Preliminaries

First: a word about the literal–metaphorical distinction. The literal–non-literal distinction marks different ways in which terms (words or phrases) in a language can be used. It falls on the speech side of the language–speech divide. When one uses a term in accordance with one of the standard meanings established within a language (to which we might add unusual lexical definitions as well), or when one uses a term in accordance with an explicitly stipulated meaning, one uses it literally. When one does not use a term in accordance with one of its standard or stipulated meanings, one uses it non-literally. One way to use a term non-literally is to use it metaphorically.

Second, some panmetaphoricists use the term ‘metaphor’ loosely, not in contrast with simile, parable, symbol, and other forms of figurative speech but rather in contrast with the literal use of language. You might say they use ‘metaphor’ metaphorically. However, whether or not panmetaphoricists agree to use ‘metaphor’ metaphorically, they all agree that an implication of their view is that none of our speech about God can be literal. If our speech about God can only be metaphorical, then none of it can be literal. I agree that this is an implication of their view. This implication is important to them; it is likewise important to me.

Third, when panmetaphoricists say that we cannot speak literally of God, they do not mean to imply that we cannot form a subject–predicate sentence with ‘God’ as the subject term and make a literal use of the predicate while intending to utter a truth. That's easy. Rather, what they mean to imply is that no such intention can succeed. No literal use of a predicate in relation to God can successfully result in a true utterance or, as I will say, no predicate applies literally to God.Footnote ¹⁰

Fourth, if a predicate applies to something literally, then there is something about it in virtue of which that predicate applies to it literally. Since I prefer an ontology according to which there are things and their properties, in what follows I will often speak as though a predicate applies to something literally (at least partly) in virtue of that predicate being associated with some property or complex of properties that the thing has. I want to stress, however, that although several of my arguments presuppose my preferred ontology, I will typically also express those very arguments without that presupposition by reverting to the much less contentious general idea that if a predicate applies to something literally, then there is something about it in virtue of which it does so. That can be the case even if there are no properties at all.

Fifth, according to panmetaphoricism, our speech about God can only be metaphorical. Note that it is our speech, ‘everything we assert’ (Keshgegian), ‘our concepts’ (Kaufman), ‘all language’ (Jefferts-Schori), presumably all of our language. Questions abound. Is it only actual human speech to which panmetaphoricists refer? Or do they mean to refer to merely possible human speech as well? What about the speech of actual non-human creatures, if there are any? What about the merely possible speech of merely possible non-human creatures? Answers to these questions may well vary depending on the reasons given for panmetaphoricism. In what follows, I will think of panmetaphoricism as a thesis about all possible creaturely speech, assertion, concepts, and language.

Panmetaphoricism is self-refuting

As it stands, panmetaphoricism seems to possess an unenviable property: if it is true, then it is false. For if speech about God can only be metaphorical, then the predicate ‘can be talked about only metaphorically’ applies to God literally, in which case it is false that speech about God can only be metaphorical. Panmetaphoricism is self-refuting.

The strictly consistent panmetaphoricist will deny the premise that her view entails that the predicate ‘can be talked about only metaphorically’ applies to God literally. After all, she will insist, since talk about God can only be metaphorical, then every predicate can only apply metaphorically to God, including the predicate ‘can be talked about only metaphorically’. What should we make of the strictly consistent panmetaphoricist?

Consider an analogy. Marcus Borg says that God is ineffable, by which he means that no concept applies to God. When we remind him of the concept of ineffability, he replies: that one doesn't either.Footnote ¹¹ In Borg's case, strict consistency results in contradiction. For if God is ineffable, then no concept applies to God; and if no concept applies to God, then the concept of ineffability doesn't either; and if the concept of ineffability does not apply to God, then it is not the case that God is ineffable – which contradicts the claim that God is ineffable. In Borg's case, not unsurprisingly, self-refutation tracks strict consistency.

Something similar holds for the strictly consistent panmetaphoricist. She says that no predicate can apply literally to God. When we remind her of the predicate ‘cannot be talked about literally’, she replies: that one doesn't either. But if that's the case, there must be something about God in virtue of which no predicate can apply literally to God, not even the predicate ‘cannot be talked about literally’. 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 is to be used literally to signify whatever that something is, in which case some predicate can apply literally to God after all. As with Borg, so with the strictly consistent panmetaphoricist: self-refutation tracks strict consistency.

Of course, the panmetaphoricist might simply deny the assumption that, if no predicate can apply literally to God, then there is something about God in virtue of which that is so. When we ask her why no predicate can apply literally to God, she might just answer: ‘No reason. It's an inexplicable brute fact.’

What should we make of this suggestion? Three thoughts seem pertinent.

First, no panmetaphoricist I know of endorses it. Rather, they say that no predicate can apply literally to God because God is not a being but the ground of being, or being-itself; or because God is undifferentiated unity, ontologically simple; or because God is transcendent, wholly other, utterly beyond. Furthermore, they say, a predicate can apply literally to God only if God is none of these things. (Hold the question of whether the predicates associated with ‘these things’ apply literally to God or merely metaphorically.)

Second, going brute comes at a price. That's because when we countenance the view that no predicate can apply literally to God, the most natural question to ask is ‘Why?’, and we expect there to be an answer even if it is difficult to discern. At any rate, I expect there to be an answer, as do the panmetaphoricists I know of.

Third, although there is no agreed-upon principle specifying when some fact is inexplicably brute, one might think that, at the very least, a fact is reasonably regarded as brute only if it strikes us as plausible, only if it has some intellectual attraction for us. But does the proposition that no predicate can apply literally to God just seem true, to anyone, independently of anything about God in virtue of which it might be true? Well, perhaps there are some people for whom that is the case. They bring the proposition before their minds eye and say: ‘Yes. That seems exactly right. No predicate can apply literally to God.’ I can only report that I'm not one of them; nor is any panmetaphoricist I know of.

So, by my lights, and those of the panmetaphoricists I know of, brute panmetaphoricism has nothing going for it. Still, I grant that it is a position in logical space, even if no one has occupied it. In what follows, however, the panmetaphoricist with whom I will engage – our panmetaphoricist, as I will call her – is not a brutist.

