Chapter 160Hokkaido Dialect Microvariation: Phonology and Morphosyntax

160.2.1 The standard accent type

In the standard type, a closed first syllable draws the high pitch to σ1; an open first syllable places it on σ2. The rules operate on the bare lexical stem; cliticized person prefixes shift the accent domain (see Chapter 26 (Accent in Compounds and under Affixation and Cliticization)). The Saru–Chitose instantiation of this type is the baseline described by Ijäs (2023) and the system treated in Chapter 23 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split).

The geographic spread of the standard type is non-contiguous. The Saru–Chitose group in south-central Hokkaido, Horobetsu on the west coast, the inland Asahikawa–Nayoro corridor, and the eastern Obihiro dialect share identical accent rules despite the intervening unaccented belt. Sōya, at the northern tip of Hokkaido, also belongs here — a significant fact given that Sōya is the Hokkaido locality most closely cognate to Sakhalin in the lexicostatistic matrices of Hattori & Chiri (1960) and the dialectometric reanalysis of Ono (2020).

160.2.2 The Yakumo–Oshamambe irregular type

When both the first and second syllables of a word are open, Yakumo and Oshamambe place the high pitch on the third syllable rather than the second. The contrast with the standard type is audible on identical lexical stems Nakagawa & Fukazawa (2022: §3.1):

The contrast recurs in verbs: Saru hopúni 'fly' vs Yakumo hopuní. Whether the irregular pattern is a phonological innovation at Yakumo/Oshamambe or an archaic retention from a stage before the standard rule spread has not been settled; neither has any synchronic account of what structural property of consecutive open syllables triggers the shift ‹contested› Nakagawa & Fukazawa (2022: §3.1).

160.2.3 Unaccented varieties

Six localities — Shizunai, Urakawa, Samani, Bihoro, Kushiro, and Shiranuka — have been described as lacking phonologically contrastive pitch at the lexical level. The description requires qualification: Samani tends toward a final-high pattern in practice; Shizunai shows no stable positional pattern across the lexicon Nakagawa & Fukazawa (2022: §3.1).

These varieties did not lose their accent through the same mechanism as Sakhalin Ainu, where the absence of pitch correlates with the retention of vowel length (and a long-vowel/high-pitch correspondence with the standard Hokkaido type; see Chapter 157 (The Final-h Problem and Its Sakhalin Vowel-Length Reflex)). In Hokkaido, the unaccented dialects appear to represent a secondary pitch-neutralization within a variety that had already developed the pitch-accent system — the opposite direction from Sakhalin's retention of length as the suprasegmental distinction ‹contested› Nakagawa & Fukazawa (2022: §3.1); Itabashi (2001). The broader prosodic comparison with Sakhalin is taken up in Chapter 24 (The Prosodic Unit and the Accent-versus-Tone Analysis).

160.3.1 Word-initial h-elision

In eastern standard-accent dialects — Obihiro, Kushiro, Bihoro, Asahikawa, and Nayoro — a word-initial h is absent when the accent falls on σ2, that is, when σ1 is open. The elision environment is exactly the standard-type accent context where the initial consonant carries no pitch-peak information; western and southern dialects (Horobetsu, Biratori, Nukibetsu, Niikappu, Samani, Sōya) retain h throughout Nakagawa & Fukazawa (2022: §3.2, Table 4):

The synchronic status of the glottal onset that phonetically precedes the initial vowel in some realizations of the eastern forms varies by dialect and speaker and connects to the analysis of Chapter 20 (The Laryngeals: Glottal Stop, /h/, and the Final-h Question); the relevant point here is that h is absent as a segment, not merely reduced ‹corpus-suggested› Nakagawa & Fukazawa (2022: §3.2).

160.3.2 Regressive cluster assimilation

The most geographically concentrated consonant alternation in the Hokkaido field is a set of regressive assimilation processes converting heterorganic clusters at morpheme junctions into geminates or homorganic sequences. The alternations are confined to the northeast and are most extensive in Bihoro Nakagawa & Fukazawa (2022: §3.3, Table 5):

Nakagawa (2024) treats these as historically altered word-forms specific to those localities — lexical entries in the Bihoro dialect — rather than as the output of a productive synchronic sandhi rule applying across the board. A community discussant takes a different interpretive direction, suggesting that the geographic concentration of the pattern in Bihoro points to an underlying "proto-phoneme" (an ancestor consonant distinct from both of the modern alternants) whose collapse produced the heterorganic clusters later simplified by assimilation ‹speculative› aynu-corpora Discord (2023–2026) (antitwilight, 2024-11-17). The relationship of the assimilated forms to the broader coda-sandhi phenomena described in Chapter 28 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters) has not been worked out.

160.3.3 The pa-:ca- alternation

One of the most discussed phonological isoglosses in the Hokkaido field is the distribution of pa- and ca- in a small set of words related to 'mouth': par (southwest) and car (northeast) 'mouth', papus ~ capus ~ patoy ~ catoy 'lip'. Nayoro and Asahikawa attest both forms, marking the interior transition zone Nakagawa & Fukazawa (2022: §3.4).

The phenomenon is formally restricted: there is no comparable pe-/ce-, pi-/ci-, or po-/co- correspondence in the lexicon, which is why Nakagawa & Fukazawa (2022: §3.4) call it a "pseudo-correspondence" rather than a sound correspondence in the strict sense. Place-name evidence corroborates the geographic split: Kirikae (1994) records eleven par-type place names against zero car-type in Ishikari; in Nemuro and Kushiro the ratio reverses (thirteen car-type). The Kuril attestations — Krasheninnikov 1738 and Golovnin 1811 — record car for 'mouth', consistent with the northeastern and eastern distribution. Tamura (1996) gives par as the Saru concept-form, with the possessed form paro-ho.

