Chapter 22Glides, Vowel Hiatus, and the Diphthong Question
Final /y/ and /w/ as codas, diphthong terminology, and competing analyses of glides between vowels.
22.1 Glides in syllables and at boundaries
Ainu /y/ and /w/ have the pronunciations [j] and [w]. They occur at syllable beginnings and endings, and glide-like transitions also occur between vowels. Two questions must therefore be distinguished: how a final glide fits into syllable structure, and whether a particular intervocalic glide belongs to the phonological form of a word Nakagawa (2024: 31, 34–35); Shiraishi (2022: §§3, 4.3).
Modern descriptions commonly treat final /y/ and /w/ as coda consonants. The treatment of glides between vowels is more disputed and depends on dialect and morphological structure Okuda (2025: 40–44) ‹contested›.
22.2 Final glides as consonants
Nakagawa analyzes aynu “person” as ay.nu and ohaw “soup” as o.haw. The final glide belongs to the preceding syllable and cannot itself bear an accent Nakagawa (2024: 31). Satō likewise treats nay “stream” and haw “voice” as monosyllabic Satō (2008: 10–11).
22.2.1 Evidence from nominalization
Shiraishi uses nominalizer selection as evidence for the consonant analysis. The shorter form -p follows a vowel, while -pe follows a consonant. His examples with final glides take -pe Shiraishi (2022: §3):
‘those who are’
Shiraishi 2022: §3; Hokkaido (dialect not further specified)
The nominalizer follows final /y/. English glosses are supplied here; Shiraishi's section does not specify a locality for these forms.
‘a bitter thing’
Shiraishi 2022: §3; Hokkaido (dialect not further specified)
The nominalizer follows final /w/. The lexical meaning follows the cited passage; English glosses are supplied here.
The glides thus pattern with other final consonants in this alternation. The nominalizing construction is discussed in Chapter 101 (P and Pe: Participants, Content, and Clause Endings).
22.2.2 Nominalization and number
The choice between -p and -pe in these examples concerns the stem's final segment. It does not itself mark a singular–plural opposition. Number distinctions in the predicate and the interpretation of the whole nominalized expression need separate analysis; the allomorphy supplies evidence about the final glide Shiraishi (2022: §3).
22.2.3 Syllables and accent
In the accentual varieties described here, a syllable ending in /y/ or /w/ is closed. This allows the usual accent rule to apply: áynu has initial accent, while kamúy has accent on its second syllable after an open first syllable Satō (2008: 15); Shiraishi (2022: §5.1). The distribution is treated in Chapter 23 (The Pitch-Accent Placement Rule and the Accented/Accentless Dialect Split).
22.3 Diphthong terminology
Older descriptions also call vowel-plus-glide sequences diphthongs. Shiraishi contrasts this terminology in Kindaichi and Chiri with the consonant analysis he adopts Shiraishi (2022: §3). Refsing describes final /w/ and /y/ as vowel-like elements forming diphthongs while listing them among the consonants Refsing (1986: §10.3).
The term can consequently refer to a phonetic sequence without deciding its structural analysis. A comparison between descriptions needs to establish whether “diphthong” means two vowel qualities in one syllable or a claim about the phonological nucleus.
22.4 What the coda evidence establishes
Syllabification and nominalizer selection give a direct basis for treating final /y/ and /w/ as consonants in the description used here. This evidence does not settle the status of every glide between vowels. The latter question requires comparisons in which the morphological boundary and dialect are known.
22.5 Glides at syllable beginnings
Onset /y/ and /w/ are the palatal and labial-velar glides. A glide heard between vowels may be a lexical consonant or a transition whose status is disputed. Nakagawa gives iwor “hunting ground” as a form containing phonemic /w/, separately from the transitions he leaves unwritten Nakagawa (2024: 34–35).
22.6 The sequences yi, wi, and wu
Descriptions commonly exclude /yi, wi, wu/ from simple roots while allowing some of them at morphological boundaries Nakagawa (2024: 44–45); Shiraishi (2022: §4.2). This is a restriction on a particular lexical domain, with consequences for how a combined form is syllabified.
22.6.1 Boundary-derived sequences
Shiraishi reproduces Tamura’s aw un mimtar “entrance space in a house” Tamura (1996: 36, via Shiraishi 2022 §4.2, example 3b) with a pronunciation joining /w/ to the following /u/. He also discusses a /y+i/ sequence across a boundary Shiraishi (2022: §4.2). Such data require the underlying elements and the surface syllables to be distinguished.
22.6.2 Competing analyses of intervocalic glides
Nakagawa treats many glide-like vowel transitions as phonetic and omits them from phonemic spelling Nakagawa (2024: 34–35, 40). Okuda compares this position with Satō's Chitose analysis, in which the presence of a glide depends on morphological structure and cannot be predicted from adjacent vowels alone Okuda (2025: 42–43). The sources therefore disagree about which glides belong to phonological representation ‹contested›.
The form meaning “thank” is a particularly informative case. Okuda argues for yayirayke in Chitose and Saru, with accent on the second syllable, against Nakagawa's yairayke analysis. He separately identifies yayrayke in the Shizunai speaker's variety Okuda (2025: 44). A single normalized spelling would conceal part of this dispute ‹contested›.
22.6.3 Shizunai evidence
Okuda reports a contrast in Orita Suteno's Shizunai speech between [suʔat] “pot bail” and [suwat] “pothook.” His interpretation assigns /w/ phonological status in the second word. He also notes that weakening of glides and glottal closure can make some phonetic realizations overlap Okuda (2025: 41).
This evidence prevents the blanket assumption that every glide between vowels is predictable from the vowel sequence. Its application to other forms and dialects remains a separate question, addressed in Chapter 29 (Glide Epenthesis and Vowel-Hiatus Resolution).
22.6.4 Interaction with person prefixes
Person prefixes can trigger semivocalization of a stem-initial vowel, changing /i/ to /y/ or /u/ to /w/. Okuda reviews Satō's proposal that this process and glide insertion apply at different levels of morphological derivation Okuda (2025: 42–43). Nakagawa uses some of the same kinds of forms to argue for a phonetic treatment of the transition. These are competing analyses of the interaction, rather than interchangeable spelling rules ‹contested›.
The relevant combinations are described in Chapter 31 (Person Markers: Contraction, Accent, and Suffix Boundaries) and Chapter 26 (Accent in Compounds and under Affixation and Cliticization).
References cited in this chapter
Nakagawa (2024) ·Okuda (2025) ·Refsing (1986) ·Satō (2008) ·Shiraishi (2022) ·Tamura (1996)