- 1 I am indebted to the following persons for very helpful comments and observations on previous vers (...)
1The alethic debate on realism and antirealism characterizes these two positions as ones concerning the nature of truth: the realist holds that truth is an evidentially unconstrained property. She argues that the extension of the predicate is fixed independently of what humans actually know, will know and would know, were their epistemic powers and/or the epistemic conditions an idealization of some sort of ours. The antirealist holds that truth is an epistemically constrained property; so she argues that the extension of “true” coincides with that of a (complex) epistemic predicate, like “warrantedly assertible”, “warrantedly assert-ible under idealized epistemic conditions”, “incontrovertibly warrantedly assertible”, etc.
2The realist is committed to denying that
(R*) it is true that p if and only if it is warrantedly assertible that p.
3(R*) asserts that “true” and “warrantedly assertible” coincide in extension. The realist is committed to rejecting it because of her endorsement of a non-epistemic conception of truth, according to which the existence of truths for which a warrant is even in principle unavailable constitutes a possibility.
4Now consider the following claim:
(R) Every reason to think that a statement is true is a reason to think that the statement is warrantedly assertible, and vice versa.
5Apparently, (R) does not entail that “true” and “warrantedly assertible” coincide in extension. It just says that, as a matter of conceptual necessity, every reason to think that a statement is true is a reason to think that the statement is warrantedly assertible, and vice versa. This means that the statements “it is warrantedly assertible that p” and “it is true that p” are cognitively equivalent; which simply means that “it is warrantedly assertible that p” and “it is true that p” have the same conditions of assertibility. So, it would seem, the realist is not committed to rejecting (R) as she is to rejecting (R*).
6In A realist Conception ofTruth W. Alston has claimed that endorsing (R) presupposes endorsing (R*). Since (R*) voices a commitment inconsistent with the realist identification of truth with a non-epistemic property, Alston has claimed that the realist must also reject (R). According to Alston, (R) is in good standing only if one presupposes that “what ES [the Equivalence Schema, according to which every instance of ”it is true that p iff p“ holds necessarily] says to be equivalent to the truth of p is some positive epistemic status of the assertion of p” [Alston 1996, 219]. If one reads (ES) in this way—Alston insists—one immediately gets to (R*). So, as long as the meta-linguistic reading of (ES) is required to have (R), the acceptation of (R) cannot be divorced from the acceptation of (R*).
7As against Alston’s suggestion, I will show (i) that (R) can be derived without endorsing the so-called meta-linguistic reading of (ES). On the contrary, I will show that (R) can be derived by uncontroversial assumptions concerning the role performed by rather than the correct reading of (ES). As a consequence, I will show that the realist is not prevented from accepting it. Finally, I will argue (ii) that the realist had better accept it; for (R) can be used by the realist to put forward an argument against the antirealist. The final part of the paper will be devoted to outline this argument.
8Recall an old-fashioned piece of philosophical exemplification. There is a person, Caroline, who knows that the first planet she sees in the sky after the sunset is Venus. She learned that the expression “the evening star” is a quite common label people uses to denote the planet, and, upon being consulted on matters of astrophysical identity, is always prone to maintain that the evening star is indeed Venus. An unlearned friend of Caroline, Tom, heard something about the existence of a star commonly referred to as the “morning star”, in virtue of the fact that it is the least star to disappear before the sunrise. Tom asks Caroline the weird question whether, because of the similarity between the two labels, the heavenly body denoted by “the evening star” and the star denoted by the “morning star” are one and the same star (actually Tom does not know which planet that “star” would be). Carolina, who does not know that the “morning star” too denotes Venus, unintendedly delivers the false answer that Venus is not the morning star. Though perhaps this is not its most interesting achievement, the philosophical literature has unanimously granted that there is nothing wrong in Caroline’s answer, nothing at least concerning her rationality (a rational agent has the right, at times, to ignore something of the world she rationally inhabits), nor her linguistic competence (the example works on the presupposition that the expression “the morning star” can be understood without mastering a full theory concerning its denotatum).
