God and metaphor

Many people think that metaphorical God-talk is not cognitive because it cannot be descriptive. For opponents of metaphorical God-talk, this is a reason to reject it and to defend literal language as the only possible way to express knowledge about God.Footnote ¹ Most defenders of metaphorical God-talk presuppose or accept that it is not cognitive at all or only in a weak sense. Do we have to accept the assumption of both opponents and defenders that metaphorical God-talk cannot be descriptive? In this article, I shall try to show that metaphors can describe God and that metaphorical descriptions of God can be true in the sense that they correspond to extra-lingual reality. Therefore, it can be cognitive.

My argument will proceed as follows. First, I will show that metaphorical description is a coherent notion because a metaphorical statement can have meaning and reference. Next, I will explore in which way metaphorical descriptions could apply to the reality of God. Then, the concept of truth, the truth-condition and the truth-criterion of metaphorical statements in general are discussed. I conclude by proposing an account of the concept of truth, the truth-condition and the truth-criterion of Christian metaphorical God-talk.

Metaphorical meaning

The meaning of a word must be distinguished from its use and from its reference. The standard meanings of words are the lexical ones in the dictionary; they belong to the code of a given language system (‘langue’, La Saussure). How these words are used in speech (‘parole’) in different contexts is another matter. Some argue that the meaning of a word depends only upon its use in different contexts, following Wittgenstein's claim that ‘for a large class of words – though not for all – . . . the meaning of a word is its use in the language’.Footnote ² I will assume that the meaning of a word cannot be reduced to its use.Footnote ³ The meaning or conceptual content of a word also differs from its reference. The relation between meaning and reference is complicated. On the one hand, we can use words to refer to certain kinds of things because words have a specific meaning in a given language system.Footnote ⁴ On the other hand, the use of a word in order to refer to other things than usual can change its meaning in the history of a language.Footnote ⁵

A word has a literal meaning.Footnote ⁶ It can also be used in a non-literal, figurative way. Metaphor is one of the ways of a figurative, indirect use of words: we refer to things by words we normally use to refer to other things. The metaphorical use of words suggests a kind of correspondence, similarity or analogy between the thing we refer to by the metaphorically used word and the thing(s) to which this word normally refers.Footnote ⁷ New, creative metaphors disclose similarities unknown until then. The metaphorical use of words creates a metaphorical statement with a metaphorical meaning. Some metaphorical statements have a meaning which cannot completely be translated into literal meanings; I call these irreducible metaphors. Following these stipulations, Janet Soskice defines a metaphor as that figure of speech in which words are used to speak about one thing in terms suggestive of another.Footnote ⁸

That irreducible metaphorical statements can have a metaphorical meaning is crucial for the truth of metaphorical God-talk: if this were not the case, metaphorical God-talk could only say something about God if it could be reduced to literal language. The claim that metaphorical statements have a meaning, however, is controversial. According to Donald Davidson, the meaning of a metaphorical word or statement can only be its literal meaning because there is no other meaning. Metaphor belongs exclusively to the use of words. When a metaphorical statement produces a meaning for the hearer, it is the hearer who arbitrarily attaches this meaning to the statement. There can be as many metaphorical meanings as there can be hearers. A metaphorical meaning cannot be the meaning of the metaphorical statement because there is no rule to infer this meaning from the statement.Footnote ⁹

However, it is impossible to separate meaning and use completely. The meaning of an utterance is created by both meaning and use of words in a communicative context. A metaphorical utterance can produce a metaphorical meaning precisely because there is no meaning without use, context and reference.Footnote ¹⁰ Paul Ricoeur even goes as far as to say that the metaphorical use of a word can create a new meaning of that word.Footnote ¹¹ Furthermore, a successful creative metaphorical statement discloses specific similarities, not just any similarity.Footnote ¹² Many metaphors give clues for finding this specific similarity.Footnote ¹³ Especially when a metaphor functions in a narrative, the narrative reduces the number of similarities the metaphor suggests.Footnote ¹⁴ Davidson is right that there are no rules which explain how a metaphorical meaning is produced. But should we try to explain this? Isn't it essential for a creative and successful metaphor that it be discovered, not deduced? This is the view of Eberhard Jüngel: if a metaphor succeeds, it conveys a specific metaphorical meaning, but whether it succeeds does not depend upon any rule. A metaphor is a free invention and discovery and in its freedom compelling – if it succeeds. A successful metaphor is a contingent event.Footnote ¹⁵

Metaphorical description

It is necessary for the truth of metaphorical God-talk that a metaphorical statement can refer: if a metaphorical statement could not refer, it could not be about God as an extra-lingual being. To what does a statement generally refer? Does it only refer to entities, or does it also refer to relations, actions and properties? These questions can be answered in different ways.

