Structuralism and Semiotics

Literary Theory An Introduction - Terry Eagleton 2008

Structuralism and Semiotics

We left American literary theory at the end of the Introduction in the grip of New Criticism, honing its increasingly sophisticated techniques and fighting a rearguard action against modern science and industrialism. But as North American society developed over the 1950s, growing more rigidly scientistic and managerial in its modes of thought, a more ambitious form of critical technocracy seemed demanded. New Criticism had done its job well, but it was in a sense too modest and particularist to qualify as a hard-nosed academic discipline. In its obsessive concentration on the isolated literary text, its delicate nurturings of sensibility, it had tended to leave aside the broader, more structural aspects of literature. What had happened to literary history? What was needed was a literary theory which, while preserving the formalist bent of New Criticism, its dogged attention to literature as aesthetic object rather than social practice, would make something a good deal more systematic and ’scientific’ out of all this. The answer arrived in 1957, in the shape of the Canadian Northrop Frye’s mighty ’totalization’ of all literary genres, Anatomy of Criticism.

Frye’s belief was that criticism was in a sorry unscientific mess and needed to be smartly tidied up. It was a matter of subjective valuejudgements and idle gossip, and badly required the discipline of an objective system. This was possible, Frye held, because literature itself formed such a system. It was not in fact just a random collection of writings strewn throughout history: if you examined it closely you could see that it worked by certain objective laws, and criticism could itself become systematic by formulating them. These laws were the various modes, archetypes, myths and genres by which all literary works were structured. At the root of all literature lay four ’narrative categories’, the comic, romantic, tragic and ironic, which could be seen to correspond respectively to the four mythoi of spring, summer, autumn and winter. A theory of literary ’modes’ could be outlined, whereby in myth the hero is superior in kind to others, in romance superior in degree, in the ’high mimetic’ modes of tragedy and epic superior in degree to others but not to his environment, in the ’low mimetic’ modes of comedy and realism equal to the rest of us, and in satire and irony inferior. Tragedy and comedy can be subdivided into high mimetic, low mimetic and ironic; tragedy is about human isolation, comedy about human integration. Three recurrent patterns of symbolism - the apocalyptic, demonic and analogical - are identified. The whole system can then be put into motion as a cyclical theory of literary history: literature passes from myth to irony and then reverts to myth, and in 1957 we were evidently somewhere in the ironic phase with signs of an impending return to the mythic.

To establish his literary system, of which the above is only a partial account, Frye must first of all clear value-judgements out of the way, since these are merely subjective noises. When we analyse literature we are speaking of literature; when we evaluate it we are speaking of ourselves. The system must also expel any history other than literary history: literary works are made out of other literary works, not out of any material external to the literary system itself. The advantage of Frye’s theory, then, is that it keeps literature untainted by history in New Critical fashion, viewing it as an enclosed ecological recycling of texts, but unlike New Criticism finds in literature a substitute history, with all the global span and collective structures of history itself. The modes and myths of literature are transhistorical, collapsing history to sameness or a set of repetitive variations on the same themes. For the system to survive it must be kept rigorously closed: nothing external can be allowed to infiltrate it lest its categories are deranged. This is why Frye’s ’scientific’ impulse demands a formalism even more full-blooded than that of New Criticism. The New Critics allowed that literature was in some significant sense cognitive, yielding a sort of knowledge of the world; Frye insists that literature is an ’autonomous verbal structure’ quite cut off from any reference beyond itself, a sealed and inward-looking realm which ’contain[s] life and reality in a system of verbal relationships’.1 All the system ever does is reshuffle its symbolic units in relation to each other, rather than in relation to any kind of reality outside it. Literature is not a way of knowing reality but a kind of collective utopian dreaming which has gone on throughout history, an expression of those fundamental human desires which have given rise to civilization itself, but which are never fully satisfied there. It is not to be seen as the self-expression of individual authors, who are no more than functions of this universal system: it springs from the collective subject of the human race itself, which is how it comes to embody ’archetypes’ or figures of universal significance.

Frye’s work emphasizes as it does the utopian root of literature because it is marked by a deep fear of the actual social world, a distaste for history itself. In literature, and in literature alone, one can shake off the sordid ’externalities’ of referential language and discover a spiritual home. The mythoi of the theory are, significantly, pre-urban images of the natural cycles, nostalgic memories of a history before industrialism. Actual history is for Frye bondage and determinism, and literature remains the one place where we can be free. It is worth asking what kind of history we have been living through for this theory to be even remotely convincing. The beauty of the approach is that it deftly combines an extreme aestheticism with an efficiently classifying ’scientificity’, and so maintains literature as an imaginary alternative to modern society while rendering criticism respectable in that society’s terms. It displays an iconoclastic briskness towards literary waffle, dropping each work into its appointed mythological slot with computerized efficiency, but blends this with the most Romantic of yearnings. In one sense it is scornfully ’anti-humanist’, decentring the individual human subject and centring all on the collective literary system itself; in another sense it is the work of a committed Christian humanist (Frye is a clergyman), for whom the dynamic which drives literature and civilization - desire - will finally be fulfilled only in the kingdom of God.

Like several of the literary theorists we have looked at, then, Frye offers literature as a displaced version of religion. Literature becomes an essential palliative for the failure of religious ideology, and supplies us with various myths which are of relevance to social life. In The Critical Path (1971), Frye contrasts conservative ’myths of concern’ with liberal ’myths of freedom’, and desires an equable balance between the two: the authoritarian tendencies of conservatism must be corrected by myths of freedom, while a conservative sense of order must temper liberalism’s tendencies to social irresponsibility. What the mighty mythological system from Homer to the kingdom of God comes down to, in short, is a position somewhere between liberal Republican and conservative Democrat. The only mistake, Frye informs us, is that of the revolutionary, who naively misinterprets myths of freedom as historically realizable goals. The revolutionary is just a bad critic, mistaking myth for reality as a child might mistake the actress for a real fairy princess. It is remarkable that literature, severed from any sordid practical concern as it is, is in the end more or less capable of telling us which way to vote. Frye stands in the liberal humanist tradition of Arnold, desiring, as he says, ’society as free, classless and urbane’. What he means by ’classless’, like Arnold before him, is in effect a society which universally subscribes to his own middle-class liberal values.

There is a loose sense in which Northrop Frye’s work can be described as ’structuralist’, and it is significantly contemporary with the growth of ’classical’ structuralism in Europe. Structuralism, as the term suggests, is concerned with structures, and more particularly with examining the general laws by which they work. It also like Frye tends to reduce individual phenomena to mere instances of such laws. But structuralism proper contains a distinctive doctrine which is not to be found in Frye: the belief that the individual units of any system have meaning only by virtue of their relations to one another. This does not follow from a simple belief that you should look at things ’structurally’. You can examine a poem as a ’structure’ while still treating each of its items as more or less meaningful in itself. Perhaps the poem contains one image about the sun and another about the moon, and you are interested in how these two images fit together to form a structure. But you become a card-carrying structuralist only when you claim that the meaning of each image is wholly a matter of its relation to the other. The images do not have a ’substantial’ meaning, only a ’relational’ one. You do not need to go outside the poem, to what you know of suns and moons, to explain them; they explain and define each other.

Let me try to illustrate this by a simple example. Suppose we are analysing a story in which a boy leaves home after quarrelling with his father, sets out on a walk through the forest in the heat of the day and falls down a deep pit. The father comes out in search of his son, peers down the pit, but is unable to see him because of the darkness. At that moment the sun has risen to a point directly overhead, illuminates the pit’s depths with its rays and allows the father to rescue his child. After a joyous reconciliation, they return home together.

This may not be a particularly gripping narrative, but it has the advantage of simplicity. Clearly it could be interpreted in all sorts of ways. A psychoanalytical critic might detect definite hints of the Oedipus complex in it, and show how the child’s fall into the pit is a punishment he unconsciously wishes upon himself for the rift with his father, perhaps a form of symbolic castration or a symbolic recourse to his mother’s womb. A humanist critic might read it as a poignant dramatization of the difficulties implicit in human relationships. Another kind of critic might see it as an extended, rather pointless word-play on ’son/sun’. What a structuralist critic would do would be to schematize the story in diagrammatic form. The first unit of signification, ’boy quarrels with father’, might be rewritten as ’low rebels against high’. The boy’s walk through the forest is a movement along a horizontal axis, in contrast to the vertical axis ’low/high’, and could be indexed as ’middle’. The fall into the pit, a place below ground, signifies ’low’ again, and the zenith of the sun ’high’. By shining into the pit, the sun has in a sense stooped ’low’, thus inverting the narrative’s first signifying unit, where ’low’ struck against ’high’. The reconciliation between father and son restores an equilibrium between ’low’ and ’high’, and the walk back home together, signifying ’middle’, marks this achievement of a suitably intermediate state. Flushed with triumph, the structuralist rearranges his rulers and reaches for the next story.

