The first and basic argument Howard-Snyder gives, and then presupposes throughout the paper, is this:
Panmetaphoricism possesses an unenviable property: if it is true, then it is false. For if our speech about God can only be metaphorical, then the predicate “can be talked about by us only metaphorically” applies to God literally. But in that case, our talk about God cannot only be metaphorical, contrary to panmetaphoricism. Panmetaphoricism is self-refuting.
Unfortunately this is much too fast, and, indeed, it becomes clear when one considers the peculiarities of this predicate; Howard-Snyder begins to consider this point, but does not press it hard enough. If I say that something 'applies to X literally', the only thing I can mean is, 'when taken literally applies to X'. But the panmetaphoricist simply does not have to take the predicate in question literally; the panmetaphoricist doesn't -- and indeed cannot -- think that talking about God is talking about God in the sense we take the phrase when taking it literally; so all our talking about God is talking about God involving at least some kind of figurative element. Howard-Snyder tries to block this move with a further argument:
She says that no predicate of ours can apply literally to God. When we remind her of the predicate “cannot be talked about literally by us,” she replies, “And that one doesn’t either”. But if that’s the case, there must be something about God in virtue of which no predicate of ours can apply literally to God, not even the predicate “cannot be talked about literally by us”. It isn’t just magic, or an inexplicable brute fact. But then we can introduce a new predicate into our lexicon—say, “is illiterable”—and we can stipulate that it signifies literally whatever that something is, from which it follows that some predicate of ours can apply literally to God after all.
This, however, is simply question-begging; if we stipulate that 'is illiterable' applies when taken literally, we are doing nothing other than stipulating the success of a kind of language the panmetaphoricist denies. The panmetaphoricist will say that 'is illiterable' applies when taken figuratively; it cannot apply when taken literally. Obviously it is true that if you stipulate that panmetaphoricism is wrong then it will follow that panmetaphoricism is wrong. But this is not a refutation. What Howard-Snyder needs, and does not give, is an argument for why the panmetaphoricist herself cannot say that there is at least some metaphoricity, some figurative aspect to such predicates.
And the seriousness of this lapse becomes worse when we consider more closely the possible panmetaphoricist response that Howard-Snyder himself considers. He imagines the panmetaphoricist saying that first-order speech about God is only metaphorical, leaving open the status of second-order speech (language about our language about God). But the fact that we are able to make any such distinction intelligible at all shows the problem with Howard-Snyder's argument: the kinds of predicate his arguments use (like 'can be talked about by us only metaphorically') involve transfers from one domain to another. Speech about God and speech about speech about God are not talking about God in the same sense of 'talking about God'. And all cross-domain transfer is metaphor in the broad sense. Far from establishing that panmetaphoricism is self-refuting, Howard-Snyder's predicates make it more secure.
The problem appears very much to be a bad theory of metaphor. We see this in another of his arguments:
One concern is this: according to Abrahamic religion, God exists, really exists. However, if our first-order speech about God can only be metaphorical—as our panmetaphoricist insists—then no first-order speech of ours can be used literally of God, including the predicate “exists”. But if the predicate “exists” cannot be used of God literally, then there is nothing about God in virtue of which the predicate “exists” can apply to him literally. And if there is nothing about God in virtue of which the predicate “exists” can apply to him literally, then the statement “God exists, really exists” is false, which is to say that God does not exist, not really.
But this is simply false. If there is nothing about God in virtue of which the predicate 'exists' can apply to him literally, all that follows is that the statement 'God exists, really exists' is false when taken strictly literally. But the panmetaphoricist has every reason to accept this. She will simply say that God does exist, does really exist, if we take those phrases metaphorically. Howard-Snyder does consider this, but argues:
If our panmetaphoricist replies that she means for her use of these predicates to be merely metaphorical, she will fail to solve the problem for which she invoked them. For if they don’t apply to God literally, and if “exists” and the like don’t either, God won’t show up anywhere on the ontological map, not as an existent or a non-existent object, not as a denizen of reality or unreality—which, on the terms of the ontology she invokes, is incoherent.
