisch, also nordwestkurisch) are spoken on both sides of the Bay of Riga, in the north of Courland as well as in the north-west of Vidzeme. The term Tamian is based on the nickname ta¯mnieki, by which the speakers of this dialect were referred to, and which seems to be derived from the adverb ta¯m ‘until now’ (standard Latvian lı¯dz šim). This adverb, now fallen into disuse, seems to have been characteristic of the dialects of Courland as a whole. The term lı¯biskais dialekts (‘Livonian dialects’), used in Latvian terminology, refers to the Livonians, the Finnic tribe originally inhabiting the Tamian area. Strictly, the term Tamian refers only to the dialects spoken in the north of Courland, but it is also used to denote the whole dialect group, and this usage will be followed here. the High Latvian Dialect (Latv. augšzemnieku dialekts, Germ. hochlettisch) is spoken in Letgalia, the easternmost part of Vidzeme bordering on Letgalia, and in Selonia, to the south of the Daugava. This dialect is the basis of a regional written standard, sometimes referred to as the ‘Letgalian language’.

This classification is still used without any modifications. However, the Central Dialect and Tamian are sometimes grouped together and referred to by the term Low Latvian, as opposed to High Latvian. Some linguists of Letgalian extraction regard this as the only major line of division among Latvian dialects, arguing that the features commonly accepted as defining characteristics of Tamian are, in fact, but a more consistent realization of features shared with the central dialect. Though this may be conceded, it must also be added that the defining features of High Latvian, though clearly opposing it to Low Latvian, are probably of relatively recent date (caused presumably by a Slavonic adstratum), so that no strong case can be made for the dichotomous classification. 2.2 The main differentiating features of the Latvian dialects The main criteria used in classifying the Latvian dialects are: – – –

differences with regard to prosodic features (tones); qualitative and quantitative changes in the vowel system; morphological innovations.

The distinguishing features of the three main dialect groups can best be shown by contrasting the Tamian and High Latvian dialects with the central dialects, on which the standard language is based. We should mention that the central dialect group is generally phonetically more conservative than the other dialects (a fact which may be connected with the

17

18

Laimute Balode and Axel Holvoet

less pervasive influence of substratum or adstratum languages). In morphology, however, at least part of this dialect has carried through a number of innovations (see 3.1.2.1–3, below, on the formation of the preterite, the future and reflexive verbs). In the domain of prosody, the conservative character of the central dialect manifests itself in the retention, in part of its area, of the original system of three tones. This can, admittedly, only be observed in a few places (in a compact area around Valmiera and Ce¯sis in Vidzeme, and in the civil parishes of Blı¯diene and Jaunpils in Semigalia), but Tamian and High Latvian have only reduced tone systems. The vowel system of the central dialect is both qualitatively and quantitatively much closer to that of Common Baltic than those of the other dialects. The vowels have undergone relatively few qualitative changes, which distinguishes the central dialect from High Latvian (with its ‘velar vowel shift’); and, compared with the Proto-Latvian state of affairs, there has been no further loss of final short vowels, and shortening (loss) of final long vowels, as observed in Tamian. The following table gives typical representations of five words in the three dialects as well as in the literary language: SL

CD

TD

HLD

meaning

mãte

mãte [ma˜te]

mãt [ma˜t]

mòte2/mùot’e2 [mf˜tjæ][mu%tjæ]

‘mother’ (nom.sg)

lu˜pa

lu˜pa [lu˜pa]

lu˜p [lu˜p]

lòupa2/lèupa2/ly`upa2 ‘lip’ (nom.sg) [loupa] [leupa] [Gupa]

labı¯ba

labı¯ba [labi˜ba]

labib/labeb labı¯ba/labeı¯ba [labib][labeb] [labji˜ba][labjeiba]

‘cereals’ (nom.sg)

situ

situ [sit˜u]

sit [sit˜]

sytu/sydl’u/s’ižn’u [sGtu][sGdlju][sji8nju]

‘hit’ (pres.1sg)

gaîlis

gaîlis [gailis]

gaîls [gails]

gaîl’s’/geîl’s’ [gailjsj][gjeiljsj]

‘cock’ (nom.sg)

As can be seen, High Latvian displays mainly qualitative changes: diphthongization of /i˜/ and /u˜/, vowel shift /a/ > /o/. The Tamian dialect shows mainly quantitative changes: reduction of long unstressed vowels to short vowels, and of unstressed short vowels to zero. The peculiarities of the three main dialects are once more illustrated in the following sentence (from a folk song) in its three varieties: Central dialect: ganuôs gãju, kre˛klu šuvu, pìe uôzuola m˜eí ru n¸êmu [Áganu%s «ga˜ju Ð Ákræklu Áwuvu Ð Ápi6 ^u%zu%la «mæ˜ru ^\e˜mu]

The Latvian language and its dialects

Tamian dialect: gane˛s gãj, kre˛kal šûj, pe uôzal2 m˜eí r j˜eí m [Áganæs «ga˜j Ð ‘krækal wu˜j Ð pë ^u%zal «mæ˜r «je˜m] High Latvian dialect: gonûs gùoju2, kraklu šyvu, p’i ûzula màru2 jiêmu [Ágfnu˜s Águ%ju Ð Ákraklu ÁwGvu Ð pji ^ulzula Áma˜ru ^ji7mu] In the standard orthography: ganos ga¯ju, kreklu šuvu, pie ozola herdsman:loc.pl go:pret.1 shirt:acc.sg sew:pret.1 by oak:gen.sg me¯ru n¸¯emu measure:acc.sg take:pret.1 ‘I went a-pasturing, I sewed a shirt, I took the measure by an oak’

Commentaries: The Central dialect is here represented by a variety distinguishing three tones. The phonetic shape of this fragment corresponds exactly to the standard language. The prothetic /T/ in n¸emt ‘take’ is idiosyncratic. The Tamian equivalent shows shortening and concomitant monophthongization of unstressed diphthongs (/uo/ > /a/ or /æ/, /ie/ > /e/ etc.) as well as the consistent loss of final short vowels, with epenthesis of /a/ where this would lead to a word-final sequence ‘obstruent + sonorant’ (/krækal/ < /krækÏ/ < /kræklu/), and with compensatory lengthening (šu¯j). High Latvian shows the High Latvian vowel shift (/a/ > /o/, /a˜/ > /uo/, /æ/ > /a/ and /æ˜/ > /a˜/) and monophthongization of diphthongs (/uo/ > /u˜/). The High Latvian palatalization is marked on p’i. In the Tamian version, the index ·2Ò warns that the broken tone may represent an original broken or falling tone: in the High Latvian version it warns that the falling tone may represent an original falling or drawn tone. Until now only vocalic features have been mentioned. If we take into account the consonant system as well, we will see that High Latvian is more distinctive here. 1. High Latvian has a full-fledged correlation of palatalization, as a result of assimilatory palatalization before front vowels (which subsequently became distinctive as a result of vowel loss, e.g., word-finally). This feature is shared by Lithuanian, but distinguishes High Latvian from the central dialects, which have no palatalization as a secondary articulation. Tamian also has a certain degree of assimilatory palatalization, but only stem-finally, probably a result of Finnic influence, as will be shown below. The central dialects have no palatalization as a secondary articulation, but they have a separate palatal series comprising the nasal /\/, the lateral /ˆ/ and the stops /c/ and /K/, as seen in the table of consonants. This is in accordance with Trubetzkoy’s observation: if a language has a correlation of palatalization, it does not have a palatal series, because palatals are identified as palatalized dentals or alveolars. Where Low Latvian has /c/, /K/, High Latvian

19

20

Laimute Balode and Axel Holvoet

substitutes /tëw/, /dëŠ/ (see 5.1.8 below) and Letgalians speaking standard Latvian substitute palatalized velars /kj /, /gj/. Widespread assimilatory palatalization seems to be associated with a Slavonic adstratum. Although most Balticists ascribe this feature to Common Baltic (or Balto-Slavonic) because it occurs in Lithuanian, High Latvian palatalization cannot have been inherited from Common Baltic, because: a.

if Latvian had once been characterized by generalized assimilatory softening and subsequently lost it, it would have lost its palatal series as well, according to Trubetzkoy’s observation, and b. wherever Letgalian has the vowel shift /æ/ > /a/ (cf. HL vacs ‘old’, standard Latvian vecs [vætës.], Lithuanian (obs., reg.) v˜etušas [«vjæ˜tuwas]), the preceding consonant is not palatalized, which means that palatalization in High Latvian cannot be older than the High Latvian vowel shift. 2. High Latvian devoices obstruents word-finally, like Lithuanian and most of Slavonic, but unlike Low Latvian, e.g. vajag ‘it is necessary’ is pronounced /vajak/ in High Latvian, while final voicing is retained in Low Latvian and the standard language. Slavonic influence seems to be involved here as well. 3. Low Latvian, unlike High Latvian, is characterized by non-distinctive lengthening of unvoiced obstruents. This occurs between short vowels, especially in disyllables, where the final vowel can be unvoiced (whispered): lapa visi

[lap˜a] or [lap˜a] [vis˜i] or [vis˜i]

It does not occur when either of the vowels is long: saka sa¯ka saka¯m sa¯ka¯m

[sak˜a] [sa˜ka] [saka˜m] [sa˜ka˜m]

‘say (pres.3)’, also [sak˜a] ‘begin (pret.3)’, also [sa˜ka] ‘say (pres.1pl)’ ‘begin (pret.1pl)’

There is a potential opposition between geminated and ungeminated obstruents on the morpheme border, as in ata¯ls ‘aftermath’ and atta¯ls ‘distant’, but usually a short vowel is heard in both cases unless a pause is inserted between the final [t] of at- and the initial [t] of -ta¯ls. Though non-distinctive, this lengthening is part of the standard pronunciation and causes notorious difficulties to speakers of High Latvian, who have only geminates in their dialects. In this respect, High Latvian resembles Lithuanian (which, however, goes further and usually even shortens geminates) and such Slavonic languages as Polish and Russian, which have only geminates, but no phonetic lengthening of obstruents. The difference between Low and High Latvian may be connected with differences in syllable structure, but this still has to be investigated in detail. Whereas the state of affairs observed in High Latvian is

The Latvian language and its dialects

reminiscent of Slavonic, the Low Latvian situation reminds us of Finnic, where a correlation of quantity in obstruents is a common phenomenon. More specifically, we are reminded of the lengthening of voiceless obstruents after short vowels in Livonian, as described by Kettunen (1938). In Livonian, this lengthening is not dependent on grade alternation conditioned by the opposition of original open and closed syllables, but by the reduction (shortening and contraction) of final syllables.

3.

The central dialects (vidus dialekts)

3.1 Common features 3.1.1 As mentioned above, the central dialects are the phonetically most conservative, so that their distinguishing features, as compared to Tamian and High Latvian, can be formulated negatively (as the lack of some phonetic change or other). The basic features of the phonemic and prosodic system coincide with those described above for the standard language. More specifically, we may note the retention of short vowels in final syllables. True, these are often unvoiced (whispered) or reduced to zero, but they may always reappear in more careful speech. In some types of polysyllabic words, however, complete vowel loss is observed in the central dialects as well, e.g. in nouns ending in -ı¯tis, -ietis: SL

CD

meaning

bra¯lı¯tis latvietis

bra¯lı¯ts latviets

‘brother’ (dim.nom.sg) ‘Latvian’ (nom.sg)

3.1.2 In morphology, on the other hand, at least some of the central dialects have carried through a number of innovations, which are also characteristic of the standard language. They distinguish most of the middle dialect, and the standard language, from High Latvian on the one hand, but also from the Curonian dialects of the central group. Of course, this does not point to a closer connection between Curonian and High Latvian; it is simply an instance of a peripheral archaism. The innovations are the following. 3.1.2.1 Common Baltic had two kinds of past tense stems, in -a¯- and in -e¯-. Lithuanian has retained both, but the central dialects of Latvian, and the standard language, have generalized the preterite in -a¯-. The changes undergone by stemfinal velar consonants in the preterite show that preterites in -e¯- must once have existed everywhere in Latvian. If standard Latvian has a preterite (3rd person) na¯ca

21

22

Laimute Balode and Axel Holvoet

from the verb na¯kt ‘come’, then this points to an original preterite na¯ce, because it is only before a front vowel that /k/ is changed into /tës/. High Latvian has retained the -e¯-preterites, and so have some of the Curonian dialects; e.g. CD (Livonian) and SL

CD (Curonian)

LITH.

meaning

augu auga¯m vedu veda¯m

augu auga¯m vežu vede¯m

áugau áugome vedžiau˜ v˜ede˙me

‘grow’ (pret.1sg) ‘grow’ (pret.1.pl) ‘lead’ (pret.1sg) ‘lead’ (pret.1pl)

3.1.2.2 The Baltic future is formed with the suffix -si-, added to the infinitive stem. If the infinitive stem ends in a dental or alveolar consonant, consonant clusters arise which are normally simplified, rendering the forms morphologically opaque. In some of the Livonian dialects this is avoided by inserting /i˜/ (borrowed from a verb class with an infinitive stem in /i˜/). Alongside daru ‘I do’ there is a future darı¯šu ‘I will do’, based on the infinitive darı¯t, and this gave rise to an analogical relationship metu : metı¯šu. High Latvian and some of the Curonian dialects have retained the original forms without insertion: CD (Livonian) and SL

CD (Curonian)

LITH

meaning

lauzı¯šu vedı¯šu

lauššu veššu

laušiu vesiu

‘break’ (fut.1sg) ‘lead’ (fut.1sg)

3.1.2.3 In the Baltic languages the reflexive pronoun has come to be integrated into the verb form. Basically, it is now affixed to it (as is the case in east Slavonic). If the verb contains a prefix, however, the reflexive marker is inserted between this prefix and the lexical verb stem, which is a relic of the original mobility of the reflexive pronoun. Lithuanian still shows this state of affairs: baigend baigend papref papref

-e˙pret -e˙pret baigend sirefl

-Ø 3 -Ø 3 -e˙pret baigend

-si refl -Ø 3 -e˙pret

-Ø 3

The majority of the central dialects of Latvian (and the standard language) have generalized a single pattern, with the reflexive marker affixed at the end, after the personal endings; but the pattern with infixed reflexive marker can be shown to

The Latvian language and its dialects

have been general in Latvian as well. The Curonian dialects have, to a certain extent, retained the original state of affairs (as have the High Latvian dialects). Often, however, the resulting verb forms have double reflexive markers, one inserted, the other affixed word-finally. The verb form in the example means ‘ends’ (intransitive, 3rd person present): CD (Liv.)/SL nopref

beidzend

-aTHEM

-s refl

CD (Cur)

-sarefl

beidzend

-aTHEM

nopref

-s refl

3.2 Subdivision of the central dialect group The central dialect group is subdivided into three subgroups: – – –

Livonian dialects (Latv. Vidzemes vidus izloksnes) Semigalian dialects (Latv. Zemgales izloksnes) Curonian dialects (Latv. kursiska¯s izloksnes)

Among these, the Curonian dialects are most specific, whereas the differences between the Livonian and Semigalian dialects are relatively slight. 3.2.1The Curonian dialects (kursiska¯s izloksnes) 3.2.1.1 The most characteristic feature of the Curonian dialects is the lengthening and/or diphthongization of the vowel in original diphthongoid sequences of the type /V + r/. Under falling and drawn tone all Latvian dialects lengthened the vowel if it was a non-high vowel (a process mentioned above, 1.5.5.4). In the Curonian dialects, this process extends to the high vowels /i/ and /u/, which, according to the dialect, are either lengthened into /i˜/ and /u˜/ or diphthongized into /ie/ and /uo/. Moreover, this lengthening occurs under all three (original) tones: SL

CD (Livonian)

CD (Curonian)

meaning

mirt zirgs burt

mi«r t zirks bu«r t

mı˜rt zîrks2/ziêrks2 bu˜rt

‘die’ (inf) ‘horse’ (nom.sg) ‘practice sorcery’ (inf)

When the first component of the original diphthongoid sequence was a non-high vowel, lengthening is general in the Curonian dialects, whereas elsewhere it did not operate under broken tone:

23

24

Laimute Balode and Axel Holvoet

SL

CD (Livonian)

CD (Curonian)

meaning

darbs dzert

darps dzert

dârps2 dzêí rt2

‘work’ (nom.sg) ‘drink’ (inf)

3.2.1.2 An archaism of the Curonian dialects is the retention of /u/ before /v/ and /b/; in the other dialects this vowel has undergone loss of lip rounding as a result of dissimilation, yielding /i/: SL

CD (Livonian)

CD (Curonian)

meaning

zivs dibens

zivs /zius/ dibíens

zuve dubíenc

‘fish’ (nom.sg) ‘bottom’ (nom.sg)

The archaic character of the forms with /u/ is shown by Lith. žuvìs, dùgnas (< *dubnas). 3.2.1.3 Another archaism, not, in this case, a regular phonetic peculiarity, but a substratum feature, is the relatively high incidence of forms with retention of the Common Baltic diphthongoid sequences of the type /V + N/. In Latvian, these were, in principle, originally changed into nasal vowels which subsequently lost their nasal resonance to yield long oral vowels or diphthongs. The now extinct Curonian language retained these sequences in their original shape. SL

CD (Livonian)

CD (Curonian)

meaning

bezdelı¯ga

bezdelı¯ga

bezdelinga

‘swallow’ (nom.sg)

A number of forms with retention of the sequence /V + N/, where loss of the nasal would be expected, are found everywhere in Latvian; these probably result from dialect borrowing (expansion of Curonian forms), e.g. menca, mence ‘cod’, corresponding to Lith. ménke˙; the regular treatment in Latvian would have been *miece. 3.2.1.4 In the Curonian dialects, /v/ is lost after /l/, with possible compensatory lengthening of the preceding vowel:

SL

CD (Livonian)

CD (Curonian)

meaning

cilve¯í ks baltgalvis

cilve¯í ks baltgalvis

cile¯í ks ba˜lgâlis2

‘man, human’ (nom.sg) ‘white-headed person’ (nom.sg)

The Latvian language and its dialects

3.2.1.5 In morphology, the Curonian dialects are characterized by a number of archaisms as compared to the innovating Livonian dialects. They have already been mentioned above, 3.1.2.1–3. 3.2.2The Livonian dialects (Vidzemes vidus izloksnes) The most characteristic feature of these dialects, occurring nowhere else, is the extension of /i˜/-insertion in the future tense (cf. 3.1.2.2) beyond the group of verbs with infinitive stems in dental consonants. Here we find /i˜/ generalized for all 1st conjugation verbs, i.e. verbs with monosyllabic stems in infinitive, present and preterite. Hence we have: SL

CD (Livonian)

meaning

na¯kt, na¯kšu liet, liešu be¯rt, be¯ršu

nãcı¯šu le¯jı¯šu b˜erı¯šu

‘come’ (fut.1sg) ‘pour’ (fut.1sg) ‘pour (sand, grain etc.)’ (fut.1sg)

3.2.3The Semigalian dialects (Zemgales izloksnes) Semigalian is characterized by a tendency toward vowel insertion between /r/ (and, less frequently, /l/) and a following consonant, a feature found nowhere in the other dialect groups. The vowel preceding /r/ is short, whereas in other dialects it is lengthened at least with some tones. It is evidently the absence of lengthening that has conditioned the vowel insertion lacking in other dialects. This feature, which is not found outside Semigalian, is rapidly receding and is now observed in aged dialect speakers only. SL

CD (Semigalian)

meaning

va¯rna kurpe bí¯erzi [bæ˜rzi] ve¯rsis malka svilpt

varana kurape bíerezi [bæræzi] víeresis malaka svilapt

‘crow’ (nom.sg) ‘shoe’ (nom.sg) ‘birch’ (nom.pl) ‘bull’ (nom.sg) ‘firewood’ (nom.sg) ‘whistle’ (inf)

4. The Tamian dialect group (lı¯biskais/ta¯mnieku dialekts) 4.1 Common features Tamian is divided into two groups: the dialects of northern Courland (Kurzemes lı¯biskais dialekts), which, as mentioned above, are the Tamian dialects proper, and those of north-western Livonia (Vidzemes lı¯biskais dialekts). Within the western

25

26

Laimute Balode and Axel Holvoet

(Courlandish) group, those spoken farthest to the north display the Tamian features in the most pronounced way; they are sometimes referred to in German terminology as strengtahmisch. 4.1.1 Tamian as a whole agrees with the majority of central Latvian dialects in having merged the falling and broken tones; both are represented by the broken tone, so that (2) and (3) are homonymous: (1) luõks

‘onion leaf ’ (nom.sg) 2

‘bow, arch’ (nom.sg)

2

‘window’ (nom.sg)

(2) luôks (3) luôgs

4.1.2 The most characteristic feature of Tamian is the reduction (loss or shortening) of unstressed vowels. Short vowels in word-final position are consistently lost. If the preceding syllable originally contained a short vowel followed by a voiced consonant, this vowel undergoes compensatory lengthening: SL

Tamian

meaning

labi kaza

la¯b ka¯z

‘well’ (adv) ‘goat’ (nom.sg)

If the consonant preceding the apocopated final vowel is a voiceless obstruent, a short vowel in the preceding syllable is not lengthened. In such cases, the obstruent is lengthened, as is generally the case between short vowels. The retention of long voiceless obstruents in word-final position is a peculiarity of Tamian and obviously accounts for the lack of compensatory lengthening of the vowel. SL saka lasa upe

Tamian [sak˜a] [las˜a] [up˜e]

sak¯ la¯s u¯p

meaning [sak˜] [las˜] [up˜]

‘says, say’ (pres.3) ‘reads, read’(pres.3) ‘river’ (nom.sg)

A tendency towards loss of final short vowels can be observed in the central dialect as well, and the extent of this phenomenon is gradually increasing. In Tamian this process has been carried through consistently. Instead of occasional reduction of final vowels (which may either be dropped or be pronounced voicelessly), we observe their complete loss in Tamian, and this dialect has also developed compensatory mechanisms restricting the impact of vowel loss on the prosodic structure of the word.

The Latvian language and its dialects

The widespread loss of unstressed vowels seems to be connected with a similar development in Livonian, so that it could be considered a substratum feature. Livonian shows widespread loss of internal and final short vowels when compared to Estonian and other Baltic Finnic dialects, cf. Livonian

Estonian

Proto-Finnic

meaning

lapst tu’lb6D

lapsid tulevad

*lapset *tulepat

‘children’ ‘they come’

4.1.3 Unstressed long vowels and diphthongs, i.e. those occurring in non-initial syllables, are consistently shortened: SL

TD

meaning

runa¯t ža¯ve¯t lası¯s telí¯ens

runat žâvet2 lasis telens

‘speak’ (inf) ‘dry (trans.)’ (inf) ‘read’ (fut.3) ‘calf ’ (dim.nom.sg)

Further reduction may lead to complete loss of unstressed long vowels and diphthongs: runa¯t uozuolin¸š

ru¯nt uozliš

‘speak’ (inf) ‘oak’ (dim.nom.sg)

The fact that even long unstressed vowels undergo shortening and may be completely lost in Tamian (a feature unknown to other dialects) may be connected with the Livonian substratum as well. Endzelin connected it with a feature of the Baltic Finnic prosodic system, which originally distinguished quantity in root syllables only, long vowels in non-root syllables being of recent date. 4.1.4 The diphthong /au/ appears as /åu/ or /ou/: saukt kauns

såûkt2/soûkt2 kåûns2

The vowel /a/ appears as /o/ before /v/:

‘call’ (inf) ‘shame’ (nom.sg)

27

28

Laimute Balode and Axel Holvoet

sava¯da¯ks

sova¯daks

‘different’ (nom.sg)

Tamian /ou/ < /au/ and /ov/ < /av/ might be of Livonian origin, cf. Livonian

Estonian

meaning

loùl ovàr

laul avar

‘song’ ‘open, spacious’

However, Livonian influence is not certain here, because this feature has been observed (less consistently) in other dialects as well. 4.1.5 In some Tamian dialects, /e˜/ develops into /ei/, especially before obstruents and with broken tone, e.g. SL

TD

meaning

¯ešana se¯d

eîšan2 seîd2

‘eating’ (nom.sg) ‘sit’ (pres.3)

4.1.6 In some Tamian dialects, dental and alveolar consonants are palatalized before front vowels stem-finally, i.e. between stem and affix, and this feature also occurs where an original front vowel has been lost. SL

TD

meaning

atna¯cis sacı¯t

atnãc’s [Áatna˜ tësj˜] sac’t [satësj˜t]

‘come’ (part.pret.act.masc.sg.nom) ‘say’ (inf)

This type of palatalization is also attested in Livonian and Estonian, where it is associated with the (original) presence of /i/. The fact that it does not occur steminitially and is restricted to dentals and alveolars distinguishes it from the High Latvian palatalization, which extends to all consonants and is independent of their position in the word. This points to the conclusion that Tamian palatalization is not historically connected with High Latvian palatalization, but rather reflects the influence of the Livonian substratum. 4.1.7 In the domain of morphology, the most striking feature of Tamian is the tendency towards reduction of the gender system. It is generally agreed upon that this must

The Latvian language and its dialects

be due to the influence of the Livonian substratum, as the category of grammatical gender is unknown to Finnic. The extent of the substitution of masculine forms for feminine ones varies from one place to another. The most common manifestation of this tendency is the use of masculine forms of the personal pronoun (vin¸ˇc ‘he’) to refer to females. The next stage of this substitution (attested in a smaller area) is the use of masculine forms of adjectives and pronouns to agree with feminine nouns, which themselves retain their characteristic feminine endings; in the case of adjectives, this occurs both in attributive and in predicative use. SL ta¯s se¯n -es these:nom.pl.fem mushroom nom.pl TD tie se¯n -s these:nom.pl.masc mushroom nom.pl ‘these mushrooms’ SL zeme ir slapj- -a earth (fem) is wet nom.sg.fem TD ze¯m i(r) slap- -š earth is wet nom.sg.masc

Finally, feminine nouns themselves may adopt endings characteristic of masculine nouns. This amounts to the virtual elimination not only of gender as a morphosyntactic category, but also of the very forms historically connected with the feminine gender. This third stage has been reached in a still smaller area than the second. SL maz-a siev- -in¸- -a little-fem wife dim nom.sg.fem TD mas (< maz-s) siev- -i- (< -in¸-) -š little-masc wife dim nom.sg.masc

4.1.8 Apart from morphosyntactically determined changes, phonetic processes have also had their impact on the morphology of the Tamian dialects. In verbal inflection, the loss of conjugational endings as a result of the loss of final vowels has led to widespread homonymy of personal forms, at least in the singular. This pattern has been extended to the plural, which results in the generalized use of the original 3rd person form for all persons. In the following table, the conjugation of iet ‘go’ is shown for the standard language and for Tamian. Note that the homonymy of the 3rd person singular and plural in all finite verb forms (the common form continues the original singular form) is an (as yet unexplained) common feature of the Baltic languages. This homonymy may have facilitated the spread of the endingless singular form into the plural.

29

30

Laimute Balode and Axel Holvoet

es tu vin¸š me¯s ju¯s vin¸i

SL

TD

ej-u ej iet ej-am ej-at iet

es tu vin¸ˇc (viš) me¯s (míes) ju¯s vîn¸2

¸ Ô Ô ˝ iet Ô Ô ˛

4.1.9 Tamian uses the accusative singular with prepositions governing the genitive singular in other dialects. TD Es bi pe bra¯l (< bra¯li) I be:pret1 at brother:acc SL Es biju pie bra¯¸la I be:pret1 at brother:gen ‘I was at my brother’s place’

As mentioned in 1.1.3, all dialects have generalized the dative plural with all or nearly all prepositions. This pattern of government is based on that of the prepositions governing a dative plural (originally an instrumental plural) as opposed to an accusative singular (which could be an original accusative or an original instrumental). The process was triggered by the coalescence of the instrumental with the accusative in the singular and the dative in the plural. Tamian seems to have carried this process to its logical conclusion by generalizing the accusative in the singular. 4.1.10 Among the syntactic properties of Tamian, one should mention the widespread use of the dative, which has partly ousted the genitive even in adnominal position, as in Dêls í valkeí te˜í vam cepur í son:nom.sg wear:pres.3 father:dat.sg cap:acc.sg ‘The son wears his father’s cap’

as compared to standard Latvian De¯ls valka¯ te¯va cepuri. son:nom.sg wear:pres.3 father:gen.sg cap:acc.sg

This is usually described as a consequence of the loss of final vowels: tí˜ev’ can be gen.sg, acc.sg or nom.pl, whereas dat.sg tí˜evam has a clearly identifiable ending.

The Latvian language and its dialects

4.1.11 The partitive genitive with quantifiers is also virtually extinct in Tamian. Thus, with a quantifier like daudz ‘much, many’, one finds the nominative (in subject function) or the accusative (in object function), e.g. SL ju¯ra¯ daudz laivu sea:loc.sg many boat:gen.pl TD ju¯reí doudz laivs sea:loc.sg many boat:nom.pl ‘There are many boats at sea.’

The genitive of negation, used as an object case with transitive verbs and as a subject case with intransitive verbs, is rare as well. In negated existential constructions, for instance, the nominative is normally used, not the genitive as would be the case in the standard language and in the central and High Latvian dialects: ju¯rmaleí nau ve¯š beach:loc be:pres.3sg.neg wind:nom.sg ‘There is no wind on the beach.’

The partitive genitive with quantifiers and the genitive of negation are much better retained in the central dialects, and their use is most consistent in High Latvian, cf. 5.1.12. 4.2 Subdivision of the Tamian dialects This dialect group can be subdivided into Tamian proper, i.e. the dialects of northern Courland, and the related dialects in the north-west of Vidzeme. It goes without saying that the former are more closely related to the Curonian subgroup of the central dialects, and the latter to the Livonian subgroup. The influence of the Finnic (Livonian) substratum language is common to both, and may be viewed as the defining feature of the Tamian dialect group as such. The following features are peculiar to Tamian proper: 4.2.1 Vowel reduction is carried further, affecting to a larger extent not only word-final, but also word-internal syllables. Short vowels may be completely lost not only in inflectional endings, but also in derivational suffixes (if they bear no secondary stress). This rule extends to short internal vowels resulting from the shortening of original long vowels and diphthongs (and still appearing as such in the Tamian dialects of Vidzeme):

31

32

Laimute Balode and Axel Holvoet

SL

TD (Vidzeme)

TD (Courland)

meaning

sacı¯t kumelin¸š

sacit [satësit] kumelin¸ˇc

sac’t [satësjt] kûmliš

‘say’ (inf) ‘horse (poet.)’ (dim.nom.sg)

4.2.2 Some Tamian dialects replace word-final unvoiced stops (fortes) with unvoiced lenes: SL

TD

meaning

iet

ied®

‘go’ (pres.3)

This seems to be due to Livonian influence, as the Finnic languages lack voiced stops (the Fenno-Ugricists’ transcription would be ieD). 4.2.3 A few Tamian dialects of Courland have retained the archaic dative plural endings of the type -Vms, replaced by -Vm in all other dialects (in the examples below it should be noted that, phonetically, the standard Latvian forms are more archaic in that they retain the original long vowels in the endings): SL

TD

meaning

sieva¯m ma¯te¯m

sievams ma¯tíems

‘wife’ (dat.pl) ‘mother’ (dat.pl)

4.2.4 The Tamian dialects of Courland display a number of features that are not peculiar to Tamian as such, but are shared with the Curonian dialects of the central dialect group to which it is contiguous, e.g., lengthening or diphthongization of /ir/, /ur/ into /i˜r/, /ier/, /u˜r/, /uor/ etc., cf. 3.2.1.1.

The Latvian language and its dialects

5.

The High Latvian dialects (augšzemnieku dialekts)

5.1 Common features 5.1.1 The opposition between falling and drawn tone is neutralized, cf. 1.5.5.3. In the examples, (1) and (2) are homonymous: (1) lùoks2/lùks2

‘onion leaf ’ (nom.sg)

2

2

‘bow, arch’ (nom.sg)

2

2

‘window’ (nom.sg)

(2) lùoks /lùks (3) luôgs /lûgs

5.1.2 Long /a˜/ acquires a more closed articulation; according to the dialect, it develops into /o˜/, /uo/, /oa/, /a¯o/, /oa¯/ etc.: SL

HLD

meaning

ma¯sa ma¯ja a¯buols skata¯s

mo¯sa // mùosa2 mòja2 oGbùels2 skotuôs

‘sister’ (nom.sg) ‘house’ nom.sg) ‘apple’ (nom.sg) ‘look’ (pres.3)

5.1.3 Long /e˜/ is changed either into /æ˜/, or into a diphthong /ie/, according to the dialect: SL

HLD

meaning

kre¯jums ¯eí du ve¯jš

krí¯ejums // kriejums êí žu // iêžu vìejs´2 [vjiejsj]

‘cream’ (nom.sg) ‘eat’ (pret.1sg) ‘wind’ (nom.sg)

5.1.4 The whole of High Latvian is characterized by a series of vowel changes conditional upon the ‘velar’ nature of the following syllable. These changes occur when the following syllable contains, or originally contained, a back vowel, as long as no /j/ nor any palatalized consonant arising from a combination with /j/ intervened between the two vowels. The scope of these changes may vary (they may, for instance, be restricted to the ‘deep’ High Latvian dialects), but, considered together, they reflect a common tendency: vowels are shifted to the back if there is no palatalizing (fronting) factor in the following syllable. These changes are sometimes referred to as the High Latvian velar vowel shift. Synchronically, it conditions a series of alternations in inflection and derivation. It should be noted that Latvian,

33

34

Laimute Balode and Axel Holvoet

as a whole, had one process dependent on similar conditions: the split of IE *e and *e¯ into /e(˜)/ and /æ(˜)/ was also connected with the presence or lack of a palatalizing factor in the next syllable. In the High Latvian vowel shift this process has been carried much farther. 5.1.4.1 Where Low Latvian has /æ/ and /æ˜/ (which is the case before original ‘back’ syllables, see 1.5.2), High Latvian has /a/ and /a˜/:

SL vecs de¯ls se¯ta

[vætës˜] [dæ˜ls] [sæ˜ta]

HLD

meaning

vacs d×als2 s×ata2

‘old’ (nom.sg) ‘son’ (nom.sg) ‘fence; farm-yard’ (nom.sg)

5.1.4.2 Original short /a/ shifts to /o/ under the same conditions as mentioned above for the occurrence of /æ/, i.e. when the following syllable contains a back vowel and no /j/ or palatalized consonant intervenes. In final syllables (whether open or closed) /a/ remains unchanged: SL

HLD

meaning

lapa maza

lopa moza

‘leaf ’ (nom.sg) ‘small’ (nom.sg.fem)

In some dialects the shift /a/ > /o/ is restricted to initial syllables: SL

HLD

meaning

vasara

vosora/vosara

‘summer’ (nom.sg)

5.1.4.3 In some of the High Latvian dialects (the so-called ‘deep’ High Latvian dialects) a short /i/ also undergoes changes under the same conditions as described above: it is replaced by the vowel /G/, usually transcribed ·yÒ in dialect recordings. This vowel does not exist elsewhere in Latvian; it is a high centralized vowel similar to Russian or Polish /y/. SL

HLD

meaning

miza tikums

myza tykums

‘bark’ (nom.sg) ‘virtue’ (nom.sg)

The Latvian language and its dialects

The same vowel is also found in some inflectional endings, where it has either taken the place of original /a/ (probably as a reduced vowel) or has been inserted between a stem-final cluster ‘obstruent + sonorant’ and the nom.sg ending -s (as /i/ is inserted in similar circumstances in the central dialects): tas kakls (< *kaklas)

tys kaklys

‘that’ (nom.sg) ‘neck’ (nom.sg)

5.1.5 There is a strong tendency to diphthongize the long closed vowels /i˜/ and /u˜/; /i˜/ becomes /ei/ (the first component may also be more closed), whereas /u˜/ yields either /ou/ or (in deep High Latvian) /yu/, /iu/ or /eu/: SL

HLD

meaning

rı¯ts vı¯rs lu¯pa

reits [rjeitës] ‘morning’ (nom.sg) vîÛi rs // v’èirs2 [vji˜jrs] / [vjeirs] ‘man, husband’ (nom.sg) ‘lip’ (nom.sg) lyupa // lòupa2 // leupa

5.1.6 In the deep High Latvian dialects the rising diphthongs /ie/ and /uo/ are monophthongized into /i˜/, /u˜/: SL ieva roka

[ru%ka]

HLD

meaning

¯ıva ru¯ka

‘bird-cherry’ (nom.sg) ‘hand’ (nom.sg)

A few general remarks may be made here on the changes in the High Latvian vowel system. These changes may be divided into two groups: (1) those dependent on the quality of the next syllable, and (2) those independent of phonetic environment. If we assume an interconnection to have existed between both groups, we could ascribe the changes in group (2) to shifts within the vowel system caused by the conditioned vowel changes in group (1). The whole series of changes might have been triggered by the velar vowel shift, the effect of which amounted to a rearrangement of the system of short vowels. The initial and final stage of this process are shown here as (A) and (B):

35

36

Laimute Balode and Axel Holvoet

(A)

(B)

i

u

i

e

(y) e

æ

a

u o

a

It is interesting to note that by adding /y/, the deep High Latvian dialects have achieved a short vowel system identical with the vowel system of the neighbouring Slavonic languages (Polish, Russian and Belarusian). A new /æ/ (from original /e/) occurs in High Latvian, but it is a variant of /e/ rather than an independent phoneme. The changes among long vowels and diphthongs might have been (at least partly) triggered by the change of /æ˜/, which was ‘swept along’ by its short counterpart /æ/ and passed to /a˜/. In some dialects its place was taken by a new /æ˜/ from /e˜/. Original /a˜/ had to shift towards /o˜/ or /uo/ in order to be kept distinct from the new /a˜/. Its passage to /uo/ must be related to the monophthongization of original /uo/ in some of the High Latvian dialects; another factor contributing to the monophthongization of both /ie/ and /uo/ must have been the loss of original /i˜/ and /u˜/ as a result of diphthongization. 5.1.7 In the consonant system the most striking feature of High Latvian is the (relatively slight) palatalization of consonants before front vowels: SL

HLD

ma¯te sieva ve¯jš lakstı¯gala

mùot’íe2 s’ı¯va v’iejs´ laks’t’eigola

meaning [Ámu%tjæ] [sji˜va] [vjiejsj] [laksjtjeigola]

‘mother’ (nom.sg) ‘wife’ (nom.sg) ‘wind’ (nom.sg) ‘nightingale’ (nom.sg)

5.1.8 As mentioned above, most High Latvian dialects do not have the palatal plosives /c/ and /L/; where these appear (normally as a consequence of interdialectal borrowing), the palato-alveolar affricates /tëw/ and /dëŠ/ are substituted: SL k¸¯eve kak¸a

HLD [ce˜ve] [kaca]

2

ˇcève kacˇa

meaning [tëwe:vje] [katëwa]

‘mare’ (nom.sg) ‘cat’ (gen.sg)

The Latvian language and its dialects

Where Low Latvian has the clusters /wc/ ·šk¸Ò, /ŠL/ ·žg’Ò, High Latvian has /wk/, /Šg/; if a front vowel follows, the velar consonant is, of course, palatalized, but it is still a palatalized velar, not a palatal. SL

HLD

šk¸audı¯t šk¸u¯nis režˇgis

[wcaudilt] [wcu:nis] [reŠLis]

škaudı¯t šky`unc2 režgis

meaning [wkaudi:t] [wkGuntës] [rjeŠjgjis]

‘sneeze’ (inf) ‘barn’ (nom.sg) ‘grating, trellis’ (nom.sg)

5.1.9 Syllable-final /l/ is often velarized into /ł/ unless followed by a palatalized consonant; sometimes this velarized /ł/ is vocalized into /u/: SL

HLD

malka celms

[tësælms]

mòłka2 caÛûms

meaning [tësaums]

firewood (nom.sg) stump (nom.sg)

5.1.10 In the domain of morphology we find the same archaisms already mentioned for the Curonian dialects: – – –

retention of -e¯-stems in the preterite (cf. 3.1.2.1); lack of /i˜/- insertion in the future of verbs with stems in dental consonants (cf. 3.1.2.2); insertion of the reflexive marker (cf. 3.1.2.3).

