Religious Language: Twentieth-Century Perspectives

Logical positivists sought to distinguish genuine factual discourse from metaphysics. A. J. Ayer states a verification principle: a proposition is factually meaningful only if it is analytic or empirically verifiable in principle. Analytic propositions are true through meanings or logical form; synthetic propositions assert possible experience. Ethical and theological utterances which meet neither test are not thereby false but lack literal factual significance.

Strong verification would require conclusive establishment and excludes much science and history. Ayer therefore favours weak verification: experience must make a proposition probable or relevantly support it. “In principle” matters because a claim can be meaningful even if current technology cannot test it. The challenge for God-language is to identify possible observations which bear on an unrestricted, transcendent being rather than reinterpret every result as compatible.

The criterion faces self-application. The verification principle is not obviously analytic or empirically verifiable, so by its own rule it appears meaningless. It may instead be proposed as a recommendation for using “factually meaningful”, but then it cannot establish that religious speakers misuse language; it stipulates a narrower practice. Scientific claims about unobservables also gain meaning through theoretical networks, not simple observation sentences. Verificationism remains valuable as a demand for evidential connection, but its global semantic verdict is hard to sustain.

John Hick uses the parable of two travellers on one road. One believes it leads to a Celestial City; the other believes it leads nowhere, and the journey's ambiguous pleasures and hardships fit either interpretation. At the end, arrival would verify the believer, while no post-mortem subject may remain to register non-arrival. Theism can therefore be verifiable without being symmetrically falsifiable.

The proposal relies on personal continuity and recognisability. An afterlife experience must be identifiable as fulfilment rather than another ambiguous stage. Hick uses replica theory to show logical possibility of resurrection: a person appearing in another world with the right continuity could count as the same. Even if coherent, the parable shows only that some theistic claims have conceivable verification conditions, not that evidence now makes them true.

Critics argue that delayed verification weakens present cognitive content and that mutually exclusive religions can each imagine confirming afterlives. The believer can reply that a condition need not be currently accessible to confer meaning; claims about remote history and future cosmology are meaningful before testing. Hick therefore challenges Ayer's exclusion, but leaves an epistemic question: rational belief now requires some present evidence or cumulative support beyond a possible future check.

Antony Flew adapts the parable of the invisible gardener. Explorers encounter a clearing; repeated failed tests lead the believer to qualify the gardener as invisible, intangible and undetectable. Flew asks how the claim now differs from no gardener. Religious assertions may begin cognitively, but “die the death of a thousand qualifications” when every apparent counterexample is absorbed.

Falsification here is not the claim that a sentence must already be false. It asks what the speaker recognises as counting against it. “God loves us” seems confronted by apparently pointless suffering. If no imaginable suffering changes confidence or interpretation, the sentence may not function as an empirical assertion. Flew's closing questions concern both the love and existence of God, though these may require different falsifiers.

A believer can distinguish ad hoc evasion from principled qualification. Scientific theories survive apparent counterexamples when auxiliary explanations have independent support; classical theism may similarly include freedom, stable law and limited human understanding. The danger is unlimited elasticity. A rigorous response states what evidence would lower probability even if it would not deductively refute God. This probabilistic interpretation preserves committed faith while conceding that evidence can count negatively.

R. M. Hare's paranoid student believes all dons want to murder him. Evidence does not revise this blik, a fundamental way of seeing the world which shapes what counts as evidence. Bliks are not ordinary explanatory hypotheses, yet they matter profoundly: sane and insane bliks lead to different lives. Hare suggests religious commitment may operate at this framework level.

The account answers the charge that unfalsifiable language is insignificant. Trust in providence can orient courage, repentance and hope even if no single observation tests it. It also explains why disputes persist when parties do not share standards of interpretation. Yet calling theism a blik appears to surrender truth. Hare's example deliberately describes a false or insane framework, and believers usually claim God actually exists rather than merely that a God-blik is useful.

A charitable development treats bliks as epistemic frameworks which can be assessed holistically for coherence, scope and fruits rather than tested one sentence at a time. Frameworks are not immune to rational comparison. Still, Hare's original move is better as an account of non-cognitive religious attitude than a defence of realist theism. It shows meaning without verification, not the truth of the religious object.

