Back to home page
Back to "Occasional Papers"

Reuven Tsur


Lakoff’s Roads Not Taken


This paper is a critique of George Lakoff’s theory and practice as presented in his “Contemporary Theory of Metaphor” (Lakoff, 1993).[1] It addresses the issue on sev­eral planes, on each plane comparing Lakoff’s approach to some alternative. The highest plane, affording the widest perspective, concerns two approaches to interpre­tation and scientific thinking: one that relies on a pre-established set of meanings, and one that assumes that “all the work remains to be done in each partic­ular case”. The two approaches involve different cognitive strategies, rapid and delayed concep­tualization. Another plane concerns the cognitive explanation for using spatial im­ages in metaphoric and symbolic processes. Here the “embodied-mind hypothesis” is confronted with the “efficient-coding hypothesis”. It is argued that the latter is more adequate, and can better account for the mental flexibility required for “delayed con­ceptualization”. On the third plane, Lakoff’s “Contemporary Theory of Metaphor” is compared to Beardsley’s “Controversion Theory of Metaphor”. I will assert that pre­cisely in those respects in which Lakoff claims superiority for his theory it is infe­rior to Beardsley’s. On the most concrete plane, Lakoff’s handling of three texts is considered, two literary and one nonliter­ary. It is argued that in two cases Lakoff’s conceptual apparatus is less than adequate to handle the arising problems; in the third case it allows him to say about the text exactly what every critic would have said about it for the past seven hundred years.



1. Rapid and delayed conceptualization

The point of departure for the following exercise is the observation that the his­tory of the inter­preta­tion of symbolic processes and of figurative language is domi­nated by two polar attitudes: one pole relies on more or less pre-de­termined meanings, the other one insists on certain “meaning poten­tials” of the sign unit, the final mean­ing(s) be­ing determined by its unforeseeable in­teraction with the signs that consti­tute the con­text. These are two alternative cogni­tive strate­gies. The former yields rather quick results and arouses in the interpreter a feeling of certainty, but tends to be rigid and maladaptive: it may miss some of the most legit­imate pos­sibilites. The latter is slower, and requires considerable tolerance of uncer­tainty; but it is much more flexible in its application. I have called these strategies “rapid” and “delayed concep­tualization”. In what follows, I will briefly explore these cognitive strategies.

In the psychoanalytic interpretation of dreams, Wilhelm Stekel compiled a dic­tionary of dream symbols. Freud regarded this as inadequate: “after warning that such a gift as Stekel’s is often evidence of paranoia, he decides that normal persons may also occasionally be capable of it” (Burke, 1957: 228). As it is frequently said, sometimes a cigar is just a cigar. Freud and some other practi­tion­ers and theoreti­cians believe that one cannot know what a dream symbol means until it is viewed in the context of the dream and the free associations of the dreamer; and the cigar, for instance, will change its meanings according to the unique stream of associations provided. In lit­er­ary theory, in the first third of our century, Richards (1929) led an assault on con­temporary aca­demic education and criti­cal practice, because they en­couraged “stock responses”, that is, some undifferentiated responses to images and symbols when­ever they occur, ir­respective of context. He vigorously insisted that symbols change their meanings and require subtle changes of response when they enter into different contexts.[2]



2. At the crossroads

I have a problem with Lakoff’s application of the “conceptual metaphor” to liter­ary texts. To indicate its nature, consider his discussion of three lines by Robert Frost:


Two roads diverged in a wood, and I-

I took the one less traveled by,

And that has made all difference.


Since Frost’s language often does not overtly signal that the poem is to be taken metaphorically, incompetent English teachers occasionally teach Frost as if he were a nature poet, simply describing scenes. (I have actually had students whose high school teachers taught them that!) Thus, this pas­sage could be read nonmetaphorically as being just about a trip on which one encounters a crossroads. There is nothing in the sentence itself that forces one to metaphorical interpretation. But, since it is about travel and encountering crossroads, it evokes a knowledge of journeys (Lakoff, 1993: 238).


There is a much larger number of equally incompetent English teachers who teach their students that whenever you encounter travel and crossroads, you have to activate, automatically, the life is a journey conceptual metaphor. (I have actually had students whose high school teachers taught them that!) Now what is so power­ful about Frost is his irony: he pretends to know nothing, not that he is metaphori­cal, not even ironical. His “language often does not overtly signal that the poem is to be taken metaphorically”. The proper response to Frost’s poem involves the uncertainty whether the image is metaphorical or not. Now what I would expect a competent English teacher to do is to teach his students, first, that in some contexts crossroads are metaphorical, and in some not; second, that crossroads may have a wide range of metaphorical meanings; third, that they should look for prin­cipled argu­ments that may support the claim that in a certain instance “cross­roads” should or should not be understood metaphorically, and in what sense(s); and fourth, that there is a stylistic difference between poems that do and those that do not overtly signal that they are to be taken metaphorically, and they require different kinds of response. The suggestion that “crossroads” can be metaphorical is trivial; it is the proper handling of these four issues that would make a competent English teacher. It is the third of these four points that poses the great­est difficulty to Lakoff. In order to find such princi­pled arguments one must admit that something is wrong with the literal meaning of the poem; for instance, that “And that has made all difference” violates some of Grice’s conversational maxims. But this would contradict one of Lakoff’s pet assumptions, namely, that metaphors do not violate communication maxims.

Ray Gibbs’s discussion of conversational maxims may account for the source of Lakoff’s problem; by the same token it may throw some light on the nature of rapid and delayed conceptualization.


One reason many scholars believe figurative language violates communica­tion maxims is that they confuse the process and product of linguistic un­derstanding. All language interpretation takes place in real time ranging from the first milliseconds of processing to long-term reflective analysis. This temporal continuum may be roughly divided into moments corre­sponding to linguistic comprehension, recognition, interpretation, and ap­preciation. Comprehension refers to the immediate moment-by-moment process of creating meanings for utterances. Recognition refers to the prod­ucts of comprehension as types (i.e., determining whether an utterance  conveys a particular type of meaning such as literal, metaphorical, ironic and so forth). Interpretation refers to the products of comprehension as to­kens (i.e., determining the specific content of the meaning type). Apprecia­tion refers to some aesthetic judgment given to a product either as a type or token (Gibbs 1993: 255-256).


