Must We Know What We Say?

  • Published on

  • View

  • Download


  • Must We Know What We Say?Author(s): Leon Andrew ImmermanReviewed work(s):Source: Religious Studies, Vol. 15, No. 3 (Sep., 1979), pp. 265-280Published by: Cambridge University PressStable URL: .Accessed: 12/09/2012 19:17

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .


    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact


    Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to ReligiousStudies.

  • Rel. Stud. 15, pp. 265-280

    LEON ANDREW IMMERMAN Professor of Religion, Princeton University


    'You cannot satisfactorily explain the meaning of your religious utterances; if you cannot explain the meaning, you don't know it; and if you, the speaker, don't know it, how could it exist?'

    So runs, in schematic form, the typical challenge to the meaningfulness of religious language. The assumption that, on pain of uttering nonsense, we

    must know what we are saying, is so deeply ingrained that more often than not it is left tacit. Yet it is an assumption which religious thinkers may well want to deny, on the grounds that the meaning of religious language must - like God himself- remain partially hidden to us, at least during our sojourn in this world. To confess in this manner our inability fully to comprehend religious language is not to deny - is, in fact, to presuppose - that it has a

    meaning (of which we are incompletely aware). An unknown meaning is a meaning all the same.

    I believe that until recently the philosophical underpinnings of such a religious response have been lacking. I hope to show that recent work on

    meaning and reference has made available a foundation for the logical possibility of this response.

    The main thesis of this paper is that religious language might have more meaning than typical speakers are aware of. A subsidiary thesis is that this partial ignorance would not necessarily interfere with the fundamental purposes for which religious language is used. Section I presents a general thesis in the philosophy of language which, in Section ii, we apply to the special case of talk about God. Section iII suggests that there is something about religion which makes the abstract possibilities imagined in Section II worth taking seriously as an account of the actual use of religious language.


    It is often thought that the meaning of a predicate is something which determines reference or extension.' If two predicates have the same meaning, they have the same extension. 'Bachelor' and 'unmarried adult male' are the stock examples of synonymous terms. Difference in extension implies difference in meaning. How would one argue against the synonymity of

    ' The views expressed in this section derive from the work of Saul Kripke and Hilary Putnam, and others. See Kripke, 'Naming and Necessity', Semantics of Natural Language, ed. Donald Davidson and Gilbert Harman (Dordrecht: D. Reidel Publishing Co., I972), pp. 3I4-23; Putnam, 'Meaning and Reference', Naming, Necessity, and Natural Kinds, ed. Stephen P. Schwartz (Ithaca: Cornell

    University Press, 1977), pp. I I9-32; Putnam, 'The Meaning of "Meaning"', Language, Mind, and

    0034-4125/79/2828-2620 $01.50 ? 1979 Cambridge University Press I0 RES 15

  • 266 LEON ANDREW IMMERMAN 'bachelor' with 'unmarried adult male'? Most likely by showing that the class of objects falling under one description differs from the class picked out by the other - that is, by demonstrating that there are bachelors who are not unmarried adult males, or unmarried adult males who cannot rightly be called bachelors. We think it quite clear that since 'book' and 'dog' are true of different sets of objects, the terms do not have the same meaning.

    Another commonly held assumption is that meaning is somehow 'in the head' of the individual using the terms. 'Bachelor' and 'unmarried adult male' have the same meaning for me because my concept of the two is the same. The divergence in meaning between 'book' and 'dog' (as I use these terms) is grounded in my knowledge of the difference between books and dogs.'

    These two assumptions, coupled with religious doctrines stressing God's radical otherness, make plausible the conclusion that religious language is somehow deficient in meaning. If God is totally transcendent, it is hard to see how a finite and fallen human being could so conceptualize his goodness,

    wisdom, power, etc., that the terms expressing these properties would be true of God. But if God's properties cannot be adequately conceptualized, can an extension-determining meaning be in our heads? And if meanings, if they are anywhere, are in our heads, how can there be any meaning at all to the divine predicates?

    Hilary Putnam has attempted to show that one or the other of these assumptions about meaning has to be abandoned: there is no consistent concept which satisfies both. Like Putnam, I prefer to abandon the assump tion that meanings are in the head, and to retain the notion that meaning determines extension.2 However, the crucial point for me is that at least one of the assumptions is false; it matters little which of the two is renounced.

    Our argument will be that the divine predicates can have a meaning even if we have no concept which determines that God is in the extension of the predicates.

    Let us imagine the case of an individual, Jones, whose idiolect includes the terms 'elm' and 'beech'. With both terms Jones associates the same concept.

    Knowledge: Minnesota Studies in the Philosophy of Science, volume viI, ed. Keith Gunderson (Minneapolis: The University of Minnesota Press, 1975), pp. 131-93. The Schwartz anthology (Naming, Necessity, and Natural Kinds) contains a good bibliography. Kripke and Putnam differ on

    many points, and my treatment of the topic in some respects diverges from either of theirs. However, it is too cumbersome to indicate all points of agreement and disagreement.

