Using Pitch and Length Information to Assess Speech Prosody: a Parallel Approach

Abstract

Recent studies on pronunciation instruction underscore the effectiveness of using pitch variation contours to enhance language learners’ awareness of English intonation. Little is known of the situation in language production: Do learners adjust their pitch levels or lengthen or shorten a sound in order to vary their prosody? The current study posits that both pitch and length variations are involved in encoding speech prosody; it purports to quantify their relative importance by using Hincks’s (2005) Pitch Variation Quotient (PVQ) and a self-created length variation quotient (LVQ). The methods section introduces four quantitative methods, using data from the participants’ verbal speech as well as the training material (a song) to compare pitch and length behaviours. Results from three of the four methods support that the learners, after undergoing a period of training (a singing class), improved their length variation but not their pitch variation over a reading task. The fourth method (a mixed between-within ANOVA) also confirms the learners’ general prosodic improvement, but the test raises curious points about the complex relationship between pitch and length that are worth further consideration by pronunciation assessors. Overall, this study provides initial evidence of the importance of length variation for assessing prosodic change.