On Christian Metz: Cinema and Language

2009. this was my encounter with semiotics and i’m glad to say now that i am unhappy about many things i said; i understand this as i have made progress in the field. i keep the post for the sake of a record of first impression


Films mean because people want them to mean. Hence the production of senses as a cognitive process relies much on the hand (mind) of the spectator. However, this is not to say that the filmmaker may present any material to make sense for anybody. In order to successfully establish the communication, the material has to be organized in a certain systematic order. Furthermore, although the system allows the communication to take place and creates meaning out of its material, it also ultimately limits the way it is interpreted. All these characteristics (channeled communication, system of interpretation with its limits) bear a strong resemblance to another system that human being is familiar with: the verbal language; hence, it is not at all inappropriate to ascribe to the cinema certain linguistic capacities.

Yet the arise of a semiotic study of cinema in the early 1970s aspires more than a modest exercise of the above possibilities. In response to a dissatisfaction with discourses in the field that are too general, with observations that are often impressionistic, semiotics wants to establish a science out of the material of cinema, and in order to do so it needs to make use of well-defined categories, to deploy rigorous analytical procedures, to highlight certain points of interest at the price of others. Hence, from the outset, the semiotics of cinema runs the danger of substituting with one methodology the concrete reality that is the object of its study, but this is a price that semiotics is willing to pay.

It is not the first chance these possibilities are brought into knowledge. In fact, the idea of cinema as a language (or anti-language, in the case of Béla Balaz) had been proposed immediately after the naissance of cinema and fell out of favor with the coming of sound. According to Metz, this is probably because now that the cinema has language in it, it can no longer be analogized to it. But even in the age when cinema was deemed as a veritable language, anyone ventured to make such an analogy would immediately and instinctively realize many incompatibilities between these two. Incompatible, yes. But not without a certain theoretical interest.

The laudable effort of Christian Metz seeks to see what was concealed by the metaphor of “cinematographic language” (to distill the turbid muddle for all), and from there, to apply the concepts of linguistics to the semiotics of cinema.

The theoretical framework in which Metz based his work on is a comparison between different forms of signifying procedure, initiated by Ferdinand de Saussure. De Saussure establishes a suture between traditional grammar and structural linguistics. In cinema, similarly, there has been efforts to put forth a grammar of the “cinematographic language” which correspond to the traditional grammar’s role of defining a good usage of the language and consisting of a list of errors to avoid, By contrast, the structural linguistic opts for an universal descriptive analysis of all the languages that are in use. And a linguistic treatment of cinema remain descriptive instead of prescriptive.

Although deemed as laying out the foundation of contemporary Semiology, de Saussure never exceeded a precautious theoretical boundary circumscribed by general observations to the relationship between sign systems. It would take the more audacious effort of Roland Barthes (eléments de sémiologie) to charter out the definite prospect of semiology’s existence as a scientific discipline. De Saussure was content to observe language as a system (he declared that dans la langue il n’y a que des différences). But later semiologists are keen to apply this Saussurean linguistic system (langage) back to its source material (langue). What Christian Metz did to cinema is that, despite all the accusations of wrongly denoting cinema as a langue, Metz nevertheless adamantly believes that the study of film can have a linguistic dimension. It is actually semiotics that we are talking about here, to which linguistics is only a branch. Yet since in fact semiotics was created from linguistics—whereas the former remains to be done, the latter is already well advanced—It is the duty of the older brother to help the younger.

Metz’s first and perhaps still the most well-known attempt to study cinema in the light of semiotics is the 1964 essay Le Cinéma: langue ou language? This lengthy and ambulatory article discusses many issues in a fashion features more rhetoric than rigorousness. And it is well worthwhile to briefly trace its various arguments here.

The article begins with a historical observation that if there certainly was an era where montage is everything (montage-roi), now the king’s power seemed to have greatly diminished. Why is that? In order to answer this question, Metz made several statements. First, he believes that montage-roi signifies the spirit of manipulation, the mind of erector-set. Second, more audaciously, he observes that manipulation implies treating cinema as a language system.

It is apparently a kind of language, but it was seen as something less, a specific language system. It allows, it even necessitates, a certain amount of cutting and montage; its organization, which is so manifestly syntagmatic, could only be derived, one believed, from some embedded paradigmatic category, even if this paradigmatic category was hardly known. Film is too obviously a message for one not to assume that it is coded. (40)

Being a Proust lover, as we all are, Metz demonstrates great use of figurative speaking:

[a message in film] becomes in time like a great river whose channels are forever shifting, depositing here and there along its course a string of islands: the disjointed elements of at least a partial code. Perhaps these islands, barely distinguishable from the surrounding flood, are too fragile and scattered to resist the sweep of the current that gave them birth and to which they will always be vulnerable. Nevertheless there are certain “syntactical procedures” that. After frequent use as speech, come to appear in later films as a language system: they have become conventional to a degree. (40-41)

Metz went on to assert his strong belief in narrativity (echoed in several of his other essays).

The rule of the story is so powerful that the image, which is said to be the major constituent of film, vanishes behind the plot it has woven—if we are to believe some analyses—so that the cinema is only in theory the art of images. (45)

The emergence of narrativity is taken by Metz as a “positive development in the history of film, and particularly in the evolution from Lumière to Mélières, from ‘cinematography’ to cinema.” (44) Narrativity is, according to Metz, capable of being the essence of film. It is at this point he unhesitatingly joints Mitry, who already points out that the Kuleshov experiment does not authorizes the theory of montage-roi, but simply demonstrates a “logic of implication”. (It is on this very ground that Mitry attacked the several indiegetic uses of montage in Eisenstein’s works.)

