Buckland - Cognitive Semiotics of Film

This page intentionally left blank
The Cognitive Semiotics

of Film
WARREN BUCKLAND
Liverpool John Moores University

The Pitt Building, Trumpington Street, Cambridge, United Kingdom

The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcn 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
Warren Buckland 2004
First published in printed format 2000
ISBN 0-511-03472-5 eBook (Adobe Reader)
ISBN 0-521-78005-5 hardback
For Thomas Elsaesser
CHAPTER ONE
The Cognitive Turn in Film Theory
We have witnessed a number of attempts to by-pass [film theorys]

most difficult conceptual problems by replacing it with something
else. The something else is sometimes film history or aesthetics;
sometimes it is a new object, such as television, popular culture,
video; and sometimes it is a question of new methodologies, which
may resemble dusted off methodologies from the social sciences,
such as audience questionnaires or interviews, procedures that
havent benefitted from the literature in the social sciences that has
interrogated its own methods and limitations. (Janet Bergstrom)1
uring the eighties, film studies gradually adopted

new methodologies from cultural studies and the social sciences, which displaced the speculative ideas of film theory.
Rather than construct hypotheses and models about the general
structure and spectators experience of film, film studies has moved
toward the something else enumerated by Janet Bergstrom. However, a number of film scholars, in both Europe and North America,
have persisted with film theorys most difficult conceptual problems, which they tackle from the perspective of cognitive science.
This book is a report on the knowledge generated by these cognitive film theorists. But because this knowledge is fragmentary and
incomplete, I have endeavored to expand and develop it in new
and unforeseen ways.
However, for the most part, I do not report on the knowledge
generated by the well-known cognitive film theorists in North
America (David Bordwell, Noel Carroll, Edward Branigan, Joseph
Anderson, among others) but discuss the much lesser known film
theorists working in the cognitive tradition in Europe particularly
Francesco Casetti, Roger Odin, Michel Colin, and Dominique Chateau.2
Despite their similarities, the two groups evidence a marked
1
THE COGNITIVE SEMIOTICS OF FILM
contrast in their work: Whereas the North American cognitivists

decisively reject the basic doctrines of modern film theory (a.k.a.
contemporary film theory, based upon structural linguistics, semiotics, Marxism, and psychoanalysis), the European cognitivists
inaugurate a revolution in modern film theory by returning to and
transforming its early stage that is, the semiotic stage.3 Both
groups therefore reject psychoanalysis and replace it with cognitive
science. However, the European cognitivists assimilate cognitive
science into a semiotic framework, whereas the North American
cognitivists work within a pure cognitive framework (one untainted by semiotics).
Treating the work of a group of individuals as representing a
homogeneous position is always risky. Nonetheless, all the North
American cognitivists I have named belong to the Institute for
Cognitive Studies in Film and Video, which to some extent unifies
the agenda of the individual authors.4 What unifies the European
cognitivists is that their work critically responds to Christian Metzs
film semiotics. This response involves transforming Metzs semiotics by means of theories of pragmatics, cognitive science, and transformational generative grammar (which is in fact one of the main
research programs in cognitive science). The European cognitivists
attempt to overcome the translinguistics of Metzs film semiotics
that is, Metzs insistence that film semiotics be based exclusively on
the methods of structural linguistics by combining semiotics with
pragmatics and cognitive science. Structural linguists overemphasize languages rigid, limiting capacity, and a semiotics
based exclusively on structural linguistics conceptualizes all other
semiotic systems in a similarly rigid manner limiting and conditioning the meaning of human experience at the expense of the
language users reflective and creative capacities to manipulate
signs. By combining semiotics with cognitive science, the European
cognitivists restore the balance and begin to conceptualize natural
language and other semiotic systems as both enabling and limiting.
Because of the dual emphasis in the work of the European cognitivists on semiotics and cognitive science, I shall call them the cognitive film semioticians.5 Figure 1 shows the relations among the
classical film theory of the 1930s1950s, modern film theory, the
North American cognitivists (from now on, simply the cognitivists), and the cognitive film semioticians.
In this book I aim to outline the common theoretical assump-
THE COGNITIVE TURN IN FILM THEORY
1. CLASSICAL FILM THEORY

(a) Montagists (Rudolf Arnheim, Sergei Eisenstein, etc.)
(b) Realists (Andr Bazin, Siegfried Kracauer, etc.)
2. MODERN FILM THEORY (a.k.a. contemporary film theory)
(a) Film semiotics (Christian Metz of Film Language, Language
and Cinema)
(b) Post-structural film theory (a.k.a. second semiotics, psychosemiotics): Marxist and psychoanalytic film theory of
Stephen Heath, Colin MacCabe, Metz of The Imaginary
Signifier, Jean-Louis Comolli, Jean-Louis Baudry, Raymond
Bellour, etc. (the transition from 2a to 2b was effected by
theories of enunciation based on the linguistics of
Benveniste)
3.
COGNITIVE FILM THEORY
David Bordwell, Nol Carroll, Edward Branigan, Joseph

Anderson, Torben Grodal, Ed Tan, Murray Smith
4. COGNITIVE FILM SEMIOTICS (development from 2a)
(a) New theories of enunciation (Francesco Casetti, Metz of The
Impersonal Enunciation)
(b) Semio-pragmatics of film (Roger Odin)
(c) Transformational generative grammar and cognitive
semantics of film (Michel Colin, Dominique Chateau)
tions held by cognitive film semioticians and clarify their relation

to the broader traditions of twentieth century intellectual thought.
Cognitive film semiotics represents the next stage and arguably
the maturation of semiotic film theory. Despite the revolution it
has inaugurated, cognitive film semiotics remains virtually unknown in Anglo-American film studies. This is unfortunate because
it develops a more informed understanding than either semiotics
or cognitive science alone of films underlying structure, together
with the way spectators comprehend films. By writing this book I
hope to introduce cognitive film semiotics to the Anglo-American
community of film scholars and, more generally, encourage a reevaluation of the role of semiotics in film theory.
Before outlining cognitive film semiotics, I shall briefly review
the cognitivists position, particularly their reasons for rejecting
linguistics and semiotics as viable paradigms for studying film. I
Fig. 1
shall also attempt to point out several problems with their purely
cognitive-based film theory.
To what extent is the dispute between modern film theory and
cognitivism based on conceptual disagreement, and to what extent
is it simply based on misunderstanding? Briefly, I shall argue that
the cognitivists criticism of the psychoanalytic dimension of modern film theory is based on conceptual disagreement and, moreover,
that this disagreement is partly justified. However, I shall also argue that the cognitivists critique of the linguistic and semiotic
dimensions of modern film theory is based on misunderstanding,
which has led them to refute its premises falsely.
If film theory is to make any advances, it needs to establish the
grounds for disagreement among its various schools and must
identify misunderstandings. Peter Lehman argues that scholars
should develop a dialogue with other scholars. He asks: How do
we teach students to respectfully argue with the perspectives of
their peers or teachers if the materials that they read encourage
them to dismiss those critical methodologies and film styles with
which they are not in agreement? And: Students should also
realize that what they can learn from someone may have little or
nothing to do with their agreement with that persons methodology
or critical judgement.6 Similarly, Noel Carroll argues that film
theorizing should be dialectical, adding: By that I mean that a
major way in which film theorizing progresses is by criticizing
already existing theory. Some may say that my use of the term
progresses here is itself suspect. However, I count the elimination
of error as progress and that is one potential consequence, it is to
be hoped, of dialectical criticism. Of course, an even more salutary
consequence might be that in criticizing one theoretical solution to
a problem, one may also see ones way to a better solution.7
Carrolls recent position is to develop a dialogue with, rather than
simply condemn, previous theories of film.
In the following review of cognitivism, I do not aim to be dismissive, but to be critical. This involves clarifying misunderstandings so that we can leave behind us the old disagreements and
make advancements by tackling new disagreements.
The cognitivists find very little of value or interest in modern
film theory, although in Narration in the Fiction Film Bordwell acknowledges the value of some early semiotic work, such as Christian Metzs grande syntagmatique.8 Yet Bordwell undermines this
acknowledgement in Chapter 2 of the same book when he asks the

following questions:
Why . . . is the employment of linguistic concepts a necessary condition of
analyzing filmic narration? Is linguistics presumed to offer a way of subsuming film under a general theory of signification? Or does linguistics
offer methods of inquiry which we can adopt? Or is linguistics simply a
storehouse of localized and suggestive analogies to cinematic processes?9
I shall take each question in turn. Moreover, I shall use my responses as an opportunity to review the previous research carried
out in the name of film semiotics.
1. Why . . . is the employment of linguistic concepts a necessary
condition of analyzing filmic narration?
The simple answer is that the employment of linguistics is not
necessary to the analysis of filmic narration. Bordwell is right to
criticize Metzs translinguistic standpoint. Metz initially made the
mistake of arguing that linguistics is a necessary condition for analyzing filmic narration because he equated film language with narrativity: It is precisely to the extent that the cinema confronted the
problems of narration that . . . it came to produce a body of specific
signifying procedures.10 However, he challenged this equation in
Language and Cinema,11 a book that marks the maturation of his
semiotic thinking on film. Perhaps we could turn this question back
to Bordwell and ask, Why is his historical poetics of cinema predominately a poetics of narration?12
2. Is linguistics presumed to offer a way of subsuming film under a
general theory of signification?
The short answer to whether linguistics subsumes film under a
general theory of signification is yes. To think of film within a
general theory of signification has many consequences, several of
which I shall outline.
Film semiotics is a project that does not consider film to be an
unproblematic, pregiven entity, but reflects on the very nature of
films existence, together with the consequences it has on culture
and society. Semioticians challenge the commonsense ideological
understanding of film as a mere form of harmless entertainment,
maintaining that it is a system of signification that articulates experience. This is a relevant framework in which to examine film be-
cause the more complex a society becomes, the more it relies upon
systems of signification to structure, simplify, and organize experience. The fundamental premise of semiotics is that the whole of
human experience, without exception, is an interpretive structure
mediated and sustained by signs.13 Semiotics offers an allembracing theory of human culture or, more precisely, of human
experience, belief, and knowledge. It is a theory in which humans
are posited to have an indirect mediated relation to their environment. I will argue that natural language plays a decisive role in
this process of mediation, of enabling individuals to control and
understand their environment. But natural language is not allencompassing, for human culture consists of numerous other semiotic systems such as film that also mediate between individuals
and their environment. Perhaps it is relevant here to note that my
discussion is limited to anthroposemiotics (the study of human
signs) and does not cover zoosemiotics (the study of animal communication), although both are united under biosemiotics (the
study of communication generated by all living organisms). Linguistics, the study of natural language, is one of the dominant
branches of anthroposemiotics but has a very small role to play in
biosemiotics and is not involved in zoosemiotics.
Studying film from a semiotic perspective does not involve comparing it to natural language (although this is one of the secondary
consequences of conducting a semiotic analysis of film), but involves first and foremost analyzing films specificity. In film semiotics, specificity is defined in terms of the invariant traits manifest
in all films, the traits that confer upon film its distinctiveness, which
determines its unique means of articulating and mediating experience. Film semioticians define specificity not in terms of films invariant surface (immediately perceptible) traits, but of its underlying (non-perceptible and non-manifest) system of invariant traits.
This semiotic perspective opposes the work of the classical film
theorists, who also studied filmic specificity. However, they defined
specificity in terms of films immediately perceptible traits, a focus
that resulted in their formulating two mutually contradictory theories of filmic specificity. Rudolf Arnheim argued that filmic specificity lies in unique distorting properties (especially montage) that
demonstrate films specific representation of perceptual reality its
presentation of a unique perspective on reality. However, Andre
Bazin argued that its specificity lies in the ability for the first time
in the history of art to record reality without the intervention of

the human hand (that is, he argued that films specificity lies in its
existential link to reality). He advocated that filmmakers not subsume films ability to record under distortive techniques such as
montage. Instead, he advocated a style of filmmaking to exploit the
recording capacity of film such as the long take, deep focus, and
camera movement which maintains the films existential link to
reality. Metz sought to surmount these two mutually contradictory
theories by defining specificity in terms of films underlying system
of invariant traits. To present an understanding of what underlying
system of invariant traits means and how it enabled Metz to surmount the contradictions of classical film theory, I need to give an
overview of semiotics.
Semiotics is premised on the hypothesis that all types of phenomena have a corresponding underlying system that constitutes
both the specificity and intelligibility of those phenomena. The role
of theory in semiotics is to make visible the underlying, nonperceptible system by constructing a model of it. A model is an
independent object which stands in a certain correspondence with
(not identical with, and not completely different from) the object of
cognition and which, being a mediating link in cognition, can replace the object of cognition in certain relations and give the researcher a certain amount of information, which is transferred by
certain rules of correspondence on the object of modelling. The
need for a model arises when for some reason immediate analysis
of an object is inexpedient or impossible.14
The first step in developing a semiotic film theory is to construct
a model of the non-perceptible system underlying films, which
involves identifying the properties and parts of this underlying
system, together with the way they interrelate and function. The
resulting model is expressed in a series of hypotheses, or speculative propositions. These propositions are not obviously true or false
but are probable. The validity of these probable propositions and
the models they construct is dependent on both internal and external criteria. Internally, hypotheses and models must display logical
consistency. Externally, they must be able to analyze existing phenomena and predict the structure of new phenomena. Semiotic
film theory can be validated or invalidated on the basis of its logical
(in)consistency, as well as its (in)ability to attribute structure to a
given or new film which involves relating the film to the semioti-
cians prior model of the underlying system. In other words, external validity is dependent upon the models possession of generality
its ability to be applied to all phenomena, given and new.
Metz attempted to construct a general model of the system
underlying all films. His first model, to be discussed, is the grande
syntagmatique; his second, developed in Language and Cinema, attempts to define filmic specificity in terms of a specific combination
of five overlapping traits iconicity, mechanical duplication, multiplicity, movement, and mechanically produced multiple moving
images.15 Taken individually, Metz realized, none of these traits is
specific to the cinema; the specificity of cinema, he argues, lies in
their specific combination. These five traits are not simply heaped
together but are organized into a particular system, which Metz
models in terms of overlapping circles, similar to a Venn diagram
(although Metz does not go so far as to visualize this model; this is
what I have done in Figure 2). Filmic specificity for Metz consists
of the five traits and of the system that organizes them. Notice that
Metz does not draw any direct comparisons between film and natural language in this semiotic model of film. Although it is possible
to question the logical consistency of Metzs mode of reasoning in
Language and Cinema, my aim in discussing this book is simply to
outline the semiotic model Metz developed there. The primary
problem with this model is its generalizability, because it leaves out
some avant-garde films that do not employ mechanical duplication
(for example, the films of Len Lye) and films that do not employ
movement (the most celebrated example is Chris Markers La Jetee).
Like other semiotic studies, film semiotics adopts the two tier
hierarchy between perceptible and non-perceptible levels of reality
and formulates probable hypotheses describing this underlying,
non-perceptible level. The ultimate objective of film semiotics is to
construct a model of the non-perceptible system underlying all
films. Whereas Saussure called the specific underlying system of
natural languages la langue, in opposition to the surface phenomena, la parole, Noam Chomsky calls the underlying system competence, in opposition to performance, and for Metz, the specific underlying system of film is called cinematic language, in opposition
to individual films.
The function of a model is therefore to mediate between a theory
and its object of study. Semioticians do not commit the fallacy of
identifying the real object with the object of knowledge because
Fig. 2
1. Iconicity; 2. mechanical duplication; 3. multiplicity; 4. movement;

5. mechanically produced multiple moving images.
they realize that each theoretical framework does not discover its
specific object of study but must construct it, precisely because the
object of study is inaccessible to perception. Saussure realized this
in relation to the specific object of semiotic study: The object is not
given in advance of the viewpoint: far from it. Rather, one might
say that it is the viewpoint adopted which creates [cree] the object.16 For Samuel Weber, This assertion marks out the epistemological space of Saussures theoretical effort, and to neglect its farreaching implications has inevitably meant to misconstrue the
status of his arguments.17 In order not to misconstrue Saussures
arguments, I need to point out that semiotics constructs a model of
its object of study; it does not create its object of study (despite
Saussures use of the verb creer in the preceding quotation).
To answer adequately Bordwells second question Is linguistics presumed to offer a way of subsuming film under a general
theory of signification? we need to go further into semiotic
theory. The underlying system is an imperceptible content lending
structure to the perceptible insofar as it signifies and conveys precisely the historical experience of the individual and group.18 Semioticians call this non-perceptible, underlying system, which lends
10
structure to the perceptible, a system of codes. One of the integral

(although by no means encompassing) codes of human culture is
natural language. It is a species-specific system that distinguishes
humans from animals and that humans use to develop a shared
understanding of the world.
A system of codes consists of the prior set of finite, invariant
traits of a language, together with their rules for combination.
Speech (la parole) is generated by two processes: Codes are selected
from the underlying system, and they are combined according to
rules. Both processes constitute the intelligibility of speech because
meaning is the product of the structural relations that exist between
the codes. Speech can then be analyzed in terms of the underlying
system of codes that generated it.19 In semiotics, code is therefore
a term that designates the underlying system that constitutes the
specificity of, lends structure to, and confers intelligibility on phenomena.
In analyzing film from a semiotic perspective, film scholars
bring to film theory a new level of filmic reality. They successfully
demonstrate that the impression of unity and continuity each spectator experiences at the cinema is based on a shared, nonperceptible underlying system of codes that constitutes the specificity of, lends structure to, and confers intelligibility on the
perceptible level of film. Early film semioticians applied the structural linguistic methodology of segmentation and classification to
identify the non-perceptible system underlying a film. The setting
up of this hierarchy between the perceptible level of film and the
non-perceptible system of codes underlying it is the main contribution semioticians have so far made to film theory. They show
that filmic continuity is a surface illusion, what Marxist critics call
the impression of reality. In effect, semiotics enables film theorists
to drive a wedge between film and its referent, to break the supposedly existential link between them, and to demonstrate that filmic
meaning is a result of a system of codes, not the relation between
images and referents.
Once film semioticians identified the hierarchy between the perceptible and the non-perceptible, what were their main objects of
study? Very simply, they began to construct models of the various
underlying systems that determine the surface perceptible level
of film. It is at this point that film semioticians moved away from
analyzing cinematic language (or filmic specificity) and created a
11
theory of textual analysis, motivating the study of underlying systems that determine the textual structure of the particular film
under discussion, including the cause-effect narrative logic, the process of narrativization, the spatio-temporal relations between shots,
the patterns of repetition and difference, and specific filmic techniques such as the eyeline match.
The identification and analysis of all these underlying systems
are a result of subsuming film under a general theory of signification. Bordwell may protest that the Russian Formalists studied
many of these filmic mechanisms, but, as is well known, Saussures
structural linguistics directly influenced the Russian Formalists.
3. Does linguistics offer methods of inquiry which we can adopt?
Linguistics does offer methods of inquiry that film theorists can
adopt. I shall refer to the most obvious example: Early film semioticians borrowed from structural linguistics the commutation test, a
deductive method of analyzing how the underlying level lends
structure to the surface level. This method consists of the activities
of segmentation and classification. In principle, a commutation involves the correlation between a change on the surface level and a
change on the underlying level. A change on the surface may be
either a variation of the same code or a new code. By means of the
commutation test, semioticians can identify the changes on the surface level that correlate with the changes on the underlying level.
The commutation test enabled Saussure to describe speech (la
parole) as an infinity of messages generated by a finite, underlying
system (la langue). The concept of identity enabled him to reduce
the infinity of speech to this finite system, for he recognized that all
speech is composed from the same small number of invariant codes
used recursively in different combinations. Saussure did not conceive of this system as a mere conglomerate of codes, but as a series
of interdependent, formal relationships. Furthermore and here
Saussure located the ultimate law of language he defined codes
only in terms of their relation, or difference, to other codes (both the
paradigmatic relations they enter into in the underlying system and
the syntagmatic relations they enter into in speech).
The theory of commutation, based on the analytic methods of
segmentation and classification, led Metz (in his essay Problems
of Denotation in the Fiction Film) to formulate the grande syntagmatique that Bordwell praised in the Preface to Narration in the
12
Fiction Film. For Metz the grande syntagmatique designates one of the
primary codes underlying and lending structure to all classical
films. It represents a prior set of finite sequence (or syntagmatic)
types, a paradigm of syntagms from which a filmmaker can choose
to represent profilmic events in a particular sequence. Metz defines
each syntagma according to the spatio-temporal relations that exist
between the profilmic events it depicts. Syntagmas are commutable
because the same events depicted by means of a different syntagma
will have a different meaning.20 Metz detected eight different types
of syntagma in total, each of which is identifiable by a specific
spatio-temporal relationship existing between its images. These
syntagmas form a finite paradigm of invariant codes to the extent
that they offer eight different commutable ways of constructing an
image sequence. The eight syntagmatic types therefore conform to
Saussures ultimate law of language, because each syntagma is
defined in terms of its relation, or difference, to the other syntagmas. For Metz: These montage figures [film syntagmas] derive
their meaning to a large extent in relation to one another. One,
then, has to deal, so to speak, with a paradigm of syntagmas. It is
only by a sort of commutation that one can identify and enumerate
them.21 Notice again that this semiotic model does not draw any
direct comparisons between film and natural language.
4. Is linguistics simply a storehouse of localized and suggestive
analogies to cinematic processes?
This final question, more rhetorical than the others, reveals
Bordwells preferred way of characterizing modern film theory.
Ideally, my responses to the previous three questions have shown
that linguistics offers film theorists more than a storehouse of localized and suggestive analogies. Moreover, linguistics does not encourage the majority of film semioticians to draw analogies between film and natural language. Metz adopted a scholastic method
of theorizing, in which he considered all the arguments that can be
advanced for or against an hypothesis in this instance, the comparison between film language and natural language to determine
its degree of credibility. The bulk of his first essay Cinema: langue
ou langage? advances arguments against the temptation to draw
analogies between film language and natural language.22 Metz realized that the two languages belong to different logical categories,
and that recognition led him to conclude that film is a langage sans
13
langue.23 In his later work, Metz went on to carry out the primary
aim of film semiotics to construct a model of filmic specificity,
films underlying system. In constructing the grande syntagmatique
he confused filmic specificity with narrativity, but later, in Language
and Cinema, he defined specificity in terms of a specific combination
of five traits, plus the underlying system that organizes them.
To summarize my responses to Bordwells four questions, the very
idea of cinematic language for film semioticians is not simply an
analogy (as it was in the prelinguistic film-language comparisons
of Raymond Spottiswoode, the filmology movement, etc.); that is,
the semioticians analysis of film is not premised on identifying any
direct resemblance between film and natural language. Instead, film
semioticians argue that film is a medium that possesses its own
distinctive, underlying system that confers intelligibility on and
lends structure to all films. Of course, semioticians do not pretend
to study everything that makes a film intelligible; instead, they limit
their analysis to the invariant traits that define films specificity.
What about the cognitivists critique of other domains of modern film theory? One of the dominant reasons the cognitivists criticize modern film theory is the behaviorism implicit in its account
of subject positioning, in which the spectator is automatically and
unfailingly positioned as an ideological subject, with no cognitive
capacity to process and manipulate the film. In other words, the
modern film theorists posited a direct, unmediated relation between the stimulus and the spectators response, which, as Bordwell observes, impute[s] a fundamental passivity to the spectator.24 This criticism is certainly valid and justifies the need for a
cognitive account of the spectators processing activity. But does it
also justify rejecting semiotics? The argument I develop is that a
cognitive theory of film that assimilates semiotics overcomes the
problems of translinguistics, the behaviorism of modern film theorys account of subject positioning, together with the cognitivists
overemphasis on the spectator as an autonomous rational self. The
cognitivists, on the other hand, argue that film theorists need to
reject semiotics and start again by developing a cognitive theory of
spectatorship untainted by semiotics. Bordwell and others ask, Why
does a cognitive theory of film need to refer to language and semiotics? My immediate answer is that we need to consider the specificity of the human mind and culture. Whereas the Enlightenment
14
philosophers argued that reasoning is specific to humanity, twentieth century philosophers belonging to the Language Analysis tradition, together with semioticians, realize that language is specific
to humanity. Language is not just another aspect of the human
mind, but is its defining characteristic. I shall have more to say on
the Language Analysis tradition later.
In opposition to modern film theory, Bordwell argues that a
film . . . does not position anybody. A film cues the spectator to
execute a definable variety of operations.25 Bordwell then proceeds
to fill in the mental blanks left by the behaviorist stance of modern
film theory. But by arguing that a film does not position anybody,
Bordwell suggests that the spectator is a context-free entity and
that film viewing is a purely rationalist activity. I agree with Bordwell that spectators are not positioned or placed by a film in the
narrow sense of the modern film theorists (as Bordwell writes, the
terms position and place lead us to conceive of the perceiver as
backed into a corner by conventions of perspective, editing, narrative point of view, and psychic unity.26) We can reject this narrow
conception of spectatorship without rejecting the proposition that a
film modifies the spectators mind in a specific way.27
To varying degrees cognitivists downplay or reject the anthroposemiotic dimension of filmic comprehension and instead focus
on its ecological dimension. The following cognitivists are key to
the ecological approach: Joseph Anderson, the Australian philosopher Gregory Currie, the Danish film theorist Torben Grodal, and
the Dutch film theorist Ed Tan.28 Edward Branigan and David
Bordwell also develop ecological theories, but to a lesser extent.
(Since its initial development in North America, cognitive film theory has therefore become international and has developed an ecological framework at the same time.) The flavour of this work can
be summed up in the following extract from Grodals Moving Pictures:
Visual fiction is viewed in a conscious state, and is mostly about human
beings perceiving, acting, and feeling in, or in relation to, a visible and
audible world. The viewers experience and the phenomena experienced
often demand explanations that imply non-conscious activities; but the
emotions and cognitions must be explained in relation to conscious mental
states and processes. For evolutionary reasons, it is improbable that the
way phenomena appear in consciousness is just an illusion caused by
certain quite different non-conscious agents and mechanisms.29
15
By ignoring each other, cognitivists and semioticians have developed unbalanced theories of the cinema. In our search to understand how films are understood, we need to maintain a balance
between cultural constraints, such as language and other semiotic
systems of human culture, and broader ecological constraints. The
cognitive film semioticians go some way to achieving this balance,
in opposition to the linguistic determinism of Metzs film semiotics
and the free will and rational autonomy the cognitivists confer
upon film spectators. Each of the following chapters charts the way
the cognitive film semioticians attempt to maintain this balance,
although there is variation among them.
The Language Analysis Tradition

The Language Analysis tradition incorporates the analytic philosophy of Frege, Carnap, Moore, Russell, Ryle, and Wittgenstein; the
structural linguistics of Saussure; the pragmatics of Habermas; and
the semiotics of C. S. Peirce. Its primary aim is to transform questions about epistemology and the mind into questions about language and meaning. Karl-Otto Apel defines the Language Analysis
tradition as (following Aristotle) a prima philosophia: First Philosophy was founded by the Greeks as an ontology of the essential
structure of things; . . . later on, in the so-called new age, it was
transformed into, or replaced by, a critical epistemology . . . ; and . . .
finally, in the 20th century, both ontology and epistemology were
questioned or transformed by language analysis.30
The three hundred year domination of epistemology, or the
philosophy of the subject, in Western philosophy began with Descartes, reached its peak in the subjective idealism of Kant and
Fichte, and was completed in the objective idealism of Hegel and
the British idealists (Bosanquet and Bradley). The significance of
the transformation from the idealism of epistemology to Language
Analysis was that language, signs, and the process of semiosis
replaced mental entities. For example, semiotics and structuralism
initiated a radical critique of Kants subjective idealism without
attacking reason and rationality (which led Paul Ricoeur to call
structuralism Kantianism without the transcendental subject). Semioticians relocated reason and rationality in language and sign
systems, rather than in the mind. For Jurgen Habermas, The structuralist approach follows Saussures lead and begins with the
16
model provided by grammatical rule systems; and it overcomes the

philosophy of the subject when it traces the achievements of the
knowing and acting subject, who is bound up in his linguistic
practices, back to the foundational structures and generative rules
of a grammar. Subjectivity thereby loses the power of spontaneously generating a world.31 Within analytic philosophy, Gilbert
Ryle wrote a devastating critique of Cartesian dualism in The Concept of Mind.32 More generally, analytic philosophers privileged language by conceiving of philosophical problems as mere confusions
and misunderstandings over language use, rather than as conflicts
or disagreements over substantive issues. Analytic philosophy
therefore set itself up as a therapeutic activity that aims to dissolve
philosophical problems by clarifying the meaning of words and
expressions. Meanwhile, Frege replaced psychologism with logical
analysis, and C. S. Peirce transformed Descartess method of introspection and Kantian epistemology into semiotics.33
Jurgen Habermas is well known for his attempts to continue the
project of modernity, and the Enlightenment project from which it
emerged. But he recognizes the limitations and failures of the Enlightenment project most notably, its idealism. One dimension of
Habermass work consists of rereading the Enlightenment project from the perspective of linguistics or, more accurately, the
pragmatic theory of speech acts.34 This enables him to update and
transform the Enlightenment by replacing its idealism with Language Analysis, effecting a shift from the purely subjective to the
intersubjective language, dialogue, and communicative reason.
His argument is that language is not simply one of many human
possessions in the world; instead, it is the primary possession, since
it constitutes the basis of humanitys understanding of the world
and orients the individual into a shared interpretation of both the
world and human actions. Habermas does not fall into a translinguistic trap because he is from the start focusing on human reasoning capacities. And he is not concerned with the purism of pure
reason, but with the way reason is concretized in language. The
problem with the philosophy of the subject for Habermas is that it
reifies subjectivity by conferring upon the individual a narcissistically overinflated autonomy because it overemphasizes purposively rational self-assertion.35 By contrast, Habermas and the
Language Analysis tradition generally decentres subjectivity by
arguing that it is not master of its own house, but is dependent on
17
something that is a priori and intersubjective language. And language is fundamental because it functions to represent (or disclose)
reality, establishes interpersonal relations, and guides personal expression. In other words, it facilitates mutual understanding and
coordinates social action: Agreement arrived at through communication, which is measured by the intersubjective recognition of
validity claims, makes possible a networking of social interactions
and lifeworld contexts . . . The stratification of discourse and action
built into communicative action takes the place of . . . prelinguistic
and isolated reflection.36 Habermas presents a strong case for replacing the idealism of the philosophy of the subject with Language
Analysis. I shall develop this point in more detail in Chapter 2,
when reviewing David Bordwells cognitive theory of the spectator.
Language Analysis in all its forms therefore rejects idealism and
mentalism, transforming the first person perspective of epistemology (Descartess method of introspection) to the third person perspective of language and signs. Thomas Daddesio clearly sums up
the issues involved:
The critique of introspection initiated by Peirce gained momentum when,
with the rise of behaviorism in the social sciences, introspection was abandoned as a reputable method because it was perceived as being unable to
provide the objective, repeatable observations that science requires. As long
as it seemed reasonable a circumstance that lasted roughly three hundred
years to believe that one could have privileged access to the contents of
ones own mind, mental processes could be taken as foundational for both
epistemology and accounts of human behavior. However, once this privilege came to be viewed as illusory, introspection was replaced by methods
relying on a third-person perspective. From this new perspective, the access that individuals have to their own thoughts could no longer be taken
as a foundation for knowledge and, consequently, private events were
replaced, in discussions of language, meaning, and reason, by events that
were open to public scrutiny such as the behavior of others, the words they
utter, and the uses to which they put words.37
The epistemologists assumptions of immediate access to the

thoughts in ones own mind and the power of the mind to disclose
reality were replaced by the Language Analysts assumptions of
indirect access to ones thoughts via language and other intersubjective sign systems. Whereas cognitivists adopt the first person perspective of epistemology (philosophy of the subject), semioticians
adopt the third person perspective of the Language Analysis tradition.
18
Noam Chomsky and the Study of Competence

In the late fifties, the mind and cognition made a decisive return
within the Language Analysis tradition, beginning with Noam
Chomskys transformational generative grammar (together with his
decisive critique of B. F. Skinners behaviorism).38 In the form of
Chomskys linguistics, the Language Analysis tradition created a
synthesis of both the mentalism of epistemology and the intersubjective nature of language, thus avoiding the idealism and first
person perspective of epistemology and the (quasi) behaviorism of
the Language Analysis tradition. In this respect, Chomskys work
represents the ideal paradigm for cognitive film semioticians.
David Bordwell has noted the absence of references to the work
of Chomsky in film theory: It is surprising that theorists who
assign language a key role in determining subjectivity have almost
completely ignored the two most important contemporary developments in linguistic theory: Chomskys Transformational Generative Grammar and his Principles-and-Parameters theory.39 He
adds that no film theorist has mounted an argument for why the
comparatively informal theories of Saussure, Emile Benveniste, or
Bakhtin are superior to the Chomskyan paradigm. For over two
decades film theorists have made pronouncements about language
without engaging with the major theoretical rival to their position.40 The truth of the matter is that over the last two decades a
number of film theorists have been engaging with Chomskyan linguistics and, furthermore, have deemed it to be superior to structural linguistics. Throughout this book I attempt to emphasize that
Chomskyan linguistics, particularly in its study of competence, has
defined the central doctrines of cognitive film semiotics. Here I shall
briefly chart the relation between early film semiotics and cognitive
film semiotics.
During the seventies, Metzs film semiotics was modified and
transformed. Its fundamental problems, as we have already seen,
lay in its total reliance upon structural linguistics. One major transformation came from post-structural film theory, which based itself
primarily upon the Marxism of Louis Althusser and the psychoanalysis of Jacques Lacan. Post-structuralists criticize structuralism
because they regard it to be the last vestige of Enlightenment reason
and rationality. Christopher Norris clearly sums up this poststructural position:
19
Structuralism renounces the Kantian transcendental subject, only to replace it with a kind of linguistic a priori, a regulative concept of structure
which seeks to place firm juridical limits on the play of signification. Such,
at least, is the critique brought to bear upon structuralist thinking by those
like Lacan and Derrida who read in it the last, lingering signs of a
rationalist tradition forced up against its own (unconscious) limits.41
Ultimately, structuralism replaces the transcendental Kantian subject with a transcendental signified.
For most Anglo-American film scholars, film semiotics takes
only one form namely, Metzs early film semiotics, ranging from
his 1964 paper Le cinema: langue ou langage? leading to his
remarkable paper on the grande syntagmatique of the image track,
and finally to his monumental book Langage et cinema, published in
1971 and translated into English in 1974. But as Metz himself acknowledged in the opening chapter of this book, By its very nature, the semiotic enterprise must expand or disappear.42 Although Langage et cinema marks the logical conclusion to Metzs
structural linguisticbased film semiotics, it does not mark the end
of film semiotics per se. In his subsequent work (particularly his
essay The Imaginary Signifier),43 Metz adopted a psychoanalytical framework, which aided the formation of post-structural film
theory. However, many of his students and colleagues continued to
work within a semiotic framework, which they combined with cognitive science. Research in film semiotics continued unabated in the
seventies, eighties, and nineties, especially in France, Italy, and the
Netherlands. Far from disappearing, film semiotics has expanded
into a new framework, one that overcomes the problems of structural linguisticbased film semiotics by embracing three new theories: (1) a renewed interest in enunciation theory in both film and
television (particularly in the work of Francesco Casetti and Metz
of LEnonciation impersonnelle ou le site du film),44 (2) pragmatics (in
the work of Roger Odin), and (3) transformational generative grammar and cognitive science generally (in the work of Michel Colin
and Dominique Chateau).
One defining characteristic of cognitive film semiotics is that it
aims to model the actual mental activities (intuitive knowledge)
involved in the making and understanding of filmic texts, rather
than study filmic texts themselves. Ultimately, the theories of Francesco Casetti, Roger Odin, Michel Colin, and Dominique Chateau
are models of filmic competence. Each theorist models this compe-
20
tence from a slightly different perspective: Casetti employs the deictic theory of enunciation, Odin employs pragmatics, and Colin and
Chateau employ generative grammar and cognitive science.
Chomskys study of linguistic competence in his generative
grammar (where grammar is defined as a theory of language) is
one of the main research programs that led to the development of
cognitive science in the fifties. Generative grammar shifted linguistic inquiry away from epiphenomena (actual language behavior)
and toward competence the intuitive knowledge that underlies
natural language behavior, together with the innate, biologically
determined language faculty that constitutes this knowledge as species-specific. Chomsky therefore follows the Enlightenment philosophers study of what distinguishes humans from non-humans.
Chomsky defines the specificity of human reasoning in terms of the
possession of a language faculty, a faculty that enables each human
to internalize a particular natural language.
Generative grammar is therefore a cognitive theory of natural
language. Its cognitive dimension consists of two stages: the study
of the language faculty in its initial state and the study of this
faculty after it has been conditioned by experience (which leads to
the internalization of a particular natural language). The study of
the language faculty in its initial state is called universal grammar
(the interlocutors initial competence), whereas the study of a particular natural language involves accounting for the structure of the
language faculty after it has been determined by experience (which
leads to the formation of the interlocutors attained competence).
To study the grammar of a particular natural language is to
attain descriptive adequacy, whereas to study universal grammar
is to attain explanatory adequacy. The aim of a descriptively adequate generative grammar is to construct a formal model (consisting of generative and transformational rules) that generates all and
only the grammatical sentences of a particular natural language. By
contrast, explanatory adequacy attempts to model the initial state
of the language faculty, which Chomsky conceived in terms of a
series of innate principles and parameters. If one studies English or
Japanese grammar, one is studying the same language faculty (the
same series of principles and parameters), but under different empirical conditions, conditions that set these principles and parameters in alternative configurations.
However, structural linguistics attains only observational ade-
21
quacy, because it merely segments and classifies linguistic behavior

into paradigms and does not attribute any cognitive reality to those
paradigms. By contrast, Chomsky defines the language faculty and
natural languages as cognitive realities, as physical components of
the mind: A generative grammar is not a set of statements about
externalized objects constructed in some manner. Rather, it purports to depict exactly what one knows when one hears a language:
that is, what has been learned, as supplemented by innate principles. UG [universal grammar] is a characterization of these innate,
biologically determined principles, which constitute one component
of the human mind the language faculty.45
In its initial formation in the fifties and sixties, generative grammar attempted to distinguish itself from structural linguistics by
defining itself as descriptively adequate, with little concern for explanatory adequacy. Descriptive adequacy was set out in what is
known as the Standard Theory of generative grammar, as outlined
by Chomsky in Aspects of the Theory of Syntax, published in 1965.46
It is this early theory of generative grammar that the cognitive film
semiotician Michel Colin used in his generative rereading of Metzs
grande syntagmatique. In Chapter 5, I will outline both Chomskys
Standard Theory and Colins re-reading of Metz.
I have noted that natural language is species-specific. But what
exactly makes it species-specific? We have already seen that for
Habermas it is through language that humans establish mutual
understanding and coordinate social action. Chomsky refers to the
property of discrete infinity (or the creativity principle); the structural linguist Andre Martinet refers to natural languages principle
of economy, its doubly articulated system of organization.47 For
Chomsky and Martinet, natural language is distinctive because it
generates infinite phenomena by finite means. Double articulation
posits a first (the higher) level of language, analyzable into a large
but finite number of meaningful units (morphemes, or monemes),
and a second, lower level, consisting of a smaller number of nonmeaningful units (phonemes), which have no semantic content in
themselves but combine with one another on the higher level to
form morphemes.48 The morphemes in turn are combined in finite
ways to produce an infinite number of sentences. Chomskys Standard Theory comprises much as semiotic models do an underlying level of reality and a surface level. The underlying level consists of a finite number of base rules that generate a finite number
22
of deep structures. These are then transformed into a potentially

infinite number of surface structure sentences by the recursive application of a finite number of transformational rules. Chomskys
model of this underlying reality balances out languages limiting
and enabling capacities.
Double articulation and discrete infinity can be made clearer by
drawing an analogy with chess, which consists of a finite number
of squares and a finite a number of chess pieces, each of which can
be moved in a finite number of ways. But when we add up all these
finites, we end up with a potentially infinite number of chess
games.
A first justification for studying film in the context of generative
linguistics is to discover whether it is also structured according to
the property of discrete infinity. Initial signs are promising, since
discrete infinity is premised on the ability to segment and classify
the discrete units of a language, as Metz did successfully in constructing the grande syntagmatique. Discrete infinity is implicit in the
grande syntagmatique to the extent that it formalizes the finite determined ordering (the eight images sequences) of an infinite number
of undetermined units (shots).
Chapter Contents
Chapter 2 reviews the cognitive theories of George Lakoff and
Mark Johnson, which are based on the premise that thought and
language are represented in the mind in the form of schemata
(cognitive structures that organize perceptual input into experiences) more specifically, image schemata, which are directly motivated by bodily experience. The chapter outlines the notion of
image schemata, and discusses the limitations of David Bordwells
nonimage schemata theory of the film spectator, before moving on
to consider the potential for developing a cognitive semantics of
film from Michel Colins essay Film Semiology as a Cognitive
Science, in which Colin perceives a close affinity between semiotics and cognitive science, since both paradigms address similar
issues language, vision, and problem solving. The chapter ends
with my attempt to develop a cognitive semantics of film based on
Lakoffs container image schema.
Chapter 3 returns to the issue of filmic enunciation. Enunciation
designates the activity that results in speech, in the production of
23
utterances (enonces). Emile Benveniste identified two forms of utterance: discourse (discours) and story (histoire). Discours employs
words such as personal, possessive, and demonstrative pronouns
that grammaticalize within the utterance particular aspects of its
spatio-temporal context (e.g., the speaker and hearer), whereas histoire is a form of utterance that excludes pronouns. The cognitive
film semiotician Francesco Casetti takes to its logical conclusions
the analysis of film in terms of personal pronouns. Using the categories I, you, and he, he develops a typology of four shot types,
which aims to describe the way film orients itself in relation to the
spectator. However, Casetti also took the translinguistic framework
to its logical conclusion and was rightly criticized by Metz for doing
so. In his final published work (LEnonciation impersonnelle ou le site
du film), Metz disputes Casettis pronoun theory of film arguing
that film can only be studied as histoire. After criticizing Casettis
translinguistics, Metz did not expand the enunciative theory of film
to incorporate general cognitive and ecological principles; instead,
he simply replaced personal pronouns with anaphora a narrow
system of orientation in which one textual element points to another. Moreover, Metz identifies anaphora as a reflexive moment in
a film, an identification that results in his identifying filmic enunciation with reflexivity. The terms of the debate between Casetti and
Metz take up most of Chapter 3, although toward the end I attempt
to argue that, in rejecting a translinguistic theory of filmic enunciation (one based exclusively on personal pronouns), we do not need
to reject the indispensable concept of deixis, which designates a
mechanism that orients individuals in relation to their environment.
The personal pronouns of natural language constitute the privileged category of orientation, which has led many film theorists
erroneously to equate deixis with pronouns. At the end of Chapter
3, I briefly develop a non-linguistic theory of deixis in film and
television, by considering the issue of colorization of black and
white films, as well as television viewers comprehension of news
programs.
In linguistics, cognitive pragmatics designates a discipline that
describes a type of linguistic competence that governs the relation
between utterances and the appropriate contexts in which they are
uttered. Cognitive pragmatics is therefore a study of the immediate
discursive nature of language. In Chapter 4 we see that, for the
cognitive film semiotician Roger Odin, film semiotics is primarily
24
pragmatic, because of the immediate discursive nature of film. According to Metz, film is a langage sans langue, that is, belongs to
the dimension of la parole. Unlike for Saussure, for Odin la parole is
not an heterogeneous domain of unconstrained language use, but
is determined by a pragmatic competence. Odin develops a threefold distinction between institutions, modes, and operations in order to characterize the film spectators pragmatic competence in
comprehending films. This chapter charts the different institutions,
modes, and operations specified in Odins work and discusses in
detail fiction films, documentaries, home movies, and what Odin
calls the dynamic mode of filmmaking.
Michel Colin made explicit the connection between film and
generative grammars concept of discrete infinity (as well as other
concepts). In his essay The Grande Syntagmatique Revisited, discussed in Chapter 5, Colin redefined the observationally adequate
grande syntagmatique as descriptively adequate. This involved redefining the eight syntagmatic types in terms of selectional (or semantic) features. Selectional features represent the inherent grammatical
and semantic components of lexical items (what the rest of humanity calls words). For example, the lexical item cat can be represented in terms of the following selectional features: Common,
Count, Animate, Human. Every lexical item can be characterized in terms of these and other selectional features. If we return to
the chess analogy, we can also analyze each chess piece into selectional features. This would simply involve encoding the grammatical and semantic components of each chess piece the moves it
can perform in terms of a finite series of components. The most
remarkable result of Colins re-reading of Metzs work is that, as
with all generative models, the actual, manifest syntagmatic types
are posited as merely the result (the epiphenomenon) of the generative process. Within the generative framework, we can identify
and analyze, not only actual syntagmas, but also possible (i.e.,
potential) syntagmas and impossible syntagmas. Once all the finite
selectional features have been identified, the potentially infinite
number of syntagmatic types can be conceived and generated.
These selectional features constitute the finite underlying level of
filmic discourse (or its system of codes) from which a potentially
infinite number of film sequences can be generated. For Colin, then,
the primary aim of the grande syntagmatique is not to identify actual
syntagmatic types, but to identify the more fundamental selectional
25
features that combine to form these syntagmatic types. Chapter 5

ends by using Colins work as a starting point for thinking through
and redefining the problem of the grammar of film in cognitive
terms. Reference is also made to the work of Dominique Chateau
and the American John M. Carroll.
All the cognitive film semioticians studied in this book are united
by the same project to combine film semiotics and cognitive
science with the aim of modelling filmic competence. The following
chapters offer my own sympathetic, but not entirely uncritical,
working through of this project. Throughout, the reader may detect
a tension between the need simply to present a project almost
unknown to Anglo-American film scholars (in the interests of being
informative, to make known what has been marginalized in AngloAmerican film studies) and the need to develop my own elaboration of a (necessarily) fragmentary and incomplete project. As film
semiotics has been marginalized and repressed in film studies, perhaps we could for a moment take the notion of working through in
its Freudian sense (Durcharbeitung) and suggest that this book aims
to overcome the resistance to film semiotics by returning it from
the repressed. Once it is returned to consciousness, the reader may
be surprised to find that what was repressed has, during its latency
period, matured and been transformed into a new object.
CHAPTER TWO
The Body on Screen and in Frame

