Moving on from Metz

by Geoffrey Nowell-Smith

from Jump Cut, no. 12/13, 1976, pp. 39-41.
copyright Jump Cut: A Review of Contemporary Media, 1976, 2004.

Christian Metz, Language and Cinema, tr. Donna Jean Umiker-Sebeok ( The Hague: Mouton, 1974). About $25 hardcover.

Christian Metz’s Language and Cinema was written between 1968 and 1970 and published in 1971. It appeared (or snuck out) in English translation in 1974 under the Mouton imprint, a couple of months after Oxford University Press had brought out another volume of Metz under the title Film Language: A Semiotics of the Cinema. This so-called Film Language is in fact a translation (and not a very good one) of Metz’s earliest essays on film semiotics, written between 1964 and 1967 and first published in French, as Essais sur la signification au cinema, in 1968. Most of Metz’s critics would appear to have only read the Film Language volume. And this state of affairs is likely to persist, if only because Language and Cinema is hard to find in bookstores around the place, and sadly expensive if and when you do find it. This is really unfortunate because Language and Cinema consists of a systematic revision, and in some cases repudiation, of the ideas expounded in fragmentary form in the earlier essays. (Language and Cinema also has the subsidiary advantage of a translator who knows French, even if she is a bit shaky on film culture and doesn't know that the title of Murnau’s most famous film is SUNRISE, not “Dawn.” )

Essentially the differences between the two Metz volumes are of two types. On the one hand, there are significant differences of emphasis and differences in the concepts employed as the author shifts from a realist problematic to a properly semiotic one. On the other hand, there are differences attributable to the writing itself. The early essais were all “learned articles” (sometimes deficient in their learning, as for example in the treatment of Eisenstein in the earliest essay of all, “Cinema, ‘langue’ or ‘langage’” ).

Language and Cinema on the other hand is a book which is written in a particular way, which some people will find infuriatingly arid and pedantic, but which in fact accurately reflects in its form, the content of the concepts elicited in the course of analysis. In the early essays Metz took for granted a lot of commonplaces about the cinema, deriving notably from the Bazinian tradition, such as the notion of a possible unmediated representation of “reality.” In the essays the notion of reality is taken as unproblematic. Here, the question is one of finding the devices (hopefully not too many) through which that reality is signified and communicated in the cinema. In Language and Cinema, however, there is no such stable implicit starting point. The book revolves entirely around the concept of signification itself, with no assurance that what is being signified is ever in any simple sense “the real.” The processes of signification in the cinema are in fact, as Metz himself had already discovered, extremely complex. The task of charting them is not made any easier if—for reasons of intellectual (and political) honesty—one deprives oneself of the support provided by a realist (bourgeois) aesthetic.

The absence of a commonsense (in fact metaphysical) appeal to reality would not matter so much in a work of linguistics. Linguistics has managed for half a century to pursue its course, and to make scientific advances, without the aid of speculative definitions of reality. It has preferred to base itself instead on a rigorous definition of its own proper object, which is language. But Language and Cinema does not have language for an object. Although the title of the book may suggest to the unwary that what it does is to take two objects, language on the one hand and cinema on the other, and then compare them, its actual procedure is quite other. For a start, you cannot just “take” cinema and compare it with language or anything else without first knowing what exactly it is in order for it to be so compared. The problem is that no one knows satisfactorily what cinema is, in this sense and for this purpose.

The work that linguists did on language fifty years ago in order to produce a viable scientific concept has to be redone for the cinema. But the work of definition cannot proceed in a vacuum. It is simply not possible to produce, out of one’s head, and without reference to surrounding conditions, a working definition of cinema (film/ movies) in its totality. The only possible mode of procedure is dialectical—to work on the concept of cinema (and, parallel, on that of language) in order to elicit from an immense array of disparate facts and impressions those aspects of cinematic and filmic practice in which the signifying function of film/ cinema is expressed. (1) This has to be done in the light of linguistics and the operations that linguistics has done on the speech/ language area, but without subordinating oneself to any crude aprioristic analogy. It is not an easy task and the exposition of how it is done does not make for easy reading. But it is a task that had to be done, and shown to be done. Which is what Metz does.

The aspect of cinema (or film) in which Metz is interested is that through which it signifies. He is not interested in the same way in its economic mode of operation, the ideologies the majority of films embody or its psychodynamics (until very recently—see below). There is no doubt that films, besides being commodities, invoking the unconscious, etc., are bearers of signification. In an elementary sense, they obviously transmit meanings to the people who go and watch them. The problem is primarily one of delimiting the area over which the process of signification (production and exchange of meaningful signs) operates.

