I posed the following problem:
A. Some sentences are true in virtue of their correspondence with extralinguistic reality.
B. If so, then reality must have a sentence-like structure.
C. Reality does not have a sentence-like structure.
London Ed solves it by rejecting (A). But let me first say why I accept (A).
Consider a true contingent sentence such as 'Tom is sad,' or the proposition expressed by an assertive utterance in appropriate circumstances of such a sentence. I maintain that the sentence or proposition cannot just be true: if true it is true in virtue of something external to the sentence. The external something cannot be another sentence, or, more generally, another truthbearer. Nor can it be someone's say-so. So the external something has to be something 'in the world,' i.e., in the realm of primary reference, as opposed to the realm of sense. The basic idea here is that some truths need ontological grounds: there is a deep connection between truth and being. There is more to a true sentence than the sentence that is true. There is that in the world which makes it true. Call it the truthmaker of the truth. Some truthbearers need truthmakers. As far as I am concerned, this is about as clear as it gets in philosophy. Which type of entity is best suited to play the truthmaker role, however, is a further question.
Please note three things. First, the direction of the truthmaking relation is from the world to language. More broadly: from external concrete reality to the realm of representations, where Fregean propositions count as representations, despite their not being tied to specific languages, and despite their independence of minds. Second, correspondence is an umbrella notion that covers two quite different relations, naming, and making-true. Naming is a word --> world relation, whereas truthmaking goes in the opposite direction. I am tempted to say that truthmaking is the converse of naming. Third, I unpack 'correspondence' as it occurs in (A) in terms of truthmaking, not naming.
Here is what Ed says in rejection of (A):
The exam question is my argument against (A), namely that some sentences are true in virtue of their correspondence with extralinguistic reality. I shall also be taking on why my reasons are properly nominalistic, given that your version of nominalism is not mine.
1. Starting with nominalism. Classic nominalism is formulated by Ockham in Summa Part I, 51. “the root [of the error of the Realists] is to multiply entities according to the multiplicity of terms, and [to suppose] that to every term [or expression] whatsoever there corresponds a thing [quid rei].”
2. My target is a formulation of the correspondence theory that violates classic nominalism, as I have defined it. There may be other formulations of the theory that are OK.
3. My formulation of the correspondence theory is that an assertoric sentence is true in virtue of naming or referring to or signifying a fact. Let that naming relation be R. Then the correspondence theory says that a sentence S (e.g. ‘Socrates is sitting’) is true iff S stands in the relation R to some fact F (e.g. ‘that Socrates is sitting’).
4. Suppose ‘Socrates is sitting’ names the fact that Socrates is sitting, and assume that it always so names. Then that fact must always exist, assuming the name is always names the fact. So ‘Socrates is sitting’ must always be true, i.e. ‘Socrates is sitting’ always stands in the relation R to the fact that Socrates is sitting. But it is not always true, clearly.
5. We might get out of this in two ways. First, by supposing that ‘Socrates is sitting’ fails to be meaningful, namely when the fact it purports to names ceases to exist, such as when Socrates stands up, or runs. This is absurd, however. The purpose of a sentence is always to mean something.
6. The other way is to suppose that the sentence sometimes names a fact, and sometimes does not. I.e. it actually names something else – a proposition – and the proposition is a fact when the sentence is true, otherwise not a fact. However we have now failed to explain the ‘correspondence’. The sentence ‘Socrates is sitting’ always bears the naming relation R to the proposition that Socrates is sitting, even when Socrates is not sitting.
7. What we really need to name is not the proposition (which may be true or false), but the reality that corresponds when the proposition is a fact. Perhaps ‘the proposition that Socrates is sitting being a fact’ or ‘the actuality of Socrates’s sitting’ or something like that. But there we have the same problem. Either the name ceases to be meaningful when Socrates is not sitting, or it continues to name something. But the former we agreed was absurd, and the latter means that we have not fully captured the relation we want.
8. The problem in general is that if the object of the relation R is something we can talk about i.e. name at all, then we have to deal with the problem of the fixity of reference. The purpose of a name is always to name what it names. But reality is not thus fixed. Whatever supposedly corresponds to the truth of ‘Socrates is sitting’ comes into existence when Socrates sits and goes out of existence when he stands up. But if ‘Socrates is sitting’ is true in virtue of naming this thing, either the sentence becomes meaningless when Socrates stands up, which is absurd, or it names something that does not go out of existence, and so does not name what the correspondence theory purports to name.
