THEURALIC

LANGUAGES
Other works in the series

The RomanceLanguages
The Celtic Languages
The SlavonicLanguages
The GermanicLanguages
The Indo-EuropeanLanguages
The SemiticLanguages
The DravidianLanguages

Forthcoming works in the series

The Turkic Languages

THEURALIC
LANGUAGES
EDITED BY
DanielAbondolo
ROUTLEDGE

LONDON AND NEW YORK

First published1998
by Routledge
2 Park Square,Milton Park,Abingdon, Oxon, OX14 4RN
270 MadisonAve, New York NY 10016

Transferredto Digital Printing 2006

© 1998 Routledge
© 1998Editorial matterDanielAbondolo

All rights reserved.No part of this book may be reprintedor

reproducedor utilized in any form or by any electronic,
mechanical,or othermeans,now known or hereafter
invented,including photocopyingand recording,or in any
information storageor retrieval system,without permissionin
writing from the publishers.

British Library Cataloguing-in-PublicationData

A cataloguerecordfor this book is availablefrom the British Library

Library of CongressCataloging-in-PublicationData
The Uralic languages/ editedby DanielAbondolo.
(Routledgelanguagefamily descriptions)
Includesbibliographicalreferencesandindex.
1. Uralic languages. I. Abondolo,Daniel Mario II. Series.
PH14.U67 1997 96-29898
494'.5-dc21 CIP

ISBN 0-415-08198-X

Publisher's Note
The publisherhas goneto greatlengthsto ensurethe quality of this
reprint but points out that someimperfectionsin the original may be apparent

Printed and bound by CPI Antony Rowe, Eastbourne

Contents

List of Figures vii

List of Maps viii
List of Tables IX

List of Contributors xiii

Preface xv
Acknowledgments xxi
List of Abbreviations xxii
1 Introduction 1
Daniel Abondolo
2 Saamic 43
PekkaSamma1lahti
3 Fennic 96
Tiit-Rein Viitso
4 Estonian 115
Tiit-Rein Viitso
5 Finnish 149
Daniel Abondolo
6 Mordva 184
GaborZaicz
7 Mari 219
EevaKangasmaa-Minn
8 Permian 249
Timothy Riese
9 Udmurt 276
SandorCsucs

v
vi CONTENTS

10 Komi 305
Anu-ReetHausenberg
11 ObUgrian 327
LaszloHonti
12 Khanty 358
Daniel Abondolo
13 Mansi 387
Laszlo Keresztes
14 Hungarian 428
Daniel Abondolo
15 Samoyedic 457
JuhaJanhunen
16 Nganasan 480
EugeneHelimski
17 Nenets 516
TapaniSalminen
18 Selkup 548
EugeneHelimski
19 Kamassian 580
PeterSimoncsics
Index 602
List of Figures

3.1 Main historicalgroupingsof Fennic 101

3.2 North Fennicinnovations 103
4.1 Homomorphemicmonophthongand stopsequences 120
4.2 Monophthongand stopsequences conditionedby
the morphemeboundary 121
4.3 StandardEstonianlong monophthongsand
diphthongs 122
4.4 Interrelationsof caseforms of weakeningstems 128
4.5 Interrelationsof caseforms of strengtheningstems 128
5.1 Finnishmorphophonemes andoperators 152
5.2 Finnishgrammaticalcasemarking 158
7.1 WesternMari palatalattraction 223
8.1 Zyrian andVotyak vowel inventory (literary
languages) 262
11.1 Mansi andKhanty dialectgroups 328
11.2 Proto-Ostyakpersonalpronounsandpersonsuffixes 343
12.1 SelectedKhanty dialectisoglosses 359
12.2 Trjffra Khanty demonstrativepronouns 369
12.3 Trjffra Khanty indicativeverb suffixes 373
12.4 Selectedverbalpersonsuffixesin Trjffra Khanty 373
12.5 Trjffra Khanty primary imperativesuffix sequences 375
14.1 Hungariancasesuffixes 440
14.2 Hungarianverb personsuffixation: agglutinatinganalogue 448
15.1 The conventionaltaxonomyofthe Samoyediclanguages 459
15.2 An alternativetaxonomyof the Samoyediclanguages 459
15.3 The proto-Samoyedicconsonantparadigm 462
16.1 Generationof allophones:suffix of the renarrativemood 489
18.1 Middle Taz Selkupvowels 552
18.2 Sandhiin Middle Taz Selkup 556

vii
List ofMaps

11 Uralie idioms of the Volga Bend xxv

11 Komi xxvi
111 ObUgrian xxvii
iv Samoyedicand ObUgrian(with Komi) xxviii
2.1 Varietiesof Saamie 51
3.1 Fennie:North/SouthandEastlWestdivisions 97
14.1 Hungariandialects 429

viii
List of Tables

1.1 Proto-Uralicconsonantsystem 12
1.2 SelectedUralic postpositions'behind,in backof 24
2.1 Backgroundand developmentof the proto-Saami
stressedvowels 55
2.2 Consonantphonemesof easternEnontekio 57
2.3 Absolutedeclension 64
2.4 Possessive declension(guos'si 'guest') 65
2.5 Proto-Saamipossessivesuffix system 69
2.6 Samplepronounparadigms:'it', 'what', 'who' 70
2.7 Samplepronounparadigms:'you', 'someone' 71
2.8 Sampleverb paradigms 72
2.9 Samplenounphrases,illustrating distribution of
caseandnumber 87
2.10 Noun phraseswith relative/interrogativepronouns 88
3.1 Proto-Fennicshortmonophthongsandtheir reflexes 107
3.2 Proto-Fenniclong monophthongsand their reflexes 108
3.3 Proto-Fennicdiphthongsandtheir reflexes 108
4.1 Comparativeand superlativeof adjectives 130
4.2 Cardinalnumerals 131
4.3 Ordinal numerals 131
4.4 Personalpronouns 132
4.5 Demonstrativepronouns 134
4.6 Voice, mood, and tensemarkersin finite forms 138
4.7 Infinite form markers 139
4.8 Sampleverb paradigm,indicative 140
4.9 Sampleverb paradigm,conditional 141
4.10 Sampleverb paradigm,imperative 141
4.11 Sampleverb paradigm,quotativeandjussive 141
5.1 Finnishvowel phonemes 150
5.2 Finnishconsonantphonemes 152
5.3 Four-by-threematrix of Finnishcases 157
5.4 Subvariantsof Finnishnominal inflection types 164

ix
X LIST OF TABLES

5.5 SampleFinnishpersonalpronounforms 169

5.6 Demonstrativepronounsand deictic adverbs 171
5.7 Subparadigmsof sampleFinnishverbs 173
6.1 Consonantinventory of literary standardErzya 185
6.2 SelectedErzyanominalparadigms 193
6.3 SelectedErzyapossessivedeclensionsubparadigms,nominative 195
6.4 SelectedErzyapossessivedeclensionsubparadigms,s3, pI 195
6.5 SelectedErzyapronominalparadigms 197
6.6 Schematicfor Mordva verb: tenseandmood suffixes 198
6.7 Erzyaconjugation:personsuffixes 199
6.8 Erzyaconjugation,personsuffixes: first pasttense 199
6.9 Erzyaindefinite andpredicativeconjugation 200
6.10 Erzyanon-finite verb forms 205
7.1 Consonantphonemesof EasternMari 221
7.2 EasternMari noun inflection 226
7.3 SelectedEasternMari possessive declensionforms:
animatenoun 228
7.4 SelectedEasternMari possessivedeclensionforms:
inanimatenoun 228
7.5 EasternMari finite verb forms 230
7.6 EasternMari primary auxiliariesulam 'is' and lijam 'is, becomes'231
7.7 EasternMari negative verb 232
8.1 The pFU initial consonantsystem 255
8.2 The pPN initial consonantsystem 256
8.3 The ZyrianNotyakinitial consonantsystem 256
8.4 Proto-Permianpossessivesuffixes 270
9.1 The Udmurt consonantsystem 279
9.2 Absolutedeclension,singular 282
9.3 Possessive paradigm:possession in singular 284
9.4 Possessive paradigm:possessor in plural 284
9.5 Udmurt personalpronouns:singular 287
9.6 Udmurt personalpronouns:plural 287
9.7 Udmurt reflexive pronouns:singular 287
9.8 Udmurt reflexive pronouns:plural 288
9.9 Finite affirmative forms of the verb 290
9.10 Finite negativeforms of the verb 292
10.1 The consonantsystemof Komi 309
10.2 Noun declensionparadigms 312
10.3 Zyrian (kZ) verb conjugation,with a few relevantkP andkY
variants 315
11.1 pOU/pOvocalismaccordingto Steinitz 329
11.2 pOU vocalismaccordingto Honti 330
11.3 pV vowels of the first syllable 334
11.4 pO vowels of the first syllable 336
 LIST OF TABLES xi

11.5 Consonantsystemsof VVj andKaz Ostyak 338

11.6 pOU personalpronounsandpersonsuffixes 342
11.7 p Vpersonalpronounsandpersonsuffixes 342
11.8 Casesuffixes 344
12.1 First-syllablevowel inventoryof Trjffra Khanty 362
12.2 Numberandpersonsuffixesof Trjffra Khanty nouns 366
12.3 SelectedTrjffra nounnumberandcaseforms 367
12.4 Paradigmsof selectedTrj personalpronouns 369
12.5 Present-tense indicativeforms of Tromaganpan- 'puts,places' 374
13.1 lllustrationsof chief dialectalisoglosses 391
13.2 Samplemorphologicaldifferences 391
13.3 Samplelexical differences 392
13.4 Canonicshapesof stemmorphemes 396
13.5 Canonicshapesof inflectional andderivationalsuffixes 396
13.6 Stemvariants 397
13.7 Present-tense indeterminatepersonsuffixes 399
13.8 Present-tense forms of the six ablautingstems 399
13.9 Preterite-tense personsuffixes 400
13.10 Preterite-tense forms of the six ablautingstems 400
13.11 Conditionalmoodpersonsuffixes 401
13.12 Determinateconjugationpersonsuffixes 402
13.13 Determinateconjugationpresent-tense forms 402
13.14 Determinateconjugationpreteriteforms 403
13.15 Conditional(present) 404
13.16 Imperatives 404
13.17 Indicativepassive,presenttense 405
13.18 Indicativepassive,preteritetense 405
13.19 Presentconditional,passive 406
13.20 Indicativepresent,narrativeform 406
13.21 Indicativepreterite,narrativeform 407
13.22 Personsuffixes,narrativeform 407
13.23 Examplesof deverbalverb derivation 408
13.24 Examplesof denominalverb derivation 409
13.25 Noun declension 410
13.26 Possessive declension,suffixes 411
13.27 Possessive declensionstemvariants 411
13.28 Mansi personalpronouns 413
14.1 Hungarianconsonantsandtheir main orthographic
representations 433
14.2 Hungarianabstractconsonants 435
15.1 Reflexesof proto-Samoyedic*k v. *s beforedifferent vowels 467
15.2 Examplesof Samoyedicvowel correspondences 468
15.3 Caseendingsin proto-Samoyedic 469
15.4 Patternsof the plural anddual declensionsin Samoyedic 470
xii LIST OF TABLES

15.5 The systemof personalendingsin pre-proto-Nenets 471

16.1 Consonantsandtheir Cyrillic graphicsymbols 483
16.2 Vowels andtheir Cyrillic graphicsymbols 483
16.3 Occurrenceof consonantsin diagnosticpositions 484
16.4 Allomorphy: suffix of the renarrativemood 488
16.5 Consonantmorphonemes(andsomemorphonemeclusters) 488
16.6 Vowel harmony 490
16.7 Nominal personalendings 496
16.8 Nominal declension:deepstructureof forms 498
16.9 Numerals 500
16.10 Personalpronounsandrelatedpronominalforms 501
16.11 Verbal personalendings(Vx) 502
16.12 Verbal inflection: deepstructureof forms andselected
fragmentsof paradigms 504
17.1 TundraNenetsconsonantphonemes:the two main systems 522
17.2 Setof personalsuffixes in the indicativeandmostothermoods 534
17.3 Case,number,andpossessivesuffixes 538
17.4 Inflection of the personalpronouns 540
18.1 Middle Taz Selkupconsonants 551
18.2 Nominal declension,substantivalrepresentation 558
18.3 Nominal declension,adjectival,verbal, andadverbial
representation 560
18.4 Declensionof personalpronouns 564
18.5 Declensionof emphatic/reflexivepersonalpronouns 564
18.6 Verbal inflection indicative,inferential 567
18.7 Verbal inflection: auditive,conditional,subjunctive 568
18.8 Verbal inflection: debitive,optative,imperative 568
18.9 Verbal inflection, substantivalrepresentation:nomenactionis I 569
18.10 Verbal inflection, substantivalrepresentation:nomenactionis
II, infinitive 570
19.1 Serialvowel harmony 583
19.2 Kamassianconsonantism 584
19.3 Singularcaseendingsby stem-type 585
19.4 Nominal paradigmsof cFaya 'river' with plural markers 586
19.5 Relativeparadigms 587
19.6 A sampleparadigmof verbalinflection 593
List of Contributors

Daniel Abondolo, Schoolof Slavonicand EastEuropeanStudies,University

of London,UK.
SandorCsucs,HungarianAcademyof Sciences,Budapest,Hungary.
Anu-ReetHausenberg,University of Tallinn, Estonia.
Eugene Helimski, Russian State University for the Humanities, Moscow,
Russia.
LaszloHonti, Finno-Ugric Seminar,Groningen,The Netherlands.
Iuha Ianhunen, Departmentof Asian and Mrican Studies, University of
Helsinki, Finland.
EevaKangasmaa-Minn,ProfessorEmeritus,University of Turku, Finland.
LaszloKeresztes,Institutefor EastEuropeanandOrientalStudies,University
of Oslo, Norway.
Timothy Riese, Institute for Finno-Ugric Studies, University of Vienna,
Austria.
TapaniSalminen,University of Helsinki, Finland.
PekkaSammallahti,University of Oulu, Finland.
PeterSimoncsics,University of Budapest,Hungary.
Tiit-Rein Viitso, University of Tartu, Estonia.
GaborZaicz, PeterPazmanyCatholicUniversity, Piliscsaba,Hungary.

xiii
This page intentionally left blank
Preface

In basic outline, this volume follows the model of previous works in the
series, most particularly B. Comrie and G. C. Corbett (eds) The Slavonic
Languages(London: Routledge1993) and E. Konig and J. van der Auwera
(eds) The Germanic Languages (London: Routledge 1994). Like those
volumes, the presentwork focuseson the languagesof a genetic unit, i.e.
languageswhich canbe saidto be relatedto oneanotherin that they havebeen
shownto be laterdevelopmentsof an antecedent,putative,protolanguage.But
whereasthe Slavonic and Germanicgroupsare themselvesgenetic subdivi-
sionsof a much largerunit, namelythe Indo-Europeanfamily, with some425
languages(Grimes 1996), the thirty-odd languagesof the Uralic family have
not beenshownto be relatedto thoseof any othergeneticgrouping.
The total numberof peoplewho speaka Uralic languageprobablydoesnot
exceedtwenty-five million. Comparedwith languagefamilies such as Indo-
European,Semitic,Tibeto-Burman,or evenDravidian,the Uralic family may
thereforebe seenas smal1. On a global scale,however,twenty-five million
speakersis a ratherlargenumber;theAlgonkian family, for example,in North
America, with a comparablegeographicdispersionand numberof languages,
hasonly about130,000speakers.
Treatment of the present-dayUralic languages,plus one, Kamassian,
whose last speakerdied in 1989, is spreadin this volume over fourteen
different chapters.Of course, in a handbookof this size not every Uralic
languagecan receiveseparatetreatmentin its own dedicatedchapter.For the
sake of compactness,for example, the nine or ten Saamic languagesare
handled in one overview chapter, which first outlines, from a historical
perspective,the main isoglosseswhich distinguishtheselanguages(and their
dialects)from one another,then focuseson the synchronicdescriptionof one
variety, the EasternEnontekio subdialectof the Finnmark dialect of North
Saami. For similar reasons,the Fennic languagesIngrian, Votic, Livonian,
Karelian, and Veps are treated together in a chapter which is primarily
historical (Fennic); western readersin searchof further detail concerning
theselanguagesmay profitably consultLaanest1982.In the caseof languages
with strong interdialectal cleavages,in particular Estonian, Khanty, and

xv
xvi PREFACE

Selkup, but to a degreein the Mari, Mordva, Komi, and Mansi chapters,as
well, contributorshaveconcentratedon onedialect,referring to otherdialects
only where this is helpful and unobtrusive.Somefurther information on the
dialects thus sidelined may be found in the relevant historical chapters
(Fennic,Permian,ObUgrian,and Samoyedic)andin the Introduction.
Throughoutthe volume, footnotes have been forborne and bibliography
kept to a minimum, in keepingwith seriesformat. As a result,the presentation
doesnot alwaysmakeclearwhich of the ideasarethe authors'own andwhich
aretransmitted;it is assumedthat suchinformationwill not troublespecialists
and would be of little interestto the generalreader.Many generalreaders,
though, will find that the bibliographical indications, particularly those of
editorshipand publisher,areoften rathermore sparsethanthoseto which they
are accustomed;this is not always a reflection of scholarly self-effacement
but indicatesthe conditionsunderwhich the bookswereproduced.
This book was preparedwith manytypesof userin mind. The primary bias
is synchronic,but there is also considerabletreatmentof the (pre)history of
elementsof each languageand genetic subdivision. Therefore,those inter-
estedin diachrony, i.e. in changethrough time, will come by much general
and specific information in this volume concerningthe developmentof the
Uralic languages,particularly in the nodal chapters on Saarnic, Fennic,
Permian,ObUgrian,and Samoyedic.
Thoseinterestedin finding out abouta particularUralic languagewill wish
to proceeddirectly to the relevantlanguagedescriptionchapter.Theseare in
the form of brief sketches,and given the range of the material and of the
theoreticalbackgroundsof the scholarsrecruitedto cover it, it is inevitable
that they vary somewhatin both style and content.All, however, strive for
succinctness.The chaptersare orderedon a geographicalmodel and proceed
roughly from west to east; thus, thoseinterestedin the geolinguisticcontext
of a given languagemight usefully nose around in adjacentchapters.For
example,after readingthe Mordva chapter,the readerwill find much that is
instructively different or similar, from the typological perspective,in the
chapterson Mari and Finnish. Parallelhistorical insightsinto Hungarianmay
be gainedby a perusalnot only of the ObUgrian chapter,but also ofthoseon
Permianand Samoyedic.
Typological comparatists,i.e. thoseinterestedin particular constellations
of linguistic phenomenasuch as vowel inventoriesand hannony,consonant
oppositionsand gradation,negation,reflexive pronouns,loanwords,or the
makeupof the noun phrase,will probablyprefer to dip into various chapters
as guided by cross-references and the subjectindex. Insofar as the diversity
of the Uralic languagespermits, the languagedescriptionchaptersfollow a
paralleldesignin orderto facilitate suchcross-linguisticchecking.
Limits on space have meant that the Introduction cannot aspire to a
balancedand critical survey of all topics, contentiousor otherwise. It is
written with the generalreaderin mind, andaims to providebasicbackground
 PREFACE xvii

and to serveas an overtureto someof the more prominentthemeswhich crop

up throughoutthe book; readersinterestedin gaining an overall impression
of the Uralic languagefamily should begin here. For further, and com-
plementary,basic backgroundreading, one might best begin with the far-
reaching articlesby B. Comrie and P. Sammallahtiin Sinor 1988; with the
compendiousHajdu 1992; and with Decsy 1965, which is concernedwith
Finno-Ugric only but is original and rich in insights. Historical and
sociolinguisticperspectiveson the Finno-Ugric languagesand their speakers
may be obtainedfrom Haarmann1974 andTaagepera(forthcoming). For the
Samoyedicpeoplesand languages,the bestoverview is still Hajdu 1963.For
a survey of the relatively insecure status of the languagesspoken in the
northernareasof the former SovietUnion, seelanhunen1991.

Notes on Transcriptions and Other Apparatus

The defaultmodeof presentationfor languageforms, evenfor languageswith
long-established orthographies, is phonemic; slant lines are therefore
eschewedsavewhereambiguitywould arise.Phonetictranscriptions,whether
broad or narrow, are given in squarebrackets.In someinstances,language
data are presentedon a more abstract plane, as well, in the form of a
morphophonemiccode. Suchforms are given in majusculeand explainedin
the relevantchapters:seeFinnish,Nganasan,and Hungarian.
Vowels are transcribedin accordancewith the useful fiction of a three-
tongue-heightspace,with basic i e a 0 u standing roughly for the vowel
qualities of Spanishor Latin. These are supplementedby characterswith
dieresis,usedto indicatea value of frontnessor backnessoppositeto that of
the plain symbol; thus ti ij are roundedfront and i" e are unroundedback
a
vowels.A front (unrounded)low vowel is written ii, and the symbol is used
to rendera backroundedlow vowel. Phonologicallydistinct vowel length is
renderedby doubledletters,e.g. ii Ii" titi uu. Refinementsanddeviationsfrom
this usagearedetailedin situ.
To simplify the typographyof the transcriptionof consonants,palataliza-
tion is indicateduniformly by j superscript,e.g. pi tj s-i, and separatelyfrom
otherfeaturediacritics, e.g. the palatalizedpendantof hushingsis written Si.
In most Uralic publications,it is traditional to combinesuch diacritics; thus
what we write as Si herewould be ~. Labializationis indicatedin this volume
by w superscript,e.g. k W tWo Affricates aretranscribed eitherwith their release
componentwritten superscript,e.g. t S dZ , or with unit symbols, e.g. c ¢,
dependingon the phonologyof the languageconcerned.
The orthographiesof Hungarian, Finnish, and Estonian are used in an
ancillary role in the chaptersdescribingtheselanguages.The writing systems
of many of the Uralic languagesspokenin Russiadiffer considerablyin their
useof Cyrillic; details will be found in the relevantchapters.For Nenetsand
Saamithis book usesspecialwriting systems,elaboratedby the authorsof the
xviii PREFACE

pertinentchaptersin this book, which depart somewhatfrom the principles

outlined above.It was deemedappropriate,in the light of the potentialfrailty
of theselanguages,to cleaveto theserather than to imposeanothersystem
from outside.
Whererelevant,forms are segmentedinsofaras the approachof the author
and the type of transcription allow. Inflectional suffixes are preceded
by hyphen, derivational suffixes by an equals sign, as in English
neighbour=hood-s. The plus sign indicates the boundary between the
members of a compound (wind+Jall); ampersandsignals reduplication
(willy&nilly). Double hyphen precedesenclitics, e.g., --ldi, Finnish e-n--kii
'and I don't', but follows a prefix, e.g., em-- in Hungarian (orthography)
em--ez'this, (closer to speakerthan ez),. The glossesthat accompanysuch
forms have beendesignedto match them morphemefor morpheme:thus in
the Mansi form aasi-;m-n~l
designed FATHER-sl-abl, the root, meaning 'father', is
aasi; the inflectional suffix form,
-;m to its right is that of the first personsingular
(sl); and the inflectional suffix at the end of the form, form,
-n~l, is the ablative. In
keepingwith commonpractice,full stop servesto link items that are separate
in the metalanguageof the gloss, e.g. PRO.sl means 'first-personsingular
pronoun'.
Deviating from commonpractice,verb stemsare glossedwith the English
third person singular presentindicative form, e.g. (Erzya) Mordva jarsa-
'eats',contrastthe infinitive jarsa=ms 'to eat' and the third personsingular
presentformjars-i '(s)heeats'.For a list of abbreviationsusedthroughoutthe
book, seep. xxii.

Bibliographical Notes
There have been eight quinquennialinternational Uralist congressessince
1960. Thesehave convenedat venuesrotating amongFinland, Hungary,and
the Soviet Union. With the exceptionof the first congress,they have been
large events, and their proceedings,when published,usually run to several
volumes.Thus, particularcitationswill havedetailssuchas 'IB', meaningthe
secondvolume (B) in a subsetof volumes.
No uniform convention of citation has emerged, but a widely used
compromise writes CXIFU, where C IFU stands for Congressus...
internationalisJenno-ugristarum(with varying capitalization)and X stands
for the numberof the congressin question.

Perhapsout of piety, the first congressis referred to simply as CIFU,

without the '1'. Details are asfollows:
CIFU Congressus InternationalisFenno-UgristarumBudapestinihabitus
20.-24.1X.1960,Budapest.
C2IFU P. Ravila, M. Kahla, A. Raisanenet al. (eds) (1968) Congressus
secondus internationalis Fenno-Ugristarum Helsingiae habitus
 PREFACE xix

23.-28.8.1965,Helsinki: SocieteFinno-Ougrienne.
C3IFU P. Ariste, V. Hallap et al. (eds) (1975) Congressustertius inter-
nationalis Fenno-UgristarumTallinnae habitus 17.-23.VIII.1970,
Tallinn: Valgus.
C41FU Congressus quartusinternationalisFenno-UgristarumBudapestini
habitus 9.-15. Septembris1975, Budapest(I 1975, II 1980, III
1981).
C5IFU Congressus quintus Internationalis Fenno-Ugristarum, Turku
20.-27.VIII. 1980, 1981.
C6IFU E.A. Saveljevaand G.v. Fedtineva(eds) (1990) MaTepHaJIbI VI
Me)J()J;yHapO}:\Horo KOHrpecca cI>HHHo-yrpoBe}:\oB [Syktyvkar
24-30July 1985], Vols I-II, Moscow: Nauka.
C7IFU L. Kereszteset al. (eds)(1990) Congressusseptimusinternationalis
Jenno-ugristarum,Debrecen.
C8IFU H. Leskinenet al. (eds) (1995) Congressusoctavusinternationalis
Jenno-ugristarum,JyvaskyHi: Gummerus.

The following festschriftsarefrequentlycited:

BereczkiFestschrift Domokos, P. and Pusztay, J. (eds) (1988) Bereczki
emlekkonyv (Bereczki Gabor 60. szuleresnapjara),
Budapest:Faculty of Philosophyof the University of
Budapest.
Hajdu Festschrift1 Bereczki, G. and Domokos,P. (eds) (1983) Uraliszti-
kai tanulmanyok(Hajdu Peter60. szuletesnapjatiszte-
lethe),Budapest:ELTE.
Hajdu Festschrift2 Bakr6-Nagy, M. Sz. and Szij, E (eds) (1993) Hajdu
Peter 70 eves [Festschrift for Peter Hajdu on the
occasionof his 70th birthday], Linguistica. SeriesA,
Studiaet dissertationes15 Budapest:MTA Nyelvtudo-
manyi Intezet.
RedeiFestschrift Dereky, P., Riese,T., Bakr6-Nagy,M., and Hajdu, P.
(eds) (1992) Festschrift flir Karoly Redei zum 60.
Geburtstag,Studiauralica6; Unilisztikai tanulmanyok
3; Linguistica Series A, Studia et dissertationes8,
Vienna - Budapest: Institut rur Finno-Ugristik der
UniversitatWien - MTA NyelvtudomanyiIntezet.

Referencesand Further Reading

Decsy, Gy. (1965) Einfiihrung in die finnisch-ugrischeSprachwissenschaft,
Wies-
baden:Harrassowitz.
Grimes,B. (ed.) (1996) Ethnologue:Languagesof the World, Thirteenth(electronic)
Edition.
Haarmann,H. (1974) Die finnisch-ugrischenSprachen.Soziologischeund politische
Aspekteihrer Entwicklung,Hamburg:Buske.
XX PREFACE

Hajdu, P. (1963) The SamoyedPeoples and Languages, Bloomington: Indiana

University.
- - - (1992) Introduzioneaile lingue uraliche, translatedand adaptedby Danilo
Gheno,Turin: Rosenberg& Seiler.
Janhunen,J. (1991) 'Ethnic death and survival in the Soviet north', Journal de la
SocieteFinno-Ougrienne83: 111-22.
Laanest,Arvo (1982) Einfiihrung in die ostseefinnische
Sprachen,Hamburg:Buske.
Sinor, D. (ed.) (1988) The Uralic Languages:Description, History and Foreign
influences,Handbuchder Orientalistik 8/1, Leiden: Brill.
Taagepera,R. (forthcoming), The Finno-Ugric Republics and the Russian State,
London: Hurst.
Acknowledgments

My fIrst thanksgo to my contributors,who not only furnishedtheir chapters

within deadlinesbut also often made thoughtful suggestionstowards the
improving of the volume as a whole. Thanksalso go to colleagueswho read
variouschaptersand sectionsand madeuseful comments,most notablyPeter
Sherwood(University of London),TapaniSalminen(University of Helsinki),
and Endre TaIos (University of Budapest).lowe a debt of gratitude to
JonathanPrice, quondam senior editor at Routledge, for originally com-
missioning this volume, and sailing with me through helpful conversations
concerningits form andcontent.I appreciate,andherewiththank, as well, all
the other professionalsat Routledge who helped me through the years,
particularlyShanMillie, LouisaSernlyen,SethDenbo, DeniseRea,andSarah
Hall. Specialthanksgo to Nicola Mooney, who typed swathesthroughmore
than half of this book, and to my copy-editor Judith Willson for her
extraordinarilygood-naturedtaking of pains.
My largest debt is owed to my teacher,Robert Austerlitz. He was the
scholarwho introducedme to the Uralic languages,and who fIrst suggested
that I teacha surveycourseon them. It was his teachingthat mostinfluenced
my own thinking about language,and his encouragementand examplethat
fIrst got me startedon the path that leadsto this book. If thesepagesbearan
editorial stamp,it is chiefly throughmy contactwith him that thatstampwas
shaped.

July 1997
DanielAbondolo
(editor) The Uralic Languages

xxi
List ofAbbreviations

A accusative d dual
A adverbial DD definite declension
abe abessive def definite
abl ablative desid desiderative
abs absolute dn denominal
acc accusative DN UpperDemjanka
act active DT Lower Demjanka
ade adessive du dual
adj adjective dur durative
adjsx adjective-forming dv deverbal
suffix E Erzya
adv adverb EE EasternEnonteki6
aff affirmative ela elative
all allative EM EasternLiterary
anaph anaphoric Mari
aor aorist enc enclitic
app approximative ess essive
Atl Atlym Eur European
attrib attributive evid evidential
B Beserman ext extent
bot botanical Fe Fennic(= 'Baltic
car caritative Finnic')
caus causative Fi Finnish
cfv comparative FP Finno-Permic
coll collective freq frequentative
com comitative FU Finno-Ugric
commis commiserative FUF Finnisch-ugrische
cond conditional Forschungen
conj conjunctive FV Finno-Volgaic
conneg connegative G genitive
cx casesuffix gen genitive
D dative ger gerund

xxii
 LIST OF ABBREVIA nONS xxiii

ill indefinite declension mod modal

id. idem Mr Mari
IE Indo-European MSFOu Memoiresde la
ill illative Societe
imp imperative Finno-Ougrienne
impf imperfect N nominative
ind indicative narr narrative
ine inessive NdV forms nounsfrom
inf infinitive verbs
ins instrumental nec necessitative
inst instructive neg negative
interr interrogative NF North Fennic
intr intransitive Ni Nizjam
ips impersonalvoice nom nominative
Irt Irtysh NSK Nykysuomen
iter iterative sanakirja
J Jugan nsx noun-formingsuffix
Jk Jukonda 0 Obdorsk
JSFOu Journal de la Societe 0 object
Finno-Ougrienne obj object
K Konda obI oblique
Kaz Kazym opt optative
Kh Khanty ord ordinal
KM Middle Konda OZ OldZyrian
KO Upper Konda P partitive
kP Komi -Permiak P Pelymka
Kr Krasnojarsk p plural
KU Lower Konda part participle
kY Yaz'vaKomi partic particle
kZ Komi-Zyrian pass passive
L Lozva pejor perjorative
lat lative perf perfect
lim limitative pF proto-Fennic
Lit Literary pFS proto-Fennic-Saamic
lit. literally pFU proto-Finno-Ugric
LM Middle Lozva plur plural
LO UpperLozva pM proto-Mansi
loc locative Po Permian
LU Lower Lozva pO proto-Ostyak
M Moksha PO Yaz'vadialectof
MB Middle Bulgarian Zyrian
Md Mordva poss possessive
ME ErzyaMordva pOU proto-ObUgrian
Mn Mansi pPN proto-Permian
xxiv LIST OF ABBREVIATIONS

PR predicate sx suffix
pred predicative Sy Sygva
pres present SyG syllabic gradation
pret preterite Syn Synja
prf perfective T Tavda
priv privative TC Candyri
pro pronoun temp temporal
prob probabilitative term terminative
prol prolative TG Gorodok
pros prosecutive TJ Janyckova
Ps personalvoice tr transitive
pS proto-Saami Tra Tromagan
pU proto-Uralic trans translative
pV proto-Vogul Trj Tremjugan
pVo proto-Votyak UAlb Ural-Altaische
px personsuffix lahrbiicher
pZ proto-Zyrian UEW Uralisches
refl reflexive Etymologisches
renarr renarrative Worterbuch
RhG rhythmic gradation V Vach
S Saamic V verb
s singular VdN forms verbsfrom
S subject nouns
Sa Samoyedic VH vowel harmony
Sal Salym Vj Vasjugan
sg singular VN North Vagilsk
Sher Sherkal Vo Votyak
Sib Siberian voc vocative
sj subjunctive VR verbalrepresentation
SN North Saami VS SouthVagilsk
So Sosva vsx verb-formingsuffix
SO UpperSysola vx verbal suffix
subj subjective WM WestemMari
subl sublative Z Zyrian
Map i Uralic Idioms of the Volga Bend

Kudymkar

Perm
Cor'kil Joskar-Ola
Izevsk

Kazan

Saransk
Ufa

Mari
Kujbysev
Mordva
Udmurt
500 km
(Komi-) Permiak

Source: Adaptedfrom GeographicalDistribution ofthe Uralic Languages,FU Societyand Helsinki University, 1980.
Mapii Komi

BarentsSea

a
Us
Pecora

Izm
a
Mez
en
e
gd
ce
Vy

Yaz'va Komi
Kama
a
tk
Va

Komi
(Komi-) Penniak

Source: Adaptedfrom Haarmann1974: 171-2.

 Map111 ObUgrian

Taz
Yen
isei
a
cor
Pe

dym
Na

a
Sosv
Tr
om
Ob'

North
an

jog

Kam
Ye

Ag
n

a
Pelym
an
Vakh
ise
i

Ko

lOZ
nd
a

an

'VA
Jng
Mansi

s
an

Ta
ug

v
sj

Irty
Khanty

da
Va Ob'

Source: Adapted from Haarmann 1974:215.

Map iv Samoyedicand ObUgrian (with Komi).

NG

NG
NG Nganasan
N E Enets
N E
N (Tundra) Nenets
FN ForestNenets
S S Selkup
N K N
N S Ye KH Khanty
nis MA Mansi
Fn ei K Komi
KH
K K MA KH
KH
KH KH
MA S
K S
KH Ob'

Irtys

Urals
1 Introduction
Daniel Abondolo

The term 'Uralic' refers to the largestlanguagefamily of northernEurasia.

This family consists of at least thirty languages,spoken in communities
scatteredover a vast areawith westernlimits in Norway and Hungary and
with easternlimits on the Taimyr peninsulaand along the Yenisei and Ob'
rivers of western Siberia. Their large number, and the considerabletypo-
logical diversity of the phonology,morphologyandevensyntaxof the Uralic
languagesmakeit impossiblefor this chapterto provide anything morethan
a brief surveyof someof the more salientsynchronicand historical features.
The newer and older names for the languagesand the names for their
reconstructedancestorsare introducedbelow; the following sectionstreat
phonology and morphology, from both a descriptive and a historical-
comparativeperspective;pp. 30-1 look briefly at someof the more common
kinds of syncretismandsuppletionin Uralic languages;syntaxis broachedon
pp. 31-3; and the final section in this chaptergives a glimpse into Uralic
vocabularyby presentingeight selectedsynonyms.
Good short general introductions to the Uralic languagesare Janhunen
1992, Comrie 1988, and Austerlitz 1968. For a better understandingof the
greaterlinguistic context, Comrie 1981b shouldbe consulted;this book has
good chaptersnot only on the Uralic languagesbut on 'Altaic' (Turkic-
Mongolic-Tungusic) and Paleosiberian(Chukotko-Kamchatkan,Eskimo-
Aleut, Yukagir, Ket, andNivkh) as well.

Internal Subdivisions and Nomenclature; Possible External

Connections
The locus from which theselanguagesemanated,in other words: the proto-
homelandof the speakersof the languagefrom which all Uralic languages
come,is unknown,but a relatively large and sparselypopulatedregion at or
near the southernend of the Ural mountainsis likely. Some of the internal
divisionsof the Uralic languagefamily arenot entirelyclear,but thereis close
to universalagreementwithin the professionthat the primary chronological
breakwas betweenSamoyedicon the onehandandFinno-Ugric on the other.
The Samoyediclanguages,spokentoday chiefly to the eastof the Urals, are
thought to descendfrom a form of the Uralic protolanguagewhich spread

1
2 INTRODUCTION

eastward,partly throughthe migrationof its speakers,partly by purelinguistic

expansion,andin the courseof this separateexistencedevelopedits own traits
and distinctive vocabulary.The populationspeakingproto-Samoyedicmust
havebeenquite small at first, andthereis no reasonto assumethatit underwent
any significantexpansionbeforethedissolutionof its linguistic unity, probably
in the centuriesimmediatelyBCE. The bulk of the vocabularywhich can be
safelycalledcommonSamoyedicis surveyedin Janhunen1977,whichcontains
650 root morphemes,of which only about 150 go back to proto-Uralic. This
relatively small proportion makes it likely that the primary split of Uralic
occurredat leastsix millennia ago, and possibly- given the sociologicaland
demographicfeaturesof the early proto-Samoyedicpopulationand the rate of
changewhich theseimply - muchearlier(Janhunen1992).
Many of the languages which resulted from the breakup of proto-
Samoyedichave doubtlessbeenlost without trace, but reports byEuropean
explorers from as early as the seventeenthcentury, and then by linguists
beginning with Castren1854, have provided us with information about six
distinct Samoyediclanguages.Listed roughly from north to south, theseare
(older designationsgiven in parentheses): Nganasan(Tavgy), Enets(Yenisei-
Samoyed), Nenets (Yurak), Selkup (Ostyak-Samoyed),Kamass(ian),and
Mator (Motor). The southernmostlanguages,Kamassand Mator, are now no
longer spoken:Mator was replacedby Turkic idioms during the first half of
the nineteenthcentury, and the fact that it is known at all today is becauseof
intensivephilological work donewith word lists; the last Kamassspeakerdied
in 1989. Of the more northerly languages,only Nenets is spoken by a
relatively large number of people (some 27,000); Selkup, which has sharp
dialectal divisions, has fewer than 2,000 speakers;Nganasan,some600; and
Enets,perhaps100.
Compared with Samoyedic, the Finno-Ugric branch is and probably
alwayswas the larger,in termsof both absolutespeakernumbersand internal
subdivisions.At leastone of thesesubdivisionsmustbe quite old, dating back
at least as far as the third millennium BCE. There is no universal agreement
aboutthe precisemembershipof the two groupswhich resultedfrom this first
break within Finno-Ugric. As the name Finno-Ugric itself suggests,the
traditional view since Donner 1879 has beenthat the two main subdivisions
of this branchare:
1 a 'Finno-' sub-branch,which consistsof Saarnic(Lapp), Fennic (more
commonly termed Baltic-Finnic), Mordva (Mordvinian), Mari (Cher-
emis), and the PermianlanguagesUdmurt (Votyak), and Komi (Zyrian,
Zyryene);
2 a 'Ugric' sub-branch, consisting of Hungarian and the ObUgrian
(Ob-Ugric) languages,Mansi (Vogul), and Khanty (Ostyak).
Although basicallycorrect,Donner'swork was a false step,takenin haste
by one whose forte was not the methodical side of historical-comparative
 INTRODUCTION 3

linguistics (it was meantas a responseto another,but intriguingly, mistaken

publication of the same year by his rival, Budenz). The problem with
Donner'ssubdivisionlies not so much in its postulationof a primary divide
between'Finno-' and 'Ugric', as in the order of the subsequentsubdivisions
of the non-Ugric branch.WhereasDonner assumedthat first Permian,then
Mari-and-Mordva,and only then Saamicand Fennic had brokenaway, work
since the 1970s comparingthe Samoyedicbranchwith the Uralic family as
a whole, but more particularly with the reconstructionof Fennicand Saamic,
has led to a revision of this view. This revision seesFennic and Saamicas
forming a nodeof westernperipherallanguages,one which broke away from
the rest of Uralic quite early, perhaps not long after the breakaway of
Samoyedic;the Saamic-Fennicprotolanguageitself had already begun to
breakup in the secondhalf of the third millennium BeE (SammallahtiSaamic
chapter2, and 1984). Working with developmentsof the consonantsin the
Finno-Ugric languages,Viitso (1996) has now proposeda refinementto this
revision, accordingto which the breakupof Finno-Ugric beganin the west,
with first Saamic-Fennic,then Mordva, then Mari, and finally Permian
leaving the core. Expressedin terms of the conventionaldownward-growing
ancestraltree,we would thenhave:

URALIC

Finno-Ugric
Saamic-Fennic
Mordva
Mari
Permian Samoyedic
CORE

in which the 'core' is the linguistic cauldron from which 'Ugric' was to
emerge.
As an alternativeto the tree type of diagram,the meiotic amoebamodel is
usefulfor renderingtransparentcertainareal/typologicalzones(Sc =Saamic,
Fe =Fennic,Mr =Mari, Md =Mordva, Pn =Permian,Hu =Hungarian,Kh
=Khanty, Mn =Mansi, Sa=Samoyedic):

Hu Hu Hu Hu
Hu
Hu Mn Hu Mn
4 INTRODUCTION

Whateverthe causeof the split in the Saamic-Fennicnode, there was a

predictably unevendemographicresult, with speakerpopulationsliving in
climatically more clementareasexpandingmorequickly than, andeventually
at the expenseof, their northern linguistic relatives. The number of people
speakinga Saamiclanguagetodayis no more than aboutthirty-five thousand,
while Fennic languagesare spokenby more than six million. Within each
branchthe figures are no more evenly distributed: while NorthernSaamihas
some thirty-thousand speakers,Akkala, Pite, and Ume Saamihavenext to no
speakersleft; while Finnish is spokenby aboutfive million and Estonianby
about one million, Veps, an easternFennic languagewith strong dialectal
subdivisions,is spokenby somesix thousandpeoplein two non-contiguous
regions of Russia, and Livonian, a westernFennic language,is spokenby
fewer than forty. Treatmentsof the subdivisionsand designationsof Saamic
andFennicmay be found in Chapters2 and 3.
The next node to breakfree from the centrewas the ancestorof Mordva,
which survives today as two varieties called 'Erzya' and 'Moksha', with a
combinednumberof speakersin excessof one million. The similarities and
differencesbetweenErzya and Mokshahave frequentlybeenexaggeratedor
minimized for extralinguistic reasons;it is in any casedifficult to quantify
divergence.What deservesto be emphasizedis the fact that there has been
extensiveborrowing and cross-fertilizationbetweenthe two main types of
Mordva, resulting in Moksha dialects with heavy Erzya influence and vice
versa.
Proto-Mordvamay have separatedfrom the core at roughly the sametime
as proto-Mari, but the idea, once widely-held, that there was a common
Mordva-Mari protolanguage(so-called'proto-Volgaic')is now out offavour.
Mari is spokenby somethingbetweensix andsevenhundredthousandpeople.
Accuratefigures for numbersof speakersby dialect are not available,but if
we divide the linguistic terrain into the traditional East-Westsplit of dialect
groupsthe overwhelmingmajority will then be classifiedas speakinga form
of Eastern(or Meadow)Mari, andprobablyfewer than onehundredthousand
speakWestern(or Hill) Mari.
Takentogether,Saamic,Fennic,Mordva and Mari may be seenasforming
the westernwing of the Uralic languagefamily. That this wing must at one
time havehadevengreaterdiversity may be deducedfrom ethnonymsknown
not only from Russianchroniclesbut also from the historian of the Goths,
Iordanes(fourth centuryCE); particularly pertinentare the mysteriousMerja
and Murom(a), peoples mentioned in connection with the ancestorsof
Mordva and Mari, usually as living to the west and north of the present-day
Mordva and Mari Republics, i.e. closer to Moscow. Beside the shared
innovationsof the Saamic-Fennicnodeandretainedinheritancesfrom proto-
Uralic, the westernUralic languagestoday show commontraits which stem
from two additional sources: common borrowings (e.g., from Baltic and
Germanic) and areal (Sprachbund) convergencephenomena,such as the
 INTRODUCTION 5

subtractiveexpressionsfor 'eight' and 'nine', e.g. *kAktA e-k-sa-n(*)TWO

NEG. VERB-PRES-3-DUAL'two are not' = 'eight'. Thesegeographicaland
arealfactorsshouldbe seenagainstthe backgroundof intensiveeconomicand
political expansion, particularly in the ninth and tenth centuries, which
involved the explosion of East Slavonic, the Varangian(Viking) conquests,
and the re-immigration of western,probably mostly Fennic (cf. the people
referredto in Russianchroniclesas 'Chud'),Uralic speakersinto what is now
centralwesternRussia.
The predecessorof the Permian languages,proto-Permian, was pre-
sumably the last to break away from what is here called the core. In its
descendantsone can seeclearly the continuationof at least one drift feature
alreadywell underwayin pre-Mordvaand pre-Mari, namely the weakening
of non-initial obstruents.For example,the intervocalic *-t- in the word for
'water', preservednicely in the Finnish oblique stem vete- but voiced in
Mordva (vedi ) and spirantizedin Mari vanished
(wiio~-), has vanishedentirely in the
Permianlanguages,viz. Komi va, Udmurt vu. Permiansymbiosisprobably
lasted well over two thousandyears, from about the middle of the second
millennium BeE to around the ninth century, when the forerunnerof Komi
beganto take shapewith the Permianexpansioninto the vast territory to the
north of the putativePermianhomelandbetweenthe Vyatka andKamaRivers.
That this expansionwas no simple one-off departure,but rather a cyclical,
back-eddying,centripetalas well as centrifugal,movementof people,goods,
andlanguageis suggestedin Austerlitz 1985.
In Hungarian, but not in the ObUgrian languages,a similar sort of
weakeningtook place;the Hungarianword for 'water' is (obliquestem)vizii-.
The relative chronologiesof obstruentweakening/voicingand of develop-
ments such as *nt > -d- in pre-Hungariandiffer, however, from those of
Permian,and the parallel development exemplified by cognatessuchas Komi
ted- = Hungarian tud- 'knows' is therefore evidence of secondary,areal
contactsat best, and not of a genetic relationshipbetweenHungarian and
Permian closer than that betweenHungarian and ObUgrian (R6dei 1988:
353-7). There must have beensomeearly contactbetweenspeakersof pre-
Hungarianand proto-Permian,however,to judge by the quantity and quality
of correspondences betweenHungarian and the Permian languagesin the
domains of morphological typology, both in root architectonics,predom-
inantly (C)VC(C), andin inflection, seenparticularlyin the hypertrophiccase
systemsdiscussedbelow; there are also lexical borrowings(e.g. the Hungar-
ian words for 'bread', 'silver', and 'threshhold')and a few striking parallels
in derivational morphology, e.g. the infinitive (Permian -ni: Hungarian -ni
from a deverbalnoun *=nA plus a lative suffix) and the suffix *=mVn, used
to form decades,e.g. Hungarian=viin, Komi =iinin in otviin = vetiinin 'fifty'.
Similarities betweenHungarian and Permian pale, however, when one
comparesHungarianwith the ObUgrianlanguages,Mansi and Khanty. The
geneticnode of thesethree languages,termed 'Ugric', has proven extremely
6 INTRODUCTION

difficult to reconstructin convincing detail. The reasonsfor this difficulty

flow from what has already been said above; we may summarizethem as
follows:
1 Hungarian,Mansi, andKhanty arethe sole survivorsof what is hereseen
as the core, i.e. most central and innovating region, of Uralic linguistic
and cultural space; we would expect languagesin such a central,
innovatingposition to undergorelatively rapid changes,evenupheavals,
in their phonologicaland lexical makeup.Put anotherway: had we no
recordsof Latin, Romancewould be more difficult to reconstructon the
basisof Frenchthan of Romanianor Portuguese.
2 The period of Ugric symbiosismust have beenfairly short. There is no
overall agreementamong scholarsconcerningthe dating of the break-
away of proto-Permianon the one hand and the separationof Hungarian
out of Ugric on the other, but a safeestimatewould placethe formerno
earlier than the middle of the secondmillennium BeE, and the latter no
later than its end, i.e. a period of five hundredyearsat the most. It is not
uninterestingto note that some scholarshave posited similar or even
identical datingsfor the beginningof the independentexistenceof both
Permianand Hungarian(e.g. Decsy1965: 154, 169,172, 183).
3 As vehicles of culture, both Hungarian and proto-ObUgrian suffered
major blows in the form of radical restructuringof genre de vie: while
speakersof proto-ObUgrian,in consequence of their migration eastand
north, were thrust back into a neolithic cultural frame, speakersof
Hungarian underwent the reverse scenario, namely the accelerated
modernizationwhich attendedtheir settling in centralEurope.The effect
on the sharedlexicon hasbeencatastrophic:in both cases,old discourse
was replacedor transmuted,usually beyondrecognition (but cf. Honti
1990).
Difficulties with the reconstructionof Ugric havebeena stumblingblock
which hashinderedprogressthroughoutUralic comparativelinguistics, from
the historical phonology to speculations concerning active/passiveand
transitive/intransitivein the verb, and such fundamental questionsas the
natureof the proto-Uralic noun phrase.Much further work within Ugric and
Permianwill be necessarybefore a useful assessment of competinghypo-
theses,including that offered here, can be made.For a rangeof other views
on the internal organizationof Uralic seeGulya (1975), Suihkonen(1987),
Hajdu-Domokos(1987), andSalminen(1993).
The precisenumber of Uralic languagescannotbe given, as this would
require a good answerto the question, 'What is the difference betweena
languageand a dialect?'(A facetiousanswer,but oneworth pondering,is: 'A
languageis a dialect with an army and a navy'.) This questionis one of the
mostimportantposedby linguistics, but like most suchquestions,linguistics
alone cannotanswerit: the questionis entangledwith extralinguisticfactors
 INTRODUCTION 7

such as nationality, identity, social strata, and other complex variables

connectedwith the hierarchizationand hieraticizationof behaviouraldiver-
sity. In any caseansweringit would lead into topics and further questionsfar
afield from the subjectof this book.
Nevertheless,it can be safely stated that some Uralic languagesare
dialectally much more fragmentedthan others.The richestin this regardare
Selkup,Khanty, andEstonian,andin all of thesecasesthe authorsresponsible
for eachof theselanguageshavechosento focus on onevariety, with reference,
whereuseful,to others.(Thesituationin Saamicis extremeandsui generis.)
Similarly, certain branchesare richer in subdivisions than others, and
limitations of space have meant that certain languages/dialectsin such
branchesdo not receiveseparatetreatment.This is especiallynoticeablein the
caseof the peripheralbranches,Fennicand Samoyedic;to compensatefor the
omission of separatechapterson e.g. Veps and Enets, nodal chaptersare
offered. There are nodal chapterson ObUgrian and Permianfor somewhat
different reasons: these chapters not only highlight the similarities and
differences betweenthe languagesin these central sub-branches,but also
introducethe readerto someof the difficulties involved in the reconstruction
of their ancestorlanguages.The chapterson Hungarianand Finnish are also
somewhatdeviant: sincenumerousdescriptionsof theselanguagesarereadily
available in English, thesechaptersaim to supplementsuch descriptionsby
focusing on aspects of these languageswhich are not so well known,
particularly as they arerelevantto Uralic typology andprehistory.
Attempts at proving externalgeneticconnectionsof the Uralic languages
have been numerous.It is of coursenot only possiblebut likely that some
protoform of pre-Uralic broke away from some older geneticunit, but such
a breakwould havehadto takeplacelong ago,andbecauseof this, greattime-
depthenterpriseswhich seekto demonstratesucha breakare not likely to be
convincing.In the caseof comparingUralic with Indo-Europeantwo kinds of
difficulty arise. On the one hand,the grossdivergencesbetweenproto-Uralic
and proto-Indo-Europeanmorphological typology mean that there is an
enormousgap to be bridged,in both form andfunction; on the otherhand,the
grossdivergencesbetweenproto-Uralic and proto-Indo-European phonology
make comparisontoo easy: becauseof the rich inventory of stops in proto-
Indo-European,for any proto-Uralic root of the shape*CVC(V) there are a
great many possible proto-Indo-Europeanconsonantalcorrespondents.For
example,to a proto-Uralic root of the shape*kVt(V)- one could plausibly
cite, as formally congruent,proto-Indo-Europeanroots of the shapes*kVt-,
*kVd-, *gVd h-, *ghVd-, and *ghVdh_, not to mention roots with initial
labiovelars(e.g. *gWVt-), or laryngeals(e.g. *h1Vd-), or with clusterswith s
(e.g. *skVt-), or with morecomplexphonotactics(e.g. *skVid-). The chances
of finding a formal match betweenproto-Uralic and proto-Indo-European
reconstructedroots are thus perilously good (for a cautiousrecentconspectus
ofIndo-Europeanphonologyand morphologyseeBeekes1995).
8 INTRODUCTION

At flrst glance,comparingUralic with' Altaic' would seemmorepromising

on both morphologicaland phonologicalgrounds,but here the comparative
methodrunsinto anotherkind of difflculty, oneof its own making.(' Altaic' is a
phylum, or stock, containing the languagefamilies Turkic, Mongolic, and
Tungusic;many scholarsalsoincludeKoreanandJapanese.) The difflculty of
comparingUralic with 'Altaic' residesin the fact that what is called 'Altaic' is
not a family in the strict sensebeing usedhere, i.e. in the sensethat Uralic,
Semitic,Tibeto-Burman,Japanese-Ryuku, or Eskimo-Aleutare families. The
historicaldevelopmentof all of thesefamilies hasbeenestablished,by meansof
the comparativemethod,to a degreeof precisionwhich is both predictiveand
productive.'Predictive'meansthatgivenform X in languageY we canpredict,
on the basisof regularcorrespondences andcrediblecoursesof development,
whattheform of its cognate,form Z in relatedlanguageW, will be. 'Productive'
meansthat such predictionswill either prove to be correct, buttressingand
fleshing out the detail of the family's geneticintegrity, or they will fail, forcing
the investigatorto rehonehis or her tools and to ask different, perhapsmore
penetrating,questionsabout the family's internal relations. On this view,
comparing Uralic genetically with 'Altaic' is a category mistake, akin to
comparingthe Florentinesonnet,say, with all of Proven~alProven~al versiflcation.For
somewhatsimilar views, presentedin greaterdetail, see Janhunen(1984),
Doerfer (1984), and Austerlitz (1978); for an opposingview see e.g.Miller
(1990). On the other hand, broadly conceivedtypological comparisonof the
belts and pockets of languagesand languagefamilies which stretch across
central and northernEurasiahas proven to be a fruitful exercise(Austerlitz
1980,Nichols 1992).
The only truly credible candidatefor productive, predictive comparison
with Uralic is Yukagir, a languagemost safely classifled, at present,as an
isolate, i.e. as having no proven linguistic relatives. Yukagir is spokenby
about 300 people scatteredin small groups in northeastSiberia, and is
traditionally bundled, with other small families and linguistic isolates of
central and northeastSiberia, into a grouping called 'Paleo-Siberian'.If
Uralic is relatedto Yukagir, any superficial similarities betweenthe two are
unlikely to reflect an old, i.e. genetic,connection.The proof will haveto come
from analysiswhich goesmuch deeper,makesmore daring hypotheses,and
therebyincurs greaterrisks. Two such analysesstandout: Harms (1977), on
morphosyntax, and Nikolaeva (1988), on historical phonologyand morpho-
logy. Cogentas eachof thesetwo attemptsis in someof its details,there are
also importantpoints at which they clash, such as the role and development
of the genitive and its surrogates;perhapsfurther work will discover an
answerwhich resolvesthe dissonance.
If Yukagir shouldprove to be relatedto Uralic, the protolanguage,'proto-
Uralo-Yukagir', will be at a temporalremove at the very edgesof recover-
ability. With our present primitive understandingof the mechanismsof
languagechange, such time-depth precludes,by implication, the detailed
 INTRODUCTION 9

reconstructionof the type that has been possiblefor Uralic, and thus also
precludes,afortiori, any predictive-productivecomparisonfurther afield. For
a parallel situationin far easternSiberia, seeComrie 1980a.

Sound Systems
The default airflow initiation for all Uralic languagesis pulmonic egressive.
Long stretchesof Finnish, however, are often spoken on an ingressive
pulmonic airstream,with whisperreplacingregularvoice (Laver 1994: 170).
Word-stressfalls on the first syllable in most languages,althoughcertain
languages(suchasEstonian)aretolerantof stresson othersyllablesin foreign
vocabulary.The most notableexceptionsto this patternare found in a belt of
west--centrallanguages(Moksha Mordva, Mari, all of PermiansaveKomi)
and in Samoyedic.The precise nature of each exception is different: for
example,stressmay be fixed, asin WestMari or Nganasan,whereit normally
falls on the penult, or in Udmurt, whereit falls on the last syllable; or it may
be mobile, as in Permiak, where its position is governedby morphemic
valence, or as in Moksha Mordva or Tundra Nenets,where its position is
governedby vowel sonority or fullness. Stresspositionin Selkupis governed
by both phonologicaland morphologicalfactors.
Consonants
Three kinds of phonological opposition are widely exploited in the con-
sonantalinventoriesof Uralic languages:quantity, voice, and palatalization.
The distribution of distinctive quantity and voice is quite skewed,and to a
large extentcoincideswith the core/peripherysplit: most languagesare rich
either in quantitative consonantaldistinctions (e.g. Finnish, Estonian, or
Selkup) or in voice distinctions(Hungarian,Permian).Voice is distinctive at
the phonological,but not at the deepestmorphonologicallevels of Tundra
Nenets; the same is true of some Saamic languages,e.g. Northern Saami
nominativesingularoabbd 'sister',genitive singularoappd; note also that the
differencein the intervocalic consonantsis not simply one of voice but also
of duration (see Saamic, Chapter 2). A parallel pattern, namely of short
intervocalic voiceless,and long intervocalic voiced, stops, occurs in Enets
(Terescenko1966: 440, Ristinen 1960: 42-3). In the core, i.e. non-borrowed
and non-affective,vocabularyof Mordva andMari, voicelessobstruentstend
to occur in word-initial position or togetherin clusters,while their voiced
counterpartstendto occurin non-initial positions,particularly after sonorants
and beforevowels, as in Erzya Mordva'month',
kudo 'house',kurgo 'month', makso
cognates
'liver', and their West Mari cognateskuo~ (with [0] - [d]), 'month',
korg~ (with [ry]
- [rg]), and moks.In Mordva and TundraNenets,suchvoicelessconsonants
as do occur intervocalically tend to be pronouncedlonger than their voiced
counterparts,e.g. Erzya Mordva koto 'six' with short geminate [ttl. The
opposition of voice is well establishedin Hungarian,partly becauseof the
greatnumberof loanwordsfrom Turkic and Slavonic,but it is perhapseven
10 INTRODUCTION

moredeeplyrootedin Permian.Examplesof miminal pairs areHungarianpap

'clergyman':bab 'bean' (both Slavonicin origin), and Komi ker 'taste':ger
'plough' (both of which go back at leastto proto-Permian,cf. Udmurt karel,
geri).
Palatalization,along with its opposites,either lack of palatalization(as in
Mordva) or velarization(asin TundraNenets),occursin oneform or anotherin
all Uralic languagesexceptFinnish.In Hungarian,palatalizationsurvivesas a
morphonologicalprocess,butpalatalizedconsonantshavegivenway to a series
of palatalconsonants,probablyas a resultof areal(Danubian)convergence.In
Tundra Nenets and Erzya Mordva, the palatalizedconsonantsinteract with
vowelsto producelimited kinds of vowel harmony;seepp.17-18.
As far as place of articulation is concerned,most Uralic languageshave
their consonantsdistributed fairly uneconomically over five oral zones,
namely (1) labi(odent)al, (2) dental, (3) post-alveolar/retroflex,(4) palatal,
and(5) velar, as in ErzyaMordva, wherethe voicelessseriesare:

1 2 3 5 5
Stops/affricates P 5 c 5 k
Fricatives (f) s s 5 (x)

with (f) and (x) restrictedto interjections and recent loans. The opposition
implied by the place-basedclassificationof columns 2-3-4 is misleading,
however, since the hissinglhushingdistinction for central fricatives and
affricates is widespreadamong the Uralic languages,and palatalizationis
commonly grafted onto either or both, e.g. Komi plain s v. hushingretroflex
s v. hushing palatalizedSi, as in sila 'fatty', sila 'awl', Sirla 'I sing'. In
languagesin which voice is distinctive, oral zones2-3-4 can becomerather
crowded; for example,Udmurt distinguishesfifteen oral consonantsin this
region. On the other hand,the distinction betweenvelar and postvelar(often
calleduvular) placesof articulationappearsto be limited to Selkup,although
it may be nascent,and due to Selkupinfluence,in Surgutdialectsof Khanty.
This distinction is doubtlesslyan import; it is typical of languagesto the east
ofUralic, including Ket, Yukagir, andthe Chukotko-Kamchatkan group.
Other than this velar/postvelar(klq) distinction, the chief deviationsfrom
the five-zone patternare found in Finnish, which lacks columns3 and 4, and
in languageswhich have in addition to columns 1-5 either interdental
fricatives (asin Saarnic,andsomeFinnishdialects),or labiovelars(asin some
Mansi, Khanty, and Selkupdialects,for exampleTremjuganKhanty, with six
nasals:columns1-5 plus labiovelarIf).
Voicelessnessin consonantswhich are normally voiced, i.e. nasals,
laterals, and trills, is found in some varieties of Saarnicand Khanty, and in
Moksha Mordva. Bilabial w has been replacedwith labiodentalv in most
languages,in keeping with the general westernEurasiantrend (Austerlitz
1976), but w is still found sporadically,e.g. in Udmurt, Mansi, and Selkup;
 INTRODUCTION 11

not coincidentally,labializedobstruentsare found in the latter two languages.

As far as non-oral, non-nasalconsonantsare concerned,the type usually
reportedfor Uralic languagesis glottal stop and relatedphenomenasuch as
st¢d (in Livonian) and 'doubling' (in Finnish). Glottal stop is particularly
important in the synchronic and diachronic morphophonologiesof most
Samoyediclanguages;it is in statu nascendiin somedialectsof Udmurt and
Mari (for the latter, seeVasikova1992).
Quantitativedistinctions among the consonantsare most readily under-
stood in terms of history. As mentioned above,in Finno-Ugric languages
nearerto the core,stopsin weakpositionwereweakened,i.e. they wereeither
(1) voiced, or changedinto (2) fricatives or glides, or (3) melted into the
surroundingvocalism.In mostof the peripheralUralic languages,on the other
hand,suchstopseither(4) persistedor, (5) were strengthened:

1 pU *witi 'water' > ErzyaMordva vedj , *pilmiHi 'dark' > Komi pemi"d
2 pU *witi > Hungarianviiz, pFU *ikii 'year,age'> ErzyaMordva ije
3 pU *witi > Komi va, pU *pilmitii > Finnishpimeii
4 pU *witi > Selkup iit, Finnish vete-en'water sIll', pFU *ikii > Finnish
ikii
5 pFU *kaata 'tent, dwelling' > North Saamigoahti, pFU *ikii > N Saami
ahki

The fortition processseen in Saamic has parallels in Samoyedic,cf. the

geminatepronunciationof non-weakobstruentsin TundraNenets.
In Saarnic,strengthenedconsonantismslike that of (5) were maintained
when at the onsetof opensyllables,but werepronouncedshorterwhen at the
onset of closed syllables, yielding one kind of consonantgradation, e.g.
Northern Saami sNgoahti 'tent' « *kaa1ii, sG goaoi « *kaata-n. Parallel
strengtheningof geminatesin weak position also occurred, with parallel
shorterpronunciationbefore closedsyllables(*-tt- > *-tt- - *-1-). In Fennic,
consonant gradation arose in reverse fashion, that is it arose through
weakening(seeChapters2 and 3), but the resultswereroughly the same.We
may schematizethe developmentsof single consonants(x), geminates(xx),
andclusters(xy) in SaamicandFennicasfollows:

Saamic Fennic
Stronggrade Weakgrade Stronggrade Weakgrade
*x> xx x x xJ2
*xx> XiX xx xx x
*xy> xly xy xy xy/2

where XiX and xly standfor a strong geminateor cluster, and x/2 and xy/2
standfor weak consonantsand clusters,the phoneticrealization of 'strong'
and 'weak' varying from languageto language, andwithin eachlanguage,
12 INTRODUCTION

dependingon (1) the segmentsinvolved and(2) their positionin the word.

In many dialects of Fennic (Votic, Finnish, Ingrian and North Karelian),
andin mostdialectsof Saarnic,the two typesof xx fell together,i.e. the strong
grade of single consonantscame to be identical with the weak grade of
geminates.For a recentsurveyof FennicgradationseeBarbera1993.
Anotherkind of consonantgradationarosein Nganasan,the northeastern-
mostUralic language.SinceNganasanmay constitutea sub-branchof its own
within Samoyedic(Chapter15), it is plausible that phonetic featuresof the
protolanguagegave riseto both the far western(Saarnicand Fennic) and the
far easternversions(Helimski 1996).
The historical backgroundof the consonantsystemsfound throughout
Uralic is clear, in the main. We may reckonwith a consonantsystemwith the
samebroadlyfive-zonesystemas that of mostpresent-day languages.
In addition, there seemto havebeentwo further segments,usually written
*b, *b j . Disagreementand uncertaintyaboutproto-Uralic consonantsusually
centreon the phonologicalstatusandphoneticnatureof these,and *c j and *x.
Since Steinitz 1952: 37, the phonemic status of *b and *b j has been
questioned, and numerous suggestionshave been made with an eye to
dispensingwith them; so far none has been deemedfully successful(e.g.
Decsy1969,TaIos 1983,Kazancev1990,Abondolo 1990,Honti 1992).
Both cacuminal *c and palatal(ized)*c j are commonly reconstructedas
affricates,i.e. as having had delayedrelease,but it is equally possible that
they were both normally realized as stops. While it is possible to reckon
without *c j , explaining correspondences which would derive from such a
segmentas affective variants of *sj, it is more difficult to managewithout
setting up *x, which is little more than a cover symbol for various phonetic
phenomenaoccurring in the syllable coda, most notably lengtheningof the
precedingvowel (or of the following consonant)in Finno-Ugric and vowel
sequences in proto-Samoyedic(Sammallahti1978).
By the time ofproto-Finno-Ugric,the consonantsystemappearsto fill out,
in part, no doubt, becausewe have a greaternumberof reliable etymologies
with which to work. Whatevertheir preciseage,the new phonemes*s and *lj
may be reconstructedwith some reliability by this stage. In general

Table 1.1 Proto-Uralic consonantsystem

1 2 3 4 5

Glides *w *j *x
Nasals *m *n *ni *1)
Stops *p *t *k
Affricates *6 *c i
Fricatives *s *si
Lateral *1
Trill *r
 INTRODUCTION 13

throughout the pU and pFU lexica as reconstructedthe stops occur more

frequently in strong (= word-initial) position, while the liquids and glides
occurmorefrequentlyin weak(= word-final andsyllable-fmal)position; thus
pU *sala- 'steals,hides' has a canonic shape,whereasa stem with a shape
suchas *IVsV- would be doubly non-canonic(thereare, in fact, no good pU
or pFU roots reconstructedwith such a shape).This distribution persists,
grossomodo, in Mordva and Mari, as mentionedabove.The relative age of
medial clusters, geminates,and the distribution of (geminate) affricates
complicatethe picture,as doesthe role of affectivevocabularyandloans.

Vowels
At least in the first syllable, vowel quantity, i.e. phonologically relevant
distinctions in the durations of vowels, is characteristicof most Uralie
languages.The chief exceptions are Erzya Mordva, with a five-vowel
inventory i e a 0 u, and Permian,where quantity is either vestigial, as in the
stress-assignment of yaz'va Komi, or nascent,as in lima and Vym Komi,
where *VI has gone to VV before pause or a consonant,e.g. soo 'salt',
soi=an+teg 'salt-box'. Note that in this book length is indicated by letter
doubling, unless otherwise indicated. In Enets and Kamass,the status of
vowel quantityis unclear.
Quantitativeoppositionsof vowels take two major typological forms in
Uralic, andin mostcasesinvestigatorsare in accordconcerningthe natureof
each system.The more widespreadtype is basedon the oppositionshort v.
long. In this type the vowel systemis madeup of two roughly equal setsof
vowels whoseprimary differenceis one of duration. ExamplesincludeForest
Nenets(Sammallahti1974: 13), with five long and five short vowels (i e U 0
a ii ee uu 00 aa), Finnish, with eight short (i e ii U 0 a U 0) and eight long
(ii ee iiii uu 00 aa UU (0) vowels, and SosvaMansi with four short (i a U 0)
and four long (ee aa uu 00) vowels. The other type is basednot only on
quantity but on fullness of articulatorydetail: systemsof this type are said to
contain full versus reduced vowels. What this means phonetically varies
greatly from languageto language,but in every casethe short duration of a
reducedvowel is intrinsic, whereasthe duration of the full vowels is much
more susceptibleto conditioned variability. Full vowels usually outweigh
their reducedcounterparts,if not in absolutenumbers,then at leastin terms
of the features neededto define their qualities. Examplesof this type of
systemare SalymKhanty, with full ii eeiiii uu 00 aa, i.e. six vowels andthree
tongue heights, and reducede ii 0 a 0, i.e. five vowels with two tongue
heights; Hill Mari, with full ii ee iiii uu 00 aa UU 00 and reduced;, ~, and
TundraNenets,with one, or, in anotherinterpretation(Janhunen1993),three,
reducedvowels opposedto eight (or six) non-reducedvowels.
Long vowels alternateparadigmatically, or arein skeweddistribution,with
diphthongsin many languages.An exampleof the former is ie - ii, eii - ee,
uo - UU, oa - 00 in North Saami,e.g. (orthographically)nominativesingular
14 INTRODUCTION

guolli 'fish', genitive/accusativeplural guliid. An example of the latter is

Finnish,wherethe long mid vowels ee 0000occurin the first syllablechiefly
in foreign or affective vocabulary,pre-Finnish*ee *66 *00 having goneto ie
UO uo in that position, e.g. tie 'road', cf. Estonian tee. Diphthongs are
prominentin the phonologiesof manyFennicdialects,in EasternandWestern
Mansi (Honti 1988: 150, Sammallahti1988: 506), in Nganasan,in southwest
Selkup dialects (Katz 1984: 41), and many dialects of Hungarian(Kalman
1966: 40-1). The alternationof long vowels with their short counterpartsis
particularly characteristicof Hungarianand SouthernMansi; in Khanty, full
and reducedvowels in the first syllable of certain stemsalternateparadig-
matically, e.g. e - 00 in Vakh KhantyjOOl);Jtw;J[ 'spinss3':jel)tii 'spin!'
Turning to the qualitative distinctions amongthe vowels, we may safely
say that the factors of tongueheight, frontnessv. backness,and lip rounding
account formost systems.The smallestinventory, that of ErzyaMordva, was
given above.MokshaMordva elaborateson this with the addition of an ii and
two schwas,one front (;J) and one back (~);e Hill (West) Mari has all these
eight vowels plus front roundeduand0:

MokshaMordva Hill (West) Mari

e u e ii u
e o e e e 6 o e e
a a a a

The vowel inventoryof Meadow(East)Mari differs from that of WestMari

in that it lacks ii and has only one reducedvowel (~).e The three mid vowels
alternatewith e~, the reducedvowel occurring chiefly in non-final position;
this is analogousto the situationin Erzya Mordva, wherenon-first syllable e
and0 are largely recoverablefrom an abstract neutral vowel; seeChapter6.
In first syllables, the vowel inventory of proto-Saamicwas probably as
follows:

*i *u
*e *e *0 *ee *00
*aa *M
*aa

where *e representsa non-front unroundedmid vowel; seeLehtiranta1989.

Severalvarietiesof Saarnicdevelopedfront roundedvowels through anticip-
atory metaphony,e.g. proto-Saarnic*koolee 'fish' > South Saarni guolie,
Inari Saarnikueli; in someof the easternSaamiclanguages(Ter, Skolt, and
Akkala) back unroundedvowels have developed secondarily, e.g. proto-
Saarnic*moore 'tree' > Skolt muer, Ter mil-r (where frepresentsa non-front
unroundedhigh vowel).
The classification of vowel types in the Saarnic languagesis further
 INTRODUCTION 15

complicatedby the feature tense/lax,which has beenused, for example,to

classify the vowels of Ter Saamiin a mannerreminiscentof the full/reduced
dichotomy mentioned above. Compare the maximal vowel systems of
Arjeplog Saami(adaptedfrom Lehtiranta1992: 73-6) andTer Saami(adapted
from Korhonen1984: 316-20):

Arjeplog Saami TerSaami

Short Long Tense Lax
ii u ie uo ii iT uu ie Ie I
e 0 ee 00
a aa aa aa ea oa a

The developmentof bichromatic(= both front androunded)vowels and of

non-low achromatic(= neitherfront nor rounded)re in Saamicis probablythe
result,in part,of relativelyrecentarealconvergence, with Norwegian,Swedish,
and Finnish providing the impetusfor bichromaticii and 0, and Russianand
Komi furnishing the adstratumfor rand e. The non-front non-low unrounded
vowelsof Permian,on theotherhand,appearto be older,andoutsideinfluences
are not obvious. Both the Udmurt and the Komi literary languageshave the
seven-vowelsystemi e u0 rea, but the prehistoryof thesevowelsis complex,
andtheir lexical distributionin thepresent-daylanguagesseverelyimpedes,or
evenprevents,mutualcomprehension. For example,Komi ecancorrespondto
any Udmurt vowel other than i, as in sep/sep'gall', iez!liz 'blue', sen/sen
'tendon',iedf/llui 'gadfly', ted-/tod-'knows',kev/kai'rope'.
Thus far we have beenlooking at systemswith three tongueheights,and
this is the most commontype in Uralic. In such systems,the presenceof 0
presupposes e: note the maximal vowel inventory
ii, and that of rpresupposes
of Literary Estonian, with e but no r, and of Nganasan,with ii but no o.
However, some of the ObUgrian languageshave vowel systemswhich lend
themselveseasily to a two-tongueheight interpretation,suchas SosvaMansi
andNizyam Khanty, both with

ii u uu ii u ee uu
o ee i.e.
a aa aa a 0 aa aa
Notice that such an interpretationdispenseswith the feature [+/-back], since
the specificationfor [+/- low] and [+/- round] is sufficient to identify all of
the vowels (cf. Katz 1975: 55, wherethe Nizyam Khanty vowels ee aa aa uu
are classifiedas [+ tense]).Very few Uralic languagesdistinguishmore than
three tongue heights,but a vowel systemwith four tongue heightshas been
proposedfor proto-Permian,and the distinction betweenopen mid 0 (as in
pon 'end') and closed mid a (as in pan 'dog') in Upper Sysola Komi is
thought to be a preservation.A synchronically similar and unexpected a
16 INTRODUCTION

seemsto exist in Enets (Mikola 1984). Other subsystemswith four tongue

heights have been reported for various dialects of Selkup and of Voru
Estonian.
Probablyno academicfield is more contentiousthan the prehistoryof the
Uralic vowels, and this is not the placefor a treatmentin depth.What follows
is merely a presentationof some of the high spots; for a brief but clear
summary see Hajdu 1992: 175-88. It is important to bear in mind that
throughoutthe history of researchinto this subject, the developmentof the
vocalismof the first syllable has occupiedcentrestage,to the virtual neglect
of syllablesfurther into the word. But the secondandthird syllablesmusthave
beenimportant,andin fact the vowel of the third syllableprobablyinfluenced
that of the second,just as much as that of the secondinfluenced the first.
There are traces of vertical vowel alternations,i.e. alternationsinvolving
tongueheight, in both the secondand the third syllable of Uralic reconstruc-
tions; while the exact form and function of these alternations remains
unknown, they appearto have been independentlymotivated. Apart from
thesealternations,the secondand third syllablesare generallyseenas rather
uninteresting, with a restricted vocalic repertoire: supposedly only two
phonologicallydistinct vocalismswere possible,and proto-Uralic bisyllables
will be referredto here as A-stems(with *a - *a) and I-stems(with *i - *i"),
accordingto the tongue-heightof the second-syllablevowel.
The primary factor dividing opinion aboutproto-Uralic vocalismconcerns
quantity. According to the view that has been the most rigorously demon-
strated,if not the most widely accepted,distinctive vowel length throughout
Uralic is a secondary development. Scholars who work within such a
framework set up a proto-Uralic vowel inventory with four high vowels (*i
*ii *1 *u), two mid vowels (*e *0) and two low vowels (*a *a); this view is
mostclearly presentedin Sarnmallahti1978 and Janhunen1981b.Thosewho
think phonemic quantity was presentin the Uralic protolanguageform a
smallerand more heterogeneous group: some,following Steinitz 1944, work
with full v. reducedvowels in the protolanguage;others,following Itkonen
1939 and its sequelae,set up a quantitative distinction only for the mid
vowels, and substitutelong *00 for *1; a third idea, first proposedby Talos
(1983) and then pursuedby this writer (1996), posits a proto-Uralic vowel
systemwith only two tongueheights,but with quantity as a featureapplying
to all vowels,at leastin the first syllableof wordswith second-syllablevowels
of equivalenttongueheight. The protovocalismassumedby the most up-to-
date compendiumof Uralic etymology, the UEW (= Uralisches Etymo-
logischesWorterbuch),is a committee-likecompromisewhich incorporates
featuresof the Itkonen andSarnmallahtiapproaches.
Uralic and Finno-Ugric protoforms set up or cited in this book are given
in eitherthe UEWor the TaIos (hereafter:'two-tiered')system.The one is in
most casesreadily convertedto the other: one must simply rememberthat
UEW *e is seen,in the two-tiered system,as the equivalentof long *aa in
 INTRODUCTION 17

A-stems and of short *i in I-stems, and, in parallel fashion, UEW *0 is the

equivalentof long *aa in A-stemsandof short*u in I-stems.Thus UEW *pesa
'nest' is, in two-tiered terms, *paasa,while UEW *wete- 'water' is *witi;
UEW *kota 'dwelling, tent' is two-tiered*kaata, and UEW *ko(nj)cjV 'long'
is *kunjtjL The vowels reconstructedby the UEW as high and short are, in
two-tiered terms, high and long, e.g. UEW *ku(nj)cje(-) 'urine, urinates' is
two-tiered*kuunhjL
The two-tieredsystemseemsto havetwo advantagesover that usedby the
UEW First, it assumesa vocalic inventory and chainsof vowel development
which better fit much of the current understandingof universals (cf., for
example,Maddieson1984: 128-30on vowel length, vowel inventories,and
vowel change).Second,the trajectoriesnecessaryto derivethe vowel systems
of all of the present-dayUralic languages,core as well as peripheral,aremore
believable,againin termsof what we now understandaboutuniversals,when
the starting-pointsinvolve both tongueheightandquantity. Oneexamplewill
have to suffice here: it is the fact that whereasUEW *e in A-stems (= two-
tiered *aa) regularly gives proto-Mansi long vowels (*aa, *ii), UEW *e in
I-stems (= two-tiered *i) regularly gives proto-Mansi short vowels (*i, *a).
For further argumentationand supportingevidence,the readeris referredto
Tcilos 1987 andAbondolo 1996; for a naturalphonology background to many
of the assumptionsupon which the two-tiered systemrests, see Donegan
1985. Approachescloser to the Itkonen-UEWtype may be found in this
volume,in the nodalchapterson ObUgrianandPermian.
High/low vowel alternations in non-initial syllables were mentioned
above; the more well-known vowel alternationseenin non-initial syllables
is of the horizontal,i.e. front v. back, variety; this is the Uralic type of vowel
harmony, a widespreadphenomenonin the world's languages.In Uralic,
vowel harmony is reconstructedas occurring in roots, derived stems, and
inflected forms; the vocalismof the root is always dominant,i.e. determines
the frontnesslbackness of the vowels of attachedsuffixes. For example,the
proto-Uralic locative suffix was presumably*-na when attachedto a back-
vocalic root such as *kala 'fish' (*kala-na), but *-na when attachedto a
front-vocalic root such as *witi 'water' (*witi-na): such twofold representa-
tions may be convenientlycapturedby the useof majusculeletters,e.g. *-nA
for the locative suffix. The Hungarianforms hIllon 'on fish' «< *kala-na)
and viizen 'on water' «< *witi-na) are thought to preservethis twofold
representation,as are the Finnish analogueskalassaand vedessii(via *kala-
s-na, *witi-s-na). In somelanguages,vowel harmony has beenextendedto
embracethe opposition rounded/unrounded, e.g. East Mari, where reduced
~ is 0 after front rounded vowels, 0 after back rounded vowels, and e
elsewhere;or Hungarian,wherethe samethreemid vowels may be recovered
from an abstractvowel (or zero): 0 after front roundedvowels, e after front
unroundedvowels, and0 elsewhere.In Mordva (andto a lesser,andlexically
marked,degree,in Nenets)the consonantsare involved in vowel harmony
18 INTRODUCTION

alternations, palatalized consonantsoccuring in front-prosodic, and non-

palatalizedin back-prosodicenvironments.Consonantalternationsare also
a side-effectof vowel harmonyin Nganasan,wherefrontlback and rounded/
unroundedharmony combine to fonn four-way (i -a - r - u) alternations.
In Hungarian and Nganasan,vowel mergers in the fIrst syllable have
renderedthe prosodicmembershipof many roots covert. On the other hand
many Uralic languages,suchas Saarnic,Estonian,Permian,Selkup,andmost
ObUgriandialects,have 'lost' vowel harmonythroughvariouscombinations
of (1) loss of the non-fIrst-syllable vowel which would have shown the
alternation, (2) neutralizations, through mergers, of frontlback vowel
distinctions,and(3) radicalrotationof the first syllablevocalism.

Morphology

Nouns and Adjectives

Nouns were probably not morphologicallydistinct from adjectivesin proto-
Uralic, althoughthe distribution of the comparativesuffIx *=mpV suggests
that an adjectivecategorymay have beendevelopingbefore the breakupof
Finno-Ugric, cf. the =mpa - =mpi of Finnish korkea=mpi and the =bb of
Hungarianmagasa=bb,both 'higher'. The original function of *=mpV was
almost certainly one of opposition, used primarily with deictics, cf. Erzya
Mordva ombo and North Saarni nubbi « *muu=mpa) 'other'. In Saamic,
morphologyfor distinguishingattributive from non-attributiveadjectiveshas
evolved,e.g. North Saarnidimis 'soft (on the inside)" attributivefonn dipma.
Otherwise,nounsand adjectivesneednot be consideredseparatelyon either
morphological or syntactic planes; in what follows, therefore, the tenn
'nouns'will refer to both.
The three nominal categoriescommon to all Uralic languagesare case,
number,and person.In the protolanguage,all threeof thesecategoriescould
be represented by a zerosuffix, with the grammaticalmeanings'unmarkedfor
case' 'unmarkedfor number' and 'unmarkedfor person';the last instanceis
traditionally called the nominative, but absolutivewould be a better tenn.
Proto-Uralic nouns thus differed sharply from verbs, which rarely, if ever,
occurredwith zero suffix.
Case
In proto-Uralic, the noun had at least two grammaticalcases:an accusative
*-m, which probably was usedchiefly to mark the defInite direct objectsof
finite verbs (i.e. verbs inflected for person),and a subordinative suffix*-n
which functioned as a genitive/prenominalizerwith nouns and as an adverb-
fonnant with verbs. There were also at least three local cases,including a
locative *-nA, a separative*tA - *-tI, andperhapsthe latives *-k (and/or*-IJ)
and *-c j (and/or*-n j ).
The local subsystemsof the caseparadigmsof most of the present-day
 INTRODUCTION 19

languages reflect a three-way spatial opposition, with stasis (locative)

opposedto motion, and, within the motion subcategory,motion towards a
target (various latives) opposed to motion away from a source (various
separatives).A few languageshave onlyone fully productivesuffix for each
spatial subtype,e.g. the three-membersystemof SosvaMansi (locative:lat-
ive:ablative), but most have at least one more. In richer case systems,
hypertrophyof the lative categoryis the norm.
Grammaticalcases.With the exceptionof SouthMansi, the corelanguages
preservelittle or no trace of the accusative*-m. It is unlikely, but just barely
possible,that the Hungarianpossessivesuffixes sl -m, s2 -d continueearlier
sequencescontaining the accusative« *-m-mV, *-m-tV). There is no trace
of an accusative*-m in Khanty. In Permian,the vocalic accusativesuffixes
of certainpersonalpronouns(e.g. -e, -e in the first-personsingularpronouns
Komi men-e,Udmurt mon-e)arethoughtto be continuationsof the stem-final
vowel, protectedby a final *-m which waseventuallylost.
At and nearthe periphery,evidencefor pU accusative*-m is fairly clear.
Southof NorthernSaami,WesternSaamichasdenasalizedreflexes,e.g. Lule
Saamigoade-v 'hut sA'. In Fennic, *-m > *-n in final position, and the *-m
accusativethus fell togetherwith the *-n genitive.Accusativesin -n arefound
throughoutmostof the present-dayNorth Fennicdialects(seeChapter3), and
morphophonemictracesof a suffix at least similarly shapedmay be found
elsewherein Fennic, e.g. weak-gradenn in Estonianvenna 'brothersG', cf.
strong-gradend in the nominativesingularvend.In Mordva, wheretherewas
a parallel, probably independent,syncretismof accusativeand genitive, both
of thesecasesare now marked with -ni, presumablyafter *m> n in final
position and with analogicalspreadof the automaticallypalatalizedvariantin
front-prosodicsequences;the details are not clear, however(seeChapter6).
There is no unambiguousreasonto doubt that the accusative-m of Mari is a
direct descendantof pU *-m. In Samoyedic,pU *-m is reflected clearly in
Tundra Nenets,e.g. myadO-mand Selkup, e.g. Taz Selkup m{ult-~2' m{ult-~2' both
'tent sA'.
Accusative*-m in the branchesofUralic:

S Mr Sa
Fe Md Mn

Despiteits broaderfunctional definition, the pU 'genitive' in *-n is slightly

lesswell attested;it hasvanishedwithout tracein all four corebranches:

S Mr Sa
Fe Md

Clear examplesof the continuationof a pU *-n genitive are Finnish vede-n,

ErzyaMordva vedj-eni,MeadowMari {3iio-;m" Taz Selkupiit-;m,2 'watersG'.
20 INTRODUCTION

In languageswith syllabic consonantgradation,a trace of earlier *-n lies in

the morphophonemicsof weak-gradeforms such as Northern Saami goatli
'tent sG', cf. nominativesingulargoahti, Estonianvenna 'sG' (homophonous
with the accusativevenna 'brother' cited above),and Nganasan,e.g. sG k;,ou
'fingernail', cf. nominativek;,tu.
Local cases.The most clearly attestedof the local casesis the locative
*-nA. The descendantsof this suffix still function as local casesin most
languages,either alone,e.g. -in of Komi tHn 'in a lake', or -on of Hungarian
tav-on 'on a lake', or combinedwith othermorphemes,e.g. the -na of Tundra
Nenets_kG-na in mya-kO-na 'tent sLoc', or the sa of -ssa in Finnish talossa
« *talo-s-na) 'in a house'.As a casesuffix, Finnish-nA now functionschiefly
as an essive('functioningas a'), e.g. sija+piiii=tt=ee-nii 'asa casesuffix', but
it still performsconcretespatial duties in postpositionsand other closedsets
of adverbs,e.g. talo-n taka-na HOUSE-G BEHIND-ess 'behindthe house',
kotona 'at home' (cf. koti 'home',kota 'Saamitent').
The separative*-tl (- *-tA) is reflectedin ablativeforms such as Moksha
Mordva oj-dj ;, 'butter sAbl' and Fennic partitives such as Estonian voi-d
'buttersP'. In Samoyedic,it occurswith the sameco-affix as the locative,e.g.
Tundra Nenetsmya-k¢-do 'tent sAb!', and in Fennic and Mordva it occurs
with the same*-s- coaffix, e.g. the elative forms Moksha Mordva kudfJ-stfJ
« *kudS-s-tS) and Finnish talo-sta « *talo-s-ta), both '(coming) out of the
house'.In Mari, a trace of the separativeremainsin the B at the endsof the
postpositionsgfJeY and deB, as in port gx '(coming) out ofthe house',je.s-fJi
deB FAMIL Y-s3 FROM 'from hislherfamily'. It is also thoughtto survivein
local-adverbial Permian casessuch as the Komi transitive, e.g. tuj vfv-ti
ROAD TOP-tr '(walking) along the road'. In Mari and in Saamiclanguages
from North Saarni north and eastwards,the initial n of the locative *-nA
becamet in caseforms built with the co-affix *-s-; this soundchangebrought
abouta collision with separativeforms, leadingto syncretismin theseSaamic
languages,e.g. North Saamidalus 'in/out of a house',and, in Mari, spurring
the formation of secondarypostpositionalconstructionswith deB and gfJeY.
Proto-Samoyediclocative *-k!2l-na and ablative *-k!2l-t(!2I) syncretized in
Selkup,as well, but at the expenseof the latter.
To judgeby the numberof different pU andpFU suffixesreconstructedfor
it, the lative category was the most used and differentiated in the proto-
language.For example,the reconstructionof the Permiancasesystementails
at least the proto-Iatives *-k, *-c i , and *-ni. The Permian egressivecase
suffixes (Komi -Siani, Udmurt -iSien.i) have been analysedas consisting,
historically, of three lative suffixes in succession(Serebrennikov1963: 63);
this is probably incorrect - see Permian, Chapter 8 - but such a poly-
morphemic backgroundto a deictic is not impossible, cf. English a-b-ove
from *ON+BY +ABOVE. The proto-Samoyedicdative *-1) presumablygoes
back to an earlier lative, as do the proto-Samoyediccoaffixes *-k!2l- and
*-nt!2l( -). The *-s- coaffix of the secondarylocal cases of the western
 INTRODUCTION 21

languagesis also thought to have beenoriginally a lative; accordingto this

line of thinking, the -s- of the Finnish interior local cases(inessive-ssA <
*-s-nA, elative -stA < *-s-tA, and illative *-sVn) is historically identical with
the -s of such fonns as ulos '(moving) towards the outside', cf. ulko-na
'(located) outside'. Mordva uses this -s as its illative case (kudo-s 'into a
house')but it also seemsto have preservedlative *-n j (in reduplicatedfonn
in the dative/allative case, e.g. Erzya kudo-njenj ), lative *-k (seen in the
prolativecase,e.g. viri-ga 'througha forest'), andlative *-1] (in the somewhat
restrictedlative case,e.g. mastor-ov'to the ground').
The case system of the protolanguagewas probably complementedby
postpositions;theseare discussedbelow (p. 23).

Number
Numberin the protolanguagewas markedwith two different suffixes, but in
all likelihood plurality was the salient semanticcomponentof neither. One
suffix, *-t, seems to have perfonned duties analogousto the absolutive
(nominative)-{j in the singular,but with additional meaningsof definiteness
and non-singularity. The other, *-j, functioned as a non-singularanalogue
both to the singularaccusative*-m, marking certaindirect objects,and to the
singular genitive *-n, marking nouns in prenominal position (qualifiers,
possessors).This *-j suffix also preceded further inflectional suffixes;
whethercasesuffixes were so usedis unclear,but the personsuffixes must
haveoccurredin suchstrings(Janhunen1981a:29).
The plural marker*-t is preservedas a suffix of nominal inflection only at
or nearthe periphery.Note the nominativeplural of 'tree': Taz Selkuppoo-t,
SosvaMansijiw-~t, Mansijiw-~t, Finnishpuu-t, MokshaMordva sufi-to Thereis no reason
to doubt that the glottal stop of the Nenetsand Nganasannominativesplural,
e.g. ForestNenets~a-1 'trees'
'trees'and NganasanmUlJku-1'forest' are also from
this *-t; the Kamassianpluralizer- lje 1- -- lji 1- - -?i seemsto havecontained
both *-t and*-j. New plural suffixeshavereplaced*-t in thenounparadigmsof
Mari (pu-wlak), Permian(e.g. Komi pu-jas),andHungarian(jaa-k), although
Mari haspreserved*-t in collectivesfonnedfrom kinshiptennsandnames,e.g.
awa-m-~t awa-m-~tMOTHER-sl-plur'my motherandherassociates'.
The other plural marker, *-j, now marks plurality of possessionin
Hungarian, e.g.faa-i-m 'my trees', cf. faa-m 'my tree'. In many Fennie
languagesit is usedin all non-nominativecases,e.g. Finnishpu-i-ssaTREE-
plur-ine 'in trees'; Estonianuseseither *-t- or *-j-, e.g. puudes/puis.Plural
genitive fonns such as Finnish kalojen « *kalo-H)-en < *kala-j-t-en) 'fishes
pG' show that the two plural suffixes could co-occur in the west, as well;
contrastthe parallel (stylistically marked)kala-i-n 'fishespG' < *kala-o-en,
Suhonen1988: 309. In Nganasanand Nenets,*-j combinedwith stem-final
vowels to producecomplexvowel alternations.
In North Saarnithereis essentiallyno agglutinativeplural fonnation; each
caseis different in singular and plural. This may beenseen,for example,in
22 INTRODUCTION

the singular/plural forms of the North Saami word for 'name', namma,
namely illative namma-i/namma-i-de(with singular *-j, plural *-i-tee-k),
locative nama-s/nama-i-n(with singular*-st, plural *-ij-n). In sharpcontrast
is e.g. AIjeplog Saami,in which singularand plural forms of all local cases
werekept distinct, cf. illative namma-j/nama-j-ta,inessivenama-n/nama-j-n,
elative nama-st/nama-j-st (Lehtiranta1992: 156).

Person
Personis a morphologicalcategoryfor nominalsin most Uralic languages.
The most widespreadfunction is to expressthe person of the possessor,
e.g. North Saamiahcca-n,Finnish isii-ni, Erzya Mordva tjetja-m, EastMari
aCia-m,Udmurt ataj-e, Hungarianapaa-m,NganasantPesr-m::J,ForestNenets
njedaa-j/njedaa-m,Taz Selkup i si -mi~i si -mi~ all FATHER-s1 'my father'. A
secondary,but quite widespread,use of the person suffixes is to indicate
definiteness.The third- and second-personsingular suffixes are most often
recruitedfor this purpose,e.g. third personsingular-(r)s in SouthernPermiak
gor-is kerku-Sii-s mun-i-s STOVE-s3 HOUSE-ela-s3 GOES-past-s3'the
stovewent out of the house',secondpersonsingular-ill in Komi starik-ill-len
pi-is OLD.MAN-s2-adeSON-s3 'the old man'sson'. Note alsothe useof the
third personplural in LipshaMari kok iioiir-iist-wlii TWO GIRL-p3- plur 'two
of the girls'.
In languageswith relatively transparentnominal morphology, person
suffixes generallyfollow a numbersuffix, e.g. Udmurtpinal-jos-r, Hungarian
djiirekii-i-m, Taz Selkup iija-ii-m(iJ, all CHILD-plur-s1 'my children',
althoughin some languages the reverseorder is usedto expressassociative
plurals, e.g. Hungarianapaa-m-eekFATHER-s1-assoc.plur'my father and
his friends'. Mari, which has one of the most transparentnominal morpho-
logies in Uralic, permitsboth orders.
The position of personsuffixes relative to that of casesuffixes is more
complex.Therearethreepatterns:(1) personbeforecase,typical ofUgric; (2)
casebeforeperson,typical of northernSamoyedic,Saamic,Fennic,andErzya
Mordva; and (3) mixed, i.e. both ordersoccurringin the samelanguage.The
languageswhich show mixed order do so in two different ways: in Mari, the
order dependson the case,while in the Permianlanguagesand in Moksha
Mordva it dependson both caseand person.For detailed discussions of the
existingpatterns,with diachronicinterpretation,seeComrie 1980bandHonti
1995.
Other nominal categorieswhich are expressedby means of suffixes
include the definite declensionin Mordva, the predestinativeof northern
Samoyedic,andthe pastandfuture clitics in Nganasan.
All present-dayUralic languageshave postpositions,and we can assume
that in the protolanguagethere were alreadya numberof nounswhich were
used primarily to specify spatial and temporal relations. The most clearly
reconstructableof thesearethe antonymspU *i1a 'spaceunderneath'and *ti1i
 INTRODUCTION 23

'space above'. We can glimpse something of the pU noun phrase by

comparingconstructionssuch as Finnish vede-nalla, TundraNenetsjidO-h
ngilna 'under(the) water', in which the locational nounsalla and ngilna are
fossilized old locativesof the pU locationalnoun *i1a and Finnish -n, Nenets
-h are the genitive suffix. Whetherthe protolanguagehad a true genitive case
or not, somekind of subordinatingsuffix *-n probably optionally, but often,
connectedthe two nounsin constructionsof this sort.
In languagescloser to the core, which have no genitive case, such
postpositionsmost frequently take the nominative, e.g. Hungarian astal-fJ
alatt, Komi pi'zan-fJ uli'n, TremjuganKhanty pfJsaani'i'lpi'nJ 'under(a) table'.
But the nominativeis the norm in Mari and Mordva, as well, in spite of the
fact that theselanguageshave genitives inherited from pU *-n. In Fennic,
many suchforms occuroptionally or exclusivelyas prepositions,e.g. Finnish
(stylistically marked) alla veden 'under (the) water', and the genitive
competeswith the partitive, the partitive usually having the upper hand in
prepositionalconstructions,e.g. Estonianveealla (WATER-sGUNDER.loc)
alongsidealla vett (UNDER.loc WATER-sP) 'under(the) water'.This useof
the partitive in Fennic is probably connectedhistorically with the use of its
cognateanaloguein Mordva, the ablative, with certain postpositions,e.g.
Moksha Mordva siedii-d;}--fJk baska HEART-abl-p1 APART.FROM 'apart
from our heart'. In Northern Saarni both postpositionaland prepositional
constructionsarecommon,andthe lexical nounis alwaysin the genitive.
Postpositions(and more rarely prepositions)often have more than one
form, and these forms can be seen as making up small case paradigms,
ranging from threeto five members.In Mari, Hungarian,and ObUgrian,the
most commontype is the three-memberset, and we may think of this as the
basicUralic minimum. Casesuffixes which are often different from thoseof
normal noun paradigms are used to distinguish stationary location from
motion, and within the motion category,goal (latives) from source(separ-
atives).Four-membersetsare typical of Samoyedic,Mordva, Fennic(but not
Estonian, whichhasthree) and- historically - Saarnic,i.e. of languagesat or
near the periphery.The present-daythree-membersetsof Saamiclanguages
are the result of syncretismof caseform and function (location and goal) or
the loss of prolative forms. In the other three groups,the fourth memberis
synchronicallya prolative ('moving along the expanse/length of X'), i.e. it is
a further subdivision of the motion category.The Permianlanguageshave
evolvedthe richestpostpositionalcasesubparadigms,with the addition of the
terminativecategory,for the terminuspostquemnon of motion.
Table 1.2 sets out the postpositionsmeaning 'behind, in back of in a
selectionof Uralic languages.
The origins of the nouns usedas postpositionsare sometimessynchron-
ically transparent,becausean unbounddoubletexists,e.g. Komi pi'zan vil-in
'on the table', mu+viv (EARTH+TOP) 'surfaceof the earth'. More usually
24 INTRODUCTION

Table1.2 SelectedUralic postpositions'behind,in backof'

Stasis Motion
Source Goal Trajectory Terminus

Hungarian mogott mogtil mogee

Vakh Khanty COI).IFI cOl).cooy cOl).c(aa)
Estonian taga tagant taha
MokshaMordva ftal~ ftaId~ ftaId~ ftaId~ ftalu ftalga
ForestNenets punniaana puuniaat puuniaal) puuniaamna
Komi saji"n saji"si saje sajti sajedzi

they are old enoughas to be invisible to the untrainedeye, e.g. Hungarian

pista mogott 'in back of Pista', pista megjott 'Pista came back' or can be
recoveredonly by the comparativemethod,e.g. North Saamicada 'through',
a fossilized lative of the root found in Moksha Mordva siedji, Finnish
siidiimme- 'heart'.
Personal pronouns.The reconstructionof the Uralic personalpronouns
encountersdifficulties specific to this word class.Like numeralsor kinship
terms, pronounsenter easily into analogicalsubpatternswhich allow cross-
infection.A further difficulty arisesfrom the relatively small sizeof the proto-
Uralic pronominalroot, which was canonicallymonosyllabic(C)V(C). This
contrastssharply from the roots of denotatives,which were all bisyllabic
(C)VC(C)V, though they probably alternated morphophonemic ally with
(C)VC becauseof root-suffix sandhi.
It is thereforewith great reservationsthat even the following imprecise
indicationsare given. First- and second-person pronounswere distinguished
by their initial consonant,which was *m in the first person,*t (? - *n) in the
second.The initial consonantwas followed by a vowel of indeterminate
quality: sl *mV, s2 *tV. In circumstanceswhich are not clear, these syllables
were extendedby an element *nV or (in Hungarian and ObUgrian) *l);
whether these elements were derivational suffixes or were historically
identical with local suffixes, perhapsserving in an ergative function, is
unknown. Insteadof the *n V-element, a plural suffix (*-k or *-t) could be
added:pI *mVk, p2 *tVk. There was also a dual, formed, perhaps,with an
element*jn, viz. dl *mVjn, d2 *tVjn.
The nominative forms of the axis-of-discoursepersonal pronouns of
Arjeplog Saami, Tremjugan Khanty, and Forest Nenets are given here for
comparison:

sl s2 dl d2 pI p2
AS mfm(na) t:1nltihna maaj(ah) taaj(ah) mij(ah) tij(ah)
TK mlHi nOl) miin niin mel) nel)
FN manj pit maj/madjii pic/picii manja? pita?
 INTRODUCTION 25

These samples present three degrees of unifonnity in personal pronoun

morphology, with the Saamic showing the most, and the Nenets the least,
evidenceof analogicallevelling.
Root suppletionis found in the personalpronounparadigmsof Hungarian,
Nenets,andEnets,e.g. the nominativeandaccusativeforms of the fIrst-person
pronoun are Hungarian eenliinge-m,Forest Nenets marJ/sa?-j, and of the
secondpersonsingular,tiilteege-dandpitlSaa?-t(-ml-j and -d/-t are fIrst- and
second-person suffIxes).Theaccusativeforms in Samoyedicseemto go backto
an earlier*kit (in ablautto *kat) 'face',and the -ge- of the Hungarianforms is
perhapsfrom the sameroot (Helimski 1982: 88-94).In FennicandKhanty, the
personalpronounsuseaspecialaccusativesuffIx -t, asin theaccusativeforms of
the fIrst-personsingularpronounFinnish minu-t, TremjuganKhanty miiiin-t.
This -t is probablyhistorically identicalwith the HungarianaccusativesuffIx -t
which is usedin most nominal forms (and optionally also in pronouns),e.g.
Mla-t 'fIsh sA' , iinge-mii-t, cf. iinge-mabove.
Theevidencefor theUralic third-personpronounis not quite asclear,but itis
probablethattheprotolanguage hada form *sF (whereinF = front vowel) which
hadanaphoricpropertiesat least.This pronoun,enlargedwith various,usually
nasal, suffixes,hasclearreflexesin theperipheralFinno-Ugric languages,butif
it occursin Samoyedicit is only in the Selkupthird-personsingularpronoun
feP2 (Hajdu 1990: 2-3). Examplesfrom Finno-Ugric are North Saami and
Mordvason,Finnishhiin, TremjuganKhanty leyW,SosvaMansitaw, Hungarian
00.
Reflexive pronounshave evolved separatelyin the various branches;the
most widespreadis seenin North Saami iehca-, Finnish itse, Udmurt and
Komi aB, a word which hasbeenconnectedwith the demonstrativepronouns
basedon *e- and *cjF, but which is more likely originally to havebeena noun
meaning '(shadow) soul', cf. SosvaMansi is 'shadow (of human); ghost'.
Samoyedic,Selkup, and Nganasanuse one root (e.g. NganasanffazSelkup
IJon~n;yon-iik 'I myself, IJon~nt;yon-iintr 'you (sg) yourself), while
Kamassand the Nenets languagesuse various words which are synchron-
ically identical or doubletsto words meaning 'body' or 'head',e.g. Kamass
(man) bos-pfJ 'I myself, cf. bos 'body', a relatively recentbifurcation of a
loan from Turkic (Joki 1952: 98-9; Hajdu 1990).
Demonstrativesand other deictics. Proto-Uralicprobablyhad a three-way
systemof demonstrativepronounswhose spatial extremeswere built onto a
base*t-, with front vocalism (*tF) correspondingto proximal reference,e.g.
Finnish tii= in tiimii 'this', and back vocalism (*tB) correspondingto distal
reference,e.g. Finnishtuo 'that'. Reflexesof thesetwo pronominalbasesmay
be found in all branchesof Uralic; in somelanguages,one or the other has
evolved into a third-personpronoun, replacing the earlier *sF mentioned
above,e.g. Estoniantema, EastMari tudo '(s)he/it', whilein othersthey are
now adverbs or have become otherwise fossilized, e.g. Udora Komi tin
'there',Hungariantee+tova 'hesitant« * 'this way + that way').
26 INTRODUCTION

The middle term of the three-waysystemhas beenreconstructedas *ciF,

i.e. *c j plus a front vowel. The function of this pronounis not quite clear: it
may have been anaphoric; it may have referred to an area closer to the
addressee;perhapsboth usesexisted side by side, as in Finnish se 'it, that
(anaphoric)',si-e-llii 'there(anaphoric,or closerto addressee)'. In Mansi and
Hungarian,all tracesof *c j= have beenlost, but it survives in all forms of
Khanty, e.g. EasternKhanty tjii(t) 'this'. In Samoyedicit has survived as a
demonstrativein Enets (setPeo 'that [one]') and developedinto the third-
personsingularpronounin Nganasan(sin").
Parallelto the pairedsetof frontlbackdemonstrativeswith initial *t, which
is well attestedin all branches,there seemsto have been another distal/
proximal pronoun (or prefIx?) pair *i- - *a- 'this', *u- - *0- 'that', as in
Hungarianii=z 'this', a=z 'that', SosvaMansi ani 'now', Udmurt oti"n 'there'.
Neither pendanthas a reflex in Saamic,and the back-vowelpendanthas no
reflex in Fennic or ObUgrian. Besidethe poor distribution of the witnesses
and the variety of functions, the scanty phonological substanceof this
reconstructionrendersthe etymologyprecarious.
At the heart of the demonstrativepronoun systemsof the present-day
languagesis the simple two-term proximal/distal dichotomy, e.g. Finnish
tiimiiltuo, Erzya Mordva tje/tona, Hungarian iiz/az 'this/that'. Different
languageselaborateon this in different ways: (la) differentiation of the
member which refers to remoteness,yielding two (or more) degreesof
remoteness, e.g. late proto-Finnic*taa alongside*too (Laajavaara1986),(lb)
similar differentation of the member which refers to proximity, e.g. Vote
ka+se 'this (closerto speakerthan se), or (1c) both, e.g. Hungarianiim--ii=z
'this (closer to speakerthan ii=z), am--a=z 'that (further from speakerthan
a=z); (2) addition of the factor of visibility and/or concreteness, e.g. Vakh
Khanty tiimii 'this (visible/concrete)" tjiit 'this (non-visible/non-concrete),;
(3) addition of, and interaction with, anaphoricpronouns,as in Finnish, or
Forest Nenets, where the distal/proximal pair is cehafjlcuhkii and the
anaphoricpronounis Cihkii. In Mari, wherethe reflex of the *cjF pronounis
chiefly anaphoric,the asymmetryof the inherited*tB : *tF :: *cjF systemhas
beenresolvedwith the analogouscreationof a more remoteanaphoricform
sade, i.e. parallel to tideltudo 'this/that' East Mari has sede/sade'this/that
(anaphoric)'.In a similar fashion, southernand easternvarieties of Khanty
evolvedback-vowelpendantsto their reflexesof *cjF, e.g. Vakh Khanty tjuut
'that (non-visible/non-concrete),.
The most complexsystemsof demonstrativesare found in Saamic,where
elaborationsof types(1) and (3) both occur, asin NorthernSaami:
 INTRODUCTION 27

Anaphoric Demonstrative

Proximal Distal

Near speaker Nearaddressee Far Further Furtheryet

dat ddt diet duot dot diU

Verbs
There is no one clear past-tensemarker reconstructablefor pU, and the
original tensesystemmay havebeenone similar to that found in Samoyedic,
where the lexical, i.e. intrinsic, aspectof verb roots determinesthe semantic
force of their finite forms. Thusan inherentlynon-perfective,stativeverb such
as 'lives' normally had present ornon-pastmeanings,while a verb such as
'dies', with inherent punctual, perfective aspect, normally had past-tense
meaning. Modifications to the basic temporal setting of a given verb were
effectedby meansof derivational suffixes, e.g. '(s)he is (in the processof)
dying' was expressedby appendinga deperfectivizingsuffix to the root.
It is perhapsfrom such derivationalsuffixes that the oldest attestedpast-
tensesuffixes evolved.Theseare pU *-sj(A)- and pFU *-i-, distributedfairly
widely across the Uralic or Finno-Ugric branches;although traditionally
treated separately,it is possible that these two suffixes are historically
identical (Helimski 1996: 40; on the suspectnatureof intervocalic pU *-sL,
seeAbondolo 1990).
Distribution of past-tense*-sL:

Mr? - Kh Sa
(Fe) Md? - Mn

Distribution of past-tense*-i-:

S MrPn Kh
Fe Md Hu

The representationof *-sj(A)- in the westernlanguagesis either restricted

(e.g. in Estonian,whereit is limited to the negativeverb) or controversial(it
is thought to be reflected in certain Mordva past-tenseforms, and in the
inflection of the so-called'second'conjugationverbsin Mari). The past-tense
suffix *-i- is clearly preservedin forms such as Finnish tul-i-n 'I came',
Mordva ud-i-n! 'I slept' (cf. the correspondingnon-pastforms tule-fJ-n, uda-
fJ-n), and may be reconstructedfrom the phonologiesof forms such as East
Mari toP~ toP~ < *toIS-jS-n < pFU **tul'i-i-n 'I came'(cf. nonpasttola-m), North
Saarni gohccon 'I called' < proto-Saarnic *kocjcjaa-je-m (cf. nonpast
gohcun).
28 INTRODUCTION

Most Uralic languageshavegone on to enrich this simple tensesystemin

diverse ways. In addition to the basic past : non-past opposition, many
languageshave evolved various kinds of future (either via the inflectioniza-
tion of derivational material, as in Tavda Mansi or in Udmurt, or by means
of auxiliary verbs, as in Hungarian (jog-) or Mordva (karma-). Different
degreesof remotenessin past time may be expressedby compound-tense
forms built with forms of a verb meaning 'is' or 'becomes',as in the
pluperfectforms North Saamile-dje-n boahta-n,Finnish ol-i-n tul-Iut 'I had
come'. In some languagesthe auxiliary verb has fused with the stem and
become an inflectional suffix in its own right; the meaning is then most
usuallyimperfective,e.g.Mordva secondpastmor-ill-inl 'I usedto sing'.Also
usually imperfective are certain secondarypast-tenseforms of other lan-
guages,in which the shortestinflected form of the auxiliary has becomea
frozen, uninflectableclitic, e.g. the WestMari third-personsingularpast-tense
form of the verb 'is', flll fl, when combinedwith the non-pastfinite form of
the lexical verb, forms a pastcontinuoustense:tola-@-m fJIl fl 'I was corning',
cf. the synonymous,and morphologicallyparallel Udmurt li'kt-islk-o val and
Hungarian(archaicand dialectal)jov-@-ok vt'llt'l. Convergencewith Turkic is
evidentin all suchconstructions(Bereczki 1983: 218-21).
Another type of remoteness,but one not tied exclusivelyto time, is that of
the inferential. This categorygoes under many different names(evidential,
non-experiential,non-eyewitnessed,narrative, auditive) and is sometimes
classifiedas a mood and not as a tense.In Estonianand Livonian it coincides
with the quotative,i.e. it is the ratio obliqua analogueof the indicative (see
Chapter 3); historically the Fennic quotative derives from a non-past
participle *=wA. In the Permianlanguagesand Mari the inferential category
has been ascribedto Turkic, specifically Volga-bend, influence (Bereczki:
1983). The inferential here is usually classifiedas a tense;the forms derive
historically from deverbalnorninals (which evolved into past participles) in
*=mA. In Mari, the syntagmsclassified as inferential are the same as the
secondary,compound-tenseforms mentionedabove. Permian on the other
hand has developeddistinct, but defective finite inferential paradigms,e.g.
Udmurt so gi'r-oz val '(s)heusedto plough (long ago)', so gi'r-oz vilem '(they
say that) (s)he ploughed; (s)he ploughed (as I remember),.Finally, infer-
entialsarealsopresentin Samoyedic,wheretheir morphologicalimplementa-
tion is usually suffixal, cf. Selkup =NT- (probablyhistorically identical with
the non-pastparticiple=NT-), Nganasan=HATU-.
Proto-Uralic had probably no more than two morphological moods, an
unmarkedindicative and an imperativein *-k(V)-. Tenseandmood were thus
in complementarydistribution in the sensethat the imperativeknew no tense
distinctions. By the Finno-Ugric stage,new mood markers, which marked
unreal--conditional-desiderative sorts of modality, had begunto develop. In
most languages,the reflexes of thesemarkersare composites.For example,
the Estonianconditional -ksi- is probablythe continuationof a Proto-Fenno-
 INTRODUCTION 29

Saamicderivational complex *=(i)ksji-, seenin Finnish frequentatives;the

Hungarianconditional -nee-/-naa-probablycontinuesa suffix chain *-ne-i-,
in which *-i- is the past-tensesuffix mentionedaboveand *-ne- is the suffix
seenin the Finnish potential (sata-ne-eRAINS-pot-s3 'it may rain') and the
Mari desiderative(Hill Mari ;;,l;;rne-z;;, LIVES-desid-s3 '(s)he would like to
live'); the Finnish conditional-isi- (as in tul-isi-n 'I would come')consistsof
past-tense-i- precededby the proto-Fenno-Saamic suffix *-nsji- seenin the
North Saamipotential-i-/-Zi-, -1515- e.g. bOtleian 'I may come'.As mentioned
above, many Uralic languageshave verb subparadigmswith epistemic
overtones,and theseand other forms which refer to the speaker'sknowledge
or attitude about the predication and the utteranceare often classified as
moods, particularly in Samoyedic;cf. Tundra Nenets, with sixteen moods
which cross-classifyvarious kinds of evidential, obligational, and attitudinal
modalities.
Most Uralic languages(but not southernSamoyedicor Hungarian)express
negationby meansof a negative(auxiliary) verb, usually with an irregular,
suppletive,or defectiveparadigm.There is considerablevariety in terms of
which verbal categoriesare encodedon the negativeauxiliary and which are
encodedon the lexical verb; for a surveyseeComrie 1981.
Most verb subparadigmsshow agreementfor subject person, and in
Mordva, Ugric, and Samoyedic,for certain features of a direct object, as
well:

Kh Sa
Md Hu Mn

The paradigmswhich encodeinformation aboutthe direct object are usually

classifiedasbelongingto 'objective'or 'definite' conjugations,the latterterm
being inspiredby the fact that suchforms are often associatedwith somekind
of definitenessin the direct object, whether intrinsic or governedby the
discourse or narrative. Since nouns in many languageshave their own
morphologicalmeansof expressingdefiniteness,the result can be redundant
definitenessmarking, as in Hungarian,wherea noun phrasemorphologically
marked for definiteness(e.g. by meansof personsuffix or definite article)
forces the selectionof 'definite' suffixation on the verb. Oth~r
verb. languages,
however, for exampleMordva, use indefinite and definite suffixation with
both indefinite and definite direct objects to expressa range of varieties of
definitenessand aspect.
The definite conjugationis thoughtto havebeennascentalreadyin proto-
Uralic; at this early stageprobably no more than third-personobjects were
encoded, by means of the encliticization of the third-personlanaphoric
pronoun*sF mentionedabove.SouthernSamoyedic,particularly Kamassian,
has elaboratedthis primitive basis the least; Hungarianhas developedfull
definite paradigms,with forms distinct from their indefinite analoguesin
30 INTRODUCTION

most subject persons,and with a formal distinction betweensecond-and

third-personobjectsif the subjectis fIrst personsingular.While the defInite
conjugations of the ObUgrian and northern Samoyediclanguagesdo not
distinguish object person, they do show number agreement,with varying
syncretismsanddegreesof obligatoriness.The defInite conjugationsof Erzya
and Moksha Mordva are the richest in Uralic, in that they show agreement,
albeit partial, with both personandnumberof the direct object.

Syncretism and Suppletion

From the typological point of view, it is conventionalto think of the Uralic
languagesas agglutinating,i.e. as conformingto onedegreeor anotherto that
broadlinguistic type in which eachgrammaticalcategoryis represented,in
the perceivablesubstance,by its own more or lessimmutablemorpheme.For
this classifIcationto work, the defInition of agglutinationmust be flexible
enough to allow the qualifIcation 'more or less', i.e. it must tolerate, or
disregard,regular and phoneticallybanal (morpho)phonologicalalternations
suchas frontlback and [+/-] roundedamongthe vowels, and assimilationsof
stricture-typeor palatalizationamongthe consonants.On the other hand we
do not expectto fInd, in a languagewhich we haveclassifIedas agglutinating,
large amounts of either syncretism or suppletion. In fact, both of these
phenomenaare ratherwidespreadin the Uralic languages.What follows is a
brief sampling of some of the subtypes,beginning with syncretism and
concentratingon the nominal paradigm.
In the nominal paradigm,there is often syncretismof casewhen person
sufftxes are involved, e.g. the Finnish sN (kiisi) sG (kiiden) and pN (kiidet)
all syncretize in possessedforms such as kiiteni 'my hand(s) sNGpN'.
Similarly, Komi -a- servesas both locative and lative in ki-a-m HAND-slne/
sIll-sl 'in(to) my hand', contrastki-i'n 'in althe hand',ki-e 'into althe hand',
and Selkup -qii- is locative, lative, and elative in utoo-qii-k HAND-sIne/sIll!
sEla-sl 'in(to)/out of my hand'.The reversescenario,syncretismof personin
certain caseforms, also occursin Selkup, e.g. sInstr nomti'sii 'with your/his
(her/its) god', a form which syncretizess2 with s3; contrastthe nominative
singularforms, in which the personforms remaindistinct, viz. nomli's2,nomti'
s3 (d23 andp23 syncretizein parallelfashion).
Occasionally,syncretismof caseis connectedwith non-singularnumber.
For examplein Nenets,which distinguishesnominative from genitive from
accusativein the singular and plural paradigmsof most nouns, there is a
completesyncretismof thesethree grammaticalcasesin the dual: for 'tent'
we have dNGA myakOh,contrastsN myaq, sG myadOh,sA myadom.Selkup
syncretizesthe dative/allativewith the illative in the non-singular,e.g. dDffil
nop-qlt-kim '(in)to two gods' contrastsD nuu-ni'k, sIll nom-ti·. In Finnish, the
accusative syncretizes with the genitive in the singular, but with the
nominativein the plural, in a mannerreminiscentof the animate/inanimate
accusativesof Russian.The reverseof this scenario,namely syncretismof
 INTRODUCTION 31

numberin certaincases,is more suspecton theoreticalgrounds.We may say

that NorthernSaamibeanan'as a dog/asdogs' is a syncretismof singularand
plural essive; but such forms look more like derivation, i.e. they resemble
adverbs, and the suppressionof the number category is thus perhapsnot
particularly noteworthy.A striking exampleof number syncretismexists in
the absolutedeclensionof Mordva, whereall non-nominativecasessyncretize
for number,e.g. sN kudo 'house'*" pN kudot 'houses',but spIne kudoso 'in
a house/inhouses',splll kudos 'into a house/intohouses',etc.
Syncretismof personin connectionwith non-singularnumberalso occurs,
e.g. in Sosva Mansi, where we have the syncretism of second and third
personsdual when the stemis markedfor dual number: saali-ay-eenis 'two
reindeer d23', but the singular paradigm distinguishessaali-jin 'the one
reindeerof you two' from saali-teen'the one reindeerof them two'; second
andthird dual forms cansyncretizein the plural, as well. The reversescenario,
namelysyncretismof stemnumberin connectionwith certainpersonalforms,
is found in Mordva, e.g. Erzya Mordva kudo-t 'your (sg) house(s)'.Such
syncretismsoccursin all personalforms savesl and s3 in Erzya, cf. kudo-m
'my house',kudo-n 'my houses',kudo-zo 'hislher house',kudo-nzo'hislher
houses'.Even thesedistinctions break down in non-nominativecases,e.g.
kudo-so-nzo'in hislher house(s)'.Syncretismof personnumbermay be seen
in a form suchas DN Khanty xoottaat'housessp3'; contrastthe singular-stem
forms xoot;}t 'houses3' *" xooteet'housep3'.
Suppletionof casesuffix in thepresenceof personinflection, apartfrom that
alreadyseenin connectionwith suchsyncreticcasesas Selkup-qii-, is rare. It
doesoccur,however,in the dativeof TundraNenets,which is (in the singular)
-n¢-h - t¢-h when usedwithout, but -x¢- when usedwith personsuffixes,e.g.
mya-tOh TENT-sD 'to a/thetent', mya-kO-n TENT-sD-s1 'to my tent'. Thelatter
form illustratesthereversesuppletivetype, thatof personsuffix in thepresence
of case,contrastmyadO-m'my tent', with -m insteadof -n for s1. This type of
suppletionis widespreadin theperipherallanguages,andmayreflectanancient
stateof affairs: it is probablethat the nominal first-personsuffixes in Fennic
(e.g.Finnish-ni) andSaamic(e.g.North Saami-m) eachreflectoneof anearlier
suppletivesuffix-pair *-mi : *-ni, with the latter suffix occurring in oblique
casesand/orwhenthepossession wasnon-singular,a distinctionstill preserved
sporadicallyin Mordva,cf. kudo-m'my house',kudo-n'my houses'citedabove
(Korhonen1981: 233-6,244).Thatthe dichotomy*-mi : *-ni wasthe resultof
phonotaxis, with *-ni coming from an even earlier *-n-mi, has not been
demonstratedto everyone'ssatisfaction,cf. Honti 1995: 59-61. The reverse
scenario, namely suppletive number suffixes in the presenceof person-
marking,is seenin Hungarian(jaa-k :faa-i-m, citedabove)andObUgrian,e.g.
VasjuganKhanty kaat-~t
KhantyHOUSE- plur 'houses': kaat-laa-mHOUSE- plur-
sl 'myhouses'.
Suppletion of case suffix in the presenceof number inflection is also
encountered.As mentionedabove, the casesuffixes for singular and plural
32 INTRODUCTION

nounsin North Saamidiffer to such a degreeas to makethe term suppletion

seem inappropriate.Elsewherein Uralic such suppletion is most usually
centredon the formation of the genitive, accusative,or both, e.g. Finnish sG
-N but pG -'tEN, ForestNenetsCioonia-m 'fox sA' but Cioonli-i 'fox pA'.

Syntax
The head noun is the centre of the typical Uralic noun phrase,in that it is
followed by any inflectional suffixes and/or postpositions,and precededby
any modifying adjuncts.Agreementbetweenheadnounsand their adjunctsis
fairly rare. Perhapsthe most glaring exceptionis Finnish, e.g. tuo-hon sama-
an iso-on rakennukse-enTHAT-ill SAME-ill BIG-ill BUILDING-ill 'into
that same big building', but Hungarian has case and number agreement
betweendemonstrativesand their head, as in the synonomousudian+ab-bt'l
a nildi eepiiliid-bii SAME+THAT-ill DEE ART BIG BUILDING-ill 'into that
samebig building'. Agreementalso occurs elsewhere,to varying degrees,
especiallyin Saamic,Fennic,and northernSamoyedic.
Possessionis expressedby the linking of noun phrases.In the peripheral
languages,the possessoris put in the genitive and the possessionis left
unmarked,e.g. Finnish lapse-nisii, North SaamimanaahcCi, TundraNenets
nyu-h nyfsya, Taz Selkup iija-n esi", all '(the) child's father'. In the
core languages,the possessoris marked in a variety of ways. Hungarian
marks the possessorwith zero, i.e. the nominative, or, less frequently, the
dative (-nAk), and marks the possessionwith the third-personsuffix, e.g. a
diiirmek(-niik az) ap-ja DEEART CHILD(-D DEEART) FATHER-s3 'the
child's father', apaa-m-(nak a) huug-a FATHER-sl(-D DEEART)
YOUNGER.SISTER-s3 'my father's younger sister'. Permian uses the
adessive,with or without s3 marking on the possessed, e.g. Komi mort-lenpi
'MAN-ade BOY 'a man's son', Permiak es=ema-Simizi"k-ven dieniga-jez
DISAPPEARS=inferential-plurMAN-ade MONEY-plur 'a man'smoneyhad
gone missing', but marks the possessorwith the ablative if the possessedis
the direct object,e.g. Komi kozi--pemen-sli-mlov LEA VES.imp PRO.sl-abl-
sl SOULILIFE 'sparemy life!', cf. also Permiaketik inlka-viSi gu-vviSi si"r-
rez Sioj=ema-Sigos-0 ONE OLD.WOMAN-abl CELLAR-subl MOUSE-plur
EATS=inferential-plurFAT-sN 'mice had eatenthe fat in an old woman's
cellar'. In the ObUgrian languages,the possessoris invariably in the
nominative.The possessiverelation may optionally be renderedexplicit by
placing the possessedin the third person, e.g. Pim Khanty juuyW toj(-;JI),
SosvaMansijiw tazix(-e), bothTREETOP(-s3) 'the top of the tree'. Mari also
often marks the possessionwith the third-person suffix, even when the
possessoris explicitly stated in its genitive form, e.g. aca-m-;Jn iio;Jr-zo
FATHER-sl-sG SISTER-s3 'my father's sister'. In Mordva, a complex
systemof degreesof definitenessmarks the possessorwith the definite or
indefinite genitive, and the possessedin threedifferent ways: with zero (as in
 INTRODUCTION 33

the peripheral languages),with s3 (as in Mari), and with its definite

declension,e.g. kvartiira-ni keIJksFLAT-G DOOR '(the) door of (the) flat'
kvartiira-ni--ti keIJkse-zeFLAT-G-def DOOR-s3 'the door of the flat', ejdie-
ni avaridiema--siCHILD-G WEEPING--sNdef'(the) child's weeping'.
Possessivesentences,i.e. possessiveconstructionswhich involve verbs,
are more complex, as is to be expected.Most Uralic languageslack a verb
meaning 'has'; the notable exceptions are the ObUgrian languagesand
Nganasan,e.g. Cingali Khanty taj- in taj-fJ-k;m pax-fJ HAS-past-d3BOY-N
'they (two) had a son', Sosva Mansi oonisi- in tas-fJ at oonisi-eeyw-;»n
VESSEL-N NEG HAS-pres-sl 'I don't have a pot', and Nganasanxon- in
b;ms::rg~j-fJ oonisi- ni'--j xon-ti~g~joonisi- ALL-dual-N WOMAN-pA HAS-present.
continuous-d3 'they both have wives'. However, normally to express
somethinglike '(s)hehas a knife' the most widespreadUralic sentence-type
usesa verb meaning'exists',and marks possessor,possession,or both, with
caseand personsuffixes. Mari and the Permianlanguageshavethe simplest
construction:a verb meaning'is/exists' (or its negativependant)is addedto
a regular possessivenoun phrase,e.g. aca-m-;m iio~r-zo ulo 'my father has
aca-m-;m
a sister', Komi ta mort-len vel-i-fJ kujim pi THIS MAN-ade IS-past-s3
THREE SON 'this man had threesons'. In Saamicand Fennic,the possessor
is markedwith a local caserather than the genitive; thus the equivalentsof
the Komi examplejust given are North Saamida-n olbma-s le-ddji golbma
bartni THIS-G MAN-Ioc IS-p3pastTHREE BOy'-G, Finnishtii-llii miehe-llii
ol-i-fJ kolme poika-a THIS-ade MAN-ade IS-past-s3THREE BOY-sP. In
Hungarian, suffixation of the dative becomesobligatory, e.g. a feerfi-fJ
haaromfi-a DEF.ART MAN-N BOY-s3 'the man'sthree sons' but afeerfi-
nak haaromfi-avol-t-fJ 'the man had threesons'.An overview of possessive
sentencesin Samoyedicin relation to nominal sentencesas a whole may be
gleanedfrom Katzschmann1986.
While the study of syntaxhas steeredmost theoreticallinguistics over the
last few decades,relatively little progresshasbeenmadein our understanding
of the Uralic sentence.Data-orientedsurveys are rare: Terescenko1973 is
limited to Samoyedic,Bartens1979to Mordva, Mari, andUdmurt, Wickman
1955 to direct-object marking, Rounds 1991 to Hungarian, Komi, and
Finnish; only Koizumi 1994 attempts a syntactic overview of the entire
family. The lag may be due, in part, to the enormity of the task of
understandingthe masses of morphophonologytypical of most Uralic
languages,some of which has been clarified only recently; but it is also
doubtlessdueto theAnglocentricity, thenEurocentricity,of much of the early
work in modemsyntactictheory.
The archetypalUralic sentencehasoften beencharacterizedas 'SOV', but
since subject and object noun phrases are regularly omitted in Uralic
sentencesit is probably more helpful to speak of '(T)FV', i.e. (optional)
Topic(s)-Focus-Verborder.The Topic, if present,may be single,like 'death'
in Hungarian (orthography) Ta haldl-fJ Flevele-t vhoz-fJ-fJ DEATH-N
34 INTRODUCTION

LETTER-A BRINGS-pres-s3'deathdeliversa letter', or 'net' in Izhma Komi

kulem-fs Ciistej 'the net is dry', or multiple, as in Ta jiatal zeneszerz;;
Torvosanaktanacsa ellenere Tszombatreggel FBecsbevindult ,Tthe young
composerTin spite of his doctor'sadviceTon Saturdaymorning v set off Ffor
Vienna'. Focusmay be shared;in this casethe usual scenariois to place one
of the focalized elementsto the right of the verb, e.g. F/gy v (rtok Fti 'Fthis is
how Fyou vwrite', Izhma Komi FCieri-fJ Vkij-fJ-enis Fkulem-en ,Ffish vthey-
catchFwith a net'.

Vocabulary
Words for core, yet non-culture-specific,conceptsacrosslanguagefamilies
are most often comparedin connectionwith attemptsto determine,by means
of various refinementsof glottochronologicalmethod,the relative degreeof
closenessof relationshipbetweenlanguagesand branches.No suchpurpose
underlies the pilot presentationhere, which aims merely to provide an
introduction to the sorts of distributional patternswhich occur. The patterns
are of two basictypes: retentionor innovation.
Retentionis a relative term: it can rangefrom total, e.g. Englishfather <
Old Englishfteder,to vestigial, e.g. the groom of bridegroom< guma 'man'.
Here we are interested more in the former extreme. We are also more
interested,for the purposesof this exercise,in the fact that the Hungarian
word for 'river' is foj=oo, i.e. historically the presentparticiple of a verbfoj-
'flows', than we are in the fact that the Hungarianreflex of the proto-Uralic
word for 'river', reconstructedas *juka, lives on in potamonymssuch as
Berettyo < *barek 'boggy grove alongside river' + *joo 'river'. As for
innovation: in core vocabulary such as we are sampling here, innovation
alwaysimplies replacementratherthannew discourse.We may recognizetwo
subtypes.An old word may be replacedwith a loanword, e.g. English table
(borrowed from French, and largely replacing Old English bord, but note
vestigial, metonymic,retentionin the phrase'bed and board') or it may be
edgedout by anotherword and/or meaningdrawn from the resourcesof the
language,e.g. English father-in-law replacing Old English sweeor, or the
expansionof the semanticsphereof bread « breead)at the expenseof that
of loaf« hlaaf).
Thereis no Uralic analogueto Buck (1949), which surveysthe formal and
semanticdevelopmentof some1,200Indo-Europeansynonyms.A greatdeal
of formal detail, however,on 43 Uralic synonymsmay be found in Veenker
1975.
As is to be expected,survival of both lexemeand sememeacrossthe entire
rangeof the family is rare. One suchis the word for 'name' (S =Saamic,Fe
=Fennic,Md =Mordva, Mr =Mari, Ud =Udmurt, Ko =Komi, Mn =Mansi,
Kh =Khanty, Hu =Hungarian,Ng =Nganasan,En =Enets,Ne =Nenets,Sl
=Selkup,Km =Kamassian,Mt =Mator):
 INTRODUCTION 35

S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
1 'name' a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a
with clear reflexesin all of the daughterlanguages,e.g. Ume Saaminamma
=Finnishnimi =ErzyaMordva IJem =Hill MariMari l~ =Udmurt njim =Sosva
Mansi nam = Hungarian neev = Nganasannjim = Kamassiannim. It is
possiblethat this word may havebeenborrowedinto proto-Uralic from some
form of proto-Indo-European,but its wide attestationthroughoutnorthern
Eurasiapoints to more general,earlier, diffusion. Other good examplesof a
similar wide distribution acrossUralic are the descendantsof proto-Uralic
*miksa 'liver' and *i1a '(space)underneath'.
The type of distributional patternshownby the words for 'hand' is fairly
common, and has traditionally been seen as one of the better pieces of
evidencein supportof the idea that Samoyedicbroke away from the rest of
Uralic early, if not fIrst:
S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
2 'hand' a a a a a a a a a a a a a a a a a a
b b b b b bb b b b b b
Thuswe havetwo setsof wordsfor 'hand':oneclearlyFinno-Ugric,e.g. Ume
Saamigiahta =Finnish kiisi (oblique stemkate-) =ErzyaMordva kedj =Hill
Mari kit (oblique stemkio;;J-) =Udmurt ki =SosvaMansi kaat =Hungarian
keez (oblique stem kiizii-), all from proto-Finno-Ugric *kati; and the other
Samoyedic,e.g. NganasantPiitii = Tundra Nenetsnguq = Taz Se1kuputi" =
Kamassianuda, from proto-Samoyedic*G)uta. Other good examplesof this
sort of distribution arethe words for 'stone'(proto-Finno-Ugric*kiwi, proto-
Samoyedic *paj), 'cloud' (pFU *pilwi, proto-Samoyedic*ti5) and the
numeralsfrom '3' to '6'.
Some Uralic etymologies point to innovation in a branch other than
Samoyedic,e.g. the new word for 'fIsh' in Permian:
S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
3 'fIsh' a a a aa a a a a a a a a a a a aa a a a a a a a a
b b

This word is 3b Udmurt Ciori"g =Komi Cieri, with possiblecognatesreferring

to specifIc speciesin Saamic and ObUgrian. Other Permian innovations
include the replacementof the pU word for 'bow', and of the proto-Finno-
Ugric word for 'leg'.
A different kind of breakin the distribution of an inheritedUralic lexeme
may be seenin the descendants of *witi 'water':
S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
4 'water' a c
b b b b b bb b b b b b b b b b b b bb b b b b b b
36 INTRODUCTION

Evenmorethantheword for 'name',theproto-Uralicword for 'water'looks as

if it might be a loan from some proto-Indo-Europeanidiom. Whateverthe
provenanceof this word, within Uralic all daughterlanguagesotherthanSaamic
and Khanty have reflexes with expectedform and with meaningintact, e.g.
Finnishvesi(obliquestemvete-)= ErzyaMordvavedj = Hill Mari W;1t (oblique
stem W;:x5;1-) =Udmurt vu =SosvaMansi wit =Nganasanbi? (oblique stem
bio;1-) =Kamassianbu. Judgingby its shapealone,the Saarnicword for water
(4a), as seenin e.g. Ume Saamitjaahtsee,probablycomesfrom somesort of
hyponymwhich referredto waterin oneof its moreforcible manifestations,e.g.
'flood' or '(sudden)thaw; freshet'; it has a possiblecognatein Khanty, viz.
Vasjugan Khanty sed '(1ate-)summer flooding; rise in water-level'. The
Khanty word for 'water' (4c), on the other hand,is a doubletto the inherited
Finno-Ugricword for 'ice',e.g.VasjuganKhantyjeIJk 'water' :jooIJk 'ice'.
It seemsmost likely that proto-Uralic had a word for 'tree' with initial *p-,
but at our presentstateof knowledgevery little else about this word may be
reconstructedwith certainty. The reflexes in Finno-Ugric point to a back
vowel in the first syllable, e.g. 5b Finnishpuu=Udmurtpu =Hungarianja;
the front vocalism of someof the Samoyedicpendants,e.g. ForestEnetspe,
is probably secondary,and points to lexically numerouscompoundsformed
with this root as secondmember.

S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
5 'tree' ac ac dde ac
bb bbb b b b b b b bb b b b b

Whateverits preciseoriginal shapemight have been,the proto-Uralic word

for 'tree' has beenreplacedin four separatelanguagezones.In Saarnic(e.g.
5a Ume Saarnimuarra) and Nganasan(5e mUIJku), the words for 'tree' are
of unclearorigin; theseare not thought to be connected,but could represent
different derivates of the same root. The Mordva word for 'tree', viz. 5c
(Erzya) cuvto =MokshasujtfJ, seemsto be a generalizationof a hyponym;its
only cognatesare in Fennic,e.g. Finnish huhta 'arableland won by felling of
heavy timber in a forest'. The ObUgrianlanguagesuse reflexesof *j'ix'i, viz.
5d SosvaMansi jiw, Vasjugan Khanty juuy; *jlxl also seemsoriginally to
havebeena hyponym,in this instancewith referenceto somesort of conifer,
e.g. TundraNenetsje,Taz Selkupcoo 'pine'.
By contrast, the replacementof the proto-Uralic word for 'fire' has a
focus:

S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
6 'fire' a a a a a a a a a a (a) a a a a a aa a a a a a
b
c c cc c c
d d
e
 INTRODUCTION 37

The proto-Uralicword for 'fire' , *tuli:, hasbeenreplacedin the corelanguages

exceptfor Udmurt (til); in Komi, only vestigesof this root remain,e.g. tiil in
tiil + kert '(pieceof) iron (kert) for startinga fire'. Komi has a new word for
'fire', viz. 6b hi, with uncertaincognates(UEW 359-60).The Mansi word 6e
uzia, known only in northernand easterndialects,is also of obscureorigin.
The Ugric word for 'fire', 6c Hungarian tiiiiz = Pelymka Mansi taawt =
Tremjugan Khanty tiiyw:Jt, is perhaps a taboo circumlocution (Abondolo
1996: 62). Taboo is probably also behind the ObUgrian use of one of the
Finno-Ugric words for' woman', *naj'i, as a cover-termfor 'fire': 6d Sosva
Mansi naaj = TremjuganKhanty niiiij; other meaningsof this word include,
beside 'sun', also 'beautiful/powerful woman; princess; queen (playing-
cards)';Hungariannadi 'big, great'is probablycognate(Meszaros1988).
In the mostcommonexpressionsfor 'white',

S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
7 'white' a ace h
bbb d g f f f f f f
f j j

Saamic and Fennic (7a) use derivatesfrom front and back isotopes of an
affective/imitativestem *wal=ke-/*wru=ke- meaning 'shines,is bright', viz.
North Saami vielgat, Finnish valkoinen, and words with related meanings
('bright', 'clear') seeminglybuilt from a stem *wal- are found in Mordva
and Mari. The Mordva and Mari words for 'white', however, are from
anothersource.They are 7b Erzya Mordva aso = Moksha Mordva aksfJ =
Hill Mari oS(:J) = Meadow Mari os(o), with possible, but not very
convincing, cognatesin either Fennic (cf. Estonianahka 'eiderduck',hahk
'grey') or Khanty (SouthernKhanty aas 'chalk, white clay'). The -k- of the
Mokshaform is thoughtto be the result of contaminationwith Turkic words
meaning 'coin, money', or 'whitish', e.g. Kirgiz aksa, from a root meaning
'white'. The Permianlanguagesdiverge from the foregoing and from each
other, with at leastthreedistinct words for white: Udmurt tedir, KomijedZia,
and Permiak BoBkem= Yaz'va Komi BoBkem= Izhma Komi BoBkem=
Vym Komi BoBkem.The last item, which means'clean' in standardKomi,
has a possible cognatein North Saami ceaskdt 'to appear(snow-)white';
the other two Permian words for 'white' are of obscureorigin, although
jedZia may be connectedwith a Permian word meaning 'raw; unripe (of
berries)' (Komiji!Z, UdmurtjeZ'). The origin of Hungarian7hfiiheer is also
unclear, as is that of the Khanty word seenin 7g SouthernKhanty niiw:J
'white' = Tremjugan neeyWi 'bright grey (e.g. fox)'; this word may be a
derivateof a verb root (proto-Khanty*nuuy- - *naay-) meaning'is (clearly)
visible'. In Mansi and Samoyedic,terms for 'white' have evolved from
words for 'ice', e.g. 7e Sosva Mansi jaa1Jk, 7i Tundra Nenets ser = Taz
Selkupseri"; this designationprobablycompetedwith and wholly or partially
38 INTRODUCTION

replaced a proto-Samoyedicsynonym, *jekV, with descendantsonly in

Nganasanand Selkup (7j). Another term for 'white', 7f sajrel], is attested
in non-northerndialectsof Mansi.
The Uralic words for 'black' are even more disparate;a superficial scan
turnedup sixteenitems:

S Fe Md Mr Ud Ko Mn Kh Hu Ng En Ne Sl KmMt
8 'black' a c ee j j k m 0 Ne Ne
b d g j j n
f h
p

The only cognateterms are (8e) Permian (Udmurt = Komi sed) and (8i),
HungarianJiikiitii = e.g. Tremjugan Khanty peyt;}. Neither set has clear
cognateselsewhere;the origin of Udmurt 8f ki"rSi is also unclear.The words
in Saamic (e.g. Northern Saami ccihppat) are all derivates of a common
Saamicroot whose shapesuggestsaffect; Finnish musta, too, representsa
pan-Fennic formation to a root *muse-/*muhe- with affective (animal
husbandry) undertones.The Mordva word, e.g. Erzya rav(u)zo, lacks a
credible etymology; its initial r- looks foreign. Mari sem has Permian
cognatesmeaning 'rust(s)', with formal difficulties (Udmurt si"nem, Komi
sim); for the semantics,compareErzyaMordva cemeni 'rust' andits possible
Fennic cognates,e.g. Finnish hiimii=rii 'dark, unclear' (Keresztes 1986:
156). Mansi dictionaries report at least three words for 'black': one (8g
sior;,s) is obscure,another(8h seem;}l) seemsto have been built from the
word for 'rust', borrowedfrom Komi; a third (8p pit), known from northern
dialects only, is a loan from Khanty (cf. 8i). The range of etymologically
unconnectedwords across Samoyedic is unusual; of eighty-seven core
sememesexamined by Helimski (1982: 129-33) only two ('black' and
'new') show such a spread,while twenty-nine (i.e. one-third) have uniform
representationsin all six Samoyedicidioms. The Samoyedicwords include
internalderivatessuchas 8j Nganasanhenkoand 81 TundraNenetspr/Jridyana
(the presentparticiple of a stative verb meaning'is black'), and a loan from
Turkic (8n Kamassiansaayar). The sourceof Selkup 8m siiiiq is obscure;
the Mator item (80) was probably a descendantof proto-Samoyedic*ktint;)
'smoke'. For a recent discussionof the relative chronology of the various
types of Turkic in Samoyediclexica seeHelimski 1987, Janhunen1989, and
Helimski 1991.

Referencesand Further Reading

Abondolo, D. (1990) 'Proto-Uralic without s-Laute: a distributional view', C71FU,
vol. III, pp. 80-85.
- - (1996) Vowel Rotation in Uralic: Obug[rJocentric evidence,SSEESOcca-
sionalPapersno. 31, London: University of London.
 INTRODUCTION 39

Austerlitz, R (1968) 'L' ouralien',in A. Martinet. (ed.), Le Langage:Encyclopediede

la Pleiade,Paris: Gallimard,pp. 1331-89.
- - - (1976) 'Az eur6pai[w] es [v] terbenesidBben',NyelvtudomdnyiKozlemenyek
7612: 250-55 [= LazicziusFestschrift,ed. P. Hajdu].
- - - (1978) 'On comparing Uralic with other languagefamilies', 455 Finno-
ugorskienarody i vostok- Trudy po vostokovedeniii(Oriental Studies),vol. IV,
Tartu Riikliku Olikooli Toimetised = Acta et CommentationesUniversitatis
Tartuensispp. 119-30.
- - (1980) 'Language-familydensity in North America and Eurasia',UAJb 52:
1-10.
- - - (1985) 'The Permian centre', in COBeTcKoe <l>HHHo-YrpoBej:\eHHe 21 =
C61FU, Texteder Plenarsitzungsvortrage, pp. 99-109.
Barbera,M. (1993)La gradazionebaltofinnica,London: Lothian.
Balazs,J. (ed.) (1983)Aredlis nyelveszetitanulmdnyok,Budapest:Tankonyvkiad6.
Bartens, R (1979) Mordvan, tseremissinja votjakin konjugaation infiniittisten
muotojensyntaksi,MSFOu 170 Helsinki: SocieteFinno-Ougrienne.
Beekes, RS.P. (1995) Comparative Indo-European Linguistics. An Introduction,
Amsterdam:Benjamins.
Bereczki,G. (1983) 'A Volga-Kama-videknyelveinekareiilis kapcsolatai',in Balazs
1983,207-36.
Buck, C.D. (1949)A Dictionary ofSelectedSynonymsin the Principal Indo-European
Languages: A Contribution to the History of Ideas, Chicago: University of
ChicagoPress.
Castren, M.A. (1854) Grammatik der samojedischenSprachen, St Petersburg.
Bloomington:IndianaUniversity, Reissued1966as no. 53 of the Uralic andAltaic
Series.
Comrie, B. (l980a) 'The genetic affiliation of Kamchadal: some morphological
evidence',InternationalReviewofSlavicLinguistics5: 109-20.
- - - (l980b) 'The order of caseand possessivesuffixes in Uralic languages:an
approachto the comparative-historicalproblem',Lingua Posnaniensis23: 81-6.
- - - (1981a) 'Negationandotherverb categoriesin the Uralic languages', C51FU,
vol. VI, pp. 350-55.
- - (1981b) The Languages of the Soviet Union, Cambridge: Cambridge
University Press.
- - (1988) 'General Featuresof the Uralic Languages',pp. 451-77 in Sinor
1988.
Decsy, Gy. (1965) Einfiihrung in die finnisch-ugrischeSprachwissenschaJt, Wies-
baden:Harrassowitz.
- - (1969) 'FinnougrischeLautforschung',UAJb 41: 33-75.
Doerfer, G. (1984) 'Prolegomenazu einer Untersuchungder dem Tungusischenund
Mongolischengemeinsamen Worter', JSFOu79: 68-85.
Donegan,P. (1978) On the Natural Phonology of Vowels, PhD diss., Ohio State
University, published1985 in the seriesOutstandingDissertationsin Linguistics,
ed. JorgeHankamer,New York andLondon: GarlandPublishing.
Donner, O. (1879) Die gegenseitigeVerwandtschaftder Finnisch-ugrischenSpra-
chen,Helsinki.
Gulya, J. (1975) 'Gabes einefinnisch-ugrischeEinheit?' C31FU, vol. I, pp. 87-92.
Hajdu P. (1990) 'Einigestiber Ftirworter', Linguistica Uralica 1: 1-12.
- - - (1992) Introduzioneaile lingue uraliche, translatedand adaptedby Danilo
Gheno,Turin: Rosenberg& Seiler.
Hajdu P. and DomokosP. (1987) Die uralischenSprachenund Literaturen,Budapest:
Akademiaikiad6.
40 INTRODUCTION

Harms,R.T. (1977) 'The Uralo-Yukaghirfocus system:a problemin remotegenetic

relationship',in P. Hopper(ed.), Studiesin Descriptiveand Historical Linguistics
(Festschriftfor Winifred P. Lehman),Amsterdam:Benjamins,pp. 301-16.
Helimski, E.A. (1982) )J,peBHeHIIIHe BeHrepcKO-CaMOIJ;HHCKHe H3bIKOBhIe
llapaJIJIeJIH,Moscow: Nauka.
- - - (1987) 'Two Mator-Taigi-Karagass vocabulariesfrom the 18th century',
JSFOu81: 49-132.
- - - (1991) 'On the interactionof Mator with Turkic, Mongolic, and Tungusic: a
rejoinder',JSFOu83: 257--67.
- - - (1996) 'Proto-Uralic gradiation [recte: gradation]: continuationand traces',
C8IFUvol.I, pp. 17-51.
Honti, L. (1984) Chrestomathiaostiacica (Osztjak nyelvjarasi szoveggyujtemeny
nyelvtanivazlattalestorteneti magyarazatokkal),Budapest:Tankonyvkiad6.
- - (1988) 'Die ob-ugrischeSprachen',in D. Sinor (ed.) The Uralic Languages:
Description, History and Foreign influences, Handbuch der Orientalistik 8/1,
Leiden: Brill.
- - (1990) 'Ugrisches',Linguistica Uralica 16: 298-9.
- - - (1992) 'Adalek a magyar1- finn t megfelelesenek es alapnyelvieiOzmenye-
nek magyarazatahoz,' in P. Dereky et al. (eds), Festschriftfur Karoly Redeizum
60. Geburtstag, Vienna-Budapest:Finno-Ugric Institute of the University of
Vienna-Departmentof Finno-Ugric, University of Budapest,Linguistics Institute
of the HungarianAcademyof Sciences,pp. 209-13.
- - (1995) 'Zur Morphotaktik und Morphosyntax der uralischenlfinnisch-
ugrischenGrundsprache',C8IFU vol. I, pp. 53-82.
Itkonen, E. (1939) Der ostlappischeVokalismusyom qualitativenStandpunktaus mit
besondererBerucksichtigungdes Inari- und Skoltlappischen,MSFOu 79, Hel-
sinki: SocieteFinno-Ougrienne.
Janhunen,J. (1977) SamojedischerWortschatz.Gemeinsamojedische Etymologien,
Castrenianumintoimitteita 17, Helsinki: Castrenianum.
- - (1981a) 'On the structureofProto-Uralic',FUF 44: 23-42.
- - (1981b) 'Uralilaisenkantakielensanastosta,'JSFOu77/9: 219-74.
- - (1984)'Altailaisenhypoteesinnykytila', Virittiijii 88: 202-7.
- - - (1989) 'On the interaction of Mator with Turkic, Mongolic, and Tungusic',
JSFOu82: 287-97.
- - (1992) 'Uralic languages',W. Bright (ed.), International Encyclopediaof
Linguistics,vol. IV, New York-Oxford: Oxford University Press,pp. 205-10.
- - - (1993) 'Options for Tundra Nenetsvowel analysis',in M. Bakr6-Nagyand
E. Szij (eds), Hajdu Peter 70 eves, [= Linguistica Series A, Studia et Dis-
sertationes,15], Budapest:Linguistics Institute of the Hungarian Academy of
Sciences,pp. 143-7.
Joki, A. (1952) Die Lehnworter des Sajansamojedischen, MSFOu 103, Helsinki:
Societe Finno-Ougrienne.
Kalman,B. (1966)Nyelvjarasaink,Budapest:Tankonyvkiad6.
Katz, H. (1975) Generative Phonologie und phonologische Sprachbunde des
Ostjakischenund Samojedischen, Finnisch-UgrischeBibliothek, Munich: Wilhelm
Fink.
- - (1984) 'SelkupischePhonologie',in P. Hajdu and L. Honti (eds),Studienzur
phonologischenBeschreibunguralischer Sprachen,Budapest:Akademiai kiad6,
pp.33-46.
Katzschmann,M. (1986) Nominal- und Esse-Satzin den samojedischenSprachen,
Fenno-Ugrica,Hamburg:Buske.
Kazancev,D.E. (1990) 'K BOIIPOCY 0 peKOHCTpYKL\HH *() H *()i B <l>HHHO-yropCKOM
 INTRODUCTION 41

rrpaSI3hIKe',Linguistica Uralica 26: 180-87.

Keresztes, L. (1986) Geschichte des mordwinischen Konsonantismus,vol. II:
EtymologischesBelegmaterial,Studiauralo-altaica27, Szeged.
Koizumi, T. (1989) 'On the deictic function of Uralic demonstratives',Uralica
8:1-16.
- - - (1994) Uraru-go toogoron, Tokyo: Daigaku Shorin (with English summary
pp. 317-57).
Korhonen, M. (1981) lohdatus lapin kielen historiaan, Helsinki: Suomalaisen
Kirjallisuuden Seura.
- - - (1984) 'Zur zentralenProblematikderTerlappischenPhonologie', in P. Hajdu
and L. Honti (eds), Studien zur phonologischenBeschreibunguralischer Spra-
chen,Budapest:Akademiaikiad6, pp. 311-25.
Laajavaara,M. (1986) Itiimerensuomendemonstratiivit vol. I, Helsinki: Suomen
kirjallisuudenseura.
Laver, J. (1994) Principles of Phonetics,Cambridge:CambridgeUniversity Press.
Lehtiranta, J. (1989) Yhteissaamelainensanasto, MSFOu 200, Helsinki: Societe
Finno-Ougrienne.
- - - (1992) Arjeploginsaameniiiinne- ja taivutusopinpiiiipiirteet, MSFOu 212,
Helsinki: SocieteFinno-Ougrienne.
Maddieson,I. (1984) PatternsofSounds,Cambridge:CambridgeUniversity Press.
Majtinskaa K.E. (1979) I1CToPllKo-corrocTaBllTeJIhHaSI MopcpOJIOrllSI
<l>llHHo-yropcKllx SI3hIKOB, Moscow: Nauka.
Meszaros,H.E (1988) 'Nagy szavunkeredeterol',in P. DomokosandJ. Pusztay(eds),
BereczkiemLekkonyv(BereczkiGabor 60. sZiiletesnapjara),Budapest:Faculty of
Philosophyof the University of Budapest,pp. 277-82.
Mikola, T. (1984) 'Einige Problemeder enzischenPhonologie',in P. Hajdu and L.
Honti (eds), Studien zur phonologischenBeschreibunguralischer Sprachen,
Budapest:Akademiaikiad6, pp. 29-32.
Miller, A. R. (1990) 'How deadis the Altaic hypothesis?',in B. Brendemoen(ed.),
Altaica osloensia,Oslo: pp. 223-37.
Nichols, J. (1992) Linguistic Diversity in Space and Time, Chicago and London:
University of ChicagoPress.
Nikolaeva, I. A. (1988) 'IIp06JIeMa ypaJIO-IOKampcKllx reHeTll'-IeCKllX cBSI3eil.,'
ABTopecpepaT: ABTopecpepaT:
}:\llccepTa~llSI Ha COllCKaHlle Y'IeHoil. CTerreHll KaH}:\ll}:\aTa
<l>llJIOJIOm'IeCKllX HaYK, Moscow: Linguistics Institute, USSR Academy of
Sciences.
Redei,K. (1988) 'Geschichteder permischenSprachen',pp. 351-94in Sinor 1988.
Ristinen,E. K. (1960) 'Samoyedphonemicsystems',unpublishedPhD diss., Indiana
University.
Rounds,C. (1991) 'Topicality and casemarking in Hungarian,Komi and Finnish',
unpublishedPhD dissertation,ColumbiaUniversity.
Salminen,T. (1993) 'Uralilaiset kielet maailman kielten joukossa',in T. Salminen
(ed.), Uralilaiset kielet tiiniiiin, Snellman-InstituuttiSeriesA, no. 13 (with English
summarypages103-105),Kuopio: Snellman-Instituutti,pp. 24-30.
Sammallahti,P. (1974) Material from Forest Nenets,Castrenianumintoimitteita 2
Helsinki: Castrenianum.
- - - (1978) 'Uber die Laut- und Morphemstrukturder uralischenGrundsprache',
FUF 43: 22-66.
- - - (1984) 'Saamelaistenesihistoriallinen tausta kielitieteen valossa' [with
English summary] in J. Gallen (ed.), Suomenviieston esihistorialliset juuret,
Helsinki: SuomenTiedeseura,pp. 137-56.
- - - (1988) 'Historical phonologyof the Uralic languages,with specialreference
42 INTRODUCTION

to Samoyed,Ugric, and Permic,' pp. 478-554in Sinor 1988.

Serebrennikov, B. A. (1963) llcTOpHQecKaH MOP<POJIOrHH IIepMCKHX H3hIKOB,
Moscow: Publicationsof the SovietAcademyof Sciences.
Sinor, D. (ed.) (1988) The Uralic Languages.Description, History and Foreign
Influences,Handbuchder Orientalistik 8/1, Leiden: Brill.
Steinitz, W. (1944) Geschichtedes finnisch-ugrischenVokalismus,Acta Instituti
Hungarici Universitatis Holmiensis, Series B. Linguistica 2, Stockholm: Uni-
versity of Stockholm.
- - - (1952) Geschichtedes finnisch-ugrischenKonsonantismus,Acta Instituti
Hungarici Universitatis Holmiensis, Series B, Linguistica 1, Stockholm: Uni-
versity of Stockholm,pp. 15-39.
Suhonen,S. (1988) 'Geschichteder ostseefinnischen Sprachen',pp. 288-313in Sinor
1988.
Suihkonen,P. (1987) 'SuomensukuisetkansatvaestOtilastojenvalossa',Terra 99/4:
209-32.
Talos, E. (1983) 'Kep sZQveg nelktil', in Uralisztikai tanulmanyok(Hajdu peter 60.
sziiletesnapjatiszteletere),ed. G. Bereczki and P. Domokos, Budapest:ELTE,
pp.409-20.
- - - (1987) 'On the vowels of proto-Uralic', in K. Redei (ed.), Studien zur
Phonologie und Morphonologie der uralischen Sprachen, Akten der dritten
Tagungfor uralische Phonologie, Eisenstadt,28. Juni-l. Juli 1984, [= Studia
Uralica 4], Vienna: Verband der wissenschaftlichenGesellschaftenOsterreichs,
pp.70-80.
Terescenko,N. M. (1966) '3HeI.\KHH H3hIK', in V. I. Lytkin et al. (eds), H3hIKH
Hapo~oBHapo~oBcccp, vol. III: <PHHHo-yropcKHeH caMo~HHcKHe
caMo~HHcKHeH3hIKH Moscow: Nauka,
pp.438-57.
- - - (1973) CHHTaKCHC caMo~HHcKHX caMo~HHcKHX H3hIKOB. IIpocToe IIpe~JIo)J{eHHe, IIpe~JIo)J{eHHe,
Leningrad:Nauka.
Vasikova,L. P. (1992) 'ACIIHpaTa 7, a<p<PPHKaTaIJk H COHaHT IJ B ropHoMapHHcKoM
H3hIKe (K rrpoB6JIeMeKOHCOHaHTH3MaB MapHHcKHX H3hIKax', Linguistica Uralica
228: 120-5.
Veenker, W. (1975) Materialien zu einem onomasiologisch-semasiologischen ver-
gleichendenWorterbuch der uralischen Sprachen,HamburgerUralistischeFor-
schungenI, Hamburg.
Viitso, T.-R. (1996) 'On classifying the Finno-Ugric languages',C81FU, vol. IV,
261-6.
Wickman, B. (1955) The Form of the Object in the Uralic Languages,Uppsala:
Almqvist & Wiksells.
2 Saamic
Pekka Sammallahti

The Saamilanguagesare spokenin an areastretchingfrom Dalarnain central

Swedento the tip of the Kola Peninsulain Russia.All the Saarnilanguagesare
fairly similar in structureand basic vocabulary.Although there are no wide
linguistic boundaries,one can distinguishten Saarnilanguageswhich differ
from oneanotherat leastto the samedegreeasthevariousGermaniclanguages.
Peripheraldialectswhich lie on two different sidesof a languageboundaryare
normally close to one another in vocabulary, and this is why the Saami
languagesform a chainin which speakersof adjacentdialectsunderstandone
anotherrathereasily. The more central dialectsof adjacentSaarnilanguages
differ enoughfor mutualcomprehension to requirea fair amountof effort.
In the literature,the variantsof Saarnihavebeentreatedasdialectsbecause
of the regular correspondencesin phonology and the similarity in basic
vocabularyand grammar.Since six of the regional variants have independ-
ently standardizedwritten forms, it is more justifiable to speakof separate
languages.The remainingfour varieties of Saami,which are generallyheld
to fall outside the six main languages,are spoken by only a few older
individuals.
It is customaryto distinguish the following main varieties of Saarni: (1)
SouthSaarni,(2) OrneSaarni,(3) Pite Saarni,(4) Lule Saarni,(5) North Saarni
(also called NorwegianSaami),(6) Inari Saarni,(7) Skolt Saarni, (8)Akkala
Saarni (also called Babino Saami), (9) Kildin Saarni, and (10) Ter Saarni.
Their territories are indicatedon Map 2.1. The majority, as many as 75 per
cent, speak North Saarni and most of these, some 10,000 people, live in
Norway (about5,000live in Swedenand about2,000in Finland). Orne, Pite,
Akkala, and probablyalsoTer Saarniare usedmainly by older people.South,
Inari, and Skolt Saami haveabout 300-500 speakerseach. Kildin Saamiis
spokenby about 1,000 speakers,and the number of Lule speakersis about
2,000-3,000.The total numberof Saarnispeakersis probablysomewhatmore
than 20,000,i.e. in Finland 3,000; in Norway 12,000;in Russia1,000; andin
Sweden7,000.All of thesefigures areestimates.
The orthographiesof the standardizedwritten languagesare basedon the
Latin alphabetin South,Lule, North, Inari, and Skolt Saarni,whereasKildin
now usesa variant of Cyrillic insteadof the Latin-basedalphabetusedin the
1930s.South Saamiorthographyrepresentssoundsin a mannercloserto the

43
44 SAAMIC

orthographiesof Swedishand Norwegianand so doesthat of Lule Saami,but

to a lesserextent.The basisfor North, Inari, andSkolt Saamiorthographywas
laid by the Danish linguist RasmusRask at the beginning of the nineteenth
century.As an amendment,in Inari and Skolt Saamidouble letters are used
to indicatelong vowels. Thereare also someadditionallettersfor distinctive
vowel andconsonantqualities.
The Saamilanguagesaremostclosely relatedto Fennic,a branchof Uralic
of which Finnish hasbeenSaami'sneighbouringlanguagesincethe breakup
of the Fennic-Saamicprotolanguage(also called Early Pre-Finnic) which
beganafter the introduction of the Indo-EuropeanBattle-axeCulture on the
coastsof Finlandabout2500BeE.
Saamidiffers from Fennicin a numberof features.In phonology,the vowel
systemhasundergonea radicalreorganizationwhereasthe consonantsystem
hasretainedmany older features,suchas palatalizedconsonants,which were
lost in Fennic as a consequenceof the Indo-Europeancontacts which
ultimately helped bring about the disintegration of the Fennic-Saamic
protolanguage.In morphology, Saami has retained the dual number in its
personal pronouns, personal endings, and possessivesuffixes, i.e. in all
elementswith the categoryperson.Furthermore,Saamipersonalendingsare
different in the presentand past tenses,and Saami lacks the external local
caseswhich Fennic developedafter the disintegrationof the protolanguage.
In Saamiconjugation,most of the present-tense personalendingsare based
on earlier nominal derivational suffixes, a fact which suggeststhat the
present-daysymmetry of the Finnish personalendingsdevelopedin proto-
Fennic,after the split from proto-Saamic.Saamideclensionseemsto indicate
that the plural local cases(illative, inessive,and elative) developedafter the
disintegration of the Fennic-Saamicprotolanguage,and that the proto-
languageprobably had only the grammaticaland generallocal cases(essive
and partitive) in the plural. Both branchesof Fennic-Saamichave developed
consonantgradation,but in Saamithis morphonologicalphenomenongrew
out of fortition processes(*kota 'hut sN' > North Saamigoahti, *kota-n 'sG'
> goatli) whereasin Fennic it is the result of weakening(*akka 'wife sN' >
Finnish akka, *akka-n 'sG' > akan). A further difference is that Saami
gradation entails all Saami consonants,whereasin Fennic only the stops
alternate.The reorganizationof the Saamicvowel systemmay be illustrated
by the following examples (all reconstructed forms are p[roto-]
F[ennic-]S[aamic]): *ajma 'needle' > tiibmi, *kati 'hand' > giehta, *appi
'father-in-law' > vuohppa, *kala 'fish' > guolli (*a sometimes>a, as in
*vanca-'walks' > vtizzi-), *poIJi 'bosom'> buokIJa(*0 sometimesis retained,
as in *joki 'river' > johka), *kota 'hut' > goahti, *n6ri 'young' > nuorra,
*meni- 'goes' > manna- (*e sometimes>ie as in *velji > vieilja), *kesa
'summer'> geassi,*keli 'language'> giella, *puri- 'bites' > borra-, *kiili-
'hears' > gulla-, *siili 'lap' > salla, *kilpi 'shielding object' > gaiba, *plri
'circle' > birra 'around'.
 SAAMIC 45

Western and Eastern Saami

The Saami languagesmay be divided into two main groups: (1) Western
Saamilanguages(South, Ume, Pite, Lule, and North Saami)and (2) Eastern
Saamilanguages(lnari, Skolt, Akkala, Kildin, andTer Saami).
Thesetwo main groupsare distinguishedfrom one anotherby the fate of
the proto-Saamiconsonantclusters*sjt and *sjk: in the Westernlanguagesthe
*sj has split into the consonantsequenceIjh/, but not in the Easternones:
'(river) rapids' in SouthSaamiis goejhke,Ume guajhkke,Pite guajhhka,Lule
guojkka,North Saamiguoika (all with the clusterIjhk/), but lnari haskuoskd,
Skolt andAkkala havekuoskk,Kildin haskuussk(orthographically:KYIIIlIIK),
and Ter Saamikyyssk(all with the clusterIsk/). The easterncoastaldialects
of North Saamistill hadsk andst in the eighteenthcentury.
Another Eastern feature with traces in the west is nz-contraction,
productive especially in diminutives: North Saami lottds 'little bird', pN
lottdzatv. Inari Saamiloddddz,pN loddddh(with contracteddd in the second
syllable).The introductionof so-calledprotheticstopsin nasalgeminatesalso
largely coincideswith this border (the exceptionis SeaSaami, a dialect of
North Saami): in the west, original *mm, *nn, *njn j , *IJIJ have changedinto
clustersbeginningwith a homotopic stop, e.g. North Saamieadni 'mother',
sG eatni v. lnari Saamienni, sG eeni (/eeni/), North Saamibatni 'tooth', sG
bani v. lnari Saami piiiini (/paani/), sG piiiini (- piine). Sea Saami has
preservedthe geminates:jien 'ne " jienne, bdnne" bdne.
WesternSaami
The WesternSaamilanguagesmay be further subdividedinto two groups:the
southerngroup (South and Ume Saami) and the northerngroup (Pite, Lu1e,
and North Saami). In the southerngroup, final soundshave usually been
preservedin unstressedsyllables,whereasthe northerngroup has lost them:
South gaetesne'in the hut', Ume gaatiesnev. Pite gaatien, Lule goaden,
North goailis. As a consequence of consonantgeminationafter short stressed
vowels, the southerngroup doesnot have short stressedsyllables: Southand
Umejohkesne'in the river' v. Pite and Lulejagan,Northjogas.Furthermore,
short stressed*i and *u havegoneto the sequences lijl andluvl beforesingle
consonantsin the southerngroup: North, Lule, and Pite juhkat 'to drink' v.
Umejovhkkedh,Southjovhkedh(*u > luv/); North gidtla 'spring',Lule gidd
v. Ume gijdtle, Southgijre (*i > lijl).
The SouthernGroup
The southerngroup of WesternSaamiconsistsof two languages:Ume Saami
(in Sweden:the districts Stille and Maskaurein Arjeplog, Mala, Gran, and
Ran in Sorsele,Umby in Tiima; Rana in Norway) and South Saami (all
variantssouthof Ume Saamidown to ldre in the SwedishprovinceDalarna).
=
Ume Saami has partial consonantgradation: gaahtie 'hut' (a 10/), sG
gaatien 'sG', gaahtaje 'sIll', guutijne 'pIne', whereasSouth Saami com-
pletely lacks gradation: gaetie 'hut', gaetien 'sG', gaatan 'sIll', gaetine -
46 SAAMIC

gootine 'pIne'. Furthermore,second-syllablelong vowels havebeenreduced

to short vowels in trisyllabic stressgroups in South Saami, whereasUme
Saamihasretainedthem: comparethe underlinedsegmentsof Southgaet~sne gaet~sne
v. Ume gaati§.sne'hut sIne', SouthbearkQsnev. Ume bierkoesne'meatsIne',
Southmaan{!,.ste
maan~ste v. Ume maan{!,.ste'child sEla' (second-syllablea = laal).
South Saami has two main dialects: the northern (or Asele) dialect (in
Sweden: Vapstenin Tlirna, Vilhelrnina and Frostviken; in Norway: Vefsn,
Grane, Hattfjelldal, Bindal, and Namdal), and the southern(or Jfundand)
dialect (in Sweden: Hotagen, Offerdal, Kall, Skalstugan,Undersak.er,and
Harjedalen;in Norway: Merak.erandthe areabetweenSnasaandVerdal). The
northern second-syllablediphthong oe (= luol « *a) correspondsto
southerna (= laal): Asele aahkoe v. Jfundandaahka 'grandmother'.The
infInitive suffIx is -dh in the north and -}h in the south: Asele Metedh v.
Jfuntland Meti}h 'to come'. Final -m has been retained in the south but
changedto -b in the north: Asele Matab, guelieb v. Jfundand Matam 'I
come',gueliem'fIsh sA'.

The Northern Group

The northern group of WesternSaami is divided into two branches,North
Saami(Tome,Finnmark,and SeaSaami)and the westernsubgroup(Lule and
Pite Saami).
In North Saamithe labial stopin the clusters*BD, *Bzi, *pc, *pcj , *ps, *psj,
and *pt has become velar (G, K; in Finnmark and Sea Saami *G has
subsequentlygoneto a fricative v): North dovdat(Tome dogdah- dogduoh)
'to recognize'v. Lule dabddat; North gokcat 'to cover' v. Lule gabttjat (-
gapttjat). There are some conspicuousmorphological differencesbetween
North Saami and the western subgroup.The North Saami locative corre-
spondsto westerninessiveandelative: Pite gaatien,Lule goaden'in the hut',
Pite gaatiest, Lule goades 'from the hut' v. North goatlis 'in/from the hut'.
In the westernsubgroup,the negativeverb hasretainedthe old present-tense
and past-tenseforms, whereasNorth Saami uses the original present-tense
forms in bothfunctions:Pite ij Matie andLule ij boade'(s)hedoesnot come',
Pite Utji) Matie and Lule ittjij boade '(s)he didnot come' v. North ii boatle
'doesnot come' and ii boahtan'did not come'.Furthermore,the genitive and
accusativesingularforms have syncretizedin North Saami,but the western
subgrouphas retainedthe distinction: Pite gaatie 'hut sG', gaatiev 'hut sA',
Lule goade 'hut sG', goadev'hut sA' v. North goatli 'hut sG/A'. In all these
features,North Saamipatternswith the EasternSaamilanguages,while Lule
andPite agreewith the South.
North Saami is divided into the following main dialects: Tome Saami,
FinnmarkSaamiand SeaSaami(the northerncoastfrom FisherPensinsulato
Troms,exceptthe PorsangerFjord).
Tome Saami has second-syllable short lui « proto-Saami *u) in
third-person singular present-tense forms of u-verbs, with ensuing
 SAAMIC 47

monophthongizationin the ftrst syllable, e.g. cirro '(s)he cries' (phono-

logically leiirrul - leiirro/; elsewherewith a long vowel from earlier *tui:
cierru, phonologicallyleierruul). Tome Saarniagreesherewith the languages
further southwest. The locative singular ending is -n in Tome, whereas
Finnmarkand SeaSaarnihave -set): Tome gielan 'in the snare'v. Finnmarkl
SeaSaarni gielas. SeaSaarni has retainedold nasal geminateswhich have
goneto clustersin Tome and Finnmark: Seabanni - banne 'tooth', biem'mu
- biem'mo 'food', and jielJlJa 'ice' v. Finnmark and Tome batni, biebmu,
jieklJa. SeaSaarniagreesherewith the Saarnilanguages further to the east.
Tome Saarni has four subdialects:the Finnish Wedge dialect (Western
Enontekio in Finland and adjacentareasin the west betweenSkibotnelva,
GaIggojavri and Nordreisain Norway); the Karesuandodialect (Konkama-
vuoma and Lainiovuoma districts in Sweden, Lyngen and Balsfjord in
Norway); the Jukkasjiirvi dialect (the districts Saarivuoma,Talma, Rau-
tasvuoma,and Kaalasvuomain Sweden,and the areasaroundVagsfjord and
Ofotfjord in Norway); and the Kaitum dialect (the districts Norrkaitum and
Mellanbynin Sweden).
In Jukkasjarviand Kaitum, the plural locative ending is -s insteadof the
-n in the rest of North Saarni: JukkasjiirviIKaitum dais variis 'in these
mountains' v. KaresuandolFinnishWedge dain variin. In Karesuando,the
monophthongalvariants ee and 00 of proto-Saarni*ea and *oa have been
rediphthongizedinto ie and uo: Karesuandobuohten'I came' and giessen'I
pulled' v. boohtenand geessenin the rest of North Saarni.Correspondingto
a in Kaitum, Jukkasjiirvi and Karesuandohave e before following j and v:
geikuot'to tear' v. KaitumlFinnishWedgegaikut. Kaitum andFinnishWedge
have Iii! and luul in the secondsyllable correspondingto standard-language
i and u, whereasJukkasjiirvi and Karesuandohave liel and luo/: Kaitum!
FinnishWedgeaski 'lap' and viessu'house'v. JukkasjiirviIKaresuandoaskie
andviessuo.The Kaitum dialecthasbeenregardedas a dialect of Lule Saarni
becauseof lexical similarities, but structurally it belongs to North Saami.
Furthermore,the Finnish Wedgedialect agreeswith WestFinnmarkSaarniin
that proto-NorthSaarni*GD and *GZj (from proto-Saarni*mn and *mzj) have
gone to Ivt/ and Ive/, respectively: Karesuandodogduoh 'to recognize',
vuogdieh'to sell', ltigiie 'strap',Jukkasjiirvi dogdat,vuogdiet,lagZie, Kaitum
dogdat, vuogdit, lagZi v. Finnish Wedgedovdah,vuovdih, lavii (and standard
dovdat, vuovdit, ltivii). Jukkasjiirvi, Karesuandoand Finnish Wedge have
generallylost the oppositionbetweenIcl andleI andbetweenthe correspond-
ing voiced affricates /zl and IZI, e.g. Karesuandoloaz'zuohl 'to obtain',
Ivaaz'ziehl'to walk', Finnish Wedgeloai'zuuhl, Ivaai'ziihl, standardoaiiut,
vazzit.
Finnmark Saarni has two dialect groups: western dialects (Eastern
Enontekio,NorthernSodankyla,and part of Inari in Finland; Kautokeinoand
Alta in Norway), andeasterndialects(Utsjoki andpart ofInari in Finland,and
Karasjok,Porsanger,andTanain Norway).
48 SAAMIC

Correspondingto westernintervocalic b and g, the easterndialects have

fricatives or zero: western stobus 'in the house', logan 'I read' v. eastern
stovus (- sto'us), loyan (- 10' an). In the easterndialects, stressedsingle
vowels have beenlengthenedbefore relatively short consonants:IjuhkenJ 'I
drank' v. IjuuhkenJ'I dealt' in the west,but IjuuhkenJfor both in the east.The
western dialects have a phonological opposition betweenlong and short
geminates,while the easternones have transferredthe difference onto the
vowels: western!kol'lii! 'gold', !kollii! 'gold sG' v. eastern!kollii! : !koollii!
(- !kollii:1 with secondarystress on the secondsyllable; in some eastern
idioms !kollii! for both).
The westernsubdivisionof the northerngroup of WesternSaamiis divided
into two languages:Lule SaamiandPite Saami.
In Pite Saami, the extensivemetaphonicalternationsin stressedvowels
dependmainly on the secondsyllable vowel; only Ii! and lui (and, in some
dialects,laa!) do not participatein metaphonicalternations.In Lule Saami,
stressedvowel alternationsare dependenton the quantity of the following
consonantcentreand the following vowel, and only the low diphthongstake
part. In Pite Saami,stresseda hasgoneto i, anda(= 10/) hasgoneto u before
i or u in the next syllable: mannat 'to go' v. minniv 'I went', harrat 'to eat'
v. burriv 'I ate'. In Lule Saami,the clusterskt, ktj, ks and ksj (tj =lei, sj =lsi)
participate in qualitative gradation, whereasin Pite they have quantitative
gradationonly: Lule tjakta 'autumnsN', tjavtav 'sA' v. Pite tjakttja : tjaktjav.
The Lule Saami qualitative alternationsbb:pp, dd:tt, gg:kk, dts:tts, dtj:ttj,
bbm:bm (= /b'm/:/pm/), ddn:dn, ggn':gn' (= Ig'IJ/:!kIJ/) correspondto Pite
quantitativealternationsp'p:pp, t't:tt, k'k:kk, t'ts:tts, t'tj:ttj, p'm:pm, t'n:tn,
k'n':kn': Lule gadde 'shoresN' : gattev 'sA', biebbmo'food sN' : biebmov
'sA' v. Pite gaat'tie: gaattiev,biep'muo:biepmuov.
Lule Saami is divided into the following dialects: northern (Sorkaitum,
Sirkas, and lillakaska districts in Sweden,Tysfjord in Norway), southern
(Tuorpon district in Sweden),and forest dialects (GaIlivare and Serri forest
Saamidistricts in Sweden).
In the northerndialect, second-syllableais labializedby first-syllable a :
northernjahttat(= Ijot'tootl) 'to start'v. southernandforestjahttat(= Ijot'taatl).
The forest dialectshavethe alternationa-e in stressedpositions:northernand
southernmannat'to go' : manniv 'I went' v. forest mannat:menniv.
Pite Saamiis divided into three dialects: northern(the district of Luokta-
Mavas in Sweden),central (the district of Semisjaur-Njargin Sweden),and
southern(the district of Svaipain Sweden).
Long (or double)aa alternateswith ee beforesecond-syllableu andi in the
centraland southerndialects,but not in the north: northernvaassiet'to go by'
: vaasij 'it went by' v. central and southernvaassiet: veesij. The southern
dialects have r where the northern and central dialects have tl: southern
aarriet 'to sleep'v. centralandnorthernaatttliet.
 SAAMIC 49

Eastern Saami
The EasternSaarnilanguagesfall into two groups: themainlandgroup (Inari,
Skolt, andAkkala Saami),andthe peninsulargroup (Kildin andTer Saarni).
The weak grades of the single sibilants, affricates, and *k have been
geminatedin the mainland group but not in the peninsulargroup (in Inari
Saarni,the geminateshave beensecondarilyshortenedin certain positions),
Inari keesist(old languagekeezzistwith the voiced sibilant Iv) 'summer
sLoc', Skolt andAkkala kVie'zzestv. Kildin andTer kie'zest,Inari ivveest'year
sLoc', Skolt ee'jjest - ii'jjest, Akkala ii'jjest v. Kildin and Ter y'gest. The
mainlandgroup has replacedthe reflexesof non-contracted*u with thoseof
*a: Inari kuarrum 'sewing' and '(having) sewn', Skolt kuarram 'idem', but
Kildin KyappaM 'sewing' (phonologically: /koarraml) v. KyppMa (phono-
logically: lkuurrmal) 'having sewn'.

The Mainland Group

The mainlandgroup consistsof two branches:Inari Saami(lnari of Finland),
and the Skolt group. The most conspicuousof the severalfeatureswhich
distinguishInari Saarnifrom the Skolt group(Skolt andAkkala Saarni)arethe
following:

1 In Inari Saarni, the strong grade of the single stop k (as well as other
instancesof k) has gone to h, and the weak gradeto v, e.g. Inari juuha
'river' :juuvVllst 'sLoc' v. Skolt andAkkalajokk:jooyyiist.
2 Skolt and Akkala have lost final unstressedvowels whereasInari has
preservedthem as a rule, e.g. Inari kyeli 'fish', tullii 'fire', aldu 'reindeer
cow' v. Skolt andAkkala kue'll, toll, and Skolt aldd, Akkala aldd.
3 Second-syllable*a has lost its lip-rounding and fallen togetherwith a in
Skolt and Akkala but not in Inari, e.g. Inari kuarrum 'having sewn' v.
Skolt kuarram, Akkala koarram - koarrmonz. Skolt and Akkala share
this feature with the peninsulargroup, but it seemsto be a relatively
recentinnovationbecauseit has also affectedcertain secondarycasesin
the mainlandgroup, in which *a goesback to *u as in Inari kuarrum <
*koarama< proto-Saarni*koaruma[nzja].The now-extinctidioms of the
mainlandgroup further south in SodankyHi., Savukoski,and Kuolajarvi
(Salla) also hadthe roundedvowel.
4 Skolt and Akkala Saarni h,ave tongue-heightalternationswhich Inari
lacks, e.g. Inari kuullM 'to hear' : kula '(s)hehears'v. Skolt andAkkala
kuullii(d) : kooll, Inari siirdij '(s)hemoved' : sirda '(s)hemoves'v. Skolt
andAkkala sii'rdi : serdd.

The Skolt group is divided into two languages:Skolt Saarni,and Akkala

Saarni(the former villages Babino and Yokostrov on ImandraLake, north of
Kandalaksha).Akkala has preservedthe nasal-plus-stop/affricateclusters,
whereasSkolt Saarnihas denasalizedthem (nd > dd, mb > bb, yg > gg, n¢
50 SAAMIC

> ¢¢, nf > ff), e.g. Akkala la'ndd 'bird' v. Skolt la'dd, Akkala soa'mbb
'stick' v. Skolt suii'bb, Akkala cuaggga 'to stick' v. Skolt cuaggad.In this
Skolt Saami agreeswith the Saamilanguages furtherwest, Akkala with the
languagesfurther east. Denasalizationspread east across the NorthlInari
Saami border after the acquisition of Christian names and terminology,
probably in the sixteenthor seventeenthcentury, cf. North andagassii,Inari
addagas'pardon'< Finnishanteeksi.
Skolt Saamihas two main dialect groups, northern, and southern.In the
northerngroup, b has turned into v in the clustersbd and bz, but not in the
southerngroup,e.g. Neidentovdad'to know', Paatsjokitovddadv. Suonikyla
- Nuortijiirvi tobddad. In the southerngroup, final g has gone to y after
sonorantsin weak-gradeclusters,but the northern group has preservedthe
stop: Neidenalggas '(going) out', Paatsjokialgas v. Suonikyla- Nuortijiirvi
aalyas. The southerngroup has preservedthe second-syllablecontracted
vowel u but it has gone to a in the northern dialects, e.g. Neiden poac'ca,
Paatsjokipuoc'ca 'reindeerpN', v. Suonikyla - Nuortijiirvi puoc'cu. The
northerngroup has preservedthe dual in conjugationbut the southerngroup
has only singularand plural, e.g. Paatsjokimoonij '(s)hewent', moonin 'the
two of them went', mo'nne 'they went' v. Suonikyla mooni '(s)he went',
mo'nne 'they (two or more) went'.
The northern group of Skolt Saami consists of two dialects: Neiden
(Neiden in Norway, now extinct), and Paatsjoki(Paatsjoki,Petsamonkyla,
and Muotka villages in the former Petsamoarea, moved over to Nellim,
Finland, after the SecondWorld War). The Neidendialect had the marker -k
in the nominative plural of nouns, whereasthe Paatsjokidialect has zero:
Neiden kuelek 'fish pN' v. Paatsjokikue'l. The Neiden dialect agreeshere
with North Saami,from which the featurewasprobablyborrowed.
The southerngroup of Skolt Saamiis divided into two dialects:Suonikyla
(in the southernPetsamoarea,now in Sevettijiirvi, Finland), and Nuortijiirvi-
Hirvasjiirvi dialect (around Lake Nuortijiirvi and south of it in the former
villages Nuortijiirvi [Notozero] and Hirvasjiirvi [GirvasozeroD.Nuortijiirvi-
Hirvasjiirvi has the stop d instead of the Suonikyla fricative a: Suonikyla
vue'aaed'to sleep'v. Nuortijiirvi vue'dded,Hirvasjiirvi vue'dde.Insteadof the
word-final voiced sibilants z and z found in Suonikyla, Nuortijiirvi-
Hirvasjiirvi hasunvoicedsands: Suonikylakorraz 'hard pN' v. Nuortijiirvi-
Hirvasjiirvi karras, Suonikyla sii'mmlaz 'Saami' v. Nuortijiirvi-Hirvasjiirvi
sii'mmlas.

The PeninsularGroup
The peninsulargroupconsistsof two languages:Kildin (in the former villages
of Kildin, Voroninsk,Varsina,Maselga,Lovozero, Lyavozero;the inhabitants
havebeentransferredmainly to Lovozero),andTer (in the former villages of
Yokanga,Lumbovsk, Ponoy, Sosnovkaand Kamensk,now scatteredon the
Kola Peninsula).The main differencesbetweenthesetwo languagesrest on
Map 2.1 Varieties of Saamic

JS JS

JS

JS
JS
JS
JS JS
JS JS

JS
JS
JS JamtlandSouthSaami
As Asele SouthSaami
JS U UmeSaami
P Pite Saami
As L Lule Saami
SN Sea(North) Saami
FN Finnmark(North) Saami
TN Tome (North) Saami
JS I Inari Saami
S Skolt Saami
A Akkala Saami
K Kildin Saami
T TerSaami

Source: Adapted fromKorhonen1981.

52 SAAMIC

vowels. In Ter, the reflexes of *uo have lost their lip-rounding, but not in
Kildin: Kildin kuu'ddo 'they left' v. Ter kiT'd'do, Kildin kuOda 'I leave(it)'
v. Ter kiodam. In Ter the reflexesof short stressed*0 and *a do not alternate,
whereasin Kildin they show qualitative and quantitativealternation: Kildin
moonno'to go' : mann '(s)hegoes' v. Ter manni": manna,Kildin poorro 'to
eat' : parr '(s)heeats'v. Ter porri": parra. Here, Kildin agreeswith Skolt and
Akkala, wherefirst-syllable *i and *u alternate,as well.

The Saami languageshave largely the same basic vocabularies.For the

Swadeshbasic list of one hundredwords, the averagepercentageof shared
vocabularyis over 80 per cent, andneighbouringlanguagesshareover 90 per
cent of this basic vocabulary, as a rule. Vocabulary pertaining to the
environment,kinship, and traditional meansof livelihood is largely the same
in all Saami languages.The most important lexical differencesstem from
transitions in meaning and use of traditional vocabulary and from recent
loanwords which have been acquired in Northern and Inari Saami from
Finnish, in the South, Ume, Pite, and Lule Saami from Scandinavian
languages,andin Skolt, Akkala, Kildin, andTer Saamifrom Russian.
Map 2.1 representsthe situation at the end of the nineteenthcentury and
the beginning of the twentieth century, when administrativemeasureshad
only minimal impacton Saamilife. Fromthe beginningof the presentcentury,
speakersof North Saamihavemovedsouthwestas far asArjeplog in the Pite
Saamiarea.In Finland,the Skolt Saamifrom the Petsamoareawereresettled
north and eastof Lake Inari as a consequence of the cedingof Petsamoto the
Soviet Union after the SecondWorld War. In Russia(at that time the Soviet
Union), the Hirvasjarvi and Akkala Saami were moved to Yona kolkhoz,
roughly half way betweentheir old winter villages. The speakersof Kildin
Saami have been concentratedin Lovozero where they are in the minority
(they make up one-quarter of the population). The Nuortijarvi Saami
dispersedin different directions when the Tuuloma hydroelectric power
station was built and (Lake) Nuortijarvi becamea regulatedreservoir.They
now live mainly in Murmanskand Lovozero.A few families havebeenable
to stayin the Verxnetulomskijkolkhoz in their traditional area.

Phonology

General Organization
Saamiwords are composedof one or more stressgroupscontainingat least
one stressedsyllable which can be followed by one or two (or in rare cases
three) unstressedsyllables.A word with more than three syllablestherefore
consistsof more than one stressgroup.As a rule, odd-numberedsyllablesare
stressed.
The stressedsyllablenucleusis calledthe vowel centreandthe consonants
 SAAMIC 53

betweenit and the next syllable nucleusare called the consonantcentre.The

first unstressednucleusis calledthe latus, and the consonantsbetweenit and
the next unstressednucleusare called the consonantmargin. The following
unstressednucleus(the third of the stressgroup) is called the vowel margin.
Final consonantsare called the finis, and initial ones the initium. In a stress
group, only the vowel centreis obligatory.
The trisyllabic word vDojasan 'I drive around' is thus divided into
positionsasfollows:

Initium Vowel Consonant Latus Consonant Vowel Finis

centre centre margin margin
v DO j a san

The word viehkalattan'I am aboutto run off containstwo stressgroups;the

initium of the non-initial stressgroup is called by a special name, limes,
becauseof phonologicalrestrictions:Iv - ie - hk -a - /1- a -tt -a - n. Normally,
a stressgroup contains two syllables. If a word has an uneven number of
syllables (i.e. is imparisyl/abic), the final syllable is the last of a trisyllabic
stressgroup, e.g. Iv - ie - hk -a -/1 - a - dd -a -m -e 'being aboutto run off' .
In somecases,the initial stressgroup is monosyllabicand the word has two
consecutivestressedsyllables,e.g.loa - mIb - ea- 11 -e 'femalecousin'.

The Vowels
The easternEnontekio subdialectof the Finnmark dialect of North Saami
(henceforth: EE) has five basic short vowels, Ii e a 0 uf, e.g. lihtel 'they
appeared',/tel'then', Imana! 'go!', loca! 'seek!',/kula! 'listen!', (= standard
languageihte, de, mana,oza, gula). Thesefive vowels can be combinedinto
double(long) vowels (written doublein this presentation)and diphthongs:Iii
ee aa 00 uuf, e.g. Itiihten/ 'I knew', /keessen/'I pulled', Ipaahcen/'I stayed
behind',/poohten/'I came',Ijuuhken/ 'I divided' (= standarddihten, gessen,
bahcen, bohten, and juhken), and lie ea oa uo/: Itiehkol 'there', Ireahka!
'sledge', loassiil 'part', Ituokkol '(going) that way' (= standarddiehko,
reahka,oassi,duokko).
The diphthongsand the doublevowel laa! can be stressedon their first or
secondcomponent.The vowel combinationswith first-member stress are
phoneticallyof greaterdurationthan thosewith second-member stress.In the
phonologicaltranscription,vowel combinationswith stresson their second
memberwill beindicatedby meansof a following': lie' ea'uo' aa'oa' uo'l, e.g.
Imaa'htel'(doesnot) know how to', Ituo'kkol '(doesnot) get tangled',loa'stel
'buy!', /kea'sel'pull!', Ivie'rrol 'foreign' (= standardwith t(, C?, q for traditional
e, 0, a mahtt(,duokkC?,oasttt(,geast(,vierrC?-; notesuchoppositionsasItuokkol
'[going] that way' v. ltuo'kkol '[doesnot] get tangled').The central nucleus
(normally the first syllable)may containall of the possibilitiesmentioned.The
lateral nucleus(normallythe secondsyllable)canhaveneithersecond-member
54 SAAMIC

stressedvowel combinationsnor diphthongs(i.e., only Ii e a0 u ii aauul occur).

Themarginalnucleus(normally the third syllable)hasno vowel combinations
(in fact only Ii e auI occur).
As a rule, the stressedsingle vowels go back to vowels which were short
in proto-Saami:Ipirraal 'around' < *pita, lkullaahl 'to hear' < *kulanek,
Imon'niil 'egg' < *mone and Inammaal 'name' < *nama (orthographically
birra, gullat, monni, namma).Before consonantcentresof quantity III (see
p. 58), the short vowels Iii leI lui and 101 can also derive from proto-Saami
diphthongs;in thesecases,there is a historically high vowel (*i or *u) in
the secondsyllable: lkir'tenl 'I endured' < *kieiDim (infinitive lkier'tahl),
Ited'denl 'I pressed'< *teairnim (infinitive Itead'diihl), kus'ken 'I touched'
< *kuoskim (infinitive Ikuos'kahl), Ikor'tnionl 'I climbed' < *koah)um
(infinitive Ikoar'tnjuuh/); orthographically girden, dedden, gusken, and
gorgyon.
The double vowel laal goes back to proto-Saarni long *a: laaj'pmiil
'triangular needle' < *aJme (standard:aibmi), and the diphthongslie/, leal,
luol and loal derive from proto-Saamidiphthongs *ie, *ea, *uo and *oa:
lkiebkal 'cuckoo' < *kieka (standardgiehka), Ikeassiil 'summer' < *kease
(standard geassi),Ikuolliil 'fish' < *kuole (standardguolli), Itoarruul 'fight'
< *toaro (standarddoarru).
The vowel combinationswith stressedsecondmemberderive from cases
in which a long vowel in the secondsyllable has been shortened:Icaa'lel
'write!' < *cjalek (standard cal'!), Ikea'sel 'pull!' < *keasek (geas'!),
poa'dan/ 'I come' < *poaoam (boailan), lkie'lttel 'deny!' < *kielDek
(gieldd,!), Ikuo'ttel 'carry!' < *kuonoek (guott'!). These combinationsare
phoneticallyof lessdurationthan the correspondingcombinationswith stress
on their first member,e.g. Icaa'lel [cjhle] 'write!' v. Ilaavel [lave] 'he usedto'
(standardcale and lave). This shorteninghas affectedcertainmorphological
categories,such as the present connegativesand second-personsingular
imperatives,most modifier componentsin compoundwords (lvaarriil 'hill' +
Icoh'kal 'top' combiningasIvaa'rre+coh'kal'hilltop', standardvarreeohkka),
and sometextually frequentword forms suchasIpoa'dan/'I come' alongside
unshortenedIpoadaan/,standardboatlan, oa'cconl'I get' alongsideunshort-
ened loaccuunl, standardoaccun. It is also fairly common in unstressed
positions. The shortened second-syllablevowels are identical with the
original short vowels, resulting in oppositionssuch as Isoadan/'I fight' v.
Ipoa'dan/'I come' (standardsoatlanandboatlan).
The backgroundof the proto-Saamistressedvowels andtheir development
into the present-dayEasternEnontekio dialect is shown by the following
examples (pFS = proto-Fenno-Saamic,pS = proto-Saarni,EE = Eastern
Enontekio. The forms given in brackets under 'standard' are in Konrad
Nielsen'sorthography[KN =Nielson 1932-62]):
By andlarge,the pFS old low (*a and *ii) andlong mid (*e and *0) vowels
have developedinto long vowels, while pFS high vowels (*1, *i, *u, *u, *u)
 SAAMIC 55

Table2.1 Background and developmentof the proto-Saami stressedvowels

PFS > pS > EE Standard

*ii *kiili- 'hears' > *u *kula- > lui Ikullaa-I gulla- (gulla-)
*u *puri- 'bites' > *0 *pora- > 101 Iporraa-I borra-(borra-)
[*o( -i) *joki 'river' > *0 *joka > 101 Ijohkaa! johka (jokka)]
*i *nimi 'name' > *a *nama > Ia! Inammaa! namma
(namma)
*ii *siili 'bosom' > *a *sala > Ia! Isallaa! salla(salla)
*e(-i) *meni- 'goes' > *a *mana > Ia! Imanna-I manna-
(manna-)
*1 *pIri 'circle' > *i *pira, > iii Ipiirraa! birra (birra)
*a *kala 'fish' > *uo *kuole > luol Ikuolliil guolli (guolle)
*kala-na sEss > *kuolene > luol Ikuolliinl guollin
(guollen)
*kala-mi sl > *kuolema > luol Ikuollaan! guollan
(guollam)
*kala-sin sIll > *kuolan > luol lkuollaaj/ guollai
(guollai)
*kala-j-ta pP > *kuolijoe > luul Ikuulijh/ giiliid (giiliid)
*o( -i) *koslJi- 'touches'> *kuoslJa- > luol Ikuos'ka-I guoska-
(guos'ka-)
*koslJi-j-i-m pastsl > *kuoslJim > lui Ikus'kenl gusken
(gus'kim)
*0 *voli- 'whittles' > *vuola- > luol Ivuolla-I vuolla-
(vuolla-)
*voli-i-i-m pastsl > *vuolim > luul Ivuullen! viillen
(viillim)
*o(-a!o) *kota 'hut' > *oa *koate > loa! Ikoahtiil goahti (goatte)
*kota-j-ta pP > *koaoijoe > 1001 Ikoodijh/ gMiid (gMiid)
*li(-ato) *lijmli 'needle' > *a *ajme > laa! laaj 'pmiil rubmi (ai'bme)
[*a(-a!o) *akka 'wife' > *ahke > Ia! lah'kaa! ahkka(ak'ka)]
*e(-ata!o) *kesli 'summer'> *ea *kease > lea! Ikeassiil geassi(gresse)
*kesli-j-tli pP > *keasijoe > leel Ikeesijh/ gesiid (gesiid)
*kejno 'way' > *keajna > lea! Ikeaj'tnuul geaidnu
(grei'dno)
*kejno-j-ta pP > *keajnojDe > lea! Ikeajnnuujh/geainnuid
(greinoid)
*kejno-lJsii diminut. > *kea)nun > leel Ikeejnnosl geinnos
(geinus)
*kejno-sin sIll > *keajnuniia > lei Ikej'tnujl geidnui
(gei'dnui)
*li( -i) *kliti 'hand' > *ie *kieta > liel Ikiehta! giehta(gietta)
*kliti-sin sIll > *kietin > Iii! Ikiihtij/ gThtii (gIttii)
*e *keli 'language> *kiela > liel Ikiella! giella (giella)

*keeli-sin sIll > *kielin > liel lkiella! gIllii (gIllii)

[*e( -i) *sielki- 'clear' > *i'9ieloa- > liel Icielkkasl cielggas
(cielgas)
56 SAAMIC

havedevelopedinto shortvowels.The pFS shortmid vowels (*e and *0) have

developedinto long vowels before second-syllablelow vowels. There is an
asymmetryin the developmentof pFS *e and *0 beforesecond-syllableshort
vowels (pFS *o-i > proto-Saami*uo-11, but pFS *e-i > proto-Saami*11-11), the
reasonfor which is not clear; there are also some exceptions(enclosedin
squarebracketsin Table 2.1) which tend towardssymmetricaldevelopment.
The developmentof the proto-Saamivowels in the other Saamilanguagesis
complex and cannotbe gone into here; some of the changesleading to the
present-daylanguagesand dialects have already been referred to in the
introductorysection.
The main courseof developmentof the vowels of the secondsyllable may
be inferred from Table 2.1. Complicationsarise,however,from the fact that
vowels of the third syllable influenced the developmentof vowels in the
secondsyllable. Third-syllable mid proto-Saami*1'5 (which becameproto-
Saami *11) causedthe descendentsof pFS *a and *ii to remain low (> proto-
Saami*a), preventingthem from fronting (proto-Saami*e), but causingpFS
*0 to developinto proto-Saami*u insteadof regularproto-Saami*a. These
second-syllablevowels, in turn, influencedthe vowels of the first syllable:
proto-Saamidiphthongswent to monophthongsbeforesecond-syllable*i and
*u. Vowel contractionsare complex processeswith ample morphological
conditioning; only somecases(in which intervocalic *s and *j go to zero)
havebeenincludedin the table. In the WestFinnmarkdialectgroup to which
EE belongs, there is a lengthening of second-syllable*11 to laa! after a
relatively shortfirst syllable, thus coinciding with the regularreflex of proto-
Saami *a: proto-Saami*t0111 'fire' > Itollaal, proto-Saami*jok11 'river' >
Ijohkaal, proto-Saami*kulam 'I hear' > /kulaan/,proto-Saami*n11m11 'name'
> Inarnmaal (standarddolla,johka, gulan, namma).

Consonants
The systemof consonantphonemesof EE is characterizedby the extensive
useof the oppositionof voice.
Extensiveuse of the voice oppositionis typical of all Saamiidioms, but
North Saamihassomepairswhich arenot normally found in others,e.g. M:m,
N:n, L:I, R:r, J:j, t:d. The West Finnmarkdialects,EasternEnontekioamong
them, lack lUi, becauseprevocalic lUi> Inj/. The standardlanguagehas IJ:
maIJIJil 'afterwards', EE Imaaninjiill. PreconsonantalIJ in forms such as
maIJga 'many' is a variant of In!: Imaan'kal.The consonantsItj dj g z zljl
occur only as geminatesof the consonantcentre,e.g. laadi'diaa! 'grandpa',
Ipealj'ljiil 'ear'.
The distribution of the consonantsis highly skewed; there are severe
restrictionson their occurrenceeverywhereexceptin the consonantcentre,
i.e. in initial, final, marginalandliminal positions.In thesepositionsthe voice
opposition is generally neutralized (only the unmarked membersoccur),
palatalization is restricted, and, in initial and final position, the dental
 SAAMIC 57

Table2.2 Consonantphonemesof EasternEnontekio

Bilabial Labiodental Interdental Post- Alveolar Palatalized Pre- Velar Laryngeal

dental palatal

Nasals
Voiced ill n ni
Unvoiced M N
Stops
Voiced b d eli g
Unvoiced p f tj k
Affricates
Voiced z Z
Unvoiced c C
Fricatives
Voiced v d j
Unvoiced f f s S J h
Liquids
Voiced 1 r Ij
Unvoiced L R

fricatives are lacking (1M is found intervocalically). The consonants1m n ni

p t k c cv f s 8 j h 1 rl can occur in initial position, plus the clustersIsp st sk
sm sn snj sl sr 8ni Sl/. In marginalpositiononly Ip k c cm n v s 8 d j h11 occur,
plus the clusters1st sk 8t rt rs lt lk 1m jk jm jn jd jstl_ In [mal position,labials,
palatalizedconsonants,and velars are lacking; the only consonantsoccurring
hereareIt n s 8 1 hi,
j plus the clustersIrs rh lh jn js jhl.

The ConsonantCentre
The consonantcentre is rich in oppositionsand morphophonemicalterna-
tions. Gradealternationand shorteninghave alreadybeenmentioned.Other
alternationsaffecting the phonology of the consonantcentre are strengthen-
ing, lengthening,and secondaryshortening.Grade alternation has doubled
original single consonantsbefore originally open syllables, e.g. pFS *kala
'fish' : *kalan 'sG' : *kalana 'sEss' : *kalasin 'sTIl' > EE /kuolliil : lkuoliil
: /kuolliinl : /kuollaaj/ (standard guolli : guoli : guollin : guolilii) , pFS
*men(i)tiik 'to go' : *menim 'I go' : *menijim 'I went' > EE Imannaah/:
Imanaan/: Imannenl (standardmannat: manan : mannen).In certain cases,
thesegeminateshave developedinto secondaryclusters:pFS *kota 'hut' »
EE /koahtiil, pFS *soni 'sinew' » EE Isuotna/.Historically, the stronggrade
originatesin an extra subglottalpulsein the consonantcentre;with time, the
strong gradeof the single consonantsbecamephoneticallyidentical with the
weak grade of the geminateconsonantsby way of *-VcCcCV- > *-VC(CV-,
where 'c' denotesa pulseboundary(therearedialectsin which this mergerhas
not occurred). In original geminates and clusters, the strong grade is
manifestedas an extra subglottal pulse within the consonantcentre before
58 SAAMIC

originally opensyllables,making the stronggradelongerthanthe correspond-

ing weak grade:Ipas'tel 'spoon' : Ipastel 'sG', kol'lii! 'gold' : /kollii! 'sGA',
in broadphonetictranscriptionwith pulsesmarked: [pas(s(te]v. [pas(te] and
[kol(l(li] v. [kol(li] (more commonly,notationssuchas [paste] v. [paste] have
beenused).In consonantclustersbeginningwith a voicedconsonant,the final
componentof a weak grade cluster has been geminated:Ivaal'taanl 'takes
perf.part' v. Ivaalttaaanl'I take'; contrast/kuos'kah/'to touch' : /kuoskanl'I
touch' (standardvaldit, valddan,guoskat,guoskkan).
Therearethreecontrastingquantitiesin qualitativelyidenticalEE consonant
centres:/kol'liisl 'gold s3 sN' : /kolliisl 'gold s3GA' : loliisl 'at, near'(standard
gol'lis: gollis: olis, KN gol'les: golles: olest),/eaal'liihl'writerpN' : leaalliihl
'to write (inf)' : leaaliihl 'to makeX write (s2pres/imp)'(standardcal 'lit : callit
: calit, KN cal'lek : callet : calet). The longest possible quantity, which
historically is in mostcasesthe stronggradeof original geminatesandclusters
and is written with' in phonological transcription (and in the somewhat
modified standardorthographyof this chapter)is called gradeIII: leuol'pma!
'knot', lah 'kuul 'grandmother', /kuos'sii! 'guest' ,/seal'kii!'back', Ivis'tii! 'flat
(standardcuolbma, ahkku, guos'si, sealgi, visti). Quantity II normally repre-
sents(1) the weakgradeof geminatesandclustersand (2) the strong gradeof
original single consonants.Examplesof 0): leuolmma! 'knot sGA', laahkuul
'grandmothersGA', /kuossii! 'guestsGA', Isealkkii! 'backsGA', lvistii! 'flat
sGA'; examplesof (2): /koahtii! 'hut', loassii! 'part', /kiella! 'language',
Isuotna!'sinew' (standardcuolmma,ahku, guossi,sealggi,vistti, goahti, oassi,
giella, suotna).Quantitylis theweakgradeof original singleconsonants:/koad
iii 'hut sGA', loasii! 'part sGA', /kiela! 'languagesGA', Isuona! 'sinew sGA'
(standardgoadi,oasi, giela, suona).
Strengtheningis a process whereby the reflexes of originally single
consonantsequal quantitatively the strong grade of original geminates,i.e.
whereoriginal simplexconsonantsgo not to gradeII, but to gradeIII. In North
Saami, strengtheningoccurs when a consonantbelonging to the stem or
derivational suffix is lost, e.g. Isul'lohl (standardsul'lot) 'islands' < proto-
Saami *suolluk < mid proto-Saami *soluj-ek < pFS *saloj-i-t, Ijuoh'kii!
(juohkki) 'one who deals' < proto-Saami*juokke < pFS *jaka=ja. If the lost
consonantwas part of an inflectional suffix strengtheningdid not occur, e.g.
Ijuuhken/ (juuhken) 'I dealt' < proto-Saami*juofcim < < pFS *jaka-j-i-m
(infinitive juohkit).
Lengtheningfrom quantity II to quantity III occursbeforesecond-syllable
long vowels (Iii aa uul) when the first syllable containsa short vowel (Ii a 0
uI) and also, if the resulting lengthenedconsonantcentredoesnot equal the
strong grade of an original geminateor cluster, a high diphthong (lie uo/).
Where the strong gradeof a single consonantis a clusterbeginningwith /hi
or a geminateotherthanIpp tt tj~ tj~ kk cc eel, lengtheningresultsin a consonant
centre equal to the strong grade of original clusters or geminates:Ijah'kii!
'year' : Ijakii! 'yearsGA' (standardjahki: jagi) , Ipal'luul 'fear' : Ipaluul 'fear
 SAAMIC 59

sGA' (standardballu : balu), /joh'taal '(s)he travels' : /jodaanl 'I travel'

(standardjohta: joaan; cf. caseswith strengtheningsuch as /joh'taal '(s)he
startsto travel' : /joh'taanl 'I start to travel', standardjohtta:johttan). In the
remainingcases,the lengthenedconsonantcentreshowsquantity III but does
not equal the strong grade of original geminatesand clusters.The phono-
logical oppositionbetweenlengthened andunlengthenedsequencesrests on
secondarylengtheningof second-syllable*a to /aa/ in EE (the normal reflex
being /aI): Iloniis/ 'ransom' : Ilot'naasahl 'pN' as opposed to /sonaas/
'shrunken': /sotnaasahl'pN' (standardlonis : lotnasatv. sonas: sotnasat).
Similarly, from Ilod'diil 'bird' is formed a diminutive llot'taaSi but from
/pod'dal 'shorttime' we have/pottaas/(standardloddi, lottas, bodda, bottaS);
from /eal'pmiil 'eye' we have diminutive /ealm'maas/,contrast/ealmmaas/
'reticulum' (standardcalbmi, calmmdsv. calmmas).
A secondaryshortening in the consonantcentre from quantity III to
quantity II may occurif the second-syllablelong vowel undergoessecondary
shortening.Normally thereis an accompanyingoverall shorteningof the long
elements(of vowel combinationsand of the consonantcentrein quantity III)
in the stressgroup: /vuol'kaal > /vuo'lkal '(s)he goes' (standardvuolga),
/aal'kaal> /aa'lkal '(s)hebegins' (standardalga), /saah'taal> /saa'htal'(s)he
might' (standardsahtta),/peas'taal> /pea'stal'(s)helets', /sad'daal> /saddal
'(s)he grows'(standardsadda),/jah'kiil 'year' > /jahke+pealliil 'half-year'
(standardjahki,jahkf?bealli), /eat'naal> /eatnal '(s)heties' (standardcatna).
Quantities II and I are not affected by shortening: lkuolliil 'fish' >
/kuo'lle+piv'tuul 'fishing' (standardguolli, guollf?bivdu). Vowel combina-
tions with stressedsecondmemberwhich occurin thesecasesare shorterthan
thosewith stressedfirst member,e.g. /raahtal'rattle' v. /saa'htal'(s)hemight'
(standardrahta, sahtta).
As a resultof the variousprocessesaffectingconsonantcentrequantity,the
following phonologicalquantititiesmay be positedfor EE clustersbeginning
with a voicedconsonant:

Cluster EEexample Gloss Standard

Il'k/ /vuol'kaanl 'havingleft' vuolgan

Ilk/ /vuo'lkal (- /vuol'kaal) '(s)heleaves' vuolga
Ilkk/ /vuolkkal 'departure' vuolgga
Ilk'k/ /vuolk'kaanl 'I leave' vuolggan

This four-way set of quantitativeoppositionsis possiblefor the following

clusters(cited in unshortenedstronggrade):/m'p n't n'e n'k d'k d'p v't v'e v'k
v'l v'L v'r j'lj'r j'kj'p j't j's j'v j'L j'M j'N l'j l'k l'p l't l's l'fl'v l'j l's l'e/.
EE gradealternationalso affects consonantquality. A weak-gradesingle
consonantalternateswith a strong-gradecluster (standard-language forms
given in brackets)
60 SAAMIC

• if the consonantgoesbackto a proto-Saamistopor affricate,e.g.Ikoahtiil

'hut' : Ikoadil 'sG' (goahti " goadi), /kiehpal 'soot' : /kiepal 'sG' (giehpa
" gieba), Iciehkal 'comer' : Iciekal 'sG' (Ciehka " Ciega), Ipaahciihl 'to
stay' : Ipaacaan/'I stay' (bdhcit,' bdzan),Ipaahciihl 'to shoot' : Ipaacaan/
'I shoot'(bdhCit,' bdian).
• if the consonantgoesback to a proto-Saaminasal,and the initium does
not contain a nasalconsonant:Iliepma I 'broth' : Iliemal 'sG' (liepma "
liema), Isuotnal 'sinew' : Isuonal 'sG' (suotna " suona), Ipotniaahl 'to
wind' : Iponiaan/'I wind' (botnjat " bonjan ), Ijietnial 'ice' : Ijienial 'sG'
(jiek-fJa " jie-fJa), and (proto-Saami*-j- » -j- alternateswith -titL , e.g.
'butter' 'butter' : lvuojal 'sG' (vuodja,' vuoja).
Ivuo~tial

Voiceless weak-gradedouble stops and affricates alternatewith voiced

strong-gradegeminates:Inub'biil 'second': Inup'piil 'sG' (nubbi,' nuppi, KN
nub'be " nubbe),Ilod'diil 'bird' : Ilot'tiil 'sG', Icag'giil 'support' : /cak'kiil
'sG', Ivaaz'ziihl 'to walk' : Ivaaccaanl'I walk' (vazzit,' vaeean),loaf'zuuhl
'to acquire': loaccuun/'I acquire'(oaiiut,' oaccun),laacP'diaal'grandfather'
: laatitiaal 'sG' (addja : adja).
Clustersconsistingof a nasalprecededby a voicedstopin the stronggrade
havea voicelessstop in the weak grade:Icaab'miihl 'to beat' : Icaapmaan/'I
beat' (eabmit " eapman),lead'niil 'mother' : leatniil 'sG' (eadni : eatni),
Ipoad'niiil 'husband': Ipoatniiil 'G' (boadnji: boatnji), ltuod'niahl 'to patch'
: Ituotnian/ 'I patch' (duog-fJat: duok-fJan).
Strong-gradeclusters with initial IkI alternatewith weak-gradeclusters
with initial Ivhl beforea stop or affricate, but with Ivl before sibilantswhich
are doubled, e.g. /cik'cuuhl 'to pinch' : Icivh'cuun/ 'I pinch' (cikeut :
civeeun), Ikok'cahl 'to cover' : Ikovhcan/ 'I cover' (gokcat : govccan),
lruok'tuul 'home' : lruovh'tuul 'sG' (ruoktu,' ruovttu), Isuok'sal 'maggot' :
Isuovssal'sG' (suoksa : suovssa),Itik'suuhl 'to tend' : ltivs'suun/ 'I tend'
(dikSut: divssun).
Weak-gradeclusterswith a final nasalgeminatealternatewith strong-grade
clusterswith a stop-plus-nasalcombination.The initial componentmay be Ij/,
Ill, or Iv/: Ical'pmiil 'eye' : Icalm'miil 'sG' (calbmi : calmmi), loaj'tniihl 'to
see': loajnnaan/'I see'(oaidnit: oainnan),etc.
In some clusters, the difference between strong and weak grades lies
mainly in the location of the syllable boundary.Someof theseclustersbegin
with Irl and end in a nasal: Ipaar'tniil 'son' : Ipaartniil 'sG', phonetically
[par(rE(lni] : [piirt(ni] (standardbdrdni " bdrtni), Icor'pmal 'fist' : Icorpmaal
'sG' (corbma : corpma), Ikoar'tniuuhl 'to climb' : Ikoartniuun/ 'I climb'
(goarg-fJut: goark-fJun).Othershave/hi as an internal segment,beginwith Ij/,
Ill, 1m!, In/, Ir/, or lvi, and end in a stop or affricate: Ipaaj'hkiil 'place' :
Ipaajhkiil 'sG', phonetically[paj(jtki] : [paiJ(ki] (bdiki : bdikki), laaj'hcahl'to
notice' : laajhcan/ 'I notice' (aieat : aieean), Ipeal'hkiihl 'to scold'
Ipealhkaan/ 'I scold' (bealkit " bealkkan), Ipol'htuuh I 'to root up' :
 SAAMIC 61

Ipolh'tuun! 'I root up' (boltut: bolttun), Ipaar'htii! 'mishap' : Ipaarhtii! 'sG'
(btirti : btirtti), lkum'hpel 'wolf : /kumhpel 'sG' (gumpe : gumppe).The
latter groupcontainsthe following clusters(given in stronggrade):Im'hp n'hk
n'hc n'hej'hp j'hkj'htj'hc v'hp v'hk[v'ht] l'hk l'hp l'ht l'hc l'her'hp r'hk r'ht
r'hc rh'e/.
The RelationshipBetweenMarginal and Final Consonants
Morphonological alternationscausemarginal consonantsto becomefinal
consonants;this in tum causesnumerousneutralizations.When they become
word final, marginal stops and Id/ are all replacedby /hi (before absolute
pausephonetically[ht]), e.g. Ipaavheakihl'to hurt' : lij paavheahl'doesnot
hurt' (btivccagit : ii btivccat, i.e. in this pair /k/-/hl is orthographically<g>
- <t», Inuorapuhl 'younger pN' : Inuorahl 'sN' (nuorabut : nuorat),
Iviel'kadahl 'white pN' : Iviel'kahl 'sN' (vielgadat: vielgat). Marginal 1m! is
replacedby In! whenfinal: lealliimisl 'sLoc' : lealliin! 'sN' (eallimis : eallin).
Marginal affricates are replaced by correspondingsibilants when final:
/lot'taaeahl'little bird pN' : /lot'taasl 'sN' (lottazat : lottas), Ismiirecihl 'to
chew the cud' : lij smiiresl 'does not chew the cud' (smirezit : ii smires).
Marginal clusters beginning with a voicelessconsonantare replacedby a
single consonantwhen in final position; marginal clusters consisting of a
voiced consonantplus a stop replacethe stop with /hi when final, examples:
Ivuo'jestihl 'to drive a little' : lvuo'jesl 'drive a little! (s2 imp)' (vuojestit:
vuojes),Ipeerostihl 'to care' : lij peerosl'doesnot care' (berostit : ii beros),
Ijaamaalkihl'to faint' : lij jaamaalhl'doesnot faint' (jamdlgit: ii jamal), etc.
Noteson the Historical Background
A few additional remarks will supplement the consonantdevelopment
implicit in the previoussections.
From a qualitative point of view, the Saami consonantstock is fairly
conservative.Proto-Fennic-Saamicpalatalizationhas been preserved,e.g.
pFS *ni6li 'arrow' > EE lniuollaf (standardnjuolla), pFS *minjii 'daughter-
in-law' > EE Imani'njii! (mannji), pFS *punii- 'twists' > EE Ipotniaa-I
(botnja-), pFS *sjilmii 'eye' > EE lealp'mii! (calbmi), pFS *ktisji- 'asks'> EE
/kaheaa-I(gahca-), pFS 'urine'>
kun~i 'urine'> EE /koz'zaf (gozza), pFS *kosjki
'river rapids' > proto-Saami*kuosjsjka> EE /kuoj 'hkaf (guoika). So haspFS
*1), as a rule (but in EE and other West Finnmarkdialectsit has gone to lnil
prevocalically),e.g. pFS *jiil)i 'ice' > North Saamijiek1Ja, EE Ijietniaf, pFS
*jiil)kii 'bog' > proto-Saami*jeaUGe > EE Ijeag'gii! (standardjeaggi),but cf.
pFS *jol)si 'bow' > proto-Saami*juoI<:sa> EE Ijuok'saf, standardjuoksa.
Some shifts have occurred, however. The post-alveolars(the probably
somewhatretroflex sibilant *s and affricate *e) have changedinto dentals
(and pFS *s thus fell togetherwith pFS *s), e.g. pFS *viei 'new snow' >
proto-Saami*vaca > EE Ivahcaaf(vahca), pFS *ktinei 'nail' > proto-Saami
*kann¢a > EE /kaz'zaf (gazza), pFS *nioeka 'depression'> proto-Saami
*njoaccke > EE Injoas'kii! (njoaski). In consonantclusters, the following
62 SAAMIC

major changesoccurredafter the split of proto-Fennic-Saamic:

pFS proto-Saami EE example(standardin brackets)

*ns > *ss *kansa'folk' > lkuos'siil (guos'si)
*sj, *sj > *ss *posjo 'bottom' > Ipoas'suuJ'innermost
part of hut' (boas'su)
*osja 'horsetail(plant), > proto-Saami
*oasjsje» EE hoas'sa(hoas'sa)

Also, pFS *k > proto-Saami*v before *j, *1, and *ni, e.g. pFS *saknia 'hole
(in ground)' > EE Isuov'tniii I 'grazing-hole'(suovdnji), pFS *vakja 'wedge'
> Lule Saamivuovjje.

