The spectral dynamics of voiceless sibilant fricatives in English and Japanese.

摘要

Voiceless sibilant fricatives, such as the consonant sounds at the beginning of the English words sea and she, are articulated by forming a narrow constriction between the tongue and the palate, which directs a turbulent jet of air toward the incisors downstream. Thus, the production of these sounds involves the movement of a number of articulators, including the tongue, jaw, and lips; however, the principal method for analyzing the acoustics of sibilant fricatives has been to extract a single "steady state" interval from near its temporal midpoint, and estimate spectral properties of this interval. Consequently, temporal variation in the spectral properties of sibilant fricatives has not been systematically studied.;This dissertation investigated the temporal variation of a single spectral property that denotes the most prominent psychoacoustic frequency, i.e. peak ERB-number. The dynamic aspects of peak ERB-number trajectories were analyzed with fifth-order polynomial time growth curve models. A series of analyses revealed a number of novel findings.;A comparison of English- and Japanese-speaking adults indicated that both language-internal sibilant contrasts are indicated by dynamic properties of peak ERB-number across the time course of the sibilants. A cross-linguistic comparison indicated that the peak ERB-number of a sibilant follows a language-specific trajectory.;Next, the development of the sibilant contrast in native English- and Japanese- acquiring children was investigated in terms of peak ERB-number trajectory. The English-acquiring children contrasted the sibilants in terms of similar aspects of peak ERB-number trajectory as the English-speaking adults. Moreover, the extent to which the children differentiated the sibilants increased with age. The analysis of the Japanese-acquiring children was complicated by an apparent developmental regression in the five-year-olds.;Effects of vowel context on peak ERB-number trajectory were examined in the English-speaking adults' and the English-acquiring children's productions. Both groups exhibited effects of vowel frontness on both sibilants: a following front vowel raised peak ERB-number across the full duration of these sibilants. Similarly, both age groups exhibited effects of vowel rounding on /s/: a following rounded vowel lowered peak ERB-number. The adults exhibited effects of vowel height on the dynamic aspects of peak ERB-number trajectory for both sibilants. Such effects were not found in the children. Furthermore, the extent of the children's vowel-context effects differentially weakened (vowel rounding and frontness) and strengthened (vowel height) developmentally.;Finally, the development of sibilant contrast was investigated in cochlear-implanted English-acquiring children. The children with cochlear implants were compared to children with normal hearing, who were matched on hearing age. The implanted children seemed to differentiate the onset and offset of frication in a more adultlike fashion than did the children with normal hearing. Furthermore, no group-related differences in the extent of sibilant contrast were found, suggesting that the children with cochlear implants differentiated the sibilants as well as the children with normal hearing.