1. Phoneme inventory

2. Epenthesis

2.1 Surface phonotactics

2.2 Chukchee syllable structure

3. Consonant alternations

3.1 Consonant harmony

3.2 Contact assimilations

3.2.1 Regressive assimilation of nasality

3.2.2 Regressive labial assimilation

3.2.3 Regressive velar to uvular assimilation

3.2.4 Alternations with alveolars - regressive assimilations

3.2.5 Regressive place assimilations affecting velar nasals

3.3 Contact dissimilations

3.3.1 Dissimilations of repeated alveolars

3.3.2 Dissimilations of velars and palatals

3.3.3 Denasalization

3.3.4 Delabialization

3.4 Metathesis

3.5 Consonant truncation

3.5.1 Causative r-

4. Vowel alternations

4.1 Vowel harmony

4.2 Vowel reduction

4.3 Final vowel truncation

4.4 Vowel cluster simplification

5. Stress

5.1 Basic pattern

5.2 Stress shift

6. Some postlexical alternations


1. Phoneme inventory









voiceless stops




























Notes: /c/ is a palato-alveolar affricate, /l/ is voiceless, /g/ is voiced velar fricative, /r/ is retroflex glide (like Standard British English). /w/ is a voiced bilabial fricative (and is often transcribed 'v').

The status of the glottal stop has occasioned some controversy. Distributionally it is aberrant as a consonant which has led some to regard it as a prosodic colouring of the vowel.






y (schwa)






Notes: e1, e2 represent a more-or-less high tense front mid vowel and a more-or-less low front vowel. The difference is important for vowel harmony. Schwa will be written as 'y'.

top of page

2. Epenthesis

2.1 Surface phonotactics

(i) NO CC clusters initially, finally
(ii) NO CCC clusters medially.

[Exceptions: #C?-, #C [son], -CC? -] There are a number of words which begin with consonant clusters, at least in the official orthography. This is said to be the result of a compromise between different dialects, one of which permits initial clusters.

Clusters resulting from word formation broken up by schwa 'y'.

Kenstowicz, Krause assume (at least) TWO separate rules needed, Initial and Final Epenthesis. BUT: no rules needed given appropriate theory of SYLLABIFICATION.

2.2 Chukchee syllable structure



























C (?)




(Occasionally we find C + sonorant clusters in onsets, e.g. plekyt 'boots', mranly yn 'mosquito net'.)

Syllabification: Syllabify from left to right, respecting the basic template and inserting schwa where necessary to break up illegal clusters.

Examples: Abs. case forms in -lgyn, -lyyn: UR has -l- stem-forming suffix plus - = Abs. case ending. [NB phonological alternation Þ g /l _____].


welo-l-n 'ear'


w e l o l n (we) (lol) (n)








mran-l-n 'mosqito net'


mr a n l n



(mran) (l) (n)


mran. l(y). (y)n







Problem: syllabification respects morpheme boundaries, thus left-to-right syllabification can be overridden as in the following case (where brackets indicate morpheme boundaries):



[...V C] [C C V...]


...V C y C C V...





*...V C C y C V...

This gives the impression that the Maximal Onset Principle (which puts consonants preferentially into Onsets rather than Codas) has been violated.

Minimal pairs:


meml-qaca-n = meml-y-qacan

water-PLACE-ABS 'place near the water' vs.


wejem-lq-n = wejem-y-lq-y-n (*wejemlyqyn)

river-FULL-ABS 'teaming with rivers'

Therefore, revise Syllabification: Syllabification: "Syllabify maximally from left-to-right, respecting Maximum Onset Principle except where this entails breaking up a morpheme." (See Spencer 1994, Kenstowicz 1994 for discussion.)


myngylgyn 'hand' mng-l-n


(mn) (gl) (gn) = m y n g y l g y n


Some words have no underlying vowels (in fact, many roots lack underlying vowels).

