When Less Is Less:
The Long Take in Documentary
Vol. 46, no. 2 (Winter 1992–93): 36–46.
There is a hidden problem in documentary film—the problem of the long camera take and what to do with it. With the exception of interview material, most of the shots in contemporary documentary films and television programs are only a few seconds long. This is in marked contrast to fiction films and television dramas, in which whole scenes are played out in a single shot. Documentary thus finds itself in the curious company of television commercials and music videos in seeking to maintain audience interest through the dynamics and variety of quick cutting. The long take has become the terra incognita of the modern documentary film, a blank space in a practice which devotes itself almost entirely to other properties of the shot. And this is contrary to its heritage, for documentary was born in the pleasures of watching such ordinary events as leaves shimmering on a tree or a train arriving at a station.
Not long ago I spent months filming in the streets of a small town in northern India. The finished film is intentionally one of counterpoints and disjunctions and not at all a smoothly flowing narrative. Yet while I was filming, something odd occurred which I still don't fully understand. I began to shoot a kind of "shadow" film alongside the main film. This notional film—notional because it remains unmade—consists of long camera takes which quite clearly could never have been used in the main film. My main justification for shooting these long takes was that we could at least extract and use pieces of them. But in the back of my mind they actually constituted an alternative film, a counter-film to the one we were making. They formed a necessary antidote, a way of holding on to qualities which are so often lost when a film is structured for its likely audiences. I remember thinking at the time: "Is it possible to go back to zero, to film as if the cinema has just been invented? What would it be like to work like Louis Lumière when he first set up his camera on the street?"
Some of these long takes last five or six minutes (200 feet of 16mm); none are shorter than a minute or two. To watch these shots one must suspend one's
This is an expanded version of ideas which I first presented in a seminar as a Visiting Fellow at the Humanities Research Centre in Canberra, Australia, in 1989. I should like to thank the HRC for its support. I am indebted to E. Richard Sorenson and Allison Jablonko for the term "digressive search," which I use in a rather different sense from theirs; and to Roger Cardinal for the stimulation provided by his essay "Pausing over Peripheral Detail." I am also grateful to Leslie Devereaux, Gary Kildea, and Paul Willemen for discussions with me of some of the issues raised here.
usual moviegoing and television-watching expectations. But these expectations serve as a frame of reference for what I want to discuss here.
Like a spark or a stab of lightning, a shot discharges most of its meaning at once, within the first few microseconds of appearing on the screen. If we close our eyes after that first instant, the meaning survives. The mind arrests it like the shutter of a camera. What follows in our response may be very different—a sudden adherence to something happening within the shot, or a kind of coasting perusal. Or so it can be if the shot continues. But most shots are not allowed to. In film-making few shots are used in their entirety. Most are shot long and cut short.
Christopher Pinney has argued that still photographs are more indeterminate than films, offering the viewer more because they dictate meanings less. Social scientists in particular, he suggests, are afraid of still photographs and prefer film because "still images contain too many meanings whereas the desirability of film lies precisely in its ability to constrain meaning through narrative chains of signification. . . . They close off plural readings in the temporal flow of succession and destruction." The temporal and sequential structure of film thus "provides a fortification against undesirable and 'unwarranted' readings."
But applying this argument to film itself produces a curious reversal of Pinney's observations. Short camera takes resemble still photographs in their fixing of a single image, but by their very brevity they disallow the kind of perusal of the image over time permitted by photographs and by longer takes. Longer takes, which create sequential chains and the narrative cloistering of meanings, also undermine these very meanings by leaving the viewer more time to ignore or challenge them. It can thus be said that the long take comes eventually to resemble the still photograph more closely than the short take, at least in these "lexical" properties.
Just as shots may be short or long, so there are short films, long films, and occasionally very long films which are rarely seen. While no one would argue that how one reads entire films is analogous to how one reads individual shots, there is perhaps a connection between the visual context within which a shot is framed and the footage from which a finished film is extracted. Dai Vaughan writes of an ideal cinema, never to be fully achieved because tending toward an impossible conflation, "something which would attain to a narrative significance whilst remaining random." The films that have come closest for him have been certain documentary films for television shot in cinéma-vérité or long-take style: "Not the rushes, yet not the fine cuts and most certainly not the transmitted versions with their cellophane wrap of commentary and captions and studio presentation, but the films as they stood when their narrative structures had just begun to emerge with the patient chipping away of the surrounding substance, yet were still perceptibly of its density and of its mass."
