Chapter 31Reduplication Phonology and Rule Interaction

31.1 Overview

Hokkaido Ainu reduplication places a phonological copy of a verbal base alongside that base. The process is most visible in iterative and distributive-plural forms, but this chapter is restricted to its phonological dimension: the shape the reduplicant takes, what happens at the boundary between reduplicant and base, how the general resyllabification and hiatus-resolution rules interact with that boundary, and how the sandhi assimilations documented in Chapter 27 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters) bear on the juncture when a suffix follows the reduplicated complex. Aspect and iterative semantics belong to Chapter 111 (Phasal, Iterative, and Habitual Aspect); verbal-number pluractionality to Chapter 57 (Verbal Number: Suppletive Stems and the Pluractional -pa); nominalizer allomorphy consulted in §5 to Chapter 99 (The Nominalizers -p/-pe and -i/-hi (Participant, Event, Place, Fact)).

31.2 The phonological shape of the reduplicant

Hokkaido Ainu reduplication is characteristically full: the reduplicant copies the complete syllable or CVC template of the base Nakagawa (2024: 204–208). The syllable canon is (C)V(C), so each copy is an independently well-formed syllable (Shiraishi (2022: §4.1); Nakagawa (2024: 41–42)); consonant clusters at the juncture arise only when an assimilation rule applies (§§3–4 below), not from the copying itself.

The copying preserves coda consonants: a CVC base yields a CVC reduplicant; a CV base yields a CV reduplicant. The coda shape of the copy is identical to the coda shape of the source form (see §5 on how coda alternation in the source propagates into the reduplicant). The table below shows canonical mappings; phonetic representations are supplied in §3 where the boundary is discussed.

Accent follows the general iambic rule (Chapter 22 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split)): if the first syllable of the reduplicated complex is closed, it carries the primary accent; if open, the accent falls on the second syllable Nakagawa (2024: 50). In ketketcep the first syllable is the closed ket, giving [ˈketketcep]. Lexicalized compounds with reduplicated elements follow the first-element priority rule governing all compounds (Chapter 25 (Accent in Compounds and under Affixation and Cliticization); Nakagawa (2024: 54)).

31.3 The reduplicant-base boundary: resyllabification blocking and glottal insertion

A reduplicant and its base do not form a seamless concatenation. Across ordinary morpheme boundaries a coda consonant resyllabifies onto a following vowel-initial syllable: sik-o → [ˈsi.ko], mat-ak → [ˈma.tak] (Kindaichi & Chiri (1936: 9); Shiraishi (2022: §4.4)). Two blocking conditions suspend this rule and trigger glottal insertion instead:

1. Secondary accent on the following syllable. When the vowel of σ_n+1 carries secondary accent — as in paˈraparak-ʔan 'cry' — the accent domain prevents resyllabification; [ʔ] is inserted at the juncture.
2. The reduplicant-base boundary itself. ˈay-ʔay 'baby' shows glottal insertion at the REDUP-BASE juncture independently of whether the base vowel happens to be accented.

Both conditions are documented by Tamura (1998: 41–42), reported in Shiraishi (2022: §4.4).

Rule statement: ∅ → ʔ / V_RED-final __ V_BASE-initial at the REDUP-BASE juncture (or when the following V carries secondary accent at any juncture); resyllabification is otherwise suspended.

Derivation for ayˈay 'baby':

Productive glide insertion at person-prefix boundaries — e.g., i= + omante → iyomante — is a distinct operation in a different prosodic domain; it is morpho-syntactically conditioned and does not extend to the reduplicant-base boundary (Shiraishi (2022: §4.3); Chapter 28 (Glide Epenthesis and Vowel-Hiatus Resolution)). The glottal option takes over at REDUP-BASE.

31.4 Sandhi at the reduplicant-affix juncture

The blocking in §3 applies specifically at the internal REDUP-BASE boundary. The junction between the reduplicated complex and a following affix is an ordinary morpheme juncture: sandhi rules operate there as they do everywhere else.

