Friday, July 20, 2007



Of the several components of post-1968 film studies, it was semiotics, or what was presumed to be semiotics, that most provoked the film criticism establishment. Semiotics and semioticians were denounced in the pages of quality dailies and weeklies: semiotics was 'a procrustean enterprise', comparable to 'painting by numbers', at once 'un­wittingly absurd' and 'insidiously political', practised by 'possessed sectarians', 'pod-people', and 'overdressed ladies bedecked in bangles and baubles', whose general demeanour had 'the poised vigilance of a lobotomised ferret1.' The outcry seems inappropriate to Saussure's original proposal for a theory of signs:

A science that studies the life of signs within society is therefore conceiva­ble. . . I shall call it semiology (from the Greek semeion, sign). Semiology would show what constitutes signs, what laws govern them. Since the science does not yet exist, no one can say what it would be; but it has a right to existence, a place staked out in advance.2

Thus stated, semiology, or semiotics, as it became more generally called, is hardly very threatening. What rattled the critics were the claims made on behalf of a semiotics of cinema, which amounted, if they were to be believed, to a notice of redundancy for those engaged in conventional criticism.

A semiotics so conceived heralded the end of all traditional aesthe­tics. Ideas of art as organic unity, as revelation, as the communication of inspired vision, were discarded and replaced by the supposition that all meanings and aesthetic effects were explicable in terms of determining structures and mechanisms. Art, in a word, was open to scientific analysis. In providing 'a scientific basis for aesthetic judge­ments in the cinema', semiotics, it was argued, would mark the end of critical impressionism.3 No longer would notes jotted down during a single screening, worked up in a literary style and larded with cul­tural references, suffice as criticism. Moreover, it was believed that the semiotic project would contribute to the political cause of demystification, of denaturalising representations and exposing them as constructed precisely of signs. The very fact of trying to forge a scientific semiotics would, Metz proposed, 'bring with it a great capacity for demystification, an irreversible break with impressionistic and idealist discourses and all claims for the ineffable'.4 Although there is no reason to suppose that Saussure himself had anything so radical in mind, his proposal for a science of signs could manifestly be given such an interpretation, and to this extent the rancour of the critical establishment was not misdirected.

Despite his call for a semiotics, Saussure's actual achievement per­tained to the more limited field of natural language, the science of which — linguistics — he is commonly held to be the founder. Even so, he committed none of his discoveries to paper, leaving it to his students to reconstruct his theories from lecture notes after his death. Such, however, has been the impact of his thought, it can be said 'we are all Saussureans now’.5

In view of the many admirable summaries of Saussure's work we shall limit our account to the broadest of outlines. The first important point to make is that Saussure broke with previous approaches to the study of language in asking hot how it developed but how it works. Philology had been content to trace the evolution of a word or sound over the centuries; Saussure sought to explain how that word, that sound produced meaning. The essence of the explanation he gave - and this is the second crucial point - is that meaning exists only within a system. In contrast to the naive view that language acquires its meaning by reference to a world of things anterior to, or independent of, signification, Saussure argued that meaning derives solely from the system within which particular utterances are articulated. The system, known as langue, and actual or potential utterances, parole, may be compared to the rule system of chess and to the set of moves that may be actually or potentially played. Langue defines both what are permissible or impermissible utterances (as do the rules of chess in relation to moves) and what their significance is (again, as in chess).

In explicating the functioning of language as a system Saussure distinguished between the signifier and the signified, which together comprise the linguistic sign (typically a word). The signifier is the actual sound (or if written, the appearance) of the word; the signified is the concept or meaning attached to it. The relationship between the two is arbitrary, since there is nothing in the nature of things to dictate that a signified should have a particular signifier - the same signified has different signifiers in different languages. But more than that, Saussure claimed, the value of a signifier is given not by its relation to a pre-given signified but by its relation to other signifiers, a concept perhaps best explained once again by analogy with chess. If the meaning of a signifier is analogous to the value of a piece on a chessboard, then it becomes evident that meaning will change according to context in the same way that the value of, say, a pawn will depend on what stage of the game has been reached, where it is in relation to other pawns, how many pieces are left on the board, and so on. In other words meaning is produced by a system of dif­ferences. Such differences may be specified in relation to two basic axes, the paradigmatic (or axis of selection) and the syntagmatic (or axis of combination). The former pertains to potential substitutes for any element in the signifying chain: the substitution of 'h' for 'c' in the word 'cat5, or the substitution of 'television' for 'mat1 in the sentence 'the cat sat on the mat'5. The latter axis, the syntagmatic, runs as it were horizontally from one signifying element to the next, and pertains to the way meaning is established by the combination of any given element with other elements in the signifying chain. The meaning of an element is therefore determined by its relation both to the present set of elements it is in combination with and to the absent set of elements that could be substituted for it. The essential point is, as Saussure put it, 'in language there are only differences'.6 This so-called 'diacritical' theory of meaning was to prove the single most influential idea operative within film semiotics.

Other approaches to semiotics in addition to Saussure's were taken up as being of possibly greater relevance to film studies, the principal among which was that developed by the American philosopher C. S. Peirce around the turn of the century. It was his philosophical rather than linguistic investigations that took him in the direction of what he called 'semiotic', specifically through a concern with sym­bols, which he saw as the woof and warp' of all thought and scientific research.

The Peircean sign points in two directions: on the one hand towards the person to whom it is addressed and in whose mind it creates an idea or secondary sign, called the interpretant, and on the other towards that which it stands for, called the object. A sign thus mediates between object and interpretant, entities that would other­wise be unrelated. Very roughly the object is equivalent to what another discourse terms the referent, the thing in the real world that the sign stands for, except that in Peirce, as has been pointed out by Silverman, the status of the real is unclear.7 Though at times Peirce maintains that there is direct experience of reality, elsewhere he argues that it can be known only through representations, so that objects are simply representations whose validity is consensually established. This latter position tends towards a Nietzschean perspectivism, as when Peirce writes 'My language is the sum total of myself; for the man is the thought.'8 In any case, the interpretant is the idea produced in the mind of the interpreter by the sign, and is not dissimilar to Saussure's signified. Yet for Peirce this too is a sign, with its own interpretant, which in turn is a sign with its interpretant, and so on - opening up a prospect of what post-structuralists would call un­limited semiosis.

Depending on whether they are considered from the standpoint of the mediating sign, or that of the object, or that of the interpretant, signs can be classified into different trichotomies, only the second of which has been taken up by film theory and therefore need concern us here. The relevant classification, then, based on the relationship of signs to their object, is that of icon, index and symbol.

For a sign to be an icon it must have some physical quality or configuration of qualities that it shares with the object. As Peirce puts it, 'Anything whatever, be it a quality, existent individual or law, is an icon of anything, insofar as it is like that thing and is used as a sign of it5.9 Resemblance, then, is the basis of iconicity. Examples of icons would be representational paintings, diagrams, statues, photographs and onomatopoeic words. Next, an index is a sign that becomes so 'by virtue of a character which it could not have if its object did not exist5, irrespective of whether it is interpreted as a sign.10 Another way of putting it is to say there is necessarily a causal relationship between object and index, so the sign is the effect of the object. Just about any calibrated instrument, such as a barometer, thermometer, speedometer or ammeter, functions as an index of what it is measuring; as also does smoke of fire, a weathercock of wind direction, a knock on the door of a visitor or pain of physical damage. Indeed, it is through a continual reading off of indexical meanings that one is able to make sense of the world at all. Lastly, a symbol is arbitrarily linked to its object 'by means of an association of ideas or habitual connection'.11 With neither resemblance nor causal rela­tion between sign and object, the basis for signification must be con­vention, and therefore most natural language and languages parasitic on it, like Morse code, as well as the codes of gesture (to some extent) and dress, consist of symbols. It should also be noted that a sign may fall into more than one category. The photograph, for instance, as was pointed out by Peirce, is both icon, in that it is similar to its object, and index, in that it is an effect on photographic emulsion of light interacting with the object.

At various times during the last two decades Peirce's ideas have been taken up by film theorists, though they have never been as in­fluential as Saussure's. The best known adoption of Peirce was by Peter Wollen in his widely read and cited Signs and Meanings in the Cinema, in which he pointed out that cinema operates with all three categories of sign: index (by virtue of being the effect of the photo­graphed real), icon (through sound and image) and symbol (in that it uses speech and writing). He berated other theorists, for example Metz, for having concentrated on only one of these dimensions to the exclusion of the others. More recently, Kaja Silverman has shown that Peirce enables distinctions to be made that are unavailable to Saussure, notably that whereas the relation of linguistic signifiers to their signifieds is preponderantly symbolic, the specific components of film, such as photography, editing, lighting and so on, are weighted equally towards the iconic and the indexical.'2 Because of cinema's typically greater involvement with these latter, it has been argued elsewhere, there is less of an apparent gap between the sign and its object than there is with symbols, which explains cinema's capacity to naturalise the most blatant stereotypes.13 Peirce's ideas have also recently been taken up by Gilles Deleuze and Teresa de Lauretis, and in view of this renewed interest they may play a more prominent role for film studies in the future.14

By and large, however, the film semiotics of the late 1960s and early 1970s was derived from Saussure. One reason for this domi­nance is almost certainly to be found in the line from film theory's political underpinning to the thesis that stressed language's status as cultural production and thence to any position, such as Saussure's, in which language was dependent on societal convention. Such a position was at least consistent with the claim - requirement, even - that language did not mirror but constructed the work, and was therefore inevitably complicit with ideology. Both the notion that meaning is produced within a signifying system and that the relation­ship between signifier and signified is arbitrary could be squared with the idea of a language as social product. This conception of language as a cultural system interposed between human beings and the real world was to receive further powerful support from the work of Lacan.

