I would like to begin by acknowledging the invaluable support of my advisor, Mark Johnson, whose level of knowledge and expertise I can only hope to achieve someday. Special thanks are also due to my other committee members: to Tom Griffiths, for arriving at Brown in the nick of time (among many other feats), and to Katherine Demuth and John Goldsmith, for providing valuable advice and additional perspectives. Eugene Charniak deserves recognition for his long-ago formative role in my research career, and for continued advice and discussion since then.

My time at Brown has been enriched by the presence of many wonderful student colleagues. I am grateful to the many past and present members of the Brown Laboratory for Linguistic Information Processing (especially Joe Austerweil, Don Blaheta, Will Headden, Matt Lease, David McClosky, Brendan Shean, and Jenine Turner), who provided friendship, stimulating discussion, and feedback on various drafts of this thesis. I would also like to thank Mr. McFrench, for his inspiring professionalism in all things; Frank Wood, for late-afternoon math cram sessions; Katherine White and Naomi Feldman, for helpful feedback on the many presentations of this work that they witnessed; and Julian Wong, for keeping me on my toes.