Morphology

General Features
The morphological mechanismsin Saami are phonological alternation,
suffixation, compounding,and encliticization. Suppletionand reduplication
arerare. Someoppositionsreston suffixes: North Saamiboahta'(s)hecomes'
v. boahtan '(having) come'. Others rest on stem phonology: North Saami
boatlan 'I come' v. boahtan '(having) come'. Still othersrest on both: North
Saamigoahti 'tent' v. gi5tliin 'in tents'.
In general,distinctionstendto be madetowardsthe centreof the word (the
vowel and consonant centre), with corresponding reductions in other
positions.Proto-Fennic-Saamic canbe reconstructedas a purely agglutinative
languagewith no or very little morphemeinteraction. Alreadyin proto-Saami,
gradealternation,contractions,and, to someextent,metaphony,had brought
abouta shift towardsfusional insteadof agglutinativemorphology.After the
proto-Saamicstage,metaphonyof stressed-syllable vowels and reductionof
unstressedpositions have madethe original morphemesand the boundaries
betweenthemlessandlesstransparent.
Saamisuffixes may be divided accordingto their positionalprivilegesinto
threetypes: markers(for mood, tense,and number),inflectional suffixes(for
case,possessor,and person), and derivational suffixes(for modification of
meaningsand recategorization,i.e. shifts in part of speech).Derivational
suffixes are attacheddirectly to stemsto form new stems,markerscomeafter
derivational suffixes, and inflectional suffixes come after markers (ditics
come after inflectional suffixes). In compounds,the last constituentis the
head,and thoseprecedingit are modifiers.
Saamiis the only Finno-Permiclanguageto have a distinct dual number.
The Saamidual is restrictedto morphologywith personalreference,viz. the
personalpronouns(North Saamimun '1', moai 'the two of us, we two', mii
'we (more than two); the personalsuffixes of verbs (North Saamibohten 'I
 SAAMIC 63

came', bi5iliime 'we (dual) came', bOtliimet 'we (plur) came'), and the
possessivesuffixes (North Saami goahtan 'my tent', goahtame 'the tent
belongingto the two of us',goahtamet'the tent belongingto us (plur),.
Inflection
Inflection is of two basic types: declension(for nouns,adjectives,numerals,
and pronouns)and conjugation(for verbs). Most adjectiveshave a special
attributive form (e.g. duojar lea ceahppi 'the craftsmanis skilful' v. ceahpes
duojar 'skilful craftsman'),the categorizationof which is not clear. The
morphology of attributive forms is irregular and thus typical more of
derivation than inflection; it is not a clear-cutderivational category,on the
other hand, for there are a substantialnumber of adjectiveswhich lack a
specialattributive form yet which function as attributivesin the sameway as
attributive forms, e.g. nieida lea nuorra 'the girl is young', nuorra nieida
'younggirl'.
Declension
Nouns,adjectives,numerals,andpronounsareinflectedfor number,case,and
possession.It is customary to distinguish between absolute declension
(possessionnot indicated)andpossessivedeclension.Therearetwo numbers,
singularand plural. The casesin standardNorth Saamiand most dialectsare
seven:N[ominative], G[enitive], A[ccusative],ll1[ative], Loc[ative], Com[ita-
tive] , and Ess[ive]. For nouns and most pronouns,genitive and accusative
singularare identical in form, andhencethe term genitive/accusative is used.
In numeralshigher than okta 'one', the accusativesingularis identical with
the nominative singular, while the genitive singular matchesthe genitive/
accusativesingularforms of nounsas far as morphonologyis concerned(e.g.
guokte'two', sA guokte,sG guovtti; cf. luokta 'bay', sGA luovtta).
Genitive and accusativeplural are identical in the standardlanguageand
in most dialects.The essivehas the sameform in the singular andthe plural.
SomeNorth Saamidialectsadd the abessiveto the list of cases,but in most
dialects this case has been replaced with a postpositional construction
(genitive+ the postpositionhaga).
The demonstrativepronounshave historical singular and plural prolative
caseforms (duokko 'that way over there',duoigo 'aboutthat way over there')
but synchronicallytheseare adverbs,togetherwith other local expressions
basedon the stemsof demonstrativepronouns(duohko '[going] over there',
duoppe'overthere').
The possessivesuffixes refer to first, second,and third personin singular,
dual, and plural. Possessiveforms for the nominative plural (first persons
only) occur, rarely, in a vocative-like function. Possessivesuffixes normally
come after caseendings,exceptin the comitative plural where they precede
the caseending-guin.
Stems may be classified into the following groups: parisyllabic, impar-
isyllabic, and contracted. Examples of inflection are given below (for
Table 2.3 Absolute declension

Singular Plural Singular Plural Singular Plural Singular Plural

Parisyllabic
N guolli guolit giehta giedat buoddu buodut beaivvadat beaivvadagat
GA guoli giiliid gieda giedaid buodu buodiiid beaivvadaga- beaivvadagaid
beaivvadat
III guollai giiliide gihtii giedaide biiddui buodiiide beaivvadahkii beaivvadagaide
Loc guolis giiliin giedas giedain buodus buodiiin beaivvadagas beaivvadagain
Com giiliin giiliiguin giedain giedaiguin buodiiin buodiiiguin beaivvadagain beaivvadagaiguin
Ess guollin giehtan buoddun beaivvadahkan
Imparisyllabic
N beana beatnagat oalis oalhisat cielus cillosat
GA beatnaga- beatnagiid oalhisa- oallasiid cillosa - cillosiid
beatnat oallas cillos
III beatnagii beatnagiidda oallasii oallasiidda cl1losii cl1losiidda
Loc beatnagis beatnagiin oallasis oallasiin cl1losis cl1losiin
Com beatnagiin beatnagiiguin oallasiin oallasiiguin cl1losiin cillosiiguin
Ess beanan oalisin cielusin
Contractedstems
N jallJlJis jalglJelt boazu bohccot
GA jalglJa jalglJaid bohcco bohccuid
III jalglJai jalglJaide bohccui bohccuide
Loc jalglJas jalglJain bohccos bohccuin
Com jalglJain jalglJaiguin bohccuin bohccuiguin
Ess jallJlJisin boazun
 SAAMIC 65

Table2.4 Possessive
declension:guos'si 'guest'

Singular Dual Plural

sN 1 guos'san guos'same guos'samet

2 guos'slit guos'sade guos'sadet
3 guos'sis guos'siska guos'siset
sGA 1 =sN =sN =sN
2 goussat goussade goussadet
3 guossis guossiska guossiset
pIll 1 giissiidasan giissiidasame giissiidasamet
2 giissiidasat gUssiidasade giissiidasadet
3 giissiidasas giissiidasaske giissiidasaset
pCom 1 giissiidanguin giissiideaskkc:tguin giissiideaskkc:tguin
2 gUssiidatguin uin
giissiideaskkc:tg uin
giissiideaskkc:tg
3 giissiidisguin uin
giissiideaskkc:tg uin
giissiideaskkc:tg

parisyllabic: guolli 'fish', giehta 'hand, ann', buoililu 'dam', beaivwidat

'sunshine';for imparisyllabic: beana 'dog', oalis 'groovein bottom of ski',
cielus 'abuse';for contractedstems:jalIJIJis 'tree stump', boazu 'reindeer'.
Whennot evidentfrom the standardorthography,doublevowels areindicated
with a macron, e.g. u; quantity III in the consonantcentreis indicatedwith
I, e.g. SiS.

The sLoe, sIll, sCorn, Ess, pGA, and pLoc possessiveforms are built by
combiningthe stems:

sLoc guossist-
sIll guos'sas-
sCorn gUssiin-
Ess guos'sin-
pGA gussiid-
pLoc gussiin-

with the suffixes (standardorthographyy, 9, ~ written hereas E, 0, A):

Singular Dual Plural

1 -an -eame -eamet

2 -at -eatte -eattet
3 -is -easkka -easet

Note the following morphophonologicalalternationsof vowels in lateral

position (secondsyllable): i-a (with pFS *i in the next syllable: guos'si -
guos'san< *kansa - *kansami) and u - 0 (with pFS *i in the next syllable:
66 SAAMIC

cielus- clllosat, buoddu'dam' - baddon'my dam' < pFS *paoo - *paoomi).

Before marginal preconsonantal/j/ (orthographici) pFS *ala has gone to
proto-Saamishort *i, which monophthongizesthe frrst-syllable diphthong,
e.g. pFS *kala-j-ta > proto-Saarni*kuolijoe > North Saarni galiid). The
second-syllablei, u, and d of the illative singular forms gfhtii, baiiiiui and
guolldi are the resultsof contraction(of pFS *-isi-, *-osi-, and *-asi-), as are
the short0 (and u beforei =/j/) in the declensionof boazu(from pFS *-oji-);
beforea lateral short i, e, 0, and u, frrst-syllablediphthongsare monophthon-
gized (uo > 11/u, oa > 6/0, ie > Vi, ea > e/e; the short variants occur before
consonantcentrequantityIII).
Case endings: Historically, the genitive and accusativesingular have
syncretizedafter the loss of the final nasals(pU *-n for the genitive and pU
*-m for the accusative).The illative singularhas acquiredthe secondarycase
ending -i, originally the marginal stem consonantin *j-stems (e.g. boazu,
proto-SaarnisN *poaz6j, sIne *poaccuste< pFS *pocoj-i-sta, sIll *poaccujin
< pFS *pocoj-i-sin); the original illative suffix has mergedwith the stemby
contraction(e.g. pFS *kati-sin > proto-Saarni*kietin> North Saarnigfhtii;
illative singular forms without the secondaryending -i have also been
attested).
The locative singular(in older orthographieswith the ending -st as in the
possessivedeclension)continuesthe former inessiveand elative, which are
preservedas separatecasesin the idioms southwestof North Saarni (pFS
inessive *-snal*-sna and elative *-stal*sta, in which *-s- was originally a
lative elementattestedalso in Mordva and Mari, and *-nal*-na as well as
*-tal*-ta continue original general local cases.The suffix *-nal*-na and
probablyalso *-tal*-ta are of pU origin, from a pU locative *-nal*-na and an
ablative *-ti.) The syncretismin the locative of North Saarniand the Eastern
Saarnilanguagesis due to loss of the final vowel followed by denasalization
of the final n in the inessiveending (proto-Saarni*-sne > *sn > -st), after
which the mergerspreadfrom this centralcaseto more peripheralpartsof the
grammarand vocabularywhere the phoneticconditionsdid not obtain. The
mergerthereforealso occurredin the plural,in the possessivedeclension,and
in adverbs,wheretherewere no phoneticgroundsfor the change(e.g. olgun
'[being] outside; [coming] from outside',contrastLule dlggon 'outside' v.
dlggot 'from outside'.At least in part, an analogicalexplanationis valid for
the mergerof the genitiveandaccusative,too.
The comitative singulargoesback to pFS *-jnal*-jna, which is originally
the essivecaseof a possessive j-adjective(thej is ultimately the sameas the
pU plural oblique casemarker *-j-). Inari Saarni, Lule Saami and most of
Fennicpresupposea pFS form *-jni yielding proto-Saarni*-jna; *-ni may be
an old variantof *-nal*-na.
The essiveending-n goesbackto the pU locative *-nal*-na.
In trisyllabic stems with grade alternation (cf. oalis in the paradigms
above)and in contractedstems,the consonantcentreis in the weak gradein
 SAAMIC 67

the essive(as in oalisin, jalI)I)isin, boazun).In thesecases,the vocalic stem

is secondary;it is basedon the analogyof the trisyllabic stemswithout grade
alternation, i.e. old vocalic stems, and the secondsyllable was originally
closed,causingthe weakgrade.Consonantalstemshavebeenattestedin (now
extinct) SodankyHiSaami, and in Lule Saamiessivessuch as btilgen (from
btilges 'path'), stem-final -s has been assimilatedinto the essiveending -n
(*-s-n > -n), just as the inessiveending -n goesback to earlier *-sn « pFS
*-snal*sna).
The plural marker -t, usedin the nominative(-k in the old orthographies,
< proto-Saami*-k < pFS *-t, cf. the -t s2 endingin the verb which similarly
< proto-Saami*-k < pFS *-t) goesback to pU *-t, as doesthe plural marker
*-i- (phonologically /j/ < *-j-) in the oblique cases.The genitive plural of
guolli 'fish' was written gulii in Konrad Nielsen'sorthography;its merger
with the accusativeplural (written guliid in KN; cf. present-daypGA guliid,
proto-SaamipG *kuolij and pA *kuolijne) in North Saamiis at least partly
basedon analogy with the singular, where the merger is due to phonetic
development,both proto-SaamisG *kuolen and sA *kuolem giving North
Saamiguoli. Ultimately, the accusativeplural endinggoesbackto the Uralic
partitive/ablativeending*-ti; Saami,Fennic,and Mordva presupposeat least
pFV *-tal*-ta: guliid < proto-Saami*kuolijne < pFS *kalajta).
The illative plural endings-de (asin guliide, giedaide,bohccuide)and-dda
(asin beatnagiidda,oallasiidda) areobscure.In North Saami,they seemto be
basedon the accusativeplural ending(pFSpartitive *-tal*-ta) to which thepFS
illative ending *-sin was attached.Intervocalic *s went to zero, as in other
illatives, and contraction produced(1) the vowel *i in marginal position,
yielding e in the present-daylanguage:pFS *-j-ta-sin > proto-Saami*-j-nin >
-de; somedialectshaveaddedthe same-i that appearsin the illative singular,
yielding guliidii insteadof standardguliide; and (2) the vowel *a in lateral
position(secondarilyshortenedto present-dayA: pFS (-j-ta-sin > proto-Saami
*-j-niin> -ddA.) SouthSaami,however,points backto pFS *-tal*-ta without
any additions,e.g.gieriehtsidie'into the sledges'),but herewe probablyhavea
secondarysyncretismof the illative plural with the accusativeplural dueto the
narrow phonologicaldifferencebetweenthe two. The comitativesingular,the
essive,andthe inessiveplural havecoincidedin SouthSaamiin the sameway.
Inari Saamipointsbackto proto-Saami*-j-nan: pIll kiedaaidv. pA kiedaidof
kieta 'hand', as well as somerelict forms in Lule Saami(e.g. ejduda 'away'
instead of * adojda). There seemsto be no clear way to reconcile these
conflicting reconstructions.The morphology suggeststhat the illative plural
might havedevelopedrelativelylatein proto-Saami.
The locative plural now covers the functions of the former elative and
inessive,but the loss of the morphologicaloppositionin the plural is not due
to phonologicaldevelopmentbut to analogywith the singular.The functions
of the elativewere transferredto the inessive(in FinnmarkSaami,SeaSaami,
and the Finnish Wedge and Karesuandodialects, and in all EasternSaami
68 SAAMIC

idioms) or to the elative(the Kaitum andJukkasjii.rvidialectsof TorneSaami).

The locative plural ending in the standardlanguagegoes back to the pU
locative *-nal*-na, precededby the plural marker.In Kaitum andJukkasjii.rvi,
the case ending is -s (guliis, gieaais instead of standardguliin, gieaain),
continuing the proto-Saami elative *-ste. The incongruity between the
inessivessingularand plural, with the singulara reflex of proto-Saami*-sne
but the plural continuing *-ne, suggeststhat they developedduring proto-
Saami.The elative plural also bearsother signs which suggestlate develop-
ment: in pFS, the suffixation of *-stal*-sta to the plural marker *-j- would
have beenpossibleonly with an epentheticvowel, producing **kati-j-i-sta;
this, in tum, would haveyielded the contractedproto-Saamiform **kietiste,
and not the *kieDajste reconstructedon the basis of Lule, Pite, Ume, and
SouthSaamiforms. In addition, the SouthSaamielative showsan unassimi-
latedplural marker-j- whereasin the inessive,the plural markeris assimilated
to the illabial stemvowel: pIne gietine v. pEla gietijste from giete 'hand'.On
the whole, the interior local casesillative, inessive,andelative plural seemto
have developedduring proto-Saami.This meansthat the symmetrybetween
the singular and the plural declensionin, say, Fennic and Mordva does not
datebackto their commonprotolanguage.
The comitativeplural (guliiguin, gieaaiguin)hasdevelopedfrom a phrase
consisting of the genitive plural plus the word guoibrni 'companion'.The
genitive plural has similar functions in all Saamilanguages;it is also usedas
a comitativeplural in Lule Saami.
10 the present-daySaamiidioms, the partitive is a regularcasein Eastern
Saamionly (e.g. loari Cittorn alged 'sevensons',with the partitivealgedfrom
alge 'son', usedwith numeralshigher than six), but there are relict casesin
WesternSaamias well, and even examplesof the partitive as a regular case
in old Ume Saamitexts. The partitive goesback to the samepFS *-tal*-ta as
that seenin the accusativeplural ending.
As a whole, Saamideclensionsuggestsa pFS systemwith no specificlocal
casesin the plural. The singularhadthe following pFS cases:the grammatical
casesnominative (zero ending: North Saami guolli, giehta), connective(=
genitive/instructive)(*-n: North Saami guoli, gieaa), and accusative(*-m:
North Saami guoli, gieaa); the general(or abstract)local casestranslative
(*-ksi, of Finno-Volgaicorigin> proto-Saami*-ssa: in North Saamiadverbs,
e.g. davas 'to the north', dalvvas 'for the winter'), essive(-nal*-na: North
Saamiessiveguollin, giehtan,andalso adverbslike olgun 'outside'),partitive
(*-tal*-ta: in North Saami adverbs,e.g. oaprnet 'as of old'), and abessive
(*-ptakl*-ptiik, of Uralic origin: in Lule Saamipronouns,e.g. dabdd 'without
it' , in North Saamiadverbslike gahperahtta'without a cap' andpostposition!
adverbhaga - taga 'without'); the specific local casesillative (*-sin: North
Saamiguollai, gihtii with a secondaryending-i), inessive(*-snal*-sna:North
Saamilocativeguolis, gieaas,Lule Saamiguolen,giedan),andelative(*-stal
*-sta: North Saamilocative guolis, gieaas,Lule Saamiguoles,giedas).The
 SAAMIC 69

plural had the following cases: grammatical cases nominative (which

probablyactedasthe caseofthe plural objectas well; plural marker*-1: North
Saamiguolit, gieaat) andconnective(plural marker*-j: North Saamigenitive
plural, KN gulii, gieaai, modem guliid, gieaaid); and general local cases
partitive (*-j-tala: North Saamiaccusativeplural guliid, gieaaid) and essive
(*-j-nala: North Saamilocativeplural guliin, gieaain).
The possessivesuffixes. The possessivesuffix system has undergone
radical simplification in North Saami. The proto-Saamisystem is shown
below (Table 2.5).
The singularsuffixeshadno numbermarker.The dual markerwas *-n. The
plural markerwas *-k, which goesbackeitherto pU *-t or to the *-k attested
in Mordva and Hungarian.The pFS systemwas isomorphic with the proto-
Saamione,the differencesbeingdue solely to soundchangessuchaspFS *ns
> proto-Saami*ss, pFS *i > proto-Saami*a, pFS *ala > proto-Saami*e.
In North Saami,the phonologicaldifferencesbetweenthe three serieshas
been lost. The first- and second-personplural suffixes are attachedana-
logically to the samekinds of stemsas the singularanddual suffixes: guollan
'my fish', guollame'the fish of us two' (insteadof *guollama),andguollamet
'our (plur) fish' (instead of expected*guollimat). The suffix vowel e in
guollame(t) was obtained from lateral positions (e.g. beatnageamt;t'our
dog'), where it is regular in the plural and third-person suffixes. In later
positions,the expectedsuffix vowel would havebeena in the first andsecond
persondual, not the attestede (*beatnagamainsteadof beatnageamt;'the dog
of us two', and *beatnagattainsteadof beatnageattt;'the dog of you two').
Here,too, the plural suffixes(beatnageattt;t,etc., with regulare) servedas the
basis for analogy. In central positions, the second-persondual and plural
suffixes have irregular geminate-tt- in the nominative insteadof expected
single -a- (e.g. beatnageattt;t 'your (plur) dog', instead of expected
geminate the geminate-tt- is regular,however,in the genitiveandother
*beatnagead~t);
oblique cases(proto-Saamipossessivesuffixes d2 *-llDan and p2 *nnek), on

Table2.5 Proto-Saamipossessivesuffix system

Singular Dual Plural

sN 1 *-ma *-ma-n *-me-k

2 *-Da *-Da-N *-De-k
3 *-se *-se-n *-se-k
sA 1 *-ma *-ma-n *-me-k
2 *-mDa *-mDa-n *-mDe-k
3 *-mse *-mse-n *-mse-k
pN and othercases 1 *-na *-na-n *-ne-k
2 *-nDa *-nDa-n *-nDe-k
3 *-sse *-sse-n *-sse-k
70 SAAMIC

which the analogicalchangewas based.In the samecentral positions, the

third-persondual and plural suffixes have irregular single -s- in the oblique
cases instead of expected geminate *-ss-, e.g. beatnageasqt'their dog',
goaaisteasqt'in their hut' instead of *beatnageass~t, *beatnageass~t, *goadisteass~t. *beatnageass~t, The
single -s- in the nominative singular, which has servedas the basis for the
analogicalchange,is regular.
Present-dayNorth Saarni has only a single series of possessivesuffixes
with slightly different forms for the different stresspositions(after a stressed
and after an unstressedvowel). The original nominative plural forms have
disappearedin North Saarni(proto-SaarnipN *nieJDana'my daughters'v. sN
*nieJDama 'my daughter'); the genitive/accusativeplural forms are used
instead, in vocative functions only, e.g. nieiddaidan! 'my daughters!'The
plural illative has the case ending -das-/-ddas-(e.g. gieaaidasan 'into my
hands',beatnagiiddasan'to my dogs') peculiar to North and parts of Lule
Saarnibut not attestedelsewhere.It consistsof the partitive (pFS *-tala) plus
illative (pFS *-sin) suffixes. The NorthlUme Saarnithird-persondual suffix
-skal-skkq(guossiska'the guestof them two') is secondary;its origin is not
clear. Its use in declensionmay be basedon an analogy with its use in
conjugation, or vice versa (manai 'he went' : manaiga 'they two went'
parallel to guossis'hislher guest', guossiska'the guest of them two'). The
vowel points backto proto-Saarni*a « pFS *ala), but this may be due to the
analogical influenceof the correspondingpresent-tensesuffix -ba/-bq « pFS
*pala+n) in verbs.
Pronouns.Pronounsshowby andlargethe samemorphologyasnouns,but
with the architectonicdistinction that pronounsmay also have monosyllabic
stemswhereasnounsare bisyllabic or longer. Somerepresentativeparadigms
aregiven in Table2.6.
In the paradigmof monosyllabicstems,the essiveending is reduplicated
(danin, manin, diinin). The interrogativemii 'what' has separateforms for
genitive (man) and accusative(maid, man); the accusativeform man is used
mainly for definite referents.The illative singularin monosyllabicstemshas
retainedthe s of the suffix and thereforedoes not show contraction(dasa,

Table 2.6 Samplepronounparadigms:'it', 'what','who'

'It' 'What' 'Who'

Singular Plural Singular Plural Singular Plural

N dat dat mii mii mat gii geat

G dan daid man maid gean geaid
A dan daid maid, man maid gean geaid
III dasa daidda masa maidda geasa geaidda
Loe das dain mas main geas geain
Com dainna daiguin mainna maiguin geainna geaiguin
Ess danin manin geanin
 SAAMIC 71

Table2.7 Samplepronounparadigms:'you', 'someone'

'You' 'Someone'
Singular Dual Plural Singular Plural

N don doai dii soam~s

soam~s soapmasat
G du dudno dIn soapmasa soapmasiid
A du dudno dIn soapmasa soapmasiid
III dutnje dutnuide didjiide soapmasii soapmasiidda
Loc dus dudnos dIs soapmasis soapmasiin
Com duinna dudnuin dInguin soapmasnn soapmasiiguin
Ess dunin dudnon dInin soamisin

masa, geasa). The personal pronouns have initial m- in the first person
(mun-mon,moai, mii), d- in the second(don, doai, dii), and s- in the third
(son, soai, sii). The dual personalpronounsoriginally had bisyllabic stemsin
the nominativeas well (*mon6j, *ton6j, *son6j < pFS *munoj etc.), but these
were reducedto monosyllabicforms in !nari, North, and Lule Saarni(North
moai, doai, soai; cf. Skolt mana,etc.). After initial m-, nasalgeminateswere
retained(mun, illative munnje;moai, genitive/accusative munno).In the dual
and plural of the personalpronouns,some casesshow plural morphology
(doai, illative dudnuide; dii, illative didjiide, comitative dfnguin). The
palatalized/nj / in illative singular munnje, dutnje, sutnje has met with no
satisfactory explanation; the fact that Skolt, Kildin, and Ter have non-
palatalizednasals(Skolt mu'nne, tu 'nne, su'nne) could indicatethat therewas
innovation in the west. The illative plurals midjiide, didjiide, sidjiide have
acquireda secondaryillative plural morphology; original forms were based
on the illative singularsuffIx, with contractionin the secondsyllable (proto-
Saarni*mi}in, *ti}in, *si}in < pFS *me-j(-i)-sin etc. > Finnish meihin 'to us'
etc.); Inari Saarni hasexpectedmijjan, tijjan, sijjan, Lule Saarnihas midjij,
tidjij, sidjij with the secondary(singular)illative suffIx -j.
The reflexive pronoun (sN ieS, pN ieia, stem iehca-) is inflected with
obligatorypossessivesuffixesin the oblique cases, e.g. sGA iehcanlieian'my
own', ieiat 'your own', sCorn ieiainis 'with himlherself', Ess iehcanis 'by
himlherself', pA ieiaideaset'themselves'.The illative and locative singular
have suppletivestemsbasedon the adverbsala 'onto the top' and aide 'on
the top', e.g. alccesan (- alccen - allasan) 'to myself' (where c is an
irregular, and s a regular, reflex of the singular illative case ending -s-),
alddan 'with/from myself'.

Conjugation
The verb is conjugatedfor two voices (active and passive,the latter used
mainly to avoid expressingthe agent,whethertransitive or intransitive),four
moods (indicative, conditional, potential, and imperative), two tenses(plus
two compoundones), three numbers,and three persons.There are also a
Table2.8 Sampleverb paradigms

Negativeverb leat 'to be' oaddit 'to sleep' muitalit 'to tell' gul'lot 'to be heard'

Indicative
Present
sl III lean oadan muitalan gul'lon
s2 it leat oadat muitalat gul'lot
s3 11 lea oadda muitala gul'lo
dl ean letne Odde muitale(tne) gul'loje(tne)
d2 eahppi leahppi oaddibeahtti muitaleahppi gul'lobeahtti
d3 eab;;t leab;;t oaddiba muitaleab;;t gul'loba
pi eat leat oaddit muitalit gul'lot
p2 ehpet lehpet oaddibehtet muitalehpet gul'lobehtet
p3 eai leat Oddet muitalit gUl'ojit
Past
sl ledjen Odden muitalin gul'lojin
s2 ledjet Oddet muitalit gul'lojit
s3 le(a)i Odii muitalii gul'lui
dl leimmt;: Odiime muitaleimmt;: gul'luime
d2 leiddt;: Odiide muitaleidde gul'luide
d3 leigg;;t Odiiga muitaleidde gul'luiga
pi leimmt;:t Odiimet muitaleimmet gul'luimet
p2 leiddt;:t Odiidet muitaleiddt;:t' gul'luidet
p3 ledje Odde muitale(dje) gul'loje(dje)
Conditional
sl liveeen oadasin muitaliveeen gul'losin
s2 liveeet oadasit muitaliveeet gul'losit
s3 liveeii oadasii muitaliveeii gul'losii
Table2.8 (Continued)

Negativeverb leat 'to be' oaddit 'to sleep' muitalit 'to tell' gul'lot 'to be heard'

dl livcciime oadaseimm~oadaseimm~ muitalivcCiime gul'loseimm~gul'loseimm~

d2 livcCiide oadaseidd~oadaseidd~ muitalivcCiide gul'loseidd~gul'loseidd~
d3 livcciiga oadaseigg~oadaseigg~ muitalivcCiiga gul'loseigg~ gul'loseigg~
pI livcciimet oadaseimm~t
oadaseimm~t muitalivcCiimet gul'loseimm~t gul'loseimm~t
p2 livcciidet oadaseidd~toadaseidd~t muitalivcCiidet gul'loseidd~tgul'loseidd~t
p3 livcce oadase(dje) muitalivcce gul'lose(dje)

Potential
sl leaccan Odezan muitaleaccan gul'lozan
s2 leaccat Odezat muitaleaccat gul'lozat
s3 leazza Odeza muitaleaZza gul'loza
dl lezze Odeze(tne) muitalezze gul'loze(tne)
d2 leazzabeahtti Odezeahppi muitaleazzabeahtti gul'lozeahppi
d3 leazzaba Odezeab~
Odezeab~ muitaleazzaba gul'lozeab~ gul'lozeab~

pI leazzat Odezit muitaleaZzat gul'lozit

p2 leazzabehtet Odezehpet muitaleaZzabehtet gul'lozehpet
p3 lezzet Odezit muitalezzet gul'lozit

Imperative
sl allon lehkon Oddon muitalehkon gul'lojehkon
s2 ale leag~ leag~ oad~oad~ muital gul'lo
s3 alios lehkos Oddos muitalehkos gul'lojehkos
dl al'lu leahkku oad'du muitaleahkku gul'lojeahkku
d2 al'li lahkki oad'di muitalahkki gul'lojeahkki
d3 alloska lehkoska Oddoska muitalehkoska gul'lojehkoska
Table2.8 (Continued)

Negativeverb leat 'to be' oaddit 'to sleep' muitalit 'to tell' gul'lot 'to be heard'

pI allot lehkot 6ddot muitalehkot gul'lojehkot

p2 allet lehket Oddet muitalehket gul'lojehket
p3 alloset lehkoset Oddoset muitalekoset gul'lojehkoset

Non-finiteforms
Infinitive leat, leahkit oaddit muitalit gul'lot
connegatives:
indic.pres leat oadE muital gul'lo
conditional livcce oadase muitalivcce gul'lose
potential leacca Odes muitaleacca gul'los
imperative leagt;: oadt;: muital gul'lo
Secondimperative lehko Oddo muitalehko gul'lojehko
Action leahkin oaddin muitalan gul'lon
Action essive leamt;:(n) oaddime(n) muitaleamt;:(n) gul'lome(n)
Action locative leamt;:s oaddimis muiatleamt;:s gul'lomis
Gerund leagt;:dettiin oadedettiin muitalettiin gul'lodettiin
Presentparticiple leahkki oadidi muitaleaddji gul'lojeaddji
Perfectparticiple lean, leamas oaddan muitalan gul'lon
Verb abessive leagt;:keahtta oadt;:keahtta muitalkeahtta gul'lokeahtta
Verb genitive oadi muital gul'lo
Supinum ama'n,-t, -s, -me, -de,
-ska, -met, -det, -set
 SAAMIC 75

numberof non-finite forms classifiableaccordingto their syntacticroles as

infinitives, participles,and gerunds.Verb stemsare bisyllabic or longer, with
two exceptions:leat 'to be' andthe negativeverb.
The endingshavedifferent forms accordingto position. There are vocalic
stemsending in a, i, or u (parisyllabic), consonantalstems(imparisyllabic),
and contractedstems,which haveboth a parisyllabicvocalic stem(endingin
d, e, or 0) anda consonantalstem(endingin -j- ). Sampleparadigmsfor North
Saarniaregiven in Table2.8.
Personalendingsand tensemarking.The first andsecondpersonssingular
have the samepersonalendingsin both tenses,and the third personhas zero
(the -a final in muitala is, at least historically, the stemvowel). First person
-n < proto-Saami*-m < pU *-m, cf. the m- of mun '1'; secondperson-t <
proto-Saarni*-k < pFS *-t < pU *-t, cf. the d- of don 'you'.
The remainingpersonshavetrue personalendingsin the pasttense,but in
the presenttenseuse endingsbasedon deverbalnominals (agentives).The
nominal suffixes on which the present-tense forms arebasedare asfollows:

• From pFS *=jala (probably < pU *=j or *=jala, a nomenagentis) were

built
the first persondual, e.g. manne'we two go' < pFS *meni=ja-n, with the
dual marker*-n;
the third personplural, e.g. mannet'they (all) go' < pFS *meni=ja-t, with
the pU plural marker*-t.
• From pFS *=pala « pU *pala, anothernomenagentis)werebuilt
the secondpersondual, e.g. mannabeahtti'you two go', probably< pFS
*meni=pa-ta-n,with the pU non-singularsecond-personending *-tala
andthe dual marker*-n;
the third persondual, e.g. mannaba'they two go' < pFS *meni=pa-n;
the first personplural, e.g. mannat 'we (all) go' < pFS *meni=pa, i.e.
without a markerfor number;
the secondpersonplural, e.g. mannabehtet'you go' < pFS *meni=pa-ta-t,
or possibly, but on conflicting evidence,without the plural marker, i.e.
*men=pa-ta.

The present-tensedual and plural second-personendings presupposea

reduplication of the personalending in proto-Saarni(mid proto-Saarnid2
*mene-Ba-Ui-oa-nand p2 *mene-Ba-ta-oa-k)with subsequentcontractionof
unstressed*-(t)a-oa- in the dual but for somereason- perhapsbecausethere
was no final consonantclosingthe syllable- not in the plural.
In the past tense, the personal endings (the third persons had zero
originally) were attachedto a stem built with the past-tensemarker *-j-,
resulting in contractionwhen the -j- was intervocalic(in the first and second
personssingular andthe third personplural). Examples:mannen'I went' <
pFS *meni-j-i-m, mannet 'you went' < pFS *meni-j-i-t, manne 'they (all)
76 SAAMIC

went' < proto-Saami*manin, with unexpectedfinal *-n insteadof *-k < pFS
*meni-j-i-t, all with the epentheticvowel *-i- to the right of the past-tense
marker.The remainingpast-tenseforms are without contraction:manai '(s)he
went' < pFS *meni-j, manaime'we two went' < pFS *meni-j-mi-n, manaide
'you (two) went' < pFS *meni-j-ti-n, manaimet'we (all) went' < pFS *meni-
j-mii-t, manaidet'you (all) went' < pFS *meni-j-tii-t. The plural markermay
havebeennot *-t, but *-k, as suggestedby Mordva (palatamk 'we kiss') and
Hungarian(vtirunk 'we wait'). The past third persondual has the secondary
peculiar peculiar to the WesternSaamilanguages.In the east(Inari
ending -gal-gg~,
and somedialectsof Skolt), the forms point backto a somewhatmore regular
pFS form *meni-j-n-i-n > proto-Saami *manajnan (note, however, the
reduplicateddual marker n), yielding moontitiin 'they (two) went' in Inari
Saami.Similar forms have also beenattestedin someNorth Saamiidioms of
the nineteenthcentury, e.g. Varanger Sea Saami suhtaina 'they (two) got
angry'. The origin of the suffix -ga/-ggq (proto-WesternSaami *-Ga) is not
clear.
In the imperative,the suffix of the first personsingularandall third-person
suffixes areidenticalwith the corresponding possessivesuffixes,andthe first-
personplural and dual suffixes are the sameas in the presentindicative. The
secondpersonsingularis zero (the original suffix *-k as in mana! 'go!', < pU
*meni-k, was the old imperativemarker).The secondpersondual and plural
show the Uralic non-singularsecondpersonending *-talii and the number
markers*-n (dual) and *-t or *-k (plural). Unlike the correspondingpresent
indicativeforms, theselack the old agentivesuffix *=palii.
In Lule, Pite, Ume, and SouthSaami,the negativeverb has separatetense
forms (e.g. Lule ittjiv - idtjiv 'I did not', ejma 'we did not') usedwith the
indicative connegative(e.g. Lule ittjij boade '(s)he did not come', contrast
synonymousNorth ii boahttin,built with the perfectparticiple.)The affricate
in ittjiv (Southidtjim etc.) goesbackoneway or the otherto the pU past-tense
marker *-sj(alii)- seenin South Estonianesin, Nenetsny{drjJmcyO 'I did not';
Saami presupposespFS *-ns-i-, but if there is a sporadiclengtheningof the
consonantcentreas suggested(*-zL > *-zjzL) the more regularpFS *-sL can
be reconstructed.
Thecompoundtenses,perfectandpluperfect,areformedwith thecopulaand
theperfectparticiple: leanboahttin 'I havecome', ledjenboahttin 'I hadcome'.
The conditional and the imperative have a compoundperfect only: livccen
boahttin 'I would have come', lehkos boahttin 'may (s)he have come'. The
potential on the other hand hasboth perfect and pluperfect,at least in some
dialects:leazztiboahttin '(s)hemight havecome',pluperfectleaccaiboahttin;
suchdialectsalso havepastpotentialforms for the copula.For more complex
temporal reference,even double perfects and pluperfectsare used, e.g. lei
leamasboahttin '(s)hehadcomealreadyon anearlieroccasion', seep. 79.
Mood markers.Stem-formationin the imperativeis basedon the following
pFS imperativemarkers:
 SAAMIC 77

1 *-0- usedin first and third personsin parisyllabic stems,e.g. dl mannu

'let us (two) go' < proto-Saami*manon, s3 mannos'let him/her go' <
proto-Saami*manose;
2 *-k usedin the secondpersonsingular, e.g. mana 'go!' < proto-Saami
*manak;
3 *-ko- used in the first and third personsof imparisyllabic stems, e.g.
oahpistehkos'may (s)heguide' < proto-Saami*oahpesteakose;
4 *-ke- usedin the secondpersondual andplural in parisyllabicstems,e.g.
d2 oahpisteahkki'guide! (you two)' < proto-Saami*oahpesteakken<
oahpistehket p2 oahpistehket'guide! (you lot)' <
mid proto-Saami*~ppastak:anan,
proto-Saami*oahpesteakenek;
5 *-13 usedin the secondpersondual and plural in parisyllabic stems,e.g.
d2 manni 'go! (you two)' < pss *mannen,with a contractedvowel from
earlier *-ane- in the secondsyllable, and subsequentgeminationof the
centralconsonant,p2 mannet'go! (you lot)' < proto-Saami*mananek.

Of thesefive manifestationsof the imperativemarker, thosebasedon *-k

and *-ke- areclearly the oldestandgo backdirectly to pU *-k and *-ka/ii. The
roundedvowel of *-ko- is probablydue to Fennicinfluence,cf. Finnishanta-
ko-on 'may (s)he give'. Of thosewithout the stop *k, at least *-0- seemsto
have arisen,throughfaussecoupe,from forms containingthe stop; this left
the vowel *0, which replaces the stem vowel, to fulfil the function of
imperative marker. The resulting forms were thus of the canonic type for
Saami,viz. parisyllabic.As for Ume, Pite, Lule, and North Saami,the forms
with *-13 seemto have an analogicalhistory in the WesternSaamilanguages,
althoughthey are usually thoughtof as having the personalendingsattached
directly to the stem.
The potentialmarker -z-/-zz-,-cc- (bi5tleian 'I may come' from boahtit 'to
come',muitaleaccan'I may tell' from muitalit 'to tell') goes backto proto-
Saami*-nzja- and ultimately to pFS *-nsji- (> proto-Fennic*-njsji- + *-j- >
late proto-Fennic*-jsi- > the Finnish conditional marker -isi-). It may have
the same origin as the nominal diminutive suffix of the same pFS shape
(*=nsji).
The conditionalhastwo markers.For North andEasternSaamiit is proto-
Saami*-k<9i- (in which the *i is originally a contractedvowel containingthe
past-tensemarker*-j-), yielding North Saamiforms suchas oailtisin 'I would
sleep' and muitalivccen 'I would tell'. For westernNorth Saami and to the
south of it (overlapping with -si-/-vcce- in West Finnmark), the marker
consistsof the present-tense marker *-k (probablyalso found in the present-
tenseconnegative)and the contractedpast-tensestem of the copula (proto-
Saami *li-, proto-Saamiinfinitive *leanek). Thus we have western North
Saamioailtilin 'I would sleep' « proto-WesternSaami*oaMvlim; in North
Saami this type is confined to parisyllabic stems).A completeconditional
form of the copula may also be suffixed, e.g. North Saami oailtilivccen 'I
78 SAAMIC

would sleep',Lule Saamigulaluluv (- more frequently, the compoundluluv

gullat) 'I would hear'. The marker *-kCii- is identical with the conditional
marker -ksi- found in Estonian(tuleksin 'I would come'); it seemsto have
originally beenidentical with a continuativesuffix, cf. pFS *-(i)ksii- > North
Saami=as- (vuojasit 'to drive along' form vuodjit 'to drive'), Finnish=(e)ksi-
(kanneksia 'to carry around', from kantaa 'to carry'). Instead of the
compoundperfectwith the conditional morpheme(livccen boahtan 'I would
havecome')an equivalentconstructionconsistingof a past-tenseform of the
copulaplus the infinitive is alsoused(ledjen boahtit).
Voice. The Saami passive differs from most passivesin that its main
function is not so much to topicalizethe patientas it is to submergethe agent
(or logical subject) of the action expressedby the verb, and is accordingly
usedwith intransitive as well as transitive verbs, e.g. el'lojuv'vui 'life went
along, peoplelived' from eallit 'to live'). Somescholarshold that the Saami
passivebelongsto the realmof derivationand not to that of inflection.
The markeris -(oj)uv'vo(j)- in North Saami,e.g. dolvojuv'vot'to be taken
somewhere'(from doalvut 'to take, lead'), muitaluv'vot'to be told' (muitalit
'to tell); such forms are conjugatedlike gul'lot 'to be heard', cf. above.
Historically, the passivemarker contains two elements.The fIrst element,
proto-Saami*-uja- (from earlier *-oje-) may originally havebeena frequent-
ative suffix; or it may historically be the sameas the secondelement,proto-
Saami*-uova- (from pFS *-ov(i)-), which was originally a reflexive suffix.
Cognatesfor the reflexive suffix are known from Mordva (-v- in Erzya
nieja=v-oms'to be seen',nieje-ms'to see'),from Mansi (Vogul; -w- in totawe
'he is brought'),and Hungarian,whereit survivesas the labiality in reflexive
suffIxes,e.g. u in epul-ni 'to be built' .
Nominalformsof the verb. The infinitive marker-t (e.g. oaililit 'to sleep')
goesbackto proto-Saami*-Dek. It is historically identical with the markerof
the Finnish fIrst infInitive (saa-da 'to get',juos-ta 'to run'), with largely the
samefunctions. Both go back to pFS *-takl*-Uik, which consistsof the pU
deverbalnoun suffix *=taia plus lative *-k, itself possiblyof pU vintage.The
connegatives(of the verb oaililit 'to sleep': indicative oail€!, conditional
oailase,potentialMes, imperativeoad€!) endedin the proto-Saamisuffix *-k
which was probably the same as the elementfound in the secondperson
singularimperative« pU *-k). Note that in the negativeimperative,mood is
encodedin the negative verb. The second imperative (/ehko, Milo etc.)
exhibits the same stem and has, accordingly, the samehistory as the third
personimperative.The action forms, e.g. nominativetallin in ahCi tal/in girji
FATHER.sGWRITES.ACTIO BOOK 'a bookwritten by father' arebasedon
the pU deverbal action noun suffix *=maia, to which the essive (tallimin
'whenwriting') andlocative(tallimis 'from writing') suffixes were attached.
The gerund,e.g. muitalettiin 'while telling', is basedon the pU deverbalnoun
suffix *=ntaia plus the comitativesuffix *-jan/a (cf. declension,above).
The presentparticiple (boahtti 'comer,onewho comes',with a contracted
 SAAMIC 79

lateral vowel, from boahtit 'to come') has acquiredverbal featuresbut is

originally clearly a derivational rather than a true conjugational form in
Saarni.Its suffix goes back to pFS *=ja/a, cf. Finnish tuli=ja 'comer' from
tule- 'comes')and is probably of pU origin. The perfectparticiple (mannan
'gone',from mannat'to go', dolvon 'taken',from doalvut 'to take,lead',lean
-leamas'been',from leat 'to be') hasthe marker-n (-masoccursin leamas
only, but there is evidencefor an older -m, and eighteenth-century-maf in
North Saarni bisyllabic stems).This -n goes back to proto-Saarni*-ma or,
with the diminutive suffix, *-ma-nzja. Proto-Saarni*-ma-nzja, in tum, goes
back to pFS *-mi-nsii (cf. Finnish mene-mi-se-,the oblique stem of
meneminen'the act of going', from mene-'goes').In pFS *-mi-nsji, the *-mi
is of Uralic origin and *-nsii is the diminutive suffix.
The supinumof the negativeverb (aman 'lestI do', amat 'lestyou do', etc.)
takesthe infinitive as its argument:(amasgahccat'lest it fall, so that it does
not fall'); functionally it resemblesthe infinitive of other verbs. Its markeris
-ma-, which is identical with the markerof the proto-Saarniperfectparticiple
*-ma(nfja), andoccursonly with personalsuffixes attached.In somedialects,
the supinumhas evolved into a conjunction,amas,which occurswith finite
negativeconstructionssuchasamasii gahca 'so that it will not fall' .
The verb abessive and genitive are gerunds: oadl!keahtta 'without
sleeping', oadi 'by sleeping'. The former derives from the abessiveof a
deverbalnoun formed with the pFS suffix *=kka/a. The latter originally had
the marker *-n (preservedin South and Inari Saarni), which has similar
functions in Finnish (cf. the second infinitive instructive kantaen 'by
carrying', probably from pFS *kanta-ta-n) and Mari (e.g. kanden 'by
carrying'). It seemsto derive from the pU connectivesuffix *-n (with more
specificfunctions suchas genitive,instructive,lative, anddual, all expressing
different kinds of connectivity).

Word Formation

Derivation
Derivation is possible both within and across part-of-speechcategory
boundaries.A substantialnumberof stemsare inflectible as both nounsand
verbs without any derivational operations,e.g. biegga(-) 'wind; (the wind)
blows'. There are also highly productivederivationalsuffixes which operate
only within a given category.The diminutive suffix, for example,which is all
but obligatory after unnaluhca 'little', is restricted to nominals (boazu
'reindeer',unna bohccos'little reindeer',guolli 'fish', unna guoMs 'little
fish'). In parallel fashion, there are causative,frequentative!continuative,
diminutive, and momentaneoussuffixes which are restricted to deverbal
derivation,e.g. vuodjit 'to drive', vuojihit 'to makeX drive', vuojasit 'to keep
driving', vuojestit 'to drive a little'. The use of frequentative!continuative
(often with conatory nuances) and momentaneousverbal derivatives is
80 SAAMIC

normal, even obligatory, wheneverthey are available and appropriatein

relation to the act to be describedby the verb; this makestheir use close, if
not equivalent,to an aspectualsystem.
A highly productiveinstanceof cross-categoryderivation isthe systemof
deverbal nouns (e.g. action nouns with -nl-pmi and -mus such as oaddin
'sleeping' and oaddimus'having to sleep',and actor nouns such as oad'di
'sleeper')and denominaladverbssuch as fiinet 'badly' (from fuotni 'bad'),
njozet 'slowly' (from njoahci 'slow'). There are also some productive
derivational suffixes borrowed from Fennic such as the deverbal adjective
ending =meahttun 'not X-ing', from Finnish =matonl=miiton, e.g. North
Saarni duhtameahttun'dissatisfied' from duhtat 'to be satisfied'), and the
denominaladjective ending =las 'belongingto or having X', from Finnish
=llise-, e.g. North Saarnisoajalas'winged',from soadji 'wing'.
Most derivational suffixes have different forms after stressed and
unstressedvowels. Thus the causativeis -h- to the right of the unstressed
secondsyllablein vuojihit 'to makeX drive', from vuodjit 'drives',but -htt(i)-
to the right of the stressedthird syllablein muitalahttit 'to makeX tell', from
muitalit 'tells'; both -h- and -htt(i)- continue pFS *=tta/a-. Similarly the
frequentativeis -d- in vuojadit 'to swim around', from vuodjat 'to swim', but
-dd(a)- in muitaladdat 'to keep telling', from muitalit 'to tell'; both -d- and
-dd(a)- continuepFS *=(i)nti-.
The most commonderivationalsuffixes, markedwith equals(=) sign, are
exemplifiedbriefly in the paragraphsthat follow.

1 Deverbal nouns may be subclassifiedinto broad functional/semantic

categoriesas nounsdenoting
(a) the act itself: oaddi=n 'sleeping',muitalea=pmi'storytelling' « pU
*=ma/a); barg=u 'work' (bargat 'to work'), < pFS *=0, perhaps
from a Finno-Volgaic*=v); goarru=mus'havingto sew' (goarrut 'to
sew'), < Fennic *=mus/*=miis; bdvkk=as 'banging' (bdvkit 'to
bang'),< pFS *=is;
(b) the actor: oad'di 'sleeper',muital=eaddji 'storyteller' « pFS *=ja/a
< pU *=jV); solo=n 'one who is eagerto pick his/herteeth' (soallut
'to pick one'steeth'),< proto-Saarni*=nja;
(c) instrument: loavdda (sG loavda=ga) 'tent-cloth' (loavdit 'to cover
the tent'), golgad=at (sG golgad=aga) 'drift-net' (golgadit 'to fish
with a drift-net'), < pFS *=ik); vuoja=n 'vehicle' (vuodjit 'to drive')
< proto-Saarni*=nja; vuoidd=as 'ointment' (vuoidat 'to smear'),<
pFS *=is; jeara=ldat 'question' (jearrat 'to ask'), < proto-Saarni
*=lDlik < pFS *=l=ta/a=ik;
(d) result:juog=us 'section'(juohkit 'to divide'), < pFS *=oksi < Finno-
Volgaic *=w(i)ksi; cala (sG calla=ga) 'writing' (callit 'to write'), <
pFS *=ik;
(e) object: vuoss=us'somethingto be cooked' (vuos'sat'to cook') , <
 SAAMIC 81

pFS *=oksi < Finno-Volgaic *=v(i)ksi; borra=mus'food' (borrat 'to

eat'),< Baltic-Fennic*=mus/*=mtis;
(f) place : njuova=hat 'butchering place' (njuovvat 'to butcher'), <
proto-Saami*=ttak < pFS *=ttala- < pU *=ktala + pFS *=ik;
(g) conditions: oainn=tidat 'seeing' (oaidnit 'to see'), < proto-Saami
*=fulDak < pFS, pU *=ntala + pFS *=ik.
2 Denominalnouns:the chief subdivisionsare
(a) diminutive: guold=s 'little fish' (guolli 'fish'), < pFS *=nsii;
(b) property: guhkk=odat'length' (guhkki 'long'), < pFS *=oti < *-voti
+ *=ik; allat=vuohta 'height' (allat 'high'), < proto-Fennic*=voti;
(c) material: bani=s 'materialfor tooth of a rake' (batni 'tooth'), < pFS
*=ksi < Finno-Volgaic *=ksi;
(d) person:suopml!=las'Finn' (suopma'Finnish'),< Fennic*=lalruse-;
(e) group: golmma=s 'group of three' (golbma 'three'), < proto-Saami
*=s < pFS,Finno-Volgaic *=ksi;
(f) order: njealja=t 'fourth' (njealljl! 'four'), < proto-Saami*=nna <
pFS *=nti < pU *=mtala;
(g) conditions: balvv=tidat 'cloudy weather' (balva 'cloud'), < proto-
Saami*=fulDak < pFS,pU *=ntala + pFS *=ik.
3 Deverbaladjectives:
(a) negation:duhta=meahttun'dissatisfied';
(b) inclination: soddj=il 'flexible' (sodjat 'to bend [intr]'), < mid proto-
Saami*=(ja)=la < pFS *=jala + *=lala; tlrro=las 'apt to cry' (Cierrut
'weeps'),< proto-Saami*=las < pFS *=lala + *=is; suhtte=s'quick-
tempered' (suhttat 'to get angry'), < proto-Saami*=i<9Cie < pFS
*=jala + pFS *=cjcjal=cicia;
(c) susceptibility:jugahahtti(juhkat 'to drink'), < proto-Saami*=tteatte
< pFS *=ttala- + *=jala < pU *=ktala + *jv.
4 Denominaladjectives:
(a) lack: guoll!=heapml! 'fishless' (guolli 'fish'), < pFS, pU *=ptala +
*mala;
(b) negation: buhtis=meahttun 'unclean' (buhtis 'clean'), < Fennic
*=malat%n;
(c) abundance:geaaga=i 'stony' (geaagi 'stone'),vuddji=i 'rich in fat'
(vuodja 'fat, oil') < mid proto-Saami*=IJa < pFS, pFU *=IJaia (>
Finnish=va/ii);
(d) character:jahkd=sas'yearly' (jahki 'year'),< proto-Saami*=sanzja
< pFS *=si=nsii; (guhkes)juolgg=at'(long-)legged'(juolgi 'leg'), <
proto-Saami*=(a)IJoa<? pFS *=IJki;
(e) comparativity: guhki=t (sG guhki=bu) 'longer' (guhkki 'long'), <
proto-Saami*=rnBe < pFS,pU *=mpala;
(f) superlativity: guhki=mus'longest',< pFS *=moksi.
82 SAAMIC

5 Deverbalverbs:
(a) causative:goaru=h-it 'to make X sew' (goarrut 'to sew'), < pFS
*=tta/a- < pU *=kta/a-; nohkka=d-it 'to put X to sleep'(nohkkat 'to
fall asleep'),< proto-Saami*=oe- < pFS, pU *=ta/a-; buori=d-it 'to
make X better' (buorranit 'to become better'), < proto-Saami
*=moa- < pFS *=m=ta/a-< pU *=mi- + *=ta/a-;
(b) reflexive: basa=d-it 'to washoneself(bassat'to washX'), < proto-
Saami*=noa- < pFS *=nti-; geassa=d-it'to withdraw (intr)' (geassit
'to draw'), < proto-Saami*=oa- < pFS *=ti-;
(c) reciprocal:dovdda=d-it 'to know one another'(dovdat 'to know'), <
proto-Saami*=noa- < pFS *=nti-; oaidn=al-it 'to see each other'
(oaidnit 'to see'),< proto-Saami*=ala- < pFS *=ili-;
(d) momentaneous:colga=d-it 'to spit once' (colgat 'to spit'), < proto-
Saami*=oe- < pFS *=ta/a- < pU *=ta/a-; njuik=e-t 'to jump once'
(v. njuikut 'to jump many times'), < mid proto-Saami*=aja- < pFS
*=iji-; bdvk=al-it 'to bang once' (v. bdvkkuhit 'to bang again and
again'),< proto-Saami*=ala- < pFS *=i1i-;
(e) subitive: borra=l-it 'to eat quickly' (borrat 'to eat'), < proto-Saami
*=le- < pFS, pU *=la/a-; njuike=st-it 'to jump quickly' (njuiket 'to
jump once'), < proto-Saami*=ste- < pFS *=sta/a-; javkk=ih-it 'to
disappearquickly' (javkat 'to disappear'),< proto-Saami*=ette- <
pFS *=tta/a- < ? pU *=kta/a-; bdrgg=dd-it 'to cry out' (bdrgut 'to
cry'), < proto-Saami*=i1rlDa-;
(t) frequentative:lahp=ad-it 'to lose(manyobjects)'(tahppit 'to lose'),<
proto-Saami*=(a)noa- < pFS *=(i)nti-; cohkan=adda-t'(many) sit
down' (cohkanit'to sit down'), < proto-Saami*=anoa-< pFS *=inti;
bdhc=al-it 'to shoot(manytimes)'(bdhCit 'to shoot'),coaskudallat'to
cool (manyobjects)'(coaskudit'to cool [one object]'),< proto-Saami
*=ru.a- < pFS *=ili-; luodd=u-t 'to keep splitting' (luddet 'to split
once'),< proto-Saami*=6- <? pFS *=0-; geas=as-it'to be pulling at
leisure'(geassit'to pull'), < proto-Saami*=acji9a-< pFS*=(i )ksji-;
(g) continuative: cucco=d-it 'to be standing' (cuof1.ut 'to stand'),
loga=d-it 'to be reading' (Iohkat 'to read'),< proto-Saami*=(a)noa
< pFS *=(i)nti-; bargg=ild-it 'to work at leisure' (bargat 'to work'),
< proto-Saami*=eloe-;
(h) diminutive: att£!=st-it 'to give a little' (addit 'to give'), < proto-Saami
*=ste- < pFS *=sta/a-;
(i) conative:bokt=al-it 'to try to wake X up' (boktit 'to wake X up'), <
proto-Saami*=ru.a- < pFS *=ili-; occo=d-it 'to try to obtain' (oaf1.ut
'to obtain'), ribadit 'to try to wrench' (rihpat 'to wrench'),< proto-
Saami*=(a)noa-< pFS *=(i)nti;
G) inchoative:lohkagoahtit'to beginto read' (Iohkat 'to read'),< proto-
Saami*=Goate- < pFS, Finno-Volgaic *=kata-/*=kata-; buol'lat 'to
catchfire' (buollit 'to bum'),<mid proto-Saami*=je-< ?pU*=j-.
 SAAMIC 83

6 Denominalverbs:
(a) causative:nama=h-it'to name'(namma'name'),<pFS,pU *=kta/a-;
(b) translative:buomi=n-it 'to becomebetter' (buorr,! 'good'),< proto-
Saami *=ma- < pFS, pU *=mi-; buoid=u-t 'to becomefat' (buoidi
'fat'), < proto-Saami*=6-< pFS *=0-; bealkh=uvva-t'to become
deaf (bealj,!heapm,!'deaf),< proto-Saami*=uova-< pS *=ovi-;
(c) essive: bodnj=ti-t 'to be twisted' (botnji 'a twist'), < proto-Saami
*=asia-;
(d) instrumental: niibbti=st-it 'to use a knife' (niibi 'knife'), < proto-
Saami*=sta-< ? pFS *=sti-;
(e) sensive:fasttti=s-it 'to considerugly' <Jasti 'ugly'), < proto-Saami
*=&&a- < pFS *=ksii-.
7 Denominaladverbs:
(a) state:rtiiggi-l 'with a hole' (rtiigi 'hole'),< proto-Saami*=lne < pFS
*=l(a/a) + *=ta/a); muohtanaga'with snow on it' (muohta 'snow'),
< proto-Saami *=naGan < mid proto-Saami*=naGi!n < pFS, pU
*=na/a+ pFS,pU *=k + (-i-) *-n;
(b) reciprocity: giehta=laga 'hand in hand' (giehta 'hand'), < proto-
Saami*=laGan < pFS *=li- + pFS,pU *=k- + (-i-) *-n-*-j;
(c) manner:ceahpi=t - cehpe=t 'skilfully' (ceahppi 'skilful'), < proto-
Saami*=kte < pFS *=kta/a < Finno-Volgaic *=kta/a.

Compounding
Compoundingis confinedalmostexclusivelyto nominalbases.Examples:

muorra 'wood' + bihttti 'piece' ltiibi 'bread'+ vtiibmil 'fond'

muorrabihttti 'pieceof wood' ltiibevtiibmil 'fond of bread'

vtirri 'hill' + cohkka'top' rukses'red [attrib]' + +nierat

vtirr,!cohkka 'hilltop' 'havingX cheeks'
ruksesnierat'red-cheeked'

eatni 'mothersG' + giella 'language' golmma'threesG' + +jahktisas'X

eatnigiella 'mothertongue' numberof yearsold'
golmmajahktisas'three-year-old'

ovda+ 'fore-' + juolgi leg, foot' coila 'through'+ cuovgi 'shining'

ovdajuolgi 'foreleg' cadacuovgi'transparent'

vuost,!+ 'which is against'+ biegga eahp,!+ 'un-' cielggas'clear'

'wind' eahp,!Cielggas'unclear'
vuost'!biegga'headwind'
84 SAAMIC

bealli 'half + +mielat 'havingan X

mind'
beallemielat'mad'

The verb base+njaddit 'to tastelike X', which doesnot occur independ-
ently, takes as its modifier component substantivesor adjectives in the
attributive form: guollt;+njaddit 'to tastelike fish' (guolli), njuoska+njaddit
'to taste raw' (njuoska, attributive form of njuoskkas'raw'). The deverbal
inchoativesuffix =goahtit comescloseto being a verb base:manna=goahtit
'to beginto go' (manna 'to go').
Calquing on Scandinavianconstructionshas producedsomeverbs which
occur with an adverb modifier, e.g. bajasgeassit'to educate'(bajas 'up' +
geassit'to pull', cf. Norwegianoppdra).

Syntax
Syntacticrelationsareexpressedby morphologicalmeans(e.g. case-marking,
derivationalendings)and,- mostly within nounphrases- by word order.The
orderingof main constituents(subject,verb, object, adverb)is guidedlargely
by pragmaticprinciples. The basic order is SVO, as in ahcCi (S) oinnii (V)
Niillasa (0) 'father saw Nils', matki (S) lea (V) guhkki 'the journey is long
- it is a long journey'. However,infinitival and participial constructionstend
towardsthe order SOY, e.g. ii diktan mu dan (0) oaidnit (V) 'did not let me
seeit', with dan 'it' coming before oaidnit 'to see'.In South Saami,SOY is
the rule in finite casesas well: dah (S) maanah(0) utnieh (V) 'they have
children'. Noun phraseconstituentsgenerally have a modifier-headorder,
e.g. dat nuorra olmmai 'this young man'.
The subjectis not an obligatory constituent.First- and second-person verb
forms do not require a subject pronoun, and third-person pronouns are
optional in subordinations,e.g. oidnen Biera 'I saw Peter',oidnetgoBiera?
'did you seePeter?',go Biera boilii, de muitalii, ahte ... 'when Petercame,
he said that .. .' Formal subjectsare not required in caseslike arva 'it is
raining', lea coaskkis'it is cold', althoughthey often do occur, as a result of
Scandinavianinfluence,e.g. dat lea coaskkis.
Questionsbeginwith the focalizedphrase.Thus we havegos don boailat?
'wheredo you comefrom?', with the questionword gos '(from) where?'in
initial position (in the reply, the adverbwhich suppliesthe information is in
normal final position, e.g. mun boailan varis 'I come from the hills'. In
oidnetgodon Biera? 'did you seePeter?'it is the verb thatis focalized;notethe
questionmarker-go and comparethe simple declarativemun oidnenBiera 'I
sawPeter', with regularSOY order.Answersrepeatthefocalizedquestionword
of a yes/no question, with appropriatemodifications concerning personal
referenceandotherdeixis,e.g.- oidnetgoBiera?- oidnen'- did you seePeter?
- yes', andsupplythe appropriateinformationin others,e.g.- gosdon boailat?
 SAAMIC 85

-Anaris '- wheredo you comefrom? - Inari', with the answerin the locative
casein keepingwith thequestionword gos.
On the other hand, if we take the morphosyntacticpropertiesof the head
to be diagnostic,then ahead-modifier(or base-argument) ordermay be seen
in numeralconstructions(with numeralshigher than okta 'one') and in verb
complexes.For example,if the noun phrasegolbmadalu 'threehouses'acts
as a subject in the sentencegolbma dalu ledje buollan 'three houseshad
burneddown', golbma 'three sN' is the headand dalu 'housesG' modifies
it: golbma ledje buollan 'three had burned' is syntactically acceptable
whereas*dalu ledje buollan is not. The sameholds for the verbal predicate
in the sentencemun ledjen boahtan 'I had come': ledjen is the baseto which
the participle boahtanis an argument.Noun phrasesare insoluble as a rule,
i.e. they cannot be intersectedby noun-phrase-external elements(verb or
sentencemodifiers).
Unlike nounphrases,verb complexessuchas compoundtensesandmoods
or other constructionsconsistingof a finite and a non-finite form cannotbe
consideredto constitutea phrasein the samenarrow sense:their components
tolerate intervening elementsand their order is not strict. They receive the
sameanalysisas other combinationsof finite and non-finite verb forms: the
non-finite form is subordinatedto the finite one and representsa separate
predicate.For someverbs, such as sahttit 'may' or leat 'be' (as copula),the
subordinationhasno distinct label becauseother argumentsare not possible,
e.g. sahtanvuolgit 'I may leave',lean vuolgan 'I haveleft', ledjen vuolgit 'I
would have left'. For others,the non-finite form may function as an object
(e.g. mun mahtan vuodjat 'I can swim', cf. mun mahtan darqgiela 'I can
[speak] Norwegian') or as an adverbial (manan cohkkat 'I go and sit', cf.
manan stohpui 'I go into the house'; ledjen vazzimen'I was walking', cf.
ledjen stobus'I wasin the house').
SouthSaarnican havepurely nominal predicates:Laara saemie'Lars is a
Saarni',dihte bt'leries '(s)heis old', Piere gaetesne'Peteris in the hut'. Some
caseshavealso beenattestedin Ter Saarni:aa'jtte si'zn saa'hp/i1Jg'therewas
a mouse(saa'hp/i1Jg)in the storehouse(aa'jtte si'zn)'. In North Saami,the
perfect participle leamascan be used alone as a predicativeverb without a
guhkki gili '(s)hehasbeento the otherend(don
finite form: dat leamasdon geahc~n
guhkki of the village'.
geahc~n)

Morphological Marking
The caseof subjectandpredicativenoun phrasesis the nominative:addja (S)
boahta 'grandpais coming',dat matki (S) leaguhkkioall~ guhkki (PR) 'this journey
is fairly long', soappit(S) ledje guhkit (PR) 'the sticks werelong'. The object
of most verbs is in the accusative:oidnen adja (0) 'I saw grandpa' (sN
addja). In South Saarni,there are also plural nominativeobjects,originally
used, it seems,with indefinite referents: aehtjie (S) treavkah (0) dorjeme
'fatherhasmade(a pair of) skis'.
86 SAAMIC

There are also a number of verbs which take argumentscognitively

equivalent to objects (i.e. which are not predicatesor adverbials or other
circumstantialmodifiers) but which areput in casesotherthanthe accusative.
Examplesinclude liikot 'to like', which takes objects in the illative (mun
liikon birgui 'I like meat'),ballat 'to fear', which takesthe locative(dat balla
guovzzas'he is afraid of the bear'), and hilbosit 'to tease',which takes the
comitative (manat hilbosedjeadjain 'the children teasedgrandpa').These
havebeencalledrection adverbialsand,recently,objects.
The first argumentof habitive constructions,correspondingto the subject
in many languages,is in the locative, the secondin the nominative,and the
verb is the existentialleat:Lasseslea beanaLARS-Ioc IS.s3presDOG 'Lars
has a dog'. In South Saami,the first argumentis in the genitive insteadof a
local case,e.g. Laaran bienje 'idem' (note the zero copula),or the transitive
verb utnedh'to have'is used,e.g.Laara bienjemLitna LARS accusativeHAS-
s3pres;this constructionis also found in Vme, Pite, and Lule Saarni. The
secondargumentin Lasseslea beanahasbeenanalysedas a predicative;the
first can be seenas a specialkind of subject.The sameregardsalso the first
argumentof existentialsentences,which is normally in the locative: garddis
ledje ollu bohccot 'there were many reindeer (bohccot) in the corral
(garddis)' .
The agentof causativeverbsandof verbsendingin -halla- is in the illative,
e.g. Biera goaruhii Birehii gapmagiid'Peterhad Berit sew him someshoes',
Mahtte borahalai beatnagii 'Matthew was (unfortunateenoughto get) bitten
by a dog' (beatnagiiis sIll of beana'dog').
The adverbialis markedby specialderivativeendings(cf. p. 83) and by a
numberof cases.Chief amongtheseare:

• the genitive, as in dat boahtd don beaivve '(s)he comesthe day after
tomorrow', addja vealla soggai calmmiid 'grandpais lying with his
eyes (calmmiid, pG of calbmi 'eye') towards the wall (soggai, sIll of
soggi 'the spacenearestthe wall');
• the illative, as in mii manaimetAnarii 'we went to Inari', dat bahce
Avvilii 'they stayedin Ivalo';
• the locative, as in Mahtte bOtlii Ldhpltluobbalis 'Matthew came from
LahpQluoppal';
• the comitative, as in mun vuojan dainna biilain Roavl!njargii 'I shall
drive that car to Rovaniemi';
• the essive, as in ii galgga bassin bargat 'one should not work on a
Sunday'(impersonalthird personsingularin ii galgga 'oneshouldnot').

A number of postpositionalconstructionsare also possible,e.g. busI sa

njuikii beavddiala 'the catjumpedonto the table', mii leimml!tjogaaide 'we
wereon the river'.
In noun phrases,a noun attribute is in the genitive, e.g. adja soabbi
 SAAMIC 87

'grandpa'sstick' (sN addjd 'grandpa'),beavddf! vuolde 'underthe table' (sN

beavdi 'table'), and an adjective attribute is in the attributive form, if
available,e.g. ceahpf!s duojdr 'a skilful craftsman'(ceahppi 'skilful'). Some
attributesare in (partial) congruencewith their head;they will be dealt with
in the sectionon the nounphrase,below.

Noun Phrase
Pronoun attributes, genitive attributes, numerals, and adjectival attributes
come beforethe noun: ddt soabbi 'this stick', adjd soabbi 'grandpa'sstick',
guoktf! soappi (sG; seebelow) 'two sticks', andguhkf!s soabbi 'a long stick'.
If severalof them are present,their order is that theseexamples:ddt adjd
guhkf!s soabbi 'this long stick of grandpa's',ddt ddjd guoktf! gUhkf!s soappi
'thesetwo long sticks of grandpa's'.Constructionscorrespondingto post-
positional ones in other languagesare of the genitive/attributive type in
Saami, e.g. vdrf! vuolldi '(going) under the hill', borgga siste 'in a snow
flurry': the adverb/postpositionis in the morphologicalform requiredby the
syntacticfunction andis thus,by definition, the headof the phrase.
The demonstrativepronouns,the numerals,andcertainindefinite pronouns
expressingan approximatenumber(e.g. rnoaddf! ' a couple') standin partial
congruencewith the noun. In constructionswith numeralshigherthan '1' and
thosewith indefinite pronounslike rnoaddf!' the head(in italics in Table 2.9)
is the numeralor pronounin the nominativesingularand the noun in the rest
of the cases.
The constructionswith the relative interrogativepronounsrnii 'which' and
gii 'who' behavein a similar mannerbut herethe noun is the modifier in the
nominativeplural (formally: the accusativeplural), as well.
Thereare also a restrictednumberof modifiers which are placedafter the
head. Especially common is the relative clause,which is connectedto its
correlate by the relative pronounsrnii 'which', gii 'who', goabbd 'which

Table2.9 Samplenounphrases,illustrating distributionof caseandnumber

Singular Plural Singular Plural

'this house' 'thesehouses' 'threehouses' 'threepairs of skis'

N ddtdallu dat ddlut golbmadalu golmmatsabt!hat

G danddlu daid ddliiid golmmaddlu golmmaidsabt!hiid
A danddlu daid ddliiid golbmaddlu golmmaidsabt!hiid
III dan ddllui daiddaddliiide golmmaddllui golmmaidesabt!hiiddq
Loc dan ddlus dain ddWn golmmaddlus golmmainsabt!hiin
Com dainnaddliiin daiguinddliiiguin golmmainddliiin golmmaiguinsabt!hiiguin
Ess danin ddllun golbmanddllun (not used)

Note: Also, sN moaddedruu 'a coupleof houses',sG moatti ddlu, sA moaddeddlu, sIll moatti
ddllui, etc., like golbmadruu. .
88 SAAMIC

Table2.10 Noun phraseswith relative/interrogativepronouns

Singular Plural Singular Plural

'which house' 'which houses' 'which priest' 'which priests'

N mii dliliiid matdliliiid giibahpaid geatbahpaid

G mandalu maid daliiid geanMhpa geaidMhpaid
A mandalu maid daliiid geanMhpa geaidMhpaid
(maid daliiid)
III mandallui maiddA daliiide geanMhppii geaiddl;l Mhpaide
Loc mandalus main daliiin geanMhpas geainMhpain
Com mainnadaliiin maiguin daliiiguin geainnaMhpain geaiguinMhpaiguin
Ess manin dallun geaninMhppan

(of two)', guhte 'who', and guhtemus'which (of a specific set)', e.g. dat
olbmot, mat manne sisa 'those people who went in' (mat pN of mii); dat
vielppis, goabbdlea stuorit 'the oneof the two puppiesthat is bigger'. Among
otherpost-modifiersare the infinitive (miis ii leat dilli vuordit 'I do not have
time to wait'), postpositionalconstructions(dat lea muitalus ovtta nieidda
birra 'it is a story abouta girl'), the partitive locative (mun dovddanovtta dii
vieljain 'I know one of your brothers'),and probably also specifiernouns
agreeing with their head in case (miis lea okta gilo vuodja PROsl.1oc
IS.s3presONE. sN KILO.sN BUTTER.sN 'I havea kilo of butter').

Negation
The negativeverb is combinedwith the connegativesto expressnegationin
the present tense. The negative verb has the same form for indicative,
conditional, and potential, and the main verb carries the mood marker
(indicative is zero). Examplesof the indicative: it boatll! 'you do not come',
it leat 'you are not'. Conditional: it boatlase'you would not come',it livcce
'you would not be'. Potential: it botld 'you may not come', it leacca 'you
may not be'. In the imperative,the mood markeris encodedin the negative
verb: all! boatll! 'do not come',all! leagl! 'do not be'. The secondimperative,
itself a connegative,is usedwith third-personforms of the negativeverb: allos
bohto 'let himlher not come',alloset bi5hto 'let them not come',allos lehko
'let himlher not be'. The indicative pastis expressedby the main verb in the
perfect participle: it boahtdn 'you did not come', it lean - it leamas- it
leamaslean- it leamasan'you were not'. The compoundtensesare negated
in a manneranalogicalto that of the copula: it leat boahtan 'you have not
come',it lean boahtan 'you had not come',it livcce boahtan 'you would not
havecome',etc.
Negation normally concernsthe verb constituentand has therefore the
entire sentenceas its scope.There are, however,ways to narrow the scope
down to a single constituent:mun ozzondan, in fal Niillasis, muhtoIl)ggas 'I
got it not from Nils but from Inga'. In participial andinfinitival constructions,
 SAAMIC 89

the matrix verb will be negatedeven thoughthe scopeof the negationis the
embeddedsentence:dat ii lohkan bierggu nohkat'he said that meatdoesnot
run out' (ii lohkan 'did not say', bierggu 'meatsA' , nohkat'to run out, to be
finished').
Participial and Infinitival Constructions
Embeddedsentencescan be representedby constructionswith the embedded
verb in a non-finite form. The action essive is used with verbs of direct
observationto indicate the presenttense, e.g. dat i5innii bdhpa boahtimen
'(s)he saw the clergymancoming'. With other verbs, the infinitive is used
for the samepurpose,e.g. dat jahkkd bdhpadiehtit vastadusa'(s)hebelieves
that the clergymanknows the answer,(s)he believesthe clergymanto know
the answer'.The past tenseof the embeddedsentenceis expressedby the
perfect participle, e.g. dat i5innii bdhpa boahtan '(s)he saw that the
clergymanhad come', dat jahkkd bdhpa diehtan vastadusa'(s)he believes
that the clergymanknew the answer,(s)he believesthe clergymanto have
known the answer'.
The nominative subject and nominative predicate of an unembedded
sentencego into the accusativecasewhen their constituentsare embedded:
]oavnna logai daid olbmuid (S; pA) leat ceavltiid (PR; pA) 'John said that
those people are haughty', contrast unembeddeddat olbmot (S; pN) leat
ceavltit (PR; pN) 'Thosepeopleare haughty'.If the subjectof the embedded
sentenceis co-referentialwith the subjectof the matrix sentence,it is replaced
by the reflexive pronoun(iehca-) in theseconstructions,e.g. ]oavnna logai
iews diehtanvastadusa'Johnsaidthat he (= John)knew the answer'.
Co-ordination and Subordination of Sentences
The co-ordinatingconjunctions,all from or via Fennic,areja 'and', sihke-
ja '(X) as well as (Y)' ,juogo- dahje 'eitherX or Y', vai 'or (in questions)"
muhto 'but'. Examples:valddE gaffala ja niibbi 'take a fork and a knife',
daga dola ja vuossagafe 'makea fire andmakesomecoffee', boahtagoeadni
vai ahcCi 'will father or mothercome?', mun curvon, muhto don it gullan 'I
shouted,but you didn't hear'. In negatedcases,co-ordinationcan also be
expressedwith the enclitic --ge attachedto the negativeverb, e.g. mun in leat
oaidnaninge gullan 'I haven'tseenor heard'.
Subordinationcanbeexpressed with anumberof conjunctions:ahte'that' «
Finnishettii), go 'when,that' « Finnishkun), goas 'when' « Finnishkonsa),
juos (cf. Finnishjos),vai - vuoi 'in orderto' ,amas(in somedialects)'lest, in
order that not', vaikko 'even though' « Finnish vaikka). Subordinated
sentences canhavemanydifferentfunctions,e.g. subject:buorr£! lei, go bi5htet
'it was goodthat you came';object: mun dieilan, ahte Niillas boahta 'I know
thatNils is coming'; adverbial:munboailan, go don dahtut 'I shallcomewhen
you ask(me to do so)'. Indirect questionsare subordinatedas subjects(ii leat
Cielggas, boahtagoNiilas 'it isn't clear whetherNils will come') or objects
(munin dieil£!, boahtagoNiilas'I don'tknow whetherNils will come').
90 SAAMIC

In subordinatedreported speech,the pronounsfor the first person are

replacedby thoseof the third, e.g. dat logai, ahte son digu njuovvatsdvzza
'(s)he said that (s)he will butcher a sheep',dat logai munnje, ahte soai
vuolgiba bajds joga '(s)he said to me that we (soai 'they [dual]') would go
up river'. Reportedspeechis often indicatedby the particlesgul (indicating
citation of someoneelse'sstatement,e.g. logai, ahte son ii gul dietlr:; '[s]he
said that [s]he doesnot know') or, rarely and in the easterndialectsonly,jat
(indicatingcitation of one'sown statement).

Lexicon
The North Saami lexicon is expandingrapidly as a result of more or less
consciousdeVelopment.Becauseof the large numberof word stemsand the
rich systemof word formation, new words are basedmostly on traditional
stems and endings, often replacing even recent loanwords. Thus the
neologismddidda 'art' (basedon ddidu 'knowing how') has supplantedthe
Scandinavianloan koansta(from Swedishkonst), usedas late as the 1980s
andpracticallythe only word for this conceptin the 1970s.The largestNorth
Saami dictionary to date contains about 35,000 entries and is far from
exhaustive.
Of the traditionalvocabulary,slightly fewer than 600 wordshavecognates
in other Uralic languages.Of these, approximately 100 date from proto-
Uralic, e.g. vuoni 'mother-in-law', vuohppa 'father-in-law', goaski 'senior
maternal aunt', ceahci 'youngerpaternaluncle', mannji 'daughter-in-law',
gdlqjeatni 'sister-in-law',natltla 'shaft',suhkat'to row'.
About 160wordsdatefrom pFU, e.g. eahki 'seniorpaternaluncle',vuovdit
'to sell', dhcCi 'father' (augmentedfrom proto-Saami*1t6ie > Kildin a;rpKh),
vuodjit 'to drive', lohkat 'to count', liepma 'broth', njuovvat 'to flay'. About
60 words may be ascribedto the Finno-Permianphase,e.g. dddjd 'grand-
father', astat 'to havetime' ,juohkit 'to divide', gdma 'shoe',got1t1it 'to knit',
boazu 'reindeer', boarti 'bark vessel'; roughly another 60 words have
cognatesin the Volgaic languages,e.g. mdksit 'to pay', gahcat 'to ask',
gealdit 'to draw (bow), to cock (trigger)" mealli 'rudder', sadjit 'to hone',
Janas 'boat'.
Finally, Saami sharesabout 200 words with Fennic, e.g. dhkku 'grand-
mother',Ciehkat'to hide',juoigat 'to sing in the Saamiway', gal 'let 'to visit'
gietkka 'cradle', vieUja 'brother', vuordnut 'to take an oath', joddu 'net',
geaidnu 'way', goarrut 'to sew', ldvzi 'rein',lohti 'wedge',bassat'to wash',
seakti 'bait', dohppa 'sheath'.
Alongside old inherited vocabulary thereare about 800 common Saami
word stems,over 600 of which are without etymology, e.g. atnit 'to use',
bivvat 'to keep warm', coagis 'shallow', caskit 'to strike', cdhppat 'black',
Ciekcat 'to kick', ddlkkas 'medicine',garvit 'to dodge',gdhccat 'to hurry',
heavdni 'spider', jal1J1Jis 'tree stump', jorrat 'to spin', ldhppit 'to lose',
 SAAMIC 91

manna'child', nagir 'sleep',njivli 'slime',ohca 'bosom',oakti 'rain shower',

ravgat 'to fall', sarrit 'blackberry',siehttat 'to make an agreement',uhcci
'small', valdit 'to take'.About 100 of the 800 are Scandinavianor Germanic
loanwords, e.g. airu 'oar', barru 'wave', dapmat 'to tame', gaica 'goat',
eallju 'zeal',gussa'cow', luovOs 'loose',mannu'moon',navli 'nail', riekkis
'ring', savza 'sheep',vievssis 'wasp', and about 150 are borrowed from
Finnish, e.g. aigi 'time', baiki 'place',dtivda illness', easka 'recently',giitit
'to thank', ihtit 'to appear',joavdat 'to arrive', ltivlut 'to sing', mearra 'sea',
neav'vu 'advice',oahppat 'to learn', reahkut 'to howl', suhttat 'get angry',
saddat'to be born', vassi 'hatred'.
In addition to the bulk of Scandinavianand Finnish loanwords(there are
severalthousandsof each),there arethe following layersof borrowings:

1 Old Indo-Europeanloanwords shared with several, mostly western

Finno-Ugric languages.There are about twenty of these, e.g. oarbbis
'orphan', coarvi 'hom', cohkut 'to comb', cuohti 'hundred', veahcir
'hammer',goahti 'hut, teepee,tent', dahkat 'to do', geassi'summer'.
2 Baltic loanwords,sharedmainly with Fennic. There are over twenty of
these; examplesare leaibi 'alder', suolu 'island', luokta 'bay', javri
'lake', luossa 'salmon',guoibmi 'companion',vuos'si 'pot handle',lasta
'leaf'. Someare not sharedwith Finnish, e.g. SouthSaamisaertie 'heart
(as food)', indicating that there were also direct contactswith Baltic
speakers.
3 Old Germanicloanwordssharedmainly with Fennic.Therearemorethan
twenty,e.g.gierdat 'to endure',bassi'sacred',soallut'to pick one'steeth',
buoidi 'fat', roavgu 'skin rug', ruovdi 'iron', guos'si 'guest', vuotta
'shoelace',lahttu 'member'.Some independentloans (e.g. luoikat 'to
borrow')indicatedirectcontactswith Germanic-speaking populations.
4 Russianloanwords.Thereare hundredsof recentRussianloansin Skolt,
Kildin, Akkala, and Ter. SomeRussianwords have spread,via Finnish
and Karelian, further west, e.g. darru 'Norwegian', radji 'border', and
gistta 'glove' have found their way down to South Saami. In addition,
North Saami has the following words of Russianorigin: savka 'cap',
spIre 'beastof prey « 3Bepb)', iskat 'to try', sibit 'domesticanimal',
ohpit 'again'.

From among the textually most frequent 100 words in North Saami
(comprising about 60 per cent of the spoken-languagetexts in the corpus
measured)11 words (i.e. 3 per cent)do not havean etymologyoutsideSaami,
e.g. boahtit 'to come', buot 'all', valdit 'to take', bidjat 'to put', orrut 'to
dwell'. Over 40 per centof the word stemsin the texts was coveredby those
most commoncore words which have an etymology connectingthem with
other Uralic languages(52 in number, e.g. the negativeverb, the personal
pronouns,nounslike ahcCi 'father', bealli 'half, eallu 'herd', beana 'dog',
92 SAAMIC

olmmos 'human being', siida 'village', and verbs like galgat 'to have to',
vuolgit 'to leave', gullat 'to hear', eallit 'to live'). There were 37 loanwords
amongthe top 100 words, comprising 14 per cent of the texts in the corpus,
e.g. ja 'and', ahte 'that', vel 'still', aigi 'time', juo 'already' from Finnish,
lavet 'to have the habit of from Scandinavian.Some items of the core
vocabularywhich were thought to have no etymologieshave recently been
shown tobe Germanicloans,e.g. oaidnit 'to see',gavdnat'to find', perhaps
also valdit 'to take'. In any event,the 600 or so word stems shared with other
Uralic languagesstill bulk largestin any Saarnitext.

North Saami Text

From HansAslak Guttorm: Idnjargga siljut.

A: text in standard orthography, segmented; B: morpheme-by-

morphemegloss;C: Englishtranslation.

AI. beatnaga-t--ge bahkke=st-edje veagal

BI. DOG-pN--ENC FORCES.ONE'SWA Y=SUBIT-p3past FORCIBLY

,olbrnu-i-d julgg-i-id gaskka vista-i.

PERSON-plur-gen LEG-plur-gen BETWEEN HOUSE-sIll

A2. Lernet doabu-i beatnagi-i-d rna-n

B2. CLEMENT GRABS-s3past DOG-plur-acc WHICH. ONE-sA

niska-i, rna-n seaiba-i fahti-i ja

NECK-sIll WHICH. ONE-sA TAIL-sIll CA TCHES-s3past AND

baIkesti-i uksa+lanja olggos. A3. beana lienju-i

THROWS-s3past DOOR+CHINK.sG OUT B3. DOG WHINES-s3past

go Lernet ceabehi-i cirvi-i.

WHEN CLEMENT NECK/fHROAT-sIll SQUEEZES-s3past

Cl. The dogs,too, forced their way betweenpeople'slegs into the house.C2.
Clementgrabbedthe dogs, caughtone by the neck, the other by the tail and
threw them out of the door. C3. The dog whined when Clementgrabbedits
throat.

Referencesand Further Reading

Abercrornby,J. (1895) 'The earliestlist of RussianLappWords',JSFOu 13/2: 1-10.
Airnli, F. (1918) Phonetikund Lautlehre des Inarilappischen,vols I-II, MSFOu 42,
43, Helsinki: SocieteFinno-Ougrienne.
Bartens,H.-H. (1980) Die Verwendungvon Potential und Konditional im Lappischen,
MSFOu 177, Helsinki: SocieteFinno-Ougrienne.
Bartens,R. (1971) 'Zur KongruenzdeslappischenAdjektivattributs' ,FUF 39: 31-40.
 SAAMIC 93

- - - (1972) lnarinlapin, merilapin ja luulajanlapin kaasussyntaksi, MSFOu 148,

Helsinki: SocieteFinno-Ougrienne.
- - - (1978) Synteettiset ja analyyttisetrakenteetlapin paikanilmauksissa,MSFOu
166, Helsinki: SocieteFinno-Ougrienne.
Bergsland,Kn. (1945) 'L'alternanceconsonantiquedate-t-elledu lapon commun?',
Studiaseptentrionalia2: 1-53.
- - - (1946) R¢ros-lappiskgrammatikk,SeriesB43, Oslo: Instituttet for sammen-
lignendekulturforskning.
- - - (1962) 'The Lapp dialects south of Lappland', in CommentationesFenno-
Ugricae in Honorem Paavo Ravila, MSFOu 125, Helsinki: Societe Finno-
Ougrienne,pp. 27-39.
- - (1967) 'Lapp Dialectal Groups and Problems of History', Lapps and
Norsemen in Olden Times, Series A26, Oslo: Instituttet for sammenlignende
kulturforskning,pp. 32-53.
- - - (1968) 'The Grouping of Lapp Dialects as a Problemof Historical Linguist-
ics', in C2lFUvol. I, Helsinki: pp. 77-85.
- - - (1973) 'Simplification of the Finno-Ugric Transcription: Lapp', in L. Posti
and T. Itkonen (eds), FU-transkription yksinkertaistaminen,Castrenianumin
toimitteita 7, Helsinki: Castrenianum,pp. 45-67.
Collinder, Bj. (1938) Lautlehre des waldlappischenDialekts von Giillivare, MSFOu
74, Helsinki: SocieteFinno-Ougrienne.
- - - (1949) The Lappish dialect of lukkasjiirvi, Skrifter utgivna av K. Human-
istika Vetenskaps-Samfundeti Uppsala 37/3, Uppsala: K. Humanistika
Vetenskaps-Samfundet i Uppsala.
Genetz,A. (1891) Kuollan Lapin murteidensanakirjaynnii kielenniiytteitii, Bidrag till
kannedomafFinlandsnaturoch folk, Utgivna av FinskaVetenskapsSocieteten50,
Helsingfors:FinskaVetenskaps-Societeten.
Grundstrom, H. (1946-54) Lulelappisches Worterbuch, auf Grund von K. B.
Wiklunds, Bjorn Collindersund eigenenAufzeichnungenausgearbeitet von Harald
Grundstrom,SeriesCl, Schriften des Instituts fiir Mundartenund Volkskunde in
Uppsala,Uppsala:Instituts fi.ir Mundartenund Volkskunde.
Halasz,I. (1885-96)Sved-lappnyelv, vols I-IV, Budapest:Ugor fiizetek.
Hansegard,N. E. (1965) 'SeaLappishand Mountain Lappish',lSFOu 66/6: 1-91.
- - - (1967) RecentFinnish Loanwordsin lukkasjiirvi Lappish,Acta Universitatis
Upsaliensis, Studia Uralica et Altaica Upsaliensia 3, Uppsala: University of
Uppsala.
Hasselbrink, G. (1965) Alternative Analysesof the PhonemicSystemsin Central
South-Lappish,Indiana University Uralic and Altaic Series 49, Bloomington:
IndianaUniversity.
- - - (1981-85) SiidlappischesWorterbuch, vols I-III, Schriften des Instituts fiir
Mundartenund Volkskundein Uppsala,SeriesC4, Uppsala:Instituts fiir Mundar-
ten und Volkskunde.
Itkonen, E. (1939) Der ostlappischeVokalismusvom qualitativenStandpunktaus mit
besondererBeriiksichtigungdeslnari- und Skoltlappischen,MSFOu 79, Helsinki:
SocieteFinno-Ougrienne.
- - - (1946) Struktur und Entwicklung der ostlappischen Quantitiitssyteme,
MSFOu 88, Helsinki: SocieteFinno-Ougrienne.
- - - (1949) 'Beitragezur Geschichteder einsilbigenWortstammeim Finnischen',
FUF30: 1-54.
- - - (1954) 'Uber die suffixalen Labialvokale im Lappischen und Ostsee-
finnischen',in Scandinavicaet Fenno-Ugrica, Studiertilliignade Bjorn Collinder
den 22juli 1954, Stockholm:Almqvist & Wiksell, pp. 183-91.
94 SAAMIC

- - - (1960) LappischeChrestomathiemit grammatikalischemAbriss und Worter-

verzeichnis,Apuneuvojasuomalais-ugrilaisten kielten opintojavarten7, Helsinki.
- - - (1968) 'Zur Frtihgeschichteder lappischenund finnischen Lokalkasus',in
C2IFU vol. I, pp. 202-11.
- - - (1969) 'Dber einige Formen der dritten Personin der lappischenKonjuga-
tion', FUF 37: 98-117.
- - (1972) 'Dber das Objekt in den finnisch-wolgaischenSprachen',FUF 39:
153-213.
- - (1973) 'Zur GeschichtedesPartitivs',FUF 40: 278-339.
- - - (1986-91) InarilappischesWorterbuch, vols I-IV, Lexica SocietatisFenno-
Ugricae 20, Helsinki: SocieteFinno-Ougrienne.
Itkonen, T. I. (1958) Koltan- ja kuolanlapin sanakirja, vols I-II, Lexica Societatis
Fenno-Ugricae15, Helsinki: SocieteFinno-Ougrienne.
Itkonen, T. (1973) 'Lisia eraan lapin vokaaliston ongelmaan',in Commentationes
Fenno-Ugricaein Honorem ErkkiItkonen, MSFOu 150, Helsinki: SocieteFinno-
Ougrienne,pp. 75-120,Germansummarypp. 120-24.
Koivulehto, J. (1992) 'Germanisch-lappische Lehnbeziehungen',in L. Honti, S.-L.
Hahmo, T. Hofstra, J. Jastrzebska,and O. Nikkila (eds), Finnisch-ugrische
Sprachenzwischendem germanischenund dem slavischenSprachraum,Amster-
dam- Atlanta, GA: Rodopi,pp. 55-95.
Korhonen, M. (1967) Die Konjugation im Lappischen.Morphologisch-historische
Untersuchung,vol. I: Diefiniten Formkategorien,MSFOu 143, Helsinki: Societe
Finno-Ougrienne.
- - (1969) 'Die Entwicklung der morphologischenMethode im Lappischen',
FUF 37: 203-362.
- - - (1974) Die Konjugation im Lappischen. Morphologisch-historische
Untersuchung,vol. II: Die nominalen Formkategorien,MSFOu 155, Helsinki:
SocieteFinno-Ougrienne.
- - - (1981) lohdatus lapin kielen historiaan, Helsinki: SuomalaisenKirjallisuu-
den Seura.
Kuruc, Rimma (1985) (ed.), CaaMcKo-PYCCK1HI.cHoBaph,Moscow: Russkij Azyk.
Lagercrantz,E. (1923) SprachlehredesSiidlappischennachder Mundart von Wefsen,
Bulletin 1, Kristiania: Kristiania EtnografiskeMuseum.
- - - (1926) Sprachlehredes Westlappischennach der Mundart von Arjeplog,
MSFOu 55, Helsinki: SocieteFinno-Ougrienne.
- - - (1929) SprachlehredesNordlappischennach den seelappischen Mundarten,
Bulletin 3, Oslo: Oslo EtnografiskeMuseum.
- - - (1939) LappischerWortschatz,vols I-II, Lexica SocietatisFenno-Ugricae6,
Helsinki: SocieteFinno-Ougrienne.
Lehtiranta, J. (1989) Yhteissaamelainensanasto, MSFOu 200, Helsinki: Societe
Finno-Ougrienne.
Nesheim,A. (1942) Der lappischeDualis mit Beriicksichtigungfinnisch-ugrischer
und indo-europiiischerVerhiiltnisse,Oslo: NorwegianAcademyof Sciences.
Nickel, K. P. (1990) Samiskgrammatikk,Berlings: Universitetsforlaget.
Nielsen, K. (1932-62) Lappisk ordbok - Lapp Dictionary, vols I-IV, B17, Oslo:
Instituttetfor sammenlignende kulturforskning.
Posti, L. (1953) 'From Pre-Finnicto Late proto-Finnic',FUF 31: 1-91.
Qvigstad,J. K. (1893) NordischeLehnworterim Lappischen,ChristianiaVidenskabs-
SelskabsForhandlingarfor 1893 no. 1, Oslo: ChristianaVidenskabs-Selskab.
Ravila, P. (1932) Das Quantitiitssystemdes seelappischen Dialekts von Maattivuono,
MSFOu62, Helsinki: SocieteFinno-Ougrienne.
Ruong, I. (1943) Lappische Verbalableitung dargestellt auf Grundlage des
 SAAMIC 95

Pitelappischen,Uppsala Universitets Arskrift 1943, 10, Uppsala: University of

Uppsala.
Sammallahti,P. (1977) NorjansaamenItii-EnontekiOn murteeniiiinneoppi, MSFOu
160, Helsinki: SocieteFinno-Ougrienne.
- - (1984) 'The Phonologyof the GuovdageaidnuDialect of North Saami',in B.
Brendemoen,E. Hovdhaugen,and o. H. Magga (eds) Riepmocdla: Essaysin
Honour of Knut Bergsland.Presentedon the Occasionof his SeventiethBirthday,
Oslo: Norvus Forlag, pp. 136-47.
- - - (1989a) Sdmi-suoma sdtnegirji. Saamelais-suomalainen sanakirja,
Ohcejohka:Jorgaleaddji.
- - (1989b)'A linguist looks at SaamiPrehistory', Acta Borealia 2: 3-11.
Sammallahti, P., and J. Mosnikoff (1991) Suomi-koltansaamesanakirja. Liiii'dd-
siiii'm siiii'nnke'rjj, Ohcejohka:Girjegiisa.
Schlachter,W. (1958) WiJrterbuch des Waldlappendialektsvon Mald und Texte zur
Ethnographie,Helsinki: Lexica SocietatisFenno-Ugricae14.
Skold, T. (1961) Die Kriterien der umordischenLehnwiJrter im Lappischen,vol. I,
Skrifter utgivna av institutionen ftir Nordiska spn'ik vid Uppsala universitet 8,
Stockholm:Almquist & Wiksell.
Wiklund, K. B. (1891) Laut- und Formenlehre der Lule-LappischenDialekte,
GoteborgsKongl. Vetenskapsoch Vitterhets SamhallesHandlingar,Ny TidsfOljd
25, Stockholm:GoteborgsKongl. Vetenskapsoch VitterhetsSamhiilles.
3 Fennic
Tiit-Rein Viitso

The Fennicbranchof the Uralic family is a dialect continuumthat is usually

divided into seven languages:Livonian, Estonian,Votic, Ingrian, Finnish,
Karelian and Veps. A Lude languagehas sometimesbeen extractedfrom
Karelian as an eighth language,but Ludes identify themselvesas Karelians.
Estonian (one million speakers)and Finnish (five million speakers)are
discussed in separatechapters in this volume. The remaining Fennic
languageshavebecomelanguagesof bilingual minorities in their traditional
territories; they are Votic (25 speakers),Ingrian (300 speakers),Karelian
(62,500 speakers)and Veps (6,000 speakers)in Northwest Russia, and
Livonian (15 native speakers)in Latvia.

Livonian
Livonian (liiv6 keelj , raandakeelj 'coastlanguage'),as spokenin Kurland, is
relatively homogenous.Three dialects, West, Central and East Livonian,
differ from eachother only slightly. The poorly attesteddialect of historical
Livonia, spokento the north of the lower reachesof DaugavaRiver, became
extinct in the nineteenthcentury.
There has beenliterature in Livonian since the translationof the Gospel
accordingto Matthew into Westernand EasternLivonian in 1863 and the
publicationofthe samegospelin a Central-and-Eastern Livonian compromise
languagein 1880. In the 1920s a new orthography was elaborated.The
Livonian literary languagewas annihilated with the Soviet occupationof
Latvia. Most Livonians were forced to leavetheir homeland(Uv6d Randa);
they now live scatteredamong other nations. Not until 1989 were a few
booklets publishedonce again in Livonian and efforts begunto revive the
language.
Livonian is the most innovativeFennic language.The chief characteristic
featuresof Livonian are:
Two distinctive lexical tones:the level tone and the glottalizedllaryngealized
one(or stf/Jd; often from *h; indicatedin this chapteras+);
Gradationof both historically long and short stressedsyllables with short
vocalism;
Long and shortfalling diphthongsuo: u o, Ie: ie;

96
 FENNIC 97

Map 3.1 Fennic:North/SouthandEast/WestDivisions

E
E

Ve
Ve F Finnish
E K Karelian
E E E Ve Veps
E I Ingrian
E Estonian
E
Vo Votic
L Livonian

E E

Source: Adaptedfrom Itkonen 1980.7,as reproducedin Szfj 1990: 154.

Rising diphthongs and triphthongs that can occur in up to five tone and
quantity patterns,e.g. k:'dgZD 'ships', k:'o+ig?J 'ship (sP)" IUozm?J 'warp
(sP)" tu+ oig?Jz 'birchbark',luoima 'warp (sNG)', aiga 'edge(sNG)', a+ig?J
'edge(sP)" azg?J 'time, weather(sP), aiga 'time, weather(sNG)" IQ+Qigi
'asunder';
Delabializationof the p[roto-]F[ ennic] labial front vowels;
Reminiscences of metaphony;
The contrastivebehaviourof bisyllabic a- and a-stems vis-a-vis U-, z- and
?J-stemsin inflectional paradigms;
The dativecase;
Near-completeloss of the externallocal cases.

Estonian
Estonian (eesti keel, older style: maakeel) is spoken mainly in Estonia.
Translationsof scriptureinto Estoniandate from 1535. Alongside the North
Estonianliterary language(formerly called the Tallinn language)the South
Estonianliterary language(the Tartu language)was usedin SoutheastEstonia
98 FENNIC

from the fIrst half of the seventeenthup to the secondhalf of the nineteenth
century;the degenerationof the Tartu languagebeganwith the publicationof
the Bible in North Estonianin 1739.
Estonian dialects are usually classifIed into three main groups: (1)
Northeasterncoastaldialect, (2) North Estonian(consisting of the Insular,
Western, Central and Easterndialects), (3) South Estonian (including the
Mulgi, Tartu, andVoru dialects).This classifIcation,however,is incompatible
with data from the other Fenniclanguages;accordingly,a classifIcationinto
fIve main groups is to be preferred:(1) Coastal,(2) Northeast,(3) East, (4)
North and (5) South Estonian. With the exception of the Coastal and
Northeastgroups, all Estoniandialects contrastthree syllabic quantitiesin
stressedsyllables. The North Estonianliterary dialect has assimilatedand
superseded almostall otherEstoniandialects.
CoastalEstonian,spokenon the southerncoastof the Gulf of Finland,was
originally closer to the Finnish dialects than to any other dialect group of
Estonian.CoastalEstonianneverhadthe backunroundedvowel e.
Northeastand East Estonian,togetherwith Votic proper, stem from one
protodialect; they sharethe merger *0 > *e in about twenty stems and the
assimilation*st > ss; NortheastEstonianhasbeeninfluencedby Coastaland,
later, by North Estonian.EastEstonianwasinfluencedfIrst by Southandlater
by North Estonian.
SouthEstonianprobablysplit directly from proto-Fennic.In somecasesit
hasretainedpF (and evenproto-Uralic) *c j , e.g. latisi 'child' < *lapcji, katisi
'two' « *kakcji < *kakti), kuudzi 'nail, claw'. On the other hand, it has
undergonethe characteristicchanges*kti > *kcji > *tjs\ *kt > *tt after the
vowel of the first syllable.

Votic
Votic (vatPtPaa ceeti, maaceeli) is presently spokenin coastal villages of
Vaipooli, Northwest Ingermanland,east of the mouth of the Luga River.
Vaipooli Votic was the westernmostdialect of Votic proper. Votic properhas
usually beendivided into West andEastVotic. WestVotic was a dialectchain
(Vaipooli, Pontizoo, Maci, Orko) characterizedby loss of word-fInal -G
« *-k) and -h, which were still retainedin EastVotic; on the otherhand,West
Votic had preservedword-fInal -n in the fIrst-person singular suffIx of the
verb (similarly to non-SouthEstonian),whereasin EastVotic the loss of -n
was total (as in South Estonianand Livonian). Other dialects of Votic were
Kukkuzi on the easternbank of the Luga River, which became extinctin the
1980s,and Krevin. The Krevin dialect was spokenby descendantsof Votic
war prisonerswho were movedto the vicinity of Bauska,Latvia, about 1445
by the Teutonic Order. Krevin becameextinct in the nineteenthcentury and
is very poorly attested.Therewas nevera Votic scripture.
Votic properprobablyseparatedfrom Eastand NortheastEstonianaround
 FENNIC 99

1000CEo In Votic proper,pF *k hasgoneto *c beforefront vowels.

Kukkuzi Votic never had the vowel *e. On the other hand, both Votic
proper and Kukkuzi have merged*ps and *ks into hs - hz and, similarly to
Northeast and East Estonian, have assimilated *st to an *ss which is
susceptibleto gradation(ss - s) in Votic properand Kukkuzi. Kukkuzi Votic
was originally a North Fennicdialectthat was first influencedby Votic proper
andlater byLower Luga Ingrian.

Ingrian
Ingrian, or Izhor (izoran keeli, earlier also karjalan keeli) may havehad five
dialects resulting from dialect split, with the possible exception of Lower
Luga Ingrian, which may well havedevelopedon the basisof a Votic proper
or Kukkuzi Votic substratum.Lower LugaIngrianis spokenalongthe Rosona
River in EstonianIngermanland(still annexedby Russia)and on both banks
of the lower reachesof the Luga River; Soikkola Ingrian is spokenon the
Sojkino peninsula;HevahaIngrian is spokenin the vicinity of the Koltusi
River. The Upper Luga, or Oredd,dialect, spokeninland, becameextinct in
the 1970s.The North Ingrian that was spokenby the Orthodox population
north of the Gulf of Finland hasbeenFennicizedand probablyservedas the
substratumfor the .Ayriimoinen dialect, traditionally classified as a form of
SoutheastFinnish.An attemptto createan Ingrianliterary languagewas made
in 1932-7.

Finnish
Finnish (suomenkieli) is spokenmainly in Finland, Sweden,and Ingerman-
land.
Finnish dialects have been traditionally classified into West Finnish
(SouthwestFinnish, SouthernTransitional,Hiime, SouthPohjanmaa,Central
and North Pohjanmaa,Deep North) and East Finnish (Savo and Southeast
Finnish). This classification seems somewhat oversimplified: West and
SoutheastFinnish as groups are too heterogeneousand should rather be
divided into separategroups.

Karelian
Karelian consistsof North (or Viena), South,Aunus (or Olonec) and Lude
Karelian.
North Karelianis spokenin North Karelia. SouthKarelianis a scatteredset
of dialects,somefew of which are spokenin Karelia, but most of which are
spokenin outlying parts of the Tver', St Petersburgand Novgorodregionsof
Russia, to which Karelians moved after 1617. The outlying regions are
Tihvina, Vessi(Ves'egonsk),Valdai, Tolmaccu,andDjorZa. Aunus(Karelian),
called livvin kieli, is spokenin SouthwestKarelia. Lude (Karelian), called
100 FENNIC

ljaadjin kielji, is spokenin Karelia in a strip eastof Aunus andSouthKarelian.

Lude is a dialect chain that can be divided into North, Central, and Kuudarv
dialects.
North and SouthKarelian (karjalan kieli) are often called Karelianproper.
North Karelian, however, is in fact more properly a dialect close to East
Finnish which has undergonestrong South Karelian influence. Similarly,
Aunus and Lude are dialects that were fIrst close to Veps but later have
undergone different degrees of South Karelian influence. In any case,
Kuudarv Lude hasremainedclosestto Veps.
The fIrst written documentin Karelianis a spell from the eleventhcentury,
written on birchbark and found in Novgorod. Attempts to create Karelian
scripture date from the nineteenthcentury. In 1804 books containing St
Matthew'sgospelandsomeOrthodoxprayerswerepublishedin bothSouthand
AunusKarelian.In 1820a StMatthew'forTver' Karelianswaspublished.Until
the OctoberRevolution, mostly Orthodox literature and a few primers were
published. Under Soviet power Karelians in Karelia were fIrst taught the
Finnish literary language.In 1931-7 a new literary languagebasedon the
Tolmaccudialectwascreatedfor Kareliansof the Tver' region; this usedLatin
script.In 1938-9in bothKarelia andtheTver' regiona new variantof Karelian
was introduced: this was a mixture of South and Aunus Karelian and used
Cyrillic script. A massiveRussifIcationof Kareliansas well as of other small
peoplesbeganafter 1940; since 1989, attemptsto revive different Karelian
dialectshavebeenmadein Karelia.A primerandsomeotherschoolbookshave
beenpreparedin Aunus Karelian. Some South and Lude Karelian has been
taughtat someschoolsalthoughthereareno schoolbooks.A projectof creating
a literary languagefor North Karelianshasevenbeenstartedanda primer has
beenprinteddespitetheclosenessofNorth Karelianto Finnish.

Veps
Veps is the easternmostFennic language.It is spokenin two different areas:
(1) on the southwesterncoast of Lake Onega in Karelia and (2) in the
northeastof the St Petersburgregion and the northwestof the Vologdaregion
of Russia.Since 1989 efforts have beenmade by Vepsiansto have the two
Veps territoriesunited undertheir own administration;theseefforts havemet
with resistanceon the part of the administrationof the St Petersburgregion.
Veps has beenclassifIedinto North, Central,and SouthVeps. North Veps
(ljadiikeP) is the closestto Lude and may havebeeninfluencedby its former,
now Russianized,Lude neighbours.South Veps and Central Veps (bepsan
kenmakeup a dialectcontinuumin which SouthVepsis the mostinnovative
dialect and the northern dialects of Central Veps share some featureswith
North Veps and with Lude Karelian. For Central and South Veps, scriptures
basedon Central Veps existed in 1931-7. In 1991 three books for Veps
children,including a primer, werepublished.
 FENNIC 101

Classification of the Fennic Dialects

The Fennic languageshavebeenconsideredto be so closely relatedwith the
Saarniclanguagesthat a commonancestorhas beenset up for both groups.
This putative common ancestoris usually called Early Proto-Fennicrather
than the more logical proto-Fennic-Saarnic. Fennieand Saamicare doubtless
close to one another: they have beenneighboursfor severalmillennia; they
both have Scandinavians and Russiansas their neighbours;there has beena
continuousFinnish and Karelian expansioninto Saarnicterritories and at the
sametime the Saamiclanguageshave beeninfluencedby Fennic. If Saarnic
andFennichavesplit from a proto-Fennic-Saarnic, the split is connectedwith
the massiveintroductionof Baltic andGermanicborrowingsinto Fennic.
Since the work of E. N. SetaHi., the Fennic languageshave usually been
divided on the basisof their most importantdialectsinto North(east)Fennic
(= Finnish, Ingrian, Karelian, Veps) and South(west)Fennic (= Livonian,
Estonian,Votic) This classificationexploits such innovationsas *e > *e in
words with back vocalism and *ns > s in South Fennic. This classification
ignores somecrucial facts, namely that (1) both Estonianand Votic include
dialects, namely Coastal Estonian and Kukkuzi Votic, which should be
classifiedas being of North Fennicorigin and(2) unlike North Fennie,which
really can be treated as a group, South Fennic is not simply more
heterogeneous than North Fennicbut rathera blanketterm for different non-
North-Fennicgroupswhich have formed, over time, variousSprachbundsin
which North Estonianhasusuallybeenthe dominatingdialect. In short,what
canbe classifiedare not the Fenniclanguages,but the Fennicdialects.
The main linguistic subgroupsof Fennic, i.e. Livonian, South Estonian,
Maa and North Fennic,can be describedin terms of the Starnmbaummodel,
i.e. as resultingfrom divergencefrom proto-Fennic,cf. Figure 3.1.
Recentstudieshaverevealedthat severalimportantinnovations,e.g. *cj »
*s and*s > *h intervocalicallyafter a non-initial syllable,and*8 > *h, *ti > *si,
which earlier wereascribedto proto-Fennic,in fact occurredafterit, but during
a periodwhencommoninnovationswerestill possible,i.e. in CommonFennic.
Livonian, alongside numerous innovations, also has some important
characteristicarchaicfeatures.For example,it has preserved(1) the contrast
of pF *ktt : *kt via *htt : *ht > +tt : +d as in the partitive and illative singular
forms of the numerals*tikti 'one' and *kakti 'two', and (2) the former stem

Figure3.1 Main historicalgroupingsof Fennic

Proto-Fennic

Neva
IMaa N Fennic

Livonian S Estonian N Estonian Chude WLadoga ELadoga

102 FENNIC

vowel *-a of severalnominals that have elsewhereshifted to *e-stems(e.g.

Fennic *jarval*j1irve- 'lake' from Baltic, cf. Lithuanianjaura). Livonian is
also the only Fennic languagein which there are distinct affirmative and
negativeimperatives2 forms of the verb, at leastfor bisyllabic a-stemssuch
as *anta- 'to give'. Schematically:

Livonian S Estonian NEstonian Finnish Veps

1 kQ{ttt~ kQ{ttt~ katte> : katte kaht(e): kahta: kaht : kahthe
kQQ+d~kQQ+d~ kahte kahteen
2 jQQra: pN jarjvj : jarve? jarv : jai'veD j1irvi : j1irvet djfu"jv:
jQQrad djfu"jved
3 aanda:ala anna?: annai?anna:lira anna:ala anda:ala
aitd(~) aitd(~) « *anna?ei?) anna anna? anda

South Estonianhas, characteristically,(1) tjsi from the pF cluster*pcj (as

in *lapcji 'child'); (2) tjsi < *kci « *kt before *i), as in *kakcji < *kakti '2';
(3) tt or D from *kt, when not followed by *i (as in *naktti 'seen'[impersonal
perfectparticiple]) (4) unmarkedpresentindicative s3 forms insteadof forms
endingin *-pi (e.g. *antainsteadof *antapi 'gives'):

S Estonian Livonian NEstonian Votic Finnish Veps

1 laijsj laps laps lahsi lapsi lapsj
2 kaijsj kaks kaks kahsi kaksi kaks
j
3 nattii na+ad~D
na+ad~D nantuD nahtu nahtu nahtu
4 arlD aandaB annaB annaB antaa andab

Dialects of the Neva group have characteristically(1) the diphthong *ei

correspondingto Livonian andSouthEstonianai in a setof words,e.g. *heina
v. *haina 'hay'; (2) the vowel *e correspondingto Livonian and South
Estonian*a in anotherlexical set, e.g. *selka v. *salka 'back'; (3) *e instead
of *e in the stem *metsav. *metsa 'forest'; (4) s insteadof the affricate *ci
as a reflex of *t in the earlier sequence*nti, e.g. *kiinsi v. *kiincji (both from
*kiinti) 'nail, claw':

NEstonian Votic Finnish Veps Livonian S Estonian

1 hein eina heina hein - hiin rona hain
2 sejlG selca selka sjeljg saalga salG
3 mets mecca metsa mjets metsa mets
4 kii:iiz ciiiisi kiinsi kjiinjz j kiinjtS kii:iiDjZ j
 PENNIe 103

The Neva groupin tum may be brokendown into North Fennic and Maa.
North Fennic has several characteristicinnovations, e.g. *sesar > *sisar
'sister' (Finnish, Lower Luga Ingrian, North Karelian sisar, Soikkola and

Figure3.2 North Fennicinnovations

West Finnish

Kukkuzi Narvusi Finnish

CoastalEstonian
Lower Luga SE Finnish

Hevaha Savakko Savo

Soikkola
N Karelian

SKarelian

Aunus

N Karelian

Kuudarv Lude

SKarelian

:··NC~~traiVep;···:o :··NC~~traiVep;···:o
N Karelian
N Karelian

key:
+ + *-ilen in genitiveplural - *-pi> *-bi > -u/-u in presents3 forms
v v *e > *0 in nominativeof pluralic personalpronouns allative => adessive
* *(*seppli: ) *seplin, *sepli (: *seplin) s z
z z *p *1 *k *s > b d g z in voicedenvironments
0 0 *Ik *rk» II rr
*iie *rk» I r in e-stemsandin words with front
vocalism '" '" *is *iz > is i);; *VlsC > ViC
C C breakingof the Proto-Fennie*0, *0, *i! x x *v > I? word-initially
*slr> Ir-dr - *j >dJ _ gJ word-initially and postconsonantally
1M breakingof the secondary*0 and i! from *oiw, *eke etc. ** (I) *-hi > -h in presents3 forms; (2) syncopation
= breakingof *Ii andli - shorteningof *0 *a *8 *11 *e
s S *sl > ss in the weakdegree 00 shorteningof *u *u *i
'DDHIJ *str> *sr ••• *pp Ott *kk > P I k everywhere
104 FENNIC

HevahaIngrian sizar, South,Aunus, and Lude Karelian and Veps sizar) and
*0> *6 in non-fIrst syllablesof wordswith a front vocalismof thefIrst syllable.
Dialectsof theMaagrouparecharacterized by severalMaainnovationssuchas
*0 > *e in more than ten stems(e.g. *oppi- > *eppi- 'learns'),and also by
innovationssharedwith Livonian andSouthEstonian,e.g.*e> *e in wordswith
backvocalism,*0> *e and *ns > *(V)s (e.g. North Estonianmaazikkaz,Votic
Propermaazikaz,Livonian mQQ~k?JzmQQ~k?Jz'strawberry').One commonfeature of
variousMaa dialectsand a part of the SouthEstoniandialect is the useof the
stem*maa 'country,land' for theidentificationof one'sown group,cf. Estonian
mfike:el 'Estonianlanguage',marahvaz'the Estonianpeople',Votic proper
maaceeli'Votic language'.
The Maa group may itself be subdivided into North Estonian and the
Chudegroup (EastEstonian,NortheastEstonian,and Votic proper).Thereis
in East and NortheastEstoniana very strong,but difficult to pinpoint, North
Estoniansuperstratumand similarly in Votic there are Ingrian and Finnish
superstrata;the preciseidentification of characteristicand commonfeatures
is thereforeoften difficult. In any event,in North Estoniantherehasbeenthe
change*e > 0 in some stems,cf. kord 'time, turn' « *kerta < *kerta), and
in Chudethe change*0 > *e hasoccurredin abouttwenty stems.
The interrelationsof dialectsof North Fennic are complex: cf. Figure 3.2,
which considerstwenty-onetypesof innovation.Contemplationof Figure 3.2
gives rise to various kinds of speculation.Most probably, North Fennic split
fIrst into two groups: into EastLadoga,which voiced *p *t *k *s into *b *d
*g *z in voiced environments,and West Ladoga, which merged the weak
members*pp *tt *kk of the gradationalpairs *pp : *pp, *tt : *tt, *kk : *kk
with *p *t *k that alternatedwith their weakervariants*p *t *k (or *B *6 *y).
Voicing of voicelesssingle obstruentsin East Ladogacannotbe considered
a late Russianinfluencebecauseit was the voicing that preventedthe spread
of the WestLadogamergerto SouthandAunus Karelian.

Phonology

Stress
In all Fennic dialects a few words (mostly conjunctions) are usually
unstressed,a few areoften unstressed,andmostwords havethe primary stress
on the fIrst syllable. Quadrisyllabicandlongerwords havesecondarystresses
on odd non-final syllables, i.e. there is a tendencytowards trochaic stress
patterning, cf. Estonianmgpala:male 'lower (sAll)" mgpala:masse'lower
(sIll)' .
The tendencytowardstrochaicstressis counteractedby certainderivational
suffixes that attract secondarystress,i.e. there is a morphologically bound
secondarystress,e.g.Estonianp~Genemi:neEstonianp~Genemi:ne 'escaping',Finnishpqkenemi:nen.
In addition, in Finnish and Votic, and locally also in North Estonian,
 FENNIC 105

quinquesyllabic andlongerwords with a shortthird anda long fourth syllable

havetheir secondarystresson the fourth syllable,e.g. Finnish fQdelli:sen 'real
(sG)' : fQdellise:ssa'real(sIne)'.In short,secondarystressis not fixed. In some
Finnish dialects there are casesof historically motivated secondarystress,
namely, of casesin which the stresshas remainedfixed on a formerly odd
syllableevenafter the syncopationof the vowel of the original secondsyllable
as in Eurajoki SouthwestFinnish muur.mi:s« *muurami:sa'in cloudberries
(pIne)': k!&.armis « *kuormisa'in loads(pIne)'),or aftertheriseof anewsecond
syllable becauseof epenthesis,cf. Kuhmalahti Harne k!&.lumassa:pa (with
secondary second syllable u, < *kglmassa:pa 'in the comer (sIne)' v.
k!&.luma:ssapa'beingworn out', with original second-syllable u.
In negative North Setu varieties of Vom Estonian, fusional negation
suffixes (from the old negationverbseil 'not [non-past]'and eS 'not [past]')
regularly attract the primary stress, so that a short initial syllable of a
bisyllabic form lacks stress, and otherwise the initial syllable receives a
secondarystress,e.g. elqil 'I do not live', elii:iis 'I did not live'.

Quantity
In pF only vowel quantity was distinctive: there were both short and long
monophthongs.Most Fennic dialectshave preservedthe distinction of short
and long vowels, but as a result of diphthongizationthe number of long
vowels may be reducedconsiderably.Centraland SouthVeps deviatein that
in thesevarietiesall long vowels have mergedwith their short counterparts;
SouthVeps then evolveda new set of long monophthongs,formed as a result
of (1) assimilativemonophthongizationof diphthongsin non-initial syllables,
rarely in initial syllables,and of (2) assimilativevocalizationof syllable-final
or word-final *1, e.g kuudaanie< *kuldainje < pF *kultainen 'golden'.
With the exceptionof the Finnishdialectsof Finland,the contrastof single
and geminate consonantshas evolved into two series of consonants.In
Livonian, South Karelian, Aunus Lude and Veps, single obstruentsbecame
voiced in voiced environments;geminateobstmentsremainedvoicelessand,
for various different reasons,at least a part of them was degeminatedinto
voicelesssingle obstruents.In most CentralVeps dialectsand in SouthVeps,
all geminateconsonantswere degeminated.Moreover (and again with the
exception of the Finnish dialects of Finland, where all geminates are
degeminatedas a side-productof apocopation)former geminateshave in
similar casesbeen renderedby final consonantsdifferent from those that
comefrom former single ones.
In most Fennic dialects original closednessof posttonic syllables has
causedthe weakeningof single and geminatestops on the boundaryof that
andthe precedingsyllable,e.g. *sepa:*sepan'neuksN:sG',*seppa:*seppan
'smith sN:sG'. This processand alternation is called radical gradation. In
Livonian a similar weakening of geminate stops arose, but for more
106 PENNIe

complicatedreasons.In Votic, Finnish, Ingrian and North Karelian the weak

grade of geminatesmerged with the strong grade of singles, e.g. *seppli:
*seplin, *sepli: *seplin); cf. Figure 3.2. In South,North, and EastEstonianthe
alternationsof geminatestopswere reinterpretedas alternationsof heavyand
light long syllables and were generalizedto all long stressedsyllables; in
Livonian they were generalizedto long stressedsyllables with no long
monophthongs.Later, syllabic weight was re-interpreted as an internal
property of long stressedsyllables that was retainedeven after the former
conditions for the rise of distinctive syllable weight were changed.Accord-
ingly in Estoniangrammarsthree syllabie quantities are distinguished:Ql
refers to short stressedsyllables,Q2 to light long stressedsyllables,and Q3
to heavy long stressedsyllables. Syllables of Ql and Q2 are obligatorily
followed by at least one unstressedsyllable, exceptin some subdialectsof
Voru Estonianwhere someverb forms of Q2 have lost the unvoicedvowel
of the secondsyllable,e.g. North Setusoozj? 'eat(conditional),, nudzi 'being
sucked'.
Tone
Tone was non-distinctivein proto-Fennic.In Livonian a contrastof level (or
slightly rising) tone and strjJd (or laryngealizedtone) was developedeither as
a result of the changeof preconsonantaland certain casesof intervocalic *h
into a tonal feature or as a concurrentfeature of geminationof weak single
consonantsin contrastto original geminateconsonantsthat occurredwith a
level tone. In South, North, and East Estonian tone can be polarized in
connectionwith the distinction betweenweak and strong long syllables: in
comparisonto syllables of Q3, syllables of Q2 have a pitch contour with a
delayedpeak.
Vowels
The pF vowel systemis reconstructedas having, symmetrically,eight short
and eight long monophthongsin the first syllable but only seven short
monophthongsin non-first syllables:

First syllable Non-first syllables

*.1 *ti *u *.1 *ti *u
*e *6 *0 *e *0
*li *a *li *a
* 11
.. *titi *uu
*ee *66 *00
*lili *aa

In proto-Fenniethe existenceof the back/frontpairs *u : *ti, *0 : *6, and

*a : *li in first syllables, and of *u : *ti and *a : *li in non-first syllables
providedthe basisfor a partial vowel harmony.The distribution of vowels in
 FENNIe 107

non-first syllablesof proto-Fennicwas regulatedby the following rules: (1)

*0 *i *e could occurin any non-first syllable; (2) *ti and *a occurredin a non-
first syllable only if not precededin the word by a syllable containingany of
the vowels *u *0 *a; (3) *u and *a occurredin a non-first syllable only if not
precededin the word by a syllable containing any of the vowels *ti, *6 *a,
but eventhenif a syllablecontaining*0 intervened.
Someof the changesin the vowel systemsof Fenniclanguageshavetended
to elaboratevowel harmony, e.g., (1) the split *0 > *0 : *6 in non-first
syllablesin North Fennic; (2) the split *e > *tf : *e in non-first syllablesin
Veps; (3) the split *e > *e : *e in Livonian, SouthEstonian,andMaa (in South
Estonian,ein the first and ~ in non-first syllablesare allophonesof the same
phoneme).On the other hand other changes,such as the rise of syllabic
quantity and tone distinctionsin the first syllable in Livonian and Estonian,
and the rise of CV-harmony (i.e., the congruenceof consonantand vowel
palatality v. non-palatality) in Veps, have often led to the loss of vowel
harmony.The youngestspeakersof North Veps often ascribethe feature of
palatality not to front vowels but to the palatalized consonantswhich
accompanythem. Theseand someotherchangeshaveyielded a considerable
variation of vowel correspondences and alternationsin the first syllablesin
different Fennic dialects; for an overview see Tables 3.1 and 3.2. Another
consequence has beena considerablereductionof the vowel systemof non-
first syllablesin Livonian andNorth Estonian.
The syllable-quantitydistinctions of Livonian and Estonian,the breaking
of initial componentsof diphthongsin Livonian, and assimilationshavebeen
sourcesof the greatvariability of polyphthongicsystemsin Fennic; seeTable
3.3. The monophthongizationoflong monophthongs(Table 3.2) andcontrac-
tions following the loss of intervocalicconsonantshavebeenothersourcesof

Table3.1 Proto-Fennieshortmonophthongsandtheir reflexes

WZi ELi Maa NEsWSa NF

*u u/ii u/ii uti u u ti ti

*0 u o/uo ill 0 u o/uo g/g dee0 oe 00 0
*a a/~oa/~o a/aa§ 0 ati a a ti ti
*ti iii iii ti ti ti
*0 ielie e/e ielie e/e ti0 ti ti
*li liIii e/e liIii e/e tili liti liti
*.1 iii iii tii i ti ti
*elF ielie e/e l'elie e/e e
ti ti ti
*elB iii e/g tie ti ti
Notes: WLi = WestLivonian, ELi = EastLivonian, NEsWSa= West Saaremandialectsof
North Estonian,NF = North Fennic.F = front vocalism,B = back vocalism.In Livonian, a
reflex standingleft from a slashoccursin a gradationlessform or in the strongdegreeand one
standingright from a slashoccursin the weak degree.Pairedreflexesalternateregularly.
108 FENNIC

Table3.2 Proto-Fenniclong monophthongsandtheir reflexes

pF ELi EsSV NEs EEs VoP Fi EFiSa SKa CSVec

*il il Ulil u/il Uly il il il il u u u u

*6 uo ~/60/ ~/60/ 0/6 ~/uo ~/uo 6 uo uo uo aa 0
*a (} ilia ala uatuii aa aa oa ua a
*U 1 ll/U WU iilJU ii ii ii ii ii ii ii ii ii ii ii
*8 ie e ij/8 6/8 ij/ii8 0 iio iio iio aa 0
*ii a ilIii Wiieateii iatiil aa aa ea ia a
*1 1 Iii Iii II! I I I I e
*elF ie e fie e/e Vie e ie ie ie ie e
*elB e vg ~/g~e/oe ~e/oe ~e/oe ~ ie ie ie ie ee

Notes: ELi =EastLivonian, EsSV =SouthVoru dialectsof Estonian,NEs =North Estonian,

EEs =EastEstonian,VoP =Votic proper,Fi =Finnish,EFiSa=Savodialectsof East-Finnish,
SKa =SouthKarelian, CSVec=Centraland SouthernVeps. F =front vocalism,B =back
vocalism.In Estonian,a reflex standingleft from a slashoccursin a syllable of Q3 andone
standingright from a slashoccursin a syllableof Q2.

the multitude of diphthongsin variousFennicdialects.

In Livonian andin most of the Estoniandialectstherehasbeena reduction
in the number of vowels occurring in non-first syllables. North Estonian
dialectsusually have only the vowels u a i e in that position. In Livonian the

Table3.3 Proto-Fennicdiphthongsand their reflexes

pF ELi NEs EEs VoP WFi NVeK

*ou ou/6 eu/eli eu/ee eu ou il

*au oUl6 au/ali aUla au au oii il
*iu iu/iu iU/ili iUlili 1U lU iiu
*eu eUlg eu/eli eUl eu eu oii iiii
*eii ieu/ie eifel d/e eii oii iiu
*aii aUlii ai/ai al/ae/ii aii aii oii au
*oii ieu/ie Oi/Oi oi/8 6i.i oii iiu
*ui ui/ui ui/ui ui/oe Ul ui Ul
*oi uoi/uoi oi/oi oe/oe oi/oe /6 oi ei oi oi ai
gi/gi eigi/gi
/~i eifel ei/ee
*ai ai/ai /(} ai/ai ae/ae al/ae/a ai ai ai ai ai
*oiIV ei eifel eifel ei/ee ei oi I
*ai ei/e ai/ai ae/al ai/ae/ii ai ai ai
*elF ei/e eifel d/e ei ei ai 1
*elB ei/g ellei ei el 1

Notes: ELi =EastLivonian, NEs =North Estonian,EEs =EastEstonian,Vop =Votic proper,

WFi =WestFinnish,NveK =North Veps. B =backvocalism,F =front vocalism,V =verb. In
Livonian andEstonian,a reflex standingleft from a slashoccursin the strongdegreeandone
standingright from a slashoccursin the weakdegree.
 FENNIC 109

same four vowels occur after a short first syllable, but after a long first
syllable of the weak degree5 a i occur, and in other non-first syllablesonly
5 and i occur. Voicelessvowels occur in non-first syllables of someFennic
dialects,e.g. in SetuSouthEstonian,Ingrian, andWestFinnish.
In Livonian, South, North, and East Estonian,and Veps, an earlier final
vowel (except *a/*ii in Livonian) has been regularly apocopatedboth in
bisyllabic words with a long first syllable (Livonian UkS, EstonianukS, Veps
uksj < *uksi 'door') and in longer words. In DjorZa South Karelian, final
vowels of all disyllabic and longer words havebeenapocopated.In the same
dialects, except Livonian, the vowel of an open short secondsyllable in a
trisyllabic or a longerword hasregularly beensyncopatedif the first syllable
was long (e.g. Estonianmire, Veps tjiitj,Jen < *mttaren 'daughter'[sG]).

Consonants
In CommonFennieafter *ti > *si and after the rise of *h therewere probably
thirteenconsonants,viz.

*m *n *IJ
*p *t *k
*s *h
*v *0 *1 *j
*1
*r

Severalinnovationshave led to the rise of new consonantseries,viz. (1) the

shift of single obstruentsto voiced b d g z, which triggereda partial or total
degeminationof geminatesin Livonian and EastLadoga(cf. Figure 3.1); (2)
apocopation,to different extents,in all Fennic dialects that has led, in all
dialectssaveLivonian, Central and SouthVeps, and Finnish, to the rise of a
long consonantseriesfrom thosegeminateconsonantsthat had precededthe
apocopatedvowel; (3) the fusion of a consonantwith a following *j that
yieldeda seriesoflong palatalizedconsonantsin Votic, Lude, and NorthVeps;
(4) the assimilativeinfluence of front vowels, most usually *i, on preceding
consonants,especiallydentals,togetherwith the subsequentapocopationor
syncopationof the vowel; theseprocessesthat have given rise to a parallel
contrastive series of palatalizedconsonantsin Livonian, Estonian (except
Northeastand CoastalEstonian),Savo Finnish, Karelian and Veps; in Voru
Estonianall consonantsexcept?andj may be palatalized.
In Votie proper *k has shifted to *c before front vowels (*kiisi > casi
'hand');in somedialects *g has similarly shifted to df,. In Aunus, Lude, and
Veps, s and z haveshifted to sandzwhen precededby i; in North and South
Karelian undersimilar conditionss andz haveshifted to ~ and zj; elsewhere
they have gone to s. In Livonian and South and East Estonian, fusion of
postvocalic i with a following dental has yielded a palatalizeddental (or
110 FENNIC

palatalizedsj in Livonian), e.g. South Estoniannaaz!e?'women',Livonian

laa~ka laa~ka 'lazy'.
The weakening of single stops initial in posttonic unstressedsyllables
within the framework of radical gradation, and the subsequentloss of the
reflexes of the weakenedstops, have led to an increaseof homophonous
forms and to contractionsthat havealteredword structure,e.g. EstonianlUGU
'story': pN lo:oD « *IUkOt) andlo:oD 'islet': pN lo:oD « *lootot).
The weakeningof syllable-initial single stops after unstressedsyllables
within the framework of suffixal gradation,and the further loss of weakened
stops,hastriggeredvarious assimilationsandcontractionsthat haveradically
changedthe structureof non-first syllableseverywherein FennicexceptVeps.
The loss of intervocalic *h in Livonian, Estonian,Votic and Finnish has had
a similar effect.

Morphology
From the typological point of view, the Fennic dialects range across an
extensivegamut, from the quasi-agglutinativeNorth Fennic and Votic to the
fusional Livonian and Estonian,where as many as five paradigmaticforms
canbe different suffixlessstemallomorphs.
Nominal Inflection
All Fennic languageshave at least three grammatical cases(nominative,
genitive, partitive) and a numberof adverbialcases,for the most part in both
singularandplural; proto-Fennichad at leastten adverbialcases.
The nominative singular is unmarked. Thenominative plural ends in a
pluralizer (-t, -d, -D or, in Voru Estonian,-? < *-t) or is unmarked(*-t has
goneto zeroin Mulgi andin Tartu SouthEstonian).
The genitive singularhasretainedits ending(*-n) in North Fennicandlost
it elsewhere.The genitive plural had the ending *-ten in North and East
Estonian,West Finnish, and probably in Livonian, and the ending *-iten (>
*-ten) elsewhere;*-iten containedtwo pluralizers,*-i- and *-t.
Both the partitive singular (*-ta) and the partitive plural (*-ita) have
undergonechangesof various kinds and exhibit a great variety of case
formativesandfusional effectsin Fennic.
The pF adverbial casesinclude (the uppercaseletters A E U render the
frontlbackharmonicpairs ala e/e>ii/u):
threeinterior local cases(illative *-sen, inessive*-snA, elative *-stA);
threeexteriorlocal cases(allative *-len, adessive*-lnA, ablative*-ltA);
the translative(*-ksi), the essive(*-nA), the instructive(*-in), the comitative
(*-ine-) and the abessive(*-ttAk).
In Common Fennic, the illative and adessiveendingsunderwentcertain
phoneticchanges:*-sen> *-hen, *-lnA > *-1lA. The suffix *-hen has been
lost in Livonian, Estonian,andVotic. The change*-snA > *-hnA occurredin
 FENNIC 111

South Estonian (with reflexes -hn, -h, or -n); a similar changein South
PohjanmaaWest Finnish is either a reminiscenceof the sameperiod or an
analogical change.The change *-len > *-llen in at least Estonian, Votic,
Ingrian and Finnish was at least partly in analogy to the new form of the
adessiveending, *-1lA. North and South Karelian have mergedthe illative
with the inessive,andthe allative with the adessive.In Aunus,Lude, andVeps,
the elative and the ablative have mergedwith the inessiveand the adessive;
secondaryelative and ablative endingshave often beenformed on the basis
of the inessive and adessiveendings and the postposition *pain 'in the
directionof. The essiveas a casehasbeenlost in Livonian andin South,East,
and mostof North Estonian(but was reintroducedinto the literary language).
The instructiveand especiallythe comitativehavebecomenon-productivein
all Fennic dialects; new comitative endingshave appearedin Estonianand
Votic (from the former postposition*kans(s)a),and in Lude and Veps (from
the postposition*kera). Livonian has a dative in -n or -~(ama (ama viiPa feb
'all
translative
lapst~ the propertyremainsto [the] children'); Livonian has alsomerged
the translative and the comitative into an instrumentalcase and uses the
exterior case forms only for some toponyms and certain lexicalized and
adverbializedforms. SomeFinnishKarelian, IngrianandVotic pronounshave
an accusativecaseform. Veps dialectshaveseveralsecondarycaseforms that
resultfrom fusion of former caseforms andpostpositions.
Verb Inflection
The Fennic verb has finite and non-finite forms. Finite forms can be
characterizedin terms of personand number,voice, aspect,mood and tense.
Non-finite forms includeparticiples,infinitive, gerundandsupines.
Finite Fonn Categories
Fennicdistinguishestwo numbers:singularandplural, andthreepersons.The
proto-Fennicfinite verb had at least two voices (personaland impersonal),
two aspects(affirmative and negative),four moods (indicative, conditional,
imperative and potential) and four tenses(present,imperfect, perfect, and
pluperfect;the perfectandpluperfectwerecomposites).
Proto-Fennichad the following personalendings:sl *-n « *-m), except
in the imperative, which had no sl form; s2 *-t, except in the imperative,
where the second personsingularhad no personalending; pI *-mV and p2
*-tV, with diversepluralizersin different dialectsaddedin the indicative, the
conditional, and the potential probably only later, e.g. Voru Estonian-mE?,
EastVotic -mmAG,HevahaIngrian -mmAn,CentralVeps -mAio
The third-person singular forms had the personal ending *-sen in the
imperative; this same ending also occurs in the reflexive conjugation of
Aunus, Lude, and Veps and in certain South Estonianverbs. Otherwise,the
s3 forms have no personalending: in SouthEstonian,many s3 forms in the
presentindicative are entirely unmarked, andelsewherethe s3 forms come
from those endingin *-pi « *-pA) from the presentparticiple suffix *-pA;
112 FENNIC

all other s3 forms end in a tenseor mood suffix sharedwith otherforms. The
p3 endedin the imperativein *-set (> *-het); elsewherethe tenseor mood
suffix was simply followed by the pluralizer*-t, in the presentindicative after
the suffix *-pA, i.e. *-pA-t.
The sl ending has beenlost in Livonian, South Estonian,and EastVotic.
Later, the Livonian unmarkedsl forms were replacedmostly by s3 forms.
Perhapsalready in proto-Pennic,the pI and p2 suffixes assimilatedto the
precedingpresent-tensesuffix: *-k-me- > *-mme-, *-k-te- > *-tte; later the
suffix *-tte- and often also the suffix *-mme- becamegeneralizedto otherpI
andp2 forms.
Correspondingto the six forms that make up the personalvoice there was
an impersonalvoice in proto-Pennic.Proma transformationalperspective,we
could say that a predicateverb in any of the threepersons,singularor plural,
could be replaced by a special impersonal form that contained (I) the
impersonalsuffix *-tA- or *-ttA-, (2) an imperfector a mood suffix and, (3)
in the indicative and the imperative, the ending *-sen. The corresponding
nominal or infinitive subjectwas simultaneouslyremoved.Suchimpersonal-
ization is still possiblein all tensesand moodsin all Penniclanguagesexcept
Livonian, e.g. Estonian kaks me:est maGaVaD kontsemil 'two men are
sleepingat the concert'»kontsemilmaGattakse'one sleepsat the concert'.
Note that in the latter casethe sentencehasno formal subject.Voru Estonian
also has a passive: a transitive clausewith the predicateverb in a personal
form of the present or imperfect indicative can be made passive by
transformingthe object into a subjectand replacingthe active predicateverb
with the correspondingpassiveverb. The s3 passiveforms are homophonous
with the correspondingimpersonalforms, while other forms have the usual
endings.
The negativeforms in Pennichavebeenformed by meansof an auxiliary
negationverb *e- (inflected in the indicative) or a prohibitive verb elii-liilii-
lala- (inflectedin the imperative)followed by the main, or lexical, verb in the
appropriatetenseor mood form. In Livonian, South, and EastEstonian,and
locally in InsularEstonian,the negationverb also has imperfectforms (used
only in the indicative mood) and followed by the main verb indicative mood
stem.Elsewherethe negationform hasonly the present-tense forms; theseare
followed in the presentindicativeby the main verb indicativemoodstem(see
below), and in the imperfect by the preterite participle. Both models have
parallelsin otherPinno-Ugriclanguages.
The indicative mood has two tenses.The presentis built with the suffix
*-k-. The imperfect is built with the suffix *-i-, exceptin the negationverb,
wherethe imperfectsuffix was *-s(i)-, cf. Livonian s3 i+z, SouthEstonianess.
The conditional mood marker comesfrom *-ksi- in Livonian, and in South,
North, and EastEstonian,and from *-isi- in Votic and North Pennic,with the
exceptionof CoastalEstonian,where the suffix -iksi- of monosyllabic and
e-stemsis thoughtto be the resultof contamination.Although *-ksi- and *-isi-
 FENNIC 113

are sometimestreatedas suffixes of different origin, this -iksi- may well be

identical with their common protoform. The potential in *-ne- has been
preservedmainly in North Fennic.The imperativehasthe markers*-k in s2,
*-kA- in p2, and *-ko- in s3, p3 andthe impersonal.The s2 imperativemarker
hasbeenpreservedas k only in EastVotic andHevahaIngrian.
Livonian and Estonian have two additional moods, the quotative (or
narrative)and the jussive.Theseare ratio obliqua analoguesof the indicative
andimperativemoodsandin Livonian aremore or lessobligatory.

Non-finite Form Categories

Proto-Fennichad four participles:the presentpersonalin *-pA, the preterite
personal in *-nUt, the present impersonal in *-tApN*-ttApA, and the
preterite impersonal in *-tU/*-ttU. Participles occurred as attributes, as
predicatives,and as componentsof compositeperfect and pluperfectforms
built with the present and imperfect forms of the auxiliary verb 'is'. In
Livonian the impersonalparticipleshavebecomepassiveparticiples.Partici-
ples are inflected as nominals. In several dialects the preterite participle
suffixes have contaminatedone other:there has beeneither loss of the final
*t of the personalparticiple or a spreadof this *t to the impersonalparticiple.
Infinitives built with the suffix *-tAk servedas both subjectsand objects.
All Fennic dialects have at least one gerund built with the suffix *-tesnA
(originally the inessive of the infinitive) that functions as an adverbial
indicating a simultaneousaction. Supines are verbal adverbs, formed by
meansof the suffix*-mA- plus a casesuffix, e.g. illative *-mAsen, inessive
*-mAsnA, ablative*-mAstA, translative*-mAksi, abessive*-mAttAk.

Syntax
All Fennic languagesare SOY languages.In contrastto most other Uralic
languagesadjectives,participles, andsomepronounsin the role of attribute
agreewith their headnounin caseand number.In Voru Estonianandin Veps
the negationverb often follows the main verb.

Vocabulary
The Common Fennic vocabularycontains (1) stems inherited from earlier
stages(protolanguages) which continueproto-Uralic,andderivatesbuilt from
thesestems;(2) stemsof unknown origin; (3) stemsborrowedfrom various
ancient Indo-Europeanlanguagesinto pF or Common Fennic, viz. ancient
Baltic, Germanic,and Slavonic borrowings; (4) stemsborrowedinto Fennic
from known contactlanguages,chiefly Swedishborrowingsinto Finnish and,
to a lesserextent,into Estonian;(5) Low and High Germanborrowingsinto
Estonian and Livonian, Latvian borrowings into Livonian and, to a con-
siderablylesserextent,into Estonian;(6) Russianborrowings(on a massive
114 FENNIC

scale) into Karelian, Veps, Votic, Ingcian and, to a lesserextent, into other
languages.
Turning to the oldest borrowings,there are about 400 loans from Baltic
(however,fewer than 200 of theseare absolutely certain,e.g. *kirves 'axe',
*heina/*haina 'hay', *jiirviJ*jarva 'lake', *ttittar 'daughter', *sampas
'tooth'); about500 loansfrom Germanic(e.g. *lampas 'sheep',*kana 'hen',
*pelto 'field', *laiva 'boat, ship', *kuniIJkas 'king', *kurkku 'throat'); about
10 borrowingsfrom Slavonic(e.g. *hauki 'pike', *vilja 'property;crop', *sirti
'log, pole'). Baltic and Germanic borrowings include a number of terms
relating to agriculture and technology, but also topographic,somatic, and
kinship terminology. The flow of Germanic loanwords must have been
continuous.There was a long time interval, though, betweenthe Baltic and
Latvian borrowings and between the oldest Slavonic and Old Russian
borrowings.

Referencesand Further Reading

Ariste, P. (1968) A Grammar of the Votic Language, Bloomington-The Hague:
IndianaUniversity: Mouton.
Hofstra, T. (1985) Ostseefinnischund Germanisch. Die Lehnbeziehungenim nor-
dlicher Ostseeraumim Lichte der Forschungseit 1961,Groningen:Drukkerij van
DenderenB.Y.
Kettunen,L. (1913) LautgeschichtlicheUntersuchungiiber den kodaferschen Dialekt,
Helsinki: SocieteFinno-Ougrienne.
- - - (1914) LautgeschichtlicheDarstellung iiber den Vokalismus des koda-
ferschenDialekts,Helsinki: SocieteFinno-Ougrienne.
- - - (1938) Livisches Worterbuch mit grammatischer Einleitung, Helsinki:
SocieteFinno-Ougrienne.
- - - (1960) Suomenliihisukukielten luonteenomaisetpiirteet, Helsinki: Societe
Finno-Ougrienne.
Laanest,A. (1982) Einfiihrung in die ostseefinnischeSprachen,Hamburg:Buske.
Posti, L. (1942) Grundziige der livischen Lautgeschichte,Helsinki: SocieteFinno-
Ougrienne.
- - - (1954) 'From Pre-Finnicto Late Proto-Finnic:studieson the developmentof
the consonantsystem',FUF 31: 1-91.
Suhonen,S. (1988) 'Geschichteder ostseefinnischen Sprachen',in D. Sinor (ed.), The
Uralic Languages:Description, History and Foreign Influences,Handbuchder
Orientalistik8/1, Leiden: Brill, pp. 288-313.
Vaba, L. (1990) 'Die baltischen Sondementlehnungenin den ostseefinnischen
Sprachen', in Itiimerensuomalaiset kielikontaktit, Helsinki: Valtion
Painatuskeskus, pp. 125-39.
Viitso, T.-R. (1978) 'The history of Finnic 0 of the first syllable', Sovetskoefinno-
ugrovedenie14: 18-106.
Viks, D. (1992)A ConciseMorphological Dictionary of Estonian,vol. I. Introduction
and Grammar,Tallinn: EstonianAcademyof Sciences,Institute of Languageand
Literature.
4 Estonian
Tiit-Rein Viitso

Estonian (native name from the nineteenth-centuryeesti keel, formerly

maakeel)is spokenby about 1 million Estonians,mainly in Estoniabut also
in Russia(mainly by descendantsof resettlersof the end of the nineteenth
century),andin Sweden,Canada,the United States,andAustralia (mainly by
refugeesfrom Nazi and Soviet terror and by their descendants).There have
beensome South Estonianenclavesalso in Latvia and in Russia;the Leivu
enclave in Ilzene and the Lutsi enclavein Ludza, Latvia, did not become
extinct until the twentiethcentury.
Estonianhas traditionally comeinto contactwith Finnish in the north and
in the northeast, with Ingrian and probably Votic in the northeast, with
Russianin the east,with Latvian in the south,with Livonian in the southwest
and with Swedishin the west. From the thirteenthcenturyuntil 1939 a sparse
Germanpopulationhas lived in Estonia(16,000in 1939) and approximately
from the sametime up to 1944, a small Swedishpopulation(8,000 in 1939),
this latter concentratedalong the northwestcoastand on Vormsi and Ruhnu
islands. Mostly beginning from the secondhalf of the seventeenthcentury
therehasbeenon the westcoastof Lake Peipsi,on the eastbankof the Narva
River south from EstonianIngermanland,and in easternSetumaaa Russian
population,at presentabout 150,000descendants of 80,000personsin 1939,
and about 450,000 Russian-speakingSoviet colonialists who arrived in
Estoniain 1944-88.

Dialects
Estoniandialectshaveusuallybeenclassifiedinto threegroups:(1) Northeast
Coastal;(2) North Estonian,including (a) the Insular, (b) Western,(c) Central
and(d) Easterndialects;and(3) SouthEstonianincluding (a) Mulgi, (b) Tartu
and (c) Voru dialects. For the purposesof this chapterboth the traditional
North and Northeast Coastal groups are reclassified into two groups, so
Estonianis consideredto consistof five main groups:(1) North, (2) East,(3)
Northeast,(4) Coast,and (5) SouthEstonian.SouthEstonian,in particularits
southeasternmost Voru dialect has been poorly understoodby speakersof
other Estoniandialects. The Voru dialect remaineduninfluenced byNorth
Estonianuntil the secondhalf of the nineteenthcentury and its easternmost
Russian Orthodox group, called Setus, were administratively united with

115
116 ESTONIAN

other Estoniansonly in the courseof Estonia'sWar of Liberation in 1919.

South Estonianprobably stemsdirectly from proto-Fennic.CoastEstonian
was originally a North Fennic dialect and probably close to East Finnish;
Northeastand East Estonianwere originally closer to ProperVotic than to
North Estonian.
As a result of the high generallevel of literacy already achievedby the
nineteenthcentury,and of the prestigeof educationand educatedspeechand
of several other, historical, factors, the fonner dialects have been almost
completely absorbed by standard (North) Estonian. Only Voru South
Estonianis to someextentstill actively used,althoughrarely by children.

Literary Estonian
The fIrst recordings of Estonian names date from the beginning of the
thirteenthcentury,the earliestrunning texts date from the sixteenthcentury,
andthe fIrst (partial) book was printedin 1535.This was a catechismwritten
by SimonWanradtin North Estonianor in the Tallinn languageandtranslated
into Estonianby JohannKoell. In the southeastern part of Estonia,namelyin
the territory of Tartu and Voru South Estonian, anotherliterary language,
usually referredto as the Tartu language,was used at least as early as the
seventeenthcentury. The fIrst North Estonian grammar(Anfiihrung zu der
EsthnischenSprach, 1637) was written by Heinrich Stahl (1600-57),and the
fIrst SouthEstoniangrammarwas written by JohannGutsleff in 1648. Both
literary languageswere developedfirst by clergymenof Gennanorigin in the
frameworkof the Lutheranideology of making the Scripturesunderstandable
and available to anyone and in one's native language. From the 1720s
calendarbooks and books on health care were printed. In 1766 the fIrst
Estonianjournal,the weekly Liihhike oppetus... beganto be publishedby the
physicianE. P. Wilde in North Estonian.In 1806 the fIrst attemptwas made
to issue an Estonian newspaperin the Tartu language: Tarto-ma rahwa
Niiddali leht. The retreatof the Tartu literary language,however,had begun
already in the eighteenthcentury, after Anton Thor Helle (1683-1748)had
publishedhis North Estoniangrammar(Kurtzgefaj3teAnweisungzur Ehst-
nischen Sprache, 1732) and especially after the Bible was translatedinto
North Estonianunder the supervisionof Helle and publishedin 1739. The
declineof SouthEstonianas a literary standardwas completedin the second
half of the nineteenthcentury.
From the 1820s more and more Estonianstook part in developing the
literary language; especially influential was Otto Wilhelm Masing
(1763-1832).Masing publishedseveraleducationalbooks on fundamentals
of Christianityandon secularsubjects,suchas the educationalcalendarseries
Marahwa Kalender (1823-6); he initiated the tradition of North Estonian
newspaperswith his weekly Marahwa Niiddala-Leht (1821-5) and was fIrst
to discussseveralcomplex problemsof Estonianpronunciationand ortho-
 ESTONIAN 117

graphy.Masingwas the fIrst to notice and to write the Estonianvowel e«0».

From 1840, the physician and writer Friedrich Reinhold Kreutzwald
(1803-82), made signifIcant contributions in creating the new medicinal,
chemical, and geographicalterminology that he used in his textbooks and
popularizedwritings on health care and generalknowledge.With the rapid
rise of the national movementin the 1860s,Estonianbeganto developfrom
being a rural languageinto a universaland unifIed national language.From
the 1870s Finnish becamean important source of lexical borrowing and
servedas a modelfor word formation.The grammarpublishedby Karl August
Hermannin 1884 put an end to severallong-lastingdisputesin the fIelds of
pronunciation and morphology and Ado Grenzstein'sEstonian dictionary
(Eesti Sonaraamat,1884), which was the fIrst dictionary addresseddirectly
to Estonians,included 1600 neologismsand internationalwords. Extremely
fruitful for literary Estonianwas the activity in the yearsbetween1910 and
1940 of two competingschoolsof languagedevelopment:thoseadvocating
language innovation (keeleuuendus), headed by Johannes Aavik
(1880-1973), and those in favour of more moderatelanguageregulation
(keelekorraldus),headedby JohannesVoldemarVeski (1873-1968).Veski's
activities include his participation in the compiling of the Estonian ortho-
graphic dictionary (Eesti keele oigekirjutuse-sonaraamat,1918), his role as
co-ordinatorin the preparationof morethanthirty terminologicaldictionaries,
and his work as chief compiler of the Estonianorthoepic dictionary, Eesti
oigekeelsuse-sonaraamat (1925-37), with about 130,000 lexemes; all this
work formed a fIrm basisfor the literary standard.

Orthography
Estonian orthographyhas passedthrough the following statesof develop-
ment:

1 Estonianwas at fIrst written inconsistently,on the modelof Low German

and Latin orthography, until the second quarter of the seventeenth
century.
2 Estonian was subsequentlywritten on the model of High German
orthography, usingmost of the letters of the Latin alphabet,especially
underthe influenceof Heinrich Stahl.
3 There followed the Old Orthography,fIrst applied by Bengt Gottfried
Forseliusin his primer (1684); this orthographydispensedwith the letters
<c>, <f>, <q>, <x>, <y>, and <z>, distinguishedbetweenii and e, but still
followed the Swedish and German method of distinguishing between
short andlong vowels in openinitial syllablesby doublingthe consonant
after a short vowel. As a result, word-pairs suchas samma[sama] 'the
same' and sama [sa:ama] 'to get' were distinguished,but intervocalic
single and geminateconsonantswere not disinguished:e.g. the spelling
warras renderedambiguouslyboth [varaz] 'thief and [vai'raz] 'rod'.
118 ESTONIAN

4 The New, or so-called 'Finnish' orthography was fIrst proposedfor

Estonianby the Finn A. J. Arwidsson (1822); it was then recommended
by Eduard Ahrens (1803-63) in the fIrst edition of his grammar
(Grammatikder EhstnischenSpracheRevalschenDalektesI, 1843) and
fIrst applied in Gustav Heinrich Schiidloffel's booklet ToomasWesten,
Lapo rahwa uso iirataja Norra maal (1844), and then in the secondand
full edition of Ahrens'sgrammar(1853). In this orthography,short and
long vowels are written by means of one and two vowel letters,
respectively;similarly, intervocalic single and geminateconsonantsare
written by meansof oneandtwo consonantletters.The New Orthography
resembled preceding orthographies in not reflecting the distinctive
palatalizationof dentalconsonantsandthe distinctive syllabic quantityof
long syllables (with one exception: the two syllabic quantities of long
syllables which have a short monophthong followed by a strong
intervocalic geminatestop or ss are indicated by writing the shorter v.
longergeminatesby meansof onev. two consonantletters).

The Alphabet
The Estonianalphabetincludesthe letters <a b d e f g h i j kIm n 0 p r s 5
z 1. t u v 0 a0 ii>; note the aberrantposition of <z> (and <1.». The letters <f z
1.> occur only in late borrowings.Similarly the letters <b d g> at the beginning
of a word are a sign that the word is a late borrowing; in all vocabulary,initial
<b d g z 1.> standfor strong voicelessconsonants[p t k s 5], elsewherethey
standfor weakvoicelessconsonants[B D/rJ G Zzi 1.]. The letters<p t k 5> stand
for strong voicelessconsonants;intervocalicallythey rendershort geminates
[pp ltlVt kk 55] when they follow a short monophthongin a stressedsyllable,
and for the most part they renderlong geminates[p:p t:t k:k 5:5] when they
follow a long monophthongor a diphthong in a stressedsyllable; they
representa geminateof ambiguouslength when they follow the vowel of an
unstressedsyllable. The letter <s> standsfor the strong voicelesssibilant [s]
at the beginningof a word, at the beginningof a consonantcluster,and after
an obstruent;it standsfor a weak voicelesssibilant [zM] intervocalically, at
the endof a clusterwhenprecededby a sonorantconsonant,andafter a vowel
at the end of a word. In renderingforeign namesborrowedfrom languages
using Latinscripts,Estonianorthographyfollows the sourcestrictly.
The letter <0> standsin standardEstonianfor a mid or, if of Q3 (seebelow),
optionally an uppermid unroundedbackvowel.

Phonology

The Syllable
A syllablein Estoniancontainsat leastonevowel optionally precededby 1-3
consonantsand optionally followed by 1-4 consonants.Except in the word
 ESTONIAN 119

praegu 'now' « *para aikoihen),an initial consonantclusterin a first syllable

signalsthat the word is descriptive,borrowed,or both. Stressedsyllablesmay
be either short or long. A short stressedsyllable has one single short
monophthong,optionally precededby consonants;a long stressedsyllable
ends in a consonant,long monophthongor diphthong. Stressedsyllables
distinguishthreesyllabic quantities,Q1, Q2 or Q3: seepp. 119-21.

Stress
In words which havestress,the primary stressis usually on the first syllable;
someinterjections,e.g. aitiih [aittIDlh] 'thanks',and a numberof borrowings
have the primary stress on a non-first syllable, e.g. idee 'idea', [id~] 'idea',
militaristlik [mi:littarj};t1ik] 'militarist' (adj)' (primary stress underlined,
secondarystressmarked with colon). There is a strong tendencytowards
trochaic stress-patterning,and to dactylic word-final stress groups if the
trochaic patterning cannot be applied; as a result, secondarystressesare
usually on odd syllablescountingfrom the stressedsyllables:vallandatavate-
legi [v~IlanDa:ltava:lsyllables
tele:Gi] counting 'even to ones being discharged',konelesime
speaking 'we spoke'; this tendencyis counteractedby certain deriva-
[k~nele:zime]
tional affixes that attract the secondarystress: elajalik ~laelajalik jali:k] 'beastly',
konelemise[k~nelemi:ze] speaking 'of speaking(sG)'.

Quantity
In all forms of EstonianexceptCoastaland Northeast,quantity is distinctive
on two levels. First, the segmentalquantity correlation SHORT v. LONG
holdsfor vowels andconsonants.In standardEstonianall vowels in primarily
stressedsyllablesand most consonants(at least aftershort monophthongsof
primarily stressedsyllables) have both short and long counterparts.Second,
there exists in stressedsyllablesa tripartite correlationof distinctive syllabic
quantities, usually referred to as quantity 1, quantity 2, and quantity 3
(hereafterQ1, Q2, and Q3). This correlationis based(1) on the contrastof
SHORT and LONG stressedsyllables,and (2) on the contrastof LIGHT and
HEAVY stressedsyllables:

I Syllable length I Syllable weight I

Syllabic quantity Ql short light
I--
Q2 long
I--
Q3 heavy

A stressgroup, a stem, a suffix, or a word consistingof one stressgroup can

be referredto as oneof Q 1, Q2, or Q3 dependingon the quantityof its stressed
syllable.
A stressedsyllable is short and, hence, of Ql, if it ends in a short
120 ESTONIAN

monophthong,e.g. the e of elu 'life';

~lu] 'life'; otherwisethe stressedsyllable is
long. A heavy stressedsyllable, or a syllable of Q3, can composea stress
group on its own, e.g. puu [py] 'tree'. A light syllable can composea stress
group only when followed by at least one unstressedsyllable; nevertheless
both light and heavy stressedsyllables can be followed by at most two
unstressedsyllables.Below, in words written in the conventionalorthography,
all light syllablesof Q 1and Q2 are markedwith an acuteaccent0 andheavy
syllablesof Q3 with the graveaccentO.
A syllable of Q2 has a full-long monophthong, a diphthong both
componentsof which are short, or endsin one or two short consonantsor the
short initial componentof a geminate,e.g. pUnad [piinan] 'pains', 'saate
[saalte] 'you (plur) get', 'auto [aiilto] 'car', 'naeris [naeriz] 'rape(bot)', 'enne
[eiine] 'omen', 'salgas [saIGaz] 'in the band', 'kaarte [kaarlte] 'arch (pG)',
neinte [helnlte] 'hay (pG)'. Note that exceptfor someforeign propernames,
the co-occurrenceof a long monophthongor a diphthong with a geminate
obstruentin a word with a syllable of Q2 is restricted(1) to genitive plural
forms of somenounsand (2) to the second-person plural present-tense forms
of monosyllabicvocalic verb stems,both of which havethe suffix -teo
A syllableof Q3 hasan overlongmonophthong,a long diphthong,or a long
consonantor long consonantcluster,cf. 'piinad [pi:inad] 'you (sing) torment',
'saate [sa:atte] 'transmission(sG)" 'autu [auttu] 'dishonest','naeris [naeriz]
'(s)he laughed', 'kallas [kallaz] 'it poured', 'enne [eime] 'before', 'salgas
[salGaz] '(s)he disavowed', 'kaarte [ka:artte] 'card(pP)'. SyllablesofQ3 may
have a more complicatedstructure than light long syllables, e.g. clusters
consistingof three or more consonantsCtursk 'cod', 'vintskleb '(s)he rolls
about'); the occurrenceof certaindiphthongsis restrictedto syllablesof Q3,
e.g. 'kaotus[kaottuz] 'loss'.
Estonianhas both short and long vowels and consonants(or geminates)
co-occurring with long vowels and consonants(or geminates)in syllables
of different types, and Estonian words are frequently, even usually,

Figure 4.1 Homomorphemic monophthong and stopsequences

Ql lagi! makkl 2 maioo3 mak4

Q2 magi5 makld.6

Q3 mag}? mitkki 8 mag9 mak!O

Key: 1 'laGi 'ceiling', 2 'maki 'taperecorder(sG)', 3 'makki 'taperecorder(sP)',4 'makk 'tape

recorder',5 'maagi 'ore (sG)', 'magician(sG)', 6 'Maacki 'Maack(sG)', 7 'maagi 'magician
(sP)', 8 'maaki 'ore (sP)','Maacki 'Maack(sP)',9 'maag 'magician',10'maak 'ore','Maack
'Maack'.
Note: In this figure and in Figure4.2, specialdiacritics are usedto indicaterelativedurationof
segments.Listed (with the vowel a) in orderof increasingquantity, theseareii a 11 aa aii.
 ESTONIAN 121

Figure 4.2 Monophthongand stop sequencesconditioned by the morpheme

boundary

makkl
Q3 makkl makkl
makkl makkl

Key: 1 'makki 'evenalthe taperecorder',2 'maagki 'evenalthe magician',3 'maakki 'eventhe

ore', 'Maackki 'evenMaack',4 'maagi 'eventhe country', 5 'maake'country (diminutive),.

polymorphemic.As a result, Estonianhas a complex systemof quantitative

patterns,especiallybecauseof complicatedsequencesof monophthongplus
obstruent:cf. Figures4.1 and4.2 (notethat in a morpheme,a long or geminate
resonantoccursonly after a shortmonophthong).

Tone
In addition to the featuressketchedabove,the contrastof Q 1, Q2, and Q3 is
usually supportedby differencesin the position of the peak of pitch in the
stressedsyllable: a syllable of Ql has a prevailingly rising pitch, followed by
a shortfall, a syllable of Q2 hasa rising-falling pitch and a syllable of Q3 has
a level or a falling pitch, sometimesprecededby a shortrise.

Vowels

Monophthongs
In flrst, primarily stressedsyllables of native words, StandardEstonianhas
short and long monophthongsof nine qualities:

i ti u
e 6 e 0
a a

In standardorthography,all long monophthongsare written double; they

can occurin syllablesof both Q2 and Q3. The vocalismof non-flrst syllables
is restrictedto four shortmonophthongs:i, e, a and u.
In severaldialects,the numberof long monophthongshasbeenreducedas
a result of the breaking (diphthongization)of long monophthongs.In West
Saaremaaand East Hiiumaa the system of short monophthongshas been
reducedas a result of the mergerof if and 0; this mergeris somehowrelated
to the labializationof long monophthongs*ee ) rere and *aa ) aa.

First syllable Non-flrst syllable

i ti u ii titi uu i u
e 6 0 ee 66 00 e
a a aa rere aa a
 122 ESTONIAN

In SouthEstonian,the numberof long monophthongsis the highestin Voru

SouthEstonian,wherethecharacteristiclong uppermid monophthongsii\ ii,ii,v,
uuv and high backunroundediTV occur only in syllablesof Q3, and ee never
occursin syllablesof Q3. In the RapinaandNorth SetusubdialectsofVoru, the
vocalismof non-firstsyllablesis therichestfound in anEstoniandialect:

First syllable
Ql Q2 Q3
6 ti 'i u ii titi uu ii titi 11 uu
ii v titi V uuv
e 6 e 0 ee 66 ee 00 ee 66 ee 00
aa 6 aa aa aa aa
Non-first syllable
i ti u
e 6 e 0
aa 6

Diphthongs
Diphthongsin standardEstonianend in u, 0, a, e, or i. All diphthongsoccur
in syllablesof Q3; diphthongsendingin u and i (and a few others)also occur

Figure4.3 StandardEstonianlong monophthongsanddiphthongs

u o a a ii a a e a
u uu2 [u02] [ua] [ue] ui 2

o ou2 00 2 oa oe2 oi 2

a
o au2 ao aa2 ae2 ai 2

a 5u2 50 5a 55 2 5e2 5i 2

iio [iio] [iia] iiii 2 [iie] iii 2

a oa 002 oe oi 2

a au2 ao aa2 ae2 oi2

e [eu2] eo ea2 ee2 ei 2

o iu 2 rio] rial riel jj2

Note: The index number2 indicatesthat the sequenceoccursin addition to syllablesof Q3 also
in syllablesof Q2. Diphthongsthat occuronly in borrowingsare given in brackets.The vowel if
is herewritten with <6>.
 ESTONIAN 123

in syllablesof Q2, cf. Figure 4.3. In non-first syllablesof native words, only
the diphthongsui, ai, and ei occur; in such casesi is either a pluralizer or
belongsto the superlativesuffix -im.
In Estoniandialectswhich have vowelharmonythere are also diphthongs
ending in U, viz. ou, au, eu. In Coastal,Northeastand northernmostNorth
Estonian dialects, as a result of the breaking of long monophthongs,the
diphthongsuo, ie, and uo occur insteadof the 00, ee, and 00 of standard
Estonian;in EastEstoniansuchbreakingwas limited to long mid vowels of
Q3.

Consonants
In standardEstonianeighteenconsonantqualitiesarephonemic:

v j
f h
s sj S
P t ~ k
m n ni IJ
I lj
r

The orthography does not distinguish the palatalized dentals from the
unpalatalized ones, and writes IJ as <0>. Long intervocalic and long
postvocalicfinal consonantsare written double in syllables of both Q2 and
Q3, with the exceptionof long intervocalicstopsandlong sin syllablesof Q2
and after a vowel other than a short monophthong,even in syllablesof Q3.
The shortand weakvoicelessstopsare written with <b d g>.
Consonant palatalization in standard Estonian and in most Estonian
dialectsis manifestedphoneticallyas a weak palatalizationof the initial part
of a denti-alveolar consonantor the presenceof a very short i-coloured
transition between the vowel of the stressedsyllable and the following
consonant; in view of this, any system of transcription (including the
traditionalFU ones)in which consonantpalatalizationis indicatedby an acute
accentwritten aboveor aftertheconsonantletteris somewhatmisleading.Thus
the pronunciationof the 'phonetically'written kuJiJi for 'kull 'hawk' is actually
[kuill]. When occurring after original long non-high back vowels, such
transitionshavedevelopedinto vocalic segmentsin Insularandlocally alsoin
WestNorth Estonian,e.g. laesfor klaas'glass'. In Voru SouthEstonian,where
consonantpalatalization is manifestedas the thorough palatalizationof a
consonantor a consonantcluster and a word-final palatalized consonant
receivesaj-colouredterminalpart,almostall consonantscanbepalatalized,cf.
vaSik'calf: vaSikj 'copper'.CoastalandNortheastEstonianandHiiumaaNorth
Estonianhaveno palatalizedconsonants.
Consonantclusters in Estonianconsist of from two to five consonants.
124 ESTONIAN

Most clustersoccuronly after the vowels of the syllableswith primary stress.

The occurrenceof certain clustersdependson the precedinglong or short
vocalismandon syllablequantity.Clustersbeginningin a weakvoicelessstop
b d g never follow a short vocalism of a syllable of Q3. Severalclusters,
amongthem biconsonantalclustersendingin a resonantand all clusterswith
s after a weak stop, always contain a morphemeboundaryor have arisenas
a result of vowel syncope('kiiiiniad 'candles',cf. 'kiiiinal 'candle'; 'kaablid
'cables',cf. 'kaabel 'cable', 'vord'len 'I compare', 'kuld'ne 'golden': sG
'kuld'se.
Morphophonological Alternations
Estonianis typologically more inflecting than agglutinating.It has numerous
morphophonologicalalternations,most of which, however,occurin a certain
restrictedsetof stems.
Gradation
Gradation in Estonian includes (1) quantitative gradation of long stressed
syllables, whereby Q3 in the strong grade alternateswith Q2 in the weak
grade, and (2) qualitative gradation, whereby an initial weak stop of the
second syllable, or t and k in the strong grade, are lost, assimilated or
weakenedinto v or j in the weak grade.Quantitativegradationis reflectedin
the orthographicalalternations<pp : p>, <tt: t>, <kk: 10, <ss: s> after a short
vocalism(note thatp, t, k, andsafter a shortvocalismrendershortgeminates)
and<p : b>, <t : d>, <k : g>, <ss : s> after a long vocalismor a resonant;there
are also cases of quantitative gradation which are not reflected in the
orthography.For qualitative gradation,t and k occur in the stronggradeonly
after the consonantss andh.
Although in most cases of gradation the strong grade is historically
primary, gradationsare classifiedon synchronicmorphologicalgroundsinto
(1) weakeningand (2) strengtheninggradations.For weakeninggradations,
the nominativeand partitive singularforms of nominalsand the da-infinitive
form of verbs are in the strong gradewhereasthe genitive singular form of
nominalsandthe first-personsingularform of the presentindicative are in the
weak grade; for strengtheninggradation the distribution of grades is the
reverse,cf.:
'Weakening'gradation:
Stronggrade Weakgrade
sN 'niigu 'face',sP 'niigu sG 'niio
sN 'jalg 'foot', sP 'jalga sG jala
sN 'viilk 'lightning', sP 'viilku sG'viilgu
sN 'vill 'wool' , sP 'villa sG'villa
sN 'keel 'tongue', sP 'keelt sG'keele
inf 'siduda 'to tie' s 1pres 'sean
inf 'anda 'to give' slpres'annan
 ESTONIAN 125

'Strengthening'gradation:
Weakgrade Stronggrade
sN 'tahe 'will', sP 'tahet sG'tahte
sN 'voti 'key', sP 'votit sG'votme
inf 'ommelda'to sew' s 1pres'omblen
inf 'iirgata 'to wakeup' s 1pres'iirkan
inf 'kallata 'to pour' s 1pres'kallan

Note that nominal stems with weakeninggradation have their partitive

singularin the strong gradeand are either (1) monosyllabicand markedwith
the case ending -tt (after a short monophthong)or -t (elsewhere)or (2)
bisyllabic and for the most part morphologically unmarked. Stems with
strengtheninggradationhave their partitive singular in the weak grade; it is
bisyllabic and marked with the case ending -to For both weakening and
strengtheninggradationall partitive plural forms of nominals and all supine
illative forms of verbsarein the stronggrade.
Alongsidecasesof an easilyrecognizablesingle-stopmutation,qualitative
gradationis also definedas characterizingcertaintypes of nominalsin which
a single stop *t in the strong-gradeforms has undergoneseveralchangesso
that it hasbeeneliminatedin mostor all environments,e.g.

sN sP sG sIll pG pP Gloss
1 'kasi 'katt 'kae 'katte 'kate 'kasi 'hand'
, , , ,
2 uus uut 'uue uude 'uute uusi 'new'
3 '66s '66nt '66ne '66nde '66nte '66si 'cavity'
, , ,
4 vars 'vart 'varre varde 'varte varsi 'shaft'

In suchcases,the sequence*ti hasbecome*si in the sN and pP forms and

undergoneapocopein the sN forms of (2), (3), (4). In addition, in the sP and
pG forms earlier stem-final *t fused with case-ending-initial*t to yield new
case endings. The short single stop has been geminated after a short
monophthongin first illative forms as seenin stem-types(2)-(3); however,
in thesetypes the stop d has been treatedas belonging to the caseending.
Thesenew sP,sIll andpG endingsoccure.g. in the following paradigmwhere
quality gradationdid not originally occur:

5 'keel 'keelt 'keele 'keelde 'keelte 'keeli 'tongue'

VowelApocope
In bisyllabic nominal stemswith a long first syllableandweakeninggradation
and in most trisyllabic stemsthe stemvowel is apocopatedin the nominative
singular,cf. sN 'sepp'smith', sP 'seppa,sG 'sepa; sN 'ring 'circle', sP'ringi,
sG 'ringi; sN Jumal 'god', sP Jumalat,sG Jumala.
126 ESTONIAN

VowelSyncope
In nominalswith strengtheninggradation,the vowel of the secondsyllable,
presentin weak-gradefonns, is syncopatedin strong-gradefonns, cf. 'tatar
'daughter':sG 'tatre; 'kamme 'ten': sG 'kamne; 'katel 'kettle': sG 'katla;
'ainus 'single': sG 'ainsa; 'motelda 'to think': 'motlen 'I think'. Syncopealso
occursin nominalswhich havea bisyllabic stemof Q3 that endsin a resonant
or, in a-stems,in the sibilant s, e.g. 'ankur 'anchor': sG 'ankru; 'tahtis
'important':sG 'tahtsa.

Stem-vowelAlternation
In addition to alternationsconditioned by apocopeor syncope of a stem
vowel, therealso occurthe stem-vowelalternationsi - e andu - o.
Bisyllabic nominalswith a shortfirst syllable havethe stemvowel i in the
nominativesingularinsteadof the stemvowel e, which is presentin mostcase
fonns, cf. 'tuti 'fire': sG 'tule, sEla 'tulest; 'nimi 'name':sGP 'nime; 'magi
'hill': sG 'mae,sP 'mage.
For bisyllabic stemswith qualitativegradationthe stemvowels u and,after
a long first syllable, the i of stronggradefonns alternatewith stemvowels 0
and e of thoseweak-gradefonns that havelost the single stopd or g, cf. 'tigu
'snail': sG 'teo; 'tegu 'deed': sG 'teo; 1ugu 'story': sG '100; pood'shop':
sP poodi: sG poe; 'saagida 'to saw' : 'saen 'I saw'.

Lowering ofShortHigh Vowels

As a result of the loss of b d g in weak-gradefonns, the short high vowels
i u a of the first syllable are loweredto the mid vowels e 0 0 before0, a, and
e, e.g. 'oad 'beans'Cuba 'bean'), poab 'cuts the hair' (pagada 'to cut the
hair') 'seon 'I bind' Csiduda 'to bind').

-ne " -s(e) Alternations

The suffix alternation -ne (sN): -s- (sP): -se(-) (elsewhere) occurs in
adjectives,in the ordinal numeralsesimene'first', teine 'second'andin nouns
=
(in complex suffixes such as lane, =tine, =mine), cf. 'sakslane 'a/the
Gennan':sP 'sakslast:sG 'sakslase.

Geminationin the First Illative (Illative-I)

In bisyllabic nominal stems with a short first syllable the intervocalic
consonantis geminatedin the illative-l singular or the 'short' illative, e.g.
'veri 'blood': 'verre 'into the blood', 'udu 'fog': 'uttu 'into the fog'. See
Nominal Inflection, p. 127.

se-Syncopationin Illative
A postvocalic stem-final se-sequenceis optionally syncopatedbefore the
illative ending-ssein a trisyllabic stemof Q3 or in a quadrisyllabicstem,e.g.
'rasku'sesse- 'raskusse,Juma1usesse - Juma'lusse,'ini'mesesse- 'ini'messe
from 'raskus'difficulty', Jumalus'deity', 'ini'mene'man,humanbeing'.
 ESTONIAN 127

Plural StemFormation
In monosyllabicvocalic stemsending in a long monophthongother than iiii
or ii, the long monophthongis shortenedbefore the pluralizer -i-, e.g. puu
'tree': pP puid, 'soo 'swamp':pP 'soid, 'tOo 'work': pP 'toid, i'dee 'idea': pP
i'deid.
For a set of bisyllabic vocalic stems with a short fIrst syllable or with
weakeninggradationthe so-calledstemplural is possible:in such casesthe
vowel is replacedby anothervowel, mostly in the partitive plural, but also in
otherplural casefonns accordingto the following scheme:

u > e, e.g. 'vork 'net', sP 'vorku, pP 'vorke;

i> e, e.g. 'varv 'colour', sP 'varvi, pP 'varve;
e> i, e.g. )arv 'lake', sP )arve, pP Jarvi;
a> u, after fIrst-syllable a(a), i(i), 0(0), ei, or iii, andafter lj after fIrst-syllable
e or a, e.g. kala 'fIsh', pP kalu; 'vaal 'whale',sP 'vaala, pP 'vaalu;
a> i, exceptafterj, after fIrst-syllable e, a, 0 or ii, or if there is 0(0) or u(u)
in the precedinglong fIrst syllable,e.g. 'koer 'dog', sP 'koera, pP 'koeri;
a > e afterj and i if there is an 0, u, or ii in the fIrst syllable or if there is an
o or u in the precedingshortsyllable,e.g. ori 'slave',sP 'orja, pP 'orje.

Morphology

Nominal Inflection
In Estonianas elsewherein Fennic,nominals,i.e. nouns,adjectives,numerals
and pronouns,are mostly inflectedin numberandcase.
There are in EstonianfIfteen nominal cases,in both singular and plural.
Three are grammatical cases, namely nominative (N), genitive (G), and
partitive (P); the remainingtwelve are adverbialcases,namelyillative-l (111-
1) (sometimescalled 'additive'),illative-2 (111-2), inessive(lne), elative (Ela),
allative (All), adessive(Ade), ablative (Abl), translative(Trans), terminative
(Tenn),essive(Ess),abessive(Abe), andcomitative(Com).
In a departurefrom tradition, here 111-1, from the proto-Fennicillative in
*-hen, which is in modem Estonianmorphologically unmarked(exceptfor
monosyllabicvocalic singularstems,whereit hasthe endings-ha, -he, -hu), is
treatedseparatelyfrom the innovative111-2, (the 'long' illative) in -sseor -de.
The 'short' illative cannotbe fonnedfrom severalstructuraltypesof nominal
stems;the 'long' illative is usedinstead,includingadjectivalattributesof nouns
in 111-1.Nevertheless,whenthereexistin a paradigmbotha 'short' anda 'long'
illative fonn, thefonneronecanbereplacedby theinterrogativepronoun'kuhu
'where to', whereas thecorresponding'long' fonn can be replacedby the
pronoun'mil/esse'in(to) what'; the choicebetweenthesetwo fonnsis not free,
but dependsratheron verb government.Thus,for example,it is possibleto say
1ahen 'soja'vakke'I go into the army', 1ahen'kooli 'I go to school', 'votan
128 ESTONIAN

Figure4.4 Interrelationsof caseforms of weakeningstems:jalg'foot'

e-Plural e-Plural Stemplural e-Plural

N 'jalu 'jalu de

P 'jalu 'jalu dest 'jalu

Ill-I 'jalu 'jalu

<"""""VvVv>
G ' jala 'jalga de - jalg! e
111-2 ' jala sse 'jalga'jalga ::::::'f:::
Ine 'jala s 'jalga des 'jalu s j' jalg! es
Ela ' jala st 'jalga dest 'jalu st j' jalg! est
All ' jala Ie 'jalga dele 'jalu Ie j' jalg! ele
Ade 'jala I 'jalga del 'jalu I j' jalg! el
Abl ' jala It 'jalga deIt 'jalu It j' jalg! eIt
Tr ' jala ks 'jalga deks 'jalu ks L ...... ~
Term ' jala ni 'jalga deni 'jalu ni
Ess ' jala na 'jalga dena
Abe ' jala ta 'jalga deta 'jalga
Com ' jala ga 'jalga dega 'jalga

'kiitte 'Itakeinthehand', pistan 'suhu 'I stickinto themouth', 1wstab'korva 'it

soundsin (one's)ear', 1wstab'korvu 'it soundsin (one's)ears', but onecannot
say *liihen 'soja'viiesse,*liihen Koolisse, *'votan 'kiiesse,*'pistan 'suusse,
* KostabKorvasse,* Kostab'korva'desse> Notethatoneusesthe 'longfonns'in
constructionssuchas'suhtun'vaenu1ikult 'soja'viiesse'I haveahostileattitude

Figure4.5 Interrelationsof caseforms of strengtheningstems:hammas'tooth'

Singular de-Plural i-plural Stemplural

N 'harnmas 'ha;Dbull
P 'harnmast ri'hambaid
111-1 'hambull

G 'hamba 'harnmaste
111-2 'hamba sse 'harnmastesse 'hamba isse
Ine 'hamba s 'harnmastes 'hamba is 'ha;Dbulls
Ela 'hamba st 'harnmastest 'hamba ist 'ha;Dbullst
All 'hamba Ie 'harnmastele 'hamba ile
Ade 'hamba I 'harnmastel 'hamba il
Abl 'hamba It 'harnmasteIt 'hamba ilt
Tr 'hamba ks 'harnmasteks 'hamba iks
Term 'hamba ni 'harnmasteni 'harbull ni
Ess 'hamba na 'hammastena 'hamba ina
Abe 'hamba ta 'hammasteta 'ha;Dbullta
Com 'hamba ga 'hammastega
 ESTONIAN 129

towardsthe army', see'puutub 'kiiesse 'it concernsthe hand'. As a result of

the restricted functions of 111-1, many nominals have no 111-1 form for
semanticreasons.On the other hand,in this chapterthe illative in -de, often
held to be a 'short' illative, since it has been abstractedfrom certain 111-1
forms (cf. 'oondefrom 'oos 'cavity', sG 'oone, wherethe -de was originally
a part of the stem), is treatedas different from 111-1. The illative in -de is
beginningto replacethe illative in -ssein bisyllabic stemswhosestemvowel
e follows a resonantprecededby a long vocalism, cf. 'keeldeand 'keelesse
from 'keel 'tongue;language'.
Apart from 111-1, the remainingelevenadverbialcasesof Estonian(111-2,
Ine, Ela, All, Ade, Abl, Trans, Term, Ess, Abe and Com) can be formed by
adding a caseendingto the genitive singularform to yield the singularform,
and to the genitive plural form to yield the plural form. In addition, the
commongenitive plural form can be formed by adding the ending -de or -te
to the genitive singularform (this is not the casefor most pronouns).Among
the grammaticalcases,the nominative singular(sN) and genitive singular
(sG) are always unmarked,while the partitive singular (sP), partitive plural
(pP), and genitive plural (pG) have severalendingsor, in the casesof the sP
andpP, may be unmarked;seeFigures4.4 and4.5.

Plural Formation
There are severaltypes of plural stem formation. The most common is the
so-calledde-plural,basedon the plural genitiveform. This is the only method
availablefor nominalswhich take only the ending-sid in the plural partitive;
for nominalstaking the pP ending -id the i-plural is also possible.For many
nominalshaving in their partitive singular anunmarkedbisyllabic form, the
stemplural is possiblealongsidethe de-plural.In suchcasesthe stemvowel
of the plural stemis different from that of the singularstem accordingto the
generalschemegiven underPlural StemFormation,p. 127.
The different types of plural formation have different stylistic values and
the stem plural, in particular, is subject to several restrictions, causing
complicatedalternationsandhomophony.

CaseFormation
Nominative, genitive, partitive, and illative-l forms are often distinguished
only by stemalternationsthat can be presentalso in morphologicallymarked
case forms. A noun can have from one to six stem allomorphs. The
interrelation of stem allomorphs in the paradigm depends also on the
deploymentof weakeningv. strengtheninggradation;seeFigures4.4 and4.5.

Comparative and Superlative of Adjectives

Adjectives take the comparativesuffix -m in the nominative singular and
-ma(-) in other caseforms; the suffix regularly follows the usual stemvowel
(transparentin sG forms) for u-, i-, ande-stemsandfor trisyllabic a-stemsand
bisyllabic a-stemsof Q3. In bisyllabic a-stemsof Q2, the stem vowel a is
130 ESTONIAN

Table4.1 Comparativeand superlativeof adjectives

sN pP Comparative Superlative Gloss

'rahulik 'rahu'likke 'rahu1ikum 'rahu1ikem 'peaceful'

'niiri 'niirisid 'niirim 1<.oige'niirim 'dull'
'sinine 'siniseid 'sinisem 'siniseim 'blue'
'uhke 'uhkeid 'uhkem 'uhkeim 'proud'
'suur 'suuri 'suurem 'suurim 'big'
'madal 'madalaid 'madalam 'madalaim 'low'
'rikas 'rikkaid 'rikkam 'rikkaim 'rich'
'must 'musti 'mustem 'mustim 'black'
'laisk 'laisku 'laisem 1<.oige1aisem 'lazy'
'vana 'vanu 'vanem 'vanim 'old'
1<.ena 1<.eni 1<.enam 1<.enim 'nice'

regularly replacedbye, in stemsof Q1 this is a rarer feature.The adjective

'hea 'good' has the suppletive comparative form parem 'better'. The
superlativesuffix is -im in the nominative singularand -ima in other case
forms; its occurrenceis regularfor adjectiveshaving the plural partitive form
ending in -id or -i and in derivativesbuilt with =ik and =lik; the superlative
suffix is -em and -ema.
Alongsidethe suffixal superlativesfor all comparableadjectivesan analyt-
ical superlativeform consistingof the sequence'koige plus the comparative
grade formis possible;Mige is thegenitivesingularform of 'kOik 'all' .
The adjectives 'alumine 'lower', 'U/emine 'higher', 'pealmine 'upper,
higher, 'esi'mene'first, foremost,front', 'viimane 'latter, recent', 'tagu'mine
'back', 'viili'mine 'outer, oppositefrom inside', 'iiiirmine 'outer,farther from
the middle' have no comparativegrade forms but may have the analytical
superlativegradeforms.

Numerals
The Estonianunderivedcardinalnumeralsandtheir patternsof declensionare
shownin Table4.2.
The correspondingordinal numeralsare mostly derivedfrom the genitive
singular stem; in the courseof derivation the vowel e is substitutedfor the
stem vowel i of certain cardinal numerals.The ordinal kolmas 'the third' is
exceptionalin that it addsthe derivationalsuffixesto the stemvowel a instead
of e. The ordinalsesimene'the first' andteine 'the second'are suppletive.
The formation of the cardinal numerals from 11 to 19 is based on
subtraction,e.g. 'iiks'teist'kiimmend'11' (i.e. 'one of the secondten'), 'kaks-
'teist'kiimmend'12', where 'kiimmendis an otherwiseobsoletepartitive form
of kiimme '10'; the correspondingordinals are 'iihe'teist'kiimnesand 'kahe-
'teist'kiimnes.The cardinal numeralsfor tens, hundreds,thousands,millions
etc. are basedon multiplication, but with singular morphology, e.g. 'kaks
 ESTONIAN 131

Table 4.2 Cardinalnumerals

sN sG sP sIll-2

o 'null 'nulli 'nulli 'nulli

1 'tiks 'tihe 'tiht(e) 'tihte
2 'kaks 1cahe 'kaht(e) 'kahte
3 'kolm 1colme 'kolme 'kolme
4 'neli 'nelja 'nelja 'nelja
5 'viis 'viie 'viit 'viide
6 'kuus Kuue 'kuut 'kuude
7 'seitse 'seitsme 'seitset , seitsmesse
8 1caheksa 1caheksa 1caheksat 1cahek'sasse
9 'tiheksa 'tiheksa 'tiheksat 'tihek'sasse
10 1ctimme 'ktimne 1ctimmet 'ktimnesse
100 'sada 'saja 'sadat 'sajasse
1,000 'tuhat 'tuhande 'tuhandet 'tuhan'desse
106 'miljon 'miljoni 'miljonit 'miljo'nisse
109 'miljard 'miljardi 'miljardit 'miljar'disse

Table 4.3 Ordinal numerals

sN sG sP I11-2 pG

1 'esimene 'esimese 'esimest 'esi'messe 'esi'meste

2 'teine 'teise 'teist 'teise 'teiste
3 1colmas 1colmanda 1colmandat 1colman'dasse 1colman'date
4 'neljas 'neljanda 'neljandat 'neljan'dasse 'neljan'date
8 1caheksas 1cahek'sanda1cahek'sandat1cahek'sandasse1cahek'sandate
1,000 'tuhandes 'tuhan'denda'tuhan'dendat 'tuhan'dendasse'tuhan'dendate
106 'miljones 'miljo'nenda 'miljo'nendat 'miljo'nendasse 'miljo'nendate

'sada '200', 'kaks 'tuhat '2,000','kaks 'miljonit '2,000,000'.Teens,tens,and

hundredsmake compounds,and their declensionpattern incasesother than
nominative, genitive, and partitive differs from that of numeralsfor thou-
sands,millions etc., e.g.:

12 200 2,000

sN 'kaks'teist(1ctimmend) 'kaks'sada 'kaks 'tuhat

sG 'kahe'teist(kiirnne) 'kahe'saja 'kahe'tuhande
sP 'kahfteist(ktimmend) 'kaht'sada 'kaht(e) 'tuhat
sIll 'kahe'teisfkiirnnesse 'kahe'sajasse 'kahte'tuhandesse
sIne 'kahe'teisfkiirnnes 'kahe'sajas 'kahes'tuhandes

With the exception of the informal fractional numerals 'pool 'half and
132 ESTONIAN

'veerand'quarter',simple fractional numerals,i.e. denominators,are derived

on the analogy of ordinal numerals by means of the suffix =ndik, e.g.
'kahendik 'a half; 'kolmandik 'a third' 'neljandik 'a quarter'. Fractional
numeralswith numeratorsare inflected similarly to cardinalsfor thousands,
millions, etc. e.g.

One-third Two-thirds
sN 'tiks 'kolmandik 'kaks'kolman'dikku
sG 'tihe 'kolman'diku 'kahe'kolman'diku
sP 'tiht(e) 'kolman'dikku 'kaht(e) 'kolman'dikku
sm 'tihte 'kolman'dikku 'kahte'kolman'dikku
sIne 'tihes 'kolman'dikus 'kahes'kolman'dikus

Table4.4 Personalpronouns

1 2 3

Singular
N 'mina - rna 'sina- sa 'tema- ta
G 'minu - mu 'sinu - su 'tema- ta
P 'mind 'sind 'teda
,
I11-2 minusse- 'musse 'sinusse- 'susse 'temasse- 'tasse
Ine 'minus - mus 'sinus- sus 'temas- tas
Ela 'minust- must ,sinust- sust 'temast- tast
All 'minule - 'mulle 'sinule - 'sulle 'temale- 'talle
Ade 'minul- mul 'sinul- sui 'temal-tal
Abl 'minult - mult ,sinult - suit 'temalt- talt
Trans 'minuks 'sinuks 'temaks
,
Term minuni 'sinuni 'temani
,
Ess minuna 'sinuna 'temana
,
Abe minuta 'sinuta 'temata
Com 'minuga- muga ,sinuga- suga 'temaga- taga
Plural
N 'meie - me 'teie - te 'nemad- nad
G 'meie - me 'teie - te 'nende
P 'meid
,
'teid 'neid
I11-2 meisse 'teisse 'nendesse- 'neisse
,
Ine meis 'teis 'nendes- 'neis
Ela 'meist 'teist 'nendest- 'neist
All 'meile 'teile 'nendele- 'neile
Ade 'meil 'teil 'nendel-'neil
Abl 'meilt 'teilt 'nendelt- 'neilt
Trans 'meiks
,
'teiks 'nendeks- 'neiks
Term meieni 'teieni 'nendeni
Ess 'meiena 'teiena 'nendena
Abe 'meieta 'teieta 'nendeta
Com 'meiega 'teiega 'nendega
 ESTONIAN 133

PersonalPronouns
Personalpronounshave,in addition to the so-calledlong forms, short forms,
most of which usually occur in unstressedpositions.In the singularthe fIrst-
and second-person short pronounshavetwo stems,one of which occursonly
in the nominative,viz. ma 'I' and sa 'thou'; the other stems,mu(-) and su(-),
occur elsewhere,exceptin the partitive, translative,terminative, essiveand
abessive;similarly, insteadof the third-personplural forms of the illative,
inessive,elative, allative, adessive,andablative,forms exploiting the partitive
stem 'nei- or (in monosyllabicforms also unstressed)nei- can be used.This
stem has been borrowed from the paradigm of the demonstrativeplural
pronoun'need'these'.In the plural, the fIrst andsecondpersonshavethe short
forms me 'we' andte 'you' only in thenominativeandthegenitive;contrastthe
third person,wheretheshortstemform ta occursin severalcaseforms.

PossessivePronoun
The possessivepronoun is oma (substitutableby 'kelle? 'whose?'), e.g.
Jutusta 'oma 1astest!'tell about your children!'; contrastthe adjectiveoma
(substitutableby 'milline? 'which?', but not by 'kelle) as in 'ma ei 'riiiigi
'omadest'ini'mestest'I do not speakabout (my) own people', where 'oma
agreeswith the following nounin caseandnumber(elativeplural).

Reflexive Pronoun
The reflexive pronoun'self hasa suppletiveparadigm.Sampleforms:

Singular Plural
N 'ise 'ise
G 'enese- 'enda 'eneste- 'endi
P 'ennast- 'end 'ennast- 'endid
I11-2 'ene'sesse- 'endasse 'enes'tesse- 'endisse
Ine ,eneses- 'endas 'enestes- 'endis

Demonstratives
The underiveddemonstrativepronounsof Estonianare 'see 'this', 'too 'that',
'sama 'same'.Of the two demonstrativellocativepronouns,most Estonians
managewith justthe fIrst, see'this' which, like thepersonalpronouns,hasboth
long and shortforms, the latter normally unstressed.Historically, the genitive
plural form and all long forms havebeenborrowedfrom the paradigmof the
third-personplural pronounnemad.Theotherdemonstrativellocative pronoun,
'too 'that', is characteristicallyused by South Estonians.Its declensionis
entirely analogousto that of 'see,e.g. sG 'tolle, sP 'toda, pN 'nood, pG 'nonde,
pP'noid. Thepronouns'seeand'samaareinflectedasshownin Table4.5.
Demonstrativepronounsform compoundswith one other, e.g. 'see'sama
'the same','too' sama'that same(one)'.
134 ESTONIAN

Table4.5 Demonstrativepronouns

Singular Plural Singular Plural

, , , ,
N ,
see need ,
sarna ,
sarnad
G selle 'nende ,
sarna ,
sarnade
p 'seda 'neid sarna sarnu
, ,
III sellesse- 'sesse 'nendesse- 'neisse ,
sarnasse 'sarna'desse
,
,
Ine selles- ses 'nendes- 'neis samas sarnades
, , , ,
Ela sellest- sest nendest- 'neist ,
sarnast sarnadest
, ,
All selle1e 'nendele- 'neile ,
sarnale ,
sarna'dele
,
Ade sellel- sel 'nendel-'neil sarnal sarnadel
, , ,
Abl sellelt - selt 'nendelt- 'neilt sarnalt ,
sarnadelt
Trans 'selleks- seks 'nendeks- 'neiks 'sarnaks sarnadeks
, , ,
Term selleni 'nendeni ,
sarnani ,
sarna'deni
,
Ess sellena 'nendena samana sarna'dena
,
Abe 'selleta 'nendeta
, ,
sarnata 'sarna'deta
Com 'sellega nendega sarnaga 'sarna'dega

Thereis also a set of demonstrativeadjectivesformed from demonstrative

pronouns either by derivation or compounding,e.g. 'selline 'such', 'see-
'sugune'such','sama'sugune'of the samekind', 'see'samane'the same'.The
demonstrativeadjective 'nii'sugune 'such, of that kind', is formed from the
adverb'nii 'so, in this way'. The demonstrativeadjectivesareinflectedon the
modelof 'selline and 'see'sugune:
Singular Plural Singular Plural
,
N 'selline 'sellised see'sugune 'see'sugused
, ,
G 'sellise 'selliste see'suguse see'suguste
,
P 'sellist 'selliseid see'sugust 'see'suguseid
,
III 'selli'sesse 'sellis'tesse- see'sugu'sesse'see'sugus'tesse
-
,
,sellis'eisse see'sugu'seisse

InterrogativesandRelatives
Estonianhas the interrogative/relativepronouns'kes 'who', and 'mis 'what',
both of which are usually inflected in the singular; they are inflected in the
plural for specialemphasis.The pronoun'kumb 'which (of the two)' inflects
in the singular or the plural dependingon the singularity or plurality of the
two sets under selection,e.g. 'kumb 'king 'which one of the two shoes'and
'kummad'kingad 'which (pair) of the two pairs of shoes'.The pronouns'kes,
'mis, and 'kumbareinflectedasfollows:
Sg Plur Sg Plur Sg Plur
, ,
N 'kes 'kes mis mis 'kumb 'kummad
G 'kelle 'kellede 'mille 'millede 'kumma 'kumbade
P 'keda 'keda 'mida 'mida 'kumba 'kumbi
III 'kellesse 'kelle'desse 'millesse 'mille'desse 'kumba 'kumba'desse
 ESTONIAN 135

There are also interrogative/relativeadjectives meaning 'which', namely

'milline, 'mis 'sugune,both formed from the pronoun'mis, and the adjective
miiherdune,which is sometimesusedin informal style.
Anothersetof interrogativesandrelativesarecaseforms of thestemku-, e.g.
illative-l Kuhu '(to) where',inessive'kus 'where','kust 'from where', Kuhuni
'as far as where', essive Kuna 'when'. Theseforms occur in appositional
compoundswith several postpositions,e.g. Kuhu'poole 'towards what, in
which direction', 'kus'pool 'in which direction', 'kust'poolt 'from which
direction', 'kusfkaudu'which way, throughwhich point', 'kust'peale'begin-
ning from which point or time', 'kust'saadik'up to which point or time'. The
samestemalsooccursin the interrogative/relativeKuidas 'how, in which way',
e.g. Kuidassa tead?'How do you know?')andin theinterrogative/relative'kui
'how',e.g.'kui vana 'how old' andtheconjunction'kui 'when;if.

Indefinite andNegativePronouns
The pronouns 'keegi 'somebody; nobody', 'miski 'something; nothing',
'mingi 'a, some;no' and 'kumbki 'either; neither(of two)' 'ukski 'at leastone;
none' are indefinite in affirmative sentencesand either indefinite or negative
in negative sentences.Example: 'keegi ei 'sulgenud 'viilis'ust means,
dependingon the context,either 'somebodyhasnot closedthe outerdoor' or
'nobodyclosedthe outer door'. The pronouns'keegi and 'miski are inflected
only in the singular,'mingi (pG mingite,pP 'mingeid), 'kumbkiand 'ukski are
inflectedin both singularandplural. Samplesingularforms:
,
sN 'keegi 'miski mingi 'kumbki 'iikski
,
sG 'kellegi 'millegi mingi 'kummagi 'tihegi
sP 'kedagi 'midagi 'mingit 'kumbagi 'tihtki - 'tihtegi
,
sIll 'kelles'segi 'milles'segi mingisse 'kumbagi 'tihtegi

When usedas negative pronouns, the negativecharacterof the pronounscan

be emphasizedby meansof the preposedword 'mitte, e.g. 'mitte 'keegi. The
compound adjective 'mingi'sugune carries negative meaning only when
precededby the word 'mitte; otherwiseit hasthe meaning'a, some'.
Somepronounsare purely indefinite, e.g. 'mani 'some','kOik 'all', 'muu
'else' and 'iga 'any, every', 'emb'kumb 'eitherthis or that'. Sampleforms:

Sg Plur Sg Plur Sg Plur Sg

, ,
N 'moni 'koik 'koik
'moned muu muud 'iga
,
G 'mone 'koige 'koigi -
'monede muu 'muude 'iga
'koikide
,
P 'mond(a) 'monesid 'koike 'koiki muud 'muid 'iga(t)
,
III 'monda 'mone'desse
'koike 'koigisse muusse'muudesse'igasse
136 ESTONIAN

Verb
The Estonianverb has (1) finite forms, which occur only as predicatesor as
the auxiliary componentsof complex predicates,and (2) non-finite forms,
which occur(a) in complexpredicatesin combinationwith a finite form (past
participles),(b) similarly to adjectivesas attributesand predicatives(partici-
ples), (c) as adverbials (supines and gerunds),(d) as subjectsand objects
(infinitives).

SimpleFinite Forms
Finite forms canencodevoice, mood,tense,aspect,personand number.
Of the two voices, the personalvoice (Ps) is indicated by person and
numbersuffixes (sl -n, s2 -d, pI -me, p2 -te, p3 -d) that are obligatory in the
indicative mood and optional in the conditional. Subjectpronounsand nouns
can be omitted when the finite form has a personalending. There are no
personor numbersuffixes in the quotativeand jussive moods.The markers
of the impersonalvoice (Ips) are -a-, -da-, and -ta- in the affirmative present
indicative, -da(-) and -ta(-) in the conditional, quotative,and jussive moods
and in the negativepresentindicative, and -d- and -t- in the imperfect tense
of the indicativemood.
Mood in Estonianincludes the indicative, conditional, imperative,quota-
tive and jussive. The indicative mood is used when the speakeraffirms or
deniesthe occurrenceor non-occurrence of a situationor an action(sa konnid
Uiga kiiresti 'you walk too quickly'). The conditionalmoodis usedto express
the speaker'sopinion that a situationor an action should or should not have
occurred,or would or would not occur, undercertain conditions(konniksin
aeglasemaltkui olekssoojem'I would walk more slowly if it were warmer')
or that the occurrenceor non-occurrenceof a situation or an action is
obligatory or preferable(riiiigiksin niiiid tott 'I should tell the truth now').
The imperativemoodexpressesthe speaker'srequest,order, or prohibition to
the listener(s), e.g. minge molemad koju! 'you both go home!') or the
speaker'ssuggestionto the listener(s)that a joint action be undertaken,e.g.
voidelgemlopuni! 'let us fight to the finish!'). The quotativemood is used
when the speakerwants to point out that (s)he is not responsiblefor the
veracity of a statement,but is only an intermediatoror reporter,e.g. sa ostvat
maja 'you are(reportedto be) buying a house'.Thejussivemoodexpresses
either that the listener(s)should convey the request,order, or prohibition to
a third party (tulen kell viis, Mari oodakumind! 'I'll comeat five o'clock, let
Mary be waiting for me!') or that one is compelledby someoneto do (or not
to do) something(meielugeguniiiid iga senti! 'we shouldbe countingevery
centnow!').
There are simple, i.e. non-compoundaffirmative present-and imperfect-
tenseforms and complex,i.e. compound,perfectand pluperfectforms in the
indicative mood and simple affirmative present-and past-tenseforms in the
conditional and quotative moods; there are also correspondingcompound
 ESTONIAN 137

forms for the non-compoundpast conditional and quotative. There are in

addition to the non-compoundpresent-tenseforms of the imperative and
jussive moods also compound perfect forms. All compound tense forms
contain afinite form of the auxiliary verb 'olla 'to be' and a personalor an
impersonalpastparticiple.
The affirmative is characterizedby (obligatory or optional) personal
endingsin the personalindicative and conditionalmood forms and by special
tense suffixes in the impersonalindicative mood forms. The negativehas
characteristiccomplexforms: (1) in the indicative, conditional, and quotative
moods, a specific negativeform, which is markedonly for voice and mood
or tense,and the preposednegationparticle ei, and (2) in the imperativeand
jussive moods, a personal form precededby the appropriateform of the
defectiveprohibition verb (s2imp 'iira, plimp 'iirgem - 'iirme, p2imp 'iirge
andjussive 'iirgu).
In the imperative, there is no first-person singular form, and the former
third-person form has been generalized in most dialects (including the
standardlanguage)to all personsto renderan indirect orderor a forced action
or situation, thus yielding a new mood, the jussive. The second-person
singular form is unmarked, the second-personplural form has the suffix
-ge/-ke,andthe samesuffix occursin the rarely usedformal first-personform,
which has the unique personalending -m. In informal style the first-person
plural form of the presentindicative is usedinsteadof the imperativemood
form; however,in contrastto the negativeforms of the indicative mood, the
first-personplural forms retain their personalending when forming complex
negativeforms with the informal first-person prohibition verb form 'iirme,
e.g. 'iirme 'ootame 'let us not wait', 'iirme 'liih(e)me 'let us not go', contrast
the formal equivalents 'iirgem 'oodakem and 'iirgem 'mingem and the
morphologicallyunmarkednegativepresentindicative formsme ei 'oota 'we
do not wait', 'me ei 1iihe 'we do not go'.
With the exceptionof the imperative mood, the finite-form voice, mood,
and tensesuffixes in the affirmative (aft) and negative(neg) are as presented
in Table4.6.

Non-finite Forms
Non-finite forms include participles,the supine,the infinitive and the gerund,
as set forth in Table 4.7. Participlesand the supinemay be either personalor
impersonal,e.g. 'valvav 'mees'a/theguardingman' and 'valvatav'mees'a/the
man who is being guarded','valvanud 'mees'a/the man who has guarded',
'valvatud 'mees 'a/the man who has been guarded'. Present participles
function like adjectives,and are inflectedin caseandnumber.Pastparticiples,
in addition to their use as attributes or predicatives,can composecomplex
tenseforms built with finite forms of the verb 'to be', namelythe perfectand
pluperfectindicative, pastconditional,pastquotativeand perfectimperative,
e.g.
Table4.6 Voice, mood, and tensemarkers in finite forms

Mood and Indicative Conditional Quotative Jussive

tense Present Impeifect Present Impeifect Present Impeifect Present

Ps aff s3 -b -i, -is,-s -ks -nuks -vat -nuvat -gu, -ku

p3 -va- -i, -si- -ks(i-) -nuks(i-) -vat -nuvat -gu, -ku
s12,
p12 -i, -si- -ks(i-) -nuks(i-) -vat -nuvat -gu, -ku
Psneg ¢ -nud -ks -nuks -vat -nuvat -gu, -ku
Ips aff -akse } -di -daks -duks -davat -dagu
-dakse
-takse -ti -taks -tuks -tavat -tagu
Ips neg -da -dud -daks -duks -davat -dagu
-ta -tud -taks -tuks -tavat -tagu
 ESTONIAN 139

Table4.7 Infinite form markers

Infinite fonns Personal Impersonal

Participle present -v -ev -day -tav

past -nud -dud -tud
Supine illative -ma -dama -tama
inessive -mas
elative -mast
translative -maks
abessive -mata
Infinitive -a -da -ta
Gerund -es -des -tes

1 Kutt on haavanudhunti 'A hunterhaswoundedthe wolf.

2 Hunti on haavatud'Onehaswoundedthe wolf.
3 Hunt on haavatud'The wolf is wounded'.

Sentences(1) and (2) have predicatesin the perfectindicative and an object

(hunti) in the partitive singular; sentence(3) has the predicateon in the
presentindicative and the subjecthunt in the nominative singular, haavatud
being a predicativecomplement.
Supineforms function as adverbials,with various caseforms expressing
intentionsor attemptsat doing something,ongoingstatesand activities,goals,
etc., for example,illative 1iihen 'malet 'miingima 'I'm going (somewhere)to
play chess', 'hakka 1ugema! 'start reading!', inessive 'olin 'k/ubis 'malet
'miingimas 'I was in the club playing chess',e1ative 'tulen 'malet 'miingimast
'I come from playing chess',translative 'oled 'liiga 'vana 'valetamaks'you
are too old for lying', abessive'iira 'lahku 'maks'mata'don't leave without
paying!' The impersonal supine is used only with the indicative and
conditional verb forms 'peab 'must' and 'peaks 'should': 'seda 'peab
'soodama'one musteat it', 'seda 'peaks'soodama'one shouldeatit'. As the
illative form of the personalsupineis always in the strong grade,it has been
the traditional citation form for verbsin Estoniandictionaries.
Infinitives serve as subject(1ugedaon meeldiv 'readingis pleasant')or
object('eelis'tame 'oodata 'we preferto wait'). Gerundsare adverbialswhich
indicate a parallel action or processCoodates'rongi ma 1ugesin '(while)
waiting for the train, I was reading').
In certain constructionsthe use of the illative supine or of the infinitive
dependson the predicateverb, cf. ta 'peab 'ootama 'he must wait', tal 'tuleb
'oodata '(s)he(adessive)hasto wait'.
140 ESTONIAN

Verbal Nouns
Most Estoniangrammarsinclude in the verb paradigmtwo (de)verbal nouns
which are inflected in both caseand number.Theseare the nomenagentisin
=ja and the nomenactionis in =mine, e.g. 'sooja 'eater','soomine '(the act
of) eating'. In fact, nominaactionis belong to a more complicatedsystemof
actor and undergoernouns that has resultedfrom a split of the former past
participles,cf. 'sooja 'eater(unspecifiedfor time)" 'soonu 'one who has or
had eaten(previously)" 'soodu 'somethingor someonethat has beeneaten
(previously)'. In certain petrified phrases,ja-derivativesbehavelike present
participlesbut remainunspecifiedfor time, e.g. 'haukuja 'koer 'a/thedog that
has the habit of barking (a lot)' and 'haukuv 'koer 'a/the dog that is (= just
happensto be) barking'.

Verb Inflection
Like nominal inflection, verb inflection is complicatedby extensivesuffix
andstemalternation,neitherof which is entirely predictable.
As an exampleof verb inflection the paradigmof the verb 'soOta 'to feed'
is presentedbelow; it representsverbswith weakeninggradation.
Infinite forms areasfollows:

Participles Personal:present'sootev,past'soOtnud
Impersonal:present'soodetav,past 'soodetud
Supine Personal: ill 'sootma ine 'sootmas ela 'sootmast trans
'sootmaks abe'sootmata
Impersonal:ill 'soodetama
Infinitive 'soOta
Gerund 'soOtes

Table4.8 Sampleverb paradigm,indicative: soota'to feed'

Present Imperfect Perfect Pluperfect

Affirmative
sl 'soodan 'sootsin 'olen'sootnud 'olin'sootnud
s2 'soodad 'sootsid 'oled'sootnud
, ,
'olid'sootnud
s3 'soodab 'sootis on sootnud ' oli 'sootnud
pI 'soodame 'sootsime ' oleme'sootnud 'olime'sootnud
p2 'soodate 'sootsite ,
'olete'sootnud
,
'olite'sootnud
p3 'soOdavad 'sootsid on sootnud 'olid'sootnud
Ips ' soodetakse 'soodeti 'on'soodetud 'oli'soodetud

Negative , , ,
Ps 'ei'sooda ei'sootnud ,
ei ole soOtnud 'ei 'olnud'sootnud
Ips 'ei 'soOdeta 'ei'soodetud ei ole soOdetud 'ei 'olnud 'soOdetud
 ESTONIAN 141

Table 4.9 Sampleverb paradigm, conditional: 'soota 'to feed'

Present Past

Affirmative
sl 'soOdaks(in) 'sootnuks(in) 'oleks(in)'sootnud
s2 'soodaks(id) 'so6tnuks(id) 'oleks(id)'sootnud
s3 'soOdaks 'sootnuks 'oleks'sootnud
pI 'soodaks(ime) 'sootnuks(ime) 'oleks(ime)'sootnud
p2 'soodaks(ite) 'sootnuks(ite) 'oleks(ite)'sootnud
p3 'soodaks(id) 'sootnuks(id) 'oleks(id)'sootnud
Ips 'soodetaks 'soodetuks 'oleks 'soOdetud

Negative
Ps 'ei'soodaks 'ei'sootnuks 'ei 'oleks'sootnud
Ips 'ei'soodetaks 'ei'soodetuks 'ei 'oleks 'soOdetud

Table 4.10 Sampleverb paradigm, imperative: 'soota 'to feed'

Present Perfect
Affirmative Negative Affirmative Negative

sl
s2 'soOda 'lira'sooda 'ole'sootnud 'lira 'ole 'sootnud
s3
pI 'sootkem 'lirgem'sootkem 'olgem'sootnud 'lirgem 'olgem
'sootnud
p2 'sootke 'lirge'sootke 'olge'sootnud 'lirge 'olge'sootnud
p3

Table 4.11 Sample verb paradigm, quotative and jussive: 'soiita 'to feed'

Present Perfect
Affirmative Negative Affirmative Negative

Quotative
Ps 'sootvat 'ei'sootvat 'sootnuvat 'ei'sootnuvat
Ips 'soode'tavat 'ei'soode'tavat

Jussive
Ps 'sootku 'lirgu'sootku 'olgu'sootnud 'lirgu 'olgu'sootnud
Ips 'soode'tagu 'lirgu'soOde'tagu 'olgu'soodetud 'lirgu 'olgu 'soodetud
142 ESTONIAN

Adverbs
Adverbs modify (1) a verb (Jookse 'ruttu 'run quickly'), adjective Cliiga
'noor 'too young') or an adverbCvaga 'ruttu 'very quickly') or (2) a clause:
theseare the interrogative/relativeadverbs.
The most frequent interrogative/relativeadverbsare 'kuhu '(to) where',
'kus 'where', 'kust 'from where', 'millal 'when', 'kunas 'when', 'kuidas
'how', and 'miks 'why'.
Local adverbsoften havethreecaseforms, onelative, onelocative,andone
separative,dependingon the direction of the verb's action, e.g. 'koju '(to)
home', 'kodus 'at home', 'kodunt 'from home'; 'alla, 'all, 'alt 'down'; 'ules,
'uleval, 'alevalt 'up'; 'ette, 'ees, 'eest 'before'; 'taha, 'taga, 'tagant 'behind';
'sUa, 'sUn, 'sUt 'here'; 'sinna, 'seal, 'sealt 'there'; 'sisse, 'sees,'seest 'in';
'valja, 'valjas, 'valjast 'out'. Othersarenot inflected,e.g. 'ara 'away'.
The mostfrequenttemporaladverbsare 'nuud 'now', 'praegu 'now', 'kohe
'immediately', parast 'after(wards)','sUs 'then', 'varsti 'soon', 'ammu 'long
ago', niljuti 'lately', (,uks),kord'once','asja 'not long ago', Just 'just', 'vara
'early', nilja 'late', 'eile 'yesterday', 'tana 'today', 'homme 'tomorrow',
'mullu 'last year', 'tanavu 'this year', nommikul'in the morning', 'paeval 'in
the daytime','ohtul 'in the evening','oosel-'oosi 'at night'.
Adverbs of mannerare most often built with the suffixes =sti or =It, and
have special comparativeforms in =mini and superlativeforms in =imini,
e.g.'kUresti - 'kiirelt 'quickly', 'kiire'mini 'more quickly', 'kiirei'mini 'most
quickly'. The adverb palju 'many; much' has suppletivecomparativeand
superlativeforms, viz. 'enamor 'rohkem 'more' and 'enim - 'koige 'enamor
'koige 'rohkem 'most'. Some adverbsof mannerhave two caseforms, one
occurringwith verbswhich indicatea changeinto a state,the other occurring
with verbs which indicate being in a state, e.g. (laheb) 'pilve '(it becomes)
cloudy': (on) 'pilves 'it is cloudy'.

Postpositionsand Prepositions
Prepositionsand postpositionsmodify nounsin variouscaseforms; many are
identical with adverbs.Like adverbs,severallocal or orientationalpreposi-
tions and postpositionshave as many as three caseforms. There are more
postpositionsthanprepositions.
Some words act as both prepositionsand postpositions,e.g. 1abi, 'ale,
'umber, 'mooda,usuallywith a differencein meaning,(1abi 'puu 'throughthe
tree', 'puu 1abi 'through the fault of a tree'; 'ule 'silla 'acrossthe bridge',
'silla 'ule 'aboutthe bridge'),lessoften with only stylistic overtonesCumber
'sorme - 'sorme 'umber 'around the finger', 'mooda 'teed - 'teed 'mooda
'alongthe way').
Nouns governedby prepositionsare in the genitive (with the prepositions
'alla 'below, beneath','labi 'through', 'ule 'above; across,beyond', 'umber
'round, around)and in the partitive (with 'enne 'before', 1igi 'closeto', piki
'along', 'keset'in the middle or, 'mooda'along','peale 'after', parast'after',
 ESTONIAN 143

'vastu 'against').The terminativeoccurswith 'Iami 'until', the abessivewith

'ilma 'without', andthecomitativewith ('iihes)'koos 'togetherwith' .
Postpositionsgovernnounsin the nominative(the postpositionliibi 'along,
throughout,asin 'piiev 1iibi 'all day long'), the genitive(Jiirgi 'accordingto',
'kaudu 'by way of, 'kaupa 'by', as in 'kahe 'tosina 'kaupa, 'by two dozens'),
Jooksul'during', 'kestel'in the courseof, 'korral 'in caseof, 'paiku 'about',
as in 'kella 'iihe 'paiku 'at about one 0' clock', piirast 'becauseof, and for
most orientationalpostpositions,e.g. 'vastu 'against,to(wards);in exchange
for', 'vastas'beyond'.The postpositions'peale 'beginningwith', and 'saadik
'up to' governthe elative.

Conjunctions
The most frequentconjunctionsareja and ning 'and', ehk and voi 'or', aga,
vaid, entandkuid 'but', 'ainult 'only', ega 'nor', kui 'if; as', 'nagu 'as','kuigi
'although',et 'that', sest 'because,for', kuna 'while, as, for', 'ometi 'still',
'siiski 'still, however,nevertheless','nii'siis 'hence,thus', 'nii('hiisti) ... kui
(ka) 'both ... and',kas ... voi 'either... or', 'ei . .. 'ega 'neither... nor'.

Syntax
Word Order and SentenceTypes
In a clause,S[ubject], V[erb], O[bject], P[redicate],and A[dverbial] usually
build the following patterns:(1) SV, (2) SVO, (3) SVP, (4) SVA, (5) SVAO,
(6) AVS, (7) AV, (8) AVP, (9) AVA, (10) OV, (11) OVA, (12) AVO, as
exemplifiedbelow.

1 Tahe-dsara-vad.'(The) starsare shining.'

STAR-plur SPARKLE-p3
2 Ttidruk mtiti-b lilli. 'Althe girl sellsflowers.'
GIRL SELLS-s3FLOWER.pP
3 Lapse-don vaikese-d.'The childrenare small.'
CHILD-plur IS.s3 SMALL-plur
4 Kutsika-d ol-i-d aia-s. 'The puppieswere in the garden.'
PUPPY-plurIS-past-plurGARDEN-ine
5 Ma lahuta-ntulu-st kulu-d. 'I deletethe expensesfrom the income.'
PRO.slDELETES-sl INCOME-elaEXPENSE-plur
6 Koera-l ol-i-d kutsika-d. 'The dog hadpuppies.'
DOG-adeIS-past-plurPUPPY-plur
Aia-s kasva-blilli. 'Thereare flowers growing in the garden.'
GARDEN-ineGROWS-s3FLOWER.pP
144 ESTONIAN

7 Viilja-s miirista-b. 'It is thunderingoutside.'

OUTSIDE-ineTHUNDERS-s3
8 Varvas-te-lhakka-skiilm. '(The/his/her...) toesbeganto feel cold.'
TOE-plur-adeBEGINS-pastCOLD
Tanava-lol-i kiilm. 'It was cold on the street.'
STREET-adeIS-pastCOLD
V60ras-te-ltule-b lahku-da.'Strangershaveto leave.'
STRANGER-plur-adeCOMES-s3LEA VES-inf
9 Me-i-l Hihe-b hasti. 'We are doing well.'
PRO.pl-sAdeGOES-s3WELL
10 Sin-d pete-t-i. 'You weredeceived.'(= 'Onehasdeceivedyou.')
PRO.s2-sPDECEIVE-ips-past
11 Rahvas-tpee-ta-kserumala-ks.'Peopleare held (to be) foolish.'
PEOPLE-sPHOLDS-ips-presFOOLISH-trans
12 Oun-te-stsaa-da-ksemahla. 'Onegetsjuice out of apples.'
APPLE-plur-elaGETS-ips-presmICE.sP

Patterns 6-9 representexistential clauses, 10-12 impersonal clauses. In

patterns 1-5, the verb agreeswith the subject in person and number; in
patterns7-9 verbs are in the singular, in 10-12 in the impersonalmood.
Pattern6 includesboth sentencesin which the verb agreeswith the subject
in personand numberand sentenceswith singular verbs and partial subject
(i.e. with the subjectin the partitive case).
In the main, word order in a clausedependsnot on subjectv. object, i.e.
grammatical,relations, but rather on topic (what is assumed)v. focus (the
centralcomponentof the commenton the topic). The topic normally occupies
clause-initialposition, and the focalized elementeither occupiesclause-final
position or attracts the strongest sentencestress in the comment. Thus
sentence5 can be reorderedinto tulust ma lahutan kulud 'from the incomeI
deletethe expenses'topicalizingthe word tulust 'from the income'.Similarly,
in kulud ma lahutan tulust 'as for expenses,I am deleting them from the
income' the word kulud 'expenses'has beentopicalizedand tulust has been
focalized.

Subject
A subjectnoun is in either the nominativeor the partitive case.The subject
noun is in the singularor plural partitive (1) when the predicateverb is in the
third personsingularand the subjectnoun, usually in final position, refers to
an indefinite amount('metsason 'loom-i FOREST-ineIS.s3presANIMAL-pP
'thereare animalsin the woods') or (2) when the clause,with predicateverb
 ESTONIAN 145

in thethird personsingularandwith thesubjectnounin final position,is negated

(,metsas'ei ole 'hunt-e 'there are no wolves in the woods'). Otherwisethe
subjectnoun is in the nominative.A subjectnoun in the partitive is called a
partialsubject;onein thenominativeis calleda total subject.

Object
Usually an object nominal is in the partitive: ema 'ostis 'leiba MOTHER
BUYS-s3pastBREAD.sP'motherboughtsomebread',kassa 'suudledmind?
'do/will you kiss me?'). An object nominal is in the nominative or genitive
only when it denotes a definite quantity in an affirmative clause whose
predicateverb is eitherin distinctively perfectiveaspector is a transitiveverb
with a final adverbial: 'ema 'ostis 'leiva 'mother bought the bread', kas sa
'suudledmu 'surnuks?'will you kiss me to death?'Note that in the first of
these two sentences,the genitive singular object 1eiva permits three
interpretationsof the clause: (1) 'I bought one loaf of (rye) bread'; (2) 'I
boughtall the breadthat was available', and(3) 'I boughtthe amountof bread
mentionedearlier'. An object nominal in the partitive is usually called a
partial object; onein the nominativeor genitiveis calleda total object.
A singularnominal,including the first- or second-person singularpronoun,
as a total object, is in the genitive if the clauseis affirmative, the predicate
verb is personal and not in the imperative or jussive mood. A singular
nominal,exceptthe first- or second-person pronoun,as a total object,is in the
nominativeif the predicateverb is impersonalor in the imperativeor jussive
mood. The total object caseof plural nominalsis the nominativeplural; only
the first- and second-personpronouns are in identical conditions in the
partitive plural.

MeasureAdverbials
Much like objects, theseoccur in the nominative, genitive, and often in the
partitive, depending on the morphological form of the predicate verb.
Examples:'olin 1wdus'terve 'niidala (sG) 'I was at homea whole week', 'ole
kodus 'terve niidal! (sN) 'be at home a whole week!', rna ei 'olnud Kodus
'tervet 'nadalat(sP) 'I was at home (but) not for a whole week'; cf.ma ei
'olnud 1wdus'terve 'niidal (sN) 'I was not at homefor a whole week'.

Lexicon
In addition to the lexicon inherited from proto-Fennic and its daughter
protolanguages,Estonianhas borrowingsfrom both Low and High German,
from EstonianSwedish,Finland Swedish,and SwedenSwedish,from Old
and Modem Russian,from Latvian, and from Finnish. Germanwas the most
significant sourceof loans from the thirteenthcenturyuntil 1940, and it was
via Germanthat the greaterpart of the internationallexicon enteredEstonian.
To this day, borrowing patternsof international lexical stock of Latin and
Greek origin follow to a great extentthoseusedwhen borrowing took place
146 ESTONIAN

via German,althoughfrom the 1920sthe original sourcelanguageshavebeen

taken into accountmore carefully. Finnish has been a significant sourceof
loanssincethe 1870s.
Low German was influential up to the sixteenth century. After the
Reformationthe local Germannoblemenand bourgeoisiebeganto shift to
High German,and with this shift the local standardform of High German
(so-calledBaltendeutsch)becamethe main sourceof borrowings. In many
casesit is impossibleto distinguishbetweenborrowingsfrom Low and High
Germanon the one hand and betweenthosefrom Low Germanand Sweden
Swedish on the other. There are about 770-850 borrowings from Low
German(e.g. aabits 'primer', haamer 'hammer',ingel 'angle', kants 'fort',
kee 'necklace',kirik 'church', klooster 'monastery',kool 'school', kaarid
'scissors',kOok 'kitchen',kuhvel'shovel',kuun 'barn',munk'monk', molder
'miller', muur 'stonewall', naaber 'neighbour',ohver 'victim', orel 'organ',
pann 'pan', pihtida 'to confess',pott 'pot', preili 'Miss', proua 'Mrs',
pooning 'attic', raad 'city council', raatus 'town hall', ruum 'room', rostida
'to roast',roovel 'robber',veerida 'to spell',vorst 'sausage',amber'bucket'),
and about 490-540borrowingsfrom High German(e.g. kamm 'comb', kett
'chain', kirss 'cherry', loss 'palace',pirn 'pear; electric bulb', sink 'ham',
vurts 'spice;flavouring') including about60 Baltic Germanborrowings(e.g.
kutsar 'coachman',redel 'ladder', sirel 'lilac', porgand 'carrot'). Loan
translationsfrom Germanareno lessimportant.
There are 100-150borrowingsfrom Swedish(e.g. hiivata 'to heave',iii
'blast of wind', moor 'old woman', norss 'smelt', plika 'girl', riik 'state',
raim 'Baltic dwarf herring',sang'bed',tasku 'pocket').
Except in Setu and the other easternmostdialects which have been in
contactwith Russianfrom asearly as the Old Russianperiod,borrowingfrom
Russianbeganin earnestafter 1710, when Russiafirst engulfedEstonia.In
literary Estonian there are about 300-350 Russian loans, including both
archaisms and Sovietisms (e.g. kapsas 'cabbage',majakas 'lighthouse',
morss 'fruit juice', munder 'uniform', puravik 'boletus', riisikas 'milk
mushroom',tassida'to carry').
Borrowingsfrom Latvianbeganto appearin Estonianin theeighthcentury.
Thereare about30-45 in literary Estonian(e.g. kauss'bowl', kouts 'tomcat',
laiits 'lentil', nuumata'to fatten [tr]', sokal 'chaff, viisk 'bastshoe').
There are about 800 Finnish loans in Estonian.Theseare mostly derived
stemsbuilt to roots which the two languageshavein common,suchas annus
'portion', built to anda 'gives';cf. the synonymousFinnishannos,built from
Finnishanta-. Thereare also about100 Finnish-originstemswhich were new
to Estonian, such as aare 'treasure',julm 'brutal', jaik 'stiff, mugav
'comfortable',riinnata 'to attack',sangar 'hero', siiilida 'to preserve',tehas
'factory'.
Since the 1920s Estonian has taken on terminology from English and
French, usually following the example and pattern of other languages
 ESTONIAN 147

(Gennan,Russian,or Finnish); thereis alsodirect borrowingfrom English.

Estonianalso has a few direct borrowingsfrom other non-Indo-European
languages,e.g. jaana+ in jaanalind 'ostrich', and koi 'stingy' from Hebrew,
velmata'to revive (tr)' from Mordvin, nulg 'spruce'from Mari.
From 1913, about 55 artificial roots have been createdquasi ex nihilo,
mainly by JohannesAavik, e.g. eirata 'ignores',kilillik 'rabbit', laip 'corpse',
liink 'gap,lacuna',meenuda'to comeinto one'smind', morv 'murder',milrsk
'missile', nentida 'states', nome 'ignorant', range 'strict', relv 'weapon',
taunida 'to condemn',veenda'to convince'.

(Northern Literary) Estonian Text

AI-3: text in orthography;BI-3: transcription(primary stressis underlined,
secondarystress is marked with colon); CI-3: morpheme-by-morpheme
glosses;DI-3: free translation.

Al laeva v6is siiski p6letada,

Bl If!eva si:i--ki
v~1-s-0 si:i--ki p6letada,
p~le=tta:-oa
Cl SHIP.sP CAN-past-s3 THEN--EVEN BURNS=VdeV-inf

kuid ainuIt lihel paeval aastas, kui

k.!!lo .!!InuIt i!he-l p~eva-l p~eva-l ~sta-z ~sta-z kui
BUT ONLY ON ONE ON DAY IN YEAR WHEN

paikesering oli j6unud oma

paikesering
p~kkeze+riDG QI-i-0 j6unud
j~ilD-nuo Qma
SUN.sG+CIRCLE IS-past-s3 REACHES-past.part OWN.sG

haripunkti. A2 suveharjaks
(h)f!rl+Bu:IJkti B2 s.!!ve+(h)a:fja-ks
RIDGE+POINT.slll-1 C2 SUMMER.sG+RIDGE-sTrans

nimetati seda paeva. A3 see

nime=tta:-lt-i s~-oa s~-oa p~evap~eva B3 s~
NAME=VdeV-imp-past THIS-sP DA Y-sP C3 THIS

on nlilidne jaani6htu.
on ni!:lio=ne jjaani6htu.
aanii+~i1tu
IS.s3pres NOW=ADJdeADV JOHN-sG+EVE

DIOne could nonethelessburn a ship, but only on one day in the year, when
the sunhad reachedits apogee.D2 This day was called the summerpeak.D3
It is the contemporarySt John'sEve.
148 ESTONIAN

Referencesand Further Reading

Erelt, M., Kasik, R., Metslang,H., Rajandi, H., Ross,K., Saari,H., Tael, K., Yare, S.
(1993-5) Eesti keele grammatika, vols I-II. Tallinn: Eesti TeadusteAkadeemia
Eesti Keele Instituut.
Hasselblatt,C. (1992) GrammatischesWorterbuchdes Estnischen(Veroffentlichun-
gender SocietasUralo-Altaica35), Wiesbaden:Ural-Altaic Society.
Kasik, R. (1996) Eesti keele sonatuletus, Tartu Ulikooli eesti keele oppetoali
toimetised3. Tartu.
Raag, R. (1997) Elementar estnisk satslara FU Uiromedel 22. Uppsala: Uppsala
universitet,Finsk-ugriskainstitutionen.
Raun, A. and Saareste,A. (1965) Introduction to Estonian Linguistics, Wiesbaden:
Harrassowitz.
Ratsep, H. (1978) Eesti keele lihtlausete tiiiibid, EmakeeleSeltsi Toimetised 12,
Tallinn: Valgus.
Saareste,A. (1958-63)Eesti keelemoistelinesonaraamat,(Dictionnaire analogique
de la langueestonienne,vols I-IV), Stockholm:Kirjastus VabaEesti.
Tauli, V. (1973-83) Standard Estonian Grammar, vols I-II, Acta Universitatis
Upsalensis;Acta Uralica et Altaica Upsalensia8, Uppsala:University of Uppsala
andAlmquist & Wiksell.
Valgma,J. and Remmel,N. (1968) Eestikeelegrammatika(Kasiraamat),Tallinn.
Viks, U. (ed.) (1992) A Concise Morphological Dictionary of Estonian, vol. I:
Introduction and Grammar, Tallinn: Eesti Teaduste Akadeemia, Keele ja
KirjandnseInstituut.
Wiedemann,F. J. (1869) Ehstnisch-deutsches Worterbuch,St Petersburg.
5 Finnish
Daniel Abondolo

Finnish (native name: suomi) is the first languageof some four and a half
million people in Finland (roughly 95 per cent of the population), and of
approximatelyhalf a million peopleliving in Sweden,Norway, Estonia,and
Russia.Censusstatisticsshow a steadydecline in native speakersof Finnish
in Russia,from about 60 per cent of 92,000 declaring themselvesFinns in
1959 to about 35 per cent of 67,000in 1989 (Ktinnap 1992). There are also
significant pockets of speakersin the United States and Canada,chiefly
around the Great Lakes; the emigratory surge to thesecountrieswas at the
tum of the nineteenthandtwentiethcenturies.
ContemporarystandardFinnish is a northern Fenniclanguagewith both
westernand easternfeatures.It is northernby virtue of isoglossessuchas the
third-personpronounhiin insteadof *tama (as in Estoniantema), and -i- in
sisar 'sister'insteadof *e (asin Estoniansosar).1tis both westernandeastern
in that it combines

1 westernFinnish dialect featuressuchas =U- detransitivizers(as opposed

to easternreflexives in =kse-, which are available to speakersof the
standard language, to varying degrees according to education and
interests,from poetic tradition) with
2 easternfeaturessuchas neutralkesii 'summer',Uta 'evening'(as opposed
to westernsuvi, ehtoo,which are stylistically chargedsynonyms).

The complexity of nominal inflection, much of which centres on the

formation of the genitive plural, is in part a consequence
of east/westdialect
amalgamation(Laakso1991: 53).
As a Fennic language, Finnish is peripheral and it is therefore not
surprisingthat it has preservedmany archaicfeaturesfrom proto-Uralic. On
the other hand its high degreeof agglutinativity (comparedwith, say, that of
North or South Estonian, or Saarnic) may suggestinnovation as well as
conservatism,and 'the choice of Finnish as the Finno-Ugric prototypemay
haveimpededprogressin the field' (Austerlitz 1993: 29).
Good short sketchesare Austerlitz 1968: 1336-47, Branch 1987, and
Karlsson1992.

149
150 FINNISH

Table5.1 Finnish vowel phonemes

Short Long

e ti u ii titi uu High
e 0 0 ee 00 00 Mid
a a aa aa Low
+ + + + Rounded
+ + + + Back

Phonology
Vowels
Finnishhaseight shortandeight long vowel phonemes;seeTable5.1.
The orthographywrites the vowels as given above, except that the high
front roundedvowels arewritten <y> and <yy>.
Vowel length is distinctive in all positions.If we assumethat proto-Uralic
had no distinctive vowel length, then in Finnish the short vowel 0 and all the
long vowels are ultimately of secondaryorigin in all positions,althoughthe
length of fIrst-syllable uu in some words is quite old. In native vocabulary,
long vowels in fIrst syllableare mostcommonlythe resultof contractionafter
the loss of a p(F)U consonant,e.g *-1]- in riihi 'drying house' (Komi ri'nis)
andjiiii 'ice' (North Saarnijiekga)and *-x- in kuusi 'spruce'« pU *kaxsi',
cf. Nganasankuo < proto-Samoyedic*kaSt).
In the fIrst syllable, both 0 and 00 signal affect (halmo 'fool', joro
'peevish',poonii 'coffee bean [folksy, jocular]') or relatively recentborrow-
ing (koli 'keel', cf. Swedish kol; kongiis 'precipitous waterfall', cf. North
Saarni geavggis) or proper-namestatus (ToolO, a northwesternsuburb of
Helsinki). First-syllableii, on the otherhand,seemsto continuepU *ti in fIve
or six words, e.g. syli '(spacebetween)outstretchedarms; lap; womb' < pU
*stili 'arms;lap' (Janhunen1981b).

VowelHarmonyandMorphophonemicCode
Within root morphemesand any suffIxes attached to them, the mixed
occurrenceof front roundedwith backvowels is restricted.In general,words
with a front rounded vowel in their fIrst syllable have no back vowel
in subsequentsyllables, and words with a back vowel in their fIrst syllable
have no front roundedvowels in subsequentsyllables; thus koulussa 'in a
school' and viiyliissii 'in a channel' occur, but forms such as *koulussa,
*koulyssa and *vaylussa do not. Non-roundedfront vowels mix with all
other vowels, e.g. perunassa'in a potato', pirussa 'in a devil', kielessii 'in
a language'.
The restrictionson the occurrencesof vowels,andthe front v. backvariants
of suffIxes such as the -ssa v. -ssii 'in' of forms such as the examplescited
 FINNISH 151

above, pennit the use of a morphophonemiccode which writes Finnish

vowels with only five symbols,I E A 0 U, and marks certainmorphemesas
front-prosodic with a prosodemicsymbol, e.g. front-prosodic #SAVU siivy
'nuance'v. unmarked,back-prosodicSA VUsavu 'smoke'.To compressthe
discussion and to render much of the morphology explicit, this chapter
presentsFinnish forms using a morphophonemiccode devised by Robert
Austerlitz (Austerlitz 1964,1965,1967,1983). The elementsof this codeare
set out in Figure 5.1; for the synchronic readings and the historical
backgroundof T, Q, and X, seep. 154.
Root morphemescontaining only the front unroundedvowels I and/or E
are assumedto be prosodically front, with the exception of the singular
partitive forms of MERE meri 'sea' and VERE veri 'sea',which are back-
prosodic,viz. MERETA mer-ta,VERETA ver-ta.
In foreign vocabulary,violations of this distribution abound,e.g. volyy-
mistii 'from/concerningvolume'; such forms must be written in code with
modulations of prosody, e.g. VOL ~UMI-STA;~UMI-STA; cf. Hungarian ~Ub~UbANS
[ntians] 'nuance'.But evenin native vocabulary,the picture is quite complex
if derivedstemsare considered:derivationalsuffixes behavedifferently from
inflectional suffixes, and the degreeof transparencyof a given derivate is
difficult to assess.For example,the nominalizer=U is front y when attached
to #KASKE- 'commands',yielding #KASKF;::U kiisky 'command',but u
when attachedto ISKE- 'strikes', yielding isk=u 'blow', and there is in
generala kind of vertical gradientaccordingto which the more openthe front
vowel in a root, the more likely it is to trigger front readingsof derivational
suffixes. On the other hand, we have ~IITTA=U~IITTA=U nUtty 'meadow', from
~ITTA-~ITTA- 'mows'; contrastniitto '(a) mowing', with back 0 in spite of the
front prosody of the root to which it is attached.This latter pair of forms
illustratesa widespreaddistributionalpatternwhich is the resultof the relative
newnessof 0 as a phoneme.SeeKarlsson1983: 98-104,with literature.
In addition to the basevowel units I E A 0 U, it is useful to work with an
operator0, which copiesthe vowel to its left. In the mostfrequentoccurrence
°
of this operator,the first vowel to the left of is immediatelyadjacent.If this
adjacentvowel is high or low, the result is a long vowel, e.g. PUD puu 'tree',
#SAD siiii 'weather'; if it is mid vowel, the result is a vowel sequence
equivalentto a high-to-mid diphthongof appropriatefrontnesslbackness and
lip-rounding, e.g. SODRA suora 'straight',#PODRApyorii 'round',TEDRA
tiera 'lump of snow andice which adheresto the bottomof a shoeor a horse's
hoof, #KEDRA kierii 'not straight'.The sequencesEE and 00 also occur in
the first syllable,but arerarein native non-affectivevocabulary;examplesare
foreign SOOLO soolo 'solo', slang ~OOKI ~OOKI rooki 'cigarette',and TEERI
teeri 'black grouse'.
°
If the first vowel to the left of is not adjacent,the result is vowel copy,
e.g. PUD_XDN puu-hun 'into a tree'; for more examplesseethe illative case
andthe discussionof X andQ (pp. 154-5).
152 FINNISH

Table5.2 Finnish consonant

phonemes

1 2 3 4 5

Nasals mn P IJ
Stops,-v P t k ?
Stops,+v b d g
Fricatives,+v v h
Fricatives,-v f s S
Lateral I
Trill r
Glide j

Note: Non-marginalphonemesare given in bold. -v = voiceless,+v = voiced. Columns

correspondto placesof articulation: I = bilabial (nasalsand stops)or labiodental(fricatives),2
=dental(but d is moreusually alveolar,andthe defaultpronunciationof s is somewhat
retroflex), 3 =palato-alveolar(and hushing),4 =dorsove1ar(h is glottal prevocalically),5 =
glottal.

Consonants
The consonantphonemesof Finnish are set out in Table5.2. The orthography
writes the consonantsas given in the table,with two exceptions:1), which only
occursintervocalicallyand long, is written <ng>; and lis not written at all.
The difference between the vowel i and the glide j may usually be
recoveredfrom the morphemicshape,e.g. #OLIU oljy 'oil' but OL.E=IO olio
'being'. The voiced stopsb and g and the fricativesf and sare restrictedto
foreignismsand slang; some,especiallyolder, speakersdo not have someor
all of these,and usep, k, (h)v, and s instead.The phonemesd, 1), and 1, on
the other hand, occur frequently in native non-affectivevocabulary,and thus
may be seenas belongingto the coreinventory of phonemes(Karlsson1983:
64-6). In a morphophonemicanalysis such occurrencesare predictable,
however, and the flrst two Can be unambiguouslyrecovered,via a process
called gradation (outlined below), from morphophonemicT and N, e.g.
RATA-N rada-n 'track sG', ONKE-N ongen [oIJIJen] 'flsh hook sG'. The

Figure5.1 Finnishmorphophonemes
andoperators

MN R
IU A PTK R
E 0 VSH R
A L
R

o x Q
t

Note: After Austerlitz 1967.For the operatorsQ X 't D, seetext.

 FINNISH 153

glottal stop ? is only one of a range of phonetic phenomenawhich may

emerge from the relatively abstract morphonemicoperator Q; see below.
Instancesof standard-language d which arise from morphophonemicT have
considerabledialectalvariation,including [w], [v], [j], [1], [r], and [0].
Consonantlength is distinctive for m n p t k sir j intervocalically;
prevocalicallyto the right of nasalsand liquids, length is distinctive for p t k
s. In intervocalicconsonantclustersof two, the fIrst consonantis pronounced
with (non-distinctive)greaterduration.SeeKarlsson1983: 104-29for details
on surfaceconsonantdistribution.

Gradation
With certain systematicexceptions,the stopsP T K are all convertedto less
fortis counterparts,calledweak grade,when they occurin the onsetof a non-
initial closedsyllablewithin the word, providedthat syllablecontainsa single
vowel or vowel-i sequence.Theselessfortis counterpartsarebriefly specifIed
below, using the suffIxation of the genitive singular suffIx (-N) as a sample
conditioningenvironment(for more detailedtreatmentsof Finnish consonant
gradation,seeFromm 1982: 49-56,Austerlitz 1967: 24):
1 The geminate sequencesPP TT KK are read as single P T K, e.g.
KORPPI-Nkorpin 'ravensG',TATTI-N tati-n 'boletussG',TADKKA-N
taakan 'hearth sG'. This sUbtype is called quantitative gradation, as
distinct from the remainingtypes,calledqualitative.
2 Intervocalically, T is read as d; P is read as v, as is K in the sequence
UKU; in other intervocalicpositions,K surfacesas a variety of prosodic
and glottalic transitional phenomena,including lengthening of the
precedingvowel, e.g. TAIKA-N taian [taiiijan] 'magicsG', andwhispery
and/or creaky voice. In three morphemes,K exceptionally surfacesas
zero in the weak grade: AIKA aika 'time', POlKA poika 'boy', and
OIKO- 'straight' as in OIKO-SSA ojossa 'outstretched; fIxed (of
bayonets)'.T is also read as d in the cluster HT, which is a by-product
of KT in a few core words: seeNumerals,p. 168.
3 P T KsG',assimilatefully to a nasal to their left, e.g. SAMPE-N samme-n
'sturgeon sG', KANTE-N kanne-n 'lid sG', HANKE-N hange-n
[haIJIJe-n] 'thick snow-cover'.
4 To the right of L and R, P is read as v; T assimilatesfully to these
consonants(e.g. KORPE-N korve-n 'backwoodssG', PARTA-N parra-n
'beardsG'). To the right of L, R, and H, K surfaceseither as zero (e.g.
~ALKA-N~ALKA-Nniilii-n 'hungersG') or, beforeE, asj (e.g. ifJALKE-N jiilje-n
'trace, track sG'). In some words, K in the cluster HK is resistantto
gradation,e.g. #SAHKO-N siihkO-n 'electricity sG'.
Note that word boundary and syllable boundary need not coincide,
e.g. VAHINKO-N+ILO DAMAGEIMISFORTUNE-G+JOY vahingonilo
[vahiIJIJonilo] 'Schadenfreude'.
154 FINNISH

Exceptions to gradation may be classified as morphological or lexical.

Systematicmorphological exceptionsare found in (la) fonns inflected for
person(seebelow); (lb) the readingst and tt for t in the genitive plural (see
below), as well as in certaindeictics, e.g. sitten 'then' : siten 'thus'; (lc) the
sP and pG of nominals derived with =ISE, and the first infinitive, perfect
active participle, and third-person imperative fonns of phonaesthemic-
affectiveverbsderivedwith =ISE, e.g. VALKO=ISE-tA valkoista 'white sP',
VIKISE-tAQ vikistii? 'to squeal/squeak/peep'; (ld) fonns in which a vowel
sequenceendingin i arisesfrom the suffixation of an -1-, with cancellationof
an underlyingvowel to the left of this -1-, fonnulaically VV-I-, e.g. ~ARKEA VV-I-,
-I-SSA tiirkeissii 'importantpIne', HAMPAX-I-SSA (> HAMPAill_I-SSA »
hampaissa'tooth pIne', contrast ilJARKE-I-SSA-~
HAMPAX-I-SSAjiirjissiiiin 'in hislher
senses(s3plne)'.Systematiclexical exceptionsare (2a) certainfonnal types
of proper noun, which have only quantitative gradation (Austerlitz 1979,
Vesikansa1989a:293-5); (2b) acronymicandstump-wordabbreviations,e.g.
alko 'Finnish state-runalcohol retail outlet', sapo 'secretpolice' < sa(la)+
po(liisi); (2c) slang,nursery,andotherwiseaffectivevocabularysuchas raaki
'a smoke',pupu 'bunny', tapa 'short (of tails)'; (2d) unassimilatedforeign-
isms,including auto 'car'.

ConsonantalMorphophonemicOperators
Theseare Q, t, and X, all entities which are more abstractthan the fifteen
morphophonemes I E A 0 U P T K M N V S H L R. Q, t, and X differ from
the vocalic operatorD (introducedon p. 151) in that they can function as
consonantsandthuscanclosea syllable,therebytriggering gradation.
The operatorQ occursonly in therebytriggering position, and hasa number
of non-segmentalrealizations.In prepausalposition, it is realizedas a glottal
stopor as relatively short durationof the precedingvowel, e.g. secondperson
singularimperativeISTU-Q [istu?], [istii] 'sit!', ANTA-Q [anna?] 'give!' To
the left of most inflectional suffixes, Q is realized as D, e.g. HUONEQ-N
huoneen'room sG'. It assimilatespartially or wholly to the initial consonants
of words andenclitics to its right, usually with transitionalstretchesof creaky
and/or breathy voice. In traditional descriptions, this phenomenenonis
usually called coda (loppu-) or onset (alku-) 'doubling' (kahdennus),i.e.
loppukahdennus, alkukahdennus, after its acousticallymost salient feature,
the increasedduration of the oral segmentalconsonantof the following
morpheme; see Karlsson 1983: 348-50. The presenceof Q in certain,
especiallyinflectional, morphemes,varies dialectally, and the behaviourof
the Q in the allative case suffix -LLEQ deviatesslightly from the pattern
outlinedhere;seepp. 157-8.
Tau, i.e. the operatorwritten here as t, occurs in both root and suffixal
morphemes.It behaveslike T exceptthat it alternates,under circumstances
which are morpheme-specific,not only with d but also with zero (and, in the
genitive plural, with tt). For examples,see the partitive (-tA) and genitive
 FINNISH 155

plural (-'tEN) suffixes of the nominal, and the infinitive suffix (-'tAQ) of the
verb.
X occurs both morpheme-final and suffix-initial. In morpheme-final
position,X is readas s at word-endandbefore't, as 0 elsewhere,e.g. # AKEX
aes 'plough', sP # AKEX-'tA aes-ta, sG # AKEX-N akee-n. For further
examplesseethe discussionof nominal and verbal X-stems. X may also be
used to capture the suffix-initial s - h alternationwhich characterizesthe
illative suffix, q.v.

Historical Backgroundofthe Consonants

Relatively minor changes in the consonantalinventory of proto-Fennic
separateFinnish from the rest of the Fenniclanguages.On the otherhandthe
proto-Fennic consonantalinventory deviates sharply from that of proto-
Saarnic, putatively its closest congener; the sole shared Saarnic-Fennic
innovationin the realmof the consonantsis the mergerof p U * 6 L and *t - into
*t- (Viitso 1996: 262). The major proto-Fennicinnovations were the loss,
through merger,of the palatalizedapicals *n : *ni > n, *1 : *li > 1, and *s :
*si (and *c i ) > s, and the mergerof initial *8- and *c- (yielding Finnish h-)
andnon-initial *t and *c (yielding Finnish t). Finnishnon-initial h hasat least
eight sources(Hakulinen 1961: 36-8).
In consideringthe developmentof the Finnish consonants,it is important
to distinguish four kinds of position: (1) root-initial, (2) central, i.e. the
position betweenthe first and secondsyllables,(3) marginal,i.e. the position
betweenthe secondandthird syllables,and (4) final position. Whereasproto-
Fennic consonantsin initial position (apart from the affricates) remained
intact in terms of constriction type, consonantsin the other positions
underwent different types of weakening. In central position, stops were
pronouncedshorter,or even as fricatives, when their following syllable was
closed (syllabic gradation); this positionally determined, morphologically
conditionedalternationis the sourceof Finnish consonantgradationas seen
in e.g. PATA-N pada-n 'pot sG'. In marginal position a similar, but not
identical, kind of weakening occurred, regardlessof the opennessof the
following syllable (rhythmic gradation);this positionally determinedprocess
led to allomorphy of the kind seenin the Finnish partitive, e.g. maa-ta 'land
sP' « *maa-ta, with *-t- in central position) v. pata-a 'pot sP' « *pata-ta,
with *-t- in marginalposition). In final position, *k and (*8 » *h became1,
and *m becamen. Massiveanalogicallevelling has coveredup most of the
traces of rhythmic gradation in Finnish; for the distinction syllabic v.
rhythmic, rather than the traditional root v. suffixal, gradation, see now
Helimski 1996. An overview of the developmentof consonantsin central
position in Finnish and North Saarniis given in tabular form by Korhonen
(1981: 189-92).
A few sampleforms andcognateswill illustrate someof the developments
sketchedabove (SN = North Saarni, ME = Erzya Mordva). Finnish syks=y
156 FINNISH

'autumn'< pFU *sjiiksi > SN cakca; Finnishsiil(=)i '(outstretched)arms;lap'

< pU *siili > SN salta; Finnish kynne-ssa< *kiinti-s-na , ME kence-se<
*kinci-s-na '(finger)nail, claw sIne', with root from pFU *kiinci > SN gazza;
Finnish kala-a 'fish sP' < pFU *kala-ta > ME kal-do 'fish sAbl'; Finnish
vene?'boat sN' < *wini=si > SN Janas; venee-t'boat pN' < *wini=si-t > SN
Jatnasa-t;Finnishnido-? 'stitchestogethers2imp' < *njiba=i-k, ME niedja-k,
SN njaile.
The functions of the operatorQ correspondto the various positional and
combinatoryfates of *k and *s (as in NITO-Q nido-? and VENEQ vene?
'boat', above) in morpheme-finalposition. Tau (r) is simply a morphono-
logically different sort of T, in terms of history: one which was susceptible
to weakeningby rhythmic gradation.The operatorX is a multiple, covering
phenomenarelatedto the prehistoryof both s andt in marginalposition.

Inflection

Nominals
As is to be expectedin a languagewithout agreementclasses,the distinction
noun v. adjectivein Finnish is mainly a semanticand distributional/syntactic
one. Eventhe comparativesuffix (=MPA) canbe attachedto stemswith noun,
usually spatial, reference, e.g. RANTA=MPA-NA SHORE=cfv-ess
ranne=mpa-na 'closer in to the shore'.A few nominals which are semant-
ically andsyntacticallyadjective-likearemorphologicallydeviantin that they
are uninflectible: PIKKU pikku 'little', ENSI ensi 'next (in sequence)',
VlIMEQ viime? 'last/mostrecent(in sequence)',ERI eri 'separate,distinct,
different', KOKO koko 'whole, entire', IOKAjoka 'each,every'.
Finnish nominals are inflected for number,case,and person.As in most
Uralic languages,zero (-0-) encodessingular(or non-plural)number,absence
of person,andnominativecase.
Nominal plural forms are built with the suffix -T (in the nominative)and
-1- (elsewhere;but seethe discussionof the genitive plural, p. 160f.). Along
with the meaning 'more than one' the suffix -T encodesa pragmatic/textual
componentof definitenessor knownness;this is not true of -1-. Examples:
TALO-0 talo 'house', pN TALO-T talo-t 'the (definite/defined set of)
houses',pIne TALO-I-SSA talo-i-ssa 'in (the) houses'.The greatestareaof
morphophonemiccomplexity in Finnish nominal inflection centres on
operationstriggeredby the plural suffix -1-; seepp. 159-62.
The casesuffixes are attachedto the right of the numbersuffix. There are
twelve cases,and in the analysisoffered here, eachcasesuffix has only one
underlying form, exceptthe genitive, which has distinct singular and plural
suffixes.The twelve casesmay be divided into four groupsas follows:
 FINNISH 157

Table5.3 Four-by-threematrix of Finnishcases

Stasis Source Goal

Grammatical -0 -N/-tEn (seetext)

Semi-grammatical -NA -tA -KSE
Local interior oSSA -STA -XDN
exterior -LLA -LTA -LLEQ

1 Three purely grammaticalcases:N[ominative] -0, G[enitive] (singular

-N, Genitive plural -LEN), and A[ccusative], with various suffixes
borrowedfrom othercases;
2 Three semi-grammaticalcases: Ess[ive] -NA, P[artitive] -LA, and
Trans[lative] -KSE;
3 Three interior-local cases: Ine[ssive] oSSA, Ela[tive] -STA, Ill[ative]
-XDN;
4 Three exterior-local cases: Ade[ssive] -LLA, Abl[ative] -LTA, and
All[ative] -LLEQ.

Following Austerlitz 1968: 1338(cf. Fromm 1982: 71) we may think of the
Finnish casesas they are set out in Table 5.3, i.e. in terms of a four-by-three
matrix defined by function and reference(grammaticalv. spatial) on the one
hand,andthe trichotomy stasis::motion away: motion towardson the other.
Grammaticalcasemarking is complicatedby the fact that nounshave no
dedicatedaccusativesuffix. In the singular,nounsuse-N (homophonouswith
the genitive) if the finite verb is inflected for person, and with active
participles;they use-0 (homophonouswith the nominative)otherwise,e.g. if
the governing verb is impersonalor the first infinitive. Plural nouns use -T
(homophonouswith the plural nominative)for all instancesof the accusative.
The cardinal numeralsuse -0. Both singular and plural personalpronouns
have a dedicatedaccusativesuffix -T, with a probablecognatein Khanty.
Figure 5.2 is an attempt at capturing the interaction of some of the
grammatical,syntactic,and aspectualfactorsat play in suffix selectionfor the
nominative,accusative,genitive and partitive cases.Notice that the partitive
standsout as being the only caseto haveits own suffix, not sharedwith any
othercase.For subject-markingtypes 1 and 2, seeSyntax,p. 176.
Many grammariansalsorecognizethreeadverbialcases: Abe[ssive] =TTA,
Inst[ructive] =(I)N, and Com[itative] =INE-, e.g. Fromm 1982: 84-5,
Karlsson21979: 133. As their nameimplies, the statusof the adverbialcases
is not entirely solid; they are of restricted distribution and stylistically
charged.
The E final in the translativesuffix is readas i whenin word-final position,
e.g. SAMMAKKO-KSE sammakoksi'frog sTrans,= turnedinto a frog'. The
Q final in the allative suffix is regularly glottal stop [?] when word-final
158 FINNISH

Figure5.2 Finnishgrammaticalcasemarking

N A G P

Definite subject

2 2

Nounsin singular I~ Finite verb inflicted

- for person

Cardinalnumerals -0 -N
-tA

Personalpronouns

Nounsin plural -N -tEN

Resultativeobject

(~ULLU-LLEQhylly-lle? 'shelf sALL'), but is cancelled before person

(~ULLU-LLEQ
suffixes; seep. 159. In the illative suffix -XoN the X is read as s, h, or zero,
°
and the element is read as simple vowel copy, ee, or ii, dependingon the
surroundingsegmentsand(in part) on the stem-typeinvolved; seestem-types,
pp. 159-63.The L of the genitive plural suffix -LEn, like all instancesof L, is
susceptibleto cancellation,i.e. alternateswith zero; it is unique, however,in
allowing the parallel readingsd and tt in forms built with the plural suffix -1-,
e.g. OSA=STO-I-LEN osastoidenlosastoitten 'departmentpG', and in resist-
ing gradationin sequencessuch as that of MERE-LEN merten 'seapG'; see
Austerlitz 1983.
The semi-grammaticalcasesstandsomewherebetweenthe local casesand
the grammaticalones;their original spatialmeaningsare still evidentin their
usewith postpositionsand otherdeictics.
The personalforms chiefly denote possession.They are built with the
suffixes (s1) -N!, (s2) -SI, (sp3) -NSA(Q) - _ON, (p1) -MME(Q), and (p2)
-NNE(Q), and are attachedin agglutinatingfashionto the right of any number
andcasesuffixes, with two exceptions:

1 the functions of the genitive and accusative singular and of the

nominativeplural are filled by the nominativesingular.Thus TALO-SI is
not only sN, but also sG, sA, andpN of 'houses2'.
 FINNISH 159

2 the Q final in the allative suffix -LLEQ, and the N final in the illative
suffix -XoN, are read as zero before personsuffixes. Examples:#ISA-
LLEQ-NI isii-lle-ni 'fathersl sAll', LAUKKU-XoN-SI laukku-u-si'bag
s2 sIll'.

Recall that person suffixes are a systematicexceptionto gradation,e.g.

KOTI-NNE(Q) kotinne(1) 'your home'.
The element(Q) final in most personsuffixes has a varied distribution
acrossthe dialects;axis-of-discourseforms without Q, andthird-personforms
with Q, are probablythe morecommon;seethe sectionon history, pp. 167-8.
The variant _oN of the sp3 suffix is usualto the right of casesuffixes that
endin a vowel (including the allative, which losesits Q to the left of all person
suffixes, including _oN, yielding the sequence-lIe-en). Further examples:
TALO-SSA-MME talo-ssa-mme'housepI sIne', LAPSEI-LLE-NNE Zaps-
i-lle-nne 'child p2 pAll'.
Inflection Types
At the most abstractlevel, all nounsinflect accordingto only one type. At the
other extreme,we have the descriptiveacribie of the Nykysuomensanakirja
(NSK), the Finnish Academydictionary, wherein eighty-two non-compound
nominal inflection types are distinguished(NSK 1: xi-xvi; re-analysedand
annotatedin Tuomi 21980: 525-39). To make the discussiontractable,here
we distinguishthe following basic types: (1) long-vowel stems;(2) stemsin
final E; (3) mutatingstems(in final A andI); (4) consonantstems(in final Q,
X, and1'); and (4) stablestems(in final U, 0, and E2). It is a fruitful exercise
to compareKarlsson 1982: 203, who also distinguishesfive basic types, but
with quite different limits and memberships(Karlsson uses the nominative
singularas his point of departure,ratherthan a P~ini-like abstraction,as here
abstraction,
or in Eliot 1890).
1 Long-vowelStems
In accordancewith the rules statedabove,manyof theseare written as having
° as their secondmember.All Finnish monosyllabic nouns belong to this
type; there are about fifty such roots, if alongsidecore vocabularysuch as
PUo puu 'tree, wood' and TEo tie 'road' we include letter-namessuch as
VEE [vee] '(the letter) v'. There are also sometwenty-oddbisyllables,none
°
native, e.g. VAPA vapaa 'free' « Slavonic), FILED filee 'filet'. Sympto-
matic of this type is the readingoft' as t in the partitive singular,e.g. VAPA 0_
'tA vapaa-ta'free sP'.
°
To the left of the plural suffix -1-, final is cancelled:MAW-I-'tA ma-i-ta
'land pP', SOW-I-'tA so-i-ta 'bog pP'. The only other sourceof inflectional
complexity involves the allomorphy of the illative suffix -XoN. When
°
attachedto monosyllables,the X is read as h and the is read as a copy of
the first vowel to its left, e.g. MAA-XoN maa-han'land sIll'; when attached
to bisyllables,-XoN is read as -seen(optionally -hen to the right of EO) in
160 FINNISH

the singular, and -hin or -sUn in the plural: VAPAD_XDN vapaa-seen'free

sIll', FfLED_XDN filee-seenlfilee-hen'filet sIll', VAPAIZi_I-XDN vapa-i-sUni
vapa-i-hin 'free pIll'.

2 E-stems
The E final in all E-stemsis readas i in the citation form (= sN); E-stemsthus
formally resemble I-stems in any dictionary which does not indicate
inflectional type. Any T penultimatein an E-stemis readas s to the left of i,
whether this i is from underlying E (VETE-0 vesi 'water') or from the
pluralizer -1- (VETE-I-SSA vesissa 'water pIne'). There are some 230
E-stems,someof which dateback to p(F)U and many of which are textually
frequent.The chief variableswhich producethe NSK sUbtypesare:

2a the cancellationv. persistenceof stem-final E to the left of the partitive

suffix (-TA);
2b the order of preferencefor parallel forms emergingfrom dual output of
variable2a;
2c double(-I-TEN) or single(-TEN) plural-markingin the pG;
2d the applicationv. non-applicationof assimilationrules to the essivesuffix
(-NA).

An exampleof the first type of variableis providedby the words ovi 'door'
andsuoni 'vein': ovi hassPove-a,with persistenceof stem-finalE (OVE-l'A),
but suoni has sP suon-ta, with deletion of E (SUONE-TA). The deletion v.
cancellation of stem-final E in such circumstancesis connectedwith the
nature of the consonant(ism)which precedesit: if this is a single dental (N
T S L R), the E is cancelled,witnesssuon-taand VETE-TA vet-ta 'watersP',
KUUSE-TA kuus-ta 'sprucesP', NUOLE-TA nuol-ta 'arrow sP',and NUORE
-TA nuor-ta 'young sP'. In eight words (NSK types 32 and 33), E is also
cancelledto the right of H, e.g. LORE-TA loh-ta 'salmonsP'; this behaviour
reflectsthe apical origins of H in most if not all of thesewords. Note also M
> n, obligatory in textually frequentLUME-TA lun-ta 'snow sG', optional in
rarerTUOME-l'A tuome-a,TUOME-TA tuon-ta 'chokecherry(tree)'.
The secondtype of variablearisesin the formation of the sPof stemsending
in E precededby a clusterwhosesecondmemberis S; the NSKlists ten such
stems(types45-50), and reportsthat E-deletionis optional for all but two of
them,viz. LAflSE-TA las-ta 'child sP' andVEI1SE-TA veis-ta'knife' (notethe
concomitantcancellationofP andT). The remainingeight stemshaveparallel
partitives singular, with various preferences,e.g. SUKSE-l'A sukse-a is
favouredoverSUKSE-TAsus-ta,both 'ski sP';with UKSE-TA us-ta,UKSE-l'A
ukse-a,both 'doorwaysP'theorderof preferenceis reversed.
The third variable reflects the prehistoryof the pG suffix -TEN, in which
the initial T was originally a plural marker. The T in this suffix behaves
uniquely in that it resistsgradationin forms such as the stylistically charged
 FINNISH 161

VETE-tEN vetten 'water pG' (historically: *wete-te-n) alongsidesecondary,

andnormal,VETE-I-lEN vesien.
The fourth variableis relevantto the languageof verbal art, in which forms
resulting from E-deletion and assimilation such as NUORE-NA > nuorra
'youngsEss'arepossible.
Certain E-final derivational suffixes are characterizedby their own
morphophonemicoperations:

(a) In nominals formed with =(I)SE, the S is read as n in the nominative

singular, and the nominative is marked with a suffix -N, e.g. NA=ISE
nainen 'woman' (cf. the sectionon numerals,p. 168);
(b) In nominalsformed with =NTE, which forms ordinal numerals,the N is
cancelledin the sP,e.g. VIITE=1lffE-tA viidettii 'fifth sP'.
(c) Stemsbuilt with =UUTE shift this suffIx to =UUKSE to the left of the
-1- plural suffIx, e.g. SUURE=UUTE suuruus '(large) size', sG SUURF.
=UUTE-N suuruude-n,pG SUURE=UUKSE-I-lEN suuruuks-i-en.
(d) In the perfectiveactive participle, =NUtE, the sequenceUt is readas ED
to the left of all suffIxes exceptthe sP, e.g. SAA=NUtE-N saaneen'that
hasacquiredsG'.

About forty-five E-stems owe their somewhatdeviant paradigms to the

presenceof a t, e.g. OLUtE olut 'beer',sG OLUlE-N olue-n. In most cases,
the sequenceUtE is synchronicallya derivational suffix, e.g. PIL VH=UtE
pilvyt 'little cloud (poetic)'.

3 Mutating StemsEnd in Either I or A

I-stems are of two basic types: thosewhich form their pG with -tEN (NSK
4) and thosewhich use both -tEN and -I-tEN (NSK 5 and 6). Stemsof the
latter type are all at leasttrisyllabic. In all I-stems,the stem-final I is read as
e to the left of the plural suffIx -1-, e.g. pP RISTI-I-tA riste-j-ii 'crosspP'.
The chief variableswhich producethe NSK subtypesof A-stems are the
mutation or cancellationof the stem-final vowel. To the left of the plural
suffIx -1-, stem-final A is normally either (a) read as if 0 (i.e. 0 or 0,
dependingon prosody),or (b) it is cancelled.The conditionswhich determine
which, if either, of the two operationsa or b is to apply to a given string are
multiples which involve the interaction of syllable-count, front v. back
prosody, the consonant(ism)immediately preceding the stem-final vowel,
homophonicpressure,and the readingof t in the pG (Karlsson1982: 282-6
is a conciseaccount,with a slightly different approach;seealsoTuomi 21980:
530, notes6 and7).
For example,the A final in LODLA luola 'cave' is cancelledto the left of
plural -1- regularly, i.e. in accordancewith a rule which prescribessuch
cancellationin bisyllables with a labial vowel in the first syllable: LODLA
-I--tA luol-i-a 'cave pP'. The A final in #KESA 'summer'is also deletedto
162 FINNISH

the left of plural -1-, again regularly, but this time in accordancewith a rule
which prescribessuch cancellationin all front-prosodic bisyllables. The A
final in SODLA 'salt' is not cancelled, eventhoughthereis a labial vowel in
its first syllable (SODLA-I--tA suoloja 'salt pP'), thus avoiding a homophonic
clashwith SODU~rI--tA SODU~rI--tAsuol-i-a 'intestinepP'.
A further, more precious,examplemay be drawn from the I-stems. The
NSK distinguishesthe paradigmsof paperi 'paper' (NSK 5) and banaali
'banal' (NSK 6) ostensibly becausein the pG of the latter, the non-
cancellationof T producesforms which are stylistically neutral: we haveboth
PAPERI-lEN and bANADLI-lEN giving paperi-enand banaali-en,but only
papere-i-den(PAPERI-I-TEN) is given as neutral,while the parallel banaale-
i-den (bANADLI-I-TEN) is given in brackets.
Certain A-final derivational suffixes are characterizedby their own
morphophonemicoperations:

(a) The A final in the comparativesuffix =MPA is readas i in the nominative

singular, e.g. KORKEA=MPA korkeampi 'taller', cf. sG KORKEA=
MPA-N korkeamma-n.Stem-finalA in bisyllablesis readas e beforethis
suffix, e.g. VANHA=MPA vanhempi'older' (cf. vanha 'old');
(b) The A final in the superlative suffix =IMPA is read as zero in the
nominativesingular,andthe resultingimpermissablefmal sequence*MP
is read as n, e.g. KORKEIFIMPA korkein 'tallest', cf. sG KORKEA
=IMPA-N korkeimma-n.Before the T of the sP, final MPA is read as n,
e.g. KORKEIFIMfA-TA korkein-ta;
(c) The morphophonemicbehaviourof the privative suffix =TTOMA is the
same,mutatis mutandis,as that of the superlative,e.g. ONNE=TTOMA
onneton 'unlucky, unhappy', sG ONNE=TIOMA-N onnettoma-n,sP
ONNE=TIOMA-TA onneton-ta.

4 Consonant-finalStemsendin Q, X, or r:
The inflectional peculiaritiesof Q-stemsall flow from the propertiesof Q,
outlined above. The sequenceQ-T, which arisesin the partitive singular, is
read tt, e.g. SATIFEQ-TA sadetta'rain sP'. The X final in X-stemssurfaces
as s before the zero of the sN and the T of the sP, but as D elsewhere,e.g. sN
HAMPAX hammas'tooth', HAMPAX-N hampaa-n'tooth sG', HAMPAX-
TA hammas-ta'tooth sP'. Tau occursfinal in very few words, of which only
#KEVAT keviit 'spring' is frequent; its behaviouris identical to that of X,
exceptthat it is readas t whereX is readas s, e.g. #KEVAT-N keviiii-n 'spring
sG', #KEVAT-TA keviit-tii 'springsP'.
A few Q-stems and X-stems are synchronically monomorphemic,e.g.
HUONEQ huonel 'room', VIERAX vieras 'guest, foreign-', but most are
derivatesbuilt with the productive suffixes =EQ, e.g. #PEITIIFEQ peite1
'cover(ing)', cf. #PEITIA- 'covers', SATIFEQ sadel 'rain', cf. SATA- 'it
rains', =IAX, e.g. ANTIFELB=IAX antelias 'generous',cf. ANTIFELE-
 FINNISH 163

'gives on several occasions',and =KKAX, e.g. LAHIA=KKAX lahjakas

'gifted', cf. LAHIA lahja 'gift'.

5 StableStemsExhibit no Stem-finalMorphophonemicPeculiarities
To this classbelong all stemsending in U, 0, or E2; in all, easily over 6,000
items. Examples are SATU 'fairy tale', #OLIU aljy 'oil', OLH=IO olio
'being', ~OLMO 2
~OLMO halma 'fool', and fAdE fade 'father (slang)'. The class
representedby this last item is fast growing; it containsneologismssuch as
ALE2 ale 'sale,periodoflower prices' and slang.
To the class of stable stems could also be assigned unassimilated
foreignismswhich endin an underlyingconsonantotherthanQ or X, suchas
SLALOM slalom, NAILON nailon 'nylon', or SANSKRIT sanskrit. These
stems append an i for all of their inflected and derived forms built with
consonant-initialsuffIxes,e.g. NAILONi-STA nailonista 'nylon sEla'.

To give an idea of the sorts of variety which are concealed by the

morphophonemiclumping outlined above,Table 5.4 setsout the singularand
plural genitive, partitive, and illative forms of thirty-two subvariantsof the
five stemtypesoutlined above.The lexical examplesare a slight reductionof
the set chosenby Sarnmallahti(1989) for his concisepresentationof Finnish
nominalinflection types.The forms are given in morphophonemiccode,with
operationsindicatedas follows: (1) cancellationof an underlying segmentis
indicatedby solidus,e.g."1; (2) segmentswhich undergootherchanges(e.g.
gradation;T > s to the left of i) are left unmarked.Vowel harmony is not
indicated, nor are the readings of ° or X D , save where these serve to
distinguishmultiple surfacings,as in korkeihinlkorkeisiin'high pIll'.

PostpositionsandPrepositions
These are for the most part spatial nouns with defective paradigms.
Postpositionsoften use the semigrarnmaticalsuffixes, or variants of these,
with spatial meaning, e.g. TAKA-NA BACK.SPACE-esstakana 'behind',
TAKA-tA takaa 'from behind',TAKA-KSE-Q taakse?BACK.SPACE-trans-
lat 'to behind'.
There are about thirty nouns of this type: most are usedas postpositions
in neutralstyles,but canalsobe usedprepositionallyin poetic or othermarked
language,e.g. VETE-N ALLA WATER-G UNDER vede-naUa 'under (the
water)" ALLA VETE-N aUa veden 'under (the) water (poetic),. Neutral as
prepositionsare ennen 'before', ilman 'without', paitsi 'besides,except',
vasten'against',all of which governthe partitive, e.g. vastenseinii-ii 'against
the wall'. The partitive is in fact the default casefor prepositions,and the
genitive for postpositions,e.g. keskeUii lattia-a, lattia-n keskeUii 'in the
middle of the floor' , but therearealsoa few nounswhich take one or the other
casein both positions,e.g. liipi seinii-n, seinii-n liipi 'throughthe wall' , kohti
seinii-ii, seinii-ii kohti 'towardsthe wall' .
Table 5.4 Subvariants of Finnish nominal inflection types

Sammallahti(1989: 503-4)
no. sN sG sP sIll pG pP pIll Gloss

6 VAL!(=O VAL!(=O-N VAL!(=O-t'A VAL!(=O-XDN VAL!(=O-I-tEN VAL!(=O-I-iA VAL!(=O-I-XDN light

valo valon valoa valaon valojen valoja valoihin
6 OSA=STO OSA=STO-N OSA=STO-iA OSA=STO-XDN OSA=STO-l-tEN OSA=STO-I-iA OSA=STO-I-XDN department
osasto osaston osastoa osastoon osastojen osastoja osastoihin
OSA=STO-I-tEN OSA=STO-I-tA
osastoiden osastoita
2 RISTI RISTI-N RISTI-iA RISTI_XDN RISTI-tEN RISTI-I-iA RISTI-I-XDN cross
risti ristin ristia ristiin ristien ristejii risteihin
5 PAPERI PAPERI-N PAPERI-iA PAPERI-XDN PAPERI-tEN PAPERI-I-iA PAPERI-I-XDN paper
paperi paperin paperia paperiin paperien papereja papereihin
PAPERI-I-tEN PAPERI-I-tA
papereiden papereita
3 KALA KALA-N KALA-iA KALA_XDN KALA-I-tEN KALA-I-iA KALA-I-XDN fish
kala kalan kalaa kalaan kalojen kaloja kaloihin
4 KOIRA KOIRA-N KOIRA-iA KOIRA-XDN KOIM-l-tEN KOIR!(-I-iA KOIM-I-XDN dog
koira koiran kairaa kairaan katrien koiria kairiin
9 OMENA OMENA-N OMENA-iA OMENA-XDN OMEN!(-I-tEN OMEN!(-I-iA OMEN!(-I-XDN apple
omena omenan omenaa omenaan omenien amenia omeniin
OMENA-I-tEN OMENA-I-tA
omenoiden omenoita
10 KULKE=IA KULKE=IA-N KULKE=IA-iA KULKE=IA-XDN KULKE=IA-I-tEN KULKE=IA-I-tA KULKE=IA-I-XDN wanderer
kulkija kulkijan kulkijaa kulkijaan kulkijoiden kulkijoita kulkijoihin
11 SOLAKKA SOLAKHA-N SOLAKKA-iA SOLAKKA-XDN SOLAKKA-I-tEN SOLAKKA-I-tA SOLAKKA-I-XDN slender
solakka solakan solakkaa solakkaan solakoiden solakoita solakoihin
SOLAKKA-l-tEN SOLAKKA-I-iA SOLAKKA-I-XDN
solakkojen solakkoja solakkoihin
12 KORK=EA KORK=EA-N KORK=EA-iA KORK=EA-XDN KORK=E!(-I-tEN KORK=E!(-I-tA KORK=E!(-I-XDN high
korkea korkean korkeaa korkeaan korkeiden korkeita korkeihin
KORK=EA-tA KORK=E!(-I-XDN
korkeata korkeis;;n
13 ISO=MPA ISO=MPA-N ISO=MPA-l'A ISO=MPA-XDN ISO=MP!(-I-tEN ISO=MP!(-I-iA ISO=MP!(-I-XDN larger
isompi isomman isompaa isompaan isompien isompia isompiin
Table 5.4 (Continued)

Sammallahti(1989: 503-4)
no. sN sG sP sIll pG pP pIll Gloss

32 #SISX=IMPA #SISX=IMPA-N #SISX=IMPX-tA #SISX=IMPA-XDN #SISX=IMPX-I-iBN #SISX=IMPX-I-t'A #SISX=IMPX-I-XDN innennost

sisin sisimmiin sisintii sisimpiitin sisimpien sisimpiii sisimpiin
#SISX=IMPX-TEN
sisinten
26 OSA=TTOMA OSA=TTOMA-N OSA=TTOMA-t'A OSA=TTOMA-XDN OSA=TTOMX-I-iBN OSA=TTOMX-I-t'A OSA=TTOMX-I-XDN portionless
osaton osattoman osattornaa osattomaan osattomien osattomia osattomiin
14 MAo MAD-N MAD_TA MAD_XDN MA"l.I-tEN MAIZl-I-tA MA"l.I-XDN land
maa maan maata maahan maiden maita mathin
15 SOD SOD-N SOD-tA SOD_XDN SO"l.I-tEN SO"l.I-tA SO"l.I_XDN bog
suo suon suota suohon soiden saita soihin
16 VAPA D VAPAD-N VAPAD_tA VAPAD_XDN VAPA"l.I-tEN VAPA"l.I-tA VAPAIZl_I-XDN free
vapaa vapaan vapaata vapaaseen vapaiden vapaita vapaisiin
18 OVE OVE-N OVE-iA OVE_XDN OVIl-l-tEN OVIl-l-tA OVIl-I_XDN door
ovi oven ovea oveen ovien ovia oviin
19 TOIME TOIME-N TOIME-t'A TOIME-XDN TOIMIl-I-iBN TOIMIl-I-t'A TOIMIl-I-XDN function
tDim; tDimen toimea toimeen toimien toirnia {aimiin
TOIMll-tA TOIMll-tEN
tointa tointen
20 KUDSE KUDSE-N KUDSll-tA KUDSE-XDN KUDSIl-I-iBN KUDSIl-I-iA KUDSIl-I-XDN spruce
kuusi kuusen kuusta kuuseen kuusien kuusia kuusiin
KUDSll-tEN
kuusten
21 LUME LUME-N LUMll-tA LUME-XDN LUMIl-I-iBN LUMIl-I-t'A LUMIl-I-X~
#KATll-i-XDN snow
fum; lumen lunta lumeen [umien [umia [umiln
22 #KATE #KATE-N #KATJl..tA #KATE_XDN #KATJl..I-iBN "KATJl..I-t'A #KATll-i-XDN hand
kiisi kiiden kiittii kiiteen kiisien kiisiii kasiin
23 #KUNTE #KUNTE-N #KUNTll-tA #KUNTE-XDN #KUNTJl..I-iBN #KUNTJl..I-tA #KUNTJl..I-XDN claw. nail
kynsi kynnen kYnttii kYnteen kynsien kynsiii kYnsiin
24 KAKTE KAKTE-N KAK1'Jl..tA KAKTE-XDN KAKTJl..I-iBN KAKTJl..I-t'A KAKTJl..I-X~
#KATll-i-XDN two
kaksi kahden kahta kahteen kaksien kaksia kaksiin
25 #KUTKll=IME #KUTKll=IME-N #KUTKll=IMll-tA #KUTKll=IMIl-I-iBN #KUTKll=IMIl-I-iBN "KUTKll=IMIl-I-t'A #KUTKll=IMJl..I-XDN coupling
kytkin kYtkimen kYtkintii kytkimeen kYtkimien kYtkimiii kYtkimiin
#KUTKll=IMll-tEN
kytkinten
Table5.4 (Continued)

Sammallahti(1989: 503-4)
no. sN sG sP sIll pG pP pIll Gloss

27 NA=ISE NA=ISE-N NA=ISE-tA NA=ISE-XDN NA=ISE-l-iEN NA=ISE-I-iA NA=ISE-I-XDN woman

nainen naisen naisfa naiseen naisien naisia naisiin
NA=ISE-tEN
naisten
28 "IANIKSE #IANIKSE-N #IANIKSE-tA #IANIKSE-XDN "IANIKSIl-tEN "IANIKSE-I-tA "IANKISE-I-XDN hare
janis jiiniksen jiinistii jiinikseen jiinisten jiiniksiii jiiniksiin
#IANIKSE-l-iEN
jiiniksien
29 OHUtE OHUiE-N OHUtE-tA OHUiE-XDN OHUiE-I-tEN OHUiE-I-tA OHUiE-I-XDN thin
ohut ohuen ohutta ohueen ohuiden ohuita ohuihinlohuisiin
30 VIITE=NTE VIITE=NTE-N VlITE=NTE-tA VIITE=NTE-XDN VIITE=NTE-l-iEN VIlTE=NsE-I-tA VIITE=Nsll-I-XDN fifth
viides viidennen viidettii viidenteen viidensien viidensiii viidensiin
31 OLE-NUtE OLIl=NUtE-N OLIl=NUtE-tA OLIl=NUtE-XDN OLIl=NUtE-I-tEN OLIl=NUtE-I-tA OLIl=NUtE-I-XDN been
ollut olleen ollul/a olleeseen olleiden olleita olleihinlolleisiin
34 #ISIi=UUTE #ISIi=UUTE-N #ISIi=UUTE-tA #ISIi=UUTE-XDN #ISIi=UUKSE-l-iEN #ISIi=UUKSE-I-tA #ISIi=UUKSE-I-XDN paternity
isyys isyyden isyyl/ii isyyteen isyyksien isyyksiii isyyksiin
33 VIERAX VIERAX-N VIERAX-tA VIERAX_XDN VIERAX-I-tEN VIERAX-I-tA VIERAX-I-XDN guest
vieras vieraan vierasta vieraaseen vieraiden vieraita vieraisiin
35 VENEQ VENEQ-N VENEQ-tA VENEQ-XDN VENEQ-I-tEN VENEQ-I-tA VENEQ-I-XDN boat
vene? veneen venettii veneeseen veneiden veneitii veineihinlveneisiin
 FINNISH 167

Most postpositionsand prepositions cantake person suffixes as well,

e.g. ALLA-SI allasi 'under you', VEDRE-XDN-NI viereeni 'to next to
me'.

Historical BackgroundofSuffixes
Both plural suffixes probably date back to pU, when they had essentially
the same distribution and function as in present-dayFinnish (Janhunen
[1981a: 30] classifies their ancestorsas plural case suffixes, pN *-t and
pGA *-j).
The historical backgroundof the case suffixes is also for the most part
clear. As for the grammaticalcases,the use of the zero-caseform to mark
certain direct objects as well as subjectsis in all likelihood an inheritance
from pU. The genitive/accusativesuffix -N is a mergerof thosetwo casesas
inheritedfrom pU, namely*-n and *-m.
Two of the semigrammaticalcaseswere originally local suffixes. The
partitive, attestedalso in Saamic,Mordva, Mari and Samoyedic,continuesa
pU separative*-tA (- *-tl). The essivecontinuespU locative *-nA; alone or
in combinationwith co-affixes, it has cognatesin all Uralic languages.The
translative is of disputed origin: it may be identical with the derivational
suffix =(U)KSE; it may be a compositeof two latives,*-k and*-s; it may even
be relatedto the predestinativeof Nganasan,Nenets,and Enets,cf. lanhunen
1989.
The present-dayFinnish local casesare composites,built by the addition
of primary casesuffixes to the elements*1 (exterior cases)and *s (interior
cases).The lines along which theseelementsdevelopedis not known with
precision, and their age and origin is disputed.In purely mechanicalterms,
however, the lines of descentof the Finnish inessive,elative, adessiveand
ablative casesare obvious, viz. *-s-nA > -SSA, *-s-tA > -STA, *-l-nA >
-LLA, *-l-tA> -LTA, whateverthe origins and original functions of *s and
*1 might havebeen.The Finnish illative is clearly a compositeof two latives,
one (*-s) with cognatesin Saamic,Mordva, and Mari, the other (*-n - *-n j )
of uncertaindistribution acrossFU (Kangasmaa-Minn1973); this particular
combination,*-sVn, is known only from Fennic and Saamic.The geminate
LL of the allative suffix -LLEQ arosevia analogy with the adessive;the Q
final in this suffix correspondsto n in Finnishdialectsandin the Mari pendant
-Ian.
The set of Finnish possessivesuffixes is a mergerof two parallel setsused
in nominative v. oblique-and-pluralin proto-Fennic.For example,whereas
the s2 suffix -SI, now used in all cases,continuesa proto-Fennics2 suffix
(*-te) which wasusedonly in the nominative,the suffixesofthe first andthird
person singular, -NI and -NSA(Q), continue proto-Fennic suffixes which
originally werenot usedin the nominative:
168 FINNISH

Proto-Fennic sl s2 s3 Finnish
Nominative *-me
*-te > -SI s2
*-sA
Non-nominative *-ne> -NI sl
*-nte
*-nsA> -NSA(Q) s3

As mentionedabove,thereis considerabledialectalvariation concerningthe

auslautof all personsuffixes savethosefor fIrst and secondpersonsingular;
the axis-of-discoursevariants with fInal Q continuean old pluralizer, either
*-k (as in Mordva pI -mok; cf. Korhonen 1981: 244) or *-t (as in SelkuppI
-mi~;cf. cf. Janhunen1981a: 32). The origin of the (Q) fInal in the third person
suffIx(es)is disputed;seeFromm 1982: 87, with literature.
Numerals
Each of the fIrst ten non-derivedcardinal numeralsdeviatesslightly from
canonin shape,morphophonemicbehaviour,or both:

1 KUUTE
~KTE yksi yhte- 6 KUUTE kuusikuute-
2 KAKTE kaksikahte- 7 #SEITSEMA seitsemiinseitsemii-
3 KOLME(Q) kolme(1) kolme- 8 KAKTE=KSA kahdeksankahdeksa-
4 mLIA neljii 9 KAKTE=KSA
~KTE=KSA yhdeksiinyhdeksii-
5 VIITE viisi viite- 10 #KUMMENE kymmenenkymmene-

The numeralfor '3' has a parallel stemKOLMA= usedin derivation; this

fInal E-A alternationis found in only two otherwords,SOoMESuomi,suomi
'Finland; Finnish (language)'- SOoMA=LAISE suomalainen'Finnish' and
ROoTSI Ruotsi, ruotsi 'Sweden;Swedish(language)'- ROoTSA=LAISE
ruotsalainen 'Swedish'.Nominal stemswith shapesresemblingVIITE and
KUUTE (i.e. long vowel plus T plus E) exist, but they are lexically and
textually rare. The cluster -lj- is unusualbetweenmid and low vowel, as in
neljii. Outside of poetic vocabulary,the abstractsequenceKT is read as ht
(- hd in weak grade) only in the numerals '1' and '2' (and in '8' and '9',
which arebasedon them),andin the infInitives of the only front-prosodictwo
verbs that end in the sequenceKE, ~AKEtAQ ~AKEtAQ niihdii 'to see' and TEKE
-tAQ tehdii 'to do'. The words for '7' to '10' take a nominativecasesuffIx
-N uniqueto themandto nominalsformed with the derivationalsuffIx =ISE.
Above '10' thereare the roots SATA sata '100', TUHANTE tuhat '1,000'
(note NTE > t, sG is TUHANTE-N tuhannen), and MILIOONA miljoona
'million', bILIOONA hiljoona '(US) billion', MILIARDI miljardi '(UK)
billion', etc. Teensare formed with +TOISEtA, the sP of TOCZ!:::ISE toinen
'other(oftwo)', e.g. mLIA+ TOCZ!:::IS-tA neljiitoista '14'. Decadesare noun
phrases with the partitive of '10' as second member, e.g. KAKTE
 FINNISH 169

#KUMMENE-TA kaksikymmentii '20', and hundreds and thousandsare

fonnedin similar fashion,e.g. VnTE TUHAMfE-TA viisituhatta '5,000'.
For the ordinals,seeE-stems,p. 161. Fractionsare fonnedby the addition
of =KSE to the ordinal stem, e.g. KOLMA=NTE-HSE kolmannes '(one)
third' , sG KOLMA=NTE=KSE-N kolmannekse-n.

Pronouns
Thereare personalpronounsfor all threepersons,singularand plural; sample
fonns are set out in Table 5.5. All personalpronounshavefront prosody,but
the stems for the first and secondpersonssingular (#MINA #SINA) have
back-voweloblique stems(MINU- SINU-). The plural pronounspI ME, p2
TE, p3 HE inflect in a manneridentical to that of the plural noun paradigm,
except in three cases: the nominative lacks the plural suffix -T, and the
accusative,partitive, and genitive casesuffixes are attachedto a plural stem
extendedwith -TA. The s3 pronouninflects like any E-stemof comparable
shape,exceptin the nominative,where its final E is cancelled:~ANE ~ANE hiin
'(s)he'.In colloquial speech,allegro sl and s2 fonns without the IN sequence
are thenonn,e.g. nominativesl ~A mii, mii, genitives2 SU-N sun,andthe third-
personpersonalpronounsare often replacedby the anaphoricpronounsSE
andNE, qq.v.
Apart from their use in contrastingor focalizing emphasis,the personal
pronouns are nonnally used only in non-nominative fonns, with one
exception: third-person pronouns always, and, in casual speech, axis-
of-discoursepronounsusually, are usedas verb subjects,e.g. hiin/se kuuli-0
'(s)heheard',m(in)ii kuul-i-n 'I heard',meo-n men=ty 'we went'.
The genitive fonns of the third-personpronounsare used to indicate a
possessorother than the salientactor (usually= the subject)in a sentence;if
the personal pronoun~NE ~NE is used, the noun indicating the possession
bears the third-personsuffix, e.g. hiin/se so-i-0 hiine-n voi+leivii-nsii/se-n
voi+leivii-n '(S)he1 ate his/her2sandwich',contrasthiin/se so-i-0 voi+leivii-
nsa '(S)he1atehis/her1 sandwich'.

Table5.5 SampleFinnishpersonalpronounforms: sl#M(IN)A - MINU-, p2

TE(-I(-TA)-), s3 #HANE

sl p2 s3

N 'MINA TE #HANE
mina te han
A MINU-T TE-I-TA-T #HANE-T
minut teidat hanet
p MINU-l'A TE-I-tA 'HANll-tA
minua teita hanta
Ine MINU-SSA TE-I-SSA #HANE-SSA
minussa teissa hanessa
170 FINNISH

The reflexive pronoun is ITSE(Q), oblique stem ITSE- itse(1) 'self. It

takescaseand person(but not plural) suffixes like a regularnoun, e.g. slsP
ITSE-tA-NI itseani,p2All ITSE-LLE~NNEITSE-LLE~NNEitsellenne.
Finnish has a three-way demonstrativesystem, with distal TOo tuo,
proximal ~A=MA tama,tama, and anaphoric-neutralSE se. The inflection of all
three of these words deviates from that of the noun in that their plural
paradigmsare basedon stemswith initial N, viz. NOD nuo, ~A=MA their nama,
NE ne. The proximal pronounis also deviantin that it containsthe morpheme
=MA only in forms which would otherwisebe monosyllabic,viz. sNG and
pN. The anaphoric-neutralpronoun SE exhibits the uniquely deviant vowel
alternationE -I -1 0 , with E occurringin sNG andpN, 10 in the interior local
cases,and I elsewhere.In casual speech,the anaphoric-neutralpronoun is
used insteadof the third-personpersonalpronouns.In addition, it has two
different but relatedfunctions: in casualspeechas a premodifierin the noun
phrase(e.g. se poika), it indicates a kind of definiteness,cf. Sulkala and
Karjalainen 1992: 269; in more formal style, it is an explicit indicator of
cataphora,and marks a following relative clauseas restrictive,e.g se nainen,
joka tul-i-{j hoitaa=ma-an laps-i-a PRO.NEUT WOMAN-N PRO.REL
COMES-past-s3CARES.FOR=inf3-illCHILD-plur-P 'the womanthat came
to look after the children'; both constructionshaveexactparallelsin Swedish
(with den/det;cf. Andersson1994: 288).
It is instructive to considerthe paradigmsof the demonstrativepronouns
in conjunction with the deictic adverbs,which similarly distinguish distal,
proximal, and anaphoric-neutralspheres.These adverbsdo not distinguish
interior from exteriorlocation; the distal forms areformally identicalwith the
exterior casesof the correspondingdemonstrativepronoun,but the proximal
and anaphoric-medialforms usea lative suffix -NTEQ -nne1 not known from
the noun paradigm,as well as stem-extendingmorphemes,-KA- (proximal)
and -KE- (anaphoric-medial), e.g. ~A-KA-LLA~A-KA-LLA taalla, #SE-KE-LLA
siella. The stem-extender-KA- also occurs, e.g., in the deictic adjective
ta=kii=la=inen 'local, of this place'.
The deictic adverbs (in italics) and sample forms of the demonstrative
pronounsmay be found in Table5.6.
Interrogativeand relative pronounsoverlap, in part. KU=KA - KE(NE)-
kuka 'who?' is only rarely used as a relative pronoun, but MI=KA mika
'what?; which' is commonin both roles. There is also a dedicatedrelative
pronoun,IO=KA joka. The distribution of the =KA elementin thesewords
is similar to that of =MA in the proximal pronounin that it occurs only in
forms which would otherwisebe monosyllabic,but the suffix order is the
reverse,e.g. sG MI-N=KA minkii 'of what?,of which' (contrastsG ~A=MA­the
N taman 'of this'). The plural nominativeis formed with -T, as in the noun
paradigm,e.g. KE-T=KA ketka 'who (plur)?' Interrogativedeictics are built
to MI, e.g. MI-SSA missa 'where?',MI-NTEQ minne1'whither?',MI=LL-
OIN milloin 'when?',MI-KSE miksi 'why?,
 FINNISH 171

Table5.6 Demonstrativepronounsand deictic adverbs

Anaphoric-
Proximal Deictic medial Deictic Distal Deictic
pronoun adverb pronoun adverb pronoun adverb

N ta=ma se tuo
G ta=ma-n se-n tuo-n
p ta-ta si-ta tuo-ta
Ine ta-ssa tii-ii-llii sii-na si-e-llii tuo-ssa tuo-lla
Ela tii-sta tii-ii-ltii sii-ta si-e-ltii tuo-sta tuo-sta
III ta-han tii-nne? sii-hen si-nne? tuo-hon tuo-lle?
Ade ta-lla si-lla tuo-lla
Abl ta-Ita si-Ita tuo-Ita
All ta-lle? si-lle? tuo-lle?

Indefinite and negative pronounsand deictic adverbs are built with the
enclitic --(KA)DN, which is addedto inflected interrogativepronounbases
(the longer form is usedafter consonants),e.g. sN KU=KA __DN kukaan, sP
(E-I) KE-LA--DN (ei) ketiiiin 'no one', sG (E-I) KENE-N--KADN (ei)
kenenkiiiin 'nobody's',(E-I) MI-SSA--DN (ei) missiiiin 'nowhere'.Thereare
also specifically indefinite pronouns, IO=KIN jokin 'some (one/thing)',
which inflects like MI=KA, and the compoundIO+KU joku 'some (one/
thing)', both of whose members inflect, e.g. sAde IO-LLA+KU-LLA
jollakulla 'some(one/thing)'.

Verb Inflection
Finnish verbs can be inflected for either tenseor mood, and for person.The
tense/moodsuffix attachesdirectly to the stem; in non-imperativeforms, the
tense/moodsuffix is followed by the personsuffix.
The tense/moodsuffixes are -0- (presentindicative), -1- (past indicative),
-NE- (potential),-ISI- (conditional),and,in the imperative,-Q (secondperson
singular) - -KAA(-) (first and secondpersonsplural) - -KO- (third person;
there is no sl imperativeform). The person,i.e. subject,suffixes are sl -N,
s2 -0 (imperative) - -T (elsewhere),s3 _D (presentand potential) - -0 (past
and conditional) - _DN (imperative),pI -MME, p2 -TTE, p3 -VAT (- _DT in
the imperative).There is also a subparadigmof impersonalinflection, used
when the subjectis unknown or to avoid stating the subjectexplicitly; see
below.
Both past-tense-1- and the secondI of conditional -ISI- continuea past-
tensemorphemewhich probablyevolvedfrom a verbal nounin proto-Uralic.
The IS of conditional -ISI- has an exactanaloguein proto-Saamic*-nj(d)zL
(Korhonen 1981: 251-2) and was homophonous,if not identical, with the
antecedentof the diminutive nominalizerseenin Finnish =ISE. At least the
-Q form of the imperativedatesback to proto-Uralic, as well. Potential-NE-
hascertaincognatesin Mari, Mansi, and Hungarian,andpossiblecognatesin
172 FINNISH

all of SamoyedicsaveNganasan.The personsuffixesareof eitherpronominal

origin (axis-of-discoursefonns) or continueverbal nouns(third person_D <
*-wV, rhythmic weakgradeof the non-perfectiveactiveparticiple =PA; cf. p3
-VAT < *=wV-t, with t plural). The long consonantsinitial in pI -MME and
p2 -MME are thoughtto continuebimorphemicsequences*-k-mV-, *-k-tV-
in which *-k- was a present-tense marker.
The NSKandits progeny(e.g.Tuomi 21980)distinguishforty-five patterns
of verb inflection. Most of the variation is due to stem-final morpho-
phonemicstriggeredby the past-tensesuffix -1- (cf. the effects of plural -l-
in nominal inflection, above). Examination of Table 5.7 will reveal that
similar, but not identical, rules governthe behaviourof the verbal analogues
of the nominal stem-types;for an exhaustiveand systematicoverview see
Austerlitz 1965. Specifically,besidestablestemsthereare E-stems,A-stems,
and long-vowel stems;there is also a rich and varied vein of X-stem verbs,
many but not all of which are built with a synchronicallysegmentablesuffix,
such as factitive =X or inchoative =EXE. Notice that the X of the fonner
behavesdifferently from the X of the latter, e.g. salasi '(s)hekept concealed'
but vanheni'(s)hegrew old'. Notice, also, that segmentationof suchsuffixes
often leavesthe analyst with a bound lexical morpheme,e.g. KATKE= of
KATKE=X- 'breaks'; such morphemes,which have been called themes
(Austerlitz 1976, Abondolo 1988), are a by-product of segmentationso
ruthlessas to resembleinternal reconstruction.They deservefurther scrutiny
nonetheless.
Impersonalfonns arebuilt with the discontinuousmorphemes-TA ... _DN
(presenttense) - -'tTA ... DN (all other tensesand moods); tense/mood
suffixes are sandwichedbetweenthe two components,e.g. impersonalpast
SANO='tTA-I-DN sanottiin 'it was said, peoplesaid', impersonalimperative
SANO-'tTA-KO-DN sanottakoon'let it be said', impersonalpresentSANO-
TA-0- DN sanotaan 'it is said, people say', impersonalconditional SANO-
'tTA-ISI-DN 'it would be said, people would say'. To the left of the
impersonalmorphemea stem-finalA is readase (cf. the comparative,p. 162),
e.g. impersonalpotentialANTA-'tTA-NE-DN annettaneen'it might be given,
peoplemight give'. The prehistoryof thesefonns is obscurein its details.
In casual speech,impersonal fonns usually replace first-person plural
personalfonns, e.g. memen-t-i-in = memen-i-mme'we went'.
Negativefonns are built with the negativeauxiliary E-, which is inflected
for person but not for tense/mood;the tense/moodof the constructionis
encodedon the lexical verb, with suffixation as follows: present-Q (the
so-called connegative, which is fonnally always identical with the s2
imperative), past =NU'tE (i.e. identical with the perfect active participle),
conditional -ISI-Q, potential -NE-Q. The negative auxiliary has deviant
third-personfonns, viz. s3 E-I ei, p3 E-IVAT eiviit. Examples,using the
verb EHTI- 'has enoughtime for X': E-N EHTI-Q en ehdi? 'I don't have
time', E-I EHTI-ISI-Q ei ehtisi? '(s)he wouldn't have enough time',
TableS.7 Subparadigmsof sample Finnish verbs

s1 Present s3 Past s3 Conditional s3 Imperative Perfectiveactive Past impersonal Infinitive 1

participle

SANO- SANO-N SANO-I-0 SANO-ISI-0 SANO-KO-DN SANO=NU~E

SANO=NU~E SANO=aAl.-I-DN SANO-tAQ
says sanon sanai sanais; sanokoon sanonut sanottiin sanoa?
MUISTA- MUISTA-N MUISTAI.-I-0 MUISTA-ISI-0 MUISTA-KO-DN MUISTA-NU~E
MUISTA-NU~E MUISTA-~T-Al._DN
MUISTA-~T-Al._DN MUISTA-tAQ
remembers muistan muisti muistaisi muistakoon muistanut muistettiin muistaa?
"TIETA- #TIETA-N #TIETAI.-I-0 #TIETA-ISI-0 TUNTE-KO-DN ~ETA-NU~E
~ETA-NU~E ~IETA-~T-Al._DN
~IETA-~T-Al._DN "TIETA-tAQ
knows tiediin tiesi tietiiisi tietiiiWon tietiinyt tiedettiin tietiiii?
KAIVA- KAIVA-N KAIVA-I-0 KAIVA-ISI-0 KAIVA-KO-DN KAIVA-NU~E
KAIVA-NU~E KAIVA-~TAI.-I-DN
KAIVA-~TAI.-I-DN KAIVA-tAQ
digs kaivan kaivoi kaivaisi kaivakoon kaivanut kaivettiin kaivaa?
SALLI- SALLI-N SALU-I-0 SALU-ISI-0 SALLI-KO-DN SALLI-NU~E
SALLI-NU~E SALLI-~TAI.-I-~
SALLI-~TAI.-I-~ SALLI-tAQ
permits sallin salli sallisi sallikoon sallinut sallittiin sallia?
VOI- VOI-N VOI-I-0 VOI-ISI-0 VOI-KO-DN VOI-NU~EVOI-NU~E VOI-tTAI.-I-~ VOI-tTAI.-I-~ VOI-~AQ
VOI-~AQ
is able vain vol vais; voikoon voinut voltUn voida?
roD_ JOD_N IOCl-I-0 IOCl-ISI-0 roD-KO-DN roD-NU~EroD-NU~E IOD_tTAI.-I-DN roD-~AQ
roD-~AQ
drinks juon joi joisi juokoon juonut juotiin juoda?
TUNTE- TUNTE-N TUNTll-I-0 TUNTll-ISI-0 TUNTE-KO-DN TUNTE-NU~E
TUNTE-NU~E TUNTE-aAl.-I-DN TUNTE-tAQ
senseS tunnen tuns; tuntis; tuntekoon tuntenut tunnettiin tuntea?
NOUSE- NOUSE-N NOUSE-I-0 NOUSE-ISI-0 NOUSE-KO-DN NOUSE-sU~E
NOUSE-sU~E NOUSE-~TAI.-I-~NOUSE-~TAI.-I-~ NOUSE-~AQ
NOUSE-~AQ
rises nousen nousi nousisi nouskoon noussut noustiin nousta?
TULE- TULE-N TULE-I-0 TULE-ISI-0 TULE-KO_DN TULE-1U~E
TULE-1U~E TULE-~TAI.-I-DN TULE-~TAI.-I-DN TULE-,AQ
comes tulen tuli tuUsi tulkoon tullut tultiin tulia?
~AKE-
~AKE- ~AKE-N~AKE- ~AKE-I-0~AKE-I-0 ~AKE-ISI-0 ~AKE-ISI-0 ~AKE-KO-DN
~AKE-KO-DN ~AhE-NU~E
~AhE-NU~E ~AKE-~TAI.-I-DN~AKE-~TAI.-I-DN ~AKE-,AQ
~AKE-,AQ
sees niien niiki niikisi niihkoon niihnyt niihtiin niihdii
SALA=X- SALA=X-N SALA=X-I-0 SALA=X-ISI-0 SALA=X-KO-~
SALA=X-KO-~ SALA=X-NU,E SALA=X-,TAI.-I-DN SALA=X-,AQ
keepsX secret salaan salasi salaisi salatkoon salannut salattiin salata?
KATKEX- KATKEX-N KATKEX-I-0 KATKEX-ISI-0 KATKE=X-KO-DN KA TKEX-NU,E KA TKEX-,TAI.-I-DN KATKEX-~AQ
KATKEX-~AQ
breaks(intr) katkean katkesi katkeaisi katketkoon katkennut katkettiin katketa?
VANHAI.=EXE- VANHAI.=EXE-N VANHAI.=EXE- 1-0 VANHAI.=EXE-ISI-0 VANHAI.=EXE-KO-DN VANHAI.=EnE-NU~E
VANHAI.=EnE-NU~E VANHAI.=EtE-,TAI.-I-DN VANHAI.=EtE-,AQ
ages vanhenen vanheni vanhenisi vanhetkoon vanhennut vanhettiin vanheta?
VALITSE- VALITSE-N VALITSE-I-0 VALITSE-ISI-0 VALITUKO_DN VALIlSE-NU,E VALITU,TAI.-I-DN VALITU~AQ
VALITU~AQ
chooses valitsen valitsi valitsisi valitkoon valinnut valittiin valita?
174 FINNISH

E-T EHTI=NUtE et ehtinyt 'you didn't havetime'. Prohibitives,i.e. negative

imperatives, are formed with the prohibitive auxiliary #ALA. The s2
prohibitive is formed by simple combinationof this auxiliary stem with the
connegativeof the lexical verb, e.g. # ALA UNOHTA-Q iilii unohda?'don't
(s2) forget'. The p2, s3 (and impersonal),and p3 prohibitives are built with
# AL-KAA, # AL_KO-DN, and # AI-KO-DT respectively,and combinedwith a
prohibitive connegativeverb form madewith the suffix -KO, e.g. #AL-KAA
UNOHTA-KO iil-kiiii unohta-ko 'don't (p2) forget!', # AL_KO-DN
UNOHTA-KO iil-ko-on unohta-ko 'let him/her not forget!', #AL-KO-DN
UNOHTA-tTA-KO iil-ko-on unohde-tta-ko 'may it not be forgotten, let
peoplenot forget', # AL_KO-DT UNOHTA-KO iil-ko-ot unohta-ko'let them
not forget! ' .
Compoundtensesand moodsof personalforms of the verb are constructed
with the perfectiveactive participle and the auxiliary OLE- 'is' (LIE- in the
potential),e.g.sl perfectOLE-N TULE-NUtE olentullut 'I havecome',sl past
conditional OLE-ISI-N SAA-NUtE olisin saanut 'I would have gotten'.
Exampleof negativecompoundconstructions:E-N OLE-ISI-Q EHTI=NUtE
en olisi? ehtinyt 'I wouldn'thavehadtime'. Impersonalcompoundtensesand
moodsusethe perfectivepassiveparticiple,the finite verb standingin the third
person singular, e.g. O-N #JARKE=STA-tTU, i.e. IS-s3 INTELLI-
GENCE=vsx-perf.pass.part on jiirjestetty 'one has organized,people have
organized',E-I OLE-Q MENE-flU, i.e. NEG.VERB-s3IS-CONNEGGOES-
perf.pass.partei ole[ mJmenty'onedidn't go, peopledidn't go'.

Infinitives and Participles

The non-finite forms of the verb are illustratedhere with forms of the verbs
MENE- 'goes',TAPA=X- 'meets',andTAPPA- 'kills'.
We may distinguishat leastfour infinitives:

1 First infinitive -tAQ, whosefinal Q descendsfrom a lative: MENE-tAQ

mennii?,TAPA=X-tAQ tavata?,TAPPA-'iAQ tappaa1.
2 Second infinitive -tE-, which occurs in instructive (-N) and inessive
(-SSA) forms, both with gerundivefunctions, e.g. MENE-tE-N mennen,
TAPA=X-tE-SSA tavatessa,TAPPA-'iE-SSA tappaessa.A stem-final E
that is not cancelledis read as i to the left of this suffix, e.g. TUNTE-
'iE-SSA tuntiessa.
3 Third infinitive =MA, which occurs in numerouscases,including the
nominative:MENE=MA menemii,TAPA=X=MA tapaama,TAPPA=MA
tappama.A form built with personsuffixes to the plural adessiveof a
derivate of this stem is sometimescalled the fifth infinitive, e.g. (OLE
-I-N) TAPPA=MA=ISE-I-LLA-NI (olin) tappamaisillani 'I was aboutto
kill'. The third infinitive is also used suppletively as perfectivepassive
participle in attributive constructionswith explicit agents: see Syntax,
p.l77.
 FINNISH 175

4 =MISE, which is a fully declinableverbal noun: MENE=MISE menemi-

nen,TAPA=X=MISE tapaaminen,TAPPA=MISE tappaminen.

The four participles are most succinctly presentedin terms of a matrix

which distinguishesperfectivev. non-perfectiveand active v. passive:

Perfective Non-perfective
Active =NUtE =VA (- =PA)
Passive =tTU =tTAVA

The non-perfective passiveparticipleis historically a compositeconsisting

of the non-perfectiveactive suffix =VA addedto the impersonalstem.The
alternate=PA of the non-perfectiveactive participleis a strong-graderelict of
rhythmic gradationalternations;its usein the present-daystandardlanguage
is restricted,e.g. #KAU=PA kiiypii as in kiiy=pii hinta 'going price' parallel
to #KAU=VA, as in koulu-a kiiy=vii 'attending school' and lexical-
izations such as #SUO=PA syopii 'cancer', cf. #SUO=VA syovii 'eats
nonperf.act.part'.Sampleparticipial forms: MENE-NUtE mennyt,TAPA=X-
NUtE tavannut,TAPPA-NUtE tappanut;MENE=VA menevii,TAPA=X=VA
tapaava, TAPPA=VA tappava MENE-flU menty, TAPA=X=flU tavattu,
TAPPA=tTU tapettu; MENE-flAVA mentiivii, TAPA=X=flAVA tavattava,
TAPPA=tTAVA tapettava. Note stem-final A > e to the left of =TTU and
=tTAVA, and comparethe impersonal(p. 172).
The N initial in the perfectiveactive participle assimilatesto dentalswhich
becomestem-final through the cancellationof final E, e.g. NOUSENUTE
noussut,cf. NOUSE-D nousee'rises s3', PURE-NUTE purrut, cf. PURE_D
'bites s3'.
The non-perfectivepassiveparticiple often expressesnecessityor obliga-
tion, e.g. LUKE=TTAVA KIRIA luettava kirja 'book which must/shouldbe
read'.

Syntax
The noun phraseis centredon a headnominal, which has optional modifiers
beforeit, after it, or both. Quantifiersand adjectives,including demonstratives
and other pronominaladjectives,and other noun phrasesin the genitive, are
all premodifiers. Examples:joku toinen henkilo 'some other person', tuo
mielenkiintoinenihminen 'that interestingperson'.Premodifiersmay them-
selvesbe precededby modifiers such as (1) adverbs,e.g. hyvin 'very', aivan
'quite', as in tii=mii aivan uusi kaupunki 'this quite new city', or (2)
subordinatedadjectival modifiers in the genitive, as in sellainen tumma-n
ruskea takki THAT. KIND DARK-G BROWN JACKET 'that kind of dark-
brown jacket', or (3) adverbialexpressionsgovernedby the premodifer,such
as the partitive of bestedcomparison,e.g. minu-a vanhe=mpimiesPRO.sI-P
176 FINNISH

OLD=cfv MAN 'a manolder than1'. This last type is especiallyfrequentwith

participle premodifers,e.g. ta-ssa kaupungi-ssaasu=va saveltaja 'a com-
poser living in this city'. In general,premodifiers agree with their head in
numberand case,e.g. ta-ssaaivan uude-ssakaupungi-ssaTHIS-ine QUITE
NEW-ine CITY-ine 'in this quite new city', kaikk-i-in na-i-hin kaupunke-i-hin
ALL-plur-ill THESE-plur-ill CITY-plur-ill 'to all these cities', kaik-i-ssa
Euroopa-n ma-i-ssaALL-plur-ine EUROPE-G LAND-plur-ine 'in all the
countries of Europe'; there are, however, a few invariable modifiers, most
notablyensi'next', viime(1) 'last',joka 'each(andevery)', koko 'whole', eri
'various', pikku 'little', e.g. koko kahde-n miljardi-n asukkaa-naluee-lla
WHOLE TWO-G MILLIARD-G INHABITANT-G AREA-ade 'in the region
of the entiretwo thousandmillion inhabitants'.
Postmodifiersare normally in the partitive, and refer to somelarger scope
of which the head is a part or the measure,e.g. kymmene=smaaliskuu-ta
TEN=ord MARCH-P 'the tenth of March', lasi olut-ta GLASS BEER-P 'a
glassof beer'.Numeralsover'l' and a few otherquantifiers(suchas PUOLE
puoli 'half) function as headwhen they are in the nominativeor accusative
(both marked with -O, cf. p. 157); the items countedare then postmodifiers
and so stand in the partitive, e.g. kolme-fJ lasi-a THREE GLASS-P 'three
glasses(subject or direct object)'; such constructionsmay be chained,e.g.
puoli-fJ tunti-a puhet-taHALF-N HOUR-P SPEECH-P'a half hour of (e.g.
recorded)speech'.In casesother than nominative and accusative,numerals
function as premodifiers, and accordingly agree with their head in case;
numberis usuallyformally singular,however,e.g. kahde-nerillise-n otokse-n
tapaukse-ssaTWO-G SEPARATE-G SAMPLING-G CASE-ine 'in a/the
caseof two separatesamplings'.
A traditional Europeansubject/objectapproachis unableto cometo grips
with the Finnish sentence,since not only subject/objectcase-marking,but
also (1) verbal aspect, (2) topic/focus, and (3) kinds and degrees of
definitenessof both subject and object are expressedby a combinationof
case-selection,constituentorder, subject-predicatenumber agreement,and
lexical valence. All of these factors therefore should ideally be treated
together.Thus the nominativecasenormally marks a definite subject(type 1
in Figure 5.2), and the accusativenormally marks a resultativeobject, as in
the sentenceromppu-fJsisiilta-a presidenti-npuhee-nCD-N CONTAINS-s3
PRESIDENT-GSPEECH-A 'the CD containsthe president'sspeech',with
S[ubject]-V[erb]-O[bject] order. (On resultativeobjectsand the verbs which
attract and produce them, see Karttunen 1975). An indefinite subject is
normally in the partitive, as in the sentenceslevy-ssao-n puhet-taDISK-ine
IS-s3 SPEECH-P'thereis (somee.g. recordedor transcribed)speechon the
disk', and levy-ssao-n puhe-i-ta DISK-ine IS-s3 SPEECH-plur-P'there are
(some e.g. recorded or transcribed) speecheson the disk', both with
A[dverbial]-V-S order; note the lack of number agreementin the latter,
so-calledexistential,sentence.VS orderis typical of sentenceswith indefinite
 FINNISH 177

subjects,but topicalization moves even an indefinite subjectinto preverbal

position, e.g. puhet-tao-n levy-ssii 'as for speech,there'ssomeon the disk';
contrastthe neutral S-V-A order of puhe-fJo-n levy-ssii 'the speechis on the
disk'. The constituentordersS-O-(A-)V andO-S-(A-)V occuras a by-product
of focalization, e.g. fussi-fJ tiimii-n hevose-nmarkkino-i-Ita ost-i-fJ JUSSI-N
THIS-A HORSE-A MARKET-plur-abl BUYS-past-s3 'it was Jussi who
boughtthis horseat the market', tiimii-n hevose-nfussi markkino-i-Ita ost-i-fJ
THIS-A HORSE-A JUSSI-N market-plur-abl BUYS-past-s3 'it was this
horse that Jussiboughtat the market' (Heinamaki1976).
The actor in constructionsbuilt with modal verbs is usually markedwith
the genitive (subjecttype 2 in Figure 5.2), and any resultativesingularnoun
direct object of such a constructionis marked with -O, thereby syncretizing
with the nominative,e.g. opettaja-ntiiyty-y kirjoitta-a? kirja-fJ TEACHER-G
IS.NECESSARY-s3WRITES-infl BOOK-A 'the teachermust write the/a
book'. The genitiveis also usedto mark the agentin constructionsmadewith
the third infinitive, e.g. opettaja-n kirjoitta=ma-ssa kirja-ssa o-n virhe-i-tii
TEACHER-GWRITES=inf3-ineBOOK-ine IS-s3 ERROR-plur-P'thereare
mistakesin the book written by the teacher'.
Finite-verbclausescan be linked by simple parataxis,by co-ordinatingor
subordinatingconjunctions,or by enclitics suchas --KA, e.g. Minii e-n ole-?,
e-n--kii halua-?ol-la?, ni-i-den jouko-ssaPRO.sl NEG.VERB-sl IS-conneg
NEG.VERB-sl--encliticWANTS-connegIS-infl PLURPRO.NEUT-plural-
pG GROUP-ine'I am not, nor do I want to be, amongthem', or --KO, which
marks both direct and indirect questions,e.g. kyse?e-i ole-? sii-tii, piiiise-
e--ko puol=ue? hallit=ukse-en QUESTION NEG.VERB-s3 IS-conneg
PRO.NEUT-elaMANAGES.TO.ARRIVE-s3--question.particleHALF=nsx
CONTROLS=nsx-ill 'the question is not whether the party will get into
government'.(There are also enclitics which expressthe speaker'sattitude
towardsthe utteranceor the speechsituation, or which invite the collocutor
to concur; in casual speech, enclitics often form chains of up to three
members.) Verbal nominals often express the equivalent of clauses, as
mentioned above under case-marking;one further example: hiin kiiv-i-fJ
Helsingi-ssii vuon-na 1984 pyyde-tty-ii-ni hiin-tii tule=ma-an PRO.s3
COMES.AND.GOES-past-s3 HELSINKI-ine YEAR-ess 1984 ASKS-
perf.pass.part-slPRO.s3-PCOMES=inf3-ill '(s)hecameto Helsinki (tempo-
rarily) in 1984whenI askedherlhim to (come)'.

Lexicon
Dialectal and folklore resourcesprovide Finnish with a huge range of
synonyms and affective-phonaesthemic vocabulary. Many domains of the
Finnishlexicon arecharacterizedby synonymicdifferentiationwhich appears
rich in a Europeancontext,e.g. threewordsfor 'or' (tai = Latin aut, eli = Latin
vel, plus vai, used in questions;similarly, three words for 'and'); kinship
178 FINNISH

tenns(eno 'maternaluncle' : setii 'uncle'); verbs of disappearing;and verbs

of possibility/sufficiency; on this last, see Flint 1980, which discussesa
network of forty-five such verbs in detail. Besides calquing and outright
borrowing, there is also a productive urban sourceof new lexical material,
namely slang (Waris 1973,Anttila 1975, Karttunen 1989);nurserylanguage
also produces new stems and stem architectonics(e.g. Austerlitz 1960).
Crisscrossingthese sources is the mechanismof derivation: the Finnish
lexicon bristles with words fonned with one or more derivational suffixes.
Becauseof the sometimesimprecise boundariesbetween derivation and
inflection on the one hand, and betweenderivation and compoundingon the
other, it is difficult to determinethe precisenumberof derivational suffixes
productive in Finnish. Nevertheless,about fifty-five nominal-forming suf-
fixes may be found in Vesikansa1978: 22-90; note that this number far
exceedsthe numberof nominal inflectional suffixes, which is lessthan thirty
evenif we include the comparativeand superlativeas inflectional categories.
Words fonnedwith derivationalsuffixes numberin the tensof thousands,and
the derivationalprofiles of Finnish have not yet receiveda treatmentwhich
is both systematicand exhaustive(but see Vesikansa 1978 and Cannelin
1932); herewe haveroom for only a small sample.
Prominentamongsuffixeswhich derive nominalsis =EQ, which originally
fonned nomina instrumenti, but is now a factotum nominalizer,e.g. KOKE
=EQ koe 'test', cf. KOKE- 'tests,tries; earlier: checksfish-traps'. Currently
namesof instrumentstend to be madewith =URI (- =ERI - =ORI in foreign
items), =IME and =KKEQ, e.g. PRINTT=ERI '(computer) printer', cf.
PRINTT=AX- 'prints (from a computerdocument)" NOSTA=URI nosturi
'crane', NOSTA=IME nostin 'lever', both from NOSTA- 'raises',#SAATA
=IME siiiidin 'regulator', cf. #SAATA- 'regulates',SULA=KKEQ sulake
'fuse', cf. SULA- melts'; PALVELE=IMEpalvelin '(computer)server'from
PALVELE- 'serves',cf. PALVELE=IApalvelija 'servant',with =IA (below).
Other deverbalnominals are fonned with =0 and =U, e.g. MENE=O menD
'going, course', cf. MENE- goes', LAULA=U laulu 'song', cf. LAULA-
'sings', #KASKE=U kiisky command, order', cf. #KASKE- 'commands'.
Nomina actoris are fonned with =IA, e.g. LAULA=IA laulaja 'singer',
KOKE=IA kokija 'fish-trap checker'(note stem-final E > i, as in the second
infinitive). Thereis also =TTARE, from which somedesignationsof women
are made, e.g. RUNO=TTARE runotar 'poetess',cf. RUNO runo 'poem',
KUNINKAX=TTARE kuningatar 'queen',cf. KUNINKAX kuningas'king'.
Locational nounsare madewith =MO, e.g. KORIAX=MO korjaamo 'repair
shop', cf. KORIAX- 'repairs', #KEITTA=MO keittiimo 'cookery', cf.
#KEITTA- 'cooks'. Some of the more common suffixes which fonn
denominalnominalsare =SE, =ISE, =ISA, =LLISE, and =LAISE, e.g. KOKE
=EQ=LLISE kokeellinen 'experimental',KALA=SE kalanen '(little) fish',
KALA=ISE kalainen,KALA=ISA kalaisa, both 'aboundingin fish' , LAULA
=U=LLISE laulullinen 'musical, vocal', #KASKE=U=LAISE kiiskyliiinen
 FINNISH 179

'subordinate,servant'. Abstractnounsare fonnedprimarily with =UDTE, e.g.

LAULA=U=LLIS=UDTE laulullisuus 'musicality, songfulness',VAPAilb
lLPTE vapaus'freedom',cf. VAPAD vapaa 'free'.
Verbsarefonnedfrom verbswith sometwentydifferentsuffixes,usedalone
or in combination.Theresultis an abundanceof fonnswhich expressa rangeof
aspectuallAktionsart distinctions and degreesof (in)transitivity, or better:
exocentricv. endocentricorientation (traditional tenninology in this areais
especiallyinfelicitous; seeAusterlitz 1982).For example,from the inherently
exocentric('transitive')root~AKE- 'shows' 'sees'(s3 niike-e)is fonned,by meansof
the suffix =U-, the less exocentricstem~AKE=U- Fennic, 'is visible' (s3 niiky-y);
attachingthe 'transitivizer' =TTA- to this stemproducesthe more exocentric
'shows' 'shows'(s3niiyttii-ii). To this stemthe 'frequentative'=ELE-
~AKE=U=TTA-
may be added,yielding ~AKE=U=TTA=ELE- Fennic, 'displays,shows,portrays(a
part on stage)' (s3 niiyttele-e).As is typical, there are also nominal derivates
made from each of these verb stems, e.g.~AKE=O Fennic, niiko 'sight', ~AKP:~AKP:
=U=ISF.=UUTEniikyisyys'showiness',~AKE=U=TTA=MOFennic, niiyttiimo 'the-
atre,arena',~AKE=U=TTA=ELE=IA=TTAREniiyttelijiitiir~AKE=U=TTA=ELE=IA=TTAREniiyttelijiitiir 'actress'.
Loanwordsare known to have come into Fennic, then Finnish primarily
from Gennanic,Baltic, Slavonic and Saamiclanguages.The relative chrono-
logy of theselanguagegroupsassourcesis disputed;someinvestigatorsclaim
to have found reflexes of Indo-Europeanlaryngealsin words borrowedinto
pre-Fennic.For a review of the debate,seeHelimski 1995.
Among the old loanwords which can be ascribed with certainty are
SAIRAX sairas 'ill' and RENKAX rengas 'wheel',from Gennanic*saira-z,
*xrelJga-z(cf. Englishsore, ring); HAMPAX hammas'tooth',IIJIEINA heinii
'hay', tlrfUTTARE tytiir 'daughter' (and its derivational suffix pendant
=TTARE, mentionedabove), from Baltic; and AHRAIME ahrain (dialect
atrain) '(forked fishing-spear)', RIST! risti 'cross', and PAPPI pappi
'clergyman',from (Old) Russian.Examplesof words borrowedfrom or via
Saamic are POUNU pounu 'large tussockin a bog, especiallyin Northern
Finland', SEITA seita 'striking topographicalfeature with traditional reli-
gious significancefor the Saami', and KAAMO(S+), as in kaamo(s+aika)
(AIKA aika 'time') 'sunlessperiodin midwinter in the north of Finland', this
last ultimately from Norwegianskam(tid).

Finnish Text
Taken from p. 170 of H. R. Nevanlenna,'Suomalaistenjuuret geneettisen
merkkiominaisuustutkimuksen valossa', in J. Gallen (ed.) (1984) Suomen
viiestonesihistoriallisetjuuret,Helsinki: SuomenTiedeseura,pp. 157-74.

A: text in orthography(but phenomenaassociatedwith Q are indicatedwith

1); B: morphophonemiccode; C: morpheme-by-morpheme gloss of B; D:
paraphraseof C; E: fairly free translation.
180 FINNISH

Al Suomenruotsalaisten vaestoe1ementtien
Bl SODME-N+RODTSA=LAISE JtyAKE=STO+ELEMENTTI-
-'tEN tEN
C 1 Finnish-G+Swedish=adj-pG PEOPLE=coll+ELEMENT-pG
D 1 of Finland Swedish of populationelements

vaikutuksesta suomalaisiin on
VAIKKUTTA=UKSE-STA SODMA=LAISE-I-DN OU-N
INFLUENCES=NdV-ela Finnish=adj-plur-ill IS-s3
aboutinfluence on Finns it is

vaikea juuri mitaan sanoa?

VAI.KEA IUURI MI-'tA-_DN SANA=O=t'AQ
DIFFICULT (particle) WHAT?-sP--enc WORD=vsx-inf
difficult actually anything to say

koska suomalaiset ilmeisesti ovat

KOSKA SODMA=LAISE-T ILMEISE=STI Om-VAT
BECAUSE Finn=adj-pN OBVIOUS=adv IS-p3
because the Finns obviously have

vaelluksensa eri vaiheissa vastaanottaneet

VAELTA=UKSE- ERI VAIHEQ-I-SSA VASTA-XDN+OTTA-
NSA NU'tE-T
WANDERS-NdV- VARIOUS STAGE-plur-ine AGAINST +TAKES-
pGsp3 perf.act.part.-pN
of their wanderings various in stages (they have)takenon

runsaasti germaanista ja skandinaavista

RUNSAX=STI GERMAAN=ISE-'tA IA SKANDINAA V=ISE-'tA
AMPLE=adv German=adj-sP AND Scandinav=adj
amply Germanic and Scandinavian

geenistoa. A2 Edella jo mainitsin, etta

GEENI=STO-'tA B2 ETE=LLA 10 MAINITSE-I-N ~TTA
GENE=coll-sP C2 FORE-ade ALREADY MENTIONS- THAT
past-sl
geneticmaterial D2 earlier o I mentioned that

suomalaiset eivat ole? havaittavissa

(=Bl) E-IVAT OLE-Q HAVAITSE:=tTA VAI-SSA
(=Cl) NEG.VERB- IS-conneg PERCEIVES=nonperf.
p3 pass.part-plur-ine
the Finns have--- not in perceivables
 FINNISH 181

olevissa miUirin siirtaneet

OLE=VA-I-SSA OMA-I-1A
~AARA=IN #SIIRTA-NUtE-T
IS=nonperf.act.part-plur- MEASURE=instr TRANSFERS-perf.act.part-
ine pN
in beings to degree(s) (they havenot) transferred

naihin omia piirteitaan ja

OMA-I-1A
~A-I-XDN OMA-I-1A PIIRF.=TA=EQ-I-LA-DN (= B1)
THESE-plur-ill OWN-plur-P CIRCLE=VdN=NdV-plur-P-sp3 (=Cl)
to these(latter) own their traits and

erikoisuuksian.
ERI=KO.lSF.=UDKSE-I-1A-DN
VARIOUS=adj=nsx-plur-P-sp3
their peculiarities

E1 It is difficult to say anything, actually, about the influence of Finland-

Swedishelementsof the populationon the Finns, becauseat various stages
of their wanderings the Finns have obviously taken on a great deal of
Germanicand Scandinaviangeneticmaterial.
E2 I have already mentionedearlier that the Finns have not transferredto
theselatter, to any degreewhich is perceivable,their own traits and peculiar
features.

Referencesand Further Reading

Andersson,E. (1994) 'Swedish', in E. Konig and J. van der Auwera, (eds), The
GermanicLanguages,London: Routledge,pp. 271-312.
Anttila, R. (1975) 'Affective vocabularyin Finnish: an(other) introduction',UAJb 47:
10-19.
Austerlitz, R. (1960) 'Two nascentaffectivesuffixesin Finnish?',in AmericanStudies
in Uralic Linguistics, edited by the Indiana University Committee on Uralic
Studies,IndianaUniversity PublicationsUralic andAltaic Series1, Bloomington:
IndianaUniversity, pp. 1-5.
- - (1964) Finnish ReferenceGrammar,Bell and Howell Microfilm Editions.
- - (1965) 'Zur Statistikund Morphonologieder finnischenKonjugationstypen',
in A.y. Isacenko(ed.), Beitriige zur SprachwissenschaJt, Volkskundeund Litera-
turforschung[Steinitz-Festschrift],Berlin: Akademie-Verlag, pp. 39-43.
- - (1967) The distributional identification of Finnish morphophonemes',
Language43/1: 20-33.
- - (1968) 'L'ouralien', in A. Martinet (ed.), Le langage, Encyclopediede la
Pleiadeno. 25, Paris: Gallimard, pp. 1331-87.
- - (1976) 'Stem-typesand segmentabilityin Finnish', TexasLinguistic Forum
5: 13-20. [= Papersfrom the Translatlantic Finnish Conference,ed. Robert T.
Harms and FrancesKarttunen]. Austin: Departmentof Linguistics, University of
Texas.
- - (1979) 'The morphologyand phonologyof Finnish given names', in I. Rauch
and G.P. Carr (eds), Linguistic Method: Essaysin Honor of Herbert Penzl, Janua
182 FINNISH

linguarum,SeriesMaior, 79, The Hague:Mouton, pp. 299-306.

- - - (1982) 'Finnish derivational profiles', Sprakhistoria och sprakkontakt i
Finland och Nord-Skandinavien.Studiertilliignade TryggveSkold, den 2 Novem-
ber 1982. Language History and Language Contact in Finland and Northern
Scandinavia. .. dedicatedto TryggveSkold,Acta Regia: SocietatisSkytteana:26:
Umea,pp. 1-9.
- - - (1983) 'Partitive,infinitive, passive,and genitiveplural in Finnish', UAJb 55:
81-91.
- - (1993) 'Agglutination, demography,Finnish, Turkic', in M. Sz. Bakr6-Nagy
and E. Szfj (eds), Hajdu Peter 70 eves[Festschriftfor PeterHajdu] Linguistica,
series A, Studiae et Dissertationes 15, Budapest: MTA Nyelvtudomany
intezet.
Branch, M. (1987) 'Finnish', in B. Comrie (ed.), The World's Major Languages,
London: Croom Helm and London: Oxford University Press,pp. 593-617.
Cannelin,K. (1932) Finska spraket.Grammatikoch ordbildningsliira, Helsingfors.
Collinder, B. (1960) ComparativeGrammar of the Uratic Languages,Stockholm:
Almqvist & Wiksell.
Eliot, C. N. E. (1890)A Finnish Grammar,Oxford: ClarendonPress.
Fromm, H. (1982) FinnischeGrammatik,Heidelburg:Winter.
Hakulinen, L. (1961) Structure and Developmentof the Finnish Language,Indiana
University Publications,Uralic and Altaic Series 3, Bloomington - The Hague:
IndianaUniversity andMouton.
Heinamiiki, O. (1976) 'Problems of basic word order', in N. E. Enkvist and V.
Kohonen(eds),Reportson Text Linguistics: Approachesto Word Order, Publica-
tions of the ResearchInstitute of the Abo Akademi FoundationNo.8, Abo: Abo
Akademi, pp. 95-106.
Helimski, E. (1995) 'CBepx~peBHlle
repMaHII3MbI repMaHII3MbI B IIpH6aJITII:ll:cKo-qJII:HcKHx
H Akademi,
~pyrllx qJII:HHo-yropcKllx ll:3bIKax. MCTOPIIll: IIJlO6JIeMbI B KpaTKoH
aHHoTllpoBaHHoH II KOMMeHTllpoBaHHoH 6116JIllorpaqJII:II', in VA. Dybo et al.
(eds),3THOll:3b1KOBall: II 3THOKYJIbTypHall:IICTOPIl:ll: BOCToqHOHEBPOIIbI, Moscow:
Indrik, pp. 3-37.
- - - (1996) 'Proto-Uralic gradiation [sic]: continuationand traces',C8IFU, vol.
I, pp. 17-51.
Janhunen,1.(1981a)'On the structureof Proto-Uralic', FUF 44: 23-42.
- - (1981b) 'Uralilaisenkantakielensanastosta', JSFOu77: 219-74.
- - - (1989) 'Samojedinpredestinatiivisendeklinaationalkuperiistii', JSFOu 82:
298-301.
Kangasmaa-Minn,E. (1973) 'Genetiv und Lativ, Adjektiv und Plural', Folia Fenno-
Ugrica 8 [FUF 40], Turku: Finno-Ugric Institute of the University of Turku,
pp.74-87.
Karlsson,F. (1983) Suomenkielen iiiinne- ja muotorakenne,Porvoo: Soderstrom.
- - (1992) 'Finnish', in W. Bright (ed.), International Encyclopediaof Linguis-
tics, vol. 2, New York-Oxford: Oxford University Press.
Karttunen,K. (1989) 'Stadi on hervotonpleisi', in Vesikansa1989b,pp. 148-64.
Karttunen, L. (1975) 'On the syntax of the word paljon in Finnish', C3IFU, vol. I,
pp.227-35.
Korhonen, M. (1981) Johdatus lapin kielen historiaan. Helsinki: Suomalaisen
Kirjallisuuden Seura.
Ktinnap, A. (1992) 'Die uralischen Volker nach den Angaben der sowjetischen
Volksziihlungen1959-1989',Linguistica Uralica 28: 50-54.
Laakso, J. (ed.) (1991) Uralilaiset kansat. Tietoa suomen sukukielistii ja niiden
puhujista,Porvoo-Helsinki-Juva:SOderstrom.
 FINNISH 183

NSK = Nykysuomensanakirja, ed. M. Sadeniemiet al. (1951-61), six volumes,

Porvoo-He1sinki:SOderstrom.
Sammallahti,P. (1989) 'Suomagielanomeniid ja vearbbaid sojahus. Deataleamos
sojahantiippaidovdamearkaMmit. Nomenat ja numeralat', in P. Sarmnallahti
(compiler), Sami-SuomasatnegirjilSaamelais-suomaZainen sanakirja, Ohcejohka
(Utsjoki): lorgaleaddji,Appendix B (pp. 503-4).
Sulkala, H. and Kmjalainen, M. (1992) Finnish, Descriptive Grammars Series,
London: Routledge.
Tuomi, T. (1980) SuomenkieZen kiiiinteissanakirja. ReverseDictionary of Modem
Standard Finnish, Suomen Kirjallisuuden Seura Toimituksia 274, Helsinki:
FinnishLiteratureSociety.
Vesikansa,J. (1978)lohdokset,NyksuomenoppaitaNo.2, Porvoo: Soderstrom.
- - (1989a) 'Etunimet',in Vesikansa1989b,pp. 276-95.
Vesikansa,1. (ed.) (1989b) NykysuomenSanavarat,Porvoo-He1sinki-luva:Soder-
strom.
Viitso, T.-R. (1996) 'On classifying the Finno-Ugric languages',CB/FU, vol. IV,
pp.261-6.
Waris, H. (1973) TyoZiiisyhteiskunnansyntyminenHelsingin Pitkiinsillan pohjoispuo-
Zelle, Helsinki: Weilin & Goos.
6 Mordva
Gabor Zaicz

Approximatelythirty-five per cent of Mordva speakers,i.e. roughly 350,000

people, live in the AutonomousMordva Republic in Russia. Here the two
primary ethnic and linguistic subgroupings,the Erzya and the Moksha, are
representedin about equal numbers. But Mordva speakersmake up only
aboutone-thirdof the populationof their own republic, the greaterpart of the
populationconsistingof Russians;TatarsandChuvashalsolive there.Outside
of the Mordva Republicproper,Mordvaslive in the Penza,Niznij-Novgorod,
Simbirsk, Orenburgand Saratovregions; in the Chuvash,Tatar, and Baskir
AutonomousRepublics;in the zonebetweenthe Volga andthe BelajaRivers;
in Siberia,and in CentralAsia. DiasporaMordva speakerslive scatteredfrom
Ukraine to VladivostokandSakhalin.
In terms of numbers,Mordva speakerstoday occupy third place in the
Uralic family, after Hungarianand Finnish. Owing to the sparsesettlement
patternof much of the population,however,the demographicfigures show a
diminishing tendency. Comparedwith the 1959 census,the 1989 census
statisticsshow a drop from 1,211,000to 1,073,000speakersin Russia(88.6
per cent). For the former SovietUnion as a whole, the drop is slightly greater:
from 1,285,000to 1,154,000(89.8 per cent). What is more, the percentageof
Mordvas who consideredMordva to be their native languagedroppedfrom
77.8 per cent in 1970 to 72.6 per cent in 1979, and continuedto fall, to 67.1
per cent, in 1989.Although the official view recognizedtwo distinct literary
languages,Erzya and Moksha, national census figures do not make the
distinction; it hasbeenestimatedthat the Erzya makeup some65 per cent of
the ethnic total.
From the standpointsof linguistics, ethnography,and physical anthro-
pology, Mordva speakersmay be divided into two groups: in Mordvinia
proper, those living around the easternbranch of the Sura Riverare called
Erzya, while a westerngroup living in the valley of the Moksha River are
called Moksha(two otherethnic groups,the KaratayandTerjuxan,no longer
useMordva). The two main dialectgroups,Erzya and Moksha,differ chiefly
in their phonology;subdialectsof both Erzyaand Mokshaalso differ chiefly
in their phonology,but also, to a lesserdegree,in their morphology(for Erzya
dialects see Ermuskin 1984: 6-27; for Moksha: Feoktistov 1966: 219-20,
1975: 258-60; for an overview and classificationof subdialects:Keresztes
1990: 14-18).

184
 MORDVA 185

The following introductionto the grammarandlexicon of Mordva is based

on the Erzya literary standard,but with reference,where pertinent, to the
Moksha literary standard and to non-literary dialects as well. Unless
otherwisespecified,therefore,Erzya forms will not be marked; the abbrevi-
ationsE[rzya] andM[oksha] signalforms cited for purposesof comparison.
The ethnonymmordvin, mordva, like E mirae, M miraii 'man, husband'
(Redei 1986:53) is presumably a loan from proto-Indo-Iranian *mi'hi-
'human',borrowedinto proto-Mordvin as *murta (> *murda> *morda). The
v of modem RussianMOP,[lBHH, collective MOp,[lBa, is secondary;compare
Moskva < Old RussianMoskY. The self-designationmay have been*morda
in commonMordva, but this term was oustedby the tribal namesE efia and
Mmoksa.

Phonology

Consonants
Literary standardErzya has twenty-eightconsonantphonemes(Redei 1984:
209). This figure includes the marginalf and x, which occur only in recent
(mostly Russian)loans: cf. E fabrika, dialect kvabrika 'factory', E kolxoz,
dialect kolkoz 'kolkhoz'. NorthernErzya dialectshave,in addition, the velar
nasal.{] where the literary standardhas innovateda v, e.g. E kov, dialect ko.{]
'moon' «pU *kul)e).
The consonantinventoryof literary standardErzyais setout in Table6.1.
Allophonic variation is the result of environment,as follows: (1) palatal-
ization of bilabials, labiodentals,and velars to the left of front vowels; (2) n
is realized as velar [I)] the left of the velar stopsk and g; (3) v is often [w]
beforeconsonantsand at word-end;and (4) the sibilants s, s, and si are often
realizedas the correspondingvoicedaffricateswhento the right of n, nl, rand
rio Examples:pe [pje] 'end',venc[vjenc] 'boat';kenks[kjel)ks] 'door',pango
[pal)go] 'mushroom';cuvto [cuwto] 'tree', kev [kjew] 'stone'; kumanza-

Table6.1 Consonantinventoryof literary standardErzya

Bilabial Labiodental Dental Alveolar Palatal Velar

Nasals m n ni
Voicelessstops p t tj k
Voiced stops b d di g
Voicelessfricatives f s S sj x
Voiced fricatives vz C Z zj
Affricates c C cj
Laterals 1 lj
Tremulants r rj
Glide C
186 MORDVA

[kuman¢a]'knee',erzia-[erj¢ja] 'ErzyaMordva'.
Within the morpheme,length is non-distinctive.In intervocalic position,
the voicelessstopsand fricatives show a slightly greaterduration, e.g. the ti
of vetie 'five' is phonetically half-long [veije]. Phonologically long con-
sonantismsoccur only at morphemeboundaries,e.g. vetJi-tie [vetjtje] 'water
sAb!,.
The present-day Erzyaconsonantsystemdiffers from that of proto-Mordva
chiefly by virtue of the new phonemesf and x, which have takenroot under
the influenceof Russian.Present-dayMoksha,on the otherhand,showsmore
significant innovations: the literary languagehas five additional phonemes,
the voicelesslaterals[i i j ], the voicelesstremulants[R Rj], and the voiceless
palatal fricative [~],c which we shall write L U R Ri J. What is more,
palatalizedsj zj and cj are phonemesin someMoksha dialects.All eight of
theseadditionalphonemesmay be presentin Erzya dialectswhich havebeen
subjectedto strongMokshainfluence.
CorrelationsofPalatalizationand Voice
In literary Erzya, the correlation of palatalizationextendsto eight pairs of
consonants:t :ti, d :tJi, n :ni, s :si, z : zi, c : d, I : Ii, r: ri. The correlation
of voice also entails eight pairs: t :d, ti : tJi, p : b, k :g, f: v, s :z, s : Z, si
: z!.
ConsonantDistribution
In word-initial position in core (= non-borrowed,non-affective)vocabulary
normally only the unvoicedmembersof the [+/-] voiced pairs occur, i.e. b d
tJi g z Z zi occur word-initially primarily in loanwordsand in onomatopoeic/
affective items. Relatively rare in this position are the newerphonemes f and
x and the voicelessaffricatesc and d. Clustersof two or threeconsonants,in
which the first consonantis a stop or sibilant, are not uncommon,e.g. pria
'head',skal 'cow', ksnli 'iron', sna- 'praises',stapo(dialect: strapo) 'naked;
only'.
All consonantphonemesoccur in word-internal position. Here a great
variety of clusters occur, consisting of two to five consonants.Both the
architectonicsandthe morphophonemics of suchclustersarehighly complex;
they await a thorough and systematicstudy. A pilot study of twenty-three
three-memberclusters (R6dei 1984: 214-18) revealed that they always
contain at least one liquid or sibilant. Four-memberclustersusually begin
with a liquid or nasal, e.g. kariske 'shoelace',vanlsitinle- 'defends(freq)'.
Clustersof five are much rarer; examplesare karks-t-neBELT-plur-def 'the
belts'.In Moksha,clustersmay containas many as six consonants.
Word-finally, p b g f x c are extremely rare, occurring only in the most
recent loans. Traditionally, a paragogica has been addedto Russianloans
ending in b, g, or in a cluster alien to Mordva, e.g. stolba 'pillar', luga
'meadow',tieatra 'theatre'« Russian).Often in bisyllabic and polysyllabic
words the literary languagehas a voiced consonantwhere dialectshave the
 MORDVA 187

voicelessanalogue,e.g. tarvaz 'sickle', riivezj 'fox', tarad 'bough' (dialect:

riivez tarat). Word-final clusters are also quite common: R6dei
tarvas, riive~,
(1984: 218-21) found twelve types oftwo-memberedclustersand nine types
of three-memberedclusters. Four-memberedclusters occur only through
concatenationof morphemes.Examples: kom~kom~ 'twenty',riivez kanj~tj kom~ 'hemp',
alks-t 'beds'.
The stopsareunaspirated.Palatalizationis moreprominentin Mokshathan
in Erzya.

Vowels
The literary languageandmostdialectshavefive vowels:

Front Central Back

High u u
Mid e o
Low a

There are no diphthongs.The back roundedvowels u and 0 show little

allophonic variation, but the front unroundedi and e have backedvariants
when to the right of distinctively non-palatalizedconsonants(t d n s z c I r),
e.g. sinj [slnj ] 'they',[senj]
se~ [senj] 'blue'; contrast~i~tJie-[senj]'breaksX' (with [i]),
~e~ks 'heron''heron' (with [e]). The phonemea has a fronted allophone[a] when
to the left of any palatalized consonant,e.g. a eriavi [ajerjavji] 'it isn't
necessary/allowed' .
In Erzya dialectswhich havebeenunder Mokshainfluencetherecan be a
phonemicii and a phonemicreducedvowel ~, as well. The Moksha literary
languagehasa rectangularsystemwith schwa(i e ii u 0 a ~).
We may reckonwith a similar system,enrichedby a frontlback schwapair
*:!!/*~, in in the first syllable of the later stagesof proto-Mordva (cf. Bereczki
1988: 319,321),andin the earlierstagesofproto-Mordvatherewas certainly
a high front rounded*ti, as well. A high vowel (*i *u *ti) in the first syllable
followed by a low vowel (*a *a) in the secondsyllabletriggereda stress-shift
(from first to secondsyllable),followed by reduction(andusually loss) of the
first-syllable vowel, e.g. (stressedvowels underlined)*usk~ *joz:!! > *~sk~
*joz:!! > EM
skal 'cow', *ktirsj! > *k:!!rzj! > EM ksi 'bread'.In Erzya, the reducedvowels
of non-first syllableshavebeenreplacedby full vowels, e.g. *juz~ *joz:!!> *joz:!! >
joto 'skin', *kilm:!! > *kelm:!! > *keljme 'cold'. In Moksha,the reducedvowel
developedfrom any full vowel in unstressed position.

Vowel Correlationsand Distribution

Erzya has the smallestvowel inventory in the Uralic family. Even so, many
oppositionsarenot heavily exploited.Of the high vowels, forexample,u does
not occurin word-final position in core vocabularyand i occursthereonly in
suffixes. Both high vowels are in fact frequentonly in the first syllable. It is
188 MORDVA

thus not surprising that it is the vowels e 0 a, which are free of such
distributionalrestrictions,which arethe most common(Veenker1981: 46).
Vowel sequencesoccuronly in loans,e.g. njeusto'really?' « Russian)or,
occasionally,as the result of loss of an intervening consonant,e.g. rauzo
'black' < *rawuzo.
Furtherexamplesillustrating vowel distribution andcanonicshapes:azoro
'lord','raspberry',
i~zjej 'raspberry',tjePe 'winter', kuvaka 'long', moP-i '(s)he goes',
srazu 'at once'« Russian).

Morphophonology

Morpheme Architectonics
Perusalof Erzyatextshasuncoveredeighteenshapesfor nominal,thirteenfor
verbal stems;with suffixes the patternis the reverse,viz. elevenverb-suffix
shapesbut only five shapesfor nominal suffixes; the distribution is roughly
identicalin Moksha(Keresztes1990: 30-1). According to eygankin(cited in
Raun 1988: 100), the mosttypical Erzyamorphemeshapesare as follows: V,
ve, vee, ev, eve.As an illustration, here are examplesof (1) the vowel-
initial nominal stemtypesand (2) the consonant-initialverbal suffix types:

1 ve ej 'ice', vev aso 'white', veve ozjaz 'sparrow',vevev uzjerie

'hatchet',veev ekse 'cool', veeve iritPes 'rib', veeveeejkaks
'child', veeeveukStor'maple';
2 e (pI-k (s2 imperative),ev -do (p2 imperative),ee-zt (p3 optative),eve
-~ek (pI(pI first past),eev=vto- (transitivizer),evev=zjeve- (inchoativ-
izer), evee-sink (p2 subject,sp3 object), eeev=kSno- (frequentativ-
izer).

Mordva is rich in morphophonologicalalternations.What follows is no

more than a sketchbasedon the availableliterature (eygankin1980: 31-43,
Redei 1984: 225-9, Keresztes1990: 31-7).
Stem Variants
All verbs and nouns have both consonant-finaland vowel-final variants,to
which suffixes attach according to various rules (Nadjkin 1981). Which
variantis selecteddependsin part on the stemitself (i.e. is lexical) andin part
on the suffix to be attached.Thus the third-personsingular first past tense
suffix -~is is always addedto the vowel-stemof verbs endingin a, e.g.jarsa-
'eats',jarsa-~'(s)he'(s)he ate'; but it is addedto the vowel stemof udo- 'sleeps',
and to the consonant-stem of kado- 'leaves(tr)', giving udo-~ udo-~ '(s)he slept',
but kad-~ kad-~ '(s)heleft'. Somewhatparallelcircumstancesexist in the morphol-
ogy of the nominal, where certain suffixes tend to require consonantstems,
and somestemseven lose their last syllable before certain morphemes,e.g.
pando 'hill', inessivepand-so, illative pand-s, plural pand-t; piljge 'leg',
 MORDVA 189

inessive pizig-se, illative pUig-s, plural piIig-tj ; siezime 'eye', inessive

siezim-se,illative siezim-s,plural sieIi(m)-ti , note also sieIi+vedj 'tear (EYE
+WATER)'.
Certainsuffixes always attachto the consonantstem,e.g. the third person
singular present -i, e.g. jarsa- 'eats',jars-i '(s)he eats', udo-, ud-i '(s)he
sleeps',kado-, kad-i '(s)heleaves'.Others always attachto the vowel stem,
e.g. genitive -ni added to moda 'earth', or siezime 'eye' gives moda-nj,
j
siezime-n. Attaching suchsuffixes to consonant-finalnounssuchas os 'city',
vi"; 'forest' triggersepenthesis:os-onj , vi";-eni .

Suffix Variants
The most salienttype of suffixal allomorphyis determinedby the frontlback
prosodyof the stem; stemprosody,in tum, dependson the presence/absence
of palatalization of stem-final consonantsand the identity and order of
vowels. For example,the vowel of the inessivesuffix is e in pice-se'(Scotch)
pine (ine)', but 0 in moda-so'earth(ine)'; the (indefinite) plural suffix is non-
palatalized-t in os-t 'cities' andpaksia-t 'fields', but palatalized-tj in vi";-t j
'forests' and velje-tj 'villages'.As theseexamplesshow, suffixal allomorphy
may entail frontlback vowels (usually elo), plain v. palatalizedconsonants
(e.g. tlt j ), or both, as in the ablativeforms paksia-do,velje-dje.There is also
automaticalternationof voice in casessuchas ablativeos-to 'from a city'; in
addition to suchvoiced/voicelessallomorphy,the prolative suffix (-ka - -ga)
has a variant -va which occursafter vowel stems:os-ka 'by way of the city',
kedj-ga 'by the hand',moda-va'along the ground'.

Other Sandhi Phenomena

Word-internally there is voice assimilation in which voicelessnessmoves
from right to left; this type of voice assimilationdevoicesvoiced stops and
sibilantswhen the morphologyplacesvoicelessconsonantsto their right, e.g.
kedj 'hand' with voiced [dj ), but keczi-sewith voiceless[t j ) ([ketjse)). Further
exampleswith 'hand': illative keczi-s [ketjs), plural keczi-tj [ketjtj ), and the
derivativekeczi=ks [ketjks) 'bracelet'.In parallelfashion,from the root calga-
'treads on (tr)' is formed a frequentativecalg=sie- [calksje). The official
orthographyis somewhatinconsistentin rendering these assimilations;for
example,'bracelet'is written Ke~bKCKe~bKC(not indicating the assimilation),while
'tramples on' is written qaAKCe-, with the letter K indicating the voice
assimilation.In nominal stemsendingin n or nj thesenasalsmay assimilate
to the plural suffix, e.g. lomanj 'humanbeing', plurallomatj-tj (-lomanj-tj ).
There is also voice assimilationin which voice movesfrom left to right;
in this type of assimilation,a non-distinctivelyvoicedconsonantismat the end
of a morphemecausesthe initial consonantismof the next morphemeto be
voiced. This type of voice assimilationoperatesnot only within the word but
also acrossword boundaries,i.e. in the compoundor phrase,e.g. tolbanidja
'campfire' (compound of tol 'fire' and panjczia 'bed'), ejzjuro 'icicle'
190 MORDVA

(compoundof ej 'ice' and flura 'horn'), erzia=ni keli [erzial]igieli ] 'Erzya

language'.Word-internal examples: niizi-tiano (with [lidiD 'we swallow',
cav-tan (with [wdD 'I strike you'. The orthographyrenders this type of
assimilationonly in compounds,i.e. not in phrasesor within inflectedforms.
A few more exampleswill illustrate and expandupon the rules outlined
above. We have voice assimilation (tF > ti ) in abessiveketF-tieme [titi]
'without hand(s)';assimilationof voice and palatalization(tF > t) in inessive
kedi-se ([tt S], phonemically /kecce/; more careful pronunciationswith [tit is]
also exist); and automaticreplacementof voiced with voicelessin the middle
of a cluster (*ndl > ntl) in frequentativekunt=zie- 'keeps grasping', from
kunda- 'seizes'.

Vowel and ConsonantHarmony

The p(F)U harmonyof vowels within the stemhasbeenfairly well preserved
in Mordva: roughly 80 per cent of the inherited word stock of Mordva has
either front or back vocalism, preservingthe original frontnessor backness
inheritedfrom p(F)U; in the remainder,front and back vowels occur mixed,
usually becauseof secondaryeffects causedby consonants,e.g. uflke 'wire'
with front e due to the fl. Borrowedand onomatopoeicvocabularynaturally
show a higher proportion of mixed-vowel words, e.g. delkovoj 'rouble',
poderidie- 'drips'.
Vowel harmonyin suffixes is a different matter.Whereasin Mokshamost
suffixes have only one harmonic shape,in Erzya over twenty have front v.
back variants (Zaicz 1993). The vocalism of these suffixes adjusts to the
prosodyof the last stemvowel; the most commonpair of alternantsis 0 - e,
e.g. kudo-so 'in a house',but pirie-se 'in a garden',tol-oz-onzo'into hislher
fire' but kezi-ez-enze'into hislherlanguage'.
Alongside this kind of rudimentaryvowel harmony Erzya has evolved a
kind of consonantharmony. Suffixes with initial apical consonants(but not
s) show an alternationplain - palatalizedaccordingto the prosodyof the stem
to which they are attached,e.g. kal 'fish', dative kal-nenl and pluralkal-t, but
kazi 'willow', dative kazi-nlenl and plural kazi-ti. The vocalismof piks 'rope'
is front, but the distinctively non-palatalizeds final in this stem selectsthe
non-palatalizedalternants,dative piks-nenl and plural piks-t; contrast viri
'forest' with viri-nieni , viri_ti.

Stress
Erzya stressis free and non-distinctivewith regardto grammarand lexicon.
This meansthat any syllable of a word suchas varakanltienl 'raven(dat)' or
flimemazo 'drinking (s3)' may be given accentualprominenceindifferently,
and that the unstressedsyllables are pronouncedfully, with unreduced
vowels. In general,compoundwords are stressedon the first syllable of their
components,e.g. kiliej+by.lo 'birch grove'; first-syllable stress in non-
compoundwords is also quite common. Moksha deviates sharply in this
 MORDVA 191

regard,placing a strong stresson the first full vowel of the word (Paasonen
1903: 114-19).
In recent Russianloanwords, stressfollows the Russianpattern in both
ErzyaandMoksha.

Morphology

Nominals

TypesofNominal, andNominal Stem-types

We may treat as nominalsall stemswhich take declensionalsuffixes: nouns,
adjectives,numerals,and pronouns.The citation form for Mordva nominals
is the nominativesingular,e.g. moda 'earth',viri 'forest',sirie 'old', od 'new;
young', vejke 'one', komsi 'twenty', sie 'that, it', mon '1'. Nominals whose
citation form endsin a consonanttake an epentheticvowel, e or 0 depending
on prosody, before certain suffixes, e.g. the genitive -nj : viri-eni, od-oni,
komsi-enj, moni-si-enj (mon-si 'I myself).
Both Erzya and Moksha distinguishindefinite from definite forms of the
nominal. The developmentof the definite declensionwas alreadywell under-
way in commonMordva, andcomplementedthe older, inheriteddichotomyof
plain v. possessive declension;examplesof this three-waydistinctionareviri -se
'in a forest', viri-senjtj 'in the forest', viri-seni 'in my forest'.
Noun Declension
Indefinite and definite paradigms. With certain syncretismsnoted below,
Erzya declension distinguishestwo numbers, singular and plural. It has
becomestandardpracticeto list elevenor twelve casesfor the Erzyaindefinite
paradigm, and ten or eleven casesfor the definite (Wiedemann1865: 42,
Bubrix 1947: 17, Collinder 1957: 232, Cygankin 1980: 220, Keresztes1990:
56). The discrepanciesin case-countsare due to the inclusion or exclusionof
the lative casein the indefinite, and of the translativecasein the definite
paradigm.
In the I [ndefinitel D[eclensionl, the opposition singular : plural is
neutralizedin all casesexceptthe nominative,e.g. sN kudo 'house', pN kudo-t
'houses',spEla kudo-sto 'out of a house- out of houses'.The ID pN suffix
-t continuesthe pU plural suffix *-t. In the D[efinitel D[eclensionl,the illative
and dative/allative casessyncretize,e.g. ID dative/allative viri-nieni 'to a
forest', illative viri-s 'into a forest',but DD viri-enjtjeni '(in)to the forest'.
The forms of the DD are distinguishedfrom thoseof the ID by -si in the
nominativesingular,-tj in the genitive/accusative, and -nitj in the other cases
of the singular; in the plural the definitenessmarker is -njel-ne. These
morphemesall derive from demonstrativepronouns(cf. Finnish se 'it, that',
ne 'those',tii- 'this'); the -ni- in the sequence-nitj- probablycontinuesthe pU
genitive *-n. Synchronically,it is no simple matterto derive the DD from an
192 MORDVA

agglutinatinganaloguemade up of thesedefinitenessmarkersand the case

suffixes of the ID, andno suchsegmentationis essayedhere.
Besides the forms given in Table 6.2 for singular and plural inessive,
elative, illative, prolative, and ablative (and singularcomparative)thereexist
parallel constructionsbuilt with the noun in the definite genitive plus the
postpositionez(- ej befores) in the appropriatecase,e.g. DD sProlviri-ganiti
or viri-eniti ez-ga 'throughthe forest'.
Taxonomyand historical backgroundo/the casesuffixes.The nominative,
genitive/accusative,and dative/allative are chiefly grammaticalin function
and thus differ clearly in function from the local cases.The nominative
markeris zero, as generallyin Uralic. The suffix -ni representsa merging of
two originally distinct cases,pU genitive *-n and pU accusative*-m. In the
definite declensionin front-prosodicenvironmentsthere occurredpalataliza-
tion (*-n-t > *-nLt j ) and assimilation(*-m-t > *-m-t j > *-nLt j ); theseforms
then generalizedto non-front environments,and the -ni- elementinvadedthe
indefinite paradigmas well. The -nienj suffix of the dative/allative(-nen; after
distinctively non-palatalizedconsonants,e.g. panar-neni 'shirt [dat/all]') is
the resultof a reduplicationof an old lative in *-n j .
The local cases may be classified according to their concrete spatial
referenceinto two subclasses,interior (inessive, elative, and illative) and
exterior (ablative, lative, and prolative). Interior casesrefer to more intimate
contactthanexteriorcases.
Historically, the interior casesareall built with a putativelylative suffix *-s
known also from Cheremis,Fennic, and Saarnic.The Mordva inessive-se -
-so continuesa suffix chainreconstructedas *-s-nA, cf. the fossilizedinessive
adverb on-sne 'in a dream', and n-allomorphssuch as vaz-ne 'calf (ine)',
attestedin Erzya dialects.The elative -ste - -sto consistsof the same*-s plus
a separativesuffix *-tV which datesbackto pU. The illative -s is the *-s suffix
alone; contrast its intervocalic -z- allomorph (kudo-z-onzo 'into hislher
house')with the -s- of the inessive« *-ss- < *-sn-) in the samephonotaxis
(kudo-s-onzo'in hislherhouse').
Turning to the exterior cases, we find the same pU *-tV separative
occurring alone in the ablative -do (with allomorphs -die -de -to -tje -te in
keeping with vowel and consonantharmony), e.g. verigiz-de 'wolf (abl), ,
os-to 'city (abl), , vent-tje 'boat (abl)" piks-te 'rope (abl)'. The prolative,
which is -ga to the right of voiced consonants,-ka to the left of voiceless
consonants,and -va to the right of vowels, continues a common Mordva
suffix *-kV; this suffix no doubt containspU lative *-k. The somewhatrare
lative suffix -v, which has Erzya dialect reflexes -1] (in back-prosodic
contexts)and -j (in front-prosodiccontexts)and is representedin Mokshaby
-u - -i - -v, continuesa nasalvariantof the same*-k lative.
The translative, abessive, and comparative may be seen as semi-
grammatical,i.e. half way betweenthe local casesand the pure grammatical
casesnominative,genitive/accusative, anddative/allative.The translativemay
Table6.2 SelectedErzya nominal paradigms: moda 'earth', v;"; 'forest'

Indefinite Definite singular Definite plural

sN moda virj moda-sj virLesj

pN moda-t virLt j modati-nje virhLnie
G/A moda-nj virLenj moda-njtj virLenitj modatLnieni virjti-nienj
Dat/All moda-nieni virj-nienj moda-njtjeni virLenitjeni modati-njenieni virhL nienjeni
Ine moda-so virLse moda-sonitj virLsenitj modati-niese virjti-niese
Ela moda-sto virLste moda-stonjti virLstenitj modati-nieste virjt L nieste
III moda-s virLs moda-njtjeni virLenitjeni modati-nies virjtLnies
Prol moda-va virLga moda-vanjtj virLganiti modati-nieva virjtLnieva
Abl moda-do virLdle moda-doniti virJ-dlenJtJ modati-niecPe virjtLniedje
Lat moda-v virLev
Trans moda-ks virLks (moda-ksoniti) (virLksenitj ) (modatLnieks) (virhLnieks)
Abe moda-vtomo virLtjeme moda-vtomonjti virLtiemenitj modati-nievtjeme virjti-nievtjeme
Cfv moda-ska virLska j
moda-skanit virLskaniti modati-njeska virhLnieska
194 MORDVA

well be simply a concatenationof two of the old lative suffixes already

mentioned,viz. *-k and *-s; or its history may be more complex, involving
a connectionwith a pU noun-formingderivationalsuffix *=ks. Cognatesof
the Mordva abessivesuffix are usually seenas adjective-formingderivational
suffixes (compare Finnish piiii=ftomii-n with Mordva pe-vtieme-n i , both
genitivesof HEAD+ABESSIVE); here againwe most probablyhaveto deal
with an old suffix-string, namely *=ptV plus *=mV. The comparativeseems
to be a string of two diminutive derivationalsuffixes, *=s and *=kA.
In Mokshathere is a thirteenth('causative')case,built with -nkse,which
hasits origin in the inessiveof a postpositioninksemeaning'on accountof.
Secondary and tertiary declension. Metadeclensionalforms may be
obtainedby attachingdefinitenessstrings such as sN -si, sGA -niti , sDAll
-nltieni to most already-inflectedcaseforms. Thus to the indefinite inessive
form of moda 'earth, ground',moda-so,one can add the definite sN ending
-si, yielding modasosi 'that which is in the ground', the definite inessive
ending -sonlti , yielding modasosonit i 'in that which is in the ground',and so
on. Each of the elevendefinite casestrings can be addedto forms inflected
for at least six of the cases(genitive, inessive,elative, translative,abessive,
comparative).Such metaformsare built to genitive basesby the addition of
a reduplicative: sN -siesi, sG -sienlti , sDAll -sienltienl, and so on, e.g.
secondarylative form of the genitive vi~-eni-siesitienlti -sienltienl, 'out of that of the
forest'.
Parallel to secondaryforms such as the secondaryinessive moda-so-si
cited abovethere exist tertiary, essentiallysynonymous,forms in which the
-siesi (-sienlti , -sienltieni , etc.) endings are added to an inflected form
augmentedby -nl-, e.g. sN moda-so-nl-siesi,sG moda-so-nl-sienit i (Erd6di
1968: 233).
Possessive declension.Alongside,andin complementarydistribution with,
the definite declensionErzyahasforms which indicatethe personandnumber
of the possessorand,in the first andthird person singular nominative,also the
numberof the thing possessed.
The literature reckonswith eight to nine distinct caseforms for the Erzya
possessivedeclension(Wiedemann1865: 52, Cygankin1980: 201, Keresztes
1990: 196).The possessive suffix alwaysfollows the casesuffix in Erzya,e.g.
window',FOREST-ela-pl'out of our forest(s)'.In Moksha,wheredistinct
vi~-ste-nlek
possessiveforms for genitive/accusativeand dative/allative haveevolved, as
well, the possessivesuffix precedesthe markers of those cases,and the
numberof the possessioncanbe distinguished,e.g. M vatimii-zieWINDOW-
slsgposs 'my window', vatimii-nle WINDOW-slplurposs 'my windows',
vatimii-zie-ni WINDOW-slsgposs-gen/acc'of my window', valimii-nle-nl
WINDOW-slplurposs-gen/acc'of my windows', valimii-zi_tii WINDOW-
slsgposs-datlall'to my window', valimii-nle-nitPi WINDOW-slplurposs-datl
all 'to my windows'. In casesbeyondthe dative/allativeevenMokshacannot
distinguishpossession number.
 MORDVA 195

Table6.3 SelectedErzyapossessive
declensionsubparadigms,
nominative:
moda 'earth,ground',vi"; 'forest'

Possessor Possessionin Possessionin

Singular Plural Singular Plural

sl moda-m moda-n virLem virLem

s2 moda-t virLe~
virLem
s3 moda-zo moda-nzo vi~-eze vi~-eze vi~-enzevi~-enze
pI moda-nok virLeniek
p2 moda-nk virLenk
p3 moda-st virLest

Table6.4 SelectedErzyapossessivedeclensionsubparadigms,
s3, pI: moda
'earth,ground',vi"; 'forest'

Third personsingular First personplural

Singularpossession Plural possession

sN moda-zo spN virLdie-niek

pN moda-nzo
Ine moda-so-nzo virLse-njek
Ela moda-sto-nzo virLste-njek
III moda-z-onzo virLezje-niek
Prol moda-va-nzo virLga-nok
Abl moda-do-nzo virLdie-niek
Trans (moda-ks-onzo) (virLks-eniek)
Abe moda-vtomo-nzo virL~eme-niek virL~eme-niek
Cfv moda-ska-nzo virLska-nok

The Erzya possessivesuffixes all go back ultimately to the pU personal

pronouns,with roots *mV, *tV, and *sV for first, second,and third persons.
Plurality of thing possessed,where expressed(e.g. kudo-n 'my houses'),is
expressedby meansof a reflex of a p(F)U pluralizer *-n-. The pI and p2
suffixes (-nok - -~ek,-nk) -nk) contain a reflex of anotherp(F)U pluralizer *-k;
the -nk of the secondpersonplural may have spreadanalogically from a
specificallyplural-possession form *-ntVk (Bereczki1988: 325).

Adjectives:Declension,Comparison,and Superlative
The morphologyof adjectivaldeclensionis identicalwith that of the noun,but
adjectivesare ordinarily inflected only when they function syntactically as
nouns,e.g. kona kudosoniti eriat, odsonjtj iIii tastosonjtj 'in which housedo
you live, in the old (one) or in the new (one)?'. Adjectives in predicate
position are inflected for numberto agreewith nouns,e.g. cuvtosipoks 'the
tree is big', cuvtnepokSt'the treesare big'. In attributive position adjectives
196 MORDVA

are not nonnallyinflectedeitherfor caseor for number,e.g.paro moda 'good

ground', genitive paro moda-ni 'of good ground(s)',plural nominative od
kudo-t 'new houses',inessiveod kudo-so 'in a new house,in new houses'.
The only exceptionis the optional, and rare, plural nominativeagreementof
attributes with their animate heads, e.g. par-t lomani-tj GOOD-plur
HUMAN.BEING-plur 'goodpeople'.
Comparisonis achievednot by meansof a comparativesuffix, but simply
by placing the bestedmemberof the comparisoninto the ablative,e.g. dora-
do pokSBOY-abl BIG 'bigger than aboy'. The thing than which something
is 'Xer' must be specified, or at least referred to with the ablative of the
demonstrativepronon ~e: repetition
~e-dje pokSIT-abl BIG 'bigger'. Various super-
latives may be fonned, by (1) repetition of the root, with the first instancein
the ablative and usually with the enclitic -jak, e.g. poks-to(-jak) pokS
'biggest';(2) more commonly,by preposingone of the particles~ex(-tje)poks «
Russian 'of all'), samajRussian« Russian 'the very'), pek « Tatar 'very') or
ve~eme-tPe 'than/of
'than/of
'than/of all', e.g. pokSpokS'biggest',ve~emedje
~extje poks 'biggest
pokspoks
of all'. Dialects also use the particle enj (cf. standardErzya inie 'big' < pU
*ena 'big, much'.
A kind of excessivesuperlativeis fonnedby threefoldrepetition,e.g.poks,
poksdi poks(di 'and' < Russian)'the very biggest'.

DeclensionofNumeralsand Pronouns
Numeralsusedattributively remainuninflected,and their headis usually put
in the plural, especiallyif the sole or last numeralis from 2 to 10, e.g. kavto
Ci-tj 'two days', kolonigemen j niilje kal-t 'thirty-four fish', kavksonigemeni
vejkseumari-t 'eighty-nineapples',~adoije-t ije-tj 'a hundredyears'.The head
j
standsin the singular, however, after numeralsending in ve(jke) '1', e.g.
kodgemen j vejke celkovoj 'sixty-one roubles', and after the compound
numeralswhich constitutethe teens,twenties,and (optionally) the hundreds,
e.g. kom~+vetje=je kom~+vetje=jekudoTWENTY+FIVE=sx HOUSE 'twenty-fivehouses',
~i~em+~ad-t~i~em+~ad-t Ci SEVEN+HUNDRED-plurDAY '700 days'.
When used as nouns, the numeralsdecline in the sameway as nouns,in
both indefinite and definite paradigms,e.g. inessivevejke-se'in one', DDsN
j
vejke-~senit 'the one', DDsine vejke-se-nit 'in the one', DDpine kavto-tj-nie-
j
senit 'in the two'.
Ordinals, fonned with the suffix =de, are similarly declined, e.g.
njiPe=cie-se 'in a fourth', ~ado=de-se-nitjombode. 'in the hundredth'. 'First' and
'second' are built to suppletive roots: va~enide, ombode.ombode.There are also
collective numerals built with =nie followed by possessivesuffix, e.g.
kolmo=nie-st 'the threeof them (-st p3)'.
Interrogativeand relative pronounsdeclinein the sameway as nouns,e.g.
genitive indefinite kona-ni 'whose', p3 kona-st 'which of them?', DDpN
mezj-tj-nie'which ones?',mezje-muP-i WHAT?-s1 EXISTS-s3pres'what do
I have?'The personalpronouns,however,deviatein that their oblique case-
 MORDVA 197

Table6.5 SelectedErzyapronominalparadigms

PRO.s3 Possessive.PRO.s3 REFLEX.PRO.s3

'(s)he' 'hislhers' 'himlherself'

N son sonLsiesi - son-zes i esi

G/A son-ze sonLsieniti - son-zeniti esLenze
Dat/All soni-enze soni-sieni~eni
soni-sesenit - son-zeni~eni son-zeseni~ esL~enze esL~enze
Ine soni-senze soni-sesenit i - son-zeseni~ son-zeseni~ esLsenze
Ela soni-stenze soni-zesenit
soni-steseni~ - son-stesenit
soni-zesenit i esLstenze
III sonLzenze soni-zesenit i - soni-zesenit
son-zeseni~ esLsenze
Prol soni-ganzo soni-gani~ soni-gani~ - son~ganj~. son~ganj~. esLkanzo
Abl sonl-d'enze soni-dieniti - sonld'enltJ esLtiedienze
(Trans sonLksenze soni-kseniti - son-kseniti esLksenze)
Abess sonLtiemenze soni-tiemeniti - sonitiemeniti esLtiemenze
Cfv sonLskanzo soni-skani~ soni-gani~ - son-skanit i esLeskanzo

forms endin redundantpersonmarkers,e.g. moni-se-niPRO.s1.obl-ine-s1'in

me', soni-se-nzePRO.s3.obl-ine-s3'in himlher'. There is also a possessive
pronoun,e.g. soni-siesi(- son-zesi)'his/hers'.The forms of the third-person
singularpersonaland possessivepronounsare setout in Table 6.S along with
the s3 paradigmof the reflexive pronoun.

The Verb

Verb Stems
The citation form of Mordva verbsis an infinitive built with -ms. The vowel-
final stem of a Mordva verb may be found by shearingoff this ending, e.g.
jarsams 'to eat', vowel stem jarsa-; jarsavtoms 'to feed', vowel stem
jarsavto-; jarsiniems 'to eat (freq)" vowel stemjarisinie-.The consonant-
final stem is obtained by lopping off the last vowel, e.g. jars-, jarsavt-,
jarisini -. The chief source of morphonological complications in Mordva
conjugationlies in stem-suffixsandhi,and the situationis not clarified by the
fact that the final vowel of somestemsis intrinsic (i.e. we might do betterto
segmentjarisinie-ms) while that of othersis epenthetic(i.e.jarsavt-oms).On
Mordva verb stemsin generalseeNadjkin 1981.

Conjugation
Mordva conjugationis exceptionallyrich in morphologicalentities,morpho-
nological rules, and paradigmaticcategories(Abondolo 1982: 11-23). The
primary category distinction is that of indefinite v. definite forms, e.g.
indefinite kund-an'I catch',definite kunda-tan'I'll catchyou'.
Use of the definite conjugationpresupposes (1) that the verb is transitive,
(2) that there is a definite direct object, explicit or implied, and (3) that the
aspectis perfective.The full definite paradigmevolved during the separate
life of Mordva, but the core aroundwhich it formed is probablymucholder.
198 MORDVA

Table6.6 Schematicfor Mordva verb: tenseandmood suffixes

Mood Tense Mood suffix Tensesuffix Personsuffixes

indicative present/future ( -f)- ) (-f)- ) present

past 1 ( -f)- ) -i-, -:/ past
past2 ( -f)- ) -i-P - past
conditional present/future -il'icPeria- present
(past -il'icPeria- past)
conjunctive present/future -vP- past
conditional! present/past -il'icP eria-vP- past
conjunctive
desiderative past -ikseP- past
optative present/future -zo- - -ze- present
imperative present/future -k - -t - -# (s2)
-do - -cPe (p2)

Crisscrossingthe definite/indefinite distinction is that of affmnative/

negative.Negativeverb forms are built in sometenses/moodswith negative
particles,in otherswith a negativeverb; comparea kund-an 'I don't catch',
ezj-i-ni kunda 'I didn't catch', a kunda-tan 'I won't catch you', ezj-i-tjinj
kunda 'I didn't catchyou'.
The Mordva grammars of the twentieth century all distinguish seven
moods, all of which maintain the definite/indefinite distinction, at least in
dialects(Keresztes1990: 45-50; contrastSerebrennikov1967: 202).
Tenseis distinguishedin only two of the moods, the indicative and the
conditional. The indicative distinguishesthree tenses:present,and first and
secondpast.The present-tense indicativecouldbe saidto be markedby double
zero,-0- forindicativeand-0- for present,e.g.sod-f:J-f:J-an'Iknow'; for thesake
of simplicity such zeroswill not be indicatedin cited forms. The past-tense
markersare -i- (-'(s)he
-~ in the third person)for the first past, -ilji- for the second
past,e.g. sod-i-ni 'I knew', soda-~-f:J'(s)he '(s)heknew', sod-ilji-nj 'I usedto know'.
Thestructureof affirmativefinite verbforms is setout in Table6.6.
When used as predicates,noun phrasesand adverbsare also conjugated;
only secondpast-tense(and not first past-tense)forms occur, and, since no
direct objectis involved, only indefinite personmarkers,e.g. kudo-so-nzo'in
hislherhouse',> kudo-so-nzo-lj-ini'I wasin hislherhouse'.Table 6.9 setsout
the present and second past-tenseforms of a verb, an adjective in the
nominative,andan inessive nounphrase.
In the presenttenseof the indefinite conjugation,the singularfirst (-n <
*-m) and second(-t) personsuffixes continuethe pU personalpronounroots
*mV, *tV, cf. the present-dayErzya personalpronounssl mon, pI mini, s2
ton, p2 tini. The third-person singular suffix -i is homophonouswith the
presentparticiple (see below), cf. lomani kand-i 'the person is carrying',
 MORDVA 199

Table6.7 Erzyaconjugation:personsuffixes: presenttense

Indefinite Definite conjugation

conjugation
Presenttense Object
s1 s2 s3 p1 p2 p3

Subject
sl -an -Tan -sa -Tadizj -sini
s2 -at -samak -sak -samizj -si~ -si~
s3 -1 -samam -Tanzat -si -samizj -Tadizj -sinizje
pI -Tano -Tadizj -sinjek -Tadizj -sinjek
p2 -Tado -samizj -sink -samizj -sink
p3 -itj -samizj -Tadizj -sizj -samizj -Tadizj -sizi

Note: (T = t tj d <Ii, accordingto morphonology)

Table6.8 Erzyaconjugation,personsuffixes: first pasttense(with past-tense

morphemes-i-, -~-) -~-)

Indefinite Definite conjugation

conjugation
First pasttense Object
s1 s2 s3 p1 p2 p3

Subject
sl -i-ni -i-tjin j -i-ja -i-diizj -i-ni
s2 -i-tj -i-mik -i-k -i-mizj -i-~
-i-ni
s3 -sLO -i-mim -i-mim -i-zje
-i-njzji~ -i-mizj -i-djizj -i-nizje
pI -i-niek -i-djizj -i-niek -i-djizj -i-niek
p2 -i-dje -i-mizj -i-nk -i-mizj -i-nk
p3 -i-dje
-sj-~ -i-mizj -i-diizj -i-zj -i-mizj -i-diizj -i-zj

kand=i lomani 'carrying people,peoplewho carry'. The -t(-) which figures

in the plural forms continuesthe pU pluralizer *-t. The secondcomponents
of both fIrst and secondpersonplural also descendfrom pronominalforms
(-Tano < *ta-mok, -Tado < *ta-tok). Negationin the presenttenseis carried
out by the particle a, e.g. a mor-an 'I don't sing', son a mor-i '(s)hedoesn't
sing'; this particleprobablydescendsfrom the baseof the pFU negativeverb
*e- - *a- (with metaphony*e/a > * a in *e wole- »avofl 'isn't').
The presenttensecould more properlybe termeda non-past,sinceit often
refers to the future. There are also compoundfutures built with the present
tenseof the verb karma- 'begins'and the -mol-meinfInitive of the main verb,
e.g. karm-an mora-mo 'I shall sing, I'm going to sing', a karm-an mora-mo
'I'm not going to sing'.
The fIrst pasttenseis built with two different tense-suffIxes,-si- in the third
200 MORDVA

Table6.9 Erzyaindefinite and predicativeconjugation:kand(0)- 'takes,

brings',mazi(j) 'beautiful',os-so'in a city'

kand(o)- mazi(j) os-so

Presenttense
sl kand-an mazij-an os-s-an
s2 kand-at mazij-at os-s-at
s3 son kand-i son mazi(j) son os-so
pI kand-tano maziG)-tjano os-so-tano
p2 kand-tado mazi(j)-tjado os-so-tado
p3 sinj kand-itj sin! mazij-tl sini os-so-~ os-so-~

Secondpasttense
sl kand-ilLini mazi-ILini os-so-ILini
s2 kand-i}i-it j mazi-ILini
mazi-ILi~ os-so-ILi~ os-so-ILi~
s3 son kand-ilLO son mazi-ILO son os-so-lj
pI kand-ilLiniek mazi-ILinjek os-so-ILiniek
p2 kand-ilLidje mazi-ILidje os-so-ILicPe
p3 sinj kand-iIL~ kand-iIL~ sini mazi-IL~ mazi-IL~ sini os-so-IL~ os-so-IL~

personand -i- in the axis of discourse.Both go backto suffixes reconstructed

for proto-Uralic (*-j, *-sj). The palatalizationof the initial consonants(-ni -tj
-cP) of the personmarkersis due to the -i- past-tensesuffix to their left. In the
first pasttense,negationis carried out by the negativeverb ez!-, e.g. ezj-i-nj
mora 'I didn't sing'. Historically, forms such as ez(i-ni are twice markedfor
pasttense,becausethe zj final in the negative verb descends from the pU past-
tensesuffix *-sL. It is possiblethat this sibilant suffix was originally used
only with the negative verb, then spreadto the affirmative paradigm (e.g.
kand-~-@)avoid to avoid homophonicclashwith the presenttense(kand-i).
The secondpasttense,which is lacking in Moksha,refers to eventsin the
pastwhich eitherlastedlong or habitually recurred.Its suffix, _ili_, developed
out of the verb ulje- 'is'; thus son mor-iP-@ '(s)he used to sing' is
homophonouswith the past-tensepredicativeform of the presentparticiple
mor=i 'singer',i.e. equals'(s)hewas a singer'.Negationis performedby the
a particle: a mor-ilj-ini 'I didn't usedto sing, I wasn'tin the habit of singing'.
Most of the personmarkersof the definite conjugationrepresentcombina-
tions of thepU personalpronounstems*m V *tV *sV, e.g.-samam,asin kunda-
samam '(s)he'll catch me', continuesthird-person*sV plus a reduplicated
fIrst-personmarker*m V. Detailsarespelledout in Bereczki(1988: 329).
The suffix of the conditional, -inidjeria-, is historically a combinationof
two morphemes.The frrst component,-ini-, is perhapsa reflex of the pU
conditional/potentialmorpheme*-ne-. The secondcomponentcontinuespre-
Erzya*tjerja-, a verb stemmeaning'tries, attempts'which survivesin dialects
to this day (cf. tjerjavto-id.). Sampleforms: mor-injtJieri-an 'if I sing', son
 MORDVA 201

mor-inidjeria-j 'if (s)he sings'; note the deviant morphophonologya-i > aj,
and contrastpresent-tense a-i > i, as in mor-i '(s)hesings'. (Past-tenseforms
of the conditionalexist in dialects;examples:mor-injdieri-i-ni 'if I sang',sini
mor-injdjeri-an 'if they sang'.)Negationis by meansof the a particle,e.g.
mor-inidjeri-e~-tj
a mor-injdjeri-an 'if I don't sing'.
The conjunctivesuffix -vP- (-voP - velj in final position) is historically a
combinationof a verb-forming derivationalsuffix =v- plus the secondpast-
tensemarkermentionedabove,and as suchis structurallyreminiscentof the
Tatar conjunctive (Bereczki 1988: 328). In more recent years conjunctive
forms are reinforcedby the particle bu « Russian6hI), e.g. mora-vP-ini (bu)
'I would sing, I would have sung', son moPe-velj-fJ(bu) '(s)he would go,
(s)hewould havegone'.The negativeconjunctiveis built with the first past-
tenseforms of the negativeverb avoP-, e.g. avolj-i-nj mora 'I wouldn't sing,
I wouldn't havesung'.
The suffix of the conditional/conjunctive,-inidjeriavlj -, developedfrom
the concatenationof the two mood markers just discussed,e.g. mor-
injdjeriavP-itj 'if you were to sing, if you had sung'.Similar sortsof unlikely
hypothesesand unreal conditions may also be expressedby the simple
conjunctiveplus the enclitic --gak (- --kak - --jak), e.g. mora-vP-itj--kak 'if
you were to sing, if you had sung'. Negation:a mor-injdieriavlj-itj 'if you .. '
The desiderativeis formed with -ikselj -. This, too, is a compositesuffix:
the first elementis identical with the (rarer) =iks variant of the first present
passiveparticiple =viks, which historically containsthe translativesuffix -ks;
the remainder(eP) descendsfrom the verb 'is'. Thus mor-ikselj-ini 'I wanted
to sing' is etymologically somethinglike *'1 was leaning-towards-becoming
(= translative)a singer".The negativedesiderativeis formedwith desiderative
forms of the negativeverb avoP-, e.g. avoP-kseP-inj mora 'I didn't want to
sing'.
The optative suffix (-zo - -ze) derives from a person marker: it is
historically identical with the third-personpossessivesuffix (kudo-zo'his/her
house'). The optative originally had only third-person forms, and to the
presentday theseforms are textually the most frequent. Its main use is as a
suppletionto the imperative,e.g. son mora-zo-fJ'may (s)hesing', mora-z-an
'let me sing', but second-person forms also occur,e.g. mora-z-tano'may you
(plur) sing'. Negativeforms are built to the negativeverb iPa-, e.g. iPa-z-an
mora 'let me not sing', iPa-zo-fJmolje 'may (s)henot go', etc.
The imperativehas only second-person forms. In the singular, -k and its
post-consonantalvariants variants
-tl-tj continue the pU imperative suffix *-k. The
suffix (-do - -dj e) of the plural imperativeforms is historically identical with
the secondcomponentof present-tense-Tado, i.e. it is of pronominalorigin
(pU second person *tV). Sample forms: mora-k 'sing!', mora-do 'sing
[plur]!', cav-t 'strike!', cavo-do 'strike [plur]!', tjej-tj 'do!', tjeje-die 'do
[plur]!'. Commandsareoften mitigatedand/ormademoreintimateby the use
of variousenclitics suchas --(j)a(k), --jat, --ka, --tia, e.g. kand-t--a(ja) 'carry
202 MORDVA

(, please)!',astje-die--jak'wait [plur] (a minute!),. The negativeimperative

(= prohibitive) is built with the negativeverb iPa-; the barestemservesin the
singular,e.g. aia mora 'don't sing!'. Furtherexamples:iPa--ka mora 'please
don't sing', iPa-do tjej 'don't [plur] do'.

Derivation
Mordva hasrich resourcesfor the productionof both nominalsandverbs,and
only a small samplingof the most importantsuffixes can be presentedhere.
Derivation may stay within part-of-speechcategories(cf. English strong>
strongish)or cross(cf. Englishstrong> strength,strength>strengthen).

Formation ofNominaIs
Here arethe moreproductiveErzyanominal-formingsuffixes:
=de forms ordinals from names of integers, e.g. kolmo=cie 'third'. This
suffix continuespU demonstrativebase*cje (> Mordva sje 'that,it').
=Ci forms abstracts,e.g. sumbra 'healthy'> sumbra=ci 'health'.From *keca
'frame, (circular) form' (cf. Finnish kehii 'ring, periphery', Mordva ci
'sun').
=ka forms nouns from adjectivesand vice versa, e.g. riizjanja 'sweet-and-
sour' > riizja~a=ka'sorrel', 'sorrel',atja 'Dad' > atja=ka 'male'; note also =aka,
which forms nominalizationsfrom verbs, e.g. pistje- 'goes without' >
pistj=aka 'wretch'.Probably(alsonext =ke) from pU *kkY.
=ke forms diminutive nounsand lends nuancesto adjectives,e.g. mekS'bee'
> meks=ke'little bee',ordakS'sensitive'> ordaks=ke'capricious'.
=ks forms metonymic/metaphoricnouns from nouns, e.g. sur 'finger' >
sur=ks 'ring', pizjol 'rowanberry' > pizjol=ks 'rowan bush'. Co-occurs
with =Ci (vide supra), as injalga 'friend' > jalga=ks=Ci 'friendship'.Also
forms result nounsand agentivesfrom verbs,e.g.jovta- 'narrates',jov=ks
'story'. From pU* =ks.
=ma and =mo - =me form nominaactionis and instrumentifrom verbs,e.g.
lovno- 'reads'> lovno=ma= '(act of) reading',iza- 'harrows' > iza=mo=
'harrow'. From different configurationsof a pU nominal-forming suffix
*=mV.
=n j forms adjectivesfrom nouns, e.g. erzia=nj 'Erzya [adj]', kev 'stone' >
kev=e~kev=e~ 'madeof stone'.Historically identicalwith the pU genitive *-n.
=nje (- =ne); =inje form diminutives from nominals,e.g. njurika 'short' >
njurik=inje 'short, little', kal 'fish' > kal=ne 'little fish'. Descendsfrom a
pU nominalizer*=nA.
=peP forms nounsfrom verbal nominals,e.g.jarsa=mo 'eating' > jarsa=mo
=peP 'food'. Sometimesheld to be a compound-formantrather than a
derivationalsuffix (cf. the nounpelj 'half).
=pulo forms collectives from nouns, e.g. tumo 'oak' > tumo=pulo 'oak
grove'. Sometimesheld to be a compound-formantrather than a deriva-
tional suffix (cf. the nounpulo 'tail; shaft;plait').
 MORDVA 203

=v - =j forms adjectivesfrom nouns, e.g. varia 'hole' > varia=v 'full of

holes', pitinie 'price' > pitinle=j 'expensive'. The j-variant favours
environmentswith front, the v- variant environmentswith back, prosody.
FrompU *=1).

Formation o/Verbs
Here are the moreimportantsuffixeswhich makeverbsfrom verbs:

=do- - =die makesintransitivesand mediopassivesinto factitives, e.g. ef.e-

'warmsup (intr)' > ef.=die- 'warmsup (tr)'. This suffix may be relatedto
the suffix =to- (- = tie- ), which hasthe oppositeeffect, e.g. stroja- 'builds'
> stroja=to- 'is built' . Perhapsfrom pU *=ttY.
=do- - =die- makes momentaneousverbs, e.g. vacka- 'stacksin a pile' >
vacko=die- 'strikes'. Probably cognate with the Mari frequentative-
formant =ede-.
=ksno- - =ksnie- makes iteratives, e.g. kekse- 'conceals',> kekse=ksnle-
'keepshiding'. A combinationof a pFV *=ks andof =no (seebelow).
=lie- makes frequentativesand duratives, e.g. korta- 'speaks'> kort=li e-
'converses'.Continuesa pU *1 frequentative.
=no - =nie makesfrequentativesandduratives,e.g. lovo- 'counts'> lov=no-
'reads'.Probablya continuationof a pU frequentative/durativesuffix with
*nt; cf. Moksha=nd-, e.g. M vane- 'looks' > vane=nde-'looks around'.
=sie- makesfrequentativesand duratives,e.g.jaka- 'goes,walks' > jak=sie-
'comesandgoes,wandersabout'.May be from p(F)U *=sj - *=cj .
=v- makes reflexives and mediopassives,e.g. cav- 'strikes' > cavo=v- 'is
smashed';this suffix has clear cognatesin Saamic, Fennic, Mansi and
Hungarian« pU *=w).
=vt- - =vti - makesfactitives andcausatives,e.g. ara- 'stands(intr)' > ara=vt-
'stands (tr)'; this suffix continues a suffix-string, either *=k=tV- or
*=p=tV-, with wide but not entirely cleardistribution acrossUralic.
=ziev- makes inchoatives, e.g. pezi- 'is afraid' > pezie=zie v- 'becomes
frightened'. Perhapsa combinationof a pFU momentaneoussuffix *=sj
with the reflexive/mediopassive =v- mentionedabove.

Thesesuffixes,mostly of p(F)U (*=t, *=j, *=k, *=1 + *=tv, *-m) or pFP([*=k

+] *=kt, [*=nt] *=sj) origin, makeverbsfrom nominals:

=d- - =di-; = ta- make instructives('provideswith X'), e.g. onks 'measure'

> onks=ta- measures';siorma 'writing; epistle' > siorma=d- 'writes'.
=gad- - =kad- (- =lgad- after vowels) « kad- 'leaves')makestranslatives
from adjectivesandnouns,e.g. aso 'white' > aso=lgad- 'turnswhite', laps
'flat' > laps=kad- 'becomesflat' .
=gavt- - =kavt- (- =lgavt- after vowels) makescausativeanaloguesto the
preceding,e.g. aso=lgavt- 'makeswhite', laps=kavt- 'makesflat' .
204 MORDVA

=ija- makestranslativesfrom nouns('becomesX-y'), e.g. sal 'salt' > sal=ija-

'becomessalty'.
=ld- makescolour statives,e.g.pize 'green'> pize= ld- 'is green'.
=m- makestranslativesfrom nounsand adjectives,e.g. cuvto 'tree; wood' >
cuvto=m- 'becomesstiff; cevtie 'soft' > cevtie=m- 'becomessoft'.
=nza- makesmultiplicativesfrom numerals,e.g. kolmo 'three' > kolmo=nza-
'trebles',niilie 'four' > niilie=nza- 'quadruples'.A combination,perhaps,
ofthe genitive *-n with pFP *=sj. Note also= nizia in pezienizia-'becomes
cloudy' (pezi=eni 'cloudy').

It is also not uncommon for related nominal and verb stems to be

homophonous,e.g. capo 'groove',cap(0)- 'cutsout', kalado 'tom, worn out',
kalad(0)- 'is ruined,collapses'.
Non-finite Verb Forms
Mordva is unusually rich in non-finite verb forms. Literary Erzya distin-
guishestwelve forms: threeinfinitives, threepresentparticiples,threeperfect
participles,andthreegerunds.Theseare set out in Table 6.10; rarerfunctions
andforms are given in brackets.
Negationis renderedby the particlesa, avo[1, and apak; their distribution
is determinedby both structural and functional factors. Thus, for example,
avoli is usedfor contrastivenegationof the infinitive: avo[1 mora-ms,a kistie-
ms 'not to sing, but/rather(a < Russian)to dance',cf. a mora-ms'to not sing';
avo[1 moramodokortatano, a liijado 'we're not talking about singing, but/
rather (we're talking about) somethingelse', a moramodokortatano 'we're
talking aboutnot singing'. Negativeparticiplesare formed with apak and the
bare verb stem, or with avoli and the suffixed form, e.g. apak mora moro -
avoli mora-zi moro 'songnot sung'.
The participles are often used as nouns; =ida then indicates the agent
(more rarely =i, =j) as in pidi=ida 'cook' (pidie- 'cooks'), car=i 'wheel'
(cara- 'turns'), cenga=j 'vodka' (cenge- 'bums'), and =st (- =t - =vt)
indicates the result of the action, e.g. kan=st 'burden' (kando- 'carries'),
poc=t 'flour' < pocodo- 'pours out'. The nominal-formingdeverbalsuffixes
=ks, =ma, and =mo/=me are alsoparticipial in origin.
Adverbs and Postpositions
Adverbs and postpositionsare often built with non-productivesuffixes from
non-productivenominal stems.Thesestemsusually have a spatial meaning,
and may function as both independentadverbsand postpositions,e.g. ike[1e
'in front (of)', udalo 'in back (of)', tombalie 'on the far side; beyond'.
Possessivesuffixes are also used, e.g. sikamo-n 'by myself, I (am) alone',
sikamo-t 'by yourself, you (are) alone', sikamo-nzo' himlherself, (s)he (is)
alone'; malaso-n(- moni malaso) 'next to me', langso-t(- toni langso) 'on
you', eksne-nze(- sonzeeksne)'behinditlhimlher' .
The caseforms of postpositionsand adverbsoften distinguish the static
 MORDVA 205

Table6.10 Erzya non-finite verb forms

Primary Negative Example Origin

junction/meaning built with

Infinitives
=mo - =me nominativeof mora=mo 'singing', pU *=mV
nomenactionis nieje=me 'seeing'
=m-s illative a,avoP mora=ms, 'to sing',
nieje=ms 'to see'
=mo-do - ablative a,avoP mora=modo'from
singing',nieje=
=me-cPe mecPe'from seeing'

Presentparticiples
=i(=da) non-pastnomen a (avoP) mor=ida 'singer', *=jA + *=ci
agentis mor=i lomani
'singingperson'
=viks passivenon-past, a (avoP) mora=viks 'being *=w + transI
passivepotential sung,singable'
=ma passivenecessive a mora=ma 'which pU*=m
mustbe sung'

Perfectparticiples
=z! perfectagent apak, avoP mora=z! 'sung', pU *-sL
(intransitives)or kulo-z! 'dead'
patient(transitives)
=ni perfectpassive (apak) mora=ni 'sung' pU *-n
participle
(folkloristic)
=vt- =vt agentivepassive (apak) mora=vt 'sung' pFU *=kt
participle nieje=vtj'seen'

Gerunds
=msto - anterior, avoP mor=amsto,'singing' =mo/e
=mste simultaneous nieje=mste'seeing' +Ela
=z! simultaneous avoP,apak mora=z! 'singing' perf.part
=do simultaneous(not avoP ~ta=do ~ta=do'standing' ? perf.part
productive) *=do<pFU
*-t?

from the dynamic, and within the dynamic, source from goal (e.g. ikelie
'in front' : ike[iev 'to the front' : ikelidie 'from the front'), but there are
many postpositionsand adverbsthat occur in only one caseform, as well.
Further examplesof adverbsbuilt from noun and adjective bases:Ci-ti 'in
the daytime',Ci-niek&ve-niek'(by) day and (by) night', tiezi-nia 'in winter',
kudo-niek 'along with all the people of the house',par-sitie 'well'; from
pronominal bases: koso 'where?', tiese 'here', kosto 'whence?', tieste
'hence',kozoni, kov 'whither?', tiezeni, tie} 'hither', kuva 'along where?',
206 MORDVA

tjija 'along here', koda 'how?', isitja 'like this', meks'why?', sieks 'for that
reason';postpositions:eziems 'insteadof, karso 'opposite',mejPe 'after';
ala 'under', aldo 'from under', alov 'to under', alga '(moving) along
beneath';pelje 'next to', peljdie 'from next to', pePev,pePej 'towards the
side of. Most postpositionsoccur with their noun in the nominative (e.g.
ezjems,alo) or the genitive (e.g. karso, pelje); a few take the ablative (e.g.
mejPe,ikePe).
Most othermodifiers andconjunctionsareof more recent(Russian)origin.
Examples:da 'yes', vot 'look', --lji '(questionparticle)', daj(tje) 'comeon!',
vedj '(invites agreementon part of collocutor),, i, di 'and', a'but (rather)', elji
'or', sto(bu) '(in order) that', butii 'if, xotj 'although'.

Syntax
Simple Sentences

Order of Constituents
The rich formal morphologicalapparatusof Mordva allows constituentorder
to fulfil semantic, logical, pragmatic and stylistic functions rather than
grammaticalones. Thus a four-membersentence,with S[ubject], O[bject],
A[dverbial complement]and V[erb] may occur with any of the theoretically
possible twenty-four orders. The basic rule is: topic(s) to the head of the
sentence,focalized elementimmediately before the finite verb; a focalized
verb must therefore stand sentence-initial.Suprasegmentals complicatethe
picture. A few examples:tjetjam kundasi kalt isiak (SVOA) 'it's my father
who caughtfish yesterday',tjetjam kalt kundasiisiak (SOVA) 'asfor what my
father caughtyesterday,it was fish', tjdam kalt isiak kundasi(SOAV) 'asfor
my father's catching fish, it was yesterdaythat he did that', tjetjam isiak
kundasikalt (SAYO) 'as for my father, it was yesterdaythat he caughtfish' ,
kalt isiak kundasitjdam (OAVS), 'asfor fish, it was yesterdaythat my father
caught them', kundasi kalt isiak tjdam (VOAS) 'my father did catch fish
yesterday'.
The order SVO is textually preponderant;it derivesfrom an earlier SOY,
which cededgroundunderforeign (Russian)influence.
Within the noun phrase,the order of elementsis fixed: modifier must pre-
cedemodified. Examples:e";ankudoso
keve~ odkudoso'I live in anewhousemadeof
stone',e";anod kudoso
keve~ kudoso'Ilivein anewstonehouse',Mordovija-sokas-itj
kudosocovo";a=zj di ljistvennoj vi";-t j MORDVINIA-ine GROWS-p3pres
piCe=~,
CONIFER=adj,MIXES=past.partAND LEAFY FOREST-pN'In Mordvinia
theregrow coniferous,mixed,anddeciduousforests'.
 MORDVA 207

Main Clause

RelationofSubjectto Predicate
We may abstract- somewhatartificially - the subjectand predicatefrom the
restof the sentencein orderto examinetheir interactionmore closely.
The subject is most usually a noun or pronoun, though adjectives and
verbal nominals(participle, infinitive) may also serve.The predicatemay be
nominal or verbal. Both verbal and nominal predicatesoccur in all persons
andtenses.Examples:varaka-sisiimev-si'the crow drankitself full' ,jaksamo
tiePe-si,psii kize-siCOLD WINTER-defHOT SUMMER-def '(the) winter is
cold, (the) summeris hot', ton dora-m-at PRO.s2SON-sl-s2pres'you are
my son', tonavtinlida-ti-nle udalo-t, uCitieP-esi stoli ekssePUPIL-plur-def
IN.BACK-p3 TEACHER-defTABLE BEHIND 'the pupils are at the back,
(and) the teacheris behind the table', sikamo-Pi-ni , kijak ezi sa ALONE-
past2-s1NO.ONE NEG.VERB.s3pastCOMES 'I was alone, no one came',
tini ombode-ti-nle-tiadoPRO.p2OTHER-pN-def-p2pres'you (plur) are the
seconds'.
When functioning as subject, toponyms appearin the definite, e.g. Sar-
ansko(je)siMordovija-nl stoliica 'Saranskis the capital of Mordvinia', as do
personalnameswhenfocalized,e.g.Niina-si sa-si'it wasNinawho came'.
Normally, subjectandpredicateagreein personandnumber.

Object-marking,Modality, andNegation
Determinacyof the object is indicatedby the choice of definite or indefinite
conjugation of the verb. The moods reflect different stancestowards the
reality of the eventexpressedby the predicate:indicative and imperativefor
the relatively real, optativefor what is not real but wishedfor, e.g. sonmolieze
'let himlher go, (s)hecan/maygo'. The otherfour moodsall refer to more or
lessunrealcircumstances.The conditionalrefersto conditionsor hypotheses,
e.g. moPinitJie~an moPinitJie~an 'if! go' in moliinltJie~anbazarov,bazarov, Pismeraman 'if! go to
market I'll buy a horse'.The conjunctiveimplies possibility or desirability,
e.g. son palavoP (bu) '(s)he might kiss/might have kissed'; it also occurs
frequently in compound sentences,e.g. isiak molievPini uma~s,uma~s, a tidi a
molian 'yesterdayI might havegone/mighthavefelt like going berrying, but
I'm not going today'. The conditional-conjunctiverefers to contrary-to-fact
conditions,e.g. son palinldieravoP 'if (s)he were to kiss (but [s]he won't), if
(s)hehad kissed(but [s]he didn't)'. In a compoundsentence:uPinltJie~avli ti won't),
ton lomanl, pePenzecarstvammaksovliijatieri 'if you werelhadbeena (real)
human being, I would give/would have given you half my kingdom'. The
desiderativecovers desired but non-occurringevents in the past, e.g. son
palikseP'(s)hewould haveliked to kiss (but didn't)'.
Negation of the verbal predicatein the indicative (present,secondpast,
future), the conditional,and the conditional-conjunctiveis carriedout by the
particle a. Conjugatedforms of the negativeverb are used in the first past
208 MORDVA

(e-), and in the conjunctive and desiderative(avol-); conjugatedforms of

the prohibitive verb (ili a-) are usedin the imperativeand optative.Negation
of nominal predicatesis normally effected with a or avozi. Examples: a
lomani '([s]he/it is) not a person',alavozi poks '([s]he/it is) not large', a
lama '(it is) not a lot', a eriavi '(it is) not necessary/allowed';avoli atia
'(he is [still]) not an old man', avoli sirie '([s]he/it is) not old', avoli isitia
'it isn't like this'. Non-existenceis expressedin Erzyawith the particlearasi,
which is also used for categoricaldenials (avozi is used to expressdoubt).
Examples: - toni uzi-i siemija-t? - arasi, mon apak urivaksto PRO.s2
EXISTS-s3presFAMIL Y-s2 NOT.EXIST PRO.sl NEG.PARTIC MARRIES
'do you have a family?' 'No, I haven'tmarried/amunmarried';- tink uzi-i
ruzoni+erziani valks?- arasi, ruzoni+erzianivalks miniekarasi, vesietiirai-
osi mije=zi PRO.p2 EXISTS-s3pres RUSSIAN-ERZYA DICTIONARY
NOT.EXIST, WHOLE PRINTRUN SELLS=part 'do you have a Russian-
Erzyan dictionary?' 'No, we don't have a Russian-Erzyandictionary; the
entire print run is sold out'; - ton umok us uc-at moni ej-se?- avozi, mon
s-i-ni vetie minuta-do ikelie PRO.s2 LONG.TIME ALREADY WAITS-
s2presPRO.sl.geniaccIN-loc NEG.PARTPRO.slARRIVES-pastl-slFIVE
MINUTE-abl BEFORE 'haveyou beenwaiting for me long?' 'No, I arrived
five minutes ago'; - kijak s-i-lii? - tiese kijak arasi. kijak ezi saksno.tieCi
konastkak ezi sakSno NO.ONE ARRIVES-s3pres--interr.part HERE
NO.ONE NOT.EXIST NO.ONE NEG.VERB.s3pastl COMES TODAY
WHO-p3--partNEG.VERB-s3pastlCOMES 'is anyonecoming?' 'No one
is here. No one camehere. Not one of them cametoday'; a meri-an karSo
val-gak NEG.PART SPEAKS-slpresAGAINST WORD-part 'I'll not say
a word againsthimlher/it'.

Direct Objects
An indefinite direct object is normally in the nominativeindefinite, but the
ablative is often used with non-count nouns. If the verbal aspect is
imperfective,the indefinite forms areusedevenif the direct objectis definite.
Examples:liisme ram-i-ni HORSE-ID.sNBUYS-pastl-sl 'I boughta horse',
lovnan jovtiniema READS-slpres STORY-ID.sN 'I'm reading a story',
varaka-si karma-si kajsie-me kevnie-ti CROW-DDsN BEGINS-s3pastl
THROWS-infPEBBLE-pN'the crow beganto throw pebbles',jars-anjam-
do, siim-an vetJi-tie EATS-slpresSOUP-ablDRINKS-slpresWATER-abl 'I
eat soup, I drink water', skal-osi tiikse(-tJie)porni-i COW-def.sNGRASS(-
abl) CHEWS-s3pres'the cow eatsgrass'.
A definite direct objectis normally in the definite declension,and usually
in the genitive/accusative,althoughthe inessiveis also used. Generally,the
definitenessof the direct objectis echoedby definite conjugationforms on the
finite verb or personalsuffixes on the infinitive. Personalnamescount as
indefinite in Erzya,definite in Moksha.Examples:tie liisme-niti tietia-m ram-
izie THIS HORSE-DDgeniaccFATHER-sl BUYS-s3s3pastl'myfather
 MORDVA 209

bought this horse (as for this horse,it's my father who bought it)', njej-an
Pisme-njtj/liisme-se-njtjSEES-s1presHORSE-DDgen/accor HORSE-ine-
def 'I see the horse', dora-~ dora-~ ~e-injzje ~e-injzje tiejtjer(tj~e-~tiejtjer(tj~e-~katka, BOY-DDN SEES-
s3p3pastGIRL-DDplur-gen/acc'the boy saw the girls', katka, uzo, kiska--~ katka,
karm-i saje-me-tj CAT WATCH.OUT DOG-DDsN WILL-s3pres
CATCHES-inf-s2'watchout, cat, the dog will catchyou'.

AdverbialComplements
These are expressedwith inflected nominal and postpositionalphrases,as
well as with infinitives. The definite and indefinite declensionalforms are
identical from the syntactic perspective.The casesrefer primarily to space
(dat/all, ine, ela, ill, prol, abl, lat) and time (dat/all, ela, ill, prol, occasionally
nom and gen/acc),but they also encodeindirect object (dat/all), instrument
(ine), cause(abl, trans),measureandotheradverbialcategories(abl, cfv). For
details see Wiedemann 1865: 33-47, Koladenkov and Zavodova 1962:
106-57, ErdSdi 1968: 209-14, Cygankin 1980: 159-78, Keresztes1990:
73-5. Examples:kardaz-soastje-~ marto vedi marto kuvsin YARD-ine EXISTS-
s3past1WATER WITH JUG 'in the yard was a jug of water',piljge-m kev-s
tomb-ija FOOT-s1-sNSTONE-ill illTS-s1s3past1'I hit my foot on a stone',
moP-ankavto vajgeljbe-tj GOES-s1presTWO VERST-pN 'I'm walking two
versts',dokov mazi=stemor-i NIGHTINGALE BEAUTIFUL=adv SINGS-
s3pres 'the nightingale sings beautifully', varaka-njtj-e~ varaka-njtj-e~ a koda ~ime-ms~ime-ms
CROW-def-dat/allNEG.PART HOW DRINKS-inf 'the crow didn't manage
to drink at all', ram-i-~ ram-i-~ knjiga ~ado-ska ~ado-ska delkovoj-dje BUYS-past1-s1
BOOK-sN HUNDRED-cfv RUOBLE-abl 'I bought a book for about a
hundred ruobles', sluz-an uCitjeP-ks velje-se SERVES-s1presTEACHER-
transVILLAGE-ine 'I serveas a teacherin a village'. The tridimensionaluse
of the elative, illative, and inessivelocal casesis one of the more archaic
featuresof Mordva, e.g. bazar-stokalaci ram-i-nj MARKET-ela LOAF-sN
BUYS-past1-s1 'I bought a loaf at (lit. from) the market', poza~ik-e~ poza~ik-e~
kacamo-s
Pepija-~ kacamo-sFIREFIGHTER-sNSUFFOCATES-s3pastl FIRE-ill 'the
firefighter suffocated in (lit. into) the fire', ~ive-die kaj-imim kaj-imim kudo-sto
COLLAR-abl THROWS-s3sl.pastlHOUSE-ela'(seizingme) by (lit. from)
the collar (s)heflung me from the house'.Thereare also verbsand nominals
which govern the inessiveor the ablative,e.g. karta-so naljkse-tanoCARD-
ine PLAYS-p1pres 'we're playing cards', mora-mo-do mePe-m moP-~ moP-~
SINGS-inf-abl FEELING-s1 GOES-s3pastl'I don't feel like singing any
more',bodka-~ peSk~e peSk~epeSk~e vina-doBARREL-def.sNFULL WINE-abl 'the cask
is full of wine' .

Attributive Constructions
Nominalsand participlesmay serveas attributes.Qualitativeandquantitative
attributesare unmarked;possessiveattributesare in the genitive/accusative.
The formal distinctionis not alwaysclear,however,becausemanyqualitative
210 MORDVA

adjectivesare formed with a suffix (=ni) which is homophonouswith that of

the genitive. If the possessorattributeis definite, the possessionmay standin
either the appropriatepossessiveor definite-declensionform, e.g. nieri-eniti
kuvalmo-zolkuvalmo-sfezi sfato BEAK-def.sG LENGTH-s3/LENGTH-
def.sN NEG.VERB.s3pastlIS.SUFFICIENT 'the length of the beak wasn't
sufficient'.
Furtherattributive examples:Mordovija-va cutF-iti kavtopokSliej-ti : Sura
di MoksaMordvinia-prol FLOWS-p3presTWO BIG RIVER-pN SURAAND
MOKSA 'two greatrivers flow throughMordvinia: the Suraandthe Moksha',
vejkinie, veekeviks dora-zo kulo-sf SOLE LOVES-part SON-s3 DIES-
s3past3 'hislher only beloved son died', ksnii-ni izamo-tinie cuvto-ni=
sfe-ti-nietFepar-t IRON-abl HARROW-pN-def WOOD-abl-sec.def-pN-def
GOOD-pN 'iron harrows are better than those (made) of wood', mezietie:
ksnii-ni vatFria kojme izii vatFria ksnii-ni kojme?WHAT ... 'what is this: a
goodiron spadeor a goodspademadeof iron?', viri Ciri e-seerizia=ni dora-sf
penig-ti keri-i FOREST EDGE-ine ERZYA=adj BOY-def.sN BILLET-pN
CHOPS-s3pres'at the edge of the forest an Erzyan boy is choppingwood',
varaka-niti sfimema-zosa-sfCROW-genTHIRST-s3ARRIVES-s3pastl'the
crow becamethirsty'.
The existential verb uli - is used in possessivesentences;the possessor
standsin the genitive and the possessiontakesthe pertinentpossessivesuffix.
Examples:moni uzi-i alasa-mPRO.sl.genEXISTS-s3presHORSE-sl 'I have
a horse', moni uzi-iti mazij(-ti ) pada-n PRO.sl.gen EXISTS-p3pres
BEAUTIFUL(-pN) SCARF-plurposs.sl'I have beautiful scarves',uCitieli -
eniti uzi=nie-sf vadiria kudo-zo TEACHER-gen EXISTS=freq-s3.pastl
BEAUTIFUL HOUSE-s3'the teacherusedto havea beautifulhouse'.

Constructionswith Verbal Nouns

Constructionsbuilt with verbal nouns are common Mordva equivalentsof
subordinateclauses.Thosebuilt with infinitives mainly expressobjector goal
clauses,e.g. ucitieli-esf meri-sf tonovtinie-msvadiria-sto TEACHER-def.sN
SAYS.s3.pastlSTUDIES-inf GOOD=adv 'the teachersaid to (= said that
we/they/youlone, etc. should) study well', son sov-i obeda-mo PRO.s3
ENTERS-s3presHAS.DINNER-inf '(s)he goes in to have dinner'. Those
built with participles serve as qualifiers, e.g. isf=eni sajezi otFiriva-sf pek
mazij YESTERDAY=adj CATCHES=part BRIDE-DDsN VERY BEAU-
TIFUL 'the bride who got married yesterdayis very beautiful', sodav=t
lomani-esf tieCi arasfezi KNOWS=part PERSON-DDsN TODAY
NOT.EXISTS-s3.past2'the person whom they (we, people) know wasn't
(there) today', kado=viks tieve-niti son ezi-izie priado LEA VES=part
WORK-DDsGA PRO.s3 NEG.VERB-s3s3.pastlFINISHES '(s)he didn't
finish the work that was left'. Those built with gerunds serve the same
function asclausesof manneror temporalclauses,e.g. anisfak ojmsfe=zi lama
a tiejat ONLY RESTS=gerMUCH NEG.PART DOES-s2pres'if you only
 MORDVA 211

rest, you don't do much',tieve-ni apakpriado a eriav-i ojm~e-msojm~e-msWORK-GA

NEG.PART FINISHES NEG.PART IS.ALLOWED-s3pres RESTS-inf 'if
you don't finish the work, you can'trest'.

Compound Sentences
Co-ordinateclausesform compoundsentenceseither asyndetically,e.g. vetP-
kepetPev-~, 'the water rose, the crow drank its fill',
e~ kepetPev-~, kepetPev-~, varaka--~kepetPev-~, ~imev-~
or with a conjunction,e.g. kardaz-soastie-~ kuvsin,kuvsin,no vedi uli=nle-~ uli=nle-~ ani~ak
kuvsin,
i
potmakske-se-nlt'in the yard was a jug, but therewas water only at its very
bottom'.
Subordinateclausesarenormally linked by meansof conjunctions,e.g.e.g. son
nie-izie riivezi-eniti , kona-~ kona-~ sal-izie saraz-onlti '(s)he saw the fox, which
(had) stole(n)the hen',a sodan,ziardo s-i meke-v'I don't know when(s)he'll
comeback'. Suchsubordination,often effectedby meansof conjunctionsof
Russianorigin (e.g. sto 'that', butii 'if, xoti 'although'),is spreadingat the
expenseof equivalentconstructionsbuilt with verbal nouns.

Lexicon
Layers of the Lexicon
Mordva roots belongto threemain etymologicalclasses,eachaccountingfor
about30 per centof the lexicon: inherited,innovated,and borrowed.The rest
of the root inventory (about 10 per cent) is without etymology. Textual
frequency is more difficult to assess,but a glimpse is provided by the text
given at the end of this chapter: of the twenty-ninewords occurring in this
text, fourteen are inherited (three times each: varaka 'crow', vetP 'water';
twice: a 'NEGATIVE PARTICLE'; onceeach:sa- 'comes',koda 'how', nleri
'beak', son 'PRO.s3',karma- 'begins', and probably sato- 'is sufficient'),
nine are Mordva innovations(~ime=ma ~ime-ms 'drinking', ~ime-ms~ime-ms 'to drink' and
'drinks'drinks one'sfill', from ~im-
~imev- from 'drinks'; marta 'with', from mar 'hill,
heap'; uli=nie- 'is FREQ', from u[i- 'is'; potm=aks=ke 'the very bottoml
inside', frompotmo 'insidepart'; kaj=~e-'throws
'throwsaround,flings', from kaja-
'throws'; kev=nie 'pebble', from kev 'stone' and kepecPe=v-'rises' from
kepecP-'lifts'); five are loans (kuvsin 'jug', three times, and no 'but' once,
both from Russian; and kardaz 'courtyard', from Baltic), and only one is
without clearetymology(astie- 'exists').
Inherited vocabularyis, in turn, of various ages, and may date back to
Finno-Volgaic, Finno-Permic,Finno-Ugric, or proto-Uralic. At eachof these
levels, a given word may be an internal innovation or a loan. Prominent
amongthe former are onomatopoeicor affective-descriptivevocabularysuch
as Erzya puva- 'blows', which probably continuesa proto-Uralic word of
onomatopoeicorigin (*puwV-, *puyV-, cf. Nenetspii- 'id'), and other items
such as nacko 'damp' (Finno-Ugric *niackV, cf. Kazym Khanty niasax
'raw'), tartPe- 'stiffens' (Finno-Permic*tara 'stiff, cf. Komi tur=d- 'stiffens
212 MORDVA

with cold'), and pizje- 'rains' (Finno-Volgaic *pisUla- 'drips', cf. Finnish
pisa=ra 'drop').
Iranian and Indo-Iranian Loans
The oldestloanswhich canbe tracedinto various stagesfrom proto-Uralic to
Finno-Volgaic are those from (Indo-)Iranian; R6dei (1986: 94-5) counted
roughly forty such items, e.g. mije- 'sells', a word dating back to Uralic
*miye- 'gives',itself a loan from Indo-Iranian;tjeje- 'does',which datesback
to FU *teke-, an Indo-Iranianor early Iranian loan. Other examplesinclude
sal 'salt', which canbe tracedback as far as Finno-Permic*salV, azija 'thill'
« FP *ajsa), sazor 'younger sister' « FP *sasare),uzjeri 'axe' « Finno-
Volgaic *wasjara). Mordva paz 'god' and tarvaz 'sickle' lack cognatesin
Uralic, but their reconstructedshapes(FP *pakase, *tarwase) suggestthat
they were borrowedfrom some form of early Indo-Iranian;similarly,Other ~eja
'goat' « FP or FV *sjaka or *sjawa) looks like an old Iranian word. Other
Mordva words with probableIranian backgroundsare kSnji 'iron' and sirinje
'gold'; their antecedentsmay have beenborrowedfrom a form or forms of
Old or Middle Iranian into Finno-Volgaic (*ktirtana, *semja); Mordvapango
'mushroom'is more difficult to date, since its reconstructum(*pankV) may
reflect FU or FV age. If a Mordva word of putative Iranian origin has a
pendantin Mari, as well, it is difficult to determinewhetherwe have to deal
with joint or separateborrowing, e.g. Moksha dong;} 'island', Erzya pusto
'pap,mush','silver'.
~ija 'silver'.
Other Indo-EuropeanConnections
Contactswith speakersof Tocharianmay lie behindthe antecedentof Mordva
'wire' « pU *wasjke '(a type of metal),); the word for 'salt', sal, also,
'wire'
u~ke
may stemfrom Tocharianandnot Iranian.
Evidence for contacts between speakersof (Uralic) Finno-Volgaic and
(Indo-European)Baltic may be found in items suchas +kirda '-fold; -times'
(FV *kerta, cf. Saamigeardi), ksna 'strop' (FV *sisna,cf. Finnishhihna), and
tjejtjeri 'daughter'(FV *ttiktare, cf. Finnish tytiir). The Mordva word for the
elm (ukso), on the other hand, may ultimately be of either Baltic or Iranian
origin; sazor 'younger sister' has also been consideredas Baltic, and not
Iranian,in origin.
SemanticFields
An analysisof 566 inheritedMordva roots revealedthe following breakdown
accordingto semanticfields: nature,flora, fauna 186; actions 124; dwellings,
clothing, meals, transport, tools 112; parts of the body 55; properties45;
kinship 23; spatialorientation17; religion 4 (Zaicz 1988: 402; seealso Mosin
1985).
Loans into Common Mordva
The mostimportantgroupsof loanwordsfrom otherlanguagesinto Common
Mordva are Iranian, Baltic, Turkic, and Russian (Serebrennikov 1965:
 MORDVA 213

242-56).Besidesthese,however,theremustalsohavebeenmanyotherdonor
languageswhich are now no longer spokenand of which we have no trace.
Furthermore,there is some small but compelling evidenceof contact with
speakersof East Germanic:pondo 'pound', and there is certainevidenceof
later (eighteenth-century)contactwith Germanspeakerswho settledin the
Volga area,e.g. keptie,J'linden bark basket',pu,Je 'mead'.Loansfrom Baltic
may havetravelledvia Muromaor Merja (two extinctwestUralic languages),
andsomeolderTurkic loans mayhavecomeinto Mordva via Mari.
Much of the vocabulary sharedby Mordva and Mari may have been
borrowed separatelyinto the two languagesfrom and via unknown donors.
Examplesincludearvo 'honeydissolvedin water',embe[i 'carpenter'splane',
lav 'dandruff' (East Mari lege), siniaro 'that much', and zaba 'child'. All
layers of the borrowed lexicon of Mordva await an up-to-datereworking,
including both those words which have been borrowed directly and those
which have entered Mordva via Turkic and Russian mediation; Mordva
calquesbasedon Turkic andRussianmodelsshouldalsobe re-examined.The
groundwork for such undertakings lies ready in the recently published
dictionaryby Paasonen(1990-96).
Iranian Loans
Old and Middle Iranian loanwordswhich enteredCommonMordva more or
less directly are usually classifiedunder the labels Scythian,Sarmatian,and
Alanic. Relatively certain membersof this layer are E a[ia M aliii 'young
man', E kani siti M kaniJ'hemp',EM lomani 'humanbeing',EM marazi 'elm
(E), maple (M)',E pitinie M pitini;) 'price', E se,JaM siii,J;) 'acorn',E tirdaz
'thrush',E vasta 'spouse',E vi,JesM viir;)s 'lamb'; the list could be doubled
with the inclusion of controversial examples. Iranian influence was not
limited to the lexical sphere; according to E. Lewy, a great deal of the
definiteness-marking andagreementphenomenaof Mordva may be attributed
to Iranian interference, altering Mordva typologically to resemble more
inflecting typesoflanguagessuchas (Iranian) Ossetian(Stipa 1973: 10).
Baltic Loans
Theseseemto be roughly equal to the Iranian elementsin number,but it is
often difficult to determinethe time andtrajectoryof the borrowing: E kardaz
'(court)yard', E [ienige M [ienig;) 'linden bark', E [iija M liijii 'other', E
mukoro M m;)k;)r 'buttocks',E panst M pand;n 'bridle', E pejeli M pej;)[i
'knife', E pu,Jginie 'thunder',E simeniM siim;mi 'root; tribe', E suro M sur;)
'millet', E tor '(ritual) knife; rod (used at weddings)',E tiozoni M tioZiini
'thousand'.
Thrkic (Tatar and Chuvash) Loans
Quite a few dozen Middle Bulgarian (= Old Chuvash) loans entered
especiallysouthern,i.e. pre-Moksha,dialectsof Mordva during the heydayof
the Volga Bolgar Empire (seventhcenturyCE). Only a few of theseChuvash
214 MORDVA

words persistinto the present-dayliterary languages;examplesareE ajira M

aj( ;,)ra 'cold, biting wind', E kenWalM kelda 'bedbug',E komoziaM komljii
'hops',E ksumanj M kusma'radish',E siirit j M siiri:Jk 'elm, ash',E siukoro,
M duk;,r 'a kind of pie'.
With the thirteenth century came the Tatar invasions, and Tatar loans
succeed Chuvash loans in Mordva. On (western) Mordva territory the
predominantform of Tatar to exert an influencewere the western,primarily
MiSer, dialects; to the east, where Mordva speakershad settledbeyondthe
Volga, KazanTatarandBashkirwerethe main influences.
Miser-Mordvacontactswere a two-way street:many Mordva words were
borrowed into Miser, as well. A more striking example of the depth of
Mordva-Turkic contact is the ethnic and linguistic duality of the Karatay
Mordva, who are descendants of Mordva speakersbut who now speakTatar.
The Moksha Mordva in the south were more directly influenced by Tatar
contact, as is witnessedby Tatar loans found only in Moksha and not in
Erzya,e.g. M aru 'clean',debiiri 'good',ezna 'brother-in-law',isa 'willow',
konak 'guest', osal 'evil', ud;JZ 'cheap'; there are also a few words with
the opposite distribution, e.g. E tolkun 'wave', uda 'spine', ulov 'the
deceased'.If we count all lexical items including obsoletewords and words
with limited dialectal distribution, the number of Tatar loans into Mordva
exceeds one thousand: for example, Tatar loans beginning with k- in
Paasonen1990-96 number close to one hundred. As for the present-day
literary languages,there may well be a hundredin Mokshaand roughly half
that many in Erzya. Among the Tatar loans of the core vocabulariesof the
Mordva literary languagesare (both Erzya and Moksha unless otherwise
indicated): ajdja- 'drives', alasa 'horse', E ajgor M ajg;,r 'stallion', ulav
'car; team of horses',E paksia M paksiii 'meadow',E emef.M im.n 'fruit,
berry', E paridej M paR.idi 'silk', dora '(young) man', dotmar 'cudgel',
cavka 'Gack)daw',kurka 'turkey', jarmak 'money; coin', E karSo M karSa
'opposite',E pek M piik 'very'. Chuvashand Tatar influence today is quite
limited and affects only a few dialects. The enormouslychallengingtask
of sorting out the chronology and the linguistic-geographical and
phonological detail of the older Turkic loans in Mordva still awaits an
appropriatelytrainedresearcher.

RussianLoans
There are no Common Slavonic elementsin Mordva, and lexical evidence
from Old Russianis scant and uncertain; tentative candidatesfor the latter
loan statusinclude rozj 'rye' and rasta- 'increases(in number)'. Russian-
Mordva contactswereclosein the eleventhandtwelfth centuries,but the best
evidencefor such contactsis in the influence exercisedby Mordva on non-
northernRussiandialects(cokanieand evenakanie; seeStipa 1973: 13-24).
In subsequentcenturiesMordva providedmany Russianplace namesin the
Vyatka area.
 MORDVA 215

From the thirteenthto the sixteenthcenturyRussianinfluenceon Mordva

rivalled that of Tatar; from the seventeenthto the nineteenthit surpassed
it; and in the twentieth century Russian influence becameexclusive. All
linguistic levels were affected: the adoption of IfI and Ix! phonemes,the
spreadof the correlationof palatalization,the useof conjunctions(borrowed
from Russian)and the resulting shifts in sentencestructure, all originated
in Russian. Towards the end of the twentieth century not only Russian
dialects but also the Russianliterary standardplayed an important role. In
the realm of the lexicon this influence has swollen into a mass of several
thousandsof words. A short sampling from literary Erzya might include
bednoj 'poor', narodnoj 'folk-', rodnoj 'domestic',glavnoj 'chief, zivotnoj
'animal',scastlivoj 'happy,fortunate',vesiola 'merry', brat 'brother',nudka
'grandchild', gositj 'guest', soldat 'soldier', jaguda 'berry', zjorna 'seed',
kalada 'loaf, vina 'wine, vodka', stolj 'table', stul 'chair', skap 'armoire',
zjerkala 'mirror', skola 'school', tjetradika 'notebook',~iga ~iga 'book', pera
'pen', pomod 'help', olja 'will; freedom' robota- 'works', stroja- 'builds',
poluca- 'gets', otveca- 'answers', ,Jisova- 'sketches', raz '+times'
(competing with the very early Baltic loan kirda from the Finno-Volgaic
age), tisca 'thousand'(replacingthe earlier Baltic loan tjozonj ). Present-day
Mordva technicalvocabulariesand the vocabulariesof the naturaland social
sciences consist, with very few exceptions, of Russian loan elements
combinedandextendedon Russianmodels.

Words of InternalOrigin
Onomatopoesisaccountsfor severalhundreditems, including verbs such as
buldordo- 'murmurs,bubbles',daxa- 'neighs',dimbePdje-'flashes',di,Jnje-
'(piglet) squeals',gPonkodje- 'drinks in a noisy fashion', karno- 'claws',
korca- 'slurps',potja- 'sucks' (- potje 'femalebreast'),t,JonkadJe-'thumps'
(- t,Jonk 'soundof thumping') and barnyardand other animal calls such as
ksu+ksu'shoo!' andkuko 'cuckoo'.
Less affective vocabularyhas been regularly made by the processesof
derivation and compounding.Derivation was broachedon pp. 202-4; here
we list a few more (Erzya) examples:from al 'lower (part)' is derived alks
'(feather-)bed', and from ava 'mother' are derived avaks 'female (animal)',
avaks 'female (bird)" and avavt 'mother-in-law'. The word for 'lock',
panzoma,is derived from the verb panzo- 'opens'; the word for 'splits',
lazo-, is derived from the noun for 'plank', laz. Further examples:jamks
'millet' (jam 'soup'), siePmukS-t'eyeglasses'(sieljme 'eye'), keve,Je-'rolls
away' (keve,J 'round'), siormav 'multi-coloured' (siorma 'writing; em-
broidery').
Compoundingmay be co-ordinative or subordinative.Examples of the
former are verbs such as kistje-+mora- 'hasfun' (DANCES+SINGS),ljisie-
+sova- 'comes and goes' (EXITS+ENTERS), kuzje-+valgo- 'swings'
(RISES+FALLS),slja-+narda- 'washesup' (WASHES+WIPES),sitja-+pra-
216 MORDVA

'lives'(STANDS.UP+PALLS), viePe-+soka- 'farms' (SOWS+PLOUGHS)

and nouns such as ponks-t+panar-t'linen' (TROUSERS+SHIRTS),ejkakS
'child' (CHILD+CHILD), suks-t+unza-t 'vermin' (WORM+BUG). Mit-
igative verbs are built by combiningthe lexical verb with the verb root tieje-
'does', e.g. astie-+tieje- 'stands around (ineffectually), (astie- 'stands'),
jarsa-+tieje- 'eatsaway at' (jarsa- 'eats').
Most subordinativecompoundsare of the order X+Y, i.e. are simple
combinationsof two root morphemes,e.g. inieveeP 'sea' (inie 'big' + vedi
'water'),kirigapari 'throat' (kiriga 'neck' + pari 'basin'),modamari'potato'
(moda 'earth' + mari 'apple'). Some morphophonemicsmay be involved,
with varying degreesof productivity, e.g. u - i in odiriva 'bride' (od 'young'
+ uriva 'daughter-in-law'),me - {} in siezived i 'tear' (siezime 'eye' + veeP
'water'). Collocationsinvolving the genitive of the modifier are besttreated
as syntagms(and the orthographywrites them as two words), e.g. ije-ni ska
'season(YEAR'S TIME)', pria-ni snamo 'self-praise(HEAD'S PRAISE)"
sanzav-onikoct 'cobweb(SPIDER'S WEAVING)' .
Quite a few dozenderivatesand compoundconstructionswhich are now
extinct may be found in eighteenth-centurysources;the semanticfields of
religion, government,societalstructure,and law predominate,and conscious
neologizing was probably the causein most cases.Examplesinclude cudii
vedi 'river (FLOWING+WATER)" iniazoro(-ni) eriamo 'Russia;the State,
the Tsardom(BIG.LORD'S+LIFE)',jovtiliida 'prophet(STORYTELLER)"
kovoni val 'calendar (MONTHLY+WORD)" ziija keliida-t 'pagans
(OTHERIFOREIGN LANGUAGE'D.ONES)" paz-oni urie 'angel (GOD'S
+SERVANT)" pukSturdi 'soldier (ONE.WHO.CAUSES.[CANNON].TO.
FIRE)" sacmo pezi 'family (TRIBE+PART)" vesie lomati 'world (ALL-
+PEOPLE),.

Erzya Mordva Text

A: text in transcription, segmented;B: morpheme-by-morpheme gloss; C:
closetranslation;D: freer translation;
nomsx= derivationalsuffix which forms nominals;vbsx = derivationalsuffix
which forms verbs.

Al varaka-njtj sa-sj sjime=ma-zo A2 kardaz-so

Bl CROW-DDsG COMES-s3pastlDRINKS=nomsx-s3B2 YARD-ine
Cl the crow's it came its thirst C2 in yard

astje-sj vedj marto kuvsin, no vedLesj

STANDS-s3pastlWATER WITH JUG BUT WATER-sNdef
it stood with - - water jug but the water
 MORDVA 217

kuvsin-senhj u1j=nie-sj potm=aks=ke-se A3 varaka-njtjenj

JUG-DDine IS=freq-s3pastlINSIDES=nomsx- B3 CROW-DDdat
=nomsx-ine
in thejug it was at the very bottom C3 for the crow

a koda sjime-ms: a sat-i

NEG.PART HOW? DRINKS-inf NEG.PART IS.SUFFICIENT-s3pres
no way to drink not it is enough

njerLeze A4 son karma-sj kaj=sje=me =kuvsin-enjtjenj

BEAK-s3 B4 PRO.s3BEGINS-s3pastl THROWS= JUG-DDdat
freq-inf
its beak C4 it began to throw into thejug

kev=nje-tj A5 vedLesj kepedje=v-sj

STONE=nomsx-plur B5 WATER-DDsN RAISES=vbsx-s3pastl
pebbles C5 the water it rose

varaka-sj sjime=v-sj
CROW-DDsN DRINKS=vbsx-s3past1
the crow it drankits fill

D1 The crow becamethirsty. D2 In a courtyardstood a jug with water, but

the water was at the very bottom. D3 There is no way for the crow to drink:
its beakisn't (long) enough.D4 It beganto throw pebblesinto thejug. D5 The
waterrose,the crow drankits fill.