Now there cometh the talkie. Metz writes that “in the period when the cinema considered itself a veritable language system, its attitude toward verbal language was one of utmost disdain.” (49) Considering the obstinate refusal to speech by silent film theorists, Metz reminds us, the ease with which speech did in fact find its way into the films is paradoxical. Another paradox, even greater, Metz notes, is that the advent of speech was not immediately reflected in theory: “films talked, and yet one spoke about them as if they were silent.” (51) Or there was a tendency to oppose sound to talking where noise and music were accepted but not speech. Here Metz points his finger towards the manifesto by Eisenstein, Alexandrov and Pudovkin. He comments,

Their attitude is positive. For them it is a matter of giving visual counterpoint an auditory dimension, of multiplying the old cinema by the new. However—and precisely because a healthy and intelligent reaction makes us regret its omission, since it provides them with such a rich frame—one notices that nowhere do the three Soviet directors take speech into consideration. (53)

In Metz’s mind, such a different treatment of sound and speech is due to the following:

..any utterance, whether governing or governed, by nature tells us something first, whereas an image, or noise or music, even when it is “telling” us a great deal, must first be produced. (53)

It is thus implied that the Soviet directors perceive image as incongruent with the speech, not with noise and music. Yet to console the montagists, Metz remarks that “the cinema as a language system is also the birth of the cinema as an art, only through its excesses— theoretical as well as practical—could the cinema began to gain a consciousness of itself.” (56)

The cinema is not a language, Metz announces, because the latter is by definition a system of signs used for intercommunication. The cinema is partly a system, with very few true signs and like other arts, it is only a one-way communication. In film, the distance between the signifier and the signified is too short to call an image a sign.

The image is first and always an image. In its perceptual literalness it reproduces the signified spectacle whose signifier it is; and thus it becomes what it shows, to the extent that it does not have to signify it. (76)

Also, “A film is always understood, but always more or less so, and this more or less is not easily quantifiable, for there are no discernible degrees, no units of signification that can be immediately counted.” (72) “A linguistic unit,” on the other hand, “is either recognized or not by the hearer, since it already exists in the language.”

Metz went on to observe that in any work of art, the world represented (denoted) never constitutes the major part of what the author has to say. It is merely a threshold. In the nonrepresentational art it is even missing.

But if the cinema is not a language system, what is it? Metz writes, “A rich message with a poor code, or a rich text with a poor system, the cinematographic image is primarily speech. It is all assertion. The word, which is unit of language, is missing; the sentence, which is the unit of speech, is supreme.” (69)

Finally, Metz lists four approaches to film: criticism, historical, theory of cinema and filmology, which is the scientific study conducted from outside by psychologists, psychiatrists, aestheticians, sociologists, educators and biologist. They consider the cinematographic fact rather than the cinema, the filmic fact rather than the film. The last two are complementary, with bordering cases (Arnheim). The Mitry book is an example of the deep reconciliation of these two.

Linguistics, Metz concludes, which can be imagined as an amiable Monsieur Norpois, has ample reason to concern with cinema, making a few extra demands that will certainly not overburden itself—the busiest people are always those who find the time to concern themselves with others.

Langage et cinéma


From Cinéma: Langue ou langage to Langage et cinéma, there is a discernible difference in tone and in style. The first one circumspectly introduces a question (it takes fifty pages to prepare the reader for the question), the second one takes the answer as granted; the first one is prosaic, resembling to that of Bazin or Mitry, the second one austere, presumably in the lineage of Spinoza. But there is one thing in common, Metz is determined to apply the full force of semiotics to cinema and expect to generate some useful results there. Although the first essay is presented in a questioning form, one cannot help but feel that the question was already answered, since the very fact that we are wondering whether cinema is a language system or is a nonsystematic language already signifies that we are applying some of the discipline’s terms to it, that we are already taking a point of view from semiotics. As the normal way of arguing goes, if cinema is evaluated as not a language system, should we not admit that a linguistic (which happens to be nothing but the study of language system) approach to cinema would be inutile (as Mitry is firmly convinced)? Metz, on the other hand, wants to posit a perspective and insists that it holds up despite unfavorable facts. That is why Langage et cinéma is indispensable in that it illustrates what Metz was only promising but unable to prove in the first essay: what can he do with the cinema, using all the linguistic terms he can think of?

Langage et cinéma is characterized by a “scientific style” where a certain methodology precedes the subject, where the question of range has to be resolved (as if it were possible) before everything else. The approach adopted by this book is determinedly top-to-down. There is no question what would be the best theoretical apparatus to cope with the structure of film (linguistics just comes handy), or even whether or not linguistic is the best candidate among alternatives (because it is the only tool he proposes). What matters is rather, to which extent a theoretical framework such as this can be bent to accommodate the specificities of the cinematic medium (we shall witness a constant effort to reshape the linguistic terms and their theoretical implications).

To illustrate this point, let us look at a matrix of choices made by Dominique Chateau in his Le cinéma comme langage. (29)

a)      Le cinéma est-il un langage?

b)      Possède-t-il une grammaire?

For these two questions, the answers of three persons are as follows:













Now this table reveals, if nothing else, that all possible choices regarding the relationship between language and cinema are legitimate. Even if we identify cinema as a langage, the idea of applying linguistic concepts does not mandate a grammar for the cinema. A theoretician may propose a grammar, or merely a rhetoric[1] (as Mitry, Bordwell do), if it is already sufficient and efficient to explicate his area of study. And if the application of a certain theoretical term proves to be problematic, why insist? It is my personal belief that the ultimate criterion which we can use to evaluate a certain approach is not how well is defines cinema (since cinema defies definition), but rather, what can this particular approach add to our understanding of the cinema.

