Download as RTF document (75KB)

How is Understanding an Advertisement Possible?1

Trevor Pateman

Abstract. This study extends the theories and methods of linguistic pragmatics (Austin, Grice, Searle, Sperber and Wilson) into the analysis of visual advertising imagery. It extends and revises the approach pioneered in semiology/semiotics by Roland Barthes.

(Superscript numbersx refer to the endnotes, bracketed names refer to the bibliography)

This essay is for the memory of Roland Barthes, who died in 1980. I learnt about semiology in his immensely pleasurable seminars at the Ecole Pratique des Hautes Etudes, VIth section, Paris, in 1971-72. I began to feel that there were not enough constraints on my semiologically inspired connotation readings while doing such readings at the Polytechnic of Central London in 1973/1974. In teaching and staff-student seminars at Goldsmiths' College, London, in 1977/78, some of the dimensions of a rather different approach, were worked out (see my "How to do Things with Images", a Goldsmiths' seminar paper, prompted by Paul Walton). Much of the literature I cite in this present essay, appeared only shortly before Barthes died but Barthes knew it or knew of it. He co-edited Communications 30 (1979) in which both a French translation of Grice 1975 and the Sperber and Wilson article of which I also make some use appeared. In the Presentation to that issue, he writes of 'des approches scientifiques de la conversation ...Elles relevent de cette partie de la semiologie, longtemps delaissee, qu'on appelle, a la suite des auteurs anglo-saxons, la pragmatique' (Barthes and Berthet 1979, p.4). I don't think that what I am doing in this paper is a break with the kind of work he pioneered, and from which everyone working in media and communication studies has learnt and will continue to learn.


Anyone who reads this essay will routinely accomplish the identification, understanding and criticism of advertisements encountered while watching TV, listening to radio, sitting in the cinema, reading a magazine, walking past billboards, and so on. In this essay I offer an analysis of some neglected conditions of possibility of this routine accomplishment. I do so because it seems to me that failure to undertake such an analysis or to realise (even to deny) its significance is a serious limitation upon the value of the hermeneutics and critique of advertisements which has been produced within broadly structuralist and semiological paradigms, say from Barthes' essay, 'Rhetoric of the Image' (Barthes 1964a) through to Williamson's book Decoding Advertisements (1978). These tend to take for granted important conditions of possibility of the routine accomplishment, proceeding directly to a hermeneutics or critique which consequently has an unnecessarily hazardous character, inviting the question 'How do you know?' Specifically, they tend to ignore those conditions of possibility which distinguish instances of speech or utterances (parole) and their comprehension from the or a language (langue), to recall the Sausurrean distinction (Saussure 1959; Barthes 1964b, Ch. I), and so analyse utterances as if they were languages, which they are not.

What I am sketching out in this essay belongs to a wider endeavour by many theorists to rethink the theory of the (formal) organisation of texts and images in terms of a theory of the active comprehension of texts and images in context,2 and thereby to complement theories of formal systems (linguistics; semiology) with a general theory of communication, itself part of a general theory of action. Such a general theory of communication has as one of its central problems, 'to build a coherent picture of the clearly heterogenous processes by which utterances are interpreted' (Smith and Wilson 1979, p.150).

For any activity X which humans accomplish, Kantian questions of the form, 'How is X possible?' can be asked. Philosophers differ as to whether answers to such questions should specify necessary, necessary and sufficient, or just sufficient conditions. I have been influenced by Sloman 1978 (especially ch. 2) and see the conditions being analysed in this essay as sufficiency conditions. But because I shall sometimes be arguing that semiological and structuralist analyses are insufficient to account for our understanding of advertisements, what I offer in their place will often read like a statement of necessary conditions. At this stage of things, it is perhaps not so important to settle this issue as to get on with providing an alternative framework to that now dominant in media and communication studies. The status of this alternative framework can come up for discussion on another occasion.

Identifying Advertisements

In general, it is easy to identify something as an advertisement, and much less easy to say how it is done, might be done, or must be done. Identification is even easier than I suppose in an earlier study (Pateman 1980c), for advertisements are rarely identified in isolation and retrospectively, but rather they are identified in a context where they have been anticipated.3 For example, if we are watching private-sector TV, we know that sooner or later we will be taken through an advertising break. In other words, viewing, like most or all human activities, has an anticipatory character in the sense that it has expected future components, and having the expectation of the occurrence of those components is part of what it means to be engaged in that activity.