Referencesand Further Reading

Abondolo, D. (1982) 'Verb paradigm in Erza Mordvinian', Folia Slavica 5:
11-24.
Bereczki, G. (1988) 'Geschichteder wolgafinnischenSprachen',in D. Sinor (ed.),
The Uralic Languages.Description, History and Foreign Influences,Handbuch
der Orientalistik8, Leiden: Brill, pp. 314--50.
- - (1994) Grundzilge der tscheremissischen Sprachgeschichte,vol. I, Studia
Uralo-Altaica35, Szeged:UniversitasSzegediensis de Attila J6zsefnominata.
Bubrix, D.V. (1947) 3p351-MOp~oBcKa51
rpaMMaTHKa.rpaMMaTHKa. MHHHMYM, Saransk:Mordo-
vian StatePublishers.
Budenz,J. (1877) 'Moksa-es erza-mordvinnyelvtan',NyelvtudomanyiKozlemenyek
13: 1-134.
Collinder, B. (1957) 'Mordvin', in Survey of the Uralic Languages,Stockholm:
Almqvist & Wiksell, pp. 227--46.
Cygankin, D.V. (1980) (ed.) fpaMMaTHKa MOp~OBCKHX MOp~OBCKHX 513hIKOB. <l>oHeTHKa,
rpaq)J{Ka,opqJOrpacpH51,MOPCPOJIOrH51,Saransk:Mordovian StateUniversity.
Erdodi, J. (1968) Erza-mordvin szovegek(magyarazatokkal,nyelvtani vazlattal es
szotarral),Budapest:Tankonyvkiad6.
Ermuskin,G.I. (1984) ApeaJIhHhIeHCCJIe~OBaHH5IHCCJIe~OBaHH5Ino BOCTOqHhIM cpHHHo-yropcKHM
218 MORDVA

H3hIKaM (3p3H-MOP/lOBCKHHH3bIK), Moscow: Nauka.

Feoktistov, A.p. (1966) 'MOP/lOBCKHe H3bIKH', in VI. Lytkin et al. (eds),Nauka,
~3bIKH
HapO/lOB CCCP 3, <l>HHHo-yropcKHe H CaMO/lHHCKHe H3bIKH, Moscow: Nauka,
pp. 172-220.
- - - (1975) 'MOP/lOBCKHe H3bIKH', in VI. Lytkin et al. (eds), OCHOBhI
<pHHHo-yropCKOro H3bIK03HaHIDI 2, IIpH6aJITHHcKo-<pHHcKHe, caaMcKHH H
MOP/lOBCKHe H3hIKH, Moscow: Nauka.
Keresztes,L. (1987) Geschichtedes mordwinischenKonsonantismus,vol. I, Studia
Uralo-Altaica27, Szeged:UniversitatisSzegediensis de Attila J6zsefnominata.
- - - (1990) ChrestomathiaMorduinica, Budapest:Tankonyviad6.
Koladenkov, M.N., and Zavodova, R.A. (eds) (1962) rpaMMaTHKa MOP/lOBCKHX
(MoKlIIaHcKoro H 3P3HHcKoro) H3hIKOB, I: <l>oHeTHKa H MOP<POJIOrIDI, Saransk:
MordovianBook Publishers.
Mosin, M.V (1985) <l>HHHo-yropCKaH JIeKCHKa B MOP/lOBCKHX H
ITpH6aJITHHcKO-<pHHCKHX H3hIKax (CeMaHTH'leCKHH aHaJIH3), Saransk: Mordo-
vian StateUniversity.
Nadjkin, D.T. (1981) 'OcHoBa rJIarOJia B MOP/lOBCKHX H3bIKax, ABTope<pepaT'
[unpublishedPh.D. thesis],Tartu: Tartu StateUniversity.
Paasonen,H. (1903) Mordvinische lautlehre, Helsingfors: Finnische Literatur-
gesellschaft.
- - - (1990-6) MordwinischesWorterbuch,compiled by K. Heikkila; edited and
publishedby M. Kahla, vols I-IV, Helsinki: SocieteFinno-Ougrienne.
Raun,A. (1988) 'The Mordvin language',in D. Sinor (ed.), The Uralic Languages.
Description, History and Foreign Influences, Handbuch der Orientalistik 8,
Leiden: Brill, pp. 96-110.
Redei,K. (1984) 'PhonologischeAnalysedes Erza-Mordwinischen',in P. Hajdu and
L. Honti (eds), Studien zur phonologischenBeschreibunguralischer Sprachen,
Budapest:AkademiaiKiad6, pp. 209-30.
- - - (1986)Zu den indogermanisch-uralischen Sprachkontakten, Wien: Verlag der
Osterreichischen AkademiederWissenschaften.
Serebrennikov,B.A. (1965) 'MCTOPIDI MOP/lOBCKOro HapO/la ITO /laHHbIM H3bIKa', in
B.A. Rybakovet al. (eds),3THoreHe3MOP/lOBCKOro HapO/la,Saransk:Mordovian
Book Publishers.
- - - (1967) MCTOpH'leCKaHMOP<POJIOrHHMOP/lOBCKHX H3bIKOB, Moscow: Nauka.
Stipa, G. J. (1973) Mordwinisch als Forschungsobjekt,Napoli: Istituto Universitario
Orientale.
Veenker, W. (1981) 'Zur phonologischenStatistik der mordvinischen Schrift-
sprachen',UAJb (NeueFolge), vol. I: 33-72.
Wiedemann,F.J. (1865) Grammatik der ersa-mordwinischenSprache nebsteinem
kleinen '" Worterbuch, St. Petersburg:Commissionlireder KaiserlichenAkade-
mie der Wissenschaften.
Zaicz, G. (1988) 'A mordvin lexika osi elemei',in P. Domokosand1. Pusztay(eds),
Bereczkiemtekkonyv,Budapest:HumanitiesFacultyofthe University of Budapest,
pp. 397-402.
- - - (1993) 'Hangrendes illeszkedesa mordvinban',in M. Sz. Bakr6-Nagy and
E. Szij (eds), Hajdu emlekkonyv,Budapest:HumanitiesFaculty of the University
of Budapest,pp. 427-32.
7 Mari
Eeva Kangasmaa-Minn

The Mari (older name: Cheremis) languageis spoken in Central Russia

around the middle stretch of the Volga and in scatteredareas in the east
towardsthe Urals. There exists a Mari Autonomous(Soviet) Republic (now,
Republic of Mari EI) extendingto both sides of the river. The Volga is also
a dialect boundary:the Mari on the western,hilly side speakkurfJk mari or
Hill Mari thoseon the low easternshorespeakolfJk mari or MeadowMari.
Only about one half of the Mari live in their own republic. A large
concentrationof the Mari populationis found in the BashkirRepublic,around
the city of Ufa. The Mari in the Bashkir areacall themselvesupo mari; they
speakfar easternvariantsof MeadowMari dialects,a fact which indicatesthat
they havewanderedeastwardsfrom the left bankof the Volga.
The Mari are very unevenly distributed: in their own republic they
compriseless than one-half of the population (323,009 or 48.3 per cent in
1989)and areoutnumberedby Russians.SpeakersofMari tendto live in rural
areas, while the cities, most notably the capital loskar ala, are almost
completelyRussian.In addition, small numbersof Tatars and Chuvashlive
among the principal nationalities. In Bashkiria the number of Mari was
reported in the 1989 censusas 106,793. Together with the Mari living in
scatteredsettlementsin neighbouringdistricts, the number of EasternMari
was 160,000in 1970. Since the Soviet censusin 1989 registeredaltogether
670,277personsof Mari nationality, the proportionof the Mari living outside
their own republic is aboutone-third(Lallukka 1990: 112-24).
The Mari have long been 3ettled in the area, but toponymsindicate that
they once lived to the west of their present-dayhabitations. They are
mentionedin the lordanesChronicle(551) by the nameof Sremnisc,together
with their close relatives the Merens. The latter name seems to be a
Latinization identical with the present-dayself-designationmario The areas
which the Mari have inhabitedhave also beenhome to Chuvashand Tatar.
The Chuvashbuilt a large and flourishing commercialempire which spread
over central andsouthernRussiaand far into the Balkans.The Tatarsarrived
somewhatlater, with GenghisKhan, and overthrewall political opponentsin
the region. After the Russianshad freed themselvesfrom Tatar supremacy
they beganto invade the areaaroundthe middle stretchesof the Volga, thus
establishingdominanceover the earlier inhabitants. The Mari people and
languagecontinuedto live side by sidewith Turkic peoplesandlanguages.

219
220 MARl

It was not until the twentieth century, and especially under the Soviet
regime, that the Russification of Mari culture began in earnest.Material
conditionsandeducationalstandardswereimproved,but at a cost: a decrease
in the usefulness,cultivation, and prestigeof the nativelanguage.Only in the
villages, and there onlyat the elementarylevel, was educationcarriedout in
Mari; all higher education was in Russian. The Mari have retained,
nevertheless,a senseof their cultural uniquenessand of the value of their
nativelanguage,perhapsbetterthan any othercentralFinno-Ugric group.
The Mari languagewas first adaptedto literary use by missionarieswho
translatedpartsof the gospelsinto the Western(Hill) Mari dialect.
From the beginning, there were two literary languages,one based on
Western(Hill), one on Eastern(Meadow) Mari. The two varieties are more
or less mutually understandable,the differences consisting mainly in the
inventory and distribution of the vowels and in certain discrepanciesin
vocabulary.The phoneticstructureof EasternMari is somewhatsimplerthan
that of the Westerntype, a fact which has tendedto favour it as the more or
lessofficial languageof Mari culture.
The Mari people have a rich body of folklore and now a flourishing
literature,aswell. Their interestin their ethnic andnationalheritageis evident
in the text fragmentincludedat the endof this chapter,a fragmenttakenfrom
a literary journal, celebratingthe achievementsof the ethnologistandfolklore
collectorTimofei Evseev.

Phonology
(Literary) EasternMari hasnineteenconsonantandeight vowel phonemes.
Consonants

Distribution
The unvoiced stopsp t k occur in all positions,i.e. word initially: koklaste
'among',tuOfm 'hislhers',pal~e
hislher 'known'; medially after both voiced and
unvoicedsounds:koklaste,Zap~stw.e
hislher 'in hislher time', skesotan 'as a matter
of fact'; and word-finally: kalfJk 'people',surt 'farm', zap 'time'. The voiced
counterpartsto the stops, written here as b/w d g, presentsome allophonic
complications. In earlier days it was sufficient simply to state that the
phonemesin questionoccurredas voiced fricatives except after homotopic
nasals,when they assumedthe quality of voiced stops.In a compromisewith
phonetic transcription these phonemeswere written as ~ 6 y, since the
fricative allophonesoccurredfar more frequently than the stop allophones.
Nowadays,however, the velar and dental fricatives are usually replacedby
stopsin all positionsand I seeno controversyin writing them as such,i.e. as
d g. For the sake of consistency,the labial should also be written with b in
all environments;but for the sakeof clarity in this chapter,w and b will be
used according to the more common pronunciation.This practice parallels
 MARl 221

Table7.1 Consonantphonemesof EasternMari

Phonemes Cyrillic equivalents

Unvoicedstops p t k II T K
Voiced b/w d g 6/B ~ r
counterparts
Affricate C q
Unvoiced s S c III
s(h)ibilants
Voiced s(h)ibilants z Z 3 )K

Nasals mn mn ni U M H Hh H
Laterals I Ii JI JIh
Trill r p
Glide j ii

WesternMari hasfour additionalphonemes:

tj Th
f x cp x
c U;

Of thesethe palatalizedtj and the two voicelessspirants,velar x and builabial or

labio-dentalJ, occur in Chuvashloanwords;the dental affricate c is found in native
words, as well.

that of the literary language,which usesthe Cyrillic letters<r> and<)J;> for velar
anddental,but two different graphsfor the correspondinglabial phoneme:<6>
after m, as in KOM60 kombo 'goose'and <B> elsewhere:JIyBIiT Zuwit 'fifteen'.
This graphic discrepancyis due to a skew in distribution, which is itself due,
in tum, to phonetichistory. Mari g and d representthe p(F)U stops *k and *t
in voiced environments,but the voiced labial representsa merger of two
p(F)U consonants:*p in voiced (and, a fortiori, non-initial) environments,
e.g.phonetic 'fox' (cf. Finnish repo of revo-n+tuZe-t'auroraborealis')and *w
in both medial andinitial positions,cf. Zuwit 'fifteen' (Finnishviite 'five') and
phonetic'maple' (Finnish vaahtera).This fact accountsfor the differencesin
distribution: whereasg and d occur only word-medial after a vowel or a
voicedconsonant;their labial counterpartis found not only word-mediallybut
word-initially as well. Therefore1 havefollowed the principle adoptedin the
literary languageandhavethroughoutthis chapterusedthe characterb for the
historical allophone after *m and the characterw elsewhere,even for a w
which hasdevelopedfrom an earlier*p.
Examplesof the occurrencesof g, d, and b/w: soga '(s)he stands',tud~tud~
nergen 'abouthim', kundem~se,kundem~se, 'hailing from the X region', Zombo 'Prunus
padus',kuwa 'old woman',wer 'place',Zuwit 'fifteen'.
The hushingaffricatechasa voiced allophone[~]tud~after ni. It is not marked
in script andis lexically rare. Example:sinicem'I know'.
222 MARl

MorphophonemicAlternations
The occurrence of voiced v. voiceless consonant phonemesis largely
conditioned by their phonetic environment. In principle, unvoiced stops,
affricates, and sibilants change into their voiced counterpartsin voiced
environmentsand vice versa. Thus the voiced fricatives g and d become
voicelessstops before voicelessconsonants.Example: ludsowlaklan 'to the
readers',derived from the verb ludem 'I read' (pronounced[lucem]), is
pronounced[lutsowlak1an].In the postpositionalconstructionawa deeposna
MOTHER FROM SEPARATE 'without mother', the initial voiced d of the
postpositiondee 'from' has developedfrom a t: cf. the historically related
locationalnoun ter-, tor- 'proximity'.
Furtherexamplesof the alternationbetweenthe two series:kio-~ 'hand 'hand
sG', kit-kx 'from the hand',jol-gx 'from the foot', kit-se 'hislherhand',jol-
to 'hislherfoot' .
However, certain suffixes do not obserVe this basic rule, but remain
unaltered in all environments,e.g. the verbal noun suffix =SV (lut=so
'reader') retains its s even after a voiced consonant:tol=so 'one who is
coming'. There are also several derivational suffixes in which a t remains
unalteredafter an I or r, e.g. pualtem 'I thresh',nortem 'I makewet', but not
after a nasal:contrastporemdem'I makewet' .
It follows from the morphophonemicrule given above that voiced
fricatives/stops and voiced sibilants do not form clusters with voiceless
consonants.Certain types of consonantcluster may occur word-finally,
especially those consisting of slllr + t (surt 'village', wost 'through') but
medial clusters which becomefinal through morphology lose their first or
secondmemberin most cases,e.g. kockam 'I eat', koc 'eat!'; sincam 'I sit',
sic 'sit!' Word-finally, voiced stops/fricativesare automaticallyreplacedby
their voicelesscounterparts:comparethe long and short forms of the present
third personsingularnegativeverb in ogeSnal- ok nal '(s)hedoesnot take'.

Vowels
EasternMari vowels:
Cyrillic equivalents
a ti u Ii y Y
e 6 a o e 6 hI o
a a

WesternMari has,in addition,the phonemes

a hI
a a
 MARl 223

EasternMari vowel alternations(allophonicrepresentations

in brackets):
Non-final syllables Final syllable
First Non-first Non-word-final Word-final
ou ieuoaS S 0
oti i e ti 0 [a][;}] [;}] 0
ase i ieuoaS S e
Figure7.1 WesternMari palatalattraction

1st syllable 2nd syllable 3rd syllable Example

a:lOU a:lOU a:lOU to!Jna

'we come'

e1 e 1 kolenii
'we die'

a :loiiei a :loiiei a ,.n,; dlenii

'we live'

The presentationof the vowel phonemeswould ideally use three dimen-

sions,sincethereare threephonologicalcontrasts:
1 Full v. reduced,the latter beingwritten S. Literary Hill Mari distinguishes
two reducedvowels, ~ and a (fronter) <7; some WesternMari dialects
distinguishfour reducedvowel qualities.
2 Front v. back. Front aree i 0 a (and <7 ii in Western);a 0 u ~ are back.
3 Labial v. illabial, i.e. 00 u av. a e i ~ (<7 ii).
Distribution
All vowel phonemesoccurin all syllables,but with restrictionsimposedby the
morphophonemicalternationsoutlinedbelow.In EasternLiterary Mari (hence-
forth: EM), the back non-high vowels a and S have front variantsif the first
syllable has 0 or a (in dialects,also after e and i); theseallophonesare not
indicatedin the orthography.In the WesternMari (henceforth:WM) literary
languagenon-fust-syllableii and ;} are similarly not indicatedin writing, but
herethey arephonemes,not allophones,sincethey occurin thefirst syllable,as
well: niild '(s)hetakes'(EMnald),t<7g 'beginning'(EM tag).
Morphophonemicalternations. Mari has two different types of vowel
alternation,eachmorphophonemic ally conditioned.The first type is rootedin
the fact that the first vowel in a word rules structurally over the following in
subsequentsyllables. This so-called vowel harmony is typical of several
Uralic, as well as of Turkic andMongolic languages.Vowel harmonyin these
languagesis usually an alternation, in correspondingstructural entities, of
back with front vowels, of roundedwith unroundedvowels, or of both. The
224 MARl

secondtype of vowel alternationis basedon the contrastbetweenreducedv.

full vowels and is characteristicof the Easterndialects.
In EM the contrastback : front, labial : illabial and reduced: full cannot
be dealt with separately.The WM vowel systemis ruled by palatal attraction
andthe situationis both more logical andmorecomplexthanthe easternone.
In EM, palatalprosodyprevailsthroughthe word only if the first vowel in
the word is front rounded0 or ii, i.e. then only front vowels (or front-vowel
allophones)occurin the rest of the word. If thereis a back vowel in the first
syllablethen therewill be only backvowels and/orthe phoneticallyfront, but
phonologically neutral vowels e and/or i in subsequentsyllables. The
alternationof full v. reducedoccursin morpheme-finalsyllables.An ~ in this
syllable remains reduced if a further suffixal morpheme is attachedbut
otherwiseit is converted toa full mid vowel. The quality of this full vowel
is determinedby that of the first precedingfull vowel in the (non-compound)
word: (1) if this vowel is unrounded,~ > e; (2a) if it is roundedand front,
S > 0, but (2b) if it is roundedand back, ~ > o. Examples: Examples:
t~lfJze 'moon',
'reader's',
t~l f JzfJ-ste 'in the moon' , sor 'milk' , sor-zo 'its milk' , lud=so 'reader' , lud=sfJ-
n 'reader's',kaj=se 'walker', kaj=sfJ-lan 'to a walker'. (The conditions for
this alternationare not uniform acrossthe subdialects,and in somedialects
the alternationis triggeredby somesuffixes and not by others.)
The situation in WM is of specialinterest. Since here the occurrenceof
back v. front vowel dependson the phonetic quality of the vowel in the
precedingsyllable, a word may changesides,as it were, in the middle of its
structure:comparetolfmam 'we seemto havecome' with koleniim 'we seem
to have died'. On the other hand, labial harmony is weak or nonexistentin
WM and there is no alternationbetweenreducedand full vowels. The great
frequencyand perseverance of the reducedvowels obscurethe phonological
patternsof WM; it is perhapspartly due to this fact that the EM varietieshave
establishedthemselvesas the primary basisfor the standardlanguage.
Word stressis non-phonemic.In EM it falls on the last phonologicallyfull
vowel, e.g. olmg 'apple' but m!1no 'egg' (phonologically/munSI), m!1nfJ-n 'of
an egg'. If a word containsonly reducedvowels, the stressfalls on the first
syllable, e.g.wilSk;nniis
t~lfJzfJ-m 'moon (accusative),.In WM stressusually falls on the
penult, even if it containsa reducedvowel and the following syllable has a
full vowel, e.g. wilSk;nniis
~stiis 'to sweep';thereis lability, however,if sucha penultis
precededby full vowel, e.g. wilSk;nniis wiisk~tiis 'to hurry', but wilSk;nniis 'haste'
(Alhoniemi 1985: 15-40).

Morphology

Morpheme Classes
Mari words can be monomorphemic,but most contain at least two mor-
phemes.Morphemesfall into three classes:bases,suffixes, and enclitics. A
 MARl 225

baseis a morphemethat can appearalone as a monomorphemicword, or as

the first memberof a polymorphemicword. Any sequenceof a baseplus one
or more affixes which,in its morphologicalfunctions,is equivalentto a base,
is a stem.
A suffix is a morphemewhich occursonly in connectionwith a baseor a
stem, i.e. bound in a polymorphemicword. Suffixes fall into three classes:
derivational,inflectional, andenclitic.
A derivational suffix is one which when addedto a baseor stemforms a
polymorphemicword that in its morphologicalfunctions is equivalentto a
monomorphemicword. The morphologicalclass(part of speech)of a derived
word is determined by the derivational suffix occurring last in it. An
inflectional suffix is one which when addedto a baseor stem indicatesthe
syntactic function of the form in question. Derivational suffixes precede
inflectional ones,with one exception:the adjective-formantsuffixes,
=s~, e.g. Morko
suffixes, 'hailing from the district of Morko' can occur to the right of
kundem~=se
local casesuffixes,e.g.pus-~s=sosuffixes,
jel) BOAT-ine-adj.sxPERSON'the person
in the boat'. Enclitics are boundmorphemesthat may occurlast in any word,
regardlessof morphologicalcategory.
Thereare a few suffixes which hoverbetweenthe categoriesof casesuffix
andenclitic, e.g. the comitative-ge (seebelow).

Morphological Categories
We may classifyMari basesinto threegroupsaccordingto their susceptibility
to suffixation: nouns,which takecasesuffixes; verbs,which take tense/mood
suffixes; and particles,which do not take suffixes (theseare usually petrified
caseforms of non-productivenouns).

NominalInflection
Nominal inflection extendsto nouns,pronouns,and numerals.Adjectives are
declinedonly when they function syntacticallyas nouns;they are declinedin
neitherattributivenor predicativeposition.
Caseforms. As indicated above, the boundariesbetweencase suffixes,
derivationalsuffixes, andenclitics arefar from clear-cut.The numberof cases
reportedby various grammariansthereforevaries considerably.Wichmann
e1923: 8) gives the numberas 13, asdoesCollinder (1957: 124), but Sebeok-
Ingemannlist 8 (1956: 33-41), and Alhoniemi has 10 (1985: 44-5). Native
grammariansusuallylist 7, e.g. SavatkovaandUcaev(1956: 803).
In Table 7.2 below, noun inflection is illustratedby five stems:kol 'fish',
wer 'place', wilt 'water',muno 'egg',andpasa 'work'.
The non-marginalcasesmay be divided into two basicgroups:nominative,
accusative,and genitive are purely grammatical cases,i.e. they indicate
syntactic role and have no semanticcontent. The rest of the non-marginal
casesare in the broadestsenselocal cases,sincethey can containa semantic
indicationof locality.
226 MARl

Table7.2 EasternMari nouninflection

kol 'fish' wer 'place' wilt 'water' muno 'egg' pasa'work'

N kol wer wtit muno pasa

A kol~mkol~m wer~mwer~m wtid~mwtid~m mun~m
mun~m pasam
G kol~nkol~m wer~nwer~n wtid~nwtid~m mun~n
mun~m (pasan)
D kollan werlan wtitlan mun~lan
mun~m (pasalan)
Ine (koles)
(kol~sto) wer~stewer~ste wtid~st6 wtid~st6 mun~ste
mun~m pasaste
Lat (koles) weres wtides munes
mun~m pasas
III (koles)
(kol~s[ko]) wer~s(ke) wer~s(ke) wtid~s(ko) wtid~s(ko) mun~s(ko)mun~s(ko) pasas(ke)
(Abl kollec werlec wtitlec mun~lecmun~s(ko) pasalec)
Marginal cases:
Mod kolla werla wtitla mun~lamun~la pasala
Com kolge werge wtio~ge wtio~ge mun~gemun~ge pasage
(Car kolde werde wtitte mun~demun~de olmade)

Note: Textually lessfrequentforms in brackets;seetext.

Within the local-casesubsystema three-waydistinction is maintained,as

is characteristicof Uralic languages:staticlocation in a placeis distinguished
from motion to (GOAL) and motionfrom a place(SOURCE).The inventory
of casesrecruitedto servethis trichotomy is lopsided:thereis only one static
case,the inessive, and only one sourcecase,the ablative, but there are as
many as four goal cases.The functions of the lative and the illative overlap
somewhat;the former seemsto designatea more permanentchangeof place
or conditionthan the latter (Alhoniemi 1967: 320-1).
Anotherfactor crisscrossesthe categoriesof case,namelythat of animacy.
The nominativeand accusativearecommonlyusedby all nouns,but genitive,
dative, and ablative are common only with nouns which refer to animates,
while inessive, lative, and illative occur more frequently with inanimates.
Deviationsfrom this basicpatterndo occur; for example,the inanimatepasa
'work' occursin the dative in a constructionsuch as pasalankaja 'goes(to
look) for work'. The lative of animatesis not uncommon in utterances
indicating a changeinto a state,e.g. kuguzanudfJZ~ udfJZ~ watesnal~udfJZ~ 'he took
the king's daughterasa wife' .
In dialects, the ablative -lee is quite common as a marker of the bested
memberof comparisons,e.g. mu-leetutlo HONEY-abl SWEET 'sweeterthan
honey'.
In addition to the core caseforms thereexist a few marginal caseswhich
may also be classifiedas declension.Most prominentamongtheseare modal
-la, comitative -ge, and caritive -de. These suffixes have a somewhat
restricteddistribution, andforms built with them thereforeresemblepetrified
adverbs;on the otherhandtheir meaningis not so clear-cutas to suggestthat
they be termedadverbialmodifiers as such.
Modal -la (sometimescalled the comparative)has lative undertonesin
 MARl 227

constructionssuch as mar-la ojla 'speaksMari', but is decidedly adjectival

(andderivational)in a constructionsuchas totar=la sequence
sol~ 'Tatarscarf'.In yet
a third role, the modal suffix is tackedon to illative forms, thus resembling
sequence kaja VILLAGE END-ill-mod GOES 'goesto
an enclitic, e.g.jal muca-sk~la
the end of the village'; the sequence sequence
-sk~la is treatedas a single suffix (the
'approximative')by somegrammarians(e.g. Bereczki 1990: 32, 39).
The comitative quite commonly has an inclusive meaning, e.g. maskan
kum igfJge satasenfJtBEAR-gen THREE CHILD-com GOES.ASTRAY-
past2-p3'the bear went astray with its three cubs'. The caritive, e.g. jol-de
'without legs', contrasts not with the comitative but with the adverbial
genitive, as injol-fm kost-d '(s)he/it goeson foot'; an exampleis trupka-de
kuzeii-em TOBACCO.PIPE-carHOW? LIVES-slpres'How (can/will) I live
without (my) pipe?'.
The above-mentioned functional gapsin the somewhatlopsidedlocal case
systemare smoothedover with the help of postpositions.The mostimportant
of theseare historically petrified caseforms built from non-productivelocal
nouns with obsoletedeclensionalsuffixes. Among the more important are
gx, with variousseparativemeanings,e.g.port gx lektas'to comeout ofthe
house'andde-, with fossilized forms det(fm) 'from', den(e) 'with, at, by',
deran 'at', deke 'to'. From the point of view of declension,the mostimportant
function of de- is supplying animateconceptswith the meansof expressing
their locationalrelations,e.g.jolmaldec serfJs 'a letter (ser;)s)from a friend',
ozakuwaideke'(comingback)to the old woman(who is their) owner(aza)';
the form den also servesto link noun phrases,e.g. maskaden pire BEAR
WITH WOLF 'a bearand a wolf.
Pronounshavea limited declensionwhenusedas headwords.
Numberand Person.Therearetwo numbers,singularandplural, andthree
persons,first, second,and third. The singular and plural personmarkersare
sl -(e)m, s2 -(e)t, s3 -(fJ)i(fJ), pI -na, p2 -da, p3 -(fJ)/I; the -e- elementof the
sl and s2 suffixes is dropped after a stem-final a. The distribution of the
various allomorphsof the third-personsuffixes is also largely phonologically
conditioned.
In the genitive and accusative,personmarkersprecedethe casesuffix, in
the local casesthey follow it; in the dative theordervaries.Cf. Table7.3.
A kind of generic,non-personalplural may be expresssed by the p3 suffix,
e.g. suspfJk-fJstmura er deneNIGHTINGALE-p3 SINGS-s3presMORNING
AT 'nightingalessing in the morning', note also tosdfJStulo kazdejeyfm
FRIEND-p3 EXISTS EVERY HUMAN-gen 'every person has hislher
friends'.
There are also severaldedicatedplural markers,occurring with different
distributions in different dialects. Chief among these in easterndialects is
-wlak, e.g. pD lud=so-wlak-lan 'to readers'.This suffix may be a loan from
Tatar, where it is a noun meaning 'cattle'; or it may be related to Mordva
ve[ie 'village' and Saarni valvi 'pack (of wolves, dogs)'. Its position in
228 MARl

Table7.3 SelectedEasternMarl possessivedeclensionforms: animatenounkol

'fish'

Nominative Genitive Accusative Dative

sl kolem koleman kolemam kolemlan

s2 kolet koletan koletam koletlan
s3 kolazo kolazan kolazam kolazlan
pI kolana kolanan kolanam kolanalan
p2 kolada koladan koladam koladalan
p3 kolast kolastan kolastam kolastlan

Table7.4 SelectedEasternMarl possessivedeclensionforms: inanimatenoun

port 'house'

Nominative Accusative Dative Inessive Lative Illative

sl portem portemam portlanem portastem portesem portaskem

s2 portet portetam portlanet portastet porteset portasket
s3 portso pOrtsam portsalan portastazo portesaze pOrtaskazo
pI portna portnam portlanna portastana portesna portastana
p2 portta porttam portlanda portastada portesta portaskada
p3 portast portastam portastlan portastast portesast portaskast

relation to person and case suffixes is rather labile. In the nominative, it

can either precedeor follow the personmarker, e.g. olma-m-wlakor olma-
wlak-em 'my apples', and in the grammaticalcasesboth orders, person-
number and number-person,are possible again, e.g. joltas-em-wlak-~ wlak-em(
FRIEND-sl-plur-acc,joltas-wlak-em( - ~) FRIEND-plur-sl-acc 'my friends
wlak-em(
(acc)'.In local cases,besidesthesetwo orders,therealso existsa third order,
number--case-person,e.g. olma-wlak-fJst-emAPPLE-plur-ine-sl 'in my
apples'.The only restrictions operatingin trisuffixal sequenceswith local
casesseemsto be that the plural marker cannotcomelast, and that the case
marker cannot come first. Other, distributionally more restricted, plural
suffixes in EM are -la, usedmainly with nounswhich designateplaces(e.g.
ola-la 'cities'; see also Derivation, p.234), and -mfJt, which is used with
(chiefly human) animatesand forms inclusives, e.g. awa-mfJt 'mother and
her associates' .
A singular form often designatesa plural conception;dependingon the
situationkol can refer to one fish or to many. Thereare also syntacticmeans
of expressingplurality, viz. a noun in the singularfollowed by a verb in the
plural: kol ijat FISH SWIMS-p3pres'fish(es) swim', contrastkol ija FISH
SWIMS.s3pres'a fish swims' andthe examplewith siispfJk-fJit'nightingales'
 MARl 229

cited above.In sum, the categoryof numberis in statu nascendiin Mari, and
is highly pragmatic.
As in many Uralic languages,referents (usually body parts) normally
occurring in pairs are encoded morphologically as singulars, e.g. the
unmarkedmeaningof kit is '(the two) hands/anns'.Disambiguationtherefore
specifiesthe markedmeaning'one hand/ann',which is expressedby means
of the word pel 'half, e.g.pel+kid=an HALF+ARM=adj 'one-anned'.
The second-andthird-person singular personmarkersalsoact asindicators
of definiteness, thus functioning as article surrogates, e.g. uzeS
kuguZan+iidfJr-fm. kuguzan+iidfJr-et mfJgfJra 'he sees a princess
(kuguzan+iidfJr-fm, sA). The princess (kuguzan+iidfJr-et, s2; lit. your
princess)is weeping.'

Verb Inflection
Finite forms. Therearetwo conjugations,calledem andam conjugationsafter
their first-personsingularpresentindicative endings.The precisemechanics
of the developmentof thesetwo inflectional patterns- discerniblein most
forms, both finite and non-finite - havenot beenadequatelyexplained.In all
probability the differenceis dueto the conspiracyof severalfactors,phonetic,
morphological,andfunctional. In this chapter,the membershipof a verb stem
to one conjugationor anotherwill be indicatedby stem-final S: presenceof
this segmentindicates the em-conjugationclass; its absenceindicates the
am-conjugation.
There are three moods: indicative, desiderative, and imperative. The
indicative has a present/future,i.e., non-past,tenseas well as two pasttenses.
The other two moods have no tense oppositions.The indicative and the
desiderativedistinguishthree persons,singular and plural, while the imper-
ative hasonly second-andthird-person(singularandplural) forms.
Thesefinite forms are setout in Table7.5.
The primary differencebetweenthe two pasttensesis neithergrammatical
nor lexical/aspectualbut rather pragmatic.The first pastrefers especiallyto
states and events which the speakerhas personally witnessed,while the
secondpastis moreor lessa recordof what hasbeenor happenedwithout any
emphasison the speaker'sattitudetowardsthe truth valueof the utterance;the
secondpast might be called a 'non-witnessedpast' ([Kangasmaa-]Minn
1960: 116-7). In many contextsthis pragmaticoppositionis neutralized,i.e.
the two tensesareinterchangeable.
The first pastderiveshistorically from the pU past-tensemarkers*-i- (in
the am conjugation) and *-sL (in the em conjugation). The former is
discerniblein Mari only as morphophonemics:stem-suffix sandhi (e.g. pI
presentludfJna, pI first pastlutna), palatalizationof stem-finaln and I (as in
naPe '(s)hetook) and the s2 suffix variant -c (as in toli-oc 'you came').The
secondpast is historically a constructionconsistingof the gerund and the
present-tense forms of ulam 'is', exceptin the third personsingular,wherethe
230 MARl

Table7.5 EasternMari finite verb forms

nalam 'I take' kodem'I leave(tr)'

Indicative
Present/future(non-past)
sl nalam kodem
s2 nalat kodet
s3 nales koda
pI nal~nanal~na kodena
p2 nal~danal~da kodeda
p3 nal~t nal~t kodat
First past
sl nalj~mnalj~m kod~s~m
kod~s~m
s2 nalj~c nalj~c kod~s~ckod~s~c
s3 nalje kod~s kod~s

pI naljna kod~sna
kod~sna
p2 naljda kod~sta kod~sta
p3 nalj~c nalj~c kod~st kod~st

Secondpast
sl nal~namnal~nam kodenam
s2 nal~nat nal~nat kodenat
s3 nal~n nal~n koden
pI nal~nna nal~nna kodenna
p2 nal~nda nal~nda kodenda
p3 nal~n~t nal~n~t koden~t koden~t

Desiderative
sl nalnem kod~nemkod~nem
s2 nalnet kod~net kod~net
s3 nalneze kod~nezekod~neze

pI nalnena kod~nenakod~nena
p2 nalneda kod~nedakod~neda
p3 nalnest kod~nestkod~nest

Imperative
s2 nal kodo
p2 nalza kod~zakod~za

s3 nalze kod~zokod~zo
p3 nal~st nal~st kod~stkod~st

gerundoccursalone.This reflectsthe fact that theseforms (e.g.koden)

nal~n, koden)
were originally nominal predicatesin which the present-tensecopula is
zero.
The third personplural imperativeis the only finite verb form which is
homophonousin both conjugations. In the main, the am conjugation is
characterizedby the occurrenceof consonantstemsand reducedvowel, the
 MARl 231

em conjugationby vowel stems,both full and reduced.The axis-of-discourse

personalendings coincide with the person markers which are attachedto
nouns.The third-personendingsof the indicativeparadigmsdiffer from those
seenin the noun, havingdevelopedfrom verbal nounsor tensestems.
The primary auxiliary verbs ulam 'is' and lijam 'becomes'have slightly
divergentparadigmsand functions. Both are conjugatedas regularam verbs
(but note the u - ~ metaphonyin the flrst past of ulam), and there is also a
specials3 presentform ulo usedin existentialllocaland possessiveconstruc-
tions, and a specialnominal form ulmasusedfor the s3 in the secondpast.
The two verbs are interchangeablein many contexts,but the presentof lijam
often refers to the future and sometimesexpresseschange.Besidestheir use
in the presenttense,ulam functions chiefly as an auxiliary and lijam chiefly
as an independentverb, but there are deviations in various dialects. The
'regular' s3 present-tense form uleSis textually quite rare (Kangasmaa-Minn
1969b: 7). The simple (flnite) forms of ulam and lijam are set out in Table
7.6.
Another auxiliary is the negative verb 0-, which has a defective and
heterogeneous paradigm.In the presenttenseit has the present-tense marker
-k (- -g-) andconjugateslike an am verb; in the first pastits stemis ;J.. (-zero)
and it conjugateslike an em verb. The negativesecondpast is a compound
form, consistingof the gerundplus forms built from the present-tense of the
negative verb plus thethe~l variant of the verb 'is'. The negative verbnever
standsalone but always togetherwith the connegativeof the main (lexical)
verb. The connegativeis always homophonouswith the s2 imperative.The
negativecounterpartof ulo 'exists, it y a' is uke, a noun of perhapsTurkic
origin which meantsomethinglike 'non-existence';this form also functions
as a sentence-substitute in the answer'no'.
There is no formal categoryof passivevoice in Mari. Passivemeaningis
achieved by means of syntactic constructions and certain derivational
elements.
Non-finite forms. Verbal nouns play an important role in Mari sentence
structure.They areformed with variousderivationalsufflxes, often combined
with caseendings.Actually their statusas membersof the verb paradigmis

Table7.6 EasternMari primary auxiliariesulam 'is' andlijam 'is, becomes'

Present First past Secondpast Desiderative Imperative

sl ulam,lijam ulmas,ulda, ulnem,lijnemof ulnem,lijnem

s2 ulat, lijat ulo,lijes ulmas,ulda, ulnet, lijnet -,Ii
s3 ulo,lijes ulo,lijes ulmas,ulda, ulneze,lijneze -,lize
pI ulna, lij~na
ulna, ulmas,ulda, ulnena,lijnena ulnena,lijnena
p2 ulda, lij~da
ulda, ulmas,ulda, ulneda,lijneda ulneda,lijneda -,lize
p3 ulo,lijes ulo,lijes ulnest,lijnest ulnest,lijnest -,lize
232 MARl

Table7.7 EasternMad negativeverb

Present First past Secondpast Desiderative Imperative

(= prohibitive)

sl +C
o(g~)moges (~)s~m+ogesC G+om~l
G+om~l ~nem+C C
s2 C+
o(g~)toges +
(~)s~coges C G +G+om~l
ot~l ~net+ C it+C
s3 ok - oges+ C ~S+C G+og~l
G+om~l ~neZ+C C ~nze~n~st++C
C
pI o(g~)naoges+ C oges
~sna+ C G+onal ~nena+ C
~nena+
p2 o(g~)daoges+ C +C
~staoges G+odal ~neda+ C
~neda+ ida+C
p3 oges C
og~t+ ~st+C + C G +og~t~lG+om~l ~nest~nest + C ~n~st+ ~n~st+ C

Note: C =connegative;G =gerund.

dependenton their syntacticroles, sincemost non-finite verb forms may also

occuras ordinarydeverbalnominalderivatives.
Verbal nouns representembeddedsentencesand as such may take their
own arguments,both objectsand adverbials.The verbal noun is fitted to the
structure of the matrix sentenceand occurs as a noun dependenton the
governing (matrix) verb. The number of forms belonging to this category
variesfrom dialect to dialect; EM recognizestwo infinitives, four participles,
andfive gerunds.Theseare:

la Infinitive =as: kodas 'to leave', nalas 'to take'. This form is homo-
phonousin the two conjugations.It is thereforeunfortunatethat it is the
citation form, sincethe conjugationmustthenbe given separately.
1 b Necessiveinfinitive =man: kod~an 'mustleave',nalman 'musttake'.
2a Active participleconjugations.
=m~: kod~o 'left; nalse 'taker'.
'leaver',
2b Passiveparticiple conjugations.
=m~: kod~o 'left; act of leaving', nalme 'taken; act
of taking'.
2c Futureparticiple =sas:
conjugations.
kod~sas 'which mustbe left' , nalsas'which must
be taken'.
2d Negativeparticiple =d~~:conjugations.
kodM-~mo 'not left' , nald~enald~e 'not taken'.
3a Mfirmative gerund-n: koden'leaving',nal~nald~e 'taking'.
3b Negativegerund=de: kodM-e 'without leaving',nalde 'without taking'.
conjugations.
3c Simultaneousgerund =s~la: conjugations.kod~s~la 'while leaving', nals~la'while 'while
taking'.
3d Anterior gerund =mek(~):conjugations.
kod~eke 'having left', nalmeke 'having
taken'.
3e Posterior gerund =mdke: conjugations.
kod~dke 'until leaving', nalmdke 'until
taking'.

The morphological makeup of many of these non-finite forms is fairly

transparent,and still other deverbalnouns might be invoked as membersof
this category,e.g. nald;Jmas~n
'without 'without taking' might be analysedas the stem
 MARl 233

'takes' plus a string of semi-productivenominalizers(mfJ, =as) and case

suffix variants(caritive, genitive).

Derivation
Derivational suffixes serveto make new words from words alreadyexisting
in the language.Somesuffixes attachonly to nominal stems,othersonly to
verb stems,but the greatestnumberattachesto both. Derivation may remain
within a part-of-speechcategory(noun>noun,verb> verb) or crossfrom one
categoryinto another(noun>adjective,noun>verb, verb> noun,etc.).
We may thus distinguishsix types of derivationalsuffix accordingto the
partsof speechbeforeandafter the derivationalprocess(N = nominal):

1 NN > N, e.g. =mfJ in pukSer=me HAZEL=mS 'hazelbush' andpalfJ=me

KNOWS=mS 'known';
2 N > N, e.g.lfJin lam=1O NAME=IS'famous';
3 V> N, e.g. =em in kuc=em SEIZES=em'handle';
4 NN > V, e.g. ust=al- BELT=al- 'puts a belt on X', mur=al- SINGS=al-
'humsa tune';
5 N > V, e.g.joskar=gfJ.. RED=gS-'becomesred';
6 V> V, e.g. kalas=kalfJ- TALKS=kaIS- 'chats'.

Derivational suffixes which do not alter the grammaticalstatusof a word

have insteada modifying influence. Modifying nominal meaningsinclude
collectives, abstracts, and diminutives, e.g. puks 'hazelnut' plus =er >
puks=er 'hazel (tree)" soVgo 'old' plus =lfJk > soVgfJ=lfJk 'old age'.
Modifiers within the verb categoryindicatethe Aktionsart, or quality of the
event,producingfrequentative,momentaneous, or durativederivates,e.g.pu-
'gives' plus =edfJ > pu=edfJ.. 'distributes', kustfJ.. 'dances' plus =altfJ.. >
kust=altfJ.. 'dances(once)'. As types (1) and (4) above illustrate, the same
suffix may operateas both modifier within a categoryand transformerfrom
onecategoryto another.

Noun-forming Suffixes
The textually most frequentnoun-formingsuffixes have alreadybeenlisted
in connectionwith the non-finite verb. All also form nouns from nominal
stems,at leastto a degree:

=as as an infinitive suffix enjoys 100 per cent productivity. It also occursin
combinationwith other suffixes, e.g. *=sS=as> =sas (future participle)
and *mS=as> =mas, the nomen action is : nal=mas 'the act of taking'. It
is productive and thus lexically frequent added to nouns, e.g. parnia
'finger' > parnias 'thimble', sorfJk 'sheep'> sorfJkas 'lamb', tfJlze 'moon,
month' > tfJlzas 'onemonthlong'.
=sfJ addedto verb stemsforms the active participle(seeabove).The same(or
234 MARl

a homophonous)suffix occurs (with an -m- element)in the formation of

ordinal numerals,e.g. nfJl 'four' > nfJlfJrnse 'fourth'.
=mfJ added to verbs makes passive participles and names of actions.
Occasionaldenominalderivatesare found referring to individuals within
a collective,e.g.pukSerme 'hazelbush'.
=fJs forms nouns from both nominal and verb stems,e.g. kut 'long', kutfJs
'length',peled- 'blooms',peled=fJs 'flower'.

Among derivational suffixes which operate only within the nominal

categoryare thosewhich form adjectivesfrom nounsor vice versa.The most
productiveof the former type is =an, e.g. lum 'name' > liim=an 'having X
name',juk 'voice' > juk=an 'loud'. Another productiveadjective-formantis
=lfJ, originally borrowed from Chuvashtogetherwith its stem, e.g. tamle
'sweet' (cf. tam 'taste'),but now productivewith native words as well, e.g.
lum=fij 'famous'.The suffix =sfJ is unusualin that it canbe addedto adverbs
or nouns which are already inflected for case. It thus may render the
equivalentsof spatialandtemporalsubordinateclauses,e.g. wud-fJstfJ=SD 'the
one which is in the water'; there is no parallel in Mordva. Purely denominal
suffixes which form nouns from adjectives are rarer, but there is non-
productive =t (e.g. kelge 'deep', kelgfJ=t 'depth'); comparethe =fJs men-
tioned just above, and the weakly productive =er, which forms collectives
and phytonyms(see p. 233), may be seen in a formation such as tfJgfxl=er
'shrub' (tfJgfxle 'small, delicate') or the neologism toster 'museum' (tosto
'old').
Collective-formingsuffixes, in particular, abound,and may even be used
in combination,e.g. koz 'spruce'> koz=er 'sprucegrove',koz=la, koz=er=la
'forest', similarly puneD 'pine', > pune=er 'pine grove' > pune=er=la 'pine
forest'; this =la is historically identical with the pluralizer mentionedabove,
p.228.
Thereare also a few adjective-to-adjectivederivations.Thesemay have a
syntactic function, e.g. =(g)e, which forms predicativesfrom colour words,
e.g. sar 'yellow', sar=e 'yellow (pred)',joskar 'red (attrib)"joskar=ge 'red
(pred)' as in sar saska, joskar saska 'yellow flower(s), red flower(s)' and
saska sare, joskarge 'flowers are yellow and red', or they merely mitigate
semantically,e.g. =alge in kande 'blue' > kand=alge 'blueish',and =rak in
joskar=gfJ=rak 'reddish'.

Verb-forming Suffixes
Most verb-forming suffixes belongunswervinglyto one of the two conjuga-
tions; deverbal verb formation may therefore entail a switch from one
conjugation to another, e.g. the em-conjugationverb tuganfJ.. 'wears out
(intr)' becomesan am-conjugationverb whenmitigating/momentaneous =al-
is added:tugan=al- 'wearsout a little'. There are in fact about a dozenverb
pairs of which the transitive memberbelongsto the em-conjugationand its
 MARl 235

intransitive counterpartbelongsto the am-conjugation,e.g. kod-em 'I leave

(tr)', kod-am 'I remain',woz-em'I drop, let fall', woz-am'I fall'. The origin
of this duality is uncertain,andthe degreeto which it might reflect an ancient
distinction is highly controversial.Whateverits pedigree,it is also seen in
some deverbal derivational suffixes, e.g. reflexive/passive-forming=alt-,
which produces am-conjugationverbs like sarn=alt-eS 'is remembered,
comes to mind', and momentaneous-forming =alt~,
which produces
em-conjugationverbs like sarn=alt-a '(s)he remembersit, calls it to mind,
mentionsit'. Whateverthe original differencebetweenthe two conjugations,
it is not to be soughtin transitivity alone,for thereis amplecounter-evidence,
e.g. kol-am 'I hear' andkol-em 'I die'.
Among suffixes which form verbsfrom both nominal andverbsare:

=d?J- (=d~
=t~ =d?J- to the right of nasals).This forms factitives from nominals,e.g.
lam 'name' plus =d?J- > lam=d~ 'names',joskar 'red' plus =t?J- >
joskar=t?J- 'makesX red'. As a deverbalsuffix it changesintransitive and
reflexive verbs into transitive ones, and builds causatives,e.g. pur-
'enters' > pur=t?J- 'lets in'. It is extremely common to the right of the
productive translative denominal verb suffix =em-, e.g. poro 'good' >
por=em- 'becomesbetter' > por=em=d?J- 'cajoles, wins over (earlier:
heals),.
'makes
=~kt~ also makesfactitives from buildscold' > k~lm~=kt~k~lm~=kt~
nominals, e.g. k~lme
'makes X cold'. As a deverbal suffix it builds causatives,e.g. soc- 'is
born' > soc=fJkt~- 'makes 'gives birth to X', builds
ku~ 'weaves'> ku=~kt~
builds 'makes
X weave'.
=al- builds denominalinstructives,e.g. upsa 'cap' > ups=al- 'putson cap' (>
ups=al-t~ ups=al-t~ 'puts cap on X'). Useddeverballythis suffix has a diminutive,
hypocoristic meaning. In keeping with this type of affect, it may occur
reduplicated,e.g. mur~ mur~ 'sings' > mur=a[-, mur=al=a[- 'humsto oneself,
singsa little'.

Suffixes which createverbsexclusivelyfrom nominalsaremany,but most

areno longerproductive.The chief productivesuffixes are:

=em-, mentionedabovein connectionwith =t~=d~), (- =d~),

=d~), e.g. molo 'other'
> mo[=em- 'changes(intr)';
dual', as in kokte=lan~
=lan~, 'becomes 'doubts, is
of two minds', from'becomes
kok(t~=) 'two,
dual', cf. 'becomes
kokt~te 'in twain';
=ay-, as in pu=ay- 'grows stiff/numb', from pu 'wood, tree', kumd=ay-
'widens(intr)' , from kumda'wide'.

There is also a non-productivedual', (- =k~),

=g~ 'becomes which may be related
historically with =ay-, e.g.
'becomes
kud~r 'curl(s)' > kud~r=g~
'becomes'becomescurly'. The
formation of a verb such as suld~ray-
'becomes'becomeswinged' 'becomes
(su[d~r 'wing',
236 MARl

suld~r=an =a1]- 'winged') is not entirely clear: is the 1] at the end of this stem a
variantof =a1]- and/or=g~? =a1]-
Among the suffixes which create verbs exclusively from nominals,
reflexive/passive-forming=alt- and momentaneous-forming=alt~ longer were
mentioned above. To these we may add the still somewhat productive
whether Note the conjugational ambiguity; it is difficult to determine
=(e)St(~)-.
whether we are dealing here with one suffix or two, since the meaningis
sometimesfrequentative,e.g. longer
'chops;
ru~- 'chops; fells (trees)' > rU=eSt- 'chops
repeatedly' and sometimes(though less often) momentaneous,e.g. torg-
'jumps' > tor=st~longer 'jumps once/suddenly'.Numerousfrequentativesarebuilt
with the no longer productive suffix =ed~, nal=ed~ e.g. nal=ed~ nal=ed~ 'keeps taking'
(from nal- 'takes'); the meaning of these derivates,as with frequentatives
generally, is often connectedwith multiple object complements,e.g. object
pu~
'gives' > pu=ed~longer 'givesto many,distributes'.
Enclitics
If we restrict to the class of clitics only those elementswhich lack any
grammaticalor semanticimplications, we are left with very few examples.
Only --at and --ak qualify; the former is especiallyfrequent.Examples: object
ikt~t
je1]--at uke 'thereis not evenone person', tud~m--ak tud~m--akobject
uz~ PRO.s3-acc--enc
SEES.slpastl'It was (s)he that I saw'. The enclitics act as emphasizersand
can be attachedto verbs as well as to nominals, e.g. kol=en--at kolt-~n~t object
DIES=ger--encFINISHES-p3past2'they certainlydid die'.
A few case suffixes, e.g. comitative -ge and modal -la, occasionally
function as enclitics, and the comparative/adjective-formant =rak may also
behavein this way. The temporal/modalconjunction independent
g~n (- k~n) 'if, when;
(evidential),hoversbetweenbeing an enclitic andan independentword.

Syntax
A great many gaps in the morphologicalsystemof Mari are conspicuously
smoothed over with the help of syntactic constructions.It is therefore
difficult, and perhapscounterproductive,to keep the two categoriesstrictly
apart.The problemis clearfrom both the nominal andverbal side.

Noun Phrases
Noun phrasesconsistof a single noun, optionally precededby anothernoun
or by an attributive phrase(seep. 237). The noun may also be followed by a
postposition;this constructionis treatedin the next section.

PostpositionalPhrases
Theseconsistof a noun or pronounfollowed by a postposition.Postpositions
arenounswith local meaningsanddefectiveparadigms,the forms oftenbeing
built with non-productivecase suffixes. The postpositionalphrasehas the
syntactic function of a case, the adjunct noun acting as a stem and the
 MARl 237

postposition(local noun) acting as a caseending. Adjunct nouns - which

could more accurately be termed conjuncts - normally stand in the
nominative, even if they refer to animates;pronouns are in the genitive.
Examples:jeg-ze-wlak kokla-stePERSON-s3-plurINTERSTICE-ine'among
the people', tudfJ-n nergen PRO.s3-genABOUT 'about himlher/it'. The
textually most frequentpostpositions,gx and the variousforms den(e),dee,
deran, dek(e)- all fossilized inflections of a noun *ter, which survivesas an
independentword tur 'edge, border' - were presentedin the morphology
section,p. 227 (Kangasmaa-Minn1966: 38).
Attributive Phrases
These may be simple, i.e. consist simply of a noun head precededby an
adjective (lumlO jeg 'famous person') or by an adjective which is itself
modified by an adverb (peS lUmlO jeg 'a very famous person').Adjectival
attributesare not declined; there is often no differencebetweena noun and
an adjectivalattribute,e.g. in mari kalfJk 'Mari people' man designatesboth
individual ('a Mari') andquality ('pertainingto Mari').
As indicated in Morphology, p. 226 the genitive is used to mark the
animateownerin possessive constructions.The possession is thenusually,but
not necessarily, marked with the appropriate person marker: kugM.a-n
udfJr(-zo) 'the Tsar's daughter'.An inanimatenoun adjunct appearsin the
nominative,andthereis no marking on the headnoun (kurfJkjol HILL FOOT
'foot of the hill'). Animate noun adjunctsalso occur in the nominative,and
the use of the genitive to mark inanimatenoun adjunctshas begunto spread
in the present-daylanguage,but the original distinction is still discernible
(Kangasmaa-Minn1966: 255).
Attributive phrasesmay also be complex,i.e. consistof two or more noun
phrasesjoined. For example,the two noun phrasesmari kalfJk 'Mari people'
and lUmlO jeg 'famousperson'may be joined by placing the headof the fIrst,
or subordinate,noun phrasein the genitive and attaching the third-person
singular suffix to the head of the secondphrase,viz. mari kalfJk-fJn lUmlO
jeg-ze 'a famous personof the Mari people'.Attributive and postpositional
phrasescan co-occur:Mari kalfJk-fJn lUmlO jeg-ze-wlakkokla-ste'amongthe
famouspersonsof the Mari people'.
Comparativeand SuperlativePhrases
The comparativephraseis specialin that in it the adjunctnoun is in a local
case:the bestedmemberof the comparisonstandsin one of the sourcecases,
separative-dec or ablative -lee, dependingon dialect. The adjective which
expressesthe quality by which the comparisonis mademay take the suffIx
=rak, but is just as often unmarked.The literary languagefavours -dee; an
example is therefore wastar kue-dee kugu(=rak) MAPLE BIRCH-sep
BIG(=cfv) 'the mapleis taller thanthe birch'.
The superlativeis usually formed with en, a loan from Tatar: en alama
'worst' (alama 'bad').
238 MARl

Verb Phrases
Thesecontaina finite verb fonn in connectionwith anotherverb fonn, either
finite or non-finite.

CompoundTenseForms
The finite verb paradigmis augmentedby two typesof compoundverb fonns.
One type, built with the negative verb, was presentedin the section on
morphology. In this section we turn to the other type, namely compound
tensesbuilt with the auxiliary ul- 'is'.
Therearefour compoundtenses,the first andsecondimperfectandthe first
and secondperfect.All four tensesrefer to the past.Thesefour tensesare the
productsof combining the·presentand second-pastfonns of the main verb
with the first andsecondpast-tensefonns ofthe auxiliary:

Auxiliary verb

Past1 Past2

Main verb Present Imperfect1 Imperfect2

Past2 Perfect1 Perfect2

Both the imperfectstranslatein the sameway, e.g. imperfect 1 kod~ functional

~lje,
ulmas are both English 'I was leaving'. The functional
imperfect 2 functional
kod~
differencebetweenthe two constructionsparallelsthe fonnal difference:first
v. secondpastin the auxiliary. The use of thesetwo tensesis the sameas the
use of the first and second past in general, i.e. they are more or less
interchangeable,but the first pastrefers especiallyto eventsthe speakerhas
personallyexperienced.The sameis true of the two perfecttenses:koden been~zie
(perfect 1) and koden ulmas (perfect 2) both translate as '(s)he had been
leaving', and the choiceof the auxiliary is more or lessa pragmaticone.
The Mari compoundtensesarebuilt on two subsequent predications,as are
the correspondingcategoriesin other languages.But in Mari there are
fonnally two kinds of compoundtense,built on two different principles. In
the first, person-markingoccurson the auxiliary. Recall thatthe secondpast
is itself a complex tense,at leasthistorically: it consistsof the main verb in
the gerundplus the present-tensepersonalfonns of the verb 'is', ul-. In some
dialectsthe ingredientsare still synchronicallytransparentin the plural, e.g.
built ulna correspondingto literary nal~na,built both 'we havetaken'.
nal~
Theperiphrasticimperfectsandperfectsarebuilt on thereverse principle, i.e.
person-markingoccurson the main verb, not on the auxiliary. The main finite
verb, on which are encodedpersonand tense,is the underlying predication,
while the auxiliary merelyrelatesthe eventto a point in time,that is to the past.
 MARl 239

Thenegativeperiphrasticimperfectsandperfectsarea sequenceof threeverbs:

personis markedon the negative verb,time (presentor secondpast)is marked
on anotherverb- thesetwo takentogetherform thelower predication- andthe
third verb, the auxiliary, standsfor the upperpredicationand relatesthe verb
processto presentor past,e.g. first imperfectok kodo ~lj e NEG.VERB -s3(pres)
LEA VES-connegIS-(s3)past'(s)he was not leaving', secondperfectnal=~nnal=~n
o-na-l ul=masTAKES=gerNEG.VERB-p1past(2)IS=nomen.actionis'wehad
not been taking'. The samesystemoperatesin the desiderativemood, e.g.
secondimperfectdesiderativekod~ne-da ul=mas 'you (plur) would like to be
desiderative
leaving',negativefirst imperfectdesiderative~ne-znal nal '(s)he
~lje '(s)hewould not
like to betaking'. Thedesiderativelacksperiphrasticperfects.
Indicative periphrasticperfectsand imperfectssometimeshave decidedly
moralizing overtones,e.g. secondimperfectivenal~am ulmasulmas'I shouldhave
beentaking' (Kangasmaa-Minn1976).

AdverbialComplementsand Converbs
Verb phrasesconsistingof a finite verb togetherwith a verbal nominal may
be divided into severalsubgroupsaccordingto both the nominal form used
and to the inner cohesionof the construction.
The infinitive in =as is found in causalcontexts;the finite verb usually
designates locomotion or change, e.g. mal=as woz-eS' SLEEPS-inf
DESCENDS-s3pres 'goesto bed',kol kuc=askaj-a FISH SEIZES-infGOES-
s3pres'(s)hegoesto catchfish'. The infinitive also servesas complementto a
few modal verbs such as 'tries', toc~ 'tries', tiilJal- 'begins', lij- 'becomes,is
permitted',kiil- 'is necessary',e.g. kol kuc=astoc-a '(s)hetries to catchfish',
mur=askust=astiilJal-en~t necessary', 'theybeganto sing(and)dance',nal=aslij-eS'? 'Isit
permittedtotake(this)?',miind~rmod=askaj=aso-g-~tkiistOFARPLAYSCOMMANDS-conneg'they don't allow one
-inf
GOES-infNEG.VERB-pres-p3COMMANDS-conneg'they don't allow one
(me, us) to go far to play',tun~kt=as tud~m tun~kt=astun~kt=as tun~kt=as kiil-eS'PRO.s3-acc
TEACHES-inf
IS.NECESSARY-s3pres '(s)hemustbetaught'.
The gerundin =n is widely usedin verb phrases.It is the usualform of the
verb complementwith a variety of verbshaving vaguesemanticcontentsuch
ascomplement able',onc~'looks,'looks, tries', kert- 'is able' as in mo-m ~ste=nmost-most-
etWHAT?-accDOES=gerIS.ABLE-s2pres'what canyou do?',ojle=n onc-a
'(s)hetries to speak',tud~m Mte=n Mte=n kert-am 'I cando that'.
But the most important function of the n-gerundis its uses in converb
constructions.Theseconsistof one or more gerundsfollowed by a finite verb
form; the semanticcoreof theconstructionmayresidein a converb,in thefinite
verb,or in both.Thefollowing morecommontypesmaybedistinguished:

1 The eventindicatedby the gerundis simultaneouswith that of the finite

verb, e.g. oksa mure=n tol-eS' ta siiske=n kaj-a MONEY SINGS=ger
COMES-s3presAND WHISTLES=ger GOES-s3pres'money comes
singing andgoeswhistling'.
240 MARl

2 The event indicatedby the gerundoccursbefore that of the finite verb,

e.g. nuno male=n kfmi el-f1t 'after sleeping,they get up'. Note that the
order of the gerundsneednot necessarilymatch theorder of events,e.g.
maska kaza-m kuce=n tol-f1n kaj-a BEAR GOAT-acc SEIZES=ger
COMES=gerGOES-s3pres'the bearcomes,catchesthe goat, andgoes'.
3 The converb constructionmay convey aspectualinformation about the
event,stressingits duration or its punctuality.In suchcasesthe semantic
contentis borneby the gerund,while the aspectis specifiedby the lexical
choice of the finite verb. Verbs used for this purpose are somewhat
colourlessandsimplein meaning,andtheir numberis limited - theremay
be as manyasforty. Imperfective,continuousactionis conveyedby finite
verbs such as sog?/- 'stands', e.g. ojf1rtemaltse wer-fJm nal=f1n sog-a
DISTINGUISHED PLACE-accusative TAKES=ger STANDS-s3pres
'will always have a distinguishedplace'. Various kinds of perfective
aspectare expressedby finite verbs designatingactions such as begin-
ning, finishing, finding, leaving, sending,readying,reaching,and going
or coming, e.g. kas tene ola-ske tol=fm su-eSEVENING AT CITY-ill
COMES=gerREACHES-s3pres'(s)he will arrive in town in the eve-
ning', joralt( = f1n) kaj-enf1tta kole=n--at kolt-enf1tFALLS.DOWN(=ger)
GOES-p3past2AND DIES=ger--encSENDS-p3past2'they fell down
anddied' (the barestemof am-verbs,asjoralt- here,may function as the
gerundin converbconstructions)(Bartens1979: 143-9).

Simple Sentences
In Mari, predicationis basedon a nounor a verb.

Noun-basedPredication
Nominal predication sets two nouns against one another. The borderline
betweenthem is markedby the copula, which in the presentindicative third
personsingularis zero; in otherpersons,tenses,andmoods,verb forms based
on both ul- 'is' andlij- 'becomes'occur. Examples:soygoul-at 'you are old',
kuwa soygo 'the woman is old', kuwa soygo ul=mas 'the woman was old'.
It is not the absence(zero) of the copulathat makesthe sentencea nominal
one; the nominality lies in the fact that the copulais merely a peg on which
to hangperson,tense,andmoodmarkers.
On the other handlij-eS (s3pres)doesoccur; but kuwa soygolij-eS is 'the
woman grows old'. This approachrecognizesanother, dynamic type of
nominal sentencealongside the static statement'the woman is old'. The
systemoftwo verbs 'to be' is discerniblein mostFinno-Ugric languages,but
the functions of the two verbshavebecometangled;in Mari, too, they are not
clearly differentiated,and there is much dialectal variation. In fact, lij- may
be found in static sentences,as well, especiallyin thosereferring to the past;
recall that standardEM has second-pastforms of lij- in suppletionfor the
 MARl 241

defective paradigmof ul-. On the other hand, ul- cannot replace lij- when
notionsof futurity or changeareinvolved (Kangasmaa-Minn1969b:6).

Verb-basedPredication
When built on a verb form the sentence structure presupposesthe
co-occurrence,actual or implied, of a noun in a local caseform. There are
severalsubtypes:
Thepossessive sentenceis mostakin to nominal-based predication.It is built
on a verb characteristicallyfound in localizing connectionsbut occurring
without a local adverbial.The verb mostoftenis ul- 'is, exists',with its special
third-person singular presentindicative form ulo and uke as its negative
counterpart;the verb always standsin the third person.The possessoris an
animatenounin thegenitivecase,andthething possessed usuallyhastheperson
marker implicated by the genitive, e.g. m~~ imni kok imni ulo PRO.s1gen
ud~r-em
TWO DAUGHTER-s1EXISTS.s3pres'I havetwo daughters',erg~n imni imniimni
~ze
ukeBOY-genHORSE-s3NOT.EXIST.s3pres'the boy doesnot havea horse',
memnanul-masport-na 'we had a house',lud=so-wlak-~knliga-zeknliga-ze(- knli-
ga-st) uke 'thereadersdo not havea book'; asthe lastexampleshows,the third
personsingularmarker(-ze)oftenappearsinsteadof thethird personplural (-st)
if theownerspossessonething together,or all of themonething each.
A possessiverelation is not necessarilystatic; it can involve loss and gain,
as well. An animategenitive is found in constructionssuchas tud~njum~zejum~zejum~ze
su-eS PRO.s3-genTHIRST-s3 ARRIVES-s3pres '(s)he becomesthirsty',
(wo)man
nun~n hislherPRO.p3-genIS.BORN-s3past2DAUGHTER-s3 'a
soc-~nlost hislher
ud~r-ze
girl was born to them',jelJ-~n (wo)man oksa-zePERSON-genGETS.LOST-
jom-~nlost hislher
s3past2MONEY-s3 'a (wo)manlost hislhermoney'.
Static possessiveconstructionsare reminiscentof nominal-basedpredica-
tions inasmuchas they may occurwithout the verb 'is': erg~n horse'.horse'. when
imnl~ze,
representinga sentence,may be translatedas 'the boy has a horse'. The
possessor nounneednot bepresent;thepersonmarker,encodingthepossessor,
thenachievessyntacticsignificance:imnlhorse', ~ze ulo '(s)hehasa horse',Ud~r-emhorse',
uke 'I do not havea daughter',jum-emsu-eS'I becomethirsty' .
Locationalpredication consistsof a verb and a local adverbial,usually a
noun in one of the local casesor followed by a postposition.The verb may
be transitive or intransitive; the occurrenceof a direct object doesnot affect
the basic locational structureof the utterance.A static verb presupposesan
adverbialin the inessiveor locative, e.g. Mari kal~k-~ kal~k-~ lUm=1O jelJ-ze-wlak
kokla-ste... T. Jewsejew... sog-a, 'T. Jewsejewstandsamong the famous
personsof the Mari people'in which sog-a 'stands',as a static verb, requires
an adverbialin a static caseform, herethe postpositionalphrasejelJ-ze-wlak
kokla-ste'amongpersons',with the inessiveending -ste. An examplewith a
static transitive verb is umbal-ne
tud~m umbal-neumbal-nekij~=kt-em
m(~)lande umbal-nePRO.s3-acc
GROUND SURFACE-at LIES=caus-s1pres'I keep himlher lying on the
ground'. As in possessiveconstructions,the third-person singular present
242 MARl

indicativeulo may be omittedin simple statementssuchas iidfJr portfJ-stO 'the

girl is in the house';such constructionsdo not representnominal sentences,
sincean adverbialmodifier is present.
A dynamic predicationrequiresadverbialsin separativeand/orin lative or
illative cases,or in postpositionalphraseswith appropriate,i.e. goal or source,
functionalendings.The verb designatesmotion of somekind, e.g. (separative)
surt-so gxfln lekt=fln kajat HOUSE-s3 FROM EXITS=ger GOES-p3pres
'they leave his house'; (lative) ko-n ter-d sine-at WHO?-genSLEIGH-Iat
SITS.DOWN-s2pres'into whosesleigh do you sit?'; (illative) koz piigelme
wol=en woz-d ik meragfln wuj-fJskfJ-zo SPRUCECONE DESCENDS=ger
FALLS.s3presONE HARE-gen HEAD-ill-s3 'a spruce-cone,falling, hits a
hare on the head', me eotl-ena puseggfJ-m fJlfJksfJ-s PRO.pl ESTEEMS.pl
TREE-accLIVING-ill 'we considera tree a living thing'; (illative) wara pop
memnan ergfJ-lan liim-fJm pu-a 'thenthe priestgives a nameto our son'.

Complex Sentences
Apart from relatively recentborrowingsMarl hasfew conjunctions;coordina-
tion and subordinationof simple finite sentencesis accordinglyrare. Instead,
utterancesarelinked togetherwith the help of non-finite verb forms.

Nominalization
This is the processwherebya verb is converted,by derivationalmorphology,
into oneof the non-finite forms introducedabove,p. 232. This non-finite verb
form, togetherwith its arguments,is thenembeddedinto the frameworkof the
matrix sentence,and anyelementswhich the two sentencesshareare deleted.
We have already met with an example of a gerund serving as a simple
embedding,in which the subjectis the samein both embeddedand matrix
sentences,viz. oksa mur=en tol-d 'moneycomessinging', a compressionof
oksa mur-a, oksa tol-d 'moneysings,moneycomes'. Evena nominalsentence
canbe nominalized:mfJj-fJn mo jegem( -fJm) sinefJm-et su-d--gfln PRO.sl-gen
WHAT? PERSON-si-accKNOWS=pass.part-s2ARRIVES-s3pres--IF 'if
you wantto know whatkind of personI am' showsa nominalpredication(mflni
mo jeg 'I amwhatkind of person')embedded asadirectobject(- fJm, optionalto
the right of s I -em-) into the matrix sentencesine= fJm-et su-d--gfJn 'IF YOUR
KNOWING IS WANTED» if you wantto know'.
A nominalizedpossessivephraseis a genitive attribute with its head; the
relation is transparentsince the samecase- the genitive - occurs in both
constructions:embeddingergfJ-n ulo imni fJ-ze 'the boy hasa horse'into imnie
kudal-d 'the horse is running' gives ergfJ-n imnjfJ-ze kudal-d 'the boy's
horseis running'.
Locational embeddingsare the most complex becausemany functional
elements may be involved: subject, object, one or more adverbials plus
adjectival attributes and modal adjuncts. The subject of the embedded
sentencestandsin the nominativeor the genitive,the objectin the nominative
 MARl 243

or the accusative,andthe local adverbialsin the appropriatecaseforms. All the

non-finite forms mentionedin the morphologysectionmay occur. Examples:
kol=sas-em
kug~za kaj=enkaj=en sog-a OLD.MAN MIND-s3 GOES=gerSTANDS-
us~ze
s3pres'the old manstandsstupefied',p~r~s
kol=sas-em
uz-~n maska-n ka tarw~/t~=m~m
kol=sas-em kol=sas-em
CAT SEES-s3past2 BEAR-gen STONE MOVES=pass.part-acc'THE CAT
SAW THE BEAR'S STONE-MOVING »> the cat saw the bear move the
stone', mozx sine-et
kol=sas-em
t~j kol=sas-em
m~j-fm kusan kol=sas-emtozo MAYBE PRO.s2
KNOWS-s2presPRO.sl-genWHERE?DIES=fut.part-slALSO 'maybeyou
know where I'm going to die, as well?', er-Ia pazar-M mij=d~=mas-Ia­'maybe
-rak lij=fm ojl=en kodo MORNING-mod MARKET-Iat GOES=car=
nomen.actionis-mod--enc IS=ger SPEAKS=gerREMAINS-s3pastl '(S)HE
REMAINED SAYING BEING SOMEHOW-WITHOUT-GOINGTO MAR-
KET NEXT-MORNING »> '(s)he persistedin saying that (s)he would
somehownot begoingto marketthenextmorning' (Kangasmaa- Minn 1969c).

Co-ordination and Subordination

Besides norninalization and embedding, Mari also uses the locational
postpositionden as a co-ordinatorbetweenequal members,as in the riddle
pire den maskawasone-atWOLF AT BEAR MUTUALLY LOOKS-p3pres
'the wolf andthe bearlook at oneanother'. Sincethe verb one-atis in the third
personplural we may takeden hereto be a conjunctiveelement.(The answer
to the riddle is torza den opsa 'window anddoor').
An enclitic can sometimeslink two co-ordinatedclauses,e.g. ik kana kana
m~j
eodra-smij-~s~--ak sometimes
peSsukopir~msukouz-~
ONE ONE TIME PRO.sl FOREST-
lat/ill GOES-slpastl--encVERY MANY WOLF-acc SEES-slpastl'once
when I went into the forest I saw many wolves'. An encliticenclitic
--k~n -enclitic
--g~n
added to the finite verb at the end of an utterancemarks a temporal or
conditional clause: meraIJ waske osem-eS--kfm,lum waske tol-eS HARE
EARLY WHITENS-s3pres-encSNOW EARLY COMES-s3pres'if the hare
becomeswhite early, therewill be an early snow'.
Finite relative clausesdo occur in Mari; they usually refer to a determi-
native pronounin the main clause,as in ko deneil-et, tud~n
dean deankoNs-~ dean
pog-et WHO AT LIVES-s2pres, PRO.s3-genFROM CHARACTER-acc
GATHERS-s2pres'you get your characterfrom the personwith whom you
live'. This type of relative clausealso appearsin ratherearly translations,e.g.
in that of the Gospel accordingto St Matthew, edited by J.F Wiedemannin
a Hill (= Western)Mari dialect: mar-Ian kej-enetsedii ketsii jakte, kuda-ze-n
Noi pur=en pos-skaMAN-dat GOES-p3past2 THAT DAY UNTIL, WHICH-
s3-loc NOAH ENTERS-s3past2 ARK-ill 'they married(= went to man) until
the day on which Noahenteredthe Ark'.

Lexicon
Three chronologicallayers are clearly discerniblein the Mari lexicon. The
basicvocabularyis Uralic or Finno-Ugric in origin; this is followed by a thick
244 MARl

layer of Turkic (Tatar and Chuvash) loans; and on top are the Russian
loanwords,which arethe easiestto detect.
At fIrst glancethe Turkic elementseemsto predominate;this reflects the
high lexical frequencyof the Turkic loans in Mari. Words of (F)U origin, on
the other hand,havebeenin the languagelonger and thus tend to havemore
derivates;for this reason,but also becausethey designateprimary concepts,
words of (F)U origin have a higher textual frequencyin most everydaytexts.
The semanticfIelds coveredby words of (F)U origin includenature,the body,
primitive genrede vie, andbasicfeelingsandactions.
An inventory of the word roots occurring in the text at the end of this
chaptergives representativeresults. If we discountproper names,there are
forty-fIve words in total. Fifteen roots of (F)U origin accountfor twenty-one
occurrences:occurring twice are mari 'Mari' (an old loan from IE to FU),
ske 'self, and lam 'name',which is of Uralic pedigree;the root lu-, with an
original meaning of 'counts', occurs three times in the text (in lud=so
'reader',lu+wit 'fIfteen', and kok+lu 'twenty'). The remainderoccur once
each:Uralic kok 'two', wit 'fIve', tide 'this', tudo 'that', ilfr 'lives', andkodfJ-
'leaves';Finno-Ugric ij 'year', sog- 'stands',wij 'strength',tiiI) 'base', fJstfr
'fInishes,does'.Of Turkic origin (or mediation)are oj fJr- 'divides', samfJdk
'mostly', sagal 'a little', palfr 'knows', ojlfr 'speaks',sfJmlfr 'studies',
kundem'region',pfJtartfr 'fInishes',gfJna 'only', nergen'about,concerning',
dokan (hypothetical/conditionalsentenceparticle), suap 'reward for good
deeds'and kalfJk 'people' (both < Tatar < Arabic), zap 'time' « Chuvash<
Arabic), sot=an 'intelligible, clear' « Chuvash< Russian),possiblyalsonal-
'takes' andjiila 'way', altogetherseventeenoccurrences.Two of the words,
jeI) 'person' and woz- 'writes', cannot be assigneda sure etymological
classifIcation. Only the sentenceparticles a and wet, used togetherin the
transitional collocation a wet 'anyway', and the pan-Europeankultur are
direct Russianloans.
Even more striking evidenceof strongTurkic influenceis providedby the
morphology. Of Turkic origin is the moderative/comparative suffIx/enclitic
--rak, as are (perhaps)the pluralizer-wlak and the modal suffIx -lao Thereare
also a numberof derivationalsuffIxeswhich Mari shareswith Chuvashand/or
Tatar. Theseoriginally enteredMari along with the roots to which they were
attached,but were then abstractedand functionalized,becomingparts of the
Mari derivational machinery. According to the degree to which these
morphemeshave beenfreed from their Turkic roots they may be classified
into two broad types: those that appearonly in words of Turkic origin and
those that occur attachedto native roots as well. Only the latter qualify as
borrowedsuffIxes in the strict sense;amongthem are the =le - =10 - =10 ,
as in liim=1O 'famous'which originally occurredonly in Chuvashloanssuch
as in the above-mentionedtam=le 'sweet'. Borrowed suffIxes often are
connectedmorelooselywith the stemthantheir native counterparts, cf. -wlak,
=sfJ, --rak and the abstractsuffIx =lfJk (from Chuvash),which can be added
 MARl 245

evento finite verb forms, e.g. kul-eS=lfJk 'necessity'(kul-eS'it is necessary'

,
a third personsingular present-tenseform). Among borrowedverb-forming
suffixes is the iterative =kalfJ-, originally occurring only in Chuvash
borrowings, e.g. jamdfJl=kalfJ- 'preparesrepeatedly',but available now for
(F)U roots, as well: liij=kalfJ- 'shootsrepeatedly'from liijfJ- 'shoots'.

Canonic Shapesfor Non-borrowed Vocabulary

Comparedwith its putativeoriginal structure,the inheritedUralic andFinno-
Ugric lexical stock of Mari is well worn. Original I-stemsand mostA-stems
have lost their final vowel, e.g. lum 'name'< *nime, cf. Erzya Mordva ljem,
Finnish nimi; kok 'two' < *kakta < *kiikta, cf. Mordva kavto, Finnish kahte-;
kol 'fish' < *kala, cf. Mordva kal, Finnishkala. Someoriginal A-stemswhich
hada roundedvowel in the first syllablehavepreservedtheir final vowel, e.g.
kudo '(summer) cooking-house'< *kota, cf. Mordva kudo, Finnish kota
'Saami tent', (Northern) Saami goahti 'tent'; lumo 'glue' < *&tima, cf.
Finnish tymii; size 'autumn' < pFU *sUltiksjV, cf. Erzya Mordva sioksi,
Finnishsyksy,SaamicakCa.
On the otherhand,Mari haspreservedUralic consonantoppositionsfairly
well. The most characteristicchangesare the following: (1) intervocalic *-k-
has goneto -j or zero, e.g. ij, WM i 'year', cf. Finnish ikii 'age',Saamijahki
'year'; (2) there is no clear evidenceof original geminates*-pp- *-tt- *-kk-;
(3) the s(h)ibilants*s *sj *s have all regularly fallen togetherinto s, savein
certain positions in certain easterndialects; (4) stops and sibilants have
becomevoicedin voicedenvironments.
Many Turkic loanwords have taken on canonic shapes in Mari; for
example,word-initial voiced stopsand sibilantshavebecomevoiceless,as in
Far EasternpagfJt 'time', from Tatar bayfJt. The voicelessvelar fricative x,
more or less at home in WM (xala 'town', from Chuvash),was lost in such
older loansinto EM (ola 'town, city'), but hasenteredwith Russianismssuch
asxarakteristika,which alsohasRussianstress.
Chuvashloanwords havebeenresponsiblefor the reintroductionof s and
the high frequencyof I and r, evenin clusters:jorlo 'lazy', soria 'sickle', but
these are features which are not unknown in native vocabulary, as well.
Becauseof structuralparallelsbetweenUralic and Turkic, foreign influence
is not always easyto ascertain.Russianelementson the other hand are very
easilyrecognized.Russianloanwordsrefer to technicalandpolitical concepts
and they are usually pronouncedas strictly in accordancewith Russian
phonology as possible.Despite the rich resourceswithin Mari to coin new
vocabulary, for political reasonsmost twentieth-centuryneologisms have
beenRussian,often down to the last suffix; thusfor example'railway station'
is zelezobetQnnfJj sooruZfZ.nij,definedin a Mari-Russiandictionary (Asylbaev
1956: 116) as 'zelezobet.Qnn~j sooruZfZ.nij,
sooruZfZ.nij,sooruz~nij'.
246 MARl

Literary Eastern Mari Text

From Kugarnia, 31 January1992.

A: text in phonological transcription, segmented; B: morpheme-

by-morphemegloss; C: close English translation; D: free English trans-
lation.
adj = 'adjective-forming suffix'; =vb = 'verb-forming suffix'; =nom =
'nominal-formingsuffix'
A1.marij kalSk-Sn lum=16 jeIJ-ze-wlak kok=la-ste

Bl.Mari PEOPLE-genNAME=adj PERSON- TWO-

s3-plur =nom-ine
Cl.Mari people's famous persons among

Morko kundem=Sse Az+jal marij

(toponym) REGION=adj (toponym)+VILLAGE MARl
Morko region's Azjal Mari

ske sot=an ojSr=t=em=alt-se wer-Sm nal=Sn

SELF SENSE=adj DIVIDES=vb=act.part PLACE-acc TAKES=ger
own clear distinct place oc-

sog-a A2. samSrSk lud=so-wlak-Ian tide

STANDS-s3pres B2. MOSTLY READS=act.part-plur-dat THIS
-copies C2. mostly to readers this

lum sagal palS=me dokan A3. tud-Sn

NAME A.LITTLE KNOWS=pass.part (hypothesis) B3. PRO.s3-gen
name little known ITMAYBE C3.ofhim

nerge-n pSt=ar=t=Ss lu+wit kok+lu

ABOUT-Ioc FINISHESvb=nom TEN+FIVE TWO+TEN
on the subject last/mostrecent fifteen twenty

ij-Ia-ste gSna ojl=as woz=as tUIJ=al-Sn-St

YEAR-plur-ine ONLY SPEAKS-inf WRITES-inf BASE=vb-past2-p3
in years only to speak to write they havebegun

A4. a wet T. Jewsejew ske zap-SstS-ze kultur-Sm

B4. BUT LO T.J. SELF TIME-ine-s3 CULTURE-acc
C4. In an event T.J. in his own time culture
 MARl 247

wij=an=1J=d=m=as-(s)te LIVES=nom-
il~=s+jtila-m
STRENGTH=adj=vb=vb=nom=nom-ineLIVES=nom-
+WAY-acc
in making strong way of life

s~ml~=m=as-(s)te s~ml~=m=as-(s)te kugusuap-~msuap-~m ~st=en

suap-~m kod-en
STUDIES=nom=nom-ine BIG BENEFIT MAKES=ger LEA VES-
s3past2
in studying great benefit he left behind

D 1. Among the famous personagesof the Mari peopleTimofej Evseevich

Evseev,from the village of Azjal in the district of Morko, standsout in his
own class,taking a glorious place. To many readersthis name may well be
unknown. Only in the last fifteen or twenty yearshavepeoplestartedto talk
and to write about him. In any event,in his era T. Evseevhad greatmerit in
strengtheningculture and studying folkways (i.e. in doing cultural and
ethnographicresearch).

Referencesand Further Reading

Alhoniemi, A (1967) Vber die Funktionender Wohin-Kasusim Tscheremissischen,
MSFOu 142, Turku: SocieteFinno-Ougrienne.
- - - (1985) Marin kielioppi, Helsinki: SocieteFinno-Ougrienne.[Hilfsmittel fur
das Stadiumderfinnisch-ugrischenSprachenXl
- - (1993) Grammatik der tscheremissischen (Mari) Sprache,Hamburg: Helmut
BuskeVerlag.
Asylbaev, A A, et al. (1956) MapmI:cKo-pYCCKHH CJIOBaph, Moscow: Gosudarst-
vennoeIzdateljstvoinostrannyxi nacionaljnyxslovarej.
Bartens,R. (1979) Mordvan. tseremissinja votjakin infiniittisten muotojensyntaksi,
MSFOu 170, Helsinki: SocieteFinno-Ougrienne.
Bereczki,G. (1990) Chrestomathiaceremissica,Budapest:Tankonykiad6.
- - - (1992) Grundzugeder tscheremissischen Sprachgeschichte,vol. II, Studia
uralo-altaica34, Szeged:Attila J6zsefUniversity.
Collinder, B. (1957) Survey of the Uralic Languages,Stockholm: Almqvist &
Wiksell.
- - - (1960) Comparative Grammar of the Uralic Languages, Stockholm:
Almqvist & Wiksell.
Galkin, I.S. (1964) I1CTopwlecKruI rpaMMaTHKa MapHHcKoro H3hIKa, vol. I, Yoshkar-
Ola: n.p.
- - - (1966) I1cTopHQecKaHrpaMMaTHKa MapHHcKoro H3hIKa, vol. II, Yoshkar-
Ola: n.p.
Isanbaev, N.I. (1978) '06m;ee H OTJIHQHTeJIhHOe B COCTaBHhIX rJIaroJIax
MapHHcKoro H rrOBOJI)KCKO-TIOPCKHX H3hIKOB', Borrpochl MapHHcKoro ~3hIKa,~3hIKa,
Yoshkar-Ola,pp. 59-90.
Itkonen, E. (1962) 'Beobachtungentiber die Entwicklung des tscheremissischen
Konjugationssytems', MSFOu 125, Helsinki: Societe Finno-Ougrienne,
pp.85-125.
Ivanov, I.G. (1975) I1cTopHH MapHHcKoroJIHTepaTypHoroH3hIKa, Yoshkar-Ola.
- - - (1981) MapHH ):IHaJIeKTOJIOrHH,Yoshkar-Ola.
248 MARl

[Kangasmaa-]Minn,E. (1956) Studiesin Cheremis,vol. IV: Derivation, International

Journal ofAmericanLinguisticsvol. 2212 = [Publication2 of the IndianaResearch
Centerin Anthropology, Folklore, and Linguistics] Indiana University Slavic and
EastEuropeanSeries,Bloomington,IndianaUniversity, I: vii + 99 pp.
- - - (1960) 'The So-Called Past Tensesin Cheremis', in American Studies in
Uralic Linguistics, Uralic and Altaic Series I, edited by the Indiana University
Committeeon Uralic Studies,Bloomington:IndianaUniversity, pp. 93-120.
Kangasmaa-Minn,E. (1966) The SyntacticalDistribution of the CheremisGenitive,
vol. 1, I, MSFOu 139, Turku: SocieteFinno-Ougrienne.
- - - (1969a) The Syntactical Distribution of the Cheremis Genitive, vol. II,
MSFOu 146, Helsinki: SocieteFinno-Ougrienne.
- - (1969b) 'Typesof Nominal Sentencein Cheremis',JSFOu70: 1-12.
- - - (1969c) 'Dber die Nominalisierung des Satzes im Tscheremissischen',
Gottingen:Symposioniiber Syntaxder uralischenSprachen,pp. 118-28.
- - - (1976) 'Les rapports temporels dans Ie systeme verbaledu tscheremisse
(mari)', EtudesFinno-Ougriennes13.
Lallukka, S. (1990) The East Finnic Minorities in the Soviet Union, Annales
AcademiaeScientiarumFennicae,series B 252, Helsinki: Finnish Academy of
Sciences.
Savatkova, A. A., and Uchaev, Z. (1956) 'KpaTKHH rpaMMaTHQeCKHH OQepK
MapHHcKoro H3bIKa', pp. 793-863in Asylbaev 1956.
Sebeok,Th. A., and Ingemann,FJ. (1961) An Eastern Cheremis Manual,Indiana
University: Uralic andAltaic series5, Bloomington: IndianaUniversity.
Sebeok,Th. A., and Raun, A. (eds) (1956) The First CheremisGrammar(1775): A
FacsimileEdition, Chicago.
Serebrennikov,B.A. (1960) KaTeropHHBpeMeHH H BH}:Ia B cpHHHO-yropCKHX H3bIKax
rrepMcKoH H BOJDKCKOH rpyrrrr, Moscow.
- - - (1961a)COBpeMeHHbIHMapHHcKHH H3b1K. MOPCPOJIOrHH,Yoshkar-Ola.
- - - (1961b) CHHTaKcHcCJIO)I{Horo rrpe}:lJIO)KeHHH, Yoshkar-Ola.
Wichmann,Y. el923) Tscheremissische Textemit Worterverzeichnis undgrammati-
kalischenAbriss, Hilfsmittel fUr das Studium der finnisch-ugrischenSprachen5,
Helsinki: SocieteFinno-Ougrienne.
8 Permian
Timothy Riese

Much hasbeenwritten aboutthe meaningand origin of the word 'Permian'.

The controversybegan in the nineteenthcentury and to this date no one
explanationhasfound generalacceptance.Here there isspacefor only a very
brief surveyof the varioustheories.It is advisableto beginwith the facts and
thenproceedto the hypotheses.
The word 'Permian' is used today to designatea subgroupof the Uralic
language family which is composedof the Komi (Zyrian) and Udmurt
(Votyak) languages,both spokenin the northeasternpart of EuropeanRussia.
It is a scholarly designation,not having been taken from the languagesin
question;thesedo not havea native expressiondenoting'Komi + Udmurt' or
'Udmurt + Komi'. The word canbe tracedbackphilologically to the Russian
IIepeMb - IIepMb, forms which occur in medieval Russianchronicles.At
ftrst, the word was used to designatecertain territories, not a people or
language.The ftrst territory in questionwas the areaof the lower courseof
the Dvina River (nearthe present-daycity of Arkhangelsk,on the White Sea),
the secondwas the areavery roughly boundedby the Urals in the eastandthe
Pechora,Vychegda,andKamaRiverson the north, west,andsouth.This latter
territory was incorporatedinto the RussianEmpire in 1478. 'GreaterPerm'
lay on and about the area of today's Komi Republic. The word 'Permian'
(rrepMCKHll:) was usedfor the non-Russianinhabitantsof this territory, for the
most part Zyrians, both before and for some time after its annexationto the
Russianrealm.Later the Russiansbeganto usethe appellation'Zyrian'. Since
the nineteenthcentury 'Permian' has been in use in scholarly writing to
designatenot only the Zyrians, but also their close linguistic relatives, the
Udmurt.
There is much evidence which points to the Russian 'Perm' being
ultimately identical with the 'Bjarma-Land'of old Scandinaviansagas.This
'Bjarma-Land'can be locatedon the Kola Peninsulaand the southernshores
of the White Sea,i.e. an areaapproximatelycorrespondingto the ftrst territory
called 'Perm' by the Russians.The inhabitantsof this area,the 'Bjarmians',
werecertainlynot Zyrians,but ratherKarelians,who very likely hadintensive
trade contacts with the Zyrians farther to the east. The fabled wealth of
'Bjarma-Land'cannothavebeendue to the richnessof the land itself, but to
the traderoutesrunning throughit andconnectingScandinavia,via the upper
Kama/Zyrian territory, with the Bulgarian Empire on the lower Kama and

249
250 PERMIAN

Volga. Even if 'Penn' and 'Bjanna' are acceptedas being etymologically

identical, this does notsolve the problemof the origin of the word. For this,
severaltheorieshavebeenoffered, the most widespreadbeing the following
three:

1 It is of Zyrian origin. The Komi word panna '(certainkind of) wooded

mountainridge' hasbeenadvancedin this regard.
2 It is of Baltic-Fennic origin. A hypothetical*perii+maa 'back country,
hinterland'hasbeenmuchdiscussed.
3 It is of Scandinavianorigin. Under this theory, it is regardedas cognate
with Gennanic *benn-, *bann- (cf. English brim), meaning 'border;
shore'.

Of thesetheoriesthe first and secondstill have their supporters,but the

third is perhapsin the (shaky)ascendancy.
At this point it is necessaryto clarify the questionas to how manyPermian
languagesthereactually are. The traditional view is that there are two: Komi
and Udmurt. In the Soviet Union, however,the official line spokeof three:
Komi-Zyrian, Komi-Pennyak, and Udmurt. The question is, therefore,
whetherKomi-Zyrian and Komi-Pennyakconstitutetwo separatelanguages
or aredialectsof the samelanguage.
Sincethe 1920stwo separateliterary languages,differing from oneanother
in no significant way, have beenin use: Komi-Zyrian for the Zyrians in the
Komi Republic, and Komi-Pennyakfor the Zyrians in the Komi-Pennyak
AutonomousRegion to the south of the Komi Republic. The Soviet official
line maintainedthat thesetwo literary fonns are distinct, separatelanguages
and that the people speakingthem constitutedistinct, separatenationalities.
Consequently,they were treatedseparatelyin censuses.The questionas to
whetherthe Komi-Pennyakconstitutea separatenationality or are a part of
the Komi people as a whole can be answeredonly by the Komi-Pennyak
themselves.National affiliation is a subjective matter which cannot be
determinedby outsiders.
Linguistic affiliation, on the otherhand,is objectiveandcanbejudgedjust
as capably by non-Komi-Pennyakas by the Komi-Pennyak themselves.
There is no doubt that dialectal differencesexist betweenKomi-Zyrian and
Komi-Pennyak. These differences are, however, much smaller than the
differencesbetweenthe dialectsof most westernEuropeanlanguagesand are
by no meansso large as to justify speakingof two distinct languages.The
opinion that we are dealing with two separatelanguageshas never been
acceptedoutside the Soviet Union and presumablythere would have been
muchmoredebateon this matterwithin the SovietUnion hadcitizensnot had
to fearthe consequences of deviationfrom the official line.
Why, then, was a separateliterary fonn, Komi-Pennyak,createdin the
1920s?Why was not the literary Zyrian languageadoptedfor all Zyrian-
 PERMIAN 251

speaking territories, as would have been more expedient?One reason is

administrativeunimaginativeness.In addition to the above-mentioneddia-
lectal differences,the Komi-Permyakare separatedfrom the Komi-Zyrians
by a swath of Russian-populated territory and havetraditionally belongedto
a different administrativeunit. Another reasonis idealism. One must bearin
mind the linguistic policy of the early yearsof Sovietpower, a policy for the
most part highly beneficial to the speakersof smaller minority languages.
New literary languageswere createdand the opportunity of using them in
education,publication, and other cultural pursuits was provided. In several
cases,however,the authoritieswent too far and fosteredthe creationof two
literary languageswhereone would havesufficed, thus splitting a nationality
into two segments.It has beenmaintainedthat this was no coincidenceand
had the purposeof weakeningminority languagesvis-a.-visRussian(divide et
impera),but it is not felt to be necessaryto impugnthe efforts of the language
reformersof the 1920s,which were undoubtedlylaudableto a high degree.
On the other hand, there is no doubt that the existenceof two literary forms
for one languagehas servedto estrangethe Komi-Permyakfrom the Komi-
Zyrians and has madethe struggleof both groupsfor linguistic development
and survival all the moredifficult.
The following data are taken from various censusesconductedin the
SovietUnion. It shouldbe notedthat in the censusesof the Sovieteracitizens
were asked to report their ethnic nationality as well as the languagethey
consideredto be their mother tongue. Due to the widespreadlanguageshift
of the last few decades(usually, minority language>Russian)the reported
mothertongueis very often not the languageof the citizen'sethnic group, i.e.
many people may feel that they are Komi or Udmurt (and not Russians,
Tatars,etc.), but give their mothertonguenot as Komi or Udmurt, but (in the
greatmajority of thesecases)as Russian.In the following statisticsthe word
Zyrian is, unlessotherwiseindicated,usedto meanKomi-Zyrian and Komi-
Permyak.
The numberof peopleidentifying themselvesas Zyrians and Votyaks has
risen in every censusin this century, but at a pacemuch slower than that of
otherpeoplesof the CIS. Indeed,the word stagnationwould now seemmore
appropriatethan growth. The growth rate of the last decadewas a mere 0.2
per cent(Zyrian) and0.5 per cent (Votyak).

Nationality affiliations:

Zyrians Votyaks

1939 422,300 606,300

1959 431,128 624,800
1989 496,579 746,793
252 PERMIAN

Of evenmoreinterestare the linguistic dataof the censuses,which purport

to tell us how many people actually speaktheselanguages.Thesestatistics
present a gloomy picture: in the past few decadesthe actual number of
speakershas beenslowly but steadilydecreasing.In 1989 a total of 360,000
people declaredZyrian, and 520,000 declaredVotyak, to be their mother
tongue. The language-retentionrate expressedin percentagesof Zyrians/
Votyaks with ZyrianIVotyak as their mothertongue,is particularlyrevealing:

Komi-Zyrian Komi-Permyak Votyak

1959 89.3 87.6 89.1

1970 82.7 85.8 82.6
1979 76.2 77.1 76.4
1989 70.4 70.1 69.6

Languageretention can also dependon (1) whetherthe ZyriansNotyaks

live in their ethnic administrativeareas,(2) whether they live in a rural or
urbanarea,and(3) sex.

1 The percentageof Komi-Zyrians living in the Komi Republic remained

stablethroughoutthe twentieth century (1926: 84.5 per cent; 1989: 84.5
per cent). Approximately two-thirds of the Komi-Permyaksstill live in
the Komi-Permyak Autonomous Region. The percentageof Votyaks
living in the Udmurt Republic has declined somewhat(1926: 78.5 per
cent; 1989: 66.5 per cent). That residencein theseareasis conduciveto
the retentionof the native languageis borne out by the figures. In 1989,
74.3 per cent of the Zyrians in the Komi RepublicspokeZyrian, and75.7
per cent of the Votyak living in their administrativeregion spokeVotyak,
as their mothertongue.The slightly higherretentionrateis due, inter alia,
to the possibilities of using Zyrian and Votyak to some extent in
education,the press,and other cultural pursuits, possibilities which are
not normally availableoutsidethe republicsin question.
2 Zyrians and Votyaks havetraditionally beenrural peoples.The language
retentionrate is also consistentlyhigher amongthosestill residingin the
countrysideas opposedto thosewho havemovedto urbancentres,which
uniformly havean overwhelmingRussian-speaking majority.
3 The number of women retaining Zyrian and Votyak as their mother
tonguehasalwaysbeen3 to 4 percentagepointshigherthanthat of men.

The causefor the declinein the numberof ZyrianiVotyak speakershasthus

not been a decreasein the numbers of ZyriansNotyaks, but a steadily
acceleratinglinguistic assimilationof ZyrianiVotyak > Russian.What have
beenthe major causesof this switch-over?
 PERMIAN 253

1 The prestigeof Russian.Russianis the languageidentified with modem-

day society,with progress.Zyrian and Votyak on the other handhaveto
someextentcometo be identified with a rural, antiquatedway of life. A
goodcommandof Russianbecameimperativein the pastfew decadesfor
anyonewishing to rise in Sovietsociety.
2 Loss of compactZyrianNotyakterritories. While still existing, the size
and numbersof areaswhereZyrianNotyakis spokenalmostexclusively
have been steadily shrinking. Previously, large-scaleethnic contact,
intermingling, and assimilationtook place only in urban areasor on the
fringes of a large, relatively homogeneousZyrian or Votyak area. Now,
a much greaterproportion of the Zyrian and Votyak peopleslive in the
'fringe' areas,and the rural exodusaffecting all peoplesof the CIS has
also led many Zyrian and Votyaks to (Russian-speaking) urban centres.
There has also been an influx of Russiansand other nationalities into
traditional Zyrian and Votyak territories.This is readily evidentfrom the
percentagesof Zyrians andVotyaks in the total numberof inhabitantsof
the Komi and Udmurt Republics.Whereasin the 1920sthe Zyrians and
Votyaks madeup about 90 per cent of the populationof their respective
republics, this figure has now plummetedto (1989) 23 per cent in the
Komi Republic, 58 per cent in the Komi-PermyakAutonomousRegion,
and 30.9 per cent in the Udmurt Republic. The offspring of the now
commonplaceethnically mixed marriagesare overwhelminglyraisedas
Russianspeakers.
3 Decline in the use of ZyrianNotyak in educationand the press. The
decadefollowing the Bolshevik Revolution saw a rapid upswing in the
use of Zyrian and Votyak in educationand print. An effort was madeto
provide all Zyrian and Votyak children who lived in their respective
republicswith the opportunityof being educatedin their native tongue-
if not for their complete schooling then at least for the first years of
primary education.Ever since the 1950s,however,the opportunitiesfor
educationin Zyrian and Votyak havebeengradually whittledaway, until
now the great majority of Zyrian and Votyak children are educated
exclusively in Russian,their native tongue being at most an optional
subject. An interesting survey has shown that rural Zyrians prefer to
speakZyrian, but to read and write in Russian.It does not need to be
stressedthat beingeducatedin Russianis highly conduciveto linguistic
assimilation.It was not until the 1920sthat Zyrian and Votyak beganto
be used extensively in printed media. The number of books in these
languageshas, however, declinedsteeplyin the past few decades.This
developmenthas also beendetrimentalto the fostering and development
of Zyrian andVotyak.

What are the prospectsnow for the Permianlanguages?Will there still be

speakersof Zyrian and Votyak fifty years hence, or will the processof
254 PERMIAN

assimilation result in complete Russification?It can safely be said at the

presentmomentthat the situationis serious,but not yet hopeless.The fall of
the Sovietregimeand the weakeningof centralpowerhavebroughtaboutthe
possibility that the Permianpeopleswill be able todeterminetheir own affairs
as regardseducation,the press,and other cultural activities. The situationis,
however, still extremely volatile, making predictionsquite hazardous.It is
also a questionof what Permiansthemselveswant and what efforts they are
willing to make to achievetheir goals.If a concentratedeffort is madenow,
a great deal can be done to offset the weakeningof the languagewhich has
set in over the recentpast.If a strengtheningof national consciousness were
to come about, one which results in a renewal of school educationin the
Permianlanguagesand a new upswingin the useof Zyrian and Votyak in the
press and in local government, the future of these languageswould be
assured.

Proto-Permian
After the westwardmigrationof the ancestorsof the Finno-Volgaic (= western
Finno-Permic)peoples,a group of Finno-Ugrians(or Finno-Permians,if we
assumethat Ugric hadalreadybrokenaway) remainedin the areaof the Kama
and Vyatka Rivers, to the east of the Volga. It has been estimatedthat this
separationtook placeabout2000 BCE. Thoseremainingbecamethe ancestors
of the Zyrians and Votyaks and are referred to as proto-Permians.The
languagethey spoke is called proto-Permian(pPN). During the long pPN
period the languageunderwentfar-reachingchangesinvolving all linguistic
domains.Thesechangesproduceda separatelanguagequite distinct from its
congeners.Proto-Permianlater evolvedinto two separatelanguagesafter the
ancestorsof the present-dayZyriansbeganmigrating northwardsin the eighth
and ninth centuries CEo We have no written records of pPN, and all our
statementsaboutit arethereforehypothetical.Studiesof pPN arebasedon the
evidenceprovidedby present-dayZyrian andVotyak, older written recordsof
theselanguages,by other Uralic languages,and by loanwordsin and from
other languages,related andunrelated.Of particularinterestare the Middle
BulgarianandIranianloanwordsin pPN.
The present-dayPermianlanguagesarerelatively closelyrelated.That they
are relatedis obvious not only to linguists, but also to naIve native speakers.
The areas of identity and similarity extend to all subsystemsof language
structure. Mutual intelligibilty over any significant stretch of discourseis,
however,not possible.

Old Zyrian (OZ) and the Yaz'va Dialect of Zyrian (PO)

In the following sectionsof this chapterlinguistic exampleswill be cited from
both Old Zyrian and the yaz'vadialectof Zyrian.
 PERMIAN 255

Old Zyrian
By 'Old Zyrian' is meantthe languageof the earliesttexts in Zyrian. These
datefrom the fourteenthcenturyand are basedon the Lower Vychegda(VU)
dialect of that time. It doesnot meanan older stageof Zyrian from which all
the modemdialectsevolved.
Yaz'vaDialect
This dialect, also known as EastPermian,is spokenby 3-4,000peoplein the
northeasternportion of the Perm' Oblast. It diverges markedly from other
Zyrian dialectsand is especiallynoteworthyin that it haspreserveda number
of featuresthat havebeenlost elsewhere.Of particularinterestare the vowel
and stresssystemsof PO. The creationof a separateliterary languagewould
havebeenmorejustified in the caseof PO thanin the caseof Komi-Permyak,
but this was not essayedbecauseof the small numberof speakers.

Phonology
In what follows, I shall attempta survey of the phonologicalstructureof the
Permianlanguages.The main emphasiswill be on historical development, i.e.
p[roto-]U[ ralic]/p[roto-]F[inno-]U[gric] > pPN > Z[yrian] andVo[tyak]. Only
a succinctreview of the mostimportanthistorical processes will be provided;
detailslie in the literaturelisted.
The ConsonantSystem

Initial Consonants
Our point of departureis, for the sakeof simplicity, the consonantsystemof
proto-FU. This systemis generally accepted,while of course beinghypo-
thetical andstill subjectto debate.
Clearly the main differencebetweenpFU and pPN is the appearance,in
pPN, of voiced analoguesfor most of the pFU obstruents.About one-sixthof
inheritedpFU words appearwith voicedinitial obstruentin pPN.
The bilabial glide *w- of pFU developedinto pPN labiodentalV-. A new,
i.e. secondary,w- arosein initial position (> *90- > wo-) in pPN from the
diphthongizationof the pPN vowel *0.

Table8.1 The pFU initial consonantsystem

Labials Dentals Alveolars Palatals Palatovelars

Nasals *m *n i
Stops *p *t *k
Affricates *6 *ci
Fricatives *s, *b? *1; *si, *b i
Glides *w *j
Liquids *1 *r *li
256 PERMIAN

Table8.2 The pPN initial consonantsystem

Labials LabiodentalsDentals Alveolars Palatals Palatovelars

Nasals *m *n *ni
Voicelessstops *p *t (*t i ) *k
Voiced stops *b *d (*di) *g
Voicelessaffricates *c *ci
Voiced affricates *¢ *¢i
Voicelessfricatives *s *1; *si
Voiced fricatives *z (*z) *(zi)
Glides (*w) *v *j
Liquids *1 *r *li

pFU *6 j > pPN *lj (Permianoffers no goodevidencefor initial pFU *6).

The consonantsgiven in brackets(w, tj , di, z, zj) are of late pPN origin
and probably played only a peripheralrole in the consonantsystemof that
time.
The initial consonantsystem of the present-dayPermian languagesis
almost identical with that of pPN. The Permian languagesare thus quite
conservativein this respect.
SecondarypPN *w- > Zyrian/VotyakV-, W-, or zero,dependingon dialect.
pPN *r- > Votyak ~- or ¢L dependingon the frontnesslbackness of the
following vowel in pre-Votyak.
z-,
Initially, ti-, tJi-, and t- are still amongthe least frequentphonemesof
the Permianlanguages.
The following examplesillustrate thesethree-stagedevelopments.Unless
otherwiseindicated,the stagesarepFU > pPN > Zyrian andVotyak.

1 pFU *p- > pPN *p-I*b- > Zyrian/Votyakp-,b-

*piIJe 'tooth' > *piIJ > ZlVo pili
*pera 'backpart' > *bar> Z ber Vo ber

Table8.3 The Zyrian/Votyakinitial consonantsystem

Labials LabiodentalsDentals Alveolars Palatals Palatovelars

Nasals m n ni
Voicelessstops p t ti k
Voiced stops b d di g
Voicelessaffricates cc i
Voiced affricates ¢ ¢i
Voicelessfricatives s 1; si
Voiced fricatives zz 1; zi
Glides v j
Liquids 1; r Ii
 PERMIAN 257

2 pFU *t- > pPN *t-I*d- > Zyrian/Votyakt-, d-

*HUwa 'winter' > pPN *tOl > Z tev Vo tol
*tarV 'proximity' > pPN *dor > Z dor= Vo dor
3 pFU *k- > pPN *k-I*g- > Zyrian/Votyakk-, g-
*kate 'hand'> pPN *ki > ZlVo ki
*kince 'nail, claw' > pPN *gUii > Z giiVo giii"
4 pFU *m- == pPN *m- == Zyrian/Votyakm-
pFU *maye 'land, earth'> pPN *mu > ZlVo mu
5 pFU *n- == pPN *n- == Zyrian/Votyakn-
pFU *nej6e 'girl, daughter'> pPN *nUl > Z niv Vo nit
6 pFU *ni- == pPN *nj == Zyrian/VotyakTi-
pFU *niele 'arrow' > pPN *njl>l > Z Tiev Vo Tiel
7 pFU *c- > pPNsmoke'
*c-I*~- > Zyrian/Votyakc-, ~­
pFU *ciIJ 'fog; smoke'> pPN *cUIJ > ZlVo Gin
pFU *cacV '(a sort of) bed' > pPN smoke'
*~o~ > Z smoke'
~~ 'cellar' Vo ~(~+ol
smoke''space
underbed'
8 pFU *ci- > pPN *ci-I*¢L > Zyrian/Votyakd-, 1-
pFU *ciaIJkV- 'breaks'> pPN *cieg- > Z deg- Vo dig-
pFU *cierV 'grey' > pPN *~orpPN > Z lor Vo lar, both 'dawn'
9 pFU *w- > pPN *v- == Zyrian/Votyakv-
pFU *wete 'water' > pPN *va > Z va Vo vu
later pPN *w- (generally)> Zyrian/Votyakv-
pFU *06e 'year' > early pPN *0- > late pPN *wo- > Z vo Vo va
10 pFU *s- > pPN *s-I*z- > Zyrian/VotyakS-, z-
pFU *soja 'arm, sleeve'> pPN *soj > Z soj Vo suj
pFU *sorV '(sort of) plant' > pPN *zl>r > Z zer 'oats' Vo zer 'Bromus
secalinus'
11 pFU *s- == pPN *s- == Zyrian/Votyaks-
pFU *siIJe=re 'mouse'> pPN *sUr > ZlVo srr
12 pFU *sj == pPN *sL == Zyrian/Votyaksi-
pFU *sjata 'hundred'> pPN *~o*~o > Z sio Vo siu
13 pFU *j- == pPN *j- == Zyrian/Votyakj-
pFU *jaIJe 'ice' > pPN *ji - *jl> > Zji Vo je
14 pFU *1- == pPN *1- == Zyrian/Votyak1-
pFU *luwe 'bone'> pPN *lU > ZlVo lr
15 pFU *lL == pPN *lL == (Zyrian?)NotyakP-
pFU *ljekkV 'tight, narrow' > pPN *ljok > Vo Puk=i'+
16 pFU *&- > pPN *lL == Zyrian/VotyakP-
pFU *&eme 'Prunuspadus'> pPN *ljl>m > ZlVo lJem
17 pFU *r- == pPN *r- > Zyrian r-, Votyak ~-, 1-
pFU *rekkV 'mush,grits' > pPN *rok > Z rok Vo ¢Uk
pFU *repa=ciV 'fox' > pPN *rVp V ci > Z ruCi Vo did'
258 PERMIAN

Non-initial Consonants
In the Finno-Ugric languagesit is of greatimportanceto differentiatebetween
the developmentof consonantsin initial and non-initial position, as these
developmentscan differ greatly. As will be seenbelow, in the development
of the Permianconsonantsthe nasalsand the liquids show no greatdifference
due to position, but the obstruents(stops, affricates, and fricatives) show a
strong divergence.Whereasthe consonantdevelopmentof p(F)U > pPN >
Zyrian/Votyak is remarkablyconservativewith regard to initial position, in
non-initial position greatchangestook place,drasticallyaltering the shapeof
pPNwords.
The stock of pFU non-initial consonantswas larger than that of initial
consonants.In addition to all the initial consonants treated on p. 257 the
following consonantscould occur in non-initial (primarily intervocalic)
position: (1) *-1)-, *-b-, *-y-; (2) intervocalic geminatestops *-pp-, *-tt-,
*-kk-; (3) many consonantclusters at the border between the first two
syllables.
In the present-dayPermian (literary) languagesthe stock of non-initial
consonantsis identical with that of the initial consonants.This stateof affairs
datesbackto late pPN. In the following surveythe intermediatestageofpPN
canbe omitted, as in mostcasestherewas no major changebetweenpPN and
ZyrianiVotyak.
The following examplesillustrate the developmentof non-initial con-
sonants,pFU > ZyrianiVotyak.
Intervocalicposition:

1 pFU *-p- > ZyrianiVotyak0

pFU *kopa 'skin, bark' > pPN *ku > 2No ku
2 pFU *-t- > ZyrianiVotyak0
pFU *sjata 'hundred'> pPN *sjo > Z sio Vo siu
3 pFU *-k- > ZyrianiVotyak0
pFU *joke 'river' > pPN *ju > Zju Vo ju+
4 pFU *-pp- > ZyrianiVotyak -p-
pFU *sappa'gall (bladder),> pPN *sap > Z sepVo sep
5 pFU *-tt- > ZyrianiVotyak -t- (in somewords,with secondarypalataliza-
tion t> ti)
pFU *witte 'five' > pPN *vit > Z vit Vo viti
6 pFU *-kk- > ZyrianiVotyak -k-
pFU *rekkV 'mush,grits' > pPN *rok> Z rok Vo i/uk
7 pFU *-m- == ZyrianiVotyak -m-
pFU *s-ieme 'fish scale'> pPN *Si6m > 2No Siem
8 pFU *-n- == ZyrianiVotyak -n-
pFU *sene 'vein' > pPN *s6n > 2No sen
9 pFU *-ui- == ZyrianiVotyak -ni-
pFU *kuuia- 'shutseyes'> pPN *kUnL > Z kuni- Vo kiW-
 PERMIAN 259

10 pFU *-IJ- == pPN *-IJ- > Zyrian -li-, -n-, -m- Votyak -1)- -li-, -n-, -m-
(Dependson dialectand vocalic environment.In general:in front-vocalic
surroundings*-IJ- > -li-, in back-vocalic surroundings*-IJ- > -n- or,
particularlyin the neighbourhoodof roundedvowels,> -m-. SomeVotyak
dialectsretain -1)-.)
pFU *piIJe 'tooth' > *piIJ > VVo pili
pFU *paIJe 'head'> pPN *POIJ > Z pom/ponVo pum/pu1)
11 pFU *-c- > Zyrian/Votyak-c-, -jt-, -s-, -z-
pFU *mucV '(sort of) illness' > pPN *mUz > VVo miz
pFU *keca 'ring, circle' > pPN *kUc > VVo kic
12 pFU *-ci- > Zyrian/Votyak(varioussibilantsor affricates)
pFU *repa=ciV 'fox' > Z rud Vo VVo ~ici"
13 pFU *-w- > Zyrian/Votyak0
pFU *luwe 'bone' > pPN *lU > VVo Ii"
14 pFU *-b- > Zyrian/Votyak-1-, -0
pFU *sjiiDa=mV 'heart' > pPN *s-iVlVm > Z sielemVo siulem
pFU *wib V =mV 'marrow' > pPN *vem > Z vemVo vim
15 pFU *-&- > Zyrian/Votyak-P-
pFU *wu&e 'new' > pPN *vUlj > VVo viP
16 pFU *-s- > Zyrian/Votyak-z-
pFU *pesa'nest' > pPN *poz > Z poz 'nest'Vo puz 'egg; scrotum'
17 pFU *-sL > Zyrian/Votyak-z!-
pFU *kusjV 'twenty' > pPN *kUzj > VVo Hz!
18 pFU *-s- > Zyrian/Votyak-z-
pFU wansa'old' > pPN *VDZ > Z vazVo vuz
19 pFU *-j- == Zyrian/Votyak-)-
pFU *woje 'adiposetissue'> pPN *vOj > Z vi} Vo vel
20 pFU *-y- > Zyrian/Votyak-0-
pFU *maye 'land, earth' > pPN *mu > VVo mu
21 pFU *-1- == Zyrian/Votyak-1-
pFU *kola- 'dies' > pPN *kul- > VVo kul-
22 pFU *-lL == Zyrian/Votyak-P-
pFU *eljV 'wet' > pPN *ul j > Z uP Vo iP
23 pFU *-r- == Zyrian/Votyak-r-
pFU *wire 'blood' > pPN *vir > VVo vir

We may summarize the major developments among the non-initial

intervocalicconsonantsasfollows: (1) the oral stopswent to zero, as did *-w-
and its velar analogue*-y-; (2) the nasals,liquids, and the palatal glide *-j-
usually remained unchanged; (3) the affricates partially lost their stop
components, sometimes acquiring voice; (4) the intervocalic fricatives
voiced, *-s- *-s-i- *-s- > *-z- *-zL *z-; (5) the interdentalfricatives merged
with the laterals,*-b- *-&- > *-1- *-lL.
260 PERMIAN

ConsonantClusters
Approximately seventydifferent consonantclustershavebeenreconstructed
for p(F)U, with roughly an additional twenty for Finno-Permic.Although
there are not always Permianreflexes for thesecombinations,we still have
somesixty-five clustersto accountfor. HereI shall do no more thanillustrate
the developmentsof the main typesof clusters.
Developmentof the clustersis of two basic types: retentionor simplifica-
tion. Simplification, usually the loss of one element, was much the more
frequent.

1 pFU clustersof two stops: pFU *-pt- was simplified to ZNo -t-; *-kt-
mostfrequentlygavethe alternationk - kt-.
pFP *saptV=rV 'currant'> pPN *soter > Z seterVo suter
pFU *kakta 'two' > pPN *kUk( t -)> Z ki"k Vo ki"k( t-)
2 pFU stopplus sibilant: in this combination,the stopwas lost in pPN.
pFU *mekse 'bee' > pPN *mos > ZNo mus
pFU *kipsV 'leg hide' > pPN *kUS > Z kis
3 pFU sibilant or affricate plus stop: retentionof the clusteris possible,but
not uniform.
pFU *musike- 'washes'> pPN *mUsik- > Z mi"Sik- Vo miSik-
pFU *kusika 'dry' > early pPN *kosi > late pPN *kwosi > Z kosi Vo
kwasi
4 pFU nasalplus stop: the nasal went to zero; the stop was subsequently
voiced.
pFU *kumpa 'wave' > pPN *gUb= > Z gib= 'ripples', Vo gib=ed
'dungheap'
pFU *lamte 'low(lands)'> pPN *lud > ZNo Iud
pFU *cial)kV- 'breaks'> pPN *cieg- > Z deg- Vo dig-
S pFU fricative, glide, or liquid plus stop: both retentionand simplification
occurred,but the latter was more frequent.The fricatives *6 *& and the
glides *w *j disappearedmore readily than the liquids, which proved to
be somewhatmoreresilient.
pFU *pilwe 'cloud' > pPN *pil > Zpiv Vo pizi=em
pFU *kertV '(sort of) wild duck' > pPN *gord=a> Z gorda Vo gurdo

Developmentof Consonantismafter Proto-Permian

Although in various dialects a number of minor changeshave taken place
sincethe breakupof Permianunity, therehasbeenonly one changeof major
importance.This changeinvolves the phonemeI which, dependingon the
dialect, has either (1) remainedunchangedor (2) developedwholly or partly
into v, w, or zero.
In Zyrian the dialectsmay be classifiedinto four groupsaccordingto the
developmentof pPN *1:
 PERMIAN 261

1 *l remainsunchanged;
2 *l > v (or w) in all positions;
3 *l> v in non-prevocalicpositionwithin the word;
4 *l> zeroin non-prevocalicpositionwithin the word; the precedingvowel
is lengthened.

We may usevariantsof CommonZyrian lol 'soul; breath' and the suffixes

-teg 'without' and -en 'with' to illustrate thesedevelopments:

1 2 3 4

Word-initial, Word-final 101 vov lov 100

Preconsonantally lolteg vovteg lovteg looteg
Intervocalically lolen voven lolen lolen

In the Zyrian literary languagesthe l- v alternation(type 3) is the accepted

norm. In the majority ofVotyak dialectsaswell asin the literary language1has
provento bestable(type 1), butin somedialects(e.g. Shoshma)we find changes
andalternationssimilar to thoseseenin Zyrian (literary Votyak nil 'girl', niltek
'without a girl' , nilez 'girl (acc)';Shoshmadialectni'w, ni'wtek,ni'wez).

The Vowel System

Vowelsof the First Syllable

The developmentof the vowel systemin the Permianlanguageshas beenthe
subjectof a greatdeal of study and of evenmore debate.Thereare few topics
more controversialwithin the field of Finno-Ugric linguistics than that of
vowel history, and the Permianvowels form one of the most contestedparts
of the whole.
It seemsto be a linguistic universalthat vowels adjustto their phonological
surroundingsmore than do consonantsand similarly, that vowels are more
susceptibleto changethanconsonants.That consonantstendto be morestable
thanvowelsis well illustratedby the present-dayFinno-Ugriclanguages, since
thedifferencesbetweendialectsaremadeup to a largepartby differencesin the
vowels and not by differencesin the consonants.Although many problems
remainunsolved,a relatively high degreeof consensushasbeenreachedin the
field of Finno-Ugric consonanthistory.Themajorpointsof Permianconsonant
history are similarly relatively free of controversy:researchbeing carriedout
now in this field is directedat details,not at basicquestions.
With the vowels, on the other hand, the situation is quite different. While
it is possibleto speakof prevailing theoriesin this field, nothing approaching
a consensuson consonantdevelopmenthas beenreached.Views vary on the
reconstructedpU, pFU, andpPN vowel systems.
Scholarswho work on p(F)U vowel history have made much use of the
262 PERMIAN

concept of 'key' languages,assuming that one or more languageshave

retained the original vowel system to a much greater extent than others.
Whereas at one time a commonly held theory spoke of the inherent
conservativeness of the EastCheremisandEastOstyakvowel systems,today
most researchersare more inclined to regard the vowel systemof Baltic-
Fennic as having preservedbetter the original characterof the pFU vowel
system.Whetherthis is correct or not, this theory automaticallyentails not
only a comparativeregularity of vowel correspondences between Baltic-
Fennic and Permian, but also a comparative irregularity of vowel cor-
respondencesbetween Permian and pFU. Such a starting point for the
examinationof the Permianvowels thereforemakes a description of their
developmentall the moredifficult.
This chaptercannotessaya comparison,or even a survey, of the various
theoriesregardingthe diachrony of the Permianvowels. Instead, thebest-
known and perhapsmost widely acceptedtheory, that connectedwith the
namesItkonen, Lytkin, and Redei, will be briefly presented.The order of
stageswill be descendentrather than ascendent,with each stagebecoming
more hypotheticaland subjectto query. We may numberthe stagesthus: (la)
Zyrian < proto-Zyrian, (lb) Votyak < proto-Votyak; (2) proto-Zyrian and
proto-Votyak < pPN; (3) pPN < pFU.
As notedabove,the consonantsystemsof both Zyrian andVotyak (literary
languages)areidentical.Interestinglyenough,the sameis true for their vowel
inventories.The sevenvowel phonemesof literary Zyrian and Votyak are
shownin Figure8.l.
Of relevancehere are also the dialectswith divergentvowel systems.For
Zyrian, theseare aboveall yaz'va(PO) and (to a lesserextent)Upper Sysola
(SO). Old Zyrian (OZ) as attestedin medievaltexts also plays an important
role in the researchof Zyrian vowel diachrony. Votyak is on the whole
somewhatmoreuniform in this connection,but the Beserman(B) dialect and
the southwest (SW) dialects diverge somewhat and must be taken into
account. The Yaz'va vowel systemis shown below. (R = rounded, UR =
unrounded.)

Figure8.1 Zyrian andVotyak vowel inventory (literary languages)

Unrounded Rounded

High i 1 u

Mid e e 0

Low a

Front Back
 PERMIAN 263

Front Back
UR R UR R

High 6 U u
Close-mid e
Open-mid e 6 0
Low a

Correspondingto (literary) Zyrian 0 Upper SysolaZyrian has two vowel

phonemes,a relatively closed(high) 0 anda moreopen(lower) o. For present-
day Zyrian 0 and e, Old Zyrian had two vowel phonemeseach,closedand
open0, 0, ande, e.
i:
Correspondingto literary Votyak e, and (pPN *6 » u Besermanhas fJe,
e, fJ and southwesterndialects(suchas Shoshma)havee, u, o.
It mustbe emphasizedthat the precisephonologicalnatureof severalof the
above-mentionedphonemesis controversial (e.g. PO e) and subject to
varying interpretation,a fact which makes historical analysis all the more
difficult. It must furthermorebe notedthat althoughthe vowel systemsin the
literary Permianlanguagesareidentical,cognatesin the two languagesdo not
necessarily,indeeddo not often, show the samevowels. As will be shown
below, there are important correspondences betweenZyrian and Votyak, but
also a greatnumberof minor correspondences andexceptions.

Zyrian < proto-Zyrian (pZ)

OZ SO PO ZLit < pZ

aa a aa e < *a
0 00 0 0 0 0 < *0
6 6u 0 u 0 0 < *6
u u uJu u < *u
e e 6 e < *0
ee e u e < *6
1 1 e 1 < *u
ee e ee e < *e
ee e e e < *e
e e e < *.1
e

The reconstructedvowel systemfor pZ is thus:

*.1 *u *u
*Ae *6 *6
*e *0 *0
*a
264 PERMIAN

The developmentfrom pZ to literary Zyrian, which representsthe majority

of present-daydialects, is thus characterizedby delabialization (*li > 1),
merger (*0, *6 > 0 and *e, *e > e), or both (*0, *0 > ii). There is a good
possibility that the distinction betweenthe pairs *0/*6, *0/*0, *e/*e was not
so much tongueheightas a differencein tensenessllaxness.
Votyak < proto-Votyak(p Vo)
B SW VoLit pVo

a a a < *a
1'1 li u < *6
u u u < *u
0 0 0 < *0
e <'> e < *0

"e
e/o
"
e
e/o
1
e
e/o
<
<
<
*li
*ii
*e
*i *i *i < *i

The vowel systemof proto-Votyak is thus:

*.1 *li *u
*0 *6
*e *0
*ii *a

The developmentsfrom proto-Votyak to literary Votyak are not as regular

as thoseof pZ > Zyrian and are thus less compelling. There are examplesof
raising (*6 > U, *ii > e), delabialization(*0 > ii, *li > i"), partial labialization
(*e > e/o) andbacking(*0 > 0).
The DevelopmentofpZ andp Vo from pPN
The following four tongue-heightvowel systemhas been reconstructedfor
pPN:

*.1 *u *u
*e *0 *6
*e *0 *0
*ii *a

In pZ, pPN *ii mergedwith its roundedneighbour*0 to give the pZ vowel

system. In the development from pPN to pVo on the other hand
neighbouring *e and *e merged as pVo *e, and pPN *u and *0 (not *6
!) merged as pVo *u. That *u and *0 underwent a merger, rather than
*u and *6, could be an indication that the differencebetweenthesevowels
 PERMIAN 265

was one of tensenessllaxness rather than tongue height, as in the case of

pZ mentionedabove.
Examplesillustrating the developmentof the vowels in pPN > Zyrian and
Votyak:

ZNo correspondence

pPN *baI) 'face' > Z ban Vo bam ala

pPN *poz 'nest' > Z pozVo puz 0/0
pPN *161 'breath,soul' > Z lov Vo luI o/u
pPN *juk 'pile' >ZjukVojuk u/u
pPN *tOl 'winter' > Z tev Vo tol e/o
pPN *sj(jm 'fish scale'> Z stemVo stem e/e
pPN *lUmi 'snow' > Z lim Vo limi" "iii
pPN *sappa'gall (bladder)'> Z sepVo sep e/e
pPN *pelj 'ear' > Z pePVo peP e/e
pPN *zer 'rain' > Z zerVo zor e/o
PPN *vez 'green'> Z veiVo vot e/o
pPN *vir 'blood' > Z vir Vo vir iii

The DevelopmentfromProto-Finno-Ugricto Proto-Permian

The vowel systemusuallyproposedfor pFU is this:

*i *[ii] *u
*e *e *0
*a *a

Many attemptshave beenmadeto derive the pPN vowel system(and its

descendants,the pZ and pVo systems)from pFU, but hitherto the task has
provenwell nigh insurmountable.Oneis juggling herewith threereconstructed
levelsof vowel diachrony,andthusfar no onehassucceeded in reconcilingthe
widely acceptedpFU reconstructedinventory with a pPN reconstruction.A
surveyof the etymologicalmaterialrevealsthatmostpPN vowelshaveat least
two pFU forebears.To list only themostfrequentreflexes:

pPN pFU pPN pFU

*a < * a,
.. * e *(j < *e,*o
*a < *a,*e,*a *U < * u,
.. * u
*0 < * a,
.. * e *e < *e, *i
*A0 *a,*e,*o * a,
.. *.1
< *e <
*u < *a,*o *.1 < *.1
*0 < *a

Given this many major reflexes (not to mention the many minor ones),the
chasmbetweenpFU and pPN has not beenbridged. What has beenpossible
266 PERMIAN

in the caseof the consonantsseemsstill out of reachin the caseof the vowels.
In conclusion,one is temptedto believe that either (1) a satisfactorytheory
accountingfor the basic outline of pFU > pPN vowels diachronywill never
be found, or (2) such a theory would involve considerablerewriting of what
mostnow assumeto be true for p(F)U andpPN.
VowelsBeyondthe First Syllable
In the precedingsectiononly the vowels of the first syllable were discussed.
It is typical for the Finno-Ugric languagesthat not all the vowels occurring
in the first syllable can occur in other syllables,or that someoccur with less
frequency.In present-dayZyrian, for example,the vowels 0, U, and e occur
only rarely in non-first syllable. For pFU only *e and *a/*a (the distribution
being dictated by vowel harmony) are usually reconstructedfor non-first
syllables.
The attentivereaderwill alreadyhavenotedthe drasticshorteningof words
which occurredin the transitionfrom pFU to pPN: amongthe examplesgiven
above are pFU *kate 'hand' > pPN *ki; pFU *sil)ere 'mouse'> pPN *SUr.
Thesetwo examplesillustrate one of the most commonfates to befall a non-
first vowel, namelyannihilation.The otherpossibility was raising of the non-
first syllable *e and *a/*a to a high vowel, probably*i, with allophones.This
high vowel was in turn lost in Zyrian, but retainedin mostVotyak dialects(as
n.
i or Furtherexamplesof annihilation:

pFU *soja 'arm, sleeve'> pPN *s6j > Z so) Vo sui

pFU *sappa'gall (bladder),> pPN *sap > Z sepVo sep
pFU *pil)e 'tooth' > *pil) > 7JVopitt

Examplesof raising (with later loss in Zyrian):

pFU *lume 'snow' > pPN *lUmi > Z liin Vo liini'

pFU *kince 'nail, claw' > pPN *gUzi > Z giz Vo gizi' (with pPN *u < pFU
*[0])

Vowel Hannony
Unlike manyotherFinno-Ugric languagesPermianhasno vowel harmony.In
p(F)U it was certainly present,but in a rudimentaryform. In some Finno-
U gric languagesit has developedinto a mechanismwhich pervadesthe
phonology; in othersit has waned away.The massivevowel changesin the
first syllable and the wholesalereduction and loss in the seconderadicated
vowel harmonyalreadyin pPN, andit hasnot sprungup anewin eitherZyrian
orVotyak.
Stress
In p(F)U stresswas probably non-distinctive.It is also probablethat stress
tendedto fallon the first syllable.We canassumethe samefor proto-Permian.
 PERMIAN 267

In the present-daylanguagesthe stresssystemsare different and form one of

the major distinguishingfactorsbetweenZyrian andVotyak.

Zyrian
In most Komi-Zyrian dialects stressis non-distinctive and the tendencyis
to stressthe first syllable. Stressin the Komi-Permyakdialectsis more fixed:
certaingrammaticalendingsare accented,othersare not. The systemin these
dialectsis not entirely uniform. In the yaz'vadialect (PO) we find a unique
stresssystem.In general,the first non-high vowel in polysyllabic words is
stressed,but in inflected words with a high vowel in their monosyllabicstem
this high vowel either loses or retains its stress in the inflected form
dependingon its history. The high vowel retains the stressif it developed
from a pZ non-high vowel, as in the case of PO i < pZ *e, u < *0, or U
< *6, e.g.

pPN *t08 'beard'> Zyrian Lit tos, PO tus, instrumentaltgs-()n

but if the high vowel continuesa pPN high vowel, then the stressmoves to
the right in accordancewith the generalrule, e.g.

pPN *purt 'knife' > Zyrian Lit purt, POpurt, instrumentalpurt-fln

In Votyak, the stressfalls on the last syllableexceptin certaingrammatical

forms. This accentuationpattern,so atypical of the Finno-Ugric languages,is
certainly due to the influence of the neighbouringTatar language.In both
Tatar and Votyak the imperativeforms of the verb form an exceptionto the
generalrule in that they are accentedon the first syllable.

Morphology
Both Permianlanguagesare typically agglutinativein that their word stems
arenormally not subjectto changewheninflectedandeverysemanticelement
is expressed,as a rule, with a single morphologicalelement.Words which
contain several semanticelementsare thus built up out of morphological
elementsof the samenumber.For example,the Zyrian equivalentof Latin in
silv-fs 'in forests' is ver-jas-i"n, with separatelyencodedplural (-jas-) and
local case(-in).

Declension
This sectionbriefly treatsnumber,case,andthe possessivesuffixes.

Number
In addition to singular and plural, for p(F)U a dual is also postulated.This
ancient dual is not to be found in Permian.The p(F)U plural markers *-t,
*-i- are similarly absent(the plural morpheme*-n will be mentionedbelow,
268 PERMIAN

under possessivesuffixes). In pPN a new plural suffix evolved from the

pPN noun *jOsV 'limb; joint; people, folk'. This word survives in both
Zyrian and Votyak both as an independentnoun (Zyrian jez 'people, folk;
joint', Votyakjoz 'limb, joint; contemporary')and as a plural suffix (Zyrian
-jas, Votyak -jos - -os). The suffix occurs in all cases of the plural,
including the possessiveforms, e.g. ZNo ki 'hand', plural nominative
Zyrian ki-jasNotyak ki-os, plural dative Zyrian ki-jas-li"Notyak ki-os-li:
dative plural third person singular ('to hislher hands') Zyrian ki-jas-is-li"f
Votyak ki-os-ez-li".

Case
Both Permianlanguagesare characterizedby a relatively large number of
cases:in the Komi-Zyrian and Komi-Permyakliterary languagesthe number
is usuallyheld to be seventeen;in Votyak it is fifteen. The discussionto follow
briefly treatseighteencasesuffixes; in Zyrian, the adverbialis not generally
viewed as being a case, and in Votyak the first and secondprolatives are
usually regardedas onecase,althoughhistorically they are distinct.

Zyrian Votyak

1 Nominative (zero) (zero)

2 Accusative zero, -es, -e zero, -e, -G)ez
3 Genitive -len -len
4 Genitive/Ablative -11sj -lesi
5 Dative -11 -11
6 Approximative -lanj -lanj
7 Inessive -"in -"in
8 Elative -"isj -"isj
9 Illative -e -G)e
10 Terminative -e¢j -ozJ
11 Instrumental -en -G)en
12 Egressive -sjau-i -("i) sjen
13 Caritive -teg -tek
14 Prolative1 -ed -G)eti
15 Prolative2 -ti -ti
16 Adverbial -ja, -ji -ja
17 Consecutive -la
18 Comitative -ked

The Permiancasesuffixesfall into two main historicalgroups,primary and

secondary.The latter group may be further subdivided into those suffixes
which are built with the co-affix *-1- and thosesuffixes which are the result
of combiningtwo casesuffixes.
Primary casesuffixes: (2) the accusative,from pU *-m. The m has been
 PERMIAN 269

lost, and its role has been taken up by the stem-final vowel. In the forms
Zyrian -es, Votyak -ez the element slz is identical with the third-person
singularsuffix; seebelow. (7) the inessivecontinuesthe pU locative *-nA. (8)
the elative: it is very likely that this suffix is identical with the homophonous
nomenpossessiderivationalsuffix, e.g. Komi-Permiakver=isi 'forest spirit' ,
derivedfrom ver 'forest' with the suffix =isi 'belonging/pertainingto X'. The
-t- elementfound in the Votyak non-final variantof the elative(e.g. gurt-i"sit-i"d
'out of your house')canbe tracedbackto the pU ablative*-tA. (9) the illative,
from pU lative *-k:. Mter the k was lost, its role was assumedby the stem-
final vowel, as in the accusative.(10) the terminative continuespU lative
*-cjv. (11) the instrumentalprobably continuesthe pU genitival/adjectival
suffix *-n. (14) the first prolativecontinuespU ablative*-tA.
Secondarycase suffixes built with *-1-: the (3) genitive, (4) genitive/
ablative, and (6) approximativecasesare innovativeformationscombining a
coaffix *-1- with the descendents of pU *-nA, pPN elative *-sj, anda pU lative
*-nj. The Permiandative (5) and consecutive(17) casesare the result of a
bifurcation; both go back to coaffix *-1- plus the descendantof a pU lative
*-k.
Other secondarycasesuffixes: (12) in Zyrian, the egressivecombinesthe
elative -si with the reflex of pU lative *-n,i; in Votyak the secondcomponent
seemsto be an allomorph of the inessivecase.(13) The caritive is probably
a combinationof the pU caritive suffix *=ttA plus the pU lative *-k, but the
retentionof the final k is irregular. (15) the secondprolative continuesa pFU
locative *-ttV plus pFU lative *-j. (16) the adverbialis a combinationof some
p(F)U lative or locative suffix with lative *-k, but the detailsare unclear.(18)
the comitativeis probablya k-coaffix with locativemeaning(cf. the Cheremis
comitative-ke - -ye) plus the first prolative.
Possessive
Declension
As in most of Uralic, in the Permianlanguagespersonalsuffixes are usedto
indicatepossession.The basicinventoryis setout below:

Zyrian Votyak

sl -eG) -e, -i:

s2 -i:d -ed, -i:d
s3 -i:s -ez, -i:z
pI -ni:m -mi:
p2 -ni:d -tldi:
p3 -ills -slzi:

In the singularthe systemsof the two languagesare closeto identical (the

-j in the Zyrian sl is a secondaryelement,of diminutive/vocativeorigin). It
is in the plural that the two systemsdiverge. For pPN a different systemhas
270 PERMIAN

Table 8.4 Proto-Permian possessivesuffixes

Possessionin singular Plural

sl *-mV *-nV-mV
s2 *-tV *-nV-tV
s3 *-sV *nV-sV
pI *-mV-k *-nV-mV-k
p2 *-tV-k *-nV-tV-k
p3 *-sV-k *-nV-sV-k

beenreconstructed,in which plurality of the possessionwas expressedby a

suffix *-n- (known from several related languages),and plurality of the
possessorwas markedwith the pFP pluralizer *-k. The basicbuilding-blocks
of this reconstructedsystemare the pU personalpronounsof the first, second,
and third persons,*me, *te, and *se.
This reconstructionis continuedintact in the singularforms of the present-
day Permianlanguages(in the first personsingular-m was lost in most cases
and its role was assumedby the precedingvowel). Zyrian plural possessive
suffixes suchas pI -niin are the continuationof the pPN singularsystemwith
plural possession,i.e. the markerof plurality of the possessioncameto mark
plurality of the possessor.In Votyak on the other hand the suffixes of
analogousfunction (e.g. -mi) continue the old pPN plural suffixes with
singular possession.The loss of word-final vowels is regular, as is the
retentionof the vowel before a consonantwhich was later lost, as in Votyak
-mf< *-mV-k.
The generalrule (from which thereis a certainamountof deviation)for the
order of possessiveand case suffixes in both Zyrian and Votyak is that
primary case suffixes precede, and secondary case suffixes follow, the
possessivesuffixes. Examples:the terminativesof 'hislher house'are Zyrian
kerka-e¢-is,Votyak korka-oz!-az;the dativesof 'hislherdaughter'are Zyrian
nil-is-li: Votyak nil-ez-lr.
Conjugation
The conjugationalsystemsof Zyrian and Votyak resembleone anotherto a
great degree,but not to the extent that their declensionalsystemsdo. The
following remarkswill concentrateon the differencesmore than on the many
similarities.
Numberof ConjugationalPatterns
In p(F)U as well aslaterin pPN therewas only oneconjugationalpattern.This
is still the casein Zyrian. In Votyak, however,a major changehascomeabout:
historical phoneticchangeshave causedverbs originally ending in a certain
commonderivational suffix to be conjugatedin a different way from other
verbs. This suffix, pre-Votyak *=al-, was originally a frequentativeformant,
 PERMIAN 271

but this meaningis now no longerobviousin mostcases.The I elementof this

suffix now appears only in intervocalic position, while the a has been
preservedin all positions. The stem-final vowel is now synchronically
diagnostic:if it is i", the verb is of the first conjugation;if it is a, the verb is
of the secondconjugation(the synchronicstemis most easily obtainedfrom
Votyak dictionary citation forms by lopping off the infinitive suffix -nil The
most importantdifferencesbetweenthe two conjugationsare the presenceor
absenceof the -a(I)- sequence,and the fact that the primary past-tensesuffix
*-i- shows up only in the first personsingular of second-conjugationverbs
(vera-j 'I said'), while it runs throughthe entire paradigmof first-conjugation
verbs(min-i 'I went', min-i-d 'you went, min-i-z '(s)hewent', etc.).

Tense
Proto-Permianhad four non-compoundtenses: present, past, perfect (=
auditive, narrative),andfuture. The presentwas markedwith zero.The suffix
of the past tensewas *-i-, inherited from pU. The suffix of the perfect was
historically identical with that of the past participle, from pU *=mA. The
future tensewas morphologicallydistinct only in the third person;this is still
the casein present-dayZyrian: mune '(s)hegoes',muneni" 'they go'; munas
'(s)he will go', munasni"'they will go'. Otherwise,presentand future forms
syncretized.
Zyrian has retained the old pPN tense system. The perfect (auditive,
narrative)is usedonly in secondand third personsand the future is restricted
to the third person. This system has been augmentedby a number of
secondary,periphrastictenses,built for the mostpart with the verb vi!l- 'is'.
In Votyak, on the otherhand,the tensesystemhasundergonerestructuring.
Once again a frequentativesuffix is to blame: the derivational suffix =sik
cameto function as a present-tense marker.This morphemenow occursin the
first and secondpersonsof the present-tense paradigm,e.g. misik-isik-o 'I am
washing',dasia-sik-odi"'you (plur) arepreparing'.The future tense,originally
formally distinct only in the third person,now cameto differ from the present
tense in all persons,i.e. alongsidethe original distinction betweenmisik-e
'(s)hewashes'and misik-oz '(s)hewill wash' therearosethe new andparallel
distinction betweenmisik-isik-o 'I am washing' and misik-o 'I shall wash'.
With the auditive perfect(usedin all persons),Votyak thus hasfour full basic
tenses;like Zyrian, it also has a number of secondary,periphrastictenses,
built with the invariablepast-tenseforms val and vilem.

Mood
Proto-Permianhad only two moods, indicative and imperative, the pU
conditional(built with *-ne-) having vanishedwithout trace.In Votyak a new
conditional mood has arisen secondarily.Synchronically,the suffix is -sal-.
Historically, it is the result of fusion of the gerund(=sa) with the past-tense
auxiliary val mentioned above, thus the syntagm *mini"=sa val became
272 PERMIAN

restructuredas the finite form mfni~salwould'I would go; I would have gone'. In
Zyrian, the categoryconditionalis either expressedby the indicative plus the
particle(v)esi-ke or implicit in the context.
Personal Endings
Nearly all personalendingsfiguring in the Permianconjugationalsystemcan
be tracedbackto the personalpronouns*me *te *se or areultimately identical
with an old participial suffix. The Zyrian andVotyak systemsarequite similar
and the deviations are in most casesof easy explanation.The sole major
discrepancyis found in the Zyrian s2 ending-n, e.g.muna-n 'you go', mun-i-n
'you went'. The Votyak pendant,-d (as in mfnisikod 'you go', mfnid 'you
went'), is the expectedreflex « pU *te, the second-person singularpronoun).
Thus far no explanationfor this anomalyhas found generalacceptance.NB:
in both Zyrian andVotyak we find -d as the s2 suffix in the nominalparadigm.
It is true that we find the sames2 ending -n in the ObUgrianlanguages,but
this doesnot solvethe problem.
Negative Verb
Like most Uralic languages,both Permianlanguagesmake use of a special
verb of negation.In this verb, which is defective, are encodednumber and
personof the subjectas well as tense.Following the negativeverb the stem
of the main, lexical verb, is used.
In both Zyrian and Votyak the negativeverb hasfewer distinct forms than
thereare grammatical person-and-number distinctions.Syncretismof subject
numberis avoidedby suffixation on the lexical verb, e.g. Zyrian oz mun '(s)he
doesnot go', oz mun-nf 'they do not go', Votyak ed mfnf 'you (s.) didn't go',
ed mfne(le) 'you (plur.) didn't go'. For the full paradigms,seepp. 315,292.
The stem of the negative verb can be traced back to the pU negation
morpheme*e-*a. The past-tenseforms seenin Permiandevelopedfrom this
stemplus the past-tensesuffix *-i-.

Syntax
The Permianlanguagesare traditionally left-branching,i.e. determining or
modifying elementsprecedethe elementsdetermined/modified.In terms of
word order this meansadjectiveprecedesnoun, noun precedespostposition,
possessorprecedespossession,and - very generallyspeaking- direct object
precedesverb. Traditional Permian sentencestructure seemsnot to have
differed from that of other Uralic languages,which in turn is for all practical
purposesidenticalwith that of the typologically similar Turkic, Mongolic, and
Tungusic languages.One salient feature of this languagetype is the nearly
completelack of relativepronounsandconjunctions;this lack, in turn, entails
a lack of subordinateclauses.Sentencesubordinationcould, of course,be
made plain enough through mere juxtaposition and context, but the more
usualway to expresswhat (Indo-)Europeanlanguagestypically expresswith
 PERMIAN 273

subordinate clauses was by means of deverbal nominal and adverbial

constructions:participles, infinitives, gerunds, nomina actionis, using the
usual nominal case and person suffixes. The Permianlanguagespossessa
richly developedstore of such deverbal constructions,the greater part of
which goesbackto pPN.
So much for the traditional picture. The past decadeshave seenthe rapid
advanceof the 'Indo-European'model of sentenceextensionvia subordinate
clauses, the catalyst being chiefly Russian. Today all Permian literary
languagesuse a goodly number of both subordinatingand co-ordinating
conjunctions,many of which have been beenborrowedfrom Russian,e.g. If
'and' > 7JVo i, HO 'and, but' > 7JVo no, 6Y,[lTO 'as if > Z bftjtje, Vo budto,
etc. The interrogativepronouns'who' andwhat' (Zyrian kod, mij; Votyak kin,
ma) are nowadays also used as relative pronouns,this use involving the
employmentof caseand plural suffixes on the patternof RussianKOTOPbIH
and qTO. Zyrian mij 'what' is also used as a subordinatingconjunction; cf.,
oncemore,RussianqTO.
Russianinfluenceon Permiansentencestructureis by no meansnew; it has
simply becomemore markedin recentyears.What Hungarianunderwentin
the Middle Ages is now occurring in Zyrian and Votyak, namely a radical
restructuringof traditional sentencepatterns.

Contacts with Other Languages

Proto-Permianand its daughterlanguageshave come into contact with a
numberof otherlanguages,andthesecontactshaveleft their tracesin present-
day Zyrian and Votyak. Thesetracesare clearestin the lexica, but they can
also be demonstratedin the domainsof phonology,morphology,and- as we
have seenin the previous section- syntax. The discussionbelow is limited
to a brief surveyof the loanwordstratato be found in Permian.
BaseVocabulary of Proto-Permian
The basic lexical stock of pPN was dominatedby words inherited from pU,
pFU, and pFP and enrichedby loanwords,mostly from branchesof Indo-
European,adoptedprior to pPN. Non-borrowedvocabularyincluded items
such as p(F)U *nie(e)le 'arrow' > pPN *nil>l > Z riev Vo riel, pFU *sappa
'gall (bladder),> pPN *sap > Z sepVo sep.
In addition,we canidentify an importantlayer of pPN words which arenot
demonstrablyinherited or borrowed.Examplesare pPN *ker 'log, beam' (Z
ker, Vokor), *dUz 'membrane'(Zdis, Vodiz), *gl>rd 'red' (Zgerd,Vogord),
*bOshi- 'takes'(Z bosit-, Vo basitf-).
Proto-Permianhad strongcontactswith Iranianlanguagesand a numberof
Iranian loanwordsin pPN havebeenidentified, e.g. *amV~amaa~), 'ploughshare'(Z
ami:/, Vo amez/; cf. present-dayPersianamaa~),amaa~), *das 'ten' (7JVo das; cf.
present-dayOssetiandiis).
Of equal importancefor pPN were the contactswith Middle Bulgarian
274 PERMIAN

(MB, an earlier stage of the branch of Turkic from which present-day

Chuvashdescends);thesecontactsbeganapproximatelyin the eighthcentury.
This stratumof Turkic loanwords, theearliestin the Permianlanguages,can
be distinguished from later Turkic loanword strata, both later Middle
Bulgarianand laterTatar,by their distribution (the oldestMB loansarefound
in both Zyrian and Votyak, later ones only in Votyak - and possibly Komi-
Permyak) and by means of phonetic criteria. For example, Votyak bam,
Zyrian ban 'face' are from pPN *bau < MB *b1i.ij - *beI), but Votyak murjo
'chimney' is a later loan from Chuvash,andVotyak azbar 'courtyard'is from
Tatar.

Zyrian Particulars
The move northwards which gave rise to the separateZyrian nation and
languagealso determinedwhich languageswould be sourcesfor new loans.
Through loss of direct contactZyrian could no longer borrow directly from
MB, or later from Tatar, but as Zyrian settlementsspreadto the north, west,
and east, they came into contact with Baltic-Fennic languagessuch as
Karelian and Veps, with Nenets, and with the ObUgrian languages.The
number of loans from theselanguagesis not great and they are to a large
extentrestrictedto the Zyrian dialectsspokenfartherto the northwest(Baltic-
Fenniccontacts),the north (Nenets),and the northeast(ObUgrian).Nonethe-
less, someof theseloanwordshave spreadto all areasof Zyrian, e.g. joma
'witch' < Baltic-Fennicjumala 'god', Pampa 'ski' < Nenets lampa, jaran
'Samoyed;northerner'< Ostyak (cf. SouthernOstyakjaran) or Vogul (cf.
NorthernVoguljaar~).
northerner'
Of much greaterimportancefor Zyrian are the loans from Russian.The
contactsbetweenZyrians and Russiansbeganlong before the conversionof
the former to Christianity, and the flow of loans grew with the passingof the
centuries.One can distinguishbetweenan older stratumof loans adoptedby
various dialectsand adaptedto someextentto the Zyrian soundsystem(e.g.
visit 'news' < north Russianv'is't', gi'rnjid 'pot', cf. Old RussianropHhl\h)
and more recentloans, adoptedin greaternumbers,which preservemore
closelythe soundqualitiesof the original.

Votyak Particulars
Three languagesshapedthe vocabularyof Votyak: Middle Bulgarian,Tatar,
and Russian.These are also the three peoplesunder whose hegemonythe
Votyak havelived to the presentday.
Unlike the Zyrians, the Votyaks remainedwithin the sphereof Middle
Bulgarian culture until the Golden Horde put an end to that kingdom.
Accordingly, there are many more Bulgarian loans in Votyak than there are
in Zyrian. Examples: busi' 'field' (cf. Chuvash pusi'i), kuno 'guest' (cf.
Chuvashxi'ina),uksio 'money' (cf. Chuvashuksa).
Until the fall of Kazan in 1552 the Votyaks formed a part of the territories
 PERMIAN 275

subjectto the Tatars,and there are many Tatar loanwordsin Votyak to attest
to this. Not all suchloansarefound in all Votyak dialectsor usedin the literary
language; many are restricted to dialects spoken in areas close to, or
co-territorial with, Tatar. Examples: batir 'hero(ic)' (cf. Tatar batir), gine
'only' (cf. Tatargenii), taza 'healthy,strong' (cf. Tatartaza).
In the sixteenth century the Tatars were supplantedby the Russiansas
rulers of the Votyak territories.As in the caseof Zyrian, one can distinguish
an older and a youngerlayer of Russianloans.The older loanwordstook part
in Votyak-specificsoundchanges,e.g. ukno 'window' < RussianOKHO, dusko
'board' < I:(OCKa. In the nineteenthand twentieth centuriesVotyak has been
inundatedby Russianloanwordswhich for the most part have retainedtheir
original shape.

Referencesand Further Reading

Csucs,S. (1990) Chrestomathiavotiacica, Budapest:Tankonyvkiad6.
Lallukka, S. (1990) The East Finnic Minorities in the Soviet Union, Helsinki:
SuomalainenTiedeakatemia.
Lytkin, v.I. (1961) KOMIHl3hBHHCKHH j:\HaJIeKT, Moscow: Academy of Sciencesof
the SovietUnion.
- - (1964) I1cTOpH'IeCKHH BOKaJIH3M rrepMCKHX 1I3hIKOB, Moscow: Nauka.
Lytkin, V. I. , and E.S. Gulaev (1970) KpaTKHH 3THMOJIOrH'IeCKHH CJIOBaph KOMH
1I3hIKa, Moscow: Nauka.
Redei, K. (1978) Syrjiinische Chrestomathie.Mit Grammatik und Glossar, Studia
Uralica 1, Vienna: Finno-UgricInstituteof the University of Vienna.
- - (1988) 'Geschichteder permischenSprachen',in D. Sinor (ed.) The Uralic
Languages: Description, History and Foreign Influences, Handbuch der
Orientalistik8/1, Leiden: Brill, pp. 351-94.
- - (1988-91)(ed.) Uralischesetymologisches Worterbuch,Budapest:Akademiai
kiad6.
Serebrennikov,B.A. (1963) I1cTOpH'IeCKali MOP<POJIOrHlI rrepMCKHX 1I3hIKOB, Mos-
cow: Nauka.
Uotila, T.E. (1933) Zur GeschichtedesKonsonantismusin denpermischenSprachen,
Helsinki: SocieteFinno-Ougrienne.
9 Udmurt
SandorCsucs

In the 1989 Soviet census747,000 people declared themselvesto be of

Udmurt nationality; of these approximately 70 per cent (thus c.520,000
people)speakthe languageof their ancestors.
Two-thirds (c.500,000) of the Udmurt (older name: Votyak) live in the
Udmurt Republic(Udmurtia),which is locatedbetweenthe Vyatka and Kama
Rivers and forms a part of the RussianFederation.The areaof Udmurtia is
42,000 km2 , its population is 1.6 million; its capital is the industrial city
Izhevsk, with 635,000 inhabitants.The Udmurt are thus a minority within
their own republic. (Breakdownby nationality: Udmurt 31 per cent, Russian
59 per cent, Tatar and other 10 per cent.) The distribution of the Udmurt
populationis not homogeneous:their proportionsare greaterin villages than
in the cities. So for example 16.5 per cent of the population of Izhevsk is
Udmurt, but in Alnasi provincethe Udmurt make up more than 80 per cent.
This disproportionis characteristicof languageuse, as well. Generallyin the
villages 95 per cent of the Udmurt population speaktheir mother tongue,
while in the cities the proportion is considerablylower. This stateof affairs
is partly due to the fact that in urban schoolsno provision for educationin
Udmurt was providedunderthe Sovietsystem.
Approximately 100,000 Udmurt live in the neighbouring republics of
Bashkiriaand Tatarstan,and in the Vyatka (Kirov) and Perm' regions.These
Udmurt havelived in the sameplacesfor severalcenturies,almostexclusively
in villages, and have in general preservedtheir languageskills well. The
proportionof mother-tonguespeakershereis between80 and90 percent.
The remainingroughly 150,000Udmurt live in scatteredsettlementsin the
former Soviet Union, chiefly in towns. This dispersionoccurredduring the
Soviet period; the motivation to move was to lesseneconomic hardship.
Among this populationthe proportion of mother-tonguespeakershas fallen
to 40-50 per cent. With time, the majority will probablybecomecompletely
Russified.
At least 90 per cent of the Udmurt are bi- or trilingual. The second
languageis generallyRussian,the third Tatar. Udmurt who live in Tatarstan
and Bashkiria speakTatar as their secondor third language.There is great
variety in the degreeof competencein the secondandthird languages;factors
include place of residence,amountof schooling,occupation,age, social and

276
 UDMURT 277

societal status and family circumstances.The young and the intelligentsia

speak Russianperfectly, but when communicatingwith one another con-
sciously strive to use their mother tongue. In the countrysidein everyday
situationsUdmurt is spoken,but at official functions Udmurt speakRussian
even amongthemselves.It will be interestingto observeto what degreethe
political eventsof the early 1990swill influencelanguageuse.
Russian and Udmurt are quite distinct from one another, but I have
occasionallywitnessedthe mixture of the two. In oneinstance, Iwas with the
headof a kolkhoz in his office. He was a native speakerof Udmurt; we were
discussing the kolkhoz. The person accompanyingme was posing the
questions,and our host duly tried to reply in his native language;but he kept
slipping into Russianor, at least, mixing Russianexpressionsand sentences
into his speech.The difficulty was causedby the fact that he had acquiredthe
habit of dealing with official matters,in his office, in Russianor Tatar, and
not in Udmurt. On many occasionsI have observedconversationsamong
studentsor teachersin which extensivecode-switchingto Russianoccurred
(phrases,half-sentences,proverbs).Such code-switchingis usual whenever
the subject-matteris betterexpressedin Russian.
BetweenUdmurtdialectsthereareno significantdifferences.The syntactic
and morphological systemsare nearly identical acrossthe entire linguistic
area.The phonemicsystemsof the centraldialectsof Udmurtia are identical
with eachother and with that of the literary standard.The only real deviation
is in the distributionof phonemes,e.g. the cacuminalaffricates(c and ~) occur
word-internally only in the central dialects, while other dialects and the
literary standardhavethe correspondingfricatives.
Common Udmurt i' has reduced,retracted,and lowered variants in the
northern and southerndialects. This feature is also typical of the dialects
spokenoutsideUdmurtia, i.e. to its west, south, and east,chiefly in the two
Turkic republics.Thesedialects are often lumped togetheras a 'peripheral'
group. Their sharedisoglossesare word-internal -1)- and the retention of
proto-Udmurt *(j and *1i, seepp. 263-4. In some peripheraldialects a new
(front openunrounded)ii phonemehasarisenthroughTatarinfluence.Strong
Tatarinfluencein thesedialectsis evidentin otherdomains,as well.
The lexicons of the various dialects differ only in non-centraldomains.
Regionalismsare usually deviant only in form or in meaning, but true
regionalismsdo occur, e.g. macPi'ni' 'sings - 'tells', mumi' - anaj 'mother'.It
is not uncommonfor one dialect to borrow a Russianword, and another
dialect to borrow a synonymousword from Tatar, e.g. zavod(from Russian)
- arberi (from Tatar), both 'stuff, rubbish'. Often both words are at homein
the literary lexicon.
Dialectal peculiarities thus present little or no obstacle to mutual
comprehension.This is not least due to the fact that the literary standard,
which is taught in schools and used in the media, is a felicitous alloy of
the dialects, basedon the central dialects but amply enlargedby northern
278 UDMURT

andsouthernfeatureswhich arenot stronglydivergent.

The developmentof literary written Udmurt began in the eighteenth
century, but Udmurt-Ianguagepublicationsdid not begin to appearuntil the
middle of the nineteenthcentury.
Before the 1917 revolution the demandfor the developmentof a unified
Udmurt literary languagedid not arise. Publicationswere written in the local
dialect, but the small interdialectaldifferencesmeantthat - as far as material
circumstancesallowed - these works were theoretically able to become
commonintellectualpropertyfor the entireUdmurt population.
During the 1920stherewaslively debateoverquestionsof orthography.As
elsewhere,among the Udmurt, too, there arosethe idea of switching to the
Latin alphabet.Numerousarticles on this questionappeared,but there had
been no tradition of the use of Latin letters in writing Udmurt (linguistic
transcriptions aside), nor was the cultural milieu favourable for such a
solution. The idea was thus more or less definitively rejected. The use of
Cyrillic on the other hand had an extensivetradition: since the eighteenth
century,every Udmurt publicationthat appearedin Russiawas in Cyrillic. A
seconddebatearoseamongUdmurt writers andlinguists over the questionof
how Cyrillic was to be implemented.One campwantedto developtheir own
orthography,basedon Cyrillic letters, but adaptedand augmentedto accord
with Udmurt phonology;the opposingcamp arguedfor the importation not
only of the letters,but of Russianorthographicprinciples, as well. The latter
sidewon the argument.This led to the adoptionof the entireRussianalphabet
along with manyof its orthographicrules.In this system,which is valid to this
day, the three Udmurt affricatesand one Udmurt vowel which Russianlacks
are indicatedby Cyrillic letters with diacritical marks q = c, i: = ¢, 3 = ~,
6 =if). A further diacritical letter, Hwas introducedto write the vowel i when
this sound was precededby a distinctively non-palatalizedconsonant,i.e.
after t, d, n, t, s, z (the letter with the analogousfunction in the Komi alphabet
is i). Thus HH = ni, but HH = nii, and MH representsmi, becausem has no
distinctively palatalizedpendant.It would be moreconsistent,andtechnically
simpler, if the distinctively palatalizedconsonantswere uniformly indicated
with the aid ofthe soft sign (h); this solutionis alreadyemployedat word-end
and before the vowel r (written hI). A further oddity of the present-day
orthography is the practice of preserving the Russian spelling of recent
Russianloanwordsintact, e.g. 'dictionary' is written CAOBaph in spite of the
fact that Udmurt hasno palatalizedri phoneme.
A good feature of the orthography, and a vital one, is its ability to
distinguishunambiguouslyevery Udmurt phoneme- albeit with the aid of a
somewhatcomplexsystemof rules.
This is not the place to go into the history of the Udmurt language,but a
brief listing of importantinnovationsandretentionswill be of use.Prominent
phonologicalinnovationsinclude the developmentof the voiced : voiceless
oppositionamongthe obstruents,the simplification of the vowel system,and
 UDMURT 279

the large-scaleerosion of the ends of words. As a result of thesechanges,

present-dayUdmurt is morphophonemic ally straightforward,a mainstream
agglutinatinglanguagewith very few stemor suffix alternations.
A prosodicinnovationis the shift of the word stress,partly as a result of
Tatarinfluence,to the last syllable.
The main morphologicalinnovationsarethe hypertrophiccasesystemand
the development,in the verb, of future andconditionalparadigms.
Against the backgroundof all theseinnovationsone should mention that
Udmurt, like Komi, has retainedthe older proto-FU distinctions amongthe
s(h)ibilants and affricates, as well as the systemof personalsuffixes. The
extensive use of verbal nominals, including that of asyndetic sentence-
linking, is perhapsalsoa retention.

Phonology

Consonants
The velar nasal (0), which probably belongedto the inventory of proto-
Udmurt, hasbeenlost in all savecertainperipheraland southerndialects.The
post-alveolaraffricatesc, written
~have (apico)retroflexarticulations.
The statusof the soundwritten w in the table is problematic.This sound
occursin threecontexts:
1 Word-initially before the vowel a in about twenty words, e.g. wani
'exists'.This is a dialectalphenomenon;the literary standardhasvani.
2 After word-initial k in a few dozen words, e.g. kwala 'family shrine',
kwini 'three'.This phenomenonis characteristicof the entirelanguage.
3 Word-internally and word-finally after vowels in the Shoshmadialect
(older name:Malmizh-Urzhum).

Table9.1 The Udmurt consonantsystem

1 2 3 4 5 6

Nasals ill n ni (IJ)

Voicelessstops p t ti k
Voiced stops b d cti g
Voicelessaffricates cc i
Voiced affricates ¢ ¢i
Voicelessfricatives vs s s si (x)
Voiced fricatives z z zi
Laterals I Ii
Rhotic r
Glides (w) s
Note: Placesof articulation: 1 bilabial, 2 labiodental,3 (apico)dental,4 post-alveolar,5 palatal,
6 (palato)velar.
280 UDMURT

The salient characteristicfeature of the Udmurt consonantsystemis the

presenceof the correlationsof voice and palatalization.Nine consonant-pairs
participatein the correlationof voice, eight in that of palatalization.In fact,
only four consonantstakepart in neithercorrelation;theseconsonantsmay be
therebydefinedas peripheral(v,j, m, r).
There are no long consonant phonemes, but geminates do arise at
morphemeboundaries,often as the result of assimilation, e.g. amedial ~ ~ini" 'to
look', li"kti[i[iam 'they came'.
With the exceptionof 1], all consonantsmay occurinitial, final, andmedial
in the word; certainconsonantsare rare, however,in certainpositions.So for
examplethe palatalstopstj and dj are quite rare in word-initial positionin the
literary standardand in the majority of the dialects,as are the post-alveolar!
cacuminalaffricatesin word-internalposition.
Clustersof two consonantsare not uncommonin word-internaland word-
final position; clustersof threealso occur,e.g. murtli" 'to a person'.In clusters
of three,oneof the consonantsis usually a liquid.

Vowels
The literary languageand the majority of the dialectshave the sameseven-
vowel system,one sharedwith Komi:

Front Back
Unrounded Rounded
High 'i 'i u
Mid e e 0
Low a

The e is open-mid [E], especially in word-final position; the 0 is

pronouncedquite farback.
In suffixes, there are no examplesof e; u is rare; and 0 occurs only in
certain verb suffixes (e.g. mi"no 'I shall go') and in the noun plural suffix
-(j)os.
Certain dialectshave vowel systemswhich differ from that given above.
Thereare also combinatoryvariantsof the above-listedvowels.
Vowel sequencesare rare, and occurusually at morphemeboundaries,e.g.
korka-os HOUSE-plur 'houses'.

Stress
The stressis of the quantitative-dynamictype and generallyfalls on the last
syllable. A stressedsyllable is approximately 1.5 times as long as an
unstressedone. Unstressed vowels do not undergoreduction.
In certain verb forms, stresslocation distinguishesmeaning: sik '(s)he
stands',sj:.1e 'stand! (plural imperative),; verg '(s)he speaks',Vf.ra 'speak!
(singular imperative)'.As theseexamplesillustrate, in imperativeforms the
 UDMURT 281

stress falls on the fIrst syllable. Initial-syllable stress also characterizes

negativeverb forms (e.g. ud mfniSike'you [plur] do not go') andreduplicative
adjectivesandadverbs(e.g. gQrd&gord 'very red').
Certain adverbsand pronouns(with first componentog-, olo-, no-, koti -)
and certainaffective-onomatopoeic words havefree stress,i.e., in thesecases
stressplacementdependson expressiveandlogical featuresof the utterance.
Morphophonemics

DeletionsandAdditions
Vowel syncope.Certain stemsshow syncopeof their last vowel (usually n
when a vowel-initial suffIx is added.Adjacent to the syncopatingvowel is
always at leastone liquid or nasal.Examples:turin 'hay', tumani" 'to mow';
polis 'oar',polsani"'to row'; koti"r 'circle', kotres 'round'.
Paragogic consonants.Certain stemsshow a paragogicconsonantto the
left of vowel-initial suffIxes. The consonantis always one of the set k t m.
Examples:niules 'forest',niulesk-in 'in the forest', but niules-mi"'our forest';
in 'sky', inm-in 'in the sky'; ki"k 'two' , ki"kt=eti 'second'.

Hiatus-blockerj
Aj-glide is epenthesized in the literary standardbetweena stem-finalvowel
andcertainvowel-initial suffIxes (e-initial suffIxes,the adjective-formant=0,
and the verb-formant =a[lJ-). Examples: kuno-e > kunoje 'my guest',
uksio=o > ukSiojo 'moneyed',gubi=al- > gubijal- 'goesmushroom-picking'
(from kuno 'guest', ukSio 'money', and gubi 'mushroom'). In dialects,
j-epenthesisis morewidespread.
Assimilation
Assimilationofvoice, general
In a sequenceof two obstruentsC)C2, C) adjustsin voicing to matchC2 . This
assimilation is not indicated by the orthography. Examples: pereSi 'old',
comparativeperef/=gem'older'; lud 'meadow',prosecutivelut-ti 'along the
meadow';goz=ja-ni" 'to write', 'meadow',
gos=ti~ni" 'to write down'.
Adaffrication
The dental stops t and d melt into, and lengthen, the stop elementof the
affricatesCi and ~ whentheseimmediatelyfollow, e.g. kut 'sandals',kutCiani"
pull 'to pull on (one's)sandals';li"d 'number',li"d¢ini"[J:?']
[c:~] pull 'to count'.
Assimilationof voice, special
This assimilationaffectsonly the possessivesuffIxes.The initial consonantof
the second-person suffIxes is t or d, and that of the third-personsuffIxes is s
or Z, in assimilationto the precedingconsonant.(To the right of vowels d Z
occur.) Notice that the direction in which this assimilationoperatesis the
reverseof that of generalassimilation,above.Examples:val-de 'your horse
(ace)',e.s-te'your friend (acc)', val-Zi" 'their horse',e.s-si"'their friend'.
282 UDMURT

Inflectional Morphology
Nominals (including postpositions,see p. 293) and verbs take inflectional
suffixes as detailed in the sections immediately following. Conjunctions,
particles,onomatopoeicwords andinterjectionsdo not inflect.

Nominals
The noun (and in certain instancesthe adjectiveand numeral)distinguishes
the following categories:number,case,and person(= possessor).All three
categoriesareindicatedby suffixes.
In terms of their stem structure,Udmurt nouns are unusuallystraightfor-
ward. With very few exceptions,all nouns belong to a single, unvarying,
declension-type.

Number
The singularis unmarked;statedin termsof the orthography,the suffix of the
plural is -jos to the right of consonants,-os to the right of vowelsU-epenthesis
is regular to the left of the -os suffix.) Examples: korka 'house',korka-os
[korkajQs] 'houses'; bekCie 'barrel', bekiJeos [bekCiejQs] 'barrels'; murt
'humanbeing',murt-jos 'people';derem 'shirt', derem-jos'shirts'.
The plural suffix alsofunctions as a collectivemarker,e.g. mumi" 'mother',
mumi"-os-izMOTHER-plur-s3 'motherandher relatives/associates'.

Case
There are fifteen cases.Every noun is declinable,but accordingto recent
grammarsnounswhich designateanimatesdo not take local cases(inessive,
illative, elative, transitive, egressive)and usagewith the approximativeand
terminative fluctuates. Observationin the field shows that thesecasesare

Table9.2 Absolutedeclension,singular:val 'horse',busr 'field'

Nominative -{} val busf

Genitive -len yellen busllen

Ablative -lesi vallesi busllesi
Dative -11 vam busHi
Accusative -O, -(j)ez valez busljez
Instrumental -(j)en, -In valen busljen
Abessive/Caritive -tek valtek busltek
Inessive -In busi-in
Elative -lsi busi-isi
Illative -(j)e buslje
Approximative -lani vallani busllani
Egressive -Isieni busi-isieni
Transitive -([jJe)ti busljeti
Terminative -ozJ buslozi
Adverbial -ja valja buslja
 UDMURT 283

in fact used with animates,but in non-local meanings,e.g. disetisJ-fn uza

TEACHER-ineWORKS.s3pres'(s)heworks as a teacher'.
The j-initial allomorphsoccur to the right of vowels. Nouns referring to
placesand ending in a, most notably korka 'house',take vowellessvariants
of the local suffixes, e.g. inessivekorka-n, elativekorka-sJ,illative korka-f1.
A closedsetof nouns(chiefly body-partandkinship designations)take the
-fn allomorphof the instrumental,e.g. ki-fn 'by hand'.Anotherclosedsettake
-rin the illative, e.g. niulesk-f'into the forest'.
The transitivevariants-eti - -ti appearto be in free variation.
The zero-accusativeis used with indefinite direct objects. For further
informationconcerningtheusesofthecasesseeSyntax,pp. 295 and297-9.
For the most part, plural caseforms are built by adding to the stem the
plural suffix followed by the samecasesuffixes as thoseusedin the singular,
e.g. abessive/caritiveval-jos-tek'without horses',approximativebusi~os-lanj
different
'towardsthe fields'. In a few cases,however,the casesuffixes havedifferent
forms in the plural:

Singularaccusative val-ez busl-jez

Plural accusative val-jos-ti: busi-os-ti:
val-jos-lz busi-os-iz
Singularinstrumental val-en busl-jen
Plural instrumental val-jos-ln busl-os-ln
Singularillative busi-je
Plural illative busl-os-l
Singulartransitive busi-jeti
Plural transitive busi-os-ti

The variants of the plural accusative(e.g.valjosiz)

valjosti~ valjosiz) are in free
variation in the literary standard(the former is characteristicof northern,the
latter of southernandperipheraldialects).
The shapeof the instrumentalsuffix in the plural (-in) meansthat plural
inessiveand instrumentalcaseshavesyncretized(e.g. busi~os-in). busi~os-in).

Possessive Forms
The allomorphic distribution is fairly straightforward:j-variants to the right
ofvowels,j-Iessvariantselsewhere.Thereare, however,exceptions,e.g. ku-e
'my skin'.
All singularpossessives with initial e have variantswith f which are used
with a closedset of nouns, most of which designatebody parts, kinship, or
(periods of) time, e.g. jfr-f 'my head',milkill-iz 'hislher mood', lui-ill 'your
soul', tusbuj-iz 'hislherexterior',drr-f 'my time', nunal-iz 'hislherday', nil-ill
'your daughter',nil-in-fm 'with my daughter'.Not all words belonging to
thesesemanticareasfollow the rule, however,e.g.jfrsJi-je 'my hair'.
284 UDMURT

Table9.3 Possessive
paradigm:possessionin singular

sI s2 s3

Nominative -(j)e, -I -(j)ed, -Id -(j)ez, -IZ

Genitive -efilen -efidlen -efizlen
Ablative -efm -efidl"i -efizl"i
Dative -efilesi -efidlesi -efizlesi
Accusative -me -te, -de -se,-ze
Instrumental -efinlm -efinld -efinlz
Abessive/Caritive -efitek -efidtek -efiztek
Inessive -am -ad -az
Illative -am -ad -az
Elative -isit"im -isi@ -isitiz
Egressive -isienim -isienid -isienlz
Transitive -efitim -efitid -efitiz
Terminative -oziam -oziad -oziaz
Adverbial -efija -efidja -efizja

Table9.4 Possessive
paradigm:possessor
in plural

pI p2 p3

Nominative -mi -ti, -di -si, -zi

Genitive -mllen -tldllen -slzllen
Ablative -mllesi -tldllesi -slzilesi
Dative -mm -tldm -slzm
Accusative -mes -tides -sIzes
Instrumental -enimi -enidi -enizi
Abessive/Caritive -mllani -tldllani -slzllani
Inessive -ami -adi -azi
Illative -amI -adi -azl
Elative -isit"imi -i'sit"idl -isitizi
Egressive -isienimi -isienidi -isienizi
Transitive -etimi -etidi -etizi
Terminative -oziami -oziadi -oziazi
Adverbial -mija -tldija -slzija

Examplesof the slz and tid alternation(specialassimilation,above):busi"-

di" 'your (plur) field', gurt-ti" 'your (plur) village', pud-de 'your foot (acc)',
gurt-se 'hislhervillage (acc)'.
The order of caseand personsuffixes is of two types.In the first type, the
person suffix precedesthe case suffix. This is the order in the genitive,
accusative,dative, ablative, abessive/caritive,adverbial, and approximative
cases.Example: val-ed-tek'without your horse'.The reverseorder, i.e. case
before person, occurs in the remaining cases,e.g. busi"-jeti-d 'along your
field'.
 UDMURT 285

The only striking allomorphy amongthe caseand personsuffixes which

resultsfrom their co-occurrenceis in the accusative(-me, -tide, -size, -mes,
-tides, -sizes),the inessive/illative(-am, -ad, -az, -ami: -adi: -azi"), and the
elative(-i"Sit- insteadof -is:i).
Plurality of possessionis expressedby the ordinary plural suffix -(j )os
followed by the possessivesuffixes as set out above, with the following
stipulations.Initial f is always used, and never initial e, in both possessive
suffixes and in the instrumental case suffix; the transitive suffix is -ti.
Examples:val-jos-fHORSE-plur-s1'my horses',pi-os-iz-lfSON-plur-s3-dat
'to his/hersons',val-jos-iit-iinfHORSE-plur-ins-p1'with our horses',uram-
jos-ti-z STREET-plur-transitive-s3'alongits streets'.
The possessiveforms are usedprimarily to expresspossessiverelationsin
both narrow andbroadsensesof the term, e.g. miitam tir-e PRO.s1genAXE-
s1 'my axe', Bagirdi Siin-jos-iz BLUE EYE-plur-s3 'his/her blue eyes',
korka-len Pipet-ezHOUSE-genROOF-s3 'the roof of the house',nil-iz-len
nil-iz DAUGHTER-s3-genDAUGHTER-s3 'his/her daughter'sdaughter',
so-lenmilkia-iz PRO.s3-genMOOD-s3 'his/hermood',pinal-jos-len nil-iz-len
kir~an­
zfCHILD-plur-gen SONG-p3 'the children'ssong',nil-len Beber-ezGIRL-
genBEAUTY-s3 'the girl's beauty'.
The first-person singular suffix is also used in affectionate forms of
address,e.g. if/, ii/ giai'k-e, nuni'ka-je, Beber-e 'sleep, sleep, my dove, my
child, my beauty'.
The second-and third-personsingular suffixes are also used to express
definiteness or determinedness.Second-personforms are rarer in this
function, and seemto have affective overtones,e.g. vatiSik-i-z sundf-jedteP
Sier-fCONCEAL.refl-past-s3 SUN-s2FORESTBEHIND-ill 'the (dear) sun
hashiddenbehindthe forest'.
There is widespreaduse of the third-person singular suffix (-ez) as a
stylistically more or less neutral meansof marking definiteness.It functions
muchasthe definite articlein Hungarianor English,e.g. guzdorvil-iit turiit-ez
BeberCLEARING SURFACE-ineGRASS-s3BEAUTIFUL 'the grassin the
clearing is beautiful'. It can be addednot only to nouns, but also to fully
declined nouns, thereby rendering them both definite and susceptibleto
further declension,e.g. gurt 'village', genitive gurt-len 'of the village', with
addeds3 gurt-len-ez'that of the village', metagenitivegurt-len-ez-len'of that
of the village', metainessivegurt-len-az'in that of the village'.

Adjective
The adjectiveis formally not distinct from the noun,but it hasmorphological
and syntacticpropertiespeculiarto it.
Chief amongtheseis comparison.The absolutedegreeis unmarked;the
comparativeis built with =ges (or =gem, in free variation). Example: kuzi
'long', kuzj=geslkuzj=gem'longer'.The superlativeis built with the help of
particlesor by meansof reduplication,e.g. piBi 'small', samojpiBi-jez 'the
286 UDMURT

smallest ones'; tuz kuzi 'longest' (tuz 'very'); sied 'black', Sied&Sied
'blackest',viP 'new', vanimiz-leSivili EVERYTHING-abl NEW 'newest(=
thaneverythingnew)'.
Plurality in the predicateadjectiveis indicatedby the suffix -eSi, e.g. kiset
liz 'the cloth is blue', kor-jos kuzi-eSi no zek-eSi'the beamsare long and
thick' .
Most Udmurt adjectivescan be used as adverbswith no changein form
(umoj 'good; well'). The meaning of some adverbs(e.g. kema 'for a long
time') canbe intensifiedby means notunlike that of the superlative(kema-leSi
no kema'long, long ago; the very longesttime ago').

Numerals
Numerals used as quantifiers in noun phrasesdo not decline or otherwise
agreewith the head; the headis usually in the singular, occasionallyin the
plural, e.g. so tue tiami's kion kut-em PRO.s3THIS.YEAR EIGHT WOLF
CATCHES-s3perf'(s)hecaughteight wolves this year', vorgoron-
ta~rte viti vorgoron-
jos Bori'ga-ni' kosk-i-zi' THIS EVENING FIVE MAN-plur FISHES-inf
DEPARTS-past-p3'this eveningfive men went fishing'.
Used as nouns, the numeralscan be fully declined, in both absoluteand
possessivedeclensions.In compound numerals only the last number is
declined.
Ordinals are formed with the suffix =eti, e.g. niil=eti klass 'fourth class'.
The definite (i.e. s3) forms of ordinals are also declined,e.g. kwinim=eti-jez-
len 'of the third'.
Fractions are formed with =mos: viti=mos '(one-) fifth', ukmi's=mos
'(one-) ninth', kwini kwati=mos 'three-sixths'.
Approximatenumberis expressedeitherwith the word og or the adjective-
forming suffix =0, e.g. og viti=ton murt 'aboutfifty people',das=o minut-jos
'aboutten minutes'.

Pronouns

PersonalPronouns
The personalpronounsdistinguishnine cases;they lack the inessive,illative,
elative,transitive,terminativeandegressivecases.
The declensionof the third-personpronoun is analogousto that of the
noun.

ReflexivePronouns
Threesimilar but distinct roots figure in the paradigmof the Udmurt reflexive
pronoun. These are as-, asi -, and aB-. The first root (as-) also occurs
uninflectedin the meaning'(one's)own', e.g. as nih 'my own daughter',and
with postpositions,e.g. as dor-az 'at hislherown home'.
 UDMURT 287

Table9.5 Udmurt personal pronouns: singular

sI s2 s3

Nominative mon ton so

Genitive mi"nam finad solen
mi"nesjti'm j
Ablative ti'nesiti"d soles
Dative mi"ni'm ti"ni"d sol'i
Accusative mone tone soje
Instrumental monen(i"m) tonen(i"d) soin
Abessive/Caritive montek tontek sotek
Adverbial monja tonja . soja
j solanj
Approximative monlan tonlanl

Table9.6 Udmurt personal pronouns: plural

pI p2 p3

Nominative mi ti soos
Genitive miljam tiljad sooslen
Ablative miljesiti'm tiljesjtYd sooslesj
Dative miljem(h) tiljed(l'i) soosl'i
Accusative miliemi"z tiljedi"z soosiz
miljemdi" tiljeddi" soostY
Instrumental miljemi"n tiljedin soosi"n
mWeni"mi" tiljeni'di'
Abessive/Caritive mitek titek soostek
Adverbial mija . tija soosja .
Approximative milan! tilani sooslan!

Table 9.7 Udmurt reflexive pronouns: singular

sI s2 s3

Nominative a()iim acjid acjiz

Genitive aslam aslad aslaz
Ablative aslesiti"m aslesjtYd aslesiti'z
Dative asl'im asl'id asl'iz
Accusative asme aste asse
Instrumental aseni'm aseni"d aseni'z
Abessive/Caritive acjimtek acjidtek aCiiztek
Adverbial aCiimja aCiidja acjizja
Approximative acjimlani acjidlani aCiizlani
288 UDMURT

Table9.8 Udmurt reflexive pronouns:plural

pI p2 p3

Nominative asimeos asiteos asiseos

Genitive asimelen asitelen asiselen
Ablative asimeles i asitelesi asiseles i
Dative asimeli: asitell asiseli:
Accusative asimemlz asitedYz asisezlz
asimedl asitedl asisezl
Instrumental asimemln asitedYn asisezln
asimeos"in asiteosln asiseosln
Abessive/Caritive asime(os)tek asite(os)tek asise(os)tek
Adverbial asime(os)ja asite(os)ja asise(os)ja
Approximative asime(os)lani asite(os)lani asise(os)lani

Other Pronouns
Reciprocal pronouns arebuilt by attaching the appropriateplural personal
suffixes to reduplicatedo(diJg 'one': og&og-mi~lr
ta-os-iz 'to one another(of us)',
og&og-en-fmf'with one another(of us)', og&og-di~tekta-os-iz 'without eachother
(of you [plur]), ,og&og-zes'eachother(ofthem[acc]), .
With the exception of the definite forms ta--iz and so--iz 'this/that
particular,preciselythis/that, this/thatvery', the demonstrativepronounsare
not inflectedwhenusedattributively. Example:ta udi"s-i"n THIS REGION-ine
'in this region'.The inflection of taiz and soiz is identical with that of nouns,
and when used as attributes these words agreewith their head, e.g. ta--iz
knjiga 'this very book', taiz-tekknjiga-tek 'without this very book', ta-os-iz
knjiga-os 'thesevery books'.
When used independently,all the demonstrativepronounsare declined.
Their paradigmfollows that of the nominal closely, but there are deviations,
e.g. ta-je 'this (acc)' (cf. mon-e 'me'). Otherdemonstrativepronounsare tace
'like this', sice 'like that', tami"nda 'this much',somi"nda'that much'.
The interrogative pronounsare also usedas relative pronouns.The most
important are kin 'who', ma(r) 'what, which', kud(iz) 'which', kice 'what
kind', ki!nia 'how much'. Apart from a few deviations (e.g. kin-e 'whom')
they decline like nouns, e.g. kin-If 'to whom', kin-en 'with whom'. Plural
personalsuffixes may be usedwith kud, e.g. kud-mf'which of us'.
Indefinite, negative, and general (universal, indifferent) pronouns are
formed, in the main, from the interrogative pronouns. Both absolute and
personal declension are frequent. Examples: kin+ke, olo+kin 'someone',
no+kin, nie+kin 'no one', kotj+kin 'anyone, no matter who', kotj+ma(r)
'anything, no matter what', no+kin-mf 'none of us', no+mfr-df 'nothing of
yours (plur),. Others include muket 'other', and vanj(miz), Mies 'all',
inflected: vanj-df 'all of you (plur),.
The addition of the suffix -na forms numeric pronouns;they are always
 UDMURT 289

inflected for person. Examples: og-na-m 'I alone', kik-na-di~lf

og-na-m 'to you
two'.

Verb

Conjugation:Finite Forms
The finite verb paradigmdistinguishesthe categoriesof mood,tense,number,
andperson.
Stem-typesand conjugations.The citation form in Udmurt dictionariesis
the infinitive, whose suffix is -nr. Lopping off this suffix gives us the verb
stem, which always ends in either r or a. Two conjugations may be
distinguished,the first conjugation (f-stems) and the secondconjugation
(a-stems).Suffix allomorphy presentedbelow will be labelled with 1 and II
accordingly.The stem-final f of first-conjugationverbs is deletedwhenever
a vowel-initial suffix is attached.
Mood and tense.Tenseandmoodarein complementarydistribution. There
are threemoods:indicative,imperative,andconditional.The indicativemood
distinguishesfour tenses:present(vetliSiko 'I go', turnaSiko 'I mow'), future
(vetlo 'I shall go', turnalo 'I shall mow'), primary past (or preterite: vetli 'I
went', turnaj 'I mowed'), and secondarypast (or perfect: vetlem '(s)he has
gone',turnam '(s)hehasmowed').The primary pastrefersto activities in the
pastwithout referenceto their completionor to any result.The secondarypast
refersto the result of activities in the past,eventswhich are often not directly
witnessed.The suffix of the primary pastis -i- in the first conjugation,-a- in
the secondconjugation.The suffix of the secondarypast is -(e)m, which is
precededin the first person,and in the secondand third personsplural, by
other morphemesof derivational origin. The -(e)m suffix is historically
identicalwith that of the perfectparticiple,seebelow.
The segmentationof the presentand future morphemesis not synchron-
ically transparent.
All four simple tense-formsmay combinewith val 'was/been'and vilem
'has been,was (evidently), to form compoundtenses.Thesetensesexpress
durativity or habitual activity, e.g. so gfr-e val PRO.s3 PLOUGHS-s3pres
BEEN '(s)hewas ploughing',so gfr-oz (s3fut) val '(s)heusedto plough(long
ago)', so grr-oz vilem '(they say that) (s)he ploughed;(s)he ploughed(as 1
remember)" min-em val GOES-past2BEEN '(s)he had (already) gone
(evidently),. Dialectal use of suchcompoundtensesvaries considerablyand
awaitsfurther investigation.
The suffix ofthe imperativeis zero in the singular.First-conjugationverbs
optionally lose their final f if this is precededby a single consonant.
Examples:vf.tlf 'go!', gfr(ij 'plough!',Vf.ra 'speak!'The second-person plural
imperativeis formed with -e(le) (I)/-le (II), e.g. min-e 'go! (plur)', persons
kfr~a-le
'sing! (plur),. Commandsand suggestionsdirected to other persons are
expressedwith the future tense;outsidethe axis of discourse,the particlemed
290 UDMURT

Table 9.9 Finite affirmative forms of the verb

First conjugation Secondconjugation

mi"ni"- 'goes' daSia- 'prepares'

Indicative mood
Present
sl mYnisjko dasjasjko
s2 mYnisjkod dasjasjkod
s3 mYne dasja
pI mYnisjkom(I) dasjasjkom(I)
p2 mYnisjkodY dasjasjkodi'
p3 mYno dasjalo
Futuretense
sl mYno dasjalo
s2 mYnod dasjalod
s3 niinoz dasjaloz
pi mYnom(Y) dasjalom(y)
p2 mYnodY dasjalodY
p3 mYnozY dasjalozY
Primarypast
sl mYni dasjaj
s2 mYnid dasjad
s3 mYniz dasjaz
pi niinimY dasjamY
p2 niinidY dasjadY
p3 mYnizY dasjazY
Secondarypast
sl mYnisjkem dasjasjkem
s2 mYnem(ed) dasjam(ed)
s3 mYnem(ez) dasjam(ez)
pI mYnisjkemmY dasjaSikemmi'
p2 mYniljl-iam(dY) dasjaljljam(dI)
p3 mYniljljam(zY) dasjaljljam(zi')
Imperativemood
s2 m:!n(Y) df!sja
p2 m:!ne(le) df!sjale
s3 m5<.d mi'noz m5<.d dasal-ioz
p3 m5<.d niinozY m5<.d dasaljozi'
Conditionalmood
sl mi'nYsal dasjasal
s2 mYnYsal(yd) dasjasal(i'd)
s3 mYnYsal(Yz) dasjasal(Yz)
pI mYnYsalmY dasjasalmY
p2 mYni'saldY dasjasaldY
p3 mYni'salzY dasjasalzY
 UDMURT 291

mustalsoprecedethe verb, e.g. m~dm~duzal-oz'let himlher work'.

The conditional suffix is -sal-. The fIrst personsingular is zero, and the
second-and third-personsuffixes may be omitted; syncretismof singular
forms is thus possible.Example:disi-sal(-iz) '(s)hewould study, (s)hewould
havestudied'.
Somenegativeverb forms arebuilt with the negativeverb (u- in the present
in future, e- in the primary past, e- in the imperative), others with negative
particles (evel in the secondarypast, ej in the conditional) or the negative
suffIx =mte (in the secondarypast,in free variation withthe useof the particle
evel).
Tables9.9 and 9.10 set out the completefInite paradigmsof mi"ni"- 'goes'
anddaFa- 'prepares'.Bracketsareusedto indicateparallelforms which seem
to differ at most stylistically in the standardlanguage.
Alongsidethe completelyregularfactotumverbverb lui~ 'is, becomes,happens;
is able, is possible;exists' there is also a constellationof suppletiveforms
which express pure existence: present vani 'exists', primary past val,
secondarypastvilem, negativepresentevel 'thereisn't'. Theseforms inflect
for neitherpersonnor number.

Verbal Nominals
Infinitive (-ni) is usedwith auxiliary verbsand as a substitutefor fInal clauses
with verbsof motion, e.g.potizi" ara-ni" 'they went out to harvest'.
Participlesfunction as both attributesandpredicates(seeSyntax,p. 299).
The imperfective active participle (=iF U=F II), e.g. vetl=iSi 'going',
vera=F 'speaking'forms manynouns,e.g. diset=iF 'teaching,teacher'. It has
its own negative form, built with =tem, e.g. (from diFt- 'is brave')
diFt=iF=tem 'timid'.
The perfect participle (=em U=m II) renders passive meanings with
transitive verbs and resultativemeaningswith intransitive verbs. Examples:
gi"r=em busi" 'ploughedfIeld', tone vaj=em murt gurt-az kosk-iz PRO.s2acc
BRINGS=perf.partHUMAN HOUSE-s3ill DEPARTS-s3pastl'the person
who brought you went home', li"kt=em kisnomurt ARRIVES=perf.part
WOMAN 'the womanwho has arrived'. This suffix also forms many nouns,
e.g. kul=em 'death'dies'),
(kuli~ 'dies'), uza=m 'work' (uza- 'works'). It hasits own
negativeform, =mte, e.g. kuti"=mte kion 'a wolf which hasnot beencaught'.
The necessitiveparticiple (=ono U=no II) expressesthe necessity of
carrying out an action, e.g. esker=ono uz 'a matter which must be
investigated',kora=no pu 'a tree which must be felled'. Its negativeform is
=on=tem U=n=tem II and expressesimpossibility, e.g. biaest=on=temuz
'work that is impossibleto carry out', Bida=n=tempeF nunal 'an unbearably
hot day'.
The potential participle (=mon VII) is the affirmative analogueto the
preceding (=(0 )n=tem), e.g. giif=mon pijas 'a young man capable of
ploughing', gi;f=mon inti" 'a placewhich canbe ploughed'.
292 UDMURT

Table 9.10 Finite negativeforms ofthe verb

First conjugation Secondconjugation

'goes'
mini~ 'goes' daSia- 'prepares'

Indicative mood
Present
sl u-g minisjkI u-g dasjasjkl
s2 u-d minisjkI u-d dasjasjki
s3 u-g mInI u-g dasja
pI u-m minisjke u-m daSiasjke
p2 u-d minisjke u-d das.iasjki:e
p3 u-gmino u-g dasjalo
Futuretense
sl u-gmini u-g dasja
s2 u-dmini u-d dasja
s3 u-gmini u-g das.ia
pI u-m mIne(le) u-m dasjale
p2 u-d mIne(le) u-d dasjale
p3 u-g mIne(le) u-g dasjale
Primary past
sl e-j mInI e-j dasja
s2 e-d mini e-d dasja
s3 e-z minI e-z dasja
pI e-m mine(Ie) e-m dasjale
p2 e-d mine(le) e-d dasjale
p3 e-z mine(le) e-z dasjale
SecondaryPast
sl minisjklmteUe) dasjasjklmteUe)
evel minisjkem evel dasjaSikem
s2 minimtejed dasjamtejed
eve! minem(ed) eve! das.iam(ed)
s3 minimteUez) dasjamteUez)
evel minem(ez) evel dasjam(ez)
pi mi:nisjkimtemi dasjasjklmtemi
eve! miniSikemmi eve! dasjaSikemmi
p2 miniljljamte(dl) dasjaljljamte(dl)
evel miniljljam(dl) eve! dasjaljljam(dl)
p3 miniljljamte(zl) dasjaljljamte(z])
evel miniJiljam(zl) eve! dasjaPljam(zl)
Imperative(prohibitive)
s2 !<n mIn(]) !<nmed~d~sja
p2 !<n mIne(le) !<n med~d~sja
s3 med~zmed~zmIn(i:) med~d~sjamed~d~sja
p3 med~zmed~zmIne(le) med~zmed~z med~d~sja
 UDMURT 293

Conditionalmood
sl ej mi"ni"sal ej dasiasal
s2 ej mi"ni"sal(i"d) ej dasiasal(i"d)
s3 ej mi"ni"sal(i"z) ej dasiasal(i"z)
pI ej mi"ni"salmi" ej dasiasalmi"
p2 ej mi"nYsaldY ej dasiasaldi"
p3 ej mi"ni"salzi" ej dasiasalzi"

The inessiveof the perfectparticiple expressesresultativepredicates,e.g.

busfgfr=em-i"n 'the field is (in a) ploughed(state),, turin tuma=m-in 'the hay
is (in a) cut (state),.
Gerunds expresscauseand result as well as various temporalrelations
(simultaneity,anteriority,duration).
The =sa gerund expressessimultaneity or anteriority, e.g. Sierekja=sa
vetl-e LAUGHS=ger GOES-s3pres'(s)he walks laughing'. This gerund
occursfrequentlyin compoundverbs,seebelow.
The =tek gerundis the negativeof the =sa gerund,e.g. veraSikf=tekpuk-e
'(s)hesits without speaking'.
Generaltemporalrelationsareexpressedwith the =ku- gerund,which may
also take personal suffixes. Examples:ara=ku ~eg ara=ku tui peSi val RYE
HARVESTS=temp.gerVERY HOT EXISTED 'at rye-harvesttime it was
very hot', bertf=ku-m 'whenI returned'.
The =to~ =to~gerundrefersto the terminuspostquemnon, e.g. iadh=to~ara=kubiz/-
iz TIRES-term.gerRUNS-s3pastl'(s)heran until (s)hewas exhausted'.
The =mon gerund, homophonouswith the potential participle, means
somethinglike 'in a mannersuch that X can be Y'd, -ably', e.g. vala=mon
ver-az'he spokeintelligibly' (vala- 'understands'), lia~f=mon lia~f=mon goit=em-in 'it
is (in a) legibly written (state),.
Furthergerundiveforms aremadefrom the perfectparticipleby addingthe
instrumentaland elative casesuffixes, e.g. berd=em-en-iznomfr vera-nfu-g
bigatfWEEPS=perf.part-ins-s3 NOTHING SAYS-infNEG.pres-s3IS.ABLE
'what with hislher weeping it is impossibleto say anything', OPos kopa=
m-i"F dugd-iz ALYOSHA DIGS=perf.part-elaCEASES-s3pastl'Alyosha
ceaseddigging'.

Postpositions
These may be classedmorphologically as either segmentable,e.g. ul-in
'under', ul-i"F 'from under', ul-e 'to under' (cf. ul 'lower part') or
unsegmentable,e.g. ponna 'for'. Personal suffixes are attached to the
postposition, not to its head, e.g. ukno ul-am WINDOW UNDER-sline
'undermy window'.
The headis usually in the nominative,but there are exceptions,e.g. tonen
serenPRO.s2insBECAUSE 'on accountof you'.
294 UDMURT

Derivational Morphology

Formation of Nouns
Someof the moreimportantnoun-formingsuffixesin the literary standardare
=li"k, whichbuild abstracts(zem'truth', zem=lik'justice', muskit'moist', musk=
i"t=lik 'moisture'),=Cii, which builds namesof occupations(vuz 'itemfor sale',
vuz=di 'merchant',argan 'accordion',argan=Cii 'accordionist'),=es (imple-
mentsand body parts: kut- 'grasps',kut=es 'threshing-rod',urd 'rib', urd=es
'side'), =on U=n II (abstracts: vera~kr- 'converses', vera~k=on 'conversa-
tion'), =et (objects,results:[iuki"- 'divides',IJuk=et 'part,piece'),=em U=m II
(abstracts;=perfectparticiple: pizi~'roasts',piz=em 'roast,joint').

Formation of Adjectives
These include =0 (roughly 'having X': jel 'milk', jel=o skal 'milk cow',
kuzJm 'strength',kuzm=o 'strong'; this suffix has a mitigating effect when
addedto intensivereduplicatedadjectives,e.g. gord 'red', gQrd&gord 'very
red', gord=o&gord=o 'reddish'), =tem (privatives: vu 'water', vu=tem
'waterless',sud 'happiness',sud=tem'unhappy'),=es (makesadjectivesand
nounsfrom nouns,adverbs,andonomatopoeics:~erem ~erem=es 'laughter',~erem=es~erem=es
'funny', ul-Iani 'downward', ul-IanJ=es 'declivity', zingir 'ting-a-ling',
zingr=es 'small bell').

Formation of Verbs
Verbal derivation is extensive, and permits nuanceddifferentiation in the
expressionof the type andmannerof the action as well as the relationof actor
to action.
The following suffixes make verbs from verbs: =li~(goz(regular, repetitive
action: giri"=li~ 'ploughs regularly, is always ploughing, ploughs and
ploughs', kusti~ 'throws out/away', kusti"=li~ 'throws around a lot, keeps
flinging'), =[1IJ a- (with similar function: loba- 'flies', 10ba=[1[1a- 'flits about',
vera- 'says',vera=[1[1a- 'keepssaying, regularly/alwayssays'), =ti~ (causa-
tives and factitives: uza- 'works', uza=ti~ 'makes[someone]work'), =i~ki~ =i~ki~
U=~ki~
(- =skr) II (reflexives,mediopassives, andreciprocals: di~a- 'dons',
di~a=~ki~ 'gets dressed',Ciupa- 'kisses', Ciupa=~ki~ 'kiss one another',
diSeti~ 'teaches', diset=ski~ 'learns'), =emja~ki~U=mja~ki~ (devalues or
questionsthe quality or validity of the action: uza=mja~ki~ 'pretendsto work;
works away (ineffectually) at').
Other verb-forming suffixes include =i~(zor(zor 'rain', zor=i~'rains','rains', viz
'bridge', viz=i~ 'crosses'),=a- (nJim 'name', nJim=a- 'names',jig&jig
'knock-knock',jig=a- 'knocks,hits'), =ma- (gord 'red', gord=ma- 'makesX
red', vozJ- 'holds, contains', vozJ=ma- 'waits'), =mi~ (iz 'stone', iz=mi~(goz
'petrifies', viti 'new', vilJ=mi~ 'is renewed',vizJ=tem 'stupid', vizJ=tem=mi~
(goz(goz
'becomesstupid/stultified'), =di~ (vij=di~ 'renews,reconstructs'),=ti(goz
~ (goz
'line, mark', goz=ti~ 'writes').
 UDMURT 295

The Morphology of Compounding

Nominal compoundsmay be co-ordinative, e.g. §iedales+liZ 'dark blue
(DARK+BLUE)" ulirlO+vilirlO 'one-storeyhouse(roughly: DOWNSTAIRS
+UPSTAIRS), or subordinative,e.g. peSi+kil 'typhus (HOT+ILLNESS)',
vajo+biz 'swallow (DOUBLE+TAIL)'. Dvandva compounds also occur,
e.g. iin+nrr 'face (MOUTH+NOSE),, anaj&ataj 'parents (MOTHER
&FATHER)'.
Co-ordinative verb compoundsmay consist of paired verbs, e.g. sudi~ kur+a~~~
nr&Sierekja- 'hasfun (PLA YS&LAUGHS)', or of NOUN+VERB syntagms,
e.g. kur+a~~~kur+a~~~ 'suffers(BADIEVIL+SEES),.
Subordinativeverb compounds,so-called'doubleverbs' (they are in fact
verb complexesbuilt with the =sa gerund), are extremely common. The
semanticsof the two verbs meld to create complex aspectual/semantic
shadings.Examples: kora=sa bidti~ CUTS=ger FINISHES 'fells (perfec-
tive)', ~a=sabrri~ BURNS=ger PERISHES 'bums down completely',
peg~r=sakoski~ ESCAPES=gerDEPARTS 'getsaway safely'.
Another type of compoundverb is built from recentRussianloans.In this
construction, the Russian verb is put in the (Russian) infinitive and the
Udmurt factotumverb kari~ kur+a~~~ 'does'is added,e.g. diktovati kari~
kur+a~~~ 'dictates'.

Syntax

Nominal Constructions
In attributive constructionsmodifier precedesmodified. Besidesadjectives,
modifiers may be deictics, numerals,nouns,and participles;participlesmay
have, in tum, their own complements.Examples:te{h derem 'white shirt',
sice Borrg 'such a fish', ukmiseti klass 'ninth class', pu korka 'wooden
house',lulo gondirez kutiinon pejSuraSi 'hunter capableof catching a live
bear'.
Attributive adjectives stand before the noun they modify and do not
generally agree with it, either in number or in case, e.g. liz kiset-jos-tek
'without blue cloths'. Exceptionallythere is plural agreement;the adjectival
plural suffix is again-eSi, e.g. kuf/-eSino zek-eSikor-jos uram-rn kitizio 'long
and thick beamslie in the street',veraSikiSitem-eSi vorgoron-jos 'silent (non-
speaking)men'. Adjectives with the s3 definiteness-markerare declined as
nouns,whetheraloneor in noun phraseapposition,e.g. viP-ez-lrNEW-s3-dat
'to the new (one)',jegit-jos-rn-iz nil-jos-rn YOUNG-plur-ins-s3GIRL-plur-
ins 'with the young ones, with girls', bur-in-iz §iinm-in-iz RIGHT-ins-s3
EYE-ins-s3'with his right eye'.
In the possessiveconstructionthe possessorstandseitherin the genitive or
(if the possessionis the direct object) in the ablative. The possessiontakes
personsuffixes. ExamplesdisetiSi-len kniiga-os-iz TEACHER-genBOOK-
plur-s3 'the teacher'sbooks', gurt-len pum-iz 'the end of the village',
296 UDMURT

muz/ej-leSi jurt-se tupat=jano MUSEUM-abl BUILDING-s3acc RENO-

VATES=nec.part'the museumbuilding hasto be renovated'.If the possessor
is encodedby a personalpronoun,this standsin the genitive: so-len vin-iz
'hislher younger brother'; the pronoun may also be omitted (vin-iz).
Furthermore,invariable as or the inflected reflexive pronoun may be used
instead:as pi-jez-li" 'to hislherown son',asladkorka-jed 'your own house'.
Two other constructions which border on the possessiveare simple
juxtaposition(~k kuk 'the leg of the table(TABLE LEG)" korka iis 'the door
of the house (HOUSE DOOR)" used with inanimates,and the use of the
elative to mark the place where somethingbelongs or comesfrom (jag-iSi
puzJm-jos 'pines from/of the forest', magaz/in-iSi vitrina-in 'in the shop
window').

Subjectsand Predicative Constructions

Any nominalcanfunction as subject.The subjectis usuallyin the nominative,
with or without personal/determinative suffixes:pi turna 'theboy is mowing',
soosturnalo 'they are mowing', pimi" disetske'our son is studying',bad{ziin-
ez pi-je zavod-in uza, piBi-jez kolxoz-in ELDER-s3 SON-sl FACTORY-ine
WORKS.-s3pres,SMALL-s3 KOLKHOZ-ine 'my older son works in a
factory, my younger son in a kolkhoz'. Paired subjects go into the
instrumental,e.g. nil-in pi-jen gorod-e kosk-i-zi"GIRL-ins BOY-ins CITY-ill
DEPARTS-pastl-p3'the boy andgirl went to the city'.
A specialform of subjectis markedwith a casesuffix (most commonly
elative orgenitive)followedby the plural suffix -jos; this constructionis used
to expressa collectivity of peoplefrom a certain place or group: otin kutsa-
Sik-o kolxoz-iSi-josTHERETHRESHES-p3pres KOLKHOZ-ela-plur 'people
from the kolkhoz are threshing there', aSime-len-josgi"ron-zes bidest-i-Zi"
PRO.REFLEX.pl-gen-plurPLOUGHING-s3accFINISHES-pastl-p3 'our
lot have finished the ploughing'.The infinitive can also function as subject:
tilled-in uza-ni" tuz kapBi PRO.p2-insWORKS-inf VERY EASY 'working
with you (plur) is very easy'.
In impersonalconstructionsthe logical subjectstandsin the dative: Kotia-/r
disetiSi dor-i" min=ono lu-i-z KOLYA-dat TEACHER NEAR-ill GOES=
nec.partIS-pastl-s3'Kolya hadto go to the teacher'.
Predicatesare of two basic types: verbal and nominal. Verbal predicates
agreewith their subjectin numberand person,and expressthe modality and
relativetime of the action.Theconditionalmoodis usedto conveydesirability
or contingency,e.g. ti dor-i"vetli~sal-mi"
'we would (like to) go to you(r place)',
so-leSiki1ziSiki~sal-zi~-ke, iijlui~sal
nomi"r PRO.s3-ablLISTENS-cond-p3--IF
NOTHING NEG HAPPENS-cond'if they had listenedto him/her, nothing
would havehappened'. Commandsandprohibitionsareexpresseddirectly by
the imperative:kosk-eletatiSi '(you plur) go away from here',en min 'don't
go!', nil-/r vera, med pi"r-oz korka-f} GIRL-dat SAYS-s2imp PARTICLE
ENTERS-s3futHOUSE-ill'tell thegirl to go into thehouse'.
 UDMURT 297

We may classify not only nouns,adjectives,numerals,and adjectives,but

also adverbial complementsas nominal predicates.Examples: viznan nier
kuf! 'the fishing rod is long', so disetiSi '(s)heis a teacher',mi doriSienBupBi
sur kitPok-i"n PRO.pl FROM CHEPCHARIVER DISTANT-ine 'the Chepcha
River is far from us', so kniiga-jez li"d~=ono 'this book mustbe read'.
Present-tenseforms of the nominal predicatetake zero copula, e.g. ton
udmurt--a 'are you Udmurt?', mon tati"n Sier murt PRO.sI HERE STRAN-
GER HUMAN 'I am a strangerhere'. In the past tense,indeclinableval is
usedfor all personsandnumbers:ti"nad brat-edsokustudientval'your (older)
brother was a studentthen', tolon bert=ono val 'yesterdayit was necessary
to return (home)'. In othertensesor to expresschangeof state,the copulaand
factotum verb lui~ is used: minam brat-e agronom lu-oz 'my (older) brother
is going to be an agronomist',zor ber-e omi"r Bilkn lu-i-z RAIN AFfER-ill
AIR CLEAN BECOMES-pastl-s3'afterthe rain the air becameclean'.
Pure existenceis expressedby the indeclinableparticle vani : ta gurt-in
odig motor nil vani 'thereis a beautiful girl in this village'; non-existenceis
expressedby indeclinableevel: kanfetevel, sakarevel, rna-in ben Baj ju-omi"
SWEETS NOT.EXISTS SUGAR NOT.EXISTS WHAT-ins INDEED TEA
DRINKS-pipres/fut 'there isn't any sweet,there isn't any sugar,with what
shall we drink tea?'.
Udmurt lacks a possessiveverb correspondingto English 'has', and vani
(negative: evel) is used instead, e.g. aSime-len tros uzaSi-jos-mi" vani
PRO.REFLEX.pl-genMANY WORKER-plur-pl EXISTS 'we have many
workers',pi-je-Ien kostium-ezevel SON-sl-genSUIT-s3 NOT.EXISTS 'my
sonhasno suit'.

Direct Objects
The direct objectcan be in the nominativeor the accusative.The generalrule
is that indefinite objectsstandin the nominative(-0) , e.g. goitet-@ gozto 'I'll
write a letter', vumurt soabi~li" Siotemajtal-@ 'the water-spiritgives that old
woman (some) soap', niulsekaSijosgondi"r-@ kutiziziam 'the hunters have
caughta bear',while definite direct objectsstandin the accusative:'{!eg-ez
~ek
oktisa bi"dtimi" inii 'we've alreadygatheredin the rye', ta gazet-jos-ft" vile
pon 'put thesenewspaperson the table'. Direct objectsmarkedwith person!
defmite suffixes always count as definite; the nominativepersonalforms are
thereforeneverusedas direct objects.Examples:baSitemajtal-ze '(s)hepicks
up his/her soap', kin so ti"pi~jez pogirtoz, so-Ii" nil-me Siot-o, su-em eksej
WHO THAT OAK-acc FELLS-s3fut PRO.s3-dat DAUGHTER-slacc
GIVES-sipres/futSAYS-past2TSAR 'whoeverfells that oak, to that one I
shall give my daughter,said the Tsar', azitem-IeSiumoj uiam-ze en vitii"
LAZY-abl GOOD WORK-s3accPROHIB.VERB-s2AWAITS'don'twait for
the good work of a lazy person,i.e. don't expect good work from a lazy
(person),.The infinitive, too, can serveas direct object: tros mediSikodlePti:.
ni" 'you want to accomplisha lot' .
298 UDMURT

Adverbial Complementsandthe Useof the CaseSuffixes

Adverbialsareeitherindependent, like ~eCi 'well': soleSit-e
leSit-e ~eCi leSit-e '(s)hedoes
it well', gerunds,or formedfrom nominalswith casesuffixesor postpositions.
Here there is spacefor only a brief survey of some of the more important
adverbialusesof the cases.
Ablativeexpressesmotion awayfrom X, origin, cause:Petrovmasina-leP
palensk-i-z'PterovJumpedaway from the machine',stal-leP kuso leSit-o
'they makescythe[s]of steel',suniY-lePlimf suna 'the snow melts from the
heat'. It is alsousedto mark the superlativememberof a comparison(sundr-
leP jugU 'brighterthan the sun') and with certainverbs (kaban-lePval u-g
kiskaSikf'a horseisn't afraid of a hayrick').
Dative is most frequently usedto indicatethe indirect object, as in es-e-li'
Siot-iSiot-i 'I gavea book to my friend', nene-li'jurtti~nf indirect 'to help mother',
k~iga
but it can also expressintendedtime-span(nunal-lf 'for a day'), objective
(Olja vu-li"mfn-e 'Olya goesfor water'), or result(CiaSi-i-z, kii/pu-ezselep-li"
permit-i-z 'lightning struck(it), the birch turnedinto (a pieceof) kindling'.
Caritivelabessiveexpressesthe lack of something: mumi"-tek&bubi"-tek
kili-i 'I was left without motheror father'.
The adverbial caseindicates accordancewith something,whether tem-
poral or qualitative: plan-ja 'accordingto plan', uzam-ja milki"d cheerful' ~tskil-i-z
'during/in the courseof the work the moodbecamemorecheerful', tus-si"-ja
no soosog&og-zi~lf
resemble kePs-o'they resembleone anotherin their appearance,
aswell'.
Instrumentalis usedboth for instrumentalproper (kaCii"-jen drawing vandi~nf 'to
cut with scissors',Pulm-in-iin sediSiko'I feel (it) with my heart') and for
accompaniment (Pedor-enuza-mf'we workedwith Pedor').
Approximativeindicatesdrawing nearto a placeor approximationin time:
nil sur-la~ bii/ bii/ -i-z 'the girl ran towards the river', guzem-la~ drawing ozjf lu-i-z
'this happenedtowardssummertime'.
Inessiverefersto location within or on a place(gurHn 'in a village', busi~busi~
os-in 'in the fields'), to point in time (vitj=eti Cias-in FIVE=ord.numHOUR-
ine 'at five o'clock'), or to state, as in disetiP-in uza '(s)he works as a
teacher'.
Illative refers primarily to movementtowards a place (Petrov kanava-je
uSi-i-z 'Petrovfell into the canal'),but also to points in time (a~a (a~a nunal-e
'on Sunday'), ~t-jos-f'evenings, 'evenings,every evening', sundijo nunal-jos-f 'on
sunnydays'.
Elative refers primarily to movementfrom a place (busi"-iSi bert-e '(s)he
comesbackfrom the field'), but is also selectedby certainverbs,e.g. va~-zeva~-ze
so-je ti sedit-odi"tatrajectory,
k~iga-iSi 'you (plur) will find all this in this book'.
Transitive indicatestrajectory, i.e. route traversedor aperturefranchised,
e.g. uWa-jeti avtobuskosk-i-z 'the bus drove along the street',so ukno-jeti
uCikeval '(s)hewas looking (out) of the window'.
Terminativeindicatesthe point in spaceor time beyondwhich something
 UDMURT 299

doesnot proceed,e.g.arama-of!bi"f/-i-mi" 'we ran asfar asthe grove', '{Itt-of!

uza-zi"'they worked untilevening'.
Egressive indicates the reverse of the tenninative, i.e. the spatial or
temporalstarting-point:Iievsk-iSienpojezd-enmin-i-mi" 'From Izhevsk,we
went by train', sif/i1-iSien '(starting)from autumn'.

Major ConstituentOrder
In neutralsentencesthe subjectis in the first part of the sentence,the predicate
is at the end, and any direct object and adverbial complementsstand
somewherein between, e.g. Mixajlov vaf! Ciukna ik uzaFjos dori" li"ktiz
MIKHAILOV EARLY MORNING DULY WORKERSTO WENT 'so, early
in the morning M. went to the workers'.Deviation from this neutralorder is
the rule, however, particularly (but not exclusively) in interrogative and
exclamatorysentences.Factorsinclude position in the discourse,communic-
ationalroles, sentence-types,
andlogical stress(emphasis).

Constructionswith Verbal Nominals

Verbal nominals (infinitive, participles,gerunds)which function as comple-
ments may take complementsof their own; they then are functionally
equivalentto subordinateclauses.The nature of the subordinationmay be
temporal, causal, or other. A sentencecan contain more than one such
construction,e.g. gorod-iSi tolon li"kt=em inzenier, uz-ze li"msor-ozi bi"desti"
=sa, gurt-e benziin-li" kosk-i-z CITY-ela YESTERDAY ARRlVES=perf.
part ENGINEER WORK-s3accNOON-term COMPLETES=gerTOWN-ill
PETROL-dat DEPARTS-pastl-s3'the engineer who came from the city
yesterday,havingfinished his work by midday,left for town to get petrol'.
Here are some further examplesof verbal nominal constructions.With
infinitive: ta uz-ez~og bi"desti"-ni"ponna, t70s kuzi"m kule luoz THIS WORK-
acc QUICK FINISHES-inf FOR, MUCH ENERGY NECESSARYIS.s3fut
'muchenergywill be necessaryin orderto finish this work'. With participles:
Ferpala !1Cikiz no urobo dorin siliP atPamijez ad~iz Oleksan 'Oleksan
looked to his side and saw the man who was standingnext to the waggon
(WAGGON NEXT.TO STANDING MAN)" vuko-iSiberl=em Oddo val-ze
jusk-e 'Oddo, having returned from the mill (MILL-ela RETURNS-
=past.partODDO) unharnesseshis horse',pinal-jos kosk=em ber-e, tatin
mezmnlu-i-z 'after the children left (CHILD-plur DEPARTS=past.part
AFTER-ill) it becameboring here'.With gerunds:nunalbitPekit mesok-jos-
ti· '{tutja=sa zatJi-i-z Mikvor Pedor 'lifting heavy sacksall day (ALL.DAY
HEAVY SACK-plur-acc LIFTS=ger) P. M. became tired', ma sui"-ni"
todi"=tek, so palensk-i-z 'not knowing what to say (WHAT SAYS-inf
KNOWS=neg.ger), (s)he left', pi-me disetski"-ni" nui"-ku-m vedra duren
kiFt=em kadi zor-i-z 'whenI took my sonto be taught(to his lesson;SON-
slacc TEACHES-inf TAKES=temp.ger-sl)it was raining as if pouredfrom
a bucket(BUCKET FROM POUREDAS)" ton niulesk-i"vui"=tof! so gurt-az
300 UDMURT

lu-oz inii 'by the time you get into the forest (PRO.s2 FOREST-ill
ARRIVES-term.ger)(s)hewill be homealready'.
CompoundSentences
Compound sentencesmay be further subdivided into co-ordinate and
subordinatetypes. In the following examplesthe conjunctionsand relative
pronouns are highlighted. Co-ordinate examples: kwaf! zoriz no, turin
vozektiz'the rain fell, and the grassgrew green',guzemnunaljoskuz/eSiluo,
a ujjos vakCiijeSi 'summerdays are long, whilelbut the nights are short',
sutetskonoval no, di"r eve! 'one ought to rest, but there isn't time', umme
uSiini" ug lui:' to puniosuto, to jegitjos ki"r~alopeople'it is impossibleto fall asleep:
either the dogs are barking or the young people are singing'. Subordinate
examples:ton vaz/ valad, kin luono tini"d 'you learnedearly who you should
become',kinn tolon giridi; olin ik voz/manoluodi" 'where you ploughed
yesterday,there you must wait', kuke sunn luiz, soku Ciori"gani" minomi"
'whenit hasbecomewarm, [then] we shall go fishing' ,pinal-joskorka-nevel
dir·ja, mon kniiga li"d~-iSik-o CHILD-plur HOME-ine NOT.EXIST TIME-
adv PRO.sl BOOK READS-pres-sl'when the children aren'thomeI reada
book', kinlen praktikajez bad~iin, solen uzezno vozin mine '(s)he who has
(had) a lot of practice[that person]hashislherwork go smoothly',adjamilen
kiee Siulmiz, siee ik uzezno 'as a person'sheart is, so too is hislher work',
uz bordi" kutskid ke, soje odno ik biaestini" tirsi" 'if you have startedto do
something,try to finish it as well'.
Thereare also asyndeticcompoundsentencesof both types,e.g. brigadaos
Sief.jiZesgine oktisa&kaltisa vuizi; vilhSi zorini" kutskiz 'the brigadeshadjust
finished bringing in the oats,it beganto rain' ,juzvilti valen vetlizi; sokemso
jun val 'they went by horseover the snowcrust,so firm wasit' .

