Robyn Anne Carston

The general consensus emerging from decades of empirical investigation of metaphor processing is that, when appropriately contextualised, metaphorically used language is no more demanding of processing effort than literally used language. However, there is a small number of studies which contradict this position, notably Noveck, Bianco, and Castry (2001): they maintain that relevance-based pragmatic theory predicts increased cognitive costs incurred in deriving the extra effects that metaphors typically yield, and they provide experimental results that support this prediction. In our study, we first surveyed and assessed the tasks and stimulus materials of many experiments on metaphor processing from the 1970's to the present day. The most telling result was an apparent disparity between the processing of metaphorical language used predicatively versus referentially. We then ran two self-paced reading experiments to test our hypothesis that when used as a predicate, metaphorical language is no more costly than literal language, but when used referentially, it does incur extra costs, even given a preceding biasing context. In the first experiment, all metaphorical referring expressions were in subject position so occurred early in the sentence; in the second experiment, we controlled for any effect of sentence position by placing metaphorical referring expressions in object position, thus later in the sentence, similar to the predicate metaphors. In both cases, metaphorical referring incurred significantly greater costs relative to literal equivalents than did metaphorical predication, with no effect of sentence position. We end with a brief analysis of why the referential use of metaphor is special and effort-demanding.

Polysemy, understood as instances of a single linguistic expression having multiple related senses, is not a homogenous phenomenon. There are regular (apparently, rule‐based) cases and irregular (resemblance‐based) cases, which have different processing profiles. Although a primary source of polysemy is pragmatic inference, at least some cases become conventionalised and linguistically encoded. Three main issues are discussed: (a) the key differences between regular and irregular cases and the role, if any, of a “core meaning”; (b) the distinction between pragmatic polysemy and semantic polysemy; and (c) the role of syntactic meaning in both generating and constraining polysemy.

Bringing together work by leading scholars in relevance theory, this volume showcases cutting-edge research within the theory, and demonstrates its influence across a range of fields including linguistics, pragmatics, philosophy of language, literary studies, developmental psychology and cognitive science. Organised into broad thematic strands that represent the latest research and debates, the volume shows the depth of analysis now possible after nearly forty years of intensive work in developing and applying the principles of relevance theory. The breadth of influence of the framework is reflected in the chapters of the volume, in some cases moving beyond the traditional realms of semantics and pragmatics to include discourse analysis, language acquisition, media and education. The volume will be essential reading for researchers in these fields, as well as for those already working within relevance theory or with other pragmatic theories.

I propose that an account of metaphor understanding which covers the full range of cases has to allow for two routes or modes of processing. One is a process of rapid, local, on-line concept construction that applies quite generally to the recovery of word meaning in utterance comprehension. The other requires a greater focus on the literal meaning of sentences or texts, which is metarepresented as a whole and subjected to more global, reflective pragmatic inference. The questions whether metaphors convey a propositional content and what role imagistic representation plays receive somewhat different answers depending on the processing route.

“Utterances and thoughts have content: They represent (actual or imaginary) states of affairs.” This is the opening statement of François Recanati’s most sustained work on kinds of representation, Oratio Obliqua, Oratio Recta (2000) and it presents the core phenomenon which it is the task of the philosophy of language to explain. A primary function of language and thought, though not their only function, is to represent how things are or might be. As well as descriptively representing entities, properties, and states of affairs in the world, our thoughts and utterances may metarepresent other linguistic or mental representations and to several levels of representational embedding. For a full understanding of how language represents (and metarepresents) we need to understand how each of the different kinds of expression in the language contributes to the proposition or thought expressed by a particular utterance. Over the last century, the philosophy of language has swung between two poles in addressing the representational properties of natural human language. Broadly speaking, at the one pole are those whose main focus is on the language system itself, rather than its users, and who emphasize the semantic power of the system, while at the other pole are those for whom it is speakers, specifically their mental states, and what they do with words that is fundamental. The two different orientations naturally lead to different ways of conceiving of the roles of semantics and pragmatics in language use. One manifestation of the division is the opposition between the ‘literalist’ view according to which the proposition expressed by an utterance of a sentence is entirely mandated by components of the sentence and the ‘contextualist’ view according to which what is said often goes beyond what is linguistically anchored. From his earliest work on pragmatics, specifically on speech acts and explicit performatives (1979, 1981, 1987), Recanati has taken a non-literalist position, which he has developed over the past 20 odd years, culminating in the publication of his latest book, Literal Meaning (2004a)..

Within relevance theory the two local pragmatic processes of enrichment and loosening of linguistically encoded conceptual material have been given quite distinct treatments. Enrichments of various sorts, including those which involve a logical strengthening of a lexical concept, contribute to the proposition expressed by the utterance, hence to its truth-conditions. Loosenings, including metaphorical uses, do not enter into the proposition expressed by the utterance or affect its truth-conditions; they stand in a relation of 'interpretive resemblance' with the linguistically encoded concept used to represent them. This asymmetric treatment is questioned here, arguments are given for an account which reflects the complementarity of these processes and several alternative symmetrical treatments are explored.