Our panmetaphoricist might respond differently to the charge of self-refutation. Perhaps she will say that I have misunderstood her thesis. Suppose someone asserts ‘All speech about unicorns is false.’ This could be read as ‘All unicorns have the following property: being such that all speech about unicorns is false.’ On this reading, ‘All speech about unicorns is false’ is self-refuting. But here's another way to read it: ‘Every sentence that implies that there are unicorns is false.’ On this reading, ‘All speech about unicorns is false’ is not self-refuting. Similarly, our panmetaphoricist asserts that ‘True speech about God can only be metaphorical.’ This could be read as ‘God has the following property: being such that true speech about God can only be metaphorical’, which is clearly spoken literally, and so self-refuting. But here's another way to read it: (MRP) ‘Every true sentence which is such that if it were spoken literally, it would imply that “God exists”, spoken literally, is true can only be spoken metaphorically.’ On this reading, ‘True speech about God can only be metaphorical’, although spoken literally, is not self-refuting.Footnote ¹²

Is that right? Does (MRP) avoid self-refutation? I'm not so sure. Suppose, for conditional proof, that (MRP) is true and God exists. Obviously enough, if (MRP) is true and God exists, then no predicate can apply literally to God. But as our panmetaphoricist, who is no brutist, will agree: if no predicate can apply literally to God, then there is something about God in virtue of which that is the case. However, if there is something about God in virtue of which no predicate can apply literally to God, then, if I introduce a new predicate – say, our friend ‘is illiterable’ – and I stipulate that it is to be used literally to signify whatever that something is, then some predicate can apply literally to God. I hereby introduce and stipulate! It follows that, if there is something about God in virtue of which no predicate can apply literally to God, then some predicate can apply literally to God. Therefore, if no predicate can apply literally to God, then some predicate can apply literally to God. And so it follows that, if (MRP) is true and God exists, then some predicate can apply literally to God. Of course, if some predicate can apply literally to God, then (MRP) is false. Therefore, if (MRP) is true and God exists, then (MRP) is false. To say ‘if (MRP) is true, then (MRP) is false’ is to say ‘(MRP) is self-refuting’. Consequently, either (MRP) is self-refuting or it is false that God exists. The theist, naturally enough, will deny the second disjunct and infer that (MRP) is self-refuting. What will our panmetaphoricist do?

Perhaps she'll deny me the power of stipulative definition, at least when it comes to God; or perhaps she'll accept atheism. But she needn't go that far. She can instead distinguish two domains of theological speech. In the domain of first-order theological speech, there is only speech about God, e.g. utterances of ‘God is merciful’, ‘God spoke to Moses’, etc., and ‘God is our fortress’, ‘God stands with us in our suffering’, etc. In the domain of second-order theological speech, there is only speech about speech about God, e.g. utterances of ‘Speech about God can only be metaphorical’, ‘No speech about God can be literal’, etc., and speech about God that is equivalent to speech about speech about God, e.g. utterances of ‘God is such that speech about God can only be metaphorical’, ‘God is such that speech about God cannot be literal’, etc. Panmetaphoricism is better seen as the view that first-order theological speech can apply only metaphorically to God. This leaves it wide open whether second-order theological speech can be spoken literally. Thus, our panmetaphoricist might say that the predicates ‘can only be spoken of metaphorically’ and ‘cannot be spoken of literally’ – not to mention, ‘is illiterable’ and ‘being such that no predicate can apply literally to God’ – belong to the domain of second-order theological speech and so can apply literally to God. If she takes this line, she can avoid self-refutation.

McFague alludes to another reason to distinguish two domains of theological speech when she observes in the second quotation above that ‘the presumption of metaphorical discourse . . . is that these metaphors . . . are of something’, that there is a ‘reality that is the referent of our metaphors’ to which ‘we apply metaphorical thinking’ and speaking. She seems to imply that when we speak metaphorically of God, we presume that our speech is of something, something we refer to with the subject term ‘God’. Of course, how we characterize its referent is up for grabs. Nevertheless, when we take ourselves to speak metaphorically of God, at least we presume that the predicate ‘can be referred to by us with our words’ applies literally to God. The distinction between different domains of theological speech allows our panmetaphoricist to endorse McFague's sensible ‘presumption of metaphorical discourse’ since she can locate utterances of, for example, ‘God can be referred to by words’ in the second-order domain because it is equivalent to ‘Words can be used to refer to God’.

Two-domain panmetaphoricism

To distinguish the initial wholly unrestricted version of panmetaphoricism from our restricted second version, let's call the latter two-domain panmetaphoricism, which we can represent like this:

Although two-domain panmetaphoricism avoids self-refutation, a serious concern remains. For, according to theism, God exists. However, if our first-order theological speech can only be metaphorical – as our two-domain panmetaphoricist insists – then none of it can be literal. And if none of it can be literal, then the predicate ‘exists’ cannot be used literally of God. But if the predicate ‘exists’ cannot be used literally of God, then there is nothing about God in virtue of which the predicate ‘exists’ can apply literally to God. And, if there is nothing about God in virtue of which the predicate ‘exists’ can apply literally to God, then the statement ‘God exists’ is false. But if the statement ‘God exists’ is false, then the statement ‘God does not exist’ is true. And, if the statement ‘God does not exist’ is true, then God does not exist. Thus, panmetaphoricism entails atheism.

How might our two-domain panmetaphoricist reply to this argument that her view implies atheism?

Suppose she accepts the implication that, on her view, God does not exist. She might yet argue that theism is compatible with that implication. ‘True enough’, she might say, ‘God does not exist. But, even so, it does not follow that there is nothing that is God. And what is essential to theism is not that God exists but that there is something – an object, you might say – which is God.’

How could it be that God does not exist even though there is something – an object – which is God? Here our panmetaphoricist might turn to Alexius Meinong who, among others, thought that statements of the form x does not exist are compatible with statements of the form x is an object. That's because the plenitude of objects includes not only those that have being and exist, the real objects, but also the unreal objects, among which there are (i) those that have being but merely subsist and don't exist, the ideal objects, (ii) those that have non-being and are non-contradictory, and (iii) those that have non-being but are contradictory. Our panmetaphoricist might concur and then posit that God is not a real object; rather, God is an unreal object, that is, either God is an ideal object, like the number three or Harry Potter, or God is an object with non-being, of the non-contradictory sort, like your computer painted hot pink or the Fountain of Youth, or God is an object with non-being, of the contradictory sort, like squircles or the set of all sets that are not members of themselves.Footnote ¹³ In any case, the predicate ‘exists’ does not apply literally to God and so God does not exist, God is not real; nevertheless, there is something, an object, that is God. And it is this claim – the claim that God is something, an object, speaking literally – that is essential to theism, not the claim that God exists. Theism, therefore, is compatible with panmetaphoricism.

