Analyzing Vocabulary Commonality Index Using Large-scaled Database of Child Language Development

Analyzing Vocabulary Commonality Index Using Large-scaled Database of Child Language Development Yan Cao1,2, Yasuhiro Minami1, Yuko Okumura2, and Tessei Kobayashi2 1The University of Electro-Communications, 2NTT Communication Science Laboratories 11-5-1 Chofugaoka, Chofu, Tokyo 182-8585, Japan, 22-4 Hikaridai, Seika, Kyoto 619-0237, Japan caoyan@sd.is.uec.ac.jp, minami.yasuhiro@is.uec.ac.jp, {okumura.yuko, kobayashi.tessei}@lab.ntt.co.jp Abstract The present study proposed a vocabulary commonality index for child language development to investigate to what extent each child acquires common words during the early stages of lexical development. We used large-scaled, vocabulary-checklist data from Japanese-speaking children (N=1,451) aged 8-48 months to estimate their age of acquisition (AoA) of 2688 words by logistic regression. Then we calculated the vocabulary commonality index for each child with two datasets. The results showed that as their vocabulary size increases, children who have the same vocabulary size tend to produce common words with the same ratio. Keywords: Large-scaled Vocabulary Database, Child Language Development, Age of Acquisition, Vocabulary Commonality Index 1. Introduction How children acquire vocabulary is among the most central issues in the fields of cognitive science and developmental psychology. Many previous studies have scrutinized what types of words young children acquire during their early stages of lexical development using a vocabulary checklist methodology, such as the MacArthur-Bates Communicative Development Inventories (e.g., Bates et al., 1995; Caselli et al., 1995; Fenson et al., 1994; Frank et al., 2017). For example, Caselli, Casadio and Bates (1999) found that 18-30 month-old English- and Italian-speaking children tend to produce more social words (e.g., people’s names, games, routines, etc.) than other types of words in their first 50 words and more common nouns (e.g., animals, foods, toys, etc.) in their first 100-500 words. Although such category analyses of vocabulary are useful for grasping children’s overall tendencies in developmental stages, they are less satisfactory for understanding more detailed changes and the individual differences of vocabulary development because the category ranges are wide-ranging and sparse (i.e., only four categories). One possible solution for grasping detailed changes and individual differences in vocabulary development is to directly compare the word items (rather than categories) acquired by each child with the word items acquired on average by many children and roughly estimate the commonality between both word sets. If we can successfully estimate the commonality in a statistically reliable way, we might be able to understand to what extent each child acquires common words and further clarify the detailed changes and individual differences of vocabulary development. In the present study, we propose a vocabulary commonality index for child language development by estimating to what extent each child can say common words. We also create a new vocabulary-checklist that includes much more word items (i.e., 2688 words) than the Japanese version of MacArthur-Bates CDI (i.e., 448 words). This is because we thought that the word range of MacArthur-Bates CDI was insufficient to grasp detailed changes and individual differences in vocabulary development. Therefore, using such a new vocabulary- checklist, we create a large-scaled database of vocabulary database in Japanese-speaking children. 2. Data Collection The data for the study came from …

Content truncated to 500 words