Chapter 58Verbal Number: Suppletive Stems and the Pluractional -pa

58.1 The category and its locus

A small, high-frequency class of Hokkaido Ainu verbs distinguishes a singular from a plural stem. The contrast is carried by the verb: ordinary nouns are number-neutral, and the plurality of a participant is read from context, quantification, or human collectives such as utar ‘people, fellows’, while the verb itself is the only locus at which number is grammaticalized Nakagawa (2024: 124–125, 140–141); Satō (2008: 99–101); Shibatani (1990: 51–54); Nakagawa (2022). The nominal side of that division belongs with Chapter 41 (Nominal Number: Transnumerality, -utar, and the Locus of Plurality), the position of the stem and the plural elements within the verb word with Chapter 56 (The Verb Word and Its Position-Class Template), and the consequences for the personal indexes with Chapter 70 (Person–Number Interactions and Plural Agreement).

Transitivity fixes which argument the plural stem tracks. An intransitive verb indexes the number of S, its sole argument, and the relation is close to agreement: a plural subject index and a plural subject phrase take the plural stem, a singular one the singular stem. A transitive verb indexes the plurality of O, or of the action construed through O, independently of subject number; the singular stem is the default and surfaces even with an overtly plural object Nakagawa (2024: 142–145); Satō (2008: 101). Following Durie, Shibatani states the cross-class generalization as selection of the principally affected argument — S of an intransitive, O of a transitive — rather than as agreement Shibatani (1990: 51–53). Nakagawa accepts the S/O split but reads the intransitive side as agreement-like and reserves the selectional, marked character for the transitive side, so the agreement-versus-selection question is best left open Nakagawa (2024: 145) ‹contested›.

In Corbett's typology of number the category is verbal number proper: a value on the predicate keyed to participant or event plurality, set apart from the agreement number of a nominal controller (Corbett 2000). The pattern is cross-linguistically uncommon, and Nakagawa places Ainu among the few languages of northern Eurasia that carry it, with the isolates Ket and Burushaski; even within Ainu the opposition reaches a minority of the verbal lexicon Nakagawa (2024: 141); Nakagawa (2022) ‹consensus›. Ijäs's learner grammar states the same architecture for the classroom: nouns do not inflect for number, and a subset of verbs have distinct singular and plural forms, set out in her Lesson 9 Ijäs (2023). Her example sentences are constructed teaching material rather than corpus attestations.

58.2 Type A: suppletive stems

Type A is total stem alternation: the singular and plural members are lexically distinct. The class is small and centres on basic existence, posture, and motion verbs, with a handful of transitives of handling and affected change Nakagawa (2024: 146–147); Satō (2008: 102). Dialect membership is part of each entry. Saru, Chitose, Mukawa, and Shizunai have oka beside okay elsewhere; Saru and Chitose have arpa for 'go' where other dialects often use oman; and Tokachi has no singular partner to ronno 'kill', so it treats that verb as number-invariant Nakagawa (2024: 146–147).

The transitive rows are the reason a simple subject-agreement rule fails. A plural object index or a plural object phrase does not force the plural stem: Nakagawa's a=eci=rayke 'I kill you (pl)' keeps singular rayke under an overtly plural object, and one narrator tells the same hunting formula with singular rayke and with plural ronnu for the same event Nakagawa (2024: 143–145). Satō reaches the same conclusion, that the conditions on the transitive plural stem remain partly obscure Satō (2008: 101) ‹contested›. A few transitive pairs depart from the O-locus: eram(u)an / eramuoka 'understand', attested as a pair only in Saru, Chitose, and Shizunai, tracks the number of the subject.

A surface-string count over the aligned corpus points in the same direction without giving a grammatically disambiguated figure. In every pair the singular member outnumbers the plural, most sharply in transitive rayke / ronnu, consistent with the singular as the default. The figures are tokens of the written surface string, and homophonous an, oka, and ek — which also serve as existential, auxiliary, and other functions — inflate the existential and motion rows Nakagawa (2024: 143–145) ‹corpus-suggested›.

58.3 Types B and C: segmentable alternations

Two further patterns have a segmentable exponent and a wider, though still lexically bounded, membership. Type B alternates a final -n in the singular with -p in the plural, and is confined to intransitives of directed motion. The matching transitives are derived with singular -ke (or -ka) and plural -te, and the number relation runs off the moving participant, with rikinka an exception to the -ke shape Nakagawa (2024: 148–149).

Type C replaces a stem-final vowel with -pa. It is the most populous class and consists mainly of transitives denoting a change of state in the object; many pair with a same-root intransitive whose S corresponds to the transitive O Nakagawa (2024: 149–151); Nakagawa (2022). The change-of-object semantics is a strong tendency with lexical exceptions: piru / pirpa 'wipe', suwe / supa 'boil', and ani / anpa 'hold, carry' impose no change on the object and have no intransitive partner Nakagawa (2024: 149–150).