Indirect Discourse
Declarative,interrogative,and exclamatorysentencesare reportedby means
of the conjunction suisa (i.e., the sa-gerundof sui:' 'says'): vara veraz, so
vanimiz SiariSi todiz sUi"sa VARA SAYS-s3pastl PRO.s3 EVERYTHING
ABOUT KNOWS-s3pastlSAYS=ger'Vara said that sheknows aboutit all',
kuzito vorttomi" sat sui"sa, purga kenez puzej-leSi juaz SHEAF DELIV-
ERS.plfut PERHAPSSAYS=ger, PURGA MRS PUZEY-ablASKS-s3pastl
'Mrs PurgaaskedPuzheywhetherwe might perhapssendsheaves'.As these
examplesshow, the subordinateclausemay standbefore or after the main
clause.
Imperativesare renderedindirectly by meansof the future forms of the
verb, precededby the particle med. As in other forms of indirect discourse,
the subordinateclauseends with sui"sa, e.g. Ciebersur-iSikalik-ez viCiak-se
tatCii" med uljzialoZi" suisa, pop kos-e CEBERSUR-elaPEOPLE-aceALL-
s3accHITHER PARTIC DRIVES.p3futSAYS=gerPRIESTCOMMANDS-
s3pres'the priestordersthat they drive all the peoplefrom Cebersurhere'.
 UDMURT 301

Lexicon
Architectonics

Loanwords
Native roots have a relatively simple structure.Two vowels occur adjacent
normally only at morphemeboundaries.Consonantclustersof two normally
occuronly root-internalandroot-final; in the latter caseoneof the consonants
is a resonant.Examples:iz 'stone',gu 'pit', uno 'many', bur 'good', bord
'wall', tili" 'feather', gondi,. 'bear'. Onomatopoieicwords typically show
(partial) reduplication,e.g. iingi,.&iangi"r 'ting-a-ling'.

Older Loans
Older loanwordsdo not deviateformally from native canonic shapes.There
is a Chuvashlayer (the Udmurt paid tribute to the Chuvashfrom the tenth to
the thirteenthcentury),e.g. ara- 'harvests',amia 'week',buskeP'neighbour',
'money'.
uk~o 'money'.Thereis also a later layer of Tatar loanwordsdating from the
thirteenthcentury.Four to five hundredTatarloansenteredcommonUdmurt,
and even more into many dialects. Examples: bati,. 'fearless; hero', kijar
'cucumber', uram 'street', jemisJ 'fruit', adiami 'person', alda- 'cheats,
deceives',bujol 'paint'. Tatar loans continueto flow into peripheraldialects
of Udmurt to the presentday.

RussianLoans
The consonantsf, x, and c occur only in recentloans, mostly from Russian
(cement,fabrika 'factory', xirurg 'surgeon').In earlier Russianborrowings,
thesesoundswerereplacedby p, k, andc, e.g.porma 'form', kalera 'cholera',
ulca 'street'.
More recent loans also retain initial consonantclusters, e.g. dvorec
'palace'; in earlier borrowings such clusters were either syncopatedor
renderedmedial by prothesis,e.g. durug 'suddenly' < B,upyr, iiiver 'wild
animal' < 3Bepb.
Northern groupsof Udmurt speakershad alreadycome into contactwith
Russianduring the Tatar reign, but the great upsurgein Russianinfluence
beganafter 1552, when the Udmurt cameunderRussiandominationand the
numbersof Russian colonistsbeganto increase.By 1917 somesix to eight
hundredRussianloans had beenborrowedinto Udmurt; someof thesewere
usedonly in translationsof religious literature,however.
After 1917 writers and scholars worked to create an Udmurt literary
language.Thoroughlanguagerenewalandhundredsof neologismswould be
necessaryif Udmurt was to be a useful and suppleinstrumentin all the new
social, economic,and cultural domainsof life. The outstandingauthor Gerd
Kuzebaj and like-minded contemporariesbeganthis work with enthusiasm,
but the 1930sand Stalinismput an endto their activity. Only a few neologisms
302 UDMURT

managedto take root, e.g. vir+ser 'artery (BLOOD+VEIN)" nJimtu[=tem

'anonymous(NAME.AND.PATRONYMIC=LESS)" kil+oz 'debate(LAN-
GUAGE+WAR),.

UdmurtText
A text in the literary language, written by the prominent linguist V. I.
Ke1jmakov.Adaptedfrom Csucs(1990: 71).
A: Cyrillic orthography;B: Latin transcription, segmented;C: morpheme-
by-morphemegloss;D: word-by-wordEnglishtranslation;E: freer English
translation.

AI. Y)KMe 6blraTaMeH YMollrec 6b1}:13C'bHHbI

BI. uz-me b'igat=em-e-ja umoj=ges bi"desja-n'i
CI. WORK-slacc KNOWS=NdV-sl- GOOD-cfv ACCOMPLISHES-
adv inf
Dl. my work accordingto my better to carry out
ability

TypTTHCbKO yrocb TO}:lHCbKO 03bbI rllH3

turtt-isjko, j tod-iSiko, ozj'i
ugos gine
STRIVES-s1pres PARTIC KNOWS-s1pres THUS ONLY
I strive after all I know thus only

KaJIbIKeJIbI 3eq JIeCbTbIHbI 6b1raTO rnYblca

kalik-e-l'i ¢jei9lesh'i-n'i b'igat-o su'i=sa
PEOPLE-sI-dat GOOD BUILDS-inf IS.ABLE-slfut SAYS=ger
my people good to build I shall be able [end quotation]

A2. Y}:IMypT KaJIbIK a3bJIaHH3 HO KbIJI33

B2. udmurt kalik aZLlanLaz no ki1-ze
C2. UDMURT PEOPLE FORE-app-s3ine/illAND LANGUAGE-
s3acc
D2. Udmurt people in the future both its language

HO KyJIbTypa33 Me}:l YTe3 rnYblca

no kulhura-ze med uti-oz su'i=sa
AND CULTURE-s3acc PARTIC FOSTERS-s3fut SAYS=ger
and its culture in orderthat it fosters [end quotation]

TblprnllcbKo A3. OCKllCbKO JIblKTHCb

t'irs-iSiko B3. osk-isjko, likt=isj
STRIVES-slpres C3. BELIEVES-slpres COMES=pres.part
I strive D3. I believe coming
 UDMURT 303

BhDKhIOCMhI HO Ta y}:\hIChIH ThlpIIIhIca

vlzl-os-nU no ta udls=ln t'irsl=sa
GENERATION-plur-pl AND OEM.PRO AREA-ine STRIVES=ger
our generations as well this in area diligently

Y)l(aJI03hI
uzal-ozl
WORKS-p3fut
they will work

E1. I try to do my work to the bestof my ability, becauseI know that this is
the only way I shall be able to do my peoplegood. E2. I work hard to ensure
that in future the Udmurt people foster their languageand culture. E3. I
believethat future generations,too, will work diligently in this area.

Referencesand Further Reading

Alatyrev, VI. (1970) (ed.) rpaMMaTHKa COBpeMeHHoro Y;lIMypTCKOro }l3hIKa.
CllHTaKCllC rrpocTororrpe;llJIO)KeHJl}I, Izhevsk: IzdateljstvoUdmurtia.
- - - (1983) 'KpaTKllH rpaMMaTll'IeCKllH OqepK Y;lIMypTCKoro }l3bIKa', in V.M.
Vaxrusev(ed.), Y;lIMYPT-'3yq CJIOBapb,Moscow: Russkij Azyk.
Csucs, S. (1970) 'A votjak nyelv orosz jovevenyszavai1.-11.', Nyelvtudomanyi
Kozlemenyek72: 323-62and 74: 27-47.
- - (1979) 'A votjak-tatar nyelvi kapcsolatok es tOrteneti hatteriik I.-II.',
NyelvtudomanyiKozlemenyek81: 365-72and 82: 135-48.
- - (1988) 'Die wotjakischeSpache',in D. Sinor (ed.) The Uratic Languages:
Description, History and Foreign Influences, Handbuch der Orientalistik 811,
Leiden: Brill, pp. 131-46.
- - - (1990a)Chrestomathiavotiacica,Budapest:Tankonyvkiad6.
- - - (1990b) Die tatarischenLehnworterdes Wotjakischen,Budapest:Akademiai
kiad6.
Decsy, G. (ed.) (1967) The First Votyak Grammar, Indiana University Uralic and
Altaic Series81, Bloomington:IndianaUniversity.
Kelmakov, VK. (1981) 06pa3l\bIy;llMypTCKOH peQll, Izhevsk: IzdateljstvoUdmurtia.
- - (1990) 06pa3l\hIy;llMYPTCKOH peQll 2, Izhevsk: Udmurtskij Institut HaL.
Munkacsi, B. (1896) A votjak nyelv sz6tara. Lexicon linguae votiacorum,Budapest:
MagyarTudomanyosAkademia.
Munkacsi,B. and Fuchs,D.R. (1952) Volksbriiucheund Volksdichtungder Wotjaken,
MSFOu 102, Helsinki: SocieteFinno-Ougrienne.
Perevosikov, P.N. (ed.) (1962) rpaMMaTllKa COBpeMeHHoro y;llMYPTCKOro }l3bIKa.
cI>oHeTllKa II MOPtPOJIOrJl}l, Izhevsk: Udmurtskoe kniznoe izdateljstvo.
Stipa, G. (1960) Funktionen der Nominalformen des Verbs in den permischen
Sprachen,MSFOu 121, Helsinki: SocieteFinno-Ougrienne.
Suihkonen, P. (1990) Korpustutkimus kielitypologiassa sovellettuna udmurttiin,
MSFOu 207, Helsinki: SocieteFinno-Ougrienne.
Vaxrusev,V.M. (ed.) (1956) PYCCKO-Y;llMYPTCKllH CJIOBapb,Moscow.
Vaxrusev, VM., Zaxarov, VN. and Kalinin, L.I. (eds) (1974) rpaMMaTllKa
cOBpeMeHHoro Y;lIMypTcKoro }l3bIKa. CllHTaKcllc CJIO)KHOrO rrpe;llJIO)KeHll}l,
Izhevsk: IzdateljstvoUdmurtia.
Vikar, L. and Bereczki,G. (1989) VotyakFolksongs,Budapest:Akademiaikiad6.
304 UDMURT

Wichmann, Y. (1893) WotjakischeSprachproben,vol. I, Helsingfors: Druckerei der

FinnischenLiteratur-Gesellschaft.
- - - (1901) WotjakischeSprachproben,vol. II, Helsingfors: Druckerei der Fin-
nischenLiteratur-Gesellschaft.
- - - (1903) Die tschuvassischen Lehnworterin den permischenSprachen,Hel-
singfors:SocieteFinno-Ougrienne.
- - - (1954) Wotjakische Chrestomathiemit Glossar, Helsinki: Societe Finno-
Ougrienne.
Wichmann, Y., Uotila, T.E., and Korhonen, M. (1987) Wotjakischer Wortschatz,
Helsinki: SocieteFinno-Ougrienne.
10 Komi
Anu-ReetHausenberg

Komi, together with Udmurt, make up the Permian group of the Uralic
languagefamily. Since ancient times speakersof Komi have lived in the
Vychegdabasinandalongrivers on the upperreachesof the Kama.
History has divided Komi into three dialect groups. The Komi-Zyrian
dialectsare spokenin the northernpart of the area,while the Komi-Permyak
and the Yaz'vagroupsdominatein the southernpart. Somelinguists treatthe
Komi-Zyrian and the Permyakgroups as separatelanguages.As the name
'Komi' is often appliedto the Zyrians only, this may causemisinterpretation
of statistics.
According to the 1989 censusthere were 344,519Zyrians and 152,060
Permyaksin the Soviet Union. 84.5 per cent of the Zyrians (291,542people)
live in the Komi Republic (capital Syktyvkar). The averagedensity of the
populationis 3 peopleper km2 , but the actual distribution is rather uneven.
The Zyrian-dominatedvillages are situatedalongrivers. Of the populationof
the Republic (1,263,000in 1989) 954,000are urban,but of theseonly about
15 per centare ethnicZyrians.
Outside the Republic relatively compactZyrian settlementsexist on the
Kola Peninsulaand in western Siberia. A major exodus (mostly from the
Izhmabasin)took placein the secondhalf of the nineteenthcentury.
The administrativeterritory of the Permyaks,the Komi-PermyakNational
Okrug (capital Kudymkar), is part of the Perm' Oblast. Unfortunatelyabout
14,000Permyaks remained outsidethe Okrug when it was created;they live
in separatevillages in the neighbouringregion. There are also some small
Permyaksettlementsin westernSiberia which date from the secondhalf of
the nineteenthcentury.The socio-political situationrendersofficial statistics
concerningthe Permyaksscarceandnot very reliable.
Speakersof the Yaz'va dialect live eastof the Permyakson the banksof
the Vishera and the Yaz'va Rivers. In the 1960s the number of Yaz'va
speakerswas estimatedat 4,000.

Dialectal Variation
The present-dayvariantsof the Komi languagereflect the impact of natural
factors as well as migrations and contacts- both a result of the language's

305
306 KOMI

linguo-geographicposition in the centreof the Finno-U gric area.Watersheds,

denseforests, and bogland have favoured differentiation. Prolongedsettle-
ment alongsiderelatedtribes, however,has supportedparallel developments
and led to lexical borrowing. The ancient trade routes which traversethe
original Komi homelandsusedto connectScandinaviawith Siberia and the
South.The Komi themselveswereenterprisingmerchants,settingup contacts
with severaldistantpeoples.Different reasonsforced the Komi to migrateto
smaller and more remote rivers and to more sparselypopulatedareas.This
brought about a mixing of dialects and a distortion of dialect borders.The
most heterogeneousarea nowadaysis situated near the Vychegda and the
Pechora.According to documentsit was in the sixteenth and seventeenth
centuriesthat the UpperVychegdaregionreceivedmanynew settlersfrom the
Udora, Vym', and Lower Vychegdaregions,and later on from the basinsof
the SysolaandLuza Rivers. The Komi settlementof the Pechorabasinbegan
at the end of the seventeenthcentury, the settlers coming mostly from the
banksof the Sysola,Vym', andMezen',andfrom the UpperVychegda.
It is traditionally emphasizedthat differencesamongthe Komi dialectsare
insignificant, as most of them are the result of phonological development
(amongthem: variationin the reflexesof a syllable-finalI, depalatalizationof
tJi and I, variation in Anlaut glide phenomenaand in suffixal vowels). While
this statementis true of the Zyrian dialects,the greaterKomi areais divided
by differences in word stress as well as in morpheme inventories and
morpheme-assignment (e.g., the numberand functional load of the cases,cf.
Baker 1985: 50-115).The three-waydialectal division mentionedabove,i.e.
a division into Komi-Zyrian (kZ), Komi-Permyak (kP) and Yaz'va (kY)
varieties,is basedprimarily on administrativeand territorial principles, and
finds linguistic supportonly in isoglosseswhich concernword stress.In kz,
stressis free; in kP, it is morphological;in kY, it is phonological.
The isogloss lines do not coincide with administrativeboundaries.The
southernZyrian and Permyakgroups are linked via the languagepocketsin
the Kirov Oblast (the Upper Kama subdialect)on the one hand, and via the
UpperVychegdaand UpperLup'ya subdialectson the other.Even the Yaz'va
dialect, despiteits notableidiosyncrasies,is linked to the southernPermyak
group via the highly divergentOn' subdialect.
The main criterion underlyingthe traditionaldivision of the Zyrian dialects
is the varied representationof original syllable-final I. In fact, however,the
I-phenomenoncould just as well be treatedas an argumentfor a linguistic
continuum, as it links the southernZyrian dialects with the Permyak and
Yaz'vagroups.Southof the Vychegda,proto-Komi *1 is generallymanifested
as I in all environments.The peripheryis characterizedby phoneticinstability
of I, the occurrenceof I or its surrogates(w, v, 0, I - v alternation) being
environmentallyconditioned(seePhonology,p. 309). Dialectalvariationseen
as a linguistic continuumwithin the whole Komi areahas beendescribedby
Baker(1985).
 KOMI 307

History and Present-dayStatus of the Literary Standards

The oldest written records of the Komi languagedate from the fourteenth
century and are connectedwith the (self-)Christianizationof the Komi. The
northern territories, with their relatively sparsepopulation and rich natural
resources,hadinterestedRussiancolonistsas early as the tenthcentury.From
the twelfth to the fifteenth century the Komi paid tribute first to the feudal
stateof Novgorod,then to the Principality of Moscow,until their territory was
finally annexedto Russia.Statepowerwas to be reinforcedby Christianity.
One of the most enthusiasticmissionarieswas Stefan Khrap, the first
bishop to the Zyrians, called St Stefanof Perm' since 1383. He devisedan
alphabetthat hasbeendescribedas a combinationof Ancient Greek,Aramaic,
and Old Slavonic letters with Komi owners' marks (pas). The so-calledOld
Komi writing systemwas usedto someextentuntil the seventeenthcentury;
later linguistic monumentsuseCyrillic. Only 225 words of Old Komi writing,
fragmentsof religious texts, have beenpreservedto this day (Lytkin 1952).
The development of a literary standard began after the 1917 socialist
revolution, togetherwith efforts to createan independentnational state.But
as the Zyrians and the Permyakshad beenseparatedgeographicallyby that
time, they were subjectedto administrativeseparationas well. In 1921 the
Komi (Zyrian) Autonomous Oblast (the later Komi Autonomous Soviet
Socialist Republic) was formed within the RussianFederation,and in 1925
the Komi-PermyakNational Okrug was establishedin the Perm' Oblast.The
cleavagewas further sharpenedby different literary standards.The Zyrian
version was basedon the Syktyvkar dialect, a natural transitional dialect.
Permyakliterary norms, on the other hand, are not basedon anyonenatural
dialect. Especiallystriking is the adoptionof the I - v alternationfrom literary
Zyrian. Suchofficial differentiationhascontributedto a further weakeningof
the Komi languagein its strugglefor existence.
Subsequentdevelopmentsare the result of Soviet languageand national
policies. In 1940 the Zyrians still constitutedan ethnic majority of about 80
per cent on their own territory. Their languageenjoyedofficial statusas the
languageof national affairs. Beginning in the 1950s,however, there was a
marked rise in Russification. The growing intensity of exploitation of the
naturalresourcesof the Komi landsbroughtwith it an influx of aliens,mostly
ethnic Russians. Accordingto the 1989censusthe proportionof Zyriansin the
Komi Republic had decreasedto 23.3 per cent. The industrial regions are
dominatedby aliens,while the Komi prevail numericallyin the villages on the
Udora, Sysola,Izhma and Upper VychegdaRivers (Lallukka 1990: 125-35).
This means, of course, that there is a great asymmetryin Komi-Russian
bilingualism: only 1.2 per cent of the local Russianscan speakKomi. This
bilingualism is on its way to the secondphrase:the Komi languagehasbeen
oustedfrom businessmanagement,services,andothernon-domesticspheres.
The prognosisfor coming generationsis not good, for Komi is spokenless
and less at home, as well. Only 39 per cent of urban Komi speakKomi at
308 KOMI

home, and no more than 3 per centuse Komi at work (Napalkovet al. 1990).
The abolition of nationalschoolshas,in tum, limited the useof literary Komi.
For example,about 60 per cent of the rural populationbetweenthe agesof
eighteenandsixty who speakKomi andRussianaboutequallyfluently, prefer
to use Komi in speechonly, while Russian is preferred for reading and
writing, sinceschoolingin literary Komi has not beensufficiently systematic
(Rogacev1984).
Although the Permyakshave retainedethnic numeric prevalence(60 per
cent) on their own territory, their languageenjoysevenfewer rights. Only the
most recentchangesin nationalpolicy haveenabledthe local intelligentsiato
beginthe work of restoringthe vernacularculture.

Phonology
Phonemes
Standardliterary Komi has thirty-three segmentalphonemes,sevenvowels,
and twenty-six consonants.The vowels are differentiated by frontnessl
backness,tongue-height,and +1- lip-rounding. To describethe Komi vowel
systemit suffices to use four distinctive features:high, low, back, and front.
The Yaz'vasystemdiffers substantially.

Literary Komi Yaz'va-Komi

ti "i u ti ti u
e
e e- 0 e o 0
a a

In the literary standard,only backvowels are rounded,whereasin dialects

one may find also labialized P' and e¢. In the Yaz'va dialect, however,there
are front roundedit (indicatedby various scholarsas Ii, W, and y) and 0 (0,
:3) insteadof back unroundedvowels. The role of non-low back unrounded
vowels is filled by a reducedvowel written here as 8. According to Lytkin
(1961: 24-6) this vowel is to be placedsomewherebetweena and 0, but the
systemwould leadoneto expecta high back(unrounded)vowel (R6dei 1978:
67-8). In the first syllableit is alwaysunstressed,correspondingto Old Komi
and Zyrian r. In non-initial syllables Yaz'va 8 can correspondto any Zyrian
vowel.
The literary standardhasnon-distinctivelyshortvowels.However,in those
dialects which have undergone *1 > 0, the preceding vowel may be
lengthened,e.g. Izma tuusoo'vernal',pukoonr'to sit', og vetlii" 'I don't walk'
(cf.literary standardtuvsov,pukavnr,og vetliv).
Vowel combinationsmay occur at morphemeboundaries,e.g. va-in 'in
water', kerka-rsi 'out of the house', nu-a 'I am taking (something some-
 KOMI 309

where),. In dialects a long vowel may develop in such contexts via

assimilation,e.g. una-an« una-en) 'many',baba-as« baba-i"s) 'his wife'.
The initial syllablemay containany vowel. In non-initial syllablesrounded
vowels are rare. In non-foreignvocabulary,e never occursin the non-initial
syllable of stems.The inflectional suffixes,if they havea vowel at all, usually
haveoneof thebackunroundedones(i~e, e, a). (In somedialects,f> i and/ore> e
in suffixes.)Thus,in textstheunroundedbackvowelsarethemostfrequent.
Consonantsare distinguishednot only by placeand mannerof articulation
but also by correlationsof voice and of palatalization(seeTable 10.1).
The soundshere transcribedas c, f, and x occur only in relatively recent
Russianloansand are not (yet) part of the Komi corephonemicinventory. (In
older loansthey hadbeenreplacedby s, p, andk.)
All consonantsexceptI and v may occur in any position. The occurrence
of I and v is restrictedby their positionandphoneticenvironment.The literary
standardand those dialects which, like the literary standard,have the I - v
alternation,do not have I in word-final position as there I > v has occurred:
vev 'horse', vel-en 'with a horse'.Word-internallyv occursonly as a substitute
for a pre-consonantalI: vev=teg 'horseless'Morpheme-internalv is not
characteristicof genuineKomi words. In the In'va subdialectofkP, I hasbeen
oustedby v in all positions.
There is no quantitativeoppositionin Komi. Long consonantsmay occur
at morphemeboundaries,however, either through accidental adjacencyor
becauseof word-internalor word-externalsandhi,e.g. 'washes misJ§ii~ 'washesself,
cf. misJkav-'washes(tr)'; [kossor] « kos sor) 'dried-upcreek';'washes li(jidi ~ 'reads'
(morphophonemic ally l'id=j'i-). In kz, morpheme-initialj is assimilatedto a
precedingpalatalizedconsonant,in kP to any precedingconsonant:kz kart-
rtas 'cats' (morphophonemic ally kanLjas), kP ri"itig}nas 'in the evenings'
(morphophonemic ally nt=ja=l-na-s).

Table 10.1 The consonantsystemof Komi

1 2 3 4 56 6

Nasals m n ni
Voicelessstops p t ti k
Voiced stops b d <Ii g
Voicelessaffricates (c) is is.i
Voiced affricates ¢ ¢i
Voicelessfricatives (f) s s sJ (x)
Voiced fricatives v z z zi
Laterals I Ii
Rhotic r
Glide 6
Note: Placesof articulation: 1 bilabial, 2 labiodental,3 (apico)dental,4 post-alveolar,5 palatal,
6 (palato)velar
310 KOMI

Consonantclusters occur word-initial only in onomatopoeicand late

Russian loans, e.g. zborgi~ 'purls', drug 'friend'. Word-finally one finds
consonantclustersof the following types: (1) liquid plus stop/sibilant:koFk
por~
'egg',purt 'knife', gerd 'red', 'pig', gors 'throat';(2) fricative plus t (kZ
only): kost 'interval',bo~t-@ 'take',suv=t-@ 'standup! (tr)' (with v < I). This
type of clusteroccursmostly astridea morphemeboundaryand to the left of
the imperative,which is zero. The clusterdoesnot occur in kP or kY, where
t > @in this position.
Word-internalconsonantismsoccur at morphemeboundaries,e.g. adjacent,
un=~fk
'more',et+medar'both'. In theory, any two consonantsmay be adjacent,but
with the following restrictions. In clusters of three the first consonantis
always either a liquid or a fricative, e.g. vorsli~ 'plays', berd-ni" 'to weep',
gort=sa 'domestic', vevtji~'covers'.Note also that thereareno combinations
of liquids or fricatives alone.
In texts the tenmost frequentconsonantsare, in descendingorder, s, n, d,
k,j, I, v, t, m, ~ (Veenker1982: 436).

Orthography
Both of the Komi literary standardsnow use the Cyrillic alphabet,with
modificationsas detailedbelow.
Palatalizationis indicatedeither by 51, e, e, 10, II, or h to the right of the
relevantconsonant.A non-palatalizedconsonantis written with a, 3, 0, y, or
i following.
The Komi-specific vowels i: e are indicatedby hI, o. Two of the Komi
affricates (c, <Y) are written with simple IJ" q. The remaining affricates are
renderedby digraphs:C = TIll, ~= TIll,
~)I(, ?I = ~3.~3. The transcription-systemsused
by variousscholarsare summarizedby Redei(1978: 65-8).
In 1935-8 an attemptwas madeto adoptthe Latin alphabet,but as in the
caseof otherSovietminorities the effort was suppressed.

Stress
Komi word stressvariesacrossthe dialects.In kZ word stressis free, i.e. one
may stressany syllable without changingword meaning.The first syllable
usually exhibits a slight prominence.In southernkZ dialectsstressis in part
morpheme-bound.
In kP, word stressis morphological:it differentiatesmorphemesand parts
of speech. Stress never falls on inflectional suffixes, whereas certain
derivationalsuffixes are always stressed.Example:ol=gm 'life', QI-em '(s)he
lived' (Batalova1975: 81-123).
Yaz'va word stressis phonological: it is dependenton the etymological
quality of the vowels involved. The non-highstem-vowelsa e 0 0 are always
stressed;e is neverstressed.Whetheror not the high vowels i if u are stressed
dependson further conditions (they are eachthe result of a merger; Lytkin
(1961: 33-4); seePermian,Chapter8 ofthis book).
 KOMI 311

The phoneticcharacterof kZ stressis more pitch-based,while that of kP

andkY stresshasbeencharacterizedas dynamic.

Morphophonology
Due to historical sound changes most non-borrowed Komi roots are
monosyllabic. The most common type of root is CVC; VC, CV, VC] C2 '
CVC] C2 occur more rarely. A longer stem suggestsforeign origin or the
presenceof a latent derivationalsuffix. The most commonsyllable structure
typesare CV andCVe. The basicshapesfor boundmorphemesareV, C, VC,
CV, CVC, all of which can be combinedinto longerchains.Variation in both
free andboundmorphemesis for the mostpart phonologicallyconditioned:

1 Regularstemvariantsarise throughthe I - v alternation.Before a suffix

beginningwith a consonant,stem-finalI > v, whereasbeforea vowel the
I remains,e.g. vel-en 'with a horse',vev=teg 'horseless'.This rule does
not apply to onomatopoeicwords,e.g. kolski:' 'smacks';niavzi:' 'miaows',
niavest- 'miaows once', nor doesit apply to adjectivesderived with the
suffix =ov (- =ev), e.g. lez=ov-esi'bluish (plur)'.
2 Certainstemsshow syncopationof the vowel in their final syllable when
certain vowel-initial derivational suffixes are attached: geti"r 'wife',
getr=asi- 'getsmarried',but geti"r=a 'married'.
3 Certain,chiefly nominal, stemsshow a paragogicconsonantto the left of
vowel-initial suffixes.The consonant,alwaysoneof the set} k t m, is only
partly predictableon the basisof the synchronicphonology.Examples:
bi"g 'foam', bi"gj=a 'foamy'; lok- 'comes', lokt-i-s '(s)he came'; mes
'cow', mesk-en'with a cow'; §in 'eye', §inm-i"Si 'out of the eye'. This
phenomenonis for the mostpartrestrictedto kZ; in kP it is quite rare.
4 In dialects,variantsof bound morphemesfollow their own idiodialectal
phonotacticrules, giving rise, for example,to alternationssuch as e -e,
i" - i, or d - t in suffixes. Theremay also be sporadicdeviation,however,
in both derivational and inflectional suffixes; in severaldialects,vowel
harmony at an earlier stage of the language has been replaced by
secondaryassimilation, resulting in allomorphs which are difficult to
classify(cf. the UpperVychegdaforms in Table 10.2).

The stem variants of the personaland reflexive pronounsand of certain

verbs are conditionedmorphologically.The plural of the personalpronouns
and the ordinalsfor 'first'and 'second'are suppletive.

Morphology
Noun Declension
The morphologicalcategoriesencodedin the Komi noun are number,case,
andperson.Personis associatedwith both possessionanddefiniteness.
312 KOMI

Table 10.2 Noun declensionparadigms(after Baker1985: 66,111)

Permyak Zyrian Zyrian Yaz'va

(standard) (Upper Vychegda)

Nominative -0 -0 -0 -0
Genitive -yen -len -len -Ian
Ablative -visi _llsi -lisi _lisi
Dative -ve -II -Ii -Ie
Accusative -0, -es -0, -es -0, -(e/aJo/e)s -0, -es
Instrumental -en -en -en -(e/aJo)n
Comitative -ket -ked -ked -ket
Abessive -teg -teg -teg -teg
Consecutive -la -la -la
Inessive -In -In -(i/aJo/u)n -en
Elative -lsi -lsi -lsi -(i/aJo/u)si _isi
Illative -e -e -e/aJo/0 -e
Approximative -vani -lani -laui _Ii ani
Egressive -siaui -siaui -siaui -siaui -siaui
Prosecutive -eti -ed -(e/a)d -et
Transitive -ti -ti
Terminativel -e¢i -e¢i -e¢i -( e/aJo/e)¢i -siaui
Terminative2 -vi
Superessive -vln
Superlative -ve
Sublative -viSi
Perlative -yeti
Superterminative -ve¢i
Comparative -sia

The categoryof numberhas two representations:singular(unmarked)and

plural. The marker for plural varies acrossthe dialects: kZ -jas (- -jes), kP
-(j)ez, -e, kY -jez.
The numberof casesis fifteen for Yaz'va, seventeenfor kz, and twenty-
four for kP. The plural suffix precedesthe casemorpheme,e.g. iJoj-jas-ked
'with sisters'.For an exampleof the variation of casemorphemesacrossthe
principal dialectsseeTable 10.2. It shouldbe notedhere thatthe consecutive
caseis absentonly in southernkP dialects, and that the comparativein -Jia
occursin southernkz dialectsas well.
Postpositionsare functionally akin to casesuffixes. They come after the
noun in its basic (sN) form and can be complementedby plural and case
suffixes of their own, in which casethey resemblethe second membersof
compounds,e.g.pizan vi'v TABLE SURFACE 'the surfaceof the table',pizan
vil-i'n TABLE SURFACE-ine 'on the table', pizan vi'v-jas-i'n TABLE
SURFACE-plur-ine'on the tables'.In southernkP dialectsthe development
of postpositionsinto case suffixes is synchronically evident (Baker 1985:
175-201).
The associationof certain nouns with certain case suffixes seemsto be
 KOMI 313

influencedby animacy.Animate nounstend to take -es in the accusative,and

their use of the illative, inessive,elative and tenninativeis restricted.The
transitiveandprosecutiveforms do not occur,their functionsbeingcarriedout
by postpositions.Usually an inanimatenoun takes the instrumentaland not
the comitativesuffix. At the sametime, it must be notedthat thereis no one-
to-one correspondencebetween natural and grammatical animacy. Thus
'cow' can be inanimate,e.g. niebi mes-0'I boughta cow'; cf. a{l{la mort-es
(with accusative-es) 'I seea man'.
The personal(possessive)suffixes are: sl -ej, s2 -rd, s3 -is, pI -nim, p2
-nia, p3 -nis. Plurality of the possessionis indicatedby meansof the regular
pluralizer-jas: vok-is 'hislherbrother',vok-jas-is'his brothers'.
The sl suffix may also perform a vocativefunction. The singularsecond-
and third-personsuffixes indicate definitenessmore often than possession,
e.g. lun-is gaza 'the day is merry'. The s2 suffix -ia tightens the sphereof
referencecloserto the collocutors,while s3 -is is neutralin this respect.
As a rule, the personalsuffix precedesthe casesuffix, but in somecases
the positionsare reversed.In the inessiveand illative, the caseendingshave
fusedwith the possessivesuffix, resultingin a syncretism:gort-am 'in(to) my
house',gort-anis 'in(to) their house'.
The personalsuffixesmay, in certainsyntacticfunctions,be attachedto the
adjective (to mark definiteness). They may also be attached to certain
pronounsand verbal nouns,and participatein the formation of collective and
emphaticpronouns.

Pronominal Declension
Komi pronouns may be divided according to their function into prosub-
stantives,proadjectives,and pronumerals.The prosubstantives(or pronouns
proper) are generally declined in the same way as nouns. Personaland
reflexive pronounsexhibit deviant stem allomorphy, however, and certain
casesareformed differently from their nominal analogues.Examples:me 'I',
accusativemene,ablativemenSirm(kP menBim);mi 'we', accusativemijanes,
ablative mijanliSi. For the third person plural the kZ dialects offer the
following variants:naja - najezda- nijii.zda - nije - nija - nia - niaa - niida
- niidajas - sijajas - sijii.zda - enZajas.
The functions of possessivepronounsare performedby the nominative,
genitive, and ablative forms of the personalpronoun and the genitive and
ablativeof the reflexive pronoun,e.g. menamjort PRO.sl-genFRIEND 'my
friend', as-la-nim kerka SELF-gen-pl HOUSE 'our own house'. The
reflexive stemas also denotesthe neutral 'one'sown', as in as kerka dor-in
'nearone'sown house'.
The reciprocalpronounseta+med,meda+med'eachother' take both case
andpersonalsuffixes; only the secondmemberis declined,e.g. meda+med-iSi
'from eachother', meda+med-nis-lr'for oneanotherof them'.
Demonstrativepronouns serve to distinguish between the closer (taje
314 KOMI

'this', tacem 'such')andthe more distant(sije, seeem).Both setscantakethe

emphaticproclitic e--, e.g. e--taje 'this (one) here',e--sicem'like that (one)
there'. Whenfunctioning as substantivesthey may take suffixes for plural and
case,but not for person,e.g. ta-i"n 'in this one',taje-jas-len'of these'.
Interrogativeandrelativepronounsmay standfor nouns(kod - kodi 'who',
mij 'what'), adjectives (kucem, mij+Siama 'what kind of) and numerals
(kiini"n, mijta, mij mi"nda 'how many'). The interrogativepronounkod(i) (kY
kin) is declined,andcantake personalsuffixes,e.g. kod-nia 'which of you?',
kod-is taje-jaspi-i"Si WHO-s3 THIS-plur BOY-ela 'which of theseboys?',kY
kin-ket 'with whom?'.In dialectsone may meetvariantssuchas kin - kini -
ked 'who', keda 'whatkind of, muj - mej 'what?',andothers.
Negativepronounsareformed by meansofthe negativeparticleni(e) - nii,
e.g. nie+kod 'nobody', ni+efi(k) 'not one' (efik 'one'), nie+kodnan 'not a
singleone'.In substantivalfunction, they aredeclined,andnie+kod(nan)may
alsotake a personalsuffix, e.g. nie+kodnan-niin'noneof us'.
Indefinite pronounsare formed by meansof the particles--ke and =Siure.
The latter forms a lexical unit with the pronoun, e.g. kod=Siure-li" 'to
someone'.The former particle, on the other hand, is attachedas a loose
enclitic to the right of any inflectional suffix, e.g. kod-liSi--ke'from someone',
andevento the right of a postposition,e.g. kod+vil-e--ke'onto someone'.
Totalizing pronouns(inter alia bia 'every(one)',stav 'all') may be declined
if in substantivalfunction. Stavmay alsotake personalsuffixes,e.g. stav-nam
(sl instrumental)'I wholly', stav-niin 'we all'.

ComparativeandSuperlative
Adjectives may be put into comparative and superlative grades. The
comparativeis formed with a suffix =~i"k
(kP =fJk, kY =~ik =~g).
- The
superlativeis most commonly constructedby means of preposedmed or
med-Sia (the latter is written separately),e.g. med+jon, med-Siajon '(the)
strongest'.The superlativemay also be formed by meansof a reduplicative
constructionwith the elative: jon-iSi jon STRONG-elaSTRONG, roughly:
'the strongestof the strong'. The negative comparativeis built by adding
the comparativesuffix to the negativeparticle abu, e.g. abu=~rkmiCia 'less
beautiful'. Occasionally, comparativesmay be formed from adverbs or
even from verbs, e.g. vo~ 'early', vo~=~rk 'more early', eni o-g={ifk viSilrfJ
NOW NEG-sl=cvfAILS-conneg'I am less(often) ill now' (Coates1982).

Verb Conjugation
The finite paradigmof the verb distinguishesthe grammaticalcategoriesof
person,number (singular v. plural), tense (presentv. various pasts),mood
(indicative, imperative),and +/- AFFIRMATIVE. The endingsare predom-
inantly syncretic,and simultaneouslyencodeseveralgrammaticalmeanings.
There is little stem/suffix allomorphy, except that stems which have an r
 KOMI 315

Table10.3 Zyrian (kZ) verb conjugation,with a few relevantkP and kY

variants:mun- 'goes'

Affirmative Negative

Indicative
Presenttense
sl mun-a o-gmun-0
s2 mun-an o-nmun-0
s3 mun-e o-z mun-0
pI mun-am(kY mun-.!!me) o-g(e)mun-ej
p2 mun-annldlad(kY mun-.!!te) o-n(e) mun-ej (kP o-de myn-e)
p3 mun-enl(kY mun-~nes)
mun-ase) o-z mun-nl (kY o-z mun-nis)
Futuretense(= present,except:)
s3 mun-as
p3 mun-asnl(kP mun-ase)
Imperfect
sl mun-i e-gmun-0
s2 mun-i-n e-nmun-0
s3 mun-i-s, mun-i-0 e-z mun-0
pI mun-i-m (kY mun-i-me) e-g(e)mun-ej
p2 mun-i-nnld (kY mun-i-te) e-n(e)mun-ej (kP e-d[e] mune)
p3 mun-i-snl(kY mun-i-nis, kP e-z mun-nl (kY i-z mun-nis)
mun-i-se)
Perfect

s2 mun-emld abu mun-emld(kY Qbe

mun-~m[a])mun-~m[a])
s3 mun-em(a) abu munem(a)(kY Qbe
mun-~m[a])mun-~m[a])

p2 mun-emnld abu munemnld(kY Qbe

mun-~masi) mun-~masi)
p3 i
mun-emaes i (kY Qbe
abu munemaes
mun-~masi) mun-~masi)

Pluperfect
sl mun-emavel-i abu mun-emavel-ille-g vev
mun-ema
s2 mun-emldvel-i(-n) etc.
s3 mun-em(a)vel-i etc.
pI i vel-i( -m)
mun-emaes etc.
p2 mun-emnldvel-i( -d) etc.
p3 i vel-i( -nl)
mun-emaes etc.
Imperative
s2 mun-0 e-nmun
pI mun-ame(j) o-g(e) mun-ej
p2 mun-e(j) e-nemun-ej
316 KOMI

before the infinitive suffix -nf lose this f before a vowel-initial suffix, e.g.
vetlf-nf 'to go', vetl-a 'I go', vetl-enf'they go', vetli'-{!J 'go!'
The negative forms are built by means of either a negative verb or a
negativeparticle.For the systemseeTable 10.3.
The perfectandpluperfect'tenses'refer moreto point of view thanto time.
Formally, they arecompound,consistingof the presenttenseof the finite verb
plus a form of the verb vev- 'is', eitherthe past-tenseform vel-i or the perfect
participle vel=em. Examples: me bi'd+lun si~lf vel-i nieb-a kalbas PRO.s1
EVERY+DAY PRO.s3-allIS-s3pastBUY-s1presSAUSAGE 'every day, I
usedto buy him a sausage',buxta-e u~-e vel-i nie+i'~'d sor BAY-ill FALL-
s3pres IS-s3past NOT+BIG BROOK 'a small brook fell into the bay',
kefajka-is vi!l=em nok~-e paB dor-in MISTRESS-s3 IS=past.part
BUSIES.SELF-s3presSTOVE SIDE-ine 'the mistresswas busying herself
near the stove'. The context often implies inferentiality, i.e. that the events
being reportedwere not directly witnessed.In addition, various shadingsof
Aktionsarthavebeenobserved(Serebrennikov1960: 58-82).

Non-finite Verb Forms

The Infinitive is formed with the suffix -nf. In dialectsit may co-occurwith
plural, comparative,and personalsuffixes, e.g. Lower Vychegda ju-ni~niin
ko~m-e DRINKS-inf-p1 GETS.DRY-s3pres'we want to drink', pukti"Si=
ist-ni~jas-teots=ist-ePLANTS=momentaneous-inf-plur-s2 HELPS=momen-
taneous-s3pres '(s)heis having a go at helping (us) plant them' (note the use
of the s2 suffix; cf. p. 313). According to Ludykova (1984: 173-7) there is a
secondinfinitive, namely the verbal derivativesbuilt with =m(=em), which
functionsin a manneranalogousto the Baltic-Fennicsupine.Traditionally the
=m(=em) derivatives are classedalong with the other productive deverbal
derivatives; this approachemphasizesthe importance of the semanticsof
eachverb, and the productivity of the suffix - it may be attachedto all verbs
(Fedjuneva1985: 66-9). Unlike the derivatives,which are fully declinedlike
ordinary nouns, the =m(=em) verbal nouns occur in the singular only, and
they take only a few casesuffixes (elative, consecutive,illative). Examples:
dugd-i-s jekt=em-i"Si CEASES-past-s3DANCES=verbal.noun-ela'(s)he
ceased (from) dancing',DANCES=verbal.noun-ela
vu~-i-s ~il=em-e BEGINS-past-s3 SINGS=
verbal.noun-ill '(s)hebeganto sing'.
Participles are formed by meansof the suffixes =isJ (presentimperfective
passive),=an (presentperfectpassive),=em (perfectactive/passive),and=tem
(negativeperfect active/passive).This traditional list has been extendedby
Cypranov(1987: 17) to include a participlebuilt with =men, which expresses
the extent of the action, e.g. eBBi'd lomtf=menpes ONCE HEATS=ext.part
FIREWOOD 'enoughfirewood to heatup (theoven) once', murt=sa ki'v=men
gelesBARELY HEARS=ext.partVOICE 'abarelyaudiblevoice'.
Verbal adverbs, or gerunds, express time relations between actions.
Simultaneityis indicatedby forms built with =emen,=sen, and =ig (kY =ki
 KOMI 317

- =k), precedence by forms built with =miSit(en) - =miSi; and successionby

The The suffix of the negativegerundis the sameas the
forms built with =te~.
nominal abessivecase,viz. =teg. The simultaneousgerundsuffix =ig usually
occurs in the instrumental or illative case and with personal suffix, e.g.
leBB=ig-a-sDESCENDS=ger-ill-s3'when(s)hedescended/descends'

Derivation
The formation of the comparativeof the adjective,and of verbal nounsand
participles,hasbeendiscussedon p. 314 becausethey are regularformations
typical of their word classes.True derivationalsuffixes, by contrast,are not
regular in this sense:they do not associatewith all membersof their word
class.
Most of the noun-formingsuffixes are unproductive.Having onceformed
nouns from both nominal and verbal stems, they have now ceasedto
participate in active derivational processes.They are characterizedby
sporadic variation of their vocalic component(all are of the shapeVC).
Examples:kuF=em - kuF=em - kuF=im - kuF=im 'spawn'(kuF- 'spawns'),
Sieter- Siefi"r 'currant',asiv - asuv 'morning' (cf. aski 'tomorrow').They are
also characterizedby a motley semanticstructure,involving a greatdeal of
synonymy, e.g. kis=ed - kis=an - kis=as 'clothes', Sioj=an - Sioj=ed -
Sioj=em - Sioj=eb 'food' (Fedjuneva1985: 43-50,60-61,116-17).
Two groups of derivational suffixes are productive in the formation of
nominals:

1 the deverbalsuffixes =i"§i and =an. The first forms agents,the second
forms nouns referring to actions, means,and objects. Examples:from
ger- 'ploughs', ger=isJ 'ploughman',ger=an 'ploughing'; from Sioj-
'eats',Sioj=an 'food'; from si"r- 'cuts',si"r=an 'scissors';
2 secondarynominal suffixes which have developedfrom independent
words and can be attachedto most word classes.The most productive
representativesof this group are from lun 'day' and tor 'piece'.
Examples: bur=lun 'goodness'(contrast the syntagm bur 19n 'Good
day!'), pov=tem=lun 'fearlessness'(pov- 'fears', pov=tem 'fearless'),
etkotF=lun 'similarity' (etkodJ 'similarly'), vuzal=an=tor 'article for
sale'(vuzav- 'sells',=an participle),em=lun 'property,somethingextant'
(em 'exists').

Komi verbs can be derived from nouns by meansof suffixes (including

zero), but also from other verbs. Phonaesthemicthemesenterinto construc-
tions with auxiliary verbsto form onomatopoieicverbs.Of the primary verb-
forming suffixes the most productive are =m, =t, =Si, =1, = (e)d, =al, and
=is(t). More frequently,however,one meetscompoundsuffixesconsistingof
two to four primary suffixes chainedtogether.Most suffixes (and all suffix
chains) are polysemantic. Thus =t and =(e)d express causativity and
318 KOMI

factitivity, =1, =al, =i1, =i'vl, =li'vl, =lal expressdurativity andfrequentativity,

=§i, =1/, ={! expressinchoativity, reflexiveness,or simplenon-transitiveness.
Examples: lebav- 'flies', lebfi:' 'takes off', lebfi'vli:' 'takes off repeatedly',
lebti:. 'raises',lebtav- 'raisesslowly', lebti"Sti:. 'raisesa little'.

Compounding and Reduplication

Komi compoundsare formed by the juxtapositionof word stems.They are
distinguishablefrom analogouslybuilt syntagmsby the fact that the meaning
of a compound differs from the mechanicalsum of the meanings of its
components. Most such compounds are nouns, e.g. paSi+kem 'clothes
(FUR.COAT+FOOTWEAR)" Siin+va 'tears (EYE+WATER)', Ciuni+kic
'ring (FINGER+HOOP)" kad+kolast 'period (TIME+INTERVAL)'. Since
the semanticshift is not always obvious the borderlinebetweena compound
and a syntagmmay be indistinct. The orthographytends to follow tradition
ratherthan semantics.
Compoundingis surpassedin productivity by reduplication and paired
words. Paired words may be found in practically all word classes,formed
from associative-metonymic,antonymous,or synonymousstems; often the
differencebetweensuchreduplicativesandtheir simplicesis one of emphasis
only.
The componentsof paired words denoting collectives may have an =a
suffix: Cioj=a&vok=a SISTER=a&BROTHER=a 'sister and brother',
mam=a&zon=a'motherandson',vel=a&mesk=a 'cattle(HORSE&COW)"
lun&voj 'day and night', ni"r&vom 'face (NOSE&MOUTH)', ber-e&vo{!-e
'here and there (BACK-ill&FORTH-ill)" pu-ni"&pef.av-ni" 'to cook and
bake',pi"r-ni"&pet-ni" 'to go in and out (ENTERS-inf&EXITS-inf),, lun&lun
'from day to day', maHn&maHn 'very close',aj=tem&mam=tem'without
father and mother', Sioj=an&ju=an 'food and drink'. In onomatopoeicpairs
the stemmay vary considerably:giina&gama + ki'v-ni" 'to rattle' (ki'v- = 'gives
off sound'),tur&bar su-ni" 'to speakhurriedly and indistinctly' (su- 'says'),
terli&merli 'nonsense',Pigi&Pogi 'staggering(ly)'.

Syntax
Komi syntaxshowsmany Uralic traits as well as innovationsthat are mostly
due to Russianinfluence. Many aspectsof the field still await systematic
exploration.

The Noun Phrase

In a sentencethe Komi nounphrasemay function as subject,object,adverbial
or predicate.The head of the noun phrasemay be a noun or pronoun, an
infinitive, a substantizedadjective or a numeral. The head may take an
attributefrom any nominal subclass,e.g. bur ni'v 'good girl', taje kniiga 'this
book', kik pi 'two boys', ger=tem mu 'unploughedland'. As a rule the
 KOMI 319

attribute precedesits head and does not agreewith it: gaza lun-jas 'merry
days',bur ni"v-jas-en 'with good girls'. A noun following a numeralis always
in the singular, varying only in case,e.g. vit pi-Ii" 'for five boys'. There are
no prepositions.Before a postpositionthe noun is usually in its basic (sN)
form.
An attribute,in tum, may be precededby membersof variousword classes,
e.g. lim je{zia dera 'snow-white shirt', ruB kotJi muder ni'v FOX LIKE
CLEVER GIRL 'a girl clever like a fox', Efiek oponi kerka din jofi EuEK
OPONJ HOUSE BASE BROOK 'brook near the houseof Eljek Oponj '. A
noun attribute denotinga possessorprecedesadjectival attributesand is put
in the genitive, or, sometimes,in the ablative,e.g. ur-jas-Ien kuZ! gen=a bef.-
jas-nis SQUIRREL-plur-gen LONG HAIR=adj TAIL-plur-p3 'the long-
hairedtails of the squirrels',batJi-liSi murtsana pet=emposnii"tJiik da neriniik
kor-jas WILLOW-abl BARELY YET EXITS=perf.partSMALL AND TEN-
DER LEAF-plur 'the small andtenderjust-openedleavesof the willow'.
Extensionsof a noun phrasemay also come afterthe head;in this position
they agreewith the headin numberand case.The plural morphemein this
constructionis -eSi.Appositionsalso agreewith the head.Examples:vaj-fJ te,
Varuk, niani-te, Beskia-esda peSi-esBRING-imp PRO.s2VARUK BREAD-
s2accTASTY-acc AND WARM-acc 'Vera, will you bring somebread,tasty
and warm'; naje, vaz kotJi veSiel-eSida ki"pia-esi PRO.p3 OLD LIKE GAY-
plur AND MERRY-plur 'they, gay and merry as before .. .'; ketjajn-is su-i-s
si"-li; Lisko-Ii" MASTER-s3 SAYS-past-s3PRO.s3-allIlSKO-all 'the master
saidto him, to Lisko ...' .

The Verb Complex

In a sentencethe verb complexfunctions as a predicate.A predicatemay be
(1) a simple verbalpredicateconsistingof finite (including compound)forms
of the verb, (2) a compoundverbal predicateconsistingof finite verb forms
and an infinitive, or (3) a nominal predicateconsistingof a copula and a
noun.
The finite predicateagreeswith the grammaticalnominative subject in
numberand person.In generic-personaland impersonalsentencesthe third-
person singular form is used: zer-e 'it's raining', pemd-i-s 'it was getting
dark', mort-ia ed bia+tor verm-as veB-ni" HUMAN-s2 PARTICLE ANY-
THING IS.ABLE-s3fut DOES-inf 'Man can do anything, after all'. If a
quantifier phraseis functioning as subject, the verb form dependson the
genusverbi (active or passive)and the animatenessof the subject: sentences
with an activeverb andanimatesubjectpreferthe plural form of the predicate
verb, and in the oppositecasethe singularis preferred,e.g. ekmismort vol-
i-sni"NINE HUMAN COMES-past-p3'ninepeoplecame',soBiY-i-skiK kerka
BURNS.INTR-past-s3TWO HOUSE 'two housesburneddown'.
Modal verbs, verbs which denotethe beginningor ending of an activity,
and several verbs of perception and cognition form the predicate in
320 KOMI

combinationwith an infinitive: sije dugd-i-s Sierav-ni' '(s)he stoppedlaugh-

ing', dietjina-jas zavodiit-i-sni' veleCiCii'-ni' 'the boys began to do their
homework'.The traditional infinitive may be replacedby the verbal noun in
=em (see under verbal nouns, p. 316), e.g. dugd-i-s majsaSi=em-i"Si'(s)he
stoppedworrying' ,per-i-{J busit=em-e'it starteddrizzling'. In the caseof the
modal verbs pof/- 'is possible; is permitted', kov- 'is necessary',10- 'is
obliged' and others,the subjectis in the dative: meniin kol-e mun-ni'PRO.sl-
dat IS.NECESSARY-s3pres GOES-inf 'I must go', si~li·poi-emun-ni'mun-ni' '(s)he
is allowedto go'.
In the function of the copulain a nominalpredicatewe find the finite forms
of the verb 'is', vev- (- 10-). In the presentaffirmative, however,the copula
is zero: lun-is zar 'the day is hot', lun-is lo-e zar 'it's going to be a hot day'.
The place of the predicate can be filled by a noun, adjective, numeral,
pronoun,or participle.Examples:Viktor Savinvel-i-{J bur artist-en 'v. S. was
a good actor (NB -en instrumental)" povodidia-is vel,=la=Si=an=a 'the
weather is changeable',pu-jas lo-i-ni' kerka sud=ta-eSi njin TREE-plur
BECOME-past-p3HOUSE HEIGHT=adj-plur ALREADY 'the trees have
grown as tall as the house',sije vel-i-{J kekjamis=ed'(s)hewasthe eighth'.
The copula of a nominal predicateagreeswith the subjectin personand
number,while a predicativeagreeswith the subjectin number.The plural of
a predicatenoun is formed by meansof -jas; other word classestake -eSi.
Examples:nioF=nan-niin morjak-jasFOUR=collective-plSAILOR-plur 'we
four are sailors',tuj-jas bur-eSi 'the roadsare good',mu-jasvel-i-ni' ger=tem-
eSi 'the fields wereunploughed'.
The modality of a sentenceis modified by modal words, verb semantics,
and their correspondingsentencestructures.Komi has severalverbs which
denotepsycho-physiologicalreactions;in sentencescontainingtheseverbs,
the experienceris put in the accusative,genitive,or dative,e.g. menekiz=ed-e
PRO.sl-acc COUGHS=causative-s3pres 'I feel like coughing', Mise-lenvun-
i-{J Sioj=em-is MISE-gen IS.FORGOTTEN-past-s3EATS-verbal.noun-s3
'Misha forgot to eat', meniin o-z ui=Sii~{J (IevlevaPRO.sl-dat NEG.PRES-s3
SLEEPS=intr-conneg'I don't feel sleepy'(Ievleva 1984:5-9).
Passive,durative,andcontinuativemeaningsare conveyedby certainpast-
tense forms of the verb. Repetition of a verb form also emphasizesthe
intensity or duration of an activity, and for the samepurposea particle or
certainnon-finite forms of the verb may be used: vot sije Siil-i-s--ke i Siil-i-s
'oh, how (s)he sang!', jez-is zu=emenzu-e 'what a crowd!' (zu- 'buzzes,
swarms').
Negationis conveyedby a negativeverb (presenttensestem0-, pasttense
and prohibitive e-) or one of the negativeparticlesabu, nie, nii. In sentence
negationthe negative verb immediately precedesthe main verb; emphatic
negationis expressedby using the negativeverb or particle beforethe focus:
piSime-see-gmegiZLETTER-s3accNEG.PAST-sl PRO.slWRITES 'itwasn't
I who wrote theletter',si~len
lo-i-{J lo-i-{J nie eli, a dasmesPRO.s3-genBECOMES-
 KOMI 321

past-s3NOT ONE BUT TEN COW '(s)hegot not one,but ten cows'. Double
negationoccursif a sentencecontainsbothanegativeverb(expressingsentence
negation)andanegativepronoun,adverb,or anemphaticnegativeparticle:naje
rti+nem o-z su-ni" PRO.p3 NOTHING NEG.PRES-p3SAYS-conneg.plur
'they don't sayanything', si~len niv niv rti pi abu PRO.s3-genDAUGHTER NOT
SONNOT.EXISTS '(s)hehasneitherdaughternorson'.

Simple Sentence
An ordinary Komi sentenceconsistsof a subjectand a predicate,with their
extensions,the usual order being S[ubject] P[redicate]O[bject] A[dverbial].
But any other order is possible,constraintsapplying only within the phrase.
Only the positionof the nominalpredicateafter the subjectis fixed: kerka i"i?id
'the houseis big' (contrastthe nounphrasei"i?id kerka 'big house').
Closer observationreveals certain regularitiesin the relations obtaining
betweenother sentenceconstituents.If a sentencebeginswith an adverb,the
predicateprecedesthe subject: sije kad-§art koP-i-@ kif! vo 'since thistime
passedtwenty years';tedli"=teg koP-i-@ vo 'a yearpassedunnoticed'(contrast
vo koP-i-@ tedlr=teg 'the year passedunnoticed'). An infinitival subject
normally precedesthe predicate,and an infinitival object usually follows it,
e.g. ker-en mun-ni" vel-i-@ le§id 'to go by reindeerwas pleasant',tan pot-e
i Cieri vugrav-ni" 'here it is permitted even to fish'. Deviant word order is
usually causedby discoursestructure.A focusedelementis usually brought
nearerto the beginningof the sentence,and new information movestowards
the sentence'send. Examples: pukal-a me gorHn SITS-lpres PRO.sl
HOME-ine 'I'm sitting at home',pu-see-g mepered-@'asfor the tree(pu-se),
it wasn't I who felled it', me vok-liSi bo§-t-i viP krtiga 'I took a new book
from brother', vok-§art pi§melokt-i-s 'a letter hascomefrom brother'.
In writing, the focus may be raised into prominenceby meansof word-
order, sometimesalso with the emphatic particles (ne, ie, inter alia). In
speech,prominenceis primarily achievedprosodically: the focusedelement
receivesgreaterstressandhasdistinctiveintonationalprofiles.
Prosodicmeansplay an importantrole in illocutionary meanings,as well.
Although thereare grammaticalmeansto distinguishvariousfunctional types
of sentence,the participationof distinctive stressandintonationis obligatory.
An imperative sentence,for example, is formed either by means of the
imperativeform of the verb or by specialparticles,but a commandcan also
be expressedas a statement(te vo-an as kad-e 'you will come on time'). An
interrogative sentenceis usually formed with a question word such as mij
'what?' as in mij te-kedlo-i-s 'what happenedto you?' or with a particle,e.g.
--e in o-z--e sije ted 'doesn't(s)he know?' But a questioncan be implied
purely by intonation, so that the stressfalls on the focusedword. In writing,
the actual focus can be discernedonly from the answer,e.g. if the focus of
- te gazet-serteb-i-n? 'did you buy a newspaper?'is on 'did you buy' (rteb-
i-n), an affirmative answerwill be - da, rteb-i 'yes, I did (I bought)'. If on
322 KOMI

the other hand the focus is on 'you' (te), the answerwill be - da, me 'yes,
it wasI (who boughtthe paper)'.

Non-simpleSentence
Co-ordination between the componentsof a compound sentencecan be
expressedeither by purely prosodic means (intonation, pause) or by
co-ordinatingconjunctionssuchas i, da, daj « da + i), nii. Examplesof the
latter: mijtem oZi"rlun tani, i stavis ve§serekujle 'so many riches here- and
all lying aboutuseless',uifavni" dugdim, daj zermis 'we stoppedworking and
it startedraining', Iti mort vu?:er nieken e-z tiaav, nP e-z ki'v neikucemsi"
NEITHER HUMAN SHADOW NO.WHERE NEG.PAST-s3IS.VISIBLE-
conneg NOR NEG.PAST-s3IS.AUDIBLE NO.KIND.OF SOUND 'not a
soul could be seen anywhere, not a sound could be heard'. Usually the
conjunctionstandsat the beginningof the clause,but the contrastingparticle
--a can be addedto the end of the clauseonly: bur jez-li" prazdniik talun--a,
mi so uifalam GOOD FOLK-dat HOLIDAY TODAY-whereas PRO.pI
BEHOLD WORKS.-pIpres'good peoplehave a holiday today, but we are
working', kucemabi-is eli voj-en pa§kal=ema, teri"t na niemtor e-z vev--a
HOW FIRE-s3 ONE NIGHT-ins SPREADS.INTR-evid, YESTERDAY
STILL NOTHING NEG.PAST-s3IS--whereas'how the fire hasspreadin one
night, eventhoughyesterdaytherewas still nothing' .
Subordinationis expressedby conjunctionsand relating words. Most of
the subordinators which relate subclausesto the main clause are of
pronominal origin, though some are borrowed.The position of a subclause
dependson the semanticsas well as on the structureof the sentenceas a
whole. As a rule the subclausecomesafter the main clauseand beginswith
a subordinator,e.g. vot i mevpalaeni, miJla me oli 'hereI am thinking now,
(wondering)why I havelived', a kodi siiSi miza, miJ kejdisis abu? 'but who
is at fault that thereis no seed?'On the other handthe subordinatorsda and
si always standat the end of their subclause,e.g. pePukroz buras rozjale si,
bur ru{leg voas 'if the rowan blossomsbloom well, there will be a good
harvest',taje kniigase me tenia §eta, lilJidia daTIl give this book to you
whenI havereadit through'.
The conditional clause usually precedesthe main clause. The relating
particle --ke never standsat the beginningof its subclause;rather,it follows
the first (rarely the second)component,e.g. ez--kesog vel, gaz eg tede 'if
there had not beensorrow, I would not haveexperiencedjoy', jen--ke§ias,
tekedloam sudajasen'if God (jen) will, we shall be happywith you'.
The subclauseof indirect speechis analogousto any other subordinated
clause. Which subordinatoris selecteddependson the characterof the
message:a statementhas mij, a commandor wish has med(im), and an
indirect questionusesthe samewords as a direct question(e.g. kucem'which,
what kind?', kimi"n 'how many?'), e.g. Stepan cektis, med viBBiSiasni"
'Stephanorderedthat they should wait', Nadia jualis, kucem kniiga niebni"
 KOMI 323

'Nadyaaskedwhich book to buy'. Indirect yes/noquestionsarebuilt with the

interrogativeparticle --e, e.g. iBe#ikPederjualis, rlebis--ecPedi"ssi1i"piscaFse
'little Peteraskedwhether Grandfatherhadboughthim the rifle'.
Insteadof a subordinateclauseone often finds a non-finite construction,
functioning as subject, object, attribute, predicate,or adverbial. Examples:
(subject)mun=i"F-jas puk§-i-sni"piz-eGOES-part-plurSITS-past-p3BOAT-
ill 'thosewho were leaving sat down in the boat'; (object, adverbial)stav-se
viil-i daFt=ema bur=a dumist=emenALL-s3acc IS-s3pastPREPARES-evid
GOOD=advCONSIDERS=gerund'one had preparedeverything,everything
having beenwell considered';Fifim ar-esBuzem-se
ti"rt=e~ na me-nemameskola-e
meded-i-sSEVEN YEAR-acc FILLS-perf.partYET PRO.sl-accMOTHER
SCHOOL-ill SENDS-past-s3'alreadybeforeI was sevenmy mothersentme
to school'; zerkalevo~-i"n
Buzem-seBuzem-sekosed=ig su-i-s ... MIRROR FRONT-
ine FACE-s3accDRIES=gerundSAYS-past-s3 'drying his face before the
mirror, he said ...'

Lexicon
The basicword-stockconsistsmostly of monosyllabicroots with the structure
CV(C), rarely Vc. Examples:va 'water', ki 'hand',vo 'year', Fin(m-) 'eye',
vir 'blood', jem 'needle',ruB 'fox', ki"k 'two', ur 'squirrel', ov- 'lives', ju-
'drinks', mun- 'goes', vaj- 'brings'. The same structureis typical of early
loans, but herebisyllabic roots begin to tum up, as well. The oldest layer of
loanwordscomesfrom various Iranian languages,of which the following is
a sampling:das 'ten',dom- 'tethers',ideg 'angel',njebeg'book',purt 'knife',
zamii 'gold'.
The contactsbetweenthe Permyaksand the Volga Bulgarsare reflectedin
the Komi lexicon by a few dozenwords suchas gob 'mushroom',ban 'face',
Barla 'sickle', and Forkni 'turnip'. The northwesternKomi dialects have
borrowedfrom Baltic-Fennic,while the northerndialectshavetakenmost of
their reindeer-breeding vocabularyfrom the Nenets.
The greatestnumber of loans comes from Russian.The earliest have a
typical Komi structure, but in more recent loans even foreign sounds are
retained.
The non-borrowedword-stockhas developedand grown richer by means
of derivation and compounding,and belles-lettreshave servedas a channel
for the disseminationof regionalisms.Furtherexamplesof word-formation:
bi"dmeg 'plant' < bi"dm- 'grows'; redmanlun 'fertility' < redm- 'bearsfruit';
lomtas 'fuel' < lomt- 'heats';ki"riinpas 'signature(ki"riin 'hand'+ pas 'sign')'.
Thesemethodshavealsobeenusedfor creatingnew terminology,e.g. njimtan
'nominative'< rlimt- 'names',ni"rpi"r su=F=an 'nasalsound'(ni"r 'nose'+ pi"r
'through';su=F- '(re)sounds(intr)" =an participle; "pronouncedthroughthe
nose"), but for the most part vernacular creations retreat before terms
borrowedfrom Russian.The resultis a pidgin well exemplifiedby a sentence
324 KOMI

taken from a university handbook: SostavnejskazujemejpodPei.asBej-ked

seglasujt=emCOMPOUND PREDICATE SUBJECT-comAGREES=perf.
part 'agreementof a compoundpredicate with the subject', in which all
morphemesare Russiansavethe suffixes of the comitative and the negative
participle.

Komi Text
Sysoladialect.Adaptedfrom Redei(1978: 110).

A: text in phonological transcription, segmented; B: morpheme-

by-morpheme gloss; C: word-by-word English translation; D: freer
English translation.

car = caritative;egr = egressive.

AI. drug sjinm-e usLi, va

B 1. SUDDENLY EYE-ill FALLS-s3past WATER
C 1. suddenly into (my) eye(s) it fell water

plck-a-s tIdal-e pa~=tempa~=tem mort,

MIDST-ine-s3 IS.VISIBLE-s3pres CLOTHING=car HUMAN.BEING
in its middle canbe seen naked person

abI va+vel dor-a-s a va

NOT.EXIST WATER+SURFACE EDGE-ine-s3 BUT WATER
not water'ssurface at but water

pIck-a-s. A2. sije vel-i tIdal-e toljke

MIDST-ine-s3 B2. PRO.s3 IS-s3pret IS.VISIBLE-s3pres ONLY
in its middle C2. it was is visible only

sitan-sjanLIs vel+dor cjash-Is, sjin-se

BUTTOCKS-egr-s3 UPPER+EDGE PART-s3 EYE-s3acc
from its buttocks upperbody its/thepart its eye(s)

kuni=ema, jUr+sji=tem, n1r=a vom=a

SHUTS.EYES-s3perf HEAD+HAIR=car NOSE=adj MOUTH=adj
it had shut(evidently) bald with a nose with a mouth

pelj=a i klrlm-jas-Is bok kuzja

EAR(S)=adj AND HAND/ARM-plur-s3 SIDE ALONG
with ear(s) and its hands/arms side along
 KOMI 325

niuzed=ema-esj A3. me vi¢jed-i sij-es

STRETCHES=perf.part-plurB3. PRO.sl WATCHES-slpast PRO.s3-acc
were stretched C3. I I watched it

zel di"r
VERY LONG.TIME
very for long

Dl. SuddenlyI saw it: a naked personwas visible in the water, not at the
surface,but in the water. Only its upperbody was visible, from the buttocks
(up); it had its eyesclosedand was bald, (but) it had nose,mouth, and ears,
(and) its armswere stretchedalong (its) side(s).I watchedit for a long time.

Referencesand Further Reading

Baker, R. (1985) The Developmentof the Komi Case System:A dialectological
investigation,MSFOu 189, Helsinki: SocieteFinno-Ougrienne.
Batalova,R.M. (1975) KOMII-IIepwrl\Kali ~lIaneKTOAOrIDI, ~lIaneKTOAOrIDI, Moscow: Nauka.
Batori, I. Sz. (1967) 'Wortzusammensetzung und Stammformverbindungim Syrja-
nischen mit Berticksichtigungdes Wotjakischen',unpublishedDr. Phil. thesis,
University of Gottingen.
Bubrix, D.V (1949) rpaMMaTIiKa nliTeparypHoro KOMII 1I3hIKa, Leningrad:
University of Leningrad.
Cemyx, VA. (1981) rnaronhHoecnoBoo6pa30BaHlieB KOMII1I3hIKe, aBTope<pepaT
~IICCepTal\lIl1,
KaH~II~aTCKOIf Tartu: n.p.
Coates,J.G. (1982) Thecomparative
-~)I(hIK comparativesuffix in contemporaryKomi usage',
Transactionsof the Philological Society,Oxford.
Cypanov, E.A. (1987) Mop<ponorlill IIpIl'laCTliH B KOMII 1I3hIKe, aBTope<pepaT
~IICCepTal\lIl1,
KaH~II~aTCKOH Tartu: n.p.
Fedjuneva,G.V (1985) CnoBoo6pa30BaTenhHhle CY<P<PIIKChI CYIn;eCTBIiTenhHhIX B
KOMII1I3hIKe, [An SSSR,Komi filial, Institut jazyka, literatury i istorii,] Moscow:
Nauka.
Ievleva, T.M. (1984) CIIHTaKclic rnarona B KOMII 1I3hIKe (CB1I3h rnarona c
IIO~ne)l(aIn;IIMII ~oIIoHeHlIeM), aBTope<pepaT ~IICCepTaIJ;llIl,
KaH~II~aTCKOH
Moscow: n.p.
Kneisl, M. (1978) Die Verbalbildung im Syrjiinischen, Veroffentlichungen des
Finnisch-Ugrischen Seminars an der Universitat Munchen, Serie C:
Miscellanea.Band9. Munich.
Lallukka, S. (1990) The East Finnic Minorities in the SovietUnion: An Appraisal of
the Erosive Trends, Suomalaisentiedeakatemiantoimituksia, Sarja B, 252,
Helsinki: FinnishAcademyof Sciences.
Ludykova, V.M. (1984) 'CKa3yeMoe c m-OBhIM IIH<pIIHIITIiBOM B KOMII 1I3hIKe'
SovetskoeFinnougrovedenie20, 3: 173-7.
Lytkin, VI. (1952) ,lJ;peBHeIIepMcKIiH 1I3hIK, Moscow: Academy of Sciencesof the
SovietUnion.
- - - (1961) KOMII-1I3hBIIHCKIiH Moscow:~lIaneKT, Moscow: Academy of Sciencesof the
SovietUnion.
aBTope<pepaTB KOMII 1I3hIKe B
Manova, N.D. (1976) CnO)l(HOIIO~'lIlHeHHhle CnO)l(HOIIO~'lIlHeHHhle IIpe~nO)l(eHlIlI
IICTOpll'leCKOM OCBeIn;eHIIIi (OCHoBHhle TIIIIhI), aBTope<pepaTaBTope<pepaT KaH~II~aTCKOH
IICTOpll'leCKOM
~IICCepTal\lIl1, Moscow: n.p.
326 KOMI

Napa1kov, A.D., Popov, A. A. , Smetanin, A.F., and E.A. Cypanov (1990)

B3aHMO~ell:CTBHeHa'IHOHaJIbHO-H3hIKOBOll:rrOJIHTHKH BYCJIOBHHX KOMH ACCP',
'Ocy~eCTBJIeHHe
in KOMi KbIB oHiH OJIOMbIH = TIp06JIeMhI CPYHKI\HOHHpOBaHHH KOMH H3bIKa B
cOBpeMeHHblxYCJIOBHHX, Syktyvkar.
Redei, K. (1978) Syrjiinische Chrestomathie.Mit Grammatik und Glossar, Studia
Uralica 1, Vienna:Associationof AustrianLearnedSocieties.
Rogacev,M.B. (1984) 'BHJIHHrBH3M CeJIbCKHX KOMH (COI\HOJIOrH'IeCKHll: acrreKT)',in
B3aHMO~ell:CTBHe cpHHHO-yropCKHXH PYCCKOro H3bIKOB, Syktyvkar.
B3aHMO~ell:CTBHe
Serebrennikov,B.A. (1960) KaTeropHHBpeMeHH H BH~aBH~aB cpHHHo-yropcKHxH3bIKax
rrepMCKOll: H BOJI)KCKOll: rpyrrrr, Moscow: Academy of Sciencesof the Soviet
Union.
Stipa, G. (1960) Funktionen der Nominalformen des Verbs in den permischen
Sprachen,MSFOu 121, Helsinki: SocieteFinno-Ougrienne.
- - - (1970) 'Impersona1iaim SyIjanischen', in Symposion iiber Syntax der
uralischenSprachen15-18Juli 1969in Reinhausenbei Gottingen,Abhandlungen
der Akademieder Wissenschaftenin Gottingen, Phi101ogisch-historische K1asse
3176,Gottingen.
Veenker,W. (1982) 'Zur phono1ogischenStatistik der syrjanischenSprache',Etudes
Finno-Ougriennes15: 435-45.
110bUgrian
LaszlOHonti

The common ancestorof the ObUgrian languages,proto-ObUgrian, was

spokenafter the breakoffof Hungarian(roughly, in the fIrst half of the second
millennium BCE) and before the split into proto-Vogul (Mansi) and proto-
Ostyak(Khanty) (roughly, in the fIrst centuriesCE).
This long period of independentexistence- nearly two millennia -
distinguishes proto-ObUgrian clearly, in terms of cladistic status, from
brancheswhich diverged more recently, such as Permian or Fennic. The
resulting enormous divergencesand independentinnovations in p[roto-]
O[styak] and p[roto-]V[ogul] make it extremely diffIcult to reconstructthe
history of proto-ObUgrianand its daughterlanguages.Although Ostyakand
Vogul share a good stock of inherited root and suffIxal morphemesand
common syntactic features, each of them also shows features which are
characteristicof it alone,or which are presentin it to a higher degreethan in
the other, and in many instancesthe sameor similar aims are achievedby
different, historically unrelated,means,as in the caseof the different Ostyak
andVogul passivemorphemes.As a resultof thesedifferences,over the years
various scholarshave called into questionthe validity of setting up a proto-
ObUgrian node, a proto-Ugric node, or both. One aim of this chapteris to
makeclearwhy this authorcannotsharesuchviews.
Howeverit must be notedthat a full and clear outline of the prehistoryof
either of the two ObUgrian languagesis yet to be written. Many questionsof
detail remain unresolved,and the following presentationwill accordingly
containlacunaeand someunevennessof emphasis.An appendixat the end of
the chaptersummarizesthe key ObUgrianandUgrian sharedfeatures.

Internal Divisions of Vogul

Most conservativefrom the phonologicalpoint of view were the southern
dialectsof Vogul, which at the beginningof the twentieth century were still
spokenalong the T[avda] River. To this group belong those of the villages
Janyckova(TJ), Candyri (TC), and Gorodok (TG); all theseforms of Vogul
are now extinct.
The easterngroup of Vogul dialectsincludesthoseforms spokenalong the
Konda River and its tributary, the Jukonda(Jk). These are the dialects of

327
328 OBUGRIAN

Vogul spokenby peopleliving along the Lower Konda (KU), Middle Konda
(KM), and Upper Konda(KO).
To the westerngroup of dialectsbelongNorth Vagilsk (VN), SouthVagilsk
(VS) and thosewhich were spokenalong the Lower (LU) and Middle (LM)
L[ozva] and the P[elymka] Rivers. It appearsthat this dialectgroup no longer
hasany speakersliving.
The northerngroup, which servedas basisfor the Vogulliterary language,
is made up of the following dialects: Upper Lozva (LO), So[sva], Sy[gva],
and Ob'.

Internal Divisions of Ostyak

Among the Ostyakdialectgroupsit is the easternthat is the mostconservative
from the phonologicalpoint of view. The easterngroup consistsof the far-
easternpair of dialectsV[ach] and Vasjugan(Vj), often cited together(VVj)
becauseof their genetic closeness,and the dialects of the Surgut region,
including Tremjugan(Trj), Pim, and J[ugan]. The Sal[ym] dialect may also
be classifiedas eastern,althoughit is in certainrespectstransitionalbetween
the easternand southerngroups.
The southerngroup consistedof dialectsspokenalong the Irt[ysh] River.
Best-known among theseare the Upper (DN) and Lower (DT) Demjanka,
K[onda], and Kr[asnojarsk] dialects. It appearsthat no speakersof these
dialectsremain.
To the northern group belong the Sherkal (Ser), Kaz[ym], and Syn[ja]
dialects,as well as, to the south,the Atl[ym] and Ni[zyam] dialects,although
thesetwo latter show southernmorphologicalfeatures.The O[bdorsk] dialect,
spoken around present-daySalexard,preserves well the vowel system of

Figure 11.1 Vogul andOstyakdialectgroups

Vogul dialect groups Ostyakdialect groups

p
u n dr e s e n t - d
aro ay
tra
n

transitional
sit
io

al
na

transit
on
l

system
iti

tran
ns

tra
tra

ns

ional
sitio

iti
on

system
nal

al
 OBUGRIAN 329

proto-West Ostyak, but is somewhatdeviant among the northern group

becauseof innovationsin its consonantism.

In what follows, I shall survey the developmentswhich brought about the

present-dayOstyakandVogul dialects fromtheir protolanguages chiefly upon
the basis of material cited from two Vogul, and two Ostyak, dialects: the
phonologicallyconservativeT Vogul and VVj Ostyak on the one hand, and
the innovativeSo Vogul andKaz Ostyakon the other.

The PhonemeSystemof Proto-ObUgrian

Vowels
We mustdistinguishthe vowel systemsof fIrst andnon-fIrst syllables.
The developmentof the vowels of the fIrst syllable in the ObUgrian
languageswas investigatedand written up in detail by W. Steinitz (1950,
1955). He also attemptedto reconstructthe vocalism of the fIrst syllable of
their commonancestor;writings on this subjecthave beenpublishedin part
(1989), but parts still remainin manuscript.Thereare seriousproblemswith
Steinitz' assumptions,however, and I have found it necessaryto attempt a
revisionof his views (seeHonti 1980, 1982a,1983bfor details).
For Steinitz (n.d.), the proto-ObUgrian (pOU) vowel inventory was
identical with that of proto-Ostyak (pO), with the exception that pOU
probablylacked*66: seeTable 11.1.
My own reconstructiondiffers primarily along structural, quantitative,
lines. Insteadof Steinitz' full/reduceddichotomy,I proposea systemof long
v. short vowels, with roughly equal membershipin the two quantitative
classes:seeTable 11.2.
I thus set up a three-way oppositionamong the short front umounded
vowels (*i : *e : *ii) where Steinitz had only *e. WhereasSteinitz reckoned
with a split of his *e in pOU which led to the correspondences pV *i = pO
*e on the onehandandpV *e =pO *e on the other, I setup respectivelypOU
*i and *e/*ii for thesecorrespondences. Similar considerationshave led me
to proposetwo tongue-heightsfor the shortback vowels *u and *0. Steinitz'

Table 11.1 pOD/pO vocalismaccordingto Steinitz

Front Back
-R +R -R +R
Full *ii *iiii *iY *uu High
*ee (*00) *00 Mid
*lUi *cece *aa *M Low
Reduced *e *6 *a *0
330 OBUGRIAN

Table 11.2 pOU vocalismaccordingto Honti

Front Back
-R +R -R +R
Long *ii *titi *iT *uu High
*ee *66 *00 Mid
*iiii *aa Low
Short *i *ti *u High
*e (*6) *0 Mid
*li *a Low

*ila and *rere are unnecessaryfor a reconstructionof pOU, since they are
allophonesof Steinitz' *aa and *00 on the one hand, and of *66 and *liii on
the other.
On the way to pV, the pOU non-highvowels merged,suchthat e.g. *00 and
*aa became *aa. Some pOU long vowels shortenedin pV, probably in
connectionwith syllablestructure,e.g. long *uu becameshort*u in the closed
syllable of pOU *juust- 'rewards'> pV *just-, but remainedlong in the open
syllable pOU *kuu/i'ir;)y 'bag' > pV *kuur;)y. On the other hand, shortening
probably also occurredin connectionwith the quantity of the vowel of the
secondsyllable,cf. the short *u of p V*tuljaa 'ring' from pOU *8uu&(VVj).
In pO, *0 and *a became*ii, and *6 and *a became*re, sporadically(but
usually in the neighbourhoodof velar consonants).Non-low pOU long
vowels becamelow in pO if the vowel of the secondsyllablewas stem-final.
The following examplesare meant to illustrate these and other regular
developmentsof the pOU vowels of the first syllable:

Proto-ObUgrian
*11 *njn1V 'arrow' >
pV *nieelV > T nieei, So niaai
pO *niaal > VVj niaai, Kaz niat'H
*uu *8uup 'oar' >
pV *tuup > T toop, So tuup
pO *iuup > V iuw, Vj juw, Kaz l(Jp
*00 *800p;)s 'net-needle'>
pV *taas> T taas 'small stick'
pO *SOOP;)s> VVj saw~, stick' Kaz S(JP;)S
*aa *aac 'sheep'>
pV *aas> So aas
pO *aac > VVj aac, Kaz ailS
*ii *8iil;)yt;)- 'mixes' >
pV *tiil;)yt;)- > T tiit-, So teei;)yt-
pO *iiil;)yt;) > V iiil;))'t;r, Vj iil;))'t;r, Kaz lil;)f-
 OBUGRIAN 331

*ee *meel= 'warm' >

pV *maru= >T miiiil=iit; So maal=tip 'thaw'
mee{;,k,> VVj mee{;,k,Kaz meel;}/(
pO *meel~k
*aa *aampV 'dog' >
pV *aampV > T iiiimp, So aamp
pO *aamp> VVj iiiimp, Kaz aamp
*titi *ctitiUk'fog'>
pV *siiUwkw > T siikw, So seeI]wkw
pO *cuuy > VVj caay, Kaz siw
*66 *66UkV 'mother' >
pV *aauwkwV > T ayCw\ So aaI]wkw
pO *aauki > VVj iiiiI]ki, Kaz aaI]ki
*u *8uy 'summer'>
pV *tuj > T toj, So tuji
pO *ioU > V 10I], Kaz lUI]
*0 soy;}S, 'comes'>
*jok~t-
pV *joy~t-jokt-, > T jokt-, Sojoxt-
pO *joy~t-jokt-, > VVjjoy;}t-, Kazjox;}t-
*a *tay= 'reindeerhide' >
pV *taw=;}P > T taw;}lj, So towl 'skin, leather'
pO *tay=ta > Kaz taxti 'reindeerhide'
*i jokt-, 'flies' >
*8iy~l-
pV *tiyl- > T tiiwl-, So tiyl-
pO *iey~l-
jokt-, > V ley;}l-, Vjjey;}l-
*e/*a *jey/*jay 'father' >
p V *jay > T jaw, jiiw, Sojiy
pO *jey > VVjjey, Kazjiw
*ti soy;}S, 'autumn'>
*8tik~s
p V *ttibs > T taks, So takw;}s
pO *s6y~s
soy;}S,> VVj soy;}S,Kaz sus
*6 *86y '(s)he,it' >
p V *taw > T tiiw, taw, So taw
pO *i6y > V lOy, Vj joy, Kaz iuw

In non-first syllables,the vowel inventory of pOU was as follows:

Front Back
Long ii 11
aa aa
Short ii ii

This systempersisted,virtually unchanged,into pV andpO.

Frontlbackvowel harmony,inheritedfrom proto-Uralic, survivedinto only
332 OBUGRIAN

someof the ObUgrian dialects:southernVogul andeasternOstyak.Elsewhere

it perished, a victim of various vowel shifts and mergers. For example,
contrast theT Vogul form-pair mini-ii '(s)he goes' : kart-ee '(s)he pulls'
(frontlback suffix-alternantsii : ee) with the correspondingSo Vogul forms
min-i, xart-i (phoneticallyfront, phonologicallyneutrali in the suffix of both
forms). Similarly, we haveVVj Ostyakfront-vocaliciiiimp-iim - iimp-;»n 'my
dog' v. back-vocalickaat-am - kuut-fnn 'my house',but Kaz Ostyakaamp-
;»n, xaat-;»n.

Word Stress
In pOU, as in the present-dayObUgrlanlanguages,word stresswas probably
on the first syllable,exceptwhendisplacedby sentencestress(emphasis).

Consonants
When non-initial, p(F)U *w became*y in proto-Ugric (Honti 1985b: 150),
e.g. pFU *kiwe 'stone'» VVj Ostyakkooy.
The non-palatalizedp(F)U sibilants*s and *8 mergedin proto-Ugricto *8,
which subsequently,after the breakupof pOU, developedinto pV *t and pO
(secondary)*i.
In the inventorybelow, the reconstructedvoicelesslateralfricatives *4 and
*V replacethe traditional *6 and *&i. It is more convincingto reckonwith *4
and *V through pOU, after which in early pV they becameapproximants(*4
> *1, *V > *lj), and in pO *4> *1 but *V> *j; later in pO, a new *4 arose,from
*8 (Honti 1992).

w j k lj
k
8
4 V
s 8
cj C
P t k
mn lj ni I)
1 lj
r

As is clear from the inventory, in pOU, as in pV and pO, there was a

j
correlationof palatalizationaffecting non-obstruentcoronals(1 : lj, n : n , 4 :
V).
In pOU, *w was restrictedto word-initial position, while *y, *yW, and *1)
did not occur word-initial. The etymological data indicate that *4 was also
restrictedto non-initial position.
There was only one labializedconsonantin pOU: *yw. This segmentwas
lexically quite rare, but occurredinvariably in first-personplural possessive
and verb suffixes.
 OBUGRIAN 333

The pOU sibilants *s and *8 are not the descendantsof p(F)U *s and *8
(which mergedas pOU *8, seeabove),but are ratherthe result of a split, and
date back to early proto-Ugric, when the depalatalizationof p(F)U *sj had
begun.This depalatalizationstartedfirst in those*sLinitial words which had
a following *j or a palatalized consonant,i.e. we may assume simple
dissimilation*sj > *s. In otherenvironments,*sj became*8.
The following examplesmay serveto illustrate the main developmentsof
the consonantsfrom pOU to the daughterOU languages:
Proto-ObUgrian
*w- *wuui;}m - *wuuimV 'dream;
sleep'>
pV *uulmV> T oo/;m, So uu/;m
pO *wuul;}m - *waal;}m > VVj wuu/;m, Kaz wuul;m
'dream',VVj aa/;m, Kaz (J(Jl;m
'sleep'
*j *jobt 'comes'(seeabove)
*-y(- ) *8uy 'summer'(seeabove)
*_yW(_) *jey-VyW - *jay-VyW 'our
father' >
pV *jay-VyW > T jiiw-;MI,jaw-;MI, SojiY-;M1
pO *jey_;}yW > V jey-oy,VjjeY-;MI, Kazjiw-;M1
*8 *8uy 'summer'(seeabove)
*-i(-) *weei;}m 'marrow' >
p V *waiil;}m > T wa/;m, KU KM wa!a!/;m, VS
waa/;m,VN P wa/;m, LU LO So
waa/;m
pO *weel;}m > VVj wee/;m,Trj J wii/;m, DN KSll
Ni weet;m,0 weel;m
*~*s 'Prunus'Prunuspadus'>
*~iIm(V)
pV *ljiImV > T /jeem, KU Paam,KM KO
/jeem,VS VN P LU LM /jeem,
LOSoPaam
pO *jafun> VVjjaam,Trj J joom, DN Ni
juum, Kazj(J(Jm
*s *siImV 'fish scale'>
pV *siImV> T seem,So saam
pO *saam> VVj saam,Kaz saam
*8 *8im 'heart' >
pV *8im> T sam,So sim
pO *sem> VVj sem,Kaz sam
j *daiir;}y - 'hurts' >
*c
p V *cjaiir;}y- > T dark-, Sotjeer;ry-
~aariy- ~aariy-
pO *daiir;}y- - *drerer;}y- > Kaz siaari-, Trj tjeer;ry-
334 OBUGRIAN

*c *ctitil]k 'fog' (seeabove)

*p *pii8- 'cooks,boils' >
pV *piit- > T piit-, Sopeet-
pO *palH:- > Vj piiiil-, Kaz paal-
*t *tay- 'reindeerhide' (seeabove)
*k *k66y 'stone'>
p V*kaw > T kiiw, So kaw
pO *k66y > VVj kooy, Kaz keew
*m *meel= 'warm' (seeabove)
*n floats' 'drifts, floats' >
*niIp~t-
pV *mlt- > So naat-
pO *naap~t-
floats' > VVj naaw;]t-, Kaz nQQP;]t-
*ni *njiI1V 'arrow' (seeabove)
*1] *ctitil]k 'fog' (seeabove)
*1 *luuntV 'goose'>
pV *luuntV > T loont, So lu(u)nt
pO *laant > VVj llulnt, Kaz IQQnt
*lj *ljtitik - 'reviles' >
pV *ljtitik- > So ljuuk-
pO *ljrerey~t-'reviles'> VVj ljrerey;]t-, Kaz Paaw;]t-
*r 'reviles'
*rtiy~t- 'mixes'
'reviles'>
'reviles'>
pV *raw~t- T riiwt-, So rawt-
'reviles'>
pO *r6y~t- VVj roy;]t-, Kaz ruw;]f-

The Developmentof the Vogul PhonemicSystem

Vowels
Onceagainwe must distinguishdevelopmentsin first from thosein non-first
syllables.
Proto-Vogul vowels of the first syllable are shown in Table 11.3. This
inventory is the author's own departure from that proposedby Steinitz
(1955).
In southernVogul, we must reckon with the following changes:(1) *ii >
ee to the left of *r; (2) *iI > ee to the right of *j or a palatalizedconsonant,

Table 11.3 P Vvowelsof the first syllable

Front Back
-R +R -R +R
Long ii titi 11 uu High
aa aa Low
Short i ti 1 u High
a a a Low
 OBUGRIAN 335

but otherwisei:l fell to ee; (3) *iiii, which was quite rare, mergedwith *ii; (4)
*uu > 00 and *u > o.
In the other three dialect groups *aa went to *00 (So. ad) and *ii went to
ee in non-palatalenvironments;*11 fell to aa, mergingwith the aa from *iHi.
For examples,seethe materialassembledabove.
In non-initial syllables,the vocalic inventory of pOD survivedintact into
pV; the frontlback opposition of the reduced vowels (*;) : (*;) *~) became
neutralized,however.
In southernVogul, high vowels fell (*ii > ee, *11 > ee); in the north, *ii and
*11 fell togetheras e; it is possiblethat there was a phonemicquantitative
distinctione :ee.
SomeVogul non-initial vowel inventories:

Tavda Sosva
Long ee ee a
aa aa e
a
Short a a

Consonants
In some dialects of pV, *k and *1) frequently, and *y always, became
labializedwhen adjacentto roundedvowels; *y thus mergedwith the already
extant but rare *yw. When such labialization occurred, the rounded vowel
itself becameunrounded.All instancesof *yW eventuallybecamew.
Among the laterals,the oppositionfricative: approximantwas neutralized
in favour of the latter, yielding *i, *1> *1, *ti, *lj > *lj. By pV, pOD *c had
lost its stopcomponent(> *8).
The pOD fricative *8 became*t.
The consonantinventoryof p Vis setforth below:

wj y yW
S
S S
j
c
t ~ kk W
P kk
mn mn ni 1) 1)w
1 lj
r

With the breakupof pV, in northern and most easterndialects *s and *8

j j
mergedas s. The palatalizedpV affricate *c lost its stop component(*c >
s-i) in all dialectsexceptthoseof the south, whereit persistedas an affricate,
particularlyin syllable-initial position.
336 OBUGRIAN

Table 11.4 pO vowels of the first syllable

Front Back
-R +R -R +R
Full 11 titi n uu High
ee 66 00 Mid
lili cece aa iiii Low
Reduced e 6 a 0

The Developmentof the OstyakPhonemicSystem

In the first syllable, pO distinguishedfull v. reducedvowels (Steinitz 1950:
2): seeTable 11.4.
Apart from sporadicchangesin a few words, the vowel inventory of VVj
Ostyakpreservesthe pOU distribution. The journeyfrom pOU to Kaz Ostyak
is much more complicated:we may sketchit in ten steps.
In proto-WesternOstyak,two major developmentsoccurred:

1 *rere became*00 when adjacentto *k or *1], but otherwiseunroundedto

*iUi;
2 *66 unroundedto ee, and *iiti unroundedto ii.

The transition from proto-Westernto Kaz Ostyak involved the following

changes:

3 to the left of pO *w, *k, or *1], *0 > *u, then*6 > *0; elsewhere*6 > *e
and *0 remained;
4 to the right of *w, *e became*u, unlessadjacentto *j, in which caseit
became*i; otherwiseit became*li;
5 *iI becameii, while *uu and *ii underwentan uncompletedsplit, with *ii
> ii, ee, and *uu > *uu, *00. At this stage, the first-syllable vowel
inventorymusthavebeen*ii *uu *ee *00 *lili *aa *iiii *e *a *0;
6 the quantitativeoppositionwas then neutralizedamongthe high vowels,
suchthat *uu x *u > u and *ii x *i > i;
7 to the right of *w, *i roundedto *u;
8 the full backmid roundedvowel *00 becameslightly fronted to (J(J;
9 *aa roundedto aa,and *a roundedto 0 (but *a remaineda to the left of
x and gx);
10 finally, the depletedslots were replenishedby the changes*lili > aa and
*li > a.

The resultingfirst-syllable vowel inventory for Kaz Ostyakwas therefore as

follows:
 OBUGRIAN 337

Full Reduced
ee QQ u u Non-low
aa aa a o Low
-R +R -R +R

For examples,seethe etymologicalmaterialassembledin previoussections.

The pO vowel systemin non-initial syllables was preservedintact into
VVj. The origin and morphophonemicbehaviourof the passivesuffix (-uj-)
of the first and second personsremainsunclear,however.
In the predecessorto the western dialects the frontlback harmonic
oppositionwas destroyed.First, each of the phoneme-pairs*ii and *iI, *aa
and *lUi mergedat the expenseof the markedmembers,yielding *ii and *aa.
Later, the quantitativepair *ii and *i mergedas *i, and *e to the left of *j
became*i. The resulting non-initial inventory in Kaz Ostyak was thus only
eeaa i e.
Consonants
From pOU to pO, the mostimportantconsonantchangesoccurredamongthe
coronals.

1 The pOU palatalizedvoicelesslateralfricative *ti became*j in pO, while

*i became*1. After this change,pO soonacquireda new *i:
2 pOU *8 becamepO *i, e.g. pOU *8iIn V 'new' becamepO *iaan.

The change *8 > *i was blocked (2a) when *8 occurred in word-initial

position and there was an *s later in the word. In this environment,total
assimilationtook place,i.e. *8-s > *s-s, e.g. pUgric *8iiysV 'autumn'> pOU
> pO *s6y~s,
*8iik~s(Honti (Honticf. pV *tiik~s
(Honti(Honti 1974: 369-70, 1986b:262; cf. also
Collinder 1960: 96).
The cacuminals(*1). *l) are the result of a secondarydevelopmentwithin
pO: they originally occurredas allophonicvariantswhenadjacentto the (non-
distinctively) cacuminal affricate *c, then spread,beginning with affective
and descriptivevocabulary,to other positions,where they becamephonemic
(ltkonen 1961: 58-9).
The consonantinventoryof pO is setout below:

Glides w j y yW
Centralfricative s
Lateralfricative i
Affricates j
c c
Stops P I). k
Nasals mnn j
I). I). I). 1J
Lateral approximants I).
lj 1
Other r
338 OBUGRIAN

Table11.5 ConsonantsystemsofVVj and Kaz Ostyak

VachIVasjugan Kaz

w j y w j
s S sJ S X
is
P j ti k p j k
m n ni l)- I) m n ni l)- I)
I Ii ! t ii I
r r

The systems of the modem dialects are the results of the following
developments:

3 The laterals *<1: and *1 mergedeverywhere:in VVj as I (exceptin initial

position in Vj, where *<1: > *j > j - fJ), in the Surgutgroup and in Kaz as
I, in all other dialectsas t. In somewhatparallel fashion, (3a) *li > P in
VVj Syn 0, Pin SurgutandKaz, and ti in Sal Irt At! Ni Ser.
4 The labialized velar glide/approximant*yW remained only in Trj; it
becamey in Vach, w elsewhere.
5 *y remainedmore or less unchangedin the easterndialects, changed
(partly) to 1] in southerndialects, and was replacedby w, j, and x in
northerndialects.
6 *c remainedunchangedin easterndialects,and changed(partly) to sin
southerndialects.*ci lost its fricative releaseto becometi in the eastand
south; it becamesi in the north. In Kaz, both affricates lost their stop
component:*c > s, *ci > si.

Furtherdetail may be found in Ronti (forthcoming).

Morphophonology
There were in all likelihood alternationsinvolving various kinds of cluster
simplification at morpheme boundaries,much as in the present-dayOU
languages,suchas *l)c+C ~ *I)C - *cc.
Proto-Ugric hada classof verbswhich I shall term unstablestems.In these
verbs,longer stemsendingin the glides *y, *w, or *j alternatedwith shorter,
vowel-final stems which lacked these semi-consonants.The majority of
unstablestemswere inheritedfrom p(F)U, e.g. pFU *sewe- 'eats' > pUgric
*8ewV- > > POU *8iiy(V)- > pV *tii- - *tiij- - *tiiy- (> T tii- - tiiiij-, So tee-
- teey-- taaj-) andpO *iii(y)- (> V lii- -liiy-, Vj ii- - iiy-, Kaz lee- - leew-);
for detailsseeRonti (1985a:51-4; 1987).It is not clearwhethereachunstable
stem had two or three variants, i.e. whether more than one distinct final
consonantcould be distinctive in any given paradigm.In all Vogul dialects
exceptT, threevariantsare attested:the w-variant« *y) is usedin the present
 OBUGRIAN 339

indicative (the picture here is somewhatclouded,becausethe present-tense

morphemeis also -y-I-w-), thej-variant is usedin the imperative,the passive,
and with certain derivationalsuffixes, and the vowel-final variant is usedin
the past tenseand the conditional. Examplesfrom SosvaVogul: teey-;m 'I
eat', taaj-;m 'eat!', tee-s-;m 'I ate', wiy-;m 'I take', woj-;m 'take!', wi-s-;m
'I took'; compareTavda Vogul tii-m, tiiiij-;m, tii-s-;m and u(w)-m, iij-;m,
u-s-;m. In proto-Ostyakunstablestemshad two variants: a consonant-final
stem, which occurred with the past tense built with *-0- and with certain
derivationalsuffixes, and a vowel-final stemwhich occurredelsewhere.The
consonantof consonant-finalstems was *j if the stem beganwith a labial
consonant,*y if it beganwith a coronal(no unstablestemsbeganwith a velar
consonant),i.e. a kind of core/peripheraldissimilation was at work, e.g. pO
*me- - *mej- 'gives' but *iii- - *iiiy- 'eats'.Examplesfrom the daughter
dialects:Vach me-s-;m'I gave',mej-iiii 'give!', lii-s-;m 'I ate', liiy-iiii 'eat!',
cf. Kazym ma-s-;m,mij-aa, lee-s-;m,leew-aa.
There was also a class of thematic verbs in pOU, characterizedby the
alternationof a low vowel, *aa or *lili, with zeroin the secondsyllable;in some
thematicstems,thevowel of thefirst syllablealternated,aswell. It is difficult to
interprethistorically theevidencefor thematicverbs:therearejust underthirty
in T Vogul, and more than fifty in Ostyak, but only nine are attestedin both
languages.An exampleis pOU *kolaa-- *kaal- 'dies'>pV *kalaa-- *kaal-, pO
*kalaa-- *kool-; for detailsseeHonti (1982a:83, 104-7).
Rich systemsof vowel alternations,occurring in both inflectional para-
digms and derivationalprofiles, are attestedin easternOstyakdialectsand in
all Vogul dialects save the northern ones. These alternations may have
originatedpartly in analogyto thoseof the thematicverbs, partly as a result
of independentdevelopments(Honti 1983b).
In Vogul we must distinguishbetweentwo basic types of alternation.One
is dependenton syllable structure,with long vowels tendingto occurin open,
and short vowels tending to occur in closedsyllables,e.g. T kiiiil;;1W 'rope',
plural kiilbt. In the other type of alternation, long fust-syllable vowels
followed by high vowels in the secondsyllable (e.g. long iiii in wiiiirii '[s]he
does,will do') alternatewith their shortpendantswhenfollowed by non-high
vowels (e.g. shortii in wiireem 'I [will] do', wiiriiiint '[s]he does').
In proto-Ostyak, a system of vowel alternations arose in derivational
profiles such as *wuul;}m 'dream' - *aal;}m 'sleep',both derived from the
verb *alaa- - *001- 'sleeps'.The historical backgroundof thesealternations
has not yet beensatisfactorilyexplained,but they must in part be connected
with the fact that there was an incompleteshift of pOU long high vowels to
long low vowels in early pO, e.g. VVj tiin, Kaz tin 'price' - VVj tiiiin, Kaz
taan 'bride-price',cf. KU So Vogul tiin 'price'. In many casesthe vowel of
the root word and that of its derivateare indicationsof a particularpO vowel
alternation, and we may associateparticular derivational suffixes with
particularalternations.Thus for exampleroots with original long high vowels
340 OBUGRIAN

show long low-vowel alternants when the momentaneous/pusillanimous

suffix *='namely
;}y/~y is attached,as in Trj kiit-, Kaz kit- 'sends',TIj kiiiit;;ry-, Kaz
kati- 'sendsfor a bit', VVj kaal- 'getsup', l«e<£l;;ry- 'getsup for a little while'.
The reverse distribution occurs in verbs formed with the momentaneous
suffix *=t'namely
;}/t~, namelyroots with original long low vowels show high-vowel
alternants when this suffix is attached, e.g. VVj maas-, Kaz mtuls- 'is
necessary;loves',VVj muustS-,Kaz m(J(Jst';}- 'appeals'.
In easternOstyak dialects,an elaboratesystemof vowel alternationshas
evolved in both verb inflection and in the nominal possessiveparadigm.
Forms showing the alternationshave not completely supplantedthe older,
non-alternatingforms, with the result that thereis much superabundance and
many parallel forms are attested,in different degreesand in different lexical
items, acrossthe dialects.In VVj, for example,the first-personsingularform
of iiiimp 'dog' is eitheriiiimpiiiim or iimp;;nn. High vowels optionally replace
low onesin parallel fashionin the formation of the imperativeand of the past
tense built with -0, e.g. V liiiiw;nl;;nn 'I feed', liiiiwtiiii or liiwtiiii 'feed!',
liiiiwt.,m or liiwt;;nn 'I fed'.

Morphology
The categoriesof the nominal are number,case,and person(possession).The
verbal categoriesare number, mood, tense,genus,and person(subject and
object). In order to compressthe discussion,pertinentreferenceto syntactic
phenomenais incorporatedin the presentation of the morphology.

Number
Proto-OU distinguishedthree morphologicalnumbers:singular, plural, and
dual. The morphemeof the singularwas 0; plural and dual eachhad a range
of markers,dependingon the lexical and grammaticalcategoryof the word
in which they occurred.
A dual-marker*-n may be seenin the sequence*iin known from the forms
of the dual personalpronoun: *miin 'we two' (> pV pO *miin), *niin 'you
two' (> pV pO *niin), *8iin 'they two' (> pV *tiin, pO *iiin), comparethe
correspondingsingularpronounspV *aam, pO *maa(n) 'I' and pV *nay, pO
*ney 'you (sg)'. No correspondingplural morphemecan be isolatedin the
plural personalpronouns,pV *maan,pO *meIJ 'we', pV *naan,pO *neIJ 'you
(plur), ,pV *taan, pO *ieIJ 'they'.
Another dual marker, *y or *yW, may be seenin nouns (1) as the dual
marker in the non-possessiveparadigm, (2) as a marker of duality of
possession,and (3) in verbs, as a marker of duality of the direct object. The
category dual has vanishedfrom the morphology of T Vogul. Otherwise,
Vogul dialectshave-y-.
In Ostyakthe dual markeris formally more substantial:the -y- is followed
eitherby -n- (in case1) or -1- - -1- - -t- (in cases2 and 3), e.g. VVj weeli-y;m
 OBUGRIAN 341

'two reindeer', kaat-k~bringing'two houses',V kiiiir-;1)' l-aam AXE-dual-s 1'my two

axes', Vj kaat-k~l-aam
bringing HOUSE-dual-sl 'my two houses',V tuu-l-~yl-aam bringing
BRINGS-pres-dO-sl'I'm bringing them(2)" Vj we-l-;1)'l-aamTAKES-pres-
dO-sl 'I'm taking them (2)'.
The -y- in all of the above-citedmanifestationsdescendsfrom the proto-
Uralic dual marker. The -n- of -y;m- is identical to the dual markerfound in
the personal pronouns. The -1- of -y;}l- is perhapsrelated to the Fennic
toponym-formant=lal=la, which in Karelian dialects has developedinto a
pluralizer. In south Ostyak dialects, the number marker of the possessive
declensioncameto mark the numberof the direct object of finite verbs, as
well: seebelow.
The plural marker of the non-possessivenominal paradigm is -(;})t in
Vogul,Ostyak.
-( ;}/~)t in Ostyak.The marker of plural possessionand the marker of
plurality of the direct object of finite verbs is -aan-/-aan- in Vogul, -1- in
Ostyak. Examples: So Vogul kWol-t 'houses',kWol-aan-;JmHOUSE-plur-sl
'my houses',V Ostyakaamp-;}t 'dogs',kiiiiri-l-aam AXE-plur-sl 'my axes',
kaat-l-aam 'my houses',tuu-l-l-aam BRINGS-pres-pO-sl'I bring them', Vj
Ostyak we-l-l-aam 'I take them'. The n of the Vogul plural marker may be
related to the n initial in the plural demonstrativepronounsof Fennic, e.g.
Finnishnama 'these',cf. tama 'this'.
In certain dialect groups,particularly westernVogul and westernOstyak,
the paradigmsmarkingpossession anddirect objecthaveundergoneconsider-
able simplification.

Person
Both nouns and verbs were inflected for person in pOU. In the nominal
paradigm,personcorrelatedmainly with possessor;the verb paradigmcould
encodeinformation aboutboth subjectanddirect object.
Personwas encodedin combinationwith number, and in most casesthe
morphemesinvolved correspondclearly to the pertinentpersonalpronouns,
as is consistentwith their putatively agglutinativep(F)U prehistory.Thereare
three exceptions,however: (1) non-singularsecond-personsubjective and
passivesuffixesreflect an older form of the personalpronoun(seebelow); (2)
the suffix of the first personplural cannotbe connectedwith any pronominal
base; and (3) in some dialects,the suffix indicating third-person singular
subject and definite singular direct object is not of pronominal origin (see
p. 342-3). In proto-Vogul, the marker for third-personplural possessorand
plural direct objectwas *-aan~l-/-aan;")l-,
third-persona form which arosethroughanalogy
(Honti 1983d).
The OU second-person pronounsdeviatefrom the rest of Uralic in having
not initial *t, but rather initial *n, e.g. the s2 pronounis T Vogul nuw, naw,
So Vogul naI], VVj OstyaknOI], Kaz OstyaknaI], contrastHungarianor Komi
te, Mordva or Udmurt ton, Finnish sina « *tina), Selkup tan. Hajdu (1985;
1987: 241) has suggestedthat the OU pronouns are the result of an
342 OBUGRIAN

Table11.6 pOU personalpronounsandpersonsuffixes

Personalpronoun Verb suffixes Nominal (possessive)

suffixes
Subject Object

sl *aam, *maa *-(V)m *-(V)m

s2 *nay, *ney *-(V)n *-(V)n
s3 *80y *-0 *-(V)8, *-8(V) *-(V)8, *-8(V)
dl *miin *-miin *-miin
d2 *niin *-niin, *-t;m *-niin, *-tan
d3 *8iin *-y *-8iin *-8iin
pI *mafu;), *maan *_(V)yW *_(V)yw
p2 *naal), *naan *-(V)n(Vn)' *-(V)n(Vn)
p3 *8VVI), *8VVn *-(V)t *-(V)8(Vn) *-(V)8(Vn)

assimilation(*tVn > *nVn) that pre-datespOU; evidenceof t-initial pronouns

may be found scatteredin the morphology,however,such as the d2 suffixes
which continuepOU *-t;}n/-t;)n, as in Vj Ostyakooyi-t;m 'the daughterof you
two', Vj Ostyakwe-l-;Jt;m TAKES-pres-d2'you two take', andthe p2 suffixes
which continue pOU *-t;}y/-t;)y, as in Vj Ostyak we-l-t;Jy 'you (plur) take'
(Hajdu 1966: 132-3,Honti 1984: 38).
Table 11.6 sets out reconstructionsfor the personalpronounsand person
suffixes of pOU; to conservespace,only front-vocalic pendantsaregiven.
This systemsurvivedlargely into pV, with modificationsasoutlinedabove;
the Vogul dialects show varying degreesof person syncretism,especially
s2-d2-p2andd2-d3-p2.Seealso the sectionon objectmarking, p. 347 ff.
In the proto-Ostyaksystem, a geminate sequence*1:-1: arose in certain
forms of the possessiveand objectiveparadigms;this was simplified to *-1:-,

Table11.7 p Vpersonalpronounsand personsuffixes

Personalpronoun Verb suffixes Nominal (possessive)

suffixes
Subject Object

sl *am *-(V)m *-(V)m

s2 *nay *-(V)n *-(V)n
s3 *tay *-0 *-ii, *-tii *-ii, *-tii
dl *miin *-miin *-miin
d2 *niin *-niin *-niin
d3 *tiin *-ay *-iin, *-tiin *-iin, *-tiin
pI *maan *_(a)yW *_(a)yW
p2 *naan *-(a)n(an) *-(a)n(an)
p3 *tVVn *-(a)t *-VVnaJ *-VVnaJ
 OBUGRIAN 343

Figure11.2 Proto-Ostyakpersonalpronounsand personsuffixes

dialects conjugation. two

conjugation. conjugation.
dialects Ostyak two two alongsidealongside
sl *miiii(n) *-(V)m *-iim *-iiiim *-(ii)m
s2 *-(V)n *-iin * -aan
.... *-(ii)n
*nolJ
s3 *iOy *-8 *-(il)V-Hi

dl *miin *-miln I *-imiln *-iiiimiln *-(ii)miln

d2 *niin *-tiln *-iiiiniln *-tiln
d3 *iiin *-Yiln I *-bn

pI *melJ *_yW *_iyw *_iiiiyw *-iiy W

p2 *nelJ *-tilY *-(il)tiln *-iiiin * *-(il)tiln
p3 *by *-(il)t *-iii *-iiiii *-iii

Note: subj cj = subjective(= indefinite) conjugation.

e.g. pO *ooyi-pl:-i > *ooyi-y;:)i > V Ostyakooyiy;:)Z 'hislher two daughters'.

The Ostyak dialects tend to syncretizethe same categoriesas Vogul (see
Figure 11.2 above).

Cases

Noun Declension
Six nouncasescanbe reconstructedfor pOD (cf. Riese1992: 387): alongside
the nominativein zero (-0) therewere

Accusative *-m(V) Ablative *-(V)l

Locative I *-naa/-naa LativeI *-(V)y
Locative II *-taa/-taa Lative II *-(V)j

Lative I probably already also functioned as a translative('turning into/

treatedas X'). Besidethesecasesuffixes therewere also threepostpositional
forms, lative *naaj, comitative *naat, and ablative *naru, built from a
pronominal base.Thesehad begun to agglutinateon to their head noun in
pOD; in present-dayOstyak may be found traces of their use both as
postpositionsandas suffixes.
The number of casesin the noun paradigmsof the various Vogul and
Ostyakdialectsvariesgreatly: while T Vogul had sevenand SosvaVogul has
six, the easternmostdialects of Ostyak have ten or eleven, while northern
Ostyakdialectssuchas Kazym haveonly three.
344 OBUGRIAN

Table11.8 Casesuffixes

Ostyak Ostyak Vogul Vogul

Kaz Vach/Vasjugan Tavda Sosva

Nominative -0 -0 -0 -0
Accusative -m, -mee/-mee
Dative/lative -aa -lilil-aa -n, -nliii/-naa -n
Locative -nii
-~n,-lilil-aa -o(o)y/-o(
-n~/-n;) -t, -tlilil-taa -t
Ablative -o(o)y/-o( o)y -nlilil/-naal o)y
-n~l

Comitative -nlilitl -naat -nlilitl-naat

Instrumental -(t)~/ naat
-( t);} -(t)~l t);} -(t)~lo)y
Translative t)~/naat
-(-~y/-;}y -( t);} (~)wt);} t);}
-~y

Approximative -(lili)plilil-(aa)paa
Abessive nilJi(
-l~y/-l;}y nilJi(
Comparative -nllJI( t)/-nilJi( t)
Distributive -nllJI(
-t~ltlilil-t;}ltaa t)/-nilJi(

The casesuffixes of our four sampleObUgrian languagesare set out in

Table 11.8 for comparison.
The zero nominativeis a continuationof the proto-Uralic stateof affairs.
The m-accusativeof southernVogul is also of Uralic origin; it is attestedin
easternVogul dialectsas well, but thereis no traceof it in Ostyak.
The functions of the three primary spatial cases- dativellative, locative,
and ablative- are performed,in Ostyakand Vogul, by morphemeswhich are
not historically connected.Thus the Ostyak dativellative suffixes are not
connectedwith thoseof Vogul. The Vogul suffixes developedeither from a
Ugric postposition*naa (Liimola 1963: 91-5, 120-3) or from a pronominal
basewhich functionedas a postposition(Riese 1992: 381); the origin of the
Ostyak*-aaJ-aalative is obscure.
The Ostyak and Vogul locative suffixes are also unrelated.The t of the
Vogul suffixes date at leastfrom Ugric times; cognateelementsare attested
in fossilizedadverbsin Ostyak.The n of the Ostyaklocativesis inheritedfrom
proto-Uralic. The vocalic componentsof both the Vogul and the Ostyak
locativesare obscure.
Finally, there is the ablative. The function of this case is carried out
in Vogul by a suffix which derives from the same postposition seen in
the dativellative (*naa), but with an ablative suffix I which dates back to
at least Ugric times (Liimola 1963: 66). In western dialects, the ablative
has taken on the function of indicating the (partial) direct object (Liimola
1963: 43-4). In Ostyak, the ablative is expressedby means of a range
of morphemes(suffixes in the eastand south, e.g. V -ooy/-ooy,DN shape -eew~,
postpositionsin the Surgut group and in the west, e.g. Pim shape iiw~i, Kaz
shape which probably all go back to a spatial noun of the shape *Vyw,
eew~i)
 OBUGRIAN 345

with or without casesuffixes (Redei 1977: 208-9, Honti 1984: 61).

Note that the less central cases,comitative, instrumental,and translative,
continuecommonpOU ancestors.The comitative,known as a suffix in Vogul
only in the south, and in Ostyakonly in the east,derivesfrom the base*nlHi
plus the t-Iocativementionedon p. 344. It alsosurvivesin onewesternOstyak
dialect, Sherkal,as a postpositionoccurringwith personsuffixes,e.g. 8ermaa
naataaIJeem'with me', naIJ naataaIJeen'with you (sg)', tuw naataaIJeet'with
himlher' (DEWOS: 1023).
In the Vogul instrumental suffix -(t)Vl, the I element derives from an
ablative suffix dating from the Ugric period. The t of the longer variant, -t;}l,
is identical with the t-Iocative mentionedabove;this is also the origin of the
t in the Ostyaksuffix (Liimola 1963: 100--101,117-18).
The translativesuffix derivesfrom a lative *-y in both ObUgrian languages
(Liimola 1963: 129-30); this *y is also attestedin adverbswith its original
lative function (Liimola 1963: 176).
At least the easternmostdialects of Ostyak innovatedthe approximative
case,which like the dativellativeindicateslative motion, but in a moregeneral
direction. The shorterform cited above (-piiiil-paa) is presumablyidentical
with a homophonousemphaticparticle; the longervariant(-iiiipiiiil-aapaa) is
a compoundof this particle with the dativellativesuffix. We may assumethat
the particlewas originally usedwith lative deictics,e.g. nuu-y-paa'upwards',
then spreadto other contexts,e.g. kaat-aa-paa 'towards the house' (Honti
1984: 62).
The comparativeis anothereastern Ostyak innovation.It derivesfrom the
postposition*nii1J;)t, with (spatial) ablative meaning- cf. Trj miiii niiIJteem
'from me', Vj tii=m-iiiil niiIJ;}t seem=l;,y IS.BORN=past.part-s3FROM!
SINCE EYE=LESS '(s)hehas beenblind sincebirth' (DEWOS: 1006; Honti
1982b: 111-12; 1984: 65, 83).
The distributive, also limited to Ostyak, is clearly a compound,but its
prehistoryis unclear.The Ostyakabessiveis alsoof obscureorigin.

DeclensionofPersonalPronouns
The casesystemsofVogul and Ostyakpersonalpronounsarefor the mostpart
identical with thoseof their nouns,and there is no reasonto suspectthat the
situationwas any different in pOU.
Apart from minor differences(suchas the lack of translative,and of course
of distributive,forms in Ostyak)the main differencebetweenthe nominaland
pronominal paradigms in both ObUgrian languagesmay be seen in the
formation of the accusative.In Vogul the relevantpersonalsuffix is attached
to an oblique variantof the pronounbase,e.g. K niina-n 'you two (acc)'.The
basemay occuralone,e.g. T iiniiii-m 'me', or reinforced(T iiniiii-m-mii 'id.');
metathesizedforms are also attested(T iimiiiin, iimiiiinmii 'id. '). Examples
from So Vogul: aan;}-m 'me', naIJ;}-n 'you (sA)', meen-men'us two'. In the
northernmostdialectsof Ostyak,the accusativeof personalpronounsis built
346 OBUGRIAN

in the sameway, e.g. Kaz man-eem'me'; elsewherein Ostyak, the suffix -t

is used, e.g. V miiiin-t 'id.'. The divergencesevidencedby the pronominal
paradigms of the various Ostyak dialects are so great as to prevent the
reconstructionof the pO system;the pOU systemis thus completely outof
reach.

Verb Inflection

Tense
The ObUgrian languageshave two non-compoundtenses,past and present.
In pOU, the morphemeof the presenttensewas *-y-, zero, or possiblyboth;
both p(F)U past-tensemarkers,*-j- and *-sL (> pOU *-s-) were used.
Vogul and Ostyak form their respective present tenses using similar
strategies,but with different materials.In mostVogul dialectsthe mostwidely
usedpresent-tense markerconsistsof a vowel plus a -y- which, in westernand
southerndialects,undergoessandhi(consonantgemination,vowel lengthen-
ing) with precedingand/orfollowing segments;this y is perhapsa continua-
tion of the pFU present-tense morpheme*-k-. Another present-tense marker,
-i-, attestedin e.g. the Sosvas3 form ool-i-tJ LIVES-pres-s3 '(s)he/it lives',
may be a reflex of the pFU deverbalnominalizer*=jV (e.g. Finnishopetta=ja
'teacher', from opetta- 'teaches').In some dialects, many forms of the
subjective conjugation contain both present-tensemarkers combined syn-
conjugation'I go' < *min-iiy-;)m < *min-
chronically into one, e.g. So min-e(e?)y-~
;)j-y-;)m, while forms of the objective conjugationcontain -i- alone, e.g. So
tot-i-l-bring
~ 'I bring it' . In the passiveparadigm,a variantof the y-Iesspresent-
tensemarkeris used,e.g. So tot-a-w-em'I am brought; (it) is broughtto me'.
In southernVogul at the beginning of the twentieth century, thesepresent-
tensemarkerswere usedto refer to future and momentaneous events;ongoing
events were indicated by means of different, historically derivational,
suffixes, notably -iiiint-/-aant- and -(iiii/aa)l-, e.g. T miniiim 'I shall go',
miniiiiinteem 'I am going (now)'.
A somewhatsimilar adaptationwas made in Ostyak, where the present-
tensemarkersof the present-daydialects(Surgut and Kaz -1-, Sal, Southern,
Ni, and Ser -t-, VVj Ber 0 -1- ) all descendfrom a frequentativederivational
suffix *=1. In the easternmost dialectsthis I-markerwas preceded,in the third
personsingular,by an element*-piUiI-paa, yielding a suffix sequence-w-;;,l >
-w;;,l, e.g. Vj we-w;;,l-tJ '(s)hetakes';the sequencew;;J(l) then spreadto the s2
and p3, e.g. menw;m'you (sg) go', menw;;,lt 'they go'; contrastmen-l-~ spread 'I
go'. This *-paiil-paa was originally a deverbal nominalizer; a parallel
developmentmay be seenin the Finnishthird-personpresent-tense forms, viz.
mene-e'([s]he) goes'< *mene=pa'goer',mene-viit'(they) go' < *mene=pa-t
'goers'.
The pU past-tensemorpheme *si (> -s-) survives in both ObUgrian
languages,e.g. So Vogul min-;;JS-tJ, VVj Ostyak men-;;,s-tJ '(s)he went'. The
 OBUGRIAN 347

other inherited past-tensemorphemeis pFU *-j-, attestedonly in Ostyak,

where it melted into the precedingstem-final vowel. The result has been
interpretedsynchronicallyas a linking vowel, and the descriptivegrammars
thereforerefer to a zero past-tense(or 'perfect') morpheme,e.g. VVj men-
({J-;»n 'I went' opposedto men-s-;»n'I was going' and men-l-;»n 'I am going'.
Only the easterndialectsuseboth pasttenses;to the west,either-({J- (southern
dialects and Nizyam) or -s- (Ser Kaz Ber 0) is used. In Vach and Vasjugan
Ostyaktwo additional tensesevolved,the so-called'historic perfect' (-yiiiis-
I-yaas-) and the 'historic imperfect' (-yiiiil-I-yaal-), e.g. men-yiiiis-;»n,men-
yiiiil-;»n 'I went'; the function of thesetenseshasnot beenclarified.

Mood
We can be certain that Proto-ObUgrianhad the three following moods:
indicative (*-0-), imperative (*-0-), and conditional (*-n Vy-). Although the
morphemesfor indicative and imperativewere both zero, the two moodsdid
not syncretize,becausethe indicative occurredonly with tensemarkers,while
the presenceof the imperativezero precludedtense-marking.
The indicative morphemeis still zero in both ObUgrian languages.The
imperative,on the otherhand,hasevolvedalongdistinct paths.In Vogul, zero
still marks the imperative,i.e. the personsuffixes are attacheddirectly to the
stem;in this Vogul contrastswith all otherUralic languages,all of which have
either a distinct imperativemorpheme,distinct personsuffixes characteristic
of the imperative paradigm, or both. Number syncretismcharacterizesthe
northern Vogul imperative, e.g. min-:m servesas singular, dual, and plural
'go!'
In Ostyak, the imperative morphemeis reflexes of *-iUi/aa- in the s2
subjective,*-ii-/-iI- elsewhere,e.g. V tuuy-aa-({J 'bring (some[thing])!',tuuy-
n~t~ 'you two bring (some[thing])!; you (dual or plur) bring (somethingor
someonedefinite)" tuuy-n~({J'bring (somethingor someonedefinite)!', tuuy-
n~yl-aa 'you (sg) bring them(dual)!'.
The pU conditional/optativemorphemewas preservedin Vogul. We may
reconstructpV *_n;'}yW_, the reflexesof which (T -nee-I-nee-,SoVogul.
-n~-) occur
in complementarydistribution with tensemorphemes,e.g T minl-nee-m,So
min-n~-;»n 'I would go'. In Ostyak the categoriesof conditional and
optative are expressedby meansof various participles,often in conjunction
with particles.

Marking of the Direct Object

As in many otherUralic languages,the pOU verb encodedinformation, in its
morphology,aboutthe direct object. This informationincludedindicationsof
definitenessand number(as in Nganasanor Nenets),but not of person(as in
Mordva, or, in part, Hungarian). Forms which indicate a 'definite' direct
object are usually called the objective,or definite, conjugation.The material
makeupof the morphemeswhich indicatedefinitenessvaries from language
348 OBUGRIAN

to language,except in the case of the third-personsingular subject forms,

wherethe morphemesignalling a definite object derivesfrom a third-person
pronoun (in easternOstyak dialects, this morphemestems from a demon-
strative pronoun).The marking of definitenesselsewherein the Ostyak and
Vogul paradigms,and the marking of object numberthroughout,is so highly
heterogeneousthat the reconstructionof the pOU objective conjugationis
wellnigh impossible.
In non-southernVogul dialects,definitenessof the direct objectis indicated
by -1- if the subjectis first or secondperson,but by -t- if the subjectis third
person;secondarydevelopmentshave alteredthis picture in southernVogul,
where-t- cameto be usedin past-tenseforms, and-1- was usedelsewhere.The
-t- or -1- element is followed by morphemeswhich indicate whether the
(definite) object is singular(sO), dual (dO), or plural (pO); thesemorphemes
are historically identical with thosewhich indicatethe numberof possessions
(cf. above). In both instancesthe -[- is historically identical with the third-
personmarker (seeabove).The vocalic elementwhich follows this -[- is of
unknownorigin andfunction (Liimola 1963: 222). To the right of consonants,
only this vocalic elementnormally survives: compare-te in Sosvatot-i-te
'(s)hebrings (sO)' with -e in tot-i-an-e '(s)hebrings (pO)'. Furtherexamples:
Sosvatot-i-an-:»n 'I bring (pO)" tot-s-an-e '(s)hebrought (pO)" tot-n;;1W-te
'(s)he would bring (sO)" tot-n;;1W-ay-e '(s)hewould bring (dO)'; Tavda tat-
ee-l '(s)hebrings (sO)" tat-ee-l-aan-:»n'I bring (pO)" tataa-s-t-aan-ee'(s)he
brought(pO)" tat-ee-l-aan-:»n
tat-n-~~l '(s)hewould bring (sO)'.
In proto-Ostyak, as in non-southernVogul, the primary split in the
morphology of the objective conjugation was along the feature [+/-] axis-
of-discourse.In forms with a third-personsubject,definitenessof the direct
object was marked by an ending which derived from the third-person
pronoun, while in forms with first-and second-personsubjects,the subject
personsuffix was precededby a full vowel, *iIlii if the (definite) object was
singular, *aaliili if dual or plural. Much analogicallevelling has occurredin
the modemdialects;for details,seeHonti (1976; 1983c;1984: 38-46).Unlike
Vogul, the Ostyakobjectiveconjugationforms for axis-of-discoursesubjects
have the same morphemic schema as those of the correspondingnon-
objectiveforms, viz. STEM-TENSEIMOOD-PERSON.The differencelies in
the vowel to the left of the personsuffix: in the non-objectiveforms this vowel
goesbackto pO *S/;}, in the objectiveforms to *iIlii - *aaliiii. Examples:VVj
tuu-l-~ bring 'I bring (something,someone)',tuu-l-iTm 'I bring (sO)"
Examples:
tuu-l-~l­
aam 'I bring (dO)" tuu-l-l-aam 'I bring (pO)' .
In southernOstyakdialects, the position of the dual and plural (definite)
object markers moved from the left of the subject person-markersuffix
(STEM-TENSEIMOOD-dO/pO-PERSON)to its right (STEM-TENSE/
MOOD-PERSON-dO/pO).The identity of these number-markingsuffixes
changed,as well: the new schemeusedthe samenumbersuffixes as thoseof
the non-possessivenominal paradigm, e.g. older Kr see1Jk-fj-eey;]t-aam
 OBUGRIAN 349

STRIKES-past-dO-slwas gradually being replaced by see.(Jk-fJ-eem-y;;m

STRIKES-past-sl-dO,both 'I struck (past) (dO)'; the correspondingforms
with plural object were see.(Jk-fJ-~t-aam probablySTRIKES-past-pO-sland see.(Jk-
probablySTRIKES-past-sl-pO.This shift probably occurredin stages;for a
fJ-eem-~t
reconstruction,seeHonti 1983c.
Genus Verbi
Both Vogul and Ostyakknow the distinctionbetweenactive and passive.The
active is unmarked;if it is to be assigneda morpheme,that morphemeis zero
(-fJ-). The passiveis markeddifferently in the two ObUgrian languages.In
Vogul the morphemeis -W-, which goesback to a proto-Uralic derivational
suffix (*=w) which built reflexives. In Ostyakthe passiveis markedby -Vj-,
in which the j-elementis thought to descendfrom anotherpU reflexivizer
(*=j). The useof the passiveis widespreadin both Vogul and Ostyak,andwe
may thereforeassumethat the categorywas presentin pOU, eventhoughits
formal expressionin the daughterlanguagesis effected by morphemesof
different origin.
In VVj and SurgutOstyak, the vowel of the passivemarker is invariably
[+ BACK] in the first and secondpersons,regardlessof the frontness or
backnessof the stemto which it is attached;in the third person,on the other
hand, this vowel reflects the harmonicpO pair *aaj/aaj, e.g. V tuu-l-uuj-fJrn
'I am brought; to me is brought',tuu-l-iT « *tuu-l-aaj) '(s)he is brought; to
himlher is brought',Vj we-l-uuj-fJrn 'I am taken;from me is taken',we-l-ii «
*we-l-aaj) '(s)heis taken;from himlheris taken'.

Derivational Morphology
Vogul and Ostyak have rich derivational morphologies,and many of the
individual derivational suffixes go back to sharedpOU morphemes,often
with p(F)U backgrounds(Ganschow1965;Sauer1967).Herewe shall restrict
our attentionto the morphemeswhich form verbalnouns.
The forms of the infinitive in the variousVogul dialectsare quite diverse,
but certainly must go back to somethinglike pV *-IJwkwe. The full form of
this morpheme persisted, for the most part, in northern dialects, while
elsewherethere was loss of at least the vowel, and often of one or the other
of the consonants,e.g. Sosvatee-.(Jkwe, Pelymkatee-x, LM tee-xw, Tavdatii-.(J,
all 'to eat'. The origin of the Vogul infinitive suffix is unclear. Gombocz
(1898) thoughtit to be a combinationof the presentparticiple (=n-) plus the
translativeending-y; while this is plausiblefunctionally (cf. the formation of
the Ostyakinfinitive, below), it doesnot explain the labiality of the infinitive
morpheme. It is in any case likely that the translative morpheme was
somehowinvolved: variantsof this suffix similar to that of the infinitive may
be found in adverbs,as well (Liimola 1963: 193-4).
In proto-Ostyakthe infinitive morphemewas *-taal-taa; in the Surgut
dialects,this basewas extendedwith the translative,yielding e.g. Trj -taayat
350 OBUGRIAN

probably
-tii iy~ (cf. the Finnish infinitive from *-ta-kl-ta-k). The pO infinitive
morphemewas probably one and the sameas that of the presentparticiple;
the forms subsequentlydiverged in most dialects, yielding distinct verbal
nominals.Thus in Vach Ostyakwe have the infinitive lii-tiiii 'to eat' distinct
from the presentparticiple lii-t;J 'eating',but in Kaz Ostyakboth forms are
iee-ti; cf. Sauer(1967: 96, 102).
In the Vogul dialectsthe morphemeof the presentparticiple derivesfrom
a different suffix, consistingof n plus a vowel, and related,inter alia, to the
infinitive suffixes of Hungarianand Permian,e.g. T tii-nii, So tee-ni, both
'eating'.
Both Vogul and Ostyak form their past participle with reflexes of pOU
*=Vm, a suffix of proto-Uralic origin, e.g. T Vogul tasaam,So Vogul tiMsam
'dry' = Trj Ostyaksasfnn,Kaz Ostyaksos;m'hard,tough',the pastparticiples
of pV *tasaa-= pO *sasaa-'drieslbecomesdry'. In addition, both languages
have a gerund which is clearly related to this *=Vm, cf. T Vogul tateel;Jm
'(whilelby) bringing', So Vogul nomim '(whilelby) thinking'. The Ostyak
pendant,which had a high/low vowel alternation in pO (*=maan/maan-
*=miInI=miin), is clearly compound,andperhapsdatesfrom the Ugric period
(Kispal 1968: 272). Examples:Vj Ostyakpoot=miTn- '(whilelby) running',
Kaz iaaw;J/=maan'(whilelby) leading'.

Syntax
Both ObUgrian languageshave preserved,by and large, the dominant pU
constituent order SOY, e.g. Sosva Vogul taw aan-;Jm waay-te s3.PRO
s1.PRO.accSEES.pres-s3s0'(s)he sees me', Vasjugan Ostyak nii jel,,",
potjiiI]kaa .!ddt=k$l-wf>l WOMAN NEW SHOE BUYS=refl-s3pres 'the
woman buys (a pair of) new shoes(for herself)'. Also of Uralic age is the
tendencyto put nouns in the singular when they are modified by numerals
over 'two', e.g. Vach Ostyakliiiiw;Jt aal SEVEN YEAR 'sevenyears',Sosva
Vogul saateet SEVEN NIGHT 'sevennights'. The dual is usually usedwith
'two' e.g. Pim Ostyak kiiiit kOltr-Y;Jfl TWO MALE. REINDEER-dual 'two
reindeer',So Vogul kit aayi-iy 'two girls'.
Both Vogul and Ostyak use the mihi est (it is to me) possessive
construction, e.g. Vj Ostyak men-n;J wal-w.,l maa~njaaljrr
innovated s1.PRO.loc
IS/EXISTS-s3presPURSE 'I have a purse',but eachhas also innovatedits
own verb meaning'has',e.g. SosvaVogul tas at oo~ ~ -eeY-;JmCONTAINER
NEG HAS-pres-sl 'I don't haveanythingto put it in', TromaganOstyakiimii
jek-k;Jfl iiiiI]kee-y;Jfl taj-;JU:J WOMAN FATHER-dualMOTHER-dual HAS-
pres-s3'the womanhasa father and mother'.
In both Vogul and Ostyak, constructionswith verbal nouns frequently
correspondto subordinateclausesin other languages.The verbal nouns
usually take person suffixes and either case suffixes or postpositions,e.g.
Pim Vogul kiiiim+min=n-iiiit wuj;J1 si'm-t;J neek;JS-w-;Js-fJOUTWARDS+
 OBUGRIAN 351

GOES=pres.part-s3 WHEN HEART-Ioc PECKS-pass-past-s3 'when (s)he went

out, (s)he felt a pang in hislher heart', Vj Ostyak Ostyak
kuuj-~Z liiiiyt=;»n piTr-n;,
nii mey+uuj leeeek-ooy iiZ;m+joyil-w;,Z HUSBAND-s3 EXITS=past.part
AFTER-Ioc WOMAN EARTH+SPACE.UNDER PATH-abl FORWARD
+GOES-s3pres'after her husbandhad gone out, the woman went along the
undergroundpassage' .
The widespreaduseof the passivein both Vogul and Ostyaksuggeststhat
it is an ObUgrian innovation. Ostyak has innovatedergative constructions,
commonespeciallyin the easterndialects;see(Kulonen 1989: 297-302).

Lexicon
The ObUgrian languagesinheriteda considerablenumberof words andstems
from earlier phases of their Uralic past.The UEW reportsjust over 700 Vogul,
and just under 800 Ostyak words of p(F)U or proto-Ugric origin. Most of
thesewords belong to the core of the lexicon, and are thus liable to wide
derivation, so that a substantialpercentageof the Ostyakand Vogullexicons
are madeup of non-borrowedvocabulary.
We have no reliable dataconcerningthe period after the breakupof Ugric
unity. In all probability the speakersof pOU had intensivecontactswith the
aborginal populationsof western Siberia, but of the genetic and linguistic
affiliations of these people we know next to nothing. It is natural, but
intellectually sterile, to supposethat someof the unetymologizedOstyakand
Vogul vocabularystemsfrom suchcontacts.
After the breakupof pOU, speakersof pV and pO took vocabularyfrom
speakersof Iranian languagesin roughly the fourth to sixth centuries CE
(Korenchy 1972), e.g. Vogul T iis;,r;»n, KU KO Pim ees;,r;»n,KM ees;,rm;"
VS VN LU jees;,r;»n, LO So eesarma'(feeling of) shame'< Middle Iranian
*afsarm, cf. Avestanfiar;»na-, Middle Persiansarm; Ostyak Kr kiiiirt, Ni
kaart;" Kaz kaarti, 0 kurti 'iron' < Middle Iranian *kart (? *k1irt), cf. Avestan
kar;")ta- 'knife', ModemPersiankaard; OstyakV Zooy;,r, Vjjooy;,r, Trj iayfJr,
DN Kr Ni tax;,r, Kaz iax;,r, 0 Zax;,r 'armour, mail (shirt)' < Middle Iranian
*zyar- 'armour',cf. Ossetezyar, zyiir, Afghan zyarah 'chain-mailarmour'.
About thirty loanwords into Ostyak attest to contactswith speakersof
variouskinds of Tungusbeforethe twelfth centuryCE (Futaky 1975),e.g. VVj
jeee(1),Trj J jee1)w, DN Ko joo1), Ni Ser KazSynjaa1),0 joo1) 'ten' < Tungus
*Joan - *Joon.
Most loans taken into Vogul and Ostyak during the past millennium are
borrowingsfrom Komi, Tatar, Samoyedic(chiefly Nenets,to a lesserdegree
Selkup), and Russian.From Komi comes,for example,the word for 'bread':
Vogul KU KM KO nieeeeni,VS VN niiini, Pim niinj, LU njenj , LO So njaanj
and Ostyak VVj Trj J DN niiiiin j , Ni Ser Syn njaani, 0 niiiiini are all from
Komi niani (which, together with its homophonousUdmurt cognate,was
originally a loan from Iranian).Also from Komi (and also ultimately Iranian)
352 OBUGRIAN

is the word for 'paper;writing': Vogul KU KM KO VS VN Pim LU neep;]/(,

LO So neepakand OstyakVVj Trj niipiik, J niipeek,DN Ko Kr neepaak,Ni
5er Kaz Syn 0 neepeek are all from Komi njebeg - nebeg 'book, paper;
birchbarkwith propertymarks'.
Example of loanwordsfrom Tatar: T Vogul aaci;,w, KU VN joonn;;),
oo~;ry, DN
Ostyak aatj ;;) 'sour', cf. Tatar acy 'sour'; T Vogul jannaa, KU KO joonn;;),
KM joor;m, Pim LU jor;m, DN Ostyakjaannaa, Ko jiiiinniiii 'grits', cf.
Tatar janna '(barley) grits'. From Nenets: LO Vogul xoopt, So xaapt, 5er
Ostyak xaapt;;), Kaz xaapti, 0 xaapti 'castratedreindeer',cf. NenetsxabtO
'id.', LO Vogul poorxa, So poorxa, Ni Kaz Ostyakpaarxa 'parka' < Nenets
parka 'id.'. From Russian:T Vogul pop, KM KO VN Pim pup, LO So puup,
V Ostyakpaap, paaw, Trj poop, Ko Kr Ni puup, Kaz P(J(JP, 0 poop 'priest'
from Russiannon, T Vogul riimkiiii, KU KM KO LU riimk;;), P rumb, So
rumka, DN Ostyak ruumkaa,DT riiiimkaa, Kr Ni 0 ruumka 'whisky glass'
< RussianplOMKa.
Loan relations have long existed betweennorthern Vogul and northern
Ostyakon the one hand, and (to a lesserextent) betweeneasternVogul and
southernOstyak on the other, for example LO So Vogul aap~i 'younger
brother' is a loan from Ostyak, cf. Kaz Syn Ostyak aap~i, i i p~i
0 'id.''id.' as
is northern Vogul kaat 'idol', cf. Ni 5er Ostyak kaat, Kaz kaalt, 0 kiiiilt
'protectivespirit'. Loans in the other direction include Kaz Ostyakxasap,0
xasap 'midge-prooftent', from northernVogul, cf. So xasap,and 5er Ostyak
aamtjikee, Kaz aamaam~i 'riddle' < LO Vogul aam~i,cf. So Ostyak
aam;;)~
(ultimately from Tatariimiilci).

Appendix
In order to supplementthe picture given in the main body of this chapter,in
the two sectionsbelow I list briefly the mostimportantfeaturessharedby (1)
the ObUgrianand (2) the Ugric languages.The presentationis basedchiefly
on Honti 1979.

ObUgrian Shared Features

1 The namesof the teens(11-19) in northern,western,and easternVogul
and in westernOstyakwere basedon the model 'X (and) lying ten', e.g.
So Vogul kit+xujp+luw, Kaz Ostyakkat+xo~+jalJ kat+xo~+jalJ both TWO+LYING-
+TEN '12' (Honti 1986a:199).
2 Widespreaduseof the passive.
3 In passivesentences,marking of the agentwith a morphemecontaining
n. The suffixes which perform this function in the modem ObUgrian
languagesareof different origin: in Vogul, it is the lative -n(aa/iiii), which
derivesfrom a Ugric postposition*naa-, while in Ostyakit is the locative
« pO *-naal-naa,from a pU locative). But the occupationof the Vogul
'lative' slot by -n(aa/iiii) is secondary.What happenedis this: in pOU, the
locativein *-n V servedto mark agentsin passivesentences.Qua locative,
 OBUGRIAN 353

this suffix was lost in proto-Vogul, where it was replacedby the newer,
probably Ugric, locative *-ttv. The potentialhomophonicclash with the
p V lative may have helpedto oust the old locative; in any event, the -n
which markspassiveagentsin Vogul is now synchronicallya lative.
4 In both ObUgrian languagesthere is a copula-like verb which is
phonologicallysimilar, but unrelated,to the verb 'is': T Vogul aas-, So
lMs-, VVj Ostyakwas-, Kaz W{J{JS- 'is someoneor something'(DEWOS:
1630-1)"# T Vogul aal-, So i'u'll-, VVj wal-, Kaz w{J{JI- 'is, exists,lives'
(DEWOS: 1577-80).
5 In both OstyakandVogul, the instrumentalmarksthe objectin sentences
with benefactiveforce; the benefitedparticipantis put in the accusative,
if there is one available. Thus alongsideTrj Ostyak liiiij;;mz miiiinteem
mej-eeAXE sl.PRO.datGIVES-imp.s2s0'give me the axe!', there is
also the benefactiveliiiij;;mz-iiiit miiiin-t mej-iiii AXE-ins.fin sl.PRO.acc
GIVES-imp.s2,roughly 'endowme with the axe!'

Ugric SharedFeatures
1 In proto-Ugric, the sibilant (*s) and shibilant (*8) inherited from p(F)U
fell togetherand changedinto *8; the resultinggap beganto be re-filled,
in proto-Ugric,by the depalatalization*sj > *s.
2 Already in proto-Ugric, *k had markedly different allophonesin front-
vocalic versusback-vocalicenvironments.The result was the develop-
ment, albeit separatelyin eachof the threeUgric languages,of *k > x (>
h- in Hungarian) in back-vocalic words. In Vogul, this change was
restrictedto the northern and Lower Konda dialects; in Ostyak, it was
restrictedto the western(= southernplus northern)dialects.
3 Non-initial p(F)U *w changedto *y in proto-Ugric.This *y subsequently
reverted to W (> v in Hungarian) in certain dialects and in certain
environments. For details, see Honti 1985b. Examples: pFU *Hilwa
'winter' (the sourceof Finnish talvi) gave proto-Ostyak *tel~y> VVj
Ostyaktel::ry, pU *luwe 'bone'(the sourceof Finnishluu) gavepOU *luy
> pV *luw (> So luw), pO *loy (> VVj loy), pFU *kiwe 'stone' (the
sourceof Finnish kivi) gave pOU *k66y > p V *kaw (> So kaw), pO
*k66y (> VVj kooy), pFU *kewbe (or perhapsbetter: *kewie) 'rope' (the
source of Finnish koysi) gave (with metathesis)pOU *k66l~y > pV
*kwaal~y (> So kWaaliy), pO *k66l~y(> Ni keet;}, Kaz keel).
4 The sequence-type p(F)U lateral plus *m (i.e. *lm, *im, *Vm; in more
traditional formulation *lm *bm *Dim) yields parallelresultsin the three
Ugric languages.
5 In certain words, the Ugric languageshave preservedevidenceof an
original high back unroundedvowel (*1), as in the word for 'arrow':
Hungariannyfl (with i < *1), pV *nj'm, pO *niaal (with aa < *i'1); contrast
Finnishnuoli.
6 In the word for '3', Hungarianand Vogul have changedinternal *-1- to
354 OBUGRIAN

-r- (Hungarianhdrom, So Vogul xuur;;mz). ContrastFinnish kolme- and

VVj OstyakkadlfJrn.
7 The gerundivesuffixes of Hungarian(=vdn/=ven) and Ostyak(*=maan/
=maan- *=rriiYnI=miin) point to U gric origin.
8 A new set of local suffixes, basedon the pronominalbase*na-, evolved
in proto-Ugric.
9 It is virtually certainthat the ablativesuffix -l is not older thanUgric.
10 Proto-Ugricinnovateda derivationalsuffix which formed momentaneous
verbs, *=pp(V)-. Cf. Hungariandll- 'stands',(fifteenth-centuryHungar-
ian) dll=ap- 'stops',So Vogul sialt- 'enters',sialt=ap- 'enterssuddenly',
JuganOstyaknjaxt- 'sneezes',niaxt=ip- 'sneezesonce'.
11 Also of Ugric origin are the Hungariandenominalverb-formant=h- and
its ObUgrianpendants.
12 The Hungarianand Vogul formulation of the caritive suffix is probably
ofUgric origin (but seealso Itkonen1992).
13 The non-attributiveform of the numeral 'two' is of parallel structurein
the three Ugric languages,viz. STEM+dual: HungariankettS« *kat;)y),
So Vogul kitiy, VVj Ostyak kiiiitk;m, cf. the attributive forms ket, kit,
kiiiit.
14 The long front rounded vowel of the Hungarian third-personsingular
pronoun,0, reflectsthe loss of an earlier *=1) suffix which is still attested
in OstyakandVogul.
15 The possessiveand verbal suffixes marking the first personplural are to
be reconstructedas 'roundedvowel plus *k' in proto-Ugric(Honti 1985a:
75--6).
16 The Hungarianword for 'flea', tetli, descendsfrom a form derived with
the samecompoundderivational suffix as in pO and pV (Honti 1985b:
153).
17 The =kdly element of Hungarian harkdly 'woodpecker' is an Ugric
derivationalsuffix-chain.
18 The developmentof the classof unstableverb stemsappearsto datefrom
the periodof commonUgric (Honti 1985a:50-52,77-8;1987).
19 The Ugric languagesshow a tendency to use distinct suffixes in the
formation of certaintypesof exocentricadjectivalcompoundsdepending
on whetherthe attributesarealienable(Hungarian=s, northernVogul =y)
or inalienable(Hungarian=u/Ii, northernVogul =p). (Kruman1983.)
20 The order of suffixes in nominal inflection is identical in the three Ugric
languages:STEM + PERSON SUFFIX + CASE SUFFIX for nouns,
STEM + CASE SUFFIX + PERSONSUFFIX for postpositions.
21 In all threeUgric languages,postpositionsreflect three-waydeixis.
22 In all threeU gric languages,coverbsplay an importantrole in the lexical,
aspectual,andpragmaticdomains.
23 Ugric preservesthe pU dichotomy of subjectivev. objectiveconjugation
(Honti 1990a).
 OBUGRIAN 355

24 Hungarianiafia 'descendants(of someone)'appearsto be a copulative

(*'daughter-s3+son-s3') compoundofUgric origin.
25 The Ugric languageshave a considerablenumber of sharedinherited
vocabularyitems (Honti 1979: 13-18).Among these,certaintypes invite
special attention, such as the vocabulary and even phraseology of
equiculture and the sharedself-designationHungarianmagy=ar 'Hun-
garian' =So Vogul maanisii 'Vogul'.
26 The Hungariancollocationkenyere-tkeres- 'earns[literally: seeks]one's
bread(= living)' is a taboo-motivatedeuphemismwhich may be retraced
to the Ugric period(Honti 1990b).

Referencesand Further Reading

General
Collinder, B. (1960), ComparativeGrammar of the Uralic languages,Stockholm:
Almqvist & Wiksell.
DEWOS= Steinitz 1966-93.
Ganschow,G. (1965), Die Verbalbildungim Ostjakischen,Ural-AltaischeBibliothek
13, Wiesbaden:Harrassowitz.
Gombocz, Z. (1898) 'A vogul infinitivus', Nyelvtudorruinyi Kozlemenyek28:
127-8.
Hajdu, P. (1966) Bevezetes az urdli nyelvtudomdnyba, Budapest:Tank6nyvkiad6.
- - (1985) 'Personalbezeichnungen fUr die 2. Personim Uralischen',COBeTcKoe
c:pHHHo-yrpoBell:eHHe1: 1-8.
- - (1987) 'Die uralischen Sprachen', in P. Hajdu and P. Domokos, Die
uralischen Sprachenund Literaturen, Budapest- Hamburg: Akademiai kiad6 -
BuskeVerlag, pp. 23-450.
Honti, L. (1974) 'Etimol6giai adalekok',NyelvtudomdnyiKozlemenyek76: 369-78.
- - (1976) 'Az osztjakszemelyje1615szuffixumokt6rtenetiattekintese',Nyelvtu-
domtinyi Kozlemenyek78: 71-119.
- - (1979) 'Characteristicfeatures of Ugric languages(Observationson the
question of Ugric unity)', Acta Linguistica AcademiaeScientiarumHungaricae
29: 1-26.
- - (1980) 'Milyen volt az obi-ugor alapnyelv teljesmaganhangz6rendszere?',
NyelvtudomtinyiKozlemenyek82: 172-90.
- - (1982a)GeschichtedesobugrischenVokalismusder erstenSilbe, Bibliotheca
Uralica 6, Budapest:Akademiaikiad6.
- - (1982b) 'A szalimi osztjak nyelvjaras hang- es alaktananakismertetese',
NyelvtudorruinyiKozlemenyek84: 91-119.
- - (1983a) 'Zur ugrischenLautgeschichte(Beitrage zur relativen Chronologie
einiger Lautwandel in den ugrischen Sprachen)',Acta Linguistica Academiae
ScientiarumHungaricae33: 113-22.
- - (l983b) 'AblautartigeVokalwechselin den obugrischenSprachen',FUF 45:
25-45.
- - (1983c) 'Alanyi es targyasigeragozasa demjankaioszjtaknyelvjarasokban',
in G. Bereczki and P. Domokos(eds), Urdlisztikai tanulrruinyok (Hajdu Peter60.
sziiletesnapjatiszteletere),Budapest:ELTE, pp. 173-8.
- - (1983d)'A harmadikszemelytje16l5szuffixumoka vogulban', Nyelvtudomd-
nyi Kozlemenyek85: 349-55.
356 OBUGRIAN

- - - (1984) Chrestomathiaostiacica,Budapest:Tankonyvkiado.
- - - (1985a) 'Szeljegyzetekinstabil tOvu igeink tortenetehez',Nyelvtudomanyi
Kozlemenyek87: 49-87.
- - - (1985b) 'OsmagyarhangtOrteneteitallinyok (Ugor hangtOrteneties etimolo-
giai jegyzetek)" Magyar Nyelv 81: 140-55.
- - - (1986a) 'Szlav hatas a magyar szamnevszerkesztesben?', Nyelvtudomanyi
Kozlemenyek88: 196-207.
- - - (1 986b) 'A vogul s es s alapnyelvi elozmenyei',NyelvtudomanyiKozleme-
nyek88: 258-63.
- - - (l986c) 'Er soIl mit BegrtiBung geschriebenwerden!', Finnisch-Ugrische
Mitteilungen 10: 155-63.
- - - (1987) 'Lautgeschichte,Etymologie, historisch-vergleichende Morphologie
(ein gemeinsamerType von Stammaltemationim Ugrischen)',in E. Lang and G.
Sauer (eds), Parallelismus und Etymologie, Studien zu Ehren von Wolfgang
Steinitz anliij3lich seines 80. Geburtstags, Linguistische Studien, Reihe A,
Arbeitsberichte1611II, Berlin: Akademieder Wissenschaften, pp. 187-93.
- - - (l990a) 'K rrpo6JIeMeB03HllKHOBeHllH 06l>eKTHOro CrrpH)KeHllH yropcKllx
H3hIKOB', in GV Fedjiineva(ed.), MaTepllaJIhI VI Me)l()l;yHapo):\Horo KOHrpecca
<pllHHO- yrpoBe):\oB,Moscow: Nauka,vol. 2, pp. 211-14.
- - (l990b) 'Ugrisches',LinguisticaUralica 16: 298-9.
- - - (1992) 'Adalek a magyar1- finn t megfelelesenek es alapnyelvieU5zmenye-
nek magyarazatahoz', in P. Dereky et al. (eds), Festschriftfor Karoly Redeizum
60. Geburtstag,Studiauralica6; Uralisztikai tanulmanyok3; LinguisticaSeriesA,
Studiaet dissertationes8, Vienna-Budapest:n.p., pp. 209-14.
- - - (forthcoming) 'Geschichtedes obugrischenKonsonantismus',in L. Honti
and M. Sz. Bakro-Nagy(eds),Uralische lautgeschichtlicheStudien,Budapest.
Itkonen, E. (1961) 'Suomalais-ugrilaisen kantakielenaanne-ja muotorakenteesta', in
E. Itkonen,Suomalais-ugrilaisenKielen-ja historiantutkimuksenalalta, Tietolipas
20, Helsinki: SuomalaisenKirjallisuuden Seura,pp. 48-84.
Itkonen, T. (1992) 'Ugrilaisten kielten karitiivista', in P. Dereky et al. (eds),
Festschriftfilr Karoly Redei zum 60. Geburtstag,Studia uralica 6; Uralisztikai
tanulmanyok3; LinguisticaSeriesA, Studiaet dissertationes8, Vienna-Budapest:
n.p., pp. 221-37.
Killman, B. (1983) 'A szerveses szervetlen kapcsolat egyik kifejezese az ugor
nyelvekben', in G. Bereczki and P. Domokos (eds), Uralisztikai tanulmanyok
(Hajdu peter60. szilletesnapjatiszteletere),Budapest:ELTE, pp. 193-205.
- - (1988) 'The history of the Ob-Ugric languages',in D. Sinor (ed.), The Uralic
Languages: Description, History and Foreign influences, Handbuch der Ori-
entalistik 811, Leiden: Brill, pp. 395-412.
Sz. Kispill, M. (1966)A vogul igenevmodattana,Budapest:Akademiaikiado.
- - - (1968) 'Uber einige Fragender ugrischenVerbalnomina',in P. Ravila (ed.),
CongressusSecundusInternationalis Fenno-Ugristarum... Pars I, Acta Linguis-
tica, Helsinki: SocietasFenno-Ugrica,pp. 267-72.
Kulonen, U.-M. (1989) The Passivein Ob-Ugrian, MSFOn 203, Helsinki: Societe
Finno-Ougrienne.
Liimola, M. (1944) 'Zu den wogulischenPersonalpronomina', FUF 28: 20-56.
- - - (1963) Zur historischenFormenlehredes Wogulischen,vol. I. Flexion der
Nomina,MSFOu 127, Helsinki: SocieteFinno-Ougrienne.
Redei,K. (1977) 'SzOfejtesek',NyelvtudomanyiKozlemenyek79: 201-16.
- - - (1988) (ed.) Uralisches etymologischesWorterbuch, vols. I-II, Budapest-
Wiesbaden:AkademiaiKiado-Harrassowitz.
Riese,T. (1992) 'Zur Entwicklung des Kasussystemsim Wogulischen',in P. Dereky
 OBUGRIAN 357

et al. (eds), Festschriftfilr Karoly Redei zum 60. Geburtstag,Studia uralica 6;

Unilisztikai tanulmiinyok 3; Linguistica Series A, Studia et dissertationes8,
Vienna-Budapest:n.p., pp. 379-88.
Sammallahti,P. (1988) 'Historical phonology of the Uralic languages',in D. Sinor
(ed.), The Uralic Languages: Description, History and Foreign Influences,
Handbuchder Orientalistik8/1, Leiden: Brill, pp. 478-554.
Sauer,G. (1967) Die Nominalbildungim Ostjakischen,Finnisch-UgrischeStudien5,
Berlin: Akademie-Verlag.
Steinitz, W. (1950) Geschichte des ostjakischen Vokalismus, Berlin: Akademie-
Verlag.
- - - (1955) GeschichtedesvogulischenVokalismus,Berlin: Akademie-Verlag.
- - - (1966-93): Dialektologischesund etymologischesWorterbuch der ostja-
kischenSprache,Berlin: Akademie-Verlag.
- - - (1989) OstjakologischeArbeiten,vol. III: Texteaus demNachlafi, Budapest-
Berlin: AkademiaiKiad6-Akademie-Verlag.
- - - (n.d.) (unpublishedessaysand notes by Wolfgang Steinitz on the history of
ObUgrianvocalism).
UEW= Redei 1988.
Zsirai, M. (1933) Az obi-ugor igekotf5k, Ertekezeseka Nyelv- es Szeptudomdnyi
Osztalykorebol, 26/3, Budapest:MagyarTudomanyosAkademia.

Contacts
Futaky, I. (1975) TungusischeLehnworterdes Ostjakischen,Veroffentlichungender
SocietasUralo-Altaica 10, Wiesbaden:Harrassowitz.
Hajdu, P. (1990) 'Einigestiber Ftirworter', Linguistica Uralica 26: 1-12.
Joki, AJ. (1973) Uralier und Indogermanen,MSFOu 151, Helsinki: SocieteFinno-
Ougrienne.
Kalman, B. (1960) Die russischenLehnworterim Wogulischen,Budapest:Akademiai
Kiad6.
Korenchy, E. (1972) Iranische Lehnworterin den obugrischenSprachen,Budapest:
Akademiaikiad6.
- - - (1981) 'Iranian contactsduring the period ofUgric division', in G. Ortutay
(ed.), C4IFU, vol. III, Budapest:AkademiaiKiad6, pp. 70-76.
Kulonen, U.-M. (1990) 'Das Nord-, Stid- und Ostobugrische:SyntaktischeUnter-
schiede und Ubereinstimmungen',in L. Jakab, L. Keresztes,A Kiss and S.
Maticsak(eds),C7IFU, vol.lB, Debrecen:CongressCommittee,pp. 139-145.
Liimola, M. (1971) 'Zur Kasuslehreder Vach- und Vasjugan-Mundartder ostja-
kischenSprache',JSFOu71: 3-24.
Redei, K. (1970) Die syrjiinischen Lehnworter im Wogulischen,Uralic and Altaic
Series 109, Budapest-Bloomington-TheHague: Akademiai Kiad6-Indiana
University-Mouton.
Riese, T. (1984) The Conditional Sentencein the Ugrian, Permian, and Volgaic
Languages,Studia Uralica 3, Vienna: Verband der wissenschaftlichenGesell-
schaftenOsterreichs.
Steinitz, W. (1980) OstjakologischeArbeiten, vol. IV: Beitriige zur Sprachwis-
senschaft und Ethnographie, Budapest-Berlin: Akademiai Kiad6-Akademie-
Verlag.
Toivonen,Y.H. (1956) 'Uber die syrjanischenLehnworterim Ostjakischen',FUF 31:
1-169.
Vertes,E. (1967)Die ostjakischenPronomina,Uralic andAltaic Series74, Budapest-
Bloomington-TheHague:AkademiaiKiad6-IndianaUniversity-Mouton.
12 Khanty
Daniel Abondolo

Khanty (older name:Ostyak)is a complexchain of dialectsspokenby people

who live in a vast,roughly L-shapedareaalongthe Ob', the lower Irtysh, and
tributaries.Accordingto the mostrecentfigures (1989census),therearesome
22,000speakersof Khanty; of these,62.9 per cent were native speakers(i.e.
c. 14,000).Khanty speakersmakeup about1 per centof the populationofthe
Khanty-MansiAutonomous Okrug (Kiinnap 1993: 86).
From a historical-typologicalperspective,Khanty presentswith a rather
odd amalgamof central and peripheral Uralic features. Like Permian and
Hungarian,Khanty has lost the proto-Uralic accusative*-m, but like Saarnic
and Samoyedic,it has preservedtraces ofthe proto-Uralic reflexivizer *=j.
Khanty is the only Finno-Ugric languagein which proto-Uralic *& has
merged fully with *j in all positions - the regular developmentin proto-
Samoyedic.The closestcongenerof Khanty is almost certainly Mansi, but
many of the KhantylMansisharedfeaturescould be Sprachbundconvergence
phenomena,and many aspects of their putative protolanguage,proto-
ObUgrian(most critically, the vocalism)havenot beenreconstructedin fully
convincing detail (Abondolo 1996: 7-15). Some of the mystery of the
morphophonemicsof the present-dayKhanty dialects is a result of our
ignoranceof this aspectof Khanty (andObUgrian) linguistic prehistory.

Dialects
On both historical-phonologicalandsyntactic-typologicalgrounds,thesemay
be broken into two major groupings,East v. West. The East group further
subdividesinto (1) the Far EasterndialectsV(ach) andVasjugan(Vj), and (2)
the Surgut group, which includes Jugan, Malij Jugan, Pim, Likrisovskoe,
Tremjugan(Trj), and Tromagan(Tra); for the differencebetweentheselast
two varieties, seepp. 362-4. The Vartovskoedialect is transitionalbetween
the Surgutandthe Far Easterndialects.Thereis also a Sal(ym)dialect,which
some see as transitional betweenthe major East and West groupings, but
which probablyis betterclassifiedwith the southernsubgroupof the western
group.
The West group subdivides into North and South subgroups.Clearly
southern are the Oemjanka (ON OT) dialects and Konda, Cingali, and

358
 KHANTY 359

Krasnojarsk.Clearly northern are the O(bdorsk) dialect and the Ber(jozov)

subgroup,consistingof the Synja,Muzhi, and Shurishkardialects,and, to the
south, Kaz(ym). TransitionalbetweenNorth and Southare the Sher(kal)and
Ni(zyam) dialects.
Reliable population figures for the various dialects are not available.
Perhapstwo-thirds of Khanty speakersuse a northerndialect, the remaining
third being speakersof one of the easternforms; southernKhanty is probably
no longerused.
Selecteddialect isoglossesare set out in Figure 12.1. The table has been
designedto dramatizethe fact that the periphery,i.e. far northwestObdorsk
and far southeastVj and V, is in severalrespectsmore conservativethan the
centre, but also that there are pocket retentionsjust short of the periphery,
such as the /- of Kazym and the Surgut group. The westerninnovation *k- >
x- before back vowels may be seenas convergencewith northern and some
easternMansi dialects,where the samechangetook place, e.g. (in the word
for 'three')Pelymka(western)Mansi kuurfim, v. Lower Konda (eastern)and
Sosva (northern) Mansi xuurfim = Tremjugan Khanty koo/-;nn v. Obdorsk
Khanty xool;nn =Nizyam Khanty xuutfim.The -r- of all Mansi dialectsin this
word is probably an example of an earlier convergence,namely of proto-
Mansi andproto-Hungarian,cf. Hungarianhaarom.
Vowel harmony of the palatovelar(i.e. front/back) type is found in the
easternmostdialects: Vach, Vasjugan, and some of the Surgut group.
Example: Vach kooy-iiiim 'my stone', kuul-aam 'my fish'. Since Southern
Mansi dialectsalso had vowel harmonyit is assumedthat vowel harmonyin
eastKhanty is a retentionfrom at leastproto-ObUgrian times.
Vowel inventoriesvary greatlyfrom dialectto dialect,with the largestin the
eastandthe smallestin the north. As in mostUralic languages,andin keeping
with the principlesof vowel harmony,the full vowel inventory occursonly in
the first syllableof the word, the vocalismof non-first syllablesbeingseverely
restricted in every Khanty dialect, regardlessof whether it has retained

Figure 12.1 SelectedKhanty dialect isoglosses

0 Ber Kaz Sher Ni Irt Sal Trj Vj V

VH - +
*kB- x- k-
*c i si ti
-s- past + - +
*i- I- i- t- i- j- I I-
*1 I i t i1 i1
*aa aa aa 00 aa aa

Note: VH =vowel harmony,B =backvowel. Fordialectabbreviations,seetext.

360 KHANTY

vowel harmony.As in all Mansi dialects,the vocalismof all Khanty dialects

is characterizedby a quantitativeopposition, termed in Ostyakologyfull v.
reduced;full vowels will be transcribedhere as long, i.e. written geminate.
Examplesof minimal pairs: Vach kaOt 'hand',kat 'distance';seem'eye',sem
'heart';jool 'sorcery',jol 'border';saart 'pike (fish), , sart 'interstice'.

SampleInventories
A relatively rich systemis that of Vach Khanty, in the far southeast,with nine
(or eleven,Katz 1975: 78-81) full, andfour reducedvowels:

ii titi 11 uu
ee 66 00 e 6 o
aa (rere) aa (M) a

ContrastObdorsk,in the far northwest,with only six and three,respectively:

11 uu
ee 00 o
aa aa a a

The smallestsystemis found in Nizyam, Sherkal,andBerezovo:

uu 11 u
ee o
aa aa a

This inventoryis identical with that of SosvaMansi, which is spokendirectly

to the west. At least two waves of rotation have shuffled the lexical
distribution of the vowels of thesetwo idioms, with the result that for each
vowel in Nizyam Khanty there are two, equally regular, correspondentsin
SosvaMansi. For example,Nizyam Khanty ee correspondsto SosvaMansi
a in seem/sam'eye', keew/kaw'stone', teel)k;Jr/tal)k;Jr 'mouse',and weet-
/al- 'kills', but to SosvaMansi aa in weet;}ffl/waal;}ffl 'marrow', eew;J/aayi
'daughter',keej/kaaP'female(animal)',and teep-/laap-'hasenoughroom to
fit' . In parallel fashion,Nizyam Khanty aa correspondsto SosvaMansi aa in
aamp/aamp'dog', paaw/paakw 'pine cone', niaar;J/niaar 'raw', and saas/
saans 'knee', but to SosvaMansi ee in niaat;}ffl/nieel;}ffl 'tongue', taal)bl
leel);;Jn 'squirrel',aal);;m/eel);;Jn 'chin', andpaat-/peet-'cooks,boils'; parallel
patternshold for the rest of the vowels, the regular Nizyam Khanty/Sosva
Mansi correspondences being ilee and iii, o/u and 0/0, uu/o and uu/aa, aa/aa
andaa/uu, u/uu and u/u, ali andala.
In Vach, Vasjugan,and the Surgutdialects,many nounsand verbsundergo
stem-vowelalternationsin their paradigms,e.g.Trj Mat 'house', processes
kuut-~ 'my
house'.The same,as well as other, alternationsunderlie many processesof
 KHANTY 361

(now no longer productive)derivation, e.g. Trj kaiir 'bark', kiir 'snowcrust';

jiilfk W 'ice', je.(]k 'water'; k<rt 'hand', kiit=;Ji 'bear's forepaw'. The
reconstructionof the vowel system of proto-Khanty (and thus of proto-
ObUgrian) is complicatedby thesealternations,whoseextentin both space
andtime is hotly disputed.

Consonants
In all Khanty dialects distinctive voice is marginal or non-existent.On the
other hand all Khanty dialectshave the correlationof palatalizationto some
degree.For example,Vach Khanty has t -:;:. I, n -:;:. Ii, 1-:;:. Pand Kazym Khanty
hass -:;:. si, n -:;:. Ii, i -:;:. I. (The Surgutdialectsand Kazym Khanty haveplain and
palatalized voiceless lateral fricatives.) Most Khanty dialects have retroflex
consonantphonemes:parallel with the hushingnon-palatalizedc and/orswhich
are well establishedin e.g. Mansi, Komi, and Hungarian,we find in e.g. Vach,
Vasjugan,and Kazym a retroflex lateral (written here() and a retroflex nasal(1]).
Minimal pair from Vach: liis '(snarefor trappingbirds)', fjis 'slack,not tight'. The
Surgutdialectsdistinguishlabializedv. plain velars(as in SosvaMansi); theseare
in perhapsall casesthe result of transfer of the feature of labiality from the
precedingvowel, e.g.TrjjayWei 'bow' < *joyei; cf. Vachjooyet.

Morphology of the Noun: CaseSuffixes

The numberof casesvaries considerablyfrom dialect to dialect. Vach, in the
far east, has ten cases; Obdorsk, in the far north, has three. All dialects
distinguishsingular, dual and plural forms of both possessorand possessed;
non-possessed nounsalso have three numbers.For example:Nizyam Khanty
xaap 'boat' hasthesenon-possessed forms:

Singular Dual Plural

N xaap xaap-lJ~nxaap-lJ~n-na xMlp-~txaap-~t-na
Lat xaap-a xaap-lJ~n-a xaap-lJ~n-na xaap-~t-a xaap-~t-na
Loc xaap-na xaap-lJ~n-na xaap-lJ~n-na xaap-~t-na xaap-~t-na

Examplesof possessed forms: xaap-eem 'my boat', xt'lap-.(]fJt-aam 'my (two)

boats',xaap-t-am 'my (3 or more) boats', xaap-eemfJn 'the boat of us (two)',
xaap-.(]fJt-fJrnfJn 'the (two) boats of us (two)'. Personsuffixes precedecase
suffixes,e.g. xaap-eem-na 'in my boat'.
Khanty verbs observeessentiallythe samecategoriesas Mansi verbs, viz.
(1) definite v. indefinite direct object; (2) within the category 'definite',
number(sg/dual/plur)is distinguished;(3) there is also a so-called'passive'.
Thus in Vach we have e.g. (-1- is the present-tensemorpheme)tuu-l-fJrn 'I
bring', tuu-l+m 'I bring itlhimlher', tuu-l-;,ylaa-m 'I bring them (two)', tuu-
l-laa-m 'I bring them (3+)" tuu-l-uuj-fJrn 'I am brought; X is broughtto me'.
Theseare the samefive distinctionsmadeby SosvaMansi, but with different
material.
362 KHANTY

Whereasall known Mansi dialects have distinct suffixes for presentand

pasttense,Khanty presentsa morecomplexpicture: northerndialectshavean
-s- past-tensemorpheme,southerndialects(and Nizyam) havea (2) past-tense
morpheme,andVach, Vasjugan,andmostSurgutdialectshaveboth.

Sketchof TremjuganfI'romagan Khanty

Tremjugan (Trj) and Tromagan(Tra) Khanty are two varieties within the
Surgutsubgroupwhich differ in time ratherthan in space:Tromaganis now
spokenwhere Tremjuganwas in the nineteenthcentury, along a tributary of
the Ob' which is called tor~the jayW;m 'sacredriver' by Trjffra speakers
(Csepregi 1993: 59). Becauseof the sparsenessof available material, data
from both varietieswill be adducedin the sketchwhich follows, with the two
temporalvariants(Trj or Tra) markedonly whenit is crucial to the discussion;
datafrom other dialectsis also adducedfor purposesof contrast,or when an
appropriateTrjffra exampleis lacking.

SegmentalInventory
Vowels.In the first syllable, Trjffra Khanty has a vowel inventory of thirteen
terms,with nearlyequalmembershipin the two quantitativeclasses.
In non-firstsyllable,TremjuganKhantyhadsix full vowels,front ii eeiiii and
backi'i'ee aa, andtwo reducedvowels,front andbackschwa:;}fJ. In Tromagan
Khanty, the frontlback opposition has been neutralized in favour of the
unmarkedmembers;fieldworkersreportonly ii eeaa ;}. This neutralizationhas
eliminatedvowel harmonyfrom the phonology:contrastTremjuganwaajfJy,
Tromaganwaaj;Jy 'animal'.Throughoutthis chapter,only theunmarkedforms
will normally be cited, with the understandingthat in earlier, i.e. Tremjugan,
instantiationstherewasalsothecorrespondingfront or backanalogue.

Noteson Pronunciation
The phonetic realization of iiii is more central than front; since the last
century, this phonemehas been moving back, in the wake of aa, which
developedfrom earlier *aa (** in the table).The reducedvowels classedhere
as mid are usually pronouncedas closemid. The durationof the low reduced
vowels often approachesor equalsthat of the full lows; the primary acoustic
cueis then vowel colour (Csepregi1993: 60).

Table 12.1 First-syllablevowel inventoryof Trjtrra Khanty

Full Reduced
High 11 11 uu
Mid ee 00 e 0 0
Low aa ** aa a re a
Rounded + + + +
Back + +
 KHANTY 363

Full vowelsmay be pronouncedshort, especiallyin non-flrst syllables.

Within the non-compoundword, stressis neverdistinctive, andis normally
on the flrst syllable, but may move to the second syllable under certain
circumstances(Honti 1988: 176). Secondarystresshas been perceivedon
non-flnal odd syllables,e.g. sii"m+ienk-k;}'I-aam EYE+WATER-dual-sl 'my
tears'.There is evidencethat stresson subsequentsyllableswithin the noun
phraseis forbidden, e.g. niuur ooy=;}]) BARE HEAD=ED 'bald', cf. w~r=;}o w~r=;}o
wiiiinim=;}p BLOOD=adj CHEEK=ED 'rosy cheeked'.
Consonantsand consonantsandhi.The TrjfTra Khanty consonantinven-
tory distinguishesthree placesof apical articulation: a dental, a palatalized,
and a cacuminal series, in rather uneconomicalfashion. This three-way
oppositionis phoneticallyat its purestamongthe nasals.Mfricate pronunci-
ation characterizesthe palatalized obstruent usually; it characterizesthe
cacuminalobstruentalways,exceptbeforethe apicalfricatives I ands.

Nasals m n ~~ I,l IJ IJw

Stops P t ~ kk yWW
Affricate ¢
Non-lateralfricatives s y yW
Lateralfricatives i i-i
Lateral approximant j
Glides w j
Trill r

The bilabial approximantor glide w and the labialized velar fricative or

approximantyW are probablyto be subsumedunderonephoneme.
Both the palatalized and the cacuminal consonantscontinue proto-
Khanty segments,where the palatalized consonantsare of proto-Uralic
origin, and all the cacuminalssavec seemto be proto-Khantyinnovations.
A pre-Surgutinnovation is the transferof labiality from a precedingvowel
to a following velar (k YO); this has given rise to an opposition between
plain and labialized velars, e.g. (eekW'ring', cf. Vach (cecek. Note also *-aIJ-
in laoaal '(s)he enters',but *-OIJ- in laow '(s)he entered'.Thesetwo velar
series contrast to the right of the two vowels which are each the result
of a merger of one rounded with one unroundedpre-Surgutvowel: flrst-
syllable a (ultimately from *1) and ii (from *ijti) , e.g. layWr 'knot in a
bough or in split wood' : layr 'wave', piiyw;}1 'heart (of the bear; a taboo
word; cf. Vach piiiiy;}l 'flshing-line') : pUy;}1 'patch on boat' (cf. Katz
1975: 88-9). Another Surgut innovation, but one shared with Kazym
Khanty in the west, is the fricativization of the proto-Khanty plain and
palatalizedlaterals, *1 > i and *lj > i-i, the cacuminal l alone remaining
an approximant.
When adjacent to kyO, the voiced velar fricatives y and yW are
automaticallyreplacedby the correspondingstopsk and kW, e.g. y-y > k-k in
364 KHANTY

jek-k;m iiiiI)kee-y;m FATHER-dual MOTHER-dual 'father and mother' (cf.

jey 'father', iiiiI)kii 'mother'), yW -k > kWk in lakw kiinitiiiii 'comparedto a
horse' (cf. layw 'horse').Stem-finaly goesto k before the adjective-forming
suffix =fJI), e.g. w<Ek=fJI) 'strong'(cf. W<Ey 'strength').
There is left-to-right assimilationamongthe distinctively D(ental), P(ala-
tal), and C(acuminal)consonantssuch that C+D > CC and P+D > PP if the
secondconsonantis an obstruent,e.g.poc- 'rakestogether'plus the infinitive
suffix -taayfJ yields poccaayfJ (with longer durationin the closedphase,not
in the release), and the infinitive of /aa/- 'stands' is laaftiaayfJ. The
sequencec-I yields [ttl becauseof the unreleasedpronunciationof cbefore
I, e.g. [til in koc-I-eem 'I tune it (musical instrument),I swaddleit (child in
cot)'.

Morphology

Architectonics
Monomorphemicnounstend to be of the shape(C)VC(C) or (C)VC(C);:!(C),
i.e. with at mostonefull vowel, andthat in the first syllable.Apart from nouns
endingin filii, which continuesa proto-Khantydiphthongwhosecomponents
more often than not straddleda morphemeboundary, deviance from this
canonusually signals a loanword or affect, e.g. jiimaas 'shirt' (a loan from
Komi), nieeyWreem 'child'. Consonantclustersin initial position seemto be
excluded,even in fairly recentloans such as poruunt 'one of severalsmall
piecesof wood usedas a guide in hollowing out a boat' « Russiandialect
sprunt, cf. standardspunt, from GermanSpund). Final clusters are always
homotopic,and their secondmemberis always a stop or affricate, e.g. mp, rt,
flC, I)k. In triconsonantalclusterswhich arise through the concatenationof
morphemes,the middle consonantis cancelled, but its voicelessnessis
transferredto the latter portion of the precedingsegment,e.g. perfectkenc=@
-fJrn 'I washunting,looking for' : presentken[N]-I-fJrn 'I'm hunting,I'll hunt'.
Stemsof the shape(C)VC(C);:!C syncopatethe schwa before vowel-initial
suffixes, e.g. kaalf1r 'nephew': kmy-fJrn 'nephew sl', jOnt;ry 'bowstring',
jonr]y-:}I 'bowstring s3', kiimp:}t 'unevenness in bark of birch; individual
scaleof cedarcone' : s3 kiim[M]I-M.
Further examplesof each shape: (C)VC(C): cuunc 'flea', iiiimp 'dog',
niiiiI)k 'larch', saart 'pike (fish)" kilar 'male (animal)', ayW 'current', tiayw
'thin ice', koos 'star', kor 'swampy place'. (C)VC(C);:!(C): pan:} 'string',
aalfJI) 'tip', iip:}l 'smell',muuyM 'liver', niiiiilfJrn 'tongue',nioorfJrn 'swamp',
kilalfJp 'net', k<Ej;ry 'female(animal)',ooyfJt 'sandbankon river'.

Ablaut
We may distinguishtwo typesof ablautseries.In the first, the high full vowels
alternatewith their low and midcounterpartsin the following sets:ii - iiii, ii
- ee; uu - aa, uu - 00. This type occurswidely in both nounsand verbs, in
 KHANTY 365

both derivation and inflection. Examples from nominal inflection: maay

'beaver': muuy-;]m 'beaversl', iiiimp 'dog' - iimp-;:nn 'dog sl', leekw 'ring':
liikw-;:nn 'ring sl', ooy 'head' : uuy-;:nn 'headsl'. In verbs,aa alternateswith
iT in the imperative: kaant- 'picks up' : kiTnm-ee 'pick it up!' and in
derivation, e.g. faalfJrn- 'steals' : fiTfm=fr;t- 'approachesstealthily'; the aa
- iT alternationalso occursin at least one noun, namely kaalfr; 'nephew' :
kiTfy-fJrn 'nephewsl'.
The secondtype is restrictedto verbal inflection. Here the ablautinvolves
the reducedsetsii - e and a - 0, e.g. wiif-f}.eem 'I killed it' : wei-ee 'kill it!',
kafaa-f '(s)he is dying' : kof-f) '(s)he died'. As mentioned above, the
roundednessof 0 or 0 is transferredto a following velar, giving rise to
alternationssuchas faIJ-aa- 'enters'- faIJw '(s)heentered', iiyw;]f_ 'believes',
s2 imperativeiiyf-iiii; thereare also analogicallevellings.
The age of the variouskinds of ablautfound in Khanty is one of the most
hotly disputed questionsin Uralic historical and comparativelinguistics;
opposingviews may be viewedin Honti (1982: 18-23) and Katz (1987-8).

Nominal Paradigms

Nominal Stem Types

Two nounshaveirregular stemallomorphy: koo 'man',with a stemkuu(j)- in
certaininflected forms, e.g. kuuj-;:nn 'my husband';andnee - niiIJ- 'woman',
e.g. niiIJ-;]t 'women'.

Number
Singularis markedwith zerodual (~; dual is -y;]n (-y;]f- beforepersonsuffixes);
plural is -t (-f- before person suffixes). Consonant-finalstemsepenthesize
schwabefore -t, and schwa-final stemsshow a final low full vowel before
both -t and -y;]n. Examples:sN iiyWii 'daughter',dN iiyWii-y;]n, pN iiyWii-t;
sN kaat 'house',dN kadt-y;;m, pN kaat-;]t; sN tayt;] 'piece',dN taytaa-y;;m,
pN taytaa-t. Stem-final ii vacillates: it either remains (as in dN iiyWii-y;;m
above)or changesto ee, e.g. iiiiIJkee-y;;m 'motherdN'.

Person
The personsuffixes are sl -(V)m, s2 -(V)(n), s3 -f, dl -m;;m, pI _yW, p3 -iii,
all others(i.e. d23p2)-iin. Sampleforms may be found in Table 12.2.

Stem-suffix Sandhi
To the left of personsuffixes with an initial consonant,stem-finalschwa>aa,
e.g. panaa-f 'string s3', taytaa-yW 'piecepI'. To the left of personsuffixes
with initial ii, all final stem-finalvowels arecancelled,e.gpan-iii 'string p3',
wiif-iin 'reindeer d23p2'. But the chief locus of allomorphy in person
inflection is to be found in the first and secondpersonssingular,where there
is a complex pattern of vowel alternationwhich is reminiscentof vertical
vowel harmony,and which is capturedhere as (V). When attachedto vowel-
366 KHANTY

Table 12.2 Numberandpersonsuffixesof a Trjffra Khanty noun (all

nominativeforms): klult 'house'

Possessorin Possessionin 1stperson 2ndperson 3rdperson

Singular
Singular kuut-dffi kuut-dn kuut-di
Dual kaiit-Ydi-aam kiiiit-Ydi-aa kiiiit -ydi-O
Plural kaat-i-aaffi kaat-i-aa kiiiit-di-O
Dual
Singular kuut-ffidn kuut-iin kuut-iin
Dual kiiiit -Ydi-dffidn kaat-Ydi-dn kiiiit-Ydi-dn
Plural kaat-i-dffidn kaiit-i-dn kiiiit-i-dn
Plural
Singular kuut-dYw kuut-iin kuut-iii
Dual kaat-Ydi-dYw kiiiit-Yd<l-dn kaat-Ydi-aai
Plural kaat-i-dYw kaiit-i-dn kiiiit -i-aai

fmal stems, suffixes with initial (V) have zero vowel; stem-final ii then
changesto ee and schwa changesto aa, e.g. wiilee-m 'reindeersl' (wiilii
'reindeer'),panaa-m'string sl' (pan;) 'string'). When attachedto consonant-
final monosyllabicstems,(V) is read(1) as a full low vowel, if the root vowel
is a reducedvowel; (2) as ee « *ii *11), if the root vowel is a high full vowel;
and (3) as schwa,if the root vowel is a low full vowel; in this last casethe
low vowel ablautsto its high full counterpart.Examplesof thesethreetypes
are (l)pay-aam'sonsl', miiyWI-iiiim 'breastsl', kar-aam 'foot sl'; (2) kuul-
eem 'fish sl', liil-eem 'breathsl'; (3) kuut-;;»n 'housesl' (lazat sN), iimp-;;»n
'dog sl' (iiiimp sN). Somekinship terms behaveas if they belongedto class
(3), e.g. uup: uup-;;»n 'my father-in-law',jey-;;»n 'my father'. Rootswith more
thanonesyllablefluctuate.
The (n) final in the secondpersonsingular suffix is realized as n when
followed by a case suffix or a postposition, as zero otherwise. Both the
debilitation of this n and the lowering of the vowel to its left are Surgut
innovations(Honti 1984: 46).

Case
Casesuffixescomeafter any numberor personsuffix. Following tradition, we
shall call the morphologically unmarked (-t'» case the nominative, even
though absolutive would be a better name. There are no other dedicated
grammaticalcases,the remainingnine being either local (locative -n;), lative
-aa, approximative -naam, ablative -ii) or more abstract, i.e. adverbial
(instructive -aat, comitative -naat, abessive-I;;,y, expletive -ptii, translative
-y;}). There is also a suffix -iin which forms temporal adverbsfrom time
words suchas loy 'summer',e.g. loy-iin 'in the summer(time),. Certainother
casesuffixes, now extinct, are fossilized in deictics,e.g. separative-I in to-I
 KHANTY 367

Table12.3 SelectedTrjffra nounnumberandcaseforms: kUat 'house'

Singular Dual Plural

N kaat kaat-yan kiiat-at

Lat kiiiit-aa kaat-yan-aa kaat-at-aa
App kiiat-naam kiiat -yan-naam kiiiit-at-naam
Loc kiiiit-na kiiat-yan-na kaat-at-na
Abl kiiiit-ii kiiiit-yan-ii kaat-at-ii
Inst kiiiit-aat kiiiit -yan-aat kaat-at-aat
Com kiiiit-naat kiiiit-yan-naat kaat-at-naat
Trans kiiiit-ya kiiiit-yan-ya kaat-at-ya

'from there'; see p. 370. There is also a distributive suffix -t;}ftaa which
occurs with some postpositionsand with numerals. Abessive plural, and
expletivedual and plural forms are not attested.For sampleforms, seeTable
12.3.
Suffix order is invariably person+ case,as in Mansi and Hungarian,e.g.
iiiiI]kee-m-n;} MOTHER-sl-loc, fuuf-;}f-ii MOUTH-s3-abl, ont-;}f-n;}
INTERIOR-s3-loc'(thought)to himlherself.Suchbisuffixal forms appearto
be relatively rare,however.

Postpositions,Preverbs,and Gerund
The true postpositionsare nounswith defectiveparadigmsand distributions.
Somehave only one form, with fossilized, synchronicallyopaquemorpho-
logy, e.g. the intrinsically lative mob '(to) as far as', e.g. flldr tom=pii peet;}
mob LAKE THERE=adj SHORE AS.FAR.AS 'as far as the opposite/far
shore of the lake'. Most, however,occur with at least one, and usually two
or moresynchronicallysegmentable local suffixes attached,e.g.lativepiTr-aa
'to behind' : locative piTr-n;} 'behind, after' : distributive pir-;}ftaa 'from
behind'; this is especiallytrue of postpositionsformed with the suffix =pii,
e.g. locative iTf=pii-n;} 'under' : lative iTf=pij-aa 'to underneath': ablative
i"i"f=pii-jii 'from under', cf. iH+piilfkw 'lower teeth'. It is probably impos-
sible to draw a sharpouter perimeteraroundthe classof postpositions,since
many nouns can enter into ad hoc postpositionalconstructions,e.g. kiim
'extent' in fuuf-eem kiim-n;} jeI]k-n;} fddf-f-;Jm MOUTH-s1 EXTENT-loc
WATER-loc STANDS-pres-s1'I'm standingin waterup to my mouth'.
Postpositionsare defective distributionally, i.e. they are syntactically
bound, in that they always occur either with a head noun or with person
suffixes. Nounsareusuallyin the nominative,e.g. fddrttJ-fJ kot-n;} SNOW-sN
INTERSTICE-loc 'in the snow', pef;}I] ont-naam CLOUD INTERIOR-app
'into (the) cloud'. In mostKhanty postpositions,personsuffixesprecedecase
suffixes just as in the inflection of fully productive nouns, e.g. kot-iin-n;}
INTERSTICE-d2-loc 'between the two of you', iifp-eem-aa
SPACE.lN.FRONT-s1-lat'to in front of me'. Here Khanty patterns with
368 KHANTY

Mansi (e.g. North Mansixal-anel-tINTERSTICE-p3-loc'betweenthem')but

not with Hungarian,where the order is casesuffix + personsuffix (koz-t-iik
INTERSTICE-Ioc-p3 'betweenthem'). In Khanty, only a very few older
postpositions show the reverse order, e.g. Synja Khanty nal) xosi-aj-;m
PRO.s2 SPACE.BESIDE-Iat-s2'to beside you', cf. the isomorphic and
synonymousHungarianhoz-zaa-d.
Another special subclassof adverb is the set of preverbs, which are
adverbial,usually lative, particlessuchas nok 'up', iTt 'down', uut~ 'back, '(up) on
to shorefrom the water; (up) inland from the shore',niik '(down) to the shore
from inland; (down) from a smallerriver to a larger one',jakW~'back,'back,home,
backhome'.Preverbscreateloosecompoundswhich are semanticallyand/or
aspectuallydistinct from their simplices,e.g. wiir- 'does,makes,puts' : iTt
wiir- 'buries', kiif- '(a)rises' : nok kiif- 'standsup'. Their normal place is
immediatelybefore the verb, but a negativeparticle usually intervenes,e.g.
'back,
toy~ iiiif tiiiifiifiitt;m 'don't (d2) drag me into it', cf. toy;:) tiiiif- 'drags
thither'. From the lexico-cultural and syntactic point of view the Khanty
preverbsare most closely akin to thoseof Mansi, but thereare closeparallels
in Hungarianand even,in termsof aspect,in Estonianand Livonian, as well.
See Sauer1992 for a review of the literature and a discussionof preverbals
from pronominalbases.
For the gerund,seep. 376.
Comparison
There is no comparativesuffix. In TIjffra Khanty, the bestedmemberof a
comparisonis marked with a postpositionborrowedfrom Komi, e.g. wiifii
tayW kiiritiiiii riaayM REINDEER HORSE COMPARED.TO SHORT 'a
reindeeris shorter than a horse'. In Vach and VasjuganKhanty the bested
member is marked with the postposition niil)iit (niil)a), e.g. Vj looy
mees+niil);:) ooy~r ooy~r HORSE COW+niiIJ~COW+niiIJ~ VALUABLE 'a horse is more
valuablethan acow'.
Pronouns
The caseinventory of the personalpronounsdiffers from that of nounsin two
key points: (1) they have a dedicatedaccusativeform, built with the suffix
-(aa)t, and (2) there are no distributive and expletive caseforms. Severalof
the casesshowdeviantformations:(1) all obliquecasesotherthanthe locative
and ablative are formed by the addition of an element =tii (with regular
alternant =t;:)j- before vowel-initial case suffixes), and in the first-person
singularpronoun(only!) this elementis followed by the first-personsingular
suffix, resultingin the sequence=t-eem; (2) to the right of this =tii, the lativel
dative suffix is zero; (3) the ablativeis formed by the addition of an element
+niil)t- (cognatewith the postpositionusedin VVj in comparativeconstruc-
tions mentionedabove)followed by the appropriatepersonsuffix.
The personalpronounsalso haveintensiveforms, especiallyin far eastern
andsoutherndialects(Honti 1984: 72-3).
 KHANTY 369

Table12.4 Paradigmsof selectedTrj personalpronouns

PRO.s1 PRO.s2 PRO.d3

N mliii nOIJ hin

A maan-t nOlJaa-t hinaa-t
LatIDat maan-t-eem nOlJaa-tii-0 hinaa-tii-0
App maan-t-eem-naam nOlJaa-tii-naam hinaa-tii-naam
Loc maa-ng nOIJ-ng hin-ng
Abl maa-niilJt-eem nOIJ-lJiilJt-ee hin-niilJ t -iin
Com maan-t-eem-naat nOlJaa-tii-naat hinaa-tii-naat
Instr maan-t-eem-aat nOlJaa-tgj-aat hinaa-tgj-aat
Trans maan-t-eem-yg ? hinaa-tii-yg
Abe maan-t-eem-igy ? ?

Personalpronounsare usedsolely for emphasis(foregrounding,contrast).

Paradigmsof threesamplepronounsare setout in Table 12.4.

DemonstrativePronounsand Other Deictics

The eight demonstrativepronouns distinguish concrete/visible(with the
consonantalframe t--m) v. abstract/invisible(with initial ti -), near(with -ee-,
-ii-) v. far (with -0-, -uu-), and independentv. attributive forms (with and
without final-ii, -t):

Figure 12.2 Trjffra Khanty demonstrativepronouns

Independent Attributive I Independent

~ tleemlii tleem tilii tiliilt

~ tleemlii tleem tilii tiliilt

ConcreteNisible AbstractlInvisible

The 'independent'forms can be used attributively, as well, and the

'invisible' category includes anaphora. Examples: tom pii/~-n;}
includesTHAT
SIDE-Ioc 'on that side', tiuu Nlar-aa tie iit-fJ-fJ THAT LAKE-Iat BEHOLD!
ARRIVES-perf-s3 'and sure enough, (s)he arrived at that (previously
mentioned) lake', teemii toyW;;m man=ma-/ taay;; THIS THITHER
GOES=past.part-s3 PLACE 'this is the placewhere(s)hewent by'.
Otherdeicticsinclude spatialadverbsbuilt from the stemste- for near,and
to- - taa- for far reference,with suffixation of -t, -I, and -y;J(-naam)
indicating stasis, motion from, and motion towards the location referred
to:
370 KHANTY

Stasis Motion from Motion to Motion towards

Near te-t te-i te-y~ te-y~ te-y~-naam te-y~-naam

Far to-t to-i tMl-Y~ tMl-Y~ taa-y~-naam taa-y~-naam

There are also variantswith palatalinitial, e.g. fee-naam'in this direction',

and modal deictics such as fee(y;,)n;, 'in this way'. 'Now' is iit, 'just now;
in a moment(Frenchtout a/'heure)' is iin, 'then' is fuut-n;,.
Interrogative and indefinite pronouns are few. Most are built to the
inherited interrogativestemsko- and me- with fossilized suffixes similar or
identicalto thoseof the demonstrativepronouns.Thuscorrespondingto tot/tet
'therelhere'is kot 'where?',to toUtei is kof(faa) 'whence?',to teyaltiMg;, is
kot (mob) 'whither?'Furtherexamples:'who?' is kojayii, 'what?'is meyWii,
'someone'is kojiikaam, and 'something'is metiiiilii. 'How' is kotn;,; 'what?'
is kot;,; 'when?' is kuunt;,. (Note: constituentorder seemsto be decisivein
such minimal pairs as men-t-:;m kuunt;, 'when you go' : kuunt;, men-t-:;m
'whenareyou going?').
The numeralej 'one' and the noun stemkoo - kuu(j)- 'man' serveas the
basefor individuativepronounssuchas ej+kuuj-m:;m ONE+MAN-d1 'oneof
the threeof us'.
Negativepronounsare also built with ej, e.g. ej+metiiiitii-p(;,) 'nothing'.

The synchronicallymonomorphemiccardinalnumeralsare:

1 ej 6 kuut
2 kiiiit 7 tiiiip;,t
3 ko(o)l;Jm 8 riiihy
4 niei;, 9 jerjeelf/iirjeeJf
5 wiit 10 jeeJf

20 koos
100 saat
1,000 for;JS

'9' is from *ej+or.t-jeel)w (*)ONE-SHORT.OF-TEN(Honti 1984: 77). Other

cardinalsare formed by (1) addition, with the particle -or;,kk(;')-, e.g.jee{r-
or;}/ck-ej '11', (2) multiplication, e.g. koot;Jm+jeeJf '30', (3) subtraction,
with the elementiir (cf. '9', above),e.g. iir+saat '90', and (4) combinations
of these,e.g. iir++jeel]w +saat '900, i.e. one (hundred)lessthanten hundred',
jeel]w+or;}/Ck+iir+jeel]w '19, i.e. ten plus one (unit) less than ten'. Note also
riiit+saat '80'. The roots meaning'100' and '1,000' are Iranian loans; the
origin of '10' is obscure.The word for '20' is a core innovation sharedwith
Mansi, Hungarian,Permian,andperhapsMordva, andbuilt to a root meaning
'(hu)man',cf. koo 'man'.The word for '8' is either relatedto the Hungarian
 KHANTY 371

word rialaab 'bundle (originally, of hides)' or - less likely - is an eroded,

ablauted,dual of '4'. Multiplicatives are formed with the suffix =paa(/iiii) or
by compoundsbuilt with the roots piT/iic or pny~r, e.g. kiit=piiii 'twice',
koo/~pny~r, kool:»npiTc 'threetimes'.
Ordinal numeralsareformed from the cardinalsmoreor lessregularlywith
the suffix =mfJI~t, riel=m~t
e.g. 'fourth', wiit=m~ 'fifth', kuut=m~t 'sixth',
liiiip~t=m~t riirl~=m~t
'seventh', 'eighth'. 'Third' is eitherkoo/~m~ or
kuuim~t. aQl~1J
'First' is (also 'tip, beginning'); 'second' is kiim~t <
*kiit=m;)t. The translativeof ordinals forms ordinal multiplicative temporal
adverbs,e.g. kiim~t-k~ 'for a secondtime'. Fractionswith the denominator
'two' are madewith piili1/( 'half; the othersconsistof the formula (numerator
= cardinal)+ (denominator= ordinal) + jokiiiin 'part'.
Verb Conjugation
Khanty verb finite forms distinguishtwo moods (indicative and imperative)
and, dependingon dialect, from two to five tenses.Conditional,optative,and
conjunctivemoodsare constructedwith the help of particlesand/ordeverbal
nominals;seep. 375. As in Mansi and NorthernSamoyedic,different setsof
person suffixes, traditionally termed definite (or subjective) v. definite (or
objective)conjugations,refer to the definitenessv. indefiniteness,of the direct
object, as well as to its number(singularv. non-singular)or its absence(the
so-called passive). Tremjugan had three tenses: present -1-, past -13-, and
perfect -s-; in Tromagan,the -s- perfect seemsto have been lost. The far
easterndialects Vach and Vasjugan have innovated two additional tenses,
-yiiiil- and -yiiiis-, with discoursefunctions. The rest of the Khanty dialects,
like Tromagan,have only one pasttense:southerndialects andNizyam have
no perfect,and the dialectsfrom Sherkal northhaveno simple past. Futurity
is expressedby meansof derivation, e.g. inchoative=yt-, or with syntagms
of infinitive plus inceptiveverbs suchasje- or (in westerndialects)pit-, both
'begins'(Honti 1984: 49-50).
Tremjugan/I'romaganConjugation
As in all Khanty dialects,three formal classesof verbal stem types may be
distinguished:theclosedsetsof thematicandunstablestems,andtheopensetof
regularstems(Honti 1984: 35-6). The thematicstemsshow an alternationof
stem-finalaa (in the infinitive andpresenttense)- 13 (elsewhere),e.g. 101(aa)-
'melts', kol(aa)- 'leaks'.Most stemswith a in their longer alternantshow an
ablautwith 0 (or its prosodicequivalent,a1Jw, seeConsonants,p. 363) in their
shorterform; examplesarekalaa-- kol- 'dies',pa1J,caa-- po1J,c-'ripens',alaa-
- 01- 'sleeps',sasaa-- sos- 'dries (intr)', kanaa- - kon- 'adheres',la1Jaa- -
la1Jw- 'enters'. Othersuchrootskeepthea in all forms,e.g. lal( aa)- 'gets/makes
wet' ,pan(aa)- 'places,lays'. Unstablestemsalsohavelongerandshorterstem
alternants;in this casethe longer forms have an additional j (if the initial
consonantis a labial) or yW (otherwise)in imperative,perfect,andmostderived
forms. Examplesincludewe(j)- 'takes',me(j)- 'gives,endows',wuu(j)- 'sees',
372 KHANTY

nii(yW)- 'is visible', lii(yW)_ 'eats', and tuu(yW)- 'brings'. Note also the
bifurcatedpairjo(yW)- 'comes',je(y)- 'begins'.
Both thematic and unstablestemshave lexical counterpartsin Mansi, but
the prehistoryof both the phonologyand the morphologyof thesestemtypes
is still unclear(for thematicstems,seeHonti 1982: 104-8).

MakeupofIndicative Mood Forms

The suffix of the indicative mood is -0-, addeddirectly to the stem. Only the
indicative mood has morphologicallydistinct tenses.A tensesuffix is added
to the right of the indicative suffix -0-, and is itself immediatelyfollowed by
a person suffix from one of three series, termed traditionally after their
agreementpropertieswith direct objectsindefinite, singulardefinite, andnon-
singular definite. Alternatively, the tense suffix may be followed by the
passive suffix -oo(j)- - -ii-, which is then followed by a suffix from the
passiveseries.The different pre-zeroallomorphsof the passivesuffix (-00-
in the second-personsingular, -U- in the third person singular) reflect the
different historiesof thesepersonsuffixes: s2 -0 reflects a recentloss of final
*n in Surgut,while s3 -0 datesbackto proto-Khantyandbeyond.The Khanty
passivesuffix is notorious in the literature for its back roundedvocalism, a
vocalismwhich is not only immuneto vowel harmonybut overturnsthe front
vocalismof a word, e.g. backuu ... ~ after e .. .iiii in Vasjuganhistoric perfect
back-roundedness
we-yiiiil-uuj-~ TAKES-hist.perf-pass-sl'I was taken (long ago)'. The
staunch back-roundednessof its vowel is due to its recency: it is the
analogicallylevelled outcomeof earlier diphthongs(*aj, *iij) in which the *j
element continues a Uralic reflexive morphemeknown from Saamic and
Samoyedic,as well (Honti 1984: 52).
The person suffixes of the Khanty passiveseries are similar to, but not
identical with, those of the indefinite series.They are most different from
those of the singular definite series, a state of affairs which parallels that
found in TundraNenets,wherethe paradigmaticanalogueof the passiveis the
reflexive conjugation(cf. Ostrowski 1983: 17).
The personsuffixesof the four seriesaresetout in Figure 12.3 (the -y suffix
variantin thethird personsingularoccursonly in theperfecttense,andis obliga-
tory only with verbs in final ii- - :Jj-, e.g. lzar:Jj-0-"Y 'it broke', stem
aarii- ). Froma glanceat thetableit is obviousthatwith theexceptionof the suf-
fixes which marka secondpersonsingularsubject,all subjectclassessyncretize
to one degreeor another,and thereis evensyncretismof subjectperson(e.g.
-tt;m canbe secondor third persondual, as well as secondpersonplural).
A more protractedexaminationof Figure 12.3 will revealthat the first and
third personsubjectclassesare mirror imagesof one anotherin terms of the
failure to syncretize (see Figure 12.4). The singular first person has a
dedicatedsuffix for non-singulardefinite objects(-aam), but the non-singular
first persondoesnot: first persondual -m;m and plural -"Yw encodenot only
 KHANTY 373

Figure12.3 Trjffra Khanty indicativeverb suffixes

Definite object
Indefinite
Passive
object Singular I Non-singular

sl -;1m -eem -aam

S s2 -0 -;In -ee -aa
U s3 -t -t;lY -0
B
dl -m;ln -t;lm;ln -m;ln
J d2 -tt;ln -;In
E d3 -y;ln -tt;ln -;In
C
pI _;lYw _t;lYw _;lYw
T
p2 -t;lY -tt;ln -;In
p3 -t -iii -aai

Figure12.4 Selectedverbal personsuffixesin Trjffra Khanty

Sub'ect
First Third
singular I dual, plural singular I dual, plural

Passive -am yan,

Indefinite -man, -9 -at
Non-singular -aam _ayw -an,
definite -aai
Singular -eem -taman, -tay -tan,
definite _tayW -iii

non-singulardefinite object, but also indefinite and passive.With the third

personthe reverseis the case:it is the non-singularthird personwhich has a
dedicatedsuffix for non-singulardefinite objects (dual third-personsubject
-;;m, plural third-personsubject-aal), while in the singularthird-personnon-
singulardefinite objectsareencodedby -O, which alsomarksagreementin the
passiveandindefinite series.
This symmetryis perhapspart of the explanationfor the singulardefinite
suffixes which Trjffra has innovated, distinct from those of other Khanty
dialects,viz. dl -t;nn;;m, pI -t;)J!w; thesesuffixes were built with the addition
of a -t- element to the template of s3 definite singular -t;)J!, which has a
pendantin VVj.
Sampleconjugationalforms aregiven in Table 12.5.
374 KHANTY

Table12.5 Present-tense
indicativeforms of Tromaganpan- 'puts,places'

Objecttype Subjectperson
1 2 3

Singularsubject
Indefinite pan-i-;)ffi pan-i-;)n pan-;)f-0
Singular pan-i-eeffi pan-i-ee pan-i-;)t;)y
Dual pan-i-;)y;)i-aam pan-i-;)y;)f-aa pan-f-;)y;)i-0
Plural pan-i-;)i-aaffi pan-i-;)i-aa pan-i-;)i-0
Passive pan-i-ooj-;)ffi pan-i-oo-0 pan-i-ii-0

Dual subject
Indefinite pan-i-;)ffi;)n pan-i-;)tt;)n pan-i-;)y;)n
Singular pan-i-;)t;)ffi;)n pan-i-;)tt;)n pan-i-;)tt;)n
Dual pan-i-;)y;)i-;)ffi;)n pan-f-;)y;)f-;)n pan-i-;)y;)i-;)n
Plural pan-i-;)i-;)ffi;)n pan-i-;)i-;)n pan-i-;)i-;)n
Passive pan-i-ooj-ffi;)n pan-i-oo-tt;)n pan-i-ii-y;)n

Plural subject
Indefinite pan_i_;)yW pan-i-;)t;)y pan-i-;)t
Singular pan-i-;)t;)yW pan-f-;)tt;)n pan-i-iii
Dual pan-i-;)y;)i-;)yW pan-f-;)y;)i-;)n pan-i-;)y;)i-aai
Plural pan-i-;)i-;)yW pan-i-;)i-;)n pan-i-;)i-aai
Passive pan-i-ooj-;)yW pan-i-oo-t;)y pan-f-aa-t

Note: Adaptedfrom datacourtesyof M. Csepregi.

Imperative
It is useful to distinguish primary (second-personactive), secondary(non-
second-personactive), and tertiary (all passive) imperative forms. Only
southernand easterndialectshave secondaryand tertiary imperativeforms.
In primary and secondaryimperativeforms, the suffix of the imperativeis (1)
-iiii/aa- with the indefinite s2 suffix -0, (2) -0- with the definite s2 suffix -ee,
and (3) -ii- (- -;,j-) elsewhere;seeFigure 12.5.
The secondaryimperativeforms consistof stemplus imperativesuffix (as
given above), followed by the regular indicative suffixes, followed by the
enclitic --iiiitiaat, which also occurs as a free sentenceparticle meaning
'letCs)!' . Thus 'let me feed' is FEEDS-imp-sl--enc iiipt-ii-m-@--aat , 'let him!
her feed' is FEEDS-imp-s3--encliipt-;,j-@-@--aat, and 'let them(2) feed them
(2)' is FEEDS-imp-d3dO--enciiipt-ii-y;,l-aal--aat. The function of non-
singularfirst-personadhortativesis usuallyfilled by ordinaryindicativeforms.
Tertiary imperativesare formed by meansof a suffix -moos-- -muus-and
the enclitic --aat. Betweenthesetwo morphemesstand the indefinite verb
suffixes, e.g. liiiip;,t-muus--;Jm--aat'let me be fed', exceptin the third person
plural, wherethe personmarkeris unexpectedly-iiii-. The provenanceof the
suffix -moos-- -muus-is alsounknown(Honti 1984: 49).
 KHANTY 375

Figure 12.5 Trjffra Khanty primary imperativesuffix sequences

Indefinite Definite
Singular Non- singular
Dual Plural

s2 -iiii-0 -0-ee -iiy;)i-iiii -ii-i-iiii

d2 -ii-t;)y -ii-t;)n -ii-y;)i-;)n -ii-i-;)n
p2 -ii-t;)y -ii-t;)n -ii-y;)i-;)n -ii-i-;)n

Nonfinite Verb Fonns

It is difficult to distinguishinfinitive from presentparticiple in proto-Khanty,
and the two categoriesare scarcelydistinct in Trjffra, where they are built
with =taa (with an optional longer form, built with the translative suffix:
=taa-y;J) and =t;J (before personsuffixes: =taa), respectively.There is also
a pastparticiple built with =( ;J)m(;J) - =( ;J)maa-, and a conditionalparticiple
=I)aa.
The infinitive functionsas a complementto verbsof motion andto modals,
e.g. (*kenc-taa» kenc=aa(-y;J)men-I-;nn 'I'm going hunting'. The partici-
ples are often used adjectivally, e.g. waaj~ where wai=t;J taayii WILD.REIN-
DEER LIVES=pres.partPLACE 'place where the reindeerlives'. An agent
can be encodedas a personsuffix, as well, e.g. toyW-;;m men=maa-Itaayii
THERE-Ioc GOES=past.part-s3 PLACE 'placewhere s3 hasgone'.The past
participle is also frequentlyusedwith passivemeaning;in this casethe agent
is normally in the nominativeif the participle is usedattributively, and in the
locative if it is used predicatively: niiiij liiyw=;nn taayiT FIRE EATS=
past.partPLACE 'a placewhich hasbeenconsumedby fire', Kazym Khanty
xojaat-;;m waant=;nn 'seen by someone'(Honti 1984: 57). However, the
locative sometimesmarks the agentwhen the participle is usedattributively,
as well, e.g. Vasjugan Khanty apaa-l-nfJ mej=miiii way FATHER-Ioc
GIVES=past.partMONEY 'moneygiven by his father'.
Both presentand past participles can have temporal reference.The past
participle refers to past or previous time, as in wej=maa-y;J kiiiisaa-I-;nn
TAKES=past.part-transDOUBTS-pres-s1s0'I doubt that it was taken',
kor=;JI) waaj tol men=maa-I LEG=GED ANIMAL THENCE GOES=
past.part-s3'after the reindeer hadgonefrom there'.With personand various
casesuffixes, thepastparticipleconstructsequivalentsto subordinateclauses,
e.g. tiiiis;JI)-kfJ je=tiiii-niiiit taj-ee RICH-trans BECOMES=pres.part-approx
HAS-s2s0imp'keep it until you becomerich!' Compoundedwith the noun
kiim 'extent',the presentparticipleforms constructionswhich refer to ability,
e.g. laal=#;J+kiim taj-I-;nn STANDS=pres.part+ EXTENT HAS-pres-s1'I
can stand',laal=# ;J+kiim ent;nn 'I/yoU/(s)hecan'tstand'.
The conditional participle, used almost exclusively with personsuffixes
attached,rendersthe equivalentof a finite verb plus the adverbial kuuntfJ
376 KHANTY

'when/if, e.g. men-IJa-f GOES=cond.part-s3'if (s)he goes', cf. men-;;,UJ

kuuntfJ GOES-pres-s3'if (s)hegoes'.It is also usedwith postpositions.
The negativeparticiple usuallyhasimperfectiveaspect.It is madewith the
suffix =[;;,'1, which is formally identical with that of the abessivecaseof the
nominal paradigm. Examples: fOIJ-iin paat=l;ry tMy-iin paat=l;ry
SUMMER-temp FREEZES=neg.partWINTER-temp FREEZES=neg.part
'(remaining) unfrozen winter and summer', kad kool=l;ry koo WORD/
LANGUAGE HEARS=neg.partMAN 'obtuse man'. Unlike its nominal
analogue,the negativeparticiple frequently occurs with person (and case)
suffixes, e.g. wal=l;ry-aam-n;;, men-43-0 IS=neg.part-sl-locGOES-perf-s3
'(s)hewent while I was not (t)here'.
There is also a gerund,i.e. a verbal adverb,formed with full productivity
from verbs by meansof the suffix =miin. It takes neither casenor person
suffixes. Examples:nayw(iip=miin PaaP-43-0 LEANS=ger STANDS-pres-s3
'(s)he standsleaning', nieeyWreem fiiiip;Jt=miin iin;;,=mt:J-f-;]m CHILD.nom
FEEDS=gerBIG=vsx-pres-sl'I raisethe child by feeding (it)', puut jeIJk-;;'f
keeyW;;,r=miin t;;,r;]m-43-0 POT WATER-s3 BOILS=ger STOPS-perf-s3'the
waterin the pot boiled away (until it was all evaporated)'.
Non-finite verb forms frequently combineto form complexnoun phrases,
e.g. GERUND + INFINITIVE + PARTICIPLE in paat=miin piiyw=taay;;,
wiir=t;;, fiiiit-n;;, FREEZES=gerBURSTS=inf STARTS=pres.partTIME-loc
'when X is about to burst, it's so cold = when it's very cold'. For more
examples,seeSyntax,below.

Syntax
The Use of the Cases
The use of casesuffixes in Khanty to mark the participantsin actions and
eventsis most easily understoodnot in syntacticterms of subjectand object,
but rather in semantic terms of agent and patient. There are three basic
sentencetypes,active, ergative,andpassive:

1 In activesentences, the agentis in the nominative;any patientnounis also

in the nominative,but a patientpronounis in the accusative(-t); the verb
takesactive (i.e. non-passive)suffixes.
2 In ergative sentences,the agent is in the locative (-n;;,); the patient is
markedas in active sentences,and the verb takesactive suffixes.
3 In passivesentences,the patientis markedas in active sentences,and any
agentis markedasin ergativesentences;the verb takespassivesuffixes.

The NP agentsof passivesentencescan be, but are not typically, omitted;

the passive nature of a sentenceis in any case clear from the passive
morphologyof the verb. A noun-phrasesubject,i.e. the agentof active and
 KHANTY 377

the patientof passivesentences,can also alwaysbe omitted if it is clearfrom

the context; subjectpersonand numberare in any caseencodedon the verb.
Sincethe only formal differencebetweenactive andergativesentencesis the
marking of the subject,it is perhapsprudentto work with the hypothesisthat
the active/ergativedistinction is neutralizedin sentenceswithout explicit NP
subjects. Note that such neutralization never occurs with direct objects:
objects known from the context are normally omitted from a sentence,but
they arethen alwaysencodedformally on the verb.
Ergativesentencesare most frequentwhen the agentis in the third person
(Honti 1984: 94), but overall they aredecidedlyrare(Kulonen 1991: 189-91),
andthis hashinderedattemptsat elucidatingtheir function. The choiceof one
sentencetype over another seems to be discourse-driven,i.e. narrative
salienceand speaker'spoint of view are leading factors. In terms of topic-
focus grammar we may simplify, with Gulya (1970: 80-3), and say that
whereasthe definite-objectsuffixes on the verb signal that the direct object
is known, ergative(i.e. locative) case-markingon the subjectsignalsthat the
subjectis known. In termsof functional sentenceperspectivewe may simplify
by saying that while passive sentencesfocalize the patient, and active
sentences,ceteris paribus, are neutral with regard to focus, in ergative
sentencesneitherthe patientnor the agentis singledout, but the subjecthood
of the agentis highlighted.It hasbeenclaimedthat both the subjectsand the
objects of ergative sentencestend to be relatively high on the animacy and
intentionality hierarchies;this is a tendencyonly, however,and not a rule, as
shownby the examplesgiven below.
The following areexamplesof eachsubtypeas outlinedabove:

la Active sentence,definite (singular-object)verb: kiTnt-;;m;;m-l1 nOl) kii"nm-

ee BACKPACK-dl-sN PRO.s2PICKS.UP-s2s0imp'you pick up our
backpack!'
Ib Active sentence, indefinite verb: kajm-~/-l1 ken-I-;;m nok ken-I-;;m
GRAZING.AREA-s3-sN UP SEEKS-pres-sl'I'll look for its grazing
area'; ki"i"nt-l1 kiinm-aa BACKPACK-sN PICKS.UP-s2imp'pick up the
backpack'.
2 Passivesentence:wiin~-l1 ral)kaltal-l1-ii-0
kuul-n~ ral)kaltal-l1-ii-0 HOOK-sN FISH-Ioc
NIBBLES-pres-pass-s3 'the hook is beingnibbledby a fish'; (jey-;;m-n~)
ken-I-;;m
piistaa-y~ iiiirjal-l1-oo-0 (FATHER-sl-loc) SWIFT-trans CREATES-
perf-pass-s2'you were createdswift, my father madeyou swift when he
createdyou'; koo-n~ mej-0-ii-0 MAN-Ioc GIVES-perf-pass-s3'the man
gave(some[thing])'.
3a Ergative sentence,definite (singularobject) verb:ken-I-;;m kuul-n~ wiin;]p rial=
laay;] wiir-0-t~ 'the fish is beginning to bite at the hook', loyW-n;]
kor;rI-l1 toy;] oyw;]rk;;mt;rl1-t~ PRO.s3-loc LEG-s3-N THITHER
STRIKES.INVOLUNTARILY-perf-s3s0 '(s)he bumpedher leg against
it' .
378 KHANTY

3b Ergativesentence,indefinite-objectverb:VasjuganKhanty riiiit-n~ miiii-n

miiiin-
miiii-n
iiii me-s-0OLD.MAN-Ioc PRO.s1-latGIVES-perf-s3'the old man gave
me (something)'(cited by Kulonen 1991: 193).

