Metaphor and Sentence Meaning

Download 74.17 Kb.
Size74.17 Kb.
Metaphor and Sentence Meaning
Facta Philosophica, Vol. 8, 2006
Mark Mercer

Department of Philosophy

Saint Mary’s University

Halifax, NS B3H 3T7

(902) 420-5825

“We are survival machines—robot vehicles blindly programmed to preserve the selfish molecules known as genes.” Reading this sentence might cause us to ask whether people who seek out performances of Steve Reich’s music can be understood thereby to be engaging in behaviour tending to increase the relative frequency with which their genes occur in the general populationCor whether homosexual behaviour, or that of those Serbs in Biha who sheltered Muslims, can be so understood. Suppose that we find that we can understand musical preferences, homosexuality, and heroism each to be ways of increasing the relative frequency of types of genes in a population. Would we then be warranted in concluding that the quoted sentence is true, that we really are survival machinesCrobot vehicles blindly programmed to preserve the selfish molecules known as genes?

Donald Davidson would say that we would not be warranted in concluding that the sentence is true. No human or any other animal lies within the extension of “machine”; “blindly programmed” is an oxymoron; and molecules, because they do not act intentionally, cannot be selfish. Thus, any sentence that implies such sentences as that we are machines or that we were blindly programmed to behave in some fashion, or that molecules are selfish, cannot be true.1

But to respond in this way, one might object, is to take the quoted sentence, from Richard Dawkins’s The Selfish Gene,2 much too literally. Of course, one would concede, if taken literally it is false; but it was not meant to be taken literally. It was, rather, meant to be taken metaphorically. To take it literally is simply to fail to understand it. If, on the other hand, one might continue, we attempt to understand Dawkins’s sentence metaphorically, then it becomes clear that it might well be trueCmetaphorically true, that is. MoreoverCand this is crucialConly if we hold that it could be true can we make sense of the fact that we think our questions about musical preferences, homosexuality, and heroism are to the point. There would be no need to ask substantive questions in order to assign a truth value to a sentence false in virtue merely of its meaning.

Davidson’s response to this objection is that metaphor is a feature of the use of sentences, not of their meaning. Dawkins used his sentence to get us to attend to something, perhaps to get us to ask just the questions we did ask. We have to reflect on more than just the fact that the sentence is false if we want to understand Dawkins, if we want to understand what he meant or intended by penning it; we have to ask, for instance, why he would pen precisely that sentence, with just the literal meaning that it has. But the sentence Dawkins penned is false and remains false. Our reading this false sentence made us attend to particular ideas, ideas such as that acting heroically tends to have the effect of increasing the relative frequency of one=s types of genes in a population. Now, should this idea indeed be true, then our reading Dawkins’s sentence has, perhaps, made us come to appreciate some hidden or novel truth. But that encountering a false sentence can have such an effect shouldn’t surprise us overmuch. After all, gaining an insight into the world on the occasion of hearing a false sentence is precisely what oftentimes happens when we hear someone exclaim, “That’s just wonderful!”

Davidson=s response is, I think, right. Furthermore, on examination, it turns out not to be at odds with most of our common beliefs about the usefulness and cognitive significance of metaphorCwith what we might call, following Max Black, the commonsense of metaphors.3 However, many philosophers, perhaps a majority of those who have considered the issue, do think that Davidson’s views offend against commonsense ideas of metaphor. In what follows, I will examine critically the reasons why two of these philosophers, Max Black and Frank B. Farrell, reject Davidson’s view. I intend to show that their arguments do not tell against Davidson’s position and that, moreover, the alternative understanding of what metaphors mean that they offer is open to an important objection. Other philosophers, some sympathetic with what they understand to be the account of what metaphors mean presented in Davidson’s paper, others unsympathetic, contend that that account is inconsistent with the theory of meaning Davidson has developed in such later papers as “A Nice Derangement of Epitaphs.”4 I will conclude by dispelling the sense that Davidson, in light of his recent work on meaning and interpretation, needs to change his mind on any point he made in the metaphor paper. I begin with a sketch of Davidson’s account of linguistic meaning and the arguments in support of it.

Davidson contends that the meaning of “We are survival machines” is given by the following T-sentence: “‘We are survival machines’ is true if and only if we are survival machines.” Neither the sentence itself nor Dawkins’s utterance of it means anything other than that we are survival machines. Since we are organisms, not machines, it is false that we are survival machines. But in uttering this false sentence, meaning just what it does, Dawkins was, let us assume, attempting to draw our attention to somethingCperhaps to a similarity between humans or organisms generally and machines designed to replicate themselves. At any rate, his utterance did draw our attention to a similarity between the two. Having had our attention thus drawn, we are left with the question whether our attention has thereby been drawn to anything interesting or important or fun. The remainder of Dawkins’s book, if I read it correctly, is given to the attempt to show that it has been drawn to something that is all three.

That the meaning of “We are survival machines” is given by that particular T-sentence follows from the claim that a theory of truth (for the language Dawkins spoke when he uttered that sentence) capable of enabling an interpreter to understand any utterance of that language would entail that T-sentence as a theorem. A theory of truth for that language would find in the sentences of it a finite set of non-logical and logical semantic primitives and a finite set of rules for assembling sentences out of these primitives. Thus the theory would show how the meanings of the innumerable sentences of the language are the products of the semantic and syntactic features of their parts.

The thesis that the meaning of a sentence depends on the meaning and arrangement of the terms in it rests easily with another thesis, one that one might think is even more evident than the first, though, as it turns out, many philosophers find this thesis remarkably difficult to keep in focus. This is the thesis that what a word or sentence literally means, that is, what its truth conditions are, is one thing, while what that word or sentence is used to do on any particular occasion of utterance is something else. That there is a distinction between meaning and use is, perhaps, obvious in the case of “That=s just wonderful” said ironically or “There once was a man from Killarney” used to begin a limerick: the first is true if and only if the thing intended by “that” is just wonderful, though the speaker is using this false sentence to inform us that the thing intended is far from wonderful; the second is true if and only if there once was a man from Killarney, something the speaker probably couldn’t care less about. However, that there is precisely the same distinction between meaning and use in cases where indicative sentences are used by speakers to assert that what they (the sentences) mean is true is often missed. “That=s just wonderful,” used to assert that that’s just wonderful, is still an indicative sentence used in a particular way to accomplish a particular end. Neither that way of use (sincere assertion) nor that end (informing the hearer of the wonderfulness of something, say) is or gives that sentence’s meaning, though we might say of that end that it is what the speaker means.

This point might be put by saying that, for Davidson, the notion of a trope, if this notion involves a contrast between the literal and the figurative drawn within the purview of linguistic meaning, is without substance. All linguistic meaning is literal meaning; the notions of trope and figurative meaning mark classes not within linguistic meaning, but instead mark classes of uses of sentences. These notions distinguish some sorts of use from other sorts of use, from, particularly, the use we might designate sincere or straightforward assertion. Moreover, as Davidson has argued, nothing, no convention, ties indicative sentences to assertions of their content, nothing ties interrogative sentences to askings of whether their content is true, nothing ties imperative sentences to commandings that their contents be made true,5 and this suggests that the distinction between meaning and use cannot be elided for practical purposes on the grounds that, often enough, we speak indicative sentences sincerely asserting their content. Now it might be true that there is a statistical sense in which indicative sentences are usually or standardly used by speakers to assert that their content is true, though, as Davidson reminds us, the prevalence of story-telling and unmeant compliments dulls the significance of this sense. Nonetheless, sincere assertion is a use of a sentence, and is no more to be identified with the linguistic meaning of a sentence than is hinting or lying. What this in turn means, then, is that if the notion of a trope, of an individual’s deviant use of a word or sentence, requires a robust notion of a community’s standard use, then there are no tropes. There is the literal meaning of a sentence, given by that sentence’s truth conditions, and there are many uses to which that sentence with that literal meaning can be put, and all the uses to which it can be put, be they straightforward or ironic, serious or playful, asserting or intimating or feigning, are, as uses, on the same footing.

I noted above that many philosophers who discuss his position have difficulty putting their finger on the precise distinction Davidson draws between the meaning of a sentence and the use to which a speaker puts a sentence. They misconstrue the distinction, thinking that for Davidson sincere assertion is a camp follower of meaning, not a type of use. We find Mary Hesse, for instance, writing that, for Rorty and Davidson, “‘Meaning’ is restricted to literal use, and is what semantics is about.”6 But, of course, for Davidson at least, meaning underlies all uses of sentences, and is neither identical with nor restricted to any sort of use. Hesse’s phrase “literal use” is either a misleading way of characterizing a speaker’s sincere assertion of the content of a sentence, in which case it refers to use and not meaning, or it is just a piece of confusion. Nelson Goodman, presenting his own view, writes, “Metaphorical use of language differs in significant ways from literal use but is ... no more independent of truth and falsity than is literal use.”7 If we again take “literal use” to mean sincere assertion, then what Goodman says also applies to Davidson, but not because uses of language can themselves be true or false. Rather, all uses of language depend equally on truth and falsity because the truth conditions of whatever sentence a speaker uses in whatever way he uses it bear on what it is that that speaker is doing with that sentence. We always need to know what the sentence itself means, if we are to know what the speaker intends to convey or accomplish by speaking it. A particularly explicit failure to see that, for Davidson, sincere assertion falls on the pragmatics side of linguistic behaviour, not the meaning side, is found in this remark by Kenneth A. Taylor: “It is especially not clear why a theory of meaning should have to do with assertion, and a theory of use to do with commands, questions, and the like.”8 Even Richard Rorty, usually a sympathetic and accurate expositor of Davidson’s philosophy, fails to get quite right the distinction Davidson has in mind.9 Rorty takes from Quine the image of meaning as a small cleared space within the jungle of use (“the tropical jungle”), a seriously misleading image when applied to Davidson, for it suggests that the distinction between meaning and use is one that divides some sentences from others, with some sentences falling on the side of meaning, others on the side of use. For Davidson, however, every sentence uttered has both a meaning and a use, sentences used metaphorically no less than ones used straightforwardly to make assertions.

That metaphors have no meaning in addition to the literal meanings of the words of which they consist does not, of course, mean that understanding a metaphor does not call for interpretation and explication, though it might mean that the interpretation of metaphor calls for no skills beyond those required for the interpretation and explication of linguistic behaviour generally. The reader of Dawkins’s sentence quite rightly can ask what Dawkins is getting at, what point, if any, he is trying to make. It is important to remember, of course, that not every making of a metaphor has a point, at least not if to have a point is to convey, whether by asserting or intimating or hinting, a more or less definite idea or claim, something about which questions of evidence or reasons can be appropriately asked. Some metaphors evoke images or feelings, not ideas, and are intended to set moods: “The ever-hooded, tragic gestured sea/ Was merely a place by which she walked to sing,” from Wallace Stevens, might be an example.10 But some do have a point, at least within the paragraphs and chapters that are their homes. There can be no objection to drawing out that point, to averring that his metaphor reveals, for instance, that Dawkins contends that our behaviour can be explained in terms of strategies to increase reproductive success. What can be objected to, however, is the idea that in drawing out the point of a metaphorCthe point that its author wanted to convey by making that metaphorCone has said what the meaning or content of the sentence itself is.

We should also object to the idea that there is some general way in which metaphors work, a way in which a sentence must work if it is to be a metaphor. We should object to the idea that there is a specific way in which metaphors convey or make their point, or evoke images or feelings, or set moods, a particular way that we can describe as the way of metaphor and about which we can theorize. Now certainly it is useless to ask just what makes any particular use of language a case of hinting, for instance, or to ask just how hints work, expecting to to gain insight into the essence of hinting. It is equally useless, I think, to ask what makes a particular use of language a metaphor, or to ask for the essence of how metaphors work. We can, of course, and we should, investigate particular uses of language that we feel are metaphors, and ask of them, in their particularity, what they accomplish or what their speakers intend to accomplish, and how they manage, or why they fail, to accomplish it. And we may very well find it enlightening to compare different metaphors with each other to see what they have in common and then each with sentences used in other ways to see how they differ from these other uses. We might even discover interesting generalizations about good and bad metaphors, though we should remain sceptical about the projectability of the generalizations we discover, and we should certainly not try to impose our findings as criteria for being a metaphor. The enlightenment we are after in our investigations and comparisons is not enlightenment as to the essential nature of metaphor, irony, sincere assertion, lying and other uses of language. Rather, our goal is simply to understand and appreciate particular, specific utterances. In the end, we mean for our efforts here to reveal to us our tastes and talents, and even, with luck, to contribute to their education and improvement.

It is for this reason that it is wrong to fault Davidson’s account of what metaphors mean for failing to describe in detail the role that our grasping the literal meanings of their words plays in our understanding or appreciation of them. Davidson, it is true, does not say much about the role of literal meaning here, except that to appreciate a metaphor one must understand its literal meaning. He makes only vague remarks to the effect that grasping the literal meaning of a metaphor nudges us into making a comparison, or intimates something, or evokes a feeling. Susan Haack, however, wants to know just how the evocation or intimation achieved through making a metaphor comes about as a result of our grasping the literal meaning of the metaphor.11 What is it, she asks, about the way the evocation or intimation comes about through metaphor that makes metaphor different from other uses of language? After all, utterances of sentences fulfil many different causal roles, “only one of which is characteristic of metaphor.” Why, though, Davidson would ask in response, should we seek for a Platonic definition of metaphor, or think that having one would be enlightening? Why should we think that there is an essence of metaphor, or that our appreciation of metaphors would be sharpened were we to know that essence? It is enough, it seems, that we have examples handy of the various ways in which metaphors convey or intimate or evoke what they do, examples we can use to shed light on new or problem cases.

It is also for this reason that it is a mistake to think of Davidson’s account of what metaphors mean as another theory of how metaphors work. Max Black has described three different such theories, the substitution, comparison and interaction theories.12 According to the substitution theory, metaphors have a linguistic meaning, and that meaning can be given by another sentence, a sentence we could call a paraphrase of the metaphor. According to the comparison view, metaphors draw comparisons, or reveal similarities, among things. Finally, according to the interaction theory, the one Black himself offers, metaphors, working as lenses or filters, enable us to see one thing as another, or one thing under an aspect of another. It is true that Davidson’s position conflicts outright with the first theory. Davidson rejects the contention central to the substitution theory, that we can find a metaphor’s meaning in another sentence, on the grounds that metaphors have no meaning in addition to their literal meaning and, thus, no additional meaning to capture in a paraphrase. But there is no conflict at all between his position and either of the other theories, for neither theory posits, at least not in its bare-bones form, that metaphors have meanings in addition to their literal meanings. Thus his is not a third, rival theory to be placed alongside these two. Partisans of particular theories of how metaphors work, like Black, often do explicitly contend that there exist metaphorical meanings, but this contention is not a necessary feature of their theories of how metaphors work. Davidson himself could be construed as endorsing the comparison theory, though only so long as we understand such a theory as simply an occasionally helpful general description of what goes on when we come to appreciate a metaphor.13 Black, however, is probably right that the comparison theory, while applicable to most metaphors, fails to uncover the workings of certain mysterious or haunting metaphors, and that in these cases his interaction theory does a better job. At any rate, the point to note here is that the two issues, what metaphors mean and how metaphors work, are different issues entirely.

