| People make mistakes when they speak,but people can also detect these mistakes and repair them,a manifestation of how people handle intrinsic problems in speech production.In interpreting,self-repairs serve as a window into this real-time problem-solving process to see what interrupts a fluent output.Drawing on Levelt’s speech production model and retrospective interviews,this study aims to explore the self-repairs in English to Chinese consecutive interpreting.Characteristics of 207 self-repairs made by 14 interpreting trainees will be presented and problems triggers for self-repairs will be investigated to offer suggestions for interpreting training.Findings suggest that interpreting trainees encounter more errors and message arrangement difficulties in output,but also show an inclination to modify their output for appropriateness.Problem triggers for self-repairs come from three dimensions: input-generated problems as the result of inefficient note-reading and attention lapses trigger message rejection and message insertion repairs;output-generated problems manifested as inefficient semantic and syntactic processing result in abundant lexical error repairs and message reordering repairs;interpreter-generated problems including knowledge gaps and bilingual transferring capabilities lead to failed lexical error repairs and unwarranted semantic appropriateness repairs... |