Continuum: The Australian Journal of Media & Culture
Vol. 7, No. 2, 1994
Screening Cultural Studies
Edited by Tom O'Regan & Toby Miller

Beyond film: Branigan's narrational world

Horst Ruthrof

Review of Edward Branigan, Narrative Comprehension and Film. London and New York: Routledge, 1992. (Distributed by the Law Book Company Ltd.). 325pp. ISBN 0415075114. (pbk), $45.00.

If Point of View in the Cinema introduced its author as one of the leading film analysts by its attention to the details of the process of cinematic presentation, Narrative Comprehension and Film establishes Edward Branigan as a creative theorist beyond the boundaries of film. The book's voice is that of a craftsman speaking from his workshop. No deconstructive symplok­ and certainly no rhetorical terrorism. Instead, we find a certain modesty of style, which is deceptive considering that Branigan offers a great deal of substance and a range of attractive speculative insights. The author's mastery of technical intricacies within the broader frame of general narrative makes Narrative Comprehension and Film an outstanding teaching book for film studies as well as other disciplines in the humanities. What makes it especially appealing is its careful elaboration of an inferential account of how we make sense of narrative. For this and other reasons, it is well worth paying closer attention than is perhaps usual for a review article to how the book's argument unfolds and in particular to how it manages to relate the double argument about narrative in film and human perception as interpretive construals.

Branigan opens his first chapter "Narrative Schema" with an emphasis on the ubiquitous nature of narrative and a perhaps somewhat tight delineation of what separates narrative from non-narrative, excluding such texts as inventories, sermons, letters, summaries, birth announcements, medical statements. Narrative is seen as a special form of 'assembling and understanding data'. Without stipulating any rigid boundaries the author suggests that we can approach the question of narrative most fruitfully by conceiving it as straddling two of four types of text: narrative fiction, narrative non-fiction; non-narrative fiction, non-narrative non-fiction (1). Narrative is regarded consistently as a broad category of which texts are only a part. In this respect Branigan makes a strong case for narrative to be understood as a feature of general perception, which 'organises data into a special pattern which represents and explains experience' (3).

By the end of the book this description will have undergone substantial modifications when the author gets to addressing postmodern text strategies. But to begin with, the reader must endure the order of Aristotelian laws. Classical narrative is shown to function as a recursive structure of representation which organises 'spatial and temporal data into a cause-effect chain of events with a beginning, middle, and end that embodies a judgment about the nature of the events as well as demonstrates how it is possible to know, and hence to narrate, the events' (3). The reference to gnosis is anything but spurious. As he says much later, 'knowing, and thus being able to tell, is a fundamental property of narration' (53). In Branigan's scheme, narrative, cognition and knowledge pertain equally to film and life.

With reference to Todorov's causal-transformation theory of narrative, Heath's theory of displacement and other accounts, Branigan gradually unfolds his central concern: the way in which spectators cannot help but construe narrative from visual data. Working from the axiomatic assumption of 'a pragmatics of comprehension' (8) he questions narrative theories that rely primarily on linguistics and formal logic as inappropriate for 'pragmatic situations' in which 'fuzzy concepts, metaphorical reasoning, and "frame-arrays" of knowledge' are employed by readers (9).

What Branigan sees as crucial in any account of narrative is that, instead of remembering detail by detail, we tend to understand narrative by 'constructing large-scale hierarchical patterns which represent a particular story as an abstract grouping of knowledge based on an underlying schema' (16). Beyond film and other specific forms of signification, the author speaks generally of 'our ability to use a narrative schema to model a version of the world' (17). In film our schematic constructions are typically guided by eight functions: 1. abstract (title) or prologue; 2. orientation or exposition; 3. initiating event; 4. goal (statement of intent); 5. complicating action; 6. climax and resolution; 7. epilogue; and 8. narration (18). The theoretical asymmetry between narration and some of the other functions is noted not only by the reader; a little later the author observes that narration 'is not really on a par with the other elements of a narrative schema, but rather stands in for the operation of a still deeper schema that drives the story' (31).

