Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives

SRB Archives

This review appeared in Volume 8 (2) of the Semiotic Review of Books.

Visual Metaphor in Contest

by John M. Kennedy

Pictorial Metaphor in Advertising. By Charles Forceville. Routledge, New York. pp 233. ISBN 0-415-12868-4

There is considerable discussion today of non-literal depiction. Curiously, two of the most interesting theories being debated are one based on an enthusiastic adoption of theory of metaphor in pictures by Charles Forceville (1996), following Black (1979) and Sperber and Wilson (1986 ), and a second based on a firm rejection of the possibility of any systematic theory of metaphor, including metaphor in pictures. The leading edge of this rejection is provided by Sonia Sedivy (1997), in sympathy with Davidson (1978). Both of these theories make some use of ideas about the relation between metaphor and class-inclusion (Kennedy, 1993; Glucks-berg and Keysar, 1990). I will sound cautionary notes here about all these lines of discussion. I will stress that there are two things for us to do in responding to metaphors in language and in pictures: Pick out what is relevant, and avoid being misled by what is irrelevant. While I think there is a great deal to modify in Charles Forceville's discussion of what is relevant in metaphoric adverts, one important point on which I will commend him is the kind of discussion he provides of our ability to discern what is irrelevant. I will point out that his discussion of what is irrelevant in a metaphoric message has a great deal in common with what creates the gulf Sedivy describes between what we might like to do in theory of metaphor and what we can ever hope to do.

Sedivy's development of Davidson accepts that "a metaphor is like a platypus", an occasion for thought but not a direct indication of an actual thought itself. The connection between occasions for thought and the content of the thought is so loose she concludes there cannot be a systematic theory of metaphors. I suggest there is useful work to be done on theory even if we accept the connections are as loose as Sedivy claims. I will also argue Forceville's application of Black's "interaction" theory is reasonable, but its very reasonableness plays into Sedivy's hands. I will also show class-inclusion analyses of metaphor stop too short to capture the meaning of metaphors.

Theories of metaphor and pictures

Forceville's book on metaphor in pictures is by far the most comprehensive examination of the topic thus far. It debates many of the major theories of metaphor, and many applications of the theories to pictures in adverts, including essays by Barthes (1986/1961), Noeth (1987) and Tanaka (1994). It argues that most analyses of metaphor conclude metaphor is expressed in language, but what is being revealed by the form of words is a metaphoric thought. Thought can be expressed in lots of media, and so metaphors can be expressed in pictures. When Forceville presents a picture of a journalist's pen being shown as a key, we may well conclude the pen is the key to freedom. If we do, pictures can be used as testbeds for theory of metaphor.

Forceville argues Black's "interaction" theory of metaphor offers a good guide to metaphor, and the interaction theory, with some extensions, can be applied verysuccessfully to pictures. Forceville thinks that metaphor involves understanding or perceiving or experiencing one thing in terms of another. Forceville's pictorial metaphors use objects such as parking meters, oars, skeletons, dolphins and airline tickets. These are often examples of everyday, "basic level" categories. Interestingly, Forceville suggests a similarity between two things can be created by the metaphor. Also, context helps tell us what is the primary subject or topic of the metaphor and what is the secondary topic or vehicle being used to comment on the topic: The key is used to comment on the journalist's craft. Context also tells us what aspects of the topic and vehicle are relevant: "They have the Great Wall! But we have Dommelsch beer!" To relay this tongue-in-cheek message in a picture, a European advertising agency set a photograph of the Great Wall rolling over the hills above a similar photograph of a gentler Dutch landscape with a small wall of full beer crates snaking into the distance for a hundred yards or so. We are amused by the claim that Dommelsch and the Great Wall are both wonders of the world! Forceville calls on Sperber and Wilson (1986) to explicate what is relevant in such pictures. The relevance-principle he favours is that messages are framed so no more effort than necessary will be needed to identify the message. Some might say, dryly, that "no more effort than necessary" is especially pertinent to most ads.

At length, Forceville examines pictorial metaphor in memorable, intelligent, attractive, ironic and often good-humoured advertisements for beer, liquor, cars, airlines, energy conservation, electronics, athletic gear, contraceptives and luxury goods. There is no doubt that these are often clever and charming, imaginative and eye-catching, and -- more to the point -- while quite, quite unrealistic they can be forceful and nuanced in their implications. The question is -- do we understand these pictures as metaphors, flagrant or discrete, and if so in what way? A picture of a man in the pose of a ski-jumper who is standing on a giant Air France ticket and flying above snowcapped peaks is surely a version of "our ticket is your skiis !" If the words are a metaphor to the listener so too is the picture to a viewer.