What is notable about this kind of analysis is that, like Formalism, it brackets off the actual content of the story and concentrates entirely on the form. You could replace father and son, pit and sun, with entirely different elements - mother and daughter, bird and mole - and still have the same story. As long as the structure of relations between the units is preserved, it does not matter which items you select. This is not the case with the psychoanalytical or humanist readings of the tale, which depend on these items having a certain intrinsic significance, to understand which we have to resort to our knowledge of the world outside the text. Of course there is a sense in which the sun is high and pits are low anyway, and to that extent what is chosen as ’content’ does matter; but if we took a narrative structure in which what was required was the symbolic role of ’mediator’ between two items, the mediator could be anything from a grasshopper to a waterfall.

The relations between the various items of the story may be ones of parallelism, opposition, inversion, equivalence and so on; and as long as this structure of internal relations remains intact, the individual units are replaceable. Three other points may be noted about the method. First, it does not matter to structuralism that this story is hardly an example of great literature. The method is quite indifferent to the cultural value of its object: anything from War and Peace to the War Cry will do. The method is analytical, not evaluative. Second, structuralism is a calculated affront to common sense. It refuses the ’obvious’ meaning of the story and seeks instead to isolate certain ’deep’ structures within it, which are not apparent on the surface. It does not take the text at face value, but ’displaces’ it into a quite different kind of object. Third, if the particular contents of the text are replaceable, there is a sense in which one can say that the ’content’ of the narrative is its structure. This is equivalent to claiming that the narrative is in a way about itself: its ’subject’ is its own internal relations, its own modes of sense-making.

Literary structuralism flourished in the 1960s as an attempt to apply to literature the methods and insights of the founder of modern structural linguistics, Ferdinand de Saussure. Since many popularizing accounts of Saussure’s epoch-making Course in General Linguistics (1916) are now available, I shall merely sketch in a few of his central positions. Saussure viewed language as a system of signs, which was to be studied ’synchronically’ - that is to say, studied as a complete system at a given point in time - rather than ’diachronically’, in its historical development. Each sign was to be seen as being made up of a ’signifier’ (a sound-image, or its graphic equivalent), and a ’signified’ (the concept or meaning). The three black marks c — a -1 are a signifier which evoke the signified ’cat’ in an English mind. The relation between signifier and signified is an arbitrary one: there is no inherent reason why these three marks should mean ’cat’, other than cultural and historical convention. Contrast chat in French. The relation between the whole sign and what it refers to (what Saussure calls the ’referent’, the real furry fourlegged creature) is therefore also arbitrary. Each sign in the system has meaning only by virtue of its difference from the others. ’Cat’ has meaning not ’in itself’, but because it is not ’cap’ or ’cad’ or ’bat’. It does not matter how the signifier alters, as long as it preserves its difference from all the other signifiers; you can pronounce it in many different accents as long as this difference is maintained. ’In the linguistic system,’ says Saussure, ’there are only differences’: meaning is not mysteriously immanent in a sign but is functional, the result of its difference from other signs. Finally, Saussure believed that linguistics would get into a hopeless mess if it concerned itself with actual speech, or parole as he called it. He was not interested in investigating what people actually said; he was concerned with the objective structure of signs which made their speech possible in the first place, and this he called langue. Neither was Saussure concerned with the real objects which people spoke about: in order to study language effectively, the referents of the signs, the things they actually denoted, had to be placed in brackets.

Structuralism in general is an attempt to apply this linguistic theory to objects and activities other than language itself. You can view a myth, wrestling match, system of tribal kinship, restaurant menu or oil painting as a system of signs, and a structuralist analysis will try to isolate the underlying set of laws by which these signs are combined into meanings. It will largely ignore what the signs actually ’say’, and concentrate instead on their internal relations to one another. Structuralism, as Fredric Jameson has put it, is an attempt ’to rethink everything through once again in terms of linguistics’.2 It is a symptom of the fact that language, with its problems, mysteries and implications, has become both paradigm and obsession for twentiethcentury intellectual life.

Saussure’s linguistic views influenced the Russian Formalists, although Formalism is not itself exactly a structuralism. It views literary texts ’structurally’, and suspends attention to the referent to examine the sign itself, but it is not particularly concerned with meaning as differential or, in much of its work, with the ’deep’ laws and structures underlying literary texts. It was one of the Russian Formalists, however - the linguist Roman Jakobson - who was to provide the major link between Formalism and modern-day structuralism. Jakobson was leader of the Moscow Linguistic Circle, a Formalist group founded in 1915, and in 1920 migrated to Prague to become one of the major theoreticians of Czech structuralism. The Prague Linguistic Circle was founded in 1926, and survived until the outbreak of the Second World War. Jakobson later migrated once more, this time to the United States, where he encountered the French anthropologist Claude Levi-Strauss during the Second World War, an intellectual relationship out of which much of modern structuralism was to develop.

Jakobson’s influence can be detected everywhere within Formalism, Czech structuralism and modern linguistics. What he contributed in particular to poetics, which he regarded as part of the field of linguistics, was the idea that the ’poetic’ consisted above all in language’s being placed in a certain kind of self-conscious relationship to itself. The poetic functioning of language ’promotes the palpability of signs’, draws attention to their material qualities rather than simply using them as counters in communication. In the ’poetic’, the sign is dislocated from its object: the usual relation between sign and referent is disturbed, which allows the sign a certain independence as an object of value in itself. All communication for Jakobson involves six elements: an addresser, an addressee, a message passed between them, a shared code which makes that message intelligible, a ’contact’ or physical medium of communication, and a ’context’ to which the message refers. Any one of these elements may dominate in a particular communicative act: language seen from the addresser’s viewpoint is ’emotive’ or expressive of a state of mind; from the addressee’s standpoint it is ’conative’, or trying for an effect; if communication concerns the context it is ’referential’, if it is oriented to the code itself it is ’metalinguistic’ (as when two individuals discuss whether they are understanding each other), and communication angled towards the contact itself is ’phatic’ (e.g. ’Well, here we are chatting away at last’). The ’poetic’ function is dominant when the communication focuses on the message itself - when the words themselves, rather than what is said by whom for what purpose in what situation, are ’foregrounded’ in our attention.3

Jakobson also makes much of a distinction implicit in Saussure between the metaphorical and the metonymic. In metaphor, one sign is substituted for another because it is somehow similar to it: ’passion’ becomes ’flame’. In metonymy, one sign is associated with another: ’wing’ is associated with ’aircraft’ because it is part of it, ’sky’ with ’aircraft’ because of physical contiguity. We can make metaphors because we have a series of signs which are ’equivalent’: ’passion’, ’flame’, ’love’ and so on. When we speak or write, we select signs from a possible range of equivalences, and then combine them together to form a sentence. What happens in poetry, however, is that we pay attention to ’equivalences’ in the process of combining words together as well as in selecting them: we string together words which are semantically or rhythmically or phonetically or in some other way equivalent. This is why Jakobson is able to say, in a famous definition, that ’The poetic function projects the principle of equivalence from the axis of selection to the axis of combination.’4 Another way of saying this is that, in poetry, ’similarity is superinduced upon contiguity’: words are not just strung together for the sake of the thoughts they convey, as in ordinary speech, but with an eye to the patterns of similarity, opposition, parallelism and so on created by their sound, meaning, rhythm and connotations. Some literary forms - realist prose, for example - tend to be metonymic, linking signs by their associations with each other; other forms, such as Romantic and Symbolist poetry, are highly metaphorical.5

The Prague school of linguistics - Jakobson, Jan Mukarovsky, Felix Vodicka and others - represent a kind of transition from Formalism to modern structuralism. They elaborated the ideas of the Formalists, but systematized them more firmly within the framework of Saussurean linguistics. Poems were to be viewed as ’functional structures’, in which signifiers and signifieds are governed by a single complex set of relations. These signs must be studied in their own right, not as reflections of an external reality: Saussure’s stress on the arbitrary relation between sign and referent, word and thing, helped to detach the text from its surroundings and make of it an autonomous object. Yet the literary work was still related to the world by the Formalist concept of’defamiliarization’: art estranges and undermines conventional sign-systems, compels our attention to the material process of language itself, and so renews our perceptions. In not taking language for granted, we are also transforming our consciousness. More than the Formalists, however, the Czech structuralists insisted on the structural unity of the work: its elements were to be grasped as functions of a dynamic whole, with one particular level of the text (what the Prague school called the ’dominant’) acting as the determining influence which ’deformed’, or pulled into its own field of force, all the others.

