Limites Et Possibilites De La Reconstruction Syntaxique

In the nineteenth century, reconstruction of syntactic relations was neglected by genetic comparative grammar, because the neo-grammarians concentrated on sound change mainly. Nowadays scholars are trying to fill the gap; but, since they encounter many theoretical problems, which some of them even believe not to be solvable, reconstruction of syntactic relations has still a long way to go. The purpose of this paper is to sort problems out, to analyse the obstacles that arise especially when reconstructing syntactic relations and to show, on the basis of Romance evidence, how these obstacles could be overcome. The fact that a reconstructed proto-language is "langue" only, never "parole," and the fact that syntactic relational features do not evolve in a regular way are the main obstacles. The means to overcome them are firstly the reconstruction of structures rather than of isolated elements, and the reconstruction of their evolution in time and space within the proto-language (provided the external history of the language family is known) and secondly, perhaps to a lesser degree, language universais and diachronic tendencies (in the absence of diachronic regularities).