Chapter 44Concept Form vs. Affiliative (Possessed) Form

44.1 Overview: two forms of the common noun

Every common noun in Hokkaido Ainu potentially has two citation shapes: a concept form (概念形) expressing the noun as a free-standing concept, and an affiliative form (所属形) expressing it as belonging to something or someone Nakagawa (2024: 217). The concept form is the stem; the affiliative is a derivative, produced by a suffix on that stem Satō (2008: 152). Earlier scholarship described the contrast under the labels 抽象形 (abstract form) and 具体形 (concrete form) Kindaichi & Chiri (1936); Nakagawa's own earlier work used 人称形 (personal form) for the possessed shape Nakagawa (2024: 217).

The alternation underlies Hokkaido Ainu's bifurcated possession system. Inalienable nouns — body parts, kinship terms, part–whole pairs, and related classes — take the affiliative form when possessed and combine directly with a possessive person-prefix. Alienable nouns remain in the concept form as the implicit object of the transitive verb kor ‘have’ Nakagawa (2024: 218); Satō (2008: 150–151). The alienable/inalienable split itself is set out in Chapter 43 (The Alienable/Inalienable Split and Possessive Classification); the morphophonology of the affiliative suffix — the echo-vowel mechanism and suffix classes — belongs to Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels); the full double-marking NP (person-prefix plus affiliative suffix) is treated in Chapter 46 (Adnominal Possession: Double Marking and the Appositive Construction); and kinship-term idiosyncrasies appear in Chapter 48 (Kinship and Honorific Possession).

44.2 Formation of the affiliative form

The affiliative suffix attaches to the noun stem. Its shape is conditioned by the final segment of the stem Nakagawa (2024: 226–228); Satō (2008: 152–153). Three types are recognised, each with a short and a long variant.

• Type ① — vowel-final stems. The short affiliative is identical to the concept form; the long affiliative adds -hV₁ where V₁ repeats the stem's final vowel: sapa ‘head’ → sapa / sapa-ha.
• Type ② — stems in -y or -w. The suffix is -e(-he): haw ‘voice’ → haw-e(-he); patoy ‘lip’ → patoy-e(-he).
• Type ③ — other consonant-final stems. The basic (Type ③-1) suffix is -i(-hi): sik ‘eye’ → sik-i(-hi). Lexically listed subsets take an echo vowel matching the stem's last vowel (Type ③-2): tek ‘hand’ → tek-e(-he); kisar ‘ear’ → kisar-a(-ha). A small group with final a takes -u(-hu) (Type ③-3): nan ‘face’ → nan-u(-hu). Types ③-2 and ③-3 must be memorised individually; Satō notes that the exceptions are numerous and must be learned one by one Satō (2008: 153).

Monosyllabic open stems ending in i or u behave as Type ② rather than ③, taking -e(-he): ru ‘track, path’ → ru-we(-he) Nakagawa (2024: 229–230). The affiliative ruwe of ru is the same form that serves as the evidential formal noun, and the possessed-noun origin underlies all four evidential nominalizers treated in Chapter 121 (ruwe ne — the Inferential / Visual-Trace Evidential) Nakagawa (2024: 258–259).

When a consonant-final noun appears as the second element of a compound, it reverts to the ③-1 pattern regardless of its isolated affiliative class: at ‘cord’ → at-u(-hu) (③-2) in isolation, but tar-at → tar-ac-i(-hi) 'carrying-rope cord' inside a compound Nakagawa (2024: 228). The distinction between an affiliative-suffixed noun in an NP and a phonologically similar compound link is discussed in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

44.3 Short and long affiliative forms

For all three types, the suffix has a short and a long variant. For consonant-final stems, the short form ends in the echo vowel alone (teke) and the long form adds a further syllable (tekehe). The two variants carry no semantic difference; Satō writes that "the difference in meaning seems absent — how they are distributed remains unclear" Satō (2008: 152–153); Nakagawa confirms that the split follows prosodic and rhythmic preference rather than grammar Nakagawa (2024: 219). The long form is more frequent in oral-narrative verse Nakagawa (2022: 484).

44.4 The possessive person-prefix system

When a common noun appears in the affiliative construction, it combines with a possessive person-prefix drawn from the transitive-subject (A) affix set Nakagawa (2024: 215–216); Satō (2008: 154); Bugaeva (2021: 52). The possessor is thus grammatically encoded as the agent-like argument of the noun. Satō captures the counterintuitive direction with a literal gloss: e=paro 'your mouth' parses as 'you [A] mouth', the possessor marked with the same form that marks a transitive subject on a verb Satō (2008: 150).

Before vowel-initial stems, ku= reduces to k= and ci= to c=; accent falls on the syllable following the prefix for ku= and e= (e.g., ku-sápa) but not for a= or eci= Satō (2008: 154). The complete person-prefix system across all argument roles — intransitive-subject, transitive-subject, and object — is documented in Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms). Relational (locative) nouns such as or ‘place, interior’ and sam ‘vicinity’ take the objective (O) affix set rather than the A-set; their possessed forms are treated in Chapter 52 (Relational and Spatial (Locative) Nouns and Their Possessed Forms).

44.5 Semantic classes of affiliative nouns

Nakagawa identifies five groupings of nouns that enter the affiliative construction when possessed Nakagawa (2024: 219–221):

• Body parts and habitually worn items. The canonical inalienable class: sik ‘eye’, tek ‘hand’, kema ‘leg, foot’, tears, nasal mucus, and so on. Clothing and footwear normally worn on the body, and even commonly carried tools, are treated grammatically like body parts: ku=tesma-ha 'my snowshoes', a=tasiro-ho 'my mountain-knife'.
• Part–whole relations. The part belongs inherently to the whole: sayo rur-i(-hi) 'the broth of the gruel'; nit op nic-i(-hi) 'the shaft of a spear'.
• Material-product relations. The output belongs to its source: ikaray ohaw-e(-he) 'nettle stew'; top cise(-he) 'bamboo house'.
• Kinship terms — a large subset whose allocation between the affiliative and the kor ‘have’ construction varies by term and dialect; see Chapter 48 (Kinship and Honorific Possession).
• Miscellaneous inalienables whose belonging is inherent or relational: atay ‘price, value’, hura ‘smell, scent’, tokuy ‘close friend’, asur ‘rumour’, kotan ‘village’ (variable; see below).

The classification is grammatically fixed: the noun's class membership determines the construction, not a speaker's inference about the nature of the possession Bugaeva (2021: 47). A sword or dog is experientially "owned" but grammatically non-possessible and must take kor ‘have’. The contrast is sharpest in minimal pairs. Satō's diagnostic example — ku=kor kani 'my money' (alienable; *ku=kanihi is ungrammatical) versus ku=teke-he 'my hand' (inalienable) — shows that the split cannot be inferred from semantics alone Satō (2008: 150–151).

Two nouns sit on the boundary. kotan ‘village’ takes both constructions roughly equivalently: a=kor kotan ~ a=kotanu(hu). mosir ‘world, land’ is restricted to the concept construction in prose (a=kor mosir; *a=mosirihi unattested in prose, though mosiri appears in verse) Nakagawa (2024: 220–221). cise ‘house’ is dialect-split: in Saru, Chitose, and Horobetsu prose the affiliative cisehe is not used for 'X's house', but verse and some other dialects allow it, and the verse-compound uncise 'home' is always affiliative (a=uncise, never *kor uncise) Nakagawa (2024: 220–221). Cultural construal can also shift the construction for the same referent: a trap-bow (ku ‘bow (trap-bow)’) is usually alienable (ku=kor ku), but when the bow belongs existentially to whoever the prey chooses, the same object takes the affiliative kuwehe Nakagawa (2024: 229). The inalienable/alienable line is therefore construal-sensitive for certain nouns, not fixed by the referent alone ‹contested›.

Community member thegodofneet (aynu-corpora Discord, 2025-09-17) proposed that utari(-hi) is the affiliative of utar ‘family, companions, kin’, with the suffixed form narrowing the sense to 'X's companions/kin', on the grounds that companions belonging to no one are conceptually difficult to realise and the form consistently presupposes a possessor aynu-corpora Discord (2023–2026) ‹contested›. The corpus count for utari is 1,502 tokens. The analysis extends — rather than contradicts — Nakagawa's observation that k=utari 'my family' is attested alongside the concept form Nakagawa (2024: 251–252); the semantic narrowing under affiliation is the community's addition.

44.6 Obligatory possession and the non-possessed-noun form

Body parts and kinship terms are obligatorily possessed: they almost never occur without a possessive prefix and affiliative suffix in natural speech Bugaeva (2021: 52–54). When elicited in isolation — and in riddles, where no specific possessor is named — speakers supply the fourth-person (indefinite) prefix a=: a=sik-i-hi 'eyes (one's/ people's eyes)', not the bare stem sik Bugaeva (2021: 53). These nouns function as bound stems in the sense defined in Chapter 34 (Nominal Subclasses and Obligatorily-Possessed (Bound) Nouns).