Basil Mitchell's partisan meets a stranger who claims to lead the resistance. The stranger sometimes helps and sometimes appears to aid the enemy. The partisan retains trust because of the initial encounter, but admits that contrary appearances count against the claim and require explanation. Faith is neither a conclusively falsified hypothesis nor an evidence-proof blik.

The parable models cumulative commitment. A foundational experience or body of testimony supplies reason for trust; subsequent evidence is interpreted within that relationship. This resembles interpersonal knowledge, where one troubling action does not erase established character. Nevertheless, enough unresolved evidence could make continued trust irrational. Mitchell therefore preserves cognitive content: the stranger really is or is not on the resistance's side.

The analogy's weakness is epistemic access. The partisan has met a publicly embodied stranger; many believers appeal to less direct religious experience. It also does not specify a threshold at which anomalies defeat commitment. Its advantage over Hare is clear: evidence matters. Its advantage over Flew is that rational beliefs are rarely abandoned after one counter-instance; assessment includes prior warrant, explanatory integration and the cost of alternatives.

The Vienna Circle sought to separate empirical science and logical analysis from metaphysics that appeared to generate disputes without possible resolution. Logical positivism combined empiricism with modern formal logic: genuine factual claims should connect, directly or indirectly, with possible experience, while mathematics and logic clarify relations between symbols. Theology was not refuted as a failed scientific theory; it was classified outside literal factual discourse.

This ambition explains both the attraction and severity of Ayer's criterion. If two speakers appear to disagree but no possible experience bears on the dispute, perhaps they do not advance rival descriptions of reality. The approach promises intellectual hygiene by asking what difference a sentence makes to anticipation. It also shifts the burden from proving “God does not exist” to showing what “God exists” asserts.

The criterion depends on an observational foundation that later philosophy of science contests. Observations are described through concepts and background theories; electrons, fields and distant cosmic events are not verified by simple unaided sensations. Their meaning comes from networks of inference, instruments and predictions. Religious claims may likewise belong to cumulative interpretive networks, although that analogy does not prove they have equivalent evidence.

A high-attainment response distinguishes a challenge from a verdict. Verificationism legitimately asks for evidential connection and exposes statements immunized against every outcome. It does not follow that one proposed criterion exhausts language. Moral judgements, historical explanations, statements about other minds and metaphysical assumptions within science all require more complex treatment than the simplest observational test.

Strong verification requires conclusive establishment by experience. Few universal scientific laws meet it, because finite observations cannot prove every future and unobserved instance. Claims about the distant past and other minds also exceed decisive inspection. A criterion designed to protect science therefore excludes much successful science if interpreted too strictly.

Weak verification asks whether experience can make a claim probable or count for its truth. “In principle” verification allows tests no present person can perform: a claim about the far side of a planet remains meaningful because imaginable observation would bear on it. This broadens the criterion, but creates uncertainty about how indirect the connection may be. Theists cite experience, order, conscience and historical testimony; Ayer can say these do not verify an infinite transcendent being. The dispute becomes evidential rather than a clean semantic exclusion.

The principle's self-reference is a famous difficulty. “Only analytic or empirically verifiable propositions are factually meaningful” is not itself an analytic truth and no sensory test establishes it. If it is a practical proposal for using the phrase “factually meaningful”, it cannot demonstrate that other communities misuse language; it recommends a linguistic policy. If it states a fact about meaning, it appears to fail its own test.

A defender can treat the principle as a clarification derived from examining successful discourse rather than a metaphysical proposition. Yet that modest status weakens its anti-theological conclusion. At most, theology is not factual under this stipulated model. The enduring value of verification is methodological: ask what observations support a claim and whether apparent support is selective. The global claim that unverifiable language has no cognitive meaning is harder to maintain.

Hick's Celestial City parable answers the idea that religious claims can never encounter verifying experience. Two travellers interpret the same ambiguous road differently; arrival at the city would settle their disagreement. Similarly, continued personal existence in a state recognizable as fulfilment could verify claims about God and afterlife. Present lack of access does not remove meaning, just as a prediction remains meaningful before its test.

