earticle

영어

Due to structural differences between Korean and English, machine translation from Korean to English faces difficulties when it processes multiple case constructions. Korean, a language with overt case markers, can have more than one Nominative/Accusative case-marked argument in a clause while English, whose case realization is abstract in general, restricts the number of Nominative/Accusative case arguments based on properties of valid case assigners in a given clause. After reviewing morphological case as well as abstract case observed in Korean and English respectively, this paper categorizes patterns of multiple case constructions in Korean based on semantic relations between arguments. An error analysis based on actual translation data from two Korean to English translators (Yahoo Babel Fish and PAT transer) makes a suggestion for morphological and syntactic processes of machine translation