The objective of this research program is to seek a better understanding of the empirical phenomenon known as """"""""locus equations."""""""" Locus equations are straight line regression fits to data points formed by plotting frequency onsets of F2 transitions along the ordinate and their midvowel target frequencies along the abscissa. Past studies have consistently shown that such coordinates emerge as linear scatterplots that are statistically distinctive across stop place categories for initial voiced stops. As such, locus equations capture, in acoustic terms, a lawful variability and relational orderliness characterizing coarticulated stop + vowel gestures. Locus equations, as quantitative phonetic descriptors capable of indexing both coarticulation and contrastive stop place categories, provide a new methodological probe to investigate the classic invariance problem in speech perception the ability of listener's to form invariant perceptual categories despite physical variation in the signal. The proposed studies are structured to experimentally assess the (l) physical, (2) perceptual, and (3) phonological reality of locus equations. This organizational framework is geared toward providing answers to questions such as: """"""""where do linear and contrastive locus equations come from?"""""""" """"""""do listener's make use of them in perception as """"""""CV prototypes?"""""""", and """"""""what is the phonological role of locus equations across the consonant inventory of a language?"""""""" The physical reality of locus equations will be investigated by three separate paradigms: (i) to ascertain if the source of the linearity exhibited by locus equations derives from inherent filtering properties of the vocal tract, regression functions will be simulated using a model that generates formant values based on vocal tract area functions; (ii) to assess whether linear locus equations for stop categories can be compensatorily generated, bite block speech will be employed to compare to normal production of CV tokens; (iii) to assess the developmental origin of locus equations a longitudinal study of infant babbling is proposed. The psychological reality of locus equations will be examined by a perception study using synthesized tokens formed from prototype locus equations. F2 and F3 onset frequencies will be systematically varied to study how perceptual judgements of category affiliation and boundaries are predicted by locus equation position in acoustic phonetic space. The phonological reality of locus equations will be investigated by testing several consonantal manner classes (stops, approximants, and fricatives), varying by place. Comparison of locus equations for a similar 'place' feature (e.g., alveolar /d/, /l, /s/) across manner classes, will help determine if locus equations are best described in feature or segment terms. Expanded phonetic contexts will include voiced stop place distinctions in syllable-final position and place distinctions for noninitial voiceless unaspirated stops.
Showing the most recent 10 out of 11 publications