To readers who have not yet read Metz’s previous articles or those who unfortunately did not agree with its conclusion, the premises this book bases itself upon may seem suspicious. But to Metz it is all too natural and consequential. What is the demolition of the previous prevailing notion if not to serve to erect a new edifice from ground zero? The problem of film/language analogy is not, as Metz saw it, the result of too much application of linguistic theory in film, but on the contrary, too little. Thus Metz calls for a return to linguistics with all its terms: code, message, system, text, syntagmatic, paradigm and so on.

The quadrants

Let us begin by looking at four closely related terms, which I organize in the form of quadrant:










According to Metz, a text is a film, or a part of the film, or a group of films. In any such text, multiple messages are delivered. How one particular message is delivered is an instance of specific codes; an actualized code is a system.

So far so good. But as Metz continues to refine what exactly each of these terms denotes and what are their interactions, problems arise. Taking up the example of Ordet, for instance, Metz argued, there are three principle types of systems: general cinematographic codes, particular cinematographic codes and the third, singular system.(58) 

First, what are the difference between general cinematographic codes and the particular ones? Metz himself noted that (60) that the distinction between them is only a matter of degrees of generality. Western is a particular code under the general tab of narrative fiction; yet Western becomes a general code if we study Italian Western, which in turn becomes general if we shift the object of our study to Sergio Leone.

But if such a pair of oppositional adjectives can be useful (like high and low) without denoting an absolute value, ultimately it is better to do without them if really scientific terminologies are wanted. In Metz’s logic, ideally, if the general cinematographic codes are the codes that apply to all films, then what is cinema can be conveniently defined as the entire body of work in which cinematographic codes are in use. This in itself is a vicious cycle .Yet what is worse is that it seems to me that cinema is indefinable in such a way, when every single aspect of film has been proved as non-essential (narration, character, camera, celluloid, image). The only trait I can think of that is still common to all films is that they are all framed. But can this be justly called a cinematographic code? Likewise, the exact distinction between cinematographic and extra-cinematographic is constantly blurred. Instead of having endless disputations here, I would rather propose that we divide codes into categories that pertain to our sensory and cognitive faculties: visual codes, auditory codes, and of course, codes that allow us to comprehend cause and effect. Each artistic discipline (painting, music, theater, film) corresponds to a different subset of these codes; and for each artist, he/she needs to employ a particular subset of codes to a satisfactory composition. Two things need to be clarified immediately: first, the codes employed by artists do not differ from the codes on which a normal person perceives his world; second, it is customary for an artist to function entirely within the regulations of a particular discipline—that is, he uses only the codes prescribed by this discipline—but it is completely acceptable that he borrows codes that are conventionally not deployed in this place. This act of transgression is often not regarded as a violation of codes, but as an expansion of the particular discipline.

Obviously, in cinema, it is more and more difficult to assert, among multiples codes that are in use, which are cinema-specific, which are not. Metz claimed that the accelerated montage is cinema-specific, while chiaroscuro is from the painting. To my knowledge it is safer to say that one is originated from cinema and the other from painting, since the codes are constantly displacing in multiple sub-cultural fields and it would be naïve to believe they still belong exclusively to one particular area of culture practice. Cinema borrows heavily from literature, music, painting, theater and others. The very notion of narration comes from extra-cinematographic sources. But does that mean voice over is just an adaptation of literary monologue? Is the accelerated montage a way of sequencing image, or a mode of narration? From Sidney Sheldon to Michael Crichton, I find the latter sense in very effective use. Ingmar Bergman, Jacques Rivette and Michael Haneke are all heavily indebted to the theater, but are their methods not cinema-specific and still remain at the stage of Marcel Pagnol? Video games, the latest means of mass entertainment, bring the full range of cinematic experience to the desktop with its own idiosyncratic mode of entertainment.

Under these circumstances, is it still advisable, if possible at all, to be content with the mere enumeration of codes into respective cabins? Nowadays every code—a bit of exaggeration granted—is culturally interchangeable and shared by a multitude of media activities. The general codes of social conduct, as well as the particular codes of a specific genre, are undergoing constant revision. Is it fair, for instance, to maintain Western as that which was established by John Ford, as a set of rigid codes of characters, events and moral values, or to allow revisions by Nicolas Ray, Eastwood, Costner, or even Ang Lee?

What is in my opinion the most confusing aspect of Metz’s tripartite is the last: système filmique singulier. According to Metz, the reason why this has to be singled out is because the study of codes is always comparative and non-exclusive whereas the study of singular filmic system is by definition not. Of course, singular is by no means the synonym of original. A banal system, Metz brilliantly argued, is yet another system that is quite “semblable à de nombreux autres systèmes. Assez semblables, et non identique.” What is problematic is that although the study of a système filmique singulier is never a study of cinematographic specificity (because this latter calls for comparison), the entity is only called such because it is a unique combination of the cinematographic specificity, that is, cinematographic codes. In other words, without the codes and their particular form of dynamic interaction, there is no way to conceive what it means by système filmique singulier. Therefore, it seems to me that such a term does not have enough ground to stand side by side with the notion of code, cinematographic or extra-cinematographic, general or particular. It is as if a scientific study of Christian Metz as a person commences by claiming that he is a man, a son, a husband, a father, a film scholar, etc.,  and finally, himself. But if we acknowledge that any aspect of human life can be generalized and coded since it is probably shared (if not, divide further) by more than one person, then naturally what constitutes Christian Metz is nothing more than a unique, organic and dynamic combination of many such codes. Ideally, there should be no mysterious, non-qualifiable core in the heart of a text; or if there is, it should be excluded from our code-oriented study since it automatically disqualifies.

