Introduction: Going Down the Anti-Formalist Road

I

It is one of the theses of this book that many of the issues in interpretive theory can be reduced to a few basic questions in the philosophy of language. Consider, for example, the discussion of "presupposition" in Ruth Kempson's Presupposition and the Delimitation of Semantics (Cambridge, 1975). Kempson begins by observing that presupposition can be defined "in one of two ways—either as a relation between statements (parallel to entailment, synonymy, etc.) or as a property of the speaker's belief in uttering a sentence" (p. 2). The difference is between a formal notion of presupposition in which it is a feature of sentences as they exist in the abstract apart from any particular occasion of use, and presupposition as a fact about what is in a speaker's mind—his understanding of the world and of the situation in which he now finds himself—at the moment of utterance. As Kempson observes, the possibility of deriving meaning from the formal properties of sentences cannot survive the serious assertion of a speaker-based theory of presupposition, for "if presuppositions in terms of speaker-belief are considered to be a part of the semantic interpretation of sentences, then it seems that the meaning of sentences must be in terms of speaker-hearer relations and not ... in terms of the relation between a symbol or set of symbols and the object or state described" (pp. 2–3). And if that is so, "one must give up the standard claim that the meaning of a sentence is a function of the meaning of its constituent parts" (p. 60). This follows because if the presuppositions of an utterance vary with the beliefs of speakers and hearers—that is, if no sentence has "a unique set of presuppositions"—then "every sentence can be analyzed with ... as many different sets of presuppositions" as there are different possible contextualizations and there is no regular and predictable way to assign a meaning, or even a range of possible meanings, to a particular sentence. Meaning, in short, becomes entirely contextual and "cannot be determined independent of the speaker of a sentence in a particular situation" (p. 60). Once this conclusion has been reached, others immediately follow:

One is thus faced with an analysis of meaning which claims that every sentence has an indeterminate number of indeterminable meaning representations. And if the meanings of sentences are indeterminable, then meaning relations between sentences such as entailment, contradiction, by definition cannot be predicted. Moreover ... it would follow that the grammaticality of sentences cannot be determined either, independent of the situation in which they are uttered. But this has the immediate consequence that one's grammar would not be predictive.

By producing this sequence of entailments, Kempson (however inadvertently and even reluctantly) makes a very important point: once you start down the anti-formalist road, there is no place to stop; remove the connection between observable features and the specification of meaning, and you also remove everything else that is supposedly independent of context; entailment, contradiction, grammaticality itself, all become as variable and contingent as presupposition. This, however, is not the conclusion Kempson reaches; and indeed it is the specter of reaching it that drives her in the opposite direction:

We are thus faced with the conclusion that a theory which incorporates a speaker relative concept of presupposition as part of its semantic representation is in principle unable to fulfill any of the four conditions I set up initially (1.1) as a prerequisite for any semantic theory and therefore must be relinquished, (p. 60)

Those conditions include "a systematic relation between the meaning of lexical items and the syntactic structure of the sentence," a "finite set of predictive rules," the mechanical separation of non-deviant from anomalous sentences, and predictability of meaning relations between sentences. Kempson is quite right to observe that a speaker-relative concept of presupposition and therefore of meaning makes the fulfilling of those conditions impossible since the variability of speakers and the difficulty of determining what is in their minds precludes generality and makes every speech situation unique. The surprise is in the last phrase: "and therefore must be relinquished." Her reasoning is that since speaker-relative presupposition, if taken seriously, would create grave (indeed, insurmountable) difficulties for a semantic theory, we cannot take it seriously. The conclusion is not reached because the evidence for speaker-relative presupposition is weak or because the connection between speaker-relative presupposition and meaning is specious, but because to pursue this line of thought would be to give up the pursuit of theory. In short, for Kempson this line of thought is unthinkable in the sense that her deepest assumptions mark it as obviously absurd. Surely any thesis that is incompatible with the assumed goal of linguistics— the goal of building a theoretical model of language—must be rejected out of hand; after all, the "four demands" that speaker-relative presupposition fails to satisfy "are agreed in principle by all linguists" (p. 2). If Kempson were to give credence to this thesis, she would be denying her membership in the community of linguists and (in effect) denying the most significant aspect of her being.

I would not be misunderstood. I am not criticizing Kempson for rejecting arguments that are stigmatized in advance by the beliefs she necessarily holds as a practicing linguist. She has no choice, if by choice one means a judgment reached independently of any predispositions or biases. It is just those predispositions and biases—those assumptions concerning what must be the case in the matter of language—that fill her judgment, and one would be making if not an impossible at least a Herculean demand if one were to ask her to set them aside.

Again, I would not be misunderstood. I am not saying that Kempson is beyond criticism simply because the context of which she is an extension prevents her from seeing certain arguments as respectable or even makable. In my very strong opinion the arguments she clings to, the arguments that underwrite the project of formal linguistics, are wrong. And it is part of my argument that I can say that despite the sympathetic analysis I make of her "epistemological condition." This does not mean that I am not in the same condition—embedded in conviction—but that precisely because I am embedded in conviction, my sense of the rightness of my arguments is no less strong than hers and is in no way diminished by my ability to give an account of its source. That at least is the burden of several of the essays in this volume, especially of those that assert the inconsequentiality (in certain terms) of theory.

This, however, is to get ahead of my story, and for the time being I would like to linger a little longer on the issues Kempson raises, for much of what I want to say builds on the thesis from which she draws back in horror, the thesis that the meaning of a sentence is not a function of the meaning of its constituent parts; or to put it another way, that meaning cannot be formally...