885095959

11:30

11:45

8SP14. The role of transition duration in the perception of speech and nonspeech analogs of place of articulation. Renće A. E. Zakia (DMLL, Morrill Hall, Comell Univ., Ithaca, NY 14853) and Richard N. Aslin (Univ. of Rochester, Rochester, NY 14627)

A previous report [R. Zakia, J. Acoust. Soc. Am. Suppl. 1 87, SI 17 (1990)) demonstrated that at a formant frequency pattem ambiguous between alveoiar and velar, subjects identify synthetie speech slimuli with longer transition durations as vclars and stimuli with shorter transition durations as alveolars. These results suggest the operation of ei-ther (1) an articulatory property characteristic of velars (longer transition durations distinguish velars because of their slowcr articulatory rclcasc), or (2) differcntial perceptual sensitivity to transition duration for spectral pattems characteristic of velars. To evaluate these possibil-ities, nonspeech analogs of a formant pattern ambiguous between an alveolar and a velar were generated with transition durations ranging between 20 and 50 ms in 5-ms słeps. In contrast to the previous iden-tification task, the nonspeech stimuli were presented in a same-different discrimination task. For comparison purposes, this same-difTerent task was also conductcd with the speech stimuli. Transition duration was an effective cue to discrimination of both speech and nonspeech stimuli, suggesting that the link to articulatory mechanisms is not essential. However, discrimination performance for the speech stimuli was gen-erally poorer than for the nonspeech stimuli, suggesting that the link to articulatory mechanisms is not essential. However, discrimination performance for the speech stimuli was generalły poorer than for the nonspeech stimuli, suggesting either that within-category judgments are morę difficult to make for broadband signals or that perception of the speech signals as phonetic segments interferes with auditory discrimination.

8SP15. The duration and perception of English epenthetic and underlying stops. Sook-hyang Lee (Dcpt of Linguistics, Ohio State Univ., Columbus, OH 43210)

In American English. an intrusive stop occurs before the fricative in words such as tense and false, making them very much likc words with underlying stops, such as tenis and faults. Ohala (1975) treats the in-serted stop as an artifact of universal physiological or aerodynamic constraints. But this approach cannot account for the fact that South African English speakers do not insert the stop between sonorant and fricative clusters (Fourakis and Port, 1986). Another approach posits a languagc- or dialect-spccific phonological rule which inserts a phono-logical segment (Zwicky, 1972). Fourakis and Port (1986) argue against this approach on the grounds that in some pairs the intrusive stop is significantly shorter than the underlying one (although the dif-ference is always very smali). This paper presents perception data and duration measurements supporting Zwicky*s approach. Phrases with intrusive and underlying stops (intense and in tenis, respectively) in citation forms produced by three speakers of midwestern dialects were presented over earphones in random order for subjects to identify. Identification was very poor, just at chance lcvcl. Also, duration measurements of the silence gap between the /n/ and /s/ in these words show no significant difference, contrary to Fourakis and Port’s findings. Moreover, token judgments in the perception experiment show very poor correlation with the durations excepc for one speaker, implying that whatever duration differences there are might not be a crucial cue that listeners exploit for labeling the words with epenthetic and underlying stops.

FRIDAY MORNING, 3 MAY 1991

LIBERTY B, 8:25 TO 11:45 A.M.

Session 8 UW

Underwater Acousdcs: Signal Processing

John Glattetre, Chair

Norwegian Defense Research Establishment, P.O. Box 115, N-3191, Horten, Norway

Cłuur^łs Introduction—8:25

Contributed Papers

8:30

8UW1. Uncertain field signal Processing. Anthony M. Richardson and Loren W. Nolte (Dept. of Elect. Eng., Duke Univ., Durham, NC 27706)

Classical matched-field processing assumes a known ocean environ-ment and known array sensor positions. It is well known that source localization using this method can be quite sensitive to this knowledge. The uncertain field proccssor approach is based on signal detection and estimation theory and allows one to incorporate, a priori, the uncertain-ties in the sound-speed profile and/or the array sensor positions. This approach is illustrated for several specific situations for a deep ocean environment. (Work supported by ONR and North Carolina Supercom-puting Facility.)

1999 J. Acoust. Soc. Am.. Vol. 89. No. 4. Pt. 2. April 1991

8:45

8UW2. Parał lei simulation of broadband umbfguity surfaces. Peter C. Mignerey (Naval Res. Lab., Appl. Ocean Acoust., Washington, DC 20375)

A connection machinę was used to implement parallcl algorithms for the solution of the normal modę equation and for matched-field Processing over a large frequency band. The all-acoustic layered model consists of piecewise linear sound-speed and density profiles with a frec surface and an acoustic half-space below the bot tom. Individual proces-sors are dedicated to one particular modę with the set of processors spanning the frcquency band for numerous sound-speed profiles. Am-biguity surfaces as a function of rangę, depth, and freąuency were generated in parallel using a linear estimator. These irnages show consid-

121 st Meeting: Acoustica! Society of America 1999