| This dissertation investigated the knowledge and learning mechanism of functional categories (e.g., determiner and auxiliary) in young children. The first part of this dissertation demonstrated the importance of sample size when measuring linguistic productivity using speech samples, which may have contributed to discrepancies in previous studies of young children's knowledge of the determiner category. It then proposed a novel probabilistic method for measuring linguistic productivity that takes sample size into account. With this new method, analyses of eight English and one German longitudinal corpora of children's naturalistic speech production showed that young children between 1;11 and 3;0 were using determiners in the same way as adults do. The results are consistent with previous findings that even 14-month-olds are processing determiners as an abstract category.;The second part of this dissertation tested the hypothesis that distributional information in the input (e.g., bigrams) is informative for categorizing function words into finer categories. The analyses demonstrated that when acoustic and phonological cues are used to identify function words and their contexts from the input, linear distributional patterns are extremely accurate in categorizing function words into substantial-sized groups of words from the same grammatical category.;In summary, this dissertation contributes to several important questions in language acquisition, including the origin of knowledge of grammatical categories, the developmental course and learning mechanism of grammatical categories. Moreover, the demonstration of the sample size problem is illuminating for future research using corpus analysis. The novel probabilistic method proposed here for measuring young children's linguistic productivity could also be adapted to study other aspects of linguistic structures. |
