The purpose of this paper is to present a complete syntactic analysis of the reflexives of English. Most of the work in this area, particularly work in the Government-Binding theory, has focused on one type of reflexive, what has come to be called the “anaphor” reflexive. Thus Koster (1984), in describing reflexives (among other things), postulates a number of essential properties characteristic of a core dependency relation between constituents. This relation, labelled “R”, is illustrated in (1). Its properties are listed in (2).

In a recent reply to my review of The Language Lottery, David Lightfoot (1985) attempts to rebut the claims which I made about the viability of language learning without task-specific innate principles. The basic thrust of Lightfoot’s book is that there are innate linguistic principles which constrain the form of grammars and play a crucial role in language acquisition. I referred to this view as “special nativism” and contrasted it with “general nativism”, the thesis that genetic structuring of the mind is of a more general sort and does not include principles or notions specific to language. I suggested that special nativism is linked to a particular syntactic theory (transformational grammar) and that a theory of language learning more compatible with general nativism could well emerge from a different type of syntactic analysis. As an illustration of this, I briefly outlined two counterproposals, one pertaining to the interpretation of the indefinite pronoun one and the other to binding theory. In his reply to my review, Lightfoot misinterprets my suggestions, pointing to supposed inadequacies which would undermine the more general point I advanced.

The aim of this paper is twofold. First, I shall propose a unified account of the subject doubling construction and point out its similarities and differences with respect to object doubling — the so-called clitic doubling. The discussion is based on Trentino, a northern dialect of Italian, and Colloquial French. I will argue that it is not desirable to establish a connection between subject doubling and free inversion as is implicitly assumed in Safir (1982). Rather, it seems preferable to assume that the subject clitics are generated in the AGR element of INFL and that they are associated with pro in subject position. The possibility of subject doubling is then related to a parameter making reference to the optionality of Case absorbtion or spell-out by the clitic.