Classical Hollywood Cinemaa formalist assessment of narrative norms.
The following essay is a good analysis of characteristic narrative conventions and norms of classical Hollywood cinema. The formalist approach is clearly restrained in its critical ambitions, but all the more useful as a basis for other perspectives more concerned with a critique of the ideology of mainstream cinema.
[Summary of the main points made in this essay]
David Bordwell, "Classical Hollywood Cinema: Narrational Principles and Procedures."
(from: Rosen 1986: 17 - 34; hypertext by CB)
Three aspects of narrative can, at least provisionally, be kept distinct. A narrative can be studied as representation, how it refers to or signifies a world or body of ideas. This we might call the "semantics" of narrative, and it is exemplified in most studies of characterization or realism. A narrative can also be studied as a structure, the way its components combine to create a distinctive whole. An example of this "syntactic" approach would be Vladimir Propp's morphology of the magical fairy tale. Finally, we can study a narrative as an act, a dynamic process of presenting a story to a perceiver. This would embrace considerations of source, function, and effect; the temporal progress of information or action; and concepts like the "narrator." This is the study of narration, the "pragmatics" of narrative phenomena. What follows concerns itself with naration in classical Hollywood cinema between 1917 and 1960, but it does not singlemindedly stick to this aspect. It is common for any one narrative analysis to focus on one aspect but to bring in others as needed. Levi-Strauss, for instance, uses a concept of narrative structure to disclose deeper levels of meaning, what the myth represents: syntax is a tool for revealing semantics. In this essay, I introduce issues of representation (especially denotative representation) and structure (especially dramaturgical structure) in order to highlight how classical Hollywood narration constitutes a particular configuration of normalized options for representing the story and manipulating composition and style. Since this is a précis of an /18/ extensive body of research, it will have an unfortunately programmatic, ad hoc air about it, but the readers can refer to the end of the article for the evidence upon which these claims are based. Unusual nomenclature is glossed [here]:
Fabula: Russian formalist term for the narrative events in causal chronological sequence. (Sometimes translated as "story.") A construct of the spectator.
Syuzhet: Russian formalist term for the systematic presentation of fabula events in the text we have before us. (Sometimes translated as "plot.")
Narration: the process of cuing a perceiver to construct a fabula by use of syuzhet patterning and film style.
Knowledgeability: the extent to which the narration lays claim to a range and depth of knowledge of fabula [story] information.
Self-consciousness: the degree to which the narration acknowledges its address to the spectator.
Communicativeness: the extent to which the narration withholds or communicates fabula [story] information.
Compositional motivation: justifying the presence of an element by its function in advancing the syuzhet [plot].
Realistic motivation: justifying the presence of an element by virtue of its conformity with some extratextual reality.
Artistic motivation: justifying the presence of an element by its calling attention to itself as a distinct device.
Transtextual motivation: justifying the presence of an element by reference to the category of texts to which this one belongs (e.g., by appealing to genre conventions).
The Straight Corridor
The classical Hollywood film presents psychologically defined individuals who struggle to solve a clear-cut problem or to attain specific goals. In the course of this struggle, the characters enter into conflict with others or with external circumstances. The story ends with a decisive victory or defeat, a resolution of the problem and a clear achievement or nonachievement of the goals. The principal causal agency is thus the character, a distinctive individual endowed with an evident, consistent batch of traits, qualities, and behaviors. Although the cinema inherits many conventions of portrayal from theater and literature, the character types of melodrama and popular fiction get fleshed out by the addition of unique motifs, habits, or behavioral tics. In parallel fashion, the star system has as one of its functions the creation of a rough character prototype which is then adjusted to the particular needs of the role. The most "specified" character is usually the protagonist, who becomes the principal causal agent, the target of any narrational restriction, and the chief object of audience identification. These features of the syuchet [plot] come as no surprise, though already there are important differences from other narrational modes (e.g., the comparative absence of consistent and goal-oriented characters in art-cinema narration).
Of all such modes, the classical one conforms most closely to the "canonic story" which story-comprehension researchers posit as normal for our culture. /19/ In fabula [story] terms, the reliance upon character cause and effect and the definition of the action as the attempt to achieve a goal are salient features of the canonic format. At the level of the syuzhet [plot], the classical film respects the canonic pattern of establishing an initial state of affairs which gets violated and which must then be set right. Indeed, Hollywood screenplay-writing manuals have long insisted on a formula which has been revived in recent structural analysis: the plot consists of an undisturbed stage, the disturbance, the struggle, and the elimination of the disturbance. Such a syuzhet [plot] pattern is the inheritance not of some monolithic construct called the "novelistic" but of specific historical forms: the well-made play, the popular romance, and, crucially, the late nineteenth-century short story. The characters' causal interactions are thus to a great extent functions of such overarching syuzhet/fabula [plot/story] patterns.
In classical fabula [story] construction, causality is the prime unifying principle. Analogies between characters, settings, and situations are certainly present, but at the denotative level any parallelism is subordinated to the movement of cause and effect. Spatial configurations are motivated by realism (a newspaper office must contain desks, typewriters, phones) and, chiefly, by compositional necessity (the desk and typewriter will be used to write causally significant news stories, the phones form crucial links among characters). Causality also motivates temporal principles of organization: the syuzhet represents the order, frequency, and duration of fabula events in ways which bring out the salient causal relations. This process is especially evident in a device highly characteristic of classical narrationthe deadline. A deadline can be measured by calendars (Around the World in Eighty Days), by clocks (High Noon), by stipulation ("You've got a week but not a minute longer"), or simply by cues that time is running out (the last minute rescue). That the climax of a classical film is often a deadline shows the structural power of defining dramatic duration as the time it takes to achieve or fail to achieve a goal.