So far the Prague structuralists might sound like little more than a more scientific version of New Criticism, and there is a seed of truth in this suggestion. But though the artefact was to be seen as a closed system, what counted as an artefact was a matter of social and historical circumstances. According to Jan Mukarovsky, the work of art is perceived as such only against a more general background of significations, only as a systematic ’deviation’ from a linguistic norm; as this background changes, interpretation and evaluation of the work change accordingly, to the point where it may cease to be perceived as a work of art at all. There is nothing, Mukarovsky argues in Aesthetic Function, Norm and Value as Social Facts (1936) which possesses an aesthetic function regardless of place, time or the person evaluating it, and nothing which could not possess such a function in appropriate conditions. Mukarovsky distinguishes between the ’material artefact’, which is the physical book, painting or sculpture itself, and the ’aesthetic object’, which exists only in human interpretation of this physical fact.

With the work of the Prague school, the term ’structuralism’ comes more or less to merge with the word ’semiotics’. ’Semiotics’, or ’semiology’, means the systematic study of signs, and this is what literary structuralists are really doing. The word ’structuralism’ itself indicates a method of enquiry, which can be applied to a whole range of objects from football matches to economic modes of production; ’semiotics’ denotes rather a particular field of study, that of systems which would in an ordinary sense be regarded as signs: poems, bird calls, traffic lights, medical symptoms and so on. But the two words overlap, since structuralism treats something which may not usually be thought of as a system of signs as though it were - the kinship relations of tribal societies, for example - while semiotics commonly uses structuralist methods.

The American founder of semiotics, the philosopher C. S. Peirce, distinguished between three basic kinds of sign. There was the ’iconic’, where the sign somehow resembled what it stood for (a photograph of a person, for example); the ’indexical’, in which the sign is somehow associated with what it is a sign of (smoke with fire, spots with measles), and the ’symbolic’, where as with Saussure the sign is only arbitrarily or conventionally linked with its referent. Semiotics takes up this and many other classifications: it distinguishes between ’denotation’ (what the sign stands for) and ’connotation’ (other signs associated with it); between codes (the rule-governed structures which produce meanings) and the messages transmitted by them; between the ’paradigmatic’ (a whole class of signs which may stand in for one another) and the ’syntagmatic’ (where signs are coupled together with each other in a ’chain’). It speaks of ’metalanguages’, where one sign-system denotes another sign-system (the relation between literary criticism and literature, for instance), ’polysemic’ signs which have more than one meaning, and a great many other technical concepts. To see what this kind of analysis looks like in practice, we may briefly consider the work of the leading Soviet semiotician of the so-called school of Tartu, Yury Lotman.

In his works The Structure of the Artistic Text (1970) and The Analysis of the Poetic Text (1972), Lotman sees the poetic text as a stratified system in which meaning only exists contextually, governed by sets of similarities and oppositions. Differences and parallelisms in the text are themselves relative terms, and can only be perceived in relation to one another. In poetry, it is the nature of the signifier, the patterns of sound and rhythm set up by the marks on the page themselves, which determines what is signified. A poetic text is ’semantically saturated’, condensing more ’information’ than any other discourse; but whereas for modern communication theory in general an increase in ’information’ leads to a decrease in ’communication’ (since I cannot ’take in’ all that you so intensively tell me), this is not so in poetry because of its unique kind of internal organization. Poetry has a minimum of ’redundancy’ - of those signs which are present in a discourse to facilitate communication rather than convey information - but still manages to produce a richer set of messages than any other form of language. Poems are bad when they do not carry sufficient information, for, as Lotman remarks, ’information is beauty’. Every literary text is made up of a number of ’systems’ (lexical, graphic, metrical, phonological and so on), and gains its effects through constant clashes and tensions between these systems. Each of the systems comes to represent a ’norm’ from which the others deviate, setting up a code of expectations which they transgress. Metre, for example, creates a certain pattern which the poem’s syntax may cut across and violate. In this way, each system in the text ’defamiliarizes’ the others, breaking up their regularity and throwing them into more vivid relief. Our perception of the poem’s grammatical structure, for example, may heighten our awareness of its meanings. Just as one of the poem’s systems threatens to become too predictable, another cuts across it to disrupt it into new life. If two words are associated together because of their similar sound or position in the metrical scheme, this will produce a sharper awareness of their similarity or difference of meaning. The literary work continually enriches and transforms mere dictionary meaning, generating new significances by the clash and condensation of its various ’levels’. And since any two words whatsoever may be juxtaposed on the basis of some equivalent feature, this possibility is more or less unlimited. Each word in the text is linked by a whole set of formal structures to several other words, and its meaning is thus always ’overdetermined’, always the result of several different determinants acting together. An individual word may relate to another word through assonance, to another through syntactical equivalence, to yet another through morphological parallelism, and so on. Each sign thus participates in several different ’paradigmatic patterns’ or systems simultaneously, and this complexity is greatly compounded by the ’syntagmatic’ chains of association, the ’lateral’ rather than ’vertical’ structures, in which signs are placed.

The poetic text for Lotman is thus a ’system of systems’, a relation of relations. It is the most complex form of discourse imaginable, condensing together several systems each of which contains its own tensions, parallelisms, repetitions and oppositions, and each of which is continually modifying all of the others. A poem, in fact, can only be re-read, not read, since some of its structures can only be perceived retrospectively. Poetry activates the full body of the signifier, presses the word to work to its utmost under the intense pressure of surrounding words, and so to release its richest potential. Whatever we perceive in the text is perceived only by contrast and difference: an element which had no differential relation to any other would remain invisible. Even the absence of certain devices may produce meaning: if the codes which the work has generated lead us to expect a rhyme or a happy ending which does not materialize, this ’minus device’, as Lotman terms it, may be as effective a unit of meaning as any other. The literary work, indeed, is a continual generating and violating of expectations, a complex interplay of the regular and the random, norms and deviations, routinized patterns and dramatic defamiliarizations.

Despite this unique verbal richness, Lotman does not consider that poetry or literature can be defined by their inherent linguistic properties. The meaning of the text is not just an internal matter: it also inheres in the text’s relation to wider systems of meaning, to other texts, codes and norms in literature and society as a whole. Its meaning is also relative to the reader’s ’horizon of expectations’: Lotman has learned the lessons of reception theory well. It is the reader who by virtue of certain ’receptive codes’ at his or her disposal identifies an element in the work as a ’device’; the device is not simply an internal feature but one perceived through a particular code and against a definite textual background. One person’s poetic device may be another’s daily speech.

It is obvious from all this that literary criticism has come a long way from the days when we had to do little more than thrill to the beauty of the imagery. What semiotics represents, in fact, is literary criticism transfigured by structural linguistics, rendered a more disciplined and less impressionistic enterprise which, as Lotman’s work testifies, is more rather than less alive to the wealth of form and language than most traditional criticism. But if structuralism transformed the study of poetry, it also revolutionized the study of narrative. Indeed it created a whole new literary science - narratology - of which the most influential practitioners have been the Lithuanian A. J. Greimas, the Bulgarian Tzvetan Todorov, and the French critics Gerard Genette, Claude Bremond and Roland Barthes. The modern structuralist analysis of narrative began with the pioneering work on myth of the French structural anthropologist Claude Levi-Strauss, who viewed apparently different myths as variations on a number of basic themes. Beneath the immense heterogeneity of myths were certain constant universal structures, to which any particular myth could be reduced. Myths were a kind of language: they could be broken down into individual units (’mythemes’) which like the basic sound units of language (phonemes) acquired meaning only when combined together in particular ways. The rules which governed such combinations could then be seen as a kind of grammar, a set of relations beneath the surface of the narrative which constituted the myth’s true ’meaning’. These relations, Levi-Strauss considered, were inherent in the human mind itself, so that in studying a body of myth we are looking less at its narrative contents than at the universal mental operations which structure it. These mental operations, such as the making of binary oppositions, are in a way what myths are about: they are devices to think with, ways of classifying and organizing reality, and this, rather than the recounting of any particular tale, is their point. The same, Levi-Strauss believes, can be said of totemic and kinship systems, which are less social and religious institutions than networks of communication, codes which permit the transmission of ’messages’. The mind which does all this thinking is not that of the individual subject: myths think themselves through people, rather than vice versa. They have no origin in a particular consciousness, and no particular end in view. One result of structuralism, then, is the ’decentring’ of the individual subject, who is no longer to be regarded as the source or end of meaning. Myths have a quasi-objective collective existence, unfold their own ’concrete logic’ with supreme disregard for the vagaries of individual thought, and reduce any particular consciousness to a mere function of themselves.

