Do they read aloud in a different way? A glimpse on the prosodic features of corpus-based second language spoken English.

Abstract

This study investigated the prosodic features of L2 spoken English. A set of ten recordings was extracted from an in‐progress learner corpus on an English proficiency test. Each recording consists of two passages as read aloud by a learner who had received a grade of 3, the median grade of the test, on a five‐point scale. A group of ten native English speakers was recruited to serve as controls. They were given the same test materials and their readings recorded under a test scenario similar to that of the L2 learners. The labeling followed the English ToBI convention. Preliminary results showed that in general, the L2 learners addressed more tones in their production. In terms of break indices, the L2 utterances were consistently segmented into a larger number of intonational units, as compared to those of the native speakers. In addition, silence was frequently employed in the L2 word‐level boundaries to facilitate enunciation, explained in that English has a more complex phonotactic structure than their native tongue, Mandarin. Finally, the L2 data exhibited more varied pitch accent types than the native‐speaker data, which corresponds to the findings of our previous study regarding an L2 read‐aloud task on simple English sentences.