Abstract
In this article, we present a selection of the results of a research on 1000 hypothetical si-constructions from the corpus of spoken French OFROM. In a first step, we establish a syntactic taxonomy consisting of three categories: micro-syntactic, macro-syntactic and (pseudo-)independent si-constructions. In a second step, we focus in particular on the first category, distinguishing constructions according to the epistemic profile of the protasis (contingent, temporal-hypothetical, counterfactual and factual reading). We compare two modes of linearization {si P, Q} vs {Q, si P} that have implications for the information structure of these constructions. In a third step, in light of the quantitatively observed trends, we identify prototypes while leaving room for marginal phenomena and ambiguous occurrences.