Chapter 39Noun–Noun Compounding: Headedness, Linking, and Semantics
Productive N-N compound formation, its headedness and morphophonological linking, and the semantic typology of compounds, drawn from the dedicated Ainu compound-noun literature.
39.1 Overview
Hokkaido Ainu forms compound nouns by combining two or more roots into a single phonological and morphological word. The process is productive across the lexicon and supplies much of the vocabulary for tools, animals, plants, social roles, and place names Nakagawa (2024: 261–272). Its dedicated structural study is Kirikae (1984), whose framework Nakagawa (2024: 261–272) reworks in terms of the valency of the constituent roots.
The base structural type is the concatenation of two nominal concept forms: a modifier root followed by a head root, the head carrying the compound's semantic class. Because intransitive verbs and nouns share the same form in Hokkaido Ainu — the same root functions as either class without morphological change (see Chapter 33 (The Word-Class Inventory and the Noun/Verb Bipartition)) — roots of any valency can serve as compound constituents. Nakagawa identifies six surface input patterns; all reduce to the same underlying N+N type once intransitive roots are treated as nominal Nakagawa (2024: 267).
The resulting compound is right-headed: the head is the rightmost constituent, which determines the grammatical class of the whole and hosts the affiliative (possessed) suffix and personal possessive affixes. The diminutive -po is explicitly excluded from compound formation — it is a derivational suffix on a finished noun, not a compound constituent (see Chapter 38 (Derivational Morphology on Nouns: Diminutive, Augmentative, and Size Modification)) Nakagawa (2024: 265).
39.2 Diagnostics: compound vs syntactic noun phrase
A two-root sequence might be either a compound noun or a syntactic noun phrase (attributive verb or modifier noun + head noun). Nakagawa (2024: 261–264) gives four diagnostics, all exploiting positional differences between one-word and two-word structures Nakagawa (2024: 261–264).
39.2.1 Person-affix position
A possessive personal affix attaches at the left edge of the word it marks. In a syntactic phrase it appears on the head noun alone; in a compound it appears outside the entire compound. The pair below has the same referent but different morphological structure.
‘his distinguished elder brother (modifier and noun are separate words)’
Nakagawa 2024: 261; Hokkaido (dialect not further specified)
Syntactic phrase: the possessive affix a= precedes yupi alone; poro is an independent attributive verb.
‘his/one's great elder brother (compound, lexicalized as a title)’
Nakagawa 2024: 261; Hokkaido (dialect not further specified)
Compound: the possessive affix a= precedes the entire compound poro-yupi, which behaves as a single morphological word.
In the compound both poro and yupi appear in their concept (bare) forms; the affix is placed outside both.
39.2.2 Affiliative-suffix position
The affiliative (possessed) suffix marks the noun it heads. In a syntactic phrase it attaches to the head noun, leaving the modifier unsuffixed; in a compound it attaches to the outermost right edge of the whole compound.
‘his wife's companions / his wives (two words: affiliative on the head mac)’
Nakagawa 2024: 262; Hokkaido (dialect not further specified)
Phrase: affiliative suffix -i attaches to mac (the concept-form stem of mat 'woman, wife'), and utar 'people/companions' is a separate syntactic head.
‘his wives (compound: affiliative on the outermost edge)’
Nakagawa 2024: 262; Hokkaido (dialect not further specified)
Compound: mat appears in its concept form inside the compound; the affiliative -i attaches to the right edge of the whole matutar unit.
39.2.3 Lexicalization and non-compositional meaning
When the meaning of the combination is not predictable from its parts, the sequence is necessarily a compound. retar ‘be white’ cikap ‘bird’ is a transparent syntactic phrase meaning 'a white bird'; retatcir ‘swan’ (where retat is a sandhi reduction of retar and cir does not occur as a free form) is a lexicalized compound meaning specifically 'swan' Nakagawa (2024: 263). The bound status of the second element and the non-compositional meaning together confirm the compound analysis.
39.2.4 The concept-form rule
The most practical criterion is the observation that two nominal concept forms placed in immediate sequence without intervening morphology always form a compound: the absence of any overt linker or predicative marking removes any phrase-level analysis Nakagawa (2024: 263–264). The series built on the head kap ‘skin, bark’ illustrates the pattern.
‘eyelid’
Nakagawa 2024: 263; Hokkaido (dialect not further specified)
sik 'eye' and kap 'skin, bark' are both in concept form; their immediate concatenation yields the compound 'eyelid'.
The same head kap combines with ni ‘tree, wood’ to give ni-kap 'bark', and with cep ‘fish’ to give cep-kap 'fish scales / fish skin' — a productive derivational series Nakagawa (2024: 263–264). Because property verbs such as retar ‘be white’ occur in concept form without any morphological marking, the only reliable way to distinguish a compound from an attributive phrase in ambiguous cases is the positional diagnostics above.
