Chapter 90Incorporation, Quasi-Incorporation, and Word Structure

90.1 Incorporation, lexical units, and complex verbs

Ainu has several ways of combining a noun with verbal material. Morphological noun incorporation places a nominal stem inside a verb. Quasi-incorporation, in Satō's account, forms a conventional lexical unit while retaining the structure of a phrase. Polysynthesis is the broader capacity to express substantial clause-level content in a complex word Satō (2022: §§1–2, 4, 6); Bugaeva (2014).

The distinctions matter for argument structure. Simple object incorporation often produces an intransitive, but incorporation can also leave a transitive frame. Related noun–verb expressions may hemselves have one-word and two-word realizations Shibatani (1990: 63–64, 68–69); Nakagawa (2024: 495–498). The basic object and subject patterns are treated in Chapter 88 (Object Noun Incorporation) and Chapter 89 (Subject Incorporation and External Possessors).

90.2 Incorporation after an applicative

An applicative can make an otherwise oblique participant an object, which can then incorporate. Shibatani illustrates this with the comitative expression “with its bones.” In his analysis, the applicative ko- introduces the bones as an object; incorporating that noun leaves the fish as an external object Shibatani (1990: 67–69).

His three-stage comparison begins with nea cep pone tura a-kuykuy, then gives nea cep pone a-ko-kuykuy, followed by the incorporated form above. Shibatani explicitly identifies the middle stage as an artificially constructed illustration following Chiri Shibatani (1990: 68, example 122). It displays the proposed derivation rather than three successive utterances from a narrative.

Instrumental incorporation follows a comparable pattern in a literary example from Kutune Shirka / Itadorimaru:

A semantic location need not begin as an oblique. Satō's kane k-φ-upsor-omare, “I put money into my bosom,” incorporates a location already selected by omare. No additional applicative occurs Satō (2023: 47, §4.2, example 13). The relevant distinction is between an available argument and an otherwise peripheral participant.

90.3 Affix order and multiple incorporated elements

When an applicative and incorporation concern the same noun, many forms place the noun outside the applicative, as in sapa-e-puni. When they concern different nouns, the applicative can occur outside an incorporated patient. Shibatani gives a-ko-tam-etaye, with incorporated “sword” and a separate location related through ko- Shibatani (1990: 68–70, examples 121, 125–127).

He uses these orders as evidence for successive derivations, but also reports Chiri's counterexamples, including ni o-pici beside o-ni-pici “fall from a tree.” He considers both an ordering irregularity and a different identity for the prefix. He found no corresponding examples in his Itadorimaru material Shibatani (1990: 70–71, example 128). The order of noun and prefix therefore needs construction-specific evidence.

Descriptions also differ over the number of nouns a verb may incorporate. Shibatani states a one-noun limit while allowing a noun and an adverbial element together. Nakagawa explicitly records rare two-object formations, including cep-ya-o-kuta, “dump fish onto shore,” and itak-mak-kus-te, “use hidden words,” literally sending words behind Shibatani (1990: 72); Nakagawa (2024: 488–489, §19.2.3) ‹contested›.

Modifier incorporation has further lexical restrictions. Nakagawa compares pirka-no a=ipere with a=pirka-ipere for feeding well or sufficiently Nakagawa (2024: 484). The status of the first element and the distribution of -no are treated in Chapter 91 (Manner Compounds and Adverbial Incorporation).

90.4 Tamura's 連他動詞 and Satō's quasi-incorporation

Tamura describes 連他動詞 as adjacent words that function together as a verbal unit. Many combine a locative or body-part noun with a transitive verb. The noun carries O-series person marking and the verb carries A-series marking. Her dictionary independently classifies par(o) oyki “feed/support” and aske uk “invite” this way Tamura (1997: 15, §C.4); Tamura (1996: entries par(o) oyki and aske uk).

Satō calls the phrasal pattern quasi-incorporation. He also connects it with the older “third class of verbs” and Nakagawa's 連動詞. The crucial distinction is between morphological words and lexical units: a conventional noun–verb expression can be two morphological words while functioning as one lexical word Satō (2016: 89–90); Satō (2021: 89–94).

Compare ordinary ku-paro “my mouth.” Satō explains the O-series marker and basic par by treating the mouth's possessor as the object of the lexical unit “support.” The structure retains a noun–verb boundary, with restrictions on what can intervene Satō (2016: 89–90); Satō (2021: 90–94).

His contrast concerns the kind of possessor marking. The rejected *a-i-par-oyki has inflectional i- referring to a literary first-person possessor; the accepted alternative is i-par a-oyki. Derivational i-par-oyki “support someone” and yay-par-oyki “support oneself” can form one morphological word. The two uses of i- must remain distinct Satō (2016: 89, examples 8–9); Satō (2022: §4, examples 25–29).

Basic noun form is not an exceptionless requirement across all descriptions. Tamura gives both e-tom k-osma “I bumped into you” and eci-tomó k-osma “I bumped into you (plural),” with long tomó in the second form Tamura (1997: 15). Satō's own lexical analysis includes a phrase with an external uncle-possessor and affiliative paro, meaning “I intend to support my uncle” Satō (2021: 91, example 23; 94).

90.5 Nakagawa's one-word and two-word realizations

Nakagawa's category 連動詞 includes both one-word and two-word realizations. His person-marked examples make the difference visible. He analyzes Ishikari e=kaopas=an and Shizunai an=e=kaopas, both with first-person narrative reference to helping the addressee, as single verb words. The person markers cannot be assigned to a separate noun and an independently inflected verb in those forms Nakagawa (2024: 495, examples 1133–1134).

