Dynamic Assessment during Suprasegmental Training with Mobile CAPT

This paper reports the results of a small-scale longitudinal study on the use of StudyIntonation, a computer-assisted pronunciation teaching environment. StudyIntonation aims to scaffold learners through their zone of proximal development by drawing eclectically on concepts, such as Vygotskian sociocultural theory, dynamic assessment and second language development. Learners perform shadowing tasks, aiming to replicate the suprasegmental prosodic aspects of model sentences. The pitch curves of the model and user attempts are displayed to help learners see their progress. We observed a group of learners who performed shadowing tasks in StudyIntonation for 24 months. The resultant corpus comprises 1050 speech records labelled with orthographic transcript, pitch readings, and similarity metrics. Longitudinal and microgenetic analysis of L2 pronunciation development was conducted on this dataset. Prosodic synchronization between speakers as well as longitudinal pronunciation assessment allows for quantitative evaluation by means of a dynamical modeling technique of cross-recurrence quantification analysis (CRQA). We located the zone of proximal development of each learner, where she/he reveals an increased responsiveness to input audiovisual stimuli, through the oscillations of pitch similarity metrics of dynamic time warping and CRQA. The rates of cross-recurrence between the model and learner are helpful synchronization indicators and performance predictors.