Two aspects of Lacan's thinking on language were to be crucial for film theory. The first of these was a Nietzschean conception of language as constitutive: 'the world of words. . . creates the world of things';15 'things only signify within the symbolic order"; 'nothing makes sense until you put a sign on it'.16 The idea that neither words nor images transmitted neutrally a pre-given reality, but offered a perspective through which reality was constituted, is readily traceable to Nietzsche. According to Nietzsche, language coerces us into think­ing in particular ways through categories that remain largely uncon­scious. Language, as he famously expressed it, secretes a mythology. By substituting 'ideology3 for 'mythology1 one gets exactly what film theorists took up from Lacan's reading of Saussure; that film is a language appearing to render the real transparently but actually sec­reting an ideology. The task therefore was to create a new language, enabling men and women to think what had previously been unthink­able.

The second aspect was Lacan's conception of meaning as produced in the exchange between subject and a set of signifiers. Here Lacan modified Saussure's idea of the linguistic sign by giving primacy to the signifier and by introducing a bar between signifier and signified (thus, S/s , where S is the signifier and s the signified), implying that there is a continual sliding of signifieds under signifiers as these enter into new relationships. In other words, meaning is not at all the stable relationship between signifier and signified presumed by Saus­sure. What stops the slide and momentarily fixes meaning is the punc­tuation of the signifying chain by the action of the subject, expressed by Lacan in the graph below.17 In this illustration the vector SS' represents the signifying chain and the vector $ represents the re­troactive construction of meaning by the subject. Meaning is always




provisional and changes as new elements are added to the signifying chain, with each successive ele­ment setting up expectations as to what will follow and retroactively changing the meaning of what precedes it. Thus meaning is produced by the subject in this process of punctuation; but, equally, the subject is produced by the meanings available in the signifying chain, for the subject is such by virtue of a self-conception that is only available within discourse. The desire of the subject engenders varying interpretations of the unfolding text; the text offers in return the condition of subjectivity. For Lacan there is, therefore, an unceasing dialectic of the subject and meaning, an idea that would recur in various guises within film theory.

The overriding question for any semiotics of cinema was first and foremost, is cinema a language? It was generally agreed that the ans­wer to this hinged on whether cinema directly imitated, or was analogous to, or was, in a Bazinian sense, an extension of reality; or whether it was a form of writing, dependent on an arbitrary and conventional sign system.

The question was answered by the pioneering film semiotician Christian Metz with an equivocal 'yes and no'. Yes, it was a language, but no, it was a language without a tongue, where tongue is under­stood in a Saussurean sense as 'a system of signs intended for inter­communication'.18 Like Saussure in relation to natural language, Metz wanted to achieve an understanding of how films are understood, but he recognised fundamental differences between language and cinema that prevented the wholesale importation of Saussure's con­cepts. The concept of tongue was inapplicable to cinema for three basic reasons. The first of these was that cinema is not available for inter-communication; if it is communication at all (rather than expression), it is one-way communication. Next, the filmic image is quite unlike the Saussurean sign, with its arbitrary relation between signifier and signified, and instead, in its reproduction of the condi­tions of perception, can be termed 'a block of reality3: The cinema has as its primary material a body of fragments of the real world, mediated through their mechanical duplication.'19 Whereas a verbal signifier acquires its significance from its place within a system, that of an image derives from what it duplicates. Moreover, as well as resemblance there is a material link between the image and its object, making it index as well as icon, and therefore motivated. However, there were qualifications even in Metz's early work, as when he acknowledged that an image necessarily involves distortion and defor­mation. Later, this idea was developed through the identification of codes at work in the image. Despite such qualifications, the general tenor of the argument was that cinema duplicated rather than articu­lated reality.

The third reason for refusing cinema the status of langue was that it lacks the double articulation that, according to Andre Martinet and other linguisticians, is the hallmark of natural language. The characteristic economy of language, through which an infinity of utterances can be generated by means of a very small number of basic units, is achieved through this double articulation. At the level of the first articulation a limited number of words (more properly mor­phemes, the lexicon of any language) are combined in different orders to provide a limitless number of utterances. But a still greater economy is permitted through a second articulation, by which mor­phemes are made up of a very much smaller number of phonemes, these being the smallest distinctive units of a language. These are without meaning in themselves, but systematised on the basis of phonological properties to produce consequential differences - in English that, say, between T and V. Cinema lacks this second articu­lation; it has nothing corresponding to phonemes. The most obvious candidate would be the shot, except that, unlike the phoneme, which by itself is without meaning, the shot possesses a meaning. But nor, contended Metz, is the shot the equivalent of the single word. A number of reasons are offered for this contention: shots, like state­ments but unlike words, are infinite in number; they do not pre-exist in a lexicon, but are to an extent the invention of the filmmaker (as statements are of the speaker). Moreover, the shot is a unit of dis­course (The image of a house does not signify "House" but "Here is a house".') and its meaning is not given by a system of paradigmatic contrasts, i.e., it is not determined by the absent units that could take its place.20 Metz concluded that the shot is more like a statement than a word, but even here the resemblance is limited in that a state­ment is reducible to discrete elements, the morphemes and phonemes, in a way that the shot is not.

If cinema is not langue, it is nonetheless language, at least 'to the extent that it orders signifying elements within ordered arrangements different from those of spoken idioms - and to the extent that these elements are not traced on the perceptual configuration of reality itself (which does not tell stories)'.21 Cinema transforms the world into discourse, and is not therefore simple duplication. But a semiotics of the cinema cannot work at the level of the image, since each image is unique, novel and analogous to reality, with its meaning produced not by its place within a system but by what it duplicates. There is no process of selection from a lexicon of images in cinema as there is from the verbal lexicon of a natural language.

It was because of this paradigmatic poverty that Metz was led to explore the semiotics of cinema in terms of syntagmatic relations. Combination, not selection, was to be the key to its understanding. While the image might not be coded the narrative certainly was, and since cinema consisted predominantly of narrative, and indeed, since its historical development had produced a number of recognisable narrative forms and structures, it was appropriate that a semiotics of cinema should concentrate on the spatio-temporal logic of narrative. Metz's so-called grande syntagmatique was an attempt to provide an exhaustive classification of the segmentation of cinematic narratives.22 Arranged in a hierarchy from the autonomous shot, the smallest seg­ment, to the sequence, the largest segment, the system of classification would permit any film's narrative syntax to be formalised. The decisive element of the classification is the so-called autonomous segment, of which the most obvious example would be the temporally continu­ous scene. Thus hierarchisation in terms of length and complexity is what constitutes the classification. In all there are eight levels within the hierarchy, and Metz approached their arrangement through a series of either/or disjunctions. For instance: did the segment consist of one or more than one shot? If more than one, was the segment chronological or achronological? If chronological, was it simultane­ous or sequential? - and so on. The resultant hierarchy of autonomous segments runs as follows:

1) the autonomous shot, which is of two kinds, either the sequence shot, where a whole scene is contained within a single shot, or the insert, for instance a subjective image within a larger segment

2) the parallel syntagm, as occurs when two motifs are interwoven in a montage in which their temporal or spatial relationship is unspecified

3) the bracketing syntagm, a montage of brief shots representative of, say, a situation or way of life

4) the descriptive syntagm, in which a series of shots comprise a composite description of a single moment

5) the alternating syntagm, which runs together two sequences in alternate shots, each with its own temporal development yet as a whole implying simultaneity, as in just about any chase sequence with shots of pursuers and pursued

6) the scene, in which a succession of shots implies temporal continuity ,

7) the episodic sequence, where there is an organised discontinuity of shots

8) the ordinary sequence, where the discontinuity is simply the omission of moments judged unimportant.

By way of illustration of its categories and to show the application of the method Metz proceeded to analyse the film Adieu Philippine, which remains the locus classicus of the grande syntagmatique.23

While the rigour of the grande syntagmatique still commands respect - David Bordwell having recently termed it 'the outstanding achievement5 in the study of cinematic narrative structure - film theorists by and large have found little application for it.24 One case study that did make use of it was John Ellis's analysis of Ealing, in which he showed that some two thirds of Passport to Pimlico consists of segments respecting a unity of time and place (scenes, ordinary sequences, and autonomous shots), a syntactic arrangement that con­tributed to the film's realism. Even Elk's, however, found there were inadequacies in Metz's classificatory scheme, a feeling that was echoed by other commentators. It was found, for example, that certain of the categories, such as that of the autonomous shot, were so broad as to include such a diversity of cinematic forms that they were of little demarcatory use. Another difficulty was deciding in which par­ticular category any given segment should be located - a scene and an ordinary sequence are often hard to distinguish. There was also the more general question as to what is to count as an autonomous segment, something that in practice can often only be settled on the basis of a reading of the film, thereby introducing a semantic element into what was conceived as a purely syntactic exercise. It was further found that many films, especially those out of the run of the mainstream, contained passages that did not fit neatly into any of Metz's categories. Consequently there was a tendency among theorists when analysing films to segment them in terms other than those given through grande syntagmatique, viz. the textual read­ings performed by Heath and by Bellour. Even without these various difficulties there existed a still more serious charge against the grande syntagmatique, namely, that it pointed towards an arid formalism that could neither account for film's specific production of meaning nor satisfy the political demand that its mechanisms for the reproduction of ideology be exposed.