His Saru e=ka a=opas pe ne na instead places person marking separately on the noun and verb. He also gives two-word forms with long kasi Nakagawa (2024: 496, examples 1136–1138). This account describes a wider morphological range than the core two-word quasi-incorporation pattern ‹contested›.

Nakagawa emphasizes that one-word members can remain transitive. He relates their behavior to the special status of the nominal element and its O-series marking Nakagawa (2024: 496–498). An expression's conventional meaning, its morphological boundaries, and its argument frame are separate parts of the comparison.

90.6 The extension to or plus a case particle

Satō's 2021 proposal extends quasi-incorporation beyond verbal phrases to or ta and related locative combinations. He compares en-or ta “at my place” with en-par oyki: both have O-series person marking on a basic noun form. He proposes that a two-word locative phrase can constitute one lexical unit Satō (2021: 91–94) ‹speculative›.

The analysis incorporates the usual contrast between cise or ta “at the house” and unarpe oro ta “at aunt's place,” while acknowledging animate uses of basic or and an inanimate phrase with oro. Satō also identifies an unresolved structural difference between the locative and verbal patterns Satō (2021: 85–96). The distribution and explanatory limits are developed in Chapter 52 (Relational Nouns, Spatial Forms, and Quasi-Incorporation).

90.7 Diagnostics of word boundaries

Person marking often distinguishes incorporation from an ordinary phrase: Nakagawa contrasts cep a=koyki with cepkoyki=an. Third-person forms without overt markers can leave the same distinction ambiguous Nakagawa (2024: 482–483).

Derivational morphology sometimes supplies further evidence, but it too requires analysis. Nakagawa contrasts a complex verb such as yay-kotan-esina with a phrase in which yaykotan is itself a noun Nakagawa (2024: 483). A prefix's presence alone cannot distinguish those structures.

Accent evidence also has to be tied to the formation described. Nakagawa leaves a general accentual diagnostic unresolved in this discussion, while Satō discusses accent changes in particular incorporating forms Nakagawa (2024: 482); Satō (2022: §2). Spacing in a transcription records an analysis; it cannot by itself decide every disputed boundary.

90.8 Competing accounts of incorporation

Shibatani argues that both semantic role and grammatical relation are needed to describe Ainu incorporation. His criticism of Mithun's functional account is that Ainu's applicative-derived objects require grammatical-relation information. He explicitly acknowledges that Mithun discusses incorporated instruments and locations across languages Shibatani (1988: 210–211); Shibatani (1990: 60–62, discussing Mithun 1984); Mithun (1984).

He also challenges predictions of Baker's syntactic incorporation account. In his presentation, the problem concerns abstract Case assignment to an applicative object with an unaccusative base. His Ainu examples show applicatives with predicates of existence or state Shibatani (1990: 66–67, discussing Baker 1988); Baker (1988). The criticism belongs to Shibatani's comparison of those theories; Ainu nouns do not overtly display the abstract Case at issue ‹contested›.

Later descriptions expand the empirical domain. Satō's natural-force transitive-subject incorporation goes beyond the intransitive-subject and transitive-object domain of Shibatani's earlier formulation Satō (2016: 83–88); Satō (2022: §§1, 3). Satō's later argument-slot account allows nominal stems, nominal derivational prefixes, and person markers to satisfy a verb's argument requirements without deriving every incorporated noun by movement from an external phrase Satō (2023: §§4.1–4.2) ‹contested›. His “slots” describe argument capacity, not successive positions in an affix template.

90.9 An accessibility proposal and its limits

Satō proposes a hierarchy from ordinary object incorporation through natural-force and possessor-related subject incorporation to natural-force transitive-subject incorporation. He relates the restrictions to backgrounding, reflexive interpretation, and ambiguity, with quasi-incorporation supplying an alternative in the inflectionally marked possessor-object case Satō (2016: 87–90); Satō (2022: §§4–6) ‹speculative›.

Satō reports the following distribution in his Chitose material, drawn from approximately 130 hours of recordings with Shirasawa Nabe Satō (2016: 83 n. 1; 88, table 2).

His proposed penalty scores are an explanation of that distribution. The 2022 presentation repeats these percentages but changes some of the numerical weights Satō (2016: 88, table 2); Satō (2022: §6, table 1). The hierarchy classifies grammatical relations and noun types; Mithun's four functional types concern different properties of incorporation Satō (2022: §1).

Specificity and lexical restrictions require further qualifications. Yoshida's small comparison of Chitose and Sakhalin texts examines contextually specific incorporated nouns and questions whether low agentivity alone is sufficient Yoshida (2013: 169–174). Nakagawa also records restricted complex nominal material in incorporating forms Nakagawa (2024: 487–488, 497). These are distinct issues from the person-marking restriction that motivates Satō's core quasi-incorporation comparison.

90.10 Polysynthesis as a broader description

Bugaeva describes Ainu as polysynthetic because a complex verb can express content that other languages distribute across a whole clause. Her literary example combines an incorporated noun, a verbal modifier, two reflexive elements, and two applicatives Bugaeva (2014: example 1, from Chiri 1974 [1936]: 169). An inventory of such resources does not imply that their individual maxima all occur in one verb.

Complex affixal words need not contain an incorporated noun. Nakagawa gives a=e=si-ko-ek-te, “I made you come to me,” as an example of verbal complexity Nakagawa (2024: 481). Bugaeva's broad characterization also leaves open differences from Baker's formal polysynthesis framework, which her abstract explicitly notes Bugaeva (2014); Baker (1996) ‹contested›.