39.3 Headedness, morphological linking, and accent
Hokkaido Ainu compound nouns are head-final (right-headed): the right constituent is both the semantic and morphological head. This follows directly from the diagnostics: person affixes and affiliative suffixes mark the word-level node — the compound as a whole — not the internal head constituent alone Nakagawa (2024: 261–264).
At the constituent juncture the modifier typically appears in its concept (bare) form, with no linking morpheme; this is the default for the N+N, 1-arg-V+N, and N+1-arg-V types. In type ⑤ below the modifier appears in the affiliative (possessed) form instead. The morphophonological behavior of the affiliative suffix at such junctures is treated in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).
Prosodically the compound constitutes a single phonological word with a single accent peak. Community observation (nukopoli, aynu-corpora Discord, 2024-12-09) records that cisekorkur 'head of household' and kotankorkur 'village chief' each carry their accent on the second syllable of the compound as a single word, with no secondary peak on either internal constituent aynu-corpora Discord (2023–2026) ‹contested›. The general account of compound prosody is in Chapter 26 (Accent in Compounds and under Affixation and Cliticization).
39.4 Formation types
Nakagawa (2024: 267–272), building on Kirikae (1984), distinguishes six input patterns by the valency of the constituent roots. The classification is arity-based: because intransitive verbs freely convert to nouns in Hokkaido Ainu, types ②–④ surface as the same N+N morphological structure as type ① Nakagawa (2024: 267). Types ⑤ and ⑥ are morphologically distinct: ⑤ uses the affiliative form of the first element, and ⑥ places a transitive verb in the head position.
| Type | Input pattern | Example | Gloss | Meaning |
|---|---|---|---|---|
| ① | N + N | sik-kap | eye–skin | eyelid |
| ② | 1-arg-V + N | rep-ni | be.offshore–tree | driftwood |
| ③ | N + 1-arg-V | menoko-monrayke | woman–work | women's work |
| ④ | 1-arg-V + 1-arg-V | — | — | reduces to types ①–③ |
| ⑤ | N.poss + 1-arg-V | ami-tanne | claw.POSS–be.long | crab |
| ⑥ | N + 2-arg-V | cikap-koyki | bird–catch | hawk |
For types ②③ the intransitive root — including property verbs such as poro ‘be big’ and retar ‘be white’ — appears in what is synchronically a nominal concept form, since no verbal morphology is available inside the compound Nakagawa (2024: 267–268).
39.4.1 Type ⑤: affiliative modifier
In type ⑤ the modifier appears in the affiliative (possessed) form, giving the compound a relative-clause reading: 'the one whose N V-s'. The two examples below share the same head verb tanne ‘be long, be tall’.
‘crab (lit. 'the one whose claw is long')’
Nakagawa 2024: 269; Hokkaido (dialect not further specified)
ami is the affiliative form of am 'claw'; the compound names the crab by a characteristic body feature. Type ⑤: N.POSS + 1-arg-V.
‘thief (lit. 'the one whose hand reaches far')’
Nakagawa 2024: 269; Hokkaido (dialect not further specified)
teke is the affiliative form of tek 'hand'. The compound lexicalizes the thief as someone whose reach is long. Type ⑤: N.POSS + 1-arg-V.
The affiliative morphology of the first constituent makes the compound inherently relational: the head noun functions as an implicit argument of the modifier verb. This pattern overlaps with the external-possession construction treated in Chapter 47 (External Possession and Possessor Raising).
39.4.2 Type ⑥: incorporated transitive object
In type ⑥ the first element is a concept-form noun serving as the implicit object of a transitive verbal head. The compound names the entity characterized by performing that action on that object type.
‘hawk (lit. 'one that catches birds')’
Nakagawa 2024: 271; Hokkaido (dialect not further specified)
cikap 'bird' in concept form is the implicit object of the transitive verb koyki 'catch, seize'; the compound names the hawk by its characteristic predatory action. Type ⑥: N + 2-arg-V.
Type ⑥ is structurally close to noun incorporation (see Chapter 88 (Object (Patient) Noun Incorporation)), the distinction being that the result here is a lexicalized nominal, not a predicate frame with an internal incorporated argument Nakagawa (2024: 271).
39.5 Semantic patterns
The modifier-head relationship in Hokkaido Ainu compounds is consistently endocentric: the head names the class of the referent and the modifier restricts or characterizes it. Three sub-patterns recur.
39.5.1 Attributive-restrictive compounds
The largest group has a modifier specifying material, origin, location, or a characteristic property of the head referent. Place names illustrate the pattern transparently: Poropet ‘Horobetsu (place name)’ derives from poro ‘be big’+pet ‘river’ ('large river'), and Siretoko ‘Shiretoko (place name)’ from sir ‘appearance, land’+etoko ‘tip, front end’ ('land's tip, cape') Nakagawa (2024: 233). The same structure governs common-noun compounds: sisam-itak 'the Japanese language (as a named category)' is restrictive with sisam ‘Japanese person, non-Ainu’ as modifier ‹corpus-confirmed›.
