Use of multiple speech dimensions in concept formation by Japanese quail

Abstract

Japanese quail (Coturnix coturnix) have learned to form a category for syllable‐initial [d], and can discriminate [d] syllables from those beginning with [b] or [g] across vowel contexts [Kluender et al., Science (1987)]. Acoustic analysis of the categorized syllables revealed no single feature that could have supported generalization, suggesting that the quail used some combination of stimulus properties to categorize syllables. Current experiments investigate two ways such a category could be formed. First, initial consonant information may be evaluated conditional upon the vowel that follows. Whether quail can learn to categorize syllables in which, for example, positive exemplars are CVs with either a voiced stop and a front vowel, or a voiceless stop and a back vowel is being tested. A second possibility is that phonetic categories are examples of polymorphous concepts, with no single necessary or sufficient condition for class membership. Two birds have learned a concept in which three stimulus dimensions of syllables are needed to make a category assignment. The best subject has learned a concept that requires her to discriminate: (1) voiced versus voiceless syllable‐initial stops, (2) front versus back vowels, and (3) sex of the talker. Responses to novel stimuli reveal a category structure remarkably similar to that hypothesized for humans. The fact that quail can learn a concept in which multiple cues are clearly required suggests that general mechanisms of audition and perceptual learning are sufficient for acquisition of speech categories. [Work supported by NICHD.]