Callier Ctr. for Commun. Disord.--UTD, 1966 Inwood Rd., Dallas, TX 75235
Within the vowel space there are subareas which correspond to specific vowel categories. Within each subarea there is a point which defines the best exemplar or prototype of that vowel. Kuhl [J. Phon. 21, 125--139 (1993)] has argued that the area immediately surrounding a prototype is perceptually shrunken, with reduced variance in responding to points in that immediately surrounding area, a magnet effect [Kuhl (1993)]. Empirically, prototype points correspond to the averages of large numbers of utterances. This leads to the hypothesis that prototypes are learned. It is equally possible that prototypes are innate [Kuhl (1993)]. To decide between these two hypotheses an experiment was run with newborns. The vowel /y/ was used in the experiment; /y/ does not occur in Texas. The newborns' only experience was provided in the experiment. ``Best exemplar'' was equated with ``most attractive'' measured by listening time in an operant choice task. Results indicated that experience did not determine preference. The prototype /y/ was preferred, regardless of the experience provided. No significant magnet effect occurred. A second, slightly different, experiment did find a magnet effect. The difference between the two experiments may explain why the magnet effect occurs.