Furthermore, some High Latvian dialects have preserved ancient locatives in -ie (for -i-stems) and -uo (for -u-stems), as opposed to the innovative forms in -ı¯, -û in the other dialects: SL

HLD

meaning

ausı¯ medu¯

ausie maduo

‘ear’ (loc.sg) ‘honey’ (loc.sg)

5.1.11 High Latvian uses another personal pronoun for the 3rd person than Low Latvian: jis//jys ‘he’ and jì2 // jèi2 ‘she’, cf. Lith. jìs, jì as well as Slavonic (where it is now used in the oblique cases only) Russ. jego, jemu…. Low Latvian has vin¸š, vin¸a instead, which no doubt contains the same pronoun *jis, the initial element *vin- being, however, unclear.

37

38

Laimute Balode and Axel Holvoet

5.1.12 In the domain of morphosyntax we may note the frequent use of the genitive of negation, i.e. the object of a negated verb is in the genitive instead of the accusative. This feature is also characteristic of Slavonic (especially Polish) and Lithuanian, and it is often viewed as a common feature of Baltic and Slavonic; in High Latvian, Slavonic influence has certainly contributed to its retention, whereas it is rare in Low Latvian. SL, CD ne-mazga¯ muti neg-wash:pres.3sg face:acc.sg HLD namåzgà2 muitis neg-wash:pres.3sg face:gen.sg ‘He doesn’t wash his face.’

Apart from the genitive of negation, the partitive genitive is also regularly used in High Latvian, more consistently than in the central dialects and especially in Tamian, where it is virtually extinct, cf. 4.1.11. High Latvian is, generally speaking, more archaic in morphosyntax than the remaining dialects. This can be seen in the government of prepositions, where High Latvian has not consistently carried through the generalization of the dative in the plural (cf. 1.1.3). The genitive plural is often retained with prepositions governing the genitive in the singular, and the deep High Latvian dialects may also retain the accusative plural (on this notion cf. 5.2). Tamian is most innovative in this respect, cf. 4.1.9. 5.1.13 In the lexicon, an interesting feature of High Latvian is the lack of the verb ‘to come’, which Low Latvian distinguishes from ‘to go’, like Germanic and Baltic Finnic (cf. Finnish mennä : tulla, German gehen : kommen), but unlike Slavonic and Lithuanian. Thus High Latvian has atı¯t’ = Lith. ateiti ‘to come’, cf. Russ. prijti, Polish przyjs´´c etc. Low Latvian has iet ‘to go’ and na¯kt ‘to come’, also in compounds. Perhaps Finnic influence may be suspected in Latvian. 5.2 Subdivision of High Latvian In the internal classification of High Latvian dialects, several lines of division are drawn. The western dialects form a kind of transition between the central dialect and eastern High Latvian, which has the most pronounced High Latvian features and is also referred to as ‘deep’ High Latvian (Latv. dzil¸a¯s augšzemnieku izloksnes, Germ. tiefhochlettisch). On the other hand, within both western and eastern High Latvian a distinction is drawn between Selonian dialects (Latv. se¯liska¯s izloksnes, in Upper Courland and south-eastern Vidzeme) and non-Selonian dialects (Latv.



The Latvian language and its dialects

nese¯liska¯s izloksnes, in Latgale and adjacent north-eastern Vidzeme). The intersection of these two lines of division yields four subgroups within High Latvian. 5.2.1 The main distinguishing feature of the Selonian dialects is that instead of the broken tone they have a rising tone which is not known in any other dialect group. It is marked with an acute over the second letter, e.g. SL

HLD (non-Selonian)

HLD (Selonian) meaning

gailis

gail’s’

gaíl’s’

[Ágailjsj]

‘cock’ (nom.sg)

5.2.2 The ‘deep’ High Latvian dialects are distinguished from the remaining ones by the following features, some of which were mentioned above: 5.2.2.1 monophthongization of /ie/, /uo/, cf. 5.1.6. 5.2.2.2 The rise of the vowel /G/, cf. 5.1.4.3. 5.2.2.3 occurrence of /G/ and /i/ instead of /a/ and /e/ in the inflectional endings -as and -es (gen.sg and nom/acc.pl): SL ma¯sas ma¯tes

deep HLD 2

mùosys mùot’is2

meaning [Ámu%sGs] [Ámu%tjis]

‘sister’ (gen.sg, nom/acc.pl) ‘mother’ (gen.sg, nom/acc.pl)

5.2.2.4 in morphology, the diminutive suffix -en¸- instead of -in¸-: SL

deep HLD

rocin¸a

rùc’en’a2

meaning [Áru˜tësjenja]

‘hand’ (dim.nom.sg)

Note * The authors express their gatitute to Oja¯rs Busš (Riga), who kindly commented on the draft version of this paper. Of course, he bears no responsibility for its shortcommings.

39



40

Laimute Balode and Axel Holvoet

References Bielenstein, August. 1863–64. Die lettische Sprache nach ihren Lauten und Formen vergleichend und erklärend dargestellt. Bd. 1–2. Berlin: Ferdinand Dümmler. Endzelin, Janis. 1922. Lettische Grammatik. Riga: Gulbis. Endzelı¯ns, Ja¯nis. 1951. Latviešu valodas gramatika. Rı¯ga: Latvijas Valsts izdevniecı¯ba. (Revised Latvian version of the preceding.) Ga¯ters, Alfreds. 1977. Die lettische Sprache und ihre Dialekte. The Hague-Paris-New York: Mouton. Kettunen, Lauri. 1938. Livisches Wörterbuch mit grammatischer Einleitung. Helsinki: SuomalaisUgrilainen Seura. (Lexica Societatis Fenno-Ugricae, V.) Latviešu valodas dialektu atlants. Leksika. Rı¯ga, 1999. Rudzı¯te, Marta. 1964. Latviešu dialektolog’ija. Rı¯ga. Rudzı¯te, Marta. 1993. Latviešu valodas v˜esturiska¯ fone¯tika. Rı¯ga. Some of the examples from Latvian dialects have been taken from the card files of the Latvian Language Institute, Latvian Academy of Sciences, Riga.



The Lithuanian language and its dialects* Laimute Balode and Axel Holvoet

General data 0.1 Number of speakers Lithuanian is now spoken by about 3,500,000 people in Lithuania and about 600,000 people spread over a number of other countries. Outside Lithuania, autochthonous Lithuanian populations can be found in north-eastern Poland near the Lithuanian border and in a few villages in Belarus. No figures are available for the number of dialect speakers compared to speakers of the standard language, or for the number of speakers of the particular dialects. Before World War II, 69.2% of the population consisted of Lithuanians; the most important minorities were: Poles (15%), Jews (8%), Germans (3.4%), Russians (2.5%). In 1989, Lithuanians constituted 81% of the population, Russians 98.3%, Poles 6.9% and Belarusians 1.7%. 0.2 Territorial division The territory inhabited by Lithuanians is often referred to as ethnic Lithuania, in order to distinguish it from the historical Grand Duchy of Lithuania, which was mainly made up of territories inhabited by Eastern Slavs (13th–16th centuries, subsequently as a part of the Polish-Lithuanian Commonwealth). Outside the Grand Duchy, Lithuanian was also spoken (up to World War II) in some areas of former East Prussia, where Lithuanians settled from the 14th century onwards (the autochthonous Prussian population, never very dense, had been decimated by war). These areas, formerly known as Prussian Lithuania (Germ. Preussisch-Litauen, now largely part of the Kaliningrad District in the Russian Federation), are often referred to as Lithuania Minor, as opposed to Lithuania Maior or Lithuania propria, i.e. the ethnically Lithuanian part of the Grand Duchy. Their inhabitants were evacuated to Germany after World War II, and this was also the case with those of the Klaipe˙da region (Germ. Memelland), an originally Lithuanian area also formerly belonging to Prussia. The Klaipe˙da region is once more inhabited by Lithuanians, of course, but for the most part they are not autochthones.

42

Laimute Balode and Axel Holvoet

Telšiai gr.

Šiauliai gr.

North

Kretinga gr.

West

Zemaitian

Kupiškis gr.

Paneve· zys gr.

Varniai gr.

Anykšciai ¦ gr.

South

Utena gr.

Raseiniai gr.

East Aukštaitian West Aukštaitian Kaunas group

Širvintos gr.

Vilnius gr.

South Aukštaitian

Lithuania Maior is divided into two regions of unequal size: –



Lower Lithuania or the žemaicˇiai region, covering the north-western part of the Lithuanian state, amounting to about 1/4 of its territory. The original žemaı˜ˇciai ‘Lowlanders’ were one of the tribes from which the Lithuanian nation was formed. The name of this region appears in Old Russian sources as Žomoitı˘, whence Polish Z˙mudz´ and Middle High German Samaythen, the form underlying the Latinized name Samogitia. Upper Lithuania or the aukštaicˇiai region, covering the remaining part of the Lithuanian territory. The aukštaicˇiai ‘Highlanders’ (whence Middle High German Eustoythen, Auxtote etc. and Polish Auksztota), were a tribe originally inhabiting the uplands of Ošmjany, to the south-east of Vilnius, a territory now predominantly Belarusian. The name acquired its modern meaning by becoming opposed to Samogitia.

Samogitia was the last ethnically Lithuanian region to be integrated into the Lithuanian state, and even after that, it retained a certain administrative autonomy and cultural identity. Until recent times the terms ‘Samogitia’ and ‘Samogitian’ were often opposed to ‘Lithuania’ and ‘Lithuanian’, which were taken to mean only Higher Lithuania. Quite often a Žemaitian language was opposed to Lithuanian, though this did not correspond to the linguistic reality; linguists distinguish

The Lithuanian language and its dialects

Žemaitian dialects, but they extend over a smaller area than former Samogitia as a territorial unit.

1.

The Lithuanian language

1.1 Lithuanian among the Baltic languages Together with Latvian, Lithuanian is said to constitute the eastern branch of the Baltic languages, as opposed to extinct Old Prussian. Lithuanian is extremely conservative in both phonetics and morphology, and, in most cases, Lithuanian forms can be taken to represent the nearest approximation to Common East Baltic and even Common Baltic forms. As compared to Latvian, Lithuanian is characterized, among other things, by the retention of free and mobile word stress and the original two-tone system, by less pervasive loss and shortening of final vowels, and by a more conservative morphology (with better retention of the inherited IE declension classes). It cannot be established with certainty whether the important part played by palatalization of consonants is an innovation of Lithuanian (influenced by a Slavic adstratum), or a feature inherited from Common Baltic. In morphology, Lithuanian has a few innovations as well: it has developed a special habitual past tense (which is, however, lacking in some dialects, cf. 4.1.10), as well as a new type of imperative with the affix -k(i)- (both lacking in Latvian). In syntax, Lithuanian is often more archaic than Latvian, using, for instance, prepositionless case forms where Latvian uses prepositional phrases. It has preserved the genitive of negation and the predicative instrumental (generally considered to be common Balto-Slavic features), whereas Latvian has lost both. Passive constructions differ in both languages: while Latvian has only an agentless passive, Lithuanian has an agented passive, with the agent phrase in the genitive (a feature regarded as an archaism of Indo-European date by some, but as a Lithuanian innovation by others). The widespread use of participles instead of finite verb forms in subordinate clauses is probably an archaic feature of Lithuanian (better preserved than in Latvian). On the other hand, the use of participles instead of finite verb forms as a special evidential form (the so-called oblique mood) is an innovation, shared with Latvian, and with parallels in Baltic Finnic. 1.2 The formation of the Lithuanian language It is known from historical sources that the Lithuanian nation emerged from a conglomerate of Baltic tribes, and that the Lithuanians were originally but one among these. This tribe apparently took the lead in organizing a unified Lithuanian

43

44

Laimute Balode and Axel Holvoet

state in the 13th century. However, no link can be established between the original tribal structure of Lithuania and its subsequent dialectal differentiation. Lithuanian is attested in writing from the beginning of the 16th century onwards. As it was never used before as an official language, and rarely in writing (the main exception being religious literature), the formation of a standard language was a slow process. The foundations of the present-day standard language were laid in the 1880s by publicists connected with the Lithuanian national revival, whose native dialect was in most cases the South-West Aukštaitian dialect of the socalled suvalkiecˇiai region (this term refers to the Lithuanian areas belonging, in Czarist times, to the province or ‘gubernya’ of Suvalki, which derived its name from the town Suwałki in present-day Poland). Another factor contributing to the role of South-West Aukštaitian in the formation of the standard language was the prestige enjoyed by this dialect as the vehicle of the tradition of Lithuanian writing in Prussian Lithuania (where the first grammars and dictionaries were also compiled). The final codification of standard Lithuanian was only possible after the establishment of an independent state in 1918. 1.3 Substrata and adstrata The oldest population whose presence on the territory of present-day Lithuania can be established were representatives of the Narew culture, which cannot be identified with any language group. A new migration wave, identified as Fenno-Ugric, entered the Baltic area in the 3rd millennium B.C. The Finnic population did not extend over the whole of present-day Lithuania (water names of Finnic origin are found predominantly in its northern and eastern parts) and was probably not very dense. It was ousted towards the close of the 3rd millennium B.C. by the bearers of the Corded Ware and Stone Battle-Axe Culture, who entered the Baltic area from the East. These were probably Indo-Europeans, but it is not clear whether they were Balts. The next migration wave, associated with the Hatched Pottery Culture, was certainly Baltic. It extended over a large area, stretching from the river Daugava in the North to the river Pripet in the South, and the upper reaches of the Volga and Oka in the East. The eastern part of this area is now inhabited by Eastern Slavs, but numerous water names testify to the former presence of Balts. Finnic substratum features cannot be pointed out as clearly in Lithuanian as they can in Latvian, though a few phenomena are reminiscent of Finnic. So, for instance, the series of local cases (illative, adessive, allative, cf. 3.2.4) attested in Old Lithuanian and the Lithuanian dialects (only fossilized remnants are found in Latvian) seems to echo the local cases of Baltic Finnic. The Lithuanians were originally one of several Baltic tribes inhabiting what would later become the Lithuanian language area. The north-western part, more or less coinciding with the territory of Samogitia, was once inhabited by Curonians, a

The Lithuanian language and its dialects

tribe whose territory extended northwards into present-day Latvia (the Samogitians themselves must originally have lived on the plains by the river Neve˙žis). Semigalians, whose territory also extended into present-day Latvia inhabited the area further to the east (north of Paneve˙žys). The Curonian and Semigalian substrata might be responsible for some features of the dialects of these areas. So, for instance, the character of the Žemaitian tones, and their similarity to those of Latvian, might somehow be connected with the Curonian substratum. Some would ascribe stress retraction in both Žemaitian and northern East Aukštaitian dialects to a Curonian or Semigalian substratum. Needless to say, it will never be possible to substantiate this. Among features which might be connected with a Slavic adstratum, so-called dzu¯kavimas, i.e., the affrication of dental plosives before high front vowels, must be mentioned. It occurs in the South Lithuanian area adjacent to Belarusian, where a similar affrication is observed (Polish affrication is of a different type). However, most scholars now explain dzu¯kavimas as an independent development in Lithuanian. 1.4 External influences From the 10th century onwards, the Lithuanian tribes, though successfully opposing the expansionism of Kievian Russia and the northern Russian principalities, were exposed to East Slavic influence. This is reflected in a number of loan words whose early date is attested by their archaic sound shape. The Lithuanian state which was formed in the 13th century expanded mainly into Ruthenian (Western and North-Western Russian) territories, and the Ruthenian element became prevalent in the culture, language and religion. The only language to be used in writing was the northwestern variety of Old Russian Church Slavic (the Ducal Chancery style). The separate development of the East Slavic dialects within the Grand Duchy of Lithuania eventually led to the rise of the Belarusian (White Russian) language. As a result of the Polish-Lithuanian Union (1569–1795), Polish language and culture became predominant among the upper classes from the 16th century onward. Lithuanian was hardly ever used in public life or for literary purposes, and even the Church did not encourage its use in religious life. Lithuanian survived only among the peasantry. In writing, Lithuanian was cultivated mainly in Prussia from the 16th century onward, under the influence of the Reformation. The symbiosis of languages and cultures within the Grand Duchy of Lithuania (Lithuanian, Belarusian, Polish) gave rise to a situation of widespread bilingualism even among the rural population, and Lithuanian was thus exposed to a strong Slavic (Belarusian and Polish) influence, which made itself felt in vocabulary, phraseology and syntax. The language developed a system of regular sound substitutions for Slavic loan words. The impact of (Great) Russian began to be felt after the incorporation of Lithuania into the Russian Empire (1795), especially after 1865, when the Czarist authorities began to enforce Russification and still further

45

46

Laimute Balode and Axel Holvoet

restricted the use of Lithuanian through administrative measures. The pervasive Slavic influence in the lexicon as well as in syntax was resisted by Lithuanian writers, philologists etc. from the 1880s onward (the Lithuanian national revival), and superfluous Slavic loan words were ousted from the standard language. A policy of Russification was again pursued after the incorporation of Lithuania into the Soviet Union in 1940. 1.5 Phonemic inventory 1.5.1The vowel system Vowels Short i ·iÒ

u ·uÒ o ·oÒ æ ·eÒ

a ·aÒ

Long /i˜/ ·y, ˛iÒ

/u˜/ ·u¯, u˛Ò /e˜/ ·e˙Ò

/o˜/ ·oÒ /æ˜/ ·e, e˛Ò

/a˜/ ·a, a˛Ò

Of the short vowels, /o/ occurs only in relatively recent loan words, such as kodas ‘code’. Some speakers of Lithuanian have a short /e/ ·eÒ, distinct from /æ/ ·eÒ, in such loan words as metras ‘metre’, but the autonomous existence of this vowel is not well enough established to justify its inclusion into the phonemic inventory. Diphthongs Rising diphthongs /ie/

/uo/

The first components are /i/ and /u/ respectively, but the second part is hard to identify (it is a centralized vowel /6/ rather than /e/ resp. /o/ in the dialects, these sounds are often realized as monophthongs. They also tend to lose tone differences in those dialects where long monophthongs lose them. For these reasons /ie/ and /uo/ are often described as non-homogeneous monophthongs rather than as diphthongs.

The Lithuanian language and its dialects

Falling diphthongs ui (oi) ei

ai

eu

au

/oi/ exists only in recent loan words such as boikotas ‘boycot’; /ui/ is exceedingly rare. Apart from loan words like Europa ‘Europe’, /eu/ exists only as a front counterpart of /au/ after palatalized consonants and is then written /au/, e.g. macˇiau [matëwjeu] ‘see’ (pret.1sg) 1.5.2Origins of the vowel system Lith. /a/ is the result of the merger of IE *a and *o. Short /o/ was reintroduced into the standard language quite recently, along with loan words (until then foreign /o/ was rendered as /a/, cf. Lith. (obs., reg.) gaspadorius ‘farmer, landlord’ from Polish gospodarz). /æ/ arises not only from IE *e, but also from original /a/ when preceded by a palatalized consonant, e.g. ve«rˇcia ‘turn’ (pres.3) is pronounced [«vjærjtëwjæ ] (/a/ and /e/ are thus in complementary distribution everywhere except word-initially). The long high vowels /i˜/ and /u˜/ are from IE *ı¯, *u¯, but also nasal /ı˜/ and /u˜/, from the sequences /iN/ and /uN/ (in the latter case, they are written ·i˛Ò, ·u˛Ò). The long mid vowels /e˜/ and /o˜/ are from IE long *e¯, *a¯. IE *e¯ was probably represented in Common Baltic by a broad /æ/, *a¯ by Common Baltic /a˜/; the narrowing occurred in the separate history of Lithuanian. The long low vowels /æ˜/ and /a˜/ have arisen from former nasal /ã/ and /«e/ (in this case, they are written ·a˛Ò, ·e˛Ò), but under stress they may also come from original short /æ/ and /a/, which were lengthened in this position while remaining short elsewhere. In the latter case, they are not distinguished from short /æ/ and /a/ in normal orthography; they automatically acquire circumflex tone, e.g. nãmas ‘house’ (compared to nami«e ‘at home’, where the unstressed /a/ has remained short).

47

48

Laimute Balode and Axel Holvoet

1.5.3The consonant system Labial

Dental & Alveolar

m

n

(])

Nasal

mj

nj

(]j) kg

Plosive

pb pj bj

td tj dj tës ·cÒ dëz ·dzÒ tësj dëŠj

Affricate Fricative

fv fj vj

Approximant

Palatoalveolar

Palatal

Velar

kj gj tëw ·cˇÒ dëŠ ·džÒ tëwj dëŠj x ·chÒ > ·hÒ xj >j

sz sj zj j

Lateral

l lj

Trill

r rj

In this table the representation of the sounds in standard orthography is given in angle brackets (unless trivial). All palatalized consonants are rendered by the addition of the letter ·iÒ if followed by a back vowel, e.g. liu¯tas [lju˜tas] ‘lion (nom.sg)’; when they are followed by a front vowel, palatalization it is not marked at all, as all consonants are palatalized in this position, e.g. dešine˙ [djæwji«nje˜] ‘right(-hand)’ (nom.sg.fem). Word-finally, palatalized consonants are replaced by their non-palatalized correlates (the opposition is neutralized in the standard language, though not in all dialects), e.g. eit [eit] ‘go’ (inf), shortened form of eiti [eitji] ‘id.’. Before another consonant, palatalization is not marked because the last consonant of a cluster always imposes its palatalized or non-palatalized articulation on the preceding ones, e.g. kélsiu [´kjæ.ljsju] ‘lift’ (fut.1sg) compared with ke˜ls [«kjæl.s] ‘id.’ (fut.3). 1.5.4Origins of the Lithuanian consonants The voiceless plosives come from similar IE consonants; voiced plosives may correspond to IE aspirated or unaspirated voiced plosives, e.g. /d/ may be from *d or *dh. The dental fricative /s/ is inherited, /z/ occurs mainly in loan words and onomatopoeia (or as an allophone of /s/). The palato-alveolar fricatives originate from IE palato-velar plosives, e.g. šimtas « ‘hundred’ (Lat. centum), žinóti ‘know’

The Lithuanian language and its dialects

(inf) (cf. Old Lat. gnosco). The soft palato-alveolar affricates arise from the clusters *-tj-, *-dj-, e.g. gaı˜džio < *gaid-j-a¯ ‘cock’ (gen.sg) alongside gaid«ys < *gaidij(a)s ‘id.’ (nom.sg). The corresponding non-palatalized affricates are rare apart from loan words. The affricates /tës/, /dëz/, /tësj/ and /dëzj/ are exceedingly rare in standard Lithuanian (except for loan words and onomatopoeia), but they are frequent in the ‘Dzukian’ dialects, see 3.2.2 below. The phonemes /x/, /xj/, />/, />j/ and /f/, /fj/ were introduced into the standard language at the end of the 19th century, along with foreign loans. All palatalized consonants have arisen either (1) by assimilatory palatalization before front vowels, or (2) from combinations with /j/. The phonological relevance of palatalization is established on the basis of minimal pairs with oppositions before back vowels, e.g. gilus [gjilus]‘deep’ (nom.sg.masc) vs. gilius [gjiljus]‘deep’ (acc.pl.masc). 1.5.5Prosodic features 1.5.5.1 Stress. Lithuanian stress is free and mobile, i.e. not only can it fall on any syllable of the word, but it often varies within inflectional paradigms. The system of word stress is closely bound up with that of tones or syllable accents (see below). This connection is of a twofold nature: (1) when stress falls on a syllable containing a long vocalic segment (on this notion cf. 1.5.5.3), it must obligatorily be realized in one of two ways: as acute or circumflex tone or syllable accent. Within lexical, derivational and grammatical morphemes, tone is, in principle, constant, i.e. whenever a given morpheme bears word stress, it displays a constant tone; change of tone (metatony) may, however, occur as a derivational device. So, for example, in laı˜svas ‘free’ the stem laisv- does not always bear stress, but when it does, it will be realized as circumflex tone; in láisve˙ ‘freedom’ the change of circumflex into acute tone is a marker of derivation (together with the change of declension type). (2) the tone of the stem syllable (or, in the case of polysyllabic stems, the syllable that bears stress within the stem) influences the pattern along which word stress may vary. Stress may be fixed (i.e., remain on the stem in all inflectional forms) only when it bears acute tone on the last stem syllable (as in vý-ras ‘man, husband’), or stress of any kind on a non-final stem syllable (as in vãsara ‘summer’, gárbana ‘curl’, nùgara ‘back’), otherwise it is always mobile within the paradigm. The patterns along which stress may vary are called accent paradigms; four of them are distinguished for nouns. 1.5.5.2 Quantity. Vowel length is distinctive in both stressed and unstressed syllables. In unstressed syllables, however, long vowels are realized as half-long, and

49

50

Laimute Balode and Axel Holvoet

there is a strong tendency for vowel length to be neutralized both in dialects and in the standard language. When a long vowel is stressed, it displays one of the two tones, (see below). Thus, under stress, length cannot be realized independently of tone. 1.5.5.3 Tone (syllable accent). Tones can be defined as alternative ways of realizing word stress in the case of long vocalic segments. These include: – – –

long vowels; diphthongs (both rising and falling); and diphthongoid sequences of the type /V + R/.

Though the sonorants /r/, /l/, /m/, /n/ (and /]/) cannot act as vowels in the sense of syllabic nuclei, their combinations with vowels behave like diphthongs as far as tonal differences are concerned. So, for instance, the sequence /um/ in trumpas « ‘short’ is marked for tone, just like the diphthong /ui/ in muı˜las ‘soap’ (nom.sg); but in lùpti ‘peel’ (inf) the first syllable simply contains a short vowel incapable of displaying tonal differences. When word stress falls on a long vocalic segment, it is realized as either acute or circumflex tone. This means that the constitutive features of stress, intensity (expiratory strength), pitch (tone height) and length, are spread over the vocalic segment in different ways. They can either be concentrated on the initial part (mora) of the vocalic segment, or they can be spread more or less evenly over the vocalic segment. Accordingly, we distinguish: – –

the acute tone, also called sharp (Lith. staigine˙) or, less appropriately, falling tone, and the circumflex tone, also called even or drawn (Lith. te˛stine˙) or, less appropriately, rising tone.

The distinctive nature of the opposition is shown by the following minimal pair: klóste˙ klõste˙

‘spread, pleat’ (pret.3) ‘pleat, crease’ (nom.sg)

Tone differences are most pronounced in the case of diphthongs and diphthongoid sequences, because these consist of articulatorily heterogeneous segments, the relative prominence of which is more clearly marked. The vocalic segment is then, as it were, split into two more or less distinct units (morae), the first or the last of which can be more prominent. Compare: láiškas laı˜kas

[¢la.iwkas] [«l6i.kas]

‘letter’ (nom.sg) ‘time’ (nom.sg)

There is a difference in relative length between both components; furthermore, the first vowel acquires a more closed and centralized articulation with circumflex tone,

The Lithuanian language and its dialects

and tends to merge with the second component to a larger extent than with acute tone. The realization of tones in the case of diphthongs seems to match the traditional notions of ‘falling’ and ‘rising’ tone most accurately. In the case of long vowels, the distinctive features described for diphthongs do not apply, as nothing seems to justify the distinction of two morae within the vocalic segment. But here as well, tone distinctions are accompanied by qualitative differences: vowels with the acute tone have been described as more tense, so that the above-mentioned forms klóste˙ and klõste˙ could almost be transcribed [´klo˜sjtje.] and [«klf˜sjtje.] respectively. Vowels with circumflex tone are, on average, also slightly longer than those with acute tone. The tones of the Žemaitian dialects differ considerably from the Aukštaitian and standard tones described here, cf. 4.1.8. In many Aukštaitian dialects, there is a tendency towards loss of tone differences in long monophthongs and in the rising diphthongs /ie/ and /uo/; at least, the distinction between the tones is perceptually slight. The tones are retained much better in the case of falling diphthongs, where their differentiation is facilitated by the differences in relative prominence of the initial and final components. 1.5.5.4 The use of tone marks. In principle, the circumflex accent (˜) is used to mark circumflex tone, the acute (´) to mark acute tone, and the grave (`) to mark a stressed short vowel, i.e. the lack of tonal differentiation (as only long segments are capable of distinguishing tones). There is, however, one exception from this rule. As mentioned above, in the case of diphthongs and diphthongoid sequences there is a marked difference in relative length between the two components. If the first component is /a/ or /æ/, it becomes half-long with the acute tone, but if the first component is /u/ or /i/, it is not perceptibly lengthened in the literary language (though it is in some dialects). In this case the acute tone is marked by putting a grave accent over the first letter in order to avoid the suggestion that it is lengthened, e.g. tvìrtas [´tvirtas] ‘firm’ (as opposed to tvártas [´tva.rtas] ‘stable, sty’). Here the grave marks acute tone, whereas in kìtas ‘other’ it marks stress and lack of tone. Moreover, the grave is used to mark the acute tone in a number of loan words, where no lengthening of the first component occurs, e.g. fèrma ‘farm’, spòrtas ‘sport’.

2.

The Lithuanian dialects

2.1 Classification The classification of Lithuanian into two main dialects, Aukštaitian and Žemaitian, has been known for centuries; they were often described as separate languages (‘Žemaitian’ being opposed to ‘Lithuanian’). The first attempt at a classification in

51

52

Laimute Balode and Axel Holvoet

modern times was made at the close of the 19th century by Antanas Baranauskas (Baranowski) and Kazimieras Jaunius. During the period of Lithuanian independence, their classification was improved by professional linguists such as Antanas Salys. The classification used nowadays was proposed after World War II by Zigmas Zinkevicˇius and Aleksas Girdenis, but in its basic shape it goes back to Baranauskas. 2.2 The differentiating features of the Lithuanian dialects The criteria used in classifying the Lithuanian dialects are: (1) In the vowel system: – – – – –

the treatment of the original sequences /aN/ and /eN/ (the main criterion in classifying the Aukštaitian dialects); the treatment of the diphthongs /ie/ and /uo/ (the main criterion in classifying the Žemaitian dialects and setting them apart from Aukštaitian); the treatment of the long vowels /e˜/ and /o˜/; the treatment of the short vowels /i/ and /u/; vowel loss in endings.

(2) In the consonant system: – the distribution of the dental plosives /t/ and /d/ and the affricates /tëw/ and /dëŠ/ (the Žemaitian Sound Law, used to set apart Žemaitian from Aukštaitian); – occurrence of palatalized and non-palatalized /l/ before non-high front vowels. (3) Prosodic features: – stress retraction (used in classifying both Žemaitian and Aukštaitian dialects); – the character of tones (sets apart Žemaitian from Aukštaitian).

3.

The Aukštaitian (High Lithuanian) dialects (aukštaicˇiu˛ tarme˙s)

3.1 Criteria of classification The main criterion used in classifying the Aukštaitian dialects is the treatment of the Baltic diphthongoid sequences of the type /aN/ and /eN/. Everywhere in Lithuanian, the nasal sonorant is preserved in such sequences when it occurs before a plosive, e.g. rankà ‘hand’ (nom.sg). However, where such a sequence occurred before a fricative or another sonorant, or word-finally, it was changed into a nasal vowel, which subsequently lost its nasal resonance and become a long oral vowel (still marked as a ‘nasal vowel’ in orthography, see above). If the vowel occurring in an original diphthongoid sequence with a nasal was *a or *e, this sequence may be of three kinds:

The Lithuanian language and its dialects

(1) in West Aukštaitian, the vowel did not undergo any change, and the diphthongoid sequence is reflected as either /aN/ and /æN/ or /a˜/ (·a˛Ò), /æ˜/ (·e˛Ò): Proto-Lith.

SL and WA

*ranka¯ *kampas *menkas *tempja *žansis *trenšja

rankà kampas « me«nkas tempia « ža˛sìs tr«e˛šia

meaning [ra]Áka] [«kam.pas] [«mjæ].kas] [«tjæmj.pjæ] [Ša.Ásjis] [«tjrj˜æwjæ]

‘hand’ (nom.sg) ‘corner, angle’ (nom.sg) ‘poor, worthless’ (nom.sg) ‘drag’ (pres.3) ‘goose’ (nom.sg) ‘fertilize, manure’ (pres.3)

(2) in South Aukštaitian, the vowel underwent no change when the sequences /aN/ and /æN/ were retained as such, but when a nasal vowel arose from such a sequence, it was narrowed into nasal /u˜˜/ and /«˜/, eventually giving rise to long oral /u˜/ and /i˜/: SL and WA

SA

rankà kampas « meñkas tempia « ža˛sìs tr«e˛šia

rankà kampas « meñkas tempia « žu¯sìs tr«yšia

[ra]Áka] [«kam.pas] [«mjæ].kas] [«tjæmj.pjæ] [Šu.Ásjis] [«tjrji˜wjæ]

(3) in East Aukštaitian, the development of the nasal vowels arising from /aN/ and /æN/ was similar as in South Aukštaitian, but, in addition, the vowels /a˜/ and /æ˜/ were narrowed into /u/ and /i/ when the nasal sonorants were also retained: SL and WA

EA

rankà kampas « meñkas tempia « ža˛sìs tr«e˛šia

ruñka kumpas « miñkas timpia « žu¯sìs tr«yšia

[«ru]ka] [«kum.pas] [«mji].kas] [«tjimjpjæ] [Šu.Ásjis] [«tjrji˜wjæ]

// runkà

[ru]Áka]

// timpja « // ž«u˛sis

[«tjimjpjæ] [«Šu˜sjis]

3.2 West Aukštaitian (vakaru ˛ aukštaicˇiu˛ tarme˙s) West Aukštaitian is in its turn divided into a southern subgroup, the dialects of the Kaunas region, and a northern subgroup, those of the Šauliai region. The latter are

53

54

Laimute Balode and Axel Holvoet

more innovative than the former: they are characterized by stress retraction and by the neutralization of vowel quantity in unstressed syllables.

˛ tarme˙s) 3.2.1The dialects of the Kaunas region (kauniškiu These dialects are sometimes referred to as dialects of the suvalkiecˇiai. The reason is that within Czarist Russia the major part of this region belonged to the province (‘gubernya’) of Suwalki (Polish Suwałki, a town situated in present-day Poland). As mentioned above (in 1.2), the standard language is based on this dialect group. The dialects of the Kaunas dialect happen to be the most conservative Lithuanian dialects, particularly with regard to the prosodic features described above for the literary language: 1. lack of stress retraction; 2. long vowels are distinguished form short ones even in unstressed position. The characteristics of this dialect can therefore, for the most part, be formulated negatively, as the lack of the features enumerated further on for the other dialects. The phonological system can be identified with the one described for the standard language in 1.5 above; as to its place within Aukštaitian cf. 3.0 above. The following features distinguish the Kaunas dialect from the standard language (most of them can also be found in one or other of the remaining dialects): 3.2.1.1 In some dialects of this group the oppositions of /a/ and /æ/, /a:/ and /æ˜/, /ai/ and /ei/ are neutralized word-initially in favour of /a/, /a˜/ and /ai/: SL

WA (Kaunas)

meaning

«ežeras eı˜ti

ãžeras a«Ht

‘lake’ (nom.sg) ‘go’ (inf)

As mentioned above (cf. 1.5.2), /a/ and /æ/ are in complementary distribution in all positions except word-initially, hence a strong tendency to neutralize the opposition in this position as well. It recurs in several dialects (cf. 3.1.2.5, 3.3.0.4 below). 3.2.1.2 In diphthongoid sequences of the type /V + R/ there is a tendency to lengthen the vowels /u/ and /i/ when the tone is acute (falling), whereas in the standard language this process is restricted to the vowels /a/ and /æ/ (see above, 1.5.2). This feature may be observed especially in the south-eastern part of the Kaunas region. In transcriptions, it is marked by using the acute instead of the grave accent (as in the case of the sequences /a + R/ and /æ + R/), e.g.

