Chapter 57Transitivity, Valence Classes, and Ambitransitivity

57.1 Transitivity as a formal property

Hokkaido Ainu classifies its verbs by the number and kind of core arguments each licenses, and the classification is decided by structure rather than by the meaning of a translation Nakagawa (2024: 125–127); Satō (2008: 96). Two diagnostics fix a verb's class. The first is the count of core arguments, where a core argument is a bare noun phrase carrying no postposition or case particle; a phrase marked by a local postposition is adverbial and can be added without altering the count, so that in cise or ta menoko hoku tura ipe 'in the house a woman ate with her husband' only menoko is an argument of ipe Nakagawa (2024: 127). The second diagnostic is person marking: an intransitive verb indexes a single subject, while a transitive verb may index a subject and an object Nakagawa (2024: 126). The indexing system that supports both diagnostics is the subject of Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms), and the order of indexes inside the verb word of Chapter 56 (The Verb Word and Its Position-Class Template).

Three descriptive vocabularies coexist in current work and must be kept apart. Nakagawa, with Bugaeva and Kobayashi, prefers numeric labels — 0-, 1-, 2-, and 3-argument verb — because the bare-noun-phrase count and the intransitive/transitive person-affix fact do not always line up and are best named separately; Satō, Tamura, and Ijäs retain the traditional terms intransitive, transitive, and ditransitive Nakagawa (2024: 128); Bugaeva & Kobayashi (2022: §3); Satō (2008: 94–98). The abbreviations used here are structural labels and carry no claim about semantic agency: an intransitive verb may take an agent-like or an affected subject, and both receive the same subject index, a point developed in Chapter 71 (The Split-Intransitive (Active–Stative) Core).

Ijäs's Lessons 2 and 3 introduce the same inventory for learners as avalent, intransitive, transitive, and ditransitive verbs, and present Hokkaido Ainu as a pro-drop, verb-final, head-marking language whose verb has argument "pockets" that are filled before, and more obligatorily than, any external noun-phrase "slot" Ijäs (2023). That image renders head-marking in the sense of (Nichols 1986) and matches Satō's argument-slot account of the same fact Satō (2023: 42–46). Two background conditions stated in the same lessons frame the classification: verb forms carry no tense inflection, and nouns are number-neutral unless plurality is supplied lexically or by context, so that cep 'fish' may be read as singular or plural Ijäs (2023); Nakagawa (2024: 124–125); Shibatani (1990: 32–33). Number enters the verbal system separately, through stem alternation and the plural elements treated in Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa).

57.2 Person marking as the transitivity test

The person rows that make the intransitive/transitive contrast visible in surface form are the first-plural exclusive and the fourth (indefinite) person. On an intransitive verb these appear as subject suffixes, =as and =an; on a transitive verb the corresponding subject markers are prefixes, ci= and a=. Satō gives the operational test in these terms: put the verb in the first-person plural, and inkar ‘look’ yields inkar=as while nukar ‘see’ yields ci=nukar Satō (2008: 96). He notes that this side-switch is highly irregular and still without a convincing explanation Satō (2008: 96). The remaining rows are uninformative: the first singular uses the same prefix ku= for subject of either class, the second person is neutral across subject and object, and the third person has no overt segment in any role.

The pair inkar and nukar is the standard demonstration that a translation is an unreliable guide to class. Both render English visual predicates, but inkar names an intransitive act of directing the gaze with no fixed object, whereas nukar takes a specified object. In one clause chain the fourth person therefore surfaces as a suffix on the intransitive stem and as a prefix on the transitive stem Nakagawa (2024: 131).

Translation cannot decide class because the verb itself carries the argument structure. Ijäs illustrates the ordinary transitive frame with a third-person subject and a third-person object: neither is segmentally marked, yet the two bare noun phrases fill the A and O positions of a two-place verb Ijäs (2023). The verb word indexes at most one subject and one object — the two-index cap — even where a clause has more arguments than that, a limit that becomes load-bearing for three-place verbs and for the debates of the final section Satō (2023: 39).