9If one asked what fault could have been attributed to Caroline beside or instead her (partial) ignorance, consider the following case (nothing wrong if, again, the reader’s mind is surfaced by the suspicion of a philosophical déjà-vu). Caroline is well aware of the fact that Tom, her unlearned friend, is not married, and is used to tell her female friends this fact, in the hope that one might get interested in the poor Tom (normally no one does). However, one day Laura—unexpectedly driven by her interest for Tom asks Carolina whether Tom is still a bachelor. This could be a good occasion for Tom; unfortunately, Carolina delivers the false answer that he is not anymore. As before, Carolina wanted to be sincere. So, if she is guilty of something, this is not to be found in her propensity to tell the truth. However she is not ignoring that Tom is a bachelor: she knows that he is not married. What is wrong? As more than one reader might have guessed, we are faced by an alternative: either Carolina does not know the meaning of the expression “bachelor”, or, if she does know it, she is being logically faulty: she does not seem to appreciate that if S has the property A, and the property A = the property B, then S has the property B.
10In both cases Caroline tells a falsehood. The second time, however, we have the strong intuition that more than ignorance is involved. We do not stay content with the supposition that she just ignores the fact that Tom is still a bachelor. What is the difference between the two cases? Why is Carolina faulty in the second case differently than in the first case? The difference, though arguably not its most interesting explanation, is that the statements
(1) The evening star is Venus, and
(2) The morning star is Venus,
(3) Tom is not married, and
(4) Tom is a bachelor,
are not cognitively equivalent.
- 2 Künne, who has recently re-called the philosophical attention to Frege’s notion of cognitive equiv (...)
11Two statements are cognitively equivalent whenever “for any context c, nobody who fully understands them can take one of them to express a truth with respect to c without immediately being ready to take the other to express a truth with respect to c as well”2. The idea is that some statements, for reasons which have to do with the meaning of the expressions they feature, are such that no one, unless excusable for a lack of semantic competence or of logical capacities, can take a different attitude toward them: if one accepts the first, then one accepts the second; if one doubts the first, then one doubts the second, etc. The second example pivots on the identity of meaning between “bachelor” and “unmarried man”. Since they mean one and the same thing, (3) and (4) are cognitively equivalent. No one can accept the first and reject (doubt, simply weight, etc.) the second, and vice versa, unless falling sort of being a competent user of both expressions (henceforth I shall drop the second alternative concerning logical mistake, and shall maintain that whenever two statements are cognitively equivalent, S’s failure to take the same attitude toward both statements signals her semantic incompetence, i.e. her inability to understand what either the first or the second statement says). The first case, on the contrary, differs because (1) and (2) are not cognitively equivalent: “the morning star” and “the evening star”, though coincide in extension, have a different meaning. So that a thinker who accepts the first and rejects the second needs not be charged of something like incompetence in the use of either expression.
12Identity of meaning (as it is arguably the case for “bachelor” and “unmarried man”) does not constitute a necessary condition for cognitive equivalence. It arguably supplies a sufficient condition for it, if the relevant statements are of the form “S is p” and “S is q” and the meaning of “p” is the same as the meaning of “q”: in such case, no one understanding both statements can, for instance, believe the first and disbelieve the second.
13However, consider a biconditional B, whose left-hand side and whose right hand-side are not of the indicated form. Suppose that B performs a definitional role of a devised form of one of the expressions, say featured either by B’s right-hand side or by B’s left-hand side. In this case B would be what is commonly called an implicit definition of much in the same way Hilbert thought that Euclid’s axioms are implicit definitions of geometrical expressions like “point”, “line”, etc. Roughly, an implicit definition is supposed to work as follows: suppose that the expression “#” is undefined, and that it is introduced within a language by the stipulation that it means whatever makes true a (partially unin-terpreted) formula featuring the expression “#” among its sub-sentential constituents. According to the theory, the acceptance of the formula as true affects that “#” acquires the intended meaning, in that whoever thinks that the formula is true is thereby affecting that the content the formula conveys is true, and so that “#” receive the intended meaning. Accordingly, “point”, in Hilbert’s story, means whatever makes true the axioms which feature among their constituents that very expression.
