Stephen Long: truth requires metaphysics

In Speaking of God, Long is eager to defend a moral and metaphysical realism within theology: Christians, he says, must make ‘full-blooded truth claims’ (Long (Reference Long2009), 268). And realism in theology is threatened, Long argues, by, inter alia, correspondence theories of truth. For Long, most threatening about correspondence theories is that they are part of a broader tradition he characterizes as ‘nominalistic’ in regard to truth. What he calls the nominalistic approach to truth is characterized by two things he finds troublesome: the notion that the primary truth-bearers are linguistic (whether propositions, sentences, or utterances), and separate explications of truth and epistemic justification (ibid., 296).

Long states that the position that ‘truth-bearing vehicles are primarily linguistic – sentences, statements, utterances . . . assume[s] that the human agent is alienated from the world and must find her way back to it’ (ibid.). This ‘alienation’ in turn leads to an epistemology ‘that mediates between the world and us primarily through language’ (ibid.). This mediation is not explicated, but Long believes it will ‘inevitably lead to a deflationary theory of truth’ (ibid.).Footnote ¹ For Long, underlying this risk of an implicit deflationism is a further risk that the truth about Jesus will be seen as being established solely by communal agreement. And this in turn would result in a kind of ‘fallibilism’, in which the community could always modify its agreement if it wished. Long, in contrast, wants a more secure foundation for theological truth claims about Jesus (ibid., 323).

The other problem Long sees with a nominalistic approach to truth is that it involves separate explications of truth and epistemic justification. Explicating these two concepts separately, Long believes, will result in either an epistemic or a deflationary understanding of truth (ibid., 272–273). In fact, any notion of truth that abstracts from metaphysical commitments, Long believes, at least implies, if not entails, that metaphysics is neither possible nor desirable (ibid., 277–279). For any such notion of truth, Long asks, ‘How is truth here not just a metaphysical regulative ideal that assumes no one has access to the real-in-itself, which we know infallibly? If so, then the real violation of the communal agreement will be to make substantive truth claims’ (ibid., 279).

Long's discussion of ‘fallibilism’ highlights the difficulty of conflating epistemology and truth. It certainly seems reasonable to think that I should hold all my beliefs open to revision should I learn something that proves one or more of them wrong. But Long says that fallibilism asks more: it asks us ‘to recognize that every claim to truth comes with an admixture of error’ (ibid., 287).Footnote ² And Long argues that makes it self-refuting.Footnote ³ Now this is confused, for Long's rhetoric masks a quantifier shift. To hold all my beliefs open to revision is simply to say that it is possible that some of my beliefs might be wrong. It doesn't imply that they are all partially wrong (which is what I take to be the meaning of ‘every claim to truth comes with an admixture of error’). If it did, then what Long calls fallibilism would be a variation of the Liar's paradox. Let's call the claim it is possible that some of my beliefs might be wrong the ‘weak fallibilist thesis’. This thesis involves no self-referential problems: even if it turned out that all my beliefs were true, this would not show the weak fallibilist thesis even to be false, much less self-refuting. For the weak fallibilist thesis to be false, all my beliefs would have to be not only true, but necessarily true.

Like Marshall, Long thinks it important primarily to speak of Christ as the truth (ibid., 295; see also ibid., 287). This coheres with Long's insight that metaphysical commitments fund much of what takes place in theology. But from this insight, Long moves to what Hector would call a substantive account of truth, drawing from Balthasar and Vico. From my perspective, however, Long is making a number of robust ontological claims, and doing so in the vocabulary of truth theory. I don't wish to object to any of Long's ontological claims (at least not here), but in my view doing so in the vocabulary of truth theory introduces needless confusion. Much of what I see as confusion in Long's account stems from his commitment to the thesis that theorizing about truth cannot abstract from metaphysical or ontological questions. And this commitment is one of the reasons he doesn't want to rely too heavily on analytic philosophers. Analytic philosophy, he says, ‘divides up truth, epistemology, and metaphysics into discrete units of analysis without attending to how they influence each other. It loses any “holistic” approach to the world we actually live in where such things inevitably relate to each other’ (ibid., 272 n. 19). And this is why he objects to separate explications of truth and epistemic justification. Yet I don't see why one can't give separate explications of truth and epistemic justification and then also give an account of how they are related. And in Long's book I don't find much in the way of argument as to why one can't do this. More generally, I don't see why treating ‘truth, epistemology, and metaphysics [as] discrete units of analysis’ causes any problems so long as ‘attending to how they influence each other’ is not precluded by the analyses themselves. As I'll mention below, in my own view taking the Trinity as a truth-maker rather than a truth-bearer explicitly attends to the way in which truth and metaphysics influence each other. Furthermore, it's more consistent with Long's insight regarding the unavoidable role of metaphysics in theology.