The Lithuanian language and its dialects

SL

WA (Kaunas)

meaning

tìltas nurìmti pìnti kùlti stùmdyti kùrmis

týltas nurýmti pýnti kØulti stØumdyti kØurmis

‘bridge’ (nom.sg) ‘abate, calm down’ (inf) ‘plait’ (inf) ‘thresh’ (inf) ‘push’ (inf) ‘mole’ (nom.sg)

3.2.1.3 Prothesis of /v/ before a word-initial vowel /o˜/ or diphthong /uo/ is observed in part of the Kaunas dialects: SL

WA (Kaunas)

meaning

uodegà óras

vuodegà vóras

‘tail’ (nom.sg) ‘air, weather’ (nom.sg)

3.2.1.4 The main prosodic feature deviating from the standard language is the circumflex tone of a few suffixes which have acute tone in standard Lithuanian: -ybe˙ ‘-ness, -hood’, -ytis and -aitis (diminutive suffixes). SL

WA (Kaunas)

meaning

bjaurýbe˙ šunýtis mergáite˙

bjaur«ybe˙ šun«ytis merga˜Fte˙

‘hideousness’ (nom.sg) ‘little dog’ (nom.sg) ‘little girl’ (nom.sg)

3.2.1.5 In the conjugation of verbs in -inti, -enti a different type of paradigmatic levelling has taken place from what is prescribed by the standard language. Sequences of the type /V + N/ normally undergo different treatments according to the nature of the following consonant (retention before obstruent, rise of a long oral vowel through an intermediary stage with a nasal vowel before a fricative or sonorant, cf. above, 3.0). This should give rise to alternations in the conjugation of the verbs in question. The standard language generalizes the stem variants in /æn/ and /in/: gyvenù gyvénti gyvénsiu

[gji.vjæÁnu] [gji.¢vjæ.njtji] [gji.¢vjæ.njsju]

‘live’ (pres.1sg) ‘live’ (inf) ‘live’ (fut.1sg)

Here the future has retention of /æN/ where /æ˜/ would be expected before a fricative. In the Kaunas dialects described here, levelling has taken the opposite direction, and the long oral vowel is generalized:

55

56

Laimute Balode and Axel Holvoet

SL

WA (Kaunas)

gyvénti gyvénk skõlinti skõlintu˛

gyv¢e˛ti gyv¢e˛k skõlyti skõlytu˛

meaning [gji.¢vjæ˜tji] [gji.¢vjæ˜k] [«sko˜lji.tji] [«sko˜lji.tu.]

‘live’ (inf) ‘live’ (imp.2sg) ‘lend’ (inf) ‘lend’ (cond.3)

3.2.1.6 In the north-western part of the Kaunas dialect group the dual is still used as a living inflectional form, whereas in the remaining Aukštaitian dialects it has virtually gone out of use (in Žemaitian, it is also retained in verbs): SL

WA (Kaunas)

m«es abù sùkame we:1pl both:masc turn:pres.1pl ‘we two/both turn’

mùdu sùkava we:1du turn:pres.1du

jÖus abù sùkate you:2pl both:masc turn:pres.2pl ‘you two/both turn’

jùdu sùkata you:2du turn:pres.2du

3.2.2The dialects of the Šiauliai region (šiauliškiu ˛ tarme˙s) 3.2.2.1 In the prosodic system, the neutralization of the opposition of quantity in unstressed vowels is observed. In the case of unstressed high vowels, this results in shortening without qualitative changes: SL su¯n«elis gyv«enimas

WA (Šiauliai) [su.«njæ˜ljis] sun«elis [gji.«vjæ˜njimas] giv«e nimas

meaning [su«njæ˜ljis] [gji«vjæ˜njimas]

‘little son’ (nom.sg) ‘life’ (nom.sg)

In the case of the unstressed mid vowels /e./ and /o./ this shortening results in short /6/ and /f/, but there may also occur narrowing into /i/ and /u/: SL

WA (Šiauliai)

meaning

sod«elis [so.«djæ˜ljis] sud«elis/sod« » elis [su«djæ˜ljis]/[so«djæ˜ljis] ‘little garden’ (nom.sg) te˙v«elis [tje.«vjæ˜ljis] tiv«elis/t˙ev«elis [tji «vjæ˜ljis]/[tje˙ «vjæ˜ljis] ‘little father’ (nom.sg)

3.2.2.2 On the other hand, the short high vowels /i/ and /u/ are somewhat lengthened under stress, yielding half-long vowels /i./ and /u./. As mentioned in 1.5.2, in the standard language this lengthening is characteristic only of the open vowels /æ/

The Lithuanian language and its dialects

and /a/. The half-long vowels resulting from this lengthening are apparently not susceptible to tone distinctions; Lithuanian dialectologists ascribe it an “intermediate” tone marked ·sÒ. Phonologically, this half-long pronunciation is evidently a variety of shortness, and the so-called ‘intermediate’ tone cannot, therefore, properly be considered a tone. SL pùse˙ kìšti

WA (Šiauliai) [Ápusje.] [Ákjiwjtji]

s

pu se˙ kisšti

meaning [pu.sje.] [kji.wjtji]

‘half ’ (nom.sg) ‘cram, thrust’ (inf)

3.2.2.3 Compared with the standard language and the Kaunas dialects, there is frequent stress retraction from endings onto stem syllables. There is some divergence between the dialects of this group, but usually stress is retracted only from a short final vowel onto a long vowel or diphthong in the stem: SL

WA (Šiauliai)

meaning

plytà rankà

pl«yta rañka

‘brick’ (nom.sg) ‘hand’ (nom.sg)

In the north of this region stress retraction is more general, and stress can be retracted from a long circumflected vowel in the ending onto a short vowel in the stem: SL

WA (Šiauliai)

meaning

šak«u˛

šàku˛

‘branch’ (gen.pl)

3.2.2.4 Word-final vowels are not preserved as well as in the Kaunas dialects: in final syllables short /a/ is usually apocopated (though short /a/ resulting from shortening of Proto-Baltic *a¯ is not affected by this process): SL

WA (Šiauliai)

meaning

výras dvãras

výrs dvárs

‘man, husband’ (nom.sg) ‘estate’ (nom.sg)

Note that as a result of apocope the sequence /ar/ acquires the status of a tautosyllabic diphthongoid sequence, which automatically acquires acute tone. 3.2.2.5 Word-initially the opposition between /a/, /a˜/, /ai/ and /æ/, /æ˜/, /ei/ is neutralized in favour of /a/, /a˜/, /ai/, as in some of the dialects of the Kaunas region (cf. 3.1.1.1):

57

58

Laimute Balode and Axel Holvoet

SL «ežeras eı˜ti

WA (Šiaul) a«žeras aı˜t

[«æŠjæras]

meaning [«a˜Šjæras]

‘lake’ (nom.sg) ‘go’ (inf)

3.3 South Aukštaitian (pietu˛ aukštaicˇiu˛ tarme˙s) This dialect group is also referred to as Dzu¯kian (dzu¯ku˛ tarme˙s), a name that is based on one of the most striking features of these dialects, so-called dzu¯kavimas (cf. 3.2.2). This use of the term is not accurate, however, as dzu¯kavimas occurs also in some East Aukštaitian dialects (the main criterion of classification being the treatment of /aN/ and /eN/ rather than dzu¯kavimas). 3.3.1 With regard to the chief criterion of classification, the treatment of the original sequences /aN/ and /æN/, the characteristic features of South Aukštaitian were already mentioned (cf. above, 3.0). These dialects have /aN/ and /æN/ without any change of the vowel, as in West Aukštaitian, but wherever these sequences have developed into nasal vowels, with subsequent loss of nasalization, the characteristic narrowing into /u˜/ and /i˜/ is observed: SL ža˛sìs vaı˜ka˛ tr«e˛šia kãte˛

SA [Ša.Ás’is] [«vai.ka.] [«tjrjæ˜wjæ] [«ka˜t’æ.]

žu¯sìs vaı˜ku˛ tr«yšia kãty

meaning [Šu.Ás’is] [«vai.ku.] [«tjrji˜wjæ] [«ka˜t’i.]

‘goose’ (nom.sg) ‘child’ (acc.sg) ‘fertilize, manure’ (pres.3) ‘cat’ (acc.sg)

3.3.2 Almost all dialects of this group are characterized by a series of sound changes referred to by the term dzu¯kavimas. This term is derived from the name dzu¯kai (singular dzu¯kas), which, in principle, is used to refer to the speakers of dialects where this dzu¯kavimas occurs, though, in practice, it usually refers to the South Lithuanians. The name dzu¯kas is onomatopoeic and suggestive of the high frequency of the affricates /tës/, /tësj/, /dëz/ and /dëzj/ which are lacking in other dialects. Dzu¯kavimas comprises two processes: (1) The affrication of /tj/ to /tësj/ and /dj/ to /dëzj/ before the high front vowels /i/ and /i˜/ (regardless of whether it developed from Proto-Baltic *ı¯ or *iN), as well as before the diphthong /ie/, the first component of which is also a high vowel.

The Lithuanian language and its dialects

SL tìk pi«r ti˛ tylùs tìek dìdis gaı˜di˛ dienà

SA [tjik] [«pjirj.tji.] [tji.Álus] [´tjiek] [Ádjidjis] [«gai.dji.] [djieÁna]

meaning

cìk pirci˛ cylùs cìek dzìdzis gaı˜dzi˛ dzienà

[tësjik] [«pjirj.tësji.] [tësji.Álus] [´tësjiek] [Ádëzjidëzjis] [«gai.dëzji.] [dëzjieÁna]

‘only’ ‘bath’ (acc.sg) ‘silent’ (nom.sg) ‘so much’ ‘great’ (nom.sg) ‘cock’ (acc.sg) ‘day’ (nom.sg)

This affrication is not impeded by /v/ intervening between the dental and the vowel: tviñko dvýlika

[«tjvji].ko.] [´djvji.ljika]

cviñko dzvýlika

[«tësjvji].ko.] [´dëzjvji.ljika]

‘swell’ (pret.3) ‘12’ (nom/acc)

Affrication does not occur before the other front vowels, /e(˜)/ and /æ(˜)/, which is obviously connected with the degree of palatalization: not only is palatalization usually weaker before the mid front vowels than before high front vowels, but it is also conceivable that assimilatory palatalization had not yet affected consonants followed by mid front vowels at the time when the affrication developed. Affrication is also lacking before a long /i˜/ arising from original *eN (as described in 3.2.1), which shows that dzu¯kavimas had ceased to operate when nasal /«e/ was narrowed. (2) The occurrence of /tës/ and /dëz/ (or /tësj/ and /dëzj/) as a result of the combinations of /t/ and /d/ with /j/, corresponding to /tëwj/ and /dëŠj/ of the other Aukštaitian dialects. Proto-Baltic SL

SA

(?) *svet-j-ai

ˇcià svecˇiaı˜

[sjvjæ« tëwjei.]

*med-j-ai

m«edžiai

[«mjæ˜dëzjei]

cià [tësjæ] sveciaı˜ svecaı˜ m«edziai m«edzai

meaning // cà [tsa] [sjvjæ«tësjei. ] [sjvjæ« tësai.] [«mjæ˜dëzjei] [«mjæ˜dëzai]

‘here’ ‘guest’ (nom.pl) ‘tree’ (nom.pl)

This second form of dzu¯kavimas extends over a slightly larger area than the affrication before front vowels (on East Lithuanian territory, its isogloss stretches north beyond that of affrication). It is clear that, given the affrication of the dentals before high front vowels, the occurrence of /tësj/ and /dëzj/ instead of /tëwj/ and /dëŠj/ implies a simplification of the morphophonological system: a binary alternation of the type /tj/ : /tësj/ is substituted for the ternary alternation /tj/ : /tësj/ : /tëwj/ which would otherwise arise as a result of affrication. It is therefore conceivable that the occurrence of / tësj/ and / dëzj/ instead

59

60

Laimute Balode and Axel Holvoet

of / tëwj/ and / dëŠj/ results from substitution rather than from regular phonetic development, but this cannot be proved. 3.3.3 A non-palatalized /l/ occurs, as opposed to the palatalized /lj/ of the standard language and most other dialects, when it is or has been followed by a mid or low front vowel (in the Lithuanian dialectologists’ transcription, this non-palatalized /l/ is rendered ·łÒ before a mid or low front vowel): SL

SA [«lje˜kjtji]

lÙekti

ł«ekc

meaning [«le˜ktës]

‘fly’ (inf)

As a concomitant of the lack of palatalization, /æ / and /æ˜/ were replaced by their back counterparts /a/ and /a˜/ (/a/ and /æ/ being in complementary distribution, as described in 1.5.2): l«edas

lãdas

‘ice’ (nom.sg)

A similar development can be observed when /l/ was originally followed by a nasal vowel continuing the sequence *eN. As described above (3.2.1), such a nasal vowel normally underwent narrowing into /i˜/; however, after /l/, which failed to undergo palatalization before a non-high front vowel, no /i˜/ could arise, as this vowel requires the preceding consonant to be palatalized; instead, a high centralized vowel arose, reminiscent of Russian y (/G/), and rendered by the corresponding Cyrillic letter in the Lithuanian dialectologists’ transcription: sáule˛

[´sa.uljæ.]

sául9

[´sa.ulG.]

‘sun’ (acc.sg)

3.3.4 In morphology, a characteristic feature of South Aukštaitian is the retention of the illative. It is a remnant of a series of local cases which are well attested in Old Lithuanian (traces can be found in Latvian as well), but have for the most part fallen into disuse in the standard language and the dialects. Four such cases can be reconstructed: –

the inessive (simply called ‘locative’), the only one to have been retained everywhere, corresponds to the IE locative, but has been renewed as to its form (usually an affix *-en, corresponding to the IE preposition *en, was added): e.g. mieste ‘in town’

The Lithuanian language and its dialects



(where the final -e results from the shortening of an original nasal vowel reflecting *-eN); the illative, being the lative counterpart of the inessive, and consisting in the original accusative with the affix -na added, e.g. miestan ‘into town’ (with -an from *-an-na);





the adessive, denoting proximity, and consisting of the original IE locative with the affix -pi added, e.g. Old Lith. Joniep (Germ. bei Johannes) ‘in (the Gospel of) John’ (lit. ‘near, close to John’); the allative, the lative counterpart of the adessive, denoting approach, movement towards a person or object, and consisting of the genitive with the affix -pi added (the modern standard language still has a few fossilized relics of this case): Old Lith. MS Lith.

Papiežauspi ‘to the Pope’ velnióp ‘to the Devil’

The adessive and the allative have fallen into disuse almost everywhere (except for fossilized relics), but the illative still occurs in some dialects as an alternative to constructions with the preposition ‘to, into’ (it may also occur in the standard language as used by those writers who have the illative in their native dialect): SL

SA

meaning

einu ˛i miesta˛

ainu miestan

‘I’m going to town’

3.3.5 In declension, South Aukštaitian has retained a number of archaisms. The old IE *i-stems and consonant stems have retained some inherited endings for which most dialects and the standard language have substituted analogical endings borrowed from other declensions: SL

SA

meaning

vãgiui píemeniui ãkmenys

vãgi píemeni ãkmenes

‘thief ’ (dat.sg) ‘herdsman’ (dat.sg) ‘stone’ (nom.pl)

Here the SA ending -i is the original ending of consonant stems, whereas -ui has been borrowed from the more productive *o-stems; nom.pl -es is the inherited IE ending for consonant stems, whereas -ys is the refashioned ending for *i-stems.

61

62

Laimute Balode and Axel Holvoet

3.3.6 In conjugation, on the other hand, the conditional shows a few innovative forms, refashioned after the pattern of the past tense: SL

SA

meaning

eı˜ˇciau eı˜tum

aitáu aitái

‘go’ (cond.1sg) ‘go’ (cond.2sg)

3.4 East Aukštaitian (rytu˛ aukštaicˇiu˛ tarme˙s) 3.4.1Common features of East Aukštaitian 3.4.1.1 With regard to the treatment of the original sequences /aN/ and /eN/ mentioned above (3.0), East Aukštaitian is characterized by consistent narrowing of the vowel, regardless whether the nasal sonorant is kept intact or disappears through the intermediary stage of a nasalized vowel: SL

EA

meaning

rankà ža˛sìs meñkas tr«e˛šia

runkà//ruñka žu¯sìs//žÖusis miñkas tr«yšia

‘hand’ (nom.sg) ‘goose’ (nom.sg) ‘worthless (nom.sg) ‘manure’ (pres.3)

3.4.1.2 As in the Šiauliai dialects of West Aukštaitian (cf. 3.1.2.2), the short high vowels /u/ and /i/ tend to be lengthened under stress; in some dialects of East Lithuanian, the opposition between original /u/ and /u˜/, /i/ and /i˜/ may be completely neutralized. In the latter case, they acquire circumflex tone; if lengthening is incomplete, dialects ascribe ‘intermediate’ tone (·sÒ) to these vowels: SL

EA

pùse˙

[Ápusje.]

mH`škas

[Ámjiwkas]

pu˜se˙ pusse˙ m«yškas misškas

meaning [«pu˜sje.] [´pu.sje.] [«mji˜wkas] [´mji.wkas]

‘half ’ (nom.sg) ‘wood’ (nom.sg)

The Lithuanian language and its dialects

3.4.1.3 As in South Aukštaitian, non-palatalized /l/ appears before mid and low front vowels and their reflexes, cf. 3.2.3. SL

EA

l«edas sáule˛

[«ljæ˜das] [´sa.uljæ.]

lãdas sául9

meaning [«la˜das] [´sa.ulG.]

‘ice’ (nom.sg) ‘sun’ (acc.sg)

3.4.1.4 As in some of the West and South Aukštaitian dialects, the opposition between /a/ and /æ/ is unstable word-initially. In the case of diphthongs, neutralization in favour of /ai/ is general, while the merger of /a(˜)/ and /æ(˜)/ into /a(˜)/ is restricted to the eastern dialects of East Aukštaitian: SL

EA

meaning

eı˜ti «ežeras

aı˜t ãžeras

‘go’ (inf) ‘lake’ (nom.sg)

3.4.1.5 In morphology, the retention of the illative is noted as in many South Aukštaitian dialects (cf. 3.2.4). 3.4.2The dialects of the Paneve˙žys region (paneve˙žiškiu ˛ tarme˙s) 3.4.2.1 Like the neighbouring Šiauliai dialect of West Aukštaitian, this dialect is characterized by a strong tendency towards stress retraction. The farther to the north, the stronger this tendency is. In the southern part of the region, stress is retracted from inflectional endings containing short vowels only (a), in the north it is also retracted from long circumflected vowels onto the stem (b):

a. b.

SL

EA (Pan.)

meaning

plytà vaikùs šakõs vaik«u˛

pl«ytu vaı˜kus šàkos vaı˜ku

‘brick’ (nom.sg) ‘child’ (acc.pl) ‘branch’ (gen.sg) ‘child’ (gen.pl)

3.4.2.2 Word stress plays a much more important part in determining the phonetic shape of a word than in the literary language and in West and South Aukštaitian. While the influence of stress on the quality and quantity of vowels in the Kaunas dialects of West Aukštaitian is relatively slight, it is enormous here, as well as in the neighbouring dialects of the Širvintos region. Not only is the distinction between long and short vowels in unstressed position virtually eliminated, but important qualitative changes occur as well.

63

64

Laimute Balode and Axel Holvoet

Long vowels are shortened in unstressed position: SL

EA (Pan.)

meaning

ãkys pelÙe

ãkis pèla

‘eye’ (nom.pl) ‘mouse’ (nom.sg)

Originally short vowels in inflectional endings are radically reduced (this is usually regarded as the most characteristic feature of this dialect). Only two vowels occur here: these are mid-high, lax, unrounded vowels noted /u/ and /i/ respectively; to the ear, they sound like something intermediary between /u/ and /o/, /i/ and /e/ respectively. /u/ represents any vowel following a non-palatalized consonant, /i/ represents any vowel following a palatalized consonant. This reduction of the original short vowel system must have originated in unstressed endings and subsequently have been generalized: SL

EA (Pan.)

meaning

šakà n«eša pelès sáuja katès

šak‚¡ n«ešu pel‚¡s sáuji katH`s

‘branch’ (nom.sg) ‘carry’ (pres.3) ‘mouse’ (acc.pl) ‘handful’ (nom.sg) ‘cat’ (acc.pl)

In the north, vowels in final syllables are often unvoiced (whispered) or may be completely reduced: SL

EA (Pan.)

meaning

výras pelès sáuja

výr’s pèl’s sáuj’

‘man, husband’ (nom.sg) ‘mouse’ (acc.pl) ‘handful’ (nom.sg)

Note that in the transcriptions the apostrophe is used to mark that (1) when the vowel is whispered, its quality cannot be established, and (2) even when it is reduced to zero, the preceding consonant behaves, phonotactically, as if a vowel was present, e.g., a voiced obstruent is not unvoiced before final /s/. 3.4.2.3 The reflexes of sequences of the type /V + N/ (cf. 3.0) vary somewhat: alongside /uN/ the reflex /oN/ » (with a narrow /o/) also occurs.

The Lithuanian language and its dialects

3.4.3The dialects of the Širvintos region (širvintiškiu ˛ tarme˙s) 3.4.3.1 Vowel shortening and reduction in unstressed syllables is comparable to what can be observed in the Paneve˙žys region. The results of the shortening of long vowels and the diphthongs /ie/ and /uo/ differ somewhat from the reflexes noted in the Paneve˙žys region. Instead of the long mid vowels /e./ and /o./ and the diphthongs /ie/ and /uo/, the low vowels /æ/ and /a/ occur. The unstressed vocalism is thus reduced to a simplified system including only the high vowels /i/ and /u/, and the low vowels /æ/ and /a/, and is, thus, identical to the original Lithuanian (and Common Baltic) system of short vowels. The existence of a reduced vowel system in unstressed syllables in combination with a richer vocalism under stress is analogous to East Slavic akan’e, especially to that of the Belarusian type. It should be stressed, however, that the occurrence of /æ/ and /a/ in the place of /e˜/ and /o˜/ is not the result of opening (lowering) in unstressed position. Since /e˜/ and /o˜/ come from Common Baltic /æ˜/ and /a˜/ respectively, the dialects of the Širvintos region have retained the original vowel quality. Apparently quantity in unstressed position became unstable at an early date in this dialect, so that original long and short vowels merged here while becoming differentiated under stress. SL su¯n«elis žmog«elis puod«elis pien«elis

EA (Širv.) [su.« njæ˜ljis] [Šmo.«gjæ˜ljis] [puo«djæ˜ljis] [pjie«njæ˜ljis]

sun«elis žmag«elis pad«elis pen«elis

meaning [su«njæ˜ljis] [Šma«gjæ˜ljis] [pa«djæ˜ljis] [pjæ«njæ˜ljis]

‘son’ (dim.nom.sg) ‘man’ (dim.nom.sg) ‘pot’ (dim.nom.sg) ‘milk’ (dim.nom.sg)

3.4.3.2 Stressed short /i/ and /u/ undergo full lengthening and acquire circumflex tone: SL bùvo mìškas

EA (Širv.) [Ábuvo.] [Ámjiwkas]

bu˜va mı˜škas

meaning [«bu˜va] [«mji˜wkas]

‘be’ (pret.3) ‘wood’ (nom.sg)

3.4.4The dialect of the Anykšcˇiai region (anykšte˙nu ˛ tarme˙s) 3.4.4.1 The most striking peculiarity (shared with the Kupiškis dialect) is the deviant treatment of original short stressed and non-final /a/, whether occurring on its own or as the first component of a diphthong or diphthongoid sequence. In the standard language, original stressed short /a/ is usually lengthened (there is a series of morphologically determined exceptions), and the /a/ occurring as the first component of a diphthong with the acute tone is pronounced as half-long. In the dialect of the Anykšcˇiai region the reflexes are narrowed: an open half-long vowel /f/ appears in both cases. It is described by dialectologists as showing the ‘intermediate

65

66

Laimute Balode and Axel Holvoet

tone’ (·sÒ). From the phonological point of view, the half-long pronunciation is, of course, a variety of shortness, and the ‘intermediate’ tone is actually lack of tonal differentiation. SL n«amas dárbas

[«na˜mas] [´da.rbas]

EA (An.)

meaning

nfs.mas dfs.rbas

‘house’ (nom.sg) ‘work’ (nom.sg)

In some of the dialects of this region, the reflexes of short stressed non-final /æ/ are narrowed as well; they are also half-long, and are ascribed “intermediate” tone. n«eša [«njæ˜wa]

ne» sša

Æ

[~nje» swa]

‘carry’ (pres.3)

3.4.4.2 As in the neighbouring dialects of the Širvintos region, the vowel system is poorer in unstressed vowel systems than in stressed ones (it is reduced to the opposition of high and non-high vowels). The unstressed long mid vowels /e˜/ and /o˜/, as well as the diphthongs /ie/ and /uo/, do not occur in unstressed syllables: instead, the low vowels /æ/ and /a/ appear; they retain their quantity, however (i.e. they are pronounced half-long), whereas they are shortened in the dialects of the Širvintos region. SL žmog«elis te˙v«elis pien«elis

EA (An.) [Šmo.«gjæ˜ljis] [tje.˜vjæ˜ljis] [pjie«njæ˜ljis]

žmageslis teveslis peneslis

meaning [wma.sgje.lji.s] [tjæ.svje.lji.s] [pjæ.snje.lji.s]

‘man’ (dim.nom.sg) ‘father’ (dim.nom.sg) ‘milk’ (dim.nom.sg)

3.4.4.3 Stress retraction is less general than in the dialects of the Širvintos region: it shifts from a short ending onto a preceding long vowel or diphthong; the latter is half-long and has ‘intermediary’ tone: SL žmonà su¯nùs

EA (An.) [Šmo.Ána] [su.Ánus]

žmfs.na sus.nus

meaning [sŠmf.na] [ssu.nus]

‘wife’ (nom.sg) ‘son’ (nom.sg)

3.4.4.4 In morphology the main peculiarity is the retention of the IE supine, a form related to the infinitive once used to describe the goal of a movement. It has disappeared from the standard language and most of the dialects. Note that the object of the supine is in the genitive, a feature which is sometimes inherited by the infinitive in those cases where it supplanted the supine, i.e. with verbs of motion.

The Lithuanian language and its dialects

SL

EA (An.)

eı˜sime ši«e no pjáuti go:fut.1pl hay:gen.sg cut:inf ‘We’ll go and cut the hay’

eismà ši«e no pjáutu˛ go:fut.1pl hay:gen.sg cut:sup

3.4.5The dialects of the Kupiškis region (kupiške˙nu ˛ tarme˙s) 3.4.5.1 The distinguishing feature of this dialect is the twofold treatment of /æ/ and /e˜/ according to whether they are followed by a palatalized consonant or not (as a rule, this is not a factor influencing the quality of vowels in Lithuanian). Where the standard language has /æ/ word-finally or before a non-palatalized consonant, the Kupiškis dialect has /a/; and where the standard language has /e˜/, it has /a˜/ under the same conditions. SL bè medùs bérnas katÙe

EA (Kup.) [bjæ] [mjæÁdus] [´bjæ.rnas] [ka«tje˜]

bà madùs bárnas katã

meaning [ba] [maÁdus] [´ba.rnas] [ka«ta˜]

‘without’ ‘honey’ (nom.sg) ‘young fellow, farmhand’ (nom.sg) ‘cat’ (nom.sg)

Note that the consonant preceding the vowel /a(˜)/ is not palatalized, which means that the change occurred before the rise of assimilatory palatalization. The treatment of /æ/ under stress when followed by a palatalized consonant is as in the dialects of the Anykšcˇiai region. 3.4.5.2 Unlike the dialects of the Anykšcˇiai region, those of the Kupiškis region have mid vowels in unstressed position; still, the vocalism of unstressed syllables is somewhat poorer than that of stressed ones, as the diphthongs /ie/ and /uo/ are not distinguished from /e˜/ and /o˜/; instead, half-long /e./ and /o./ appear: SL žmog«elis puod«elis te˙v«elis pien«elis

[Šmo.«gjæ˜ljis] [puo«djæ˜ljis] [tje.«vjæ˜ljis] [pjie«njæ˜ljis]

EA (Kup.)

meaning

[Šmo.sgje.lji.s] [po.sdje.lji.s] [tje.svje.lji.s] [pje.snje.lji.s]

‘man’ (dim.nom.sg) ‘pot’ (dim.nom.sg) ‘father’ (dim.nom.sg) ‘milk’ (dim.nom.sg)

3.4.5.3 Stress is retracted from a short ending onto a long vowel or diphthong, as in the Anykšcˇiai region.

67

68

Laimute Balode and Axel Holvoet

3.4.5.4 In morphology, the dialects of the Kupiškis region share the retention of the supine with those of the Anykšcˇiai region.

˛ tarme˙s) 3.4.6The dialects of the Utena region (uteniškiu 3.4.6.1 The most characteristic feature of these dialects is that the Common Baltic long open vowels /a˜/ and /æ˜/ have not undergone narrowing into /o˜/ and /e˜/ either in stressed or in unstressed position. Instead, we find a long /a˜/, slightly shifted to the back, or an open /f˜/, with intermediate stages: SL žõdis

[«Šo˜djis]

EA (Ut.)

meaning

[«Ša˜djis]//[«Šf:djis]

‘word’ (nom.sg)

Similarly, the equivalent of the /e˜/ of the standard language is a much broader vowel, ranging from /e˜/ to /æ˜/: te¢˙vas

[´tje˜vas]

[´tje˜vas]// [´tjæ˜vas]

‘father’ (nom.sg)

The retention of original /æ˜/ and /a˜/ is undoubtedly connected with the fact that the treatment of short stressed /æ/ and /a/ is similar to that in the Anykšcˇiai and Kupiškis regions: these vowels have a half-long pronunciation, but are phonologically short. There is thus usually no merger of original /æ/ and /æ˜/, or /a/ and /a˜/. In unstressed syllables, original /æ˜/ and /a˜/ are represented by half-long /æ./ and /a./, without narrowing. The diphthong /uo/ is usually preserved in unstressed position, but /ie/ has coalesced with /æ˜/: SL

EA (Ut.)

meaning

te˙v«elis pien«elis

[tjæ.svjæ.lji.s] [pjæ.snjæ.lji.s]

‘father’ (dim.nom.sg) ‘milk’ (dim.nom.sg)

The half-long /i./ which can be seen in the ending /-i.s/ in the above examples reflects a morphological fact: the Utena dialect uses the ending /-i.s/ for masculine nouns of this declension class, whereas the other dialects have the alternative ending /-is/. 3.4.6.2 As in the dialects of the Vilnius region, before the vowels /æ/, /æ˜/ and /e˜/ and the diphthong /ei/ the consonants /s/ and /r/ are not palatalized. The low vowels /æ/, /æ˜/ are correspondingly replaced by their back counterparts /a/ and /a˜/:

The Lithuanian language and its dialects

SL

EA (Ut.)

meaning

s«enas r«etas

sãnas rãtas

‘old’ (nom.sg) ‘rare’ (nom.sg)

3.4.6.3 The stress retraction characteristic of the East Aukštaitian dialects thus far described is not observed here. 3.4.6.4 The nominative forms of the personal pronoun jìs, jì ‘he, she’, plural ji«e, jõs, are replaced by the forms ãnas, anà, ãnys, ãnos, whereas in the oblique cases the forms derived from the stem of jis are retained. The nominative forms ãnas etc. are actually forms of the demonstrative pronoun meaning ‘yonder’. It is related to the Slavic demonstrative pronoun on, ona, ono, which is also used to supply the nominative forms of the 3rd person personal pronoun. The Lithuanian development therefore probably reflects Slavic influence (though an(a)s occurs also in Žemaitian, cf. 4.1.11). 3.4.7The dialects of the Vilnius region (vilniškiu ˛ tarme˙s) 3.4.7.1 These dialects share with those of the Utena region, among other things, the lack of stress retraction (unlike the remaining East Aukštaitian dialects) and the lack of palatalization of /s/ and /r/ before mid and low front vowels or their reflexes. Here also, we find instances of the unnarrowed pronunciation of Common Baltic and Proto-Lithuanian /æ˜/ and /a˜/, i.e. of the occurrence of /æ˜/ and /a˜/ where the standard language has /e˜/ and /o˜/, but the reflexes of these sounds show quite considerable variation in the Vilnius region. They range from /æ˜/ and /a˜/ to /ie/ and /uo/. The original diphthong /ie/ itself is not merged with the reflex of /æ˜/, but kept intact. Among the peculiarities distinguishing the dialect of the Vilnius area from the remaining East Aukštaitian dialects, we must mention: 3.4.7.2 Like the South Aukštaitian dialects on which they border, the dialects of the Vilnius area are characterized by dzu¯kavimas (cf. supra, 3.2.2). 3.4.7.3 In morphology, a series of particularly archaic endings is found. There are a few archaic consonant stem endings already cited from the South Aukštaitian dialects, where they also occur: SL

EA (Vil.)

meaning

ãkmenys akmeñs

ãkmenes akmenès

‘stone’ (nom.pl) ‘stone’ (gen.sg)

69

70

Laimute Balode and Axel Holvoet

Extremely archaic endings, used for all or nearly all declensions, can be found in the dative and locative plural. In the dative plural, some dialects have the ending -mu, corresponding to Old Lithuanian -mus (the inherited ending), whereas all other dialects have -m(s) (with loss of /u/). In the locative plural, the Vilnius dialects have the inherited endings in -su, whereas all the other dialects have introduced endings in -se after the pattern of the locative singular, which usually ends in -e. SL

EA (Vil.)

meaning

laukuosè vilkáms

laukuosù // laukúosu vilkãmu

‘field’ (loc.pl) ‘wolf ’ (dat.pl)

3.4.7.4 A curious feature in morphology is the use of two different imperatives: alongside the imperative derived with the affix -k(i)-, which is known in all Lithuanian dialects and in the standard language, there is a second imperative which is said to express a command that has to be obeyed immediately. SL

EA (Vil.)

meaning

rašýk

1. rašýk 2. rãšai

‘write’ (imp.2sg)

The imperative in -k(i) is a Lithuanian innovation, unknown even to Latvian (it is thought to contain an exhortative particle comparable to Russian -ka), whereas imperatives of the type rãšai seem to continue an original optative (they are also found in Old Lithuanian, and they are the only ones to be used in Latvian).

4. The Žemaitian (Low Lithuanian) dialects (žemaicˇiu ˛ tarme˙s) 4.1 Common features 4.1.1 Probably the oldest isogloss setting Žemaitian apart from Aukštaitian is that of the so-called ‘Žemaitian sound law’. It is reflected in the following forms, where Aukštaitian has the affricates /tëwj/ and /dëŠj/, whereas Žemaitian has /t/ and /d/: SL

Žemaitian

meaning

jáucˇiai m«edžiams

jáute¯ mèdems

‘ox, bull’ (nom.pl) ‘tree’ (dat.pl)

The mechanism underlying this divergence is the following.

The Lithuanian language and its dialects

At a certain stage in the prehistory of Baltic, /j/ was lost between a consonant and a front vowel. The result of this can be seen in the following alternation: vertì ve«rˇcia

< *vertie < *vert-j-a

dial. Lith p«asogà ‘dowry’) that is noted here for the first time, and one loan from Belarusian (aby, cf. Pol. ˙zeby ‘in order’). These traces of Polish or Belarusian influence on the Vilnius Russian publications are generally extremely rare. Some features of the Vilnius sociolinguistic situation were reflected in fictional works of Russian and Jewish authors. Baron Vasiliy Aleksejevich fon Rotkirx (1819–1891, pen-name ‘Teobal’d’) was one of the best known Russian writers in Vil’na in the second half of the 19th century. He wrote about everyday life in 19th century Vil’na (Vospominanija Teobal’da: 1890). In a novel Jacek Kryštalevicˇ, Vil’na’s jurodivyj’ ‘God’s fool’, for example (first published in Vilenskij Vestnik, 1888, N 259), he told the story of a legendary Vil’na ‘God’s fool’ and rhymester of the 1830s and 40s. The educated characters in the novel speak Polish while the speech of the countrymen and ordinary town folk is represented (in this and other novels) by a stylised Belarusian dialect, cf.: Sluxajt’e, xlopcy. ˇci vy bacˇili koli …malpu? …Tak heto malpa i jos’c’? … A lixo jaje golove, jakaja jana paskudnaja! …Kab jaje vovk zarezav! ‘Listen, chaps. Have you ever seen… an ape? …Is this an ape? … (liter.) I’ll luck for its head, how ugly it is! …May a wolf it devour!’ etc. (1996: 154). Teobal’d’s texts attest that Russian was not the everyday spoken language of the common people in Vil’na. L. O. Levanda (1835–1888) wrote the Jewish Russian feature-story (Goldberg 1900: 4). He published his sketches and novels in the first Russian-Jewish publications (Rassvet, Sion, Russkij Jevrej); the best of them were reprinted in central

Russian varieties: Urban Russian

Russian publications (Levanda: 1875). The characters of his works commonly speak “Jewish-German jargon”. This becomes clear when they are going to forward a petition to a Russian administrator: it turns out they needed to translate this petition into Russian. Everyone of them, the author remarks, “was internally, to himself aware it would not be easy to do, especially if you could not lay your hands on the Russian-German and German-Russian Schmid’s dictionary, this handbook of every Jew who begins to teach himself Russian” (Levanda 1875: 58). But it cannot be determined from the style of their direct speech that they are using another language than Russian. Some Jewish words can be found in Levanda’s prose (for instance: mexuten ‘bridegroom, fiancé’, xupa ‘bride’s croun’, gviry (Ydish gvirim) ‘reach people’, mašures ‘servant’ etc.), but their meanings are always explained and the language could be that of any of the educated Russian authors of that time. The same can be said about Levanda’s novel “The hot times” (1875) about Polish problems and life where the well-spoken Russian-speaking characters use some Polish words (usually spelled in Polish). Thus, Vilnius Urban Russian was the language of imperial administration, the press, literary works and the theatre. The very phenomenon of L. O. Levanda and other Jewish authors who wrote in Russian reflected the strengthened social position of Russian during the 19th century as one of the main results of Russification. There is no evidence to suppose that any social sections in 19th century Vilnius used the Lithuanian language and claim, as Semenova (1977: 213) does, that PolishRussian-Lithuanian trilingualism existed in the town. Moreover, it is doubtful whether there was any stable tradition of spoken Russian among the uneducated people here. The peculiarities of the local Polish, Belarusian and Jewish languages are not, as far as we know today, reflected in the Russian language of Vilnius publications. We cannot agree with Semenova (ibid.) that the Lithuanian language influenced Russian in 19th century Vilnius (and Lithuania generally). The official Russian language was obviously a source of lexical borrowings into the local Polish and Lithuanian dialects. By the end of the 18th century (Kurzowa 1993: 480) Polish authors from Lithuania (J. Chod’ko, A. Mickiewicz, F. Mickiewicz, the so-called Filomats) had used Russian loans such as bezporza˛dki ( si of Proto-Finnic is regarded by Posti (1953: 54f.) as a result of the Baltic influence. According to traditional opinion this change is regular both in inherited Finno-Ugric words and in Baltic loanwords but not in later strata of words, cf. Fin. syvä ‘deep’ /NSam. davvi ‘north’; Fin. sika ‘pig’ /ErzaMrd tuvo, from *tiwä and *tika respectively, and Fin. silta ‘bridge’ < Balt., cf. Lith. tiltas. As the reason for this change, Posti (1953: 54f.) suggests the palatalization of consonants in Baltic languages, Lith. tiltas ‘bridge’, e.g., being pronounced [t’i-]. The palatalized [t’], alien to the Proto-Finnic phonological system, was supposedly regarded as prestigious by the speakers of Proto-Finnic — in the same way as Baltic words for familiar notions — and therefore they imitated it but pronounced it more like an affricate [ts] from which the sibilant s later developed. Posti (1953: 54) is well aware that his hypothesis depends on the age of the palatalization of plosives before front vowels in Baltic languages. He bases his theory on Vaillant, according to whom the palatalization in this case would have been at hand in Proto-Balto-Slavic or even in periods before that. Today this hypothesis does not appear to be quite that uncomplicated. First of all, a development, like that assumed by Posti, i.e. *ti > *t’i > *t’s´i, has taken place also in East Baltic languages, cf. Lith. žodis ‘word’ (NSg.): žodžio (GSg.). There are, however, strong arguments against the high age of this development, e.g. writings of Lithuanian place-names in older sources, which do not show any affricate. This “indicates that in the 13th c[entury], perhaps even in the 14th c[entury, A.D.!]” they were still pronounced as plosives (Zinkevicˇius 1996: 99). Furthermore, the representation of this sound combination in the Žemaitish dialects of Lithuanian does not indicate that the change is of Proto-Baltic origin. This is probably supported by the testimony of Prussian texts — probably since it is difficult to tell what sound the writing in Prussian represents (Stang 1966: 103ff.). The early palatalization of *t’ in Baltic assumed by Posti on the basis of Vaillant consequently seems rather uncertain. The development of the affricate in Lithuanian is obviously a very late phenomenon and even if Stang (1966: 88f.) does not deny the possibility of *t’ in Proto-Baltic, it seems to me to be a far more important argument that palatalization is a spreading phenomenon. There is reason to believe that palatalization of consonants was not common in Proto-Baltic but spread in later times under the influence of Slavic languages. If this is correct, then Posti (1953) in his classic article has not given any example of Baltic influence on the phonological system of ProtoFinnic but several examples of Germanic influences.

