A STUDY INTO ACCURACY ASSESSMENT FORMULA OF INFORMATION IN E-V CONSECUTIVE INTERPRETING OF ULIS FOURTH-YEAR STUDENTS=

Abstract

ABSTRACT

The purpose of this study was to create a formula for accuracy assessment of information on the basis of NER-model (N: number, E: edition error, R: recognition error) which is used to evaluate the accuracy of live subtitles using speech recognition. The subject of the research was 27 ULIS fourth-year students majoring in translation and interpretation taking their interpreting exam. Although the exam included both simultaneous and consecutive interpreting sections, the study focused only on the consecutive performance of students. Using the mixed method consisting of qualitative and qualitative analysis via questionnaire, interview and document analysis, the author attempted to judge the function of the new formula and found out related results. The questionnaire was carried out right after students sitting their exam. The researcher did the interview with the exam-maker of the test to gain more insight understandings towards its criteria and requirements. More importantly, students’ recordings of their exams were transcribed for document analysis. The accuracy formula included two sub-formulas regarding idea and word unit. Each of them had their own set of criteria. After testing the formula, the author had come to the conclusion that the new formula for accuracy assessment developed by the researcher worked out. Besides, most of students were aware of their performance and errors they had made. In short, the study has been successful to some extent in the effort of developing an accuracy assessment formula; however, certain limitations still remains and further studies need implementing to improve these weaknesses.