Kevin Hector: metaphysics ruins truth

The threat that Long fears – epistemic or coherence notions of truth – is openly embraced by Kevin Hector. Long resists the influence of Donald Davidson; Hector welcomes it. Hector believes we should ‘understand truth in terms of the practice of taking true’ (Hector (Reference Hector2011), 201). Long rejects Davidson's view that ‘nothing can count as a reason for holding a belief except another belief’ (Davidson (Reference Davidson and LePore1986), 310, quoted in Long (Reference Long2009), 296). In contrast, Hector adopts it into his own understanding of truth: ‘judging some belief to be correct on the basis of one's other beliefs, and thus using it to judge still other beliefs’ (Hector (Reference Hector2011), 201).Footnote ⁴

Hector embraces Davidson because he wants believers to have direct access to God with nothing standing between them. And he construes correspondence theories such that they permit concepts to ‘stand between’ us and God. And the crucial thing that sets up such conceptual barriers is the isomorphism between mind and object that Hector sees as integral to those correspondence theories he wants to reject: ‘one's belief about an object is true if and only if it corresponds to an object's form, and that object is itself true if and only if it corresponds to that same form’ (ibid., 205).Footnote ⁵ Hector is aware that many correspondence accounts of truth do not have an isomorphic understanding of correspondence.Footnote ⁶ Thus, he objects to ‘a rather narrow family of accounts’ of truth, and he says it might be wise to call them ‘correspondentist’ accounts. Nonetheless, having made this distinction, he alternates, sometimes calling the objectionable theories ‘correspondentist’ and sometimes ‘correspondence’ theories (ibid., 205–206).

But beyond isomorphism, Hector objects to ‘the very idea of explaining truth in terms of such an appeal [to isomorphism], that is, of thinking that such correspondence could be called upon to explain what one is (or ought to be) doing when one takes some belief to be true’ (ibid., 206; emphasis in original). Hector objects because he thinks it makes one's ideas the measure of a thing (including God), thus ‘cutting it down to the size of one's categories’ (ibid.). Such an understanding of truth can offer us only a God of our own construction (ibid.).

Hector takes this Feuerbachian problem to be related to a more general problem, typically raised by epistemic theorists. How am I to determine whether my statement or thought corresponds to some extra-mental reality? How can I determine whether my belief that This apple is red is true, i.e. whether it corresponds to reality? Hector thinks I can't, because doing so requires perception, and perception requires the application of concepts, which Hector takes to be a species of belief formation (ibid., 208).Footnote ⁷ If Hector is right about this, he believes, it raises an infinite regress problem. To determine whether That apple is red is true, I have to make a judgement about whether the statement corresponds to the reality of the apple. But then I have to make a meta-judgement about whether my judgement on the correspondence relation is true. And then I must decide (or assume) that my meta-judgement is true. And so on, ad infinitum. Hector believes this will be an insurmountable problem, not only for correspondence theories, but for any ‘substantive’ understanding of truth (ibid., 210–211). The solution Hector proposes is an epistemic approach to truth: ‘to invert the customary order of explanation, so that instead of using a substantive theory to explain the practice of taking-true, we can use the practice of taking-true to explain whatever needs to be explained about the concept of truth’ (ibid., 211). He will then add to this a theological overlay whereby the Holy Spirit conforms our beliefs to Christ.

Though Hector's objections to correspondence theories are very different from Long's, their strategies bear some structural similarities. Long introduces metaphysical issues into his understanding of truth theories, because he fears correspondence theories in the analytic tradition imply an anti-metaphysical stance. Hector objects to those correspondence theories that have isomorphism as an integral part of them. Now where one sees isomorphism as a part of a ‘correspondentist’ theory, it's often because the theory has been integrated into an overarching metaphysical scheme (e.g. in Thomas). And Hector rejects such theories as part of his project of doing theology without metaphysics. But he also introduces epistemological issues that he believes raise problems for correspondence theories across the board. To solve these he rejects correspondence theories in favour of an epistemic approach to truth.

