On Distribution Of Conversation Repairs Of English Majors

This thesis is a corpus-based study of conversation repair in the framework of conversation analysis with the focus on the distribution of repair typology. The mini-corpus for study is taken from SWECCL (Spoken and Written English Corpus of Chinese Learners), which includes transcriptions of English oral tests for English majors. Conversation by English majors whose native language is Chinese provides a special setting for the study of repair and may contribute some insights to EFL teaching, learning and assessment.Based on the previous findings and also for the purpose of research, the present study designs a typology for repair that consists of four types (namely, self initiated self repair, self initiated other repair, other initiated self repair and other initiated other repair) and nine subtypes, among which more emphasis is attached to the distribution of the same turn self-repair with four types and six subtypes. They are same repair, different repair, appropriate repair (appropriate replacement repair, appropriate supplement repair), and error repair (lexical, syntactic, phonetic, morphological error repair). All the types and subtypes are marked in the mini-corpus by the researcher herself, and statistical analysis is done about their frequencies with the main tool being Chi-square tests. Some empirical findings are then attained based on the statistic analysis:First, other initiated repairs are really rare and self-initiated repairs are significantly more than other types. Second, in same turn self-repairs, same repairs are significantly more than different repairs, appropriate repairs and error repairs. Third, on the whole gender does not significantly affect the distribution of repair types either in the same turn or in different turns. Fourth, participants' grouping mechanism significantly affects the distribution of repair types in the same turn and finally, as far as the data are concerned, no delayed-other-initiated repairs are detected.For these findings about distribution of repair typology, explanations are proposed from four perspectives. In the first place, it is turn taking-rules that determine the preference of self-initiated repairs to other-initiated repairs. Human interacts on a turn-to-turn basis; there are rules governing the change of turns and the rules contribute to the preference of self-initiated repairs. In the second place, other repairs are face-threatening acts, and the strategies for doing these acts are decided by social variables and certain payoffs. In order to save face, speakers are unwilling to do other repairs. In the third place, gender differences are not born or absolute, and gender identity is a dynamic construct accomplished through the interaction with other social members in social activities. The special setting in which female and male students are conversing with each other leads to the significant difference in the distribution of the same turn self-repair between groups. Girls are more gender-aware when conversing with boys, thus improve their control over language forms while boys are affected with more outcome of self-repairs. Finally, non-native speakers are different from native speakers in their language proficiency, and this imparity leads to zero occurrences of delayed-other-initiated repairs.The author insists that the above findings possess some pedagogical values for EFL learning, teaching and language assessment. First, in EFL teaching and learning, though gender is an involving factor, its role should not be exaggerated by language teachers and learners. Second, in language teaching, when a trouble source occurs, ample opportunities should be given to learners considering that they are able to do successful self repair. Third, phonetic and syntactic errors should be given special attention to as these two types of errors may be more demanding for learners to repair. Fourth, in language activities, female and male learners may be grouped together to stimulate their language awareness. Finally, in EFL language assessment, the task types should be more simulative and more interactive to encourage real communication, and female and male participants may be assigned in one group to witness higher awareness of language form.