Phoneme categorization of synthetic vocalic stimuli by speakers of Japanese, Spanish, Persian, and American English.

Abstract

A set of 69 synthetic vowel sounds were generated which covered a wide range of F1/F2 frequency combinations. Subjects from four language groups (American English, Spanish, Japanese and Persian) were asked to listen to these stimuli in random order and to assign each one either to a vowel phoneme of their native language (as represented in a key-word) or to a class of sound which was not represented in that language. The results showed that the subjects were prepared to accept synthesized sounds for this purpose and were able consistently to assign many of the sounds to phonemes. The method therefore provides a means of arriving at the vowel phonemes for a number of languages, except in the case of sounds where time changes, such as diphthongisation, form an important cue.