Interrogative Possessors and the Problem with Pied-Piping in Chol*

In this paper I present new data from possessive phrases in Chol (Mayan) and discuss the problem they present for standard analyses of pied-piping (see Heck 2004 and references cited therein). I argue that a theory of pied-piping in which features of a wh-word “percolate” up to a higher maximal projection is unable to straightforwardly account for the ordering facts found in interrogative possessive constructions in Chol (also described for Tzotzil (Aissen 1996) and San Dionicio Zapotec (Broadwell 2001)). It will be shown that in certain derivations, feature percolation would have to both occur and not occur from the same wh-word at different stages in a single derivation. A derivation in which percolation happens consistently at each step results in ungrammaticality. I argue that these facts provide evidence against an account involving feature percolation.1 I adopt instead an analysis along the lines of that proposed in Cable 2007, in which feature percolation is eliminated from the grammar and wh-movement to Spec,CP is always the result of a relationship between C and a projection called Q(uestion)P. Under this analysis, so-called pied-piping is simply an instance of more familiar phrasal movement. Furthermore, while additional stipulations are required to prevent incorrect ordering under a feature percolation story, it will be shown that in the QP account the ungrammatical forms are ruled out as a natural consequence of the semantics of Q. I propose further that the apparent free choice between possessor extraction and pied-piping constructions in Chol can be explained as a result of where the Q head is merged, rather than by a special operation of feature percolation.