We might well be unimpressed. For starters, we might deny the distinction between objects that exist (the real ones) and objects that do not exist (the unreal ones). Every object exists; every object is real. Alternatively, we might endorse the distinction but reject the recommended application to God. For example, we might insist that God has being, and so does not ontologically neighbour either your computer painted hot pink or the Fountain of Youth, or squircles or the set of all sets that are not members of themselves; and we might even insist that God does not neighbour either the number one or the number three, nor (sadly) Harry Potter.

But a deeper difficulty lies in the vicinity. According to the reply to the objection, at least some of the predicates that signify these ontological categories – e.g. ‘is an object’, ‘is unreal’, ‘is an ideal object’, ‘has being’, ‘has non-being’, ‘is a non-contradictory object with non-being’, ‘is a contradictory object with non-being’, and others besides – apply literally to God, even if God does not exist, even if God is unreal. In that case, some first-order speech can apply to God literally after all. If our panmetaphoricist replies that her use of these categorial predicates is metaphorical, she will fail to explain how it is that even though God does not exist, there is something – an object – that is God. For, on the ontology she invokes for that explanation, if those predicates don't apply to God literally, God won't show up anywhere on the map of being, not as something that exists or something that does not exist, not as a real object or as an unreal object – which, on the terms of the ontology she invokes, is incoherent.

Our panmetaphoricist might pursue a different tack altogether. Perhaps she will deny that if the predicate ‘exists’ cannot be used of God literally, then God lacks the property of existence. She might insist that, even if ‘exists’ is a predicate, there is no property of existence.

By way of reply, there is no premise in my argument according to which if the predicate ‘exists’ cannot be used of God literally, then God lacks the property of existence. My argument contains the premise that 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 God literally, and it contains the premise that, if there is nothing about God in virtue of which the predicate ‘exists’ can apply to God literally, then the statement ‘God exists’, when used literally, is false. These premises, and those surrounding them, are compatible with existence not being a property; indeed, they are compatible with there being no properties at all.

The panmetaphoricist will notice that, even if my argument does not presuppose that existence is a property, it does presuppose that ‘exists’ is a predicate, a presupposition open to challenge independently of whether existence is a property. Perhaps statements of the form ‘x exists’, where ‘x’ is a singular term, are to be understood as disguised existentially quantified statements (e.g. ‘God exists’ is to be understood as ‘there exists an x such that x is numerically identical with God’, or ‘God exists’ is to be understood as ‘there exists an x such that x is . . .’, where the ellipsis is filled in with some suitable general description). Instead of recasting my argument so as to avoid the presupposition that ‘exists’ is a predicate, I will simply reply that, without loss of force, I could substitute ‘contingently exists or necessarily exists’ for ‘exists’, which is a predicate and not a disguised quantifier.Footnote ¹⁴

Our panmetaphoricist might adopt another strategy. She might say something like this: ‘At one point in your argument, you used these words: “if there is nothing about God in virtue of which the predicate ‘exists’ can apply literally to God, then the statement ‘God exists’ is false.” While I agree that my view has the consequence that there is nothing about God in virtue of which the predicate “exists” can apply literally to God, it does not follow that the statement “God exists” is false, full stop. All that follows is that the statement “God exists” is false provided that the predicate “exists” in that sentence is used literally. If “exists” in the statement “God exists” is used metaphorically, your premises leave it open whether “God exists” is true. And that's the way I intend to use “exists” in “God exists”. Thus, by the time we arrive at the end of your argument, all you've shown is that, on my view, God does not exist, provided that we are speaking literally in uttering that statement. That's hardly surprising, however, since, on my view, first-order theological speech can apply only metaphorically to God.’

What should we make of the suggestion that ‘exists’ can apply only metaphorically to God?

Recall that one of the reasons our panmetaphoricist moved from the wholly unrestricted to the two-domain version of her view was McFague's ‘presumption of metaphorical discourse’ about God, that is, that ‘these metaphors . . . are of something’, something we who participate in that discourse take ourselves to refer to with the subject term ‘God’. But to say that those metaphors are ‘of something’ while at the same time saying that ‘exists’ does not apply literally to what they are of may seem incoherent. After all, how could David's utterance of, for example, ‘God is my refuge and strength’ be of something and yet not be of something that, speaking literally, exists? So, if we endorse McFague's sensible ‘presumption of metaphorical discourse’, it seems that we cannot say that ‘exists’ can apply only metaphorically to God.

Our panmetaphoricist might well demur. She might deny McFague's ‘presumption of metaphorical discourse’ that these metaphors are of something. Alternatively, she might affirm that these metaphors are of something, alright, but insist that they are ‘of something’ only metaphorically speaking. What should we make of these two suggestions?

As for denying that these metaphors are of something, there are two points to consider. The first is that, with rare exception (e.g. ‘The average family has 2.3 children’), it seems to be a presupposition of any sincere assertion of a simple subject–predicate sentence, whether spoken metaphorically or not, that the speaker takes it that the subject term is ‘of something’, that the subject term at least has a referent. At any rate, those who use God-talk in lived religion frequently intend to say something true when, speaking metaphorically, they use subject-predicate sentences with ‘God’ as the subject term, and they frequently intend the subject term to be ‘of something’ in the minimal sense of having a referent. The second point is that to deny that these metaphors are of something in the minimal sense of having a referent is to deny that ‘God’ refers in uses of subject-predicate sentences with ‘God’ as the subject term. But if ‘God’ lacks a referent when such sentences are used – whether the term is used metaphorically, literally, or otherwise – then those sentences are false. Consequently, panmetaphoricists who take this line are atheists. Perhaps theists will be forgiven if they don't follow suit.

What about the second suggestion that these metaphors are ‘of something’, alright, but that they are ‘of something’ only metaphorically speaking? Suppose David sincerely asserts ‘God is my refuge and strength’, and suppose his assertion is ‘of something’, something referred to by his use of the term ‘God’. What would it be for it to be ‘something’ only metaphorically speaking? What sort of being is it, exactly, that our panmetaphoricist, on this second suggestion, would say David's assertion presupposes? She might answer as follows: ‘When I say that “exists” in “God exists” can only be used metaphorically but that, nevertheless, in saying “God is our refuge and strength”, we presuppose that we are speaking of something, where “of something” is also used metaphorically, I do not imply that the referent of the subject term “God” is any sort of being; nor do I imply that it is an object either. Speaking literally, it is not one object among others; and so, speaking literally, it is neither a sort of object nor an object. Only when we speak metaphorically can we speak truly when we say “It exists”, only when we speak metaphorically can we speak truly when we say that our discourse about it – that is, God – is “of something”.’