The claim that a theory of truth for a language can supply a statement of the meaning of each of its sentences rests on some general ideas about just what a language is and is not. One of these ideas, we have seen, is that the meaning of a sentence depends, in a systematic way, on the meanings of the elements within it. This idea grounds Davidson=s contention that the meaning of “We are survival machines” is given by the T-sentence “‘We are survival machines’ is true if and only if we are survival machines,” and not by the T-sentence “‘We are survival machines’ is true if and only if we are organisms who are like machines that have been constructed to do what promotes their survival.” The right hand of the latter T-sentence might well capture what Dawkins had in mind to convey to his readersCand we might, though at the risk of confusion, say that it captures the speaker’s meaning of that utteranceC, but that T-sentence cannot give the meaning of his sentence, for it could not be generated by a theory of truth that displays the semantic connections between Dawkins’s sentence and other sentences with which it shares terms.

The idea that the meanings of sentences depend, in a systematic way, on the meanings of the elements within themCthe compositionality thesisCis certainly more than plausible, but not everyone finds it entirely beyond question. For instance, those philosophers and cognitive scientists who attempt to find the meanings of sentences in the typical uses of or reactions to sentences uttered take sentences themselves, or the speech acts of uttering sentences, not words, to be the smallest units of meaning analysis; they thereby see the relation between sentences and the words out of which they are composed to be similar if not identical to the relation between words and the letters or phonemes of which words are composed. Just as “chin” is not a semantically significant stretch of “machine,” but merely part of a convention that prevents people from confusing talk of machines with talk of mates, so, too, within this project, “machines” in “We are survival machines” is merely an element in a convention for writing that we are survival machines. What “We are survival machines” means is, on this account, given either by the illocutionary or perlocutionary intentions of the speaker, or by the idea or image typically engendered by it in those who know the language.

But, most likely, the thesis that sentences are signs or symbols that take their meaning from the extra-linguistic intentions of their speakers or from the reactions of their hearers is not true. Typically we have to revert to a description of the truth conditions of sentences when we attempt to explain why their speakers used the specific sentences they did to do the job they wanted done or to explain why those sentences had the effect they had on their hearers or readers. Typically, that is, it seems that use or effect comes by or through meaning and, thus, cannot constitute it. We must take linguistic meaning to be autonomous of use if we are to cite the meanings of sentences in our descriptions and explanations of linguistic behaviour.

It is, however, precisely the view that meaning is given by use or effect that is implied by the positions of many critics of Davidson=s account of metaphor, or at least so I intend to show below in the case of two critics. Thus, to the extent that one finds sound the argument sketched above why this view is inadequate, one has reason to prefer Davidson’s account of what metaphors mean to those of these critics.


Max Black, I mentioned, rejects the popular view that metaphors are either disguised similies or implicit comparisons and proposes instead that metaphors are instances of seeing as, of seeing one thing as another. There is much to say in favour of Black’s interaction view, even though, as Black admits, there is much that is vague in it.14 In any case, as I urged earlier, I do not think that Black’s positive account of how metaphors work is inconsistent with Davidson’s dismissal of the notion of metaphorical meaning. What is inconsistent with that dismissal, though, is Black’s claim that in making a metaphor, at least when our making is in earnest, we affirm the content of the metaphor we make.15 If Davidson is right, there=s no content to affirm.

That we, when in earnest, affirm the content of our metaphor is one among a number of assertions concerning metaphor that Black lists. According to these assertions, Dawkins affirmed that we are survival machines when he wrote his sentence down, given that he meant what he wrote. Dawkins was saying something, not just, to use Davidson’s locution, nudging his readers to find similarities between humans and survival machines. At least, says Black, this is the commonsense of metaphorsCand philosophers, as always, should be wary of contradicting common sense.

That Dawkins, in writing his sentence, was affirming that we are survival machines means that “machines,” perhaps along with “survival,” cannot have had only its literal meaning, at least not if, in affirming that we are survival machines, Dawkins was affirming something that might well be true, something worth considering, not something false in virtue of its meaning alone. “Machines” cannot have had its literal meaning, or have had only its literal meaning, for then Dawkins’s sentence would be false in virtue of its meaning. Instead, “machines” has an additional or, perhaps, extended meaning, one that fits it for application to entities lacking metal, gears, circuitry. But this should not incline us to think that a new meaning has been added to the English word-type “machine,” Black insists. What has happened, rather, is that the word “machine” acquired a nonce-meaning on the occasion of Dawkins’s utterance, an additional or extended meaning in that context only.