Bruce Marshall and Davidson's slingshot

Marshall's most sustained critique of correspondence theories is set out in Trinity and Truth (Reference Marshall2000), and I'll limit my comments to that nuanced and erudite text. Marshall wants to preserve the central aspects of the theological method Frei and Lindbeck appropriated from Barth. Frei in particular maintained that to argue for the truth of any theological claim by reference to some independent notion of truth is to submit to an ‘all fields-encompassing scheme’ that inevitably distorts the Christian message. Yet contrary to evangelical critiques, Frei wanted to remain a theological realist, and Marshall follows him in this (Marshall (Reference Marshall2000), 117). Much of Marshall's book on truth is concerned with the epistemic justification of theological beliefs. He doesn't focus on the concept of truth until chapter 8 (the next-to-last chapter). And there, the first section in the chapter is ‘Realism in search of a truth-bearer’. By the end of the chapter he will say that a theological account of truth will need some version of correspondence, for he wants to defend postliberal theology as realist when it comes to truth. And the most common way to preserve realism in truth theory is via a correspondence theory. So Marshall does not want correspondence theories (and a fortiori the notion of truth) to be ‘annihilated’. But he does want them to be ‘radically disciplined and changed’, because, he argues, no (non-theological) correspondence theory of truth can work unless it is subjected to Trinitarian discipline (ibid., 242; quoting MacKinnon ([Reference MacKinnon1979] 2011), 81).

For assistance in making this argument, Marshall turns to Donald Davidson.Footnote ⁸ Marshall uses Davidson's famous ‘slingshot’ argument to describe what Marshall takes to be fatal deficiencies in any non-theological correspondence theory of truth.Footnote ⁹ This type of argument was developed long before Davidson and often has been seen as establishing limits on metaphysical theories regarding facts. In consequence, it has been appropriated in criticisms of a variety of theories that have facts as a constituent, such as theories of truth, correspondence, and the like.Footnote ¹⁰ Davidson, and Marshall following him, use it to argue that correspondence theories of truth are ‘not so much false as empty’, as Marshall summarizes it, ‘and so useless for saying what truth is’ (Marshall (Reference Marshall2000), 226).Footnote ¹¹

The slingshot argument seems to have originated with Alonzo Church, who developed it using certain ideas of Frege (see Church (Reference Church1943), 298–304; Idem (Reference Church1956); see also Frege (Reference Frege and Black1948)). The argument turns on two crucial assumptions. First, logically equivalent sentences have the same reference (see Barwise & Perry (Reference Barwise and Perry1983), 24). This is an extrapolation from the formal Principle of Substitutivity for Logical Equivalents (PSLE). Second, the reference of a sentence does not change if a component singular term is replaced by another with the same reference. Again, this is an extrapolation from the formal Principal of Substitutivity for Singular Terms (PSST). Using these two assumptions, proponents of the slingshot will identify two sentences with different meanings and attempt to show that they refer to the same fact (Young (Reference Young2002), 124). This is exactly the way Marshall uses it. According to PSST, co-referential singular terms can be substituted salva veritate (without any change in truth value) in sentences that create an ‘extensional context’ (Marshall (Reference Marshall2000), 228). As for PSLE, as Marshall explains it, ‘in an extensional context, logically equivalent sentences, like co-extensive terms, may be substituted without changing the truth value of the expression in which the substitution occurs’ (ibid.). Two questions naturally arise: ‘What is an extensional context?’ and ‘Do factual contexts count as extensional?’ As Marshall suggests, extensional contexts can be simply defined as those in which substitution of co-referential terms necessarily preserves truth value. Unfortunately, Marshall is imprecise when explaining PSLE. He says that ‘sentences are logically equivalent which must always have the same truth value’ (ibid.). According to the standard definition of logical equivalence, however, ‘two sentences are logically equivalent if and only if they are true in precisely the same models or possible worlds’ (Young (Reference Young2002), 124). In defining extensional contexts, Marshall uses the term ‘must’, but leaves out the modal term ‘necessarily’. This is potentially important, because, as we'll see, the facts Marshall uses as evidence that factual contexts are extensional are not necessarily true. Thus, since I have only one brother (Jim), are the two sentences ‘John threw the ball to Jim’ and ‘John threw the ball to his brother’ logically equivalent? They have the same truth value in the actual world. But there is a possible world W in which Jim is John's cousin and not his brother, and in which John did indeed throw the ball to Jim. In W, then, the two sentences have different truth values, and this shows that they are not logically equivalent.