Metz's revised semiotics was at once more complex and more flex­ible than this early model.25 While retaining the concept of cinema as a language he abandoned the attempt to locate a specific set of rules underlying the articulation of each cinematic text. Instead he came to treat cinema and the cinematic text as fields of signification in which a heterogeneity of codes, some specific to the cinema and others not, interacted with one another in ways that were specific, systematic and determinate at certain specified levels of cinematic discourse (individual films, particular genres) and hence at certain specified levels of analysis. Among specifically cinematic codes he distinguished codes of editing and framing, of lighting, of colour versus black and white, of the articulation of sound and movement, of composition, and so on. Non-cinematic codes included costume, gesture, dialogue, characterisation and facial expression. A further important distinction was made between cinematic codes and cinema­tic sub-codes, where the former organise elements potentially or actu­ally common to all films, say lighting, and the latter refer to specific choices made within a particular code, say that of low-key in prefer­ence to high-key lighting. Codes, therefore, do not conflict, whereas sub-codes do, it being a matter of one choice rather than another. Different codes and their sub-codes are in a syntagmatic relation of combination; sub-codes from the same code are in a paradigmatic relation of substitution. The codes of genre and authorship are addi­tive, as in the Westerns of Ford or the comedies of Hawks; but the sub-codes Ford and Hawks within the code of authorship, and the sub-codes Western and comedy within the code of genre, are (gen­erally speaking) only commutable. A cinematic code is often defined predominantly by the meanings of its sub-codes, a weighting that contrasts with the codic stability of natural language. Cinematic codes may be seen as consisting of signifiers awaiting the signifieds that come from their mobilisation in sub-codes. For example, the general significance of the code of authorship cannot be grasped outside of the specific operation of the sub-codes Ford, Hawks and the rest.

The film semiotics propounded by the novelist and film director Pier Paolo Pasolini was utterly different in inspiration and conception from that of Metz, being the theoretical corollary of the uncom­promisingly anti-bourgeois realism of his films. While he agreed with Metz that cinema has no langue, it was for a very different reason. For Pasolini 'the cinema is a language which expresses reality with reality. So the question is: what is the difference between the cinema and reality? Practically none.' In addition to this, he stated: When I make a film . . . there is no symbolic or conventional filter between me and reality, as there is in literature.'26

Where Pasolini parted company with Metz was in his contention that cinema is a language with a double articulation, albeit one quite distinct from that of natural language. Cinema's smallest units, cor­responding to phonemes, are, according to Pasolini, objects, actions or events that are unaltered by being reproduced on film. Termed 'cinemes', these possess their own meaning, one that is natural rather than conventional, and are combined into larger units - shots - that are the basic significant units of cinema and correspond to morphemes in natural language. It is through this second articulation, the selec­tion and combination of objects and events from the real world (the so-called 'profilmic - that which is in front of the camera), that the cinema is able to articulate reality. As distinct from phonemes, how­ever, cinemes are both infinite, or at least countess, and have the feature that they are, as it were, compulsory: 'we cannot but choose from among the cinemes that are there, that is to say, the objects, forms and events of reality which we can grasp with our senses'.27

Yet another approach to cine-semiotics was offered by Umberto Eco, who was critical of both Pasolini and Metz. Pasolini came in for criticism on a number of grounds, among them his conception of reality and his thesis that the objects and events of the real world, quite outside any cultural code, provide the primary constituents of cinematic discourse. Far from being presented with reality in the cinema, Eco argued, we are presented with signifiers subject to cultural codes through which they are read as signifieds. Where Pasolini perceived nature, Eco detected culture. Events, actions, objects, forms of human interaction such as gesture, far from having the supposed extra-cultural rawness Pasolini imagined, are inextricably imbricated with convention, code, system and, by extension, ideology. Pasolini's error was in effect to conflate signifier, signified and referent under the rubric of reality. A second criticism was that the analogy between cinemes and phonemes did not stand up, because phonemes only have meaning in combination, not in themselves, whereas cinemes, in being recognisable objects, do have meaning in themselves. And the further supposed equivalence between the shot and the morpheme was also open to question. Like Metz, Eco suggested that the shot was much closer to an utterance than to a single word. A shot of Sylvester Stallone naked to the waist firing a rocket launcher does not signify 'Rambo', but rather 'Here is Rambo', or more probably 'Here is Rambo single-handedly defeating the Evil Empire'.

Eco also criticised Metz along similar lines. He, too, had failed to appreciate the extent to which images were not mere simulacra or duplicates of reality, and therefore non-arbitrary and motivated, but were indeed the habitat of codes. Metz's assumption of this irreduc­ible primacy of the image, analogous to reality, had supported his conception of cinema as a language without a langue and had led to a concentration on how images are combined in syntagmatic struc­tures. Eco contended that, far from inhabiting a domain below the level of codic organisation, images owe their very existence to the workings of cultural codes, of which no fewer than ten are potentially operative in the communication of the image: codes of perception, codes of transmission, codes of recognition, tonal codes, iconic codes, iconographic codes, codes of taste and sensibility, rhetorical codes, stylistic codes and codes of the unconscious.

Eco's contention can be illustrated by considering just one theme, that of the cultural training necessary in order to perceive a similarity between a physical object and an iconic representation of it. Because the physical properties of the two are very different, and therefore give rise to different perceptual experiences, similarity can only be perceived as a result of a cultural background that both includes a knowledge of the conventions of, say, painting and has specified what should be regarded as pertinent in determining similarity. For instance, a single line image of a horse's profile relies on a cultural decision that it shall count as an image and as such requires a trained eye to see it as a horse's profile. In other words, 'similitude is produced and must be learned'.28 Indeed, the history of the visual arts abounds with examples of works that to us seem self-evidently realistic but confounded their contemporaries because they transgressed the con­ventions of the time. Even (and from the standpoint of a cine-semio­tics, especially) the photographic image, seemingly so analogous to reality, requires training to be recognised.

From the above it will have become apparent that Eco's principal quarrel with both Pasolini and Metz was their too narrow delimita­tion of the bounds of the cultural. In fact Eco's declared aim was to explain every case of signification 'in terms of underlying systems of elements mutually correlated by one or more codes'.29 His stress on the importance of convention and culture is taken over into his con­ception of the sign, which he defines as everything that, on the grounds of a previously established social convention, can be taken as something standing for something else. Convention, therefore, is the necessary condition for signification. Such conventionality, as we have seen, extends even to iconic signs, which seem at first sight to elude it. Their apparent indubitable motivation yields to an appreciation of the codes that make them possible as signs. On the other hand, Eco does allow that iconic signs are not completely arbit­rary and that they reproduce some but only some of the conditions of normal perception of an object. Those features that are held in common between image and object, for example the stripes on a drawing of a zebra, are still subject to the normal perceptual code through which we perceive the object, in this case, as striped.

In developing his argument apropos cinema, Eco registered his disagreement with both Metz's contention that it had a single articu­lation and Pasolini's that it had a double articulation, and claimed for it instead a triple articulation. At the first level of articulation the total image can be broken down into meaningful units, recognisable in themselves, known as 'semes'. For example, 'man in bizarre blue outfit with cape' and 'New York skyline' are semes that combine to form a shot of Superman flying over New York. Semes can in turn be analysed into smaller iconic signs, such as 'clenched fist" or 'deter­mined jawline', which are only recognisable in the context of the seme since they are part of a graphic continuum and appear as non-dis­crete. This is the second articulation. Then, finally, iconic signs can be analysed as comprising a third articulation of the conditions of perception, that is things such as angles, curves, textures, effects of light and shade, and so on. These in themselves have no meaning, and in this respect are analogous to phonemes, being defined simply in differential and oppositional terms. But they are essential to the construction of meaning in that their progressive alteration will at a certain point articulate a different iconic sign. The overall effect of this unique triple articulation of the cinematic code is to permit a far greater degree of realism than any other form of representation. 'Confronted with a conventionalisation so much richer, and hence a formalisation so much subtler than anything else, we are shocked into believing we stand before a language which restores reality to us'.30

Like Metz, Eco was to subject his early thinking to an auto-critique and to move away from what became seen as a static conception towards one with a greater degree of flexibility. Instead of thinking of signs in terms of elementary units with fixed values (the basis of the triple articulation), which was rejected as emphasising structure at the expense of process, Eco now argued that signs are better thought of as sign-functions correlating a unit of expression with a unit of content in a temporary encoding: 'Signs are the provisional result of coding rules which establish transitory correlations of ele­ments, each of these elements being entitled to enter - under given coded circumstances - into another correlation and thus form a new sign.'31 It is context that determines what is and what is not to count as an element of a sign. Signs are therefore sensitive to context.