Evidence that /y/ is sometimes an underlying phoneme (Kenstowicz, 1979, Krause, 1979):

Plural allomorphy: -ti/te after ...V C[+coronal]

-t elsewhere [pace Kenstowicz]

tintin Þ tintinti 'ice'

ener Þ enerti 'star'

ococ Þ ococte 'leader'

q?awal Þ q?awalti 'corner'



titi Þ titit 'needle'

jara-y Þ jarat house'

kuke Þ kuket 'kettle'


r?ew Þ r?ewyt 'whale'

?aacek Þ?aacekyt 'youth'[-cor]

rileq Þ rileqyt 'spine'


kej-yn Þ kejyt 'brown bear'

pojgyc?-yn Þ pojgyc?yt 'stalk'

renm-yn Þ renmyt 'wall'

What happens with ...VCC[+cor]?

qepyl (>qepl) Þ qeplyt 'ball' (*qepylti)

rytyn (>rtn) Þ rynnyt 'tooth' (*rytynti)

lewyt (lewt) Þ lewtyt 'head' (*lewetti)

Now contrast:

memyl Þ memylte 'seal'

mimyl Þ mimlyt 'water'

This can be understood if URs are /memyl/ and /miml/ resp.

top of page

3. Consonant alternations

[Taken principally from Irina Muraveva's class notes ('Grammatika neindoevropejskogo jazyka - chukotskij jazyk', 1995-96, Russian State Humanities University, Moscow), supplemented by Skorik vol. I, Krause, Kenstowicz.] Skorik differentiates between automatic (phrase phonology, 'phonetic') alternations and lexical, morphophonemic alternations (which he calls 'historical').

Many of these alternations (including some postlexical ones) are reflected in the orthography (which can make it difficult to perform morphemic analysis on orthographic representations), though many are not. Where an alternation is not reflected in the orthography I indicate this.

3.1 Consonant harmony

[This is not mentioned in Skorik.] Some morphemes are 'palatalizing', and induce the alternation l, t Þ c

vilu- 'ear' + -tkyn 'end' Þ vecotkyn 'tip of the ear'
tytl 'door', cycc-?yt?yjoca 'in front of the door'

/l/ and /c/ alternate under circumstances which can't be associated easily with palatalization and there are a fair number of lexical doublets amongst roots and affixes which differ only in the presence of /l/ or /c/, e.g. (from Bogoraz's dictionary) quliletyk 'to shout' qucicetyk 'to scream, make a row'.

3.2 Contact assimilations

[Skorik I:43f] Tend to be found morpheme internally and at affixation boundaries, but are optional in compounding, e.g. incorporation (Skorik I:48; see also Muravyova, 1988).

3.2.1 Regressive assimilation of nasality

p + m/n/ Þ m

t + m/n/ Þ n

k,g,q + Þ

Assimilation to following nasal: /p, t/ become nasals before any nasal (/m, n, /):

jyp- 'put on (clothing)'



jymnen 'he put it on'

pyne- 'sharpen'



gemnelin 'sharpened'

tyni- 'sew'



gennilin 'sewn'

rytyn 'tooth'



rynnyt 'teeth'

migciret- 'work'



nymigcirenmuri 'we work'

tyiwy- 'send'



geniwylin 'sent'


Optional assimilation in compounds:

janot 'front' inqej 'boy'


janotinqej/ janoninqej 'front boy'


/k/ assimilates only before //:





tepleyk 'to make footware, cobble'


g/ undergoes the same assimilation but this is optional and rare according to Skorik (I:45). Nonetheless, the following example is reflected in the official spelling:

lig- 'egg' yto- 'come out'



leyto- 'hatch; lay egg'

[On rn Þ nn see below]

3.2.2 Regressive labial assimilation

Velars before labials become labials (not reflected in orthography):

k/g + p/m Þ w

myg 'many' + penwel 'stag'


myw-penwel 'many stags'

ytlyg- 'father' + pojgyn 'spear'


ytlyw-pojgyn 'father's spear'

ceg 'egg' + mak 'shell'


cew-mak 'eggshell'

gw is a special case since this becomes ww, which then becomes kw (see below)

3.2.3 Regressive velar to uvular assimilation

k q Þ q q (not orthographic)

gynnik 'animal' -qej 'diminutive, young of' Þ gynniqqej

3.2.4 Alternations with alveolars - regressive assimilations

(Generally not indicated in the orthography)

c, r Þ t /_____t, l, r

mac 'fairly' tot?ety 'stupid' Þ mat-tot?ety

g-oc-len 'gnawed at' Þ gotlen

mac + rayky 'somewhat further Þ mat-rayky

pykir-tyk 'you came' Þ pykit-tyk

qapar 'wolverine' + -cyyn 'suffix' Þ qapat-cyyn

aar 'star' + -lyyn 'suffix' Þ aat-lyyn

l Þ t /_____r

gyrgol 'upper' ramkyn 'people' Þ gyrgot-ramkyn

r Þ n /___n

tur 'new' nelgyn 'hide' Þ tun-nelgyn

Muraveva treats some of these as complete assimilations followed by dissimilation (see below), e.g.:

