Rich Davis has responded to a piece written by Guillaume Bignon and myself. I would like to thank him for taking the time to interact substantially with our initial response. In this piece, I am going begin with a few points of clarification on the arguments that Bignon and I give. The clarifications will be important to understand what moves I am making in the final section, which ends on matters more philosophically substantial. As a result of this discussion, I hope to make it more clear what challenges exist for Davis’ case against Calvinism. It is now impossible for anyone to follow this discussion without having read the pieces linked above. If you have not yet done so, do that first. But in case you do not, a main point of dispute regards a principle called TRIVIAL, which will be mentioned throughout this piece.
(TRIVIAL) For any x, if x is elect, then x will not perish and x will have eternal life.
Bignon and I discuss whether Jesus can assert an analytic statement because we took Davis to be denying that it would be befitting for Jesus to assert one. We thought, apparently incorrectly, that Davis held for Jesus to perform a speech act involving a trivial – analytic – statement is for Jesus to make some kind of a performative speech error. “To put (3:16b) into the mouth of Jesus either has him teaching falsehoods or empty tautologies, neither of which is befitting to the Savior.” Davis agrees with us and so there is no need to continue on this point.
The argumentative strategy Bignon and I employ is the same for each of the examples we raise – i.e., the Pharisee-Sadducee dispute, whether the Holy Spirit could become incarnate, whether Daniel Speak is right. Apparently, it was not clear what we were up to in our discussion of whether TRIVIAL is analytic. I will clarify that strategy now.
Although we suggest this only after the first example, our strategy is to give defeasible evidence for denying that a statement is analytic. If seeing that a statement is true requires much argumentation despite understanding the meaning of the terms involved, then that is defeasible evidence against its analyticity. That is plausible because if one could see the truth of a statement just by virtue of understanding the meanings of the terms in a sentence, then there is no reason to have argumentation for the proposition expressed by that sentence. We deny that this method is a proof of syntheticity because we are not taking this method to amount to a definition or criterion. When it comes to TRIVIAL, all that we needed to point out is that seeing the truth of TRIVIAL depends on more substantial matters, such as any of the three disputes highlighted. So when we note the “speculative” (our word) example about the Holy Spirit becoming incarnate, we are not suggesting that the Holy Spirit could have done this. And when we appeal to Speak’s paper, we are not saying that what he says is true, or even possibly (i.e. metaphysically possibly) true. We only point to the fact that there is a substantial dispute (that explains why we left Speak’s paper only with a link). Those are substantial matters for which arguments are needed, they are not settled by definition, and since so needed gives defeasible evidence against TRIVIAL’s analyticity.
This is similar, although in one important way different, to a move that Benson Mates raises in response to Quine and White in “Analytic Sentences” (1951). They rejected the analytic-synthetic distinction. In response, Mates does not take on the task of arguing for a definition of analyticity, as if Quine and White failed to consider an alternative definition. Instead, Mates argues that we have good reason to believe that there is such a thing as analyticity because there is much agreement about which sentences are analytic; of course he is not claiming there is agreement is about all sentences. Convergence of opinion about a statement’s being analytic is defeasible evidence of its so being analytic. For Bignon and myself, divergence of opinion about truth is evidence against analyticity. The analogy to Mates’ move could be tighter had we argued that disagreement about a statement’s being analytic is evidence against its analyticity, but we do not argue in that way. I am do not endorse that tighter route.
What exactly is analyticity? What does it mean to say that a sentence is analytic? More importantly, upon which analysis of analyticity does Davis’ argument depend? Upon further reflection about Davis’ argument and the response Bignon and I give, it now seems to me that this has not been clear. Probably, that should not have been surprising because the very notion of analyticity is not clear. From Kant up through the logical positivists, the notion of analyticity has often been a mix of semantic, metaphysical, and epistemological ideas.
