Chapter 40Deverbal and Denominal Noun Derivation (Lexical Nominalizer Morphology)
The morphology of forming nouns with -p/-pe, -i/-hi, and -kur, treated here as lexical word-formation, with syntactic nominalization/relativization deferred to Part XVI.
40.1 Scope: lexical word-formation and the syntactic cut
Hokkaido Ainu has no separate class of deverbal suffixes standing apart from the rest of the morphology. The same formal elements that nominalise whole verbal phrases at the syntactic level also operate at the word-formation level to produce stable lexical nouns. Three morphemes dominate the lexical deverbal layer: the nominalizer -p/-pe, the possessive/affiliative suffix -i/-hi (with its vowel-harmonic variants -e/-we), and the lexical class-noun heads kur ‘person’ 'person' and usi ‘place, time’ 'place/time'. The diminutive -po and the full range of noun-to-noun compounding belong to Chapter 38 (Derivational Morphology on Nouns: Diminutive, Augmentative, and Size Modification) and Chapter 39 (Noun–Noun Compounding: Headedness, Linking, and Semantics); the reverse direction — noun-to-verb conversion — belongs to Chapter 60 (Denominal and Deverbal Verb Derivation).
The boundary between a lexical deverbal nominal and a syntactic nominalized phrase turns on the arity of the base. When all argument slots of a transitive verb are filled by bound person morphology, the resulting -p-final form has no open relativizable gap and is reanalysable as a dictionary-listed noun. When a gap remains — when one slot is left open for the matrix argument — the construction is a syntactic relative clause or complement, treated in Chapter 101 (The Nominalizers -p/-pe and -i/-hi (Participant, Event, Place, Fact)). Nakagawa formalizes this with an arity calculus: a clause whose arity sums to zero supplies a complement; one summing to +1 heads a relative clause Nakagawa (2024: 503–504). The formal-noun subclass that these heads belong to, and the full complementizer syntax, are developed there and in Chapter 102 (Lexical-Head and Formal-Noun Nominalization (kur, uske, ruwe/hawe/siri/humi)).
40.2 The -p/-pe nominalizer at word level
40.2.1 Allomorphy
The nominalizer appears as p after a vowel-final form and as pe after a consonant-final one Satō (2008: 177); Ijäs (2023). The conditioning is phonological and holds across Hokkaido dialects ‹consensus›.
| VP-final segment | allomorph | example |
|---|---|---|
| vowel | p | ipe yakne pirka p — 'if you ate it, good' |
| consonant | pe | rok=an pe ne na — 'one ought to sit quietly' |
40.2.2 Lexicalized agent/patient nominals
The most common word-formation output of -p/-pe at the lexical level is a patient or agent nominal produced when both argument slots of a transitive verb are filled by bound person morphology. aep ‘food’ 'food' is the canonical illustration: synchronically a common noun listed as a dictionary headword Tamura (1996: s.v. aep), its morphological shape is a=e-p — the indefinite fourth-person agent a=, the transitive verb e ‘eat (transitively)’, and the nominalizer — 'that which one eats'.
aep
‘food’ (lit. ‘that which one eats’)
constructed example
Patient-nominal type. The subject slot is filled by the indefinite fourth-person agent a= and the patient slot is closed by -p; the result is a stable common noun with no open argument gap. The morphemic analysis is the standard decomposition; the surface form is aep.
When a gap is not lexically filled and one slot remains open, the form heads a syntactic relative clause rather than a stable noun. Nakagawa's example makes the contrast:
‘those whom I raised’
Nakagawa 2024: 214; Saru
Nakagawa ex. 230. Here p heads a relative clause with an accessible object gap (the raised persons); this is the syntactic use. The lexicalized aep differs by having no gap and a conventionalized meaning. Both shapes are morphologically identical; the distinction is arity and degree of reanalysis.
The lexicalized forms are patient nominals when the subject slot is filled and the nominalizer closes the object position, and agent nominals when the object is lexically saturated and the subject supplied by a person prefix. The instrument or manner reading — 'the thing associated with this action' — arises when neither argument is left open but the predication describes an activity rather than a state; it is determined by the semantics of the verb, not by morphological marking ‹consensus›.
