Skip to content

Chapter 73Valency, Voice, and Noun Incorporation: Overview

The architecture of the valency-changing system: the inventory of operations (causatives, the e-/ko-/o- applicatives, antipassive, reflexive, reciprocal, middle/anticausative, and noun incorporation) and how they compose, stack, and order on the verb.

73.1 The valency system as an arity calculus

Hokkaido Ainu reworks the argument frame of the verb through a dense layer of derivational morphology. Nakagawa presents the system as an arithmetic over the verb word: every derivational element carries an arity value, the derived verb’s argument capacity is the sum of those values, and the person indexes fill whatever argument positions remain Nakagawa (2024: 136–140, 184–212). The Handbook covers the same ground as verbal valency and voice — the applicatives, the causatives, the reciprocal, the reflexive, the antipassive, the middle, and a set of valency-neutral alternations Bugaeva & Kobayashi (2022: §1–5). Noun incorporation belongs to this architecture as well, since an incorporated nominal satisfies an argument slot inside the verb word and is diagnosed by the position of person marking Satō (2022: §1–2); Shibatani (1990: 60–66).

The diagnostic substrate is the transitivity system of Chapter 57 (Transitivity, Valence Classes, and Ambitransitivity) and the four-person paradigm of Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms). A derived intransitive takes the suffixal S indexes =as (first-plural exclusive) and =an (fourth); a derived transitive takes the prefixal A indexes ci= and a= Satō (2008: 96); Nakagawa (2024: 137, 140). Placing a verb in the first-person plural is therefore the one operative test of its derived transitivity. Satō’s pair pukusa a=ta ‘we dig pukusa ‘lily bulb’’ beside pukusa-ta=an makes the point: incorporating the object intransitivizes the verb, and the fourth-person S suffix =an becomes available Satō (2022: §2). A verb hosts at most two person affixes, one subject and one object, so every operation works under a two-slot ceiling Satō (2023: 39); Bugaeva & Kobayashi (2022: §2). Ijäs’s learner image of argument “pockets” on the verb beside external “slots” renders the same head-marking intuition (Nichols 1986); her terms are pedagogical and her example sentences constructed Ijäs (2023) ‹consensus›.

(1)
numan yesterday
hekaci boy
cep fish
koyki catch

‘The boy caught fish yesterday.’

constructed exampleIjäs 2023

Ijäs Lesson 3 introduces the two-argument transitive frame with constructed sentences of this type; the third-person subject and object are segmentally unmarked.

73.2 Inventory of operations

The working inventory for the part is given below. The values in the first column are arity effects in Nakagawa’s sense, and they predict the person-marking diagnostics used by Satō and the Handbook. Several surface strings recur across rows, so the arity and the host verb class settle how a given occurrence is glossed. Each operation is developed in its own chapter, listed in the final column. Bugaeva (2012: §15.4.4.1)

directionoperationmarker or materialtypical outputdetail
+1productive causative-re, -te, -ecauser added as subject; old subject becomes objectChapter 74 (The Morphological Causative -re / -e / -te)
+1applicativee-, ko-, o-a peripheral participant becomes a core objectChapter 78 (The Applicative System: Overview and the Preverb-vs-Applicative Debate)
+1transitivizer-kar on an intransitivean object is addedChapter 75 (The Transitivizer -ka and Lexical Valency Pairs)
+1 / 0direct causative, transitivity-pair formation-V, -ke, -kalexical or direct-causative alternationChapter 75 (The Transitivizer -ka and Lexical Valency Pairs)
0indefinite causative-yar, -arcauser added while the causee stays unexpressed and unindexedChapter 77 (The Permissive/Sociative Causative -yar)
0sociativeuko-, uwe-joint participation, no net arity changeChapter 86 (The Reciprocal u-)
0affiliative part prefixe- ‘head of’, o- ‘bottom of’fills a part slot but requires its own possessorChapter 78 (The Applicative System: Overview and the Preverb-vs-Applicative Debate)
−1antipassive / indefinite objecti-generic object absorbed; transitive becomes intransitiveChapter 83 (The Antipassive / Detransitive i-)
−1reflexiveyay-, si-object position filled by a self participantChapter 84 (The Reflexive yay-); Chapter 85 (The Reflexive-Possessive si- and the yay-/si- Division)
−1reciprocalu-object position filled by reciprocal participants; plural subjectChapter 86 (The Reciprocal u-)
−1middle / object-resultativeci-the affected participant becomes the sole argumentChapter 87 (Middle, Anticausative, and the (Non-)Passive Question)
−1lexical body-region prefixhe- ‘head’, ho- ‘bottom’body-region object supplied by the prefixChapter 87 (Middle, Anticausative, and the (Non-)Passive Question)
−1 (×N)noun incorporationbare noun stem, and in places adverbial materialan argument or adjunct joins the verb wordChapter 88 (Object (Patient) Noun Incorporation)