Narratology consists in generalizing this model beyond the unwritten ’texts’ of tribal mythology to other kinds of story. The Russian Formalist Vladimir Propp had already made a promising start with his Morphology of the Folk Tale (1928), which boldly reduced all folk tales to seven ’spheres of action’ and thirty-one fixed elements or ’functions’. Any individual folk tale merely combined these ’spheres of action’ (the hero, the helper, the villain, the person sought-for and so on) in specific ways. Drastically economical as this model was, it was possible to reduce it even further. A. J. Greimas’s Semantique structural (1966), finding Propp’s scheme still too empirical, is able to abstract his account even further by the concept of an actant, which is neither a specific narrative even nor a character but a structural unit. The six actants of Subject and Object, Sender and Receiver, Helper and Opponent can subsume Propp’s various spheres of action and make for an even more elegant simplicity. Tzvetan Todorov attempts a similar ’grammatical’ analysis of Boccaccio’s Decameron, in which characters are seen as nouns, their attributes as adjectives and their actions as verbs. Each story of The Decameron can thus be read as a kind of extended sentence, combining these units in different ways. And just as the work thus comes to be about its own quasi-linguistic structure, so for structuralism every literary work, in the act of apparently describing some external reality, is secretly casting a sideways glance at its own processes of construction. In the end, structuralism does not only think everything through again, this time as language; it thinks everything through again as though language were its very subject matter.

To clarify our view of narratology, we may look finally at the work of Gerard Genette. In his Narrative Discourse (1972), Genette draws on a distinction in narrative between recit, by which he means the actual order of events in the text; histoire, which is the sequence in which those events ’actually’ occurred, as we can infer this from the text; and narration, which concerns the act of narrating itself. The first two categories are equivalent to a classic Russian Formalist distinction between ’plot’ and ’story’: a detective story usually opens with the discovery of a body and finally backtracks to expose how the murder happened, but this plot of events reverses the ’story’ or actual chronology of the action. Genette discerns five central categories of narrative analysis. ’Order’ refers to the time-order of the narrative, how it may operate by prolepsis (anticipation), analepsis (flashback) or anachrony, which refers to discordances between ’story’ and ’plot’. ’Duration’ signifies how the narrative may elide episodes, expand them, summarize, pause a little and so on. ’Frequency’ involves questions of whether an event happened once in the ’story’ and is narrated once, happened once but is narrated several times, happened several times and is narrated several times, or happened several times and is narrated only once. The category of ’mood’ can be subdivided into ’distance’ and ’perspective’. Distance concerns the relation of the narration to its own materials: is it a matter of recounting the story (’diagesis’) or representing it (’mimesis’), is the narrative told in direct, indirect or ’free indirect’ speech? ’Perspective’ is what might traditionally be called ’point of view’, and can also be variously subdivided: the narrator may know more than the characters, less than them, or move on the same level; the narrative may be ’non-focalized’, delivered by an omniscient narrator outside the action, or ’internally focalized’, recounted by one character from a fixed position, from variable positions, or from several character-viewpoints. A form of ’external focalization’ is possible, in which the narrator knows less than the characters do. Finally there is the category of ’voice’, which concerns the act of narrating itself, what kind of narrator and narratee are implied. Various combinations are possible here between the ’time of the narrative’ and the ’narrated time’, between the action of recounting the story and the events which you recount: you may tell of the events before, after or (as in the epistolary novel) while they happen. A narrator may be ’heterodiegetic’ (i.e. absent from his own narrative), ’homodiegetic’ (inside his narrative as in first-person stories), or ’autodiegetic’ (where he is not only inside the narrative but figures as its principal character). These are only some of Genette’s classifications; but one important aspect of discourse to which they alert us is the difference between narration — the act and process of telling a story - and narrative - what it is you actually recount. When I tell a story about myself, as in autobiography, the ’I’ who does the telling seems in one sense identical with the ’I’ whom I describe, and in another sense different from it. We shall see later how this paradox has interesting implications beyond literature itself.

What are the gains of structuralism? To begin with, it represents a remorseless demystification of literature. It is less easy after Greimas and Genette to hear the cut and thrust of the rapiers in line three, or feel that you know just what it feels like to be a scarecrow after reading The Hollow Men. Loosely subjective talk was chastized by a criticism which recognized that the literary work, like any other product of language, is a construct, whose mechanisms could be classified and analysed like the objects of any other science. The Romantic prejudice that the poem, like a person, harboured a vital essence, a soul which it was discourteous to tamper with, was rudely unmasked as a bit of disguised theology, a superstitious fear of reasoned enquiry which made a fetish of literature and reinforced the authority of a ’naturally’ sensitive critical elite. Moreover, the structuralist method implicitly questioned literature’s claim to be a unique form of discourse: since deep structures could be dug out of Mickey Spillane as well as Sir Philip Sidney, and no doubt the same ones at that, it was no longer easy to assign literature an ontologically privileged status. With the advent of structuralism, the world of the great aestheticians and humanist literary scholars of twentiethcentury Europe - the world of Croce, Curtius, Auerbach, Spitzer and Wellek - seemed one whose hour had passed.6 These men, with their formidable erudition, imaginative insight and cosmopolitan range of allusion, appeared suddenly in historical perspective, as luminaries of a high European humanism which pre-dated the turmoil and conflagration of the mid-twentieth century. It seemed clear that such a rich culture could not be reinvented - that the choice was between learning from it and passing on, or clinging with nostalgia to its remnants in our time, denouncing a ’modem world’ in which the paperback has spelt the death of high culture, and where there are no longer domestic servants to protect one’s door while one reads in privacy.

The structuralist emphasis on the ’constructedness’ of human meaning represented a major advance. Meaning was neither a private experience nor a divinely ordained occurrence: it was the product of certain shared systems of signification. The confident bourgeois belief that the isolated individual subject was the fount and origin of all meaning took a sharp knock: language pre-dated the individual, and was much less his or her product than he or she was the product of it. Meaning was not ’natural’, a question of just looking and seeing, or something eternally settled; the way you interpreted your world was a function of the languages you had at your disposal, and there was evidently nothing immutable about these. Meaning was not something which all men and women everywhere intuitively shared, and then articulated in their various tongues and scripts: what meaning you were able to articulate depended on what script or speech you shared in the first place. There were the seeds here of a social and historical theory of meaning, whose implications were to run deep within contemporary thought. It was impossible any longer to see reality simply as something ’out there’, a fixed order of things which language merely reflected. On that assumption, there was a natural bond between word and thing, a given set of correspondences between the two realms. Our language laid bare for us how the world was, and this could not be questioned. This rationalist or empiricist view of language suffered severely at the hands of structuralism: for if, as Saussure had argued, the relation between sign and referent was an arbitrary one, how could any ’correspondence’ theory of knowledge stand? Reality was not reflected by language but produced by it: it was a particular way of carving up the world which was deeply dependent on the sign-systems we had at our command, or more precisely which had us at theirs. The suspicion began to arise, then, that structuralism was only not an empiricism because it was yet one more form of philosophical idealism - that its view of reality as essentially a product of language was simply the latest version of the classical idealist doctrine that the world was constituted by human consciousness.

Structuralism scandalized the literary Establishment with its neglect of the individual, its clinical approach to the mysteries of literature, and its clear incompatibility with common sense. The fact that structuralism offends common sense has always been a point in its favour. Common sense holds that things generally have only one meaning and that this meaning is usually obvious, inscribed on the faces of the objects we encounter. The world is pretty much as we perceive it, and our way of perceiving it is the natural, self-evident one. We know the sun goes round the earth because we can see that it does. At different times common sense has dictated burning witches, hanging sheep-stealers and avoiding Jews for fear of fatal infection, but this statement is not itself commonsensical since common sense believes itself to be historically invariable. Thinkers who have argued that the apparent meaning is not necessarily the real one have usually been met with scorn: Copernicus was followed by Marx, who claimed that the true significance of social processes went on ’behind the backs’ of individual agents, and after Marx Freud argued that the real meanings of our words and actions were quite imperceptible to the conscious mind. Structuralism is a modern inheritor of this belief that reality, and our experience of it, are discontinuous with each other; as such, it threatens the ideological security of those who wish the world to be within their control, to carry its singular meaning on its face and to yield it up to them in the unblemished mirror of their language. It undermines the empiricism of the literary humanists - the belief that what is most ’real’ is what is experienced, and that the home of this rich, subtle, complex experience is literature itself. Like Freud, it exposes the shocking truth that even our most intimate experience is the effect of a structure.