Although passivesentencesof the classictype, i.e. in which the agentand

patient have syntactic roles opposite to those of their active-sentence
counterparts,are frequent,e.g. iiiimp-n~ miiii-n
por-s-ooj-;mzDOG-Ioc BITES-past-
pass-s1'I wasbitten by a dog', cf. Vachpor-s-iim niiilm-;mz BITES-past-s1s0
TONGUE-s1-sN'I bit my tongue',passivesentenceswith intransitiveverbs,
especially statives and locomotives, are also quite common, e.g. juuyW-n~
miiii-n
iinm-tJ-ii-0 TREE-loe GROWS-perf-pass-s3'a tree was growing', miiii-n
lii"I:=m~t-0-oo-0 PRO.s1-loc APPROACHES.STEALTHIL Y-perf-pass-s2
'I sneakedup on you'.
The pictureis only slightly complicatedby the fact that for many verbsthe
endsareexpressedby the means,i.e. they takedirect objectsin the instructive,
e.g. kuul-aat wiinqJ=I-~/-;mz FISH-inst ANGLES-pres-s1'I'm angling for
fish', riiiik-aat liiiiy/~ks;}-/-0 GRUEL-inst WAITS.AROUND-pres-s3
'(s)he'swaiting for gruel'. This use of the instructive is probably connected
with its appearance in benefactiveconstructions(seebelow).

Comitativeand Instructive
The comitative is used to express accompaniment,e.g. tje iimii-naat
'(s)helived with that/thewoman'. Both comitativeand instructive
instrumentals,
wall~;m
are usedas instrumentals,but the instructive has an additional nuance:with
ditransitive verbs it also expressesalienability from the agent(to the benefit
of the patient, if there is one). Examples: comitative/instrumentalbenefit
jayW~1
t'Ulhy-naatpayW ~/-taa wiir-0-ii-0 BOW TIP-comPOKES-infBEGINS-pres-

Iiiliiiip~-I-;m
pass-s3'(s)he beganto poke with the tip of (his/her) bow', instructive tom
niayw;}-jaat THEREITHEN MEAT-inst BEHOLD FEEDS-
pres-s2'then you'll feed (someone)with meat', moo nOIJ-aat me-I-;mzjem
jiirnaas-aat PRO.s1PRO.s2-accGIVES-pres-s1GOOD SHIRT-inst 'I'll give
you a good shirt (I'll endowyou with a good shirt)' (seeHonti 1984: 63,96).

Noun Phrase
The membersof noun phrasesshow no agreement,exceptthat numbermay
optionally be expressedformally on the head noun in noun phraseswhich
contain a numeral higher than ej '1', e.g. kiiiit nieeyWreem(-y;m)TWO
CHILD(-dual) 'two children', wiit wont(-~t)
higher FIVE HIGH.FOREST(-plur)
'five woodedridges'.
Unmodified nouns are co-ordinatedby simple juxtaposition(seebelow),
or, if they shareone or more semanticcomponents,by putting both in the
dual, e.g.siiyW~s-y;mkOr;]/C-k;m OSPREY-dualEAGLE-dual 'an ospreyand
an eagle'.
Noun phrasescan be linked in two basicways. One type of linkage forms
 KHANTY 379

possessive constructions;in these, the possessoris unmarked, and the

possessedis optionally in the third person, e.g. tor:wr iiiij pay(-;,I) GOD
YOUNG SON(-s3) 'God's young(est)son'. In the other type of linkage, the
two nouns are simply juxtaposed,e.g. lakw kdat HORSE HOUSE 'stable',
poom kaat HAY HOUSE 'house made of hay'. In Tremjugan Khanty, if
the fIrst noun was inflected for case or even followed by a postposition,
the two nouns were linked with the attributivizing particle -y;" e.g. ej liir-
n;,-y;, kdat-;,t ONE ROW-Ioc-y~
ROW-Ioc-y~ HOUSE-plur 'housesin a row' ,juuyW ont-
ii-y;, /eekw FOREST INTERIOR-abl-y~INTERIOR-abl-y~ PATH 'a path (leading) out of the
forest'. It is not clear whether or not this constructionis now extinct, but
in Tromagana noun in the locative can function as an attribute to a second
noun, e.g.#aar;,s-n;Jjeyk-aaSEA-Ioc WATER-Iat 'into the seawater' (Honti
1984: 87).
Direct-objectMarking
The formal encoding,onto the verb, of semanticandpragmaticfeaturesof the
direct object was outlined in the verb morphology section above. We may
summarizethe deploymentof this agreementsystembriefly as follows. A
direct object may be indefInite; or it may be defInite, either intrinsically
(unique global entities, e.g. the sky, names, deictics, nouns with person
suffIxes) or by virtue of its occurrenceearlier inthe discourse.In active and
ergativesentences,this defInitenessmay optionally be encodedon the fInite
verb. If the defInitenessis encoded,the non-singularnumber of the direct
objectmay optionally be encoded,as well. (Note: in northerndialectsnumber
agreementwith non-singularobjectsis obligatoryfor defInite verbs.)
For sentenceswith defInite direct objectsthereare thus threescenarios,A,
B, andC:

A B C

1 DefInitenessencodedon + +
verb
2 Non-singularnumber 0 +
encoded
Nominal andAdverbialPredicates
Sentenceswith nominal predicatesdisplay a rangeof formal alternatives.

1 In sentenceswith non-past reference, a nominal predicate in the

nominativecasemay combinewith the copulawos-, which has no tense
inflection and is zero in the third person,e.g. (miiii) tiilaay wos-:wr 'I'm
healthy',(lOyW) tiilaay '(s)he'shealthy'.
2 A2 Anominal predicatein the translativecasemay combinewith the copula
wal-, which takestensemarkersand is otherwisefully inflected, e.g. the
synonymous(miiii) tiilaay-b wal-I-:wr, (loyW) tiilaay-k;, wal-I-@.
380 KHANTY

3 Contaminationsof types 1 and 2 are also attested,e.g. Vasjuganmaa

mor;w-bwas-;;mz 'I'm healthy' (Honti 1993: 138).

In all cases,numberagreementbetweensubjectandpredicateis the norm, e.g.

pak-k~l-aam
aaj-y~'my two sonsareyoung'.Existentialsentencescan also
have zero verb, e.g. k(uln~l]-n~
oyW~r wont SHORE-IocTALL FOREST 'on
the shorethereis a tall forest'.
Sentenceswith adverbialpredicatesare characterizedby optional number
agreementbetween subject and predicate. The regular nominal number
suffixes are then added to the case form, e.g. miin kaat-naa-y~
PRO.d2
HOUSE-Ioc-dual 'we two are in the house',contrastnon-predicativekaat-
Major 'in two houses',with the reversesuffix order.
y~-n~

Major ConstituentOrder
Within the simple declarativeor interrogative sentence,the neutral major
constituentorder seemsto be SOY, no matterwhetherthe sentenceis active,
ergative,or passive.Sentencescanbe co-ordinatedor subordinatedby simple
juxtapositionor by meansof conjunctionssuchas paan 'and' or clitics such
as --p~.andAs is to be expected,topics are normally at the beginningof clauses,
andthe placefor focus is immediatelybeforethe verb, as in nOl] ayw-e-0jem-
and
y~ taj-l-e, maan-t waaj~l~-l-~ PRO.s2DAUGHTER-s2-sNGOOD-trans
HAS-pres-s2s0,PRO.sl-accMISTREATS-pres-s2'you love your daughter,
(but) me you mistreat' or nOl] kiim=~koo-n~kot~ war-l-ii-0 PRO.s2
EXTENT=adj MAN-Ioc WHAT? DOES-pres-pass-s3 'what can a man like
you do?'
Negationand Prohibition
Negation of constituentsor of the sentenceas a whole is effected by the
negative particle ent~,
e.g. ent~
taj-l-;;mz 'I don't have any'. Existenceis
~r
negatedwith ent;;mz, e.g. lal]wkw ent;;mz 'there is no mouse'.Prohibition is
expressedwith the modal negativeparticle aM plus the imperative,e.g. teem
riTt-aa aallel-aa THIS BOAT-Iat DON'T SIT-s2imp 'don't sit in this boat';
Trj Khanty had distinct prohibitive forms in all personsand numbers,e.g.
third personsingularpassiveprohibitive aM prohibitive 'may it not break'
aar~j-~j--aatprohibitive
(Honti 1984: 89).
Substitutesfor SubordinateClauses
Equivalentsto the subordinatedsentencesof other languagesare regularly
constructedwith the help of verbal nouns;many of thesehavebeenillustrated
abovein the sectionon non-finite verb forms. Here, two more exampleswill
illustratethe principle underlyingthis type of construction.

1 Equivalentsto 'if/when' clausescan be built as noun phraseswith the

locative of laat 'time' as head, modified by the participle of the main
verb. The participle, in tum, may have its agent explicit, e.g. teyW~t
 KHANTY 381

piiyW;;mz=t;} liiiit-n;} FIRE GIVES.OFF.SPARKS=pres.part TIME-Ioc

'when/if a/thefire gives off sparks'.
2 Another type of subordinateclause-equivalent,a temporalone, is made
up of the pastparticiple (plus any explicit agent)plus a postpositionsuch
as piTr-n~ 'after', e.g. iikii pii"'-n~
soocm=~ OLD.MAN LEAVES=
past.partBEHIND-Ioc 'after the old manleft'.

Alongside the noun-phrasesubstitutesfor subordinateclauses,in Trjtrra

Khanty we may distinguish at least two kinds of parallel construction
involving finite verb forms: (1) conditionals,which merely stateconditions,
built with kuunt~ kuunt~
'if/when', e.g. men-I-;;mz GOES-pres-sl'if I go',
kuunt~ARGUES-pres-d2
feeyw;}t-I-;}t;Jn IF 'if you two are going to argue',
and (2) conjunctives,which refer to contrary-to-factstatesand actions,built
with the particleol;}I], e.g. ol;}I] men-I-;;mz'I would go'.

Lexicon
The natureof Khanty (andMansi) traditionalpoeticdeviceshascreatedlexica
which are particularly rich in synonyms(e.g. SherkalsaI]k : maas 'sweat')
andparallelwords,both natural(Sherkaltil;}si 'moon':xatl 'sun')andcultural
(Sherkal taap;}t 'seven' : xuut 'six'). The vocabularypertainingto the bear
andthe bearfestival, as well as to otherspiritual matters,further supplements
the basic lexical stock with a large array of secret words and taboo
paraphrase.
The lexical stock is also enlargedby processesof derivation,only a small
samplingof which canbe touchedupon here.
Deverbal Derivation
In addition to the fully productive non-finite forms treated above, both
norninalsand, to a rathergreaterextent,verbsare formed from verbs.Among
the most common verb-forming suffixes are the frequentative/multi-
objectival =1- and =nt-, e.g. iiyw;}t- 'cuts',iiyw;}f=I_ 'cuts into many pieces',
kool- 'hears',kool=;Jnt- 'listens, obeys',the inchoative =m-, e.g. keeyw;}r-
'(water) boils (intr)" keeywr=;;mz- '(water) beginsto boil', the transitivizer/
causative =t-, e.g. keeyw;}r=t- 'cooks (tr)', and the momentaneous/
pusillanimous/repetitive=y-, e.g. kiil- 'rises', keel=;Jy- 'gets up for a little
while; gets up usually; makesa short side-trip'. Often two or more suffixes
sharea root morphemewhich does not occur otherwise(synchronically;cf.
HungarianandFinnishthemes),e.g.paas= paas=
~- 'drips once' : paas=;Jy- 'drips
repeatedly',soj=;;mz- 'spitsonce' : soj=;Jy- 'spitsrepeatedly'.
Verb-forming deverbal suffixes occur in clusters and chains at least as
frequently as they do singly. For example, alongsidethe simple derivates
keeyw;}r=t- and keeywr=;;mz- mentioned above there is also keeyWr=
;;mz=;}l=t;J- 'brings to the boil'. Further examples:laI]- 'enters',laI]=;}l=t-
'causes(animalsor people)to enter';wiil- 'kills', wiil=ii=l- 'kills on several
382 KHANTY

occasions'; uus=ii=p- 'yawns once', uus=ii=/- 'yawns (repeatedly)',

uus=ii=p="'!=;:J/- 'yawnsa little'.
With the exceptionof the non-finite verb fonns treatedabove, deverbal
nominal derivation is sparseand unproductive.Historically, many nounsare
secondaryfonnations built from verb stemswith =P;:J, =;:JP, =<7S and their
combinatoryvariants=p<7S and =pS;:J, e.g. ayW_ 'flows', ayw=;:Js 'lower stretch
of a river', /iiyW_ 'eats',/iiyw=p<7S 'food'.

Denominal Derivation
Suffixes which fonn verbsfrom nominalsare for the mostpart homophonous
with their deverbalcounterparts,e.g. tOf",! 'winter', to/y=;:Jm- 'it beginsto be
winter', /01) 'summer', fo1)=;:Jt- 'it is/will be summer',wiin;:JfJ 'fish-hook',
wiin;:Jp=/- 'angles'.
Denominal nominals,on the other hand, are both lexically and textually
frequent.Particularlysalientare the suffixeskuuf= =~1) and =p~ - kuuf=
kuuf= =~, which build
nomina possessoris and bahuvrihi compounds,respectively,e.g. kuuf= ~1)
kuuf=
'rich in fish', ej+siim=;:JfJ ONE+EYE=adj 'one-eyed'.The frequent use of
thesesuffixes is due,in part, to their role in the fonnationof circumlocutions,
e.g. kor;:J=1) 'havinglegs' (kOr;:J 'leg'), kOr;:J=1) w£itlj",! 'deer(leggedanimal)"
oyw;:Jr kor;:J=p ot LONG LEG=adj THING '(tabooword for) wolf, but also
becauseof their role in the fonnation of parallel lines of versein traditional
Khanty verbal art, e.g. (recited, i.e. not sung, version; from Csepregi1995:
272): si'is=;:J1) waaj",! kow jem siis-;:J/!miir;:J=1) waaj",! kow jem miir;r/ 'the
long(-lasting) good persistenceof the persistentanimal/the long(-lasting)
good perseverenceof the persevering animal'. There are also secular
circumlocutions,many of which are areal calques,e.g. 'mirror' is wiiiiti;:Jm
saaf=ii way FACE ISJUST.VISIBLE=adjMETAL, with parallelsin Mansi
and Selkup(Austerlitz 1990: 50).
Khanty (with Mansi) deviatesfrom the Uralic nonnin havinga verb which
means'to have',e.g. Trj taj(aa)- - toj-, SosvaMansi ootisi-. Neitherverb has
a credibleetymology,but it is likely that eachis a separateborrowing during
a period of bi- or multilingual convergencewith an autochthonouslanguage
or languagesnow lost.

Loanwords
It is difficult to distinguishloans from Iranian languagesinto proto-Khanty
after the breakupof ObUgrian from thoseborrowedinto proto-ObUgrian and
subsequentlylost in Mansi. An exampleof the fonner is probablypent 'path';
this item was also borrowed into Komi (cf. pad+veZ 'crossroads')and
Gennanic.With all due cautionone can speculate,on the basisof the number
anddistributionof Iranianloansin the two ObUgrianlanguages,that Iranian-
Mansi contactswere more intensive, longer lasting, or both, than Iranian-
Khanty contacts.If thereis truth in sucha speculation,it is echoedin the fact
that the numberof well-distributedKomi loansis greaterin Khanty than it is
 KHANTY 383

in Mansi (Toivonen,Redei); from this Salminen(1989: 19) has inferredthat

at the time of contactwith Komi speakers,speakersof pre-Khanty lived to the
north of speakersof pre-Mansi.Examplesof loans acquiredby both Khanty
and Mansi from or via Komi include iiniiiir 'saddle' (acquired by proto-
Permian from a Turkic languageof Chuvashtype), and the northern and
easternwords for 'cow', e.g. Obdorskmus,Trjffra mas(originally an Iranian
loan into proto-Permian).Komi loansattestedonly in Khanty include notonly
words limited to westerndialects,such as Obdorskkorti, Kazym karti 'iron'
or to southerndialects,suchas Demjankapuus'sieve' (cf. Yaz'va Komipuz),
but also widely distributeditems such as Trj and Demjankamet;}/(, Obdorsk
maz~x 'fist' (cf. Lower VychegdaKomi mi"fJk 'punch'),Vach riim;}/(, Nizjam
rim~x 'dark (cf. Komi rem 'dark(ness)').
The questionof the ageandextentof contactsbetweenspeakersof Khanty
and various Tungus languagesremainsopen; possibleTungus loans areare aj~
'luck', siiy(=)~ 'short brook betweenlakes', and iileem 'shame,shyness'
(Futaky 1975; Katz 1977).
From Turkic languagescome, e.g., Vasjuganjaarmaa 'barley grits' and
Demjanka muuy~t 'haycock'.A Turkic word for 'cow' was the sourceof the
word found in southern Khanty, e.g. Demjanka say~r (the word is also
attestedin non-northernMansi dialects).Note alsojiiiim 'postingstation'(via
south Khanty as early as the twelfth century, and ultimately from Chinese,
perhapsthroughMongolic, seeFutaky 1992: 18-20and Clauson1972: 933).
From Ket: k{uln~(J began'(edgeof) river bank'.
Loans from TundraNenetsare most numerousin the northerndialectsof
Khanty (and Mansi); the Khanty Surgut group has had closer contactswith
speakersof Forest Nenets (Sal 1976: 336). An example of a Selkup word
borrowedinto (eastern)Khanty is (Taz) Selkupaqsil '(mushroom)" cf. Vach
axs~l, began
Trjffra aks=~(J.
Russianloans beganto enter Khanty in the seventeenthcentury. The first
and strongestimpact was on southern and western dialects, and Russian
contacts have always been most intensive here, e.g. Krasnojarsk sufniiik
'onion' « qeCHoK), Nizjam ruupuutaa,Kazym roopaataa 'work' < pa60Ta,
Nizjam iisifeTJiiii, Demjankasfeniiii 'wall' < CTeHa,but many old loanshave
a wide distribution acrossthe dialects,e.g. Demjanka,Trj iiiir;mt, VVj iiiir;m,
NizjamRussian
i i T1J~, Obdorskiiiir;m 'debt' < apeH.n;a'lease'.In terms of semantic
fields, older Russianloans in Khanty resemble Latinloans into Old High
German (cf. Mauer 'wall', Tisch 'table', Zwiebel 'onion'); more recent,
especiallySoviet,loansresembletwentieth-centuryEnglishloansinto French
or Hungarian.

Tromagan Text
Excerptedfrom Honti 1984: 165.
384 KHANTY

A: segmentedtext, in transcription; B: morpheme-by-morpheme

gloss; C:
periphrase;0: metaphrase(free translation).

Al iikii-0 taj-;}i-0, iimii-0 wej-0-0

B 1 OLO(ER).MALE- HAS-pres-s3 OLO(ER).FEMALE TAKES-perf-s3
sN
Cl an older male he has, a woman he took
relative

A2 iimii-0 jek-k;}n afu;Jkee-y;}n taj-;}i-0

B2 OLO(ER).FEMALE FATHER-dual MOTHER-dual HAS-pres-s3
C2 (his) wife father- and-mother shehas

A3 uup-;}i + iikii-n;} afu;Jk=aayt;}=miin

B3 FATHER.IN.LAW-s3 + OLO(ER).MALE-Ioc NAGS=ger
C3 by his father-in-law naggingly

koi;} taj-i-ii-0 A4 n1im=t-0-ii-0

always HAS-pres-pass-s3 B4 NAME=vsx-perf-pass-s3
always he is had C4 he is called

sey+jel)k=ii 100ywtiaa=0 kon-0

BURBOT+WATER=adj SPLASHES=pres.part BELLY-sN
burbotsoup splashing belly,

iaar;}y+jel)k=ii 100yW tj aa=0 kon-0

RUFFE+WATER=adj SPLASHES=pres.part BELLY-sN
ruffe soup splashing belly.

A5 nom=s-;}i-aa ent;} joyw;}t-;}i-0

B5 THINKS=NdeV-s3-lat NEG.PART ARRIVES-pres-s3
C5 to his mind it doesn't come

01 He has a father-in-law, (because)he has taken a wife. 02 His wife

has a father and a mother. 03 He is constantlybeing naggedby his father-
in-law. 04 He is called burbot-soup-belly,ruffe-soup-belly.05 He doesn't
like it.

Referencesand Further Reading

Abondolo, D. (1996) Vowel Rotation in Uralic: Obug[rJocentric Evidence,SSEES
OccasionalPapersno. 31, London: Schoolof Slavonicand EastEuropeanStudies.
Austerlitz, R. (1958) Ob-ugric metrics: The Metrical Structureof Ostyakand Vogul
Folk-Poetry, Folklore Fellows Communications174, Helsinki: Finnish Academy
of Sciences.
 KHANTY 385

(1990) 'Contact, space, history', C7IFU, vol. IB (Sessionesplenares et

symposia),pp. 47-54.
Clauson, G. (1972) An Etymological Dictionary of Pre-thirteenth-centuryTurkish,
Oxford: Clarendon.
Csepregi, M. (1993) 'Az osszeferhetetleneger (Tromagani osztjak mese)', in M.
Bakr6-Nagyand E. Szij (eds), Hajdu Peter 70 eves[= Festschrifton the occasion
of P. Hajdu'sseventiethbirthday], LinguisticaSeriesA, Studiaet dissertationes15,
Budapest:HungarianLinguistics Institute, pp. 59-64.
- - - (1995) 'Surgutin hantien karhulaulujen kielenpiirteitli', C8IFU, vol. IV,
pp.271-8.
Futaky, I. (1975) TungusischeLehnworterdes Ostjakischen,Veroffentlichungender
SocietasUralo-Altaica 10, Wiesbaden:Societas Uralo-Altaica.
- - - (1992) 'EtymologischeBeitrage zu den obugrisch-russischen Sprachbezieh-
ungen 1', in L. Honti et al. (eds), Finnisch-ugrischeSprachen zwischendem
germanischenund dem slavischenSprachraum. Vortriige des Symposiumsaus
Anlaj3 des 25-jiihrigen Bestehensder Finnougristik an der Rijksuniversiteit
Groningen/2-15 November1991,Amsterdam-Atlanta:Rodopi, pp. 17-24.
Ganschow,G. (1962) 'Zur FragedesreduziertenAuslautsvokalsim Ostostjakischen',
UAJb 34: 4-7.
Gulya, J. (1966) Eastern OstyakChrestomathy,Indiana University Uralic and Altaic
Series51, Bloomington: IndianaUniversity.
- - - (1970) 'Aktiv, Ergativ und Passivim Vach-Ostjakischen',in W. Schlachter
(ed.), Symposion iiber Syntax der uratischen Sprachen 15-/8 Juli 1968 in
Reinhausenbei Gottingen,Gottingen:pp. 80-83.
Haarmann,H. (1974) Die Finnisch-ugrischenSprachen:Soziologischeundpolitische
Aspekteihrer Entwicklung,Hamburg:Helmut Buske.
Honti, L. (l982a) Geschichtedes obugrischenVokalismusder ersten Silbe, Bib-
liothecaUralica 6, Budapest:Akademiaikiad6.
- - - (1982b) 'VergleichendeAnalyseder Phonologieder nordlichenMundartender
obugrischenSprachen',FUF 44: 11-22.
- - - (1984) Chrestomathiaostiacica,Budapest:Tankonyvkiad6.
- - - (1988) 'Die ob-ugrischeSprachen,II.: die OstjakischeSprache',in D. Sinor
(ed.) The Uratic Languages: Description, History and Foreign Influences,
Handbuchder Orientalistik, 811, Leiden: Brill.
- - - (1993) 'A nevsz6i allitmany alaktanaaz osztjakban',in M. Bakr6-Nagy,and
E. Szij (eds), Hajdu Peter 70 eves [= Festschrifton the occasionof P. Hajdu's
seventiethbirthday], Linguistica SeriesA, Studia et dissertationes15, Budapest:
HungarianLinguistics Institute, pp. 135-42.
Katz, H. (1975) Generative Phonologie und phonologische Sprachbiinde des
Ostjakischenund Samojedischen,Finnisch-UgrischeBibliothek 1, Munich: Wil-
helm Fink.
- - (1977) Review of Futaky 1975, NyelvtudomdnyiKozlemenyek79: 1-2,
444-9.
- - (1987-8) Review of Honti 1982, Neprajz es Nyelvtudomtiny,Acta Uni-
versitatisSzegediensis de Attila J6zsefnominatae31-2, pp. 251-61.
Korenchy, E. (1972) lranische Lehnworterin den obugrischenSprachen,Budapest:
Akademiaikiad6.
Kulonen, U.-M. (1989) The passivein Ob-Ugrian, MSFOu 203, Helsinki: Societe
Finno-Ougrienne.
- - - (1991) 'Ober die ergativischenKonstruktionenim Ostostjakischen',JSFOu
83: 181-201.
Klinnap, A. (1993) 'Siperianuralilaisetkielettanaan',in T. Salminen(ed.), Uralilaiset
386 KHANTY

kielet tanaan,A-sarja 13, Kuopio: Snellman-Instituutti,pp. 85-8.

Ostrowski, M. (1983) Zur Nomen:Verb-Relationierung im Wogulischen,lurakischen
undlukagirischen,Arbeiten des Kainer Universalien-Projekts51.
Redei, K. (1970) Die syrjanischenLehnworterim Wogulischen,ResearchCenterfor
Language Sciences, Uralic and Altaic Series 109, Bloomington-The Hague:
IndianaUniversity-Mouton.
Sal, E. (1975) 'Uber die Stammendvokaleim Ostostjakischen',C3IFU, vol. I,
pp.674-6.
- - (1976) 'ITeKcllKa', chapter(pp. 332-9), with bibliographycompiledby E. Sal,
J. Gulya, and E.I. Rombandeeva(pp. 339-41)in V.I. Lytkin, K.E. Majtinskaii and
K. Redei (eds), OCHOBhI tPllHHo-yropcKoro H3bIK03HaHIDI. MapllHcKllH,
rrepMcKlie II yropcKlie H3bIKll, Moscow: USSRAcademyof Sciences.
Salminen, T. (1989) 'Classification of the Uralic languages',in R. Grunthal, S.
Penttinen and T. Salminen (eds.), IFUSCO 1988: Proceedings of the Fifth
International Finno-Ugrist Students' Conference, Helsinki, 22-26 May 1988,
Castrenianumintoimitteita 35, Helsinki: Snellman-Instituutti.
Sauer, G. (1968) 'Nominalstiimme auf *-a/*-ii im Ostjakischen',C2IFU, vol. I,
pp.459-61.
- - (1992) 'Zur Verbalpriifigierung im Ostjakischen',in P. Dereky, et al. (eds),
Festschriftfiir Karoly Redei zum 60. Geburtstag,Studia Uralica 6, Uralisztikai
tanulmanyok3, Linguistica SeriesA, Studiaet dissertationes8, Vienna-Budapest:
Institut fUr Finno-Ugristik der Universitiit Wien - MTA nyelvtudomanyiintezet,
pp. 399-402.
Schlachter,W. (1970) Symposioniiber Syntaxder uralischen Sprachen15-18 luli
1968in Reinhausenbei Gottingen,Gottingen.
Steinitz, W. (1950) Geschichte des ostjakischen Vokalismus, Berlin: Akademie-
Verlag.
Tereskin, N.I. (1981) CJIOBapb BOCTOqHO-XaHTbIHcKliX ;L\llaJIeKTOB, Leningrad:
Nauka.
Toivonen, Y.H. (1956) 'Dber die syrjiinischenLehnworterim Ostjakischen',FUF 32
1-169.

In addition, the following are invaluablelexical sources:

DEWOS = Steinitz, W., et al. (1996-93) Dialektologischesund etymologisches

Worterbuchder ostjakischenSprache,Berlin: AkademieVerlag.
KT = Karjalainen, K. F. (1948) OstjakischesWorterbuch, bearbeitet und her-
ausgegeben von Y. H. Toivonen,Lexica societatisFenno-ugricae10, vols I and II,
Helsinki: SocieteFinno-Ougrienne.
13 Mansi
LaszlOKeres'lJes

History of the Investigation into Mansi

The fIrst written attestationof Mansi is in the form of personalnames,found
sporadicallyin Russianchronicles.Interestin the Mansi languagedatesfrom
the eighteenthcentury, when travellersfrom various lands (Messerschmidt,
Strahlenberg,Fischer,Schl6zer,Klaproth, Pallas)beganto report vocabulary
they hadcollected.This early lexical evidenceis both linguistically naive and
of great philological importance;it must be used, therefore,with linguistic
judiciousnessandacumen.
The fIrst investigator to visit the Mansi for the express purpose of
collecting linguistic and ethnographicdata was Antal Reguly (1819-58;
fIeldwork 1843-4).At that time aboutfIfty Mansi still lived westof the Urals.
Reguly travelledeastof the Urals, as well, wherehe managedto makecontact
with an informant from along the Lozva River, from whom he collected
extremelyimportantmaterial.After Reguly,AugustAhlqvist (1826-89)made
three trips to the Mansi region, but managedto bring back relatively little
material.
Reguly died young. It fell to his contemporary,Pal Hunfalvy (1810-91),
to work up the texts Regulyhadcollected,somethinghe did with only limited
success.It was not until Bernat Munkacsi (1860-1937)that Mansi studies
were put on a fIrm footing. Munkacsi sortedout Reguly'smaterial, and in a
year of his own fIeldwork (1888-9) collectedfurther materialfrom all Mansi
dialects.The longeststretchof fIeldwork was carriedout by Artturi Kannisto
(1874-1943),who spent nearly six years (1901-6) among the Mansi, and
visited every areawhereMansi was spoken.
After the 1917Revolution,foreign accessto the Mansi waseffectively shut
off; betweenthe two world wars and after, the only foreigner permittedto
enter the region was Wolfgang Steinitz (1905-67). Fortunately, Soviet
linguists, as well, visited the area, among them v.I. Chernetsovand A.N.
Balandin. 'Western'investigators- amongwhom the most importantis Bela
Kalman - were allowed to travel no further than St Petersburg,where they
were able to collect fresh material from Mansi, studying at the Herzen
Institute. Other investigatorsfortunate enoughto carry out such fIeldwork
with native speakersincludeGy6rgy Lak6 andthe presentauthor.

387
388 MANSI

The first researcherto venture beyond St Petersburgafter the Second

World War was Eva Schmidt, who reachedthe Mansi lands by dint of
persistencein the face of natural adversityand bureaucraticobstacles.Since
the beginningof the 1990sit hasbecomepossibleonceagainfor researchers
from the Westto do Mansi fieldwork.
Among native-speakerresearchersof Mansi, M.P. Vakhrusheva,A.I.
SainakhovaandE.I. Rombandeeva deservespecialmention.At the endof the
twentieth century a new brigade of indigenousMansi languagescholarsis
beginningto testits wings.
Mansihasalwaysenjoyeda favouredpositionamongFinno-Ugric scholars.
This is evidencedin the first instanceby the attentionpaid it by scholarsin
Hungary(in the nineteenthcentury:Moric SzilasiandZoItanTrocsanyi;in the
first half of the twentieth century: Oliver Hazay, OdanBeke, David Fokos,
ZoItan Gombocz,Mano Kertesz,Antal Klemm, Dezs()SzabO,MagdaKavesi;
in the secondhalf of the twentiethcentury: MagdolnaKispal, Eva Sal, Odan
Lavotha, Karoly Redei, JanosGulya, Laszlo Honti, Eva Schmidtand Laszlo
Keresztes).But in Finland,too, prominentscholarshavedevotedtheir attention
to this languagewhich, for them,is only a distantrelative(Matti Liimola, Kaisa
Hakkinen,VuokkoEiras,Ulla-MaijaKulonen).
The work of Wolfgang Steinitz standsout among that done by scholars
whose native languagewas not a Finno-Ugrian one. BesidesSteinitz, one
should also mention the namesof Wolfgang Veenker,RobertAusterlitz, and
Giuliano Pirotti.

Mansi Linguistic Sources

Great caution must be used in exploiting the wordlists from eighteenth-
century expeditions.It was theseearly perceptions,however,which led the
first dedicatedlinguistic investigatorsto the Mansi regions. They returned
with extensive folkloristic texts. The chain of scholarly collectors was
unbroken, from Reguly to Hunfalvy (who made promising stabs at inter-
pretation)to BernatMunkacsi,who was the first to makea thoroughanalysis
and to publish the data. Munkacsi also did his own fieldwork; he beganto
work up the texts he had collected and part of a commentary,and after his
death Bela Kalman finished the commentary. By this time Kalman had
already beencollecting texts himself: he publishedthese,and worked their
lexical material into the corpus of the Mansi dialect dictionary basedon
Munkacsi'snotes.
Another important, albeit smaller, source for Mansi is the nineteenth-
century material collected by August Ahlqvist. The material collected by
Kannistohas beeneditedand published,for the most part, by Matti Liimola.
The most extensiveMansi dialect dictionary is still in preparation,underthe
direction of Vuokko Eiras.
E.I. Rombandeeva's monographs(1976, 1979) offer amplestorehouses of
examplesfor investigators.Sheis currently working on an edition of folklore
 MANSI 389

texts and on a dictionary. Schooltextbooksand university teachingmaterials

producedduring the Sovieteraare alsoilluminating.

The Developmentof the Mansi Languageand People

The Mansi - in their own language,maanjsii-live chiefly along rivers in the
areabetweenthe Urals andthe left bankof the Ob'. With its closestcongener,
Ostyak,it forms the so-calledObUgrianbranchofthe Uralic languagefamily.
The term Ugric is usedto refer to ObUgrianandHungariantogether.
After the speakersof proto-Hungarianbroke away(roughly seventhto fifth
century BCE), the linguistic ancestorsof the Khanty and the Mansi remained
in western Siberia, where they spoke a common language for some
considerabletime. Although linguistically they are the closestrelativesto the
Hungarians,in terms of physical anthropologythey are quite distinct. This
differentiationmay be due to geneticadmixturewith other,unknown,groups.
The distinction, in name,of 'Mansi' from 'Khanty' is datableto no earlier
than the fourteenthcentury. This is doubtlessdue to the great similarities in
the physical and spiritual culture of the two groups. The linguistic split,
however,musthavebeenmuchearlier,perhapsas early as the first half of the
first millenium CEo The period betweenthe breakawayof Hungarianand the
breakupof Khanty andMansi is calledthe ObUgrian period,andthe common
language,proto-ObUgrian(Honti 1982a:15).
During the ObUgrianperiod a significant numberof speakerslived to the
west of the Urals. The ethnonymVogul is first attestedin 1396: the Russians
thus namedthe first mansispeakerswith whom they cameinto contact,mansi
who were then living along the Vogulka River. The name then spreadto
designateall the Mansi. The self-designationmaanisii (western maanisi,
eastern ma:a:njsi, southern miinjdi) is perhaps an Iranian loan meaning
'human',but it is also connectedwith the nameof a Khanty/Mansiphratry
(Mansi moosi, Khanty Vakh maaniti , Kazym maasi) and with a Hungarian
tribal name (Megyer) and with the self-designationof the Hungarians
(magyar) as whole.
Between the tenth and fifteenth centuriesCE the northern and western
groups of Mansi speakerswere subject to Komi, then Russianinfluence;
southern and easterngroups experiencedintensive contact with Siberian
Tatar. The Russianstook western Siberia from the Tatars in the sixteenth
century; thereafterRussiancolonizationhit the Mansi populationhard. In the
eighteenthandnineteenthcenturiestherewerenew wavesof Komi migration;
theseKomi settledbetweenand among the Mansi, Khantak, and Samoyed
population.During their entire history, the Mansi have beenin contactwith
the Khanty, and northern groups of Mansi speakershave had contact with
Nenetsspeakers,as well (Honti 1982a:13-17; Kalman 1988: 395-412).
390 MANSI

Mansi Dialects
Mansi is spoken along the left-bank tributaries of the lower Ob', on the
Siberian side of the Urals. The dialects are traditionally classified into
Northern,Western,Easternand Southerngroups(Kalman 1976: 10-11).
Northern Group
Thesedialects,uponwhich the literary standardis based,arespokenalongthe
Sosva and Sygva Rivers, also formerly along the upper Lozva and in the
Berezovoregion (it is here that the Vogulka River is located).This grouping
of dialects is characterizedby strong Russian,Komi, and Nenetsinfluence.
The areais adjacentto that of northern Khanty, with which there has been
intense contact. Phonological innovations include: the backing of certain
vowels (e.g. *a > a, *aa > aa); geminationof intervocalicconsonants(upper
Sosva); 1JX > x and kW > k and k > t' before front vowels (Sygva); in Ob'
subdialects,Si insteadof si, and fricative laterals (I, P) insteadof approx-
imants (l, [1) - this latter feature a result of northern Khanty influence.
Throughout the Northern dialect area the accusative is unmarked, i.e.
homophonouswith the nominative. Dual forms of the direct object and of
possessions aredistinguished(Rombandeevea 1964: 51-8).
Western Group
Thesedialectswere formerly spokenalong the mid and lower reachesof the
Lozva, also along the Vagilsk and Pelymka.In Reguly'sday Westerngroup
dialectswere spokenwest of the Urals, as well (Chusovaya);they are now in
the final stagesof extinction. Russianand Komi influence was significant.
They are characterizedby a withering away of the dual, and,in commonwith
dialectsof the Easterngroup,by the dipthongizationof long vowels.
Eastern Group
Easterngroup dialectsare spokenby some100-200peoplealong the Konda
and YukondaRivers.Therewas closecontactwith Khanty speakersalong the
lower Konda;considerableTatarinfluenceis alsopresent.Vowel harmonyhas
been preserved.Word-initial s is more common in Yukonda subdialects.
Correspondingto Northernaa andSoutherniiii is the characteristica?a?, which
is frequentlydiphthongized.
Southern Group
Thesedialects were still spoken,along the Tavda River, at the beginningof
the twentieth century. Vowel harmony and the greaterpart of the original
vocalismare bestpreservedhere; on the otherhandthe dual hasdisappeared.
Thereis strongTatarinfluence.The main stressis on the secondsyllable.

Table 13.1 provides an overview of the chief dialectal isoglossesfor the

Northern,Western,Easternand Southern dialectgroups(Honti 1982a).
Therearedifferencesin the morphology,as well (cf. Kalman 1976: 12), as
shownin Table 13.2.
 MANSI 391

Table 13.1 Illustrationsof chief dialectalisoglosses

Gloss, Rontipageno. Northern Western Eastern Southern

grandfather123 aasj reresj aacji

daughter123 aayi aa rere aaw
dog 126 aamp aamp reremp aamp
pulls out 127 aal)xw- eel)kw- aal)xw- aal)k-
thin 130 sjaaliy sjaaliy sjrereli cjaal~w cjaal~w
hill 131 saal)kw sax W silk
sax
poursout l3l soos- soos- soos- saas-
flea 133 sus sus sons
w sos
fog l33 seel)kw seex seel)kw siikw
stone147 kaw kaw kaw kilw
dies 149 xool- kool- xool- kaal-
wave 150 xuump kup xop kop
claw 152 kWoss,kos klis kWas kilns
breast163 maay~lmaay~l mayl mreyl mawl
wades181 suus- suus- suus- soos-
cuts 183 saayr- sayr- sreyr- sawr-
heart 184 sim sim sim sam
hundred186 saat seet seet saat
god 190 toor~mtoor~m tooraamt toor~mtoor~m taar~m taar~m
fat 192 wooj wooj wooj waaj
lives, exists 193 001- 001- uul- aal-

Table 13.2 Samplemorphologicaldifferences

Northern Western Eastern Southern

Accusative -m(~) -m(~) -m -mi - -me

Translative -iy -a -j -0 --u
Ablative -ax
-n~l -n~-m(~) -n~l -naal - -naal
Dual -iy- -a- -aa-- -~j--aa-
Conjunctive H [+] [+] H
Future [-] [-] [-] [+]
Infinitive -I)kwe -ux -~x -ax
- -ax -I)
Pres.part. -ne
-m(~) -n~ -m(~) -p -ni - -ne

If alongside phonological and morphological differences we consider

lexical ones, as well, it becomesobvious that interdialectalcommunication
canbecomeproblematic(cf. Kalman 1976: 12).
392 MANSI

Table 13.3 Samplelexical differences

Northern Western Eastern Southern

[negator] at at u:u:t aa,ak

fire naaj, ulja taawt tu:u:wt taw;}t
small maani wiisj wisi musi
bread niaanj niaani nju:u:ni itimax
cow mis salr su:y;}r saw;}r
reindeer saali kunna xon;}j kOIJka
wolf saaliuj sees saas cjeesj

Phonology
Vowels
The vowel systemof Northern (Literary) Mansi is summarizedbelow (cf.
Rombandeeva1973: 17-28; Kruman 1984: 73--6):

11 e u uu
ee e ;) o 00
a aa

Here are the valuesof thesesymbolsin somephoneticdetail:

luI: short back roundedhigh: ut 'something',rusi 'Russian',laawuokwe'to

say'.
luu/: long backroundedhigh: uus 'town', suuj 'forest'.
Ii!: short front unroundedhigh: iO 'still', kit 'two', toti- 'brings'; retracted
variantsoccurafter x andbeforey (piy [p"iy] 'boy', xili- [xi1i] 'digs').
Iii!: long front unroundedhigh: iitii 'at night' ,jiiw 'tree'; retractedbefore y:
wUy;,r [wriy;)r] 'red'.
10/: shortbackroundedmid: niol 'nose',jom:JS'good',low 'ten'.
100/: long back rounded low [iiii]: ooli- 'is, lives', poor 'fish roe'. Often
diphthongizedas [mil.
leI: short front unroundedmid [E]. Occurs rarely in first syllable; example:
ness'only'. In non-first syllable it is more frequent,especiallyin absolute
final position: jaate 'its river', laawuokwe 'to say', luwkwel 'with a little
horse'.In non-first syllablesand when non-final, the e phonemeis usually
pronouncedlong. Velarized and diphongized variants also occur, e.g.
mineyen [mineeyeen] 'you (plur) go', taajen [taajeen] '(you plur) eat!
(imperative),minimen[minimeEn] 'let's go!'.
lee/: long front unroundedlower-mid [E:]. Examples: nee 'woman', keelp
'blood', seem;,l'black'.
Ia!: short back unroundedlow: aji- 'drinks', kan 'space',josa 'ski'. Fronted
variantsoccurafterj, e.g.janiy [jiin"iy] 'big'.
 MANSI 393

faa/: long back unroundedlow: aamp 'dog', taal 'winter', maa 'earth',saat
'hundred'.
I:J/: A reducedvowel, occurring in the literary languageonly in unstressed
syllables:joxt:Js 's3 came'. Beforem p, roundedvariantsoccur,e.g. toor;»n
[to:rum] 'sky, god', lilt;p [liltup] 'air'. In the environmentof palatalized
consonants[i]-like variantsoccur, e.g. aakani:J1 [a:kanjil] 'with a doll'; in
certain other environments[a]-like variants occur, e.g. xans:JIJ [xansaIJ]
'many-coloured',min:Js [minas] 's3 went'. Thesethree phonetic variants
of :J - [u], [i], and [a] - are indicated by the Cyrillic orthography,e.g.
TOpyM, JIHJITyrr, aKaHHJI, xaHcaHr,MHHac.

Oppositionsamongthe Vowels
The mostpervasivecorrelationin the vocalismis that of quantity.Witnessthe
pairs: sam 'eye' - saam'fish-scale;region', xot (perfectivizingcoverb)- xoot
'six; where?',tur 'voice' - tuur 'lake'. Qualitativedifferencesaccompanythe
quantitative distinctions; for example, short 0 is higher than its long
counterpart00, and short u is higher than its long counterpart uu. The
quantitative opposition is old; it is reconstructedfor proto-Mansi and for
proto-ObUgrian(cf. Steinitz 1955: 154; Honti 1982a:26).
In all surviving Mansi dialects,the correlationfront: backis irrelevant,i.e.
[+1-] backcanbe considereda trivial byproductof [+1-] labial.

Distribution o/Vowels
The phonologicallylong vowels tend to occur in the first, stressed,syllable;
contraste, which tendsto occur as (phonetically)long in non-first, non-final
syllables.Shortu and 0 arerare in unstressedsyllables;in word-final position
only a e i occur.The reducedvowel :J occursonly in unstressedsyllables.
In descendingorder of textual frequency,the vowels are a i :J 0 ee aa 00
u e uu ii. The proportion of back to front vowels has beenmeasuredas 66 to
34 (cf. Kalman 1976: 36).

Consonants
The consonantsysteminventory of North (Literary) Mansi is summarized
below (cf. Rombandeeva1973: 28-37; Kalman 1984: 76-9):

m n ni I)
pt r ~ kk kk
W

s sj x
w j y
r Ij
r

Hereare thevaluesof thesesymbolsin somephoneticdetail:

394 MANSI

1m!: voicedbilabial nasal:maam'my land'.

In!: voiceddentalnasal:naan 'you (plur)' .
Inj/: voicedpalatalizeddentalnasal:njaani 'bread'
ITJ/: voicedvelar nasalstop: distinctiveonly word-internalandword-final, e.g.
sia.(]sii 'sparrow',xuri.(] 'good-looking'.Frequent,in all subdialects,before
/k/ andlx!, e.g. paa.(]k 'soot', noo.(]x 'up'.
Ip/: voicelessbilabial stop: puut 'pot', tuup'oar'.
It/: voicelessdentalstop: taal 'winter', tuur 'lake'.
IV/: voicelesspalatalizeddentalstop: waatji 'short'.
/k/: voicelessvelar stop: kol 'dwelling', keer 'iron'.
/kw I: voiceless labialized velar stop: kW onkwaalu.(]kwe 'to exit', eekwa
'woman'.More frequentalong the Sosva;along the Sygva,replacedby /k/,
especiallyin initial position, e.g. kol 'dwelling', kon 'out'
Iw/: voiced bilabial fricative or approximant:wooj 'fat',jiw- - juw- 'comes'.
In somesubdialectsthereis word-internalfree variationof Iwl with Iyl, e.g.
puwi- - puyi- 'seizes';in others,Iwl alternatesfreely with zero in word-
initial position,e.g. Ob' uPa - wuPa 'fire'.
Ij/: voicedpalatalglide:joxti- 'arrives',wooj 'fat'.
Ix!: voicelessvelar fricative: xanisii- 'knows', xum 'man', nox 'up'; in some
(idio )lects, also occurslabialized,e.g. see.(]xw(- see.(]kW)'fog'.
Iy/: voiced velar fricative: janiy 'big', piy 'boy'. Does not occur word-
initially; cf. Iw/. Along the Ob', frequently correspondsto Iwl in other
areas,e.g. paay;;,l (other dialects: paaw;;,l) 'village', moyiniti- (mowinti-)
'laughs'.
lsi: voicelesshissingsibilant: saaw 'many',pos 'light, shining'.
lsi: voicelesshushingshibilant: occursonly in recentRussianloanwordsand
in Ob' subdialects,e.g. uuS(otherdialects:uus) 'town'.
Isj/: voicelesspalatalizedsibilant, varying subdialectallyand idiolectally with
a palatalizedhushingsibilant [sj]: siuni 'wealth',pasia 'greeting'.
/1/: voiced dental lateral approximant:lili 'breath,life', laawi- 'says'.Along
the Ob', as a result of Khanty influence, most words have a voiceless
lateralfricative (i).
/lj/: voicedpalatalizeddentallateral approximant:ljuuP 'bad';Along the Ob',
as a result of Khanty influence, most words have the corresponding
palatalizedvoicelesslateralfricative (~),e.g.e.g. PuuPi- 'stands'.
Ir/: voiced strongapical trill: roo.(]xi- 'shouts',xuri.(] 'good-looking'.

Oppositionsamongthe Consonants
An importantcharacteristicof the Mansi consonantsystemis the correlation
of palatalization,which is relevant for all the consonantshere classified as
dental, e.g. saaj 'spacebehind', siaaj 'tea'; sun 'sled', siuni 'wealth'; naan
'you (plur)', niaani 'bread'; paal 'side', paP 'ear'; waati- 'picks', waatji
j
'short time'. The palatalizedstop IVI is rare, althougha new t is commonin
Sygva subdialects,where a changeke, ki> tje, tji has occurred,e.g. Sy (kit
 MANSI 395

» tJit 'two', contrasttit 'here'.The soundsometimestranscribedas c', which

occursin a few loanwords,is to be interpretedas a sequenceof tJ plus ~, e.g.
'sewing-bagmadeof reindeerhide' (cf. Honti 1977: 412).
tutJ~a1)Mansi
No Mansi dialect knows a correlation of voice. The voicelessstops and
sibilants generally lack voiced counterparts,and the sonorantsand j lack
voicelessones.The only pair distinguishedsolely by the featureof voice are
the velar fricatives Ix! : Iy/; the functional load is slight, however, since Iyl
never occurs word-initially and these phonemesdo not appearto contrast
word-internally,either.
Quantitative oppositionsamong the consonantsare also few: geminates
occuronly at morphemeboundaries,e.g. witt 'in water' (wit 'water'),xass~xass~
'I wrote' (xansi- 'writes'), laawwes~laawwes~'they said to me =I was told' (laawi-
'says').
The contemporarynorthern (and western) dialects preservethe proto-
Mansi consonant system essentially unchanged, the only restructuring
stemming from the disaffrication of p V *ci and the realignment of the
sibilants(cf. Honti 1982a:24-5 andChapter11 in this book).
Distribution ofthe Consonants
Most consonantsoccurin all positions,with the exceptionof y and 1), which
do not occurword-initially. Mansi also tendsto avoid word-initial clusters;a
limited numberof word-final clustersare tolerated.
In descendingorder of textual frequency, the consonantsof Northern
Ii, ni, 1), tJ (cf. Kalman 1976: 36).
Mansi are: t, I, m, s, w, r, n, p, k, y, x,j,~,k, y, x,j,~,

Morpho(no )logy

The PhonologicalMakeup of Morphemes

Free Morphemes(Stems)
The canonicMansi word is bisyllabic, but monosyllabicandtrisyllabic words
are commonas well. The most frequenttypesof stemmorphemesare shown
in Table 13.4.
Nominal stems end in vowels, consonants,and consonantclusters, but
most verb stemshave a final consonant(ism).There are just six verb stems,
all monosyllabic (e.g. wi-, 'takes'), which have vowel-final forms before
certainsuffixes; but these,too, haveconsonant-finalallomorphs.
BoundMorphemes(Suffixes)
The most commonshapesof simple inflectional and derivationalsuffixes are
shownin Table 13.5.
The Syllable
The numberof syllablesin a Mansi word-form is conditionedby the number
of vowels present.Syllablesbegin with a consonantwheneverpossible,and
396 MANSI

Table 13.4 Canonicshapesof stemmorphemes

Noun Verb

ve ur 'mountain' uur- 'waits'

ev maa 'earth' Wl- 'takes'
eve tul 'cloud' laaw- 'says'
vev ulja 'fire'
vee oln 'money' uunl- 'sits'
veev oojka 'oldman'
evev saali 'reindeer'
evee suunt 'opening' roOI)X- 'shouts'
eveev soomji 'gold'

Table 13.5 Canonicshapesof inflectional andderivationalsuffixes

Nominal suffixes Verbal suffixes

V -e (s3 px) -i (s3 vx)

ve -iy (dual) -ew (pI vx)
vev -awe (pass)
veev -ul)kwe (int)
e -t (loc) -s (pret)
ev -te (s3 px) -te (s3 VX)
eve -n;}l (ab1) -nuw (cond)

the last element in any word-internal cluster belongs to the subsequent

syllable,e.g. molwallaalli 'laughs',roogixaxltely;)t 'they shout'.

MorpbonologicalAlternations

StemAlternations
Nominalstems.Nominal stemswhich endin a vowel do not alternate:suffixes
simply attachto the final vowel.
Consonant-finalnominal stems,on the other hand, are augmentedby the
reducedvowel ;;J or, rarely, by i when certain suffixes are attached:contrast
kol;rt 'dwellings' with kol-t 'in a/thedwelling' (kol 'dwelling'); xumi-te 'her
husband'with xum-n;;J1 'from a man'.
When vowel-initial suffixes are attached,an analogousloss of ;;J and i in
final syllables characterizesthe syncopatingclass of stems, e.g. xuur;;»n
'three', xuurm=it 'third'; sooj;;»n 'brook', soojm;rt 'brooks'; eeriy 'song',
eery-e'songs3'; laaY;;J1 'foot, leg', laayl-e 'foot/leg s3'.
The word maa 'earth' showsa unique alternationof y with zero, e.g. the
instrumentalcaseform maay;rl.
Verbal stems.There are no non-alternatingvowel-final verb stems. All
 MANSI 397

consonant-finalverb stemsare characterizedby the alternationof zero, ~, i,

and a to the left of various suffixes, e.g. zero inwaar-s-~ 'I made',waar-
nuw 's3 would make',waar-we-s's3 was made',and ~in waar;rs 's3 made',
waar;rs-te 's3 mades3', i in waari-l~ 'I make s3', waari-te 's3 makess3',
anda in waara-we 's3 is made'.
In verbalstemsendingin nasalplus homotopicobstruent(sibilant or x), the
nasalgoesto zero when consonant-initialsuffixes are attached(denasalizing
type): liuunisii- 'weeps', liuusi-s;rm 'I wept'; sunsi- 'looks', sus-s;rm 'I
looked';xaa1Jxi- 'climbs',xaax-s;rm'I climbed'.
Six verbal stemsare characterizedby alternationsof final consonants(y,
w, j) with zero, and, in somecases,by vowel alternationsas well. Table 13.6
sets out the stem variants of these verbs to the left of present,preterite,
conditional,imperative,andpassivesuffixation.

Table 13.6 Stem variants

Present Preterite, conditional Imperative,passive Gloss

miy- ml- maj- 'gives'

liy- li- laj- 'throws'
wiy- Wl- woj- 'takes'
teey- tee- taaj- 'eats'
waay- waa- waaj- 'sees'
juw- - jiw- ju- - ji- jaj- 'comes'

SuffixAlternations
Suffix allomorphy is dependenton whether the preceding segmentis a
consonantor a vowel (cf. Kalman 1976: 40).
The third-person singularpossessivesuffix is -te after vowels, -e after
consonants:aayi-te 'hislher daughter',piy-e 'hislher son'. A homophonous
pair is found in the third-personsingular suffix of the definite conjugation:
waari-te '(s)hemakesit', waari-jay-e '(s)hemakesthem (dual)', waari-jan-e
'(s)hemakesthem(plur)'.
The suffix of the instrumentalis -I in the singular and plural,pots'
-t~l in the
dual: puut;rl 'with a pot' ,puut;rt;rl 'with pots',puut-iy-t~l
diminutive 'with two pots'.
The distribution of the alternants of the diminutive suffix is unclear:
aayi=ke-m 'my (little) girl', aayi=kwe 'herlhislittle girl'.

VowelDistribution and Vowel Harmony

The present-daydialects of Mansi have lost vowel harmony.The Southern
(Tavda)dialectdatacollectedat the tum of the lastcenturyshowthat this form
of Mansi was archaicin this respect,e.g. Tavdakiil-niiiil 'from the/ahouse',
kaap-naal 'from the/aboat'; kiiiit-tiiii 'in the/ahand',maa-taa 'in the earth'
(cf. Honti 1975: 122).
398 MANSI

Stress
The primary stressfalls on the first syllable. There is also a secondary,but
considerable,stress on the third, but no stress falls on a final syllable.
Examples: lu "jyante 'y;]t 'they chirp', kaa"sala'san;]l 'they noticed itlhiml
her'. The Tavdadialectdeviatedfrom this pattern:it had stresson the second
syllable, apparentlyas a resultof Tatarinfluence.

Verbs

Stemtypes
Mansi verb stemsnormally endin one or two consonants.Stemaugmentation
comesin the form of an additional ;], i, or a; the selectionis conditionedby
the suffix. Only six stemshavevowel-final variants(seep. 397).

Conjugation
Active v. passiveforms are distinguished.All forms encodetense/mood.In
addition, passiveforms encodethe personand number of the experiencer,
while activeforms encodethe personandnumberof the subject.Active forms
further encodewhetheror not thereis a definite object;if the objectis definite,
its number is also encoded.Number can be singular, dual, or plural (cf.
Balandin-Vakhrusheva1957: 111-39; Rombandeeva1973: 111-45; Kalman
1976: 56-64).

Indeterminate(Indefinite, Subjective)Conjugation
Threemoodsaredistinguished:indicative,conditional,and imperative.
The indicative distinguishes two tenses: present and preterite. The
indeterminateperson-suffixesin the presenttenseare as follows:

Singular Dual Plural

1 -ey;)m -imen -ew
2 -ey;)n -ey;)/en -ey;)/en
3 -i -ey -ey;)t

The -ey- componentof certain of the person suffixes was originally a

present-tense morpheme(?pFU *-k-). The y andw final in the ablautingstems
appearto havealwaysbelongedto the stem,however.
Most of the fIrst- and second-personsuffixes derive from personal
pronouns. The -n of the second person, which has pendantsin Khanty,
Permian,and Nenets,is attributedby somescholarsto dialectal variation in
proto-Uralic, by others to a soundchange(*t > *n; cf. Hajdu 1966: 132-3,
142). In eitherevent,the distribution seemsto indicatean arealphenomenon.
In general,second-person forms - particularly dual and plural ones- tend to
syncretizein Mansi.
The third-person forms of the indeterminate conjugation are not of
 MANSI 399

pronominalorigin. The third-personsingular-i is historically the suffix of a

now non-productiveimperfective/continuousparticiple (pFU *=j). The six
ablauting verbs have zero in the third personsingular. The dual and plural
third-personforms consist of dual and plural suffixes (-y, -t) addedto the
stem;this stateof affairs is bestpreservedin the forms of the ablautingstems.
The first-person plural suffix -w is perhaps a continuation of a pFU
pluralizer *-k.
Future time is indicatedeither by present-tense forms, often with the aid
of a temporaladverb,or with the auxiliary verbpati-, e.g. eery-esujtugkW akw
tox pari 'hislher/itssongwill always(re)sound'.

Table13.7 Present-tenseindeterminatepersonsuffixesooli- 'is, lives', waari-

'makes',xansi-'writes'

Singular Dual Plural

1 ool-ey;}m ool-imen ool-ew

2 ool-ey;}n 00I-ey;)/en ool-ey;}/en
3 ool-i ool-ey ool-ey;}t
1 waar-ey;}m waar-imen waar-ew
2 waar-ey;}n waar-ey;}/en waar-ey;}/en
3 waar-i waar-ey waar-ey;}t
1 xans-ey;}m xans-imen xans-ew
2 xans-ey;}n xans-ey;}/en xans-ey;}/en
3 xans-i xans-ey xans-ey;}t

Table13.8 Present-tense
forms of the six ablautingstems

'gives' 'throws' 'takes' 'eats' 'sees' 'comes'

sl miy-;}m liy-;}m wiy-;}m teey-;}m waay-;}m juw-;}m

s2 miy-;}n liy-;}n wiy-;}n teey-;}n waay-;}n juw-;}n
s3 miy Iiy wiy teey waay juw - jiw
dl miy-men Iiy-men wiy-men teey-men waay-men juw-men
d2 miy-en Iiy-en wiy-en teey-en waay-en juw-en
d3 miy-iy Iiy-iy wiy-iy teey-iy waay-iy juw-iy
pI miy-uw Iiy-uw wiy-uw teey-uw waay-uw juw-uw
p2 miy-en Iiy-en wiy-en teey-en
waay-en waay-en juw-en
p3 miy-;}t Iiy-;}t wiy-;}t teey-;}t waay-;}t jUW-;}t

Note: SeeTable 13.6 for ablautingstems.

Thepreteriteforms arebuilt with the suffix -s « pU *-si), which is addedto

the plain (unaugmented)stemin all forms savethethird personsingular,where
either~ or (morerarely) a occurs.Thepersonsuffixesin thepreteriteare:
400 MANSI

Singular Dual Plural

1 -~men
-~m -uw
-~men
-~men

2 -iy
-~n -en-en
3 -0
-iy -iy-iy
-~t

The third-personsingularsuffix is always -0 in the preterite;the dual and

plural third-personforms are built with the dual and plural suffixes found in
the nominalparadigm.
Dual andplural syncretizein the secondperson.

Table13.9 Preterite-tense
personsuffixes:ooli- 'is, lives', mini- 'goes',sunsi-
'looks'

Singular Dual Plural

1 ool-s-en
ool-s-~m ool-s-~mensus-s-~men ool-s-uw
2 ool-s-en
ool-s-~n ool-s-en ool-s-en
3 ool-s-en
ool-(~)s-O ool-s-iy sus-s-en
ool-s-~t

1 ool-s-en
mina-s-~m mina-s-~men sus-s-~men mina-s-uw
2 ool-s-en
mina-s-~n mina-s-en mina-s-en
3 mina-s-O mina-s-iy sus-s-en
mina-s-~t

1 ool-s-en
sus-s-~m sus-s-~mensus-s-~men sus-s-uw
2 ool-s-en
sus-s-~n sus-s-en sus-s-en
3 ool-s-en
suns~-s-O sus-s-iy sus-s-en
sus-s-~t

Table13.10 Preterite-tense
forms of the six ablautingstems

'gave' 'threw' 'took' 'ate' 'saw' 'came'

sl mi-s-~mmi-s-~m li-s-~m li-s-~m wi-s-~mwi-s-~m tee-s~mtee-s~m waa-s-~mwaa-s-~m juw-~mjuw-~m

s2 mi-s-~n mi-s-~n li-s-~n li-s-~n wi-s-~n wi-s-~n tee-s-~n tee-s-~n waa-s-~n waa-s-~n juw-~njuw-~n
s3 mi-s-O li-s-O wi-s-O tee-s-O waa-s-O ji-s-O
dl mi-s-men li-s-men wi-s-men tee-s-men waa-s-men juw-men
d2 mi-s-en li-s-en wi-s-en tee-s-en waa-s-en juw-en
d3 mi-s-iy li-s-iy wi-s-iy tee-s-iy waa-s-iy juw-iy
pI ml-S-UW li-s-uw wi-s-uw tee-s-uw waa-s-uw juw-uw
p2 mi-s-en li-s-en wi-s-en tee-s-en waa-s-en juw-en
p3 mi-s-~t mi-s-~t li-s-en
li-s-~t li-s-en
wi-s-~t li-s-en
tee-s-~t li-s-en
waa-s-~t li-s-en
juw-~t

Conditional. The suffix of the conditional mood is -nuw- « pFU *-nek).

The personsuffixesin the conditionalare:
Singular Dual Plural
1 -en -amen -uw
2 -en -en -en
3 -en -iy -en
-~t
 MANSI 401

Table 13.11 Conditionalmoodpersonsuffixes:toti- 'takesaway,brings',mini-

'goes',xanli- 'glues',mi'Y- 'gives'

Singular Dual Plural

1 tot-nuw-gm tot-nuw-amen tot-nuw-uw

2 tot-nuw-gn tot-nuw-en tot-nuw-en
3 tot-nuw-0 tot-nuw-iy tot-nuw-;}t
1 min-nuw-gm min-nuw-amen mm-nuw-uw
2 min-nuw-;}n min-nuw-en min-nuw-en
3 min-nuw-0 min-nuw-iy min-nuw-gt
1 xanl;}-nuw-gm xanl;}-nuw-amen xanlg-nuw-uw
2 xanl;}-nuw-gn xanl;}-nuw-en xanlg-nuw-en
3 xanlg-nuw-0 xanlg-nuw-iy xanlg-nuw-;}t
1 mi-nuw-;}m mi-nuw-amen mi-nuw-uw
2 mi-nuw-gn mi-nuw-en ml-nuw-en
3 mi-nuw-0 mi-nuw-iy mi-nuw-gt

i.e., identicalwith thoseof the preteriteexceptfor the initial vowel in the first
persondual.
Somescholars(e.g. Kalman 1976: 56) use the term 'conditional', others
call this mood 'conjunctive' (e.g. Rombandeeva 1973: 128-9) or
'conditional-optative'(Riese1984: 75).
Conditions mayalso be expressedby meansof a particle ke, usedwith the
indicative,e.g.jo[;:,[ ke mini 'if (s)he/itgoesup there' (cf. Rombandeeva1973
129).
Imperative.The imperativeis formed by direct suffixation of the second-
personsuffix to the stem,i.e. its suffix is zero. This may preservean archaic
(pU) stateof affairs, or the imperative*-k- may havebeenlost throughsound
change(cf. Hajdu 21966: 136).
Examples:waari- 'makes',plus the ablautingverbs:

s2 waar-.m maj-gn laj-gn woj-:m taaj-.m waaj-:ln ji/aj-:ln

d2 waar-en maj-en laj-en woj-en taaj-en waaj-en ji/aj-en
p2 waar-en maj-en laj-en woj-en taaj-en waaj-en ji/aj-en

Imperativesfor the other personsare constructedby meansof a particle

(w)os plus the presentindicative,e.g. (w)os ujj-ey;Jm 'may I sink'.

Determinate(Definite, Objective)Conjugation
Fonnsof the detenninate(definite, objective)conjugationencodenot only the
personandnumberof the subjectbut alsothe numberof a direct objectwhich
is 'definite' (seep.417).
The detenninateconjugation has the same tenses and moods as the
indetenninate.The personsuffixes are shownin Table 13.12.
402 MANSI

Table 13.12 Determinateconjugationpersonsuffixes

Definite objectnumber
Singular Dual Plural

sl -l;}m -(j)ay;}m -(j)an;}m

s2 -l;}n -(j)ay;}n -(j)an(;}n)
s3 -te -(j)aye -(j)ane
dl -Iamen -(j)ay(a)men -(j)an(a)men
d2 -len -(j)ayen -(j)an(en)
d3 -ten -(j)ayen -(j)anen
pI -luw -(j)ayuw -(j)anuw
p2 -len -(j)ayen -(j)an(en)
p3 -an;}l -(j)ayan;}1 -(j)an(an);}l

Indicative forms of the determinateconjugationare shownin Table 13.13.

The paradigmsof xanli- 'causesto adhere' and miy- 'gives (away)' in
Table 13.14illustrate the preteriteforms of the definite conjugation.
The personsuffixes of the determinateconjugationare difficult to analyse
from both descriptive and historical points of view. In forms which encode

Table13.13 Determinateconjugationpresent-tense
forms: uunti- 'occupies(a
place)',tee'Y-'eats(up)'

Singular Dual Plural

Definite object Definite object Definite object

sl uunti-l;}m uunti-jay;}m uunti-jan;}m

s2 uunti-l;}n uunti-jay;}n uunti-jan(;}n)
s3 uunti-te uunti-jaye uunti-jane
dl uunti-Iamen uunti-jaymen uunti-janmen
d2 uunti-Ien uunti-jayen uunti-jan(en)
d3 uunti-ten uunti-jayen uunti-janen
pI uunti-Iuw uunti-jayuw uunti-januw
p2 uunti-Ien uunti-jayen uunti-jan(en)
p3 uunti-Ian;}l uunti-jayan;}l uunti-jan(an);}l

sl teey-l;}m teey-ay;}m teey-an;}m

s2 teey-l;}n teey-ay;}n teey-an(;}n)
s3 teey-te teey-aye teey-ane
dl teey-Iamen teey-aymen teey-anmen
d2 teey-Ien teey-ayen teey-an(en)
d3 teey-ten teey-ayen teey-anen
pI teey-Iuw teey-ayuw teey-anuw
p2 teey-Ien teey-ayen teey-an(en)
p3 teey-Ian;}l teey-ayan;}1 teey-an(an);}l
 MANSI 403

Table 13.14 Determinateconjugationpreteriteforms: xanli- 'causesto

adhere',miy- 'gives(away)'

Singular Dual Plural

Definite object Definite object Definite object

sl xanl~-sl~mxanl~-sl~m xanl~-say~mxanl~-say~m xanl~-san~mxanl~-san~m

s2 xanl~-sl~n xanl~-sl~n xanl~-say~nxanl~-say~n xanl~-san(~n) xanl~-san(~n)
s3 xanl~-ste xanl~-ste xanl~-saye xanl~-saye xanl~-sane xanl~-sane

dl xanl~-slamen
xanl~-slamen xanl~-saymen xanl~-sayen xanl~-sanmen xanl~-sanmen
d2 xanl~-slen xanl~-slen xanl~-sayenxanl~-sayen xanl~-san(xanl~-san(en)
d3 xanl~-sten xanl~-sten xanl~-sayenxanl~-sayen xanl~-sanen
xanl~-sanen

pI xanl~-sluwxanl~-sluw xanl~-sayuwxanl~-sayuw xanl~-sanuw

xanl~-sanuw
p2 xanl~-slen xanl~-slen xanl~-sayenxanl~-sayen xanl~-sanxanl~-san(en)
p3 xanl~-san~lxanl~-san~l xanl~-sayan~l
xanl~-sayan~l xanl~-san(an)~lxanl~-san(an)~l

sl mi-sl~m mi-sl~m mi-say~mmi-say~m mi-san~mmi-san~m

s2 mi-sl~n mi-sl~n mi-say~nmi-say~n mi-san(~n)mi-san(~n)
s3 mi-ste mi-saye mi-sane
dl mi-slamen mi-saymen mi-sanmen
d2 mi-slen mi-sayen mi-san(en)
d3 mi-sten mi-sayen mi-sanen
pI mi-sluw mi-sayuw mi-sanuw
p2 mi-slen mi-sayen mi-san(en)
p3 mi-sayen
mi-san~l mi-sayen
mi-sayan~l mi-sayen
mi-san(an)~l

fIrst- and second-personsubjects, a singular defInite object is generally

indicatedby the element-1-, which is of uncertainorigin; in analogousforms
with third-personsubject-marking,a singulardefInite object is indicatedby
-to, which descendsfrom eitherpU *se, a personal,or pU *tl:i, a demonstrative,
pronoun.Duality of the defInite objectis encodedby -(j )ay- « pU *-ka-/ka-),
andplurality by -(j )an- « pU *-n).
To the right of the elementwhich encodesthe numberof the defInite object
come the (subject) person suffIxes. Some of these - those of pronominal
origin - are identical to those found in the indeterminate conjugation.
Syncretisms,especially among second-personforms, are common here as
well. Throughout the defInite conjugation, third-person plural subject is
encodedby the historically obscuresuffIx -an;]l.

Passive
Formsof the passiveconjugationare all built to a stemformed with the suffIx
-(a)we-, which continuesa proto-Uralic reflexive-passivederivationalsuffIx
*=w. The personencodedin passiveforms is the experiencer;there is no
direct object and a fortiori no need to distinguish defInite from indefInite
objects.Passiveforms may be built to both transitiveandintransitiveverbs.
404 MANSI

Table 13.15 Conditional(present):waari- 'makes',li,,{- 'hits by throwing,

shoots(dead)'

Singular Dual Plural

Definite object Definite object Definite object

81 waar-nuwl::lm waar-nuwaY::lm waar-nuwan::lm

82 waar-nuwl::ln waar-nuwaY::ln waar-nuwan(en)
83 waar-nuwte waar-nuwaye waar-nuwane
dl waar-nuwlamen waar-nuwayamen waar-nuwanamen
d2 waar-nuwlen waar-nuwayen waar-nuwan(en)
d3 waar-nuwten waar-nuwayen waar-nuwanen
pi waar-nuwluw waar-nuwayen waar-nuwanen
p2 waar-nuwlen waar-nuwayen waar-nuwan(en)
p3 waar-nuwan::ll waar-nuwayan::ll waar-nuwan::ll

81 li-nuwl::lm li-nuwaY::lm li-nuwan::lm

82 li-nuwl::ln li-nuwaY::ln li-nuwan(en)
83 li-nuwte li-nuwaye li-nuwane
dl li-nuwlamen li-nuwayamen li-nuwanamen
d2 li-nuwlen li-nuwayen li-nuwan(en)
d3 li-nuwten li-nuwayen li-nuwanen
pi li-nuwluw li-nuwayuw li-nuwanuw
p2 li-nuwlen li-nuwayen li-nuwan(en)
p3 li-nuwan::ll li-nuwayan::ll li-nuwan::ll

In the Present conditional the w' s of the conditional and the passive
suffixes merge(-nuw-w- > -nuw-).

Table 13.16 Imperatives:toti- 'takesaway,brings',tee,,{- 'eats(up)'

Singular Dual Plural

Definite object Definite object Definite object

82 tot-eln tot-eY::ln tot-en

d2 tot-e1en tot-eyen tot-en
p2 tot-elen tot-eyen tot-en
82 taaj-eln taaj-eY::ln taaj-en
d2 taaj-e1en taaj-eyen taaj-en
p2 taaj-elen taaj-eyen taaj-en

Imperative.The imperativepassivecan be fonned periphrastically:(w)os

'let!' + the indicative presentpassive- e.g. naIJ wos alawen 'may you be
killed' (more literally: 'let + you.are.killed'),taw wos alawe 'may (s)he be
killed' (moreliterally: 'let + [s]he.is.killed').
 MANSI 405

Table 13.17 Indicative passive,presenttense:keeti- 'sends(away)',joxti-

'arrives',waa'Y- 'knows'

keeti- joxti- waay-

'sends(away)' 'arrives' 'knows'

sl keet-awem joxt-awem waa-wem

s2 keet-awen joxt-awen waa-wen
s3 keet-awe joxt-awe waa-we
dl keet-awemen joxt-awemen waa-wemen
d2 keet-awen joxt-awen waa-wen
d3 keet-awey joxt-awey waa-wey
pI keet-awew joxt-awew waa-wew
p2 keet-awen joxt-awen waa-wen
p3 keet-awet joxt-awet waa-wet

Table 13.18 Indicativepassive,preteritetense:loti- 'takesaway,brings',alisili-

'kills (game)',mi'Y-'gives'

toti- alisili- miy-

'takesaway, brings' 'kills (game)' 'gives'

sl tot-wes;}m alisil-awes;}m maj-wes;}m

s2 tot-wes;}n alisil-awes;}n maj-wes;}n
s3 tot-wes alisil-awes maj-wes
dl tot-wesamen alisil-awesamen maj-wesamen
d2 tot-wesen alisil-awesen maj-wesen
d3 tot-wesiy alisil-awesiy maj-wesiy
pI tot-wesuw alisil-awesuw maj-wesuw
p2 tot-wesen alis.il-awesen maj-wesen
p3 tot-wes;}t alisil-awes;}t maj-wes;}t

Occasionallysimple forms in the secondpersonsingular are also found,

e.g.

s2 miy-wen liy-wen wiy-wen tey-wen waay-wen

Narrative
Narrative verb forms express events not eyewitnessedby the speaker.
Translationsinclude 'allegedly', 'they say', 'it seems'.
The indicative presentis built with personalsuffixes from the continuous/
non-perfectiveparticiple. Its suffix is -ne- « pFU *-n, which formed nouns
from verbs; cf. pp. 414-15).In transitiveverbs,the active/passiveopposition
is neutralized: tot-ne-m can be either 'allegedly I bring' or 'allegedly I am
brought/tome is brought'. The categoriesof the definite object can also be
406 MANSI

Table 13.19 Presentconditional,passive:xanii!lti- 'teaches',mi-y- 'gives'

'teaches'
xanji~ti- miy-
'teaches' 'gives'

sl xaniisita-nuwem miy-nuwem
s2 xaniisita-nuwen miy-nuwen
s3 xaniisita-nuwe miy-nuwe
dl xaniisita-nuwemen miy-nuwemen
d2 xaniisita-nuwen miy-nuwen
d3 xaniisita-nuwey miy-nuwey
pI xaniisita-nuwew miy-nuwew
p2 xaniisita-nuwen miy-nuwen
p3 xaniisita-nuwet miy-nuwet

Table 13.20 Indicativepresent,narrativeform: por-yi- 'jumps',toti- 'brings'

(Definite) sO (Definite) dO (Definite) pO

sl pory;)-nem tot-nem tot-nel;)m tot-ney;)m tot-nen;)m

s2 pory;)-nen tot-nen tot-nel;)n tot-ney;)n tot-nen(;)n)
s3 pory;)-nete tot-nete tot-nete tot-neye tot-nene
dl pory;)-nemen tot-nemen tot-nelamen tot-neyamen tot-nenamen
d2 pory;)-nen tot-nen tot-nelen tot-neyen tot-nenen
d3 pory;)-neten tot-neten tot-neten tot-neyen tot-nenen
pI pory;)-new tot-new tot-neluw tot-neyuw tot-nenuw
p2 pory;)-nen tot-nen tot-nelen tot-ney;)nen tot-nenen
p3 pory;)-nen;)1 tot-nen;)1 tot-nen;)1 tot-neyan;)1 tot-nen(an);)1

expressed,as in tot-ne-l-;nn 'I bring himlher/it, allegedly'.

In the preterite narrative, the oppositionof voice is not neutralized.The
active indicative preteriteis built with personalsuffixes from the perfective
participle. Its suffix is -m- « pFU *=m, which formed nounsfrom verbs; cf.
p. 415). Therecanbe epenthesis,to the left ofthis suffix, of an :J, U, or a.
With personsuffixes, the numberof a definite direct object may also be
expressed,as in the present.
The passiveindicative narrativeis formed with personalsuffixes from the
gerund,whosesuffix is -ima- (cf. p. 415).

sl tot-imam 'I was brought,allegedly'

s2 tot-iman
s3 tot-ima
 MANSI 407

dl tot-imamen
d2 tot-iman
d3 tot-imay
pI tot-imaw
p2 tot-iman
p3 tot-imat

Table13.21 Indicativepreterite,narrativeform: ooli- 'is, lives', ali- 'kills'

ooli- ali-
'is, lives' 'kills'

sl ool-m;)m ala-m;)m
s2 ool-m;)n ala-m;)n
s3 ool-um ala-m
dl ool-mumen ala-mamen
d2 ool-men ala-men
d3 ool-miy ala-miy
pI ool-muw ala-muw
p2 ool-men ala-men
p3 ool-m;)t ala-m;)t

Table13.22 Personsuffixes,narrativeform: loti- 'brings'

sO dO pO

sl tot-aml;)m tot-may;)m tot-man;)m

s2 tot-aml;)n tot-may;)n tot-man;)n
s3 tot-am(t)e tot-maye tot-mane
dl tot-amIamen tot-mayamen tot-manamen
d2 tot-amlen tot-mayen tot-manen
d3 tot-am(t)en tot-mayen tot-manen
pI tot-amluw tot-mayuw tot-manuw
p2 tot-amlen tot-mayen tot-manen
p3 tot-aman;)l tot-mayan;)l tot-man(an);)l

Precative
Although somegrammars(e.g. Kalman 1976: 56-61) treat the precativeas a
mood, it is more accurate,on formal grounds,to distinguishtheseinteresting,
affect-ladenverb forms from the true moods(cf. Rombandeeva 1973: 176-8).
Formally, the precative suffixes lie somewherebetween inflection and
derivation; they are colouredby positive affect (-ke-, -kwe-; dialectally -tje-;
e.g. xaajtikWe '(s)he/it is running, the dear') or pity (-ris i -). To the right of
408 MANSI

these suffixes may occur temporal or modal suffixes, or both; the passive
occurs,as well (Honti 1977: 414-6).
Indicativepresent.Example:xaajti- 'runs'.

sl xaajtikem xaajtirisj;}m
s2 xaajtiken xaajtirisj;}m
xaajtiri~;}n
s3 xaajtikWe xaajtirisjen
xaajtiri~
dl xaajtikemen xaajtirisj;}men
d2 xaajtiken xaajtirisjen
d3 xaajtikew xaajtirisjiy
pI xaajtikew xaajtirisjuw
p2 xaajtiken xaajtirisjen
p3 xaajtiket xaajtirisjen
xaajtiri~;}t

Furtherexamples:teeytiem'I, cute little thing, am eating'; sunsiytiewes-

men 'we two deariesare being looked at'; sunsirisilas::nn 'I looked around,
poor me'; sunsiyrisiawem'I, miserablething, am being looked at'; woje-
risilen 'catchit, you wretch!'.

Verb Formation
Verbs may be formed from verbs or from nominals (Rombandeeva1973:
148-76;Kalman 1976: 55--6).

Table 13.23 Examplesof deverbalverb derivation

Suffix Function Origin Example Gloss

=axt- reflexive <pFU *=kt paajt=axti- cooks(intr)

=xat- reflexive <pFU *=kt lowt=xati- washes(refl)
=aj- momentaneous <pFU *=k por=aji- flashes
=iyl(aal)- frequentative suns=iyl(aal)i- looks around
=iyt- frequentative xaajt=iyti- runs around
=1- frequentative <pFU *=1 pee1;J=li- pokesat
causative xan=li- causesX to adhere
=Iaal- frequentative laawi=laali- keepssaying
=It- inchoative jeekw;J=lti- beginsto dance
causative oj;J=lti- putsto sleep
frequentative laaw;J=lti- mentions
=ltaxt- reflexive/inchoative rooIJxa=ltaxti- (suddenly)shouts
=m- momentaneous <pFU*=m muur=mi- sinks
=mt- momentaneous puw;J=mti- seizes
=ml- inchoative xaajt;J=mli- startsrunning
=nt- frequentative <pFU *=nt mina=nti- goesalong
=p- momentaneous <pFU*=pp sialta=pi- popsin
=s- frequentative <pFU*=Sl jala=si- goesaround
=t- causative <pFU *=tt jooIJx=ti- turns (tr)
 MANSI 409

Table13.24 Examplesof denominalverb derivation

Suffix Function Example Gloss

=m- becomesX janiymi- grows (janiy big)

kantmi- grows angry (kant anger)
=t- provideswith X ooxti- tars (oo(O)x resin, tar)
=1- becomesX sakwali- breaks(intr; sakw small piece)
aaygmli- becomesill (aaY;Jmsickness,
pain)
provideswith X xoramli- decorates(xoram decoration)
tinisiaI)li- lassoos
=ltt- transitiveact samgltti- notices(sameye)
=j- intransitiveact uulm=aji dreams(uul;JfTI dream)
namaji- names;offers (to gods)(nam
name)
=laxt- reciprocity pasialaxti- greetsgreets
(pa~ 'reciprocally')
rumalaxti- becomefriends (ruma friend)

Besides the more common derivational suffixes and suffix-sequences

(such as =iylaal-) there are also numerous nonce-combinations,e.g.
joo1Jxiypapti- 'turns on heel',joo1Jxataxtiyli- 'turns round and round (intr)',
juuntsaxti-,nom;]lmati- 'thinks, ponders',teesi;]flti- 'keepseating',etc.
Nominals

Nouns
The base form of nouns is the nominative singular. This can end in a
consonantor a vowel; amongthe lattergroup,nounsendingin i form a special
subclass.Thereis no grammaticalgender.
Absolutedeclension.Threenumbersare distinguished:singular,dual, and
plural (Balandin-Vaxruseva1957: 50--72; Rombandeeva1973: 38--64; Kal-
man 1976: 41-7).
Singularis unmarked.The suffix of the dual is -y « pU *-kA); that of the
plural is -t « pU *-t). Examples with a vowel-final stem: eekWay 'two
women', eekWat 'women'. Stemsin final i show a hiatus-blockingj in the
dual: aayi 'girl', aayijiy 'two girls'. Consonant-finalstemsareextended,e.g.
luw 'horse',luwiy 'two horses',luw;]t 'horses'.Syncopatingstemslose their
stem-internalvowel, e.g. siax;]l 'pile, heap', siaxliy 'two heaps', siaxl;]t
'heaps'.In northerndialects,reduplicatedforms arealso attested.
NorthernMansi noundeclensiondistinguishessix cases:
Nominative -0 (= Accusative,Genitive)
Locative -(;J)t <pFU *-ttA
Lative -(;J)n < pUg *-na-k/j
Elative/ablative -n;Jl <pUg *-na-l
Instrumental -(;J)1/-t;J1 <pUg *-1
Translative/essive -iy <pU*-kA
410 MANSI

The translative/essiveis not frequently usedin the dual and plural. The -t-
ofthe instrumentalis the result ofJaussecoupein the plural. Certainnorthern
dialects also have a vocative in -aa (Liimola 1963: 24-131; Riese 1992:
379-88).
Possessivedeclension. Suffixes expressthe person and number of the
possessor,and the numberof the possession;seeTable 13.26.
Consonant-finalstemstake the variantswith initial a (or e); i-final stems
take the j-initial variants, and stems ending in other vowels take the
consonant-initialvariants.
Personalsuffixesprecedecasesuffixes (example:puut 'pot'):

NING Gloss Loc Lat ElalAbl Ins

puut;)m my pot puut;)mt puut;)mn puut;)mn;)l puut;)mt;)l

puutay;)m my 2 pots puutay;)mt puutay;)mn puutay;)mn;)l puutay;)mt;)l
puutan;)m my 3+ pots puutan;)mt puutan;)mn puutan;)mn;)l puutan;)mt;)l

Table13.25 Noun declension:ala 'roof',aawi 'door',puut'pot',paart 'board',

iaa-yai 'foot, leg'

ala aawi puut paart paart

laay~l
'roof 'door' 'pot' 'board' 'foot, leg'

Singular
NIAIG ala aawi puut paart laaY::lI
Loc alat aawit puutt paart::lt laaY::llt
Lat alan aawin puutn paart::ln laaY::lln
ElalAbl alan::ll aawin::ll puutn::ll paartn::ll laaY::lln::l1
Ins alaI aawil puut::ll paart::ll laayl::ll
Trans/Ess alay aawijiy puutiy paartiy laayliy
Dual
NIAIG alay aawijiy puutiy paartiy laayliy
Loc alayt aawijiyt puutiyt paartiyt laayliyt
Lat alayn aawijiyn puutiyn paartiyn laayliyn
ElalAbl alayn::ll aawijiyn::l! puutiyn::ll paartiyn::ll laayliyn::ll
Ins alayt::ll aawijiyt::ll puutiyt::ll paartiyt::ll laayliyt::l!
Trans/Ess
Plural
NIAIG alat aawit puut::lt paart::lt laayl::lt
Loc alatt aawitt puut::ltt paart::ltt laayl::ltt
Lat alatn aawitn puut::ltn paart::ltn laayl::ltn
ElalAbl alatn::ll aawitn::ll puut::ltn::ll paart::ltn::ll laayl::ltn::ll
Ins alat::l! aawit::ll puut::lt::ll paart::lt::ll laayl::lt::ll
Trans/Ess
 MANSI 411

Table 13.26 Possessive

declension

Possessionin singular Possessionin dual Possessionin plural

sl -(;})m -(Ola)y;}m -(Ola)n;}m

s2 -(;})n -(Ola)y;}n -(Ola)n
s3 -(t)e -(Ola)ye -(Ola)ne
dl -men -(Ola)yamen -(Ola)namen
d2 -(e)n -(Ola)yen -(Ola)nen
d3 -(t)en -(Ola)yen -(Ola)nen
pI -(Olu)w -(Ola)yuw -(Ola)nuw
p2 -(e)n -(Ola)yen -(Ola)nen
p3 -(Ola)n;}i -(Ola)yan;}i -(Ola)n(an);}i

declensionstemvariants:ula 'bow',anyi 'girl'

Table13.27 Possessive

Possessionin Possessionin
Singular Dual Plural Singular Dual Plural

sl uiam uiay;}m uian;}m aayim aayijay;}m aayijan;}m

s2 uian ulay;}n ulan;}n aayin aayijay;}n aayijan;}n
s3 ulate ulaye ulane aayite aayijaye aayijane
dl ulamen ulayamen ulanamen aayimen aayijayamen aayijanamen
d2 ulan ulayen ulanen aayin aayijayen aayijanen
d3 ulaten ulayen ulanen aayiten aayijayen aayijanen
pI ulaw ulayuw ulanuw aayijuw aayijayuw aayijanuw
p2 ulan ulayen ulanen aayin aayijayen aayijanen
p3 ulan;}l ulayan;}l ulan(an);}l aayijan;}l aayijayan;}l aayijan(an);}l

In general,the personalsuffixes are pronominal in origin; the dual and

plural suffixes arealsoof FU or U origin (-m -n -te < pU *-me, *-te, *-se; dual
-y- < pU *-kA-, etc.).
The personal suffixes in the noun are similar to those of the definite
conjugation of the verb; they are of a common origin (cf. p.402; Liimola
1963: 202-41).

Adjectives
Adjectival declension.In attributive position, adjectivesremain uninflected.
In predicativeposition,adjectivesagreewith their subjectin number,andthus
take dual and plural suffixes, e.g. towl-ay-;nn osisia-yWING-dual-s1THIN-
dual 'my (two) wings are thin', laayl-an-;an xosa-tLEG-plur-s1 LONG-plur
'my legs are long'. The translativecaseis frequent, e.g. xoopsimmos<11J-iy
jeemt<1S'my lung becameill' (mos<11Jill; Rombandeeva1973: 82).
Comparativeand superlative.There is a comparativesuffix -nuw, e.g. ta
412 MANSI

xum kar;;,s-nuw 'that man is taller', but in full-blown comparisonsthe plain

form is at leastas usual.The bestedmemberof the comparisonis then in the
ablative, e.g. nalJ aa~-;;nn-n;;,lKalman janiy PRO.s2FATHER-s1-ablBIG 'you are
bigger than my father' (cf. Kalman 1976: 48). The superlativeis renderedby
particles,e.g. am ~ar
kar;;,s kar;;,s s1.PROPARTIC TALL 'I am the tallest'.
Numerals
Cardinalnumeralsare as follows:

1 akW(a) 20 xus
2 kit(iy) 21 waatnop<ll akwa
3 xuur<lm 30 waat
4 njila 40 naliman
5 at 50 atpan
6 xoot 60 xootpan
7 saat 70 saatlow
8 njololow 80 njolsaat
9 ontolow 90 ont<lrsaat
10 low 100 (janiy) saat
11 akwxujplow 1000 soot<lr

The non-primary numeralsare built either by compounding,e.g. '95' is

ont;;,rsaat at, '991' is ontolowsaatont;;,rsaatakwa, or by 'adding' the smaller
numberto the larger. The largernumeralis then put into someform of lative,
either the lative case, e.g. waat-n akwa THIRTY-Iat ONE '21 (i.e., one
towardsthirty), or by meansof the postpositionnop;;,l 'towards',e.g. naliman
nop;;,l akwa FORTY TOWARDS ONE '31 (i.e. one towardsthirty). Numerals
occur most frequently in the nominative, translative, and instrumental
(Rombandeeva1973: 91-8; Kalman 1976: 48-9). The numeralsfor 1-6, 10,
20, and 100 areFU in origin; '7', '8', and '1,000' datefrom the Ugric period,
as do someof the now synchronicallyopaquedecades.
Ordinal numerals. 'First' is oow;;,l, 'second'is moot or kit=it. All other
ordinals are formed, like kit=it, from the correspondingcardinal with the
suffix =it, e.g. xuurm=it 'third', saat=it 'seventh',low=it 'tenth'.
Pronouns
Table 13.28 attempts to capture the formal system of Mansi personal
pronouns(Com =comitative).
There is also an intensivizing clitic, -ki (-tji) , which is particular to the
personal pronouns, e.g. amki 'I myself, neenki 'you two yourselves'
(Rombandeeva 1973: 103-5).
The intensive forms of the personal pronouns are used as reflexive
pronouns, e.g. accusative amkinaam 'myself, takWinaate 'himlherself,
dative amki(naa)mn 'to myself, takWi(naa)ten 'to himlherself. For the
declensionoftheseforms seeRombandeeva1973: 103-4.
 MANSI 413

Table 13.28 ~ansipersonalpronouns ~ansipersonalpronouns

Person
First Second Third

Nom am nalJ taw

Ace aan;}m nalJ;}n tawe
Dat aan;}mn nalJ;}nn tawen
Ab! aan;}mn;}! nalJ;}nn;}! tawen;}!
Com aan;}mt;}! nalJ;}nt;}! tawet;}!
Dual
Nom meen neen teen
Ace meen(a)men neenan teen(a)ten
Dat meen(a)menn neenann teen(a)tenn
Ab! meen(a)menn;}! neenan;}nn;}! teen(a)tenn;}!
Com meen(a)ment;}! neenant;}! teen(a)tent;}!
Plural
Nom maan naan taan
Ace maana(nu)w naanan(en) taanan(an);}!
Dat maana(nu)wn naanann taanan(an);}!n
Ab! maana(nu)wn;}! naanan(;}n)n;}! taanan(an);}!n;}!
Com maana(nu)wt;}] naanan(;}n)t;}! taanan(an);}!t;}!

Demonstrativepronounsmay be subdividedinto proximal and distal sets,

e.g. tiji 'this', ta(ji) 'that'; akWti 'this same(one)', akWta 'that same(one)';
tamlie, tixurip 'this kind of', taxurip 'thatkind of' (Kfumin 1976: 51).
Interrogative and relative pronounsmay distinguish animatefrom inan-
imate, e.g. xoo1Jxa'who?',man;)r 'what?'Othersincludemana(x) 'what kind
of?', mansaawit 'how many?'. They decline like nouns, xoo1Jxa-t;)1 'with
whom?', man;)r-n;)l 'from what?' (Rombandeeeva1973: 107-8; Kalman
1976: 51). The interrogativepronounsare also usedas relative pronouns.
Indefinite and negativepronounsinclude forms suchasxootpa 'someone',
neemxootpa 'no one', xotixootpa 'no matter who'; mat;)r 'something',
neemat;)r 'nothing', mat;)rsirmat 'somekind of', neemat;)rsir 'of no kind',
mataxkem'somewhat'.
Nominal Formation
Examplesof denominalderivationof nominals:

Noun-formingsuffixes:
=it abstracts <pFU *=tt xos=it length
=k(W)e diminutive/affectionate <pU *=kk piy=kwe little boy
=risj diminutive/pejorative piy=risj little boy

Certainroots occurso frequently as the secondmemberof compoundsthat

they begin to take on derivational status,e.g. kol 'house;place' in pumkol
414 MANSI

'grassland'(pum grass),ut 'thing' in teenut'food' (teen eating), sakw 'small

piece' injiwsakW 'splinter' (jiw tree,wood).

Adjective-formingsuffixes:
=i position <pU *=j num=i upper
jol=i lower
=kapaj augmentative kapaj 'giant' mil=kapaj very deep
=kwe diminutive <pU *=kk FuuF=kwe pretty bad
=tal privative <pFU=pt= way=tal weak(way
strength)
paF=tal deaf(paP ear)
=1) providedwith X <pU *=1) xuul=g rich in fish
wit=;;}g wet, watery
=p(a) bahuvrhicompounds< pU *=p wit=q:> -watered
xuul=pa -fished

Numeral-formingsuffix:
=it ordinals < ?pU *=mt nom=t
~il=it fourth

Examplesof deverbalderivationof nominals:

=m abstract = perf. mat=;;Jm old age
participle (mat- ages)
=p nomeninstrumenti <pU *=p saayra=p axe (saayri-
cuts)
=t nomenresultatis <pU *=tt nom=t thought
(nomi- thinks)
=tal privative <pFU*=pt= joxt=tal inaccessible
(joxt- arrives)

Verbal Nouns
Infinitive, participle, and gerund are distinguishedin Mansi (Rombandeeva
1973: 145-7; Kalman 1976: 62).

Infinitive
The suffix is -(u/a)gkwe.The initial vowel is u with verbswhosestemhas an
odd number of syllables,a with verbs whose stem has an even number of
syllables. The consonant-initial variant, -gkWe, occurs only with the six
unstablestems(seeTable 13.6).Examples:laaw-ugkWe'to say',pot;;}rt-agkWe
'to speak',tee-gkWe 'to eat'.

Participles
The present(imperfective)participle suffix is -ne « pU derivational *=n +
?*-k-*-j lative; the participial role is a Mansi innovation(Sz. Kispal1966:68,
 MANSI 415

185). This suffix attachesto the consonant-stem of all verbssavethe unstable

six, e.g. laaw-ne 'saying', pot~rt-ne'speaking',tee-ne'eating'.Theseforms
are also used as nouns denoting actions, e.g. laaw-ne 'a(n instance of)
saying'.
With personalsuffixes, -ne forms the presentnarrative(seep. 406).
The older imperfectiveparticiple, built with =i « pFU *=j or *=k, which
madedeverbalnouns,Sz. Kispa11966:28-9), now servesas the third-person
singular ending in the finite paradigm (laawi '(s)he says'). In folkloristic
texts, however,it may still be found in its original participial function, e.g.
teeli '(child-)bearing'.
The privative suffix =tal may be thoughtof as a kind of negativeparticiple,
e.g. sam+pos joxt=tal xosa woolj EYE+UGHT ARRIVE=priv LONG
RIVER-STRETCH'stretchof a river longerthanthe eyecansee'.
The perfectiveparticiple suffix is -m, preceded,exceptwhen addedto the
six unstablestems,by one of the vowels i-a - ~, e.g. laaw-;;m 'said',added
pot~rt­
am 'spoken',waar-im 'made',tee-m 'eaten'.According to Sz. Kispal (1966:
193),this suffix datesbackto pU *=m, which formed norninalsfrom verbs.
With personalsuffixes, -m forms the preteritenarrativeactive (seep. 407).
The perfectiveparticiple can syncretizewith variants of the gerund (Sz.
KispaI1966:337).
Gerund
The suffix is -im(a) « *-j + *-m, a combinationof deverbalnoun-forming
suffixes, plus a lative ?*-k (cf. Sz. Kispal 1966: 336-7». Examples:juuns-
im(a) 'while sewing',pot~rt-im(a)suffixes,'while speaking',taaj-im 'while eating'.
With personalsuffixes, -im(a) forms the preterite narrative passive(see
p.407).
Adverbs
This class may be defined to include forms built from interrogative and
demonstrativestems,both productive and non-productive,by meansof the
addition of caseand othersuffixes. Meaningsrangefrom spacethroughtime
and other abstracts(Rombandeeva1973: 186-92; Kalman 1976: 52). A few
examplesmust suffice here.

Place:xoot 'where?''thence',
, xot~l 'whence?',xottaP 'whither?', tit 'here', tot 'there',
tiyl 'hence',tuwl 'thence',tiy(lje) 'hither', tuw(Pe) 'thither'.
Time: xu~ xu~'when?',tuwl 'then',akWmateert(n)'once',moolal 'recently',anj
'now', xoli 'in the morning',eetji 'at night', teeli 'in winter'.
Manner: xumus'how?', tox 'like that, thus', maa~~i1Ji~ maa~~i1Ji~ 'in Mansi', moljax
'quickly' ,jor~l
,jor~l
'very; powerfully', taak~~ 'very'.
Cause:manriy 'why?'.
State:xomi 'prone',sisi 'supine'.
Quantitative: manax 'how many times?','-fold' ~os '-fold' (xuur;;m '-fold'
~os 'three
times').
416 MANSI

Preverbs
Thesewere originally independent,concrete-spatial,adverbs.They gradually
grew together with their verbs, first modifying them semantically, then
aspectually;some among them have now becomepure perfectivizers.The
most frequent preverbsare: xot (perfectivizer),eel 'away', jol 'down', juw
'inlback', k(W)on 'out', lakwa 'asunder',lap 'down!away, shut', nal 'down!
riverward', nO(1J)x 'up', paay 'outlriverward', pal 'apart'. (Kalman 1976:
53-4.)

Postpositions
Some were originally (spatial) nouns with defective case paradigms.All
Mansi postpositionsfollow the noun in the nominative. Among the most
frequent are: xalt 'in between', xaln~1
'from between',xaln 'to between',
xoojt~1 'in the mannerof', xoPt 'as',jot 'with', kiw~m
'into', lialit 'against',
maay~s 'becauseof', muus 'as far as', muw~/,onts~l
'around', nup~ltarm~l
'towards',sist 'behind', sisn~l
'from behind', sisiy 'to behind', tarm~l
'on',
taara 'through',uultta 'across'.
Fitted with personalsuffixes, postpositionsfunction as adverbialcomple-
ments,e.g.jot;;nn 'with me' (Rombandeeva1973: 193-5; Kalman 1976: 54).

Particles
These expressthe speaker'sattitude towards the utterance(Rombandeeva
1973: 196-8; Kalman 1976: 53). Their chief functional categoriesare:

• interrogative:-a (signalsyes/noquestions);aman 'or, whether';

• assentand denial: sooli 'yes,indeed';atxuni 'no way!';
• uncertainty:tel 'probably';jar 'as if'; tup 'hardly'; eer~1J eer~1J'perhaps';
• negation,prohibition: aatii 'no' at 'not'; ul 'don't!';
• command:(w)os 'let!';
• hypothesis:ke,p~1eer~1J'if' .

Someparticleshavebecomefiller-elementswithout lexical meaning:ti, ta, i,

ja, etc.

Conjunctions
The conjunctionstie togethersentencesand sentence-parts.
They originatein
part from independentwords, in part from adverbs (Rombandeeva1973:
198-200). The most frequent conjunctionsare os 'and/also',i 'and', xunl
'when',ke 'if', akWtup'asif', kos 'although'.

Interjections
Alongside the greetingpasia 'hi' theseinclude kaaj 'oh!', axaa 'aha',ja ti
'thereyou are!', ananaa'alas!',a-a 'now you'retalking'.
 MANSI 417

Syntax
The Simple Sentence

PredicativeConstruction:Subject-Predicate
The subjectof a sentenceis generallya noun, but it can also be anothertype
of nominal, including infinitives andparticiples.Thereis no article. Definite-
nessand its oppositeis implicit in the context,thus neejuunt-i out of context
canbe 'a womanis sewing' or 'the womanis sewing'.The suffix of the third
personsingular can serve to mark for definiteness,however: nee-tejuunt-i
'thelhis woman is sewing', woot-e xoot WIND-s3 WHERE 'where is the
wind?', eekWa-telaaw-i 'the woman says' (Kalman 1976: 65; Rombandeeva
1979: 104-11).
Predicate: the most common type is the verbal predicate, e.g. xootal
kaatkWewitkWetjOIJy-i SUN BEAM WATER.dimin.loc PLAY-s3 'the sun-
beamis playing on the water', but nominal predicatesare also common(as
in Hungarian),e.g. njool-;;mzpeelpNOSE-sl POINTED 'my noseis pointed'.
Zero copulais usedin all threepersons(unlike Hungarian),e.g. naIJkar;Js
~ar kar;Js
PRO.s2SUPERL TALL 'you are the tallest' (Kalman 1976: 65; Romban-
deeva1979: 9, 32-6, 87-98).
Direct ObjectConstructions
The direct object standsin the nominativeif a noun, in the accusativeif a
pronoun,e.g. xum xaap waar-i MAN BOAT MAKE-s3 'the man is making
a boat', xaap wit-n jol+xarti-te BOAT WATER-lat DOWN+DRAG-s3s0
'(s)he drags the boat down to the water', taw aan;;mz waay-te PRO.s3
PRO.s1.accsee-s3s0'(s)he sees me'. Definitenessof the object may be
implicit in the context, or may be encodedby a personalsuffix on the noun
and/ordefinite endingson the verb, e.g. puki-te lap+juunt-;JS-teSTOMACH-
s3 COVERB+SEW-past-s3s0'(s)he sewedhislher/its stomachshut', saa-
li+ur=ne xum jomas~ar jomas saali-jay-e sun-n keer-s-ay-e REINDEER
WATCHES=pres.part MAN SUPERL GOOD REINDEER-dual-s3
SLEDGE-lat HARNESS-past-dO-s3DEF 'the reindeerherderharnessedhis
two best reindeerto the sledge'.Objects of first and secondpersoncan be
definite, as well, e.g. manriy at wiy-lan 'why don't you take (me) away?',
eerapti-lam'I love (you)' .
Adverbials
Adverbial complementsmay consistof adverbs,nominalswith casesuffixes
or postpositions,or verbal nouns,e.g. aas waata-t PuuPey;;mz'I standon the
bankof the Ob", aayit eeryimjaleY;Jt'the girls go a-singing'.
The following is a summaryof the adverbialfunctionsof the cases:
The nominative (-f)) expressestemporal relations, e.g. eet-f} xuml-awe-f}
NIGHT LIGHTNING-pass-s3'thereis lightning in the night' .
The locative (-t) refersto stationarylocation, e.g. witkasi mil aaY;Jr-t ooli
418 MANSI

'the water-monsterlives in a deep whirlpool' or to time, e.g. am ta pora-t

maansiup=talxaajtiyta-s-;;nnPRO.slTHAT TIME-Ioc UNDERPANTS=priv
RUNS.AROUND-past-s1'at that time I ran aroundwithout underpants'.
The lative (-n) refers to motion to a place,e.g. nee kol-n sialt-i WOMAN
HOUSE-IatENTERS-s3'the womancomesinto the house';it also functions
asindirect objectmarker,e.g. am ti xum-noln miy-;;nn PRO.slTHIS MAN-Iat
MONEY GIVES-sl. In passiveconstructions,the lative marksthe agent,e.g.
woot-n tot-we-s-@WIND-I at BRING-pass-past-s3'it was carried (away) by
the wind', aamp-;mjoxt-we-s-fJ DOG-Iat ARRIVE-pass-past-s3'(s)he was
approachedby a dog', neemxotjut-nat puwa-we-m,neemxotjut-nat ala-we-m
NO.ONE-Iat NOT SEIZE-pass-s1NO.ONE-Iat NOT KILL-pass-sl 'no one
cancatchme, no onecankill me'.
The elative/ablativerefersto motion from a place,e.g. niaawram-;;,t woor-
n;;,l xaajt-ey;]f CHILD-plur FOREST-ela/ablRUNS-p3 'the children run out
of the forest', or from a person,e.g. am asi-;;nn-n;;,l neepak-;;,ljoxt-we-s-;;nn
PRO.s1 FATHER-sl-ela/abl LETTER-ins ARRIVE-pass-past-sl'I got a
letter from my father'. It also hasa prolativefunction: am suup-;;nn-n;;,l kWaal-
;;m 'comeout throughmy mouth!'
The instrumental refers to both means and accompaniment.Examples:
man kasaj-;;,l jakt-ew 'we cut with a knife', niaawramaakweekwa-I uunl-i 'the
child is sitting with the grandmother'.In passiveconstructionsthe instru-
mental encodesthe direct object of the correspondingactive, cf. neepak-;;,l
'letter' in the precedingparagraph.
The translativelessiverefersto resultsand statesinto which things change,
e.g. am aasi-iy jeemt-s-;;nn 'I became a father', and to goals (manr-iy
'why?').
All of the casesparticipatein lexicalized verb rections, e.g. the elative/
ablative with pil- 'fears', the translative/essivewith namtal- 'names',the
instrumentalwith niult- 'swears(by)'.

Attributive Constructions
Any nominal (including the participles)may standin attributive position; it
then takes no agreementsuffixes, e.g. eery=;;'1) maake-m-nSONG=FUL
HOMELAND-s1-lat 'into my songful homeland', xurmit klass-;;,t
THREE=ordCLASS-Ioc 'in the third class'.
In possessiveconstructionsthe possessoris similarly unmarked. The
possession,on the other hand,may be markedwith the third-personsingular
suffix, e.g. eel;;nnxolassaatas-eHUMAN SEVENAPERTURE-s3'the seven
openingsof the human (head)', waanika maani xaap-e VANYKA SMALL
BOAT-s3 'Vanyka's little boat'. Zero marking of both possessorand
possession is alsopossible,at leastwith inanimates,e.g.poriypanekWata tuur
waata-ti1) ooli FROGTHAT LAKE SHORE-IocSTILL EXISTS-s3'the frog
still lives on the shoreof the lake', (Kalman 1976: 66).
 MANSI 419

Habeoconstruction
Unlike most Uralic languages(but like Khanty), Mansi has a verb which
means 'has': taw saali, aamp oonisi-i '(s)he has a reindeer (and) a dog';
oojka=lJ eeka tin at oonisi-~-fJwoman HUSBAND=adj WOMAN MONEY NOT
HAS-past-s3'the marriedwomanhadno money'.

Agreement
Subject and predicateagreein number and person: nalJ tinisialJ-ay-:m kol
muw.,l joxt-ey PRO.s2 LASSO-dual-s2 HOUSE AROUND ARRlVE-d3
'your two lassosreacharoundthe house'.
Within the noun phrase,the existenceof a numeraltriggers the singular:
waatxumTWENTY MAN 'twenty men'; the predicateto sucha nounphrase
agrees with the sense rather than the form, viz. waat xum saayr-ey.,t
TWENTY MAN CHOPS-p3 'twenty men are chopping'. Dvandva com-
poundsdecline in the dual, and their predicatesagreewith them in number,
e.g. eekWa-y&oojka-y xuu tot-ey WIFE-dual&HUSBAND-dual FISH
BRlNG-d3 'the marriedcouplebring (a/some)fish'.
If the verb is in the definite conjugation,it agreeswith its direct object in
number, e.g. laayl-ay-e eelali nar=iyta-s-ay-e LEG-dual-s3 FORWARD
STRETCHES-past-dual-s3 '(s)he stretched hislher legs forward',
saayrapsup-an-eta tiiwt-ulJkwe pat-s-an-eAXE-plur-s3 THEN HONES-inf
BEGINS-past-plur-s3'then (s)he began to sharpenhislher axes' (Kalman
1976: 66-7; Rombandeeva1979: 17-21).

Modality
Non-singular first-person indicative forms are hortative in force, e.g.
tuujtiylaxt-imen PLAYS HIDE-AND-SEEK-d1 'let's (the two of us) play
hide-and-seek!',min-ewGO-pI 'let's (all) go!' (Rombandeeva 1979: 24).
The narrativemood refersto activities and statesrecountedby others,e.g.
xaaroojka al-am-e '(s)he killed, it seems,the reindeer-stag'(Rombandeeva
1973: 137-44).
The conditionalmoodrefersto unreality: tot-nuw-:m 'you might bring/you
might have brought'. With the particle ke it may expresshypothesis,e.g.
jomsiakWke xuuntla-nuw-;;mt,puss:mtoryamta-nuw-l;;mtWELL IF LISTEN-
cond-s1ENTIRELY UNDERSTAND-cond-slDEFsg'if I had paid attention,
I would haveunderstoodeverything'.Notice the neutralizationof tense:any
temporalreferencemustbe understoodfrom the context(Riese1984: 87).
The affectionate/commiserativemoods reflect the speaker's affective
stancetowardsthe utterance.Usedwith the imperative,the commiserativehas
precative undertones, e.g. wooje-risi-l-en, taaje-risi-en TAKE-commis-
DEFsg-s2EAT-commis-s2'takeit andeat(you miserablewretch)!'
420 MANSI

Negationand Prohibition
The generalnegativeparticle is at: at teey-;m 'I'm not eating'. Prohibition
is expressedby ul: ul eessiyxat-;m'don'tboast!'
Negation of existenceis expressedby predicativeaatiim, which agrees
with its subject in number: naIJ ooma-n tit aatiim PRO.s2 MOTHER-s2
HERE NON.EXlST 'your motherisn't here',niaawram-iyneemxotaatiim-iy
CHILD-dual NOWHERE NON.EXlST-dual 'the two children aren't any-
where'.
There is also a negationparticle which is equivalentto a sentence:aatJi.
This is usuallyfollowed by a negatedsentence,e.g.- naIJ xuuzitey;m?- aatJi.
at xuulitey;m '- are you staying?- no, I'm not staying'.

Aspect
As in the other U gric langauges,perfectiveaspectis expressedin Mansi by
meansof coverbs,e.g.jakti- 'cuts',xot+jakti- 'cuts off, cutscompletely'.The
developmentof the individual coverbsfrom spatial adverbsseemsto have
occurredindependentlyin eachU gric language.

ConstituentOrder
Mansi is basicallySOY. An indicativepredicatenormally standsat the end of
the sentence,e.g. taaliy&tuwiy sis mat;mwoorajanxumsawsiruuj ti pisazi-;;Jl
aliyla-s-0 WINTER&SUMMER TIME OLD HUNTER VARIOUS AN-
IMAL THIS RIFLE-ins KILLS-past-s3 'winter and summer,the old hunter
killed all kinds of animalswith this rifle' (Kalman 1976: 65; Rombandeeva
1979: 52-67).

Constructions Involving Verbal Nominals

Embeddedin the sentence,verbal nominalsoften replacesubordinateclauses.
Verbal nominals take the same complementsas finite verbs. The most
importanttypesare listed here.

Infinitive
This can replace an object-clausein an analogue to the accusativeand
infinitive (indirect statement)construction: akmateertnmaataaprisi kW on;;Jl
kW aal-uIJkwe nom;;Jlmata-s-0 SUDDENLY MOUSE OUT GOES-inf
THINKS-past-s3'suddenly,the mousethoughthe would go out'. It can also
replacegoal-clausesand respect-clauses, either concrete,e.g. am xuul alisi-
aIJkwemina-s-;mPRO.s1 FISH KILLS-inf GOES-past-s 1 'I went fishing' , or
abstract: uus-iy laaw-uIJkwe maani naaIJk-i CITY-trans/ess SAYS-inf
SMALL APPEARS-s3pres'it seems(too) small to be called a city' (Kalman
1976: 70).

Participle
This, too, can replacean object-clause,e.g. ani xaapjii-ne ta kaasala-s-an;;Jl
NOW BOAT COMES-pres.partTHEN PERCEIVES-past-p3'now they
 MANSI 421

noticed that a boat was coming', but it may also function as subject, e.g.
tjiwtxa-ne-te sujt-i SHARPENS-pres.part-s3 IS AUDIBLE-s3pres 'one can
hearthat he is sharpening(his teeth)'.But participlesmostcommonlyreplace
descriptiveor determinativeclauses,e.g. piykWe-tjoom-ne xara njol BOY-
plur WANDERS-pres.partBARRENNESS'a barren promontorywhereboys
wander', maasit~r saaxi mas-~ aayi-t EXCELLENT FUR.GARMENT
DONS-perf.part GIRL-plur 'girls who have put on excellent furs'. The
original imperfectiveparticiple, now the marker of the third personsingular
present(seep. 399), survivesas a participlein archaicizingconstructionssuch
as nieelm-e xul=i ~eelm=~IJ uuj TONGUE-s3 FALLS.OFF=imperf.part
TONGUE=adj BIRD 'the tongued bird whose tongue perishes' (Kalman
1976: 70--1; Rombandeeva1979: 44).

Gerund
This replacestemporal clauses:am ta sis waanjka maa~maa~ xaap-e sunsiyla-
ma-m kaasala-s-~ 'meanwhile,while (I was) watchingVanyka'slittle boat,
I noticed ...'. The following sentenceillustrates both the gerund and the
perfective participle: joomant=ima-teakWmateertnaamp xil=~
wooIJxa-n
raayat-i RUSHES-ger-s3SUDDENLY DOG DIGS=perf.part HOLE-Iat
FALLS-s3pres'as(s)herushes(along),(s)hefalls in a hole dug by a dog'.The
gerund also replacesclausesof manner and of state, e.g. poriypaneekwa
njeelm-e jol+xanuwjane xurip=a, akWtop eeljm-~lsart=ima FROG
TONGUE-s3 DOWN+STICKING SURFACE=adj, AS.IF GLUE-ins
SMEARS-ger'the frog's tongueis sticky asif it hadbeensmearedwith glue',
peesjakti=m jiiw+aaIJkwal tuujt-n pat=ima LONG.AGO CUTS=perf.part
TREE+STUMP SNOW-Iat FALLS=ger 'the tree stump, cut long ago, has
beencoveredby snow' (Kalman 1976: 71).

Compound Sentences
Co-ordination
Co-ordination is usually effected by simple asyndetic concatenation,e.g.
laayl-e waaytal, kaat-ewaaytal 'hislherlegs areweak, (and) hi/her handsare
weak', turpate taaylak sar~l toxrite, kosamtite, mossakeastal uunli, tuwl
oow~lti '(s)he fills hislher pipe full of tobacco, lights it, sits for a while
without speaking,thenbegins'(Rombandeeva 1979: 122-34).
There are native conjunctions(os 'and, also', man 'or'), but the majority
are of Russianorigin (i 'and',a 'but').

Subordination
Alongsidethe use of participlesto expresssubordination(p. 420) Mansi also
usesclauseswith finite verbs,linked by conjunctions.Thesearederivedfrom,
or equivalentto, interrogativepronounsandadverbs,e.g. xoot 'where', xomos
'how'. Examples:naIJ waayl~, xoot oln waarawe? 'Do you know where
moneyis made?'In sentencescontaininga hypothesis,the particle ke (tje) is
422 MANSI

used,e.g. ~i~naajteey, tie teey, asj:»nayjeemti 'if the auroraborealisis burning,

it will be cold' (Rombandeeva1979: 135-40).

Lexicon
Etymologicalresearchhasdeterminedthe origin of roughly 60 per centof the
Mansi lexicon.A significantportion of the basevocabularyis eitherinherited,
i.e. Uralic, Finno-Ugric, or Ugric in origin, or has arisen internally by
morphologicalmeans(derivationandcompounding)or by sound-symbolism.

Inherited Vocabulary
The entriesof the UEWdemonstratethat a greatpart of the known Uralic and
Finno-Ugric lexicon has beenpreservedin at leastone of the Mansi dialects
(cf. p.390). Approximately 155 Mansi words may be seen as Ugric
innovations;thesehavecognatesonly in Hungarian,Khanty, or both. Half are
found in all three Ugric languages,e.g. saat 'seven',niololow 'eight', tur
'throat',jomas 'good',joxti- 'arrives', kWon 'out', maani~i 'Mansi',posim
'smoke',mil 'deep',tawt 'quiver', aamp 'dog', luw 'horse',nlayir 'saddle';
roughly one quarter are found only in Mansi and Hungarian, e.g. jeekw::Jr
'root', puwi- 'seizes',kol::Js 'grain (millet)" mowinlt- 'laughs'cf. Hungarian
gyoker,jog-,kales,andnevet-.

LoanWords
We may reckon with Proto-Iranian (saat 'seven') and Proto-Turkic (luw
'horse') influence as early as the Ugric period. Both Iranian and Turkic
influencecontinuedafter Hungarianbroke away.

Iranian Loans
The influenceof Old Iranian languageson ObUgrian is attestedby a body of
loanwords in diverse semanticfields. This vocabulary must have entered
proto-ObUgrianno later than the fourth to sixth century CEo About a dozen
survive in Mansi, e.g. aara~ 'hearth',waasiy 'calf', eeriy 'song', ees::Jrma
'shame',oot::Jr 'chief, price', puuy 'wealth(y)', ~ak '(sledge)hammer', woot
'wind', waar- 'makes',tuujt- 'hides' (Korenchy 1972: 46-84).

Tatar Loans
Thereis evidencefor SiberianTatar influenceon Southerndialectsof Mansi
from the fourteenthcentury onwards.There are more than 500 Tatar loans,
embracing most social and economic semantic spheres:alongside words
pertinentto animalhusbandry,agriculture, handicrafts,andcommercewe also
find vocabularyfrom thedomainsof clothing,food, medicine,andeven Islamic
religious life. The main currentof Tatarinfluencewas brokenby the Russian
conquestof westernSiberiain thesixteenthcentury.Only about5 percentof the
Tatar loans are found in all Mansi dialect areas,and only about 10 per cent
reached the northern dialects. Among these latter loans are e.g. xoosax
 MANSI 423

'Cossack',xoon 'Khan',japak'silk' , suuntax'box', toor 'pieceof cloth', eelak

'sieve',tolmasi 'interpreter',aam;,si'riddle', (Kannisto1925: 1-264).

RussianLoans
Thereappearsto havebeensporadicRussianinfluenceas early as the twelfth
to the fifteenth century. With the conquestof western Siberia, however,
Russianinfluence grew significantly; since then Russianloans have flowed
into Mansi consistentlyanddirectly. Russianloansnumberover600 in dialect
texts collected in the last century; the number in the standardlanguageas
currently used is demonstrablygreater. From the semanticpoint of view,
Russianloans extendand elaboratethe fields representedby Tatar loans; an
important novelty are the terminologiesof Russiancentral governmentand
orthodox religion. Examples: aark;,ri 'bishop', Peekkar 'physician', mir
'people', isiwesisia 'candle', laampa 'lamp', turpa '(tobacco)pipe', puska
'cannon',peet;,r 'pail', tas 'vessel',ras 'case,instance',siaaj 'tea', kaalasi
'bun', kurka 'hen',paania 'bath-house'(Kalman 1960: 1-327).

Komi Loans
The earliestKomi merchantsand settlersappearedin westernSiberia in the
tenth century, and we may date the first phaseof Komi influence on Mansi
to the period from the tenth tothe fifteenth century,whencommercialcentres
and towns arose.The secondphaseof Komi influence on Mansi occurredin
the eighteenthand nineteenthcenturies.In both phases,the greatestcontact
was in the northernMansi linguistic area.Of the more than 300 Komi loans
in Mansi, 85 per cent may be found in northerndialects.Only about 10 per
cent pervadedto all Mansi dialects. Prominentamong the Komi loans are
words from the semantic domains of building, furnishings, foodstuffs,
clothing and transport,e.g. isnas 'window', kuur 'oven', ulas 'chair', pasan
'table', niaani 'bread', kaasi 'women'sundergarment',toos 'cross-strutin
boat'. Other examplesinclude moojt 'tale', pupakwe 'bear', siaar;,si 'sea',
kaasal- 'notices', sart- 'smears',turap 'storm', aakani 'doll', oojka 'old
man', rusi 'Russian'.
It shouldbe notedherethat Udmurt loanspenetratedthe Chusovayadialect
of Mansi describedby Reguly (Kalman 1988: 411).

SamoyedLoans
Roughly thirty Nenetsloanwordsconcernedwith reindeer-breedingand life
on the tundra have enteredNorthern Mansi. Such are xalew 'gull', tinisiaI]
'lasso',tutisialJ 'sewingbag madeof reindeerhide', siowal 'chuval' (Steinitz
1959: 426-53).

KhantyLoans
The Northern Mansi dialect groups are in contact with Obdorsk (now
Salekhard)and Kazym Khanty speakers,and EasternMansi dialect groups
wereoncein contactwith the KondaKhanty. In the caseof two closelyrelated
424 MANSI

languagessuch as Mansi and Khanty it is of coursedifficult to determine

whether in a given instance we have to deal with a cognate or a later
borrowing; the entire question still awaits monographic treatment. It is
beyonddoubt, however,that there are many Khanty loans in many layers of
the Mansi lexicon. This is especially true of the vocabulary of the bear
ceremony: to this day the Khanty are the custodiansof the world's most
elaborate bear cult. Examples of probable Khanty loans in Mansi are:
niaawram 'child', aapsii 'youngerbrother;nephew',kur 'leg', laj 'dark', uunt
'forest', leelJk;Jr 'mole', luuli 'wild duck', luulJi 'dog', maay 'honey', ooxi
'bear's head', xos 'bear's eyes' (Sz. Bakr6-Nagy 1979: 23-59; Steinitz
1996-93s.vv.).

Words of Internal Origin

Thereis much sound-symbolicvocabulary,e.g. liaty- '(fire) crackles',miry-
'thereis thunder'.The most productiveform of word-formation,however,is
regularderivation (cf. p. 408, 414). The oldestroots havethe most extensive
derivationalprofiles, e.g. xuuli - 'leavesbehind',xuulit- 'remains',xuuzitiyl-
'keepsgetting left behind',xuuliiyl- 'suddenlyis there', xuuziiylant- 'keeps
croppingup'. But recentloans are also susceptibleto derivationalprocesses,
e.g. tumaj 'thought', tumajt- 'thinks'.
The secondmost commontype of word-formationis compounding,e.g.
aayi+piy 'child (GIRL+BOY)', maa+xum 'people(EARTH+MAN)', numi-
+toor;Jm 'god (UPPER+SKY)',niuli+kis 'rainbow (PITCHPINE+HOOP)"
woot+piy 'storm (WIND+SON)'. For a time, efforts at centrally directed
languagereform createdamalgamsof compoundingand derivation such as
xaniisitan+xum 'teacher (TEACH=ING+MAN)', xanlisita=n+kol 'school
(TEACH=ING+HOUSE)',towl=;JlJ+xaap 'aeroplane(WING=ED+BOAT)'.

Words of Unknown Origin

A significantportion of the basevocabularyof Mansi consistsof words with
no known cognates.Sucharexanisi- 'knows',laaw- 'says',oonisi- 'has',pat-
'falls', raat- 'strikes',woow- 'calls', li apa 'near',oxsar 'fox'.

Text
(For a slightly different version, in Cyrillic orthography,seeKalman 1976:
92).

seer&moor woor-t siir&siir towl=;)1J

DENSE&DENSE FOREST-Ioc VARIOUS&VARIOUS WING=ED

uuj-;)t ool-ey;)t. ti-t siop;)r kisup

ANIMAL-plur LIVE-p3pres HERE CAPERCAILLIE HAZEL-GROUSE
 MANSI 425

kukuk maal)kwla xoont-ul)kwe weerm-ey~n.weerm-ey~n. uumi-t-t

uuj=risL~t
CUCKOO OWL FIND-inf IS.ABLE-s2pres BIRD=sx-plur

suj~I)=~sj suj~I)=~sj lujy-ey~t lujy-ey~t eery-ey~t. eery-ey~t. uumi-t-t

jiiw+tow-~t
LOUD=adv CHIRP-p3pres SING-p3pres TREE+BRANCH-plur

paal-t niirsi axl-~t axl-~taxl-~t

-t uumi-t-t taan
SURFACE-loc BUSH-plur-loc HOLLOW-plur-loc PRO.p3

dense waar-ey~twaar-ey~t muul)i-t pin-ey~t pin-ey~t

dense MAKE-p3pres EGG-plur LAY-p3plur

niaawram-~t niaawram-~t janmalt

janmalt
-ey~t.
CHICK-plur RAISE-p3pres

'In the dense dark forest many kinds of birds live. Here you can find
capercaillie,hazel-grouse,cuckoos,andowls. The little birds loudly chirp and
sing. On tree-branches,in bushes,in hollows they build (their) nest(s),lay
eggs,raise(their) young.'

Referencesand Further Reading

Ahlqvist, A. (1891) WogulischesWorterverzeichnis,MSFOu 2, Helsingfors: Societe
Finno-Ougrienne.
Bakr6-Nagy, Sz. Marianne (1979) Die Sprache des Biirenkultes im Obugrischen,
Budapest:Akademiaikiad6.
Balandin, A.N., and Vakhrusheva, M.P. (1957) MaHcllllcKllll H3bIK, Leningrad:
YQrrej:\fIl3.
- - (1958) MaHcllllcKlIll-PYCCKllll CJIOBapb.Leningrad:YQrrej:\rll3.
Chemetsov,VN. (1937) 'MaHCIIllCKllll (BOryJIbCKllll) H3bIK', in r.H. IIpoKocpbeb
(ed.), 5I3b1Kll II rrllCbMeHHOCTb Hap0j:\OB CeBepa. 1.JacTb, vol. I, Leningrad:
YQrrej:\fIl3, pp. 163-92.
Collinder, B. (1957) Survey of the Uralic Languages, Stockholm: Almqvist &
Wiksell, pp. 319-44.
Hajdu, P. (1966) Bevezetes az urdli nyelvtudomdnyba,
Budapest:Tanktinyvkiad6.
Honti, L. (1975) Systemder paradigmatischenSuffixmorphemedes wogulischen
Dialektesan der Tawda,Budapest:AkademiaiKiad6.
- - (1977) review of KaIman 1976.
- - (1980) 'Zur Phonemanalysedes Tavda-Wogulischen', Finnisch-Ugrische
Mitteilungen4: 61-8.
- - (1982a)Geschichtedes ObugrischenVokalismusder erstenSilbe, Budapest:
AkademiaiKiad6.
- - (1982b) 'Der Passivin der ObugrischenSprachen',Acta Linguistica Hungar-
ica 32: 39-5l.
- - (1984) 'Versucheiner Beschreibungdes Phonembestandes im wogulischen
Dialekt an der Pelymka', in P. Hajdu and L. Honti (eds), Studien zur phonolo-
gischen Beschreibung uralischer Sprachen, Budapest: Akademiai Kiad6,
pp.87-102.
- - (1988) 'Die wogulischeSprache',in D. Sinor (ed.), The Uralic Languages:
426 MANSI

Description, History and Foreign Influences, Handbuch des Orientalistik 8/1,

Leiden: Brill, pp. 147-71.
Kalman,B. (1961) Die russischenLehnworterim Wogulischen,Budapest:Akademiai
Kiad6.
- - - (1965) Vogul Chrestomathy,IndianaUniversity Uralic and Altaic Series46,
Bloomington-TheHague:IndianaUniversity-Mouton.
- - - (1975) ChrestomathiaVogulica, Budapest:Tankonyvkiad6.
- - - (1976) WogulischeTextemit einemGlossar,Budapest:AkademiaiKiad6.
- - - (1984) 'Das nordwogulischePhonemsystem' , in P. Hajdu and L. Honti (eds),
Studien zur phonologischen Beschreibung uralischer Sprachen, Budapest:
AkademiaiKiad6, pp. 73-9.
- - (1988) 'The history of Ob-Ugric languages',in D. Sinor (ed.), The Uralic
Languages: Description, History and Foreign Influences, Handbuch der
Orientatistik8/1, Leiden: Brill, pp. 395-412.
Kannisto,A. (1919) Zur Geschichtedes Vokalismusder erstenSilbe im Wogulischen
vom quantitativenStandpunkt,MSFOu46, Helsinki: SocieteFinno-Ougrienne.
- - - (1925) Die tatarischenLehnworterim Wogulischen,FUF 17: 1-264.
Kannisto, A. and Liimola, M. (1951-63) Wogulische Volksdichtung, vols I-VI,
MSFOu 101, 109, 111, 114, 116, 134, Helsinki: SocieteFinno-Ougrienne.
Kannisto, A., Liimo1a, M. and Eiras, V. (1982) Wogulische Volksdichtung,vol. VII,
MSFOu 180, Helsinki: SocieteFinno-Ougrienne.
Keresztes,L. (1972) 'Uber das Vokalsystemdes Jukonda-Dialektsim Wogulischen',
FUF 39: 277-94.
- - - (1973) 'Vogulin Jukondan(Ala-Kondan)murteenkonsonanttijiiIjestelmasta',
JSFOu72: 167-74.
Kispal Sz., M. (1966)A vogul igenevmondattana,Budapest:Akademiaikiad6.
Korenchy, E. (1972) lranische Lehnworterin den obugrischenSprachen,Budapest:
Akademiaikiad6.
Kovesi, M. (1933) 'Igem6doka vogulban',Finnugor Ertekezesek1 Budapest:Kirelyi
MagyarEgyetemNyomda.
Kulonen, U.-M. (1989) The Passivein Ob-Ugrian, MSFOu 203, Helsinki: Societe
Finno-Ougrienne.
Lak6, G. (1957) NordmansischeSprachstudien,Acta Linguistica Hungarica 6:
347-423.
Liimola, M. (1963) Zur historischenFormenlehredes Wogulischen,vol. I: Flexion
der Nomina,MSFOu 127, Helsinki: SocieteFinno-Ougrienne.
Munkacsi, B. (1894) 'A vogul nyelvjarasok sz6ragozasukbanismertetve', Ugor
Fiizetek 1.
Munkacsi,B. andKalman,B. (1952)Manysi(vogul) nepkoltesigyujtemeny,vol. IIU2,
Budapest:Akademiaikiad6.
- - (1963) Manysi (vogul) nepkoltesi gyujtemeny, vol. IV/2, Budapest:
Akademiaikiad6.
- - - (1986) WogulischesWorterbuch,Budapest:Akademiai kiad6.
Pirotti, G. (1972) Grammatica vogula, Mantu: Associazione culturale Italo-
UnghereseTaddeoUgoleto da Parma.
Redei,K. (1970) Die syrjiinischenLehnworterim Wogulischen,Budapest:Akademiai
kiad6.
Riese, T. (1984) The Conditional Sentencein the Ugrian, Permian and Volgaic
Languages,Vienna: Institut fUr Finno-Ugristikde UniversitatWien.
- - - (1988) 'Zur Herkunft des Konditionalis im Siidwogulischen', FUF 48:
51-61.
- - - (1992) 'Zur Entwicklung des Kasussystems im Wogulischen',in P. Dereky et
 MANSI 427

al. (eds), Festschrift flir Karoly Redei zum 60 Geburtstag, Vienna-Budapest:

Institut fUr Finno-Ugristikder Universitiit Wien-MTA Nye1vtudomanyiIntezet.
Rombandeeva, E.I. (1954) PyccKo-MaHcHHcKHHCJIOBaph,Leningrad:ceBepHhIX yqne~rH3.
- - - (1964) <l>oHeTHQeCKHeoco6eHHOCTHB ~Ha.rreKTax ceBepHhIXceBepHhIXMaHCH, in B.A.
Serebrennikov(ed.), Bonpochl <l>HHHo-yropcKoro ll:3hlK03HaHIDI, rpaMMaTHKa H
CPOJIhKJIOpaMoscow-Leningrad:HaYKa.
~eKCHKOJIOrIDI,
(1973) MaHcHHcKHH (BorYJIhCKHH) ll:3hlK, Moscow: HaYKa.
- - - (1979) CHHTaKcHcMaHcHHcKoro (BOryJIhCKOro)ll:3hlKa, Moscow: HaYKa.
- - - (1993) I1cTopHll: Hapo~aHapo~aMaHCH (BOryJIOB) H ero ~yxoBHall: ~yxoBHall: KyJIhTypa (no
CPOJIhKJIOpa
~aHHhlM CPOJIhKJIOpaH 06Pll:~OB), 06Pll:~OB), Surgut: CehepHhlu~OM
Rombandeeva,E.I., and Kuzakova, E.A. (1982) CJIOBaph MaHCHll:CKO-PYCCKHH H
PyccKo-MaHCHHCKHll:, Leningrad:IIpocBe~eHue.IIpocBe~eHue.
Steinitz, W. (1955) Geschichte des wogulischen Vokalismus, Berlin: Akademie-
Verlag.
- - - (1959) 'Zu den samojedischenLehnwortem im Ob-ugrischen',UATb 31:
426-53.
Steinitz, W. et al. (1993-96)Dialektologischesund etymologischesWorterbuch der
ostjakischenSprache[= DEWOS],Berlin: AkademieVerlag.
Szab6, D. (1904) 'A vogul sz6kepzes',NyelvtudomanyiKozlemenyek34: 55-74,
217-34,417-57.
Veenker, W. (1971) Riickliiufiges Worterbuch der vogulischen Schriftsprache,
Wiesbaden:Otto Harrassowitz.
14Hungarian
Daniel Abondolo

Hungarian is by far the most widely spoken Uralic language,with some

thirteen million speakersin Hungary and adjacentcountries and roughly
anothermillion scatteredaroundthe globe. The oldest continuoustext dates
from the end of the twelfth century, and from the fifteenth century a
considerablephilological recordbeginsto accumulate.
Hungarianis traditionally assignedto the eastern,more commonly called
U gric, branchof Finno-Ugric; it brokeawayfrom this nodebetweentwo-and-
a-half and three-and-a-halfmillennia ago, leaving behind proto-ObUgrian,
the ancestorof Khanty and Mansi.
Hungarianprobablybeganas a core, ratherthan as a peripheral,dialect in
proto-Uralic.Although typically Uralic in many respects(broadly agglutinat-
ing, with vowel harmony, postpositions,possessivesuffixes, and minimal
noun-phraseagreement), Hungarian is also strongly deviant in certain
regards,in particularin the developmentof its vowels and in the detailsof its
noun and verb finite paradigms.In terms of historical phonology,Hungarian
is the only Finno-Ugric languageto spirantizeoriginal initial *p-; in this it
patterns,coincidentally,with the northernmostand southernmostSamoyedic
idioms, Nganasanand Mator. Coincidental, too, are the independentspir-
antizations of initial *k before back vowels in Hungarian, Northern and
EasternMansi, and WesternKhanty. On the other hand soundchangessuch
as *mp > b, sharedwith Permian,and the (partial) marking of object person
in the finite verb, sharedwith Mordva, point to Sprachbundconvergences. In
fact what is traditionally termed 'Ugric' may itself have been more a
Sprachbundthan a nodein the Uralic family tree (seenow Viitso 1996, who
speculates along these lines on the basis of the development of the
consonants).
Much has beenwritten aboutHungarianin the readily availablelinguistic
literature of the past few decades.Accordingly, this chapterconcentrateson
some of the aspectswhich have tended to be neglected,with particular
emphasison the morphological typology and prehistory of Hungarianqua
Uralic language.

428
 HUNGARIAN 429

Map 14.1 HungarianDialects

99
T
5
6

B
1
2
4
7
88
3

D 1 Transdanubian
2 Transdanubiandialects
3 Southerndialects
4 Tiszadialects
5 Northern(PaI6c)dialects
6 Northeasterndialects
B Lake Balaton 7 CentralTransylvaniandialects
D Danube 8 Transylvanian(Szekely)dialects
T Tisza 9 Csiing6 dialects

Source: Simplified from Hajdu-Domokos1987: 44.

Phonology
Primary word stressis assignedto the first syllable; sentence-stress
rules then
heighten,dampen,or deletethis stressin accordancewith syntactic,aspectual,
andpragmaticvariables.SeeHetzron 1982for a detailedexploration.

Vowels
Dependingon the variety, there are two main vowel inventories,one with
fourteen and one with fifteen vowels. Perhapsas many as two-thirds of
speakersusethe smallerinventory: this consistsof sevenshortandsevenlong
vowels. The inventory usedby most of the remaining one-third of speakers
hasan additionalshortvowel, front mid unroundede:
Short Long
Front Back Front Back
-R +R -R +R -R +R -R +R
High i ti u ii titi uu
Mid (e) ij 0 ee ijij 00
Low a a aa
430 HUNGARIAN

In the standardorthography,both aande are written with <e>; in specialist

contexts (linguistics, lyrics of folksong) e is differentiated by means of
diaeresis, <e>. In older specialist works, a is sometimes written <£>.
Otherwise,the orthographicsymbols are as given above, except that long
vowels are written not double, as in both the phoneticand the phonological
transcriptionsusedin this book, but with an acutediacritic, thus <3.> <0> <u>
<e>. The acutereplacesthe dot of <i>, viz. <f>, and a double acutereplaces
the diaeresisof <ti> and <0>, viz. <i1> <C».
On the whole, the duration of the long vowels is from 50 to 60 per cent
greater than that of the short vowels; the chief deviations from this
a,
distribution are whosedurationoften approachesthat of a long vowel, and
ee, whosedurationusually is as low as that of a shortvowel (for somerecent
figures see Gosy 1989: 80). From the point of view of phonetics these
a
deviations are not unexpected,since is quite open and ee is rather close.
Distinctivenessof length amongthe high vowels is the leastconsistent;many
speakershaveit only in orthoepicsocialcontexts,or not at all.
Syllable structurehas only restricted effects on vowel quantity; see on
marginalphone(me)s, below. The effectsof word stresson vowel quantityhas
not beenadequatelyinvestigated(see,however,Nadasdy1985: 229).
The long mid vowels ee ijij 00 are pronouncedrathermoreclosethan their
short counterparts(e) ij o. This is also true, thoughto a lesserdegree,of the
high vowels.The tongue-heightof ais often quite low, evenfor speakerswho
lack e. Of all the vowels, aa is pronouncedthe most open. Note also that
althoughaa is hereclassifiedas 'back',it can be pronouncedrathercentrally
or evensomewhatfront, especiallyby urbanyoung women.
There are also severalmarginal vowel phones;we may group theseinto
threetypes.One type is representedby a phoneof externalorigin: it is a short
central-to-frontlow unroundeda, occurringmostfrequentlyin the pronuncia-
tion offoreign items suchas Weis(s)[vajs] andArafat [arafat]. Whenfollowed
by j, this a remainsunchanged,but any a in words of the Arafat-typereverts
a
to core when the word is inflected, e.g. singulardative Weisnak[vajsmlk]
but Arafatnak[iirMatnilk].
The secondtype of marginal vowel phoneinvolves the short low vowels
a and a. Frequently in the pronunciation of the deictics arra 'this way,
thereabouts',erre 'this way, hereabouts' and merre '(by) which way?
whereabouts?'the first vowel is pronouncedlong, i.e. we have [aiir(r)a],
[aar(r)a], [maar(r)a]. (The duration of the rhotic varies considerablyin these
forms and is non-distinctive;the length of the precedingvowel, however,is
distinctive,cf. arra [mal 'onto that'.)This kind of pronunciationappearsnow
to be spreadingto acronyms,e.g. EK [aakaa] 'E(uropean)C(ommunity), and
otherquotativeinstancesof letter-names.For a recentdiscussionof thesetwo
typesof marginalvowel seeKeresztes1993.
The third, somewhatcontroversial,type of marginal vowel phone is the
mirror image of the secondtype: it involves the relatively short duration of
 HUNGARIAN 431

the pronunciationof phonologically long vowels when theseare followed,

within the morpheme,by homotopicclustersof resonant(particularly nasals)
plus obstruents,e.g half-long [a·] in bdnt- [ba·nt] 'harms,irritates' but fully
long in bdn-t-{j [baant] 'treated,dealtwith'.
To varying degreesand in various registers, any short vowel may be
pronouncedlonger when to the left of a tautomorphemicresonantfollowed
by an obstruentor pause,e.g. bors [bo·rs] 'pepper',bor [bo·r] 'wine', and any
long vowel may be pronouncedshorter in a closed syllable, particularly
beforenon-geminateclusters.
Onemustalsodistinguishphenomenaassociatedwith affectivepronuncia-
tion, in which any vowel may be renderedwith (much) increasedduration.
Evenin this type of durationaldeviation,tongue-heightremainsconstant,e.g.
the long vowel of hooogy?[hoooeV] 'how?!' (with a quick dip, followed by
a slow rise, in pitch, and with optional creaky voice) has the sametongue-
heightas that of its non-affectivependanthogy?
At the phonologicallevel, Hungarianvowelsexhibit a rangeof alternations
involving length, frontness/backness, the presence/absence of lip-rounding,
and tongue-height.These are briefly presentedunder morphophonology,
below.
Much is known about the history of the Hungarianvowels over the last
millennium, becausethe philological record has been studied and well
summarized(e.g. Barczi 1958). On the prehistoricalside, many of the details
of the backgroundof the Hungarianvowels have been worked out by the
comparative method (e.g. Sammallahti 1988: 500-501, 513-15). Never-
thelessthe overall picture remainsunclear,and the following must therefore
a
be cursory.Labial low was probablystill a positionalvariant of unrounded
*a in the thirteenthcentury. The long mid vowels ee 00 00 are probably all
secondary,the result of (1) contractionafter the loss of glides, e.g. osz[oos]
'autumn'< *oyOs < pUgric *Stibs,« pFU *stiksii; (2) monophthongization
of diphthongs;or (3) compensatorylengtheningafter the loss of the vowel of
the secondsyllable, e.g. ver [veer] 'blood' «< *weri < pFU *wiri), or kez
[keez] 'hand'.On the otherhandthe length of the fust-syllablehigh vowel in
many words, at leastin parts of their paradigm,may be a preservationfrom
p(F)U, e.g. hUgy [huueV] 'urine' < *kuunisil, nya [niiil] 'arrow' < pFU *niml
< pU *nilxll. The low vowels (a a aa) usually date back to p(F)U non-high
vowels, as in hal [ha·l] 'fish' < pU *kala, Mz [haaz] 'house'< pFU *kaata,
k6z (accusativekeze-t[kazat]) < pFU *kati, but in words which were built
with derivationalsuffixes alreadyin p(F)U, they are often from (short) non-
lows, e.g. eger, accusativeegere-t([ageer],[agiirat]) < pFU *siIJiri (> Finnish
hiiri, Komi sfr).

Vowel OppositionsandAlternations,and the MorphophonemicCode

From the morphophonologicalpoint of view, the vowel inventoriesas given
above are extravagant. Productive or extensive morphophonemicvowel
432 HUNGARIAN

alternationson the one hand,and numerousrestrictionson the distribution of

vowels on the other, enablethe extractionof prosodicfeaturessuchas length,
frontness/backness, and lip-rounding. Such extractions,in tum, allow us to
reducethe vowel inventoriesgiven above to five, written in this chapterin
code(given in capitalletters):I E A 0 U. The trivial readingsfor theseabstract
segmentsare i e(or, in varietieslacking this segment,a) it 0 u when single,
ii ee aa00 uu whendouble;in front-prosodiccontexts(markedwith prefixed
# where necessary) the readingsare i e(/a) a 0 ti and ii ee ee00 titi. Note that
the abstractdescriptionof the standardlanguage,in which the mergerof e and
a has been completedat the surfacelevel, is best carried out with distinct
underlying E (= e) and # A (= a) units (cf. Szepe1969; Abondolo 1988: 34;
Kornai 1994: 14).
In bound morphemes,the short mid vowels 0 0 e (with a counting as
a mid vowel for those speakers who lack e) may in most cases be
unambiguouslyrecoveredfrom a cover symbol, written here as 3. Front
prosody, indicated by # at the start of a root morphemein the mannerof
a key signaturebefore a sequenceof tonal music, is necessaryonly if the
morphemecontains one of the vowels A 0 U, i.e. those with dual surface
prosodic readings,and with verb roots and themeswhose sole vowel is I,
which normally have back vocalism. Examples: BOR bor [bo·r] 'wine',
~OOR ~OORbOr [boor] 'skin, hide, leather',IRT-0-3K irtok 'I extirpate',IfJIINT-
0-3K hintek 'I sprinkle'. In front-vowel roots, the symbol 3 may be used
to indicate superabundantforms, e.g. #P3L fellfol 'up(wards)'. The code
renders transparent much of the phenomena which operate in the
morphophonology, and will therefore be used here. The fact that this
procedureaccommodatesmany unassimilatedforeign items awkwardly -
e.g. niiansz [niians] would have to be written ~bANS, ~bANS, with change of
prosodyin mid-morpheme- is of interestto diachronists.
In terms of lexical frequency,the most pervasivevowel alternationis AA
- A, i.e. aa - it and ee - a; this alternationis most characteristicof, but not
restrictedto, nominal inflection and derivation, e.g. utca 'street' - utcan 'on
(the) street',feje 'hislher/itshead' - fejen 'on hislher/itshead',szoba'room'
- hdromszobds'three-roomed'.In all suchcases,the shortvowel is the result
of reductionof stem-finalAA, either in word-final position (as in utca,feje,
szoba) or to the left of certain derivational suffixes, e.g. UTCAA = I utcai
'pertainingto the street',contrastUTCAA-I-NKAutcaink'our streets'.The
cancellationof the A in this last exampleis the most pervasivealternationin
termsof textualfrequency;suchcancellationbefallseveryA whenit is in final
position in the non-compoundword, and henceforthwill not be indicatedin
the code.
In non-final position in large, but closed,setsof core nominal vocabulary,
the peripheralvowels I U A alternatewith their long pendantsII UU AA,
while A in morpheme-finalposition alternateswith zero.Examples:non-final
I is readas II (ii), and final A as zero, in NJILA ny{[ [njiil] 'arrow'; similarly,
 HUNGARIAN 433

final A is read as zero, non-final U is read UU (here titi becauseof front

prosody)in ~UZA~UZAtilz [ttitiz] 'fire'; andfinal A is readas zero, but non-final
A is readasAA, in MADARA maddr[madaar] 'bird'. Contrastthe isochronic
readingsof thesesegmentsin the correspondingaccusativeforms NJILA-T
nyilat [njilat], #TUZA-T tuzet [ttizat], MADARA-T madara-t [madarat].
Many nominals (belonging to other large closed lexical sets) and many
(mostly derived) verbs show epenthesisof a short mid vowel, e (3), 0, or ij
(cover symbol: 3) betweenthe consonantsof their stem-final cluster, e.g.
AALM dlom 'sleep;dream',TITK titok 'secret',IKR iker 'twin', #OKR okor
'ox', MOZ:=G-mozog [mozog] 'moves'. As these examples suggest, the
quality of the epenthetic vowel depends on frontlback prosody, [+1-]
roundednessof the vowel of the precedingsyllable, and the nature of the
surroundingconsonants(Abondolo 1988: 108-10,205-9).A few words show
both epenthesisand quantitative alternations,e.g. ~ALKA ~ALKA zelek [leelek]
'soul'.
In many people'sspeech,the long high vowels II and UU alternatewith
their short pendantsin subsystemsof the derivationalmorphologyof certain
univocalic verbs, e.g. HUUZ- hUz- [huuz] 'pulls, draws', HUlL!Z=AT huzat
[huzat] 'draught', BIIZ- biz- [biiz] 'entrusts', BIIZ=ALMA bizalom
[bizaIom] 'confidence,trust'.
Consonants
The inventory of Hungarianconsonantsis set out in Table 14.1. The places
of articulation are (1) bilabial, (2) labiodental, (3) apicodental (nasalsand
stops), lamino-alveolar (sibilants, and release portions of affricates), (4)
palatals, (5) larnino-palatalveolar(hushing fricatives and affricates), (6)
dorsovelar,and(7) glottal. The stopsareunaspirated.
The orthographic representationsgiven in Table 14.1 are the default
spellings; numerous other spellings also occur because Hungarian

Table14.1 Hungarianconsonantsand their main orthographicrepresentations

12 3 3 4 563 7 Orthography

Nasals mn ni ni ill n ny
Stops p t tj k P t ty k
bdd ni nij g b d gy g
Fricatives f s (I;) s (x) h f sz ch s ch h
vz p Z v z zs zs
Affricates c C c cs
dZ dZ dz dzs
Lateral p I
Glide j j/ly
Trill r r
434 HUNGARIAN

orthographyis for the most part morphophonemic.Most notably, the output

of all ofthe morphophonemicprocesses(1-5) outlinedbelow is not normally
reflectedin writing.
The palatal glide j is written <j> or <ly> on historical grounds(the shift
from *lj to j was not completed until the nineteenthcentury, when the
orthographyhad begun tojell). Long consonantsare written double, but in
digraphsonly the first memberis written twice, e.g. <ggy> for djdj . In a few
foreign words, the digraph <ch> is usedto write the segments~~ (in front-
vowel environments)and xx (elsewhere),e.g. pech [pe/a~~] permits'hard luck' and
krach [knlxx] 'financial collapse'.Thesephones(and their voiced pendants,
which arisein sandhi;seebelow) are marginal, occurmost typically in slang!
foreign vocabulary, and are long insofar as the environmentpermits (see
below). Certain instancesof ~ are derived from I, e.g. LOP-IA-0 lopj permits
[lop~]
'steal!' Otherwise, ~ and x may be derived morphophonemic ally from
geminatesequences of an abstractfricative H, which whensingleis the source
of h (Abondolo 1988: 67-9). Similarly, the glide j may be derived from the
short vowel I. The voiced labiodentalfricative v, on the other hand, is more
firmly rootedand musthaveits own abstractsegmentV.
In general,consonantlengthis distinctiveintervocalicallyandbetweenthe
last vowel of a word and pause.Long consonantsoccur most frequently
astridemorphemeboundaries,e.g. KANALA kanal [kanaal] 'spoon',KAN-
NAAL kannal [kannaal] 'than a male (animal)', MEN-IA-0 menj! [menini]
'go!', but are also quite frequentroot-final, particularly in recently acquired
vocabulary,e.g. SHOKK sokk [sokk] 'shock', cf. SHOKA sok [sok] 'much,
many'. Length in affricatesis realizedin the stop component:LAZAC-VAL
lazaccal [h\ziittsaI] 'with salmon'.There seemto be no instancesof long vv
or zz in prepausalposition; on the other hand, dZ and dZ occur short only
where length is disallowed. Consonantswhich are long at the morphopho-
nemic level are shortened wherever their occurrence is precluded, e.g.
WSHT-TOOLjiisttol [ftist60l] 'from smoke'.
If we minimalize the redundanciesin the matrix given abovewe arrive at
an abstractconsonantalinventory with twenty-five members(plus j, which
impingesfrom the realmof the vowels, seeTable 14.2).
The main oppositionsare thus (1) [+/-] voiced and, amongthe core conso-
nants,theoppositions(2) [-1+] palatalizedfor thestopsand(3) hissing:hushing
for thefricativesandaffricates.Themostimportantsynchronicconsonantalter-
nationsinvolve neutralizationsof theseoppositionsin variousforms of sandhi.
Theseare asfollows (notethat noneis reflectedin the orthography).

1 Obstruentvoice assimilation,in which an obstruentassimilatesin voice

to an obstruentto its right, e.g. HAT-BAN hatban [hiidbiin] 'in six';
excludedis v, to the left of which obstruentsdo not assimilatein voice,
e.g. HAT=VAN hatvan[hiitvan] sixty'.
2 Adaffrication, in which any sequenceof core stop or affricate plus core
 HUNGARIAN 435

Table 14.2 Hungarianabstractconsonants

Peripheral core
a b

L LJ
M N NJ
PK C T TJ
BG C D DJ
F H S SH
V Z ZH
C CH
X XH
R j

fricative yields a geminateaffricate, e.g. HAT-S3R hatszor[haccor] 'six

times'; with obstruent voice assimilation: NEEDJ=S3R negyszer
[neetitjser] or [neeccer]'four times'.
3 Adpalatalization,in which [- palatal]consonants(columna in Table 14.2)
are replaced by their palatal equivalentswhen to the left of palatals
(column b) or j, e.g. t is replacedby t j to the left of ni in #OT NJULA
Ot nyul 'five hares'. Adpalatalizationis most consistentin the caseof the
oral and nasalstops;its applicationin sequenceswhosefirst memberis
I, or whose secondmemberis j, or both, is more complicated,mostly
becauseotherprocessescompete;seeKiefer 258 ff.
4 j-Assimilation: when the input of adpalatalizationinvolves j as second
member, impermissablesequencessuch as *njj are the output. These
sequences,as well as all sequencesconsisting of a core fricative or
affricate plus j, are automaticallyfed into j-assimilation,which converts
them into geminatesof the first consonant,e.g. *njj > nini, *sj > ss.
Examples:AD-IA-NAK adjanak[acPdjfulak] 'let them give; they should
give', RAAZ-0-IAA razza[raazza] '(s)heshakesit'.
5 Hissing-to-hushingassimilation,parallelto adpalatalization,involves the
assimilation of a fricative in column a to a following fricative from
columnb, e.g. OLAS #SHORolaszsor [0Iass6·r] 'Italian beer'.

There are also morphophonemicoperationswhich apply only to specific

morphemes, such as (6) complete assimilation of the v initial in the
instrumentaland translativesuffixes to a consonantto their left, e.g. VASH-
VAL vassal[vassa·l] 'with iron', #TOK-VAL tokkel [t6kkli·l] 'with pumpkin'
and (7) completeassimilationof the z final in the demonstrativeprounouns
az 'that' and ez [liz] 'this' to a consonantinitial in a morphemeattached
to its right, e.g. effajta [affajta] 'this kind of. As the examplesindicate, the
output of complete assimilation is normally indicated in the orthography.
436 HUNGARIAN

Seealso the subjunctive,underverb morphology,p. 449.

The historical backgroundof the Hungarianconsonantsis complexbut for
the mostpartclear.The bestsurveyis Kalman 1965; seealsoLak6 (1965) and
Sarnmallahti(1988: 490-94,501-2,510-11,515-20).
The correlation of palatalizationamong the consonantsis built upon the
proto-Uralic state of affairs. Distinctive voice on the other hand is an
innovation, and must have developedafter Hungarianbroke away from the
rest of Ugric. We may outline the most characteristically Hungarian
developmentsas follows:

1 The voiced palatal stop dj comesfrom *j in initial position (gyalog 'on

foot'), but root-internallyit can comenot only from *-j- (fagyal 'privet')
but also from a variety of palatalized consonantismsincluding *-&-
(hagy- 'leavesbehind'), *-cL (vigyaz- 'watchesclosely') *-njcL (magyar
'Hungarian')and *-lL (meggy[medjeP] 'morello cherry').
2 The remaining voiced stops (b d g) stem from clusters of nasal plus
homotopicstop,e.g. eb 'hound' = TIj (= Tremjugan)Khanty iiiimp 'dog',
had 'army, host' = Trj Khanty kant=;;JY 'Khanty', szeg- 'slices (bread);
hems' = Trj Khanty siiIfkw - 'strikes'. There was also some voicingof
initial stops, particularly in roots containing such medial clusters, e.g.
dug- 'inserts' = Finnish tunke-. Voiced stops later becamecommon in
initial position in affective vocabulary(bub 'crown, crest', baba 'doll',
bab 'puppet') after the example of first Turkic, then Slavonic loans.
Hungarianz is of parallel origin, viz., orginally from p(F)U intervocalic
*+ (haz 'house'),thenspreadingto initial positionin Slavonicloanssuch
as zalog 'pledge'.
3 p(F)U initial *p- gave Hungarianf- lfagyal) and *k- gave h- to the left
of originally back vowels (had; h{z- 'grows fat' = Vach Khanty kaat=
;;Jyt-). Of the p(F)U initial stops, only *t- persists uniformly into
Hungarian(tel, oblique stemtele- 'winter' =SosvaMansi taal).
4 The non-palatalizeds(h)ibilants p(F)U *s- and pFU *8- fell together,
presumablyin Ugric, althoughSamoyedicshowsno clearevidencefor an
*S/*8 distinction, either. In any eventthe two s(h)ibilantshaveboth been
lost in Hungarian.Thus we havezero initial in e.g. {n (oblique stemina-)
'sinew' = Trj Khanty lacin = Komi sen, and eger (oblique stem egere-)
= TIj Khanty liilfkw;}r = Komi sfr. Present-dayHungarians reflects *sj
in p(F)U vocabulary(szem[sem] 'eye'),while scomesprimarily from the
p(F)U affricates *c- (sovany 'thin'), *-c- (kes 'knife'), and *-d- (Bs
'ancestor').The asymmetryof the pU s(h)ibilantlaffricatesystem,*s ::
*sj : *d :: *c, is perhapsreflected in two kinds of Hungariandoublets
s
involving which have been ascribedto internal borrowing: (a) those
with s- - so, such as szav- 'weaves'- saveny'hedge',or szem 'eye' -
samar 'herpes',and (b) those with s- - Co, such as sajog- - csillog-
'glitters, shines'.
 HUNGARIAN 437

5 From pU to pre-Hungarian,non-initial consonantstendedto weaken.For

single consonantsthis meant at least a decreasein the degree of
constriction. Thus p(F)U intervocalic *+ gives Hungarian z, and the
regulardevelopmentof p(F)U intervocalic *-m- is v « earlier *w); non-
initial -m- seemsalwaysto havehad supportfrom an adjacentconsonant,
as in szem[seml 'eye' = Finnish silma. In fact, most original consonant
clusterspreserveone memberat the expenseof the other, e.g. kOt- 'ties,
knits' = Finnish kytke-, kh 'two'= Udmurt kikt-, mos- 'washes' =
Estonianmoske-,()sZ 'autumn' = North Saamicakca. The development
of the clustersmentionedabove at (2), and of the long consonants(6,
below) may also be seenin this light.
6 Hungarianprovidessomefairly goodevidence- betterthanthat found in
the ObUgrian languages- for a distinction betweenshortandlong medial
stops.The latter mostly date from pFU; they are especiallyfrequent in
affective vocabulary (onomatopoeic verbs, kinship terms) such as
fak=ad- 'bursts'< pFU *pakkV- (re-createdin Hungarianpukkan-'pops,
explodes'),ap=a 'father' - (dialect) ip=a 'father-in-law' = Finnish appi
(oblique stemappe-) 'father-in-law'.
7 Some of the morphophonemicpropertiescharacteristicof certain con-
sonantsmay be tracedto their phonologicalroles in earlier stagesof the
language.For example, v, although in present-dayHungarian an obs-
truentopposedin voice to f, doesnot induceobstruentvoice assimilation.
This is in keeping with the fact that in morpheme-initialposition v has
developedfrom non-distinctivelyvoiced *w. With h, the reverseis true:
this segment now has no voiced phonemic pendant, yet it induces
obstruentvoice assimilation.This morphophonemicbehaviourreflectsits
earliersystemicstatus:h developedfrom *x, which in pre-Hungarianwas
opposedto *y « non-initial *w and *k).

Morphology

StemArchitectonics
Most Hungarianmonomorphemicverbsare monosyllabic;all verbsareeasily
accommodated, in a synchronicanalysis,as endingin a consonant.Nominals
are more various. Severalbisyllabic and even a few trisyllabic stemshave
cognatesin ObUgrianor beyond,e.g. fekete 'black' = Vach Khanty peyt;} <
*p#VkkVttV.
The morphophonemicoperations triggered by suffixation are largely
predictablein terms of interaction betweenvarious kinds of stem-typeand
suffix type. The primary stemtypes are (A) stablev. (B) unstable,and within
the latter, (Bl) epenthesizingv. (B2) non-epenthesizingstems (see the
sectionon vowel oppositionsand alternations,p. 431-3). The greaterformal
diversity of nominal stems is enhancedby the presenceof A-final stems,
438 HUNGARIAN

which havenumerous,but complementarilydistributed,subtypes(Abondolo

1988: 185).
In inflection, the sole morphologicalprocessis suffixation. Derivational
morphologyis also overwhelminglysuffixal, but thereare also variouskinds
of compoundand there is one prefix: see Pronounsand the deictic system,
p.443.
Inflection of Nominals
The word-classof nominalsconsistsof nouns,adjectives,numerals,deictics,
and pronouns.The inflection of all of thesecategoriesoverlaps,but eachhas
distinctive features; the inflection of the personal pronouns is the most
deviant. Within the subcategorynoun, names show distinct inflectional
properties.
Person(= Possessive Suffixes)
A nominal may be markedfor any of six person-numbercategories,or left
unmarkedfor person(-0-). The possessivesuffixes are sl -MA, s2 -DA, s3
-IAA, pI -UNKA, p2 -T3KA, p3 -IUKA.
A few sampleforms will illustrate the types of sandhiwhich occur when
thesesuffixes are attacheddirectly to the stem.As notedabove,AA alternates
with A, and A with zero, e.g. SOBAA-MA szobtim [sobaam] 'my room',
SOBAA-IAA szobtija [sobaaja] 'hislher room'. Epenthesisof 3 occurs
betweenconsonants,e.g. ABLAK-MA ablakom [ablakom] 'my window',
#GOOG-DA gogod 'your arrogance'.Suffix-initial U cancelsanA to its left,
e.g. HAAZA-UNKA kazunk[haazuuk] 'our house',but is itself deletedby a
preceding AA, e.g. SOBAA-~KA SOBAA-~KA szobtink [szoba·uk] 'our room'. The
initial I of the third-personsuffixes -IAA, -IUKA is deletedto the right of
stem-finalpalatalsandcorefricatives andaffricates,but persiststo the surface
(as j) to the right of all vowels saveA, e.g. LAANJ-IUKA lanyuk [laaniuk]
'their daughter', TANAACH-lAA tanacsa [tanaaca] 'hislher advice',
ORVOSH-IUKA orvosuk[orvosuk] 'their doctor',but KAAVEE-IAA kdveja
[kaaveeja]'hislher/itscoffee'. In the sequenceA-I, the destructionis mutual,
e.g. HAAZAIAA kaza [haaza] 'hislher house'.Otherwise,whetherthis I is
deletedor not is not entirely predictablefrom the shapeof the stem (Papp
1975: 109--63,Abondolo 1988: 213-21).
In six kinship terms,stem-finalAAis deletedbeforetheI of the third-person
suffixes, e.g. AN JAA-IAA anyja [an.-ittia] 'mother s3'. This operation has
analoguesin Mari andKhanty,wherethepersondistributionis thereverse.
Number
Plurality is specifiedby the suffixes -KA- or -1-. The latter suffix is usedwhen
possessivesuffixes are also present.Examples:SOBAA-KA szobtik[sobaak]
'rooms',SOBAA-I-MA szobtiim[sobaaiim] 'roomssl', #ASH=AT-KA esetek
[asatek] 'cases',#ASH=AT.IA-I-IM esetei [asatl=Pi] 'casess3'. As the last
example shows, certain sectors of the plural possessiveparadigm resist
 HUNGARIAN 439

agglutinating-analogue analysis.In the analysisoffered here, the possessive

plural morphemeis seenas intolerantboth of vowels to its right (witnessthe
wholesale cancellation of the entire bulk of the s3 suffix -IAA) and of
consonantsto its left. This latter intolerancerequires the addition of the
thematic sequence.IA to consonant-finalstemswheneverthey are inflected
with the plural possessivemorpheme.This subsectionof the Hungariannoun
paradigm has been interpreted in several other ways in the descriptive
literature,including the positing of a discontinuousmorpheme,e.g. Melcsuk
1968.
The history of both of the plural numbersuffixes is disputed.The -KA- is
either from a FU derivational suffix *=kkV which formed resultative and
collective nounsfrom nouns,e.g. North Saamictizet 'summerfootwear' «
*cjaaccee'water' plus *=k), Finnishpunakka'rubicund'<puna= 'red=' plus
*=kkA), or it is the same morphemeas that seenin the plural possessive
suffixes, e.g. pI -UNKA, as in htizunk [haazul)k] 'our house'< *kaata-ml-k
HOUSE-l-plur. For the historical backgroundof the possessivesuffixes, see
below.
The origin of the possessiveplural suffix -1- is also disputed.According to
one line of thinking, it is the same morphemeas that of the third person
singularpossessive,which doesindeedhave -i- allomorphsin a few areasof
nominal inflection. The other view holds that this -i- is historically identical
with the -i- which occurs in the oblique cases of Saamic, Fennic, and
Samoyedic(Papp1968: 136-8,Kulonen 1993: 56-7).
The possessivesuffixesdo not derivedirectly from proto-Uralic analogues.
The set of Hungarianpossessivesuffixes is perhapsbestreconstructedas the
resultof analogicallevelling which favouredoneor the otherof two originally
complementarysets of suffixes, one set marking singular, the other non-
singular possessions(Kulonen 1993: 72). In one version of this theory, the
proto-Uralic, or at leastthe proto-Finno-Ugric,noun paradigmis thought to
havehada suffix *-n- which indicatednon-singularityof the possession,with
the result that there were opposingforms such as *kala-ti: 'your (one) fish' :
*kala-n-t'i 'your (several)fish' (cf. Korhonen 1981: 234). In Hungarian,no
traceof *kala-ti: remains(but cf. North Saamiguolltit, Finnishkalasi), andthe
form halad 'your fish (one or several,indifferently)" which lacks reference
to explicit plurality, is the direct and regulardescendantof a form which was
orginally explicitly non-singular.A parallel collapse occurred in Finnish,
leading to syncretism:Finnish has distinct forms for 'a fish' and 'fishes' but
only kalasi 'your fish(es)'. In Hungarian,on the other hand, the collapseof
the original singular/pluraldistinction, basedon *-0-/*-n-, was remediedby
its replacementwith the -i- plural mentionedabove.
If this theory is correct, the levelling which occurredin Hungarianis in
nearly perfectcomplementarydistribution with the levelling thoughtto have
occurredin Finnish.An exampleis providedby the second-and third-person
singularforms:
440 HUNGARIAN

pU Finnish Hungarian
*kala-u kalasi
*kala-n-t'i halad

*kala-sa hala « *xal::>j::»

*kala-n-sa kalansa
Case
The precisenumberand inventory of the casesuffixes is a matterof dispute,
but we will not err grosslyin positingsixteen,of which ten (groupB in Figure
14.1) are primarily spatialin meaning.
The categoriesand cells in Figure 14.1 have been arrangedso as to
highlight formal and functional correspondences and overlaps.The column-
labelsStasis,Source,and Goal arethus somewhatexaggeratedwhentakento
refer to the more grammatical,more abstract, suffixes of section A. The
traditional namesfor the casesare (1) nominative, dative; (2) accusative,
instrumental,translative; (3) causal/final; (4) inessive, elative, illative; (5)
superessive, delative,sublative; (6) adessive, ablative, allative; and (7)
terminative.Therearealsotwo otherspatialsuffixeswith limited distribution,
a locative -T(T) used only with certain toponyms and with deictics (e.g.
mogott 'behind', Vac-ott 'at (the town of) Vac (in Pest county)" and a
separative/locative-L found in postpositionsand as a building block in
composite case suffixes (e.g. mogUl 'from behind'; for elative -BOOL,
delative-ROOL, ablative-TOOL, seebelow).
In unguardedspeech,the illative usually performs the functions of the
inessive.
The morphophonemicoperationstriggered by the case suffixes vary in
accordancewith (1) phonological or (2) morphological (i.e., historical)
conditions. Broadly speaking,all case suffixes (other than the exceptions
discussedbelow) attachto a noun stem by meansof operationsidentical to

Figure14.1 Hungariancasesuffixes

Stasis Source Goal

1 gramm-l -0 -NAK
A 2 gramm-2 -T -VAL -VAAH
3 cause/aim -EERT

4 interior -BAN -BOOL -BAA

5 surface -N -ROOL -RAA
B
6 proximity -NAAL -TOOL -H3Z
7 terminus -IG
 HUNGARIAN 441

those seen in the nominative, e.g. SHAATRA sator [saator] 'tent sN',
SHAATRA-BAN satorban[saatorbi'm]'in a tent'. The exceptionsare:

1 The accusative,which attachesin the mannerof the possessivesuffixes,

e.g. SHAATRA-MA satram [saatrfun] 'my tent', SHAATRA-T satrat
[saatrat] 'tent sA', exceptthat no epenthesisis required to the right of
stem-finalpostvocaliccore continuantconsonants,e.g. BETON-T betont
[batont] 'concretesA' but CITROM-T citromot [citromot] 'lemonsA';
2 The superessive,which selectsthe non-nominativeshapesof epenthetic
stems,e.g. TITK-N titkon [titkon] 'secretsSUP',and,in poeticlanguage,
of A-stems, e.g. #AGA-N egen [agan], cf. standardegen [eegen] 'sky
sSUP';
3 The instrumentaland the translative,whoseinitial V assimilatesto any
preceding consonant, e.g. CITROMH-VAL citrommaZ [citromma.I]
'lemonsINST', VASHA-VAAH vassa[vassaa]'iron sTRANS'.

As is suggestedby their behaviour in the present-daylanguage, the

Hungarian case suffixes fall into two basic historical types, primary and
secondary.The primary suffixes all derive from morphemesdating from at
leastthe FU period.Theseare:

1 The accusative-T, which was originally an encliticized demonstrative

pronoun, either proximal or, more likely, distal. Hungarian HALA- T
haZat[Mlat] 'fish sA' was thus originally somethinglike *kala--tV FISH-
-THAT 'that fish', wherein 'that' had at least as much anaphoric as
demonstrativeforce. The closesttypological parallel is found in Mordva,
where encliticized demonstrativeshave servedas the building blocks of
an entire definite paradigm,e.g. Erzya Mordva kaF--si 'fish sNdef', kaZ-
O1t--f 'fish sGdef'.
2 The superessive-N, which continuespU *-nA, as in Finnish koto-na 'at
home',North Saarnigodi-i-n 'tentpLoc', Selkupnloo-qi:n 'floor sLoc'.
3 The instrumental-VAL, which probably derives from an old locativel
separativeform of the FU root *woo 'strength;people'.A form suchas
KEESH_VAL kessel [keessa·l] 'knife sINST' would then come from
earlier(*) KNIFE + STRENGTH-Ioc,i.e. *ktici + woo-IV» *keec:rya1il
» *ked)'iil > [keessa·l].
4 The translative -VAAH, as in #KOVA-VAAH kove [k66vee] 'stone
sTRANS', KEESH-VAAH kesse[keessee]'knife sTRANS'. This suffix
is usually explainedas simply a continuationof the pU lative suffix *-j,
with pendantsscatteredthoughoutthe adverbial and deictic portions of
the Hungarianlexicon, e.g. the e/a of ide 'hither', oda 'thither' and the
e of mage 'to behind'; see the sectionson postpositionsand deictics
pp. 443 and 445. Its initial V has beenexplainedasfaussecoupe (Papp
1968: 155-6), but given the assimilatory behaviourof this V it seems
442 HUNGARIAN

more likely that -VAAH is actually the sum of two latives, *-kI plus *-j,
e.g. *ktici-ki-j » *keec~y.:lj*keesyej
» *keesyej> [keessee].
5 The terminative-IG. The origin ofthis suffix is unclear.On the onehand,
its morphophonemicbehaviouris characteristicof secondarysuffixes; on
the other hand, it lacks postpositionaland person-inflectedpendants,in
the manner of primary suffixes. It may continue the lative suffixes
mentionedin connectionwith the translative,but in the reverseorder, or
it may be the result of a strengtheningof the pU lative *-1). Neither
explanationaccountsadequatelyfor the vowel. Paralleldevelopmentsof
*1) in deictics, e.g. Hungarianmog- < *mul)a 'behind', renderthe latter
explanationslightly morecredible.

The remainingcasesuffixesareall clearly secondary.The adessive-NAAL

is a composite,madeup of a pronominalbase*na plus a locative/separative
*-1 (as seenin the instrumentalsuffix, above).This pronominalbasemayor
may not be connectedwith *na, a pronominalbasewhich servedas the initial
componentof the Hungariandative (-NAK; cf. Selkup -ni'kl-ni'g) and was a
building block for local casesin CommonUgric (Riese 1992; Mikola 1975:
164-170). The interior suffixes -BAN, -BOOL, -BAA are eroded and
fossilized compositesof a nominal base (cf. #J3ALA 'interior, gut') plus
locative *-nA, locative/separative*-1, and lative *-j (Kulonen 1993: 84). The
delative (-ROOL) and sublative (-RAA) are also thought to derive from a
noun plus local suffixes, but the identity of this noun is disputed.It is most
likely that the noun in questionis found in Mansi (UEW 883; Kulonen 1993:
84) andin the Mordva superficialpostpositionlag-, e.g. ErzyaMordva lag-so
'on', lag-s 'on to'. The ablative has a parallel background:it continues a
locative/separativeform of FU *ttil)i '(area around the) base (of X)" cf.
EasternMari tUg 'base(incl. of tree)' and the Komi postpositionaldin, as in
pu din-iSi 'awayfrom the tree'.The allative -H3Z is formed from a basewhich
is identical with a Khanty spatial noun, seen in e.g. Tremjugan Khanty
kuutJ=fJg 'adjacentspace'and the SherkalKhanty postpositionxosi-aa (with
an -aa lative); seeHonti (1984: 81). Causal/final-EERTis presumedto be an
old locative form of a noun with a spatial/local meaning, but no credible
etymonhasbeenput forward.
Personalforms ofcasemorphemes.Of the sixteenproductivecasesuffixes
presentedabove, twelve have independentanalogueswhich take person
suffixes,therebysuppletingdefectivepersonalpronounparadigms.

-NAK -VAL -EERT

-NAK
~AK- IIyALA- II£ERTA-

-BAN -BOOL -BAA

#J3ANNA- #J3ALOOLA- #J3ALAA-
 HUNGARIAN 443

-N -ROOL -RAA
RAITA- ROOLA- RAA-

-NAAL -TOOL -H3Z

NAALA- ~OOLA
~OOLA HOZZAA-

Note the stemRAITA- « *raI)-tV, with old locative *-t) suppletiveto -N

in the superessive.Examples: #TOOLA-MA t/JZem [tool fun] 'from me',
ROOLAUNKA r6Zunk [rooluIJk] 'about us', HOZZAA-DA howld
[hozzaad]'towardsyou'. The dative~AK- ~AK-IAA standsout as the only stemin this
classto have afinal consonant.It is probablynot a coincidencethat its third-
personforms are deviant, e.g. ~AK-IAA~AK-IAA neki [naki] 'to himlher/it' instead
of expected(*neke [niika]), but for anotherinterpretationseeKulonen (1993:
56,85-6).

Pronounsand the Deietie System

Hungarianhasthreearticles,definite AZ (- A beforeconsonants),singulative
EDJDJ egy, and zero. The definite article implies that its noun exists. Thus a
zany-omis 'my daughter(who exists)" whereaszany-om,with zero article,
leaves the question open, witness Nines Zany-om DOES.NOT.EXIST
DAUGHTER-s1 'I haveno daughter'.The singulativearticle individuatesan
instanceof a memberof a categorynot yet alluded to (Sherwood1996: 26),
e.g. egy zanyom'a daughterof mine'. The developmentof the definite article
from the demonstrativeAZ can be followed through thephilological record.
The singulativearticle is etymologicallyidenticalwith the numeralEDJDJegy
'one'.
Somewherebetweeninflection and derivationhoversthe suffix =EE. This
suffix descendsfrom diphthongsarising from the combinationof stem-final
vowel with the lative case suffix *-j, but in the present-daylanguageit
convertsany noun stem, with or without personsuffixes, into a possessive
pronoun,e.g. HAAZA-UNKA=EE htizunke[haazuIJkee]'that (e.g. the roof)
of our house'. Such pronouns may then be fully inflected for case,
e.g. TART=ALMA-IAA=EE-NAAL HOLDS=NdV-s3=EE-adetartaZmaenaZ
[tfutaImaaeenaal]'(e.g. the impressiveness of its form is greater)than that of
its content'.Iteration of =EE (Peteree'that of that of Peter'),andpluralsbuilt
to it with both -1- and -KA (Peterei 'thosethings of Peter',Peterek'Peterand
his associates')vary from variety to variety; seeLotz (1967) and, for a recent
stab at an analysis,Kornai (1994: 124-6). The non-occurrenceof this suffix
in the philological record until the mid-fifteenth century is attributed to
chance(Bemir 1957: 20).
The basesof the personalpronounshave only nominativeand accusative
forms, the restof the paradigmbeingsuppliedby person-inflectedstem-forms
of the case suffixes (see above). The first- and second-personsingular
pronounsexhibit aberrantconsonantalandvocalic alternations,andin all non-
444 HUNGARIAN

axis-of-discourseaccusativesthe appropriate person suffix is obligatory

(while, at leastin the singular, the casesuffix may be omitted):

Nominative Accusative

sl #AAN en #ANG-MA( -T) engem(et)

s2 TE te TEEG-DA(-T) teged(et)
s3 #OVA 0" #OVA-T ot

pI MI mi MI-UNKA-T minket
p2 TI ti TI-T3KA-T titeket
p3 #OVA-K ok #OVA-KA-T oket

There are also correspondingpossessivepronouns sl #ANJEEMA, s2

TIADA, s3 #OV=EE, pI MIEENKA, p2 TIAT3KA, p3 #OVA=EE-KA; these
occur always with the definite article, e.g. AZ #ANJEEMA az enyem'(that
which is) mine'. Note that the final sequenceOVA of the third-person
pronounbehavesin a mannerwhich is the reverseof that seenin the noun,
cf. #KOVA-T kovet [k6vat] 'stone(ace)',#KOVA=EE k(/e [kooee] 'that of a
stone'.
Demonstrativepronounsdistinguishproximal #AZ from distal AZ; eachof
these may be intensified by the prefixation of AM--, viz. AM--# AZ emez
[amiiz] 'this (oneevencloserthan ez)',AM--AZ amaz[funaz] 'that (oneeven
further away than az)'. The Z final in thesepronounsassimilatesfully to the
initial consonantof mostcasesuffixes (and of six othermorphemes,Szilagyi
1980: 115), e.g.AZ-BAN abban [abban] 'in that', #AZ-NAAL ennel[anneel]
'at this, than this', #AZ+KOR ekkor [iikkor] 'this time', AZ+FAITAA affajta
[Mfajta] 'that kind (of)'. There is no such assimilationin the accusativeor
superessive,e.g. #AZ-N ezen[iizen] 'on this', AZ-T azt [ast] 'that (ace)'; the
terminativeexhibitsthe uniquealternationZ - DD, viz. #AZ-IG eddig [addig]
'up until this'.
The rest of the spatial deictic systemconsistsof parallel proximal/distal
pairs expressedby frontlback pairs of forms, e.g. #AZ=NJI : AZ=NJI ennyi :
annyi 'this much' : 'that much'; I-TT : O-TT itt: ott 'here' : 'there';I=DAA
: O=DAA ide: oda 'there' : 'thither'; I=e=AN: O=e=AN ilyen: olyan 'like
this' : 'like that'; II=D J : UU=DJ fgy : ugy 'in this way' : 'in that way'.
Interrogativepronounsare built to the basesKI, MI, and HO=, all from
Uralic stems. They distinguish animate (ki 'who?') from inanimate (mi
'what?'), and coultablefrom non-countablequantity (mennyi 'how much?',
hany 'how many?').There are interrogativependantsfor all proximal/distal
deictic pairs, usually but not always built with identical or similar suffixes,
e.g. ME=WNJI me.myi'how much?',MI=eAN milyen 'whatkind of?', HO=
NNAN honnan'whence?',HO=DJhogy 'how?' (cf. ennyilannyi,ilyen/olyan,
innen/onnan,fgy/ugy above).
 HUNGARIAN 445

Relative, negative, indefinite, and indifferent pronouns are built to the

samebases.Most relative pronounsconsistof the appropriateinterrogative
base precededby invariable A, e.g. A+KI aki [aki] 'who', A+ME=NJNJI
amennyi [ameninji] 'as much as'; negative pronouns have SHE(N)+, e.g.
SHE+HOL sehol [seho·l] 'nowhere', SHEN+KI senki [seIJki] 'no one';
indefinite forms have VALAA+, and indifferent forms have AKAAR+, e.g.
VALAA+KI valaki [vi'tlaki] 'someone',AKAAR+KI akdrki [akaarki] 'any-
one (it doesn'tmatterwho)'. The derivationof severalparadigmaticmembers
deviatesfrom the agglutinatingmatrix just outlined,yielding forms which are
synchronicallymore or less opaque,e.g. soha [soM] 'never' « *SHEN 'not
+ HA 'if/when').

Postpositions,Coverbs,and OtherAdverbials
There are about twenty-five widely used postpositions.Most are perhaps
betterclassifiedas dependentadverbs(Lotz 1939: 101), for they occur only
in combinationwith nounsor with personsuffixes. Thereare also a few pure
postpositions,i.e. forms which occur only with nounsand neverwith person
suffixes, e.g.OOTA ota [oota] 'since'.
A small subgroupof eight postpositionsdistinguishesthree(or, sometimes,
two) locational/directionalmodesby meansof marginal casesuffixes, e.g.
ALA- TTA alatt '(located)under',ALA-OOLA aMl 'from under',ALA-AAH
ala 'to under'. The remaining postpositionshave only one form each, e.g.
DJANAANT gyanant[dianant] 'as a', or, if two forms are used,at least one
of them has only abstract, non-spatial, reference,e.g. #YEEGA-IG vegig
'along' : #YEEGA-TT vegett'with an aim to'.
Most postpositionsoccur with their lexical noun in the nominative,but a
few govern the superessive,e.g. TUUL 'beyond',AAT 'across,through',
#BALUL 'within', as in AZVAAROSH_N ~ALUL ~ALUL a varosonbeW 'within
the city'. In stylistically marked utterances,a subset of this group of
postpositionscan also occurbeforetheir noun, e.g. TUUL AZFOUOO-N tul
a folyon 'beyondthe river'. One postposition,KEEPASHT 'comparedwith,
considering', takes the allative, e.g. KOR-IAA-H3Z KEEPASHT korahoz
kepest 'for hislher age'. In older varieties of Hungarian, postpositions
commonly enteredinto possessiveconstructions,e.g. szeretet-nekmiatt-a
insteadof szeretetmiatt 'on accountof affection'; seeSyntax,p. 450.
Coverbs are a special class of adverbial modifiers which form loose
compoundswith verbs. There are about forty coverbs; many share root
morphemeswith postpositions,e.g. #ALOO ela [aloo] 'to the fore', cf.
postposition#ALOO-TT elOtt [aloott] 'in front of, before'. The productive
noun casesuffixes -RAA, -IG and -BAA are evident in a few coverbs,e.g.
#ALOO-RAA elOre 'in advance',#KOZ-RAA kozre 'public', e.g. #KOZ-
RAAIAD- kozrelGIVES- 'publishes',#YEEGA-IGIFUT vegigfut- 'runs the
length of' , cf. the postpositionvegig 'along'. Othercoverbsare (often shorter
variantsof) third-personforms of casemorphemes,e.g. neki 'to it' in ~AK-(often
446 HUNGARIAN

IAAILAAT- nekilat- 'seesto', rajta 'on it' in RAIT~-rAAIKAP-RAIT~-rAAIKAP- rajtakap-

'catchesin the act'. Still othersaredeictic primitives, e.g. KI 'out(wards)',LE
'down(wards)',and MEG 'correctly; back (to correctplace or state)',which
functionsmostoften as a perfectivizer.
Besidestheir semanticand aspectualfunctions within the lexicon, coverbs
also play an aspectualrole within syntax;seethe sectionon constituentorder,
p.4S1.
In their semanticand aspectualfunctions Hungariancoverbsoverlapwith
those of Mansi and Khanty, but there are few cognates.Aspectualparallels
may be found in Estonianand Slavonic, and Germanverbal prefixes have
often servedas modelsin calquing,e.g.fellfog-UPIGRASPS'comprehends'
(auffassen),belIat- INISEES 'understands'(einsehen),utanalnez-AFTERI
SEES 'checks(whetherX is the case)'(nachsehen).
Adverb formation. Many textually frequent adverbs are synchronically
monomorphemic,e.g. ma 'today',most 'now', mar 'now/then(as opposedto
before)" meg 'now/then (as before)" reg 'long ago', rogtOn 'immediately'.
Most commonly, however, adverbsare derived by one of three suffixes. In
decreasingorderof frequency,theseare as follows:

1 =AN, e.g. OJORSH=ANgyorsan [diorSan] 'quickly', cf. gyors 'quick',

SEEP=AN szepen 'nicely', cf. SEEP szep 'good-looking, nice'. This
suffix is thoughtto have arisenthrough a split in the function of the old
locative *-nA; cf. the superessivesuffix of the noun caseparadigm(and
OJORSH-Ngyorson [djorSon] 'on a fast (train)'). There is a parallel in
Permian:cf. KomilUdmurt instrumental-en/-en,inessive-in.
2 =UL, e.g. MAO JAR=UL magyarul 'in Hungarian',FINN=ULfinniil 'in
Finnish', KONOK=UL konokul 'stubbornly'.As the examplessuggest,
ethnonymsand pejorativespredominate.This suffix is often classifiedas
a noun case,with the meaning'as a, qua', i.e. as an essive.Historically,
it is probably from a suffix sequence *-kI-lV, i.e. LATIVE +
LOCATIVE/SEPARATIVE. Example:TAARSH-UL tarsul [taarsul] 'as
a companion'.
3 =LAG, e.g. KIIZAAR=OO=LAG OUTICLOSES=pres.part=LAGkizar-
61ag 'exclusively', VISOW=LAG RELATION=LAG viszonylag 're-
latively', *RAMEEL=HAT=OO=LAG HOPES=potential=pres.part=
LAG remelhet81eg'hopefully'.Adverbsformed with this suffix occasion-
ally contrastminimally with adverbsmadewith =AN, e.g. egyhanguan
[e~ha1Jguuan] 'monotonously', egyhangulag [e~ha1JguuIag] 'unani-
mously',cf. EOJOJ 'one',HANG 'sound,voice'.

Verb Inflection
Non-compositefinite forms of verbs in ContemporaryStandardHungarian
can be inflected either for mood or for tense.The moods are subjunctive
(-IA-) and conditional (-ANAA-), both neutral with regard to tense (=
 HUNGARIAN 447

universal tense). The tensesare past (-TTA-) and non-past(-0-, -S-), both
neutralwith regardto mood (= indicative).
Person suffixes are added to the right of the tense/moodsuffix. It is
traditional to speakof two conjugationsin Hungarian,the differenceresiding
in the type of direct object encoded.Here we shall simplify by saying that
there are two setsof personsuffixes, eachof which encodesa different sort
of relationshipbetweensubjectand object person.Suffixes of one set, here
termedcentrifugal,are usedwhenthe personof the objectis greaterthanthat
of the subject(e.g. first-personsubject/second-person object: 'I seeyou', or
second-personsubject/third-personobject: 'you seeher'). Suffixes from the
other set, termedcentripetal,are usedwhen the reverserelation obtains(e.g.
'she seesme', 'they seeyou'). Centripetalsuffixes are also usedwhen there
is no object at all ('they stand')or when the object is indefinite ('he makes
coffee'). This formulation is correct as far as it goes, but the selectionof
suffixal morphemesand the morphophonemicdetails involved in their
realization are both more complex than a binary split would suggest.For
example,the centripetalsuffix for first-personsubjectsis -3K, but with two
types of exception: (1) in the past tense, it is -3M, syncretizing with its
centrifugalpendant,and (2) for a large setof verbs(the deponents,calledikes
igek 'verbswith ik [in the third personsingularnon-past]'in Hungarian)it is
-3M or -3K in both tensesandboth moods,the choicebeing madeon the basis
of a complex interaction of sociolinguistic, lexical, and stylistic factors
(Abondolo 1988: 97-101). Figure 14.2 sets out the two sets of suffixes in
morphophonemiccode,with neitherphonologicalnor morphologicalopera-
tions carriedout, andpresentsa simplified picturein thatdeponentsuffixation
is not indicated. The following sample forms will illustrate some of the
complexitiesentailedin the productionof actualforms from theseanalogues:
OLVASH-0-IAA olvassa[olvassa] '(s)he/itreadsit (s33)',NEEZ-0-IAA nezi
[neezi] '(s)he/it beholds himlher/it (s33)', MEN-ANAA-EEK mennek
[menneek] 'I would (fain) go (s1cond)',JlyAT=IIT-ANAA-IUK vet{tenok
[vatiitanook] 'we would project it (p13cond)',JlyAT=IIT-IA-IUK vet{tsuk
[vatiicctik] 'let's/we should project it (p13sj)', LEV-IA-ANAK legyenek
[ledianii.k] 'they should be (p3sj)', MEN-TIA-M mentem[mentii.m] 'I was
going'.
There are also composite verb forms. These are rare in the standard
contemporarylanguage,with one exception: the past conditional, which is
formedby the addition of the fixed form volna to a finite past-tenseform, e.g.
MEN-TTA-M VOLNAA mentemvolna 'I would havegone(but 1didn't)'.
Suppletionoccursonly in the verb 'to be', which is also uniquein showing
ablaut of its stem vowel: VAL- (- past-tenseand conditional stem VOL-) :
subjunctivestemLEV-. The stemLEV- has a full paradigmwith the meaning
'becomes'.Two other stems,MEN- 'goes' and ilJov- 'comes',also show a
few irregularities.
448 HUNGARIAN

Figure14.2 Hungarianverb personsuffixation: agglutinatinganalogue

sl s2 s3

pI
-3M Hungarian Hungarian llliJ
-3D -0
pI -3M -3D I-IAA
Ips I sj ICd I pt Ips I sj I cd I pt Ips I sj I cd I pt

pI -UNK -T3K -ANAK I-K

:-UNK: -IAAT3K -IAAK
pI :-UNK:
-!UK

pI p2 p3

Key: s1, s2, etc. = subjectperson;ps = present,sj = subjunctive,cd = conditional,pt = past,P =

centripetal,F =centrifugalobjectmarking.

Non-finite Verb Forms,History of Verb Inflection

Thesearethreeparticiples,two gerunds,andan infinitive. Of these,the future
participle and the past gerund are stylistically marked and have a limited
distribution. In modal constructionsthe infinitive can take personsuffixes in
agreementwith the agente.g. ~AM SABADSABAD SEM-T HUNJ-NI-IAA NOT
PERMITTED EYE-acc SHUTS-inf-s3 nem szabadszemethunynia '(s)he
mustnot close(his/her)eye(s)'.
The infinitive -NI appears tobe an old formation consistingof a deverbal
noun in *=nA inflected with a lative suffix *-j; there is an exact parallel in
Permian (Kulonen 1993: 65). The gerunds =VAA, =VAAN are probably
cognatewith the Khanty gerundreconstructedas *=mAAn - *=rnIIn (Honti
1984: 59). The presentparticiple =00 probablydescendsfrom a verbal noun
built with *=pA, with cognatesin Fennic (e.g. -b in Estonianthird person
singular kuuleb '(s)he hears') and Saamic (e.g. -p in AIjeplog Saarni first
personplural kulla-p 'we hear'). The future participle =ANDOO is thought
to derive from the presentparticiple of stemsderived with the suffix chain
=AM3D-, which onceformed inchoatives,e.g. fut=amod-ik 'beginsto run'.
For the pastparticiple,seeundertense,below.
The historical backgroundof Hungarianverb inflection is not clear in all
of its details; here we briefly summarizea few of the less controversial
highlights. The past-tensesuffix usedin contemporary standard Hungarian
(-TTA-) is an innovation; it is basedon the inflection, with personsuffixes,
of the past participle =TTA. The past participle, in tum, descendsfrom a
 HUNGARIAN 449

proto-Uralic verbal noun, *=(n)tA. An older pasttensehas beenall but lost:

it survives only in ceremonialand jocular language,and is formed with the
suffix -AA-, which is a continuation of diphthongs arising from the
combinationof the stem-finalvowel plus a past-tensesuffix *-i-, which seems
to have evolvedfrom a verbal noun (Janhunen1982: 36). This *-i- has clear
cognatesin Permian and the languagesto the west, but it has also been
detectedin the final vocalism of the Hungarianconditional suffix -ANAA-.
The long unroundedvowel of this latter suffix probably derives from an
i-final diphthong;the conditional would then be the result of a relatively late
compoundingof the Finno-Ugric *-ne- conditional/potentialsuffix with the
past-tensemorpheme(Kulonen 1993: 64). We thus have a double shift in
function, from tenseto moodandfrom participleto tense:

StageI StageII ContemporaryHungarian

Mood suffix .)"*-ne-i conditional-mennel 'you would go'
Tensesuffix *-i- past/'.)" *-tta- past • mentel 'you were going'
Verbal noun *=(n)tA /'

The present-tensezero suffix -0- has parallels in all Finno-Ugric lan-

guages.The presenttensein -S-, which is usedonly with the seventextually
frequent verbs TEV- 'does/makes',VEV- 'takes ([-] locomotion)" JtyIV-
'takesaway ([+] locomotion)" EV- 'eats',IV- 'drinks', IIJIIv- 'believes',and
LEV- 'becomes',is the continuationof a p(F)U durative,or de-perfectivizing,
derivationalsuffix *=s\ and may also be concealedin the present-tense stems
of MEN- megy-'goes'(with gy < *n-sj ) andVAL- vagy- 'is' (with gy triggers
< *l-~).
The subjunctive (-IA-) descendsfrom proto-Uralic*-k(A-); it triggers
several morphophonemicprocesses,including the above-mentionedad-
palatalizationandj-assimilation,but alsoprocesseswhich areuniqueto it. For
example, to the left of the subjunctive suffix a verb-stemfinal T > C H if
precededby a consonantor a long vowel, but> SH if precededby a short
vowel, cf. #VAT=IIT-IA-IUK vet{tsiik [vatiicctik] 'we should project it',
JtyAT-IA-IUK vessiik [vasstik] 'we should cast it'. Another example: in six
stems,final V melds with the initial I of the subjunctivesuffix to form the
segmentdj , e.g. LEV-IA-3N legyen[lecPen] '(s)he/itshouldbe'.
SinceHungarianfinite verb forms encodenot only subjectpersonbut also
featuresof the direct object, it is not surprisingthat many of the suffixes are
innovations, and that some of them have no exact analogueselsewherein
Uralic. Oldest are the singular centrifugal suffixes -3M, -3D, and -IAA, all
three of which are historically identical with the correspondingpossessive
suffixes.Thus the third-personcentrifugal-IAA seenin, e.g.,adja '(s)hegives
it' in all likelihood continuesa verb form with a third-personsuffix *-sA,
e.g. *'imta-sa » *ad;)j;) > *adj;) > [iicPcPii]. The suffix which encodesfirst-
personsingular subjectand second-personobject is clearly a relatively late
450 HUNGARIAN

compositemade up of the (secondary)second-personmarker -3L plus the

first-personsuffIx -3K, for both of which seebelow. Mordva, which is the
only Uralic languageto havea paradigmaticslot correspondingto this suffIx,
encodesit with etymologically different material, but in analogousfashion,
viz. ErzyaMordva katltlan =HungarianhagyiliakLEAVES-s2-s1'I'll leave
you' (cf. Tlilos 1975: 46). The plural centrifugal suffIxes all appearto be
secondaryformations: the -j- elementthat has crept into them all probably
did so by analogy from the third person singular, but the preciselines of
descentfor the remaining portions of these suffIxes remain unclear (see,
however,Honti 1984: 75-6 for a Ugric etymology of the first-personplural
form).
The centripetalforms are for the most part straightforward.The unmarked
second-and third-personforms, i.e. respectivelythe imperative and indica-
tive, were marked with zero. The marked forms were built up later, in the
second-personnon-imperativeby meansof suffIxes which were originally
durative/deperfectivizingin function (-3L, -AS; seealso derivation, p.452),
andin the third-personimperativeby meansof suffIxation of the third-person
pronoun *sVn, i.e. *"imta-ka-sVn » *ad~~j~n[iididion]
» [iididion] adjon '(s)he
should give' (cf. Kulonen 1993: 92-3); the third personplural -ANAK is a
later secondarypluralizationof this form. The first and secondpersonsplural
appearto be continuationsof the pluralizedpersonsuffixes *-mI and *-tl.
The first person singular -3K (along with its offspring, the K of s12
-ALAK), however, is an enigma. Ku10nen lists three explanations,none of
them convincing (1993: 89-90); a few more may be found in Papp (1968:
174-7).

Syntax
In its simplestform, the Hungariannoun phraseconsistsof a nominal head
(with inflectional suffIxes and/or postpositions,any or all of which may be
zero) precededby slots (any of which may be empty) for quantifiers (QN),
counters(CN), qualifiers (QL), emphatic pronouns(PRO), articles (ART),
and demonstratives(DEM). Some of the combinatory possibilities are
illustratedbelow, where the English equivalentsare (1) 'my key', (2) 'these
threelargekeysof mine',(3) 'concerningthesekeys of mine',(4) 'concerning
this rusty bunchof keys', (5) 'concerningthesethreelargekeys'.

DEM ART PRO QN QL CN QL N-sxx (PP)

1 az en kulcs-0-om-0
2 ez a harom kulcs-0-om-0
nagy
3 ez-ek-rol a kulcsa-i-m-rol
4 er-rol a rozsdascsomo kulcs-0-0-rol
5 er-ro1 a harom nagy kulcs-0-0-rol
 HUNGARIAN 451

As theseexamplesshow, the only membersto show agreementwithin the

noun phrase are the demonstratives,which agree with the head noun in
numberandcase(3); furthermore,the presenceof a quantifierhigherthanegy
'one', or of a counter,precludesplural marking (2, 4, 5). As examples2-5
suggest, occupancy of the demonstrativeslot by AZ or # AZ forces the
presenceof the definite article in the following slot, but in stylistically marked
registersthere are other demonstrativeforms which precludethe article, e.g.
(2a) e Mrom nagykulcsom.
Noun phrasesmay be linked in a possessiverelation by the suffixation of
the s3 personsuffix to the possessed, with or without suffixation of the dative
to the possessor,e.g. a kalap-om(-naka) karima-ja D.A HAT-sl(-dat D.A)
RIM-s3 'the rim of my hat'. The longer constructionis obligatory in the
expression of multiple possessive relations, e.g. a rovid hangzo-k
idfJ+tartam-a-nakarany-a D.A SHORT VOWEL-plur TIME+DURATION-
s3-datPROPORTION-s3'the ratio of the durationof long vowels' and when
the possessed precedes the possessor (hUg-a az apa-m-nak
YOUNGER.SISTER-s3D.A FATHER-sl-dat 'my father's younger sister')
or is separatedfrom the possessorby the predicate(apa-m-nakmeghalta
hUg-a FATHER-sl-dat DIED D.A YOUNGER.SISTER-s3 'my father's
younger sister died'). Otherwise both constructionsare possible, but with
pragmaticand textual distinctions, some of which entail the anaphoricand
global functionsof the article.
Qualifiers in the noun phrasemay be participleswith their own adverbial
andobjectcomplements,e.g. a nepi muveltseg-ben mutatkoz=okulonbseg-ek
D.A FOLK CULTURE-ine IS.EVIDENCED=pres.partDIFFERENCE-plur
'differencesevidencedin folk culture', or they may be qualifiers derivedby
meansof the suffix =1 from postposition-finalnoun phrases,e.g. a vacsora
utan-i ora-k-ban D.A DINNER AFTER=adjsxHOUR-plur-ine 'in the hours
after dinner'.
More sophisticatedaccountsof the noun phrasemay be found in Komai
1985 and Szabolcsi1987, both with good bibliography; on countersseenow
Beckwith (1992).

SyntacticRoles,the Useof CaseSuffixesand ConstituentOrders

Subjectsare normally markedwith zero (-0) and direct objectswith -T, e.g.
a jel=z=B-f1 megleI8=z-f1-ia jel=z=ett szo-t D.A SIGN=VdN=pres.part-N
PERFIFORE=VdN-pres-s33D.A SIGN=VdN=past.partWORD-acc '(the)
modifier precedesthe modified word'. A direct object which is inflected for
first or secondpersonsingular,however,can also go into the nominative,e.g.
leveleld-f1 meglkap-ta-mLETTER-s2-N PERFIGETS-past-s13'I got your
letter'. Indirect objects are put in the dative (-NAK), e.g. malac+sul=te-t
ad-ott-f1 a fiu-k-nak PIGLET+ROASTS=past.part-ACC GIVES-past-s3D.A
BOY-plur-dat '(s)hegaveroastsucklingpig to the boys'.The dative hereacts
like any otherlative in that it is interchangeable
with the accusativein parallel
452 HUNGARIAN

instrumental constructions,e.g. megIkfnal-t-a a fiu-ka-t malac+siil=t-tel

PERFIOFFERS-past-s33 D.A BOY-plur-acc PIGLET+ROASTS=past.part-
ins '(s)heofferedthe boys roastsucklingpig'.
The dative is also usedto mark higher participantsin constructionswith
auxiliaries, e.g. Kalman-nak Buda-ra kell-f)-f} men-ni(-e) KALMAN-dat
BUDA-subl IS.NECESSARY-pres-s3GOES-inf(-s3) 'Kalman has to go to
Buda', and in sentenceswith non-personalverb forms suchas kover-neklen-
ni vidam dolog FAT-dat IS-infMERRY THING 'beingfat is fun'.
The neutralorderfor major constituentsis TOPIC(S)- FOCUS- FINITE
VERB, e.g.Andrasgruzul tud ANDRAsGEORGIAN-advKNOWS '(As for)
Andrew, it's Georgianthat he knows'. If the finite verb is the focus, it comes
first, e.g. Tud Andrasgruzul or Tud grUzul Andras.The rules applying to the
order of (aspectualv. spatial) coverbsand of auxiliary verbs are of special
interestandcomplexity; seeKalmanet al. (1990).

Lexicon
Derivation of four basic types is widespread,viz. (1a) denominalnominals
(e.g. sar 'mud' : sar=os 'muddy',sarga 'yellow' : sarga=s 'yellowish'), (lb)
deverbalnominals (e.g. ad- 'gives' : ad=as 'act of giving' : ad=at 'datum'
: ad=ag 'portion' : ad=o 'tax'), (2a) denominalverbs(e.g. sarg=ul- 'becomes
yellow', sargas=od-ik 'becomesyellowish', ad=at=ol- 'provides/backsup
with data') and (2b) deverbalverbs. Spacepermits us only a brief glanceat
this last type.
A large and open set of affective, quasi-descriptiveverbs are formed with
the suffix =(3)G-, usually from themes, e.g. kop=og- 'knocks', tip=eg-
waddles, toddles', doc=og- 'trundles along'; most of these verbs refer to
repeatedactivities or to activities which consist of numerousrecurrences.
Many have semelfactiveanaloguesformed with =$AN- which refer to single
events($ is an operatorwhich occursonly in derivationand which lengthens
a preceding consonant, wherepossible), e.g. kop=pan- 'knocks (once)',
doc=cen- 'jolts, jerks (once)'. Productive,also,is the detransitivizing=OOD-,
e.g. emel=od-ik 'rises (of its own accord)' alongside emel=ked-ik 'rises,
climbs', both from emel- 'raises'.
Many verbs which are synchronicallymonomorphemiccontainthe relicts
of older derivationalsuffixes, e.g. nez- 'beholds'from *niiki=ta-, a durative/
continuativederivateof *niiki- 'sees',known from e.g. Finnish niike- 'sees';
keres-'seeks',built with a detransitivizing/deperfectivizing suffix *=si - *=ci ,
cf. ker- 'asksfor' and Komi kor- 'asksfor', kor=Si- 'seeks'.Reflexesof other
Finno-Ugric suffixes with deperfectivizing,frequentative,or multiobjectival
force are the =sz- found in met=sz- [macc] 'slices thinly', cf. met=el=t
'noodles' and the present-tense -S- suffix (seep. 447), as well as the =d- in
marad- 'remains'« *=ntV-).
Loanwordscome from four major sourcesand may be grouped,in rough
 HUNGARIAN 453

chronologicalorder, into Iranian, Turkic, Slavonic, and WesternEuropean,

although there is some overlap, e.g. early Turkic loans into pre-Hungarian,
contemporaneouswith the earliest Iranian layers, or the thin layer of
sixteenth-centuryTurkish loans (see below), which postdate the early
Slavonic items. Loans from Romanian and Romani are relatively few in
number,but rich in semanticand stylistic content.
Iranian loans into Hungariandate back to the time immediatelyafter the
breakup of Ugric unity or contact. Early examples are #TIZA tlZ 'ten',
#TAHANA tehen 'cow', and ~AIA'milk', tej 'milk', in all of which pre-Hungarian
hassubstitutedits sole, voiceless,initial dentalt- for the voiced soundsof the
donorlanguage;by the time Hungarianborrowedthe Slavoniccognateof the
last of these, viz. dajka 'wet nurse', it had acquiredd- in initial position.
Contact with Iranian idioms must have lasted well over a thousandyears,
culminating with loans such as mid 'reed', from a late Middle Iranian
languageof the southern Russian steppe.
The threelayersof loansfrom Turkic languagesare surveyedin R6na-Tas
(1988: 751-60).Theseare(1) so-calledpre-conquestloans,i.e. borrowedinto
Hungarianbeforeits speakersarrived in Europe,(2) medievalloans, and (3)
occupationloans.The first layer is particularlylarge andvaried, and includes
words which must have been borrowed from Turkic idioms of a Chuvash
type, e.g. borj';' 'calf', cf. ChuvashpSru v. Turkish buzagl, del 'midday;
south',cf. Chuvashtel v. CommonTurkic 'tUS'. The secondlayer is madeup
primarily of words borrowed from the languagesof the Pechenegsand the
Cumanians,Turkic-speakinggroupswhich settledin Hungary in the twelfth
and thirteenth centuries; examples are koboz 'lute-like instrument', and
komondor 'breedof dog'. The most recentlayer enteredHungarianwith the
Turkish occupation(roughly 1526-1698).Thesewords, of which only about
twenty survive, are usually also well representedin the languagesof the
Balkans, through which the Turks had passedearlier on their way into the
heart of Europe. Examples are zseb 'pocket', with pendantsin Serbian,
Albanian, Bulgarian,and Greek,and tepsi 'baking dish', with pendantsin all
thoselanguagesplus Romanian.
A conspicuousnumber of the items borrowed from or via Slavonic are
religious or domestic/agrarian in content.For example,from Latin comeboth
pogany'pagan'(via SouthSlavonic)andits doubletpohdnka'buckwheat'(via
Czech); note also paradicsom 'paradise';[from 1856] 'tomato', borrowed
direct from Latin, with the Latin neuterending (-um) intact. Other Slavonic
doublets: vacsora 'evening meal', vecsernye'vespers';colOp 'peg', oszlop
'column';genny'purulentmatter',ganej 'manure'.Not only doublets,but also
interdialectalsynonymshaveenteredHungarianfrom Slavonic,e.g.bab 'bean'
and(dialect)paszuly'bean'(via Slovenefrom Greek,which hadit from Italian).
Many items must,to judgefrom their meaningandform, be Slavonic,but their
antecedents areasyetunknown,e.g.poggyasz'luggage', drusza'namesake' .
Loans from German(ic) date from the first arrival of the Hungariansin
454 HUNGARIAN

Europe, perhapseven earlier (Rot 1988: 697). Among the older loans are
pelda 'example'(Modem GermanBUd), cerna 'thread' (Zwirn), and bognar
'cartwright' (Wagner).
Until recently,westernEuropeanloansentered primarilyby way of German
(e.g. kozmosz'cosmos',tik()r 'liqueur', meteor'meteor',nabob 'nabob',and
puder 'powder'),althoughtherewas also somedirect borrowing from Italian
and from French (there were Walloon merchantsfrom at least the twelfth
centuryin the largerHungariancities [Barczi 1958: 111]). Examplesof direct
Italian borrowings,probablyadoptedduringtheAngevinperiod(1308-82)are
pajzs 'shield' (Italian pavese'of Pavia[famousfor its shields])',palya 'prize,
contest, racetrack; career' (patio). Examples from French: kilincs 'latch'
(clenche),targy 'object(earlier:type of shield;target),.

Hungarian Text
From a letter written 1910by Artur Elek to Miksa Feny{) (Vezer 1975: 336).

A: orthography;B: phonologicaltranscription; C: morphophonemiccode,

segmented;D: morpheme-by-morpheme gloss;E: loosetranslationof D;
F: freer translation.

Al Ezzel a levelemmel egyidejUleg

Bl az-zal a lavalammal edidjidajti(ti)lag
Cl #AZ-VAL AZ ~AVALA-MA-VAL EDJ+#JDAIA=IUU=LAG
~AVALA-MA-VAL
Dl THIS-ins D.A LETTER-sl-ins ONE+TIME=adjsx=adv
El with this the with my letter simultaneously

postara adorn a Fogazzaro-forditast

postaara adorn a fogaddzaroofordiitaast
POSHTAA-RAA AD-0-3M AZ F+I\FOR=D=IIT=AASH-T
POST-subl GIVES-pres-s13D.A F.+TURN=vbsx=vbsx=NdV-acc
on to post I give it the Fogazzaro-translation

A2Nem tudom hogy eleg lesz-e

B2nam tudom hodj aleeg lasse
C2D2NOT
~AM TUD-0-3M HODJ #ALAGA LEV-S-0--E
D2NOT KNOWS-pres-sI3 CJ ENOUGH BECOMES-pres-s3-I.Q
E2not I know it that enough whetherit will be

nem is hiszem A3 Legjobb lenne

nam is hisem B3 lagjobb lenna
ISH HIV-S-3M
~AMbest C3 ~AG+IO=BBA best LEV-ANAA-0
NOT PTL BELIEVES-pres-s13D3 SUP+GOOD=cfvBECOMES-cond-s3
not duly I believeit E3 best it would be(come)
 HUNGARIAN 455

kiszedetni masik megtudjam esetleg

kisadatni masik mektudidjam asatlag
KIISED=AT=N! masik MEG ITUD-IA-M #ASH=AT=LAG
OUTISETS=caus=inf PERFIKNOWS-sj-s13 FALLS=NdV=adv
to have(it) set (so) that I shouldfind it out if suchis the case

mekkora teIjedelmu masik novellat

makkora taIjadalmii(ii) maasik novallaat
~AK+KORAA #"TARI=AD=ALMA=IUU
~AK+KORAA MAASH=IK NOVELLAA-T
HOW.BIG "EXTEND=vsx=NdV OTHER= NOVELLA-ace
=adjsx INDIV
how big extensive other novella

fordftsak hozzaja.
fordiiccilk hozzaaja
"FOR=D=IIT-IA-3K HOZZAA-IAA
"TURN=vbsx=vbsx-sj-sl TO-s3
I shouldtranslate to (go with) it

Fl I'm posting the Fogazzarotranslationat the sametime as this letter (of

mine). F2 I don't know whetherit'll be enough;I doubt it (in fact). F3 The
bestthing would be to haveit typesetso I can seewhetherI haveto translate
anothernovellato go with it, andif so, how long it'll haveto be.

Referencesand Further Reading

Abondolo, D. (1988) Hungarian Inflectional Morphology, Bibliotheca Uralica 9,
Budapest:Akademiaikiad6.
Barczi, G. CZ1958)Magyar hangtOrtenet,Budapest:Tankonyvkiad6.
Beckwith, C.I. (1992) 'Classifiers in Hungarian', in Kenesei et aI., vol. IV,
pp. 197-204.
Berrar,J. (1957) Magyar tortenetimondattan,Budapest:Tankonyvkiad6.
G6sy, M. (1989) BeszMeszleles,Linguistica Series A, Studia et dissertationes2.
Budapest:A magyartudomanyosakademianyelvtudomanyiintezete.
Hetzron, R. (1982) 'Non-applicability as a test for categorydefinitions', in F. Kiefer
(ed.), Hungarian Linguistics,Linguistic and Literary Studiesin EasternEurope4,
Amsterdam-Philadelphia: Benjamins,pp. 131-83.
Honti, L. (1984) Chrestomathiaostiacica,Budapest:Tankonyvkiad6.
Janhunen,J. (1982) 'On the structureofProto-Uralic',FUF 44: 23-42.
Kalman, B., (1965) 'A magyar massalhangz6-rendszer kialakulasa',Magyar Nyelv
61,285-398.
Kalman, G.c. et al. (1990) 'A magyar segedigekrendszere',Altalanos Nyelveszeti
Tanulmanyok17: 49-103.
Kenesei,I. et. al (eds) (l985-92)Approachesto Hungarian [=ATH], vol. 11985,vol.
II 1987,vol. IV 1992,Szeged:Atti1a J6zsefUniversity.
Keresztes,Laszl6 (1993) 'Uj maganhangz6-fonemak a magyarban?',in Hajdu Peter
70 eves[Festschriftfor P. Hajdu], Linguistica SeriesA, Studiaet dissertationes15,
ed. Sz. M. Bakr6-Nagyand E. Szij, Budapest:Linguistic Instituteof the Hungarian
Academyof Sciences.
456 HUNGARIAN

Kiefer, F. (ed.) (1992) Strukturdlis magyarnyelvtan,vol. II: Fonologia [= SMN, 2],

Budapest:Akademiaikiado.
Korhonen, M. (1981) lohdatus lapin kielen historiaan, Helsinki: Suomalaisen
Kirjallisuuden Seura.
Kornai, A. (1985) 'The internal structureof noun phases',in Keneseiet al. vol. I,
pp.79-92.
- - - (1994) On Hungarian Morphology, Linguistica Series A, Studia et
dissertationes,Budapest: Linguistics Institute of the Hungarian Academy of
Sciences.
Kulonen,Ulla-Maija (1993)lohdatusunkarin kielen historiaan,Suomi 170, Helsinki:
SuomalaisenkiIjallisuuden seura.
Lako, G. (1965) A magyar hangdllomdnyfinnugor ellJzmenyei,Nyelvtudomanyi
ertekezesek 47, Budapest:Akademiaikiado.
Lotz, J. (1939) Das ungarische Sprachsystem,(Old) Publication Series of the
HungarianInstitute of Stockholm3, reissued1988 as EurasianLanguageArchives
1, ed. G. Decsy,Bloomington: Eurolingua.
Lotz, J. (1967) 'Egy nyelvtani modell (Ket fejezet a magyarnyelvtanbol)',Magyar
Nyelv 63: 394--408.
Melcsuk, I. (1968) 'A magyar fBnevragozasegy ujabb modellje', Nyelvtudomdnyi
Ertekezesek 58: 502-5.
Mikola, T. (1975) Die alten Postpositionendes Nenzischen(luraksamojedischen),
Budapest:Akademiaikiado.
Niidasdy,A. (1985) 'SegmentalPhonologyand Morphophonology',in Kenesciet aI.,
vol. I, pp. 225--46.
Papp, F. (1975) A magyarfonev paradigmatikusrendszere,Budapest:Akademiai
kiado.
Papp, I. (1968) Unkarin kielen historia, Tietolipas 54, Helsinki: Suomalaisen
kirjallisuudenseura.
Rona-Tas,A. (1988) 'Turkic influenceon the Uralic languages'in D. Sinor (ed.), The
Uralic Languages:Description, History and Foreign Influences,Handbuchder
Orientalistik 8, Leiden: Brill, pp. 742-80.
Rot, S. (1988) 'Germanicinfluenceson the Uralic languages',in D. Sinor (ed.), The
Uralic Languages:Description, History and Foreign Influences,Handbuchder
Orientalistik 8, Leiden: Brill, pp. 682-705.
Sammallahti,P. (1988) 'Historical phonology of the Uralic languages,with special
reference to Samoyed, Ugric, and Permic', in D. Sinor (ed.), The Uralic
Languages:Description, History and Foreign Influences,Handbuchder Orien-
talistik 8, Leiden: Brill, pp. 478-554.
Sherwood,P. (1996)A ConciseIntroduction to Hungarian, SSEESOccasionalPapers
no. 34, London: University of London.
Sima, F. (1971) Magyar nyelvtortenet,vol. I, Bratislava: Slovenskepedagogicke
nakladatel'stvo.
Szabolcsi,A. (1987) 'Functionalcategoriesin the noun phrase',in Keneseiet al. (eds)
vol. II, pp. 167-89.
Szilagyi, S. N. (1980) Magyar nyelvtan,Bucharest:Edituradidacticii §i pedagogicii.
Tiilos, E. (1975) 'A magyarhang- es alaktan nehiiny kerdesehez',Uralica (Journal
of the Uralic Societyoflapan) 3: 39-53.
Vezer, E. (1975) Feljegyzesekes leveleka Nyugatrol, Uj magyarmuzeum:irodalmi
dokumentumokgy{fjtemenye 10, Budapest:Akademiaikiado.
Viitso, T.-R. (1996) 'On classifying the Finno-Ugric languages',C8IFU vol. IV:
261-6.
15 Samoyedic
luha lanhunen

Geographically occupying the easternperiphery of the language family,

Samoyedicis conventionallyclassifiedas one of the two principle branches
of Uralic. Mter the breakupof proto-Uralic linguistic unity, the speakersof
pre-proto-Samoyedic graduallycameto be concentratedin the region located
betweenthe middle coursesof the Ob' and Yenisei Rivers in southwest
Siberia. Unlike the Finno-Ugric branch, which shows a high degree of
internal diversity, Samoyedic has no surviving early sub-branches,the
historically known Samoyedic languages forming a coherent group of
relatively closely related idioms. In terms of absolute dating, proto-
Samoyedicseemsto have dissolved as recently as the last centuriesBeE.
Consecutivewavesof ethnic andlinguistic expansionthen spreadthe various
forms of Samoyedicboth northwards,along the Ob' and Yenisei basins,and
southwards,in the directionof theAltai andSayanMountains.It is reasonable
to assumethat in the course of this expansion severalunidentified non-
Samoyediclanguagesspokenearlier in the sameregions became extinct due
to linguistic assimilation. Most of the modem Samoyedic-speaking ethnic
groups are thereforecomposedof two major ethnohistoricalcomponents:a
local componentcorrespondingto the earlier indigenouspopulation,and an
immigrantcomponentresponsiblefor the Samoyediclanguage.
The Samoyedicbranch may rather uncontroversiallybe divided into six
main entities or independentmajor languages:Nenets (Yurak), along the
Arctic coastfrom the White Searegion to westernTaimyr; Enets (Yenisei-
Samoyed),in the lower Yenisei region; Nganasan(Tavgy), on the Taimyr
Peninsulafrom the lower Yenisei in the westto the KhatangaBay in the east;
Selkup (Ostyak-Samoyed), in the region betweenthe Ob' and Yenisei from
the Taz and Turukhan in the north to the Chaya and Chulym in the south;
Kamas (Kamassian),in the easternpart of the upper Yenisei region; and
Mator (Motor), in the easternSayanMountainsfrom the upperYenisei in the
west to the Baikal region in the east. Considerabledialectal differencesare
presentwithin Nenets(with Tundra Nenetsand ForestNenets),Enets (with
Tundraor Madu Enetsand Forestor Bai Enets),and Selkup(with a complex
dialect continuum). In addition, a separate,transitional idiom, technically
termed Yurats, once existed betweenNenets and Enets. Both Kamas and
Mator are historically known by a variety of alternative names,including

457
458 SAMOYEDIC

Taigi andKaragasfor Mator, and Koibal for Kamas.Thesenamesare mainly

of geographicaland chronologicalinterest; linguistically Kamas and Mator
are two well-delimited entities with a minimum of internal dialectal
differentiation. Some sources recognize a transitional idiom, technically
termed Abakan, between Kamas and Mator, but this is better seen as a
philological conglomerationof early KamasandMator lexical material.
In termsof their geographicalenvironment,the speakersof the Samoyedic
languagesform three distinct ecological groups: the Tundra Samoyeds,the
Taiga Samoyeds,and the Mountain Samoyeds.The Taiga Samoyedsare
essentiallyidentical with the speakersof Selkup, who up to the presentday
occupya territory moreor lesscoterminouswith the original Samoyedhabitat
in the westernSiberianforestzone.The TundraSamoyedscomprisethe major
part of the speakersof Nenets,Enets,andNganasan,who togetheroccupythe
whole of the Arctic tundrazoneof EuropeanRussiaand westernSiberia.The
Mountain Samoyedscomprisethe speakersof Kamas and Mator, who used
to inhabit the wooded highlands of southern Siberia. The ethnolinguistic
developmentduring the last few centurieshas beenleast favourablefor the
Mountain Samoyeds,whoselanguagesare now extinct, Mator sincethe first
half of the nineteenthcentury and Kamas sincethe deathof the last speaker
in 1989.Extinction also threatensthe TaigaSamoyeds,althoughSelkup(with
under 2,000 speakers)still survives,albeit dialectally highly fragmented,as
a living language.The TundraSamoyedshavebeengenerallylessexposedto
external threats, and Nenets (with some 25,000 speakers)has emerged
(togetherwith NorthernSaarniand Greenlandic)as one ofthe threeprincipal
Arctic languagesof the world. Unfortunately,the expansionof Nenetsin the
lower Yeniseiregion hasbeenat the expenseof Enets,which (with fewer than
100 speakersremaining) is now facing imminent extinction. The Nenets
expansionwas also one of the factors that causedthe disappearance of Yurats
in the nineteenthcentury. Nganasanwas never seriously affected by these
developments,but the small size of the Nganasancommunity (with hardly
more than 500 speakers)is becoming an obstacleto the survival of this
languagein the future.