My point is that Gibbs’s distinction between process and product works both ways: it may suggest that in some instances metaphor need not violate communica­tion maxims; but it may also explain why insistence on this distinction may lead Lakoff to the mishandling of Frost’s metaphor. When Lakoff and myself disagree as for what is a competent or incompetent En­glish teacher, we do, in fact, disagree about the proper uses of Gibbs’s sequence “lin­guis­tic comprehension, recognition, interpretation, and ap­preciation”. Judging from the above example, for Lakoff, liter­ary response con­cerns only comprehension; for me, it concerns much of the whole sequence. What is more, at each step, deci­sion in­volves uncertainties—not only in the process of deci­sion, but also in its product. That is what my foregoing descrip­tion of a competent English teacher im­plies. The moment-by-moment account of the process by Gibbs may illuminate the nature of rapid and delayed conceptualiza­tion, but with a twist. Confining the process to its first step allows relatively rapid conceptualiza­tion. As much is evi­dent. But the abil­ity to give a thorough account of the whole process is not neces­sarily evidence of delayed conceptualization. Delayed conceptual­ization implies that one is capable of perceiving much of the process “in a flash”. (I cannot tell, how­ever, how the phe­nomenological quality “in a flash” should show through Gibbs’s experimental pro­cedures). One may, of course, pro­duce experimen­tal evidence that readers or listeners are able “to create some interpre­tation for a trope during the earli­est moments of comprehen­sion” (Gibbs 1993: 255). My point is that creating “some interpretation for a trope” is not necessarily a competent re­sponse to a piece of literature.

Or, let me introduce another example: Oedipus the King. “Laius was slain where three highroads meet”. One could plausibly argue that the overwhelming importance of the location is, first, its uniqueness: that the place is almost uniquely identifiable, and for Oedipus it leaves little doubt as for the identity of the murderer; and second, that it “sounds” somehow very significant, partly because of the meeting of three highroads (not two and not four). This meeting of the roads, in turn, may indi­cate metaphorically some “strange coincidence” (which is not the same as the impli­ca­tions of the life as a journey metaphor). Indeed, the play is governed by some strange coincidences. Iocasta had a child, killed long before, who was said to be des­tined to kill his father when he grows up; and Oedipus received a similar prophecy. Laius was killed where three highroads meet; and Oedipus killed an old man at ex­actly such a place. But the strangest coincidence is when everything “ties in”: the analogies are revealed as identities. The existence of the life as a journey conceptual metaphor too may have to do with this “air” of significance; but its specific aspects are of low salience at best. One could argue, of course, that the play is concerned with Oedipus’ life, and his life is a journey to discover its hidden meaning; and that this crossroads has significantly changed it. But then, any person travelling on a road also has a life, and this road could be symbolic of this life. With some good will any crossroads and journey can be forced to become symbolic. Let me put it dif­ferently: Is there for Lakoff any way for a crossroads to escape being symbolic? Or to put it yet differently, one may claim that the meeting of three highroads too can be derived from the life as a journey metaphor. But then any­thing can be derived from anything, and the whole system becomes utterly trivial. Briefly, it is not at all clear what are Lakoff’s constraints.



3. The “Embodied Mind” and the “Efficient Coding” hypotheses

My present suggestion is that we use visual spatial imagery for a variety of rea­sons, one of them being that it is a very efficient coding of many kinds of informa­tion. According to Neisser (1968: 320),


the amount of information may require less capacity coded in terms of spatial relationships than in terms of temporal sequence. […] This assumption would explain the predominance of visual im­agery in dreams, and perhaps also our preference for visual models and meta­phors for thinking, from ‘.x.insight;’ to ‘.x.point of view;’.


In the chapter “The Concrete and the Abstract in Poetry” of my book On Metaphoring I have discussed at great length this aspect of the use of concrete im­ages, also quoting Brooks and Warren:


The word peach implies certain qualities: a certain shape, a certain colour, a certain kind of sweetness. But peach implies these qualities as “grown together” as we should actually find them embodied in a peach. (The Latin word from which concrete comes means literally “grown together”). We can, of course, abstract (this word literally means “to take away”) these qualities from the actual peach and refer to them in isolation: sweetness, fuzziness, softness. Isolating these qualities in such fashion, we get a set of abstract words. Sweetness is a quality common to peaches, of course, and to many other things; the quality is thought of as an idea in its own right (Brooks and Warren 1958: 298).


In a concrete noun or verb a wide range of features are “grown together”, which constitute its “meaning potential”. One or several of them may be actualized in a specific context. In this way, several meanings may be encoded in one expression. “Where three highroads meet” provides, first, a precise, identifiable description of the loca­tion of the murder; and second, it indicates some outstanding significance. In ad­dition, one may evoke, of course, some implications of the life as a journey metaphor. But one must be aware that only a small subset of these implications, if at all, is relevant to the text, and at a very low salience. Such a conception of con­crete images in literature allows the reader or critic to move from one aspect to an­other, yielding great flexibility and considerable accuracy in interpretation. Such an attitude requires strongly delayed conceptualization, to allow a differential response to the image, moving from one aspect to another, choosing the relevant ones, and to respond differentially to the rel­ative salience of the various aspects. To sum up this discussion of roads, I would like to quote L. C. Knights’s (1948: 229) comment in a very different con­text: “But to say this is to admit that all the work remains to be done in each partic­ular case”. Lakoff, by con­trast, smuggled back through the back door the stock re­sponses so forcefully exor­cised by Richards.