57.3 Zero-place and one-place verbs

The zero-place class contains predicates with no grammatical subject and no object. Most are ambient or weather expressions built on a bound element such as sir- 'surroundings' or a non-free root plus a verb: sirpirka 'be fine weather', sirkunne 'get dark', and mean 'be cold (weather)', the last analysable as me- 'cold' plus an 'exist' Nakagawa (2024: 128–130); Shibatani analyses the type as subject incorporation Shibatani (1990: 33). Ijäs groups the same predicates as avalent and notes that descriptive meanings belong to the verbal system, since Hokkaido Ainu has no adjective class distinct from verbs and a stative verb such as hure 'be red' carries the inchoative reading 'become red' as well Ijäs (2023); Nakagawa (2024: 124); Satō (2008: 2). Where the contributing element is itself a free noun the analysis is open: Ijäs treats apto as 'it rains' as a single avalent unit, whereas Nakagawa leaves it undecided, since apto 'rain' is an autonomous noun and the string may be a two-word subject-plus-verb clause Nakagawa (2024: 129–130); Ijäs (2023) ‹contested›.

The boundary between zero-place and one-place verbs is sharpest in the temperature vocabulary. Climate predicates take no person index; the felt-temperature counterparts take a subject. Nakagawa (2024: 129–130)

One-place verbs cover agentive activities such as apkas 'walk' and ipe 'eat a meal', changes of state such as rewke 'bend', and properties such as pirka 'be good' and pewre 'be young'. The single argument is grammatical and need not be a semantic agent: a patientive subject still takes the subject index, as in rittursere=an 'I am rolled by the waves' and ku=merayke 'I feel cold', where the affected participant is marked with the ordinary S set Nakagawa (2024: 126–127). The active/unaccusative cut does resurface, but in derivation and incorporation rather than in the person paradigm, and it does not split the intransitive subject marking Bugaeva & Kobayashi (2022: §5).

57.4 Two- and three-place verbs

A two-place verb licenses a subject and a single object, both bare noun phrases, whatever the translation suggests about patient, goal, or other role. Hokkaido Ainu draws no formal distinction between a direct and an indirect object: what a Japanese rendering marks with を or に is the same bare object in Ainu Nakagawa (2024: 127). This is why cep ipe cannot mean 'eat fish': ipe is the intransitive 'eat a meal', and a specified food object requires a transitive such as e 'eat (something)' Nakagawa (2024: 127). Several near-basic pairs show that close semantic correspondence does not entail formal identity; the most basic of them have to be learned as distinct stems Satō (2008: 97); Nakagawa (2024: 131).

The copula ne belongs with the two-place verbs in argument structure, relating a subject to a predicate nominal, and it takes transitive-type subject marking. Its person behaviour is exceptional, however: it indexes only the subject and admits no object index, so that okkayo ku=ne 'I am a man' is well formed while an object-indexed *en=ne is not Nakagawa (2024: 132); Shibatani (1990: 50). Predicate-nominal syntax is treated in Chapter 96 (The Copula ne and Predicate-Nominal Clauses), and the copula reappears in the final section as Satō's central case for distinguishing several measures of valence.

Three-place verbs license a subject and two bare objects, both omissible and without a fixed grammatical order between them. With kore 'give', the theme and the recipient are both bare third-person phrases and only the subject is indexed below. Almost all three-place verbs are derived, by an applicative such as ko- or the causative -re, the chief base-form member being the locative o 'put X into a place' Bugaeva & Kobayashi (2022: §4.2); Nakagawa (2024: 134). Satō analyses the object index on kore as tracking the recipient, a secundative pattern, a point on which Bugaeva's account differs; the indexing options are set out in Chapter 66 (Object Indexing: Monotransitive and Ditransitive Clauses) and Chapter 93 (Argument Realization: Affix Indexing and Free NP Order) ‹contested›.

Ijäs presents the ditransitive frame as subject, indirect object, and direct object, with the recipient placed before the theme Ijäs (2023). Nakagawa's examples of epakasnu 'teach' show the two objects and the subject in several preverbal orders with one set of argument roles, so the recipient-before-theme sequence is a teaching convention rather than a grammatical requirement Nakagawa (2024: 363–365). The productive operations that build and rearrange these frames are surveyed in Chapter 73 (Valency, Voice, and Noun Incorporation: Overview).