The proposal requires a criterion of personal identity. If death destroys the organism and a psychologically continuous person appears elsewhere, is that the same subject or a duplicate? Hick's replica thought experiment argues that bodily continuity need not be uninterrupted if the new person has the right memories, character and unique causal relation. Duplication remains a challenge: if two equally exact replicas appeared, both could not straightforwardly be numerically identical with one individual.

Recognition is another condition. An afterlife may itself be ambiguous; unfamiliar experiences could be dreams, simulations or further natural stages. Verification requires enough conceptual and experiential continuity to identify divine fulfilment rather than simply survival. Different religions also predict conflicting destinations. Arrival might confirm a broad transcendent hypothesis while falsifying particular doctrines.

Hick accepts asymmetry: arrival verifies, whereas failure may leave no surviving subject to observe that there is no city. This shows that verifiability does not entail falsifiability. Critics may regard one-sided tests as weak, but many existential claims have asymmetrical evidence. The argument succeeds against blanket meaninglessness; it does not supply present warrant or prove that the proposed conditions will occur.

Popper uses falsifiability to distinguish empirical science from non-science, not meaningful language from nonsense. A scientific theory takes risks by ruling out observable states. Flew's theological challenge is related but broader: what does a believer's assertion exclude? If “God loves every person” is compatible with every conceivable pattern of suffering, the assertion may cease to function as it initially appeared.

The invisible-gardener parable traces this death through qualification. A gardener becomes invisible, intangible, scentless and undetectable after every failed test. Each qualification might be intelligible alone; together they remove the contrast between the asserted gardener and no gardener. Flew's concern is not that sophisticated concepts are forbidden but that revision is driven solely by counterevidence without independent reason.

Theists can respond that God was never proposed as a physical gardener. Non-detection by wire fences is irrelevant to a non-embodied cause, so rejecting that test clarifies rather than retreats. Likewise, divine love need not predict a pain-free life if freedom and stable law belong to the doctrine independently. The relevant question is whether those qualifications were already motivated and whether any evidence could lower confidence.

A probabilistic reading improves the debate. Few major worldviews have a single decisive falsifier, but evidence can count for or against them. A believer may retain faith after tragedy while admitting that apparently gratuitous evil is negative evidence outweighed by experience or other arguments. The assertion remains cognitively alive when it exposes itself to graded revision, even if no one event deductively refutes it.

Hare's lunatic believes every university don is plotting murder. Reassuring encounters are interpreted as deceptive performance; hostile encounters confirm the suspicion. The conviction is not an ordinary hypothesis revised by evidence but a blik, a basic way the world is seen. Bliks determine what counts as evidence and can be sane or insane, yet they profoundly shape conduct.

The idea explains why verification is not the only form of significance. Trust that reality is meaningful, that other people are persons or that moral obligation is real may structure inquiry before particular tests. Religious commitment can function similarly, orienting guilt, hope, gratitude and risk. No isolated observation contains the whole framework's meaning.

Hare's concession is substantial: bliks are not assertions and so are not true or false in the ordinary factual way. Most theists do not merely adopt a helpful interpretive stance; they believe God exists independently of it. Worse, the example shows a harmful false framework. Practical influence proves importance, not rationality or truth.

A contemporary development can assess frameworks holistically. Coherence, explanatory reach, openness to correction and moral fruits distinguish responsible from paranoid bliks. Still, once those criteria count as evidence, the view moves towards Mitchell rather than remaining entirely non-cognitive. Hare's lasting contribution is to expose the background level at which faith operates. It is a partial account of commitment, not a full defence of realist theology.

Mitchell's partisan trusts a stranger encountered during occupation. The stranger sometimes helps the resistance and sometimes appears to assist its enemies. The partisan does not deny these troubling facts; they count against trust. An initial meeting, however, supplies a reason to resist immediate abandonment. Faith is an assertion and a commitment whose evidence is assessed across a relationship.

This resembles ordinary personal knowledge. One anomalous action does not erase years of known character, yet accumulating unexplained betrayal eventually can. Prior warrant affects how new evidence is weighted. Religious experience, testimony and a cumulative case may play the role of the encounter, while suffering and hiddenness are acknowledged as counterevidence rather than defined away.

