A SECCL-based Study Of Self-repairs In English Majors’ Oral Production

Self-repair occurs when a speaker detects erroneous or inappropriate output, halts the speech flow, and executes a correction without intervention from interlocutors. It is a common phenomenon in spoken language, the study of which possesses great theoretical and pedagogical values.This paper is a corpus-based study of the use of self-repairs in Chinese EFL learners’ spoken language. Taking Levelt’s Perceptual Loop Theory of Monitoring and theories on automatization and attention as the theoretical foundation, this paper aims to investigate the general characteristics of self-repairs in English learners’oral production as well as the relationship between self-repair behavior and language proficiency.Data used for the study were drawn from Spoken English Corpus of Chinese Learners (SECCL2.0), in which ninety samples of three proficiency levels were randomly selected from the oral data of TEM8Oral Test task3in the year2007. Based on Levelt’s (1983) classification of self-repairs with slight modifications, self-repairs in the collected data were identified and annotated. Then detailed analysis of these self-repairs was conducted, generating findings as follows:Firstly, the overall characteristics of self-repairs were retrieved and analyzed. The frequency of self-repairs detected in the data is relatively high (4.36self-repairs every100words). In distribution, S-repair (55.96%), which belongs to covert repair, occupies the largest proportion, and within overt self-repairs, E-repair is the most frequently used (21.01%), followed by A-repair (13.59%), D-repair (6.97%), and F-repair (2.47%). The results indicate that the subjects are more concerned with the correctness of linguistic forms than the appropriateness of content. EL-repair constitutes the most commonly seen subtype of E-repair, which echoes the findings of previous studies that L2learners are particularly sensitive to the correctness of lexical items. In structure, each type of self-repair has its preferred repair structure. For instance, D-repair concentrates in the structure of fresh start, while the structure of instant repair and anticipatory retracing are abundant in E-repair and A-repair. Secondly, the effect of language proficiency level on self-repairs was probed into. Generally speaking, with the improvement in language proficiency level, the subjects tend to make less self-repairs. The results of ANOVA indicate that there exists significant difference in the use of S-repair and A-repair among the subjects at different proficiency levels. Analysis of the subtypes of E-repair and A-repair shows that the high-level subjects make significantly more ES-repairs and AR-repairs than the low-level subjects. The results suggest that with a higher degree of automatization, high-level learners may become more capable of monitoring the complex syntactic and discourse level of the speech.The findings of the present study have significant pedagogical implications. Being an effective learning strategy and a stimulus for autonomous learning, self-repair deserves due attention from English teachers. They should encourage students to be conscious of self-repairs and foster their ability in autonomous learning. By paying more attention to students’ self-repair behaviors, teachers can also know better about students’ weak and strong points and their proficiency levels.