I want to examine this problem of the ideal and the actual, the object within grasp yet somehow lost, and draw a broad analogy between the way in which shots are reduced in length in films and the way in which an entire body of footage shot for a film is reduced to produce the finished film. On the way I hope to question some of the assumptions which underlie these practices.
Disquiet in Documentary
Long takes were not always the exception. In the early days of the cinema, when all films consisted of a single shot, they were the norm. Louis Lumière's first films ran for up to a full minute uncut—the length of a roll of film at the time. Some of Georges Demeny's shots (filmed in the 60mm gauge as early as 1895) ran to 40 seconds. That is very long by today's standards, even in fiction films, although a few directors (Jancsó, Jarmusch) have created distinctive styles around very long takes. In television documentary the average length of a shot is closer to five seconds, excepting interviews and "talking-head" presentations. These shots tend to be cut automatically at the point where it is assumed audience attention drops, or where there is any suggestion of a pause in narrative flow.
The great enemy of documentary (and oddly, rather a taboo topic of discussion among film-makers themselves) is the "dead spot" in which nothing seems to be happening. Film producers are terrified of such moments, for they are terrified of audience impatience. I suspect that the taboo status of this topic goes back to an inherent contradiction in documentary principles. In the early days of cinéma vérité and Direct Cinema, the prevailing ideology had it that dead spots weren't supposed to exist. Ordinary life was deemed to be interesting and worthy of everyone's attention. But documentary film-makers still contrived to avoid dramaturgical dead spots, cutting around them or focusing on exciting events and famous people. Documentary, whatever its ideology, still took its shape from fiction or journalism. It had to defend its interest in the ordinary by making sure that the ordinary played well. There was a tacit understanding that you didn't talk publicly about this. Who cared to admit that documentary actually concealed the lacunae characteristic of ordinary life and chose only the best bits, just like the fiction film-makers?
What constitutes a "long take" is obviously an artificial and somewhat arbitrary concept, formed in relation to an average notion of shot length and affected by content and position as well as by duration. Long takes are perhaps better defined by their structural qualities than by their length. Does the shot, for example, form an entire sequence in the film, or is it merely part of a more extended, edited sequence? In this analysis, the term "long take" refers more
to a method of film construction than to actual length. Brian Henderson has pointed out that although Murnau uses a long-take style, his shots are actually quite short. In his films the viewer's attention tends to be focused more upon developments within shots than upon linkages between them.
It is also evident that shots of long duration are not necessarily more revealing than if they had been shorter—for example, shots of repetitive activities or shots containing limited information which is rapidly grasped by the viewer. It is no use comparing generically different materials. Duration is perhaps the least important criterion in comparing a static, practically empty frame and a frame crowded with activity. And yet . . . and yet, as I shall argue later, absolute duration does finally matter. It is not wholly subjective and has its own measure of influence upon our reading of shots.
The Viewing of Images
It seems almost self-evident that how long we look at an image affects what we see in it and how we interpret it. Even if there were no other evidence of this, it has been shown that the eye successively scans an image in a series of fixations. If the time for doing this is cut short, the eye fixes on fewer points and the mind creates a less extensive version of what David Marr has called the "primal sketch." In talking about viewing film images it is useful to place the process in the context of viewing practices generally. How does film-viewing differ from viewing other kinds of images, such as still photographs?
Sometimes the length of time we devote to a still photograph is determined for us, as when a train we are on flashes by a billboard with a photograph on it. The frequency with which we view photographs is also often beyond our control: it is the aim of advertisers to expose us to the same pictures as often as possible, although many photographs, such as those in newspapers, we see only once, and then usually briefly. Others, such as family snapshots, may be seen again and again—and we may choose to study them for quite long periods.
When we watch films we exchange the role of private consumer of images for that of public participant at a spectacle. Our choices become more limited. Not only are the still photographs of the film regulated to 24 or 25 per second, but the length of time we have to view each shot is precisely dictated. We thus surrender an important part of our control over the image, although not all of it. There is still the possibility of searching the shot and interpreting it to some degree independently—for example, by looking for "peripheral detail." How we interpret it depends upon who we are and what assumptions we bring to it. This is a fertile process, the text of the film interacting with the texts of personality, culture, and society that define us. Despite that, there are
habits of film-viewing which will hold broadly true for audiences with a shared set of cultural expectations. If the following description is in any way recognizable, it is because it applies to a quite specific set of film-making and film-viewing conventions.