    1 The assumption that meaning is in the head has come under concerted attack by philosophers in the Wittgensteinian tradition. But it should be clear that the attack developed in this paper is along quite different lines from the Wittgensteinian ones. A few contrasts between Putnam and

    Wittgensteinians are drawn by Robert Hollinger, 'Natural Kinds, Family Resemblances, and Conceptual Change', The Personalist 55 (I974), pp. 323-33. See also Eddy Zemach, 'Putnam's Theory On the Reference of Substance Terms', The Journal of Philosophy 73 (1976), pp. 124-27. Zemach believes that Putnam's hypothesis of the 'division of linguistic labor' is not a challenge to

    Wittgensteinian views of reference. 2 See Putnam, 'Meaning and Reference', op. cit. pp. I I9-24, and 'The Meaning of "Meaning"',

    op. cit. pp. 134-44.


    Jones believes each term to denote a species of deciduous tree. Indeed, he believes that the species so denoted are distinct from one another. However, he knows nothing which would enable him to make the distinction. Our claim is that the poverty ofJones' individual conceptual resources is consistent with his using the terms to refer to distinct species, provided he receives a sort of linguistic public assistance.

    Suppose that Jones has acquired the term 'elm' by reading in a newspaper that many Minneapolis elms are dying of Dutch elm disease. Remarks like 'it's a pity about the Minneapolis elms' are soon on his lips. His ability to refer to elms, and not to refer simultaneously to beeches, is borrowed from the newspaper reporter.

    In the ordinary course of events, one acquires the term 'elm' from a speaker who possesses the ability to say something to the effect of: 'Elms are

    whatever have the same nature as that', where, upon uttering 'that' the speaker ostensively picks out an elm. (Non-circularity is best served by the assumption that the present reader and writer both associate differentiating concepts with 'elm' and 'beech'.) The speaker from whom one learns 'elm' stands in an appropriate sort of relation to a particular elm, and intends 'elm' to refer to anything of the same natural kind as that elm.

    If the person from whom one learned 'elm' did not enjoy such a relation to elms, then perhaps it was the person from whom he learned the term, or the person from whom the person from whom... The point is that ultimately in the social-linguistic chain will be found a person who can justly claim such a relation to elms. Jones borrows the term 'elm' from whoever occupies this link in the social-linguistic chain leading to his own use of the term.

    And when he borrows the term, he uses it with the same reference as did the original 'owner'. Persons occupying the pivotal positions in the language community for certain words I will call 'guarantors' of the meaning of those words.

    Note that there might conceivably be different persons serving as guaran tors of ' elm' and 'beech' for Jones. In principle, there need be no one in the community who associates differentiating concepts with 'elm' and 'beech'. (Suppose that through some bizarre fluke Jones is the first person in the community to have acquired both 'elm' and 'beech'; everyone else has at

    most one of these terms in his vocabulary.) Indeed, the entire notion of 'differentiating concepts' may be misplaced

    here. If 'elm' is a natural kind term, it is at least as important that the guarantor stand in a real, nonconceptual relation to elms as that he have accurate concepts of elms. (The notion of a natural kind term is explained somewhat more fully below.)

    Jones does realize that 'elm' refers to elms, whereas 'beech' refers to those other sorts of things, i.e., beeches. So Jones does associate a different concept with each word, doesn't he? The problem is whether this difference



    in concept, such as it is, can account for the difference in reference. Jones tries to explain the difference in reference: 'I use the world "elm", unlike the word "beech", to refer to elms; and it is part of my concept of elms that they, in contrast to beeches, are referred to by "elm".'

    But what does Jones mean by the italicized word 'elms'? His only recourse, I think, is to urge that elms are what 'elm' refers to. And this seems to make his attempted differentiation between 'elm' and 'beech' transparently circular. It is as if he had said: 'The word " elm ", unlike the word " beech", refers to whatever the word "elm" refers to.' But without an independent account of what 'elm' refers to, there is no assurance that 'whatever the

    word "elm" refers to' does not include beeches. No assurance within the concept, that is. What is needed to guarantee that 'elm' has no beeches in its extension is the presence of appropriate social links.

    But perhaps Jones has other conceptual distinctions up his sleeve. Maybe he associates with 'elm' the concept: kind of tree that is dying in Minneapolis of Dutch elm disease, a kind distinct from beeches, and which I read about in a newspaper. But imagine Jones to have forgotten that the elms were in

    Minneapolis, that he read about them in a newspaper, and that it was Dutch elm disease killing them. In that case, Jones might still want to say

    ' elms are dying somewhere', but he would not have a distinguishing concept for elms - Jones surely believes that beeches are dying somewhere too. It seems to me that what makes Jones' utterance about elms, not beeches, is that the utterance was prompted by the half-forgotten article on elms.

    But this suggests another trick for Jones: ' Elms are trees of the kind which I heard was dying; and hearing this now prompts my use of "elm".' If we allow this much to Jones, he does have a concept to distinguish elms from beeches. But two points need to be made.

    First, for a concept of this sort to do the trick, Jones must be related to other members of his community, members who can make 'primary' references to elms. It is a conceptual difference that differentiates only on condition that Jones stand in an appropriate relation to the person from whom he received the term 'elm', and that this person stood in an appro priate relation to elms (or stood in an appropriate relation to someone who stood in an appropriate relation to elms, or . . .