40.2.3 Arity-slot nominalizations: ci= and a= with -p
A set of lexicalized nominals was formed when both argument slots of a transitive base were filled — the subject slot by the first-person exclusive prefix ci=, the patient slot by -p — before the form stabilized as a dictionary entry. The word cep ‘fish’ is the most-discussed example: nukopoli (aynu-corpora Discord, 2023-12-06; aynu-corpora Discord (2023–2026)) proposes the analysis ci=e-p 'the thing that we (exclusive) eat', with phonological reduction cie-p → cep on a par with the attested contraction kuani → kani ‹contested›. No primary source states the etymology directly; Tamura's Saru dictionary lists cep without internal analysis Tamura (1996: s.v. cep). The same community discussion (nukopoli, 2023-12-08; aynu-corpora Discord (2023–2026)) identifies cikarkarpe 'manufactured thing, handcraft' as ci=kar-kar-pe — exclusive subject, reduplicated stem kar-kar 'make', nominalizer pe after the consonant-final stem — consistent with the allomorphy and the compositional meaning ‹contested›.
A denominal analysis for canup ‘teaching, moral precept’ 'teaching, lesson' rests on the same skeleton: nukopoli (aynu-corpora Discord, 2024-11-10; aynu-corpora Discord (2023–2026)) reads ca-nu-p, with ca 'mouth' as the base, an associative element -nu, and -p as nominalizer — 'the thing one gives with the mouth, moral precept'. The compound verb ecanupkar 'to teach' supports the noun status of canup; the internal segmentation remains unconfirmed in the accessible grammars ‹contested›.
Alongside the ci=-initial type, the fourth-person agent prefix a= fills the subject slot to produce indefinite-subject nominals: a=e-p → aep 'food'. The alternation between ci= (exclusive: we-not-you) and a= (indefinite: one in general) reflects the person system described in Chapter 33 (The Word-Class Inventory and the Noun/Verb Bipartition), not a difference in the nominalization itself.
40.2.4 Instrument nominalization: ani X p
A productive word-formation pattern builds instrument nouns via the postpositional adverb ani ‘with, by means of’ followed by a verbal stem and the nominalizer: ani [V] p/pe = 'the thing with which one V-s'. Nukopoli (aynu-corpora Discord, 2026-05-29; aynu-corpora Discord (2023–2026)) documents this as a regular instrument-nominalization frame: ani e=kanpinuye p 'pencil' (lit. 'the thing to write characters with'), ani humi a=aste p 'musical instrument' (lit. 'the thing to make sound with') ‹corpus-confirmed›. In each form the verb's argument slots are filled and p closes the zero-arity predication; the instrument reading is contributed by ani. The construction shows that the word-level nominalizer can co-occur with adverbial adjuncts — not only with person prefixes — to saturate the argument frame before nominalization.
40.2.5 The lexical–syntactic boundary
Two criteria distinguish a lexical nominal from a syntactic one. Argument opacity: a lexicalized form has all slots filled and admits no modification by a relativizing matrix argument; a syntactic form projects an accessible gap. Compositionality: the meaning of a syntactic nominalization follows from the predicate and the gap's role; a lexicalized item may carry conventionalized meaning that has drifted from any compositional reading — 'fish' is not synchronically recoverable as 'what we eat'.
The community observation (nukopoli, 2024-11-09; aynu-corpora Discord (2023–2026)) that -p(e) nominalizes without dropping an argument, unlike relativized clauses, aligns with Nakagawa's arity account Nakagawa (2024: 503–504): a=i=ye p 'the thing that one says' is well-formed because both the subject slot (a=) and the object slot (i=) are filled, whereas substituting a bare head noun is impossible (*a=i=ye kamuy) ‹consensus›. The full syntax of the construction — the headless-relative and complement continuum, TAM marking inside the nominalized clause, insubordination in Tokachi — belongs to Chapter 101 (The Nominalizers -p/-pe and -i/-hi (Participant, Event, Place, Fact)).
‘It is the thing to put on the fire!’
Nakagawa 2024: 505; Saru
Nakagawa ex. 1165. Both argument slots of e=o are filled via person morphology; p closes the zero-arity predication with a patient reading. The directive force is contextual, not contributed by p itself.
