Religious Language

Before asking whether "God is good" is true or known, there is a prior question: what does it mean? Religious language uses ordinary words — good, wise, powerful, loves, exists, father — of a being said to be infinite, immaterial, timeless, and simple. If those words keep their ordinary sense, they seem false or category-mistaken (an immaterial being has no hands, a timeless one cannot act); if they lose it entirely, religious talk says nothing. This is the semantic question of the three that organize the field, and for a stretch of the twentieth century it was thought to be decisive: if religious sentences are literally meaningless, then the arguments for and against God, and the problem of evil, are all disputes about nonsense.

This page sets out the verificationist challenge and the falsification debate (Flew's gardener; the Flew–Hare–Mitchell exchange), the classical theories that make God-talk meaningful but non-literal (analogy, the via negativa, symbol, metaphor), and the Wittgensteinian reconception of religious language as a "language-game." The further question of whether religious claims are true/false at all (cognitivism vs. non-cognitivism, realism vs. anti-realism) is taken up in realism and anti-realism.

The verificationist challenge

The sharpest attack came from logical positivism and its verification principle (the Vienna Circle; A.J. Ayer, Language, Truth and Logic, 1936): a statement is factually/cognitively meaningful only if it is either analytic (true by definition, like mathematics) or empirically verifiable (confirmable, at least in principle, by sense experience). Statements meeting neither test are not false but meaningless — pseudo-statements that look like assertions but express no proposition.

Ayer's verdict on theology was blunt: "God exists," "God loves us," and equally "God does not exist" are all cognitively meaningless, because no possible experience would confirm or disconfirm them. Religious language is, on this view, in the same boat as metaphysics generally — neither true nor false, but empty (at most, an expression of feeling). This is a more radical attack than the problem of evil: it does not say theism is false but that it never managed to say anything.

The verification principle, however, self-destructed. Applied to itself, it is neither analytic nor empirically verifiable — so by its own criterion the verification principle is meaningless. Attempts to weaken it (to "confirmability in principle") either let metaphysics and theology back in or excluded legitimate science (universal laws, the past, other minds, are not strictly verifiable either). With the collapse of the principle, the blanket verificationist dismissal lost its force — but it bequeathed a sharper, more durable challenge: the falsification version.

The falsification debate

Antony Flew ("Theology and Falsification," 1950s) recast the challenge around Popper's falsifiability: a genuine assertion about reality must rule something out — there must be some conceivable state of affairs incompatible with it, such that, were it to obtain, the assertion would be false. Flew's worry: religious believers seem to let nothing count against their claims. He adapts John Wisdom's parable:

Two explorers find a clearing with both flowers and weeds. One says a gardener tends it. They watch — no gardener appears. The believer says the gardener is invisible. They set traps, electrified fences, bloodhounds — nothing is detected. Still the believer says: an invisible, intangible, eternally elusive gardener tends the garden. At last the skeptic asks: "How does what you call an invisible, intangible, eternally elusive gardener differ from no gardener at all?"

Flew's point: as believers qualify "God loves us" against every apparent counter-evidence (a child dying of cancer is met with "God's love is not merely human love…"), the assertion "dies the death of a thousand qualifications" — it is eroded until it forbids nothing and so asserts nothing. If no conceivable suffering could count against "God loves us," the sentence is not a factual claim.

The University discussion (Flew, R.M. Hare, Basil Mitchell) gave the classic replies:

R.M. Hare — the "blik." Hare concedes religious utterances are not falsifiable assertions but denies they are meaningless. They express a blik — an unfalsifiable fundamental attitude or frame through which one sees the world (his example: the paranoid who is certain all dons want to murder him, whom no friendly don can reassure — that is an unfalsifiable blik, and we all have sane ones, e.g. trust in the uniformity of nature). Bliks are not factual claims, but they matter and differ (sane vs. insane). Religion is a blik. Objection: this seems to concede Flew's point — that religion makes no factual claim — and to reduce God-talk to attitude, which most believers reject (they mean to assert that God really exists). It slides toward non-realism.
Basil Mitchell — the partisan. Mitchell agrees religious claims are assertions and that evil does count against "God loves us" — but argues believers can rationally maintain them despite counter-evidence, as one trusts a friend. His parable: in wartime, a partisan meets a Stranger who tells him he is on the partisan's side; thereafter the Stranger is sometimes seen helping the resistance, sometimes apparently aiding the enemy. The partisan, having once trusted, holds "he is on our side" through the ambiguous evidence — not letting it be conclusively falsified, yet acknowledging it as evidence against that he must bear. Faith is like this: the believer counts evil against God's love (so the claim is falsifiable in principle, contra Flew) but does not treat it as decisive (so the claim survives, contra the demand for easy falsification). This preserves religious language as genuinely assertoric while explaining its resilience — the most influential of the three replies.

