¹ I would like to thank the members of the Haitian Creole Project at the Université du Québec à Montréal, where I was a visitor during part of the writing of this paper. I am also grateful to Barbara Brunson, Jila Ghomeshi, Diane Massam, the members of the Toronto Syntax Reading Group, the editors of this volume, and in particular Betsy Ritter, for helpful discussion. This work was partially supported by grant number 410-89-1529 from the Social Sciences and Humanities Research Council of Canada.

² I assume that not in English is an adjunct element, not the head of a projection. This makes it more like French pas than like ne. For Pollock (1989), ne heads NEGP, and pas is its specifier.

³ Beukema and Coopmans (1989) give a different analysis of matrix imperatives, focussing primarily on the properties of the subject position.

⁴ Tremblay (1990) claims that have is a copula, inserted only to assign case to an argument.

⁵ This account of the meaning of the have to construction points the way to a straightforward account of the rather similar be to construction, illustrated below:

Mary is to wash the car before she goes to the beach. A full account of this construction is beyond the scope of this paper.

⁶ I am ignoring the differences between bridge verbs like believe, which allow a null complementizer, and non-bridge verbs like mutter, which require a lexical complementizer. A discussion of this issue would take us outside the scope of this paper.

⁷ Borer (1989) argues that while if is a complementizer, whether, like the other wh-phrases, appears in the CP-specifier. This does not affect anything in this paper.

⁸ Note that (25b) is expresses an obligation not to leave, not an absence of obligation. Thus we are not dealing with matrix negation and the so-called British have shown below:

I haven’t time to talk to him.

Rather, the negation is situated in the infinitival clause.

⁹ I am informed that for at least one speaker of a British dialect, I haven’t to leave can mean ‘I must not leave’. While this apparently contradicts the analysis presented here, there is the possibility that have, for this speaker, exhibits the phenomenon formerly known as neg-raising. I have not yet investigated this possibility.

¹⁰ Palma dos Santos (1989) proposes an analysis of Portuguese inflected infinities involving SA but no T, in which SA alone is involved in nominative Case assignment. See also Raposo (1987) for a treatment of this phenomenon.

¹¹ Haik, (1985) argues for a similar treatment of some sentences in French which superficially appear to contain relative clauses, but which, she claims, involve a small clause whose subject is the immediate post-verbal NP, and whose predicate is the apparent relative clause. An example is the following:

J’ai ma fille qui va à l’école.