Secondarily there is a problem of delimiting the signification that is specifically filmic or cinematic from that which is originally external to it but is incorporated into the meaning production of actual films. Structural linguistics operates with the distinction between langue and parole in which langue is a self-contained and to some extent abstract system through which one can give a rigorous account of some of the aspects of concrete speech acts manifested at the level of parole. No such distinction is operative with movies. Individual movies (or film texts) may be like acts of parole. But there is no single or even dominant signifying system to which they can be referred in the last instance in the same way as a novel like Wuthering Heights can be referred back to the language (English) that it’s written in. (It should also be noted that films are never strictly speaking and in the full sense parole, but are always writings. Even when people speak in movies that speech is always “inscribed” on soundtrack.)

But there does exist a socially instituted form which helps to designate the messages of films as film messages and not any other kind of message. This form, which is present as a social fact, is what Metz calls cinema. (2) Thus the various acts of filmic parole come into their own in the context of the fact of cinema. Or rather than strictly parole, or speech, it is writing, spread over a wide area of possible practices (“film”). Film, therefore, is not a language, and neither is cinema. But the possibility for spectators of grasping the messages contained in film texts is organized around the social fact that cinema is instituted, and that a variety of signifying systems and subsystems are, as it were, clustered around the central cinematic fact.

Now the most striking feature about this way of looking at things is that it is absolutely non-essentialist. There is no need to assume any “essence” of cinema or film, let alone any “human essence” of which cinema or film would be manifestations. On the contrary we are dealing solely with determinate practices at a given moment in history. And insofar as there exist specifically filmic or cinematic practices, their specificity is guaranteed not in terms of any essence of the medium but in terms of an interplay between signifying practices and the material apparatuses through which they are inscribed.

This said, it is still the case that in Language and Cinema, Metz remains preoccupied with that part of the cinematic fact that cannot properly be described as linguistic, or, as he himself puts it elsewhere, the “metaphor” of cinematic “language.” It is only very recently, in his forays into psychoanalysis (see “The Imaginary Signifier,” Screen 16; 2, Summer 1975), that he has concerned himself in any detail with the other constituent apparatuses of cinema. But right from the beginning his interest in “cinematic language” (langage, in the generic sense) has been predicated on the conviction that cinema cannot be seen as a language (the French langue) or as analogous to language (langue) in the sense of a single dominant, self-contained, and systematic communicative code. His early formulation of the question in 1964 was not satisfactory. It was among other things still essentialist, postulating an essence of language on the one side and an essential reality on the other. But the basic fact—cinema is not langue and therefore the key to its semiotics must be found elsewhere—has remained a guiding thread throughout his work, and receives systematic treatment in Language and Cinema.

When I say that the question receives systematic treatment, I do not mean that cinema as a whole is treated by Metz as if it were a system, or even that its signifying aspect (or semiotic apparatus) can be conceived on the model of the system. That would be incompatible with the whole thrust of his analytic method. What I mean, quite literally, is that the treatment is systematic. It is, if you like, a systematic treatment of an object which is not a system (or not a single system or even a systematic set of interlocking sub-systems). Rather, cinema tends, by virtue of its status in the world, to defy systematization. It is true that Metz uses the word system quite a lot, and he even makes nods in the direction of systems theory. But his use of the word system (except in the phrase “language system,” where it is introduced by the translator as a not very accurate rendering of the French langage) never attributes an overall systematicity to cinema as such. Systematicity is a property of the enquiry conducted into the object by the author, and to some extent it is the property of the concepts used to situate the object. It is not, because it cannot be, a property of the object itself.

A systematic enquiry then, but into an object that is not itself a system. But if the object is not a system (that is to say internally organized in a systemic fashion) what else is it? The key to the Metzian method lies in its replacement of the concept of a unitary langue by that of a plurality of codes and the simultaneous preservation of the distinction between langue and parole, now displaced as a relation between code and text. Films are concrete objects whose nature is to be not just strips of celluloid but texts. These texts are readable because they contain the messages of various codes or subcodes.