A model for what is involved in this kind of projective activity is provided by Schank and Abelson's theory of script based understanding. They describe a script as follows:

A script is a structure that describes appropriate sequences of events in a particular context. A script is made up of slots and requirements about what can fill those slots. The structure is an interconnected whole, and what is in one slot affects what can be in another. Scripts handle stylized everyday situations. They are not subject to change, nor do they provide apparatus for handling totally novel situations. Thus, a script is a predetermined, stereotyped sequence of actions that defines a well-known situation.

(Schank and Abelson 1977, p.41)

Because we actively anticipate that something will occur to fill a script-defined slot, we can use minimal and diverse cues to decide that something now occurring fills a slot in a script. This is certainly recognised by advertisers: they certainly don't have to do much to get us to recognise anything appearing on the back-cover of a Sunday newspaper colour supplement as an advertisement. However, it should be stressed that however 'obvious' it may be that something is an advertisement, there is always an inference to be made from the cue provided to the decision that something does indeed fill an advertising slot (i.e. count as an advertisement). Even when an advertisement is labelled with the word 'Advertisement', the reader has to add the premise or assumption that the word is being used with its literal meaning to perform an act of classification, etc. in order to conclude that the word 'Advertisement' labels an advertisement, since the word could be used otherwise. (The word 'Advertisement' incorporated in a fine-art poster does not label an advertisement.)4

No doubt it is because identification of advertisements is so easy and 'automatic' that its mechanisms are taken for granted in many analyses. I have said something about these mechanisms just because I want to argue that if we did not accomplish the identification of an advertisement as an advertisement it would be strictly impossible for us to understand or criticise it. This is a claim I attempt to substantiate in the next two sections.

What is Advertising? and What is an Advertisement?5

When on the basis of some clue something is taken to fill a slot in a script, various sorts of knowledge about the kinds of thing which fill that slot can and will be accessed and used in the analysis of the particular slot-filler - if it is analysed at all: we often we skip past slots filled by advertisements.

This knowledge will be (part of) our knowledge of advertising as a practice or activity type, to borrow a concept from an important paper by Levinson, who describes it as follows:

I take the notion of an activity type to refer to a fuzzy category whose focal-members are goal-defined, socially constituted, bounded, events with constraints on participants, setting and so on, but above all on the kinds of allowable contributions. Paradigm examples would be teaching, a job interview, a jural interrogation, a football game, a task in a workshop, a dinner party, and so on.

(Levinson 1978, D-5)

To this list of activity types I am now adding advertising.

Now it is clearly the case that each of us knows different things about any given activity type, and that on different occasions, different bits of our knowledge are accessed. In the case of advertising, it is often argued in justification of media studies courses that the more we know about it (history and structure of the advertising industry; production processes; campaign strategies; visual and verbal techniques; etc. etc. ), the better off we are in uderstanding and criticising individual advertisements as we encounter them. Doubtless this is true. However, what I want to stress is the (minimal) knowledge about advertising without which we would not be able even to get started on the business of understanding and criticising advertisements. Such (minimal) knowledge forms part of the competence of most of the world's population, very young children included. It does not have to be taught in school.

This (minimal) knowledge concerns the point or purpose of advertising, which assigns the point or purpose with which any individual advertisement is produced. This point or purpose is to sell products. Individual advertisements are paradigmatically a means whereby the sale of particular products is promoted, and through the advertisement the (anonymous) advertiser addresses (or interpellates: Althusser 1971) any actual reader or viewer as a potential consumer of the product in question.6

This minimal knowledge is required to get started on understanding and criticising an individual advertisement. Of course, we also require knowledge of a language and knowledge of a culture in order to accomplish an interpretation of an advertisement. But knowledge of a language and a culture are insufficient to understand an advertisement. We must also assign the point or purpose with which we believe a given text or image has been produced, and to understand such a text or image as an advertisement is precisely to assign it to the activity type, ADVERTISING.

It is often argued that a text or image must possess certain 'formal' properties in order to count as belonging to a particular genre - say, haiku or limerick. However, even if necessary, such formal properties are not sufficient to classify a text in a particular genre, for one can have 'accidental' haikus and limericks.The formal properties of a text can be mapped into a genre regardless of authorial intention or activity type. More radically, it can be argued that it is only because of the genre assignment that we pick out certain formal properties as the relevant properties which then confirm or disconfirm our initial genre assignment. This is the position I want to defend in the next section, by showing that the relevant 'formal' properties of texts and images used in advertisements can only be specified on the basis of the recognition that they are being produced in advertisements, that is as belonging to the activity type, ADVERTISING.7 This does constitute a fairly major anti-formalist claim and should be considered as opposed to those versions of semiotics, semiology and stylistics which present themselves as formalisms.


