Chapter 14Valency-changing morphology

Causatives, applicatives, the antipassive, reflexive and reciprocal prefixes, and noun incorporation.

Since every Sakhalin Ainu verb stem is lexically intransitive or transitive (Chapter 12), all adjustment of valency is done by derivational morphology, and the language maintains a full double set of it. Valency is increased by the causative suffixes -te/-re, -ke, -ka and -yara and by the applicative prefixes e-, ko-, o-; it is decreased by the antipassive i-, the reflexives yay- and si-, the reciprocal u-, and noun incorporation. Chiri catalogued both sets for Sakhalin in one programmatic paragraph (Chiri 1942: §33), and Dal Corso confirms the same inventory in Piłsudski's east-coast corpus (Dal Corso 2024: 70). The richness and mutual combinability of these ‘voices’ is the celebrated typological trademark of Ainu ((Shibatani 1990); (Bugaeva 2022)); this chapter documents what of it is attested in Sakhalin sources, ending with the combinations and a productivity assessment. Two further markers touch valency only obliquely: the resultative ci-, which detransitivises by deleting the agent (ci-ranke ‘be lowered’ (Chiri 1942: §49), (Dal Corso 2024: 67)), and the impersonal passive built on the indefinite person, treated with person marking in Chapter 13.

Each operation below also has a fixed effect on Chiri's classification of predicates into complete (object-less) and incomplete (object-requiring) verbs (Chapter 6). The increasers — applicatives and causatives — turn a complete verb into an incomplete one, which thereafter takes the prefixal agreement of a transitive; the decreasers — antipassive, reflexive, reciprocal, incorporation — fill or remove the object slot and return an incomplete verb to the complete class, with suffixal agreement (Chiri 1942: §33). The whole chapter can accordingly be read as the derivational traffic between Chiri's two classes, each ‘voice’ a one-way crossing.

14.1 Causatives

The productive causative is -te ~ -re: -re after vowel-final stems, -te after consonant-final ones (Chiri 1942: §55), (Tangiku 1998: 229–230), (Tangiku 2022). For the verbs in -rV the choice follows from the two stem classes established in Chapter 5: a Class I verb (underlying /r/) counts as consonant-final and takes -te with the regular shift of r to n (kor-tekonte ‘give’), while a Class II verb (lexically vowel-final) takes -re like any vowel stem. Citation forms that look identical therefore part ways under the suffix — nukaranukante but kirakirare — and the causative is in practice the readiest test of a verb's class membership. Hokkaidō's third allomorph -e after r (kor-e, nukar-e) is replaced in Sakhalin by -te (Chiri 1942: §55), so the -nte causatives are among the clearest Sakhalin-defining forms in the verbal morphology. Table 1 lists base ~ causative pairs for which both members are attested in Sakhalin sources.

Attested base ~ causative pairs
basecausativeattestation of the causative
ee ‘eat’eere ‘feed sb sth’(Murasaki 2001: text 5)
nu ‘hear’nuure ‘let hear, tell’(Tangiku & Ogihara 2001: Letter 2)
ipe ‘have a meal’ipere ‘give sb a meal’(Murasaki 2001: text 4)
mahnuu ‘take a wife’mahnuure ‘get sb a wife’(Murasaki 2001: text 34)
tuuri ‘stretch’tuurire ‘stretch sth out’(Murasaki 2001: text 18)
pecika ‘ford, cross’pecikare ‘take sb across’(Murasaki 2001: text 18)
koro ‘have’konte ‘give’(Tangiku & Ogihara 2001: Letter 1); (Dobrotvorsky 1875); (Sakaguchi 2021)
nukara ‘see’nukante ‘show’(Murasaki 2001: text 51)
mokoro ‘sleep’mokonte ‘put to sleep’(Murasaki 2001: text 4)
oman ‘go (sg)’omante ‘send’(Murasaki 2001: text 51); (Tangiku & Ogihara 2001: Letter 1)
asin ‘go out (sg)’asinke ‘take out’(Murasaki 2001: text 5)
ahun ‘go in (sg)’ahunke ‘put in’(Murasaki 2001: text 52)
makan ‘go up (sg)’makanke ‘send up’(Murasaki 2001: text 35)
tuhse ‘jump’tuhseka ‘send flying, kick’(Murasaki 1979: 32); (Murasaki 2001: text 10)