Metz himself is probably aware of the awkwardness of the above terminology. From a certain point(97) he substituted code cinématographique particulier with the notion of sous-code and consequently code cinématographique géneral with simply code cinématographique. As for système filmique singulier, it was already previously (88) renamed as système filmique textuel. The nuance between singularity and textuality, as Metz demonstrates, is that “la singularité n’est pas exactement le caractère définitoire du textuel, mais plutôt un corollaire de cette définition: ce n’est pas parce qu’il est singulier qu’un texte est un texte, c’est parce qu’il consiste en un déroulement manifeste antérieur à l’intervention de l’analyste.” (88) Later, Metz further classifies both code and sub-code to système filmique non-textuel (112). To simplify the matter, one is led to believe that what can be codified is textual; and the rest (Metz doesn’t specify whether permanently or temporarily) goes to the “unclassified” bin that is labeled “non-textuel”. But again, does it help to improve the overall clarity of the whole system of classification? I think not.

The Syntagmatic and the paradigmatic

Stephen Heath once remarks “the focus on syntagmatic relations ‘saves’ Semiology in the face of the paradigmatic poverty of cinema.” (4) There are four messages in this sentence: the cinema is paradigmatically poor; Semiology focuses on syntagmatic relations; syntagmatic relations in cinema is richer (implied); by focusing on a richer aspect of its study, Semiology saves its face. For these four messages, I only accept the second, which is a fact. But let me first resume what Metz said about the pair of syntagmatique and paradigmatique.

The idea comes from de Saussure, where he observes linguistic properties implied in the text as:

En dehors du discours, les mots offrant quelque chose de commun s’associent dans la mémoire, et ils se forment ainsi des groupes, etc.… ainsi le mot enseignement fera surgir inconsciemment devant l’esprit une foule d’autres mots (enseigner, renseigner, etc.).

Metz believes, however, that de Saussure wrongly interprets this phenomenon as “dans le cerveau”. He regards this is the kind of “psychologisme” and “associationnisme” as “parmi les aspects de la pensée saussurienne qui ont le plus vieilli.” (126) Metz rightly asserts that semiotics should be concerned only with the study of discourse, not the psychology of discourse. Hence, this phenomenon refers to the “paradigme”, which is by definition “une classe d’éléments dont un seul figure dans le texte: c’est donc le proper du paradigme que de ne jamais être déployé dans toute sa surface au niveau textual.” (124) The “syntagme”, on the other hand, is always textual and given, albeit it might feature both simultaneity (temporal and spatial) and succession.

Apparently, the pair of “le syntagmatique” and “le paradigmatique” presents a virtual/actual contrast reminiscent of the quadrants that have been the focus of previous chapters. And the study of both (la syntagmatique and la paradigmatique), Metz stresses, are both “codique”. (124) The study of syntagmatic departs from a given text to arrive at syntagmatic regularities; the study of paradigmatic is accomplished when the paradigm is discovered. In this way, the latter complements the former by extending the notion of the text into an ensemble of texts, a big text, if you wish, where both syntagmatic and paradigmatic can be found.

Again one must note that proposing these two studies to cinema actually acknowledges that the cinema has a grammar (and two!) and that we can apply something to cinema that is firmly associated with the study of langue (from de Saussure, grammaire associative et syntagmatique) while Metz has already claimed that the cinema is not a langue. It wouldn’t be a surprise, therefore, that the result is multi-folded. On the one hand, the notion of syntagmatic works well with cinema; on the other, the paradigmatic does not.

In the verbal language, if we substitute a minimal unit in the sentence (a word, or a letter) by another, there is a high chance that the sentence would not make the same sense. According to Odin (93), we say the result is wrong, or even it no longer belongs to that language. In cinema, however,  we do the same (no matter what the minimal unit is, cineme, iconeme, videme…), and the result still makes sense and is still cinema. Therefore, Odin argues, if we can substitute one element of the message for indefinite others, the value of opposition of this element is considered weak. Moreover, in the paradigm scenario established in linguistics, elements from the same class are mutually exclusive: you cannot have two subjects for the same sentence; this, again, is not true for the cinema.

It is easy to arrive from here the conclusion that cinema is paradigmatically poor. But let us take another perspective of the issue, temporarily. Cinema and language both model after the real world. And in doing so they adopt different approaches, making distinct abstractions and compromises. Both excel in depicting the world. Yet the ways they succeed are sometimes compatible, sometimes incompatible. Certainly we can substitute an image or a framed-object in a film and still make some sense, at least for someone, but this sense is always perceived as in contrast to a norm[2]. The syntax of cinema is indeed very loose, but the fact that we are looking for syntagmatic regularities proves that the making of senses is achieved in relation to a norm. In this sense, none of the filmic image is completely arbitrary. It is true that compare to the cinema, verbal languages seem to be better “regulated” and have equipped themselves with a repertoire of laws. But let us not forget that not every linguistic law can be applied to all languages. Some languages have more inflections (conjugation and declension) and are consequently more flexible in word order (Latin, for example); but the same result can also be attained by not using any inflections at all (Chinese, for example). The assertion that cinema is paradigmatically poor is a poor assertion itself since it implies that one has to use exactly the same means for a similar result, for example, to sustain causality, substituting subject with only subjects, not a predicate. It also suggests, rather “arbitrarily”, what is language and what is cinema without specifying which language, which cinema. The poem of Isidore Isou (Odin, 94), for instance, is considered not belonging to the language while similar productions of Stan Brakage is considered remaining in the cinema.