At the same time the spectator realises filmic elements as a series of causal evaluations: as consecutive (nexus 'and'); as chronological (nexus 'then'); as aligned to social or generic convention; as remote causes mediated by intervening causes, the last of which is understood as the proximate cause; as enabling cause necessary for others; as direct cause sufficient for others (if there is more than one direct cause we experience the filmic world as overdetermined); as unique cause that is both necessary and sufficient (26f.). Yet such causal explanations of film are unsatisfactory unless they are complemented by an account of '"top-down" frames of reference' (27) that are not generated by specific clusters of screen data but rather in relation to a broader cognitive matrix.

This kind of argument cannot be contained within the confines of film. With reference to Schank and Abelson's Scripts, Plans, Goals and Understanding the author highlights 'a narrative schema, together with a host of related schemas', as symptomatic of the 'interest we take in the world as humans'. Such schemas are interpretive grids with which to sift 'cultural assumptions and values'. As a result, '"causes and effects" emerge, as it were, after the fact' as descriptions of how a chain of events coheres from the perspective of a chosen schema (29).

"Story World and Screen" unfolds the moves which film viewers typically make in transforming the 'bare facticity of graphics on the screen' (which is no more than a flat 'wash of light and dark') into the cinematic quasi-world. Branigan's constructivist view presupposes a double interpretation of screen data. At its simplest, our judgment construes from the same data either a square or a box, depending on how we locate the data in the story world. More generally, we are always dealing with at least two interpretive horizons: 'the space and time of a screen' and 'the space and time of a story world' (33). Again, these same principles are shown to apply equally to our construals of causality in our everyday world.

Branigan also addresses the traditional, more specific distinction between the diegetic and non-diegetic: the former consisting of data 'represented as being at least potentially accessible to a character', while the latter is made up of features which 'are not accessible to any of the characters'. Branigan adds the proviso, however, that the diegetic is nevertheless also 'about the diegetic world of a character' and so assists viewers with their interpretive inferences (35).

What is important then is how data are translated from story to screen and visa versa. Pertinent to this description are three kinds of relationship: the relation between 'diegesis and story' (Todorov's transformations; or Branigan's narrative schema); the relation between 'diegesis and screen' (Branigan's kinds of narrative causes: enabling, intervening, remote, unique, direct); and the relation between the diegetic and non-diegetic (modes of narration and questions as to fictional and nonfictional reference) (36).

At this point Branigan introduces the distinction borrowed from cognitive science between 'bottom-up' and 'top-down' perception, the first being 'serial and "data-driven"', and the second indirect impositions on data, reframing given information in a number of possible ways. Much as in ordinary cognition, this allows viewers to traverse their experience with considerable freedom, experimenting with various 'syntactical, semantic, and referential hypotheses' (37f.). The many possibilities of interaction between bottom-up and top-down processing suggest the description of perception as a systemic interaction between more or less compatible interpretations of data.

Here cognitive science particularises phenomenological insights about the relation between theme and horizon, whereby every shift in attention to detail involves, or even depends on, a shift in our overall horizonal world. This applies to perception as well as to large-scale cultural explanations, active especially in the dependence of the description of single phenomena on large-scale theoretical frames, be they conscious or non-conscious. In this, Branigan is rightly sympathetic to both the technical observations made in recent cognitive science and the phenomenological emphasis on the constructive role of the spectator.

Narrative in such forms as conversation, pictures, dance, music, is regarded as a coherent construction of cognitive effects. Yet the relation between top-down and bottom-up processing is presented as anything but smooth. This is why Branigan describes the interpretation of narrative as 'the moment by moment regulation of conflicts among competing spatial, temporal, and causal hypotheses' (39). Such interpretive accounting for data is said to produce narrative in film as an effect.

Going more deeply into the implications of this umbrella description, Branigan distinguishes eight different kinds of temporal relationships which schematize 'varieties of story time created as a spectator relocates the on-screen time' of a spatial fragment (B) 'relative to the on-screen time of a spatial fragment' (A). The result is shown to be a 'new and imaginary temporal order in the story, relationship ABn' (41). In this way, Branigan is able to account for the construction by the spectator of 'temporal continuity, ellipsis, overlap, simultaneity, reversal, and distortion'. Applied to Jean-Luc Godard's Weekend (1967), the schema generates a persuasive analysis of distorted time as 'a retrospective, nonsubjective flashforward' (42). Likewise, spatial patterns such as chains, gaps, reversals, and distortions make sense in relation to filmic temporality.