A first view is that a statement by its subject term refers only to one entity; the predicate of the statement does not refer, but merely express a subjective view or perspective on that entity. Another view is that the predicate in a statement refers to a property which exists independently of its bearer and is instantiated in this particular entity. ‘Alvin is a philosopher’ is taken to refer both to a person, Alvin, and to the property of being-a-philosopher. The ontological presupposition of this view is that properties exist in the same objective and independent way as entities. A third possible view is that a statement refers to a state of affairs, for example the state of affairs that Alvin is a philosopher. To deny any reference of the predicate can easily lead to scepticism and relativism. Reference of the predicate to an independent property implies the existence of such properties, which is a very robust, Platonic ontological claim. In the following discussion, I will assume that a statement as such and as a whole refers to a state of affairs, that is to say, to an entity represented by the subject term and to the relations, actions, properties and circumstances of that entity. I call the first reference identification, and the second reference description.

The reference of a metaphorical statement can be explained in three ways: by a theory of double reference, a theory of split reference and a theory of single reference. According to the theory of double reference, which is held by Henle, Black and Alston,Footnote ¹⁶ we look at the thing which is mentioned or suggested by the metaphorical word as a model for the subject matter of the metaphorical statement. In order to understand ‘The chairman ploughed through the discussion’, we have to think about somebody who ploughs, and to look at the chair through the ‘filter’ or ‘screen’ of this model.Footnote ¹⁷ Black stresses that it is not the factual model which is important in metaphorical description, but the system of commonplaces which are associated with it in a given language community. ‘Man is a wolf’ can only be understood if we know what people think about wolves.

Paul Ricoeur proposes a theory of split reference.Footnote ¹⁸ His explanation focuses on the process of creating and discovering a new metaphor. The meaning of a new, creative metaphorical statement cannot be produced without its literal meaning and its literal reference, but the literal meaning and reference appear to be impossible. So the metaphorical meaning and reference do not simply replace the literal ones. A metaphor creates a new metaphorical meaning and reference by destroying the literal ones; while the literal meaning and reference are being destroyed, they remain in a way present and active in the metaphorical utterance.Footnote ¹⁹ The process of destruction is part of the event which creates the new metaphorical meaning. Therefore, the meaning of a metaphorical statement is a matter of tension or conflict between literal and metaphorical meaning.Footnote ²⁰

Soskice proposes a theory of single reference. In her view a successful metaphorical statement has only a metaphorical meaning and one subject matter. Although a model is implicitly supposed, a metaphorical statement does not refer to it.Footnote ²¹ In some cases, a metaphorical statement is our only access to reality, for instance the metaphorical description of a private and particular grief.Footnote ²²

These three theories are focused on different phenomena and provide solutions for different problems. Therefore, I think they are not mutually exclusive. All of them are useful, but in different situations. When we are confronted with a completely new metaphor, which we initially don't even recognise as a metaphor, the idea of split reference is the best one. If somebody hears ‘The Lord is King’ for the first time and is puzzled by it, the theory of split reference explains how he discovers the meaning and reference of this statement. When we concentrate on the meaning of a metaphorical statement in a communicative situation between speaker and hearer, the idea of single reference is a good explanation. When somebody has known ‘The Lord is King’ for a long time and shares this statement with others in a language community, it will have a single meaning and reference in that community. When we use a metaphor as a model in order to think about something else we cannot directly and completely grasp, double reference is appropriate. If somebody tries to reflect on God's reign in this world, he has to reflect on the similarities and differences between the reign of God and the reign of the kings in the world. The creation of completely new metaphors and the use of metaphors as models are exceptional cases. Repeated use of metaphorical statements in a language community is normal; in this case, the theory of single reference is the best explanation.

