SUMMARY This volume is a collection of papers previously published by Sperber and Wilson. This book contains a sample of the current status of relevance theory, as it has developed since originally present in the authors' 1986 book ''Relevance: Communication and Cognition''. The preface presents the general hypothesis that linguistic meanings provide evidence of a speaker's intended meaning but do not actually encode this meaning.
The introduction defines 'pragmatics' as ''the study of how contextual factors interact with linguistic meaning in the interpretation of utterances'' (p.1). The authors begin with Grice's work on implicature, bridging linguistic meaning and broader communicated meaning. Then they present three approaches to pragmatics which follow from Grice's work: the 'literalist' approach, the 'contextualist' approach, and the 'cognitive-relevance approach'. The 'literalist' approach describes a process of recovering explicit sentence meaning through assignment of referents. The 'contextualist' approach describes recovery of speaker meaning through a significant amount of pragmatic inference. The 'cognitive-relevance' approach describes a process which results in optimally relevant interpretation of utterances. The authors also relate these principles to the question of semantic interpretation. The 'literalist' approach sees explicit sentences as expressing (near) propositions. The 'contextualist' approach sees these as expressing only partial propositions, such that even complete truth-conditional meaning is achieved through pragmatic inference.
Relevance theory introduces the idea of 'explicature', the speaker's explicit meaning developed from the logical form represented by a sentence. 'Implicature' on the other hand refers a message implicitly communicated by the speaker by uttering the utterance. Degrees of explicitness of a message relate to the relative degrees to which decoding and pragmatic inference contribute to utterance interpretation. The final interpretation of an utterance is a choice from several possibilities which arise from the details of implicatures that arise.
The introduction continues with a discussion of lexical pragmatics. A comparison of Gricean and relevance theory approaches to metaphors, trope and vagueness is presented. Relevance theory suggests a process of adjustment of explicatures and implicatures of an utterance as an explanation.
Finally, the authors suggest that combining a contextual semantics and a relevance-theoretic pragmatics allow for explanation of speech acts, presuppositions and indexicals. Interpreting these parts of utterances requires selecting 'higher order' explicatures from those arising as directed by 'indicators' in the utterance. The introduction concludes with a reiteration of the cognitive nature of the relevance theory approach to sentence interpretation and the advantages of such an approach.
The book is divided into three parts. The first, 'Relevance and Meaning', deals with the human ability to communicate beyond the bounds of an agreed linguistic code. It includes 6 chapters. Each discusses the relevance theory approach to a specific language phenomenon.
In Chapter 2, Dan Sperber and Deidre Wilson discuss 'The mapping between the mental and the public lexicon' the mapping between 'public words' and ' mental concepts'. There is no one-to-one mapping between words and concepts; the most appropriate of the possible mappings is contextually chosen. The 'literal' mapping is just one of these possibilities, and not necessarily favored. A concept is defined as 'an enduring elementary structure, which is capable of playing different discriminatory or inferential roles on different occasions in an individual's mental life' (p.35). An 'inferential theory' of communication is presented. Relevance is defined in terms of a relationship between inputs and cognitive processes. From this they derive a 'cognitive principle' and a 'communicative principle'. The result is a comprehension procedure of this type: construct a conceptual structure from the linguistic code, explicitly enrich this structure and implicitly complement it until an 'expectedly' relevant interpretation is reached. The variability of the mapping from linguistic code to interpretation, both within and between speakers, does not impede communication since an encoded message merely provides an indication of the speaker's communicative intent, and not a complete explicit message.
In Chapter 3, 'Truthfulness and relevance', Wilson and Sperber argue that what is 'communicated' is assumed to be true, and not what is 'said'. Expectations that utterances are true result from the expectation that these utterances are relevant. Different explanations of metaphor, irony, loose and figurative speech follow. The emphasis is on the truth of the conclusions drawn by the hearer based on the utterances which may have practical implications for the hearer's behavior.