This confusion is evident, I think, in Davis’ formulation of the argument and in our response. For example, at one point Davis characterizes analytic truths as truths by definition, and he says that they “tell us nothing that we did not already know by definition.” Analyticity, then, is very strongly epistemic; analytic truths are not informative. They are already known. So analytic truths are not news. And so the gospel according to Calvinism is not good news. (Actually, the topic is more complicated than this lets on. One famous proponent of the distinction, A. J. Ayer, thought that analytic statements could be surprising and he restricts analyticity to sentences that give no empirical information.) In response to this, Bignon and I assumed the account of analyticity underlying the above thought and then argued by the method in clarification 2 above that TRIVIAL is not analytic. However, notice what Davis asks of us when we deny that TRIVIAL is analytic. He asks if we mean to assert (1) or (2):
(1) The proposition The elect will have eternal life could have been false;
(2) The persons who (as things in fact stand) are elect could have been non-elect?
However, the relevant concept at work in (1) and (2) is modality, an arguably metaphysical concept. Why must denying a sentence to be analytic imply either (1) or (2) if analyticity is at least partly epistemic?
Another example. Davis writes in his response to us that “it is not as if analytic truths are analytic for some but not others.” Alright; so perhaps the relevant notion of analyticity is such that a sentence is analytic iff the statement is true in virtue of its meaning in all possible worlds. Kripke stipulates this definition in Naming and Necessity. Given this view, whether an analytic sentence is informative or not is irrelevant. Informativeness is at least a two-part relation between a proposition’s content and an epistemic agent. So it is also irrelevant whether the sentence can come to be known just by understanding the meaning of the terms. But then the whole point about whether the Gospel is news, that is to say, informative, is irrelevant.
It seems to me, then, that there are different notions of analyticity at work in Davis’ argument. The one that really matters, as far as I can tell, is his strongly epistemic one. How can the debate move forward regarding the points pertaining to TRIVIAL’s analyticity?
(a) Davis should tell us which notion of analyticity he assumes. He probably should argue that this notion of analyticity is the one we should accept as well. As has already been suggested here, there are different notions of analyticity in the history of the philosophy, and not everyone assumes the same one. Some Calvinists, such as Greg Bahnsen, have been quite impressed with Quine’s argument against the analytic-synthetic distinction. Unlike Bahnsen, Bignon and I did not dispute that there is some such notion of analyticity. In order to support more than a conditional, viz., given such and such view of analyticity, Calvinism makes the Gospel tautologous, Davis will need to argue for that notion of analyticity. I think this is too difficult a task.
(b) There should be an argument that TRIVIAL is analytic, whatever the right account of analyticity is. I will close on this.
Bignon and I took Davis to be assuming that analyticity was partly epistemic. I do not see how else his argument can possibly yield the conclusion that Calvinism fails to make the gospel good news. So why think that TRIVIAL is analytic? Davis thinks he is on safe ground to assert that TRIVIAL is analytic (given Calvinism). He writes in his response to Bignon and I:
Looking at my original post (see here), I think we can safely conclude that (on Calvinist premises) I take the concepts not perishing and having eternal life to be included in the concept of being elect. That is to say, you can impute to me the view that the proposition anyone who is elect has eternal life is an analytic truth on Calvinism. For all those who are elect (on the system) receive irresistible grace, and thus cannot help believing in Jesus (or perhaps, on B&G’s startling proposal, the incarnate Holy Spirit who dies for our sins). On either scheme of salvation, the elect won’t perish but have eternal life.
I do not understand this argument. It is well-known that the idea of one concept being “contained in” or “included in” another is only a metaphor, and it is not clear what that means. Frege questions Kant’s use of the containment metaphor, and I don’t think Davis should have a pass on this either.
Perhaps what follows the “For…” is more helpful. So “everyone who is elect will have eternal life” is analytic because all those who are elect (on the system) receive irresistible grace… and so on. I think Davis intends the qualifiers, “on Calvinism”, “on the system,” and “on Calvinist premises” to be doing a lot of work here. But it is not entirely clear to me what work that is. Why must TRIVIAL be analytic on Calvinism? After all, Bignon and I are two Calvinists explicitly rejecting that is so.