Usually the classical syuzhet [plot] presents a double causal structure, two plot lines: one involving heterosexual romance (boy/girl, husband/wife), the other line involving another spherework, war, a mission or quest, other personal relationships. Each line will possess a goal, obstacles, and a climax. In Wild and Wooly (1917), the hero Jeff has two goalsto live a wild western life and to court Nell, the woman of his dreams. The plot can be complicated by several lines, such as countervailing goals (the people of Bitter Creek want Jeff to get them a railroad spur, a crooked Indian agent wants to pull a robbery) or multiple romances (as in Footlight Parade and Meet Me in St. Louis). In most cases, the romance sphere and the other sphere of action are distinct but interdependent. The plot may close off one line before the other, but often the two lines coincide at the climax: resolving one triggers the resolution of the other. In His Girl Friday, the reprieve of Earl Williams precedes the reconciliation of Walter and Hildy, but it is also the condition of the couple's reunion.
The syuzhet [plot] is always broken up into segments. In the silent era, the typical /20/ Hollywood film would contain between 9 and 18 sequences; in the sound era, between 14 and 35 (with postwar films tending to have more sequences). Speaking roughly, there are only two types of Hollywood segments: "summaries" (compromising Metz's third, fourth, and eighth syntagmatic types) and "scenes" (Metz's fifth, sixth, seventh, and eighth types.) Hollywood narration clearly demarcates its scenes by neoclassical criteriaunity of time (continuous or consistently intermittent duration), space (a definable locale), and action (a distinct cause-effect phase). The bounds of the sequence will be marked by some standardized punctuations (dissolve, fade, wipe, sound bridge). Raymond Belour points out that the classical segment tends also to define itself microcosmically (through internal repetitions of style or story material) and macrocosmically (by parallels with other segments of the same magnitude). We must also remember that each film establishes its own scale of segmentation. A syuzhet [pot] which concentrates on a single locale over a limited dramatic duration (e.g., the one-night-in-a-haunted-house film) may create segments by character entrances or exits, a theatrical liaison des scenes. In a film which spans decades and many locales, a series of dissolves from one small action to another will not necessarily create distinct sequences.
The classical segment is not a sealed entity. Spatially and temporally it is closed, but causally it is open. It works to advance the causal progression and open up new developments. The pattern of this forward momentum is quite codified. The montage sequence tends to function as a transitional summary, compressing a single causal development, but the scene of character actionthe building block of classical Hollywood dramaturgyis more intricately constructed. Each scene displays distinct phases. First comes the exposition, which specifies the time, place, and relevant characterstheir spatial positions and their current states of mind (usually as a result of previous scenes). In the middle of the scene, characters act toward their goals: they struggle, make choices, make appointments, set deadlines, and plan future events. In the course of this, the classical scene continues or closes off cause-effect developments left dangling in prior scenes while also opening up new causal lines for future development. At least one line of action must be left suspended, in order to motivate the shifts to the next scene, which picks up the suspended line (often via a "dialogue hook"). Hence the famous "linearity" of classical constructiona trait not characteristic of Soviet montage films(which often refuse to demarcate scenes clearly) or of art-cinema narration (with its ambiguous interplay of subjectivity and objectivity).
Here is a simple example. In The Killers (1946), the insurance investigator Riordan has been hearing Lieutenant Lubinsky's account of Ole Anderson's early life. At the end of the scene, Lubinsky tells Riordan that they're burying Ole today. This dangling cause leads to the next scene, set in the cemetery. An establishing shot provides spatial exposition. While the clergyman intones the funeral oration, Riordan asks Lubinsky the identity of various mourners. The /21/ last, a solitary old man, is identified as "an old-time hoodlum named Charleston." Dissolve to a pool hall, with Charleston and Riordan at a table drinking and talking about Ole. During the burial scene, the Lubinsky line of inquiry is closed off and the Charleston line is initiated. When the scene halts, Charleston is left suspended, but he is picked up immediately in the exposition of the next scene. Instead of a complex braiding of causal lines (as in Rivette) or an abrupt breaking of them (as in Antonioni, Godard, or Bresson), the classical Hollywood film spins them out in smooth, careful linearity.
The linkage of localized causal lines must eventually terminate. How to conclude the syuzhet? There are two ways of regarding the classical ending. We can see it as the crowning of the structure, the logical conclusion of the string of events, the final effect of the initial cause. This view has some validity, not only in the light of the tight construction that we frequently encounter in Hollywood films but also given the precepts of Hollywood screenwriting. Rule books tirelessly bemoan the pressures for a happy ending and emphasize the need for a logical wrap-up. Still, there are enough instances of unmotivated or inadequate plot resolutions to suggest a second hypothesis: that the classical ending is not all that structurally decisive, being a more or less arbitrary readjustment of that world knocked awry in the previous eighty minutes. Parker Tyler suggests that Hollywood regards all endings as "purely conventional, formal, and often, like the charade, of an infantile logic." Here again we see the importance of the plot line involving heterosexual romance. It is significant that of 100 randomly sampled Hollywood films, over 60 ended with a display of the united romantic couplethe cliche happy ending, often with a "clinch" and many more could be said to end happily. Thus an extrinsic norm, the need to resolve the plot in a way that provides "poetic justice," becomes a structural constant, inserted with more or less motivation into its proper slot, the epilogue. In any narrative, as Meir Sternberg points out, when the syuzhet's end is strongly precast by convention, the compositional attention falls on the retardation of outcome accomplished by the middle portions; the text will then "account for the necessary retardation in quasi-mimetic terms by placing the causes for delay within the fictive world itself and turning the middle into the bulk of the represented action." At times, however, the motivation is constructed to be inadequate, and a discordance between preceding causality and happy denouement becomes noticeable as an ideological difficulty; such is the case with films like You Only Live Once, Suspicion, The Woman in the Window, and The Wrong Man.[l3] We ought, then, to be prepared for either a skillful tying up of all loose ends or a more or less miraculous appearance of what Brecht called bourgeois literature's mounted messenger. "The mounted messenger guarantees you a truly undisturbed appreciation of even the most intolerable conditions, so it is a sine qua non for a literature whose sine qua non is that it leads nowhere."
