Chapter 18/s/ and the s ~ š Palatalization Alternation

18.1 The single fricative and its positional variants

Hokkaido Ainu has a single fricative phoneme, /s/ (see Chapter 16 (The Consonant Inventory and Its Phonetic Realization)). Its phonetic realization depends on syllabic position. In onset function /s/ appears as a laminal sibilant with a degree of palatalization that varies by speaker and vowel context. In coda function the realization splits by the preceding vowel: after /i/ the coda /s/ surfaces as [ɕ], merging perceptually with the initial consonant of Japanese シ; after /a/, /e/, /o/, and /u/ it surfaces as a partially palatalized laminal [sʲ]~[ʃ], intermediate between Japanese ス and シ. A third context overrides the coda rule: immediately before the glide /w/ at a following syllable or word boundary, the coda /s/ surfaces as a plain [s] Nakagawa (2024: 27–28); Shiraishi (2022: §3).

Because /s/ is the only fricative in the Hokkaido Ainu phoneme inventory, the alternants [sʲ], [ʃ], [ɕ], and [s] are all allophones of one underlying segment, never contrastive with each other. The distribution does, however, produce perceptual near-mergers that led early transcribers working within a Japanese kana frame to hear palatal colouring in unexpected environments, and it creates systematic difficulty for Japanese-background listeners distinguishing morphological structures that differ in syllabic affiliation rather than in the consonant itself.

18.2 Onset realization

In onset position /s/ is realized as a laminal alveolar fricative. A community phonetics survey characterizes it as [s̻] — with the blade of the tongue making the primary constriction rather than the tip — and notes a perceptible difference from the palato-alveolar [ɕ] of Japanese シ even when both are heard as "sh-like" aynu-corpora Discord (2023–2026) (nukopoli, 2023-12-11) ‹speculative›.

The palatalization tendency of onset /s/ in older recordings extends beyond the environment of a following /i/. Community phonetic analysis of early twentieth-century audio reports that the onset /s/ was palatalized in a wider range of vowel environments than is typical of the better-documented mid-century speakers aynu-corpora Discord (2023–2026) (nukopoli, 2023-12-05) ‹corpus-suggested›. Shiraishi describes Ainu /s/ as "often palatalized" Shiraishi (2022: §3), citing Kindaichi (1931: 9) and Tamura (1988b: 13) for the forms sini [ʃini] and as [aʃ] as early printed attestations.

Early Japanese transcriptions of Ainu rendered the onset sa, su, se, so syllables with the digraphs シャ, シュ, シェ, ショ rather than サ, ス, セ, ソ, reflecting a palatal impression before non-high vowels. Tamura (1996) states that in Saru "the fricative is /s/ alone, close to the consonant of シ, but with individual variation; sa, se, so, su vary between a サ-type and a シャ-type pronunciation, while si is invariably シ" (reported via aynu-corpora Discord (2023–2026), nukopoli, 2023-12-11; Shiraishi (2022: §3) corroborates the general pattern) ‹speculative›.

18.3 Coda allophony: the /i/-conditioning

The realization of coda /s/ is conditioned by the vowel of the syllable nucleus it closes. When that vowel is /i/, the coda /s/ surfaces as [ɕ]; when it is /a/, /e/, /o/, or /u/, the coda surfaces as a partially palatalized [sʲ]~[ʃ]. The rule as stated in Nakagawa (2024: 27–28) and corroborated by Shiraishi (2022: §3) can be formulated:

/s/ (coda) → [ɕ] / i_ ; → [sʲ]~[ʃ] / (a, e, o, u)_

The following table gives attested examples in word-citation (pause) form. All coda forms are taken from Nakagawa (2024: 27–28); the general rule applies equally to coda /s/ after /u/, though no dedicated example is cited there.

The near-merger between coda /s/ after /i/ and onset /s/ before /i/ generates a systematic perceptual difficulty. The pair tures 'younger sister' — with coda /s/ in the syllable /res/ — and tures-i 'his/her younger sister' — in which the same /s/ resyllabifies as the onset of the following possessive suffix -i, giving the syllable /si/ [ɕi] — both approach [tureɕ] at their right edge as heard by Japanese listeners, even though they are structurally distinct Nakagawa (2024: 27–28). For Ainu speakers the difference is in the syllabic affiliation of /s/ (coda vs. onset), recoverable from morphological structure. Resyllabification at morpheme boundaries is treated in Chapter 21 (Syllable Structure, Phonotactics, and Word-Edge Constraints); the morphophonology of the possessive suffix is taken up in Chapter 45 (Morphophonology of the Affiliative Suffix and Echo/Copy Vowels).

The onset /s/ before /i/ participates in the same [ɕ] realization. The reflexive prefix si- ‘reflexive’ surfaces as [ɕi], the same phonetic output as coda /s/ after /i/. The two environments converge on [ɕ] from different structural positions (coda conditioned by preceding /i/, onset conditioned by following /i/) without forming a phonemic /ɕ/ distinct from /s/.

18.4 Coda /s/ before /w/: the override constraint

At a juncture before the labio-velar glide /w/, the coda /s/ surfaces as plain [s] rather than the [sʲ]~[ʃ] expected from the non-/i/ conditioning rule. Tamura (1996) describes as 'stand' in isolation as アㇱ ([aʃ]), but the same verb followed by the conjunctive particle wa ‘and then; conjunctive’ becomes アㇲワ ([aswa]), with the fricative shifting to a plain sibilant before the /w/ onset aynu-corpora Discord (2023–2026) (aomidori, 2023-12-11, reporting Tamura 1996); see also Nakagawa (2024: 27) for the coda-/s/ allophony frame within which this constraint operates ‹corpus-suggested›.

