Syntactic patterns of embedded wh- clauses

Abstract

This thesis investigates constructions commonly referred to as embedded or indirect questions. Evidence is presented (in the form of several diagnostics) that shows a clear behavioral distinction among constructions that are canonically analyzed as being identical. Specifically, these diagnostics (e.g. polarity item licensing, clause fronting, etc.) show that one class of verb (including wonder and ask) embeds clauses that behave like true questions, while another class (including discover and know) embeds clauses that do not behave like questions. The non-question clauses under discover-class verbs are analyzed as free ("headless") relative clauses (DPs) while the question clauses under wonder-class verbs are analyzed as true embedded questions (CPs). The structural distinction between these homophonous clauses is supported by positive empirical evidence. The ultimate implications of this thesis are a restriction of the term "embedded/indirect question" to wonder-class complements and an expansion of the literature on free relatives to include discover-class complements.