Abstract:

Cross-language similarities and differences in the acoustic variability of vowels as a function of speaking ``style'' (citation versus sentences) and phonetic context were explored by comparing the productions of four adult male native speakers each of American English (AE) and Japanese. Multiple instances of the 11 AE vowels /i, (small capital eye), e(small capital eye), (cursive beta), (ae ligature), (open aye), (inverted vee), (open oh), o(inverted cap omega with diaresis), (inverted cap omega with diaresis), u/ andthe 10 Japanese vowels /i, ii, e, ee, a, aa, o, oo, (inverted em), (inverted em)(inverted em)/ produced in citation-form bisyllables in /hVba/ and in CVC syllables /bVb, bVp, dVd, gVg, gVk/ imbedded in a carrier sentence were analyzed. Formant values at three temporal locations within the vocalic nucleus (25%, 50%, 75%) of the CVC syllables were compared with ``canonical'' /hVba/ values to determine the amount of ``target undershoot'' and changes in dynamic formant structure as a function of consonantal context. Vocalic durations were measured to determine the extent to which speaking style and consonantal context influenced relative vowel length information in the two languages. Linear discriminant analyses were conducted to quantify the acoustic overlap within and across languages. Implications of these results for cross-language studies of vowel perception will be discussed. [Work supported by NIDCD and ATR.]