The phonetic interpretation is straightforward: the labial co-articulation of /w/ suppresses the palatal colouring that the coda /s/ otherwise carries. Whether this is best analysed as anticipatory labial assimilation blocking palatalization, or as a context-sensitive allophone rule that takes precedence over the general coda rule, is not resolved in the sources consulted. The broader sandhi environment in which /s/ participates at word junctures — including the /r/+/s/ cluster that produces progressive effects (see Chapter 28 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters)) — is treated there.

18.5 Ainu [ɕ] and Japanese /ɕ/: a phonetic note

The [ɕ] that appears as the coda /s/ allophone after /i/ and as the onset /s/ allophone before /i/ is phonetically distinct from the Japanese palato-alveolar [ɕ] of シ. Community phonetic analysis characterizes the Ainu surface [ɕ] as having less high-frequency friction than its Japanese counterpart, placed somewhat further back in the oral cavity — on the boundary between Japanese [ɕ] and an English postalveolar [ʃ] — while never reaching a full retroflexed or postalveolar [ʃ] aynu-corpora Discord (2023–2026) (nukopoli, 2023-12-11) ‹speculative›. The analysis characterizes it as a laminal sound, with the tongue blade rather than the tip engaged, which distinguishes it from the Japanese [ɕ] even within the palato-alveolar region.

This distinction has no phonemic status: there is no /ɕ/ phoneme in Hokkaido Ainu separate from /s/; the [ɕ] surface form is entirely predictable from the vowel context. The use in earlier descriptions of /ʃ/ as a coda symbol (as in Shiraishi's notation as [aʃ], and Tamura's representation of coda -s after /i/ as シ) reflects the impressionistic transcription of the fully palatalized allophone, not a phonemic /ʃ/ distinct from /s/ Shiraishi (2022: §3).

18.6 Speaker and dialect variation

The degree of coda palatalization shows inter-speaker variation within and across dialects. Analysis of NINJAL recordings identifies Hiraga Sadamo (平賀さだも, Saru dialect) as producing "pure [ɕ]" at the coda and onset-before-/i/ position, while Mukawa dialect recordings from the same archive frequently show a reduced variant [ɕ̱̥] — a partially devoiced, laminal palato-alveolar — in the same environment aynu-corpora Discord (2023–2026) (nukopoli, 2024-03-24) ‹speculative›. The Mukawa variant suggests an interaction between coda /s/ palatalization and the vowel devoicing that affects /i/ between voiceless consonants (treated in Chapter 17 (The Vowel Inventory and Vowel Realization)): when /i/ devoices in the environment of the coda /s/, the palatalization feature is retained on the fricative but the overall phonetic output moves toward a devoiced palatal fricative [ç] or a devoiced [ɕ̥].

These inter-speaker differences have not been quantified from published acoustic measurements in any directly-read source. The phonetic detail is graded accordingly. The vowel devoicing pattern that creates the relevant environment in Mukawa is independently attested in Shiraishi (2022: §2.1) and Nakagawa (2024: 28).

18.7 Progressive palatalization in sirsak

A sporadic case of progressive palatalization has been observed in NINJAL corpus recordings of the compound word sirsak NINJAL (2003). The underlying form /sir.sak/ has initial /si/ → [ɕi] by the regular onset rule, followed by the coda /r/ of sir and then the onset /s/ of sak. In the speech of 木村きみ (Kimura Kimi, Saru), the coda /r/ after /i/ is palatalized ([ɾʲ]), and this palatalization carries forward progressively onto the following /s/, shifting it from the expected laminal [s̻] toward a laminal postalveolar [ᶘ], as in [ɕiɾʲᶘak˚]; an alternative surface realization is [ɕiɾʲɕak˚] aynu-corpora Discord (2023–2026) (nukopoli, 2026-05-11) ‹speculative›.

The speaker 小田イト (Oda Ito, also documented in Saru at the same NINJAL site) does not show the same progressive shift: her realization is [ɕiɾʲis̻ak˚], with an echo vowel after the /r/ coda (a property of coda /r/ discussed in Chapter 19 (The Rhotic /r/ and Coda-r)) and a plain laminal [s̻] — not a palatalized [ᶘ] or [ɕ] — as the onset before /a/ aynu-corpora Discord (2023–2026) (nukopoli, 2026-05-11) ‹speculative›.

The contrast between the two speakers illustrates two independent strategies for the /ɾʲ.s/ juncture: progressive coarticulation of palatality from [ɾʲ] onto the following /s/ (Kimura), versus insertion of an echo vowel that interposes a vocalic buffer between the palatalized rhotic and the onset /s/ (Oda). The observation is idiolect-specific in the available NINJAL data, has not been confirmed in other compounds or other speakers in any read source, and the IPA transcription reflects community analysis rather than published spectral measurement ‹speculative›. The broader /r/+/s/ sandhi environment — where a coda /r/ interacts with a following /s/ onset — is treated in Chapter 28 (Assimilation and Cluster Simplification: Coda /r/, Nasals, and Clusters).