Here our panmetaphoricist seems to have taken a Tillichian turn. In taking this turn, she has a different answer to our question of what it is that these metaphors are about when she says that they are ‘of something’, metaphorically speaking. According to Tillich, God is not one being among others; rather, God is being-itself. As such, God is ‘the ground of the ontological structure of being without being subject to this structure himself’, God is ‘beyond the subject–object structure of everything that is’; ‘[t]herefore, if anything beyond this bare assertion is said about God’, it is said metaphorically or symbolically and not literally.Footnote ¹⁵ So it is that ‘God exists’ is not true when spoken literally, and it is not true that when we speak metaphorically of God our metaphors are ‘of something’, speaking literally. Tillich's idea seems to be that we can apply a term literally only if what it is to which we apply it is ‘subject to the structure of being’, that is, only if what it is to which we apply it is structured in such a way that a subject–predicate sentence, spoken literally, can be true of it. But being-itself is not so structured. Thus, given the identification of God with being-itself, it follows that we cannot speak literally of God; it follows that ‘every assertion about being-itself [= God] is either metaphorical or symbolic’.Footnote ¹⁶ With exactly one exception, the assertion ‘God is being-itself’.Footnote ¹⁷

Suppose our two-domain panmetaphoricist deploys Tillich's theology in an effort to explain why ‘God exists’ can be true only when it is used metaphorically, and why our metaphorical discourse about God can be ‘of something’ only when ‘of something’ is used metaphorically. Then, by my lights, her position is either internally inconsistent or incoherent.

For suppose she speaks literally when she says ‘God is being-itself’, which is how Tillich understood his claim.Footnote ¹⁸ Then two things follow. First, some first-order speech applies literally to God, which is inconsistent with our panmetaphoricist's claim that no first-order speech applies literally to God. (As I parenthetically hinted at earlier, any serious attempt to say what it is about God in virtue of which no terms can apply literally to God seems bound to be internally inconsistent, if spoken literally.) Second, according to the logic of identity, any open sentence of the form ‘a = b’, spoken literally, is true only if ‘a exists’ and ‘b exists’ are both true, spoken literally. ‘God is being-itself’, spoken literally, as it is by Tillich, is an open sentence of the form ‘a = b’. Therefore, ‘God exists’ is true, spoken literally – which is inconsistent with our panmetaphoricist's claim that ‘God exists’ is not true, spoken literally.

Of course, our two-domain panmetaphoricist might well reply that her use of ‘is being-itself’ is metaphorical, not literal. But in that case she will fail to explain why ‘God exists’ can be true only when it is used metaphorically, and she will fail to explain why our metaphorical discourse about God can be ‘of something’ only when ‘of something’ is used metaphorically. For, on the terms of the Tillichian ontology she invokes for that explanation, if ‘is being-itself’ does not apply literally to God, God won't show up anywhere on the ontological map, not as a being and not as being-itself – which on the terms of that ontology is incoherent.Footnote ¹⁹

It appears, therefore, that our two-domain panmetaphoricist's Tillichian turn is either internally inconsistent or incoherent.

Let's take stock. I argued that two-domain panmetaphoricism, the view that all first-order theological speech is metaphorical, implies atheism. I then considered several replies, each of which failed for one reason or another, by my lights. Towards the end of this article, I will articulate what I regard as a better way for our two-domain panmetaphoricist to reply to my argument that her view implies atheism, a reply that is internally consistent, coherent, and explains why all (or, as I will explain shortly, nearly all) first-order theological speech can apply only metaphorically to God. For now, however, I want to explore another option she might take in response to my argument.

Suppose our two-domain panmetaphoricist endorses McFague's ‘presumption of metaphorical discourse’ and grants that metaphors about God are, speaking literally, of something, and suppose she grants that, speaking literally, God exists. Even so, she might adopt a friendly suggestion from John Hick, who observed that ‘classical thinkers who have affirmed the ultimate ineffability of the divine nature’, ‘need not have worried’ since these points – the points we've just been discussing, namely whether ‘exists’ applies literally to God, to which we might add ‘is self-identical’, ‘is such that modus ponens is valid’, ‘is either pentagonal or non-pentagonal’, etc. – are just ‘logical pedantries’. He continues:

Of course, our panmetaphoricist will want to remind Hick that, although our substantial predicates cannot apply literally to God, they can apply metaphorically. With that caveat in place, she can adopt his recommendation as follows: whereas many purely formal and logically generated predicates apply literally to God, and whereas negative substantial predicates do as well, this need not concern us any more than it concerns the friends of ineffability. Yes, formal predicates – all those ‘logical pedantries’, as Hick calls them – belong to first-order theological speech, as do negative substantial predicates; and, yes, they apply literally to God. And the same goes for ‘exists’ and all the rest. But none of this undermines the thrust of panmetaphoricism, which is that positive substantial first-order theological speech can apply only metaphorically.

Restricted two-domain panmetaphoricism

We began with panmetaphoricism, the view that our (true) speech about God can only be metaphorical. Self-refutation and McFague's ‘presumption of metaphorical discourse’ led us to restrict the view to first-order theological speech. Incompatibility with theism and Hick's distinction between positive substantial predicates, on the one hand, and formal predicates and negative substantial predicates, on the other hand, have now led us to restrict the view further to positive substantial speech about God. Call the result restricted two-domain panmetaphoricism, which we can represent as follows:

I have several concerns about restricted two-domain panmetaphoricism. By my lights, addressing them adequately will, oddly enough, provide reason to lift the restriction of metaphoricity to positive substantial speech about God. As a result, I suspect we'll gain a clearer view of the God of panmetaphoricism.

