Adaptation patterns

In the northern Japanese dialects that supplied most of these words — Tōhoku outside the Sanriku coast and southern Fukushima, northern Niigata, and the coastal belt of Hokkaidō — the i-row and u-row vowels are centralized ([ï], [ɯ̈]) and the two rows fall together after sibilants and affricates, し ~ す, ち ~ つ, じ ~ ず alike. Ainu has no central vowel, so one and the same donor syllable is taken over now as i, now as u: つ gives ci in cikenki ← 付け木 and cu/tu in cunpo ← 壷. The u → i half of this is already stated as a rule for Japanese loans in Ainu, conditioned by a preceding s, c or z.

Ainu has no *ti, so ち has ci as its only possible outcome and carries no weight here; inotu ← いのち is the one form where a ち syllable surfaces with u.

Outside these syllables the vowel of a Japanese source form carries over unchanged. る gives ru in cirunpe ← 釣瓶, kuruma ← 車 and asinŋaru ← 足軽, and o in saro ← 猿 and ontaro ← 大樽; kiseri ← キセル is the only ru → ri. The -uy words kamuy ← 神 and pasuy ← 箸 belong to an older layer of contact — kamuy answers to Old Japanese kamï — and the bank leaves the direction of borrowing open for both.

Northern Japanese keeps the onset nasal on medial voiced obstruents — [mb], [nd], [ŋg] — that the standard language has lost, and southern Hokkaidō recordings still carry it. The loans show that nasal: さば → sampa, ふだ → hunta, むぎ → munki, あざみ → ancami. The donor is the same speech behind the i ~ u of the sibilant syllables, so the two patterns point at one source. Word-initially, where the donor has no prenasalization, a plain stop stands: peko ← べこ, taka ← だか, kotco ← ごちそう, conko ← じょうご. ～がつ → -ŋaci gives the nasal by itself, matching the ガ行鼻濁音 of the donor. Ainu obstruents carry no voicing contrast, and the p, t, k and c here follow the transcription of the dictionaries this bank draws on. Twenty of these words appear with b, d, g or z elsewhere in the corpus: Batchelor writes 「Peko … or Beko」, Komatsu's Samani recording has コンド, and 麦 is mungi in both Batchelor and Chiri. The nasal is the part every tradition writes. A borrowing may also keep the donor's own distinctions, as a stratum at the edge of the receiving system that carries a contrast with it (Itô & Mester 1999).

Ainu allows m, n and ŋ in the coda before a stop, so the donor's nasal has a place to stand.

Twelve forms have no nasal: usupa ← 薄刃, micipa ← 三つ葉, tesapa ← 停車場, simotori ← 相撲取り, isatono ← 医者殿, kaysotono ← 会所殿, sirokane ← 白金, sitoki ← 粢, kurumakay ← 車櫂, icicikan ← 一時間, níci-han ← 二時半, kanuci ← 雁の字. Several are transparent compounds, with the donor's own word boundary before the consonant.

The consonant of Japanese ハ行 was [ɸ], and before that [p]; [h] is the late outcome. The loans divide along that history. One set has Ainu p — pito ← 人, pera ← 箆, pone ← 骨, pukuru ← 袋, piwci ← 火打ち, sippo ← しほ — and another has h — hanasi ← 話, huku ← 服, hatto ← 法度, hoso ← 放送. 袋 sits in the bank twice, as pukuru and hukuro, one donor word taken over on both sides of the shift. Ainu has p and h alike, so neither reflex is forced by its phonology; what separates them is when the word crossed. The newest layer takes ひ [çi] to s: sikoki ← 飛行機.

Ainu uses both p and h in the onset, so the split is not a repair of an unavailable sound.

は gives h throughout, in all ten forms that carry it. Its one p is pasuy ← 箸, where the bank leaves both the source form and the direction of borrowing open, as it does for konru ← こほり.

Hokkaidō Ainu has no length contrast in its vowels, so a long donor vowel simply comes over short: じょうご → conko, ほうそう → hoso, さとう → sato. Three forms put an accent where the length was — tépu ← テープ, tóhumame ← とうふまめ, teppó ← 鉄砲 — and a long ū before a vowel comes over with a glide: しゅうと → siwto, ちゅうはん → cuwan. Weight in the consonants behaves the other way round. A donor sokuon is kept as a geminate (はっと → hatto, とっくり → tokkuri, ラッパ → rappa), and a geminate also turns up in forms whose donor has none: ひしゃく → pisakku, しほ → sippo, つとこ → tutukko, おとな → ottena.

An Ainu syllable may close in p, t, k, s, m, n, ŋ, r, w or y, so a geminate spans a coda and the following onset.

Sakhalin Ainu does have long vowels, from a lost coda; the forms here are Hokkaidō. Where an affricate follows the coda the cluster is tc rather than a true geminate: mutci ← 鞭, kotco ← 御馳走, pakutciki ← 博打.

A Japanese verb is never taken over as a verb that Ainu then inflects. It arrives as a fixed stem in the donor's た-form with -ro after it — あづかる → ancikattaro, まける → maketaro, かつ → kattaro, ちょす → cositaro — and the Ainu verb is built from there. A borrowed noun or adjective is verbalized by native means instead: ne 'be' in pakane ← 馬鹿, hantásine ← 裸足, yantone ← 宿; kar and ki 'do' in pakakara ← 馬鹿, hukasikar ← 吹かす, pakutciki ← 博打. Ainu prefixes then apply to the borrowed stem like any other: esimpay ← e- + 心配, ewkoyakamasi ← e- + u- + ko- + やかましい, upakanere ← u- + pakane + -re. Nouns compound freely with native elements: kunne + すみ, ni + おけ, yar + たび.

Ainu inflection — personal affixes, the plural, applicatives — attaches to the derived stem, never to the Japanese material inside it.

maketa ← まける stops at the た-form with no -ro. The bank records the -ro without settling where it comes from.