Why is HOW in Korean Insensitive to Islands?: A Revised Nominal Analysis

Abstract

Korean HOW, i.e. ettehkey, unlike its English counterpart, i.e. how, does not show island effects. There have been several approaches entertained in the literature to account for the island insensitivity:T. Chung's (1991) VP adjunct analysis and an ECP account; D. Chung's (1996) nominal analysis and a binding account; Yang's (1997) d-linking analysis and an ECP account; and D. Chung's (2000a) predicate analysis and an ECP account. As reviewed in Chung (2000a), the first two analyses bear some critical empirical and/or theoretical problems. Pointing out that the third and fourth approaches also face some empirical and/or theoretical problems, this paper proposes an alternative analysis, a revised nominal analysis, in which ettehkey is decomposed into four sub-parts: e-tte-ha-key, 'Det-N-do-adverbializer'. The second element, i.e. -tte, is diagnosed as a nominal element because the first element, i.e. e-, as a determiner, requires a nominal complement and the third element, i.e. -ha, as a transitive verb, requires a nominal complement. The proposed analysis gains support from the morphological paradigms that Korean WH-elements display. Given this revised nominal analysis, the scope of HOW in Korean can be licensed via binding, accounting for the lack of island effects.