We might happily accept that pictures in ads are versions of what would be metaphors in language. But does this tell us much about theory of metaphor? Black (1979) argues that the topic "men" and the vehicle "pigs" interact in a metaphor such as "men are pigs." He meant that the term "pigs" is understood in this claim in a way it is not understood in "pigs are farm animals". Also, properties of "men" that become relevant to entertain the metaphor are very different than the properties that come to the fore in "men are mammals" and " The men allowed the women and children to the lifeboats first." The comprehension of each term is affected by the presence of the other in the claim. Also, the metaphor works by claiming some properties of the vehicle can be asserted of the vehicle. This theory appears innocuous. Indeed, it is little more than the obvious.


Sperber and Wilson claim that messages are framed to be relevant to the listener. This is certainly one of the goals of the communicator. But it does not follow that no more effort than necessary will be needed to understand the message. Politicians are often elusive. So are parents talking to their children about delicate matters. Lovers often hint. Irony is often grasped after the event. Few of us appreciate serious text on one reading. We can use ambiguity to appear to be enjoined while we are actually avoiding commitment. Adverts too are puzzling on first inspection at times. "Silk Cut" cigarettes are advertised simply by showing folds of expensive fabric, slashed. It takes some initiation to identify the picture's commodity. Another British cigarette is advertised via pictures that involve "hidden figures". The cigarette packet is present but it is hard to discern. Often the hidden packet has to be pointed out to the viewer. Forceville favours the Sperber and Wilson claim that the first meaning that occurs to the recipient is the one that is relevant. This cannot apply well to "hidden" messages, or indeed to any message that requires work even if it is a genuine reward once it is found.

The central features of Black's interaction theory and Sperber and Wilson's relevance theory are useful descriptions of key issues. Black points out how terms shift their emphasis in metaphors. Sperber and Wilson point out the shift ensures the terms are relevant to the message. But in the long run these are just descriptions, and, to be frank, post hoc. They are not actual theories of how interaction occurs, or how relevance is established.

Metaphors in theories of metaphors

Like Forceville, many authors have written that metaphors allow "one object to be experienced in terms of another", or metaphors "project one object onto the field defined by another" or metaphors "let us perceive one object as another". Alas, this is metaphor being described by metaphors. It often gives rise to confusion. Indeed Forceville himself is driven to wonder whether Black suffered slips of the pen because he uses one metaphor of projection -- the topic is projected onto the vehicle -- when Forceville preferred another -- the vehicle is projected onto the topic. Tut! Tut! In fact both Forceville and Black mean precisely the same thing when the metaphors they use are unpacked! Both mean that features of the vehicle are attributed to the topic. This can occur because, metaphorically, the topic has been moved onto the vehicle's field and some of the topic's characteristics then became significant (Black's version of projection). Or, it can occur because the topic received the vehicle, and then the relevant factors were emphasized (Forceville's version).But actually nothing is really moved, and so there is no way to decide between these metaphors of transportation. This is the danger in resting a theory on metaphors -- that readers will fight over them without appreciating what is undecidable. Unwary readers, having trouble deciding what is relevant, do not discern the relevant implicit properties. The lesson to learn here is that in a metaphor the topic and the vehicle are said to have some common properties. Talk of projection or experiencing or perceiving is just a metaphoric way of explaining this matter of common properties. Buyer beware!

One wonders if perhaps there is more to Forceville's championing of Black's argument that novel metaphors "create similarity" -- is this a more "momentous insight", as Forceville puts it? An example Forceville considers is the odd sentence " a poem is a pheasant", from Wallace Stevens. Puzzled, we may find ourselves turning the Stevensline around this way and that looking for a stable meaning. Much the same happens at the edge of knowledge in the sciences, Forceville notes. Are we actually creating a similarity in either case? In the clinch, Forceville retreats from this intriguing possibility to a gentler , but less provocative position, admitting we might say the similarity was "pre-existent-but-waiting-to-be-discovered". As a result, we are left with the conclusion that a metaphor, like a novel scientific theory, asserts a topic A has something in common with a vehicle B.

