885095994

occurrencc of some coarticulatory cflfccts of thc prcccding phoncme in the conversation modę Chan in che text-reading modę.

9:00

4SP5. Voiced versus voiceless inaspirates: Acoustie measurements of English stops. Laura L. Koenig, Arthur S. Abramson, and Leigh Lisker (Haskins Labs., 270 Crown St._f New Haven, CT 06511-6695)

Accention is focused on inCervocalic stops in trochaic words. Em-bedded in semantically plausible sentences, cokens of six pairs of words like robing and roping were recorded by three speakers of American English. These included only places of articulation, labial and dorso-velar, for which che voicing opposition is reliable for all dialccts in natural speech. Four acoustie properties were examined because they are considered relevant and they are amenable to cditing for perceptual experiments. They are closure pulsing, closure duration, stressed-vowel duration, and release-burst amplitudę. The only entirely reliable corre-late is presence versus absence of closure pulsing. A fairly good correlate is vowel duration. Vowels before voiced stops are somewhat longer, although there is overlap between the two ranges. Another fairly good correlate is closure duration. VoiceIess elosures are somewhat longer, although herc, too, therc is overlap. The amplitudę of the burst is a poor correlate, even though voiceless bursts tend to be morę intense. These data furnished a baseline for our paper on perception. [Work supported by NIH Grant HD-0I994.J

9:15

4SP6. Voiced versus voiceless inaspirates: The perception of English stops. Leigh Lisker, Arthur S. Abramson, and Laura L. Koenig (Haskins Labs., 270 Crown St., New Haven, CT 06511-6695)

Preliminary data on four acoustie variables as possible cues to the contrast between voiced and voiceless unaspirated stops in fluent American English speech were reported at the Il9th Meeting. They were burst amplitudę, duration of the preclosure vowel, duration of the closure interval, and the naturę of the closure signal. A single English speaker recorded the sentence pair He’s knoion as Jack the Ribber and He’s known as Jack the Ripper, which servcd as sources of waveform-edited stimuli that were jdentified by a jury of ten American English speakers. Additional word identification data have sińce been collected for three additional sentence pairs recorded by the same single speaker, with the new word targets chosen to include dorsal as well as labial stops after both the short vowel [i] and the long vowel (oj. For each acoustie variable two values were tested, one norma! for thc word as originally produced and the other typical of the minimally contrasting word. The data allow us to rank the four acoustie properties in respect to their roles in perception: burst amplitudę < vowel duration < closure duration < closure signal. [Work supported by NIH Grant HD-01994 to Haskins Labs.]

9:30

4SP7. lite production of “breathy voice.^w Katherine Davis (Dept. of Linguistics, Morrill Hall, Comell Univ., Ithaca, NY 14853)

The naturę of voicing in the Hindi **voiced aspirated” phonemes has been a subject of discussion in the literaturę for a number of years. In particular, the question of whetheror not “breathy voice” or “murmur” is present during the aspirated portion of the stop has been at issue. The present study looks at productions of the velar voiced aspirated stop (/gh/) in initial position from the speech of ten adult female speakers of Hindi. Amounts of lead time (prevoicing) and lag time (post-release VOT) in the voiced aspirate were compared to lead and lag times in the voiced unaspirate (/g/) and thc voiceless aspirate (/kh/). In addition, the presence or absence of voicing during the post-release portion of the stop was determined. Both lead and lag time values of the voiced aspirate were found to be significantly different from those of the other velar stop types, and the presence of voicing during aspiration (breathy voice) was found to be speaker dependent. Implications of these find-ings for a description of voicing contrasts in various languages will be discussed.

9:45-10:00

Break

lOtfO

4SP8. Production and perception of vowel duration as a cue to the word-final English /t/-/d/ contrast by native and Chinese subjeets. J. E. Flege (Univ. of Alabama at Birmingham, Box 503, UAB Station, Birmingham, AL 35294)

Vowel duration was measured in seven minimally paired English words such as heai-beadand bat-badasspoken by native English (NE) subjeets and four groups of Chinese subjeets (ten per group). The Chinese subjeets were either relatively inesperienced (MA, TA) or experi-cnced (TB) adult learners of English, or else had learned English as an L2 before the age of 12 years (TC). The early learners in TC closely resembled the NE speakers in making vowels much longer before /d/ than /t/. The inexperienced adult learners (MA, TA) produced a voic-ing eflfect, hut one that was far smaller than that of the subjeets in NF. and TC. The experienced adult learners in TB produced a significantly larger voicing eflfect than the inexperienced subjeets in TA and MA, but their eflfect was also smaller than that of the subjeets in NE and TC. Sensitivity to vowel duration as a perceptual cue to the English /t/-/d/ contrast was invcstigated in three ways. The subjeets identified the

1917 J. Acoust. Soc. Am., Vol. 89. No. 4, Pt. 2. April 1991 members of natural continua in which vowel duration was varied by deleting glottal pulses. One continuum ranged from bead to beat, the other from bad to bat. The subjeets imitated the members of both continua, and also chose which members of the continua represented thc best exemp!ars of bead, beat, bad, and bat. As expected from the production data, the early learners in TC resembled the NE speakers per-ceptually, whereas the inexpcricnced adult lrarners (MA, TA) showed little evidence of perceptual sensitivity to vowel duration. Some experi* enced adult learners in group TB, on the other hand, were perceptually sensitive to the vowcI duration cue. The relationship between their production and perception will be discussed.

10:15

4Sł*9. VOT values fn bilingtials ver»i» monolinguals of Hindi and English. Carol E. Carey, H. S. Gopal, and Heidi Affentranger (Dept. of Speech and Hearing Sci., Univ. of California, Santa Barbara, CA 93106)

This study compared the VOT values of stop consonanis produced 121 st Meeting: Acoustical Society of America 1917