Back to list

Detail of contribution

Auteur: Chris CUMMINS

Co-Auteur(s): J. P. DE RUITER, Bielefeld University, Germany

Multi-level contextual effects on the scope of negation

Abstract/Résumé: A lengthy debate in the recent literature concerns the relation between the semantic and the syntactic behaviour of negation. Musolino et al. (2000) argue that language users preferentially take the semantic scope of negation to match its syntactic scope – e.g. “some Xs are not Y” to mean “there exist Xs that are not Y” rather than “it is not the case that some Xs are Y”. Conversely, Gualmini et al. (2008) argues that this apparent “preference for isomorphism” is a preference for the interpretation that best answers the Question Under Discussion (QUD). In this presentation we distinguish two levels of contextual influence on the choice of expression. The approach of Gualmini and colleagues operates at a high level, in that it requires the language user to correctly identify the QUD and respond accordingly. However, there is also evidence that lower-level priming effects influence the choice of utterance in a given discourse context (Branigan et al. 2000, Pickering and Garrod 2004). Speakers tend to reuse syntactic constructions and lexical items that are contextually activated. Priming effects of this type have recently been argued to influence the usage of numerically-quantified expressions (Cummins et al. 2012). We extend this approach to the interpretation of scopally ambiguous negation, and posit an analysis in which a general preference for isomorphism interacts with low-level priming. We support this argument by appeal to new production data, in which participants are asked to correct erroneous descriptions of visual displays, and display both a tendency towards isomorphism, where applicable, and distinct influence from the preceding linguistic context, which cannot easily be attributed to QUD or similarly high-level discourse factors. Finally, we discuss the implications of this analysis for metalinguistic negation. This approach distinctively predicts that there is not a clear line between echoic and non-echoic usage; instead, certain linguistic structures are preferentially ‘echoed’. There is independent evidence that hearers are able to take this usage preference into account when drawing pragmatic enrichments – for instance, that “more than 90” is understood to mean “between 90 and 100” unless 90 is a contextually salient number. However, the hearer cannot generally be certain whether a given instance of usage reflects priming or whether it is a free choice on the part of the speaker (and thus admits pragmatic enrichment). We consider the possibility that metalinguistic negation may be seen as the end-point of a continuum of priming, at which the hearer may be entirely certain that the speaker’s choice of expression is echoic and may therefore confidently suppress implicatures.