Peter W. Culicover and Wendy K. Wilkins, Locality in linguistic theory . New York and London: Academic Press, 1984. Pp. viii + 276.

Clearly there is a close relationship among these three items, and a claim made about any one of them usually has consequences for hypotheses about the other two. This is by now reasonably well established for Universal Grammar (UG) and particular grammars, but there has been a curious silence about the triggering experience. Current literature describes' parametric differences' between, say, English and Chinese, but it almost never mentions how the parameter would be fixed, what the triggering experience would need to be for the Chinese or English child. Despite this incomprehensible omission, there is an intrinsic relationship among the units in (1). If the Primary Linguistic Data (PLD) were rich and well organized, correspondingly less information would be needed in UG, and vice versa. These are not aesthetic swings and roundabouts, and there are facts of the matter which limit viable hypotheses. It is clear, for example, that the PLD for a particular child are in many ways less rich than information found in a typical issue of Linguistic Inquiry, which includes non-occurring sentences, data about ambiguity, analogous constructions in other languages, and much more. In addition, UG must be able to support the acquisition of any human grammar, given the appropriate triggering experience, and grammars must be not only attainable but also usable for purposes such as parsing and speech production, appropriately vulnerable for the kinds of aphasias that one finds, and they should provide part of the basis for understanding the developmental stages that children go through. Of course, UG need not be seen as homogeneous, and may emerge piecemeal, parts of it being available maturationally only at certain stages of a child's development. There is no shortage of empirical constraints on hypotheses about (1). Work on UG over the last fifteen years has focused on locality restrictions, of the form that no rule may affect items in separate clauses unless the lower one is contained in the COMP or is the subject of an infinitive. More formally: