¹ This paper is the development of work done by both authors between the years 1985 to 1987. The work was presented at WCCFL VI and a much briefer version appeared in the proceedings. The body of the paper remains faithful to the work at this time, but because of theoretical changes in the field since 1987, and recurring questions, the paper has been updated and embellished in footnotes and the appendix. The first author is grateful for support from the following research grants: FCAR 91-ER-0578 and SSHRCC 410-90-0523.

of nominative Case, and in Section 6, we sum up our results. Finally, Section 7, the appendix, represents an attempt to bring our view of Case into the context of Larson’s (1988) phrase structure for VPs.

² We are only concerned with null Case in this paper since it is only null Case which is subject to the ECP. Lexically realized Case, we claim, is much freer since it is only restricted by a “complement of” relation.

³ Instances (pointed out to us by Richard Kayne) where that deletion is possible in a non-adjacent clause we assume are due to reanalysis as shown by the awkwardness of that deletion when the intervening material has been extracted.

1. (i)

   (i) We should tell them (that) Mary will win.

2. (ii)

   (ii) Who should we tell ??(that) Mary will win.

⁴ Fillmore(1968) also assumes that NPs contain a category K, however, in his system K is not the head of the nominal projection in X’-terms. It also bears a much closer relation to theta-roles. Lefevre and Muysken (1988: Chapter 5) suggest that there is a “COMP-like CASE position” at the X’” level of an NP which is much closer in spirit to our proposal.

⁵ Fabb uses matching of features for both affix hopping and case assignment rather than feature transmission. In other respects, however, our correlation of case and verbal affixation owes much to his account.

⁶ Prepositions in other languages such as à in French (Jaeggli 1982) also lend themselves to this K analysis.

⁷ See Cowper (this volume) for another way of looking at wanna-contraction.

⁸ We leave open the question of wh-movement in other languages. Obviously there must be some parameterization in this area.

⁹ This is similar to Borer’s (1986) version of I-subjects.

¹⁰ Vainikka (1986) exploits this correlation but arrives at a very different solution.

¹¹ Pollock (1989) argues that the apparent lack of adjacency in French is due to the head movement of the verb. In cases, however, where a participial form of the verb is used, it is not as clear that verb movement has occurred. There must be another account, then, for these facts. The interaction of adverbs and adjacency will be discussed in the Section 7.

¹² See Guilfoyle (1985, 1990) where it is argued that modals such as tharla are base generated in INFL and lexically select quirky case (QC) marked subjects.