But what about factual contexts? Do they count as extensional contexts? Marshall thinks so based on the following example. Since St Olaf College is located in Northfield, the definite description ‘the town where St Olaf is located’ can be substituted (at least in the actual world) for Northfield and retain the same truth value. So, Marshall concludes that the sentential operator ‘. . . corresponds to the fact that . . .’ creates an extensional context. An extensional context, he quite rightly says, is typically contrasted with an ‘intensional’ (or ‘hyperintensional’, as many philosophers call it) context, in which substitution of co-referential terms does not necessarily preserve truth value. For example, ‘Mary believed John hit his brother with the ball’ may or may not be substituted for ‘Mary believed John hit Jim with the ball’. If Mary believed that Jim was John's cousin, then substituting one sentence for the other would not preserve truth value. In drawing this contrast, and as his discussion proceeds, Marshall drops even his allusions to the mode of necessity in discussing extensional contexts. Thus: ‘The goal of the slingshot argument is to establish the truth of the general point, suggested by our example, that in expressions of the form “the sentence s corresponds to the fact that p,” the values of s and p can be any true sentences’ (Marshall (Reference Marshall2000), 228–229).

I'll return to this line of criticism later. For now, let's see how Marshall sets up his argument, which he structures as a reductio ad absurdum. It goes like this:

1. (1) ‘Northfield is east of Lonsdale’ corresponds to the fact that Northfield is east of Lonsdale.

   (Premise)

2. (2) ‘Northfield is east of Lonsdale’ corresponds to the fact that the town where St Olaf is located is east of Lonsdale.Footnote ¹²

   (PSST)

3. (3) â(a = a and Northfield is east of Lonsdale) = â(a = a). Footnote ¹³

   (PSST)

4. (4) Northfield is east of Lonsdale = (3).

   (PSLE)

5. (5) ‘Northfield is east of Lonsdale’ corresponds to the fact that â(a = a and Northfield is east of Lonsdale) = â(a = a).

   (PSLE)

6. (6) ‘Northfield is east of Lonsdale’ corresponds to the fact that â(a = a and Red Square is in Moscow) = â(a = a).

   (PSST)

7. (7) â(a = a and Red Square is in Moscow) = â(a = a).

   (PSST)

8. (8) Red Square is in Moscow = (7).

   (PSLE) . . . and finally

9. (9) ‘Northfield is east of Lonsdale’ corresponds to the fact that Red Square is in Moscow.

   (PSLE) (ibid., 229–230)

What conclusion does Marshall draw from this? Marshall says that the correspondence theory ‘is supposed to explain truth; . . . [to tell] us something about truth which we would not otherwise know’ (ibid., 230).Footnote ¹⁴ But, he says, as an explanation of truth for sentences, the correspondence theory is hopeless. It ‘tells us nothing useful we do not already know, and know better, from whatever grasp of “true” we already have’ (ibid., 231). On Davidson's (and my own) view, though, the correspondence theory is not designed to explain something about truth we don't already know. Rather, it articulates a pre-theoretical understanding of truth that realists about truth do indeed already have. But in addition, it creates a relationship between semantics and ontology.

Be that as it may, Marshall's gloomy prognosis for the correspondence theory presupposes the success of the slingshot argument. Yet, in its Davidsonian instantiation, I don't think it is successful, which is why few analytic philosophers have followed Davidson in using it to attack the correspondence theory.Footnote ¹⁵ There are a couple of obvious responses that on their own should make us suspicious about Davidson's and Marshall's use of the slingshot. First, if the slingshot works in this context, it would work not only on the sentential operator ‘. . . corresponds to the fact that . . .’, but also equally well on ‘. . . is true if and only if . . .’. That is, Marshall argues that the slingshot shows that ‘in expressions of the form “the sentence s corresponds to the fact that p,” the values of s and p can be any true sentences’ (Marshall (Reference Marshall2000), 228–229). If Marshall is right, then the slingshot would also show that in expressions of the form ‘the sentence s is true if and only if p’ (i.e. T-sentences), the values of s and p can be any true sentences as well. And if that's the case, then the slingshot could be thought to undermine not only the correspondence theory, but any theory at all about truth. Now this would not bother Davidson (though it might trouble Marshall), since Davidson is famously a primitivist about truth – he takes the concept of truth to be properly primitive, or basic, and not subject to definition, explanation, or explication at all (Burgess & Burgess (Reference Burgess and Burgess2011), 84–90). Thus, Davidson's use of the slingshot is closely related to his programme of truth-conditional semantics. Davidson famously maintains that the meaning of a sentence is constituted by its truth conditions. And if that's the case, then to know the meaning of a sentence is to know the sentence's truth conditions. But one can't know what any sentence's truth conditions are without possessing the concept of truth. Thus, Davidson says, the concept of truth is a primitive, or basic, concept (Davidson (Reference Davidson1996) ).