As a consequence of this new emphasis, semiotic analysis is no longer a matter of identifying a fixed number of articulations in fixed interrelationships because, depending on context and point of view, 'an element of first articulation can become an element of second articulation and vice versa'.32 As de Lauretis puts it:

Even if in a given iconic continuum, an image, one can isolate pertinent discrete units or figurae, as soon as they are detected, they seem to dissolve again. In other words, these 'pseudo-features' cannot be organised into a system of rigid differences, and their positional as well as semantic values vary according to the coding rules instituted each time by the context.33

Signs, or rather sign functions, are to be seen as texts whose elemen­tary units can be identified only within a signifying process. The various codes then become purely temporal devices posited in order to explain a certain message rather than a secure ground to meaning.

Under such a conception, 'the classical notion of "sign" dissolves itself into a highly complex network of changing relationships' and meaning becomes an effect of a continual process of codic readjust­ment without any final referent or closure.34 Eco's revised semiotics can be seen as consistent with post-structuralism and its denial of all fixity of meaning, something we shall come to later in this chapter.

The effect of semiotics, or more particularly its central thesis that meaning is produced by a system of differences, was to call into ques­tion existing modes of thinking about cinema. The insistence that cinema is production ran against its innocent reception 'as natural, as life, as beauty' unfolding before the spectator.35 More particularly, it challenged conventional modes of thinking, dependent as they were on such notions as 'source', 'origin', 'centre', 'expression', 'represen­tation', 'full subject5 and so on. If meaning was produced through, and only through, signification, it could not have a source or origin elsewhere that needed merely to be expressed in language or film. Similarly, there was no longer any question of signification represent­ing that which already existed. For anything to exist as an object it had to be encoded as such, and therefore it was a matter of one sign for, or against, another. Such a refusal of pre-significant or extra-dis­cursive objects extended to the human subject, who was also denied the satisfaction of apparent self-sufficiency outside of language. For if the subject's self-conception (an essential constituent of its being) could be formed only in signification it could no longer be the 'full subject" of humanist ideology.

According to Stephen Heath there were two traditions in thinking about the relationship of language and cinema prior to semiotics. The first of these was a cinematic purism that saw in language, 'that dangerous supplement', with its power to misrepresent as well as represent, something that clouded the direct truth of the image.36 The aesthetic correlative to this epistemological apartheid between word and image was realism. The second tradition held that cinema was, precisely, a language, and it was because of this that it was en-tided to claim parity with the other arts. With resources as extensive or more so than those of literature or painting, film could become the means of expression of the artist's vision. The camera was not merely the mechanical device for recording pre-existing reality but the pen, the brush, of the creative cineaste. Here, in other words, was an aesthetic of authorship.

Since authorship and realism are considered at length in chapters 4 and 6, here we shall simply signal the effect semiotics was to have on thinking about each of them. As far as realism was concerned, the central challenge was to its assumption of transparency. Far from being a window on the world, which for some theorists was it. unique capacity and true vocation, cinema is, according to the semioticians, a work of construction, always within signification. The seeming real­ity it constructs is only accepted as such because it coheres with the prevailing ideology's version of reality. The reality effect is no more than a set of codes subservient to ideology. Film is not reproduction and representation of a pre-existent reality but the production and construction of an imaginary one. Semiotics similarly intervened against traditional ideas of the author as the origin of the meaning of the text. Because meanings do not pre-exist the text, there can be no question of expression of authorial intent or communication of meaning. Any ascription of authorship can only be in terms of a construction from the codes functioning within a text or corpus of texts. Whatever the intentions, conscious or unconscious, of a John Ford, the distinctive Fordian structure is to be found in the films he directed and not elsewhere. For both realism and authorship, then, the impact of semiotics was to render the conventional position untenable.

It was indeed the work of Stephen Heath that was to give the decisive orientation to the study of semiotics in film. In relation to the debates among the previously mentioned film semioticians, Heath sided with Eco against the early Metz and Pasolini, who, he main­tained, in supposing that the image duplicated reality, ignored the specific activity of cinema. Underlying the apparent naturalness of the image, there are, as Eco had shown, processes of codification and conventionalisation. Nevertheless, for Heath it was Metz who had to be the focus of attention, since in his revised semiotics he succeeded in bringing out the social basis of cinema through codes and conventions that were normally invisible. Heath's concern with the relation between spectator and film was, more than anything else, what moved film semiotics in a new direction. Here he parted company with Eco, who had turned away from any such project, and instead followed Julia Kristeva in her call for a theory of the speaking subject constituted within language. This would entail, she said, leaving behind the study of formalised meaning systems as the instrument of transcendental subjects, and entering a phase under the sign of psychoanalysis in which signification was seen as involved in the construction of subjectivity.37 Heath's deliberations on the implications of this shift from the structures of the text to the process of reading marked him out as considerably more than an astute com­mentator on continental semiotics.

The development of these ideas occurred in two distinct phases, with Althusser and Lacan respectively as the predominant influences. In the remainder of this chapter we shall be concerned with detailed elaboration of each phase and with some of the debates to which they gave rise.

In the first phase the institution of cinema was held to constitute individuals as subjects, in a manner analogous to ideological state apparatuses. Addressed by the text, the spectator accepts the identity assigned and is thereby fixed in a position where a particular mode of perception and consciousness appears natural. He or she is locked into a structure of misrecognition, into an imaginary relationship to the real conditions of his or her social existence. The ideological perspective imposed by the text makes it, according to this concep­tion, seem like a window on reality. During this phase, therefore, there was an attempt to relate the materiality of ideology to that of signification, where ideology was understood not as a system of ideas but as a practice of representation producing 'the subject as the place where a specific meaning is realised'.38 In a word, the reader was interpellated by the text, the spectator by the film.

One inflection of the theory was towards a typology of texts and drew upon the work of Emile Benveniste. His central contention was that linguistics, and in particular the analysis of discourse, could no longer afford to ignore the role of the subject within signification, for the very good reason that language is so deeply marked by sub­jectivity. One aspect of this is the role played by personal pronouns, which, unlike common nouns with their relatively fixed meanings and referents, signify entirely according to context. Pronouns and other so-called shifters like 'here', 'now5 and 'this' lack a stable, con­tinuous significance, but are nonetheless integral and indispensable to language. Without the possibility of each speaker being established as a subject through the use of shifters, notably the first person pro­noun, language might even cease to function at all. Because of the entrenchment of subjectivity in language, Benveniste urged linguis-ticians to turn their attention from the abstract system of langue to the operation of discourse within specific social contexts.

To this end he proposed a distinction to supplement that between langue and parole, namely, between enunciation (enonciation), the act of speaking, and the enounced (enonce), what is spoken. This then allows a further distinction between the subject of the enunciation, the subject who speaks, and the subject of the enounced, who is represented within the utterance. In most social exchanges there is no need to distinguish between the two subjects and they are taken as coinciding. But there are evidently occasions when the two do appear as distinct, as when a speaker utters a self-referring lie, a pointed instance of which is the liar paradox, the statement 'I am lying5. If the subject of the enunciation is telling the truth then the subject of the enounced is lying, and vice versa; they cannot both be telling the truth, therefore they are necessarily distinguished.

Benveniste's distinction between the two subjects and his example of the liar paradox were taken up by Lacan as support for his conten­tion that personal identity necessarily involves an element of mis-recognition. For Lacan, the subject of the enunciation and the subject of the enounced were disjunct not only in such special cases but were so in principle. Whenever the subject utters the word T and identifies with the T so represented there is always a discrepancy between the two, because what is and must be absent is that which alone could close the gap, namely, the unconscious.

The idea that the representation in discourse of the subject was thereby a misrepresentation was, according to the film theorists who adopted it, exactly applicable to what happened in certain forms of cinema. The argument ran as follows. Though all discourse has a subject of enunciation who produces meaning, this subject is not, as might be supposed, necessarily the author of the text. It is rather the individual who occupies the place of the subject of enunciation in what Benveniste terms 'the unceasing present of enunciation'.39 While this position may be occupied by the author writing or director directing the text, it is also occupied by each reader or spectator in making sense of the text. With certain texts the so-constituted subject of the enunciation misrecognises his or her situation because, though both text and positioning by it are produced, the text does not permit an awareness of such production on the part of the spectator. Unlike other texts, which avow their origins in contingency and acknowledge their viewpoint as perspectival, texts of this kind do not appear to have been produced by an act of enunciation at all. Rather, they present themselves as unenounced, as writing themselves, and so the spectator, unaware of his or her positioning, accepts their represen­tations as reality.