244 Lars-Gunnar Larsson

5.

Baltic influence on Proto-Finnic syntax

Loanwords are not simple additions to the inherited stock of words, but they also influence the usage and functions of already existing vocabulary. This is obvious, if one considers borrowed adjectives denoting colours, e.g. Fin. keltainen ‘yellow’ < Balt., cf. Lith. geltas and Fin. harmaa ‘grey’ < Balt., cf. Lith. širmas. There are no natural borderlines in the spectrum of colours and the taking over of an adjective denoting a colour leads to restructuring of this semantic field. A good example of the close connection between a loanword and changes in syntax is Fin. ja ‘and’. Originally, Finno-Ugric languages had no conjunctions. Instead of subordinate clauses, nominal constructions were used, cf. Fin. Kallen tullessa Pekka lähti ‘when Kalle arrived, Pekka left’, lit. ‘in Kalle’s coming…’, and to express the co-ordination of elements clauses could be repeated, cf. Chanty ta]k6 wet6s, n´o˘h6s wet6s ‘he killed squirrels and sables’, lit. ‘squirrel he killed, sable he killed’; Nnts nisew ha¯´s, nebew ha¯´s, næ¯w ha¯´s ‘my father, my mother and my brother died’, lit. ‘my father died, my mother died, my brother died’ (examples from Collinder 1960: 249). Against this background I regard the obvious syntactical similarities that can be found between Finnic and Baltic languages as a result of the restructuring of the inherited, FinnoUgric usage depending on the heavy Baltic influence on Proto-Finnic that is testified by the loanwords. 5.1 The case of the subject and the object The Finnic variation of nominative ~ partitive as the cases of the subject and accusative ~ partitive as the cases of the object resembles the variation of nominative ~ genitive and accusative ~ genitive respectively in Baltic languages. In contexts, where Finnic languages have the partitive, Baltic languages often have the genitive. One cannot expect any exact correspondence, since we are dealing with the traces of an ancient language contact, that took place maybe 3000–4000 years ago, and since the morphological systems of the languages in question differ from each other. The differences in morphology can be illustrated by the variable of definiteness. In the Finnish standard language, definiteness can be expressed in several ways in different contexts: by word order, by the variation nominative ~ partitive (in existential clauses) etc., but there is no unambiguous morphological means of expressing definiteness. In Lithuanian, however, there is such a means, viz. when a noun has an adjective attribute, cf. baltas namas ‘(a) white house’: baltasis namas ‘the white house’. In both instances Finnish would have valkoinen talo, and word order would express difference in definiteness. Some examples will illustrate the similarities.

Baltic influence on Finnic languages 245

Partitive/Genitive as the case of the subject: (1) a.

b.

c.

d.

(2) a.

b.

Finnish Metsässä on susia. forest:iness.sg be:3sg.pres wolf:prtvpl Lithuanian Miške vilku˛ yra. forest:loc.sg wolf:gen.pl be:3.pres ‘There are wolves in the forest’ cf. Finnish Sudet ovat metsässä. wolf:nom.pl be:3pres forest:iness.sg Lithuanian Vilkai yra miške. wolf:nom.pl be:3pres forest:loc.sg ‘The wolves are in the forest’ Finnish Minulla ei ole isää I:adess neg:3sg be father:prtv.sg Lithuanian Ne˙ra man te˙velio beneg I:dat father:gen.sg ‘I have no father’

Partitive/Genitive as case of the object: (3) a.

b.

(4) a.

b.

(5) a.

Finnish Juon vettä. drink:1sg.pres water:prtv.sg Lithuanian Geriu vandens. drink:1sg.pres water:gen.sg ‘I drink water’ Finnish Tapasin vieraita. meet:1sg.impf visitor:prtv.pl Lithuanian Radau svecˇiu˛. meet:1sg.impf visitor:gen.pl ‘I met (some) friends’ Finnish En anna isälle rahaa. neg.1sg give father:all.sg money:prtv.sg

246 Lars-Gunnar Larsson

b.

(6) a.

b.

Lithuanian Aš te˙vui neduodu pinigu˛. I father:dat.sg neg:give:1sg.pres money:gen.pl ‘I don’t give (any) money to father’ Finnish Etsitkö veistä? search:1sg.pres.int knife:prtv.sg Lithuanian Ar peilio ieškai? int knife:gen.sg search:2sg.pres ‘Do you look for the knife?’

The use of the Finnic partitive and the Baltic genitive as the case of the object is not restricted to the verb ‘search’ but can also be found at other verbs denoting an aterminative action, e.g. Fin. tarvita, Lith. reike˙ti ‘need’; Fin. odottaa, Lith. laukti ‘wait (for)’; Fin. surra, Lith. gede˙ti ‘mourn’; Fin. pyytää, Lith. prašyti ‘ask (for)’; Fin. katsoa, Lith. žiu¯re˙ti ‘look at’; Fin. opiskella, Lith. mokytis ‘learn, study’. There are certainly more such examples, but it is necessary to define each case very carefully. Vagueness in this respect would justify criticism about a Finnic taking over “en masse”. To avoid such criticism I have in my own study (Larsson 1983) tried to delimit exactly the similarities to be treated. The similarity between clauses like the examples (1)–(6) seems systematic indeed and cannot be explained as a result of pure coincidence, especially not when taking the old Baltic loanwords from Proto-Finnic into consideration. Finnish words of Baltic origin, such as morsian ‘bride’ (cf. Lith. marti), sisar ‘sister’ (cf. Lith. sesuo), reisi ‘thigh’ (cf. Lith. rietas), heimo ‘tribe’(cf. Lith. šeima ‘family’), clearly indicate close contacts across the language border in very ancient times. I do not see any other explanation for the systematic, syntactical similarities between the Baltic and Finnic languages. The distributions of old Baltic loanwords and of the use of subject and object cases are similar: both are spread fairly evenly among the Finnic languages and can be clearly distinguished from the influences that depend on later contacts. One basic result of my investigation (Larsson 1983) is that the use of the partitive as the case of the subject and object in Finnic languages is, in fact, very similar to that of the Baltic genitive, whereas it differs considerably from the distribution of the etymologically corresponding case in Mordvin. In Mordvin, a use of the object case can be shown that is in clear contrast to the use in Finnic, e.g. MrdErza Sas´ viravas´ i karmas´ ´simime piva ‘the Forest Mother came and began to drink beer’. The object here, piva ‘beer’, is in the nominative. In such a clause Finnic would definitely take a partitive object: the verb ‘drink’ in itself can have both nominative/accusative and partitive objects, but after a verb denoting movement the partitive would be preferred (Larsson 1983:139f.). Furthermore, the object word ‘beer’ is in the partitive case, unless it denotes a definite quantity.

Baltic influence on Finnic languages 247

This inevitably leads to the conclusion that the change that has taken part in Proto-Finnic has made its case syntax more similar to that of Baltic languages. Even if the circumstances in Sami may be taken as counterevidence to such an explanation, rejecting it would entail a double standard in that lexical similarities are seen as evidence of near contacts and influence whereas syntactical similarities between the same languages are ascribed to mere chance. Mordvin and Finnic form the endpoints of a scale of development and a gradual approaching to the Baltic system. The next task is to put the Sami language on this scale, taking into consideration its position within the Finno-Ugric language group, Finnish influence on Sami and its indirect and direct Baltic loanwords etc. A traditional diachronic study of language change in the Baltic geographical area seems to me to offer a better explanation than theories based on more diffuse Sprachbund ideas (cf. Mathiassen 1985: 274). The Sprachbund theory should be supported by more evidence of a mutual influence in the area (cf. Larsson in press), but the direction of the loanwords in the Baltic-Finnic language contact clearly speak in favour of a Baltic influence on Proto-Finnic, whereas relatively few words have wandered in the opposite direction. In the lexicon of Baltic languages there do exist words that probably are of Finnic origin, e.g. Lith. bure˙ ‘sail’ and laivas ‘ship, vessel, craft’, cf. Fin. purje and laiva, but these few examples are an exception. There is no doubt whatsoever that the dominating direction of influences has been from Baltic to Finnic. 5.2 The agent participle of Finnish The resemblance in distribution of the subject and object cases is not the only example where Baltic and Finnic languages go together. One more instance should be mentioned, even if it is more complicated. In Finnish, there is what is called an agent participle, ending in -mA(-) that in its function as well as its form is very similar to the passive present participle of Lithuanian (Larsson 1996a), e.g.: (7) a.

b.

c.

Finnish Isän ostama auto. father:gen.sg buy:part car Lithuanian te˙vo perkamas automobilis father:gen.sg buy:part.masc.sg car ‘the car bought by father’ Finnish Auto on isän ostama. car be:3sg.pres father:gen.sg buy:part

248 Lars-Gunnar Larsson

d.

Lithuanian Automobilis (yra) te˙vo perkamas. car (be:3pres father:gen.sg buy:part.masc.sg ‘The car is bought by father.’

There is also in other Finno-Ugric languages a participle ending in *-m(V). In many languages it has preterite meaning and can be formed from both transitive and intransitive verbs, e.g. Udmurt gyrem busy ‘a ploughed field, a field that has been ploughed’ and lyktem kišnomurt ‘the arrived lady, the lady who has arrived’. Languages with both a present and a preterite participle tend to connect the present participle with active meaning and preterite participle with passive meaning, e.g. Hungarian and Swedish. The m-participle in Mari denotes a preterite passive, e.g. in MarEast omtam poˇc´ mo ‘the door (has been) opened’. The Mari language is crucial in the explanation of Finnish ma-participles. The Mari m-participles can function as Finnish agent participles, e.g. MarEast t6j kal6k6n mond6mo ulat ‘you are forgotten by the people’, having the agent in the genitive as in Finnish, but they also differ from Finnish, e.g. in being formed also from intransitive verbs, e.g. MarEast memnan [gen.] tolmo korno ‘the road that we have come’, lit. ‘the by us come road’. So even if the material, the participle in *-mA(-), is of Finno-Ugric origin, its functions in Finnic could have been influenced by Baltic languages, thus providing us with a good parallel to the distribution of the partitive in Finnic languages. Quite a few indications point to this as an instance of Baltic influence. First of all, the mA-participle in Finnish cannot be derived from intransitive verbs as it can in other Finno-Ugric languages. It should also be observed that there is no participle with the ending -m(V) in Mordvin, although it is one of the closest cognates of Finnic. Furthermore, in descriptive Finnish grammars the participle in -mA(-) tends to end up outside the participle system. The system is instead regarded as consisting of four forms, e.g. from the verbal stem sano- ‘say’:

Present Perfect

Active

Passive

sanova ‘saying’ sanonut ‘having said’

sanottava ‘that has to be said’ sanottu ‘having been said’

Apart from these participles there is also, according to the general description of Finnish grammar (e.g. Karlsson 1979: 193, 206f.), an agent participle in -mA. The root of the trouble is, of course, the term passive in Finnish, which denotes an impersonal form (by many researchers called impersonal passive). Fin. sanotaan, the present passive of the verb ‘say’, means ‘they say, people say’, i.e. a form that can in some languages be rendered by a passive verb form, e.g. Swd. det sägs or Lith. sakyma. Such so-called passive forms can in Finnish be formed also from intransitive

Baltic influence on Finnic languages 249

verbs, e.g. tullaan ‘they come’, and the frequent use of this so-called passive has somehow turned the attention away from the agent participle. Many studies have been devoted to the question of whether there is a passive in Finnish, but they have all basically dealt with the impersonal form. Now there certainly is one real passive in Finnish, as can be seen from the example: (8) a.

b.

Finnish Kalle ostaa auton. Kalle:nom buy:3sg.pres car:acc.sg ‘Kalle buys a car’ Finnish Auto on Kallen ostama. car:nom.sg cop Kalle:gen.sg buy:part ‘The car is bought by Kalle’.

It can be questioned whether a passive construction like this, in a language that is said to have no passive, could be interpreted as a sign of its foreign (Baltic) origin. One problem is that this participle in -mA is missing in Estonian. Another problem is that the Baltic participle in -ma does not represent the most frequent IndoEuropean ending of a passive participle, even if it has, in fact, parallels in other Indo-European languages (Szemerényi 1957: 107f.). The similarity between Lithuanian and Finnish is, however, so obvious, that the possibility of a Baltic influence cannot be neglected.

6. Later, separate contacts According to the traditional picture of language contacts in the Baltic area, the period of Baltic influence on Proto-Finnic was followed by a period of Germanic influences. Later on, when both Baltic and Proto-Finnic were split up into separate languages and dialects, influences from different directions can be distinguished in the Finnic language area: Swedish influence on Finnish, Russian on Veps and Karelian, and Latvian on Livonian and south Estonian dialects. Today, Livonian is spoken on the shores of northern Courland, but, until the middle of the 19th century, it was also spoken east of the Gulf of Riga. Since Livonian has been yielding to the Latvian language for a very long time, it is easy to understand that it has been heavily influenced by Latvian and that it is, e.g., abundant in loanwords from Latvian. Kettunen’s (1938) Livonian dictionary contains some 12,000 words. More than one fifth of these are Latvian loans. In such a large number of words one will, of course, find words pertaining to most areas of human activity: professions and occupations, housing, food and clothing, personal names, fauna and flora. Verbs make up a quarter of the Latvian loanwords. In spite of the

250 Lars-Gunnar Larsson

high absolute numbers of Latvian loanwords (more than 20% of the lexicon!), these loans amount to only 5% in a running Livonian text (Suhonen 1973). The Latvian loanwords in the southern dialects of Estonian have been thoroughly investigated by Vaba (1977). He has found more than 1,100 words of Latvian origin, but only a few — 254 words — are to be found in the dialects spoken in Estonia. The absolute majority of these words are restricted to Estonian dialects spoken in Latvia (Vaba 1977: 254). When compared to the influence on Livonian this number is surprisingly low. Both in Estonian dialects and in Livonian one can find examples of words that have been borrowed twice: first into Proto-Finnic and much later from Latvian into Estonian and Livonian separately (cf. Section 2 above). As a further example of such words one could mention Est. kauss ‘wooden bowl’ (Vaba 1977: 42) and Liv. koùš ‘Schüssel’ (Suhonen 1973: 136), which was also taken over by Proto-Finnic, as is attested by Fin. kauha ‘scoop’ (Kalima 1936: 112f.). Such words bear witness to the long duration of the contacts between the Baltic and Finnic languages. In the sound system of Livonian it has been suggested, though perhaps not convincingly (Kettunen 1938: xxxviii), that the glottal stop of Livonian is due to Latvian influence. The extensive use of the dative case in Livonian is obviously connected with Latvian influence, even if the material — once again — is inherited (Kettunen 1938: xli; Laanest 1982: 169). There is one case where an influence from Latvian on the southern Finnic languages and a Swedish influence on Finnish seems to explain a difference between the Finnic languages, namely the use of reflexive possessive pronouns (Nilsson 1988). In Estonian one reflexive possessive pronoun is used for all persons, as in Lithuanian, e.g.: (10) a.

b.

(11) a.

b.

Estonian Ma loen oma raamatut. I read:1sg.pres refl.pron book:prtv.sg Lithuanian Aš skaitau savo knyga˛. I read:1sg.pres refl.pron book:acc.sg ‘I read my book’ Estonian Sina loed oma raamatut. you read:2sg.pres refl.pron book:prtv.sg Lithuanian Tu skaitai savo knyga˛. you read:2sg.pres refl.pron book:acc.sg ‘You read your book’

Baltic influence on Finnic languages

(12) a.

b.

Estonian Tema loeb oma raamatut. he/she read:3sg.pres refl.pron book:prtv.sg Lithuanian Jis skaita savo knyga˛. he read:3pres refl.pron book:acc.sg ‘He (/She) reads his (/her) book’.

This system differs considerably from the Finnish system which distinguishes the possessor only in the 3rd person, as does to the Swedish system, in cases like (13) Finnish a. Hän luki kirjaa-nsa. he read.3sg.pret book:prtv.sg-poss.3sg ‘He read in his book’ (the reader = the owner of the book) b. Hän luki hänen kirjaa-nsa. he read his book:prtv.sg-poss.3sg ‘He read in his book’ (the reader π the owner of the book).

The word oma certainly exists also in Finnish, but there it expresses possession: (minun) oma kirjani is my own book and belongs to nobody else but me. In contrast, Est. oma, as is the case also with Lith. savo, does not stress any ownership but only signals that the subject person has the book at hand.

7.

Further fields of study

The study of linguistic contacts between the Baltic and Finnic languages is by no means exhausted by the examples given above. As a matter of fact, it seems that the Baltic influences on Finnic have had less attention paid to them than the Germanic influences, one of the reasons for this being that a larger number of Finnic scholars are trained in Germanic languages. This could explain, why, e.g., there is an almost total unanimity in regarding Fin. ranta ‘beach’ and its counterparts in other Finnic languages as a Germanic loanword. Nobody can, however, tell, what sounds were omitted in the initial consonant cluster of the source language when the word was taken over: it could have been st-, cf. Swd. strand ‘beach’, as is usually assumed, but it could equally well have been k-, cf. Lith. krantas ‘bank, sea-shore’, as suggested — to my knowledge — only by Jaunius (1972: 264–265, 323) and Terent’ev (1990: 30). There are certainly still more similarities between Baltic and Finnic languages to be discovered. Baltic influence has played an important rôle in the development of the present day Finnic languages from an older Finno-Ugric structure.

251

252 Lars-Gunnar Larsson

References Collinder, Björn. 1932. Die urgermanischen Lehnwörter im Finnischen. 1. Skrifter utg. av K. Hum. Vetenskaps-Samfundet i Uppsala 28: 1. Uppsala. Collinder, Björn. 1945. The Affinities of Lapp. In: Språkvetenskapliga Sällskapets i Uppsala Förhandlingar. Uppsala Universitets Årsskrift 1945: 12. Uppsala & Leipzig. Collinder, Björn. 1960. Comparative Grammar of the Uralic Languages. Stockholm. Hansegård, Nils Erik. 1967. Recent Finnish Loanwords in Jukkasjärvi Lappish. AUU: Studia uralica et altaica upsaliensia 3. Uppsala: A&W. Hertzen, Erik von. 1973. Itämerensuomen lainasanakerrostumien ikäämisestä. In: Journal de la Sociéte Finno-Ougrienne 72, 77–105. Helsinki: Suomalais-ugrilainen seura. Huss, Leena. 1991. Simultan tvåspråkighet i svensk-finsk kontext. AUU: Studia Uralica Upsaliensia 21. Uppsala: A&W. Jaunius 1972 = Kalbininko Kazimiero Jauniaus rankrastinis palikimas. Katalogas ir publikacijos. Vilnius. Kalima, Jalo. 1936. Itämerensuomalaisten kielten balttilaiset lainasanat. Helsinki. Karlsson, Fred. 1979. Finsk grammatik. 2nd ed. SKST 339. Helsinki: Suomalaisen kirjallisuuuden seura. Karsten, Torsten Evert. 1915. Germanisch-finnische Lehnwortstudien. Ein Beitrag zu der ältesten Sprach- und Kulturgechichte der Germanen. Helsingfors. Kettunen, Lauri. 1938. Livisches Wörterbuch mit grammatischer Einleitung. Lexica societatis fenno-ugricae 5. Helsinki: Suomalais-ugrilainen seura. Koivulehto, Jorma. 1988. Lapin ja itämeren suhteesta. In: Virittäjä 92, 26–51. Helsinki. Korhonen, Mikko. 1981. Johdatus lapin kielen historiaan. SKST 370. Helsinki. Laanest, Arvo. 1982. Einführung in die ostseefinnischen Sprachen. Hamburg: Buske Verlag. Larsson, Lars-Gunnar. 1981. Three Baltic Loanwords in Finnic. On dating pre-literary linguistic material. I: Acta Universitatis Upsaliensis. Acta Societatis Linguisticae Upsaliensis. Nova series. 3: 2, 17–55. Stockholm: A&W. Larsson, Lars-Gunnar. 1983. Studien zum Partitivgebrauch in den ostseefinnischen Sprachen. AUU: Studia Uralica et Altaica Upsaliensia 15. Uppsala: A&W. Larsson, Lars-Gunnar. 1984. Estnisch piim und finnisch piimä — ein baltisches Lehnwort? In: Fenno-ugrica suecana 7, 129–140. Uppsala: Finsk-ugriska institutionen, Uppsala universitet. Larsson, Lars-Gunnar. 1990. Zum Problem der uralischen Urheimat. I: Linguistica Uralica 26, 230–238. In Memoriam Paul Ariste. Tallinn: Akad. Nauk Est. Larsson, Lars-Gunnar. 1993. Zum Problem der urfinnisch-ugrischen Vogelbezeichnungen. In: Hajdú Péter 70 éves, 255–259. Szerk. Bakró-Nagy M. & Szíj E. Budapest: MTA Nyelvtudományi intézet. Larsson, Lars-Gunnar. 1996a. Bemerkungen zur Geschichte der mA-Partizipien im Finnischen. In: FU-8. Congressus Octavus Internationalis Fenno-Ugristarum. Pars III: Sessiones sectionum. Phonologia & Morphologia, 145–152. Red. H. Leskinen et al. Jyväskylä: Moderatores. Larsson, Lars-Gunnar. 1996b. Sprachen, die dem Lauf der Zeit trotzen? In: Ünnepi könyv Mikola Tibor tiszteletére, 222–226. Szerk. Mészáros E. & Janurik T. Szeged: JATE Finnugor tanszék. Larsson, Lars-Gunnar, in press: Quantitätsausdruck und Hauptwort in den ostseefinnischen und baltischen Sprachen. Methodologische Überlegungen. In: Symposium Konfrontation und Identifikation. Göttingen 14.–17. November 1997. Mathiassen, Terje. 1985. A Discussion of the Notion ‘Sprachbund’ and its Application in the Case of the Languages in the Eastern Baltic Area (Slavic, Baltic, and West Finnish). In: M. Flier & D. Worth (eds.), International Journal of Slavic Linguistics and Poetics 31–32. Slavic



Baltic influence on Finnic languages 253

Linguistics, Poetics, and Cultural History. In Honor of Henrik Birnbaum on his Sixtieth Birthday 13 December 1985, 273–281. ’s-Gravenhage. Mägiste, Julius. 1959. Gibt es im Tscheremissischen baltische Lehnwörter? In: Ural-Altaische Jahrbücher 31, 169–176. Wiesbaden: Otto Harrassowitz. Nesser, Anne. 1986. “Se oli kurja käya koulusa”. Bruket av inre och yttre lokalkasus i sverigefinska skolbarns språk. AUU: Studia Uralica et Altaica Upsaliensia 16. Uppsala. Nilsson, Torbjörn. 1988. Reflexive Possessive Pronouns in the East Baltic Area. In: Ural-Altaische Jahrbücher. N.F. 8, 18–37. Wiesbaden: Otto Harrassowitz. Ojanen, Muusa. 1985. Adjektiivikategoria venäläis-lyydiläisissä kontakteissa. Lingvistinen interferenssitutkimus. MSFOu 188. Helsinki: Suomalais-ugrilainen seura. Plöger, Angela. 1973. Die russischen Lehnwörter der finnischen Schriftsprache. Veröffentlichungen der Societas Uralo-Altaica 8. Wiesbaden: Otto Harrassowitz. Posti, Lauri. 1953. From Pre-Finnic to Late Proto-Finnic. In: Finnisch-ugrische Forschungen 31, 1–91. Helsinki. Qvigstad, Just Knud, 1893: Nordische Lehnwörter im Lappischen. Christiania VidenskabsSelskabs Forhandlinger for 1893: 1. Christiania. Sajantila, Antti et al. 1995. Genes and Languages in Europe: An Analysis in Mitochondrial Lineages. In: Genome Research 5, 42–52. Cold Spring Harbor Laboratory Press. Sköld, Tryggve. 1961. Die Kriterien der urnordischen Lehnwörter im Lappischen. 1. Skrifter utg. av Institutionen för nordiska språk vid Uppsala universitet. 8. Uppsala: Institutionen för nordiska språk vid Uppsala universitet. Sköld, Tryggve. 1996. Finn. meri ‘Meer’. In: Lapponica et Uralica. 100 Jahre finnisch-ugrischer Unterricht an der Universität Uppsala. Vorträge am Jubiläumssymposium, 311–316. Hrsg. von L.-G. Larsson. AUU: Studia Uralica Upsaliensia 26. Uppsala. SSA = Suomen sanojen alkuperä. Etymologinen sanakirja. 1-. 1992-. Helsinki: Suomalaisen kirjallisuuden seura. Kotimaisten kielten tutkimuskeskus. Stang, Chr. S. 1966. Vergleichende Grammatik der baltischen Sprachen. Oslo, Bergen & Tromsö: Universitetsforlaget. Suhonen, Seppo. 1973. Die jungen lettischen Lehnwörter im Livischen. MSFOu 154. Helsinki: Suomalais-ugrilainen seura. Svonni, Mikael. 1993. Samiska skolbarns samiska. En undersökning av minoritetsspråksbehärskning i en språkbyteskontext. Acta Universitatis Umensis. Umeå Studies in the Humanities 113. Umeå. Szemerényi, Oswald. 1957. The Problem of Balto-Slavic Unity. In: Kratylos 2, 97–123. Wiesbaden. Terent’ev, V. A. 1990. Corrections to the ‘Suomen kielen etymologinen sanakirja’ concerning Germanic, Baltic and Slavic Loanwords. In: Uralo-Indogermanica. II. Materialy 3-ej baltoslavjanskoj konferentsii, 18–22 ijunja 1990 g, 30–32. Moskva: Akademija Nauk SSSR. Thomsen, Vilhelm, 1890: Beröringer mellem de finske og de baltiske (litauisk-lettiske) Sprog. København. Vaba, Lembit. 1977. Läti laensõnad eesti keeles. Tallinn: Valgus. Wiklund, Karl Bernhard, 1892: Die nordischen lehnwörter in den russisch-lappischen dialekten. In: Journal de la Société Finno-Ougrienne 10. Helsingfors: Suomalais-ugrilainen seura. Zinkevicˇius, Zigmas. 1996. The History of the Lithuanian Language. Vilnius: Institute of Lithuanian language.



Part 3

Contact phenomena in minor Circum-Baltic languages



The role of language contact in the formation of Karelian, past and present Stefan M. Pugh

1.

History and the Linguistic Context

The earliest history of the Karelians is one common to the Finnic group as a whole: that is, the Finns, Karelians, Estonians, Vepsians, Ingrians, Livonians, and Votes. The exact time of their appearance in the Baltic region cannot be identified with any great certainty; thus, while some scholars assume their arrival to have taken place as early as the III or IV millennium BCE, others place the largest waves of settlement in the first millennium BCE. Contact between Baltic Finns and Eastern Slavs (the Slavic branch to which Russian belongs) can be posited as early as the sixth century CE. From the earliest days the relationship between (Proto-) Russians and (Proto-) Karelians was one of apparently peaceful coexistence; the northern territories which they jointly inhabited were sparsely populated, and there was no evident conflict between them. Indeed, as the Karelians lived alongside Russians, they adapted to the growing Russian presence by participating in the Russian cultural sphere: they adopted (Russian) Orthodox Christianity and presumably spoke Russian — judging by the large number of early Russian borrowings in their language. By the beginning of the twentieth century, after a millennium of Russian growth and development, the position of the Karelian people gradually shifted from one of coexistence to de facto subordination (but without suppression) as an ethnic and linguistic minority within a large and powerful state. Centuries of contact between the (Proto-) Russian and Karelian peoples have left a distinct mark on the language of the latter. To the casual reader or listener, this mark will be perceived as lexical in nature, as the spoken — and now written — language of Karelians contains a vast number of lexemes borrowed from Russian. At the very earliest stages such borrowings would have been limited to semantic spheres specific to the Russian culture or to technological advances made by

258 Stefan M. Pugh

Russians, for which native Finnic words were lacking. The extensive and intensive nature of language contact between these peoples led very quickly, however, to the presence of massive numbers of Russian lexemes, from every semantic sphere, in the everyday speech of the Baltic Finns of Russia. Given that the Karelians are and have been bilingual in their L1 and Russian, it is extremely easy for Russian forms to be used in L1 without causing any difficulties of understanding. The Karelians who attended school were educated in Russian — the only language of education at all levels before the present day. Not all words of Russian origin used in Karelian can be termed ‘borrowings’, of course, and any number of labels can be attached to lexemes casually used in speech. I have applied the term ‘usings’ to such forms: a ‘using’ is a lexical form that is in use in the L1 of a given speaker, but that cannot with certainty be labelled a borrowing. A test that might confirm that a particular lexeme has achieved the status of a borrowing is its occurrence in a number of dialects of the same language, or in the language of a significant number of speakers of the same dialect. The investigative linguist who goes beyond the lexicon discovers that evidence of Russian influence on a language like Karelian extends to much deeper levels of the language, however: to phonology, morphology, and even to word-formation; in the following analysis I shall briefly examine each of these linguistic components in turn. The present day is the subject of the final section of the study, as it is concerned not only with the current state of the language, but with the direction it appears to be taking given current attempts to codify it, use it in newspapers and literature, and use it as a medium of instruction in schools.

2.

The Lexicon

2.1 The oldest layers of Russian borrowings clearly reflect the nature of the early Russian-Karelian relationship, as borrowings include such culturally important notions as ‘cross’ (Krl risti) and ‘pagan’ (Krl pakana), etc. Interestingly, the old Finnic word for ‘Bible’, based on ORus gramota ‘writing, written document’ has been superseded by Krl bibl’ii, based on Rus biblija (cf. Fin raamattu ’Bible’). This relationship continued and intensified over a period of centuries as the populations became more mixed — but mainly in urban centres; the influx of Russians in large numbers to the northwest began in earnest after the foundation of St Petersburg (1703) and Petrozavodsk (1777). With the growth of towns came changes to the early Finnic way of life; as a consequence, the development of the Karelian lexicon now came to be centred on such semantic fields as agriculture, government, the military, and the like. With the advent of the Russian revolution of 1917 came the last and most intensive phase of interaction: the imposition of absolute Soviet control over the economy and political life of the Soviet peoples led to the adoption

The role of language contact in the formation of Karelian, past and present 259

of great numbers of lexemes related to these two key spheres of life in the USSR. It must be remembered that all the while such contact was progressing, Karelian was an unwritten language, and that all lexical transferrence occurred in the spoken language. By the 1930’s many Soviet minority languages were ‘given’ the Cyrillic alphabet and codified; some of these were successful and continue to be successful literary languages, e.g., Komi and Udmurt. The success of any codification process is often a function of several factors, not least important of which is the number of speakers using a given language. Some of the codified languages, e.g., literary Komi and Udmurt, have survived and are widely used by relatively large numbers of people. The ultimate choice of alphabet was not, of course, made on purely linguistic grounds; in practical terms, writing in Udmurt in Cyrillic made the transferrence of Russian lexemes into literary Udmurt extremely easy. This somewhat cynical statement is of course valid only for lexical terms characteristic of sophisticated semantic domains: it does not apply to the hundreds or even thousands of Russian-based words that were already present in the spoken language. The attempt to codify Karelian failed, however, for a variety of reasons, political, sociolinguistic, and purely linguistic: 1) Finnish was instituted as a language of instruction and publication in 1940; 2) a suitable dialect base for the literary language was difficult — if not impossible — as literary Karelian was intended for speakers of three major dialects; 3) the geographic separation of speakers of these dialects; 4) the further dislocation and loss of speakers in World War II; and 5) the Cyrillic orthography — as it was applied to Karelian — did not suit the phonological structure of the language. For more detail on the sociolinguistic and linguistic issues involved, see Pugh (forthcoming) and Austin (1992). If one compares the lexical base of Karelian with that of, e.g., Udmurt, it was the sheer weight of the dominance and necessity of Russian that resulted in the presence of large numbers of Russian-based lexemes in Karelian, while the lexicon of Udmurt was expanded by Russian via both the bilingualism of its speakers and language planners/lexicographers. Evidence of the massive penetration of Russian lexemes into Karelian is found in all semantic spheres, far beyond those of politics and economics; thus, any given utterance by a Karelian can contain Russian-based lexemes for which native Finnic equivalents exist: (1)

Tihvin Krl a. voina ‘war’ < Rus vojna; cf. Fin sota Rukajärvi Krl b. pol’n’ittša ‘hospital’ < Rus bol’nica; cf. Fin sairaala

260 Stefan M. Pugh

All Krl dialects c. proit’i-e, proid’i-e ‘go, go through or past-INF’ < Rus projti

2.3 We also find that many Russian numerals are used in place of Karelian ones, especially large numerals in the expression of years, with a native Finnic case form at the very end: (2)

Aunus Krl a. sorok p’at -oil 40 5 -ADESS/ALL ‘in the year (19)40’ b.

d’evätnatcat -oil 19 - ADESS/ALL ‘in the year (19)19’

vuvv -el year - ADESS/ALL

vuvv -el year - ADESS/ALL

This particular phenomenon is described as reflecting a very significant level of interference by Thomason and Kaufman, and even as one of the signals of eventual language death (1991: 74–6). 2.4 Russian conjunctions (e.g. i, a ‘and’, no ‘but’, štoby ‘in order to’) and particles (že ‘emphasis’, [z]nacˇit ‘so, then’) are also heavily used in Karelian. 2.5 Even more interesting from a structural point of view is the occasional incorporation of Russian prepositions into the Karelian system, as they represent elements that have penetrated the system beyond the surface level of the lexeme, into the morpho-syntactic level: (3)

Salmi Krl a. kromi hän-däh except him-PRTV ‘except for him’, (Rus krome) b.

mesto hän-däh instead of him-PRTV ‘instead of him’ (Rus vmesto)

We note in both examples that it is the Karelian partitive that occurs with a following form, whereas it is the genitive that is required in Russian; this is not altogether surprising, as there are numerous Russian-Finnic parallel constructions in which these two cases occur as functional equivalents (e.g., the case of nouns/noun phrases following numerals; the Finnic partitive vs. Russian partitive genitive, etc.).

The role of language contact in the formation of Karelian, past and present 261

2.6 Finally, code-switching must be mentioned as crucial to Karelian speakers: Russian words (unchanged in form from the original, often including Russian case endings) and entire Russian phrases occur throughout the speech of many Karelians, especially of those living in larger and more russianized towns (for a detailed study of this phenomenon, see Pyöli 1996). Russian therefore continues to have an extremely high profile in everyday spoken Karelian, beyond the presence of Russian borrowings/usings discussed above.

3.

Phonology

Over the centuries, contact with Russian has changed the face of Karelian phonology. The two most important features of contemporary Karelian to be considered here are (1) consonantal palatalization and (2) consonant clusters. Consonant palatalization is present in a number of Baltic Finnic languages, partly as a result of internal processes, partly as a result of Russian influence; if palatalized consonants originally arose before i and j, they can now occur before e (and in Russian borrowings or usings before a, o, and u). This is now a feature not only of Russian borrowings/usings, but significantly also of native lexemes; there is some evidence that palatalized consonants are even gaining phonemic status (Pugh 1994: 47). Compare: (4)

Russian-based forms zvon’iw ‘he/she calls’ pl’et’t’ie ‘to weave’ ud’ivie ‘to surprise’ l’upoi ‘any, whichever’ l’ohkoi ‘simple’

(5)

Finnic-based forms ol’i ‘he/she was’ l’äks’imä ‘we left’ tul’i ‘he/she came’ n’enä, n’en’ä ‘nose’ n’euvonta ‘instruction’, etc.

Note that labial consonants generally resist palatalization in Karelian, although some speakers of one dialect (Tver’ Karelian) do have palatalized labials. The proliferation of consonant clusters in the Russian lexicon has led to their presence in large numbers in Karelian; they are often reduced by one element in the process of transferrence, but if the original Russian form contains a 3-element cluster, than the resulting Karelian word will still have a 2-element cluster that

262 Stefan M. Pugh

might have been unacceptable to the original Finnic system. In many cases we find that even 3-element clusters are retained in the Karelian form, sometimes side-byside with a form in which it has been reduced.The following make up but a small sampling of this widespread phenomenon: (6)

srojie

(e.g., Aunus Krl) ‘to build’ < Rus. stroit’; but strojie is also attested (Ludic, Valdai Krl) striettie (Ludic Krl) ‘to meet’ < Rus. vstretit’; note that there is still a 3-element cluster in the new form (the next stage, were there to be one, would presumably be *sriettie) tvettie (Valdai Krl) ‘to bloom’ < Rus. cvesti, stem cvet-

Finally, a feature of Karelian phonology — vowel harmony — is retained in that language. But Russian can also play a role in the realization of this feature in borrowings/usings: in verbs in particular, the presence of a prefix (which is not a structural feature of Finnic) can dictate the front-back colouration of the new Karelian form, as a function of the vowel of that prefix. Otherwise it is the desinence of vowel(s) of the root which is (are) important. (7) Examples without prefixes (where -i- is the marker of the verbal stem-type): a.

brod -i -u (root back vowel > desinential back vowel) wander -SM -3SG ‘he/she wanders’

b. sluav -i praise -SM ‘to praise’

-da (root back vowel > desinential back vowel) -INF

c.