Though I don't have time to argue the point here, the trouble with the Davidsonian programme in semantics is that no truth-conditional theory of meaning has any hope of succeeding (see Knight (Reference Knight2008); Idem (Reference Knight2013), 229–234; see also Soames (Reference Soames1992), 26–29). That in itself should make us suspicious of Davidson's (and Marshall's) claim that the slingshot tells us anything about truth.Footnote ¹⁶ But we should also be suspicious of the slingshot's prospects of success in accomplishing what Davidson and Marshall want it to accomplish. One global reason is fairly straightforward: crudely stated, Marshall argues that if the slingshot works, then ‘Northfield is east of Lonsdale’ corresponds to the fact that Red Square is in Moscow. He takes it to work, and that therefore ‘Northfield is east of Lonsdale’ corresponds to the fact that Red Square is in Moscow, an obviously absurd result. And what this accomplishes, Marshall argues, is to demonstrate that correspondence theories tell us nothing about truth. But, to borrow Alvin Plantinga's quip, one person's modus ponens is another's modus tollens. Thus, I could agree that if the slingshot works, then ‘Northfield is east of Lonsdale’ corresponds to the fact that Red Square is in Moscow. But it should be obvious that ‘Northfield is east of Lonsdale’ does not correspond to the fact that Red Square is in Moscow. Therefore, the slingshot does not work. We could refine the argument somewhat by substituting ‘factual contexts are/are not extensional’ for ‘the slingshot works/does not work’. If factual contexts are not extensional, and neither Marshall nor Davidson provides any adequate reason to believe that they are, then the slingshot cannot undermine the correspondence theory.

A slightly more technical reason to believe that factual contexts are not extensional has to do with PSLE used by Marshall and Davidson. As I mentioned above, ‘two sentences are logically equivalent if and only if they are true in precisely the same models or possible worlds’ (Young (Reference Young2002), 124). Yet several statements in Marshall's argument are not logically equivalent in this way. And the reason is quite straightforward. Marshall says that ‘. . . corresponds to the fact that . . .’ creates an extensional context so that PSST and PSLE will apply. And to demonstrate this, he notes that ‘we may safely infer’ (2) above from (1), since Northfield is the town in which St Olaf College is located. But it's obviously a contingent fact that St Olaf is located in Northfield. There is any number of possible worlds in which Northfield is east of Lonsdale, but St Olaf is not located in Northfield. At the very least, then, we'd need more than inferring (2) from (1) in the actual world to establish that factual contexts are extensional.

Indeed, not many philosophers have endorsed Davidson's slingshot argument, and many are unwilling to allow Davidson's and Marshall's assumption that factual contexts are extensional.Footnote ¹⁷ Particularly dubious is the phrase ‘the a such that . . .’ or ‘the set a such that . . .’. Philosophers such as Nathan Salmon argue that these operators are non-extensional and that expressions within the scope of such operators are not amenable to substitutivity (see Salmon (Reference Salmon1986), 6; Idem (Reference Salmon, Anderson and Zelëny2001), 349; see generally Idem (Reference Salmon2005) ). Others object to Davidson's assumption that facts are extensions or can be treated as such (see e.g. Krüger (Reference Krüger1995) ). Still others contest the assumption that the correspondence relation is essentially a reference relation (e.g. McFarland (Reference McFarland2013), 430 n. 59).Footnote ¹⁸ Indeed, Davidson himself seemed to reject this assumption in his earlier use of the slingshot.Footnote ¹⁹ Finally, if one adopts a Russellian quantificational account of definite descriptions, rather than a referential account, PSST won't apply to such descriptions, and the slingshot won't work (see Brandl (Reference Brandl, Schurz and Dorn1991) and Read (Reference Read1993) ). One might argue that Davidson (and Marshall) could avoid this critique by simply adopting a referential account of descriptions. Perhaps so, but a correspondence theorist could also avoid the slingshot by adopting a Russellian account.Footnote ²⁰

Finally, contrary to his earlier use of the slingshot, in his argument against the correspondence theory, Davidson makes the substantive assumption that sentences refer to facts. In this, he departs from the classic slingshot arguments used by Frege, Church and Gödel. Their arguments advanced only a semantic thesis – namely, that sentences with the same truth values have the same referents. All true sentences refer to the true, while all false sentences refer to the false (Frege (Reference Frege, Quinton and Quinton1956) ). But these are both unique semantic entities and abstract completely from any metaphysical commitments. By switching the referents of sentences to facts, and then assuming that the context will remain extensional, Davidson smuggles a metaphysical conclusion into what is essentially a semantic argument. Thus, he is only able to draw metaphysical conclusions from the argument (viz., that there is only one Eleatic fact) by building into the argument precisely those assumptions he needs for the argument to work. Consequently, as Andrew McFarland argues, however interesting the semantic results of properly deployed slingshots, ‘they tell us very little (if anything) about metaphysics’ (McFarland (Reference McFarland2013), 432).

