I mean, I truly believe there exists some combination of words. There must exist certain words in a certain specific order that can explain all of this, but with her I just can't ever seem to find them. --- WALTERWHITE(Bryan Cranston), in AMC’s Breaking Bad, Season 3 Episode 10 (‘Fly’)
Relevance theory (RT) makes assumptions which are broadly similar to Recanati’s model. As with Recanati, the RT point of departure is psychological rather than philosophical. RT assumes that ‘what is said’ is very different from what is semantically encoded – in fact, it introduces new terminology in order to explicitly separate itself from the Gricean framework. The outcome of the semantic decoding process is called the ‘logical form’ – this is an incomplete, subpropositional representation, computed entirely through “linguistic decoding processes” (Carston 2002a: 9). Contextually, it is developed into the ‘explicature’ in roughly the same way as Recanati’s ‘what is said’ is
derived – not only through reference assignment and disambiguation, but also through the other types of pragmatic enrichment outlined above (Wilson & Sperber 2004: 615; see Carston (2002a: 21-28) and (2010) for discussion). Implicatures are also derived in much the same way as they are under Recanati’s account: they are generated by using the explicature as one of the premises (and combining it with contextual assumptions which are available to the hearer) (Hall 2009: 112). Explicatures, on the other hand, are developed ‘locally’, as they are on Recanati’s account – free enrichment and saturation, for instance, modify “subparts of the linguistic logical form”. Implicatures, by contrast, are derived ‘globally’ – they take the entire explicature as a premise in the inferential process (Carston & Hall 2012: 69). In both cases, inferential processes play a role – in the case of explicatures, to develop the logical form; in the case of implicatures, to derive the implicitly communicated assumptions using the explicature as a premise (Blakemore 2002: 74).
There do not seem to be many differences between RT and Recanati’s account (see Huang 2007: 241). However, RT rejects the assumption that there is a difference between the way in which explicatures (Recanati’s ‘what is said’) are derived on the one hand, and implicatures on the other. Recanati argues that the first process is associative and subdoxastic, and the latter properly inferential and conscious; RT argues that both processes are subpersonal (that is, unavailable to consciousness) but properly inferential (Sperber & Wilson 2002/2012: 267; Carston 2010: 267-268). Recanati also assumes that his primary and secondary pragmatic processes occur sequentially – first, ‘what is said’ is derived, and only then can implicatures be generated. RT, on the other hand, assumes that there is a ‘mutual adjustment process’ here – the explicature, of course, determines which implicatures are derived, but the derivation of implicatures also enables the hearer to adjust his explicature (Carston 2002b: 142-143).32 So-called
‘bridging implicatures’, for instance, necessarily precede the derivation of the (full) explicature:
(28) a. Jane has a new house. The front door is blue. (from Wilson & Matsui 2000/2012: 187)
In (28a), ‘the front door’ is obviously part of the ‘new house’ which Jane has bought. However, this is an implicature of the utterance:
(28) b. Jane’s new house has a front door.
32 Carston (2007) notes that Recanati’s more recent work accepts the ‘unitary’ account which RT provides with
(28b) is trivial, of course, but it is still an implicature which has to be derived in order to determine the full explicature for (28a):
(28’) a. Jane has a new house. The front door [of that new house] is blue.
In (28a’), free enrichment has taken place, but it has taken place under the guidance of the bridging implicature in (28b). A bridging implicature, as such,
“is a contextual assumption, warranted by the explicit content of previous discourse, needed to introduce an intended referent which has not itself been explicitly mentioned.” (Matsui 2000: 199)
(28), then, provides evidence for the assumption that the derivation of explicatures and implicatures cannot be strictly sequential – implicatures and explicatures can be derived in parallel (Carston 2002b: 143; also Carston 2007).33
The most important difference between Recanati and RT, however, is the guiding principle underlying the derivation of ‘what is said’/explicatures and implicatures. According to Recanati, ‘what is said’ is derived through considerations of accessibility, and implicatures through conscious reflection on speaker intentions. According to RT, both explicatures and implicatures are derived through the same inferential process, and both are guided by the same heuristic. As such, Recanati’s pragmatics is binary, while RT’s is unitary (Carston 2002b: 141-143). RT assumes that all pragmatic processes are handled by a single “pragmatic comprehension system”, guided by the following heuristic:
(29) a. Follow a path of least effort in computing cognitive effects: Test interpretive hypotheses (disambiguations, reference resolutions, implicatures, etc.) in order of accessibility.
b. Stop when your expectations of relevance are satisfied (or abandoned). (Wilson & Sperber 2004: 613)
(29) is known as the relevance-theoretic comprehension heuristic, and it underlies all aspects of utterance interpretation. It retains the core of Recanati’s accessibility principle, but matches it up against a stopping rule (relevance) which ensures that not just the most accessible interpretation is chosen, but the most accessible interpretation which is acceptable. RT is still very much focused on speaker intentions: the hearer has
33 Note that this does not mean that the derivation of implicatures can be a local process – a bridging
to derive an interpretation which fits in with the communicator’s abilities and preferences (Wilson & Sperber 2004: 612; Sperber & Wilson 1995²: 268-269), but he will derive the one which is most accessible (that is, least effort-intensive) given those parameters. More specifically, the hearer will accept the first interpretation which comes to mind and which satisfies his expectations of relevance. How should these expectations be understood?