Monosemy, polysemy, and the French imparfait
Recent years have seen an increasing interest in polysemy as a theoretical model, both in lexical semantics (Blank 1997, 2001) and in the theory of grammar. In particular, it has been claimed that polysemy is a model suitable for the description of all sorts of "procedural" items, such as discourse markers (Hansen 2002), derivational suffixes (Booij 1986, Meibauer 1995) and grammatical constructions (Goldberg 1995, Croft & Cruise 2004). A major problem with many of these approaches is that their concept of polysemy is far from clear. In fact, as has been shown by Hansen (1998), many such approaches really are based on monosemic conceptions of meaning rather than on genuinely polysemic ones. The aim of my paper is twofold. First, I will present a simple model which conceives of polysemy as the synchronic outcome of diachronic change. Second, I will show that polysemy relations in grammar are ruled by particular constraints which have to do with the nature of grammatical meaning, especially with the fact that grammatical meaning is procedural, not conceptual. The empirical phenomenon with which I will be concerned is the French imparfait, a past tense paradigm whose major function is to mark background events within past tense narratives, and which, among other things, is also used in irrealis conditional clauses. Classical analyses (Coseriu 1976) as well as state-of-the-art descriptions (e.g. Berthonneau & Kleiber 1993, 1999), however different in detail, are essentially monosemic in that they try to reduce the various functions of the imparfait to a single, underlying “basic value”. Comparing the French imparfait with other Romance past tense paradigms both synchronically and diachronically, I will present an account which dispenses with the notion of “basic value”, but which, in turn, emphasizes the differences between lexical polysemy on the one hand and grammatical poly-functionality on the other.