The Two-level morphological analysis model is a general model for morphological analysis and synthesis first proposed by Kimmo at 1983. The strength of the model lies in the rules that are bidirectional, parallel, and easy to implement. It however has been to be difficult to apply this model to Korean. Since Korean is a syllable-based language and has too many inflected forms.
In this thesis, the two-level morphological rules for Korean presented. We found that the regularity of irregular inflections of verbs and adjectives could be captured in the model. Nonetheless the development of rules is not easy as multiple spell changes may occur in a word phrase. By devising phoneme based Korean code the rule description became much efficient. Three boundary markers have been used to distinguish regular and irregular verbs and adjectives.
The developed rule set includes 52 rules for processing 17 inflections that can account for standard Korean. The test results show that the two-level model can be used to handle effectively diverse inflected forms of Korean.