Four analyses of the alternation have been proposed ‹contested›:

• Kindaichi, Chiri, and Naert 1958 (via Nakagawa & Fukazawa (2022: §3.4)): a diachronic change pa → ca, or the reconstruction of a single Proto-Ainu syllable *Xa (with an indeterminate initial consonant).
• Kirikae (1994): Proto-Ainu already had two distinct 'mouth' lexemes; the par and car forms are not cognates but separate vocabulary items with overlapping reference.
• Nakagawa 1996 (via Nakagawa & Fukazawa (2022: §3.4)): the ca- form is the older one, evidenced by its appearance in place-names and formulaic expressions (sanca or ta) inside otherwise par-using regions — an ABA distribution where a central innovation displaced the peripheral retention.
• Fukazawa (2017): ca- and pa- were always distinct morphemes, with irreducible minimal pairs can:pan 'thin' and cas:pas 'run' coexisting in the same dialects with distinct phonological histories.

Alonso de la Fuente (2022) §2.2.6 reconstructs PA *pʸaɾ for 'mouth', treating the p/c split as a relic of a Proto-Ainu palatalized consonant series in which *pʸ shifted toward c in the northeast; this reconstruction is typologically motivated rather than philologically forced ‹speculative›. A community discussant proposes a related but distinct mechanism: that the alternation reflects a Proto-Ainu syllable *ti, whose labialized variant underwent palatalization via an intermediate glide (*pj → *tj → c), overlapping with Alonso's palatalized-series account but positing a glide intermediate ‹speculative› aynu-corpora Discord (2023–2026) (nukopoli, 2023-12-08). None of the four analyses has achieved consensus.

160.3.4 Metathesis: -iw and -uy

Bihoro and Kushiro show a systematic alternation in which the sequences -iw and -uy are transposed relative to the forms recorded elsewhere Nakagawa & Fukazawa (2022: §3.5, Table 7):

The base form has a glide following the vowel (iw); the Bihoro/Kushiro form has the vowel following the glide (uy = u + y). The alternation is limited to these two unaccented eastern dialects; standard-accent dialects do not participate.

Additional, lexically specific metatheses occur in other dialects without forming a regular alternation: piyar / puyar / puray 'window' (three variants across the field), nipek / nupek / nikep 'light', and askepet / aspeket 'finger' (Horobetsu) Nakagawa & Fukazawa (2022: §3.5, Table 8). Their relationship to the -iw/-uy set is unclear.

160.3.5 Other consonant alternations: y/w and n/m

Two further consonant alternations are documented across the dialect field. First, y and w alternate in a small set of lexical items in the Yakumo area: puyar and nuye where most dialects have puwar and nuwe Nakagawa & Fukazawa (2022: §3.6, Table 12). This remains lexically specific.

Second, n and m vary in particular words and localities: nimak ~ mimak 'tooth' (Yakumo, Nukibetsu, and Niikappu favour mimak); and ómke ~ ónke 'cough' at Sōya versus other dialects Nakagawa & Fukazawa (2022: §3.6, Tables 13–14). A community discussant observes that the change mi → ni (or its inverse) appears broadly in Hokkaido and suggests the epicentre may lie in the Horobetsu area, where long-distance assimilation could be the operative mechanism ‹corpus-suggested› aynu-corpora Discord (2023–2026) (nukopoli, 2024-12-06). That hypothesis is consistent with the nimak/mimak distribution but requires systematic corpus verification before it can be stated as established.

An isolated consonant-copying phenomenon produces Sōya purár against the more widespread puwar 'fog/mist'; and the alternation hútne ~ Bihoro húpne 'narrow' represents a C:p correspondence that remains lexically isolated Nakagawa & Fukazawa (2022: §3.6, Table 15).

160.3.6 Allophonic microvariation in the 1967 field record

Simeon's fieldwork of June–July 1967 on Hiratori, Horobetsu, and Higashi-Shizunai — the informants were Kayano Shigeru, Itaku Haru, and Sasaki Taro — records a set of sub-phonemic differences among three southern dialects that later descriptions rarely separate Simeon (1968: 13–17). Intervocalic voicing of the obstruents runs along a gradient: Higashi-Shizunai shows [b d g dʑ] most freely (hapo [habo] 'mother', ante [ande] 'put down', yakun [jagun] 'if', eci [edʑi] 'you (pl.)'), Horobetsu less often, and Hiratori hardly at all. The [ɸ] realization of /h/ before /u/ is constant in Horobetsu and Higashi-Shizunai and optional in Hiratori (hure [ɸure] 'red'); final /n/ surfaces as [ŋ] in Horobetsu and Higashi-Shizunai (kotan [kotaŋ]); and in Horobetsu the contrast of /h/ and /s/ is neutralized before /i/, giving free variation of the type unihi ~ unisi '(my) home' Simeon (1968: 13) — the same sibilant–laryngeal instability treated in Chapter 18 (/s/ and the s ~ š Palatalization Alternation).

These observations carry a caveat. Except for Kayano, Simeon's informants had seldom spoken Ainu for decades, all were bilingual, and no acoustic verification was made; a gradient of intervocalic voicing in particular could reflect Japanese interference or moribund-speaker variability as easily as an old isogloss. The record is best read as a dated snapshot of individual speakers, valuable precisely because each is localized and named ‹speculative›.