885095995

by bilingual speakers of Hindi-English with those produced by mono-lingual speakers of Hindi and English. Specifically, it investigated whether native Hindi speakers use their nativc VOT values of stop consonants when producing English stops, or if they use the correct/ appropriate VOT values as produccd by native English speakers. Two subjects in each of the three speaker groups (Hindi monoKngual, English monolingual, and Hindi-English bilingual speakers) produced CYC test syllables in a sentence contexL Th esc scntences were acousti-cally analyzed for the VOT values of the initial stop consonants in the CVC syllables. Results will be discussed within the framework of the equivalence classification hypothesis which claims that, when there are similar phones between two languages, bilingual speakers of those lan-guages will not produce the phones in the second language accurately as they will substitute the similar phone from their native language [J. Flege, J. Phon. 15, 47-65 (1987)].

10:30

4SP10. Factors influencing the spectral representation of front-back vowels in American English. H. S. Gopal, Joyce Manzella, and Carol Carey (Dept. of Speech and Hearing Sci., Univ. of Califomia, Santa Barbara, CA 93106)

It has been proposed that front vowels in American English have an auditory distance of less than 3 critical bands between the second and third formants, whcrcas, in back vowels, this distance is greater than 3 Barks (A. K. Syrdal and H. S. Gopal, J. Acoust. Soc. Am. 79, 1086-1100 (1986)]. The current study investigates the success of classifying front and back American English vowels into two separate classes using this auditory model given the variations in the acoustic signal duc to speaking ratę, post-vocalic coasonantal voicing, and different speakers. Sevcral researchers have shown that these factors influence vowel for-mant frequencies. Measurements were madę of vowel El, FI and dura-tions in /pVC/ target syllables containing one of four pairs of front-back vowels and voiced or voiceless stops or fricatives. Target syllables were produced in sentence context at three different speaking rates by two małe and two female native American English speakers. Results will be reported and discussed for an invariant auditory representation of front and back vowels in AE under these influences.

New York Univ., 719 Broadway, New York, NY 10003)

Most English unstressed vowcls are perceived as varying between schwa and /i/. To investigate this variation, which can be only partially explained by coarticulation with consonantal contexts, words containing such vowels in a wide variety of C __C environments were randomly selected from the conversation of a małe speaker of American English. These were c om pa red to the same words in the speaker’s reading of sentences from his conversation. Frequency levels of the first two formants were measured at three points within each vowel. A statistically significant percentage of word pairs demonstrated larger Fl/F\ ratios in the sentence-reading modę, particularly for formant values obtained at vowel midpoints. Because this result occurred with all coasonantal con-texts, it can be concluded that unstressed vowels are, in generał, morę /t/-like in formal speeech. Examination of the word pairs that did not conform to this predominant pattem revcaled unexpected correlations between formant frequencies of an unstressed vowel and two higher level linguistic factors: position in the word and proximity to a primary stress.

11:15

4SP13. Effects of speaking ratę on the vowel length distinction in Korean. Harriet S. Magen and Shcila E. Blumstein (Dept. of Cognitive and Linguistic Sci., Brown Univ., Box 1978, Providence, RI 02912)

Variations in speaking ratę aflfect the perception of tcmporally de-fined phonetic distinctions in consonants, such as voicing in stops, and the stop-glide and stop-affricate contrasts. The production data on these contrasts show asymmetrical effects on temporally distinctive pairs, with the longer members showing the effects morę markedly. These data further suggest that the shorter member of the pair serves as a phonetic anchor. This study investigates whether similar effects emerge for vow-els. To this end, production of short-long vowel pairs was examined across speaking rates in Korean, a language that has phonemic vowel length. Results show that both long and short vowels vary across speaking rates, such that the symmetrical effects found for consonants are not found for vowels. Short vowels do not provide a phonetic anchor: In fact, the durations of short vowels produced at a slow ratę nearly always overlap those of long vowels produced at a fast ratę. [Work supported by NIH.]

10:45

4SP11. A comparison of the first and second formants of vowels common to English and French. Dawn M. Behne (Speech Res. Lab., Dept. of Psychol., Indiana Univ., Bloomington, IN 47405)

English vowels have been characterized as being lower, morę central, and less rounded than the predominantly high, front, rounded vowels of French [P. Dclattre, Comparing the Phonic Features of English, French, German and Spanish: An Interim Report (Groos Verlag, Heidelberg, 1965)]. In the present study, Delattre's description is ex-perimentally investigated by comparing the first and second formant frequencies of English and French vowels that occur in both languages (i.e., /i,e,t,u,o,D,o/). The results suggest that: (1) high vowcls tend to be higher in French than in English; (2) /n/ tends to be lower in English than in French; (3) high and mid vowcls tend to be morę central in English than in French; (4) rounded vowels tend to have greater liprounding in French than in English; (5) front vowels are higher and morę fronted in French than in English; and (6) back vow-els have greater pharyYigeal constriction in English than in French. Although Delattre’s description of English and French vowels is gen-erally supported by the results, it does not fully characterize the differ-ences demonstrated between the vowe!s common to English and French.

11:00

4SP12. Formant frequencies of lexically unstressed English vowels in two speech modes. Karen L. Wallace (Dept. of Linguistics/ISALS,

1918 J. Acoust. Soc. Am., Vol. 89, No. 4, Pt. 2. April 1991

11:30

4SP14. Vowel effects of pharyngeal and uvular consonants in Interior Salish. Nicola J. Bessell (Dept. of Linguistics, Univ. of British Columbia, 4616 Larchwood Ave., Philadelphia, PA 19143)

Formant values of the vowels /i/, /u/, /a/, and schwa in the envi-ronment of the pharyngeal and uvular consonants of several Interior Salish languages are plotted and the results arc compared with the formant values of the same vowel phonemes in nonuvular/pharyngeal environments. The data represent several speakers of each language. Results show a consistent lowcring of vowcls in uvular/pharyngeal en-vironments, with El rising considerably. The rise in El is greatest for vowels in pharyngeal cnvironmcnts. Backing (or E2 lowering) is not consistent across vowels or place of articulation. In fact, the voicelcss pharyngeal shows some tendency to front the Iow central vowel /a/. The data arc compared to similar materiał from several Arabie dialects.

11:45

4SP15. Study of vowel coarticulation in British English. James L. Hieronymus (Ctr. for Speech Technol. Res., Edinburgh Univ., 80 South Bridge, Edinburgh, Scotland)

Coarticulation in continuous speech causes the vowel formant tracks to be altered by nearby phonemes. The present study concentrates on 660 phonetically hand-labelcd read sentences from one małe talker of the RP accent of British English. The 14 monothongal vowels of RP

121 st Meeting: Acoustical Society of America 1918