In this section I give two examples to show how identification of something as an advertisement is involved in specifying the operative structure, meanings and illocutionary force (see Searle 1969; 'illocutionary force' is roughly equivalent to 'function') of texts and images used in advertising. And I begin with an example which shows how identifying something as an advertisement can allow the reader or viewer to complete an incomplete text or image.

Example 1: The Absent Product

To know that something is an advertisement is to know that it is an advertisement for something (some product). Because of this, advertisers are able to leave out of advertisements the name or any icon of the product they are advertising, relying on the reader or viewer to make a 'default assignment' (Schank and Abelson 1977, pp.38-42) and infer the name of the product being advertised on the basis of whatever clues are provided by the advertiser in the text or image, together with knowledge about the world of products, past advertisements, etc. (At the time this essay was first written, examples were provided by White Horse whisky and Benson and Hedges cigarette advertisements).

As far as I can see, this default assignment of a product, prior to further interpretation of the explicit text or image, is strictly incomprehensible within the structuralist or semiological frameworks which have generally been employed in media and communication studies.

One facilitator of the process of default assignment is the reader's or viewer's knowledge that an advertisement is a thoroughly prepared act of communication which is unlikely to have omitted a product's name by mistake (as a 'performance error' to use Chomsky's expression), but rather has done so deliberately. Consequently, to be compatible with the point or purpose of advertising, the advertisement must and will contain deliberate clues permitting the recovery of the product name. This makes the process of default assignment of a product into a rational (and often enjoyable) activity in which to engage (compare Pratt 1977 on the preparedness of literary texts).

From the point of view of further understanding of the advertisement, identifying the product is indispensable - it is around the product that the discourse of the advertisement is built. The product is thus a structuring principle or function as is shown in the examples below.


Suppose the four words and full stop "Jingle bells this Christmas." were given to you, written on a piece of paper, and you were told that they were a transcript of a continuous stretch of writing occurring in an undefined context (so the situation is like Katz's 1972 'zero-context' semantics). Several surface structure syntactic analyses of this fragment are possible, including at least two ((2) and (3))8 which yield well-formed sentences, and one (1) which renders a well-formed noun-phrase.

  Jingle bells this Christmas
1 nominal (song title) deictic noun
2 imperative
(direct object)
deictic noun
3 imperative
deictic noun

I did not see these words in a zero-context but

(a) on a billboard poster

(b) at Christmas time (1979)

(c) accompanied by the British Post Office's Buzby character (this at a time when the Post Office ran the telephone service)

I now argue that this is sufficient information to allow a reader to assign analysis (2) to the words 'Jingle bells this Christmas' as the operative reading in this instance, and this may be the only reading which occurs to readers, though they would have missed something if they did not also think of structural analysis (1). In analysis (1) the words 'Jingle bells', appear as a nominal compound or 'fixed syntagm' (Saussure 1959) and this is how they are often encountered. In the advertisement, however, the fixed systagm is broken up in an amusing way and given an unusual operative syntax, semantics and pragmatics by the Post Office. The nominal compound becomes a verb and direct object which addresses the billboard viewer.

In a moment, I shall consider what difference is made by the further textual material - 'Make someone happy with a phone call.' - and the fact that Buzby is holding a telephone. For the present, consider how even without this material, analysis (2) above rather than analysis (3) can be assigned, in virtue of the fact that among other things 9 we (as readers) know that:

Known fact i) 'Jingle bells this Christmas.' appears in an advertisement, and advertisements are attempts to get us to do something -paradigmatically, to buy products.

Known fact ii) the Post Office (identifiable as the advertiser from Buzby alone) sells telephone calls, and 'bells' can be analysed as metonymic for telephone calls.

This analysis can be novel, that is, not one we have previously produced and we don't need a dictionary of metonyms to produce it. (See next section.) In contrast, the Post Office has no interest in appealing to bells to ring, which is what the vocative analysis (3) - roughly equivalent to "Jingle, oh ye Bells!" - is about.

Now in Known fact i) above we refer to facts about advertising and in Known fact ii) to facts about a product advertised. Neither of these are facts about language. But if I am right, they are sufficient to decide (or disambiguate) the operative syntax, semantics and pragmatics of a text. In the present case, they could have determined what is the operative syntax ('Bells' as direct object rather than vocative); the operative semantics ('Jingle bells' means '(you) jingle bells!' rather than naming a song or meaning 'Let bells jingle'); and the operative pragmatics (the textual utterance is addressed to us who read the advertisement rather than to bells). That's quite a lot of linguistic work done by non-linguistic knowledge!