14Suppose, as more than one does, that the meaning of some expression is successfully introduced by means of an implicit definition. Suppose further that our B is indeed a implicit definition of “#”. If B is an implicit definition, acceptance of B constitutes a necessary condition for being a competent user of “#”. Since “#” means whatever makes B true, falling short of appreciating the truth of B is to fall short of understanding the meaning of “#”. Now remember that B is of a bi-conditional form. So it is of the form p iff q, where “#” is either a constituent of “p” or a constituent of “q”, no matter whose constituent it is. If it is so, failure to accept the transition from p to q and from q to p is a failure to count as competent user of the expression “#”. Acceptance of a bi-conditional may in fact be broken down into the unconditional acceptance of the transition from its left-hand side to its right hand, and of the transition from its right hand-side to its left-hand side. So, whoever accepts p and rejects q, or accepts q and rejects p, falls short of being a competent user of “#”. Moreover, what the implicit definitional status of B seems to rule out is the consistency of a difference in attitude toward its left-hand side and its right-hand side, and the supposition that this difference is not to be attributed to one’s linguistic incompetence. For every attitude φ, a different attitude ξ is ruled out by the acceptance of B, whose bi-conditional form guarantees p’s and q’s covariance in truth-value. Whoever accepts B, and takes the attitude φ toward p is thereby committed to taking the same attitude toward q (and vice-versa). Accordingly, if B is a biconditional, and performs the semantic role of an implicit definition, its left-hand side and its right-hand side are cognitively equivalent: the acceptance of B constitutes a necessary condition for understanding and the acceptance of B rules out the possibility of a difference in attitude toward B’s left-hand and right-hand sides, so that whoever takes different attitudes toward them falls short of being competent with “#”. Which is what the cognitively equivalence of a pair of statements requires.
15Does (ES) plays an implicit definitional role along the lines of the preceding section? Assessing the question whether a context “B” is to be regarded as an implicit definition of one of its constituting expressions “#” arguably requires assessing two distinct sub-questions: whether mastery of “#” can be credited independently of the doxastic attitude one is disposed to take toward “B” (i.e. whether acceptation of “B” constitutes a necessary condition on the understanding of “#”), and whether mastery of “#” is entailed by the acceptation of “B” (i.e. whether acceptation of “B” constitutes a sufficient condition on the understanding of “#”).
16To say that acceptation of “B” provides a necessary condition on the understanding of “#” is to say that every case in which one is reluctant to accept “B” is to be regarded as a case in which one does not understand Acceptation of (ES) seems, in light of the following example, to possess this feature. Tom (again) tells Caroline an unbelievable story concerning his relationship with Laura: “We have been to New York for the week-end”. Caroline is not convinced because she knows that both Tom and Laura live in Austria. So she asks Tom whether it is true that they have been to New York. Tom apparently weights the question, and delivers the answer that it is not true. However, he finally adds: “but as I already told you, we have been to New York”. What should we think of Tom? His linguistic performance reveals that he fails to appreciate the fact that “We have been to New York” does entail “it is true that we have been to New York”. Differently than before, however, we are not disposed to treat Tom’s defiance as immaterial to the question whether he understands the predicate “true”: he is not sensitive to the fact that to assert is to present as true, so that his asserting that p commits him to assert that it is true that p. So, we are disposed to treat Tom as one who fails to grasp the concept of truth. It would seem that this example could be easily generalised to every p. So that, it might be suggested, the acceptation of (ES) seems to provide a necessary condition for the understanding of the truth-predicate.
17As against the foregoing suggestion, it might be argued that having the disposition to accept every instance of (ES) provides too strong a condition on the understanding of “it is true”. After all, it might be suggested, no rational thinker can be disposed to accept the transition from “what I now say by uttering this sentence is false” to “it is true that what I now say by uttering this sentence is false”. Such transition would entail the utterance of a paradoxical claim, which never seems to be possibly true. If what I say is true that is to say if I what I now say by uttering the first sentence is false — then what I say when uttering it is false. If false, however, than what I say is true.
18This rejoinder can seemingly be dealt with by imposing the constraint that it is the disposition to accept all non-pathological instances of (ES) that constitutes a necessary condition on the understanding of the truth predicate. However, once such constraint is in place, it seems that understanding the truth predicate requires having the disposition to recognise as such all pathological instances of (ES). Can the latter observation be made compatible with the claim that the qualified acceptation of (ES) provides a necessary condition on the understanding of the truth predicate?
19A negative answer is plausibly enforced by the observation that if one is in a position to comply to the qualified condition, one must already be in possession of the concept of truth (as constituted by the acceptation of the claim that if a statement is true, then it is not false). If one reads the “already” in a logical manner, then the possession of the concept of truth must be constituted much independently of the qualified disposition to accepting every non-pathological instance of (ES). Accordingly, the acceptation of every non-pathological instance of (ES) cannot constitute a necessary condition for understanding the truth-predicate.
20However, it might be suggested that the concept of truth has an holistic nature, such that its possession is constituted by several conditions, of which the condition that if something is true, then it is not false, and the condition that every non-pathological instance of (ES) hold good constitutes just two different instances. Were it so, possession of the concept of truth would be constituted by the joint acceptation of all the basic conditions.