According to the description theory of reference, which claims that the meaning of a term determines to which things this term can refer, metaphorical statements cannot be descriptive because they have no complete and definitive meaning. Only terms with a complete and definitive meaning can refer: the definition of ‘gold’ is the enumeration of the properties of gold, and we can use the term ‘gold’ to refer to the things that possess all these properties. Following Kripke and Putnam, Janet Soskice and Wentzel van Huyssteen have argued that this is false. If we discover new properties of a natural kind like gold or a physical process like electricity, the meaning of the terms ‘gold’ and ‘electricity’ changes. Still, although their meaning has been changed, we can use these terms to refer to the same things; this shows that terms without complete and definitive meaning can refer.Footnote ²³ Therefore, although they have no complete and definitive meaning, metaphorical predicates can be descriptive if the metaphorical statement in which they occur produces a specific metaphorical meaning which is understood by the hearers or readers. This is often the case. In a language community, successful metaphors acquire a standard meaning by repeated use.

There is no need to restrict the term description to literal predication, and to contrast literal description with metaphorical depiction or with metaphorical imagination, as Soskice and Van Huyssteen do. This contrast suggests that metaphorical language as such is less objective and descriptive than literal language. But description is not by definition literal. If everybody in a community perfectly understands what is meant by ‘Julia is the sun in her family’, this is a metaphorical description. If we use a metaphor as a model to understand invisible physical reality and the model proves to have explanatory and predicting power, we are entitled to say that the model offers a partial and indirect, that is, a metaphorical description of invisible reality. So there is literal and metaphorical description.

Metaphorical description and the reality of God

Metaphorical description is a coherent notion, but metaphorical description might have ontological and epistemological implications which jeopardise realistic and cognitive God-talk. If we compare metaphorical God-talk with other forms of metaphorical description, such as works of literary art or scientific models, we might think that the metaphorically described God necessarily becomes a fictional or hypothetical entity. So the question arises, to what extent these other forms of metaphorical description can illuminate metaphorical God-talk.

Ricoeur considers a work of literature as the proposition of a world.Footnote ²⁴ This world is the result of imagination, but it is not merely unreal. A work of literature describes how our human world could have been; it is a possible world with regard to our real world; not a different world, but our world differently. Ricoeur calls it a redescription. It discloses our human possibilities to inhabit this world. In this sense, literature offers an indirect, metaphorical description of our world. Sallie McFague has applied Ricoeur's notion of metaphorical redescription to God-talk. In her view, God-talk refers to our world, to our way of being-in-the-world, our way of relating to God.Footnote ²⁵ God can only be talked about as an implication or presupposition of our existence, experience and interpretation. To construe metaphorical God-talk as a redescription of our world risks reducing it to talk about our world, lives and experiences. It has great difficulty avoiding the non-realist consequence that God is a fictional character.Footnote ²⁶

Ricoeur himself is more prudent and avoids subsuming God-talk to literary redescription of our world. In his important article ‘Nommer Dieu’, he argues that scripture is in a sense literature, but that it differs from all other kinds of literature in that it names God.Footnote ²⁷ Naming God is not just one of the many ways of imaginative redescription of our world. God is the ultimate and primordial referent of all the biblical texts.Footnote ²⁸ The metaphors by which we indirectly describe God are models of his Name and his relationship with us.Footnote ²⁹ These remarks of Ricoeur's suggest an understanding of biblical, metaphorical predicates of God as indirect, non-conceptual, but nonetheless realistic descriptions of the God to whom the Bible bears witness.Footnote ³⁰ This means that metaphorical description of God is not a particular case of literary metaphorical redescription of the world.

