Identifying Segmental Substitutions in Spontaneous Speech of L3-French/L1-Japanese Learners: A Corpus-based Pilot Study

Abstract

This article illustrates current research in corpus-based L3-French/L1-Japanese interphonology. It uses one of the first annotated corpora of spontaneous speech of Japanese learners of French. Its goal is to identify persistent segmental ‘errors’ (or indices of ‘non-targetlike’ segments), through perceptual annotations of segmental substitutions at different levels of speech proficiency. Our results highlight the difficulty for Japanese learners in producing the French vowel /u/-[u] at all levels, among other segments. Since the task we use is close to what teachers and automatic speech recognition systems must handle, our pilot study paves the way for further research with comprehensively controlled data.