A crosslinguistic study of postposing in discourse

In this paper we examine constraints on the use of seven sentence-types permitting the non-canonical appearance of the logical subject in post-verbal position: inversion in English and in Farsi, presentational and existential there-sentences in English, presentational ci-sentences and subject inversion in Italian, and es+subject postposing in Yiddish. We show that these sentence-types share a common discourse constraint: each requires the NP in non-canonical (i.e. postposed) position to represent information that is unfamiliar in some sense. The discourse function of postposing is contrasted with that of another sentence-type involving post-verbal subjects: right-dislocation in English. Unlike postposed NPs, the marked NP of English right-dislocation represents information that is familiar within the discourse: concomitantly, a pronoun coreferential with the marked constituent appears in this constituent's canonical position. We argue, then, that the function of postposing is to place subjects representing unfamiliar information in sentence-final position. On this analysis the functional difference between these sentence-types and English right-dislocation can be straightforwardly accounted for. Given that the marked NP of right-dislocation is coreferential with an intrasentential pronoun, we would expect this NP to represent a discourse-old entity, as do anaphoric pronouns in general. Thus, it is not accidental that right-dislocation does not serve to keep unfamiliar information out of subject position; the presence of the pronoun rules out such a function.