Another aspect of the paradigmatic is that, although for a scholar self-charged with the study of a filmic text, the paradigmatic dimension seems intangible and arbitrary, for the author of this text, this is probably not the case. A director has to choose between different takes of a same shot (paradigmatic) not only by their internal virtue, but also how it functions inside the syntagmatic order of the bits he shot. These apparently two different tasks in reality are inevitably intertwined and the choice of one affects another. A director sees a different text in front of him/her not only because he/she is the author of it, but also because, he/she sees more syntagmatic and paradigmatic dimensions. In contrast, an average moviegoer would not experience a same film in different orders or some parts of it substituted with different takes. He accepts the film as it is, without bothering himself with the endless question of how to improve it. Naturally, there are cases where the paradigmatic dimension can be larger than what a director can see. A remake, for instance Van Sant’s Psycho, provides a paradigmatic dimension of which even Hitchcock is certainly unaware.

The best thing of the syntagmatic and the paradigmatic is that they help to define the code: une code est une machine paradigmatique et syntagmatique à la fois. (128) In fact, Metz claims, “il est impossible dans le domaine cinématographique, de définir la notion de séquence autrement que par ses différences avec d’autres formes d’agencement syntagmatique.” Consequently, a sequence cannot be said to have been established unless both of its internal (the exact images that it is composed of) and external (the exact position of this sequence in the entire film) syntagmatic order have been established. Obviously, here the syntagmatic establishment has as prerequisite numerous paradigmatic decisions.

Perhaps in order to salvage the “paradigmatic poverty of cinema”, Metz claims that in an alternating montage (his favorite example) sequence A-B-A-B, the simple enunciation of A/B is not only a syntagmatic order, but also a paradigmatic one, since A/B signifies a logical opposition (129). But here I believe Metz contradicts himself. According to his own definition (see above), the presence of A/B is only a case of simultaneous syntagmatic order, but not paradigmatic, for the simple reason that they are already given in the text. Lévi-Strauss’ mythemes, Propp’s formula on Russian fables, they are all extra-textual and typical studies of the paradigmatic. A myth, a fable, on the other hand, contains a syntagmatic order actualized in one text. Therefore, if a film presents us in itself different endings of a same story, as they sometimes do these days, this is always a syntagmatic strategy, not a paradigmatic one. Likewise, if Eisenstein presented us in Alexander Nevsky the allegorical opposition of black and white, it cannot be paradigmatic exactly because it is quasi-arbitrary. (134)

What I do agree with Hearth’s observation is that in cinema the process of making meaning relies more on syntagmatic order than on paradigmatic ones. A film can be comprehended by itself to a certain point without the help of paradigmatic dimensions (who made it, under which circumstances, how, etc.). In the production notes of Kill Bill, we read “WE DO A QUICK SHAW BROTHERS ZOOM INTO HER EYES.” Do we really have to know who the heck are Shaw brothers? Or even what is a zoom? Certainly not. A film is given to us; and it is what is given that counts. This is especially true for narrative fictions, which happens to be Metz’s object of study. Semiology, at least that of Metz, focuses on the syntagmatic order of cinema, not because it is richer (would you say a limited selection is richer than an unlimited one?), but because in cinema, especially in the narration films, the syntagmatic is more tangible for both the spectator (which includes the scholar) and the task of analysis. Cinema is in no way poor in paradigmatic dimension. On the contrary, cinema is probably the richest in paradigmatic dimension amongst all art and all communication. The only problem is that what we used to qualify this aspect of the “cinematographic language” does not work very well any more. Since cinema does not have a dictionary, it certainly cannot have a thesaurus or an associative grammar. The notion of grammatical category cannot exist in cinema, so does a transformational grammar of the cinema.

A discussion on the system of signs

Although there might be different opinions in regard to in which way the Semiology of the cinema can be progressed, the first problem that was identified in Metzian semiotics involves the very notion of sign. Peter Wollen, in his comments on the matter, published as early as 1969, already points out that the inflexibility of the Metzian notion of sign lies where de Saussure started the whole enterprise. This was made clear by a comparison with the works of Charles Sanders Peirce, who developed independently another conception of the system of signs. Peirce’s works are massive and cover extensive grounds. But his theory of signs is often resumed as follows. According to Pierce, there are three kinds of signs: an icon is a sign which represents its object mainly by its similarity to it; an index implies an existential bond between itself and its object; finally, a symbol presents an arbitrary connection. With the help of this trichotomy devised by Peirce, Wollen proposes that the cinema is mostly iconic and indexical whereas the verbal language is by and large symbolic. Another point Wollen stresses here, also from Peirce, is that these three categories are not mutually exclusive. In fact, they frequently or even invariably overlap and are co-present. Photography, Peirce noted, is the good example to illustrate this point.

To outline make my understanding of the matter, I would like to inquire about the said trichotomy as follows.

First, if indeed the three classes of sign do not have a clear-cut boundary between them and co-exist in many instances, the term trichotomy is probably not the right word to use here, because it signifies a division. Instead, I propose we use the notion of layer. Layers overlap by definition and generate hence a combinatory effect where the receiver is not always aware of its composite nature.

