(Université Libre de Bruxelles)

The aim of this paper is to question the well-established view in constraint based grammars on Romance causatives, i.e. that they are functionally monoclausal, as well as its most important corollary, that their lexical representation is underspecified with regard to argument structure (Alsina 1996, Abeillé, Godard & Sag 1998) that gets fully specified by merging with another predicate. This is the generally accepted view, even though it contradicts initial claims on lexical integrity. The discussion will rest upon a few problematic sentences of which the monoclausal theories give an unsatisfactory account. I argue that the monoclausality of causatives is only a surface (i.e. a c-structure) monoclausality. The ‘merging effect’ observed at the level of surface realisation, and the apparent extra arguments of the causative predicate will be accounted for by some other means. The idea is to resort to the previous conception of a biclausal f-structure, but with a more ‘complex ’ use of structure sharing. I will mainly be concerned with the implementation of this concept in Lexical Functional Grammar (LFG) in this paper, but a short comparison with a similar Head-Driven Phrase Structure Grammar (HPSG) approach will support the idea that the proposal has some empirical relevance.