Responding to the Shot
For Western, middle-class viewers (at least), the initial response to a shot is determined both by its content and placement in the context of the film and by various plastic and compositional elements of the shot itself. The audience, from its grasp of the context, quickly identifies the intended center of signification of the shot. In a typical character-centered film, for example, imagine that a person whom we have already seen walks down a street and encounters a stranger. Our attention attaches to this person, is then transferred to the stranger, and then perhaps shifts back again to the familiar character. We take in certain background details, but we identify the primary meaning of the shot as residing in what happens to the major character. This primary meaning—perhaps related to what Eisenstein termed the "dominant"—need not be a person, or even a specific visual object. For example, a slow pan over a city may simply signify "a sunny morning in San Francisco" (as in Hitchcock's Vertigo , 1958).
Dramatic films thus extend to us a challenge which is a little like a game. We are invited to participate in creating the meaning of each shot by recognizing its narrative or expository center. The length of the shot is gauged so that we must carry this out fairly quickly, leaving little time for other considerations. This contrast between a centered meaning and other coded and uncoded information in the shot may be thought of as a figure-ground relationship. What is identified as figure, and what as ground, is a result of placement and, as Nick Browne has shown for sequences of shots, may also shift and depend upon duration. Previously noted details may be brought forward retrospectively by a new context. Centered meanings may be forgotten in a process Browne calls "fading."
However, there is a certain threshold of narrative or expository efficiency beyond which the motivated meaning of the shot is exhausted. If the shot unexpectedly remains on the screen without further developments, we may feel impatience or annoyance, during which we perhaps look away or withdraw our attention. If the shot continues still longer we may move to a third stage of what might be called "digressive search," when we begin to bring a very different and more idiosyncratic kind of interpretive process to bear upon the shot. In films like Andy Warhol's Sleep (1963) and Blow-Job (1964) our expectations are deliberately confounded and we are provoked into supplying
the images with meaning. Audiences, however, are generally asked to stretch the rules only so far. And when they are asked to do so they are usually offered compensations.
How we respond to a shot is shaped not only by our conventional expectations but also by the rules that the film itself establishes. In Stanley Kubrick's 2001: A Space Odyssey (1968), for instance, shots early in the film are purposely lengthened considerably beyond the norms of Hollywood editing. The result is that when the climaxes come we accept that they develop at an almost dreamlike pace.
The viewing of film shots may also be affected by largely neurophysiological processes which are still not well understood. There may be a point at which the recognition of any sign becomes subject to a certain cognitive loss or slippage. For example, after a period of time our attention may automatically shift from a particular visual figure or thematic focus to "ground" or background material. This process may be related to the directional switching that occurs when we study the diagram of a cube, or to the experience of figure-ground switching familiar from such examples in Gestalt psychology as seeing a picture alternately as a vase or two symmetrical faces. It may have to do with the different functions of the two hemispheres of the brain, or with the way in which different cells in the visual cortex respond to highly specific shapes in the environment. It is possible that digressive search is triggered by such processes, so that a search for alternative configurations and meanings follows the "saturation" of an initial act of recognition.
Such a schematic description of how we read film images cannot, of course, pretend to deal with the many convolutions of pattern recognition or the kinds of layered responses that may be part of reading the denotative and connotative content of complex images.
From Rushes to Films
Few documentary film-makers would deny that their films are highly selective and expressive of a particular culture and ideology. At the same time, when film-makers measure their films against their experience of the world they often find them lacking. What has recently been referred to as a "crisis of representation" in a broad spectrum of human studies has resulted from just such a sense of discrepancy between experience and the existing paradigms for representing it. Such an intellectual climate may now encourage film-makers to pursue this sense of discrepancy a little further.
It is true that documentary film-makers have periodically questioned the inherited assumptions of their vocation. This has happened notably around the
years 1935, 1960, and 1975. Ethnographic film-makers, in their brushes with other arts and rhetorics, have perhaps been particularly inclined to do so. But even they have seemed unwilling to confront perhaps the most deeply seated assumption of all: that films are necessarily superior to the raw materials shot for them.
