Chapter 10Numerals and quantification
The vigesimal numeral system, classifier suffixes, ordinals, and quantifiers.
Sakhalin Ainu numerals fall into two formally distinct series: a bare adnominal series (sine ‘one’, tu ‘two’, re ‘three’…), which can only stand before a noun, and a nominal series derived from it (sineh ‘one’, tuh ‘two’, reh ‘three’…), which is used for counting and stands in the clause like any other noun. Chiri Mashiho 知里真志保 built his classic account of Ainu numerals around exactly this opposition Chiri (1942: §§105–113), and the Sakhalin corpus bears it out throughout. On top of the inherited base-twenty arithmetic, the Sakhalin variety developed a productive decimal system around the borrowed base words kunkutu ‘ten’ and tanku ‘hundred’ — the fullest studies are Murasaki (Murasaki 2009) and Sakaguchi (Sakaguchi 2022). This chapter sets out the cardinal numerals (§10.1), the vigesimal system and the decimal innovation (§10.2), the counting of humans and other unit-noun constructions (§10.3), the general quantifiers (§10.4), numeral syntax and calculation practice (§10.5), and the expression of ordinal rank (§10.6).
10.1 Cardinal numerals one to ten
The basic numerals are adnominals: they modify a following noun and never stand alone as an argument Chiri (1942: §105). To count objects, or to use a numeral as a noun phrase by itself, the adnominal is nominalized with the ‘thing’ suffix that appears as -h after vowels and -pe after consonants; the -h of sineh, tuh corresponds to Hokkaidō -p by the regular coda neutralisation described in Chapter 4 Chiri (1942: §113), (Murasaki 2009: 74). Table 1 gives both series.
| adnominal | nominal | sources | |
|---|---|---|---|
| 1 | sine | sineh | (Murasaki 2025: 24–25); (Chiri 1942: §106); both pervasive in the corpus, e.g. (1), (2) |
| 2 | tu | tuh | (Murasaki 2025: 24–25); corpus tuh pis, see (3) |
| 3 | re | reh | (Murasaki 2025: 24–25); corpus re horokewpo ‘three men’ (Murasaki 2001: text 51) |
| 4 | iine | iineh | (Murasaki 2025: 24–25); (Chiri 1942: §106) (ine) |
| 5 | asne ~ asisne | asneh ~ asisneh | (Murasaki 2025: 24–25); asisne in the east and in Yamada Hayo's speech, cf. (12), asisneh in Dobrotvorsky (Sakaguchi 2022: 69) |
| 6 | iwan | iwampe | (Murasaki 2025: 24–25); corpus iwan monimahpo ‘six girls’ (Murasaki 2001: text 51) |
| 7 | arawan ~ aruwan | arawampe | (Murasaki 2025: 24–25); aruwan in Yamabe's text, cf. (Sakaguchi 2022: 73) |
| 8 | tupesan | tupesampe | (Murasaki 2025: 24–25); (Chiri 1942: §112) |
| 9 | sinepisan ~ sinepesan | sinepisampe | (Murasaki 2025: 24–25); sinepesan (Chiri 1942: §112) |
| 10 | wan | wampe | (Murasaki 2025: 24–25); (Chiri 1942: §106) |
Murasaki records the nominal series as the counting chant itself: sineh, tuh, reh, iineh, asneh, iwampe, arawampe, tupesampe, sinepisampe, wampe Murasaki (2025: 24). Two points of variation cut across the table. First, ‘five’: the west-coast Rayciska form asneh has simplified the cluster of expected *asisneh, while Dobrotvorsky's mid-nineteenth-century west-coast records and east-Sakhalin speakers keep asisne(h) Murasaki (2009: 74), (Sakaguchi 2022: 69); both shapes occur in the corpus, cf. (12). Second, the final consonant: Naert's informants from Taraytomari and Nairo pronounced final -p (sinep), so the -h shapes of Table 1 must not be projected onto every locality Sakaguchi (2022: 69).
