(5) Barbara is a lawyer or Barbara is not a lawyer.
(6) Barbara is a lawyer or Barbara is a doctor.
But (5) is a simple logical truth, while (6) is a straightforward synthetic truth. Second, no connection has been provided between truth-maker theory and epistemology. Knowing a truth need not involve knowing its truthmaker; one can know (6) without knowing which disjunct is true (Barbara works in a building where only lawyers and doctors work). No account has been given as to why it should be easy from an armchair to know a truth with no truthmaker, or a truthmaker only of the special sort supposedly appropriate to analytic truths.
Nevertheless, at least one clear difference between paradigms of “analytic” and paradigms of “synthetic” is in the vicinity. For meaning that barristers are lawyers is sufficient for being true, whereas meaning that Barbara is a barrister is not. More generally, call a meaning sufficient for truth just in case necessarily, in any context any sentence with that meaning is true.11 Thus the meaning of “Barristers are lawyers” is sufficient for truth; the meaning of “Barbara is a barrister” is not. One proposal is to explicate “analytic truth” as “truth whose meaning is sufficient for truth.” Call this “modalanalyticity.”12 For non-skeptics about meaning and necessity, the notion of modal-analyticity is quite intelligible. But what are its consequences?
Consider any non-indexical sentence s that expresses a necessarily true proposition. Necessarily, in any context, any sentence with the actual meaning of s expresses that necessary truth and is therefore true. Thus s is a modal-analytic truth, because its meaning is sufficient for truth. In that sense, it is true in virtue of meaning. But how little has been achieved in so classifying it! Nothing has been done to rule out the hypothesis that it expresses a profound metaphysical necessity about the nature of the world, knowable if at all only through arduous a posteriori investigation, for instance. No reason has been provided to regard s as “merely verbal” or “insubstantial” in a pretheoretic sense, unless one already had independent reason to regard all necessities as merely verbal or insubstantial. Similarly, mathematical truths count as modal-analytic; their so counting is by itself no reason to regard them as merely verbal or insubstantial. Indeed, for all that has been said, even “Water contains H2O” is modal-analytic, given that “water” has a different meaning as used on Twin Earth to refer to XYZ, a different substance with the same superficial appearance.
To make the point vivid, call a meaning temporally sufficient for truth just in case at all times, in any context any sentence with that meaning is true. Read the quantifiers “at all times” and “in any context” non-modally, so they do not range outside the actual world. Thus any sentence which expresses, in a time-independent way, an eternally true proposition, however contingent, has a meaning temporally sufficient for truth. For example, the meaning of “No hotel ever has a billion rooms” is presumably temporally sufficient for truth. We can call the sentence “temporal-analytic” if we like, but that in no way implies that it is somehow insubstantial, because there is no background connection between eternity and some sort of insubstantiality. Similarly, calling a sentence “analytic” in the sense of modal-analyticity does not imply that it is somehow insubstantial, in the absence of a background connection between necessity and some sort of insubstantiality. Yet the account of analyticity was what was supposed to substantiate the claim of insubstantiality. If we already had a background connection between necessity and insubstantiality, there would be little to gain from invoking modal-analyticity in order to argue that core philosophical truths are insubstantial, since we could do it more simply just by arguing that true philosophical sentences in the core express necessarily true propositions.
Admittedly, not all modal-analytic true sentences express necessarily true propositions. Examples of the contingent a priori such as “It is raining if and only if it is actually raining” are modal-analytic, since the truth of “It is raining” as uttered in a given context is necessarily equivalent to the truth of “It is actually raining” as uttered in that context, because “actually” refers rigidly to the world of the context, but the biconditional does not express a necessary truth, since the weather could have been relevantly different, in which case it would have been not raining if and only if it is actually raining. Thus modal-analyticity violates Kripke”s constraint that analyticity implies necessity; in this respect it may diverge from the traditional conception. Conversely, not all sentences that express necessarily true propositions are modal-analytic: consider examples of the necessary a posteriori such as “I am not Tony Blair.” Nevertheless, such examples seem marginal to the envisaged conception of core philosophical truths, most of which will both express necessarily true propositions and be modal-analytic.
A core of philosophical truths may indeed be modal-analytic. Some philosophers seek to articulate necessary truths without essential reliance on indexicals; if they succeed, the sentences they produce are modal-analytic. Even if contextualists are right, and key philosophical terms such as “know” shift their reference across contexts, the relevant sentences may still both express necessarily true propositions and be modal-analytic: consider “Whatever is known to be the case is the case.” The answers to philosophical questions of the forms “Is it possible that P?” and “Is it necessary that P?” will themselves express necessary truths, given the principle of the widely accepted modal logic S5 that the possible is non-contingently possible and the necessary non-contingently necessary; if the answers can be phrased in non-indexical terms, they will then be modal-analytic. But outside the envisaged core many philosophically relevant truths will not be modal-analytic, as the examples near the start of the chapter show.
Unfortunately, even for modal-analytic philosophical truths, classifying them as modal-analytic does not unlock their epistemology, any more than classifying a truth as necessary explains how we can know it. Of course, if a sentence is modal-analytic, then one is safe from error in uttering it with its given meaning. In that sense, one”s utterance is reliable. But such reliability falls well short of what knowledge requires, since otherwise any true mathematical assertion would count as an expression of knowledge, no matter how fallacious the “proof” on which it was based. “Vixens are female foxes” is utterly misleading as a paradigm for the epistemology of modal-analytic truths in general. To say that s is a modal-analytic truth whose constituent words and grammar we understand does very little way to explain how we can know or justifi ably believe s.13 In particular, it does not imply that the mere linguistic understanding of s, which every competent speaker possesses, provides any insight into the truth of s, or constitutes more than the minimal starting-point for inquiry it does for ordinary synthetic truths.
4
Issues related to those just raised for modal-analyticity arise for what is sometimes called “Frege-analyticity.”14 A sentence is Frege-analytic just in case it is synonymous with a logical truth. For example, “All furze is furze” is a logical truth, roughly speaking because everything of the form “All F is F” is true. “All furze is gorse” is not a logical truth, because not everything of the form “All F is G” is true (“All fungus is grease” is false). However, “All furze is gorse” is Fregeanalytic, because it is synonymous with the logical truth “All furze is furze,” since “furze” is synonymous with “gorse.” In “Two Dogmas,” Quine admits the notion of logical truth, and therefore allows that if “synonymous” were legitimate, so would be “analytic” in the sense of Frege-analyticity. By present standards, the notion of Frege-analyticity is quite intelligible. But what are its consequences?