Chapter 8Number

8.1 The plural word utah ~ utara

West-coast utah ‘people’ ~ uta, East-coast utara, corresponds regularly to Hokkaidō utar (coda r → h ~ rV, Chapter 4). Three uses must be kept apart (Sakaguchi 2020a: 188). First, it is an ordinary noun ‘kin, companions, people’, with the possessive form utari(hi) (Chapter 7). Second, placed after an adnominal, a noun or a relative clause it builds a plural noun phrase — nea utara ‘those people’, aynu utara ‘the people’, kotan ohta okay utara ‘the people who are in the village’ (Sakaguchi 2020a: 188) — and in this use it has become the language's general plural marker. Third, unlike Hokkaidō utar, Sakhalin utara also occurs bare, with no modifier at all, as a pronoun-like plural ‘they, people, we, you’; this use was first described by Satō in 1987 and has been reanalysed by Sakaguchi (2020a), whom we follow.

8.2 The nominal plural =ahcin

The second nominal plural is the suffixal clitic =ahcin, with allomorphs =hcin after vowels (the usual realisation, since its hosts end in the possessive suffix -hV), =ahcin after consonants, and southern and west-coast variants in s (=(a)hsin, =sin, =cin) (Sakaguchi 2020b: 4–5). Nothing like it exists in Hokkaidō; within Sakhalin it has been recognised since Piłsudski, who called cin ~ sin a plural sign typically of inanimates (Piłsudski 1912: 36), while Kindaichi mistook it for an emphatic particle — refuted by Chiri, who showed that it stands before case particles, where no such particle can stand (Chiri 1942: §67). Its distribution is rigidly restricted: it attaches only to the possessive (“belonging”) form of nouns, never to the bare conceptual form and never to proper names (Sakaguchi 2020b: 5–8, 13). Hattori's Rayciska paradigm shows the system (Hattori 1961: 6–7):

Within the possessed paradigm, then, number marking is close to obligatory: a possessed noun without =ahcin is singular — or distributively singular, one per possessor, as an-seta-ha shows. The same distributive singular is textually attested: Yamabe writes an-tek-i naa uk-ahsi ‘they took our hands (each one hand)’ (Sakaguchi 2020b: 6), and Hattori elicited:

The suffix is equally at home in the west-coast narrative corpus, on kin terms, body parts and belongings:

Semantically Sakaguchi (2020b: 8–9, 16–17) shows that =ahcin expresses additive plurality only — sets of like objects: siblings, fingers, shoes, pots. This is why it never attaches to personal names, whose referent is unique, and it is the key to the contrast with -utara drawn by Murasaki from her Fujiyama Haru materials: ku=mici-hi=hcin ‘my grandchildren’ (each one my grandchild) against mit-utara ‘the grandchildren (and such small fry)’, where the associative reading is available (Sakaguchi 2020b: 15–16). A marginal associative residue does exist: an-(h)ok-ho=cin, literally ‘my husbands’, refers in its tale to the husband and his younger brother, and Chiri glossed precisely this type ‘the husband and his companion’ (Sakaguchi 2020b: 8–9). In Asai Take's tales we likewise find aaca-ha-hcin ‘her father and his household’ rendered associatively by Murasaki; we analyse such kin-term tokens as the same marginal associative extension, expected on relational nouns whose referents come in households, and keep the additive characterisation for the core. One further host type is the evidential nominalizers hawe(he) ‘voice’, humi(hi) ‘sound, feel’, sir(ihi) ‘appearance’, ruu(he) ‘trace’ (Chapter 16, Chapter 20): formally possessive forms, they take =ahcin when the nominalized event is plural — and then the verb inside the clause carries no -hci, the two markers being in complementary service (Sakaguchi 2020b: 9–12):

Outside these types only the reciprocal verb uriwahne ‘be siblings to one another’ takes the suffix, in nominal use: uriwahne-hcin ‘brothers and sisters’ (Hattori 1964: 42), (Sakaguchi 2020b: 12–13). The element is formally similar to the verbal plural -hci of §8.3, and Chiri treated the two as one morpheme (Chiri 1942: §41); with Sakaguchi (2024: 103) we keep them apart synchronically — their hosts and shapes differ — while accepting the historical connection.