The meaning runs the whole causative range from coercion to giving and letting: konte ‘cause to have’ is the ordinary verb ‘give’, and Chiri analysed apparent datives as causees on exactly these verbs (Chiri 1942: §68). The causee is the indexed object (in=konte ‘give me’, eci=nu-re ‘I will let you hear’; Chapter 13).

(1)
orowa then
nean that.one
ipe-re ANTIP.eat-CAUS
teh and
taa EMPH
orowa then
suy again
taa EMPH
mokon-te sleep-CAUS
manu REP

‘Then she fed him, and then put him to sleep again.’

Murasaki 2001: text 4; West Sakhalin, Odasu (Asai Take)

(2)
oman-te-hci go.SG-CAUS-PL
taa EMPH
nukan-te-hci see-CAUS-PL

‘They would send him over and show him (to the girls).’

Murasaki 2001: text 51; West Sakhalin, Odasu (Asai Take)

Both causatives take the plural -hci; the singular stem oman agrees with the single causee.

(3)
acahcipo old.woman
taa EMPH
nean that
unkayoh monster
taa EMPH
pecika-re cross-CAUS
teh and
kema-ha leg-POSS
tuuri-re stretch-CAUS

‘The old woman ferried that monster across: she stretched out her legs.’

Murasaki 2001: text 18; West Sakhalin, Odasu (Asai Take)

pecika and tuuri are Class II (true vowel-final) stems and take -re (Tangiku 1998: 230).

Three minor causative formatives partition the lexicon with -te/-re. (i) -ke pairs with the n/p number stems (Chapter 12): singular ahun-ke ‘put one in’ against plural ahup-te ‘put many in’, so that the suffix choice itself signals object number (Chiri 1942: §§36, 55); the corpus supplies asinke, ahunke, makanke and the east-coast an-e-makan-ke ‘I sent you uphill’ ((Piłsudski 1912: text 22); (Sakaguchi 2024: ex. 3-33)). (ii) -ka derives causatives chiefly from property and posture verbs: tuhse-ka ‘kick’ above, hosipi-ka ‘send back’ (Chiri 1942: §55). (iii) The thematic vowels -a/-i/-u/-e/-o transitivise bare roots (mak-a ‘open’, as-i ‘stand sth up’) (Chiri 1942: §55). Finally -yara is the indefinite-causee causative, ‘have people do’: Chiri's e-yara ‘have (people) eat’ (Chiri 1942: §55), and lexicalised in sinukarara ‘show oneself off’ from si-nukara-yara ‘have people look at oneself’ (Murasaki 1979), (Tangiku 1998: 229–230).

Stacked causatives of the shape -te-re or -re-te do not occur in the corpus or in the descriptions we control. Double causation is instead expressed by a causative over an already derived stem (ipe-re on antipassive ipe, sinukarara with reflexive plus -yara) or analytically with the light verb ki. We take the gap to be real and systematic — the causee slot, once filled, is not reopened — though a form like nukante-yara ‘have people show’ would be the expected shape if it were; finding one in unpublished recordings would settle the point.

14.2 Applicatives

The applicative prefixes promote an oblique participant to object, transitivising intransitives and ditransitivising transitives. The three are not interchangeable: the choice tracks the semantic role of the promoted argument (Dal Corso 2024: 71). e- covers location, goal, topic (‘about’), path, instrument, purpose, cause and accompaniment — Chiri illustrates each for Sakhalin, e.g. neampe an-e-mina ‘we laughed about it’ from mina ‘laugh’, oya kotan e-oman ‘he went to another village’ (Chiri 1942: §§33, 56). ko- adds a goal or addressee, typically a person (Chiri 1942: §58); o- adds places only (Chiri 1942: §57). For the southern Hokkaidō dialects Bugaeva's corpus study (Bugaeva 2006: 188–193) quantifies the same division of labour — ko- the most and o- the least frequent, with about a third of all verb lexemes containing an applicative — and the Sakhalin corpus matches that ranking, as set out below.