I have said that structuralism contained the seeds of a social and historical theory of meaning, but they were not, on the whole, able to sprout. For if the sign-systems by which individuals lived could be seen as culturally variable, the deep laws which governed the workings of these systems were not. For the ’hardest’ forms of structuralism they were universal, embedded in a collective mind which transcended any particular culture, and which Levi- Strauss suspected to be rooted in the structures of the human brain itself. Structuralism, in a word, was hair-raisingly unhistorical: the laws of the mind it claimed to isolate - parallelisms, oppositions, inversions and the rest - moved at a level of generality quite remote from the concrete differences of human history. From this Olympian height, all minds looked pretty much alike. Having characterized the underlying rule-systems of a literary text, all the structuralist could do was sit back and wonder what to do next. There was no question of relating the work to the realities of which it treated, or to the conditions which produced it, or to the actual readers who studied it, since the founding gesture of structuralism had been to bracket off such realities. In order to reveal the nature of language, Saussure, as we have seen, had first of all to repress or forget what it talked about: the referent, or real object which the sign denoted, was put in suspension so that the structure of the sign itself could be better examined. It is notable how similar this gesture is to Husserl’s bracketing of the real object in order to get to closer grips with the way the mind experiences it. Structuralism and phenomenology, dissimilar though they are in central ways, both spring from the ironic act of shutting out the material world in order the better to illuminate our consciousness of it. For anyone who believes that consciousness is in an important sense practical, inseparably bound up with the ways we act in and on reality, any such move is bound to be self-defeating. It is rather like killing a person in order to examine more conveniently the circulation of the blood.

But it was not just a matter of shutting out something as general as ’the world’: it was a question of discovering some toehold of certainty in a particular world where certainty seemed hard to come by. The lectures which make up Saussure’s Course in General Linguistics were delivered in the heart of Europe between 1907 and 1911, on the brink of an historical collapse which Saussure himself did not live to see. These were precisely the years in which Edmund Husserl was formulating the major doctrines of phenomenology, in a European centre not far removed from Saussure’s Geneva. At about the same time, or a little later, the major writers of twentieth-century English literature - Yeats, Eliot, Pound, Lawrence, Joyce - were developing their own closed symbolic systems, in which Tradition, theosophy, the male and female principles, medievalism and mythology were to provide the keystones of complete ’synchronic’ structures, exhaustive models for the control and explanation of historical reality. Saussure himself was to posit the existence of a ’collective consciousness’ underlying the system of langue. It is not difficult to see the flight from contemporary history in the recourse to myth of the major writers of English literature; it is less obviously detectable in a textbook of structural linguistics or an esoteric piece of philosophy.

Where it is more obviously detectable, perhaps, is in structuralism’s embarrassment with the problem of historical change. Saussure looked at the development of language in terms of one synchronic system following another, rather like the Vatican official who remarked that whether the Pope’s imminent pronouncement on the question of birth control turned out to uphold the previous teaching or not, the Church would nevertheless have moved from one state of certainty to another state of certainty. For Saussure, historical change was something which afflicted the individual elements of a language, and could only in this indirect way affect the whole: the language as a whole would reorganize itself to accommodate such disturbances, like learning to live with a wooden leg or like Eliot’s Tradition welcoming a new masterpiece to the club. Behind this linguistic model lies a definite view of human society: change is disturbance and disequilibrium in an essentially conflict-free system, which will stagger for a moment, regain its balance and take the change in its stride. Linguistic change for Saussure seems a matter of accident: it happens ’blindly’, and it was left to the later Formalists to explain how change itself might be grasped systematically. Jakobson and his colleague Yury Tynyanov saw the history of literature as itself forming a system, in which at any given point some forms and genres were ’dominant’ while others were subordinate. Literary development took place by way of shifts within this hierarchical system, such that a previously dominant form became subordinate or vice versa. The dynamic of this process was ’defamiliarization’: if a dominant literary form had grown stale and ’imperceptible’ - if, for example, some of its devices had been taken over by a sub-genre such as popular journalism, thus blurring its difference from such writings - a previously subordinate form would emerge to ’defamiliarize’ this situation. Historical change was a matter of the gradual realignment of fixed elements within the system: nothing ever disappeared, it merely changed shape by altering its relations to other elements. The history of a system, Jakobson and Tynyanov comment, is itself a system: diachrony can be studied synchronically. Society itself was made up of a whole set of systems (or ’series’, as the Formalists called them), each of which was powered by its own internal laws, and evolved in relative autonomy of all the others. There were, however, ’correlations’ between the various series: at any given time the literary series would encounter several possible paths along which it could develop, but which path was actually selected was the result of correlations between the literary system itself and other historical series. This was not a suggestion which all later structuralists took up: in their resolutely ’synchronic’ approach to the object of study, historical change sometimes became as mysteriously inexplicable as the Romantic symbol.

Structuralism broke with conventional literary criticism in many ways, while remaining mortgaged to it in many others. Its preoccupation with language was, as we have seen, radical in its implications, but it was at the same time a familiar obsession of academics. Was language really all there was? What about labour, sexuality, political power? These realities might themselves be inextricably caught up in discourse, but they were certainly not reducible to it. What political conditions themselves determined this extreme ’foregrounding’ of language itself? Was the structuralist view of the literary text as a closed system really much different from the New Critical treatment of it as an isolated object? What had happened to the concept of literature as a social practice, a form of production which was not necessarily exhausted by the product itself? Structuralism could dissect that product, but it refused to enquire into the material conditions of its making, since this might mean surrendering to the myth of an ’origin’. Nor were many structuralists worried about how the product was actually consumed - about what happened when people actually read works of literature, what role such works played in social relations as a whole. Moreover, was not structuralism’s stress on the integrated nature of a sign-system just another version of the work as ’organic unity’? Levi-Strauss spoke of myths as imaginary resolutions of real social contradictions; Yury Lotman used the imagery of cybernetics to show how the poem formed a complex organic totality; the Prague school developed a ’functionalist’ view of the work in which all the parts laboured inexorably together for the good of the whole. Traditional criticism had sometimes reduced the literary work to little more than a window on to the author’s psyche; structuralism seemed to make it a window on to the universal mind. The ’materiality’ of the text itself, its detailed linguistic processes, was in danger of being abolished: the ’surface’ of a piece of writing was little more than the obedient reflection of its concealed depths. What Lenin once called the ’reality of appearances’ was at risk of being overlooked: all ’surface’ features of the work could be reduced to an ’essence’, a single central meaning which informed all the work’s aspects, and this essence was no longer the writer’s soul or the Holy Spirit but the ’deep structure’ itself. The text was really just a ’copy’ of this deep structure, and structuralist criticism was a copy of this copy. Finally, if traditional critics composed a spiritual elite, structuralists appeared to constitute a scientific one, equipped with an esoteric knowledge far removed from the ’ordinary’ reader.

At the same moment as structuralism bracketed off the real object, it bracketed off the human subject. Indeed it is this double movement which defines the structuralist project. The work neither refers to an object, nor is the expression of an individual subject; both of these are blocked out, and what is left hanging in the air between them is a system of rules. This system has its own independent life, and will not stoop to the beck and call of individual intentions. To say that structuralism has a problem with the individual subject is to put it mildly: that subject was effectively liquidated, reduced to the function of an impersonal structure. To put it another way: the new subject was really the system itself, which seemed equipped with all the attributes (autonomy, self-correction, unity and so on) of the traditional individual. Structuralism is ’anti-humanist’, which means not that its devotees rob children of their sweets but that they reject the myth that meaning begins and ends in the individual’s ’experience’. For the humanist tradition, meaning is something that I create, or that we create together; but how could we create meaning unless the rules which govern it were already there? However far back we push, however much we hunt for the origin of meaning, we will always find a structure already in place. This structure could not have been simply the result of speech, for how were we able to speak coherently in the first place without it? We could never discover the ’first sign’ from which it all began, because, as Saussure makes clear, one sign presupposes another from which it differs, and that another. If language was ever ’born’, Levi-Strauss speculates, it must have been born ’at a stroke’. Roman Jakobson’s communicative model, the reader will remember, starts from an addresser who is the source of the transmitted message; but where did this addresser come from? To be able to transmit a message at all, he or she must already be caught up in and constituted by language. In the beginning was the Word.