39.5.2 Qualitative-possessive compounds: the N-kor-kur type
A productive sub-pattern combines a noun, the possession verb kor ‘have, own, hold’, and the formal-class noun kur ‘person’ to name a person who possesses or is in charge of the first noun. This N-kor-kur template is one of the few fully productive three-constituent patterns in the language.
The compound accent data (nukopoli, aynu-corpora Discord 2024, aynu-corpora Discord (2023–2026)) confirms that cise-kor-kur and kotan-kor-kur are single words with second-syllable accent ‹contested›. The same template extends to mosir-kor-kur 'ruler of the land' and similar formations. This construction must be distinguished from the syntactic alienable possession phrase a=kor cise 'my house', where kor is a fully inflected transitive verb taking its own person prefix, not a bound medial element Nakagawa (2024: 217–218). The possession constructions are treated in Chapter 46 (Adnominal Possession: Double Marking and the Appositive Construction).
39.5.3 Attributive NP vs compound: the scope distinction
Two surface-similar structures differ in their grammatical and semantic status. The syntactic NP kane cise 'a metal house' uses kane ‘metal, iron’ as a free-standing attributive noun; the first element cannot receive a personal affix for this relationship (*ku=kane cise as 'my metal's house' is not a grammatical possessive frame here), and the combination is compositional and productive with any house material Nakagawa (2024: 218). A genuine compound, in contrast, passes the person-affix and affiliative-suffix tests and typically names a culturally fixed category. The same applies to sequences with kamuy ‘spirit, deity; bear’: a contextual phrase kamuy cise 'a spirit's house (in this context)' is attributive; a lexicalized compound naming a specific object type would be non-compositional and semantically restricted.
39.5.4 Coordinate and exocentric compounds
Pure coordinate compounds (copulative; both elements of equal rank) and exocentric compounds (whose semantic head is not expressed) are marginal in Hokkaido Ainu. Community discussion (antitwilight, aynu-corpora Discord, 2026-06-02) proposes that this sparsity reflects the verb-centred character of Ainu nominal derivation: because most roots can serve as intransitive verbs, noun formation proceeds through verb+noun or noun+verb frames rather than noun+noun coordination aynu-corpora Discord (2023–2026) ‹contested›. The attested examples cited in the same discussion — haca-hapo ‘parents (lit. father and mother)’ as coordinate and po-at ‘womb (lit. child-place)’ as exocentric — function as lexical items rather than templates for new formation aynu-corpora Discord (2023–2026).
39.6 The compound–phrase boundary
Deciding whether a given modifier+noun sequence is a compound or a syntactic phrase has direct orthographic and grammatical consequences: compounds are typically written solid (no space), phrases with a space between elements. The diagnostics in the preceding section provide the formal criteria; their application requires attention to both morphological form and meaning.
A minimal orthographic contrast illustrates the semantic stake: sisam itak (two words) means 'speech/language of a Japanese person (in context)', while sisamitak (one compound word) means 'the Japanese language (as a named category)' (nukopoli, aynu-corpora Discord, 2024-11-05) aynu-corpora Discord (2023–2026) ‹corpus-suggested›. The compound reading adds a categorical, language-name interpretation absent from the phrase. Accent reinforces the distinction: the compound bears a single peak, the phrase retains peaks on both members (nukopoli, aynu-corpora Discord, 2024-12-20) aynu-corpora Discord (2023–2026). Full treatment of the prosodic test is in Chapter 26 (Accent in Compounds and under Affixation and Cliticization).
Ijäs (2023) presents the compound–phrase distinction as a key point for learners of Ainu, treating the presence or absence of internal syntactic modification (TAM morphology, person marking) on the first element as the diagnostic: a root bearing its own grammatical morphology is in a phrase; a root appearing directly before the head with no such morphology is a compound constituent. This matches the morphological diagnostics above and makes the distinction accessible without reference to accent.
The boundary also marks the limit of noun incorporation. The object incorporation construction shares the surface appearance of an N+V compound, but the incorporated noun remains an internal verbal argument rather than a constituent of a new nominal word. Community discussion (nukopoli, aynu-corpora Discord, 2023-12-25) notes that verbs taking an intransitive-type modifier present an ambiguity between incorporation and lexicalized compounding that cannot always be resolved by surface form alone aynu-corpora Discord (2023–2026) ‹contested›. Deverbal and denominal noun derivation via -p/-pe and related suffixes is treated in Chapter 40 (Deverbal and Denominal Noun Derivation (Lexical Nominalizer Morphology)).
References cited in this chapter
aynu-corpora Discord (2023–2026) ·Ijäs (2023) ·Kirikae (1984) ·Nakagawa (2024)