Against this line of analysis, it can rightly be pointed out that the advertisement I am considering contains both a disambiguating text ('Make someone happy with a phone call') and a disambiguating icon (Buzby is holding a telephone). These double the message of 'Jingle bells this Christmas' and 'anchor' its operative meaning (cf. Barthes 1964a, esp. 38-40). So even if 'Jingle bells this Christmas' could be understood in the way I have suggested, maybe it can also be understood in the way semiologists have suggested - through semantic material (text and image) actually present in the advertisement. For some readers, this material may be redundant, in the sense of information theory, but that is just their particular good fortune or bad luck, depending on how you evaluate the possession of competence in decoding advertisements.

However, it seems to me that an account which says that one text or image 'anchors' or 'disambiguates' another is vulnerable to a set of objections consistent with the alternative approach I am sketching here. In particular, how do we know that 'Make someone happy with a phone call' disambiguates 'Jingle bells this Christmas' and does so in just the way required? First, it is necessary to see the two texts as connected and then to see them as connected in a particular way. Here we make use of our pragmatic knowledge about texts in general and advertisements in particular. And, second, far from the second text disambiguating the first, which would require that in some mental dictionary 'Jingle bells' was entered as a semantic equivalent of 'Make ... a phone call', it is rather because we can work out pragmatically that 'Jingle bells' means 'Make a phone call' that we can use the second part of the text to confirm a reading of the first part. In other words, the disambiguation has already been done by the time we are able to say that one sentence disambiguates another one. The 'anchorage' one text provides for another is essentially illusory, since 'anchorage' is only of use to those who know how to anchor (compare the earlier remarks on genre and n.7, and my subsequent remarks on the Dictionary Fallacy). It is, of course, entirely possible that some advertisers have the same mistaken theories of textual understanding as some analysts of advertisements, and this would account for the fact that they put into advertisements material which apparently provides evidence for the analysts' theories.

The fact that Buzby is shown holding a telephone can be analysed along the lines indicated in the preceding paragraph and in the next section.

A rather briefer way of putting all this would be to say that this advertisement could have still worked without the supplementary text, "Make someone happy with a phone call" or without the Buzby icon. It may have needed something to locate the advertisement as an advertisement for phone call making, but it is not fixed in advance what would be sufficient to achieve that. It depends on the product's previous advertising history.

Pragmatic Structuration of the Semantics of the Photographic Image

Parallel arguments to those just given apply to iconic or analogic material in advertising, of which photographs form a sub-class.10

Imagine a typical still-life painting - say, a composition of apples, cheeses, a pheasant, a wine-bottle, etc. Now imagine reproductions of the same still-life used in advertising, successively, apples, cheese, pheasant, wine - in each case, let just one word be added outside the frame of the image. If the cheeses are Stiltons, the word could be 'Stilton'. If the wine is "Beaujolais" then "Beaujolais" , and so on. My argument is that as the signifying matter of the image is transposed from the activity 'fine art' to 'advertising', so the relationship between the 'elements'11 of the image changes, according to the product which is made thematic or topical.12 In other words, the changes in context effect a structural reorganization of the operative meaning (the operative semantics) of what is (in some sense) 'the same image'.

The determinations of meaning here are contextual (pragmatic) rather than structural (systematic in the sense of belonging to la langue: compare Pateman 1973). The products which are not advertised (not topical or thematic) assume the structural relation of Comment or Support for the product advertised, and as the topic or theme changes, so too does the comment made or support offered. In some ways, this process is analogous to the way in which changes in stress-assignment alter the focal scale of semantic entailments of an utterance (Smith and Wilson 1979, ch.7).13 Before we can understand the kind of comment which is being made on the theme or topic of the advertisement (the product being promoted) we have to understand the topic/comment structure, which always exists even though the topic/comment elements may not be materially separable (as when highlighting or the angle at which it is photographed constitutes the comment on the thematic product).

Beyond Connotations

The contrast between denotative (or literal) and connotative meanings plays a central part in structuralism and semiology (Barthes 1964a; 1964b). I have already argued that operative denotative meanings of an advertisement are not specifiable without reference to contextual (or pragmatic) variables. It remains to say something about connotative meaning.