What kinds of insight can we get into poetic language, according to these exam­ples, with the two kinds of conceptual apparatus based on the “Embodied Mind” and the “Efficient Coding” hypotheses, respectively? Lakoff’s apparatus can help us to reduce a wide variety of expressions to one underlying image. It can point out that the opening lines of Dante’s Divine Comedy


In the middle of life’s road

I found myself in a dark wood...,


as well as Bunyan’s The Pilgrim’s Progress, along with Frost’s poem, the three highroads meeting in Oedipus Tyrannos, and such colloquial expressions as “Look how far we’ve come. It’s been a long, bumpy road. We can’t turn back now. We are at a crossroads. We may have to go our separate ways. The relationship isn’t going anywhere. We’re spinning our wheels. Our relationship is off the track. The mar­riage is on the rocks. We may have to bail out of this relationship” (Lakoff, 1993: 206) are all derived from the same underlying conceptual metaphor. The apparatus I have proposed may show, by contrast, how several potential meanings are condensed in one expression; by this, in turn, it may account for its expressive force and uniqueness among the other terms derived from the same conceptual metaphor. Fi­nally, it can indicate an image’s potential to combine with a wide variety of addi­tional images at the same time; that is what I call its combinational poten­tial.[3] 

The reader may ask whether I can produce some experimental evidence for my conception of concrete images in figurative language. Ray Gibbs, I believe, has provided some very convincing experimental evidence—unintentionally, though. In a paper presented on 31.12.1997 at the Porter Institute, Tel Aviv University,[4] Gibbs spoke of elementary bodily experiences as the basis of metaphors, such as “balance”, or “the body as a container”. “The idea is that balance is something that we learned with our bodies and not by grasping a set of rules [...] And the point here is that it’s not just an arbitrary thing, that we happen to use balance and talking about these other kinds of things”, such as “people’s personalities are balanced or out of bal­ance”, or that “certain views are balanced”. “It’s motivated because it makes sense in terms of talking about these things, given the nature of our valued experience of balance”. Likewise, “we have a very strong sense of ourselves as bottles and con­tainers”. Such conventional metaphors as “he spilled the beans” or “he blew his stack” can be produced and understood by our interlocutors, because “we have a very strong sense of ourselves as bottles and containers”; hence we all share the underly­ing conceptual metaphor “the body as a container”.

In order to test the psychological reality of this hypothesis, Gibbs conducted the following experiment. He gave three paraphrases of the expression “he blew his stack”, that pointed out one, or two, or three ingredients it shared with “anger”. He asked observers to make preference judgments. He found, not surprisingly, that pref­erence increased with the number of shared ingredients. “So this suggests that our understanding of what “blew his stack”, for instance, means is not just that John got very angry. It means more specifically that John got very angry because he has felt a great deal of internal pressure; perhaps he unintentionally expressed his anger, and he did so very quickly and forcefully”. Gibbs interpreted these re­sults as evidence for the “mind-body” hypothesis. It seems to me, however, that this exper­iment pro­vides lit­tle evidence for that hypothesis. To be sure, this specific idiom does have one ingre­dient of “internal pressure”, because we are speaking of a mental state that is inter­nal, and does involve a sense of pressure; and we are using the image of “blowing up”, that does entail the building up of internal pressure. In this sense, “internal pressure” is predetermined by the specific image and the specific referent of this spe­cific idiom, and is accidental to the use of metaphors in general. But as for meta­phors and idioms in general, the experiment provides, rather, evidence for the cogni­tive hypothesis proposed here, namely, that people use spatial imagery in thinking or in figurative language, because spatial imagery is an efficient coding of informa­tion. Preference of spatial imagery increases with the amount of information coded in it. Subjects prefer that interpretation of “he blew his stack” which contains the meaning components “perhaps he unintentionally expressed his anger, and he did so very quickly and forcefully” not because it corroborates the body as a container conceptual metaphor (it does not); but because it increases the number of meanings encoded in one spatial image. If the metaphor concerns basic bodily ex­periences, its effectiveness may receive additional reinforcement.

Such a con­ception of figurative language, in turn, entails different modes of prac­tical criticism, that yield much subtler insights into poetic language. Thus, for in­stance, there is also a practical consideration for pre­ferring my hypothe­sis to the “mind-body” hypothesis. Subtler and more flexible in­tertextual or intratextual dis­tinctions can be made, and of greater aesthetic signifi­cance, by pointing out the amount of information coded in the various spatial im­ages than by pointing out that such expressions as “he spilled the beans” or “balance sheets” can be traced back to some basic body expe­rience.[5]



4. Stock responses

My argument has two facets. On the one hand, I point out some general problems with Lakoff’s theory of metaphor from the logical and cognitive point of view, as compared to other theories of metaphor; on the other hand, I claim that its literary application may be harmful. One of the issues relevant in the latter respect is Richards’ discus­sion of stock responses. I have suggested that Lakoff smuggled back through the back door the stock re­sponses exor­cised by Richards.


A stock response, like a stock line in shoes or hats, may be a convenience. Being ready-made, it is available with less trouble than if it had to be spe­cially made out of raw or partially prepared materials. And unless an awk­ward misfit is going to occur, we may agree that stock responses are much better than no responses at all. Indeed, an extensive repertory of stock re­sponses is a necessity. Few minds could prosper if they had to work out an original, “made to measure” response to meet every situation that arose [...]. But equally clearly there are in most lives fields of activity in which stock responses, if they intervene, are disandvantageous and even dangerous, be­cause they may get in the way of, and prevent, a response more appropri­ate to the situation. These unnecessary misfits may be remarked at almost ev­ery stage of the reading of poetry, but they are especially noticeable when emotional responses are in question (Richards 1929: 228).


