Knowledge base · Part III — Segmental Phonetics, Phonology, and Phonotactics · this topic as a network
20. The Laryngeals: Glottal Stop, /h/, and the Final-h Question
How do laryngeal consonants function in the language?
The glottal stop [ʔ]
What is the phonetic distribution of the glottal stop?
In Murasaki's phonemic analysis of Sakhalin Ainu, vowel-initial syllables are not recognized, and a weak glottal consonant phoneme ' is posited before all vowels (proposed in Murasaki 2025). This phoneme is realized as [ʔ] at syllable onset and as weak laryngeal tension intervocalically, yet /V'V/ and /VV/ remain distinctly contrasted (Murasaki 2025: 3). A morpheme-initial ' is deleted after -n, yielding the sequences -nV- or -niV- (Murasaki 2025: 89).
generated 3 claims ·
Phonetic distribution
Where does the glottal stop occur phonetically?
Vowel-initial syllables phonetically take an onset glottal stop across Ainu generally (Shibatani 1990, Shiraishi 2022: 13.2). In Hokkaido southwest dialects, /ʔ/ occurs syllable-initially when there is no other initial consonant (Bugaeva 2012: 464–465). In Sakhalin Ainu, vowel-initial syllables are analyzed as having an onset /ʔ/, yielding C1V, C1VV, or C1VC2 structures (Itabashi 2001), and the same analysis applies in Kuril Ainu, where such syllables are treated as C1V or C1VC2 (Itabashi 2001). The glottal stop [ʔ] is pronounced by breaking the transition between vowels or between a consonant and a vowel (Satō 2008). Nakagawa, however, asserts that the glottal stop does not distinguish lexical meaning word-initially, as in [e] versus [ʼe], or across consonant-vowel boundaries, as in [itakʼan] versus [itakan] (Nakagawa 2024: 32).
generated 6 claims ·
Phonemic status
Is the glottal stop a phoneme?
The phonemic status of the glottal stop is disputed. Nakagawa asserts that it is not an independent phoneme but a phonetic boundary marker indicating a syllable boundary before a vowel, and that it has no contrastive function in monomorphemic forms (Nakagawa 2024: 32, 33, Shiraishi 2022: 13.4). Refsing reports that Kirsten Refsing (1986) does not recognize the existence of the glottal stop in Ainu at all (reported in Nakagawa 2024: 31), while Refsing also reports that Asai (1969) recognizes glottal closure but does not treat it as a separate phoneme (reported in Refsing 1986: 68). By contrast, Refsing reports that Hattori (1964) posits a glottal plosive /ʔ/ before all word-initial vowels and intervocalically, treating it as a separate consonant phoneme ' (reported in Refsing 1986: 68), and Nakagawa reports that in the 1950s Hattori and Tamura similarly recognized the glottal stop as a phonemic consonant occurring before all vowels (reported in Nakagawa 2024: 31). Satō proposes that the glottal stop can be analyzed as a phoneme before word-initial vowels and between vowels, surfacing as throat tension or as an intervocalic vocal break (proposed in Satō 2008: 10). Nakagawa notes that recognizing the glottal stop as a phonemic consonant eliminates vowel-initial syllables and simplifies the syllable template of Hokkaido Ainu to CV and CVC (Nakagawa 2024: 31), but also argues that positing /'/ before all vowel-initial syllables is uneconomical because it requires deletion rules at various stages of word formation to account for pitch accent in words like parunpe (Nakagawa 2024: 44). The phonological status of the glottal stop remains unclear in the Chitose dialect (Bugaeva 2022: 20.2).
generated 9 claims, 1 sentences without a claim ·
Juncture gradation and prosodic integration
How does the glottal stop vary with prosodic boundaries?
In Hokkaido southwest dialects, /ʔ/ tends to weaken in intervocalic positions but does not weaken in accented syllables (Bugaeva 2012: 466–467), and in Saru Ainu specifically, the glottal stop is often omitted in the syllable immediately following an accented syllable (reported in Bugaeva 2012: 466–467). A glottal stop or tension is clearly audible when V2 bears accent and/or when V2 is followed by a consonant in a closed syllable (Shiraishi 2022: 13.4), whereas when V2 does not bear accent and is not in a closed syllable, no clear glottal stop is audible and the transition between the two vowels is gradual (Shiraishi 2022: 13.4). When the suffixes -as and -an attach to verbs ending in a consonant other than n, a glottal stop may be inserted before them, whereas after verbs ending in n they are generally pronounced continuously (Satō 2008: 111).
generated 5 claims ·
The Shizunai minimal pair
Does the glottal stop contrast with other sounds in any dialect?
