A Corpus-based Analysis of Shadowing Speech: Case of L2 English by Japanese Learners

In this study we intend to investigate the typical phenomena in shadowing speech and work out a tentative scheme for shadowing speech labeling. Our aim is two-fold: a) to give useful feedback to students and teachers who are using shadowing as a way of language learning; b) to explore the possibility of automatic assessment and error detection of shadowing speech. We firstly labeled a shadowing corpus, which includes data from 10 male and 10 female students, where their speaking proficiency ranges from low, to intermediate to high level. Then we did statistical analysis on the labeled phenomena and our preliminary results are: 1) roughly defining ten prototypes of speech phenomena in shadowing, including substitution, insertion, omission, filled pause, repetition, grammatical errors(tense, voice, etc.), whispering, rearrangement of words’ order, prosodic shadowing (shadowing without grasping the actual content of the speech) and combing parts from several words into one; 2) analyzing the subtypes within each prototype and finding that some types are gender/speaker dependent, and others are gender/speaker independent; 3) finding good correlation between our defined phenomena and shadowers’ overall language proficiency.