-ü (root front vowel > desinential front vowel) -3SG

d’ern’ burst

-i -SM

d. sluuž -i serve -SM ‘to serve’

-mah -INF

(root back vowel > desinential back vowel)

e.

-mäh -INF

(root front vowel > desinential front vowel)

pl’et’t’ -i weave -SM ‘to weave’

(8) Examples with prefixes; most telling are those forms with a back vowel in the prefix, front vowel in the root (exx. c-e): the former overrule the latter:

The role of language contact in the formation of Karelian, past and present 263

a.

pere-živ -i -imö survive -SM -1PL ‘we survive’

(front prefixal vowel, front desinential vowel)

b. pr’i-miet’ -i -tää notice -SM -3PL ‘they notice’ c.

u-id’ -i leave -SM ‘he/she leaves’

d. za-d’ien’ put ‘we put’ e.

-u -3SG

(back prefixal vowel, back desinential vowel)

-i -mma -SM -1PL

na-l’ei -cˇemma pour -1PL (no SM in this stem-type) ‘we pour out’

It is striking that the prefix, a linguistic element that is not historically part of the L1 Finnic system, should play such a commanding role in this Karelian phonological domain. The inference that has to be made is that the verbal prefix is psychologically real to the Karelian speaker, and — perhaps because it is so semantically important in Russian, and therefore in Russian borrowings — it attracts phonological attention as well; perhaps more than ‘attention’, as the vowel of the prefix is clearly in control. In one respect, however, there is true interaction between the Russian prefixal and Karelian phonological systems: the vowel of the prefix does not control the quality of the desinential vowel if the prefixal vowel is i or e and a back vowel occurs in the root. In other words, these two ‘neutral’ vowels — which can occur with back or front vowels in Karelian and Finnish — will be overruled by a back vowel in the root. Compare example (f): f.

pri-nuud’ -i -u compel -SM -3SG ‘he/she compels’

(neutral-back-neutral-back)

For a more detailed study of the whole phenomenon, see Pugh 1996. As many of these and later examples reflect, Baltic Finnic consonant gradation is retained in Karelian, and in fact more widely applied than in standard Finnish; Russian-based forms with stem-final voiceless consonants take part in this process.

264 Stefan M. Pugh

4. Morphology In the strictest sense of the word, the ‘morphology’ of Karelian is unaffected by the influence of Russian — if we discount such immutable elements as conjunctions and particles. The native case system is stable (although syncretism between particular cases has occurred in various dialects, without external influence), as is the conjugational system of personal endings and tenses. However, there is a certain amount of systemic convergence between Karelian and Russian that has affected the morphological profile of Karelian to some extent, again in the verb. The existence of a large number of distinct Russian verbal stem types has meant that the transformation of Russian verbs into Karelian ones must involve a process of assimilation to native Karelian stem types. In particular, two large classes of verbs in Russian are those with stem-markers -i- and -Vj- (in which V = vowel). The key is that both of these stem types exist in the native Finnic system, but they represent considerably smaller sets of verbs in languages like Finnish (and originally in Karelian) than they do in Russian. In addition, all Russian verbs that are not of the -Vj-type (and therefore not just original Russian -i-type verbs) are assimilated into Karelian as -iverbs; this fact, and given that the Russian -Vj-class is so large, means that the profile of both types has grown considerably in Karelian. Thus, the change is not qualitative, but quantitative in nature. The following examples are given as stems (not the infinitive, as that form masks the identity of the stem for a great number of verbs), according to their verbal class in Russian: (9)

RUSSIAN a.

KARELIAN

‘read’ ‘dance’ ‘have’ ‘dig’

cˇitajtancuj-, tansujimejroj-

‘call’ ‘meet’ ‘receive’

zvon’i(v)striet’t’polucˇ’cˇ’i-, etc.

‘sign’ ‘punish’ ‘gather’ ‘break through’ ‘begin’ ‘reach’ ‘transport’ ‘help’

potpiššinakažisoberiprorvinacˇn’idoid’iperevez’ipomogi-, etc.

‘i’ type zvonivstretipolucˇi-

c.

>

‘Vj’ type cˇitajtancujimejroj-

b.

>

‘All others’ podpišnakažsoberprorvnacˇndojdperevezpomog-

The role of language contact in the formation of Karelian, past and present 265

5.

Word-formation

Prefixes were mentioned in the context of phonology above, specifically in regard to the regulation of vowel harmony in Russian-based verbs. They could also have been mentioned in the context of verbal morphology, as the prefix is a vital element in the structure of Russian. Russian verbs are borrowed with prefixes, and the prefix is a semantically real element to speakers of Karelian; thus, I have shown that the use and distribution of Russian verbs of motion in Karelian is in fact often dictated by the meaning of a given prefix (Pugh 1991). There are also indications, however, that some Russian prefixes are able to be affixed to native Finnic stems: thus, we find pere- and do- in Vepsian, and pere- in isolated Karelian forms. Such a phenomenon is evidence of extreme systemic interference; it is impossible to tell at this point whether or not the tendency will grow or become more widespread, especially given the current attempts to codify such languages as Karelian and Vepsian (see 5 below).The following are examples of Russian prefixes found with Finnic stems: (10)

Vepsian: a. b. c. d. e.

do-panda do-soda i do-ajand pere-hüppähtada pere-panda

f. g.

do-r’ikot’he do-l’ämptab

Karelian h. i.

pere-katšella pere-pat’t’ie

‘place as far as’ ‘finish eating’ ‘did not reach’ ‘jump across’ ‘place across’ (examples a-e from Oinas 1958) ‘finished off, killed’ ‘finish warming up’ ‘look at or examine thoroughly’ ‘thoroughly, utterly’ (examples f-i identified by SSMPugh; see Pugh forthcoming [b])

The native Finnic system of verbal (suffixal) derivation is clearly intact and productive; one comment we could make here is that the full participation of a Russian lexeme in such word-formational processes signals the complete assimilation of that lexeme into the Karelian system (and is therefore certainly a borrowing, not a using). Perusal of Karjalan Kielen Sanakirja (the Dictionary of the Karelian Language) reveals that vast numbers of Russian-based forms, especially verbs, do take part in this process (see Pugh 1992). (11)

muani> a. muan betray

-i -SM

-kse-nd-ele-FREQ+DUR (compound suffix, incorporating the notion of ‘passivity’) ‘be betrayed many times, for a long time’

266 Stefan M. Pugh

>

b.

muan -i betray -SM ‘be betrayed’

-vu-PASS

>> c.

muan -i -vu -kse-nd-elebetray -SM -PASS -FREQ+DUR ‘be betrayed many times, for a long time’

>

d.

muan -i betray -SM ‘dupe, deceive’

>> e.

>

muan -i -tbetray -SM -CAUS ‘deceive many times’

-ele-FREQ

>> f.

>

muan -i -ta betray -SM -CAUS ‘deceive many times’

-kse-nd-ele-FREQ

-t-CAUS

These are just a few of the many possible permutations that have been generated in the Karelian dialects; note that a number of different suffixes can express the same notion from dialect to dialect — and even within the same dialect.

6. The present The Karelian people finds itself at a linguistic crossroads as we speak. Since 1991 there has been a serious move to establish Karelian (as well as the much more seriously endangered Vepsian language) as a literary language, for use in literature, the media, and the schools. Primers have been prepared, in two major Karelian dialects, and are being used as of the first year of schooling in the Karelian Republic. There are, of course, immense difficulties involved in these developments, not least of which is the presence of large numbers of Russian speakers throughout the republic; this is most true of the capital, Petrozavodsk, but also of the relatively large town of Olonets, where at least a good half of the population speaks Russian as their L1 (Karelian is still commonly heard on the streets, however). Beyond the Russian presence there are financial implications — resources are scarce, especially for largescale production of books for use in schools. It is also not enough for primers and other basic language books to be introduced for the instruction of Karelian: if Karelian is to survive (at least in the schools), then it will have to become the medium of instruction for all or most basic school subjects: science, maths, and so on. For this to occur there are not only financial implications to consider (which would be substantial) but also questions of staffing; teachers must be found, trained in Karelian if they are not fluent L1 speakers already, and placed in schools, from

The role of language contact in the formation of Karelian, past and present 267

primary school up through to the high schools. This is obviously a daunting task, and only time will tell if it will succeed. At this point, however, there are also linguistic concerns. As mentioned above, materials are being prepared for speakers of two main dialects of Karelian. This was obviously the choice that had to be made: (1) codify one variant for all, which would entail reconciling features of two variants with each other for the greater good; and (2) codify two variants so that speakers would feel that this language really was theirs, and therefore could and would be usable. The disadvantage of option (1) is that the codifiers would be creating a supradialectal literary language, in a form that is actually spoken by no one Karelian. The disadvantage of (2) is clearly one of resources: two efforts are progressing instead of one, with a smaller audience for each, with greater training implications for the schools; but the fact that each variant mirrors the living language for the groups in question must outweigh the negative factors. A large concern for the future of the Karelian literary language(s) lies in the development or acquisition of new lexical items in semantic fields for which the Karelian lexical pool is unable to produce native alternatives. In other words, language planners will have to settle on which prestige language to use as the lexical source of new borrowings. In the field of popular culture, this may well turn out to be English, as the study of English is growing in popularity in Karelia as it is in the rest of the former Soviet Union. But in other fields a decision will have to be made. Logic dictates that Russian could continue to be this source: it is spoken throughout Karelia, will continue to be used in education (especially higher education), it has always been such a source (as a fact of life as well as by design in later years), and retains prestige. The latter point is not insignificant, as the Karelians are not essentially anti-Russian — as the populace might be perceived to be in such newly independent republics as Estonia or Lithuania. But in developing Karelian society, Karelians now look west, to Finland. Finland gives Karelia assistance in many ways, but especially in the area of the economy (as far as they are able to contribute, given current financial constraints in Finland) and the development or maintenance of Karelian national identity. Finnish is a prestige language, with a comparatively long literary and cultural history; it has been one of the official languages of Karelia (off and on) over the last several decades; and, what is extremely important, it is closely related to Karelian. Thus, Finnish is an obvious and easy resource. In fact, recent Karelian-language newspaper articles have indeed reflected a tendency to rely on Finnish as such as a source language. However, more importantly, the lexical items used represent no new coinages in new or developing semantic fields: the Finnish items in question were actually replacing already existing lexemes used in the spoken language that had come into the language from Russian. This phenomenon is unfortunate from the point of view of the linguist. On the one hand, it will in the end make Karelian more ‘Finnic’, if not more ‘Finnish’;

268 Stefan M. Pugh

but on the other, the replacement of lexemes in use in the spoken language with lexemes in use in the written language can ultimately result in a separation between the written and spoken forms of the language. In fact, it is not impossible to foresee the development of a diglossic situation in the medium-range future. Should this tendency become policy in any profound manner, it would mean adding diglossia to what can currently be described as ‘stable bilingualism’ (in Karelian and Russian) among many Karelians; in addition, there is no reason to suppose that there will be an end to their use of Russian, given demographic and practical considerations. This might be a recipe for eventual linguistic disaster for the maintenance of the Karelian language.

7.

Modelling language contact

There are numerous ways in which we could create a theoretical model for language contact, with diagrammes and pictures. Here I would like to reflect the RussianKarelian situation over the centuries in a very simplistic way — simplistic and yet relevant. In the following chart (chronologically, from the earliest phase at the top) K = Karelian, R = Russian, F = Finnish; when appearing in upper case, this represents the strength of the given component — vitality, stability, and degree of influence in language contact situations. Two letters appearing together, e.g. in the case of Kr, means that K is still strong, while R is playing a role in the development of K; the fact that it appears in lower case suggests less than major influence, whereas an upper case letter would suggest strong interference. If they are separate, arrows indicate in which direction the interference (however minor) is progressing. 1. 2. 3. 4. 5. 6. 7.

Early Phase (before 17th century) 17th century–1917 1917–1930’s Early 1940’s Late 1940’s–1991 1991–present The future

K < < < < < R, but becoming Kr Kr, becoming KR KR, becoming KRf then KR again KRf kR KRf KrF literary language??? KRf spoken language???

Such a description cannot of course be entirely accurate, because it concentrates on general tendencies, ignoring the stability of Karelian among older generations in the villages. In general, however, it is valid as a global perception of the state of Karelian in terms of language contact. Phase 3 recognizes the strong russifying activities of the Soviet government vis-à-vis small minorities; the addition of the component F reflects the widespread use of Finnish in Karelia (in schools and public life) until 1938, when F was banned from use (as a ‘protective’ measure against bourgeois/

The role of language contact in the formation of Karelian, past and present 269

fascist Finland!). Phase 4 adds F again, during the war years and occupation by Finnish forces. The post-war attempt to reintroduce Finnish in any thorough fashion failed in part because of the complete destruction of Finnish-language resources in 1938, therefore Phase 5 has lost F; the weakened state of K, whether a result of the Second World War, migrations, or government policy, is reflected here as well. Phase 6 recognizes the limited influence of F in the post-Soviet period, consisting currently of the replacement of some R-based K lexemes by F ones. What the future will hold is certainly up to language planners, grammarians, government support for indigenous languages, and the willingness of the people to use and maintain their language — and to pass it on to successive generations. Phase 7 reflects the potential co-existence of two forms of the language: for the literary language, it represents a move to choose F as a source for future lexical expansion, replacing R as the prime giver-language but not entirely supplanting it; given the bilingual status of Karelians, and the presence (alluded to above) of large numbers of Russians in the Republic, this could not reasonably occur. The spoken language, on the other hand, could continue as it is at present: strong R influence, a stable K component, and some F influence. Were this two-fold development to take place, the situation would clearly come under the category ‘potential diglossia’, the ultimate results of which cannot be predicted, as there are so many variables in play. In terms of the Circum-Baltic languages as a whole, Karelian (as it is currently in the process of being codified) represents one that is truly in a state of flux, to a greater extent than any other language of this region.

References Austin, Paul M. 1992. Soviet Karelian: the Language that Failed, in Slavic Review, v. 51, n.1 (Spring, 1992), pp. 16–35. de Bray, R. G. A. 1980. Guide to the East Slavonic Languages (Guide to the Slavonic Languages, Third Edition, Revised and Expanded, Part 3). Slavica Publishers: Columbus, Ohio. Kalima, Jalo. 1955. Die slavischen Lehnwörter im Ostseefinnischen. Osteuropa-Institut an der Freien Universität Berlin, Slavistische Veröffentlichungen, Band 8, Berlin. Oinas, Felix J. 1958. Russian and Eastern Balto-Finnic Linguistic Contacts, American Contributions to the Fourth International Congress of Slavists. The Hague: Mouton, 267–78. Pugh, Stefan M. 1991. A Balto-Finnic Semantic Field in Motion. Semantic and Structural Aspects of Language Contact. Finnisch-Ugrische Forschungen. Vol. L (50), 1, pp. 1–19. Pugh, Stefan M. 1992. Balto-Finnic Verbal Derivation and the Assimilation of Russian Lexemes. Linguistica Uralica, XXVIII 2, 89–101. Pugh, Stefan M. 1994. The Fennicization of Russian Verbs in Karelian and Veps: a Question of Degree, Ural-Altaische Jahrbücher, Neue Folge, Band 13, pp. 45–55. Pugh, Stefan M. 1996. Vowel Harmony in Karelian and Veps: Rules for Russian Verbal Borrowings. Papers from the Fourth World Congress for Soviet and East European Studies (Harrogate-July 1990). Volume III in the Occasional Series: Papers in Slavonic Linguistics. Ed. JI Press and FE Knowles. London. 213–223.



270 Stefan M. Pugh

Pugh, Stefan M. (forthcoming [a]). Minority L1 Literacy in Sweden and Karelia: Challenges and Strategies. Acquisition of literacy in a multilingual context: A cross-cultural perspective, in press, Erlbaum. Pyöli, Raija. 1996. Venäläistyvä Aunuksenkarjala. Joensuu. Thomason, S. G. and T. Kaufman. 1991. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press.



Syntactic code-copying in Karaim Éva Ágnes Csató

1.

The Karaim language

For more than six hundred years, Karaim has been spoken as a community language in what is today Lithuania. Due to political measures taken by the postwar Soviet regime, the community is now dispersed and the maintenance of its language has become endangered. The number of Karaims in Lithuania is about two hundred, but only a fourth of them, mostly members of the eldest generation, still have a communicative competence in the language. Karaim belongs to the Kipchak branch of the Turkic language family, more closely to the sub-branch which also includes Karachay, Crimean Tatar and Kumyk. Three dialects of Karaim are distinguished. The eastern dialect, spoken in the Crimea until the beginning of this century, is now extinct. There are two western dialects, the northwestern Lithuanian dialect, which is often called the Troki (in Polish) or Trakai (in Lithuanian) dialect, and the southwestern dialect of Galicia in the modern Ukraine, is now spoken by only six Karaims in the town of Halich. The Lithuanian dialect shares a number of typological properties with cognate Turkic languages. It is also influenced by the non-Turkic dominant languages of the linguistic area. The adoption of new structural and lexical properties has led to significant convergence with the linguistic type dominating in the area. This study will focus on the nature of linguistic processes involved in contactinduced typological changes in general, and present the case of adpositional constructions in particular. My aim is to illustrate that processes of code-copying may introduce new grammatical patterns, i.e. that code-copying can lead to framechanging innovations at all levels of the grammar. For the following analysis, spoken material recorded recently in Lithuania has been analysed. All my informants speak the Lithuanian dialect, and their average age is above 50.1

2.

Processes of code-copying

Karaim speakers in Lithuania are multilingual, also having command of the regionally and super-regionally dominant languages: Lithuanian, Polish and Russian. A few

272 Éva Ágnes Csató

Karaims also have some fragmental knowledge of the religious language, Hebrew. The functional domain of the Karaim language is, however, restricted to everyday conversational situations in the family and with a few Karaim friends. Karaim also plays an important role in religious practice, since songs and prayers are both in Karaim and Hebrew. The speakers are trained in writing and reading Karaim, but their access to written literature is limited. Code-shifting is an everyday communicative habit of many speakers in multilingual Lithuania. The Karaim speakers would not be able to manage without it. Most families include speakers of different subsets of the dominant languages. The relatives living in Poland do not, for instance, speak Lithuanian but they understand Russian. The ones living in the Crimea understand neither Polish nor Lithuanian. Thus, the members of the families have to alternate from one code to the other according to the needs of a given speech situation, e.g. from Karaim to Polish, or from Polish to Russian. The processes I am going to discuss in this paper are not the ones involved in this type of code-shifting, i.e. shifting from one language to another, but those involved in code-copying, i.e., that a speaker of Karaim copies foreign linguistic items and properties into his Karaim speech. The notion of code-copying was introduced by Johanson (1993) and developed in several publications, see references for further titles. “The central concept of code-copying is that elements of one code are copied and the copies inserted into another code. Copies are not ‘switched’, not just juxtaposed to the elements of the other code, but adjusted to its structure in some way and to some degree, even if they are only used occasionally, as so-called ‘momentary copies’” (Johanson 1999: 39). The notation, code-copying, is used by Johanson instead of the terms “borrowing”, “interference” and “transfer” used by other authors. Partly because of their multilingual habits, and partly because of the insufficiency of the Karaim lexicon, Karaims have a well-developed habit of copying foreign lexical items. Examples (1) and (2) illustrate elements copied from Russian (< Russ) and Polish (< Pol) respectively. It is not in every case clear which Slavonic language served as the source language for copying. In such cases, the source will simply be called “Slavonic” (< Slav). (1) Nu to bašladï iš’l’˙am’˙a anar well then (< Slav) begin:past do:inf she:dat bu d’ixan’iye this respiration (< Russ) yattïrγïzdï yer üs’n’u, ˙ kuxn’˙ad’˙a lay down:DI-past floor on.postp kitchen (< Russ):loc ‘Well, then he started to give her artificial respiration, he laid her down on the floor, in the kitchen.’

Syntactic code-copying in Karaim 273

Vïzvat’ et’t’im skoranï, call (< Russ) DI-past.1sg ambulance (< Russ):acc k’el’m’it’. com:neg:A-nonpast ‘I called the ambulance, it doesn’t come.’ L’il’˙ag’˙a pazvont et’t’im, to Lilia:dat make a call (< Russ):DI-past.1sg to.part (< Slav) skoray k’el’d’i. ambulance (< Russ) come:DI-past ‘I called Lilia and the ambulance came.’ (2) Nu n’e maya γal’e avur bunu barïn, well (< Slav) what I:dat now difficult this:acc all:3poss.acc bunun ücˇ’un’ ˙ k’i t’uv’ ˙ ul’ ˙ barïn because that not all:3poss.acc m’en’ pam’entac’ et’˙am, I remember (< Pol): A-nonpast.1sg maya pl’ontac’ et’˙at’ barï. I:dat confuse (< Pol):A-nonpast all:3poss ‘Well, it is now difficult for me (to tell) it all, because I don’t remember everything, I confuse it all.’ Nu n’ecˇ’ik b’ir bala ošukac’ et’t’i well (< Slav) how a young man cheat (< Pol):DI-past b’ir tašcˇanï … Nu da ga]lax uže a girl:acc well (< Slav) and Catholic priest already (< Russ) spov’edac’ et’˙at’ alarnï, hear confession (< Pol):A-nonpast they:acc k’i pov’in’n’i ed’l’˙ar buley da buley … that ought to (< Pol):3pl be.cop:DI-past.3pl so and so ‘Well, how a young man gets a girl into trouble, and the Catholic priest hears their confession so that they ought to do so and so.’

The term “code-copying” is used here to refer to processes of creating a Karaim copy of a foreign lexical item or linguistic feature. The Karaim copy of the Slavonic lexical element is, however, never identical with the original item. It is, as a rule, both phonologically, grammatically and semantically adapted to the Karaim system. The adaptation is rule-governed. In the following example, the Karaim speaker uses a copy of the lexical unit zakuska ‘snack’ from Russian. (3) Iš’l’eyd’l’˙ar zakuskanï make:A-nonpast.3pl snack (< Russ):acc ‘They prepare the snack.’

The copied item, zakuska, is declined as a Karaim noun, and, consequently, the harmonic form of the accusative suffix, -nï, is attached to it. The copied item is

274 Éva Ágnes Csató

classified as a [−front] stem and the Karaim sound harmony rules are applied to it. This is a global copy, implying that both phonological material and certain grammatical, combinational, lexical and frequency properties are copied. Copying may also be selective, involving not all but only some selected aspects of a linguistic item. Example (4) includes an essive expression meaning ‘in the function of ’ that is formed in Karaim with the postposition -bA ‘with’. The use of this postposition with this function is the result of the selective copying of the syntactic function of the instrumental case which is used in corresponding Russian essive constructions. The function of the Russian case suffix but not the material form is copied selectively onto a Karaim postposition. The example also contains a global copy of the Russian expression vagonnï d’is’pecˇer ‘superintendent of waggons’. The selective copy and the global copy in this example together constitute a so-called mixed copy. (4) Iš’l’˙ad’im t’em’ir yolda, vagonnï work:past.1sg railway:loc waggon (< Russ adj) d’is’pecˇ’erbcˇ6. superintendent (< Russ):with.postp ‘I worked at the railway as superintendent of waggons.’

Syntactic patterns, for instance, the structure of a Polish copular sentence, may also be copied, as the following examples (5) and (6) illustrate. The predicate is introduced by the particle to in both the Polish and the Karaim sentences. (5) Polish Ten wysoki blondyn to mój brat. that tall blond man to.part my brother ‘That tall blond man is my brother.’ (Rothstein 1993: 729) (6) Karaim Bunda n’e jazγan, here what write:gan.part to astrï kart s’oz’l’˙ ˙ ar. to.part (< Slav) very old word:pl ‘The words which are written here are very old ones.’

Code-copying involves creative, productive and rule-governed processes. Codecopying strategies are a component of the language competence of Karaim speakers. Code-copying results in changes involving various aspects of the Karaim language system. The following discussion will discuss examples of contact-induced changes involving Karaim adpositional constructions.

Syntactic code-copying in Karaim 275

3.

Global and mixed copies of foreign prepositional syntax

Turkic languages exhibit postpositional phrases, which is in harmony with their typical basic SOV word order. Karaim also employs postpositions, as, for instance, üs’n’˙u ‘on’ in (7). (7) Cˇïpcˇïxlar olturadlar t’er’˙ak’l’˙ar üs’n’u. ˙ bird:pl sit:A-nonpast.pl fruit-tree:pl on.postp ‘Birds are sitting on the fruit-trees.’

The prepositional phrases used in Karaim are mostly copied from foreign languages. The following example (8) includes a global copy of a Slavic prepositional phrase, s Petrograda ‘from Petersburg’. (8) Troxta anuz yox ed’i Trakai:loc yet not existing be.cop:DI-past gramafon, a atam gramophone (< Slav) but father:1sg.poss k’el’t’ir’d’i s Petrograda bring:DI-past from.prep (< Slav) Petrograd (< Russ):gen gramafon. gramophone (< Slav) ‘There was no gramophone in Trakai, but my father brought a gramophone from Petersburg.’

Copied prepositions may also be used in mixed copies, i.e. in structures in which the copied preposition is used together with a Karaim noun. The combinational properties of the Polish preposition are copied onto the Karaim construction. Consequently, in the following example, the noun Basia (girl’s name) is in the genitive. (9) I uže bu fotograf and already (< Russ) this photographer (< Slav) turat kolo Bas’yanïn. stand:A-nonpast by (< Pol) Basia:gen ‘And this photographer is already standing near Basia.’

The mixed copy in example (10) contains the global copy of the Slavic preposition za ‘for’. The whole expression is a selective copy of an idiomatic Slavic expression meaning ‘What kind of …is it?’. (10) N’e bu za karay? what this for.prep (< Russ) Karaim ‘What kind of Karaim is this?’

As the above examples demonstrate, copied prepositions and prepositional phrases occur in Karaim. The use of such copies has not, however, changed the original typological property of Karaim. It has remained a predominantly postpositional language.

276 Éva Ágnes Csató

4. Disharmonic properties of the Karaim basic word order The other originally postpositional language spoken in the Baltic area, Estonian, has also preserved its predominantly postpositional nature. The resistance of these languages to change from postpositional to prepositional syntax is interesting given the fact that so many of their basic syntactic properties have been changed. In a previous study (Csató 1996), I have discussed syntactic changes in Karaim and concluded that its original Turkic SOV order has been changed to SVO. This might already have taken place before the Karaims settled in the Baltic area. Karaim has today word-order properties similar to those of the dominant languages spoken in Lithuania. It has adopted the word order noun+genitive (NG) as a free variant of the original genitive+noun (GN) order. In some stylistic variants, mainly in pre-war written language, the NG order may even be dominant. (11) Lithuanian Russian Polish Karaim

SVO SVO (free variation) SVO (free variation) SVO

Prep GN Prep NG Prep NG Postp/(Prep) NG/GN

AN AN AN/NA AN/(NA)

The Postp, NG, AN order in Karaim contradicts a word-order universal suggested by Greenberg (1966 and 1980), according to which prepositional languages have NG and postpositional languages GN as the basic order. The combination Postp+NG may, however, occur if and only if the adjective-noun order in the language is NA, thus Postp fi (NA fi NG). Karaim seems, however, to contradict this assumption, since it has a basic AN order, the NA order having a very low frequency. Copies of foreign prepositions and prepositional phrases are used in Karaim. One would expect that these copies would change the Karaim adpositional syntax. The fact that the use of postpositions is disharmonic with the other syntactic properties of Karaim gives further reason to assume that this component of Karaim syntax has undergone changes. In the following, I will present arguments supporting the assumption that the syntax of postpositional constructions has been changed in Karaim. The path of syntactic change in Karaim is, however, not postposition fi preposition, but rather noun-like postposition fi particle-like postpositions. Particle-like postpositions may be cliticised and thus come to constitute a phonological word together with the noun. This path is well documented cross-linguistically. There is historical evidence that case suffixes have developed from postpositions in several languages, for instance Hungarian.

Syntactic code-copying in Karaim 277

5.

The effect of selective copying on the syntactic properties of Karaim postpositions

5.1 Turkish postpositional constructions A closer look at the syntax of Karaim postpositional phrases reveals that the dominant type of postpositional phrases in Karaim is the one that is most compatible with SVO constituent ordering. I will first present the types of postpositional phrases typically found in Turkic languages by giving examples from Turkish and then argue that some constructions are more compatible with SVO word order than others. Several types of postpositional constructions can be distinguished in Turkish. I will present here two of these that have particular relevance with respect to the Karaim data. The first, very frequent, type has a structure that looks similar to possessive constructions. In spite of the similarities, there are significant differences between the two constructions (see Johanson 1974). Compare the postpositional structure in (12a) to the possessive construction in (12b). (12) Turkish a. ev(in) içinde house(:gen) in:poss.3:loc ‘in a/the house’ b. evin içi house:gen inside:3poss ‘the inside of the house’

The Turkish postposition içinde ‘in’ acts as the syntactic head of the nominal construction in (12a). The adverbial case-suffix (locative, dative or ablative) required by the semantics of the matrix construction is attached to this head. It also bears a possessive suffix agreeing with the attributed noun. The noun, ev ‘house’ can be assigned the genitive, as is usual in possessive constructions, but it can also be absent, which is an option only in postpositional constructions. This case-marking of postpositional phrases is an indicator of the altered relation between the postposition and the nominal category. Note that the structure ev içi [house inside:3poss] is a compound in Turkish. No syntactic category may intervene in either the postpositional construction without genitive-marking, e.g. ev içinde [house inside:3poss.loc], or in compounds, such as ev içi [house inside:3poss]. While the genitive-marked noun evin may take a postverbal position outside the postpositional phrase, ev without genitive-marking has a fixed position. Thus, whereas the postposition behaves as a nominal head in this type of postpositional constructions, there is a syntactic cohesion between the postposition and the attributed nominal category that is characteristic of nominal compounds. The postpositions used in the second type of postpositional constructions are morphologically unanalysable except for any synchronically segmentable possessive

278 Éva Ágnes Csató

or case suffix. Whereas the class of the first, noun-like postpositions is relatively open, this second type of particle-like postpositions constitutes a strictly closed class in Turkish, including only a few postpositions such as, için ‘for’, gibi ‘like’. Nouns may never be assigned the genitive when followed by a particle-like postposition, e.g. ev gibi ‘like a house’ and never *evin gibi [ev:gen like].2 No syntactic element may, as a rule, intervene between the noun and the postposition. Whereas a nounlike postposition may constitute a postpositional phrase on its own, i.e. without an overt attributed noun, a particle-like postposition may occur only together with a nominal category. Thus içinde [inside:3poss.loc] ‘in it’ is a well-formed expression but *Gibi ‘Like’ is normally not. The syntactic function of a particle-like postposition is to anchor a nominal phrase or a clause syntactically in the matrix construction, a function shared with case suffixes. The similarity to case-suffixes is even more conspicuous in the case of the postposition ile ‘with’, which can be cliticised to the noun and obey the morphophonological rules of sound harmony, e.g. babamla [father:poss.1sg.with] ‘with my father’, trenle [train:with] ‘by train’. Johanson (1974) observes that, in Turkic, the order of determination at the syntactic level is just the opposite of the order of determination at the morphological level. A left-branching syntax is based on a right-branching morphology.3 Types of postpositional constructions can be characterised as being closer to left-branching syntactic structures or to right-branching morphological ones. A scale may be defined by setting left-branching postpositional phrases headed by noun-like postpositions on the one extreme and right-branching postpositional phrases based on particle-like postpositions on the other. A number of intermediary sub-types, not demonstrated here, may be placed on the scale between these two extremes. Karaim has acquired right-branching syntactic properties, as demonstrated above. The left-branching properties of postpositional constructions based on noun-like postpositions are not in harmony with this new SOV order. Thus, we would expect Karaim to raise the frequency of the second type of postpositional constructions, which is not in conflict with the acquired right-branching principle. In the following, I will argue that this is actually what happens in Karaim. 5.2 Karaim postpositional constructions The dominant type of postpositional construction in Karaim is the one in which the case marking of the postposition is either standardised or missing. In addition to the postposition bïla ‘with’, the postposition usnu/üs’n’˙u ‘on, on to’ can also be cliticised to the nominal stem. They both have harmonic variants. See, for instance, the examples in (13). (13) a.

B’ir k’iš’i d’˙a n’erv’˙al’˙ar üs’n’u˙ olturmït. a person also nerve:pl on sit:neg:A-nonpast ‘Nobody annoys me [sits on my nerves].’

Syntactic code-copying in Karaim 279

b.

M’en’ macˇ’i k’ib’ik aγacˇ usnu I cat like tree on m’in’˙al’m’im. climb:pot.neg:A-nonpast.1sg ‘I cannot climb a tree like a cat.’

The form of this postposition is standardised. It is invariable in every person. In the following example (14), the postposition üs’t’˙un’˙a ‘at’ carries the fossilised third-person suffix in the expression m’en’im üs’t’˙un’˙a ‘at me’, and thus does not agree in person with the genitive attribute m’en’im [I:gen]. The postposition, which lacks a case suffix, can be used both in a locative meaning, ‘on’, and in a directional meaning, ‘onto’. Some postpositions may optionally bear a case suffix, as in example (14). The form without case suffix is, however, more frequent in spoken language. (14) S’en’ m’en’im üs’t’un’˙ ˙ a/ üs’n’u˙ you I:gen on:dat on baγïnmïs. look at:neg:A-nonpast.2sg ‘You are not looking at me.’

A dative case suffix may be an invariable part of the lexicalised form of certain postpositions. The postposition t’˙ub’˙un’˙a ‘under’ contains both a possessive and a dative suffix [under:( ej; (iii) (regional) raising of au > ou; (iv) (regional) half-fronting of ou > oj. Three of the four features (ii, iii, and iv) are qualified as “regional”, i.e., as occurring only in a subregion of the broader NEY dialect area; the remaining feature, (i) made only partial inroads (phonologically) into zameter yidish [ZY]/CourlYid (see below). Thus, the more Yiddish approaches the Baltic, the more we see the maintenance of the earlier state of affairs (retention of vowel length, of rounded øy, etc.). However, in spite of the clear differences between northern (i.e., closer to the Baltic) NEY and non-northern NEY, the two sub-areas do share much common development, and Yiddish linguistic scholarship is justified in linking these to a common earlier NEY stage. The historical changes in the vowel systems across the Yiddish map are Yiddish-internal developments in this area, reflecting Yiddish-internal topography, rather than external contact. Herzog et al. (1992: 41) write: “In large parts of NEY and SEY, [ji] is a free or contextual variant of word-initial /i/.” One also frequently finds in NEY (and SEY) that StY jid ‘Jew’, jing6r ‘younger’ are realized as NEY id, ing6r. In most NEY, the insertion of j as a hiatus breaker is a general process; cf. StY r6fu6 ‘remedy’, to6s ‘mistake’, NEY r6fuj6, toj6s (see Herzog 1965: 209–210; Herzog et al. 1992: 22). Note that glide j thus often reappears as a hiatus breaker in NEY: id ‘Jew’, but di + idn as [di-j-idn] ‘the Jews’. Much of NEY also generally inserts a glide between a or e and a following velar nasal; thus, StY bre]g6n ‘to bring’, la]g ‘long’, NEY brej]g6n, laj]g (Herzog et al. 1992: 22). In the consonantism the broader NEY speech territory shared the historical confusion of hushers and hissers (š–s, ž–z, tš–ts, etc.) — the so-called sabesdiker losn ‘Sabbath speech’ (based on the historical presence of hushers in the words realized in StY as šab6s ‘Sabbath’, lošn ‘language, speech, tongue’). This feature was under

Yiddish in the Baltic region 291

retreat throughout the broader NEY area (though to varying degrees; additionally, the manifestation and description of sabesdiker losn varied in different NEY subareas; see Mark 1951: 436–438), under the influence of emerging StY (U. Weinreich 1952). Weinreich (1964a) showed the independence of Yiddish sabesdiker losn from a similar phenomenon in coterritorial languages (e.g., Polish regional mazurzenie). Lithuanian and Latvian maintain a clear distinction between hushers and hissers; thus, the presence of the (retreating) feature of sabesdiker losn in BY is not to be attributed to CB influence.13 The voicing sandhi whereby a sequence of obstruents takes the voicing of the last obstruent in the sequence (e.g., fus ‘foot, leg’ + benkl ‘bench’ Æ fu[z]benkl ‘footstool’) is attributed to Slavic influence (U. Weinreich 1958: 7), and is found generally across the NEY region. Thus — to the degree that this feature is found in BY — we are not justified in claiming a specifically CB source, in spite of the fact that this feature is present in Lithuanian (and Latvian). NEY generally preserves obstruent voicing distinctions word-finally (Mark 1951: 434–435), e.g., vajz ‘(I) show’, vajs ‘white’. 4.2.2Morphology and syntax Perhaps most striking in NEY vis-à-vis other Yiddish varieties is the restructuring from an earlier three-gender system (masculine, feminine, neuter) to a system lacking the neuter gender. Early treatments attributed the rise of the “two-gender” system in NEY to contact14 with Lithuanian and/or Belarusian.15 The Baltic languages coterritorial with BY — Lithuanian and Latvian — have two-gender systems — masculine and feminine.16 The gender system of modern Lithuanian (Mathiassen 1996: 37ff) is heavily semantically-based; for example, biological sex overrrides morphology in gender assignment. A residue of an older neuter is found in certain impersonal constructions. Furthermore, modern Lithuanian contains classes of nouns which only occur in the singular (singularia tantum) or the plural (pluralia tantum). The class of singularia tantum includes abstract nouns like ‘love’, ‘conscience’; collectives like ‘youth’, ‘intelligentsia’; materials like ‘iron’, ‘vodka’, ‘honey’, and names like Lietuvà, Vìlnius. The class of pluralia tantum includes names of ceremonies and feasts, many mass nouns like ‘paint’, ‘flour’, ‘gas’, compass directions, and more. Mathiassen (1996: 51) finds noteworthy the high number of mass nouns in the pluralia tantum group, and writes: “The reasons for this development remain obscure.” Some parallels in the NEY gender system may possibly be relevant to discussion of the Lithuanian pluralia tantum group; see also Lemkhen (1995: 44–45). In three-gender StY, nouns are assigned (in the singular) a gender — masculine, feminine, or neuter. There is definite article/adjective agreement for all cases. All plural noun phrases show (definite article di and) adjective suffix -6.

