Retour vers liste

Détail de la contribution

Auteur: Chris CUMMINS

How not to trigger a presupposition

Abstract/Résumé: Presuppositions are traditionally held to project from under the scope of negation. However, in cases such as (1) and (2), this does not occur. The presuppositions are said to undergo ‘local accommodation’ – that is, they are satisfied in a local domain and do not project globally. (1) I didn't realise that whales are fish, because whales are not fish. (2) John hasn't quit smoking; he never used to smoke. Recent experimental approaches to this topic address whether local accommodation is more effortful than global accommodation, and whether the process of comprehension involves the projection and cancellation of a global presupposition. This approach parallels the experimental treatment of scalar implicature (Bott and Noveck 2004, and many others). In my research I supplement this by looking at the role of the speaker. I argue that the use of typical presupposition triggers is disfavoured for sound pragmatic reasons in cases where the speaker does not wish to commit to the presupposition, i.e. where it is intended to be 'locally accommodated'. However, this preference competes with general considerations of discourse relevance, which encourage a speaker to re-use a specific lexical item in order to challenge it, thus transparently linking the correction to the existing discourse content. I present experimental data from the domain of quantification in support of both these usage preferences. In cases such as (1) and (2), I argue that the felicity critically depends on the presence of the presupposed content in the prior discourse context. I discuss recent experimental data in which felicity judgments are elicited for discourse fragments in which the presupposition is absent from the prior context (and accommodated when the trigger is encountered) versus those in which it can be inferred from the prior context (and is merely confirmed by the presence of the trigger). Discourses in which the presupposition is already present (but 'not at issue') are judged more felicitous than those in which it is not. Under this analysis, cases of 'local accommodation' can in fact be construed as involving the denial of existing background content. Moreover, the hearer is expected to use contextual information to determine how the presupposition should be treated. I discuss the implications of this for experimental research, arguing that the paradigms used potentially confound three distinct tasks: the actual performance of the semantic/syntactic operation of 'local accommodation' (the desired object of study), the process of imagining a context that justifies the use of a presupposition trigger, and the cancellation of the presupposition that exists within this licensing context. I consider how we might model and investigate the process of accommodation under these fresh assumptions.