In standard Relevance Theory, hyperbole and metaphor are categorized together as loose uses of language, on a continuum with approximations, category extensions and other cases of loosening/broadening of meaning. Specifically, it is claimed that there are no interesting differences between hyperbolic and metaphorical uses. In recent work, we have set out to provide a more fine-grained articulation of the similarities and differences between hyperbolic and metaphorical uses and their relation to literal uses. We have defended the view that hyperbolic use involves a shift of magnitude along a dimension which is intrinsic to the encoded meaning of the hyperbole vehicle, while metaphor involves a multi-dimensional qualitative shift away from the encoded meaning of the metaphor vehicle. In this article, we present three experiments designed to test the predictions of this analysis, using a variety of tasks (paraph.

A cognitive pragmatic approach is taken to some long-standing problem cases of negation, the so-called presupposition denial cases. It is argued that a full account of the processes and levels of representation involved in their interpretation typically requires the sequential pragmatic derivation of two different propositions expressed. The first is one in which the presupposition is preserved and, following the rejection of this, the second involves the echoic (metalinguistic) use of material falling in the scope of the negation. The semantic base for these processes is the standard anti-presuppositionalist wide-scope negation. A different view, developed by Burton-Roberts (1989a, 1989b), takes presupposition to be a semantic relation encoded in natural language and so argues for a negation operator that does not cancel presuppositions. This view is shown to be flawed, in that it makes the false prediction that presupposition denial cases are semantic contradictions and it is based on too narrow a view of the role of pragmatic inferencing.

The interpretation of metaphorical utterances often results in the attribution of emergent properties, which are neither standardly associated with the individual constituents in isolation nor derivable by standard rules of semantic composition. An adequate pragmatic account of metaphor interpretation must explain how these properties are derived. Using the framework of relevance theory, we propose a wholly inferential account, and argue that the derivation of emergent properties involves no special interpretive mechanisms not required for the interpretation of ordinary, literal utterances.

The generative grammar approach to language seeks a fully explicit account of the modular systems of knowledge (competence) that underlies the human language capacity. Similarly, the relevance-theoretic approach to pragmatics attempts an explicit characterisation of the sub-personal systems involved in utterance interpretation. As an on-line performance system, however, it is subject to certain additional constraints; this is demonstrated by the way in which matters of computational (processing effort) economy are currently employed in the two types of theory. A sub-module of “discourse competence” is shown to be compatible with and complementary to the wider system of pragmatic processes.

In (1), the talking event described in the matrix clause is elaborated on in the following adjunct: the arguing about the data and the theories makes up the content of the talking referred to in the matrix clause. In (2), on the other hand, the events (or sub-events of a single complex event) described are in a causeconsequence relation, a result of the action described in the matrix clause being that the porcelain vase breaks. These two examples illustrate a central issue for the interpretation of -ing adjuncts: the relation holding between the event described in the adjunct and the event described in the matrix clause is not always the same – in fact, as we’ll see, there is quite a range of possibilities. So the question is: how is the particular relation arrived at?

The idea is that, in a wide range of contexts, utterances of the sentences in (a) in each case will communicate the assumption in (b) in each case (or something closely akin to it, there being a certain amount of contextually governed variation in the speaker's propositional attitude and so the scope of the negation). These scalar inferences are taken to be one kind of (generalized) conversational implicature. As is the case with pragmatic inference quite generally, these inferences are defeasible (cancellable), which distinguishes them from entailments, and they are nondetachable, which distinguishes them from conventional implicatures. The core idea is that the choice of a weaker element from a scale of elements ordered in terms of semantic strength (that is, numbers of entailments) tends to implicate that, as far as the speaker knows, none of the stronger elements in the scale holds in this instance. The pattern is quite clear in (1) and (2), where the weak/strong alternatives are some/all and five/six respectively. In the case of (3), the stronger expression must be intelligent and good-hearted which entails intelligent; what Y's utterance implicates is that Mary does not have the two properties: intelligence and good-heartedness, so that, given the proposition expressed (Mary is intelligent) it follows, deductively, that she is not good-hearted, in Y's opinion. The example in (4) involves a scale inversion due to the negation, so that the weak/strong alternatives are not necessarily/not possibly; the negation which the scalar inference generates creates a double negation, which is eliminated giving possibly.

A standard view of the semantics of natural language sentences or utterances is that a sentence has a particular logical structure and is assigned truth-conditional content on the basis of that structure. Such a semantics is assumed to be able to capture the logical properties of sentences, including necessary truth, contradiction and valid inference; our knowledge of these properties is taken to be part of our semantic competence as native speakers of the language. The following examples pose a problem for this view of semantics.