57.5 Locative-object verbs

A small closed class takes a place as a bare object, with no ta or or ta. These locative-object verbs cut across the two- and three-place distinction: oma 'be in, enter a place', osma 'plunge into a place', and kus 'pass through a place' are two-place, while o 'put X into a place' is three-place and is the chief base-form ditransitive of the language Nakagawa (2024: 212–213). The defining property is the absence of a local postposition, so that *cise or ta osma is ungrammatical against intransitive cise or ta ahun 'enter into the house', where ta is required Nakagawa (2024: 128).

The place noun within such a frame works like a covert case marker. With omare 'put X into a place', ikayop omare is read as 'put the quiver somewhere', since a quiver is a theme, while ikayop or omare is read as 'put something into the quiver', with or supplying the locative object Nakagawa (2024: 213). The applicative o- derives further locative-object verbs from non-locative bases, and the local-postposition system that these verbs bypass is treated in Chapter 53 (Postpositions, Local Case, Motion Events, Comitative, and Instrumental) and Chapter 81 (The o- Applicative).

57.6 Arity and the derivational calculus

Nakagawa generalizes valence into an arithmetic over the verb word. Every element — the base verb, each derivational affix, an incorporated noun, and each person index — carries an arity value, and the sum predicts how many bare external arguments the surface predicate still licenses Nakagawa (2024: 136–140). A base intransitive contributes +1, a base transitive +2, and a base ditransitive +3; applicatives and the productive causative add +1; a noun root or a concept-form prefix that fills an argument slot subtracts 1; a single person index subtracts 1 and a subject-plus-object pair subtracts 2.

The calculus earns its keep where one segment has several functions. Built on intransitive itak 'speak' (+1), the applicative ko- gives ko-itak 'speak to' (+2), the subject index in ku=ko-itak brings the count back to +1, and the reflexive yay- in ku=yay-ko-itak fills the object position to leave 0, so that a further bare object is excluded Nakagawa (2024: 138). The same arithmetic separates the three uses of prefixal e-: it must be the applicative (+1) in a=ureepunpa 'I kicked it up with my foot' for the count to admit a bare object, and the affiliative prefix (0) in a=koinawepunpa 'we offered inau' Nakagawa (2024: 138–139). Satō credits this arity concept as Nakagawa's central theoretical contribution Satō (2025: 20). The valency-changing operations that feed the calculus — the applicatives, the causatives, the antipassive i-, and the reflexive and reciprocal prefixes — are catalogued in Chapter 73 (Valency, Voice, and Noun Incorporation: Overview), Chapter 74 (The Morphological Causative -re / -e / -te), Chapter 78 (The Applicative System: Overview and the Preverb-vs-Applicative Debate), Chapter 83 (The Antipassive / Detransitive i-), Chapter 84 (The Reflexive yay-), and Chapter 86 (The Reciprocal u-) Bugaeva & Kobayashi (2022: §4); Bugaeva (2021); Nam (2021).

One commitment of the system bears directly on the debates below. Third person contributes no affix and therefore no negative value; assigning −1 to an unpronounced third-person affix would throw off the arithmetic, so Nakagawa treats third person as having arity 0 Nakagawa (2024: 137). Number morphology is likewise arity-neutral, while causatives and applicatives change the verb's class, so the verbal-number system can be kept apart from valence and is developed in Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa) Nakagawa (2024: 137, 140).

57.7 Ambitransitivity and lability

Hokkaido Ainu has little productive ambitransitivity of the English type. A single bare form serving as both intransitive and transitive is vanishingly rare; Satō's inventory of correspondence types yields effectively one identical-form pair, the plural roski 'stand' and 'stand something up'. The other pairs show either partial derivational relatedness, as with transitive nunpa 'wring' beside intransitive inunpa, or no formal resemblance at all, as in the basic vocabulary of the preceding table Satō (2008: 97). Valence alternation is thus coded by derivation, incorporation, or a separate lexical stem rather than by lability.