But the structure of the text is not the same as the structure of the codes which underlie the messages. This is because the codes are plural while the texts, although they embody messages from a plurality of codes, are themselves singular objects. Each text or group of texts is constituted on the basis of codes, but it also constitutes its own system—or, indeed, several systems. Strictly speaking it is wrong to say that the codes are at the basis. Although the codes may pre-exist particular films-that-have-been, every new film both confirms existing codes (constructed on the evidence of previous films) and displaces or deconstructs them. There is a constant interplay between the codes, leading to their perpetual deconstruction and reconstruction through the action of various texts which combine them in new ways. Furthermore, the codes brought into play are not all specifically cinematic. Thus, the boundaries of the filmic area are in constant flux, and infinitely subvertible.

This analysis into groups of terms—cinema and film, code and message, system and text—is necessitated (and I use the word advisedly) by the complex properties of the object of the analysis. It remains now to ask three more questions. What methodological principles at the back of this analysis derive from linguistics? What is the purpose of the analysis? And, finally, what if anything is wrong with it (preferably in relation to the other two questions)?

The linguistic derivation of many of Metz’s concepts is unquestionable. Not only, as we have seen, are the couplets code and message and system and text elaborations of the linguists’ distinction between langue and parole, but underlying practically every item in Metz’s construction is the basic Saussurean notion of the (linguistic) sign. As is well known, Saussure saw the fact of the sign (or the embodiment of the signifying relation) as the irreducible heart of language. The essence of the sign is that it is always two-faced. On the one side is the signifier—that whereby the sign signifies. And on the other side is the signified—that whereof it signifies. This concept was extended by the Danish linguist Louis Hjelmslev into the distinction between the forms of expression on the one hand and those of content on the other. Such a distinction permits a realization that the signifying relation, as a fact of language, is not necessarily embodied in each and every individual sign as such, but it is a function of language as a whole. Metz’s most radical step in relation to linguistics consists in splitting the signifying relation further open still.

Since cinema is not language but is, as already mentioned, pluri-codic, there is no reason to prohibit cinema from containing entire codes which, although they contribute to the ability of the cinema to signify, are themselves situated exclusively on one side only. Thus light, and therefore lighting, natural or artificial, are necessary to the form of expression of the cinema. Different forms of lighting are in coded (paradigmatic or syntagmatic) relation to each other. But this does not mean that for every lighting effect (signifier) there is a thing meant by that effect (or signified). On the contrary, different forms of lighting only mean different things when taken in conjunction with a homologous organization of the elements of other codes situated on the other side. Thus the high key lighting of film noir takes its value not from an intrinsic equivalence between the lighting and “its” content, but from a process whereby an original potential for meaning (dark shadows = mysterious, etc.) becomes encoded and part of the languages of cinema.

With this example in mind we can now turn to the question of the rationale of Metz’s approach. By applying some of the concepts of linguistics to an object that is not in itself a language, he is able systematically to describe certain processes which up to now have been subject only to ad hoc empirical verification. By modifying the concepts in the light of the new object’s properties, he has on the one hand avoided the fallacies of the old “film grammarians.” But on the other hand, he has been forced into a situation where a rigorous method confronts an object which constantly refuses to be stably systematized.

This problematic has two major corollaries. The first is that the method used by Metz is absolutely not a method for film interpretation. Metz’s semiotics is not concerned with how to understand, and therefore to interpret, either films in particular or cinema in general. On the contrary, understanding is taken for granted and, as he himself puts it, the problem is not to understand films but to understand how it comes about that films are understood. And films undoubtedly are understood—by the general public as much as by critics and semioticians—although understandings are apt to vary (not surprisingly) from spectator to spectator.

Film language is not something that is ever going to be perfectly possessed by anyone. At most, some people are going to be in possession of the knowledge (and use) of more or different codes than others. Film semiotics therefore comes to concern itself with the conditions that permit the establishment of different readings. But none of these readings can aspire to the dubious status of a once-and-for-all interpretation. The most semiotics can do is say what validates one reading over another and which possibilities of differential readings are generated within the field of the codes—and which have to be referred outwards—to the social situation and political stance of the readers themselves.

The second corollary concerns the nature of the concept of production. The semiotic approach practiced by Metz says very little about production at either the economic or the technical level. And what it does say, it says from the standpoint of the signification present to the viewer. Personally I think this is a correct line to take and that it is very important clearly to demarcate, through a choice of standpoint, that which exists at the level of the sign and that which exists at the level of productive structures and processes.