31.5 Coda alternation in paired reduplicants: rikin and rikip

Because the reduplicant copies the phonological shape of the source form, any alternation visible in the base propagates into the reduplicant. A community observation on the verb rikin ‘go up, ascend’ illustrates this point.

The verb 'go up / ascend' appears in two forms: rikin, with final nasal coda /n/ (attested in corpus; see example below), and rikip, with final stop coda /p/. In reduplication, either form may serve as the base that is copied, yielding alternant shapes; the alternation is summarized as rikin~rikip (aynu-corpora Discord (2023–2026) nukopoli, 2026-04-27) ‹speculative›.

The causative suffix -ka ‘causative’ attaches to the nasal-coda form only: rikinka is attested; *rikipka is not found (aynu-corpora Discord (2023–2026) nukopoli, 2026-04-27) ‹speculative›. This asymmetry indicates that the nasal-coda form is the morphological base for further derivation, while the stop-coda form is a secondary shape.

The stop-coda form rikip is consistent with the nominalizer allomorphy: the nominalizer -p/-pe ‘nominalizer’ surfaces as -p after a vowel-final stem (Shiraishi (2022: §3); Nakagawa (2024: 29)). If the root is vowel-final riki-, then riki-p = root + nominalizer. The reduplicant then mirrors whichever form of the verb is being used as the base — nominalized or plain — without neutralizing or canonicalizing it. The coda alternation in the reduplicant is thus an instance of morphological conditioning of the base shape feeding into the copying operation. For the nominalizer system itself, see Chapter 99 (The Nominalizers -p/-pe and -i/-hi (Participant, Event, Place, Fact)) and Chapter 44 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

31.6 Person-prefix attachment to reduplicated stems

Personal prefixes ku=, e=, and ci= attach to the full reduplicated complex as if it were a single prosodic word. The iambic accent-placement rule — first closed syllable gets the nucleus; if the first is open, the second does — applies to the [prefix + REDUP + BASE] string as a unit (Nakagawa (2024: 50–54); Chapter 22 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split); Chapter 25 (Accent in Compounds and under Affixation and Cliticization)).

The vowel-deletion rule of ku= and ci= — prefix vowel deletes before stem-initial /a e u o/, but not before /i/ (Nakagawa (2024: 52); aynu-corpora Discord (2023–2026) nukopoli, 2024-11-28) ‹corpus-confirmed› — applies at the prefix-reduplicant junction exactly as it does before any vowel-initial stem. If the first syllable of the reduplicant is vowel-initial, the prefix vowel deletes; if consonant-initial, it does not. The full morphophonology is treated in Chapter 30 (Personal-Affix Junctural Sandhi, =an/a= Allomorphy, and Connected-Speech Reduction) and Chapter 62 (First- and Second-Person Singular Affixes (ku=/en=, e=/e=)).

31.7 Corpus illustration

No sandhi applies at the internal REDUP-BASE boundary in this example: the last syllable of the reduplicant (ri) is open, so there is no coda consonant to trigger the rules in §4. The following particle wa attaches at the right edge of the whole reduplicated form; the only potential sandhi there is the progressive nasal-place assimilation documented for this particle (Nakagawa (2024: 37)).

31.8 Dialectal variation

The regressive cluster assimilations of eastern Hokkaido dialects — tk → kk, pt → tt, rs → ss in Bihoro and related varieties (Nakagawa & Fukazawa (2022: §3.3)) — are treated by Nakagawa as word-specific historical consolidations rather than active synchronic rules (Nakagawa (2024: 36); aynu-corpora Discord (2023–2026) antitwilight, 2024-11-17). No specific reduplicated compound showing these eastern assimilations is attested in the sources consulted here ‹corpus-suggested›. The broader inventory of dialect-specific morphophonological isoglosses is treated in Chapter 158 (Hokkaido Dialect Microvariation: Phonology and Morphosyntax).