Chapter 60Denominal and Deverbal Verb Derivation

60.1 Overview

Hokkaido Ainu derives verbal stems from nominal and verbal sources through mechanisms that must be distinguished from the valency and voice operations treated in the chapters routed from Chapter 73 (Valency, Voice, and Noun Incorporation: Overview). Valency operations — applicatives, causatives, antipassive, reflexive, and reciprocal — alter the verb's external argument count in a principled and largely productive way. The processes documented below either place a base into an argument class without changing that count, or build a new stem whose argument class is determined by the source. They include the combination of bound nominal roots with verbal heads, the copula ne ‘be, become’ and the transitive verb kar ‘make, do’ as routes for nominal predication, adverbial prefixes, and the deverbal stem modifications that alter lexical aspect or number without altering argument structure.

The architecture is head-marking in the sense of (Nichols 1986): a derived stem must enter one of the four argument-count classes — zero-place ambient predicates, one-place intransitives, two-place transitives, or three-place verbs with two bare objects — and the person-index set then diagnoses the class of the result. Nakagawa assigns every formative a numerical arity value so that the derived form's external argument capacity is computable from the visible morphology Nakagawa (2024: 136–139). Satō accepts the same head-marking premise but separates three simultaneously active valency dimensions: an invariant argument-slot count, macrorole transitivity, and the morphological class of the derived form; the copula ne and three-place verbs motivate the split Satō (2023: 42–61). The two accounts are presented together where they produce the same classification and distinguished where they diverge ‹contested›.

60.2 Visual paradigm for one stem

For data modelling, it is useful to visualize an Ainu stem the way Japanese grammars visualize intransitive–transitive pairs: the stem may branch into a direct transitive, a productive causative, an anticausative, an autocausative, a reflexive, and an antipassive. The branches are not symmetric and should not be assigned one flat tag. Bugaeva's 2025 analysis is especially important here: -ke is the primary anticausative marker, while verbal si- is mainly autocausative and only secondarily anticausative Bugaeva (2025: §1.2, §4).

60.3 Denominal ambient predicates

The most clearly documented denominal verbs in the Hokkaido evidence are ambient zero-place predicates built from bound nominal roots. Nakagawa classifies sirpirka ‘be fine weather’ (< sir ‘surroundings, weather’ + pirka ‘be good’), mean ‘be cold (weather)’ (< me- 'coldness root' + an ‘exist’), hawas ‘it is said’ (< haw ‘voice’ + as ‘stand’), sirkunne ‘get dark’, and sitcuk ‘become autumn’ as zero-place verbs: they take neither a subject nor an object, and therefore no person index appears on them Nakagawa (2024: 128–130). Shibatani describes the same ambient and weather predicates as a form of subject incorporation, with the nominal element supplying the domain that an ordinary intransitive would predicate of an external subject Shibatani (1990: 33). The fuller treatment of the incorporative analysis is in Chapter 89 (Subject Incorporation and Possessor-Stranding).

The autonomy of the source noun determines whether the string constitutes a single word or a two-word sequence. The roots sir- and me- cannot stand as ordinary nouns in the relevant predicates — *sir anak pirka and *me ka an are ungrammatical — so sirpirka and mean are single zero-place verbs. The noun apto ‘rain’, by contrast, is free, and Nakagawa leaves apto as 'it rains' undecided between a zero-place compound and a two-word noun-plus-verb clause Nakagawa (2024: 129–130) ‹contested›. The same bound-vs-free diagnostic applies to other weather-like expressions.

A set of temperature terms shows the zero-place and one-place contrast side by side. The zero-place member describes the ambient climate and carries no person index; the one-place member describes a felt bodily state and carries the intransitive subject index. Nakagawa (2024: 128–130)

Property predicates — forms such as poro ‘be big; become big’ and pirka ‘be good; become good’ — are intransitive verbs with no separate adjective class; a stative reading ordinarily carries an inchoative one as well Nakagawa (2024: 124); Shibatani (1990: 33–34). Their clausal behaviour is treated as a subclass in Chapter 37 (The 'No Adjective Class' Thesis: Property Concepts as Stative Verbs) and Chapter 99 (Property-Concept (Adjectival) Predication without a Copula). The denominal ambient predicates above are their zero-place counterpart: the nominal root takes the same finite-verb template but licenses no external argument.