The two series divide the labour cleanly in texts. The adnominal counts a following noun:
The nominal form stands alone, as subject, object, or — very commonly — followed by the quantifier particles of Chapter 24:
With ‘two’ and ‘three’ the West-coast narratives very often add the unit noun pis ‘piece’ — tuh pis, reh pis — whether the referents are objects or people; the combination is the default way Asai Take counts pairs (24 tokens of tuh pis in her tales). Chiri records the same element for Hokkaidō as an optional extension tu-p(-pis) (Chiri 1942: §113).
sine doubles as an indefinite article-like determiner ‘a certain, one’ Chiri (1942: §115): story openings of the type sine horokewpo an manu ‘there was a man’ and the temporal formula sine too ‘one day’ (over a hundred tokens in the Asai corpus) are the normal way a new participant or scene is introduced. Beside sine, the old proclitic ar- ‘one (of a pair)’ survives in fixed formations: ‘once’ is ar-suy, never *sine suy Chiri (1942: §113).
10.2 The vigesimal system and the decimal innovation
From eleven to nineteen every recorded Sakhalin variety, from Dobrotvorsky's dictionary to Hattori's Rayciska survey, uses one and the same additive pattern: the unit numeral comes first, then ikasma ‘to be left over’, then wampe ‘ten’ — sineh ikasma wampe ‘eleven’, tuh ikasma wampe ‘twelve’, literally ‘one left over, (and) ten’ Chiri (1942: §107), Sakaguchi (2022: 70), (Murasaki 2025: 25). The element ikasma is the ordinary verb ‘be left over, remain’, alive in the letters of Sentoku Tarōji 千徳太郎治:
10.2.1 The score hoh
‘Twenty’ is hoh ‘twenty’ (Hokkaidō hot), and unlike the numerals of Table 1 it is a noun; its adnominal is derived with -ne, hohne Chiri (1942: §108). The inherited system counts in scores: tu hoh ‘forty’, asisne hoh ‘hundred’, iwan hoh ‘hundred and twenty’ Chiri (1942: §109). The intervening tens are ‘overcounted’ toward the next score with the applicative e-: wan e-tu-hoh ‘thirty’, literally ‘ten (going) to forty’ Chiri (1942: §110), and units are undercounted with ikasma, asisne ikasma hohne ‘twenty-five’ (Chiri 1942: §111).
The two devices are not in free variation: they divide each score at its midpoint. Below it, the count hangs from the score just passed — twenty-one to twenty-nine are ikasma-forms on hohne; above it, the count reaches for the score ahead — thirty-one is sine ikasma wan e-tu-hoh, ‘(one over ten) toward forty’. Ochiai, who states the rule in these terms, observes that Hokkaidō and Sakhalin agree on the switch point in detail and therefore reconstructs the undercounting-to-overcounting alternation itself for late Proto-Ainu, rather than treating the e- forms as a local elaboration; her closest typological parallels are the overcounted vigesimal systems of Tibeto-Burman languages Ochiai (2021).
This machinery was fully productive in the west-coast speech that Dobrotvorsky recorded around Kusunnay in 1867–72 (Dobrotvorsky 1875), and there it scales: hoh serves as base only up to 199, and each higher base is the tenfold of the one below it Sakaguchi (2022: 78–80).
| value | base | formation | serves as base for |
|---|---|---|---|
| 20 | hoh | inherited score | 20–199 |
| 200 | sinewano | fused sine-wan-hoh ‘one ten of scores’ | 200–1,999 |
| 2,000 | waysinewano | ‘ten two-hundreds’ | 2,000 upward |
| 4,000,000 | waysinewano waywanho | 2,000 × 2,000, multiplicative | ceiling of the record |
At the top the dictionary simply multiplies the largest base into itself — waysinewano waywanho ‘four million’, a figure far beyond anything recorded for Hokkaidō Sakaguchi (2022: 79–80). Below the bases, units are slotted in with ikasma at each level:
‘two hundred and twenty-one’
Dobrotvorsky 1875: 305; Sakaguchi 2022: 79; West Sakhalin, Kusunnay area
Dobrotvorsky writes ⟨sine⟩; the final -h is supplied by Sakaguchi.