The analogy is stronger than Hare's because truth remains at issue: the stranger really is loyal or disloyal. It is less vulnerable than Flew's target because qualifications arise from an independently grounded relationship. However, the story does not specify how much adverse evidence should defeat trust. Without a possible threshold, “we do not yet understand” can continue indefinitely.

The initial encounter also needs scrutiny. A direct meeting with a visible agent differs from culturally mediated religious experience. Competing faiths cite different encounters. Mitchell offers a structure for rational commitment, not proof that any believer possesses adequate initial warrant. Its success therefore depends on an epistemology of experience and testimony supplied outside the parable.

Flew asks whether a religious sentence retains assertive content. Hare preserves existential significance by giving up ordinary assertion. Mitchell preserves assertion by allowing evidence to count within durable commitment. Their answers should not be flattened into three opinions about the same test. They prioritize different dimensions: empirical risk, framework orientation and relational trust.

Their parables also model different pathologies. The gardener illustrates ad hoc retreat; the lunatic illustrates interpretive immunity; the partisan illustrates loyalty that may be rational or may become stubborn. Each can illuminate religious and non-religious positions. Naturalism can function as a blik, and scientific communities also protect promising theories while anomalies are investigated. The question is not whether a belief is resilient but how resilience is earned.

A robust cognitive faith should incorporate all three insights. It makes claims about reality; it recognizes experiences that count negatively; it belongs to a wider orientation shaping interpretation; and it weights evidence in light of prior grounds rather than responding to one isolated fact. These features do not guarantee truth. They specify what intellectual responsibility would look like.

Exam evaluation improves when it asks which religious utterance is under discussion. “God exists”, “God loves me”, “I trust God” and a liturgical acclamation do not have identical grammar. Flew's test is strongest against apparent descriptions, Hare illuminates orientation and Mitchell illuminates trust. A universal verdict on all “religious language” ignores this internal variety.

The early Tractatus presents meaningful propositions as pictures of possible states of affairs. Logical form enables a sentence to represent how the world might be. Ethics, aesthetics and the mystical lie beyond what can be stated as facts, though Wittgenstein does not call them unimportant. This setting partly resembles the later positivist concern with the limits of factual representation, even though the Vienna Circle's appropriation was not identical with Wittgenstein's purpose.

The later Philosophical Investigations rejects one essence of language. Naming and describing are only some activities among commanding, joking, praying, thanking and calculating. Meaning is found in public rule-governed use within human life. The shift is from looking for a universal representational mechanism to examining diverse practices.

This development matters for theology. The early framework may tempt a division between factual propositions and ineffable value. The later approach asks how words such as “God”, “soul” or “last judgement” are learned and what role they play. It does not follow that they refer to nothing. Describing use is compatible with speakers making truth claims, just as the use of scientific language includes representation.

Invoking “language-games” as an automatic defence therefore misreads the method. Wittgenstein is not granting each community a private reality. He is warning that philosophical confusion arises when words are detached from the circumstances giving them grammar. Whether the practice is justified or its objects exist remains a further question, but it should be asked after the language has been accurately understood.

A language-game is language interwoven with action. In prayer, words may petition, confess, praise, lament and reshape attention. Their sense depends on relations to stories, rituals, expectations and conduct. A person who can translate every word but cannot distinguish praise from bargaining has not fully grasped the game. Understanding is practical competence, not only dictionary substitution.

A form of life is the shared human background in which rules have grip. It includes embodied needs, social responses and patterns of training. Religious forms of life are not wholly sealed: believers and non-believers share pain, historical inquiry, moral vocabulary and ordinary standards of sincerity. Games overlap through family resemblances and can borrow or contest one another's concepts.

Public rules prevent radical subjectivism. A worshipper cannot make any use correct by private declaration; communities distinguish orthodox and mistaken uses, genuine confession and performance, responsible and manipulative interpretation. This also makes internal critique possible. Reformers expose contradictions between proclaimed mercy and actual exclusion by using the tradition's own grammar.

External critique remains possible where claims cross into shared practices. If a community says a miracle occurred in public history, historical evidence matters. If a ritual harms children, moral and legal language-games are relevant. Context protects against category mistakes but not accountability. The best Wittgensteinian defence therefore explains distinctive criteria without turning them into immunity.