What looks like lability is in most cases an uncoded alternation of semantic role within a fixed argument count. Bugaeva and Kobayashi, after Kobayashi's work on valency, reduce the roles to agent, theme, and space-location, and observe that some intransitives admit either an agent or a theme as their single argument: as 'stand' takes a standing person or a standing tree, whereas apkas 'walk' is agentive only and tuy 'be cut' theme-like only Bugaeva & Kobayashi (2022: §5.1.1). The alternation becomes visible in causativization, where a role-flexible intransitive forms two causatives: the direct asi 'stand something up' for the theme reading and the indirect aste 'make someone stand' for the agent reading Bugaeva & Kobayashi (2022: §5). The same active/theme split governs which noun a verb can incorporate, a domain pursued in Chapter 88 (Object (Patient) Noun Incorporation).

Verbal number gives a further reflex of the formal nature of transitivity. The intransitive plural stem tracks the number of the subject and behaves much like agreement in Nakagawa's description, while the transitive plural stem tracks the plurality of the object or of the event, the singular remaining the default even when a plural object is overt; Shibatani, following Durie, treats the alternation as a selectional restriction on the principally affected argument — subject of an intransitive, object of a transitive Nakagawa (2024: 142–145); Satō (2008: 101); Shibatani (1990: 51–53). A small set of cognition transitives is exceptional in that the plural stem tracks the subject instead, as in the plural understanding verb below.

57.8 Arity against the three-valency analysis

What kind of valence the grammar should count is the central open question of this domain, and the two leading positions are to be held side by side rather than adjudicated. Nakagawa's arity is a single surface-computable property: the value of the final form predicts how many bare noun phrases remain and how person indexes reduce that capacity, and it disambiguates homophonous prefixes, incorporated nouns, and person indexes in the same calculation Nakagawa (2024: 136–140).

Satō argues that one number collapses distinctions a head-marking language needs to keep, and proposes three valencies operating at once. The first is the argument-slot count, an invariant abstract property of the lexical or derived verb, filled equally by a person affix, a nominal-derivational prefix, or an incorporated noun — so that incorporation is slot-filling and needs no head-movement analysis of the kind associated with (Baker 1988). The second is M-transitivity, the number of macroroles available for person indexing in the sense of (Foley & Van Valin 1984), capped at two. The third is morphological valency, the intransitive or transitive behaviour of the final form Satō (2023: 42–52). The copula displays all three: its argument-slot count is two, but the predicate-nominal term is not a referential undergoer, so its M-transitivity is one and it takes no object index, while its morphological valency is transitive, as the incorporating nispa a=ne 'I am a master' shows. This derives the excluded *en=ne from a principle Satō (2023: 52). Satō notes that his own scheme is complex and remains to be justified Satō (2023: 61) ‹contested›.

Two narrower disputes follow from the same divide. Nakagawa records a marginal four-place category from causativized ditransitives such as epakasnure 'make someone teach someone something', korere, and a textual omarere 'made someone put something into something' Nakagawa (2024: 135). Satō declines to raise so rare a configuration to a regular class, given that the ordinary three-place verb already exceeds the two-index cap and is syntactically special on its own Satō (2025: 22–23) ‹contested›. The second dispute is the third-person zero. Nakagawa's arity excludes any third-person affix, while Satō holds that a zero third-person affix is required to explain why a verb cannot host three person indexes — why, for instance, *en-e-raykere 'he makes me kill you' is impossible, the two overt affixes having already filled the cap with no room for the third-person causer Nakagawa (2024: 137); Satō (2023: 39); Satō (2025: 23–24) ‹contested›. Ijäs's zero sign for the unmarked third person is compatible with either side if it is read as the absence of an overt segment Ijäs (2023).

The observable diagnostics remain bare-noun-phrase valence, person-index behaviour, and derivational morphology, and these are stable across the analyses. Arity serves as the working calculation for stem formation and argument realization; Satō's three-valency account is the reference wherever the copula, the ditransitives, incorporation, or the zero-affix question make a single count inadequate. The realization of overt and omitted arguments continues in Chapter 93 (Argument Realization: Affix Indexing and Free NP Order), and the voice operations that alter these frames in Chapter 73 (Valency, Voice, and Noun Incorporation: Overview).