Korean sentential nominalization, like English gerund phrases, exhibits internal verbal properties and external nominal distribution. Within the tradition of transformational grammar, it has been widely accepted that Korean sentential nominalization involves a syntactic process where the nominalization affix converts a verbal phrase to a nominal phrase (Han 1987, Yoon 1989). This article argues that the syntactic analyses have difficulty explaining the mixed properties without undermining important linguistic generalizations. Taking a lexicalist view (HPSG, Pollard and Sag 1994), this paper suggests an alternative lexicalist approach built on the mixed category analysis in Malouf 1998 and the type-hierarchical morphology in Kim 2004. This alternative proposes a way to explain Korean sentential nominalization without violation of lexicalism and endocentricity.