Chapter 162Sakhalin and Kuril Ainu: The External Comparison

162.1 Ainuic: three varieties, two branches

Ainu has no demonstrated genetic relatives and is treated as a language isolate Janhunen (2022: §2); the failed macro-comparisons are surveyed in Chapter 5 (The Genetic Position of Ainu: Isolate Status and Macro-Comparison Controversies). Internally the language forms a small family — sometimes labelled Ainuic — of three entities: Hokkaido (HA), Sakhalin, and the Northern Kuril varieties Janhunen (2022: §2). The tripartition was formalised on geographic grounds in the early twentieth century and appeared to receive objective support from the lexicostatistic studies of Hattori & Chiri (1960) and Asai (1974). Ono (2020) demonstrates with spectral-clustering methods, however, that the three-way "Major Division" carries no statistically robust support: the cleaner two-way structure is Hokkaido (including the Kuril varieties) versus Sakhalin. The internal Hokkaido classification and the dialectometry underlying that finding are treated in Chapter 159 (Hokkaido Dialect Classification and Dialectometry); the boundary-dialect methodology is refined further by Fukazawa & Ono (2024).

The branching order follows the archaeological record of the Satsumon culture's northward expansion from Honshu: the language reached Hokkaido from around the seventh century CE, southern Sakhalin around the thirteenth century, and the Northern Kuriles around the fifteenth to sixteenth century Janhunen (2022: §3a–3b). The Sakhalin branch diverged from Common Ainu approximately 800 years ago; the Kuril varieties diverged later and cluster more closely with northeast Hokkaido in the lexicostatistic data Alonso de la Fuente (2022: §4); Janhunen (2022: §3b). Both the Sakhalin and Kuril varieties ceased to be community languages during the twentieth century. The primary Sakhalin documentation consists of the grammar and texts of Murasaki (1979) and the folkloric corpus of Piłsudski (1912); the Kuril varieties are known only from questionnaire records assembled by eighteenth- and nineteenth-century travellers. Within Hokkaido, the dialect geographically and lexically closest to Sakhalin is Sōya, at the northern tip of the island Hattori & Chiri (1960).

162.2 The prosodic correspondence: length and pitch

The Sakhalin varieties have a vowel-length contrast on the five vowels a e i o u, with long vowels occurring only in open stem syllables; Hokkaido has a pitch-accent system and no distinctive vowel length. Hattori (1967) first established the correspondence: a Sakhalin long vowel corresponds to a Hokkaido high pitch, and a Sakhalin short vowel in a closed-syllable stem likewise corresponds to Hokkaido high pitch in the same environments Alonso de la Fuente (2022: §2.3); Tangiku (2022: §2.1). The pitch-accent rules are described in Chapter 23 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split); the diachronic account of the length–pitch relationship is the subject of Chapter 157 (The Final-h Problem and Its Sakhalin Vowel-Length Reflex).

Three competing reconstructions of Proto-Ainu prosody account for this correspondence. (a) Hattori (1967) and De Boer (2010) held that Proto-Ainu had vowel length alone and Hokkaido developed pitch accent through contact with Japanese. (b) Vovin (1993) and Itabashi (2001) argued that Proto-Ainu had both length and pitch, with the Sakhalin branch losing accent under Amur-Sakhalin contact. (c) Satō (2015, 2016) proposed Proto-Ainu had pitch and no length, the Sakhalin vowel length being a phonetically conditioned open-syllable phenomenon. Alonso de la Fuente endorses view (c) in the body of his chapter but (a) in the conclusion, leaving the question unresolved Alonso de la Fuente (2022: §2.3–2.4) ‹contested›. Community discussion (nukopoli, aynu-corpora Discord 2023; aynu-corpora Discord (2023–2026)) arrives at the same impasse, recognising both the Hattori and the Satō positions as defensible.

162.3 Final-consonant debuccalization in Sakhalin

The phonological signature of the Sakhalin varieties relative to Hokkaido is systematic debuccalization of coda obstruents. Where Hokkaido has a word-final or syllable-final p, t, or k, the core Sakhalin varieties produce h; final m becomes n; and final r receives a following vowel, keeping the syllable open Tangiku (2022: §2.3); Alonso de la Fuente (2022: §1). A secondary rule applies within Sakhalin: the h produced by debuccalization becomes s following i (h > s / i__), yielding alternants such as sis for HA sik 'eye'. Alonso de la Fuente uses this set of alternations as his illustration of what internal reconstruction can accomplish for a language isolate without recourse to the comparative method Alonso de la Fuente (2022: §1). The correspondences are set out in Table 2; the coda consonants of the Hokkaido system are described in Chapter 16 (The Consonant Inventory and Its Phonetic Realization).

The Hokkaido forms in the first column are corpus-attested throughout; the constructed example below places cep in a transitive clause to show the coda -p in context. The corresponding Sakhalin Raiciska form ceh is the regular debuccalization reflex Alonso de la Fuente (2022: Table 18).

Community corroboration: the kor > koh change and the general pattern of coda-consonant debuccalization in Sakhalin are noted, with examples, in aynu-corpora Discord discussion (nukopoli 2024-12-19, aldehyde8908 2025-04-22; aynu-corpora Discord (2023–2026)), consistent with Tangiku (2022: §2.3).