Such thinking issued in the crucial distinction between those texts that effaced the marks of their own production and of their construc­tion of spectators and those, usually self-reflexive avant-garde texts, in which this did not happen. The categorisation rested on two further opposed terms borrowed from Benveniste: histoire, a mode of enun­ciation where the pronouns 'I' and 'you' only occur in reported speech and all marks of subjectivity are suppressed; and discours, in which such shifters are present, and which exhibits 'the imprint of the pro­cess of enunciation in the utterance'.40 Impersonal and atemporal in address, histoire is the typical mode of enunciation of written history and of novels adopting this historical form, with events narrated in an indefinite past tense by an absent narrator. It would include a novelist such as Balzac, a passage from whom Benveniste cited as an example, in which 'no one speaks here; the events seem to narrate themselves'.41 Discours, on the other hand, is typified by conversation, letters, speeches and those forms of writing in which the narrator is in the foreground addressing the reader, such as the epistolary novel.

A large proportion of all narrative films falls within the category of histoire, and indeed it could be put forward as the defining feature of traditional cinema that it effaces all traces of its enunciation, pre­senting stories 'from nowhere, told by nobody, but received by some­one (without which it would not exist)'.42 What that someone cannot do is recognise that in occupying the position of invisible enunciator he or she is constituted by the film. Instead the spectator experiences him or herself as a pure subject, empty and absent, a pure capacity for seeing what appears to be simply there. All 'content^ is seemingly on the side of the film, elsewhere, anywhere but inscribed as subjec­tivity. Of course discours is not entirely absent from narrative cinema. It exists both as characters' dialogue and as characters' point of view shots, but such explicit instances of enunciation are contained within a supervening narrative that specifies who is speaking and looking, and also establishes that their words and vision are partial. The nar­rative itself, locating the enunciations of the characters, is always at pains to present itself not as discours but as histoire, unfolding before the spectator as completed, comprehended, resolved and impartial. The representations on screen appear not as one point of view but as reality itself. The narrative doses around a world as it would appear to the all-seeing eye of God - a privileged position that the spectatorial ‘I’ is only too pleased to occupy.

To sum up this first phase of thinking about the relation between text and subject, the central emphasis throughout was on the text's power to determine the subject's response. Watching a film necessar­ily entailed adopting the spectator position that was inscribed in the text; to comprehend was to be positioned. In the second phase a far more complex and sophisticated theory was elaborated, once again pioneered by Stephen Heath, who revised his earlier opinions as he came to appreciate the limitations of Althusser's theory of ideology.

In appropriating Lacan to help explain how ideology functions, Althusser had concentrated on the mirror phase, in which the child perceives and identifies with an idealised self-image, and from this had derived his notion of interpellation, in which the individual is called to an image of him or herself, is caught up in a structure of misrecognition and thereby becomes constituted as a subject within ideology. For Althusser, Heath wrote, 'the subject is thus the indi­vidual always held in the identity - the identification - of interpella­tion'.43 This entails, first of all, that people are fixed in positions of subjection through their self-conceptions operating within a system of representation; and secondly, that they are locked into illusion. Determined and deluded, they 'go all by themselves, like so many automata'.44

There were in this account, Heath argued, insurmountable prob­lems that went beyond those recognised by Hirst in his critique. Hirst's perception that interpellation involved a pre-given subject pointed to a fundamental incompatibility between Althusser's and Lacan's views on subjectivity. Because of Lacan's emphasis on the primacy of the signifier there can be no individual prior to the pro­cesses of language to become the support for the identification. But further, not only is the subject not pre-given, it is always necessarily in process, being constituted through the process that is language. These criticisms were given expression in an influential article in the Edinburgh Magazine of 1976, where Heath wrote:

There is no subject outside of a social formation, outside of social processes which include and define positions of meaning, which specify ideological places.

There is a concrete history of the construction of the individual as subject and that history is also the social construction of the subject; it is not in other words that there is first of all the construction of a subject for social / ideological formations and then the placing of that constructed subject-sup­port in those formations, it is mat the two processes are one in a kind of necessary simultaneity - like the recto and verso of a piece of paper. The individual is always entering, emerging, as subject in language.45

In underlining the importance of process Heath drew attention to two related aspects of the constitution of the subject. The first was that the subject is as much constituting as constituted because, according to Lacan, the subject halts the slide of the signifier, thereby becoming the producer as well as the product of meaning. There was therefore, Heath said, a 'dialectic of the subject',46 in which lan­guage is not the sole determination but is rather 'an area of determi­nations. . . the condition-and-effect of social practice'.47 Although the subject is always implicated in discursive practices, this is not to say that the effects of meaning and subjectivity are produced by the organisation of discourse alone. Signifying practices should rather be thought of as 'subject productions', a phrase which implies produc­tions both by and of the subject.48 Meaning and subjectivity come into being together, each engendering the other in a process of endless dialectic.

This complex idea of the dialectic of the subject, which will recur frequently in the course of this book, is perhaps best explained by reference to concrete examples. In each case there is an exchange between the subject and the other - whether that other be an aspect of the world, a person, or a film - involving an act of interpretation. For instance, different people confronted by the same stretch of open moorland will respond differently: for the walker it is an enticing prospect for vigorous exercise, for the agoraphobic a terror-inducing void. In each case the subject is constitutive in that the interpretation is not inherent in the object - the landowner interprets it differently again; but at the same time the subject is constituted by the object, in that, according to the interpretation, he or she is to a lesser or greater extent transformed by it. The subject is at once the producer and the product of the meaning. Two further examples would be the condition of the lover in the erotic relation and that of the analysand in the analytic situation. In the former instance, the lover constitutes the beloved as the idealised object of desire, while being constituted in the condition of infatuation by this interpretation of the independently existing object. Similarly, in the analytic situation, the (mistaken) presumption that the analyst is the one who knows the meaning of the analysand's discourse engenders a mode of sub­jectivity in which the analysand can discover something of what he or she wants. When applied to the relations between spectator and film this conception entails that both make their contribution, what Heath called 'give-and-take'.49 Without this, the subject can only be thought of as either inescapably determined by the text or as voluntaristically creating meanings (which, though politically more hopeful than immobility, leaves little or nothing to be achieved in the realm of the textual). In each case political intervention becomes a redun­dancy, in the one because meanings are unalterably fixed, in the other because they are already fluid. Instead the relation of subject and text is a movement of exchange: 'the subject makes the meanings the film makes for it, is the turn of the film as discourse'.50

The other aspect of the stress on process was that the subject's constitution within language is a moment of division. Entry into language is the condition for subjectivity and identity, but also for the unconscious, which functions to invalidate any attempt to capture the subject's reality. Not unity but division, not identity but non-iden­tity, these are the terms for the constitution of the subject. Nothing of this, however, pervades Althusser's account of subjectivity. For him the pre-given individual is constituted as a subject by a pre-given representation; the self-image presented through interpellation is the subject's identity. What vanishes in this is the distinction between subject and representation, for they are one and the same. Psychoanalysis, by contrast, insists on the interminable elusiveness of the subject, the impossibility of ever fully defining it, by virtue of the unbridgeable gulf between the subject of the enunciation, who speaks, and the subject of the enounced, who is spoken of. Thus the subject is never completely positioned or captured by its representa­tions. Although existing in society and in ideological formations, the subject always exceeds any representations of such formations, and the attempt to limit and unify it belongs to the fictive realm of the imaginary. Since the individual is not one but is process, heterogeneity, multiplicity and unfinished, there is no possibility of ideological apparatuses simply constructing subjectivity. Therefore the cinema as institution does not position but contains. Its attempt to achieve a complete closure that will hold the subject in position never abso­lutely succeeds. Its representations can never fully represent, its bids to fix in place must finally fail. So although a film may adopt and construct forms of interpellation the notion of interpellation is inadequate as an account of the relations of film and spectator. Instead, these should be thought of as 'signifying practice, as so many relations of subjectivity, relations which are not the simple "property" of the film nor that of the individual-spectator but which are those of a subject production in which film and individual have their specific historical and social reality as such'.51

Heath's intervention concluded with a call to shift the analysis away from text as system towards text as process, away from 'the object cinema' towards 'the operation cinema'. In its fixation on the formal attributes of texts, semiotics risked blocking the understanding of how cinema is related to other practices as well as the more general relations among signification, ideology and history. Semiotics was in danger of becoming an obstacle rather than the royal road to the analysis of the text's political functioning. To circumvent this danger Heath elaborated a conception of film (proposed originally by Jean-Louis Comolli) as a 'specific signifying practice'. Each term in the formulation was explicated as follows: 'Signifying indicates the re­cognition of film as system or series of systems of meaning, film as articulation'; 'specific is the necessity for analysis to understand film in the particularity of the work it engages, the differences it sustains with other signifying practices', which requires a semiotic analysis that attends to the heterogeneity present in particular textual sys­tems and to the range of codes at work.52 Practice, the crucial term, was conceived in Althusserian terms as the processes of transforma­tion of a determinate, given raw material into a determinate product. The term 'practice' entailed that film was not some neutral medium transmitting a pre-given ideology, but was the active production of meaning, with its own materiality and effectivity. As such, it broke with reductionist accounts that conceived films as mere reflections of pre-existing social forces or as having political effectivity only in so far as they communicated explicitly political messages. The term also carried the further implication that since film is the work of production of meanings, the question of the positioning of the spec­tator enters into the analysis of film. In this way the classical, and by now obscurantist, opposition between form and content could be bypassed in favour of the operations of film and the relations of subjectivity so entailed. As specific signifying practice film was to be studied in terms not of langue and parole but of discourse, thereby implicating a subject (to be theorised by psychoanalysis). Under such a conception cinema is one of a number of 'machines' generating ideology, specifically in that through its mechanism 'the spectator is moved, and related as subject in the process and images of that move­ment'.53 Or in the words of Teresa de Lauretis, cinema is 'a work of semiosis: a work that produces effects of meaning and perception, self-images and subject positions for all those involved, makers and viewers.'54 The work of Lacan was to figure centrally in thinking through the complexities of this second phase. Although Heath was not entirely uncritical of Lacan (his occlusion of history being one cause for concern), he was on the whole confident that Lacanian psychoanalysis could overcome any difficulties. Others, however, as we shall see, were somewhat less confident.

