885096000

lin’s avcraging procedurę in the variational form is used so as to derive the one-dimensionai goveming equations of torsional motions. The gov-eming equations that are expressed both in local and variational forms incorporate all the significant effects of motion; the warping of cross section, the geometrical nonlinearity, and the fluid effect by the added mass concept of ship arc all taken into account. Further, (3) special cases are discussed and, in particular, the fully linearized goveming equations are studied. The sufficient conditions are enumerated for a unique solution of the linearized goveming equations by use of the logarithmic convexity argument.

WEDNESDAY AFTERNOON, 1 MAY 1991    INTERNATIONAL B, 1:30 TO 4:30 P.M.

Session 5SP

Speech Commumcation and Psychological and Physiological Acoustics: SP and PP Potpourri

(Poster Session)

Doug H. Whalen, Chair

Haskins Laboratories, 270 Crown Street, New Haven, Connecticut 06510

All posters will be on display from 1:30 to 4:30 p.m. To allow contributors an opportunity to see other posters, contributors of odd-numbered papers will be at tłieir posters from 1:30 to 3:00 p.m. and contributors of even-numbered papers will be at their posters from 3.00 to 4:30 p.m.

5SP1. Acoustic target zon es for naturally produced vowels in running speech. Frank E. Kramer, Donald J. Meyer, Marios Fourakis, and James D. Miller (Central Inst. for the Deaf, 818 S. Euclid Ave., St. Louis, MO 63110)

Acoustic measures of 2304 vowel tokens of Midwestem American English were collected by one of us (Fourakis). The vowels (i, i, e, ae, u, d, u, u, a] occurred in [bVd] or [hVd] syllables imbedded in sentences spoken with different rates of speech and different stress pattems. There were four małe and four female speakers. Two methods of automatic classification of this corpus of acoustic measures into vowel categories are evaluated. One method derives from parametric statistical theory and has been used often in studies of speech sounds. This method is linear discriminant analysis (LDA). The other method assigns each point in the acoustic space a most likely “vowel classification" by a "spatial windowing” procedurę. Then an edge-detection algorithm au-tomatically finds the edges of regions with similarly classified interior points. This is called the target-zone meth<xl (TZM). Both methods are applied completely automatically to the acoustic descriptions of the vowels. When the vowels are described acoustically in terms of Miller's auditory-perceptual space (J. D. Miller, J. Acoust. Soc. Am. 85, 2114-2134 (1989)), the target zones automatically generated by TZM classify about 85% of the tokens correctly. The LDA method does not achieve this level of performance.

5SP2. Effects of experimental manipulatioos of auditory feedback upon vowels. Emily A. Tobey, Dawn Cooper, Heidi Swilzer (Commun. Sci. Lab., Dept. of Commun. Disord., LSU Med. Ctr., New Orleans, LA 70112), and Mario Svirsky (MIT, Boston, MA)

The role of auditory feedback upon intermediate vowe! production was explorcd by manipulating the amount and type of information pro-vided by a Nucleus multichannel cochlear implant. Four experimental • conditions.were examined: (a) two types of control conditions, the implant turned on versus off, and (b) two experimental conditions that varied the channels being stimulated. Twcnty repetitions of the vowels [l], [r], [i], [u] in an hVd context were acquired from four adventitiously deafencd adults using multichannel implants. Two of the vowels, (i) and (e], were selected because of previous work indicating significant changes in formant frequencies when the implant was turned on veisus off. The vowels (i) and [u] were included as Controls sińce previous work indicatcd that their formant frequencies failed to change when the implant was turned off. Recordings were digitized at a 20-kHz ratę and analyzed using CSpeech. Fundamental frequency, formant frequencies, and duration measures were acąuired for each token. Data were statis-tically treated using a repeated measures ANOVA. Preliminary data indicates that the intermediate vowels rely upon re!evant auditory in-formation morę hcavily for their production than the control, point vowels. [Work supported by NIDCD.) 5SP3. Intelligibility in competing noise by males with partial laryngectomies. Be mice Gerdeman, M. Trudeau, D. Pearl, and R. Wilhelms (Speech and Hcaring Sci., Ohio State Univ., 110 Pressey Hall, Columbus, OH 43210)

Although Ievels of intelligibility among alaryngeal speakers using various speech modes have been invcstigated [J. G. Clark and J. C. Stemple, J. Speech Hearing Res. 25, 335-338 (1982); M. Kalb and M. Carpenter, J. Speech Hearing Disord. 46, 77-80 (1981)], little data are available for speakers with partial laryngectomies. The specific research questions of the investigafion were "To what dcgrcc do ihc paiicnls with hemilaryngectomy or supraglottic laryngectomy maintain intelligibility when speaking in various levels of competing noise?" and “Are there differences in speech intelligibility as a function of whether the partial laryngectomy was in the vertical or horizontal piane?” Ninę hcmilaryn-gectomees, nine supraglottic laryngectomees, and nine normal laryngeal małe speakers recorded 20 SPIN test sentences with no competing noise and with 75 dB A whitc noise. Intelligibility scores (identification of each sentence’s finał word by 30 normal-hearing listeners) were analyzed using logistic regression and indicated that the two surgical groups did not differ significantly from each other in either noise con-dition and that in both conditions the two groups differed significantly from the normal laryngeal group.

1934

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

121 st Meeting: Acoustical Sooety of America

1934