Film and Cognitive Semantics
Reason is embodied in the sense that the very structures on which

reason is based emerge from our bodily experiences. Reason is
imaginative in the sense that it makes use of metonymies, metaphors, and a wide variety of image schemas. (George Lakoff).1
odern (or contemporary) film theory marks a decisive break with the classical film theory of Rudolf Arnheim, Andre Bazin, Bela Bala`sz, Siegfried Kracauer, and Victor
Perkins. We can characterize this break as a transition from an
extensional semantic to an intensional (or immanent) semantic theory of meaning. Classical film theory is a theory of cinematic representation premised upon mechanical duplication (upon the existential relation between film and referent) and is therefore based
upon an extensional semantic theory of meaning. In contrast, modern film theory, beginning with the film semiotics of Christian
Metz, corresponds to an intensional semantic theory of meaning,
for it drove a wedge between mechanical duplication and cinematic
representation and then defined the latter, not externally as the
effect of an existential link between film and referent but internally as an effect of codes.
Furthermore, the evolution of modern film theory can be characterized as progressing from an intensional semantic to a pragmatic theory of filmic meaning. Pragmatists study how successful
communication is achieved on the level of language use. This is
because, they argue, languages meaning is not determined in advance of its use. In the early days of pragmatic theory, deictic words
such as personal and demonstrative pronouns became the privileged example of how language only gains meaning when used.
For instance, the meaning of the personal pronoun I is dependent
on the person uttering it at the time of speaking. For pragmatists,
26
THE BODY ON SCREEN AND IN FRAME
27
language is structured around, and gains its meaning from, external

absent causes (such as the speaker), and the purpose of early theories of pragmatics was to study how these external absent causes
are formally represented in linguistic structure. In the seventies this
early theory entered modern film theory through Emile Benvenistes theory of enunciation. In particular, the modern film theorists
used Benvenistes theory of enunciation to argue that narrative film
is structured around, and gains its meaning from, the external absent centre that spectators come to occupy. Within this theory the
spectator is regarded as being formally represented in the film by
the subject position (traced onto the filmic image by means of the
perspectival-optical system of the camera and reproduced across
potentially disruptive cuts by means of conventions of spatial construction, such as the 180 degree rule, eyeline matches, and so on).
The modern film theorists went beyond Benvenistes purely linguistic theory by conceiving the subject position in terms of Louis
Althussers theory of interpellation and Jacques Lacans theory of
the mirror phase.
In the previous chapter I pointed out the main problem with
these theories their implicit behaviorism, in which the spectator
is theorized as passive, with no cognitive capacity to process information. Furthermore, it is a common assumption of psychoanalytic
film theory that the cinema projects the spectator into a spatiotemporal elsewhere. That is, cinema releases the spectator from her
immediate spatio-temporal limits by transporting her vision
through space and time (while keeping the body motionless). Psychoanalytic film theory in effect dematerializes the spectators body
while privileging the unconscious mind and vision. This follows
Descartes, who also disembodied vision by making it a property of
the (conscious) mind. My aim in this chapter is to develop my own
theory of cinematic perception, one that grounds perception in the
physicality of the body. To achieve this I summarize the film theories of Metz, David Bordwell, and especially Michel Colin; identify their limitations (specifically relating to the disembodiment
of perception); and attempt to overcome these limitations by reviewing and deploying George Lakoffs and Mark Johnsons cognitive semantic theory of meaning, a theory that extends and reformulates some of the basic assumptions of the Language Analysis
tradition.
28
Disembodied Structures and Schemata in Film Theory:

Metz, Bordwell, Colin
Saussures principle of the arbitrary and conventional relation between signifier and signified implies a disembodied realm of conceptual structure (the signified), since the body is not seen to motivate the meaning of signifiers. To sustain the theory of the arbitrary
sign, Saussure developed the concept of a language as an underlying system, because the sign consists of the combination of two
entities without connection, and only a system, consisting of necessary conventions, can explain how the signifier and signified are
combined.
In Cinema: Language or Language System?2 Metz failed to
establish a film semiotics on the level of the image because of his
inability to articulate a new epistemological position on film by
means of the notion of the arbitrary relation between signifier and
signified, for he believed the filmic image to be motivated. In order
to achieve his primary aim to define filmic specificity in semiotic
terms Metz sought, in Problems of Denotation in the Fiction
Film,3 arbitrariness on the level of image sequences, which confer
upon the profilmic events a meaning that goes beyond their analogical relation to the image. These image sequences conform to the
principle of arbitrariness because there is no strict motivation governing the choice of one image sequence over another in representing a particular profilmic event.
In Cinema: Language or Language System? and Problems of
Denotation in the Fiction Film, Metz worked with the conception
of the image-as-analogue (as a mechanical duplication of reality).
Stephen Heath called this the blind spot of Metzs formulations,
for it is the point at which the articulation of significance collapses
in the face of analogy.4 But in Language and Cinema, Metz overcame
this blind spot under the influence of Umberto Eco. Eco was one of
the first semioticians to study the image successfully in terms of
codes; that enabled him to define its apparent analogical (nonsemiotic) nature as a system of iconic codes organized into a triple
hierarchy.5
The basic terms that Metz worked with were therefore arbitrariness, motivation, code, and iconicity. Arbitrariness is opposed to
motivation and is correlated with code, whereas code is opposed to
iconicity and is correlated with arbitrariness (although in Language
29
and Cinema Metz defined iconicity in terms of codes).6 Metz began

from the folk psychological premise that film is motivated and
iconic, then proceeded to demonstrate that it is in fact coded and
conforms to the principle of arbitrariness. Furthermore, structural
linguistics begins from the premise that language is a system of
uninterpreted (meaningless) symbols. For structural linguistics,
meaning arises out of non-meaning, from the combination of phonemes (which have no semantic content, but only signify difference)
into morphemes (which do have semantic content).
In Narration in the Fiction Film David Bordwell rejected a theory
of film based on structural linguistics and instead developed a
schema-based theory of filmic comprehension. Cognitive psychologists define schemata as abstract, transcendental, static, top-down
(rather than bottom-up) structures of the mind that organize perceptual input into coherent mental representations.7 Schemata are
therefore finite abstract structures that interact with an infinite
amount of perceptual data to form experiences. In this sense, schemata constitute the generative capacity of the mind to comprehend
perceptions recurrently. Bordwell develops his theory within the
Constructivist school of cognitive psychology, which studies the
activity of the perceiver in generating hypotheses and inferences
in order to make sense of inherently fragmentary and incomplete
perceptual input. From this perspective, Bordwell conceptualizes
the narrative film as an inherently incomplete form of discourse
(as with all other forms of discourse). More specifically, he argued that the narrative films logical form is incomplete, but is
enriched, or completed, by the spectators activity of inference generation.
Narrative film conforms to a psychologically motivated causeeffect logic. Bordwell describes this logic in terms of the two tier
fabula-syuzhet opposition as formulated by the Russian Formalists:
Presented with two narrative events, we look for causal or spatial or temporal links. The imaginary construct we create, progressively and retrospectively, was termed by Formalists the fabula (sometimes translated as
story). More specifically, the fabula embodies the action as a chronological, cause-and-effect chain of events occuring within a given duration and
spatial field. . . . The syuzhet (usually translated as plot) is the actual
arrangement and presentation of the fabula in the film. It is not the text in
toto. It is a more abstract construct, the patterning of a story as a blow-byblow recounting of the film could render it.8
30
In constructing a fabula, the perceiver defines some phenomena as events

while constructing relations among them. These relations are primarily
causal ones. An event will be assumed to be a consequence of another
event, of a character trait, or of some general law. The syuzhet can facilitate
this process by systematically encouraging us to make linear casual inferences. But the syuzhet can also arrange events so as to block or complicate
the construction of casual relations.9
To complete the films narrative logic, the film spectator must possess a cause-effect schema that will serve as the context in which
the narrative film is processed. The cause-effect schema is the context most accessible during the spectators processing of a narrative
film, for (in the terms of Sperber and Wilsons principle of relevance, discussed in Chapter 4) it enables the spectator to process
the inherently incomplete logical form of the narrative film with
the least amount of processing effort yielding an optimal contextual
effect.
Bordwells theory is primarily a top-down account of information processing; from this perspective, perceptual data (in this instance, narrative films) are conceived merely as a set of cues interacting with the spectators cognitive capacity (in this instance, the
cause-effect schema), triggering and constraining her activity of
inference generation. In Bordwells terms, the cues in the narrative
film are organized in such a way as to encourage the spectator to
execute story construction activities. The film presents cues, patterns and gaps that shape the viewers application of schemata and
the testing of hypotheses.10 The narrative film cues and constrains
the story construction activities of the spectator in such a way as to
enable her to form the space, time, and cause-effect dimensions of
the film into a single, coherent mental representation.
The cause-effect schema constrains the film spectator to interpret the narrative films set of cues in a particular way in terms
of a character-centred causality, based on the psychological motivations and rational actions of characters. The schema is, in effect,
an interpretive strategy for the spectator, enabling her to recognize
certain perceptual data as relevant cues, to generate the necessary
inferences, and to combine cues and inferences into a coherent
mental representation.
The result of the interface between narrative film and causeeffect schema the films fabula is not a pre-existing entity, but a
31
mental representation constructed by the spectator during her ongoing experience of the films syuzhet.11 The spectators construction of the fabula is therefore a procedural (rather than a static)
process. The fabula is in a constant state of change, due to the
spectators ongoing generation of new inferences, strengthening of
existing inferences, and abandonment of existing inferences.
The combined schemata plus procedural account of information
processing accurately describe the Constructivist theory of perception, a theory Bordwell explicitly adopts. The strength of Constructivism for film studies lies precisely in its emphasis on the procedural, top-down, defeasible nature of the perceivers activity in
information processing.
I have discussed Bordwell at length here because he can be
credited with pioneering a cognitive theory of filmic meaning. Yet
recent advancements in cognitive theory inevitably reveal the limitations of Bordwells theory: He conceives schemata as transcendental, functioning to construct literal meaning only, and isolates
them from both language and the body. George Lakoff and Mark
Johnson argue that schemata conceived in this way are static structures simply imposed upon perceptual input to give it meaning.
Lakoff insists that such traditional theories of schemata attempt to
provide a format for representing human knowledge in computational models of the mind. They attempt to do so by providing
conventional propositional structures in terms of which situations
can be understood.12 Because they only account for literal meaning, such theories exclude the imaginative projective devices that
both Lakoff and Johnson posit as being fundamental to cognitive
reasoning, devices such as metaphor and metonymy. As we shall
see, both authors call for an enrichment of the theory of schemata,
to move them away from a transcendental conception by taking
into account imaginative projective devices, and also by conceiving
them as dynamic, not static.
In contrast to Bordwells, Lakoffs and Johnsons schemata are
image-based, are embodied and inherently meaningful (are constituted by the structure of the body) rather than being transcendental,
are based on metaphor and metonymy, and are dynamic rather
than static. But such schemata are not made up of what we generally think of as mental images, since an image in this traditional
sense always refers to particular objects and events. Instead, the
32
emphasis remains with the term schema: Image schemata are nonrepresentational spatial structures; they delineate the abstract structure of images.
Bordwells cognitivism follows the philosophy of the subject in
that both involve disembodiment and the subjects isolation from
language. Firstly, in terms of disembodiment, Bordwell only considers the physiological characteristics of the eye when outlining
his schema theory. Like Descartes, Bordwell connects the eye only
to the mind, thereby separating both from the body. Secondly,
Bordwells isolation of language is evident in his rejection of a
communication model of narration. He asks, Must we go beyond
the process of narration to locate an entity which is its source?13
and answers, To give every film a narrator or implied author is to
indulge in an anthropomorphic fiction. . . . [This strategy takes] the
process of narration to be grounded in the classic communication
diagram: a message is passed from sender to receiver.14 As opposed to this communication model, Bordwell argues that narration
presupposes a perciever, but not any sender, of a message.15
By rejecting the narrator and the communication model of narration, Bordwell confers too much autonomy upon the spectator in
constructing filmic meaning. More generally and to express it in
Habermass terms cognitivism assigns to the spectator a narcissistically overinflated autonomy and purposively rational selfassertion. What this means is that cognitivism focuses on isolated,
purposively acting subjects at the expense of intersubjective communicative practices. By contrast, Habermas (and the Language
Analysis tradition generally) rightly stresses the need to conceive
of subjects as necessarily bound up in intersubjective, communicative relationships, which involve reciprocal recognition and social
interaction, made possible by language. In rejecting the narrator,
Bordwell adheres to the first person perspective of the philosophy
of the subject, isolating the spectator from intersubjective, communicative relationships (the third person perspective advanced by
the Language Analysis tradition), which is fundamental to the coordination of social action in human society.
In Film Semiology as a Cognitive Science16 the cognitive film
semiotician Michel Colin begins by taking Metzs statement to
understand how film is understood17 as the central aim of film
semiotics. This involves investigating the correspondence between
semantics and visual representation, without falling into the trap of
33
identifying semantics as specifically linguistic (the trap, Katz and

Fodor argue, that Chomsky falls into). The alternative is to align
semantics with conceptual structure (as do Katz and Fodor). For
Colin, to approach the issue of how film is understood, cognitive
film semiotics must describe the rules governing the links between
the visual representations of 3-D scenes and conceptual structures,
a description that largely implies the analysis of the links between
shots.18 This explains why Colin reread in great detail Metzs
grande syntagmatique, primarily within Chomskys Standard Theory
of transformational generative grammar (see Chapter 5).
But in Film Semiology as a Cognitive Science, Colin moves
closer to the position of Katz and Fodor by employing the framework of non-linguistic domains of cognitive science (but without
rejecting linguistics altogether, as we shall see). For Colin, Metzs
view that the aim of film semiotics is to understand how film is
understood presents film semiotics as being closely related to cognitive science, since (in Terry Winograds definition) cognitive science specializes in research on language, vision, and problem solving. From this premise, Colin maintains that
there is no need to insist on the fact that film semiology is concerned with
problems of vision and language; only the third field can seem unrelated
to the semiological approach. It is, however, reasonable to consider that
the interpretation of semantic phenomena, such a spatial-temporal relationships or co-reference, imply the solving of problems. Thus, it can be argued
that the understanding of the shot-reverse shot implies a solution to the
problem of determining the position of the characters within the diegetic
space on the basis of the filmic configuration (alternation between two
shots). More generally, it could be said that the codes postulated by the
semiological approach are an explanation of the knowledge necessary to
the solving of the problems raised by the interpretation of filmic configurations. For instance, a code such as the grande syntagmatique of narrative
film can be considered to represent the knowledge that the spectator needs
in order to solve the problems of spatio-temporal relationships within the
image track.19
Colins identification of filmic comprehension as a problem-solving

activity enables him to posit his object of study as the spectators
active, cognitive processes that constitute filmic meaning. This cognitive conception of the film-spectator interface posits that the
filmic text does not signify its own meaning, but is constituted by
those spectators who possess the necessary knowledge.
34
Following Gilbert Ryle, cognitive scientists argue that comprehension involves two fundamental types of knowledge: declarative
and procedural knowledge (i.e., knowing that and knowing how,
respectively).20 One of the tasks of cognitive film semiotics is to
identify the specific knowledge needed to understand cinematic
language, and how spectators use that knowledge in the process of
comprehension. Cognitive film semiotics therefore strives to describe the filmic intuitions of the ideal film spectator, to characterize
what a film spectator knows when she is able to understand a film.
Colin approaches this task by considering how artificial intelligence
(AI) conceptualizes the representation of knowledge, which he considers under three headings:21
(a) Acquiring more knowledge
(b) Retrieving relevant facts from knowledge
(c) Reasoning about the facts in search of a solution
I shall outline and extend Colins discussion of these categories.

Acquiring More Knowledge. Category (a) involves the relation
between knowledge that is given and that which is new. Colin
argues that film semiotics has previously concentrated only on
given knowledge that is, has never considered how knowledge
(both given and new) is acquired. Colin refers to the notion of
encyclopedic competence, a term deriving from the distinction between the encyclopedia and the dictionary.22 But such a conception
is based almost entirely on declarative knowledge. That is, encyclopedic knowledge, consisting of a series of interconnections between
manifest chunks of information, does not take into account the
active, generative capacity of the mind. But recent research into
knowledge and cognition has shifted emphasis toward the dynamic, generative workings of the mind. For example, in a discussion of their principle of relevance, Dan Sperber and Deirdre Wilson argue that people have not only the knowledge that they
actually entertain, but also the knowledge that they are capable of
deducing from the knowledge that they entertain.23 For Sperber
and Wilson, an individual does not have to be aware of all her
knowledge (an assumption implicit in the encyclopedia model of
knowledge), since an individuals knowledge consists of what she
knows together with the generative capacities that allows her to
generate additional knowledge from what she knows. And the
35
individual is not aware of this additional knowledge until it is

generated.
In order to advance on the question of given/new knowledge,
film semiotics needs to shift its emphasis to procedural aspects of
knowledge and must also identify mechanisms of knowledge generation and acquisition.
Retrieving Relevant Facts from Knowledge. Category (b) is dependent on how knowledge is organized. Colin refers to semantic
networks (a hierarchy of levels of knowledge, from general to specific) and to frames and scripts (both of which are types of schemata). But again, these are static, declarative models of knowledge.
In Sperber and Wilsons dynamic model, the retrieval of knowledge
also involves the generation of knowledge, as the following example shows:
Suppose Mary and Peter are looking at a landscape where she
has noticed a distant church. She says to him,
(49) Ive been inside that church
She does not stop to ask herself whether he has noticed the
building, and whether he assumes she has noticed, and assumes she has noticed he has noticed, and so on, or whether
he has assumed it is a church, and assumes she assumes it is,
and so on. All she needs is reasonable confidence that he will
be able to identify the building as a church when required to:
in other words, that a certain assumption will be manifest in
his cognitive environment at the right time.24
Sperber and Wilson are arguing against the mutual knowledge

hypothesis, which states that, in order for the receiver to retrieve
the same meaning from the message as that encoded by the sender,
the sender and receiver must not only share the same system of
codes, but know that they share this system. Mutual knowledge is
therefore knowledge that is shared and known to be shared. But
knowing that the knowledge is shared leads to an infinite regress,
or an infinite process of checking. In the preceding example, for the
knowledge that the building is a church to be mutually known,
Mary must know that Peter knows that the building is a church,
and Peter must know that Mary knows that the building is a
church. But then, Mary must know that Peter knows that she
(Mary) knows that the building is a church, and Peter must know
36
that Mary knows that he (Peter) knows that the building is a

church, and so on. For mutual knowledge to be established, Mary
and Peter must not only know the same knowledge, but must know
that they share this knowledge, and know that they know that they
share this knowledge, and so on ad infinitum.25 It is the psychological unreality of the mutual knowledge hypothesis that makes it
implausible, and it seems that the static, encyclopedic model of
knowledge is also based upon the mutual knowledge hypothesis.
Reasoning about the Facts in Search of a Solution. In category (c),

Colin refers to procedural knowledge to ways in which facts are
processed by individuals. As well as procedures, Colin also mentions the following strategies: analogy, formalization, generalization, and abstraction. Lakoff and Johnson have reduced these processes down to two metaphor and metonymy whereas Sperber
and Wilson have identified one process the principle of relevance.
Colin attempts to overcome the lack of procedural knowledge processes in film semiotics in an analysis of an ideal spectators comprehension of spatial relations in a scene from The Barefoot Contessa
(Joseph Mankiewicz). To do this he employs (without introducing)
the general space planner (GSP) of Charles M. Eastman,26 an early
attempt at computer aided design. It is a computer program designed to overcome space planning problems the organization
and arrangement of objects (furniture, wall partitions, etc.) within a
bounded space. The program contains the necessary number of
constraints (the distance needed between furniture, locations where
furniture cannot be placed, and so on), and is asked to compute the
optimal solution.
The appeal of the GSP in explaining how spectators mentally
construct filmic space should be obvious. The GSP is a procedural
process explicitly constructed as a problem-solving activity. By using the GSP, Colin imputes that each spectators construction of
filmic space is a problem-solving activity, in which filmic space is
gradually pieced together by the spectator, who is following a series of constraints. Such constraints correspond to the rules of continuity editing. Failure to construct a coherent space will be due to
ill-defined constraints (either an insufficient number or contradictory constraints), or by the spectators improperly locating a shot
within the bounded space (the film scene). But the more the con-
37
straints are well defined, the more efficient the construction of the
bounded space will be.
The scene Colin analyzes occurs near the beginning of the film,
as four characters enter a cabaret (the bounded space) and sit
around a table. Much of the scene consists of shots depicting one
character only, so the spatial position of each character in relation
to the others needs to be worked out (and reinforced) by each
spectator in a procedural manner. Colin notes: It can then be
assumed that the spectator goes through the mental process of
reconstructing the spatial relationships between the characters on
the basis of successive shots of each one.27 Colin also mentions
that the entrance and the table serve as stable anchoring points of
the scene, even though they are not shown at the same time.
The general process Colin is examining is that of constructing a
coherent mental representation of what is not actually experienced
in toto the spatial layout of the cabaret, together with the four
characters sitting around a table. Colin argues that, once a spectator
has mentally constructed this space, a few contradictory cues (such
as the breaking of the 180 degree rule) can be rectified, since it
stands out as an anomaly (or the contradictory cues can be said to
be displaced codes that leave traces of their original position). However, it is possible that a series of contradictory cues will force the
spectator to revise her mental representation of the scene.
As does Bordwell, therefore, Colin acknowledges that comprehension is a procedural, top-down, defeasible process continually
under review. The spectator is ceaselessly attempting to solve problems when watching a film constructing the optimal filmic space
(and, of course, time and narrative) from the cues offered by the
text. Colin and Bordwell differ only in the way this process is
theorized. Although Colin moves into non-linguistic areas of cognitive science, he does not reject linguistics as a paradigm for film
studies. At the end of Film Semiology as a Cognitive Science, he
addresses the issue of the relation between cognitive and linguistic
faculties. In effect, the debate is between modular and non-modular
theories of the mind. The modular view is advocated most strongly
by Jerry Fodor,28 who posits a theory of the mind consisting of
separate cognitive domains (or input systems) with the following
properties: Each system is domain-specific (it specializes in processing a particular type of stimulus only), innately specified (i.e.,
it is not learnt, but already known); hardwired (it can be associ-
38
ated with neural systems), and autonomous (it does not share
memory, attention, and so on, with other cognitive systems).29 No
module has access to information in other parts of the mind, including those containing general cognitive abilities. For Fodor, grammar
is a module whose sole domain is the processing of linguistic input;
as such, it is insensitive to all other cognitive capacities.
Bordwell implicitly adopts at least a number of the tenets of the
modular theory of the mind especially the domain-specific and
autonomous nature of information processing. In Narration in the
Fiction Film he states: It will come as no surprise that I do not treat
the spectators operations as necessarily modeled upon linguistic
activities.30 Furthermore, he does not consider the rhetorical tropes
of metaphor and metonymy to be relevant to the construction and
comprehension of the films fabula, since he considers metaphor
and metonymy to be linguistic tropes.31
But for Colin: As far as film semiology is concerned, the problem is then knowing whether the homologies between linguistic
and filmic structures depend on linguistic faculties, or derive from
the fact that language and film call for the same mental faculties.32
Colin confronts this issue by quoting Ray Jackendoffs claim There
is a single level of mental representation, conceptual structure, at
which linguistic, sensory, and motor information are compatible.33
This view of the mind then allows for the possibility that linguistic
structure interacts (i.e., is compatible) with visual data, that language does have a role to play in the mental processing of nonlinguistic information. Jackendoff calls this the cognitive constraint: There must be levels of mental representation at which
information conveyed by language is compatible with information
from peripheral systems such as vision, nonverbal audition, smell,
kinesthesia, and so forth. If there were no such levels, it would be
impossible to use language to report sensory input. We couldnt
talk about what we see and hear.34 But this is not to advocate a
return to the translinguistics of the sixties, of reducing all semiotics
to linguistic meaning, far less advocating substantive parallels between film and language.
Bordwells (untheorized) modular view of the mind forecloses
on questions of the role of language in the processing of filmic
texts. It seems that he has adopted this position on the basis of his
criticism of past results on the application of linguistic concepts to
film.35 But even if all of Bordwells criticisms were justified, this
39
does not, a priori, rule out the applicability of other linguistic concepts to film. The problem with Bordwells cognitivism is that he
has rejected the communication model of narration, the role of the
narrator, and has developed a disembodied theory of schemata.
Colins cognitive film semiotics avoids the first two problems by
remaining within a communicative framework, although he does
not take into consideration the role of the body in filmic comprehension.
Cognitive Semantics
In his essay Semiotic Foundations of the Cognitive Paradigm,36
Winfried Noth considers the affinities between semiotics and cognitive science. Noth asks, Is semiotics one of the paradigms that
have been replaced by cognitive science, has the cognitive turn in
the humanities resulted in a paradigm shift within semiotics, or has
semiotics remained unaffected by the cognitive turn?37 Noth goes
on to argue that the paradigms cannot simply be compared and
contrasted since there are many forms of cognitivism just as there
are many schools of semiotics. Nonetheless, Noth does argue that,
although cognitive science has many affinities with Peircean semiotics, it has very little in common with Saussurean semiotics. This
is particularly evident when he compares Saussurean semiotics to
George Lakoffs experientialist view of cognition, which is based
on the premise that thought and language are fundamentally motivated by bodily experience.
Lakoffs view is shared by a growing number of researchers
who study the interface between cognitive science and linguistics
including Mark Johnson, Mark Turner, Ronald Langacker, Gilles
Fauconnier, and Eve Sweester.38 (It is generally regarded that
George Lakoff and Mark Johnson are the dominant representatives
of cognitive semantics, so I shall focus on them.) Lakoff and Johnson trace the origins of intangible abstract thought back to the body,
rather than consider thought to exist in an autonomous realm. They
therefore challenge what is known within Western philosophy as
objectivism.
One dominant issue that marks the difference between objectivism and cognitive semantics is that of mental representation.
Whereas objectivists believe that knowledge is represented in the
mind by propositions consisting of meaningless symbols (which
40
gain their meaning via a direct correspondence to external reality),

cognitive semantics posits that knowledge is represented in the
mind in the form of schemata (cognitive structures that organize
perceptual input into experiences). More specifically, cognitive semanticists such as Lakoff and Johnson have posited a series of image
schemata that structure perceptual input into experiences. These
schemata are inherently meaningful because they gain their meaning directly from the bodys innate sensory-motor capacities. Image
schemata therefore represent the body in the mind (to use the title
of Johnsons book) and are posited as being cognitively real because
they are directly motivated (non-arbitrary) and inherently meaningful. Cognitive semantics therefore challenges the dualism (the
mind-body problem) of Cartesian philosophy. Descartes defined
the body as a machine determined by mechanical laws, which can
be described by means of the principles of mechanics. But introspection reveals that the mind cannot be described in the same
way. Descartes therefore posited the mind to be made up of a
second material (res cogitans, or thinking substance), which is not
reducible to the principles of mechanics. It is this second substance
that distinguishes humans from non-humans, according to Descartes. But if the mind and body are two completely separate substances, how do they interact? This dualism, one of the dominant
principles of objectivism, is overcome in cognitive semantics by
establishing a determinate link between the body and cognition. In
effect, cognitive semantics mediates between dualism and its polar
opposite, eliminative behaviorism, which repudiates the existence
of any mental events or states by reducing them to bodily behavior.
In rejecting both dualism and eliminative behaviorism, cognitive
semantics defines itself as a form of mental materialism, in which
mental phenomena are dependent on (or are realized by) the material (the body), but are not mechanically reducible to the material.
The problem both Lakoff and Johnson have with objectivism is
that it bypasses humans. Reasoning is believed to be based entirely
on propositional structure and literal meaning operating autonomously from human experiences, with little or no room for image
schemata that gain their meaning from (or are directly motivated
by) the body, and imaginative processes of thinking, such as metaphor and metonymy. The objectivist approach conceives cognition
to be transcendental and abstract as disembodied, a premise that
perpetuates the Cartesian mind-body problem, as we have already
41
seen (a problem, incidentally, maintained by cognitivists who study

the mind as a computer). Lakoff and Johnson instead emphasize
the human nature of the mind and cognition or, more specifically,
argue that cognition is embodied. Lakoff writes that where objectivism defines meaning independently of the nature and experience
of thinking beings, experiential realism characterizes meaning in
terms of embodiment, that is, in terms of our collective biological
capacities and our physical and social experiences as beings functioning in our environment.39 Lakoff is using a broad definition of
the term experience, one that includes everything that plays a role
in constructing human understanding, from the genetic makeup of
our bodies to the way the body interacts with its social environment. Lakoff stresses that this is not simply a return to classical
empiricism, in which experience is understood as passively received sense impressions, since experience does not determine conceptual systems, but only motivates them.40 Johnson similarly rejects the classical empiricist definition of experience, defining it
instead in socio-historical terms.41
To specify what they mean by embodiment, Lakoff and Johnson
make the distinction between conceptual structure and preconceptual bodily experiences and employ the notion of embodiment to
argue that conceptual structure arises from (or is motivated by)
preconceptual bodily experiences. The distinction is not therefore
based on the opposition between structured concepts and unstructured experiences, for both authors argue that experience is itself
structured and is already meaningful.
Lakoff and Johnsons cognitive semantics therefore challenges a
number of the basic assumptions of the Language Analysis tradition. Firstly, cognitive semantics presents a striking alternative to
the structural linguistic assumptions that the sign is arbitrary and
that meaning arises out of non-meaning. Secondly, Lakoff developed cognitive grammar from generative semantics, which he
helped to formulate in the sixties. Briefly, generative semantics was
a response to Noam Chomskys Standard Theory of Transformational Generative Grammar (outlined in Chapter 5). Generative semantics took many of its premises from the Standard Theory but
refuted the claim that the deep structure is a formalized system of
uninterpreted (meaningless) symbols that later receive a semantic
interpretation. For the generative semanticists, the deep structure is
the semantic interpretation (is already meaningful). Lakoffs cogni-
42
tive semantics discards all traces of transformational grammar to

develop the idea that language is already meaningful and is motivated by bodily experiences. This emphasis on bodily experiences
is also adopted by Johnson, although he refers to his position as
descriptive or empirical phenomenology.42
Lakoff characterizes experience in terms of kinesthetic image
schemata, which are simple structures that arise from the body
updown, backfront, centreperiphery, partwhole, insideoutside, paths, links, forces, and so on. These schemata are directly
constrained by the dimensions of the human body. And because
the dimensions of the fully grown body are shared (uniform and
constant), any discussion of conceptual structure in terms of the
body does not fall into radical relativism and subjectivism. Image
schemata are not, therefore, arbitrary, but are directly motivated by
a shared and constant bodily experience.
The structure of our shared bodily experience then becomes the
basis for rational, abstract thought by means of image based schemata and creative strategies such as metaphor and metonymy,
which project and extend this structure from the physical domain
into the abstract domain of concepts. As Johnson observes:
Through metaphor, we can make use of patterns that obtain in
our physical experience to organize our more abstract understanding43 (examples are provided later).
The basic premises of experientialism are represented schematically in Figure 3. This figure shows that what is taken to be an
autonomous realm in objectivism (conceptual structure) is shown
in experientialism to be the end result of a series of processes. We
see that preconceptual experience is constituted by image schemata.
Lakoff outlines several kinesthetic image schemata and shows how
they determine the structure of abstract conceptual thought:44
1. The container schema, which structures our fundamental awareness of our bodies, is based on the elements interior, boundary,
and exterior. In terms of metaphorical extension, Lakoff notes
that the visual field is commonly conceived as a container, since
things come into and go out of sight. Also our understanding
of our emergence out of various states of mind (drunkenness,
dizziness, and so on) is also a metaphorical extension of the
container schema. Theories are commonly understood as containers for example, we say that a theory is watertight, or cannot
43
Conceptual structure
(abstract reasoning)
Metaphor, metonymy
Fig. 3
Preconceptual experience
(image schemata, basic level categorization)
Social and physical reality, the body, etc.
hold water, or we refer to a theorys framework (as we shall see,

the frame as container is fundamental in the comprehension of
films). Sometimes, theories are understood in terms of a particular type of container, a building: For example, we commonly say
that a theory has no foundations, needs shoring up, has collapsed, and so on.
Johnson calls the container schema the inout schema, structured on our physical experience of our bodies as both containers
and as objects that can be contained.45 However, these two schemata are not identical: The inout schema has a broader scope,
defining the boundary between what is contained (in) and what
is excluded from the container (out).
2. The partwhole schema, also fundamental to our self-awareness,
is based on our perception of our bodies as wholes made up of
parts. It is extended in numerous ways, by metaphorical projection, into the realm of abstract thought. Lakoff mentions that
families are understood in terms of the partwhole schema.
3. The link schema is based on our awareness of our position in
relation to others, in which this sense of position enables us to
establish a link between self and other. Social and interpersonal
relations are metaphorically conceived in terms of making con-
44
nections or breaking social ties, whereas freedom is understood

as the absence of any links tying us down.
4. The centreperiphery schema is similar to the partwhole schema
in that it is based on our awareness of our bodies as having
centres (the trunk and internal organs) and peripheries (limbs,
hair, etc.). The centre gains its importance from the basic fact that
the preservation of the body is more important to survival than
the peripheries. Again, many abstract entities (such as theories)
are metaphorically understood in terms of this schema.
5. The source-path-goal schema is based on our experience of bodily
movement in a particular direction from one point to another
along a path. Its many metaphorical extensions include its structuring of ones long-term aims and ambitions, which become
sidetracked or blocked by obstacles.
In addition to covering some of these schemata, Johnson discusses

the balance schema. As are other image schemata, it is based upon
our bodily experience (a symmetry of forces relative to an axis) and
can be metaphorically mapped onto other domains to give them
structure. Johnson considers how balance operates in visual artworks and other phenomena (including theoretical arguments) that
are understood in terms of balance. Whereas our bodily sense of
balance is based upon gravitational and physical forces, in artworks
this physical bodily experience is metaphorically mapped onto vision (and in theoretical arguments, balance is mapped onto concepts). From the balance schema, cognitive semantics is able to
answer the question why symmetry and balance are pleasing in
artworks because they imitate the symmetry and balance of the
perceivers body. The process of reasoning and comprehension can
be described using Lakoffs theory of experientialism in terms of
the ability to form complex concepts by projecting several image
schemata into the conceptual domain.
A Cognitive Semantics of Film