To see language in this way is a valuable advance on seeing it simply as the ’expression’ of an individual mind. But it also makes for severe difficulties. For though language may not be best understood as individual expression, it certainly in some way involves human subjects and their intentions, and it is this which the structuralist picture leaves out of account. Let us go back for a moment to the situation I outlined earlier, where I tell you to close the door when a gale is howling through the room. I said then that the meaning of my words was independent of any private intention I might have - that the meaning was, so to speak, a function of the language itself, rather than some mental process of mine. In a certain practical situation, the words just do seem to mean what they mean whatever I might whimsically want them to mean. But what if I asked you to close the door having just spent twenty minutes roping you to your chair? What if the door was closed already, or there was no door there at all? Then, surely, you would be quite justified to ask me: ’What do you mean?’ It isn’t that you don’t understand the meaning of my words', it is that you don’t understand the meaning of my words. It will not help if I hand you a dictionary. Asking ’What do you mean?’ in this situation is indeed asking about the intentions of a human subject, and unless I understand these then the request to close the door is in an important sense meaningless.

Asking about my intentions, however, is not necessarily asking to peer into my mind and observe the mental processes going on there. It is not necessary to see intentions in the way that E. D. Hirsch does, as essentially private ’mental acts’. To ask in such a situation ’What do you mean?’ is really to ask what effects my language is trying to bring about: it is a way of understanding the situation itself, not an attempt to tune into ghostly impulses within my skull. Understanding my intention is grasping my speech and behaviour in relation to a significant context. When we understand the ’intentions’ of a piece of language, we interpret it as being in some sense oriented, structured to achieve certain effects; and none of this can be grasped apart from the practical conditions in which the language operates. It is to see language as a practice rather than as an object; and there are of course no practices without human subjects.

This way of viewing language is on the whole quite foreign to structuralism, at least in its classical varieties. Saussure, as I have mentioned, was interested not in what people actually said but in the structure which allowed them to say it: he studied langue rather than parole, seeing the former as an objective social fact and the latter as the random, untheorizable utterance of the individual. But this view of language already encodes a certain questionable way of conceptualizing the relations between individuals and societies. It sees the system as determined and the individual as free; it grasps social pressures and determinants not so much as forces active in our actual speaking, but as a monolithic structure which somehow stands over against us. It presumes that parole, individual utterance, really is individual, rather than an inevitably social and ’dialogic’ affair which catches us up with other speakers and listeners in a whole field of social values and purposes. Saussure strips language of its sociality at the point where it matters most: at the point of linguistic production, the actual speaking, writing, listening and reading of concrete social individuals. The constraints of the language system are consequently fixed and given, aspects of langue, rather than forces which we produce, modify and transform in our actual communication. We may also notice that Saussure’s model of individual and society, like many classical bourgeois models, has no intermediate terms, no mediations between solitary individual speakers and the linguistic system as a whole. The fact that someone may not only be a ’member of society’ but also a woman, shop-steward, Catholic, mother, immigrant and disarmament campaigner is simply slid over. The linguistic corollary of this - that we inhabit many different ’languages’ simultaneously, some of them perhaps mutually conflicting - is also ignored.

The shift away from structuralism has been in part, to use the terms of the French linguist Emile Benveniste, a move from ’language’ to ’discourse’.7 ’Language’ is speech or writing viewed ’objectively’, as a chain of signs without a subject. ’Discourse’ means language grasped as utterance, as involving speaking and writing subjects and therefore also, at least potentially, readers or listeners. This is not simply a return to the pre-structuralist days when we thought that language belonged to us individually as our eyebrows did; it does not revert to the classical ’contractual’ model of language, according to which language is just a sort of instrument essentially isolated individuals use to exchange their pre-linguistic experiences. This was really a ’market’ view of language, closely associated with the historical growth of bourgeois individualism: meaning belonged to me like my commodity, and language was just a set of tokens which like money allowed me to exchange my meaning-commodity with another individual who was also a private proprietor of meaning. It was difficult on this empiricist theory of language to know how what got exchanged was the genuine article: if I had a concept, fixed a verbal sign to it and threw the whole package across to someone else, who looked at the sign and rifled through his own verbal filing system for the corresponding concept, how could I ever know that he was matching up signs and concepts in the way that I was? Maybe we were all systematically misunderstanding each other all of the time. Laurence Sterne wrote a novel, Tristram Shandy, exploiting the comic potential of just this empiricist model, not long after it had become the standard philosophical view of language in England. There was no question for the critics of structuralism of returning to this sorry state in which we viewed signs in terms of concepts, rather than talking about having concepts as particular ways of handling signs. It was just that a theory of meaning which seemed to squeeze out the human subject was very curious. What had been narrow-minded about previous theories of meaning was their dogmatic insistence that the intention of the speaker or writer was always paramount for interpretation. In countering this dogmatism, there was no need to pretend that intentions did not exist at all; it was simply necessary to point out the arbitrariness of claiming that they were always the ruling structure of discourse.

In 1962, Roman Jakobson and Claude Levi-Strauss published an analysis of Charles Baudelaire’s poem Les chats which has become something of a classic of high structuralist practice.8 With toothcombing tenacity, the essay dug out a set of equivalences and oppositions from the poem’s semantic, syntactic and phonological levels, equivalences and oppositions which extended right down to individual phonemes. But as Michael Riffaterre pointed out in a famous rejoinder to this critique, some of the structures Jakobson and Levi-Strauss identified would simply have been imperceptible to even the most vigilant reader.9 Moreover, the analysis took no account of the reading process: it seized the text synchronically, as an object in space rather than a movement in time. A particular meaning in a poem will cause us retrospectively to revise what we have learnt already; a word or image which is repeated does not mean the same as it did the first time, by virtue of the very fact that it is a repetition. No event occurs twice, precisely because it has occurred once already. The Baudelaire essary, Riffaterre argues, also overlooks certain crucial connotations of words which one could recognize only by moving outside the text itself to the cultural and social codes on which it draws; and this move, of course, is forbidden by the authors’ structuralist assumptions. In true structuralist fashion, they treat the poem as ’language’; Riffaterre, by appealing to the reading process and the cultural situation in which the work is apprehended, has gone some way towards regarding it as ’discourse’.

One of the most important critics of Saussurean linguistics was the Russian philosopher and literary theorist Mikhail Bakhtin, who under the name of his colleague V. N. Voloshinov published in 1929 a pioneering study entitled Marxism and the Philosophy of Language. Bakhtin had also been largely responsible for what remains the most cogent critique of Russian Formalism, The Formal Method in Literary Scholarship, published under the names of Bakhtin and P. N. Medvedev in 1928. Reacting sharply against Saussure’s ’objectivist’ linguistics, but critical also of ’subjectivist’ alternatives, Bakhtin shifted attention from the abstract system of langue to the concrete utterances of individuals in particular social contexts. Language was to be seen as inherently ’dialogic’: it could be grasped only in terms of its inevitable orientation towards another. The sign was to be seen less as a fixed unit (like a signal) than as an active component of speech, modified and transformed in meaning by the variable social tones, valuations and connotations it condensed within itself in specific social conditions. Since such valuations and connotations were constantly shifting, since the ’linguistic community’ was in fact a heterogeneous society composed of many conflicting interests, the sign for Bakhtin was less a neutral element in a given structure than a focus of struggle and contradiction. It was not simply a matter of asking ’what the sign meant’, but of investigating its varied history, as conflicting social groups, classes, individuals and discourses sought to appropriate it and imbue it with their own meanings. Language, in short, was a field of ideological contention, not a monolithic system; indeed signs were the very material medium of ideology, since without them no values or ideas could exist. Bakhtin respected what might be called the ’relative autonomy’ of language, the fact that it could not be reduced to a mere reflex of social interests; but he insisted that there was no language which was not caught up in definite social relationships, and that these social relationships were in turn part of broader political, ideological and economic systems. Words were ’multi-accentual’ rather than frozen in meaning: they were always the words of one particular human subject for another, and this practical context would shape and shift their meaning. Moreover, since all signs were material - quite as material as bodies or automobiles - and since there could be no human consciousness without them, Bakhtin’s theory of language laid the foundation for a materialist theory of consciousness itself. Human consciousness was the subject’s active, material, semiotic intercourse with others, not some sealed interior realm divorced from these relations; consciousness, like language, was both ’inside’ and ’outside’ the subject simultaneously. Language was not to be seen either as ’expression’, ’reflection’ or abstract system, but rather as a material means of production, whereby the material body of the sign was transformed through a process of social conflict and dialogue into meaning.

