초록 열기/닫기 버튼
English allows inanimate subjects with theta-roles such as Causer, Theme, Source, Instrument, Locative, and Time while Korean generally allows inanimate subjects with only Theme. Machine translation of English into Korean, however, does not reflect this mismatch and only considers the syntactic position. This paper examines the current status of machine translation and propose a way which can be helpful in making an algorithm to reduce errors. This study uses three machine translators available online to translate English sentences with inanimate subjects into Korean. Only 3% of all the Korean translations by machine translators were acceptable. English verbs should be classified based on theta-role frames to reduce translation errors. Nouns should be assigned an animacy feature: [+animate] or [-animate]. A machine translator should have an algorithm whereby (1) [-animate] subjects are translated adverbially according to the theta-roles of the subjects, (2) objects are translated as subjects in Korean, and (3) verbs are translated into either intransitive verbs or passivized verbs morphologically or syntactically based on the context and the characteristics of Korean verbs.
키워드열기/닫기 버튼
machine translation, inanimate subject, animacy, theta-roles, feature