21What about the sufficiency condition? If we allow for the holistic nature of the truth concept i.e. if we allow the acceptation of more than one principle to constitute the possession of the truth-concept it hardly constitutes a possibility that one of those principles, besides providing a necessary condition, also provides a sufficient one for possessing the truth concept. For if a condition for possessing a concept is both necessary and sufficient, then meeting the condition is all that is required to possess the concept. Accordingly, if we salvage the contention that the acceptation (ES) provides a necessary condition for understanding the truth predicate by constraining the range of the instances one is to be disposed to accept just to its non-pathological instances, then it must be at the cost of denying that acceptation of (ES) also provides a sufficient condition for the possession of the truth concept.
22Does this result impede the characterisation of (ES) as an implicit definition of the predicate “it is true that”? According to the foregoing provisos, this question deserves an affirmative answer. Acceptation of (ES) does in fact not provide a necessary and sufficient condition for understanding the truth predicate. However, it resulted from the foregoing considerations that the qualified acceptation of (ES) does indeed provide a necessary condition. If we allow that a context partly constitutes the understanding of an expression if accepting the context constitutes a necessary condition for understanding it, we can then propose to regard ( ES) as a partial implicit deinition of the truth predicate.
23Nothing significant happens concerning the relation between the notion of implicit definition and that of cognitive equivalence if partial implicit definitions are allowed as legitimate implicit definitions of an expression. It remains true that if a (partial or non-partial) implicit definition has a biconditional form, its left-hand side and its right-hand side are cognitively equivalent.
24Let us illustrate the point by concentrating on the case most interesting for us, that is to say on (ES). Along with the foregoing considerations, suppose that the acceptation of every (non pathological instance) of (ES) constitutes a necessary (though non-sufficient) condition of the possession of the concept of truth. That the qualified acceptation of (ES) constitutes a necessary condition to possess the truth concept guarantees that whoever falls short of the disposition to accept all (non-pathological) instances of (ES) does not possess the concept of truth. Accordingly, for any arbitrary p that does not give rise to a pathological instance of (ES), acceptation of “p” and rejection of “it is true that p” is not consistent with the possession of the concept of truth. Conversely, whoever possesses the concepts necessary to grasp both what “p” says and what “it is true that p” says cannot but be disposed to take the very same propositional attitude towards both statements. For if she has the concepts to understand both “p” and “it is true that p”, a fortiori she possesses the concept of truth, and the possession of concept of truth, so long as it is also constrained by the qualified acceptation of the covari-ance in truth value of “p” and “it is true that p” encapsulated by (ES), guarantees the necessary convergence of propositional attitudes.
25In Alston’s own terminology, the Equivalence Schema is referred to as the T-schema. Accordingly, every statement that derives from the uniform substitution of “p” with a declarative sentence, is, in Alston’s terminology, a T-statement. Does Alston attribute to the acceptation of every T-statement the role of constituting the understanding of the truth-predicate? This would be equivalent to attributing a constitutive role to the acceptation of (ES). Given the result of the preceding section, an affirmative answer to this question would then show that Alston is committed to the claim that “p” and it is true that “p” are cognitively equivalent.
26The following quotation from A Realist Conception of Truth seemingly provides promising evidence as to Alston’s endorsement of the implicit definitional role of (ES):
27The suggestion is that if we understand that any T-statement is conceptually, analytically true, true by virtue of the meaning of the terms involved, in particular the term ‘true’, then we thereby understand what it is for a proposition to be true. [Alston 1996, 27]
28However, there is one difficulty in construing Alston as endorsing the conception that (ES) implicitly defines ”true“. The problem is that Alston seemingly requires one to appreciate that every T-statement (every instance of (ES)) is conceptually true for her to understand the predicate ”true“. However, the attribution of an implicit definitional role to (ES) just requires a thinker to accept every instance thereof, much independently of the recognition that every such instance is conceptually, not to say analytically true.
29More than this, Alston is quite explicit in declaring that the T-schema is not a definition, not even a contextual definition, of ”true“. The reason is that, unlike in a proper definition, the right-hand side and the left-hand side of (ES) are not synonymous.