Marshall responds to only one objection to the slingshot, that of John Searle.Footnote ²¹ Searle argues that any set of logical constraints must take into account features of the meaning of ‘fact’ set out by the particular correspondence theory to which it is applied. And on Searle's account, a fact is that which makes a statement true or in virtue of which a statement is true. In other words, a fact is a truth-maker. In addition, any set of logical constraints must respect the intuitive features of ‘true’ and ‘correspond’. But the slingshot fails to do this, Searle argues, in applying principles such as PSLE. Another way to put the objection, Searle says, is that PSLE does not preserve the identity of facts. And it should be obvious that the self-identity of all objects has nothing to do with the fact that Northfield is east of Lonsdale.

Marshall's response is twofold. First, he says that an objection that blocks the inference from (1) to (5) would also block the inference from (1) to (2). In consequence, he says, Searle effectively treats the operator ‘. . . corresponds to the fact that . . .’ as extensional in (2) but intensional in (5) (Marshall (Reference Marshall2000), 231–232). But so far as I can see, Marshall is wrong about this. For the inference from (1) to (2) to be valid, we must assume a suppressed premise: namely, that Northfield is the town where St Olaf is located. And the reason we need this suppressed premise is that it's a contingent fact about Northfield that St Olaf is located there. But given this premise plus (1), we can infer (2) whether or not the context is extensional. But without the suppressed premise, we can't infer (2) from (1), and that in itself should make us suspicious of the notion that the correspondence operator creates an extensional context.

Marshall's second response challenges Searle's method of differentiating one fact from another. ‘The difficulty remains,’ he says, ‘of individuating facts or states of affairs and of specifying when they are identical’ (ibid., 232). The most obvious initial reply, he says, is to say that facts are differentiated by true statements. And he notes that Searle himself advocates this reply.Footnote ²² Marshall argues that when facts are individuated in this way, the correspondence theory fails as a strategy for explaining truth, because we must already understand the concept of truth in order to identify the fact that makes a statement true. For his part, Searle seems to recognize this but is untroubled by it: ‘Facts in short are not extralinguistic things, but facts already have the notions of statement and truth built into them, because in order to specify a fact we have to have a true statement’ (Searle (Reference Searle1995), 205). This construal of facts is contested, of course, but it indicates that Searle's correspondence theory is an articulation of our already (somewhat) understood, pre-theoretical notion of truth.

Marshall also objects to Searle's perceived violation of Occam's razor. Individuating facts this way leads to an infinitely multiplying Platonic menagerie of facts. He illustrates this with the following examples:

1. (10) Northfield is ten miles east of Lonsdale.

2. (11) Northfield is 16 kilometres east of Lonsdale.

3. (12) Lonsdale is 10 miles west of Northfield.

Now (10)–(12) all describe an aspect of the geographic relationship between Northfield and Lonsdale. Yet at least on Marshall's reading of Searle, they all describe different facts – too many facts for Marshall's taste. For every true statement, we have a separate corresponding fact, and in this situation, he says, ‘substitution which preserves identity of fact becomes impossible’ (Marshall (Reference Marshall2000), 232). Marshall takes this to imply that Searle's method for individuating facts is untenable as a way to distinguish which substitutions are acceptable and which are not, and thus we have ‘no clear idea what is wrong with the slingshot argument’ (ibid., 233). So far as I can see, though, Marshall's conclusion is too extreme. We have a very good idea about what is wrong with the slingshot argument, and I've mentioned several things wrong with it. The basic problem, though, which Searle and others have pointed out in different ways, is that we have no reason to assume that factual contexts are extensional, especially in the absence of any argument that they are. But beyond that, I think Marshall's dispute with Searle highlights a broader point. When the co-referential principle is applied to sentences (or propositions) rather than singular terms, it creates problems whose analysis is clarifying. And what it clarifies is that the dispute between realists and antirealists is not a difference over truth, but over metaphysics (see Soames (Reference Soames1999), 32–39). And indeed, here Marshall's dispute with Searle seems to be as much about the metaphysical status of facts as it is about truth.Footnote ²³

There is one final flaw in Marshall's use of the slingshot that I'd like to highlight. And this, so far as I can see, is a fatal flaw in Marshall's and Davidson's use of the slingshot, as well as any use that deploys the sentential operator ‘. . . corresponds to the fact that . . .’. That is not so much the move from (4) to (5) (which I do think problematic as well), but the move from (5) to (6). So to recall:

1. (5) ‘Northfield is east of Lonsdale’ corresponds to the fact that â(a = a and Northfield is east of Lonsdale) = â(a = a).