When an obligatorily-possessed noun is incorporated into a verb and the possessor is non-coreferential with any core argument, the derivational prefix i- ‘antipassive / unspecified-possessor prefix’ encodes the unspecified possessor: i-sapa-kar usi 'a barber shop' (lit. 'a place where one does someone's head') Bugaeva (2021: 53–54). This is the non-possessed-noun (NPN) marker, distinct from the antipassive i- that demotes the transitive object; the two homophonous prefixes are treated together in Chapter 83 (The Antipassive / Detransitive i-). The fourth-person prefix a= (which retains affiliative morphology on the free noun) and the NPN i- (which enables possessor-less incorporation) are two distinct de-specifiers Bugaeva (2021: 54, 81).

44.7 The personal form and its syntax

The combination of person-prefix plus affiliative-marked noun constitutes what Satō calls the personal form (人称形): the form the noun takes when it is related to a particular possessor Satō (2008: 155). Because the possessor is indexed on the noun itself, the personal form functions as an autonomous NP. The third-person form siki(-hi) alone means 'his/her/their eye(s)' without any overt possessor pronoun — unlike Japanese 目, which requires 彼の to express the same content Satō (2008: 155).

A consequence of this indexical autonomy is that the possessor NP and the possessed noun form a looser constituent than a Japanese genitive phrase. Satō demonstrates that they can be separated by particles and intervening words, with the personal form presupposing only that some possessor NP is accessible in the discourse context Satō (2008: 155–156). Bugaeva's formulation is parallel: the possessive suffix "registers a possessor but does not index its properties", and "the possessor NP can easily be omitted without ambiguity as the possessed suffix signals its presence" Bugaeva (2021: 52). Stacking of personal forms — where the possessor of one is itself a personal form — is structurally transparent, as the third-person (zero) prefix on the inner noun is sufficient to license retrieval of the outer personal form as its possessor.

44.8 The affiliative constraint: specificity, grammar, and non-possessive uses

Tamura identified the trigger for the affiliative as the specificity of the possessor N₁: kamuy rusihi 'that specific bear's pelt' (affiliative) versus kamuy rus 'bear-pelt in general' (concept form, approaching a compound) Nakagawa (2024: 231). Satō challenged this with the counter-example aynu opitta honihi kor wa okay pe ne 'all humans have a belly' (honi-hi affiliative; possessor 'all humans' = non-specific) Satō (2008: 155–156) ‹contested›.

Nakagawa argues that the constraint is grammatical rather than semantic: the affiliative appears wherever a person-prefix or a relative-clause structure demands it, regardless of the possessor's specificity Nakagawa (2024: 227, 230–233). Three classes of evidence bear on this.

• General-warning contexts. a=teke munin pe ne na 'your hand will rot' (a warning about touching rainbows) uses the affiliative teke with the fourth-person prefix even though no specific person is addressed; the prefix alone forces the affiliative Nakagawa (2024: 231).
• Relative-clause structure. 'A dog with a long tail' requires sara-ha tanne seta (affiliative sara(ha)), not *sar tanne seta, even with an indefinite dog: the NP structure in which the dog possesses its tail forces the affiliative on sar Nakagawa (2024: 231–232).
• Quantifier and pro-form triggers. The quantifier usa 'various' and the pro-form nep 'what, whatever' independently induce the affiliative: usa kamihi usa cepihi 'meat and fish of various kinds' (Chitose); nep aynuhu e=ne 'whatever human you are' (aynuhu = affiliative of aynu) Nakagawa (2024: 232–233).

Nakagawa concludes that the received claim — that only some nouns possess an affiliative form — is overstated: given the right relational structure, potentially any common noun can take one Nakagawa (2024: 227, 230) ‹contested›. The phonological boundary between a suffixed affiliative and a bare compound second element is discussed in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels). The possessed form also feeds noun incorporation as discussed in Chapter 89 (Subject Incorporation and Possessor-Stranding).

Community member nukopoli (aynu-corpora Discord, 2025-12-03) parsed iteki e=hawe aste 'don't raise your voice' as [iteki [[e=hawe] aste]], with e=hawe (2SG possessive prefix + Type ② affiliative of haw 'voice') serving as a plain possessed-noun argument of aste 'raise' aynu-corpora Discord (2023–2026) ‹contested›. The analysis is consistent with the attested non-evidential, lexical use of hawehe in the example below.