In the mid-1980s the semiotics concerned with formal systems (as opposed to semiotics relating to the subject) survives within film theory only in the form of specific detailed studies around narrative, point of view, editing and so on. Although such work is admirably rigorous it no longer occupies the position of centrality that semiotics briefly enjoyed in the aftermath of 1968, when its apparent scientific status threatened to displace all other filmic discourses and to become the necessary precondition for advance in any area of film study.

Of the various reasons for the eclipse of semiotics, the most import­ant for film theory was the failure to integrate it with historical materialism. Initially it had seemed a relatively straightforward matter to bring together Althusser's notion of the materiality of ideology with Saussure's of the materiality of signification. After all, each could only apparently gain from the other: historical materialism lacked an adequate theory of signification (this lack was especially felt with the growing emphasis on ideology and its existence as representations), and semiotics lacked the theoretical means of relating signification to its social context. Semiotics would fill the lack in historical materialism, and vice versa. Furthermore, the rigour of semiotics chimed nicely with the Althusserian promise of scientificity. The out­line of such an integrated theory would at once figure how a text produced an ideological representation of the world and how it con­stituted the individual as subject for that representation.

But as the enterprise proceeded it became evident that relating the signifying process to history and the social formation was more dif­ficult than had been anticipated. It became apparent that Saussure's theory of meaning was less of a foundation stone than a stumbling block, the problem being that the logic of the model was at odds with the intention behind its deployment. Instead of allowing the exchange between spectator and text to be related to history, the model had the effect of evacuating history from that exchange. The exit to social reality was blocked both via the text and via the subject: if meaning was produced by a system of differences then the process of signification became autonomous and therefore difficult to relate to other social practices; the spectator as constituted by the text was difficult to relate to the subject as constituted elsewhere. It was these issues that underlay two important debates in the late 1970s: between Ros Coward and the Centre for Contemporary Cultural Studies (CCCS); and between Paul Willemen and Colin MacCabe.55

The broad project of CCCS was not dissimilar to Screen's, in that they were using Marxist theories of ideology, pre-eminently Althusser’s theory of relative autonomy and ideological state apparatuses, to analyse various cultural phenomena such as the press and television. Coward's basic contention was that in studying representation they disregarded Saussure, in that they took representations to be both expressive of pre-given meanings and determined by the class interests of their source. What this omitted was firstly that signification pro­duces meaning rather than simply expresses pre-given meaning, and secondly that it has its own specificity and is irreducible to any other practice. Coward did not argue that signifying processes are fully autonomous, and indeed insisted that they are not. It was rather that although the means of representation has its conditions of existence, these are not expressed or represented by it. Institutions like cinema or television are clearly shaped by social forces, but because of the determining action of the means of representation such institutions do not necessarily represent the economic interests in which they are constructed. Nor is there any easy way to relate signification to such forces.

In reply, CCCS denied that their position was as reductionist as Coward had suggested, giving various reasons to show they respected the relative autonomy of signifying practice. Coward's position, they said, despite her gestural attempts to reconcile specific signifying prac­tice and the social formation, effectively gave complete autonomy to signification. The question then arose as to whether she was still working in any sense within a Marxist framework; for their part they would continue the immensely difficult task of relating culture to the class society in which it existed, a task that required the develop­ment of a general social theory based on the work of Gramsci and Althusser. Coward, for her part, remained unphased by the sugges­tion that she had ceased to be a Marxist since, like Hindess and Hirst, she saw Marxism as having unduly privileged certain concepts that blocked the way towards necessary political analyses, not least among them those pertaining to the oppression of women. Rather than con­tinue to rely on the incoherent concept of relative autonomy, the new politics was better served by thinking of the social formation in terms of conditions of existence rather than determination (especi­ally determination by the economic in the last instance) and by con­centrating on the specificity of signification. In women's struggles, for instance, a prime concern must be to combat and transform exist­ing systems of representation. The role of theory was therefore not to provide tantalisations of society but to intervene in particular strug­gles as a tactical instrument. As usual in such debates matters were not finally resolved, but of the two positions it was Coward's that was to be the more influential. For our purpose the importance of the debate was in the indication it gave of the unravelling of the theoretical fabric holding together Althusser, Saussure and Lacan, further evidence of which was apparent in the debate between Willemen and MacCabe.

The debate turned on the vexed question of subject position, a concept we have already seen developed through two phases, one based on interpellation, the other on the dialectic of the subject. Wil­lemen perceived that for all the sophistication of the second phase, there was a major problem with it in that if the subject was already in place, produced by other social practices and able to work on the text, it became impossible to specify a text's effectivity on the basis of its structure. Because of this problem there was a tendency to fall back on the earlier phase, whereby the text unilaterally determined the spectator's reading. Here, though, the spectator was a mere func­tion of the text, locked into position by the unalterable chain of signifiers. It was this formalism, with its implications of 'subjugation. . . not to say terrorism', that Willemen wanted to challenge.56 His immediate target was a paper by Edward Branigan contrasting two films by Fellini and Oshima. It exemplified that formalist criticism which identified and described the structural codes present in films, such as point of view, spatial organisation and editing, which were in themselves supposedly of determinate effect on spectators. Analyses of this kind, Willemen argued, became immanent, hypostatising the text and hence the reader, with both frozen into immobility. Their effect was to evacuate ideology, the social formation and history, and to block 'the theorisation of the construction of subjectivity in social practices'.57 Because the text inscribes a reader, irrespective of his or her social and historical placement, in an inevitable fixity, there is no room for the workings of ideology except through this monolithic determination. Formalism ignored the outside of the text, 'an outside consisting of discourses in struggle, discursive formations cohering into conjunctures of ideologies'.58 In other words, it ignored the multiplicity of social forces and practices at play, at work, in the reading of any text. Against formalism's hypostatisation, Willemen contended that texts were open to a multiplicity of readings. Meaning is a product not simply of the text, but of historically and socially constructed subjects engaging with the text. Whereas for the for­malists the text is a unified structure with determinate effects, for Willemen it is unstable, offering only provisional coherences that vary according to context and reader. Any code functions within an ideological configuration that not only gives it meaning but also specifies its place within ideological practice. Instead of being deter­mined once and for all by the codes of the text, the political effectivity of the text is a function of the mode of reading. Under the pressure of diverse and variable readings, texts may be transposed into more or less any ideological space, may be commandeered for even quite contradictory critical ends. The best any purely textual analysis can do is open up problematic areas of that ideological space by activating the repressions, contradictions and latencies within it. In making these claims for the real plurality of readers and readings Willemen's views coincided to a large extent with those of Umberto Eco mentioned earlier.

Like Eco, Willemen saw the formalist identification of the textually constructed reader with the real reader as being of no value in the analysis of the text's political functioning. No such analysis was pos­sible unless it took the historically formed subject into account as well. Not only are the textually constructed reader and the real reader radically divergent, but the constructed reader is itself a multiplicity formed through the various subject positions offered by the text -as too is the historically formed real subject. Thinking through the relations between them was, Willemen appreciated, a formidable task, but he was confident that relative autonomy provided the framework with which to do so. Ultimately the organisation of the ideological and discursive formations in which the subject was situated was a function of 'the real', and this in turn was to be identified with the relations of production. It is the place occupied within the relations of production that determines which institutions and discursive regimes will be encountered. Each individual reads texts in terms of his or her 'concrete experience', that is, in terms of the ideologies and discourses he or she has encountered.

Willemen was concerned that this traditional, even commonsensical, viewpoint should neither fall back into a mechanistic determinism nor detract from the effectivity and productivity of signifying practice. He insisted that both the subject and reality are constituted in discourse. The relations among the real (which is logically prior to discourse), reality (which exists for a subject), subject and discourse are to be thought of in terms of a dialectic, whereby the real deter­mines the encounter with discourses, which produce reality and affect the subject's passage through the real. Subjects work both on and in discourses: their positionality in the real and in reality must be distinguished yet 'thought together in a dialectical movement of mutual determination'.59

This analysis led him to the conclusion that ideological struggle would involve two simultaneous components. One, by challenging and displacing existing discourses, would alter the balance of forces within institutions; the other, in parallel, would replace the existing discourses with new ones, so providing alternative subject positions within ideology. Analysis and politics, theory and practice, were inextricably bound together.