Despite this, it is not uncommon to hear film-makers say: "The real film was in the rushes." I have an instance of this from Roger Graef, the maker of many documentary films for British television. In an interview with Alan Rosenthal he says, "In one film we shot 100,000 16mm feet. That's fifty hours. . . . These vérité films are usually best in the rushes. All fifty hours tend to be interesting. It's like a long-running serial. Strangers wandering through our viewing rooms tend to sit there and come back, and back, and back because they want to know what's going to happen next. It's got that kind of excitement to it. There is a problem in structuring them. The films tend to be next best at something like six or eight hours. . . . And then there's a terrible problem because all of the subplots, all the nuances, all the things that aren't going to survive, but do feed the sense of reality, all have to be cut."
There is in descriptions like this, and in the experience of many film-makers, a pervasive sense of loss which is not about a quantitative difference but a qualitative one. It is as though once a film has been pruned to achieve what it actually sets out to achieve—a coherent narrative or analysis—certain qualities perceived in the rushes have been edited out of it. This contradicts the accepted notion of creative economy that "less is more," and "the work is greater than the sum of its parts." The feeling seems to be that the work has clearly become less than the sum of its parts. It is not merely a reduced semblance of the longer work but has been reduced in other important ways in achieving its final statements.
This is not to revert to the naïve view that film footage is some kind of unmediated evidence which contains the "truth" about external reality. If that were so there would be little point going beyond the rushes themselves. It would also deny that editing does in fact introduce its own higher order of truth and understanding. Rather, the sense of loss seems to identify positive values perceived in the rushes and intended by the film-maker at the time of shooting but unachieved in the completed film. It is as though the very reasons for making films are somehow contradicted by the making of them. The processes of editing a film from the rushes involve both reducing the length overall and cutting most shots to shorter lengths. I believe there is a connection between these two processes, in that they both progressively center particular meanings. Sometimes film-makers appear to recognize this when they try to preserve some of the qualities of the rushes in their films, or reintroduce those qualities through other means.
Much of what is lost from the rushes is a sense of the historical contingency of the images—the actual conditions under which films (and meaning) are produced. Film rushes are as much a chronicle of a film's production as they are of its supposed subject. The excitement Graef describes—of wondering what will happen next—is really the excitement of sensing that in the rushes anything can happen next. While finished films suggest a past tense, rushes seem to unfold in the present tense of a camera running. What editing removes are the stigmata of this historical moment. The shots that remain have been domesticated. They are neither tangential nor contradictory nor incontinent nor otherwise incapable of being marshalled to the film's purpose.
What does one lose, then, from the rushes? One loses, I think, qualities of spaciousness, context, and historicity, and these can be described in four different ways.
Qualities Sacrificed to the Film
First, one loses excess meaning —meaning in excess of what the film expresses and requires. This is not merely what remains unexplored in the subject that can still be found in the rushes, but all material which escapes from what might be called the "economy of signification" of the film.
Second, there is a loss of interpretive space —a closing-off of the legitimate areas in which the viewer is invited to supply meaning. The film dictates a certain standard of relevancy. As it moves toward its final form, the background around the centered subject is gradually whittled away. This controls the viewer's relationship to the footage in two ways: first, the background is made to appear incidental and subservient to what the film designates as a sufficient reality. Second, the background itself is physically thinned out by cutting, thus further reducing the opportunities for "irrelevant" intervention. Although different films provide different kinds of interpretive space for the viewer, this space often merely allows the audience to endorse the film-maker's meaning rather than to participate more actively in creating it. Viewers of rushes, by contrast, constantly interject their own interpretations.
Third, there is a loss of the sense of encounter . As the film becomes a polished, professional work, its connections with the historical act of filming, which were so evident in the rushes, gradually disappear. This is especially true of television documentaries, which typically begin with a title sequence whose purpose is to characterize the program as a fully packaged (and therefore predigested) institutional product.
Fourth, there is a loss of internal contextualization . In editing a film to its final length there is an inevitable loss of material which would otherwise clarify and extend the meaning of the material which is retained.
Throughout the editing process there is a constant tension between maintaining the forward impetus of the film and providing enough contextual information so that the central narrative or argument continues to make sense. As the film becomes shorter, the analysis becomes cruder. Film-makers continually sacrifice footage which they know would permit a more complex understanding of the subject but which, for reasons of length, the film cannot afford. To solve this problem, such gaps and elisions are often roughly patched over with spoken voice-over commentary.