Some significant work has followed in our own time from this radical antistructuralist perspective.10 It also has remote relations with a current of Anglo-Saxon linguistic philosophy which is far from concerned with such alien concepts as ’ideology’. Speech act theory, as this current is known, began in the work of the English philosopher J. L. Austin, and especially in his jocosely entitled How to Do Things With Words (1962). Austin had noticed that not all of our language actually describes reality: some of it is ’performative’, aimed at getting something done. There are ’illocutionary’ acts, which do something in the saying: ’I promise to be good,’ or ’I hereby pronounce you man and wife.’ There are also ’perlocutionary’ acts, which bring an effect about by the saying: I may succeed in convincing, persuading, intimidating you by my words. In the end, interestingly, Austin came to admit that all language is really performative: even statements of fact, or ’constative’ language, are acts of informing or affirming, and to communicate information is as much a ’performance’ as naming a ship. For ’illocutionary’ acts to be valid, certain conventions must be in place: I must be the kind of person licensed to make such statements, I must be serious about it, the circumstances must be appropriate, the procedures must be correctly executed, and so on. I cannot baptize a badger, and will probably have made things worse if I am not a clergyman anyway. (I choose this baptismal image because Austin’s discussion of appropriate conditions, correct procedures and the rest has an odd and not insignificant similarity to theological debates about sacramental validity.) The relevance of all this to literature becomes clear when we realize that literary works themselves can be seen as speech acts, or as an imitation of them. Literature may appear to be describing the world, and sometimes actually does so, but its real function is performative: it uses language within certain conventions in order to bring about certain effects in a reader. It achieves something in the saying: it is language as a kind of material practice in itself, discourse as social action. In looking at ’constative’ propositions, statements of truth or falsity, we tend to suppress their reality and effectivity as actions in their own right; literature restores to us this sense of linguistic performance in the most dramatic way, for whether what it asserts as existing actually exists or not is unimportant.

There are problems with speech act theory, both in itself and as a model of literature. It is not clear that such theory can finally avoid smuggling in the old ’intending subject’ of phenomenology in order to anchor itself, and its preoccupations with language seem unhealthily juridical, a matter of who is allowed to say what to whom in what conditions.11 The object of Austin’s analysis is, as he says, ’the total speech act in the total speech situation’; but Bakhtin shows that there is rather more involved in such acts and situations than speech act theory suspects. It is also dangerous to take ’living speech’ situations as models for literature. For literary texts are not of course literally speech acts: Flaubert is not actually talking to me. If anything, they are ’pseudo’ or ’virtual’ speech acts - ’imitations’ of speech acts - and as such were more or less dismissed by Austin himself as ’non-serious’ and defective. Richard Ohmann has taken this characteristic of literary texts - that they imitate or represent speech acts which themselves have never happened - as a way of defining ’literature’ itself, though this does not in fact cover all that ’literature’ is commonly taken to denote.12 To think of literary discourse in terms of human subjects is not in the first place to think of it in terms of actual human subjects: the real historical author, a particular historical reader and so on. Knowing about this may be important; but a literary work is not actually a ’living’ dialogue or monologue. It is a piece of language which has been detached from any specific ’living’ relationship and thus subject to the ’reinscriptions’ and reinterpretations of many different readers. The work itself cannot ’foresee’ its own future history of interpretations, cannot control and delimit these readings as we can do, or try to do, in face- to-face conversation. Its ’anonymity’ is part of its very structure, not just an unfortunate accident which befalls it; and in this sense to be an ’author’ - the ’origin’ of one’s own meanings, with ’authority’ over them - is a myth.

Even so, a literary work can be seen as constructing what have been called ’subject positions’. Homer did not anticipate that I personally would read his poems, but his language, by virtue of the ways it is constructed, unavoidably offers certain ’positions’ for a reader, certain vantage-points from which it can be interpreted. To understand a poem means grasping its language as being ’oriented’ towards the reader from a certain range of positions: in reading, we build up a sense of what kind of effects this language is trying to achieve (’intention’), what sorts of rhetoric it considers appropriate to use, what assumptions govern the kinds of poetic tactics it employs, what attitudes towards reality these imply. None of this need be identical with the intentions, attitudes and assumptions of the actual historical author at the time of writing, as is obvious if one tries to read William Blake’s Songs of Innocence and Experience as the ’expression’ of William Blake himself. We may know nothing about the author, or the work may have had several authors (who was the ’author’ of the Book of Isaiah, or of Casablanca?'), or to be an acceptable author at all in a certain society may mean writing from a certain ’position’. Dryden could not have written ’free verse’ and still have been a poet. Understanding these textual effects, assumptions, tactics and orientations is just to understand the ’intention’ of the work. And such tactics and assumptions may not be mutually coherent: a text may offer several mutually conflicting or contradictory ’subject positions’ from which to be read. In reading Blake’s Tyger poem, the process of building up an idea of where the language is coming from and what it is aimed at is inseparable from the process of constructing a ’subject position’ for ourselves as readers. What kind of reader do the poem’s tone, rhetorical tactics, stock of imagery, armoury of assumptions imply? How does it expect us to take it? Does it seem to expect us to take its propositions at face value, thus confirming us as readers in a position of recognition and assent, or is it inviting us to assume a critical, dissociating position from what it offers? Is it, in other words, ironic or satirical? More disturbingly, is the text trying to stand us ambiguously between the two options, eliciting from us a kind of consent while seeking simultaneously to undermine it?

To see the relation between language and human subjectivity in this way is to concur with the structuralists in avoiding what may be called the ’humanist’ fallacy - the naive notion that a literary text is just a kind of transcript of the living voice of a real man or woman addressing us. Such a view of literature always tends to find its distinguishing characteristic — the fact that it is written - somehow disturbing: the print, in all its cold impersonality, interposes its ungainly bulk between ourselves and the author. If only we could talk to Cervantes directly! Such an attitude ’dematerializes’ literature, strives to reduce its material density as language to the intimate spiritual encounter of living ’persons’. It goes along with the liberal humanist suspicion of all that cannot be immediately reduced to the interpersonal, from feminism to factory production. It is not, in the end, concerned with regarding the literary text as a text at all. But if structuralism avoided the humanist fallacy, it did so only to fall into the opposite trap of more or less abolishing human subjects altogether. For the structuralists, the ’ideal reader’ of a work was someone who would have at his or her disposal all of the codes which would render it exhaustively intelligible. The reader was thus just a kind of mirror-reflection of the work itself - someone who would understand it ’as it was’. An ideal reader would need to be fully equipped with all the technical knowledge essential for deciphering the work, to be faultless in applying this knowledge, and free of any hampering restrictions. If this model was pressed to an extreme, he or she would have to be stateless, classless, ungendered, free of ethnic characteristics and without limiting cultural assumptions. It is true that one does not tend to meet many readers who fill this bill entirely satisfactorily, but the structuralists conceded that the ideal reader need not do anything so humdrum as actually exist. The concept was merely a convenient heuristic (or exploratory) fiction for determining what it would take to read a particular text ’properly’. The reader, in other words, was just a function of the text itself: to give an exhaustive description of the text was really the same thing as to give a complete account of the kind of reader it would require to understand it.

