885095955

whether thc echo sound came from the loudspcaker at 35* or 55* righl by pressing one of two buttons; correct feedback was provided on every trial. This test noise burst was presented in isolation, or preceded by several conditions. Lead ouly had a train of nine noise bursts from the Icadirig loudspcaker preccding thc test noise burst; Lag only had a train from the lagging loudspcaker at 45* right; Precedence effect (PE) had a train from both leading and lagging loudspeakers at 45* left and right.

The PE condition had the highest echo threshold (14.9 ms), followed by the isolated burst (11.2 ms), with the lowest threshold for Lead only (6.8 ms). Increased threshold uuder the PE condition indicated that repetition of the echo during the train was necessary to drive up echo threshold comparcd to the isolated burst. The single source train ap-pcared to enhance the test burst echo following the train. (Work sup-ported by NSF.J

FRIDAY MORNING, 3 MAY 1991    INTERNATIONAL B, 8:00 A.M. TO 12:00 NOON

Session 8SP

Speech Communication: Consonant and Vowel Perception

Sally G. Revoile, Chair

CASS/MTR, Gallaudet Uniuersity, Washington, DC 20002

Contńbuted Papers

8:00

8SP1. Thresholds for formant-freąuency discrimination of vowels in consonantal context. Dianę Kewley-Port and Charles S. Watson (Dept. of Speech and Hearing Sci., Indiana Univ., Dloomington, IN 47405)

The discrimination thresholds for shifts in formant frequency were shown to bc in the rangę of l%-2%, in a recent report to this society [Kewley-Port, J. A co ust' Soc. Am. Suppl. 1 87, SI 59 (1990)]. Thresholds for fi and FI obtained from well-trained subjects listening to vowcls under minimal stimulus uncerlainty were a factor of 3 lower than earlier estirnates. The present experiment extends that study to examinc the effects of placing a vowei in a consonantal context. The vowel /i/ was synthesized in CVC syllables for the consonants /b/, /d/, /g/, /z/, /m/, and A/. For /•1=450 Hz, thc threshold, AF, was thc same for isolated /l/ as the AF averaged over all CVC contexts, about 12 Hz. For 77=2300 Hz, AFwas significantly larger (45 Hz) for the vowel in the average CVC context than in isolation (25 Hz). Thresholds for individual CVC’s were significantly difłerent from the threshold for isolated /i/, in about one-half of the cases examined. These differences are discussed in terms of the cxtent of thc formant transitions and thc durations of the steady-state vowel formants. (Research supported by NIH and AFOSR.]

8:15

8SP2. Are articulations integrated in the perception of vowel height? John Kingston (Linguistics Dept., South College, Univ. of Massachusetts, Amhcrst, MA 01003)

Using the Ganncr paraidgm (W. Gamer (1974)], Kingston (Pho-netica (in press)] demonstrated that the acoustic effects of differences in velum height (the frequency separation of the nasal pole and zero = nasalization) and ratę of vocal fold vibration (fundamental fre-quency) which covary with tonguc height in vowels are integrated per-ceptually with the acoustic effect of that articulation (first formant frequency), perhaps because they exaggerate the perceptual value of the latler articulation. The failure to separate perceptually the acoustic effects of these three articulations challenges the claim of direct realists (e.g., C. Fowler, J. Phonet. 14, 3-28 (1986)] that articulatory gestures are the objects of speech perception, but in only a limited way, sińce the stimuli were brief and simple enough that they may not have allowed listeners to attribute these acoustic effects appropriately to their articulatory sources. Experiments are currently in progress to test whether similar perceptual integration occurs even when other aspects of the stimuli would allow the acoustic effects to be attributed to coarticulation (R. Kraków et ai, J. Acoust. Soc. Am. 83, 1146-1158 (1988)], e.g., is nasalization integrated with first formant frcquency in nasal as well as orał consonant contexts? Integration will also bc tested morę rigorously than in the earlier work, using preccpts of signal detection theory.

8:30

8SP3. Assessing the role of F0 in vowel perception via linear logistic modeling. T. M. Nearey (Dept. of Linguistics, Univ. of Alberta, Edmonton, Alberta T6G 2E7, Canada) and J. E. Andruski (Brown Univ., Providence, RI 02912)

Logistic models provide powerful tools in evaluation of stimulus-response relationships in speech perception (T. Nearey, J. Phonet. 18, 347-373 (1990)]. Excellent fits result when such models are applied to vowel perception data. These models allow insight into the possible normalizing role of FI) in vowel perception. If (as in many current perceptual accounts) thc role of F0 is restricted to an additive formant normalization factor (for sonie nonlinear transform of the frequency axis), then optimized logistic models using formant frequencies and the fundamental as predictors should show certain spccific pattems of cor-relation among certain estimated parameters across vowel categories. Preliminary results from the analysis of data in these laboratories indi-cate that this is indeed the case. Furthermore, FO normalization appears to occur nearly independently in “head” and “taił” sections of “hybrid” syllables, wherc, e.g., the ending portion of a syllable from a małe speaker is spliced (following a period of silence) to the beginning portion from a female. (Work supported by SSHRC.]

8:45

8SP4. An approach to the classification of American English diphthongs. Michael Gottfried (Central Inst. for the Deaf, 818 S. Euclid Ave., St. Louis, MO 63110)

1996

J. Acoust. Soc. Am., Vol. 89, No. 4, Pt. 2, April 1991

121st Meeting; Acoustical Society of America

1996