According to epistemic brands of truth-conditional semantics, which dominate the philosophy of language, to know the meaning of a statement is to know its truth-conditions or the set of possible worlds in which it would be true.1 Against this view, henceforth "truth-conditionalism", I shall argue that in many cases:
(a) We know, for a given statement P, the meaning of P.
(b) We do not know the truth-conditions of P.
(c) Therefore knowledge of meaning ≠ knowledge of truth-conditions.
I begin by citing cases where we know the meaning of a statement yet do not know its truth-conditions, and afterwards I respond to some possible objections.
Cases. To begin with, consider the truth-conditions for ethnic slurs as in:
(1) Hitler was a kraut.
According to the Simple Conjunction Theory, (1) is equivalent to "Hitler was German, and Germans are despicable", which is false. According to the Indexical Conjunction Theory, (1) is equivalent to "Hitler was German, and I despise Germans"; its truth-value varies from speaker to speaker. According to the Bracket Theory, pejoration is a parenthetical side feature, and (1) is truth-conditionally equivalent to just "Hitler was German"; it is true. According to the Stereotype Theory, (1) is equivalent to "Hitler meets the negative stereotype of Germans", which is also true, though it differs from the Bracket Theory in treating "Einstein is a kraut" as false. According to Disquotational Semantics, (1) is true if and only if Hitler was a kraut, which is hardly informative; even given complete knowledge of the world, the truth-value of (1) remains unknown. According to various Non-proposition Theories, (1) either rests on a false presupposition or functions expressively; it is neither true nor false.2
The existence of such disagreement means that no one really knows what the truth-conditions of (1) are. Even if someone happened to have the right idea, that wouldn't count as knowledge. For example, just suppose that the Conjunction theory happens to get the right truth-conditions for (1). Then Conjunction theorists would seem to be lucky, not particularly justified. They could say "I knew it all along", and they could think that they had sound reasons to support the theory, but all the rival theorists can do the same. Of course Conjunction theorists could add that they were smarter than everyone else, which might even be true; but unless they had objective evidence for this, and also reason to think that extra intelligence is effective at getting at the right theory of vagueness, they would have insufficient justification because justification has a social dimension: one's awareness of controversy serves as a defeasible defeater to one's own knowledge claims. Even if we granted that Conjunction theorists know or grasp the truth-conditions of (1), it would then follow that rival theorists do not know or grasp it. Yet they all understand English; therefore knowing the language does not entail knowing or grasping the truth-conditions.
The variety of theoretical positions demonstrates that no one knows the truth-conditions of "Hitler was a kraut". Not only that, the truth-conditions are so unspecified that we could stipulate any precisely defined world, and we would still not know its truth-value. We do not know any of its truth-conditions. There is no conceivable condition under which (1) counts as categorically true according to all theories (for the Indexical Theory treats truth as relative while some Non-proposition Theories treat truth eliminatively), there is no condition under which (1) counts as categorically false by all, there is no condition under which (1) counts as relatively true by all theories, and so forth. This means that not a single truth-condition is known.
This is not a case of borderline fuzziness; this is a case where the concept at its center is inscrutable, if concepts are truth-theoretic categories. But of course the concept or meaning of "kraut" is notinscrutable; we understand someone who uses the word, we understand who is being referred to, we understand what the speaker's attitude toward the referent is, we understand not only what the speaker's feelings are but something about the discourse's level of civility (or lack thereof). We know everything there is to know about the conventions behind the word. We know its meaning, though we do not know its contribution to truth-conditions.
Truth-conditionalists may reply that pejoratives are recherché. They may say that pejoratives represent a marginal aspect of language that can fairly be ignored by semantic theory. This move, however, makes multiple mistakes. First, it is a mistake to dismiss recalcitrant data or logical difficulties by equating either rarity or small size with inconsequentiality. Observational discrepancies at the sixth decimal place prove that Einstein's theory is superior to Newton's, and Russell's single paradoxical sentence proves that naïve set theory is inconsistent. There is no such thing as a "little" inconsistency; one counter-example fells a theory.
Second, it is a mistake to think of pejoratives as aberrations. Elsewhere I relate pejoratives proper to connotation more generally, which is ubiquitous. It is not just racist and sexist slurs, but any expression having point-of-view content, that raises questions about truth-conditions.
Quite aside from connotation, the problem of identifying truth-conditions emerges for almost every kind of construction. It famously arises for definite descriptions (Russellians hold that under actual conditions "the king of France is bald" is false, Strawsonians that it is neither true nor false). The problem arises for so-called category mistakes (category errorists regard "rectangularity drinks procrastination" as obviously neither true nor false, others regard it as obviously false). The problem arises for non-declarative sentences and for attitude contents other than belief (imperatives are analyzed as simple futures by Chomsky's standard theory, as performatives by generative semanticists, as veiled threats by Bohnert 1945 and others, as sequences of declaratives by Davidson 1979…). The problem notoriously arises for belief ascriptions, for subjunctive conditionals and causal statements and probability statements, and for figurative language. Absolutists like Unger 1975 deny the strict truth of statements like "my desk is flat"; while others, relativizing to loose standards, insist on their complete truth. In the realm of moral and esthetic philosophy, normative judgments are interpreted by Kantians as consistency-claims, by utilitarians as consequence-claims, by Stevenson 1944 as exclamations, by Hare 1952 as prescriptions, by Phillips 1998 as indexical claims,… Indeed, the whole history of philosophy starting with Socrates can be read as primarily the history of trying to find truth-conditions. It is a history that, fascinating and valuable though it is, has yet to yield knowledge.
To all appearances, then, knowledge of meaning rarely if ever correlates with knowledge of truth-conditions: you can have the former without the latter. To this I have time to discuss three possible replies. The Double Ignorance Defense admits that we lack knowledge of truth-conditions, but insists that we also lack knowledge of meaning. The Double Knowledge Defense admits that we know the meanings of the sentences we utter, but insists that we actually know the truth-conditions. The Dialect Defense denies that there exists any collective "we" to which the question of knowledge applies; a given individual S understands a sentence P in one idiolect, in which case S knows P's truth-conditions, but does not understand the homophonous P* from a different idiolect, in which case S does not know P*'s truth-conditions.
These positions will be described and discarded in order.