60.4 The copula ne as a denominal predicate head

The copula ne ‘be, become’ is the primary path from a nominal expression to a finite predicate. A noun or noun phrase precedes ne as a predicate nominal, and the result is a two-place construction with a subject and a predicate nominal argument. Nakagawa classifies ne as a two-place verb, yet it admits only the subject person index: the object-indexed form *en=ne is excluded in his diagnostic example, while the subject takes the transitive-subject prefix set Nakagawa (2024: 132); Shibatani (1990: 49–50).

Satō uses this pattern to motivate the separation of three valency dimensions. The argument-slot count of ne is two, since the construction predicates a class or identity; its macrorole transitivity is one, since the predicate nominal is not a referential undergoer and therefore licenses no object affix; its morphological valency is transitive, which explains the transitive-subject-prefix pattern. Lexicalized denominal compounds such as nispa a=ne 'I am a master' and inne 'be numerous' (< ir-ne) follow the same transitive-prefix pattern Satō (2023: 52) ‹contested›. Nakagawa's surface observation that ne never takes an object index and Satō's principled account of why it does not are two descriptions of the same morphological fact.

Because Ainu property verbs carry an inchoative reading alongside their stative one, a noun plus ne can similarly express both 'be X' and, in the appropriate aspectual context, 'become X'. A productive denominal suffix -ne detached from the copular construction has no independent support in the available evidence; lexicalized forms such as inne are best analysed as the copula incorporated into a compound stem rather than as the output of an autonomous verbalizing suffix ‹corpus-suggested›. The syntax and semantics of predicate-nominal constructions are treated in Chapter 96 (The Copula ne and Predicate-Nominal Clauses).

60.5 kar and the word–phrase boundary

The transitive verb kar ‘make, do’ supplies a second denominal route. A noun or nominalized expression may function as the object of kar, and frequent collocations may lexicalize as single verb stems. The synchronic default is that kar is an independent lexical transitive verb. Nakagawa's example tanpe kar pe en=ne, cited to show the copula's blocked object index, shows kar functioning as the verb of a relative clause 'the one who made this'; it takes ordinary transitive morphology and is clearly separate from ne Nakagawa (2024: 132). Horobetsu kar-ici 'make, do' — with the eastern plural suffix ci inside the person index — is a verbal-number fact about the stem itself, not evidence for a productive denominal verbalizer Nakagawa (2024: 154–155).

A transparent N kar sequence is therefore a two-word clause with kar as its verb. A single derived verb stem arises only where morphological diagnostics confirm it: fixed wordhood, phonological fusion, or person and number morphology showing that the nominal element has been incorporated into the verb word rather than occupying an external object slot. Nakagawa applies the same bound-vs-free diagnostic to the weather predicates: bound roots such as me- and sir- support a single-word analysis, while free nouns leave the boundary unsettled Nakagawa (2024: 129–130). Without those independent tests, the default analysis of N kar is a two-word clause ‹corpus-suggested›. The light-verb and pro-verb uses of kar, including its behaviour with incorporated objects, are treated in Chapter 59 (Light Verbs, Pro-Verbs, and the Auxiliary-Verb Construction) and Chapter 88 (Object (Patient) Noun Incorporation).

60.6 Deverbal derivation without valency change

Several deverbal processes change the shape or the lexical-aspect profile of a verb stem without altering its argument frame. Nakagawa assigns arity 0 to adverbial prefixes such as ar- 'completely' and ru- 'slightly': they modify the degree or thoroughness of the denoted event and leave the external argument count unchanged Nakagawa (2024: 137). Their position in the verb template and their interaction with person indexing are set out in Chapter 56 (The Verb Word and Its Position-Class Template).

Reduplication adds iterativity, continuity, or intensity without altering the argument count. Nakagawa distinguishes root reduplication, partial-stem reduplication, and full-stem reduplication: turi ‘stretch (sth)’ yields tutturi (root-reduplicated, marking repeated action), turiri (partial, marking result-state persistence), and turituri (full, marking sustained repetition). He treats reduplication as encoding lexical aspect rather than verbal number, though the two interact at the surface Nakagawa (2024: 151–153); Nakagawa (2022: §1).