    Second, if conceptual differences of this sort can suffice to determine divergent extensions for the associated words, difficulties surrounding the divine predicates can be cleared up in a similar manner. Our problem was

    that it seemed impossible, given a certain theological picture, for believers to have adequately extension-determining concepts associated with the divine predicates. The core idea in Section ii below will be that possibly, e.g., 'good', applies to God in the sense in which it does because the believer stands in an appropriate relation to God and intends to use the word in the sense in which God understands it. God perhaps needs to know the meaning


    of the terms, or to stand -in some 'primary referring' relation to himself, but the believer does not. If concepts like 'the property God associates with the word "good" when this word is applied to him' figure into the believer's conceptual resources, then there should be no problem in conceptualizing the sense in which 'good' is true of God. But such a concept, to do its referential or denotative job, presupposes what Putnam calls a 'division of linguistic labor' - in our case, a division of linguistic labour between believer and God. The believer presumably has an overriding intention to use the terms in the sense God would find correct, an intention that would overrule his intention to use the term in the sense he himself would find correct.

    Many believers would of course hold that their own understanding of the sense corresponded to the sense understood by God. My point is that a non idolatrous believer would be willing to let his own understanding be overruled by God's, in case of a conflict between the two.

    'Elm' and 'beech' seem to be natural kind terms; and the hypothesis I have been sketching has developed out of a consideration of such terms. But with some modifications the same considerations apply to what might be dubbed 'descriptive predicates'. This is important since it is unclear whether theological predicates are closer to natural kind or to descriptive terms.

    By 'natural kind term' I mean essentially what Paul Teller has in mind by 'special general term'. Teller, however, reserves the expression 'natural kind term' for a subset of the special general terms:

    There are respects in which certain general terms such as 'tiger', 'lemon', 'star', 'electron', 'length', and 'temperature' are strikingly like proper names. Having introduced by example the type of term in which I am interested, I will loosely refer to them as special general terms. By and large, they include what others have referred to as natural kind terms ('emerald'), species terms ('tiger'), substance terms ('gold', 'water') and terms of quantity ('temperature'). As many have argued, special general terms do not have analytic definitions just as proper names are not disguised definite descriptions. Kripke has extended these arguments to persuasively support the claim that, just as proper names do not have a meaning analysable in terms of some 'cluster' of characteristics, special general terms cannot be given a 'cluster concept' analysis.'

    A very rough account of the natural kind/descriptive distinction is this. Suppose we were to arrive at the belief that most of the ideas we had been associating with 'p' were false. If 'p' is a natural kind term, we would - consulting our deepest intuitions - be inclined to say that p exists (or that p's exist) but that we had previously held radically mistaken ideas about it (or them). Should we undergo a similarly sweeping alteration of the concepts

    we associate with a descriptive predicate 'p' our tendency, provided we grasp the meaning of the term, would be to deny that there even are any p's.

    I Paul Teller, 'Indicative Introduction', Philosophical Studies 3I (I977), p. I74.


    What if we came to the conclusion that those fuzzy-seeming creatures known as 'cats' were in reality robots operated from Mars? We ought then to allow that our former opinions had been radically faulty - our former opinions about cats, that is. We would not, I think, relinquish our belief in the existence of cats. This suggests that 'cat' is a natural kind predicate.

    It might be argued that 'cat' is a descriptive term which in the situation just described is merely being used to refer to objects which are not among its semantic referents. Perhaps - unlikely as it seems - it is analytically true that all cats are animals. If so, the robots could not be cats (unless of course

    we were mistaken in thinking that robots can't be animals). However, we could still use the word 'cat' to refer to these objects.' For our purposes it is of little importance whether any given terms actually are natural kind predicates; the more important issue is whether they can be used this way. And since we are not committed to taking the divine predicates as natural kind terms, it is not even crucial that there be terms which can be used as natural kind terms.

    Now suppose it one day dawned on us that those cats we term 'fat' actually weighed quite little, had small amounts of adipose tissue, fit easily into small spaces, etc. It would then be natural to surrender our belief in the existence of fat cats. What gives terms like 'fat' their extension or reference is associated descriptive content, rather than (as with natural kind terms) the relations between speakers and representatives of the kind.

    However, communal determination of extension applies here in roughly the same way that it applies to natural kind terms. The transmission of the terms from one speaker to another is carried out in the same way, but the guarantor of meaning possesses a different sort of ability. The guarantor's key to success is not the ability to say 'things of that nature are p'. Rather, the requisite talent is competence to describe what it is for an object to be p (or to be a p).

    The guarantor is the one who decides if we should drop our commitment to the existence ofp's when our concepts change drastically. In the case of 'fat'

    we are virtually all guarantors, since knowledge of the term's meaning is nearly universal among speakers of English. But suppose Smith is ignorant of the meaning held by the slightly more esoteric term 'corpulent'. Davis tells Smith: 'Tabby is a corpuJent cat'. When asked to describe Tabby, Smith repeats the sentence, believing it to express a truth. It seems to me that by virtue of Davis' knowledge of the meaning of 'corpulent', Smith's relation to Davis, Smith's intention to use the term in the Davisean sense, and perhaps Smith's own (admittedly weak) command of English, Smith's tokens of 'corpulent' have the standard, Davisean reference.