(4)
suma rock
kema foot
e-tuhse-ka APPL-jump-CAUS

‘He kicked the rock with his foot.’

Murasaki 1979: 32Dal Corso 2024: 71; West Sakhalin, Rayciska

The instrument kema ‘foot’, expressed obliquely with ani ‘with’ in the unprefixed construction, is promoted to object by e- (Dal Corso 2024: 71).

(5)
rekoro very
ituyaskara feel.affection
rekoro very
an=e=ko-yay-irayki INDEF.A=2SG.O=APPL-REFL-be.grateful

‘I feel great affection, and I am deeply grateful to you.’

Dobrotvorsky 1875Sakaguchi 2021; West Sakhalin, West coast, central (recorded 1867–1872)

ko- adds the addressee, indexed by e= ‘you’, to the intransitive yay-irayki ‘be thankful’.

Piłsudski's corpus also shows ko- in a use where no applied object can be recovered, on zero-valent weather predicates; Dal Corso reads this as a relic of a broader high-transitivity function of the prefix (Dal Corso 2024: 73):

(6)
ko-sistono APPL-dawn
to-noske day-middle
ko-oman APPL-go.SG
kane while

‘Day had dawned and, when it had gone on to midday, …’

Piłsudski 1912: 146Dal Corso 2024: 73; East Sakhalin

ko-oman ‘(the day) went on to (midday)’ is a regular goal applicative; ko-sistono ‘it dawned’ has no retrievable applied object.

The prefix o- is marginal in the Sakhalin corpus: we find it only inside lexicalised place-object verbs (Chiri's locative sense, §57), never as a live alternation parallel to the e- pair above. We treat it as a closed lexical class in Sakhalin, against its modest productivity in Hokkaidō (Bugaeva 2006: 193). Applicatives also feed the rest of the system: an applied object is incorporable (§14.6), and the applicative slot is what reciprocal and reflexive prefixes fill in combinations like u-ko- and yay-ko- (§14.7).

14.3 The antipassive i-

Prefixed to a transitive verb, i- suppresses the object, yielding an intransitive of generic activity: ‘the unspecified, the usual thing’ is understood. Chiri glosses it ‘things’ and lists Sakhalin examples in which the conventionalised object has lexicalised: i-ku ‘drink (i.e. liquor)’ from ku ‘drink’, inkara ‘look (about)’ from i-nukar(a), i-ruska ‘be angry’ from ruska ‘resent sth’ (Chiri 1942: §51). The pair ipe ‘have a meal’ beside transitive ee ‘eat’ (from i-e) is the showcase: the two verbs even split as complete versus incomplete predicates in Chiri's classification (Chiri 1942: §30). All four are richly attested in the corpus; inu ‘listen’ (i-nu ‘hear things’) joins them in the Rayciska recordings.

(7)
okay=an exist.PL=INDEF.S
teh and
ikuu=an ANTIP.drink=INDEF.S
raapoke while
ta in

‘I sat at home, and while I was smoking…’

Kitahara et al. 2003; West Sakhalin, Rayciska (Fujiyama Haru)

iku ‘drink/smoke (the usual thing)’ takes the intransitive person suffix =an, confirming detransitivisation.

