There are three ways to refer to a fact from the complement of afactive verb: (1) Via abstract object anaphoric reference, or, witha full sentential complement that will be interpreted either (2) asa bound presupposition or (3) as triggering a presupposition of afact that will have to be accommodated. Spoken corpus examplesreveal that these three possibilities differ in relation to thetype of information they tend to contribute, and this has twoeffects. First, the information status of the fact and its role (...) inthe discourse seem to affect the preference for one constructionover another in a particular context. Second, presupposed factivecomplements that need to be accommodated tend to be hearer-new andthe focus of the utterance, meaning that information structureseems to contribute to the felicity of accommodation ofpresupposed facts. (shrink)

We argue that contrastive statements have the same underlying semantics and affect the context in the same way as denials. We substantiate this claim by giving a unified account of the two phenomena that treats contrast as a subtype of denial. This analysis crucially requires a dynamic semantics view of context-dependence with a multi-dimensional representation of information.

We present a unified treatment of contrast and denial as slightly different instantiations of the same discourse schema. Both denial and contrast are analysed as involving a revision operation, what sets them apart is merely the type of information being retracted. The formal analysis requires a representational framework that separates different types of information and is therefore implemented in Layered DRT. One of our selling points is the account of the uses of rectification vs. contrastive particles (like German sondern/aber) we (...) get for free with our unification. (shrink)

Some apparent differences: (i) Denials are essentially a dialogue phenomenon as is obvious from the fact that (ii) denials taken out of their dialogue context are often plain contradictions (Horn, 1989), and for this reason (iii) their analysis necessarily involves nonmonotonic operations. Contrast on the other hand is (i’) a discourse relation frequently occuring in monologue, (ii’) never involving overt contradictions (*I am hungry but I am not hungry)1 and therefore (iii’) often treated as an essentially monotonic phenomenon: what licenses (...) a contrastive conjunction is not overt contradiction, but rather a conflict between what’s defeasibly implied by the first and second conjuncts. (shrink)