Stock responses have a semantic and a pragmatic aspect. Words enter into verbal contexts, and assume different meanings in different contexts. These verbal struc­tures may convey emotionally and ideologically significant situations. One of the signif­icant differences between the poetic and the non-poetic uses of language is that in the poetic use the semantic changes of words are somehow more significant, and re­quire subtler and more frequent shifts of “mental sets” on the understander’s part. These semantic changes need not be figurative, they may well be literal. Consider the case of the panda who orders a steak sandwich; after eating it he shoots the bar-tender and walks out with­out pay­ing. The manager stops him at the entrance: “What do you think you are doing?” The panda answers: “As you may have noticed, I am a panda; look me up in the dic­tio­nary; it says “bearlike animal; eats shoots and leaves”. In this instance, the words “shoots and leaves” change their literal mean­ings, including their word class and syntactic function. Even “eats” undergoes an admittedly more evasive change of meaning—from a transitive to a middle verb. The phe­nomenological qual­ity of this shift is “wit”. Emotional qualities are generated by streams of subtler and more fre­quent shifts. One aspect of stock responses involves a refusal to shift mental sets as required by the stream of shifting meanings. Another aspect concerns a refusal to shift one’s mental attitudes towards certain emotionally or ideologically loaded situations when the wider context is changing. Richards’ at­tack is directed against both kinds of refusal; but he applies the term “stock re­sponses” only to the latter. One source of the enormous appeal of conceptual meta­phors is that whenever one encounters a certain spatial image in a text, say, “jour­ney”, or “cross­roads”, one need not shift one’s mental sets neither across con­texts, nor within some evolving context. This is, in fact, the main point of my foregoing criticism of Lakoff’s handling of Frost’s crossroads and of what he consid­ers an “incompetent teacher”.

There is an inherent paradox in Lakoff’s work. He and his associates invested al­most unprecedented subtlety and intellectual rigour into working out the “meaning potential” of such concep­tual metaphors as life is a journey, seeing is know­ing, or the body is a container. By the same token they have prepared “an ex­tensive repertory of stock re­sponses”. By this they made an enormous service and disservice at one and the same time to literary studies. On the one hand, “we may agree that stock re­sponses are much better than no responses at all”; and Lakoff’s system allows crit­ics intolerant of “delayed conceptualization” to do reasonable prac­ti­cal criti­cism. On the other hand, “stock responses, if they intervene, are disandvan­tageous and even dan­gerous, because they may get in the way of, and pre­vent, a re­sponse more appro­pri­ate to the situation”. As the foregoing discussion may suggest, the enormous lure of stock responses may prove irresistible not only to some of Lakoff’s less sensitive disci­ples, but to Lakoff himself too. What is at stake here is cogni­tive strategy: rapid or delayed conceptualization. The road not taken would have re­quired exposure to uncertainty for too long; and that might be particularly offen­sive to some per­sons.



5. The “Contemporary” and the “Controversion” theory of Metaphor

Lakoff claims in his paper under discussion that he and his followers have radically changed the state of the art in the study of metaphor. He claims that three of their most outstanding achievements are: (1) the discovery that a wide range of metaphors in colloquial language, literature, dreams etc., can be reduced to a relatively small number of underlying “generative meta­phors”; (2) that the main bulk of these generative metaphors rely on spatial im­agery: abstract, mental and social processes are typically expressed in spatial images; and (3) that their theory is more adequate than the traditional Controversion Theory of metaphor. I have no quarrel with the first claim, though I have my doubts as for its usefulness regarding the analysis of literature. I have no quarrel with the second claim either, except that, as I have already argued, their spatial model is not as ade­quate as it could be. In what follows, I will devote some attention to the third claim.

Consider Lakoff’s following statement:


A major difference between the contemporary theory [i.e., Lakoff’s theory] and the classical one is based on the old literal-figurative  distinction. Given that distinction, one might think that one “arrives at” a metaphorical inter­pretation of a sentence by “starting” with the literal meaning and employ­ing some algorithmic process to it (see Searle, this volume). Though there do exist cases where something like this happens, this is not in general how metaphor works, as we shall see shortly (Lakoff 1993: 205).


Lakoff argues, then, against “the old literal-figura­tive distinction”. In figurative lan­guage, just as in literal language, he says, we need not assume contradictions and meaning-cancelations. Research in pragmatics and ar­tificial intelligence during the past twenty years or so suggests, rather, a reverse posibility: the processing of lit­eral language too is riddled with inferences, implied expectations confirmed or re­futed, contradictions and conditions in which meanings are or are not cancelled. In his “Conceptual Dependency Theory”, Schank (1975: 68-70) uses the “but-test” to distinguish between straightfor­ward assertions and in­ferences. From “John went yesterday to the movies”, an AI program will assume that John saw a movie; however, “John went yesterday to the movies but all the tickets were sold” involves no logical contradic­tion, whereas “John went yesterday to the movies but he stayed at home” does. Notice that the “but-test” cancels in the first instance an inference and not a feature. Das­cal distin­guishes “sentence-based and con­text-based ‘excesses’”, and points out that cancelation is a matter of degree.[6]

It is difficult to argue against Lakoff by counter-examples, since he admits in ad­vance that “there do exist cases where something like this happens”; so, apparently, the only way to refute him would be to analyse all metaphors in the world, and show that the majority of metaphors don’t behave as he claims they do. Since I am not prepared to do that, I propose to try an alternative way, which has its logical shortcomings, but considering the alternative, this would be a viable way to take. I am going to take some of Lakoff’s own examples, see what he does to them, and see what one could do to them with some other, pre-Lakoffean theory. This method does not rule out the possibility that Lakoff has ill-chosen his examples, and that they are among those few “cases where something like this happens”. But, again, consider­ing the alternative, this is the best I can do. In fact, the issue at stake is not factual at all, as this paragraph might suggest. Paraphrasing Morris Weitz (1962), theories of metaphors are no factual statements, but crucial recommendations as for what to look for in metaphors, and how to look at it. When we have two or more such theories, we need not verify them against the facts, but rather carry out the various recommenda­tions and compare outcomes.

I am going to compare Lakoff’s approach to Beardsley’s Controversion Theory of metaphor. According to this theory, “a metaphor is a significant attribution that is either indirectly self-contradictory or obviously false in its context, and in which the modifier connotes characteristics that can be attributed, truly or falsely, to the sub­ject” (Beardsley 1958: 142). The opposition between the two approaches is obvi­ous. One of the advantages Beardsley claims for his theory is this:


The Controversion Theory explains one of the most puzzling and important features of metaphor, its capacity to create new contextual meaning. [...] Sometimes we invent, or hit upon, a metaphor and find that it gives us a new idea. The reason is that the connotations of words are never fully known, or knowable, beforehand, and very often we discover new connota­tions of the words when we see how they behave as modifiers in metaphor­ical attributions. The metaphor does not create the connotations, but it brings them to life (Beardsley 1958: 143).