292 Neil G. Jacobs

Standard Yiddish definite article and adjective suffix

nom acc dat

masc

fem

neut

pl

der -6r dem -n dem -n

di -6 di -6 der -6r

dos -6 dos -6 dem -n

di -6 di -6 di -6

Allomorphs of adjective suffix {-n} are -n, -6n, and -6m. Positional variants of the -n are [n], [m], and []]. The uniform (singular) indefinite article is a/an for all genders and cases. With one exception, the substitution of the definite article by the indefinite article leaves the adjective suffix unchanged; thus: der gut6r man ‘the good man’, a gut6r man ‘a good man’. The exception is with neuters; here, the adjective ending is zero in indefinite NPs: dos šejn6 lid ‘the pretty song’, a šejn lid ‘a pretty song’. The indefinite article in the plural is zero; thus: di šejn6 lid6r ‘the pretty songs’, šejn6 lid6r ‘pretty songs’. Additionally, StY allows contraction of the masculine and neuter definite article dem to -n after a preposition; thus, mit ‘with’, af ‘on’, far ‘for’, etc. + dem Æ mitn, farn, afn, etc. With the prepositions in ‘in’, and fun ‘of, from’ the contracted forms are in6m, fun6m (or in–ø, fun–ø). With the prepositions ba ‘by’ and tsu ‘to’, the contracted forms are bam, tsum. Note that it is only the post-prepositional dem which may contract; thus, in StY only masculine and neuter NPs may show contraction, a point which is relevant to our discussion of the NEY system. U. Weinreich (1961) showed that the new NEY gender system that arose was more than simply a “two-gender system.” The new system contained masculine, feminine, mass nouns (vs. regular count nouns), in addition to regular plurals. In addition, masculine and feminine are further subdivided semantically into ±male, ±female, and/or morphologically marked/unmarked. Nouns which lack both semantic maleness/femaleness and morphological marking are either intermediate masculine or intermediate feminine.17 The pairings of definite article + adjective suffix in NEY are like those in StY: der -6r, di -6, dem -n (with no dos -6 pair). However, two points should be noted. First, NEY generally shows a reduction of the older (sg) acc-dat > obl. Mark (1951: 454) calls this combined case the objective case. He claims that in the pronouns the dative forms have survived, while in the definite articles the older accusative forms have survived. However, the situation is less clear-cut across the Yiddish map, as seen from Wolf ’s (1969) treatment of the relationship of gender and case systems. Second, NEY allows a contracted form even with intermediate feminines (as long as there is no adjective in the NP), e.g., di gas ‘the street’, afn gas ‘on the street’ (but: af der brejt6r gas ‘on the broad street’). Thus, the contracted intermediate feminine NPs allow a door for the reinterpretation of the gender of contracted

Yiddish in the Baltic region 293

masculines and neuters as intermediate feminine nouns, and of intermediate feminine nouns as (intermediate) masculine nouns. Suggestions are made that relate the origin of the neuterless gender system of NEY to a borrowed rule of case/gender syncretism in the Polish loc.sg, surfacing in NEY as the contracted form of prep + def (e.g., NEY intermediate feminineafn gas, masculine afn tiš ‘on the table’); see U. Weinreich 1961; Herzog 1965: 132). In light of Wolf ’s (1969) findings, suggestions of a Baltic source for this NEY development seem even weaker, since the Baltic languages in question (Lithuanian and Latvian, as well as Estonian) are rich in cases and case marking; rather, contact with Polish/Slavic is the preferable suggestion. Many of the structural changes in the NEY gender system can be shown to be internally motivated (U. Weinreich 1961; Herzog 1965: 124; Wolf 1969; Jacobs 1990a). While some parallelism exists between new gender assignment in varieties of NEY and the Lithuanian equivalents (see Mark 1944: 90; U. Weinreich 1961, cited in Herzog 1965: 122), there is also a substantial corpus of examples where the newly-assigned gender in NEY π Lithuanian gender. However, the influences of language contact — Yiddish and Lithuanian — are not to be ruled out as contributory factors in the development of the NEY gender system (U. Weinreich 1958: 12), particularly in the treatment of mass nouns,18 which appear similar to a sub-class of the Lithuanian pluralia tantum-substances, e.g., ‘water’, ‘jam’, ‘bread’, etc. Characteristic for the mass nouns in NEY is a plural NP, but a third person singular for the verb. Thus, in di gelt ‘the money, the definite article remains di after a preposition: mit di gelt. However, the verb is in the singluar: di gelt ligt afn tiš ‘the money is lying on the table’. In Jacobs (1990a: 97) the NEY gender system is represented as a series of flowchart decisions, as follows: 1. Is the noun [+count]? If NO, then assign it mass gender: di vas6r ‘the water’, di brojt ‘the bread, etc. 2. If YES ([+count]), is the noun [+singular]? If NO, then assign it noun-plural status. 3. If YES (i.e., noun is [+singular]), is it marked for biological sex [+sexus]? If NO, then ask a follow-up question: Does the noun contain any formal marking (phonological/morphological)? If NO, then assign to the class of EMPTY nouns, which ultimately receive intermediate masculine or intermediate feminine gender; here, the gender assignment is ad hoc, and open to regional (within the NEY area) gender reinterpretation. If, however, the [−sexus] noun does contain formal marking, then use that as a non-ad hoc basis for gender assignment, e.g., nouns ending in -6r are assigned masculine gender, nouns ending in -6 are assigned feminine gender. 4. If, however, the noun is [+sexus], then gender assignment is based on the biological sex, overruling noun-endings; thus: zejd6 ‘grandfather’ is masculine, in spite of its ending in 6, and švest6r ‘sister’ is feminine, in spite of its ending in 6r.

294 Neil G. Jacobs

For further details of the NEY system, the reader is referred to Jacobs (1990a), and the works cited therein. Yiddish has a present-participle construction consisting of verb “infinitive” + suffix -dik, e.g., laxn ‘to laugh’ + dik Æ laxndik ‘laughing’, redn ‘to speak’ + dik Æ redndik ‘speaking, talking’.19 When the verb takes the reflexive pronoun zix,20 as in, e.g., vašn zix ‘to wash (oneself)’, we observe a partial distinction between NEY and general EY. In the construction X-n-dik, non-Lithuanian EY shows either vašndik zix or zix vašndik, while Lithuanian Yiddish allows vašnzixdik. In non-NEY the suffix dik is appended directly to the verb form, forming a present participle; the reflexive pronoun zix may then either precede or follow the present participle. In NEY, however, the zix may be suffixed directly as part of the verb, to which the suffix dik is then attached. Mark (1951: 457) attributes this NEY construction to influence from Lithuanian. In Lithuanian gerund constructions the reflexive particle is inserted between the (last) prefix and the verb root (Mathiassen 1996: 154); this may have served as the model for the NEY innovation of laxnzixdik, bodnzixdik ‘while bathing (oneself)’, etc.21 NEY permits use of the present participle form to express “while X-ing”, whether referring to the subject of the main sentence or the subject of the embedded sentence (Mark 1951: 457). The distinction relies here on position in the sentence; thus: NEY (1) gej6ndik hob ix im g6zen walking have I him seen ‘I saw him (while I was) walking’ (2) ix hob im g6zen gej6ndik I have him seen walking ‘I saw him (while he was) walking’

Non-NEY, on the other hand, uses the gej6ndik form to refer only to the subject of the main sentence (I). When the walking is performed by the subject of the embedded sentence [= object of the main sentence] (he/him), this is expressed either by use of the infinitive gejn, or by a circumlocution (Mark 1951: 457): Non-NEY: (3) gej6ndik hob ix im g6zen walking have I him seen ‘I saw him (while I was) walking’ (4) a.

b.

ix hob im g6zen gejn I have him seen go.inf ‘I saw him (while he was) walking’ ix hob g6zen, vi er gejt22 I have seen how he goes

Yiddish in the Baltic region 295

Further, Mark (1951: 457) notes that the Yiddish literary language may express relative time of action by using either the present participle form (arajnkum6ndik) or the past participle form (arajng6kum6n): (5) arajn- kum6n- dik in štub, hob ix im derzen into come ing into room have I him pfv-seen ‘(while) entering the room I caught sight of him’ (6) arajn- g6kum6n in štub, hob ix im derzen into come[past.part] into room have I him pfv-seen ‘having entered the room, I caught sight of him’

Mark claims that in NEY (and in the non-literary language outside the NEY area) only the present participle form is used, and no distinction in time of action is possible; thus, arajnkum6ndik is used to express both present and past action. Here, NEY does not show influence from Lithuanian, where relative time of action between main and subordinate clause is marked in the gerund (Mathiassen 1996: 154).23 Historically, Yiddish forms the past tense by use of an auxiliary — a form of either hobn ‘to have’ or zajn ‘to be’ — plus the past participle of the main verb. There is a strong tendency in NEY to use the verb hobn ‘to have’ as the universal auxiliary verb in past tense constructions (Mark 1951: 457): NEY ix hob g6gang6n ‘I went’, ix hob g6zung6n ‘I sang’; cf. StY ix bin g6gang6n, ix hob g6zung6n. No suggestions of areal influences are made here, even though Swedish has undergone a similar change to universal use of ‘have’ as the auxiliary verb in past tense constructions (cf. English, as well). 4.3 Subdividing NEY While there is reason to speak of a broad NEY dialect, there is likewise reason to subdivide NEY. Thus, for example, U. Weinreich (1969) describes the situation of Yiddish in Belarus, as distinct from other NEY. Wolf (1969) demarcates a northern area from other NEY. This northern area — mentioned by others in the Yiddish linguistic literature as well — is what we might call here Baltic Yiddish. The term is justified as a shorthand notation; however, it is not clear that the term is justified as a separate classification. Mark’s (1951: 432) discussion of NEY distinguishes among Samogitian Yiddish — zameter yidish [ZY], “plain” NEY — Stam-Litvish, and Suvalker Yiddish. Mark (1951) will serve as the point of departure for the current discussion. Using Mark (1951: 440), we may summarize the general picture as follows. The main subdivisdion in NEY is between ZY to the west and Stam-Litvish to the east. ZY is bounded by Germany (actually, East Prussia) on the west, Courland to the north, Stam-Litvish to the east, and Suvalker Yiddish to the south. Mark (1951: 441)

296 Neil G. Jacobs

claims that the lines of demarcation between ZY and its neighbors are clear to the north, west, and south. To the east, there is a gradual transition from ZY to StamLitvish. Courland Yiddish (CourlYid) to the north is seen as a branch of ZY. Suvalker Yiddish to the south is basically Belarusian Yiddish, and is linked historically to the influences of the Yiddish of Grodne/Grodno.24 In subdividing NEY, Mark (1951:439ff) states that Semiticisms are more evident in ZY than elsewhere in NEY (and, as noted earlier, generally higher in NEY than in other EY). Mark also claims that ZY contains more Lithuanian loan words than does other NEY (see also Lemkhen 1995: 12). Thus, before we get into discussion of concrete linguistic elements and features, it can be noted that the NEY dialect area seems to subdivide culturally into two main areas: Zamet, and “not-Zamet”.25 In distinguishing the two culture sub-areas, Zamet and not-Zamet, it should be cautioned not to look in the first hand to an external, non-Jewish explanation. As Mark (1951: 447) points out, there is no direct parallelism between the main dialect subdivisions of Yiddish with those of the coterritorial dialects of Lithuanian. The Yiddish name Zamet corresponds to the Lithuanian Žemaitija. However, the geographic territories covered by Jewish Zamet and Lithuanian Žemaitija are not identical (though for conflicting views on the relative sizes of the respective areas, see Mark 1951: 447, and Lemkhen 1995: 12). In terms of Yiddish linguistics, ZY is a conservative, “older” type of NEY, while Samogitian Lithuanian is an innovative, “younger” dialect of Lithuanian. Furthermore, for the most part there is a lack of areal correspondence of specific linguistic features in Samogitian Lithuanian and Zameter Yiddish. For example, in some varieties of Samogitian Lithuanian original diphthongs > long monophthongs;26 ZY, on the other hand, has richly preserved original diphthongs (Mark 1951: 447).27 Lemkhen (1995: 21) does provide some examples of Samogitian Lithuanian loans in ZY; however, once having entered Yiddish, the geographic or social diffusion of Lithuanian loans is no longer constrained by the dynamics of Lithuanian linguistic geography, but rather, may diffuse along Yiddish linguistic-geographic pathways. Ashkenazic geography in general, and Yiddish linguistic geography in particular, must be looked at on their own terms, and not as subsets of the coterritorial non-Jewish cultural or linguistic geography (see Mark 1951: 450; M. Weinreich 1980: 47). When dealing with notions of contact, we must distinguish between internal and external contact. Thus, note Mark’s (1951: 442–443) general impression that (as far as Yiddish-Yiddish contact is concerned) ZY speakers for many generations had contact only with other NEY — in this instance, stam-litvishspeakers; the latter, on the other hand, were in ongoing contact not only with ZY speakers, but also with speakers of Polish Yiddish and Belarusian Yiddish. As regards external contact, Mark (1951: 442) notes that Stam-Litvish contains many more Slavic-origin words than does ZY, especially from Polish; e.g.:

Yiddish in the Baltic region 297

Stam-Litvish

ZY

gloss

past6x (< Slavic) paš6 (< Slavic) štšur (< Slavic) pozetsk6s/porzetsk6s (< Sl) (~ vajmp6rl6x < Germanic)

šef6r (< Germanic) vajd (< Germanic) rats (< Germanic) vajmp6rl6x (< Germanic)

‘shepherd’ ‘pasture’ ‘rat’ ‘currants’

Furthermore, Mark (1951: 450) notes that there is much more Slavic influence found in the Yiddish of Belarus (presumably, coterritorial Slavic only) than in Lithuania (historically, Yiddish in contact with both Baltic and Slavic, largely, Polish). The topography of NEY is shaped both by internal and external contact. We must keep in mind that even with our subdivision of NEY, there was an ongoing dialect levelling taking place. In general, urbanization, along with other political, social, national, and economic factors in the 19th and 20th cc. led to the loss of many older dialect forms in EY (Herzog 1964). For NEY, Mark (1951: 449–450) writes of four historical factors which contributed to dialect levelling in NEY: (1) the two major road-building projects of the 19th c.; (2) the spread of rail communications; (3) export business to Prussia brought in Jews from the eastern part of Lithuania; (4) major urbanization from shtetlekh ‘market towns’ to cities (e.g., Kovne/Kaunas, Shavl/Šiauliai, Ponevezh/Paneve˙žys) within the two most recent generations (pre-World War II). Thus, the face of NEY was undergoing significant change from the end of the 19th c. into the early 20th c., with consolidation of local dialects toward a broader, more generalized NEY (along with the tendency toward emerging StY). Mark (1951: 441) notes that ZY was once spoken over a wider area, and, in the decades before World War II, broader NEY gained ground. Mark (1951) provides a number of examples of features which distinguish varieties of NEY. We will focus at this point mainly on the features which distinguish ZY from Stam-Litvish. Page reference to Mark is given in the form: [YM: 439]. 1. Compared with Stam-Litvish, ZY largely retains the distinction between long and short monophthongs [YM: 439]: ZY bi˜n ‘bee’ — bin ‘am’, zu˜n ‘son’ — zun ‘sun’. Distinctive vowel length is also found in Lithuanian and Latvian (and Estonian), as well as in Baltic German. This might suggest that the coterritorial languages played a role in the maintenance of long vowels in northernmost NEY; however, contact with the coterritorial languages does not account for the instability of long vowels in CourlYid (see §4.5, below). 2. Vowels are frequently longer in stressed, open syllable in ZY [YM: 439–440]: mam6 ‘mother’, lib6 ‘love’. 3. Reduced or cliticized form of 3sg.fut marker [YM: 439–440]: in ZY -at, in Stam-Litvish -et (cf. non-reduced StY vet, reduced form -(6)t. Mark gives the

298 Neil G. Jacobs

sentences: ZY az m’at lebn, ‘at’n zen, Stam-Litvish az m’et lebn, ‘et’n zen ‘if one shall live, one shall see.’ 4. Survival of r-coloring in second-degree dim suffix28 in ZY mejdal6 vs. StamLitvish mejd6l6 ‘little girl’, from base form NEY mejd (StY mojd) ‘maiden, lass’. Note also lowering of schwa before x in Stam-Litvish mejdlax vs. ZY mejdl6x. 5. Maintenance of vowel color in clitic form of du ‘you’ in ZY, versus reduction to schwa in Stam-Litvish: ZY štejstu ‘you stand’ vs. Stam-Litvish štejst6. 6. Maintenance of diphthongal back/round glide u in ZY vs. tendency toward universal glide j in Stam-Litvish; thus: ZY

Stam-Litvish

gloss

heux feugl (cf. ZY plural [fejgl] boux

hejx fejgl bujx

‘high’ ‘bird’ ‘belly’

The presence of a back/round glide in Yiddish dialects correlates positively with the presence of phonemic long monophthongs; see Jacobs (1994b: 137). Thus, the presence of the back/round glide in ZY, and its absence in Stam-Litvish, adds support to the analysis of ZY as a dialect possessing phonemic vowel length, and to the analysis of Stam-Litvish as lengthless.29 4.4 Lithuanian influences in NEY 4.4.1Mark (1951) Mark (1951: 465ff) devotes discussion to Lithuanian influences in NEY. He claims that there are almost one thousand Lithuanian loanwords in NEY, and provides a partial list.30 Mark emphasizes social factors in Lithuanian-Yiddish language contact: Yiddish speakers in villages in contact with Lithuanian-speaking farmers, along with some contact among shtetl (market town) Jews with Lithuanian farmers who came into the shtetlekh. The shtetlekh also had their own Lithuanian-speaking element. Mark (1951: 466) writes that Lithuanian loan words in Yiddish are found in several semantic spheres, and do not lend themselves to precise summary; however, he claims that most of all these come from dorf (village) life, and heavily from daily life. Some examples from Mark’s (1951: 466ff) partial list of Lithuanian loans found in NEY include:

Yiddish in the Baltic region 299

NEY

Lithuanian

Yiddish gloss

pipk6n6n bug6 bájd6n6n zix gájš6n6n ~ gájs6n6n

pypkiúoti bugù baidýtis (su)gaı˜šti, (su)gaišìnti

‘to smoke, giving off much smoke’ ‘terror, fear’ ‘to be frightened (animal term)’ ‘to waste time’

The amount of Lithuanian loan words varied with the amount of contact, and many loans were localized — i.e., limited to one specific sub-area, and not known throughout Jewish Lite (Mark 1951: 465–466). Mark (1951: 466) provides the following examples of Lithuanian loans which are socially or geographically limited in their usage in Yiddish: only older (in 1951) Jews from Kovne and Yanev understood matréng6 ‘a flood’; only a true Zameter understood bukst6 ‘frightened (a horse term)’; possibly only a Jew from Palonge/Palangà called the wind from the north sóm6n6s, the wind from the south sáks6n6s;31 only cattle-cunning Jews understood grém6l6 (farlirn) ‘to cease chewing the cud’ or krék6n6s ‘first milk after calving’ < Lithuanian krekenà. Other scholars before Mark have noted Lithuanianisms in NEY. Prilutski (1924: 239) provides a small note on Lithuanian loans in Yiddish, recorded from the teacher Khane Rozin from Shavl; his list includes: Shavler Yid

StY

Lithuanian

gloss

set6n6 tak6 pelk6 šlaj6 plik6 gandriš šmag6n6s

brukv6 štegl kaluž6 a grojs6r šlitn nak6t bušl moj6x

se˙tin˜ys tãkas pélke˙ šlãjos plìkas gandras smãgenys33

‘turnip’ ‘footpath’ ‘puddle’ ‘a big sled’ ‘naked, bare’ ‘stork’32 ‘brains, mind’

However, we can go further than merely noting the presence of Lithuanian loans in NEY. Mark (1951: 466) notes that Lithuanianisms in Yiddish: (a) assimilate phonetically to Yiddish; (b) follow the grammar of the (Yiddish) dialect; (c) are adapted/integrated in idiomatic expressions; and (d) are thus Yiddish words. Thus, of deeper interest to us are the nature and degree of integratedness of Lithuanian influences in NEY. 4.4.2Lemkhen 1995 Lemkhen (1995) takes us significantly further in the discussion of Lithuanian influences upon Yiddish.34 However, even in Lemkhen’s detailed study the bulk of the Lithuanian influences to be found are lexical; pages 57–120 are devoted to a list

300 Neil G. Jacobs

of Lithuanian loan words in LithYid. Lemkhen (1995: 18) dates the appearance of Lithuanian loans in Yiddish to the 18th c., based on consideration of these loans in Yiddish in Lithuania, as well as in the Yiddish of Courland and Estonia. Like Mark (1951), Lemkhen (1995: 42) emphasizes the relatively large number of Lithuanian loan words in Jewish Zamet (as well as in the areas around Shavl, Ponevezh, and Birzh-Posvel). Lemkhen (ibid.) also gives examples of Lithuanian dialect variants in NEY, e.g., ZY/CourlYid kleb6 < Samogitian Lith kle˙b~ y s, non-ZY gleb6 ‘armful’ < Aukštaitian Lith gle˙b~ y s. This example shows a parallelism in this particular word between Jewish and non-Jewish Samogitia. Lemkhen (1995: 42) states that the variation in the Lithuanian loans within Yiddish is attributed to Yiddish borrowing from different Lithuanian dialects; however, once in Yiddish, the loan may diffuse beyond the geographic bounds of its Lithuanian source dialect. 4.4.2.1 Phonology. Lemkhen’s (1995: 26ff) discussion of phonology provides interesting data, both for issues of language contact and for the phonology of Yiddish. As expected, Lithuanian loans in Yiddish show, for the most part, adaptation to Yiddish phonology. Page references for data will be given in the form [L,26]. Liquids r, l. Lithuanian lingual r is realized with uvular r in Lithuanian loanwords in LithYid [L,26], except in CourlYid, which has a lingual r [L,34]. Similarly, NEY generally realizes Lithuanian non-palatal l as hard; CourlYid — presumably from contact with Baltic German — has a soft, alveolar l; thus, Lithuanian loans in CourlYid have a soft l. Lemkhen (1995: 33) notes that, in general, both Lithuanian l and Yiddish l are realized as progressively harder in the geographic sweep from west > east. Lemkhen (1995: 34–36) discusses the problem of husher-hisser confusion (sabesdiker losn; see discussion, above). He claims that for Yiddish in Zamet and the adjoining area (= bay-zamet), as well as in northeastern Lithuania, there was, traditionally, no set of hushers distinct from hissers; rather, there was an unclear inbetween sound. The confusion is not limited to older components in Yiddish (e.g., Semitic, Germanic, Slavic), but rather, is evident in Lithuanian loans as well. Thus, Lemkhen (1995: 35) states that he, as a native speaker, cannot say whether there is š or s in, e.g., NEY špejg6 ~ spejg6 ‘a big frost’ < Lithuanian speigas. Note also confusion in examples gandriš, šmag6n6s, mentioned above. Lemkhen (1995:29ff) discusses issues concerning vocalism. Lith au is integrated in various ways into Yiddish [L: 30], as ou or ei; thus, speakers “choose” identification with one or the other historical Proto-EY vowel. In CourlYid, Lith au remains au; thus, Lith ka˜upas ‘supremacy, preponderance, excess’ > LithYid koup ~ kejp ~ (CourlYid) kaup. The general NEY insertion of glide j between a, e and a following velar nasal is found in Lithuanian loans as well: LithYid trej]k6n6n (zix) ‘to shake’ < Lithuanian treñktis, trenkýti; cf. Gmc.-origin StY [la]g], NEY [laj]g] ‘long’.

Yiddish in the Baltic region 301

Lithuanian loans generally show reduction of the vowel to schwa in post-tonic position (L,29): Lithuanian vãnagas, rãgana, pãšaras > LithYid ván6g6 ‘sparrow hawk’, rág6n6 ‘witch’, (ZY) páš6r6 ‘fodder’. Only in a small number of words is the vowel color retained in post-tonic position (L,29): e.g., LithYid déndarz6s < Lithuanian di˜endaržis ‘enclosure for animals (in a field)’. This type represents a less integrated, newer loan. Lemkhen (1995: 29) also notes as an example of a newer loan — due to the non-reduction of a post-tonic vowel — Lith viršininkas > LithYid vírš6ni(n)k ‘boss’. However, it may be the case here that non-reduction of the posttonic vowel in the final syllable is related to higher-level phonological or morphological structure in Yiddish — that is, it may be seen as part of a suffix -ink or -(n)ik. Consideration of pretonic vowels in the Lithuanian loans also provides insight into foot structure in Yiddish. Commonly, pre-tonic e in Lithuanian loans becomes a in LithYid; thus: Lithuanian pelÍeda, kepùre˙ > LithYid paléd6 ‘owl’, kapúr6 ‘a nonfitting hat’ [L,30]. However, when the word contains two pre-tonic vowels, we find reduction of the second vowel to schwa for reasons related to Yiddish rhythm and foot structure; thus: Lithuanian šaltanõšiai, šimtakõjis > LithYid šalt6nós6s ‘blintzes, dumplings filled with cottage cheese’, šimt6kój6 ‘centipede.’ 4.4.2.2 Morphology. Lemkhen (1995: 44) deals with the issue of NEY gender, and opts for internal causation — rather than appeal to external influence — as the main impetus for the NEY development. Still, Lemkhen (1995: 44) is aware of the possibility of external, i.e., contact-induced change, and cites examples of change from historical neuter > feminine gender in Polish dialects of western Lithuania (around Wilkomir and Kejdan/Ke˙dainiai); in Polish spoken in the Vilnius area, the change of neuter > feminine gender was accompanied by the change of word-final -o# > -a# (generally these are markers of neuter and feminine, respectively, in Standard Polish), e.g., ucho, serco > ucha ‘ear (fem)’, serca ‘heart (fem)’. Likewise, Lemkhen (1995: 44) notes that if a Lithuanian loan ends in schwa in Yiddish, its gender is feminine in Yiddish, even if it was originally a masculine in Lithuanian.35 LithYid shows a rather free variation in how it deals with the Lithuanian noun suffixes -is, -es, -e. For example, LithYid can have an s-suffix where Lithuanian has -e: LithYid bjojríbis ‘rotten scoundrel’ < Lithuanian bjaurýbe˙ [L: 37]. Regionally, however, an s-suffix has established itself as productive in ZY, spreading to new Yiddish words, e.g. krextsis ‘a person who groans/moans a lot’ < krextsn ‘to groan/ moan’. See also kamatsis ‘cheapskate’ [L: 37], cf. StY kamtsn; here, there is direct replacement of one agentive suffix (-n, inherited from the Semitic component in Yiddish) with another (-is). Lith -ys is similarly used as an agent marker (Frank Silbajoris, p.c.). The use of -is as a type of noun marker is also found in Swedish; e.g., tjockis ‘fatso’, kändis ‘celebrity’ < adjectives tjock ‘thick, fat’, känd ‘known’.36 As concerns Lithuanian adjectival suffixes -is, -us, -as, however, LithYid shows a different development. Here, LithYid generally dumps the Lithuanian suffix

302 Neil G. Jacobs

entirely, and usually adds the native Yiddish inflectional suffixes (6, -6r, -n) as needed [L: 38]: Lithuanian plìkas > LithYid plik- ‘bare’. However, when the original Lithuanian adjective contained a (non-final) suffix -in-, LithYid incorporates it: Lith šešk-in-is ‘from a skunk’ > LithYid šesk6n6r.37 Lithuanian loan verbs frequently are placed in the Yiddish class of verbs containing a thematic vowel, and show -6n6n in the infinitive [L: 38].38 4.4.2.3 Semantics. Lemkhen (1995: 47ff) provides a list of Lithuanian loans in LithYid, organized by semantic area, e.g., nature, crafts, tools, agriculture, etc. Two examples will be noted here in the way of illustration. Lemkhen (1995: 56) notes that, often, a Lithuanian loan in Yiddish can take on a pejorative sense, especially when the Lithuanian loan coexists in Yiddish with an equivalent from the Slavic or Semitic component. He additionally notes semantic narrowing, as in Lith baidýti ‘to scare’ > LithYid bajd6n6n ‘to scare (animals).’ Lemkhen (1995: 56) also provides an example of a widespread Slavic-origin word in Yiddish taking on the meaning of its Lithuanian cognate. Generally, StY prost ‘simple’ derives from Slavic, and retains a meaning similar to the Slavic meaning. In Lithuanian prãstas has the meaning ‘bad’. Lemkhen gives the examples of NEY prost6 šix ‘bad shoes’, prost6 ib6rzetsung ‘bad translation’, etc., analogous to the Lithuanian meaning. 4.5 Courland Yiddish This section provides a brief discussion of the Yiddish of Courland, Western Latvia; see M. Weinreich 1923;39 Kalmanovitsh 1926; Jacobs 1994a. Linguistically, CourlYid is an offshoot of ZY. Jewish settlement in Courland is dated to the seventeenth (Kalmanovitsh 1926) or sixteenth century (Ariste 1937). The closing of Courland to new Jewish settlement in 1829 is important in that it means that Lithuanianisms in CourlYid may generally be assumed to predate that year (Lemkhen 1995: 19). Further, Kalmanovitsh (1926) emphasizes the importance of the presence of a significant Slavic component in CourlYid, tracing back to a population of Yiddish speakers in earlier contact with Slavic (Belarusian and Polish). Thus, CourlYid was “born” elsewhere, outside of Courland. There remains much work to do on the perceptual geography of Courland Jews, who referred to any Jew not from Courland as a Zameter (M. Weinreich 1923: 195). Lazerzon (1942) states that inland Jews from the Latgalia province called the Baltic-coastal Jews švejdn ‘Swedes’ or švejdl6x ‘little Swedes’, perhaps owing to the Swedish presence in the area from 1660. Lazerzon (1942) also provides interesting discussion of Yiddish toponyms in Latvia, which show distinct geographic patternings, cultural orbits, and orientations. The basic language setting in which CourlYid existed primarily consisted of contact with Baltic German (the language of official administration, commerce,

Yiddish in the Baltic region 303

urban life) and with Latvian. Russian did not arrive on the scene until much later. The German presence in the Baltic is divided into two main periods: a period of Low German settlement (Teutonic knights) from the thirteenth century, and a High German settlement dating from approximately the sixteenth century. High German eventually supplanted Low German by approximately the end of the eighteenth century, with survival of some Low Germanisms. Courland Jews were often seen as “Germanized” in speech40 and custom (see Rivkin 1951). The CourlYid word lists provided in Weinreich (1923) and Kalmanovitsh (1926) contain many (Baltic) German loans; e.g., CourlYid u(˜)r ‘clock, watch’, farpasn ‘to miss’, taug6ništ ‘good-for-nothing’ < German Uhr, verpassen, Taugenichts (cf. StY zejg6r, farzam6n gorništ). However, the German influences are for the most part lexical. In Jacobs (1994a), I argue that the changes of structural import occurring in CourlYid of the early twentieth century were mostly due to the “pull” of the general ongoing consolidation of EY speech during the modern period, toward broader NEY, or toward the emerging StY language.41 Thus, German loans show adaptation to Yiddish structure; e.g., German prefix ver- > CourlYid far-; German africate pf is adapted to Yiddish phonotactics: f in word-initial position, p non-initially: German Pfingsten ‘Pentecost’, Schnupfen ‘sniffles’, stumpf- ‘dull’ > CourlYid fingstn, šnupn, štump-. Like ZY, CourlYid represents a relatively conservative subdialect of NEY. On the one hand, CourlYid, like ZY generally, has preserved a number of features otherwise lost in NEY, notably, distinctive vowel length, the preservation of a frontround diphthong øy, and the non-merger of øy and ej (CourlYid [æj]), which are merged in broader NEY ej. While distinctive vowel length in CourlYid is still evident in the archival tape recordings of the Language and Culture Atlas of Ashkenazic Jewry (see Herzog et al. 1992), it is clearly a system in collapse, with occurrences of free variation (Jacobs 1994a: 95). This collapse is not attributable to influence from German or other CB languages, which maintain long vowels, but rather, to influences from broader NEY and/or the emerging standard language (Jacobs 1994a: 95–96). On the other hand, like broader NEY generally, CourlYid shows loss of the historical neuter gender, and the rise of intermediate-gender constructions (e.g., di land–afn land). Further, CourlYid shares with the rest of NEY the historical sabesdiker losn. However, here M. Weinreich (1923: 202) notes an interesting development: sabesdiker losn remains in CourlYid only in words of non-German origin (i.e., Slavic and Semitic); thus:

304 Neil G. Jacobs

CourlYid

StY

gloss

cep6blondz6n šuk6 šæjxl

ˇcep6blondž6n suk6 sejxl

‘badger, touch’ ‘to stray’ ‘booth for Sukkoth’ ‘logic; sense’

(< Slavic) (< Slavic) (< Semitic) (< Semitic)

In words of German origin the original hushing-hissing state of affairs has generally been restored, based — it would appear — on the influence of (Baltic) German; thus, CourlYid vas6r ‘water’, kusn ‘kiss’, šrajbn ‘write’, StG Wasser, küssen, schreiben; cf. StY vas6r, kušn, šrajbn, though the CourlYid doublet: kirzn6r ~ kiršn6r ‘furrier’ < German Kirschner provides an interesting counter-example (see Jacobs’ 1994a: 95 discussion of Weinreich 1923: 203). There is reason to look at CourlYid in a broader context of language contact in the Baltic region. For example, in Kiparsky’s (1936) discussion of Yiddish influences in Baltic German (see also Ariste’s 1937 review of Kiparsky), we note that Baltic German blondzen shows the results of NEY sabesdiker losn (cf. StY blondž6n ‘to stray, ramble (having lost one’s way)’ < Polish błondzic´). Furthermore, we find some loan words common to an array of languages in the region. Thus, Weinreich (1923: 235) claims that CourlYid raut ‘window’ traces back to a Middle High German form *rûte (the etymon is found in Courland German as well), but says that the source into Middle High German is unknown. Jacobs (1994a: 93) proposes Swedish ruta ‘quadrangle, window pane’ as the source for both Courland German and CourlYid. The claim of a Swedish source (rather than a Low German source, cf. Dutch ruit) is supported by the presence of this loan in Estonian ruut, which has a stem-final u in non-nominative forms, suggesting that the word was borrowed as an oblique form of an older Swedish weak noun (nom -a, acc/dat/gen -u), something typical of such Swedish loans in Finnish as well. M. Weinreich (1923: 224) notes that the Latvian influence on CourlYid is strikingly small, especially as compared with that of Lithuanian in the neighboring Kovner province. Weinreich suggests that this may owe to the presence of German in Courland. Lemkhen (1995: 22) claims to find only three Latvian loans in CourlYid which do not also have Lithuanian equivalents: priad6 ‘pine tree’, bruklin6s ‘mountain cranberry’, and kaudz6 ‘pile’. On the other hand, contact with Latvians should not be minimized; it existed, exerted itself economically and socially, and influenced Jewish toponymics (Lazerzon 1942) as well as the structure of Jewish households (in the eighteenth century; see Plakans and Halpern 1981). In summary, it may be said that CourlYid of the early twentieth century exhibits a “pull” from three directions: first, from the broader EY speech community; second, from Baltic German; third, though probably weakest, from general Baltic areal influences.

Yiddish in the Baltic region 305

4.6 Yiddish in Estonia Historically, Yiddish in Estonia is seen as part of CourlYid, which in turn is seen as a sub-branch of ZY. Lemkhen (1995: 23) cites Ariste’s (1970: 250–252) list of Lithuanianisms in Estonian Yiddish, further emphasizing the connection of Estonian Yiddish with northermost NEY. However, Lemkhen (1995: 23) notes that these Lithuanian loans were used primarily by the older generation of Estonian Jews. Verschik (1999) gives an overview of Yiddish in Estonia, and provides discussion of the current sociolinguistic situation for Yiddish speakers. Verschik (1999) outlines the historical, social, and educational background of Estonian Jewry. Historically, the Jewish migration to Estonia came from Courland and Lithuania. The Jewish community has been, and is, quite small; it was first organized in 1829; by 1918 it numbered 1523 Jews; 1929 Jews in 1922; 4389 Jews in 1934 (Verschik, 1999). Verschik emphasizes the multilingualism of Estonian Jews: “It is quite evident that Yiddish monolingualism has always been and is impossible in Estonia due to the small size of the Jewish community: multilingualism is an essential characteristic feature of Estonian Jews.” Verschik provides discussion and transcripts of interviews with informants. One possible contact feature which I noticed had to do with word order of subject and finite verb. Generally in Yiddish the subject appears before the finite verb, unless another sentence unit (e.g., direct object, indirect object, adverbial) is fronted, in which case the finite verb moves back to second position, and the subject now follows; thus, StY ikh leyen dos bukh ‘I read the book’, but dos bukh leyen ikh ‘It’s the book that I’m reading’ (and not, for example, the newspaper). In Verschik (ms), we find subject before finite verb, even with fronting of adverbial (transcription as found in Verschik): reidn reidn zei nit, ober in idiš me [Subject] ken [finite verb] zei nit farkeifn ‘as for speaking, they don’t speak, but one cannot sell them away in Yiddish’. Verschik also provides detailed discussion of the occurrence of loan vocabulary from Russian, German, and Estonian. Finally, Verschik mentions as a topic for future investigation a style of mixed speech — Russian, Estonian, and Yiddish — used by some Jews in Estonia.

5.

Conclusions

In this paper I have attempted to place the discussion of Yiddish in a Baltic context. The Baltic contact features in Yiddish were primarily lexical. In the Thomason & Kaufman (1988) framework, the overwhelmingly lexical nature of the contactinduced influences (Lithuanian Æ Yiddish) suggests a more superficial type of influence. The Slavic impact upon Yiddish has been much deeper, encompassing all levels of the grammar, as shown by U. Weinreich (1958). Yet the Baltic data in



306 Neil G. Jacobs

Yiddish can provide us with valuable data on phonological and morphological nativization processes, as well as shed light on the general synchronic description of Yiddish, for example, in the description of syllable and foot structure in Yiddish. We have seen repeated examples of the important role played by the macrosociolinguistic orientation of Yiddish-speaking Baltic Jews. While there are significant CB influences observable, the pull is strongest from the consolidation processes in EY speech and writing. At the same time, the CB presence has had its impact — beyond the merely lexical — on Yiddish in the Baltic region; this needs to be taken into account in linguistic descriptions of varieties of Yiddish in the region. This present paper is, of course, only one step in the investigation of Yiddish in its Baltic context. Some questions are raised. Why are there so few Latvian loans in CourlYid? The answer does not seem to rely solely on the number of years Jews lived in Courland, for Lithuanian loans were taken up after comparable time in Lithuania. M. Weinreich (1923) noted the presence of a dominant Baltic German; this certainly played a considerable role in the speech both of Jews and of Latvians. However, the levels of historical Jewish-Latvian contact would perhaps lead us to expect a larger Latvian-origin corpus in CourlYid. It would perhaps be most useful to look more deeply at the interplay of multiple and shifting “pulls” — linguistic orientations and orbits, over time and geography.

