Chapter 25Lexical/Contrastive Accent and Minimal Pairs

25.1 Predictable and lexically specified accent

The placement rules of the accent dialects — if σ1 is closed (CVC), σ1 bears high pitch; if σ1 is open (CV), σ2 bears high pitch — assign accent to the large majority of the Hokkaido Ainu lexicon Nakagawa (2024: 50). Shiraishi terms this the iambic (LH) pattern: words with a heavy (closed) first syllable are trochaic in accent, and words with a light (open) first syllable are iambic Shiraishi (2022: §5.1). The full statement of these rules and their interaction with syllable weight is the subject of Chapter 23 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split).

The rules leave a residue of words with open σ1 that are accented on σ1 against the prediction. These carry lexically stored accent and must be individually memorized. Nakagawa (2024: 51) lists roughly twenty such items: cápe 'cat', húci 'grandmother', húra 'smell', húre 'red', káni 'I (emphatic)', kéra 'taste', míci 'father', mína 'laugh', nína 'gather firewood', nísap 'suddenly', núman 'yesterday', páse 'heavy', sápo 'elder sister', sések 'hot', síno 'really', sísam 'Japanese person', túki 'cup', túnas 'fast', úsey 'hot water', yúpo 'elder brother', and further items beyond the list given there. Every word in the accent dialects is either rule-assigned or lexically specified; there is no third status.

The person-marking prefixes ku=, e=, and ci= are incorporated into the prosodic word and can shift accent relative to its base; a= and eci= do not trigger a shift. Compounds suppress the accent of all but the initial element. These morphological and compositional effects on accent placement are treated in Chapter 26 (Accent in Compounds and under Affixation and Cliticization) and, for the person-affix interaction, in Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms).

25.2 Minimal and near-minimal pairs

The exception-list words create lexical contrasts wherever a segmentally identical (or near-identical) word follows the regular rule. Three such pairs are established in the descriptive literature Nakagawa (2024: 51); Shiraishi (2022: §5.1):

In each pair, form A carries lexical exception accent on σ1 and form B is rule-derived with σ2 accent. The critical phonetic property is the rising nucleus (上り核アクセント): what is distinctive is where the pitch ↗ rises, not where it subsequently falls. A word accented on σ1 begins with high pitch and may remain high throughout with no change of meaning, because the system encodes the onset of the high span. A word accented on σ2 requires a ↗ rise at σ2; pronouncing it with an σ1-initial high instead misidentifies it as the σ1-accented form Nakagawa (2024: 48–49). This stands in contrast with Standard Japanese, whose falling nucleus (下り核アクセント) marks the syllable after which pitch drops Shiraishi (2022: §5). A community observation from the Saru dialect captures the asymmetry: the pronouns ku↗ani and ku↗a↘ni are both acceptable, but ku↘ani (Japanese-style falling nucleus) is perceived as deviant — the rise position, not the fall, is what matters ‹corpus-confirmed› (aynu-corpora Discord (2023–2026); nukopoli, 2024-12-09).

The nísap : nisáp pair also has a diachronic dimension. Sakhalin Ainu preserves vowel-length distinctions in exactly the positions where Hokkaido Ainu carries an accent distinction: Itabashi (2001)

The long Sakhalin vowel occupies the syllable that carries initial accent in the Hokkaido exception word; the short Sakhalin vowel corresponds to the Hokkaido rule-derived form. This correspondence underpins the reconstruction of a proto-system in which these syllables were prosodically marked — by vowel quantity, retained in Sakhalin and restructured as pitch in Hokkaido Nakagawa (2024: 27, 51); Itabashi (2001). The direction of that change and whether the ancestral contrast was quantity-based or pitch-based remain disputed ‹contested› (Nakagawa & Fukazawa (2022: §3.1)). The SA–HA length–pitch correspondence independently confirms that the three minimal pairs above represent genuine lexical contrasts.

25.3 Functional load of the accent contrast

The contrastive yield of the accent system is limited. The placement rules account for the large majority of the lexicon; the exception list is closed and memorizable rather than an open accidental class. The three pairs documented above appear to constitute the best-established minimal/near-minimal set in the literature; Nakagawa (2024: 51) acknowledges the list to be incomplete, which implies further near-minimal contrasts in principle — any exception word whose segmental shape matches a rule-derived word — but such additional pairs are not systematically tabulated in the available sources. Lexical disambiguation relies primarily on morphosyntactic context, and the segmental form alone resolves ambiguity in most encounters.

