This book goes to the heart of contemporary pragmatics by addressing the debate between neo- and post-Gricean pragmatics and highlighting some controversial issues in more recent developments of Relevance Theory.
After a concise and clear introduction by the editor, the book proper opens in chapter 2 with Jay Atlas's contribution, ‘On a pragmatic explanation of negative polarity licensing’. This chapter is a straightforward criticism of Laurence Horn's (2002) pragmatic account of the licensing of negative polarity items and his suggestion to replace the notion of downward entailment by that of downward assertion. Atlas gives linguistic evidence against Horn's analysis and points out the theoretical difficulties raised by Horn's account.
In chapter 3, ‘Regressions in pragmatics (and semantics)’, Kent Bach complains that the weaknesses of pragmatics at its beginnings, notably ‘an illicit mixing of pragmatics with semantics’ (25), are still extant. While even a straightforward utterance may have to be enriched to access the speaker's meaning, ‘a sentence has its content independently of being uttered’ (27), which will be recovered through parsing. Pragmatic interpretation is always needed, but this fact should not be interpreted as undermining the pragmatic/semantic distinction. Bach describes nine contemporary approaches which he sees as embodiments of the confusion between pragmatics and semantics, and advocates the term ‘impliciture’, rather than ‘explicature’, for speaker meaning as a way of making the distinction clear.
In chapter 4, ‘Constraints, concepts and procedural encoding’, Diane Blakemore considers the Relevance-Theoretic notion of procedural meaning. After a discussion of the relevance of parentheticals, she reminds us that procedural expressions (i) are not subject to charges of untruthfulness, (ii) cannot be semantically composed with other expressions, (iii) are very difficult to translate or to paraphrase, and (iv) can be dropped from the explicature of the utterance once they have played their role in its recovery.
Reinhard Blutner's contribution, ‘Optimality Theoretic pragmatics and the explicature/implicature distinction’ (chapter 5), tackles the explicature/implicature distinction from the perspective of applying Optimality Theory to pragmatics. Blutner describes two Optimality-Theory-based systems, which are roughly equivalent to the global and the local approaches to implicature, and defends the idea that the local account (or its Optimality-Theoretical equivalent) is derived from the global account (or its Optimality-Theoretical equivalent) via an evolutionary process, which he calls fossilization. Blutner concludes by proposing this process as an explanation for the data presented by Noveck (Reference Noveck2001), which shows that implicatures are less easily accessed by children than by adults.
In chapter 6, ‘Varieties of semantics and encoding: Negation, narrowing/loosening and numericals’, Noel Burton-Roberts reproaches Relevance Theory for advocating an ‘unstable’ linguistic semantics, which provides ‘too little to be consistent with the traditional assumption that particular languages have semantics as ordinarily understood, and too much to be consistent with LoT [Language of Thought] being the sole locus of real semantic properties’ (91). Burton-Roberts proposes two options: either give semantic properties to linguistic expressions as well as to LoT or, alternatively, restrict semantics to LoT and deny it to linguistic expressions. After a critique of the ambiguous notion of encoding in Relevance Theory, Burton-Roberts chooses the second option and defends the ‘Wittgensteinian’ view that linguistic meaning merely amounts to use.
Relevance Theory is also attacked in chapter 7, ‘Relevance Theory and shared content’, by Herman Cappelen & Ernie Lepore, who claim that it ‘implies that content sharing is impossible’ (115). This, according to the authors, renders mysterious the interpretation of speech reports, belief attributions and assessments of assertions. However, the authors acknowledge that ‘communicated content is deeply context sensitive’ (132) and propose, as a solution, to abandon the idea ‘that each utterance of a sentence says … just one thing’ (133). The semantic content of a sentence can then be defined as the common proposition between sets of interpretations that are determined by different contexts.
Marjolein Groefsema's chapter, ‘Concepts and word meaning in Relevance Theory’ (chapter 8), discusses the ambiguity inherent in the Relevance-Theoretical view of the content of concepts (does it correspond to the concept's logical entry, to both the encyclopaedic and the logical entry, or does it constitute an atomic concept?) and the consequence that this has for the explicature/implicature distinction. Groefsema notes that a fourth possibility as to how to define concept content has arisen with the notion of ad hoc concepts. She then turns to the explicature/implicature distinction, adopting cancellability as the distinguishing criterion, and argues that this criterion provides support for the hypothesis that concept content includes encyclopaedic information rather than for the hypothesis according to which concepts are atomic.
In chapter 9, ‘Neo-Gricean pragmatics: A Manichaean manifesto’, Laurence Horn proposes to reduce the four Gricean maxims to two principles (Q and R), which jointly support a ‘division of pragmatic labour’ and constitute a minimax constraint on interpretation. He illustrates the workings of the two principles with a variety of linguistic phenomena – from the avoidance of synonymy and homonymy to different types of pragmatic adjustment through scalar implicatures. Horn points out that Relevance Theory, which defines relevance in the dual terms of effort and effect, is a dual theory in disguise and relies on a minimax algorithm just as Manichaean neo-Gricean pragmatics does.
Ira Noveck & Dan Sperber's contribution, ‘The why and how of experimental pragmatics: The case of “scalar inferences”’ (chapter 9), begins with some insightful remarks about the difference between semantic intuitions (which are themselves semantic facts) and pragmatic intuitions, which are ‘educated guesses … about hypothetical pragmatic facts, but are not themselves pragmatic facts’ (186). The authors argue that an experimental approach can help choose between alternative theories of the cognitive processes behind a (common) interpretation of an utterance. Using the example of scalar implicatures, they describe a number of experimental studies, conducted mainly by Noveck and his colleagues, involving both adults and children. They review experiments by other teams, concluding that all of the experimental data point toward or are at least compatible with a processing cost of scalar implicature computation that is predicted by the Relevance-Theoretic account but not by the theory of Generalized Conversation Implicature.
