Taking a Wittgensteinian line, D. Z. Phillips construes the question of the reality of God as like the question of the reality of physical objects in general, and unlike the question of the reality of any particular physical object such as a unicorn. Phillips would therefore have a bone to pick with Edward 'Cactus Ed' Abbey who writes,
Is there a God? Who knows? Is there an angry unicorn on the dark side of the moon?
Abbey's meaning is clear: It is as idle to suppose that there is a God as to suppose that there is an irate unicorn on the far side of the moon. Of course, there could be such a unicorn. It is logically possible in that there is no contradiction in the idea. It is also epistemically possible in that the supposition is consistent with what we know. (Perhaps a clever extraterrestrial scientist synthesized a unicorn, put him in a space suit, and deposited the unfortunate critter on the moon.) But there is no positive reason to believe in something so outlandish. The same goes for God according to Abbey, Russell, and plenty of others. Such theists think of God as just one more being among beings, as something in addition to all the other things that exist.
How might a theist respond to this puerile conception? (And to such cognate 'objections' as Russell's Teapot? ) One way to respond is that of the Wittgensteinian fideist. A fideist like Phillips would take Abbey to have misconstrued the very sense of the theist's affirmation. Abbey takes the theist to be adding a weird object to the ontological inventory: hence the comparison of God to an irate lunar unicorn. Phillips, however, clearly sees that this is a mistake. His positive theory, however, is just as bad. Phillips thinks that the claim that God exists is more like the claim that there are physical objects in general. That there are physical objects in general is presupposed by any inquiry into whether a particular physical object exists. It is a presupposition without which such an inquiry would make no sense. As Phillips puts it:
Similarly, the question of the reality of God is a question of the possibility of sense and nonsense, truth and falsity, in religion. When God's existence is construed as a matter of fact, it is taken for granted that the concept of God is at home within the conceptual framework of the reality of the physical world. . . . to ask a question about the reality of God is to ask a question about a kind of reality, not about the reality of this or that, in much the same way as asking a question about the reality of physical objects is not to ask about the reality of this or that physical object. ("Philosophy, Theology, and the Reality of God," Phil. Quart., 1963, reprinted in Rowe and Wainwright, p. 281)
Phillips blunders badly in this passage when he says that construing the divine existence as a matter of fact takes it for granted that the concept of God belongs within the conceptual framework of the reality of the physical world. For if anything is clear, it is that God for the theist is not a physical object. Surely, in claiming that God exists as a matter of fact, the theist who understands his doctrine is not claiming that God exists as a physical object. What Phillips should have said is that construing God's existence as a matter of fact presupposes, not that God is a physical object, but that God is a being or an existent.
Phillips would want to deny that too. His view is that the reality of God is not the reality of a special being, but the reality of a presupposition that is not and cannot be questioned from within a religious language-game (or at least from within a theistic religious language-game). The reality of God has to do with what a religious believer is prepared to say: ". . . the religious believer is not prepared to say that God might not exist. It is not that as a matter of fact God will always exist, but that it makes no sense to say that God might not exist." (280)
Perhaps the following analogy will clarify what Phillips is driving at. Consider the reality of checkmate in chess. The existence of checkmate is not a matter of fact in the way that it is a matter of fact that Karpov opened a certain game with 1. d4. For within the game of chess it makes no sense to say that checkmate might not exist. Checkmate and the rules governing it are defining features of the game. They cannot be questioned from within the game, and to question them from without the game is simply to reject the game. For that reason, it makes no sense to demand proof of these rules, nor can one raise the question whether they are reasonable. Is it reasonable that one cannot castle out of check, into check, or though check? It is neither reasonable nor unreasonable. The question of reasonableness cannot arise. Similarly, for Phillips, it is neither reasonable nor unreasonable that God exists. To play a theistic language-game is to presupposes the meaningfulness of God-talk just as to play chess is to presuppose the meaningfulness of talk of checkmate. And just as God exists in theistic language-games, checkmate exists in chess. But also: just as checkmate does not exist outside chess, God does not exist outself theistic language-games. For if God does exist apart from theistic language-games, then there would be a fact of that matter as to the existence of God.
At this point one can see what is wrong with Phillips' view. Every sane person is an anti-realist about checkmate, but to be an anti-realist about God, as Phillips' view seems to require, is to make a joke of theistic belief. Phillips' claim that "theology is the grammar of discourse" (282) is therefore as preposterous as the claim that botany is the grammar of discourse about plants. There is of course a sense in which for the theist the existence of God is necessary, but this is not the sense in which a rule is necessary for a language-game. Chess is not chess without checkmate, so checkmate is necessary within chess. God, however, is not a rule, nor a linguistic presupposition, nor concept, nor anything dependent on human talking and acting. So the necessity of God is not the necessity of a rule. God is a necessary being, which implies that he is a being, which implies that he exsts independently of human talk and speech if he exists at all. God cannot be reduced to God-talk and God-ritual. Chess just is chess-talk and chess-ritual: chess has no reality outside chess conventions and the chessic form of life. Not so with God.
These points are frightfully obvious, but one can understand why Phillips was driven to contravene them. Surely God is not a physical object, and it is arguable that he is not a being among beings. What then is God, and how understand his reality? His is not the reality of any sort of abstract object, nor that of any sort of collection; thus he is not the world-whole. So Phillips is driven to say something equally untenable, namely, that God is immanent to certain language-games, as a sort of framework truth of those language-games.
The question of the reality of God is a hard nut to crack, but surely it won't do to say that the reality of God is a matter of talk and practice, as if God were merely a feature of certain language games and forms of life. If God exists, then he is a reality transcendent of any Sprachspiel or Lebensform.