The typological character that emerges is a pitch-accent system with a predominantly predictable accent assignment and a limited lexically stored residue. The formal typology of rising versus falling nuclei and the mora-syllable question are treated in Chapter 24 (The Prosodic Unit and the Accent-versus-Tone Analysis).

The contrastive relevance of accent varies across dialects. In Shizunai, Urakawa, Samani, Bihoro, Kushiro, and Shiranuka, pitch placement is not phonemically distinctive Nakagawa & Fukazawa (2022: §3.1); Nakagawa (2024: 49), and the three contrasts above are neutralized. Whether Shizunai should be classified alongside the accent dialects — the position of Refsing (1986: 73) — or treated as accentless, as Nakagawa (2024: 49–50) argues on the basis of variable archival transcriptions, has not been settled ‹contested›. The practical consequence is that the minimal pairs in this chapter are attributes of the Saru-type accent system and cannot be assumed for the accentless eastern and southeastern varieties. The dialect typology is addressed in Chapter 160 (Hokkaido Dialect Microvariation: Phonology and Morphosyntax).

The Yakumo and Oshamambe varieties retain pitch distinctiveness but apply a rightward-shifted rule: when both σ1 and σ2 are open, accent falls on σ3 rather than σ2 Nakagawa & Fukazawa (2022: §3.1); Nakagawa (2024: 50–51). Forms that contrast in Saru by the exception-versus-rule distinction may therefore pattern differently in Yakumo, since their accent is calculated from a shifted baseline — Saru notákam corresponds to Yakumo notakám, and Saru sinúye to Yakumo sinuyé (Nakagawa (2024: 50–51)). Whether the Yakumo rule interacts with the exception list in the same way as the Saru rule — and so whether the same minimal pairs surface in Yakumo — is not addressed in the available sources. The Yakumo system is described in Chapter 23 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split).

25.4 Accent of loanwords and reduplicated forms

Loanwords are adapted to the Hokkaido Ainu (C)V(C) syllable canon — consonant-cluster repair, epenthesis, and syllable-boundary insertion — before entering the phonological system, as described in Chapter 21 (Syllable Structure, Phonotactics, and Word-Edge Constraints). Once adapted, they are in principle subject to the same placement rules as native words. The Japanese borrowing tampaku 'tobacco' has its initial syllable closed by coda -m and would receive σ1 accent by 原則1 (támpaku); the Russian-origin form potoloko 'ceiling' begins with open σ1 and would receive σ2 accent by 原則2 (potóloko). Neither Nakagawa (2024) nor Shiraishi (2022) states a loan-specific accent rule, and the application of the placement rules to adapted loanforms is an inference from the general description of the system rather than a directly documented observation. Whether any loanword has joined the exception list with lexically stored non-rule accent is unattested in the available sources.

A community discussion observed a potential tension with the compound kamuynomi 'prayer ceremony' (kamúy + nomí → kamúynomi by first-element priority, Nakagawa (2024: 54)). The Ainu accent on the first element is expected; the observation concerned the different prosodic placement that this term acquires when it is borrowed into Japanese-language usage ‹speculative› (aynu-corpora Discord (2023–2026); palestine_2026, 2024-08-12). That is a property of the borrowing in its receiving language. Within Ainu, kamuynomi follows the first-element-priority pattern of all compounds, detailed in Chapter 26 (Accent in Compounds and under Affixation and Cliticization).

For reduplicated forms, direct evidence from phonetically transcribed accent-dialect texts is sparse. A community observation reports that somósomo — the reduplicated form of somó 'even, also' — functions as a single prosodic word accented on σ2, matching the base, and that the two-word string ka somó ki 'even … too' assigns accent on the base word independently ‹speculative› (aynu-corpora Discord (2023–2026); nukopoli, 2024-12-09). The pattern is consistent with first-element priority — in full reduplication the first copy retains its accent and the echo copy is de-accented — but primary-source confirmation from the phonetically transcribed descriptive literature is lacking. The prosodic-word domain and its role in accent assignment are treated in Chapter 26 (Accent in Compounds and under Affixation and Cliticization).