First concern. My initial concern is that some negative predicates can apply literally to God only if certain positive substantial predicates can also apply literally to God. Consider the predicate ‘is unlimited’, which our panmetaphoricist, following Hick, regards as a negative substantial predicate that can apply literally to God.Footnote ²¹ This predicate seems incomplete: is unlimited with respect to what? Presumably, unlimited with respect to whatever it is that might be limited. And what might that be? Presumably, things like power, knowledge, compassion, etc. But there's the rub. Nothing can be unlimited with respect to power, knowledge, compassion, etc. unless it has some power, knowledge, and compassion, in which case the positive substantial predicates ‘has some power’, ‘knows something’, and ‘has some compassion’ apply literally to it. Therefore, if negative substantial predicates such as ‘is unlimited’ can apply literally to God, then certain positive substantial predicates can also apply literally to God.

Our restricted two-domain panmetaphoricist might well reply: ‘Not so fast.’ ‘After all’, she might say, ‘predicative statements such as “God is unlimited with respect to power” are to be understood as disguised impredicative statements. Thus, for example, “God is unlimited with respect to power” just means “It is not the case that God is limited with respect to power”.’ And now our panmetaphoricist can argue as follows: ‘Since “God is unlimited with respect to power”, spoken literally, just means “It is not the case that God is limited with respect to power”, spoken literally, and since “It is not the case that God is limited with respect to power”, spoken literally, can be true even while “God has some power”, spoken literally, is false, it follows that “God is unlimited with respect to power”, spoken literally, can be true, even while “God has some power”, spoken literally, is false. But “God has some power”, spoken literally, is false only if “has some power” does not apply literally to God. Thus, “God is unlimited with respect to power”, spoken literally, can be true even while “has some power” does not apply literally to God. Therefore, “God is unlimited with respect to power”, spoken literally, does not imply that “has some power” applies literally to God – contrary to the argument expressed in your first concern. And what goes for “God is unlimited with respect to power” goes for any other statements according to which God is unlimited with respect to something. None of them imply that God has some of that with respect to which God is unlimited.’

What should we make of this speech? It seems to me to fail for two reasons.

First, while some negative predications – e.g. ‘Santa Claus does not exist’ – may best be understood impredicatively, ‘God is unlimited with respect to power’ is not to be understood in that way. That's because the predicative ‘God is unlimited with respect to power’ implies that God exists while the impredicative ‘It is not the case that God is limited with respect to power’ does not imply that God exists. Thus, the premise according to which the former ‘just means’ the latter is false. Of course, our panmetaphoricist can sidestep this worry by saying that the predicative ‘God is unlimited with respect to power’ just means ‘God exists and it is not the case that God is limited with respect to power’, which, naturally enough, implies that God exists. Suppose our panmetaphoricist amends her argument accordingly.

Second, the amended argument has this premise: ‘God exists and it is not the case that God is limited with respect to power’, spoken literally, can be true even while ‘God has some power’, spoken literally, is false. This premise seems untrue. For to suppose it is true is like supposing that ‘The number one exists and it is unlimited with respect to successors’ does not imply ‘The number one has some successors’. To suppose it is true is like supposing that ‘You exist and you are unlimited with respect to future times at which you exist’ does not imply ‘You exist at some future times’. These suppositions are false, and so I conclude that my first concern stands.

But perhaps our panmetaphoricist does not mean ‘God is unlimited’ in the way I have understood her. When she said God is unlimited, speaking literally, I took her to mean that, speaking literally, God is unlimited with respect to whatever it is that might be limited. But perhaps she meant something else, something like this: if God is F, speaking literally, for some positive substantial predicate ‘is F’, then God is limited by the property of being an F, or God is limited by falling under the concept of an F, or God is limited by having ‘is F’ apply literally. For God to be unlimited, speaking literally, is for God to be such that, for any substantial positive predicate ‘is F’, God is non-F, speaking literally. And this avoids the difficulty for my previous understanding of our panmetaphoricist's claim that God is unlimited since, naturally enough, something can be non-F, speaking literally, without having at least some F. My toenails, like those of most people, are non-knowers and non-compassionate even though they lack knowledge and compassion altogether. This leads to my second concern.

Second concern. According to this way of understanding our restricted two-domain panmetaphoricist, negative substantial predicates such as ‘is a non-creator’, ‘is non-compassionate’, and, more generally, ‘is non-personal’, as well as ‘is a non-platypus’, apply literally to God; indeed, on her view, for any positive substantial predicate ‘is F’, God is non-F, speaking literally. But recall that our panmetaphoricist rejects brute panmetaphoricism. Thus, on her view, there is something about God in virtue of which, like every other negative substantial predicate, ‘is non-personal’ and ‘is a non-platypus’ apply literally to God. But what is that something? How can it be characterized, on her view?

Is it something to which only formal predicates apply literally? It seems not. Even though we lack a precise definition of ‘formal predicate’, it seems that any uncontroversial instance of one that applies literally to God also applies literally to everything else, and so every such formal predicate applies literally to persons and platypuses: ‘exists’ and ‘is self-identical’, ‘is such that modus ponens is valid’ and ‘is either pentagonal or non-pentagonal’, etc. But none of these or their uncontroversial formal siblings can be, singly or jointly, what it is in virtue of which ‘is non-personal’ and ‘is a non-platypus’ apply literally to God. For, if they were, then, since they apply literally to persons and platypuses as well, it would follow that persons are non-personal and platypuses are non-platypuses, speaking literally – which they are not.

So then: is that in virtue of which negative substantial predicates apply literally to God something to which only formal predicates and negative substantial predicates apply literally? I have no knockdown argument against this possibility. Nevertheless, it seems that, for any other thing, negative substantial predicates apply literally to it in virtue of at least some positive substantial predicate applying to it as well. Gold and hydrogen, for example, are non-personal in virtue of their specific atomic structure; cats and dogs are non-platypuses in virtue of their specific DNA; and so on. To be sure, we might initially say that, for example, my wedding band is a non-platypus in virtue of its being a non-animal; but that won't satisfy us as an explanation of why it is a non-platypus unless we go on to explain why it is a non-animal and we will be satisfied on that score only if, eventually, something positive is said (e.g. something about the atomic structure of the atoms of which it is composed). Indeed, is there anything that we can think of – aside from our restricted two-domain panmetaphoricist's God – to which a negative substantial predicate applies literally, but to which no positive substantial predicate applies whatsoever?

Our panmetaphoricist might well agree: positive substantial predicates are required if we aim to understand what it is about God in virtue of which negative substantial predicates apply literally to God. But, she will insist, they can be applied only metaphorically to God. However: is that enough to gain the understanding we seek? Again, I have no knockdown argument against this possibility. Still, something like the thought of the last paragraph strikes me as plausible. For any x, if there is something about x in virtue of which some negative substantial predicate applies literally to it, then there is something about x in virtue of which some positive substantial predicate applies literally to it as well. Put simply: if we want to understand what it is about something in virtue of which it is a non-F, we ultimately need more than metaphors.