Utterances are defined as relevant only in the case that they result in the cognitive benefits of information (knowledge) gain and/or revision of currently held assumptions. Utterances are also evaluated in terms of the cost or effort required to achieve the interpretation. Human communication aims for 'optimal relevance', based on relevance relative to hearer effort and relative to the speaker's characteristics. The chapter includes an illustration of how this works. Linguistic decoding is seen as an automatic, almost trivial part of the process. The chapter concludes by redefining the concepts 'explicit', 'literal' and 'what is said' within the framework of relevance theory.
In Chapter 4, 'Rhetoric and relevance' by Sperber and Wilson, the dilemma of rhetoric is described. The proposed solution to the dilemma is to retain the notion of literal meaning without the presupposition of literal use. The application of relevance theory to metaphor and 'loose' uses of language is described. In interpreting these utterances, no literal interpretation is examined and rejected; the hearer interprets directly to the 'non-literal' interpretation relevant to the discourse. Metaphors and 'figurative' speech are not merely seen as decorations but as triggering cognitive effects, by basically the same mechanism as is used in interpreting non-figurative speech.
In terms of speech acts, there is not a direct correspondence between grammatical mood and illocutionary act. In echoing an utterance the speaker is communicating his attitude to this utterance and thus the echo is relevant. Irony is seen as echoing a thought while expressing an ironic attitude.
Chapter 5, 'A deflationary account of metaphors', is also by Sperber and Wilson. Relevance theory views metaphor as important, but not distinctive from other, more 'literal' expressions. The chapter reviews the role of language in communication. As opposed to the traditional orientation, relevance theory does not view human communication as the process of decoding and sees language as a defective code.
When interpreting an utterance, the concepts available include broadening and narrowing, as well as the continuum between them. Similarly there is a continuum between 'literal' to variously inferred interpretations so that the interpretation of metaphors does not require a special mechanism. Poetic effects are achieved by promoting the inference of a large number of weak implicatures.
In Chapter 6, 'Explaining irony', Wilson and Sperber describe the point of irony as the speaker showing that the statement uttered is grossly inadequate. In this chapter for the first time a reference is made to the developmental and clinical applications of pragmatic theory. They point out three puzzles connected with irony. First, relevance theory suggests that irony involves the expression of an attitude rather than an assertion literal or otherwise. Irony relates to attitude while other types of figurative expressions are thoughts about to states of affairs. Second, irony appears to have a normative bias where is can be used to deride but not to praise. Thirdly, ironic expressions have a specific intonation pattern.
Relevance theory provides an echoic account of irony in which irony is seen as an example of attributive use of language in which the speaker does not make an assertion but communicates an attitude about an assertion. Ironic utterances are distinctive in that the attitudes they convey are dissociative (the thought is considered to be false or grossly inadequate). Metaphors are distinguished from irony in that metaphors express a thought about a state of affairs while irony expresses an attitude about a thought.
The second part of the book, 'Explicit and Implicit Communication', discusses arguments in support of the major contribution of pragmatic processes to explicit communication, beginning with Chapter 7, 'Linguistic form and relevance' by Wilson and Sperber. Utterance interpretation involves two phases. A linguistically encoded logical form is enriched first by inference. This enriched 'propositional' form, which may be seen as an explicature, is then the input to an inference mechanism. Relevance theorists argue that among Gricean maxims, relevance is sufficient to motivate this mechanism and the remaining three maxims are therefore unnecessary.
Pragmatics deals with the distinction between decoding and inference while semantics then deals with questions about the decoding phase. A distinction is made between informative and communicative intentions. One may be covert while the other is overt. Linguistic constructions not only include decoded information (conceptual information) but also the procedures for dealing with this information. This 'conceptual representation' acts as we might have assumed a logical form would act in terms of entailment and contradiction relations and truth-conditions.
Four types of information are described: conceptual, truth-conditional information coded by 'content' words, conceptual, non-truth-conditional information inferred by adverbials as higher level explicatures, procedural, non-truth-conditional information inferred from discourse connectives which constrain implicature inferences and procedural and truth-conditional information encoded by personal pronouns which restrain the explicature inferences arriving from an utterance in which they participate.