The World within the Shot
Films and shots are complex structures, each evoking a larger world. Just as there are levels of contextual material within the footage shot for a film, so there are levels of contextualization within the shot itself. Loss of context can occur in discarding footage, but it can also occur when individual shots are made shorter.
Film-makers are aware of this. Sometimes they include an occasional long take simply to reinject into their film some of the qualities perceived in the rushes. But for a few film-makers the long take becomes a way of redefining the terms in which the film addresses its audience. Such an approach does not necessarily imply a realist aesthetic of the kind championed by André Bazin and many of the Italian neorealists. Brian Henderson notes this in the case of Ophuls, whose long takes can be highly choreographed, and in the case of Godard. He describes a long take in Godard's La Chinoise (1967), which tracks past the shacks of Algerian workers in Nanterre to a modern university complex. "Eisenstein," he writes, "would have cut from a shot of the one to a shot of the other, making the juxtaposition for the viewer, obliterating time and space relations to make a clearcut social relation. Godard observes the time and space relation and lets the viewer make the social relation. . . . He does this by virtue of the long take's continuity of dramatic space and time, which this usage reveals as itself a form of argumentation or demonstration; the shot has its own internal relations, its own logic. This instance of the shot seems Bazinian but, far from fidelity to the real, Godard rips this bit of footage from its grounding in the real and puts it down in the midst of a highly abstract film essay."
In this shot Godard uses a long take to create what Walter Benjamin called a "dialectical image"—an internal contextualization of a specific kind, in which one foreground element is qualified by another. But long takes permit contextualization of several kinds. They reveal relationships which link foreground with background, they reemphasize the objective presence of disparate physical objects in the shot, and they provide the "stage" for the enactment of human behavior which reveals individual identity.
Foreground/Background Relationships: A simple instance of linking between foreground and background within a shot is the way in which a moving camera defines the geography of a space. The perception of spatial relations is always a problem in the cinema because of the monocular vision of the camera, but by shifting the perspective, camera movement allows us to make sense of these relations. This movement must of necessity occur over time. A similar kind of spatial linking is produced by the quite different movement of people or objects within the frame. Thus the long take may be crucial to defining the geographical context within which a character exists or an action takes place. It is also obviously important in delineating actual matters of time, such as how long it takes someone to perform a particular task—something which is normally masked by the condensation of edited sequences.
Long takes can also reveal the relations between simultaneous actions and co-existing objects in one setting. These may be complex personal interactions or (as in the Godard example) connections between people and their surrounding social and economic environment. The objective conditions and historical processes which shape people's lives may often be more effectively demonstrated by appearing in the same frame than by being shown in the juxtapositions of editing, provided we are given a sufficient intellectual framework for interpreting them. This interpretation may require a conscious reapplication of detail from the margins of the film to its center.
The Persistence of the Physical: As I have mentioned, we sometimes subject an image to a process of digressive search. We inspect details which escape the film's inscriptions of meaning, resisting what Roger Cardinal calls the "fixation on congruity." Such details can play a role in film which goes beyond either the pleasures of discovery or a merely supportive "authenticity."
Realist documentaries have tended to rely on background detail to legitimate their choice of what is significant in the foreground, just as historical dramas provide set dressing of the proper period to make us accept their version of history. The long take, however, can serve the opposite purpose: to assert the independence and autonomy of a physical, "background" world and the constantly shifting relations, or lack of them, between material and social being. Presented in this way, physical objects reassert their stubborn and oblivious existence—what Barthes would call their "obtuse" presence. They may even appear as surreal, not because they invoke the irrational or the unconscious, but because they force upon us recognition and confrontation with the unnamed and the unremarked.
The Dimensions of Personhood: Finally, the long take can make possible a contextualizing behavior which may be essential to recognizing individual human identity. Over time, details about other human beings accumulate for us and eventually coalesce into distinct personalities. There is perhaps a parallel here with John Berger's distinction between the private photograph, produced
and consumed in a context of familiarity, and the public photograph, torn out of its context and presented to strangers. The challenge for the photographer, says Berger, is to restore context to the public photograph.