Common properties and accidents

To assert that some A has some properties in common with a B is not to get far, because everything has something in common with everything else. Poems and pheasants have many things in common, but virtually all of what they share is irrelevant to the metaphor. Both are not bowling balls, for example. Also, neither make much use of glue. Further, neither is likely to be confused with Ugandan mountain whales. Metaphors therefore involve more than a blind search for any common features of any kind. Metaphors invite us to find apt common features -- and ones we can reasonably think were intended in the context. In practice, the sorts of ad hoc categories the Stevens proposition stimulates in most minds are not likely to contain imaginary "Ugandan mountain whales". The search for suitable categories often requires considerable knowledge of the culture of the speaker. Tellingly, it cannot rely on the rather limited set of referents we find as dictionary or literal definitions. Often an encyclopedia is a better guide than the dictionary. But even encyclopedias are not sufficient: "She has a bun in the oven" means she is pregnant for reasons to do with common thought, not Britannica. While we can tell someone "you can dish it out but you can't take it", we cannot say "you can't dish it in". We say "eat up!" and not "eat down!" for reasons that have little to do with "up" as a direction. The reason "he burns me up " means anger and "she lights my fire" means admiration is that a cultural accident has occurred, and there is no surety that someone learning English could guess which means what.

Acknowledging the breadth of influences on metaphors, Davidson (1978) argued metaphors cause us to think, but they themselves do not contain the means to come to a firm understanding of their reference. They are like platypuses -- curious affairs, which stimulate us and occasion ideas. But they do not have a decided meaning. Davidson's account is supported by the fact that we would not be able to decide on the meaning of "lights my fire" and "burns me up" in the absence of general knowledge of the context in which the phrases were used. Recently Sedivy (1997) developed a strong extension of Davidson. She argues Davidson's point should be accepted even if we concur with writers like Forceville that pictures in advertisements can be said to be making claims, and to be metaphoric. In a controversial and sophisticated discussion -- one which is likely to be the centre for swirling debate for some time -- she supports the thesis that in their daily use pictures can be said to have propositional content, and that pictures can be metaphoric. But she says that we can accept this thesis about pictures but firmly deny the prize being sought by analysts of metaphoric pictures -- a systematic theory. In a radical argument, she claims there is nosystem to be found in this domain whatsoever. For Sedivy, there cannot be a theory of metaphor in pictures because there could not be any systematic principles mediating between literal content, contextual factors and background knowledge. Instead there are happenstances to do with the shapes of airline tickets in a particular time, cultural accidents such as the Great Wall, historical particulars about the shapes of beer containers, idiosyncrasies about ski-jumping and boxes upon boxes of tricks. To know metaphor is to know a culture we live within, not a set of general rules that can generate all metaphors.

Sedivy's point is a challenge to metaphor theory of course. Part of its radical import for theory of metaphor in pictures arises from its being aligned with still greater challenges to wider matters. The foundations of cognitive science as a coherent discipline are uncomfortable places to be these days as challenges are raised to the idea that we can systematize any major area of thought in any thoroughgoing way. If metaphor theory goes Sedivy's way, so too may theory of analogy, problem-solving, inference, memory and perhaps even perception. Vervaeke (1997) does not tackle theory of metaphor head-on , but he applies discoveries about the importance to cognition of specific knowledge of particulars to virtually all the areas of literal expression and thought. Vervaeke concludes that the imperative behind three decades of cognitive science to find systematic, natural laws of thinking that can be applied widely is facing defeat. If Sedivy dissolves an empire, Vervaeke may dissolve a world.

Context and intermediate-level accounts

If Sedivy and Vervaeke are correct we are about to live in interesting times for theory of signs. Their accounts of the difficulties facing theories of comprehension are often different, but they converge on this: Comprehension is not dictated by reactions to sentences and pictures in a context-free fashion. We react to any particular event by virtue of its relation to a highly-variable context. We apply our knowledge of context in a reasonable way, and there is no set of rules for what is reasonable.