40.3 The affiliative suffix as nominalizer: -i/-hi and -e/-we
The suffix that forms the affiliative (possessed) form of common nouns — -i/-hi in general notation, with -e/-we as the shape after certain base types — also operates as a derivational morpheme when applied to sensory and perceptual noun bases, producing formal nouns capable of nominalizing a preceding clause. The resulting form is morphologically a possessed nominal: ru-we is 'the trace of —', its implicit possessor being the event described by the nominalized clause. The same possessed-noun morphology underlies all four members of the Hokkaido evidential set Nakagawa (2024: 258, 392); Tamura (1996: s.v. ru, sir, hum, haw).
40.3.1 Conditioning of the suffix allomorphs
The possessed-form suffix follows the same five-type conditioning that governs the general affiliative paradigm Nakagawa (2024: 223–227); Nakagawa (2022: 484). The short form and the long form carry no semantic difference; the choice is prosodic Nakagawa (2024: 219). For the four sense-noun bases that grammaticalize as evidential formal nouns, the suffix type follows directly from the base-final segment:
| base (concept form) | base shape | affix type (Nakagawa 表51) | formal noun | grammaticalized information source |
|---|---|---|---|---|
| ru 'track, path' | CV open | ③-like → +we(he) | ru-we | inferred / recognized fact |
| sir 'appearance, scene' | CVC (C ≠ y, w) | ③-1 → +i(hi) | sir-i | directly visible situation |
| hum 'sound, sensation' | CVC (C ≠ y, w) | ③-1 → +i(hi) | hum-i | non-visual sensation |
| haw 'voice' | CVCw → type ② | ② → +e(he) | haw-e | verbal report, hearsay |
Each short form (ruwe, siri, humi, hawe) is the form that heads the evidential construction; the corresponding long forms (ruwehe, sirihi, humihi, hawehe) exist as ordinary lexical possessed nouns but are blocked from the evidential slot — Tamura marks this explicitly for each item Tamura (1996: s.v. ruwe). The blocking of the long form from the evidential slot shows that the formal-noun use has been grammaticalized beyond the general affiliative paradigm ‹corpus-confirmed›.
Refsing, from Shizunai, reanalyses the same set as nominalizers rather than possessed nouns, treating them as one of four sentence-conversion types alongside conjunctionalization, quotationalization, and adnominalization Refsing (1986: 259–264). Nakagawa retains the formal-noun word-class label and treats the evidential role as a grammatical function (complementizer) that certain formal nouns fill when arity is zero Nakagawa (2024: 503–505). The two framings are compatible at the descriptive level, differing in whether the category label or the functional role is foregrounded ‹contested›. Full treatment of the evidential semantics and syntax belongs to Chapter 121 (ruwe ne — the Inferential / Visual-Trace Evidential) and the sibling evidential chapters; what matters here is that the deverbal process is the possessive-suffix morphology, and the evidential construction is the grammaticalization endpoint of a path that begins at lexical word-formation.
40.4 Lexical class-noun heads: kur and usi/uske
Two common nouns operate as productive denominative heads: kur ‘person’ 'person' and usi ‘place, time’ (also usike) / uske ‘place’. Neither is an agreement morpheme; Hokkaido Ainu has no numeral classifier system Nakagawa (2024: 461). They are ordinary common nouns that follow a modifier (NP or VP) and supply a semantic class specification for the resulting compound.
kur forms person-denoting nominals from place names via the allative particle un: Sar un kur 'a Saru person', Iyoci un kur 'a Yoichi person' Nakagawa (2024: 234). Corpus rank 98 (2,456 tokens in the Hokkaido texts) reflects how productively it combines. The word-class diagnostics that distinguish kur from the formal-noun subclass — it heads an independent NP, takes attributive modifiers, and takes possession marking — are set out in Chapter 35 (The Formal / Defective Noun Set (ruwe, hawe, siri, humi, hi, pe, kur)).
‘a Saru person’
Nakagawa 2024: 234; Saru
Nakagawa §5.2.6. Place name + allative particle un + class-noun head kur; the pattern is productive across place names.
usi(ke) and uske head locational nominals from a verbal phrase. Nakagawa lists them under the formal-noun inventory (2024: 257–258; Nakagawa (2024: 257–258)); their full syntax as heads of VP nominalization constructions belongs to Chapter 102 (Lexical-Head and Formal-Noun Nominalization (kur, uske, ruwe/hawe/siri/humi)). The example below shows usi nominalizing a VP to produce a location noun that serves as goal of a motion-verb sequence:
‘I went (and came back) to the face-shaving place.’
