You are on page 1of 18

Christian Metz and the Semiology of the Cinema

Author(s): Alfred Guzzetti


Source: Journal of Modern Literature, Vol. 3, No. 2, Film as Literature and Language (Apr.,
1973), pp. 292-308
Published by: Indiana University Press
Stable URL: http://www.jstor.org/stable/3831037
Accessed: 15/12/2009 11:28

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=iupress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Indiana University Press is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Modern Literature.

http://www.jstor.org
ALFREDGUZZETTI
HARVARD UNIVERSITY

ChristianMetz and the


Semiology of the Cinema

THE THREE BOOKS1 and numerous articles that Christian Metz has
written since 1964 give him fair claim to be considered the most
important film theorist since Andre Bazin. To Bazin's question What
is Cinema? he has replied cinema is a language. Though he is not
the first to give this answer, he is the first to inform it with a sophis-
ticated understanding of modern linguistics. Using models drawn
from this discipline, his work attempts to confer on film theory the
virtues of systematization, objectivity, and precision.
Metz is, to use his own term, a semiologist, and, as such, groups
natural language with signifying systems as diverse as those of myth,
dress, food, cinema, kinship, politeness, painting, poetry, and car-
tography. These systems are taken not simply to serve a function
analogous to natural language but to be open to the sort of analysis
developed in its study. For these reasons, linguistics plays a double
role in semiology. As the source of method, it is the parent disci-
pline; and as the study of a single signifying system, it is a branch of
the more general, though still embryonic, science.
Since neither semiology nor the school of linguistics on which it
is based is very well known in the United States, I shall try to give a
brief outline of its premises and the place of Metz's work with rela-
tion to them. Semiology derives its linguistic framework from Ferdi-
nand de Saussure, who furnished the indispensible distinction
between the general term langage (e.g. French, English, Russian,

'These are Essais sur la Signification au Cinema (Paris, 1968), Langage et Cinema (Paris, 1971), and Prolo-
sitions Methodologiques pour L'analyse du Film (Universitatsverlag Bochum, Germany, 1970), the last of
which I have not been able to locate. My translations of quotations from the first two of these books
(abbreviated as Essais ... and Langage .. .) are incorporated in the text of this essay.
293

etc.) and langue, the system within a langage that comprises such
elements as the phonetic, syntactic, and semantic sub-systems,
which together permit the intelligibility and multiplicity of utter-
ances. Forthe term langue, which refers specifically to natural lan-
guage, Metz substitutes the less prejudicial word "code," which is
defined in opposition to the complementary term "message." The
code is composed of signifying elements and the message of signi-
fied elements, called respectively signifiers and signifieds. A code,
to merit the name, must be instanced in more than one texte, a
word that I shall leave in French in order to indicate its pertinence
to any division of any discourse, whether a poem, a painting, or all
the films of JerryLewis. A texte may embody a single code, many
codes, or no code at all. In any case, the respective messages will be
the counterparts of the codes and not coextensive with the texte.
Although code and message are complementary terms, their signi-
fiers and signifieds may not be isomorphic. Metz illustrates this
point by citing Gilbert Ryle's observation that the division of the
verbal signifier "Fido" into its four constituent phonemes does not
entail a corresponding division of the signified, which is, of course,
a mental response to the signifier, not the real animal. By contrast,
in the case of the cinematic image, a signifier which shows a dog
will, if divided, effect a corresponding division of the signified; that
is, it will signify a portion of a dog. Thus the relation of signifier to
signified, and of code to message is a complex matter informed by
problems specific to particulartextes and languages.
Finally,there is the distinction between syntagm and paradigm.
Following Roland Barthes's suggestion in Elements of Semiology,
this may be envisioned as a pair of Cartesiancoordinates in which
the syntagm is the horizontal dimension and the paradigmthe verti-
cal. The syntagm is the dimension, whether temporal as in the case
of music or a-temporal as in painting, along which the message un-
folds, and the paradigm is the system of alternatives that comprise
the code. The paradigm portraysthe code as a system in which a
given signifier takes its meaning through opposition to others
which are possible but absent in a given place on the syntagm.
Semiologists preferto organize these oppositions in binaryor bifur-
cating structures.
Metz's use of this general theory is characterizedby his resistance
to a literal interpretationof the analogy between natural and cine-
matic language. From the outset he has denied that there is any-
294 ALFREDGUZZETTI