In Chapter 8, 'Pragmatics and time', Wilson and Sperber contrast the Gricean interpretation where the temporal (including interval), causal and sequential information provided by conjunction is purely inferential (implicatures) on one hand and how relevance theory views this information as the pragmatically inferred parts of the truth-conditional content of the utterance. The authors argue that this latter approach allows this phenomenon to be analyzed within the same general explanation based on inference as other phenomena discussed.
Chapter 9 is 'Recent approaches to bridging: truth, coherence, relevance' by Wilson and Tomoko Matsui. It discusses inferences based on 'bridging assumptions', or assumptions made by the hearer to allow him to find the reference of a noun phrase. The authors argue for a relevance-based explanation which incorporates cognitive effort and communicative effect rather than 'coherence' and truth-based approaches, or combinations of the two. In this chapter too, experimental results are brought to illustrate and support the arguments.
Chapter 10, 'Mood and the analysis of non-declarative sentences' by Wilson and Sperber, begins with a discussion of 'mood' as the logical properties of declarative versus other types of sentences. 'Illocutionary force' is then taken to be a pragmatic property of utterances. Here too, the relationship between mood and force is seen as inferential. Mood seems to indicate a group of utterance uses. Semantically, imperative sentences are considered to describe possible worlds which the speaker sees as 'potential' and 'desirable'. Depending on the use of the utterance, desirability may be defined from the point of view of the speaker or the hearer. The authors describe 'descriptive' representation as 'a relation between thoughts or utterances and possible or actual states of affairs which make them true' and 'interpretive' representation as 'a relation between thoughts or utterances and other thoughts and utterances that they resemble in content.' (p.218). Thus, an explanation of imperative utterances relies on an interpretive representation. A similar line is presented to explain the interpretation of interrogative utterances. The authors claim that the linguistic form of imperatives, interrogatives and other non-declarative sentences is restricted to indicating the relevance of the utterance and plays a minimal role in utterance interpretation.
Chapter 11, 'Metapresentation in linguistic communication' by Deidre Wilson, deals with an integration of a relevance theory account of understanding speaker meanings and a more general cognitive account of 'theory of mind'. The part of language under discussion is the comprehension of reported speech, direct or indirect, and this chapter again includes a summary of relevance theory. Developmental stages in inference are described. Progress from stage to stage parallels general stages of development of 'theory of mind'. In this approach interpretation is achieved by 'resemblance', either metalinguistic or interpretive. The chapter also shows how this approach deals with attributive utterances.
The third part of the book, ''Cross-Disciplinary Themes', explores broader implications of relevance theory. Chapter 12, by Sperber and Wilson, 'Pragmatics, modularity and mindreading', begins with a summary of the nature and purpose of pragmatic study. Pragmatics is seen as a metapsychological process, specialized for communication. Support for comprehension being a general metapsychological process is found partly in the correlation of various degrees of difficulty (first, second, third order, etc.) with other cognitive abilities, particularly of a 'theory of mind'. An argument is made for the modular nature of pragmatics. The rationalized inferential steps suggested by Grice can then be a 'test' for the pragmatic nature of the inference (i.e. conversational implicature), but may not be a model of how the inference is actually made. The currently proposed model suggests a sub-module for comprehension within a more general theory of mind module. At least some of the specific functions of the mind-reading module may be innate.
The authors take an evolutionary approach to human cognition. A gradual evolutionary process steers the cognitive system to relevance while a discrete evolutionary process results in the 'relevance -- based comprehension module'. The search for relevance guides attention, memory retrieval and inference.