It seems to me that at best, the idea of connotation is a 'dummy' concept which identifies a domain the mechanisms at work in which have still to be mapped out in detail. At worst, and quite often, connotation is taken to be a straightforward semantic function or operation in which connotations figure as properties or entailments of texts or images, rather than as the result of structured operations performed upon texts or images by knowledgeable readers. It is the latter, pragmatic, view which I believe is the correct one, though some of the things called connotations do indeed have a semantic character.14 In the rest of this section I shall focus on those connoted meanings which clearly have to be handled pragmatically, and shall do so exclusively in connection with the relation between topic and comment.

Once again I shall ask you to imagine a fragment of an advertisement rather than use a real advertisement. I do this because actual advertisements are usually quite complex, whereas I want to make my points as simply as possible. So imagine an advertisement for cigarettes in which some people are shown smoking beside a natural, rurally-located waterfall. In this advertisement, the brand of cigarette advertised forms the topic (or theme), and the waterfall is (part of) the comment. Of such an advertisement, semiologists might say something like either A or B:

A. 'The denoted image of the waterfall connotes a healthy, outdoor life'

B 'The denoted image of people smoking by a waterfall connotes smoking as part of a healthy, outdoor life.'

If semiologists go on to consider the 'psychological' effects of this advertisement, they may say things like:

C By some metonymic substitution, we can come to believe that cigarette smoking stands for the healthy outdoor life;

D By some logic of identification, we can come to believe that cigarette smoking is the healthy, outdoor life (the problematic of totemism: see Levi-Strauss 1962).

I don't want to consider C and D, fascinating though they are, but instead concentrate on A and B.

The first set of points I wish to make arises from the fact that, at the very least, making statement A supposes an ability to select from the possible connotations of waterfalls the operative one. This selection may be facilitated by a verbal message (which Barthes says 'anchors' the meaning: Barthes 1964a), but such verbal messages are neither necessary or sufficient to achieve disambiguation, which also has a pragmatic aspect. For what is also required is a concept of the relevance of the comment to the topic or theme, compatible with assumptions about the point or purpose with which the image is being produced.

In important work, Sperber and Wilson have sought to reduce Grice's four famous conversational maxims15 to a single axiom of relevance. They argue that in the standard case, the hearer (reader, viewer, etc) treats as axiomatic that, 'Le locuteur a fait de son mieux pour produire l'enonce le plus pertinent possible') (Sperber and Wilson 1979, 89; see also Smith and Wilson 1979, ch.8) - the speaker has done their best to produce the most relevant utterance possible. Sperber and Wilson are thinking of 'relevance' as relevance to an ongoing discourse, and I think this can be extended to cover the bearing of a comment on an (already posited) theme or product - that the connexion is not temporal but spatial in an image does not matter. 'Relevance' is a very difficult concept to analyse, and it would take me too far afield to analyse it here.16 (See on this website my critical notice of Sperber and Wilson's later book, Relevance). The points I want to make can be made informally, using the earlier-introduced idea of point or purpose.

The point of advertising cigarettes is to sell them. The main obstacle to selling cigarettes is consumers' beliefs that cigarettes ruin your health. The most relevant thing a cigarette advertiser can do, given the point of advertising, is to attempt to modify, eliminate, or repress that belief. It is the consumers' ability to work this out (using knowledge of activity types, and something like Sperber and Wilson's axiom of relevance) which allows the correct (intended) connotation of the waterfall to be located - a connotation which bears directly on the chief obstacle to selling cigarettes.

Furthermore, and importantly, this way of analysing how we figure out connotations allows advertisers to make, and readers to recognize, novel connotations which do not conceivably have a place in some pre-existing 'Dictionary of Connotation' (cf. Pateman 1973). The Dictionary Fallacy, as a particular form of the semantic or formalist fallacy which thinks that all meaning is 'in' the text or image, bedevils semiology. For example, in the propaedeutic analysis of a single advertisement for Goodyear tyres, with which she begins her book, Williamson (Williamson 1978, pp.18-19) discusses the part played by a jetty as a comment (in my terms) on Goodyear tyres. She concludes, 'A system of meaning must already exist, in which jetties are seen as strong, and this system is exterior to the ad - which simply refers to it' (p.19). Now this is just wrong since it is incompatible with the idea that there can ever be new connotations (cf. Fraser 1979 on novel metaphors). What I would say in this case is that using (or even inferring) our non-semantic knowledge about jetties, advertising, Goodyear tyres, and our pragmatic axiom of relevance we can figure out (infer) that in a given instance this jetty on this occasion connotes strength.17 This inferential approach is superior to the semantic / formalist approach not least because it avoids assigning to each of us pre-existing dictionary knowledge of a potentially infinite number of semantic properties possessed by any given object .