(8)
orowa then
munaynu grass.man
taa EMPH
cise house
kaa-ta top-LOC
rikin climb.SG
teh and
nay river
onne to
inkara-ha ANTIP.see-POSS
neampe when

‘Then the grass-man climbed onto the roof, and when he looked out toward the river, …’

Murasaki 2001: text 10; West Sakhalin, Odasu (Asai Take)

The antipassive output re-enters derivation freely: ipe-re ‘give a meal to’ (§14.1) causativises it, and inkara behaves as an underlying r-stem (inkan rusuy, Chapter 12). The prefix is segmentally identical to the indefinite object marker i-, from which it historically derives; the synchronic division — referential first-person or logophoric reading for the agreement marker, generic-object reading here — is drawn in Chapter 13. In our corpus counts, productive (non-lexicalised) antipassives are rare: the construction survives chiefly in the conventionalised activity verbs listed above, which the dictionaries already treat as lemmas ((Hattori 1964)).

14.4 Reflexives yay- and si-

The reflexive yay-, from the root ‘self’, intransitivises a transitive verb, identifying agent and patient: yay-rayki ‘kill oneself’, yay-nuyna ‘hide oneself’ (Chiri 1942: §53). The valency decrease is visible in person marking: east-coast yay-reske-an ‘I raised myself’ takes the intransitive suffix where reske ‘raise’ takes prefixes (Dal Corso 2024: 73–74).

(9)
sine one
yay-reske REFL-raise

‘A girl lived all alone (lit. raised herself).’

Murasaki 2001: text 39; West Sakhalin, Odasu (Asai Take)

(10)
yay-nuyna REFL-hide
teh and
taa EMPH
nukara watch
manu REP

‘He hid himself and watched.’

Murasaki 2001: text 37; West Sakhalin, Odasu (Asai Take)

A second pattern leaves valency intact: yay- marks the possessor of an incorporated noun, coreferent with the subject — the ‘quasi-possessive reflexive’ (Bugaeva and Kobayashi, in (Bugaeva 2022)). Piłsudski's yay-cise-koro ‘have a house of one's own’ (Piłsudski 1912: 238), (Dal Corso 2024: 74) recurs verbatim in Asai Take's tales:

(11)
tu two
yay-cise-koro-hci REFL.POSS-house-have-PL
okay-ahci exist.PL-PL
manuyke and.then

‘The two men lived each keeping a house of his own, and…’

Murasaki 2001: text 51; West Sakhalin, Odasu (Asai Take)

Lexicalised yayko- (reflexive plus applicative) means ‘by oneself, within oneself’ (Chiri 1942: §25); the nineteenth-century west-coast conversations use it in verbs of mental settlement:

(12)

Соннока а яйкурамуосьма.

sonnoka truly
an-yayko-ramuosma INDEF.A-REFL.APPL-be.convinced

‘Truly, we are persuaded.’

Dobrotvorsky 1875Sakaguchi 2021; West Sakhalin, West coast, central (recorded 1867–1872)

ramuosma ‘consent’ is itself ramu ‘mind’ incorporated into osma ‘rush into’.

The parallel prefix si- covers the middle range: with plain transitives ‘oneself’ (si-etaye ‘draw back’), with causatives ‘get oneself V-ed’ (si-kasuy-re ‘get oneself helped’) and ‘feign’ (si-ray-re ‘play dead’) (Chiri 1942: §53); Sakhalin sinukarara ‘show oneself off’ (§14.1) builds on it. In the corpus yay- outnumbers si- by an order of magnitude, and only yay- shows the possessive-reflexive use — a distributional statement that is our own.

14.5 Reciprocal and sociative u-

The prefix u- ‘each other’ intransitivises a transitive verb just as the reflexive does (Chiri 1942: §52). Corpus pairs include sam ‘marry’u-sam ‘marry each other’, nukara ‘see’u-nukara ‘meet’ ((Murasaki 2001: text 34)), koyki ‘attack, beat’u-koyki ‘fight one another’:

(13)
re three
okore all
mahnuu wife.take
re three
u-sam-ahci RECP-marry-PL

‘The three men all took wives; the three girls were wedded (to them).’

Murasaki 2001: text 44; West Sakhalin, Odasu (Asai Take)

mahnuu ‘take a wife’ is denominal: mah ‘wife’ with the verbaliser -nu(u) ‘get, have’, beside hoku-nu ‘take a husband’ (Chiri 1942: §92).