Black tried to give a general account of metaphors, and ended by describing some of what happens in comprehension of metaphors. Much of what he said is simply uncontroversial. If we try to go further and try to develop a theory about comprehension in the fashion of every theory in cognitive science for several decades we will be trying to develop a general and systematic theory. It will try to explain what people do and when they will do it. This is the theory Sedivy blocks. She says this is impossible. People are influenced by accidents, and contexts. Some theorists like Forceville attempt to argue we do what is "relevant." We react to specific messages, most especially novel ones, in ways that show we are flexible and appreciate the circumstances, Forceville says. But this is simply a way of rephrasing Sedivy and Vervaeke. Far from finessing Sedivy's argument about metaphors and our inability to give rules for reasonableness, Forceville's acknowledgement of the observer's flexibility plays into her hands. Forceville notes that we are "sensible" when we come across novel uses of oars or pianos or references to marriage in pictorial metaphors. We try to find what is pertinent in each use, and what is pertinent changes from use to use. He goes so far as to say vehicles always go through some transformation from one metaphoric use to another. This transformation is only explained in a post hoc fashion. Sedivy claims it has to be. Anything stronger she calls a hopeless quest. And if Forceville goes down so too must the rest of cognitive science, if Vervaeke is correct.

Neither Sedivy nor Vervaeke are denying determinism, I hasten to add. Both think that our decisions and the interpretations we come to are influenced by events. But the events and our comprehension are comparatively open-ended. What is relevant one time is not another. There is no clear way to establish what will inevitably be relevant. Both leave open the way for influences from our own encyclopedic autobiographical knowledge. Interestingly, this makes cognitive science more like a discipline one could call the cognitive study of individuals, historical periods and cultures. The goal of finding natural general universal laws is greatly diminished in importance by Sedivy and Vervaeke, and what is increased in stature is the goal of finding out how people think and communicate in particular locales , in particular times and about particular topics.

Political humour is probably a good case in point for Sedivy. A great deal of satire is bound to a particular country -- or election. Very little of any subtlety can carry over to other periods. Likewise, many degrees of nuance in advertisements are likely to be bound to a special historical current in the affairs of our own particular marketplaces in Bremen, Boston or Bolton. Nuance is specific to a time and a place.At best, only the obvious will be stated in any general laws, the argument can run.

Is there anything of value that is intermediate between Black's mundane general comments on metaphor in the interaction theory, on the one hand, and the alarming attacks from the contextualists, on the other?

I believe there may be. Firstly, we should not overemphasize novelty. Metaphors in pictures, like verbal metaphors, do use stock figures like pigs and fire a lot. Also, the types of figurative expression can be described, and they are not an indeterminate set. Forceville usefully examines metaphor and simile, for example and asks whether there is a sense in which the two can be present, distinctively, in pictures. His question may be answerable , even if Sedivy blocks us from an ultimate, fully-systematic theory. Further, depiction and literal and figurative expression may be characterized as having both a mixture of openness to context, as Sedivy and Vervaeke argue, and yet also definitely closed aspects capable of being described with finite resources, the "invariable forms" Forceville wants to delineate.

Lego is a system with a single block shape that can be combined into as many permutations as we wish.The element and the rules for building are a closed set that offers an open set of permutations. Likewise, arithmetic offers a few rules that permit an infinite number of Arabic numbers to be envisaged. The discovery of polar perspective was based on squares and diagonals but it enables us to portray a landscape stretching into the indefinite distance with an infinite number of objects on it, with an infinite variety of shapes. These systems use a few rules and elements, and so are closed, but each allows infinite numbers of examples to be generated. Further, each example can be subjected to a set of criteria for relevance at a particular time -- a relatively changeable set we must admit under pressure from Sedivy and Vervaeke.This is the kind of analysis that the study of types of tropes can enjoy, with a role for the infinite and a role for restrictions.

The value of this analysis into open and closed sets, in which there is infinity and restriction, is evident when we turn to new media for metaphors. The traditional verbal medium offers a set of types of tropes, like metaphor, simile, tautology, irony and hyperbole. Hyperbole is readily given a pictorial equivalent -- the man as big as a house. Can each of the other tropes be squeezed into pictures? If we pursue this question we may be able to do useful work, intermediate between Black's bland characterization of metaphors in general and Sedivy's barrier to context-free explicit laws of the mind. The lessons to be gained from such an intermediate analysis could surely be enlightening. For example, there may not be easy transfer from every form of verbal trope to pictorial versions, and the failures of some transfers may teach us general lessons.