    1 For relevant discussion see Jerrold J. Katz, 'Logic and Language: An Examination of Recent Criticisms of Intensionalism', Language, Mind, and Knowledge: Minnesota Studies in the Philosophy of Science, volume vii, ed. Keith Gunderson (Minneapolis: The University of Minnesota Press, I975), pp. 97-8.


    As I said, the natural kind/descriptive distinction is crudely drawn. The lines dividing one from the other are not completely clear, and perhaps cannot be made clear without distorting the facts. It may be that there is often no precise answer to the question: how wrong are we allowed to be about ostensible p's before we give up our belief in their existence? And if descriptive terms can be used as if they were natural kind terms, the lines become even more hazy. But fortunately for us, it is relatively unimportant how the lines are to be drawn. For the focus of our interest is the relation of the believer to the guarantor, not the ability possessed by the guarantor himself. The relation of guarantor to the other believers seems to be the same whether the terms are descriptive or natural kind. I hope this will become clearer as we consider the applications of the thesis of communally determined extension to talk about God.

    Note: the fact that God is a member of no natural kind is irrelevant to the question of whether theological predicates function as natural kind terms. God is the sole member of his own kind, a supernatural kind let us say. Anything that is of the same kind as God is, necessarily, God. Of course there are many things that have the same general nature as any particular elm (all other elms, I think). But even if there had been only one elm, there

    might have been more. God's nature or supernature is such that there can be at most one representative of it. But this need not affect the linguistic status of the divine predicates; it simply means that there are some natural or supernatural kind terms which necessarily have no more than one object in their extension. It may be that such natural kind terms would in reality be proper names. But this only shows that some natural kind terms are proper names, not that they are not really natural kind terms. Moreover, even if we decided that no natural kind term could be a proper name, we

    might simply speak of the possibility that the divine predicates are proper names (instead of speaking of the possibility of their being natural or super natural kind terms). The communal determination of extension holds as much for proper names as it does for natural kind terms.


    The conclusion we hope for is that terms such as 'good', 'person', 'powerful', etc., as used by ordinary believers, possibly have God in their extension, even if the concepts associated with the words by the believers fail to deter

    mine that God falls within the extension. It has often been assumed that the concept a speaker associates with a term determines what, if anything, the speaker's tokens of a term have in their extension. By denying this assump tion, we can also deny that believers must adequately conceive the meaning of the predicates in order for the predicates to refer to or be true of God.

    Most of what the believer believes can be false; perhaps he need believe no descriptions which are uniquely satisfied by God, at least none which he


    fully comprehends. Meaning cannot both be in the head and determine reference. Though I myself find it plausible to assume that meaning deter

    mines reference, and hence that meaning is not in the head, the alternative is coherent. One might argue, against my preference, that when Jones associates the same concepts with 'elm and 'beech' the two terms have the same meaning in his idiolect; contextual factors, such as Jones' relation to other members of his community, determine the difference in extension despite the sameness in meaning. (Compare: 'that' means the same when ever it is spoken non-deviantly; but of course its reference varies with context.) But whichever of the two assumptions (meaning is in the head,

    meaning determines extension) one gives up, the crucial point remains: the point that reference or extension can be determined communally.

    A guarantor of meaning for the divine predicates is called for. If the predicates are descriptive the guarantor must understand the meaning of the predicates. If they are natural kind terms, he must stand in an appropriate real relation to a representative of the kind (that is, to God - the sole repre sentative of his kind). In either case, the guarantor must be linked to ordinary believers, and perhaps believers must intend to use the predicates in the guarantor's sense.

    A number of candidates for guarantor or possible guarantor exist. Most salient are the community's religious specialists: the priests and the prophets.' The priests, who need undergo no exceptional personal experience of God, receive special training in the sacred tradition. This is in contrast to the prophet, whose authority is derived from a claim to 'personal revelation and charisma'.

    Specialists of both varieties can be found in many religious communities, and social links between them and the laymen are straightforwardly speci fiable. The straightforwardness of these links is the advantage of employing guarantors from these categories. But the advantage seems to come dear, for these would-be guarantors are so much a part of their communities that they seem to share in the same fallenness or finitude that renders the ordinary believer unfit to understand the divine predicates.

    The prophet shows more promise than the priest, however; for the prophet is allegedly vouchsafed a special personal relation to God. As one would expect, his social role within the visible community tends to be more peripheral than that of the priest. But he is nevertheless clearly a member of the visible religious language community. Yet, at least if the predicates are descriptive, it seems that the prophet is more a mediator than a primary guarantor. It is not the prophet's own understanding that guarantees

    1 For Max Weber, 'the personal call is the decisive element in distinguishing the prophet from the priest. The latter lays claim to authority by virtue of his service in a sacred tradition, while the

    prophet's claim is based on personal revelation and charisma. It is no accident that almost no prophets have emerged from the priestly class.' The Sociology of Religion, trans. Ephraim Fischoff (Boston: The Beacon Press, I964), p. 46.


    meaningfulness, but God's understanding and the prophet's relation to God. This suggests that the ultimate guarantor is God himself, not the prophet. On, the other hand, if the predicates are natural kind terms, the prophet would not need adequately to conceptualize the properties referred to; he would merely need to stand in the appropriate relation to God. (The notion of appropriateness will be discussed more fully below.)