Later Wittgenstein rejects the idea that every meaningful word gains meaning by naming an object. “Meaning is use” directs attention to rule-governed practices. A language-game is not trivial play but language interwoven with activity: promising, calculating, praying, confessing and mourning have different norms. A form of life is the wider human practice and shared response within which such rules make sense.

Religious utterances may therefore be misunderstood when treated as primitive scientific hypotheses. “God watches over me” may express trust, shape conduct and connect worship, narrative and community. Understanding requires competence in that grammar, just as understanding a chess move requires more than identifying the material movement of wood. This does not automatically make religion non-cognitive; practices include assertions, but their criteria are not assumed to be identical to laboratory prediction.

The charge of fideism says language-games become insulated islands, each true by its own rules. Wittgenstein does not say every use is correct or that games cannot overlap and be criticised. Religious and ordinary moral language share concepts of evidence, suffering and honesty. Outsiders can learn a practice, and internal inconsistency remains possible. Still, if “true” is wholly game-relative, disagreement between believer and atheist risks being dissolved instead of addressed. A realist use of language-games should clarify grammar while allowing extra-linguistic reality to constrain claims.

Verification asks what could support a proposition; falsification asks what could count against it; language-game analysis asks what role the words perform and what rules govern them. These questions are not interchangeable. A sentence may have clear use without observational verification, or possible verification without a simple falsifier. Hare and Mitchell respond to Flew in different registers: Hare relocates significance to an unfalsifiable framework, while Mitchell preserves a truth-claim exposed to cumulative evidence.

Religious language is also heterogeneous. “God exists”, “God loves me”, “Blessed be God” and “I promise before God” perform different acts. One criterion applied to all will misclassify some. Existence and historical miracle claims appear cognitive; praise and commitment have expressive or performative dimensions; parables may reshape perception. Mixed function does not entail that factual dimensions disappear.

A strong judgement therefore avoids announcing one universal winner. Ayer exposes the need for evidential connection but over-restricts meaning. Flew exposes ad hoc immunity but may model faith too narrowly on isolated hypotheses. Hare protects life-orienting significance but weakens realism. Mitchell gives the most balanced account of evidence-sensitive commitment. Wittgenstein explains why this diversity is unsurprising, while needing a realist supplement if religious and non-religious speakers genuinely disagree about reality.

Wittgensteinian fideism is the view that religious belief is intelligible and assessable only within its own form of life, so external demands for evidence misconstrue it. D. Z. Phillips often emphasizes that belief in resurrection, providence or immortality gains sense from a moral and spiritual practice, not from speculative observation of invisible mechanisms.

This protects religion from reduction to failed science. Asking where a soul is spatially located may misunderstand what talk of a person's eternal significance does. It also takes formation seriously: someone may need to learn practices of forgiveness and worship before grasping a doctrine's depth, just as musical judgement requires trained participation.

The danger is circularity. If only participants can judge the practice and its standards are validated internally, no evidence could show that the community is mistaken. Converts, apostates and comparative scholars demonstrate that people can understand and evaluate across boundaries. Religions themselves make missionary and universal claims rather than treating belief as a local game equivalent to a hobby.

Later Wittgenstein does not entail relativism. Grammar identifies what a claim means; it need not guarantee truth. A moderate reading treats participation as epistemically useful but allows rational assessment through overlapping standards, testimony, coherence and consequences. Fideism is strongest against crude external tests and weakest when used to avoid genuine disagreement about a shared world.

Cognitive language presents propositions capable of truth or falsity. Non-cognitive language expresses attitudes, commits a speaker, performs an act or evokes a way of seeing. The categories describe functions, not mutually exclusive vocabularies. “God created all things” asserts dependence, expresses gratitude and places the speaker under an ethic of stewardship at once.

Speech-act theory helps. Saying “I promise” does not report an inner promise but enacts one under appropriate conditions; it still creates facts and can be sincere or insincere. Liturgical words may praise or confess rather than hypothesize, yet presuppose beliefs about their addressee. Expressive force does not eliminate ontological commitment.