Tropes in pictures

Tautology is easy in language -- a rose is a rose -- and easy to extend -- a rose is a rose is a rose. But surely tautology in pictures is impossible, for good reasons. A picture of a rose is just a picture of a rose. And a picture of a picture of a rose is just that too -- just a picture of a picture of a rose. Embedding is easy in pictures, but what Ricoeur called the metaphoric "is", evident in tautology, is not distinct in pictures. "This is" and "this is not" are not explicit in pictures. Pictures deal with dimensions, such as long and deep and broad, and so can present shapes in realistic or non-realistic ways. They can present objects in a scene, and so can present objects together that would not normally be juxtaposed. The objects can be found in different locations or times or size-scales normally, so space, time and size can be violated in pictures, for metaphoric purposes. But reference itself is not shown, just pictorial reference. To tackle reference itself we need to put a picture of a pipe, say, above a caption denying the picture is a pipe, or a picture of a rose above the tautology referring to roses. But denial and tautology themselves are not within depiction.

Evidently, we can accept that cognition is open to context while closing some limits on cognition expressed via certain media. In particular, we can determine what kinds of communication are impossible, in addition to showing how open we are to the infinite variety within particular species of metaphor. Metaphoric pictures do assuredly have infinite variety. But they operate within limits -- they only call on a few principles.

Metaphors in general use apt, intentional distortions, in the sense that there is always an extra nuance of meaning in a metaphor. Metaphoric pictures in general use two kinds of distortions -- the object shapes in the picture carry an extra meaning or the use of the depicting elements in the picture involves a non-standard function. Distortion of shape is evident when we can modify a central actor's posture or accoutrements in the picture, showing the hero as having feet of clay, or angelic wings or a platform like an airline ticket instead of skiis. Alternatively, we can use the line element in the picture, which normally stands for an edge of a surface, to stand for, say, an aroma from a flower. But if we want to say something is not to occur or if we want to say reference is to be entertained we need symbols. To announce that the viewer is not to eat here, forexample, we have to add a symbol such as a slash to a picture of food. Notice that the context -- not the picture -- tells us the boundaries to which the "do not" applies. We understand the picture tells us, for example, "do not eat sandwiches here in this library" and it does not tell us "do not eat sandwiches ever again!" The when and the where to which the picture applies are not specified by the picture. In principle, the locus of the reference must be partly specified by the context.

Specificity of reference requires context for the picture. Is it also true that the picture can be specifically one kind of trope rather than another? At one extreme, paraphrasing the picture into one specific statement rather than any other is impossible. It is always possible to rephrase the statement and have an acceptable verbal equivalent for the gist of the picture.At the other extreme, we can be reasonably definite that a given picture uses hyperbole. But can some subtle distinctions in language be reflected , distinctively, in pictures? Consider Forceville's claims about metaphors and similes in pictures.

Metaphors and similes are subtly different in figurative language. Their cousins in literal language have more readily separated functions. The claim "this object is like a banana" is not the same as "this object is a banana". The first literally means it is not a banana, just an object that has some key properties in common with a banana. The second means that it has all the key properties of a banana, and no properties that would exclude it from being a banana. However, in figurative speech, "Jeff is a fish" (a metaphor) and "Jeff is like a fish" (a simile) often are interchangeable.

Forceville contends metaphors make us think the tenor and vehicle are fused (as if Jim had gills) more than similes (which can fit with side-by-side lists of properties of Jim and fish). Consequently, he thinks a picture with a Dutch gin bottle masquerading as a traditional Dutch house is a pictorial metaphor but a picture showing a girl diving and a dolphin leaping, side by side, may be a simile. We could accept that "Bokma gin is my home" is a metaphor, and "a girl in Adidas is like a dolphin" is a simile and both are reasonable verbal versions of such pictures. Forceville asks us to accept no more. But I think we can push a little deeper. Metaphors are not understood as fusing anything. This is a common terminus of many tempting accounts of metaphor, but as a stop it is a misleading temptress -- a semiotic fraud.

Fusion is not the aim of metaphor. "Man is a wolf" is not about a werewolf. The aim of the metaphor is to draw attention to common features of the two, not to mix the two. Hence, arguments that pictorial metaphors are those that fuse the two can be quite inappropriate. Also, the absence of any hint of fusion is no bar to an expression being a metaphor. Accordingly, for principled reasons, we can readily say the symmetry of form shown by a diving girl and a dolphin can be expressed by a metaphor such as "that Adidas girl is a dolphin".