8.3 Number in the verb: the plural clitic -hci

Verbs mark number twice over: a small set of stems alternates or supplements for number (§8.4), and any finite verb may take the plural clitic -hci (-ahci after consonants; east coast also -(a)hsi, -si, west coast south -cin), whose form rules Chiri established (Chiri 1942: §41) and whose syntax Sakaguchi (2024: 77–102) has now clarified. The clitic attaches outside the person suffixes (paye-an=ahci ‘we went’), follows even auxiliaries, and disappears under nominalization, exactly like person marking — Sakaguchi therefore ranks it with the clitic-like person markers rather than with derivational plural stems (Sakaguchi 2024: 79–80, 102):

What does it index? Murasaki listed it as an optional third person plural (Murasaki 1979: 49–51), but it co-occurs with overt person markers of any person, so it is a number marker, not person agreement (Chapter 13). Its controller is a core argument: normally the subject, but Piłsudski already glossed an-kuyra-si ‘I stole up on them’ as plural “because the object (the gods) is in the pl.” (Piłsudski 1912: 40–41). Sakaguchi's synthesis of Hattori's elicitation and the texts is that -hci supplements person marking wherever person affixes leave plurality of a human subject or primary object unexpressed: it is usual with plural third persons (zero-marked) and with the formally ambiguous indefinite series, and it is not used where eci- 2PL or the CI-series already settle the number (Sakaguchi 2024: 97–102). The controller is animate; plural inanimate subjects take a plural stem but not -hci (see example (20) below). In narrative it is pervasive:

8.4 Suppletive and alternating verb stems

A closed set of intransitive verbs of motion, posture and existence distinguishes number in the stem, by suppletion or by alternation of the final consonant (Chiri 1942: §34–35). Sakhalin keeps the inherited core but has fewer pairs than Hokkaidō: the dialect dictionary records 21 singular–plural pairs for Rayciska against 42 for Saru, and the productive Hokkaidō plural suffix -pa survives only in a handful of transitive pairs (tuye/tuypa ‘cut’ type, (Murasaki 1979: 42)) (Sakaguchi 2024: 122–124). The table lists the pairs we can verify in Sakhalin sources; every plural stem in the right-hand column is also attested in the Asai Take corpus.

The -n/-p rows are a regular coda alternation confined to verbs of oriented motion; the plural coda surfaces as h before a consonant (makah sirihcin above; Chapter 4). The corresponding transitives split the same way, with singular -ke against plural -te (ahun-ke ‘let one in’, ahup-te ‘let many in’) and oman-te against paye-re (Chiri 1942: §36) (Chapter 14). The Hokkaidō pair rayke/ronnu ‘kill’ is not a number pair in Sakhalin: Hattori's dictionary marks the two as stylistic variants, and in the texts plural killing is expressed by rayki-hci (Sakaguchi 2024: 123) — in Asai Take's tales raykihci occurs 23 times, ronnu never. Asai's plural of ‘go’ is consistently paye(hci); a form payeh with final h is unattested in the corpus.

The choice of stem follows grammatical rather than referential number. The CI- and AN-series select plural stems even for one referent (Chapter 13); conversely the plural stem with -an is the normal ‘we’:

8.5 Numeral phrases, optionality, and mismatches

Numerals modify the unmarked noun directly — sine aynu ‘one person’, tu monimahpo ‘two girls’, re horokewpo ‘three young men’ — or stand free with the classifier suffixes -n (humans) and -p ~ -pe(h) (things) (Chiri 1942: §105, §113), (Sakaguchi 2022) (Chapter 10). A numerally quantified noun normally takes no plural marker, and since Kindaichi it has been reported that it also prefers the singular verb stem, the numeral making plurality explicit enough (Sakaguchi 2024: 131). The corpus bears this out:

But it is a preference, not a rule. Sakaguchi (2024: 133–140) shows numeral and poronno ‘many’ phrases with either stem, and argues that the real conditioning is construal: a quantity taken as one bounded mass selects the singular stem, individuated members select the plural. A single sentence of Yamabe's makes the contrast:

The same looseness runs through the whole system. In Asai Take's tales we find tu monimahpo utah an manu ‘there were the two girls’, a plural-marked noun phrase with the singular existential (text 51), and ahun ‘enter (sg.)’ predicated of the plural wives in example (17) above; conversely fully redundant strings like example (1), where demonstrative, utah, plural stem and -hci all co-occur. We analyse the situation thus: number in Sakhalin Ainu is everywhere optional and semantically driven — markers are recruited to make plurality (or its associative and distributive shadings) explicit where it matters to the discourse, and left off where it is recoverable, the only near-grammaticalised corners being the possessed-noun paradigm of §8.2 and the grammatical concord of person series with stem number. The genuinely open question is the division of labour between plural stem and -hci with third persons; our working hypothesis, after Sakaguchi (2024: 140), is that the stem tracks construal (mass vs. individuated) while -hci tracks animate participants, and texts in which inanimate plurals take -hci would falsify it.