The classical ending may be a sore spot in another respect. Even if the ending resolves the two principal causal lines, some comparatively minor issues may /22/ still be left dangling. For example, the fates of secondary characters may go unsettled. In His Girl Friday, Earl Williams is reprieved, the corrupt administration will be thrown out of office, and Walter and Hildy are reunited, but we never learn what happens to Molly Malloy, who jumped out a window to distract the reporters. (We know only that she was alive after the fall.) One could argue that in the resolution of the main problem we forget minor matters, but this is only a partial explanation. Our forgetting is promoted by the device of closing the film with an epilogue, a brief celebration of the stable state achieved by the main characters. Not only does the epilogue reinforce the tendency toward a happy ending; it also repeats connotative motivs that have run throughout the film. His Girl Friday closes on a brief epilogue of Walter and Hildy calling the newspaper office to announce their remarriage. They learn that a strike has started in Albany, and Walter proposes stopping off to cover it on their honeymoon. This plot twist announces a repetition of what happened on their first honeymoon and recalls that Hildy was going to marry Bruce and live in Albany. As the couple leave, Hildy carrying her suitcase, Walter suggests that Bruce might put them up. The neat recurrence of these motifs gives the narration a strong unity; when such details are so tightly bound together, Molly Malloy's fate is more likely to be overlooked. Perhaps instead of "closure" it would be better to speak of a "closure effect," or even, if the strain of resolved and unresolved issues seems strong, of "pseudo-closure." At the level of extrinsic norms, though, the most coherent possible epilogue remains the standard to be aimed at.
Commonplaces like "transparency" and "invisibility" are on the whole unhelpful in specifying the narrational properties of the classical film. Very generally, we can say that classical narration tends to be omniscient, highly communicative, and only moderately self-conscious. That is, the narration knows more than any or all characters, it conceals relatively little (chiefly "what will happen next"), and it seldom acknowledges its own address to the audience. But we must qualify this characterization in two respects. First, generic factors often create variations upon these precepts. A detective film will be quite restricted in its range of knowledge and highly suppressive in concealing causal information. A melodrama like In This Our Life can be slightly more self-conscious than The Big Sleep, especially in its use of acting and music. A musical will contain codified moments of self-consciousness (e.g., when characters sing directly out at the viewer). Second, the temporal progression of the syuzhet makes narrational properties fluctuate across the film, and these too are codified. Typically, the opening and closing of the film are the most self-conscious, omniscient, and communicative passages. The credit sequence and the first few shots usually bear traces of an overt narration. Once the action has started, however,the narration becomes more covert, letting the characters and their interaction take over the transmission of information. Overt narrational activity returns at certain conventional moments: the beginnings and endings of scenes /23/ (e.g, establishing shots; shots of signs, camera movements out from or in to significant objects, symbolic dissolves), and that summary passage known as the "montage sequence." At the very close of the syuzhet, the narration may again acknowledge its awareness of the audience (musical motifs reappear, characters look to the camera or close a door in our face), its omniscience (e.g., the camera retreats to a long-shot), and its communicativeness (now we know all). Classical narration is thus not equally "invisible" in every type of film nor throughout any one film; the "marks of enunciation" are sometimes flaunted.
The communicativeness of classical narration is evident in the way that the syuzhet handles gaps. If time is skipped over, a montage sequence or a bit of character dialogue informs us; if a cause is missing, we will typically be informed that something isn't there. And gaps will seldom be permanent. "In the beginning of the motion picture," writes one scenarist, "we don't know anything. During the course of the story, information is accumulated, until at the end we know everything." Again, these principles can be mitigated by generic motivation. A mystery might suppress a gap (e.g., the opening of Mildred Pierce), a fantasy might leave a cause still questionable at the end (e.g., The Enchanted Cottage). In this respect, Citizen Kane remains somewhat "unclassical": the narration supplies the answer to the "Rosebud" mystery, but the central traits of Kane's character remain partly undetermined, and no generic motivation justifies this.
The syuzhet's [plot's] construction of time powerfully shapes the fluctuating overtness of narration. When the syuzhet adheres to chronological order and omits the causally unimportant periods of time, the narration becomes highly communicative and unselfconscious. On the other hand, a montage sequence compresses a political campaign, a murder trial, or the effects of Prohibition into moments, and the narration becomes overtly omniscient. A flashback can quickly and covertly fill a causal gap. Redundancy can be achieved without violating the fabula world if the narration represents each story event several times in the syuzbet, through one enactment and several recountings in character dialogue. Deadlines neatly let the syuzhet unselfconsciously respect the durational limits that the fabula world sets for its action. When it is necessary to suggest repeated or habitual actions, the montage sequence will again do nicely, as Sartre noted when he praised Citizen Kane's montages for achieving the equivalent of the "frequentative" tense: "He made his wife sing in every theater in America." When the syuzhet uses a newspaper headline to cover gaps of time, we recognize both the narration's omniscience and its relatively low profile. (The public record is less self-conscious than an intertitle "coming straight from" the narration.) More generally, classical narration reveals its discretion by posing as an editorial intelligence that selects certain stretches of time for full-scale treatment (the scenes), pares down others a little, presents others in highly compressed fashion (the montage sequences), and simply scissors out events that are inconsequential. When fabula duration is expanded, it is done through crosscutting /24/
Overall narrational qualities are also manifested in the film's manipulation of space. Figures are adjusted for moderate self-consciousness by angling the bodies more or less frontally but avoiding to-camera gazes (except, of course, in optical point-of-view passages). That no causally significant cues in a scene are left unknown testifies to the communicativeness of narration. Most important is the tendency of the classical film to render narrational omniscience through spatial omnipresence. If the narration plays down its knowledge of effects and upcoming temporal developments, it does not hesitate to reveal its ability to change views at will. Cutting within a scene and crosscutting between various locales testify to the narration's omnipresense. Writing in 1935, a critic claims that the camera is omniscient in that it "stimulates, through correct choice of subject matter and set-up, the sense within the percipient of 'being at the most vital part of the experienceat the most advantageous point of perception throughout the picture." Whereas Miklos Jancso's long takes create spatial patterns that refuse omnipresence and thus drastically restrict the spectator's knowledge of story information, classical omnipresence makes the cognitive schema we call "the camera" into an ideal invisible observer, freed from the contingencies of space and time but then discreetly confining itself to codified patterns for the sake of story intelligibility.
By virtue of its handling of space and time, classical narration makes the fabula world an internally consistent construct into which narration seems to step from the outside. Manipulation of mise-en-scene (figure behavior, lighting, setting, costume) creates an apparently independent profilmic event, which becomes the tangible story world framed and recorded from without. This framing and recording tends to be taken as the narration itself, which can in turn be more or less overt, more or less "intrusive" on the posited homogeneity of the story world. Classical narration thus depends upon the notion of the "invisible observer." Bazin, for instance, portrays the classical scene as existing independently of narration, as if on a stage. The same quality is named by the notion of "concealment of production": the fabula seems not to have been constructed but appears to have preexisted its narrational representation. (In production, in some sense, it often did: for major films of the 1930s and thereafter, Hollywood set designers created three-dimensional tabletop mockups of sets within which models of cameras, actors, and lighting units could be placed to predetermine filming procedures.)
This invisible-observer narration is itself often fairly effaced. The stylistic causes of this I shall examine shortly, but we can already see that classical narration quickly cues us to construct story logic (causality, parallelisms), time, and space in ways that make the events "before the camera" our principal source of information. For example, it is obvious that Hollywood narratives are highly redundant, but this effect is achieved principally by patterns attributable to the story world. Following Susan Suleiman's taxonomy, we can see that the narration assigns the same traits and functions to each character on her or his /25/ appearance; different characters present the same interpretive commentary on the same character or situation; similar events befall different characters; and so on. Information is for the most part repeated by characters' dialogue or demeanor. There is, admittedly, some redundancy between narrational commentary and depicted fabula action, as when silent film expository intertitles convey crucial information or when nondiegetic music is pleonastic with the action (e. g., "Here Comes the Bride" in In This Our Life). Nevertheless, in general, the narration is so constructed that characters and their behavior produce and reiterate the necessary story data. The Soviet montage cinema makes much stronger use of redundancies between narrational commentary and fabula action. Retardation operates in analogous fashion: the construction of the total fabula is delayed principally by inserted lines of action (e. g., causally relevant subplots, interpolated comedy bits, musical numbers) rather than by narrational digressions of the sort found in the "God and Country" sequence of October. Similarly, causal gaps in the fabula are usually signaled by character actions (e.g., the discovery of clues in detective films). The viewer concentrates on constructing the fabula, not on asking why the narration is representing the fabula in this particular waya question more typical of art-cinema narration.
The priority of fabula causality and an integral fabula world commits classical narration to unambiguous presentation. Whereas art-cinema narration can blur the lines separating objective diegetic reality, characters' mental states, and inserted narrational commentary, the classical film asks us to assume clear distinctions among these states. When the classical film restricts knowledge to a character, as in most of The Big Sleep and Murder My Sweet, there is nonetheless a firm borderline between subjective and objective depiction. Of course, the narration can set traps for us, as in Possessed (1947), when a murder that appears to be objective is revealed to have been subjective (a generically motivated switch, incidentally); but the hoax is revealed immediately and unequivocally. The classical flashback is revealing in this connection. Its presence is almost invariably motivated subjectively, since a character's recollection triggers the enacted representation of a prior event. But the range of knowledge in the flashback portion is often not identical with that of the character doing the remembering. It is common for the flashback to show us more than the character can know (e.g. scenes in which s/he is not present). An amusing example occurs in Ten North Frederick. The bulk of the film is presented as the daughter's flashback, but at the end of the syuzhet, back in the present, she learns for the first time information we had encountered in "her" flashback! Classical flashbacks are typically "objective": character memory is a pretext for a nonchronological syuzhet arrangement. Similarly, optically subjective shots become anchored in an objective context. One writer notes that a point-of-view shot "must be motivated by, and definitely linked to, the objective scenes [shots] that precede and follow it." This is one source of the power of the invisible-observer effect: the camera seems always to include character subjectivity within a broader and definite objectivity. /26/
Even if the naive spectator takes the style of the classical Hollywood film to be invisible or seamless, this is not much critical help. What makes the style so self-effacing? The question cannot be answered completely until we consider the spectator's activity, but we may start with Yuri Tynyanov's suggestion: "Pointing to the 'restraint' or 'naturalism' of the style in the case of some film or some director is not the same as sweeping away the role of style. Quite simply, there are a variety of styles and they have various roles, according to their relationship to the development of the syuzhet." Three general propositions, then.
(1) On the whole, classical narration treats film technique as a vehicle for the syuzhet's transmission of fabula information.
Of all modes of narration, the classical is most concerned to motivate style compositionally, as a function of syuzhet patterning. Consider the very notion of what we now call a shot. For decades, Hollywood practice called a shot a "scene," thus conflating a material stylistic unit with a dramaturgical one. In filmmaking practice, the overriding principle was to make every instantiation of technique obedient to the character's transmission of fabula information, which would invariably make bodies and faces the focal points of attention. Given the recurrent causal structure of the classical scene (exposition, closing off of an old causal factor, introduction of new causal factors, suspension of a new factor), the filmmaker can deploy film techniques isomorphically with respect to this structure. The introduction phase typically includes a shot which establishes the characters in space and time. As the characters interact, the scene is broken up into closer views of action and reaction, while setting, lighting, music, composition, and camera movement enhance the process of goal formulation, struggle, and decision. The scene usually closes on a portion of the spacea facial reaction, a significant objectthat provides a transition to the next scene.
While it is true that sometimes a classical film's style becomes "excessive," decoratively supplementing denotative syuzbet demands, the use of technique must be minimally motivated by the characters' interactions. "Excess," such as we find in Minnelli or Sirk, is often initially justified by generic convention. The same holds true for even the most eccentric stylists in Hollywood, Busby Berkeley and Josef von Sternberg, each of whom required a core of generic motivation (musical fantasy and exotic romance, respectively) for his experiments.
(2) In classical narration, style typically encourages the spectator to construct a coherent, consistent time and space of the fabula action.
Many other narrational norms value disorienting the spectator (albeit for different purposes). Only classical narration favors a style which strives for utmost denotative clarity from moment to moment. Each scene's temporal relation to its predecessor will be signaled early and unequivocally (by intertitles, conventional cues, a line of dialogue). Lighting must pick out figure from ground; color must define planes; in each shot, the center of story interest will tend to be centered in relation to the sides of the frame. Sound recording is perfected so as to allow for maximum clarity of dialogue. Camera movements aim at creating an unambiguous, voluminous space. "In dollying," remarks Alan Dwan, "as a rule we find it's a good idea to pass things.... We always noticed that if we dollied past a tree, it became solid and round, instead of flat." Hollywood makes much use of the anticipatory composition or camera movement, leaving space in the frame for the action or tracking so as to prepare for another character's entrance. Compare Godard's tendency to make framing wholly subservient to the actor's immediate movement with this comment of Raoul Walsh's: "There is only one way in which to shoot a scene, and that's the way which shows the audience what's happening next." Classical editing aims at making each shot the logical outcome of its predecessor and at reorienting the spectator through repeated setups. Momentary disorientation is permissible only if motivated realistically. Discontinuous editing, as in Slavko Vorkapich's sequence of the earthquake in San Francisco, is motivated by the chaos of the action depicted. Stylistic disorientation, in short, is permissible when it conveys disorienting story situations.
(3) Classical style consists of a strictly limited number of particular technical devices organized into a stable paradigm and ranked probabilistically according to syuzhet demands.
The stylistic conventions of Hollywood narration, ranging from shot composition to sound mixing, are intuitively recognizable to most viewers. This is because the style deploys a limited number of devices, and these devices are regulated as alternative depictive options. Lighting offers a simple example. A scene may be lit "high-key" or "low-key." There is three-point lighting (key, fill, and backlighting on figure, plus background lighting) versus single-source lighting. The cinematographer also has several degrees of diffusion available. Now in the abstract all choices are equiprobable, but in a given context, one alternative is more likely than its mates. In a comedy, high-key lighting is more probable; a dark street will realistically motivate single-source lighting; the closeup of a woman will be more heavily diffused than that of a man. The "invisibility" of the classical style in Hollywood relies not only on highly codified stylistic devices but also upon their codified functions in context.
Or recall the ways of framing the human figure. Most often, a character will be framed between plan-american (the knees-up framing) and medium closeup (the chest-up framing); the angle will be straight-on, at shoulder or chin level. The framing is less likely to be an extreme long-shot or an extreme closeup, a high or low angle. And a bird's eye view or a view from straight below are very improbable and would require compositional or generic motivation (e.g., as an optical point of view or as a view of a dance ensemble in a musical).
Most explicitly codified into rules is the system of classical continuity editing. /28/ The reliance upon an axis of action orients the spectator to the space, and the subsequent cutting presents clear paradigmatic choices among different kinds of "matches." That these are weighted probabilistically is shown by the fact that most Hollywood scenes begin with establishing shots, break the space into closer views linked by eyeline-matches and/or shot/reverse shots, and return to more distant views only when character movement or the entry of a new character requires the viewer to be reoriented. Playing an entire scene without an establishing shot is unlikely but permissible (especially if stock or location footage or special effects are employed); mismatched screen direction and inconsistently angled eyelines are less likely; perceptible jump cuts and unmotivated cutaways are flatly forbidden. This paradigmatic aspect makes the classical style, for all its "rules," not a formula or a recipe but a historically constrained set of more or less likely options.
These three factors go some way toward explaining why the classical Hollywood style passes relatively unnoticed. Each film will recombine familiar devices within fairly predictable patterns and according to the demands of the syuzhet. The spectator will almost never be at a loss to grasp a stylistic feature because s/he is oriented in time and space and because stylistic figures will be interpretable in the light of a paradigm.
When we consider the relation of syuzhet and style, we can say that the individual film is characterized by its obedience to a set of extrinsic norms which govern both syuzbet construction and stylistic patterning. The classical cinema does not encourage the film to cultivate idiosyncratic intrinsic norms; style and syuzhet seldom enjoy prominence. A film's principal innovations occur at the level of the fabulai.e., "new stories." Of course, syuzhet devices and stylistic features have changed over time. But the fundamental principles of syuzhet construction (preeminence of causality, goal-oriented protagonist, deadlines, etc.) have remained in force since 1917. The stability and uniformity of Hollywood narration is indeed one reason to call it classical, at least insofar as classicism in any art is traditionally characterized by obedience to extrinsic norms.
The Logic of Classical Spectatorship
The stability of syuzhet processes and stylistic configurations should not make us treat the classical spectator as passive material for a totalizing machine. The spectator performs particular cognitive operations which are no less active for being habitual and familiar. The Hollywood fabula is the product of a series of particular schemata, hypotheses, and inferences.
The spectator comes to a classical film very well prepared. The rough shape of syuzhet and fabula is likely to conform to the canonic story of an individual's goal-oriented, causally determined activity. The spectator knows the most likely stylistic figures and functions. The spectator has internalized the scenic norms /29/ of exposition, development of old causal line, and so forth. The viewer also knows the pertinent ways to motivate what is presented. "Realistic" motivation, in this mode, consists of making connections recognized as plausible by common opinion. ("A man like this would naturally. . .") Compositional motivation consists of picking out the important links of cause to effect. The most important of transtextual motivation are recognizing the recurrence of a star's persona from film to film and recognizing generic conventions. Generic motivation, as we have seen, has a particularly strong effect on narrational procedures. Finally, artistic motivationtaking an element as being present for its own sakeis not unknown in the classical film. A moment of spectacle or technical virtuosity, a thrown-in musical number or comic interlude: the Hollywood cinema intermittently welcomes the possibility of sheer self-absorption. Such moments may be highly reflexive, "baring the device" of the narration's own work, as when in Angels Over Broadway a destitute playwright reflects, "Our present plot problem is money."
On the basis of such schemata the viewer projects hypotheses. Hypotheses tend to be probable (validated at several points), sharply exclusive (rendered as either/or alternatives), and aimed at suspense (positing a future outcome). In Phil Rosen's Roaring Timber (1937), a landowner enters a saloon in which our hero is sitting. The owner is looking for a tough foreman. Hypothesis: he will ask the hero to take the job. This hypothesis is probable, future-oriented, and exclusive (either the man will ask our hero or he won't). The viewer is helped in framing such hypotheses by several processes. Repetition reaffirms the data on which hypotheses should be grounded. "State every important fact three times," suggests scenarist Frances Marion, "for the play is lost if the audience fails to understand the premises on which it is based." The exposition of past fabula action will characteristically be placed within the early scenes of the syuzhet, thus supplying a firm basis for our hypothesis-forming. Except in a mystery film, the exposition neither sounds warning signals nor actively misleads us; the primacy effect is given full sway. Characters will be introduced in typical behavior, while the star system reaffirms first impressions. ("The moment you see Walter Pidgeon in a film you know he could not do a mean or petty thing." The device of the deadline asks the viewer to construct forward-aiming, all-or-nothing causal hypotheses: either the protagonist will achieve the goal in time or s/he will not. And if information is unobtrusively "planted" early on, later hypotheses will become more probable by taking "insignificant" foreshadoing material for granted.
This process holds at the stylistic level as well. The spectator constructs fabula time and space according to schemata, cues, and hypothesis-framing. Hollywood's extrinsic norms, with their fixed devices and paradigmatic organization, supply the viewer with firm expectations that can be measured against the concrete cues emitted by the film. In making sense of a scene's space, the spectator need not mentally replicate every detail of the space but need only /30/ construct a rough relational map of the principal dramatic factors. Thus a "cheat cut" is easily ignored because the spectator's cognitive processes rank cues by their pertinence to constructing the ongoing causal chain of the fabula, and on this scale, the changes in speaker, camera position, and facial expression are more noteworthy than say, a slight shift in hand positions. The same goes for temporal mismatches.
What is rare in the classical film, then, is Henry James's "crooked corridor," the use of narration to make us jump to invalid conclusions. The avoidance of disorientation we saw at work in classical style holds good more broadly as well. Future-oriented "suspense" hypotheses are more important than past-oriented "curiosity" ones, and surprise is less important than either. In Roaring Timber, imagine if the landowner had entered the bar seeking a tough foreman, offered the job to our hero, and he had replied in a fashion that showed he was not tough. Indeed, one purpose of foreshadowing and repetition is exactly to avoid surprises later on. Of course, if all hypotheses were steadily and immediately confirmed, the viewer would quickly lose interest. Several factors intervene to complicate the process. Most generally, schemata are by definition abstract prototypes, structures, and procedures, and these never specify all the propertics of the text. Many long-range hypotheses must await confirmation. Retardation devices, being unpredictable to a great degree, can introduce objects of immediate attention as well as delay satisfaction of overall expectation. The primacy effect can be countered by what psychologists call a "recency effect" which qualifies and perhaps even appears to negate our first impression of a character or situation. Furthermore, the structure of the Hollywood scene, which almost invariably ends with an unresolved issue, insures that an event-centered hypothesis carries interest over to the next sequence. Finally, we should not underestimate the role of rapid rhythm in the classical film; more than one practitioner has stressed the need to move the construction of story action along so quickly that the audience has no time to reflector get bored. It is the task of classical narration to solicit strongly probable and exclusive hypotheses and then confirm them while still maintaining variety in the concrete working out of the action.
The classical system is not simpleminded. Recall that under normal exhibition circumstances the film viewer's rate of comprehension is absolutely controlled. The cueing of probable, exclusive, and suspense-oriented hypotheses is a way of adjusting dramaturgy to the demands of the viewing situation. The spectator need not rummage very far back into the film, since his or her expectations are aimed at the future. Preliminary exposition locks schemata into place quickly, and the all-or-nothing nature of most hypotheses allows rapid assimilation of information. Redundancy keeps attention on the issue of immediate moment, while judicious lacks of redundancy allow for minor surprises later. In all, classical narration manages the controlled pace of film viewing by asking the spectator to construe the syuzhet and the stylistic system in a single way: construct a denotative, univocal, integral fabula./31/
Implications and Avenues
By virtue of its centrality within international film commerce, Hollywood cinema has crucially influenced most other national cinemas. After 1917, the dominant forms of filmmaking abroad were deeply affected by the models of storytelling presented by the American studios. Yet the Hollywood cinema cannot be identified with classicism tout court. The "classicism" of 1930s Italy or 1950s Poland may mobilize quite different narrational devices. (For instance, the happy ending seems more characteristic of Hollywood than of other classicisms.) But in most such cinemas classical narration's principles and functions can be considered congruent with those outlined here. A group of Parisian researchers has come to comparable, if preliminary, conclusions about French films of the 1930s. Noel Burch has shown that in the German cinema, a mastery of classical style is displayed as early as 1922, in Lang's Dr. Mabuse Der Spieler. As a narrational mode, classicism clearly corresponds to the idea of an "ordinary film" in most cinema-consuming countries of the world.
The many variants of classicism make any overall periodization of the mode very difficult. Even the history of Hollywood norms is notoriously hard to delineate with any precision. This is partly because significant periods in the history of studios or technology will not necessarily coincide with changes in stylistic or syuzhet processes. Broadly speaking, we could periodize classical Hollywood narration on three levels. With respect to devices, we could trace changes within classical narrational paradigms, according to what options come into favor at certain periods. Here we should look not only for innovations but for normalization, patterns of majority or customary practice. Connecting scenes by dissolves is possible but rare in the silent cinema, yet it is the favored transition between 1929 and the late 1950s. On the dimension of narrational systems, we could study the principles that constitute narrative causality, time, and space. Spatial continuity within a scene can be achieved by selecting from several functionally equivalent techniques, but such continuity rests on broader principles, such as the positing of the 180 degree-line, or axis of action; and changes in this postulate can be traced across the history of cinema. We could also study the fluctuations of the more abstract narrational properties over time. For instance, narration in the 1920-1923 American silent cinema tends to be somewhat more self-conscious than in the later 1920s, chiefly because of a greater use of expository intertitles in the earlier period. Similarly, an insistently overt suppressiveness emerges in many films associated with the grouping known as film noir. I can here only hint at the manifold possibilities; we await a thorough history of classical storytelling and style.
Where, we might finally ask, does all this leave two important critical issues: authorship and ideology? In this space, only sketchy answers can be suggested. It seems evident that an auteur's work can be identified by its characteristic narrational principles and patterns. Hitchcock and Fuller's films are more selfconscious than, say, those of Hawks and Preminger. Moreover, we can associate /32/ consistent stylistic choices with directorial signatures: Ophuls' preference for tracking shots over cut-ins, Lubitsch's use of closeups. Most important is the fact that any distinct authorial approach to narration typically remains within classical bounds, creating extrinsic norms that conform to or amplify intrinsic ones. Authorial difference in Hollywood thus dramatizes the range and limits of the classical paradigm. As for the ideological significance of classical narration, all the principles and procedures I have considered could be analyzed in this regard. The goal-oriented hero; the appeal to principles of unity and realism, the functions of temporal and spatial coherence, the centrality of the invisible observer, the arbitrariness of closureeach bears the traces of social-historical processes of production and reception. The predominance of three-point lighting appeals to canonized conceptions of glamor and beauty; the treatment of heterosexual romance links Hollywood classicism to dominant conceptions of sexual relations. The 180-degree system not only bears the traces of a mode of production seeking speedy and economical filmmaking; it also continues a tradition of spatial representation at work since the Greek theater. Each film works with, or with and against, ideological and economic protocols.
What is important, however, is that even in this most ordinary cinema, the spectator constructs form and meaning according to a process of knowledge, memory, and inference. No matter how routine and "transparent" classical film viewing has become, it remains an activity. Any alternative or oppositional cinema will mobilize narration to call forth activities of a different sort.
 V. Propp, Morphology of the Folktale (Austin: University of Texas Press, 1968).
 This essay refers to material discussed at length in chapters 1 to 7 of David Bordwell, Janet Staiger, and Kristin Thompson, The Classical Hollywood Cinema: Film Style and Mode of Production to 1960 (New York: Columbia University Press, 1985). A general background for the discussion is David Bordwell and Kristin Thompson, Film Art: An Introduction (Reading MA: Addison-Wesley, 1979).
 Perry W. Thorndyke, "Cognitive Structures in Comprehension and Memory of Narrative Discourse," Cognitive Psychology (1977), 9:84-96. For an example of an approach to a different narrative cinematic mode, see my article "The Art Cinema as a Mode of Film Practice," Film Criticism (Fall 1979), 4(1):56-64.
 Eugene Vale, The Technique of Screenplay Writing (New York: Grosset and Dunlap, 1972), pp. 135-60; Stephen Heath, "Film and System: Terms of Analysis," Screen (Spring 1975), 16(1):48-50.
 See Bordwell, Staiger, and Thompson, The Classical Hollywood Cinema, chapters 14-18.
 Rick Altman stresses the need to consider the importance of character parallels as "paradigmatic" relations in the classical text. It is true that analogies and contrasts of situation or character occur in classical films, but these relations are typically dependent upon logically prior causal relatiorls. Rick Altman, "The American Film Musical: Paradigmatic Structure and Mediatory Function," in Altman, ed., Genre: The Musical (London: Routledge and Kegan Paul, 1981), pp. 197-207
 Christian Metz, "Problems of Denotation in the Fiction Film," in Film Language, tr. Michael Taylor (New York: Oxford University Press, 1974), pp. 108-46
 Raymond Bellour, "The Obvious and the Code," Screen (Winter 1974-75), 15(4):7-8. See also Alan Williams, "Narrative Patterns in 'Only Angels Have Wings,' " Quarterly Review of Film Studies (November 1976), 1(4):357-72. .
 Raymond Bellour, "To Analyze, to Segment," in Altman, ed., Genre pp. 107-16
 Thierry Kuntzel, "The Film-Work 11," Camera 0bscura (1980), 5:25.
 Parker Tyler, The Hollywood Hallucination (New York: Simon and Sehuster, 1970), p. 177
 Meir Sternberg, Expositional Modes and Temporal Ordering in Fiction (Baltimore: Johns Hopkins University Press, 1978), p. 17X.
. See Richard Dyer, Stars (London: British Film Institute, 1979), p. 65, and David Bordwell, "Happily Ever After, Part 11," in The Velvet Light Trap (1982), no. 19, pp. 2-7 i
 Bertolt Brecht, Collected Plays, ed. Ralph Manheim andJohn Willett (New York: Vintage, l977), 2:331. ..
 Vale, Technigue of'Screenplay Wrihug, p. 81. ~F
 Jean-Paul Sartre, "Quand Hollywood veut faire penser," L'Ecran francais (3 August 1945), no. 5, p. 3. .,
 I borrow the term from Seymour Chatman, Story and Discourse: Narrative Structure in Fiction and Film (Ithaca: Cornell University Press, 1978), p. 103.
 A. Lh1dsley Lane, "The Camera's Omniscient Eye," American Cinematographer (March 1935), 16(3):95. F
 The clearest statement of the "invisible observer" notion is to be found in V. 1. Pudovkin, Film Technique (New York: Grove, 1960), pp. 67-71.
 Andre Bazin, What Is Cinema? trans. Hugh Gray (Berkeley: University of California Press, 1966), p. 32.
 See Hal Herman, "Motion Picture Art Director," American Cinematographer (November 1947), 28(11):396-97, 416-17; Herman Blumenthal, "Cardboard Counterpart of the Motion Picture Setting," Production Design January 1952), 2(1):16-21.
 Susan Rubin Suleiman, Authoritarian Fictions: The Ideological Novel as a Literary Genre (New York: Columbia University Press, 1983), pp. 159-71.
 Herb Lightman, "The Subjective Camera," American Cinematographer (February 1946), 27(2):46, 66-67
 Yury Tynyanov, "Fundamentals of the Cinema," in Christopher Williams, ed., Realism in the Cinema (London: Routledge and Kegan Paul, 1980), p. 149 I have modified the translation slightly.
 Peter Bogdanovich, Alan Dwan (Berkeley: University of California Press, 1970), p. 86.
 Quoted in Thomas Elsaesser, "Why Hollywood," Monogram (April 1971), no. 1, p. 8.
 Because norms are guidelines that rank options probabilistically, we ought not to be too quick to disclose "transgressions" of classieal style. For instance, Peter Lehman claims that subjective framings of a character's to-camera stare in Dr Jebyll and Mr Hyde (1932) are "quite at odds with the usual Hollywood paradigm." Yet optical point-of-view shots are not forbidden by classical protocols; they are just less likely than other alternatives. Similarly, Lehmann points out a diseontinuity when Jekyll leaves an establishing shot and supposedly turns his back; cut to Ivy looking at the camera and tossing a garter at it. I would suggest three things here. First, the cues seem ambiguous as to whether Jekyll in fact turns his back; he could still be watching offscreen. A later shot, of his feet turned toward Ivy as the garter lands before him, reinforces j some such spatial hypothesis. Seeond, the playfulness of the point-of-view pattern is not unlike the whimsical play with space in Lubitsch and other innovative classical directors. Finally we should reeall that Dr. Jekyll and Mr Hyde begins with a lengthy traveling shot from Jekyll's optical point of view, before we have been introduced to the character. Optical subjeetivity thus constitutes an important part of the film's intrinsic norm. One could argue that Ivy's glance into an ambivalent offscreen eye simply amplifies the film's narrational norm. See Peter Lehman, "Looking at Ivy Looking at Us Looking at Her: The Camera and the Garter," Wide Angle (1983), 5(3): 59-63
 "There are, of course, periods tending toward maximally attainable harmony and stability; they are usually called periods of classicism." Jan Mukarovsky, "The Aesthetic Norm," in Structure, Sign, and Function: Selected Essays by Jan Mukarovsky, trans. and ed. byJohn Burbank and Peter Steiner (New Haven: Yale University Press, 1978), p. 54.
 Frances Marion, How to Write and Sell Film Stories (New York: Covici-Friede, 1937), p. 144.
 Richard Mealand, "Hollywoodunit," in Howard Haycraft, ed., The Art of the Mystery Story (Ncw York: Grosset and Dunlap, 1946), p. 300.
 It is thus somewhat misleading for Vance Kepley to assert that the restaurant scene in His Girl Friday creates "a shifting cinematic space not unlike what Burch finds in Ivan the Terrible and what other theorists find in such non-classical directors as Ozu." Eisenstein and Ozu make mismatches more prominent than does Hawks. The point is not that Hawks's scene has no spatial incompatibilities, but the classical spectator is simply cued to overlook them. See Vance Kepley, Jr., "Spatial Articulation in the Classical Cinema: A Scene From His Girl Friday," Wide Angle (1983), 5(3):50-58.
 Sternberg, Expositional Modes, p. 71.
 Michele Lagny, Marie-Claire Ropars, and Pierre Sorlin, "Analyse d'un ensemble filmique extensible: Les Films francais des annees 30," in Theorie du film, ed. J. Aumont and J. L. Leutrat: (Paris: Albatros, 1980), pp. 132-64.
 Noel Burch, "Fritz Lang: German Period," in Richard Roud, Cinema: A Critical Dictionary, ed. (New York: Viking, 1980), 2:583-88.