In daily life it is our observation of people over time which causes us to transform undifferentiated strangers (or human types) into known individuals. Film shots, unlike still photographs, can provide the necessary time frame in which sequences of behavior can unfold, allowing us to distinguish what Gombrich calls the "likeness" from the "mask." Likenesses emerge as continuities in the midst of variations. According to Gombrich, "[T]he film shot can never fail as signally as the snapshot can, for even if it catches a person blinking or sneezing the sequence explains the resulting grimace which the corresponding snapshot may leave uninterpretable."
In documentary, the long take can help redress the decontextualization of the fragmentary public image, or in Berger's terms, restore the context of the private. Such recontextualization can be seen clearly in ethnographic films, which for much of their history have defined people of non-Western cultures by their roles or occupations, as anonymous actors in exotic social mechanisms. They were almost always mask, never likeness. Longer camera takes with synchronous sound and subtitled dialogue provided a means of refiguring the relationship between the person on the screen and the viewer. More effective than narrative or other humanizing strategies, it was these uninterrupted passages of behavior which, despite cultural differences, gave the necessary clues to discovering the person within the indigenous social actor.
The long take has been associated with the very earliest motion pictures and with two recent periods in the history of documentary. In the cinéma-vérité and Direct Cinema films of the 1960s it was used to record extended events and conversations. In the political and biographical films of the 1970s and 1980s it was used largely to record interviews. These latter films were strongly influenced by television journalism, which produced its own special use of the long take in building programs around eminent talking heads. The long take proved equally serviceable in discovering these heads in European cathedrals, on Andean railway journeys, or amidst the flora and fauna of African jungles.
In each of these cases, the long take has been used for quite specific and, arguably, quite narrow purposes. Even so, some of these uses have been equivocal or self-contradictory. In the period of cinéma vérité and Direct Cinema, an important model for documentary was Italian neorealist cinema—films like La Terra Trema (1948) and Umberto D (1952), which themselves borrowed ideas
from earlier styles of documentary. But the attempt to reproduce in documentaries the literary qualities of fiction (as in Salesman , 1969) tended to confine the use of the long take to largely narrative, quasi-fictive, or authenticating functions. Paradoxically, many of the other potentialities of the long take—for articulating space and time, relating people to their environment, exploring human personality—were being more adventurously investigated in fiction, in the work of directors like Godard, Antonioni, Resnais, and Rossellini. In the second period of interview-based documentaries, the long take seems to have been devoted almost entirely to creating an oral narrative and establishing the authority of the interviewees.
There have of course been alternative tendencies and exceptions to this pattern. Leacock's Queen of Apollo (1970), Rouch's Tourou et Bitti (1971), Wiseman's Hospital (1970), and Kildea's Celso and Cora (1983) all use long takes in distinctive and sometimes idiosyncratic ways. Experimental (and "underground") films have provided other kinds of explorations. One should include here, along with the films of Andy Warhol, the work of Michael Snow, Stan Brakhage, and Trinh T. Minh-ha. In other recent documentaries, film-makers like Amos Gitai and Claude Lanzmann have used the long take to subvert the traditional construction of foreground and background. However, for most film-makers there remain serious obstacles to developing these possibilities. Film length is one of them. Films must either conform to conventional lengths, using fewer shots, or develop into much longer films. But who will watch longer films, especially if they willfully include the dreaded "dead spots" of ordinary life as legitimate content?
Segmentation suggests one possible strategy. There have been a number of experimental documentary series for television, such as Craig Gilbert's An American Family (1972), Roger Graef's Police (1982), and Melissa Llewelyn-Davies' Diary of a Maasai Village (1984), but so far these have tended to reproduce in documentary the interlocking story structures of drama series or have been composed of essentially self-contained episodes. In neither case has segmentation led to a noticeable expansion of conventional film time to allow for longer takes.
Ultimately the problem of film length is related to the larger problem of how to articulate longer shots to produce meanings. Without commentary, conventional documentary editing usually finds long takes intractable unless they are tracing a clear narrative line, as in Graef's films. Using longer takes gives fewer opportunities to signify by means of the cuts between them. Longer takes are also likely to be complex entities, creating problems of intellectual focus. They characteristically contain ambiguities, interruptions, and competing centers of attention. The content is mingled in ways which make it difficult for the film-maker to isolate "signal" from "noise." In scripted fiction,
"noise" is generally present only when it is put there on purpose to create verisimilitude, but documentary footage is rarely so tidy. Voice-over commentary has traditionally provided one means of superimposing meanings upon such material, but always at the cost of distancing it and reducing the viewer's engagement with its physical immediacy.
