Lingvisticæ Investigationes - Volume 7, Issue 1, 1983

The semantic representation of adjectives constructed with the suffixes -phile and -gène (i. e. the combination of arguments and predicates underlying the surface form) is examined here. It may lead to interesting observations about the surface use of such adjectives, as well as of others.

The use of 'support-verbs' is not specifically limited to Modern French. Old French displays some striking similarities on this point with further development of the French language. This is what emerges from the study of the verb doner {donner) throughout a corpus regrouping the works of Chrétien de Troyes. In many cases the relation shows both a morphological connection and a semantic equivalence, e.g. doner consoillconseillier {i.e. to advise). But the semantic equivalence may appear regularly without any morphological connection, e.g. doner cop/ferir {i.e. to strike). A large selection of substitutes for cop (blow), hyponyma of the head-noun, offer a rich variety of possible expressions. Such pairs tend to form semantically related groups. The structure base on the 'support-verb' yields less ambiguity than the one using the derived verb. Several verbs in Old French including bailler, which is partly synonymous with doner, may imply the meaning of "recevoir" as much as that of "donner". This construction falls within the scope of general linguistics and constitutes a permanent feature, which is noticeable throughout the different stages of the evolution of the French language.

A recent attempt within the EST framework to account for the facts of the French causative construction is that of Rouveret and Vergnaud 1980. Here, we discuss two major flaws contained in their proposal and the relevance of the issues we raise to the Opacity Condition of Chomsky 1980. One of these flaws involves the various devices which are used to account for the distribution of y and en in the causative construction. Rouveret and Vergnaud fail to recognize that there are significant differences among these devices with respect to core grammar. The other problem with their proposal that we discuss concerns a serious mistake in overgeneration. In addition, we point out that the Opacity Condition of Chomsky 1980 allows an ambiguous interpretation of the notion 'domain of subject' and that the French causative construction provides a set of empirical facts which suggest a resolution of this ambiguity. This is accomplished through a slight modification of the OC. Furthermore, we show that there is a close relationship between this reformulation and the devices used to account for y and en. Only when certain of the devices used to account fory, en are recognized as being outside of core grammar is an empirically adequate account possible.

This paper is an attempt to circumscribe the set of English prefixes. The coverage of the lexicon is intended to be as complete as possible; i.e. the author's goal is to write a grammar of prefixes with extensive coverage both of phenomena and of words.The type of grammar which is being built should relate sound and meaning under superficially varied conditions. In particular, although distinctions are made in the lexicon in terms of Greek, Germanic and Latin (or French) types of words, the rules should be general. A number of morphophonologi-cal derivations of prefixed words are given which illustrate the approach.

The present paper has two intimately connected goals; It aims at contributing to Chomsky's "Government Binding Theory" and also at providing a fairly detailed comparative analysis of French and English impersonal constructions. Its contribution comes under the guise of (a) an Agreement Theory (see section 2), (b) a general constraint on impersonal chains (see (72)) and (c) a new nominative Case assignment rule: it is suggested that in French and Italian (but not in English) "ergative verbs" (in Burzio (1981) 'sense) can assign nominative Case to their "object". Furthermore, as has become standard in recent comparative work in the GB framework, the paper attempts to isolate the parameters that are responsible for the minimally distinct properties of the constructions under investigation. It is shown here that they can be traced back to the interplay of a Case parameter, the morphological properties of expletive elements (il vs there) and the properties of Universal Grammar.