I would like to make two comments on this statement: that, as we have seen, Lakoff’s system has been devised, by contrast, to create meanings that show little sensitivity to changing context; and that Beardsley’s statement is highly compatible with the conception of spatial imagery I have suggested above. Unforeseen contexts create unforeseen contextual meanings by “bringing to life” unrealised “features” (in my terminology) or “connotations” (in Beardsley’s) terminology. There appears to be only one (rather trivial) difference between Beardsley’s and my formulation, namely, that “connotations” are better suited to “words”, whereas “features” are bet­ter suited to “spatial images”.

Notice that the sentence “John went yesterday to the movies but he stayed at home” serves above as an example of logical contradiction. But, in certain cir­cum­stances, it can be used figuratively to make one of several perfectly meaningful state­ments as, e.g., “John went yesterday to the movies but he couldn’t enjoy it, be­cause he was thinking of his sick child left at home with the baby sitter”; or, “John has a rich imagination: he imagined seeing a full movie without leaving his home”. This is what the Controversion Theory is about. The understander applies the Prin­ci­ple of Congruence to an incongruent text, generating unforeseeable meanings in un­foresee­able cir­cumstamces. Now suppose we supply some relevant conceptual meta­phors (such as going to as imagining, or staying at a place as emo­tion­ally  loaded thinking of it), we also have to supply some rules or princi­ples to account for the understanders’ amazing ease to switch from one underlying metaphor to an­other. We must assume perhaps that when the situation changes, the under­stander perceives some incongruence between the metaphor and its context which, in turn, requires him to substitute one underlying metaphor for another. But then we are back at the Controversion Theory, complicated by the Conceptual Meta­phor Theory.

Now consider Disraeli’s remark, quoted by Lakoff from Searle:


I have climbed to the top of the greasy pole.


I will fully reproduce here Lakoff’s discussion, to make sure I am not making some unfair selection—assuming that Lakoff executes the crucial recommendations of his theory. Then I will do my best to carry out the crucial recommenda­tions sug­gested by Beardsley’s conceptual apparatus as for what to look for in this sentence, and how to look at it; then I will compare the two in light of the two the­ories. Let us begin with the question how can we know that this sentence must be understood metaphorically?


This could be taken nonmetaphorically, but its most likely metaphorical interpretation is via the career is a journey metaphor. This metaphor is evoked jointly by source domain knowledge about pole climbing, which is effortful, self-propelled, destination-oriented motion upward, and knowledge that the metaphor involves effortful, self-propelled, destination-oriented mo­tion upward (Lakoff 1993: 238).


Lakoff admits that “this could be taken nonmetaphorically”, but tacitly assumes that it is metaphorical, then proceeds to the assertion that “its most likely metaphor­ical interpretation is via the career is a journey metaphor”, skipping the ques­tion why we should think of it as metaphorical at all. Beardsley would have a very straightforward and principled answer to that, namely, that it is “obviously false in its context”. Why did Lakoff not come to a similar conclusion? Not because such a conclusion is beyond his logi­cal capacities; rather, I think, because he explicitly  set out in the first place to show that “this is not in general how metaphor works” (Lakoff 1993: 205). So, as far as this specific instance is concerned, Lakoff is forced tacitly to assume that which Beardsley can support by principled arguments. As far as Beardsley is con­cerned, there are here two incongru­ous domains. One of them is a long-term social process, requiring great men­tal and social effort; the other one is a relatively short-term physical movement in space, involving great physical effort. That is why they are incongruent. Now the Principle of Congruence requires us to render the incon­gruent domains congruent. In this case, the modifier (climbing to the top of the greasy pole) connotes such characteristics as “effortful, self-propelled, destination-oriented motion upward” that can be attributed, truly or falsely, to the subject (career).

I have already argued that the existence of a conceptual metaphor does not guar­antee that it is applicable in a certain instance. By claiming that the career is a journey metaphor is evoked jointly by source domain knowledge and by target domain knowledge, Lakoff is assuming that which he has set out to prove in the beginning. If career is a journey, it does not guarantee that a journey is career; and even if the journey is career, it does not guarantee that greasy pole climbing is career. As we have seen, one of the most important achieve­ments Lakoff claims for his theory is that it has superseded the incongruence con­ception of metaphor. This forces him to play down two crucial stages of the pro­cess. First, he must tacitly assume that an expression is metaphorical though it could be nonmetaphorical as well. And second, he must tacitly skip a rather illumi­nating stage of his argument: that mapping must take place between two incongru­ent domains, so as to render them congruent.

The Controversion Theory can be complemented by the “feature-cancellation theory of metaphor”. Those features of the source that conflict with features of the target are cancelled. The cancellation of the conflicting predicates abstracted from the source foregrounds the relevant ones; at the same time it may facilitate the fusion of the two domains. In the present instance, such abstract predicates as difficulty, effort, insecurity etc. are foregrounded, whereas such features as the material of the pole, the colour of the grease, the climbing movements of the climber are cancelled.

But let us follow the rest of Lakoff’s argument:


Part of the knowledge evoked is that the speaker is as high as he can get on that particular pole, that the pole was difficult to climb, that the climb pro­bably involved backward motion, that it is difficult for someone to stay at the top of a greasy pole, and that he will most likely slide down again. The career is a journey maps this knowledge onto corresponding knowl­edge about the speaker’s career: he has as much status as he can get in that particular career, it was difficult to get that point in that career, it probably involved some temporary loss of status along the way, it will be difficult to maintain this position, and he will probably loose status before long (Lakoff 1993: 238-239).


