Aspect-Switching and Subordination: the Role of It-Clefts in Discourse

In an attempt to construct a general explanation for botb sets of cases, we examine Prince's [1978] suggestion that it-clefts serve to mark a piece of information as KNOWN FACT. We propose that tile ASPECTUAL EFFECT of using a cleft both explains tbe Known Fact Effect observed by Prince, and accounts for the various kinds of discourse subordination associated with it-clefts. We then turn our attention to a third set of data: it-clefts that have a contrastive effect in the discourses in wbicb they appear, and what goes wrong with those discourses when they are de-elefted. In some cases, de-clefting causes no ill-effects. In other cases, however, no contrastive relation can be retrieved upon de-clefting and the discourse becomes incoherent. In the final section of the paper, we make some speculative comments on an apparently related phenomenon: the fact that it-clefts cannot take it as the clefted constituent, which we feel is amenable to a discourse-structural explanation.