The Type-C vowel-replacement alternation creates plural stems from singular ones without changing argument count: tuye ‘cut (sg)’ / tuypa ‘cut (pl)’, yasa ‘tear (sg)’ / yaspa ‘tear (pl)’. Nakagawa assigns arity 0 to the suffix -pa in this pattern, so the plural stem enters the same argument class as the singular source Nakagawa (2024: 137, 149–150). The full verbal-number system — Types A (suppletive), B (-n/-p), and C (-V/-pa), the auxiliary pa, and the relationship between subject vs object plurality and stem choice — is treated in Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa).

Derived intransitives formed by prefixes such as si-, he-, and ho- on transitive bases inherit the verbal-number type of their source. Nakagawa lists si-turi / si-turpa 'stretch (intr)', ho-puni / ho-punpa 'get up, fly up', and ho-yupu / ho-yuppa 'run' as derived intransitives that still undergo the Type-C alternation Nakagawa (2024: 150–151). Number agreement in these forms follows the subject — the single argument of the intransitive — because the prefix has converted the transitive base to an intransitive. The argument class of the output, not the source, determines the number-agreement target Nakagawa (2024: 145–146). The prefixes themselves, insofar as they create anticausative or middle-like readings, are treated in the voice chapters via Chapter 73 (Valency, Voice, and Noun Incorporation: Overview).

60.7 Transitivity pairs

Hokkaido Ainu has almost no productive lability: a single stem serving as both intransitive and transitive is extremely rare. Satō reports only roski ‘stand (pl) / stand sth up (pl)’ as the standard case of a form that fills both slots, and he presents it as an exception rather than a productive pattern Satō (2008: 97). The near-absence of lability means that most intransitive–transitive correspondences are lexical pairs, partial derivations, or suppletive-looking forms that require separate learning Satō (2008: 97); Nakagawa (2024: 131–132).

Three formal correspondence types occur. First, identical surface forms: only roski. Second, partial formal similarity, as in nunpa ‘strain, wring’ (vt) and inunpa ‘strain liquor for ritual offering’ (vi), where the intransitive member is analyzable as the antipassive prefix i- attached to the transitive base, though the pair has lexicalized to distinct meanings Satō (2008: 97). Third, no formal similarity at all — suppletive-looking pairs of basic vocabulary where the relationship between the two stems is not synchronically derivational Nakagawa (2024: 131). Selected pairs of the third type are listed below.

The class membership of a form is diagnosed formally, not semantically. A vi takes the suffixal fourth-person subject marker (=an) and the suffixal first-person-plural exclusive marker (=as); a vt takes the corresponding prefixal forms (a=, ci=). The contrast between inkar and nukar — glossed by the same Japanese verb 見る — appears within a single sentence:

Bugaeva and Kobayashi identify a related pattern of uncoded role alternation. Certain one-place verbs may take either an agent-like or a theme-like subject without any change in form: as ‘stand’ can express both 'stand there (agent subject)' and 'stand upright (theme subject)'. The alternation becomes visible under causativization — an agent subject yields the indirect 'make sb stand' formation, a theme subject the direct 'stand sth up' formation Bugaeva & Kobayashi (2022: §5). That distinction intersects with the transitivity-pair suffix system, where the inventory of -V, -ke, and -ka formations and their productivity are treated in detail.

60.8 Valency-changing derivation routed outward

Productive valency-changing operations form a separate system whose paradigms are documented in the voice and valency chapters. Applicatives e-, ko-, and o- add or promote an object and raise the verb's arity by one; the productive causatives -re, -e, and -te add a causer argument; antipassive i- absorbs the object and yields an intransitive; reflexives yay- / si- and reciprocal u- fill the object slot internally so that a subject and an object of the same person are not both indexed by person affixes Bugaeva & Kobayashi (2022: §4); Nakagawa (2024: 137–139, 183). Their paradigms are in Chapter 78 (The Applicative System: Overview and the Preverb-vs-Applicative Debate), Chapter 74 (The Morphological Causative -re / -e / -te), Chapter 83 (The Antipassive / Detransitive i-), Chapter 84 (The Reflexive yay-), Chapter 85 (The Reflexive-Possessive si- and the yay-/si- Division), and Chapter 86 (The Reciprocal u-).

