3aSC36. Relational perception of formants and stop consonants: Evidence from missing first formant stimuli.

Session: Wednesday Morning, December 4


Author: Krishna K. Govindarajan
Location: Speech Commun. Group, RLE, MIT, Cambridge, MA 02139


If the first formant (F1) is missing, does the listener treat the second formant (F2) as the effective ``F1'' or does he/she substitute a default value for F1? Is stop consonant identification dictated by relational or absolute formant frequencies (e.g., locus theory)? To answer these questions, an experiment was carried out in which subjects identified synthetic consonant--vowel (CV) stimuli where F1 had a transition that rose into the vowel, had no transition (i.e., a flat transition), or was absent. Listeners identified stimuli with a rising F1 transition as either /ba/ or /da/ depending on the F2 transition, and stimuli with no F1 transition as /a/. For stimuli where F1 was absent and F2 had a rising transition, listeners perceived either /ba/ or /da/ depending on the F3 transition; and for stimuli with F1 absent and no F2 transition, listeners perceived /a/. Thus preliminary findings suggest that when F1 is missing, subjects do indeed treat F2 as ``F1,'' and respectively, F3 as ``F2,'' leading to the conclusion that stop consonant identification is not dictated by absolute formant frequencies, but uses relative measures. The results have implications for models of stop consonant perception. [Work supported by NIH grant DC00194-02.]

ASA 132nd meeting - Hawaii, December 1996