In this paper I introduce a new set of data on predicate-repeated alternative question in Korean, say, V-not-V question, and claim that this construction should not be treated by a bi-clausal analysis applying for a usual alternative question but be dealt with by a simple clause analysis. A variety of word orders observed in this construction reject the bi-clausal approach and further opt for the SVO hypothesis for Korean rather than the usual SOV hypothesis. In particular, I suggest that a kind of operator movement is involved in the construction in question, thereby accounting for island effects as well. The results of this paper lend support for the universal specifier-head-complement order initiated by Kayne (1994) and help eliminate the head-parameter, thereby reducing computational complexity.