(14)
cisekoro house.own
tura with
taa EMPH
u-koyki-hci RECP-fight-PL
u-koyki-hci RECP-fight-PL
omantene and.then
taa EMPH
cisekoro house.own
rayki kill
teh and
taa EMPH
ociwe throw

‘He grappled and grappled with the master of the house, killed him, and threw him out.’

Murasaki 2001: text 50; West Sakhalin, Odasu (Asai Take)

The co-participant is expressed with comitative tura while the verb itself is reciprocal — the regular discontinuous construction.

Beside true reciprocity, u- has a sociative-collective use that Chiri singles out as characteristically Sakhalin: on intransitives, with or without a causative, it means ‘in numbers, all together’, sometimes amounting to mere plurality — utara u-mokon-te ‘the people slept’, u-ariki-re ‘they came along’ (Chiri 1942: §52). Sentoku's letters supply both this and the combination with ko-:

(15)

обісьта увекарі укобакарі анкіхці

opista all
uwekari RECP.gather
ukopakari RECP.APPL.consult
an=ki-hci INDEF.A=do-PL

‘We will all gather and take counsel together.’

Tangiku & Ogihara 2001: Letter 1; East Sakhalin, Otasan (Sentoku Tarōji)

uwekari is u- plus e-kari with regular glide insertion; ukopakari is u-ko-pakari ‘reckon with one another’.

Lexicalisation has gone furthest in ukoyki itself. In the same letter Sentoku writes cep ukoyki an=e-hci, which Tangiku and Ogihara translate ‘we catch a little fish and eat it’ (Tangiku & Ogihara 2001: Letter 1) — here ukoyki is not ‘fight each other’ but the collective activity verb ‘take (fish), do the fishing’, built on koyki ‘attack, catch (game)’; in Letter 3 it serves as a plain noun, ukoyki porono an ‘the catch was plentiful’ (Tangiku & Ogihara 2001: Letter 3). The reciprocal, sociative and lexical readings shade into one another along a path the corpus shows still open.

14.6 Noun incorporation

A noun in its bare conceptual form may be compounded into the verb, ceasing to be a syntactic argument (Chiri 1942: §§24–25), (Tangiku 2022). The canonical case incorporates the object of a transitive activity verb:

(16)
toyaynu earth.man
wahka-taa water-draw
kusu PURP
san go.down.SG
manu REP

‘The earth-man went down to draw water.’

Murasaki 2001: text 10; West Sakhalin, Odasu (Asai Take)

(17)
eani 2SG
wahka-taa water-draw
eani 2SG
wahka-taa water-draw

‘You fetch the water! You fetch the water!’

Murasaki 2001: text 9; West Sakhalin, Odasu (Asai Take)

(18)
orowa then
nean that.one
niina firewood.gather
kusu PURP
kina-taa plant-pick
kusu PURP
makan go.uphill.SG

‘Then she went up into the hills to gather firewood and pick butterbur.’

Murasaki 2001: text 37; West Sakhalin, Odasu (Asai Take)

niina ‘gather firewood’ (nii ‘wood’) is fully lexicalised; kina-taa ‘pick plants’ is built like wahka-taa.

That these are words and not object–verb phrases is shown by three convergent facts, which we state from our corpus analysis. First, the plural suffix treats the whole as one stem: cehkoyki-hci ‘they caught fish’, ehahtaa-hci ‘they dug lily roots’, with -hci outside noun plus verb. Second, the incorporated noun is non-referential and unmodifiable: a counted or possessed object always stands outside (ponno ponno cep ukoyki ‘catch a few fish’, Letter 1). Third, the complex serves as a bare complement of kusu ‘in order to’ exactly like a simple verb, as above. The same nouns occur loose with the light verb (ceh kii, ehah taa as two words, Chapter 12), so univerbation is a gradient the speakers exploit.

