Chapter 12Verb structure and transitivity

12.1 The verbal word and its template

The maximal verbal word of Sakhalin Ainu has the structure in (i): person prefixes or proclitics; one or more valency-changing prefixes (reciprocal u-, reflexive yay-/si-, antipassive i-, applicative e-/ko-/o-); an incorporated noun; the root; valency-changing suffixes (causative -te/-re, -ke, -ka); the desiderative rusuy; the person suffixes -an, -as, -yan; and, outside them, the plural clitic =hci (Chapter 8). Chiri demonstrated the prefixal expansion with built-up forms like u-ko-itak ‘talk to each other’ and usa-oruspe a-e-u-ko-itak ‘we talk to each other about various news’, where the applicative ko- licenses the reciprocal and e- introduces the topic of conversation (Chiri 1942: §25); the polysynthesis that made Ainu famous in typology is this machine at full stretch, though he is careful to note that such piling-up is rare outside epic diction (Chiri 1942: §25).

The suffixal field is rigidly ordered. Causative suffixes attach directly to the stem, before the echo vowel of -rV stems is supplied (kor-te → konte ‘give’; Chapter 5); the desiderative rusuy likewise reaches the bare stem (an-kon rusuy ‘I want to have’), and the plural suffix -(a)hci attaches outside it. Tangiku concludes from this ordering that rusuy is a suffix rather than a free auxiliary in Sakhalin Ainu (Tangiku 1998: 231–232):

Beyond the word so defined comes a short chain of postverbal auxiliaries and particles, unmarked for person, which carry aspect, modality and evidentiality. Table 1 gives the commonest ones with corpus attestations; their semantics is the business of Chapter 15 and Chapter 16.

The auxiliaries are person-blind: in the corpus only the plural -(a)hci ever intervenes onto them (rusuy-ahci above), and person markers stay on the lexical verb or on a supporting ki ‘do’ (§12.5). Where the literature is silent on kusu kara and hemaka, the statement above is our generalisation from Asai Take's tales, where both are frequent and never person-marked.

12.2 Transitivity

Every verb stem of Sakhalin Ainu is lexically specified for its valency. Chiri's classification divides predicates into complete (requiring no complement: intransitives, including all property verbs) and incomplete (requiring one: transitives, plus the copula ne, the single underived verb that takes a complement rather than an object) (Chiri 1942: §§27, 32). The diagnostic is morphological, and it is the pivot of the person system: a first-person or indefinite subject is suffixed to a complete verb but prefixed to an incomplete one — ahun-an ‘I went in’ against an-ahunke ‘I put it in’ (Chiri 1942: §28) (Chapter 13). Homophonous stems are kept apart by exactly this test (Chiri 1942: §29):

There is no lability. A stem never alternates between transitive and intransitive use without overt marking; valency is adjusted only by the derivational morphology of Chapter 14 — compare asin ‘go out’ with asinke ‘take out’, ahun ‘enter’ with ahunke ‘put in’ (Chiri 1942: §§33, 55). Where English uses one verb in ‘the door opened’ and ‘I opened the door’, Sakhalin Ainu pairs makke with maka, both derived from the bare root mak- (Chiri 1942: §55) (§12.4). Even lexical semantics that straddles the divide is split over two stems: ‘think’ is complete yaynu with quoted thought but incomplete ramu with a complement clause, as ‘say’ is itah against ye and ‘eat’ is ipe against ee ‘eat’ (Chiri 1942: §30). Chiri knew only two verbs used in both classes, roski ‘stand (pl)’ and kari ‘turn’ (Chiri 1942: §31).

12.3 Verbal number

A closed set of verbs distinguishes singular and plural stems. Number here is the number of the event's absolutive participant — the subject of an intransitive, the object of a transitive — and, more exactly, of the event itself: one entering against many enterings (Chiri 1942: §§34–36), (Sakaguchi 2024: 124). Three formal types occur in Sakhalin (Chiri 1942: §35), (Sakaguchi 2024: 122–123): suppletion (Table 3); the -n/-p alternation of seven motion verbs (Table 4); and a residual class with plural -pa (tuye ~ tuypa ‘cut’, nasa ~ naspa ‘split’, (Murasaki 1979: 42)), which is far less productive than in Hokkaidō — most Hokkaidō -pa plurals have no Sakhalin counterpart, the plural suffix -hci doing their work (Tangiku 2022), (Sakaguchi 2024: 123).

Hokkaidō's pair rayke ~ ronnu ‘kill’ does not belong here: in Sakhalin rayki ~ rayke are regional variants, ronnu is recorded only as a stylistic alternative, and the Sakhalin prose corpus has no number opposition between them (Hattori 1964: 29), (Sakaguchi 2024: 123). In the corpus we likewise find no token of ronnu.