These obstacles are of course only obstacles in the context of a specific set of film-making conventions and viewing practices. The real test is whether long takes can find a place in quite new communicative structures. New technologies and shifts in popular culture at least open up certain possibilities for this to happen.
First, viewers' expectations of films are likely to change as some film-making practices which are now marginal enter the mainstream. This could alter the ways in which people actually "read" long camera takes. At the moment, the tendency in commercials and music video seems toward ever shorter takes, but this could contribute to a greater tolerance for associative, non-narrative editing and eventually for more films patterned on structures other than conventional stories or arguments. Films may emerge which require greater retrospective reconstruction in the mind. Against this current must be put the way in which the formats of television journalism seem actually to have narrowed the structural repertoire of documentary.
Second, unexplored opportunities exist for combining words with images, perhaps especially with long takes. One could cite the use of multiple voices on the sound track, voices used in less regular patterns, voices addressing us in new registers. There is no equivalent in documentary, so far as I know, to the whispered commentary which accompanies live golf telecasts. Words may also be deployed more effectively in titles and intertitles, as they once were in Soviet silent films and have been occasionally since, in such films as The Village of Furuyashiki (1982), by Shinsuke Ogawa. The history of documentary contains other experiments worth examining and pressing further, such as the use of spoken verse in the documentaries of the 1930s. One might expect certain parallel developments to evolve from the emergence of rap videos.
Third, one can imagine more complex layerings of sound and image. As precedents one can point to Godard's "middle period" films (British Sounds , 1969), Clément Perron's Day After Day (1962), and several of Gitai's documentary films (Ananas , 1984; and House , 1980). Sounds can make us reinterpret what is nominally background and, on some occasions, reconstitute it as thematic foreground.
A fourth strategy open to documentary is to make a much more consciously analytical use of the camera. Reframing with the camera resembles a form of montage which selects, connects, and juxtaposes different images, but in "real time." In fiction films it is possible for such an approach to be scripted, as in
Hitchcock's experimental Rope (1948). In documentary the situation is very different, requiring on the film-maker's part an ability to impose a process of thought on the camera's movements while filming unpredictable material. So far few film-makers have adopted such a demanding interpretive stance while filming or have developed the skills to accomplish it.
But camera movement within a shot allows for certain kinds of irony which are not possible with shorter takes. In Claude Lanzmann's Shoah (1985), the long take makes particular reference to the fact that however long one pans over landscapes where atrocities took place, one still sees only landscapes. In effect, one looks in vain for the signified in the signifier. In a number of Godard's films (Weekend , 1967; British Sounds ) long tracking shots, instead of following characters, as is usual in fiction films, track past them, fixing them not in relation to the film but in relation to their physical and social setting.
Lastly, it is worth noting that new technologies may have a profound effect upon viewing practices and, eventually, upon film form. We have yet to absorb the full implications of television. Video, an even more recent phenomenon, combines the privacy of television viewing with much greater control over the selection of viewing material. This could make possible longer works, organized in chapters or clusters of related films. Video also makes it easier to recast old films in new forms, or produce new commentaries on them. As for interactive video, it may, by giving the viewer even greater control over the investigation of material, generate much more exploratory viewing practices and eventually stimulate the new film-making practices that would allow for them.
Neither film-making nor documentary will be revolutionized by the long take alone, nor should it be claimed that the long take is in any sense the special province of documentary. But the question of what to do with the qualities which are found in long takes, and yet not found in the films derived from them, is perhaps the quintessential problem of documentary. It brings us closer to the paradox of reduction which lies at the heart of all representation, but which has none of the same implications in fiction that it does in documentary. For within every documentary is a kind of cavity, the negative imprint of the missing persons and events which are not there. In struggling with this material, the documentary film-maker is struggling not only with signs but quite literally with the shadows of the living and the dead. If photography does not steal the soul it steals something very like it, something deeply enough felt to generate the fraught ethical debates which uniquely surround the making of documentary films and photographs. These debates more commonly concern what is shown than what is left out. But for the film-maker the problem is truly one of disposing of the human remains.