Frequency Distribution in Phoneme Learning Across the Lifespan

One hypothesis to explain perceptual narrowing in speech perception is the distributional learning account. This account claims that both infants and adults are able to infer the number of phonemic categories through observations of frequency distributions of individual phones in their speech input (Maye, Werker, & Gerken, 2002). Although the distributional learning account indeed provides an insight on how listeners learn phonemic categories in their native language, it is not well understood how it interacts with other sources of information. The aim of this thesis was to address how listeners utilize distributional cues with other sources of information in learning phonemic categories. The series of experiments described in this thesis examine (a) how infants use distributional information in the presence of multiple speakers and (b) whether the experiences of native language(s) or music, that have been known to be beneficial to some linguistic tasks, interact with distributional information in adults. In the first two experiments (Experiment 1, 2) of Empirical Investigation 1, I investigated whether the use of multiple cues (acoustic variability, visually presented cues, and statistical information) to infer proper phonemic categories is within the capacity of infants. When multiple speakers are presented, computing speaker-specific distribution may be necessary. Providing visual context can guide infants to extract the appropriate phonemic categories. In the following investigations (Empirical Investigation 2 & 3), similar questions were addressed in adults. First, I investigated whether bilingual and/or musical experience affords adults advantages in their sensitivity to frequency distribution in learning non-native stop phonemes (Experiment 3). Second, I examined speakers of two languages, English and Korean, that differ in their use of specific phonetic cues for classifying stop consonants, in their discrimination capacity to non-native stop phonemes and speculated how the presence of specific phonetic features affected sensitivity to the frequency distributional information (Experiment 4, 5). Overall results suggest that while it was not evident that infants could infer phonemic categories in the presence of speaker variability, adult listeners seem to utilize other sources of information with distributional cues in learning non-native phonemes. Taken together, the findings provide important insights into how listeners utilize frequency distributional information with other sources of information to systematically acquire phonemic categories in native and/or non-native phonology across the lifespan