It goes without saying that the sign operates differently in verbal language and the cinema. How different? In verbal language (that is, I need to stress, in alphabetic languages), icon seldom makes an appearance. In cinema, the iconic is omnipresent and immensely effective. Eisenstein, we are told, went into great length picking out the right actors in order to make effective use of their image, especially their faces (typage).

The indexical layer becomes apparent when an object refers to another object. In the domain of verbal language this phenomenon is often called metonymy and almost invariably involves a concept. In cinema however, an indexical sign must also point to a concrete object. A rolling gait hanging in the closet points to the identity of sailor, not necessarily which sailor; but an added dimension of the real owner of the costume is often desirable. The pince-nez in Potemkin is such an example. It can of course be wore by no matter who[3]. But in order for the sign to function properly, it is better regarded as pointing to the rank and pro-Tzarist aristocracy; or better still, points to a certain Dr. Smirnov. And the general notion of “ruling class being thrown overboard” cannot exist without this singular sample of the ruling class being thrown overboard first. Whatever the signified idea, the represented object must be in the first place confirmed as an object. This way, and only this way, can a notion be introduced into the context of film and functions properly.

It is customary for us, in painting, to talk about whether a portrait resembles its object or not. Because there is a high chance that, due to the lack of expertise or purposeful “artistic” rendition for the part of the painter, it does not. In photography, however, the indexical side is unquestionable because it is carried out in a mechanical fashion, free of human intervention.

By using our metaphor of layer, we can illustrate the differences between these two in a much more intuitive way—that is, in painting, the indexical layer is opaque whereas in photography it is transparent. Therefore, what Bazin finds especially pleasing in Italian Neorealism and repulsive in German Expressionism, for example, The Cabinet of Dr Caligari, Nibelungen, is the opacity value of this indexical layer. In Rossellini the opacity is zero (things are, why manipulate); whereas in Lang, or even The Third Man and Ivan the Terrible, this value is rather high. Bazin developed an aesthetic which was founded upon the indexical character of the photographic image.

As for the symbolic, according to Pierce, it refers to all the signifying instances where the signifier and the signified bear neither physical resemblance nor existential bond. Under this circumstance, the connection is perceived as “arbitrary.” Yet obviously, the arbitrariness of the signification does not mean that I can arbitrarily refer one object to another and expect the others to understand. Naturally, the connection is not really arbitrary, but only conventional; it exists by a contract, so to speak, shared between different parties of the communication. If ever, as it often happens across radically different cultures, the contract is invalidated and the sign will not function.

What are the symbolic signs in verbal language? The favorite example given by structural linguistics is the connection between articulation and meaning. The fact that cinema does not have a double articulation signifies one thing: cinema does not have, or at least does not encourage symbolic signs.

“In the cinema,” Wollen has argued, “indexical and iconic aspects are by far the most powerful.” Contrary to the verbal language, in the cinema it is rather difficult to name any instance that is purely an arbitrary sign. The symbolic does find its usage there, but it is never without the cooperation, or rather the foundation of indexical or iconic facts. The God sequence in October shows why symbolic signs along can not be effective in cinema. And theoretically, Wollen remarks, if we are to take one aspect (as Bazin with index, Metz with symbol) as The essential sign of cinema, we are “impoverish the cinema”. The rhetoric of cinema is based on all of them and especially, the interaction within the three.

Despite the rigorous thinking, Metz does not question an over-restricted Saussurian notion of sign. Compared to Pierce, Saussure limits the sign system as consists of only arbitrary ones, that is, symbols. The onomatopoeia, he argues, are never organic elements of a linguistic system. Besides, their number is much smaller than is generally supposed. In short, they do not contribute to the complex and dynamic structure of the system but rather, are the surplus, marginal residue of the language in its primitive days. Jakobson corrected this view by pointing out that both the iconic and the indexical can be found in syntactic structure of the language. He argued that the poet’s use of language is a challenge between the arbitrariness between a sign and its signifier.

If in alphabetic languages the link between spelling and meaning is perceived as arbitrary, in hieroglyphic languages this is not the case. In Chinese, for instance, a symbol is never wholly arbitrary. There is always a natural bond (most of the time the rudiment of it since it has been five thousand years) between the signifier and the signified. The words, when they were first invented, are icons.  Every elementary school kids can testify to this point, since their first lesson of writing can hardly be distinguished from a lesson of drawing. Even in cases where the connection is indeed arbitrary, it is not arbitrary in the sense that it can be substituted by something else.

Emblems are labile; they may develop into symbolic signs or fall back into the iconic. In fact, there is not clear-cut point where we can say an icon has definitely become a symbol. The only thing we can say is one icon is more symbolic than the others due to the fact that this icon is more often connected to a fixed signified.

A poet aims to challenge and destroy these established symbols and reduce them to their iconic state. In the meantime, he inevitably establishes new symbols, which is a vocabulary of his own. A word, in the hand of a poet, becomes an image where the icon is again visible. The craft of working out a poem, then, is similar to the montage process, where images form a sequence. A word signifies nothing alone; it is where the poet puts it that makes all the difference.

In fact, if in medieval paintings the symbols and their deciphering occupy an important place in the overall meaning-making, for impressionists, symbols are for a considerable period of time completely ousted. Wollen, for one, notices that Courbet echos Bazin and Rossellini: nature is there, why manipulate?

The beautiful exists in nature and may be encountered in the midst of reality under the most diverse aspects. As soon as it is found there, it belongs to art, or rather, to the artist who knows how to see it there. As soon as beauty is real and visible, it has its artistic expression from these very qualities. Artifice has no right to amplify this expression; by meddling with it, one only runs the risk of perverting and consequently of weakening it.

