In analyzing Korean language for machine translation, the output of morphological analysis is morphemes, and the basic composition of syntactic analysis is sentence constituent. Korean language is an agglutinative language, and sentence constituent consists of morphemes. Thus, it is necessary to transform morphemes into sentence constituents to do syntactic analysis.
In this thesis, the processing to transform morphemes into sentence constituents is presented. The processing consists of six phases, that is, the numeral processing, the derivative processing, the compound word processing, the combination of stem and ending of a word, the auxiliary processing, and the combination of independent word and postposition.
The processing reduces by half the number of input tokens of syntactic analysis. Many undefined words in the dictionary can be constructed by using morphemes such as prefixes and suffixes to form derivatives and by combining adjacent nouns to form compound nouns.