However, when the gap between screen and story space cannot be successfully closed by the spectator, the book speaks of 'degrees and kinds of "impossible" space'. This is a situation in which space is not experienced as a fully integrated part of diegesis. As a result of impossible space, the spectator is forced to re-examine earlier 'hypotheses about time and causality' (44). Similarly, story time is not a straightforward consequence of screen time, but rather is inferred with the assistance of 'top-down processing seeking a new order which will be sensitive to other constraints on the data, e.g. presumed causality and event duration' (45). More generally, all inferences concerning the filmic object world are primarily determined by top-down processing rather than by the immediate absorption of retinal information. As the author puts it, 'evidence on the screen cannot speak for itself' (46). Indeed, there is always also an absence of data; what is given is never complete. As a result, the spectator always performs an interpretive selection dictated by top-down master frames chosen to satisfy specific interpretive demands. In this context, Branigan highlights the persistence of top-down readings. In terms of general perception, his position also bears on the description of the self. A one paragraph potted history of theories of the fragmented self from Hume to Minsky and the poststructuralists is a good example of how Branigan's theory of film narrative leaks fruitfully into much broader perspectives (46).

What does it mean then to see an event? It involves decisions as to 'where actions begin, how they break off, and which actions belong together' (48). We require hypotheses concerning time if we wish to construe causation and, vice versa, hypotheses concerning causality in order to project temporality. Moreover, 'camera movements, matches on action, perspective relations, attached and cast shadows, optical transitions, screen direction, sound overlap, off-screen sound, voice-over and voice-under' conspire to make our inferences a complex web. For example, something as seemingly straightforward as 'the break between shots' does not normally amount to a meaning in itself; 'it is merely a catalyst for us to proceed to generate hypotheses' (48). And more often than not, such hypotheses turn out to be misguided in the light of further information. For in film and, one might add, also in ordinary perception, Branigan's reference to Roland Barthes makes sense, since it is he who observed that narrative operates on the principle of the logical fallacy post hoc, ergo propter hoc (50).

Stressing the active role of the film viewer, Branigan draws our attention to what above all else distinguishes for instance an eyeline match from a point-of-view shot. It is not the mere technicalities of production that counts, says Branigan, but rather the guided, interpretive, inferential work the spectator performs. Accordingly, the eyeline-match shot demands the construal of 'what a character sees and when the character sees it'; whereas the point-of-view shot constrains us to construct, 'in addition, the perspective of the character' (51). Hence Branigan's description of point-of-view shot constructions as 'semi-subjective effects' (51).

This leads the author to the specifics of the glance. Talking about Fritz Lang's Dr Mabuse, The Gambler (1922) he asks, 'Why should a glance have so much power to generate a narrative?' His answer is worth quoting:

a glance bristles with implications about space, time, and causality. A glance leaps across space...A glance implies temporal relationships...a glance may be linked directly to a character's intention, or to a forthcoming act by the character, or to a reaction...A glance implies an interaction with an object...the only glance that is generally avoided is the glance into the lens of the camera. A look into the camera breaks the diegesis because it makes the conventional reverse shot or eyeline match impossible. (Such a match would reveal the camera itself; its absence would be just as revealing.)...In a deeper sense a character's glance is an important measure of the acquisition of knowledge by character and spectator. (53)

If we accept the general thesis that story is always primarily constructed from the top down then it follows that story 'is most certainly not visible' (61). According to Branigan, there are four basic ways in which this invisible construct comes about as the result of the relation of hypotheses and screen data: 1. reinforcement; 2. as completion of a pattern analogous to a story pattern (typical of Hollywood narrative film); 3. independence; and 4. contradiction (typical of experimental film) (61). These four types are responsible for a 'basic tension' in our construction of film narrative. And finally, every story is also always the result of a double construction. All narrative is bound to tell 'the story of its own telling' and, vice versa, we as viewers cannot escape 'retelling a story which exists less on the screen than in our predisposition to make sense' (62).

