Input-based phonological acquisition

Abstract

This dissertation contrasts two theories of language acquisition. The first theory (Universal Grammar Hypothesis - UGH), which has been dominant in generative linguistics, argues that acquisition is primarily mediated by innate properties of language provided by UG. The opposing view (General Pattern Learning Hypothesis - GPLH) is that language is acquired according to the patterns in the ambient language or input. These theories are contrasted by examining children's acquisition of coda consonants in CVC words. The UGH was based on the universally preferred codas. These data came from previous research and from frequency analyses of codas in CVC words from 35 languages. Results showed that languages prefer coronal codas and sonorant codas. These cross-linguistic preferences are interpreted as reflecting UG. Predictions of the GPLH were established through an examination of English codas in CVC words from a number of different sources. This examination revealed the frequency of codas in the input, upon which the GPLH was based. In order to determine which hypothesis better predicts children's coda acquisition, data were then collected from previously published research, from CHILDES, and from an experiment designed to test children's productions of English codas. To evaluate the UGH, children's coda productions were analysed to determine whether the preferred codas were coronals or sonorants. Results showed that neither of these codas were favoured. To evaluate the GPLH, analyses determined whether there were significant correlations between children's coda productions and the frequency of codas in English. Results showed that these relationships were significant. The role of the input was further examined in an experiment designed to test children's productions of the identical coda in non-words controlled for phonotactic probabilities. Results showed that phonotactic probabilities played a significant role in accounting for children's productions of the identical coda in different words. The results support an input-based account of phonological acquisition. Thus, language acquisition is best characterised with respect to patterns in the ambient language, where frequently occurring properties of the input serve to organise children's linguistic representations. The research here illustrates the importance of considering the input in children's acquisition of phonological structures.