    To choose God himself as guarantor is to raise problems which are in a sense the converse of those facing the choice of the all-too-human priests and prophets. If priests and prophets unquestionably stand within the human language community, they are perhaps too limited to conceive or even to be properly related to God's goodness, wisdom, power, arid so on. If God surely possesses the capacity to know himself fully, it is less clear that he can be within any human community. Worse still, sentences like 'God understands the meaning of the terms predicated of him' themselves employ predicates (e.g., 'understands') whose meaningfulness God is supposed to guarantee.

    Four strategies come to mind for avoiding the problems with assigning to God the role of guarantor. The first strategy involves a vicious regress, though not a clearly vicious one. What insures that 'knows' has meaning in an utterance of 'God knows the meaning of the divine predicates'? It is, according to this strategy, God's knowledge of the meaning which 'knows' has in this utterance. And God knows the meaning of 'knows' in 'God knows the meaning of "knows" in "God knows the meaning of the theological predicates".' And so on to infinity.

    God knows if this strategy is incoherent. It is in any case not particularly likely. The second strategy is more promising only if we are willing and able to sacrifice some of the power which our hypothesis about predication seemed to lend to the defender of religious language's possible meaningful ness. Under this strategy, we choose a few privileged predicates whose meaning we can secure without God's assistance; and, building on this handful, we manage to have God insure for us the meaningfulness of the rest. For example, if we know what it means to say 'God knows the meaning of such and such predicates' (where such and such predicates do not include 'knows' or 'knows the meaning of. . .'), and we know what it means to say 'God is within our language community', then perhaps we can without

    circularity employ God as guarantor of such predicates as 'powerful', 'good', or 'personal'.

    The main problem here is to winnow out those chosen few predicates whose meaningfulness we human beings can insure for ourselves. It is hard to see how the meaning of 'knows', for example, could be more compre hensible than that of 'powerful' (as these terms apply to God). This is especially true if God's simplicity entails that the divine power is identical to the divine knowledge.


    The third strategy also involves certain sacrifices. But the sacrifices here are less onerous, I think, and they produce greater success. Strategy three has us give up the idea that the divine predicates might be descriptive; we assume instead that they are all of the natural kind variety. This means that God does not strictly speaking need to know the meaning of the predi cates; all he needs is to associate the term with the extension (i.e., with himself) and to stand in an appropriate relation to himself and to some believers.

    My feeling is that what is given up in strategy three is not particularly valuable. Though, as I said, it is unclear whether the divine predicates are closer to natural kind or to descriptive predicates, the natural kind side

    may be a slightly better bet. Natural kind terms are more basic or 'primitive' than descriptive ones in that their reference is not determined by other predicates which they somehow abbreviate. If 'fat' is descriptive, it is a necessary truth that fat things have most (or at least many) of the qualities we think that they, by virtue of being fat, must have. These other qualities, other than fatness itself, might include such things as weighing a great deal, fitting only into large clothes, and so on. Terms expressing these other qualities are needed to fix the extension of 'fat'. Natural kind terms, however, are like proper names in having no meaning apart from reference; they do not abbreviate sets of descriptions, even in a loose sense of 'abbre viate'. This basicness or primitiveness is what one would expect to find in the divine predicates. For if God is simple none of his intrinsic qualities can be explained in terms of any other qualities, and a term expressing one quality could not rely for its reference on terms expressing other qualities. God has only one property, it seems to me, if he is simple.

    Another relevant consideration is that natural languages seem to have relatively few descriptive predicates. If so, it may be better to assume that a predicate is not descriptive unless one has a special reason to think that it is. The topic of natural kind terms is enormously complex, and the application of any position on natural kind terms to the question of the divine predicates adds still more complexity. Though I suspect that the hypothesis that the divine predicates are (or are very much like) natural kind terms is true, it would be premature to think that the above discussion establishes this hypothesis.

    It is likely to be objected that, even if the divine predicates can safely be assumed to be natural kind terms, strategy three suffers from an intolerable vagueness. What is all this talk about 'appropriateness'? What makes one relation more appropriate than another, and how do we tell whether a relation is appropriate? To the best of my knowledge, no one has yet produced an intuitively satisfying set of necessary and sufficient conditions for the appropriateness of connections between guarantor and ordinary speaker. The usual approach is anecdotal and unsystematic: one tells a story


    about how speaker and guarantor are related, and about how guarantor is related to the alleged extension, and consults one's intuitions as to whether such a connection can serve the purposes of communal determination of extension.