Notes * My original research on Yiddish in Courland, conducted at the YIVO Institute for Jewish Research, and at the archives of the Language and Culture Atlas of Ashkenazic Jewry, was funded by the Melton Center for Jewish Studies at The Ohio State University. I thank Prof. Mikhl Herzog, director of the LCAAJ for access to the Courland (and other) tapes, as well as for personal discussion about Courland Yiddish. I also thank the following individuals for discussion on relevant issues: Mr. Zachary Baker and Ms. Dina Abramowicz of the YIVO, Prof. Mark Louden, and Prof. Ilse Lehiste. For research support for the present paper, I thank the Center for Slavic and East European Studies at The Ohio State University. I also thank Ms. Anna Verschik for access to her unpublished paper. Any mistakes or shortcomings in the present article are the responsibility of the author. 1. The number of years of contact is not always the determining factor in the degree of influence in language contact; see U. Weinreich (1958: 40–41) on the differing degree of Slavic influence in eastern Europe upon Yiddish and Colonial German. 2. The ethnographic present is used, reflecting developments before the Nazi-German genocide in World War II. 3. On earlier use of the term topography in another sense, see by Chambers (1994). Topography is used here to refer to the gradualness, or Staffelung in variation across the Yiddish dialect map; see Jacobs (1993). 4. The term Baltic Yiddish is not, to my knowledge, used in Yiddish linguistic scholarship. Rather, varieties of Yiddish spoken around the Baltic rim are generally seen in the context of their

Yiddish in the Baltic region 307

respective dialects — generally, as subsets of Northeastern Yiddish in present-day Lithuania, Latvia, and Estonia, and, historically, of Western Yiddish in Scandinavia. In the present paper the term Baltic Yiddish is used both as a shorthand for easy reference, and as a (straw?) construct for examining to what extent Yiddish along the eastern Baltic rim differs significantly from its sister subdialects within the given dialect region. Use of the term Baltic Yiddish is not meant to imply sufficient cohesiveness or distinctness that would force a special classification within the field of Yiddish dialectology. 5. As a toponym, the term was used in Medieval Hebrew to refer to German lands. Later, the term came to refer to the home territory of Ashkenazic Jewry. The first known use of the name Ashkenaz was as an anthroponym, referring to a son of Gomer (Genesis 10: 3). 6. The term Ashkenazic Jewry refers to that Yiddish-speaking Jewry which arose in Central and Eastern Europe, and their progeny. The Ashkenazic branch of the Jewish people is thus distinguished from other branches, which developed their own civilizations, specific Jewish languages, etc., as for example Sephardic Jewry, which arose on the Iberian Peninsula and created a distinct Judeo-Spanish language — Judezmo. 7. Both Katz and Wexler view the Jewish settlements in Loter as giving rise to a separate Jewish Germanic-based idiom, which Wexler (1981) terms Ashkenazic German. 8. For detailed discussion of Yiddish dialect classification, the reader is referred to Herzog (1965), Katz (1983), Herzog et al. (1992). 9. On the problematic nature of the Stammbaum classification of SEY, see M. Weinreich (1980), U. Weinreich (1969). 10. Though the specific Slavic-origin features need not be identical in NEY and SEY. 11. Language names are generally neuter in non-NEY varieties of Yiddish. In NEY, which lost the historical neuter gender, language names are generally masculine. In the present paper the standard romanization system employed by the YIVO Institute for Jewish Research will be used when referring to Yiddish names (e.g., personal names, toponyms), general concepts (e.g., sabesdiker losn), where phonetic detail is not at issue. Linguistic examples will be given in phonetic symbols. However, automatic processes such as the assimilation of /n/ to place of articulation of an adjacent consonant will not be indicated unless germane to the point being made. Likewise, the symbol 6 represents a set of variants across the NEY area. Unless otherwise indicated, stress in Yiddish words is on the penultimate syllable. 12. In Yiddish, the term Litviner refers to a non-Jewish Lithuanian. 13. The suggestion could be raised that the very recession in NEY of sabesdiker losn, and restoration of the husher-hisser distinction, might have been encouraged by the distinctions made in coterritorial Lithuanian and Latvian. However, this is not borne out by the data. Lithuanian loanwords in NEY often show the hushing-hissing confusion. Of more importance in the recession of sabesdiker losn in NEY is the influence of StY (U. Weinreich 1952). In CourlYid, however, contact with Baltic German did play a role as well (see M. Weinreich 1923: 202, and discussion, below). 14. Shapiro (1939, cited in Herzog 1965: 156, Note 4) suggested a Lithuanian substrate Jewish language from which Lithuanian Jews later shifted to Yiddish; this suggestion is refuted, among other things, by the findings of Yiddish historical phonology. See also Herzog (ibid.), citing Sapir (1915), Mark (1944, 1951: 446). 15. While Belarusian is a three-gender language (masculine, feminine, neuter), there has occurred some realignment of gender based on the phonetic similarity of fem -a nouns and neut -a (< -o) nouns; see also Jacobs 1990a.

308 Neil G. Jacobs

16. Yiddish came into direct contact with Estonian much more recently. Historical simplification of the NEY gender system likewise cannot be attributed to Estonian influence. 17. U. Weinreich (1961) proposed intermediate feminine category. The parallel category intermediate masculine was proposed in Jacobs (1990a). 18. The possibility of Yiddish influence upon Lithuanian as concerns the mass nouns should also be investigated. All too often, linguistic treatments of language contact have ignored the possibility of areal influences from Yiddish, while recognizing other languages in the “areal soup”; see Jacobs 1996:211–212 on contact features in Vienna Urban German. On Yiddish influences upon Colonial German, see U. Weinreich (1958: 15) on word order, and (ibid.: 16–17) on vocabulary. 19. The exact analysis of this form is more complicated. One could possibly analyze it as 1st/3rd p. plural + suffix dik, alternatively as present stem of verb + suffix (6)ndik, or as infinitive + dik. Any analysis chosen leaves problematic residue. See Mark 1978: 335. 20. Use of the vowel i reflects the Yiddish spelling; the vowel in the reflexive pronoun is usually realized as e, a, or schwa. 21. In Lithuanian, the reflexive element is placed at the end of the participle if the verb stem does not contain any prefixes and the whole participle is in the nominative case. Generally, the reflexive element in Lithuanian is placed between prefix and verb root (Maria Koptjevskaja-Tamm, p.c.). 22. NEY speakers use this circumlocution as well. 23. The source of the distinction possible in the literary language is thus a separate matter. 24. Mark (1951: 448) writes that Grodne was the seat of the Jewish Va’ad d’Lita ‘Council of Lithuania’, and had the Suvalker kehiles ‘Jewish communities’ under its jurisdiction; as well, there were strong business ties, similar networks of trade-fairs, rabbi selection, arranged marriages, etc. 25. On a different slicing of the perceptual geographic pie by Courland Jews, see below. 26. However, it has been pointed out to me by Maria Koptjevskaja-Tamm (p.c.) that the situation is somewhat more complex: “some Samogitian dialects have changed original diphthongs not to monophthongs, but to other diphthongs (e.g. /uo/ became /ou/ in North Samogitian, cf. Standard Lithuanian duona ‘bread’ vs. North Samogitian douna).” Koptjevskaja-Tamm also provides an example of an innovative diphthong in Samogitian Lithuanian kuoja ‘leg, foot’, vs. Standard Lithuanian koja. 27. Mark (1951: 447) does, however, provide some examples of limited parallelism in certain smaller phonetic details. Thus, he claims that in the Ezherener region the hard l sound is found in both the local Lithuanian and the local Yiddish. However, he points out the possibility that here both languages were influenced by contact with Russian. 28. The second-degree diminutive suffix represented in StY by -6l6 is reconstructed for ProtoYiddish as *-6rl6. The /r/ is evident in varieties of Austro-Bavarian German. On analogous r-coloring in some varieties of CY, see Jacobs 1990b: 82. The term r-coloring here refers to the realization of the 6 preceding the r as some sort of full vowel (a or o, depending on which variety of Yiddish). 29. However, the unrounding of the glide must be seen as fairly recent in many locations, as recently as the early part of the twentieth century; see Herzog 1965: 11. 30. U. Weinreich (1958:37) speaks of “relatively few recorded elements of Lithuanian… or Lettish origin…” citing Mark (1951) for Lithuanian, and M. Weinreich (1923) and Kalmanovitsh (1926) for Latvian loans. U. Weinreich is concerned here with the relatively higher impact of Slavic over other coterritorial influences in Eastern Europe.

Yiddish in the Baltic region 309

31. Maria Koptjevskaja-Tamm (p.c.) suggests that the Lithuanian words corresponding to som6n6s and saks6n6s seem to be restricted to the area of Palangà. 32. Lemkhen (1995: 71) writes that the Lithuanian loan gond(6)r6s~gandris~gandriš is a possible word for ‘stork’ in Yiddish in Lithuania, alongside German-origin štorx and Slavic-origin bušl; however, in mocking speech in Zamet, only the Lithuanian-origin word is used (in describing a tall person). 33. Lemkhen (1995: 118) notes the Lithuanian dialect variant smãgenys alongside sme4genys. 34. Lemkhen (1995) in Yiddish follows in large part, though not entirely, the author’s earlier publication in Lithuanian (Lemchenas 1970). 35. This conforms with a general tendency not limited to NEY; thus, e.g., Polish neuter b³oto ‘mud’ is regularly a feminine in Yid blot6, with final schwa. 36. Note also Swedish barkis (in Stockholm Swedish bergis) ‘a type of braided white bread’, a loan from WY berx6s~ barx6s. Here the Yiddish suffix -6s < Hebrew abstract noun suffix -u˜þ has likely been reinterpreted as the -is suffix under discussion. Whether there is a contact explanation linking Swedish -is and Lithuanian -ys, or whether this is merely coincidence, is unknown to me. 37. Note also augmentative adjectival suffix in NEY -6n6, as discussed in Mark (1951: 459): NEY grejs-6n6 ejgn ‘big eyes’. 38. See also Stankiewicz 1993 on Yiddish thematic verbs. 39. In Jacobs (1994a), I mistakenly referred to Weinreich (1923) as his doctoral dissertation. 40. M. Weinreich (1923: 207) notes that Jews from Zamet would make fun of Courland Jews for “speaking German” for using the verb mext rather than volt in subjunctive constructions. What is at issue here are perceptions, not necessarily the facts, since Weinreich points out the antiquity of the use of mext in Yiddish documents. 41. By the late nineteenth century there had occurred, for some Courland Jews, a shift from Yiddish to German as the home language. However, the present discussion is limited to consideration of CourlYid.

References Ariste, Paul. 1937. Tsu der hashpoe fun yidish oyf nit-yidishe shprakhn. [Includes a review of Kiparsky 1936]. Yivo bleter XI.1–2, 83–85. Ariste, Paul. 1970. Review of Lemchenas 1970. Baltistica 6.2. Chambers, Jack. 1994. “An introduction to dialect topography.” English world-wide 15, 35–53. Herzog, Marvin. 1964. “Channels of systematic extinction in Yiddishdialects.” In: For Max Weinreich, 93–107. Herzog, Marvin I. 1965. The YiddishLanguage in Northern Poland: Its geography and history. Bloomington: Indiana University Press. Herzog, Marvin, Uriel Weinreich z”l, and Vera Baviskar z”l. 1992. The language and culture atlas of Ashkenazic Jewry, Vol. 1. Tübingen: Max Niemeyer Verlag. Jacobs, Neil. 1990a. Northeastern Yiddish Gender-Switch: Abstracting dialect features regionally. Diachronica VII: 1, 69–100. Jacobs, N. 1990b. Economy in YiddishVocalism: A study in the interplay of Hebrew and nonHebrew components. Volume 7: Mediterranean Language and Culture Monograph Series, P. Wexler (ed.). Otto Harrassowitz Verlag: Wiesbaden. [Revised from Neil G. Jacobs. 1984.

310 Neil G. Jacobs

Unpublished Columbia University dissertation: “Economy in YiddishVocalism: The Case of Central Yiddish.”] Jacobs, Neil. 1993. Linguistic geography and phonological instability: Vowel length in East Central Yiddish. Paper delivered at Methods VIII Conference on Dialectology. University of Victoria, Victoria, B.C., Canada, August 3–7, 1993. Jacobs, Neil. 1994a. Structure, Standardization, and Diglossia: The Case of Courland Yiddish. In: Dagmar C. G. Lorenz & Gabriele Weinberger (eds.), Insiders and Outsiders. German-Jewish, Yiddishand German Literature and Culture in Contact, 89–99. Wayne State. Jacobs, Neil. 1994b. On a Structural ‘Fifth Column’ in Sociolinguistic Change: The Diffusion of a Standard Yiddish Feature in Yiddish Dialects. In: Howard I. Aronson (ed.), NSL. 7: Linguistic studies in the non-Slavic languages of the Commonwealth of Independent States and the Baltic Republics, 133–150. Chicago Linguistic Society: Chicago. Jacobs, Neil. 1996. On the investigation of 1920s Vienna Jewish Speech: Ideology and Linguistics. American Journal of Germanic Linguistics and Literatures 8: 2, 177–217. Kalmanovitsh, Zelik. 1926. Der yidisher dialekt in kurland. Landoy-bukh: D”r Alfred Landoy tsu zayn 75stn geboyrntog dem 25stn november 1925 fun zayne gute-fraynt un talmidim. Shriftn fun yidishn visnshaftlekhn institut, Band 1, Filologishe serye 1, 161–188. Katsenelnbogen, U. 1951. Litvakes. In: M. Sudarski (ed.), Lite. Katz, Dovid. 1983. Zur Dialektologie des Jiddischen. In: W. Besch et al. (eds.), Dialektologie: Ein Handbuch zur deutschen und allgemeinen Dialektforschung, 1018–1041. Berlin & New York: Walter de Gruyter. Katz, Dovid. 1987. The proto dialectology of Ashkenaz. In: Dovid Katz (ed.), Origins of the Yiddishlanguage. [= Winter Studies in Yiddish, Vol. 1]. Pergamon Press: Oxford. Kiparsky, V. 1936. Fremdes im Baltendeutsch. Mémoires de la Société Néophilologique de Helsingfors, XI. Helsinki. King, Robert. 1979. Evidence of the German component. Paper presented at First International Conference on Research in YiddishLanguage and Literature, August 6–9, 1979. Oxford Centre for Postgraduate Hebrew Studies. Lazerzon, M. 1942. A bisl baltish-yidishe geografye. Yidishe shprakh 2: 3, 84–87. Lemchenas, Ch. 1970. Lietuviu˛ kalbos ˛itaka Lietuvos Žydu˛ tarmei. Vilnius: Mintis. Lemkhen, Khatskl. 1995. Di hashpoe fun litvish oyfn yidishn dialekt in Lite. In: Oksforder Yidish 3, 6–130. Mark, Yudel. 1944. Di litvishe yidn un der neytraler min. Yidishe shprakh 4.83–94. Mark, Yudel. 1951. Undzer litvisher Yidish. In: Mendl Sudarski (ed.), Lite, 429–472. New York: Kulturgezelshaft fun livishe yidn. Mark, Yudl. 1978. Gramatik fun der yidisher klal-shprakh. Alveltlekher yidisher kultur-kongres. New York. Mathiassen, Terje. 1996. A short grammar of Lithuanian. Slavica Publishers: Columbus. Miller, David Neal. 1990. Ashkenaz: Paradigm and resistance. Paper delivered at conference: The role of geography in Jewish civilization — Perceptions of space, place, time, and location in Jewish life and thought. October 21–22, 1990. The Ohio State University. Plakans, Andrejs, and Halpern, Joel M. 1981. An historical perspective on eighteenth century Jewish family households in Eastern Europe. In: Paul Ritterband (ed.), Modern Jewish Fertility, 18–32. Brill: Leiden. Sapir, Edward. 1915–6. Notes on Judaeo-German Phonology. Jewish Quarterly. Review N. S. 6.231–236. (Reprinted in David G. Mandelbaum (ed.), Selected Writings, 1949, 252–272. Berkeley/Los Angeles: University of California Press.) Shapiro, Moyshe. 1939. Der gramatisher min in yidish. Afn Shprakhfront 3rd series, 3.111–163.



Yiddish in the Baltic region

Stankiewicz, Edward. 1993. The Yiddish thematic verbs. In: David Goldberg (ed.), The field of Yiddish: Studies in YiddishLanguage, folklore, and literature. Fifth collection, 1–10. Northwestern University Press and the YIVO Institute for Jewish Research: Evanston. Thomason, Sarah G. & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley, Los Angeles, London: University of California Press. Verschik, Anna. Ms. Some aspects of the multilingualism of Estonian Jews. Weinreich, Max. 1923. Dos kurlender yidish. Chapter 4 in: Shtaplen: fir etyudn tsu der yidisher shprakh-visnshaft un literatur-geshikhte, 193–240. Berlin: Wostok. Weinreich, Max. 1980. History of the Yiddishlanguage. Translated by Shlomo Noble, with the assistance of Joshua A. Fishman. Chicago and London: University of Chicago Press. [Partial translation of his: Geshikhte fun der yidisher shprakh: bagrifn, faktn, metodn. 1973. New York: YIVO Institute for Jewish Research.] Weinreich, Uriel. 1952. Sábesdiker losn in Yiddish: A problem of linguistic affinity. Word VIII, 360–377. Weinreich, Uriel. 1958. Yiddishand colonial German in Eastern Europe: the differential impact of Slavic. American contributions to the fourth international congress of Slavicists, 369–421. The Hague. Weinreich, Uriel. 1961. The Seven Genders of Yiddish. Unpublished paper read before the annual meeting of the Linguistic Society of America, December 29, 1961. Chicago. Weinreich, Uriel. 1964a. Four riddles in bilingual dialectology. American Contributions to the Fifth International Congress of Slavicists, 335–359. The Hague. Weinreich, Uriel. 1964b. Western traits in Transcarpathian Yiddish. In: For Max Weinreich, 245–264. Weinreich, Uriel. 1969. The geographic makeup of Belarusian Yiddish. In: Marvin I. Herzog, Wita Ravid & Uriel Weinreich (eds.), The Field of Yiddish Vol. 3, 82–101. London-The HagueParis: Mouton & Co. Wexler, Paul. 1981. Ashkenazic German. International Journal of the Sociology of Language 30, 119–130. Wexler, Paul. 1987. Explorations in Judeo-Slavic linguistics. Brill: Leiden. Wexler, Paul. 1991. Yiddish — The fifteenth Slavic language. A study of partial language shift from Judeo-Sorbian to German. International Journal of the Sociology of Language, v–150. Wolf, Meyer. 1969. The Geography of Yiddish Case and Gender Variation. In: Marvin I. Herzog, Wita Ravid, & Uriel Weinreich (eds.), The Field of Yiddish Vol. 3, 102–215. London/The Hague/Paris: Mouton & Co.

311



The North Russian Romani dialect Interference and Code Switching* Aleksandr Yu. Rusakov

1.

Introduction

In addition to the traditional classification of Romani dialects in terms of genetic relationships,1 attempts have recently been made to create a new taxonomy based on on criteria having to with language contact phenomena. Besides the “paraRomani” dialects (in which a “switch” to an “alien” grammar has taken place, cf. Bakker & Courthiade 1991), we could distinguish different types of “inflected” Romani dialects, depending on the degree of interference from other languages (cf. Courthiade 1995: 97–106). In the present article,2 I shall try to show some characteristic features of a dialect belonging to one such “contact type” — an inflected dialect with strong interference, namely the North Russian Romani dialect (NRRD). Two major closely related points will be discussed. First, I shall look at some interference processes that affect the grammatical system, more precisely the system of verbal aspect. Then, I shall turn to a more general discussion of code-mixing and code-switching phenomena. NRRD, which belongs to the Baltic group of non-Vlach Romani dialects, is spread mainly in the northern part of Russia. One of the largest groups of NRRD native speakers live in the suburbs of St. Petersburg. The number of Roma (Gypsies) in St.Petersburg district (Leningradskaja oblast’) exceed 4000 according to the 1989 census, but not all of them are speakers of NRRD. The autonym for the ethnic group is russka roma, haladytka roma. Practically all speakers of NRRD are bilingual. NRRD is the first language of Roma children, but Russian seems to be the dominant language in the situation of Romani-Russian bilingualism. NRRD is a very typical example of strong linguistic interference (stage 4 or 5 of interference: moderate and heavy structural borrowing, according to Thomason & Kaufman 1988). Nearly all levels of its phonological and grammatical system resemble the Russian one. The phonological system of NRRD is practically identical to the Russian one (Eloeva & Rusakov 1990: 11–13). The interference in the grammatical system is manifested: (a) in meaning changes of the elements of

314 Aleksandr Yu. Rusakov

already existing morphological categories; (b) in the structural reshaping according to the Russian models of the forms of already existing morphological categories; (c) in the borrowing of the Russian grammatical markers. Such borrowings sometimes are and sometimes aren’t accompanied by the change in the system of grammatical categories. The word-order syntax of NRRD is practically identical to that of Russian.

2.

Interference in the NRRD verb system: The rendering of aspect

The original tense-aspect system common to the all dialects of Romani seems to be characterized by the aspectual opposition of aorist to imperfect and the single form of present-future, realizing aspectual meaning depending on verb lexical semantics and on the context: (1) bagand’a ‘he, she sang (perfective)’ bagala ‘he, she sings, will sing’

bagavas ‘he, she sang (imperfective)’

In the new verbal system arising under Russian influence, the opposition imperfect : aorist disappears, with the old aorist form serving as a general past, and the imperfect surviving as a special and rarely used aspectual form with iterative meaning, the aspectual opposition are expressed by borrowed prefixes, and the future is expressed by two new constructions: (2) bagand’a s-bagand’a ‘he, she sang (imperfect.)’ ‘he, she sang (perfect.)’ bagala ‘he, she sings’ lela te bagal: s-bagala ‘he, she will sing (imperfect.)’ ‘he, she will sing (perfect.)’

All the examples in schemes (2)–(3) are taken from the syntactic questionnaire. But these schemes reflect rather an idealized picture of the possibilities of express the aspectual meaning. The real picture is more complicated. 2.1 The invasion of Russian verb prefixes General. Verb forms formed with Russian prefixes are frequent in NRRD texts. In the texts recorded in 1984, the following distribution of prefixed forms is found:

The North Russian Romani dialect

Among them: Aorist Synthetic future Present

All finite verbal forms (without Russian borrowings)

Prefixed forms

Ca 500

107

160 072 168

056 019 018

The borrowing of alien prefixes is not an exclusive feature of NRRD: it is observed in some other Romani dialects in contact with languages with rich prefixal systems (cf. Boretzky, Igla 1991: 21–25). Practically each NRRD verb may be used with the borrowed prefixes on the Russian model: (3) a. b. c.

te ot-des ‘to give back’, cf. Russ. ot-dat’ te vy-des ‘to give out’, cf. Russ. vy-dat’ te roz-des ‘to distribute’, cf. Russ. roz-dat’

The Russian prefixes in NRRD preserve all their grammatical and lexical meanings: aspectuality, actionality, possibility to change the verb valency structure and so on. However, it has been noted in the literature that the copying of the Russian aspectual system in NRRD is not complete. N. Boretzky (1989: 368) writes correctly that in NRRD as well as in the other dialects with borrowed prefixes, imperfectivization as a derivational means is absent, cf. also: “Some dialects of Romani have developed a type of prefixal aktionsart similar to that which is found in Slavic or German, but this does not appear to constitute a true superordinate aspectual system” (Friedman 1985: 387). A careful analysis of the functioning of prefixed verbs in texts and in the answers to the syntactic questionnaires makes it possible to modify these statements. Equivalence of prefixed and simple forms. Prefixed and simplex forms of one and the same verb lexeme are often found in the texts and in the answers to the questionnaire without any obvious difference in meaning. Cf. the following pair of examples, drawn from the same text: (4) I avne roma, u-galyne so joj buty kerd’a ‘And the Roma came, (they) discovered that she worked’ (5) Nu dote gyne pal latyr te roden i vdrug galyne… ‘And then (they) went to look for her, and suddenly discovered…’

Commenting upon the questionnaire sentence “He asked money from me, but I

315

316 Aleksandr Yu. Rusakov

didn’t give it him”, the informant notes that it is possible to translate ‘I didn’t give’ either as na dyjom or as na ot-dyjom. On the basis of such examples, we may conclude that the use of verb prefixes for the expression of perfective meanings is not obligatory. On the other hand, prefixed verbs are used (though not very often) to express the meaning of the past imperfect tense. This should probably be seen as a calque of Russian imperfectivized prefixed verbs. (6) I vot, priježali ke joj but moly svatat’. Joj na gyja i daj la na ot-dyja ‘And so, (they) came to her many times trying to marry her. She didn’t go and her mother didn’t give (imperf) her (to them)’ (cf. Russian ne ot-davala “(she) didn’t give (imperf)”.)

Loan translations. N. Boretzky has pointed out that Romani verbs with Russian prefixes are modelled on individual Russian verbs. We are thus dealing with a kind of loan translation. Thus, differences in lexical or actional meaning between the members of Russian verb pairs are often transferred to Romani, as in: (7) phenava “I say, I’ll say” ros-phenava “I retell, I’ll retell”, cf. skažu (Rus) ras-skažu (Rus)

In the same way, Russian aspectual differences carry over to Romani verb pairs: (8) po-pychne “They asked (perf)”; u-cherde “have stolen (perf)” po-prosili (Rus) u-krali (Rus) (Boretzky 1989: 368, cf. also: Beskrovnyj 1971; Barannikov 1934).

Like in Russian, we find in NRRD combinations of “actional-lexical” and aspectual modification of verbs by prefixation: (9) A phen lakiri … vy-gyja pale rakleste pale rom. ‘And her sister … married non-Roma’ (lit ‘went out to the husband’)

An important conclusion can be drawn from the above: grammatical changes in the sphere of aspect representation are brought into NRRD “by lexical means”. The significance of the semantics of the verb. The usage of prefixed vs. unprefixed verb forms is greatly influenced by the semantics of the base verb. Thus, terminative verbs (f.ex. te aves “to come”) are as a rule used in the past and future tenses without prefixes (we mean only the aspectual prefixed forms), and have perfective aspectual meaning: (10) jov javja khere ‘He came home (perf)’

Such verbs may be used sometimes with the imperfective meaning also: (11) Me javja ke jov but molo ‘I came to him many times.’

The North Russian Romani dialect

Non-terminative verbs, on the other hand, may be used with perfective meaning with as well as without prefixes. (12) joj gyja ucˇit’s’a ‘She went (perf) to study’ joj u-gyja lesa ‘She went away (perf) with him’

Are prefixed verbs stored or constructed on-line? A very important, question arises, however. Are prefixed verbs part of the NRRD lexicon, or they are constructed online? It is nearly impossible to answer these questions without direct access to NRRD speakers’ linguistic competence. We may suppose, however, that NRRD speakers use both strategies. The case of prefixed verbs does not differ from the general situation of Russian borrowings in the dialect (see below) or from the processing of word-forms in languages with a highly developed morphology in general. The rather numerous cases of co-existence between Russian prefixed verbs and their Romani equivalents in NRRD texts may be considered as an indirect confirmation of the reality of the “on-line construction” strategy: (13) jov saro pro-igryvaet, graf Cˇornyj, … pro-khella saro ‘…he loses (Rus) everything, count Chornyj, (he) loses everything’

Similar examples are found among the answers to the syntactic questionnaire: (14) (the birds) pri-letajut, pri-urn’ana ‘fly in (Rus), fly in’

There are, however, some prefixed verbs in NRRD which may hardly be supposed to be constructed on-line. I am referring, first of all, to the prefixed derivatives of the polysemantic verb te les “to take, to begin; semi-auxiliary, forming the analytic future”. Consider text examples such as: (15) Daj la pri-lyja… ‘Mother received her’

where Russian: vz’ala “he, she took” corresponds to NRRD lyja and Russian prinjala “he, she received” to NRRD pri-lyja. From the synchronic point of view the Russian prefixed verbs are very obscure opaque derivatives with a non-independent root. On-line calquing is here almost excluded as a possibility. A more likely supposition is the lexical identification was done at an earlier stage of dialect development, perhaps, in the environment of an other Slavic language. Other examples found in our texts are the forms of the verb te ot-keres “to open (the door)”. The model for this verb was no doubt the Russian verb ot-tvorit’ (lit. from-do, work). It is highly unlikely for the interlanguage identification to take place on-line every time the forms are used — the “normal” counterpart of the Romani verb te keres is Russian verb delat’ “to do”, tvorit’ is a rather archaic form.

317

318

Aleksandr Yu. Rusakov

We may suppose, however, that the very correspondence between the Russian and the Romani verbs (here between keres and tvorit’) is “stored” in the lexicon, but the concrete prefixed forms may be constructed on-line according to existing models. Are Russian verb prefixes productive in NRRD? An other important question is whether the loan translation of certain Russian verbs is the only possible strategy for the NRRD speakers, or if they can have a strategy of forming new verbs (or of aspectual forms of existing verbs) by “adding” Russian prefixes to Romani verbs according to productive models, but without association to specific Russian lexemes. In other words, do the borrowed prefixes have their own derivational and/or inflectional potential in NRRD?3 It is very difficult to answer this question. We have, however (both in our texts and in the answers to the syntactic questionnaire), prefixed verb forms which may hardly be explained as calques of specific Russian verbs. See f.ex.: (16) te piri-pas’os ‘to spend a night, lit. through-lie’ (cf. Rus pere-nocˇevat’ ‘to spend a night’ lit. through - spend. a. night)

Consider also the set of prefixed forms based on the verb te chuves ‘to put’, which all denote actions applied to clothing (te ras-chuvespe “undress”, s-chuves ‘take off (one’s clothes)’ Such examples may be treated, on the one hand, as lexical residues of earlier calques of the certain words of other dialect or chronological variants of Russian which thus reflecting preceding stages of NRRD language contacts. It is possible, on the other hand, that NRRD speakers have the strategy of the constructing of new verbs (as well as, possible, of the aspect forms of the existed verbs) with the help of the borrowed prefixes according to productive models. It should be noted, that in the examples cited above the prefixes retain the transparent derivative meaning that they have in Russian. It is clear, of course, that strategies may be combined in different ways and play different roles in the competence of certain speakers. 2.2 The fate of the imperfect The imperfect is very rarely used to express temporal relations.4 In our texts there is only one fragment where imperfect forms were used. The subject is the past nomadic life of the Roma and the imperfect forms designate habitual actions, alternating freely with the forms of the preterite (aorist) and historical present: (17) Nu syr de šatry, nu dzhinde (pret) de šatry roma, rikirde (pret) gren, dadyves rat’kirnas (imp) de dava foro, tas’a tradenas (imp) de javir foro… Kernas (imp) o šatry, roma beshte (pret) po kham, tat’kirnape (pres), romn’a psirna (pres), piri gav dzhana (pres)… ‘How did the Roma live in tents, well, the Roma lived in tents, had the horses, today spent the night in this town, tomorrow went to the other town… Made

The North Russian Romani dialect 319

the tents, the Roma lived with the sun, spent nights, the Roma women go out, go through the village’

In the answers to the syntactic questionnaire, imperfect forms are used to express the combination of two elements — emphasis on the past character of the action and iterative meaning, also expressed by additional, usually lexical means: (18) Ran’še jov cˇasto bagavas ada gily ‘He sang this song often before’

In NRRD the imperfect seems to have the status of a special and rarely used aspectual form with iterative meaning. Thus, the former purely aspectual opposition aorist/imperfect has undergone a fundamental change in NRRD. 2.3 The status of the analytic future in NRRD The degree of grammatical adaptation of the elements which result from Russian interference is often rather difficult to determine. This holds for instance of a wellknown “Slavic” element of the NRRD verbal system such as the analytic future of the type lasa te bagas “we’ll sing”.5 This form seems to be very close to the corresponding Russian construction but is really very rare. In our texts we have only two examples of analytic future compared to 72 examples of future time reference which are not formally different from the present (19 of which have prefixes). We may suppose, however, that the model for this NRRD construction is the alternative half-grammaticalized Russian future model with the verb stat’ ‘become, begin’: (19) Russian ja stanu pet’ ‘I’ll sing’

If it is true, we might suppose that the NRRD analytic future should be treated rather as an inchoative (and, maybe, a not fully grammaticalized) construction, than a future one, cf. (20) Russian ja stal pet’ ‘I began to sing’ (21) NRRD me liya te bagav ‘I began (took) to sing’

2.4 Discussion The extremely heterogeneous system for expressing tense-aspect relations that has been forming in NRRD, is characterized by a high level of opacity and markedness.

320 Aleksandr Yu. Rusakov

Indeed, it is almost impossible to determine the aspect meaning of a verb form taken without a context. If we do not have a sufficient context and a corresponding Russian verbal pair for a given pair (prefixed and unprefixed verbs) of Romani verbs we cannot determine what kind of relations exist between the original and the derived words. On the other hand, as we have seen, in the dialect there is an impressive selection of different means, both pure grammatical and ranging from grammatical to lexical, which can express and in fact do express aspect meanings. There is no doubt, that the situation be found in NRRD is highly marked. Thus, the development of the means of expression of tense-aspect relations from those in “common-Romani language” up to those in NRRD,6 can be described in terms of a transition from transparency to opacity and high level of markedness. In principle, we could expect such a system to be unstable. However, if we look at textual materials from the middle of the 19th century, we find, that on the contrary, the expression of tense-aspect in the dialect have not changed very much since then. For instance, in the collection of Dobrovol’skij (1908), whose materials reflect a state of the dialect which is very close to that of modern NRRD, tense-aspect relations are expressed in a similar way. 65 out of 300 finite verb form taken from the Dobrovol’skij’s collection (Russian lexical loans were not included in the analysis) were prefixed. Like in the modern dialect, prefixed verbs are most often used in the forms of preterite (41 out of 145) and in the synthetic form of the future tense (this form formally does not differ from that of present tense; 12 out of 26). Both in our records and in Dobrovol’skij’s collection imperfect forms are extremely rare. There is also no important changes in the functional use of prefixes: they can modify both lexical and aspectual meanings of the verb. We could also point out many similarities in details.7 In other words, during the last century the means for expressing tense-aspect relations in the dialect have not substantially changed. The marked component of the language system demonstrates a high degree of stability. This is very well in accordance with the fact shown by Thomason and Kaufman. They found out, that in the situation of intensive language contacts, which can result in interference of the fourth or the fifth degree, interference phenomena can increase the degree of the markedness of the language system (Thomason & Kaufman 1988). In a situation where Russian becomes more and more dominating, and Romani language is preserved due to communicative-pragmatic needs, the Russian language system starts playing “a double role.” Russian elements penetrate the system of the dialect, promote the processes of language attrition and hamper the language potency of NRRD itself.8 At the same time, this penetration, paradoxically, increases the viability of the dialect. Russian language can be seen as an inexhaustible reservoir from which the dialect can “scoop” elements from different language levels. It is interesting, that, as we tried to show, the majority of interference phenomena penetrate the language system of NRRD via the lexical level. As for the borrowing

The North Russian Romani dialect

of Russian prefixes, the lexical character of the process can be postulated not only for the origin of this phenomenon, but also synchronically. At the same time (perhaps precisely for this reason), NRRD has not copied completely the Russian tense-aspect system. This is well in accordance with N. Boretzky’s opinion that complete copying or borrowing of a system is not obligatory in interference situations (cf. Boretzky 1989). Actually, in NRRD there is a possibility to express grammatical meanings almost in the same way as in Russian. However, we do not know to what degree the grammatical system of the dialect has incorporated these “russified” types of grammatical expression. We just do not have a method to determine “degree of grammaticalization.” Perhaps the relatively low degree of grammaticalization can be explained by the complexity of the Russian process of perfectivation, with tight interweaving of “lexical” and “grammatical” elements. Another explaining factors is the high level of Russian-Romani bilingualism in the situation where Russian is the dominating language. NRRD native speakers, having a high degree of language competence in Russian, can borrow Russian syntactic and word-formation models on-line. Thus, their “Romani” competence includes not the concrete models, borrowed from Russian, but the possibility of this kind of borrowing itself. A similar situation can be found not only in the system of expressing tenseaspect relations but also, for example, in the voice system. The original system of passive voice expression with -jov suffix now belongs to the domain of nonproductive word-formation. The expansion of the reflexive verbs formed with -pe (-pes) particle was no doubt strengthened by Russian influence. As in the case of prepositions, this refers primarily to loans of specific Russian verbs. Romani reflexive verbs exhibit a plethora of meanings carried over from the Russian sources: direct-reflexive, indirect-reflexive, passive, etc. (cf. Ventcel’ 1964: 74–75). At the same time, NRRD has a special analytical passive construction formed with a linking verb and a participle. which, although rarely used, are known from other Romani dialects and are also supported by Russian influence. Unfortunately, the insufficient number of attested passive forms does not allow us to reach a final conclusion concerning the relationship between these two “new” types of voice meaning expression, their possible connection to the category of aspect (as in Russian), etc. In the nominal domain, the question of direct calques vs. grammaticalized changes in the language system can be put for each case meaning change separately (cf. Ventcel’ 1964: 52–54).

321

322 Aleksandr Yu. Rusakov

3.