The second set of points I want to make bears on the specific way in which theme and comment are related, as in connotation B of the cigarette advertisement ("Smoking as part of a healthy life"). How is that in B the producer of the connotation is not seen to intend it ironically or satirically? How is it that we interpret the comment as intended (in virtue of what we believe the advertiser believes we believe about waterfalls) to enhance the status of cigarettes, not diminish it in virtue of the blatant incongruity of the juxtaposition of cigarettes and waterfalls? Quite clearly, it is our knowledge of the point or purpose with which cigarette advertisements are issued which is indispensable to working out the kind of connotation which is appropriate, together with our sense of the relevance which the comment has to the discourse of the advertisement.18

Understanding that the comment be interpreted as enhancing or supporting the theme or product does not, of course, commit the reader or viewer to acceptance of the relation communicated, any more than the advertiser is committed to believing in the relationship they sincerely attempt to communicate. The anonymous and non-reciprocal nature of advertising makes it generally impossible for the consumer to dialogically challenge the advertiser's relation to the claims made and connotations produced, though this is a handicap to advertisers as well as an asset. For advertisers also have few ways of defending themselves in an advertisement against a reader's disbelief in the sincerity with which the beliefs vehicled are held. One way they do have, however, is precisely to present the theme/comment relationship self-reflexively ('these are the sorts of ludicrous claims advertisers make about this product'), thus thematising the relationship between advertisers and consumers and presenting themselves as 'honest brokers'. This is also a way of selling products (the best way to sell second-hand cars is to say, 'Who'd buy a second-hand car from someone like me?').

The third set of points I want to make concerns the elusiveness of connotations and the felt hazardousness of interpretations and critiques of advertising. I think I have an explanation for this felt hazardousness, and this explanation in turn illuminates some interesting features of ideology and ideological conflict. For the hazardousness is real and, in principle, ineliminable.

Basically, connotations are hazardously attributed just because they are pragmatic implications of a produced text or image rather than semantic entailments. (Smith and Wilson 1979,chs. 6 and 7). The difference is roughly this: a semantic entailment of a sentence (text, image, discourse, etc,) cannot be denied by someone who uses that sentence in an utterance without it being the case that they contradict themselves. So, for example, I cannot both assert that 'Elizabeth I was Queen of England' and deny that there was someone who was Queen of England. Nor it seems can I say of a woman, 'She is just a housewife' and deny that I'm thereby expressing a low opinion of the woman and/or housewifery - it is part of the semantics of 'just' to express that kind of attitude. You could point this out to a foreigner learning English, for instance. (Of course, 'just' is ambiguous and also means something like 'precisely, wholly, exactly', but that is not a particular problem). In contrast, pragmatic implications of an utterance are worked out by hearers on the basis not only of linguistic knowledge, but on the basis of assumptions they make about the speaker's intentions, the principles (e.g. of relevance) governing the conversation, activity type, point or purpose, and so on. Not only is there a considerable possibility of error based on mistake about speaker's intentions, about operative conversation principles, activity type definition, etc., but also it seems that the inference to pragmatic implications is probabilistic rather than a matter of strict entailment. One major consequence of this is that it is always possible, though not always plausible, for a speaker to deny that they intended a pragmatic implication drawn by the hearer, and to make this denial without self-contradiction. In other words, some implications are cancellable. Logic, in contrast, is not like that!

For example, a hearer who makes the inference that a particular conversational remark has, say, racist or sexist implications, and draws attention to them, will often encounter the response, 'But I didn't intend them', and this response is neither necessarily insincere or necessarily irrelevant. In many cases, the best the hearer can do in response to the speaker's defence is to say, 'Well, now you know that if you say things like that you are likely to be misunderstood'. One thing this example shows is that even in a perfectly homogeneous speech-community, in which every speaker paired sounds and meanings in exactly the same way, misunderstandings could arise between speakers on the basis of different world knowledge and employment of different inferential strategies.

In relation to advertising, the possibilities of misunderstanding, and of its strategic exploitation (see below), are greatly increased because of the anonymous and one-way character of advertising communication. The reader or viewer who draws a particular implication cannot usually check out whether the implication was intended or whether they are 'reading into the ad. more than is there', and their position is doubly difficult because it is unclear to whom intentions (which are not the same as assumed point or purpose) should be attributed, if indeed to anyone (cf. Grice 1957 on traffic lights). It seems to me that the ideological significance of these banal facts has not been sufficiently appreciated. I shall try briefly to spell out what I take it to be.

