Optimal implementation setting for computerized visualization cues in assisting L2 intonation production

Research has established that computerized visualization cues as operationalized via waveform and spectrogram can assist L2 intonation production. However, whether such effects would vary as a function of learners' attentional state remains largely unexplored. Similarly, whether waveform and spectrogram should be displayed in discrete sentences or in contextualized discourse also remains unclear. This study examined the moderating effects of different attentional states (explicit vs. implicit) and notation contexts (discrete sentences vs. contextualized discourse) in computer-assisted pronunciation teaching (CAPT) on 85 L2 learners' diachronic intonational gain in scripted (planned) and unscripted (spontaneous) speech. The combination of the two variables led to four CAPT implementation conditions for scripted and unscripted speech: (1) explicit-sentence, 2) explicit-discourse, 3) implicit-sentence, and 4) implicit-discourse. The results show that when the objective is to promote intonational gain in scripted speech, the key is L2 learners’ attentional states; learners whose attention was explicitly directed to the computerized visualization cues exhibited more significant intonational gain. However, when the goal is to promote intonational gain in unscripted speech, the notation context in which the computerized visualization cues are embedded is crucial; L2 learners receiving the cues displayed in discrete sentences showed marked intonational gain. Pertinent pedagogical implications are discussed.