The homophony of the prefix e- shows why holding these categories apart matters. In one derived verb it is the applicative with arity +1 (adding an object); in another it is the affiliative nominal prefix with arity 0 (contributing a body-part relation); in another it is a second-person object index with arity −1. Nakagawa's worked examples a=ureepunpa 'I kicked it up with my foot' (applicative, arity +1) and a=koinawepunpa 'we offered inau' (affiliative, arity 0) are distinguished only by the arity computation, since the resulting argument count differs Nakagawa (2024: 138–139). The affiliative nominal prefix is defined in Chapter 44 (Concept Form vs. Affiliative (Possessed) Form).

Locative-object verbs form a small class whose distinguishing property is that their object is a bare place expression. Nakagawa lists oma ‘be in / enter a place’, osma ‘plunge into a place’, and kus ‘pass through a place’ as two-place verbs of this type. The single base-form three-place verb in his inventory is o ‘put X into a place’; derived three-place locative-object verbs such as omare ‘put into’, osmare ‘plunge into’, and kuste ‘pass through’ are formed by applicative or causative processes from non-locative bases Nakagawa (2024: 212–213).

The derivational chain that Satō documents on the base ruska ‘be angry at’ illustrates the regularity of the full valency system: applicative ko- raises the base to a three-place verb (ko-ruska); reflexive yay- reduces it (yay-ko-ruska); antipassive i- yields a one-place intransitive (i-ruska); applicative ko- re-applies (ko-i-ruska); and causative -re raises that in turn (i-ruska-re, ko-i-ruska-re, and further) Satō (2023: 40) ‹consensus›. Each step alters the verb's argument class by a calculable amount; the individual formatives are treated in the chapters named above.

60.9 Ijäs's L2–L3 framework

Ijäs's Basic Lessons 2 and 3 supply a pedagogical framing of the same system. Her zero-place, one-place, two-place, and three-place classes correspond to Nakagawa's N-place inventory, Satō's argument-slot count, and the Handbook valency chapter Ijäs (2023); Nakagawa (2024: 125–139); Bugaeva & Kobayashi (2022: §§2–3). Her "pockets" image — person markers fill positions on the verb itself before and more obligatorily than free noun phrases — is a learner-facing rendering of head-marking, and is especially close to Satō's argument-slot language Satō (2023: 42–51). Her statements that bare verb forms carry no tense and that nouns are number-neutral agree with all primary sources Nakagawa (2024: 124–125); Shibatani (1990: 33).

Her Lesson 2 zero-place weather verbs match Nakagawa's zero-place category and correctly include the inchoative reading of descriptive predicates such as hure ‘be red; become red’. One qualification on her treatment: she presents apto as alongside the bound-root zero-place forms, but apto ‘rain’ is a free noun and Nakagawa leaves the word-boundary status of the expression unresolved Nakagawa (2024: 129–130) ‹contested›.

Two Ijäs simplifications require a caveat from the primary sources. Third person is segmentally zero in all attested forms, and her explicit Ø- notation is descriptively harmless; the theoretical status of that zero is nonetheless disputed, with Nakagawa treating it as the absence of an affix and Satō positing a zero third-person affix to preserve the two-index cap Nakagawa (2024: 137); Satō (2023: 39); Satō (2025: 23–24) ‹contested›. Her presentation of the indirect object of a three-place verb in a fixed position between subject and direct object reflects a teaching default; Nakagawa's examples with epakasnu ‘teach’ show that the two objects and the subject are freely ordered by discourse conditions Nakagawa (2024: 363–365) ‹consensus›.

Her forward reference to verbs with distinct singular and plural stems belongs with Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa); her adverbial -no examples with Chapter 61 (The Adverbializer -no and Adverb Formation from Verbs). Neither stem alternation nor the adverbializer alters the argument structure of the base.