Relevance and Linguistic Meaning: The Semantics and Pragmatics of Discourse Markers (review)

Relevance theorist Diane Blakemore here undertakes a provocative reassessment of the bedeviled category ‘discourse marker’. The book’s subtitle will draw the right audience; however, B’s ultimate aim is to show that ‘there is no justification for writing a book about discourse or discourse markers at all … from the point of view of a cognitively grounded theory of utterance interpretation’.

B approaches her topic by reexamining two assumptions that have informed much research on discourse markers: that they are best characterized as non-truth conditional and thus fall exclusively in the purview of pragmatics, and that they connect units of discourse.

Ch. 1, ‘Meaning and truth’, questions the view that semantics should be concerned with truth conditions and pragmatics then defined as ‘meaning minus truth conditions’ (Gazdar 1979). B notes that this popular formula spreads pragmatics across the linguistic/nonlinguistic divide, lumping together the encoded meaning of expressions like but with inferentially derived implicatures. She argues that Gazdar’s formula, in identifying semantics with truth conditions, runs against extensive evidence (summarized in Carston 2002) that semantic representations provided by the grammar underdetermine the propositional content of utterances and therefore cannot bear truth conditions. From B’s relevance-theoretic perspective, propositional content is extensively context-dependent, and this dependency cannot be handled by a ‘pragmatic competence’ which, being unable to integrate linguistic with nonlinguistic knowledge, could not explain how utterance interpretations are actually recovered. For the recovery task, B enlists the cognitive performance mechanisms posited in relevance theory (Sperber & Wilson 1995), by which the hearer takes input from the linguistic parser, from the senses, and from memory, and inferentially derives representations of speakers’ communicative intentions. Truth conditions remain, but in such a framework attach not to semantic representations but to conceptual representations of thoughts.

In Ch. 2, ‘Non-truth conditional meaning’, B introduces a range of non-truth conditional expressions and discusses problems in determining truth conditionality; she then turns to problems with the speech-act-theoretic distinction between describing and indicating, focusing particular attention on H. Paul Grice’s use of this distinction in formulating the controversial notion ‘conventional implicature’. B argues that there is no significant difference between claiming that an expression like but or so carries a conventional implicature and claiming that it is non-truth conditional: neither claim illuminates what it is that the expression contributes to.

Ch. 3 outlines a more explicit, cognitively based approach to encoded meaning. The chapter opens with an introduction to relevance theory emphasizing features of the theory that bear on B’s claims (§3.1–3.2). Though not intended as comprehensive, these sections stand well as a coherent short presentation of the theory. On the view sketched here, pragmatic inference is served by the linguistic system rather than auxiliary to it and is crucial in determining both implicit and explicit utterance content. In §3.3, B introduces the conceptual/procedural distinction, which emerged from her earlier work and now figures routinely in discussions of discourse markers both in and out of the relevance framework. Roughly, this distinction provides that a linguistic expression may encode either a constituent of a conceptual representation or information that guides the inferential processing of an utterance. This is a cognitively based distinction between two types of linguistically encoded meaning and is conceived of as crosscutting the distinction between truth conditional and non-truth conditional meaning. B claims that the conceptual/procedural distinction is fundamental in semantics, and that since items referred to as discourse markers appear to fall on both sides of the fence, the unity of the discourse marker category is left without a cognitive leg to stand on. The discussion in this chapter is necessarily provisional since, as B acknowledges, the precise nature of procedural meaning remains under negotiation; however, B’s comments here and in Ch. 4 greatly help in clarifying the general grounds for a (suitably elaborated) notion...