What Is Meaning?

Yeh.

Chapter 1

Meanings

IN WHAT FOLLOWS, I will take it for granted that words, phrases, and sentences have meaning, that for each meaningful expression there are correct answers to the question What does it mean?, and that two expressions mean the same thing when the answer to this question is the same for both. Theories of meaning envisioned by philosophers attempt to answer questions of this sort in a systematic way. The targets of such theories include formal languages of logic and mathematics, extensions of these languages incorporating philosophically important intensional and hyperintensional notions, fragments of natural language, and full-fledged natural languages. Although much progress has been made with these theories, much remains to be done.

This is as it should be. It’s not the job of philosophers to inventory all of the semantically significant structure of English, or to note how it differs from that of Urdu. It is their job to articulate theoretical frameworks in which such investigations can profitably take place. Happily, these foundational efforts have not been wasted. Much of the progress so far achieved has occurred in frameworks descended from Frege, Russell, Tarski, Carnap, Kripke, Montague, Lewis, Stalnaker, Davidson, and Kaplan. However, the picture is not entirely rosy. There is, I shall argue, an unsolved problem at the heart of our conception of what meaning is, and what we want from a theory of meaning. My task will be to state the problem, indicate why it has been difficult to solve, and sketch the beginning of what I hope will prove to be a solution.

The problem involves the relationship between sentence meaning and the entities, called propositions, with which such meanings have traditionally been identified. This notion is implicitly defined by the following assumptions.

A1.  Some things are asserted, believed, and known. These attitudes relate agents—those who assert, believe, or know something—to that which they assert, believe, or know.

A2.  The things asserted, believed, and known are bearers of (contingent or necessary) truth and falsity.

A3.  Propositions—the things satisfying A1 and A2—are expressed by sentences. The proposition expressed by S can standardly be designated by expressions such as ⌈the proposition that S⌉, ⌈the statement/claim/assertion/belief that S⌉, or simply ⌈that S⌉.

A4.  Since different sentences may be used to assert the same thing, or express the same belief, and since different beliefs or assertions may result from accepting, or assertively uttering, the same sentence, propositions are not identical with sentences that express them. Intuitively, they are what different sentences, or utterances, that say, or are used to say, the same thing have in common.

A1-A3 are little more than platitudes. By contrast, the abstraction implicit in A4 may seem worrying. What, after all, do different utterances that say the same thing have in common? Although the answer to this question is not obvious, the abstraction involved in attempting to answer it is not fundamentally different from other, presumably innocent, abstractions. What are the infinitely many meaningful sentences of a language, if not abstractions from the finitely many utterances of its speakers? Since there is no science of language without such abstraction, abstraction itself isn’t a serious worry about propositions. As we shall see, however, there are deep and serious worries about propositions that go to the heart of the question of whether they can play the roles required of them in theories of language and mind, or whether, on the contrary, they should be dispensed with altogether.

In order to appreciate the force of this dilemma, one needs some sense of the roles that propositions do play in our theories. Within semantics, propositions are needed (i) as referents of certain names—‘Logicism’, ‘Church’s Thesis’, ‘Goldbach’s Conjecture’; (ii) as referents of demonstratives in utterances of sentences like ‘That’s true’; (iii) as entities quantified over, as in sentences like ‘At least six of the theses advanced by Professor Wyman are unsupported by evidence’; and (iv) as objects of the attitudes, as indicated by the sentence ‘Mary defended several of the claims that Bill denied and Susan questioned’. Propositions are also needed to state the goals of semantic theory, and to relate semantics to the interpretation of speakers. Even if a language lacks propositional attitude constructions, we still need to know what speakers assert and believe, when they sincerely utter, or assent to, a sentence. Since propositions are that which is asserted and believed, and since semantics is charged with specifying the contribution of the meaning, or semantic content, of a sentence to what is asserted by utterances of it, propositions are presupposed by our best account of what we want semantic theories to do.

