Vowel Categorization and the Critical Band

Using the concept formation paradigm, two series of experiments were conducted to test the hypothesis that the critical band (CB) was a factor in learning to make absolute discriminations of vowels. The specific hypothesis being examined was that the CB is a psychoacoustic boundary in learning to make vowel categorizations, and that learning absolute discriminations of pairs of vowels that differ in one of their formants by one bark or more is significantly easier than learning absolute discriminations of vowels that differ by less than one bark. Subjects were given the task of learning to identify paired sets of synthesized vowels that differed in either F 1 , or F 2 by 0.4, 0.6, 0.8, 1.0, or 1.2 bark. The results of these experiments suggest that the critical bandwidth is not a natural psychoacoustic boundary in the learning of vowel categories, and also that subjects seem better able to learn distinctions involving differences in F 1 than distinctions involving differences in F 2. The discussion raises the possibility that vowels that differ by less than one bark may not be perceptually viable because of such factors as ambient noise, articulatory constraints, and coarticulatory influences. Some evidence from Dutch and French is presented in support of this conjecture.