為提升第二語語調習得效益之電腦視覺化輔助最佳操作及呈現環境

Existing research, albeit still limited in number, has established that exposure to the visual display of intonational features through waveform and spectrogram can effectively promote the production of L2 intonation in discrete sentences when such exposure is supplemented with explicit instruction. However, little is known if the efficacy of waveform and spectrogram depends on explicit instruction; and whether L2 learners’ explicit cognitive processing of the speech visualization notation cues is the key to their success/efficacy. Additionally, much remains unknown if the two supports need to be presented and encoded in discrete sentences or in contextualized discourse for optimal learning outcomes. To fill this gap in existing computer-assisted pronunciation teaching (CAPT) literature, the study was set out to explore the optimal CAPT implementation settings for the use of waveforms and spectrograms. In particular, two CAPT implementation variables are of particular interest to this study: 1) the optimal encoding (presentation) context for waveforms and spectrograms (discrete sentences vs. contextualized discourse); and 2) the desirable cognitive processes that should be invoked in encoding (learning) the waveforms and spectrograms (explicit vs. implicit processes). Throughout a time-series CAPT treatment on 65 L2 learners and the analysis of their intonational gains in scripted and unscripted speeches, this study argues that the optimal CAPT settings the one in which speech visualization notation cues are embedded in discrete sentence stimuli and where learners’ attention is being explicitly directed to attend to the support cues.