30The meaning of ”The proposition that lemons are sour is true“ cannot be the same as that of ”Lemons are sour“. The former has conceptual content absent from the latter. One could understand ”Lemons are sour“ perfectly well without having any concept of truth whatever. [Alston 1996, 34]
31Two considerations are in order. The first one concerns the role Alston attributes to (ES). If my reading of the foregoing quotation is correct, Alston is implausibly making the capability to understand the truth-predicate depend on the capability of entertaining propositional attitudes of the form ”it is conceptually true that —“ and ”it is analytically true that —“. The implausibility stems from the fact that Alston’s account of the way the understanding of the truth-predicate can be imparted circularly requires, on the part of those who still lack the linguistic resources to understand it, the capability of entertaining complex attitudes toward — and therefore of understanding — contents featuring the very concept expressed by the predicate. Where the understanding of ”true“ to depend on the satisfaction of Alston’s condition, no one could understand that predicate.
32Second observation: both the claim that the truth-predicate ”is the only predicate such that, when one appreciates that the statement generated by that insertion [within the schema: for any p, p is_iff p] is conceptually true, one is thereby in possession of the concept expressed by that predicate“ [Alston 1996, 54], and the claim that (ES)’s left-hand side and right-hand side are not synonymous, are consistent with the cognitive equivalence of the left-hand side and right-hand side of (ES).
33As to the first contention, consider, as Alston seemingly suggests, that S is able to understand every statement of the form ”it is true that p“, only if S understand that every instance of (ES) is conceptually, analytically true. If it is so, for every S possessing that capability, we might derive the consequence that S knows, for every p, that if it is true that p, then p, and that if p, then it is true that p. Knowledge that every instance of (ES) is conceptually, analytically true certainly entails knowledge that the transition from its left-hand side to its right-hand side, and vice versa, holds good.
34As to the second contention, if ”p“ and ”it is true that p“ are not synonymous, then one being in a position to understand ”p“ can plainly be in a position not to understand that ”it is true that p“. For in addition to possessing the conceptual resources necessary to understand that p, she also has to possess the conceptual resources necessary to understand that is true that___. However, once it is granted that, for some S, S does possess both kinds of conceptual resources, and once it is granted that the conceptual resources necessary to understand the truth predicate are acquired in the indicated way, it follows that ”p“ and ”it is true that p“ are cognitive equivalent. No one understanding both of them can take one of them to express a truth (a falsehood, or whatever) without at the same time being ready to take the other one to express a truth (a falsehood, whatever) as well.
35I conclude this section by maintaining that Alston’s stance toward (ES) does indeed commit him to the view that ”p“ and ”it is true that p“ are cognitively equivalent. Therefore, the following argument, taking as its first premise the claim that both statements are cognitively equivalent, is acceptable by Alston’s own lights.
36To say that ”p“ and ”it is true that p“ are cognitively equivalent is to say that
(i) R is a reason to believe that p iff R is a reason to believe that it is true that p.
The cognitive equivalence thesis ensures that no consideration possibly enjoins acceptance of p unless it also enjoins acceptance that it is true that p, and vice versa. Existed an R whose substitution within (i) gave rise to a false statement, there would be circumstances under which a thinker fully understanding both sides of (ES) would be rationally required (permitted) to take different propositional attitudes toward them; which is inconsistent with the statements’ cognitive equivalence. Now consider the following premise:
(ii) R is a reason to believe that p iff R is a reason to assert that p.
- 3 (ii), as a consequence, excludes as reasons for an assertion prudential considerations, or motivat (...)
(ii) should be taken to be a definitional truth. With this I mean that (ii) just codifies the purely epistemic sense in which the notion of ”reason for an assertion“ is referred to within the debate over semantic realism and anti-realism3.
37By substituting (i)’s left-hand side with (ii)’s right-hand side we yield
(iii) R is a reason to assert that p iff R is a reason to believe that it is true that p.
If every reason to assert is a reason to believe, and vice versa, and every reason to believe in a proposition is a reason to believe that the proposition is true, and vice versa, then every reason to assert a proposition is a reason to believe it true, and vice versa, that is to say (iii). As a last premise consider now the following one:
(iv) R is a reason to assert that p iff R is a reason to believe that it is warrantedly assertible that p
- 4 (iv) involves no commitment as to the contents a thinker must be prepared to accept if she is to c (...)
- 5 It must also be noted that “reason“ and “warrant“ are not used in this context as synonymous. This (...)
