Chapter 30Citation vs Combining Stem Shapes and Support Vowels

30.1 Overview

In Hokkaido Ainu the shape a morpheme takes when listed in a dictionary — its citation form — and the shape it takes when combined with an adjacent morpheme — its combining form — frequently differ. The alternations are phonologically conditioned: the coda of one morpheme interacts with the onset of the next to produce assimilation, deletion, or consonant mutation. Four processes govern the principal alternations treated here: (1) the behavior of coda /r/ at morpheme junctions and its vocalic release in citation position; (2) the categorical palatalization of coda /t/ before a following onset /i/; (3) vowel elision in the person prefixes ku= and ci= before vowel-initial stems; and (4) the historical compound assimilation visible in the elm-bark robe name attus (underlying at-rus). Shiraishi (2022: §4.4)

The sandhi rules that propagate across word boundaries are described in Chapter 28 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters); the echo-vowel diagnostics for distinguishing /r/-final from /rV/-final words are analyzed in Chapter 19 (The Rhotic /r/ and Coda-r); glide epenthesis at vowel hiatus is covered in Chapter 29 (Glide Epenthesis and Vowel-Hiatus Resolution); the accent consequences of combining prefixes are treated in Chapter 26 (Accent in Compounds and under Affixation and Cliticization). The affiliative suffix and the support vowels it introduces — the /i/, /e/, /we/ alternants in possessed forms such as hum-i, tek-e, ru-we — constitute a distinct citation/combining opposition routed to Chapter 44 (Concept Form vs. Affiliative (Possessed) Form) and Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

30.2 Coda /r/: citation release and combining forms

The rhotic /r/ is phonologically exceptional because its citation (pre-pausal) realization and its realization before a following morpheme diverge in two distinct ways. In citation position, coda /r/ releases into a vocalic echo that copies the preceding nucleus: kor [koɾ(o)], kur [kuɾ(u)], kar [kaɾ(a)]. The echo-vowel carries no accent, does not feed the nominalizer allomorphy, and disappears when /r/ resyllabifies before a following vowel-initial morpheme Nakagawa (2024: 30); Shiraishi (2022: §3.2). In combining position before a consonant-initial morpheme, /r/ instead triggers regressive assimilation (r-sandhi):

Sources: Nakagawa (2024: 35–37); Shiraishi (2022: §6.1); Kindaichi & Chiri (1936). The full assimilation paradigm and its domain conditions — including the asymmetry between n-sandhi (bounded by the phrase) and r-sandhi (crossing phrasal and sentence boundaries) — are treated in Chapter 28 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters).

30.3 The t + i → ci rule at morpheme junctions

The palatalization rule /t/ → /c/ [ʨ] before a following onset /i/ is the one sandhi alternation that applies categorically in all Hokkaido dialects without exception Nakagawa (2024: 36); Shiraishi (2022: §6.1); Kindaichi & Chiri (1936). Its formal statement:

t → c / _ i (at a morpheme junction, where /t/ is the coda of morpheme 1 and /i/ is the onset of morpheme 2)

Resyllabification is prerequisite: the morphemic sequence …V.ti… resyllabifies to give …V.C, placing the erstwhile coda /t/ in onset position before the /i/ syllable; the resulting /ti/ then palatalizes to /ci/ [ʨi] Kindaichi & Chiri (1936); Shiraishi (2022: §5.2). The rule applies at compound boundaries, derivational boundaries, and inflectional (possessive) boundaries alike.

Sources: Nakagawa (2024: 36); Shiraishi (2022: §6.1); Kindaichi & Chiri (1936).

The derivation of macikor from mat ‘woman’ + ikor ‘treasure, goods’ illustrates the two-step process:

1. Underlying: /mat-ikor/ — coda /t/ of mat at the left edge of the /i/-initial morpheme ikor
2. Resyllabification: /maˈtikor/ — coda /t/ moves to onset of the /i/ syllable, giving the sequence /ti/
3. Palatalization: /ti/ → /ci/ [ʨi] → macikor

The compound is well attested in Saru narrative speech:

The rule is subject to one phonotactic constraint: if the /i/-initial morpheme begins a closed syllable whose onset would create an impermissible three-consonant sequence, the resyllabification is blocked and palatalization does not apply. See Chapter 21 (Syllable Structure, Phonotactics, and Word-Edge Constraints) for the (C)V(C) canon and Chapter 29 (Glide Epenthesis and Vowel-Hiatus Resolution) for the parallel blocking of glide weakening before closed /i/-initial syllables. The possessive-boundary instances of the same rule are documented in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

30.4 Vowel elision in the combining forms of ku= and ci=

The first-person singular agent prefix ku= and the first-person plural agent prefix ci= each have a reduced combining form before vowel-initial stems. In the Saru, Chitose, and Mukawa dialects, the rule is:

/u/ → ∅ / ku= _ {a, e, u, o} (the prefix vowel elides before a stem-initial /a/, /e/, /u/, or /o/; parallel rule for ci=)

Before a stem-initial /i/, elision does not apply. Instead, the stem /i/ weakens to the glide [y], giving the combining form ku=ype (from underlying ku=ipe) rather than *k=ipe Nakagawa (2024: 52); Satō (2008: notation p. xii); aynu-corpora Discord (2023–2026) (nukopoli, 2024-11-28) ‹corpus-confirmed›.

Sources: Nakagawa (2024: 52–53); Satō (2008: notation); aynu-corpora Discord (2023–2026) (nukopoli, 2024-11-28).