The inventory also marks the homophony hazards that run through the part. Applicative e- and o- add an object; affiliative e- and o- fill a part slot while keeping a possessor; second-person e= is a person index. The prefix ci- may be the middle, while ci= is the first-plural-exclusive A index. The suffix -kar transitivizes on an intransitive base, while a homophonous -kar on transitives is a verse-filler with no arity effect Nakagawa (2024: 190–193, 209–210). None of these can be glossed without first reading off the arity.

73.3 Valency-increasing morphology

The productive causative is one suffix with three principal allomorphs: -re after a vowel or glide, -te after most consonants, and -e after r Nakagawa (2024: 200–201); Bugaeva (2014: 134). It raises arity by one. On an intransitive base the original S becomes the object of the causative verb; on a transitive base the result is a double-object verb, and double causatives are admitted as well, both of which are taken up in Chapter 76 (Causee Marking, Ditransitive and Double Causatives). The suffix covers coercive and permissive causation alike, and Bugaeva’s Saru sample shows it as the productive default, taking in many direct-causative uses that older descriptions assigned to non-productive transitivizing suffixes Bugaeva (2014: 130, 139–144).

(2)
matkaci girl
sinot play

‘The girl played.’

Nakagawa 2024: 431; Saru, Nakagawa ex. 954, after Tamura 1984

(3)
hapo mother
matkaci girl
sinot-te play-CAUS

‘Mother made the girl play.’

Nakagawa 2024: 431; Saru, Nakagawa ex. 954′

The base subject matkaci ‘girl’ is the object of the derived causative.

The older, less productive direct-causative suffixes -V, -ke, and -ka build many valency pairs and are split out into Chapter 75 (The Transitivizer -ka and Lexical Valency Pairs). Bugaeva analyzes most of this morphology as direct causative in function, where Nakagawa keeps the forming suffixes, the productive causative, and the special motion-verb causatives distinct by their distribution Nakagawa (2024: 198–203); Bugaeva (2014: 129–144) ‹contested›. The indefinite causative -yar, which adds a causer while leaving the causee unexpressed and unindexed, is valency-neutral and is treated separately in Chapter 77 (The Permissive/Sociative Causative -yar). Ijäs’s Lesson 9 statement that -re “turns intransitive verbs into transitive verbs” gives the right entry intuition but states no allomorphy and conflates causation with transitivization Ijäs (2023); Satō (2008: 234–238) ‹consensus›.

The applicatives e-, ko-, and o- add a non-agent participant to the object set, turning an intransitive into a transitive and a transitive into a double-object verb Bugaeva (2006: 188); Nakagawa (2024: 194). Their applied objects range well beyond a narrow instrumental or benefactive: Bugaeva’s role tables cover content, theme-patient, instrument, location, range, addressee, goal, comitative, benefactive, malefactive, recipient, source, and purpose Bugaeva (2006: 188–196). The three prefixes, the diagnostics, and the older 充当相 and case-relation-prefix labels they displace are set out in Chapter 78 (The Applicative System: Overview and the Preverb-vs-Applicative Debate), with the individual prefixes in Chapter 79 (The e- Applicative), Chapter 80 (The ko- Applicative), and Chapter 81 (The o- Applicative).

