Chapter 17The Vowel Inventory and Vowel Realization

17.1 The five vowels

Hokkaido Ainu has five vowel phonemes, /a, i, u, e, o/ Nakagawa (2024: 26); Satō (2008: 10); Shibatani (1990: §2.1). The inventory distinguishes vowel quality without a phonemic opposition between short and long vowels. The phonetic realization of each vowel varies with speaker and context.

This diagram gives broad phonological categories. It does not imply identical vowel positions across speakers or precise equivalence with vowels in another language.

17.2 Phonetic quality

Shiraishi surveys acoustic studies beginning with Obata and Teshima's recordings of three male speakers, two from Asahikawa and one from Sakhalin. Later studies examined Saru and other Hokkaido speakers Shiraishi (2022: §2). The small, differently composed samples are useful evidence about particular speakers; they should be kept distinct in comparisons of vowel quality.

17.3 Devoicing

The vowels /i/, /u/, and sometimes /e/ can be devoiced between voiceless consonants. Shiraishi cites realizations of cise “house,” sutu “root,” and seta “dog” with a devoiced first vowel Shiraishi (2022: §2.1). Devoicing changes the production of a vowel; it need not remove the syllable from the word's phonological structure.

The frequency and distribution vary markedly. In Tamura's comparison of two female Saru speakers, one produced approximately twenty-five times as much devoicing as the other. Shiraishi's acoustic study of Saru recordings found it most readily in si and ci, with no instances in tu, pi, pu in that study Shiraishi (2022: §2.1). These findings describe the recordings examined; they do not justify an exceptionless rule for every Hokkaido speaker.

Japanese and Ainu devoicing overlap in some environments but differ in others. Shiraishi reports Tamura's observation that Japanese-speaking learners can extend devoicing to Ainu forms where it is inappropriate Shiraishi (2022: §2.1). Pronunciation should therefore be learned from the relevant Ainu recording as well as from its transcription.

17.4 Length and lengthening

A vowel may be long in an utterance even though vowel length does not distinguish Hokkaido words. A monosyllabic open word can be prolonged in isolation; accented vowels in longer words also tend to be lengthened. Shiraishi illustrates these contexts with to “lake” and poro “big,” among other forms Shiraishi (2022: §2.2).

Emphatic forms can have a different structure. The forms transcribed [weʔen] “very bad” and [toʔo] “far away” contain a glottal interruption and are analyzed as two syllables Shiraishi (2022: §§2.2, 4.1). They should be distinguished from a single vowel prolonged in performance; see Chapter 20 (Glottal Stops and the Distribution of /h/).

Sakhalin Ainu provides a comparative contrast because it has phonemic vowel length. Correspondences between Sakhalin length and Hokkaido accent are discussed by Nakagawa and Itabashi Nakagawa (2024: 26–27); Itabashi (2001). The comparison requires attention to the particular words and dialects; length and high pitch do not correspond uniformly in every case.

17.5 Vowel sequences and final glides

A sequence written with two vowel letters in an older transcription may be represented with a final glide in modern spelling. Shibatani writes the sequences pronounced [aj], [uj], and [aw] as ay, uy, aw Shibatani (1990: §2.1). Under the consonant analysis, the glide occupies the syllable coda. Morphological evidence for this treatment is considered in Chapter 22 (Glides, Vowel Hiatus, and the Diphthong Question).

17.6 Interpreting vowel counts

A count of written vowels depends on the transcription and the unit being counted. Counting each dictionary headword once gives a different measure from counting every word occurrence in connected speech. Person forms and common particles can strongly affect the latter measure. Any comparison of vowel frequencies needs to identify the sample and counting convention; see Chapter 9 (The Dialect Sample and the Corpus-Quantification Method).

17.7 Vowels in derivational patterns

Certain transitive-verb and possessive-noun formations show restrictions on the combination of the stem vowel and a following suffix vowel. Examples include kay-e “break” and mos-o “wake someone,” discussed in Shiraishi's review of Chiri's analysis Shiraishi (2022: §4.5). The restrictions concern particular morphological patterns, rather than the choice of vowels throughout the language.

The interpretation as vowel harmony is disputed. Shiraishi reviews objections by Shibatani, Vovin, and Satō: the patterns have limited morphological scope and do not fit a single familiar harmony feature such as backness Shiraishi (2022: §4.5) ‹contested›. The data and competing analyses are developed in Chapter 45 (Affiliative Suffix Patterns).

17.8 Dialect comparison

A shared inventory of five vowels allows differences in the realization or distribution of those vowels. Nakagawa and Fukazawa describe lexical correspondences involving /u/ and /o/ in Sōya and /o/ and /e/ in Yakumo Nakagawa & Fukazawa (2022: §3). Such correspondences should be distinguished from small phonetic differences within one vowel category. Claims about either require locality and speaker information.