   (PSLE)

2. (6) ‘Northfield is east of Lonsdale’ corresponds to the fact that â(a = a and Red Square is in Moscow) = â(a = a).

   (PSST)

Let's let N = ‘Northfield is east of Lonsdale’ and R = ‘Red Square is in Moscow.’ Then we can restate the steps as follows:

1. (5) N corresponds to the fact that â(a = a &N) = â(a = a).

2. (6) N corresponds to the fact that â(a = a &R) = â(a = a).

Marshall takes (6) to follow from (5) as a result of PSST. That can only be the case if â(a = a &N) = â(a = a) and â(a = a &R) = â(a = a) are co-referential – in non-technical terms, if they are names for the same fact. Consequently, to defeat the slingshot, all a correspondence theorist needs to do is to find some criterion of fact identity according to which the two facts are not identical. And we don't need to search too far to find one. One quite plausible criterion is the existential criterion of fact identity set out by Kenneth Olson, according to which two facts are identical if and only if they are instantiated in exactly the same possible worlds (see Olson (Reference Olson1987) ). Now Olson's biconditional raises some potentially thorny problems, so let's modify the criterion such that existing in the same possible worlds is a necessary but not sufficient condition of fact identity.Footnote ²⁴ But since N and R are both contingent facts, there will be possible worlds in which N is instantiated but R is not, and vice versa. And this means in turn that there are possible worlds in which â(a = a &N) = â(a = a) is false but â(a = a &R) = â(a = a) is not. And therefore it's possible that in (5) N corresponds to some fact, while in (6) N corresponds to nothing. Consequently, substitution is not warranted, and the slingshot fails. As a result, I see no reason to accept Marshall's conclusion ‘that a view like Davidson's, which accepts realism's fall but not anti-realism's rise, is the most plausible outcome currently available of the long philosophical debate about what truth is’ (Marshall (Reference Marshall2000), 241).

Concluding remarks

These three theologians all remain suspicious of correspondence theories of truth. On my reading their suspicions arise from concerns distinct from the correspondence relation itself. For Long, insofar as correspondence theories abstract from any substantive metaphysical and epistemological commitments, they imply that metaphysics (including talk about God) is impossible and undesirable. Hector objects to them precisely insofar as they don't abstract from such commitments, because he believes metaphysics does violence to the objects it analyses and makes God a projection of ourselves. Marshall wants realism in truth theory to be dependent on Christology, taking as his clue one statement about truth in John's Gospel: ‘If Jesus Christ is the truth, then truth is borne, not only or chiefly by sentences and beliefs, but by a person. More than that: if the New Testament is right, then in the end truth is a person’ (ibid., 242). I don't have space in this article to scrutinize Marshall's theological disciplinary proposal – that will have to await another article. But let me simply suggest the anticipated direction of that article. Marshall argues that his disciplinary proposal is necessitated by two things: first, that some version of correspondence is needed for a theological understanding of truth; and, second, that philosophical versions of the correspondence theory have been crippled by Davidson's slingshot argument. I've tried to show in this article why I think the correspondence theory has withstood the slingshot attacks. If I'm right, this opens the door to a different kind of Trinitarian gloss on the notion of truth than that put forward by Marshall.

Rather than the kind of intellectual gymnastics required to defend the notion that the primary truth-bearer is Jesus, why not say that the primary truth-maker is the Trinitarian God? Certainly at first glance, this would seem as consistent with the enigmatic Johannine statements about truth as the notion that the primary truth-bearer is a person. As an added bonus, it would take us into a conversation about ontology, which is what we are doing when we are discussing a concept or doctrine of God. And out of that conversation about truth-makers, we can address disputes over differing accounts of facts or states of affairs (see generally Armstrong (Reference Armstrong2004), 48–49; Hochberg (Reference Hochberg, Lowe and Rami2009); Newman (Reference Newman2002), 140–171). I can't address those issues here. Instead, I've simply tried to show why I remain unpersuaded by the objections to correspondence theories of truth put forward by Long, Hector and Marshall.