(4)
sirkorkamuy standing.tree
emus sword
e-tuye APPL-cut

‘He cut a standing tree with a sword.’

Nakagawa 2024: 193; Saru, Nakagawa ex. 173, after Monbetsu-chō 1969: 237

The instrument emus ‘sword’ is the applied object licensed by e-, which raises tuye ‘cut’ from a two-place to a three-place verb.

73.4 Valency-decreasing and rearranging voices

The prefix i- fills the object position of a transitive verb with a generic ‘thing’ or ‘someone’ and intransitivizes it, so that huraye ‘wash something’ yields i-huraye ‘do the washing’ and ku ‘drink something’ yields i-ku ‘drink alcohol’ Nakagawa (2024: 190–192). Its label divides the literature. Bugaeva analyzes it as an antipassive on syntactic, semantic, and discourse grounds, and Nakagawa adopts the term after her; Ijäs follows the same usage in Lesson 9. Satō’s textbook treats the identical prefix as a derivational i- ‘people, someone’ and uses no antipassive label Bugaeva (2021); Bugaeva & Kobayashi (2022: §4.2.3); Ijäs (2023); Satō (2008: 232–233) ‹contested›. Its origin is a separate dispute: the older line relates it to the fourth-person object index i=; Bugaeva derives the antipassive from incorporation of a generic noun and then the pronoun from the antipassive; Nam reconstructs the source as *hi or *ihi ‘thing, place, time’, the generic noun he also ties to the nominalizer and the possessed suffix Nam (2021: 63–74) ‹contested›. Ijäs’s further suggestion that isam ‘not exist’ contains this i- over a lost *sam is idiosyncratic and unverified.

(5)
i-ri=an ANTIP-skin=4.S
wa CONJ
or-o place-POSS
wa from
sini=an rest=4.S
na FIN

‘I will skin the catch and then rest.’

Bugaeva & Kobayashi 2022: §4.2.3; Shizunai, Shizunai Board of Education 1991–1995, O0130054_6; Handbook ex. 24b

The intransitive S suffix =an shows that i-ri ‘do skinning’ is intransitive; the patient ‘catch’ is left out of the argument frame.

The reflexives yay- and si- both mean ‘self’ and both reduce arity by filling the object slot Nakagawa (2024: 188–189). Their division of labour is among the most disputed points in Ainu derivation. Earlier accounts split them by intentionality or by an agentive against a patientive self; Satō’s Chitose analysis, adopted by Nakagawa and the Handbook, separates direct self-involvement, marked by yay-, from indirect involvement or dependence on another participant, marked by si-, so that yay-rayke ‘commit suicide’ stands against si-rayke ‘be ruined by ill fortune’ Satō (2007: 31–39); Nakagawa (2024: 188–189); Bugaeva & Kobayashi (2022: §4.2.2) ‹contested›. The two prefixes are treated in Chapter 84 (The Reflexive yay-), and the possessive use of the same morphology, where si- supplies a subject-coreferent possessor to an incorporated noun, in Chapter 85 (The Reflexive-Possessive si- and the yay-/si- Division).

Reciprocal u- fills the object slot with reciprocal participants and yields an intransitive verb with a plural subject. Its readings reach past ‘each other’: u-tures-kor means ‘have a sister, be siblings’ and u-mina-re ‘laugh together’, with no causative-reciprocal sense Nakagawa (2024: 187–188); Bugaeva & Kobayashi (2022: §4.2.1). Ijäs’s Lesson 9 gloss ‘mutually, each other’ captures the core reciprocal Ijäs (2023) ‹consensus›. The valency-neutral sociative uko- / uwe-, built from u- plus an applicative, is taken up with the reciprocal in Chapter 86 (The Reciprocal u-).