In chapter 11, ‘Indexicality, context and pretence: A speech-act theoretic account’, François Recanati examines context-shifting and distinguishes contextual features that depend on speaker's intentions (and which can be shifted at will) from those that do not (e.g. the time and location of utterance). The latter can be shifted only through pretence, such as is involved in fiction but also in echoic or reported speech/thought. Finally, Recanati draws a distinction between the locutionary context (the context of the utterance) and the illocutionary context (the context of the speech act, e.g. assertion).
In the final, twelfth chapter, ‘A unitary approach to lexical pragmatics: Relevance, inference and ad hoc concepts’, Deirdre Wilson & Robyn Carston present the results of their ongoing work on lexical pragmatics, outlining the notion of ad hoc concepts. Their view is that pragmatically guided lexical adjustment plays a central role in interpretation, as seen in approximation and neologism, and also constitutes a force in lexical change, when such ad hoc concepts become new lexical senses. Wilson & Carston argue in favour of a unitary inferential (rather than an associative or mixed) account of the lexical adjustment process, pointing out that associations are a type of inference, which means that a unitary inferential account will be more constrained than alternative theories.
In my general discussion, I will concentrate on two issues: (i) the neo-/post-Gricean debate on (scalar) implicatures; and (ii) Relevance-Theoretic developments concerning implicatures and explicatures as well as the nature and role of concepts in explicatures.
Regarding the neo-/post-Gricean debate, the most relevant papers in this collection are Blutner's and Noveck & Sperber's. Blutner considers Relevance Theory to be a local approach to implicatures, which I think is debatable. While Relevance Theory admittedly does not consider the derivation of implicatures to entail comparison between the actual utterance and its possible alternatives, it differs from local approaches, such as, for example, the theory of Generalized Conversational Implicature, proposed by Levinson (Reference Levinson2000), in that it strongly rejects default accounts. Moreover, Blutner's proposal that the developmental data can be accounted for by postulating a fossilization process that leads from global to local processes is called into question by the data reported in Noveck & Sperber's paper, which show that implicatures come at a cost for adults and children alike. Noveck & Sperber thus brilliantly succeed in demonstrating the usefulness of experimental approaches for adjudicating between alternative accounts.
Let us now consider the more general questions relating to the content of concepts and the explicature/implicature distinction. Here, I agree with Groefsema's analysis, according to which the main issue concerns the content of concepts, with the content of the concept to be understood as what it contributes to the explicature of an utterance. As Groefsema points out, the Relevance-Theoretic view of concepts is not easily compatible with the atomic view of concepts, which pragmatic theory borrowed from Fodor (see Fodor Reference Fodor1998 for the latest version). However, I strongly disagree with her suggestion that this tension should be resolved by abandoning concept atomicity and putting both encyclopaedic and logical entries in the content of the concept. There are quite good reasons to preserve concept atomicity, as shown by Fodor's and Millikan's work (see Fodor Reference Fodor1998, Millikan Reference Millikan2000). Moreover, understanding the content of a concept to be both the encyclopaedic and the logical entries blurs the distinction between these two types of information. Despite the fact that neither the encyclopaedic nor the logical information associated with a concept can be thought of as constitutive of the concept or as being its content on an atomic account of concepts, we should not be blind to the (further) fact that these are quite different sorts of information: if you have both the concept of CAT and the concept of ANIMAL, you have ipso facto the (logical) information that cats are animals. This is entailed by the conceptual hierarchy in which both concepts occur and which represents the inclusion of the extension of CAT in the extension of ANIMAL. In other words, it is a priori information that is directly accessible from the conceptual hierarchy. But you may have the concept of CAT (and even the conceptual hierarchy) without knowing that cats have retractile claws (encyclopaedic information), because such information is empirical information that cannot be derived from (atomic) concepts as such. On such a view, the atomic concept should be considered as entering the explicature (though the logical entry can be used to derive further explicatures if necessary), while the encyclopaedic entry should be used to derive implicatures. This leads us directly to the notion of an ad hoc concept (which has shifted some items from the set of implicatures to the set of explicatures). Here, I once again agree with Groefsema that there exists a contradiction in the Relevance-Theoretic description given of the derivation of an ad hoc concept and the claim that ad hoc concepts are atomic. Hence, it is not clear that the notion of an atomic ad hoc concept is not self-contradictory, and thus it is not clear, despite Wilson & Carston's claim, that what an ad hoc concept contributes to the explicature is an atomic concept. This is problematic, given that ad hoc concepts should provide the answer to the concern, expressed in the majority of this book's chapters, as to what is the derivation of the logical form of an utterance, as well as help clarify the distinction between explicatures and implicatures.
The notion of an ad hoc concept has brought into the open a long-standing tension in Relevance Theory, viz. that between the adoption of an (atomistic and externalist) view of concepts (such as Fodor's) and the description that Relevance Theory in fact gives to concepts. This is an issue that has to be addressed if Relevance Theory is to continue to be of central importance in contemporary pragmatics – a role which it clearly has at present, as illustrated by the fact that most papers in the book adopt this particular framework.