In this final section, I want to go beyond the work of Metz, Bordwell, and Colin by developing a cognitive semantic theory of film
one that takes into consideration the role of the body in filmic
perception. In applying Lakoff and Johnson to film, I want to refer
to Figure 3 and emphasize that the causality is not just one way.
45
That is, abstract reasoning is not merely an effect of bodily experience, since I shall argue that abstract reasoning (or, in this
instance, comprehension of a film) changes the way we experience
our bodies.
A general point that needs to be repeated about image schemata
is that they are directly meaningful, since they put us in touch
with preconceptual structures in our bodily experience of functioning in the world.46 Just as kinesthetic image schemata are not
arbitrary, but are directly motivated by (and directly obtain their
meaning from) bodily experience, so the metaphorical projection of
image schemata onto conceptual structure is also motivated by
bodily experience. Here we see Lakoff and Johnson eliminating the
semiotic principle of arbitrariness from the fundamental processes
of meaning and cognition.
How are the metaphorical projection processes of meaning employed in film viewing? I shall limit myself to discussing what I
consider to be the most fundamental kinesthetic image schemata
relevant to the comprehension of film the container (and in-out)
schema.
Whereas in daylight we are able us to orient our bodies in
relation to the environment (since we can perceive distance, depth,
extension, surface, and so on), in the dark we become disoriented,
because we cannot perceive distance between us and the environment. This breaking down of the relation between self and other
affects ego boundaries, as Eugene Minkovski points out:
[Dark space] does not spread out before me but touches me directly, envelops me, embraces me, even penetrates me, completely passes through me,
so that one can almost say that while the ego is permeable by darkness it
is not permeable by light. The ego does not affirm itself in relation to
darkness but becomes confused with it, becomes one with it. In this way
we become aware of a major difference between our manners of living
light space and dark space.47
This means that, in the dark, the individuals consciously experienced tactile body image can no longer function as the basis for
meaning, since it cannot establish a relation to its environment. In
the darkened auditorium, the bright, delimited screen serves to
rescue the spectators tactile body image; it re-engages the spectators body, offers it a stable reference point, and returns the body
to its role of generating meaning. As perceived in the auditorium,
46
a film works against the darkness that surrounds it. Correlatively,

each spectator seeks out the bright image on screen in order to
become oriented and to create a meaningful environment again.
Unlike everyday life outside the cinema, the darkness of the auditorium works in a superlative manner to engage each spectators
body image, and thus her kinesthetic image schemata.
It is not, of course, only the darkness that primes the spectator
it is also the screen, frame, narrative structure, and the process of
suture. These terms form the basis of the modern film theorists
concept of subject positioning. Crucial to this concept is the films
frame, which serves to fix the spectators attention to the filmic
image projected on screen. It achieves this by determining the
boundary between what is included and what is excluded from the
spectators vision at any one moment. The function of the frame is
therefore literally to position the spectator in a spatial relation to
the image, fixing her visual and aural attention on screen.48 Stephen
Heath, one of the most prominent of the modern film theorists,
then metaphorically mapped the function of the frame onto narrative. (Like Lakoff and Johnson, therefore, Heath begins with a literal, spatial experience, then metaphorically maps this onto the
abstract domain of narrative.) Heath argues that narrative is a
decisive instance of framing in film, since it checks the images,
centring and containing, prescribing a reading as correlation of
actions and inscribing a subject as, and for, the coherence of that
operation, carried through against possible dispersion, the multiple
intensities of the text of the film.49 Heaths main model here is a
tension between containment and dispersion a containment of
excess carried out by means of frame and narrative, which in their
turn ensure a stable subject position (now understood in both literal
and metaphorical terms).50
In cognitive semantic terms, basic filmic comprehension consists
of the automatic projection onto film (filmic space and time, as well
as narrative) of preconceptual kinesthetic image schemata, by
means of metaphor and metonymy, forming the films basic structure and level of meaning. This process primarily involves the
projection of image schemata onto vision, which engages the image
on screen. Some of the schemata are invariant (in-out, container
schemata), whereas others are variant (such as source-path-goal).
In the final pages of this chapter I shall begin applying the container
schema to the cinema.
47
We have seen that Heath has already used the term containment when theorizing subject positioning in terms of both frame
and narrative. Using Lakoffs container schema, I shall explore both
the frame and the diegesis. To say that the frame and diegesis are
understood in terms of a kinesthetic image schema is to suggest
that they are comprehended in terms of our experience of our
bodies as containers. The frame in particular is analogous to (or is
a reduplication of) sight itself, which, as Lakoff remarks, is also
understood in terms of the container schema. But before discussing
frame and diegesis, it is necessary to contextualize them. The term
diegesis was first introduced into film studies by Etienne Souriau,
who distinguished seven levels of filmic reality:51
1. Afilmic reality (the reality that exists independently of filmic reality)
2. Profilmic reality (the reality photographed by the camera)
3. Filmographic reality (the film as physical object, structured by
techniques such as editing)
4. Screenic (or filmophanic) reality (the film as projected on a
screen)
5. Diegetic reality (the fictional story world created by the film)
6. Spectatorial reality (the spectators perception and comprehension of a film)
7. Creational reality (the filmmakers intentions)
Souriaus discussion of the way these levels of reality relate to one

another can be reformulated in terms of the container and inout
schemata. The inout schema delimits the first boundary between
afilmic and profilmic reality. The afilmic exists outside the realm of
the cinema, whereas the profilmic exists inside. But both these
levels are then defined as existing outside the filmic text. This
establishes a new boundary between extra-textual and textual
reality. Whereas the afilmic and profilmic are extratextual, the following three types of filmic reality are textual and are structured in
relation to one another in terms of the container schema: The filmographic contains the screenic, and the screenic contains the diegesis. The final two levels of filmic reality are cognitive, referring to
the spectators comprehension of film and the film as conceived by
the filmmaker.
In Souriaus terms, the frame exists on both the filmographic
and screenic levels of filmic reality and serves as a container for the
48
diegesis. But the relation between frame and diegesis in the cinema
is complex. In painting and film, the frame serves as a boundary
between what is seen and what remains unseen. Nonetheless, there
is a fundamental difference between the function of the frame in
painting and that in film. In painting, the frame acts as an absolute
boundary; it unequivocally severs the bounded space from its surroundings. But in cinema, the frame is mobile. The boundary between bounded and unbounded space is equivocal. In filmic terms,
there is an opposition between on-screen space and off-screen
space. On-screen and off-screen space are themselves understood
in terms of containment, since they contain the films diegesis, or
fictional story world. So the container schema operates at two levels
of the film the frame (which acts as a boundary between onscreen and off-screen space) and the diegesis (which acts as a
boundary between fiction and non-fiction). Off-screen space has an
unusual semiotic specificity, since it exists between the filmographic
level and the screenic level of filmic reality. It is an invariant property of filmographic reality, but a variant property of screenic reality. In these terms, we can think of it as the potential, non-manifest
stage of on-screen space.52
Off-screen space is therefore that part of the first container (the
diegesis) that does not appear on screen and in frame ( the second
container) at any one moment. The space of the auditorium, discussed earlier, can then be considered as the third and ultimate
container. Although visual stimuli are contained within the frame
and diegesis, sound is only contained by the third container, the
auditorium. The auditorium is a container that contains the screen
and frame. The frame is a container that contains the diegesis, the
fictional story world (characters, settings, and actions). The auditorium, frame, and diegesis are container objects; what they contain
are container substances the auditorium contains the screen and
frame (filmographic and screenic reality); the screen and frame
contain the diegesis; the diegesis contains the fictional story world.
But the meaning of the container (or in-out) schema is equivocal
when understood in terms of the cinema. In can mean something
hidden, unavailable, or unnoticed, the direct bodily meaning of in.
But in the cinema, it means the opposite object x is in the image.
Out also has two meanings something previously hidden is
brought to the forefront (again, the direct bodily meaning of out),
or the opposite, as in object x is out of the image. This equivoca-
49
tion disappears only when we understand contained in the cinema

as a transparent container. So the cinema cannot be understood in
direct analogy with the body, but in modified terms, since the body
as an opaque container does not directly map onto the frame and
diegesis, which are understood as transparent containers. It should
now be quite evident that this transparent container is not a preexisting entity (i.e., part of what is projected on screen), but is a
default setting, a part of the spectators competence, and it is precisely its status as a default setting that encourages the spectator to
construct the films diegesis mentally in her ongoing viewing of the
film.
The frame can be comprehended in more specific ways as well.
For example, whereas the framing of an objective (non-focalized)
shot is comprehended as representing the vision of a non-diegetic
narrator (unless of course you argue, like Bordwell, that nondiegetic filmic narrators do not exist), a subjective (or focalized)
shot is comprehended as embodied as representing the vision of a
character existing in the diegesis. The frame in a subjective shot is
therefore comprehended as being controlled within the diegesis,
whereas the frame of an objective shot is perceived as disembodied
and non-diegetic but embodied in the film spectators body. Of
course, the subjective shot is also ultimately embodied in the film
spectator, but that embodiment is duplicated in the films diegesis.
Furthermore, what consequences does the mobile filmic frame
have in terms of the body? Firstly, the fixed frame of painting
reaffirms the fixed boundary of the body. But in the cinema, the
frame is potentially mobile (its potentiality therefore defines it as a
sub-code). But a mobile frame does not inherently signify anything.
However, in conjunction with the narratological coding of the
frame as subjective or objective, camera movement can be said to
affect the bodys absolute boundary between container and contained. If comprehended as subjective, the mobile frame represents
the movement of a character within the diegesis (the frame is embodied in a mobile character). But when comprehended as objective, the mobile frame gives her vision a sense of transcendence.
The enormous amount of writing devoted to the subjective (point
of view, or focalized) shot attests the importance in perceiving a
shot as objective or subjective.53 A number of directors (such as
Fritz Lang and Steven Spielberg) and genres (such as the horror
film) play on the ambiguity of shots, in terms of whether they are
50
subjective or objective, which determines the way we experience

their films.54
The narrative film in itself is understood as a container, as is
evident in everyday comments people make after seeing a film
There was not much in the film (the evaluation of a film employs
terms of quantity the more it contains, or the more that happens,
the better); I became immersed in (or absorbed into) the film (here
the spectator metaphorically views herself as an object drawn into
the film as container); and so on. Heath employs the term suture
to describe the process that stitches the spectator into an imaginary relation to the filmic image. By contrast, from a cognitive
semantic perspective we can argue that the spectators projection
into the film is in fact the result of the container schema. This
schema, directly motivated by the spectators body, is metaphorically projected onto the film by the spectator in her ongoing
construction of a mental representation of the films diegesis as a
coherent container. In Colins example from The Barefoot Contessa,
the spectator is involved in constructing the cabaret room as a
container, and it is the successful completion of this activity that
absorbs the spectator into that space. Furthermore, the activity is
facilitated by the frame, which necessarily excludes as well as includes. The frames very act of exclusion prompts the spectator to
begin mentally constructing the space of the room.
Lakoffs other schemata can also be applied to narrative films,
with variable results. In terms of the partwhole schema, a narrative film is of course understood to be a whole made up of parts
shots and scenes. One of the main issues addressed by the structural linguisticbased film semiotics was to identify these parts by
means of segmentation and classification. The partwhole schema
is closely aligned to the link schema, since a collection of parts can
only be perceived as a whole if the parts are linked coherently.
Many of the techniques of continuity editing achieve this impression of coherence; they can therefore be understood in terms of the
link schema. Some shots are central to a scene, such as the establishing shot, whereas others are less central, such as an analytical cutin. A hierarchy of shot types could be established using the centre
periphery schema. Finally, narrative trajectories can be understood
in terms of the source-path-goal schema, as Jan Simons has already
demonstrated.55
This chapter only represents the beginnings of a cognitive se-
51
mantic theory of film. When developed further, cognitive semantics

will enable film scholars to overcome one of the primary limitations
of both cognitive and psychoanalytic theories of spectatorship
their disembodied accounts of perception. Perception is not a process that only involves a relation between the eye and the mind
(whether conscious or unconscious); more fundamentally, it involves the metaphorical projection of the body on screen and in
frame.
CHAPTER THREE
Not What Is Seen through the Window but the

Window Itself
Reflexivity, Enunciation, and Film
In film, when enunciation is indicated in the utterance, it is not, or

not essentially, by deictic imprints, but by reflexive constructions.
. . . The film talks to us about itself, about cinema, or about the
position of the spectator. (Christian Metz)1
hroughout his career, Christian Metz was preoccupied

either directly or indirectly with reflexivity in film. It is
possible to identify three key moments: (1) his study of mirror constructions (construction en abme) in Fellinis film 812; (2) a theory of
enunciation based on Emile Benvenistes distinction between histoire (hi/story) and discours (discourse); and (3) his seminars of the
late eighties, culminating in his book LEnonciation impersonnelle ou
le site du film.2 In this chapter I will be investigating Metzs cognitive
film semiotics (his theory of impersonal filmic enunciation), outlining the concepts it is based upon particularly reflexivity, but also
metalanguage and anaphora and contrasting it with Francesco
Casettis cognitive film semiotics, a deictic theory of enunciation
based on personal pronouns.3 I shall attempt to articulate what I see
to be the main difference between Casetti and Metz: Whereas Casetti models his theory on speech (on face-to-face conversation),
Metz implicitly models his theory on writing. Metz rejected a pronoun-based theory of enunciation because, he argues, pronouns are
not appropriate to describing the reality of film. But Metz also rejected all deictic concepts (concepts that designate how film is oriented to its contexts of production and reception). I agree with Metz
in rejecting Casettis deictic theory of filmic enunciation (based on
personal pronouns), but I do not agree with his rejecting deixis as a
concept for describing the reality of film and other audio-visual
media. In contrast to Metz and Casetti, I suggest we understand
deixis as a broad cognitive concept, rather than a narrow linguistic
52
REFLEXIVITY, ENUNCIATION, AND FILM
53
one (which equates deixis with pronouns). Once we have freed

deixis from its purely linguistic manifestation, we can use it to analyze how other audio-visual media are oriented to their contexts of
production and reception. Toward the end of this chapter I shall reprieve the concept of deixis from Casettis pronoun-based theory
and from Metzs critique by demonstrating its value in describing
the experience of watching television, as well as its usefulness in explaining the recent process of digital colorization of black and white
films. Finally, I shall consider the implications of Metzs theory of
impersonal enunciation for the textual analysis of films.
Reflexivity, Metalanguage, Anaphora

To identify a filmic text as reflexive is to suggest that it refers to
itself that is, foregrounds its own processes of production (its
production as text). Here we see the explicit link between reflexivity
and enunciation, because enunciation is precisely a process of production or mediation that transforms the underlying language system, la langue, into a text. This process of transformation leaves
traces on the text, and it is the foregrounding of these traces that
makes a text reflexive.
This connection between reflexivity and enunciation is completely absent from Metzs discussion of Fellini. However, in his
paper Histoire/Discours this link becomes more apparent, although Metz uses the term exhibitionism instead of reflexivity.
In its initial stages, the theory of filmic enunciation was dependent
almost exclusively on Benvenistes distinction between enunciation
and the utterance and, within the utterance, the distinction between
histoire and discours. Metz transfers Benvenistes two forms of utterance to a psychoanalytical theory of vision: identifying exhibitionism with discours and voyeurism with histoire. The exhibitionist
knows that she is being looked at and acknowledges the look of the
spectator, just as discours acknowledges the speaker and hearer of
the utterance, whereas the object of the voyeurs gaze does not
know that she is being watched. The voyeurs look is secretive,
concealed, like the marks of the speaker and hearer in histoire. Metz
argued that classical narrative film is primarily voyeuristic hence
histoire for it conceals its own discursive markers (here, the spectators look), an activity other post-structural film theorists identified with films ideological function and feminists with films patri-
54
archal function (for the look is traditionally conceived as masculine).

Benvenistes theory of enunciation therefore fits in quite well
with the hierarchy dominating film semiotics. Enunciation denotes
a process that mediates between the underlying non-perceptible
system and the perceptible level. In fact, we could argue that enunciation denotes the major process mediating between the nonperceptible and the perceptible. The other concepts suture, disavowal, patterns of repetition/alternation, and so on arguably
designate more specific processes of enunciation that mediate between the non-perceptible and the perceptible.
Within the theory of filmic enunciation, all films were considered to be constituted by marks of production and reception; the
classical narrative film was then identified as a type of filmic text
that attempts to conceal these marks (discours posing as histoire).
Foregrounding the process of enunciation (making the text reflexive) breaks the illusion of continuity and reveals the codes that
constitute this illusion. This is why the concept of enunciation enabled film theorists in the seventies to distinguish classical from
modernist films those that conceal their traces of enunciation, and
those that emphasize these traces, respectively. Modernist films
were then valorized because they implemented the same strategy
as semiotics that is, an investigation of the inner logic of filmic
discourse.
Metzs theory of impersonal enunciation continues to discuss
film in terms of reflexivity and enunciation, but this time with the
concepts of metalanguage and anaphora, rather than Benvenistes
distinction between histoire and discours. Metzs shift in terminology
also marks a shift in his conception of filmic enunciation, since the
first approach, based on Benveniste, finds a place of deictic terms,
whereas the second, based on metalanguage and anaphora, rejects
all reference to deixis.
As with most original theoretical texts, Metzs theory of impersonal filmic enunciation does not serve as an adequate introduction
to its own concepts. Here I shall outline these concepts before moving on to discuss Casettis deictic theory of filmic enunciation.
For meaning to be determined, every language must have the
capacity to refer to itself, as well as to extra-linguistic reality. By
referring to itself, language becomes metalanguage, in opposition
to object language, language referring to extra-linguistic reality.
55
Metalanguage is therefore one of the dominant ways in which

language is used reflexively. In the following sentence (a constant
source of irritation during my childhood)
(1) You should never say never
The first use of never belongs to the object (or first order) language, and the second to the meta- (or second order) language.
Whereas signs functioning as object language are denotational,
signs functioning as metalanguage are non-denotational, since they
refer to language, not extra-linguistic reality. Metalanguage is not
completely autonomous from the object language, since the same
signs are used by both languages. The object languagemetalanguage distinction is therefore a functional, rather than an ontological, distinction. Metalanguage cannot be reduced to a number of
particular textual features, but refers to one potential function of all
textual features. Metz transfers this insight to film: Cinema does
not have a closed list of enunciative signs, but it uses any sign . . .
in an enunciative manner, so that the sign can be removed from the
diegesis and immediately come back to it. The construction will
have, for an instant, assumed an enunciative value.4 This is one of
the most radical and far-reaching conclusions of Metzs theory of
impersonal enunciation, and I shall discuss it in more detail at the
end of this chapter.
It is self-evident, then, that metalanguage is reflexive. For most
linguists, reflexivity and metalanguage are essential attributes of
natural language. But can these structures be located in cinematic
language? Metz certainly believes so, and in his application of them
to film, he emphasizes that they are forms of enunciation: All
figures of enunciation consist in metadiscursive folds of cinematic
instances piled on top of each other.5 Metz then gives an example:
In subjective framing, the gazing and at the same time showing
character duplicates [that is, reflects] both the spectator and the
camera.6
Even back in 1964, in his paper Cinema: langue ou langage?
Metz implicitly acknowledged the reflexive (and therefore metadiscursive and enunciative) nature of filmic discourse. He gave the
example of an image of a revolver, which, Metz tells us, does not
signify revolver but Here is a revolver. If the image simply
signifies revolver it would be purely denotational (i.e., functioning as object language). But by signifying Here is a revolver the
56
image also functions reflexively as metalanguage, since it acknowledges its own presentation of the revolver to a spectator. Furthermore, Metz conceives the reflexive dimension of the image deictically, since here is a deictic term (more specifically, an adverb of
place) that means proximate to the speaker (or spectator).
Anaphora designates another way in which language becomes
reflexive. In fact, anaphora is a special form of metalanguage, since
the referent of an anaphor is not an object or concept, but another
linguistic sign. In the sentence
(2) John said that he is ill
he (the anaphor) is bound to John (the antecedent). Moreover,

the antecedent is the governing category, since it determines the
meaning (or the reference) of the anaphor. Anaphora therefore expresses the reference back of one linguistic sign (the anaphor) to
another linguistic sign (the antecedent).
The function of anaphora is to shorten and simplify sentences
by avoiding the repetition of words within the same sentence. It
also creates cohesion between sentences, as in the following example:
(3) John is ill. He will not be coming to work today
The substitution of he for John in the second sentence creates

cohesion, since the anaphor in the second sentence is bound to the
antecedent in the first sentence. Although Metz refers to anaphora
on a number of occasions, he is concerned only with its reflexive
and metadiscursive functions (its function as a mark of enunciation), not with its creation of economical, cohesive texts. Furthermore, he emphasizes the opposition between anaphora and deixis.
Anaphora therefore refers to a binding relation between two
linguistic signs. It constitutes a purely internal (i.e., textual) relation,
in which reference is made to information already contained in the
utterance. But deixis signifies a relation between linguistic signs
and their context. Deictic terms including personal pronouns (I,
you), demonstrative pronouns (this, that, these, those), and
adverbs of time (now, then) and place (here, there) refer
directly to the circumstances of enunciation, primarily to the spatiotemporal position of speaker and hearer. In contrast, Saussure argued that meaning is generated from la langue, the underlying
system consisting of pre-existing paradigmatic relations between
57
linguistic signs. Saussure assumed that such signs are then simply
manifest and organized in la parole (speech), the pragmatic dimension of language (langage) he excluded from his Course in General
Linguistics. However, Benveniste in particular realized that a number of linguistic signs (deictic terms) only gain meaning when they
are manifest in la parole (i.e., when they are used in concrete discursive situations). For example, the meaning of I depends entirely
on the person uttering it at that time. With the realization that some
words gain meaning only when manifest, Benveniste understood
that the relation between la langue and la parole needed investigating. He used the concept of enunciation to designate a process or
activity that mediates between la langue (virtual/non-manifest system) and la parole (actual utterances). Enunciation is the process
presupposed by the very existence of utterances.
By taking again the previous example
(4) John is ill. He will not be coming to work today
we note that, whereas he is anaphoric, coming and today are

deictic. The verb coming in the phrase coming to work suggests
that the unidentified speaker is located in Johns workplace (contrast it with He will not be going to work today). Furthermore, the
word today is also deictic. Its meaning or reference is the day the
speaker utters the word today. When uttered on another day, its
reference will obviously change. Its reference is therefore dependent on the circumstances of enunciation. Note, by contrast, that
the anaphoric he in the example will always refer to John, on
whatever day the two sentences are uttered. Anaphors are not,
therefore, dependent for their meaning on the circumstances of
enunciation.
The shift within modern film theory from a film semiotics to
post-structural theory closely shadowed the shift from structural
linguistics to a theory of enunciation, particularly Benvenistes theory of deixis. Because deictic terms only gain their meaning when
manifest, they obviously serve to bond or link the utterances to the
speakers and hearers who produce and receive them. Indeed, language is not self-enclosed, but is oriented toward, or centred
around, its users and their spatio-temporal context. Deictic words
grammaticalize the reference points of production and reception
within the utterance itself. The theory of deixis influenced the poststructural film theorists concept of subject positioning, which was
58
defined in terms of those reference points (especially the spectators

look) grammaticalized (i.e., formally represented) in the film itself.
Rather than conceptualized as a closed, autonomous entity, the
filmic text in post-structural film theory is organized around an
external reference point occupied by the spectator. This theory of
meaning (which merely theorized the spectator as an absent centre
point) was then enriched by Althussers theory of interpellation (a
discursive process that attempts to address each individual as a
coherent and unified subject) and Lacans linguistic re-reading of
Freudian metapsychology.
The distinction between deixis and anaphora is crucial to understanding Metzs disputation with Francesco Casetti.
Francesco Casetti on Personal Filmic Enunciation

Metzs theory of impersonal filmic enunciation is directly opposed
to and formulated against Francesco Casettis rigorous deictic theory of filmic enunciation, culminating in his book Dentro lo sguardo
in 1986.7 It may seem paradoxical, then, that Metz wrote the Preface
to the French translation of Dentro lo sguardo.8 However, when
reading this Preface, we come to understand the influence Casetti
had on Metzs own theory. Metz notes that, before he read the
original Italian version of the book in 1986, the issue of filmic
enunciation was running through his head, although his mind was
unfocused. But after reading Casettis book, Metz acknowledges
that it persuaded him to work on a study of filmic enunciation.
Metzs debt to Casetti sounds very much like Kants debt to
Hume. Casetti certainly jolted Metz from his dogmatic slumbers,
but his influence on Metz is indirect. Just as Kant did not become
an empiricist, so Metz did not embrace Casettis deictic theory of
filmic enunciation. However, in his Preface Metz does not express
his opposition to Casettis deictic theory, but instead draws attention to a different opposition between his work and Casettis:
Whereas Casetti proposes a general conceptual framework to study
filmic enunciation (supported by examples), Metz chose the complementary direction (as Metz puts it) to write about the numerous concrete forms of filmic enunciation.9 Metz directly expressed
his reservations about Casettis deictic theory of filmic enunciation
in a 1987 paper, The Impersonal Enunciation, or the Site of Film
(quoted at the beginning of this chapter), which constitutes part 1
59
of LEnonciation impersonnelle. Before I discuss Metz reservations, I

shall outline the main argument of Casettis work in the eighties.
In his essay Looking for the Spectator, published in English
in 1983 and constituting Chapter 1 of Dentro lo sguardo, Casetti
charts the emergence of the spectator as a legitimate object of inquiry in film theory (rather than an unproblematic pregiven entity).
When identified as an object of study (implicitly in the sixties,
explicitly in the seventies), the spectator was conceptualized, Casetti notes, either as a decoder someone who can and must decipher a set of images and sounds or as an interlocutor someone
to whom one can address propositions and from whom one may
expect signs of intelligence.10 The spectator as decoder is an implicit premise of film semiotics, because the spectators role was
marginalized in the production of filmic meaning (since films were
conceived as autonomous entities).
Challenges to the conception of reader/spectator as decoder
emerged from both literary and film theory; the most notable manifestations (listed by Casetti) were Barthess theory of reading as rewriting, German theories of interpretation (based on phenomenology and hermeneutics), the implicit reader (Wolfgang Iser, M.
Corti, Seymour Chatman), and the virtual reader (Jurij Lotman and
Gerard Prince). Casetti also notes that in Language and Cinema, Metz
admits, in his attempt to describe filmic codes and systems objectively, a plurality of levels of reading, each of them submitted to a
given principle of deciphering, and thus open to subjective choice
and intervention.11 Here, Casetti alludes to the tension in Metzs
book between an empirically driven attempt to identify and classify
filmic elements and the semiotic activity of modelling the underlying system that organizes these elements. Casetti then refers to
Metzs well-known statement that the aim of film semiotics is to
understand how films are understood, a process that both Metz
and Casetti characterize as a meta-reading of a film. (Of course,
despite these allusions to spectators, only in Psychoanalysis and Cinema did the spectator take center stage in Metzs work.)
Placed center stage, the film spectator is conceived as an interlocutor (although to varying degrees by different film theorists).
For Casetti, the most important aspects of conceptualizing the spectator as interlocutor are twofold: (1) [the spectator] is in charge of
weaving the threads of a fabric which is ultimately meant for him/
her; consequently, (2) the film is now conceived, not as an ordered
60
and complete set of images and sounds that simply need to be

attested (or decoded), but as a text, an organism submitted to and
influencing its context.12 This definition of text implies that a film
is not hermetically sealed from the outside, but is necessarily
bonded to or oriented toward the outside world, and therefore
is necessarily open and incomplete.13
It is at this point, or interface between spectator-as-interlocutor
and film-as-text that Casetti makes his original intervention into
the theory of filmic enunciation. To conceive spectators as interlocutors and films as texts strongly implies the films necessary orientation toward a spectator:
At every moment, the film indicates a point where it can anchor its own
moves and find a response; it directs its looks and voices, beyond the limits
of the scene, toward someone who presumably (or rather, pretendingly)
has to collect them and to answer back. Briefly stated, the film offers itself
for viewing instituting its own destination as a goal to be reached, or a
backboard on which to bounce again.14
In response to a film offering itself for viewing, the spectator commits him/herself to viewing: s/he responds to the availability of
the world-on-the-screen by taking his/her own responsibility in
response to the propositions of a destination by the film.15
Finally, Casetti briefly indicates that pragmatics is able to act as
a framework for his study, since it also theorizes the relation between text and its contexts (particularly its context of reception)16
and enables the film theorist to ask about how the film entertains
its interlocutor, how it founds his/her presence, organizes his/her
action, etc, in a word, how the film says you.17
It is to this question that Casetti addresses himself in his subsequent essay Face to Face18 (published in 1983 and containing the
central hypotheses of Dentro lo sguardo). Casetti points out that film
is not a self-enclosed work but a text that opens in itself a space
ready to receive whomever it is addressed to.19 The essay is dedicated to the task of describing this space in detail, employing deictic
categories of person.
Casetti begins by considering the openings of two films Riso
Amaro (Guiseppe de Santis, 1949) and The King of Marvin Gardens
(Bob Rafelson, 1972). Both begin with a character looking directly
at the camera, a technique that Casetti characterizes as a form of
direct interpellation that transgresses a well-known interdiction and
61
functions metadiscursively by drawing attention to the process of

filmic enunciation. Casetti then evokes the following series of oppositions to distinguish films (or segments of films) that draw attention to their enunciation and films that attempt to disguise their
enunciation: discourse/story, commentary/narrative, and enunciative utterance/utterative text, respectively. He holds to the principle that there are permanent marks [of enunciation] accompanying the filmic text all along its development, in which the addressee represents one of these marks.20 A characters look at the
camera is a textual mark that affirms the presence of an addressee/
spectator. In Casettis words, this is one filmic technique that says
you.
Through an analysis of the opening sequence of The Kid from
Spain (Leo McCarey, 1932), Casetti develops, using the deictic categories I, you, and he, his typology of shot types, a typology that
goes some way toward describing the space that opens out a film
(or a scene) to each spectator (or her look):
The relation between a look and a scene allows us to appraise at the same
time the enunciator, addressee and the discourse they control, that is to
say, a group of three elements that correspond to, respectively, the gesture
of actualization thanks to which one sees, the gesture of destination that
gives itself to be seen, and the thing (or the character) that is seen. In fact,
if we want to establish indicative identities, we have an I, a you and a he
(character or thing).21
For Casetti, this system of three gestures (or elements) is obligatory

in all films (it constitutes the films conditions of possibility). Although these elements are invariant, Casetti allows for a shift in
emphasis on one at the expense of the others; indeed, his typology
is based on the hierarchical re-configuration of these elements.
The first type of shot Casetti identifies is based on a fundamental equilibrium between the elements. Attention is drawn neither to
the enunciator (I) nor to the addressee (you), but only to the film
itself (the utterance, or he).
Opposite a he that shows itself for what it is, there is an I and a you that
are present but which do not make their presence explicit. In particular,
the addressee must assume the position of a witness: he is the one who is
led to look, and therefore who is permitted to look, but without this
mandate being made explicit and without this task (of looking) intervening
in the events.22
62
This is the objective shot, or film as histoire/narrative/utterative

text. What Casetti means here is that, as with all discourse, the
objective filmic shot presupposes an enunicator or an I (the
gesture of actualization) and an addressee or a you (the
gesture of destination), although attention is not drawn to these
two gestures. According to Casetti, their presence is not, therefore,
explicit, but implicit, in the objective shot. We can take this to mean
that these two gestures are understood by the spectator to be present, even though they are not marked in the shot. Examples of the
object shot are numerous, and Casetti refers to a moment in The Kid
from Spain when a group of girls talk to one another without acknowledging the look of the camera.
The second type of shot is interpellation, which ruptures the
equilibrium apparent in the first type, since both the enunciators
(I) and the addressees (you) presence become explicit in the
utterance (although unequally). A characters look at the addressee
is a seminal case of interpellation. Casetti formally defines interpellation as follows:
An I (who looks and sees) coincides with a he (who is seen but at the same
time looks at he [the spectator] who is led to look), whereas a you (who is
meant to be looked at and is looked at, but is not seen) enters in the game
without assuming any precise form. The enunciator is represented in a
character, who depends on a question of action (the act of looking) and a
question of framing (reaching the spectator), effecting a slippage from the
level of enunciation to the level of the utterance.23
What Casetti means in effect is that the enunciator (I) enters the
film through the intermediary of a characters look (he), which is
directly addressed to the spectator (you).
The third type of shot is the subjective shot, consisting of two
moments a characters act of looking and the addressees being
shown what the character is looking at (these two moments can be
shown in one, two, or three shots). Casetti conceives these two
moments as a series that goes from You and I see him to You
and he see what I show you.24 Here, the character (he) and
addressee (you) become prominent.
Finally, Casetti calls the fourth type of shot an unreal objective
shot, referring to unusual camera angles. Casetti gives the example
of those shots found in Busby Berkeley musicals in which the camera is placed perpendicular to the horizon when it is pointing
63
downwards on a group of dancers who form an abstract pattern

(an example of which is found in The Kid from Spain). According to
Casetti, the deictic formula is as if you were I: the addressee
renounces his own competence in order to slide into the other. He
confines himself to a pure faculty of seeing, to a look without a
determinate place.25 This type of shot is characterized by the inability to attribute it to a character, and by the absence of enunciator
and addressee (only the cameras look is present).
How do these four canonical shots characterize their addressee?
Casetti argues the following:
The addressee [assumes] respectively the posture of the witness (a confirmed you opposite a confirmed I: the objective shot), the spectator set
aside (a you installed opposite an I combined with a he: interpellation), the
character (a you combined with a he opposite a confirmed I: the subjective
shot) and the camera (a you combined with an I: the unreal objective
shot).26
In other words, we have the following schema:

Shot
Addressee
Objective
Interpellation
Subjective
Unreal objective
Witness
Spectator set aside
Identification with character
Identification with camera
This is Casettis basic framework for describing the films orientation toward the spectator-as-interlocutor.
Figure 4 shows how Casetti complexifies Metzs application to
film of Benvenistes distinction histoire/discours. Although shot types
1 and 4 (like 2 and 3) use personal pronouns, only the third person
pronoun, which is an anaphoric term, not a deictic term (and it is
anaphoric because it is an indirect form of address, unlike the first
and second persons), is visible. This is why 1 and 4 conform to
histoire, not discours.
Christian Metz on Impersonal Filmic Enunciation

One of the main arguments in Metzs last published book is that
filmic enunciation is resolutely non-deictic, and therefore does not
require pronouns to describe it: Personal pronouns can only lead
toward a deictic conception of enunciation in cinema, which in my
64
Enunciation
Utterance
Fig. 4
histoire
discours
1. objective shot
2. interpellation
3. subjective shot
4. unreal objective shot
opinion is not suitable to the realities of film.27 As we have seen,

Metz argues that the realities of film are conducive to anaphora (in
which one piece of text refers to another piece of text) but not deixis
(in which a piece of text must refer to its context of enunciation for
its meaning). For Metz, the reality of oral discourse allows for a
strong distinction between deixis and anaphora; written language
allows for a weak distinction; whereas in filmic discourse, there is
only anaphora.
Although the linguistic theory of enunciation began as a theory
of personal pronouns, Metz resolutely rejects all notions that the
link between enunciation and pronouns is determinate. Casetti,
however, identifies enunciation with pronouns. Metz lists three
main risks in doing this: (1) anthropomorphism, (2) artificial use of
linguistic concepts, and (3) transformation of enunciation into a
dialogue (involving real, extra-textual relationships).28 Metzs
65
main criticism of Casetti revolves around point 2. Before discussing

this point, I shall briefly go over points 1 and 3.
First of all, to prevent anthropomorphism, Metz prefers to speak
of source and target of enunciation, rather than enunciator and
addressee.29 Furthermore, he argues that these terms designate
nothing more than parts of texts, or orientations. What he means
by this latter term is that the source is the text as seen from its
origin to its end (for Casetti, the film in the process of being made);
the target is the same text, but this time seen from its end point
being undone and freed in imagination30 (for Casetti, the film in
the moment it gives itself to be seen).
With reference to point 3, Metz notes that, whereas dialogue is
an empirical activity, involving actual interlocutors, enunciation is
a purely textual phenomenon:
It is of no use to pretend with all the required discretion that knowledge of
the enunciator and the addressee would give us at least probabilities or
general frames to understand the authors intentions and the spectators
reactions. For these forecasts are so general that no empirical analysis
would take them into account and they could prove to be wrong for any
given spectator, even if they express a partial tendency common to everyone. The reason why is that you deal with two heterogeneous orders of
reality, a text (that is, I repeat, a thing) and persons.31
To prevent misunderstanding, Metz adds the following rejoinder:

I dont pretend here that enunciative configurations are deprived
of any influence on the observable behaviour of the spectators (this
hypothesis is as improbable as that of its unmistakable determination by the film).32
We now come to what Metz characterizes as Casettis artificial
use of linguistic concepts. In particular, Metz questions Casettis
specific use of personal pronouns to describe aspects of filmic reality before rejecting outright a deictic theory of filmic enunciation.
First of all, he questions Casettis identification of the enunciator
with the I that is, the subject of the utterance (the filmmaker);
Metz argues that it is more plausible to identify the subject with
the spectator, as film theorists in the seventies had. Moreover, Metz
rejects the symmetry posited by Casettis deictic formulas, which
suggest that enunciator and addressee have equal status, that both
are roles to be invested with bodies. But Metz argues that this
66
symmetry is merely a symptom of the use of the first two persons

of natural language (of Casettis speech model of film):
Where the enunciator is concerned, there is no body. And since it is true
that roles (or their equivalents in another theoretical frame) call for an
incarnation the nature of this call still remains enigmatic the enunciator is incarnated in the only available body, the body of the text, that is, a
thing, which will never be an I, which is not in charge of any exchange
with some You, but which is a source of images and sounds, and nothing
else. The film is the enunciator, the film as source, acting as such, oriented as
such, the film as activity. Casettis idea, that a body would be needed for
the enunciator as well as for the addressee, is inspired by the first two
persons of the verb in languages.33
Metz comes close to Bordwells rejection of the narrator (see Chapter 2), but he does not, of course, reject linguistics, as Bordwell
does. Instead, Metz simply repeats that film theorists should not
draw analogies between film and natural language. This is evident
when he challenges Casettis identification of film/character with
the third person pronoun. Metz finds this paradoxical because, in
Benvenistes famous definition, the third person is the absent nonperson, in opposition to the first (I) and second (you) persons,
who are present. But in the cinema, the film, far from being an
absent instance stuck between two present ones, would resemble
rather a present instance stuck between two absent ones, the author,
who disappears after the fabrication, and the spectator, who is
present but does not manifest his presence in any respect.34
Metz is equally critical of Casettis attribution of the verbs to
look/watch and to see to the I (the filmmaker): [The filmmaker]
does not watch, he has watched (which is still not entirely accurate:
he has filmed, and, therefore, watched; the utterer [emetteur] does
not watch his film; he makes it).35 What Metz clearly means is that
the filmmaker, when watching his film, becomes its addressee (a
you in Casettis terms). In this sense, then, the I does not watch:
From the source, nothing either watches or sees, the source produces, expands, shows.36
Finally, Metz attempts to demonstrate the arbitrariness of Casettis deictic formulas. He argues this point by using the orientational approach he developed to combat the anthropomorphism of
a deictic theory. Metz simply reorients himself to the opposite side
of the filmic text. For example, this is how he discusses Casettis
formulation of the subjective shot:
67
According to Casetti, the enunciator plays a silent role and the addressee
on the contrary is very much highlighted, since he is syncretized with a
concrete character, through whose very eyes we see what we see, and who
is therefore, like the addressee, a watcher. . . . That is beyond doubt. But it
is also true, if you turn over the text, that the enunciator regains his
importance, in that the source is figurativized in a character who is not
only a watcher (as the spectator), but is also someone who shows, like the
filmmaker who stands behind him. This character has one eye in front and
one eye in back, he receives rays from both sides, and the image can be
perceived in two different ways, as in some drawings in which form and
background can be inverted.37
Metz is then at pains to emphasize that the reversibility he speaks

of is totally different to the reversibility of interlocutors in a dialogue a reversibility made possible by personal pronouns! In fact
Metz is quite right; he points out that, in a dialogue, the deictic
signifiers change referents, constituting the actual movement/reversibility of oral exchange, whereas the reversibility he speaks of
is an analyst simply changing his theoretical viewpoint.
Metz concedes that not all figures of enunciation can be reversed, since some are precisely imprinted in a film (this means
then, that the subjective shot, which Metz did reverse, is not precisely imprinted on a film). In Casettis formulas, Metz mentions
that the unreal objective shot cannot be reversed (Metz refers not to
the Busby Berkeley examples, but to the systematic use of low
angles in the films of Orson Welles).
We can now begin to articulate the fundamental difference between Casetti and Metz. Whereas Casetti models film on the immediacy and symmetry between filmic enunciator and addressee,
as in a dialogue, Metz argues for the mediate and non-symmetrical
nature of the relation between filmic enunciator and addressee as
in writing. This relation is non-symmetrical because one of the
functions of writing is to dispense with the presence of the enunciator (or allow a spatio-temporal displacement between the enunciator and her utterance). Similarly, Metz dispenses with the filmic
enunciator because of the way he conceives the realities of the filmic
medium it resembles the recording activity and permanence of
writing rather than the immediacy and impermanence of speech.
More specifically, Metz conceives film as a particular type of writing, namely, histoire, which Benveniste defined by its absence of
deictic markers.
A consequence of the reality of the filmic medium (its reality as
68
histoire) is that it records unique spatio-temporal events (and

therefore no longer makes the perception of these events unique).
Yet Metz does not explore this aspect of film or indeed of television. More generally, Metz does not explore the relation between
the space and time depicted in a film/on TV and the space and
time of the film/TV viewers situation. But this is crucial in the
comprehension of both film and TV images and represents a deictic
bond between the spectator/viewer and the image, a bond Metz
refuses to acknowledge. By rejecting Casettis use of deictics, Metz
has rejected all forms of deictic orientation of images toward their
contexts of production and reception. It is to this limitation in
Metzs theory that I shall now turn.
Deixis, Television, and Digital Colorization

According to Metz, film does not orient itself in relation to the
spatio-temporal circumstances of its production and reception, but
can only orient itself in relation to its own internal spatio-temporal
relationships relations between its shots and between its autonomous segments: The film is able to express space and time relationships of some kind, but only anaphorically, within the film
itself, between its different parts, and not between the film and
someone or something else.38 This is why, for Metz, Enunciation
is the semiological act by which some parts of a text talk to us about
this text as an act. However, resorting to the complicated and quasiinimitable mechanism of the deixis is not a necessity.39 Instead, he
argues that cinematic enunciation is always enunciation on the
film. Reflexive, rather than deictic, it does not give us any information about the outside of the text, but about a text that carries in
itself its source and its destination.40
Metzs theory of impersonal enunciation only functions within
a specific (or narrowly defined) framework. I have mentioned that
Metz aligns film to writing (or to histoire) rather than speech. Because of the absence in histoire of deixis and its select use of a
limited set of tenses (the aorist, imperfect, and pluperfect),41 histoire
represents events that are not spatio-temporally related to the context of the writer/enunciator. The events narrated by the historian
are not related, by deixis and/or by tense, to the present (to the
context of utterance).
69
On the other hand, discours is related to its context of utterance.