Another example of metaphorical description is the use of models in the natural sciences. The most fruitful models are theoretical models, sets of terms and formulae.Footnote ³¹ Models do not refer directly to experience; they refer putatively ‘to possibly real entities, relations, and states of affairs’.Footnote ³² A model can be suggested by a metaphorical statement like ‘the brain is a computer’. This model itself suggests new metaphors, such as ‘input’, ‘feedback’ and ‘programming’. These can become terms in a theory which hypothetically explains how the brain works. Theories generate existential hypotheses: the billiard ball model of gases suggests the hypothesis that molecules exist.Footnote ³³ If the model has explanatory power, it is reasonable to believe that the metaphorical theory terms it generates ‘may be “partially denoting” or reality depicting’.Footnote ³⁴ Soskice has used this form of metaphorical description to explain metaphorical God-talk. Metaphors for God can become models when they suggest many characteristics and implications. ‘Father’ is the central model for God in Christian God-talk because it suggests many relations, acts and properties of God. Therefore, for Christian believers, the term Father depicts the reality of God and our relationship to him.Footnote ³⁵

However, there are differences between metaphorical description in natural science and in God-talk. First, the non-perceivable entities, relations and processes scientific models indirectly describe do not exist independently of matter. If metaphorical description of God would be an instance of metaphorical description in science, the ontological status of God would not be different from that of non-perceivable physical reality, and ultimately his existence would be as hypothetical as the existence of molecules. But, according to the Christian faith, God is the creator of visible and invisible things (Col. 1:15; Nicene Creed). Second, the explanation of regularities in physical phenomena is different in nature than the explanation of the personal relation between God and believers, as Vincent Brümmer has rightly pointed out. Footnote ³⁶ Thus, although it is an indirect description of invisible reality, metaphorical description of God is not a particular case of metaphorical description in science.

Both literary imagination and scientific models are no more than analogies of metaphorical God-talk. This has two consequences. First, metaphorical description in literature and science shows its importance in acquiring knowledge about human existence and physical reality, but is not sufficient to justify the possibility of true metaphorical God-talk. This possibility can only be based on two premises: that metaphorical description is possible, and that God can be known and characterised as he is. Second, metaphorical description is used in three different areas of being: in literature about an imagined, possible world, in natural science about physical reality, and in faith and theology about the reality of God. If we do not ontologically distinguish the reality of God from imagination and physical reality, God can only exist as part of our imaginative mind or as part of physical reality. In both cases he ceases to be the creator.

The truth of metaphorical descriptions

Truth is a relation between a statement and the reality the statement is about. A concept of truth is a conception about what kind of relation this is. The concept of truth as correspondence I try to defend here only means that a state of affairs really is as it is described by a statement; it does not imply that a true statement is an image, mirror or copy of reality. The correspondence between statement and reality need not be complete: it neither requires that every element of a statement represents an element of reality, nor that every element of reality is represented in the statement. Could there be any correspondence between a metaphorical statement and reality, given the fact that in a metaphorical statement a metaphorically used predicate has lost the complete definitive meaning it normally has and does not describe the entities it normally describes?Footnote ³⁷ Can we specify the conditions which have to be met by the statement in order to be true? And how could we know whether this correspondence holds or not? Can we specify the criteria for the truth of a metaphorical statement? If this were impossible, metaphors could not be true and cognitive and only express subjective views and perspectives.

In order to avoid a subjectivist account of metaphorical language, Ricoeur has developed a theory of ‘metaphorical truth’, according to which a metaphorical statement is at the same time literally false and metaphorically true. He argues that the split and tension between literal and metaphorical meaning and reference in a metaphorical statement are also important for its truth: it is a tensional truth. ‘Bill is a wolf’ is literally false, but can nevertheless be true metaphorically. In that case, Bill is a wolf in a metaphorical way: he is like a wolf. We cannot separate the metaphorical ‘is’ from the literal ‘is not’; because both remain present and active in the statement, there is tension between them. The ‘is’ poses an identity, the ‘is not’ a difference; the result is ‘is like’, or ‘being-as’.Footnote ³⁸ This means that a metaphorical statement is true in a different way than a literal one. A metaphorical statement requires a special concept of metaphorical truth. It is not true because it corresponds to reality, but because being manifests or discloses itself in such a statement. The idea of truth as manifestation of being has been taken from Heidegger.Footnote ³⁹ According to Heidegger, the primordial event is not that we talk about reality, but that being comes into speech. Descriptive language, which classifies parts of reality according to literal concepts, is derivative. Heidegger thinks that being itself must be a kind of event because it manifests itself in language. Thus, Heidegger's concept of truth postulates ontologically another kind of being, prior to the objective reality we talk about.