Despite Willemen's insistence that he nowhere lost sight of the effectivity of signification or the dialectic of the subject, Colin MacCabe was unconvinced. For MacCabe detected what had troubled Coward in the work of CCCS, namely, the presumption that cinema stands in a representational relation to the ideological, the political and the economic. As a practice with its own specific effectivity, cinema can be neither the representation nor the expression of any­thing pre-given, but must be understood in terms of discourse and the production of subject positions. This does not lead back to for­malism, because textual structures do not of themselves determine readings but are, as MacCabe put it, always imbricated with the ideological. This adherence to the notion of the dialectic of the subject entailed that the text neither preceded ideology, nor the reverse; the imbrication of the two meant that reading is a function both of the text's formal organisation and of ideology, which itself can only exist within a discursive formation. Willemen, said MacCabe, failed to think of the relation of text and ideology in terms of imbrication, but in maintaining a general theory of discourse and politics held fast to a belief in the separate fields of the textual and the extra-textual. The one comprised the formal articulation of the text, its codes and structures, operating to produce effects in the reader; the other was an historically given ideological and political space outside of cinema determining how any film would be read. The separation of these two, according to MacCabe, resulted in a siphoning-off of determinacy from the textual to the extra-textual, ultimately allowing no determining reality for the discursive as against 'the real'. Rather than thinking in terms of inside (cinema) and outside (ideology), the con­cept of discourse enabled one 'to think the operations by which cinema is constantly transforming the outside inside, and that inside a further element in the outside'.60 Classic narrative cinema is perpetu­ally referring to an outside which is 'pulled into place, into space' in the film's address to the spectator, thereby becoming an inside that confirms the outside.61

As an illustration of this process MacCabe cited Nashville, a film that has as its theme the respective realities behind the country music scene and populist politicking. While the music of the performers, the speeches of the politicians and the diegetic television reportage of both tell little of, indeed conceal, what is really going on, the picture track renders this invisible visible for the spectator. Taken behind the rhetoric of pork-barrel politics, and the clich├ęs and stereotypes of the television reporter, the spectator is offered a com­prehensive and omniscient vision, a position of knowledge. Now although the address to the spectator comes from the articulation of shot and narrative, it is not solely the work of the codes scrutinised by formalist analyses, but rather relies on an audience already sub­scribing to a belief in the 'falsity and idiocy of Middle America'.62 Nashville thus pulls its public 'into the place that it already occupies'.63 The spectator is shown what she or he already believes, namely that all political activity is a futile charade incapable of changing social reality, and therefore succumbs to the political apathy that the film promotes as its ideological project. Truth and transformation are re­presented as mutually exclusive; showing it like it is to show that nothing can be done about it; knowledge is impotence.

MacCabe's article bears testimony to the increasing and finally insoluble difficulties facing the post-1968 project of determining the political effectivity of the text. In proposing such effectivity as a func­tion, firstly, of subject position, and secondly and more sophisticatedly, of the constituted-constituting subject, the model had con­fronted many problems (some of which we have already discussed, others to be taken up later), but the one on which the enterprise foundered was that to which it had returned time and again, namely the subject. If the dialectic of the subject was a reality, if the subject was as much producer as produced, then the political effectivity of the text could not be determined by an analysis of the text alone. That is, immanent analyses of its structures, organisation, strategies and codes would not permit the final determination of the relations of subjectivity it would constitute. The exchange, therefore, between text and society, between discourse and history, became so complex as to preclude anything beyond the most provisional and gestural of generalities.

In coming to appreciate these difficulties MacCabe, in his most recent writings (Theoretical Essays), has in certain respects moved closer to Willermen's position, but more importantly towards one that could be called postmodernist (a topic we shall return to in chap­ter 7). With the possibility of the single revelatory analysis of the text foreclosed, there can only be readings for readers, different models of its modus operandi for different theoretical approaches adopted. Both text and reader take their form from 'a dizzyingly vast series of determinations'.64 There is no longer any question of ascer­taining the unique meaning of the text, any more than there is of establishing the single subject position constituted by it or the sole political effect of its textual operations. The spectre of endless differ­ence [can] not be dispelled theoretically (for every reader a different meaning, for every reading a different meaning).'65 In which case, it may be wondered, is all possibility of analysis thereby precluded? To this MacCabe answered no: though analysis could no longer claim to be final or unique, there was a more limited, provisional and mod­est option based on what he has called shared questions. In practice, MacCabe writes, we share questions about sexuality and power, edu­cation and the national language. In our professional, academic or other shared practices, such questions find their validity, as do, by implication, their answers.

On the strength of the debates outlined above, it became evident that the semiotic enterprise as originally conceived was in need of re-evaluation. While still available for admirably rigorous close textual reading, it could no longer be seen as the way forward in the under­standing of the political functioning of cinema. Developments within post-structuralism powerfully reinforced this trend. We shall here limit our elucidation to certain pertinent aspects of the work of Jacques Derrida, leaving to one side both the bulk of his critique of Western metaphysics, and other post-structuralists (notably Foucault and Lyotard, who are treated elsewhere in this book).

Our starting point will be Derrida's when he writes: 'This is my starting point: no meaning can be determined out of context, but no context permits saturation.'66 It is fundamental to any signifier, word or text that it can be repeated; indeed signification precisely depends on this condition of repeatability, or iterability. But what determines the meaning is the context in which the signifier, word or text occurs; and since this is variable and boundless, the meaning of successive iterations are equally so. Just as no one person or institu­tion can finally control the contexts in which a text will be situated, no one person or institution can specify the limits of meaning accruing to the text. Readers constantly relate any given text to others, so producing new meanings, new interpretations. The possibility of' the text overrun[ning] all the limits assigned to it' entails that meaning is always potentially both different and deferred (Derrida's term differance has this dual sense as well as designating that spatio-temporal difference is the condition of meaning).67 The absence of any fixed or final meaning, the constant entering into new textual relationships, which Derrida refers to as 'nonmasterable dissemination', is what deci­sively distinguishes Derrida's account of language from that of Saus-sure.68 In effect, Saussure had been blind to the radical implications of his conception of the sign, appreciating that signification is an effect of a system of differences but failing to conclude (in large measure because of his privileging of speech over writing, with the implication of a meaning-intending presence behind speech) that there could therefore be only endless difference.

Derrida also took issue with Lacan, who, while in principle recog­nising the implications of the primacy of the signifier, did not (said Derrida) do so in practice. The focus of Derrida's disagreement was Lacan's reading of a short story by Edgar Allan Poe, The Purloined Letter'. In the story, a letter (whose content is never revealed) has been sent to the Queen, who, in order that it may not be discovered by the King, hides it by pretending there is nothing to hide, laying it openly on the table when the King enters while she is reading it. The King sees nothing, but the Minister realises what is going on and steals it by gathering it up with his other papers, thereby gaining power over the Queen. The Minister in turn hides it by placing it where it can be seen in a card rack on his mantelpiece, a ploy that fools the secret police the Queen has ordered to retrieve it, but not, subsequently, the detective Dupin. Correctly identifying the crum­pled piece of paper as the purloined letter, Dupin returns the follow­ing day, distracts the Minister's attention and takes the letter, sub­stituting for it a facsimile on which he has inscribed a pointed quo­tation, thereby settling an old score with the Minister. For Lacan, the story functions almost as an allegory for a number of psychoanaly­tic truths, summed up by the formulation 'a letter always arrives at its destination'.69 While Lacan meant several things by this, among them that the subject is caught up in the compulsion to repeat, that the unconscious is never silent, and that the role of the analyst is to ensure that the messages sent by the analysand are delivered to their true addressee, for our purposes the important theme is the one dis­cussed in this chapter, that of the role of the symbolic in the constitu­tion of the subject. In the same way that the subject is not master of the signifier but is rather subjected by it, so are the characters in the story denned by their relation to the letter, changing as this rela­tion changes. Thus in respect of the two triadic scenes of King, Queen, Minister and police, Minister, Dupin, the positions instituted by the letter are the same but are occupied by different characters - in the first scene the Minister is the robber, the Queen is the dispossessed, and the King notices nothing; in the second these positions are taken up by Dupin, the Minister and the police respectively. As their posi­tions change, so do not only their actions but also their characteristics - when Dupin occupies the third place he becomes the aggressor, while the hitherto aggressive Minister takes on the feminised role previously enacted by the Queen. In short, then, the tale illustrates the centrality of the signifying chain in the constitution of subjectivity. It is, of course, this notion on which so much of 1970s film theory turned.

Although Derrida did not directly challenge Lacan's psychoanalytic theses, he did question this reading of Poe on the grounds that Lacan found in the text only what he wanted to find there. By treating the text as a mere representation of pre-given psychoanalytic truths Lacan was violating his own Saussurean precept that the signified continually slides beneath the signifier, a point that he made in the course of the analysis itself. Instead, however, of following the logic of his position and allowing that it is impossible to pronounce the final word on a text, Lacan did the reverse, deciphering the hidden mean­ing of the story. At one point in the analysis, for example, Lacan likened the letter to the phallus; but this, claimed Derrida, was a classic piece of Freudian reductionism, in which the phallus is accorded the status of a transcendental signified, a meaning given prior to signification. Rather than acknowledging the dissemination of meaning, Lacan sought to master the text using the phallus as the key. In so doing he failed to perceive the possibility that 'a letter can always not arrive at its destination' - though equally, of course, it may arrive.70 Meanings, in other words, can never be safeguarded against the vagaries of interpretation; they can always go astray.