First of all, there is little readers can do to reduce their uncertainty about the implications of an advertisement. They cannot ask the advertiser, 'What exactly do you mean?'19 And even if they could, the advertiser could reply 'Nothing and Everything' - in other words, deny that they intended an actual interpretation, and accept all the different interpretations offered by creative receivers - thus adopting the position of the guru who says, 'Il faut y mettre du sien'. 20 This tends to place the locus of interpretation solely in the receiver, and this is just as much an error as to place it wholly in the text. It is essentially a denial of responsibility, and the passing of responsibility where it does not (in this case21) rest.

Advertisers get consumers to do their dirty ideological work for them, and keep their own hands clean. Of course, if the consumers refused to play the game advertisers would have to change direction (cf. Parkin 1971).

Second, the anonymous and non-reciprocal character of the communication opens the doors wide to what Habermas calls 'strategic action', in contrast to 'communicative action' (Habermas 1979b, esp. pp.116-123; see also Bach and Harnish 1979, pp.97-103). If strategic action (manipulation) is defined in terms of intention to achieve an effect where the successful achievement of the effect is partly dependent on the hearer's non-recognition of the intention to achieve that effect, then advertisers are well-placed to engage in strategic action. This is because (a) they cannot be required to spell out their intentions; (b) in the light of what was said under the first point above, it doesn't make much sense for readers to go looking for intentions; (c) many of the intentions which advertisers have are undoubtedly deniable without self-contradiction in virtue of the 'open ' pragmatic implications which they are getting people to draw. We can do something about (a) and (b), but (c) will always be with us, and interpretation will always be a hazardous affair, even if its uncertainty can be reduced by attention to the conditions of possibility I have written about in this essay.

The Pleasure of the Advertisement22

A possible objection to the relevance of my emphasis on the cognitive work people do in understanding and criticising advertisements is to say that people don't bother to understand or criticise them: either they consume them passively or ignore them. Actually, I think this an empirically false claim, and that most people actively enjoy advertisements. What is enjoyable about advertisements is relevant to this chapter, and I shall therefore briefly discuss it.

First, many advertisements are just visually pleasurable - they are quite possibly a substitute for encounters with other art forms.23

Second, they are pleasurable as discourses (both verbal and visual) partly because they call upon some of our more sophisticated linguistic and cognitive competences: the anatomy of advertising may well be the key to the anatomy of human communication! All the figures of classical rhetoric can be found in the texts and images of advertisements (see Bonsiepe 1961 for a classification of visual metaphors in advertisements), and just understanding such non-literal messages (or, more accurately, understanding messages non-literally) gives us the pleasure of exercising our competences. Again, in their incompleteness, many advertisements provide the same kinds of intellectual pleasures as crossword puzzles.


The structure of some of the competences advertisements allow us to exercise with pleasure has been the subject of this paper. Those competences cannot be studied as part of la langue. They are not part of it. In philosophy of language, linguistic pragmatics, and cognitive psychology these competences are being actively studied. Reference to them is involved in any science of signs-in-use, that is, in any theory of communication. The development of such a theory, as part of a general theory of social action, must surely be the first priority for teachers in the emerging fields of media and communication studies. Only by thinking through the conditions of possibility of their own activity can they move from appearances to ever deeper levels of reality, which it is the object of all sciences to uncover.


For bibliographic references, click here

1. Publication details for an earlier version of this essay are given at the end of the Endnotes. This website version is based on a version prepared around 1990 for a book which never appeared.

2 I owe this way of putting things to a lecture by Giovanni Carsaniga at the University of Sussex, December 1979. He also commented in detail on the drafts of this essay.

3 Though I had read about the anticipatory character of human cognition, it was a conversation with Susan Ervin-Tripp which brought home to me the significance of the idea.

4 Though I have stressed contextual anticipation and minimal cues it is, of course, true that people can figure out that something is an advertisement either out of context or in context when all the usual cues are missing, but not both together ( in which case the status of a text or image becomes undecidable). Schank and Abelson discuss non-script based understanding in terms of 'plans' and 'goals', and I refer the reader to their book for further discussion.

5 This section corrects the analysis offered in Pateman 1980c, though elements of that earlier analysis can be incorporated into this new account.

6 For present purposes, I can legitimately ignore other kinds of advertisement which are designed to promote an image, activity or doctrine, rather than sell a product. Clearly, we are quite competent to understand the differences among sub-types of the general activity type which is advertising.

7 The concept of 'genre' is inadequate just because it conflates the concept here called 'activity type' and concept of 'formal property of the text'. My position is that activity type is specified in terms of point or purpose, independently of formal properties, and that relevant (or operative: see Bach and Harnish 1979) formal properties are only specifiable as a function of activity type. The sets of formal properties a text has out of context can be specified at some levels (e.g. syntactic), but these simply remind us of features of la langue, and don't say anything about the particular text as an instance of la parole.