Thus Lakoff. Now what would Beardsley say about the same issues? He would probably say word by word the same, except one sentence. Instead of “The career is a journey maps this knowledge onto corresponding knowledge about the speak­er’s career” he would have said: “the modifier connotes characteristics that can be at­tributed, truly or falsely, to the subject”. And Beardsley’s modifier does, indeed, con­note all those things enumerated by Lakoff. If so, Lakoff and his followers’ claims for great innovativeness is less than warranted. Thus far, then, Lakoff’s con­ceptual system added nothing to Beardsley’s, except substituting the predicate “maps this knowledge onto” for “connotes characteristics that can be attributed, truly or falsely, to”. On the other hand, it had tacitly to skip a range of issues which Beards­ley’s sys­tem can handle in a principled way. Then Lakoff concludes:


All this follows with nothing more than the conventional career-as-journey mapping, which we all share as part of our metaphorical sys­tems, plus knowledge about climbing greasy poles (Lakoff 1993: 239).


Now suppose we ask a panel of judges to list the properties of career and of journey, and then rate them for relevance and salience. Of all the features enumer­ated above by Lakoff, perhaps “self-propelled, destination-oriented” would turn up in journey, and not as the most salient ones. All the burden of un­derstanding this metaphor is laid on the ad-hoc, pre-theoretical notion of “plus knowledge about climbing greasy poles”. Apart from this, Lakoff can merely enu­merate a list of spe­cific “connotations”. Beardsley’s conceptual system, by contrast, is tailor-made for this situation: “The Controversion Theory explains one of the most puzzling and important features of metaphor, its capacity to create new contex­tual meaning. [...] The connotations of words are never fully known, or knowable, beforehand, and very often we discover new connota­tions of the words when we see how they behave as modifiers in metaphor­ical attributions. The meta­phor does not create the connota­tions, but it brings them to life”. The career-as-journey mapping can explain very little about Disraeli’s sentence; it rather adds the noisy el­ement of a journey. Whereas Beardsley’s crucial recommendations provide every­thing one needs for understanding it, from identifying it as a nonliteral statement, through bringing to life dormant connotations, to at­tributing them, truly or falsely, to the subject. What is more, when the conceptual metaphor system breaks down, Lakoff himself resorts to those in­structions.

As Odette de Mourgues (1953) remarked, pigeonholing gives certainty but no insight. Pigeonholing the sentence “I have climbed to the top of the greasy pole” as an in­stance of the career-as-journey metaphor gives, indeed, little insight. Insight is gained not because of it, but in spite of it. In the present instance, the best we can say about it is that it didn’t prevent Lakoff from acting, intuitively, upon the in­structions implied by Beardsley’s theory.



6. Whose theory and principles become un­necessary?

The business of this paper has been to point out two opposite approaches to the handling of meaning in figurative language and symbolic systems, that dominate the history of interpretations both in psychotherapy and literary criticism. One of them works with pre-established meanings; the other one with an indefinite range of potential meanings, changing subsets of which may be realized in changing contexts. These are two alternative interpretative strategies. The respective cognitive attitudes are rapid and delayed conceptualization. They have different advantages and disadvantages. The former is advantageous when speed of response is required, while accuracy and subtlety are less important; the latter is advantageous when the obverse is the case. In light of this summary, let us consider Lakoff’s following paragraph (with reference to another example: “the hours crept by as we waited for the plane”):


Searle accounts for such cases by his Principle 4, which says that “we just do perceive a connection” which is the basis of the interpretation. This is vague and doesn’t say what the perceived connection is or why we “just do” perceive it. When we spell out the details of all such perceived connections, they turn out to be the system of conceptual metaphors I have been de­scrib­ing. But given that system, Searle’s theory and his principles become un­necessary (Lakoff 1993: 239). 


Going back to our discussion of Disraeli’s “I have climbed to the top of the greasy pole”, we should remember this: when we spell out the details of all such perceived connections, they don’t turn out to be the system of conceptual meta­phors. On the contrary, we are forced “to admit that all the work remains to be done in each partic­ular case”. To put it more explicitly: if we ask whether some connec­tion between journeys and climbing greasy poles (e.g., movement in space) can be pointed out, the answer is yes (they turn out to be the system of conceptual metaphors); if we ask whether this exempts us to do all the work in this particular case, the answer is no.

We might re-write, then, Lakoff as follows: “given Beards­ley’s theory and prin­ciples, Lakoff’s system of conceptual metaphors becomes un­necessary”. The rule appears to be this: when one set of principles can account for a metaphor, the other  set becomes unnecessary. Now how can we decide, whose principles or system become unneces­sary? It depends on the circumstances and the purpose of our inquiry. In well-prac­ticed circumstances, rely­ing on the system of conceptual metaphors may be quite sufficient, and Searle’s or Beardsley’s principles may become unnecessary. When, however, unfore­seen circumstances arise, Lakoff’s system may become unneces­sary, while Beardsley’s theory and principles may become all-important. It is illu­minating to observe Lakoff’s own performance in the two test cases we have consid­ered above. When fac­ing Frost’s verse lines, he displays excessive self-confidence, but little literary sub­tlety. When confronted with Disraeli’s sentence, he cannot tell why he thinks it is to be understood metaphorically, but takes it for granted. In his principles of interpreta­tion he is forced to fall back on the ad-hoc, pre-theoretical no­tion of “plus knowl­edge about climbing greasy poles”; whereas in practice, he acts upon the recommendations of Beardsley’s (supposedly false) theory. Obviously, in these test cases, the rival models are better suited.