The debuccalization applies across the Sakhalin varieties with notable subdialect variation. The eastern coast varieties Taraika and Tarantomari preserve the coda stops p/t/k like Hokkaido; the Esutoru subdialect distinguishes -is from -ih Tangiku (2022: §2.3). The phonological status of SA coda h itself is disputed: Alonso de la Fuente treats it as a velar fricative /x/ (with realisations [ax], [iɕ], [uf]), while Tangiku and community members analyse it as a laryngeal breath h Alonso de la Fuente (2022: §1); Tangiku (2022: §2.3) ‹contested›.

162.4 Word-initial consonants in the Nairo subdialect

Within Sakhalin, the Nairo subdialect (eastern Sakhalin) shows word-initial t- where Hokkaido and the remaining Sakhalin varieties have r-: Nairo teera 'wind' corresponds to HA rera and non-Nairo SA reera; Nairo taarak 'smooth' corresponds to HA raarak Tangiku (2022: §2.5). Alonso de la Fuente interprets this as retention of a Proto-Ainu lateral *l that merged with r in most dialects but with t in Nairo — on this account Nairo preserves the most archaic reflex for the relevant set Alonso de la Fuente (2022: §2.2.1–2.2.2) ‹contested›. The PA consonant reconstruction, including the evidence from Ainu→Ghilyak loans showing Proto-Ghilyak *l- opposite these forms, is treated in Chapter 156 (Proto-Ainu Reconstruction: Segments and Accent Classes).

162.5 Morphological and lexical divergences

162.6 The Kuril varieties

The Northern Kuril varieties are known only from questionnaire records compiled by eighteenth- and nineteenth-century travellers — abbreviated in Alonso de la Fuente's discussion as KD (Dybowski), KK (Krašeninnikov), KL (Klaproth), KT (Torii), and KV (Voznesenskij) Alonso de la Fuente (2022: §2.2). The lexicostatistic reanalysis of Ono (2020) finds the Kuril varieties clustering with the northeast Hokkaido group — Samani, Obihiro, Kushiro, Bihoro, Sōya — rather than with Sakhalin, supporting a two-way Hokkaido-plus-Kuril versus Sakhalin division over the classic three-way one. A separate analysis using nineteenth-century Kuril material by Satō and Bugaeva (2019) places the Kuril items closer to southern Hokkaido (Saru/Chitose); the two results rest on different data sets and are not directly comparable Ono (2020) ‹contested›.

Phonologically the Kuril varieties largely preserve the coda stops p/t/k that Sakhalin debuccalises, but they also show prothetic h- in some forms alongside loss of original h- in others — an independent innovation parallel to the prothetic h- found in the Bihoro dialect of Hokkaido Alonso de la Fuente (2022: §2.2.5). The word for 'mouth' appears as car in Kuril records (aligning with northeastern HA and Sakhalin against the southwestern HA form par), an isogloss documented in place-name data by Kirikae (1994) (via Nakagawa & Fukazawa (2022: §3.4)) and diachronically accounted for as a palatalized-series relic *pʸaɾ by Alonso de la Fuente (2022: §2.2.6). The diachrony of that split is treated in Chapter 156 (Proto-Ainu Reconstruction: Segments and Accent Classes). Hokkaido phonological microvariation, including the Bihoro and northeastern features that align partially with Kuril, is described in Chapter 160 (Hokkaido Dialect Microvariation: Phonology and Morphosyntax). A Kuril contrast text appears in Chapter 172 (A Glossed Sakhalin (Enciw) Text with Hokkaido Contrast).

162.7 Northern contact and the Hokkaido residue

Sakhalin was the primary zone of Ainu contact with Nivkh (Ghilyak), Uilta (Tungusic), and Manchu-mediated trade vocabulary from Qing China. The multilingualism was "extremely limited and localised", concentrated among individual interpreters rather than being community-wide Shiraishi & Tangiku (2022: §2.5). Most of the resulting loanword layer remained specific to the Sakhalin varieties; the full contact account belongs in Chapter 165 (Northern Contact: Nivkh, Tungusic, and Manchu). The items with a direct reflex in Hokkaido are:

• kaya ‘fish-skin garment; sail’ — borrowed on Sakhalin from Nivkh, reaching Hokkaido with the specialised sense 'sail' only Shiraishi & Tangiku (2022: §3.2);
• sintoko ‘lacquerware cask’ — morphologically unanalysable in Ainu, possibly connected to the Eurasian sanduk Wanderwort, also appearing as Nivkh sinduχ Shiraishi & Tangiku (2022: §3.4);
• san ‘offering shelf’ — Nivkh cʰaŋ, Uilta saan Shiraishi & Tangiku (2022: §3.6);
• tunakay ‘reindeer’ — from Pre-Proto-Ghilyak *tola-ŋa.y, also the source of Japanese トナカイ Janhunen (2022: §3c).

In the reverse direction, several Ainu lexical items passed into Nivkh and Uilta on Sakhalin Shiraishi & Tangiku (2022: §3.2–3.6); the directionality in each case is argued from morphological analysability, plausible sound correspondences, and the dialect distribution of the forms across Sakhalin versus the continent.