Recent advances in our understanding of the relationship between semantics and pragmatics have added to the weight placed on propositions. For many years, the governing assumption about the relationship between meaning and assertion was that what a speaker asserts when uttering a sentence is typically nothing more than the semantic content of the sentence uttered, plus, perhaps, a few trivially obvious consequences of this (e.g., the conjuncts of a conjunction). In the last two decades that picture has been complicated by the recognition (i) that the semantic contents of some sentences are incomplete, and require the contribution of pragmatic information available in the context of utterance in order to generate truth-evaluable candidates for assertion, and (ii) that even when the semantic content of a sentence is complete, that content can often be further enriched by contextually salient information. As a result of this semantic-pragmatic interaction, a single utterance of a sentence often succeeds in making multiple assertions, some of which go well beyond the semantic content of the sentence uttered; and in some cases, the semantic content of the sentence uttered is not even among the propositions asserted.¹

These observations pose a serious challenge to semantic theories that try to do without propositions by assigning each utterance, or sentence-in-context, a single set of truth conditions that determines a single truth value.² Since the truth value of an utterance can only be the truth value of the conjunction of all the assertions it makes—which, in turn, have been shown to contain a great deal of nonsemantic information provided by features of the context—any strictly semantic theory that assigns truth conditions directly to utterances faces the dilemma of either incorrectly assigning such utterances impoverished truth conditions (that leave out the pragmatic contributions of context), or incorrectly building nonsemantic information into meaning, often by mischaracterizing pragmatic contributions to assertion as semantic ambiguities (as in attempts to semanticize Donnellan’s referential/attributive distinction). By contrast, the newfound recognition of the complex relationship between semantic and assertive content is made to order for a theory that takes the semantic content of a sentence to be a structured complex that provides a skeleton for the proposition, or propositions, asserted—which are arrived at by pragmatically fleshing out the bare bones provided by the semantics.³ Without structured propositions, as contents of sentences, objects of attitudes, and bearers of truth value, it is hard to see how any suitably complex theory of the relationship between semantics and pragmatics could get off the ground.

Propositions are also ubiquitous in other areas of philosophy. For example, epistemological questions about what can be known, and about whether there are necessary truths that are knowable only aposteriori and contingent truths that are knowable apriori, presuppose that certain things that are known are also bearers of contingent, or necessary, truth. These are propositions. But if there is no denying the existence of propositions, there is also no denying the puzzles to which they give rise, and the problematic answers to those puzzles provided by traditional conceptions of them. Appeals to propositions in theories of language and mind incur the debt of explaining what it is for us to entertain, or otherwise bear cognitive attitudes to, them—and to identify that to which propositions owe their ability to represent the world, and so bear truth conditions. Traditional answers to the first of these questions are variations on the familiar story of propositions as denizens of a third realm (beyond mind and matter), which are grasped by a mysterious intellectual faculty of platonic extrasensory perception. Answers to the second question typically assume that propositions are both intrinsically representational, and that from which all other representational bearers of truth conditions—sentences, utterances, and mental states—inherit their representationality.

In what follows, I will explain why I have come to think that propositions in this traditional sense do not exist. If I am right, we face a dilemma. Either we must learn to conceptualize our philosophical, linguistic, and cognitive problems and theories without appealing to propositions, or we must conceive of them in a fundamentally different way. I will argue for the second horn of this dilemma by offering a new account of propositions that reverses traditional explanatory priorities. Propositions, as I understand them, can play the roles for which they are needed in semantics, pragmatics, and other areas of philosophy. However, they are not the source of that which is representational in mind and language. Sentences, utterances, and mental states are not representational because of the relations they bear to inherently representational propositions. Rather, propositions are representational because of the relations they bear to inherently representational mental states and cognitive acts of agents.

The view I will outline locates meaning in thought, perception, and the cognitive acts of agents. Although language is the locus of our ordinary and theoretical talk about meaning, meaning itself, or more properly, propositions—which are the meanings of sentences—are understood in terms of the explanatorily prior notion of agents predicating properties—of objects, other properties, and even of propositions—in all forms of cognition. For this reason, I reject the pervasive language of thought fantasy, which wrongly takes linguistic meaning as the fundamental model—to be extended to theories of cognition—for understanding all intentionality. By contrast, I start with perception. One who sees an object x as red and tastes it as sweet thereby predicates redness and sweetness of it, just as one who feels an object y as vibrating and hears it as humming predicates those properties of it. As a result, the perceptual experience of