한국어의 음절화

Abstract

In this paper I will discuss several phenomena related to syllabification in Korean, which have been regarded as phonological rules in traditional analyses. First, I will argue that the syllable in Korean is not a prosodic constituent, which serves as a domain of lexical rules in Prosodic Phonology (Selkirk 1986; Selkirk and Shen 1990; Nespor and Vogel 1986; Shih 1985; Hayes 1989). Second, information on syllabification does not need to be listed in the underlying representation but core syllabification first applies after ω's are formed, because syllable structure in Korean is simple and predictable. Note that Korean allows maximally one consonant either in the onset or in the coda. Third, universal constraints on syllabification, onset maximization and sonority sequencing principle, play a key role in Korean syllabification. If onset maximization cannot apply in the compound boundary or prefix+stem boundary, the consonant in question is neutralized as a coda. The surface coda realization of the underlying tautomorphemic consonant clusters is determined by coda licensing; only sonority and place specifications listed in the underlying representation are licensed to the coda in Korean. Resyllabification is an extension of onset maximization across prosodic word boundaries. Therefore, coda neutralization and stray erasure are not distinct rules, but they are trigged by syllable structure in Korean.