cl Þ ll Þ tl; cr Þ rr Þ tr; rc Þ cc Þ tc; rl Þ ll Þ tl

3.2.5 Regressive place assimilations affecting velar nasals

Þ m /_____p, w (not usually in orthography)

p?op?o Þ p?omp?o 'mushroom'
ta 'good' + wagyrgyn 'life' Þ tam-wagyrgyn

Before alveolars Þ n

yto- 'go out' Þ ga-nto-len 'went out'
ta 'good' cotcot 'pillow' Þ tan-cotcot
lili 'heart' Þ linli

tele 'ancient' remken 'people' Þ telen-remken
tele, jep 'still' Þ telen-jep 'long ago'

3.3 Contact dissimilations

3.3.1 Dissimilations of repeated alveolars

cc Þ tc
ll Þ tl
rr Þ tr

mec + cytc' 'nearÞ met-cytcy 'fairly near'
n?el 'become' Þ ge-n?et-lin 'became'
tur 'new', rycq- 'sword' Þ ga-tot-rycq-yma 'with a new sword' (NB vowel harmony)

Cf. also rj Þ tj:

kur- 'buy'+ jo Þ kot-jo 'bought' (NB vowel harmony)

3.3.2 Dissimilations of velars and palatals

k, j Þ g /_____ alveolars

kty- 'hard' Þ ny-gty-qin

myk- 'many', liglig 'egg' Þ myg-liglig

nennet 'otters' Þ myg-nennet

remkyn 'people' Þ myg-ramkyn (NB vowel harmony)

inqej 'boy' + -ti 'plural' Þ inqeg-ti

w?ej 'meadow' -curm- 'edge' Þ w?eg-curmyn

w?aj 'grass' -lyyn 'suffix' Þ w?ag-lyyn

w?aj + ra- 'house' Þ w?ag-ran 'grass house'

gg, gw Þ kg, kw:

menig 'cloth' -gypy 'ABL' Þ manik-gypy 'from the cloth'

ytlyg- 'father' gyjiwqew 'mark' Þ ytlyk-gyjiwqew

atlag- 'sweet' wetgaw 'speech' Þ atlak-wetgaw

3.3.3 Denasalization

Þ g /_____ m, n (not orthographic)

tara- 'build a house' Þ nytarag-more 'we build a house'

tejyli- 'stroke' Þ tejylig-nin 'stroked it'

3.3.4 Delabialization

ww Þ kw

This is more common than might be thought if we regard the gw Þ kw and especially the wg Þ kw alternations as passing through a stage gw Þ ww. The sequence wg occurs very frequently because there are several sets of g-initial agreement suffixes and -ew is a common verb stem final suffix.

tyw- 'communicate' -gyrgyn 'nom.' Þ tyk-wyrgyn

ejmew- 'approach' -g?i '3sg.' Þ ejmek-w?i 's/he approached'

3.4 Metathesis

q Þ ?/_____C, then V?C Þ ?VC

jaqjaq 'seagull' (Þ ja?jaq) Þ j?ajaq

3.5 Consonant truncation

Root initial consonants in clusters are often dropped word initially:

/tkur/ 'buy' Þ kuryk 'to buy', Þ ge-tkur-lin 'bought'

A list of such roots is found at the back of Moll and Inenliqej's dictionary.

3.5.1 Causative r-

Alternatives with /n/ when non-initial:

wiri- 'go down' ry-wiri- 'take down' ge-n-wiri-w-lin 'took down'

top of page

4. Vowel alternations

4.1 Vowel harmony

Vowels divided into recessive and dominant sets:









Either: schwa

Confusingly, /e/ belongs to both sets. There is no clear concensus in the literature as to exactly how the vowels are pronounced, though /i, u/ are lax rather than tense. The dominant and recessive /e/ vowels appear to have the same pronunciation. Where necessary I represent recessive /e/ as e1 and dominant /e/ as e2.

