Input for Learning Japanese : RIKEN Japanese Mother-Infant Conversation Corpus(COE Workshop Session 2)

概要

論文の詳細を見る
The input for infant language learning is the speech that mothers and other adults produce around them. In order to investigate language acquisition, it is critical to understand the nature of input that infants are presented with. When speaking to infants (Infant-Directed Speech: IDS), adults modify their speech in certain ways-including prosodic, phonological and lexical level changes. Although a rich literature exists that documents various aspects of Japanese IDS properties, previous studies on phonological and prosodic aspects of Japanese IDS have been based on relatively small sample sizes. In this talk, we will introduce a new corpus of Japanese mothers' speech to infants. The RIKEN Japanese Mother-Infant Conversation Corpus (R_MICC) consists of 22 mothers' conversations with their 18-24 month old children as well as with adults. The corpus consists of approximately 45 minutes of recording from each mother-infant dyad, totaling 14 hours. Using the coding criteria developed for the Corpus of Spoken Japanese (2004, National Institute for Japanese Language) each mother's speech is transcribed phonologically, and coded for further phonetic details as well as morphological and prosodic characteristics. Analysis of this corpus will allow us to capture the basic characteristics of Japanese infants' language input, such as frequency of consonants and vowels, types of syllables, word types, parts of speech, intonational characteristics etc. For purposes of illustration, we will present a comparison of speech rates between IDS and ADS..
社団法人電子情報通信学会の論文
2006-07-08

Input for Learning Japanese : RIKEN Japanese Mother-Infant Conversation Corpus(COE Workshop Session 2)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク