A Distinctive Collexeme Analysis of -(u)m and -ki in Korean

This paper investigated the properties of Korean nominalizer -(u)m and -ki using a distinctive collexeme analysis (DCA; Gries and Stefanowitch, 2004b). In this study, all the sentences with Korean nominalizer -(u)m and -ki were extracted from the Sejong Treebank. Then, a DCA was applied to the extracted corpus data. A DCA is known to be a statistical analysis which identifies the lexemes with a strong preference for one of the alternations. The analysis results demonstrated which predicates prefer -(u)m and which ones prefer -ki. Through the analysis, the followings were observed: (i) The collostructional strengths (CS) were differentiated in two dimensions (-(u)m vs. -ki and matrix predicate vs. complement predicate), (ii) -(u)m nominals have stronger CS values than -ki nominals, (iii) a lot of predicates can combine with -(u)m or -ki to form a nominal but that only some predicates -(u)m or -ki nominals as a complement, and (iv) Matrix predicates have stronger CS values than complement predicates.