    If, as I think, the ultimate court of appeal on 'appropriateness' is the linguistic intuitions of the speakers involved, the vagueness of our position is not particularly troublesome. The important question would be: would believers regard the relation between God and the religious community as the sort which permits a communal determination of extension, with God guaranteeing the meaningfulness of some predicates? If believers - even believers in God's unutterably radical transcendence - would tend to say yes to this question, strategy three seems extremely viable despite its vagueness. The intuitions of the speakers have a privileged status not shared by any set of necessary and sufficient conditions for appropriateness.

    - If God is appropriately related to certain members of the religious language community, then those members are appropriately related to him. If only a handful of believers (namely, the prophets) are in this relation, and the other believers have the meaningfulness of their talk guaranteed by the prophets, the situation is one in which God is the ultimate guarantor and the prophets are linguistic mediators.

    The fourth strategy I want to mention attempts to combine the advantages of employing prophets with those of employing God. This strategy, unlike the third, would allow the predicates to be either descriptive or natural kind terms. The basis for strategy four is the observation that in many religious communities it is thought that certain unassailably visible members are identical to God. A prophet identical to God would, it seems, be in an excellent position to understand the divine predicates and to be appro priately related to God.

    One problem with this line of thought is: do we understand the sense in which, e.g., Jesus is said to be or to be identical to God? Now, I do not think it is clear how such an identity is possible. Yet it strikes me that the sense of the statement is clear enough. On the face of it, it is an identity statement involving two proper names; there should not be any problem about the sense of the statement, certainly not if there is no problem about the reference of the names. To get strategy four working, we need only an

    understanding of the sense of the identity statements; our knowledge of the nature of such identities can remain pitifully inadequate. Another problem with strategy four, and one which I cannot resolve, is that it is unavailable for communities which lack the belief that one or more of their members is identical to God. However, the motif of identity to God is remarkably

    widespread, so the problem may be less serious than it seems. (The above remarks on identity are consistent with but do not presuppose

    the widely held view that identity statements are not predications. These


    remarks would presuppose that identity statements are not predications only on the assumption that proper names are not predicates. But it seems consistent to hold, with Michael Lockwood, both that proper names are predicables and that proper names have no meaning 'over and above' their denoting the objects which they denote.)1


    As an exercise in odd possibilities the preceding discussion might have some interest. But I want to argue that our application of the communal predi cation hypothesis to religious language is more than a fanciful thought experiment. It seems to me that our position is illuminating as a description of how religious speakers actually understand themselves to be using their religious language.

    I would locate the position of this paper squarely within an ancient tradition of interpreting religious language, a tradition whose insights are too often ignored by analytic philosophers. Those who consider religious language somewhat mysterious in meaning are not necessarily under the sway of positivistic superstition. Many of religion's most acute interpreters have found religious language to be rife with 'hidden meanings'. As Peter Brown suggests, figures as diverse as Augustine and Freud are united in this perception of religion:

    For Augustine and his hearers, the Bible was literally the 'word' of God. It was regarded as a single communication, a single message in an intricate code, and not as an exceedingly heterogeneous collection of separate books. Above all, it was a communication that was intrinsically so far above the pitch of human minds, that to be made available to our senses at all, this 'Word' had to be communicated by an intricate game of signs... Once it is thought possible for something larger than our conscious awareness to be capable of active communi cation, whether this be the 'whole' personality, conscious and unconscious of the

    modern psychoanalyst, or the ineffable 'Word' of the Early Christian Exegete, an attitude similar to that of Augustine occurs quite naturally.2

    For many religious thinkers, the Bible is a transcription by humans of God's words. Though human amanuenses may have distorted the message by filtering it through their own minds, fundamentally, to understand the sacred text is not to know what the human writers had in mind. It is, rather, to know what God had in mind. Whether the text must remain forever partially inaccessible to humans this side of the beatific vision depends on the extent of human potentials. Many thinkers have held a very low opinion of these potentials. But in any case the partial mysterious

    1 See Michael Lockwood, 'On Predicating Proper Names', The Philosophical Review 84 (I975), P. 495.

    2 Peter Brown, Augustine of Hippo: A Biography (Berkeley: University of California Press, I967), pp. 252-3.


    ness of the language to many speakers would be insisted on by a large number of authors.

    More familiar to some readers are the social-scientific versions of the 'hidden meaning' theory; for example, the Freudian views to which Peter

    Brown alludes. According to Freud,

    the truths contained in religious doctrines are after all so distorted and systemati cally disguised that the mass of humanity cannot recognize them as truth. The case is similar to what happens when we tell a child that new-born babies are brought by the stork. Here, too, we are telling the truth in symbolic clothing, for we know what the large bird signifies. But the child does not know it.

    He hears only the distorted part of what we say, and feels that he has been deceived.1

    This remarkably suggestive passage admits of a number of interpretations. One reading is that sentences like 'Babies are brought by the stork' are ambiguous, but the child (unlike the adult) fails to perceive the double

    meaning. Failing to perceive the double meaning, the child would express only the one meaning were he to utter the sentence. While the sentence can be used to express this other meaning, it does not when spoken by the child.

