Back to list

Detail of contribution


On the encoded content of logical connectives: a cognitive reanalysis

Abstract/Résumé: Relevance theorists have for quite some time pursued the argument that discourse markers encode procedures, i.e. instructions which guide pragmatic inference by creating cognitive "shortcuts" that the hearer takes advantage of during utterance interpretation. For example, 'but' has been proposed to encode an instruction to process the clause it introduces as contradicting and therefore eliminating an assumption that should be readily available to the hearer at that processing stage (e.g. Blakemore 2002). This type of meaning is clearly different in nature from the conceptual (denotational) one that open-class words are generally taken to encode within the framework. Given the conviction that what is important in a psychologically plausible account of utterance interpretation is the “kind of cognitive information” that a linguistic expression encodes, rather than whether it “contributes to something with truth conditions” (Blakemore 2000:464) and the amount of research that has been conducted on the meaning of discourse connectives, it is quite surprising that relevance theorists have only minimally dealt with what logical connectives encode, accepting pretty much without argument the traditional truth-tabular approach to their meaning and even assuming, as Wilson recently mentioned (2011:24), that they are essentially conceptual encodings. In this talk, I wish to reconsider this view by approaching the relevant terms along procedural lines. My argumentation will be based on the idea that, from a cognitive perspective, there is no particular reason to differentiate between logical and discourse connectives, since both types of expressions can be taken to have evolved for argumentation rather than reasoning purposes (e.g. Sperber 2001). What is more, the analysis of logical operators in truth-functional terms may have been pivotal for the philosophical investigation of their semantic content, but there is again hardly any reason to assume that the meaning of their natural-language counterparts is directly inherited from them when cognitive considerations enter the picture; hence the fact that in some languages the meaning of a single logical operator is carried by more than one words, which are nevertheless used to mark semantic distinctions that are unidentifiable from a logical point of view (e.g. Mauri 2008). Against this background, I will suggest that the relevant natural language terms should be investigated on a par with discourse connectives, since both types of expressions are effectively used to mark some relation between two or more propositions in everyday conversation. Finally, I will discuss the position that logical analysis can and should occupy in this picture.