(19)
niina-hci firewood.gather-PL
koh whenever
kuru-hci burn-PL
ceh-koyki-hci fish-catch-PL
koh whenever
e-hci eat-PL
ehah-taa-hci lily.root-dig-PL
koh whenever
e-hci eat-PL

‘They would gather firewood and burn it, catch fish and eat, dig lily roots and eat.’

Murasaki 2001: text 18; West Sakhalin, Odasu (Asai Take)

Which argument incorporates is constrained by grammatical relation. Shibatani's survey of Ainu concluded that incorporation targets objects (including derived, applicative objects) and patient-like intransitive subjects, obliques only after applicative promotion (Shibatani 1988: 210–211); Satō's Hokkaidō counts give object incorporation 85.9 percent, natural-force subject incorporation (sir-pirka type) 6.8 percent, with possessor-stranding and transitive-subject incorporation marginal (Satō 2022: ch. 16, in (Bugaeva 2022)). The Sakhalin corpus fits: object incorporation as above; subject incorporation in the weather verbs (siri-pirika, Chapter 12); and incorporation of a promoted locative object in cip-oo ‘ride in a boat’ (cip ‘boat’ with o(o) ‘sit in’):

(20)
mah upstream
wa from
taa EMPH
sine one
aynu person
taa EMPH
cip-oo boat-ride
wa and
san go.down.SG
manu REP

‘From upstream a man came down riding in a boat.’

Murasaki 2001: text 37; West Sakhalin, Odasu (Asai Take)

Incorporation also interlocks with the other voices: yay- rides on an incorporated possessed noun (yay-cise-koro, §14.4), and body-part incorporation underlies lexical prefixes like he- ‘head’ and ho- ‘rear’, which Chiri groups with the detransitivisers (he-pen-u ‘raise one's face’) (Chiri 1942: §§33, 41).

14.7 Combinations and productivity

The valency machinery is freely recursive within fixed scope: each prefix construes with everything to its right, so the outermost prefix corresponds to the outermost derivation. Attested Sakhalin stackings include u-ko- (reciprocal filling the applicative's object slot: u-ko-itak ‘talk to each other’ (Chiri 1942: §25), u-ko-pakari above), yay-ko- (§14.4), ko-yay- (applicative over a reflexive base: an=e=ko-yay-irayki, §14.2), causative over antipassive (ipe-re), reflexive plus indefinite causative (si-nukara-yara), and applicative over reflexive-applicative:

(21)
cise-ta house-LOC
paye-hci go.PL-PL
ike and
tani now
taa EMPH
acahcipo old.woman
reekoh very
e-yayko-nopuru APPL-REFL.APPL-be.delighted
manu REP

‘They came to the house, and the old woman was overjoyed at it.’

Murasaki 2001: text 51; West Sakhalin, Odasu (Asai Take)

yayko-nopuru ‘rejoice (within oneself)’ takes a further e- introducing the stimulus.

Our productivity assessment from the corpus, text frequencies weighed against type counts, is as follows. Fully productive: causative -te/-re (new formations on loanwords and derived stems; no lexical gaps observed) and the reciprocal u-. Productive but lexically anchored: yay-, applicative ko- and e- — most tokens belong to recurring lexemes, yet fresh combinations occur in single texts. Closed or fossilised: applicative o-, antipassive i-, si-, the causatives -ka, -ke (alive only inside the number-paired motion verbs) and -yara. Object incorporation sits in between: the pattern is well represented but its productive core is a small set of subsistence-activity verbs. Dal Corso reaches a compatible conclusion from the other direction, observing that the secondary, valency-preserving functions of e-, ko- and yay- common in Piłsudski's 1900s east-coast corpus have all but vanished from the west-coast recordings of the 1960s, an erosion he connects with the general loss of productive verbal morphology under Japanese contact (Dal Corso 2024: 70–71). The system the last speakers kept intact was the core: causative, reciprocal, reflexive, and the lexicalised stock of everything else.