    This reading is suggested by the passage but is probably untrue to Freud's intentions. Ordinary believers are, for Freud, like the child in being unaware of the less obvious construals of their doctrines - the construals on which, to Freud's mind, the doctrines might come out true. Yet I think Freud would want to say that the religious doctrines espoused by ordinary believers do express these true propositions which are hidden from the conscious minds of the speakers. I think Freud most likely wants to say that religious language expresses notions of which the speaker is, typically, unaware.

    On one level, the passage from Freud seems to credit the psychoanalyst with the ability to explain the meanings concealed in religious language. But the analogy between the adult (parent) and the child suggests that at a deeper, perhaps unconscious level, Freud meant God the father as the individual aware of the hidden meanings. (Recall the Freudian insistance that God is father writ large.)

    It is irrelevant for us whether it is thought that the religious speaker has unconscious knowledge of meaning, or whether he has no knowledge at all. If the Bible represents God's words, why must the prophets have understood all they said when the divine afflatus was upon them? This point is perhaps obscured by a psychoanalytic view according to which the prophets would have at least unconscious knowledge of meaning. But the point is clarified by taking a sociological analogue to the psychoanalytic position. To maintain, for example, that by 'God' the believer somehow refers to society is not

    1 Sigmund Freud, The Future of an Illusion, trans. W. D. Robson-Scott (Garden City, N.Y.: Doubleday and Company, Inc., I96I), p. 73.

  • 278 LEON ANDREW IMMERMAN necessarily to attribute to the believer even an unconscious knowledge of this fact. It might merely be to point to the function which talk about God has in the life of the individual and the community, a function unperceived by even the unconscious minds of the believers.

    It would be misguided for a philosopher of language to argue that if certain psychologists are right, 'God is our heavenly father' means 'My childish sense of helplessness in the face of an indifferent universe is assuaged by attributing to a cosmic force the qualities of a super-father'. This sort of 'meaning' seems well-founded in ordinary language (e.g., 'What do you really mean by that?') but its direct interest to semanticists is doubtful. A distinction needs to be made between what a person means by uttering certain words, and what the words themselves mean. Much of what, in the broader sense, is meant is not linguistic meaning. (A simple illustration: when Jim explains a theory to his audience, he means for them to think him a clever fellow. However, nothing he says actually means what is expressed by the sentence 'I am a clever fellow'. Indeed, were Jim to utter these words his audience would be less likely to consider him a clever fellow, and Jim knows this.)

    Yet when we speak of God possibly guaranteeing the meaningfulness of the terms predicated of him, the meaningfulness we are concerned with should be more a matter of what the speaker is perhaps using his words to express than of what the words themselves mean. While I do not propose any translation of religious utterances into what they 'really' mean, the sense of meaning in which one might say 'what you really mean is. . .' is relevant to us.

    - I am not committed to any particular position on what the hidden meanings within religious language are. What I am stressing is that many sensitive observers have held that such meanings are essential to religious language. The reason I feel I must be so insistent about this is that there seems to be a growing prejudice among analytic philosophers that religious language wears its meaning on its sleeve, and that anyone who worries about the meaningfulness of religious language has been victimized by residual verificationism.

    The provision of 'meaning' to life is often cited as a function of religion. The view that religious language is partially mysterious in no way contradicts this insight into the role of religion.

    Clifford Geertz writes that the religious response to the suspicion that the world, and man's life in the world, has no genuine order,

    is not to deny the undeniable - that there are unexplained events, that life hurts, or that rain falls upon the just - but to deny that there are inexplicable events, that life is unendurable, and that justice is a mirage. The principles which constitute the moral order may indeed often elude men, as Lienhardt puts it, in the same way as fully satisfactory explanations of anomalous events or effective

  • MUST WE KNOW WHAT WE SAY? 279 forms for the expression of feeling often elude them. What is important, to a religious man at least, is that this elusiveness be accounted for, that it be not the result of the fact that there are no such principles, explanations, or forms, that life is absurd and the attempt to make moral, intellectual or emotional sense out of experience is bootless.'

    Religious explanations need not so much explain as affirm explicability. An all good, all powerful God perhaps guarantees that no suffering is in vain. But how exactly is my particular suffering redemptive? This question my religion may be unable to answer. God ensures that the suffering 'makes sense'. Though I may not know what sense this is, I have faith that it exists, if I am religious.

    The very incomprehensibility of God sometimes serves to 'make sense' of things: it is all sensible because God, after his ineffable fashion, makes sure that it is.

    Language about God might matter to an individual who was incompletely aware of its meaning. A belief that God's goodness is somehow analogous to human goodness might help the self conform to God, though the sense in which God was good could be seen at best dimly. The faith that God knows the sense in which he himself is good might help the believer to trust that somehow the world makes sense and life is endurable.