8 The vocative reading was suggested to me by Richard Coates.

9 Time of utterance (just before Christmas) also serves to produce the reading intended, but I shan't discuss this.

10 Sloman provides a general definition, 'an analogical representation has a structure which gives information about the structure of the thing denoted, depicted or represented' (Sloman 1978, p.165). Fregean representations do not give such structural information. Sloman's use of 'analogic' corresponds to Peirce's use of 'iconic'. For Peirce, photographs are iconic and indexical (because causally linked to that which they represent). See Peirce 1940, pp.98-119.

11 Because the photographic signifier is continuous, talk of 'elements' must have to do with the signifieds - in this case, the objects denoted by the image. So the structural reorganization spoken of here has to do with the semantics of the image.

12 In Pateman 1980c, I use Searle 1969 and call the thematic or topical content the 'referential content' of the advertisement (which I treat as a single speech-act). Since advertisements are clearly discourse acts, rather than speech acts in Searle's sense, I abandon the use of the concept of referential content in this essay, using the less articulated ideas of 'theme' and 'topic'. In a further development of the lines of thought represented by the two studies, it might be possible to reintegrate the idea of referential content.

13 Roughly speaking, different stress-assignments produce utterances which can be heard as answers to different implicit questions, as in the following list (where italics represent heavy stress): John stole three horses (answers the question, Who stole three horses?) John stole three horses (What did John do with three horses?) John stole three horses (How many horses did John steal?) John stole three horses (John stole three what?) (Compare Smith and Wilson 1979, ch.7). In Pateman 1980c I call the comment 'predication', a term I have abandoned here for reasons explained in fn. 12.

14 On the semantic/pragmatic distinction, see Grice 1975; Kempson 1975; Wilson 1975; Smith and Wilson 1979, chs. 7 and 8; Bach and Harnish 1979: Gazdar 1979; Ortony 1979 - a small selection of excellent work from a large and fast-growing literature.

15 The four maxims are (1) Make your contribution as informative but not more so than is required (for the current purposes of the exchange); (2) Try to make your contribution one which is true; (3) Be relevant; (4) Be perspicuous (Grice 1975, pp.45-6). These maxims are seen by Grice to derive from a general Cooperative Principle of Conversation which he states informally as, 'Make your conversational contribution such as is required, at the stage at which is occurs, by the accepted purpose or direction of the talk exchange in which you are engaged' (p.45).

16 Smith and Wilson offer the following informal definition of relevance (to an ongoing discourse): A remark P is relevant to another remark Q if P and Q together with background knowledge yield new information not derivable from either P or Q, together with background knowledge, alone (Smith and Wilson 1979, p.177, see now Sperber and Wilson 1986 and my website review of this book).

17 One might say: there is no way in which the strength of jetties is semantically conventionalized the way the romance of roses is conventionalized. Conventionalization shortens the inferential circuit, but doesn't eliminate it (cf. Bach and Harnish 1979, chs. 6 and 7).

18 These intentions we impute to the advertiser are what Bach and Harnish 1979, following Grice 1957 and subsequent work, call R-intentions: intentions to achieve an effect by means of recognition of the intention to produce that effect .Habermas makes such intentions criterial for communicative action: see Habermas 1979a and 1979b, pp. 116-123).

19 The founders of the London Women's Liberation Workshop launched it in 1969 by enhancing London Underground advertisements with campaign stickers one of which read, 'What exactly are you selling?' (used particularly on advertisements for Elliott boots). I remember this because I was one of those employed to put on the stickers - for some reason, we did this very early in the morning, before the commuters arrived, and when our furtive behaviour must have been quite conspicuous.

20 Lacan.

21 In both art and science, the demand for answers to 'What exactly do you mean?' questions can be thoroughly destructive of original and creative work. See, for example, Boyd 1979 for a discussion.

22 Compare Barthes 1973.

23 I owe this point to Patricia Holland.

The original version of this essay, under the same title, appeared in Howard Davis and Paul Walton, editors, Language, Image, Media (Oxford: Basil Blackwell 1983) pp 187 - 204. It was revised around 1990 for a book which never appeared, and has been lightly revised from that second version for this new website version. The original pre-dated the appearance of Sperber and Wilson's major book on Relevance (1986) my discussion of which appears elsewhere on this site under the title "Relevance, Contextual Effects and Least Effort".

Valid XHTML 1.0!