The Semantic Contributions of Wh-words and Type Shifts: Evidence from Free Relatives Crosslinguistically

Free relatives (henceforth, FRs) are embedded non-interrogative wh-clauses like what Adam cooked in lie tasted what Adam cooked. They are attested crosslinguistically. So far, I have found them in twenty-eight languages besides English. The main goal of this paper is to show that a crosslinguistic investigation of the semantic behavior of FRs sheds light on the semantic contribution of wh-words and gives further empirical support to the existence of type-shifting rules in the grammar whose purpose is to deal with type-mismatches. The paper is structured as follows. First, I give a definition of FRs and show their crosslinguistic distribution (§ 2). Then, I move on to investigate their semantic behavior (§ 3). The main conclusion is that FRs follow three different semantic patterns. There are FRs that always exhibit maximality, like definite descriptions (§ 3.2), there are FRs that semantically behave like narrow scope indefinites and never exhibit maximality (§ 3.3), and, finally, there are free relatives that may or may not exhibit maximality (§ 3.4). In the last part of the paper (§ 4), I sketch a semantic analysis according to which all FRs denote a set and phrasal wh-words like who, what, where, when, how (and their crosslinguistic equivalents) behave like set restrictors . The different semantic behaviors among the different kinds of FRs result from the presence or the absence of a type-mismatch between the denotation of a FR (i .e. a set) and the semantic requirements of the matrix clause. I argue that, if a type-mismatch occurs, it is repaired by the type shifters iota (t) or existential closure (3), which Partee ( 1987), Chierchia ( 1 998), and Dayal (2004) have argued for in order to account for a completely different set of data (i .e. the semantic behavior of DPs like bare plurals and bare singulars) .