Categories and common properties

Pictures often use common articles, such as pigs and pianos, Forceville notes, but invite us to use some creativity in putting objects into the categories suggested by the common articles. Some might want to go so far as to say the meaning we extract from metaphors, in both pictures and words, is an ensemble of common scripts (such as being on a bus) and common categories (such as chairs, wheels and lifebelts). Like Berlin and others (1974), Forceville speaks warmly of the idea that we use many objects , and engage in many activities, that can be described as fitting common "basic-level" categories. These can be turned to account in metaphors , especially in adverts that must communicate to a wide audience.

The idea that we have many "basic level" categories is hardly more than the point that we have many common experiences, from breathing and eating to holding babies and wearing hats. But the idea that chairs, hands, walls and keys are used in metaphors does not take us far.We do use these common articles in metaphors, but each time we do the issue arises: Which aspect is relevant? These are not fixed elements that can be juxtaposed to get a predictable meaning, a point which Forceville emphasises perfectly correctly. The wall of Dommelsch beer crates would not have the same significance if it were paired with the Berlin wall or Hadrian's wall. Hence, the issue of meaning is not solved by listing common scripts or common objects. Basic-level categories are not Lego blocks of the mind. They are variable in their significance. Indeed, any property--such as drinkable--can vary--for example from drinkable-for-pleasure to drinkable-for-sustenance. If properties themselves are not atomic, Vervaeke's point goes deep.

Forceville is quite correct that we are invited by objects in pictures to put topics of interest such as Adidas, IBM, and the London Underground into new categories."IBM is our beacon", one picture suggests, for example. But can this line of description solve many of the puzzles in metaphor? Categories involve modifiers such as ALL, SOME, THE and A. Are these pertinent to pictorial metaphor? If so, does this indicate the heart of metaphor is class-inclusion?

All roses belong to the category of flowers, and some roses are Peace roses. Language labels levels of classification, and adds modifiers such as ALL and SOME, as part of its explicitness about reference. How can a picture deal with levels of classification? Tropes manipulate levels of categorization figuratively. "Bill and Hillary are just people in the end" and " people are just Jack and Jill in the end" do just this. The first can be used to indicate celebrities are like us. The second can be used to say that everyone, individually, faces the same issues in their life. On the surface, one claim relates a specific Bill and Hillary to a general category and the other relates the general category to the specific Jack and Jill.

Perhaps pictures in certain contexts take advantage of our skill with classification. On the car-ad pages in newspapers, below the CARS FOR SALE heading is often a sketch of a car.What is a reasonable verbal equivalent of this drawing? Not THIS CAR IS FOR SALE. Rather, this is "a car", and it is merely " an example of a car". Noone could be sued for advertising in this car-ad column but not selling a car like the one in the sketch. On the other hand, if there is a photograph of a 1997 Jaguar inside one of the boxes with phone numbers in the column, but the would-be-purchaser finds the vendor has a 1992 model of the car for sale, complaints will fly: "This does not show THE car!"

Evidently, pictures in certain contexts are taken as general and in other contexts as evidence of particulars or groups. The level of classification is to some extent set bythe picture and its context. The context of pictures tells us what is a reasonable interpretation of their purpose. That is Forceville's undeniable point. The fact that what is reasonable is itself open to debate, and changes like quicksilver with the observer's purposes, is what context-ualists hammer home. Between the Scylla of being no more than obvious and the Charybdis of denying any fixed reactions to individual examples lies some fruitful work of analysis. We can assess how different tropes do and do not play in different media, and what this says about reference, propositions, paraphrase and, it is clear from the newspaper examples, conceptualization.

Is conceptualization the cornerstone of metaphor? Is this any different from the contention that metaphors claim topics have properties the vehicle allows us to entertain? I do not think much, if anything, is to be gained by changing from talk of common properties to talk of concepts, alas.

Forceville, in a nutshell, argues the topic and vehicle terms in "men are pigs" stand for categories that have been adjusted slightly from their common use, providing an interaction between the topic and the vehicle. The new categories are ad hoc categories, then, and the claim is that "men" (as adjusted) belong to the category of which "pigs" (as adjusted) are a prime example. The extra use of "pig" is like the use of animal in "a dog is an animal" , which indicates the "class" into which dog should be put. It is not like the use of "platinum" in "platinum is like silver", which is a comparison between two members of one class. We capitalize on this skill with class-inclusion when we see a butterfly in a picture with a thin, expensive, exotic, admirable and elegant watch, and imagine " Lassale watches are butterflies", categorizing the watches as butterflies - exotic, admirable and elegant.

Let us examine the argument here carefully and see what it gains us, if anything.

What exactly is a statement about a class of items? We say "this item is a banana", if it has all of the hallmarks of the class of bananas. In contrast, we say an item is "like a banana" if it only has some of the hallmarks of bananas. If it has only some of the features of bananas, the item and bananas only have "a few common features". Let us call statements based on a few common features statements about similarity. If both metaphors and similes are claims about classes, as a class-inclusion theory avers, then it follows they can be used interchangeably. Of course, if they are only figurative statements then they are actually not true as claims that the topic possesses all the hallmarks of the class indicated by the vehicle taken literally. But, we could protest, if we were defending class-inclusion accounts of metaphor, they could be true about the class indicated by the vehicle if the vehicle is understood as indicating a class in the second or non-literal way. "Pig" might be meant as indicating the class of rude, brutish animals in this second, and non-literal way.

Few would dispute the idea that "pig" can be used in two ways, but other parts of this class-inclusion thesis are contentious. For example, even if "Jim is a pig" is understood by means of taking this statement to be a claim about a class, actual comprehension requires us to go further. "Jim and pigs belong to a class indicated by the word "pig" " is not enough. Rather, the claim is understood when we realize Jim and pigs have common features. The features are ones exemplified by pigs. That is, it is not enough to say "pigs are good examples of a category". Rather, some properties are to be found, we realize. Appreciating the common features may or may not involve some matters of classification over and above the common features themselves. Indeed it is hard to see what extras are bought by adding that the common features can be said to help define a class.

It is easy to see that a term like "pig" can function to have secondary meanings -- it does so in cliches and idioms. But further work needs to be done to establish how a relatively uncliched term (such as seagull) could function as an indicator of a secondary class in addition to its primary meaning:"Jim is a seagull" is obscure or ambiguous. If the obscurity is dispelled by convention, why not simply say so? If it is dispelled by finding common features why add talk of classes? If finding common features is guided by context, is any role played at all by use of "classes" rather than common features?

Some may argue that if metaphors inevitably invoked the modifier ALL then they could be said to use the machinery of class-inclusion. But this line of enquiry proves to be a red-herring, it can be shown. Any feature can be the basis of a statement about classes, as in " my face is red". This is just a statement about a property, but an entailment is that my face belongs to a class of red things. If all talk of properties can be turned into talk about classes, is there anything specific to talk of classes? Perhaps the force of the argument is the role of " all" and "some," since "people are pigs" means all of them belong to the class of pigs and only some pigs are those people. If so, the class-inclusion theory rests on a claim about an implied use of "all". But this is a losing move. The use of "all" is at best figurative here. The "all" is not meant to be taken literally. So "all" actually has to be translated into "some". Then the task is to find why "some" people and "pigs" have common properties. The wheel has come full circle.

It seems there is no advantage to notions of class-inclusion over and above notions of common properties. The topic of a metaphor (pictorial or verbal) tells us what we are to entertain a comment on (people, IBM, Dunlop tires etc) and the vehicle, in concert with the topic, provides an indication of the properties that are the focus of the comment. Talk of class-inclusion is a roundabout way of saying just this. The pictorial metaphor usually shows us a commodity, or an indication of a supplier of some purchasable item, and also many kinds of other objects. Our task as viewers is to ascertain the relevant common properties. The advert provides evidence of the topic and we know there is to be a comment on it: "Mobil oil is..."? The advert gives us features that help us complete the proposition: "...is life-sustaining". The common properties we are to entertain, or else remain at a surface level staring at what appears to be a surgical drip delivering motor oil, are matters to do with supplying vigour and value.