9. Bringing this back to nominalism. The problem above arises from the supposition that ‘Socrates is sitting’ is the name of some fact, and thus from supposing that every expression (‘Socrates is sitting’) has a name or referent or whatever.
Ed does two things above. He confronts the truthmaker theorist with a certain (supposedly insoluble) problem, and then he explains how this problem arises by way of a false assumption. First, the problem. I will summarize it as I understand it.
Since Socrates is a past individual, but nothing in this discussion has to do with time, I will change the example to 'Tom is red.' Tom is a tomato of my present acquaintance. We assume that the sentence is true. And of course, if true, then contingently true. My type of TM-theorist holds that contingent true predications such as 'Tom is red' have worldly correspondents called facts. These concrete facts are the truthmakers of contingent predications. Note that the fact corresponds to the sentence as a whole. So not only does 'Tom' have a worldly correspondent, and presumably also the predicate 'red'; the sentence has a worldly correspondent as well.
Note also that the sentence is not about the fact; it is about Tom, or, if you insist, it is about Tom and the property of being red. Still, there is some relation R that connects the sentence and/or the proposition it expresses and the fact. Notice, I wrote 'and the fact,' not 'to the fact.' 'To the fact' suggests a direction from language to world, and not vice versa, whereas 'and the fact' leaves the directionality open. Is the truthmaking relation R naming? Ed thinks it is, but this is not clear. Indeed, I will argue in a moment that the truthmaking relation is not the naming relation. It is clear that 'Tom' names Tom. It is not clear that 'Tom is red' names anything. Suppose it doesn't. This doesn't exclude the possibility that the sentence has a truthmaker. Maybe it has a truthmaker, but that truthmakers cannot be named. Note also that what Ed says above is nothing like what any TM-theorist has ever said. Truthmaking is a relation that runs from the world to representations, whereas naming and referring and 'signifying' run from representations to the world. Truthmaking is more like the converse of the naming relation. We shall see.
But let us suppose arguendo that the truthmaking relation R is naming. On this supposition, Ed sets up a clever little dilemma. It is based on three plausible theses.
T1. If N is a name, then N cannot be vacuous: it must have a nominatum or referent.
T2. If N is a name, then it has an existing referent. That is, there is no naming of nonexistent objects, pace Meinong.
T3. If a name N names an object O, then at every time at which N names something, it names O. So the following is impossible: at some times at which 'Kripke' is in use as a name, it names Kripke, at other times Shkripke. I think this is what Ed means by "the fixity of reference."
The Dilemma. Either sentence S names fact F or it doesn't. On either alternative, trouble. Remember, Ed is assuming that the truthmaking relation is a naming relation and that declarative sentences name facts.
Horn One. If S names F, then, by the conjunction of the three plausible theses, F exists at every time at which S exists, which is plainly false. Clearly, 'Tom is red' both as type and as token can exist at times at which the fact of Tom's being red does not exist. (I might assertively utter 'Tom is red' while Tom is green, or after Tom has been dunked into molten chocolate.) If you say instead that S is meaningless when the fact does not exist, then truthmaking is not naming (by T1), which is all it can be on Ed's (mis)understanding of truthmaking.
Horn Two. If S does not name F, then there is no truthmaking. For truthmaking is a naming relation.
It is clear that Ed does not understand truthmaker theory. The key idea is not that sentences name facts, but that facts make sentences true. That truthmaking is different from naming is clear from the different directions of the relations, but also because truthmaking is a many-many relation whereas naming is a many-one relation. That truthmaking is many-many can be seen as follows. One and the same truth can have many different truthmakers. For example, 'Something is red' is made true by a's being red, b's being red, c's being red, etc. And one and the same truthmaker can make true many truths. For example, Tom's being red makes true 'Tom is red,' 'Tom is red or Shlomo is sad,' etc. (Cf. Armstrong 1997, pp. 129-130.)
Ed has an understanding of nominalism which contemporary analytic philosophers will find idiosyncratic and vacuous to boot. No philosopher today thinks that for every bit of language there is a corresponding bit of reality. So we are all nominalists in this vacuous sense. And no one is a realist if “the root [of the error of the Realists] is to multiply entities according to the multiplicity of terms, and [to suppose] that to every term [or expression] whatsoever there corresponds a thing [quid rei].” And surely it is a bad joke to claim or suggest that TM-theorists straightaway infer the existence of facts from the existence of declarative sentences.