Davidson complains of the vacuity of the idea of nonce-meanings, saying that explanations in terms of nonce-meanings are like explanations in terms of dormative power.16 But the deeper criticism here is that the idea of a nonce-meaning is the idea of a meaning at once both linguistic and nonsystematic. It is linguistic in that the word “machine” itself had, on that occasion, an additional meaning, perhaps an extended extension. And it is nonsystematic in that no theory of truth, no theory of truth for the language to which Dawkins’s utterance belongs that a radical interpreter could confirm, would reveal that that word had that extended meaning on that occasion of use. This, however, is to deny that sentence meaning is a systematic function of word meaning and arrangement; the sentence has a linguistic meaning not systematically generated by word meaning. Its meaning, therefore, resides in the effect that the sentence, as a whole, has on us. One who holds that the effect of a sentence on its interpreter is to be explained by adverting to its meaning, rather than that its meaning is given by its effect, has, then, a solid reason for rejecting Black’s account of metaphor and nonce-meanings.

Is Davidson, then, denying that Dawkins affirmed what he wrote? If so, is Davidson holding that Dawkins did not mean what he wrote, that he was not in earnest? Davidson is denying that Dawkins affirmed what he wrote, but this does not imply that he must hold that Dawkins was not, in writing his sentence, acting in earnest. Much of what Black lists as the commonsense of metaphor can be asserted in a Davidsonian account of metaphor, so long as one is careful to clarify a few key terms, terms such as “affirmed,” “saying something,” “meant,” and “in earnest.”

As Black would insist, the phrase “We are survival machines” surely did occur to Dawkins before he decided to keep it in his book, and his decision to keep it in his book, we can assume, was made in earnest. But this does not mean that Dawkins affirmed that we are survival machines or even that he entertained the thought that we are, in the sense of wondering whether it was true. He meant what he wrote in that he intended to write a sentence true if and only if we are survival machines, and thought that writing a sentence with those truth conditions would adequately or superbly get a particular job done, the job of preparing his readers to find patterns where they might not have looked before, or types of patterns they might not have perceived before. He thought he could do this by writing the sentence he wrote because it is a metaphor that draws attention to similarities between two things or, as Black might have it, promotes the seeing of something as another thing.


Frank B. Farrell, for his part, is happy to avoid the idea that words in metaphors take on additional or extended meanings.17 He agrees with Davidson that each word in a metaphorical sentence possesses only its literal meaning. Nonetheless, Farrell maintains, the metaphorical sentence itself, as a whole, does possess a meaning in addition to its literal meaningCit possesses a metaphorical meaning. Of course, because its words possess only literal meanings, the metaphorical meaning of a sentence cannot have been generated by the meanings of its constituent words. What determines the metaphorical meaning of a sentence that has one, then, according to Farrell, is its location within metaphorical schemes, within, that is, patterns of other sentences, some literal, some metaphorical. Moreover, because the metaphorical meaning of a sentence is a function of its relation to other sentences and, thus, to linguistic entities, the notion of metaphorical meaning remains a citizen in good standing of the city of semantics, not a renegade or nomad haunting the hinterlands of use and pragmatics.

Consider the sentence Farrell offers as an example of metaphor, “Robinson’s book on the history of philosophy is a cheap TV dinner.”18 We understand the metaphorical meaning of this sentence, Farrell says, when we understand that the sentence, taken metaphorically, implies that Robinson’s book does not provide a scholarly, in-depth account of its subject matter, the history of philosophy. The sentence that describes this implication is a sentence within the metaphorical scheme that determines the meaning of the metaphorical sentence. Whether the metaphor, understood as a metaphor, is true or warranted has to do with whether other sentences in this scheme are true or warranted.

This view, Farrell contends, enables us to see how a metaphor can be false when taken literally and true when taken metaphorically, without requiring us to equivocate between the strict semantic meanings and other, looser, meanings of “true” and “false.”

Farrell’s particular examples, however, because they verge on cliché or idiom, easily lend themselves to another interpretation. Davidson could, it would seem, allow that “Robinson’s book is a cheap TV dinner” does imply that Robinson’s book fails to provide an in-depth account of its subject matter and that its truth value is tied to the fortunes of that implication. This is because “is a cheap TV dinner” can, in the context of sentences similar to Farrell’s example, literally mean fails to provide an in-depth account. The phrase “is a cheap TV dinner” might, in the language of the sentence, be ambiguous between what its constituent words add up to mean and an idiomatic meaning it has as a single term. To call a theory a house of cards, to take another of Farrell’s examples, might well be literally to call it a theory based on insufficient or tenuous evidence, because that is what it is, literally, for a theory to be a house of cards. Farrell’s examples might be examples of dead metaphors, and dead metaphors are not metaphors at all.

Davidson, Farrell says, “ignores his own general strategy for semantics” in claiming that metaphors have no meaning in addition to their literal meaning; actually, according to Farrell, “his general strategy provides a way to defend the semantic status of metaphor.”19 Davidson’s general strategy, of course, is to assign meanings to sentences on the dual assumption that speakers’ beliefs form strongly coherent sets and most of what speakers believe is true. Were Davidson to follow his general strategy more rigorously, he would, according to Farrell, realize that many metaphors have a stable semantic location in the speakers’ languages and are true. Farrell’s particular examples of idioms and clichés, though, have stable semantic locations and easily figure in true sentences simply because they function as terms, as sub-sentential semantic primitives.