Benveniste argues this by referring to the perfect tense:
The perfect creates a living connection between the past event and the
present in which its evocation takes place. It is the tense for the one who
relates the facts as a witness, as a participant; it is thus the tense that will
be chosen by whoever wishes to make the reported event ring vividly in
our ears and to link it to the present. Like the present [tense], the perfect
belongs to the linguistic system of discourse, for the temporal location of
the perfect is the moment of the discourse while the location of the aorist
[a tense exclusive to histoire] is the moment of the event.42
In rejecting the presence of deixis in film, Metz limits his discussion of filmic enunciation to the articulation of space and time
within narrative fiction films (as he did when formulating his
grande syntagmatique in the sixties). He argues that spectators do not
need to know anything about the circumstances of enunciation to
comprehend a films internal spatio-temporal relations. It is indeed
the case that each fiction film does not require any deictic relation
to the space and time of its production and reception to comprehend its internal spatio-temporal relations, since the films own
space-time relations constitute an imaginary elsewhere and remain
an imaginary elsewhere each time and place in which it is projected. Yet there is more to understanding a narrative film than
simply comprehending the internal spatio-temporal relations of its
diegesis. I propose to steer a course between Casettis personal
pronoun theory of filmic enunciation and Metzs theory of impersonal enunciation based only on anaphora. In other words, we need
to supplement Metzs theory with deictic terms as long as we
understand deixis to be a general cognitive category, not a narrow
linguistic one. In the 1930s Karl Buhler used the term deixis in this
broad sense to develop what he called a situation model of action,
which studies the meaningful behaviour of the living being.43
More specifically, Buhler used the term deixis to study the systems
of orientation that enable an individual to orient herself in relation
to her environment. Buhler gives the example of an individual
attempting to orient herself in a new city. Each individuals mode
of orientation is not limited to deixis in speech, but also includes
the physical act of pointing (ocular deixis, or pointing that takes
place within a common space of actual perception) and what Buhler
calls imagination-oriented deixis (orientation in fictional spaces). Of
course, these various types of deixis work in conjunction with one
70
another to give the individual a detailed mental map of her environment. Interestingly, in outlining the way ocular deixis interacts
with linguistic deixis, Buhler emphasizes the role of the individuals
body: When [a] person uses words like in frontbehind, rightleft,
abovebelow [a] fact becomes apparent, namely the fact that he senses
his body, too, in relation to his optical orientation, and employs it to
point. His (conscious, experienced) tactile body image has a position
in relation to visual space.44 Buhler anticipates the work of George
Lakoff and other cognitive semanticists by fifty years.
I want to use this broad understanding of deixis to begin explaining how spectators comprehend documentary films, home
movies, and TV news programs. Whereas in narrative fiction films
the internal spatio-temporal relations are specific, the films relation
to the extra-filmic space-time of its production and reception is nonspecific, as we have already seen. But in the comprehension of
home movies and TV news, both the internal and external spatiotemporal relations are specific (or perhaps we should say, they need
to be specific for comprehension to take place). That is, the relation
between the space and time of an event depicted in the TV news
(for example) and the space and time of the TV spectator must be
specified, since news images obviously gain their meaning from the
spatio-temporal context in which they are manifest (they are deictically bound to their context of reception).
This can be understood more clearly by referring to Daniel
Dayan and Elihu Katzs work on the televisation of live events.45
They begin by making a distinction between spectacle, ceremony,
and festival. Whereas spectacle involves a minimum level of interaction between performers and audience, festival involves a strong
interaction (with ceremony falling somewhere in between). In this
sense, both film and TV are spectacles because they minimize interaction between performers and audience through spatio-temporal
displacement. In festivals, on the other hand, there is minimal displacement because the audience becomes part of the event (they are
as much part of the event as the performers); in other words, it is
difficult to separate audience from performers. Ceremonies depend
on a reaction from the audience (shouting, applause, etc.), although
this is their only level of interaction.
Katz and Dayan then ask, What happens when ceremonies are
televised and broadcast live? Their main argument is that television
71
enables spectators to attend the whole of the event as it unfolds and

maintains a sense of continuity and immediacy between spectators
and performers. If recorded, events become spectacles, but if they
are broadcast live, the activity of watching them on TV allows the
audience to feel a part of (rather than apart from) the event. There
is therefore a determinate link between the TV image and the audience, since the spatio-temporal relations between the TV audience
and the televised event become specific: The audience is oriented
toward the event as it unfolds in real time and in continuous space.
Under these conditions, I would characterize the link between the
event and the TV audience as deictic (more specifically, ocular
deixis), since the broadcast gains its meaning (as deictic terms do)
by maintaining a direct and continuous link between an event and
its context of reception its audience. In live broadcasts, the whole
of the TV apparatus functions deictically.
It appears, then, that it is the recording aspect of film and TV
that disqualifies us from defining them in terms of deixis. This is
what Metz argues when he aligns film with writing. In opposition
to Metz, I argue that the act of recording, like other means of
production, also creates deictic links between the image and its
context (although in this instance, the image creates a link with its
context of production, not reception). It is quite reasonable to argue
that all film and TV images are embodied with their technological
mode of production, which binds each to an historical space and
time. This phenomenon is particularly accentuated in early films,
by innovations in film technology (black and white emulsions,
Technicolor, colorization), camera technology (such as the development of wide screen), and projection technology (Dolby stereo,
etc.). Indeed, what does the recent practice of digital colorization of
black and white films signify except a disparity between the reception of films by their contemporary audience and the purported
intolerance of todays audiences, who perceive black and white
images as dated? The so-called need for colorization is an indication that the black and white nature of filmic images is a mark of
enunciation, a mark of the films historical context of production.
For todays audiences, brought up on color images, the black and
white nature of film images becomes reflexive or metalinguistic (is
rendered reflexive by historical time). Digital colorization seems to
me, then, to be a mechanism of disguise that attempts to conceal
72
a form of filmic enunciation (here, black and whiteness). Paradoxically, however, the black and white nature of these films is accentuated through its very effacement.
The Impossibility of Textual Analysis?

Metzs theory of impersonal filmic enunciation can be taken in
another direction. I shall return to his notion that marks of enunciation only have a functional, rather than an ontological, status in
the filmic text, then briefly mention the work of a number of American film scholars who have independently adopted a functional
approach.
In the previous section I highlighted the negative side of Metzs
rejection of deixis. But this rejection has a positive side it discards
the empiricist claim that marks of enunciation are inherent in a text
and simply need to be identified and read off the films surface.
According to deictic theories of filmic enunciation, a films formal
configuration (its framing, inherent perspectival system, mise-ensce`ne, and so on), gang up to designate a place or position for each
spectator who watches the film. But Metzs work on impersonal
filmic enunciation brings into question the power of these formal
configurations, which are treated as if they are empirical objects
that automatically and unfailingly position each spectator. A theory
of enunciation can no longer point to a particular textual figure and
argue that it constitutes a mark of enunciation, that it positions
every spectator or determines her psychic disposition toward the
text.
This theoretical position, in which texts do not (or cannot) determine their own reading, has been developed within both deconstruction and reception studies. For Paul de Man, the rhetorical
structure of a text will in the end prevent the analyst from applying
to that text a preconceived interpretive method or reducing it to a
prior set of general theoretical assumptions. Each textual reading is
also a misreading, because it focuses only on those elements of the
text that conform to these general theoretical assumptions at the
expense of discrepant elements, a residue of indetermination that
resists generalization. Although reception studies also posits that
texts are inherently indeterminate, its advocates argue that textual
meaning is temporarily fixed in different ways by different interpretive communities. Whereas reception studies focuses on indeter-
73
minate texts as they are consumed and interpreted by various communities, deconstruction focuses on textual indeterminacy from a
vantage point that is not in itself interpretive.
In contrast, the primary aim of structural-based textual analysis
was to replicate the structure of a text in a theoretical language, to
model a structure not immediately manifest to perception because
only the effects of structures are perceived, rather than the structures themselves. But this process of replication in fact diminishes
the text, reduces it to a preconceived structural grammar. The idea
of a progressive filmic text, first advocated by the editors of Cahiers
du cinema in the late sixties, acknowledges discrepant details in a
text details that cannot be reduced to a preconceived grammar.
This reading strategy codifies these discrepancies as contradictory,
progressive moments that are inscribed on (and can simply be read
off) a films surface.
Metzs comments on the functional status of marks of enunciation challenge the assumptions of structuralism, the progressive
text, as well as reception studies. Metz questions the notion that a
text can be labelled as inherently feminist, progressive, reactionary
or even reflexive (although Metz did not draw out these consequences from his comments). It seems to me that Metzs theory of
impersonal filmic enunciation is an enunciative theory implicitly
based on the premises of deconstruction most notably, the inherent indeterminacy of texts. This is evident in his refusal to consider
relations between the text and its extra-textual contexts of production and reception, as well as in the way he reverses some of
Casettis deictic formulas, thereby rendering them indeterminate.
Although I concur with Metzs deconstruction of Casetti, I would
not go so far as to deconstruct all attempts to relate a filmic text to
extra-textual reality.
Metzs deconstructive position foreshadows recent work by
American film theorists, including Judith Halberstam, David Rodowick, and Tom Conley. All three authors in their own way approach the question, How does textual analysis progress, after realizing that its cherished filmic text is indeterminate? In a chapter
on The Silence of the Lambs (Jonathan Demme, 1991), Halberstam
argues that we are at a peculiar time in history, a time when it is
becoming impossible to tell the difference between prejudice and
its representations, between, then, homophobia and the representations of homophobia.46 In relation to Demmes film, Halberstam
74
avoids the temptation to advance generalized claims that seek to

reduce the film to some preconceived interpretive method:
I resist here the temptation to submit Demmes film to a straightforward
feminist analysis that would identify the danger of showing mass audiences an aestheticized version of the serial killing of women. I resist the
temptation to brand the film as homophobic because gender confusion
becomes the guilty secret of the mad man in the basement. I resist, indeed,
the readings that want to puncture the surface and enter the misogynist
and homophobic unconscious of Buffalo Bill, Hannibal the Cannibal and
Clarice Starling. This film, indeed, demands that we stay at the surface and
look for places where the surface stretches too thin.47
However, Halberstam presents a peculiar mix of reception studies,

deconstruction, and the theory of intentionality. The preceding excerpt is structured by a rhetoric that, firstly, asserts the different
readings of the film by various communities (gay men, heterosexual
feminists, and lesbians) and, secondly, at the same time resists these
readings by declaring the inherent indeterminacy of the film itself.
These two strategies can be seen to complement one another, for
both deconstruction and reception studies are based on the concept
of textual indeterminacy. But Halberstam then undermines this emphasis on indeterminacy by ascribing intentionality to the film: The
film indeed demands that we stay at the surface (my emphasis). The
fallacy of ascribing agency to a text is compounded by the logical
conflict between the theory of intentionality (which fixes textual
meaning in advance) and the deconstructive and reception studies
emphasis on textual indeterminacy. Halberstam is undecidable
when it comes to choosing between these incompatible theories of
textuality.
Both Rodowick and Conley combine deconstruction and reception studies with textuality in a more positive and decidable manner than Halberstam does. They advocate a practice of politically
creative re-reading of films, one that does not passively reproduce
the dominant values indeterminately perpetuated through filmic
texts. Rodowick argues for textual criticism that must be understood not as repeating what a text means, but as presenting the
opportunity to construct positions from which it can always be read
and understood in new and unforeseen ways.48 For Rodowick,
then, critical reading becomes an act of creative intervention in
which the reading encounters the text in a relation of difference,
not identity.
75
Similarly, Conley develops a creatively political practice of reading by analyzing the presence of alphabetical writing in narrative
films. Following Derridas critique of the subservience of writing to
speech and his subsequent emphasis on the alterity of these two
channels of discourse, Conley places the image on the side of spoken language against written language and in so doing posits an
essential difference between writing and filmic image. Conleys
main aim is to trace the movement of difference between writing
(film titles, shop and road signs, etc.) and the film images in which
it appears. He notes that these two channels of discourse result in
an activity a pleasure of analysis allowing spectators to rewrite and
rework discourses of film into configurations that need not be determined
by what is immediately before the eyes. Reworking of this order can lead,
it is hoped, to creatively political acts of viewing. In this way, the literal
aspect of film writing can engage methods of viewing that need not depend
entirely upon narrative analysis.49
Conley wants to articulate and politicize the personal fantasies of

each spectator when she sub-consciously experiences the marginal
(non-narrative) details within the filmic image. Conleys and Rodowicks answers to the question, How does textual analysis progress, after realizing that its cherished filmic text is indeterminate?,
are important because they offer a way out of the impasse expressed by Halberstam and, more generally, refute commentators
who reduce deconstruction to a theory focusing on the mere play
of signifiers.
Yet it is unfortunate that Metz did not develop this second
functional dimension of his work; unfortunate because it can be
taken in the direction independently developed by Rodowick and
Conley. Metzs innovation therefore seems to be primarily negative
a critique of Casettis analogies between film and natural language. To this extent, Metz has returned to and repeated the negative tone of his first paper Film: langue ou langage? in which he
rejected a comparison between film and la langue. In LEnonciation
impersonnelle, Metz also committed himself to a more reflexive style
of writing, a style he gradually developed since the mid-seventies.
It would take another chapter to discuss this issue in Metzs works;
I shall simply note here that, at the beginning and ending of The
Imaginary Signifier, Metz writes about the very activity of theorizing about the cinema and highlights its paradoxes.50 And in The
76
Impersonal Enunciation he combines his reflexive style with a

more personal style of writing (the paper and the book it is part of
serve to acknowledge the support his friends and colleagues have
given him over the years). It is fitting, then, that toward the end of
the paper Metz argues that the validity of research into the cinema
is, in the end, dependent on the personal qualities of the individual
researcher: If you assume that the analyst has the necessary training (knowledge, method), the whole value of his work depends on
his personal qualities, since he is at the same time the scholar, and
(together with the film), the very terrain of the research.51
The question posed previously How does textual analysis
progress, after realizing that its cherished filmic text is indeterminate? has not been tackled only by American film scholars. The
cognitive film semiotician Roger Odin has made the truism that
texts cannot determine their own reading a fundamental premise
of his semio-pragmatic approach to film. Odins work is analyzed
in the following chapter.
CHAPTER FOUR
The Institutional Context

A Semio-pragmatic Approach to Fiction and
Documentary Film
I do not think, I no longer think today, that it is interesting to pose

the problems of the relations between the immanentist paradigm
and the pragmatic paradigm in terms of exclusion or conflict . . .
The consequence of these reflections is that to understand how film
is understood necessitates the inclusion of the immanent and pragmatic approaches in the same theoretical framework. As its name
indicates, semio-pragmatic attempts to articulate these two paradigms. (Roger Odin)1
The most promising [definitions of pragmatics] are the definitions
that equate pragmatics with meaning minus semantics, or with a
theory of language understanding that takes context into account,
in order to complement the contribution that semantics makes to
meaning. (Stephen Levinson)2
n conjunction with that of the other cognitive film semioticians I have already discussed in this book, Roger
Odins semio-pragmatic film theory is preoccupied with researching the film spectators competence, the tacit knowledge that constitutes each spectators psychic disposition (or mode of attention)
when engaged in the activity of watching a film. Odin regards
filmic competence as predominantly pragmatic, with the result that
meaning is determined, not by the internal, semantic constraints of
la langue, or contingent grammatical rules, or by deictic markers,
but by a multitude of external constraints.
Odin calls these external constraints institutions, a concept central not only to pragmatics, but also to sociology and anthropology.
For Bronislaw Malinowski, an institution
implies an agreement on a set of traditional values for which human beings
come together. It also implies that these human beings stand in definite
relation to one another and to a specific physical part of their environment,
natural and artificial. Under the charter of their purpose or traditional
77
78
mandate, obeying the specific norms of their association, working through

the material apparatus which they manipulate, human beings act together
and thus satisfy some of their desires, while also producing an impression
on their environment.3
One fundamental component of Odins semio-pragmatics is that

the institution-spectator-film interaction is not predetermined and
failsafe. The ontology of a film is not automatically fixed in advance, but is determined through the process of watching (or reading) a film. This means that no film can be defined as inherently
fictional, documentary, or avant-garde. Odin shows that such
modes of film are not given in advance, but are the effect, or end
result, of a series of operations. A crucial feature of Odins work
is to delineate the various filmic modes and operations.
Another aspect of semio-pragmatics is that the modes of film
are not seen to be mutually exclusive categories, but share a number of operations. Odin argues that only one operation (fictivization) separates the fiction and documentary modes of filmmaking.
To this extent, Odins work reinforces recent debates concerning
the close relation between fiction and documentary films.4 Later I
shall outline all the primary modes Odin identifies, and the operations they share and discuss three particular modes of filmmaking central to Odins work the documentary, home movie, and
dynamic modes. But first, I shall outline the foundations of Odins
film theory, which is based on a three-way relation between semiotics, pragmatics, and cognitivism.
Meaning, Context, and the Mind

Pragmatics is not a unified, homogeneous discipline, but is multiple
and diverse. Francesco Casetti identifies four dominant schools of
pragmatics:5 (1) empirical pragmatics (based on psycho- and sociolinguistics), (2) transcendental (or universal) pragmatics (as formulated by Jurgen Habermas), (3) illocutionary pragmatics (based
on speech act theory), and (4) enunciative pragmatics (theories of
enunciation, which I have discussed in Chapter 3). Whereas empirical pragmatics studies many of the contingent, ad hoc influences
on speech (class, gender, geographical location), and universal
pragmatics studies an ideal speech situation, illocutionary and
enunciative pragmatics study linguistic competence, delimiting its
domain of research to specific dimensions of language (such as
pronouns, which encode context in language form), as well as in-
A SEMIO-PRAGMATIC APPROACH TO FILM
79
trospective (or cognitive) dimensions of language (for example, inference generation). Despite their diversity, all schools of pragmatics propose that language meaning is not fixed in advance.
Pragmatists study the relation between language and context, and
all agree that understanding language is not simply a matter of
knowing the meaning of words and the grammatical relations between them, but depends on knowing how to use language in
various contexts.
Odins film theory possesses a cognitive dimension and a pragmatic dimension, which firmly align it to a recent paradigm within
linguistics, one Casetti does not identify what Melina Sinclair calls
cognitive (or mentalist) pragmatics.6 She identifies this position in
the work of a number of prominent linguists, including Jan Nuyts,
Diane Blakemore, Asa Kasher, and Dan Sperber and Deirdre Wilson.7 For Sinclair, cognitive pragmatics develops a theory of competence that determines appropriate language use (or the competence necessary to the appropriate construction and comprehension
of utterances, or indeed films).
We can only determine the aims of Odins semio-pragmatics by
contextualizing it within Metzs film theory, to which it has a complex relation. Odin re-reads and transforms Metzs work on several
levels. On one level, Odins pragmatic framework is based on the
immediate discursive nature of film. This concept is influenced by
Metzs famous definition that film is a langage sans langue, which
places film on the side of parole (speech).8 Later I shall indicate how
a linguistics of la parole is inherently pragmatic (for it studies language use), and a linguistics of la langue is inherently semantic (for
la langue constitutes language as a closed circuit, consisting of a
shared system of codes).
On another level, Colins mentalist reading of Metzs grande
syntagmatique (to be discussed in Chapter 5) is complemented by
Odins assertion that Metzs Language and Cinema9 also formalizes
the spectators competence. Odin bases this assertion on the oftenquoted passage in Language and Cinema in which Metz compares
the film semioticians work to the film spectators reading of a film:
The path that the semiotician follows is (ideally) parallel to that of
the film viewer. It is the path of reading, not of composition. But
the semiotician forces himself to make explicit this procedure, step
by step, while the viewer practices it directly and implicitly, wanting above all to understand the film. The semiotician, for his part,
would also like to be able to understand how film is understood.10
80
On the basis of this passage, Odin argues that Metzs ambition is

not to propose an immanent description of the codes of cinematic
language, but to pose the framework of the cognitive work of the
spectator (to understand how film is understood). It is interesting
to note, for example, that what Metz calls codes, in Language and
Cinema, are the questions that the spectator must ask when watching a film.11
On yet another level, Odins semio-pragmatics combines Metzs
taxonomic work in Language and Cinema with his psychoanalytic
work. In Psychoanalysis and Cinema: The Imaginary Signifier,12 Metz
employed psychoanalysis to examine the general (or generic) psychic disposition the spectator adopts when watching a fiction film.
He characterized this mode of attention in terms of Lacans concept
of the imaginary, which portrays the relation between fiction film
and spectator as a fantasy relation. The fiction films imaginary
signifier aims to transform the spectators consciousness to displace her attention away from the immediate space and time of her
context (including the screens material surface) and toward the
space and time of the events depicted in the diegesis (the films
fictive, imaginary elsewhere). In contrast, in Language and Cinema,
Metz employed the taxonomic procedures of structural linguistics
(among other things) to distinguish the system of cinematic codes
from filmic texts and, within the latter category, identify the different levels at which the term filmic text operates: It refers not
only to individual films, but also to groups and classes of films.13
Odin combines these two approaches by detailing the different
modes of attention spectators adopt when watching different
groups and classes of films (what Odin calls modes of film). However, Odin does not adopt a psychoanalytical approach to discuss
the film spectators psychic disposition, but a cognitive approach.
The result of Odins formulations is that the multiple modes of film,
together with their multiple external constraints, create multiple
dispositions, or modes of attention. Each mode of film requires a
different disposition, and one of Odins aims is to outline the different institutions that condition the various dispositions necessary to
understand different modes of film.
Meaning Minus Semantics

One dramatic way to define the pragmatic theory of meaning is to
say that it is premised on non-communication in which failure to
81
communicate is the norm, and what needs to be explained is how

successful communication takes place. Less dramatically, pragmatists argue that meaning is not determined in advance, but is temporarily fixed when language is used in various contexts. By contrast, the semantic theory of meaning describes successful
communication as an automatic process regulated and determined
in advance by a system of codes. This system is based on the
metaphor of language as a closed circuit along which messages can
be encoded by the sender and decoded by the receiver, both of
whom mutually share the same system of codes, ensuring no loss
of meaning in the transmission of the message. Pragmatists, however, contest this model of language as a closed circuit, with sender
and receiver sharing the same system of codes, because it is based
on the mutual knowledge hypothesis. This hypothesis, as we saw
in Chapter 2, states that knowledge is not only shared, but known
to be shared, then must be known to be shared, and so on, ad
infinitum.
Roy Harris observes that the semantic metaphor of language as
a closed circuit was first developed by Ferdinand de Saussure in
his Cours de linguistique generale: The Cours is the first treatise on
language to insist that speech communication is to be viewed as a
circuit, and to attach any theoretical significance to the fact that
individuals linked by this circuit act in turn as initiators of spoken
messages and as recipients of such messages.14 The metaphor of
language as a speech circuit borrowed two claims made by the
seventeenth century psychological theories of oral communication
(as developed by John Locke, and others): (i) that communication
is a process of telementation (that is, of the transference of
thoughts from one human mind to another), and (ii) that the necessary and sufficient condition for the successful telementation is
that the process of communication, by whatever mechanisms it
employs, should result in the hearers thoughts being identical with
the speakers.15 The concept of the speech circuit enabled Saussure
to propose the existence of a system of signs shared by speaker and
hearer (la langue), a shared system that made the speech circuit
operative.
The importance for Saussure of the concept circuit lay in its
use of the notion of energy conversion as studied within the natural sciences. It enabled language to be described scientifically in
terms of a symmetry of message conversion between speaker and
hearer, and Saussure needed a scientific metaphor in order to con-
82
fer upon his particular object of analysis, la langue, scientific status.

Harris remarks that the metaphor may have been suggested to
Saussure during the time he presented his course in general linguistics (between 1907 and 1911) by the major technological innovations
in communication, innovations that were forms of energy conversion applied to the transmission of verbal messages telegraphy,
telephony, and broadcasting. Harris is critical of Saussures metaphor, remarking that by presenting speech as a closed, causally
determined process in every way analogous to energy-conversion
processes of physics and chemistry, linguistics was provided with
a forged carte dentree to the prestigious palace of modern science.16
One of the pragmatists alternatives to the closed-circuit code
model of communication (a model that constitutes language as a
failsafe semantic algorithm) is based on the concept of inference.
For pragmatists, no failsafe semantic algorithm exists between
sender and receiver. The message the sender wishes to communicate cannot be automatically encoded into a message and then
automatically decoded by the receiver. Instead, the senders utterance merely modifies the cognitive environment of the receiver,
and it is on the basis of this modification that the receiver infers or
constructs the message the sender purportedly wishes to communicate. Pragmatists are crucially concerned with answering the
question, How does the receiver infer the correct message without
the benefit of a shared system of codes?
The Space of Filmic Communication

The objective of a semio-pragmatics of film and the audio-visual is to
attempt to understand how audio-visual productions function in a given
social space. According to this approach, the act of making or seeing a film
is not immediately a fact of discourse, but a social fact obtained by adopting a role that regulates the production of the film text (which means film
as a construction endowed with meaning and generating affects). A role
can be described as a specific psychic positioning (cognitive and affective)
that leads to the implementation of a certain number of operations that
produce meaning and affects. A priori, there is absolutely no reason for the
actant director and actant reader to adopt the same role (the same way of
producing meaning and affects). However, it is only when the same role is
adopted by these two actants that what can be called a space of communication is created. A space of communication is a space in which the production of meaning and affects are harmoniously formed during the film-
83
making and the reading. From this space of communication derives the
feeling of mutual comprehension between the actants, which gives the
impression that communication resides in the transmission of a message
from a Sender to a Receiver. (Roger Odin)17
In the middle of this long quotation, Odin states that a priori, there
is absolutely no reason for the actant director and actant reader to
adopt the same role (the same way of producing meaning and
affects). Firstly, Odin is using A. J. Greimass term actant, which
designates an agent who acts or is acted upon, and Odin conceives
the film spectator as an active actant, since she is designated by the
term reader. This sentence clearly illustrates the pragmatic dimension of Odins work, since the activities of the actant director and
actant reader are not conceived as being constrained by a closed
circuit of shared codes, or, in Odins terms, the actants are not
compelled to adopt the same role. For Odin, a role is a specific
psychic positioning (or mode of attention) created by the social
space in which films are seen, a space consisting of institutions
and modes. Odin employs the term modes to cover what Metz
calls groups and classes of films (Odin identifies eight modes in
total: fiction films, documentaries, home movies, and so on). Each
mode consists of several operations. Throughout the 1980s, Odin
identified up to seven operations (diegetization, narrativization, fictivization, and so on), which combine in different ways to constitute
the different modes of film.
Collectively, these eight modes and seven operations constitute
the specifically filmic dimension of the spectators competence, the
tacit knowledge necessary to the comprehension of all the various
groups and classes of films. Singularly, each mode (and its specific
set of operations) establishes a role that in turn constitutes the
spectators mode of attention towards that particular mode of film.
In Odins terms: A role can be described as a specific psychic
positioning (cognitive and affective) that leads to the implementation of a certain number of operations that produce meaning
and affects. Only when the director actant and the reader actant
adopt the same role (for example, when both implement the operations linked to the role established by the fictional mode) is a
space of communication . . . created. Adopting the same role creates
the impression that semanticists believe to be pre-established the
impression that communication resides in the transmission of a
message from a Sender to a Receiver. Semanticists reify this im-
84
pression of successful communication (i.e., conceive of it as a pregiven entity), whereas pragmaticists demonstrate that this impression is the result of a series of procedures. However, Odin does not
give a clear and thorough account of these procedures. I shall try
to explain why the actant director and actant reader adopt the same
role by means of Sperber and Wilsons principle of relevance. Although Odin makes no reference to this principle, his semiopragmatics shares most of its features (to the extent that both are
cognitive theories of pragmatics) and can be effectively described
within its framework.
The principle of relevance is based on the notion of maximal
utility of cognitive resources. It can aid film scholars in describing
the cognitive disposition a spectator needs to adopt in order to gain
pleasure from watching a classical film, just as Metz in his later
work employs psychoanalysis to describe the spectators psychic
disposition. The basis of Sperber and Wilsons principle of relevance lies in the relation between what they term contextual effect
and processing effort. The addressee will process information only
if it creates contextual effects that is, if it is new and relates to
information already acquired by the addressee. But the new information will not be processed at all unless the processing effort is
small.
If the new information has a large contextual effect and if, at the
same time, its processing effort is small, then it is what Sperber and
Wilson call relevant. More specifically, they define relevance in
terms of two extent conditions:
Extent condition 1: an assumption is relevant in a context to the extent that
its contextual effects in this context are large.
Extent condition 2: an assumption is relevant in a context to the extent that
the effort required to process it in this context is small.18
Relevance is therefore a comparative concept, involving a balancing

of input (processing effort) against output (contextual effects).
Sperber and Wilson argue that, when an addressee is confronted
with new information as presented, for example, in an utterance
she processes it by generating inferences in order to complete the
logical form of that utterance. Many inferences, of course, can be
generated. But, in accordance with the principle of relevance, the
inference the addressee generates will be the one that produces the
optimal contextual effect with only a small (adequate) amount of
85
processing effort. To communicate a particular message (or assumption), the sender produces an utterance that will be optimally
relevant to the addressee. The addressee assumes the utterance is
optimally relevant and will thus process it according to the principle of relevance that is, complete it by generating an inference
that enables the utterance to create the optimal contextual effect
with only a small amount of processing effort.
For Sperber and Wilson, then, the principle of relevance is mutually manifest to both sender and addressee. This is because it is
in the interest of the sender to communicate her message by producing an utterance that will yield in the addressee the optimal
contextual effect with only a small amount of processing effort.
This, of course, is also in the interest of the addressee, who will
thus automatically generate the inference in which such a result can
be achieved, which is precisely the inference corresponding to the
senders message.
It is therefore reasonable to assume that both sender and addressee abide by the principle of relevance when generating and
processing information. Yet there is no guarantee of this, for relevance is not built into the logical structure of language, but is a
contingent pragmatic principle. It is a rational but nonetheless fallible belief the sender holds about the information processing strategy of the addressee, and vice versa. In more formal terms, the
inferences generated by the addressee are not deducible from the
semantic content of utterances (i.e., are not demonstrative), but are
non-demonstrative hypotheses always open to revision (they cannot be proved or disproved, merely confirmed or disconfirmed).
Despite its contingent status, Sperber and Wilson argue, their principle of relevance is a universal cognitive-psychological principle of
information processing, rather than a principle that originates in
socio-cultural norms.
We can now argue that, within each of Odins filmic modes, the
textual cues make one role more relevant than other roles. We can
even go on to argue that the principle of relevance constitutes the
main institutional constraint on the process of making and reading
a film.
Odins binary conception of the space of communication (in
which the director actant and reader actant adopt/do not adopt the
same role) needs to be extended, since it is not simply a matter of
either/or either the actants roles match up and the space of
86
communication is created, or they dont match up, thus blocking

communication. The space of communication does not simply operate on one level, as Odin suggests, but on several levels. To
extend the conception of the space of communication beyond
Odins binary opposition, we can turn to David Bordwells fourfold
hierarchy of filmic meaning, as outlined in Making Meaning.19 Bordwells hierarchy consists of the following levels: (1) referential
meaning (the films diegesis, or spatio-temporal world), (2) explicit
meaning (the films conceptual meaning directly stated in the film),
(3) implicit meaning (the films symbolic or allegorical level of
meaning, commonly called its theme), and (4) symptomatic meaning (the films repressed level of meaning only involuntarily
yielded by the film through its re-reading, or interpretation, by
means of a master narrative, usually Marxism or psychoanalysis).
Bordwell gives the example of Psycho to illustrate these four levels:
You might assume that Psychos referential meaning consists of its fabula
and diegesis (the trip of Marion Crane from Phoenix to Fairvale, and what
happens there), and you might take its explicit meaning to be the idea that
madness can overcome sanity. You might then go on to argue that Psychos
implicit meaning is that sanity and madness cannot be easily distinguished.
. . . Taken as individual expression, symptomatic meaning may be treated
as the consequence of the artists obsessions (e.g., Psycho as a worked over
version of a fantasy of Hitchcocks). Taken as part of a social dynamic, it
may be traced to economic, political, or ideological processes (for example,
Psycho as concealing the male fear of womans sexuality).20
The first three levels of meaning (referential, explicit, and implicit)

are intentional, according to Bordwell, who develops as I have
argued in Chapters 1 and 2 a rational-agent model of filmmaking
and film reception. The space of communication can operate on any
of these three levels, whether the actants operate in the same level
or not. In making a film, the director actant may adopt the role
constructed on the level of explicit meaning, for example, and the
reader-actant may adopt the role constructed on the level of implicit
meaning; but despite these different roles, a space of communication is still created. It is only when the reader-actant adopts the role
constructed on the fourth level, that of symptomatic meaning (in
which the reader-actant identifies meaning involuntarily revealed
by the text), that we see the space of communication break down
(as it does in other cases i.e., when the spectator is bored), since
the reader-actant is breaking with the films fictional contract. In
87
Metzs terms, such a reader-actant (who is usually a theorist) attempts to disengage the cinema-object from the imaginary and to
win it for the symbolic,21 that is, attempts to produce a knowledge,
or a little more knowledge, of films discursive properties, rather
than reify its imaginary pleasures. Similarly, Bordwell comments,
When operating within the institution of film criticism, perceivers
are likely to use the film to produce implicit and symptomatic
meanings, regardless of the filmmakers intent.22
Identifying different levels of meaning has not only a synchronic
dimension (Bordwells approach), but also a diachronic, or historical, dimension. In Melodrama and Meaning,23 Barbara Klinger analyzes the spectators reading strategies at different historical periods
that confer upon Douglas Sirks films different meanings: Sirks
films have been historically characterized as subversive, adult,
trash, classic, camp, and vehicles of gender identification.24 Klingers book is concerned with the genesis of these historical readings
of Sirks films, and, in a move that parallels Odins work, she
locates meaning in institutions (although she does not theorize the
concept of the institution): I hope to contribute to recent research
focused on the manner in which institutional contexts most associated with the Hollywood cinema (such as academia, the film industry, review journalism, star publicity, and the contemporary mass
media in general) create meaning and ideological identity for
films.25 In her first chapter, Klinger offers a history of how Sirks
films have been received within the academic context, concentrating on the way critics and theorists privileged Sirks Brechtian
intentions, enabling them to label moments of subversion in his
American melodramas as intentional, or as forming part of the
films implicit meaning.
However, Klinger also addresses the issue of Rock Hudsons
star image in Sirks 1950s melodramas in light of his announcement
in the 1980s of his homosexuality, dramatically revising his star
persona as the quintessential heterosexual male. This reversal created a new level of meaning in Sirks films: This reversal has
undoubtedly helped define contemporary responses to Sirks films
in ways totally unforeseeable by their original creators.26 Klinger
suggests here that the revision of Hudsons star persona breaks the
space of communication between director actant and reader actant,
since the announcement of his homosexuality creates a meaning for
contemporary spectators that subverts his 1950s star persona. Run-
88
ning parallel to, but distinct from, the Brechtian subversion that
forms part of the intentional meaning of Sirks films, the subversion
of Hudsons star persona is unintentional (does not form part of
Sirks intentional meaning). The contradiction between past star
persona and present extra-filmic knowledge of Hudsons private
sex life creates artifice around Hollywoods representation of heterosexual romance: A self-reflexive and distancing element is thus
introduced into the spectatorial experience of these films. As a
result, Sirks films were most likely made strange in ways totally
unforeseeable by him or his critics.27
Modes
Odins term mode encompasses Metzs divisions within the category of filmic texts classes and groups of films. However, whereas
Metzs categories remain resolutely taxonomic, Odins modes are
cognitive they refer to a major part of each spectators discursive
competence. The space of communication is created when sender
and receiver employ the same competence the same mode in
the making and reading of a film.
In his essay Semio-pragmatique du cinema et de laudiovisuel:
Modes et institutions,28 Odin identifies and defines sometimes
vaguely eight different modes operative in the filmic field (several
of these modes, together with examples, will be discussed later in
this chapter):
1. The spectacle mode, in which the spectator perceives the film as
a spectacle. Odin is using spectacle in the same way that Dayan
and Katz define this term in their tripartite distinction between
spectacle, ceremony, and festival (see Chapter 3): Spectacle involves a minimal level of interaction between spectators and
events; in this sense, film constitutes a spectacle because it minimizes interaction between the spectator and filmed events as a
result of the spatio-temporal displacement inherent in films.
2. The fictional mode, in which the spectator resonates to the
rhythm of the narrated events ( faire vibrer au rhythme des evenements racontes). In other essays, Odin uses the phrase mise en
phase. Both resonates and mise en phase suggest that there is
a correlation between the film-spectator relation and the relations
manifest in the diegesis.
89
3. The dynamic mode, in which the spectator resonates to the

rhythm of the films sounds and images. The difference between
the fictional mode and the dynamic mode is that, in the fictional
mode, the spectator resonates with the diegetic events, whereas
in the dynamic mode, the spectator resonates with the images
and sounds as autonomous (non-representational) dimensions of
the film. Odin gives the example of Moroders version of Metropolis, which is colorized and has a rock soundtrack (more on this
later).
4. The home movie mode, which a spectator adopts when the film
depicts her or others past lived experiences.
5. The documentary mode, which in this instance at least Odin
defines as informing the spectator of real events. We shall see
later that Odin also defines the documentary in terms of the real
(as opposed to fictional) status of its enunciator.
6. The instructional (or persuasive) mode, from which the spectator
draws a moral, or truth.
7. The artistic mode, in which the film is perceived usually by a
cinephile spectator as the work of an auteur.
8. The aesthetic mode, in which the spectator focuses on the technical work that has gone into the creation of the images and
sounds.
Collectively, these modes constitute what Odin calls a sort of

polylectal grammar of film,29 a structured heterogeneity of interacting (conflicting and overlapping) modes. This position challenges
any attempt to develop a semiotics of the cinema: The cinema as
homogeneous and unique institution does not exist; there only exist
cinemas relevant to different institutions: classical fiction film, instructional film.30 Odins concepts of modes and institutions of
film attempt to challenge reductive studies that use the term film
in an absolute sense.
One consequence of a cognitive-pragmatic film semiotics is that
the modes (and operations) are not presented as textual features
embedded in films. No film is inherently a fictional, documentary,
or avant-garde film, because its mode is only determined through
the process of reading. A spectator who traditionally reads a film
within one institution and mode can therefore read it in the context
of another institution and mode. This, in fact, is what academic film
scholars do when they produce symptomatic readings of films:
90
They study a fiction film in terms of its non-fictional context. For

Edward Branigan, this is the basis of filmic interpretation:
It is apparent that one aim of a global interpretation is to propose (new,
nondiegetic) contexts in which a fiction may be seen nonfictionally, that is,
seen to have a connection to the ordinary world. Under an interpretation,
a fiction may offer knowledge about the real world and may come to
possess, if not a truth value, then at least relevance, significance, appropriateness, value, plausibility, rightness, or realism.31
Branigans claim is premised on the assumption that fiction establishes an indirect relation to the ordinary world, and that the practice of filmic interpretation aims to explicate that indirect relation.
Jan Simons has also expressed this pragmatic notion by arguing
that not only do different kinds of films require from their viewers
different dispositions, but different dispositions towards the same
film are also possible.32 Simons identifies four dispositions/modes
of attention a spectator can take toward the same film: fictivization
(in which the spectator fully engages with the films diegetic illusion), poetic (in which the spectator focuses on the formal dimensions of a film), documentizing (in which the spectator comprehends a film in terms of her real world beliefs and extra-filmic
reality), and allegory (in which the spectator perceives a film as the
real expression of an actual directors beliefs). Simons gives the
example of Fritz Langs The Blue Gardenia:
A film like The Blue Gardenia for instance can be appreciated because of a
maximal involvement in the diegetic illusion (fictivization), but can also
be interpreted as Langs comment on human fate (allegorizing), analyzed and criticized as a document of sexist and patriarchal ideology (documentization), but also an example of exercises in style (poetic).33
It is instructive to compare this fourfold reading of The Blue Gardenia with Bordwells fourfold characterization of Psycho: The most
direct overlap is between Bordwells implicit meaning and Simonss
allegorizing reading, and Bordwells symptomatic meaning and Simonss documentization. Both frameworks outline how to do
things with filmic texts, although Simons is clear to state (as does
Odin) that no spectator is free to read any film within any institution and mode, because of contextual constraints. In other words,
something like the principle of relevance guides this pragmatic
process. (However, Bordwell maintains that a symptomatic reading
91
simply involves the arbitrary imposition of theoretical frameworks

onto films.)
After outlining the main modes of filmmaking, Odin goes on to
characterize the specificity of each mode in terms of a combination
of operations.
Operations
In his latest essays, Odin identifies six operations, which he groups
under three categories. In earlier work, he identified up to seven
operations and simply listed them, one by one. Here I shall first
present Odins list of seven operations and then his more recent
reformulations, briefly noting any major differences between early
and later work.
Odin identifies the characteristics of each operation and emphasizes that no mode (and, therefore, no single filmic text belonging
to that mode) simply embodies one operation (operations cannot,
therefore, be identified with single films). In his essay Du spectateur fictionalisant au nouveau spectateur: approche semiopragmatique34 Odin lists the following operations:
1.
2.
3.
4.
5.
6.
7.
Figurativization
Diegetization
Narrativization
Monstration
Belief
Mise en phase
Fictivization
Figurativization. It is the most general or least specific operation and is responsible for visual analogy. Figurativization also
constitutes a procedure in Greimassian semiotics, in which it is
defined more precisely than in Odins work: It is necessary to
distinguish . . . at least two levels of figurativization procedures: the
first is that of figuration, that is, of the setting-up of semiotic figures
(a sort of phonological level); the second would be that of iconization, which aims at decking out the figures exhaustively so as to
produce the referential illusion which would transform them into
images of the world.35 Umberto Ecos analysis of iconic signs into
ten discrete codes organized in terms of a triple articulation went
92
some way to detailing both figuration (the ten codes) and iconization (their triple articulation).36
Diegetization. This operation involves the construction of an imaginary world inhabited by characters. It refers to the literal dimension of the film the space, time, and events experienced by
characters. (As such, it presupposes the prior operation of figurativization.) In Bordwells terminology, diegetization operates on a
films most basic level referential meaning. Odin emphasizes that
diegetization is an operation constituting part of the director actants and reader actants discursive competence, suggesting that
the literal level of filmic meaning is not simply embedded in the
film, but is part of an actants judgment. Edward Branigan comes
to a similar conclusion: Diegesis is not something that the film
either possesses or lacks, but rather is a way of describing an interlocking set of judgments we make about the presentation of sensory
data in the film at a particular moment.37
Narrativization. To make the concept of narrativization specific,
Odin distinguishes between the narrative effect and the story.38
Whereas the narrative effect is a micro-narrative element and can
be found in the most abstract, avant-garde films, the story refers to
a films overall (or macro) narrative structure. Furthermore, Odin
characterizes the story in Greimassian terms as two complete actantial structures:
It seems, in particular, that all stories require at least the intervention of
two antagonistic Subjects (a Subject and an Anti-Subject), and therefore,
since there is no Subject without an Addresser and without an Object, no
operational Subject without a Helper, and no Object without an Addressee,
two complete actantial structures (one for the Subject, the other for the
Anti-Subject).39
We have seen that diegetization presupposes figurativization. However, diegetization does not presuppose narrativization, since the
construction of an imaginary world can be organized around description rather than narration. But narrativization cannot operate
without diegetization, since the two actantial structures that constitute narrativization require an imaginary world in which to function.
Odin refers to Gerard Genettes claim that narration does not
93
exist in the cinema because of absence of a mediator who introduces

a distance between the activity of telling and the story told.40 Genettes claim is based on the analogical nature of the filmic image
(on its operation of figurativization). However, Odin refers to two
further operations (belief and monstration) that challenge Genettes
claim.
Belief. Any reference to this concept today must go beyond the

facile gloss that film viewing is based upon a suspension of disbelief, for, as John Searle has argued: Every subject matter has its
catchphrases to enable us to stop thinking before we have got a
solution to our problems. . . . It is easy to stop thinking about the
logical status of fictional discourse if we repeat slogans like the
suspension of disbelief or expressions like mimesis. Such notions
contain our problem but not its solution.41 The notion of suspension of disbelief is based upon the splitting of the spectator actants
psychic disposition, in which her attention is drawn away from (or
in which she disavows) the mediator. The notion is therefore useful
in refuting Genettes claim, but does not sufficiently characterize
the spectators mode of attention.
Richard Allen has clarified film theorys facile gloss of the concept of disavowal as the suspension of disbelief by distinguishing
between normal and pathological forms of disavowal, and by disengaging the concept of disavowal from the concept of fetishism.42
These two aims are, of course, interrelated, because pathological
disavowal is based upon fetishism and involves a fundamental
contradiction in the spectators beliefs what Freud called a splitting of the ego in which one set of beliefs are conscious and an
antithetical set are unconscious. (This is the form of disavowal
traditionally adopted by psychoanalytic film theorists.) Normal disavowal, on the other hand, is benign and does not involve a contradiction of beliefs. In the cinema, this means that pathological disavowal involves a spectators actually believing in the reality of a
fiction films illusion, whereas normal disavowal involves the spectators merely entertaining in thought the fiction films illusion
while all the time knowing that it is an illusion. In normal disavowal the spectator remains a conscious, rational agent, whereas
pathological disavowal involves a conflict between a spectators
conscious and unconscious reactions to a fiction film:
94
A spectator of a given disposition might respond to a film as if it were real

and behave as if the world on the screen and the world of experience
formed an undifferentiated continuum. Fraught with the anxiety correlative to such an acting out of fantasy, his pathological spectator would at
once ward off the anxiety and keep his experience intact through a splitting
of the ego: I know this is a film, but all the same I believe it to be real. A
case like this may illuminate the character of cinemagoing in general, but
it would be absurd to claim that this is really the way in which the cinema
spectator characteristically behaves. . . . The movies do not cause the ego to
lose its grip on reality and unconsciously defend against this loss by splitting.43
Monstration. Odin borrows this term from Andre Gaudreault, for

whom it means to show, and who opposes it to narration (to
narrate).44 Furthermore, monstration is equated with spatial representation and the theater, and narration with temporal representation and written (or scriptural) narrative. For Gaudreault, cinema
combines monstration and narration, since the shot shows space,
whereas the cut (and, more generally, editing) introduces temporal
and narrative articulation between shots.
Mise en Phase. This concept has undergone significant transformation in Odins work. He first introduced it in his 1983 paper
Mise en phase, dephasage et performitive dans Le Tempestaire de
Jean Epstein.45 All of the three concepts found in the title of the
essay function to specify different relations established between the
film and the spectator. Mise en phase designates the moments in a
film when the relation between film and spectator resonates with
the relations manifest in the films diegesis: The homogeneity between the positioning of the spectator and of the narrative dynamic
manifest in the diegesis constitutes one of the essential characteristics of the functioning of the fiction film (one of the operations of the
fiction-effect).46 Mise en phase is not to be equated with the conjunction between film and spectator. Instead, it involves a correlation between the film-spectator relation and the relations manifest
in the diegesis. In other words, mise en phase is created when a
conjunction within the diegesis is matched by a conjunction in the
film-spectator relation, and also when a disjunction in the diegesis
is matched by a disjunction in the film-spectator relation. In this
way, the spectator is said to resonate to the rhythm of the events
told.
95
Odin offers two examples of mise en phase in Du spectateur

fictionalisant:
If the operation of mise en phase has a meaning, it is just a matter of
mobilizing all the materials of film rhythm, music, all the dynamics of
montage, all the plays on looks and framings, etc. to make the spectator
resonate to the rhythm of the narrated events. Thus, a shipwreck scene
[in order to create mise en phase] will be filmed using blurred images,
staccato camera movements, misframings, etc. to the point of illegibility, to
realize in the film-spectator relation the aggressivity manifest in the storm.
. . . Inversely, carefully framed long shots, a softly filtered light delicately
modelling faces, a poetic type of dialogue (with rhythm, assonances, timing), a diction directly anchoring speech in the musical inflections, will
make the spectator feel an extraordinary complicity with the secrecy between characters, as in the admirable scene in Partie de campagne where
Henriette and Juliette confide in one another under the cherry tree.47
By creating resonance, mise en phase functions as the principle of

intelligibility in the cinema, for it designates a correlation between
textual structures and spectator competence with one miming or
echoing the other.
Dephasage, or phase displacement, simply designates the opposite, in which the relation manifest in the diegesis (conjunction or
disjunction) is not mirrored in the film-spectator relation. The spectator does not, therefore, resonate with the diegetic events. According to Odin, phase displacement is created in Le Tempestaire through
Epsteins use of non-professional actors, who never conform to the
impression of reality evident in other fiction films. For Odin, this
alienates many spectators, leading to the breaking of the fiction
effect (since, as Odin points out in the preceding excerpt, mise en
phase is one of the important operations of the fiction effect). The
choice spectators face is then simply to condemn the film as badly
made or to comprehend the use of non-professional actors as a
reflexive device, drawing attention to filmic discourse, rather than
to story. Odin takes the second option, arguing that Le Tempestaire
is reflexive or, in his terms, it is a performative text. Later we
shall see that, in his later work, Odin distinguishes between two
types of mise en phase (fictional and dynamic) and changes its
status.
Fictivization. Unlike the other operations Odin discusses, fictivization is unique to one mode of film the fiction film. The function
96
of this operation is to confer a fictional status upon the enunciator

and addressee of a fiction film:
Fictivization is a modality which is applied to the enunciative structure of
film, an intentional modality that characterizes the status, or position, the
spectator attributes to the fiction films enunciator. Fictivization consists of
considering that the enunciator does not intervene as a real origin but as
a fictive origin: it accomplishes the act of enunciation without assuming
the commitments which are normally required for this act [John Searle].48
Similarly, the spectator no longer feels interpellated as a real person having to take seriously what is narrated to him.49
Odin characterizes fictivization in Greimassian terms as a modality that is, in terms of the subjects modification of meaning
according to its mode of existence. In fictivization, the modification
of the films meaning derives from the enunciators modalization
as fictive, and the reader actants modalization as imaginary. The
result is that the addressee need not take as serious (as real) the
meanings articulated in the film, but comprehend these meanings
in terms of non-deceptive pretense. Documentary films meaning is
modalized differently: Both enunciator and reader actant are modalized as real, with the result that the addressee must take seriously
what is articulated on screen. One consequence of this theory is
that the fictive/documentary character of a film is determined, not
by the (un)reality of the profilmic events, but by the modal status
attributed to the addressee and enunciator. Within a semiopragmatic framework, a films status is defined in terms of its mode
of address and the modality conferred upon its enunciator.
Odin defines the fictional mode of film as possessing all seven
operations, which combine to create what he calls the fiction effect
(and what Noel Burch significantly calls the Institutional Mode of
Representation). Other modes are defined by the presence or absence of these various operations.
In his 1994 essay Semio-pragmatique du cinema et de laudiovisuel: Modes et institutions,50 Odin identifies the following
three types of operations:
(a) Operations concerning the representational status of images and
sounds
(b) Discursive operations (diegetization, narrativization, discursivization)
97
(c) Enunciative operations (construction of an enunciative structure),

which includes the status conferred upon the dominant enunciator
Category (a) simply refers to figurativization (operation 1 in the

earlier list); category (b) refers to operations 2, 3, and 4 (it seems
that Odin has exchanged Gaudreaults term monstration for the
more general term discursivization); category (c) refers directly to
operation 7, fictivization, and all reference to belief structures is
dropped. The main difference in Odins reformulation is that mise
en phase is divided into two categories: the fictional and the dynamic. Furthermore, fictional mise en phase is defined as the result
of all the other operations (it is equated with the fiction effect,
rather than being conceived as one operation among others that
lead to the fiction effect).
Institutions
A priori, the spectator can make any mode function on any film (or fragment of film); in reality, we are always subject to constraints that limit this
possibility. Textual constraints: all films tend to block the working of certain
modes of production of meaning and affects; contextual constraints: seeing
a film is always made in an institutional framework which governs our
way of producing meaning and affects. In fact, . . . these two types of constraint can be reduced to one: institutional constraints.51
The space of filmic communication is delimited, in the final analysis, by institutions. The starting point of Odins reflections on institutions is (not surprisingly) the work of Metz. For Metz, institutions
consist of two dimensions, one mentalistic and internal, the other
material and external: The institution is outside us and inside us,
indistinctively collective and intimate, sociological and psychoanalytic, just as the general prohibition of incest has as its individual
corollary the Oedipus complex, castration or perhaps in other states
of society different psychical configurations, but one which still
imprint the institution in us in their own way.52 For Odin, the
external and material dimension of institutions consists of the
places in which the film is projected commercial cinemas (for
films realized within the fictional, spectacle, and dynamic modes),
art house cinemas (films of the aesthetic and artistic modes), schools
(films of the instructional mode) and so on whereas the internal
98
and mentalistic dimension comprises the modes and operations

constituting each spectators discursive competence, which determine the appropriate mode of attention necessary to the comprehension of a film.53 Odin adds that each film is to some extent
imprinted with an institutional identity, and he specifies two procedures of inscription: the films display of its preferred institution
in its credit sequence and the spectators recognition in the film of
a thematic and stylistic system characteristic of a certain mode of a
certain institution. Furthermore, as mentioned in the discussion of
modes, the reading of a film is also determined by the principle of
relevance. This principle encourages a cinephile to allegorize an
auteurist film, whereas its fictionalization is the preferred reading
of naive (non-cinephile) spectators; film academics work against
the principle of relevance in order to study a films form or to
produce a symptomatic reading (leading to a deliberate break in
the space of filmic communication).
Each institution consists of several modes for example, the
institution of commercial cinema comprises the spectacle, fictional,
and dynamic modes, whereas the institution of non-professional
cinema includes the home movie, aesthetic, and artistic modes.
Institutions determine the mode in which a particular film is to be
comprehended. For Odin, an institution is a bundle of determinations which govern the production of meaning in selecting, hierarchising, and structuring the modes of production of meaning which
are put to work.54
Finally, just as no operation is specific to a particular mode (with
the exception of fictivization), no mode is specific to a particular
institution. The relations among institutions, modes, and operations
are relations of inclusion: Operations are included in modes, and
modes are included in institutions. Figure 5 presents a summary of
the main institutions, modes, and operations that Odin identifies.
The Documentary, Home Movie, and Dynamic Modes of

Filmmaking
The Documentary Mode. As pointed out, for Odin, the only essential difference between the fiction and documentary modes of film
lies in the operation of fictivization the modal status conferred
upon the enunciator and addressee. In the fiction film, the enunciator and addressee are modalized as imaginary (or absent), whereas
99
INSTITUTIONS
External:
commercial cinemas, schools,
art house cinemas, etc.
Internal:
discursive competence
(comprising modes and operations)
MODES
spectacle, fictional, dynamic, home movies,

documentary, instructional, artistic
OPERATIONS
figurativization, diegetization, narrativization,

monstration, belief, mise en phase, fictivization
in the documentary, they are modalized as real. Here I shall develop this idea further, with particular reference to Odins essay
Film documentaire, lecture documentarisante.55
Firstly, in distinguishing fiction from documentary modes of
filmmaking, Odin notes that the documentary film has no privileged relation to reality, since both modes employ the same technologies mechanics, optics, and photochemistry. All modes that
employ these technologies therefore record real events. The fiction
mode, for example, is a record of actions and events performed by
actors at a certain time in a certain place (either on location or in a
studio). Rejecting a semiotics of realization as a criterion for
defining the documentary mode, Odin instead opts for a semiotics
of reading to define its specificity.56 In Greimassian terms, Odin
rejects a referential theory of truth (study of the relation between
signs and their extra-textual reality) for a study of veridiction the
modality of truth/reality as articulated by enunciator and addressee.57 Most of Odins essay is concerned with characterizing the
specificity of the documentary mode according to the documentarizing reading strategy adopted by film spectators, and with outlining how this documentarizing reading is triggered by the film and
the institutions in which it is screened.
To posit an imaginary/absent enunciator for the fictional mode
involves not only the texts concealing its own marks of enunciation
Fig. 5
100
(the film as histoire) but, more radically, the spectators refusing to

posit the existence of any process of enunciation at all. The first is
an internal constraint, the second an external constraint: The opposition story vs discourse is an effect of the text (it is identifiable
through the marks inscribed in the utterance), whereas the opposition fictivizing reading vs documentarizing reading is an effect of
the positioning of the reader facing the film, the result of an operation external to the film: a strictly pragmatic operation.58 As noted,
because of its pragmatic status, the mode of reading adopted by
the reader actant is not tied to one mode of filmmaking. The documentarizing reading is therefore one level on which spectators can
read any mode of filmmaking; it is not a reading strategy limited
to one mode. But the documentarizing reading is privileged only in
films generated by the documentary mode and institution. Odin
begins to consider textual cues that privilege the documentarizing
reading in the documentary film.
Firstly, he considers the different types of real enunciators that
spectators can posit:59
1. The reader can take the camera to be the real enunciator. In this
framework, all the profilmic events are subjected to the documentarizing reading. Odin finds the best example of this argument in
the work of Siegfried Kracauer.
2. The reader can take the cinema (or the cinematic apparatus) to be
the real enunciator. Here, Odin refers to the work of Stephen
Heath.
3. The reader can take as the real enunciator the society in which
the film is produced. Although it is not noted by Odin, this
position can be attributed to theorists who practice symptomatic
interpretations, particularly those who take symptomatic meaning to be part of economic, political, or ideological processes of
the social dynamic. In the Psycho example discussed earlier, the
enunciator of the film is deemed to be patriarchy, which attempts
to conceal male fear of womans sexuality.
4. The reader can take the cameraperson to be the films real enunciator. (This category, which is usually applied to the reportage
film, will be discussed later.)
5. The reader can take the director to be the enunciator, as auteur
critics do. In Bordwells terms, an auteurist reading operates on
the implicit level of meaning, since auteur critics are not compre-
101
hending the text literally, but allegorically. This is a non-fictional

reading to the extent that auteur critics are reading the film as
part of an actual directors oeuvre.
6. Finally, the reader can take the specialist responsible for the film
to be its enunciator, a reading strategy relevant to instructional
films, for example.
Despite the pragmatic account Odin offers of the readers construction of these various enunciators, his most persuasive example,
relating to the reportage film, is primarily a textual account. In
reportage films, a number of specific textual figures prompt the
reader to take the cameraperson as the embodiment of the real
enunciator. These figures include a blurred image, jolting camera
movements, hesitant pan shots, abrupt editing, long sequence shots,
insufficient light, film grain, direct sound (as opposed to studio
sound), and real background ambience. (Odin notes that these figures are also imitated by fiction films that want to look like reportage films.) These textual figures serve as an index of the camerapersons real presence:
The function of this set of figures is clear: to mark in the very structure of
the film the real existence of the cameraman; to make known to the reader
that the cameraman is to be taken as the real Enunciator. The figures cited
testify to the difficulty the cameraman had filming in the conditions he
found himself in, and to his physical interaction with the event, to see the
risks he took.60
Whereas the technical perfection of many fiction films enables the

reader to posit the enunciator as imaginary or absent, the documentary film (and reportage film in particular) signifies its concrete and
contingent location in the profilmic world. In other words, the
cameraperson is always locatable or grounded on a specific point,
and it is this anchoring in a specific point that leads the spectator
to infer a real enunciator (a hand-held camera, for example, marks
the actual anchoring of the cameraperson in the space and time of
the events she is filming). In news bulletins, the cameraperson is
subject to the forces of the events she is filming; she does not exist
outside or apart from the events being filmed, but is subject to their
laws. She can therefore attest to the existence of the events filmed,
and the preceding textual figures act as an index of her presence,
her status as witness. This in turn modalizes the addressee as a real
102
spectator as one who must take seriously what is presented to

her.
The Home Movie Mode. The study of the home movie seems to
be one of Odins greatest passions, and also the motivation for his
work: It is, at least in part, because of this work on the home
movie that I came to experience the necessity of constructing a
semio-pragmatic model [of film].61 Here I shall present an overview of Odins work on the operations that characterize the home
movie, making particular reference to his essay Rhetorique du film
de famille.62
The institutional characteristic of the home movie is that it is
shown within the domestic space, in which both the enunciator and
addressees are family members: A home movie is a film made by
a member of a family, with regard to objects or events linked in
one way or another to the history of this family, and for the privileged viewing by members of this family.63 Odin then discusses
eight textual characteristics that prompt the spectator to read a film
as a home movie:64
1. The absence of closure (the film does not embody the operation
of narrativization).
2. Discontinuous linear temporality. The home movie is dedicated
to chronological order (there are no flashbacks, flashforwards, or
alternating montage); however, that chronology is indeterminate
(the temporal relations between the fragments of the home movie
are discontinuous). The home movie does not, therefore, embody
the operation of diegetization.
3. Spatial indeterminacy. As with (2), the operation of diegetization
is not adhered to because it is often difficult to determine the
spatial relation between shots.
4. Dispersed narrative. This point is a continuation of (1); not only
does the home movie lack closure, but its narrative elements do
not constitute a macro-narrative (it therefore only contains micronarrative elements).
5. Jumps. The home movie disregards the 180 degree axis of action
line and ignores the 30 degree rule, shot/reverse shot, and the
other codes of spatial coherence. As with (2) and (3), this textual
figure signifies a disregard for the operation of diegetization.
6. Blurred images, jolting camera movements, hesitant pan shots,
103
and so on. As with the reportage film, the enunciator of the home
movie is modalized as real. However, Odin resists comparing the
home movie to the reportage film, because they are shown in
different institutions.
7. Address to the camera. The family members in home movies
frequently look at the camera, thus blocking the fiction effect by
acknowledging its presence (the camera and its operator are
therefore located in the diegesis through acknowledgement by a
direct address).
8. The sound of a home movie may be inaudibe, variable, or completely absent, with the effect as with (2), (3), and (5) that it
only presents fragments of a diegesis (nonconformity to the operation of diegetization).
Odin then returns to a pragmatic perspective on the home movie

to discuss in more detail its relation to the operation of diegetization. From a pragmatic perspective, the diegesis not only is conceived as a textual construction, but is also a product of the spectators competence. If the home movie is projected in its institutional
context (in the domestic space of the family who is represented),
then we can argue that the operation of diegetization is put fully
into effect. Odin notes, The home movie, in fact, has as its particularity to be made to be seen by those who have lived (or seen)
what is represented on the screen.65 The home movie itself does
not need to establish a coherent diegesis because the events filmed
are already known by the intended addressees. All the home movie
needs to do is revive the addressees memory of the actual experiences. The home movie recalls a previous series of events; it does
not need to narrate those events. The absence of the operations of
diegetization and narrativization is therefore an essential characteristic of the home movie, and this is why Odin is critical of cineclubs that encourage home movie makers to adopt the techniques
of the fictional mode. It is only outsiders (nonfamily members)
who experience the home movies of others as fragmented. Operating within its intended institution, the home movie can function
without the need for the operations of diegetization and narrativization. This leads Odin to state the paradox generated by the home
movie: For the relevant members of the family, the home movie
generates pleasure through the diegetic and narrative poverty of its
images. By not presenting the addressee with a complete and co-
104
herent diegesis, the home movie has less chance of conflict with, or
contradict, the addressees experience.
Finally, Odin rather confusingly ends his analysis of the home
movie by arguing that it is still involved in the creation of a fiction,
despite the fact that it does not embody all the operations necessary
for the creation of the fiction effect. The home movie creates a
fictional representation of a particular family through its selective
portrayal of events in that familys life the special occasions and
anniversaries, rather than everyday reality. The home movie functions to guarantee the institution of the family (and this is why, in
French, the home movie is called film de famille): The home movie
is not an innocent film; it contributes in its own way to the maintenance of a certain order.66 However, I do not think we should
understand Odins use of the term fiction in the context of the
home movie to mean the same as the fiction effect as defined earlier.
The home movies creation of a fiction involves its presentation of
an imaginary unity (the family) rather than its utilization of the
mode of attention a spectator adopts when watching a fiction film
(the mode of attention that encourages the spectator to read the
film in front of her as a fiction film).
The Dynamic Mode. In the second half of his essay Du spectateur

fictionalisant, after outlining the seven operations that constitute
the fiction effect, Odin writes about the emergence of the dynamic
mode of filmmaking, which derives from the institution of commercial cinema. Although shown in commercial cinemas, films belonging to the dynamic mode do not create the fiction effect, because
they do not embody all of its operations. Odin mentions films such
as Star Wars, Tron, Cobra, and the Mad Max and Rocky series but
gives particular attention to Moroders modified version of Langs
Metropolis.
Odin charts the modifications that transformed this film from
one conforming to the fictional mode to one conforming to the
dynamic mode. The main difference between these two modes is
that, in the fictional mode, the spectator resonates with the narrated
events and, in the dynamic mode, resonates with the sounds and
images. In the terminology established in Chapter 2, the dynamic
mode operative in Moroders Metropolis prevents the spectator from
comprehending Langs Metropolis in terms of the container schema,
105
since the spectator cannot enter into the diegetic space of the new
version of Metropolis.
Moroders modifications, which transformed Metropolis from a
fiction to a dynamic film, include the addition of tinting, the rewriting of intertitles, the addition (sometimes in the form of still
photographs) of new material edited out of the original version and
the introduction of a rock music soundtrack. For Odin, all these
modifications work to block the fiction effect: In fact, he argues that
six of the seven operations listed in the first half of his essay are
affected, and only figurativization remains intact.
Odin argues that whereas black and white contributes to diegetization and the effect of depth, colorizing a film today draws
attention to film as film: Colorization appears to be a process of
derealization opposing at the same time monstration, diegetization,
and belief. With colorization, the image effect takes over from the
fiction effect.67 The spectators look does not go beyond the surface of the image into the imaginary elsewhere of the films diegesis, but remains on the surface of the image. We can say that the
transformation taking place from the fictional mode to the dynamic
mode is that between a gaze aesthetic and a glance aesthetic, respectively.
With the addition of still photographs into the film, a rupture of
the impression of movement takes place. Many of these still photographs are filmed by using a moving camera, a technique that, for
Odin, simply highlights the absence of movement in the images.
Furthermore, the manipulation of the intertitles (making the letters
dynamic, and so on) draws attention to them as lettering. In general, the plastic elements in the image track have a semiautonomous status in relation to the operation of narrativization;
they do not function to further the films narrative.
But it is primarily the addition of a rock music soundtrack that
transforms Metropolis into a film belonging to the dynamic mode.
Odin argues that the organization of the soundtrack into musical
movements opposes the original films organization into narrative
segments, with the result that the story merely illustrates the
soundtrack, rather than the reverse, as is the case in films of the
fictional mode. One technique that privileges sound over image is
what Odin calls fine (or micro-) synchronization. In films of the
fictional mode, synchronization between extra-diegetic music and
the story only operates on the macro level, whereas diegetic sound
106
and music are synchronized on the micro level.68 But in Moroders

Metropolis, the non-diegetic music is synchronized with the story,
with the result that small actions in the story particularly gestures
are synchronized to the rock soundtrack. A non-fictional dimension of the film (the rock music soundtrack) seems to be conjoined
to a fictional dimension (character action). For Odin, this blocks the
operations of narrativization and diegetization and aligns Moroders Metropolis with other films dominated by the soundtrack:
It is this relation (micro-synchronization of the elements of the
image which are not sources of the sound) that dominates in dance
films, musical comedies, opera films, burlesque films.69
Odin conceives a reversal taking place in Moroders Metropolis,
in which its narrative is a mere adjunct to the plastic elements of
the image and soundtrack. The result is that the film acts directly
on the spectator, a spectator who no longer resonates with the
narrated events (the fiction-effect) but resonates with the variations
of rhythm, intensity and color of the images and sounds.70 In the
dynamic mode, the space of filmic communication is not governed
by the production of meaning but by the production of affects.
Odin ends his analysis of Moroders Metropolis by generalizing
about the dynamic mode of filmmaking and the new spectator it
constructs. He links the emergence of the dynamic mode to the
decline in the role of fiction, metanarratives, and the end of the
social in postmodern culture. However, it is equally possible to take
his argument in another direction: to link it to the more traditional
genre of the decorative arts. Bordwell has established a mode of
filmmaking called parametric (or style-centred) narration by referring to Gombrichs work on decorative arts: Borrowing from E. H.
Gombrichs account of decorative art, I have argued that parametric films organize film techniques in patterns that may create an
ongoing spectatorial engagement independent of narrative action.
The films of Yasujiro Ozu, with their nonnarrative structures of
locale and shot composition, offer good examples of what Gombrich calls order without meaning.71 Bordwell also refers to the
serial music of Messiaen, Pierre Boulez, Karlheinz Stockhausen,
Luigi Nono, and Jean Barraque, together with the work of the
nouveaux romanciers, as other instances of style-centred artistic practices.72 Bordwell also notes that parametric narration in the cinema
is not linked to a single national school, period, or genre of filmmaking. . . . In many ways, the pertinent historical context is less
107
that of filmmaking than that of film theory and criticism. To some

extent, then, this mode of narration applies to isolated filmmakers
and fugitive films.73 For this reason, it is possible to consider the
dynamic mode of filmmaking to be a style-centred form of cinema
operating within the context of New Hollywood Cinema, rather
than to be a symptom of the end of fiction and the social.
Theories of pragmatics focus critical attention away from texts in
themselves and toward the actants that create and use these texts.
By means of the concepts of institution, mode, and operation,
Odin is able to avoid falling into the trap of positing the actants as
the sole producers of filmic meaning. Instead, for Odin, the actants
are simply points of passage for a multitude of determinations.
Nonetheless, Odins conception of actants requires refinement, especially in the light of comments made earlier concerning Bordwells discussion of symptomatic meaning in Psycho and Simonss
gloss on a documentizing reading of The Blue Gardenia. In brief, it
is misleading to modalize actants (the enunciator and addressee) of
the fictional mode of filmmaking as imaginary/absent and the actants of the documentary mode as real. A symptomatic/documentizing reading of a fiction film will posit its actants as real although on a latent level since this type of reading attempts to
relate that fiction film to its real, socio-historical context (usually
capitalism and patriarchy).
This suggests that whereas documentaries modalize their actants as real on the surface level, fiction films modalize their actants
as real on a latent level. This reformulation enables us to begin
refining our understanding of the relation between fiction and reality. As I mentioned in the introduction to this chapter, for Odin
fiction and documentary films are separated by only one operation,
that of fictivization. If my suggestion, that the fiction films actants
are modalized as real on the latent level, whereas in documentaries
they are modalized as real on the surface level, is viable, this would
suggest that fiction and documentaries are even closer to one another, since the operation of fictivization would not separate them.74
Both fiction and documentaries are attempting to achieve the same
aim, although on different levels: Both are attempting to represent
social reality, but fiction achieves this end in an indirect manner.
Fiction is not to be opposed to reality, but is a means by which
reality can be indirectly represented.
108
Here we reach an anthropological discussion of fiction, an under-developed area in film studies at the present time. In literary
theory, however, Wolfgang Iser has recently developed a study of
literary anthropology.75 Iser does not ask, What is fiction?, but Why
do we need fiction?76 For Iser, only a pragmatic (rather than a
semantic) response will adequately answer this question. Whereas
semantics is confined to offering a definition of fictions explanatory
function free of its context (situation), a pragmatic answer addresses the contextual impact of fiction: Impacting as the pragmatics of fiction never loses sight of its situational function, whereas
explanation as the semantics of fiction aims precisely to make its
situational necessity disappear. Thus the pragmatic function unfolds the special use of fiction, and the special use determines the
individual nature of fiction.77 One pragmatic function of fiction,
noted previously, is defined by its relation to reality: It can present
a simplified image of a complex reality, but can also go beyond the
given into the realm of the potential or unrealized. Odins proclamation of the end of the social and the fictional is unfounded
because, if we take a pragmatic definition of fiction to its logical
conclusion (that is, take it into the realm of anthropology), we see
that it is an anthropological constant. Rather than positing its disappearance, it is more accurate to argue that the function of fiction
(together with the function of the social) is constantly renewed by
successive generations.
CHAPTER FIVE
All in the Mind?

The Cognitive Status of Film Grammar
The problem for the linguist . . . is to determine from the data of

performance the underlying system of rules that has been mastered
by the speaker-hearer and that he puts to use in actual performance.
Hence, in the technical sense, linguistic theory is mentalistic, since
it is concerned with discovering a mental reality underlying actual
behavior. (Noam Chomsky)1
My aim . . . is to argue for what I will gladly call a grammar of film,
and adhere at the same time to a conception of scientific work
broadly comparable to that of Metz (and [Sol] Worth), but advancing different theses as to the general nature of rules committed to
representing this model of the structure of film. (Dominique Chateau)2
ne indication that linguistics has become a mature science is its ability to handle in a formalized manner the
problem of grammaticality, or to define in formal terms the
boundary between grammatical and ungrammatical sentences. A
technical formulation of this boundary was one of Noam Chomskys primary aims in his transformational generative grammar
(TGG). We can claim that film theory has reached a relatively mature stage (in comparison with other disciplines in the humanities)
thanks in part to attempts to develop a film grammar based on
Chomskys early theories of TGG.
David Bordwell has expressed his surprise that theorists who
assign language a key role in determining subjectivity have almost
completely ignored the two most important contemporary developments in linguistic theory: Chomskys Transformational Generative Grammar and his Principles-and-Parameters theory.3 In this
chapter I shall focus on two cognitive film semioticians Michel
Colin and Dominique Chateau who have, contrary to Bordwells
assertion, been working with Chomskys theories since the early
109
110
seventies. One conclusion we can draw from their work is that

Metzs grande syntagmatique (GS) does not merely observe and organize filmic data, but formalizes in a set of explicit rules the film
spectators competence, or intuitive knowledge of film.
We saw in Chapter 2 how Michel Colin returned to Metzs
semiotic theory of film and subjected it to a cognitive re-reading,
transforming it into a cognitive semantics of film. In this chapter
we shall review how Colin re-read Metzs grande syntagmatique from
a cognitive perspective. Colins re-reading is heavily influenced by
what is generally known as the Standard Theory of transformational generative grammar (TGG), as presented by Chomsky in
Aspects of the Theory of Syntax. After outlining the basic components
of Chomskys Standard Theory and presenting Colins long and
detailed Chomskyan re-reading of Metzs GS, I shall focus on the
way deviant film sequences (those that seem to break the implicit
grammatical rules of film) are comprehended. I shall end by considering the cognitive status of Metzs GS.
Transformational Generative Grammar

The fundamental principle of TGG is elegant in its simplicity. TGG
answers the question, How can a language user understand a potentially infinite number of sentences? Infinity has no boundaries
or limits and cannot therefore be understood in human terms. In
other words, it has no cognitive reality. TGG reduces this infinity
to a finite series of steps, which can be understood in human terms.
In other words, these finite steps have a cognitive reality. A TGG
of film also begins with the issue of how a film spectator can
understand a potentially infinite number of films and similarly
reduces this infinity to a finite series of steps.
In TGG, what are these finite steps? The elegance of the Standard Theory lies in the following components: It includes a finite
set of recursive rules that generate deep structure sentences (or
basic strings) and a finite number of transformational rules that
transform these deep structures into a potentially infinite number
of surface structure sentences. Moreover, the grammar must be
constrained so that it generates all and only the sentences that native
speakers judge to be well formed, or grammatical. So, for Chomsky,
grammatical is a technical, rather than an evaluative or prescriptive term. In this technical sense, a grammatical sentence is one that
is generated according to the rules of a particular grammar (it is a
THE COGNITIVE STATUS OF FILM GRAMMAR
111
sentence recognized by that grammar). Such a sentence can thereby

be described as well formed with reference to that grammar.
The grammar generates deep structures by means of a series of
re-writing (or categorical) rules, which expand a single grammatical
category into a string of other grammatical categories, such as
S NP VP
NP Det N
This reads, Re-write Sentence as Noun Phrase and Verb Phrase,
and re-write Noun Phrase as Determiner and Noun. Re-writing
rules can be represented in the form of a tree diagram. Tree diagrams interrelate three elements: a root node (the topmost node);
preterminal nodes, which are dominated by the root node; and
terminal nodes (the lexical entries). At the end of re-writing rules,
the category symbols are replaced by lexical entries (e.g., the category symbol N is re-written as a particular noun). Strings containing grammatical categories are called preterminal strings, whereas
those containing lexical entries are called terminal strings (see
Figure 6).
Category symbols such as N cannot simply be re-written as any
lexical entry belonging to the class of nouns. Each category symbol
is analyzable into a set of complex symbols, a set of specified selectional features. Thus, a particular noun category symbol may be
represented in terms of the following selectional features: [N,
Common, Count, Animate, Human]. Only a noun that possesses these particular selectional features (such as dog cat) can
then be inserted into the preterminal string to form a terminal
string. The selectional features therefore act as restrictions, for they
specify the constraints placed on the choice of lexical items that can
be inserted into preterminal strings.
The transformational component mediates between the finite
deep structure sentences and potentially infinite surface structure
sentences and formally represents the structural relations between
sentences. Whereas structural linguistics employs the taxonomic
processes of segmentation and classification (or bracketing and labelling) to analyze the surface structure of language, TGG analyzes
the more abstract rules underlying these sentences. In Chomskys
terminology, structural linguistics is only observationally adequate (it is only able to describe a corpus of utterances), whereas
the Standard Theory of TGG is descriptively adequate. Within the
context of a descriptively adequate theory of grammar, the formal
112
NP
Det
[Root node]
VP
[Preterminal nodes]
NP
[More preterminal nodes;

V = verb]
Fig. 6
The
boy
kicked
Det
[More preterminal nodes]
the
ball [Terminal nodes,

complete with lexical entries]
rules constructed by the linguist represent the native speakers competence, a point Chomsky emphasizes throughout his work: On
the basis of a limited experience with the data of speech, each
normal human has developed for himself a thorough competence
in his native language. This competence can be represented, to an
as yet undetermined extent, as a system of rules that we can call
the grammar of this language.4 A grammar can be regarded as a
theory of a language; it is descriptively adequate to the extent that it
correctly describes the intrinsic competence of the idealized native
speaker. The structural descriptions assigned to sentences by the
grammar, the distinctions that it makes between well formed and
deviant, and so on, must, for descriptive adequacy, correspond to
the linguistic intuition of the native speaker.5 In a 1983 interview,
he stated: I would still want to resist what is a very common
assumption, and I think one that is totally wrong, namely that the
study of the abstract structure of language cant tell us anything
about what is sometimes called psychological reality or biological
nature. On the contrary, it is precisely telling us about psychological reality in the only meaningful sense of that word.6 Finally,
Chomsky has recently written that there can be little doubt that
knowing a language involves internal representation of a generative procedure that specifies the structure of the language.7 Ulti-
113
mately, a descriptively adequate grammar like TGG is not justified

by internal criteria of logical consistency, but must be adequate to
a psychological reality. This position, known as mentalism, advocates a strict identity (isomorphism) between theory and its object
of study, the mental or cognitive reality external to it.
In a critical assessment of the cognitive status of Chomskys
linguistics, Scott Soames identifies two types of competence models
and a performative model.8 For Soames, the aim of a performance
model is to account for virtually all aspects of speaker judgment,
sentence production, and sentence comprehension. This includes
not only the nature and operation of internalized linguistic rules
and strategies, but also the influence on linguistic behavior of nonlinguistic cognitive systems such as those involving attention, perception, memory, and general ability to reason.9 By contrast,
broadly construed theories of linguistic competence (Soamess
term) focus primarily on those aspects of the speaker-hearers cognitive capacity involved in the mastery and use of language: Their
subject matter includes the speaker-hearers internalized linguistic
rules, plus any processing or heuristic procedures whose sole or
primary applications are restricted to internal structures representing linguistic material.10 Finally, Soames mentions theories of linguistic competence, narrowly construed: These are theories of the
speaker-hearers internalized linguistic rules, minus any claims
about the processing techniques or heuristics that determine how
they are used.11 Soames characterizes Chomskys TGG as a theory
of competence narrowly construed, that is, a highly idealized (cognitively unreal) theory. I shall return to Soamess characterization
of Chomskys theory in the final section of this chapter.
The Grande Syntagmatique Revisited (Michel Colin)

Colins re-reading of Metzs grande syntagmatique (GS) consists of
two steps: (1) a reconsideration of its formal properties in terms of
the formal properties of the Standard Theory of TGG and (2) an
attempt to determine its cognitive status.12 The most striking aspect
of Colins writing is that it is exact (it aims to make explicit the
implicit assumptions of Metzs film semiotics, and to follow them
through to their logical conclusions), frequently fragmentary and
detailed (it exhaustively deals with particular problems), and, more
than any other film theory makes numerous references to special-
114
ized disciplines not only TGG, but also other areas of cognitive
science: artificial intelligence, philosophy of language, and computational linguistics. Thus his style and breadth of knowledge place
great demands on the reader and make discussion of his work a
long and arduous process. This explains the largely expositional
nature of my discussion of his work, and the mainly technical
nature of my criticisms.
For Colin, the GS consists of three types of object, which are
related in terms of the three types of object in TGG: a root node
(the autonomous segment), preterminal nodes (the six classes of
syntagmas), and terminal nodes (the eight types of syntagmas).
Colin begins his essay by representing Metzs formulation of the
relation between these objects in terms of the tree diagram represented in Figure 7.
As in TGG, this tree diagram represents a series of re-write
rules. The root node (autonomous segments) can be re-written as
syntagmatic type 1, the autonomous shot, or the syntagmatic class
of syntagmas. The class of syntagmas can be re-written as two other
syntagmatic classes achronological syntagmas and chronological
syntagmas. Achronological syntagmas can be re-written as two syntagmatic types: 2, the parallel syntagma, and 3, the bracket syntagma, whereas the class of chronological syntagmas can be rewritten as syntagmatic type 4, the descriptive syntagma, or the class
of narrative syntagmas. The class of narrative syntagmas can be rewritten as syntagmatic type 5, the alternate syntagma, or as the
class of linear narrative syntagmas. The class of linear narrative
syntagmas can be re-written as syntagmatic type 6, the scene, or as
the class of sequences. Finally, the class of sequences can be rewritten as two syntagmatic types: 7, the episodic sequence, and 8,
the ordinary sequence.
The relations between the syntagmas can be represented by the
following re-write rules:13
Autonomous segments
Syntagmas
autonomous shot
syntagmas
achronological syntagmas
chronological syntagmas
, etc.
115
Autonomous
segments
Syntagmas
Chronological
syntagmas
Achronological
syntagmas
Narrative
syntagmas
Fig. 7
Linear narrative
syntagmas
Sequences
The eight types of syntagma: 1 = autonomous shot, 2 = parallel syntagma,

3 = bracket syntagma, 4 = descriptive syntagma, 5 = alternative narrative
syntagma, 6 = scene, 7 = episodic sequence, 8 = ordinary sequence.
Colin points out that whereas the tree diagrams and re-writing
rules used in TGG represent a relation of conjunction, his tree
diagram and re-writing rules represent a relation of disjunction. In
other words, in TGG tree diagrams represent the syntactic relation
between concatenated symbols, whereas in the GS the tree diagram
represents alternative ways of re-writing the preterminal objects
(the synatagmatic classes) either as other syntagmatic classes or
as terminal objects, the eight syntagmatic types. Moreover, in the
GS the re-writing rules denote a relation of inclusion rather than
concatenation. So the syntagmatic types episodic sequence and
ordinary sequence are included within (are elements of) the syntagmatic class of sequences, which is itself included within the
syntagmatic class of linear narrative syntagmas, and so on.
Colin goes on to emphasize the difference between a deductive
reading and an inductive reading of the GS (a distinction intro-
116
duced, but not elaborated, by Metz). A deductive (or top-down)

reading of the GS considers it in terms of a series of re-writing
rules, for we begin from an abstract root autonomous segments
and continually re-write it until we reach the terminal objects, the
actual syntagmatic types. An inductive (or bottom-up) reading
starts from the syntagmatic types and defines them in terms of their
specific characteristics (it spells out those features of syntagmatic
types that make them identifiable). So an inductive reading considers each syntagmatic type in terms of its membership (relation of
inclusion) to syntagmatic classes. Colin outlines these two types of
reading in the following way: The [relation of inclusion] depends,
of course, on the direction in which the tree-diagram is being read,
corresponding to what Metz calls an inductive method, starting
from the syntagmatic types observed in the film (the terminal
nodes). . . . Metz goes on to show another way of reading the tree,
corresponding to what he calls the deductive method, starting from
the root of the tree and progressing towards the terminal nodes.14
Later in the essay Colin formalizes in terms of selectional features the information obtained from both types of reading. But
before he does this, he attempts to resolve some of the anomalies
and asymmetries in the GS. For Colin, the dominant asymmetry of
the GS is apparent from a deductive reading, which shows that,
whereas most of the syntagmatic classes are re-written as a syntagmatic class and a syntagmatic type, the class of achronological
syntagmas is re-written as two syntagmatic types (the parallel and
bracket syntagmas), with no syntagmatic category (i.e., no defining
characteristic) to distinguish them. (This information is manifest in
the complete representation of the GS reproduced in Figure 7 and
in Metzs table of the GS.15) Colins aim is to make the GS completely symmetrical.16
For Metz, the bracket and parallel syntagmas are similar in that
they present non-diegetic events, that is, shots of generic events
inserted into the spatio-temporal order of the story world to illustrate a concept rather than to advance the story. Metz bases the
distinction between the two syntagmas on the fact that the parallel
syntagma consists of alternating shots of contrasting events (shots
of the rich alternating with shots of the poor), whereas the bracket
syntagma consists of a series of shots illustrating one concept, with
the shots separated by optical devices. Metz gives the example of
the beginning of Godards Une femme mariee (1964), consisting of a
117
series of erotic images illustrating the concept of modern love.17