It is not an easy task to catch Lakoff doing this: the reader accepts his analysis, without noticing that the theory does not cater for it. We have just seen how Lakoff does this. He suggests some abstract logical connection between journeys and climbing greasy poles, indicating that his theory is relevant to this instance; that is just enough to induce the reader to a willing suspension of disbelief. By mentioning “plus knowledge about climbing greasy poles”, he appeals to the trained reader’s willingness to bracket certain issues to preserve the lucidity of the argument: the reader has the knowledge, is shown how Lakoff uses it; and does not notice that he has not been told the principles according to which he is to use it. Finally, suppose Lakoff is right in his claim that “when we spell out the details of all such perceived connections, they turn out to be the system of conceptual metaphors I have been de­scrib­ing”. In such a case he ends up with the barren piegeonholing of the metaphor. The productive stage is that of “spelling out the details” which, as we have seen, is not necessarily taken care of by the system. Now suppose Lakoff adds career is climbing greasy poles to the system of conceptual metaphors. There will al­ways be unforeseen cases that can be related to the existing system in the final act of pigeonholing—after working out the unpredicted details with the help of some model as the ones suggested by Beardsley or Searle. Indeed, I have just been told that Lakoff and his colleagues have recently assimilated the “feature-cancellation theory of metaphor”. Turner’s work on blended spaces using Lakoff’s “invariance hypothe­sis” attempts to formulate exactly how and why such feature cancelletion happens. If you can’t beat them, join them.



7. Metaphor Processing and the Stopwatch

Thus, such New Critics as Richards and L. C. Knights elaborated on the speed element in literary response, anticipating, in the first quarter of the Twentieth Century, the notion of rapid and delayed closure elaborated by psychologists of perception and personality in the late forties. My own contribution in this respect lies merely in pointing out a close relationship between the two. In the second half of the century all this was translated into terms of information processing and experimental procedures measured by the stopwatch in the psychological laboratory. An enormous number of empirical studies was produced, proving that the understanding of a metaphoric expression takes the same time as the understanding of a literal expression. This, of course, was more consistent with Lakoff’s “Contemporary Theory”, that assumed direct understanding, than with the stages postulated by the “Controversion Theory”. Then, however, an enormous amount of empirical research emerged that demonstrated the opposite: the understanding of figurative expressions takes longer than that of literal expressions. Rachel Giora reviewed all this experimental evidence, and came exactly to the same conclusion as I arrived above speculatively. “Both approaches ac­count for only a lim­ited number of findings”. Her concluding paragraph is most illumi­nating (Giora, 1997: 50):


At this stage it seems possible to formulate the conditions under which var­ious processing models apply. Thus, direct process assumed by contempo­rary cogni­tive psychologists, seems to apply when highly salient meanings are intended. For example, the salient figurative meaning of highly conven­tional idioms is processed directly (Gibbs, 1980). Parallel processing ap­plies when alternative meanings are equally salient, as in the case of con­ventional metaphors (Blasko & Connine, 1993), or when less conventional referring expressions are used in­novatively (Gerrig, 1989). Sequential pro­cessing, assumed by the traditional pragmatic model, applies when language is used innovatively, as in the case of novel metaphors (Blasko & Connine, 1993), novel uses of highly conventional language (Gerrig, 1989), novel re­ferring expressions (Gibbs, 1990), or literal uses of highly conventional id­ioms (Gibbs, 1980).


The above speculative conclusions, supported by stopwatch experiments in the psychological laboratory have been reinforced by an fMRI study in the neurological laboratory, suggesting that different brain centers are involved in understanding novel and conventional metaphors: “a unique network, consisting of the right homologue of Wernicke’s area, right and left premotor areas, right and left insula and Broca’s area, is recruited for the processing of novel metaphors but not for the processing of conventional metaphors”.



8. Conclusions

One of the most impressive parts of Lakoff’s theory (for me at least) was its hierarchic organization. “Metaphorical mappings do not occur in isolation from one another. They are sometimes organized in hierarchical structures, in which ‘lower’ mappings in the hierarchy inherit the structures of the ‘higher’ mappings” (Lakoff 1993: 222). Consider the following three levels:


Level 1: The event structure metaphor

Level 2: a purposeful life is a journey

Level 3: love is a journey; a career is a journey


Now, after having written the foregoing discussion, I am somewhat less enthusiastic about it. Judging from Lakoff’s handling Disraeli’s metaphor, it didn’t help him very much to get an insight into its meaning. But after having received an insight by other means, it helped him to pigeonhole it as another instance of the conceptual metaphor system. This may throw some light on the inherent merits of Lakoff’s performance regarding some literary metaphors in this paper. Consider the opening lines from Dante’s Divine Comedy:


In the middle of life’s road

I found myself in a dark wood...,


What does Lakoff have to say on these lines? That light affords seeing, and seeing is knowing, while darkness suggests the opposite of knowing, confusion. The phrase “In the middle of life’s road”, in turn, suggests that life is a journey. I don’t think there was a single critic since Dante’s time to our own times, of any critical school, or pre-theoretical, who thought otherwise. Doesn’t Dante himself say, after all, that it was “life’s road”? The only new information Lakoff contributes is that this is an instance of the conceptual metaphor system. Now I agree that it is difficult to say something different on this example. But then, I didn’t choose the examples to illustrate the explanatory power of Lakoff’s theory. Lakoff chose them.


Tel Aviv University





Beardsley, M.C. 1958. Aesthetics: Problems in the Philosophy of Criticism. New York & Burlingame: Harcourt, Brace & World.

Brooks, C. and Warren, R.P. 1958. Modern Rhetoric. New York.

Burke, K. l957. “Freud—and the Analysis of Poetry”, in The Philosophy of Literary Form. New York: Vintage. 221-250.

Dascal, M. 1983. Pragmatics and the Philosophy of Mind I: Thought in Language. Amsterdam/Philadelphia: John Benjamin.

Gibbs, R.W. 1993. “Process and Products in Making Sense of Tropes”. In A. Ortony (ed) Thought and Metaphor, 2nd ed. Cambridge: Cambridge UP. 252-276.

Gibbs, R. 1997. “Conceptual Metaphors Underlying Conventional and Poetic Language”, paper presented at the “Cognitive Theories of Intertextuality” research workshop at the Porter Institute for Poetics and Semiotics, Tel Aviv University: December 30, 1997 – January 1, 1998.