thing in cinema resembling a langue. However, he also denies that


the intelligibility of the cinema can be entirely explained through its
iconicity, that is, the resemblance of the sound and image to the
perceived world. Cinema is a discourse and as such must depend on
codes. Metz has tried to show that these codes are irreducibly mul-
tiple and that no one of them occupies a central or definitive place.
Yet they can be, at least initially, divided into those which are found
only in films and those which are also found elsewhere. The first
group he calls cinematic ("cinematographique") and the second
filmic ("filmique"). Among the second group there is a sort of heir-
archy, since there exist codes that cinema shares with its double,
television, and others with the novel, and still others which it em-
bodies uniquely. Metz denies, however, that this heirarchy suffices
to give any of the codes a place like that of the langue in natural
language.
This position has consequences both for the semiologic study of
the cinema and for non-semiologic efforts at theory and criticism. In
the first case, it puts a decisive end to fruitless attempts to identify
the langue of cinema, to isolate its minimal units, and to enumerate
the cinematic equivalents of the phoneme, morpheme, sentence,
and grammar. In the second, it shows the futility of trying to base
normative definitions of the cinema on arguments about the nature
of the medium. Metz is firm in his insistence that the medium, or in
his terms the matiere de I'expression, is manifest only through the
codes it will support and permit, and that it is only in terms of these
codes that cinema as a signifying system can be defined. Since no
one or group of these occupies a controlling position, the semio-
logic study of the cinema will not authorize a normative position,
thus superseding "theories" like Eisenstein's and Bazin's, which are
at best criticism in disguise.

More now needs to be said about the seminal concepts code and
systeme. The code is defined in relation to the message: "The code
is that which is not the message" (Langage ..., p. 65). The opposi-
tion code/message is parallel to systeme/texte: "The word 'systeme'
for us has no sense other than that given by its opposition to
'texte' " (Langage .. , p. 65). The systeme is the code-like configu-
ration, the formal principle of a texte. Every code is a systeme and
every message a texte, but not the reverse, since a systeme is a sin-
CHRISTIANMETZAND SEMIOLOGY 295

gular instance bound to a given texte while the code occurs in more
than one texte. Both systeme and code are abstractentities: "A code
(in theory) is defined exclusively as a relationship of logic, as pure
form" (Langage. . ., p. 165). The "relationship of logic" obtains be-
tween the elements of the paradigm,the signifiers.The systeme has
a similar definition: "What defines the systematique (that is, the
non-textual) is its characteras an ideal object constructed by analy-
sis; the systeme has no materialexistence, it is nothing more than a
logic, a principle of coherence; it is the intelligibility of the texte:
what one must suppose in order that the texte be comprehensible"
(Langage..., p. 57).
Using this vocabulary,Metz distinguishes the work of the semio-
logist from that of both film-maker and viewer. The semiologist's
task begins where the film-maker'sfinishes; it does not lead toward
the film, but away from it, toward its systemes. This movement par-
allels that of the viewer but differs in one respect: while the viewer
wishes to understand the film, the semiologist wishes to know in
addition how the film is understood ("en outre comprehendre com-
ment le film est compris").His readingis a "meta-lecture"as opposed
to the "lecture 'naive' (en fait: a la lecture culturelle)" of the viewer.
Inthe passagesquoted, Metz's own text supportstwo distinct read-
ings. Inthefirst,systeme isthe namegiven to thefactof intelligibilityin
the texte. It is the process of comprehension ("comment le film est
compris")and is picturedas the interactionof readerand texte within
the culture that they share. The semiologist makes a naive-that is,
cultural-reading and a simultaneous meta-lecture; he is the reader
who, as he comprehends, enunciates the process of comprehension.
This enunciation is the systeme or, in the case of multiple textes, the
code. Bothterms refer not to the texte, which can be isolated from its
cultural context, but to the lecture, which cannot. The meta-lecture
does not displace but rather subsumes the lecture. Therefore, the
analysisremainsculturelle,and the propositionsof the semiologist are
linked on some level to psychology, sociology, or political theory, as
Chomsky'slinguistics is to psychology.
In the second reading, the systeme is a fact logically (though not
chronologically) prior to the intelligibility and comprehension of
the texte. It is not implied by the texte but constructed ("construit")
by the analyst in opposition to the texte. To accomplish this, the
semiologist must first pass through and discard a cultural lecture in
296 ALFREDGUZZETTI