The block on elision before /i/ is phonotactic. If the prefix /u/ were deleted before a closed syllable beginning in /i/ (e.g. ikra), the result *k=ikra would require a three-consonant onset, violating the (C)V(C) syllable canon. Nakagawa (2024: 52–53) confirms the ungrammatical forms *k=ipe and *ku=ykra and *k=ikra alike, with only ku=ikra grammatical before the closed /i/-initial stem.

The same elision process accounts historically for the pronoun kani ‘I (pronoun)’. Community analysis traces kani to an underlying ku=an-hi (first-person + existential root + nominal suffix), with the prefix vowel /u/ elided before the /a/-initial stem — parallel to the reduction ci=ep → cep 'fish' aynu-corpora Discord (2023–2026) (gengojiro, 2024-02-05) ‹speculative›. On this account the historical combining form and the present-day independent pronoun are related by the same morphophonological rule, now lexicalized in kani.

The elision rule is restricted to southern Hokkaido (Saru, Chitose, Mukawa); most other dialects retain ku= and ci= uncontracted before all vowels Nakagawa (2024: 52). For the personal-affix accent interactions triggered by prefix integration into the prosodic word, see Chapter 26 (Accent in Compounds and under Affixation and Cliticization) and Chapter 31 (Personal-Affix Junctural Sandhi, =an/a= Allomorphy, and Connected-Speech Reduction).

30.5 at-rus → attus: a compound assimilation

The elm-bark robe name attus (アットゥシ) derives from a compound at-rus, with at designating the elm-bark cloth. Two dialect reflexes are reported:

• General Hokkaido: attus — the onset /r/ of rus assimilates progressively to the preceding coda /t/ of at, producing a geminate /tt/ and collapsing the /t.r/ cluster: /at + rus/ → /at.tus/ → attus. The form attus is corpus-confirmed (24 tokens) ‹corpus-confirmed›.
• Saru dialect: ahrus — a different outcome where the coda /t/ of at weakens to /h/ rather than geminating, while the /r/ is preserved. This Saru form is reported by the community and has not been independently verified in a read primary source ‹speculative›.

aynu-corpora Discord (2023–2026) (nukopoli, 2024-12-26). The progressive assimilation /t + r/ → /tt/ producing attus is consistent with the cluster- simplification isoglosses documented in eastern dialects by Nakagawa & Fukazawa (2022: §3.3) (e.g. tupte → tutte in Bihoro; hotke → hokke 'lie down'). Nakagawa (2024: 36) treats such changes as historical word-form changes rather than synchronic alternations — a classification that applies equally to the lexicalized attus. Broader dialect consonant-assimilation correspondences are treated in Chapter 160 (Hokkaido Dialect Microvariation: Phonology and Morphosyntax).

30.6 koré: two analyses of a lexicalized junction

The verb koré ‘give (to someone)’ presents a case where the synchronic combining-form analysis and a competing historical account remain unsettled.

Nakagawa's morphological analysis Nakagawa (2024: 53): koré is the combining stem kor 'have' plus the causative suffix -e, parallel to kay-e 'break (tr.)', tuy-e 'sell'. The surface accent koré (σ2 high) rather than *kóre (σ1 high) follows from the iambic placement rule reapplying to the whole as if monomorphemic once lexicalization has obscured the morpheme boundary: /CV + CVC/ → σ2 accent. The same accent pattern holds for the parallel lexicalized causatives treated in Chapter 74 (The Morphological Causative -re / -e / -te) Shiraishi (2022: §5.2).

The community contraction account: koré is a contraction of korpare, the ditransitive root 'give to sb' (corpus frequency 153) aynu-corpora Discord (2023–2026) (nukopoli, 2024-12-26). Both koré and korpare are productive in attested narrative speech; the accent koré (not *kóre) is agreed under both accounts. The corpus attests koré (accent-marked) 30 times and kore (unaccented, orthographic variant) 3120 times; the disparity reflects editorial marking practice rather than speaker frequency ‹contested›.

The two analyses make different predictions about the combining system: Nakagawa's account treats kor-e as a still-productive, if now lexicalized, causative combination; the contraction account treats koré as an opaque form derived from korpare by phonological reduction, making the kor + suffix boundary morphologically inert in this lexical item. Neither is settled by the accessible sources.

30.7 The affiliative suffix as a support-vowel paradigm

A further class of citation/combining opposition involves the suffix vowel that consonant-final nouns receive when they take the possessed (affiliative) form: hum → hum-i 'his/her sound', haw → haw-e 'his/her voice', ru → ru-we 'the trace of —', tek → tek-e 'his/her hand'. The vowel that appears after the root consonant is conditioned by the root's final vowel, following co-occurrence restrictions that are not simple spreading Shiraishi (2022: §4.5); Nakagawa (2024: 226–230). This constitutes a systematic citation/combining opposition — the citation form is the bare noun, the combining (possessed) form adds the suffix vowel — and is treated in Chapter 44 (Concept Form vs. Affiliative (Possessed) Form) and Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

An additional morphologically restricted /h/-insertion between a vowel-final root and a following possessive vowel — ona-a → ona-h-a 'his father', siki-i → siki-h-i 'his eyes' (Kindaichi & Chiri 1936: 14, reported via Shiraishi (2022: n. 8)) — is also treated in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

30.8 Summary of combining-form alternations

The table below collects the principal citation/combining alternations documented in this chapter. For each rule, the underlying and surface forms are given; chapters treating the full paradigm are indicated. Nakagawa (2024: §1.2.1)