Wh-words have been claimed to behave like existentially quantified expressions, or universally quantified expressions, or definite descriptions, or indefinites. The main reason for such a heterogeneous set of proposals is that wh-words have been almost exclusively studied in relation with interrogative clauses. But our semantic intuitions about interrogatives are not particularly sharp to begin with. For instance, what are our intuitions about the semantic contribution of "what" in "What did Adam cook?" or "Jie wonders what Adam cooked?"
In this talk, I suggest a different approach to this problem. We can better understand the semantic behavior of wh-words like "who", "what", "where", "when", "how" and their equivalents crosslinguistically, if we look at them in non-interrogative constructions like "free relatives", about which we have better semantic intuitions (e.g. "Jie ate what Adam cooked"). Different kinds of free relatives are presented and discussed from more than twenty languages from three different language families.
I will discuss the reasons that support the existence of QR as an operation in the LF component of the syntactic module. I will then discuss in detail some of the previous theories of QR and show some drawbacks that need to be fixed. After offering a concise introduction to Chomsky's (1999/2001) theory of phases, I will then defend a version of this theory in which access to the phonological component is cyclical but access to the semantic component is not. I will finally look at the locality conditions of QR in English and Italian and propose a comprehensive theory of QR within the version of the theory of phases introduced.