38(iv) too holds true on conceptual grounds. It just spells out the iterative nature of the reasons for an assertion. If one has a reason to assert a proposition, then one has a reason to believe (assert) that the proposition is assertible. Conversely, if one has a reason to believe (assert) that a given proposition is assertible, then one has a reason to belie (assert) the proposition4,5. It suffices to substitute (iv)’s right-hand side for the left-hand side of (iii) to yield
(iv) R is a reason to believe that it is warrantedly assertible that p iff R is a reason to believe that that it is true that p,
- 6 In his recent Conceptions of Truth Wolfgang Kunne offers a new version of Fitch’s argument against (...)
39which is the very condition (R)6.
40In the light of the foregoing argument, it should be clear that acceptance of (R) does not presuppose the acceptance of the meta-linguistic reading of (ES). So, also if one accepts Alston’s claim that such reading is false, acceptance of (R) is not inconsistent with the acceptance of (ES) (in its correct reading). For (R) derives from the further principle (i) that is consequent upon the supposition — arguably shared by Alston — that acceptation of (ES) is indeed necessary to grasp the concept of truth. Contrary to Alston’s suggestion, then, (R) does not beg the realist’s question. Its acceptation does not depend on the further claim, unacceptable to the realist, that ”it is true that p“ and ”it is warrantedly assertible that p“ are extensionally equivalent.
41Before concluding, I would like to stress an important respect under which the realist should welcome this result. The next section is in fact devoted at outlining a very general argument against the semantic anti-realist, taking as its main premise (R).
42As it is commonly suggested, ”the essence of an anti-realist conception of meaning [is] given by the thesis that a sentence’s meaning or content is given in terms of its assertion conditions“ [Skorupski 1993, 133]. Such conception straightforwardly derives from the antirealist identification of truth with an epistemic notion: understanding meaning is in fact held both by the realist and by the antirealist to consist in grasp of truth-conditions; if, along with the antirealist, truth is (some kind of) assertibility, then understanding meaning is to grasp conditions of assertibility. Let us state the antirealist claim in the following form:
(a) Meaning is assertibility conditions.
43If meaning is assertibility conditions, then two sentences having the same assertibility conditions have the same meaning. Now (R) says that if R is a reason to assert that it is true that p, then R is a reason to assert that it is warrantedly assertible that p. This claim can be put equivalently as a claim to the effect that ”it is true that p“ and ”it is warrantedly assertible that p“ have the same assertibility conditions. In fact, (R) guarantees that every reason in the light of which ”it is true that p“ is assertible is a reason in the light of which ”it is warrantedly assertible that p“ is assertible, and vice versa.
44So, acceptation of (a) commits the antirealist to accept that
(b) “it is true that p” and “it is warrantedly assertible that p” have the same meaning.
45Now consider a given p which, under an available body of information E, is still undecided (where agnosticism is the right epistemic attitude toward p). In this case, it is both assertible that it is not the case that it is assertible that p and it is not the case that it is assertible that ¬p. If it is assertible that it is not the case that it is assertible that ¬p, then it is not assertible that ¬p is true. The antirealist is in fact committed to the claim that ¬p’s truth consists in (some form of) warranted assertibility enjoyed by ¬p. Accordingly, when p is undecided it can be assertible that it is not the case that it is assertible that p, and at the same time it can fail to be assertible that it is not the case that p is true (equivalently, that ¬p is true). So,
(c) “it is not the case that it true that p” and “it is not the case that it is warrantedly assertible that p” have not the same assertibility conditions, hence they do not have the same meaning
46(b) and (c) are actually inconsistent with
(d) The meaning of a complex sentence is a function of the meanings of its constituents.
- 7 The argument, in its essential lines, is already sketched in [Skorupski 1993].
47The statements mentioned in (c) result form embedding the statements mentioned in (b) with the phrase ”it is not the case that". According to (b), the constituent statements have the same meaning. Therefore, (d) entails that also the complex statements they constitute should have the same meaning7.
48If the first conclusion is sound, then the antirealist is committed to accepting premise (b) of the foregoing argument. Accordingly, the only options available to her seem to reduce either to reject, along with (d), the compositionality of meaning, or to reject the very premise (a), voicing her essential commitment to an epistemic conception of meaning. This is not the right place to press the intuitive point that the latter alternative does not look as much as available to the antirealist. The compositionality of meaning doesn’t seem to be a negotiable belief: its rejection in fact threatens to make language learning an impossible task. The main conclusion I want to draw is rather the following: so long as the foregoing argument is intuitively appealing, it apparently forces the antirealist to recede from her commitment to the epistemic nature of meaning. Since the argument essentially features premise (R), its very existence constitutes good reason, in contraposition with the worries voiced by Alston, for the realist to accept it.