This concept of metaphorical truth is problematic in three respects. First, it presupposes Heidegger's notion of being, which is easily construed as a quasi-divine ultimate reality. Second, it is based on the theory of split reference, which, as we have seen, is only useful for the special class of surprising, creative metaphors. However, the reference of most metaphors can be explained by single reference. Third, if a metaphorical statement is a manifestation of being, it is always true; its meaning is its truth.Footnote ⁴⁰ Nevertheless, it is possible to understand a metaphor and think it is false.Footnote ⁴¹ So the replacement of the concept of correspondence by the concept of manifestation offers no solution.

According to William Alston, the concept of truth as correspondence can be maintained for the propositional content of a metaphorical statement, which can only be formulated in literal terms. When the truth-conditions of ‘the chairman ploughed through the discussion’ are specified, they are described literally: the chairman went slowly, he got tired, etc. In other words, only reducible metaphors can be true or false. This claim is based on the distinction between language and speech: there are no literal or metaphorical meanings in a language; there is only literal or metaphorical use of words in speech. Metaphor is about how we say something, not about what we say.Footnote ⁴² However, in actual speech, meaning and use cannot be totally separated because, as I have said before, the meaning of a particular sentence is determined both by the meaning and by the use of words in a specific context. Moreover, Alston seems to presuppose that we can always have direct access to the reality which makes a metaphorical statement true apart from the metaphorical description itself. But this is not true in the case of non-observable physical reality, as Soskice has pointed out.Footnote ⁴³

Alston argues that a non-reducible metaphorical statement can only be true in the very general sense that something is in some respect similar to something else. In this sense, all metaphors are true. ‘God is a Father’ is as true as ‘God is a tyrant’ because the only possible truth-claim of both statements is that there is some similarity between God and a human father and between God and a human tyrant.Footnote ⁴⁴ However, in a successful metaphor, a specific metaphorical meaning is created, which excludes other specific metaphorical meanings.

I think Soskice is right when she argues that metaphorical statements can be true or false if and only if they have a specific metaphorical meaning that can be grasped by the members of a language community. The truth-condition of a metaphorical statement is not different from the truth-condition of a literal statement: a statement is true if a state of affairs is as it is described.Footnote ⁴⁵ The truth of a metaphorical statement is a kind of partial correspondence: an analogy between the implicit model of the metaphor and the reality it describes.Footnote ⁴⁶ When we understand the specific metaphorical meaning of a metaphorical statement, we can assess its truth.Footnote ⁴⁷ If people understand ‘Julia is the sun in her family’, and if they know Julia, they can decide whether it is false or not. If it is false, ‘Julia is not the sun in her family’ is true. If the positive metaphorical statement ‘Julia is the sun in her family’ is false, another metaphorical statement may be true, for instance ‘Julia is the blizzard in her family’.Footnote ⁴⁸ Consequently, a metaphorical statement can have a truth-value.

When we have to decide whether a metaphorical statement is true, we must test whether the reality it describes is as the statement describes it. This test cannot always be carried out in the same way. If we test whether ‘Bill is a lion’ is true, we look at Bill's behaviour. Observational data are the criteria for the truth of an empirical statement. This means that the condition and the criterion of truth of an empirical statement are the same. In statements about non-observable physical reality, observation cannot be the criterion because we have no direct access to this reality apart from the models by which we describe it metaphorically. If a theory explains physical processes and predicts physical events, we are entitled to assume that physical reality is, at least in part, as the theory describes it. The theory is made true by physical reality, but the criterion for its truth is its explanatory and predicting power. In this case, the truth-condition and the truth-criterion are different. The issue of the truth-condition is closely related to ontology: what part, kind or aspect of reality is it that makes a statement true? The question of the truth-criterion is epistemological in nature: how can we know whether a statement is true or false? The distinction between condition and criterion of truth is also relevant for the possibility of true metaphorical God-talk.