Analogy, negation, symbol, metaphor

Independently of the modern verification debate, the tradition long recognised that God-talk cannot be literal-univocal and developed theories of how it means. The problem: if "wise" means the same applied to Socrates and to God (univocal), we anthropomorphise and limit God; if it means something wholly different (equivocal), we say nothing and cannot reason about God at all.

Analogy (Aquinas). Between univocity and equivocity lies analogical predication. When we say "God is good," "good" is used neither in exactly the human sense nor in a totally unrelated one, but in a proportionally related sense. Two forms: the analogy of attribution (a quality is named from its cause or source — as food is "healthy" because it causes health in the body, God is "good" as the cause of all goodness) and the analogy of proper proportionality (goodness is to God as goodness is to creatures, each proportioned to the kind of being that has it — divine goodness is to the divine nature as human goodness is to human nature). Analogy lets religious language be meaningful and partly informative without either caging God in human concepts or collapsing into nonsense. It is the mainstream Catholic account.
The via negativa (apophatic theology). Maimonides, Pseudo-Dionysius, and the mystical tradition hold that God so exceeds our concepts that we can only say what God is not — not finite, not mutable, not corporeal, not ignorant. Positive predications are reinterpreted as negations of privations (God is "living" = God is not inanimate). This safeguards divine transcendence and the ineffability of mystical experience, but risks emptiness: a being of which only negations are true threatens to become an unknowable blank (Aquinas's worry about pure negation, which motivates analogy as a positive supplement).
Symbol (Tillich). Paul Tillich holds that religious language is symbolic: a religious symbol "participates in the reality to which it points" and opens up levels of reality (and of the self) otherwise closed. The only non-symbolic statement we can make about God, for Tillich, is that God is "being-itself" (the ground of being), not a being; everything else — God as "father," "king," "person" — is symbol. Symbols, unlike signs, cannot be replaced at will and "die" when they lose their power. Objection: if all positive God-talk is symbolic and only "being-itself" is literal, it is unclear that anything determinate (and true/false) is being asserted — Tillich's view edges toward the non-realist.
Metaphor. Modern theology (Sallie McFague, Janet Soskice) treats much religious language as metaphor — "God is a rock," "a shepherd," "a fortress" — irreducibly figurative yet cognitively significant (metaphors can say things no literal paraphrase captures, as in poetry and science). The question is whether metaphor is reducible (a decorative stand-in for a literal claim) or irreducible but still referential (Soskice's "reality depicting" without literal description), which connects to the realism debate.

Wittgensteinian language-games

The later Wittgenstein's philosophy of language reframed the whole question. Meaning is use: words have meaning within language-games embedded in forms of life — rule-governed practices (giving orders, joking, praying, thanking) — and a word's sense is fixed by its role in such a practice, not by naming an object or stating a verifiable fact. D.Z. Phillips and other "Wittgensteinian" philosophers of religion applied this:

Religious language is its own language-game, with its own "grammar," learned by participating in the religious form of life (worship, prayer, confession). To understand "God exists" is to understand its use in that practice — not to treat it as a quasi-scientific hypothesis about an extra object whose probability the arguments assess.
The verificationist (and, ironically, the natural theologian) misunderstand the grammar of religious language by importing the standards of one game (empirical fact-stating) into another. "Does God exist?" asked as if about a hidden physical entity is a grammatical confusion; within the religious game, belief in God is more like a commitment, an orientation, a "picture" that regulates a life.

This is liberating but double-edged. It protects religious language from external (verificationist) refutation — and from the problem of evil as a factual disproof — by denying that religion competes with science at all. But critics (including many theists) charge that it deflates religion into a self-contained practice cut off from truth: if "God exists" makes no claim assessable from outside the game, it seems to slide into non-realism (Phillips is often, though he resisted it, read this way), abandoning the believer's conviction that God really, mind-independently exists. The cost of immunity from refutation is the loss of factual content — the same trade-off the independence model of faith and reason faces.

Where religious language sits

The semantic question is logically prior to the existence and knowledge questions: if religious sentences are meaningless (verificationism) or make no factual claim (Hare's blik, Wittgensteinian and non-realist readings), then the arguments for God and the problem of evil are confused from the start. The verificationist attack collapsed with its self-refuting principle, and the falsification challenge was met — most persuasively by Mitchell — by showing that religious assertions can be genuine, evidence-sensitive claims that are nonetheless rationally resilient. The constructive theories — analogy, the via negativa, symbol, metaphor — secure that God-talk is meaningful but non-literal, occupying the middle ground between an anthropomorphic univocity that limits God and an equivocity that empties the words. The Wittgensteinian turn offers the most radical reframing, protecting religious language from external refutation at the risk of severing it from truth. Which of these is right determines whether one is a realist or non-realist about God — the question taken up next — and it conditions everything in the epistemology (one cannot be justified in believing what one cannot meaningfully state) and in the pluralism debate (whether rival traditions contradict one another depends on whether their language asserts incompatible facts).