Infant Word Comprehension-to-Production Index Applied to Investigation of Noun Learning Predominance Using Cross-lingual CDI database

Infant Word Comprehension-to-Production Index Applied to Investigation of Noun Learning Predominance Using Cross-lingual CDI database Yasuhiro Minami1, Tessei Kobayashi2 and Yuko Okumura2 The University of Electro-Communications1, NTT2 1-5-1 Chofugaoka, Chofu, Tokyo 182-8585, Japan1, 2-4 Hikaridai Seika-cho Soraku-gun, Kyoto, 619-0237, Japan2 minami.yasuhiro@is.uec.ac.jp1, (kobayashi.tessei, okumura.yuko)@lab.ntt.co.jp2 Abstract This paper defines a measure called the comprehension-to-production (C2P) index to investigate whether nouns have predominance over verbs in children’s word learning that identifies a partial word learning period from comprehension to production. We applied the C2P index to noun predominance using cross-lingual child communicative development inventory databases and confirmed that it indicates noun predominance in word learning, suggesting that the process between a word’s comprehension and its production is a significant factor of predominance in noun learning by children. Keywords: vocabulary learning, word acquisition, cross-lingual CDI database 1. Introduction At around one year of age, infants produce their first words and rapidly start acquiring more of them. Almost every child follows this tendency. However, the content of a child’s first few words depends on the individual. These variances might reflect different culture, language, and individual environments. In the children word learning process, identifying what parts of speech children learn first is important; this result would be a step toward resolving the children's word acquiring mechanism and would provide clues for engineering solutions for teaching words to computers. Gentner argued that nouns should predominate verbs since they appear more often than verbs in the early development stages of children (Gentner, 1979; Gentner et al., 2001). There are two hypotheses for this predominance: the existence of a word continuum and constraints (biases). Gentner explained the former hypothesis using a division of dominance continuum. She assumed a word continuum in abstract space that varies from cognitive to linguistic dominance and introduced two assumptions, natural partitions and relational relativity, to explain that children learn words in the cognitive dominance region earlier than words in the linguistic dominance region. Since the natural partitions insist that concrete objects and entities are easier to individuate in the world, children easily acquire nouns. Relational relativity insists that a verb’s meaning is not isolated and that it depends on the surrounding words. Maguire et al. explained similar reasons for noun predominance using the shape, individuation, concreteness, and imageability (SICI) continuum (Maguire et al., 2006). These factors are assigned to the one dimensional abstract space axis (from left to right). The instances of words are arranged on the axis where the far left instance has an easy shape, simple individuation, high concreteness, and high imageability. Considering the meaning of nouns and verbs, we can allocate nouns to the left hand side and verbs to the right hand side. The SICI continuum has distributions of nouns and verbs that describe their individual difficulty differences. However, no direct evidence has connected the concrete measure of word difficulty to abstract spaces because no index, other than the occurrence rate of the part of speech, has expressed word difficulty. Since the occurrence rate is strongly affected by such environments …

Content truncated to 500 words