Methods for testing vowel perception in the field

Abstract

Based on procedures developed by Johnson etal. [J. Acoust. Soc. Am. 91, 2387(A) (1992)] programs have been written for Macintosh and MS-DOS portable computers for experiments on the perception of vowel quality in the field. In one version, vowels determined by 15 different values of F1 and 22 values of F2, with F3 determined by algorithm, were digitized and stored. In the basic experiment listeners try to match one of these stimuli to a selection of their own vowels. Listeners can hear any of the stored stimuli by clicking with a mouse on a square in a 22×15 matrix. In other experiments, listeners estimate the distance between pairs of stimuli, or select the better match to one of their own vowels from pairs of vowels that are algorithmically varied from trial to trial in accordance with the previously selected better match. A later version of these programs, not yet tested in the field, generates vowels in real time. The advantage of this system is that the fundamental frequency, the vowel duration, and the third formant can be manipulated by the subject. This is important when matching vowels in languages in which tone or length or rhotacization is contrastive.