Furthermore, for Metz, the generic characteristic of the parallel and
bracket syntagmas is manifest in the fact that there are no spatiotemporal relations between the events depicted in them (both are
therefore achronological).
But Colin argues that the surface structure characteristic of alternating images is insufficient to distinguish the two syntagmas and
observes that if the optical devices are removed from the bracket
syntagma, it can be read as a descriptive syntagma (the events can
be read as diegetic and spatially co-existent). This leads Colin to
emphasize the strong relation between the bracket and descriptive
syntagmas, in which the only essential difference is the presence of
optical effects in the bracket syntagma to deny a reading of spatial
co-existence between the events depicted in each shot. Colin concludes that the bracket syntagma must be diegetic:
It may be seen that the possibility of confusion between the bracket and
descriptive syntagmas implies that, unlike the parallel syntagma, the
bracket syntagma is diegetic. The problem raised by Metzs table concerning the chronological/non-chronological distinction, which prevented any
representation of the feature differentiating the bracket syntagma from the
parallel syntagma, has thus been solved. The tree representation has become homogeneous: Each node is the starting point of two strings, one
leading to a lower node, the other to a terminal node.18
Colin then produces a new, completely symmetrical tree diagram

representing the core of the GS (see Figure 8).
This re-reading of Metzs GS permits Colin to represent the
syntagmatic types in terms of selectional features, where each type
is re-written as a set of specific traits:19
Parallel syntagma diegetic, linear
Bracket syntagma diegetic, specific
Descriptive syntagma diegetic, specific, narrative,
linear
Alternate syntagma diegetic, specific, narrative,
linear
Scene diegetic, specific, narrative,
linear, inclusive
Sequence diegetic, specific, narrative,
linear, inclusive20
118
Syntagma
diegetic
specific
1parallel
2bracket
Fig. 8
narrative
linear
etc.
3descriptive
4alternate
Colin also proposes the following selectional features for the class
Syntagma:21
(1)
Syntagma syntagma, diegetic
(2) syntagma linear
(3)
diegetic specific, narrative, inclusive
Colins use of these selectional features allows him to re-read
the GS in a completely different way to other commentators. Early
on he points out, It must be said, first of all, that our purpose here
is not, as it so often is with our discussions of the GS, to know
whether or not Metzs list of syntagmatic types is a comprehensive
119
one, whether it should be shortened or expanded. Our purpose is

to determine what properties can help to define syntagmatic types,
and what relationships exist between them.22 As with all generative models, Colins model does not merely describe existing syntagmatic types, but is also able to generate these existing syntagmatic types, together with possible/potential types and impossible
types (a tripartite distinction familiar to linguistics, as we shall see
ahead). This is probably one of the most innovative results of a
generative re-reading of Metzs GS, and Colin begins to put this
novel reading to work:
With the help of such rules, one can easily see that it is possible to deduce
many more types than are represented in the table, for instance: a syntagma
syntagma, diegetic, specific, linear, corresponds neither to the
parallel syntagma (it is diegetic), nor to the alternate syntagma (it is not
specific), although it has the same property as those two types (linear).
This, therefore, solves the problem of knowing how comprehensive the GS
is, since it predicts the possibility of as yet undetected types. This, however,
does not mean that everything is possible. The rules assume that certain
types are impossible, for instance syntagma, linear [i.e., a non-linear
sequence shot!].23
In the remainder of this chapter, I want to extend Colins discussion of actual and non-actual syntagmatic types by using Chomskys distinction between grammatical and acceptable sentences.
More specifically, I would like to use this distinction to investigate
grammatically deviant film syntagmas that spectators nonetheless
judge to be acceptable or innovative, rather than simply confusing.
Generative Theories of Film Grammar

In 1974, at the First Congress of the International Association for
Semiotic Studies (in Milan), Metz stated:
It is necessary to note that the work of film semiotics, in the vast majority
of cases, is inspired more or less directly by models of structural or distributional linguistics, and not by generative linguistics. This is not by chance.
. . . Each film is an original product [un produit germine], in the sense in
which one does not find on the first approach grammatical nor ungrammatical sequences, but only attested combinations. In the midst of these
filmic orderings that already exist, the first activity which attracts the analyst is that of calling attention to the regularities, co-occurrences, oppositions, recurrent motifs, progressions, etc., whose final synthesis constitutes
120
a structural analysis. The generative semiotic perspective comes across an

obstacle: the difficulty of defining in film a criterion of grammaticality.24
I want to argue that Metzs pessimism regarding the application of

TGG to film was premature, because it is possible to define a criterion for filmic grammaticality.
In Le cinema comme langage, Dominique Chateau has developed
a transformational generative theory of film. At first, he seems
skeptical about positing a theory of filmic competence in Chomskys sense of the word. Chateau notes that, for the most part,
grammars developed by contemporary linguists are limited to the
study of standard language, which represents the competence of a
native speaker in an idealized situation of ordinary communication.
But for Chateau, there are strong reasons against positing a standard cinematic language:
The most serious among them is incontestably . . . our incapacity to theoretically interpret the notion of a series of ungrammatical filmic images. A first
advance by [Sol] Worth, then repeated by Metz, it is each time approached
in relation to a discussion of the usefulness of the Chomskyan dichotomy
competence/performance for the understanding of cinematic language.25
And
The deprivation of a standard cinematic language, due to the inadequacy
in film of supposed ordinary communication, signifies that it is in vain to
directly research a standard grammar for film, especially since the absence
of a criterion of grammaticality does not allow . . . us to distinguish a subset of well-formed filmic segments.26
For Metz and Worth, because of the absence in film of a criterion of

grammaticality, and therefore the absence of a standard cinematic
language, it is implausible to posit a theory of filmic competence (a
representation of the tacit knowledge directors/spectators possess
that enables them to construct and comprehend filmic messages).
On the other hand, Chateau believes that a study of cinematic
language can nonetheless result in a plausible theory of filmic competence. This immediately leads to an apparent paradox in Chateaus position but one he is well aware of. Stated bluntly, the
paradox is, How can one reconcile the existence of filmic competence without positing the existence of a standard cinematic language and grammar? Chateau attempts to overcome this paradox
by arguing that whereas natural language presupposes a number
121
of strong principles regulating communication (principles that constitute the interlocutors competence), in film the contract is based
on much freer principles, to the point where the spectator will
accept, as part of her tacit knowledge, new principles that are
contradictory with those she already possesses. In other words, the
principles that regulate cinematic language are variable, contingent,
and continually open to revision. This is why throughout Le cinema
comme langage Chateau calls them diverse rules of the game.
Under these conditions, it is impossible, Chateau argues, to establish with any degree of certainty a standard cinematic language
based on a stable grammar. But the diverse rules of the game
still constitute the tacit knowledge directors/spectators need to possess to construct and understand films.
Chateau then notes that Metzs GS is prescriptive, since it refers
to only one stylistic variation of cinematic language classical
narrative cinema. Chateau characterizes this type of cinema as one
overdetermined by the super-structural rule of narrative logic, and
the GS as prescriptive because it represents this type of filmmaking
as the language of film. In Language and Cinema, Metz attempts to
counteract this imbalance by suggesting that the GS is merely a
sub-code of the montage code. Without wishing to return to Metzs
conflation, in Problems of Denotation in the Fiction Film, of classical narrative film with cinematic language, I would still want to
privilege the eight syntagmatic types identified by the GS as representing standard cinematic language (and its grammar) because of
the reified status of these eight syntagmatic types. Furthermore,
Chateaus positing of a lack of a standard cinematic language leads
him to argue that it is impossible to identify semi-grammatical film
sequences.27 But my privileging of the GS as this standard will
enable me to define semi-grammatical film sequences in cognitive
terms. To do this, we need to return to Chomsky.
For Chomsky, a descriptively adequate grammar must be able
to define degrees of grammaticalness, a phrase he defines as follows: The degree of grammaticalness is a measure of the remoteness of an utterance from the generated set of perfectly well-formed
sentences.28 Jerrold J. Katz has expanded Chomskys study of
degrees of grammaticalness in his essay Semi-Sentences29 (his
term for semi-grammatical sentences). He raises the question of
whether TGG (the primary aim of which is to generate all and only
the sentences that native speakers judge to be well formed, or
122
grammatical) can only explicate the speakers knowledge of the

structure of fully formed sentences, or whether it can also explicate
the speakers knowledge of semi-sentences. Within TGG, the
speakers understanding of a particular sentence is based upon her
possessing the grammar that generated that sentence. By definition,
this grammar only produces and comprehends (allows the speaker
to comprehend) fully grammatical sentences. But clearly, Katz argues, speakers can understand semi-sentences as well. Katz then
attempts to explain how speakers can comprehend semi-sentences.
A semi-sentence contains elements of grammaticality and ungrammaticality (a sentence that is fully ungrammatical or incomprehensible is called by Katz a nonsense string). The main question,
for Katz, is what formal condition determines when a string has
sufficient structure to be comprehensible30 (a formal condition that
enables the analyst to distinguish semi-sentences from nonsense
strings). His answer involves the speakers applying rules of association between fully grammatical sentences and semi-sentences:
A speaker knows (in the sense in which he knows the rules of the
grammar of his language) a system of rules that enables him to
associate a non-null set of grammatical sentences with each semisentence. This association is performed on the basis of the structure
that the semi-sentence has. And the speakers understanding of the
semi-sentence is nothing other than his understanding of the sentences in the set with which the semi-sentence is associated.31 For
Katz, therefore, the semi-sentence, to be comprehended, must possess sufficient grammatical structure that it can be associated with
a fully grammatical sentence.
In Aspects of the Theory of Syntax, Chomsky stresses that wellformedness/grammaticality does not wholly determine acceptability: Acceptability is a concept that belongs to the study of
performance, whereas grammaticalness belongs to the study of
competence. . . . Grammaticalness is only one of many factors that
interact to determine acceptability.32 Constraints on acceptability
include memory limitation and intonational factors. Keeping in
mind the interdependence of grammar and acceptability, we can
employ the well-known tripartite distinction between actual, possible/potential, and impossible sequences to distinguish between linguistic and filmic constructions that are grammatical and acceptable, ungrammatical but acceptable, and ungrammatical and
unacceptable (see Figure 9).
1. Actual
2. Possible/
potential
3. Impossible
123
Phonology
Grammar
Narrative fiction film
pet
lon
a year ago
a grief ago
atp
a the ago
the 8 syntagmatic types

Metzs example from
Godard (see text)
non-linear sequence shot
Category 1 refers to grammatical and acceptable sequences, 2 to

ungrammatical but acceptable, and 3 to ungrammatical and unacceptable sequences. A linguistic construction can also be grammatical and unacceptable, insofar as it is produced by the generative
and transformational rules of the grammar but it overstretches certain faculties of performance such as memory. This applies to very
long sentences (produced by the recursive application of a particular transformational rule, for example) or to sentences containing
multiply embedded structures. An open question is whether a
filmic sequence can be grammatical and unacceptable.
In Figure 9, category 1, of course, is from everyday discourse
(all the grammatical examples are taken from Chomsky). The
eight actual film syntagmas outlined in the GS define grammatical and acceptable film sequences, which constitute the hypothetical norm of narrative film sequencing, as embodied in classical Hollywood films. Here I am using the term grammatical in
Chomskys technical sense (discussed previously), which allows
us to describe the eight syntagmatic categories as grammatical in
that they denote sequences generated by the filmic re-write rules
and are therefore recognized by competent film spectators as being well formed.
Category 2 defines semi-sentences, together with those sequences that are possible according to the rules of grammar (i.e.,
like those in category 1 they are regular sequences) but that, for
accidental, historical, and/or psychological reasons, have not been
employed in everyday discourse. (Chomskys syntactic example is
from Dylan Thomass poetry, although examples from Lewis Carrolls nonsense verse or from Joyces Finnegans Wake, for example,
also belong to this category.) Metz makes a passing reference to
possible/potential constructions in the cinema in his essay Modern Cinema and Narrativity,33 in which he rejects the common
assumption that the modern (i.e., the French New Wave) cinema
Fig. 9
124
rejected narrative. Instead, he argues that it adopts a new type of

narrativity:
Far from demonstrating the nonexistence of syntax, [the French New
Wave directors] are really discovering new syntactic regions while remaining (at least as long as they are intelligible, as is the case almost always)
entirely submissive to the functional requirements of filmic discourse. Alphaville and Last Year in Marienbad are still, from one end to the other,
diegetic films, and they were still conceived in relation to the requirements
of narrative fiction, despite their undoubted originality, their editing, and
their montage.34
Except for the comment in parentheses, Metz is referring to the

formal dimension of French New Wave films. But immediately he
adds a cognitive dimension to his argument (or extends the preceding cognitive statement in parentheses):
Impossible constructions do exist in the cinema. . . . But such orderings have
never been seriously tried by film-makers, unless perhaps and even then
one would have to examine the matter more closely by some extreme
avant-gardist who had deliberately abandoned the effort to make himself
understood (and, then, usually in cinematographic genres initially foreign to the narrative fiction film). And the reason that the other film-makers
never attempt to construct such combinations, or even to imagine that they
might exist, is precisely because the main figures of cinematographic intelligibility inhabit their minds [emphasis added] to a much greater extent than
they are aware of.35
Clearly, in the terminology I established in Figure 9, Metz means

possible or potential rather than impossible constructions (and
later in the essay, he uses the term potential). The main point of
this passage, however, is Metzs insistence that the French New
Wave directors challenged the norms that constitute conventional
cinematic intelligibility, epitomized in classical Hollywood cinema.
This is not surprising from directors brought up on Hollywood
films who nonetheless attempted to produce their own distinctive
style.
As an example of a possible/potential construction, Metz cites
the sequence from Godards Pierrot le fou in which the two protagonists, Pierrot and Marianne, hurriedly leave Mariannes apartment. This sequence juxtaposes shots of the protagonists escaping
from the apartment with shots of the protagonists several minutes
into their journey. After rejecting a classification of this sequence in
terms of his eight syntagmatic types, Metz concludes:
125
In the midst of the frenzy of the hasty departure ([signified] of the denotation), it presents as equal possibilities which implies a sort of selfconfession to narrativity, an awareness of its own fablic nature several
slightly different variations of a frantic escape, sufficiently similar to each
other nevertheless for the event that really did occur (and which we will
never know) to take place among a class of quite clearly outlined occurrences. . . . [Godard] is able to suggest with a great deal of truth, but without determining the outcome, several possibilities at the same time. So he
gives us a sort of potential sequence an undetermined sequence that
represents a new type of syntagma, a novel form of the logic of montage,
but that remains entirely a figure of narrativity.36
I shall briefly discuss category 3 before returning to this example. Category 3 of Figure 9 illustrates irregular sequences that violate phonemic, syntactic, and filmic rules of sequencing. These sequences are called impossible because the phonemic sequence
cannot be pronounced, the syntactic sequence cannot be processed
by the speaker (i.e., given a semantic interpretation), and the filmic
sequence cannot be produced (as Colin argued).
Category 2 is the most interesting in filmic terms (and, indeed,
in linguistic terms), for it expresses a complex relation between
structure, cognition, and aesthetics. What makes such grammatically deviant sequences acceptable is that their remoteness from
well-formed sequences can be measured, or calculated by the
speaker. Chomsky writes: Given a grammatically deviant utterance, we attempt to impose an interpretation on it, exploiting whatever features of grammatical structure it preserves and whatever
analogies we can construct with perfectly well-formed utterances.37 In filmic and linguistic terms, category 1 requires the
standard (or optimally relevant) amount of processing to construct a semantic comprehension, whereas category 2 requires an
additional amount of processing effort to yield a semantic interpretation (usually accompanied by an aesthetic payoff). The examples
in category 3 will not yield to any semantic comprehension because
they do not preserve any grammatical features.
In Pierrot le fou, the characters escape is comprehensible because
it is possible for spectators to relate, using what Katz calls the rules
of association, this semi-sequence to a fully grammatical sequence.
This challenges Metzs suggestion that the spectator will never
know the actual event that did occur.
This process of reading a potential sequence of film within the
context of actual filmic sequences has been partially formalized by
126
Chateau. In a discussion of Metzs analysis of the sequence in Pierrot le fou, Chateau argues that the director dismantles a well-formed
sequence and then rearranges the shots into a new order, to express
the characters sense of panic and haste. Similarly, the spectator
must attempt to reconstruct the linear order that has been disrupted, in order to grasp the meaning of the sequence. Chateau
seems to argue that this particular sequence from Pierrot le fou is an
ordinary sequence (in Metzs sense of the term) that has undergone
a transformation. Although Chateau does not pursue this issue any
further, it is possible to explain this transformation by referring to
Chomskys trace theory. And the best way to explain the trace
theory is by means of the derivational history of the following
sentence:
(1a) John was expected [John to hurt himself]
(1b) John was expected [t to hurt himself]
(1c) John was expected to hurt himself
(1a) is a deep structure sentence made up of two clauses. Both

clauses contain the same subject (John). For this deep structure to
yield a grammatical surface structure sentence, the subject of the
embedded clause (the embedded clause is in parentheses) must be
deleted, yielding sentence (1c). Although the second subject has
been deleted, it still leaves a trace in the deep structure of the
sentence, sentence (1b). But why must it leave a trace? For Chomsky, the trace represents an important piece of information in the
competent speakers mental representation of the surface structure
sentence: The trace t, though not pronounced, is actually present
in the mental representation of the sentence. It is seen by the mind
as the mind computes the structure of the sentence . . . but it is not
pronounced by the vocal mechanisms because it contains no phonetic features.38 The trace is an empty category, but it functions in
the same way as the proper noun it has replaced (the trace possesses the same syntactic and semantic properties as the category it
has replaced).
What is important for my argument here is that in the sentence
John was expected to hurt himself, John is understood to be the
subject of the verb hurt as well as the subject of the verb expect.
The deep structure trace (which has no surface structure representation) represents the competent speakers underlying knowledge
that John is the subject of the verb hurt.
127
The sequence from Pierrot le fou is an ungrammatical but acceptable filmic sequence. It is acceptable because it can be associated
with a grammatical sequence. Or, in the terms just used, it can be
understood in relation to a grammatical sequence.
I shall attempt to reconstruct this sequence by putting its shots
(back) into a grammatical sequence. The sequence contains fourteen
shots, each of which corresponds to fourteen events. The following
description outlines the fourteen shots as they present the order of
events in the film.
Shot
No.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Events
Marianne (Anna Karina) and Pierrot (Jean Paul Belmondo) in Mariannes apartment. Frank (Mariannes
lover) enters and is ambushed by Marianne and Pierrot. Marianne prepares to leave the apartment, and
Pierrot begins to drag Frank away.
Marianne driving a red car. Pierrot gets in.
Marianne and Pierrot exit the bathroom (Frank, unconscious, has been put in the bath).
The red car is seen speeding along a street. It drives
under a height restriction barrier.
Marianne and Pierrot on the apartment roof. They look
down. Cut to:
Two men running toward the apartment.
Cut back to Marianne and Pierrot on the roof (continuation of 5).
Cut to Marianne and Pierrot climbing down a drain
pipe.
Repetition of shot 2 (slightly truncated).
The red car driving along a street. The height restriction barrier can be seen ahead in the distance.
Marianne gets into the red car and drives off.
The car approaches the height restriction barrier.
Point of view shot from the car wind screen of a replica
of the Statue of Liberty. The height restriction barrier is
briefly seen in the shot.
The red car pulls up at a petrol station.
Each action can be assigned a letter to illustrate the disjunction

between shot number and event order:
128
Actual sequence
Shot number: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Order of events: a h b k c d e f h i g
l m
The shots can be re-arranged to conform to an ordinary sequence

(can be associated with an ordinary sequence):
Re-arranged order of sequence
1 3 5 6 7 8 11 2/9 10 12 4 13 14
a b c d e f g
i j k l m
In terms of the events depicted, we can divide the sequence into

two: events ag (escape from the apartment) and events hm (both
protagonists driving away). In the actual sequence, three events (g,
h, and k) have been displaced and re-inserted into the sequence.
But this only makes five displaced events (rather than six) because
event h (Pierrot getting into the car) has been duplicated before
being displaced. In terms of the sequences linear narrative order,
the progression from a to b (the ambush of Frank and placing him
in the bath) has been interrupted by h (Pierrot getting into the car),
and the progression from b to c (Marianne and Pierrot exiting the
bathroom and going onto the roof of the apartment) has been interrupted by k (the car driving under the height restriction barrier).
The progression from c to f (Pierrot and Marianne on the roof of
the apartment, seeing the men below, and climbing down a drain
pipe) remains intact. Event g (Marianne getting into the car) has
been displaced two shots forward. Events h and i (Pierrot getting
into the car and driving along the street) remain intact. The progression from i to j (the car approaching the height restriction
barrier) is interrupted by g (Marianne getting into the car). The car
going under the barrier (k) has been displaced into the first half of
the sequence, between b and c (Marianne and Pierrot exiting the
bathroom and going onto the roof of the apartment).
What this analysis reveals is that two events from the second
half of the sequence (h, Pierrot getting into the car, and k, the car
going under the height restriction barrier) have been displaced into
the first half of the sequence, whereas one event in the first half (g,
Marianne getting into the car) has been displaced into the second
half of the sequence.
129
It is possible to assign letters to the events depicted in each shot

and reconstruct the sequence because the original position of the
events and shots is (in Chomskys terms) marked by a trace. A trace
contains the same syntactic and semantic features as the shots/
event it replaces. In a shot, as Metzs GS emphasizes, these features
include spatio-temporal coordinates, together with the cause-effect
narrative logic linking the events. In terms of h (the repeated shot
of Pierrot getting into the car), it is possible to argue that its original
position is shot 9, but that it has been displaced to shot 2. Moreover,
it has not left a trace, but has been duplicated before being displaced. Hence, whereas the other displacements cause two disruptions (their old location and their new location), this duplication
creates only one disturbance (the new location).
This process of reconstruction shows that the sequence can be
understood by the spectator: The spectator understands that shot
2, for example (Pierrot getting into the car), jumps ahead to show
future events (in shot 1 Pierrot is shown dragging the unconscious
Frank, whereas in shot 2 Pierrot is getting into a car; such a jump
depicts spatio-temporal as well as narrative displacement). Such an
understanding would not be possible if shot 2 depicted a deep sea
diver searching for buried treasure, for example. I am therefore
arguing that the re-constructed ordinary sequence is present in the
spectators mental representation of the sequence present not in
its pure form, of course, but in the sequences actual form together
with traces. And it is from these traces that the ordinary sequence
can be re-constructed. When seen at the cinema, the grammar of
this sequence conveys the confusion of the protagonists escape. But
the grammar does not arrange the shots haphazardly (they do not
result in a nonsense string of shots). Instead, it still has a structure,
which can be associated with a fully grammatical structure (as
represented in the GS).
Chateaus claim that this sequence from Pierrot le fou represents
the re-writing of a grammatical filmic sequence has two effects: (1)
It challenges Metzs reading of the sequence, in which he states that
it represents several slightly different variations of a frantic escape,
sufficiently similar to each other nevertheless for the event that
really did occur (and which we will never know).39 Contrary to
Metzs reading, the sequence presents only one variation of the
escape, which, when re-constructed, can be known by the spectators. (2) Secondly, Chateaus re-reading of this sequence reinforces
130
my argument that the GS can be taken to represent a standard

grammar of film, one that generates well-formed sequences, and in
relation to which deviant (or semi-) syntagmas can be comprehended. We saw that for Chomsky and Jerrold J. Katz semisentences are comprehensible according to their association with
fully grammatical sentences (the remoteness of semi-sentences from
well-formed sentences can be measured, or calculated by the
speaker). In analogous fashion, I am claiming that semi-syntagmas
(or deviant syntagmas) are those that can be associated with one or
more of the eight syntagmatic types of the GS. If no association is
possible, then the collection of shots must be categorized as a nonsense string. The fully grammatical syntagmas of the GS are defined
and distinguished according to their spatio-temporal coordinates
(which represent the films diegesis in a coherent, or comprehensible, manner). Deviation is calculable in terms of an ungrammatical
syntagmas difference from the spatio-temporal relations represented in the fully grammatical syntagmas of the GS. As long as
that difference is measurable (as it is in the sequence from Pierrot le
fou), then the syntagma can still be comprehended.
But we can go further than Chomskys and Katzs discussions
of the degrees of grammaticalness. The GS constitutes well-formed,
or grammatical, film sequences, not so much because of any formal
criteria they may conform to, but because its eight syntagmas constitute the hypothetical norm of narrative film processing, or comprehension. Ill-formed (ungrammatical) syntagmas require from
the spectator an additional amount of processing effort. Although
Metz defines the syntagmas in formal terms, it seems to me that
they conform to Sperber and Wilsons cognitive principle of relevance.
The spectators processing of narrative film is governed by the
goal of constructing a coherent literal meaning or a coherent
narrative world (diegesis). As David Bordwell argues, the films
fabula (a concept that parallels the concept of the diegesis) is inherently fragmentary and must be completed by the cognitive activity
of the spectator. In Metzs film semiotics, the eight categories of the
GS are constructed according to the principle of relevance they
are the optimal shot orderings, for they permit the spectator to
employ the least amount of processing effort while yielding an
optimal contextual effect ( the spectators construction of a diegesis). Ill-formed syntagmas thwart the principle of relevance, be-
131
cause they require additional processing effort form the spectator

in order to construct/comprehend a coherent narrative world.
However, if this additional processing effort is successful, it results
in an aesthetic pay-off (which explains the select audience for the
international art cinema and its corresponding characterization as
aesthetic). But if this additional processing effort is unsuccessful, it
may simply lead to boredom and confusion (hence the international
art cinema is not inherently aesthetic; its aesthetic effects are partly
achieved from the mental effort or, more formally, competence
of the spectator).
We can now apply the principle of relevance to the three categories outlined. In category 1, actual sequences, the principle of
relevance operates automatically. In category 2, potential sequences, it operates with conscious effort (and this additional effort
is usually rewarded by an aesthetic pay-off). But in category 3,
impossible sequences, the principle of relevance does not operate,
leading to incomprehension.
What exactly are the increased processing effort and aesthetic
pay-off in category 2 sequences? Firstly, the spectator must work
harder to construct a coherent diegesis must work harder simply
to determine what is happening in the film. We have already seen
this in relation to Pierrot le fou. A further example is given by David
Bordwell in Narration in the Fiction Film, in an analysis of Alain
Resnaisa film La Guerre est finie (1968).40 Bordwell concentrates on
the first nineteen shots of the film. It begins with Diego (a left-wing
political agitator) driving back across the Spanish-French border
with his collaborator Jude. A number of the shots are overlaid with
a non-diegetic voice-over expressing Diegos anxieties about
crossing the border. This is followed by Jude speaking off-screen.
As he does so, the spectator sees nine shots depicting Diego performing a number of actions (catching a cab, meeting his friend
Juan, missing his friend Juan, waiting for a cab, walking down a
train corridor, catching a train, missing a train, and being greeted
by Juan). Bordwell writes that the mode of presentation [of these
nine shots] must give the spectator pause. Seeing that shots 917
rupture the spatiotemporal continuum of the scene in the car and
hearing Judes chatter continuing in voice-over, the viewer versed
in the convention of the art cinema hypothesizes a temporal disjunction between sound (the present) and image (not the present).
But the ambiguities of the sequence thwart any easy comprehen-
132
sion.41 Translated into Sperber and Wilsons terms, what Bordwell

means here, in effect, is that this sequence of nine shots cannot be
automatically processed (i.e., processed in accordance with the
principle of relevance). As with the sequence from Pierrot le fou, this
sequence cannot be categorized in terms of any of the eight syntagmatic types represented in Metzs GS. The nine shots do not,
therefore, represent a fully grammatical sequence.
Bordwell then goes on to explore whether the sequence can be
comprehended at all. He notes that the shots cannot be comprehended as presenting a single event chain; rather, they represent
mutually exclusive alternatives Diego catches/does not catch a
cab quickly; Diego calls on Juan/misses Juan/Diego calls and Juan
arrives later; Diego catches/does not catch the train (it is this structure of mutually exclusive alternative events that Metz wrongly
attributed to Pierrot le fou). Bordwell then ponders whether the
shots represent past actions, frequentative actions (depicting Diegos regular routine), or future actions. Although the least likely
the last hypothesis, Bordwell argues, is the only one that can sufficiently explain/motivate these shots: Only one construction accounts for everything in these shots. It is an unlikely one, but it is
the one that later passages will confirm. These shots may be taken
to represent various possible future events. Put in simple fashion:
I might grab a cab right away or have to wait in line, I might miss
Juan, what if I cant catch a train? La Guerre est finie will explore
character subjectivity according to one principle of the art film; we
justify what we see and hear by reference to psychological motivation.42 That the shots represent future events is not an inference
automatically generated by the spectator when watching the film.
The spectator must work harder than normal to find the optimally
relevant inference the one that will motivate these nine shots. Like
all non-demonstrative inferences, the one claiming that the shots
are of future events is open to revision. However, as Bordwell
notes, this inference is strengthened by later sequences.
The spectators additional processing effort involved in comprehending semi-syntagmas leads not only to the construction of a
coherent diegesis, but also to the construction of an auteur who has
manipulated the shots for aesthetic effect. (In Roger Odins terms,
semi-syntagmas are prominent in the artistic mode of filmmaking,
since the film is read as the product of an auteur.) For Bordwell, on
the one hand, art cinema narration is based on objective and ex-
133
pressive realism. On the other hand, it is highly self-conscious, as a

result of the flaunting of narrational procedures. Bordwell writes
that when these flauntings are repeated systematically, conventions ask us to unify them as proceeding from an author.43 Although Bordwell rejects the communicational model of film particularly its emphasis on the sender of a message (see Chapter 2)
he argues that the positing of an author is necessary in Resnaiss
film: The average spectator of La Guerre est finie (1968) is likely to
approach the film with some expectations about the principal narrational manipulations the film will offer and to attribute those to
an authorial intelligence.44
In discussing Pierrot le fou and La Guerre est finie, I have begun
to define the boundaries between grammatical (and acceptable),
ungrammatical (but acceptable), and ungrammatical (and unacceptable) film sequences, a definition not based purely on formal criteria but also on cognitive criteria (how spectators comprehend grammatical and deviant syntagmas). One conclusion we can reach is
that there is a marked difference in processing effort between classical narrative films (whose syntagmas conform to the hypothetical
norm outlined in Metzs GS) and films identified as belonging to
the avant-garde or to international art cinema (whose syntagmas
deviate from those in the GS). Correlatively, classical films conform
to the principle of relevance (the hypothetical norm of processing
activity), whereas films of the international art cinema deviate from
the principle of relevance, since they require additional processing
effort, which is usually rewarded with an aesthetic pay-off.
However, we should not conclude that spectators automatically
abide by the principle of relevance in comprehending films. This
principle simply expresses a rational norm, from which actual
groups of spectators may deliberately prefer to deviate. The increase in processing effort may not simply be demanded by the
filmic text, but may be instigated by the spectator. This is precisely
the conclusion that Lerdahl and Jackendoff reach in their generative
theory of tonal music. They find that the specificity of music cannot
simply be described in terms of grammatical rules, but also requires
a series of what they call preference rules:
We have found that a generative music theory, unlike a generative linguistic theory, must not only assign structural descriptions to a piece, but must
also differentiate them along a scale of coherence, weighting them as more
or less preferred interpretations (that is, claimimg that the experienced
134
listener is more likely to attribute some structures to the music than others).
Thus the rules of the theory are divided into two distinct types: wellformedness [i.e., grammatical] rules, which specify the possible structural
descriptions, and preference rules, which designate out of possible structural
descriptions those that correspond to experienced listeners hearings of any
particular piece.45
Preference rules therefore represent a series of alternative, but plausible, structural descriptions of a piece of text. Whereas sentences
of natural language (except ambiguous sentences) are subject to one
unique structural description, music, Lerdahl and Jackendoff argue,
does not demand a singular grammatical description, but a series
of preference rules.
Such preference rules very likely operate in the comprehension
of films as well because, as Dominique Chateau makes clear, there
is no highly formalized grammar of the cinema to demand a singular grammatical description. This means that, under specific circumstances, the comprehension of a film may be governed by preference rules, of which the principle of relevance is only one
(although the dominant one). This can then explain how cinephiles
can identify authorial structures in classical Hollywood films
they prefer to select different cues and experience a different structure than non-cinephiles. To read a Hollywood film for its authorial
structures, rather than simply for its literal meaning, is a nonrelevance-determined strategy preferred by cinephiles. We can
draw an analogy with the choice readers face when confronted with
an ironic or allegorical text. Some readers may interpret the text
literally, whereas others are able to comprehend it figuratively.
With their theory of preference rules, Lerdahl and Jackendoff restate the principle of relevance as follows: Prefer to assume that
the speaker is conveying something relevant.46 As a preference
rule, the principle of relevance should not be taken to represent the
necessary and sufficient conditions for all film spectators in comprehending a film, but to represent one preferred way among many
although admittedly it governs the most basic and literal level of
comprehension. Unlike the idealized normal spectator, cinephiles
prefer to go beyond the principle of relevance to concentrate on
what the filmmaker is not conveying directly.
We can therefore posit that film has a special (or rather loose)
type of grammar: For Chateau, for example, the grammar of film is
based on diverse rules of the game; or we can say that it is no
135
more than a preference rule (reflecting the inherent ambiguity of

filmic structure, as with tonal music); or we can say that it is based
on the pragmatic (therefore defeasible) principle of relevance. For
Metz, the grammar of film is inseparable from rhetoric: The nature
of the semiotics of film is that grammar and rhetoric are not separate in it.47 Metz refers specifically to the classical rhetorical trope
of dispositio, which consists of prescribing determined orderings to
undetermined elements.48 Whereas shots are undetermined, their
ordering does pertain to a degree of determination, and the semiotic analysis of the grammar of film attempts to discover both its
nature and its degree of determination. The trace theory and preference rules have been introduced in this chapter to outline a number of cognitive determinations of film. These two cognitive rules,
together with the principle of relevance, complement the distinctions made in this chapter, among actual, possible (potential), and
impossible filmic sequences, and go some way to predicting why
those sequences identified by the GS are frequently employed in
narrative films, why semi-sequences are rarely employed, and why
others not employed at all. These predictions in turn confirm the
cognitive reality of filmic sequences as analyzed by the cognitive
film semioticians, for they suggest that frequently used sequences
are more easily comprehended than difficult sequences, and some
sequences are not employed at all because spectators cannot process them.
Film Semiotics and Cognitive Science

Chomsky favored grammars that captured . . . linguistically significant
generalizations [Chomsky, Aspects, 42] but the question that immediately
struck many psychologists was whether linguistically significant generalizations were also psychologically significant.49
Within the context of TGG, film semiotics is characterized as positing a relation of identity between its semiotic descriptions and
cognitively real structures, namely, cognitive states and processing
operations in the spectators mind. We need to follow the cognitive
interpretations of Colin to answer the following question: What are
the reasons for making cognitive claims about film semiotics? One
immediate answer suggests itself to increase the power of film
semiotics, that is, to make it descriptively adequate rather than
merely observationally adequate. Whether a descriptively adequate
136
theory of film semiotics is an improvement on Metzs observationally adequate theory is a question I shall begin to answer in the
following pages.
In the conclusion to The Grande Syntagmatique Revisited
Colin makes two contradictory assertions concerning the cognitive
reality of Metzs GS. Colin stresses throughout his essay that the
primary motivation for his re-reading of the GS is pedagogic how
can it be taught? It is in answer to this question that he at first
asserts the cognitive unreality of the GS: It thus seems quite reasonable to use segmentation as a starting point when teaching the
GS. This, however, does not imply that spectatorial competence
proceeds in the same way. It is even conceivable that an explicit
theory on this competence would not have to distinguish between
these two procedures.50 But on the next page, after briefly considering the function of sound in the analysis of film, Colin asserts:
In this sense, there is no reason to distinguish between image
track and sound track, since the GS can be considered as an
explicit analysis of syntagmatic relationships between the mental
images constructed by the spectator on the basis of the visual and
acoustic information provided by the film.51 Here, Colin confers
on the GS a cognitive reality, suggesting that it is a descriptively
adequate theory, a theory of the film spectators underlying competence.
Like the TG grammarians, Colin is arguing that a TGG-based
film semiotics does not merely offer a formal way of generating
and representing filmic structures, but also represents cognitively
real structures in the spectator (in other words, the formal filmic
structures explained by the semiotician are cognitive representations). Again, as with the TG grammarians, Colin is equating formal
rules of grammar with cognitive states.
In an essay already cited, Scott Soames (among many others)
argues that grammars (grammatical rules) and theories of competence are independent of one another. Whereas grammars characterize languages as formal systems of symbol manipulation, theories of competence attempt to uncover cognitive structures, which
cannot be characterized simply or primarily in terms of formal
systems of symbol manipulation, for linguistic cognitive structures
involve far more than a narrowly construed notion of competence.
Soames is following many researchers in artificial intelligence in
making a strong distinction between linguistics and cognitive psy-
137
chology, for he denies that grammar is cognitively real (i.e., he

denies it plays a role in language processing). He offers two
straightforward reasons: There are linguistic facts which are not
psychological in nature, and There are psychological facts which
are non-linguistic.52 Here Soames is rehearsing the modular theory
of the mind, which posits that the mind consists of separate information processing devices, each geared to the processing of particular types of information. By contrast, the non-modular view models the processing of all information in terms of one general
cognitive apparatus. Because of the richness and complexity of
natural language, it is regarded to constitute the general cognitive
apparatus.
But we do not need to adopt a fully modular theory of the mind
to avoid the translinguistics of semiotics and structuralism, or to
deny that grammar plays a role in the processing of language. A
number of grammarians have argued that the Standard Theory of
TGG was simply the wrong theory. Thomas G. Bever notes that
three trends developed during the 1980s in an attempt to renew the
relation between linguistics and psychology: (1) Chomskys Government and Binding theory; (2) Joan Bresnans lexical-functional
grammar; and (3) the Generalized Phrase Structure Grammar of
Gerald Gazdar.53 All three models are notable for having greatly
reduced (or abandoned completely) the transformational component of grammar.
In a review of Metzs Film Language, John M. Carroll takes a
different approach to Colin in assessing the cognitive status of film
semiotics. Carroll writes: The categories and binary features of the
grande syntagmatique are not aspects of consciousness in film
viewing. They are theoretical entities, aspects of a model, and constitute a break with experience. Of course, if the taxonomy has any
efficacy, it will ultimately be relevant to models of conscious experience.54 Here Carroll indicates the cognitive unreality of the GS
but also stresses that a formal model of film becomes a satisfying
theory only to the extent that it has cognitive implications. Later in
the review, he makes a tentative suggestion:
It would be odd indeed if any purely formal analysis of film turned out to
be psychologically significant independent of general facts about human
perception. If the film theory includes the facts about perception in its own
statements, we risk confusing generalizations about perception with generalizations about film structure. A more reasonable hypothesis is that a
138
formal analysis of film becomes psychologically significant through the

interaction with various behavioral systems (e.g. perception).55
In Toward a Structural Psychology of Cinema,56 Carroll attempts, less

successfully than Colin, to develop a TGG of film and to theorize
the interactions of film structure, perception, and aesthetics. He
argues that an aesthetic theory of filmic perception must be based
on a formal model and proposes that TGG can offer the most
explicit formal model of film, hence can serve as the most adequate
framework for an aesthetic theory of film. The most important
question that such an aesthetic theory must confront is, How do
grammar and aesthetic theory interact, how do they relate to each
other in the integrated theory?57 Carroll stresses that grammar
cannot by itself determine the aesthetic value of a filmic (or linguistic) sequence, for it is an aesthetically neutral theory of cognition.
He then feels justified to distinguish the filmic and the cinematic:
The term filmic will be used to refer to sequences which are
structurally correct and therefore well-formed . . . the term cinematic will hereafter be used to refer to cinema sequences that are
artistically and stylistically valuable.58 A filmic sequence according to Carroll is one that obeys the rules of filmmaking (obeys the
180 and 30 degree rules, constructs readable point of view sequences, maintains an unambiguous relation between the space of
each shot and the space of the sequence it forms part of, etc.),
whereas a cinematic film is structurally incorrect and ill formed.
Although Carroll notes that these terms do not correspond to aesthetic judgements, he nonetheless implies that filmic means routine and pedestrian, and cinematic means artistically valuable
(precisely the term he uses to define the cinematic).
In a programmatic essay, Carroll cites several examples of cinematic sequences (all based on confusing spatial relations). Here I
shall cite one of his examples:
In Hitchcocks Psycho while the detective climbs the stairs (just before he is
murdered), there is a cut to the base of a door just beginning to open (as
we later find out, the bedroom door). Where the door is, the viewer does
not know at the time of the cut.
. . . . It is very clear that Hitchcock, for example, wanted the geography
of the door to be unspecified this unclarity is the basis for the suspense
of the sequence. . . . This [unclarity] is quite irrelevant to a formal model of
film but crucial to an aesthetic. . . . The sequences we have just discussed
are unacceptable but they may all be very successful aesthetically.59
139
Moreover, in his book Carroll defines aesthetically pleasing in

