Chapter 29Glide Epenthesis and Vowel-Hiatus Resolution

29.1 Vowels at morphological boundaries

Combining Ainu forms often brings two vowels together. The resulting pronunciation may include a glide, a glottal closure, or the loss of a vowel. Some sequences remain as two adjacent vowels. The outcome depends on the forms being combined and on the variety being described Shiraishi (2022: §4.3); Okuda (2025: 40–44).

Two glide processes are especially important. In insertion, a glide appears between vowels that remain in the form. In glide formation, also called vowel weakening, an existing /i/ or /u/ becomes /y/ or /w/. These processes change the syllable structure differently Bugaeva (2012: 468–469); Nakagawa (2024: 34, 40).

29.2 Phonetic transitions and phonological glides

Nakagawa treats the [j] heard in [ijomante] as a transition between vowels in the phonological form /iomante/. His spelling consequently omits it. Similarly, the first [w] of [uwenewsar] is transitional in his account, while the /w/ inside newsar belongs to the word's phonological form Nakagawa (2024: 34–35).

Satō's Chitose analysis assigns some inserted glides phonological status, because their presence depends on morphological structure in addition to the adjacent vowels Satō (2006a: 47–49). Okuda compares the two approaches explicitly Okuda (2025: 42–43) ‹contested›. A glide can be introduced by a phonological rule without being part of the deepest underlying form; “inserted” and “phonemic” therefore refer to different questions in this comparison.

29.3 Insertion after particular prefixes

Satō's Chitose descriptions include /y/ after prefixal i- and /w/ after reciprocal u- before a vowel. His examples are iyomare “pour liquor” and uwomare “gather” Satō (2008: 13). In both, the prefix vowel remains before the inserted glide.

The Chitose trigger set also includes si-. Satō illustrates si-y-osmak “behind oneself” in his later discussion Satō (2008: 248–249). His published comparison distinguishes Chitose insertion after this prefix from its absence after the corresponding prefix in Horobetsu Satō (2006a: 49).

29.4 When vowel weakening is blocked

In Saru and Chitose, a stem-initial /i/ can become /y/ after certain prefixes. Nakagawa gives ku=ipe pronounced kúype “I eat,” and e=itak pronounced éytak “you speak” Nakagawa (2024: 52). The former vowel becomes the coda of the preceding syllable.

A closed initial stem syllable has a different outcome. In inkar “look,” converting /i/ to /y/ would create an impermissible consonant sequence. Nakagawa therefore gives a retained vowel:

Nakagawa gives the same blocking pattern with ikka “steal” Nakagawa (2024: 53). This condition concerns a process that removes a vowel nucleus by turning it into a consonant. Insertion preserves that vowel and has its own morphological conditions.

29.5 Vowel weakening and accent

Nakagawa also gives ko-itak becoming koytak “speak to,” and e-ukoytak becoming ewkoytak “discuss something.” He identifies these as common patterns in Saru and Chitose, leaving their wider dialect distribution open Nakagawa (2024: 40).

Satō relates weakening to the accentual effect of the preceding prefix. Alongside iyomare and uwomare, he gives ku-yomare “I pour liquor” and ku-womare “I gather” Satō (2008: 13–14). His introductory account also preserves the vowel in o'ika “go over,” whose stem-initial /i/ already bears accent, and notes that /u/ weakening is less regular than /i/ weakening Satō (2008: 13–14).

The contraction of ku= to k= and ci= to c= before certain vowels is a further process: the prefix loses its vowel. Nakagawa distinguishes this from stem-vowel weakening and identifies its dialect limits Nakagawa (2024: 52–53). See Chapter 31 (Person Markers: Contraction, Accent, and Suffix Boundaries).

29.6 Glottal realization between vowels

A vowel sequence can also involve glottal closure or tension. Shiraishi's survey reports clearer glottal realization in some accented or closed following syllables, alongside individual variation Shiraishi (2022: §4.3). Audible closure is therefore not a necessary feature of every token lacking a glide.

Okuda discusses alternative representations in terms of a glottal consonant or a syllable boundary Okuda (2025: 43–44) ‹contested›. The dispute concerns the analysis of the observed contrasts, treated in Chapter 20 (Glottal Stops and the Distribution of /h/).

29.7 Contraction and vowel deletion

Published descriptions also record shorter forms resulting from vowel contact. Shibatani gives kera plus an as keran “be tasty,” and ine-an-kur as inankur “which person” Shibatani (1990: §2.4, 13). Kindaichi and Chiri present corresponding contraction patterns in their older account Kindaichi & Chiri (1936: §V.10, 6–7).

The first comparison brings identical vowels together; the second loses the first vowel of an unlike pair. These are documented outcomes in particular forms. The prefix restrictions described above show why a single deletion rule cannot replace the separate morphological patterns.

29.8 Comparing dialects and analyses

Okuda's Shizunai observations add a lexical dimension. He reports a distinction in Orita Suteno's speech between [suʔat] “pot bail” and [suwat] “pothook,” crediting the discovery to 志賀雪湖. Some realizations become less distinct through weakening Okuda (2025: 41).

The same speaker's forms also distinguish lexicalized expressions from transparent combinations. Okuda reports a glide in the narrative term uwepeker, but its absence in the transparent expression for “talk together,” and no inserted [j] in iomante Okuda (2025: 41–43). He relates these differences to lexical integration.

The comparison follows Okuda's table Okuda (2025: 43). It involves both different pronunciations and different representations of a pronunciation ‹contested›. Refsing's broader Shizunai insertion description and Shibatani's general hiatus rules provide earlier accounts, but do not set out all these later distinctions Refsing (1986: §10.5, 72–73); Shibatani (1990: §2.4).