10.2.2 The decimal bases kunkutu, tanku, wantanku
Alongside this, every source from Dobrotvorsky onward attests a decimal system built on kunkutu ‘ten’ and tanku ‘hundred’, which are nouns usable directly as numerals. Their origin is partly clear: tanku matches Manchu–Tungusic words for ‘hundred’ (Manchu tanggū) and Piłsudski was told it came from the ‘Olcha’ with the marten-trapping trade Piłsudski (1912: 141), (Murasaki 2009: 75), while kunkutu has no good Tungusic match and may be an internal Ainu word pressed into service for trade counting Sakaguchi (2022: 66). By the twentieth century the decimal forms had displaced the scores everywhere: Naert's mid-century informants from Taraytomari and Nairo knew only kunkutu forms for ‘thirty’, and his Rayciska informant offered wanpe ikasma sine hoh only as an archaism Sakaguchi (2022: 71). Hattori's dictionary likewise lists hohne ‘twenty’ as ‘elders' speech’ (onne itah) Hattori (1964: 262), (Murasaki 2009: 74).
| form | sources | |
|---|---|---|
| 10 | wampe | (Murasaki 2025: 25); (Hattori 1964: 262); never *sine kunkutu (Murasaki 2009: 74) |
| 20 | tu kunkutu (elders hohne) | (Murasaki 2025: 25); (Hattori 1964: 262) |
| 30 | re kunkutu | (Murasaki 2025: 25); (Sakaguchi 2022: 71) |
| 50 | asne kunkutu | (Murasaki 2025: 25) |
| 100 | sine tanku | (Murasaki 2025: 25); (Chiri 1942: §112) |
| 110 | sine tanku orowa wampe | (Hattori 1964: 264); (Sakaguchi 2022: 76) |
| 200 | tu tanku | (Murasaki 2025: 25) |
| 1,000 | sine wantanku | (Murasaki 2025: 25); (Sakaguchi 2022: 72–73) |
| 2,000 | tu wantanku | (Murasaki 2025: 25); also from an Odasan speaker in 1967 (Sakaguchi 2022: 73) |
Note the asymmetry preserved at ‘ten’: the bare decade is the native wampe, never *sine kunkutu; the borrowed base begins at twenty Murasaki (2009: 74), (Sakaguchi 2022: 71–72). ‘Thousand’ is the transparent compound wantanku ‘ten hundreds’, so Murasaki could elicit the year 2008 as tu wantanku orowa tupesan paa ‘two thousand and then eight years’ (Murasaki 2009: 75).
The decimal system in running text is best observed in the autobiography that Yamabe Yasunosuke 山辺安之助 dictated in 1912 (あいぬ物語), analysed in detail by Sakaguchi. Score forms do occur there, mostly with years and days:
‘the year Meiji 26 (= 1893)’
Sakaguchi 2022: 73; East Sakhalin, Yamabeci (Yamabe Yasunosuke)
From Yamabe's あいぬ物語 (1913: 5); hohne too ‘twenty days’ recurs there twice. Only six hoh-forms occur in the whole book, against sixteen with tu kunkutu.
But the workaday counting is decimal, and the same value can be assembled in two ways — by bare juxtaposition on the Sino-Japanese model, or with native ikasma:
‘twenty-five men’
Sakaguchi 2022: 75; East Sakhalin, Yamabeci (Yamabe Yasunosuke)
あいぬ物語 p. 18 — the same value as (7), built with ikasma; Murasaki (1979: 88) records the mirror order sineh ikasma tu kunkutu ‘21’ from Rayciska, so the decade constructions wavered even within one speaker's usage.
‘eight hundred and fifty people’
Chiri 1942: §112; Sakaguchi 2022: 77; East Sakhalin, Yamabeci (Yamabe Yasunosuke)
Large sums show the two systems interleaved. In Chiri's example of current usage the hundreds are counted by a score:
No base for ‘ten thousand’ is recorded anywhere (Hattori 1964: 265), (Sakaguchi 2022: 64); beyond a thousand, speakers either stretch the multiplier — adnominal hohne ‘twenty-fold’ on wantanku — or juxtapose number nouns as an outright multiplication:
‘money amounting to twenty or thirty thousand yen’
Sakaguchi 2022: 74; East Sakhalin, Yamabeci (Yamabe Yasunosuke)
hohne wantanku ‘20×1000’ uses the adnominal of hoh as a coefficient; re kunkutu wantanku ‘30×1000’ juxtaposes the number noun ‘thirty’ with ‘thousand’ — Sakaguchi's ‘multiplicative’ type.