terms of a theory of tension (in which the value of art is located in
the conflicts it creates). According to Carroll, tension is created in
film through a conflict between event/action boundaries and shot
boundaries. In his own TGG-based theory of film (outlined in the
central chapters of Toward a Structural Psychology of Cinema), Carroll
conceives the base component as a device that generates events,
each consisting of a series of actions (some of which can be optionally re-written as preparatory and focal sub-actions).60 Each action
is represented in the form of a proposition. Furthermore, the
boundary between two or more propositions (actions) represents a
major constituent boundary (the boundary between preparatory
and focal actions represents a minor constituent boundary). Transformational rules then transform these actions into filmic sequences. The most important rule involves representing the actions
in a sequence of shots (or placing cuts within the chain of actions
and sub-actions). The relation between action boundaries and shot
boundaries allows Carroll to develop a theory of motivated and
unmotivated cutting, as follows: If a cut coincides with a major
boundary, the cut is motivated by the event. If the cut coincides
with a minor boundary, the cut is motivated by an action in the
event. If a cut coincides with a nonboundary, it is not motivated.61
For Carroll, tension in the cinema occurs when shot boundaries do
not coincide with action/event boundaries. Carroll establishes five
different sequence structures for the same event structure that is,
devises five different ways of placing cuts within a sequence consisting of three actions.62 Zero tension is created when shot boundaries coincide with action boundaries, and maximum tension is
created when there is little or no coincidence.
Carroll develops a very simplistic, almost quantitative, theory
of aesthetics based on a formal model. The less coincidence between
action and shot boundaries, the more tension and so the more
aesthetic pleasure created by the sequence. The arithmetic is elegant
but hardly bears up to scrutiny. For example, one tension-creating
sequence consists of the three actions being filmed in one shot!
There is certainly no coincidence between action boundaries and
shot boundaries (at least within the sequence); according to Carroll,
therefore, it must create a great deal of tension. But this conclusion
fails to take into account the structure within the image that is,
camera movements, and so on (for a camera movement can be
140
functionally equivalent to a cut); nor does it consider the arbitrary

division of filmic events into action, sub-actions, and so on. Carroll
seems to uphold the classical film theorists distinction between
montage and the long take, in which the two devices are distinguished according to their materiality, not their function.
My extension of Colins work on actual and non-actual syntagmatic types (through the introduction of the work of Chomsky and
Katz on semi-sentences) makes a tentative link between grammatical structure, cognitive processing effort, and aesthetics, indicating
that structure does at least have a partial cognitive reality. However, unlike the early psycholinguists, I do not propose an isomorphic relation between structural complexity and processing complexity. The link is far more indirect and mediated, although I
would not go as far as Scott Soames (and others) and assert that
grammatical structure has no role to play in cognition.
In noting that there is a measurable difference in processing
effort between classical Hollywood films (as represented in the
GS) and the films of the French New Wave, I have shown that a
films structure does partly determine processing complexity. But
the descriptions of film structure offered by Metzs structural linguisticbased film semiotics or Colins cognitive film semiotics
based on the Standard Theory of TGG are insufficient in themselves
in explaining this processing complexity. This is why I have updated Colins Standard Theory of film grammar with concepts from
Chomskys trace theory, Sperber and Wilsons principle of relevance, and Jackendoff and Lerdahls preference rules. Film theory
can only benefit from exploring in more detail these and other
recent linguistic models.
Conclusion
The compatibility of Chomskys theory with semiotic views of symbolic function remains to be explored, but will probably find its
explanation when both can be integrated into the fabric of a more
comprehensive cognitive science. (Thomas Sebeok)1
The problem for us is not . . . to complete semiotics, but to transform it. (Michel Colin)2
y aim in this book has been to outline the film spectators cognitive capacity as theorized by the cognitive
film semioticians, whose work is united by the same project: to
combine film semiotics and cognitive science, with the objective of
modelling filmic competence that is, the spectators knowledge or
intuitions about filmic meaning. To offer an outline of this work I
have had to mediate between the insights of the Language Analysis
tradition and cognitive science (the twentieth century version of
epistemology). As we saw in Chapter 1, these two traditions are
usually opposed to one another, since the Language Analysis tradition replaces the epistemologists assumption that we have immediate access to our own thoughts with the assumption that we
only have indirect access to our thoughts via language and other
intersubjective sign systems. Chomskys linguistics creates a synthesis between epistemology and Language Analysis, thus avoiding
the idealism and first person perspective of epistemology and the
(quasi) behaviorism of the Language Analysis tradition. Chomskys
work on competence therefore epitomizes what this book is all
about.
More specifically, the cognitive film semioticians study filmic
competence by integrating with film semiotics insights from enunciation theory, pragmatics, as well as Chomskys transformational
generative grammar. To this extent, the cognitive film semioticians
141
142
are exploring the compatibility of Chomskys theory with semiotics

that Thomas Sebeok calls for. Enunciation theory, pragmatics, and
transformational grammar have given the work of the cognitive
film semioticians theoretical coherence explicit, rigorous, precise,
and clear aims and objectives. Throughout this book I have used
their work as the starting point to study specific and fundamental
issues concerning filmic competence, including the spectators intuitions about filmic space (the construction of a coherent space, the
status of off-screen space, intuitions about the frame and diegesis
as containers); intuitions about how the film addresses the spectator
(different modes of address and spatial orientation); the spectators
different modes of attention required to comprehend different
modes of film; and finally, intuitions concerning film sequencing
and the comprehension of deviant but acceptable filmic sequences.
But before suggesting that it is possible to develop a complete
rapprochement between semiotics and cognitivism, we need to address one final issue the cognitivists critique of Grand Theory.
Post-Theory?
The term theory is used in this book to mean speculative
thought, or thought that goes beyond phenomena to model the
non-perceptible reality underlying phenomena. Semiotics is the
quintessential speculative theory because, as Thomas Sebeok remarks, What a semiotic model depicts is not reality as such, but
nature as unveiled by our method of questioning.3 Among the
dominant aspects of nature that semiotics and transformational
grammar unveil are the scope and limits of human reasoning. The
disciplines and authorities that constitute this books foundations
do not, therefore, exclusively study small scale, medium-level issues, but continue to construct universal frameworks. This is why I
continue to use the term theory in the singular.
For David Bordwell and Noel Carroll,4 localized, middle-level
(or piecemeal) theories should replace film theory. Moreover, they
argue that theorizing should be conceived as an activity (accounting for their use of the present participle theorizing) that, finally,
should be problem-driven rather than doctrine-driven. There are
two immediate responses to this re-configuration of film theory,
one serious, the other less serious. First, I am reminded of the story
of the elephant and the six blind men. The one who felt the ele-
CONCLUSION
143
phants leg said that it was like a tree trunk; the second, who felt
its tail, said it was like a rope; the third, who touched the elephants
trunk, said it was like a hose; and so on. Piecemeal theorizing may
not be able to see the wood for the trees if it completely abandons
the tendency to develop a unifying theory. My second response
suggests that theory and piecemeal theorizing are not incompatible.
The initial stage of the theoretical activity involves the simplification (abstraction and idealization) of the domain under study.
For Robert de Beaugrande: On first inspection, the phenomenal
domain is composed of an unruly aggregate of objects and events
for which the theorists hope to discover categories, regularities, and
principles of order. This step can never be achieved without rarefying and abstracting the phenomena by judging at least some
concrete aspects to be irrelevant, accidental, and insignificant.5 For
example, structural linguists propose that only a single description
of the minimum units of linguistic meaning is necessary to
describe the inner logic of natural languages structure. This accounts for Saussures exclusive emphasis upon langue at the expense of parole, which he regarded as being too individualistic to
come under scientific scrutiny.
This activity of simplifying is the necessary, initial first step in
constructing a theory. The early stages of any theory are governed
by attempts to obtain the maximum amount of simplicity by studying only the essential determinants of a particular domain. But as
research progresses, those determinants deemed inessential and irrelevant at the early stages take on a greater importance and make
the research domain complex in the negative sense that the theory
cannot successfully subsume important factors within its framework. For research to progress, this negative complexity must be
translated into positive complexity in other words, the theory
must expand to take into account these additional factors.6
In attempting to establish itself as a science, structural linguistics
followed the simplicity principle and attempted to reduce, by as
much as possible, the negative complexity of its research domain,
by studying only those characteristics of natural language posited
to be linguistically specific. This move toward the study of linguistic specificity resulted in the organization of linguistic data into
binary oppositions, which in turn resulted in the development of a
segmental methodology to identify these binary oppositions. This
approach worked well with the identification of minimal units (dis-
144
tinctive features and phonemes) but reached its saturation point at

the level of morphemes.
This process of simplifying eventually and inevitably leads to a
crisis, as the inessential phenomena take on importance. Robert de
Beaugrande points out that there are two ways of thinking about
this crisis. First, the crisis can be understood in terms of Thomas
Kuhns theory of scientific revolutions.7 For Kuhn, normal science
involves the activity of problem solving. A crisis develops when
standard puzzles are exhausted and new ones cannot be solved
with the existing framework (or paradigm). A scientific revolution,
in Kuhns definition, involves the setting up of a new paradigm
with its own puzzles to be solved. The older paradigm is seen to be
incompatible with the newer one, and as a result the older one is
simply discarded. Second, de Beaugrande understands the crisis in
terms of the concepts of negative complexity and positive complexity. For him, A productive revolution . . . brings about a real major
rise in positive complexity, building upon the facts and observations made available by the preceding paradigm and inheriting its
achieved positive complexity. Unification revolutions provide the
most glowing example of this continuity.8
The cognitivists take the Kuhnian approach to the relation between different film theories by suggesting that they (the cognitivists) are starting again, rather than building on previous film theories. By contrast, the advantage of cognitive film semiotics is that
it involves a productive revolution by building upon the insights of
film semiotics (while discarding psychoanalytic film theory). Cognitive film semiotics inaugurates a productive revolution in relation
to modern film theory. It recognizes its tendency to oversimplify
and attempts to compensate for this oversimplification by adding
cognitive theory to its theoretical framework, leading it away from
observational adequacy toward descriptive adequacy (and eventually, explanatory adequacy). The cognitivists also recognize modern
film theorys tendency to oversimplify, but their reaction has been
to debunk its whole approach to theory by replacing it with piecemeal theorizing. I suggest that this is an overreaction, since piecemeal theorizing can be integrated into a general theoretical approach. To recognize and break down general theoretical problems
into smaller, piecemeal theoretical problems does not involve a
rejection of theory, as Bordwell and Carroll suggest, but represents
its next stage of development.
Notes
Chapter One. The Cognitive Turn in Film Theory

1.
2.
3.
4.
5.
6.
7.
Janet Bergstrom, American Feminism and French Film Theory, Iris,

10 (1990): 189.
Much of the work of these European authors remains untranslated.
However, a sample of translations can be found in The Film Spectator:
From Sign to Mind, ed. Warren Buckland (Amsterdam: Amsterdam
University Press, 1995).
I use the term modern rather than contemporary because this type
of film theory belongs to the discourse of modernity that emerged
from the Enlightenment period of European history.
The Institute for Cognitive Studies in Film and Video was set up by
Joseph Anderson in 1994 in the Department of Theatre and Film at the
University of Kansas.
It is important to note that David Bordwell writes distinguished
works in both film theory and film history. He does not, therefore, see
film history as a threat to film theory. Rather, he regards them to be
complementary approaches to the cinema (as, indeed, would most film
scholars). The threat to film theory comes mainly from the study of
new objects and from the adoption of the new methodologies enumerated by Janet Bergstrom in the opening quotation.
In my anthology The Film Spectator I called these European film theorists the new film semiologists. However, the term semiology is now
an archaic term, which only refers to the structural linguisticbased
work of the sixties. Throughout this book I shall use the label film
semiotics. Furthermore, the addition of the word cognitive indicates
the equal status these European film theorists accord to cognitive and
semiotic mechanisms in filmic comprehension.
Peter Lehman, Politics, Film Theory, and the Academy, Journal of
Film and Video, 40, 2 (1988): 49; 50.
Noel Carroll, Theorizing the Moving Image (Cambridge: Cambridge University Press, 1996): 319. See also Carrolls Prospects for Film Theory:
A Personal Assessment, Post-Theory: Reconstructing Film Studies, ed.
David Bordwell and Noel Carroll (Madison: The University of Wisconsin Press, 1996): 3768; esp. 5668.
145
146
NOTES TO PP. 412
8. David Bordwell, Narration in the Fiction Film (London: Methuen, 1985):

xii.
9. Ibid., 23.
10. Christian Metz, Film Language: A Semiotics of the Cinema, trans. Michael
Taylor (New York: Oxford University Press, 1974): 95.
11. Christian Metz, Langage et cinema (Paris: Larousse, 1971); Language
and Cinema, trans. Donna-Jean Umiker-Sebeok (The Hague: Mouton,
1974).
12. In the Preface to Narration in the Fiction Film, Bordwell writes, My
aim, stated broadly, is to set forth a poetics of narration (xiii). In the
Conclusion, he writes, None of what I have said is definitive, but this
account of narration may encourage the growth of a valuable realm of
knowledge: the historical poetics of cinema (336).
13. John Deely, Brooke Williams, and Felicia E. Kruse, Introduction to
Frontiers in Semiotics (Bloomington: Indiana University Press, 1986): xi.
14. Boris Glinskij, Boris Grjaznov, Boris Dynin, and Evgenij Nikitin,
quoted in P. N. Denisov, Principles of Constructing Linguistic Models
(The Hague: Mouton, 1973): 154.
15. Metz, Language and Cinema, Chapter 10.
16. Ferdinand de Saussure, Course in General Linguistics, trans. Roy Harris
(London: Duckworth 1983): 8.
17. Samuel Weber, Saussure and the Apparition of Language: The Critical Perspective, Modern Language Notes, 91 (1976): 916.
18. John Deely, Brooke Williams, and Felicia E. Kruse, Introduction to
Frontiers in Semiotics, xiv.
19. One fundamental difference between the underlying realities is that
natural languages system of codes (la langue) is of a different logical
type from films system of codes (cinematic language). La langue is of
a much higher level of organization than cinematic codes and indeed,
of all other semiotic codes.
20. Raymond Bellour made this point in his interview with Janet Bergstrom: Its obvious that a film can be very different depending on
whether its constructed more or less by scenes or sequences, or by
means of alternating syntagmas and episodic sequences. Alternation, Segmentation, Hypnosis: Interview with Raymond Bellour,
Camera Obscura, 3/4 (1979): 86.
21. Metz, On the Notion of Cinematographic Language, Movies and
Methods, ed. Bill Nichols (Berkeley: University of California Press,
1976): 587. Metz developed the grande syntagmatique in Chapter 4 of
Film Language: A Semiotics of the Cinema, trans. Michael Taylor (New
York: Oxford University Press, 1974).
22. In general terms, Metz argued that the cinema is not a language
system, because it contradicts three important characteristics of the
linguistic fact: a language is a system of signs used for intercommunication. Now, like all the arts, and because it is itself an art, the cinema is
one-way communication. Film Language, 75.
In more specific terms, Metz avoided the temptation to compare
the film image to the sentence. He argued that there is no structural
NOTES TO PP. 1314
23.
24.
25.
26.
27.
147
similarity between the two, because the sentence, unlike the image, is
also analyzable into codes: The difference [between image and sentence] is that the sentences of verbal language eventually break down
into words, whereas in the cinema, they do not: A film may be segmented into large units (shots), but these shots are not reducible (in
Jakobsons sense) into small, basic, and specific units. Film Language,
88.
Christian Metz, Essais sur la signification au cinema (Paris: Klincksieck,
1968): 51. Michael Taylor translates this phrase as a non-system language. Film Language, 44.
Bordwell, Narration in the Fiction Film, xiii.
Ibid., 29.
Ibid.
In reviewing the cognitivists critique of modern film theory, one must
at the very least acknowledge Noel Carrolls condemnation of Marxist
and psychoanalytic phases of modern film theory in his Mystifying
Movies: Fads and Fallacies in Contemporary Film Theory (New York: Columbia University Press, 1988). As I have already discussed this book
in detail elsewhere, I shall not discuss it here (see Warren Buckland,
Critique of Poor Reason, Screen, 30, 4 [1989]: 80103). However, it
may be worthwhile to respond briefly to Carrolls Cognitivism, Contemporary Film Theory, and Method: A Response to Warren Buckland (in Theorizing the Moving Image, 321335), his reply to my review.
Firstly, Carroll misunderstands my review of his book as an attack on
cognitivism. But my primary aim was to investigate the reasoning
behind his extreme interpretation of modern film theory, to examine
the logic of his claims and the conditions that make such an extreme
interpretation possible. I find Mystifying Movies a fascinating read because the extremities of its arguments are innovative (a breath of fresh
air) and seductive (as a result of their appeal to common sense and to
philosophical modes of reasoning such as the reductio ad absurdum).
However, the results are frequently misleading (the breath of fresh air
becomes a cold, biting wind), as my review attempts to demonstrate.
Secondly, I criticized the alternative cognitive theory Carroll presented in Mystifying Movies because it was insufficiently developed
and out of place. It seems that the cognitive sections were hurriedly
added as an afterthought. Furthermore, my review did not try to
defend the aims of modern film theory in the face of Carrolls critique.
After all, I stated near the beginning of my review that contemporary
film theory needs to be critically analysed at its foundations. However,
I differ with Carroll in stipulating the manner in which this critique
should be carried out (Critique of Poor Reason, 83). Both Carroll
and Edward Small fail to acknowledge this aspect of my review; for
example, Edward Small calls it reactionary (Introduction: Cognitivism and Film Theory, Journal of Dramatic Theory and Criticism, 6, 2
[1992]: 171). Such dismissive talk simply attempts to close off any
attempt to develop a debate between modern film theory and cognitivism, suggesting that the reader should either accept Carrolls argu-
148
28.
29.
30.
31.
32.
33.
34.
NOTES TO PP. 1416
ments or keep quiet. Is Smalls attitude to be taken as an exemplary

instance of the dialectical reasoning Carroll calls for? The tone of my
review simply implies that Carrolls critique of modern film theory is
exaggerated. When a new player enters the intellectual marketplace,
she establishes her identity by overstating her case (just think of the
slogans propounded by the Russian Formalists in their formative period). This is exactly what Carroll aimed for (and achieved) in his
review of Stephen Heaths Questions of Cinema in 1982 (the review
upon which Mystifying Movies is based). Carroll wanted to throw the
baby out with the bath water without getting wet. My review argued
that we need not go quite so far, but agreed that the bath water needed
to be changed.
Joseph Anderson, The Reality of Illusion: An Ecological Approach to Cognitive Film Theory (Carbondale: Southern Illinois University Press,
1996); Gregory Currie, Image and Mind: Film, Philosophy, and Cognitive
Science (Cambridge: Cambridge University Press, 1995); Ed Tan, Emotion and the Structure of Narrative Film: Film as an Emotional Machine
(Hillsdale, N.J.: Lawrence Erlbaum, 1996); Torben Grodal, Moving Pictures: A New Theory of Film Genres, Feelings, and Cognition (Oxford:
Clarendon Press, 1997).
Torben Grodal, Moving Pictures, 6. I review Grodals book in European
Journal of Communication, 13, 4 (1998): 577581.
Karl-Otto Apel, The Transcendental Conception of LanguageCommunication and the Idea of First Philosophy, History of Linguistic
Thought and Contemporary Linguistics, ed. Herman Parret (Berlin: Walter de Gruyter, 1976): 34.
Jurgen Habermas, Postmetaphysical Thinking: Philosophical Essays, trans.
William Mark Hohengarten (Cambridge: Polity Press, 1992): 208.
Gilbert Ryle, The Concept of Mind (London: Hutchinson, 1949).
Apel discusses how C. S. Peirce transformed Kantian epistemology
into Language Analysis in From Kant to Peirce: the Semiotical Transformation of Transcendental Logic, Towards a Transformation of Philosophy, trans. Glyn Adey and David Frisby (London: RKP, 1980): 7792.
Speech act theory, founded by J. L. Austin and systematized by John
Searle, is one of the dominant areas of research in pragmatics. For
Austin, Language Analysis involves explaining the way utterances are
embedded in social institutions, and also specifying what actions utterances perform. Language does not simply express ideas that describe
existing states of affairs in the world (the view of logical positivism,
which Austin is refuting). In addition, speakers use utterances to perform actions to promise, create obligations, influence the action of
others, and establish social relationships. For example, a priest who
utters to a couple, You are now man and wife isnt simply describing an existing state of affairs in the world; he is creating a new state
of affairs establishing a social relation between a couple by means
of his utterance. Moreover, this utterance performs its function only if
spoken by a genuine priest representing the institution of the church.
NOTES TO PP. 1628
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
149
More generally, speech act theory attempts to establish what types of

acts are performed by speakers and under what conditions a particular
act is successful. See J. L. Austin, How to Do Things with Words (Oxford:
Clarendon Press, 1962), and John Searle, Speech Acts (Cambridge: Cambridge University Press, 1969).
Jurgen Habermas, The Philosophical Discourse of Modernity, trans. Frederick Lawrence (Cambridge: Polity Press, 1987): 315.
Ibid., 322323.
Thomas Daddesio, On Minds and Symbols: The Relevance of Cognitive
Science for Semiotics (Berlin: Mouton de Gruyter, 1995): 4950.
Noam Chomsky, A Review of B. F. Skinners Verbal Behavior, Language, 35, 1 (1959): 2658.
David Bordwell, Contemporary Film Studies and the Vicissitudes of
Grand Theory, Post-Theory: Reconstructing Film Studies, ed. David
Bordwell and Noel Carroll (Madison: The University of Wisconsin
Press, 1996): 22.
Ibid.
Christopher Norris, The Contest of Faculties: Philosophy and Theory after
Deconstruction (London: Methuen, 1985): 222.
Language and Cinema, 19.
Metz, The Imaginary Signifier, Psychoanalysis and Cinema: The Imaginary Signifier, trans. Ben Brewster et al. (London: Macmillan, 1982): 1
87.
Metz, LEnonciation impersonnelle ou le site du film (Paris: Meridiens
Klincksieck, 1991).
Noam Chomsky, Knowledge of Language: Its Nature, Origin, and Use
(New York: Praeger, 1986): 24.
Noam Chomsky, Aspects of the Theory of Syntax (Cambridge, Mass.:
MIT Press, 1965).
Andre Martinet, Elements of General Linguistics, trans. Elizabeth Palmer
(London: Faber and Faber, 1964).
Roman Jakobson and Morris Halle, in Fundamentals of Language, 2nd
ed. (The Hague: Mouton, 1971), then broke down the phoneme into a
bundle of distinct features, which were defined in terms of binary
oppositions.
Chapter Two. The Body on Screen and in Frame

1.
2.
3.
4.
George Lakoff, Women, Fire, and Dangerous Things: What Categories

Reveal about the Mind (Chicago: University of Chicago Press, 1987): 386.
Christian Metz, Cinema: Language or Language System? Film Language: A Semiotics of the Cinema, trans. Michael Taylor (New York:
Oxford University Press, 1974): 3191.
Metz, Problems of Denotation in the Fiction Film, Film Language,
108146.
Stephen Heath, The Work of Christian Metz, Screen Reader 2: Cinema
and Semiotics (London: SEFT, 1981): 141.
150
NOTES TO PP. 2834
5. Umberto Eco, Articulation of the Cinematic Code, Movies and Methods, ed. Bill Nichols (Berkeley: University of California Press, 1976):
590607. For additional semiotic studies of iconicity, see Eco, A Theory
of Semiotics (Bloomington: Indiana University Press, 1976): 191217;
Sren Kjrup, Iconic Codes and Pictorial Speech Acts, Danish Semiotics, ed. Jrgen Dines Johansen and Morten Njgaard (Copenhagen:
Munksgaard, 1987): 101122.
6. These four terms can be represented in Greimass semiotic square as
follows:
arbitrariness
code
iconicity
motivation
7. Whereas the bottom-up approach to comprehension is empirical, for it

emphasizes perceptual input in the perceivers comprehension process, the top-down approach is cognitive, for it emphasizes the perceivers cognitive capacity that is, prior knowledge, expectations,
and ability to form hypotheses and make inferences in the comprehension process.
8. Bordwell, Narration in the Fiction Film (London: Methuen, 1985): 4950.
9. Ibid., 51.
10. Ibid., 33.
11. In relation to this point, Bordwell writes: It would be an error to take
the fabula, or story, as the profilmic event. A films fabula is never
materially present on the screen or soundtrack. Ibid., 49.
12. Lakoff, Women, Fire, and Dangerous Things, 116.
13. Bordwell, Narration in the Fiction Film, 6162.
14. Ibid., 62.
15. Ibid.
16. Michel Colin, Film Semiology as a Cognitive Science, The Film Spectator: From Sign to Mind, ed. Warren Buckland (Amsterdam: Amsterdam University Press, 1995): 87110.
17. Metz, Language and Cinema, trans. Donna-Jean Umiker-Sebeok (The
Hague: Mouton, 1974): 74.
18. Colin, Film Semiology as a Cognitive Science, 106.
19. Ibid., 8889.
20. See Edward Branigan, Narrative Comprehension and Film (London:
Routledge, 1992): 65.
22. See Umberto Eco, A Theory of Semiotics, Chapter 2, esp. 98114.
NOTES TO PP. 3441
151
23. Dan Sperber and Deirdre Wilson, Relevance: Communication and Cognition (Oxford: Basil Blackwell, 1986): 40.
24. Sperber and Wilson, Relevance, 4344.
25. The concept of mutual knowledge has been conceived and developed
in the following publications: David Lewis, Convention (Cambridge,
Mass.: Harvard University Press, 1969): 5260; Stephen Schiffer, Meaning (Oxford: Clarendon Press, 1972): 3042; and Neil Smith (ed.), Mutual Knowledge (London and New York: Academic Press, 1982).
26. Charles M. Eastman, Representations for Space Planning, Communication for the Association for Computing Machinery, 13 (1970): 242250;
Eastman, Automated Space Planning, Artificial Intelligence, 4 (1973):
4164.
28. Jerry Fodor, The Modularity of Mind (Cambridge, Mass.: MIT Press,
1983).
29. Ibid., 3637.
31. Ibid.
33. Ray Jackendoff, Semantics and Cognition (Cambridge, Mass.: MIT Press,
1983): 17.
34. Ibid., 16.
35. Bordwell, Narration in the Fiction Film, Chapter 2 (discussed in detail
in Chapter 1 of this book).
36. Winifred Noth, Semiotic Foundations of the Cognitive Paradigm,
Semiosis, 19, 1 (1994): 516.
37. Ibid., 5.
38. Mark Johnson, The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason (Chicago: University of Chicago Press, 1987); Mark
Turner, Reading Minds: The Study of English in the Age of Cognitive
Science (Princeton, N.J.: Princeton University Press, 1991); Ronald Langacker, Foundations of Cognitive Grammar, Volume 1, Theoretical Prerequisites (Stanford, Calif.: Stanford University Press, 1987); Volume II,
Descriptive Application (Stanford, Calif.: Stanford University Press,
1991); Langacker, Concept, Image, and Symbol: The Cognitive Basis of
Grammar (Berlin: Mouton de Gruyter, 1991); Gilles Fauconnier, Mental
Spaces: Aspects of Meaning Construction in Natural Language (Cambridge:
Cambridge University Press, 1994); Eve Sweester, Etymology to Pragmatism: Metaphorical and Cultural Aspects of Semantic Structure (Cambridge: Cambridge University Press, 1990).
39. Lakoff, Women, Fire, and Dangerous Things, 266267. Lakoffs and Johnsons theory of embodied meaning resembles Thomas A. Sebeoks
theory of the biosemiotic self. For Sebeok, each species constructs,
according to its own unique sensory and bodily structure and functions, its own Umwelt (its own perception of the outer world). Because
of the variation in the biological makeup of species, it is plausible to
argue that different species live in different sensory worlds. See Tho-
152
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
NOTES TO PP. 4148
mas A. Sebeok, An Introduction to Semiotics (London: Pinter Publishers,

1994), esp. 5, 1112, 113, 122123.
Lakoff, Women, Fire, and Dangerous Things, 310.
Johnson, The Body in the Mind, xvi.
Ibid., xxxvii.
Ibid., xv.
Johnson, The Body in the Mind, 3037.
Eugene Minkovski, quoted in Sue L. Cataldi, Emotion, Depth and Flesh:
A Study of Sensitive Space Reflections on Merleau-Pontys Philosophy of
Embodiment (New York: State University of New York Press, 1993): 50.
Crucially for modern film theorists, the process of production is excluded, leaving on screen the finished product. The frames ability to
draw the spectators attention away from the processes of production
is then defined as ideological.
Stephen Heath, On Screen, in Frame: Film and Ideology, Questions
of Cinema (London: Macmillan, 1981): 13.
Heath spells out this opposition between containment and dispersion
in the following terms:
containment imaginary, signified (positive meaning), stability
dispersion symbolic, signifier (negative meaning), flow
He argues that, in a narrative film, the spectator continually oscillates

between containment and dispersion. To explain how this oscillation
operates, Heath employs the concept of suture, which refers to an
interminable process that continually re-positions the spectator in an
imaginary relation to the film. This process designates an oscillation
between the spectators gradual recognition of the image as signifier
(as part of the symbolic that is, a lack, or absence) and the replacement of that image with another that temporarily cancels out that lack
or absence (thus re-establishing the spectators imaginary relation to
the film). Suture therefore designates a process mediating between the
imaginary and the symbolic. Moreover, this oscillation does not simply
operate from image to image, but from image to image through the
spectator (who poses the lack in one image, motivating the cut to the
next): The major emphasis in all this is that the articulation of the
signifying chain of images, of the chain of images as signifying, works
not from image to image but from image to image through the absence
that the subject constitutes [or poses]. Heath, Questions of Cinema, 88.
51. Etienne Souriau, La structure de lunivers filmique et le vocabulaire
de la filmologie, Revue Internationale de Filmologie, 78 (1951): 231240.
My discussion of Souriau is indebted to Edward Lowry, The Filmology
Movement and Film Study in France (Ann Arbor, Mich.: UMI Research
Press, 1985): 8486.
52. In a few instances, space can be in frame but off-screen as when a
mirror reflects a space not contained on screen.
NOTES TO PP. 4958
153
53. For the most comprehensive theory to date on the point of view shot,
see Edward Branigan, Point of View in the Cinema: A Theory of Narration
and Subjectivity in Classical Film (Berlin: Mouton, 1984), and Narrative
Comprehension and Film (London: Routledge, 1992).
54. For the equivocal use of point of view shots in the films of Fritz Lang,
see Raymond Bellour, On Fritz Lang, Fritz Lang: The Image and the
Look, ed. Stephen Jenkins (London: BFI, 1981): 2637. I have written
about Spielbergs use of equivocal point of view shots in Raiders of the
Lost Ark in A Close Encounter with Raiders of the Lost Ark, Contemporary Hollywood Cinema, ed. Steve Neale and Murray Smith (London:
Routledge, 1998), 166177.
55. In Film, Language, and Conceptual Structures. Thinking Film in the Age of
Cognitivism (Amsterdam: Academisch Proefschrift, University of Amsterdam, 1995), Jan Simons presents a detailed application of the
sourcepathgoal schema to Dutch political election campaign films.
Chapter Three. Not What Is Seen through the Window but the
Window Itself
1.
2.
3.
4.
5.
6.
7.
8.
Christian Metz, The Impersonal Enunciation, or the Site of Film, in

Warren Buckland (ed.), The Film Spectator: From Sign to Mind (Amsterdam: Amsterdam University Press, 1995): 145146.
Christian Metz, Mirror Construction in Fellinis 812, Film Language:
A Semiotics of the Cinema, trans. Michael Taylor (New York: Oxford
University Press, 1974): 228234; Metz, Story/Discourse (A Note on
Two Kinds of Voyeurism), Psychoanalysis and Cinema: The Imaginary
Signifier, trans. Ben Brewster, Celia Britton, Alfred Guzzetti, and Annwyl Williams (London: Macmillan, 1982): 9198; Metz, LEnonciation
impersonnelle ou le site du film (Paris: Meridiens Klincksieck, 1991).
Throughout this chapter, I shall primarily refer to the first section of
this book, translated into English as The Impersonal Enunciation, or
the Site of Film, in The Film Spectator: From Sign to Mind, 140163.
For a more general overview of Francesco Casettis work (but one that
does not discuss his deictic theory of filmic enunciation in any detail),
see Giuliana Muscio and Roberto Zemignan, Francesco Casetti and
Italian Film Semiotics, Cinema Journal, 30, 2 (Winter 1991): 2346.
Metz, The Impersonal Enunciation, 146.
Ibid., 159.
Ibid.
Francesco Casetti, Dentro lo sguardo: Il filme e il sou spettatore (Milano:
Bompiani, 1986), translated into French as DUn regard lautre: Le film
et son spectateur, trans. Jean Chateauvert and Martine Joly (Lyon:
Presses Universitaires de Lyon, 1990); English translation forthcoming
from Indiana University Press.
Metz, Sur une traversee des Alpes et des Pyrenees . . . , Preface to
Casetti, DUn regard lautre, 510.
154
NOTES TO PP. 5868
9. See LEnonciation impersonnelle, part II, 37172, where Metz outlines ten
forms of filmic enunciation.
10. Francesco Casetti, Looking for the Spectator, Iris, 2 (1983): 19.
11. Ibid., 21.
12. Ibid., 22.
13. For a more detailed account of what Casetti means by text, see his
Le texte du film, Theorie du film, ed. Jacques Aumont and Jean-Louis
Leutrat (Paris: Albatros, 1980): 4165.
14. Casetti, Looking for the Spectator, 24.
15. Ibid., 25.
16. See also Casetti, Pragmatique et theorie du cinema aujourdhui, Hors
Cadre, 10 (1992): 99109.
17. Casetti, Looking for the Spectator, 29.
18. Francesco Casetti, Face to Face, in The Film Spectator, 118139.
19. Ibid., 118.
20. Ibid., 124. Casetti is aware of the empirical implications of this position, since he notes, This [position] results in hypotheses that one
must conform to, traces that indicate the unfolding of the action, and
parameters that must be obeyed (ibid., 126). In particular, this position suggests that only one meaning can be attributed to these textual
marks, a premise Metz deconstructs in The Impersonal Enunciation.
21. Face to Face, 128.
22. Ibid., 128129.
23. Ibid., 129.
24. Ibid., 130.
25. Ibid.
26. Ibid., 131.
27. Metz, The Impersonal Enunciation, 141.
28. Ibid., 149150.
29. Metz also challenges Casettis characterization of the film spectator as
interlocutor because this term is dependent upon the idea of immediate interaction, which, Metz argues, is absent in film.
30. Metz, The Impersonal Enunciation, 154.
31. Ibid., 152.
32. Ibid., 153.
33. Ibid., 150.
34. Ibid., 150151.
35. Ibid., 151.
36. Ibid.
37. Ibid., 154155.
38. Ibid., 147148.
39. Ibid., 146.
40. Ibid., 153. Yet, at the same time, Metz does stress that film vaguely
acknowledges the extra-textual spectator, both in his revolver example
(the here is acknowledges the presence of a spectator), as well as in
his later work, particularly The Imaginary Signifier, where Metz
states that film exists for the spectator. This is quite evident in the
NOTES TO PP. 6878
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
155
frontality of the mise-en-sce`ne, in camera angles and camera movement,

and in the shifts from omniscient to restricted narration, all designed
to aid (or deliberately block) the spectators access to the films diegesis.
We need to remember that Benveniste introduces the distinction between histoire and discours in order to offer an adequate classification
of the various tense forms in modern French, since (and this is the
main argument behind Benvenistes paper see note 42) the traditional
grammarians exclusive reliance upon temporal divisions is insufficient. Discours and histoire therefore represent two different but complementary planes of utterance, each manifesting a distinct tense system in modern French.
Emile Benveniste, The Correlations of Tense in the French Verb,
Problems in General Linguistics, trans. Mary Elizabeth Meek (Coral Gables; Fla.: University of Miami Press, 1971): 210.
Karl Buhler, quoted in Achim Eschbach, Editors Introduction Karl
Buhler: Sematologist, Karl Buhler, Theory of Language: The Representational Function of Language, trans. Donald Fraser Goodwin (Amsterdam: John Benjamins, 1990): xxix.
Karl Buhler, Theory of Language, 145 (emphasis in the original).
Daniel Dayan and Elihu Katz, Electronic Ceremonies: Television Performs a Royal Wedding, On Signs, ed. Marshall Blonsky (Baltimore:
The Johns Hopkins University Press, 1985): 1632.
Judith Halberstam, Skin Shows: Gothic Horror and the Technology of Monsters (Durham, N.C.: Duke University Press, 1995): 167.
Ibid., 166167.
David Rodowick, The Difficulty of Difference: Psychoanalysis, Sexual Difference and Film Theory (London and New York: Routledge, 1991): 135.
Tom Conley, Film Hieroglyphs: Ruptures in Classical Cinema (Minneapolis and Oxford: The University of Minnesota Press, 1991): xxxi.
Metz, The Imaginary Signifier, 15, 17, 79, and 80. See also Story/
Discourse, 9293.
Metz, The Impersonal Enunciation, 157.
Chapter Four. The Institutional Context

1.
2.
3.
4.
Roger Odin, Approche semio-pragmatique, approche historique, Kodikas/Code, 17, 14 (1994): 2728.
Stepehen Levinson, Pragmatics (Cambridge: Cambridge University
Press, 1983): 32.
Bronislaw Malinowski, A Scientific Theory of Culture and Other Essays
(Chapel Hill: The University of North Carolina Press, 1944): 39.
For an outline of these recent debates, see the opening of Noel Carrolls
essay, From Real to Reel: Entangled in Nonfiction Film, Theorizing
the Moving Image (Cambridge: Cambridge University Press, 1996): 224
252.
156
NOTES TO PP. 7886
5. Francesco Casetti, Pragmatique et theorie du cinema aujourdhui,

Hors Cadre, 7 (1989): 104.
6. Melina Sinclair, Fitting Pragmatics in the Mind: Some Issues in Mentalist Pragmatics, Journal of Pragmatics, 23 (1995): 509539.
7. Jan Nuyts, Aspects of a Cognitive-Pragmatic Theory of Language: On Cognition, Functionalism, and Grammar (Amsterdam: John Benjamins, 1992);
Diane Blakemore, Understanding Utterances (Oxford: Basil Blackwell,
1992); Asa Kasher, On the Psychological Reality of Pragmatics, Journal of Pragmatics, 8 (1984): 539557; Dan Sperber and Deirdre Wilson,
Relevance: Communication and Cognition (Oxford: Basil Blackwell, 1986).
8. Roger Odin, La semio-pragmatique du cinema sans crise, ni desillusion, Hors Cadre, 7 (1989): 8384.
9. Christian Metz, Language and Cinema, trans. Donna-Jean UmikerSebeok (The Hague: Mouton, 1974).
10. Ibid., 7374.
11. Odin, La semio-pragmatique du cinema, 84.
But Jan Simons is critical of Odins direct cognitive re-reading of
Language and Cinema (Jan Simons, Cognitivism and Pragmatics, unpublished ms., 1991). He begins by agreeing with Odin: Odin is . . .
right when he claims that the description of the codes of a language
(e.g. the cinematographic language) has some cognitive relevance, insofar as the understanding of film can be considered as the result of
some decoding (23). But he questions Odins assumption that the
spectator decodes the filmic text in pretty much the same way as the
semiologist does (which means with the same cognitive tools and
procedures of segmentation and commutation) (3). Simons is critical
of this assumption because the object of description and explanation
of structuralist linguistics is not the knowledge of language-speakers,
but the immanent structure of the language system (5). Finally, he
argues that Odins assumption is too literal because even the most
simple textual analysis of even a very small segment of a film very
quickly reaches a degree of complexity a natural spectator is very
unlikely to perform in one viewing of a film (7).
12. Christian Metz, Psychoanalysis and Cinema: The Imaginary Signifier,
trans. Ben Brewster, Alfred Guzzetti, Celia Britton, and Annwyl Williams (London: Macmillan, 1982).
13. Language and Cinema, 121128.
14. Roy Harris, Reading Saussure (London: Duckworth, 1987): 2425.
15. Ibid., 205.
16. Ibid., 216.
17. Roger Odin, A Semio-Pragmatic Approach to the Documentary
Film, The Film Spectator: From Sign to Mind, ed. Warren Buckland
(Amsterdam: University of Amsterdam Press, 1995): 227.
18. Sperber and Wilson, Relevance, 125.
19. David Bordwell, Making Meaning: Inference and Rhetoric in the Interpretation of Cinema (Cambridge, Mass.: Harvard University Press, 1989):
89.
NOTES TO PP. 8696
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
157
Ibid.
Metz, Psychoanalysis and Cinema, 3.
Bordwell, Making Meaning, 270.
Barbara Klinger, Melodrama and Meaning: History, Culture, and the Films
of Douglas Sirk (Bloomington: Indiana University Press, 1994).
Ibid., xv.
Ibid.
Ibid., xix.
Ibid., 127.
Odin, Semio-pragmatique du cinema et de laudiovisuel: Modes et
institutions, Towards a Pragmatics of the Audio-Visual, Vol. 1, ed. Jurgen
Muller (Munster: Nodus Publikationen, 1994): 3346.
Odin, For a Semio-Pragmatics of Film, The Film Spectator: From Sign
to Mind, 222.
Odin, Rhetorique du film de famille, Revue dEsthetique, 12 (1979):
368.
Branigan, On the Analysis of Interpretive Language, Part I, Film
Criticism, XVII, 23 (1993): 8.
Jan Simons, Pragmatics, Deixis, and the Political Election-Campaign
Film, Towards a Pragmatics of the Audio-Visual, Vol. 1, 79.
Ibid., 82.
Roger Odin, Du spectateur fictionalisant au nouveau spectateur: approche semio-pragmatique, Iris, 8 (1988): 121139.
A. J. Greimas and J. Courtes, Semiotics and Language: An Analytical
Dictionary, trans. Larry Crist et al. (Bloomington: Indiana University
Press, 1982): 119.
Umberto Eco, Articulation of the Cinematic Code, Movies and Methods, ed. Bill Nichols (Berkeley: University of California Press, 1976):
590607.
Edward Branigan, Diegesis and Authorship in Film, Iris, 7 (1986):
44.
Odin, Du spectateur fictionnalisant, 124.
Ibid.
Genette, discussed on page 125 of Du spectateur fictionalisant.
John Searle, The Logical Status of Fictional Discourse, Expression and
Meaning (Cambridge: Cambridge University Press, 1979): 6061.
Richard Allen, Projecting Illusion: Film Spectatorship and the Impression
of Reality (Cambridge: Cambridge University Press, 1995): 135143.
Ibid., 138139.
Andre Gaudreault, Du litteraire au filmique: Syste`me du recit (Paris:
Meridiens Klincksieck, 1988), Chapter VIII, and Narration and Monstration in the Cinema, Journal of Film and Video, 39, 2 (1987): 2936.
Roger Odin, Mise en phase, dephasage et performitive dans Le Tempestaire de Jean Epstein, Communications 38 (1983): 213238.
Ibid., 225.
Du spectateur fictionnalisant, 127128.
Ibid., 128.
158
NOTES TO PP. 96108
49. Ibid.
50. Roger Odin, Semio-pragmatique du cinema et de laudiovisuel:
Modes et institutions, Towards a Pragmatics of the Audio-Visual, Vol. 1,
3346.
51. Ibid., 39.
52. Metz, Psychoanalysis and Cinema, 7.
53. Odin, Semio-pragmatique du cinema et de laudiovisuel, 41.
54. Odin, Approche semio-pragmatique, approche historique, 28.
55. Odin, Film documentaire, lecture documentarisante, Cinemas et realites, ed. J. Lyant and Roger Odin (Saint-Etienne: Cierec, 1984): 263280.
56. Odin, Film documentaire, lecture documentarisante, 276.
57. Greimas and Courtes, Semiotics and Language: An Analytical Dictionary,
307308.
58. Odin, Film documentaire, lecture documentarisante, 266.
59. Ibid., 268269.
60. Ibid., 273.
61. Odin, Semio-pragmatique du cinema et de laudiovisuel, 34. More
recently, Odin has edited a book on the home movie: Le film de famille,
ed. Roger Odin (Paris: Meridiens Klincksieck, 1995).
62. Odin, Rhetorique du film de famille, Revue dEsthetique, 12 (1979):
240273.
63. Ibid., 345.
64. Ibid., 348353.
65. Ibid., 356.
66. Ibid., 366.
67. Odin, Du spectateur fictionalisant, 132.
68. Odin, Cinema et production de sens (Paris: Editions Armand Colin, 1990):
244.
69. Odin, Du spectateur fictionalisant, 133.
70. Ibid., 134.
71. Bordwell, Making Meaning, 271. Bordwells theory of parametic narration is developed in Chapter 12 of Narration in the Fiction Film (London:
Methuen, 1985); his analysis of Ozus films is to be found in Ozu and
the Poetics of Cinema (Princeton, N.J.: Princeton University Press, 1988).
73. Ibid., 274.
74. Richard Allen (personal communication) suggests that it would also
be possible to begin thinking of the actant of the documentary film as
fictional (the documentarys implied author), a process that would
bring fiction and documentary even closer together.
75. Wolfgang Iser, Prospecting: From Reader Response to Literary Anthropology (Baltimore: The Johns Hopkins University Press, 1989), and The
Fictive and the Imaginary: Charting Literary Anthropology (Baltimore: The
Johns Hopkins University Press, 1993).
76. Iser, Prospecting, 265.
77. Ibid., 267.
NOTES TO PP. 109117
159
Chapter Five. All in the Mind?