Nakagawa 2024: 258; Hokkaido (dialect not further specified), Kushiro
Nakagawa ex. 330, citing Yaegashi 2001: 13. The VP nanu memke 'shave the face' is nominalized by usi to produce 'the face-shaving place (barber's)', which then heads the goal argument of the motion-verb sequence ku=oman wa ek. The ruwe ne closes the matrix clause as inferential evidential.
40.5 Zero-conversion and the verb–noun interface
Many Hokkaido Ainu nouns have corresponding intransitive verbs with no overt marking of the class difference. Property words like poro ‘be big’ 'be big' serve as noun modifiers — poro cep 'a big fish' — without any derivational suffix; word-class membership is assigned by the syntactic slot, not the form Nakagawa (2024: 115–116); Nakagawa (2022: 473). This zero-conversion is the default mechanism in compound nouns where an intransitive verb appears as the first element: Nakagawa's taxonomy of compound types (2024: 267–272) — 1-arg-verb+N, N+1-arg-verb, and 1-arg-verb+1-arg-verb — all reduce to N+N because the verbal element is freely reanalyzed as nominal in the modifier position Nakagawa (2024: 267–272).
inonno-itak ‘prayer discourse, liturgical speech’ 'prayer discourse' (compound of inonno ‘pray, offer prayers’ + itak ‘word, speech’) illustrates this path: the intransitive verb inonno 'pray' modifies itak 'speech' as the first element of a V+N compound with no nominalizing suffix. What signals the nominal reanalysis is the compound's function as a common-noun argument and the person-affix position outside the whole compound, not between its elements. The formal diagnostics for whether a sequence is a compound or two syntactic words belong to Chapter 39 (Noun–Noun Compounding: Headedness, Linking, and Semantics).
Antitwilight (aynu-corpora Discord, 2026-06-02; aynu-corpora Discord (2023–2026)) observes that Ainu has very few pure-noun coordinate compounds or exocentric N+N combinations, because noun-formation is verb-centred — nearly every noun can also serve as an intransitive verb, so N+N and V+N compounds are morphologically indistinguishable at the surface ‹contested›. This sharpens Nakagawa's (2024: 116) principle that word-class membership is defined by syntactic role rather than form. The verbal-number suppletion patterns that some deverbal nominals feed (e.g. singular ek vs. plural arki 'come') belong to Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa).
40.6 Grammaticalization cline and the Sakhalin contrast
The morphemes treated here as word-formation devices extend along a grammaticalization cline into clause-level nominalization and then into evidential and sentence-final positions. Takahashi's study of the Horobetsu dialect shows that the formal-noun nominalizers — including p/pe and the possessive-suffix set — undergo a further stage in Horobetsu, functioning as adverbial clause markers and conjunction-like linkers in contexts that the better-documented southern dialects (Saru, Chitose) handle with other strategies Takahashi (2018). The lexicalized deverbal nominals treated here are the initial pole of the cline; the Horobetsu data mark the far end within Hokkaido ‹corpus-suggested›.
Bugaeva's comparison of Hokkaido and Sakhalin nominalization Bugaeva (2016) identifies Sakhalin Ainu as having innovated in the nominalization system relative to the common ancestor. One index of this difference is the treatment of the evidential formal nouns: in Sakhalin varieties the long possessed form (ruwehe, sirihi) is used in the evidential slot, whereas Hokkaido restricts that slot to the short forms and blocks the long form from evidential use (see the conditioning table above). Whether this blocking is a Hokkaido retention or an innovation is an open question; the Sakhalin data frame it as a candidate for Hokkaido-internal change ‹speculative›. Full comparative treatment is in Chapter 121 (ruwe ne — the Inferential / Visual-Trace Evidential) and the sibling evidential chapters.
References cited in this chapter
aynu-corpora Discord (2023–2026) ·Bugaeva (2016) ·Ijäs (2023) ·Nakagawa (2022) ·Nakagawa (2024) ·Refsing (1986) ·Satō (2008) ·Takahashi (2018) ·Tamura (1996)