(6)
nea that
nispa rich.man
utar PL
u-koyki RECP-attack
kor while
oka exist.PL
siri appearance
a=nukar 4.A=see

‘I saw those rich men fighting each other.’

Bugaeva & Kobayashi 2022: §4.2.1; Saru, Nakagawa corpus K8010281UP.168; Handbook ex. 11a

The middle ci- presents an event from its affected side and yields an intransitive: cari ‘scatter something’ gives ci-cari ‘be scattered’, with the affected noun as subject, against a=cari, which keeps that noun as the object of a transitive clause with an indefinite agent Nakagawa (2024: 190–191, 429–430); the Handbook glosses this ci- object-resultative Bugaeva & Kobayashi (2022: §4.2.4). It is normally restricted to third-person participants, which keeps it apart from the first-plural-exclusive index ci=.

(7)
aynu human
kam flesh
ka even
upas snow
ka even
ta LOC
ci-cari RES-scatter

‘Human flesh and snow too lay scattered on the ground.’

Nakagawa 2024: 189; Chitose, Nakagawa ex. 167

Hokkaido Ainu has no subject-promoting passive. The work of a passive is carried by the fourth, indefinite transitive subject a=, dialectally an=: the agent is defocused while the patient stays the object of a transitive verb, and a demoted agent may appear in an or wa ‘from’ phrase Nakagawa (2024: 428–429); Bugaeva & Kobayashi (2022: §1). Satō’s subjecthood test confirms the structure: a ‘want to be loved’ clause cannot treat the patient as an intransitive subject and instead requires a nominalized, object-marked frame Satō (2008: 205–206) ‹consensus›. The middle, the anticausative patterns that overlap it, and the boundary with this indefinite-agent construction are pursued in Chapter 87 (Middle, Anticausative, and the (Non-)Passive Question).

(8)
seta dog
or place
wa from
aynu person
a=kupapa 4.A=bite

‘A person was bitten by a dog.’

Nakagawa 2024: 428; Saru, Nakagawa ex. 945″

The dog appears in an or wa ‘from’ phrase; the indefinite transitive subject a= stays on the verb and the patient aynu ‘person’ stays the object.

73.5 Noun incorporation and wordhood

Noun incorporation forms one verb word from a nominal stem and a verb Nakagawa (2024: 480); Satō (2022: §1). The wordhood diagnostic is again the outer position of the person index. With a phrasal object the index sits on the verb, cep ku=koyki ‘I catch fish’; under incorporation it precedes the whole noun-verb complex, ku=cep-koyki ‘I fish-catch’; and with a fourth-person subject the test is decisive, because the intransitive suffix =an cannot attach to a transitive verb Nakagawa (2024: 482–483); Satō (2022: §2).

(9)
pet river
or place
ta LOC
ran=an descend.SG=4.S
wa CONJ
suy again
cep-koyki=an fish-catch=4.S
rok PL.PRF
ayne then

‘I went down to the river and caught fish again, and at last …’

ILCAA Ainu materials 1976; Saru, Kawakami Matsuko, uwepeker 13; ILCAA materials aa-irc/016#14

The intransitive suffix =an on cep-koyki proves that the incorporated cep ‘fish’ has detransitivized koyki ‘catch’; the whole noun-verb sequence is one word.

Object incorporation is the unmarked, lexical type and the overwhelming majority in Satō’s Chitose count; it answers to Mithun’s Type I compounding (Mithun 1984); Satō (2016: 83–84); Satō (2022: §3). Subject incorporation is marked and divides into natural-force and possessor-requiring subtypes, the latter stranding a possessor that the subject recovers — the material of Chapter 89 (Subject Incorporation and Possessor-Stranding). Oblique nouns do not incorporate directly: an instrument or a location first becomes an object through e-, ko-, or o-, and only the derived object incorporates Shibatani (1990: 62, 66–67); Nakagawa (2024: 484), the feeding relation pursued in Chapter 82 (Applicative Stacking, Ordering, and the Feeding of Relativization).