10.3 Counting humans and other unit nouns
Hokkaidō Ainu counts humans with a dedicated series in -n (after vowels) and -iw (after consonants): sinen, tun, iwaniw Chiri (1942: §113). For Sakhalin, Murasaki states that this series is absent and that humans are counted with adnominal numeral plus aynu ‘person’ — sine aynu ‘one person’, iwan aynu ‘six people’ Murasaki (2009: 75), (Murasaki 2025: 25). The corpus does confirm the aynu pattern as the productive one:
The statement of absence does not survive the record, however. Sentoku's east-coast letters use sinen in a negative-polarity idiom:
‘There is not even a single person (left).’
Tangiku & Ogihara 2001: Letter 2; East Sakhalin, Otasan (Sentoku Tarōji)
More decisively, Chiri — whose grammar describes Sakhalin, not Hokkaidō — sets the bound series out as a complete paradigm: sinen, tun, ren, inen, asiknen with -n after vowels, then iwaniw, arawaniw, tupesaniw, sinepesaniw, waniw with -iw after consonants, closed by the interrogative hempakiw ‘how many people’ Chiri (1942: §113). These are his stated Sakhalin forms, and they do duty in his plain prose examples:
‘One woman came.’
Chiri 1942: §113; Sakhalin (dialect not further specified)
The bound counter even supplies the unit half of the additive teen frame, where the counted noun is named only once:
The narrative corpus agrees with Chiri rather than with the claim of absence. In Asai Take's tale of the boat-builders, the flying chips are counted off in a chant sinen kuh ray, tunne kuh ray, renne kuh ray ‘one man dead, two men dead, three men dead’ (Murasaki 2001: text 32), preserving sinen and, we take it, tun and ren extended by -ne; and Piłsudski's east-coast corpus has iwan-niw ‘six (rowers)’ in a clause that also shows aynu enko ‘half of the men’ with enko ‘half’ Piłsudski (1998: 287), (Dal Corso 2024: 75). Sakhalin therefore kept both strategies side by side: the inherited, phonologically conditioned -n/-iw counter — complete in Chiri, persisting in chants, formulae, and the low numerals — and the transparent innovation with aynu, which had become the default by the last-recorded generation. Murasaki's ‘absent’ captures the end state of that competition among her Rayciska informants, not a gap in the variety; the pairing of an opaque inherited counter with a semantically transparent renewal is a familiar one cross-linguistically, and here the renewal won.
Other unit nouns combine with the adnominal series in the same way as pis and aynu. Days: sine too ‘one day’, iwan too ‘six days’, with suppletive tutko, rerko ‘two, three days’ (Chiri 1942: §113). Years and months: paa and cuh, as in (6) and in Sentoku's sine cup okta ‘per month’ (Tangiku & Ogihara 2001: Letter 1). Times: suy, with ar-suy ‘once’ (§10.1). Fathoms: tem, in the corpus fused in iwanten kosonto ‘a six-fathom robe’ (Murasaki 2001: text 51). Chiri further records a Sakhalin-only collective atuyta ‘ten head of game’ for counting catch (Chiri 1942: §113).
10.4 Quantifiers
The general quantifiers are adverb-like words that stand before the verb or before the noun phrase they measure. ‘Many, in numbers’ is poronno ‘many’ (from poro ‘big’) and, in the West-coast narratives, above all renkayne ‘plenty’, which in Asai Take's tales is the routine word for ‘in great numbers’ (64 corpus tokens) and which Murasaki's course book likewise uses for ‘a lot’ (Murasaki 2025: 22); its other, clause-linking life (‘because of’) is taken up in Chapter 11.
‘A little’ is ponno ‘a little’ (from pon ‘small’), often reduplicated ponno ponno ‘just a little’ (Murasaki 2001: text 39):
‘All’ is expressed by okore ‘all’ — the high-frequency exhaustive of the West-coast corpus, ‘all of them, completely’ (Chapter 8) — and by emuyke ‘all’, which quantifies over a preceding noun phrase:
The Hokkaidō shape opitta ‘all’ is absent from the Sakhalin corpus, though its regular reflex opista (Chapter 4 (Phonology)) does appear in the Sentoku letters (Chapter 14 (Valency-changing morphology)); okore and emuyke are the usual exponents. Chiri cites opitta only from Hokkaidō sources, distinguishing it from epitta ‘all over’ (Chiri 1942: §123(104), (131)). Negative quantification uses ka ‘even’ with the negative existential, as in (13) and nee aynu ka isam ‘there is nobody’ (Murasaki 2025: 17); see Chapter 17. The interrogative quantifier is hempah ‘how many’ (adverbial hempahno ‘how much’, (Murasaki 2025: 24)), which with following ka yields an indefinite plural: hempah too ka ‘for some days’, hempah poo ka korohci ‘they had several children’ (Murasaki 2001: texts 48, 12).
10.5 Numeral syntax and calculation
Within the noun phrase the adnominal numeral immediately precedes its noun (re horokewpo ‘three men’), like any other adnominal of Chapter 6. When a counted noun phrase is built additively, the classical pattern repeats the noun or replaces its first occurrence with a nominal numeral: sine-p ikasma wan suma ‘eleven stones’, literally ‘one (thing) left over, ten stones’ Chiri (1942: §114). With the borrowed bases this constraint lapses — kunkutu and tanku are themselves nouns, so the counted noun follows the whole numeral phrase without being repeated, as in (7)–(9) Sakaguchi (2022: 75–76).
Used without a noun, the nominal numerals function as floating quantifiers in apposition to an argument, as in (3) and (14); the distributive particle ranke ‘each’ (Chiri 1942: §125(172)) combines with them:
As for calculation practice, Sakaguchi's survey shows a system that could be operated at scale. Addition within a base is handled by ikasma and, between bases, by orowa ‘and then’ — sine tanku orowa sineh ikasma wampe ‘111’ (Hattori 1964: 264), (Sakaguchi 2022: 76): orowa marks the point where the otherwise descending sequence of values gives way to the fixed ascending unit-ikasma-ten block. Multiplication is expressed either by an adnominal coefficient (hohne wantanku ‘20,000’) or by juxtaposing number nouns (re kunkutu wantanku ‘30,000’, (11)); Dobrotvorsky's dictionary runs the same devices up to waysinewano waywanho ‘4,000,000’ Sakaguchi (2022: 79–80). The trade context that drove this — counting marten pelts and cash — is visible in the dictionary's own glosses, where kunkutu and tanku first appear as tallies of marten snares Sakaguchi (2022: 66), and in the money counting of Murasaki's conversation course, sine wantanku en ‘a thousand yen’ (Murasaki 2025: 24).
10.6 Ordinals
There is no ordinal affix. The Hokkaidō pattern with ikin ne (iwan ikin ne ‘sixth’) has no Sakhalin counterpart: the cognate ikinne survives only in its older adverbial sense ‘in a row, all together’ (sine ikinne ‘all as one’), never as an ordinal former (Chiri 1942). The previous descriptions give no Sakhalin ordinal paradigm, and the texts show why none was there to elicit: ranking is handled by two lexical items at the salient low end and by nothing at all above them. This section is our generalisation from the corpus.
‘First’ is suppletive. It is carried by the positional hoski ‘first’ ‘first, earlier’ (adverbial hoskino), with no formal connection to sine ‘one’ — the same dissociation of ‘first’ from ‘one’ seen in English first or Latin prīmus. Its standing narrative opposite is yooponi ‘later, after’:
‘Second, the next’ is the single derived form, (i)otutanu, built on tu ‘two’ (cf. Hokkaidō (o)tutanu); in the Asai corpus it names the second-eldest of a set of siblings. For sets, the lexical series kiyanne ‘eldest’, inoskun ‘middle’, poniwne ‘youngest’ does the same work — the standard framing of the three-brother tales (Murasaki 2001: text 5).
From ‘third’ on, the bare adnominal cardinal stands before the noun and the ordinal reading comes from the narrative sequence alone. In Asai's tale the hero works through the corners of a house one by one, and the third is simply re ‘three’:
The Sakhalin ordinal system is thus suppletive at the bottom and zero-marked above: hoski(no) ‘first’, (i)otutanu ‘second, next’, and thereafter the plain cardinal disambiguated by the order in which the referents were introduced — typologically the commonest of ordinal systems, since ‘first’ (and often ‘second’) are cross-linguistically the positions most prone to suppletion while the rest of the series goes unmarked.