Code switching and code mixing phenomena in NRRD

3.1 The concepts of code switching and code mixing With regard to the use of NRRD in different communicative situations, we should note that the dialect shows a very high degree both of code switching and of code mixing phenomena. Whereas the term “code switching” is understood by scholars in more or less the same way, that is, in terms of a shift from one language to another during a monologue or a dialogue, there is no such unanimity about the meaning of the term “code mixing”. Some scholars use it of unrestricted insertion of lexical elements of one language into another language (Hock 1986: 480). Others define it as “the alternate use of two linguistic systems within a clause” (Bhatt 1997: 223). The latter approach treats code mixing as a variant of code switching and implies the possibility of speaking two languages at once. The issue is not merely terminological; its answer relies on the theoretical understanding of the phenomenon. Below we will understand the term code mixing as “the use of unadapted lexical elements of another language in the speech.” Thus, we admit that the possibility of an opposition between code switching and code mixing, though this has yet to be demonstrated. The phenomena of code switching and code mixing have been extensively discussed in recent literature on language contact. On one hand, many investigations have studied the specific factors that determine code switching from a sociolinguistic or pragmatic point of view: what the conversation is about, what real-world objects are referred to, who is spoken to, what is the internal structure of the text. Some attempts have been made to place the phenomenon of code switching in a general anthropological approach to linguistics. On the other hand, much attention is paid to different kinds of structural and functional constraints favoring or disfavoring code mixing (on any understanding of the term). Such constraints include: isomorphy/non-isomorphy of syntactical constructions in the contacting languages, matching/non-matching of grammatical categories, constraints in terms of the notions of Government and Binding theory, etc. (from the recent works on code switching and code mixing cf. Myers-Scotton 1987; Bhatt 1997; Auer 1998). Much less attention has been paid to another aspect of the problem: the possibility of a connection between degree of interference — first of all, grammatical interference — and the occurrence of code switching or code mixing, and if such as connection exists, what its nature is (cf. e.g. Matras 1998: 286).

The North Russian Romani dialect 323

3.2 Code mixing in NRRD 3.2.1General In the NRRD materials, we find, in addition to inter-sential code switching, where the shift from one language to another takes place on the border of two sentences, there are also some examples of intra-sential code switching. One of the most distinct features of NRRD, which sets it apart from many other Romani dialects, is the extensive use of grammatically unadapted Russian lexical elements: (22) Da nat, me prosto na dumind’om, so me tut date vstrechu ‘No, I didn’t just think that I meet you here’

This is exactly what we called code mixing above. In order to reach an understanding of this phenomenon in NRRD we have to answer two closely related questions which have already been touched upon: 1. Can a NRRD text which includes includes unadapted Russian lexical elements really be seen as a text in one language? 2. Can a strict border-line be drawn between phenomena of this kind and classic code switching? Before answering these questions we have to discuss some characteristics of the use of Russian lexemes in NRRD. 3.2.2Adapted and unadapted forms First of all, we should note that in all Roma dialects there are very strong mechanisms of adapting borrowed words. The use of special morphemes inserted between the stem and the inflections in borrowings motivates treating them as a special component of the lexicon which differs markedly from the inherited vocabulary (for the most detailed account of this problem see Boretzky & Igla 1991). In the nominative singular, masculine and feminine borrowed nouns take the endings -o and -a, respectively. (Neuter Russian nouns are assimilated to the masculine gender when adapted.) Since in borrowed nouns the last syllable cannot be stressed, the stress is moved to the first syllable of the stem whenever necessary. In the oblique cases of the singular, borrowed masculine and feminine nouns take the endings -os and -a (without palatalization of the previous consonant), respectively. In the plural, all nouns take -y/-i in the direct case and -en (without palatalization of the previous consonant) in the oblique cases. Borrowed adjectives form a special inflexional class in NRRD as well as in other Romani dialects. The words which belong to this class are characterized by with a possible shift of a stress and lack of differentiation in gender. In the oblique cases, a suffix -on- is inserted between the stem and the ending. In verbs, a suffix -in- is inserted between the Russian verb stem and the Romani inflection.

324 Aleksandr Yu. Rusakov

Thus, the general adaptional mechanism in NRRD are the same as in the other Roma dialects. However, there is one very peculiar difference. Russian adjectives and nouns are practically always used in NRRD in adapted form: (23) tu dzhasa de foro… peskire podrugenca uzhe ‘you’ll go to the city… with his/her female friends already’

Russian verbs, on the other hand, are mostly used in an unadapted form and are conjugated according to the Russian model: (24) tu poedeš de foro ‘You’ll go to the city’ pišem po-romanes ‘We write Romani’

Thus, in the syntactic questionnaire there are 103 unadapted and 17 adapted occurrences of Russian verbs. In one of the texts, there were 35 cases of unadapted verb forms and not a single adapted one. 3.2.3Why the differences between the parts of speech The question thus arises in what manner we should interpret the difference in use of Russian verbs and of adapted nouns and adjectives. Common sense and the necessity to build a simple and psycholinguistically feasible model makes it preferable to regard all the cases of use of Russian nouns, adjectives and verbs as linguistically similar and generated by the same mechanism allowing native speakers of NRRD to use in their Romani speech practically all Russian words (for some exceptions to this rule, see below). But for the nouns an additional rule exists, which is probably active on the sentence production level rather than in the lexicon. Support for this assumption is found in the cases of where the singular and plural forms of nouns are independently adapted: (25) drugo (nom.sg), cf. Russ. drug ‘friend’ druzji (nom.pl), cf. Russ. druz’ja ‘friends’

A separate question is that of the cause of this difference between nouns and verbs. A possible explanation is the simpler nature of noun adaptation mechanisms. Indeed, for the masculine nouns the adaptation mechanism consists only of the addition of the -o suffix to the original noun form, for the feminine nouns in the application of a simple stress rule. The Romani noun has a clear agglutinative structure, thus the formation of the indirect case forms consists of the adding of case affixes to the adapted original form. Adjectives are adapted also by replacing the Russian ending by a Romani one. The adaptation mechanism for verbs is more

The North Russian Romani dialect 325

complicated, it includes the separation of the Russian verb stem, a complicated operation due to the complex character of the Russian verb morphonology. The Roma verb paradigm also contains rather difficult morphonological rules. A further point can be made. Because of the high degree of interference we can nearly always establish a simple correlation between the Romani and the Russian grammatical forms. The absence of the infinitive in the Romani grammatical system constitutes an exception to this generalization which might provoke the intrusion of some unadapted Russian infinitive forms at an early stage. This supposition is supported indirectly by the Hungarian Romani dialects, where Hungarian finite verb forms are borrowed in an adapted form, while the infinitives are unadapted (Leksa Manuš 1979: 252). Then the use of unadapted Russian infinitives might trigger the unadapted use of finite verb forms. 3.2.4A grammatical model for code mixing in NRRD There has been an intensive discussion in the literature of the questions whether similar “mixed codes” have one or two different underlying grammatical structures, and whether there exist some kind of transformational rules, which help to make a transition from one structure to another one (cf. Sankoff & Poplack 1981; Woolford 1983; Bokamba 1988). In cases of strong interference, such as NRRD, this question seems somehow irrelevant, though. The following grammatical model may be postulated for NRRD: The surface syntactical structure of the dialect is identical to that of Russian one. The sentence in NRRD represents by itself the prosodic, syntactical and semantic frame, which coincides with the corresponding Russian sentence9 and into which both Romani and foreign (Russian) lexical elements can be put. The Russian elements may be used in a Russian grammatical setting. The assumption that the examples which were define as “mixed code” can be interpreted as a Romani speech with unlimited insertion of Russian lexemes, receives at least indirect support from metalinguistic comments by native speakers. In the answers to the syntactic questionnaire, which implies the usage of Romani language par excellence, we can find many unadapted Russian lexical elements: (26) me tas’a poedu de foro ‘tomorrow I’ll go to the city’

Interestingly enough, answers of this kind are often followed by comments such as: (27) u nas tak i govorjat: poedu (Russ) ‘we often say it just like this: “pojedu”’

proving, on one hand, that the examples of code mixing do not change the Romani character of the text, and on the other, that speakers are nevertheless able to identify the inserted elements as Russian. The monitoring of the translation provoked by the

326 Aleksandr Yu. Rusakov

request to translate a text from Russian into Romani makes the speaker look for a translation which would not contain borrowed lexemes: (28) atasya me poedu de foro, atasya me dzhava de foro, možno skazat’, so tradava de foro atasya. ‘I’ll go (Russ) to the city tomorrow, I’ll go (Rom) to the city tomorrow, it is possible to say I’ll go (Rom, “archaic” word) to the city tomorrow’

Very often this search of purist translation variants is accompanied by comments in Russian: (29) A kak bystrej, tak i govorjat ‘It’s faster, that’s why we say like this’ (30) My russkie cygane, u nas mnogo russkix slov ‘We’re Russian Roma, we have many Russian words’ (31) Redko, kto znaet cˇisto cyganskij jazyk, potomu cˇto my živ’om sredi russkix i uzhe kak-to samo po sebe slova pereputalis’ ‘It is very rare, that someone knows pure Romani language, because we live among Russians and somehow all the words got mixed up’

It is worth mentioning that not only unadapted but adapted Russian elements also are “introspected” as Russian inclusions: (32) psirlas, — hodine, èto po-russkomu. ‘psirlas, — hodine, that’s in Russian.’

This fact gives indirect support to the hypothesis put forward above of unadapted verb and adapted nouns being elements of the same language level in NRRD. 3.2.5Clear cases of code switching in NRRD Maybe we can see even more clearly the common nature of adapted and unadapted Russian lexemes if we compare these cases with undisputed examples of code switching. In our recordings, there are numerous examples of switches from Russian to Romani and back. We can single out several factors influencing such shifts. On the textual level, to begin with, such shifts can be found when the narrator makes a transition from his own narration to the direct speech of one of the characters of the story:10 (33) Yake ev avya, stadina nahmureno leste, nadvinuto po shero i do zalo dorik, kogda ev de peskiri komnato, o ray pal: nu cˇto, cˇto Pjotr, cˇto ty takoj? Kak cˇto, ved’ ona menja vygnala, von otlupili menja pljotkami… ‘And he came. The hat is on his head, and (goes) into the hall. When he (moves) to his room, the count (goes) after him: “What has happened to you, Petr?” – “What do you mean, she chased me away, they beat me up with a whip.”’

This kind of shift occurs very often in narrative texts (short stories, fairy tales, etc.). In the stories recorded by us, wherever such a transition from the narration to the

The North Russian Romani dialect 327

direct speech takes place, we find in the vast majority of cases typical examples of classic code switching. An analysis of these cases shows us that they most often involve a transition to the direct speech of non-Roma characters. Thus, this transition can be explained purely by pragmatic needs. However, the transition to the direct speech of non-Roma characters is not always followed by code switching. We think that this phenomenon plays a very specific stylistic role, marking the contrast between the direct speech of the characters (including those of fairy tales) and the narration itself. Interestingly, this kind of transition shows up in the most important and culmination parts of a story. For instance, in one short story about a man who was buried alive (the plot is quite widespread), we can find many examples of direct speech but they are all in Romani, no matter which nationality the character has. The only exception is found in the most culminational moment of the story, when the man who was buried alive addresses the robbers who excavated him. In this connection, it is interesting to point out that the end of a story is also marked very often by the transition to Russian: (34) na ètom i koncˇilos’ ‘that’s how it ended’ etc.

An additional proof of the fact that code switching in a story marks the border of stylistically different fragments of the text, can be found in our records: often the transition to Russian language takes place after the end of the direct speech. Another question, which is quite interesting, is to what degree a narrator realizes that he makes the shift from one language to another. It is very typical when a speaker continues to tell the the story in Russian after the direct speech is over. Then he suddenly recollects this fact and after a little pause switches back to Romani. In general, after listening to the records of the Romani texts, one gets an impression that, for some degree, it does not really matter for the narrator in which language to speak, and that the transitions from one language to another are made because of the need to make the text more “picturesque” stylistically, rather than to tell some specific fragments of the text in one language or another because of their genre.11 This supposition does not deny, of course, the fact that the transition is made in some particular parts of the texts (direct speech, commentaries, etc.). At the same time, at every single moment the narrator can realize which language he speaks, though sometimes he seems to forget to speak Romani, as in the above mentioned examples, when he keeps on talking in Russian, though the original setting was to tell this part in Romani.12 A slightly different situation can be found in the texts dealing with everyday life. In this case the transition to Russian language seems to be caused not by stylistic but by thematic reasons. A certain topic can make the speaker to use many Russian words. That, in its turn, causes the speaker to make a transition to Russian language, because it is easier to talk about this in Russian, the situation quite typical for code switching.

328 Aleksandr Yu. Rusakov

3.2.6The borderline between code switching and code mixing Let us return to the second question put above: Is there any strict boundary between code-mixing and code-switching? This question has been answered in the negative by quite a number of specialists (e.g. Myers-Scotton 1997: 23–24). Nevertheless we can try to distinguish between the two phenomena from a formal point of view. As code-mixing is typical of the Romani speaking only (when they speak Russian they do not use Romani elements in their Russian speech), we should distinguish between the cases of Romani speaking that includes Russian lexical elements and the cases of speaking Russian. The following test criterion of the Russian character of the text and, thus, the switch to Russian could be proposed. In the speech of NRRD-speakers there are the elements which mark speaking Russian. These elements are never used in the texts we have the reasons to suppose to be Romani. From the “operational” point of view we can postulate that the word directly preceding and the word immediately following such an element inside one sentence could not be both the Romani ones Among these elements first of all are Russian nouns in oblique cases in Russian grammatical form. We have already mentioned above that Russian nouns, when used in NRRD, are nearly always used with Romani grammatical setting. The exceptions are those of the nouns which are used in the idiomatic adverbial and prepositional phrases. (35) so vremenem ‘as the time passes by’ rano utrom ‘early in the morning’; the use of nouns in oblique cases in the meaning of adverbs (36) rundja golosom ‘yelled, lit. cried with the voice’

also the cases when nouns are used with Russian numerals: (37) ev dodzhindya do sto let ‘He lived till he was one hundred years old’.

For nouns in the direct case it is often difficult to define whether they are adapted or not, for example for Russian feminine gender nouns with non-final stress (e.g. doróga) “adaptation” flexion coincides with the Russian one and there is no stress shift. Thus, if we come across the form of Russian nouns inflected according to Russian grammatical system, we can conclude that this fragment is pronounced in Russian. Indeed, in our texts it is almost impossible to find the Russian oblique case forms in the environment of Romani lexical elements. Russian personal pronouns as well as adjectives in oblique forms inflected according to the Russian grammatical system could be also considered as the indicators of the Russian character of the text. The texts are lacking Romani forms of the Russian verb byt’ ‘to be’.

The North Russian Romani dialect 329

There are, however, Russian elements which being used in speech of Roma do not indicate code-switching; such elements are either in the relations of free variation with their Romanified variants or simply do not have Romani equivalents. Among these elements are 1. some verb forms, already mentioned above. As we have pointed out earlier, Russian verbs can be used in an adapted form, though the cases of unadapted use are much more frequent. It is interesting to note, that our oldest informant, a 90-year-old Roma woman L. M. Stepanova, used more adapted verb forms than other members of her family (below we discuss how to distinguish different styles or registers in Romani speech). On one hand, it is possible that that the use of adapted/unadapted forms depends on some individual lexical characteristics of the verbs. Every single verb can in principle be used in its adapted as well as in unadapted form. There are some verbs however, which are used in their adapted form much more frequently than in unadapted form. 2. adverbs formed from adjectives, which occur both in adapted and unadapted forms. Cf.: bystro ‘quickly’ — bystres. We could not find any regularity in the distribution of these forms. 3. function words: a very complicated relationship exists in NRRD between Romani and Russian function words. So, for example, we have in our texts both Romani kai and Russian gde ‘what’. Of course, this method of differentiation between Romani and Russian “pieces” of a NRRD text is preliminary and rather relative. Sometimes this method does not allow to achieve an unambiguous result. Interestingly, the code mixing situation in modern NRRD is very much like that found in the collection of Dobrovol’skij (1908). (38) a my dumali, so rashaj, azhno kozlo (Dobrovol’skij 1908: 5) ‘we thought, that a priest, but it’s a goat’

However some differences can be found. For instance, in the texts Dobrovol’skij (1908), practically all Russian nouns and adjectives are used in an adapted form. As for verbs, the ratio between adapted and unadapted forms in the texts collected by Dobrovol’skij is a little bit different than that in the modern NRRD texts. In the prose text of the collection the above mentioned ratio is around 80 to 70. The same verb forms could be used both in their adapted and unadapted forms. Nevertheless, some verbs tend to be used in adapted forms, when another ones almost always are used in their unadapted variant (e.g. moch’ ‘to be able’, dolzhen ‘to have to’, and other modal and half-auxiliary verbs). It is interesting that in the prose texts of the collection we found only two examples of code switching, despite the fact that many of the texts are quite similar with the texts recorded by us.

330 Aleksandr Yu. Rusakov

Another curious fact is that these two cases of code switching in the Dobrovol’skij’s texts are found on the border between plain narration and direct speech: (39) Lyja te schupine. “Schupine sygydyr”, kricˇit na zasekax sedja, “Da u buje da gazhes” (Dobrovol’skij 1908: 64) ‘Started to touch. “Touch faster”, he yells, sitting near an abattis, “you damned man!”’

In this example the narration is marked by the transition to Russian language. Here too, as in the case with the borrowing of Russian prefixes, the dialect shows very high degree of stability. Thus, in NRRD speech Russian lexical elements can be used with practically no restrictions. The number of Russian words can reach such a level that there is an illusion of a “mixed Russian-Romani text.” However, as we tried to show, there is a possibility to differentiate Romani texts and the cases of code switching. Most probably, this differentiation has some kind of psycholinguistic basis. We think, that both code mixing and grammatical interference (at least, some of the phenomena, which are traditionally considered to be grammatical interference) have a similar nature. Both code mixing and grammatical interference are possible because of the existence of some common processes, which allow the speaker to use the resources of Russian language right in the moment of locution. In the first case, the speaker uses “material” lexical elements (some of them, especially nouns and adjectives, are adapted according to special rules). In the second case we deal with translation loan. It can be a word-formation model of individual Russian words which is borrowed, or it can be a whole syntactic model. It is important to note that this kind of phenomena are very wide-spread and are used practically by all the native speakers of the dialect, thus being an integral part of their language competence (cf. for example some speculations on code mixing as a discourse mode in Pyemi 1993). We should also point out that this approach erases the distinction between interference processes in language and interference processes in speech, which was noticed by Weinreich (1953). The observed phenomena being realized in the moment of speech surely influence the language system. Word-formation and syntactical loan translations gradually replace the previous modes of grammatical expression, and code-mixing-like phenomena lead to the erosion of the lexicon. Undoubtedly interference phenomena and the code-mixing influence each other. Maximal approximation of NRRD syntactical structure to that of Russian results in the abolishment of the restrictions to use code-mixing, and code-mixing in its turn provides growing similarity of two language systems (by inclusion of Russian grammatical elements into Romani texts first of all).

The North Russian Romani dialect

3.2.7The domain of validity of the NRRD situation In connection with the situation described above two questions arise: 1. To what extent is the code switching and above all the code mixing situation with code mixing in NRRD typical also of other Romani dialects and other languages with intensive language contacts? 2. Can we assert that this situation is typical for NRRD in general, and not for some specific social register? When answering the first question we have to note that code mixing phenomena are found in some other Romani dialects such as the Ukrainian and Southern Russian dialects analyzed by A. P. Barannikov (1934: 139 and elsewhere) and in non-Romani languages of the previous Soviet Union such as some German dialects (cf. Verešcˇagin 1965; also Smirnitskaja & Barotov 1997: 83–96). At the same time code mixing of such a type is not common in a large number of Romani dialects,13 although “classic” code switching is found almost everywhere. A very interesting situation can be observed in the Romani dialects of Kabudzhi in Albania and the Agia Barbara dialect in Greece. Some time ago the native speakers of these dialects were in a very close contact with the native speakers of Turkish. At the moment only some members of the elder generation of these two dialects (Kabudzhi and Agia Barbara) are somewhat familiar with Turkish (cf.: Cortiade 1992: 8; Messing 1988: 27–28). Nevertheless, in both dialects we can find many originally Turkish verb with the Turkish grammatical setting. Thus, we can assume, that earlier, while these dialects were still in close contact with Turkish, code mixing type of phenomena were typical for them. Then, after the contact was over, some kind of grammaticalization process took place and Turkish verbs, which became real borrowings, formed a separate paradigm. It is interesting, that the modern state of both dialects is not characterized by code mixing. We cannot find any unadapted loans either from Albanian, or from Greek. It is also difficult to say if code mixing is typical only of some specific (social of stylistic) variant of NRRD or it is a feature of NRRD in general. Practically all our informants use code mixing in their speech. The situation in the dialect described by Dobrovol’skij proves that this phenomenon has undergone some kind of “institutionalising”. Nevertheless, we cannot deny that there may exist a variant of NRRD without use of unadapted Russian loans.14 Researchers postulate that for some Romani dialects there are two “styles” or registers of speech: “higher” and “lower”. For instance, such a situation is reported to exist in Finnish dialect of Romani language (cf. Valtonen 1972). In this case “the higher style” is similar to “an older, more original form of the language, while “the lower style” refers to a more recent, grammatically simplified Romani, which is heavily influenced by Finnish” (Vuorela & Borin 1994: 10–11). However, as the authors report, these language forms are better seen as a continuum, and the degree

331

332 Aleksandr Yu. Rusakov

of interference depends on individual speaker characteristics, above all age (ibid.). Possibly NRRD also represents such a continuum, as suggested, for instance, by the following. In spite of the absence of texts without code mixing in our material, the number of unadapted Russian lexemes varies between individual informants15 and depending on text type. Unfortunately, the material is not large enough to allow a statistical analysis, but presumably the situation in question can be described by Labov’s variable rule model. It would be interesting to carry out a sociolinguistic investigation of NRRD, analyzing how linguistic variables (including usage of unadapted Russian lexemes/adapted borrowings/genuine Romani words) are used depending on social characteristics of informants and type of communicational situation. Such an investigation could help to answer in a more precise way the question of the psycholinguistic reality of on one hand code switching-models, implying bilingualism in the speaker and on the other the Labovian model, which involves a monolingual speaker using variable rules (cf. Labov 1970), in consonance with the code mixing notion. Presumably, a concrete speaker can use both strategies in his language performance. Very interesting is the question of the relationship between the NRRD type of dialects on one hand, and mixed Romani dialects, on the other. Dialects such as Finnish Romani are evidence for a gradual transition from a dialect with strong interference to a dialect which has lost its original, “genuine” grammatical component. 19th century British Romani may be another case in point (cf. Yeloyeva & Rusakov 1990: 39–40). There is an apparent contradiction here with the rather widespread viewpoint according to which para-Romani dialects are artificial or semi-artificial formations like argot (a review of the origins of this type of dialects can be found in Bakker & van der Voort 1991; cf. also Boretzky & Igla 1994). We must here make a clear distinction between the history of the “para-Romani dialects” and their current synchronic state. Indeed, at least some of the paraRomani dialects are very close to argot both functionally and structurally, as they – – –

are used as secret languages; are not the first language to be acquired by children; the ways of the creation of new words in these dialects resemble argot (periphrastic constructions are widely used, some word-formation suffixes are used in a specific way, existence of “cryptolalic” formations, etc.) (cf. Eloeva & Rusakov 1990: 41–43, Bakker 1995: 134).

Another question is how the current state of these dialects came about. We think that the most “attractive” view is that of D. Kenrick, according to which present-day British Romani is a result of a gradual development from a Romani dialect, during which a great number of English grammatical constructions penetrated into the language system (Kenrick 1979; a similar view is expressed in Thomason & Kaufman 1998: 103–04).

The North Russian Romani dialect 333

Can we consider the NRRD type of dialects to be a stage in the development of para-Romani dialects? In other words, can a wide use of code mixing stimulate the borrowing of the grammatical system of a neighbouring language? We can imagine that the paradigms which were borrowed into NRRD along with unadapted Russian lexical elements were later on extended to genuine Romani words. However, it seems that both code mixing phenomena and “the borrowing of another grammatical system” should be regarded in the frame of one process, that is a convergence of a Romani dialect and a neighbouring language. The NRRD language system is becoming more and more similar to that of Russian. (In reality this process realizes in the competition between the elements of the original system and the elements inducted by interference.) Grammatical meanings can be expressed both by Romani and Russian elements: At some point, the system of the dialect allows Russian lexical elements, which have the same syntactic functions as Romani ones, to be used freely. It is possible that in other Romani dialects the equivalence between Romani and non-Romani grammatical elements, at some stage of development of a dialect, could result in their variational usage in Romani texts. Later the original Romani ways of expression of grammatical meaning would be replaced by non-Romani ones (cf. also Myers-Scotton 1998).16 In our opinion, one of the advantages of the suggested model is that the language changes are seen and treated as rather homogenous. It can be seen, for example, from the priority of the lexical changes in the interference processes. Some details of this process, however, are still to be uncovered. It is unclear, for instance, exactly in which way the original grammatical system is replaced. In this connection, we would like to point out the following: 1. From the psycholinguistic point of view, para-Romani dialects are extreme points on the way to the type of language contact situation what in the beginning of the century Šcˇerba called “one language with two terms”.17 In this case it is not clear if we can speak about a bilingual situation (in the broad sense) or if we are dealing with one language system, which has lexical language variables (as it was understood by Labov). In any case, it is quite evident that some kind of psycholinguistic research of Romani dialects should be carried out. 2. From a sociolinguistic point of view, there is no doubt that those dialects which have a high degree of code mixing as well as para-Romani dialects are two stages on the way to what is called with an elegant term “language death”. Indeed, out of 13 symptoms of “language death”, listed in Dressler (1996), the dialects of NRRD type share 11. Those 11 symptoms are: – – – – –

massive lexical loans from the dominant language; borrowed lexical items that are not integrated but are used like citation words; borrowed words that are not integrated and are used like regular words; indigenous words replaced by borrowed words; speakers inability to use the indigenous lexicon appropriately;18



334 Aleksandr Yu. Rusakov

– – – – – –

loss of use of proper names of the recessive language; loss of productivity of word formation rules; phonological alignment with the dominant language; decreased use of subordinate clauses; replacement of synthetic by analytic constructions; use of only a casual, conversational style of the language.

As for the remaining symptoms on Dressler’s list, it is questionable if Romani dialects of this kind are characterized by “decay of inflectional morphology.” Also, the speakers of these dialects definitely do not have “negative attitudes … toward their language”. The absence of the last feature can be explained by the role of the Romani language as an “ethno-preserving” factor. Because of the heterogeneous nature of para-Romani dialects, the application of the symptoms to them is more problematic, in particular those which deal with lexical borrowings. We cannot deny, however, that they represent probably highly striking examples of “language attrition”. At the same time, both NRRD type of dialects and para-Romani dialects show a relatively high degree of language stability. The reasons for this stability are evident. There is a need to preserve the Romani language as a special, secret language, which cannot be understood by outsiders (cf. Thomason & Kaufman 1998: 103–04). That is why this stability exists despite the dominance of the language of the neighbouring population and the consequent very high degree of interference. In this connection we find very interesting N. B. Vaxtin’s ideas about the possibility of preservation of a language as an “ethnical identifier”, notwithstanding drastic changes of the language. It would very interesting to investigate the role of intentional efforts of language community members in the process of preserving the language (cf. Vaxtin 1998: 122). For example, all the original grammatical elements in para-Romani dialects have been replaced, but that is not in contradiction with the goal of preserving the Romani character of these dialects.

Notes * This work was supported by the Research Support Scheme of the OSI/HESP, grant No: 616/1996. 1. According to this traditional division, the Romani dialects in the circum-Baltic area, like those in the rest of Europe, are divided into two main branches, on one hand, non-Vlach dialects, whose speakers migrated to Central and Western Europe in the 15th century, and Vlach dialects, which spread from Romania to other parts of Europe in the second part of the 19th century and later. Here, we shall speak only about the non-Vlach dialects, which are divided into several groups: – – –

the Baltic group: the dialects of Central Poland, Lithuania, Latvia and Estonia, the North Russian Romani dialect and the very close related dialect of Belarusian Roms; the Sinto dialect; the dialects of Finnish Roms.

The North Russian Romani dialect 335

In the area, we also find Swedish and Norwegian “para-Romani” varieties (see below). All these dialects could according to some linguists be united in the North branch of Romani (non-Vlach) dialects (Boretzky, Igla 1991: 7–8). 2. The materials for this article (syntactic questionnaire and texts of different genres) were mainly collected through fieldwork in 1984–88 and 1996 in the Vyrica and Mixajlovka villages near St. Petersburg, in collaboration with F. Eloeva, E. Perexval’skaja and O. Abramenko. 3. Verb derivation by prefixation is treated in traditional Russian grammar as a word-formation process, regardless of whether the prefixes modify only verbal aspect, or the lexical meaning. 4. Imperfect is used rather regular as the conditional in NRRD. 5. It is possible that besides the Russian model budu pet’ ‘I shall sing’ the NRRD model was influenced by ” the Ukrainian one as in xot-im-u want-take-1sg.pres. ‘I’ll want’ (Boretzky, 1989: 369). If it is true, we have here a very interesting example of a “contamination” of the influence of the two synonymous models, one of which influenced the syntactic structure while the other the “lexical” semantic structure of the future construction in NRRD. It is interesting that the other model of analytical future more close to the Russian one exists in NRRD: me avava te bagav ‘I’ll sing’ (lit. ‘I’ll be sing’). This form is videly represented in Dobrovol’skij’s texts (1908) and rarely in our materials. 6. Of course, we should clearly realize, that this “common-Romani language” is, firstly, more a linguistic construct, and, secondly, is a product of interference influence of Balkan languages (first of all, of Greek language) too. 7. For example, in the texts collected by Dobrovol’skij as wel as in our texts we can find quite a popular verb te hodinas ‘to walk’, which was borrowed from Russian and then adapted. It is interesting to note, that this verb forms a quasi-aspectual pair with a Romani verb te dzhas. This pair is parallel in all respects to the Russian verb pair khodit’ – idti ‘to walk’ (in general) – ‘to walk’ (purposefully). In both pairs, the verbs are opposed to each other by the distinctive feature of unidirectionality. In the dialect we can find a few more pairs which would contain two verbs of movement, where the first verb is a “genuine” Romani one, the second one is an adapted Russian verb. For example: te prastas – te begines ‘to run’, te urjes – te letinas ‘to fly’. Unfortunately, we do not have enough material to decide if an opposition similar to the above mentioned one exists between the members of these pairs. 8. In that connection, note that NRRD is characterized by loss of productivity of word formation rules, the feature which was mentioned by Dressler (1996) as one of most symptomatic ones for a situation of language death. Unfortunately, we cannot examine this issue in detail here. 9. Pfaff (1979) expressed similar views on the factors favorable for code switching (cf. also Thomason & Kaufman 1988: 154). 10. We should note, however, that the majority of the texts analyzed are quite homogenous genrewise — fairy tales, tales, “true stories”, etc. 11. Of course, we should take in account that the researcher is Russian, and this factor can provoke the informant to start speaking Russian. At the same time the informant is “controlled” by the original directions (also given by the researcher) to speak Romani. 12. Sometimes, however, it seems that it is easier for the narrator to speak Russian and he forgets to return to Romani. 13. It should be noted, that very often there is no reliable evidence to decide whether code mixing exists in the dialect. 14. A perfect example of such a variety is the artificial “standard Rom language” created in the Soviet Union in the 1920s on the basis of NRRD. It is interesting, however, that in the plays (that

336 Aleksandr Yu. Rusakov

is the genre of literature which is oriented mostly towards spoken language) written in this language by Rom-Lebedev, there are cases of code mixing: Gadzhengire buty kerna, kos’at, zhnut. (Rom-Lebedev 1931: 29) They do gajos’ work, mow, reap Unadapted verb forms are used here in the place of rarely used Romani verbs. (On the problems of standard language based on NRRD cf. Rusakov 1997.) 15. For example, there are fewer cases of code mixing in the speech of our oldest informant L. M. Lobanova, whom we have mentioned already. 16. We have mentioned above extensive overlapping of the Romani grammatical system (in the NRRD type of dialects) and the grammatical system of the dominating language. We can suppose, that in para-Romani dialects grammatical system coincides with that of the dominating language even on the surface level of morphology, that is on the level of “material” signifiers. 17. In modern psycholinguistics the notion of compound bilingualism can be considered as an equivalent to the Šcˇerba’s term. 18. This can be also found in the fact that very often speakers of NRRD (as the speakers of other Romani dialects) use the borrowed words instead of Romani words, familiar to them, even in the answers to the questions of the questionnaire.

References Auer, Peter (ed.). 1998. Code-Switching in Conversation: Language, Interaction and Identity. London: Routledge. Bakker, Peter, Cortiade, Marcel (eds.). 1991. In the margin of Romani. Gypsy languages in contact. Amsterdam: Institute for General Linguistics. Bakker, Peter, van der Voort, Hein. 1991. Para-Romani languages: an overview and some speculations on their genesis. In: Bakker & Cortiade, 16–44. Bakker, Peter. 1995. Notes on the genesis of Calo and other Iberian Para-Romani varieties. In: Matras, Yaron (ed.), Romani in Contact. Amsterdam: Benjamins, 125–150. Barannikov, A. P. 1934. The Ukrainian and South Russian Gypsy Dialects. Leningrad. Beskrovnyj, V. M. 1971. O vtoricˇnoj prefiksacii glagola v indoarijskom jazyke (na materiale cyganskogo jazyka. In: Vostocˇnaja filologija, Moscow, 14–20. Bhatt, Rakesh M. 1997. Code-switching, constraints, and optimal grammars. Lingua 102: 223–251. Bokamba, Eyamba G. 1988. Code-Mixing, Language Variation and Linguistic Theory: Evidence from Bantu Languages. Lingua 76: 21–62. Boretzky, Norbert. 1989. Zum Interferenzverhalten des Romani. Zeitschrift fur Phonetik, Sprachwissenschaft und Kommunikationsforschung. 42: 357–374. Boretzky, Norbert, Igla, Birgit. 1991. Morphologische Entlehnung in den Romani- Dialekten (Arbeitspapiere des Projektes “Prinzipien des Sprachwandels”, Nr.4). Essen. Boretzky, Norbert, Igla, Birgit. 1994. Romani Mixed Dialects. In: Peter Bakker & Maarten Mous (eds.), Mixed Languages. 15 Case Studies in Language Interwining. Amsterdam: IFOTT, 35–68. Cortiade, Marcel. 1992. I Rom in Albania. Un profilo storico e sociale. Lacio Drom, 28: 3–14. Courthiade, Marcel. 1995. Phonologie des parlers rom et le diasystème graphique de la langue Romani. Thèse pour le doctorat. Paris. Dobrovol’skij, V. N. 1908. Kiselevskie cygane. St. Petersburg.

The North Russian Romani dialect 337

Dressler, Wolfgang. 1996. Language Death. In: Singh, Rajendra (ed.), Towards a Critical Sociolinguistics, 195–210. Amsterdam: Benjamins. Eloeva, Fatima A. & Rusakov, A.Ju. 1990. Problemy jazykovoj interferencii. Cyganskije dialekty Evropy. Leningrad. Friedman, Viktor A. 1985. Balkan Romani Modality and Other Balkan Languages. Folia Slavica 7: 381–389. Hancock, Ian F. 1988. The development of Romani linguistics. In: Jazayery, Mohammad Ali & Winter, Werner (eds.), Languages and Cultures: Studies in Honor of Edgar C. Polome, 183–223. Berlin: Mouton de Gruyter. Hock, Hans H. 1986. Principles of Historical Linguistics. Berlin: Mouton de Gruyter. Kenrick, Donald. 1979. Romani English. In: Hancock, Ian F. (ed.), Romani Sociolinguistics (International Journal of the Sociology of Language 19). The Hague: Mouton. Labov, William. 1970. The study of language in its social context. Studium Generale. 23: 30–87. Manuš, Leksa. 1979. O kontakte cyganskogo i vengerskogo jazykov. Sovetskoe finnougrovedenie. 15: 248–255. Matras, Yaron. 1998. Utterance modifiers and universals of grammatical borrowing. Linguistics 36: 281–331. Messing, Gordon M. 1988. A Glossary of Greek Romany as Spoken in Agia Varvara (Athens). Ohio: Slavica Publishers, Inc. Myers-Scotton, Carol. 1997. Duelling languages: grammatical strucure in codeswitching. Oxford: oup. Myers-Scotton, Carol. 1998. A way to a dusty death: the Matrix Language turnover hypothesis. In: Grenoble, Lenore A. & Lindsay J. Whaley (eds.) Endangered Languages. Cambridge: cup, 289–316. Pfaff, Carol W. 1979. Constraints on language mixing: intrasentential code-switching and borrowing in Spanish/English. Language 55: 291–318. Rom-Lebedev, Ivan. 1931. Kham dro blato. Moscow. Rusakov, Aleksandr. 1992. Imperativ v severnorusskom dialekte cyganskogo jazyka. In: Xrakovskij, V. S. (ed.), Tipologija imperativnyx konstrukcij. S.-Petersburg: Nauka. Rusakov, Alexandre. 1997. The education of Russian Gypsies: The problem of dialects. In: Vassiliadou, Maria (ed.), Education of Gypsies. Development of Teaching Material. Athens. Sankoff, Gillian & Poplack, Shana. 1981. A formal grammar for code-switching. Papers in Linguistics, 14: 3–46. Smirnitskaya, Svetlana V. & Barotov, M. A. 1997. Nemeckije govory Severnogo Tadžikistana. S.Petersburg: Institut lingvisticˇeskix issledovanij RAN. Thomason, Sarah Grey, Kaufman, Terrence. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Valtonen, P. 1977. Trends and special traits in Finnish Romani. Studia Orientalia, 47: 281–286. Vaxtin, Nikolaj B. 1998. Iscˇeznovenie jazyka i jazykovaja transformacija. In: Kozinceva N. A. & Ogloblin, A. K. (eds.), Tipologija — Grammatika — Semantika. K 65-letiju Viktora Samuilovicˇa Xrakovskogo. S.-Petersburg. van der Voort, Hein. 1991. The Romani dialect(s) of the Finnish Gypsies. In: Bakker & Cortiade, 132–151. Ventcel, Tatiana V. 1964. Cyganskij jazyk (severnorusskij dialekt). Ìoscow. Verešcˇagin, E. M. 1965. K voprosu raznojazycˇnogo oformlenija ispol’zuemoj leksiki v recˇevyx proizvedenijax bilingvov. In Voprosy obšcˇego i romano-germanskogo jazykoznanija. Tezisy dokladov. Vyp. 1. Ufa. Vuorela, Katri, Borin, Lars. 1994. The Finnish Gypsies and Finnish Romani (ms.). Weinreich Uriel. 1953. Languages in Contact. The Hague: Mouton. Woolford, E. 1983. Bilingual code-switching and syntactic theory. Linguistic Inquiry 14: 520–36.