Maybe the thought is this: general theories give meanings to terms. A term’s meaning is something like the functional role it plays in a theory. Calvinism is a theory that uses the term, “elect,” and as that term is used by the theory it is a settled matter that everyone who is elect has eternal life. So the meaning of the very term, “elect,” on Calvinism, “includes” having eternal life. If that were Davis’ thought and I don’t know that it is, I should point out that this is very Quinean and not at all what the adherents of the analytic-synthetic distinction would wish to say. For another thing, this relies on a deeply controversial theory of meaning that has a long dispute between internalists and externalists about semantic content. If Davis’ argument is supposed to depend on this sort of move, that’s pretty bold and I doubt he would want it to depend on as controversial a view as that. And even if he did, it would at most show a conditional conclusion: given such and such view of meaning and analyticity, Calvinism makes the Gospel tautologous. Most Calvinists, I think, would not even blink at a conclusion that weak.
So I don’t understand his argument that TRIVIAL is analytic. But given that analyticity (whatever that is) is partly epistemic, Bignon and I are clear about what evidence would justify the denial of statement’s analyticity.
Let me close with a related but different point.
Consider the following:
(3) On modern chemistry, water is H2O.
In fact, (3) is false because modern chemistry does not say that; but this is a classic example used repeatedly and for that reason I am using it. Is what follows the prefix in (3) an analytic truth? No; it is a classic a posteriori synthetic necessity. It does not follow that “water is H2O” is analytic even if (3) were analytic. If (3) were analytic, all that would tell us is what modern chemistry says about water. It would not result in seeing the truth that water is H2O just by virtue of understanding those terms of (3): “modern chemistry”, “water” and “H2O”. At best, one would understand a conditional: that “water” means “H2O” according to modern chemistry. Or worse, modern chemistry says nothing about the meaning of “water” but instead picks out which properties in the world (i.e. which properties are referred to) by the use of the word.
(4) On Calvinism, the elect are persons who have eternal life.
Even if (4) were analytic, it is a non-sequitur that what follows the prefix is analytic under the notion of analyticity being assumed. If Davis’ argument for the analyticity of TRIVIAL depends upon the prefix being in (4), at best he has identified a conditional, and no one thinks Jesus is asserting a conditional in John 3. And so no one, Calvinist or not, is forced to think that Jesus is saying something that even a Calvinist must regard as uninformative in John 3.
I think Davis might resist this by two possible moves.
Davis might insist that
(5) According to the English language, all bachelors are males.
is closer to (4) than (3). That is because what follows the prefix in 3 is an analytic statement. But that would be blatantly question-begging and would require an argument.
The second move I think Davis could make is to point out that the meaning of terms evolve over time. “Madagascar” is one of the examples raised by Gareth Evans and Saul Kripke. On this thought, although “elect” may have meant “persons chosen by God to be saved” in some vague sense of “saved” (political salvation from Rome, perhaps, or eternal life?) during the life of Jesus, the meaning of “elect” has evolved within Calvinist circles to include having eternal life.
Response. Assume that is part of the very definition of “elect” according to Calvinism, although I reject that this is a truth by definition. Then a Calvinist can understand the conditional on Calvinism, the elect have eternal life. It is yet a further question of whether their so understanding the meanings of the terms involved is sufficient for them to see that the elect have eternal life is true. After all, it is one thing to say that “having eternal life” is part of the definition of “elect”. It another thing altogether to think that one should accept this definition. What is required to make the elect have eternal life even putatively uninformative is an argument that Calvinism is true. But then seeing that the elect have eternal life is true is no longer a function of the meaning of the terms alone, i.e. the terms following the prefix. And that is yet a further reason, consistent with the strategy outlined above, for denying that TRIVIAL is analytic.
James A. Gibson