More generally, there's something fishy about using metaphors alone to explain why something is the case, speaking literally. We wonder why water moves downhill, speaking literally. We are told that it seeks the lowest point it can find. That might be a start at gaining understanding, but we should not be satisfied. Why? Because water does not ‘seek’ the lowest point it can ‘find’, speaking literally. Those are metaphors. We want more. Fully successful metaphysical enquiry – in theology as elsewhere – ultimately demands the stone-cold sober truth, spoken literally. And the same goes for our panmetaphoricist's God. Speaking literally, God is non-personal, she tells us. Well, what is it about God in virtue of which, speaking literally, God is non-personal? One answer: ‘God is Energy, connecting, interacting points of light in motion’.Footnote ²² Literally? ‘Well, no, not literally. It's a metaphor, as all positive substantial speech about God is a metaphor.’ Of course, if God, speaking literally, really is Energy, connecting, interacting points of light, then we'd understand what it is about God in virtue of which, speaking literally, God is non-personal. It makes perfect sense. But once we're told that ‘God is Energy, connecting, interacting points of light’ is spoken metaphorically, all bets are off. And the same goes for other panmetaphorical answers to the question of what it is in virtue of which, for any positive substantial predicate ‘is F’, God is non-F, speaking literally.

Third concern. So far I've expressed two rather lengthy concerns about our restricted two-domain panmetaphoricist's claim that negative substantial predicates can apply literally to God. My third concern is the epitome of brevity itself: restricted two-domain panmetaphoricism implies atheism. For, if the predicate ‘is non-personal’ applies literally to God, as our panmetaphoricist says it does, then God exists and God is non-personal, speaking literally. But, according to theism, God exists and God is personal, speaking literally. Therefore, restricted two-domain panmetaphoricism is incompatible with theism, and so panmetaphoricism entails atheism.

Does my claim that ‘according to theism, God exists and God is personal, speaking literally’ beg the question against our restricted two-domain panmetaphoricist? No. For no premise of mine is the mere denial of what she asserts or the mere assertion of what she denies. If I'm right about what theism is, then the theist will deny an implication of our panmetaphoricist's view, but surely that does not suffice for my having begged the question. Theism is not the view that God exists and God is whatever you wish. It is the view that God exists and God is personal, speaking literally. This points leads to my fourth concern.

Fourth concern. Restricted two-domain panmetaphoricism does not sit well with the ecumenical spirit that imbues its adherents. Theistic religious traditions across times and cultures have affirmed that, speaking literally, God is F, for a great many positive substantial Fs. According to them, God is experienced as personal. According to them, God is, speaking literally, almighty, compassionate, and wise, among other things. To say that none of these personal predicates can apply literally to God is, therefore, tantamount to rejecting such traditions in favour of those according to which, speaking literally, God is non-personal. It is tantamount to conceiving of God as Dharmakaya, Brahman, or the Tao rather than Yahweh, the Father, or Allah. Giving pride of metaphysical place to such traditions violates panmetaphoricism's ecumenical spirit.

Perhaps our panmetaphoricist might turn to Hick for help in restoring the spirit of ecumenism to her position and, as a bonus, avoid implying atheism. Recall that Hick, speaking of what he calls ‘the Ultimate’, and at other times ‘the Real’, says:

Hick's line of thought here seems to be this: suppose you want to affirm those traditions whose members experience the Ultimate as personal as well as those whose members experience the Ultimate as non-personal. You can't do so by saying both experiences are veridical. For then the Ultimate ‘would have mutually contradictory attributes’. But neither do you want to say that just one is veridical since you would not be regarding with sufficient equanimity the ‘fruits in human life’ of both. So what to say? Answer: ‘postulate a reality to which these predicative dualisms do not apply’. That is, the Ultimate is neither personal nor non-personal, neither a creator nor a non-creator, neither compassionate nor non-compassionate, etc.

Perhaps our panmetaphoricist can say something similar. When she is at her ecumenical best, she wants to affirm those traditions whose members apply ‘is personal’ to God as well as those whose members apply ‘is non-personal’. She can't do so, however, if she says that all such predicates apply to God literally since God would then have contradictory attributes. But neither should she say that ‘is non-personal’ applies literally while ‘is personal’ does not. That violates her ecumenism. Moreover, she should recognize that the members of each tradition have a stake in preserving their speech about God because of the moral, social, and spiritual value embedded in their linguistic and liturgical practices. So what to do? Answer: our speech about God, whether it is positive or negative substantial speech, can only be metaphorical. Or, more accurately at this point in the dialectic: we can postulate that each predicate of a staked-out predicative dualism applies only metaphorically to God, where a predicative dualism is an instance of the schema ‘is F or non-F’ and a predicative dualism is staked out if and only if different religious traditions have a stake in applying each of their constituent predicates to God; otherwise, a staked-out predicative dualism applies literally to God. Upshot: even-handed ecumenism is restored, at least when it comes to speaking literally about God. For, on the panmetaphoricism at which we have arrived, ‘is non-personal’ applies literally to God no more than ‘is personal’, which is to say neither applies literally to God at all.

To sum up: restricted two-domain panmetaphoricism allowed some negative substantial predicates to apply literally to God that, on further reflection, are better seen as applying only metaphorically: specifically (i) those negative substantial predicates whose literal application implies the literal application of certain positive substantial predicates and (ii) those negative substantial predicates that partly constitute a staked-out predicative dualism, for example, ‘is personal’ and ‘is non-personal’. Moreover, restricted two-domain panmetaphoricism allowed some formal predicates to apply literally to God that, on further reflection, must apply only metaphorically – namely staked-out predicative dualisms, for example, ‘is either personal or non-personal’. So it is that we recover some of the ‘pan’ in panmetaphoricism that restricted two-domain panmetaphoricism gave up. This should be good news for our panmetaphoricist.

I now want to argue that she should recover even more of the ‘pan’ in panmetaphoricism. To see why, note that the recovery project as it stands still allows some predicative dualisms to apply literally to God, namely those not all of whose constituent predicates are staked-out by some religious tradition. For example, since no tradition has a stake in thinking of God as a platypus, it allows the negative substantial predicate ‘is a non-platypus’ to apply literally to God, and consequently it allows the staked-out predicative dualism ‘is either a platypus or a non-platypus’ to apply literally as well. Here a difficulty begins to emerge.