In the 1950s, Shiro Hattori and Suzuko Tamura recognized the glottal stop as a phonemic consonant occurring before all vowels in Ainu (reported in Nakagawa 2024: 31).
generated 1 claims ·
The phoneme /h/
What is the phoneme /h/ and its distribution?
A sound change or alternation s → h occurs in certain words (Kindaichi & Chiri 1936). In Sakhalin Ainu, when morphophonemic /h/ is followed by a vowel or y-, an underlying plosive or liquid — -t-, -p-, -k-, or -r- — surfaces (Murasaki 2025: 89).
generated 2 claims ·
Onset distribution and allophony
How does /h/ behave in onset position?
The phoneme /h/ shows considerable variation across dialects and prosodic environments. In Hokkaido southwest dialects, /h/ is hardly distinguishable in intervocalic position, where it is frequently weakened and voiced to [ɦ] (Bugaeva 2012: 466–467), and before /u/ it is often realized as a voiceless bilabial fricative [ɸ] (Bugaeva 2012: 466–467); in Sakhalin Ainu, /h/ is [h] at syllable onset, sometimes voiced [ɦ] intervocalically, [ɦ]~[ç] after /i/ in coda, [ɸ] after /u/ in coda, and after other vowels a voiceless breath with the shape of the preceding vowel (Murasaki 2025: 2). More generally, the fricative h palatalizes to [ç] before i and labializes toward [F] before u, while remaining the same phoneme (Kindaichi & Chiri 1936). Cases of h-insertion are restricted to specific morphological contexts such as possessive formation (Bugaeva 2022: 20.6). Across word boundaries, an initial hV- sequence can drop h- after a consonant-final word (Satō 2008: 13), and particles and formal nouns beginning with h- or y- often drop the initial consonant and are pronounced continuously with the preceding word (Nakagawa 2024: 39). In the Shizunai dialect, the [h] of the nominalizer hi is usually elided in all but very slow speech (Refsing 1986: 233); initial h- of unstressed suffixes tends to disappear without exception in rapid speech and almost always in normal speech unless there is a pause or hesitation (Refsing 1986: 73); hine frequently drops word-initial [h], and in rapid speech [i] may also drop so that only [ne] remains (Refsing 1986: 242); and hike similarly tends to drop initial [h] in rapid speech (Refsing 1986: 242). In Hokkaido more broadly, in normal-speed speech the word-initial h- of hine is not pronounced, and i- often forms a single syllable with the coda consonant of the preceding word (Nakagawa 2024: 338). In dialects such as Tokachi, Eastern Hokkaido, and Ishikari, initial h has a strong tendency not to occur, especially at the beginning of unaccented words (Nakagawa 2024: 39), and in Tokachi, Kushiro, and Ishikari, unaccented word-initial h- regularly drops (Nakagawa 2024: 109); in the Ishikari dialect specifically, hine may be written as ine (Nakagawa 2024: 338). When accent falls on the second syllable because the first syllable is open, word-initial /h/ is often elided in geographically eastern dialects (Nakagawa & Fukazawa 2022: 8.3). Conversely, in the Urakawa and Samani dialects there is a tendency to add an initial h- to words that begin with a vowel in other dialects (Nakagawa 2024: 110), and insertion of word-initial /h/ is prominent in unaccented-type dialects in the northeastern part of Hokkaido (Nakagawa & Fukazawa 2022: 8.3). In Bihoro, initial syllables that had high pitch accent in Proto-Hokkaido Ainu developed a prothetic /h-/, while those with low pitch did not (reported in Alonso de la Fuente 2022: 5.2). Early sources of Kuril Ainu attest both prothetic h- and the loss of original h- (Alonso de la Fuente 2022: 5.2).
generated 20 claims, 1 sentences without a claim ·
Coda /h/ and the final-h question
Does /h/ occur in coda position?
The occurrence of /h/ in coda position is treated differently across dialects and analyses. Refsing asserts that /h/ is not found as a final consonant in Hokkaido Ainu, where it often becomes [ɸ] before /-u/ (Refsing 1986: 68). In Sakhalin Ainu, by contrast, syllable-final /-h/ after /i/ is phonetically neutralized with /-s/, fluctuating between [ɕ] and [ʃ] but usually closer to [ʃ] (Murasaki 2025: 3), and Nakagawa similarly notes that syllable-final -x (-h) in Sakhalin Ainu, when preceded by i, may be pronounced from a sound close to [çi] to a sound transcribed by small ㇱ (Nakagawa 2024: 64). Morpheme-initial h- is deleted after -n in Sakhalin Ainu, and word-initial h- can alternate with glottal stop plus vowel ('V-) (Murasaki 2025: 89).
generated 4 claims, 1 sentences without a claim ·