기술문서의 한일기계번역 문제에 대한 통제언어 연구 —‘-되다’ 구문의 기계번역수월성 제고를 위한 통제규칙 (A Study of Controlled Language in Korean-Japanese Technical Document Translation —Control Rules For the Facilitation of Machine Translation Regarding ‘-되다’—)

서지정보

· 발행기관 : 한국번역학회
· 수록지 정보 : 번역학연구 / 11권 / 4호 / 191 ~ 238페이지
· 저자명 : 함수진, 유수린

초록

Technical documents refer to documents produced for the purpose of recording knowledge or information in the area of education, science and technology. Translation of such documents becomes more and more important as their use and application keeps expanding along with globalization reaching readers of different languages. However, the structural ambiguity and vagueness found in such technical texts, as well as several other factors, hinder the comprehension, and therefore, the effective translation of technical documents, especially in the case of machine translations which are used for mass translation.
Many errors appear in terms of style, lexicon, and grammar when machine translation is used for translating technical documents from Korean into Japanese. Among the 1,473 [되다] examples of syntax extracted from the Korean-into-Japanese technical document corpus, which is the subject of this study, errors(16.09%) have been found where [N-이/가 되다] syntaxes indicating ‘passive voice’ have been automatically translated into [N-になる] which indicate ‘a change of state’. A deeper inquiry into previous studies revealed that these errors were caused by a structural ambiguity lying at a grammatical layer where the [N-이/가 되다] syntax can imply both ‘a change of state’ as well as ‘passive voice’.
This study proposes an approach that uses controlled language to avoid errors caused by polysemic and structural ambiguity of the [N-이/가 되다] syntax which are often found in Korean-Japanese machine translations. The controlled language approach aims to effectively control language from the initial stage of text production with the purpose of enhancing comprehensibility and the facilitation of machine translation.
First, the distinct properties and characteristics of said syntax have been analyzed from reference studies to deduce formalized standards and values to resolve the ambiguity of [N-이/가 되다]. Second, controlled grammar and lexicon writing rules presented in previous studies have been considered or applied to form and implement a set of control rules that include writing rules and writing proscriptive rules.
There should be follow-up studies to categorize errors found in various layers of style, lexicon, grammar that can affect the quality of Korean-Japanese machine translation and continuous research should be conducted regarding the cause-and-effect relationship in applying the rules of the controlled language to ensure that the rules function complementary to one another.

영어초록

Technical documents refer to documents produced for the purpose of recording knowledge or information in the area of education, science and technology. Translation of such documents becomes more and more important as their use and application keeps expanding along with globalization reaching readers of different languages. However, the structural ambiguity and vagueness found in such technical texts, as well as several other factors, hinder the comprehension, and therefore, the effective translation of technical documents, especially in the case of machine translations which are used for mass translation.
Many errors appear in terms of style, lexicon, and grammar when machine translation is used for translating technical documents from Korean into Japanese. Among the 1,473 [되다] examples of syntax extracted from the Korean-into-Japanese technical document corpus, which is the subject of this study, errors(16.09%) have been found where [N-이/가 되다] syntaxes indicating ‘passive voice’ have been automatically translated into [N-になる] which indicate ‘a change of state’. A deeper inquiry into previous studies revealed that these errors were caused by a structural ambiguity lying at a grammatical layer where the [N-이/가 되다] syntax can imply both ‘a change of state’ as well as ‘passive voice’.
This study proposes an approach that uses controlled language to avoid errors caused by polysemic and structural ambiguity of the [N-이/가 되다] syntax which are often found in Korean-Japanese machine translations. The controlled language approach aims to effectively control language from the initial stage of text production with the purpose of enhancing comprehensibility and the facilitation of machine translation.
First, the distinct properties and characteristics of said syntax have been analyzed from reference studies to deduce formalized standards and values to resolve the ambiguity of [N-이/가 되다]. Second, controlled grammar and lexicon writing rules presented in previous studies have been considered or applied to form and implement a set of control rules that include writing rules and writing proscriptive rules.
There should be follow-up studies to categorize errors found in various layers of style, lexicon, grammar that can affect the quality of Korean-Japanese machine translation and continuous research should be conducted regarding the cause-and-effect relationship in applying the rules of the controlled language to ensure that the rules function complementary to one another.

참고자료

· 없음