Rule inversion, opacity, conspiracies: French liaison and elision

Abstract

The purpose of this paper is to present an analysis of French liaison and elision based on the notion of rule inversion, introduced by Vennemann (1972). It is contended that the synchronic rule for liaison is not the continuation of the historical preconsonantal and prepausal consonant deletion, but rather its inverse, consonant epenthesis in prevocalic position. A number of factors are adduced in support of epenthesis, including relative frequency, simplicity of analysis, acquisition of language, and historical data. Elision, however, is not considered describable by inversion, the diachronic vowel deletion rule here also being motivated synchronically. Therefore, liaison and elision are seen as distinct processes of French, a conclusion unlike that of Schane (1968), but one that is also proposed in Schane (forthcoming), although reached by entirely different means. Finally, liaison and elision are found to form a conspiracy, according ti Kiparsky (1973), because both rules are transparent. A deletion formulation of liaison would make it opaque, and thus inadmissible for a conspiracy. Since the functional unity (conspiracy) of liaison and elision cannot be doubted (they both eliminate the sequence VV), the epenthesis analysis of liaison receives strong theoretical backing.