But in another sense Metz is talking about production, because he is talking about the way signs are produced, not by the filmmaker but by the spectator. The signification present to the viewer is not something inert which has been already produced and is standing there waiting to be appropriated. It is the product of the reading of the film text as signs. Signs do not have intrinsic values. In addition, there is no writing that does not to some extent propose a reading and no reading which is not itself to some degree writing. The process of deconstruction and reconstruction that is continually taking place within the codes is inseparable from an activity of deconstruction and reconstruction present in the relation of the spectator to the text. The filmmaker may produce movies and throw them onto the market, like oven ready chickens, with everything present for the act of consumption of the sign value of the movie to take place. But in the last analysis it remains the spectator who determines the role of the product as sign.

However, this notion of the production of sign values reduces itself to a tautology the moment it is framed in economic terms. It is normally the case with any commodity that it is produced as a commodity (i.e., as exchange value) and that it becomes a use value only through consumption. The food industry doesn't produce food as such. It produces products which have food value when eaten. The makers have to assume that the product will have a use value, otherwise they stand to lose money on making it. But the use value relation is, so to speak, suspended, until it is realized in consumption. It is therefore true, but irrelevant to the purposes of political economy, to point out that the production of a chicken (say) as food takes place in the kitchen, the mouth and the stomach (as productive consumption, in Marx’s term), rather than in the egg, on the factory-farm or in the slaughterhouse.

The political economy of signs is, of course, not quite the same as that of non-signifying commodities. But there is a clear sense in which the relations that Metz and other semioticians are talking about have to be distinguished from those of production proper, which is prior, both logically and chronologically, to the “work of the text,” which takes place within the sphere of circulation. It is an essential feature of film that it does have such a thing as production. At is this fact of production, distinct from the circulation of signs, which marks it off most clearly from language.

The science of linguistics basically does not deal with the production of signs but only with their exchange. Certain sub-branches of the science do of course deal with the material conditions under which language speakers generate (or fail to generate) the signs of a language. But most studies of language, even of its pathology, are predicated on the notion of language itself as a system of exchange. Use of language (via motor acts, cognitive processes, etc.) refers to participation in this system of exchange. Signs are “produced,” by individuals, in order to take their place within this system. The status of linguistics as positive science cannot be divorced from its success in conceptualizing its object as a set of pure exchange relations. Language is a patrimony of exchangeable goods. In it, we all have an equal birthright even if, mysteriously, some of us get more mileage out of the system than others.

There are features of language that favor its analysis as an exchange system—the fact that hearers are also speakers and vice versa, the fact of reciprocity in exchanges, the fact that exchanges can he more or less instantaneous, the fact that language itself is relatively ahistorical (it’s always been there, and it changes very slowly), the fact that languages are convertible, like coinages. All these aspects of language are all absent from cinema. Cinema is deferred: there is a time lag between production and consumption. There is no reciprocity: when you are watching a movie you aren't making one. Although filmmakers (producers) can also be spectators (consumers), most consumers are not producers (even with super-8). And cinema has a history: there were no movies before 1895.

The emergence of sign values in the cinema, therefore, takes place against the background of a productive apparatus that resembles that of language about as much as modern capitalism resembles a hunting-and-gathering economy. Now a society that lives by hunting and gathering does practice exchanges other than economic ones, and so does a social formation in which the capitalist mode of production is dominant. The facts of a signifying system cannot be totally subsumed under the facts of the mode of production. But if language escapes the impact of capitalism only lightly, cinema does not. For the signs that the cinema produces (as opposed to those signifying systems into which it inserts itself) are not the coinage of individual free exchange. They bear the stamp of a mode of production in which free exchange exists only as a residual but tenacious myth.

It is not a question here of saying that every sign must be singly analyzed for the traces that appear in it of the mode of its production, or of claiming that where and how a film was made—in a socialist or capitalist country, in industrial or artisanal conditions—totally determines the value of the sign. The assertion I would like to make is a more complex and at the same time a more modest one. It is that signifying systems have a real, but only relative autonomy vis-à-vis other levels of a social formation, and the mode of production in particular. And one feature of this “only relative” autonomy is that the social formation tends to throw up as one of its effects a notion that autonomy can be total. Linguistics is an ideology precisely in its very pretension to be science.

Linguistics refuses to inscribe into its own operations an account of how it produces itself as science (and I mean this in a strong sense: it does not “fail,” it actively refuses). This refusal is in return motivated by a fetishization of exchange relations apart from production. What is at stake, therefore, is not the character of the sign within exchange (which may or may not be what the linguists take it to be) but the question of the deformation produced by an ideology which insists on seeing signs only as they exist in exchange. To my knowledge no form of linguistics yet exists which fully takes into account the fact of production either in the sense of linguistic signs being produced as well as exchanged or in the sense of the relations of production governing the way linguistics itself can conceive (“produce”) its own object.