When a derivational prefix si-, he-, or ho- builds an intransitive on a Type-C transitive, the alternation is preserved and the derived verb agrees with its subject: situri / siturpa 'stretch' (< turi / turpa 'stretch something'), hopuni / hopunpa 'rise, get up', hoyupu / hoyuppa 'run' Nakagawa (2024: 150–151).

Derived transitives inherit the number relation of the base that supplies their affected participant. A transitive built on an intransitive keeps subject-type agreement, as with hosippa 'return.PL' (< ho-sipi); a causative in -re tracks the causee, which is the original intransitive subject, so hopunire is used for sending off one slain bear and hopunpare for three wolves Nakagawa (2024: 145–146). The Type-C suffix -pa is therefore a stem-forming, root-bound suffix, the pa1 of the Handbook account, and is kept apart from the outer plural clitic pa2 treated next; its place in the verb word is set out in Chapter 56 (The Verb Word and Its Position-Class Template), and the derivation in Chapter 74 (The Morphological Causative -re / -e / -te) Nakagawa (2022).

58.4 The postverbal plural clitics

Plurality is also marked outside the stem. Western Hokkaido — Saru, Chitose, Ishikari — uses the postverbal element pa2; eastern Hokkaido uses ci; and Sakhalin has =hci / =ahci, the three in rough complementary distribution Nakagawa (2024: 151–159); Nakagawa (2022). Reduplication occupies the same descriptive neighbourhood but ordinarily contributes iterative, continuative, or intensive lexical aspect, and stays outside the number category proper Nakagawa (2024: 151–153). The diagnostic that separates the root-bound suffix pa1 from the clitic pa2 in Saru is position.

The outer plural element primarily indexes participant plurality, especially of the agent, which is the device behind one cross-linguistically unusual property of Ainu: plural-marked transitives can co-occur with a plural transitive subject, as in i=res-pa 'they raised me' Shibatani (1990: 53). The eastern ci behaves the same way and gives the inside/outside test directly: in Horobetsu it is a fossilized suffix inside the person clitic (iki-ci=as), while in Shizunai it is a productive auxiliary outside it (hotke=an ci 'we slept') Nakagawa (2024: 154–155). The interaction of these elements with the second-person-plural index is taken up in Chapter 65 (The eci= Portmanteau (1A→2O) and Second-Person Plural Syncretism).

The status of pa2 is disputed. Nakagawa and the Handbook treat it, eastern ci, and Sakhalin hci as a single clitic class distinct from the stem suffix, on the positional evidence above, with Sakaguchi reaching the same conclusion for hci Nakagawa (2024: 156–159); Nakagawa (2022). Satō's Chitose data make the clitic analysis less secure there: =an pa is rare, a consultant rejected *yaynu-as pa in favour of the bound yaynupa-as, and okaypa is marginal Satō (2025: 25–26) ‹contested›. The descriptive generalization is consistent: the postverbal marker supplies plurality where the stem does not, and adds emphatic or event-plural force where the stem already carries number; Nakagawa reads forms such as kuspare and korpare as evidence that pa2 is drifting toward affixhood.

The Sakhalin marker serves as a contrast. Murasaki analysed -hci as a third-person plural affix, but its optionality and its appearance with non-third-person reference led Nakagawa to class it with optional plural markers rather than person indexes Murasaki (1979: 49–52); Nakagawa (2024: 153–155). It can over-mark a stem that is already suppletively plural.

58.5 Mismatches between semantic and stem number

Two environments separate semantic number from stem number. A semantically singular referent takes the plural stem in the polite imperative: a bare singular stem commands one addressee (ek! 'come!'), but the polite imperative in yan requires the plural stem even for one addressee, so arki yan is well formed and *ek yan is not Nakagawa (2024: 160). The honorific fourth person used for a single respected addressee, and in many dialects a single story protagonist in the fourth person, likewise take the plural stem; these are person-system effects, routed to Chapter 114 (Imperative and Prohibitive (Directive Mood)), Chapter 68 (Honorific and Logophoric Uses of the Fourth Person), and Chapter 70 (Person–Number Interactions and Plural Agreement).

The reverse mismatch is a semantically plural subject specified by a numeral, which ordinarily takes the singular intransitive stem. Tamura's rule of thumb, reported through Nakagawa, is that two, three, and sometimes four favour the singular Nakagawa (2024: 160–162). The threshold is not firm: the culturally salient iwan 'six' often takes the singular as well, and the quantifier poronno 'many' favours the plural yet is recorded with singular predicates by both Nakagawa and Satō, so no predictive rule covers the full range of numeral, quantifier, dialect, and stance Nakagawa (2024: 161–162); Satō (2008: 101) ‹contested›. The quantifier syntax itself belongs with Chapter 55 (Numeral Classifiers, Ordinals, and Quantifier Syntax).