The bare plural stems of Table 4 end in a stop and therefore surface with the coda h when nothing follows (yah ‘they went ashore’); before vowel-initial suffixes the stop reappears (yap-an, makap-ahci; Chapter 5). In the corpus the bare plurals are rare: they are almost always clothed in -ahci or a person suffix, an observation we owe to our own counts.

Stem number and the plural suffix -hci are independent systems that stack. -hci marks plurality or collectivity of a core participant on any verb (Chapter 8; for its interaction with person, Chapter 13); on a verb with stem number it attaches to the appropriate stem, so plural reference is regularly marked twice:

With transitive members of the set, stem number tracks the object while -hci or yan tracks the subject independently: uyna-hci ‘they gathered them up’ (Murasaki 2001: text 51), and in the plural imperative:

Sakaguchi notes a single east-coast singular ‘come’, eki ~ iki, used in Yamabe's narrative for plural subjects as well, and even combined with the plural suffix (eki-hsi) (Sakaguchi 2024: 122) — a hint that for ‘come’ the suppletion was already decaying in the early twentieth century, while the west-coast tales of Asai Take keep all the pairs of Tables 3–4 intact.

12.4 Stem classes

For the morphology of the preceding sections, four stem classes need to be distinguished. (i) Vowel-final stems: ee ‘eat’, tura, kira ‘flee’, tuuri ‘stretch’. They take post-vocalic allomorphs throughout: causative -re, plural -hci. (ii) Consonant-final stems whose final stop or m is neutralised word-finally: itah ‘speak’ (itak-), ahkas ‘walk’, ahun. They take -te, -ahci, and their underlying coda reappears before vowels (itak-ahci, yap-an; Chapter 5). (iii) Stems in -rV, which split in two: underlying r-final stems (koro ‘have’, nukara ‘see’, mokoro ‘sleep’) behave as consonant-final under suffixation — konte, nukante, kon rusuy, nukan rusuy — while true vowel-final ones (poro, kira, tuuri) do not; this is Tangiku's demonstration, set out with the phonology in Chapter 5 (Tangiku 1998: 225–230). The same r may neutralise to coda h instead of taking the echo vowel — utah beside utara ‘people’, mahnekuh ‘woman’ (Chiri 1942: §16), (Dal Corso 2024: 64). (iv) Bare roots of shape CVC that surface only with a derivational suffix: mak- ‘open’ in mak-a (transitive, thematic vowel) and mak-ke (intransitive), per-e ‘tear’ beside per-ke ‘be torn’ (Chiri 1942: §55) — the seedbed of the valency pairs of Chapter 14.

Citation forms in this grammar follow the surface word: -rV verbs are cited with their echo vowel (nukara, not nukar), coda-neutralised verbs with h (itah, yah), and suppletive verbs as a pair with the singular first (oman ~ paye). Where an underlying stem matters we add it in parentheses (itak-, kor-).

12.5 The light verb ki

The verb ki ‘do’ — Sakhalin kii with the regular lengthening of monosyllables, shortened again under suffixation (ki-hci) (Tangiku 2022) — is the language's all-purpose support verb. Chiri's classic diagnostic for the unity of verbs and property words turns on it: focused predicates of either kind are resumed by ki, as in oman ka ki ‘he did go’ and an ka somo ki ‘he is not there at all’ (Chiri 1942: §21) (Chapter 6); negation with a focus particle takes the same shape, hosipi kayki han ki ‘he did not so much as return’ (Piłsudski 1912: 13), (Chiri 1942: §49) (Chapter 17).

Beyond focus support, ki is a dummy predicate over a lexical verb or verbal noun, which keeps its own internal morphology while ki carries the clausal trappings. Dobrotvorsky's nineteenth-century conversations build whole exchanges on it:

The construction can even split the person marking of a single semantic predicate: in Yamabe's autobiography the object person sits on the lexical verb and the subject person on kii, the lexical verb serving as the dummy's object (Sakaguchi 2024: 52):

Finally, ki forms lexicalised activity predicates with a bare noun: ceh ki ‘fish’ (literally ‘do fish’), yuh ki ‘hunt’ (‘do game’), beside the quotative pro-verb nah ki ‘do thus’. Asai Take strings them into the formula that opens many of her tales:

The same collocations occur univerbated, with the plural suffix treating noun plus verb as one word (cehki-hci ‘they fished’, (Murasaki 2001: text 4)) — the bridge into noun incorporation, taken up in Chapter 14. The variant iki appears in the future periphrasis kusu iki (Table 1) and as a free variant of ki in the east-coast letters ((Tangiku & Ogihara 2001: Letter 1)).