동음이의어의 사용 양상

Abstract

We examine various aspects in the use of homonyms (ambiguous words) in Korean. In a comprehensive Korean dictionary, about 30% of the entries are ambiguous words and most of them are nouns. In the use of the language, 30% of the used nouns are occurrences of forms with two or more senses (not in the sense of polysemy but in the sense of homonymy). Thus lexical ambiguity might cause potential difficulty in human and computer processing of Korean. But closer examination of the usage shows that the problem of ambiguity might not be so devastating. Among ambiguous words of the same form, people use a single word (sense) most of the time. About 98% of the noun usage can be accounted for by the most frequently used word (sense) among ambiguous words. This tendency is reinforced within a discourse context, where two or more homonymous words are rarely used together. To get these and related results we collect data from and experiment on Sejong Sense Tagged Corpus (5.5 million words) and (sense tagged) Korean Modern Novel Corpus (1 million words). Results from some preliminary WSD(word sense disambiguation) experiments, which are based on collocation, frequency, and discourse contexts, are reported.