order to achieve an analytical meta-lecture, the systeme. Thus the


systeme, and the code along with it, liberated from its cultural cir-
cumstances, becomes an "objet ideal."
Despite hints of the first interpretation,it is the second that pre-
vails in Langage et Cinema. Whatever the validity of this position, it
has several inescapable consequences for the central concepts code
and systeme. First,it weakens the meaning of "intelligibilite," "lec-
ture," and comprehension by locating these processes exclusively
within the texte. Though the lecture is required and presupposed,
only the meta-lecture is expressed, or even reflected, at the level of
the achieved theory. This isolation of meta-lecture from lecture re-
quires an opposition that Metz willingly accepts of "analytique"to
"culturelle."The analytic statement, the systeme or code, can there-
fore be tested only against the texte, within which the processes of
the lecture have, somehow, been subsumed. In addressing the
texte, the semiologist can dismiss as "cultural"the problems of how
its messages were encoded and how they are to be decoded. The
process of signification thus becomes a synchronic textual fact
whose author is superfluous, his work finished, and whose "reader"
is demoted to the role of witness.
The abstract, difficult concept of a code that emerges from this
analysis can perhaps be better understood through two examples.
The first, drawn from the Essais sur la Signification au Cinema (pp.
121-146 and 105-107), is Metz's most ambitious attempt to describe
a particularcinematic code. According to his scheme, the image in
fiction film can be sorted into eight categories, or syntagms.2One of
these, the montage alterne or alternatingsyntagm, is sufficient here
to suggest the character and structure of the code as a whole. It is
defined at the level of the signifier by two types of shots, called, in
Langageet Cinema, A and B. The alternatingof the signifiersA and
B yields at least three signifieds with respect to temporal denota-
tion. It may indicate that the two signifieds are continuous in time
(e.g. alternating views of the two tennis players in a game of sin-
gles), or simultaneous (alternatingviews of fugitives and their pur-
suers in a chase), or a-temporal (the alternating shots of the
situation comedy and singing auditions in Milos Forman'sTaking
Off). The syntagm "A'-B'-A2-B2-A3-B3-etc."
is made possible by the

2The eight syntagms are described in "Film and Language, Film and Literature,"IML, II (September 1971),
154-160.
CHRISTIANMETZAND SEMIOLOGY 297

sorting of the shots into two classes according to a criterion like


"pursuerversus pursued,"that is, by a paradigmaticrelation "A/B."
This relation obtains not between shots or types of shots but be-
tween two classes defined by their mutual opposition. Its basis is
logic, not the texte of a given sequence.
The second example, described in Langage et Cinema (pp. 149-
150), is based on Bazin'sessay "Lameilleure femme ne vaut pas un
bon cheval." Bazindescribes the well-known pattern in Westerns in
which the figures of woman and horse are charged with symbolic
meanings (domesticity versus freedom, for example) and presented
to the hero for choice. This configuration, according to Metz, is a
genuine code since it is common to numerous texts and depends
not simply on the presence of a woman and a horse in the story but
on the condition of their co-presence, which qualifies the opposi-
tion woman/horse as a paradigm.The modest code thus formed,
unlike that of the eight syntagmaticcategories, is extra-cinematic,or
filmic, because it occurs in novels and songs as well as films.
Itis true by definition that a code is common to multiple textes, but
the claim that those textes maybelong to different"languages"-one a
song and another a film-requires some discussion, particularlycon-
cerning the concepts of the message and the signified. It should be
recalledthat the message is not the global meaningof the texte, but the
set of meanings produced by a single code. A texte maybe the locus of
more than one code (for example, as Metz points out, the sentence
"Voudriez-vous tenir ceci, s'il vous platt?" ["would you hold this
please?"] displays, among others, a code of phonemes and a code
of politeness) and hence of more than one message. In this respect
the message, like the code, is a function of the point of view of the
analysis.
I admit to some confusion on the question of exactly where or
what the signified and message are. In the case of the alternating
syntagm one of the possible signifieds is "simultanaeity."But it is
not, evidently, the word I have written on the page nor what was
projected on the screen to provoke it-since these can only be signi-
fiers. It is ratherthe idea of simultanaeitywithin what Metz calls the
diegesis, that is, the fictional world of the film. Metz assures us that
one cannot speak of the diegesis or its constituent signifieds di-
rectly or in isolation from their signifiers. In other words, the mes-
sage and its signifieds belong neither to the texte nor in the head of
its reader,but to the explanatoryabstractionof the lecture.
298 ALFREDGUZZETTI