The point that colourful idioms are not necessarily metaphors is important to this discussion because of the centrality in Farrell’s account of metaphorical meaning of the notion of a metaphorical scheme. According to Farrell, we determine what a metaphor means, and then evaluate its truth value, by finding its location within a metaphorical scheme, and then evaluating the sentences within that scheme that it implies. There is a hope of definiteness in this proposal at odds with the suggestiveness or imagistic nature of many of the metaphors we value. To which metaphorical scheme, on any helpful sense of the term “scheme,” should Yeats’s “Imagining in excited reverie/ That the future years had come,/ Dancing to a frenzied drum,/ Out of the murderous innocence of the sea” be assigned?20 If we can find a scheme at all, it would seem, we would do so on the basis first of our understanding and appreciation of the force of these lines; that is, the scheme would emerge after we have achieved an understanding, not as a vehicle conveying to us that understanding. On the other hand, Farrell’s proposal might well explain our ready understanding of those turns of phrase, fresh and colourful or stale and dull, that lack the richness or force of metaphors.21

These remarks do not add up to a direct argument against Farrell’s view, for they are first of all about the sorts of examples Farrell uses to illustrate his view. What they at least do, though, is to help to bring that view into focus. If Farrell is right, then metaphorical meaning is just as definite as literal meaning, even though, unlike literal meaning, it is not a function of a finite stock of words and ways of arranging them. We might fail in our endeavour to find the meaning of a metaphor, but that is because time and resources are limited, and not because, as Davidson maintains, there are no meanings to find while the process of explication is, in principle, endless. But for Farrell to be right, it must be true that the sentence that is a metaphor has, as a whole, independently of our understanding of its literal meaning, a particular effect on us; if it does not have this effect, we have not understood it. Those of us who are suspicious of a semantics of sentences that does not draw upon a semantics of words and their arrangements, then, have reason to favour Davidson’s account of metaphor over Farrell’s.


I said in section II that some philosophers misconstrue Davidson’s distinction between sentence meaning and sentence use as a result of their miscategorizing sincere assertion, of their placing it in the meaning rather than the use category. Other philosophers, however, misconstrue this distinction as a result of their misconceiving the role Davidson assigns to our appreciation of speakers’ uses of words in our attempt to interpret their sentences. These philosophers tend to think that Davidson’s writings on meaning since “A Nice Derangement of Epitaphs” (or perhaps since “Communication and Convention”) mark a departure from his earlier contention that a Tarski-style theory of truth for a language yields a theory of meaning for that language. Davidson, they think, now conceives of meaning almost wholly in terms of use. Thus they contend that Davidson’s recent writings imply that metaphors mean what they are used to do and, when used successfully, are true. My view, on the other hand, is that the idea of language Davidson now explicitly rejects has never really been an idea he accepted, or at least not one he ever drew upon in philosophizing about language. The meaning of a metaphor is the literal meaning of the sentence, as given by the truth conditions of that sentence as understood by an ideal radical interpreter. The theories of meaning radical interpreters construct do not make reference to the habits of the speech communities of those whose sentences they interpret, and thus the languages they describe are not communal languages, and the word meanings they find are not (that is, they need not be) average or standard or preferred meanings.

Part of the reason for this misunderstanding of Davidson’s recent thought is an ambiguity in the term “use.” When contrasting use with meaning, Davidson’s concern is with the illocutionary, perlocutionary, or other extra-linguistic intentions of speakers, their intentions to use sentences to assert and inform, for instance, or to hint or mislead, and with their intentions to use words to create the sentences with which to do these things. However, when Davidson speaks of word use in discussions of radical interpretation, he is usually speaking of speakers’ applications of their words to objects and events in the world. A radical interpreter, by assuming that most of what the speaker believes is true, fashions and tests hypotheses about which sounds or markings go with which things. (Of course, to assume that most of what a speaker believes is true is not itself to assume that most of what he says is true, or even that most of what he says he believes to be true.) Investigating a speaker’s use of words, in the sense of his application of them to things, is a necessary part of coming to understand the meanings of those words and, thus, of the sentences that can be formed with them. It is important to realize, however, that, for Davidson, despite the importance he places on interpreters’ observations of speakers’ word application, use in the sense of application does not itself give or constitute word meaning. Word meaning is constituted by the speaker’s correct application of his words to things, by, more accurately, his standards of their correct application. We all sometimes do things like apply our word “cat” to a boot half hidden by a door, thinking wrongly that it is a cat, but our word “cat” does not include such boots in its extension. Now for most of our words, most of the time our applications of them to things will conform to our linguistic standards; if this were not true, it would be false that most of what we believe is true. Because most of a speaker’s applications will conform to her standards, how a speaker in fact applies her words to things provides an interpreter with excellent evidence as to their correct application and, thereby, to their meanings. But, we must be clear, evidence of their meaning, and not their meaning itself, is all that actual application provides.

