Transcription and annotation of a Japanese accented spoken corpus of L 2 Spanish for the development of CAPT applications

This paper addresses the process of transcribing and annotating spontaneous non-native speech with the aim of compiling a training corpus for the development of Computer Assisted Pronunciation Training (CAPT) applications, enhanced with Automatic Speech Recognition (ASR) technology. To better adapt ASR technology to CAPT tools, the recognition systems must be trained with non-native corpora transcribed and annotated at several linguistic levels. This allows the automatic generation of pronunciation variants, new L2 phoneme units, and statistical data about the most frequent mispronunciations by L2 learners. We present a longitudinal non-native spoken corpus of L2 Spanish by Japanese speakers, specifically designed for the development of CAPT tools, fully transcribed at both phonological and phonetic levels and annotated at the error level. We report the results of the influence of oral proficiency, speaking style and L2 exposition in pronunciation accuracy, obtained from the statistical analysis of the corpus.