Irrelevant properties, semantics and physical models

Any topic and any vehicle have many properties. We might feel queasy seeing motor oil running through a surgical drip! But Forceville points out that just as we are flexible at finding what is relevant so too we are adept at discarding the irrelevant. How do we "filter out" the irrelevant properties? Saying "peaches-and-cream corn is champagne!" does not imply we drink the corn. If we see the corn in a champagne glass in an advert we search for the apt connection -- the excellence of both. Forceville, admirably, avoidsmany of the dangers of theoretical metaphors such as "filtering" when discussing this issue. Like "fusing", "filtering" is a physical process. It is tempting to discuss sorting out and dispensing with irrelevancies in physical terms such as filtering, discarding and inhibition. But just as "fusion" is not the route to reasonableness, so too irrelevance is not a matter of filtering. Relevance and the search for meaning cannot be reduced to physical terms. This is, ultimately, the basis for the arguments on which Sedivy and Vervaeke insist. It seems to me that when Forceville examines Mobil Oil pictures, and other examples of advert pictures with many irrelevancies ready to come to hand, he does so adroitly, steering away from the dangers of reductionism.

Semantics is not accommodated in physical models, which do not fit the kind of interaction Black described. For example, aspects of "inhibition" may be involved in some models of attention and " filtering out". But we can perfectly well entertain several aspects of a vehicle on a given occasion and realize what is pertinent and what is not. We can actively consider champagne as French, and made from grapes, and something with bubbles, and something liquid, all of which are irrelevant to the claim that one kind of corn is as special as champagne. Indeed, if we became confused and forgot the differences between corn and champagne, and proceeded to fuse them we might end up imagining a kind of mushy, hot, semi-liquid, bubbling yellowish goo -- an unappetising product and not the aim of the advert. The point of this example is that a property can be entertained -- "activated" in a physical metaphor -- and deemed to be "irrelevant". This is not the same as "inhibition."

As Forceville writes, interpretation is an affair of exploration, adjustment, making things coherent and being reasonable. We set aside as immaterial the fact that the skier on the Air France ticket is on one slip of paper, but we ski on two skiis. We are sensible people when we look at metaphoric pictures. We do not hold to unchangeable basic categories. We look for the relevant common properties. Precisely how we do this is a challenge, and the challenge is a profound one for all of cognitive science. If we cannot even see how we might develop the outlines of a theory of reasonableness, cognitive science will likely become a kind of cognitive studies. In cognitive studies, physical theories of the determination of relevance and irrelevance would be understood to be a kind of clockwork toy model of the mind -- a reductionist's model. And other tactics, less reductionist, for ascertaining the common properties underlying metaphors expressed in statements and pictures would be sought. I suggest the search for those other tactics will be enlightening, even if we were to abandon the naturalistic imperative, discussing the meaning of pictures without being quite sure how meaning per se is present in any system.


Barthes, R. (1986/1961) The responsibility of forms. (Trans R. Howard) Oxford: Blackwell

Berlin, B., Breedlove, D.E. and Raven, P.H. (1974) Principles of Tzeltal plant classification. New York: Academic Press

Black, M. (1979) "More about metaphor." In A.Ortony (Ed.), Metaphor and thought. Cambridge: Cambridge University Press

Davidson, D.(1978) "What metaphors mean." In S. Sacks (ed.) On metaphor. Chicago:University of Chicago Press

Forceville, C. (1996) Pictorial metaphor in advertising. New York: Routledge

Glucksberg, S. and Keysar, B. (1990) "Understanding metaphorical comparisons."Psychological Review , 97, 3-18

Kennedy, J.M. (1993) Drawing and the blind. New Haven: Yale press

Noeth, W. (1987) "Advertising: The frame message." In J. Umiker-Sebeok (ed) Marketing and semiotics: New directions in the study of signs for sale. Berlin: Mouton de Gruyter, 279-94

Sedivy, S. (1997) "Metaphoric pictures, pulsars, platypuses." Metaphor and Symbol 12, 95-112

Sperber, D. and Wilson, D. (1986) Relevance: Communication and cognition. Oxford: Blackwell

Tanaka, K. (1994) Advertising language: A pragmatic approach to advertisements in Britain and Japan. London: Routledge

Vervaeke, J.(1997) "The naturalistic imperative in cognitive science." Doctoral thesis, Department of Philosophy, University of Toronto

John M Kennedy is professor of Psychology at the University of Toronto. He is the editor of Metaphor and Symbol . He was President of Psychology and the Arts, Division 10 of the American Psychological Association. He is the author of Drawing and the Blind (Yale University Press, 1993). He recently co-authored a book (with Yvonne Erickson and Monica Strucel) for parents of blind children, From Object to Tactile Picture (SIH: Stockholm, 1997).

Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives