Friday, July 20, 2007

Historical Poetics of Cinema by David Bordwell

Historical Poetics of Cinema
by David Bordwell


--------------------------------------------------------------------------------

From The Cinematic Text: Methods and Approaches, ed. R. Barton Palmer (New York: AMS Press, 1989), pp. 369-398.




1. Introduction

The volume you hold in your hands belongs to a genre that came into currency during the postwar boom in college literary criticism. In the late 1940s, two major works, Wellek and Warren's Theory of Literature and Stanley Edgar Hyman's The Armed Vision, set forth the premise that literary studies played host to distinct "methods." Intrinsic/ extrinsic; textual/ contextual; sociological/ psychological/Marxist/psychoanalytic/archetypal/formalist/ deconstructionist/ reader-responsiveness: as such categories have accumulated over the last forty years, the literary institution has believed itself moving beyond the doctrines of New Criticism. While the field was being carved up methodically, the "anthology of approaches" moved into the library and the classroom. The genre became a going concern in the 1950s and 1960s, and it continues to flourish. This volume reminds us that film studies has, as part of its entry into the academy, come to subscribe to critical Methodism--an affiliation testified to by the title of one of the most popular anthologies, Movies and Methods. A recent collection of approaches to television criticism may signal the legitimation of tv studies under the same auspices.

There would be much to remark on in this process, not least the extent to which film and television studies may seek to establish their seriousness by asserting that, whatever the intrinsic importance of the object of study, a set of up-to-date approaches constitutes adequate credentials. But I sketch this institutional background for another reason: to establish that historical poetics does not grow organically out of this history, and this for a very good reason. What I shall be discussing is not a method at all.

In film studies, as in its literary counterpart, "method" has been largely synonymous with "interpretive school." An interpretive school, I take it, consists chiefly of: (a) a semantic field with which particular theoretical concepts are associated; (b) a set of inferential procedures that render certain features of films salient and significant on a priori grounds; (c) one or more conceptual maps of textual progression across which salient features enact a transformation of the semantic field; (d) a set of characteristic rhetorical tactics for setting forth the writer's argument. For example, the psychoanalytic critic posits a semantic field (eg, male/ female, or self/ other, or sadism/ masochism) with associated concepts (eg, the deployment of power around sexual difference); concentrates on textual cues that can bear the weight of the semantic differentiae (eg, narrative roles, the act of looking); traces a drama of semantic transformation (eg, through condensation and displacement the subject finds identity in the Symbolic); and deploys a rhetoric that seeks to gain the reader's assent to the interpretation's conclusions (e.g., a rhetoric of demystification). Every recognized "method"--phenomenological, feminist, Marxist, or whatever- can be described along these lines. They all aim to produce interpretations--that is, the ascription of implicit or symptomatic meanings to texts.

A historical poetics of cinema does not fit this description. It does not constitute a distinct critical school; it has no privileged semantic field, no core of procedures for identifying or interpreting textual features, no map of the flow of meaning, and no unique rhetorical tactics. It does not seek to produce interpretations. What, then, does it do?


2. Poetics and Historical Poetics

Aristotle's fragmentary lecture notes, the Poetics, addressed what we nowadays recognize as drama and literature. Since his day we have had Stravinsksy's Poetics of Music, Todorov's Poetics of Prose, a study of the poetics of architecture, and of course the Russian Formalists' Poetics of the Cinema. Such extensions of the concept are plausible, since it need not be restricted to any particular medium. "Poetics" derives from the Greek word poiesis, or active making. The poetics of any medium studies the finished work as the result of a process of construction--a process which includes a craft component (e.g., rules of thumb), the more general principles according to which the work is composed, and its functions, effects, and uses. Any inquiry into the fundamental principles by which a work in any representational medium is constructed can fall within the domain of poetics.

By adding the predicate "historical" I mean to narrow the field somewhat. Poetics of literature has long been the province of sterile taxonomies and dogmatic prescriptions. In the twentieth century, German-language art studies and Slavic literary theory laid the groundwork for a historical poetics. Heinrich Wolfflin, Alois Riegl, Erwin Panofsky, and later E. H. Gombrich showed how one could systematically describe forms and styles in painting and go on to explain their changes causally. The Russian Formalists--most notably Viktor Shklovsky, Yuri Tynianov, Boris Eikhenbaum, and Roman Jakobson--and the Prague Structuralists--e.g., Jan Mukarovsky and (again) Jakobson--proposed both concrete analyses of literary works and larger explanations for how they functioned in historical contexts. This tradition has been alive in film studies as well, crossing periods and doctrinal schools and recently emerging as a significant force in academic work.

A historical poetics of cinema produces knowledge in answer to two broad questions:

1. What are the principles according to which films are constructed and by means of which they achieve particular effects?

2. How and why have these principles arisen and changed in particular empirical circumstances?

Historical poetics is thus characterized by the phenomena it studies--films' constructional principles and effects--and the questions it asks about those phenomena--their constitution, functions, consequences, and historical manifestations. Poetics does not put at the forefront of its activities phenomena such as the economic patterns of film distribution, the growth of the teenage audience, or the ideology of private property. The poetician may need to investigate such matters, and indeed many others, but they become relevant only in the light of more properly poetic issues. Underlying this hierarchy of significance is the assumption that, while in our world everything is connected to everything else, one can produce novel and precise knowledge only by making distinctions among core questions, peripheral questions, and irrelevant questions.

Andre Bazin's "Evolution of the Language of the Cinema" can illustrate how a project within historical poetics works. The essay relies upon some fundamental conceptual distinctions, such as inter-shot effects vs. intra-shot effects, types of montage, distortion vs. fidelity, spatiotemporal unity vs. discontinuity, shallow space vs. depth. Bazin holds these to be principles determining the stylistic construction of any film whatsoever. They yield categories which permit the analyst to correlate devices with particular effects--eg, a linearization of meaning with "visible" montage versus a more natural conveyance of meaning through Welles' or Wyler's depth of field. Furthermore, Bazin offers a historical account which employs these categories to trace the development of Western cinema from primitive filmmaking to Neorealism. Bazin argues for a dialectical movement whereby the struggle between a realistic style and a more distorted style reached a compromise in the Hollywood decoupage of the 1930s, and then was transcended by a new synthesis in the deep-focus style.

While Bazin's remains the most influential history of film style currently available, I am not here concerned with the persuasiveness of his argument; the point is to show how the essay exemplifies the possibilities of a poetics. For one thing, it self-consciously constructs its analytical categories while also referring to a range of data by which the arguments can be assessed. Bazin supplies concrete historical evidence that subjects his claims to revision or rejection. This appeal to empirical evidence, or "facts," does not make poetics an "empiricism," at least, in any interesting sense of that term. A poetics can be rationalist or empiricist, Kantian or phenomenological, deductivist or inductivist, idealist (as Bazin probably was) or positivist (as Barry Salt seems to be). Insofar as a poetics seeks to explain historically existing works, whatever its ontology or epistemology or discovery procedures, it requires an appeal to intersubjectively accepted data which are in principle amenable to alternative explanation. Just as in the philosophy of language, a Chomskyan nativist must confront the fact that people seem to acquire significant aspects of language through experience, so even the post-Structuralist film theorist must recognize the existence of apparent motion or characterization or editing. Every enterprise within film study necessarily draws upon facts in this sense. Whether such facts are "socially constructed" is an open question. (Indeed, it is partly an empirical question.)

Bazin's "Evolution" essay also illustrates the extent to which a poetics takes as its object a body of conventions. Conventions, in film as in other domains, lie at the intersection of logical distinctions and social customs. Admittedly, Bazin's realist aesthetic leads him to range stylistic devices along a continuum whereby some are less "conventional" than others. Nevertheless, he is studying constructive choices which have collectively recognized functions within definable contexts; editing and deep-focus, he argues, constitute reciprocal choices in the history of Western cinema. These conventions can be regarded as leading to preferred choices--that is, norms or rules, two more concepts valuable for a poetics.

Bazin's essay exemplifies still other aspects of a cinematic poetics. He refuses a division of labor among theory, history, and criticism; the essay is all three at once. It mixes intensive scrutiny with extensive viewing. Bazin considers both "texts" and "contexts" (technology, genre). He offers descriptions, analyses, and explanations: he seeks to establish what happened, how it happened, and why it happened. Finally, Bazin presupposes that the phenomena he studies are the results of filmmakers' choices. (Welles could have cut Citizen Kane as if it were It Happened One Night.) A historical poetics will thus often be concerned to reconstruct the options facing a filmmaker at a given historical juncture, assuming that the filmmaker chooses an option in order to achieve some end.

My initial questions and my exposition of Bazin should raise several questions. What, for example, is the status of the "principles" studied by poetics? At what level of generality are they pitched? Are they conceived as "specific" to cinema in some sense? My replies here must be brief. I would argue that the principles should be conceived as underlying concepts, constitutive and/or regulative, governing the sorts of material that can be used in a film and the possible ways in which it can be formed. The degree of generality will depend upon the questions asked and the phenomena to be studied. If you want to know what makes Hollywood narratives cohere, "personalized causality" may suffice as one constructive principle; if you want to know what distinguishes the Western from the musical, that principle will not do the job. For some poeticians, some principles are held to be "laws" on the model of covering laws in physical science; but one need not push this far. One could assert that a concept or category--eg, intra-shot/ inter-shot relations--is conceptually stable but that the constructive principles that utilize it are so historically variable that they constitute empirical generalizations or tendencies.

On the matter of specificity, suffice it to say that although certain poeticians have assumed a distinction between the cinematic and the non-cinematic, this is by no means constitutive of poetics as such. One could assume that any film could be studied by poetics, with no film lying any closer to the essence of the medium than others. One could, however, argue that the distinction is not a substantive but a functional one, to be filled out in different periods with different content. Or one could use the cinematic/ non-cinematic distinction in an explanation by seeking to show that in particular circumstances this pair of concepts entered into the norms of filmmaking practice.

Since poetics is often assumed to be merely a descriptive or classificatory enterprise, the range of explanations it offers also requires some elaboration. There is no need to assume any one model of causation and change. Bazin argues for a suprapersonal dialectic through which cinema evolves toward an ever more faithful capturing of phenomenal reality. This is a teleological explanation. One could also propose an intentionalistic model that centers on more localized acts of choice and avoidance. Two collaborators and I have argued for a functionalist model of explanation, whereby the institutional dynamics of filmmaking set up constraints and preferred options that fulfill overall systemic norms.

Nor need poetics confine itself to "immanent" explanations that refuse to leave the field of cinema or art or signifying practice. Nothing prevents the poetician from arguing that economics, ideology, the class struggle, or inherent social or psychological dispositions operate as causes of constructional devices or effects. There is likewise no need to cast poetics as offering "scientific" explanations (although, again, some poeticians have done so). Poetics has the explanatory value of any empirical discipline, which always involves a degree of tentativeness about conclusions. On the other hand, one should not discard "scientific" pretensions too quickly, since there are many sorts of science, such as geography and meteorology, which are low in predictive power but high in ex post facto explanatory power. Poetics can, in short, be considered either as a science offering knowledge in some strict sense, or as a discipline that aims at Verstehen, or "understanding."

Finally, explanation in poetics does not confine itself to issues of what films mean. Of course, meaning in one (very general) sense comprises a big part of what poetics describes, analyzes, and explains; but meaning in the more specific sense that is the product of film interpretation ("readings") can be considered only one domain of inquiry. Films produce many effects, ranging from perceptual ones (eg, the phenomenon of apparent motion, the perception of color or shape) to conceptual ones (eg, how we we know that X is the protagonist) that film interpretation never seeks to explain. Historical poetics offers explanations, not explications. In the modern critical institution, of course, explications need explaining.


3. Domains and Tendencies

The core questions of poetics have led to the formulation of distinct domains and tendencies within the field. Traditional poetics distinguishes among three objects of study.

1. Thematics considers subject matter and theme as components of the constructive process. Not every study of such matters automatically qualifies as a contribution to poetics, since many such studies are unconcerned with films' constructive principles; the film may be ransacked for discrete items of "content" (eg, the representation of social stereotypes) which are then used to answer questions about, say, the film industry or cultural values. Thematics would study motifs, iconography, and themes as materials, as constructive principles, or as effects of constructive principles. For example, several scholars have revealed how genres present recurring imagery, myths, and themes, while other writers, inspired by art-historical research, have shown the importance of iconography in popular cinema.

2. Constructional form. We lack a term for those trans-media architectonic principles that govern the shape and dynamics of a film. The most prominent research domain here is the theory and analysis of narrative, which is a fundamental constructive principle in films. Current interest in this subject should not, however, lead to a neglect of other compositional principles, such as argumentative form, categorical form, associational form, and abstract form.

3. Stylistics deals with the materials and patterning of the film medium as components of the constructive process. Bazin's "Evolution" essay is a model of stylistic history.



We could carve up the domain of historical poetics in still other ways. Following R. S. Crane, we could distinguish studies of precompositional factors (sources, influences, cliches, received forms) from compositional ones (normalized principles of combination and transformation within works) and from postcompositional ones (effects, reception, varying responses in different contexts). For example, Noel Burch's To the Distant Observer treats Japanese cinema as the legatee of stylistic practices from earlier centuries, while Vance Kepley's In the Service of the State, using a different precompositional focus, traces more proximate influences on Dovzhenko's films. The work of Charles Musser, Tom Gunning, and Andre Gaudreault has demonstrated that pre-1915 films obey coherent compositional norms. And recent work in reception by Janet Staiger has revealed how audiences' varying construals of the same film presuppose historically variable viewing conventions. In my own studies of Dreyer and Ozu, I have tried to relate the three domains by suggesting historically determinate gaps among them. In the works of Ozu, for instance, source material and contemporary conventions are transformed by specific compositional procedures, but the results have been appropriated differently by various audiences.

Recognizing that linguistic analogies are notoriously shaky in film studies, I will risk one more mapping of the field. Like linguistics, film poetics has its "semantics," the study of how meaning is produced. It has its "syntactics," the study of rules for selecting and combining units (with respect to style, Raymond Bellour's micro-analyses; with respect to compositional form, Thierry Kuntzel's study of openings, Peter Wollen's applications of Propp, or Rick Altman's "dual-focus" narrative). And poetics has its "pragmatics," the study of how relations between viewer and text develop in the process of the film's unfolding (e.g., accounts of narration or of filmic "enunciation"). Meaning, structure, and process--these three aspects of any representational system are also central to poetics.

These equable mappings of the terrain conceal, of course, how much territory is in dispute. I have already suggested several issues about which poeticians wrangle; two more divergences seem to me worth brief discussion.

Across history, poetics has had to steer a course between strictly "immanent" accounts and strictly "subsumptive" ones. Few poeticians have been willing to accept the consequences of an utterly intrinsic account of constructional processes; even Wolfflin, mistakenly treated as the model of the pure formalist, explained changes in artistic styles partly by changes in a culture's visual habits. On the other side, very few poeticians have sought to account for every phenomenon by appeal to processes in other social domains; even the Zhdanovite recognizes some special quality in art. For most poeticians, the constructional principles studied are not self-sealed, but they are also not in every respect subsumable to other principles.

Assuming that the escape hatch of "relative autonomy" is of no help, we can distinguish two tendencies within poetics. One tendency hypothesizes that the phenomenon we study has a considerable degree of self-regulated coherence. The early Shklvosky seems to hold this view; he seeks to explain the laws of fairy tale composition by purely poetic principles like repetition, retardation, and so forth. He gives theoretical priority to such factors. In film poetics, perhaps Burch's Theory of Film Practice approaches this position. The second tendency, articulated by the later Russian Formalists and the Prague Structuralists, gives immanent factors only a methodological priority. For example, as Tynianov and Jakobson point out, even if the immanent evolution of literature can explain the direction of change, it cannot explain timing, which must be governed by extraliterary causes. A comparable position is taken by Staiger, Thompson, and myself in studying the history of the classical Hollywood cinema. Here the analyst looks first to the "immanent" factors that might be the most proximate and pertinent causal factors but also assumes that virtually every explanatory task will require moving to those mediations that lie in "adjacent" domains.

To continue the geographical metaphor, poetics is less a field with distinct boundaries than a kind of Alsace-Lorraine constantly being claimed by interested neighbors. On one side is Aesthetics, which, in the eighteenth century, replaced the study of poetic praxis with a concern for the philosophical problems involved in the creation and appreciation of beauty. On another side lies Semiotics, which seeks to subsume poetics into a general theory of the production of meaning. Interestingly, poeticians have been drafted into both camps. Aristotle, the Russian Formalists, and the Prague Structuralists can play roles in the history of aesthetics, as in Beardsley's survey history, or they can be promoted to the rank of proto-semioticians, as Peter Steiner does.

In my view, the tension between semiotics and aesthetics has been immensely fruitful. There remains, however, a core of questions and issues that cannot be wholly absorbed into the adjacent areas. It is useful to differentiate between the practical theory of an art and the philosophy of it. The "practical theory" of music or poetry, for instance rests upon a posteriori questions, involving empirical generalizations about conventions and practices in these arts. From this perspective, film poetics is a systematizing of theoretical inquiry into cinematic practices as they have existed. The philosophy of an art, on the other hand, inquires into the a priori aspects of it; it involves conceptual analysis of its logical nature and functions. On the whole, aesthetics concentrates upon such matters. As for semiotics, it concentrates on matters of meaning, which is only part of the effects for which a poetics seeks to account; on the other hand, if semiotics seeks to explain "the life of signs in society," it encompasses far more than any poetics can. Yet one should not discourage border crossings; if Barthes' S/Z offers a semiotics and Goodman's Languages of Art offers an aesthetics, both are splendid contributions to poetics.<



4. Neoformalism

One trend within the domain of historical poetics has been dubbed by Kristin Thompson "neoformalism." It is associated with research she and I have done over the past dozen years or so. The trend derives principally from Slavic poetics, particular the Russian and Czech thinkers, but it is also influenced by the more or less oblique "return to Slavic theory" one finds in Todorov, Genette, the 1966-1970 Barthes, and contemporary Israeli poeticians like Meir Sternberg. It draws heavily upon the writings of Bazin, the Soviet filmmakers, and Burch, without being committed to a "phenomenological" or "materialist" or "serialist" theory of film. In fact, neoformalism is not a theory of film at all, if we take that to consist of a set of propositions explaining the fundamental nature and function of all cinematic phenomena.

Neoformalism has even less in common with what has been called "Grand Theory," that development in the humanities that has embraced ever more wide-ranging intellectual programs. Under these auspices, the study of film has become "only a part" of the theory of ideology or of sexual difference or, most abstractly, of "the human subject." The principal issue here is not whether there is something "inherently filmic" that must be addressed, for, as mentioned above, the specificity of cinema may be conceived as more social and functional than substantive. The point is that concepts constructed at this level of generality and abstraction are not well suited to answering questions pitched at lower levels. Neoformalism, which addresses the latter sort of questions, is thus not a general theory of film, let alone a Grand one.

Nor is it, once again, a method. It is a set of assumptions, an angle of heuristic approach, and a way of asking questions. It is frankly empirical and places great emphasis on the discovery of facts about films. Since recent film theory usually claps the word "fact" within sneer-quotes, my claim is apt to seem a recourse to naive empiricism; but this conclusion, already jumped to by one writer, is itself naive. As I indicated above, any poetics---indeed, any descriptive or explanatory project--is committed to some grounding in intersubjective data. Furthermore, one can consider a fact to be an accepted claim about what there is in the world, including theoretical or unobservable entities--something that positivism rules out. Moreover, there is no question of letting the facts speak for themselves. Neoformalism presumes that one cannot discover factual answers to questions about films' construction without carefully devising analytical concepts appropriate to these questions. But it also assumes that not all concepts are equally precise, coherent, or pertinent, and so we may evaluate competing conceptual schemes; it also assumes that not all concepts explain the data with equal clarity, richness, and economy; and, crucially, it assumes that we are not complete prisoners of our conceptual schemes, that we may so construct them that anomalous and exceptional phenomena are not invisible but actually leap to our notice. In sum, neoformalist poetics makes theoretically-defined, open-ended, corrigible, and falsifiable claims.

This is a direct result of its not being a general theory of film. If I am bent on substantiating the belief that every film constructs an ongoing process of "subject positioning" for the spectator, nothing I find in a film will disconfirm it. Given the roomy interpretive procedures of film criticism, I can treat every cut or camera movement, every line of dialogue or piece of character behavior, as a reinforcement of subject positioning. The theory thus becomes vacuous, since any theory that explains every phenomenon by the same mechanism explains nothing. On the other hand, I can ask how Hollywood films secure unity among successive scenes, and answer with something more concrete--say, that one scene often ends with an unresolved causal sub-chain that is soon resolved in the following scene. Here I have said something that is informative: it is not self-evident, it is not discoverable by deduction from a set of premises, and it is fruitful, leading to further questions. (Does this suggest some hypotheses about the nature of narrational norms in Hollywood? Do films in other filmmaking traditions utilize more self-contained episodes?) Most important, the answer I supply is empirically disconfirmable. If it is disconfirmed, I need to rethink the data and indeed, the question itself. Shklovsky's counsel of skepticism should be our guide: "If the facts destroy the theory--so much the better for the theory. It is created by us, not entrusted to us for safekeeping." Neoformalism's hypotheses are grounded in a theoretical activity rather than a fixed theory. This theorizing moves across various levels of generality and deploys various concepts and categories. It does not presume global propositions to which the researcher pledges unswerving allegiance and which automatically block our noticing recalcitrant data.

In being question-centered and focused on particular phenomena, neoformalism does resemble the practices of science as many theorists are coming to understand them. Stephen Jay Gould writes:

Progress in science, paradoxically by the layman's criterion, often demands that we back away from cosmic questions of greatest scope (anyone with half a brain can formulate 'big' questions in his armchair, so why heap kudos on such a pleasant and pedestrian activity?). Great scientists have an instinct for the fruitful and the doable, particularly for smaller questions that lead on and eventually transform the grand issues from speculation to action....Great theories must sink a huge anchor in details.

This is not to grant neoformalism the status of a science, only to suggest that as compared with Grand Theory, its approach and spirit are closer to certain scientific practices. It is in this frame of reference we can best understand Boris Eikhenbaum's defense of the Russian Formalist group:

In our studies we value a theory only as a working hypothesis to help us discover and interpret facts; that is, we determine the validity of the facts and use them as the material of our research. We are not concerned with definitions, for which the late-comers thirst; nor do we build general theories, which so delight eclectics. We posit specific principles and adhere to them insofar as the material justifies them. If the material demands their refinement or change, we change or refine them.

With no set point of arrival, committed to no a priori conclusions, seeking to answer precisely posed question with concepts that will be refined through encounter with data, neoformalism deploys "hollow" categories. While the "Oedipal trajectory" or "looking equals power" carry interpretable meaning whenever they appear, other concepts mark out fundamental constructive principles that have effects but not a priori meanings. An instance of such a "hollow" principle that of norms. The neoformalist assumes that every film may be placed in relation to sets of transtextual norms. These operate at various levels of generality and possess various degrees of coherence. For instance, in most studio-made narrative films, the credits sequence characteristically occurs before the first scene, but it may also, as a lesser option, occur after a "pre-credits sequence." Such norms, while "codified," are not reducible to codes in the semiotic sense, since there is no fixed meaning attached to one choice rather than the other. And no particular meaning automatically proceeds from Godard's decision, in Detective, to salt the credits throughout the first several scenes.

A great deal of theorizing about norms remains to be done. (Are there, for instance, fruitful distinctions among convention, norm, and rule?) But even at this stage neoformalist poetics has put forward fairly detailed and comprehensive accounts of norms of narration and style in Hollywood cinema, "art cinema," Soviet montage cinema, and other modes. These are not definitive analyses; they are attempts to chart the range of constructional options open to filmmakers at various historical conjunctures, and the results are always open to revision. At this point, however, several analytical concepts seem well-founded. For example, Neoformalist poetics has established the usefulness of distinguishing between stylistic or narrative devices (e.g., the cut or the motif) and systems (e.g., spatial continuity or narrative causality) within which they achieve various functions. Establishing a unified locale is a function which different devices and different systems have fulfilled in various ways across history. But even this function is not historically invariant. (Some norms do not make unity of locale a salient feature.) In practical research terms, neoformalism's emphasis on historically changing norms, devices, systems, and functions requires that the analyst complement the scrutiny of single films by studying a wide range of films.

An orientation toward transtextual norms allows the analyst to be sensitive to the abnormal. Neoformalist poetics has been especially interested in how, against a background of conventions, a film or a director's work stands out. Kristin Thompson has been concerned to demonstrate how the works of Eisenstein, Ozu, Tati, Godard, Renoir, and others provide not wayward deviations from norms but rather systematic innovations in thematic, stylistic, and narrative construction. Neoformalism balances a concern for revealing the tacit conventions governing the ordinary film with a keen interest in the bizarre film that, subtly or flagrantly, challenges them. Accordingly, new concepts will often have to be forged. To account for Ozu's editing, Thompson and myself had to propose the concept of the "graphic match" and to spell out how Ozu's across-the-line shot/ reverse-shots do not sporadically transgress rules but rather achieve perceptual functions within a larger, idiosyncratic system of 360-degree space.

The construction of concepts in accord with empirical data leads to historical explanations for the phenomena in question. Neoformalist poetics has relied upon three explanatory schemes, adjusted to cases at hand: a rational-agent model, an institutional model, and a perceptual cognitive model. The first follows from the concept of the filmmaker's choosing among constructional options. Here the task becomes that of reconstructing, on the basis of whatever historical data one can find, the choice situation which the filmmaker confronts. This is not to say, however, that the filmmaker becomes the sole source of the film's construction and effects.

The institutional dimension--most proximately, the social and economic system of filmmaking, involving tacit aesthetic assumptions, the division of labor, and technological procedures--forms the horizon of what is permitted or encouraged at particular moments. It is not just that the filmmaker's choices are constrained; they are actively constituted in large part by socially structured factors. In the Hollywood studio system of the 1920s and 1930s, for instance, the continuity script not only became a way to rationalize production but also encouraged workers to think of a film as assemblage out of discrete bits (shots, scenes), and the individual filmmaker found choices and opportunities structured accordingly. By the same token, an institution centered conception would seem the most promising basis for the study of how spectators use appropriate films in different historical contexts (though I would argue that the "microfoundations" of such a study would have to include some rational-agent assumptions).

Most recently, a perceptual-cognitive model has been used to describe and explain the effects of various constructional tactics. I have proposed that a Constructivist theory of psychological activity yields the most discriminating and detailed explanation of such narrational principles as syuzhet/fabula construction and such stylistic processes as continuity editing. In a work in progress, I consider how the routine practices of film interpretation can be partly explained through principles of inference and problem-solving set forth by cognitive theory. In all cases, the models are not absolute; the neoformalist poetician does not treat individual phenomena as instantiating laws but rather as demanding an inferential argument "to the best explanation," which always remains in principle corrigible.

Some discussions of Godard's Sauve qui peut (la vie) provide a convenient occasion to exemplify the richness of the neoformalist approach. For example, in his review of the film Colin MacCabe followed the conventional journalistic format: teaser (description of the notorious "sex machine" scene), one-paragraph plot synopsis, background on Godard's career (over a page), mention of themes (town vs. country, prostitution, masculine vs. feminine), discussion of "form" (the image track dominates the soundtrack), reflections on the author as person (Godard's dissatisfaction with the familiar "economic and aesthetic constraints," as confided to MacCabe), and a final, unexplained invocation of "the exhilaration of actually watching the movie." As in his contemporaneous book on Godard, MacCabe's discussion relies upon a straightforward thematics (Godard's fetishization of woman) and an "empty" formalism (e.g., the celebration, in all contexts, of moments when sound dominates image).

Or, to take a more substantial example, consider the articles on Sauve qui peut (la vie) gathered in Camera Obscura. Although more detailed than MacCabe's, they are plagued by errors of description; several also make some questionable assumptions (e.g., that the protagonist Paul Godard stands for the director, or that all cuts "cannot be seen"). One essay, by perhaps the most influential textual analyst, describes the film's editing this way:

The images seem to hit each other, musically, pictorially, striking each other admirably and thus making impossible any continuity of movement which would produce, the moment it is a question of bodies and of the sexes, an imaginary ideality that has simply ceased to exist.

It must be recalled that 1970s film theory never tired of attacking exactly such writing as "impressionistic."

In its commitment to explaining difficult films precisely, neoformalist poetics offers signal advantages. Kristin Thompson's lengthy analysis of Sauve qui peut (la vie) sets out to answer some specific questions: what makes the film so complex, and how and why have critics made it seem far simpler than it is? She situates the film within its institutional context, that of the promotion of a new, apolitical, "accessible" Godard. She goes on to show how the recurrence of characteristic themes (e.g., prostitution) and attitudes (e.g., misogyny) lent the film an easy recognition along the lines that MacCabe in fact took. She argues that Godard deliberately solicits art-cinema comprehension strategies, of exactly the kind that the Camera Obscura writers employ (without displaying any awareness of those as normalized strategies). Thompson goes on to reveal how gaps and dislocations in the syuzhet prompt such thematizing. In place of MacCabe's hackneyed country/ city opposition, Thompson shows that the film employs a continuum of settings: a city, a town, a village, a farm, and the countryside. Instead of a plot synopsis, Thompson offers a segmentation that points up the temporal construction of the syuzhet. Rather than positing a form and a content, Thompson argues that the film transforms its thematic material by means of an overall organization of parallel parts which compare different characters. Having established all these macrostructural factors, Thompson is able to explain functionally what most critics ignore or interpret atomistically: the stop-motion sequences that interrupt the film and (contra MacCabe) the insistent and ambiguous organization of sounds. It is not just that Thompson's analysis of narrative, narration, and style has a finesse not approached by any other discussion. The real point is the range and depth of the conceptual scheme she employs. Neoformalist poetics, while concentrating on historical context, narrative form, and cinematic style, does not exclude thematic interpretations. It absorbs them into a dynamic system--here, one that reveals why discrete meanings can be the bait at which critics will snap, and how a clever filmmaker has set the trap for them. Historical poetics, in its concern for constructional effects, thereby comes to include the study of the conventions of film criticism itself.



5. Grand Theory, SLAB Theory, and Poetics

Although Grand Theory and historical poetics operate at different levels of generality, they invite comparison, if only because most people studying film have been influenced by one particular version of the former. This version treats cinema study as an instance of the study of the "human subject," employing tenets based upon Saussurean semiotics, Lacanian psychoanalysis, Althusserian Marxism, and Barthesian textual theory. I shall therefore call this version, acronymically and a little acrimoniously, SLAB theory. SLAB theory is most clearly identified with the main current of work in Cahiers du cinema during the early 1970s and Screen later in the decade. It is handily codified in Rosalind Coward and John Ellis' Language and Materialism. Most subsequent survey texts, such as Dudley Andrew's Concepts of Film Theory and Kaja Silverman's Subject of Semiotics, treat this trend as central to contemporary film studies. Although SLAB theory is subject to internal revision, and although it now seems close to a skirmish on its left flank with the burgeoning area called "cultural studies," I shall treat it as the mainstream position within film theory at present. I am not here concerned with laying out conceptual problems in SLAB theory as such, only with contrasting its intellectual modus operandi--its methods, if you like--with the aims of a historical poetics.

1. Whereas poetics is problem- and question-centered, SLAB theory is doctrine-centered.

SLAB theorists assume that they possess a general theory of social and psychic life which can subsume cinematic phenomena under broader laws. But this theory constitutes an ad hoc assemblage of pieces from various thinkers' works: some chunks of Lacan, a little of Althusser, etc. Hence Jonathan Rée's description--the nouveau melange. The effect is most clearly seen in those syntheses or textbooks that cut up pieces of doctrine and then provide an exposition that patches them together.

Likewise, SLAB theory changes by adding on new pieces of doctrine. The fact that one bit of any thinker's work can always be linked, somehow, with a bit of any other's underwrites the project of theory as bricolage. The absorption of a few terms from the Christian, anti-Freudian, and neo-Kantian Mikhail Bakhtin into psychoanalytic, feminist, culturalist, and orthodox lit-crit "methods" is only the most recent example. Even a single word can trigger the bricolage impulse. Teresa de Lauretis finds that both the physiologist Colin Blakemore and the semiotician Umberto Eco use the term "mapping." She concludes from this that both theorists' works support the idea that "perception and signification are neither direct or simple reproduction (copy, mimesis, reflection) nor inevitably predetermined by biology, anatomy, or destiny; though they are socially determined and overdetermined." This is an unwarranted inference from Blakemore's discussion, which stresses physiological invariants and evolutionary adaptation; a commitment to the social overdetermination of perception is hard to square with Blakemore's assertion: "Human perception depends ultimately on activity within the nerve centers of the brain." De Lauretis might reply that even if Blakemore does not believe in the social construction of perception, his evidence supports it. But then one could counter that he has marshalled the bulk of his evidence, which she does not examine, to demonstrate exactly the opposite position, which she denounces but does not attempt to refute.

It is thus not surprising that challenges to SLAB theory are typically cast in the form "My Continental thinker can lick yours": Deleuze against Lacan, Benjamin against Althusser, Frankfort versus Paris. Since doctrines age faster than ideas, there emerges an urge to stay on the cutting edge. How the SLAB theorist does so is, again, most clearly seen in the summarizing texts. Here the author functions as a tipster, assuming that having the most recent word in a debate means having the last word on the subject. In a review of Language and Materialism, Rée describes many kindred efforts:

In fact most of Coward and Ellis' fallacies are of a slightly different kind: they involve not so much referring to a particular authority, as watching the ways in which the currents of opinion are flowing: a kind of punting on what future authorities will say, based on ideas of what can be "seen" and seen "only now."

In its grim determination to keep abreast, SLAB theory reveals its only open-ended side: almost anything may become grist for the doctrinal mill.



2. Poetics, in its contemporary form, conducts systematic research; SLAB theory does not.

Systematic research consists of posing questions, reflecting on the historical factors that lead to the questions' becoming salient, broaching alternative answers, and weighing them in the light of evidence; it also presents arguments that seek to demonstrate that some answers are better than others. By these canons, SLAB theory does not constitute systematic research.

As a rule, SLAB theory does not ask particular questions and reason out possible answers, rejecting and refining them and weighing the comparative advantages of competing explanatory frameworks. The writer instead starts with a doctrinal abstraction and draws on cinematic phenomena as illustrative examples. Thus Silverman's Subject of Semiotics employs filmic and literary texts as audio-visual aids in laying out claims about the Oedipus complex or condensation and displacement; she does not, by and large, cite evidence that would establish the claims as holding good about general human phenomena of the sort that the theory aspires to explain. Nor does she consider how the same cinematic processes might be explained by rival theories. Nor does she consider counterexamples that might challenge her premises or inferences. The point is important because any belief, including astrology and a trust in dowsing rods, can be illustrated by particular phenomena. Marshall Edelson calls this "enumerative inductivism," the notion that adducing instances of a hypothesis will support it; in fact, such a notion is vacuous because any number of hypotheses can be supported by a set of instances. The real test involves "eliminative inductivism": "No conjecture about the world is in and of itself confirmed by evidence. It is always evaluated relative to some rival. The degree of its acceptance is simply the extent to which at any particular time it is considered better than its comparable rivals."

The focus upon doctrine can blind one to the most obvious counterinstances. Instead of asking what the everyday ideology of vision might be, John Ellis starts from the premise that the cinematic institution necessarily imitates a phenomenological model. So he informs us that projection in a movie theatre "exactly parallels" our ideology of vision, "one that thinks of the eyes as projecting a beam of light, like a torch-beam, that illuminates what we look at, making it visible and perceptible." Ellis' commitment to SLAB theory has apparently made him oblivious to people's habit of switching on the lights when they enter a darkened room.

If SLAB theory is largely uninterested in posing questions and examining a range of evidence, it is no more keen on doing homework in the history of its concepts. Its canonical texts arrive untainted by any larger context (save perhaps that of "1920s Soviet culture" or "Paris after 1968"). Freud is not situated within the history of psychology, nor Saussure in the history of linguistics; Lacan's ties to Surrealism are passed over, as is Althusser's complicated relation to the French Communist Party. In the endless exposition of these texts, the writer has license to remake history. One can skip from a schematic account of Descartes' conception of the "subject" to an account of Freud's, as if everyone in the intervening centuries, including minor thinkers like Hume, Kant, and Hegel, were blundering about in rationalist darkness. To read SLAB theory, one would never know that such books as Sebastiano Timpanaro's Freudian Slip or B. A. Farrell's Standing of Psychoanalysis or G. A. Cohen's Karl Marx's Theory of History: A Defence exist.

Such provincialism cripples SLAB theory as an intellectual endeavor. Any theorist who really wanted to pose questions about language would grapple with the work of Locke, Humboldt, Sapir, Whorf, Wittgenstein, Quine, Chomsky, Montague, Grice, Putnam, Kripke, Davidson, Dummett, Searle, Katz, and Sperber. Any theorist claiming an interest in psychology would certainly need to consider the contending ideas of Piaget, the Russian reflexologists, Vygotsky, Bruner, Fodor, et al. Any theorist seeking the sociopolitical functions of cinema cannot ignore Weber, Durkheim, Mauss, Parsons, Elster, and Giddens. A theorist who pronounces upon whether semiotics or psychoanalysis is a science ought to be familiar with the history and philosophy of the sciences. Yet inspection of current "theoretical" texts in our field reveals an embarrassing ignorance on all these scores. As it stands, SLAB theory constitutes a convenient way of not knowing a lot of things. Paradoxically, a movement that makes novelty its chief appeal seems unaware of recent developments in the fields to which it lays claim. SLAB theory wants to be new without being current.

The rhetoric of SLAB theory can be seen as a strategic concealment of the conceptual problems I have noted. Despite its persistent use of the phrase "X argues," SLAB theory does not characteristically offer arguments. Argument presupposes a dialectical confrontation with potential or actual opponents. Assuming that s/he writes for a skeptical reader, the writer anticipates objections, refutes antagonists, and advances her/ his thesis as the most plausible candidate. SLAB theory is instead largely expository, summarizing and synthesizing claims made by previous theorists. There was a short flurry of pseudoscientific rigor in the early 1970s, but this, which enraged so many opponents, now emerges as a momentary vogue. Once Barthes rejected his pre S/Z work as tainted by "scientificity," he ratified a movement back to the intuitive belletrism we have already seen in the Sauve qui peut (la vie) discussions and which comes virtually second nature to people of literary training.

SLAB theory has found its most comfortable rhetorical mode in a form of commentary whose components include the following: exegesis through quotation and paraphrase, the rectification of this or that point in the light of recent developments (Rée's "only now can we see" syndrome), the extrapolation of other points on the basis of conceptual or terminological association (e.g., de Lauretis on "mapping"), the interpretation of illustrative examples from films, and above all the striking of a stance that means business. The essay, chapter, or book is likely to end with some tough talk, when the writer invokes something new and dangerous: a recently translated book demanding to be assimilated, a just-finished film to be interpreted, or a new mode of filmmaking.

What the exposition-rumination-illustrations format blocks, of course, are the massive critiques that have been launched at SLAB theory and its cinematic adherents. Reading SLAB work, one could not learn that there are standard arguments against Saussure, Lacan, Althusser, and Barthes, for no SLAB theorist bothers to defend these thinkers' ideas in any engaged way. One would scarcely know that many writers have pointed out conceptual difficulties in SLAB arguments about film. Unlike Shklovksy, SLAB expositors usually regard theory as entrusted to them for long-term safekeeping. Still, if one holds some power, as SLAB theory does, ignoring all opponents, however complacent it may seem, is the safest rhetorical recourse.



3. Whereas poetics uses concepts to construct explanatory propositions, SLAB theory uses concepts to construct interpretive narratives.

If SLAB theorists are uninterested in debating their views within wider contexts, it is for the very practical reason that theorizing seldom lies at the center of their concerns. Theory becomes not explanation but a guide for explication. As applied to cinema, SLAB theory tells stories. Or rather, a story with few variants--the tale of stable personal identity, lost and (perhaps) found (but differently). This is a perennially popular tale among humanist academics, and SLAB theory draws upon psychoanalysis (that trove of great stories) in order to deck it out in different costumes. By means of traditional interpretive tactics, such as analogy and personification, any aspect of a film (setting, camera position, editing) can be assigned a meaning within this drama of subjectivity. SLAB theory yields a scheme for interpreting films that is close enough to traditional semantic fields (order/ disorder, identity/ loss of identity, self/ other, male/ female) to seem comfortable but also new enough in its particular working out to rejuvenate thematic criticism (as when the Mirror Stage underwrites critics' penchant for looking for reflections and doublings). SLAB theory as theory can escape scrutiny because it is made to be used, to let the critic come to the desired conclusions about this text's conventionality or that text's transgressiveness.

Historical poetics needs no such stories to guide its work. It offers explanations, not the recasting of films into the form of a master narrative; and insofar as metacriticism is part of its purview, it may take as part of its business the study of how SLAB theory has become geared toward interpretation.

As SLAB theory has incorporated many diverse ideas into its bricolage, so has it included historical poetics. Characteristically, however, the evidence mobilized by the Soviet filmmakers, the Russian Formalists, Arnheim, Bazin, and more recent poeticians have become, in the hands of SLAB theorists, yet more illustrations of the same received doctrines. True, the theorist often gets it wrong--shot/ reverse shot and point-of-view editing seem surprisingly difficult to grasp- but even in the muddles, there is a recognition that if SLAB doctrine is to be mapped onto cinema, it needs at least the vocabulary and concepts provided by some poetics.



In light of these points, I conclude that contemporary film studies, thought to be dominated by abstract theory, is actually quite untheoretical--if theory is understood not as the routine exposition of cryptic doctrines but as an active, open-ended enterprise that poses clearly-defined questions, seeks empirical evidence that will help decide them, analyzes alternative explanations of that evidence, and systematically argues for the best answer. Film theory, I take it, demands wide reading, constant reflection, intimate acquaintance with the history of the problems posed, and a degree of skepticism that compells the researcher to seek out difficult challenges, either in the data or in the form of opposing arguments. But these qualities are not characteristic of SLAB theory. Its doctrinaire quality has led to dogmatism; its inadequate research has made it blinkered; and its streamlined schematism has rendered it simply another method for interpreting films. Unfortunately, this modus operandi is encouraged by several institutional factors, most recently those publication ventures which reward academics for dashing off homogenized summaries of Grand Theory aimed at student consumption. This might be called the Methuenization of the humanities.

SLAB theorists commonly counterpose "theory" to "history," as if historical research could not also be theoretical. I propose a more informative opposition. SLAB theory and its offshoots, in their deepest assumptions and their concrete practices, have consolidated a new scholasticism, a ceaseless commentary on authoritative sources. Poetics, on the other hand, frankly offers scholarship--an open-ended, corrigible inquiry that respects the reciprocal claims of conceptual coherence and empirical adequacy. Lacking a substantive doctrine, it does not have the answers ready before anyone has asked the questions.

To commit oneself to scholarship is, at this point in history, openly to commit oneself to academic institutions. Although SLAB theorists have been reluctant to acknowledge it, their theory depends crucially upon the university; indeed, Saussure, Lacan et al. produced most of their work in academic circumstances which were, by contemporary American or English standards, leisurely. Historical poetics can succeed only if colleges, universities, and archives give the researcher the resources to work steadily on questions that cannot be answered from the depths of the armchair.

I am certainly not recommending that we embrace a cozy professionalism. What matters now is that we exploit the academicization of film study for scholarly ends. If we recall that Bazin and the Formalists produced brilliant insights within academic conditions we would consider materially barren, we can appreciate the enormous opportunity which most Grand Theorists of film neglect. We can, for the first time in history, study cinema according to the stringent demands of scholarly inquiry. We have the time to fight with each other about ideas and enthusiastically pursue answers to truly demanding questions. We can do this best, I think, by transcending that Methodist division of labor initiated in Hyman's and Wellek and Warren's time. In this respect, historical poetics becomes not one method but a model of basic research into cinema. It offers the best current hope for setting high intellectual standards for film study.