Reduction in either direction loses meaning. Treating prayer only as a set of factual sentences misses address and relationship. Treating resurrection only as a symbol of hope may ignore a tradition's historical assertion. Analysis should identify the dominant act, its presuppositions and the criteria appropriate to each dimension.

This plural account answers a weakness in global verification debates. Some religious sentences can be assessed historically or philosophically; others are evaluated as promises, commands or narratives. The same discourse links them. A worldview can fail if its factual presuppositions collapse, its practices become morally corrupt or its symbols no longer disclose. Rational assessment is richer, not weaker, when these dimensions are distinguished.

Contemporary philosophy often treats evidence probabilistically. Evidence supports a hypothesis when it is more expected if the hypothesis is true than if a rival is true. This avoids requiring a single conclusive verifier or falsifier. Order, religious experience, hiddenness and suffering can each change relative confidence without mechanically deciding the question.

The approach vindicates part of Flew's demand: a claim predicts nothing if every possible observation is equally expected on it. It also vindicates Mitchell: prior probability and an existing evidence base affect how strongly a new anomaly should move belief. Qualifications are legitimate when independently supported and ad hoc when introduced only to save a threatened conclusion.

Exact numbers are rarely available. Theists disagree about what a loving creator would be expected to produce, while naturalistic hypotheses vary. The framework is most useful conceptually: identify alternatives, state expected observations, avoid counting the same evidence twice and admit disconfirming data. It replaces a binary “meaningful or meaningless” dispute with a graded question about responsible belief.

Bayesian treatment cannot capture all language. Praise and commitment are not probabilities, though their background beliefs can be assessed. Nor does probabilistic support guarantee existential adequacy. It provides a bridge between analytic argument and lived faith: a commitment may exceed a numerical calculation while remaining accountable to evidence that genuinely changes rational confidence.

Different traditions use overlapping words—God, liberation, enlightenment, revelation—within distinct narratives and practices. Language-game analysis helps prevent false equivalence: “salvation” in one tradition may not perform the same role as “awakening” in another. Understanding requires attention to internal grammar before comparison.

Difference also challenges insulation. If incompatible communities each appeal to a self-authenticating form of life, internal meaningfulness cannot determine which claims are true. Eschatological verification may be imagined differently by each; bliks can organize lives in conflicting ways; symbolic systems can all be powerful. Public reasons become important precisely because sincere use is plural.

Comparison need not impose one crude scientific test. Traditions can be assessed for coherence, explanatory scope, historical grounding, capacity to incorporate counterevidence and moral fruits. These criteria themselves require debate, and no result will be neutral from every worldview. Still, shared human forms of life make dialogue possible: people recognize suffering, contradiction, manipulation and intellectual virtue across boundaries.

The lesson for twentieth-century theories is balanced. Positivism demanded public accountability but narrowed rationality excessively. Fideism recovered context but risks closed circles. Mitchell's evidence-sensitive commitment and a moderate language-game account offer a middle route: the insider's practice contributes understanding, while claims reaching beyond the community remain open to common inquiry.

Ayer supplies a rigorous question about possible evidence but mistakes one type of test for the boundary of cognitive meaning. Hick shows that future and indirect conditions can make theological claims verifiable, though possible verification is not present justification. Flew reveals the danger of endless qualification but overstates the need for isolated decisive falsifiers.

Hare identifies world-orienting commitment but gives away too much of the believer's realism. Mitchell best models trust that remains answerable to adverse evidence, provided the initial warrant and a threshold for revision can be defended. Wittgenstein explains why all these accounts go wrong when they assume every utterance has one function, but language-games do not make communities infallible.

The theories can be ordered as questions. First ask what the utterance does in its practice. Then identify any factual presupposition and its possible evidential relations. Ask what would count against it, whether qualifications are independently motivated and how commitment affects interpretation. Finally test its coherence and ethical use across overlapping forms of life.

This sequence avoids a superficial “each has strengths and weaknesses” conclusion. It explains why no universal criterion wins: the object of assessment is diverse. It also avoids relativism: diversity of function does not remove truth, evidence or criticism. Religious language is meaningful where speakers can show its rules, contrasts, referential commitments and consequences rather than protecting it through unexplained exception.