The truth of metaphorical God-talk

God must be the ultimate truth-condition of God-talk. Metaphorical God-talk is true if God is as he is described by it. But who is ‘God’? How can we fix the referent of God-talk? How can we identify God? When we identify God as an absolutely transcendent reality, we will never be able to know if he really is as we describe him. This is demonstrated by Soskice's identification of God. She rejects the description theory of reference, and follows the causal theory of reference, according to which some descriptive terms, such as ‘gold’, can refer in the same way as proper names; a name can refer because an entity, or a kind of entities, has once been baptised by this name.Footnote ⁴⁹ For Soskice, ‘God’ is the name for the source and cause of all there is.Footnote ⁵⁰ The consequence of this identification is that we can name him, but not really describe him: ‘the apophatic insight that we say nothing of God, but only point towards him, is the basis for the tentative and avowedly inadequate stammerings by which we attempt to speak of God and his acts’.Footnote ⁵¹ The apophatic ‘is-not’ is dominant; the metaphorical description becomes subjective: we cannot describe God as he is.Footnote ⁵² Although Soskice has made a strong case for metaphorical description and the truth of metaphorical statements in general, she cannot allow for cognitive metaphorical God-talk because she identifies God as the absolutely transcendent source and cause of all there is.

However, according to the Christian faith, God is not only transcendent, but has become a part of created reality in Jesus Christ. We can avoid the agnostic corollary of Soskice's analysis if we identify God as the one who has made himself known in the historical reality of Jesus Christ. At this juncture, Jüngel is important because he has grounded his account of metaphorical God-talk on God's coming into this world in Jesus Christ in which he has revealed himself.Footnote ⁵³ In the context of our discussion, two aspects of Jüngel's concept of revelation are important. First, Jüngel understands revelation as an event in which God communicates himself to and with created being; God's coming into the world in Christ is a transfer of being, a metaphorical process which is continued by metaphorical language.Footnote ⁵⁴ Second, Jüngel understands revelation as address. In an address a person makes herself present and known to another person. God addresses us in the life and the cross of Jesus Christ, in Jesus' parables of God's coming kingdom and in the gospel. He does not exist for us apart from this; he exists in the event of his speech; he exists as Word. In God's address, reference takes place, but reference to what God performs by speaking and not to something extra-linguistic.Footnote ⁵⁵

Address is a metaphorical event which takes place in, and by addressing, metaphorical language. According to Jüngel, this mode of speech should sharply be distinguished from naming language by which we describe things. In naming, a subject talks about an object; in addressing, two subjects participate in one language-event. In naming, literal, conceptual, classifying language is dominant; addressing takes place in metaphorical, disclosing, appealing language. The truth of naming is correspondence; the truth of addressing is manifestation. Manifestation is an event in which something new addresses us and lets itself be discovered. In this ‘language-event’ subject and object cannot be isolated. On the ‘objective’ side, a being addresses and lets itself be discovered; on the ‘subjective’ side, a being is addressed and discovers another being. Objectivity and subjectivity are sides of one and the same communicative event. When we discover, in such an event, something that was unknown until then, we talk about it metaphorically. This also applies to God's revelation in Jesus Christ.Footnote ⁵⁶ Knowledge of God is grounded in the experience of being addressed; this experience is the subjective side of an event in which God lets himself be discovered. That it is God whom we discover in this event cannot be proved apart from this experience.Footnote ⁵⁷ It is the presupposition of Christian God-talk. Because God addresses us in a totally new event, his coming into created being, which is itself a metaphorical transfer, we can only talk about him in metaphorical language. In this way, God's coming into our world is continued and completed by God's coming into our language, first in Jesus' parables, and second in our God-talk, modelled on Jesus' parabolic language. Thus, metaphorical God-talk is the continuation of God's metaphorical coming into the world and into language.

Jüngel's account shows that the God who has revealed himself in Jesus Christ is the truth-condition of Christian metaphorical God-talk. However, his concept of truth is manifestation. As I have pointed out before, this concept of truth cannot allow for false metaphors. Whereas Soskice offers a satisfactory account of the concept of truth of metaphorical God-talk as partial correspondence, but not of its truth-condition because God is identified as absolutely transcendent, Jüngel gives a satisfactory account of the addressing and self-disclosing God as the truth-condition of Christian metaphorical God-talk, but not of its truth as correspondence.

