Inverse scope and unaccusativity alternation

Abstract

In this paper, I first review evidence for the claim that there is an inverse correlation between Japanese scrambling and QR: that is, Japanese is scope rigid because it allows scrambling as a syntactic option (Szabolcsi 1997, Bobaljik and Wurmbrand 2012). According to Bobaljik and Wurmbrand's approach, QR is blocked in Japanese because Japanese has scrambling. There are, however, cases in which apparent inverse scope is easily available in Japanese (Oku 2010), which is problematic to Bobaljik and Wurmbrand and to any theoretical attempt to account for Szabolcsi's inverse correlation. To explain this conflicting situation, I will propose that verbs involved in such apparent counterexamples are actually unaccusatives so that the surface subject is the underlying complement of the verb: the apparent inverse scope is read off at the underlying structure where the surface subject is c-commanded by the PP. As independent evidence for the unaccusasivity of the verbs in question, I will show that the floating quantifier fact (Miyagawa 1989) confirms what I propose. Further, the unaccusativity alternation in Japanese (reported by Yamada 1998) is endorsed by the inverse scope facts explored in this paper.