For given what some traditions have actually deemed special foci of God's relation to the natural order, it is not all that far-fetched to consider what would have been the case if our species had evolved in such a way that some culture had a stake in thinking of God as specially related to platypuses – say, by becoming one of them – and so had a stake in applying ‘is a platypus’ to God. Or consider what would have been the case if our species had evolved in such a way that no religious tradition had a stake in thinking of God as personal and so had no stake in applying ‘is personal’ to God. In the first case, restricted two-domain panmetaphoricism implies that ‘is a non-platypus’ would not have applied literally to God; God would not have had the property of being a non-platypus, although he actually does have it. In the second case, it implies that ‘is non-personal’ would have applied literally to God; God would have had the property of being non-personal, although he actually does not have it. But surely we can't be expected to accept that simply by virtue of the historic accident of ‘is a platypus’ not being staked-out, ‘is a non-platypus’ applies literally to God; surely we can't be expected to concede that simply by virtue of the historic accident of ‘is personal’ being staked-out, ‘is non-personal’ does not apply literally to God.

Think of it this way. On restricted two-domain panmetaphoricism, sometimes there's a third option to God's being F or non-F, namely in those cases where, due to historical contingencies, different religious communities have a stake in whether each constituent of the predicate ‘is either F or non-F’ applies to God. In those cases, God is neither F nor non-F, and so the predicate ‘is either F or non-F' does not apply to God. That's the third option. But whether or not there's a third option has to do with God, surely, and not the chanciness of terrestrial evolution in general or human biological or cultural evolution in particular. In that case, the sensible thing for the panmetaphoricist to say seems to be that no substantive predicates apply literally to God, whether positive or negative, in which case no substantive predicative dualism applies literally to God – full stop. Of course, she might yet say that some first-order formal speech can apply literally to God. But other than that, first-order speech about God is only metaphorical.

Partly recovered two-domain panmetaphoricism

Suppose that our two-domain panmetaphoricist says that both positive and negative substantial speech about God can apply only metaphorically to God. Call the resulting view partly recovered two-domain panmetaphoricism, which we can represent like this:

At this point, we might well wonder whether the recovery will put the panmetaphoricist back on the sickbed. For how could God fail to fall into one of the mutually exclusive and jointly exhaustive classes expressed by a predicative dualism? After all, everything is either a platypus or a non-platypus. Everything is either personal or non-personal. How could it be otherwise? These are important questions, one our panmetaphoricist must answer. So, then, how can it be that God is neither a platypus nor a non-platypus, neither personal nor non-personal, and so on? How could there be a third option?

The only way in which there could be a third option, so far as I am able to see, can be illustrated with a homely example. Consider the predicate ‘is bald’. Now imagine a man who is a borderline case of baldness, a man who is such that no amount of empirical research or armchair philosophizing can decide the question of whether the quantity and distribution of his hair renders him bald. In such a case, some philosophers – in particular those who characterize vagueness as metaphysical rather than epistemic or linguistic – will say that there is no determinate fact of the matter about him in virtue of which he has the property of being bald.Footnote ²⁴ Or, without properties: there is no determinate fact of the matter about him in virtue of which the predicate ‘is bald’ applies literally to him. Thus, the proposition that he is bald is neither true nor false. Likewise, the proposition that he is either bald or non-bald is neither true nor false. The predicative dualism ‘is either bald or non-bald’ does not apply literally to him. Why? Because there is no determinate fact of the matter about him in virtue of which it could. And the same goes for most other positive and negative substantial predicates in a natural language. We can imagine borderline cases of their application and get similar results.

Here our panmetaphoricist might well take note. For what these philosophers say about our borderline case of a bald man and the predicate ‘is bald’, she can say about God and every substantial predicate. (Well, every substantial predicate that admits of a borderline case. Perhaps ‘is a circle’ and other mathematical predicates do not admit of borderline cases. I leave this qualification implicit.) For any positive substantial predicate ‘is F’, there is no determinate fact of the matter about God in virtue of which God has the property of being F. Or, without properties: for any positive substantial predicate ‘F’, there is no determinate fact of the matter about God in virtue of which ‘is F’ applies literally to God. Thus, for any proposition of the form God is F, it is neither true nor false that God is F. And the same goes for negative substantial predicates. Likewise, for any proposition of the form God is either F or non-F, it is neither true nor false that God is either F or non-F, where ‘is F’ is a substantial positive predicate and ‘is non-F’ is a substantial negative predicate. Thus, for example, the predicative dualism ‘is either personal or non-personal’ does not apply literally to God since there is no determinate fact of the matter about God in virtue of which it could. At best, we can apply substantial predicates only metaphorically to God, whether positive or negative.

This way of understanding panmetaphoricism strikes me as a significant advance since it nicely explains why our substantial predicates cannot apply literally to God, whether positive or negative. On the explanation offered here, there simply is no determinate fact of the matter about God in virtue of which our substantial predicates could apply literally to God. When it comes to substantial predication, we are left with metaphorical application alone, which is exactly what our partly recovered two-domain panmetaphoricist would like to say. Indeed, it is difficult to see how else it could be that, for any positive substantial predicate ‘is F’, God is neither F nor non-F, speaking literally.

Might our two-domain panmetaphoricist fully recover her position? Might she extend what I have suggested that she say about substantive predication to formal speech about God? Maybe. At any rate, philosophers who say vagueness is metaphysical sometimes say that self-identity and existence are vague.Footnote ²⁵ Suppose they're right. Then our panmetaphoricist might insist that something similar goes for God. There is no determinate fact of the matter about God in virtue of which God has the properties existing and being self-identical. Or, without properties: there is no determinate fact of the matter about God in virtue of which ‘exists’ and ‘is self-identical’ apply literally to God. Thus, the proposition that God exists is neither true nor false, as is the proposition that God is self-identical. Likewise, the predicative dualisms ‘is either existent or non-existent’ and ‘is either self-identical or non-self-identical’ do not apply literally to God since there is no determinate fact of the matter about God in virtue of which these predicates can apply literally to God. At best, we can apply predicates of existence and self-identity only metaphorically to God. We might well suspect that, if there is no determinate fact of the matter about God in virtue of which predicates of existence and self-identity can apply literally to God, then there is no determinate fact of the matter about God in virtue of which any formal predicate can apply literally to God. There just isn't enough determinately there, so to speak, for them to latch onto.