Because when making metaphors speakers usually violate their linguistic standardsCthat is, because they usually apply one or more words to things those words do not, given those standards, correctly apply toCmost metaphors are false. Were one to neglect, or in any case to downplay, the distinction between a speaker’s actual application of his words and the correct application of them, one might well come to the conclusion that most metaphors are not obviously false but rather trivially true. And this, indeed, is the conclusion E.M. Zemach comes to.22 According to Zemach, metaphors, at least good ones, are literally true, though they might not constitute useful knowledge for our practical purposes. Metaphors compare one thing with another, or place a thing in a category. But since everything is like everything else in one respect or another, these comparisons or categorizations are not just often apt, but true. When we understand the literal meaning of the sentence, an understanding we are to get by discerning what things the words of the sentence are being applied to, we understand the metaphor, and immediately see that it is true. Comparisons or categorizations, however, derive their interest and usefulness from our purposes; some comparisons serve a particular purpose better than others, and some fail to serve it at all. Comparisons meant to serve our everyday and practical purposes are not metaphors, and those among them that fail to serve those purposes are, perhaps, false, while those comparisons meant not to serve mundane or standard purposes are simply metaphors, and are usually true.

According to Zemach, there will be some purpose somewhere that categorizing things in that particular never-done-before way will serve well. This means that any property by which to categorize things can be observationally basic; given that purpose, the world can be seen through the lens of that categorization. Metaphors turn out to be literally true sentences in which the observationally basic properties that give those sentences their meanings are novel propertiesC“novel” in the sense, perhaps, of newly recognized, perhaps in the sense of newly created, but in any case properties the recognition of which is impractical from the point of view of commerce or science.

Let us recall Davidson’s contention that a speaker’s application of a word to a thing is only evidence of that word’s meaning, for the speaker might well be misapplying his wordCmisapplying it according to his own standards, of courseCto that thing. When a door is pulled back to reveal a boot, not a cat, I count myself mistaken in my own terms in having said that a cat is behind the door. But that is a case of sincere assertion; what about in a case of metaphor? Zemach’s position, it seems, makes all cases of mouthing metaphors, or almost all, cases of sincere assertion. Thus, when I call a former friend who lately has let power corrupt him a “festering lily,” I am sincerely asserting something that is true if and only if, in my language, he falls into the class of festering lilies. But Zemach is wrong that I am here sincerely asserting the content of my sentence, for in my language it is false that my former friend is a festering lily; despite all the similarities between the two, he simply does not have enough attributes, or the proper attributes, to be what I take to be a festering lily. A radical interpreter who comes to understand my standards for the correct application of “festering lily” will understand that what I have said is, in the particular language I spoke at the time, false.

Zemach is certainly on to something here, however, although it is not that metaphors express strange, impractical, but true claims about what in the world goes with what. What he is on to is that the question whether some utterance is a metaphor rather than something else is not always a question with an easy answer, and might in some cases be a question with no unique correct answer. Whether an utterance is the speaking of a metaphor or the making of a sincere assertion or whatever has to do with the language the speaker was speaking at the time and her extra-linguistic intentions. An interpreter seeking to make sense of a speaker’s behaviour will have to determine what meaning to assign the sentence uttered and what point to assign the speaking of it. Whether the sentence is to be counted as obviously false (or trivially true) and a metaphor, or strangely true and a sincere assertion, will depend on the interpreter’s sense of how best to account for all the evidence he has. Consider again the sentence “We are survival machinesCrobot vehicles blindly programmed to preserve the selfish molecules known as genes.” I have throughout this paper assigned this sentence specific truth conditions, and declared it both false and a potentially illuminating metaphor. But I might well have misunderstood Dawkins; his sentence might not be a metaphor at all. “Machine,” in the language Dawkins was then speaking, might include organisms in its extension along with machines. If that is the case, his sentence would be true if and only if, as we might put it in our language, we are survival machines or survival organisms. (Or perhaps our language will have to develop to look more like his if it is to possess the conceptual resources required to express the meaning of his sentence.)

My grounds for taking Dawkins to have used his sentence metaphorically have to do with my ideas about how best to understand his book and what he intended in 1976 to do through it. Were Dawkins to pen another instance of that same graphic shape today, I might instead decide that he is sincerely asserting its content, a content different from that which that shape had back then. His language might have changed, its conceptual resources having expandedCor, perhaps, contracted. The important lesson to take from Zemach’s paper is that whether a sentence is a metaphor is not something discernable by looking only at the sentence itself, by itself.


To conclude: Both Max Black and Frank B. Farrell draw attention to the fact that metaphors are (at least sometimes) appropriate objects for explication and evaluation. They take this fact to require that metaphors be interestingly true or false, and this, in turn, they take to require that metaphors have linguistic meanings in addition to their literal meanings. Black finds this additional meaning in the idea that nonce-meanings attend certain words in the metaphor; Farrell finds it in the idea that metaphorical sentences have, as wholes, locations among other sentences not determined by the semantics of words and their arrangements. Each of these ideas, I have argued, implies that linguistic meaning can reside outside of the systematic contribution the parts of sentences make to the whole, outside in the effects sentences supposedly typically can have or properly should have on us. But this, I claim, is to get the order of linguistic understanding backward: illocutionary or perlocutionary effects do not give locutionary meaning; locutionary meaning, rather, is a basis from which illocutionary or perlocutionary effects arise.