Yet to say that I'm not persuaded by such objections is not to defend any particular correspondence theory, and especially not any and all of them. So I shall close simply by pointing in the direction of the kind of account I would hope to articulate and defend in future work. I would begin with the basic truth-maker thesis:

This thesis can be understood as a basic, minimum articulation of realism in truth theory. It will be compatible with some correspondence theories and not with others. But generally correspondence theories will involve three components – propositions (statements/beliefs), facts, and a correspondence relation. To defend any account of each of these components will involve answering a number of objections, which I don't have space to attempt here. But it seems most promising to me to adopt a rather minimalist version of each component. Thus a proposition is simply the content of an assertion or propositional attitude. This is our basic intuition about propositions, though it is not an explanation. I would not take them as having any mode of being independent of illocutionary acts or propositional attitudes, or, as William Alston would say, ‘independent of their content-bearing involvements’ (Alston (Reference Alston1996), 19).

Regarding facts, Andrew Newman says that generally speaking there are two approaches to facts – compositional views (such as those of Russell, the early Wittgenstein and Armstrong) and linguistic views (more like Austin's) (Newman (Reference Newman2002), 140). A linguistic view, Newman says, ‘starts with certain linguistic expressions and singles out some as describing states of affairs, states of affairs being introduced merely as what those expressions describe, and then facts are explained as states of affairs that obtain’ (ibid., 154). My own view is closer to the linguistic approach. To put it more colloquially, a fact is simply a slice of the actual world identified under a description. It seems to me that a fact bears a mereological relation to the actual world similar to the relation that my left leg bears to me. Yet this linguistic view of facts doesn't imply that they are not real, but only that the identification of a fact does not on its own increase the inhabitants of the world.

And finally, the correspondence relation can be articulated in a variety of ways, but, to use George Pitcher's distinction (Pitcher (Reference Pitcher and Pitcher1964), 9–14), very often it is done in either of two ways. The first is ‘correspondence-as-correlation’, in which facts and propositions are correlated according to some rule or function. The second is ‘correspondence-as-congruity’, in which facts and propositions correspond just in case they have the same components and these components are related to each other in the same or sufficiently similar way, i.e. they have the same or similar structure (David (Reference David, Lowe and Rami2009), 139–143). My own preference would be to abstract from these for purposes of an account of truth, as William Alston has done. For Alston, all that is needed for a minimal but sufficient account of correspondence is:

Such an understanding of correspondence may seem trivially obvious and uninformative, and it certainly doesn't tell us all we might want to know about correspondence. But it does play an important role. First, in requiring that the same substitution be made for ‘p’ on both sides of the biconditional guarantees that however one construes the correspondence relation, it will be truth-preserving. Thus, ‘it is enough to assure us that truth and correspondence are tightly connected’ (ibid., 39). Second, although it is neutral regarding competing metaphysical accounts of facts, it is compatible with those views that regard them as real parts of ‘the extralinguistic, extraintentional world’ (ibid.).

All this requires further articulation, of course. But I would envision a rather minimal correspondence theory of truth that is metaphysically neutral on a variety of issues, though not all issues. That neutrality should satisfy Marshall's objection to Searle's view of facts. It leaves theologians free to articulate a theistic ontology unconstrained (or less constrained) by the requirements of a more metaphysically involved truth theory. A correspondence theory, even a minimal one, identifies a truth-maker for true propositions. Conflicts can arise out of the thought that there is only one truth-maker (a particular fact) for any true proposition. But truth-maker theories generally argue that there may be a number of truth-makers for any true proposition. Thus they are complementary, so that truth-maker theory connects a realist ontology with truth theory. And this relation between minimal correspondence theories and truth-maker theories should alleviate Long's concerns that truth theories with linguistic units as truth-bearers surreptitiously eliminate the possibility of a realist ontology. On the contrary, truth-maker theory is explicitly concerned to preserve metaphysical realism. Furthermore, a minimalist (and non-isomorphic) correspondence theory should alleviate Hector's concerns (as he acknowledges) with conceptual limitations on the concept of God. Finally, a metaphysically neutral understanding of truth should satisfy Marshall's concern that a doctrine of God be articulated without methodological constraints. And for any doctrine of God that views a Trinitarian God as the creator of all that is, since true propositions may have more than one truth-maker, a truth-maker theory leaves open the possibility that such a God can be construed as the Maker of all truth.