Within film theory Derrida is perhaps best conceived of as a struc­turing absence. Although initially cited by the post-1968 theorists as support for the materiality of language, references to his work became subsequently less frequent. In the structuralist phase there was obviously a serious incompatibility between the requirement that texts determine readers' responses (through interpellation) and Derrida's project. Once it is allowed that the human subject has a degree of agency in the reading of a text (and in the last chapter we showed that it must be), then the notion of dissemination renders any posi­tioning by the text untenable. For if interpellation can only operate through an act of interpretation on the part of the reader, then there is no guarantee that a text will always interpellate in the same way. After Derrida there can be no question of specifying the text's effectivity independently of the context of reception: readings differ; a letter does not always reach its destination. Structuralists, therefore, could only lose by co-opting him. For post-structuralists, on the other hand, committed as they were to a plurality of readings of texts, the grounds for maintaining a distance were less evident, but may be traced to a general preference for a Lacanian perspective.

Even when it entered its post-structuralist phase film theory was concerned above all else with the text's political effectivity, and was therefore disinclined to indulge in the flamboyant propagation of meanings associated with many Derrideans engaging with literature. For film theorists the important thing was that film, as Frank Lentricchia said of literature, 'makes something happen'.71 While accept­ing that there is no final meaning to the text, no limiting its meaning to the demands of any one institutional framework (of relevance, among other things, in challenging the established protocols of teach­ing texts), they were insistent that in any given, historically defined instance the text is read in a particular way and that this has an effect upon the reader. Just as the open stretch of moorland that induces terror in the agoraphobic has a very different meaning for and effect on the rambler, the landowner or the birdwatcher, so too with any text: openness of meaning, yet determinacy of effect. Derrida's con­cern was not with how meanings are produced, but rather with how they can miscarry; for film theorists, however, the dominant concern was to study the determinations at work in particular historical moments giving rise to particular meanings, and they needed there­fore a model of signification that would take account of what Derrida termed 'external constraints'. This need was seemingly met by Lacan's conception of the dialectic of the subject: that in the act of enunciation subject and meaning come into being together.

To recall Lacan's graph reproduced earlier in this chapter, the sig­nified slides beneath the signifying chain, which the subject punctuates retroactively to produce meaning. Thus, in so far as film theory was concerned with accounting for the effectivity of significa­tion, it tended to look more to Lacan than Derrida. However, as we shall show in the following chapter, there were serious problems attaching to Lacan.

2 Semiotics

1 These phrases appeared in the polemics of Kevin Brownlow, John Col-eman, Nigel Andrews, and Clive James.

2 Ferdinand de Saussure, Course in General Linguistics, trans. Wade Baskin (Glasgow: Fontana/Collins, 1974), p. 16.

3 Colin McArthur, 'Analysing cinematic sign language', inDialectic! (Lon­don: Key Texts, 1982), p. 38.

4 Christian Metz, quoted by Jim Hillier, Movie 20, spring 1975, p. 25.

5 Geoffrey Sampson, Schools ofLinguistics: Competition and Evolution (Lon­don: Hutchinson, 1980), p. 48.

6 Saussure, Course in General Linguistics, p. 120.

7 Kaja Silverman, The Subject of Semiotics (New York: Oxford University Press, 1983), pp. 16-17.

8 C. S. Peirce, quoted in Silverman, The Subject of Semiotics, p. 18.

9 C. S. Peirce, quoted in Robert Almeder, The Philosophy of Charles S. Peirce: A Critical Introduction (Oxford: Basil Blackwell, 1980), p. 24.

10 Almeder, The Philosophy of Charles S. Peirce, p. 25.

11 Ibid., p. 25.

12 Silverman, The Subject of Semiotics, p. 22.

13 Doane, Mellencamp and Williams, Re-vision, p. 6.

14 Gilles Deleuze, Cinema 1: The Movement-image, trans. Hugh Tomlin-son and Barbara Habberjam (London: Althone Press, 1986); de Lauretis, Alice Doesn't.

15 Jacques Lacan, Ecrits:A Selection, trans. Alan Sheridan (London: Tavis-tock Publications, 1977), p. 65.

16 Jacques Lacan, quoted in Juliet Flower MacCannell, figuring Lacan:

Criticism and the Cultural Unconscious (London: Croom Helm, 1986), p. 46.

17 The graph is reproduced from Lacan, Ecrits, p. 303.

18 Christian Metz, quoted in Stephen Heath, 'Film/cinetext/texf>, Screen Reader 2 (London: Society for Education in Film and Television, 1981), p. 104.

19 Christian Metz, quoted in Heath, 'Film/cinetext/text', p. 106.

20 Christian Metz, Film Language (New York: Oxford University Press, 1974), p. 116.

21 Ibid., p. 105.

22 See Metz, Film Language, chapter 5.

23 The analysis can be found in Metz, Film Language.

24 David Bordwell, Narration in the Fiction Film (London: Methuen, 1985), p. xii.

25 See Christian Metz, Language and Cinema, trans. D. J. Umiker-Sebeok (The Hague: Mouton, 1974).

26 Oswald Stack, Pasolini on Pasolini (London: Thames and Hudson in association with the British Film Institute, 1969), p. 29.

27 Pasolini, quoted in de Lauretis, Alice Doesn't, p. 42.

28 Umberto Eco, A Theory of Semiotics (London: Macmillan, 1977), p. 200.

29 Ibid., p. 3.

30 Umberto Eco, 'Articulations of the cinematic code', in Movies and Methods: An Anthology, ed. Bill Nichols (Berkeley: University of California Press, 1976), p. 604.

31 Eco, A Theory of Semiotics, p. 49.

32 Ibid., p. 235.'

33 de Lauretis, Alice Doesn't, p. 47.

34 Eco, A Theory of Semiotics, p. 49.

35 Heath, 'Film/cinetext/text', p. 100.

36 Ibid., p. 103.

37 See Julia Kristeva, The system and the speaking subject', The Times Literary Supplement, 12 October 1973, p. 1249.

38 Rosalind Coward and John Ellis, Language and Materialism: Develop­ments in Semiology and the Theory of the Signifier (London: Routledge and Kegan Paul, 1975), p. 68.

39 Emile Benveniste, quoted in Antony Easthope, Poetry as Discourse (London: Methuen, 1983), p. 43.

40 Emile Benveniste, ibid., p. 41.

41 Emile Benveniste, ibid., p. 43.

42 Christian Metz, 'History/discourse: note on two voyeurisms', trans. Susan Bennett, Edinburgh Magazine 1976, p. 24.

43 Stephen Heath, The turn of the subject", Cine-tracts 7/8, summer/fall 1979, p. 33.

44 Ibid., p. 43.

45 Stephen Heath, 'Screen images, film memory', Edinburgh Magazine, 1976 p. 40.

46 Stephen Heath, 'Anato Mo', Screen 16, 4, winter 1975/76 p 50

47 Heath, The turn of the subject1, p. 41.

48 Ibid., p. 43.

49 Ibid., p. 43.

50 Stephen Heath, Questions of Cinema (London: Macmillan 1981} d 88. ' V'

51 Heath, 'The turn of the subject1, p. 44.

52 Stephen Heath, "'Jaws", ideology and film theory5 Film Reader 2 1977, p. 167.

53 Heath, Questions of Cinema, p. 62.

54 de Lauretis, Alice Doesn't, p. 37.

55 The principle articles in the former debate were Rosalind Coward, 'Class, "culture" and the social formation', Screen 18, 1, spring 1977; lain Chambers et al., 'Marxism and culture', Screen 18, 4, winter 1977/78; Rosalind Coward, 'Response', Screen 18, 4.

56 Paul Willemen, 'Notes on subjectivity - on reading "Subjectivity under siege5", Screen 19, 1, spring 1978, p. 45.

57 Ibid., p. 44.

58 Ibid., p. 43.

59 Ibid., p. 69.

60 Colin MacCabe, The discursive and the ideological in film - notes on the conditions of political intervention', Screen 19, 4, winter 1978/79, p. 35.

61 Ibid., p. 35.

62 Ibid., p. 38.

63 Ibid., p. 38.

64 Colin MacCabe, Theoretical Essays (Manchester: Manchester University Press, 1985), p. 24.

65 Ibid., p. 24.

66 Jacques Derrida, 'Living on: borderlines', in Harold Bloom, Paul de Man, Jacques Derrida, Geoffrey H. Hartman and J. Hillis Miller, Deamstruc-tion and Criticism (London: Routledge and Kegan Paul, 1979), p. 81.

67 Ibid., p. 84.

68 Jacques Derrida, 'White mythology', in Margins of Philosophy, trans. Alan Bass (Brighton: Harvester, 1982), p. 248.

69 Jacques Lacan, 'Seminar on The Purloined Letter"', Tale French Studies 48, 1972, p. 72.

70 Jacques Derrida, The purveyor of truth', Tale French Studies 52, 1975, p. 65.

71 Lentricchia, Criticism and Social Change, p. 105.