NEG-raising does not involve syntactic reconstruction

Abstract

NEG-raising concerns the phenomenon by which certain negated predicates (e.g. think, believe, hope) can give rise to a reading where the negation seems to take scope from an embedded clause. The standard analysis in pragma-semantic terms goes back to Bartsch (1973) and is elaborated in Gajewski (2005, 2007), Romoli (2013) and others. Recently, this standard approach has been attacked by Collins and Postal (2014), who argue, by providing some novel arguments, that NEG-raising involves syntactic movement of the negation from the embedded clause into the matrix clause. The syntactic structure of ‘I don’t think you’re right’ would then be: I do[n’t]i think you’re ti right, and the NEG-raising reading would result from syntactic reconstruction of the negation. In this talk I present four novel arguments against this account. First, following up work by Horn (2014), I show that Collins and Postal (2014), and their reply to Horn (Collins & Postal 2015), predict that every negated predicate that can license so-called Horn clauses (non-negative clauses containing NPIs and subject–auxiliary inversion) should receive a NEG-raising reading, contrary to fact. Second, Collins and Postal (2014) adopt phonological deletion of negative operators – a necessary ingredient for their account – but this cannot be independently motivated. Third, it turns out that for certain constructions, Collins and Postal (2014) must also allude to the original Bartschian approach. Finally, I demonstrate that the standard approach actually explains the grammaticality of Horn clauses better than the syntactic alternative presented by Collins and Postal (2014).