E.M. Zemach agrees with Davidson that metaphors mean only what their words mean, but he argues that, as it turns out, almost all of them are true. Metaphors are true because they express comparisons, and everything is like everything else. Zemach takes seriously, as he should, the idea that the words of a speaker’s language get their meaning from their place in that speaker’s own language, not from their place in some other person’s or persons’ language. However, he neglects the fact that it is not simply its actual use that reveals a word’s meaning, but its correct useCcorrect, of course, according to the speaker’s standards at the time of utterance. For Zemach, a speaker making a metaphor is always sincerely asserting its content. Actually, though, when making a metaphor, speakers are not, or not usually, using their words to express their ideas about the novel properties or categories things possess or fit into. Typically, they are deliberately applying their words to things in ways out of keeping with their current linguistic standards. What they say, then, is usually false, though by saying it they might come to achieve great things.

Donald Davidson’s thesis that metaphors mean only what their words mean, only what their T-sentences tell us they mean, T-sentences got through an appreciation of the speaker’s linguistic standards, is not inconsistent with the fact that metaphors are appropriate objects for explication and evaluation. Realizing that a sentence is obviously false or trivially true, we also realize that its speaker likely is not asserting that its content is true, or is interested in doing so; so we ask what other thing might the speaker be doing speaking a sentence with that particular meaning. Having some answer to this question in mind, we can then decide whether speaking that sentence was a good, bad or indifferent way of doing it. We can, if it seems called for and worthwhile, explicate the metaphor using further sentences, and attempt to ascertain the truth value of these other sentences. But we do not have to find in the meaning of any of these other sentences the meaning of the metaphor. And it is best that we do not, for otherwise we blur the distinction between what sentences can be used to do and what they mean.23


1 Donald Davidson, “What Metaphors Mean,” Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984).

2 Richard Dawkins, The Selfish Gene New Edition (Oxford: Oxford University Press, 1989), v. (The sentence is found also in the first edition, from 1976.)

3 Max Black, “How Metaphors Work: A Reply to Donald Davidson,” Critical Inquiry 6 (Autumn 1979), 132.

4 Donald Davidson, “A Nice Derangement of Epitaphs,” in Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson, ed. Ernest LePore (Oxford: Blackwell, 1986).

5 Davidson argues for this point in “Communication and Convention,” Inquiries into Truth and Interpretation.

6 Mary Hesse, “Tropical Talk: The Myth of the Literal,” Proceedings of the Aristotelian Society Suppl. Vol. 61 (1987), 297 (my italics).

7 Nelson Goodman, “Metaphor as Moonlighting,” Critical Inquiry 6 (Autumn 1979), 125.

8 Kenneth A. Taylor, “Davidson’s Theory of Meaning: Some Questions,” Philosophical Studies 48 (1985), 104 (n. 4).

9 Richard Rorty, “Unfamiliar Noises: Hesse and Davidson on Metaphor,” Proceedings of the Aristotelian Society Suppl. Vol. 61 (1987), 285-286.

10 Wallace Stevens, “The Idea of Order at Key West,” The Collected Poems of Wallace Stevens (New York: Alfred A. Knopf, 1968), 129.

11 Susan Haack, “‘Dry Truth and Real Knowledge’: Epistemologies of Metaphor and Metaphors of Epistemology,” in Aspects of Metaphor, ed. Jaakko Hintikka (Dordrecht: Kluwer, 1994), 9.

12 Max Black, “Metaphor,” Chapter 3 of Models and Metaphors (Ithaca: Cornell University Press, 1962). Black summarizes the three models on 44-45.

13 Toward the end of “What Metaphors Mean,” however, Davidson discusses Black’s interaction theory in a friendly way. He criticizes only Black’s claim that paraphrase fails because it strips metaphor of its power; for Davidson, as we have seen, paraphrase fails because there is no such thing as metaphorical meaning and, thus, nothing to capture in a paraphrase. “What Metaphors Mean,” 260.

14 Black, “How Metaphors Work,” 142-143.

15 Black, “How Metaphors Work,” 132-133.

16 Davidson, “What Metaphors Mean,” 247.

17 Frank B. Farrell, “Metaphor and Davidsonian Theories of Meaning,” Canadian Journal of Philosophy 17, #3 (September 1987).

18 Farrell, 625.

19 Farrell, 627.

20 W.B. Yeats, “A Prayer for My Daughter,” The Variorum Edition of the Poems of W.B. Yeats, eds. Peter Allt & Russell K. Alspach (New York: Macmillian, 1971), 403.

21 Farrell seems aware that he is, for the most part, discussing idioms, not metaphors, and, further, that his view will not apply to metaphors for just the reasons I’ve given. He writes that many metaphorical sentences are not locatable in semantic systems, that many metaphors play no semantic role. Presumably, he holds with Davidson that these sentences have no meaning in addition to their literal meaning. Farrell, 630-631.

22 E.M. Zemach, “Metaphors and Ways of Life,” in Aspects of Metaphor.

23 I thank Sophie Beaulé, David Braybrooke, Søren Häggqvist, and David Sturdee for their helpful comments.

Download 74.17 Kb.

Share with your friends:

The database is protected by copyright © 2022
send message

    Main page