1. Noam Chomsky, Aspects of the Theory of Syntax (Cambridge, Mass.:
MIT Press, 1965): 4.
2. Dominique Chateau, Le cinema comme langage (Brussels: AISS
Publications de la Sorbonne, 1987): 50.
3. David Bordwell, Contemporary Film Studies and the Vicissitudes of
Grand Theory, Post-Theory: Reconstructing Film Studies, ed. David
Bordwell and Noel Carroll (Madison: The University of Wisconsin
Press, 1996): 22.
4. Noam Chomsky, Current Issues in Linguistic Theory (The Hague: Mouton, 1964): 89.
5. Chomsky, Aspects, 24.
6. Chomsky, in Robert Reiber, Dialogues on the Psychology of Language and
Thought (New York: Plenum Press, 1983): 42.
7. Chomsky, Linguistics and Adjacent Fields: A Personal View, The
Chomskyan Turn, ed. Asa Kasher (Oxford: Basil Blackwell, 1991): 9.
8. Scott Soames, Linguistics and Psychology, Linguistics and Philosophy,
7, 2 (1984): 155179.
9. Ibid., 165.
10. Ibid.
11. Ibid.
12. Michel Colin, The Grande Syntagmatique Revisited, The Film Spectator: From Sign to Mind, ed. Warren Buckland (Amsterdam: Amsterdam University Press, 1995): 4586.
13. Ibid., 50.
14. Colin, The Grande Syntagmatique Revisited, 4748.
15. Christian Metz, Film Language: A Semiotics of the Cinema, trans. Michael
Taylor (New York: Oxford University Press, 1974): 146.
16. For Metz, on the other hand, the dominant asymmetry in the GS lies
within the autonomous shot, which is divided into two heterogeneous
sub-types the sequence shot and four types of insert. The tree diagram of the GS also shows the class of sequence re-written as two
syntagmatic types. But Colin finds this acceptable (presumably because it marks the closure of the GS).
17. Metz, Film Language, 126.
18. Colin, The Grande Syntagmatique Revisited, 67.
19. Ibid., 73.
20. Colin distinguishes the scene from the sequence in terms of the way
they each represent space. The scene is founded on the unity of what
Colin calls a spatial frame, whereas the sequence is founded on the
unity of an action that may take place across different spaces (the
action creates a path, not a spatial frame). Colin indicates this distinction by means of the inclusive feature: This inclusive feature
means that the scene shows relationships of inclusion between the
spaces (ibid., 74).
160
NOTES TO PP. 118137
21.
22.
23.
24.
Ibid.
Ibid., 58.
Ibid., 75.
Metz, Rapport sur letat actuel de la semiologie du cinema dans le
monde (debut 1974), A Semiotic Landscape/Panorama Semiotique, ed.
Seymour Chatman, Umberto Eco, and Jean-Marie Klinkenberg (The
Hague: Mouton, 1979): 151.
Chateau, Le cinema comme langage, 52.
Ibid., 59.
Ibid., 58.
Noam Chomsky, Degrees of Grammaticalness, The Structure of Language: Readings in the Philosophy of Language, ed. Jerry A. Fodor and
Jerrold J. Katz (Englewood Cliff, N.J.: Prentice-Hall, 1964): 387.
Jerrold J. Katz, Semi-Sentences, The Structure of Language, ed. Fodor
and Katz, 400416.
Ibid., 410411.
Ibid., 411.
Chomsky, Aspects, 11.
Metz, Modern Cinema and Narrativity, Film Language, 185227.
Metz, Modern Cinema and Narrativity, 211.
Ibid.
Ibid., 219 (emphasis in the original).
Chomsky, Degrees of Grammaticalness, 384.
Noam Chomsky, Language and the Problems of Knowledge, 81.
Metz, Modern Cinema and Narrativity, Film Language, 219.
David Bordwell, Narration in the Fiction Film (London: Methuen, 1985):
213228.
Ibid., 217218.
Ibid., 218.
Ibid., 211.
Ibid., 213.
Fred Lerdahl and Ray Jackendoff, A Generative Theory of Tonal Music,
9.
Ibid., 310. This is, in fact, Lerdahl and Jackendoffs restatement of H. P.
Grices conversational maxim of relation, from which Sperber and
Wilsons principle of relevance derives.
Metz, Film Language, 117.
Ibid.
Eric Wanner, Psychology and Linguistics in the Sixties, The Making
of Cognitive Science: Essays in Honour of George A. Miller, ed. William
Hirst (Cambridge: Cambridge University Press, 1988): 144.
Colin, The Grande Syntagmatique Revisited, 80.
Ibid., 81.
Scott Soames, Linguistics and Psychology, 162.
Thomas G. Bever, The Psychological Reality of Grammar, The Making of Cognitive Science: Essays in Honour of George A. Miller, ed. William
Hurst, 132133.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
NOTES TO PP. 137144
161
54. John M. Carroll, Linguistics, Psychology, and Cinema Theory, Semiotica 20, 12 (1977): 180.
55. Ibid., 183184.
56. John M. Carroll, Toward a Structural Psychology of Cinema (The Hague:
Mouton, 1980).
57. Ibid., 194.
58. Ibid., 69.
59. Carroll, A Program for Cinema Theory, Journal of Aesthetics and Art
Criticism, 35 (1977): 344.
60. In the central chapters of his book, Carroll uses TGG in a purely
analogous manner, arbitrarily converting filmic events into re-write
rules. The cognitive film semioticians discussed in this book do not
use linguistics in such an arbitrary and forced manner.
61. Carroll, Toward a Structural Psychology of the Cinema, 118.
62. Ibid., 200203.
Conclusion
1.
2.
3.
4.
5.
6.
7.
8.
Thomas Sebeok, An Introduction to Semiotics (London: Pinter, 1994):

112.
Michel Colin, Langue, film, discours: prolegome`nes a` une semiologie generative du film (Paris: Klincksieck, 1985): 15.
Sebeok, An Introduction to Semiotics, 4.
David Bordwell and Noel Carroll, eds., Post-Theory: Reconstructing Film
Studies (Madison: University of Wisconsin Press, 1996).
Robert de Beaugrande, Complexity and Linguistics in the Evolution
of Three Paradigms, Theoretical Linguistics, 17, 13 (1991): 45.
Negative complexity characterizes a situation in which the phenomena
in the domain under study become too heterogeneous to be theorized.
Positive complexity characterizes a situation in which the phenomena
in the domain can be theorized successfully.
Thomas Kuhn, The Structure of Scientific Revolutions (Chicago: University of Chicago Press, 1970).
Robert de Beaugrande, Complexity and Linguistics, 48.
Bibliography of Works Cited
General
Apel, Karl-Otto, The Transcendental Conception of Language-Communication and the Idea of First Philosophy, History of Linguistic Thought
and Contemporary Linguistics, ed. Herman Parret (Berlin: Walter de
Gruyter, 1976).
From Kant to Peirce: the Semiotical Transformation of Transcendental
Logic, Towards a Transformation of Philosophy, trans. Glyn Adey and
David Frisby (London: RKP, 1980): 7792.
Austin, J. L., How to Do Things with Words (Oxford: Clarendon Press, 1962).
Benveniste, Emile, The Correlations of Tense in the French Verb, Problems in General Linguistics, trans. Mary Elizabeth Meek (Coral Gables,
Fla.: University of Miami Press, 1971): 205215.
Bever, Thomas G., The Psychological Reality of Grammar, The Making of
Cognitive Science: Essays in Honour of George A. Miller, ed. William Hirst
(Cambridge: Cambridge University Press, 1988).
Blakemore, Diane, Understanding Utterances (Oxford: Basil Blackwell, 1992).
Buhler, Karl, Theory of Language: The Representational Function of Language,
trans. Donald Fraser Goodwin (Amsterdam: John Benjamins, 1990).
Cataldi, Sue L., Emotion, Depth and Flesh: A Study of Sensitive Space Reflections on Merleau-Pontys Philosophy of Embodiment (New York: State
University of New York Press, 1993).
Chomsky, Noam, A Review of B. F. Skinners Verbal Behavior, Language,
35, 1 (1959): 2658.
Degrees of Grammaticalness, The Structure of Language: Readings in the
Philosophy of Language, ed. Jerry A. Fodor and Jerrold J. Katz (Englewood Cliffs, N.J.: Prentice-Hall, 1964).
Current Issues in Linguistic Theory (The Hague: Mouton, 1964).
Aspects of the Theory of Syntax (Cambridge, Mass.: MIT Press, 1965).
Knowledge of Language: Its Nature, Origin, and Use (New York: Praeger,
1986).
Linguistics and Adjacent Fields: A Personal View, The Chomskyan
Turn, ed. Asa Kasher (Oxford: Basil Blackwell, 1991).
Daddesio, Thomas, On Minds and Symbols: The Relevance of Cognitive Science
for Semiotics (Berlin: Mouton de Gruyter, 1995).
163
164
BIBLIOGRAPHY
De Beaugrande, Robert, Complexity and Linguistics in the Evolution of

Three Paradigms, Theoretical Linguistics, 17, 13 (1991): 4373.
Deely, John, Brooke Williams, and Felicia E. Kruse (eds.), Frontiers in Semiotics (Bloomington: Indiana University Press, 1986).
Denisov, P. N., Principles of Constructing Linguistic Models (The Hague:
Mouton, 1973).
De Saussure, Ferdinand, Course in General Linguistics, trans. Roy Harris
(London: Duckworth 1983).
Eastman, Charles M., Representations for Space Planning, Communication
for the Association for Computing Machinery, 13 (1970): 242250.
Automated Space Planning, Artificial Intelligence, 4 (1973): 4164.
Eco, Umberto, A Theory of Semiotics (Bloomington: Indiana University Press,
1976).
Fauconnier, Gilles, Mental Spaces: Aspects of Meaning Construction in Natural
Language (Cambridge: Cambridge University Press, 1994).
Fodor, Jerry, The Modularity of Mind (Cambridge, Mass.: MIT Press, 1983).
Greimas, A. J., and J. Courtes, Semiotics and Language: An Analytical
Dictionary, trans. Larry Crist et al. (Bloomington: Indiana University
Press, 1982).
Habermas, Jurgen, The Philosophical Discourse of Modernity, trans. Frederick
Lawrence (Cambridge: Polity Press, 1987).
Postmetaphysical Thinking: Philosophical Essays, trans. William Mark Hohengarten (Cambridge: Polity Press, 1992).
Harris, Roy, Reading Saussure (London: Duckworth, 1987).
Iser, Wolfgang, Prospecting: From Reader Response to Literary Anthropology
(Baltimore: The Johns Hopkins University Press, 1989).
The Fictive and the Imaginary: Charting Literary Anthropology (Baltimore:
The Johns Hopkins University Press, 1993).
Jackendoff, Ray, Semantics and Cognition (Cambridge, Mass.: MIT Press,
1983).
Jakobson, Roman, and Morris Halle, Fundamentals of Language, 2nd ed. (The
Hague: Mouton, 1971).
Johnson, Mark, The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason (Chicago: University of Chicago Press, 1987).
Kasher, Asa, On the Psychological Reality of Pragmatics, Journal of Pragmatics, 8 (1984): 539557.
Katz, Jerrold J., Semi-Sentences, The Structure of Language: Readings in the
Philosophy of Language, ed. Jerry A. Fodor and Jerrold J. Katz (Englewood Cliffs, N.J.: Prentice-Hall, 1964): 400416.
Kjrup, Sren, Iconic Codes and Pictorial Speech Acts, Danish Semiotics,
ed. Jrgen Dines Johansen and Morten Njgaard (Copenhagen;
Munksgaard, 1987): 101122.
Kuhn, Thomas, The Structure of Scientific Revolutions (Chicago: University
of Chicago Press, 1970).
Lakoff, George, Women, Fire, and Dangerous Things: What Categories Reveal
about the Mind (Chicago: University of Chicago Press, 1987).
Langacker, Ronald, Foundations of Cognitive Grammar, Volume I; Theoretical
BIBLIOGRAPHY
165
Prerequisites (Stanford, Calif.: Stanford University Press, 1987); Volume

II, Descriptive Application (Stanford, Calif.: Stanford University Press,
1991).
Concept, Image, and Symbol: The Cognitive Basis of Grammar (Berlin: Mouton de Gruyter, 1991).
Lerdahl, Fred, and Ray Jackendoff, A Generative Theory of Tonal Music (Cambridge, Mass.: MIT Press, 1995).
Levinson, Stephen, Pragmatics (Cambridge: Cambridge University Press,
1983).
Lewis, David, Convention (Cambridge, Mass.: Harvard University Press,
1969).
Malinowski, Bronislaw, A Scientific Theory of Culture and Other Essays
(Chapel Hill: The University of North Carolina Press, 1944).
Martinet, Andre, Elements of General Linguistics, trans. Elizabeth Palmer
(London: Faber and Faber, 1964).
Norris, Christopher, The Contest of Faculties: Philosophy and Theory after
Deconstruction (London: Methuen, 1985).
Noth, Winfried, Semiotic Foundations of the Cognitive Paradigm, Semiosis, 19, 1 (1994): 516.
Nuyts, Jan, Aspects of a Cognitive-Pragmatic Theory of Language: On Cognition,
Functionalism, and Grammar (Amsterdam: John Benjamins, 1992).
Reiber, Robert, Dialogues on the Psychology of Language and Thought (New
York: Plenum Press, 1983).
Ryle, Gilbert, The Concept of Mind (London: Hutchinson, 1949).
Schiffer, Stephen, Meaning (Oxford: Clarendon Press, 1972).
Searle, John, Speech Acts (Cambridge: Cambridge University Press, 1969).
The Logical Status of Fictional Discourse, Expression and Meaning
(Cambridge: Cambridge University Press, 1979): 5873.
Sebeok, Thomas A., An Introduction to Semiotics (London: Pinter Publishers,
1994).
Sinclair, Melina, Fitting Pragmatics in the Mind: Some Issues in Mentalist
Pragmatics, Journal of Pragmatics, 23 (1995): 509539.
Smith, Neil (ed.), Mutual Knowledge (London and New York: Academic
Press, 1982).
Soames, Scott, Linguistics and Psychology, Linguistics and Philosophy, 7,
2 (1984): 155179.
Sperber, Dan, and Deirdre Wilson, Relevance: Communication and Cognition
(Oxford: Basil Blackwell, 1986).
Sweester, Eve, Etymology to Pragmatism: Metaphorical and Cultural Aspects of
Semantic Structure (Cambridge: Cambridge University Press, 1990).
Turner, Mark, Reading Minds: The Study of English in the Age of Cognitive
Science (Princeton, N.J.: Princeton University Press, 1991).
Wanner, Eric, Psychology and Linguistics in the Sixties, The Making of
Cognitive Science: Essays in Honour of George A. Miller, ed. William Hirst
Weber, Samuel, Saussure and the Apparition of Language: The Critical
Perspective, Modern Language Notes, 91 (1976).
166
BIBLIOGRAPHY
Film Studies
Allen, Richard, Projecting Illusion: Film Spectatorship and the Impression of
Reality (Cambridge: Cambridge University Press, 1995).
Anderson, Joseph, The Reality of Illusion: An Ecological Approach to Cognitive
Film Theory (Carbondale: Southern Illinois University Press, 1996).
Bellour, Raymond, On Fritz Lang, Fritz Lang: The Image and the Look, ed.
Stephen Jenkins (London: BFI, 1981): 2637.
Bergstrom, Janet, Alternation, Segmentation, Hypnosis: Interview with
Raymond Bellour, Camera Obscura, 3/4 (1979): 71103.
American Feminism and French Film Theory, Iris, 10 (1990): 183197.
Bordwell, David, Narration in the Fiction Film (London: Methuen, 1985).
Ozu and the Poetics of Cinema (Princeton, N.J.: Princeton University Press,
1988).
Making Meaning: Inference and Rhetoric in the Interpretation of Cinema
(Cambridge, Mass.: Harvard University Press, 1989).
Contemporary Film Studies and the Vicissitudes of Grand Theory,
Post-Theory: Reconstructing Film Studies, ed. David Bordwell and Noel
Carroll (Madison: The University of Wisconsin Press, 1996): 336.
Bordwell, David, and Noel Carroll, eds., Post-Theory: Reconstructing Film
Studies (Madison: University of Wisconsin Press, 1996).
Branigan, Edward, Point of View in the Cinema: A Theory of Narration and
Subjectivity in Classical Film (Berlin: Mouton, 1984).
Diegesis and Authorship in Film, Iris, 7 (1986): 3754.
Narrative Comprehension and Film (London: Routledge, 1992).
On the Analysis of Interpretive Language, Part I, Film Criticism, XVII,
23 (1993).
Buckland, Warren, Critique of Poor Reason, Screen, 30, 4 (1989): 80103.
(ed.), The Film Spectator: From Sign to Mind (Amsterdam: Amsterdam
University Press, 1995).
A Close Encounter with Raiders of the Lost Ark, Contemporary Hollywood
Cinema, ed. Steve Neale and Murray Smith (London: Routledge, 1998):
166177.
review of Torben Grodal, Moving Pictures, European Journal of Communication, 13, 4 (1998): 577581.
Carroll, John M., Linguistics, Psychology, and Cinema Theory, Semiotica
20, 12 (1977): 173195.
A Program for Cinema Theory, Journal of Aesthetics and Art Criticism,
35 (1977): 337351.
Toward a Structural Psychology of Cinema (The Hague: Mouton, 1980).
Carroll, Noel, Mystifying Movies: Fads and Fallacies in Contemporary Film
Theory (New York: Columbia University Press, 1988).
Theorizing the Moving Image (Cambridge: Cambridge University Press,
1996).
Cognitivism, Contemporary Film Theory, and Method: A Response to
Warren Buckland, Theorizing the Moving Image (Cambridge: Cambridge University Press, 1996): 321335.
BIBLIOGRAPHY
167
Prospects for Film Theory: A Personal Assessment, Post-Theory: Reconstructing Film Studies, ed. David Bordwell and Noel Carroll (Madison:
University of Wisconsin Press, 1996): 3768.
Casetti, Francesco, Le texte du film, Theorie du film, ed. Jacques Aumont
and Jean-Louis Leutrat (Paris: Albatros, 1980): 4165.
Looking for the Spectator,Iris, 2 (1983): 1529.
Pragmatique et theorie du cinema aujourdhui, Hors Cadre, 7 (1989):
99109.
DUn regard lautre. Le film et son spectateur, trans. Jean Chateauvert and
Martine Joly (Lyon: Presses Universitaires de Lyon, 1990).
Pragmatique et theorie du cinema aujourdhui, Hors Cadre, 10 (1992):
99109.
Face to Face, The Film Spectator: From Sign to Mind, ed. Warren Buckland (Amsterdam: Amsterdam University Press, 1995): 118139.
Chateau, Dominique, Le cinema comme langage (Brussels: AISS
Publications de la Sorbonne, 1987).
Colin, Michel, Langue, film, discours: prolegome`nes a` une semiologie generative
du film (Paris: Klincksieck, 1985).
The Grande Syntagmatique Revisited, The Film Spectator: From Sign to
Mind, ed. Warren Buckland (Amsterdam: Amsterdam University
Press, 1995): 4586.
Film Semiology as a Cognitive Science, The Film Spectator: From Sign to
Mind, ed. Warren Buckland (Amsterdam: Amsterdam University
Press, 1995): 87110.
Conley, Tom, Film Hieroglyphs: Ruptures in Classical Cinema (Minneapolis
and Oxford: The University of Minnesota Press, 1991).
Currie, Gregory, Image and Mind: Film, Philosophy, and Cognitive Science
Dayan, Daniel, and Elihu Katz, Electronic Ceremonies: Television Performs a Royal Wedding, On Signs, ed. Marshall Blonsky (Baltimore:
The Johns Hopkins University Press, 1985): 1632.
Eco, Umberto, Articulation of the Cinematic Code, Movies and Methods,
ed. Bill Nichols (Berkeley: University of California Press, 1976): 590
607.
Gaudreault, Andre, Narration and Monstration in the Cinema, Journal of
Film and Video, 39, 2 (1987): 2936.
Du litteraire au filmique: Syste`me du recit (Paris: Meridiens Klincksieck,
1988).
Grodal, Torben, Moving Pictures: A New Theory of Film Genres, Feelings, and
Cognition (Oxford: Clarendon Press, 1997).
Halberstam, Judith, Skin Shows: Gothic Horror and the Technology of Monsters
(Durham, N.C.: Duke University Press, 1995).
Heath, Stephen, The Work of Christian Metz, Screen Reader, 2, Cinema
and Semiotics (London: SEFT, 1981): 138161.
Questions of Cinema (London: Macmillan, 1981).
Klinger, Barbara, Melodrama and Meaning: History, Culture, and the Films of
Douglas Sirk (Bloomington: Indiana University Press, 1994).
168
BIBLIOGRAPHY
Lehman, Peter, Politics, Film Theory, and the Academy, Journal of Film
and Video, 40, 2 (1988).
Lowry, Edward, The Filmology Movement and Film Study in France (Ann
Arbor, Mich.: UMI Research Press, 1985).
Metz, Christian, Essais sur la signification au cinema (Paris: Klincksieck,
1968).
Film Language: A Semiotics of the Cinema, trans. Michael Taylor (New
York: Oxford University Press, 1974).
Problems of Denotation in the Fiction Film, Film Language: A Semiotics
of the Cinema, trans. Michael Taylor (New York: Oxford University
Press, 1974): 108146.
Language and Cinema, trans. Donna-Jean Umiker-Sebeok (The Hague:
Mouton, 1974).
On the Notion of Cinematographic Language, Movies and Methods, ed.
Bill Nichols (Berkeley: University of California Press, 1976): 582589.
Rapport sur letat actuel de la semiologie du cinema dans le monde
(debut 1974),A Semiotic Landscape/Panorama Semiotique, ed. Seymour
Chatman, Umberto Eco, and Jean-Marie Klinkenberg (The Hague:
Mouton, 1979).
The Imaginary Signifier, Psychoanalysis and Cinema: The Imaginary Signifier, trans. Ben Brewster et al. (London: Macmillan, 1982): 187.
Story/Discourse (A Note on Two Kinds of Voyeurism), Psychoanalysis
and Cinema: The Imaginary Signifier, trans. Ben Brewster et al. (London:
Macmillan, 1982): 9198.
Sur une traversee des Alpes Pyrenees . . . , Preface to Casetti, DUn
regard lautre: Le film et son spectateur (Lyon: Presses Universitaires de
Lyon, 1990): 510.
LEnonciation impersonnelle ou le site du film (Paris: Meridiens Klincksieck,
1991).
The Impersonal Enunciation, or the Site of Film, in The Film Spectator:
From Sign to Mind (Amsterdam: Amsterdam University Press, 1995):
140163.
Muscio, Giuliana, and Roberto Zemignan, Francesco Casetti and Italian
Film Semiotics, Cinema Journal, 30, 2 (Winter 1991): 2346.
Odin, Roger, Rhetorique du film de famille, Revue dEsthetique, 12
(1979): 340373.
Mise en phase, dephasage et performitive dans Le Tempestaire de Jean
Epstein, Communications 38 (1983): 213238.
Film documentaire, lecture documentarisante, Cinemas et realites, ed. J.
Lyant and Roger Odin (Saint-Etienne: Cierec, 1984): 263280.
Du spectateur fictionalisant au nouveau spectateur: approche semiopragmatique, Iris, 8 (1998): 121139.
La semio-pragmatique du cinema sans crise, ni desillusion, Hors Cadre,
7 (1989): 7792.
Cinema et production de sens (Paris: Editions Armand Colin, 1990).
Semio-pragmatique du cinema et de laudiovisuel: Modes et institu-
BIBLIOGRAPHY
169
tions, Towards a Pragmatics of the Audio-Visual, Vol. 1, ed. Jurgen

Muller (Munster: Nodus Publikationen, 1994): 3346.
Approche semio-pragmatique, approche historique, Kodikas/Code, 17,
14 (1994): 2736.
For a Semio-Pragmatics of Film, The Film Spectator: From Sign to Mind,
ed. Warren Buckland (Amsterdam: Amsterdam University Press,
1995): 213226.
A Semio-Pragmatic Approach to the Documentary Film, The Film Spectator: From Sign to Mind, ed. Warren Buckland (Amsterdam: Amsterdam University Press, 1995): 227235.
(ed.), Le film de famille (Paris: Meridiens Klincksieck, 1995).
Rodowick, David, The Difficulty of Difference: Psychoanalysis, Sexual Difference and Film Theory (London and New York: Routledge, 1991).
Simons, Jan, Cognitivism and Pragmatics, unpublished ms., 1991.
Pragmatics, Deixis, and the Political Election-Campaign Film, Towards
a Pragmatics of the Audio-Visual, Vol. 1, ed. Jurgen Muller (Munster:
Nodus Publikationen, 1994): 7792.
Film, Language, and Conceptual Structures. Thinking Film in the Age of
Cognitivism (Amsterdam: Academisch Proefschrift, University of Amsterdam, 1995).
Small, Edward, Introduction: Cognitivism and Film Theory, Journal of
Dramatic Theory and Criticism, 6, 2 (1992): 165172.
Souriau, Etienne, La structure de lunivers filmique et le vocabulaire de la
filmologie, Revue Internationale de Filmologie, 78 (1951): 231240.
Tan, Ed, Emotion and the Structure of Narrative Film: Film as an Emotional
Machine (Hillsdale, N.J.: Lawrence Erlbaum, 1996).
Index
Allen, Richard, 9394

Alphaville (Godard), 124
Althusser, Louis, 18, 27
Anderson, Joseph, 1, 3, 14
Apel, Karl-Otto, 15
Arnheim, Rudolf, 3, 6, 26
Austin, J. L., 148149n34
avant-garde, 8, 78, 124
Bala`sz, Bela, 26
Barefoot Contessa, The, 36, 50
Barthes, Roland, 59
Bazin, Andre, 3, 6, 26
Beaugrande, Robert de, 143144
behaviorism, x, 13, 14, 18, 27, 40
Bellour, Raymond, 3, 146n20
Benveniste, Emile, 3, 18, 27, 5254,
57, 63, 67, 69, 155n41
Bergstrom, Janet, 1
Berkeley, Busby, 62, 67
Bever, Thomas G., 137
Blakemore, Diane, 79
Blue Gardenia, The (Lang), 90, 107
Bordwell, David, ix, 1, 3, 414 passim, 18, 22, 27, 2932, 37, 38
39, 44, 49, 66, 8687, 90, 92,
100, 106, 107, 109, 130, 131
133, 142, 144, 145n4
Bosanquet, Bernard, 15
Bradley, Francis Herbert, 15
Branigan, Edward, 1, 3, 14, 90, 92
Bresnan, Joan, 137
Buhler, Karl, 6970

Burch, Noel, 96
Carnap, Rudolf, 15,
Carroll, John M., 25, 137140
Carroll, Noel, 1, 3, 4, 142, 144,
147n27
Casetti, Francesco, x, 1, 3, 19, 20, 23,
523, 5863, 6467, 73, 78
Chateau, Dominique, x, 1, 3, 19, 20,
25, 109, 120121, 126, 129, 134
Chatman, Seymour, 59
Chomsky, Noam, x, 8, 1822, 33,
41, 109, 113, 121, 122, 123,
125, 126, 129, 130, 135, 137,
140, 141, 142
see also Transformational Generative Grammar
classical film theory, 2, 3, 6, 26
Cobra, 104
cognitivism, x, 14, 2932, 142, 144
Colin, Michel, x, 1, 3, 19, 20, 21, 22,
2425, 27, 3239, 44, 50, 109
110, 113119, 135136, 140,
141, 159n20
colorization, 23, 53, 71, 105
Comolli, Jean-Louis, 3
Conley, Tom, 73, 7475
contemporary film theory, see modern film theory
Corti, M., 59
Currie, Gregory, 14
171
172
Daddesio, Thomas, 17
Dayan, Daniel, 70, 88
deconstruction, 73
de Man, Paul, 72
Derrida, Jacques, 19, 75
Descartes, Rene, 15, 16, 27, 32, 40
descriptive adequacy (Chomsky), x,
20, 21, 24, 111, 112, 113, 121,
135, 144
see also explanatory adequacy;
observational adequacy
documentary, 24, 78, 96, 98102,
107
dynamic mode of filmmaking
(Odin), 24, 78, 89, 104107
Eastman, Charles M. (general space
planner), 36
Eco, Umberto, 28, 9192
812 (Fellini), 52
Eisenstein, Sergei, 3
enunciation, 3, 19, 20, 2223, 5258,
61, 6365, 67, 68, 72, 78, 96
97, 99100, 142
explanatory adequacy, 20, 21, 144
see also descriptive adequacy;
observational adequacy
INDEX
grammar, x, 20, 38, 109113, 119

123, 134, 136
universal, 20, 21
see also Transformational Generative Grammar
Greimas, A. J., 83, 91, 92, 96, 99
Grodal, Torben, 3, 14
guerre est finie, La (Resnais), 131133
Habermas, Jurgen, 15, 1617, 21, 32,
78
Halberstam, Judith, 7374
Harris, Roy, 8182
Heath, Stephen, 3, 28, 46, 47, 50,
100, 148n27, 152n50
Hegel, Georg Wilhelm Friedrich, 15
home movies, 24, 70, 78, 89, 102
104
Hudson, Rock, 8788
implied author, 32, 158n74
institutions, see Odin, institutions
Iser, Wolfgang, 59, 108
Jackendoff, Ray, 38, 133134, 140
Johnson, Mark, 22, 27, 36, 3947
Fauconnier, Gilles, 39
femme mariee, Une (Godard), 116
Fichte, Johann Gottlieb, 15
film language, 5, 8, 10, 12, 13, 120
121
Fodor, Jerry, 33
modular theories of the mind, 37
38
Frege, G., 15, 16,
French New Wave, 123125, 140
Kant, Immanuel, 15, 19

Kasher, Asa, 79
Katz, Elihu, 70, 88
Katz, Jerrold J., 33, 121122, 130
Kid from Spain, The (Leo McCarey),
6163
King of Marvin Gardens, The (Bob
Rafelson), 60
Klinger, Barbara, 87
Kracauer, Siegfried, 3, 26, 100
Kuhn, Thomas, 144
Gaudreault, Andre, 94, 97

Gazdar, Gerard, 137
Genette, Gerard, 9293
Gombrich, E. H., 106
Lacan, Jacques, 18, 27, 58, 80

Lakoff, George, 22, 26, 27, 36, 39
47, 50, 70
Lanacker, Ronald, 39
INDEX
Language Analysis tradition, ixx,

14, 1517, 27, 32, 141
Last Year in Marienbad (Resnais),
124
Lehman, Peter, 4
Lerdahl, Fred, 133134, 140
Locke, John, 81
Lotman, Jurij, 59
Lye, Len, 8
MacCabe, Colin, 3
Mad Max series, 104
Malinowski, Bronislaw, 77
Marker, Chris, 8
Martinet, Andre, 21
metaphor, 26, 36, 38, 40, 42, 43, 51
metonymy, 26, 36, 38, 40, 42, 43
Metropolis (Moroder version), 89,
104106
Metz, Christian, ix, x, 2, 3, 4, 515,
1819, 24, 26, 27, 32, 44, 79
80, 84, 86, 88, 97, 109110, 119
120, 135, 140, 146147n22
Cinema: Language or Language System? 12, 19, 28,
55, 75
grande syntagmatique, 4, 8, 1112,
13, 19, 21, 22, 24, 33, 69, 79,
110, 113119, 126, 129130,
132, 133, 135, 136, 137, 159n16
Imaginary Signifier, The, 18,
19, 75, 80
Language and Cinema, 5, 8, 13, 19,
28, 59, 7980, 121
Modern Cinema and Narrativity, 123126
theories of filmic enunciation,
19, 23, 5259 passim, 6369,
7273
Minkovski, Eugene, 45
modern film theory, ix, 2, 3, 13, 14,
57, 145n3
modes, see Odin, modes
173
Moore, George, E., 15

Morris, Charles, x
mutual knowledge hypothesis, 35
36, 81
narration, 5, 32
narrative, 2930, 50, 9293, 102, 123
126
narrator, 32, 49, 66
Norris, Christopher, 18
Nuyts, Jan, 79
observational adequacy (Chomsky),
x, 2021, 24, 111, 135, 144
see also descriptive adequacy; explanatory adequacy
Odin, Roger, x, 1, 3, 19, 2324, 76,
108, 132, 156n11
institutions, 24, 7778, 83, 87, 97
98, 107
modes, 24, 78, 80, 83, 85, 8891,
97, 98107
operations, 24, 78, 82, 83, 89, 91
97, 98, 107
operations, see Odin, operations
Partie de campagne (Renoir), 95
Peirce, C. S., ix, 15, 16, 17, 39
Perkins, Victor, 26
Pierrot le fou (Godard), 124130,
131, 132, 133
poetics, 5
post-theory, 142144
pragmatics, x, 2, 16, 19, 2324, 26,
60, 7782, 103, 107108, 141
Prince, Gerard, 59
principle of relevance, see Sperber,
Dan, and Deirdre Wilson
Psycho (Hitchcock), 86, 90, 100, 107,
138
psychoanalysis, 2, 4, 18, 19, 27, 51,
58, 80, 86, 93, 97
174
reflexivity, 5256, 68
Ricoeur, Paul, 15
Riso Amaro (Guiseppe de Santis),
60
Rocky series, 104
Rodowick, David, 7374, 75
Russell, Bertrand, 15
Russian Formalism, 11
Ryle, Gilbert, 15, 16, 34
Saussure, Ferdinand de, ix, 9, 11,
15, 39, 81
arbitrary relation between signifier and signified, 28
langue/parole, 8, 10, 11, 24, 5657,
79, 143
metaphor of language as a
speech circuit, 8182
syntagmatic/paradigmatic, 11
schemata, 26, 2931, 4047, 48, 50
Searle, John, 93, 96, 148149n34
Sebeok, Thomas, 141, 142
the biosemiotic self, 151152n39
semantics, x, 26, 3233, 3944, 77,
79, 81, 82, 8384
semiotic theory, 2, 517, 142
codes, 10, 28, 81, 82
commutation test, 11, 12
perceptible/non-perceptible
(surface/underlying) hierarchy, 811, 2122, 54, 111
semiotic modelling, 79, 25
see also Saussure
Silence of the Lambs, The (Jonathan
Demme), 7374
INDEX
Simons, Jan, 50, 90, 107, 153n55,

156n11
Sinclair, Melina, 79
Sirk, Douglas, 8788
Skinner, B. F., 18
Small, Edward, 147148n27
Smith, Murray, 3
Soames, Scott, 113, 136137, 140
Souriau, Etienne, 47
Sperber, Dan, and Deirdre Wilson
(principle of relevance), 30,
3436, 79, 8485, 98, 130131,
133, 134
Spielberg, Steven, 49
Spottiswoode, Raymond, 13
Star Wars (Lucas), 104
suture, 50, 152n50
Sweester, Eve, 39
syntax, see grammar
Tan, Ed, 3, 14
television, 23, 68, 7071
Tempestaire, Le (Epstein), 9495
Transformational Generative Grammar (TGG), x, 3, 18, 19, 20,
41, 109115, 119, 121122,
126, 135137, 141 142
see also Chomsky
translinguistics, 2, 5, 13, 16, 23, 137
Tron (Lisberger), 104
Weber, Samuel, 9
Winograd, Terry, 33
Wittgenstein, Ludwig, ix, 15
Worth, Sol, 109, 120

Buckland - Cognitive Semiotics of Film

Uploaded by

Copyright:

Available Formats

Buckland - Cognitive Semiotics of Film

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Buckland - Cognitive Semiotics of Film

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

The Cognitive Semiotics

For Thomas Elsaesser

The Cognitive Turn in Film Theory

We have witnessed a number of attempts to by-pass [film theorys]

uring the eighties, film studies gradually adopted

THE COGNITIVE SEMIOTICS OF FILM

contrast in their work: Whereas the North American cognitivists

THE COGNITIVE TURN IN FILM THEORY

1. CLASSICAL FILM THEORY

COGNITIVE FILM THEORY

David Bordwell, Nol Carroll, Edward Branigan, Joseph

tions held by cognitive film semioticians and clarify their relation

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

acknowledgement in Chapter 2 of the same book when he asks the

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

in the history of art to record reality without the intervention of

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

1. Iconicity; 2. mechanical duplication; 3. multiplicity; 4. movement;

THE COGNITIVE SEMIOTICS OF FILM

structure to the perceptible, a system of codes. One of the integral

THE COGNITIVE TURN IN FILM THEORY

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

The Language Analysis Tradition

THE COGNITIVE SEMIOTICS OF FILM

model provided by grammatical rule systems; and it overcomes the

THE COGNITIVE TURN IN FILM THEORY

The epistemologists assumptions of immediate access to the

THE COGNITIVE SEMIOTICS OF FILM

Noam Chomsky and the Study of Competence

THE COGNITIVE TURN IN FILM THEORY

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

quacy, because it merely segments and classifies linguistic behavior

THE COGNITIVE SEMIOTICS OF FILM

of deep structures. These are then transformed into a potentially

THE COGNITIVE TURN IN FILM THEORY

THE COGNITIVE SEMIOTICS OF FILM

THE COGNITIVE TURN IN FILM THEORY

features that combine to form these syntagmatic types. Chapter 5

The Body on Screen and in Frame

Reason is embodied in the sense that the very structures on which

THE BODY ON SCREEN AND IN FRAME

language is structured around, and gains its meaning from, external

THE COGNITIVE SEMIOTICS OF FILM

Disembodied Structures and Schemata in Film Theory:

THE BODY ON SCREEN AND IN FRAME

and Cinema Metz defined iconicity in terms of codes).6 Metz began

THE COGNITIVE SEMIOTICS OF FILM

In constructing a fabula, the perceiver defines some phenomena as events

THE BODY ON SCREEN AND IN FRAME

THE COGNITIVE SEMIOTICS OF FILM

THE BODY ON SCREEN AND IN FRAME

identifying semantics as specifically linguistic (the trap, Katz and