Since the signified can be approached only through its signifiers,


which are, in turn, bound to a particular signifying system, how can
it be said that two or more languages may embody one and the
same code? This problem is approached through a complex argu-
ment entitled "Interferences Semiologiques entre Langages" (Lang-
age..., p. 160ff). Its premise is that a code, and hence a message, is
by definition a logical relation without physical existence. However,
this premise has a limiting condition: namely, that codes always oc-
cur in a matiere de I'expression, a physical material, which affects
and restricts their possible configuration. Thus it is impossible for a
code with temporal signifiers (though not temporal signifieds) to
exist in a matiere like painted canvas which does not extend
through time. This limiting condition is outlined by three paradigm
cases. The first, relatively inconsequential for semiology, concerns
the borrowing by one language of a fragment of a code belonging
to another. Metz's example is Faulkner's use in prose fiction of a
narrative technique resembling montage alterne, which is, as ex-
plained earlier, a fragment of the code of the image in narrative cin-
ema. The second, and most important case is the manifestation of
one and the same code in two languages (or arts). Metz's example is
the code of chiaroscuro, light and dark, in painting and color pho-
tography. These languages, like any two, necessarily differ with re-
spect to their sensory base, or matiere de I'expression. The critical
point-and the one that distinguishes this category from the third-
is "that the differences between the two matieres concern traits not
pertinent to the code in question" (Langage, p. 163). The third cate-
gory fails to meet this condition and hence is a weaker version of
the second; it concerns two languages (or arts) whose differences of
matiere will not permit the manifestation of the same code, but
only of similar, or "isomorphic," codes. For example, the code of
light versus dark may be shown directly in a painting, but in a novel
its signifiers must be translated into words, which render the code
in a similar, though by no means identical, form.
Can the condition of the second category be met in reality? I have
difficulty considering this in the context of Metz's example, since I
do not know precisely what the code of chiaroscuro is, unless it is
the opposition between light and dark. It seems true enough that
such an opposition is possible equally in a color photograph and a
painting. Yet the two languages cannot have the same signifiers,
since analine dye is not the same as oil paint, nor therefore the
same signifieds. Metz's contention, I think, would be either that the
CHRISTIANMETZAND SEMIOLOGY 299

differences between these matieres would not truly change the sig-
nifiers (any more than speaking French with an American accent
does) or that the opposition of signifiers, which is logique relation-
nelle, remains identical despite changes in the signifiers. This iden-
tity, however, is postulated, rather than proved; hence the argument
is circular. As logic, codes may be exempted from the material con-
ditions which embody them, but is it true that one may perform
analogous surgery on the mati'ere de I'expression, taking from it
only the traits relevant to the codes? Standing before a particular
painting and a particular color photograph, is it possible to regard
the two systems of light-dark opposition apart from their material
context? This may be conceivable as a mental operation, but is this
operation other than a thesis whose meaning consists in a contin-
uing relation and return to the matiere? In the absence of this rela-
tion, how is the code capable of signifying? If the code is severed by
definition from the matiere, this does indeed authorize the erection
of the second category, but does it not at the same time annul its
power as an analytic tool?
Perhaps a more central question is why Metz needs such a cate-
gory in the first place. His position in the Essais is that cinema is a
langage without a langue, and his argument throughout is keyed to
this denial. Apart from a few hesitant hypotheses, the work is unre-
mittingly critical and distrustful of any wide-reaching claims for the
role of codes in cinema. Cinematic imagery is "a rich message with
a poor code, a rich texte with a poor systeme" (Essais, p. 74). Or
"Goodbye, paradigm! Its poverty is the counterpart of a richness
distributed elsewhere: the film-maker, different in this respect from
the speaker of a language, can express himself by showing us di-
rectly the variety of the world; in this way the paradigm is quickly
overwhelmed" (Essais, p. 75).
Langage et Cinema gives these denials the positive terms of a
theory by pursuing to its conclusion the definition of cinematic lan-
guage as an "ensemble" of codes. To do'this, the three categories of
"interference" are by themselves insufficient, since they only serve
to indicate the further problem of describing the limiting condi-
tions imposed by the matiere de I'expression. Metz's response is to
classify the languages bordering on cinema-still photography,
painting, the photo-roman, comic strip, radio and television-ac-
cording to such traits of matiere as iconicity, temporality, and prov-
enance of the image. The resulting taxonomy, in conjunction with
the crucial second category of "interference," suggests a method of
300 ALFREDGUZZETTI

defining the relations between languages in terms of their shared


codes. In this way, it is possible to identify the set of codes common to,
say, cinema and the photo-roman and distinguish it from the set
common to cinema and still photography. This network of inter-rela-
tions replaces the dichotomy "specific/non-specific," which is the
descriptive basis for such normative statements as "Cinema is the art of
movement." The analysis also implies that the "ensemble" of codes
comprising a given language is not simply an inventory, but has a
structure that may, Metz suggests, itself be code-l ike. Thus, one could
complete the definition of cinematic language bywriting the "formule
codique de sa specificite"-a prospect that both resolves and reflects
the skepticism of the Essais concerning the explanatory power of
codes regarded individually. However, Langage et Cinema stops short
of attempting such a definition since Metz's position is sufficiently
established simply by indicating the possibility.
* * * * *

Is this theory truly without norms? To approach this question, I


want to sketch an alternative explanation of the syntagm of the film
image and compare Metz's to it. In my account, the telecast image
of a football game may stand as a paradigm case. In order to under-
stand what he sees, the viewer must know that the syntagm on his
TV screen alternates among the views of a small, fixed number of
cameras. It is-with apologies to Chomsky-a surface-structure
whose deep structure may be visualized as the configuration of
monitors in the control room, each of which displays the view of a
single camera. It is unnecessary for the viewer to have the slightest
knowledge of the circumstances of production in order to under-
stand either this structure or the rules that link it with what appears
on his screen at home-just as the native speaker of a language is
said to know its deep structures even if he cannot describe them. The
proof that this understanding is at work is that when the program
deviates from the rules that transform the deep structure into the
surface structure, this must be signalled; for instance, an instant replay
must be identified either with superimposed titles, spoken comment,
or the clear presence of slow motion, so as not to be confused with the
live transmission; otherwise, the telecast lapses into unintelligibility.
The film-image is a more complex, but similar case. Again, the
structure has two layers: one which appears on the screen, and its
deeper counterpart, which may be represented by the unedited
footage. Here too it may be said that the viewer understands the
CHRISTIANMETZAND SEMIOLOGY 301

deeper structure even if he knows nothing about the production of


motion pictures. If he did not infer it while watching films, he
would, like certain early audiences, be confused and disoriented by
every disjuncture of time and space, even when they are masked by
the continuity of action within the diegesis. To understand the
deep structure means, in part, to know that the syntagm is not a
transmission, but an assemblage of pieces filmed discontinuously in
time and space. This knowledge is reflected in the word "cut," which
signifies not simply a change like the switch between television cam-
eras, but a joint. To object that the deep structure is here a superfluous
concept because the viewer only "knows" the portions of it that reach
the surface is to miss the point; the viewer's ability to comprehend
consists of his knowledge of the rules that connect the two. The
lecture, therefore, in certain respects parallels the composition of the
film, rather than occurring after it, as in Metz's model.
To take the simplest example, consider a dialogue scene in which
bits of two camera-shots, or takes, A and B, alternate. Undoubtedly
the viewer infers, as Metz maintains, on the basis of continuity in
the dialogue that the alternating image signifies continuity in time.
However, neither this observation nor its implied oppositions be-
tween temporal continuity and simultanaeity and between alternat-
ing and linear syntagms accounts for the structure of alternation. It
explains, so to speak, how the structure is absent, but not how it is
present. One must add that the viewer also understands the prove-
nance of the alternating shots, the model of which is the unedited
takes A and B. If this were not so, dialogue scenes articulated by
jump-cuts, like those in Breathless, would be, at least for the origi-
nal audience, not simply novel or shocking, but unintelligible, or-
what comes to the same thing-they would require a new category
in the taxonomy of the image-syntagm. In fact, they are intelligible
because their relation to the deep structure is intelligible. The role
of this deep structure in the specific case of the alternating syntagm
is demonstrated clearly by the "dialogue" scene in Persona in which
the camera-takes A and B, which show in one-shot Liv Ullmann and
Bibi Andersson respectively, are not intercut and do not alternate,
but run in their entirety one after the other: A, then B. This repeti-
tion in the diegesis does not, however, impair the intelligibility of
the sequence. Rather, its effect is to point to and illuminate the
deep structure in a way that is not only locally intelligible, but in-
dicative of the director's concern elsewhere in this work with the
processes of cinema and representation.
302 ALFREDGUZZETTI

Metz is also aware that exceptions test the rule, and in the Essais
(pp. 213-215) he cites an image-sequence from Pierrot le Fou which
does not fit any of his eight categories; it is the sequence, consisting
of many short shots, in which the protagonists, Jean-Paul Belmondo
and Anna Karina, flee Paris. For Metz, the problem is two-fold: first,
the shots are out of chronological order, though the glimpses of the
action that they permit allows the viewer to reconstruct the likely
chronology, and second, they show some bits of the action more
than once and in slightly differing versions. Metz takes the se-
quence to be an instance of the diachronic processes of history at
work, altering and adding to one of the codes of cinema. His re-
sponse as a semiologist is to give the excerpt the provisional name
of a "potential" syntagm, since it allows neither a definitive nor
chronological version of the action.
Whatever the paraphrasable meaning of its structure, the se-
quence is intelligible. As in the much simpler scene from Persona,
the viewer is referred to the deep structure of the shots, which is, as
always, open to an organization that contradicts the chronology of
the diegesis. Godard goes further than Bergman, however, when he
includes corresponding bits from different camera-takes in which
the action happens to differ slightly. Though few viewers would de-
scribe this process as I have, in terms of shooting and cutting, every-
one would, I think, recognize the director's gesture toward the deep
structure and would understand that our normally unconscious
knowledge of it is here being invoked. Presumably Godard makes
such a gesture because at this moment he sees no alternative but to
ask us to participate in his work of examining the world by means of
the cinema, not because he is unsure how things happened and can
give only a "potential" account. For after all, nothing "happened";
the story is made up, and its author would seem intolerably pre-
tentious if he demanded that we forget this fact in the midst of a
sequence that so clearly shows his hand.
The differences between my account of this sequence and Metz's
suggest, among other things, the extent to which his thinking is
bound to the narrative film. In all of his writing, I could not cite a
single example from another source. In the Essais (pp. 96ff.) he de-
fends the historical basis of this commitment. His recognition of the
relevance of this diachronic question is remarkable in light of his
usual contention that his is an exclusively synchronic discipline. His
attitude elsewhere toward writing history is the same as toward the
creation of individual textes: the semiologist begins his work where
the film-maker, and likewise the film-historian, finish theirs. He
CHRISTIANMETZAND SEMIOLOGY 303

minimizes the importance of the moments at which the work of


production, history, and semiology must touch a common point; he
maintains, for example, that the semiologist's choice of a texte for
study may be influenced by the problems currently seen as urgent
within his discipline, but is in itself as indifferent as the linguist's
decision to study a particular dialect; and, as I have tried to show,
his notion of signification neither incorporates nor permits refer-
ence to the composition of the texte.
The version of film history given in the Essais stresses the central-
ity of "la rencontre du cinema et de la narrativite," the meeting of
cinema and narrative. It holds that in the beginning film was purely
iconic-that is, it signified exclusively by means of the resemblance
between its imagery and the visible world-and that its association
with narrative resulted in the gradual invention and super-
imposition of various codes over its iconic mechanisms. The reper-
tory of codes pertaining to the image was complete by the time of
D. W. Griffith, and those of image and sound formulated during the
thirties, stabilizing in the style Bazin called "decoupage classique."
In this view, the Soviet school of the twenties, despite the magnifi-
cence of its production, is a cul-de-sac. Metz is clearly hostile
(though respectful) both to Eisenstein's attempt to give montage
the status of a "langue" (though, of course, Eisenstein would not
use the term) and to his reading of film history, which groups Grif-
fith with the Soviets as the dominant school. In short, Metz's history
is like Bazin's, except that he does not acknowledge Bazin's charac-
terization of the two warring tendencies in film history. Nor does he
concede that the reliance of his synchronic theory on this version of
history in any way compromises its claim of objectivity.
What is at stake here is not simply the importance of narrative
film but the controlling place of narrative within film theory. The
contrast between Metz's account of the articulation of the image
and the alternative that I sketched should make this clear. The basis
of his analytical method is to refer the structure of the image to the
narrative. In cases where the practice of the cinema itself challenges
the explanatory power of the narrative, as does the sequence from
Pierrot le Fou, his response is to avoid the problem by inventing a
new narrative category. The history of film as he rehearses it conve-
niently minimizes examples of this kind and at the same time au-
thorizes his methods of dealing with them. Supported by this
version of history, his synchronic theory affirms a notion of signifi-
cation in which the spectator, in order to understand the structure
of the image, refers not to the real world where films are financed,
304 ALFREDGUZZETTI

shot, cut, printed, sold, projected and criticized, but to the fantasy
world that the films present-what is clinically called the "diegesis."
At first glance, this tendency is less clear in the vastly more ab-
stract text of Langage et Cin&ma. There (pp. 142-143) Metz denies
that the syntagmatic categories of the image in narrative film merit
the status of a langue, or even an especially important code, of the
cinema. The denial harmonizes with the book's argument that the
codes of cinema form an "ensemble," to which the notions of hier-
archy, centrality, or specificity are simply not applicable. On the sur-
face, this seems a neutral, a-historical, and objective definition. Yet
on examination it turns out, I believe, to be only the position of the
Essais slightly modified and projected into more abstract terms.
To support this contention, let me reopen the discussion of the
crucial second category of "interference," the instance of one code
in two languages, by adducing an example from the Essais. Though
in context the example relates to some notions of Pasolini's that
seem remote from "interference," it presents issues that are, I think,
easier to grasp than those of the light-dark code in paintings and
color photographs:
Certainly, the total comprehension of any film would be impossible if
we did not carry within us this vague but quite real dictionary of "im-
segni" of which Pasolini speaks, if we did not know-to take only one
example-that Jean-Claude Brialy'scar in Les Cousins is a sports-car, with
everything that this signifies in twentieth-century France, the diegetic
period of the film. But we would know all the same, because we would
see it, that it is a car, and this would be sufficient for the comprehension
of the (tenotecd meaning of the passage. One may not object that an
Eskimo with no priorcontact with industrial civilization would not even
recognize it as a car! For what that Eskimo would need is acculturation,
not translation .... (Essais, pp. 209-210)
To analyze this example in terms of Metz's scheme of "interference,"
we may note that the sports-car belongs both to a "language" of
consumer items in twentieth-century France, like those described by
Roland Barthes in Mythologies, and to the filmic language of Les
Cousins. The matieres de I'expression of these two languages, in the
first case the visual, three-dimensional presence of the real objects
and in the second their two-dimensional image on film, do not differ
in the traits relevant to the code in question. For example, the para-
digm sports-car/sedan, in which the sports-car signifies luxury, is
possible equally in twentieth-century France and in a film. Like most
of the visual codes of culture, the one that includes this paradigm can
be rendered in the iconic terms of the cinematic image, and thus is an
instance of the second category of "interference": the manifestation
of one and the same code in two languages.
CHRISTIANMETZAND SEMIOLOGY 305

The argument that establishes this instance also-and more im-


portantly-opposes the notion that Metz calls "cinematographisa-
tion." His analysis of "interference"implies that a code like the one
of our example does not become "cinematic" by virtue of its mere
presence in a film; it is an instance of the second category, not the
third. The sports-car in Les Cousins not only has the significance it
would also have in twentieth-century France-at least in part-but
obtains this significance by the same means: "Thus is it necessary to
insist on the fact that filmic code is not necessarily cinematic, since
a code (in theory) is defined exclusively as a relationshipof logic, as
pure form, and is therefore not 'ied to a particular matiere de
I'expression, for example, the cinema's" (Langage..., p. 165).
Although it is a subtle and difficult matterto say how looking at a
painting differs from looking at a color photograph, everyone can
tell the difference between a sports-carand a picture of one. Metz
does not deny this difference. Rather,he passes over it in silence,
speaking only about the place the object occupies in a discourse.
This silence is the counterpartof his contention elsewhere that one
can sever the code from its mati'ereand hence that none of its
meaning consists in, or arises from, its situation in the matiere. Here
the difficulties begin. For-to return to the example-when we are
before the actual automobile, the paradigm sports-car/sedan and
the code which subsumes it can hardlybe said to inhere exclusively
in "logique relationnelle," but belongs at least partly to the con-
crete relation between the object and us. We are in the world
where we may not address the object except as, say, an owner, cus-
tomer, envier, producer or viewer-all relations in which the code,
admittedly abstract, is at work. None of these relationshipscan en-
ter unaltered into a film, where there is no longer the possibility of-
to keep to the example-an economic relation between us and the
object, but only the quite different opportunity to project this rela-
tionship onto, or extractit from, the world that is represented.
Yet filming is not simply the loss of a certain sort of relationship.
The object filmed, it is true, is in one sense reduced to a phantom.
But at the same time it becomes another sort of object-a picture.
The code that describes our relation to an actual automobile has a
proper analogy only in our relation to the commodity we call a
movie, not to what is signified by the movie. Going to the movies
and buying a sports-carcan even be said to form a paradigm,since
they have, as an opposition, the power to signify, and there exist
situations in which such a choice may be read as a discourse. I do
306 ALFREDGUZZETTI

not mean to insist on the economic basis of this analysis (it is only
an illustration chosen to suit Metz's example), but rather on the
centrality in signification of the relation between the viewer and
the matiere, from which the code as a signifying mechanism is not
detachable. Consequently, all codes, even those of natural lan-
guage, are "cinematographized" when they pass into films; hence
Metz's second category is without true examples.
This argument does not deny the points of similarity between,
say, a film and a song that share the theme "The best woman isn't
worth a good horse." Rather, it refuses such similarities the status of
language, as that term is understood by Metz. This disagreement is
reproduced and elaborated in Metz's critique of Pasolini's semiot-
ics, from which the discussion of Les Cousins is taken. Pasolini
maintains that cinematic language has two articulations, the second
more or less equivalent to Metz's specific codes, and the first a
rather ill-defined and partly to-be-invented lexicon of "im-segni"
(image-signs), which have meanings within culture and therefore
comprise a kind of language prior to the intervention of the film-
maker. Metz argues that this first articulation is not only question-
able but superfluous, since it can be adequately replaced by the
concept of iconic analogy; he insists on "the perceptual and cul-
tural status of these 'im-segni,' opposing it to the language-like
character of the codes that are properly cinematic" (Essais...,p.
210n).
I do not wish to associate myself with Pasolini's position, least of
all with the problemmatical concept of the "im-segno." But I do
sympathize with Pasolini's motive, which is, I think, less to intro-
duce economies into the semiologic study of cinema than to place
its most fundamental premise, and hence the discipline itself,
within a perceptual, cultural, and ultimately political context. Metz
is not only unpersuaded of the necessity of this step, but sees an
antithesis between the terms "cultural and perceptual" on the one
hand and "language" on the other.
In thus opposing the signifying process to its cultural and percep-
tual setting, Metz exposes the basis of his conception of the cin-
ema. Though his approach to particular questions is always
distinguished by the sophisticated application of linguistic method-
ology, the effect of these methods is to attribute to the cinema a
sort of transparency, to imagine it as a glass distinct from culture but
through which culture is visible. Cinema emerges in his portrait as
an instrument that is as indifferent to ideology and value as he sup-
CHRISTIANMETZAND SEMIOLOGY 307