Wollen suggests that de Saussure dismissed natural signs as unexpressive and less meaningful. He also suggests that, obscured beneath Metz’s semiological analysis is a very definite and frequently overt aesthetic parti pris. Because like Barthes and Saussure, he perceives only two mutually exclusive modes of existence for the sign: natural and cultural, uncoded (which is named, as we have seen, textual) or coded.

If we restrain ourselves to the domain of real world in the discussion of sign systems, we may say a sign (iconic, indexical, symbolic) is a link between two objects. Whether or not this link is to be acknowledged is another question: sometimes it is; sometimes it is not; it depends on the user. A portrait, for example, may resemble its subject or not, depending on my attitude; a straitjacket may signal a patient, or nothing at all; as for symbolic signs, I reserve my right to deny any of them in case I need to associate the signifier to another object.

Verbal language is different: by using it, we undertake a thousand conventions of its signifying system. Words are merely words. How can they signify anything without a grammar, a vocabulary, a syntax and a million other things?

Film lies somewhere between the verbal language and the real world. It encourages the establishment of conventions: there are cases when not only are film images and sounds no longer to be thought of as fragments of reality, they now do not even refer to the real—in this sense we may say the film has become a reasonable language, albeit a language still in the process of making itself clear. But unlike language, a film can function without any convention by falling back easily to the real world (the level of natural signs) with all its arbitrariness. This can happen when a spectator refuses to recognize signs that are perfectly conventional, as reported by André Maurois (cited in Kracauer’s Theory of Film, p175);  or it can happen when a filmmaker makes use of his private signs into a film that is destined to public viewing, where his signs would most probably not function as he has envisioned them to be. But here lies the power of cinema: between the thoroughly public symbolic sign and the potentially private natural sign lies a whole range of ambiguous signs about whose status we are unsure. Thus the spectator is free to exert his power of interpretation over these signs, from entirely rejecting as incomprehensible, to partially accepting as potentially insinuating, to wholeheartedly intaking as a manifestation of his own concrete experience.

Film is a language-like communication which strives to be art; but it is none of these three in its simplest form. Unlike verbal language, a film represents a one-way communication where the message unrolls before a silent spectator. And unlike art, which has to be communicative, a film can be communicative without being art. Finally, exactly because film is not a language, and has neither grammar nor vocabulary, film as a means of communication is characterized by a linguistic inefficiency. All languages have exceptions and aberrations. In this sense film is the poorest among them, whose incorrigible imperfection should have been begging a linguist’s effort to become a “pure” one such as Esperanto. Instead, not only does it already make perfect sense in an intrinsically ambiguous way, it makes sense despite everything else, and will always do.



Any examination of Metz’s work without paying attention to its theoretical context is incomplete. In this respect the single most important name to pronounce is Jean Mitry. In fact, it is profitable to look at what Metz has chosen to do as a counteraction to Mitry’s monumental work. This counteraction is described by Dudley Andrew as:

Despite Mitry’s thoroughness, indeed all the more because of it, Metz is convinced that film theory must achieve a reorientation. For him the first fifty years produced at best a diverse, intelligent, and sometimes astounding view of the medium, but in every sense a view which must be termed “general”. (212)

What exactly is too general in Mitry’s gigantic effort? Is it not exhaustive, not well organized, or not scientific enough? It seems to me that what Metz meant by “general” here can be better understood as “too ambitious”. It is simply impossible, in Metz’s view, to encompass all areas of phenomena that relate to cinema. That is why in Cinema and Language, Metz lavishes (for the length of three chapters) on the careful delimitation on his subject matters: what is cinema, what is film, film inside cinema, cinema inside film, and so on. All these distinctions would seem to have served no purpose (does one do the same for literature/novel?) if it were not for its context—Mitry’ work.  Mitry has done film theory an incomparable service by touching virtually every problem film theorists need to encounter, Metz might well proclaim, “but let us now get to the real business.” In opposition to Mitry’s aesthetic and psychology (human, too human) of cinema, Metz proposed a neutral and specific approach.

In response to Mitry, Metz asks, does a film theory need a statement on what is man’s state in the universe and the function of art? Naturally, such a statement, once established, will enable the critic, with utmost coherency, to tell good art from bad art. But Metz finds this too subjective. What he proposes instead is something totally impersonal, and just by the look of it, more scientific. But what makes science scientific is not that it is neutral, but that it can be verified or falsified. Humanities, on the other hand, do not possess such an absolute value, no matter how scientific it might seem to be.

Metz claimed to establish a description of the processes of signification in the cinema. How scientific can this description be? The system/code/message/text quadrants, for example, find no solid correspondence, as we can in physics, to concrete objects in the real world. Instead, the whole project can be at best a system of conjectures. Metz’s method is what a discourse typically does, that is: mark out what appears to be a new area of human experience for preliminary analysis, define its contours, identify the elements in its field, and discern the kinds of relationships that obtain among them. It is simply a discourse which adds to many others. And in circumstances like this, there are always legitimate grounds for differences of opinion as to what they are, how they should be spoken about, and the kinds of knowledge we can have of them.

Also in response to Mitry’s extensive coverage, Metz boasts of specific inquiries. Instead of sitting back, pipe in mouth, ruminating about the origin, purpose and general laws of his subject, Metz prefers field and laboratory work which, unconsciously or not, suits better to what he conceives as scientific research.