If a dominant vowel occurs anywhere at all in the word, all the recessive vowels are replaced by their dominant counterparts, as indicated in the table above. This applies even across very long and complex incorporation domains (in contrast, say, to Turkic or Finno-Ugric vowel harmony).

milute 'hare' ga-...-ma 'Comitative circumfix' Þ ga-melota-ma

-qin 'adjective suffix' e.g. ny-tur-qin 'new', om- 'warm' Þ n-om-qen

tur 'new' nelg- 'hide' Þ ga-tor-nalg-y-ma 'with a new hide' (incorporation structure)

Schwa never itself alternates but it may trigger harmony. All intances of epenthetic schwa are recessive. 'Recessive' lexical schwa: ytlyg- 'father' (Ergative case ytlyg-e, *ytlyg-a). 'Dominant' lexical schwa: -gyrgyn 'nominalizer', e.g. tylek 'to move' tylagyrgyn 'step, path'.

Many roots lack a vowel altogether and some of these are dominant, e.g. -tm- 'kill'

tm + nin '3sgSubj/3sgObj' Þ tym-nen

Cf. also the adverbial comparative suffix -:

cymce 'near' cymca- 'nearer'

4.2 Vowel reduction

Word final /e, a/ reduced to schwa:

epe-: epy 'grandfather' (cf. epe-t 'grandfathers'

wala-: waly 'knife' (cf. wala-t 'knives')

4.3 Final vowel truncation

Many roots with apparently final consonants end in vowels which are truncated word finally:

?aacek(e) 'youth': ?aacek Abs. sg., but ?aaceke-t Abs. pl., ?aaceke-gty All.

ekk(e) 'son': ekyk Abs. sg., but ekke-t Abs. pl., akka-gty All. [NB geminate cluster broken by epenthesis in ekyk.]

Adjective suffix -qin(e): nilgy-qin 'white-ADJ', nilgy-qine-t 'white-ADJ-PL'

4.4 Vowel cluster simplification

V1 + V2 Þ V2 where V2 is 'stronger' than V1 according to hierarchy:

u/o > i/e2 > e1/a

ge-it-lin 'was' Þ gitlin

ge-umeket-linet 'met' Þ gumeketlinet

ga-okwanaw-ma 'with tar' Þ gokwanawma

r?yra-emlypgat 'murky pool' Þ r?yremlypgat

No truncation after consonant cluster, even if the following morpheme begins with a long vowel:

gemge-ekyk 'every son'; gamga-otkyn 'every corner'; tyme-ottoot 'the usual tree'; qente-uwil 'dull echo'; mejy-ilir 'large island; emytlo-ajmyjocgyn 'leaking bucket'; tany-er?yn 'foreign clothing'

gemge-uul 'every chisel'; gamga-eekelgyn 'every edible root'; tyme-aak 'the usual lamp'; majy-aatger 'large brook'

top of page

5. Stress

5.1 Basic pattern

(I:67-71) Chukchee has a stress system which appears to be contrastive. There is very little information about stress. Stress is not indicated in the orthography (probably because Russian stress isn't indicated orthographically). None of the dictionaries I've seen except that of Bogoraz gives stress information and Skorik doesn't indicate stress in any of his examples (except for transcribed examples in the section on stress). However, Bogoraz's transcriptions all include stress information and are a valuable source of data for anyone wishing to attack this problem.

Skorik claims that stress always falls on the final syllable of the root or stem, and never on inflectional suffixes or on reduplicates. Thus we have:

pójg-a 'spear-INSTR'

base /pojg/

íw-y-k 'to speak'

base /iw/

jará-y 'house-ABS'

base /jaja/

reqoká-lgyn 'arctic fox-ABS'

base /reqoka/

ekwét-yk 'to set out'

base /ekwet/

migcirét-yk 'to work'

base /migciret/

túmg-y-tum 'friend.ABS'

base /tumg/

káw-kaw 'biscuit.ABS'

base /kaw/

nym-nym 'village.ABS'

base /nym/

nuté-nut 'earth.ABS'

base /nute/

However, a number of reduplicated forms with disyllabic bases have initial stress:

wánewan 'not', wéniwen 'bell', céricer 'dirt', kélikel 'book'

Stress doesn't, apparently, fall on epenthetic schwas inside bases (Skorik seems unaware that not all schwas are lexical and includes schwa in representations of roots) (I:70) (ý is stressed schwa):

mycýkw-yn 'shirt-ABS'

base /mycykw/ (Skorik)


actually /myckw/ or even /mckw/

rykgýt-yk 'to get stuck'

base /rykgyt/ (Skorik)


actually /rykgt / or even /rkgt/

What would be interesting is information about suffixed forms of words such as mémyl 'seal' in which stress falls on the first syllable in the unaffixed form, but in which the schwa is lexical.