Fortunately, the concept of truth as manifestation is not logically implied in the notions of address and self-revelation. Jüngel's concept of manifestation is based on two suppositions. The first one is that God exists in and by his addressing speech, that is, as Word. In a sense, God himself becomes a lingual entity.Footnote ⁵⁸ But the fact that we have no access to God apart from the experience of being addressed does not imply that God only exists in and by this address. Nor does the fact that God is only accessible by speech about his coming into the world imply that he exists only in this speech. That God is a speaking God does not mean that he literally exists, in a literal sense, as Word. When John calls Jesus Christ the ‘Word’ of God, the term ‘word’ does not refer to an element of human language, as Augustine has pointed out.Footnote ⁵⁹ Jüngel's second supposition is that our metaphorical God-talk is a continuation of God's metaphorical coming into the world and into language. Nevertheless, God-talk is a response to God's address which cannot be a direct continuation of it because it is of a different kind. God's address may be self-referential, in that he reveals and communicates himself by it, but our response to God cannot be self-referential or self-revealing in the same way. It is a response and an address in which God sometimes can be described in the third person at the same time, especially in praise: ‘You are good’ becomes ‘He is good’. God-talk is not a direct continuation of God's address, but an extension of our response to it.

We can ground metaphorical God-talk on revelation and maintain a concept of truth as correspondence, when we modify Jüngel's concept of God's self-revelation by removing the assumptions that God exists as Word, and that God-talk is a direct continuation of his coming into the world. The God who addresses us in Jesus Christ is the transcendent creator, whose existence does not depend upon immanent forms of his revelation such as language and history. This transcendent God himself is not totally different from the immanent God for us; otherwise his address would not have been self-revelation. In response to his address we can describe how he is for us and how he is in himself. The truth-condition of metaphorical God-talk is the God who has revealed himself in Christ, and its truth is a partial correspondence with the transcendent creator.

This account can only make sense if we can distinguish true metaphors from false ones. Is there a criterion for the truth of metaphorical God-talk? The truth-condition itself cannot be the truth-criterion, as is the case with empirical statements, because the experience of God's self-revealing address in Jesus Christ, which is the work of the Holy Spirit, is not accessible for sensory perceptions. Nor can the truth-criterion be totally different from the truth-condition as is the case with scientific models: God-talk is not a hypothetical explanation of a postulated transcendent divine reality, nor is God a hypothetical explanation of our world, experiences and practices. Therefore the criterion of metaphorical God-talk cannot be its illuminating, explanatory or puzzle-solving power, as McFague, Soskice and Van Huyssteen have suggested.Footnote ⁶⁰

If God-talk is a response to God's self-revealing address in Jesus Christ, its truth can only be tested by this address. If this address itself is not directly accessible, the only criterion can be something in which it is documented in a reliable way, something that gives an indirect access and a reliable witness to God. For the Christian belief this is scripture. Scripture is the criterion of the truth of metaphorical God-talk. Because scripture as witness is not identical with revelation, this truth-criterion is not the same as the truth-condition. Because scripture is itself an indirect expression of God's address and a human response to it, this truth-criterion is not of a different kind than the truth-condition.

To be sure, this criterion is itself highly diverse and complex and therefore difficult to apply. In order to be used as criterion, the coherence and unity of the biblical witness about God, his acts, relations and properties need to be made explicit and defined. Part of this task is to determine the meaning and implications of indispensable biblical root-metaphors as ‘Father’ and ‘King’, which, in the context of the witness of scripture as a whole, give access to the God of Israel.Footnote ⁶¹ In this way, we might be able to establish criteria by which it can be tested whether new metaphorical predications of God such as ‘God is an elephant’ and ‘God is the sea in which I drown’ are true.

The result of these reflections is a coherent theological account of the truth of Christian metaphorical God-talk: Christian metaphorical God-talk can correspond to the reality of God because its truth-condition is the self-revealing address of the transcendent creator, and because its truth-criterion is biblical God-talk.Footnote ⁶²