It has long been noted that children tend to omit syllables in their early speech. Although several proposals have been made as to why this occurs, none accurately captures either the English or crosslinguistic data, nor do they provide a developmental account for how children's early word shapes change over time. In contrast, a statistically and phonologically grounded perspective on these issues, employing insights from Prosodic Phonology and Optimality Theory along with consideration of the input, achieves better empirical coverage of the data. It also provides a more theoretically coherent understanding for why children omit the syllables (or parts of syllables) they do, how children eventually arrive at more adult-like speech productions, and the implications this holds for the emergence of grammatical function morphemes. The main focus of the project is to provide a coherent developmental theory of early prosodic phonology and morphology. The languages to be investigated are English and French - languages with very different prosodic and morphological structures. The study has three goals: The first goal is to conduct longitudinal case studies of 12 children's spontaneous speech productions from the onset of speech until around 3 years of age. The second goal is to design a set of linguistically sophisticated experimental procedures designed to test children's level of prosodic development. The experiments will be administered to both cross-sectional and longitudinal subjects. Both sets of data will be subject to fine-grained linguistic analysis, supplemented with acoustic analysis, of children's early speech productions at the level of Syllables, Feet, Prosodic Words, and Phonological Phrases. The third goal is to address the behavioral data from a quantitative perspective within Optimality Theory. This will be carried out by evaluating statistical variation in the data using both stochasitically ranked constraints and conditional exponential models. The longitudinal data, experimental tools and statistical methods developed will be invaluable for future study of both normal and language impaired populations, and will break new ground in exploring quantitative approaches to language variation.
Showing the most recent 10 out of 11 publications