    The notion of analogy provides a clue to the sort of partial knowledge of meaning which I think must be attributed to ordinary believers. Whatever one thinks of religious mystery, one must account for the disparity between the believer who cannot quite grasp the sense in which God is good, and the non-English speaker who has no idea whatever even what the ordinary sense of 'good' is. It may be that the English-speaking believer cannot understand the precise sense in which God is good, or that he cannot have God in the extension of his tokens of 'good' in the absence of communal chains. But what the believer might still know is that the sense of 'good' is somehow analogous to the ordinary sense, or that God's goodness is analogous to the property ordinarily denoted by 'good'. This provides an epistemic edge over the non-English speaker, but does not tell the believer how the analogy is supposed to work. Nor does it show that the believer understands

    what respects of resemblance if any exist between God's goodness and its ostensible reflection in the created realm.

    Does this weaken the sense of ' analogy' to the point where we cease to use even that word in its ordinary sense? I doubt it. But if the answer is yes, the problem can be skirted by a slight alteration in our statement of what the believer knows. Instead of attributing to the believer a knowledge that the

    1 Clifford Geertz, 'Religion as a Cultural System', Anthropological Approaches to the Study of Religion, ed. Michael Banton (London: Tavistock Publications, I969), pp. 23-4. The subtle interplay of mystery with explanation is portrayed more astringently by Peter Berger, The Sacred Canopy: Elements of a Sociological Theory of Religion (Garden City, N.Y.: Doubleday and Company, Inc., I 969), p. 90.


    sense is analogous, we can instead hold: the believer knows that it is somehow fitting to regard the sense of 'good', when predicated of God, as analogous to the commonplace sense. If it is unlikely that we shifted the sense of 'analogy' in the paragraph preceding this one, it is still less likely that we

    are now tampering with the sense of 'fitting' or 'somehow fitting'. I have been speaking as if the meaning of the divine predicates would be

    mysterious if the extension of the predicates were determined communally. But if the predicates are natural kind terms this is probably not quite right, since natural kind terms seem not to have any meaning, and hence seem not to have any meaning to be ignorant of. However, natural kind terms have extensions, and there is some analogy between failing to be directly or appropriately related to the extension, or failing to associate the extension

    with the term, on the one hand, and on the other hand failing to understand the meaning of a term. Moreover, it can be mysterious to someone how an extension of a natural kind term is determined, or exactly which extension is determined. (It is fairly clear how to recognize elms, fairly clear which things are elms. But God cannot be pointed to in the way elms can, and it is hard to say whether other things than God are good in the sense in which God is good.)

    One final point concerning natural kind terms. If the divine predicates are natural kind terms, and if each member of the community of faith stands in an appropriate relation to God, the hypothesis of the communal determi nation of extension would become unnecessary for us. The believer need not understand God in order for the natural kind terms he predicates of God to be meaningful (meaningful in the sense of determining extension). And he would not require the services of any guarantor of meaning. God, in the circumstances we are now imagining, is no more a guarantor of meaning than the beech tree is a guarantor of meaning for the person who makes a primary reference to it. God is in the extension of the terms because of the believer's relation to God, independently of any special relation God may bear to himself.

    But even if this were the case, much that is said in this paper would be worthwhile. Whether or not it is useful to take the extension of divine predi cates to be determined communally, our main thesis stands: that terms can be true of God in virtue of the relation of believers to God, and that serious deficiencies in our conceptualizations of God need not interfere with the meaningfulness of the predicates.'

    1 Earlier drafts of this paper were read to the philosophy departments of The University of Connecticut-Storrs (Spring, I977) and The University of Illinois-Urbana (Spring, 1978). I am grateful for the many helpful comments made on those occasions.

    Article Contentsp. [265]p. 266p. 267p. 268p. 269p. 270p. 271p. 272p. 273p. 274p. 275p. 276p. 277p. 278p. 279p. 280

    Issue Table of ContentsReligious Studies, Vol. 15, No. 3 (Sep., 1979), pp. 265-428Must We Know What We Say? [pp. 265-280]Divine Omniprescience: Are Literary Works Eternal Entities? [pp. 281-287]Divine Conservation and Spinozistic Pantheism [pp. 289-302]Divine Omniscience and Human Freedom [pp. 303-316]Price, Hick, and Disembodied Existence [pp. 317-325]The Dimensions of the Self: Buddhi in the "Bhagavad-Gt" and "Psych" in Plotinus [pp. 327-342]Zen and San-Lun Mdhyamika Thought: Exploring the Theoretical Foundation of Zen Teachings and Practices [pp. 343-363]Sri Aurobindo's Integral View of Other Religions [pp. 365-377]Is God Really in History? [pp. 379-390]Divine Independence and the Ontological Argument: A Reply to James M. Humber [pp. 391-397]Does Elusive Becoming in Fact Characterize H. D. Lewis' View of the Mind? Comment on H. D. Lewis' Paper: 'Mind-Body Duality' at the Annual Meeting of the Metaphysical Society of America, Emory University, 18 March 1978 [pp. 399-405]Reply to Professor Bertocci [pp. 407-409]ReviewsReview: untitled [pp. 411-413]Review: untitled [pp. 413-414]Review: untitled [pp. 414-417]Review: untitled [pp. 417-418]Review: untitled [pp. 419-421]Review: untitled [pp. 421-423]Review: untitled [pp. 423-425]Review: untitled [pp. 426-428]

    Back Matter


View more >