Taxonomic Relationships
The conventional view holds that the ecological trichotomy of the
Samoyedic-speaking populationscorrelateswith a geneticdivision (cf. Figure
15.1), in that the TundraSamoyedsare linguistically supposedto correspond
to a separateNorthern Samoyedicsub-branch,as opposedto a Southern
Samoyedic sub-branch.Within the latter, the languagesof the Mountain
Samoyedshave been assumedto form a shallow-level sub-branchtermed
SayanSamoyedic.This taxonomyhas beenquestionedmore and more often
in recentyears,as it has becomeevidentthat much of the internal coherence
within eachproposedsub-branchmustbe dueto secondarymutualinfluences.
 SAMOYEDlC 459

Figure15.1 The conventionaltaxonomyof the Samoyediclanguages

Proto-Samoyedic

NorthernSamoyedic SouthernSamoyedic

SelkupSamoyedic

Nganasan Enets Nenets Selkup Kamas Mator

(TundraSamoyeds) (Taiga Samoyeds) (Mountain Samoyeds)

Figure15.2 An alternativetaxonomyof the Samoyediclanguages

Proto-Samoyedic

Nganasan Enets Nenets Selkup Kamas Mator

No conclusivenew taxonomyhas yet emerged,but there are indicationsthat

the first entity to havebrokenoff from proto-Samoyedicunity may have been
Nganasan,possiblyfollowed by Mator (cf. Figure 15.2).This situationwould
correlatewell with the geographicalfact that Nganasanand Mator represent
the northeasternand southeasternextremitiesin the geographical continuum
of the Samoyediclanguages.In this continuum, each languageshares a
numberof diagnosticfeatureswith its immediateneighbour:Nganasanwith
Enets, Enets with Nenets, Nenets with Selkup, Selkup with Kamas, and
Kamas with Mator. However, there are also features which, at least
superficially,link the extremitieswith eachother: Nganasanwith Mator, and
NorthernSamoyedicwith SayanSamoyedic.

Documentation
Of the six Samoyediclanguages,only Nenets may be regardedas well
documented for both descriptive and comparative purposes, though a
considerableamountof additional documentationwill still be requiredfor a
detailedunderstandingof Nenetsdialectology.Somewhatless infonnationis
460 SAMOYEDIC

available on Selkup, especially if only printed sources are taken into

consideration.Additional material on both (Tundra) Nenetsand (Northern)
Selkupis providedby the modemliterary languages, which haveexisted,with
many vicissitudes,since the early 1930s. By contrast,Nganasanand Enets
have until recently remainedpoorly documentedat all levels of linguistic
substance,the results of the latest field researchbeing still unpublished.
Nevertheless,the grammatical and lexical information now available on
Nganasanand Enets is quite sufficient to allow thesetwo languagesto be
seriously analysedin a comparativecontext. The same is true of Kamas,
although the material collected from the last few generationsof Kamas
speakersreflects a somewhatrudimentary commandof the language.For
Mator, only lexical information was noted down in time, leaving the
morphology and syntax of this languagevirtually unknown. In terms of
absolutetime, the availablenotes on Mator cover the period extendingfrom
the early eighteenthto the early nineteenthcentury,while Kamasmaterialwas
collectedfrom the early eighteenthcenturyuntil the 1970s.The earliestnotes
on the otherSamoyediclanguagesderivefrom the seventeenthcentury.

Diachronic Evaluation
The close cognatenessof the Samoyediclanguagesallows us to reconstruct
proto-Samoyedicas a ratheruniform idiom with a relatively elaborateset of
unambiguouslyestablishedphonologicaland morphologicalcharacteristics.
The conventional view has been that a feature which is presentin both
NorthernSamoyedicand SouthernSamoyedicis to be recognizedas deriving
from proto-Samoyedic,but this criterion will have to be reviewed if a
different approach is adopted to the internal taxonomy of Samoyedic.
Although there is no single Samoyediclanguagewhich would provide a
simplekey to proto-Samoyedic,a greatdeal of diachronicinformation canbe
drawn from Nenets by the method of internal reconstruction. In the
comparativeframework, an importantplaceis occupiedby Nganasan,which
in somefundamentalways deviatesfrom the patternsexhibitedby the rest of
the Samoyediclanguages.At the oppositeextreme,Mator would also be of
great potential importance,were it not so poorly documented.Each Samo-
yedic languageis characterizedby a basicallyidiosyncraticset of archaisms
and innovations,but there are indications that Nganasanand Mator are in
certain ways particularly conservative,while the more centrally located
languagesare generally more innovatory. Nevertheless,even Selkup occa-
sionally preserves diachronically relevant information which has been
obscuredin the restof the Samoyediclanguages.

Typological Variation
In the light of the available reconstructions,and possibly becauseof
systematic distortions connected with the comparative method, proto-
 SAMOYEDIC 461

Samoyedicappearsto havebeena languagewith an abundantsystemof word

forms createdby quasi-mechanical agglutination.A few morphophonological
alternations can also be reconstructed,but morphophonology plays a
considerablymore importantrole in the modemSamoyediclanguages.There
is a clear tendency towards increasing morphophonologicalcomplexity
towards the north and northeast,with Nganasanexhibiting an exceptionally
large number of cumulative morphophonologicalphenomena,including
consonant gradation in both stems and suffixes, stem-final vowel and
consonantalternations,andvowel harmonyof a highly idiosyncratictype. As
far as the systemof morphophonologicalcategoriesis concerned,Northern
Samoyedic seems to have preservedthe original proto-Samoyedicstate
somewhat better than Southern Samoyedic has done, but variation is
considerablebetween the individual languages.The same is true of the
patterns of segmental structure, which differ greatly not only between
adjacent languages,but even between the dialects of a single language.
Areally and typologically relevant peculiaritiesof the segmentalparadigm
appearto be particularlynumerousin Nenetsand Selkup.

Proto-Samoyedicand Proto-Uralic
For a long time in the history of Uralic comparativestudies,it was customary
to approach proto-Uralic from the Finno-Ugric side alone. Insofar as
Samoyedicis to be recognizedas a branch parallel with Finno-Ugric, it is
obvious that a definitive understandingof proto-Uralic will only be possible
after binary comparisonhasbeenmadebetweenproto-Samoyedicand proto-
Finno-Ugric. However, as there still seemto be considerabletaxonomicand
reconstructionalproblems to be solved for the easternbranchesof Finno-
Ugric, a simplified but very useful approximationof proto-Uralic can be
obtainedin the meantimeby comparingproto-Samoyedicwith proto-Baltic-
Fennic, with additional information drawn, as required, from Saami and
Mordva. On the historical map of the Uralic languages,proto-Samoyedicand
proto-Baltic-Fennicare two roughly contemporaneous idioms, both of which
are fairly uncontroversiallyreconstructable by a comparativeanalysisof their
modem descendants.Also, as these two branchesrepresentopposite geo-
graphicalextremesof the Uralic languagefamily, it may be assumedthat any
diachronic feature shared by proto-Samoyedicand proto-Baltic-Fennicis
likely to derive from proto-Uralic. The number of such sharedfeaturesis
fairly high, especiallyas far as morphologyis concerned,a situation which
suggeststhat Samoyedicand Baltic-Fennic are in many important respects
equally and similarly conservative.The easternbranchesof Finno-Ugric
obviously representthe more innovatory centreof the languagefamily, and
they may well prove to be of less relevanceto the reconstructionof proto-
Uralic thanthe two peripheries.
462 SAMOYEDIC

Further Reading
Classicintroductoryworks on Samoyedicwith good bibliographiesareHajdu
(1963, 1968, and 1988). A generalevaluationof the ethnohistoricalposition
of the Samoyedic-speaking peoplesis Xelimskij (1983). Problemsof genetic
taxonomyaredealtwith in Xelimskij (1982a: 27-47)(Samoyedicin general),
Xelimskij (1976) (the status of Yurats), Janhunen(1991) (the position of
Nganasan).Dialectologicaltreatmentsof the individual Samoyedic languages
include Salminen (1990) (Nenets), Janurik (1985) (Nenets and Selkup),
Xelimskij (1985a)(Enets),Janurik (1978) (Selkup), Ktinnap (1985) (Selkup
andKamas).

Phonology
ConsonantParadigm
There is a widespreadconsensustoday that proto-Samoyedicpossessed
thirteen consonant phonemes, which represent four or five places of
articulation and five to sevenmodesof articulation (cf. Figure 15.3). Four
placesof articulation(labial: dental: palatal:velar) are clearly presentin the
nasalseries *m *n *ii *ng, while the five obstruents*p *t *c *s *k may be
defined either in terms of five separateplacesof articulation (labial : dental
: cacuminal:palatal: velar) or, alternatively,by assumingdistinctive modes
of articulation for *s (continuant and/or sibilant) and *c (retroflex and/or
affricated). The interpretation of *s as belonging to the palatal series is
probablycorrectfor pre-proto-Samoyedic, but by the proto-Samoyedicperiod
this segmentseemsto have becomemore closely connectedwith the dental
series.The segment*c is preservedonly in part of the Selkupdialects,where
its quality variesbetweena dental affricate and a retroflex stop, while in the
rest of the Samoyedicidioms it hasinvariably mergedwith the dentalstop *t.
The two liquids *1 *r may be characterizedin terms of a complexdistinction
involving both place and mode of articulation (lateral : vibrant), but
morphophonologicalevidence suggests that their primary phonological

Figure 15.3 The proto-Samoyedicconsonantparadigm

*p *r *r *r *k
*r
*r
*m *n *f1 *ng
*1
*r
*w *y
 SAMOYEDIC 463

oppositionmay haveconcernedthe featureof continuance(stop: continuant).

The two glides *w *y correspondto two separateplacesof articulation(labial:
palatal).

Vowel Paradigm
Proto-Samoyedicseemsto have had a systemof one reducedand ten full
vowel phonemes(seeparadigmbelow).

Full *u *1 *ti *.1

*0 *e *6 *e
*a *a
Reduced * !IS

The reduced vowel *!IS was probably the least marked member of the
paradigm,distinguishedfrom the full vowels by being quantitativelyshorter
and prosodically weaker.The full vowels representthreedegreesof opening
(high v. middle v. low) and four different combinationsof lip-rounding with
frontnesslbackness(rounded back v. unroundedback v. rounded front v.
unroundedfront). In pre-proto-Samoyedicthesedistinctions were probably
fully employedonly in the high series*i *u *1 *ti, while the mid seriesseems
originally to have comprisedonly the three vowels *e *0 *e, the fourth mid
vowel *6 appearingas a marginalnew phonemeat the proto-Samoyediclevel.
The low series consistedof the two vowels *a *a; these may have been
distinguishedphoneticallyeitherby lip-rounding (roundedbackv. unrounded
back) or by frontnesslbackness (unroundedback vs. unroundedfront). The
vowel *a, in particular,may have beenqualitatively unstable:its reflexesin
the modemlanguagesare variously eitherfront vowels (in Nenets,Enetsand
Mator) or backvowels (elsewhere).

Vowel Sequences
An important characteristicof proto-Samoyedicvowel phonotaxiswas the
fact that in initial syllablesany full vowel could be followed by the reduced
vowel, in a sequenceof the type *V!IS. There is evidencethat thesevowel
sequenceswere in proto-Samoyedicrhythmically equivalent to bisyllabic
sequencesof the type *VC!IS, but at the pre-proto-Samoyedic level they may
be derivedfrom monosyllabicunits involving a syllable-final consonant.The
consonantin question has been reconstructedas *x, and it is probably to
be classifiedas a velar glide, paradigmatically parallel to the othertwo glides
*w *y. The non-syllabic origin of the reducedvowel in vowel sequences
explains the absenceof any such sequenceswith a full vowel as the latter
component.For relateddiachronic reasonsit also seemsthat a sequenceof
two reduced vowels was not permitted in proto-Samoyedic.Among the
modem languages,consistenttraces of the vowel sequencesare preserved
only in Nganasanand Enets, the two idioms which synchronicallypermit
464 SAMOYEDIC

the linear accumulationof two or more syllabic segments.Nganasanand

Enetsalso possesssecondaryvowel sequences,sometimesanalysedas 'long
vowels' and 'diphthongs', which have arisen through the loss of an
intervocalic consonant or glide. Somewhat similar secondary vowel
sequencesare present in Nenets, where original vowel sequenceshave
generallymergedwith the paradigmof singlevowels.In the otherSamoyedic
languagesvowel sequencescan only occasionallybe traced in otherwise
inexplicablequalitativedevelopments.

Morpheme Structure
The structureof underivednominal andverbal stemsin pre-proto-Samoyedic
may be describedby the generalformula *(C)V(X)(C)(CV), where X stands
for the glides *w *y *x. The samestructuraltypes were preservedin proto-
Samoyedic,with the modificationthat by then *x was vocalizedinto *!Il. Like
Baltic-Fennic,proto-Samoyedichad many lexical items of the proto-Uralic
bisyllabic type *(C)V(C)CV, as in *m!llna 'egg' (proto-Uralic *muna), *yent!ll
'sinew' (proto-Uralic *yanti). However, structurally innovative monosyl-
lables of the type *(C)VC, as in *wit 'water' (proto-Uralic *weti), were also
conspicuouslyabundant.An importantnew type of morphemestructurewas
* (C)V, which developedregularly in items containing a proto-Uralic inter-
vocalic *x, as in *to 'lake' (proto-Uralic *tox'i). The occurrenceof vowel
sequencesin underivedstemswas primarily restrictedto the type *(C)VXC,
as in *ka!llt 'fir' (proto-Uralic *kaxs'i), but for reasonsnot yet explainedthere
also appearedthe type *(C)VX, as in *ka!ll- 'dies' (proto-Uralic *kax=l1-).
Any more complicatedstructures,as in *ka!llsa 'man' « *ka!ll=sa 'mortal'),
imply the presenceof an etymological morphemeboundary.A morpheme
boundarymay also be assumedto have beenoriginally presentin any other
casesin which a bisyllabic sequenceis followed by an extensionof the type
*-C«C)V). It seems,however,that a few original complexstemshad already
been transformedinto monomorphemicentities in proto-Samoyedic.This
must have been the casein, for instance,*Yn!ll(=)p!ll 'parent-in-law' (proto-
Uralic *Yna+YppY).

ConsonantPhonotaxis
The complete proto-Samoyedicparadigm of consonantsis attested only
intervocalically; other positionswere subjectto various phonotacticrestric-
tions. A restriction of areal relevance,deriving from proto-Uralic and still
valid for the modemSamoyediclanguages,excludedthe liquid r from word-
initial position. In pre-proto-Samoyedic thereseemsto havebeena tendency
to extendthis restriction to the other liquid 1, as well, for most of the Uralic
items with *1- show *y- in proto-Samoyedic,as in *yom- 'to snow' (proto-
Uralic *lomY-). There are, however, a few instancesin which *1- has been
preservedintact, as in *115 'bone' (proto-Uralic *IYxY), and in the post-
proto-Uralic part of the proto-Samoyediclexicon *1- is fully permitted.
 SAMOYEDIC 465

Interestingly,the word-initial situationis echoedword-internally,in that only

*1, and not *r, was pennittedto occur as the latter componentof a consonant
clusterin proto-Samoyedic.Of the otherconsonants,only the velar nasal*ng
was excludedfrom initial position, a phonotacticgap which has beenfilled
in Northern Samoyedicby the introduction of a prothetic (*)ng- before all
original initial vowels. In fact, it would be tempting to assumethat *ng was
on the way to becomingcomplementarywith *ii in proto-Samoyedic,for the
possible word-internal and word-final occurrencesof *ii are difficult to
distinguish from the sequence*yn, while *ng is well attestedboth word-
internally andword-finally. On the basisof evidencefrom Selkup(andFinno-
Ugric) it appears,nevertheless,reasonableto reconstructan intervocalic*ii in
a few items, as in'tame'
*lii~ 'tame' (proto-Uralic *liil). A stem-final *ii is perhaps
to be reconstructedin a few items which in Nenetsexhibit an alternationof
*n with *y, as in Nenetstoh: toy- 'blanket'< pSAM *toii.
Vowel Phonotaxis
The maximal proto-Samoyedicparadigmof vowels hasthus far beenverified
as valid only for the initial syllable; how many distinct vowel phonemes
occurredin non-initial syllablesis still a matterof controversy.In the lexical
items inheritedfrom proto-Uralic only the three vowels *a *a *~*a (and zero)
are reliably attestedin non-initial syllables,as in*y~ka 'river' (proto-Uralic
*yuka), *pita 'nest' (proto-Uralic *pesa), *kuns~ 'urine' (proto-Uralic
*kun~'i). In the post-proto-Uralic part of the proto-Samoyediclexicon,
however,there are undeniableexamplesof other vowels, notably *u and *i,
'knife''knife' > Enetsbru v. NenetsxrjJr' (formerly regarded,probably
as in *k~ru
incorrectly, as a Uralic item), *wapi 'luck' > Nganasanbaxi : babi- 'wild
reindeer'v. NenetsyabO 'luck'. On the otherhand,evidencefrom the internal
reconstructionof Nenetsand Enets suggeststhat many of the instancesthat
would superficially suggestthe reconstructionof vowels other than *a *a *~*a
in non-initial syllables actually require the reconstruction of sequences
containing*a *a 'tame'
*~ in variouscombinationswith the glides *w *y. Contracted
vowels of varioustypes and origins in non-initial syllablesare also presentin
the otherSamoyediclanguages,althoughthe diachronicdetailsstill remainto
be worked out. The principal problemto be solvedin the future concernsnot
so much the sequentialorigin of suchvowels as the relativechronologyof the
contractivedevelopments.
Vowel Harmony
Although vowel harmony as a productive process is attested only in
Nganasan,relatedphenomenaare presentin NenetsandEnets.The combined
evidence of these three languagessuggeststhat proto-Samoyedichad a
residual harmony of the palatovelartype, which affectedthe use of the low
vowels *a *a in suffixes.The structureof stemmorphemesin themselveswas
not subject to any harmonic restrictions, for various innovations had
introducedviolations of vowel harmonyevenin items of proto-Uralic origin,
466 SAMOYEDIC

as in proto-Samoyedic*kala 'fish' « proto-Uralic *kala). Suffixal harmony

in proto-Samoyedicwas conditionedby the quality of the vowel in the initial
syllable of the stem morpheme.Thus stemscontaining any of the non-low
front vowels *i *ti *e (and possibly *6) in their initial syllable required a
suffixal *a, while stemscontainingany of the backvowels *u *1 *0 *e or the
low vowels *a *a in the initial syllable were combinedwith *a, as is still
evidentfrom modemexamplessuch as Nenetsyilt'nya 'in the water' < *wit-
kl/i-na v. myalt'na 'in the tent' < *mat-kl/i-na. It is of diachronicrelevanceto
note that *a in the initial syllable (a post-proto-Uralicinnovation) behaved
like a back vowel, while *a in non-initial syllables (proto-Uralic *a) was a
front vowel. In otherwords,the two vowels were not diachronicallyidentical,
althoughthey occupiedan identicalplacein the paradigm.The reducedvowel
*I/i also seemsto havebeenambivalent,in that it was able to behaveboth as
a back vowel and as a front vowel dependingon the stem, as in *tl/i=ta- 'to
bring' v. *tl/i-na 'there'. Such examplessuggestthat pre-proto-Samoyedic
may havehadtwo separatereducedvowel phonemes.

Consonant Correspondences
The phonologicalcorrespondences amongthe Samoyediclanguagesare for
the most part transparentto the professionaleye, although a naIve native
speakerof any single Samoyediclanguagewould often find it impossibleto
recognizecognateitems evenin the immediatelyneighbouringlanguage.The
innovations responsible for the differences in the development of the
consonantsinclude, amongothers:

• the prothesis of (*)ng- in Northern Samoyedic,as in *opa 'glove' >

Nganasannguxuv. Kamasuba;
• the spirantizationof *p into x in Nganasanand Mator, as in *payma
'boot' > Nganasanxajmu v. Nenetspy{wa;
• the developmentof *w into (*)k (> k, q) in Selkup, as in *Wl/ita 'hook'
> Selkupkot¢ v. Nenetsw¢da;
• the sporadicdevelopmentof *y into (*)k (> k, q) in Selkup, as in *yora
'deep'> Selkupqor¢ v. Nenetsyorya;
• the neutralization of syllable-final obstruents into a glottal stop in
NorthernSamoyedic,asin *kat 'face' > Enetsseqv. Selkupqaat;
• the simplification of all consonantclusters in Enets, as in *mir-Wl/i
'weapon'> Enetsmim¢ v. NenetsmyirwO;
• the assimilationof *t « *t and *c) after dentalsonorantsin Nenets,as in
*Sl/ilcl/i 'stump' > Nenetss¢Zov. NganasanseZte;
• the translocationof palatality from any original front vowel on to the
precedingconsonantin Nenets,as in *pir=ka 'high' > Nenetspyircya v.
Mator xirge.

An exampleof a less trivial set of correspondences

is offered by the proto-
 SAMOYEDIC 467

Table 15.1 Reflexesof proto-Samoyedic*k v. *s beforedifferent vowels

Nganasan Enets Nenets Selkup Kamas Mator Before:

s C; sy C; v. s C; v. s k *i *e
k v. s C; sy C; v. s C; v. s k *ti *0
k v. s s sy (*)k v. s k v. s k k v. s *li
k v. s k v. s (*)k v. (*)s (*)k v. s k v. s k v. s others

Samoyedicoppositionbetweeninitial *k and *s (cf. Table 15.1). It may be

seen that, although *k has undergone assibilation in all the Samoyedic
languagesexcept Mator, the developmentshows three different contextual
patterns:onefor Nganasan(neutralizingassibilationbefore*i *e), anotherfor
Nenetsand Enets(neutralizingassibilationbefore any front vowel, including
*a), and a third for Selkup and Kamas (non-neutralizingassibilation before
any front vowel except *a). The most curious situation is encounteredin
Mator, where *s hasdevelopedpositionally into *k (neutralizingocclusiviza-
tion beforeany front vowel except*a).

Vowel Correspondences
The vowel distinctions originally present in the initial syllable are best
preservedin Selkup,which dialectally still retainsthe proto-Samoyedicvowel
paradigmalmostintact. Apart from the effect of the translocationof palatality,
Nenetsnormally also allows a reliable reconstructionof vowels in the initial
syllable to be made,while the rest of the Samoyediclanguagesexhibit cases
of completeparadigmaticneutralization.Nganasan,for instance,has merged
*e and *a into a, *0 and *6 into u, and *1 and *ti into i (which also represents
original *i). Such neutralizationsare often connectedwith major tendencies
affecting the whole vowel system,suchas the Nganasan'vowel shift' , which
is also responsiblefor the developments*a > 0 and *u > ii. On the otherhand,
the history of individual lexical items occasionallyrevealsinterestingminor
phenomena,such as the qualitative levelling in *s'ira 'snow' > Nganasan
(eighteenthcentury) siru > (nineteenthcentury) sirii > (twentieth century)
siirii v. Nenetssira. Thereare particularlymanyproblemsstill connectedwith
vowel sequences,as well as with sequencesinvolving the palatalglide *y; in
many casesthesehave reflexes differentfrom thoseof the single vowels (cf.
Table 15.2). Proto-Samoyedicalso had syllable-final instancesof the labial
glide *w, as in *kaw 'ear', but such examplesseemto have beenso few in
number that they had no relevanceto the evolution of the modem vowel
paradigms.

FurtherReading
Some aspectsof the phonological reconstructionof proto-Samoyedicare
discussedin Janhunen(1976) (vowels of the initial syllable), Xelimskij
468 SAMOYEDIC

Table 15.2 Examples of Samoyedicvowel correspondences

Nganasan Enets Nenets Selkup Kamas Mator

*u ii u u u uu u
*uy uj u U ii: ii uj
*u0 ue ua U u: uu u
*U0y iie ua 0 u: ii u

(1978) (vowels of the non-initial syllables), Janhunen (1986: 149-68)

(syllable-final consonants),Janurik (1982) (consonant correspondences)
Terentjev (1982) (various topics). A summaryof the phonologicaldevelop-
ments from proto-Uralic down to the individual Samoyediclanguagesis
presentedin Sammallahti(1988: 484-6,494-9).

Morphology

Morphophonology
Most of the morphophonologicalphenomenathat canbe datedbackto proto-
Samoyedicare peculiarto stemsending in a consonant.When followed by a
suffix consistingof a singleconsonant,suchstemsrequirethe insertionof the
reducedvowel *-~- with as a connectivesegment,as in *nim 'name': (genitive)
*nim-~-n; it is, of course,a matter of interpretationwhetherthis connective
segmentis viewedas belongingto the stemor to the suffix. On the otherhand,
if a stem-finalconsonantis followed by a suffix beginningwith a consonant
cluster,the initial segmentof the suffix is deletedin orderto avoid a sequence
of threeconsonants,asin *op 'one': dative *op-t~(-ng) 'together'< pre-proto-
Samoyedic *o(=)p-nt~(-ng). Also, according to the rules of consonant
phonotaxis,a suffix-initial *r was replacedby *1 after a stem-finalconsonant,
as in *wen 'dog': *wen-l~ 'your dog' < pre-proto-Samoyedic*wen-r~.used All of
these phenomenaare preservedas synchronic processesin the modem
Samoyediclanguages.As far as stemsending in a vowel are concerned,the
final *a and *a of bisyllabic stemsalternatemorphophonologicallywithwith *~.
In the caseof nouns,the alteredstemseemsto havebeenusedas an accusative
plural in proto-Samoyedic,as in Nenetstyon°« *tont~),used accusativeplural of
tyonya 'fox' « *tonta). In the caseof verbs,the stemalternationwas perhaps
primarily connectedwith the transitive/intransitiveopposition, as in Nenets
xaye- « *kaya-) 'leaves(tr)' : xayo- « *kay~=y=~y-) 'remains'.At the pre-
proto-Samoyediclevel, the stem alternants with with *~ probably involved
sequencesof a final vowel plus the glide *y, but the diachronicdetailsremain
obscure.The entire phenomenonis preservedin Northern Samoyediconly,
whereit hasbeenaffectedby varioustypesof morphologicalrestructuring.
 SAMOYEDIC 469

Table 15.3 Caseendingsin proto-Samoyedic

Spatialnouns Ordinary nouns

Nominative *-0
Accusative *-m
Genitive *-n
Dative *-ng *-k¢- *-nt¢(-n§)
Locative *_na2 *-k¢-na2 *-nt¢-na
Ablative *-t(¢) *-k¢-t(¢)
Prosecutive *-m-na2 *-m¢-na2

Note: 2The digital index standsfor the variation connectedwith vowel harmony.

CaseDeclension
Proto-Samoyedicnominal declensionseemsto have had a system of one
unmarkedand six suffixally markedcases(cf. Table 15.3).With the exception
of a few lacunae of documentationfor Kamas and Mator, this system is
preservedas the core of the case systems of all the modem Samoyedic
languages.Three of the casesexpressgeneralgrammaticalrelationshipsand
have proto-Samoyedic shapes deriving directly from proto-Uralic: the
unmarked nominative (absolutive) in *-13, the genitive in *-n, and the
accusativein *-m. The otherfour casesexpresslocal relationshipsand occur
with two principal setsof endings,dependingon whetherthey are attachedto
an ordinary noun or to a spatial noun usedas an adverbor postposition.The
declension of spatial nouns is structurally more simple and incorporates
archaic endingsderiving from proto-Uralic: *-ng for the dativellative, *-na2
for the locative, and *-t(p) for the ablative.The declensionof ordinarynouns,
on the other hand, was basedon the coaffixal use of the elements*-kp- and
*-ntp( -), which function basically as dativellative endings with a varying
distribution in the modemlanguages.Nganasanis alonein using the locative
complex *-ntp-na2 as opposed to *-kp-na2 elsewhere,a situation which
suggeststhat the local casesystemwas still being formed at the time when
Nganasanbroke off from the rest of proto-Samoyedic.The ending of the
remaining local case, the prosecutive/prolative,may also have a coaffixal
origin, in that it possibly containsa combinationof the accusative*-m- with
the locative *-na2 • The fact that the prosecutivecomplexappearsas *-mp-na2
in the declensionof ordinary nouns may be explainedas due to a rhythmic
analogywith the locativecomplex*-kp-na2 •

Number
Apart from the unmarkedsingular(absolute)numberof nominal declension,
proto-Samoyedicretained the proto-Uralic markers for the dual and plural
numbers.Thesewere originally not combinablewith the normalcaseendings;
various types of plural paradigm were constructedsecondarilyout of the
availableelements(cf. Table 15.4). It is importantto note that the plural was
470 SAMOYEDIC

Table 15.4 Patterns of the plural and dual declensionsin Samoyedic

Proto-Samoyedic Nenets Nganasan

Plural nominative *-t

accusative *-y
genitive *-y-q
dative *-k0-q *-nt0-y-q
locative *-k0-q-na2 *-nt0-y-(q-)na2
ablative *-k0-q-t0 *-k0-y-(q-)to
prosecutive *-Y-Q-m0-na
Dual *-k0-fi

from the beginning marked by two different suffixes depending on the

morphologicalcontext: *-t for the nominative (absolutive) and *-y for the
accusative(oblique). This differenceis synchronicallypreservedin Northern
Samoyedic,where the formation of the accusativeplural, in particular, is
connectedwith a multitude of morphophonologicalcomplications,including
reflexes of the proto-Samoyedicstem-final alternationof *a and *a with *!Il.
The accusativeplural is also used as a basefor severaltypes of denominal
verbs, such as the captativesin *=y=n- and the possessivesin *=y=s-. The
NorthernSamoyedicgenitive plural is formed by addinga glottal stop (-q) to
the accusativeplural. The form has beenexplainedas a combinationof the
two plural markers *-y and *-t, but there is little actual evidencefor the
precise original identity of the latter component,which might just as well
have been *-s or *-k. The genitive plural, in tum, is used to form the
prosecutive plural, a circumstancewhich superficially would appear to
suggesta postpositionalorigin of the prosecutiveending. However, in view
of the secondarynatureof the whole plural paradigm,the prosecutiveplural
is likely to be a recentstructuralinnovation with no relevanceto the origin
of the prosecutiveas a category.

PersonalEndings
All the Samoyedic languageshave at least two main types of personal
endings:the predicative andthe possessivesuffixes. The predicativesuffixes
are usedto expressthe actor of a finite verb (the subjectiveconjugation)or
the subjectof a predicativelyusednoun (the nominal conjugation),while the
possessivesuffixes expressthe possessorof a noun (the possessivedeclen-
sion) or the actor of a transitive verb with a definite object (the objective
conjugation).The possessivesuffixes are also used, togetherwith a special
marker,to expressthe recipientof a noun(the predestinativedeclension).The
details vary from languageto language,but there are indications that the
highly differentiatedsystemsynchronicallyattestedin Nenetsmost closely
reflects the proto-Samoyedicstate of affairs. In this system,the possessive
suffixes appearin four different variants,dependingon which other suffix, if
 SAMOYEDIC 471

Table 15.5 The systemof personalendingsin pre-proto-Nenets

Predicative Possessive Accusative Oblique Plural/dual Reflexive

Nominative

Singular 1 *-m *-ml'l *-0-ml'l *-0-nl'l *-nl'l *-m-l'l-q

2 *-n( -tl'l) *-rl'l *-m-tl'l *-n-tl'l *-tl'l
3 *-0 *_ta2 *-m-ta2 *-n-ta2 *-ta2 *-0-q
Dual 1 *-mi-fi * -Illi-n
'- *-0-mi-fi *-0-ni-fi *-ni-fi
2 *-ti-fi *-ri-fi *-m-ti-fi *-n-ti-fi *-ti-fi
3 *-kl'l-fi *-ti-fi *-m-ti-fi *-n-ti-fi *-ti-fi
Plural 1 *-ma-t2 *-ma-t2 *-0-ma-t2 *-0-na-t2 *-na-t2
2 *-ta-t2 *-ra-t2 *-m-ta-t2 *-n-ta-t2 *-ta-t2
3 *-0-t *-ton *-m-ton *-n-ton *-ton *-0-t-l'l-q

any, precedesthem(cf. Table 15.5).Thereis also a fragmentarysetof endings

used in a separateparadigm of reflexive finite forms (the reflexive con-
jugation).As for the material shapesof the personalendingsin the different
categories,many problemsstill await a definitive solution. From the proto-
Uralic point of view, one of the most interesting features is that the
second-personsingularpredicativeending seemsto have been*-n in proto-
Samoyedic,as opposedto *-t in most sub-branchesof Finno-Ugric. The
simple shape*-n is, however,preservedonly in Nganasan,while the other
Samoyediclanguageshave *-n-t!1l, possiblyas a result of the influenceof the
correspondingpossessivesuffixes. In the possessivesets,the second-person
endingsfor all the three numbersshow a morphophonologicalalternationof
*-t- with *-r-. In addition,the suffix-initial *-r- (usedafter stem-finalvowels)
participatesin the normal alternationwith *-1- (after stem-finalconsonants).
The third-personendingsin the predicativeset are identical with the number
markers of the nominal declension. The plural number marker *-t- also
appearsas a componentof the third-personplural reflexive ending.
Finite Conjugation
One of the formal differencesbetweenverbs and nounsin Samoyedicis that
a verbal stemis normally not combinablewith the personalendingswithout
an element of stem enlargement.In this respect, Samoyedicseemsto be
innovative,for it may be assumedthat the useof an unmarkedfinite stemwas
possiblein proto-Uralic, as it still is in Finno-Ugric, as in Finnish mene-n'I
go' (proto-Uralic *meni-m). Most types of stemenlargementin the modem
Samoyediclanguagesmay be tracedback to non-finite derivationalsuffixes
which have developedsecondarilyinto temporal and modal markersof the
finite conjugation. In other words, the origins of the various finite sub-
paradigmsin Samoyediclie in the predicativeconjugationof verbal nouns.
There are considerabledifferencesbetweenthe synchronic systemsof the
individual languages,suggestingthat the functions of the morphological
472 SAMOYEDIC

elementshadnot yet beenfully crystallizedat the proto-Samoyediclevel. One

of the few elementsoccurringin severallanguagesin an essentiallyidentical
function is the past-tensemarker *-sa2-, as in Selkup ap-sa-p 'I ate' of am-
'eats' (proto-Samoyedic*!/lm-). In Nenetsand Enets,this elementhas been
restructuredinto a modal marker(the interrogativemood), but it still retains
a connotationcorrespondingto the earliertemporalfunction. As a verbalnoun
suffix, *=sa2 is attestedonly in a few lexicalized relicts, as in Selkup aps¢
'meat'< 'food' < 'edible' (proto-Samoyedic*!/lm=sa).
Aorist Formation
A central role among the various finite sub-paradigmsformed by stem-
enlargementis played by the aorist, which may be defined as a temporally
ambivalent category referring to either on-going or completed action
dependingon the intrinsic aspectualcontentof the verb, as in Nenets(third
personsingular) yilyeO '([s]he) lives' v. xaO '([s]he) died'. In the functional
sense,the aorist is an archaicfeaturepreservedonly in NorthernSamoyedic,
but the categoryhasformal parallelsin SouthernSamoyedic. Aoristformation
is anothermorphologicalfeature by which Nganasanstandsalone, opposed
to the rest of the Samoyediclanguages.The Nganasanaorist, as in (third
person singular objective) xan-u-q-a-tu of xan- 'places' (proto-Samoyedic
*pen-), is formed synchronicallyby suffixing a vowel (-a- or -e- ) to a stem
showing material propertiesidentical with those observedin the genitive
plural of nominaldeclension.The materialbackgroundof the Nganasanaorist
remainsobscure,but a verbal noun suffix containingthe palatalglide *-y-is
likely to havebeeninvolved. In the otherSamoyediclanguages,two typesof
aorist formation may be observedsynchronically:stem-final consonantsare
normally combinedwith *-nga-, asin Nenets(third personsingular)nger-nga
of nger- 'to drink' (proto-Samoyedic*er-), while stem-final vowels require
a vocalic element,as in Nenets(third personalsingular)xada-o : (objective)
xada-o-daof xada- 'kills' (proto-Samoyedic*ka[=]ta-). A vocalic element,at
least superficially identifiable with *-!/l-, is also presentin combinationwith
the reflexive and plural possessive(objective) personal endings, which
require the additional segment*-y- as a markerof reflexivity or plurality, as
in Nenets(third personsingularwith plural object) xada-y-O-da.There have
beensuggestionsthat the vocalic elementin questionmight alsoultimately go
back to *-nga-, but this identification seems to involve insurmountable
phonological problems. The element *-nga-, however, is also attestedin
Nganasan,where it occurs in a modal function (the interrogative mood).
There are indications that *-nga-, also, was once a verbal noun suffix, still
preservedin relicts, as in Nenets yeng¢ 'river' < *wen=nga from *wen-
'flows'.
Periphrastic Conjugation
It cannotbe ruled out that the formation of the aorist actually involves traces
of an obscuredperiphrasticconjugation,in that the vocalic elementof the
 SAMOYEDIC 473

aorist, as observedin both Nganasanand the other Samoyediclanguages,

might representthe stemof an originally independentauxiliary verb, perhaps
identical with proto-Samoyedic*i- 'to be'. Whetheror not this is so, Nenets
and Enets exhibit a transparentcase of periphrastic conjugation in the
formation of their normal finite past-tenseparadigm,which has the rather
unusual formula of STEM + AORIST FORMATIVE + PERSONAL
ENDING + PAST-TENSESUFFIX, as in Nenets(secondpersonsingular)
yilye-o-nfjJ-syo 'you lived'. Diachronically, this type of conjugationinvolves
the suffixation of the past tense auxiliary form *+i-sa '(it) was' to a fully
conjugatedaorist form of the main verb. The appearanceof the periphrastic
pasttense has,incidentally, diminishedthe formal differencebetweenverbal
and nominal inflection, in that it can also be formed from nouns used
predicatively, as in Nenetsxasawa+syO 'he was a man'. Another possible
exampleof the periphrasticconjugationis involved in the SayanSamoyedic
present/futuretense paradigm, which is based on a Southern Samoyedic
gerund suffix, as in Kamas kal-la-m 'I (will) go' < *kan=la+V-m (proto-
Samoyedic*kan- 'goes').

Non-finite Forms
While many of the pre-proto-Samoyedicverbal noun suffixes were restruc-
tured into temporal and modal markersof the finite paradigm,there always
existed a system of productive non-finite forms, as well. At the proto-
Samoyediclevel, threefunctional categoriesof suchforms seemto havebeen
distinguished:infinitives (verbal substantives),participles(verbal adjectives),
and gerunds (verbal adverbs). It is difficult to reconstruct the proto-
Samoyedicsystem becauseof cross-connecteddevelopmentswhich have
affected the functional distinctions. Proto-Samoyedicverbal noun types
datingback asfar as proto-Uraliccomprise:

• the infinitive in *=ma2 , as in Nenets to=wa '(the act of) coming' <
*toy=ma (proto-Samoyedic*toy- 'comes');
• the perfectparticiplein *=m!1l-, as in Nenetsto=wi '(one who has)come'
< *toy=m!1l=y!1l;
• the negative participle in *=m!1l=ta=ma2 - (with the abessiveelement
*=ta2- < proto-Uralic *=kta2-), as in Nenetsto=wodawe=yO '(one who
has)not come' < *toy=m!1l=ta=ma=y!1l.

The proto-Uralic verbal noun formatives *=ta2 and *=pa2 survive in

Samoyedicas *=n(=)ta(=) and *=m(=)pa(=), with an apparentlysecondary
nasal element, as in the Selkup imperfect and perfect participles ilfjJ-ntfjJ-[i
'(one who is) living' < *iHi=n(=)ta=y!1l v. ilfjJ-mpfjJ-1J '(one who has) lived' <
*ila=m(=)pa=y!1l. Among the gerunds,there are only two which can with
reasonablecertainty be consideredas proto-Samoyedic:the co-ordinative
(modal) gerund in *=ki( -), as in Nenets yilye-syO '(by way of) living' <
474 SAMOYEDIC

*ila=ki, and the subordinative(conditional)gerundin *=put-, as in Nganasan

tuj-xaq '(under the condition of) coming' < *toy=put. In view of their
phonologicalshapes(involving the vowels *i and *u), both seemto be post-
proto-Uralic innovations. Both also share the characteristicof having a
fragmentary nominal paradigm, suggestingthat they may originally have
functioned as verbal nouns. This is particularly obvious in the caseof the
subordinativegerund, which in combination with the possessivesuffixes
requiresthe presenceof a local caseending(dative or locative) of the archaic
simple type, as in Nenets(dative first personsingular) to=bf/Jq-nO 'underthe
condition that I come' < *toy=put-ng-njli and (locative) to=bOq-na-n° 'id.' <
*toy=put-na-njli.
Imperative
In the Samoyedicsystemsof verbalconjugation,the imperativesub-paradigm
occupiesan exceptionalposition, in that it not only variesconsiderablyfrom
languageto language,but it also shows a conspicuouslack of structural
coherencewithin each individual language. There is no doubt that the
imperativein pre-proto-Uralicoriginally existedas a specialcategoryfor the
secondperson,andin the modernSamoyediclanguagesto this day it remains
somewhat unclearwhetherit is at all possibleto regardthe imperativesub-
paradigm as comprising other persons;the third person is notably ques-
tionable. The only Samoyedicimperative form that can be unambiguously
datedback to proto-Uralic is the secondpersonsingularin *-k, as in Nenets
nger_O-q '(thou) drink!' < *er-jli-k. The synchronicstatusof the element*-k
and its reflexes in the modern languagesprovides an interestingobject for
synchronicmorphologicalanalysis,for it shouldperhapsnot be considereda
personalending in the samesenseas the ordinary personalendingsof the
finite paradigm.However, it contrastswith the true personalending *-tjli in
the second-personsingular imperative form with a definite object, as in
Nenets nger-t> '(thou) drink it!' < *er-tjli. This form is also of structural
interest, for it is a rare example of finite conjugation without a stem
enlargement,a peculiarity shared only by the other Northern Samoyedic
second-personsingular imperative forms (reflexive and plural/dual objec-
tive). In contrast,the third-personforms contain the normal aorist enlarge-
mentin combinationwith specialimperativemarkers.The imperativemarker
for the simple third-personsingularand plural forms may havebeen*-ya in
proto-Samoyedic,as in Nenets nger-nga-ya 'may he drink' < *er-nga-ya.
Curiously, an entirely different marker, reconstructableas *( -jli)-m-, appears
beforethe possessive(objective)andreflexive personalendings,as in Nenets
(third-personsingular objective imperative) nger-nga-m-da < *er-nga-m-ta.
The original identity of both *-ya and *(-jli)-m- remainsobscure.
Negation
With the exceptionof Selkup,all the modernSamoyediclanguagespreserve
the proto-Uralic featureof expressingnegationby meansof a negativeverb,
 SAMOYEDIC 475

probablyto be reconstructedas proto-Samoyedic*i-. With somevariation in

the synchronic details of the system, the paradigm of the negative verb
basicallycomprisesall the normalmorphologicaldistinctionsconnectedwith
the categoriesof personalconjugation,including tenseand aspect,as well as
the formation of verbal nouns and related forms. The main verb, which
normally follows the conjugatednegativeverb, hasthe invariableconnegative
form in *-k, identical with the second-personsingular imperative ending.
However, synchronic evidence from the modem languages,particularly
Nenets,suggeststhat the negationof verbal nounsin proto-Samoyedietook
place by using the negative verb not in the connegative but in the
correspondingverbal noun form, as in Nenets(imperfectparticiple) nyi-nya
yilye-nya 'not living'.

Further Reading
Up-to-dateliterature on Samoyediediachronic morphologyis scarce,but a
wealth of information on earlier researchmay be obtained from Ktinnap
(1971 and 1978). Some useful material is also found in Mikola (1988:
236-59).Aspectsof comparativemorphosyntaxare dealt with in Terescenko
(1973) (general syntax), Katschmann(1986) (expressionsfor 'to be'). The
proto-Uraliebackgroundof Samoyedicmorphologyis discussedin lanhunen
(l98lb).

Lexicon

Common Uratic Vocabulary

In accordancewith the assumptionthat Samoyedicforms one of the two
principal branchesof Uralic, the Samoyediclanguageshave relatively few
lexical items that can uncontroversiallybe regardedas direct inheritances
from proto-Uralic.Evenby a very optimistic estimation,the numberof proto-
Uralic underived stems surviving in Samoyediccannot possibly be much
more than 150.Apart from pronounsand other auxiliaries,the commonitems
areconfinedto basictermsfor body partsandbodily functions,kinship terms,
celestialobjects,meteorologicalphenomena,and topographicalconcepts.Of
some chronological and areal relevanceare items pertaining to primitive
technology, apparently reflecting the neolithic level of development,and
termsfor faunaandflora, pointing to the borealforestenvironmentof the Ural
region.
Common SamoyedicVocabulary
The number of stem morphemesattestedin both Northern Samoyedicand
SouthernSamoyedicseemsto be approximately700. In addition, some 150
lexicalizedderivativeswith a similar distribution are known. If items shared
by any two Samoyediclanguages,including adjacentones (such as Nenets
andEnets,or Kamasand Mator),were also to be recognizedas deriving from
476 SAMOYEDIC

the proto-language,the numberof proto-Samoyedicstemmorphemeswould

probably rise to over one thousand.This vocabularycoversa wide range of
socio-technologicalconcepts, many of which are connected with post-
proto-Uralic cultural innovations.Importantin the NorthernEurasiancontext
are the proto-Samoyedicitems relating to reindeer-breeding,such as *tejlj
'reindeer' and *Cjljrkjlj( =)y 'reindeer calf', as well as those relating to
shamanisticbeliefs, suchas *cadi=pa 'shaman', *pe=n+kir 'shaman'sdrum',
*sampjlj- 'to shamanize',*kaykjlj 'spirit'. Of ethnohistoricalinterestare also
the two proto-Samoyedichydronyms *yentjlj(=)si(=ng) 'the Yenisei' and
*yam 'sea, large river Othe Ob' ?Lake Baikal)" which define the main
geographicaldimensionsof the proto-Samoyedicterritory. As far as vocabu-
lary pertainingto the details of natural environmentand material culture is
concerned,a massivesemanticand material differentiation has taken place
betweenthe three ecological groups of the Samoyedic-speaking peoples.A
casein point is proto-Samoyedic*mjljya 'land, terrain', which in the modem
languageshas meanings ranging from 'tundra' to 'taiga' to 'mountain'.
Particularly many idiosyncratic items of cultural vocabulary seem to be
presentin Nganasan.

Numerals
Although the mutual relationshipsof the main sub-branchesof Finno-Ugric
are still a matterof discussion,it is of considerabletaxonomicimportancethat
the Finno-Ugric languagespossesscommon basic numeralsfrom 1 (or, at
least, 2) to 6, of which only two have cognatesin Samoyedic.Moreover, of
these,only the item for '2' occurs in identical functions in Samoyedicand
Finno-Ugric, while the cognateof the Finno-Ugricitem for '5' appearsin the
function of '10' in Samoyedic.The restof the Samoyedicbasicnumeralsvery
probably representpost-proto-Uralicinnovations (cf. the list of numerals,
below).

1 *o(-)p derivativeof *0(-) '1'

2 *kita < proto-Uralic *kektiil*kakta
3 *naku(-)r ? derivativewith unidentifiedelements
4 *tettjlj
5 *Sjljmpjlj-Iangkjlj derivativeof an otherwiseunknownroot
6 *mjljktu( -)t ? derivativewith unidentifiedelements
7 *seytwjlj < *seyptjlj < pre-proto-Tocharian
8 *kita( -y-n)+tettjlj '2x4', compositumof *kita and *tettjlj
9 *amay-tumjlj derivativeof *amay 'other'
10 *wiijljt < proto-Uralic *wixti '5'
100 *yiir < proto-Bolgar-Turkic *yiir

Among the numeralswith no cognatein Finno-Ugric, only the item for '4'
seemsto consistof a single indivisible indigenousstemmorpheme,while the
 SAMOYEDIC 477

other items are either composita, derivatives, or loanwords. The overall

structureof the numeralsystemsuggeststhat therewas a time whenthe count
extendedonly from 1 to 6, with the original item for '5' perhaps also
functioning as 'lO'. After the borrowing of the item for '7', the expressions
for '8' and '9' were createdat sucha late datethat their representationin the
modem languagesis in fact not fully uniform. The expressionsfor '9', in
particular, vary considerably,with the proto-Samoyedicitem being reliably
attestedin only two languages(NganasanandKamas).

Foreign Contacts
After the break-upof proto-Uralic, the early stagesof pre-proto-Samoyedic
probably long continued to maintain an areal contact with Finno-Ugric,
especiallyUgric. As traces of this interaction, Samoyedic possesses a few
lexical items which may be explainedas early loanwordsfrom Ugric or pre-
Ugric, such as *num 'sky, god', *kalma 'dead body, grave'. Another
concretelyidentifiable linguistic entity in the neighbourhoodof early pre-
proto-Samoyedicseems to have been some kind of pre-proto-Tocharian,
which also yielded a few loanwords, notably the numeral *seytW\?i '7'.
Contactswith Indo-Europeanthen probably continuedduring the period of
the early Iranianpresencein southernSiberiain the secondto first millennium
BeE. During the ethnic movementsof the subsequent'Hunnic' period, just
around the time of the break-up of proto-Samoyedicunity, Samoyedic
received an important layer of proto-Bolgar Turkic loanwords, including
items pertaining to animal husbandry,such as *yunt\?i 'horse' and *kapt\?i-
'castrates' , as well as the numeral*ytir '100'. It cannotbe ruled out that proto-
Samoyedicalso had direct contactswith early forms of Mongolic, Tungusic,
and Yeniseic, but the linguistic evidenceof these contactsremains incon-
clusive. In any case,after the break-upof proto-Samoyedic,the individual
Samoyediclanguageshave enteredinto contact with all the neighbouring
languages,which include idioms of the Turkic (Khakas, Tuva, Dolgan),
Mongolic (Buryat), Tungusic (Evenki), and Yeniseic (Ket, Kott, Arin)
families, as well as of the Finno-Ugric branch (Khanty, Mansi, Komi). To
some extent, this interaction extends beyond the limits of the lexicon.
Particularly close areal bonds of typological interaction have been formed
betweenSelkupandYeniseic(Ket), as well asbetweenSayanSamoyedicand
Turkic (Khakas,Tuva). As the most recentcontactphenomenon,all Samo-
yedic languageshaveincorporatedlexical elementsfrom Russian.In addition
to lexical impact, Russianinfluence is currently also penetratinginto the
syntacticandmorphologicalsystemsof the surviving Samoyediclanguages.

Further Reading
The lexical items sharedby NorthernSamoyedicandSouthernSamoyedicare
listed in Janhunen(1977a).The proto-Uralic items surviving in Samoyedic
are discussedin Janhunen(1981a)and listed in Sammallahti(1988: 536-41).
478 SAMOYEDIC

Other studies pertammg to Samoyedic comparative lexicology include

Xelimskij (1986 and 1992-3) (Mator etymologies),Joki (1975) (numerals).
A pioneeringwork on foreign lexical elementsin Samoyedic(Kamas and
Mator) is Joki (1951). Recent discussionsof the topic include Janhunen
(1983) (contacts with Indo-European),Janhunen(1977b) (contacts with
'Altaic'), R6na-Tas(1980) (contactswith Turkic), Xelimskij (1982a)(paral-
lels with Ugric), Xelimskij (1982b)(contactswith Yeniseic),Janhunen(1989)
andXelimskij (1991) (selectedcontroversialissues).An attemptat an analysis
of structuralinteraction(with Ugric) is Katz (1975).

Referencesand Further Reading

Hajdu, P. (1963) The SamoyedicPeoplesand Languages,Indiana University Uralic
andAltaic Series14, Bloomington-TheHague:IndianaUniversity-Mouton.
- - (1968) ChrestomathiaSamoiedica,Budapest:Tankonyvkiad6.
- - (1988) 'Die SamojedischenSprachen', in D. Sinor (ed.), The Uralic
Languages: Description, History and Foreign Influences, Handbuch der Ori-
entalistik 8/1, Leiden: Brill, pp. 3-40.
Janhunen,J. (1976) 'Adalekok az eszaki-szamojed hangtOrtenethez.Vokalizmus: az
elsa sz6tagimaganhangz6k', Neprajzesnyelvtudomany19-20: 165-88.
- - (l977a) SamojedischerWortschatz.Gemeinsamojedische Etymologien,Cas-
trenianumintoimitteita 17, Helsinki: SocieteFinno-Ougrienne.
- - (1977b) 'Samoyed-AltaicContacts:PresentState of Research',in Altaica,
MSFOu 158, Helsinki: SocieteFinno-Ougrienne,pp. 123-9.
- - (l981a) 'Uralilaisenkantakielensanastosta', JSFOu77: 219-74.
- - (1981b) 'On the structureof proto-Uralic', FuF 44: 23-42.
- - (1983) 'On early Indo-European-Samoyed contacts',in 1. Janhunenet al.
(eds), SymposiumSaeculareSocietatis Fenno-Ugricae, MSFOu 185, Helsinki:
SocieteFinno-Ougrienne,pp. 115-27.
- - (1986) Glottal Stop in Nenets, MSFOu 196, Helsinki: Societe Finno-
Ougrienne.
- - (1989) 'On the interaction of Mator with Turkic, Mongolic, and Tungusic',
JSFOu82: 287-97.
- - (1991) 'HraHacaHhI H paClIa~ paClIa~ IIpacaMo~HHCKOll:IIpacaMo~HHCKOll: 5I3hIKOBOH 06~HOCTH',
06~HOCTH', in
Seminar«IIpo6JIeMhI IIpoHcXmK~eHH5IIIpoHcXmK~eHH5IHapo~oBHapo~oB ypaJIhCKOH 5I3hIKOBOIl: ceMhH»,
Izhevsk,pp.16-19.
Janurik, T. (1978) 'A szOlkup nyelvjarasok osztaIyozasahoz',Nyelvtudomanyi
Kozlemenyek80: 77-104.
- - (1982) 'Szamojedhangmegfelelesek.I. Massalhangz6k',Nyelvtudomanyi
Kozlemenyek84: 41-89.
- - (1985) 'Kriterien zur Klassifizierung der Dialekte der Samojedischen
Sprachen',in W. Veenker (ed.), Dialectologia Uralica, Veroffentlichungender
SocietasUralo-Altaica20, Wiesbaden:Harrassowitz,pp. 283-30l.
Joki, A.J. (1951) Die Lehnworterdes Sajansamojedischen, MSFOu 103, Helsinki:
SocieteFinno-Ougrienne.
- - (1975) 'Uber einige Zahlworter im Samojedischen',C3IFU, vol. I,
pp.729-32.
Katz, H. (1975) Generative Phonologie und phonologische Sprachbiinde des
Ostjakischen und Samojedischen,Finnisch-Ugrische Bibliothek 1, Munich:
Wilhelm Fisk.
Katschmann,M. (1986) Nominal- und Esse-Satzin den samojedischenSprachen.
 SAMOYEDIC 479

Dargestellt anhand ausgewiihlten Belegmaterials,Fenno-Ugrica 9, Hamburg:

Helmut BuskeVerlag.
Ktinnap, A. (1971, 1978)Systemund Ursprung der kamassischen Flexionsuffixe.vol.
I: Numeruszeichen und Nominalflexion; vol. II: Verbalflexion und Verbalnomina,
MSFOu 147, 164, Helsinki: SocieteFinno-Ougrienne.
- - - (1985) 'Zur K1assifizierungder Dia1ektedesSe1kupischenund Kamassichen' ,
Dialectologia Uralica, VeroffentIichungender SocietasUralo-Altaica 20, Wies-
baden:Harrassowitz,309-16.
Miko1a, T. (1988) Geschichteder samojedischenSprachen,in D. Sinor (ed.), The
Uralic Languages:Description, History and Foreign Influences,Handbuchder
Orientalistik8/1, Leiden: Brill, pp. 219-63.
R6na-Tas,A. (1980) 'On the earliest Samoyed-Turkishcontacts',C5IFU, vol. III,
pp.377-85.
Salminen,T. (1990) 'Phonologicalcriteria in the classificationof the Nenetsdialects',
C7IFU, vol. mc, pp. 344-99.
Sammallahti,P. (1988) 'Historical phonology of the Uralic languages,with special
reference to Samoyed, Ugric and Permic', in D. Sinor (ed.), The Uralic
Languages: Description, History and Foreign Influences, Handbuch der Ori-
entalistik8/1, Leiden: Brill, pp. 478-554.
Terentjev, v.A. (1982) 'K Borrpocy 0 peKoHcTpyK~llllrrpaCaMO):lllHCKOro rrpaCaMO):lllHCKOro H3hIKa',
Sovetskoefinno-ugrovedenie 18: 189-93.
Terescenko,N.M. (1973) CllHTaKcllc CaMO):lllHCKllX H3hIKOB, Leningrad:Nauka.
Xelimskij, E.A. (1976) '06 O):lHOM rrepeXO):lHoM ceBepHOCaMO):lllHCKOM):IllaJIeKTe (K
llCTOpll'IeCKOH ~llaJIeKTOJIOrllll
rrepeXO):lHoM HeHe~KoroHeHe~Koro H3hIKa)', IIpollcxo~eHlle):IllaJIeKTe
a6opll-reHoBCll6llPllllllX H3hIKOB 3: 89-93.
- - - (1978) 'PeKoHcTpyK~llH 'PeKoHcTpyK~llHrrpaceBepHOCaMO):lllHCKllX(PSS)JIa6llaJIll30BaHHhlx
rJIaCHhIX HerrepBhlx CJIOrOB', in KOH<pepeH~llHKOH<pepeH~llH«IIpo6JIeMhI peKoHcTpyK~llll»,peKoHcTpyK~llll»,
Moscow: Institut Yazykozhaniya,pp. 123-6.
- - - (1982a) .o.peBHeHrrIlle BeHrepcKo-CaMO):lllHCKlle H3hIKOBhIe rrapaJIJIeJIll,
(JIllHrBllCTll'IeCKaH II 3THoreHeTll'IeCKaH 3THoreHeTll'IeCKaH
llHTeprrpeTa~llH), Moscow: Nauka.
- - - (1982b) 'Keto-Uralica', in KeTcKllH C60pHllK (3), aHTpOrrOJIOrllH,
3THorpaq)llHMllcpOJIOrllH, JIllHrBllCTllKa, Leningrad:Nauka,pp. 238-50.
- - - (1983) 'PaHHlle 3TarrhI 3THoreHe3aII 3THll'IeCKOH llCTOPllll CaMO):lllH~eB CaMO):lllH~eBB
CBeTe H3hIKOBhIX ):IaHHhIX', in IIpo6JIeMhI 3THoreHe3aII 3THll'IeCKOH llCTOPllll
CaMO):lllHCKllX HapO):lOB, Omsk,pp. 5-10.
- - - (1985a) 'Die Feststellung der dialektalen Zugehorigkeit der enzischen
Materialien', in Dialectologia Uralica, Veroffentlichungender SocietasUralo-
Altaica 20, Wiesbaden:Harrassowitz,pp. 303-8.
- - - (1985b) 'CaMO):lllHCKo-TYHryccKlle JIeKCll'IeCKlle CBH3ll II llX 3THOllCTO-
pll'IeCKlle llMrrJIllKa~ll',
Novosibirsk: in YpaJIO-aJITallCTllKa. ApXeOJIOrllH, 3THorpacpllH,
H3hIK, Novosibirsk: Nauka,pp. 206-13.
- - (1986) 'Etymologica1-48' NyelvtudomdnyiKozlemenyek88, p. 119-43.
- - - (1991) 'On the interactionof Mator with Turkic, Mongolic, and Tungusic: a
rejoinder',JSFOu83: 257-67.
(1992-3) 'Etymologica 49-79', MaTepllaJIhI rro 3TllMOJIOrHll
MaTopcKo-TaHrllHcKo-KaparaccKoro H3hIKa', Nyelvtudomdnyi Kozlemenyek,
pp. 101-23,Budapest.
16Nganasan
EugeneH elimski

The Nganasanlanguage,known also as Tawgi or Tawgi-Samoyed,is spoken

by the northernmostethnic group in Eurasia.The Nganasancall themselves
Nya (daa, formally an adjectivefrom da 'mate').The appellation'Nganasan'
is basedon lJanasa (plurallJanasan;)?) 'humanbeing, person' (never used,
however as a self-designation);the appellation 'Tawgi' is basedon Nenets
tawi? 'Nganasan'.
Until recentlyall Nganasansled the nomadiclife of tundradwellers- wild
reindeerhunters,fishermenanddomesticreindeerbreeders- in the centraland
northernpartsof the Taimyr Peninsula(excludingits coastalareas).Sincethe
1960sthe majorityof thepopulation,which accordingto censusdatanumbered
about900 personsin 1979 and about 1,300 personsin 1989, is concentrated
mainly in threevillages which lie to the southof the original ethnic territory -
Ust-Avam, Volochanka, and Novaya. Only several dozens of Nganasan
families continueto live ashunterson isolatedspotson thetundra.
The transition to the settled way of life together with other rapid and
dramatic social and economic changes - all carried out in a purely
administrative way - broke the normal transmissionof cultural values to
subsequent generationsandposeda mortal threatto the languageandnational
heritage of Nganasans.The sociolinguistic situation around 1990 was as
follows: peopleagedforty to fifty yearsand over were all perfectly fluent in
their native languageand often unable to communicateeffectively in any
otherlanguage,while amongthoseunderthirty, who had goneor were going
through the experience of living in villages with an ethnically mixed
population,andof studyingin Russian-language boarding schools,only about
one-thirdwas able to understandNganasanandhardly more than ten per cent
were able to speakit. According to my own estimate(the official censusdata
are in this respectheavily distorted), the total number of Nganasannative
speakerswas about 600 in 1990, a numberwhich is permanentlydecreasing
despitethe remarkablyhigh birth rate and naturalgrowth.
The Nganasanlanguageis usedin traditional spheresof the economyand
in families (within the limits determinedby the lack of linguistic proficiency
amongyoungerage groups),but not in administration,culture, trade, or any
other activities of village communities.Recently it began to be taught at
primary schools (practically as a second language even for Nganasan

480
 NGANASAN 481

children). Native-languageliteracy remains the privilege of only a few

Nganasans,mainly graduatesof teacher-trainingcolleges.

Bilingualism
As the sociolinguisticdataaboveshow,Nganasan-Russian bilingualism(with
varying degreesof proficiency in the two languages)is at presentespecially
typical of middle-agedNganasans(between thirty and sixty); this bilin-
gualism is strictly unilateral: among the Russian-speakingpopulation of
Taimyr (predominantlynewcomersor temporarydwellers) thereseemsto be
not a single personwho speaksNganasan- this may, however, be at least
partly dueto the extrememorphophonemiccomplexityof the language.
In all threevillages mentionedabovethe Nganasanslive togetherwith the
Dolgans, an ethnic group of mixed (predominantlyTungusic) origin whose
languagecan be classifiedas a dialect of (Turkic) Yakut. Many Nganasans,
especiallyin the eastern partof Taimyr (Novaya), are able to understandor
evento speakDolgan(but usuallynot vice versa).
The river basinof Pyasina(WesternTaimyr) was for many generationsan
area of Nganasan-Enetsbilateral bilingualism (to some extent also of
trilingualism, with Nenetsas the third language),but the transitionto settled
life meant an end to thesevery close contacts(though there are still many
mixed Nganasan-Enets oldermarriedcouples).
An importantelementof the ethnolinguisticsituation in the areawas the
creation and functioning of the so-called 'Govorka' -a Taimyr Pidgin
Russianwith analytic grammar,which follows Ural-Altaic morphosyntactic
patterns(seeHelimski 1987). Until the first half of our century it servedas
an important meansof interethnic communicationin the area, but has now
largely beenoustedby StandardRussian.Many elderly Nganasans, especially
males,still speak'Govorka'fluently.
In earlier times, before the formation of the Dolgan ethnicum,there also
existed Nganasan-Evenkicontacts, probably with a certain degree of
bilingualism.

Dialects
Nganasanis split into two very close dialects,Avam (spokenin the western
and centralpartsof the ethnic territory by c.75 per centof all Nganasans)and
Vadey (spokenin its eastern partand under strongerDolgan influence). Still
smallerare the differencesbetweenthe two subdialects,PyasinaAvam(now
spokenin Ust-Avam and Volochanka) and Taimyra Avam (now mainly in
Volochanka).
The dialectal distinctions in phoneticsare not structural and mostly are
restrictedto the choice of preferredvariants: Pyasina,Vadey 8 : Taimyra 4
(cacuminald) or 8, dependingon position; Avam Ji (idiolectally j): Vadeyj
(idiolectally Ji); the reducedmid vowel ;} is in Vadey more closedthan in
482 NGANASAN

Avam). Other dialectal distinctions have never been specially studied. It is

known, however, that they do not go beyond very few lexical and
morphological items (e.g. Vadey torautu 'works': A vam IJojbautu 'works'
versus torautu 'is good for, is useful'), and the Nganasandialects are
completelymutually intelligible.
The descriptionbelow refersto the PyasinaA vam subdialect.

Historical Development
There are sufficient reasonsto believe that an early form of the Nganasan
language was spoken by reindeer-breedingSamoyedswho reached the
Taimyr peninsulain the first millennium CE from the south,moving alongthe
Yenisei River. They assimilated linguisticallythe local population of wild-
reindeerhunters,adoptingcertain elementsof material and spiritual culture
from them.
Nganasandiffers from all other Samoyediclanguages,including also its
NorthernSamoyedicneighbours(NenetsandEnets),in at leasttwo important
points of morphology, namely the marking of the locative caselanhunen(*-nt~-na
instead of *-b-na) and the formation of the presenttense (see lanhunen
1991).
Of utmostimportancefor Uralic linguistics is the obviousandfar-reaching
similarity between Nganasanconsonant gradation (see below) and the
correspondingphenomenonin Baltic-PennicandSaamic.The co-existenceof
two different mechanisms,syllabic gradation and rhythmic gradation (cf.
so-called 'stem gradation' and 'suffixal gradation' in Baltic-Pennic) is
particularlysuggestive.In view of the absenceof typologicalparallelsin other
languagesof the world it seemshighly improbablethat thesemechanisms
could be separateand wholly independentdevelopmentson the westernand
the easternperipheryof the Uralic language-familytree. The similarity must
be explained either by the parallel influence of a common unknown
substratum,or - and this is much more likely - by the phonologizationof a
phoneticmechanismthat alreadyoperatedin the Uralic proto-language.
Otherwise Nganasanshows numerous innovations, both shared with
Nenets and Enets (such as the rise of word-initial prothetic nasalsbefore
vowels, the developmentof glottal stopandits morphophonemicalternations,
the wide usageof incorporatedparticlesand modal markers)and individual.
Unique to Nganasanis the relatively recent - partly already after the
beginning of Russian-Nganasanlinguistic contacts, i.e. not before the
seventeenthcentury- vowel chain-shift*ti > i, *u > ii, *0 > u, *a > 0 and also
*e > r, with numerousspecial casesof retentionsand deviations from this
generalschemedueto phoneticpositions.
 NGANASAN 483

Table 16.1 Consonantsand their Cyrillic graphicsymbols

Labials Dentals Palatals Velars Glottals

Voicelessstops p<6>1,<II> t<T> ti<q> k<K> 7<">

Voicedstops b<6> d<A> di <A>2 g<r>
Voicedfricatives 6<3> j <ii>
Voicelessfricatives s<c> sj <C>2 h<x>
Nasals m<M> n<H> nj <H>2, <If>
I)
<Hh>3
Laterals l<.rr> Ij <JI>2,
<JIh>3
Vibrants r<p>
Note: 1 Beforevoicelessstopsandfricatives. 2 Beforethe 'yotated'vowel letters- <e>. <e>.
<H>, <"a>, <10>, <lOa>, <H>. 3 Beforeconsonants.

Table 16.2 Vowels and their Cyrillic graphicsymbols

Front Nonjront
Unrounded Rounded Un rounded Rounded

High i <i>l, <H> ii <y>, <10>2 "i <hI> ii <hI>

Mid e <e> ii <ia>l, <"a>
;} <;}>,<e>2
Low
Mid ia <ia>l, <"a> a <a>, <H>2 ii <ia>l, <"a>
Note: 1 Usedafter <11>, <C>, <H>, <]I> to denotetheir dental (non-palatal)character.2 Used
after <11>, <C>, <H>, <]I>, to denotetheir palatalcharacter.

PhonologyandMorphophonology
Phonemesandtheir Distribution
The phonemeinventory of Nganasanconsists of twenty-one consonants,
including three 'subphonemes'(phonetically distinct units, which are in
complementarydistribution with other units in genuine words and partly
independentin borrowedwords), and ten vowels. They are shown in Tables
16.1 and 16.2,

Consonants
In Nganasan,as well as in other Northern Samoyediclanguages,there are
systematicrestrictionson the occurrenceof consonantsin different positions.
The maximal differentiationis found in intervocalic position, less free is the
distribution of consonantsword-initially, word-finally, and in clusters(which
occur only word-medially, and until very recently were regularly eliminated
from the two other positionsin borrowings,as well); seeTable 16.3 for the
484 NGANASAN

Table 16.3 Occurrenceof consonantsin diagnosticpositions

[p] t ti k ? s si hbd d [b] di [j] g m n Ii I] Ii r

#_V (-) (+) (+) (+) (-) (+) (+)(+) (+) (-) (+) (+) (+)(+) (+) (+)(+) (+) (+) (-)
(+) (+) (+)
C- V (-) (+) (+)(+) (-) (+)
(+) (+)(+)
(+) (+)(+) (-) (+)
(+) (+) (-) (+) (+)(+) (-)
(+) (+)(+)(+)
(+)(+)(+)
V - V (-) (+) (+) (+) (+) (+)
(+) (+) (+)
(+) (+) (+) (+) (+)
(+) (+) (+) (+) (+)
(+) (+) (+)(+)
(+) (+) (+)(+)
V - C (+) (-) (-) (-) (+) (+)(-) (-) (-) (-) (-) (-) (+) (+) (+)(+) (+) (+) (+)(+)
(+) (+) (+)
(-) (-) (-) (-)
V _# (-) (-) (-) (-) (+) (-) (-) (-) (+) (-) (+)(-) (-) (+) (-) (-)(+)

Key: (+) occursonly in non-nativewords, including old and otherwiseadaptedborrowings.

(-) occursonly in recentborrowings.- in free alternation.

occurrenceof consonantsword-initially, syllable-initially after consonants,

betweentwo vowels, syllable-finally beforeconsonants,andword-finally.
This table lists also the 'subphonemes'[p] [6] [j] and thus allows for the
possibility of treating[p] in genuinewords as an allophoneof b (or of h, which
is lessplausiblephonetically,but historically adequate:Nganasanh < *p), of
treating [6] in genuinewords as an allophoneof d (or vice versa), and of
treating [j] as an allophone(partly in complementarydistribution, and partly
in free alternation)of cPo
Which of the phonological interpretations of these 'subphonemes'is
adoptedis not crucial. As a matter of fact, it is questionablewhether the
phonological level itself (as a stage between deep morphophonemicand
surfacephoneticrepresentations) is a psychologicalreality for the speakersof
Nganasan,andwhetherphonologyis a necessarypartin the descriptionof this
language.
The other allophonic phenomenacharacteristicof the consonantsare as
follows:

I is in free variation with the palatal affricate [(9]; in fact, there is but slight
phonetic difference between these two consonantsin Nganasanpro-
nunciation;
the glottal stop ? is realizedword-medially usually as a breakin articulation
after a vowel; optionally even this break may be absent,so that a word-
medial glottal stop is reducedto zero - this optional effect is especially
frequentin the TaimyraA vam subdialect;
b betweenvowels is idiolectally pronouncedas the correspondingvoiced
fricative
vowels
[~];
in Taimyra A vam Nganasan,t5 is usually pronouncedas the corresponding
voiced stop, cacuminal[<;1] or postdental[d] (in M.A. Castren'sNganasan
recordsfrom the l840s, t5wasnot distinguishedfrom d at all);
the sibilantshaverelatively rare optional free alternants:[8] (voicelessdental
non-sibilantfricative) for s, and [si] for si;
in archaicpronunciation,which is still fairly frequentamongelderly women,
 NGANASAN 485

a labial fricative ([<1>], [f]) or a labiovelar[hW] is usedinsteadof h (Castren's

materials and evenrecords from the beginning of the twentieth century
have <f> in correspondingwords); moreover,the consonanth is realized
as a voicedvelar fricative [y] betweentwo vowels;
alliabials andvelars, 8, and r are strongly palatalizedbeforefront vowels.

Vowels
The inventory of vowel phonemesincludes eight short monophthongsand
two diphthongoidswith very short initial elements,ia and U a. The diph-
thongoidsare more liable to positional and idiolectal variation than the other
vowels. In particular,idiolectally in casualpronunciationthey lose their initial
element(and thus their contrastwith a): k;J/)t"asu~ syllables '(it) faded',
syllables - bbtasu~
kiiii?a - kiiiila '(s)he died'; in final syllables before a glottal stop they are
a:
monophthongizedinto ii and k;J/)t"al [bpta?] 'fade!', kiiii?al [ktiti?a?]
'they died'. After labials, velars, 8, or r the vowel ia is often pronouncedas
a or ii with a strong palatalizationof the precedingconsonant:hial] [Ilia1)],
[Ilia1)] 'palm'.
Thereare certainrestrictionson the occurrenceof vowels:

the phonemese and 0 occur (in genuineor adaptedwords) only in the first
syllable; the sounds [e] and [0] in non-first syllables are the phonetic
manifestationsof the phoneme~;
the front vowels (i, e, i a, ii) do not occurin the first syllable after dentals;
the vowels ~, i~ u do not occur in the first syllable (i" and u - also in other
syllables) after palatals,being in this position neutralizedwith e, i, and ii,
respectively;
the vowels ~ and 0 are neutralized after labial consonantsand h; the
correspondingsequenceshave the phonetic shapes [bo] , [mol (before
labial consonantsand labialized vowels),(otherwise),
[b~], [m~] (otherwise),[ho] (in
the first syllable), [h6] (after a syllable with ii), [ho] or [hW~] (otherwisein
(otherwise),
non-first syllables);
the vowels a and ia are neutralizedin favour of a after palatals.

The phonetic stock of Nganasan is very rich in long vowels and

diphthongs,but (apartfrom the caseof ia and u a, seeabove)it is reasonable
to treat and to transcribe them as vocalic sequences.This approach is
supportedby abundantevidencefrom phonetics(normally a long vowel or a
diphthongis approximatelytwice as long as a single vowel), accentologyand
metrics (a long vowel or a diphthong usually counts as two morae, as seen
alreadyby Hajdu (1964», morphophonology(when necessary,the morpho-
phonemic rules are applied separatelyto each part of long vowels and
diphthongs),and morphology(very often the parts are divided by morpheme
boundaries);seerelevantexamplesin the following sections.
The diphthongshave i, ii, u, a, ia and (only after high vowels) ~ as their
486 NGANASAN

secondcomponent.Contractionsare not typical, but the reducedvowel ;} is

assimilatedby first componentsinto the correspondingmid vowel, and then
optionally contractedwith first componentsinto the correspondinglong mid
vowel: sii;} [sjie - sjee] 'hole'; hii;} [hiuo - hi60] 'year'; honsi";}a'[hons'ieO'i-
honseec,'i] '(he) had it'; niomu;} [niomuo - njomoo] 'hare (attributive),
(derivedfrom niomu 'hare').This assimilation(and contraction)accountsfor
the presenceof the vowels [0], [e] (and also of [e], [0] in non-first syllables)
in the phoneticstockof Nganasan,thoughthey do not belongto the phonemic
system.The assimilationoccursalso in ;}U [ou] and ;}ti [oti].
The simple vowels i, ii, i~ u, a, ;} participatealso in the formation of long
vowels: hii 'night'; b;}jkiiii?, pG of b;}jku;} 'old'; hiTm- 'to be frightened';tuu,
sG of tuj 'fire'; niaa 'Nganasan'(derivedfrom nia 'mate');s;};} 'heart'.Other
long vowels, namely [eel [60] [eel and [00], are phoneticallysecondary(see
above).

Orthography
The graphic system of Nganasan(Terescenko 1986, with modifications
officially adoptedin 1990),basedon the RussianCyrillic system,is presented
in Tables 16.1 and 16.2. Its slightly modified version (e.g. with <1> instead
of <">, <0> instead of <3>, and without <b» is used in Helimski (ed.)
(1994).
Becauseit hasinheritedthe Russianorthographictradition, which denotes
'hard' and 'soft' consonantswith the samesymbol, anddistinguishesthemby
the use of different symbolsfor the following vowels, the graphic systemof
Nganasanis not phonological.Nevertheless,in principle it is automatically
convertible into phonemic transcription,if both the conventionsconnected
with the 'yotated' vowel letters and the above mentioned phonotactic
restrictions are taken into account.For example,the Cyrillic <e> is usedin
the first syllable for e, and in non-first syllables, where e does not occur, it
is used to indicate the vowel ;} plus the palatal quality of the preceding
consonant;thus <MeHb,IJ;5I,n:e>=menidiadi ;} 'new'.
Long vowels and diphthongsare written as sequencesof graphemes(as in
phonemictranscription).Besidethe graphic symbolsin Tables 16.1 and 16.2,
otherCyrillic letterscan be usedto write borrowedwords.

Accent and the Rhythmic Organization of Words

The main stressis normally placedon the penultimatevowel or on the vocalic
sequencewhich includes the penultimate vowel: kQru? 'house', kor!!:.o;}?
'houses',kiimaa [kitimaa] 'knife', baarb;} [kitimaa]
[baarb~] 'master,chief', b;}/;}Ub
[kitimaa] 'a kind of movable dwellingon runners'.This generalprinciple is
[b~loub]
optionally violated by the retraction of stressfrom a high vowel or ;} to the
vowel (usually an open one) in the preceding syllable: baT?!:.sii - b!:J:.rusii
'devil'.
Longerwords(with five syllablesandmore- suchwords arevery common
 NGANASAN 487

in Nganasan,cf. the text at the end of this chapter)are usually divided into
two, three, or four - potentially even more - rhythmic groups. Each group
typically containstwo syllables(two phonologicalvowels), andthe last group
has two or three syllables.It is very common- especiallyfor verbal fonus -
that the stemand the derivationalsuffix (or suffixes) arebisyllabic, so that the
boundariesbetween groups in most cases coincide with the boundaries
betweenmorphemes,while the last group includes a cluster of inflectional
suffixes (sometimestogetherwith a monosyllabicderivationalsuffix).
The boundariesbetweenrhythmic morphemic groups are very distinct:
sometimesthe breakin articulationcreatesthe acousticimpressionof a glottal
stop (which, as mentionedabove,is also realizedword-medially as a break).
The last group has, according to the general rule, the main stress on its
penultimatevowel, and all precedinggroups receive additional stresseson
their first vowels. In the following examples,rhythmic groupsare separated
by the symbol'('; secondary,group-initial stressesareindicatedby a colon):

ko: (5 u(siu(5fl:.ni ;} 'I will kill them'

ki:nt;}(l;}: bti(ky.tjiI] 'you are smoking'
ki:nt;}J;}: bti(kutiin;} 'I am smoking'
b:ri(g;}:PiAni 'in marches'
b:ri(g;}: Pi(tinin;} 'in my marches'
b:ri(g;}: Pi(ri:· a(tinin;} 'only in my marches'.

The rhythmic organizationof words plays an important role also in the

morphophonologyof Nganasan,regulating the phenomenonof rhythmic
gradation;seebelow.

Morpbopbonology

Alternations and Morpbopbonemes

The abundanceand complexity of productive morphophonologicalalterna-
tions place Nganasanin an exclusive position among the Uralic languages.
This uniquenessis the result of the co-existenceand cumulative effect of
severalrelatively independentmechanisms,including glottal stop alternation,
truncation,rhythmic gradation,syllabic gradation,nunation,stemalternation,
vowel hannonyand accommodation.Most of thesemechanisms- syllabic
gradation and vowel hannony in particular - have largely lost their
synchronicphoneticmotivation; this does not preventthem, however,from
remainingregularandproductive.
As a resultof the above-listedprocesses,Nganasanexhibits greatdiversity
of both radical and suffixal allomorphs;to serveas an illustration, Table 16.4
and Figure 16.1 accountfor the distribution and generationoften allomorphs
Table 16.4 Allomorphy: suffix of the renarrativemood

Featuresofthe precedingstem(= sequenceofmorphemes)

Evennumberofvocalicmorae Odd numberofvocalic morae
Vocalic Consonantal Vocalic Consonantal
V-stem I-stem V-stem I-stem V-stem I-stem V-stem I-stem

Secondsyllable of suffix is open bal)hu bial) hi" hUal)hu hi al) hi" bahu biahi" hUahu hiahi"

Secondsyllable of suffix is closed bambu biambi" hUambu hiambi" bahu biahi" hUahu hiahi"

Table 16.5 Consonantmorphonemes(andsomemorphonemeclusters),their transformationsand phonemicrealizations

Morphoneme(s) HTKS SJ SJ SJ MH NT IJK NS NJSJ cH cT cTJ cK oS cSJ

RhG, weakgrade b [) g eli eli CH cT OK cS cSJ
SyG, weakgrade b [) g eli eli mb nd I)g nieli nidi
Nunnation I)h nt niti I)k ns nidi
Word-finally ? ?
Syllable-finally (7) (7) (7)
Otherwise(prior to accommodations)h t k s si mh nt I)k ns nisi h LL Lti L k s si

Morphoneme(s) TJ ? B DJ J M N IJ NJ J
LL LL R s
IJ
Word-finally j 1),0 1 j
Syllable-finally (7) p j
Betweenvowels 0 0
Otherwise(prior to accommodations)t i ? b eli di m n ni I) r Ii r 0
An emptycell in this table meansthat the correspondingmorphonemeundergoesno transformationin the correspondingposition or doesnot occur there.
Note: I 1] in monosyllabicwords, 0 (sometimes1] - 0) in longer words.
 NGANASAN 489

Figure16.1 Generationof allophones:suffix of the renarrativemood

Rhythmic Syllabic Vowel Accommodation

gradation gradation harmony
68 57
HAMHU I HAMHU 6 8
I bAMHU~bArnhU 7
gradation:;:bamhu-baJ)hu
7 biarnhi:-biaIJhi:
68
7
bAmbU~bambuVowel
7
8 b1ambi:
7 7
HAcHU --bAhU~bahu Vowel
8 biahi:
7 57 7
hAMHU 3 -hAhU~harnhu~harnhu -huaIJhu
hAMHU~hArnhU
68
h1amru-h1aIJhi:
7 7
hAmbU~hambu-hUambu harmony
74
8 hiambi:
7
hACHU -hAhU ~hahU---.
hiahi: hUahu
8 hiahi:

Key: I after vocalic stems. 2 after consonantalstems. 3 after stemswith evennumber

of vocalic morae. 4 after stemswith odd numberof vocalic morae. 5 secondsyllable of
suffix is open. 6 secondsyllable of suffix is closed. 7 after stemsof the first harmonic
class.8 after stemsof the secondharmonicclass.

of the renarrativemood suffix, allomorphswhich rangefrom -huambu-,as in

kou-huambu-l'they are said to remain', to -biahi~,
as in kolioi~biahi~
1 'they
are saidto be fishing'. Otheralternations,with their occasionallyparadoxical-
seemingoutcomes,are illustratedbelow, as well as by paradigmatictablesin
the sectionon morphology.
The characterof Nganasanmorphophonologymakes it convenient to
describethe compositionof morphemesin termsof morphophonemes. Below,
this approachis appliedmainly to suffixes,but whennecessary,partsof stems
can also be transcribedmorphonemically.The morphophonemes, as distinct
from surfacelevel phonemes,are written in capital letters; for example,the
generalizedmorphonemicshapeof the above mentionedrenarrativemood
suffix is written HAMHU.
It appearssufficient to distinguish, instead of the twenty-one surface-
level consonantphonemesand 'subphonemes'(Table 16.1), 18 deep-level
consonantmorphonemes:H T T J K ? S SJ B J DJ M N N J 1] L U R c.
Perhapsat a still deeperlevel of description five of these (Tj ? sj Di c)
could be eliminated - but at the expenseof making the description more
artificial and complicated.Most of these consonantmorphonemes,if they
are not affected by any regular and obligatory rules applied to
490 NGANASAN

Table 16.6 Vowel harmony:phonemicrealizationsof vowel morphonemesin

suffixes(prior to accommodations)

Morphoneme A Al Ao U Uo D Do I ;) UA iA

After U -stems Precedingvowel is a a a u u ii ii ii ii ii iiii ii 1 i d u a ia

(first class) not i or ii
Precedingvowel is aaa a ii u l ii ii ii ii 1 i d u a ia
iorii
After I-stems Precedingvowel is ia 1 a 1 u i ii 1 i d u a ia
(secondclass) not i or ii
Precedingvowel is ia i a i ul i ii 1 i d u a ia
iorii

Note: 1 In TaimyraAvamsubdialectu(at leastin certainsuffixes with Uo)'

individual units and their strings, are rewritten in the correspondingnon-

capital letters. The morphonemeC ('zero-consonant')correspondsto a
surface-levelzero: it is used to denote a conventionalunit which occurs
syllable-finally after vowels and imparts to the (phonetically) open syllable
the phonologicalpropertiesof a closed one; it also participatesin certain
morphophonologicalrules. Like other morphonemes,C is a phonological
reality from the viewpoint of proto-Nganasan(proto-Samoyedic)reconstruc-
tion. The most important transformationsand surfacelevel realizationsof
consonantmorphonemesareindicatedin Table 16.5.
The treatmentof vowels in terms of morphonemesis connectedalmost
exclusively with the phenomenonof vowel harmony, see pp. 493-40 and
Table 16.6.

Rhythmic Gradation (RhG)

The gradealternation,or gradation,of word-medialconsonantsand clusters
consists in the alternative appearanceof two grades, strong and weak,
dependingon the (morpho)phonologicalenvironment.
In general,the stronggradeof RhG surfacesafter an odd numberof vocalic
morae(= phonologicalvowels) in the precedingpart of a word, and its weak
grade surfacesafter an even number of vocalic morae. RhG affects five
morphonemes(H T K S SI) and clustersof thesewith precedingnasals(MH
NT .oK NS NISI), seeTable 16.5. RhG manifestsitself very clearly in the
allomorphy of suffixes, cf.ni~ti~bi"ni~Ot', hi(uF ;rti", k;'}ri"g;'}[ii-Ot' 'his wife, rope,
thumb, march' (all with the s3 possessivesuffix); ni~r;'}gi~ brni~r;'}ki~
hi(uf;r
r;'}gi", k;'}ri"g;'}[ii-r;'}ki" 'similar to a woman, rope, thumb, march' (all with the
samesimilative suffix) etc., cf. also Table 16.4. Historically, the functioning
of RhG is attestedalso in roots, cf. hiit;'}8;'} 'trunk' from *put;'}1;'} (= Nenets
pudiid, Enetspu&J&J).
The functioning of RhG is restrictedby severaladditionalrules:
 NGANASAN 491

1 Only the stronggradesurfacesafter consonants(including c), unlessthe

resulting clusteritself participatesin RhG: t;]r-tu 'hislher hair', ka8ar-tu
'hislherlight';
2 Only the weakgradecomesout directly after vocalic sequences:lat;];}-ou
'hislherbone', biri;}-Qr 'hislherwound';
3 Only the weakgradeof the clusterNT occursbeforean opennon-second
syllable: hiit;]o;}-t;]11U 'in the trunk' (with the locative suffix NTgNU).
Closed non-secondsyllables preservethe strong grade of this cluster,
which then undergoessyllabic gradation: hiit;]o;}-ndinii 'in the trunks'
(with NTIcNU).

The number of vocalic morae is normally countedon the phonological

level and may differ from the number of phonetic vowels and phonetic
syllables.So for examplethe stemshUaa 'tree' or biai 'target',both phonetic
monosyllables,are treatedfrom the viewpoint of RhG as having two morae:
hUaa-r~u,biai-r~i~ 'similar to a tree, to a target'. However, it is typical,
perhaps even normal, for a word-medial (and non-first-syllable) vocalic
sequenceto be counted as only one vocalic mora: lat;];}-r~u,
biri;}-r;]ki"
'similar to a bone,wound'.

Syllabic Gradation (SyG)

In general, the strong grade of SyG surfacesbefore the vowel of an open
syllable, and its weak grade surfacesbefore the vowel of a closed syllable.
SyG affectsthe samemorphonemesandclustersas RhG, but the weakgrades
of clustersresulting from SyG differs from those resulting from RhG; see
Table 16.5. In terms of rule-orderingSyG follows RhG and is applied only
to thoseunits that arepreservedin their stronggradeby RhG (the weakgrades
of RhG, e.g. 0 from T, or cT from NT, do not participatein SyG).
SyG functions in both nominal and verbal paradigms,cf. (sN : pN) kuhu
: kubu? 'skin, hide', basa : batJia? 'iron', k;]11t;] : k;]11d;]? 'sledge',kaoar :
katar;]? 'light', henitJiir : hensi,.;]? 'shaman'sdrum'; (verbal adverb :
connegative)kotutJia : kolJu? 'kills', tJiembi?sii : tJieljhitJi;]? 'getsdressed'.For
the functioning of SyG in suffixes (combinedwith that ofRhG) seeTable 16.4
andFigure 16.1.
Like RhG, SyG does not affect postconsonantalconsonants,unless the
clusterparticipatesin SyG itself, cf. lj;»nsu 'meat',pN lj;»nsu?'
If the closedcharacterof a syllable is determinedby a deeplevel c, SyG
may surfaceas an 'unconditioned'alternation:k;]11t;] 'sledge',sG k;]11d;] (with
the sG casesuffix _c). It must also be noted that an intervocalic glottal stop
alwaysmakesthe precedingsyllableclosed.
The morphemesthat endin C ('pseudo-vocalic'or 'quasi-consonantal') are
twice abnormalfrom the viewpoint of SyG. Not only does the weak grade
surfacein phoneticallyopensyllables,but alsothe stronggradeis usedin final
syllables followed by a glottal stop or a second(suffixal) c, as well as in
492 NGANASAN

certain types of word-medial phonetically closed syllables: koOu 'snow-

storm', sG kotu, pN kotu?; diagusa 'being lost', diaku? 'gets lost (con-
negative)" diaku?;mz 'I got lost' (stems: KOTUc , DJUKUc-). Probablythis
paradox is to be accountedfor by a special rule that changesthe zero-
consonantC into the zero-vowelv in the relevantpositions (C?# > v?#, CC#
> vC#, etc.)
Nunation
This morphophonologicalmechanism,known as 'Nunation' since M.A.
Castren(1854: 162), consistsin the changeof C into a nasalconsonantwhen
C is followed by H, T, TJ, K, S, or SJ and is precededby a nasal, being
separatedfrom that precedingnasalonly by a vowel or a vowel sequence;see
Table 16.5. In terms of rule-orderingit follows both RhG and SyG. Thus, on
the one hand, some deep level clusters are, after the application of RhG,
'restored'by nunation, and, on the other hand, neither type of gradationis
applied to the output of nunation. Therefore there are such forms as e.g.
mununtum'I say', in which the deeplevel clusterNT appearsto be preserved
in its stronggradecontraryto the conditionsof both RhG andSyG. The actual
paththat generatessuchforms looks as follows:

munu-NTU-M> munu-cTU-M> mununtum

(RhG) (Nun)

The functioning of nunationis evidentin pairs such as: kunduasu;J'[s]he

slept' : tanUansu;J'[s]he was accustomed'(stemskunduac-, tanUac-), niaaOu
'hislherNganasan': niantu 'hislhermate' (stemsniaa and niaC ; the suffix is
-TU); basu-tu'hunts' : ni"a-ntu 'savesup' (the suffix is -NTU, transformedby
RhG into cTU).
The rule of nunationis less obligatory (especiallyin derivation)than other
morphophonologicalrules, and sometimesnunatedand non-nunatedforms
coexistas variants:k;mz;Jut;]- - k;mz;Junt;]- 'to wrestle',l;Jl]inti" - 1;J1]i"ti" 'bums
[intr]' .
VowelHarmony(VH)
Fromthe viewpoint ofVH all mono-andbisyllabic stemsaredivided into two
harmonic classesof approximatelythe samesize. Historically, this division
correspondsto the opposition 'front: back', and wasdeterminedby the last
vowel or (if this vowel was harmonicallyneutral) by the first vowel. Today,
however,this original principle is so overshadowedby secondarychangesof
vowel quality, that it is more convenientsimply to ascribeto each stem its
harmonicclass,first ('U-stems')or second('I-stems'),as a lexical morpho-
phonologicalcharacteristic.The synchronicphoneticcompositionof stemsis
in many casesnot sufficient to determinetheir harmonicclass: e.g. hon- 'to
plait' is aU-stem(honsu;JOu '[s]he plaited it'), while hon- 'to have' is an
I-stem (honsi";J8i" '[s]he had it'); cf. also diint;J 'archer'sbow', and k;mz;]- 'to
 NGANASAN 493

catch' (V-stems)v. kint;;, 'smoke'and k;;,r;r 'to moor' (I-stems).

Another way of dealing with VH would be to distinguishtwice (or nearly
twice) as many vowel morphonemesas there are vowel phonemes.In this
presentation,however,the descriptionin morphonemictermswill be applied
only to the vocalismof suffixes, in order to accountfor the impact of VH on
it (Table 16.6). The impacton suffixal vowels is also not systematicfrom the
synchronicpoint of view. For example,the vowel u is in harmonicalternation
with rin somesuffixes,but remainsstablein others:basu-gu-mu?'let us hunt'
and niili"-gu-mi? 'let us live'. This means that u representstwo different
morphonemes,Vo in -gu- andV in -mu?l-mi?(andhistorically, that u results
from a mergerof two different vowels).
The situation is further complicatedby a kind of 'secondary'VH, which
is due to the fronting influenceof i and if on high vowels in the next syllable,
so that each harmonic class becomesfurther divided into two subclasses
(Table 16.6).
The assertionthat 'thereis no vowel harmonyin Nganasan'(Terescenko
1979: 48) is true in the sensethat what is called VH here lacks synchronic
motivation and does not create a real 'harmonic unity' of words. Besides,
within bisyllabic primary stemsand old derivativesthe principlesof VH can
be violated.
Like othermorphophonologicalnorms,the principlesofVH aresometimes
disregarded,usually by younger and less capablenative speakers.In such
cases the prevailing direction of neutralization is in favour of the first
harmonicclass(V-stems).

StemAlternation
While the alternationsdescribedabovehavephoneticmotivation (synchronic
or historical), the choice betweenstem variants can be determinedby the
morphological environment, as well. It is convenientto distinguish three
variantsof nominal and verbal stems:

SI, which occurs,in particular, in the nominative singular of nouns and in

verbal adverbs,andcanbe consideredthe basicstemform;
S2, e.g. in the genitive singular and nominativeplural of nouns, and in the
connegativeform of verbs;
S3, e.g. in the genitive plural of nouns,and in the presenttenseof perfective
verbs.

Sometypesof relationshipbetweenstemvariantsare shownon p. 495. Nouns

are cited in sN for SI, in sG for S2, in pG for S3. Verbs are cited in the forms
of verbaladverbfor S1, in connegativeforms for S2, in the forms of indicative
presents3 (suffix ?;J) or of perfectiveverbalnoun (suffix ?MU;J) for S3.
494 NGANASAN

1 Vocalic basicstems:

IJUsi' IJUsi' lJusi'? lJusi'? 'work' nansu<:lii nansu? nansu?a 'standsup'

lJuta lJulia lJuba? 'berry' lJonsi'dii lJoncPi? lJonctii?a 'goesout'
koli' koli' kola? 'fish' koni'ctii koni'? kona?a 'goes'
maku magu maga? 'back' kontudia kondu? konda?a 'carries'
hiiii hiiii hia? 'tinder' biiiictia bii? bia?mua 'departs'
ctiab ctiaga diagii? 'twin' hotactia holia? holiii?a 'writes'
lataa lataa latai? 'bone' hursactii hursa? hursii?a 'returns'

2 Consonantalbasicstems:

bi'? bi'lia bi'lii'? 'water' nasa nalia? naliu?a 'scours'

hia? hiactia hiadii? 'fur overcoat'diembi?sii dielJhictia? ctielJhictii?a 'dresses'
hiactiir hiasi'ra hiasi'ri'? 'fishing-rod' bi'lii'rsi' bi'ti'ra? bi'ti'ri?mi'a 'drinks'
tuj tuu tuu? 'fire' tuGlsia tu? tuu?a 'comes'
lJoj lJua lJua? 'foot' me(j)sii mi'? mi'i'?a 'makes'

3 Pseudo-vocalicbasicstems:

koliu kotu kotu? 'snowstorm' ctiiigusa ctiiiku? ctiiiku?a 'get lost'

It is also possible to describe stem alternationsas resulting from the

applicationof specialoperators(vowel addition, vowel replacement,etc.) to
a basic deep level stem form. For certain stemstwo alternants,or even all
threeof them,may syncretize.
Stem-medialconsonantstypically occurin stronggradesof SyG in vocalic
SI, and in its weak gradesin S2 and in consonantalS1. In keepingwith this
general tendency, the secondary symbol SI' is used to denote those
morphologicalpositionsin which the oppositedistribution obtains,i.e. where
weak gradesof SyG are used in vocalic SI, and similarly the symbol S2'
denotespositionswhereS2 hasstronggradesof SyG.
Within suffixal strings many suffixes (for example, mood and tense
markers) undergo alternations which are essentially the same as stem
alternations.

Other Morphophonological Rules

The rule of truncation operatesevenbeforeRhG and SyG. Truncationworks
as follows: 1) C > 0 / _CC; 2) C > 0 / C_c. That is, (1) a suffix that
begins in a consonantcluster ousts a final C in the precedingmorpheme
or (2) losesits first consonant,if the precedingmorphemeendsin a non-zero
consonant.
Table 16.5 reportsthe word- andsyllable-finalchangeof the morphonemes
T, K (which doesnot occurword-finally), and S into aglottal stop, which can
then be reducedphoneticallyto zero (this reductionis almostregularwithin
rhythmic groups, optional at boundariesbetweenthesegroups, and rare in
word-final position). This results in alternationssuch as mal 'house', sG
mab;;J, sEla mak;;Jt;;J (stem maT); Ipmsa?su;;J '(s)he ate', I);;Jmsadi;;J? 'eat! (s2
 NGANASAN 495

imp)' (stem l):nnsaS-).At a deeperlevel of descriptionit is also possibleto

treatevery word- and syllable-finalglottal stop asT (or S), evenwhenit does
not participatein suchalternations.
Numerous accommodative processes follow, in terms of rule-ordering,
the above-describedmechanisms,and are usually applied to the phonemic
realizationsof morphonemes(seeTables 16.5 and 16.6). They operateboth
within morphemesand acrossmorphemicboundaries(internal sandhi).The
most important rules of accommodationare listed below. Symbols: C =
consonant,CJ = palatal consonant,[a b] = a and b correspondingly,{a
b} = a or b, (a) = a may be present or absent, -a = not a, (» = the
rule is optional or restricted to certain morphological positions. The rules
are arrangedso that those under (B) operate after those under (A) and
before those under (C), while the rules under (D) appear to be neutral
to rule ordering.

(A) (C)
t(»0/r_ {?lnIJ} >01 _{llj}
[u i] (» [ti i] 1 (ti i}(C(C))_ j(»01 _C J
[u 1] (» [ti i] 1 _ (C(C)){ti i) [n 1] > [ni lj] 1_C J
[e 0] > ['iu] 1_;)
(D)
(B)
[m n ni IJ] > 01_[m n nj IJ]
r> 11 (-i -ti}C_ {mnnj IJ}>0/{mIJ}
r>lj/{iti}C_ {nni} >IJI _{kh}
i (» ti 1_(C(C))ti m(» IJ I_h
i (» ti 1 ti(C(C))_ ;) > a1 {a U a ia}(?)_
[t s n1] > [V sj ni lj] 1j_ a> ua/h_
[s 1] > [sj lj] 1_ {i ti ia}
['iu ia] > [i ti a] ICJ_

Morphology

Nominal Categoriesand Nominal Inflection

The grammaticalcategoriesof nounsin Nganasanare as follows:

Number: singular,dual, andplural are distinguished.Plural forms occurmore

frequentlythanin manyotherUralic languages, often renderingthe general
idea of abundance(kam;;,? 'bloods = a lot of blood') or collectivity
(niemiW;;, 'my mothers= my mother and other female relatives'). Plural
forms also may combinewith numerals:nakiir;;,? basutu;;,?'threehunters,
as many as threehunters'.
496 NGANASAN

Case: nominative(= absoluteform), genitive, accusative,lative (= dative or

dativellative), locative (= locative/instructive),elative (= ablative), prola-
tive (= prosecutive),and comitative (= sociative)are distinguished.There
are also severalforms that standon the borderlinebetweencaseforms and
postpositionalconstructions(the allative with the marker -cJiaa, joined to
the genitivessingularandplural), betweencaseforms andnon-finite verbal
constructions(the essive/translativewith the marker -isia, joined to the
nominativesingular),and betweencaseforms anddenominaladverbs(the
caritive [= abessive],with the suffix -KAJ or -KA eLJI, joined to stem
variantSl).
Possessivity:non-possessive (= absolutive)forms and possessiveforms with
the subcategories of possessor's number(singular,dual, plural) andperson
(first, second,third) are distinguished.The possessiveforms of the second
personsingularare often usedto specify nounsas definite and co-referent
with their predecessors in the discourse(ratherthan implying collocutor's
possession),and occasionallydefinitenessis expressedalso by possessive
forms of the third personsingular.
Predicativity: with the subcategoriesof subject's number (singular, dual,
plural) and person (first, second, third). Predicativeforms are used as
predicateswithout copulaeor, if anothergrammaticalmeaningthan that of
indicativepresentmustbe expressed,with correspondingforms of the verb
ij- 'to be': niaam
m;m~ niaam 'I am Nganasan',niaam m;m~ niaam isiu;;»n 'I was
Nganasan'.

Tables 16.7 and 16.8 show the morphonemiccompositionof suffixes for

the above-listedgrammaticalcategories,the structure of the paradigmatic
forms, and an exampleof nominal declension.There are also specialshorter

Table 16.7 Nominal personalendings

Predicative Possessive
Pxi Px3 Px4 Px2
sN sA dpNGA Other cases,adverbials

si M Md Md Nd Nd Nd
s2 J] Rd MTd Td NTd
s3 0 TV MTU TU NTU
di MI C MI C MI C NI C NI C
d2 RIC RIC MTI c TIc NTI e
d3 KdJ TIc MTI c TIc NTI c
pi MU? MU? MU? MU? NU? NU?
p2 RU? RU? MTU? TU? NTU?
p3 ? TUJ] MTUJ] TUJ] NTUJ]
 NGANASAN 497

forms of local casesuffixes which are usedmainly in adverbials(local and

temporaladverbs,postpositions,adverbialpronouns):lative _c, locative -NU,
elative-T~.
The combinability of categoricalmeaningsis partly restricted.The dual
paradigmlacks local cases,thesemeaningsbeing renderedby constructions
with the dual genitive and correspondingcaseforms of the postpositionna-:
kuhugi na 'to two skins', kuhugi nanu 'in/with two skins',etc.). Thereare no
possessiveforms of the comitativecase.The predicativeforms haveno case
or possessiveinflection. Further restrictions are found primarily in special
categoriesof nominal words (pronouns,adverbials).
Some descriptions of Nganasanalso reckon with a category of (pre)-
destinativeness: kuhuO;;mt;] 'skin for me', etc. (cf. Terescenko1979:102-7).It
seems,however,moreappropriateto treatthe so-called(pre)destinativeforms
as possessiveforms of substantivesthat have beenbuilt with the clitic -T~,-K~~
which shifts the temporalreferenceto the future: kuhuo;] 'what is going to be
_DJ~~
a skin, future skin'; cf. the clitic with the oppositetemporalreference:
kuhutP;);} 'what usedto be a skin, former skin'.

Adjectives
The semanticdifference betweensubstantivesand adjectivesis supported
derivationally (most adjectivescontain specialadjectival suffixes, e.g. -K~~-K~~
and -ceIcKUo in qualitative, and or -~ _~DJ~ in relative words), but from
syntactic and inflectional viewpoints they are identical. Like adjectives,
substantivescanbe usedattributively andthenhavea defectivecaseparadigm
with only three cases(see Syntax, p. 511), while in non-attributiveposition
adjectivesundergocompletesubstantivizationand have all nominal inflec-
tional forms.
A small group of qualitative adjectiveshas formally distinct attributive-
positive, attributive-comparativeand predicative forms: tan;]g;];] 'wide' :
tandutF;] 'wider (attrib)' and tandUa 'is wide', b;]Piiku 'short' : bimtF;]
'shorter(attrib)' : bim 'is short'. The singular prolative and plural genitive
case forms of qualitative adjectives are used as qualitative adverbs, e.g.
tant;]g;];;mt;Jnu, tant;]g;]i? 'widely'.

Numerals
Primary cardinal numeralsand correspondingordinal numeralsand numeral
adverbs(answeringthe question'how many times?')are listed in Table 16.9
(seep. 500). Othernumeralsare compounds,formed without any connecting
elements additively (bii? siiti '12'), multiplicatively (siiti bii? '20'), or
multiplicatively andadditively (siiti bii? siiti '22').

Pronouns
Personalpronounsand relatedpronominalforms (Table 16.10, p. 500) have
many formal peculiarities. Basic personal pronouns actually have no
Table 16.8 Nominal declension: deepstructure of forms and the paradigm of kuhu 'skin, hide'

Non-poss Poss s1 s2 s3 d1 d2 d3 p1 p2 p3

sN Sl kuhu Sl-Pxl kuhum;) kuhur;) kuhuou kuhumi kuhuri kuhuoi kuhumu? kuhuru? kuhuouI)
sG S2-"or Sl-Px2, kuhun;) kubut;) kubutu kuhuni kubuti kuhunu? kubutu? kubutuI)
S2-0' Sl'-Px22
kubu(I))
sA S2-"or SI-Px3, kuhum;) kubumt;) kubumtu kuhumi kubumti kuhumu? kubumtu? kubumtuI)
S2-M' S2-Px33
kubu(m)
sLat Sl'-NT;;1' Sl '- kubut;)n;) kubut;)t;) kubut;)tu kubut;)ni kubut;)ndi kubut;)nu? kubut;)ndu? kubut;)nduI)
kubut;) NT;J-Px2
sLac Sl'- Sl'- kubut;)nun;) kubut;)nunt;) kubut;)nuntu kubut;)nuni kubut;)nunti kubut;)nunu? kubut;)nuntu? kubut;)nuntuI)
NT;}NU NT;}NU-
kubut;)nu Px2
sEla Sl-K;}7;} Sl- kuhug;)t;)n;) kuhug;)t;)t;) kuhug;)t;)tu kuhug;)t;)ni kuhug;)t;)ti kuhug;)t;)nu? kuhug;)t;)tu? kuhug;)t;)tUI)
kuhug;)t;) K;fT'd-
Px2
sProl Sl-M;}NU Sl- kuhum;)nun;) kuhum;)nunt;) kuhum;)nuntu kuhum;)nuni kuhum;)nunti kuhum;)nunu? kuhum;)nuntu? kuhum;)nuntuI)
kuhum;)nu M;}NU-
Px2
dN Sl-K;}l Sl-K;}[- kuhug;)ini;) kuhug;)iti ;) kuhug;)inii kuhug;)inii kuhug;)itji kuhug;)iniii? kuhug;)inii kuhug;)ititiI)
kuhug;)j l-Px4
dG Sl-K[C Sl-K;}[- kuhug;)in;) kuhug;)it;) kuhug;)itu kuhug;)ini kuhug;)iti kuhug;)inu? kuhug;)itu? kuhug;)ituI)
kuhugi 7-Px4
dA =dG =dN
pN S2-7 kubu7 S3-J-Px4 kubania kubatia kubatiii kubanii kubatii kubaniii7 kubatiii7 kubatiiilJ
pC S3-7 kuba7 S3-7-Px4 kubana kubata kubatu kubani kubati kubanu7 kubatu7 kubatulJ
=pN
pA S2-1kubuj
pLat SI'-NTfl4 SI'-NTr kubutina kubutita kubutitii kubutini kubutindi kubutinii7 kubutindii7 kubutindiilJ
kubuti7 -Px2
pLac SI '- SI '- kubutiniina kubutiniinta kubutiniintii kubutiniini kubutiniinti kubutiniinii7 kubutiniintii7 kubutiniintiilJ
NTrNU NTfCNU
kubutinii -Px2
pEla SI-KFT;J' SI-KlcTJ kuhugitina kuhugitita kuhugititii kuhugitini kuhugititi kuhugitinii7 kuhugititii7 kuhugititiilJ
kuhugita _Px26
pProl S3-7-M;JNU S3-7 kuba7manunakuba7manuntakuba7manuntukuba7manuni kuba7manunti kuba7manunu7kuba7manuntu7kuba7manuntulJ
kuba7manu -M;}NU
-Px2
sCorn S2-NA
kubuna
dCarn SI-KF-NA
kuhugina
PCarn S3-7-NA
kuba7na

Note: For the explanationsof symbolsusedin presentingdeepstructuresofforms (stemsSI, SI', S2, S3; morphonemes)see 'Morphophonology'pp. 487-495.Px1-4 are different seriesof possessive
suffixes, seeTable 16.7.
I The secondvariantis archaicand occursmainly in the languageof folklore.
2 S I-Px2 in possessiveforms of first person,S I'-Px2 in possessivefonns of secondand third persons.
3 SI-Px3 in possessiveforms of first person,S2-Px3in possessiveforms of secondandthird persons.Cf. bigaj 'river': sA + Pxsl bigajm;}, sA + Pxs3bikaamtu.
4 Thereis an optional variantS3-NTI?for certainstems:bigajl'i? or bikauti? 'to the rivers'.
5 Variants with lessfrequentoptional forms of casesuffix: S I-KlcTI?, SI-K;}cTI?(kuhugiti?,kuhugati?).
6 The deepstructurecan also be interpretedas S I-KI'Tl'"-Px2. Its optional variantis SI-KifT/-Px2 (or S I-KifTl'"-Px2).
Predicativeforms are as follows:
sl kubum dl kuhumi pI kuhumu?
s2 kubulJ d2 kuhuri p2 kuhuru?
s3 kuhu d3 kuhug;>i p3 kubu?
500 NGANASAN

inflection: their nominatives,genitives, and accusativesformally coincide,

and the roles of their local casesare playedby the correspondingpossessive
caseforms of the postpositionna- (usedalso to supply surrogatecaseforms
in the dual of nominalinflection).

Table 16.9 Numerals

Number Cardinal numeral Ordinal numeral NumeraladverbI

1 IJu?gi?- IJu?gj niergolti"g, niergbtgg IJu?gou?

2 siiti siioimti( g) siioi?
3 nagiir naggmtu(g) nakiirii?
4 tietg tietgmti"(g) tieti?
5 sgIJhgliaIJkg sgmb;}mti"(g ) sgIJhgliaIJgi?
6 mgtii? mgtgmt1(g) m;}tii&ii?
7 siajb;} siajb;}mti"(;} ) siajbi?
8 siitiogtg sJitio;}tgmt1(g) siitiogti?
9 IJamiatiiim;} IJamiati iim;}mt1(g) IJamiatiiimi?
10 bii? biimti(g) biioi?
100 diir diir;}mti"( g ) diiri?

Note: to Identicalto the pG of cardinalnumerals.

When combined with clitics (and such combinations are used very
frequently), basic personalpronounsare replacedwith bound pronominal
stems:MIN for sdp1,TIN for sdp2, and SI? for sdp3.Thesestemsare then
followed by a clitic (in Table 16.10 exemplifiedby RA]AO 'only') and Px2
(cf. Table 16.7).
The indeclinablepersonalemphaticpronouns('myself etc.) containthe
bound stem yon;r and Px2. Both structurally and syntactically they are
similar to personalforms of nominal adverbials.
In other categories,nominal pronounsare usually inflected like ordinary
nouns (and distinguish all inflectional categories).These categories also
include pronounsthat are formally and functionally similar to adjectives,
numerals,andadverbs.
Demonstrativepronounsdistinguishdeixis of proximateobject (;]I11ti ;]I11k;}t;}
~ ;]111-

'this', ;]I11;}rii;} 'this one', ;]111;} 'here [lative]" ;]I11ni" 'here [locative]" ;]I11k;}t;}
'from here', ;]I11ZaJii 'like this, such', ;]I11i?f'a 'like this, so', etc.), deixis of
distant object (tak;};} 'that', taanii;} 'that one', tab;} - taba?a 'there [lative]"
tamnu 'there [locative]" tabk;}t;} 'from there',etc.), and anaphoric(t;}ti 'this/
that, it', t;mii;} 'this/thatone', t;md;} and t;mii 'there [lative]', t;mi" and t;miini
'there[locative]" t;}g;}t;} and t;miio;} 'from there',t;}raJii 'such',t;mii?iia 'so',
etc.).
Interrogativepronounsinclude maa and maayuna 'what', siti" and si1i"yuna
 NGANASAN 501

Table 16.10 Personalpronouns and related pronominal forms

N,G,A Lative Locative Elative Prolative Personal Personal

pronouns+ emphatic
cUtic pronouns
('only')

sI m;}n;} nan;} nanun;} nag;}t;}n;} nam;}nun;} rriiljian;} l)on;}n;}

s2 t;}n;} nant;} nanunt;} nag;}t;}t;} nam;}nunt;} tlljiat;} l)on;}nt;}
s3 slti nantu nanuntu nag;}t;}tu nam;}nuntu siljiatl l)on;}ntu
dI mi nani nanuni nag;}t;}ni nam;}nuni mlljiani l)on;}ni
d2 ti nandi nanunti nag;}t;}ndi nam;}nundi tnjiati l)on;}nti
d3 si'ti nandi nanunti nag;}t;}ndi nam;}nundi slljiati l)on;}nti
pI mil) nanu? nanunu? nag;}t;}nu? nam;}nunu? miljianl? l)on;}nu?
p2 til)
p3 nandu? nanuntu? nag;}t;}ndu? nam;}nundu? tlljiati? l)on;}ntu?
p3 Sltll) nandul) nanuntul) nag;}t;}ndul) nam;}nundul) siljiatll) l)on;}ntul)

'who', ku;;, 'what (attrib)', kan;;, 'how many', kan;nntu(;;,) 'which (in order),
das wievielte', kanu? 'how many times', kunii;;, 'what kind of, kund;;, and
kunii 'where [lative]" kunu and kuniini 'where [locative]" kuniio;;, 'from
where',kur;xlii 'what, what kind of, kuniiliia 'how', kaOg;;, 'when',maatfaa
'why', etc.
Negative pronouns (or rather emphatic interrogative pronouns,used in
negativesentences)and indefinite pronounsare correlativewith the interrog-
ative onesand incorporateclitics: maag;;'Piti;;, 'nothing, anything',kaOk;;,g;;,l-
iiti;;, 'never, whenever';maagu;;" maati;;" maati;;,ku;;, 'something',kaok;;,gu;;"
kaOg;;,ti;;" kaOg;;,ti ;;,ku;;, 'somewhere',etc.
Determinativepronounsinclude b;;Jfls;;, 'whole, all' (dN b;;Jfls;;,g;;,j 'both',
pN b;;Jfltf;;,? 'all'), oamiaj 'another',oamiatfum 'the other', huns;;,;;, 'other',
malams;;, 'whatever,different', etc.

Adverbs and Syntactic Words (Adverbials)

The vast and heterogeneous classof adverbialscomprises,from the syntactic
viewpoint, adverbsproper (with the semanticsubgroupsof local, temporal,
qualitative,modal,gradationaladverbs,etc.),postpositions,andconjunctions.
From a morphologicalviewpoint it is possibleto single out the subclassof
inflected nominal adverbials,which have personaland, to an extent, case
forms as well. This subclassincludesnot only local andtemporaladverbsand
postpositions(cf. OiP;;Jfli" 'below', Oili ;;,0;;' 'from below', hUaa OiP;;Jfli" 'under
the tree', hUaa OiP ;;,0;;, 'from under the tree', OiP ;;Jfli"n;;, 'underme', OiP ;;Jfli"nti"
'underhimlher', etc.), but also such words as brut:1- 'simply, just so', On:1-
'still', oon;;,;;,o;;, 'once again', etc.: On;;Jfl;;' '1 still', OiI;;,ti" '(s)he still'. The
morphological subclassof indeclinable adverbials is the more numerous,
however.
502 NGANASAN

Verbs

ConjugationType, Person,andNumber
There are five conjugation types (voices) in Nganasanwith five partly
different setsof finite personalendings:(1) subjective,(2-3-4) objectivewith
singular,dual, andplural object, and (5) objectless(or reflexive). Transitive
verbs are conjugatedin all five types: accordingto three objective types (if
their object in the correspondingnumberis definite), to subjectivetype (if
their object is indefinite and in someother situations),and to objectlesstype
(if they renderreflexive meaning).Intransitive verbs are divided into those
conjugatedaccordingto objectlessor, alternatively,subjectivetypes (reflex-
ive, reciprocal, passive, many inchoative and finitive verbs) and those
conjugatedonly accordingto subjectivetype.
Verbal personalendings(Table 16.11) distinguishthreepersonsand three
numbersof the subject.In the subjectivetype theyare similar to the nominal
predicativeparadigm,andin the objectivetypesthey resemblethe possessive
endings of nouns. The personalendings used in the imperative are partly
differentfrom theendingsin othermoods.Mood andtensesuffixesthatdirectly
precedethe endings of the objective-plural and objectlesstypes undergo
changessimilarto thosefoundin S3 ascomparedwith S 1(seepp. 494-5).
Table 16.11 Verbal personal endings (Vx)

Subjective Objectiveconjugation Objectless

conjugation with singular with dual with plural conjugation
object object object

sl M M;:) K;:)I-J-N;:) J-N;:) N;:)

s2 l) R;:) K;:)I-J-T;:) J-T;:) l)
s3 0 TU K;:)I-J-TU J-TU ? orT;:)C
dl MI C MIC K;:)I-J-NIC J-NIC NI C
d2 RIC RIC K;:)I-J-TIc J-TIc NTI e
d3 K;:)J TIC K;:)I-J-TIc J-TIc NTI e
pI MV? MV? K;:)I-J-NV? J-NV? NV?
p2 RV? RV? K;:)I-J-TV? J-TV? NTV?
p3 ? TUl) K;:)I-J-TUl) J-TUl) NT;:)?

Imp pres sl T;:)M M;:) K;:)I-J-N;:) J-N;:) N;:)

Imppress2 ?1 T;:)4 K;:)I-J-N;:)2 J-N;:)3 TIl)'
Impfut s2 K;:)l) orK;:)2 K;:)cT;:)2 K;:)I-J-T;:) J-T;:) KIl)2

Note:
I Joineddirectly to the secondstemof the verb - S2.
2 Joineddirectly to the first stemof the verb -S 1.
3 Joineddirectly to the third stemof the verb - S3.
4 Joineddirectly to the verbal stem:usually to sl, but in someverbswith consonantalSI- to
S20rS3.
In otherforms Vx arejoined after mood and tensesuffixes: seeTable 16.12.
 NGANASAN 503

Aspect
The aspectual distinction between perfective and imperfective verbs is
displayedformally in the indicative present,whereperfectiveverbstake only
the suffixes of the presentperfect (and renderthe correspondingmeaning),
andwhereimperfectiveverbstakeonly the suffixes of the presentcontinuous.
In otherverbal forms the [+1- perfective] distinction remainson the semantic
level (this is easilyrenderedin Russian,but not in Englishtranslations).From
many verbal roots both perfective and imperfective stems are derived (cf.
from'catches':
k;m~ 'catches':k;mi?;m;J 'I have caught[pres perf]' and k;mii&itiim
'I am catching[pres cont]'. Thereare also rareexamplesof biaspectualstems
(hUaOku- : hUaoka?am'I got drunk' and hUaokutum'I am drunk').
Mood and Tense
The systemof moods includesindicative, imperative,interrogative(usedin
interrogativesentencesand in relative clauses),inferential (or latentive,used
for reporting eventswhich the speakerdid not witness directly), renarrative
(mood of reported speech),irrealis (used to denote imaginary results of
unfulfilled conditions), optative (used to denotedesirable,but hypothetical
events), admissive-cohortive(with several functions, including those of
expressinggnomic constatation[with neutral intonation], doubt [with inter-
rogative intonation], and polite recommendation[with exclamatoryintona-
tion]), debitive (or obligative, used to denote a future action which is
considered proper), abessive (not-yet-accomplishedaction), prohibitive
(warning againstan action); this list is not exhaustive.
Tensesaredistinguishedin threemoods:

• in the indicative (present, with the complementarilydistributed sub-

categorical meanings of present perfect and present continuous: see
above;past,pastperfect,future, and future-in-the-past- this last denotes
an action that was supposedto be done);
• in the imperative(presentand future, with the meaningsof, correspond-
ingly, immediateandremotefuture);
• in the interrogative(present,past,future).

Suffixes and selectedforms of verbalinflection aregiven in Table 16.12.

Non-finite Forms
The main non-finite verbal forms are indicated in Table 16.12 (with
translationsandin somecaseswith the constructionsin which they are used);
the termsusedandespeciallythe distinctionsmadebetweenverbal nounsand
verbal adverbsare conventionaland provisional. Like other nominals,most
non-finite verbal forms take possessivesuffixes with reference,in this case,
to the subjectof the verbal action. Verbal nouns also take casesuffixes, and
participlestake numberandcasesuffixes.
The connegativeform is usedin analytic constructionswith the negative
504 NGANASAN

Table 16.12 Verbal inflection: deepstructure of forms and selectedfragments

of paradigms (kotu- [S2 kooo-, S3 kofu- ] 'to kill; (in objectlessconjugati~n) conjugati~n) to
ruin oneself',kooot;)- 'to be [in the processof) killing, ruining oneself',n'i-
auxiliary negativeverb)

1 2 3 4 5

Indicativepresentperfect
S3-?;J-Vxand S3-?I-VXl
sl kolia?am kolia?am;) kolia?abini;) kolia?iui;) kolia?in;)
s2 kolia?al) kolia?ar;) kolia?abiti;) kolia?itiu kolia?il)
s3 kolia?a kolia?atu kolia?ak;)iuiu? kolia?itiu kolia?ili;) -
kolia?i?
dl kolia?ami kolia?ami kolia?ak;)iuii kolia?inii kolia?ini
d2 kolia?ari kolia?ari kolia?ak;)itii kolia?inii kolia?indi
d3 kolia?ag;)j kolia?alii kolia?ak;)itii kolia?inii kolia?indi
pi kolia?amu? kolia?amu? kolia?ak;)iuiu? kolia?iuiu? kolia?inu?
p2 kolia?aru? kolia?aru? kolia?ak;)iuiu? kolia?itiu? kolia?indu?
p3 kolia?a? kolia?aliul) kolia?ak;)itiul) kolia?itiu? kolia?ind;)?

Indicativepresentcontinuous
SI~NTU-VxSI~NTU-Vxand SI~NTAI-VXlSI~NTAI-VXl
sl koliut;)ndum koliut;)tum;) koliut;)tug;)ini;) koliut;)ndaui;) koliut;)tan;)
niindYm niint"im;) niint"ig;)iui;) uiindYui;) uiintYn;)
s2 koliut;)ndul) koliut;)tur;) koliut;)tug;)iti ;) koliut;)ndaui;) koliut;)ndal)
uiindYI) uiintYr;) niintYg;)iti ;) uiindYui;) niindYI)
s3 koliut;)tu koliut;)tuliu koliut;)tug;)itiu koliut;)ndatiu koliut;)tali;) -
koliut;)nda?
uiintY uiintYliY uiintYg;)itii uiindYtii uiintYli;) - niind"i?
dl koliut;)tumi koliut;)tumi koliut;)tug;)iuii koliut;)ndauii koliut;)tani
uiintYmi uiintYmi uiint"ig;)iuii uiindYuii uiintYni
d2 koliut;)turi koliut;)turi koliut;)tug;)iuii koliut;)ndati koliut;)ndati
uiintYri uiintYri uiint"ig;)iuii uiuiindYti
indY~i uiindYti
d3 koliut;)tug;)j koliut;)tulii koliut;)tug;)itii koliut;)ndati koliut;)ndati
uiintYg;)j niintYlii uiintYg;)itii uiindYtii uiindYti
pJ koliut;)tumu? koliut;)tumu? koliut;)tug;)iuiu? koliut;)ndauiu? koliut;)tanu?
uiintYrnY? uiintYm"i? uiintYg;)iuii? uiindYnii? uiintYnY?
p3 koliut;)turu? koliut;)turu? koliut;)tug;)iuiu? koliut;)tug;)iuiu? koliut;)ndatu?
uiintYr"i? uiintYrY? uiintYg;)itii? uiindYti? uiindYti?
p3 koliut;)ndu? koliut;)tuliul) koliut;)tug;)iuiu? koliut;)ndatiul) koliut;)ndat;)?
uiind"i? uiintYliYI) koliut;)tuliul)
uiintYg;)i~il) uiindYti? uiindYt;)?

Indicativepast
Sl-SU;J-Vxand SI-s'UU-VXl.2
sl kotudiu;)m kotudiu;)m;) kotudiu;)g;)ini;) kotudiiiiiui;) kotudiUUn;)
niisY;)m uiisY;)m;) uiisY;)g;)iui;) uiisiiiui;) niisiiin;)

Indicativepastperfect
Sl-SU;J-IY:J;J-Vx and Sl-SU:J-IY:JI_VXl
sl kotudiU;)di;);)m kotudiU;)di;);)m;) kotudiu;)di;);)g;)iui;) kotudiu;)di;)iui;) kotudiU;)di;)in;)
uiisY;)di;);)m uiisY;)di;);)m;) uiisY;)di;);)g;)iui;) uiisl;)di;)iui;) uiisY;)di;)in;)
 NGANASAN 505

Table 16.12 (Continued)

2 2 3 4 5

Indicativefuture
Sl '-?SUT;J-Vxand Sl '-?SJOT/-Vxl ,2
sl kOOu?suO;lm koOu?sut;lm;l koOu?sut;lg;lini;l koou?sitioini;l - koou?siiitin;l -
koou?siiiOiini;l koou?siiitiin;l
niisi'O;lm niisi'O;lm::l niisi'O;lbini;l niisiioini;l niisiioin;l

IndicativeJuture-in-the-past
Sl '-?SUT;J-IY;};J-VX and Sl '- ?SUT;}-IY;}I- Vx 1
si kOtUI);l;lliu kOtUI);l;lliu koOU?sut;lcti;l;l- kOtUI);l;lliu kOtUI);l;lliu
cti;l;lm cti;l;lm;l g;lini;l cti;lini;l di;lin;l
niisi'O;lcti;l;lm niisi'O;ldi::l;lm;l niisi'O;ldi::l;lg;lini;l niiSi'O;lcti;lini;l niisi'O;ldi;lin;l

Imperativepresent
sl/dl/pl: SI-KUo-Vx
d2/p23: Sl-{]U-Vx and SI-{]A/-Vx 1
1
s3/d3/p3:Sl-{];J;J-Vx and Sl-{];}I-Vx
si kotuguli;lm kotugum;l kotuguk;lini;l kOtUgUni;l kotugun;l
niikuli;lm niikum;l niikug;lini;l niikuni;l niikun;l
s2 koliu? kotuli;l kOtUg;lini;l koliani;l kotuliil)
nii? niint;l niibini;l niini;l niiliil)
s3 kOtUI);l;l kOtUI);l;lliu kOtUI);l;lg;litiii kOtUI);liVii kOtUI);li? -
kOtUI);lili;l
niil);l;l niil);l;llii' niil);l;lg;liVi niil);liVi niil);li? - niil);lili;l
di kotugumi kotugumi kotuguk;linii kotugunii kotuguni
niikumi niikumi niikug;linii niikunii niikuni
d2 kotul)uri kotul)uri kotul)ubitii kotul)aVi kotul)andi
niil)i'ri niil)i'ri niil)i'g;litii niil)i'Vi niil)i'nti
d3 kotUI);l;lg;lj kOtUI);l;llii kotUI);l;lg;liVi kotUI);litii kotUI);linti
niil);l;lg;lj niil);l;llii niil);l;lg;liVi niil);liVi niil);linti

pi kotugumu? kotugumu? kotuguk;liniii? kotuguniii? kotugunu?

niikumi? niikumi'? niikug;linii? niikunii? niikuni'?
p2 kotul)uru? kotul)uru? kotul)ubiVii? kotul)aVii? kotul)andu?
niil)i'ri'? niil)i'ri'? niil)i'g;litii? niil)i'Vi? niil)i'nti'?
p3 kotul);l;l? kotul);l;lliul) kotUI);l;lg;litiiil) kotUI);litiiil) kotul);lint;l ?
niil);l;l? niil);l;llii'1) niil);l;lg;liVil) niil);liVil) niil)::lint::l?

ImperativeJuture(exceptingS24)
Sl-KUo;}-Vx and Sl-KOoOo-Vxl , 2
si kotugu;lm kotuku;lm;l kotUgU;lg;lini;l kotugiiiini;l kotugiiiin::l
niiku::lm niiku;lm;l niiku;lg;lini;l niikiiiini;l niikiiiin;l
s2 kotUg;l(I)) kotUg::lt;l kotUgU;lg;lit-i;l kotugiiiiti;l kotugil)
niig;l(I)) niig::lt;l niiku;lg;liti;l niikiiiiti;l niigil)

Interrogativepresent
Sl-{]U-Vx and Sl-{]A/-Vx 1
si kotul)um kotUI)Um;l kotul)ubini;l kotUl)ani;l kotul)an;l
niil)i'm niil)i'm;l niil)i'g;lini;l niil)i'ni;l niil)i'n;l
506 NGANASAN

Table 16.12 (Continued)

2 2 3 4 5

Interrogativepast
Sl-HU-Vx and Sl-HAJ-Vx1
sl kotubum kotubum;! kotubuk;!ini ;! kotubani;! kotuban;!
niibim niibini;! niihig;!ini;! niibini;! niihin;!

Interrogativefuture
S l'-NT;J-lJU-Vx and Sl'-NT;J-l]A,-Vx 1
sl kol\ut;!l)um kol\ut;!l)um;! kol\ut;!l)ug;!ini;! kol\ut;!l)ani;! kol\ut;!1) an;!
niint;!l)in;! niint;!l)ini;! niint;!l)i"bini;! niint;!l)ini;! niint;!l)in;!

Inferential
Sl-HATU-Vxand Sl-HATA,-Vx 1
sl kotubal\um kotubatum;! kotubatug;!ini;! kotuba1\ani;! kotubatan;!
niint;!l)in;! niint;!l)ini;! niihi a1\ibini;! niihi a1\ini;! niihi a1\in;!

Renarrative
Sl-HAMHU-Vxand Sl-HAMHA,-Vx l
sl kotubambum kotubal)hum;! kotubal)hug;!ini;! kotubambani;! kotubal)han;!
niint;!l)in;! niint;!l)ini;! niibiahibini;! niibiahini;! niibiahin;!

Irreali5
Sl-HA,AoT;J;J-Vxand Sl-HA,AoT;JI-Vx1• 2
s1 kotubaal\;!;!m kotubaa1\;!;!m;! kotubaal\;!;!g;!ini;! kotubaal\;!ini;! kotubaa1\;!in;!
ni ihia1\;!;!m ni ihia1\;!;!m;! niihial\;!;!g;!ini;! niihia1\;!ini;! niihia1\;!in;!

Optative
Sl-HA,Ao-Vxand Sl-HA/-Vxl. 2
51 kotubaam kotubaam;! kotubaag;!ini;! kotubaini;! kotubain;!
niihiam niihiam;! niihiag;!ini;! niihiini;! niihiin;!

Admi55ivelcohortative
Sl'JK;J;J-Vx and Sl'JK;JI-Vxl
51 kol\ub;!m kol\uk;!;!m;! ko1\ub;!g;!ini;! kol\uk;!ini;! ko1\ubin;!
niil)b;!m niil)k;!;!m;! niil)b;!g;!ini;! niil)bini;! niil)k;!in;!

Debitive
Sl'-BSUT;}-Vx and Sl'-BYUTl-Vx1• 2
51 ko1\upsu1\;!m ko1\upsut;!m;! ko1\upsut;!g;!ini;! ko1\upsiii1\ini;! - ko1\upsjiitin;! -
ko1\upsiii1\iini;! ko1\upsiiitiin;!
nj ipsi1\;!m niipsi1\;!m;! niipsil\;!k;!ini;! niipsjil\in j ;! ni ipsii1\in;!
5
Abes5ive
Sl-M;JTUMA,lAo -Vx and Sl-M;JTUMA,lI-VxL 2
51 kotum;!tuma?am kotum;!tuma- kotum;!tuma?a- kotum;!tuma- kotum;!tuma?in;!
tam;! g;!inj;! ?ini;!

Prohibitive
sl-Ll-Vx
51 kotullm kotulim;! kotullk;!ini;! kotullni;! kotulin;!
niilim niillm;! niiJ'ig;!ini;! niillni;! njilln;!
 NGANASAN 507

Table 16.12 (Continued)

Nonjinitefonns:
Imperfectiveverbal noun
SI-MUN(-Cx) (-Px)
kotumu
sG kotumun;}hir;} 'worth killing'
sLat hiiinsi";Jm kotumund;}'I was afraid to kill'
sEfa + sf kotumugg;n;Jn;}'so that 1 do not kill', etc.
Perfectiveverbal noun
S3-?MUd(-CX)(-Px) or SI'-?MUd(-CX)(-Px)
kooa?mu;}- koou?mu;}
sLoc + sl kooa?mu;Jnt;Jnun;}'where1killed', etc.
Preterite verbal noun
SI'-NTU(-Cx)(-Px); the mostcommonlyusedform is sLat without or with Px
kooutu
sLat koOutund;} 'whenkilled'
sLat + sl kooutund;Jn;}'whenI killed', etc.
Supine
SI-NAKd(-Px2)
kotunak;}
s3 kotunag;nu'in orderthat he killed', etc.
Presentparticiple
SI'-NTUd(-NCx)(-Px)
koOutu;} 'which haskilled - which hasbeenkilled'
Preteriteparticiple
SI-SUd-.D'dd(-NCx)(-Px)
kotucFu;}(P;};} 'which killed - which was killed'
Future participle
SI'-?sUTd(-NCx)(-Px)
koou?sut;}'which will kill- which will be killed'
Future-in-the-pastparticiple
SI'-?SUTd-DJdd(-NCx)(-Px)
kotum;}tuma?a'which was to kill- which was to be killed'
koou?sut~;};}
Abessiveparticiple
SI-MdTUMA]?Ao(-NCx)(-Px)
kotum;}tuma?a'which hasnot killed - which hasnot beenkilled'
Preterite abessiveparticiple
SI-MdTUMA]?Ao-DJdd(-NCx)(-Px)
kotum;numa?d j ;};} 'which did not kill- which was not killed'

Passiveparticiple
SI-Mdd(-NCx)(-Px)
kotum;};} 'killed'
Preteritepassiveparticiple
SI-Mdd-DJdd(-NCx)(-Px)
kotum;};Jdi;};} 'which was killed'
Verbal adverb
SI-SA]
kotudja 'havingkilled', niisi" kotucFa 'havingnot killed'
Verbal adverbofimmediateprecedence
SI-KAJ-SAj •
'having 'havingjust killed'
kotuga~a
508 NGANASAN

Table 16.12 (Continued)

Conditional-temporalverbal adverb
SI-HUl(-Px4)
kotubii? '(if - when) to kill',kotubiiniin~ kotubiiniin~ 'if - when I kill', etc
Preterite temporalverbal adverb
SI-HO?-d(-Cx) (-Px)
'whenkilled',kotubiiniin~ kotubiiniin~ 'whenI killed', etc.
kotubiiniin~ kotubiiniin~
Future conditional verbal adverb:
SI-HO?-NO-Px2
kotubiiniin~ kotubiiniin~'if I will kill', etc.
Connegative
S2-1
koOul

Key: 1-5 =conjugationtypes (subjective,objectivewith singular/dual/pluralobject,object-

less);S1-3 = variantsof stems;capital letters= morphonemes,seeMorphophonology,p. 489;
Vx = verbal personalendings(Table 16.11);Px (Pxl, Px2, Px4) = nominal personal
(possessive)endings(Table 16.7); Cx and NCx = nominal caseand nominal number+ case
suffixes (Table 16.8).
Notes:
1 Among two patternsthe first is usedin 1-3, and the secondin 4 and5.
2 From the viewpoint of morphophonologyit is possibleto treat -SJOd- as -SUI-, -lSJOTI- as
-lSUTI- -KOoOo - as -KUd-, -HAJAoTd(l)- as -HAJdTd(/)-, -HAJAO- as -HAJd-, -BSJOTI- as
-BSUTI-, -MdTUMAIA o- as -MdTUMAJld-.
3 For the form of imperativepresents2 seeTable 16.11.
4 For the form of imperativefuture s2 seeTable 16.11.
5 Negativeforms of this mood are hardly used.

verb nii- (for paradigmaticforms seeTable 16.12) and semanticallyrelated

auxiliaries, which are also used to render emphatic affirmative meanings:
niisi'd kUdf '(s)hedid not die' (cf. kuddiu:J '(s)hedied'), kasadiu:J kU:Jf '(s)he
almostdied',kU:Jfniili' '(s)hemustdie' (with negativeverb in the prohibitive),
IJu:Jlr kUdf 'how canit be that (s)hewill not die =(s)he will certainlydie'.

Derivation

Clities
Typical of Nganasanare the elities with meaningswhich in other languages
areusuallyexpressedby particlesor by attributes.Normally, Nganasanclitics
are attachedto S 1of both nounsand verbs, and occupythe position between
stemsand inflectional elements;they thus standon the borderline between
word-compoundingand derivation. The most common universal (nominal
and verbal)clitics are as follows:

-RA1AO 'only, still': kolfishes',

i~ri'a-f 'only fishes', i-Pia-dia 'only/still being' (from
ij- 'to be');
 NGANASAN 509

-BTA 'or': kuxFiimu nf-bfa 'man or woman',basu-bta-bulJnii-bfa-bii] 'are

you hunting or not?';
-KgUITIg 'even': tJiint~g;;,Pi#~m;;,huntinghunting 'even my archer'sbow', niisf lJ;]ffllu-
g;;,Pi#~? 'evenbeingnot eaten';
-IJALg 'also': kurxFi-lJal;;, 'any' (from kurxFi 'what kind of'), dior~lJal~tu lJ;]ffllu-
hitJii-Jjal~taagiimii
tf 'both cries andlaughs';
-KOocMUo 'but': taagiimii 'but the reindeer',niigiimiintf tu? 'but (s)he has
not come';
-TIg 'at least':b;;,(fa-#~bieven/at
ini n;;, 'if even/atleastI will grow';
-KUog (with emphaticoppositivemeaning):hir;;,g;;,;;,gii;;, 'asfor the high one'
or 'high (s)hereally is, but ...'.

Severalother clitics occur mainly or exclusively with nominals: -Tg (with

referenceto a future state),_DIgg (with referenceto a former state),-RgKU
(similative), -DIUoM (selective).

DenominalDerivation
The following patterns of derivation and their suffixes are especially
productive(someof themuseas their basesof derivationalmostany nominal,
including evenparticiplesandothernon-finite verbalforms, i.e. approach100
per centproductivity):

diminutives (S2-A?KUo with the final vowel(s) of S2 truncated,in certain

casesalso Sl'-KUo and S2-AIJKU with truncation):kubalku 'little skin'
(from kuhu), niilffa?ku 'the little living' (from the participle niilrtf;;,);
augmentatives(S3-RBA]?gor -?g after various stemmodifications):kubar-
ba?a 'big skin', niili"ta?a 'the big living';
denominaladjectives(S3-?BALg) with the meaning'abundantin X' (where
'X' is the baseof derivation): kuba?bal;;,'abundantin skins';
desubstantival locative adjectives, correlative with locatives singular
(Sl'-NTgg) and plural (Sl'-NTIc<l): korult;;,;;, 'which is in the house',
korulti?;;, 'which is in the houses';
denominal verbs with the basic meaning 'to acquire X' (S3-S-): si"rajku;;,
kuba?-'getswhite skin, becomeswhite-skinned';
denominalverbs with the basic meaning 'to have X' (S3-?Tg-): anikalim- k;J1ldiilt~
'goesby sledge'(from bnt;;, 'sledge');
denominalverbs with the basic meaning 'to becomeX' (S3-M-): anikalim-
'increases[intr]' (from anika?a 'big');
denominal verbs with the basic meaning 'to make X out of something'
(S3-MTU-): anikalimtii- 'increases[tr]';
denominal verbs with the basic meaning 'to lose X' (Sl-KAcUI-M-):
kuhugaPim-'losesone'sskin'.
510 NGANASAN

DeverbalNouns
At leastfour deverbalsubstantivescanbe derivedfrom practically any verbal
stem (except for some types of derived deverbal stems): nomen agentis
(S2-?SJI), nomen amatoris (Sl-KUTg), nomen instrumenti (Sl'-BSAN),
nomenloci (Sl-RgMU). Examples:basu?sii 'hunter', basur;;mtu
basugut~ 'hunt-lover,
fond of hunting', basupsa 'means/toolsfor hunting', basur;;mtu 'hunting
area'.The productivity of other deverbalnominal suffixes (-Tg for nomina
actionis, -HUg for adjectives)is restricted.

Intransitive and Transitive, Passive,CausativeVerbs

In primary verbal stemsthe (in)transitivity remainsunmarked,but in derived
verbs the stem-forming suffixes often render, along with other lexical and
grammaticalmeanings,the meaningof (in)transitivity. Thereare also regular
patternsby which transitiveverbsarederivedfrom intransitiveones(with the
suffix -BTU-: n~nsu- n;msupti~
'stops [intr]', 'stops [tr]') and of deriving
t~l;r
intransitive,especiallystative,verbsfrom transitiveones,e.g. 'closes[tr]',
t~ljtic-
'is (in a) closed(state)" b~tu- b~uac_
'grows[tr]', 'grows[intr]'.
Passiveverbs are regularly derived from transitive bases(SI-RU-). They
are perfectiveor imperfective(dependingon the aspectof their bases)andare
usually conjugated according to the subjectlesstype: yanasandi? yuam~ yanasandi?
diarbbtalurunda? 'people are knocking at my door', literally 'my-door
to-peopleis-knocked-at'.The high textual frequency of passiveforms and
constructionsin Nganasanis determinedby the stylistic rule which seeksto
preservethe samegrammaticalsubjectin a sequenceof sentencesthat form
a discourse.
Causativeverbs are regularly derivedfrom both intransitive andtransitive
bases(Sl-RU-BTU-, so that causativescan formally be viewed as transitive
derivatives of passive derivatives). They are perfective if their base is
perfective,and biaspectualif their baseis imperfective:r!ilibiaspectual
~ri"pti~ btr 'haslet
him live', r!ili~ri"biaspectual
pti~ti"ti" 'lets him live'.

Mode ofAction (Aktionsart)

The mostproductivemodesof actionandtheir suffixes are asfollows:

Iterative (imperfective verbs; Sl-Kg-): kotug;r'kills many times, repeat-

edly';
Habitual (imperfectiveverbs;Sl-MUMHA c-): kotumumba-'kills usually';
Durative (imperfective verbs, derived mainly from non-primary transitive
bases;Sl-KUoJ-): koturubtuguj- 'continuouslycausessomeoneto kill';
Non-perfective, denoting that an action remained unfinished or is only
planned (imperfective verbs, derived from perfective bases;Sl'-NTg-):
kOOut;r 'is in the processof killing, is trying to kill';
Inchoative(perfectiveverbs;SI'-?Kg-): ko8u?k;r 'startskilling';
Attenuative (derived from transitive bases;Sl'-BTU-): bikilling';
ar~pti"- 'opens a
 NGANASAN 511

little' (from biar;r 'opens');

Muitisubjective, denoting a collective action (Sl'-?NAR-): biiiilnar- '(of
many persons)leave;disperse'(from biiii- 'leaves,departs').

There are also two completelyproductiveimperfectivederivativeswhich can

be viewed as modal modes of action: the volitive (SI-NANTU-) and the
intentional (S2-?HAN- or S3-?HAN-), e.g. kotunantu- 'wants to kill',
ko8ulhuan- - ko&llhuan- 'is going to kill'.

Syntax
Noun Phrase
Within the noun phrase the attributes normally precede the head noun.
Adjectives and participles (often with subordinatewords of their own) can,
however,be put into focus by placing thempostpositively.
Adjectives(including adjectivalpronounsand participles)agreewith their
headnounsin number;adjectivesand numeralsagreewith their headin case.
The case agreementof attributes is complete in three cases (nominative,
genitive, accusative),but if the headnoun is in anothercasethe attributehas
the appropriate(singular or plural) form of the genitive: sN n:;>1Jh:;> taa 'bad
reindeer',sG n;m,b:;> taa, pA n;m,b:;>j taaj, pG n;m,bultaal,sLat n;m,b:;> taat:;>,
pLat n;m,bu1taatil.
The link between possessorand possessedis expressedby genitive
constructions(with genitive onlyin preposition)or by attachinga possessive
suffix to the possessed.
Simple Sentence
The dominatingword order is SOY; other types of word orders are usedfor
shifting the sentencefocus, especiallyin emphaticspeech.
The caseof the logical subjectis nominative,or genitive (in constructions
with most non-finite verbal forms), or lative (with passive and causative
verbs, when the nominative is reserved for the logical object, i.e. the
causator),or locative (in constructionswith passiveparticiplesin -Mgg and
-Mgg-DJgg). The case of the direct object in non-passivesentencesis
accusative,but nominative in special cases(e.g. when the verb is in the
imperative or when the direct object is non-focal and typical for the
correspondingverb); for many words, including personalpronouns, thereis
no formal distinctionbetweennominative,genitive, and accusativesingular.
Within the verb complexthe negativeauxiliary verb normally precedesthe
main verb, which standsin the connegativeform (but the negativeauxiliary
can be placed postpositively in emphatic affirmative constructions, see
p. 508). In compoundnominal and verbal predicatesthe conjugatedauxiliary
verb (usually ij- 'is') is placedpostpositively.
Due to the existenceof the interrogativemood,interrogativesentencescan
be both syntacticallyandintonationallysimilar to narrativesentences.
512 NGANASAN

Sentence Combining
Co-ordination or subordination of sentencescan be achieved both by
parataxisand with conjunctions:tujhUao~;Jm
diaggu?
diaggu?
(t~?), diaggu?
taani~ diaggu? 'I would
have come, but my reindeers are absent (= I have no reindeers),; the
adversativeconjunction t~?my belongs intonationally to the first part, and is
optional. The most typical methodof sentencecombiningconsists,however,
in transformingsubordinateclausesinto constructionswith non-finite verbal
forms; for examples see two such constructionsin the last sentenceof
the text on p. 514.

Lexicon
It can be said that the entire structureof Nganasanis dominatedby a strong
preferencefor rhythmically canonic trochaic word forms, in which the
boundariesbetweenbisyllabic (bivocalic) rhythmic groups tend to coincide
with morphemeboundaries.This tendencyis dueto the following factors:
1 A prevalenceof bisyllabic primary stems(= roots); in some casessuch
rootsresultfrom irregularlengtheningof vowelsin original monosyllabic
roots (bii? 'ten' insteadof bi?from Samoyedic*wtit) or from stem-final
vocalic extensions (m;m~ t~n~
'I', 'you' from Samoyedic*m::!n, *t::!n);
2 A prevalenceofbisyllabic suffixes (or suffixal clusters)with derivational
andnon-syntacticalinflectional meanings,as well asofbisyllabic incorp-
oratedclitics; thereareinstancesof rhythmically motivatedreduplication
(_CK~CK~_
of suffixes and clitics _cK~_
insteadof in inchoativeverbs,
T~T~-
- insteadof -T~- for expressingtemporalreferenceto the future in
(-~ _~DJ~
nouns)andof suffixal synonomy or in relativeadjectives).
This rhythmic preferenceis, on the other hand,also obviously relatedto the
metrical requirementsof the much-respected traditional versificationsystem,
and especiallyto the canonic metrical pattern of shamanisticincantations,
which normally consistedof octosyllabic lines with four bisyllabic feet in
each, with a caesuraafter the secondfoot and with no foot-medial word
boundaries,cf.: m;m~niim~ bio::! tagu / ;Jm~ dierioibinin~ 'motor( / l~timiaku
'motor(
baarb;Jm~/ ganasanu? siiidiamind~/niaag~?niili?sro~g 'My nameis Water
Eagle/ If I have guessedit / 0 my masterLaptimyaku/ until the expiration
of humans/ you wi11live well'.
The paradigmatic and syntagmatic rules of Nganasanphonology are
observedthroughout the inherited vocabulary (with violations allowed in
sound-imitative interjections) and, to a great extent, even in very recent
borrowingsfrom Russian.Cf. the phoneticadaptationin k~lku~s~ 'kolkhoz',
s;J/)i~sb 'Soviet (RussiancOBeTcKHH')', m~tu~r~
'motor('motor(-boat)" etc. Note,
however,that initial voiced consonantsare preservedeven in older Russian
loans: g~r;xl~ 'town (Russian ropop;)', d~r;J/)atu- 'greets (Russian
3p;0pOBaTbC5l)'.
 NGANASAN 513

SeveraldozenRussianwords were borrowedinto Nganasanin the period

from the beginningofthe seventeenthcenturyuntil the 1940s.Sincethen, the
influx ofloanshasgreatlyincreased.
Other foreign influences on the Nganasanlexicon have not been ade-
quately studied. Their sources are Dolgan (buluni ;) 'bastard', cf. Yakut
buluninia); Tungusic (turkutan;) 'sledgefor transportinggoods', cf. Evenki
turku; see Futaky (1983, 1990), Katzschmann(1986», Enets (ukucfarr
'white-nosedloon', cf. Enetsu)Qseri), Nenets(me1Jgia 'depressionin ground
with water', cf. Nenetsmie1Jta),and possibly even Ket (biia 'wind', cf. Ket
bei) as well as unknownsubstratumlanguage(s).The extralinguisticevidence
suggeststhat the layer of Enets loans (together with words borrowed into
Enetsfrom Nganasan)must be the most significant,but it is often difficult or
even impossible to distinguish such loans from the (Northern) Samoyedic
heritagewhich is commonto both EnetsandNganasan.

NganasanText
Source:Fieldwork.

A: morphophonemictranscription(stemsare not transcribedmorphophonem-

ically, but have an index showing their harmonic class); B: phonemic
transcription;C: morpheme-by-morpheme gloss; D: closetranslation;E: freer
translation.

nidv = deverbal nomen instrumenti; vadv = verbal adverb; vadvcond =

conditional/temporalverbal adverb; vidn = denominal intransitive verb;
vnpret = preteriteverbal noun; vtdn = denominaltransitiveverb.

Al <Pes12-Mg I) a<PaI-KgJ-TU t::l2+ij I-SUg-KgJ

Bl <Pesl-m::l l)a<Pa-g::lj-tju t::l+i -SjU::l-g::lj
Cl FATHER-sl YOUNGER.SISTER-N-s3 THAT+BE-pret-d3
Dl my father two youngersisters they two were

A2 djak::ll-K~}J<Pak::l-g::lj W-SUg-KgJ sjitF ru2 A31)amiajl

B2 <Pak::l-g::lj i-sju::l-g::lj sjiti ru B3 l)ami aj
C2TWIN-N BE-pret-d3 TWO WOMEN C30THER
D2 two twins they two were two women D3 one

<Pak::ll manul=RiAI? ku::lI-SUg ijl-HAMHU 1)amiaj I

<Pak::l manu=riai? kU::l-<Pu::l i-bahu I) amiaj
TWIN EARLY=adv DIE-pret(-s3) BE-renarr(-s3) OTHER
twin died early (so they say), the other
514 NGANASAN

diak;)l nokuI=RAIAo=NU nii1i'2-SUg

diak;) noku=raa=nu niil'i-dii;)
TWIN NEAR=ONLY =loc.adv LIVES-pret(-s3)
twin only recently was (still) alive

A4 t;)2=NU=A 1'lA o+W-SAI niim2-TU diak;)I=OJgg

B4 t;)=nji=i7 i a+i-sia niim-ti diak;)=di;);)
C4 THAT=loc.adv=lat.adv+BE=vadv NAME-s3 TWIN=FORMER
04 that beingthe case her name (was) 'Former1Win'

A5 kobtUaC1=MO-HU'l-g-TU dies'i2-Ng k;)nt;)I_C

B5 kobtUa=mti-bti'l-;)-tu dies'i-n;) bnd;)
C5 GIRL=vidn-vadvcond-pret-s3 FATHER-s1sG SLEOGE-sG
05 when shewas a girl my father's sledge's

tier;)2=OJgg mat1-U=S=BSAN-U='lTg-NTU+djaa C
tj er;)=dj ;);) mab-u='l=sian-u='lt;)-tu+djaa
LOAO=FORMER TENT-S3=vtdn=nidv-S3-vtdn-vnpreH TO
former cargo to (therebeing) a tent-festivity

biu1-HO'l-TU
btiti-bti-tti
OEPART-vadvcond-s3
whenhe left

E1 As for my father, he had two younger sisters.E2 They were twins, two
girls. E3 One of the twins died young, the other was still alive not long ago.
E4 Thereforeher namewasOiakadiaa('former[ly a] twin'). E5 Whenshewas
a girl, she used to be taken in my father's sledge, when he was going to
celebratetent-festivities(,Festivitiesofthe PureTent').

Referencesand Further Reading

Castren, M.A. (1854) Grammatik der samojedischenSprachen, St. Petersburg:
Buchdruckereider KaiserlichenAkademieder Wissenschaften.
Futaky, I. (1983) 'Zur Frage der nganasanisch-tungusischen Sprachkontakte',in O.
Bereczki and P. Domokos (eds), Uralisztikai tanulmdnyok (Hajdu Peter 60.
sziiletesnapjatiszteletere),vol. I, Budapest:ELTE, pp.155-62.
--(1990) 'EtymologischeBeitragezum Nganasanischen', SpeciminaSibirica 3:
51-5.
Hajdu P. (1964) 'Samoiedica',NyelvtudomdnyiKozlemenyek 66: 397-405.
Helimski, E.A. (1987) 'PYCCKHH roBopKa MeCTO Ka3aTb 6y~eM6y~eM (TaHMblpcKHH
IIH~)J{HH)', IIH~)J{HH)', in B03HHKHOBeHHe H rpYHK~HoHHpoBaHHe
rpYHK~HoHHpoBaHHeKOHTaKTHblX Jl3bIKOB,
Moscow: Nauka,pp. 84-93.
- - (in press) '0 MOPCPOJIOrHHHraHacaHCKoro513bIKa', paperread at the Fourth
InternationalSymposium'Uralic Phonology'(Hamburg,4-8 September,1989).
 NGANASAN 515

- - - (ed.) (1994) TaHMhIpcKHH 3THOAHHrBHCTHQeCKHH C60pHHK, BhlII. 1 :

MaTepHaAhI rro HraHacaHcKoMY maMaHcTBY H ll:3hIKY, Moscow: Rossijskij
gosudarstvennyjgumanitarnyjuniversitet.
lanhunen,1. (1991) 'HraHacaHhIH pacrra):lrrpaCaMO):lHHCKOHll:3hIKOBOH 06W:HOCTH',
in CeMHHap 'TIp06JIeMhI rrpOHCXO}K):IeHHll: HapO):lOB ypaJIhCKOH ll:3hIKOBOH
ceMhH', Izhevsk,I pp. 16-19.
Katzschmann,M. (1986) Tunguso-Nganasanica', Finnisch-UgrischeMitteilungen
10: 173-87.
- - - (1990) Vorliiufiges NganasanischesWorterverzeichnisauf der Grundlage
alter und neuer Quellen, Tei1 1: Nganasanischjremdsprachig,n.p. (private
edition).
Miko1a, T. (1970) 'Ada1ekok a nganaszannye1v ismeretehez',Nyelvtudomanyi
Kozlemenyek72: 59-93.
- - - (1986) 'Beitragezum nganasanischen Sprachgeschichte', Finnisch-Ugrische
Mitteilungen1O: 243-8.
Prokof'ev, G.N. (1937) 'HraHacaHcKHH (TaBrHHcKHH) ):IHaJIeKT', in 5l3bIKH H
IIHCbMeHHOCTb HapopoB CeBepa, lJacTh, I Moscow-Leningrad : Ucpedgiz,
pp.53-74.
Terescenko,N.M. (1973) CHHTaKcHc caMopHIIcKHX H3bIKOB, Leningrad:Nauka.
- - - (1979) HraHacaHcKHH H3bIK, Leningrad:Nauka.
- - - (1986) 'AmpaBHT HraHacaHCKoro ll:3hIKa' , in P.A. Skorik (ed.),
IIaJIeoa3HaTcKHe H3bIKH, Leningrad:Nauka,pp. 45-7.
17Nenets
Tapani Salminen

The traditional territory of the TundraNenetslanguageextendsalong a vast

tundra zone from the Kanin Peninsulain the west to the Yenisei River delta
andthe YeniseiBay in the east.The northernboundaryis formedby theArctic
Ocean,TundraNenetsbeingalsospokenon severalof its islands.In the south,
the languageboundaryextendsjust beyondthe tree line. In terms of present
administrativeunits of the RussianFederation,the areathus definedincludes:
(1) the whole Nenetsdistrict, including the Kolguev andVaigachIslands,and
part of the Mezen' county in the ArkhangelskProvince;(2) parts of the four
northernmostcountiesin the Komi Republic; (3) practically all of the Yamal,
Nadym, and Taz counties,about half of the Ural county, and minor parts of
the remaining three counties of the Yamal Nenets district in the Tyumen'
Province; (4) most of the Ust'-Yeniseiskcounty of the Taimyr district in the
Krasnoyarskregion.
The dialects of TundraNenetsexhibit relatively little diversity. There are
no grave obstacles to mutual comprehensiondespite the geographical
distance.This must be due to both the relatively recentoccupationof much
of the presentterritory and the greatmobility typical of the nomadicway of
life. Nevertheless,severalphonologicaland lexical, and a few morphological
isoglossescrossthe languagearea.
Three dialect groupsmay be recognized,viz. Western(to the west of the
Pechora,with the subdivisionsof Far Westernon the Kanin Peninsula,and
Mid Westernin the Malaya Zemlya); Central (from the Pechorato the Ural,
i.e. in the Bol'shayaZemlya); and Eastern(on the Siberian side, with the
subdivisionsof Mid Eastern,including the Ob' areaand the Yamal Peninsula,
and Far Eastern, to the east of the Ob' Bay). Phonologically, the main
bifurcation is between the Western dialect group, which exhibits several
peculiar innovations, and the Central-Easterncluster, which, though less
innovative, possessesa couple of common soundchanges.By contrast,the
Urals tend to divide morphological and lexical variants, so that it is often
justified to talk about specifically Europeanv. Siberian featuresof Tundra
Nenets.The actualisoglosses,however,vary from onecaseto another.Unless
statedotherwise,the materialpresentedin this chapteris in accordancewith
the Centraldialects.
The languagesthat historically border, and partly mingle in, the Tundra

516
 NENETS 517

Nenetscountry are Russianin the far west, Komi on most of the European
side,Mansi to a limited extentin the Ural area,Khanty in the Ob' area,Forest
Nenetsin the Nadym and the Pur areas,Northern Se1kupalong the Taz after
the historically attestedarrival of the Selkupsthere,andForestEnets,Tundra
Enets,Evenki, Ket, Yakut (Dolgans),and Nganasanin the east,eversincethe
gradualexpansionof TundraNenetsto the Yenisei area.Komi and Northern
Khanty are the two languagesthat are known to havehad the most extensive
contactswith TundraNenetsfor a lengthyperiod,while Russianinfluencehas
now by far surpassedtheir effectson TundraNenets.
The numberof TundraNenetsspeakershas beengrowing throughoutthe
historical era. Sincethe TundraNenetsareahasbeenmostly expandinguntil
recently, both the growth of the national population and the absorptionof
membersof other nations have contributedto an increasein the numberof
speakers.In recentdecades,however,the numberof speakershas remained
fairly constant,becausepopulationgrowth is offset by linguistic assimilation.
Currently, there are approx. 25,000 Tundra Nenets speakers.The official
populationfigure for the Nenetspeoplewas 34,665in the 1989Sovietcensus,
andthe numberof first languagespeakersamongthemwas 26,730,i.e. 77 per
cent,including approx.2,000ForestNenetsof whom maybe1,500spokeit as
their first language.Theaveragepercentage of nativelanguageproficiencytells
little of thereal situation,asit variesenormouslyfrom onedistrict to another.
The above-definedtraditional Tundra Nenets territory comprisesareas
that are known to have been inhabited by other peoplesin the beginning
of the historical era. Firstly, the areas west of the Yenisei were formerly
populated by speakers of Yurats. By now, the Yurats appear to have
completely adopted the Tundra Nenets language and identity, and the
recordingsof their original vernacularare meagre. Secondly, seventeenth-
century explorers reported that other, linguistically unrelated people had
beenliving side by side with the Nenetsin the westernmostareasof Kanin
and Kolguev. Presumably,an extensive part of the modern Nenets area
was inhabited by a more aboriginal population in not too remote
prehistorical times. There is a Tundra Nenets word syixortya referring to
the aborigines,vividly describedin Nenetsfolklore, but there is no material
evidenceof their languageor languages.
In more recenttimes, TundraNenetssettlershavecontinuedto expandby
inhabiting further areas lying beyond the traditional territory as defined
above. Theislandsof NovayaZemlya in the ArkhangelskProvincereceived
their first inhabitants only in the nineteenth-century,when the Russian
governmentbrought in Nenets families in order to strengthenits claim to
sovereigntyoverthe islands.SomeNenetsalsofollowed the IzhmaKomi who
emigratedto the Kola Peninsulain Murmansk Province, though it is not
known to what extentNenetsratherthan Komi was usedby them. In the east,
the Tundra Nenets-speakingareanow extendsacrossthe Bay of Yenisei to
larger parts of the Taimyr district, where a processof Nenetsization,similar
518 NENETS

to the completedabsorptionof Yurats, is underwayamong both groups of

Enets.
While continuously expandingin the east, the Tundra Nenets area has
lately beenrecedingon the Europeanside. Not only is the Russianpresence
most influential there,but a numberof Izhma Komi have also immigratedto
Nenetsareas,often taking a leadingpositionin economicspheres;this hasled
to many communities shifting to the use of Komi. Becauseof nuclear
experimentsbeginningin the 1950s,the inhabitantsof Novaya Zemlyawere
resettledin urban settlementson the continent,which effectively led to the
loss of native language commandamong the Nenets in question. The net
result is that while some of the local dialects in the vicinity of Komi areas
have alreadybecomeextinct, many if not all forms of EuropeanNenetsmust
be regardedas moribund. In the 1989 census,only 2,875, or 45 per cent, of
the 6,423 ethnic Nenetsin the Nenetsdistrict were fIrst languagespeakers,a
figure that cannotinclude very many adolescents,while 2,474 Nenetswere
listed as Russian speakersand 1,074 as speakersof another language,
obviously Komi.
The survival forecast onthe Siberianside is, as is to be expected,much
brighter. In the ~b'area,
area,the relative vigour of the aboriginalTundraNenets
and Northern Khanty communities together with the diversity of Komi,
Russian,and Tatar immigrant groupshavetraditionally favouredwidespread
multilingualism rather than the domination of a single language. In the
more eastern areas, it is Tundra Nenets that has functioned as a lingua
franca, gradually replacing other vernaculars. During the Soviet era,
however,RussifIcationpolicies and the massiveinflux of Russian-speaking
colonizers nearly eliminated both the multilingual tradition and the
interethnicuse of Tundra Nenets,leaving only the home and the traditional
economy, based on nomadic reindeer breeding, for the native language.
The deliberatealienationof children from their native languageand culture
through the Soviet schooling system is, of course, deeply felt among the
Siberian Nenets, as well, so that not even Nenets homes have avoided
RussifIcation. Nevertheless,the traditional Nenets way of life is still a
competitive alternative to the adoption of Russianhabits and, eventually,
of the Russian language. Many younger Nenets seem to be devoted to
the maintenanceof their national culture, and to the Nenets languageas
its expression.According to the 1989 census,the rate of native language
retention was a remarkable94 per cent (19,713 of 20,917; the fIgures also
include ForestNenets)in the Yamal Nenetsdistrict, and a fair 81 per cent
(1,990 of 2,446) in the Taimyr district.
Despite many positive indications, even the Siberian Tundra Nenets
community is still very much threatened,and perhapsmore so now that the
heartlandsof the Nenetscountry on the Yamal Peninsulaare being invaded
by oil andgasprospectors.Becauseof continuingRussiancultural oppression
and economicexploitation,only a wide-scalenational awakening,leadingto
 NENETS 519

a real ethnic autonomy witha strict control of the native territory, may secure
the long-termexistenceof the TundraNenetspeopleandtheir language.

Phonology

Syllable Structure
The basic syllable structureis CV(C), i.e. a syllable consistsof an initial
consonant,a medial vowel, and an optional final consonant,e.g. ya 'earth',
myaq 'tent', wada 'word', ngarka 'big', nyaX'r 'three',xampol 'litter'.
There exist, however,vowel sequences,with the schwa0 or the reduced
vowel ¢ as the latter segment.Such sequencesare bestdivided into separate
syllables,which yields an additional,non-initial syllablestructureo/f'J(C), e.g.
xooba 'cradle',nyaOra 'inner part of hide', ng¢¢bf'q 'poison',w{h 'tundra' :
sG w{)h, to- 'to come' : subj sl to¢dom: s2 to¢no : s3 too.
The basic syllable structureimplies that word forms do not begin with a
vowel. However, many dialects seemto contain a few words with an initial
vowel. For instance, ¢mke 'what' is widely used instead of ng¢mke. By
contrast,the Westerndialects, becauseof the loss of initial *ng, possessa
largenumberof initial vowels, and thus an initial syllablestructureV(C), e.g.
Westernarka 'big' - Central-Easternngarka.
Monosyllabic word forms cannotend with a reducedvowel, so there are
no word forms of the structure *Cf'J. In other respects,the basic syllable
structureholds good for all dialects, i.e. there are no diphthongsor double
vowels, no initial or final consonantclusters, and no medial consonant
clusterswith more thantwo consonants.
StressPattern and Vowel Reduction
Stressis predictablefrom syllableandsegmentalstructure.It falls on an initial
syllable, syllablesprecedinga syllable with a schwa,and non-final syllables
precededby an unstressedsyllable. Stressdoes not fallon a final syllable,
syllableswith a schwa,and syllablesfollowing a stressedsyllableunlessthey
precedea syllablewith a schwa.Most typically, the first syllable of eachtwo-
syllable string is stressed,or in other words, the stressfalls on non-final odd
syllables.
However, the predictability of stressholds good only if the oppositionof
the reducedvowel and the schwais respected,becausethe stressrelations
which govern vowel reductionare partly lexical and morphological.Taking
into accountthose complications,vowel reduction is an automatic phono-
logical processwhere¢ ~ 0 in unstressed positions.It yields alternationssuch
as xa~d¢da.
x¢~ 'knife' : poss. sNs2 x¢r¢~x¢r¢~ : s3 x¢~da, x¢~da, xar¢d° 'house' : xa~d¢~xa~d¢~ :
xa~d¢da.xa~d¢da.
Undercertainconditions,however,the unstressedschwavowel appearsin
an odd syllable. Suchcasesinclude, in the first place,vowel sequencesafter
a stressedsyllable, unlessthey precedea syllable with a schwa,e.g. xada- 'to
520 NENETS

kill' : sOs3 xadaoda : sOs2 with a clitic particle xadaorrjJ-wa (instead of

*xadarjJ~-wa; *xadarjJ~-wa; cf. xadarjJ~
*xadarjJ~-wa;without a clitic particle).
A numberof sequences with aconsonant,in mostcasesa suffix-initial glide,
following a vowel, show a similar effect, e.g. xada- 'to kill' : pOs3xadeyOda,
tyenye- 'to remember': partic.futtyenyewOnta, srjJwa 'good' : sProssrjJwawona,
yedeyO 'new' : sProsyedeyOwrjJna, => yedeyOmta- 'to renew',tyorye- 'to shout':
subord s3 tyoryebOta. In the last mentioned case, there is variation, and
tyoryebrjJta is alsoattested.However,manifestingthemorphologicalcondition-
ing of the stresspattern,a similar sequenceretainsthe reducedvowel in other
morphological structures,e.g. nyadayrjJ- 'to smell of lichen' : partic.impf
nyadayrjJda, tyenyewrjJ- 'to know' : partic.impftyenyewrjJna.
Further,secondaryvowel stemsformed from liquid stemshave the added
vowel unstressedwheneverpossible, often yielding a schwa in the third
syllable,e.g. yayol- 'to grow turbid' => iter yayoZOngkrjJ-, syibyel- 'to turn pale
andearthy' => syibyelOngkrjJ-, poyol- 'to get mixed up' => tr poyoZOta- 'to mix
up'. Such casescontrast with primary vowel stems, e.g. wenolrjJ- 'to get
frightened(an animal), => iter wenolrjJngkrjJ-, pyidyelrjJ- 'to becomepliable' =>
pyidyelrjJngkrjJ-. In other instances,though, it is the final vowel of primary
vowel stemsor the vowel precedingthe stem-finalconsonantthat undergoes
reduction, e.g. lrjJkadrjJ- 'to snap (fingers)' => tr lrjJkadOta-, nyancyalrjJm- 'to
becomedumb' : s3 nyancyaZOma, => freq nyancyalOwor- 'to grow dumb', =>
tr nyancyalomtye- 'to makedumb'.
In a few loanwords,a schwamay also appearin an initial syllable, e.g.
pOrasyin° 'tarpaulin', roronyi 'peasant's sledge (for carrying wood)',
~ryfs°tyanye-ny~na 'club (playing card)'. The primary stress is then on the next
syllable.
In compounds,eachpart has a separatestresspattern,e.g. nye-nya 'sister'
: sLoc nye-ny~nanye-ny~na (ratherthan *nyenyaxrjJna).

Vowels
The general system of Tundra Nenets vowel phonemesis shown below.
(Distinct subsystemsfor plain and stretchedvowels arejustifiable mainly on
historical grounds.)

Schwa Reduced Plain Stretched

u u f u
e o
o
u a re

Someof the FarWesterndialectsseemto lack *re (> e).

Pronunciation
1 Frontnesslbackness. All vowels exceptce (*Cyce not existing) have two
basic allophones,a front one when precededby a palatalconsonant,and
 NENETS 521

a back one when precededby a non-palatalconsonant.In other words,

palatalityis a propertyprevailingwithin the syllable as a whole.
2 Quantity. The schwais pronouncedeither as an over-shortvowel or not
at all; nevertheless,it is always reflectedin the phoneticsubstance.The
reducedvowel is generally a relatively low and short vowel. The high
plain vowels vary from half-long to short.The mid plain and all stretched
vowels are pronouncedlong or half-long. In stressedpositionsa is long
but subject to reduction when unstressed.The vowel sequencesare
invariably over-long or long vowels, with the possibility of two syllable
peaksin their pronunciation.
3 Phoneticvowel reduction. When unstressed,a is pronouncedshort, or
half-long at best, and more central, close to the quality but not the
quantity of the mid vowels. (As a relic of the former view that the
reductionof a shouldbe regardedas phonemic,early publicationsby the
authorusethe symbolafor the reduceda.)
4 Phonetic schwa. In sequencesof qC', an over-short vowel is often
pronouncedbetweenthe glottal stop and the consonant.
5 Diphthongization.Most typically, and especiallyin the Easterndialects,
(R is pronouncedas a slightly rising diphthong,while e and, after palatal

consonants,a, canbe slightly falling.

6 Vowel harmony.After x and, in the few existing cases,q, the quality of
the following 0, ¢ and, occasionally,a matchesthat of the preceding
vowel. The quantitative oppositions remain, so that the phonemic
distinctions are preserved.A phonetic metathesisoccurs in unstressed
sequencesof VXO so that the pronunciationof the schwais almostthat of
a full vowel while the precedingvowel is phoneticallyreduced.However,
no neutralizationtakes place, so that pairs likereduced.
xor¢~q 'oven' pD and
xOr'xaq 'birch' simulativepN arekept apart.
7 Especiallyin the Easterndialects,unstressede and 0 appearas relatively
high vowels, quite close tothe allophonesof i and u. However, in most,
if not all, dialects,the contrastis retained.

Orthography
The basic standardorthographyrecognizesonly five vowel units, with a
double numberof vowel signs for indicating the palatality of the preceding
consonant,i.e. ¢ & a = <ai5f.>, e & (£ = <3/e>, i & (= <hI/H>, 0 = <ole>, u & U
=<y/IO>. The schwausually has no overt marking, exceptafter x and q where
it is written according to (phonetic) vowel harmony, and, inconsistently,in
other positions,suchas betweenconsonantclusterswhereit is written like ¢.
In dictionaries,a refined version of standardorthographyis used.It could in
principle distinguishall vowel phonemes,but fails to do so in practice.Most
consistently,(back) e and (R are differentiatedby a dot on the former, i.e. e
= =
<3>, (R <3>. The high stretchedvowels are, by contrast,only rarely written
with a macron, i.e. (= <if>, U = <y>. As a further complication, the macron
522 NENETS

is alsousedto indicatevowel sequences. Thecrucialdistinctionof ¢ v. a is more

often thannot renderedwith a micron overthereducedvowel, i.e. ¢ =<i:i/H>, a =
<a/5l>. Vowel harmonyis indicated,e.g.noxa'Arctic fox' =<HOXO>, nixo 'power'
=<HbIXbI> «HblXbI> ),pyix"nya'nightsLoc' =<IIHXHH5l> (<IIHXHH5l> ). Thereduced
vowel is written allophonically alsobeforea pre-schwalabial glide, e.g.pad¢wo
'bagsNsI' =<IIa}l;YB> ( <IIa}l;YB> ). Thephoneticschwabetweenaglottal stopanda
consonantprecedinga schwais reflectedin the orthography,though incon-
sistently,e.g. waqwO 'bed' = <Ba"aB> «Ba"aB»,myaqmO'tent sNsI' = <M5l"aM>
«M5l"aM> ).
Phonotactics
As explainedin connectionwith the stresspattern, the schwa is practically
absent in the first syllable, and there cannot be two schwa vowels in
consecutivesyllables.As the latter part of a vowel sequence,only the schwa
or the reducedvowel, not mutually contrastivein that position, may appear.
The stretchedvowel ce does not appear after palatal consonants,and the
reducedvowel ¢ not after labiopalatalconsonants.The distribution of non-
initial stretchedvowels is not entirely clear: in the presentdescriptionnon-
initial ce is restrictedto the essive-ngce and the high stretchedvowels do not
appearin non-initial syllablesat all, with the possibleexceptionof dialectal
accusativeplural forms.
Consonants
Palatal consonantsare marked by Cy digraphs. Word-initially and post-
vocalically, whereno confusionwith palatality markeris possible,the palatal
glide is written y insteadof y. In the morphologysection,p. 526, suffix-initial
y always indicates the palatalizationof the precedingconsonant.The dual
marking of the glottal stop as eitherq or h is explainedbelow.
Historically, the Central-Easternsystem has changedfrom the proto-
system only by *wy ~ by. By contrast, the Western systemhas not only
retained wy, but has also acquired four secondary consonantsthrough

Table 17.1 ThndraNenetsconsonantphonemes:the two main systems

Central-Eastern Western

Nasals m my n ny ng m my n ny ng
Stops strong p py t ty k qlh p py t ty k qlh
weak b by d dy b by d dy g
Affricates strong c cy c cy
weak t jy
Fricatives s sy x s sy x
Semivowels w y w wy z y
Liquids lateral t ly t ly
tremulant r ry r ry
 NENETS 523

denasalization,i.e. *nt ~ d whereby*d ~ z, *nc ~ j, *ncy ~ jy, and *ngk

~ g; *mp, *mpy and *ndy have presumablymerged with b, by, and dy,
respectively. Two additional systems are also known to exist, viz. a Far
Easternone where *c ~ s, *cy ~ ty, and a Westernone with wy but without
denasalization,thus lackingj,jy, g, andz.

Pronunciation
1 Palatalizationlvelarization.While the palatal counterparts of dental
consonantsare also phoneticallypalatal, the labiopalatalconsonantsare
palatalized.On the other hand, the non-palatalcounterpartsof palatal
consonantsare frequently velarized. Cf. frontnesslbackness of vowels,
p.520.
2 Postnasalobstruent weakening. Especially in the European dialects,
obstruentsareoften voicedafter a nasal.Furthermore,postnasalaffricates
may lose their closure, yielding phonetic voiced sibilants, which is also
the usualpronunciationof the Westernweakaffricatesj jy.
3 Fricativization.The Central-Easternweakobstruentd andthe historically
identical Westernz are typically pronouncedas a fricative. Other weak
obstruentsare subjectto slighterfricativization.
4 Gemination.All consonantsexcept the weak obstruents (includingthe
Westernz) and x are half-long and often transcribedas short geminates
in intervocalicpositions;the sameis true of obstruentswhenprecededby
a liquid.
5 The added glottal stop. A glottal stop is pronounced after other
consonantsin final position,i.e. b I m r. The resultingphoneticsequences
differ from sequencesof a consonantfollowed by a schwaand a glottal
stop, i.e. C'q (C'h), mainly by the consonantsbeing pronounced
markedlylongerin the latter case.
6 Most typically in the Far Easterndialects,the palatalobstruentssy and ty
« *ty and *cy) arepronounced'with a morehushingquality.

Orthography
Spelling of consonantsis for the most part phonemic,but in a few casesit is
phonetic.Postnasalobstruentweakeningis reflectedin standardorthography
with <M6>, <M;n;>, <M3>, <Mr>, <H;n;> , <H3>, <qr> for mp, mt, me, mk, nt, ne,
ngk. In normativeorthography,the glottal stop is written with separateletters
for the different sandhi variants, i.e. q = <">, h = <'>, though many
publicationsare contentwith a single letter. The addedglottal stop is written
with the sameletters, so that m is followed by <'>, e.g. num <HYM'> 'sky',
and b, I, and r arefollowed by <">, e.g. ngob <qo6"> 'one',xampol<xaMbOA">
'litter' , yur <lOp"> 'hundred'.It is vital that the addedglottal stophas an overt
expressionin the orthography,becausefrom the phonemicpoint of view, it
denotesthe absenceof the schwain the final position.
524 NENETS

Sandhi
Consonantsandhiis an automaticphonologicalprocess,valid irrespectiveof
word boundaries.It includesthe following interconnectedsubprocesses:

1 Postvocalicobstruentweakening:p py t ty --t b by d dy --t - V, e.g. ya

'earth' : sNs3yada (cf. yam 'sea': yamta,yar 'side' : yarta);
2 Postconsonantal continuantstrengthening:s sy x --t c cy k /C _, e.g. yam
'sea' : sLoc yamf<Dna, :=:} com yamcaweyO;yar 'side' : sLoc yarf<Dna, :=:}
comyarcaweyO;(cf. ya 'earth' : yaxOna,:=:} yasaweyO);
3 Preobstruentalloss of the non-nasalizableglottal stop: q --t 0 / _
C[obstruent],e.g. yaq 'strandof hair' : sNs3 yata: sLoc yaf<Dna, :=:} com
yacaweyO;
4 Preobstruentalnasalizationof the nasalizableglottal stop: h --t m n ng /
_ C[obstruent],e.g. yah 'soot' : sNs3yanta: sLoc yangf<Dna;
5 Presonorantalloss of the nasalizableglottal stop: h --t 0/_ C[sonorant],
e.g. yah 'soot' : sNs2yalO (cf. yaq 'strandof hair' : yaqlO).

Phonotactics
On the basis of consonantsandhi, consonantsare divided into primary and
secondaryconsonants.The secondaryconsonants,b by d dy c cy k and the
Westernjjy g z, arein all instancesderivedfrom primaryconsonants,
exceptthe
caseswhereb andby aredueto themorphophonological palatalization of w.
Phonotacticdistribution of Central-EasternTundra Nenetsconsonantsis
as follows:

I_V m my n ny ng p py t ty s syx I ly wy
CV m my n ny ng p py b by t ty c cy k I ly wy
Vj m my n ny ng p py b by t ty d dy c cy s sy k x q I ly r ry wy
V_C mn h ng b q I r
Vj m h b q I r

Restrictions on initial consonantsare strictly observedin the present-day

language,as seenin such Russianloanwords as Eur pangkor - Sib pakor
'gaff « bagar), syasO 'hour, watch' « cas), xosoka 'cat' « koska). The
initial vibrants r ry are nowadaysallowed in the Europeanbut not in the
Siberiandialects, e.g. Eur ryesOka- Sib lyesOka 'unleavenedflat cake, pie,
(Sib also) dough'. Initial secondaryconsonantsmay also appearin recent
loanwordsbut only in restrictedareas,viz. b by d dy g in the Westerndialects,
and c cy k in the neighbourhoodof Komi and Khanty dialects. Post-
consonantalby appearsonly as a resultof the morphophonologicalpalataliza-
tion of w, as in myirwo 'weapon' : pA myirbye, syfqwO 'seven' :=:} ord
syfqbyimtyeyo'seventh'.
Vowel sequences cannotprecedex. This is reflectedin verbal morphology
in connectionwith x- initial suffixes.
 NENETS 525

The intervocalic glottal stop is present only in a few cases, notably

augmentativeforms like p¢ncyeqO'louse'andthe (typically Eastern)adverbal
stemtyuq¢- 'up'.
Because of consonant sandhi, the glottal stop is excluded and the
oppositionof the non-labialnasalsis neutralizedbeforeobstruents.
There are three interpretationsof the glottal stop soundin final position
accordingto its behaviour,but only the nasalizableand the non-nasalizable
are phonologically significant, the added glottal stop being an automatic
concomitant of a prepausal consonant. The nasalizable and the non-
nasalizableglottal stop, transcribedh v. q, are to be understoodas not
phonemicbut nonethelessphonological,their oppositionbeingmanifestedby
distinct sandhipatterns,e.g. nyeh x¢no =nyeng_k¢n°'a woman'ssledge'v.
nyeqx¢no =nyeng_k¢n°'a women'ssledge',toh war" =to_war" 'a shoreof
a lake' v. toq war"q =toq_war"q 'shoresoflakes'.The nasalizableglottal stop
is, consequently,presentonly prepausally.In all likelihood, its pronunciation
coincideswith that of the non-nasalizableglottal stop. A reservationmay be
in order here becausespeakersare clearly aware of the dual phonological
nature of the glottal stop, a circumstancethat may give rise to some
difference,real or pretended,in the pronunciation.In any case,the description
obviouslybenefitsby havingdistinct symbolsfor the two phonologicalglottal
stops.
The number of consonant clusters is further restricted by morpho-
phonological processes,as a residue of assimilative sound changes.In a
few cases,there remains a possibility of analogicalrestorations,the extent
of which varies from one dialect to another. The acceptable clusters
include:

1 q + a sonorant.On the basisof Lehtisalo'srecordings,it is often assumed

that the glottal stophasbeenlost preconsonantally in the Siberiandialects
but this doesnot hold true for all if any of them.
2 m, b, l, r + any consonant.Geminatetype clusters,i.e. mm(y),bp(y), ll(y),
rl(y), are, however, in principle excluded, though they may emerge
analogicallyin somedialects.The clusters*mw, *mby, *bw, *bby are, by
contrast,unattested,andprobablyunacceptable.
3 n, ng + a homorganicobstruent.
4 n + y. This cluster appears to be of analogical origin. Dialectally,
analogicalnl(y) may also appear.

Acrossword boundaries,geminatetype clustersexcludedaboveas well as

m_w and b_w occur freely when the first word ends in m, b, l, or r. By
contrast, *ny, *n_l, *n_ly are impossible.It is anotherissue whether the
outcomeof a nasalizableglottal stop plus an initial sonorantdiffers from a
single consonant.While there may be some evidence that points to an
affirmative answer, the question does not arise when words are spelled
526 NENETS

separately,the nasalizableglottal stop being consequentlyrenderedwith its

own symbol,i.e. h.

Morphology
Within inflection, thereis oneword-form, the absolutenominativesingularof
nouns, where no morphological processis involved, so that the form is
identical with the basic stem, disregardingmorphophonologicalprocesses.
For a few stemtypes, no morphologicalprocesstakesplacein the formation
of the absoluteaccusativeplural, either. In derivation, there are instancesof
noun-verbconversion.All otherword forms andderivativesexhibit morpho-
logical processes,i.e. suffixation, modification, or (partial) suppletion.
Suffixation is the mostfrequentof them,thoughmodification is by no means
uncommon.Partial suppletion,by contrast,is rare and confined to a number
of irregularverbalforms.
Morphological Word Classes
The two major word classesare verbsand nouns.Alongsidethe nouns,there
are minor classeswhich exhibit somenominal categories.Theseinclude the
personalpronounsand various groups of adverbsand postpositions.Adjec-
tives do not form a word class distinct from nouns on the basis of their
inflection, thoughthey may havederivationalpeculiarities.The sameapplies
to numerals,i.e. ngob '1', syidya '2', nyax"r '3', tyet' '4', s¢mpOlyangftY(in
Europeandialects s¢mOfyangk,n)'5', m¢t'q '6', syfqwO '7', syidontyet' '8',
xasu-yuq (Easternxasawa-yuq)'9', yuq (Easternalso luca-yuq) '10', yur
'100', yonOr '1,000'. Non-personalpronounsalsoconformto normalnominal
inflection, e.g. demonstrativetyukuo 'this', takio 'that', tyikio 'it', t¢rcya 'that
kind', or interrogativexfbya 'who', ng¢mke'what',xurka 'what kind'.
The residualmorphologicalclassis the particles,i.e. non-inflectedwords.
In syntacticclassification,they would mostly be includedamongthe adverbs,
as there are no true conjunctions in the language. Some particles are
synchronicallyunanalysable,e.g. t¢ryem 'thus,so', ngoq - ngodOq'also',but
othershavemorphologicalstructure,evidentin derivation,e.g. tyedah 'now'
~ lim tyedaryih 'for a while', x¢n-cyer)q 'how' ~ lim x¢n-cyelyiq
'anyhow'.
StemTypes
The basic stem types are (1) vowel stems,(2) glide stems,(3) mixed stems,
and (4) consonantstems, the major division being between vowel and
consonantstems.The numberof glide stemsis very small, and they differ
from vowel stemsonly becauseof the appearanceof a glide when a suffix
with an initial vowel is attached.Mixed stems,including all 0- and some¢-
stemsof verbs,show, besidesspecific peculiarities,a mixture of propertiesof
vowel and consonantstems;the remaining¢-stemsas well as all monosyl-
labic, a- and e-stemsof verbs belong to vowel stems.Two verbs, Xa?- 'to
 NENETS 527

depart' andnglX- 'to be', may be characterizedas irregular.The negativeverb

nyi- 'not' alsoexhibits idiosyncrasies.

morphophonological
~orphophonology
morphophonologicalprocessesinclude morphophonologicalassimilations,
~orphophonological
alternations,and changes,vowel stemformation, truncation,and (de)palatal-
ization. They differ from phonologicalprocesses,i.e. (primarily) consonant
sandhi,in that they do not occurat word boundariesandthat they arenot valid
stem-internally.Cf., for instance,ngarkam¢,-oq'big town', and nyema'sleep'
or xamada- 'to understand'despite a morphophonologicalprocessm > w
intervocalically.Analogicalexceptionsare alsopossiblein certaincases.
Assimilations
1 r ry > Ily / C _, e.g. nyum 'name' : sNs2nyumlO, syer 'thing' : syelO (cf.
nya 'friend' : nya,-o).
2 t s > q / _Cor #, e.g.myaq 'tent' : sNs2myaqlO: s3 myata(cf. sG myadOh,
pA myado),m¢nOq'lump' : m¢n¢qIO: m¢nOta(cf. m¢n¢soh,m¢nOso).
3 n ng > h / _C, excepty (and, dialectally, liquids) in most cases,or #, e.g.
peh- 'to put' : modal gerundpencyO: s3 penga (cf. pOs3penjFda), weh
'dog' : dN weng~h : sNs2we/°(- wenlO) : sGsl wenD (cf. sG wenoh,pA
wenD or odorativeweny¢-),wih 'tundra' : sD win~h: sNs2 w{[O : sGsl
win° (cf. pA wingo), tideh 'sembrapine'=> odorativetidey¢- 'to smell of
sembrapine'.
4 ng > n / _ yin certain cases,e.g. nyenecyOh'person,human being' =>
pejor nyenecy¢nye'poor man'.
5 ng> 0/_ ¢ (in vowel stemformation), e.g. win 'tundra' : sG w{'h.
6 m> w /V _ V, e.g. ngum 'grass': pN nguwOq: pA nguwo,ng¢m- 'to eat'
: connegng¢wOq : inf.perf ng¢woqma,xo- 'to find' : inf.impf xowa (cf.
xoq- 'to fetch' : xoqma).
7 Degemination:m> 0/_ m, e.g. nyum 'name' : posssNsl nyumo; p--7
0/ _p, e.g. ngob 'one' => mod ngopoyO'the one'; 1 r--7 0/ _lor r, e.g.
ser 'salt' : sNs2 selo. Dialectally, degeminationmay be suppressedby
analogy.
8 a> e / _y¢ in non-initial syllablesundercertainconditions,e.g. xada- 'to
kill' : pOs3 xadeyOda, ord nya~r 'three' => nyaxoromteyo'third' v.
nyabyi 'other' => nyabyimtyeyO'second'.
9 1/l--7 u / _ w
in prosecutivesingular,e.g. x¢,-o 'knife' : sProsx¢ruwona.
Alternations
1 n-y-stems:n _ C, Y _V, e.g. toh 'blanket' : pN toyOq : sD ton~h toyo, : sLoc
tong~natoyo, : pA toyo, peh- 'to put' : partic.impfpenta: s3 penga : pOs3
peny-oda : connegpeyoq :info perfpeyoqma.
2 e-i-, e-i¢-, O-U¢-, 0-y¢-stems:the latterin final syllablesandoptionally
precedingthe preteritesuffix, e.g. ti 'reindeer' : sNs2 te,-o : pred pret s3
tisyO (- tesyO),me- 'to be' : connegmiq;p¢niO 'dress': sNs2p¢ne,-o: pred
528 NENETS

pret s3 pf/lnif/lsyo (- pf/lnesyO); tyukuo 'this' : sD tyukonoh; ngopoyO 'the

one' : sNs2ngopor'.
3 Glide stems,with 0 - w or y: the glides appearbefore a vowel, e.g. xa
'ear' : pA xawo, yf 'wits' : pA yfbye, myf- 'to prepare' : info perf
myfyeqma.
4 Suffixes with initial n - t: the latter attachesto consonantand mixed
stems, e.g. luca 'Russian' : pred s2 lucan° : sD lucanoh V. nyenecyOh
'person, human being' : nyenecyf/ln~nyenecyf/ln~h, : nyenecyf/ln~h, nyenecyf/ln~h, nu- 'to stand' :
partic.impfnuna V. myih- 'to go' : myinta (- myintya) V. xonyo- 'to sleep'
:xonyoda.

Suffix-initial Change
In the formation of the generalfinite stem for m-stems, the ng of the suffix
is lost: ngf/lm- 'to eat' : s3 ngf/lma. This is not a question of a simple
assimilation, as mng generally remains intact, e.g. ngum 'grass'~simple ess
ngumnga?

Internal Changes
The few casesincludethe accusativeplural stemsof xasawa 'man,male' : pA
xasyewO, yf/lxa 'river' : yesyi, and the generalfinite stemand the connegative
of the mixed stemsyoxo- 'to disappear': s3 yuxu : connegyuxuq, toxo- 'to
learn' : tuxu : tuxuq. As can be seen,internal vowel changesdo not occur
independentlyof a modificationaffectingthe final vowel of the stem.

Vowel Stem Formation

Consonantstemshave a secondaryvowel stemusedbefore certain suffixes,
formed by addingf/l to the final consonant.For ng-stems, removalof the final
ng takesplace simultaneously,which yields a vowel sequence.The suffixes
in questioninclude thoseconsistingof a single consonant,viz. sA -m, sG -h,
and pN -q, e.g. myaq 'tent' : pN myadOq, syuh 'navel' : syuOq, and conneg=
imp s2 -q, e.g. myiq- 'to give' : myisOq. Other inflexional suffixes involved
are the imp pOs2-n-f/lq, e.g. ngf/lm- 'to eat' : ngf/lwf/lnOq, andimp refl s2 -t-f/lq,
e.g. sf/ll- 'to return' : sf/llf/ldOq. There are more caseswithin derivation, for
instance the comparative -rka, e.g. Sf/lngkowoq- 'to be heavy' ~
Sf/lngkowosorka- 'to be heavier'.

Truncation
Certain suffixes and suffix combinations with two initial consonantsare
subjectto the processof truncation,wherebythe first consonantis lost when
attachedto a stemwith a final consonant.The casesinclude the second-and
third-person accusativeand genitive possessivesuffix combinations, e.g.
myaq 'tent' : posssNs3 myata = sA = sG (contrastya 'earth' : sN yada : sA
yamta : sG yanta), and the necessitativeand durative suffixes, e.g. ngf/lm- 'to
eat' : nec s3 ngf/lmcu (cf. pya- 'to begin' : pyabcu), myiq- 'to give' ~ dur
myipf/l- (cf. xo- to find' ~ xompf/l-).
 NENETS 529

Palatalizationand Depalatalization
There are two processeswhich affect the palatality of a consonant.(1)
obligatory palatalization or depalatalization of a consonant occurs in
connectionwith certain morphologicalprocesses;(2) optional palatalization
concernsthe initial consonantsof certain suffixes. In both cases,the non-
palatal v. palatal pairs of consonantsare m H my, n H ny, ng ~ y, p
H py, t H ty, k ~ cy, b ~ by, d H dy, c Hey, S H sy, x ~ sy,
w H by, r H ry, l H ly; H indicates both palatalization and
depalatalization,~ only palatalization.There is a tendencyfor palatality
within the stem to favour palatalization,e.g. nyum 'name' : nyubyev. ngum
'grass' : nguwo, but there are counterexamples:syer 'thing' : syero or
tf/Jr 'body-hair' : tf/Jrye. The two palatality processesdo not necessarily
conform, cf. syer : sProssyermOnya.
Obligatory palatalization and depalatalization are met with in four
contexts:

1 accusativeplural stemformation, e.g. nguda 'hand' : pA ngudyi, tyonya

'fox' : tyon°, myir 'price' : myirye, and a number of derivational
operations,e.g. ngf/Jdyim- 'to appear'=> freq ngf/Jdyibyer-;
2 generalfinite stemformation for m-stems,e.g. ngf/Jdyim- 'to appear': s3
ngf/Jdyimya;
3 mixed steminflection, e.g. xonyo- 'to sleep': s3 xoni;
4 the dual personalsuffix, e.g. sNd2 -r-yih (cf. s2 -r_O andp2 -r-aq).

Optional palatalizationyields suffixal variantswith a palatalconsonant.It

occurs in conjunction with certain suffixes when attachedto a number of
stems,which requirelexical marking. Examplesof nominalforms includeyiq
'water' : sLoc yi~nya
yiqmOnya : sProsyiqmOnya: posssNs3yitya : p3 yityoh, sy{ 'hole'
: sLocsy{wOnya;
sy{~nya sy{wOnya;
: sPros sy{wOnya; ord nyabyimtyeyO'second' (cf. ny~­
romteyO 'third'). Examplesof verbal forms involve mainly the imperfective
participle,e.g. yilye- 'to live' : yilyenya,myih- 'to go' : myintya,ptef/Jr- 'to do'
: pte°rtya,ptewf/J- 'to be dark' : ptewOdya.A non-palatalizedvariantis always
possible,thougha few forms are so lexicalizedthat suchvariantsarerare,e.g.
pf/Jryidyenya'black', yf/Jng~nya
yiqmOnya 'excessive(usedfor the secondten numer-
als)', ty{rtya 'flying; (Eur) bird'. At leastthe verb Xte- 'to depart' has forms
like partic.fut xtewontya: inf. impf xtebya : info perf xteqmya.In the caseof
the negativeverb nyi- 'not' there appearforms like subords3 nyib°tya: ind
sOs3nyidya : interr s3 nyisya : obligative sl nyibcyakedom.Otherverbsmay
occasionallyhavesimilar forms, e.g.pyisyf/Jh-'to laugh' : info impf pyisyOmya.
Certain postpositionsinvariably have a palatal consonantin certain case
suffixes, notably nyi- 'on' : loc nyinya : pros nyimnya, and myu- 'in' : loc
myunya: prosmyumnya.
530 NENETS

Verbal Inflection
The finite inflectional categoriesare mood, tense, conjugation(subjective,
objective, and reflexive), number of object in the objective conjugation,
personof subject, and numberof subject. There are also severalnon-finite
forms.

Mood
Accordingto the presentcount, thereare sixteenmoods.The formation of the
indicative,imperativeproper,andoptativeis basedon variousmorphological
substemsand distinct sets of personal suffixes. The other moods have
characteristicsubstemswith suffixal markers,andthe samepersonalsuffixes
as in the indicative. The moods, exemplified, where possible,by the third
personsingularof the verb nu- 'to stand', are:

I The indicative,e.g. nu '(s)hestands';

G

2 The imperative,comprisedof threesubmoods,which are

(a) the hortative-x!/! in the first person,e.g. sl nu.x¢dom'let me stand';
(b) the imperativeproperin the secondperson,e.g. s2 nuq 'stand!';
(c) the optativein the third person,e.g. s3 nuOya 'let him/herstand';
3 The conjunctive-yi, e.g. nuyi '(s)hewill stand(request)';
4 The necessitative-psu,e.g. nubcu '(s)heshall stand(demand)';
5 The interrogative-sa,e.g. nusa 'did (s)hestand?';
6 The imperfective probabilitative-n-ta-qxe-i!/!, e.g. nunakio '(s)he may
stand';
7 The perfectiveprobabilitative-me-qxe-i!/!, e.g. nuwekio '(s)hemay have
stood';
8 The obligative -psa-qxe-i!/!, e.g. nubcakio '(s)he should stand(expecta-
tion)' ;
9 The imperfective approximative-n-ta-r!/!xa, e.g. nunar¢xa '(s)he seems
to stand';
10 The perfective approximative-me-r!/!xa, e.g. nuwer¢xa '(s)he seemsto
have stood';
11 The futuritive approximative-m!/!nta-r!/!xa, e.g. nuwOntar'xa'(s)heseems
to be going to stand';
12 The superprobabilitative-ma-n-t!/!h-x!/!bya, e.g. nuwanOngk¢bya'(s)he
probablystands';
13 The hyperprobabilitative-r!/!xa-me-i!/!, e.g. nu~xawiOnu~xawiO '(s)he must have
stood';
14 The narrative-me-i!/!, e.g. nuwio '(s)hehasstood';
15 The reputative-m!/!na, e.g. nuwOna '(s)heis supposedto stand';
16 The desiderative-r!/!wa, e.g. nu~wanu~wa '(s)heis encouragedto stand'.

The mood systemcould of coursebe presentedmore hierarchically. Many

moods have complex markers consisting of participial and derivational
 NENETS 531

suffixes, and their morphological behaviour depends on the individual

suffixes. For instance, the choice of the suffix-initial consonantin the
imperfectiveprobabilitativeand approximativefollows that of the imperfec-
tive participle. The superprobabilitativemood, in tum, consistsof the dative
of the imperfectiveinfinitive and the noun x¢bya 'sign; sense'.However, as
long as the formations in question are, on the one hand, conjugatedin all
conjugations,and do not, on the other hand, possessnon-finite forms, their
statusas moodsis not in question.The habitive, regardedas a derivative in
this description,may dialectally behavelike a mood. Thereis also a peculiar
constructionwith a postverbalnegativeverb plus a clitic particle that could
be regardedas a compoundmood,e.g. nga:q nYI-woh 'it certainlyis', maqnYI-
wah '(s)he certainly said' (also notice the difference from the connegative
forms ngaqandmanOq).
Tense
The inflectional categoryof tensecomprisestwo tenses,the aorist and the
preterite.While the aorist has no marking, the preteriteis expressedby the
suffixation of -syj.'} after the personalsuffixes, e.g. nu- 'to stand' : aor s1
nu¢dom : s2 nu¢n°: s3 nuo 'I : you: he stand(s)' : pret sl nuOd¢mcyo: s2
nuOn¢syO: s3 nu¢syO'I : you: he stood'.Despitethe morphotacticpeculiarity
of preterite suffixation, there is no doubt about its true inflectional status.
While the category of tense exists in conjunction with the indicative,
conjunctive,and narrative,it doesnot appearin the imperative,interrogative,
or necessitative,andis marginalin the othermoods.
All verbs are divided into two groups with regard to their temporal
relations.For momentaneous verbs,the indicative aoristexpressesimmediate
past, and the indicative preteriteexpressesmore remotepast.For continuous
verbs, the indicative aorist expressespresent, and the indicative preterite
expressessimple past. In the conjunctive, the aorist expressesconditional
future, and the preterite expressesconditional past. In the narrative, the
oppositionis basicallyperfectv. pluperfect.
The pastexpressedby the indicativepreteritealwaysrefersto the speaker's
personalexperienceof the action. For the expressionof an action which was
not observedbut the resultsof which are still observable,the narrativemood
is used.
For the expressionof non-conditionalfuturity, a particular derivative is
used,as explainedbelow.
The future derivativeco-occurswith inflectional tense,e.g. lad¢- 'to beat'
: fut sOsl pret ladOngkuw¢syO'I was going to beathim'.
Conjugation
Verbsbelongto one of four'conjugationalgroups.

1 Intransitiveverbshaveonly the subjectiveconjugation.

2 Transitiveverbshaveboth the subjectiveandobjectiveconjugation.
532 NENETS

3 Reflexiveverbshaveonly thereflexive conjugation.

4 Transitive/reflexiveverbshaveall threeconjugations.

The category of conjugation is connectedwith four sets of personal

suffixes.

1 The first setis usedin the subjectiveconjugation.

2 The secondset is usedin the objectiveconjugationwhen the object is in
the singular.
3 The third set isusedin the objectiveconjugationwhenthe objectis in the
dual or plural.
4 The fourth setis usedin the reflexive conjugation.

Morphological Substems

GeneralFinite Stem
1 The vowel stems,andthe mixed andirregular stemsin the optativeof the
subjectiveconjugation,add -0, exceptbefore a suffix with an initial x,
where -nga is added,e.g. to- 'to come' : s3 too: d3 tongaxoh,yilye- 'to
live' : yilyeO: yilyengaxoh.
2 The mixed stems,exceptin the optative of the subjectiveconjugation,
changetheir final vowel into i or u, occasionally accompaniedby a
changeof palatalityof the precedingconsonant,e.g. nyen¢- 'to be angry'
: s3 nyeni : d3 nyenix'h,pcew¢- 'to be dark' : s3 pcebyi, ng¢dy¢- 'to be
visible' : ng¢dyi, yak¢- 'to itch' : yaku, ngeso- 'to camp' : ngesi,xonyo-
'to sleep' : xoni, l¢bcyo- 'to stick together' : l¢bcyi, yangko- 'to lack' :
yangku,pyiryencyo- 'to do cooking' : pyiryencyu(notice toxo- 'to learn'
: tuxu, yoxo- 'to disappear': yuxu).
3 The consonantstemsadd -nga, e.g. pce¢r- 'to do' : sOs3pceorngada.In
m-stems,the ng of the suffix is lost, e.g. ng¢m- 'to eat' : ng¢mada.The
verb mah- 'to say' exhibits an irregular vowel stem, viz. s3 ma, but d3
mang~h(- m~h).
mang~h ng~h.
4 The irregularstems,exceptin the optativeexhibit partial suppletion,viz.
xce- 'to depart' : s3 x¢ya
ng~h. : d3 x¢ya~h, ng~h. ngce- 'to be' : nga : ng~h.ng~h. The
negativeverb is alsoexceptional,viz. nyi- 'not' : ny{: ny{~h. ng~h.

Dual ObjectSubstems
The dual objectsubstemsareformed by adding-x0Yu to the basicstemin the
imperativeproper, the generalfinite stemin the indicative and optative, and
the modal substemsin other moods, e.g. xada- 'to kill' : imp dOs2
xadax¢yun° q :ind dOs3xadangaxOyuda: interr dOs3xadasaxOyuda,(a mixed
stem)tamp¢- 'to be giving' : ind dOs3 tampyix¢yuda.
 NENETS 533

SpecialFinite Stem
1 The ¢-stemschangetheir final vowel into i and add -S'!, exceptbefore a
suffix with an initial x (present, incidentally, only in the reflexive
conjugation,so that this variantdoesnot exist for transitiveverbs),where
-yS'! is added,e.g. yurk¢- 'to standup' : refl s3 yurkiOq : d3 yur~y¢xoh. yur~y¢xoh.
2 The other vowel and all consonantstemssimply add -YS'!, e.g. peda- 'to
be tired' : refl s3 pedeyOq,s¢l- 'to return' : s¢ly-oq.
3 For the mixed stems,the generalfinite stemis usedinstead,e.g. tamp¢-
'to be giving' : pOs2tampyidO'you aregiving them' (cf. sO tampyi,o).

SpecialModal Substems
The special modal substerns correspondto the special finite stem of the
indicative and optative. They are formed from modal substemsmostly by a
final vowel change,as exemplifiedby the pOs2forms of xada- 'to kill'.

1 The interrogative:a ~ ¢ - y¢ e.g. xadas¢do- xadasy¢dO'did you kill

them?'(cf. sOs2xadasa,o).
2 In the objective conjugation the probabilitatives, obligative, hyper-
probabilitative,and narrative: e-iS'! ~ i¢, e.g. narr xadawi¢do'you have
killed them' (cf. xadawe,o).
3 The approximatives,superprobabilitative,reputative,and desiderative:a
~ i, e.g. approximativeimpf xadana,oxido'you appearto kill them' (cf.
xadana,oxa,o): superprobxadawanOngk¢byidO'you probablykill them'
(cf. xadawanOngk¢bya,o).
4 The hortative, conjunctive, and necessitative,and in the reflexive
conjugation,the moodslisted in (2): no changefrom the modal substem,
e.g. conj pOs2 xadayidO 'you will kill them' (cf. sOs2xadayi,o), s¢na-
'to jump' narr refl s2 s¢naweno:s3 s¢nawiOq.

Personand Number

Indicative
In the subjectiveconjugation,the first set of personalsuffixes is attachedto
the generalfinite stem, e.g. (a transitive/reflexiveverb) yemp¢q-'to dress' :
sl (xtbyax¢wam)yempOqngadOm'I dressed(somebody),: s2 yempOqngan° :
s3 yempOqnga: dl yempOqnganyih: d2 yempOqngadyih: d3 yempOqngaxoh:
pI yempOqngawaq : p2 yempOqngadaq : p3 yempOqngaq.
In the objective conjugation,the secondsetof personalsuffixes is attached
to the general finite stem, e.g. sOsl yempOqngawO'I dressedhim' : s2
yempOqnga,o:s3 yempOqngada: dl yempOqngamyih:d2 yempOqngaryih:d3
yempoqngadyih: pI yempoqngawaq: p2 yempoqngaraq: p3 yempoqngadoh.
The third set, with a dual object, is attachedto a dual object substem,e.g.
dOsl yempOqnga~yun°
yempOqnga~yun° 'I dressedthem (two)' : s2 yempOqngaxOyudO: s3
yempOqngaxOyuda: dl yempOqngaxOyunyih: d2 yempOqngaxOyudyih: d3
534 NENETS

yempOqngaxOyudyih: pI yempOqngaxOyunaq : p2 yempOqngaxOyudaq: p3

yempoqngaxOyudoh.
In the objective conjugation with a plural object, the third set and in
the reflexive conjugation the fourth set of personal suffixes are attached
to the special finite stem of the vowel and consonantstems, e.g. pOsI
yempOqyl/Jn°'I dressedthem (many)' : s2 yempOqyl/Jdo: s3 yempOqyl/Jda
: dl yempOqyl/Jnyih: d2 yempOqyf/Jdyih: d3 yempOqyl/Jdyih: pI yempOqy¢naq
: p2 yempOqyf/Jdaq: p3 yempOqYl/Jdoh: refl.sl yempOqy¢wOq'I got dressed'
: s2 yempOqy¢n°: s3 yempl/JqYOq: dl yempOqy¢nyih: d2 yempOqYl/Jdyih
: d3 yempOqy¢~h
yempOqy¢naq: pI yempOqy¢naq: p2 yempOqyf/Jdaq: p3 yempOqy¢dOq,
and to the general finite stem of the mixed stems, for which no special
finite stemexists.

ImperativeProper
The imp.s2 is identical with the connegative, e.g. yempl/Jq- 'to dress'
(x{byaxl/Jwa) yempl/JsOq 'dress (somebody),. The other s2 forms exhibit
peculiar suffixes,attachedto the basic stem,viz. sO -till, dpO -n-lIlq, and refl
-t-lIlq (the latter two requiring vowel stemformation), e.g. sO yemp¢f' 'dress
him' : dO yempokl/JyunOq'dressthem (two)' : pO yempOs¢nOq'dressthem
(many)' : refl yemposf/JdOq'get dressed'.The d2 and p2 forms arereplacedby
the respectiveindicativeforms. The indicativeforms of the stemnyo- perform
the imperative function of nyi- 'not', e.g. nyon°tuq 'do not come', nyor'
xadaq 'do not kill it'.

Optative
There are peculiaroptative setsof personalsuffixes distinct from thoseused
in the other moods. The first and secondsets are usually attachedto the

Table 17.2 Setsof personalsuffixes in the indicative andmostothermoods

Subjective sO dpO Reflexive

sl -t-¢m -m-¢ -n-¢ -m-¢q

s2 -n-t¢ -r-¢ -t-¢ -n-t¢
s3 0 -t-a -t-a -q
d1 -n-yih -m-yih -n-yih -n-yih
d2 -t-yih -r-yih -t-yih -t-yih
d3 -x¢h -t-yih -t-yih -x¢h
pI -m-aq -m-aq -n-aq -n-aq
p2 -t-aq -r-aq -t-aq -t-aq
p3 -q -t-oh -t-oh -t-¢q

Note: (Morpho)phonologicalprocessesnot executed.

The sl suffix is -m-mphratherthan -t-prn in the Europeandialects.
 NENETS 535

generalfinite stem, e.g. yilye- 'to live' : s3 yilyeOya, pce¢r- 'to do' : sOs3
pceorngamta;notice mah- 'to say' : s3 mangaya(cf. the irregular indicative
ma). The exceptionare the mixed and irregular stemswhen the first set is
attached: they appear as vowel stems presentinga similar variant of the
general finite stem, e.g. syu~mp¢- 'to run' : s3 syu~mp¢Oya (cf. ind
syu~mpyi), xonyo- 'to sleep' : s3 xonyoOya(cf. ind xoni), xce- 'to depart' :
xceOya(cf. x¢ya), ngce- 'to be' : ngceOya(cf. nga). The stemnyo- is usedalso
in the negativeoptative,e.g. nyoOyatuq 'let him not come'.The third and the
fourth setsare attachedasin the indicative.
The setsof personalsuffixes in the optative are shown below; (morpho-)
phonologicalprocessesarenot executed.

Subjective sO dpO Reflexive

s3 -ya -m-t-a -tjll-m-t-a -m-t-jIlq

d3 -ya-xjllh -m-t-yih -tjll-m-t-yih -Xjll-m-t-jIlq
p3 -ya-q -m-t-oh -tjll-m-t-oh -tjll-m-t-jIlq

The dpO forms are not used in the Siberian dialects, the corresponding
indicativeforms beingusedinstead.

OtherMoods
The samesetsof personalsuffixes as in the indicative are attachedto modal
substems,exemplifiedin the list of moods.

Non-finite Forms
There are two infinitives (imperfective and perfective), four participles
(imperfective,perfective,negative,futuritive), two gerunds(modal andfinal),
three subordinates(the subordinative,the auditive, and the evasive),and a
connegative.There are no verbal nouns of the actio or actor type, but the
infinitives and participles fulfil their function as well. The infinitives and
participles are, nevertheless,verbal inflectional forms rather than deverbal
nominal derivatives, since they take normal verbal qualifiers such as
accusativeobject. The infinitives and participles are inflected like nouns
except that they lack predicative forms. The gerunds are not further
declinable.The subordinateshavean absoluteform andpossessive forms with
the oblique singular co-affix -n; it seemsthat the auditive may be further
inflectedin tense.

Infinitives
The imperfectiveinfinitive hasthe suffix -ma, e.g. nu- 'to stand' : nuwa. The
perfective infinitive has the suffix -qma, which for the vowel and mixed
stems is simply added to the stem, e.g. nu- 'to stand' : nuqma. For the
consonantstems-0 is addedfirst, e.g. pce¢r- 'to do' : pceoroqma.The unique
536 NENETS

glide stem myi- 'to prepare'shows the suffix variant -ye before -qma, viz.
myiyeqma.

Participles
The imperfectiveparticiplehasthe variablesuffix -n-ta. The vowel stemsuse
-na, e.g. nu- 'to stand' : nuna, exceptto- 'to come' : toda (also tona). Both
the consonantand the mixed stemshave -ta, e.g. pcer;r- 'to do' : pceortya,
xonyo- 'to sleep' : xonyoda. Of the irregular stems,xce- 'to depart' prefers
xcena,while ngce- 'to be' showsngceda.The negativeverb nyi- 'not' always
hasnyinya.
The perfectiveparticiplehasthe suffix -me-i~,-m~tawa(y~),e.g. nu- 'to stand': nuwio.
The negative participlehas the suffix -m~tawa(y~),-m~tawa(y~), e.g. nu- 'to stand'
nuwodaweyo.
The futuritive participle has the suffix -m~nta, -m~nta, e.g. nu- 'to stand'
nuwOnta.

Gerunds
The modal gerund has the suffix -sy~ -sy~ for the consonant stems and
monosyllabicvowel stems,e.g.pcer;r- 'to do' : pcer;rcyo,nu- 'to stand': nusyo,
andthe suffixfor -~ for the polysyllabic vowel stems,including the mixed stems,
e.g. yilye- 'to live' : yilyeO, xonyo- 'to sleep': xonyoo.In the Westerndialects,
however,the suffix -sy~theis invariably used.
The final gerundhasthe suffix -m~nsy~,-m~nsy~, e.g. nu- 'to stand' : nuwr;ncyo 'in
orderto stand'.

Subordinates
The subordinativehas either the suffixsuffix
-p~q or the suffix combination-p~q­-p~q­
na, e.g. nu- 'to stand': s3 nubOtaor nubOqnanta'if/when he stands'.
The auditive has the suffix variants-manohand -moh, e.g. ye- 'to ache' :
s3 yewanonta- yewonta'it feels like it aches'.
The evasivehas the suffix -moh followed by the ablative ablative
-x~-t~, e.g. nu-
'to stand': s3 nuwongkr;dOnta'lesthe stands'.

Connegative
The connegative,used with the negativeverbs, notably nyi- 'not', has the
suffix -q. The s2 imperativeis formally identical.

1 Vowel stemsgenerallydo not require anything else, e.g. yilye- 'to live'
: yilyeq, but notice the e-i-stemme- 'to be' : miq, and the irregular to-
'to come' : tuq.
2 Mixed stems have their final vowel changed into u, the preceding
consonantbeingpalatalif eitheror both of the basicstemandthe general
finite stemhave a palatal consonantbefore the final vowel, e.g. nyenr;-
'to be angry' : s3 nyeni : conneg.nyenuq,pcewr;- 'to be dark' : pcebyi :
 NENETS 537

pfRbyuq, ngeso- 'to camp' : ngesi : ngesuq, xonyo- 'to sleep' : xoni :
xonyuq (noticealso toxo- 'to learn' : tuxuq, yoxo- 'to disappear': yuxuq).
3 The irregular stemsexhibit decidedirregularities,viz. XfR- 'to depart' :
xanyOq, ngfR- 'to be' : ngaq.
4 Consonantstemshavethe regularvowel stemformation, dictatedby the
shape of the suffix, as the only complication, e.g. ngf/Jm- 'to eat'
ngf/JwOq.

Nominal Inflection
The nominal inflectional categoriesare number,case,declension(absolute,
possessive,and predestinative),and, in the non-absolute declensions,
person
and numberof possessoror predestinator.Besidesnominal declension,there
are alsopredicativeforms of nouns,or the nominalconjugation.
Number and Case
There are three numbers,singular, dual, and plural. Of the sevencases,the
grammaticalcases(nominative, accusative,and genitive) combine with all
three numbers, while the local cases (dative, locative, ablative, and pro-
secutive),appearonly in singular and plural, the missing local dual forms
being replaced by expressionswith the correspondingcase forms of the
postpositionnya- 'at'. No morphological processtakes place in the nom-
inative singular, e.g. myaq 'tent'. Someforms have simple suffixes, e.g. sA
myadom: sG myadoh: sProsmyaqmOna: dN mya/{'h =dA =dG: pN myadOq.
Most local forms exhibit a systemof multiple suffixation, e.g. sD myadoq
mya~h :
sLoc mya/{'na : sAbl myakf/Jdo : pD mya/{'q: pLoc mya/{'qna: pAbl myakf/J~.
myadoq
The accusativeplural stemis usedfor the rest, e.g. pA myado : pG myadoq
: pProsmyadoqmf/Jna. However,the prosecutiveplural of monosyllabicvowel
stemsis basedeitherinvariably on the basicstem,e.g.pya 'tree' : pyaqmOna
(not *pyiqmOna), Ii 'bone' : leqmOna (not *liqmona), or variably on both
stems,e.g.ya 'earth' :yaqmOna - yoqmOna.
Accusative Plural Stem
The vowel stemseither require no morphologicalprocessor have their final
vowel changed. Consonantand glide stems attach a vowel to the final
consonantor glide.
The vowel stemsrequiring no changein the form of the accusativeplural
include (1) most monosyllabicstems,e.g. to 'lake' : to, nyu 'child' : nyu; (2)
somef/J-stems,e.g. SfRWo 'eye' : SfRWo; (3) i- and u-stems, e.g. ngesi 'camp'
: ngesi, suyu 'calf: suyu.
Othervowel stemsshow variousvowel changes:

1 monosyllabica- ande- i-stems:

a --7 0, ya --7 yi, e-i --7 i, e.g.ya 'earth' : yo, nya 'friend' : nyi, ti 'reindeer'
: ti(: poss.sNs3teda: pN tida);
2 somef/J-stems:
538 NENETS

Table 17.3 Case,number, and possessivesuffixes

Declension Personofpossessor
Absolute Possessive I 2 3

Singular
Nominative 0 0- ~ -m- -r- -t-
Accusative -m 0- ~ -m- -m-t- -m-t-
Genitive -h 0- ~ -n- -h-t- -h-t-
Dative -n-t0-h -X0- ~ -n- -h-t- -h-t-
Locative -x0-na -x0-na- ~ -n- -h-t- -h-t-
Ablative -x0-t0 -x0-t0- ~ -n- -h-t- -h-t-
Prosecutive -m0na -m0na- ~ -n- -h-t- -h-t-
Dual
Nominative -x0h -X0yu- ~ -n- -t--t- -t--t-
Accusative -x0h -x0yu- ~ -n- -t--t- -t--t-
Genitive -x0h -X0yu- ~ -n- -q-t- -q-t-
Plural
Nominative -q ¥- ~ -n- -t- -t-
Accusative ¥ ¥- ~ -n- -t- -t-
Genitive ¥-q ¥-q- ~ -n- -q-t- -q-t-
Dative -x0-q -x0-q- ~ -n- -q-t- -q-t-
Locative -x0-q-na -x0-q-na- ~ -n- -q-t- -q-t-
Ablative -x0-q-t0 -x0-q-t0- ~ -n- -q-t- -q-t-
Prosecutive (¥)-q-m0na (¥)-q-m0na- ~ -n- -q-t- -q-t-
J, J, J,
Numberofpossessor
Singular -0 -0 -a
Dual -yih -yih -yih
Plural -oh -q-t- -oh

Note: (Morpho)phonologicalprocessesnot executed;preconsonantal h representsan

archiphonemefor nandng, i.e. -h-t- yields nt as -q-t- yields t andpostvocalic-t-d; the
formation of the accusativeplural stemsymbolizedby ¥.

~ ~ 0, y~~ yo, ~ ~ ye, e.g. x¢,o 'knife' : x¢ro, nyanyO'bread': nyanyo,

syun°'steam': syunye;
3 somea- stems:
a ~ ¢, ya ~ ¢, ya ~ Y¢, e.g.xoba 'fur' : xobo(noticexasawa'man,male'
: xasyewO),tyonya 'fox' : tyon°,yesya'iron, money' : yesyO;
4 othera- stems:
a ~ i, a ~ yi, ya ~ yi, (two words only) ya ~ e, e.g. ngaw°ka'pet
reindeer' : ngawoki, nguda 'hand' : ngudyi (notice y¢xa 'river' : yesyi),
ngodya'berry' : ngodyi,xalya 'fish' : xale (andyalya 'day' : yale);
 NENETS 539

5 e- and0- stems:
e ~ i, 0 ~ u, e.g. yake 'smoke': yaki, ng¢no 'boat' : ng¢nu;
6 alternatingstems:
e-ifj ~ i¢, o-ufj ~ u¢, 0-yfj ~ )i¢, e.g. p¢nio 'dress': p¢nio (: posssNs3
p¢neda : pNs3 p¢niOda), tyukuo 'this' : tyukuo (: tyukoda : tyuku°da),
ngopoyO'the one' : ngopoyO(: ngopoda:ngopoyOda).

Dialectally, i- and u-stemsmay also exhibit vowel change,yielding pA

ngesi,suyu.
Consonantstemsadd -0 or -ye, e.g. myaq 'tent' : myado,nyum 'name' :
nyubye.Glide stems followthe consonantstems,viz. xa 'ear' : xawo,yi 'wits'
: yibye, syo 'throat' : syoyo,x¢byi 'Khanty; servant': x¢byiye.Notice that the
baseused for ng¢mke 'what' is in this caseng¢m-, yielding pA ng¢wo. At
leastfor mostfunctions,the dl possessiveforms take over the sl forms in the
Siberiandialects.

Possessive Declension
Thereareforms for threepersonsand threenumbersof the possessor for each
absoluteform. They are formed through complex suffixation, with number
and case suffixes partly different from the absolute ones. The accusative
plural stem is used in the nominative plural, too. For example: ya 'earth' :
sNsl yawo : s2 yar' : s3 yada : dl yamyih : d2 yaryih : d3 yadyih : pI yawaq
: p2 yaraq : p3 yadoh; sAs3 yamta : sG yanta : sD y~nta : sLoc y~nanta
yaxOyuta
: sAbl ya~ d¢nta : sProsyawOnanta: nom.du yaxOyuda=dA : dG yaxOyuta
: pN yoda =pA : gen.pluryota : pD y~ta : pLoc yaxOqnata: Sabl yaxOyuta
y~t¢ta
: pProsyaqmOnata- yoqmOnata.

PredestinativeDeclension
There are forms for threepersonsand three numbersof the predestinatorfor
each singular grammaticalcaseform of the absolutedeclension.They are
formed by suffixing -tfj, followed by the respectivepossessivesuffixes. For
example,x¢r' 'knife' : nom s3 x¢r'd¢da : acc x¢r'd¢mta : gen x¢r'd¢nta 'a
knife for him' .

Nominal Conjugation
When predicates,absoluteforms areconjugatedfor person,andboth absolute
and possessiveforms are conjugatedfor tense.In the aorist, the third-person
forms of thesepredicativeforms of nouns coincide with the corresponding
nominativeforms of declension.For example,nye 'woman' : predic aor sl
nyedOm : s2 nyenO : s3 nye 'I am : you are : she is a woman' : pret sl
nyed¢mcy<}: s2 nyen¢syO: s3 nyesyO'I was: you were: she was a woman',
nya 'friend' : posssl aor sg nyawO : pret nyaw¢syO'he is : was my friend' :
pI aor nyin°: pret nyin¢syO'they are : were my friends'.
540 NENETS

Table 17.4 The inflection of the personal pronouns

1 2 3

Nominative
sl m!/lnyo s2 pid!/lro s3 pida
dl m!/lnyih d2 pidOryih d3 pidyih
pI m!/lnyaq p2 pidOraq p3 pidoh
Accusative
sl syiqmO s2 syitO s3 syita (- syitya)
dl syidOnyih d2 syidOdyih d3 syidOdyih
pI syidOnaq p2 syidOdaq p3 syidOdoh
Genitive
sl syiqn° s2 syitO s3 syita (- syitya)
dl syidOqnyih d2 syidOtyih d3 syidOtyih
pI syidOqnaq p2 syidOtaq p3 syidOtoh

Personal Pronouns
The s1 variants acc syiqmyih : gen syiqnyih are common especiallyin the
Easterndialects.The 2/3 N stemhasan Easternvariantwith prjJ- and a further
Far Eastern variant with pu- instead of pi-. That the forms have inner
morphological structureis seenin derivation, e.g. mrjJnyOryinaq 'only we',
pidOryidoh 'only they'.The local caseforms are takenover by the possessive
forms of the postpositionnya- 'at'. The nominative formsare usedonly for
emphasis,and can in that role occur also before the accusativeand genitive
forms as well as the postpositionalforms.
Adverbs and Postpositionswith Partial Declension
There are severalgroupsof adverbial stems,eachwith particularcategories
of nominal declension.For thosewith a local function, a specialset of local
casesuffixes exists,viz. dat -h, loc -na, abl -to, and pros -mna, e.g. nyah 'to'
: nyana 'at' : nyadO 'from' : nyamna'along, about',nyih 'on to' : nyinya 'on'
: nyidO 'off' : nyimnya'over'.
Local postpositionslike nya- 'at', nyi- 'on', myu- 'in', ngilrjJ- 'under',
tyaxrjJ- 'behind',pu- 'after', yeq- 'towards',xi- 'near',yer- 'in the middle of',
nyerrjJ- 'before' have both absolute case forms and a full possessive
declension,formed with the sameco-affix as the nominal singularlocal case
forms, e.g. nyamnanta'aboutit', nyinyantoh'on them'.
A numberof nominal stemsoccurin conjunctionwith nya-, e.g.XteWO-nya-
'beside'.Postpositionalstems often have compoundforms with nya-, e.g.
myu-nya-'inside'.Thereare alsoderivedstemslike nyayurjJ- andnyaku-from
nya- 'at'.
Local adverbshave only absolutecaseforms, e.g. tyuqrjJ- 'up' : tyuqoh :
tyuqona : tyuqrjJdo : tyuqomna, trjJsyi- 'down' : trjJsyih : trjJsyina : trjJsyidO :
 NENETS 541

t¢syimna.This groupincludesseveralcompoundforms with nya-, e.g. t¢-nya-

'there',x¢-nya- 'where',syata-nya-'left', m¢xa-nya-'right'.
Pronominal stems in their local case forms also render an adverbial
meaning, e.g. tyukuo 'this' : tyukox¢na 'here'. Some nominal stems are
lexicalized in their adverbialfunction, but they still exhibit normal nominal
case inflection, occasionallyeven in the plural, e.g. nga- 'far' : ngaxOq :
ngaxoqna :ngaqmOna.
ngax¢~ : ngaqmOna.
Non-local postpositionshave absoluteand possessiveforms but no case
inflection, e.g. xawOna'except': xawOnanta'excepthim' (a petrified nominal
prosecutive).
There are also adverbswith only possessiveforms, often fulfilling the
function of conjunctions of other languages,e.g. ng¢dyObya- 'because':
ng¢dyObyanta'becauseof that'. From the morphologicalpoint of view, the
reflexive pronounx¢r¢q- belongshere,e.g. x¢yDta 'he himself'.

Derivation

DeverbalNouns
1 Local nouns,e.g. xanye- 'to hunt' =:} xanyel¢wa'hunting ground', yo¢r-
'to fish' =:} yoOI¢wa 'fishing hamlet'.
2 Instrumentalnouns,e.g. nglRdalyo- 'to travel' =:} nglRdalyosyOh'travel-
ling sledge',padon¢- 'to be writing' =:} padOn¢bcyoh'pen', yenyer- 'to
shoot' =:} yenyercyOh'gun'.
3 Potentialnouns,usuallyin possessiveforms, e.g. xet¢- 'to tell' =:} xe~yiq xe~yiq
'possibility to tell', yoq- 'to lose' : poss.s3yoqyita 'the possibility of
losing it'.
4 Potentialadjectives,e.g. tlRW¢- 'to reach': tlRwonana'within reach'.
5 Inclinative adjectives,e.g.pyfn¢- 'to be afraid' =:} pyfnOxado'coward'.
6 Other deverbal nouns, e.g. pyirye- 'to boil' =:} pyiryebco 'something
boiled' =:} pyiryebcodO 'something to be boiled'; yilye- 'to live' =:}
yilyebcO 'subsistence'; xanye- 'to hunt' =:} xanyeya'huntingoccupation',
yo¢r- 'to fish' =:} yoOrya 'fishing occupation'.

DenominalVerbs
1 Possessiveverbs, e.g. s¢wa 'cap' =:} s¢byiq- 'to have a cap, to use as a
cap'.
2 Translativeverbs,e.g. ngayD 'largeness'=:} ngar¢m- 'to becomelarger'.
3 Captativeverbs,e.g. noxa 'Arctic fox' =:} nosyih- : conneg.nosyiyOq'to
huntArctic foxes'.
4 Caritive verbs,e.g. myaq 'tent' =:} myacy¢-: s3 myacyi 'to be tentless'.
5 Odorative verbs, e.g. xalya 'fish' =:} xalyay¢- : s3 xalyayi 'to smell of
fish'.
542 NENETS

DenominalAdverbs
1 Caritives,e.g. myaq 'tent' ~ myacyiq'without a tent'.
2 Predestinatives, e.g. ng¢no 'boat' ~ ng¢nodO'a boat forsomeone'.
3 Essives,e.g. Ii 'bone' ~ lengce 'as a bone,for a bone',syidya 'two' ~
syidyangce'both together'.
4 Other denominal adverbs,e.g. sarmyiTt> 'animal, wolf ~ sarmyik¢dO-
ryem - sarmyik¢d°ryewOh'like a wolf; tyet' '4' ~ tyet'l¢d°h'four at a
time'; yuq '10' ~ yucyanO'aboutten'.

DenominalNouns
1 Comitativenouns,e.g. nye 'woman' ~ nyesaweyO'married(man)'.
2 Various adjectives,e.g. wa~wa~'edge,shore' ~ Wal"xiO '(what is) on the
shore';war) ~ wariO 'outermost';limp¢do 'swamp' ~ limpOd¢lyangkO
'paludified'.
3 Ordinal numerals,e.g. nya~r 'three' ~ nyaxoromteyo'third', tyet' 'four'
'three'
~ tyetyimtyeyo'fourth'.
4 Relationalnouns,for semanticreasonsnot usedin the singular,e.g. nya
'friend' ~ du nyas¢xOh : pI nyasOq 'friends (to each other)'; ny{sya
'father' ~ N p3 ny{syan¢doh'their (respective)fathers'.

DeverbalVerbs
1 Futureverbs,with an incompleteparadigm,e.g. (vowel stemsandmixed
¢-stems)me- 'to be' ~ mengko-: s3 mengku'is going to be', nyen¢- 'to
be angry' ~ nyenOngku-;(mixed o-stemsand consonantstems)xonyo-
'to sleep' ~ xonyod¢-, mah- 'to say' ~ mant¢-; notice xce- 'to
depart'~ xanOt¢- (but ngce- 'to be' ~ ngcengko-),to- 'to come' ~ tut¢-,
ta- 'to bring' ~ t¢t¢-.
2 Habitive verbs,alsowith an incompleteparadigm,e.g. tur- 'to come' freq
~ turcyOt¢- : s3 turcyOti 'is in the habit of coming' : conneg.turcy°tuq.
3 Precative verbs, with a fragmentary paradigm, mainly used in the
imperative,e.g. to- 'to come' ~ tox¢r-: imp s2 tox¢~q 'please
'pleasecome'.
4 Intensiveverbs,e.g. t¢nya- 'to exist' ~ t¢nyax¢ya-'to really exist'.
5 Intransitiveverbs,e.g. tola- 'to read' ~ tolangko- 'to do reading',pyirye-
'to cook' ~ pyiryencyo-- pyiryengko-'to do cooking',peh- 'to put' ~
pent¢- 'to do loading'.
6 Transitive verbs, e.g. nyesey¢m-'to change'~ nyeseyOmta-'to change
(tr)', ng¢dyim- 'to appear'~ ng¢dyimtye-'to bring forth', t¢rp¢- 'to exit'
~ t¢rpora- 'to take out', y¢ngk¢m- 'to separate'~ y¢ngTt>mla- 'to
separate(tr)', tira- 'to dry' ~ tirabta- 'to dry (tr)'.
7 Imperfectiveverbs,e.g.pyi- 'to boil' ~ pyin¢- 'to be boiling', wadyo- 'to
grow' ~ (Western-Central)wadyod¢n¢-, (Eastern) wadyod¢- 'to be
growing'.
8 Durativeverbs,e.g. myiq- 'to give' ~ myip¢- 'to keepgiving' : s3 myipyi
: connegmyipyuq,xada- 'to kill' ~ xadab¢- 'to keepkilling'.
 NENETS 543

9 Frequentativeverbs, e.g. ngf/Jm- 'to eat' => ngf/Jwor- 'to have a meal',
xayo- 'to stay' => xayur- 'to remain'.
10 Iterative verbs, e.g. tyu- 'to enter' => tyungkf/J- 'to enter frequently,
regularly', ngamtf/J- 'to sit down' => ngamfYngkf/J- 'to sit regularly'.
11 Inchoative verbs, e.g. (vowel stems)yilye- 'to live' => yilyel- 'to start
living'; (consonantand mixed stems)pyisyf/Jh- 'to laugh' => pyisyOIf/J- 'to
startlaughing'.
12 Incompletiveverbs,e.g. nu- 'to stand'=> nuy%tye- 'to standfor a while',
ngf/Jwor- 'to havea meal' => ngf/Jworyf/Jbtye- 'to havea snack'.
13 Momentativeverbs,e.g. tesf/J- 'to drip' => tesoxf/Jl- 'to drop'.
14 Passiveverbs,e.g.xada- 'to kill' => xadara- refl 'to get killed'.
Omnibased Derivatives
1 Comparatives,e.g. sf/Jwa 'good' => sf/Jwarka 'better',sf/Jngkowoq- 'to be
heavy' => sf/Jngkowosorka- 'to be heavier'.
2 Moderatives,e.g. ngarka 'big' => ngarkampoyO 'ratherbig'.
3 Augmentatives,e.g. ngarka 'big' => ngarkaqya 'very big'.
4 Diminutives, e.g. sf/Jqla 'moron' => sf/Jqlako 'fool', tuq 'animal fat' =>
tudako 'mushroom',wada 'word' => wadako 'tale'.
5 Pejoratives,e.g. ti 'reindeer'=> tekocya 'poor little reindeer';nyenecyOh
'person,humanbeing' => nyenecyf/Jnye 'poor man'.
6 Limitatives ('only'), e.g. ngf/Jmkeryi 'whatever;thing'.
7 Simulatives('asif), e.g. syunorf/Jxa 'steam-like;blue'.
8 Concessives('even'),e.g.xloyaxf/JrfY 'anybody'.
9 Affirmatives ('indeed'),e.g. xloyaxf/Jwa 'somebody',xadaxf/Jwao 'to kill
indeed'modalgerund.

Syntax
Word Order
The word order is predicate-final.A regular transitive sentenceappearsas
(Time adverbial) Subject noun phrase (Place adverbial) Object noun phrase
(Manner adverbial) Predicate verb. Any focusedconstituentmay be placed
in preverbal position, but otherwise the order is quite rigid; only heavy
emphasismay result in a postverbalconstituent.Notably, questionwords do
not causechangesin the word order. In negative sentences,the two final
word-forms are, in this order, the negativeauxiliary verb and the main verb
in the connegative.Within noun phrases,the attribute always precedesits
head.
Constituent Structure and Agreement
The head of a subject noun phraseis in the nominative. Subject personal
pronouns are used only for emphasis,while the person is expressedby
conjugation. Subjectlessconstructionsinclude sentenceswith the second
544 NENETS

personimperativeand impersonalsentenceswith the verbs tara- and sif/Jr- in

the sense'must'.
The subjectof an embeddedclause,the possessorattribute, and the head
of a postpositionarein the genitive. Personalpronounsin thesefunctions are,
however, in the nominative, and used only for emphasis,while person is
expressedby possessive declension.The genitiveof personalpronounsis used
only in thoserare instanceswhich do not allow possessivedeclensionof the
main word.
The headof an object noun phraseis in the accusative,exceptif the verb
is in the secondperson imperative, when the object is in the nominative.
Personalpronounsare,however,invariably in the accusative.
Within a noun phrase,an attribute never agreeswith its headin case,but
agreementin numberis possible,the choicedependingon the particularfocus
relations. In a special form of agreement,an attribute may duplicate the
possessivesuffix of its head.
A predicateverb or noun agreesin personand numberwith the subject.A
predicateverb also agreesin numberwith the object if it is in the objective
conjugation.The choice of conjugationdependson the focus of the object.
When introducedas new information, the object usually standsimmediately
before the verb, which is then in the subjective conjugation. When non-
focused,the objectmay appearapartfrom the verb or be completelyomitted,
the verb being obligatorily in the objectiveconjugation.
A predicatenoun is followed by a form of the copulangte- 'to be' if and
only if the sentenceis negative,non-indicative,future or habitive.
There are no conjunctions,but subordinationis expressedby subordinate
non-finite forms or infinitives and participles in local case forms and
postpositionalphrases.Simple parataxisoften servesfor co-ordination,but
variousconnectiveadverbsarealso available.Yes-noquestionsareexpressed
(1) by the interrogativemood, when referring to pasttime, or (2) by a special
intonation,when referring to the presentor future time. A few clitic particles
are alsousedfor specialemphasis.

NenetsText
Excerptedfrom Susoi1990: 20.

A: transcription,with segmentationof suffixes; B: morphemologicalglosses,

unmarkedfeaturesnot indicated;C: freer translation

Al nyewoxio nyenecy~y-e-q nyenecy~y-e-q syo-q

B1 ANCIENT PERSON-plur-G song-pN

A2 xurka=ryi song-pN yar~bcOyar~bcO

B2 WHAT.KIND=lim song-pN YARABTS.pA
 NENETS 545

syudObj/lbcO ngj/ldyibyelye=wa-n-t-oh xawona

SYUDBABTS.pA PRESENTS=inf(impf)-sG-p3 BESIDES

nyenecYj/lO-q yilye=wa-n-t-oh yampj/l-nO-h

PERSON-pN LIVES=inf(impf)-sG-3-plur LONG-sD

Xj/lrO-toh yf-xOn-toh syoy-o syerta=bj/l=wiO-q

SELF-p3 MIND-sD-3-plur SONGpA MAKES=dur=narr-p3

A3 ya-h syarO-h nyi-nya ngoyak°-q

B3 EARTH-sG SURFACE-sG ON-Ioc RARE-pN

nyenecyj/l=lyi-q syarO-h
Xj/l~-t-oh syerta=wiO
PERSON-lim-pN SELF-3-plur MAKES=partic(perf)

wadyi-d-oh syo-ngre me=cyo nyf-doh

VVORl).pA-3-plur SONG-ess PERFORMS=ger(mod)NEG=-pO-p3

pyirj/lsO_q syo-ngre syo-m me=cyo

IS.ABLE-conneg me=cyo SONG-sA PERFORMS=ger(mod)

nyi=nya pyirO=ta-q xfbya-q

NEG=partic(impf) IS.ABLE=partic(impf)-pN VVHO-pN

A4 tj/lrcya nyenecyOh ngulyiq tyanyo

B4 SUCH PERSON VERY LITTLE

Cl Traditionalfolk songs.
C2 Besidespresentingvariouskinds of tales(lax<>nako ), lamentrecitatives
(yar¢bcO), andheroic recitatives(syucPb¢bcO), the people,in the courseof
their lives, havemadesongsin their own minds.
C3 On the surfaceof the earth,only few peoplecannotperformwords made
by themselvesas a song- they are the onesunableto performa song.
C4 Suchpeoplearevery few.

Referencesand Further Reading

Major introductions and grammatical treatmentsappearin Castren (1854
[21966]), Tereseenko(1947), Decsy (1966), and Hajdu (1968 [21982]).
Textbooksinclude TereSeenko(1959),Almazova (1961), Barmie and Kupri-
janova (1979), and Kuprijanova and Barmie and Homie (1985). Tereseenko
(1956) and Hajdu (1975) presentwide selectionsof articleson varioustopics.
Among dictionaries,Castren(1855) is the earliest, Lehtisalo (1956) exten-
sive, phonetic,anddialectological,andTereseenko(1965) large andstandard.
546 NENETS

Pyrerka and Terescenko(1948) is the only larger dictionary from another

languageto Nenets. Text collections focusing on folklore include Castren
(1940), Castrenand Lehtisalo (1960), Lehtisalo (1947), Kuprijanova(1965),
Terescenko(1990). For a literary history in NenetsseeSusoj (1990). Large
bibliographieshavebeenpublishedby Hajdu (1968 eI982], 1988).
For thephonology,an earlypublicationis Lehtisalo(1927).A majorwork in
the spirit of generative phonology is Janhunen(1986), with an ample
bibliography.Later articlesinclude Helimski(1989),Janhunen(1993), Salmi-
nen (1990a, 1990b, 1993a, 1993b). Morphology is the focus of all Nenets
grammarsandtextbooks.Honti andZaicz (1970)is a reverselisting of suffixes
andsuffix combinationscompiledon thebasisof Hajdu(1968[21982]).Mikola
(1975) is a thorough survey of the postposition system. Hajdu, Lahadi,
Labanauskas, Perfil'eva,Sebestyen,andScerbakova,amongothers,havebeen
active in publishingarticles,as seenin the the bibliographyby Hajdu (1988).
Salminen(1997a)is a reversedictionary with a key to inflectional paradigms,
and Salminen(1997b)a monographon inflection. The only major publication
on syntaxis Terescenko(1973).

Almazova,A. V. (1961)CaMoYQHTeJIbHeHe~Koro
HeHe~KoMy 513bIKa,Leningrad:Ucpedgiz.
HeHe~KoMy
Barmic, M. Ja. and Kuprijanova, Z. N. (1979) IIpaKTlIKyM no HeHe~KoMy HeHe~KoMy 513bIKy,
Leningrad:Prosvescenie.
Castren,M. A. (1854 e1966]) Grammatik der samojedischenSprachen,Hrsg. A.
Schiefner, St Petersburg[Indiana University Uralic and Altaic Series 53, Bloo-
mington: IndianaUniversity-TheHague:Mouton].
- - (1855) Worterverzeichnisseaus den samojedischenSprachen, Hrsg. A.
Schiefner,StPetersburg.
- - (1940) SamojedischeVolksdichtung,ed. T. Lehtisalo, MSFOu 83, Helsinki:
SocieteFinno-Ougrienne.
Castren, M. A. and Lehtisalo, T. (1960) SamojedischeSprachmaterialien,in T.
Lehtisalo(ed.),MSFOu 122,Helsinki: SocieteFinno-Ougrienne.
Decsy,G. (1966) Yurak Chrestomathy,IndianaUniversity Uralic andAltaic Series50,
Bloomington:IndianaUnIversity-TheHague:Mouton.
Hajdu,P. (1968 [21982]) ChrestomathiaSamoiedica,Budapest:Tankonyvkiad6.
- - (1975) SamojedologischeSchriften, Studia Uralo-Altaica 6, Szeged:J6zsef
Attila Tudomanyegyetem.
- - (1988) 'Die samojedischen Sprachen',in D. Sinor(ed.),The Uralic Languages:
Description, History and Foreign Influences, Handbuch der Orientalistik 811
Leiden: Brill, pp. 3-40.
Helimski, E. A. (1989) TJIy6HHHO-q,oHOJIOrnQeCKHHH30CHJIJIa6H3M HeHe~Koro Moscow:
CTlIxa' ,JSFOu82, 223-68.
Honti, L. andZaicz,G. (1970) 'Jurakatergotoldalektar', NyelvtudomdnyiKozlemenyek
72,363-98.
Janhunen,J. (1986) Glottal Stop in Nenets,MSFOu 196, Helsinki: SocieteFinno-
Ougrienne.
--(1993)'Optionsfor TundraNenetsvowel analysis', inM. Sz.Bakr6-NagyandE.
Szfj (eds), Hajdu Peter 70 eves, Budapest: MTA Nyelvtudomanyi Intezet,
pp. 143-7.
Kuprij anova,Z. N. (1965)3nHQeCKHeneCHH Moscow:
HeH~eB, Moscow: Nauka.
 NENETS 547

Kuprijanova, Z. N., Bannie, M. Ja., and Homie, L. V. (1985) HeHe~KHH HeHe~KHH }I3hlK,
Leningrad:Prosveseenie.
Lehtisalo, T. (1927) Ober den vokalismusder ersten silbe im juraksamojedischen,
MSFOu56, Helsinki: SocieteFinno-Ougrienne.
- - (1947)luraksamojedische Volksdichtung,MSFOu90, Helsinki: SocieteFinno-
Ougrienne.
- - (1956) luraksamojedisches Worterbuch,Lexica SocietatisFenno-U gricae 13,
Helsinki: SocieteFinno-Ougrienne.
Mikola, T. (1975) Die alten Postpositionendes Nenzischen(luraksamojedischen),
Budapest:Akademiaikiad6.
Pyrerka, A. P. and Terescenko,N. M. (1948) PyccKo-HeHe~KHH NenzischenCJIOBapb, Moscow:
Ogiz-Gis.
Salminen, T. (1990a) 'Samoyedologyin Finland 1985-1989: the glottal stop', in
ProblemsofUralistics, vol. I, Moscow: Instituteof Scientific Informationon Social
Sciences,USSRAcademyof Sciences,pp. 216-37.
- - - (I 990b) 'Phonologicalcriteria in the classificationof the Nenetsdialects', in L.
Kereszteset al. (eds),C7IFU, vol. mc, Debrecen,pp. 344-9.
- - (1993a) 'On identifying basic vowel distinctionsin TundraNenets',FUF 51,
pp.I77-87.
- - - (1993b) 'A phonemizationof Tundra Nenetslong vowels', in M. Sz. Bakr6-
Nagy and E. Szfj (eds), Hajdu Peter 70 eves,Budapest,NyelvtudomanyiIntezet,
pp.347-52.
- - - (1997a) A Morphological Dictionary of Tundra Nenets, Lexica Societatis
Fenno-Ugricae26, Helsinki: SocieteFinno-Ougrienne.
- - (1997b) Tundra Nenets Inflection, MSFOu 227, Helsinki: Societe Finno-
Ougrienne.
Susoi,E. G. (1990)HeH3~He"
HeH3~He" JIHTepaTypa,Leningrad:Prosveseenie.
Tereseenko,N.M. (1947) OqepK rpaMMaTHKH HeHe~KoroHeHe~Koro(JOpaKo-caMoe~cKoro) (JOpaKo-caMoe~cKoro)
}I3blKa, Leningrad:Uepedgiz.
- - - (1956) MaTepHaJIblH HCCJIe~OBaHH}IHCCJIe~OBaHH}IllO }I3hlKY HeH~eB,HeH~eB, Moscow-Leningrad:
AkademijaNaukSSSR.
- - - (1959) B llOMO~bHeH3~He"caMOCTO}lTeJIhHOH3yqaJO~HMH3yqaJO~HM HeHe~KHHHeHe~KHH }I3bIK, Leningrad:
UCpedgiz.
- - - (1965)HeHe~Ko-pyccKHH CJIOBaph,Moscow: SovetskajaEnciklopedija.
HeHe~Ko-pyccKHH
- - - (1973) CHHTaKcHccaMo~HHcKHX caMo~HHcKHX}I3hlKOB: llpocToellpe~JImKeHHe,llpe~JImKeHHe, Leningrad:
Nauka.
- - - (1990)HeHe~KHH 3110C,3110C,Leningrad:Nauka.
18 Selkup
EugeneHelimski

The Selkup language,formerly known also as Ostyak Samoyed(ic),is now

spokenby c.2,000 descendantsof an indigenouspopulationof a vast West
Siberiantaiga arealocatedmainly betweenthe Ob' and Yenisei rivers. The
total numberof Selkupsaccordingto the Soviet censusesof 1979 and 1989
was the same,about3,600persons,but the level of native-languageretention,
and the proficiency of its speakers,are both steadily decreasingdue to
linguistic assimilation:in almostall partsof their original ethnic territory the
Selkupsarenow outnumberedby Russians.Beyondthis territory, thereareno
sizeablegroupsof Selkupresettlers.
Selkup is usedin family life and in traditional domainsof the economy
suchas fishing andhunting,but not in administrationor otherofficial spheres.
Severalattemptsto introduceit as the languageof educationin elementary
schoolsand to develop Selkup literacy, madeespeciallybetweenthe 1930s
and 1950s, have never been supportedby any systematicefforts. Having
recently been reintroducedin several schools, Selkup found itself in the
position of a secondand foreign languageas a result of low native-language
proficiency among schoolchildren. The activity of the national cultural
society Qatt;;, Qup ('Man of the [Ob'] River') since the late 1980shas been
a help in preservingSelkup national identity (see Pusztay: 1992), but the
chancesfor the retentionof Selkupby coming generationsare closeto zero
for SouthernSelkups(in theTomskregion) andnot very high in the North (the
Tyumenand Krasnoyarskregions).

Bilingualism
Selkup-Russianbilingualism, with varying degreesof competencein the
native and second languages,and, typically, with a transitory stage of
semilingualism,in which the commandof neither languageis perfect, is
universally spreadamong Selkup speakersin the South (where in younger
generationsmost ethnic Selkups are Russian monolinguals) and at least
amongyoungerand middle-agedNorthernSelkups(herethereare still some
Selkupmonolingualsamongolderpeople,especiallywomen).
While this type of bilingualism is unilateral and leads ultimately to
Russification, there exists also, now mainly among the people connected

548
 SELKUP 549

with traditional occupations,bilateral bilingualism on both sides of the

borderlinesof the Selkupethnic territory: Selkup-Khantyin the river basins
of Vasyuganand Vakh, Selkup-Ketin the basin of Yelogui, a left tributary
of the Yenisei River, Selkup-Tungus in the Krasnoyarskregion and the
adjacentparts of the Taz river basin. The reindeer-breedingSelkupsin the
middle part of the Taz river basin usually speakboth Selkup and Nenets,
and there are cases of Selkup-Turkic (Chulym Turkic, Siberian Tatar)
bilingualismin the South.
Up to the beginningof the twentiethcentury Selkupusedto be spokenas
a lingua franca, widely known by non-Selkups,in the northeasternparts of
West Siberia, and this role, played through centuries,may have contributed
to the relative simplicity of its grammaticalsystemin comparisonwith the
otherSamoyediclanguages.

Dialects
The dialectsof Selkupform a chain,originally continuous,but now no longer
unbroken.The differencesbetweenany adjacentidioms are minimal, but for
example the Taz dialect (North) and the Middle Ob' dialect (South) are
mutually incomprehensible,differing approximatelyas much as Russianand
Polishor Udmurt andKomi.
The main dialectgroupsare:

1 Northern (Taz), with the dialectsof Middle Taz, UpperTaz, Baikha-and-

Turukhan,Karasino,Yelogui (the first two are spokenin the easternpart
of the Yamal NenetsAutonomousdistrict of the Tyumenregion, and the
last three along the tributaries of the Yenisei River in the Krasnoyarsk
region). The dialectsof Middle Taz and Baikha-Turukhanare very close
and accountfor aboutone-halfof today's native speakersof Selkup.They
have servedas the dialect basefor all main attemptsto promoteSelkup
literacy and for the publication of school textbooks(with the exception
of the most recentones,which lean towardsthe Upper Taz dialect). The
descriptionbelow refers, unlessotherwisespecified,to the Middle Taz
dialect.
2 Central, in the northwesternpartsof the Tomsk region, with the dialects
of Tym (closer to the Northern group) and of Narym (closer to the
Southern group). Central dialects are or were spoken also by minor
Selkupgroupsin areasadjacentto thoseof the Khanty (Ostyaks),along
the rivers Vasyugan,Parabel,Vakh, andtheir tributaries.
3 Southern,in the central part of the Tomsk region, with the dialects of
Middle Ob' (spokenalong the Ob' River betweenthe towns Narym and
Kolpashevo, and standing closer to the Central group), Upper Ob'
(upriver from Kolpashevo),and the now extinct dialects of Chaya and
Chulym (along the Ob' tributaries).The Middle Ob' and Chayadialects
550 SELKUP

were usedin severalSelkup books publishedby Orthodox missionaries

in the late nineteenthcentury.
The Selkupdialect spokenat one time still further to the southeast,in
the vicinities of Tomsk, andoustedas early as in the eighteenthcentury,
mustprobablybe classifiedas representinga separatedialectgroup.
4 The Kef group, in the northeasternpart of the Tomsk region along the
Kef River, with the dialectsof Middle Kef (closeto the Southerngroup)
and Upper Kef (Nat-Pumpokolskdialect). Thesedialects are preserved
relatively better and look more viable than other Selkup dialects in the
Tomskregion.

The grouping set out above correlatesclosely with the distribution of

Selkup self-designations.In the absenceof any single autoethnonymfor all
territorial groups the Northern Selkups call themselvesso!jqup, :joPqup
(hence 'Selkup'), the Central ones, cum~! qup, the Southernones siisoq;J(j)
Selkups
qum (but Chulym tjujqum), andthe Kef Selkupssiis(s)ii qum.

Historical Development
According to the opinion sharedby specialistsin archaeologyand anthro-
pology, at least the Southern Selkups continue to live in the area which
belongedto the original Samoyedicproto-homelandtwo thousandyears ago
and more (seeHelimski 1991). The present-dayrepartition of dialects is the
result of the gradualpenetrationof Selkup speakersin the northwesternand
northerndirections.The NorthernSelkupstook over their present-dayhabitat
(oustingor assimilatingthe previousEnetspopulationof this area)as recently
as the secondhalf of the seventeenthcentury. Still, the earliest linguistic
recordsfrom the beginningof the eighteenthcenturyprove that all the main
distinctive traits of all Selkupdialectgroupswere alreadypresentat that time
(seeHelimski: 1985a).
Structurally, Selkup looks in many respectscloser to Ugrian languages
than its Northern Samoyedicrelatives; this may be at least partly due to
secondarycontactsbetweenSelkupandObUgrian.
In the domainof phonologyboth innovationsand uniqueretentionscanbe
found. Selkup has developeda secondarycorrelative opposition between
short and long vowels, and hasundergonean extensiveprocessof qualitative
vowel reductionin non-first syllables;a seriesoflabializedplosivesappeared,
later to be lost in the Northern dialects.On the other hand, the vowels of the
first syllable preserve,in the main, the quality of their Proto-Samoyedic
monophthongalprototypes,and the dialectsof the Tomskregion are the only
Samoyedicidioms in which the distinction between*t and *c is preserved
(thesetwo consonantsmergedlong ago in other Samoyediclanguages,and,
aslate as the first half of the twentiethcentury,alsoin NorthernSelkup).
The most characteristicinnovationsin Selkup morphologyare the radical
 SELKUP 551

extensionof the case system(mainly through the fusion of postpositional

constructions),the loss of a conjugatednegative verb, the developmentof
preverbs(separableverbal prefixes), the developmentof the morphological
meansfor mutual transformationsof syntactic roles of nominal, adjectival,
and verbal stems.Furtherdevelopmentsin individual dialectspertainin most
casesto inventoryratherthanto structure.
The original syntactic system, with extensiveuse of non-finite verbal
forms, has been replacedin most non-Northerndialects (partly also in the
North) by the use of subordinateclauseswith conjunctions,patternedafter
Russian.Other domains of syntax, including word order, display stronger
retentiveproperties.

Phonology

Phonemesand their Modifications

The phonemicinventory of Middle Taz Selkupincludesforty-one segmental
units, with the vowels far outnumberingthe consonants(twenty-five against
sixteen),seeTable 18.1 andFigure 18.1.
This numericalsuperioritymanifestsitself only paradigmatically, i.e. in the
phonemic system,while syntagmatically(i.e. in texts) the averageratio of
phonemesis roughly 132 consonantsto 100vowels.

Consonants
From the viewpoint of phonologythe sole affricate (:i, which has [tj] as its
optionalvariant(occurringidiolectally in all positions),forms jointly with the
stops a single series of plosives. Numerousempty cells in the system of
consonants,especially in the fricative series and in the postvelar (uvular)
zone, as well as the absence of distinctively voiced and palatalized
consonants,openwide the possibilitiesfor considerableallophonicvariation.
The phonologicallyvoicelessplosivesandfricatives areusuallyrealizedas
voiced or weak (half-voiced) betweentwo vowels or after resonants,cf. itt
'water': sG ittfn [tiDtn - tidtn]. At morphemicboundariesthis voicing may
often be absent, cf. uukf [-G- - -g-] 'tip', but uukmti [-k-] 'having no

Table 18.1 Middle Taz Selkup consonants

Labials Dentals Palatals Velars Postvelars

Nasals ill n ni IJ
Stops P t k q
Affricates ci
Fricatives s iii
Laterals I Ii
Trills r
Glides w r
552 SELKUP

Figure 18.1 Middle Taz Selkupvowels

Front Central Back

(unrounded) (rounded)

unrounded rounded

tense lax tense lax

High i ii III ii iiii 1 iY u uu

Mid e ee e ee 0 00 e ee 0 00
00
aa

Low a aa a aa

ptannigan' (uu 'ptannigan');this phenomenonis better accountedfor by

differentiating morpheme-internaland morpheme-initial(like word-internal
andword-initial) positions,ratherthanby ascribingphonemicstatusto voiced
consonants.On the otherhand,in Narym Selkup and someotherCentraland
Southern dialects the loss of word-final reduced vowels and some other
secondaryphonetic developmentshave resulted in the rise of a phonemic
contrast betweenvoicelessand voiced sounds: Narym pet 'by night': ped
'nest', qop 'man': qob 'skin' (in Taz Selkup respectivelypit: piti: qup :
qopi).
Most (non-palatal)consonantsbecomepalatalizedwhenfollowed by front,
especiallyfront tense,vowels. The stopsareoptionally realizedas implosives
word-finally or beforeotherstops.Also optionalis the fricative pronunciation
of stops,when they are followed by s or Si: apsi' [-fs- - -ps-] 'food', rakSia
[-xsL - -ksL] 'jingling pendant'.
Especiallydispersedare the positional (obligatory or optional) allophones
of the sole postvelarphonemeq. Besidevoiced, half-voiced,palatalized,and
implosive postvelar stops, as well as the postvelar fricative [x], their set
includesthe postvelarnasal [ij] (occursbefore nasals)and the postvelartrill
[r] (occurs word-internally mainly in the vicinity of liquids: qaqli' [qarll]
'sledge').
m
Idiolectally Si is realized as a non-palatalalveolar [s], a cacuminal or
this
[~]; this last pronunciationis typical of the Upper Taz dialect. Still another
caseof optionalvariationis thatbetween'Russian'non-palatal[I] andslightly
palatalized[1], both representingthe phonemeI.
 SELKUP 553

Vowels
The richnessof the phonemicinventory of vowels is due, in the main, to the
vast scope of the quantitative opposition (unlike Northern Samoyedic,in
Selkuplong vowels canhardly be treatedas phonemicsequences),and to the
contrast,in five pairs, of tenseand lax vowels: i : I, ii : II, e :10, ee : flO, 00
: aa. The rise of this last contrastappearsto be a NorthernSelkupinnovation.
The tensellaxcontrastmay originally havebeenrestrictedto the front vowels
in sequences like Gii- : Gil (where[I] was originally merely a manifestationof
i: due to the restriction on the occurrenceof r after palatals),but later the
distribution of front lax vowels becamemore complicated(e.g., lax but not
tense front vowels occur after dentals), and in a number of instances
morphophonemicdevelopmentsor borrowings led to the penetrationof the
opposition to other positions (cf. merka 'measurefor gunpowder' (from
Russian):merka 'the wind is blowing'). The contrastbetween00 andaa is the
resultof the development*aa > aa (modemaa is of secondaryorigin).
Phonetically lax vowels differ from correspondingtense ones by a less
strainedarticulationand by a concomitantshift towardsthe centralzone(I, II,
10, 1010) and/orlower tongue-height(10, flO, aa).
Long front vowels and 00 are optionally diphthongizedin open syllables.
In allegro speech,and especially in unstressedpositions, long vowels are
pronounced half-long or even almost short; this tendency towards the
phonetic neutralization of the quantitative opposition is very typical of
CentralandSouthernSelkupdialects.
The phonemic symbol r in non-first syllables denotesa reducedvowel
(common Selkup *~);[;:)]. it is realized asm or, more seldom, [5], and, after
palatals,[l] or [;:)]. While thereis no vowel harmonyin Selkup,non-Northern
dialectstend to assimilatethe reducedvowels in their phoneticquality to the
vowel of the first syllable: Ket' suur~[eH!gu][suurilm] 'animal',Tym[eH!gu]
el~gu [eH!gu]
'to live', andthis createsan effect similar to vowel harmony.

Stress
The dynamicstressin Selkupis mobile. Its placementin a word is determined
by a combinationof phoneticand morphologicalfactors.
According to the most widespreadphoneticrule of accentuation,stressis
placedon the last long vowel; if thereareno long vowels in non-first syllables
it is placed on the first vowel: p!!:.rqr 'smoke', uuGirqo 'to work', aqqaali
'rein', uuGiaami't 'we work', aatiiiiqd![ 'your two reindeers'. Regular
deviationsfrom this rule are determinedby the presenceof inflectional and
derivationalsuffixesof two accentologicallyspecifictypes.Onetype of suffix
containsvowels, which, althoughphoneticallyshort, behaveas though they
were long and attract the stress(unlessthere are long vowels in following
syllables): koggltrqo 'to tell', puutQntr '(to the) inside'. The other type of
suffix createsthe phenomenonof doublestresswithin a single phoneticword:
tf1.1da 'closerhere', t§.nis!'mi'P 'clever'; eachaccentologicallyseparatepart is
554 SELKUP

thengovernedby the basicphoneticrule of accentuationgiven on p. 533.

Differences in the morphemic composition of words account for the
relatively rare examplesof contrastivestress,such as Cir.ziCialqo 'to stamp
down': BEPBg}qo 'to stamp(once)', krkmzi 'loving': krkrtil j 'riverless'.It is
possible that certain accentologicalrelics are preservedby second-person
singularimperativeforms, in which there are such unusualcontrastivepairs
as e.g.
'lose!'.
!1rtii~ 'makefat!':'lose!'.
urtii~ 'lose!'.
In non-NorthernSelkup dialects there is often a strong tendencyfor the
stressto migraterightwardson to shortlow vowels of non-first syllables:Tym
kangk 'dog', iiwii 'mother'.

Distribution of Phonemes
Among the consonants,g does not occur word-initially, j occurs in this
position only in borrowings,and word-initial r occursonly in borrowingsand
onomatopoeticwords. Word-finally Ci and w do not occur at all (w is rare
word-internally, as well), and q and ~ are extremely rare. According to
phonostatisticdata,the most frequentconsonantsare (in descendingorder of
frequency)t, n, m, q,p, I, k.
Consonantclusters (consisting of two, seldom of three phonemes)are
normal only word-internally. Most common among them are the combina-
tions of a nasal or a liquid plus a plosive, and particularly the homotopic
clustersmp, nt, njCi, gk, It, IjCi, and the geminateclusterstt, qq, mm, nn, SS,
etc. etc. (these are the result of sandhi phenomenain many cases; see
~~,
pp. 555-6).Othertypesof clustersareconsiderablylessfrequent.
All vowels can occur in the first syllable, though there are numerous
restrictionson the occurrenceof vowels after certainconsonants.In non-first
syllableslong uu and iT, andlong and shorti!lee, iiluu, 0100, e/eedo not occur
at all, while short u and long and short ilii occur only in a few suffixes. In
general, the reduced vowel r and the short full vowel a are much more
frequentthan any of the other vowels; the relative frequenciesof ee, aa, uu,
iiii, ande are closeto zero.
Vowel clusters are possibleonly at morphemeboundaries,but there are
always optional variants with hiatus eliminated by contraction or the
epenthesisof a glide: kanaiilr - kanajiilr - kanEElr 'your (sg) dogs'.Very often
the first of two adjacentvowels in word groupsis apocopated:soma EEga -
somEEga 'is good'.

Morpbopbonology

Nasal - Stop Alternation

Perhapsthe most characteristicmorphophonemicphenomenonof Selkup is
the alternationof nasalswith homotopicstops(m - p, n - t, g - k), in some
casesalso with zero in word-final position, cf. qontam - qontap 'I'll find',
 SELKUP 555

suurim - suurip - suuri' 'animal',karman- karmat 'pocket',somaI] - somak

'well', aaI] - aak - aa 'mouth',etc. Suchvariantsare functionally equal and
they are almost equally frequent, for example,in pre-pausalposition. The
alternationis not automatic:word-final nasalsand stopsmay be also stable,
e.g., kem 'blood', toop 'edge', iYesi'n 'snare', kat 'ten', taaI] 'taiga', tuk
'beetle' etc. Dependingon the presence/absence and the conditions of this
alternation, severaltypes of stems and suffixes can be distinguished;their
morphophonologicaldifferencesmanifestthemselvesin both inflection and
derivation, cf. sG aa-n 'of a mouth',symbols
taaI]i~n 'of taiga',
symbols
tuki~n 'of a beetle'.
In orderto accountfor this difference,the following symbolswill be used(N
=nasal,T =stop):
Nl Tl N2 T2 N 2x N3
Alternationin word-final none none N-T N-T N-T N-T-0
position
Consonanttypically N T N T (-) 0
preservedin inflection

(The morphophonemicN2x doesnot occurin inflection.) We may therefore

write the examplesgiven above morphophonemic ally as qontam2, suurim3,
karman2, somaI]2x'aaI]3' kem], toop], iYesi'n], kat], taaI]], tuk].

Sandhi
The sandhiprocessescausedby the assimilativeinteractionof two consonants
belongingto different morphemes(stem+ suffix, suffix + suffix, compounds
and fused word groups), are shown in Figure 18.2. It will be seenthat the
assimilation can be complete or partial, progressive,regressive,or even
bilateral (as in the caseof ~-nnln> nlni ). Sandhiis determinedby morphological
rather than purely phonetic factors; the samecluster may be 'tolerable' in
nominal forms, but eliminated through sandhi in verbal ones (puri"Siti' 'his
torch', but suuri"iYiYi' 'it turns out that he is hunting' from *SUurlS-t'i), or
'tolerable' in derivation, but not in inflection (cf. kiitsan 'grandson'and the
regulartransformationts > ssin both nominal andverbal inflection).
A specialsandhirule transformsthe clustersNP (whereN is a nasal and
P is a plosive, not necessarilyhomotopic)into nasalclustersNNp (whereNp
is a nasalhomotopicwith P), if they are followed, after a vowel, by another
clusterconsistingof a nasaland a plosive: qomtii 'money': qomniinti' 'of his
money', soqiWiYi'qo 'to ask': soqiWnlimpa'he asked':soqininlimmanti''you
asked'.As the last example shows, this rule applies recursively from the
beginningto the endof a word. The morphemeboundary(or boundaries)may
be locatedat any placein the string -NPVNP-.
Someothersandhiphenomenaarecausedby phoneticprocesseswhich are
no longer productiveand regular (for example,the loss of initial t in certain
556 SELKUP

Figure18.2 Sandhiin Middle Taz Selkup:assimilativeinteractionof the

consonantsC1 and C2 whereC1 and C2 belongto different morphemes

C
C k t q s iii rn n 1

p mm mn [~1] [~1]

t ss gigi nrn T]n 11

k qq urn un [yl]

q kk ijrn ijn [fl]

s si gi

si ci Ci(l) si si ni ni(1) Ii Ii (1)

rnl [~1] [~1]

nl Uk ni iii 11

ni nici nini

Ul [y 1]

rn2,P) pk(3) pt(3) pq ps psi [~1] [~1]

n2,3(2) tk(3) tt(3) tq ss gi si 11

U2,3(2) kk(3) kt(3) qq ks ksi [yl]

Ii l i ci iiii

j jci

Notes: This figure accountsonly for thoseoutputclusterswhich differ from the simple sum of
their input components.If the outputis phonesratherthanphonemes,theseare given in square
brackets(i.e, phonologically[~l][~l] is equalto Ipl/ or Iml/, [ijm] to Iqm/, etc.).
(I) This sandhiassimilationoccursonly in verbal inflexion. (2) Final nasalsin bound
morphemesusually behavelike No' (3) For N2 this assimilationis optional.

verbal suffixes after verbal stems ending in r), They shall therefore be
relegatedto the domainof suffixal allomorphy.

StemVariants
While the nominativesingularservesas the main stem (first stem) of nouns,
certainparadigmaticand derivationalforms are basedon its modified variant
 SELKUP 557

(secondstem).If the fIrst stemendsin a vowel, the two stemsare identical.

If the fIrst stemendsin a consonant,the secondstem acquiresan additional
fInal vowel 0"), or losesits fInal consonant(usually N3 ) and, in monosyllabic
nouns,may containa vowel quantitativelyor qualitatively different from that
of the fIrst stem. In the following examples,the secondstemis exemplifIed
by plural nominatives(suffIx -t): qum 'man' : qumrt, kanalJ 'dog' : kanat,
aqsil 'mushroom': aqsilrt, ruF 'Russian': ruuFi"t, ser 'bog, marsh' : siTrrt,
nom 'God, heaven': nuut, qOlJ 'chief: qoot.
The modifIcation of verbal stems occurs mainly in derivation (paradig-
matically it is found only in severalmonosyllabic verbal stems). Here the
morphophonemicprocesseswhich producestemvariantsare lessregularand
more diverse; they include lengthening or shortening of stem vowels,
truncation, and geminationof intervocalic stem consonants,e.g. ousted,Ciuuri~ 'to
ousted,
cry' : Ciurqil- 'to startcrying', qo- 'to fInd': qooqil- 'to fInd many', saari~ 'to
bind' : sarrEl:;- 'to bind (already,intensively),.
In both nominal andverbal stemsthe last vowel (usually n is often ousted,
if the suffIx that follows beginswith a vowel: uti" 'hand' : utam 'my hand',
live'
ili~ 'to live' : ilcntak Tlllive'.

Morphology

Parts ofSpeechand CategoryofRepresentation

The Selkup language,with its more or less banal set of parts of speech
(substantives,adjectives,numerals,pronouns,verbs, adverbswith particles,
postpositions,preverbs,interjections),has developeda system of morpho-
logical meansfor transformingthe typical syntacticroles of parts of speech
into one another.While for verbs such transformations(into verbal nouns,
participles, and verbal adverbs)are a common phenomenonin the Uralic
languages,therearein Selkupalso specialparadigmaticforms of substantives
with the meaningsand syntacticfunctions of adjectives,verbs, and adverbs,
e.g. (qum 'man, person'): qumaak 'I am a human being', qumil} 'human',
qumi'k 'human(ly)" etc. The regularity and productivity of such formations
makesit expedientto treat them as inflectional ratherthan derivational,and
to use in the descriptionof (at least) substantives,adjectives,and verbs the
notion of categoryof representation,with four categoricalmeanings(sub-
stantival,adjectival,verbal, andadverbialrepresentations).
Substantives
Number.The singularis unmarked.The dual hasthe suffIx -QI (to the left of
which reducedi" goes to 00, and other vowels are lengthened),in several
words also -AAQI, -OOQI, and (with reduplication)-QAAQI. The plural has
the suffIx -T2 (joined to the second stem), in the vocative case -N, in
possessiveforms -ii- (joined to the secondstem). The collective form has
-eMl (joined to the secondstem),so that formally it is a combinationof the
558 SELKUP

Table 18.2 Nominal declension,substantival representation: nom 'God,

heaven'

Singular Dual Plural Collective

Non-possessive
forms
N nom nopqI nuut nuulimi
G nuun nopqm nuutin nuulimYn
A nuum nopqlm nuutim nuulimYm
Ins nopsa nopqIsa nuussa nuulimYsa
Car nomkaalik nopqlkiiruik nuutkaalik nuulimYkiiruik
Trans nuutqo nopqltqo nuumko nuulimHqo
Co-ord nuuSiSiak nopqlSisiak nuutisisiak nuulimYSisiak
DatJAll nuunik nopqltkim nuumklm nuuliminik
III nomti nopqltkim nuut'itklm nuulimlnri
Loc nopqin nopqilqin nuutqin nuulimHqin
Ela nopqini nopqilqini nuutqini nuulimHqini
Prol nomrriin nopqilmin nuunmin nuulimHmYn
Voc nomee nopqee- nopqiI nuunee

sl Possessive
forms
NA nommY nopqlm(i:) nuuiim(i) nuulimim'j
G nomni - nuuni nopqmi nuuiini nuulimi'ni
Ins nomnisa- nuunisa nopqmisa nuuiinisa nuuliminisa
Car nomnikftftlik - nopqmikaalik nuuiiniki'truik nuuliminikaalik
nuuniki'tftlik
Trans nomnoo(qo)- nopqmoo(qo) nuuiinoo(qo) nuuliminoo(qo)
nuunoo(qo)
Co-ord nomnisiak- nuunisiak nopqmiSiak nuuiiniSiak nuuliminisiak
DatJAll nomninik - nuuninik nopqmikim nuuiinikim nuulimininik
IlllLoclEla nopqlik nopqilqlik nuuiiqlik nuulimHqlik
Prol nommak nopqilmlik nuuiimlik nuulimllmlik

s2 and s3 Possessive
forms
N (s2) noml1 nopqIli nuuiill nuulimiTi
(s3) nomti nopqlti nuuiiti nuulimltl
G nomti nopqmti nuuiinti nuuliminti
A nomti nopqlmti nuumti nuulimimti
Ins nomtisa nopqmtisa nuuiintisa nuulimYntisa
Car nomtiki'tftlik nopqmfIki'tftlik niinfIkftruik nuuliminfIki'tftlik
Trans nomtoo(qo) nopqmtoo(qo) nuuiintoo(qo) nuulimintoo(qo)
Co-ord nomtiSiak nopqmtisiak nuuiintiSiak nuulimintisiak
DatJAll nomtinik nopqmftkim nuuiinftkim nuulimintinik
IlllLoclEla
(s2) nopqlinti nopqnqanri nuuiiqant'i nuulimHqanti
(s3) nopqinti nopqnqinti nuuiiqinti nuulimHqinti
Prol(s2) nommlinti nopqnmanti nuuiimanti nuuJimHmanri
(s3) nommYnri nopqnminti nuuiim'inti nuulimHminti
 SELKUP 559

di Possessive
fonns
NA nommrr nopqlmn nuuiimn nuuJimlmn
G nomnn- nuunn nopqmn nuuiinn nuuJimlnn
Ins nomnnsa- nuunnsa nopqmnsa nuuiinnsa nuuJimlnnsa
Car nomnnkiiiUi'k - nopqmnkiiiili'k nuuiinrrkiiiili'k nuuJimlnnkiiiili'k
nuunnkiiiili'k
Trans nomnnqo- nuunnqo nopqmnqo nuuiinnqo nuuljminnqo
Co-ord nomnns-iak- nuumIs-iak nopqmnsjak nuuiinns-iak nuuljminnsjak
DatJAll nomnnni'k- nuunnni'k nopqmnkim nuuiinnkim nuuljminnni'k
IlllLoclEla nopqimI nopqnqinn nuuiiqinn nuulj ml1qinn
Prol nomminn nopqnminn nuuiiniinn nuuli ml1minn

d2 and d3 Possessive
fonns
N (d2) nomln nopqdn nuuiiln nuulimlln
(d3) nomtn nopqltn nuuiitn nuuliniitn
G nomtn nopqintn nuuiintn nuuliniintn
A nomtn nopqimtn nuuiimtn nuuliniimtn
Ins nomtnsa nopqintnsa nuuiintnsa nuuliniintnsa
Car nomtnkiiiili'k nopqintnkiiiUi'k nuuiintnkiiiUlk nuulimlntnkiiiUi'k
Trans nomtnqo nopqlntnqo nuuiintnqo nuuliniintnqo
Co-ord nomtnsiak nopqintnsiak nuuiintnsiak nuuJimlntnsiak
DatJAll nomtnni'k nopqlntnkim nuuiintnkim nuuliniintnni'k
IlllLoclEla nopqintn nopqnqintn nuuiiqintn nuuJiml1qlntn
Prol nommintn nopqnmlntn nuuiimintn nuuli ml1mlntn

pi Possessivefonns
NA nommlt nopqlmlt nuuiimlt nuulimlmlt
G nomnlt - nuunlt nopqmlt nuuiinlt nuulimlnlt
Ins nomnlssa- nuunissa nopqmlssa nuuiinlssa nuuljminissa
Car nomnltkiiiili'k - nopqmltkiiiili'k nuuiinltkiiiUi'k nuulimlnltkiiiili'k
nuunltkiiiili'k
Trans nomnltqo- nuunltqo nopqmltqo nuuiinltqo nuulimlnltqo
Co-ord nomni"Sisiak- nopqmlsisiak nuuiinlsisjak nuuljmlnlsisjak
nuunls-is-iak
DatJAll nomnlnni'k - nuunlnni'k nopqmltkim nuuiinitkim nuuljmlnlnni'k
IlllLoclEla nopqlnit nopqnqinlt nuuiiqinlt nuuli ml1qlnlt
Prol nomminlt nopqnniinit nuuiiniinlt nuuli ml1minit

p2 andp3 Possessivefonns
N(d2) nomllt nopqdlt nuuiiht nuuli mi1lt
(d3) nomtlt nopqltlt nuuiitlt nuuliniitlt
G nomtlt nopqmtlt nuuiintlt nuuJimlntlt
A nomtlt nopqlmtlt nuuiimtlt nuulimlmfit
Ins nomtlssa nopqmtlssa nuuiintissa nuuliniintlssa
Car nomtltkiiiUlk nopqmtltkiiiUi'k nuuiintltkiiiili'k nuulimlntltkiiiili'k
Trans nomtltqo nopqmtltqo nuuiintltqo nuuliniintltqo
Co-ord nomtlSis-iak nopqmtlsisjak nuuiintlsjsjak nuulimlntlsis-iak
DatJAll nomtlnni'k nopqmtltkim nuuiintltkim nuuliniintlnni'k
IlllLoclEla nopqlntlt nopqnqintlt nuuiiqlntlt nuulj ml1qintlt
Prol nommlntlt nopqnniintlt nuuiiniintlt nuuliniilniintlt
560 SELKUP

Table18.3 Nominal declension,adjectival,verbal,and adverbial

representation:nom 'God,heaven'

Relativeadjectival Co-ordinative Locativeadjectival

form adjectivalform form

Non-possessive nuuli nuusisi ali nopqi"li

Possessive
s1 nuuni1i nuuni'siali nopqiili
s2 nomtili nomtlsiali nopqiinti1i
s3 nomti'li nomti'siali nopqi'ntlli
d1 nuunnli nuunnsiali nopqi'nnli
d2 nomtnli nomtnsiali nopqi'ntnli
d3 nomtnli nomtnsiali nopqi'ntnli
p1 nuuni'tili nuuni'sisiali nopqi'nYti1i
p2 nomti'ti'li nomti'sisi ali nopqYntltlli
p3 nomti'ti'li nomti'sisiali nopqYnfitlli

Verbal representation(predicativeforms)
s1 nomiiak d1 nomi'l)mn p1 nomi'l)mi't
s2 nomiiiinti' d2 nomi'l)ln p2 nomi'l)llt
s3 nom d3 nopqr p3 nuut

Adverbial representation
nuuk

relative adjectival form (see below) with the generic pronoun Ml (the
orthographicconventionis to write collectivesand other similar casesas two
words, but the two componentsareinseparable).
The collective form, as distinct from the plural, rendersthe meaning of
integral and often uncountableplurality: pu[1mi" 'stones,heapof stones,stony
place', cf. pUt '(several)stones';qumilj mi" 'people,crowd', cf. qumi"t 'men,
humanbeings'.It is more commonto usecollective forms of inanimate,and
plurals of animatenouns.
SuffIxes of number always precedethose of case and possession;see
Tables18.2 and 18.3.
In mostnon-NorthernSelkupdialectsthereareno collectiveforms, andthe
plural suffIx is -LA.

Case.The casesystemincludes:

Nominative(unmarked,i.e. -0);
Genitive(-N2 , joined to the secondstem);
Accusative(-M2 ,joinedto the secondstem);
Instrumental(-SA) , renderingboth instrumentaland sociativemeanings;
 SELKUP 561

Caritive (-KAAI1N3 , or optionally -KAAL);

Translative(-QO,joinedto the form of the genitive; in first, second,and third
singular possesiveforms the precedingreducedr changesto 00, and the
suffix -QO itself may be then omitted) , denotesthe stateof the sentence
subjector the destinationof the sentenceobject;
Co-ordinative(-SjAN 2x' joined to the form of genitive) , denotesmeasureor
point of reference;
Dative/allative(in the singularand the collective -NIN3 , joined to the second
stem, andin the dual and plural-KiNI, with severalvariantssuchas -KiN!,
-KiNIN3 , -KiNTI,joinedto the form ofthe genitive);
Illative (in the singular and the collective -NT! after vowels and -TI after
consonants;before -NTI reducedr goesto stressedshort0, and other short
vowels becomestressed;in the dual and plural this casesyncretizeswith
the dative/allative,andin possessiveforms with the locative);
Locative (-QIN2; to the left of this suffix reducedr goes to 00, and other
vowels becomelengthened);
Elative (in non-possessive forms -QINI with the samemorphophonological
propertiesas -QIN2; in possessiveforms syncretizeswith the locative);
Prolative(-MIN2, with the samemorphophonologicalpropertiesas -QIN2);
Vocative (has only non-possessive forms of singular,dual, andplural, and is
marked with a long, commonly pronouncedoverlong, final or added
vowel, usually -E£).

In adverbs,postpositions,and pronounslocal casesare often markedwith

-A or zero (illative), -N2 (locative),and -N! (elative).
Other Selkup dialects often lack some of the above-listedcase forms
(especially the co-ordinative and the elative) or have additional and alter-
native case forms of secondaryorigin (e.g., elative or adessivein -NAN!,
destinativein -WU, etc.); seeBecker(1978).
Nominals which designateanimateslack locative and elative forms. This
appearsto be the only obligatory morphologicaldistinction betweenanimate
andinanimatenounsin Middle Taz Selkup.

Possessivity.Possessive
forms distinguishthreepersonsandthreenumbers
of possessors
with the help of two seriesof possessivesuffixes, oneusedwith
nominativeforms and the otherwith all oblique cases:

si s2 s3 di d2 d3 pi p2 p3
Nominative -MI. -LI, -TI -MIl -LII -TIl -MIT2 -LIT2 -TIT2
-(A)M 2-(A)L
Oblique -I -TI -TI -II -TIl -TIl-TIl -IT2 -TIT2 -TIT2

In oblique cases,possessivesuffixes either follow casesuffixes or (in the

instrumental,caritive, translative,co-ordinative,and dative/allative)precede
562 SELKUP

it. Locative and prolative have specialfused possessiveforms of sl and s2.

Furtherdetails concerningthe combinationsof caseand possessivesuffixes
canbe extractedfrom the datain Table 18.2.
The possessiveforms of the third person(and, in imperativesentences,of
the secondperson)often renderthe meaningof definitenessratherthan that
of possession.
Non-substantiveRepresentations
The markerof adjectival representationis -e. In relative adjectival forms it
is joined to the secondstem,in co-ordinativeand locative adjectivalforms it
replacesthe final consonantsof the relevantcasesuffixes (-!FAN2x > -!FALJ ,
-QIN2 > -QIe): alako 'boat' > alakozilapr 'boat-oar',alakoSiSial}lapr 'an oar
good for a boat',alakooqil} lapr 'an oar (which is) on the boat'.The structure
of the noun phraseis preservedin adjectival forms: somaalakomr 'my good
boat', somaalakonil} lapr 'oar of my good boat' (and not *'good oar of my
boat'),somaalakooqiizilapr 'oar (which is) on my goodboat',etc.As the two
last examplesshow, adjectivalforms canrenderpossessive meanings.
The forms of verbal representation(usually treatedin descriptivegram-
marsas predicativeforms of substantives)usein the axis-of-discourseperson
suffixes which are similar to the verb suffixes of the subjectiveconjugation
(sl -N2x' s2 -NTI, dl -MIl, d2 -UI, pI -MIT2' p2 -IlT ); they are preceded
by a special marker of V[erbal] R[epresentation] (-XL - -IJ- - MIJ- -
-IJAA-). In the third person the singular, dual, and plural nominatives of
substantivalrepresentationareemployed.
The verbal forms of substantivesare used as predicates:tan krpa iija-
gaa-ntr PRO.s2 LITTLE BOY-VR-s2 'you are a little boy'. When it is
necessaryto rendermeaningsof moodandtenseotherthanindicativepresent,
the relevantforms of the verb EE- 'to be' are used, e.g. tan krpa iijagaantr
EppintrPRO.s2 LITTLE BOY-VR-s2 BE-narrative-inferential-s2'it turns out
that you were a little boy'.
The suffix of adverbial representationis -N2x' which is joined to the
secondstem. It functions most often as a replacementfor relative adjectival
forms in compoundpredicates,e.g. tmolr 'cloud' > tmolil} Cleelr 'cloudy day',
tmoliy Esiinpa 'it becamecloudy'.
Adjectives
Inasmuchas relative adjectival forms belong,in Selkup,to the paradigmsof
substantives,the classof adjectivescomprisesmainly words with qualitative
meanings,both primary (soma 'good') and derived(and then usually ending
in _LJ: soopitJ 'sharp').When usedattributively, adjectivesare not inflected.
When used non-attributively as independentsubstantives,they are declined
like nouns(= substantivalrepresentationof adjectives).
The adverbial representationof adjectivesis regularly formed by adding
the suffix -N2x or replacing -e with -N2x: somak 'well', soopik 'sharply'.
Combinationsof theseadverbialforms (or of truncatedadjectivalstems)with
 SELKUP 563

existentialverbsareusedaspredicatesandcanbe viewedascompoundverbal
representationsof adjectives:som(a) EEl)a 'is good', soopii] EEl)a 'is (in a)
sharp(state)'.
There are no morphologicaldegreesof comparison.The meaningof the
superlativedegreeis renderedwith the help of the prepositiveparticlepoosi":
poosi"soma'best', poosi"soopili 'sharpest'.
Numerals
The primary cardinalnumeralsare:

1 ukkir 6 muktft2
2 Fitti" (non-attributively:FittiiiiqI) 7 seeliCii"
3 naakir 10 kat}
4 tutti" 100 toon2
5 sompila 1000 tiSiSia « Russian)

Other cardinal numerals are formed subtractively (Siitti" ci iiiil)kml i kat

TWO LACKING TEN 'eight', ukkir Ciiiiil)kml i kat ONE LACKING TEN
'9'), additively (ukkir keeli kat ONE EXTRA TEN '11 '); (toon Ej kat
HUNDRED AND TEN '110'), multiplicatively (siitti" tEEsar TWO FORTY
'80'; sompila toon '500'), by suffixation (Fitti"-sar '20'; nas-sar'30', with a
reductionof the first component),or by combinationsof thesemethods.Like
adjectives,cardinal numeralsare inflected (accordingto nominal patterns)
only when they are usednon-attributively.
Ordinal numerals(except ukoozi, poos+ukoozi 'first') are built with the
suffix -MTALle Goinedto the secondstem): Fittiintiilil i 'second',kaatiintii-
li1i 'tenth'. The suffix -MTAL forms multiple numerals:Fittiintiil 'again,for
the secondtime', kaatiintiil 'for the tenthtime'.
Pronouns
The classof pronouns(or ratherof 'prowords')can be viewed as comprising
not only pronoun-substantivesand pronoun-adjectives,but also pronoun-
verbs (e.g. qattII- 'to get where?') and pronoun-adverbs(e.g. tiinti" 'here').
Their relationship,however, is different from the relationshipsbetweenthe
forms of different representations,becausemorphosyntacticallydifferent
pronounsnormally arebuilt to different stems.
The personal pronouns (see Table 18.4) and the emphatic reflexive
personal pronouns (see Table 18.5) have case paradigms with fewer
distinctionsthan thoseof nouns, and often with irregular (partly suppletive)
relationshipsbetweencaseforms, especiallyin the axis-of-discoursepersons.
Third-personpronounsare usedonly with referenceto persons,the meaning
'it' beingrenderedby the demonstrativeanaphoricpronounnami".
Otherpronoun-substantives are declinedlike ordinary nouns(and have,in
particular, numerativeand possessiveforms, e.g. tammiiiiqlmi" 'thesetwo of
mine', from tammi" 'this', kutiintissii 'with which personsof theirs', from kuti"
564 SELKUP

Table 18.4 Declensionof personal pronouns

First person Secondperson Third person

sN man tan tep

sG man tan tepi"n
sA (ma)si1m (ta)shnti" tepi"m
sins massli tassli tepsli
sCar matkMli'k tatkiliHi"k tepkMlik
sTrans matqo tatqo tepi"tqo
sDatiAll mlikkli - matqlik tlinti" - tatqlinti" tepi"nik
dN mee tEE tepliliqI
dG mee tEE tepliliqm
dA (mee)shmn- (tEE)shntn tepliliqlm
(mee)shnn
dins meesli tEESli tepliliqIsli
dCar meekMlik tEEkfullIk tepliliqlkiliilik
dTrans meeqo- meetqo tEEqO -:- tEEtqO tepliliqltqo
dDatlAll meeqi"nin tEEqYnJcJn tepliliqltkim
pN mee tEE tepi't
pG mee tEE tepltln
pA (mee)si1mi't- (tEE)shntIt tepmm
(mee)simH
pins meesli tEESli tepi"ssli
pCar meekMlik tEEkiiiilIk tepi'tkiliH'ik
pTrans meeqo-:- meetqo tEEqO -:- tEEtqO tepi"tItqo
pDatlAll meeqi"nJH tEEqi"nJcJj"t tepi"tkim

Table 18.5 Declensionof emphatidreflexive personal pronouns

First person Secondperson Third person

sNG onlik 'myself' onlinti" 'yourself(sg)' onti" 'himlherself'

sA onliqqosianoo( qo) onlinti"qosiantoo(qo) onti"qosii"ntoo(qo)
sins onaksli onlinti'sli onti"sli
sDatlAll onliqqlik- onlinti"qlinti" - onti"qi'nti' -
onliqqosiaqlik onlinti"qosaqlinti" onti"qosiooqYnti"
dG onn 'ourselves(du)' ontn 'yourselves,themselves(du)'
dA onnqos.itnnqo ontnqosii"ntnqo
dins onnsli ontnsli
dDatlAll onnqi"nn - ontnqi"ntn- ontnqosiooqi"ntn
onnqosiooqYnn
pNG onH 'ourselves(pI)' ontIt 'yourselves,themselves(pI)'
pA onHqosii"nHqo ontItqosiYntItqo
pins oni'ssli onti"ssli
pDatiAll onHqi"nlt - ontItqYntIt - ontItqosiooqi"ntIt
onHqosiooqi'nH
 SELKUP 565

'who?'). Pronoun-verbsare conjugated;pronoun-adjectivesand pronoun-

adverbsarenormally not inflected.
The demonstrativepronounsdistinguishproximal anddistal deixis:

tam 'this (attrib)', tammi" 'this (non-attrib)" ti" 'here (lat)', tiinti" 'here (loc)',
tiTni" 'from here',etc.,

versus

toonna 'that (attrib)', toonnami"'that (non-attrib), too 'there (lat)', toonniinti"

'there(lat)', tooni" 'from there',etc.,

andanaphora

(na 'this/that(attrib)', nami" 'this/that(non-attrib),it', niT 'here/there[lat]',

niinti" 'here/there[loc]', niTni" 'from here/there,afterwards',etc.).

Occasionally,ftrst-personor second-person deixis is also expressed:ilnam

'this (of mine or nearme, attrib)', ilna 'this (of yoursor nearyou, attrib), Latin
iste'.
Interrogativepronouns(which are also usedas relative pronouns) include
qaj 'what', kuti" 'who', qajilJ 'what kind of, kutil} 'whose',qattII- 'to get to
where?',qatampi~ happen','what? to happen',qatamu- 'what? to have happened',
kuBBa 'where(lat)', kun 'where(loc)', kuuni" 'from where',kuSiSian'when'.
Negativepronounsarederivedfrom interrogativeoneswith the prepositive
particle NJ ] or with the sufftx -N£J: nii qaj, qajnEj 'nothing', nii kuSiSian
'never'etc.
Indeftnite pronounsare derivedfrom interrogativeoneswith the particles
KOS (pre-or postpositive),££MA (postpositive),QAM2 (postpositive).These
particles render different shadesof indeftniteness:qaj kos, kos qaj 'some-
thing', qaj EEma 'something,anything',qaj qap '(at least)something',etc.
The genericpronounmi" and its correlativesmil} (adjectival),mi"- (verbal),
mik (adverbial) function as surrogates,i.e. as universal indicators for the
correspondingparts of speech.The pronoun mi" occurs, for example, as a
replacementfor a missing or forgotten noun ('thingamy') and as a sub-
stantivizing element with adjectives and participles (qiftpil) mi" 'what is
killed'). Its verbal correlatesimilarly replacesverbsin sentencessuchas man
manni~mp-ap,manni~mp-ap, tan aSiSia manni~mp-ap,
mi~1J-al PRO.sI SEE-dur-slobj PRO.s2 NEG
DO-pres-s20bj'I seeit, you don't'.
Determinativepronouns include muntik 'all, entire(ly)' (functions as a
substantive,or adjective,or adverb),isi" 'every',jarik 'other'.
Pronoun-numeralsbelong to different morphosyntactic (adjectives,
adverbs)and semantic(demonstrative,interrogative... ) categories:naSiSiali
'so many', naSiSiak'so much', kuSiSialJ 'how many', kusisiak 'how much',
kuSiSiamtalil}'which (in order),Germandas wievielte',etc.
566 SELKUP

Verbs
Category of aspect. Verbs in Selkup are perfective or imperfective (this
distinctionis morereadily renderedin Russian:laokali - 'KPHKHYTb, 'to shout,
to have shouted' : laokisJ- 'KpHQaTb, 'to shout, to be shouting'. The same
aspectualmeaning is preservedthroughout the paradigm of verbal forms
derived from the same stem. In particular, it determines the temporal
referenceof the presenttense(seebelow). On the formal plane,the paradigms
of perfectiveverbs, as distinct from imperfective ones,lack presentpartici-
ples. Otherwise, perfective and imperfective verbs with primary (non-
derived) stems do not differ formally, while verbal derivational suffixes
usuallyrenderaspectunambiguously.

Categoriesofmoodand tense.The systemof moodsincludes:

Indicative (unmarked);
Inferential, or latentive, usedfor reporting eventswhich the speakerdid not
witness directly (suffix -NT-, often in combinationwith the prepositive
particleNA; the suffix surfacesas -T- after most consonantalstemsand as
zero after stemsendingin r);
Auditive, usedfor reporting eventswhich the speakerwitnessedby hearing
(suffix -KUNA-);
Conditional, usedfor indicating real conditionsin presentand future (suffix
-MMA- after vocalic and-MA- after consonantalstems);
Subjunctive,used for indicating unreal conditions and consequences in the
past (postpositiveparticle £NA, joined to slightly modified forms of the
pastindicative);
Debitive, usedfor qualifying future eventsas obligatory or enforced(suffix
-PSAAT- after vocalic and -SAAT- after consonantalstems);
Optative, usedfor qualifying future eventsas desirableor appropriate(suffix
-LA-, often in combinationwith the pre- or postpositiveparticleSA);
Imperative (with no special suffix, but with a set of finite suffixes for the
secondandthird persons,which differ from the finite suffixesusedin other
moods).

Different tensesare distinguishedonly in the indicative (present,future,

past,pastnarrative)and the inferential (present,future, pastnarrative).
The presenttenserendersthe meaningof 'presentperfect' with perfective
verbsand that of 'presentcontinuous'with imperfectiveones:the verb qo- 'to
find' is intrinsically (lexically) perfective, so qooai means'you have found
it (just now, and what has been found is at your disposal)' whereas the
correspondingpresent-tense form of the intrinsically imperfectiveverb pee-
'to searchfor' ,peeoal,means'you are searchingfor it'. The presenttensehas
no specialsuffix, but the finite suffixesin the presenttenseare separatedfrom
their stem with the elements(suffixes) -N- after consonantalstemsand -N-
after stemsendingin non-reducedvowels (sometimesalso after f·stems).The
 SELKUP 567

sameelementsoccuralsobeforethe finite suffixes in the imperative.

The future tense,etymologically relatedto the non-perfectivederivatives
(seep. 573), has the suffixes or suffix clusters-T-, -NT-, -t:NT-, -Tt:NT- etc.,
dependingon the phoneticstructureof the stem to which it is attached.The
suffix of thepasttenseis -S-, andthat of thepastnarrative tenseis -MP- after
consonantaland -P- after vocalic stems.Tensesuffixes always follow mood
suffixes andprecedefinite ones;for further detailsseeTables18.6-18.8.

Categoriesof person, number, and conjugation type. The finite suffixes

indicate the person and number of the subject and form two (partly

Table 18.6 Verbal inflection: indicative,inferential(qo- 'to find, to see')

Present Future Past (Past) Narrative

Indicative
sI qOlJak qontak qoosak qompak
sIobj qOlJam qontam qoosam qompam
s2 qOlJantY qonnantY qoosanti qommanti
s20bj qOlJal qontal qoosal qompal
s3 qOlJa qonta qoosa qompa
s30bj q01J1tl qontYIJ1tl qOOS1tl qompatY
dl qOlJimn - qontYmn - qoosimn- qompimn-
qOIJEj qontEj qOOSEj qompEj
d2 qOlJlIn qontlln qooslln qomplln
d3 qOlJaaqr qontaaqr qoosaaqr qompaaqr
d30bj qOlJltn qontYlJltn qoosltn qompatn
pI qOlJimlt qontYmit qoosimit qompimit
p2 qOIJ1l1t qontllit qoos1l1t qomp1l1t
p3 qOlJaatlt qont-aatlt qoosaatlt qompaatlt

Inferentiatl
sI qontak qonnintak qommintak
sIobj qontam qonnintakm qommintam
s2 qonnantl qonninnantY qommintal
s20bj qontal qonnintal qomminnanti
s3 qontY qonnintY qommintl
s30bj qontlti qonnintlti qommintlti
dl qontYmn - qonnintYmn- qommintYmn-
qontEj qonnintEj qommintEj
d2 qontlln qonnintlln qommintllu
d3 qontaaqr qonnintaaqr qommintaaqr
d30bj qontltn qonnintltn qomminti:tu
pI qontYmit qonnintYmit qommintYmit
p2 qontllit qonnint1l1t qommint1l1t
p3 qontaati:t qonnintaatit qommintaatit

Note: 1The inferentialmoodis often additionallymarkedwith the prepositiveparticlena.

568 SELKUP

Table 18.7 Verbal inflection: anditive, conditional, sUbjunctive

Auditive Conditional Subjunctive

sl qokunak qommak qoosal)Ella

sIobj qokunam qommam qoosamElla
s2 qokunantl qommantl qoosantElla
s20bj qokunal qomma-l qoosalElla
s3 qokuna qomma qoosanElla
s30bj qokunatl qommati" qoosi"t Ella
dl qokunamn qommamn qoosi"mnElla
- qokunEj -qommEj - qOOSEj Ella
d2 qokunaln qommaln qoosi1nElla
d3 qokunaaqI qommaaqI qoosaaqIElla
d30bj qokunatn qommatn qoosi"tn Ella
pi qokunlimi"t qommami"t qoosi"mi"n Ella
p2 qokunali"t qommali"t qoosmnElla
p3 qokunaati"t qommaati"t qoosaatlnElla

Table 18.8 Verbal inflection: debitive, optative, imperative

Debitive Optative) Imperative

sl qopsaiitak qollik
sIobj qopsaiitam qolam
s2 qopsaatanti" qollintl qOl)as(ik) - qOl)ik
s20bj qopsaatal qolal qotl
s3 qopsaata qola qOI)i"ja
s30bj qopsaati"ti" qolatl qOI)i"mtlja
dl qopsaatlmn qollimn
- qopsaatEj - qolEj
d2 qopsaatiln qolaln qOI)i1n
d3 qopsaataaqI qolaaqI qOI)i"jaaqI
d30bj qopsaati"tn qolatn qOI)i'mtljaaqI
pi qopsaiiti'mi"t qolami't
p2 qopsaatili"t qolali't qOI)i"t - qOl)mt
p20bj qopsaiitmt qolali't qOl)aati'- qOI)i'li't
p3 qopsaiitaati't qolaati"t qOI)i'jliliti"t
qopsaiitaati"t qolaati"t qOI)i'mtljaati"t

Note: 1 The optativemood is often additionally markedwith the pre-or postpositiveparticle

sa.

coinciding) series:the subjectiveand the objective conjugation.Intransitive

verbs always take sUbjective suffixes, transitive verbs can take suffixes of
eithertype (objectiveforms areusedmainly if the sentencefocus is the object,
 SELKUP 569

Table18.9 Verbal inflection, substantivalrepresentation:nomenactionis II

Nominative Genitive Translative Locative

Non-possessive qopta qoptan qoptatqo qoptaaqi'n

Possessive
sl qoptam(i') qoptani' qoptanoo(qo) qoptaaqak
s2 qoptali' qoptanti' qoptantoo(qo) qoptaaqanti'
s3 qoptati' qoptanti' qoptantoo(qo) qoptaaqi'nti'
dl qoptamn qoptann qoptannqo qoptaaqHllIr
d2 qoptaln qoptantn qoptantnqo qoptaaqi'ntn
d3 qoptatn qoptantn qoptantnqo qoptaaqi'ntn
pl qoptamn qoptann qoptannqo qoptaaqi'ni't
p2 qoptal"it qoptanti't qoptanti'tqo qoptaaqi'nti't
p3 qoptati't qoptanti't qoptanti'tqo qoptaaqi'ntlt

Note: Only the commonlyusedcasefOffils of nomenactionis are given here.

or in elliptical sentenceswith the object omitted). The suffixes of the

subjectivetype (I) are similar to the predicativesuffixes of nouns,thoseof the
objectivetype (II) to the possessivesuffixes of nouns:

sl s2 s3 dl d2 d3 pi p2 p3

sl -(A)1J2x -(A)NTr (-A) -(AA)QI

-(AA)MII -(AA)LII -(M)M'tr 2 -(AA)LIT2 -(AA)TlT 2
- £J
II -(A)M 2 -(A)L -(A)Tr -(AA)TII

Special sets of both subjective and objective suffixes are used in the
imperative;seeTable 18_8.

Non-verbalRepresentations
As in the case of the finite verbal forms, the substantivalforms of verbs
distinguish the personand numberof the subject with the help of suffixes
which are formally similar to the possessivesuffixes of nouns.This is so in
the caseof nominaactionis 1 (suffix -PTA after vocalic stems,-TA after most
consonantalstems, -A after r-stems) and II (suffix -KU; the two nomina
actionis occur in different constructions),and in the case of the infinitive
(suffix -QO, in possessiveforms -QI ___ QO).
The adjectival representationof verbs comprisesfive participles: present
(suffix -TIe or -NTIe), past (-pIe or -MPIe), debitive (-PSMTIeafter
vocalic and -SMTIe after consonantalstems), destinative (denotes the
destinationof an objectfor a certainaction; suffix -PSOafter vocalic and -SO
after consonantal stems), caritive (denotes non-realized action; suffix
-KUNJeJITIe). Participles of transitive verbs render, dependingon the
570 SELKUP

Table18.10 Verbal inflection, substantivalrepresentation:nomenactionis lIt,

infinitive

Nomenactionis II Infinitive
Genitive Co-ordinative

Non-possessive qOkUll quokusjsjak qoqo

Possessive
sl qOkUll"i qOkUll"iSjak qOq"illOO(qo)
s2,s3 qOkUlltl qokulltlsJak qOq"illtOO(qo)
dl qokullrr qokullrrsjak qoq"illrrqo
d2, d3 qokulltrr qokulltrrsJak qoq"illtnqo
pI qOkulltt qOkUll"iSjSjak qOq"ill"itqO
p2,p3 qOkUllttt qOkUlltISJSJak qOq"illtttqO
Note: Only thosecaseforms of the nominaactionis which arecommonlyusedare indicated
here.Potentiallythe paradigmsof nominaactioniscontainotherforms, as well.

Adjectival representationis as follows:

Preselltparticiple (peentiP,from pee- 'to (Perfectiveverbs [qo- amollg
searchfor') them] do not form present
participles.)
Pastparticiple qopili - qompili
Debitive participle qopsaatlll
Destinativeparticiple qopso.. .
Caritive participle qokun'CllfiIJ
Adverbial representationis as follows:
Presentverbal adverb qola
Pastverbal adverb qola puula
Caritive verbal adverb qokun'Claalik

constructionsin which they are used, both active and passive meanings:
qorqr-m qo-pil} ima BEAR-acc FIND-past.partWOMAN 'the woman who
found the bear', qorqi"-n qo-piti ima BEAR-gen FIND-past.partWOMAN
'the womanwho was found by the bear'.
The presentverbal adverb (suffix -LA), like the presenttense,denotesa
simultaneousaction with imperfectiveverbs and a directly precedingaction
with perfectiveverbs.When supplementedwith the particlePUULA, it serves
as the preteriteverbal adverbanddenotesa previousaction.
While the verbal system principally and in most details is the same
throughout the Selkup language territory, there are numerous dialectal
peculiarities in the formation and usage of non-verbal representations.In
particular, at least some SouthernSelkup dialects use the pure verbal stem
(which for NorthernSelkupis a mereabstraction)as a verbal adverbin verbal
compounds(seeHelimski 1983: 46).
 SELKUP 571

Adverbsand SyntacticWords
While many adverbialforms with qualitative meaningsbelong in Selkup to
the paradigmsof substantivesand adjectives,the class of adverbs,which
comprisesindeclinableforms that serveas attributesof adjectivesand verbs
(orsa 'very, very much',kenpila 'quickly') or as modifiersof whole sentences
(qaclti" 'maybe,perhaps',iTri"k 'still'), is frequentlyformally intermingledwith
classesof syntacticwords.
Somelocal adverbs,especially(il)lative adverbsin -A, are employedalso
as preverbs,i.e. prepositiveparticlesin phrasalverbs, e.g. IlIa 'down' in IlIa
mee- 'to bury' (lit. 'to make down'). Preverbscan be separatedfrom their
main verbsby pronominalobjects,negativeand other grammaticalparticles,
and in imperativesthey are placedafter the main verbs: IlIa Siinti" meeliik 'I
will bury you', meelJiintiJaIlIa 'let him bury'. An exampleof a non-adverbial
preverbis IJam 'one another,eachother': ziam ora- 'to wrestle',cf. ora- 'to
seize,to catch'.
About twenty nominal stems,someof them bound, are employedin four
local cases(illative, locative, elative, prolative) as10caVtemporaladverbsand
postpositions,cf. the themeil- in ilqi"n 'below', poo-n ilqi"n 'underthe tree',
ilqi"ni" 'from below', poo-n ilqi"ni" 'from under the tree', etc. Like other
nominals,suchpostpositionsalsohavepossessiveforms, which are usedwith
personalpronouns:man il-qa-k PRO.sl UNDER-loc/ilVela-sl'(to) underme,
from underme'. In addition,thereare manynon-adverbial(non-nominal)and
indeclinable postpositions,e.g. tara 'like, as', Blzati" 'for, about'. Most
postpositionsoccurwith the genitivecaseof the precedingnoun,cf. -n in poo-
n ilqi"n 'underthe tree', cited above.

Derivation

DenominalDerivatives
The most productive - or, in some cases,maximally productive within a
closedsemanticgroup- patternsof denominalderivativesin Selkupandtheir
suffixes are as follows:

1 Nouns:
Diminutive nouns (=eA, joined mainly to the secondstem): iija=zia
'baby' (iija 'child'), maati"=zia 'little house'(maatz 'house');diminutives
are also derived with several other, less productive or non-productive,
suffixes (usuallycontainingthe consonantK);
Singulativenouns (=IAKA, lit. 'piece, lump'; for round objects =SAJI,
lit. 'eye'; for oblong objects =QU, lit. 'stalk'; all theseformations are
derivedmainly from adjectival forms of substantivesand can be treated
as compoundsrather than suffixal derivatives): ulqa-I=laka 'ice-floe'
(ulqa 'ice'), uti~zi=sajr water''drop' and uti~IJ=quwater''jet of water' (ut 'water');
Associative(connective/reciprocal)nouns,with the meaning'X plus the
572 SELKUP

person(s)for whom he or she is X', where X is the baseof derivation

(=Sl, followed by dual or plural suffixes): ima=si"-qaaqI 'marriedcouple'
(ima 'woman,wife'), esi"=si~t
'father 'fatherwith his children' (esi" 'father').The
basesof derivationaremostly kinship terms;
Instructive nouns, with the meaning 'person who has X' (=SlMA):
moolmi"=si"ma 'liar' (moolmi" 'lie, deception');
Caritive nouns, with the meaning 'person who lacks X (=KITA):
eeti"=krta 'dumbperson'(eai" 'word, language').
2 Adjectives:
Instructive adjectives,with the meaning'having X' (=SIMILJ): saji"=kmti
teni"=si~
mil} 'clever' (teni" 'mind, intellect');
Caritive adjectiveswith the meaning 'lacking X' (=KITle): saji"=kmti
'blind' (saji" 'eye');
Destinative adjectives (=Tle): picJr=titi 'for making an axe' (picJi"
'axe').
3 Verbs:
Verbs with the basic (but not the only) meaning 'to becomeX' (=N'" or
=M-, added to the second stem; also used with adjectives): aalfg-
'becomesweak' (aali" 'weak'), ira=m- 'gets old (of males), (ira 'old
man');
Verbs with the basic meaning 'to transformsomethinginto X' (=MTI-,
addedto the secondstem; also usedwith adjectives):aali"=mti~ 'makes 'makes
weak',paari"=mti~
'arranges'arrangesa feast' (paari" 'feast');
Verbs with the basic meaning 'to do/make with X' (=T- or =TTl- :
CiImi"=t- 'glues' (cJImi" 'glue'), kekki'beheads'
"=tti~ 'suffers' (kekki" 'trouble, tor-
ment');
Verbs with the meaning'to searchlhuntfor X' ('captitives',=SJ_, added
to the secondstem):qorqi"=sJ- 'huntsbear' (qorqi" 'bear');
Verbs with the meaning'to lose X' (=KILlM-): saji"=kili"m- 'losesone's
sight, goesblind' (saji" 'eye');
Verbs with the meaning 'to deprive of X' (=KILTl-):'beheads'
oli"kilti~ 'beheads'
(oli" 'head');
Verbs with the meaning'to smell of X' ('olefactives',=NJl, addedto the
adverbs 'to smell of dog(s)' (kanay 'dog').
secondstem):kana=n}i~
4 Adverbs:
Instrumental adverbs (=N2 , added to the second stem; of limited
productivity): topi"=n 'by foot' (topi" 'foot'), qeqqi"=n 'at a walk' (qeqqi"
'step').

DeadjectivalDerivatives
Besidethe deadjectivalverbs mentionedabove,the derivativeswith restric-
tive meaning are very productive, both adverbs (suffix =LAAQl) and
adjectives(=LAAQl=e): soma=laaqi" 'rather well', soma=laaqi"=l} 'rather
good' (soma 'good'). Occasionallythesesuffixes are joined to adverbsand
 SELKUP 573

even to verbal adverbs, as well: ira=m-lii=lculqf 'getting a little older'

(ira=m-lil 'gettingolder', ira=m- 'getsolder').
Intransitive and Transitive, Causative,ReflexiveVerbs
The valence (+/- transitivity) in primary verb roots is unmarked, and
occasionallythe sameverb can be usedboth as intransitive and as transitive:
mi~'is 'is suspended;hangs (tr)'. In derived verbs, however, the derivational
suffixesusuallyrenderunambiguously,often togetherwith othergrammatical
meanings,the meaningsof intransitivity or transitivity. Thereareseveraltypes
of correlativederivatives.For example,intransitiveperfectiveverbsbuilt with
the suffix =MMT- systematicallycorrelatewith transitive perfective verbs
built with the suffix =ALT1-: nieninifmaat- 'gets angry' : nieninialti- 'makes
X angry', tarqiinaat- 'shakes(intr)' : tarqalti~
'studies,'shakes(tr)'. Many borrowed
roots add =1T1- to form intransitiveimperfectiveverbs,and add=N- to form
transitiveperfectives:uCimi~
'studies,
'studies,learns': uCiii]- 'teaches'(Russiany'uJ./,
Th, -ThC»).
The causative(curative) suffix clusters(=RALT1-, =TALT1-, =£f£fALT1-,
=ALIALT1-) canbejoinedto most verbroots andto manyderivedverb stems.
The causativeverbs indicate that their subject (the causator,in the nom-
inative) doesnot coincidewith the subjectof the action itself (the performer,
usually in the dative/allative): mat tlmnia-ni~nik na alako-m iitf=ralti~s-am
on-ilk Ciaatf PRO.sl BROTHER-sl-dat/allTHIS BOAT-acc TAKE=caus-
past-slobjSELF-slgenFOR 'I mademy brothertake this boatfor me'.
The reflexive verbs are derived from transitive (sometimesalso from
intransitive) verb stems with the suffixes =1-, =CJ1- or (with additional
intensive/perfectivemeaning)=II-, =I£fCJ1-:panal+ 'breaks[intr]' (panal-
'breaks [trn, tott=IliCi ~
'brings self to vertical position' 'stands').
(totti~ 'stands').
Many reflexive verbswith the samesuffixes are deriveddirectly from bound
roots. There are certainpeculiaritiesin the conjugationof reflexive verbs,cf.
ili~s-a'(he) lived' , but panal=i~s-@
'(he) broke [intr]'.
Mode ofAction (Aktionsart)
The mostproductivemodesof actionandtheir suffixes are asfollows:

Iterative (imperfective verbs; =kiPolipi~,

repeatedly' with numerous variants):
Ciattf=kiliolipi~
'throwsmany times, repeatedly'(Ci'throws');
'throws atti~ 'throws');
Habitual, often used in gnomic sentences(imperfective verbs; =KK1- after
vocalic and =K1- after consonantalstems): Ciattf=kki~'throwsusually';
Durative (imperfectiveverbs; =MP1- after vocalic and =p1- after consonantal
stems): Ciattf=mpi~
'is throwing'. With their logical object treated as
syntacticsubject,the durativederivativesof transitiveverbsacquire(in the
subjectiveconjugation) a passivemeaning:pii Ciattf=mp-a 'the stone is
thrown';
Non-perfective,denoting that an action remains unfinished (imperfective
verbs; = T1-, =NT1-, =E:NT1-, =TE:NT1-, etc., dependingon the phonetic
574 SELKUP

structureof the basestem): Batt=wti~'makes'makesan attemptat throwing, is

in act of throwing without finishing the movement'.The future tenseof
verbsis relatedto the non-perfectivemodeof action,but in modemSelkup
the two forms are independent,as is provenby the existenceof the future
tenseof non-perfectiveverbs: Batt=wn-wt-a-m 'I will make an attempt
at throwing';
Intensive!peJfective (perfective verbs; =LJ(;Jl_, =E.:E.:, =E.:LJ(;Jl_, =II,
=ILJ(;Jl_, depending on both phonetic and morphological factors):
Batt=n:- 'hasalreadythrown with an intensiveeffort';
Multiobjectival, denoting that an action is applied to many objects or to
different partsof a single object (perfectiveor imperfectiveverbs,derived
only from transitive bases;=QIL- or =Q1LAAL-, =AL- or =ALAAL-, =L-
or =LAAL-, depending on the phonetic structure of the base stem):
Batt=iil( ddl)- 'throwsmanythings';
Multisubjectival, denoting that an action is performedsimultaneouslyby a
multitude of homogeneoussubjects (perfective or imperfective verbs,
derived only from intransitive bases;the same suffixes as from multi-
objectival verbs,followed by the reflexive suffix =1-, then followed by the
durative suffix =MPl- or an intensive/perfectivesuffix): aaljB=iili"=mpi~ followed
'falls in largequantities(e.g. ofleavesin autumn),(aaPBi~ 'falls');'falls');
Inchoative(perfectiveverbs,derivedfrom imperfectivebases;=AL J-, =ALl-
andseveralother suffixes): quut=iili- 'falls ill' quut=iili-
(quuti~ 'is ill');
Attenuative(mainly perfective verbs; =E.:(;'1- in intransitive and =APT1- in
transitive verbs): qant=EBi~'falls'); 'gets slightly frozen' 'becomes(qanti~ 'becomes
frozen'), Batt=iipti~'throws'throwsslightly, to not far away'.

In many instancesthesesuffixes arejoined to themes,ratherthan existing

verbs. The suffixes, and meanings,of different modes of action can be
combinedin clusters,asin Batt=iilt=twti"=kiljoPpi~throwing manyTHROW=multiobj=non-
perf=iter 'makesnumerousattemptsat throwing manythings'.

Other DeverbalDerivatives
There are severalsuffixes for forming deverbalnouns,for example,=SAN3
for nomina instrumenti(mir=san 'carpenter'splane' from miir- 'planes')or
=MO for nomina loci (minir=mo 'hunting area' minir- 'hunts'),but none of
them is productivein present-daySelkup. On the other hand, the combina-
tions of participles with qum 'person' for nomina agentis (suuri"CiBil j qum
'hunter') and with the generic pronounmi" for nomina acti (tEEpilj mi" 'rot,
rotten scrap') occur very frequently and transfunctionallyapproachsuffixal
derivatives.
The derivation of adjectives and adverbs from verbs, as well as any
derivational processes with adverbsor syntacticwords as their base,are only
sporadicphenomenain Selkup.
 SELKUP 575

Syntax
Noun Phrase
Within the noun phraseattributes(adjectives,numerals,pronoun-adjectives,
subordinatenounsin the genitive, appositivenounsin the nominative,etc.)
normally precedethe headnoun. Occasionally,however,attributesthat have
subordinatewords of their own (participlesandsomeothertypesof adjectival
words) are placed postpositively: tan werqi" tnnnla-l, Siitti" poo-t kunti"
CiiiiiI]I]i"=mpi"-till , moqi"nii tii-I]-a PRO.s2BIG BROTHER-s2TWO YEAR-
gen DURING BE.ABSENT-pres.partBACK COME-pres-s3 'your elder
brother,who was absentfor two years,has comehome'.The namesof units
of measurementare treated as attributes to the namesof substances,and
placedprepositively:poCika itt 'a caskof water',poCika itti"-m ii-s-ap CASK
WATER-accTAKE-past-s lobj 'I took a caskof water'.
Thereis no agreementbetweenadjectives(and other attributes)and head
nouns. Normally there is also no agreementin number with attributive
numerals: ukki"r qum 'one person',Siitti" qum 'two persons'(but also Siitti"
qumooqI, in dual) , ntu'iki"r qum 'threepersons'.
Thelink betweenpossessor andpossessed
is expressed
eitherby placingthe
possessor in thegenitiveprepositivelyor by attachinga possessive suffix to the
possessed. Theuseof bothmarkerswithin thesamenounphraseoccursin non-
NorthernSelkupdialects,especiallywhen the possessed is a term of kinship:
(LowerChaya)era-n ii-d;J OLD.MAN-genSON-s3'theold man'sson'.

Verb Complex
Within the verb complex, inflected verbal auxiliaries are normally placed
postpositively. This statementholds for both compound predicateswith
substantival, adjectival, or adverbial first components(ittqi"n ffI]a 'is in
water', siiiiq fsiinpa 'becameblack', Ciasi"q ffI]a 'it is cold'; cf. also puu-t
CiiMti" ff-till AFfERWARDS-genFOR BE-pres.part'left for later need',and
puu-t Citulti" u-lii AFfERWARDS-genFOR IS-verb.adv'for later need',and
for numerousverbal analytic constructionswith 'non-finite' first components
(quntilii oriWn1a 'is dying', ili"qo olapsak 'I beganto live', qenqo ffI]a 'it is
necessaryto go', itti"r-qo fsiinp-ak DRINK-inf BECOME-sl 'I am thirsty',
etc.).
The sameorder of constituentsalso occursin negativeconstructionswith
verbal nouns (nomina actionis I), where the conjugated component is
actually the verbal noun with its possessivesuffixes: qo-ptii-mi" CiiiiiI]k-a
FIND-verb.noun-sl BE.ABSENT-s3 'I did not find', qo-ptii-li" CiiiiiI]k-a
FIND-verb.noun-s2BE.ABSENT-s3 'you did not find'. Such constructions
are used when reporting real events in the past. In other situations the
negation is expressedby the preverbal particles i"ki" (with optative and
imperative forms) and aSiSia (with forms of other moods); there is no
negativeverb in Selkup.
576 SELKUP

In interrogative sentencesthe verb often has the form of the inferential

mood. With the prepositive unstressedparticle qaj (cf. qaj 'what', which
normally has sentencestress)the verb complexexpressesgeneralquestions:
tiip koptooqi"n qaj ippi"nti" 'is it true that (s)helies in bed?(cf. koptooqi"n qaj
ippi"nti" 'whatlies in bed?',with a focal, sentencestresson qaj).
Simple Sentence
In syntacticterms,the dominantword orderin Selkupis SOv. Butas long as,
in terms of discourserelations,the topic is placedsentence-initiallyand the
focus is placeddirectly before the verb, other types of word order can also
occur: OSV (with topicalizationof the direct object), SVO (with the subject
focalized). Typically, interrogative words (as subjects,objects, or circum-
stantial modifiers in interrogative sentences)directly precede sentence
predicates.
The caseof the sentencesubjectis nominative,but genitive in construc-
tions with non-verbal representationsof verbs (see examples in the next
section).
The choice of the caseof the direct object dependson severalmorpho-
syntacticfactors. We may say with somesimplification that the direct object
is (la) always in the nominative, if the verb is an imperative, and (lb)
predominantly in the nominative, if the object is indefinite; but (2a)
predominantlyin the accusative,if the object is definite, and (2b) always in
the accusative,if the objectis a personalpronoun.
As mentionedon pp. 568-9, a significant role in shapingthe topic/focus
structure of the simple sentenceis also played by the conjugation type
(objectiveor SUbjective)of transitiveverbs.
SentenceCombining
Typically in Selkup the co-ordination or subordination of sentencesis
achievedby parataxis,without conjunctions,but sometimeswith the correl-
ative use of verbal moods: ima-ti" moqi"na qal-a, ira-ti" karra qiinn-EE-ja
OLD.WOMAN-s3 BACK REMAIN-s3 OLD.MAN-s3 TOWARDS.RIVER
GO-perf.intensive-s3'his (=the) wife remainedat home,her (= the) husband
went down to the water'; apstiiti; apsal EEmma 'feed him, if you have food'
(lit. ' ... your food is [conditional mood]'). Subordinatingand co-ordinating
conjunctions are partly borrowed from Russian, but partly result from
secondaryfunctions of native words (e.g., from employing interrogative
pronounsin the role of relative ones);they occurin the present-daycolloquial
languagemore often than in traditional folklore texts, and still more often in
non-NorthernSelkup dialects,so sentence-combining by meansof conjunc-
tions mustbe attributedmainly to the influenceof the Russianlanguage.
Anothermethodof sentence-combining consistsin transformingthe finite
verb in the subordinateclause into a form of non-verbal representation:
qumitit kit qantr tiiptaiiqi"n Ciasi"q Esi"kka 'when the peoplewere approaching
the river, it was getting cold', lit. 'people (pG) river to-bank in-coming
 SELKUP 577

(nomen actionis I) cold becomes-usually';tan njeninjanti" SiiitpiP porqal

tokkaltiiti" 'put on the fur-coat sewnby your sister',lit. 'you your-sister(sG)
sewn your-fur-coat put-on'; mat kuttar ilcntak iisiini" SiIp pEitikunjGitul1ik?
'how canI live without father'shelp?',lit. 'I how will-live my-father(sA) me
(sG) not-helping(ger)'.

Lexicon
The canonic shapeof primary (non-derived) stems is (C)V(C) for mono-
syllabicsand(C)VC(C)V(C) for bisyllabics,with the structures(C)VC(C)VC
occurring in nominal, but not in verbal primary stems.There are no stem-
initial or stem-finalclusters,and all other phonotacticrestrictionsmentioned
above(seeDistribution of Phonemes,p. 554) arestrictly observed.Deviations
from this canonicshapearefrequentin interjectionsandotheraffectivewords
(ksa 'comeon!', aa? 'yeah',ki"pil j oo.oo 'very-very little'), but they virtually
never penetrateinto non-affectivelexical domains. Such deviations occur,
however,in somerecentRussianloans,suchas stado 'herd' or metr 'meter',
though there usually are (unlessthe word is a nonce-usage)parallel adapted
forms (istati: metra).
The older strataof borrowingsin the lexicon of Selkupinclude:

Turkic loans,which areespeciallynumerousin SouthernandKet' Selkup,but

more than a dozen of which occur in Northern Selkup: maati"r 'warrior,
hero', siTri" 'cow', Giaarik 'bifurcation, tributary', etc. This topic receives
monographictreatmentin Filippova (1991);
Ostyakloans, more typical of Central and Northern Selkup: purqi" 'smoke',
kuras 'view, exterior',nurik 'straight',etc. (seeR6dei 1972,andUralic and
Ostyaketymologicaldictionaries).
Severalwords of Zyrian origin must havereachedSelkupby way of Ostyak:
niiini 'bread',ruSi 'Russian[noun]', mojtak- moniGiak'soap'.
Ket (and other Yeniseic) loans. It is often difficult to distinguishthesefrom
Selkuploansin Ket: kupak'fist', qeq 'pine forest',etc. (seeHelimski 1982,
with further bibliography).
Tungusic(Evenki) loans, more typical of Northern and Ket' Selkup: oljqan
'small woodedareasin tundra', Giaawiri~
berries','to step aside',kuja 'birch-bark
box for beating down and gathering berries', etc. (see e.g. Helimski
1985b).
Early Russianloans, beginning in the sixteenthcentury: qam 'linen' (Old
RussianXaM'b), sipiWGia 'pig', saaqir 'sugar',etc.

There are also sporadicborrowingsfrom Mansi (? Siiiq 'salt'), Mongolic

(torqi" 'receptacle'),Nenets(waOkna 'barrenreindeercow'), Enets(Raatta, a
place-name),as well as numerouswords of unknown, possibly substrate,
origin.
578 SELKUP

No systematicefforts to developthe vocabularyof Selkupby coining new

terms for culture and technologyhave ever beenmade,and the speakersof
Selkup apparently prefer direct borrowings from Russian to descriptive
coinages.

Text
A: phonemic transcription, segmental;B: morpheme-by-morpheme
gloss;
C: free translation.

infmtl = inferential; inf = infinitive; vidv = derivational suffix making

intransitiveverbsfrom verbs; lat.adv= lative adverb;dn = denominal;dec
= decade.

Al nioma-li porqY ira karr=a na

BI HARE-adj COAT OLD.MAN DOWN.TO.WATER=lat.advinfrntl

paI=sLmi"-nt-Y laniniY-m-tY mannY=mpY-qY-ntoo-qo.

COME.DOWN=vidv-narr- FISHWEIR-acc-s3 LOOK=dur-inf-s3-inf
infmtl-s3

A2 na (9 asYq-i1i &eelY-tY na EE-ppY-nH.

B2 THIS COLD-adj DAY-s3 infmtl BE-narr-infmtl-s3

A3 qO-I)-1t1 takkY=t qoltY-t

B3 SEE-pres-s30bj NORTHlDOWNSTREAM=loc.adv RIVER-gen

&iipY-qYt purqY-sa taat-i"k: ciiqqY=maat=EE-mp-a.

END-Ioc SMOKE-ins DIRECT-adv SMOKE=vidv=int-narr-s3

A4 seepY=la-k ira mannY=mpY-la tattY=raltY-mp-atL

B4 ENOUGH= OLD.MAN LOOKS=dur- BRlNG=caus-
dn-adv verb adv narr-s30bj

A5 na qaj i
nas=sar-i1 qalY-li mutY tU-I)-aatYt.
B5 THIS WHAT THREE=dec- NENETS- TROOP COME-pres-p3
adj adj

CI Old Hare Coat camedown to the river to checkhis fishweir. C2 It was a

cold day. C3 He saw that downstream,at the end of the (visible part of the)
river, there was directly rising steam(from the mouths of reindeer).C4 The
old man looked long enoughto let (the newcomers)approachhim. C5 That's
it: thirty Nenetswarriors havecome.
 SELKUP 579

Referencesand Further Reading

Becker, E.G. (1978) KaTeropIDI rralle)l{a B ceJlhKyrrcKoM H3bIKe, Tomsk: Tomsk
University Press.
Erdetyi, I. (1969) SelkupischesWorterverzeichnis(Tas-Dialekt), Budapest:Akade-
miai kiad6.
Filippova, T.M. (1991) 'TIOpKcKHe 3aHMCTBOBaHIDI B CeJIbKyrrcKoM H3b1Ke', Novosi-
birsk: unpublishedmanuscript(dissertation).
Helimski, E.A. (1982) 'Keto-Uralica', in E.A. Alekseenko et al. (eds), KeTcKHH
C60PHHK, Leningrad: Nauka,pp. 238-50.
- - (1983) The Language of the First Selkup Books, Szeged: J6zsef Attila
University.
- - (1985a) 'K HCTOpH'IeCKOH lIHaJIeKTOJIOrHH CeJIbKyrrcKoro H3bIKa', in
ITeKcHKa H rpaMMaTHKa H3b1KOB CH6HPH, Barnaul: Altai University Press,
pp.42-58.
- - - (1985b) 'CaMollHHCKO-TYHryccKHe JIeKCH'IeCKHe CBH3H H HX 3THOHCTO-
pH'IeCKHe HMrrAHKaI\HH', in YpaAo-aATaHCTHKa: ApXeOJIOrIDI, 3THorpaqlHH,
H3b1K, Novosibirsk: Nauka,pp. 206-13.
Irikov, S.V. (1988) CJIOBapbCeJIbKyrrcKo-pyCCKHH H PyccKO-CeJIbKyrrcKHH,Lenin-
grad: Prosvescenie.
Janurik,T. (1978) 'A szOlkupnyelvjarasokosztalyozasa',NyelvtudomanyiKozleme-
nyek 76/1: 77-104.
Katz, H. (1975-88) Selcupica,vols. I-IV, Munich: Finnisch-UgrischesSeminarder
UniversitatMtinchen.
- - - (1979) Selkupische Quellen: Ein Lesebuch, Vienna: Verband der
wissenschaftlichen GesellschaftenOsterreichs.
Kazakevic,O.A. (1990) llcrrOJIb30BaHHe3BM lIJIH HCCJIelloBaHIDI 6ecrrHcbMeHHblx
H MJIallOrrHCbMeHHblX H3b1KOB (Ha MaTepHaJIeCeJIbKyrrcKoroH3b1Ka), Moscow:
University of Moscow Press.
Kuz'mina, A.I. (1974) rpaMMaTHKa ceJlhKyrrcKoro H3b1Ka, pt. I, Novosibirsk:
NovosibirskUniversity Press.
Kuznecova, A.I., Helimski, E.A. and Gruskina, E.V (1980) O'lepKH rro
CeJIbKyrrcKoMY H3b1KY: Ta30BcKHH lIHaJIeKT, pt. I, Moscow: Moscow University
Press.
Kuznecova,A.I., Kazakevic,O.A., Ioffe, L. Ju., and Helimski, E.A. (1993) O'lepKH
rro ceAbKyrrcKoMY H3bIKY: Ta30BcKHH lIHaAeKT, pt II, Moscow: Moscow Uni-
versity Press.
Prokofev, G.N. (1935) CeAbKyrrcKaH (OcTHKo-caMOellCKaH) rpaMMaTHKa, Lenin-
grad: Izdatel'jstvoInstitutaNarodovSeveraCIK SSSR.
Pusztay,J. (1992) 'Lesz-eszolkup ujjasztiletesT,in FestschriftfurKaroly Redei,in
P. Dereky et al. (eds),Vienna- Institut fUr Finno-Ugristik der UniversitatWien -
MTA NyelvtudomanyiIntezet,pp. 367-70.
Redei, K. (1972) 'Ostzjiik jovevenyszavak a szelkupban', Nyelvtudomanyi
Kozlemenyek74: 186-93.
19 Kamassian
PeterSimoncsics

Although Kamassianis no longer used - the last speaker,Klavdija Plotni-

kova, died on 20 September1989 - it is perhapsone of the bestdocumented
amongthe Samoyediclanguagesthanksto the endeavours,perseverance, and
hard work of its collectors:MathiasAlexanderCastren(1813-52)fIrst of all,
then Kai Donnerand his meticulouseditor, Aulis Joki (1888-1935),and last,
but not least,Ago Kiinnap.
Kamassian,togetherwith severalothercognateandnon-cognatelanguages
of Russia,hoveinto the view of studentsof languagein the eighteenthcentury
when mapping and assessingthis vast Eurasiaticempirebecametechnically
possible, economically necessary,and culturally desirable.The travellers,
discoverersand, scholarswho collectedand/orpublishedKamassianmaterial,
most of them Russian subjects, are the following: G.F. Miiller (1735),
F. Ade1ung (1735), Johann Eberhard Fischer (1739-47), P.S. Pallas
(1772/1776/1786-89, 1811),andJ. von Klaproth (1823).
A Finn, MathiasAlexanderCastren,the founder of Samoyedology,made
his recordsof Kamassianmaterial(some870 words) in November1847.The
bulk of his reuvrewas publishedposthumously,by Anton Schiefner,in 1855.
On two occasionsanotherFinnish scholar,Kai Donner,visitedAbalakovo,
a village north of the SayanMountainswhereKamassianspeakerslived. The
fIrst visit was in 1911; the second,longer, visit was in the summerof 1914,
when he stayedfor almosttwo months.Donnercollectedsome1,550lexical
items as well as texts. His collection was also publishedposthumously,by
Aulis Joki, in 1944.
In 1926 a Russianscholar,A.J. Tugarinov, recordedand publishedsome
thirty Kamassianwords in an article entitled 'IIocJIe,n;Hl1e KaJIMa)l(l1', i.e.
'The last Kamassians'(in CeBepHa5IA3115I 1926: 1:73-8).
It was only thirty-sevenyearslater, in 1963, on a toponomasticexpedition
arrangedby the Uralic State University in Sverdlovsk (formerly, and now
once again, Ekaterinburg),that it becameapparentthat the Kamassiandeath
notice hadbeenpremature:therewere two elderly ladies still living who used
Kamassian.
From 1964 to 1970 the EstonianscholarAgo Kiinnap worked with these
last two speakersof Kamassian,Alexandra Zibjeva (born in the 1880s or
1890s)and Klavdja Plotnikova(born as Andzigatovain 1895), and was able
580
 KAMAS SIAN 581

to collect preciousandlinguistically uniquematerial.

In the early 1950s two exemplarystudieswere publishedon Kamassian
versificationby J. Lotz, an Americanlinguist of Hungarianorigin.
Thereare soundrecordingsof Kamassiandating from the beginningof the
twentieth century (Donner used the phonograph).The Uralic University
expeditionand Kiinnap both carried out extensivetape-recordingduring the
1960s.
Among the South Samoyedicor SayanSamoyediclanguagesKamassian
seemsto be more closely affiliated to Koibal than to the linguistic complex
known as Mator-Taigi-Karagass-Soyot. All of theselanguages/dialects died
out ante Kamassian,probablyin the nineteenthcenturyor evenearlier.
The ethnonymKamassianmay well go backto a compound,kama(1) + as
'mountain+ Az [an ancientSiberianethnonym]),. Sucha compoundwould fit
in well with neighbouringethnonymssuchasKaragass(? kara + as 'blackAz')
andHakass(hak + as 'white Az'). In another,equallyjustifiable interpretation
theethnonymKaragasscomesfromkara+ kas'black Kas', thelattermemberof
thecompoundbeinganotherSiberianethnonymmeaning'Tatarfrom theregion
of Minusinsk; Kaca-Tatar' and deriving, perhaps,from a proto-Samoyedic
word meaning 'human being, man' (Hajdu 1950: 95-6). Both of these
interpretationsreceivesomesupportfrom a third ethnonymof the area,Kyzyl
(Turkic 'red'), which forms - togetherwith the other two ethnonymswhich
contain colour terms -a tricolour of 'black-white-red'.Yet anotherinter-
pretationis possible,one which derivesthe kara of Karagassnot from Turkic
kara 'black' but from anotherKamassianstem(alsoof Turkic origin), namely
kara1 - karad 'steppe'.We would then have a correlative pair, Kamass/
Karagass'mountainAz [people]/steppeAz [people)', cf. the entry kard;}j in
Donner'sdictionary: 'Tatarin the village ofUgumakova;thesepeopleearlier
lived in the steppe,asopposedto the Kamass,who lived in the taigaandin the
mountains'(seealso Joki 1952: 163). The etymon 'mountainpeople'may be
echoedin the Latin term Monticolae Sajanensesof P.S. Pallas, one of the
earliestreferencesto this population(possiblydatingfrom1735).Spontaneous
mixing of explanationsandetymologiescannotbe excludedeither.Alongside
an ethnic triad (perhapsSamoyeds-Tatars-Mongols), the tricolour white-
black-redmay well refer to the habitat of thesespeakersand cover a trio of
topographicdescriptors(snow-cladmountains,barrensteppe,red-sanddesert).
In the following pagesa somewhatspeculativedescriptionof the language
as if it were still alive is attempted.The categoriesare traditional (phonology,
morphology,syntax,stylistics), but the emphasisis on processescategorized
as morphophonology.

Phonology
In its final phaseKamassianhad nine vowels, eight full andonereduced.The
reducedvowel;} seemsto have alternatedmostly with [+ high] vowels, and
582 KAMASSIAN

!Reduced! a

high a a a
Full mid a a a
low a a

was in all likelihood originally only a combinatoryvariant of these,even in

the middle of the nineteenthcentury: Castrencountedonly eight vowels. In
sum,the phonemicizationof ;) is a recentprocess.
The alternationof ;) with high vowels is certainly not independentof the
general reduction of the short, unstressed,high vowels of Hakassian,a
neighbouringTurkic languagewhich influenced Kammassianand to which
most speakersof Kamassianfinally switched.
Donner'sdatashow that the full, back, low vowel a is more often than not
rounded[a].
Since vowel harmony dominatesthe languageit seemsunnecessaryto
reckonwith pairssuchasi - r, e - ii, and;) - .1. It is sufficientto considerthethree
phonemesi, e, and;)asneutralwith respecttovowel harmony(seebelow).
With respect to quantitative oppositions, in addition to the opposition
FULL: REDUCED mentionedabovethereis also a tendencyfor stem-final
syllables to be lengthened,e.g. b;)raa-n 'sack (dative), amnoo-bi '[(s)he]
lived'. This non-phonological phenomenonis due to the influence of
neighbouringTurkic idioms.

Vowel Harmony
We may distinguishtwo types: tonal andserial. Tonal vowel harmonyis based
upon the oppositionof front and back vowels. In Kamassianthe vowels fall
into three tonal-harmoniccategories,back (u 0 a), front (ii 0 a), and neutral
(i e ;)). The domain in which tonal vowel harmony operatesis the non-
compoundword (a stem followed by one or more suffixes, including zero).
Within this domain, back vowels co-occuronly with back vowels, and front
vowels only with front, while neutral vowels co-occurwith either as well as
with themselves.A neutral vowel has no intrinsic phonologicalvalue unless
it occursin the companyof anotherneutral vowel, when togetherthey count
as front. Otherwise, neutral vowels obtain their value from the vocalism
occurringin the samedomain.The following examplesillustrate someof this
by meansof suffixal allomorphy:

Bisyllabic stem Plural suffix Tonal domain Example

allomorph
1 back+ back back back tura-zay'houses'
2 front + front front front back iiiii-zay 'caps'
3 back+ neutral back back kal;)s-(z)ay 'swords'
4 neutral+ front frontback front sira-zay 'snows'
 KAMASSIAN 583

Neutralvowels occurringaloneusually makefront-tonal words:

5 neutral+ neutral front front nere-liii-m 'I am
frightened'

Serial vowel harmonyis doubtlessdue to Turkic influence. Its effects are

more limited than those of tonal vowel harmony, in that it affects only the
syllables occurring in certain subcategoriesof stem-suffix sandhi: serial
harmony determinesthe [+/-] labial feature of thesevowels in the singular
non-third-personforms of the possessiveparadigmsof nouns.

Table19.1 Serial vowel harmony:num 'thunder;God',miit 'liver', kama?

'mountain',siirgiit 'elbow'

Singular Plural
Pxl Px2 Px3 Pxl Px2 Px3

num-bT num-nT num-dT v. num-b1? num-n1? *num-dTn

miit-p T miit-l T miit-t T v. miit-p1? miit-U? *miit-t Tn
kama-b1 kama-l1 kama-de v. kama-b1? kama-U? kama-den
siirgiit-p 1 siirgiit-11 siirgiit-U v. siirgiit-p1? siirgiit-U? siirgiit-ten

Key: T marksa high vowel which is governedboth by tonal and serial harmony.
1 marks a low vowel which is governedonly by tonal harmony.
* marksa form (P3) which is not ruled by serialharmony.Note, however,that the s3 form upon
which it is basedis so ruled.
- connectsstemwith relationalsuffix.

The codedforms abovereadas follows in a quasi-phonemictranscription:

numbwnumnwnumdwnumbaVnumnaVnumdun
miitpwmiitlwmiittwmiitpii?Imiitlii ?Imiittiin
kamabalkamalalkamadelkamaba?lkamala?lkamaden
siirgiitpiilsiirgiitliilsiirgiitte!siirgiitpii?lSiirgiitlii?lSiirgiitten(in

Consonants
The consonantsare setout in Table 19.2:
The tenseoral stopsare strongly aspirated.The palatal(ized)(in Donner's
terms: 'postdental')stopsti and di are more often hushing affricates(c, di)
thanhissingones(d, dzi ).
The inventory given in Table 19.2 is a cautious one. With some
morphophonemictrickery a tighter inventory of consonantsis possible
(Castren,for example,assumedonly twenty-four).
584 KAMASSIAN

Table 19.2 Kamassianconsonantism

2 2 3 4 25 6 56
Nasals m n Ii 1)
Tensestops p t ti k ?
Lax stops b d & g
Glides w j
Tensefricatives s ~ Si x
Lax fricatives z z! i! r
Laterals I P
Trill r

Note: Columns1-6 containbilabial, dental,palatal(ized),palatalizedhushing,velar, and glottal

consonants.

Glottal Stopin Kamassian

When it occursin prepausalposition, the glottal stop in Kamassian'canbest
be comparedwith the non-nasalizableglottal stop of Tundra Nenets,for it
alternates with obstruents only' (Janhunen 1986: 155). Elsewhere, the
Kamassianglottal stop is the 'mirror image', as Janhunenputs it, of the
so-called 'added' glottal stop of Nenets, inasmuch as it can precedethe
obstruentwith which it paradigmatically alternates,e.g. ma?- ma?d 'tent'
(both sN). This and other, similar, occurrencesindicate that the Kamassian
glottal stop may have arisen from the implosion of obstruents(rather than
their explosion, as in the case of Tundra Nenets). Since such implosion
probably involved some staggeringin the onset of the three closures(oral,
velic, and glottal), it is probably not unconnectedwith the neutralizationof
word-final nasality in Selkup (Janhunen1986: 167) and may also be related
to the reduction of stem-final vowels as well. In this connectionJanhunen
refers to the 'pharyngealization'of Tuva and Tofa short vowels before a
prepausalor a preconsonantalobstruent as an indication of specifically
Kamassian,andgenerallySayan-Samoyedic, influenceon theseneighbouring
languages(Janhunen1986: 168). It is possible that the etymology of the
ethnonym Kamass(ian)itself reflects such staggeredtiming: if the root in
questionis the sameas that seenin kama?- kawa? 'high place, heap;edge
of a ditch; height, hill, mountain'(from Turkic, wherethe forms havemedial
-b-), the neutralization of word-medial nasality is probably due to the
imperfect synchronizationof glottal and velic closure. The -lJ- in the self-
designationofthe Kamassians,kalJmaaZ;J,is probablyyet another,segmental,
manifestationof this glottalic activity, althoughit it is the only known instance
of nasalizationof a glottal stop before m. For a parallel to the lenition ?b >
w one can cite m;mz"Tlt:/S'(Donner 1944: 85, 101) 'is cooking', in which the
final t:/S' representsthe auxiliary verb i?b;J 'to lie'. This verb form is a
composite, made up of the gerund m;mZ;J=lii 'cooking' plus i?b;J 'lies'
expressingcontinuity (i.e. i ?b;J > t:/S').
 KAMASSIAN 585

Morphology

Nominal Inflection

AbsoluteParadigm
Traditionally, grammarslist seven casesfor the absolute (non-possessed)
singular nominal paradigm: N[ominative] -0, A[ccusative] -0/-m/-bV/-pV,
G[enitive] -n, L[ative] -n;J/-d;J/-t;J, LO[cative] -Nn/-pn/-g;Jn/-k;Jn,Ab[la-
tive] -Y;Jl etc. and INS[trumental] -ze l/-sel. The suffix variantsare selected
according to stem-type, which may be determined from the segments
occurring final in the stem. Stemsare either (V) vocalic or (C) consonantal,
andif the latter, endeither(Cl) in glottal stop with or without a following lax
consonant(usuallyd) or (C2) in all otherconsonants,which is alwaystense.
The case endings may be divided according to their relative age into
three classes:(1) old (A, G, L), (2) more recent (LO, AB[lative]), and
(3) most recent (INS). In the paradigmof nominal stem-type(Cl), which
is the most problematic and fairly recent, the three age-classesof case-
ending also differ in the morphonologicalprocesseswhich they undergo
in inflection. Thus the accusativeending is attachedto a Cl-stem either
by deletion of stem-final 1, e.g. sN mal, sA *ma1-m > mam, or by the
opposite processof extensionof the suffix itself, viz. syllabification, e.g.
-m > -ba in sA maldba. In this stem-type,deletion is also characteristic
of the sG (man < *ma1-n), while syllabification, togetherwith simultaneous
denasalization,characterizesthe suffixation of the L morpheme(*ma1-n;)
> mald;J). Another process, akin to syllabification, is reduplication,
characteristicof the LO forms of the Cl paradigm,in which the stem-vowel
is repeatedin the suffix, e.g. mal-an.
The superabundance of A forms for stem-classCl may reflect the relative
newnessof this form class, but it may also be seen as reflecting a much
older, areal tendencyto neutralizationand alternation of nasals and their

Table 19.3 Singularcaseendingsby stem-type

Vowel Consonant
-? -? + Lax consonant Tense:-k,-t, -p, -s, -s

A
2 A
N djaya 'river' maJorma?d'house' kilrilp 'pit, hole'
A tPayam mamor ma?dba kilrilbilm
G djayan man kilrilbiln
Lat djayay;m
tPayan~ ma?an
ma?d~ kilrilbiln
kilrilpt~

2 LO djayay;m ma?an kilrilpkiln

AB djayay~?
djayay;m ma?an
ma?g~? kilrilpkil?
3 INS tPayaze? ma?ze? kilrilpse?
586 KAMASSIAN

correspondingoral stops.The morphophonemics entailedin the shortsA form

mamparallel thoseof the sG (man), while the processesentailedin the long
sA form ma?dbaresemblethoseofthe sLat (ma?d:1).

Dual
Only onedual form, that of the nominativeabsolute,wasrecordedby Donner.
The rest of the paradigmmusthaveperishedearlier, sinceno recordsof such
forms exist in Castren'sor Ktinnap'smaterials.In a form suchas kaftlaZ:1gej
'two brothers',the dual ending -Z:1gej consistsof two elements:-gej is a dual
markerof Samoyedicorigin, andthe precedingmorpheme-Z:1 is a connective-
reciprocalsuffix, also of Samoyedicorigin.

Plural
Kamassianhas two plural markers.In one of them -zaIJ (-zaIJ - -saIJ - -saIJ),
the sameconnective/reciprocalsuffix seenin the dual (-Z:1) servesas a base.
The plural marker-zaIJ is usedin both absoluteandrelativenominalinflection
(see the next section). The other plural marker is -?je? (or -?ji?) in final
position and - n- in non-final position. There appearsto be a semantic
complementarityin the useof the two plural markers,in that -zaIJ is usedonly
with membersof a nominal classthat may be labelled [+HUMAN]: persons,
parts of the body, and domesticatedanimals, while -?je?- (and its variants)
occurswith all nominals.
The plurals of two nouns, essaIJ 'children' (singular di) and kopsaIJ
'daughters'(singular ko ?bdo), are traditionally classified as irregular. As
indicated above, a morphophonemicanalysis of the Kamassianconsonant
paradigmwhich is rather more generouswith rules but more parsimonious
with underlying segments(e.g. only one bilabial oral stop and only one
unpalatalizedsibilant) would accountfor such forms descriptively while at
the same time offering insights into the internal reconstruction of the
language.

The RelativeParadigm
The semanticand morphologicalcohesionof the threegrammaticalcasesN,
A, and G is apparentalso in the relative paradigm: here, these three cases

Table19.4 Nominal paradigmsof tPa'Ya 'river' with plural markers

-zal)l-sao -lje?

N tJia)f1zao tJia)f1 ?je?

A tJia)f1zaO;Jffl d'a'}f1?im
G tJia'}f1zaO;Jfl tJia '}f1 ?in
L tJia'}f1zaod:1 tJia'}f1?in:1
LO tJia '}f1zaopn tJia '}f1?i "Pfl
AB tJia '}f1zaOp? tJia'}f1?ip?
INS tJia'}f1zaoze? d'a'}f1?ize?
 KAMASSIAN 587

Table 19.5 Relative paradigms

Possessor Possessed
Singular Plural

SyncreticformsofNIAIG
Singular 1 djayam djayazaIJba
2 tPayal tPayzaIJna
j
3 tPayat d a yazaIJda
Dual 1 tPayawej tPayazaIJbij
j
2 tPayalej d a yazaIJnij
3 tPayaduj djayazaIJduj
Plural 1 tPayawa? djayazaIJba?
2 tPayala? djayazavna?
j
3 tPayaden d a yazaIJden
Syncreticforms ofULO
djayanii j
Singular 1 d a yazaIJani
djayanan j
2 d a yazavanan
djayanda j
3 d a yazaIJanda
Dual 1 djayaniiwej tPa yazaIJaniwej
2 tPayaniitej djayazavanilej
3 tPayandej tPayazaIJandej
j
Plural 1 tPayaniiwa? d a yazaIJaniwa?
j
2 tPayaniila? d a yazaIJanita?
j
3 tPayanden d a yazavanden
Syncreticforms ofAB
Singular 1 tPayatje j j
d a yazaIJati
2 djayattan tPa yazavattan
3 djayatte djayazavatte
Dual I tPayatjiwej djayazavatjiwej
2 tPayatjitej djayazaIJatjitej
3 tPayattej djayazaIJattej
j j
Plural 1 tPayatjiwa? d a yazaIJatiwa ?
tPayatjia? j j
2 d a yazaIJatita?
j
3 tPayatten d a yazaIJatten
Syncreticforms ofINS ®
Singular 1 tPayamze?
2 djayaize?
3 dj a yatj sie?

Note: The symbol ® indicatesthat the forms includedin the table herediffer in their
morphologicalstructurefrom the rest of the forms cited, as they are in fact postpositional
phrasesratherthan suffixatednominal forms.

syncretize.The more covert affinity of the L and LO casesalso manifests

itself here,againas a formal syncretism.TheAB andINS havedistinct forms,
the latter consistingof a postpositionalphrase.
588 KAMASSIAN

According to the available data the categoryof the dual appearsin this
paradigm only in connection with the possessor,and never with the
possessed;
contrastthe personalpronouns,below.
Pronominal Inflection
Demonstratives
Kamassianhasa more archaic,three-way,deictic systemthan the binary this-
that type of most European languages such as English, Russian, and
Hungarian.We may best describethe Kamassiansystem,borrowing Lotz's
term, as a T-shapedstructure:

dil '!hi,' di'that'

se 'yonder'
This three-waydemonstrativesystemis extendedby two remonstratives,
dum 'and this (even closerto speaker)'and iid;;, 'and that (even farther from
speaker)'. The form iid;;, ' ... andthat' may well be a compoundof the Russian
conjunctionIf 'and' plus the Kamassiandemonstrativedi. The initial long ii-
may be due to the foreign origin of the morphemeor, what is more probable,
by the characteristicallydeviantphonologicalbehaviourof deictic systems.
Personalpronouns
Unlike nouns, Kamassian personal pronouns do have dual forms. The
nominativeforms are:

s d p
1 man miste mi?
2 tan siste si?
3 di diside dizfu]

We evidently have here anotherT-shapedstructure, in which the third

personis opposedto the tandemof fIrst and secondpersons.Recall that the
lone third-personform, di '(s)he; that', also participatesin the demonstrative
system.

man man

di
 KAMASSIAN 589

The axis-of-discourseforms build yet anotherT-structure:the singularfIrst-

and second-person forms (ma-Ita-) on the one hand, and the dual and plural
forms on the other(mi-Isi-).
The forms of the s1, s2, and pi personalpronouns are transparent.Of
Uralic origin are the stems(ma-, ta-), the pronoun-formingsuffIx =n, and
probably also the plural suffix -1 « *-t). In the d2 form we have a suffIxal
element-ste (from the numeralside 'two'); the basesi- of the d2 andp2 forms
is thought to come from proto-Samoyedic*te, the non-singularbase for
second-person pronouns.BesideKamassian,Koibal alsohasan initial sibilant
in this pronoun(p2 ce 'you, your'); Mator doesnot.
The dual forms dl miste, d2 siste, and d3 diside clearly consist of the
pronominalstemsmi-, si-, and di- plus the dual markerside - ste mentioned
above. The case forms of the singular axis-of-discoursepronouns are as
follows:

sl s2
N/G man tan
AlLILO mana tanan
AB manatie1 tanatan
INS manzie1 tanzie1

In the axis-of-discourseparadigm, the syncretism of the N and G is

opposedto that of the A, L, LD, resulting in only a partial overlap with the
relative paradigmof nominal inflection. In the pronominalparadigm,the A
patternswith the oblique cases(LO and L) and is set againstthe unmarked
NIA form; this contrastssharplywith the relative nominalparadigm,in which
NIA/G is set againstLOlL. It would appearthat the markednessof the A is
more important in axis-of-discoursecontextsthan it is at the more neutral
level of substantives(grammatically:third person).
Whereasthe paradigmof axis-of-discoursepronominalsshowssimilarities
with the relative nominal paradigm,the inflection of the s3 pronoundi, which
is, at the sametime, a demonstrative,is identical with the nominal absolute
paradigm,i.e. thereis no syncretismwhatsoever:

s3
Nominative di
Genitive din
Accusative dim
Lative dime
Locative di1;m
Ablative di1;]1
Instrumental dizie1
590 KAMASSIAN

OtherPronouns
The reflexive pronoun (sl bosp~, bost~, s2 bosl~, bost~, s3 bost~, bost~, etc.) consists of the
relative paradigm of a noun of Turkic origin (bos 'body'). Interrogative
pronouns include simd~
'who?', ;nnbi 'what?', ;nnbi ser 'what kind of?',
sombi 'what (else)?',gin? 'which (of the two)?', mo(?) 'why?', kajet 'what
kind of?', kad~
'how?',kiimen 'how many?',kamen'when?'.

Adverbials
Theseconsistof adverbsproper and postpositions.Adverb-forming suffixes
are -i - -j, as in paardej 'around' (cf. paarziam 'I tum [myself]'), njeeri 'by'
(cf. ~ eer~ziem 'I miss [when shooting],),pinzi 'over', sam~j
'near,along'.
Most postpositionsnormally occur with their noun in the genitive, e.g.
dja}U-n kunzuRIVER-sGALONG 'alongthe river', pa-n ~ukut
FOREST-sG
THROUGH 'through the forest', kuza-n naam;m MAN-sG FOR 'for the
man'. The accusativeis required by the postpositionnieeri, e.g. koonu-m
nieeri BEAR-sA BY '(passing)by a bear'.Postpositionsmay also occurwith
their noun in the nominative,and may take caseand personsuffixes, as in ki
jil-g;m-d~ MOON UNDER-LO-s3 'underthe moon'.

Verbal inflection
Alongside the categoriesof person and number, which they share with
nominal and pronominalinflection, Kamassianverb stemsalso encodetense-
and-moodand the category of the definite object. All four categoriesare
representedat the morphologicallevel, e.g. ilbe-m LIES-sl 'I lie', i?be-je?
LIES-p3 'they lie', i ?be-? LIES-s2imp 'lie!', paarga-t-(i) CUTS-dO-s2imp
'cut it!'
The categoryof aspect,on the otherhand,is expressedby syntacticmeans,
i.e. with the help of auxiliaries; this contrastssharplywith the situationin the
other Samoyediclanguages.Such aspectualsyntagmsare made up of the
gerundof the lexical verb plus auxiliary verbs such as i?be- 'lies' (imperfec-
tive aspect)orwalja- 'leaves,abandons'(perfective).Examples:imperfective
kuja djemd~laa?b~ 'the sun is shining' < djemd~=laa
tight'
i?b~ SHINES=ger
~jerlaawaziam
LIES-s3, perfective 'I have tied it (up) tight' < ~fer=lljaa tight'
walja-m TIES=gerLEA VES-sl.
This spillover of verbal semanticsinto the syntacticdomainis perhapsthe
result of a functional push-chain:the tensemarkersfor the presentand past
tense are themselvesrelatively recent intruders into the domain of verbal
inflection, being of derivational origin (they are historically participles and
gerunds),e.g. presenttense kallam 'I go' < kan=la-m GOES-pres.part-sl,
pasttensekambiam'I went' < kan=bi-amGOES-past.part-sl.
From a strictly morphological point of view there are only two modaV
temporalformations of the Kamassianverb: the imperativeand the adhorta-
tive. Thesetwo moodsare in complementarydistribution: the imperativehas
forms for the secondand third person, and the adhortativehas only first-
 KAMASSIAN 591

person forms. Examples of imperatives: kana-l 'go! (s2)" kay-g;}-j 'may

(s)he go', kay-g;}-l;;Jj 'may they (both) go', kay-ga-ji1 'may they (all) go'.
Adhortative:kan-za-b;;Jj GOES-adhort-d1'let's both go!'
The conditional, on the other hand, is a syntheticform consistingof the
conjunctivemorpheme-na- (- -nii- - -da- - -dii- - -ta- - -tii-) plus a fossilized
archaicpreteriteform i-zii from the verb i-zet 'is', e.g. ilbeniimzii 'if! lay' <
i1be-nii-m i-zii-fJ LIES-conj-sl IS-pret-s3.The conjunctive suffix probably
continuesthe Uralic conditional/potentialmodal suffix *-n V-.
As a result of the influence of neighbouringTurkic idioms, converbal
phrasessuch as m;;Jllaand;;J'}f1m 'I go' « m;;Jfl=la + xand;;J='}f1-m WAN-
DERS=lst.ger+ GOES=part-s1)occur quite frequently. The auxiliary verbs
met with most often in such constructionsare (cited in the first person
singular)kalam, xand;;J'}f1m, (infinitive kanzet) 'I go', ilbem 'I lie', amnam 'I
sit', nU'}f1m 'I stand',and waPam 'I leave'; theseare usually precededby the
lexical verb in its first gerund form, built with =la(- =lii; see p.592). As
mentionedabovethesephrasesalsofunction as meansof expressingaspectual
distinctions.
The modal auxiliary moolJam (- ma[iam) 'I am about to, wish to, would
like to, am able to X' is also frequently used,e.g. kamn;;Jld;;J moolJam 'I am
going to [cure with] smoke, 1 would like to [cure with] smoke', in which
kamn;;Jl=d;;J is the participle of kamn;;Jl=[ia-m 'I smoke'.SeealsoAffirmative
Auxiliaries, pp. 593f.

Derivation
Aktionsart,genusverbi, and otherVerbal Constellations
The complementarycategoryto aspect,that of Aktionsart or mode of action,
is expressedin a typically Uralic way, viz. by meansof derivationalsuffixes.
Thus the derivational suffix =r forms augmentative-frequentatives, as in
amo=r=la-m 'I eat frequently' (cf. am=nJa-m 'I eat'), and the suffix =lu1
forms momentaneous-inchoatives, as in dioor=lu1=[ia-m 'I begin to cry,
burstinto tears'(cf. dioor=IJa-m 'I weep').
The categoryof genus verbi has the passiveas its markedmember.This
is expressedby the suffix =00-, as in Unii saroona 'the horseis tethered',in
which sar=oo=na is the first participle of the passivependantto sar=IJa-m
'I tie'. Reflexivity is alsoexpressedby this suffix, asin amn=oo=[ia-m 'I seat
myself, 1sit (down)', cf. amn=na- 'sits'.
Other constellationsof subject and object, such as causatives,are also
expressedby meansof derivation.Straightforwardcausativesareformed with
=d;;J-(- =t;;J-), e.g. diemd;;JlJem 'I heat X', cf. dieem=nJ;}. 'X is warm'; an
example of a more complex, quasi-causativerelation is tusu[ilJom «
tusu=I=IJo-m ) 'I teach[= 1 makeX understand]" formed with the suffix =1-,
cf. tusu=[ie-m 'I understand,1 learn'.
592 KAMASSIAN

Modality is also expressedby means of derivation. For example, the

desiderative,a mood roughly synonymouswith constructionsbuilt with the
modal auxiliary moolia- (above),is formed with the suffix =nz~ =z~), (- =z~),=z~),
e.g. kono=nza=li~m~m'I would like to sleep',cf. kuno=lia-m 'I am sleeping'.
Gerunds and Participles
The distinction betweenthe categoriesof gerund and participle is one of
syntactic function rather than morphology: the term 'gerund' applies to
deverbaladverbs,while 'participles'aredeverbalnouns(andadjectives).
The mostfrequentgerundivesuffixes are:

1 =la(?) - =lii( ?), after nasals: =na(?) - =nii( ?), which expresseseither
simultaneity,e.g. nee kunoolamna< kunoo=la + am=na-fJ 'the woman
is sleeping'or finality, e.g. ka=la?nerbe-?GOES=gerTELLS-s2imp 'go
[ =havinggone] tell (him/her)';
2 =bi - =pi, followed by personalforms of the lative case,also expresses
simultaneity,e.g. man amor=bi=nii di soo-bi-fJsl.PROEATS=ger-sllat
s3.PROCOMES-pret-s3'while I was eating,(s)hecame';
3 =bi (- =pi) plus =za (- =zii) expressesanteriority,e.g. di nii-m ku=bi=za
tunold;;,=la soo-bi-fJ THIS CHILD-A SEES=having RUNS=ger
COMES-pret-s3'havingseenthis child, (s)hecamerunning'.

Suffixes (1) and (2) also function as markers of presentand past tense,
respectively. In the case of suffix (1), there is an additional fine point:
palatalizationof the lateral (li) seemsto mark presenttense,while the plain,
unpalatalizedl indicatesfuture, e.g. (1) FUTURE nie?-le-m 'I shall blow',
(11) PRESENTnie?-zie-m'I blow', (2) PAST soo-bi-fJ '(s)hecame'.
The mostfrequentlyattestedparticipial derivativesare

4 =na - =nii, after consonants:=da - =dii, which forms nomina agentis,

e.g. bil-}'iin nu=na pa WATER-Ioc STANDS=partTREE 'a tree which
standsin water';
5 ='}Cl- =,a, after consonants:=ga - =gii, which forms presentparticiples
with active or passive meaning, e.g. pu?cP;;,='}Cl til? 'stinking bug',
tiemd=gii ma?'heatedtent'.

Suffix (5) is also used as a present-tensemarker, e.g. nU='}Cl-m 'I am

standing'.

Syntax
The basic order is modifier before modified. Thus adjectiveprecedesnoun:
bilii kuza 'poor man', and adverb precedesverb: sojcPoo-ze? sojcPoo-ze?
su?bt~bi-fJ
BASKET-INS SCOOPS-pret-s3'(s)he scooped(it) with a basket'. Object
precedesverb: siljmil-bil xeer-bi-fJ MARE-A TIES-pt-s3 '(s)he tied the
 KAMAS SIAN 593

Table 19.6 A sampleparadigmofverbalinflection: l1ntn 'to lie', infinitive

ilbezet,gerundilbelii?, '-ing' ilbij

Singular Dual Plural

IndicativePresent 1 i ?b;}ffl i?beb;)j i?bebiiii?

2 i ?b;)l
3 i?b;) i?bebiiii?
Preterite 1 i?bebijiim
2
3 i?bebi
Adhortative 1 i?beib;)j
Imperative 2 i?b;)? i?begul;)j
3 i?b;)g;)j
Conditional 1 i ?beniimzii
2 i ?beniilii?zii?
3

mare',as doesthe negativeauxiliary: ej surarga '([s]he) doesn'task'. Subject

generallyprecedespredicate:me1Jgejto ?btoobi 'the fox came',but the reverse
order occursin sentenceswith focalization, e.g. axsa?i-bi-fJ sejmu-tLAME
IS-pret-s3MARE-s3 'lamewas his mare'.
Certain featuresof the direct object may be encodedmorphologically on
the verb. This category is traditionally termed [+/-] definite, but the
'definiteness'of a Kamassianobject is evidently dependenton semanticand
pragmaticfactors about which we have only limited information. Whatever
its precise content, the formal distinction [+/-] definite obtains only if the
object is in the third person(but objectsof imperativesmay be both definite
and axis-of-discourse)and then only in forms with a third-personsubjectin
the indicative present/futureand conditional, and in second-person singular
forms of the imperative.Definitenessof the direct object is indicatedby the
morpheme -t (- -d, from the third-person pronominal stem *te), non-
definitenessby its absence.So for example we have in the presenttense
paarga-1Ja-fJ-t'(s)hecuts it' as opposedto paarga-lia-fJ '(s)hecuts', and in
the imperativepaarga-t 'cut it!' versuspaarga-?'cut!', but in the pasttense
there is only paarga-bi-? '(s)he cut (it)'. The definitenessmarker occurs
betweenthe two elementsof the conditional complex (above), e.g. paarga-
na-t-sa '(s)hewould cut it' v. paarga-na-za'(s)hewould cut'.
Three basic types of auxiliary verb may be distinguished: aspectual,
affirmative, and negative. The first of these was discussedunder Verbal
Inflection, p. 590.
The primary affirmative auxiliary is i- gii- 'is'. This may be usedas copula,
e.g. man tan ka}f2-1 i-gii-m 'I am your brother', with zero variant in the third
person,e.g. di ko?bdos3.PRODAUGHTER 'sheis a daughter/girl',but it is
594 KAMASSIAN

also used to express possession, e.g. ko?bdo-t buz!~n na'fUr ko?bdo-t i-bi-0
OLD.MAN-G THREE DAUGHTER-s3 IS-pret-s3 'the old man had three
daughters'.
In either of its two roles the verb 'is' is normally situatedat the end of the
sentence,although various textual/pragmaticeffects (focus, foregrounding)
may be achievedby inversion, as in to?bdZi-t mana, muner-dmana, i-ga-0
mana 'strike me down, breakme up, I still have.. .' .
We may alsoclassify the verb mooPa-,first mentionedabovein connection
with converbal phrases,as an affirmative auxiliary. The lexical verb then
stands(1) in the infinitive, as in man am=z;n moolja-m 'I would like to eat',
(2) in the participle in =na, e.g. samo=nu mooPa-m 'I am going to
shamanize',or (3) in the gerundin =la, e.g. o?blej « o?b=la + ej) moofia-m
COLLECTS=gerNEG.PARTAFFIRM.AUX-sl 'I can'tcollect'.
Negative auxiliaries are used to express negation. We know of two
negativeverbsin Kamassian:oneis the inflectablepair a-/e- (non-imperative)
- i- (imperative,i.e. prohibitive), for generalnegation;the otheris na}!l-, with
a defectiveparadigm,usedto negateexistenceandpossession.
Generalnegationis expressedin two ways. In the older construction,the
negative auxiliary a-/e- - i- is inflected for mood and tense, number and
person; it also encodesthe [+1-] definite distinction for direct objects. The
lexical verb standsin the connegative,a form built with the suffix -? and
thus homophonouswith the second-person singular (indefinite) imperative.
As Kamassianbecamemoribund, this constructionbeganto be rivalled by
one built from the invariable negativeparticle ej (itself historically from the
negative auxiliary) plus the regularly inflected form of the lexical verb. In
both constructionsthe negatornormally precedesthe lexical verb. Examples:
man e-m so-? sl.PRO NEG.AUX-sl COMES-conneg 'I don't come',
tan e-l-l;;J su-? s2.PRO NEG.AUX-pres-s2 ENTERS-conneg 'you don't
enter', bos-tu ej di xam-bi-0 SELF-s3 NEG.PART s3.PRO GOES-pret-s3
'(s)he himlherself didn't go', i(-?) tFoora-? PROHIB.AUX(-s2imp)
CRIES-conneg'don'tcry!'
The negativependantto the verb 'is' (i-, above)is na}!l-, with a somewhat
defectiveparadigm.It is usedto negatepredicates,as in bila kuzamanna}!l-m
BAD MAN sl.PRO ISN'T-sl 'I am not a poor man', and to expressnon-
possession,e.g. ipak di-n na}!l-0 BREAD s3.PRO-GISN'T-s3 '(s)hedoesn't
have any bread'; notice also di-za-n am=Z;;Jt na')O?-bi-0 s3.PRO-plur-G
EATS-infISN'T-pret-s3'they didn't haveanythingto eat'.
Historically, the generalauxiliary negativea-/e- - i- may well be related
to the verb 'is' (i-). On the synchronic plane, we would then have the
opposition indicative v. prohibitive based upona metonymic axis, i- 'is'
occurring in sentence-finalposition, i- 'don't!' occurring non-final; while
within the domain of negation,the opposition of negationv. prohibition is
expressedby quite different means,namely the metaphoricalaxis of vowel
alternation, a-/e- v. i-. Examples: (sentence-finali-, existence) di i-ga-0
 KAMAS SIAN 595

s3.PROIS-pres-s3'(s)heexists',(non-final, prohibition) i so-1'don'tcome!';

(e-, negation)e-m so-1NEG.AUX-slCOMES-conneg'I'm not coming', (i-,
prohibition) i(-1) kuro-1 PROHIB.AUX(-s2imp) IS.ANGRY-conneg'don't
be angry!'

Stylistics
Thanksto the analysesof John Lotz we know Kamassianversification well:
parallelismof numerically defined phrasesand/or lines is the central device
of the Kamassianversificator, who in this respectdiffers little from other
Siberianbards. Becauseof the relative scarcity of Kamassianlinguistic and
folkloristic materialwe know little of the differencesbetweenthe varioususes
of the language: everyday v. ritual; if ritual, sung v. recited; if recited,
numerical or free; if free: everydayv.... e cosi via, da capo. For the same
reason,we know next to nothing of the miniaturedeviceswhich wordsmiths
(shamans,singers,storytellers)might haveusedfor the purposesof enchant-
ment, entertainment,or deception. I have already cited an example of
emotionallymotivatedmetathesisat the syntacticlevel in riddle 29 ( to 1bdzit
mana, miinerd mana, igii mana 'evenif you strike me, evenif you breakme,
I still have...' ,in which we seean atypical sort of regens-rectumword order.
Sincethe text is a folkloristic one, it is perhapsnot too daring to proposethat
breaking the basic word-order rule is part of the arsenalof artistic word-
arrangement,i.e. poetry. This is all the more true becausethis metathesisis,
at the same time, part of a parallelism which is built upon the syncretic
pronominalform mana 'me (A)' = 'to me (L)' as rectum and the preceding
verbal forms as regens. In the first two phrasesthe sequencemodifier-
modified is justified by the universalrule accordingto which a verb in the
imperative always comes first, but in the third phrasethe verb is in the
indicative, so the metathesismust be explainedby somethingelse. In this
instance stylistic/poetical reasonsunderlie. The syntactic manipulation of
linguistic material for poetic endsis frequentin the folk-poetry of the other
Samoyedicpeoples,as well.
As a further illustration, we shall now examineanothertype of manipula-
tion of linguistic materialcommonin the folklore of the Samoyedicpeoples.
The hero of the tale chosen as our sample text is called tarcaab;Jrdzii.
Mentioning this nameis part of the tale'sopeningformula: 'Therewas a poor
man. His name was -'. Perhapsnot unexpectedly,it is part of the closing
formula, as well. At the end of the tale, the antagonistkaan (the Emperor)
surrenders,and addressesour hero as follows: 'tartiaa-bartia, pol bartia'.
This somewhatdistortedvariant of the nameis embeddedin a parallelismof
which the frame is the mysterioussequencetia:

tar tiaa tar tiaa

pol tar tiaa
596 KAMASSIAN

We can identify the elementtj in the sequencetja as the third-personsingular

possessivesuffix of the syncretic forms N/G/A and LILO/AB. The former
also occurs in the form ti, the latter in the formform
ti~, according to Kiinnap
(1971: 157, 159). On the otherhandthe final a cannotbe identified with any
grammaticalelement:it is an inetymologicalelementwhich functions as the
nucleusof an expletive syllable which, at the sametime, indicatesthe end of
a grammaticalmorphemeor stem.
Having identified the frame of the parallelism, we may now tum to its
content.The stembar 'all' is repeated:this repetitionconstitutesthe bearing
pillars of the parallelism,i.e. the identical elements.The differing elements
(the parallel elements)occupythe slots precedingthe identical elements.The
first of theseparallelwords is possiblyto be identified with tar 'hair; feather'.
As for the second,thereare severalcandidates.In Donner's(1944) dictionary
we find (1) pOJI 'ground,floor' (from RussiannOJI), (2) POJIU?/i em 'to dive',
and (3) po?tiam 'to bathe, to dip, to swim, to run (water); to be on heat'.
Consideringthe hero'sfinal trick, by which he compelsthe kaan and all his
peopleto surrenderby drenchingthem with water pouredfrom a sack, thus
washingthem away, we can excludeitem (1). We are thus left with items (2)
and (3). Perhapsit is not too far-fetchedto supposethat thesetwo verbs have
in common a root pol-, with the meaning 'is immersedinto a stream; is
overcomeby water/sexualdesire'.A more literal translationof our parallel
lines would thenbe:

FEATHERIHAIR-s3 ALL-s3
OVERCOME ALL-s3

or, in paraphrase:'One who is coveredall over in featherslhairis sweptaway

by passion',or: 'The featheredman/hairyman is in a trance'.
Obviously both of these interpretationsrefer to, and are appropriate
descriptionsof, the shaman.The parallel arrangementof the constituentsin
this miniature couplet are reminiscentof a techniquein Nenets verbal art,
wherebythe words arecut up into their constituentsyllables/morphemes with
the help of expletives,therebyfilling out the metrical schemeand creatinga
patternin which the original messageis woven in and out of a quasi-syllabic
texture,like inlaid-work:

tar- bar-
tjaa- tia
pol bar-
• tia
 KAMASSIAN 597

i.e. schematically:

Seenin this framework,the fIrst variant of the name,tarcaab;}rdzii, is even

more interesting.Unlike the variant examinedabove,here the two halvesof
the name are set against one another as velar and palatal (back and front)
constituents,i.e. oppositepolesin tonal vowel harmony;this contrastin vowel
colour is accompaniedby the oppositionunvoiced : voiced in the variantsof
the third-personpossessivesuffIx, -tJa(a): it appearsas -caa and -dzii. Let us
now set asidetheseconstants (disguised as variables)in order to concentrate
on the real variablesin the parallelism: tar and b;}r. Even if the fIrst of these
(tar) is the samemorphemeas in the other variant of the name, the second
one, b;}r, is certainly not: it is a palatalword, as its suffIx -dzii plainly shows.
The interpretationsgiven abovewill not work here,then. If on the otherhand
we revert to our original procedure,and strip away the pseudo-possessive
suffIxe(l)expletivesto exposethe real variables,we arrive at the constella-
tion:

tar- b;}r-
caa- dza

in which we have not the compoundtar+bar 'hair/featherallover', as in the

earlier variant of the name,but rather tar-b;}r. This sequenceseemsto be a
cover for the word taarb;} 'shaman'.If we try to force taarb;} into the frame,
the fInal -r is assignedto the following pseudo-possessive suffIx, but for the
formal parallelismto remain,the fInal -r of the fIrst half must be re-assigned,
as well:

ta- b;}-
rcaa- rdza

Although the formal parallelismis maintained,the semantictransparencyis

blurred onceagain,sincethis time the final -r of tar hasslippedaway in spite
of the fact that it belongsto the etymologicalmaterialof the word. The result
is an acoustichide-and-seeksimilar to certainkinds of optical illusion based
on the complementarityof black-and-whitepatterns which one can only
identify one at a time: when you focus on one, the other is blurred, and vice
versa.
598 KAMAS SIAN

The parallelism tartiaa-bartiaJpol bartia and the variant of its first part,
tarcaab;;,rdia, both refer to the samedenotatum,but they do it in different
ways. The first parallelismtells a story while describingits topic; the second
tries to utter and, at the sametime, to conceala word. Yet both do the same:
poetry, since neither speaksdirectly. In addition to manipulations at the
syntacticlevel, Kamassianpoetic speechalso relies on transformationsat the
level of morphology(break-upof the word by expletives,splitting words into
constituentmorphs and intermixing them with inetymologicalphonic mate-
rial, etc.)just as do the northernSamoyedictraditions,NenetsandNganasan.

Sample KamassianText
Adaptedfrom Marchenno. 2, pp. 88-90,in Donner1944.

numak Tale
bila kuzaamnoo-bi-{j Therewas a poor man,
n;m-d;;, tarcaa+b;;,rdia his name[was] Tarcab;,rdia.
bila sejmu-budiaab-pi He took a poor mare,
axsa?i-bi-{j sejmu-t lame washis mare.
konzan=de-bi-{j He equipped[it]:
saay,Jrb;;,raa-ba saar-bi-{j [a] black sack[he] tied [to its
saddle],
sojdioo-basaar-bi-{j [a] birchbarkbaskethe tied [to it].
m;;,l=le 0 lbt;rbi-{j Going he started(= he setoff).
urtaba to ?btoo-bi-{j 'Grandpa'(= bear)came[along].
manaji-t hiilaa=s-t;;, 'Me take as a companion,
pada-?say,Jrb;;,raa-n;;,! put [me] into [the] black sack!'
pada-bisaapzb;;,raa- n;;, He put [the bear] into [the] black
sack
m;;,l=le 0 ?bt;rbi-{j Going he started,
meggejto ?btoo-bi-{j [a] fox came[along].
manaji-t hiilaa= 3;rt;;, 'Me take as a companion,
pada-?saay,Jrb;;,raa-n;;,! put [me] into [the] black sack!'
pada-bi-{j saay,Jrb;;,raa-n;;, He put [the fox] into [the] black sack
di-}i?-ti kusto ?doo-bi-{j Then a wolf came[along].
manaji-l hiilaa-z;rt 'Me takeas a companion,
pada-?saay,Jrb;;,raa-n;;,! put [me] into [the] black sack!'
di-}it-ti bu tolktoo-bi-{j Thenwatercame[along].
d;;, bu-msu?lej-bi-{j saay,Jrb;;,raa- This waterhe scoopedinto his black
nd;;, sack,
soj(]ioo-ze?sulbt;rbi-{j with a baskethe scooped[it],
sejme-nd;;,sar=lawa?-bi-{j to his marehe tied [it].
ku=bi-nd;, He sees:
talaj-nd;;, to ?-ond;;,jada nu-ta-{j on the sea'sshore[a] village stands.
 KAMASSIAN 599

jadaa-n;,ej tu-)fJ-@ Into [the] village he doesnot go.

iizii-bi-@ He got off [the horse],
sii [hJiim-bi-@ [a] fIre he made,
siijmii-bii xeer-bi-@ [the] marehe tied,
am-na-@ he sits [down to wait].
kaan njilnen m;,l=le?-bi-@ The Tsarwas walking outside,
ku=lalwa-bi-@ [and] saw:
sindi di-}in am-na-@ 'Who is sitting there,
inii-b;, xeer=lawa-bi-@ [who] tied [the] horse,
man kiik noo-batoonola?-bi-@ [who] trod my greenlblue/golden
grass
ej surar=ga not asking [for permission]
to ?-ond;,m;,nd;,njii-bi-@ on the side/roadpassed[his day]
;,?I;J-bi-@ [and] let [it] go [by]?
kal=la suraar-d;, Going, askit!'
sindi d;,rgit ii-gii-l 'What kind of man are you,
ej suraar=ga kaan-;mz [who] without askingthe Tsar
kaan-;,n nolb-to toolna-bi-@ trod the grassof the Tsar?
;mzbi tan-an karii? What do you want?'
kam-bi-@suraar=la?-bi-@ [The servant]went [and] asked:
kaj;,t kuzai-g;J-l 'What kind of man are you?
;mzbii-mpi=lii?m;,{)-g;J-l What are you looking for,
wandering?'
kaan-;,n ko?btoo-biii-zit soo-bija-m 'I cameto take the Tsar'sdaughter.'
d;, kuzapaar-bi-@ This man (= the servant)returned[to
the Tsar],
kaan-d;,neerbii-bi-@ to [the] Tsarhe answered:
kaan-;,n kolttoo-boii-sit soo-bija-m 'I cameto take the Tsar'sdaughter.'
kala- 1, neerbii-? 'Go, tell [him]:
jaks;J-l-zi paara-? 'Of your [own] good [will], go
back!'
;J-mpaara 'I won't go back.'
jakS;J-Z;'ej m;,=bi-nd;, 'If he doesn'tgo backwillingly,
bilii?-z ii-/i-m I shall take [it] with bad [feelings].
eeli-t kaaz;,r asbr-i-m Releasethe wild horses,
tal;,j-din tal;,j-din
s;,jm;J-t-~i with your marethey will knock you
down,
nje?le-din they will tearyou apart.'
u?dulu-bi-@kaaz;,r asbr-za-d;,n He let their wild horsesrun,
ask;,r-ie?tunoo=la xand;J-)fJ-i? [When] wild horsesarerunning
saaprb;,raa-bo Cik;J-bi-@ he opened[the] black sack
saaprkoon;J-multullu-bi-@ he let out the black bear.
ur)fJaba ask;,r-iji-m pja{)d;,l=la? 'Grandpa',following the horses,
kuna-mbi chasedthem away.
600 KAMASSIAN

lbtul-gu-t buga-li-m, kaaz;,r-iji-m 'Releasethe bulls [and] horsesto

sejmit-t(~e buga-tjit}tl mitla trample[him] to deathwith his
mare!'
saaprboraa-b;,tiib-bi-fJ [The] black sackhe opened,
kus-tu ;,l-la m;J-bi-fJ his wolf he released.
kusbu}tl-ji-m pa1Jd;,=lal ku-mbi [The] wolf, chasingthe bulls,
pa-m;'1Jg;m pursuedthemin [to] the forest.
kaan maala kojoo-bi-fJ [The] Tsar,remaining,stayed,
iil obt;J-bi-fJ paard;,j [his] peoplegatheredaround[him]:
i-s-piil d;J-m 'Let's take him!'
paar;,ldlul-bi-?ji They surrounded[him],
diab;,=n;, mazia-?ji wantedto grab [him].
saaprboraa-b;,tiike i-bi-fJ, niept;J- [The] black sackhe opened,pulled,
bi-fJ
bit-bit kolmn;,j ba?pt;J-bi-fJ shaking[the] waterpoured[it onto
them].
iilii-m bit kund;J-}tl-t xand;J-}tl-t [The] waterchasesthe folk away.
tartiii&bartia, pol bartia 'tartia&bartja,pol bartia,
piil mal-b;, m;J-lii-m half of my cattle1 shall give [you],
piil baj-bu m;J-zii-m half of my empire1 shall give [you],
ko?bdoo-mm;J-zii-m my daughter1 shall give [you].'
ej karii manatan mal-I;, 'I don't need/wantyour cattle,
ej karii tan baj-l;, 1 don't need/wantyour empire,
tozikoabakaj kolbdo-l karii mana only your princessdaughter1 do
need/want.'
il-bi-fJ sojdioo-bo He took [the] basket,
bit-bit sulgda-bi-fJ poured[out the] water.
kaan kolbdoo-b;, m;J-bi-fJ The Tsargave[him his] daughter.
axsas;,jm;J-nd;,Silt;J-bi-fJ He sat [her] on to the lame mare,
boku-lla u?bt;J-bi-fJ fastening,he mounted,
mal-and;,kunnaaam-bi-fJ to his tent he took [her].

Referencesand Further Reading

Note: all examplescited in this chapterare takenfrom Donner1944.
Castren, M.A. (1854) Grammatik der samojedischenSprachen, St. Petersburg.
(Re-issued 1966,with a foreword by P. Hajdu, as volume 53 of the Indiana
University Uralic andAltaic Series,Bloomington:IndianaUniversity).
Donner, K. (1944) KamassischesWorterbuchnebstSprachprobenund Hauptziigen
der Grammatik, bearbeitet und herausgegebenvon A. Joki, Helsinki, Lexica
SocietatisFinno-Ugricae8, Helsinki: SocieteFinno Ougrienne.
Hajdu, P. (1950) 'Die Benennungender Samoyeden',JSFOu5411: 1-112.
- - (1963) The SamoyedPeoplesand Languages,Bloomington: Indiana Uni-
versity; The Hague:Mouton.
- - (1968,21982)ChrestomathiaSamoiedica,Budapest:Tankonyvkiad6.
 KAMAS SIAN 601

- - (1975) 'Dber die Herkunft des kamassischenPluralsuffixes -salJ', UAJb 47:

85-8.
Janhunen,J. (1977) SamojedischerWortschatz. Gemeinsamojedische Etymologien,
Castrenianumintoimitteita 17, Helsinki: Castrenianum.
- - (1986) Glottal Stop in Nenets, MSFOu 196, Helsinki: Societe Finno-
Ougrienne.
Joki, A. (1952) Die Lehnworter der Sajansamojedischen, MSFOu 103, Helsinki:
SocieteFinno-Ougrienne.
Ktinnap, A. (1971) Systemund Ursprung der Flexionkamassischensuffixe, vol. I:
Numeruszeichenund Nominalflexion, MSFOu 147, Helsinki: Societe Finno-
Ougrienne.
- - - (1978) Systemund Ursprung der kamassischemnFlexionsuffixe, vol. II:
Verbalflexion und Verbalnomina, MSFOu 164, Helsinki: Societe Finno-
Ougrienne.
- - - (1993) KaMacllHcKliH H3hIK in: YpaJIhCKlie H3hIKll, Moscow.
Lotz, J. (1953) Structural Presentationof the KamassianLament,Stockholm:Thesis
Instituti Hungarici UniversitatisHolmiensis.
- - - (1954) 'Kamassianverse',Journal ofAmericanFolklore 67: 369-77.
Simoncsics,P. (1986)AenigmaCamassica,Pomaz:Editio propria. 7 pp.
- - - (1988) 'Rejtelyes li1iom', in: Studia in honorem P. Fabian, F. Racz, I.
Szathumanob/ataa coUegiiset discipulisBudapest:n.p., pp. 125-32.
- - - (1993a)'Sziberiaidelikatesz'[= 'A Siberiandelicacy.Yet anotherKamassian
riddle analysed'],Nyelvtudomdnyikozlemenyek 91: 195-202.
- - (1993b) 'Kamasszvilag - ahogy a talal6skerdesekmutatjak' [= The World
of the Kamassian- as it is seenthroughriddles'], in M.Sz. Bakr6-Nagyand E. Szfj
(eds), Hajdu Peter 70 eves, Linguistica Series A, Studia et dissertationes15,
Budapest:MTA NyelvtudomanyiIntezet,pp. 371-82.
Index

abessivemood Udmurt 281,286,297,298,446

Nganasan503 Aktionsart seemodeof action
adhortativemood anaphoricpronounsseepronouns,
Kamassian590-1, 593 anaphoric
Khanty 374 aorist
adjective18-27 Samoyedic472
Estonian126, 127, 129-30, 134, 137, aspect
142 Estonian446
Hungarian438 Fennic III
Kamassian592 German446
Komi 313, 314 Hungarian446, 451
Mansi 411-12,414 Kamassian590,591
Mari225,234,237,242 Khanty 368, 376
Mordva191,195-6, 198,205,207, Mansi 420
210 Mari 229
Nenets526 Mordva 197, 208-9
Nganasan497,499-500,511 Nganasan503
Saamic63,81,84,87 Saamic80
Selkup560, 562-3, 574, 575 Selkup566
Udmurt 281,282,285-6,294,295, Slavonic446
297 auditive mood
adverb Nenets535
Estonian136, 139, 142,145 Selkup566, 568
Fennic 110, 113 auxiliary verbs28, 29
Finnish 157, 170, 171,446 Finnish 172
Hungarian445-6,451 Hungarian451
Kamassian590, 592 Kamassian591,593-4
Khanty 367-8,369, 379-80 Mari 231,238,239
Komi 446 Nenets543
Mansi 415, 417-18 Nganasan508,511
Mari 232, 237,239-40,241,242,243 Samoyedic475
Mordva 198, 204-6, 209 Selkup575
Nenets525, 526,540-1, 542
Nganasan497,501 Baltic-FennicseeFennic
Permian446
Saamic63, 68, 83, 84, 85,86 CheremisseeMari
Selkup561, 562, 571, 572, 574, 575 clitics

602
 INDEX 603

Finnish 177 Komi 322

Hungarian441 Mansi 416, 421-2
Komi 314 Mari 242, 243
Mansi 236, 412 Nenets544
Mari224-5,236,243 Nganasan501, 512
Mordva201 Saamic89
Nenets520, 531,544 Selkup576
Nganasan497,500,508-9 conjunctivemood
Saamic62 Khanty 371
conditionalmood 28 Mordva 201,207,208
Estonian136, 137, 141 Nenets530-1, 533
Fennic 111, 112 consonantclusters483, 490, 491,492
Finnish 171 Estonian119, 123-4
Hungarian446-7 Finnish 153
Kamassian591, 593 Hungarian433
Khanty 371 Khanty 364
Mansi 398, 400, 419 Komi 310
Mordva 198,200,201,207 Mansi 395, 396
ObUgrian347 Mari 222
Saamic71, 77, 88 Mordva 186, 187
Selkup566, 568 Nenets521, 525
Udmurt 289-91 Permian260
conjugationof nominals Saamic57, 60, 61
Nenets539 Selkup 554
Samoyedic470 Udmurt 280,301
conjugation,definite/indefinite consonantgradation
Mansi 401-3 Estonian124-5, 126, 127, 129
Mordva 197-202 Fennic 105-6, 110
conjugation,determinate/indeterminate Finnish 152, 153-4, 155, 159, 160,
Mansi 398-401,401-3 175
conjugation,finite Nganasan487, 490-2,494
Samoyedic471-2 Saamic59, 57-8, 66-7
conjugation,objective consonantharmony
Hungarian447 Mordva 190, 192
Khanty 371 consonantlength seelength
Mansi 401-3 copula
Nenets532, 533, 534 Komi 320
Nganasan502 Mari 230, 240
ObUgrian347-8 ObUgrian353, 355
Samoyedic470 Saamic76, 77-8, 85, 86, 88
Selkup568, 576 Cyrillic script 43,100,221,222,278,
conjugation,periphrastic 310, 393,483, 486
Samoyedic472-3
conjugation,subjective debitive mood
Hungarian447 Nganasan503
Nenets532, 535 Selkup566, 568
Nganasan502 definiteness
Samoyedic470 Finnish 156
Selkup568, 576 Kamassian593
conjunctions Khanty 371
Estonian143 Mansi 417
Fennic 104 ObUgrian347-9
Kamassian588 Selkup 562
604 INDEX

deixis 25-6 Nganasan481-2,484

Finnish 154, 170,171 ObUgrian332, 338, 339,340, 341,
Hungarian438, 442, 443-5 343,345,348,353
Kamassian588 Saamic56, 79, 462, 561, 576
Khanty 369-71,379 Selkup549-50
ObUgrian354 Udmurt 279,280,281,289,301
Saamic84 Votic 109
Selkup565 dual
Udmurt295 Kamassian586, 588, 589,593
demonstrativepronounsseepronouns, Khanty 361, 365, 366, 367
demonstrative Mansi 398, 399,400,401,402,403,
denominalderivationseederivation, 404,410,419
denominal Nenets529, 532, 533, 537
derivation,denominal Nganasan495, 502
Hungarian452 ObUgrian340-1,348
Khanty 382 Saamic62, 69, 70, 75, 76,77, 90
Mansi 409 Samoyedic469-70
Mari 233-4 Selkup558-9
Mordva202
Nenets541-3 Enets167,457-79,481,482,513,550,
Nganasan509 577
Saamic81, 83 consonants9-13, 462-3,466--7
Selkup571-2 diachronicevaluation460
derivation,deverbal documentation459-60
Hungarian452 Forest517
Kamassian591-2 lexicon 475-8
Khanty 381-2 morphology18-32,468-75
Komi 317-18 phonology462-8
Mansi 408, 414 taxonomicrelationships458-9
Mari 234--5 Tundra517
Mordva 203-4 typologiccalvariation460-1
Nenets535, 541-3 vowels 13-18,463-4,465-6,467
Nganasan510 Estonian76, 78,96-114,115-48,149,
Saamic79-80, 80-2 437,446,448
Selkup574 adjective126, 127, 129-30,134, 137,
desiderativemood 142
Kamassian592 adverbs142
Mari 229 alphabet118
Mordva 201,207,208 aspect446
Nenets530-1, 533 conditionalmood 136, 137, 141
deverbalderivationseederivation, conjunctions143
deverbal consonants9-13,109-10,123-4,
dialects6--7 124--5, 126
Estonian108-9, 115-16,121-3, 137 demonstrativepronouns133-4
Fennic 105, 107, 109,111, 113, 168 dialects101-4, 108-9, 115-16,
Hungarian429 121-3,137
Khanty 358-60,371, 383 gerunds136, 137, 139,140
Komi 305-8,310, 311, 312, 314,316 imperativemood 136, 137, 141
Livonian 108-9 indefinite pronouns135
Mansi 362, 383,390-2,395,422,423 indicativemood 136, 137, 140
Mari 226 inferential mood 137, 139, 140
Mordva 185, 186, 187,192 interrogativepronouns127, 134-5
Nenets519, 521, 523, 524, 535,539 jussivemood 136, 137, 141
 INDEX 605

lexicon 145-7 loan words 113-14

literary Estonian116-18,147 morphology18-32, 110-13
loan words 113, 145-7 negativeverbs 105
measure145 nominal inflection 110-11
morphology18-32, 110-13,124-7, non-finite verbalforms 113
127-43 participles Ill, 113
negativepronouns135 passive112
negativeverbs 113 phonology104-10
nominal inflection 127-9 plural 110, 111, 154
non-finite verbalforms 137-9 potentialmood 111
Northern 147 quantity 105-10
numerals126, 130-2 quotativemood 113
objectnouns145 simple sentence113
participles135, 137, 140 stl1ld 106
personalpronouns132-3 stress104-5, 106, 110
phonology118-27 syntax 113
plural 125, 127, 129,133, 134, 135 tone 106
possessivepronoun133 verb inflection 111-13
postpositionalconstructions135, vocabulary113-14
142-3 vowel harmony106-7
prepositions142-3 seealso Estonian;Finnish; Ingrian;
preverbs368 Karelian; Livonian; Veps; Votic
quantity 105-10,119-21 Finnish44, 71, 77, 78, 79,80, 89,91,
quotativemood 136, 137,141 96-114,115,117, 118, 145, 146,
reflexive pronouns133 149-83,184,202,220,221,245,341,
relative pronouns134-5 351-5,381,388,431,436,437,439,
simple sentence143-5 441,446,452
stress104-5, 106, 110, 119, 120, 133 anaphoricpronouns170
subjectnouns144 auxiliary verbs 172
syllable 118-19 c1itics 177
syntax 113, 143-7 conditionalmood 171
text 147-8 consonants9-13,109-10,152-6
tone 121 definiteness156
verbs 111-13,136-41 deixis 154, 170,171
vocabulary113-14 demonstrativepronouns170, 171
vowels 13-18, 106-7, 121-3,125, dialects101-4
126 gerunds174
word order 143-4 glottal stop 154
imperativemood 154,171
Fennic44,66,68,77,80,81,89,90,91, indefinite pronouns171
96-114,127,145,155,167,179,192, indicativemood 171
203,327,341,439,448,461,482 inferential mood 154,174-5
aspect111 infinitival constructions174-5
Baltic-Fennic262, 274, 316, 323 inflection 156-75
conditionalmood 111, 112 interrogativepronouns170
conjunctions104 lexicon 177-9
consonants9-13,105,109-10 loan words 113, 147, 179
dialects101-4, 105, 107, 109,111, morphology18-32, 110-13
113,168 negativepronouns171
gerunds111, 113 negativeverbs 172
imperativemood 111, 112, 113 nominal inflection 110, 156-71
inferentialmood 111, 113 non-finite verbalforms 174-5
jussivemood 113 numerals157, 168-9
606 INDEX

participles 154, 161,174-5 clitics 441

passive 174, 175 conditional mood 446-7
persona l pronouns 157,169 consonant s 9-13, 361,433-7
phonology 150-6 deixis 438,442,443- 5
plural 159,160,163,167,169 , 170 demonstrativ e pronouns 435,444
postpositiona l construction
s 163-7 denomina l derivation 452
potential mood 171 deverba l derivation 452
prepositions 163-7 dialects 429
prohibitive mood 174 gerunds 448-50
quantity 105-10 indefinite pronouns 445
reflexive pronouns 170 inflection of nominals 438-46
relative pronouns 170 interrogative pronouns 444,445
simple sentenc e 175-7 lexicon 452-4
stress 104-5,106, 110 loan words 452-4
syntax 113,175-7 morphology 18-32,437-5 0
text 179-81 negative pronouns 445
verb inflection 111-13,171- 4 non-finite verb forms 448-50
vocabulary 113-14 noun phrase 450-1
vowel harmony 106-7,150-1,16 3 number 438^10
numerals 370-1,438
gerunds objective conjugation 447
Estonian 136, 137, 139,140 participles 448-50,451
Fennic 111, 113 persona l pronouns 438,442-3,443^ 1
Finnish 174 phonology 429-37
Hungarian 448-50 plural 438^10,450
Kamassia n 584, 590, 592, 593, 594 possessiv e construction s 438,439,
Khanty 367-8, 376 451
Komi 316-17, 320 postpositiona l construction
s 334-6,
Mansi 42,415,421 368,442,445- 6
Mari 229-30, 232, 239, 242 preverbs 368
Mordva 204,205, 210-11 relative pronouns 445
Nenets 535, 536 subjective conjugation 447
ObUgrian 354 suffix order 367
Permian 273 syntax 450-2
Saamic 75, 78, 79 text 454-5
Samoyedic 473 verb inflection 446-8,448-5 0
Udmurt 293, 298, 299 vowels 13-18,429-3 3
glottal stop 11
Finnish 154 imperative mood 28
Kamassia n 584,585 Estonian 136, 137,141
Nenets 522, 523, 525, 526, 584 Fennic 111, 112,113
Nganasa n 484,487,491,494,49 5 Finnish 154,171
Selkup 584 Kamassia n 590-1,593, 594,595
Khanty 371, 374-5
hortative mood Komi 314,315
Nenets 533 Mansi 398,401,404-5,41 9
Hungarian 69,76,78, 151, 171, 184, Mari 229
203, 273, 285, 327, 341, 347, 350, Mordva 201-2, 207,208
351-5, 358, 359, 361, 381, 383, 388, Nenets 530-1,532, 533,534, 536
389,417,422,428-56 , 588 Nganasa n 502, 503,511
adjective 438 ObUgrian 339,347
aspec t 446,451 Permian 271
auxiliary verbs 451 Saamic 54,71,76-7, 78, 88
 INDEX 607

Samoyedic 474,475 Nganasan503,511

Selkup 562, 566, 568, 575 jussivemood
Udmurt 280-1, 289-91, 292, 300 Estonian136, 137,141
indicative mood 28 Fennic 113
Estonian 136, 137,140
Fennic 111,112 Kamassian(Kamas)457-79,580-601
Finnish 171 adjective592
Khanty 371, 372^1 aspect590, 591, 593
Komi 314, 315 auxiliary verbs591, 593-4
Mansi 398,402-3,405- 6 conditionalmood591, 593
Mari 229 conjunctions588, 591
Mordva 198, 207 consonants9-13, 462-3, 466-7,
Nenets 530-1, 532, 533 583-4,585
Nganasa n 503 definiteness593
ObUgrian 347 deixis 588
Permian 271 demonstrativepronouns588, 589
Saamic 71, 76, 88 derivation591-2
Selkup 566, 567 desiderativemood592
Udmurt 289-91, 292 deverbalnouns592
inferential mood 28 diachronicevaluation460
Estonian 137,139,14 0 documentation459-60
Fennic 111, 113 dual 586,588,589,593
Finnish 154,174-5 gerunds584,590, 591,592, 593, 594
Kamassia n 593, 594 glottal stop 584,585
Mansi 414,420 imperativemood 590-1, 593, 594,
Mari 233, 239 595
Mordva 204, 205, 207,209,21 0 inferential mood593, 594
Nenets 535-6, 544 interrogativepronouns590
Nganasa n 503 lexicon 475-8
ObUgrian 349-50 modeof action (Aktionsart) 591-2
Permian 273 morphology18-32,468-75,585-91
Saamic 75, 78,79, 84, 89 negative594
Selkup 566, 567, 576 nominalinflection 585-8
Udmurt 291, 299 participles590, 591, 592, 594
infinitival forms passive591
Finnish 174-5 personalpronouns588-9
Nenets 535 phonology462-8, 581-4
Saamic 89 plural 586-8, 589, 593, 594
Ingrian (Izhor) 96-114,115 postpositionalconstructions590
consonant s 9-13,109-10 pronominalinflection 588-90
dialects 101-4 reflexive pronouns590
loan words 114 reflexive verbs591
morphology 18-32,110-1 3 stylistics 595-8
nominal inflection 110-11 syntax592-5
quantity 105-10 taxonomicrelationships458-9
stress 104-5,106,11 0 tense590, 592, 594
syntax 113 text 598-600
verb inflection 111-13 typological variation460-1
vocabulary 113-14 verbal inflection 590-1
vowel harmony 106-7 versification595, 598-600
interrogative mood vowel harmony465-6, 582-3,597
Estonian 134-5 Karagassee Mator
Nenets 530-1, 533, 544 Karelian91, 96-114,274, 341
608 INDEX

consonants9-13, 109-10 text 383-4

dialects101-4 TremjuganffromaganKhanty 362-4,
loan words 114 383-4
morphology18-32, 110-13 verbs371-6,419
nominalinflection 110-11 vowel harmony362, 365
quantity 105-10 Komi (Zyrian) 150,211,267,278,279,
stress104-5, 106, 110 305-26,341,351,361,364,383,389,
syntax 113 390,423,431,442,446,452,517,
verb inflection 111-13 518,549
vocabulary113-14 adjective313, 314
vowel harmony 106-7 clitics 314
Ket383,513,517,549,550,577 compounding318
Khanty (Ostyak) 157,211,327-57, conjunctions322
336-8,358-86,389,390,398,422, consonants9-13,308,309-10,361
423-4,428,436,437,438,446,448, copula320
517,518,549 demonstrativepronouns313-14
anaphoricpronouns369 derivation316, 317-18
aspect368, 376 deverbalderivation316
conditionalmood371 dialects305-8,310, 311,312,314,
consonants9-13, 361, 363, 364, 365, 316,524
366 gerunds316-17,320
definiteness371 imperativemood 314, 315
deixis 369-71,379 indicativemood 314, 315
demonstrativepronouns369-71 interrogativepronouns314
denominalderivation382 lexicon 323-4
deverbalderivation381-2 loan words 323-4,382
dialects358-64,371, 383, 524 modeof action (Aktionsart) 316
duaI361,365,366,367 morphology18-32,311-18
gerunds367-8, 376 morphophonology311
imperativemood 371, 374-5 negativepronouns314
indefinite pronouns370 negativeverbs315, 316,320-1
indicativemood 371, 372-4 non-finite verb forms 316-17,323
interrogativepronouns370 non-simplesentence322-3
lexicon 381-3 noundeclension311-13
loan words 364, 382-3 noun phrase318-19
morphology18-32,361-2,364-76 numerals311, 314, 320
negativepronouns370 participles316
nominal paradigms365-7 passive320
non-finite verbalforms 375-6,382 personalpronouns311, 313
nounphrase376-7, 378-9, 380-1 phonology308-11
numerals370-1 plural 311, 312, 313, 316,320
objectiveconjugation371 possessiveconstructions313
optativemood, 371 postpositionalconstructions312, 313
participles375-6 pronominaldeclension313-14
passive371, 372, 376-81 reciprocalpronouns313
personalpronouns368-9 reduplication318
plural 361, 365, 366, 367 reflexive pronouns313
postpositionalconstructions367-8, relative pronouns314
376 simple sentence321-2,322-3
preverbs367 stress308, 310-11
sampleinventories360-1 syntax318-23
simple sentence380 text 324-5
syntax376-81 verb conjugation314-16
 INDEX 609

verb phrase319-21 ObUgrian351-2

vowels 13-18,280, 311 Permian273-5
Saamic90-2, 477
Lapp seeSaamic Selkup576, 577
length,consonant Udmurt 295,301-2
Estonian119-21 Veps 114
Finnish 153 Votic 114
Hungarian433,437 Lude seeKare1ian
Komi 309
Nenets528 Mansi (Vogul) 171,203,327-57,334-5,
Nganasan487,494 358,361,387-427,428,436,442,
Saarnic58-9 446,517,577
length, vowel adjective411-14
Estonian119-21 adverbs415
Fennic 105 aspect420
Finnish 159 c1itics 236, 412
Hungarian429,430,431 compoundsentences421-2
Khanty 362-3 conditionalmood398, 400, 419
Mansi 392, 393 conjunctions416, 421-2
Mari 224 consonants9-13,220-2,361,393-5,
Nenets519-20,521, 522 396
Nganasan485,486 definiteness371, 417
ObUgrian329, 330,335, 336-7 demonstrativepronouns403, 413
Saamic53, 54,58 denominalderivation409
Sarnoyedic467, 468 developmentof Mansi languageand
Selkup553, 557 people389
Livonian 96-114, 115 deverbalderivation408, 414
consonants9-13,109-10 dialects359, 360, 362, 383,390-2,
dialects101-4, 108-9 395,422,423
morphology 18-32,110-13 dual 398, 399,400,401,402,403,
nominal inflection 110-11 404,410,419
preverbs368 gerunds 42,415, 421
quantity 105-10 history anddevelopment387-9
st!ild 106 imperativemood 398, 401, 404-5,
stress104-5, 106, 110 419
syntax 113 indefinite pronouns413
verb inflection 111-13 indicative mood398, 402-3,405-6
vocabulary113-14 inferential mood414,420
vowel harmony 106-7 interjections416
loan words interrogativepronouns413
Estonian113, 145-7 lexicon 422-4
Fennic 113-14 linguistic sources388-9
Finnish 113, 147, 179 loan words 382, 383,422-4
Hungarian452-4 morpho(no)logy395-416
Ingrian 114 negativepronouns413
Karelian 114 noun phrase236-7,419
Khanty 364, 382-3 nouns409-11
Komi 323-4,382 numerals370, 412, 414
Mansi 382, 382,422-4 objectiveconjugation401-3
Mari 227, 244-5 participles414-15,418, 420-1
Mordva 185,186,188, 191,211-15 passive398, 403-5,406
Nenets520, 524 personalpronouns398,412-13
Nganasan512-13 phonology392-5
610 INDEX

plural 398, 399,400, 401, 402, 403, postpositionalconstructions236-7,

404,410 241,243
poetic devices381 simple sentence240-2
possessiveconstructions410, 411, stress224
418 subordination243
postpositionalconstructions368, 416 syntax236-43
preverbs368, 416 verb phrases238-40
prohibitive mood420 vowels 13-18,222-4
reflexive pronouns412 Mator (Motor) 428, 457-79,589
relative pronouns413 consonants9-13, 462-3,466-7
simple sentence240-2,417-20 diachronicevaluation460
stemtypes372 documentation459-60
stress393, 398 lexicon 475-8
suffix order367 morphology18-32,468-75
syntax417-22 phonology462-8
text 424-5 taxonomicrelationships458-9
verbal nouns414-15,420-1 typological variation460-1
verbs361, 382,398-409 vowel harmony465-6
vowels 13-18,390,392-3,395-6, modeof action (Aktionsart)
397 Kamassian591-2
Mari (Cheremis)66, 79,147,167,171, Komi 316
192,203,212,213,219-48,438, Mari 233
442 Nganasan510-11
adjective225, 234, 237,242 Selkup573-4
aspect229 Mordva 66, 68, 69,76, 78, 155, 167,
auxiliary verbs 231, 238,239 168,184-218,245,341,347,428,
clitics 224-5, 236, 243 441,442,450,461
complexsentence242-3 adjective191, 195-6, 198,205,207,
conjunctions242, 243 210
consonants9-13, 220-2 adverbs204-6
coordination243 aspect197, 208-9
copula230, 240 clitics 201
derivation233-6 compoundsentences211
deverbalderivation235 conditionalmood 198,200,201,207
dialects226 consonants9-13,185-7,190
gerunds229-30,232, 239,242 derivation202-4
imperativemood 229 dialects185, 186, 187, 192
indicativemood 229 gerunds204, 205,210-11
inferential mood 233, 239 imperativemood 201-2, 207, 208
lexicon 243-5 indicativemood 198,207
literary EasternMari text 246-7 inferential mood 204,205,207,209,
loan words 227, 244-5 210
modeof action (Aktionsart) 233 interrogativepronouns196
morphology18-32,224-36 lexicon 211-16
negativeverbs231, 232, 239 loan words 185, 186,188, 191,
non-finite verbal forms 231-3, 242-3 211-15
noun phrase236-7 morphology18-32, 191-206
numerals225, 234 morphophonology188-91
participles232, 234 negativeverbs 198,202,207-8
passive231, 234 nominals191-7
phonology220-4 non-finite verbal forms 204
plural 227, 228 noun phrase198, 206
possessiveconstructions241 numerals191196-7,370
 INDEX 611

optativemood 207, 208 auxiliary verbs543

participles198,204,205,207,209 clitics 520, 531,544
personalpronouns196-7, 198 conjunctions544
phonology185-8 consonants9-13, 462-3, 466-7,519,
plural 189, 191, 192, 195,199,201 521,522-6,537
possessiveconstructions194-5,204, definiteness347
209,210 demonstrativepronouns526
possessivepronouns197 denominalderivation541-3
postpositionalconstructions204-6, derivation535,541-3
209 deverbalderivation535, 541-3
prohibitive mood 208 diachronicevaluation460
reflexive pronouns197 dialects519, 521,522, 523, 524, 526,
relative pronouns196 535,539,540
simple sentence206-11 documentation459-60
stemvariants188-9 dual 529, 532, 533,537
stress187, 190--1 Forest383, 457, 517, 518
suffix variants189 gerunds535, 536
syntax206-11 glottal stop 522, 523,525-6, 526, 584
text 216-17 imperativemood530--1,531, 532,
verbs 197-202 533,534,536,544
vowels 13-18, 187-8, 190, 192 inferential mood535-6, 544
Mordvinian seeMordva infinitives 535
Motor seeMator interrogativemood 530--1, 531, 533,
544
narrativemood lexicon 475-8
Mansi 405-9,419 loan words 520, 524
Nenets530--1 morphology18-32,468-75,526-43
necessitativemood morphophonology527-30
Nenets530--1, 533 negativeverbs535, 536,543
negativeverbs27, 29 nominal inflection 537
Estonian113 non-finite verb forms 531, 535-7,
Fennic 105 544
Finnish 172 nounphrase543, 544
Kamassian593, 594 numerals526, 542
Khanty 380--1 objectiveconjugation533, 534
Komi 315, 316, 320--1 optativemood530, 532, 533,534-5
Mansi 420 participles535, 536,544
Mari 231, 232,239 personalpronouns526, 540, 543-4
Mordva 198,200,201,202,204, phonology462-8, 519-26
207-8 plural 522, 528,532, 534,537-9
Nenets536-7, 543 possessiveconstructions538, 539,
Nganasan503-8 541,544
Permian272 postpositionalconstructions526, 529,
Saamic76, 79, 88-9, 91 540--1,544
Samoyedic474-5 pronunciation523
Selkup575 reflexive conjugation533, 534,535
Udmurt 281,291,292-3 Siberian524
Veps 113 simple sentence543
Nenets76, 167,274,323,351,352,389, stress519-20,521, 522
390,398,423,457-79,480,481,482, subjectiveconjugation532, 533,535
513,516-47,549,577,596,598 syntax543-4
adjective526 taxonomicrelationships458-9
adverbs540--1 text 544-5
612 INDEX

Tundra372, 383,457,460,516-47, personalpronouns500

584 phonology462-8,483-7
typological variation460-1 plural 495, 502
verbal art 596 possessiveconstructions496
verbal inflection 530-2 postpositionalconstructions501, 511
versification598 prohibitive mood 503, 508
vowels 13-18,463-4,465-6,467, sentencecombining512
519-20,520-2,524,528,537 simple sentence511
word order543 stress486-7
Nganasan(Tavgi Tawgi, Tawgi- subjectiveconjugation502
Samoyed)150,167,172,428, syntax511-12
457-79,480-515,517,598 taxonomicrelationships458-9
adjective497, 499-500,511 text 5l3-14
aspect503 typological variation460-1
auxiliary verbs508, 511 verbs502-8
bilingualism481 versification598
clitics 497, 500, 508-9 vowels 13-18,465-6,483,485-6,
conjunctions501,512 487,490,492-3
consonants9-13, 462-3, 466-7, non-finite verbalforms
483-5 Estonian137-9
definiteness347 Fennic 113
demonstrativepronouns501 Finnish 174--5
denominalderivation509 Hungarian448-50
derivation508-10 Khanty 375-6,382
determinativepronouns501 Komi 316-17,323
deverbalderivation509, 510 Mari 231-3,242-3
diachronicevaluation460 Mordva204
dialects481-2,484 Nenets535-7, 544
documentation459-60 Nganasan503-8,509, 511, 512
dua1495,502 Saamic75, 89
glottal stop484, 487,491, 494,495 Samoyedic473-4
historical development482 noun phrase
imperativemood502, 503,511 Hungarian450-1
indicativemood503 Khanty 376-7, 378-9,380-1
inferential mood503 Komi 318-19
interrogativemood 503, 511 Mansi 236-7,419
interrogativepronouns501 Mari 236-7
lexicon 475-8,512-13 Mordva 198, 206
loan words 512-13 Nenets543, 544
modeof action (Aktionsart) 510-11 Nganasan511
morphology18-32,468-75,495-510 Saamic84, 85, 86-7, 87-8
morphophonology483-7,487-95 Selkup562, 575
negativepronouns501 numberseedual and plural
negativeverbs503-8 numerals
nominalinflection 495-501 Estonian126, 130-2
non-finite verb forms 503-8, 509, Finnish 157, 168-9
511,512 Hungarian438
noun phrase511 Khanty 370-1
numerals497, 500 Komi 311,314,320
objectiveconjugation502 Mansi 412, 414
optativemood503 Mari 225, 234
participles509, 511 Mordva 191,196-7
passive509-10,511 Nenets526
 INDEX 613

Nganasan497,500 Mordva 207, 208

ObUgrian352 Nenets532, 533, 534-5
Saamic63, 85, 87 Nganasan503
Samoyedic476-7 Selkup566, 568,575
Selkup563, 575 OstyakseeKhanty
Udmurt 282, 286, 295,297 Ostyak-Samoyed seeSelkup

obligative mood participles

Nenets530-1 Estonian135, 137, 140
ObUgrian272,274,327-57,382,389, Fennic 111, 113
550 Finnish 154, 161,174-5
conditionalmood 347 Hungarian448-50,451
consonants9-13, 332-4,335, 337-8 Kamassian590, 591, 592,594
copula353, 355 Khanty 375-6
definiteness347-9 Mansi 414-15,418,420-1
deixis 354 Mari 232, 234
demonstrativepronouns348 MordvaI98,204,205,207,209
derivationalmorphology349-50 Nenets535, 536, 544
dialects332, 338, 339, 340,341, 341, Nganasan509,511
343,345,348,353 ObUgrian350
dual 340-1,348 Saamic75, 76, 78-9, 84, 88-9
gerunds354 Samoyedic473
imperativemood 339, 347 Selkup569-70,574
indicative mood 347 Udmurt 291,295,299
inferential mood 349-50 passive
internal divisions of Ostyak328-9 Fennic 112
internaldivisions ofVogul327-8 Finnish 174, 175
lexicon 351-2 Kamassian591
loan words 351-2 Khanty 371, 372, 376-81
morphology 18-32,340-9,349-50 Komi 320
morphophonology338-40 Mansi 398,403-5,406
numerals352 Mari 231, 234
objectiveconjugation347-8 Nganasan509-10,511
ObUgriansharedfeatures352-3 ObUgrian339,349,351,352
Ostyak336-8 Selkup570
participles350 pasttense27-8
passive339, 349, 351, 352 Finnish 172
personalpronouns341-3,345-6 Hungarian447
plural 340-1,348, 354 Kamassian590, 592,593
postpositionalconstructions354 Mari 229, 238
Proto-ObUgrian329-40 Mordva 198, 199-200
proto-Ugric 353-5 Nenets531
simple sentence350-1 ObUgrian339, 346
stress332 Saamic77, 78,88
syntax350-1 Selkup567
Ugric sharedfeatures353-5 Udmurt289
verb inflection 346-9 Permiak250-75,312,323
Vogu1334-5 Permian249-75,327, 350, 358, 398,
vowels 13-18,329-32,334-5,336-7 428,446,448,449
seealso Khanty and Mansi conjugation270-2
Old Zyrian 254-5 consonants9-13, 255-61
optativemood contactswith otherlanguages273-5
Khanty 371 declension267-70
614 INDEX

dialects 254-5 Nganasa n 496

gerunds 273 Permian 269-70
imperative mood 271 Saamic 63,65, 69,70
indicative mood 271 Samoyedi c 470,472
inferential mood 273 Selkup 561-2,571, 575
loan words 273-5 Udmurt 283-5,295, 297
morphology 18-32,267-7 2 postpositional construction
s
negative verbs 272 Estonian 135,142-3
numerals 370 Finnish 163-7
Old Zyrian 254-5 Hungarian 334-6,442,445- 6
persona l pronouns 227 Kamassia n 590
phonology 255-67 Khanty 367-8, 376
plural 267-8, 270 Komi 312, 313
possessiv e constructions 269-70 Mansi 416
Proto-Permia n 254, 2 7 3^ Mari 236-7, 243,241
simple sentenc e 272-3 Mordva 204-6,209
stress 266-7 Nenets 526,529, 540-1, 544
syntax 272-3 Nganasa n 501,511
Votyak 274-5 ObUgrian 354
vowels 13-18,261-6 Saamic 86, 87
Yaz'va Dialect 254-5 Selkup 561, 571
Zyrian 274 Udmurt 293
seealso Komi and Udmurt potential mood
persona l pronouns seepronouns Fennic 111
plural Finnish 171
Estonian 125, 127,129,133,134,13 5 Saamic 71, 77, 88
Fennic 110, 111, 154 prohibitive mood
Finnish 159,160, 163, 167,169,17 0 Finnish 174
Hungarian 438-40,450 Mansi 420
Kamassia n 586-8, 589, 593 Mordva 208
Khanty 361, 365, 366,367 Nganasa n 503,508
Komi 311, 312, 313, 316, 320 pronouns , anaphoric
Mansi 398, 399,400,401,402,403 , Finnish 170
404,410 Khanty 369
Mari 227, 228 Selkup 565
Mordva 189,191, 192,195,199,20 1 pronouns , demonstrativ
e 25-6
Nenets 528, 532, 534, 537-9 Estonian 133^
Nganasa n 495, 502 Finnish 170,171
ObUgrian 340-1, 348, 354 Hungarian 435,444
Permian 267-8, 270 Kamassia n 588,589
Saamic 63, 67, 69,70, 71,75, 76,77, Khanty 369-71
85 Komi 313-14
Samoyedic 468,469-70,47 2 Mansi 403,413
Selkup 557, 558-9, 560 Nganasa n 501
Udmurt 280, 282,283, 286, 287, 288 ObUgrian 348
possessiv e constructions 32-3 Saamic 63, 87
Estonian 133 Selkup 565
Hungarian 438,439,45 1 Udmurt 288
Komi 313 pronouns , indefinite
Mansi 410,411,41 8 Estonian 135
Mari 241 Finnish 171
Mordva 194-5, 197,204, 209,210 Hungarian 445
Nenets 539 Khanty 370
 INDEX 615

Mansi 413 Selkup563, 564

Saamic87 Udmurt 286-7,288
Selkup565 pronouns,relative
Udmurt288 Estonian134-5
pronouns,interrogative Finnish 170
Estonian127, 134-5 Hungarian445
Finnish 170 Komi 314
Hungarian444, 445 Mansi 413
Kamassian590 Mordva 196
Khanty 370 Saamic87, 88
Komi 314 Se1kup576
Mansi 413
Mordva 196 quotativemood 28
Nganasan501 Estonian136, 137, 141
Saamic87, 88 Fennic 113
Selkup565, 576
Udmurt288 reportedspeech
pronouns,negative Nenets535
Estonian135 Nganasan488, 489, 503
Finnish 171 Saamic90
Hungarian445 Udmurt300
Khanty 370 reputativemood
Komi 314 Nenets530-1,533
Mansi 413
Nganasan501 Saamic43-95,101, 149, 150, 155, 167,
Selkup565 171,179,192,203,212,227,245,
pronouns,personal24-5 358,439,448,458,461,482
Estonian132-3 adjective63, 81, 84, 87
Finnish 157, 169 aspect80
Hungarian438,442-3,443-4 clitics 62
Kamassian588-9 conditionalmood 71, 77, 88
Khanty 368-9 conjunctions89
Komi 311, 313 consonants9, 13,56-62
Mansi 398,412-13 coordinationof sentences89-90
Mordva 196-7, 198 copula76, 77-8, 85, 86, 88
Nenets526, 540,543-4 deixis 84
Nganasan500 demonstrativepronouns63, 87
ObUgrian341-3,345--6 denominalderivation81, 83
Permian272 deverbalderivation79-80, 80-2
Saamic62, 71, 91 dialects56, 79, 462,561,576
Selkup563, 564 dual 62, 69, 70, 75,76,77,90
Udmurt 286,287,296 gerunds75, 78,79
pronouns,reciprocal imperativemood 54, 71, 76-7, 78, 88
Komi 313 indefinite pronouns87
Udmurt 288 indicative mood71, 76,88
pronouns,reflexive 25 inferential mood75, 78, 79,84, 89
Estonian133 infinitival constructions89
Finnish 170 inflection 63-79
Kamassian590 interrogativepronouns87, 88
Komi 313 lexicon 90-2
Mansi 412 loan words 90-2
Mordva 197 morphology18-32,62-84,85-7
Saamic71, 89 negativeverbs76, 79, 88-9, 91
616 INDEX

non-finite verbal forms 75,89 adverb562, 574

noun phrase84, 85, 86-7, 87-8 anaphoricpronouns563, 565
numerals63, 85, 87 aspect566
participles75, 76,78-9, 84, 88-9 auxiliary verbs575
personalpronouns62, 71,91 bilingualism548-9
phonology52-62 causativeverbs573
plural 63, 67, 69, 70,71, 75, 76, 77, Central553
85 conditionalmood566, 568
possessiveconstructions63, 65, 69, conjugation567-9
70 conjunctions576
postpositionalconstructions86, 87 consonants9-13, 462-3,466-7,
potentialmood71, 77, 88 551-2,554,555,556,557
reflexive pronouns71, 89 deadjectivalderivatives572-3
relative pronouns87, 88 definiteness562
simple sentence84 deixis 565
stress52-3, 55, 59,62, 70, 80 demonstrative pronouns 563, 565
subordinationof sentences89-90 denominalderivation571-2
syntax84-90 derivation571-4
text 92 determinativepronouns565
varieties43-52 deverbalderivation574
vowels 13-18,53-6, 372 diachronicevaluation460
word formation 79-84 dialects549-50,551-3,554, 560-1,
Samoyedic167, 172,358,371,372,553 570,575,576
auxiliary verbs475 documentation459-60
consonants9-13, 462-3,466-7,468 dual 558-9
diachronicevaluation460 genericpronouns565
dual 469-70 glottal stop 584
gerunds473 historical development550-1
imperativemood 474,475 imperativemood562, 566, 568,569,
lexicon 475-8 575,576
loan words 477 indefinite pronouns565
morphology18-32,468-75 indicativemood566, 567
negativeverbs474-5 inferential mood566,567,576
non-finite verbal forms 473-4 interrogativemood576
numerals476-7 interrogativepronouns565, 576
objectiveconjugation470 intransitiveverbs573
participles473 lexicon 475-8,577-8
phonology462-8 loan words 576, 577
plural 468, 469-70,472 measurement 575
possessiveconstructions470, 472 modeof action(Aktionsart) 573-4
Proto-Samoyedic461 morphology18-32,468-75,557-74
Proto-Uralic461 morphophonology554-7
subjectiveconjugation470 Narym552
typological variation460-1 negativeconstructions575
vowels 13-18,463-4,465-6,467, negativepronouns565
553 negativeverbs575
seealso Enets;Kamassian;Mator; nounphrase562, 575
Nenets; Nganasan; Selkup numerals563, 565
Selkup(Ostyak-Samoyed)168,341, objectiveconjugation568, 576
351,383,441,457-79,517,548-79, optativemood566, 568, 569, 575
584 participles565, 569-70,574, 575
adjective560, 562-3,565, 569, 574, passive570
575 person567
 INDEX 617

persona l pronouns 563, 564 Nganasa n 487

phonology 462-8,551— 4 ObUgrian 332
plural 557,558-9, 560, 562 Permian 266-7
possessiv e construction s 561-2,563, Saamic 52-3,55, 59, 62,70, 80
571,575 Selkup 553-4
postpositiona l constructions 561, 571 Udmurt 280-1
preverbs 571 Zyrian 267
reflexive pronouns 563, 564 subjunctive mood
reflexive verbs 573 Hungarian 446-7,449
relative pronouns 576 Selkup 566,568
sentenc e combining 576-7
simple sentenc e 576 Taigi seeMator
stress 553^4 Tavgi seeNganasa n
subjective conjugation 568, 576 Tawgi-Samoye d seeNganasan
syntax 575-7 tense 27-30
taxonomic relationships 458-9 Estonian 138
tense 566-7 Fennic 111-13
text 578 Finnish 171^1
transitive verbs 573 Hungarian 446-7,449
typological variation 460-1 Kamassia n 594
Upper Taz 552 Khanty 372
verbs 562, 575-6 Mansi 398
vowels 13-18,465-6 , 551, 552, 553, Mari 238-9
554,557 Mordva 198
word order 576 Nenets 531
sentenc e combining Nganasa n 502,503
Nganasa n 511-12 ObUgrian 346-7
Selkup 576-7 Permian 271
simple sentenc e 33-4 Saamic 62,75-6, 85
Estonian 143-5 Samoyedi c 473
Fennic 113 Selkup 566, 567
Finnish 175-7 Udmurt 289-91
Khanty 380 tone 9
Komi 321-2, 322-3 Estonian 119-21,120,13 3
Mansi 240-2,417-2 0 Fennic 104-5,106,11 0
Mari 240-2 Finnish 153,159
Mordva 206-11 Hungarian 429,430,431,433,43 7
Nganasa n 511 Khanty 362-3
ObUgrian 350-1 Komi 308, 309, 310-11
Permian 272-3 Mansi 392, 393, 398
Saamic 84 Mari 224
Selkup 576 Mordva 187,190-1
Udmurt 299 Nenets 519-20,521 , 522,528
st0d Nganasa n 485,486,487,49 4
Livonian 106 ObUgrian 329, 330, 332, 335, 336-7
stress 9 Permian 266-7
Estonian 119, 120,133 Saamic 52-3,54,55, 58-9,62,70, 80
Fennic 104-5,106, 110 Samoyedi c 467,468
Komi 308, 310-11 Selkup 553-4,557
Mansi 393, 398 Udmurt 280-1
Mari 224 Zyrian 267
Mordva 187,190-1
Nenets 519-20,521,52 2 Udmurt (Votyak) 276-304, 305, 341,
618 INDEX

351,423,437,446,549 syntax 113

adjective281, 282, 285-6, 294, 295, vocabulary113-14
297 Vogul seeMansi
assimilation281 voice, active
conditionalmood: Udmurt 289-91 Finnish 175
consonants9-13, 279-80,281, 301 Khanty 376-81
deixis 295 ObUgrian349
demonstrativepronouns288 Saamic71
derivationalmorphology294 voice, impersonal
dialects279, 280,281, 289,301 Estonian136, 138
gerunds293, 298,299 Fennic 112
imperativemood 280-1, 289-91, voice, passive
292,300 Fennic 112
indefinite pronouns288 Finnish 174, 175
indicativemood 289-91,292 Kamassian591
indirect discourse300 Khanty 371, 372, 376-81
inferential mood 291, 299 Komi 320
inflectional morphology282-93 Mansi 398,403-5,406
interrogativepronouns288 Mari 231, 234
lexicon 301-2 Nganasan509-10,511
loan words 295, 301-2 ObUgrian339, 349, 351, 352
morphology18-32,282-93,294, Saamic71, 78
295 Selkup570
morphophonemics281 voice, personal
negativeverbs281, 291,292-3 Estonian132-3, 136, 138
numerals282, 286, 295,297 Fennic 111-12
participles291, 295, 299 Kamassian588-9
personalpronouns286, 287,296 Nenets540
phonology279-81 ObUgrian345-6
plural 280, 282,283,286, 287,288 Saamic75-6
possessiveconstructions283-5,295, Udmurt286
297 voice, reflexive
postpositionalconstructions293 Estonian133
reciprocalpronouns288 Fennic 107, 108, 110
reflexive pronouns286-7, 288 Finnish 170
reportedspeech300 Kamassian590,591
simple sentence299 Komi 313
stress280-1 Mansi 412
syntax295-300 Mordva 197
text 302-3 Nenets532, 533, 534,535
verb 289-93 Nganasan502
voice 281 Saamic71, 78, 89
vowels 13-18,280, 281 Samoyedic472
Ugric 389 Se1kup563, 564,573
Udmurt 286-7,288
Veps 96-114,274 Votic 96-114,115
consonants9-13,109-10 consonants9-13,109-10
dialects101-4 dialects101-4, 109
loan words 114 loan words 114
morphology 18-32,110-13 morphology18-32, 110-13
negativeverbs 113 nominal inflection 110-11
nominal inflection 110-11 quantity 105-10
stress104-5, 106, 110 stress104-5, 106, 110
 INDEX 619

syntax 113 Mari223-4

verb inflection 111-13 Mordva 190, 192
vocabulary113 Nenets465-6,521, 522
vowel harmony106-7 Nganasan465-6,487,490,492-3
Votyak seeUdmurt ObUgrian331-2,337
vowel clusters Permian266
Selkup554 Selkup553
vowel harmony17-18 vowel lengthseelength
Enets465-6
Estonian123 word orderseesimple sentence
Fennic 106-7
Finnish 150-1, 163 Yurak seeNenets
Kamassian582-3,597 Yurats seeNenets
Khanty 362, 365
Komi 311 Zyrian seeKomi
Mansi 390, 397 ZyryeneseeKomi