Chapter 13, by Jean-Baptiste Van der Henst and Sperber, 'Testing the cognitive and communicative principles of relevance', surveys experimental data supporting the cognitive and communicative principles of relevance. It recounts the basis of relevance theory using a previously appearing example. The experimentation requires manipulating the effort required to achieve an interpretation while holding the effect constant, or by manipulating the effect while holding the effort constant. This results in different degrees of satisfaction of a hearer's expectations of relevance. Determinate and indeterminate problems are presented. Subjects are asked what follows from the premises. Answers of 'nothing follows' from the indeterminate problems show that irrelevant or trivial conclusions are not considered, thus supporting the notion of relevance in inference. The effort required to reach a conclusion affects its perceived relevance. The conclusions from these reasoning experiments lead to a theory that relevance generally influences performance on cognitive tasks and results may reflect task relevance rather than an individual's ability. Manipulation of effort and effect independently showed that both affect reasoning. Further experiments specifically address the role of the communicative principle of relevance in speech production. Results show that speakers formulate their utterances such as to achieve maximum relevance.
Chapter 14, 'The why and how of experimental pragmatics: the case of 'scalar inferences' by Ira Noveck and Sperber, further discusses experimental support for relevance theory, specifically the case of 'scalar inferences'. The problem of basing pragmatic theory on pragmatic intuitions of researchers is brought. Experiments clarify these intuitions and further help sort out differences between different theoretical accounts. The chapter reviews generalized conversational implicatures (GCIs). Four claims of GCIs are brought and then refuted. The first, that GCIs are default, is refuted since cancellation of GCIs in a significant percentage of cases would require too much effort. The context independent argument is refuted as a scalar term such as 'some' would vary in its reference (the quantity referred to) depending on the context of the utterance. Relevance theory argues that these inferences are not scalar -- not really indicative of a pre-existing scale since the 'literal' meaning can be extended or restricted, and are not implicatures but in most cases explicatures.
GCI and relevance theory accounts of scalar term interpretation predict different experimental results. Developmental experiments are described which show that children are overall less likely to interpret scalar terms with pragmatic enrichment than adults. Experiments showing that enriched interpretations take longer than literal interpretations are counterevidence for default scalar inferences and support a relevance theory approach. Specifically, these results do not support a default -- and thus speedier -- nature of the scalar inferences. The main aim of their chapter is to encourage students of pragmatics to consider experimental work in their theoretical investigations.
The final chapter, 'A pragmatic perspective on the evolution of language' by Sperber and Gloria Origg, begins with an overview of the differences between a 'code' model of communication and an 'inferential' model. The discussion distinguishes animal communication, which involves codes, from human communication in which the 'code' plays only a minor role. In inferential communication, a mutual goal or conclusion is required, not a mutual code. From an evolutionary point of view, changes and developments of the existing grammar of the code are felicitous in inferential communication but detrimental in non-inferential, 'code' communication.
EVALUATION The chapters originally appeared as independent papers. Still, the chapters cohere since the editors authored or co-authored all of the chapters. There is repetition of basic definitions and concepts of 'relevance theory' in most chapters. Often, there are slight differences in emphasis and definition of central concepts (e.g. compare the two main principles of relevance as described in Chapters 2, 4 and 5). In each case the explanation assumes no prior knowledge on the part of the reader. The result is on one hand highly repetitious, while on the other, guarantees that by the end of the book the reader is well familiar with these principles and mechanisms. Throughout, examples clarify the theory and to support its explanatory value. Several of the examples recur in different papers. Other topics recur in different chapters.
The book is cleanly produced, though I found one small typo on page 318 where 'some' should be replaced by 'or'.
The authors have met their proposed goal of providing a survey of relevance theory in recent years. Interesting issues include a solution to the interval problem presented in Chapter 8, which could have interesting applications to the 'allover' implicature affecting colors and other adjectives. I am left with a question about the place and process of decoding. The actual words are seen by relevance theory as a trigger, however I did not gain a clear picture of how the nature and the details of the trigger are seen within the theory.
Overall, this book is well organized and gives a comprehensive overview of relevance theory.
ABOUT THE REVIEWER:
Leah R. Paltiel-Gedalyovich is a speech language clinician with a special interest in clinical applications of theoretical linguistics. Her doctoral thesis, and research since, investigated the semantic-pragmatic interface in first (Hebrew) language acquisition, specifically scalar implicatures. Her current research projects involve the development of a comprehensive battery for developmental Hebrew language acquisition.