But this again is simply false; the panmetaphoricist will obviously reply that they do show up on the ontological map in such a way that only metaphorical expressions can describe them. Howard-Snyder is assuming that reality -- and thus ontology -- is literal. But neither reality nor ontology have any intrinsic connection to literalness; literalness is a matter of the language we use, not reality or ontology. We can talk about reality in expressions to be taken literally, but we can also talk about reality in expressions to be taken metaphorically. Our ontological map can be drawn using expressions that are literal or expressions that are figurative. The panmetaphoricist is saying that God does show up 'on the ontological map' -- but that there are parts of the ontological map that can only be drawn with metaphors and God shows up there. Or, indeed, the panmetaphoricist may say that there are really several different ontological maps that human beings lack the means to reconcile into a single map; and God is on an ontological map only metaphorical expressions are robust enough to trace. 'Really' and 'literally' are not synonyms; nor do they mutually apply each other. They don't even belong to the same domain.
The more disastrous issue is that Howard-Snyder is begging the question in every step of the argument, because subject terms are talk about things, too. Consider an analogy. Suppose I were to say, "Colloquial, conversational English can only talk about black holes metaphorically." Is this self-refuting in the way Howard-Snyder had suggested? No. If we use the predicate 'can only be talked about metaphorically' of something we are already metaphorically calling 'black holes', we are still only talking about it metaphorically. The same is true if we say, 'black holes are illiterable'. The same is true if we say 'black holes exist'. In order for Howard-Snyder's argument to work, he must first assume that 'God' in 'God can only be talked about metaphorically' involves no implicit metaphor -- that is, he must first assume that panmetaphoricism is false. (But even if we set this aside, we run into the cross-domain transfer problem noted above, once the panmetaphoricist makes a distinction between first-order and second-order language.) And the same is true of 'God exists'.
But suppose we even ignore this. Consider our analogy again, ignoring the fact that the subject term is a metaphor. Is there anything inconsistent about saying that 'Black holes really exist' is true only if we take the predicate to apply metaphorically? There is not; whether or not it is true, it is entirely possible to have a version of English in which the words 'really exist' only apply literally to things very different from black holes. Does this claim commit us to saying that black holes are not on our 'ontological map'? It does not. At the very least, Howard-Snyder needs an argument to show that it does, and he has provided no such argument. He is merely assuming it. And in the same way, if we say, "Human languages can only talk about God metaphorically," we seem neither to have any inconsistency, nor to have any problem with saying that God exists -- if understood in the right way.
We see the same problem later when Howard-Snyder considers a version of panmetaphoricism using Thomistic language:
if , as my friends insist, his doctrines of analogical predication and divine simplicity imply that the predicate “is personal” can only be predicated analogically of God and humans, and if, as my friends insist, that implication itself implies that the predicate “is personal” cannot apply literally to God, then God is not personal, not really.
Howard-Snyder's friends are simply wrong about Aquinas, but it doesn't really matter, anyway, since Howard-Snyder is simply not justified in drawing th econclusion. If the predicate 'is personal' cannot apply literally to God, but it does apply to God, then God is personal, really, when 'is personal' is not taken literally but in the appropriate figurative way, in the same way that when we rightly say that 'Such and such astronomical phenomenon is a black hole', the astronomical phenomenon is a black hole, really, as long as we take 'is a black hole' in the appropriate figurative way.
And, indeed, this is the whole problem of Howard-Snyder's argument: it involves a shoddy understanding of the distinction between the literal and the figurative that lets Howard-Snyder slide back and forth between between treating the distinction as linguistic and treating it as somehow ontological. This is quite easily visible in Howard-Snyder's continual sliding between 'literally' and 'really'. This is simply untenable, and cannot be seriously maintained after a moment's thought. Howard-Snyder's argument, however, depends crucially on it: it repeatedly comes up, and can't be eliminated from the argument without eliminating the argument.
Of course, the panmetaphoricist view fails, but for precisely the reason Howard-Snyder's argument against it fails: it's based on a false view of the literal/figurative distinction. But that's another argument.