Title:
Speech Shadowing Method in Language Distant Learning

Speaker:
LEE, Carson (JAIST)

Abstract:
There exist a plethora of software fashioned to improve a person's language skill, from reading to grammar to writing. For learning to speak however, many software take the conversation and/or memorising approach. Speech shadowing is one of the technique that can be used to improve fluency of a person’s spoken language.  As compared to traditional “repeat-after-me” techniques, the speech shadowing method forces a learner to keep up with the pace of the native speaker and intonation. If we take a look at how children learn more to speak more complex words and sentences, we can see that a child mimics, or parrot his/her parents’ speech. Speech shadowing works not unlike that. This is one of the teaching method used by Japanese language teachers in my previous university for more advanced classes.

The goal of this research would be to develop a software that will enable a user to acquire a new speaking language by listening to recordings of native speakers and shadowing it. The shadowing process will be picked up by the software to measure the speaker’s tempo and intonation, as well as his/her pronunciation accuracy. The user would then be given a score to show his/her progress. The proposed method of doing this would be to compare the user’s voice waveform to the native speaker’s voice waveform.

Users would get the meaning and clarity of the recording via transcriptions. Memorising the transcription would however, somewhat defeat the purpose of this method as it becomes the traditional memorise and regurgitate method instead. In this research, the target language would be English, but adding more languages should not be a problem should there be sufficient time.