5.2 Stress shift

If there are no vowels in the suffix(es) attached to a base, then stress is retracted to the penultimate syllable of the base (or to the first syllable, i.e. as far back as possible,where the base has fewer than three syllables):

tití-y 'needle-ABS'but títi-t 'needle-ABS.PL'  


melotá-lgyn 'hare-ABS'

base /milute/

but mélute-t 'hare-ABS.PL'

[Disturbingly, Bogoraz's dictionary gives the stress melótalgyn for 'hare']

It is unclear from Skorik's account whether this is retraction to the antepenult or retraction to the root initial syllable. (On typological grounds one would expect the latter.)

Similar retraction is observed with unsuffixed bases in which stress would fall word finally:

ricít-te 'belt-ABS.PL'

but rícit 'belt.ABS.SG'

base /ricit/

warát-te 'people-ABS.PL'

but wárat 'people.ABS.SG'

base /warat/

jatjól-te 'fox-ABS.PL'

but játjol 'fox.ABS.SG.'

base /jatjol/

Vowel reduction and stress: Retraction of word final stress seems to have resulted in vowel reduction for a number of roots:


walá-jpy 'knife-ABL'omqá-jpy 'polar bear-ABL'

úmqy 'polar bear.ABS'

Prefixation generally doesn't affect stress placement:

kój-yn 'cup-ABS'

ga-kój-y-ma 'COM-cup-COM.II', 'with a cup'

qulil?ét-yk 'to shout'

ge-qulil?ét-lin 'PAST.II-shout-PAST.II' 'shouted'

However, where the root lacks a vowel then stress may retract to a prefix (some of these examples are from Bogoraz's dictionary):

tým-yk 'to kill', base /tm ~ nm/

tým-gyrgyn 'killing'

tým-jo 'killed (passive participle)'

tym-ý-rkynen 'kills him'

q-ená-nm-y-ge 'kill me!'

gá-nm-y-len 'killed (PAST.II)'


Stress in incorporation: incorporated roots tend to get secondary stress (primary stress falling on the rightmost lexical stem):

ny-qorá-gynrét-qen '3pl-reindeer-guard-3sg' 'they guard reindeer'

ga-tór-máj-y-walá-ta 'COM-new-large-knife-COM.I' 'with a large new knife'


Skorik observes that some suffixes may bear secondary stress:

ytó-sqycát-g?e 'go.out-SUDDEN-3sg' 'he dashed out' where primary stress falls on ytó-.

mémyl-y-t?ól-a 'seal-meat-INSTR' 'with seal meat' primary stress mémyl-

top of page

6. Some postlexical alternations

1. k Þ g /___C (except /w/) (I:53)

kyjew- 'wake up' ge-kjew-lin Þ gegjeklin 'woke up

2. Pronunciation of schwa (not represented orthographically)

Schwa Þ unrounded [u] adjacent to labial consonant

Schwa Þ i adjacent to /j/

[Krause also treats this as a highly opaque morphophonemic alternation.]

This alternation gives rise to surface violations of vowel harmony.

3. Intervocalic glide deletion. Glides or glide-like elements may be deleted between vowels. Usually optional postlexical process:

jilyjil 'tongue, language'


ew-ekyk 'daughter'


gagalalen 'went past'


qoray 'reindeer'


[Note in last case, qaa- is usual Abs. pl. stem form and also incorporative form.]

4. Consonant allophony (Asinovskij 1991).

The basic pronunciation of /c/ is as a prepalatal sibilant (like Polish /s'/), despite the orthography and the usual transcription system (based on the orthography). (A better transcription for /c/ would therefore be /s/.)

Both /l/ and /c/ are (slightly) affricated intervocalically.

Voiceless stops are aspirated in word final position.

/w r j g/ are devoiced after voiceless stops.

There is some palatalization of oral and nasal stops before /i/

/w/ is vocalized to very short [u] in word final position and before /i/.


Chukchee homepage

chapter 1

chapter 3

chapter 4

chapter 5

chapter 6

chapter 7

chapter 8

chapter 9

chapter 10


Andrew Spencer's homepage

top of page

This page created on 11 July 1999.

Last modified 16 July 1999.