Ettehkey 'how' as a Small Clause Head

WH-words are not uniform in their syntactic behaviors. Adjunct WH-words such as HOW and WHY are said to be more restricted in their distribution than argument WH-words such as WHO and WHAT. It is observed, however, that HOW in some East Asian languages behaves more like argument WH-words and does not display ECP effects. In this paper, I try to account for the HOW vs. WHY difference in Korean. First, I briefly review two previous attempts. T. Chung (1991) ascribes the difference to the positional variance: ettehkey 'how' is generated VP-internally, whereas way 'why' is generated in an IP(AgrP)-adjoined position, i.e., above subject. The trace of island internal ettehkey, but not of way, satisfies the ECP under the assumption that subject (or INFL) is a (special) antecedent governor for adverbs, which he motivates based on the fact that adverbs agree with subject in number and may take plural morpheme tul when subject is plural. Now the trace of ettehkey, but not of way, is antecedent governed by subject due to the hierarchical (c-command) relation. Another attempt was made by D. Chung (1996), who proposes to decompose ettehkey into etteh-ki-ey and attributes the lack of the ECP effects to the nominal feature associated with the nominalizer ki contained in ettehkey. Both approaches fail when more data is considered. Crucially, it will be shown that etteh, the main part of ettehkey, is a predicate, (i.e., it is neither an adjunct nor a nominal element,) but it does not show the ECP effects. I extend the predicate analysis to ettehkey, analyzing it as the predicate head of an adjectival small clause. Now the question is why predicate WH-words do not show the ECP effects like argument WH-words. I provide a theta-theoretic account under the assumption that the theta-identification between a predicate and its arguments is a bilateral relation in the sense that they identify or restrict each other. Thus, ettehkey/etteh, as predicates, are theta-identified and do not show the ECP effects. In contrast, way, as a pure adjunct, is not theta-identified and does show the ECP effects. As for the ECP effects that English how displays, I suppose that it is not a predicate but a pure adjunct, based on the observation that adjectival small clause heads are replaced by what, but not by how, in WH-questions or echo-questions.