Giora, Rachel (1997) “Understanding figurative and literal language: The graded salience hypothesis”, in Yeshayahu Shen (ed.), Cognition and Figurative Language. Tel Aviv: The Porter Institute for Poetics and Semiotics. 30-54. Reprinted: (1997) “Understanding figurative and literal language: The graded salience hypothesis”. Cognitive Linguistics, 7/1: 183-206.

Knights, L.C. 1964 [19281] “Notes on Comedy”. In E. Bentley (ed), The Importance of Scrutiny. New York: New York UP. 227-237.

Lakoff, G. 1993. “The Contemporary Theory of Metaphor”. In A. Ortony (ed) Thought and Metaphor, 2nd ed. Cambridge: Cambridge UP. 202-251.

Mashal, N., M. Faust, T.Hendler "The role of the right hemisphere in processing nonsalient metaphorical meanings: Application of Principal Components Analysis to fMRI data". forthcoming in Neuropsychology, Available http:

Mourgues, O. de. 1953. Metaphysical, Baroque, and Précieux Poetry. London: Oxford UP.

Neisser, Ulric (1968 [19631]) “The Multiplicity of Thought”, in P. C. Wason and P. Johnson-Laird (eds.), Thinking and Reasoning. Harmondsworth: Penguin. 307-323.

Ortony, A. (ed). 1977, [19932] Thought and Metaphor. Cambridge: Cambridge UP.

Richards, I. A. 1929. Practical Criticism. New York: Harcourt, Brace and Company.

Schank, R.C. 1975. Conceptual Information Processing. Amsterdam & New York: North Holland/American Elsevier.

Tsur, R. 1987. On Metaphoring. Jerusalem: Israel Science Publishers.

Tsur, R. 1988. “‘Oceanic’ Dedifferentiation and Poetic Metaphor”. Journal of Pragmatics 12: 711-724 (reprinted in Tsur, 1987: 177-190).

Tsur, R. 1998. “Light, Fire, Prison: A Cognitive Analysis of Religious Imagery in Poetry.” PSYART: A Hyperlink Journal for the Psychological Study of the Arts, article 980715. Available HTTP:


Tsur, R. (2002) “Some Comments on the Lakoffean Conception of Spatial Metaphor”. In Thomasz Komendzinski (ed.), Metaphor: A Multidisciplinary Approach. Special issue of Theoria et Historia Scientiarum VI. No. 1. 245-267. Also available online:

Weitz, M. 1962. “The Role of Theory in Aesthetics”, in J. Margolis (ed) Philosophy Looks at the Arts. New York: Scribner. 48-59.








­[1]   This paper does not presume to offer  a wholesale criticism of “cognitive linguis­tics”. It is focussed on one of its forcible statements. Lakoff (1993) modestly claims to speak on behalf of the Contemporary Theory of Metaphor, challenging much that appears in the other papers in Ortony’s (1977, 19932) collection, “many of which make certain assumptions that were widely taken for granted in 1977” (204). Thus, I am not going to consider all the examples ever discussed by cognitive linguistics, only three of them which, Lakoff claims, illustrate the explanatory power of his theory. The business of this paper is to show that the challenge is less of a challenge than it purports to be.

[2]   I have elsewhere discussed at considerable length “rapid” and “delayed con­ceptualiza­tion” with reference to interpretation in psychotherapy and to the interpre­tation of poetic metaphor. See Tsur (1988).

[3]   My “alternative” proposal was elsewhere worked out at great lentgth. See Tsur 1998; Tsur 2002 and, in an earlier version, Tsur 1987.

[4]   I am quoting the transcript of a tape-recording of the session.

[5]   In a recent paper (Tsur 1998) I discussed some additional reasons. There I have also shown in considerable detail that the “efficient coding” hypothesis is far more illuminating of poetic metaphor on the one hand, and of the creative process on the other; that the “conceptual metaphor” points out merely one possible meaning in a wide range of meanings, and not necessarily the most important one.

[6]   Consider the fol­lowing paragraph:


     One characteristic property of the excesses or modifications of the sense [in cases in which the utterance has meaning in excess of (or distinct from) the sense of the sentence uttered] is that they can be usually canceled without generating a contradiction. Thus one can say The dog is on the carpet, but it will not piss because it just came back from the garden, thus canceling the possible warning and request aspects of the first part of the assertion. It is hard to imagine, however, how one could say, without contradiction, The dog is on the carpet, but it is not touching the carpet’s surface. It is important to notice that sentence-based ‘excesses’, though cancelable, are more difficult to cancel  than context-based ones. The ‘hint’ that there was an expectation that John would not come, conveyed by an utterance of Even John came, is very hard to eliminate. In order to eliminate it, one has to imagine a very special context of utterance which is able to retain the assertion (John came) and to filter out the hint. Similarly, the suggestion that the captain is usually drunk, conveyed by The captain was sober today, is not easy to cancel, in spite of the fact that its denial is not logically inconsistent with the ‘sense’ of the assertion. Cancelability, therefore, seems to be a matter of degree (Dascal, 1983: 26).

[7]   Consider the fol­lowing paragraph:


     One characteristic property of the excesses or modifications of the sense [in cases in which the utterance has meaning in excess of (or distinct from) the sense of the sentence uttered] is that they can be usually canceled without generating a contradiction. Thus one can say The dog is on the carpet, but it will not piss because it just came back from the garden, thus canceling the possible warning and request aspects of the first part of the assertion. It is hard to imagine, however, how one could say, without contradiction, The dog is on the carpet, but it is not touching the carpet’s surface. It is important to notice that sentence-based ‘excesses’, though cancelable, are more difficult to cancel  than context-based ones. The ‘hint’ that there was an expectation that John would not come, conveyed by an utterance of Even John came, is very hard to eliminate. In order to eliminate it, one has to imagine a very special context of utterance which is able to retain the assertion (John came) and to filter out the hint. Similarly, the suggestion that the captain is usually drunk, conveyed by The captain was sober today, is not easy to cancel, in spite of the fact that its denial is not logically inconsistent with the ‘sense’ of the assertion. Cancelability, therefore, seems to be a matter of degree (Dascal, 1983: 26).

Back to home page
Back to "Occasional Papers"