If our suspicions are correct, then our partly recovered two-domain panmetaphoricist will have made a full recovery: our first-order speech about God, whether substantial or merely formal, can only be metaphorical. This is panmetaphoricism worthy of the name!

But won't our fully recovered two-domain panmetaphoricist still be subject to our earlier objection, according to which, 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 literally to God, in which case the statement ‘God exists’ is false – which is to say that God does not exist? And so won't we have to conclude that her view entails atheism?

In short, the answer is ‘No’. For according to our fully recovered two-domain panmetaphoricist, the inference from ‘there is nothing about God in virtue of which the predicate “exists” can apply literally to God’ to ‘the statement “God exists” is false’ is invalid. That's because, on her view, although there is no determinate fact of the matter about God in virtue of which ‘exists’ can apply literally to God, and although, as a consequence, the statement ‘God exists’ is not true, it is also the case that there is no determinate fact of the matter about God in virtue of which the statement ‘God exists’ is false, and so it does not follow that God does not exist. Our fully recovered two-domain panmetaphoricist is no atheist.

Let's return to partly recovered two-domain panmetaphoricism, according to which some formal predicates can apply literally to God, but no substantial predicate ‘is F’ or ‘is non-F’ can apply literally to God since there is no determinate fact of the matter about God in virtue of which they can do so. I have three concerns about this view.

My first concern is that it is impossible. That's because there are pairs of substantial predicates, ‘is F1’ and ‘is F2’, such that, necessarily, for any x, if x is a borderline case of the application of ‘is F1’, then x is not a borderline case of the literal application of ‘is F2’. For example, if something is a borderline case of the application of ‘is located all and only in Australia’, then it is not a borderline case of the literal application of ‘is located all and only in Brazil’. That's because, necessarily (and holding fixed the actual locations of Australia and Brazil), if something is indeterminate enough to be a borderline case of the application of ‘is located all and only in Australia’, then it is determinate enough not to be a borderline case of the literal application of ‘is located all and only in Brazil’; speaking literally, it is not located all and only in Brazil. And the point here holds not just for relational predicates. For example, if something is a borderline case of the application of ‘is perfectly loving’, then it is not a borderline case of the literal application of ‘is obstinately wicked’. That's because, necessarily, if something is indeterminate enough to be a borderline case of the application of ‘is perfectly loving’, then it is determinate enough not to be a borderline case of the literal application of ‘is obstinately wicked’; it is determinate enough to be not obstinately wicked, speaking literally. Similarly, if God is indeterminate enough to be a borderline case of the application of ‘is perfectly loving’, then God is determinate enough not to be a borderline case of ‘is obstinately wicked’; God is determinate enough to be not obstinately wicked, speaking literally. And the same goes for other pairs of predicates, e.g. ‘is omnipotent’ and ‘is wimpy’, ‘is omniscient’ and ‘is an ignoramus’, etc.Footnote ²⁶

My second concern is that if no substantial predicate can apply literally to God, then no substantial predicate applies more aptly to God than any other. You say God is compassionate and a non-platypus; I say God is a platypus and non-compassionate. According to our partly recovered two-domain panmetaphoricist, there is no determinate fact of the matter about God that could settle the question of whose substantial predicates more aptly apply, yours or mine.

She might well reply: ‘Although no substantial predicate can apply literally to God, it does not follow that no substantial predicate applies more aptly to God than any other. That's because even if no substantial predicate can apply literally to God, some substantial predicate can apply metaphorically to God more aptly than some other. You have not ruled out this possibility.’

But is it really possible for a substantial predicate to apply metaphorically to God more aptly than another if there is no determinate fact of the matter about God in virtue of which it does? I can't see how. Return to our borderline case of a bald man. Suppose, as we did above, that there is no determinate fact of the matter about him in virtue of which the predicate ‘is bald’ applies literally to him. The number and distribution of hairs on his head don't allow it. The proposition that he is bald is neither true nor false; and the same goes for the propositions that he is non-bald and that he is either bald or non-bald. There isn't enough determinately there, so to speak, for ‘is bald’ or ‘is non-bald’ to latch onto. In that case, might it nevertheless be more apt to apply ‘is bald’ to him than ‘is non-bald’? Presumably. For example, his body might be otherwise hairless. But then, it is only because ‘is hairless everywhere else’ applies literally to him that it is more apt to apply ‘is bald’ metaphorically to him than ‘is non-bald’. There is some determinate fact of the matter about him – his being hairless elsewhere – in virtue of which ‘is bald’ applies metaphorically more aptly than ‘is non-bald’. Absent any such explanation, there is no basis for supposing that ‘is bald’ more aptly applies metaphorically to him than ‘is non-bald’. The same goes for God. The predicate ‘is compassionate’ can apply metaphorically more aptly to God than ‘is non-compassionate’ only if there is some determinate fact of the matter about God in virtue of which it does so. There is no such fact, however, according to our partly recovered two-domain panmetaphoricist. And the same point holds for any substantial speech about God. So it is that our panmetaphoricist's ban on literal substantive predication frustrates any expectation that God is more aptly metaphorically described one way rather than another.

So, it appears that partly recovered two-domain panmetaphoricism implies that no substantial predicate can apply metaphorically to God more aptly than any other. Three things follow.

To sum up metaphorically: panmetaphoricism entails acute religious anaemia – or worse.

My third concern about partly recovered two-domain panmetaphoricism is this. It implies that there is no determinate fact of the matter about God in virtue of which either ‘is personal’ or ‘is non-personal’ can apply literally to God. Therefore, the statement ‘God is personal’, spoken literally, is not true. Moreover, it implies that there is no determinate fact of the matter about God in virtue of which ‘God is personal’, spoken metaphorically, is more apt than ‘God is non-personal’, spoken metaphorically, and so the former is no more apt as a metaphor than the latter. But theism implies that the statement ‘God is personal’, spoken literally, is true, or at least that ‘God is personal’ is more apt as a metaphor than ‘God is non-personal’. Thus, partly recovered two-domain panmetaphoricism is not theism – even though, unlike its fully recovered sibling, it does allow ‘exists’ and other formal predicates to apply literally to God.

The upshot is that our partly recovered two-domain panmetaphoricist fails to be a theist every bit as much as her fully recovered sibling. They are both non-theists, you might say, even if they are not atheists.