Knowledge base · Part III — Segmental Phonetics, Phonology, and Phonotactics · this topic as a network
22. Glides, Vowel Hiatus, and the Diphthong Question
How does Hokkaido Ainu treat word-final sequences of a vowel followed by y or w?
The analytical question
What are the two competing analyses for word-final vowel-plus-y/w sequences in Hokkaido Ainu?
Two competing analyses bear on word-final vowel-plus-glide sequences in Hokkaido Ainu. Itabashi reports that the inventory and number of semivowels are identical across all dialects, with y and w as the relevant segments (Itabashi 2001). Murasaki asserts that /w/ is realized as [w] at syllable onset but weakens in syllable-coda position, so that careful speech yields [-uʔV] where rapid speech yields [-uwV] (Murasaki 2025: 2); similarly, /y/ is [j] at syllable onset and weakens in coda position, with careful [-Vʔi] becoming [-Vy] in rapid speech (Murasaki 2025: 2).
generated 3 claims, 1 sentences without a claim ·
The consonant analysis
What evidence supports treating post-vocalic y and w as coda consonants in Hokkaido Ainu?
The consonant analysis holds that post-vocalic y and w function as coda consonants rather than as vowels or epenthetic segments. Nakagawa asserts that phonetically /-y/ is almost identical to [i], but phonologically it behaves as a consonant that forms a single syllable with the preceding vowel, cannot bear accent, and is pronounced more weakly than /i/, as in aynu and kamuy (Nakagawa 2024: 31); the same holds for /-w/, which is phonetically almost identical to [u] but phonologically a consonant forming a single syllable with the preceding vowel, unable to bear accent and pronounced more weakly than /u/, as in ohaw and kewtum (Nakagawa 2024: 31). A transitional glide between vowels is not a phoneme, so non-phonemic [y] and [w] appearing between vowels are not written in phonemic transcription (Nakagawa 2024: 34); for example, the [y] in [iyomante], from prefix /i-/ and verb /omante/, and the [w] in [uwenewsar], from prefixes /u-/ and /e-/ and verb /newsar/, are non-phonemic and are transcribed phonemically as /iomante/ and /uenewsar/ (Nakagawa 2024: 34). By contrast, the [w] in /iwor/ 'hunting ground' is the realization of the phoneme /w/ and not a transitional glide (Nakagawa 2024: 35), and the word for 'eloquence' has the stem /paw/ rather than pa with an epenthetic glide, meaning there are two roots for 'mouth', /pa/ and /paw/ (Nakagawa 2024: 35). The phonemic form of the verb for 'to hurt' in the Chitose and Saru dialects is /iunin/, and the phonetic [y] in [iyunin] is a transitional glide rather than a phoneme (Nakagawa 2024: 40). Murasaki reports semivocalization in clusters such as -n + s- becoming -ys-, -V + 'i- becoming -Vy-, -V + 'u- becoming -Vw-, -V + 'a becoming -Vya-, and 'u + 'V- becoming -wVV- (Murasaki 2025: 89). In the y:w consonant alternation, Yakumo characteristically takes w where other dialects take y, and Horobetsu and Obihiro take y where other dialects take w, as seen in puyár / puwár and suyóp / suwóp (Nakagawa & Fukazawa 2022: 8.3).
generated 9 claims, 1 sentences without a claim ·
Syllabification and pitch-accent placement
How does first-syllable weight determine pitch-accent placement in Hokkaido Ainu?
Bugaeva asserts that when the prefixes i- (indefinite object) or u- (reciprocal) occur intervocalically in an accent-bearing position, they turn into a glide and form a syllable with the following vowel (Bugaeva 2022: 20.6).
generated 1 claims ·
The diphthong analysis
How did early researchers analyze word-final vowel-plus-y/w sequences in Hokkaido Ainu?
Early researchers generally analyzed word-final vowel-plus-glide sequences as diphthongs. Kindaichi and Chiri assert that diphthongs include ai, ui, oi, au, iu, eu, ou, ei, and ae, all of which are falling diphthongs (Kindaichi & Chiri 1936); among these, ou, ei, and ae are extremely rare, with ou occurring only in kokou 'son-in-law', ei formed when short e is emphasized, and ae occurring only word-initially through rapid contraction of a- and e- (Kindaichi & Chiri 1936). Refsing asserts that in syllable-final position, semivowels /w/ and /y/ are realized as [u] and [i], forming diphthongs with the preceding vowel (Refsing 1986: 68), and further that when an unstressed /u/ or /i/ follows a vowel in the Shizunai dialect, it may be realized as non-syllabic [w] and [y] respectively (Refsing 1986: 73). Itabashi asserts that in Hokkaido Ainu the semivowels /y, w/ serve as the second vowel of a diphthong in syllable-coda position (Itabashi 2001), and makes the same claim for Sakhalin Ainu (Itabashi 2001) and Kuril Ainu (Itabashi 2001). Shibatani asserts that diphthongs such as ai, ui, and au involve devocalization of the second vowel, yielding pronunciations [aj], [uj], and [aw] (Shibatani 1990). Satō asserts that y and w can occur in syllable-coda position as well as onset position (Satō 2008: 10), and that words such as nay and haw are monosyllabic, with y and w pronounced lightly and briefly following the vowel (Satō 2008: 11).
generated 10 claims, 1 sentences without a claim ·
Onset glides
What is the phonological status of word-initial and post-consonantal y and w in Hokkaido Ainu?
Itabashi asserts that in Kuril Ainu, word-initial /y/ hardly ever occurs, and when it does, other records almost always transcribe it with another vowel (Itabashi 2001). Satō asserts that although ku-ipe can be theoretically assumed as the base form, its actual pronunciation is ku-ype (Satō 2008).
generated 2 claims ·
Marginal sequences: yi, wi, wu
Which onset-vowel sequences are absent from monomorphemic roots in Hokkaido Ainu?
Shibatani asserts that the semivowels /w/ and /y/ do not occur before [u] and [i] respectively, so the sequences [wu] and [ji] do not occur in Ainu (Shibatani 1990).
generated 1 claims ·
Boundary-derived occurrences
Under what conditions do the sequences [ji] and [wu] appear in Hokkaido Ainu?
Murasaki reports several boundary-derived alternations in Sakhalin Ainu: initial y- is deleted following a coda -n (Murasaki 2025: 89); initial y- drops after -m or -s, and initial w- assimilates to a preceding -m (Murasaki 2025: 89); and initial y- alternates to -n- after coda -h (Murasaki 2025: 89). Nakagawa asserts that when another element precedes a word beginning with vowels i or u, vowel weakening can occur, shifting i → y and u → w (Nakagawa 2024: 40), and that this process is frequently observed in the Saru and Chitose dialects (Nakagawa 2024: 40).
generated 5 claims ·