Stefanie Jannedy
Dept. of Linguist., Ohio State Univ., 222 Oxley Hall, 1712 Neil Ave., Columbus, OH 43210
Jun and Beckman (ICSLP94) explain vowel devoicing in Japanese and Korean in terms of gestural overlap [Browman and Goldstein, LabPhon1 (1990)]; the glottal gestures for preceding and following voiceless consonants overlap to a greater or lesser extent with the glottal gesture for high vowels. The devoicing of the four short high Turkish [i (barred eye) y u] can be explained by the same model which predicts that vowels are more likely to undergo devoicing if they are short and the adjacent voiceless consonants have strong and large glottal opening gestures. This study evaluates influencing factors such as preceding and following environment, rate, stress, and syllable type. Subjects read 210 words positioned utterance initially in a carrier phrase at three rates. Faster rates and lack of stress facilitated devoicing most. As in Japanese and Korean, there were more devoiced vowels before following stops than before following fricatives, since vocal fold vibration ceases gradually before fricatives, but abruptly before stops. Devoicing also occurred in utterance-initial position. This result is less easy to interpret in terms of the model, since only the following laryngeal gesture can overlap with the vocalic gesture. More devoicing occurred in open syllables, where vowels are significantly shorter in Turkish.