sures alone were able to classify place with 85% accuracy, showing that automatic measurements can reproduce the result of Blumstein and Stevens [J. Acoust. Soc. Am. 64 (1978)]. With seven measures, the error rate was only 10%. Most errors could be attributed to a strong labial burst and rapid formant transitions in /i/ or /r/ context, or coloration of an alveolar burst by back-cavity coupling. [Supported by a grant from NIDCD.] 2aSC3. Locus equations as phonetic descriptors of consonantal place of articulation. Harvey M. Sussman and Jadine Shore (This experiment tested whether locus equation coefficients, slope, and y intercept could serve as indices of place of articulation for obstruents sharing the same place of articulation across different manner classes. A previous study [C. A. Fowler, Percept. Psychophys. 55, 597-610 (1994)]reported that locus equations provided poor information for place because the mean slope for/d/(0.47) was found to be statistically different than the mean slope for/z/(0.42) across ten speakers. Locus equations from 22 speakers were derived from CV/t/words with initial voiced oral stop/d/, voiceless aspirated stop/th/, nasal stop/n/, voice fricative/z/, and voiceless fricative /s/ preceding ten vowel contexts. Post-hoc tests revealed /d/= /z/= /n/ for slope means. Slopes for/d/and/th/were also equivalent 2891

This experiment tested whether locus equation coefficients, slope, and y-intercept could serve as indices of place ofarticulation for obstruents sharing the same place ofarticulation across different manner classes. Locus equations for 22 speakers were derived from eVltl words with initial voiced stop Idl, voiceless aspirated stop It/, nasal Inl, voiced fricative Izl, and voiceless fricative lsi preceding 10 vowel contexts. Post hoc tests revealed Idl = Izl = Inl for slope means. Voiced Idl and voiceless It I were also equivalent when F2 transition onset measurement points were equated. Scatterplots of locus equation coefficients revealed three nonoverlapping and distinct clusters when the diverse coronal group was compared with labials and velars. A discriminant analysis using slope and y-intercept as predictors successfully categorized all five coronals into one alveolar group with 87.1% accuracy. The collective results support the contention that locus equations can serve as effective phonetic descriptors of consonant place of articulation across manner classes.Locus equations are linear regressions of the onset of F2 transitions on their offsets measured in the vowel nucleus (Lindblom, 1963). Frequency coordinates coding F2 transitions ofallophonic variants ofa given stop place category (e.g., beat, bet, bat, boot, but, boat, bought, bait, etc.) This orderly plot captures a higher order and emergentlike acoustic commonality-namely, a vowel normalization of the F2 transition within each stop place category. Higher order locus equation coefficients-slope and yintercept-when serving as predictor variables in a discriminant analysis, were shown to correctly classify labial, alveolar, and velar stop place of articulation with 100% accuracy (Sussman et al., 1991 the slope of the regression function is proportional to the extent ofCV coarticulation being used by the speaker (see Figure 10 in Sussman et al., 1993). The consistently linear scatterplots, as documented across languages, gender, and speaking conditions, have recently been interpreted as possibly serving a functional role in speech perception and the representational encoding (viz., mapping) of acoustic-based place of articulation categories (Sussman, Fruchter, & Cable, 1995).The acoustic correlates of consonantal place of articulation, however, are many and varied. Burst information, onset spectra, and formant transition motion have all been implicated, to greater or lesser extents, as perceptual cues for place ofarticulation (see, e.g., Blumstein & Stevens, 1979, 1980Dorman, Studdert-Kennedy, & Raphael, 1977;Eek & Meister, 1995;Furui, 1986; KewleyPort, 1983;Lahiri, Gewirth, & Blumstein, 1984; Nossair & Zahorian, 1991;Ohde, Haley, Vorperian, & McMahon, 1995;Walley & Carrell, 1983). Locus-equation-based analyses can offer additional insights into the role of the F2 transition onset-offset combination as a contributing cue for consonant place of articulation. Though the taxonomic value of locus equations to date has primarily been as a phonetic descri...

Syllable-final stops have different coarticulatory and perceptual properties compared to syllable-initial stops. Locus equations have previously been used to acoustically characterize place of articulation for initial stop consonants [H. Sussman, H. McCaffrey, and S. Matthews, J. Acoust. Soc. Am. 90, 1309–1325 (1991) and H. Sussman, K. Hoemeke, and F. Ahmed, J. Acoust. Soc. Am. 94, 1256–1268 (1993)]. Since locus equations also encode degree of coarticulation, they might also provide an adequate phonetic description of VC events for final /b/, /d/, and /g/ following varied vowel contexts. Each of ten speakers, five male and five female, produced three repetitions of 90 CVC tokens. For each final stop (/b,d,g/) there were ten medial vowels and three initial consonants—bVb, dVb, gVb, bVd, dVd, gVd, bVg, dVg, gVg. Three points along the second formant were measured—F2 onset (Hz), F2 midvowel (Hz), and F2 offset (Hz). Offset locus equations (VC) were generated for each syllable-final stop place, both across initial consonantal contexts and as a function of each initial stop. In addition, 3-D plots were generated to determine how F2 onset (x), midpoint (y), and offset (z) could acoustically capture and differentiate lexical contrasts. [Work supported by NIDCD.]