The last issue is the research method. Andrew has already pointed out that Metz’s reliance on the work of others. He wrote, “by attending conferences on semiotics, exchanging manuscripts and bibliographies with scholars from virtually every country, and by arguing with colleagues in other branches of semiotics, Metz’s theory consciously builds on gains made by others. In this way he has indeed given to film theory at least the outward appearances of a progressive human science.” Such is the reality of film studies as a scholarly discipline today. It is Christian Metz, among others, whose effort establishes an example of how scholarly work can and should be done in the cinema.


Semiotics tends to treat realism as a specific mode of signification which has no special rights or privileges. By doing so, semiotics automatically devalues any debate around the virtues of implementing a cinematic realism—neither the faithful duplication nor the intentional distortion of the real implies any intrinsic value. And if the latter, represented by a modernist cinema, is readily recognizable, the former, still the dominant form of our time, yields interesting results once the code on which it operates is exposed.

Semiotics reveals the cinematic reality as an implicit point of reference in film criticism. By extending this line of thought, we might also see that the so-called reality is also built up by perceptive codes, albeit differ from those in cinema. The inquiries we should be making, therefore, are not how much film approximates reality, but how much the codes of vraisemblance approximate codes of actuality.

The grand project of the semiotics of cinema, as envisioned by Metz, was to progressively illuminate, on the one hand, the codes that all films must draw on to signify anything, and on the other, how in particular films, genres, or periods, these codes interweave with together. But one important underestimation of the task is that, the cinema functions not so much as a system of codes/rules, but as a place where multiple codes come together and lend each other its signifieds. A consistent meaning does not exist (consider how, for instance, by the virtue of hindsight many perfectly harmless Hollywood fabrications are deemed as highly subversive). Nor does a construction of codes exist where meaning whatsoever is totally absent.

The various scientific terms that Metz introduced into the cinema studies seek to understand how films mean. Curiously, although it is vital to understand how films mean, the often brilliant analysis supplied by Metz do not illuminate our understanding of the cinema in several ways. First, the terms also apply to other forms of communication, if they happen to rely on the power of visual narration. Second, the complex relationship between sound and image, one of the definite specificity of cinema, is hardly mentioned. Metz treats the films, I regret to say, as if they are completely silent. This is not to say Metz is not aware of the importance of the soundtrack. On the contrary. It is the theoretical framework in which Metz works does not provide a ready-to-wear for such purposes. Finally, such an approach does not concern at least what are the purposes (or meanings proper) of the cinema, and intentionally so. Yet I cannot help but to feel that meaning and purpose are two sides of the same coin. As long as we are still looking for meaning, the study of cinema probably still belongs to humanity, not science. Granted, linguistic terms are much more impressive than impressionistic observations. Yet one wonders why Noel Burch can operate on films in a formative way comparable to semiotics without borrowing anything from linguistics. For example, in his Theory of Film Practice, Burch churned out fifteen possible relationships between succeeding shots. Apparently, there is a similar exercise associated with Metz, by the name of la grande syntagmatique[4]. In this sense we might say Burch’s structural approach achieves more or less the same thing with Metz’s linguistic analysis. By staying formative, both expel the realistic or “natural” as the ultimate reference point. But in Burch’s way, films do make a difference: they are well crafted or badly crafted; in Metz’s way, however, all films are same: I strongly believe that this implication will inevitably invalid any serious study on the subject.





Works Cited


Christian Metz, A Semiotics of the Cinema.

------------------, Langage et cinéma.

Peter Wollen, Sign and Meaning in Cinema.

Stephen Heath, The Works of Christian Metz. Screen, v14, n3, Autumn 1975.

Andrew Dudley, Major Film Theories, Oxford Univ. Press: 1976.

-------------------, Concepts in Film Theory, Oxford Univ. Press: 1984.

Jean Mitry, Sémiologie en question.

Francesco Casetti, Theories of Cinema 1945-1990.

[1] Metz once asked the same question, rhetoric or grammar, in Problems of denotation in the fiction film.

[2] The norm is often “what happens in the real world.” A man wearing a boat signifies rather “a man should be wearing a hat, but instead, he is wearing a boat.” Here the image refers to an improbable situation in the real world. But if Bruce Willis beats up a woman in Die Hard, which is quite probable in the real world, the norm is now “what usually happens on the screen.”

[3] In Potemkin, pince-nez is seen on a pensioner and a schoolmistress. See Bordwell, The Cinema of Eisenstein, p70.

[4] Problems in the denotation of fiction film. p119.



Anonymous said...

The term "cinematography" (?) is interesting; I remembered narratologist Garrett Stewart's term "narratography" in his recent study of filmic narrative. seems that both sound similar to each other, despite their different disciplinary backgrounds.

No offense but you sound a little "nuts" at the beginning. I will read more to understand why.

Allan said...

It is a perplexing fact that in cultural
studies countless publications discuss
the American Dream while there is no
mention of a British Dream. This paper
proposes to use the term British
Compromise, which will be sketched
around three features which
characterize life in contemporary Britain:
the search for financial security and
decent living; the search for educational
opportunities; appreciation of cultural
The essay further explores the
representation of the British
Compromise in contemporary British
film in order to show how films discuss
social matters and how they log into
mentalities and living conditions to
reconstruct, reinterpret and reinvent
them. The scarcity of resources in the
British film industry has led to films
which allow, and also challenge, viewers
to concentrate and reflect upon the
artefact. Such films are seen as ideal
objects of film education because they
facilitate didactically motivated reduction
and concentration on the basic elements
of film language.
In the end, this article is meant to show
how foreign language teaching can
benefit from the use of British-Asian
films in particular. They provide learners
and teachers with extraordinary
perspectives on culture.
Italian Language Lessons

Popular Posts

Blog Archive