The ideal reader or ’super-reader’ posited by structuralism was in effect a transcendental subject absolved from all limiting social determinants. It owed much as a concept to the American linguist Noam Chomsky’s notion of linguistic ’competence’, by which was meant the innate capacities which allowed us to master the underlying rules of language. But not even Levi- Strauss was able to read texts as would the Almighty himself. Indeed it has been plausibly suggested that Levi-Strauss’s initial engagements with structuralism had much to do with his political views about the reconstruction of post-war France, views about which there was nothing divinely assured.13

Structuralism is among other things one more of literary theory’s series of doomed attempts to replace religion with something as effective: in this case, with the modern religion of science. But the search for a purely objective reading of literary works clearly poses grievous problems. It seems impossible to eradicate some element of interpretation, and so of subjectivity, from even the most rigorously objective analysis. How, for example, did the structuralist identify the various ’signifying units’ of the text in the first place? How did he or she decide that a specific sign or set of signs constituted such a basic unit, without recourse to frames of cultural assumption which structuralism in its strictest forms wished to ignore? For Bakhtin, all language, just because it is a matter of social practice, is inescapably shot through with valuations. Words not only denote objects but imply attitudes to them: the tone in which you say ’Pass the cheese’ can signify how you regard me, yourself, the cheese and the situation we are in. Structuralism conceded that language moved in this ’connotative’ dimension, but it shrank back from the full implications of this. It certainly tended to disown valuations in the broader sense of saying whether you thought a particular literary work was good, bad or indifferent. It did so because this seemed unscientific, and because it was tired of bellelettristic preciosity. There was thus no reason in principle why you should not spend your life as a structuralist working on bus tickets. The science itself gave you no clue as to what might or might not be important. The prudishness of structuralism’s evasion of value-judgements, like the prudishness of behaviourist psychology, with its coy, euphemistic, circumlocutory avoidance of any language which smacks of the human, was more than just a fact about its method. It suggested the extent to which structuralism was the dupe of an alienated theory of scientific practice, one powerfully dominant in late capitalist society.

That structuralism has in some ways become complicit with the aims and procedures of such society is obvious enough in the reception it has received in England. Conventional English literary criticism has tended to divide into two camps over structuralism. On the one hand there are those who see in it the end of civilization as we have known it. On the other hand, there are those erstwhile or essentially conventional critics who have scrambled with varying degrees of dignity on a bandwagon which in Paris at least has been disappearing down the road for some time. The fact that structuralism was effectively over as an intellectual movement in Europe some years ago has not seemed to deter them: a decade or so is perhaps the customary timelapse for ideas in transit across the Channel. These critics operate, one might say, rather like intellectual immigration officers: their job is to stand at Dover as the new-fangled ideas are unloaded from Paris, examine them for the bits and pieces which seem more or less reconcilable with traditional critical techniques, wave these goods genially on and keep out of the country the rather more explosive items of equipment (Marxism, feminism, Freudianism) which have arrived with them. Anything unlikely to prove distasteful in the middle-class suburbs is supplied with a work permit; less well-heeled ideas are packed back on the next boat. Some of this criticism has in fact been sharp, subtle and useful: it has represented a significant advance in England on what existed before, and at its best displays an intellectual adventurousness which has not been greatly in evidence since the days of Scrutiny. Its individual readings of texts have often been remarkably cogent and rigorous, and French structuralism has been combined with a more English ’feel for language’ in valuable ways. It is simply the extreme selectivity of its approach to structuralism, one not always acknowledged, which needs to be highlighted.

The point of this judicious importation of structuralist concepts is to keep literary criticism in a job. It has been evident for some time that it is a little short on ideas, lacking in ’long perspectives’, embarrassingly blind both to new theories and to the implications of its own. Just as the EC can help Britain out in economic matters, then, so structuralism can do in intellectual ones. Structuralism has functioned as a kind of aid scheme for intellectually underdeveloped nations, supplying them with the heavy plant which might revive a failing domestic industry. It promises to put the whole literary academic enterprise on a firmer footing, thus permitting it to surmount the so-called ’crisis in the humanities’. It provides a new answer to the question: What is it that we are teaching/studying? The old answer - Literature - is not, as we have seen, wholly satisfactory: roughly speaking, it involves too much subjectivism. But if what we are teaching and studying is not so much ’literary works’ but the ’literary system’ - the whole system of codes, genres and conventions by which we identify and interpret literary works in the first place - then we seem to have unearthed a rather more solid object of investigation. Literary criticism can become a kind of w^tacriticism: its role is not primarily to make interpretative or evaluative statements but to step back and examine the logic of such statements, to analyse what we are up to, what codes and models we are applying, when we make them. ’To engage in the study of literature,’ Jonathan Culler has argued, ’is not to produce yet another interpretation of King Lear but to advance one’s understanding of the conventions and operations of an institution, a mode of discourse.”4 Structuralism is a way of refurbishing the literary institution, providing it with a raison d’etre more respectable and compelling than gush about sunsets.

The point, however, may not be to understand the institution but to change it. Culler seems to assume that an investigation of how literary discourse works is an end in itself, requiring no further justification; but there is no reason to suppose that the ’conventions and operations’ of an institution are less to be criticized than gush about sunsets, and enquiring into them without such a critical attitude will certainly mean reinforcing the power of the institution itself. All such conventions and operations, as this book tries to show, are the ideological products of a particular history, crystallizing ways of seeing (and not just of ’literary’ seeing) which are far from uncontroversial. Whole social ideologies may be implicit in an apparently neutral critical method; and unless studying such methods takes account of this, it is likely to result in little more than servility to the institution itself. Structuralism has demonstrated that there is nothing innocent about codes; but there is nothing innocent in taking them as the object of one’s study either. What is the point of doing this anyway? Whose interests is it likely to serve? Is it likely to give students of literature the impression that the existing corpus of conventions and operations is radically questionable, or will it rather intimate that they constitute some neutral technical wisdom which any student of literature needs to acquire? What is meant by the ’competent’ reader? Is there only one kind of competence, and by whose and what criteria is competence to be measured? One could imagine a dazzlingly suggestive interpretation of a poem being produced by someone who entirely lacked ’literary competence’ as conventionally defined - someone who produced such a reading not by following the received hermeneutical procedures but by flouting them. A reading is not necessarily ’incompetent’ because it ignores a conventional critical mode of operation: many readings are in a different sense incompetent because they follow such conventions all too faithfully. Still less is it easy to assess ’competence’ when we consider the way literary interpretation engages values, beliefs and assumptions which are not confined to the literary realm. It is no good the literary critic claiming that he is prepared to be tolerant about beliefs but not about technical procedures: the two are far too closely bound together for that.

Some structuralist arguments would appear to assume that the critic identifies the ’appropriate’ codes for deciphering the text and then applies them, so that the codes of the text and the codes of the reader gradually converge into a unitary knowledge. But this is surely too simple-minded a conception of what reading actually involves. In applying a code to the text, we may find that it undergoes revision and transformation in the reading process; continuing to read with this same code, we discover that it now produces a ’different’ text, which in turn modifies the code by which we are reading it, and so on. This dialectical process is in principle infinite; and if this is so then it undermines any assumption that once we have identified the proper codes for the text our task is finished. Literary texts are ’code- productive’ and ’code-transgressive’ as well as ’code-confirming’: they may teach us new ways of reading, not just reinforce the ones with which we come equipped. The ’ideal’ or ’competent’ reader is a static conception: it tends to suppress the truth that all judgements of ’competence’ are culturally and ideologically relative, and that all reading involves the mobilizing of extra-literary assumptions for the measuring of which ’competence’ is an absurdly inadequate model.

Even at the technical level, however, the concept of competence is a limited one. The competent reader is one who can apply to the text certain rules; but what are the rules for applying rules? The rule seems to indicate to us the way to go, like a pointing finger; but your finger only ’points’ within a certain interpretation I make of what you are doing, one which leads me to look at the object indicated rather than up your arm. Pointing is not an ’obvious’ activity, and neither do rules carry their applications on their faces: they would not be ’rules’ at all if they inexorably determined the way we were to apply them. Rule-following involves creative interpretation, and it is often not at all easy to say whether I am applying a rule in the way that you are, or even whether we are applying the same rule at all. The way you apply a rule is not just a technical affair: it is bound up with wider interpretations of reality, with commitments and predilections which are not themselves reducible to conformity to a rule. The rule may be to trace parallelisms in the poem, but what is to count as a parallelism? If you disagree with what counts for me as a parallelism, you have not broken any rule: I can only settle the argument by appealing to the authority of the literary institution, saying: ’ This is what we mean by a parallelism.’ If you ask why we should follow this particular rule in the first place, I can only once more appeal to the authority of the literary institution and say: ’This is the kind of thing we do.’ To which you can always reply: ’Well, do something else.’ An appeal to the rules which define competence will not allow me to counter this, and neither will an appeal to the text: there are thousands of things one can do with a text. It is not that you are being ’anarchistic’: an anarchist, in the loose, popular sense of the word, is not someone who breaks rules but someone who makes anoint of breaking rules, who breaks rules as a rule. You are simply challenging what the literary institution does, and although I might ward this off on various grounds, I certainly cannot do so by an appeal to ’competence’, which is precisely what is in question. Structuralism may examine and appeal to existing practice; but what is its answer to those who say: ’Do something else’?