The status of incorporation is contested. Shibatani argues that a purely grammatical-relation account and a purely semantic-role account each fall short, and that Ainu incorporation must reference both the incorporated noun’s low agentivity and its grammatical position Shibatani (1988: 210–211); Shibatani (1990: 60–66). Satō revises the unaccusative account because transitive natural-force subjects also incorporate, and takes low agentivity as the defining property, ranking the four types by a markedness calculus Satō (2008: 221–227); Satō (2022: §5–6). Bugaeva places Ainu among the polysynthetic languages in Baker’s sense while recording where it departs from the Mohawk prototype — its word order, its in-situ wh, and its unrestricted morphological causatives (Baker 1988); Bugaeva (2014) ‹contested›. These questions are developed in Chapter 88 (Object (Patient) Noun Incorporation) and Chapter 90 (Oblique Incorporation and the Syntactic-vs-Lexicalist Debate).

73.6 Stacking, ordering, and the verb-word template

The valency prefixes sit inside the person-index layer and before the root, while the causative suffixes follow the stem and its forming material. Nakagawa’s position table gives the order as person index, nominal or applicative prefix, root, forming and plural suffix, causative suffix, and suffixal person index Nakagawa (2024: 210–212), the template developed in Chapter 56 (The Verb Word and Its Position-Class Template) and Chapter 69 (The Personal-Affix Template: Position Classes and Affix Ordering). Because the person index is the outermost shell, any element standing outside it begins a new word, which is the same fact that underlies the incorporation diagnostic above.

Stacking is extensive and bounded by the template. A single verb may carry up to two applicatives, two causatives, two incorporated objects, and a lexical prefix, alongside a reciprocal, reflexive, or antipassive and the agreement indexes Bugaeva (2014). The classic example loads an incorporated object (ram ‘heart’), an adverbial element (tuyma ‘far’), one reflexive proper, one reflexive-possessive prefix, and two applicatives onto one transitive root:

(10)
usa-oruspe various-rumor
a=e-yay-ko-tuyma-si-ram-suy-pa 4.A=APPL-REFL-APPL-far-REFL.POSS-heart-sway-PL

‘We keep swaying our hearts afar over various rumours.’ (lit. ‘we sway our hearts about ourselves far over various rumours’)

Shibatani 1990: 73Bugaeva 2014; Hokkaido (dialect not further specified), after Chiri 1936: 169; Shibatani ex. 131 / Bugaeva ex. 1

One transitive root suy-pa ‘sway (PL)’ carries an incorporated object, an adverbial, an outer reflexive proper, an inner reflexive-possessive si- before ram ‘heart’, and two applicatives; the whole reads ‘we wonder about various rumours’.

The Handbook’s yay-ipe-e-ko-sunke ‘eat meagre meals’ shows the converse, a three-object applicative frame re-intransitivized by incorporating one object and reflexivizing another Bugaeva & Kobayashi (2022: §4.1.1). Valency morphology also interacts with verbal number, since what a plural stem or the plural -pa tracks depends on transitivity: an intransitive plural tracks subject number, a transitive plural tracks object or event number Ijäs (2023); Nakagawa (2024: 142–159) ‹consensus›. That system belongs to Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa) and Chapter 70 (Person–Number Interactions and Plural Agreement), and a valency operation can shift which participant a plural form tracks.

References cited in this chapter

Baker (1988) ·Bugaeva (2006) ·Bugaeva (2012) ·Bugaeva (2014) ·Bugaeva (2021) ·Bugaeva & Kobayashi (2022) ·Ijäs (2023) ·ILCAA Ainu materials (1976) ·Mithun (1984) ·Nakagawa (2024) ·Nam (2021) ·Nichols (1986) ·Satō (2007) ·Satō (2008) ·Satō (2016) ·Satō (2022) ·Satō (2023) ·Shibatani (1988) ·Shibatani (1990)