The development of a semiotics distinct from linguistics (i.e. not merely as an extension of classical linguistics to cover areas other than language proper) has to some extent led to a confrontation with this difficulty. In the work of Julia Kristeva and some of the Tel Quel writers, for example, there is evidence of a constant reading in of the facts of production into the exchange of signs.

But, unless I am much mistaken, this reading in takes place within the context of a complex play of metaphor which never really clarifies the relation between two possible senses of production. It does not consider both the production of signs as analogous to commodity production (with further analogies to do with “surplus” production, etc.) and the production of signs within commodity production, i.e., the social relations immanent in a certain mode of production of sign values. Failure to confront this ambiguity (apart from providing political cover, in the Tel Quel case, for petty bourgeois anarchist subjectivism and for a kind of intellectual terrorism of the displaced sign) leads paradoxically to a reading of sign production within exchange rather than (as no doubt originally intended) a reading of exchange (of signs or anything else) within overall relations of production.

For the future development of film theory, however, a proper clarification of the role of production is a prime necessity. Linguistics-influenced film semiotics has a built-in tendency to retreat into formalism, caused by the ideological deficiencies of the parent discipline. The current trend is to fight this formalism by recourse to the notion of a dynamic generated either inside the system (the text that “produces” itself) or in the spectator-text relationship (activation of desire). While there is a lot to be said for both strategies, they still leave a number of major problems untouched, principally at a social level. The new psychoanalytical semiotics attempts to cope with the relations of a semiotic level proper and another level of determination—that of psychodynamics—by giving an account of the inscription of the desiring subject into the work of the film.

But as for the desiring subject also as class subject—spectator position also as social position—these are problems which continue to evade the scope of the model. And as long as semiotics fails to include them, it not only points up its own theoretical incompleteness, it also leaves their treatment in the hands of practitioners of other, less potentially fertile disciplines. This weakness is evident in Language and Cinema in the insipid liberal pluralism with which Metz treats the possibility of other approaches than a semiotic one. For Metz society exists outside the world of the sign, whereas it would be more correct to say that the sign exists within the world of the social formation. The difference is vital. Materialist semiotics does not yet exist. But it must be invented. Soon.


1. The idea that one can separate signifying from non-signifying aspects of a given practice is fundamental to the semiotic project. Ultimately all film work aims at signification and everything that is done is capable of signifying itself somehow. But a lot of work can still be non- or pre-signifying or signify itself only in a marginal way. For example, there is the whole area of film technology. There are enormous differences of a technical and economic order between Technicolor and other color processes such as Eastman Color. But these differences are of interest to semiotics only insofar as they find expression in the signifying material of film as viewed. The technical specificity of Technicolor (triple negative dye-transfer printing, etc.) is strictly irrelevant to signification. What is relevant is what can be, or is, perceived as coded difference by the viewer—e.g., whether there is color at all, whether it is garish or muted, whether or not color elements function symbolically, etc. All these features have technical determinants somewhere along the line (garish vs. muted is basically a question of saturation) but what is at stake is not the determination as such but its role (if any) in signification.

2. Metz in fact gets into quite a few terminological muddles at the beginning of the book when he is trying to decide how he would like to define his terms, but he sorts himself out eventually. A lot of the discussion at the beginning has to do with the reasons for the difference between his semiotic use of terms and ordinary usage. But since the ordinary usage he has in mind is of course French, and French usage differs somewhat from British and U.S., a lot of the subtlety (and the justification for the whole operation) is going to get lost on the English speaking reader. Basically the end result of Metz’s careful distinction of “cinema” and “film” is to confirm the existence of what he calls “cinematic codes” (forms of expressive organization such as cutting, camera movement, etc.) which are on the one hand purely formal, expressive elements (they have no necessary implications for “content”) but are yet exceedingly material.

In my opinion the materiality of the cinematic codes can be established more simply than it is by Metz (who relies on Durkheim on the one hand and Hjelmslev on the other) by reference to Marxist concepts of labor and production. We tend to think of production as the production of films, but in point of fact cinema itself is also the product of labor. The emergence of cinema as socially instantiated form is clearly not just a question of “Who did what work where and when?” thus involving questions of ideology, representation, etc.. But in the last instance, it is labor performed on and in the material world that is determining.