I am indebted to an extremely large number of people who have contributed ideas and/or judgments of well-formedness to this project. The 1998 Synners discussion group at OSU, in particular Martin Jansche, Nathan Vaillette and Shravan Vasishth, provided valuable critical input and an abundance of robust counterexamples to many of the claims in TIE. Much of this work is an outgrowth of a research program carried out in conjunction with Thomas Hukari and Mike Calcagno, some of whose results are reported in Levine et al. (in press) and Levine et al. (1999). In addition, I have benefited greatly from discussion of this material with Peter Culicover, Shalom Lappin, Geoffrey Pullum and Ivan Sag. I should also like to express my sincere appreciation to the large number of faculty and graduate students in the Department of Linguistics at OSU who served as on-call informants in checking the judgments reported below. None of those cited should be assumed to agree with the author on the views reported in the following, or with anyone else.