poses natural language to be, as a system whose workings are logi-


cal rather than material and whose history is the immaterial
accretion of these processes in the realm of pure logic and pure
form. It is a conception that, despite the linguistic vocabulary in
which it is stated, recalls Bazin. For Bazin bequeathed to Metz not
simply the question What is Cinema? but an answer of his own,
namely that cinema is an image and substitute for reality, whose
history is the gradual aesthetic and technological perfection of the
power of illusion. Metz rejects the wishful, almost religious aura of
this position. He discards its normative arguments, fills in its si-
lences on the structureof cinematic discourse, and banishes its aes-
thetic of illusionism if not from the description of particularcodes,
then at least from the higher, more abstractlevels of the theory. Yet,
despite all this, he preservesand elaborates Bazin'sassumption that
the cinema is something one looks through but not at and studies
in isolation from the circumstanceswhere it is made and seen.
This shared assumption indicates what is truly at issue in Metz's
distinction of his work from that of other theorists and critics. For
example, Eisenstein's concept of montage, which Metz patron-
izingly calls "montage-roi,"is, unlike any of the seminal concepts of
Metz's theory, rooted in the materialworld of culture and percep-
tion. It locates the process of signification in the interaction of
spectator and screen. The relation of frame to frame, of image to
image, and of the parts of an image to each other is not merely a
fact of the texte, but a relation which the film-maker creates and
which produces ideas, emotions, and comprehension only by acting
on, and being re-enacted by, the viewer. In a comparable way, Paso-
lini, whose conception is in other respects close to Metz's own,
maintainsthat from the outset the discourse of cinema is entangled
in its cultural origins, from which not even the image, for all its
iconicity, can be entirely extricated.
Faced with such profound and consequential disagreements, no
reader can ignore either the need for judgment or the search for
criteria on which a judgment may be based. Despite the subordi-
nate place habitually given to the discussion of particularfilms and
partsof films in Metz's writing, and despite his repeated denial that
such discussions are either urgent or germane to his work as a
semiologist, I would argue that it is not only inevitable but fair that
they should stand as a measure of his theory. Must one not ask what
light is shed on the sequence from Pierrot le Fou by giving it the
name of "potential" or by testing it against the eight supposed syn-
308

tagmatic categories one at a time? Are we told any more about The
Outlaw by calling its theme of woman-versus-horse a code? Or can
any plausible rationale be conceived for preferring the long analysis
of Intolerance in Langage et Cinema (pp. 81ff.) to Eisenstein's sup-
posedly unscientific and normative discussion of the same work in
"Dickens, Griffith, and the Film Today"? Can one deny that Metz's
detailed description of Griffith's parallel montage is simply less than
Eisenstein's effort to relate it to the film's ideology, than to submit
that ideology to a critique based on a reading of history? Whatever
the talent of these two men as critics, and whatever one's attitude
toward what they value, is there any possible conclusion but that
Metz's ideas produce what is by comparison to Eisenstein a diminu-
tion and abortion?
But Eisenstein is not to be taken as the measure of Metz's writing.
The point is rather that Metz's effort to excise from cinema the
process of signification and, in the name of language, to set it in a
realm of abstract logical relations is in the end both meaningless
and futile. The analogy between natural and cinematic language is
not perfect enough in itself to authorize or sustain such an effort.
Cinema as a signifying system lacks, as Metz concedes, both arbi-
trariness in Saussure's sense and the paired systems that permit both
distinctive and significative articulation. Moreover, it is, unlike nat-
ural language, the outcome of industrial production-a fact with
inescapable consequences for its means of signification. When the
world, whether coded or not, passes into the sounds and images of
cinema, it is not merely reduced to the configurations of an abstract
system, but transformed into a concrete object of another kind. The
difference between these sounds and images and the objects and
codes that they reflect, though it has no stable meaning, nonethe-
less always and of necessity enters into the process of signification.
It is not within the power of the film-maker's intentions, or of their
reflection in the film's style, to dissolve this difference. Even if it
were, we as spectators are not simply prisoners of those intentions,
but are free and co-present in the world where the images and
sounds exist as objects. The conditions of this world, its social, eco-
nomic, and political structure, impose on us, spectators, critics,
theorists, and film-makers alike, the obligation to analyze, interpret,
and criticize the cinema as it reflects and embodies these condi-
tions. In this light, Metz's retreat to the realm of logic and abstrac-
tion that he calls "language" can be seen as not simply anti-
normative, but, despite its methodological precision, anti-critical
and anti-analytical as well.

You might also like