A structural asymmetry in intervention effects

This paper documents a previously unnoted structural asymmetry amongwh-phrases with respect to interventioneffects. I show ina series of experiments (phonetic and syntactic) that inKorean (andpotentially in JapaneseandTurkish), onlywh-phrases that arebasegenerated in a structurally low position trigger intervention effects while wh-phrases generated in a high position do not. I show that these structural positions do not align with expectations from work on weak island effects: it is the syntactic position, not the semantic type, of the wh-phrase that matters. This result supports an important insight that the sensitivity to IEs does not necessarily depend on the argument-adjunct difference (Rizzi, 1992; Miyagawa, 2002). I further show how these empirical contrasts are accounted for along the lines of theearlierworkon interventioneffects,whichmadeuseof structural constraints statedover LF representations. 2010 Elsevier B.V. All rights reserved. * Corresponding author. E-mail address: [email protected]. 1 Among various interveners, this paper concentrates on the interaction of wh-phrases and negation/NPIs in Korean, since these form the traditional core of the facts to be explained (see Pesetsky, 2000 for a relevant discussion). 2 Although I present a syntactic analysis along the lines of Beck and Kim (1997) and Ko (2005), as a reviewer notes, the results of the current experiments work equally well with the semantic accounts. 3 There have been some discussions on the idiosyncrasy of there being no IEs for reason why for Korean (Ko, 2005), Turkish (Ko p.c.) and Japanese (Hoji, 1985; Takahashi, 1990; Tanaka, 1997; Kuwabara, 1998; Miyagawa, 2002). 4 The term low vs. high wh-phrase is adopted for reasons to become clear later in section 4.