The purpose of this study is two-fold: (1) to show what problems exist in machine translation post-editing performed by translation students and (2) to suggest what should be considered in an undergraduate course on post-editing. To achieve this purpose, the researcher analysed data from a real post-editing project in which five junior students majoring in English-Korean translation post-edited Google Translate output. The source text is a medical text (consisting of about 780,000 words), written by a U.S. medical centre for the benefit of the general public. The product of the students’ post-editing was to be used for a smartphone application that would provide multilingual diagnosis services for ordinary people. The post-editing output was evaluated, using Mossop’s (2014) revision parametres: accuracy, completeness, logic, facts, smoothness, tailoring, sub-language, idiom, mechanics, layout, typography, and organisation. A close analysis of sample data revealed four problems: the students (1) performed monolingual post-editing in an inappropriate manner, (2) tried to resolve word-level issues without considering co-text, (3) paid scant attention to unnatural or ungrammatical language forms, and (4) had great difficulties in revising (translating) medical terminology. The implications of the findings for post-editor training are also briefly mentioned in the context of undergraduate translator training.
목차
Abstract I. 서론 II. 선행 연구 1. 포스트에디팅 품질 2. 포스트에디팅 교육 III. 연구방법 1. 연구의 배경 및 윤리 2. 참가자 3. 번역 브리프 및 분석 텍스트 4. 연구 방법 IV. 분석 결과 1. 정확성(accuracy) 2. 완결성(completeness) 3. 논리(logic) 4. 사실 관계(facts) 5. 연결 관계 및 자연스러움(smoothness) 6. 사용 적합도(tailoring) 7. 하위 언어(sub-language) 8. 연어, 관용구(Idiom) 9. 문법(Mechanics) 10. 레이아웃(layout) 11. 타이포그래피(typography) V. 논의 및 결론 1. 분석 결과에 관한 논의 2. 연구의 한계와 의의 참고문헌
키워드
free online machine translation (FOMT)Google Translatequality evaluationtranslation revisionpost-editor training