In "Narration" Branigan poses two interrelated questions: 'how is narrative comprehension affected by the particular way we imagine we are seeing events?' (64) and 'How is it possible for us to possess the knowledge we come to possess in a narrative?' (64f.). His answer involves two crucial concepts by which to analyse narration as well as narrative theories. The how of narration concerns 'how an event is presented, how it happens, rather than what is presented or what happens'. The how perspective is important in considering the problem of readability (65).

The how and the what of film, its narration and its narrative, are two different domains of knowledge: 'procedural knowledge and declarative (or postulated) knowledge' (65), the former producing the latter in the sense that it is the filmic procedures which facilitate the viewer's construction of narrative events. Complicating the picture further, Branigan draws the reader's attention to what he calls 'the notion of a disparity of knowledge' (66). This takes up a traditional critical tool from the analysis of drama, the 'discrepancy of awareness'. If we add this basic asymmetry of knowledge to the how and what structure of film narrative, we can say that narrative occurs when 'there is a disturbance or a disruption in the field of knowledge'. The most general principle which follows from this for Branigan is that narration depends at least on 'a subject in an asymmetrical relationship to an object' (66). In this respect the author notes the excess of narration in 'flamboyant genres, such as melodramas and television soap operas' in which spying, eavesdropping, and gossip abound (67). Melodrama in particular exploits this relation by representing the "same" event from different points of view, contributing different degrees of disparity. These principles are illustrated by an analysis of eight distinct camera positions each of which restricts the visual data in a different way (67f.). Thus narration is a technique of producing knowledge which Branigan describes as 'intermittent' (69).

"Hierarchies of Knowledge" offers the view that theories relying on bottom-up processing of screen data along categories such as distance, depth or alignment are misleading in that they cannot account for the large amount of inferential work which a spectator must perform in order to make a story out of visual data. Obvious as it may seem, it is important to stress that the spectator always knows a great deal more than is provided by the screen. Competing information has to be sifted and thematically arranged and rearranged, with the result of a marked discrepancy between the epistemic status of movie apparatus and film viewer.

In "Forgetting and Revising", Branigan shows how the perceptual processes of erasure of image data become productive in film narrative. Film comprehension very much depends on the fact that 'the spectator routinely sees what is not present and overlooks what is present' (83). This of course flies in the face of the idea of transparency which, the author points out, is as slippery an issue in film as is has been for semantics. This should of course not be too surprising, given that the notion of identity underlying such observations is an idealisation which has no bearing on what goes on in the actual practice of social semiotic. Branigan could perhaps have followed this line of thought a little longer. In language too, top-down processing of non-transparent materials is ubiquitous.

Branigan now returns the reader to the question of how narration links up with human cognition. The crucial term here is 'epistemological boundary'. When we speak of characters, implied author, narrators, or cameras we choose labels in order to mark 'epistemological boundaries, or disparities, within an ensemble of knowledge' (85). This observation is refined by the stipulation of eight hierarchically organised levels of film narration each of which constitutes an epistemological domain for the description of data. They extend from the 'internal dynamics of a character to a representation of the historical conditions governing the manufacture of the artifact itself': text (historical author, historical audience); fiction (extra-fictional narrator, extra-fictional narratee); story world (non-diegetic narrator, non-diegetic narratee); event/scene (diegetic narrator, diegetic narratee); action (character, character); speech (external focalisation, observer); perception (internal focalisation [surface], identification); thought (internal focalisation [depth], identification) (87).

From this perspective Branigan argues that communications model explanations of film narrative are not able to capture fully what goes on. 'If a text is sometimes a "communication", it is almost certainly operating in other ways as well' (110). He gives the reasons for his critique of the communications model in the section "Texts under a Description" by drawing our attention to 'the transformation of one epistemological context into another, a movement from (embedded) level to (embedding) level' which he considers to be at the heart of film narrative (111). Branigan supports this claim with reference to his eight levels from historical author to internal focalisation in Hitchcock's The Wrong Man (1956) and concludes that narration takes place whenever the spectator transforms screen data from one epistemological level to another: 'the central activity of narration is the redescription of data under epistemological constraints' (112).

The advantage of describing film viewing in terms of levels and boundaries is that it allows us to speak of 'the relationships of embedding and hierarchy, which, in turn, provide mechanisms for a fundamentally different kind of contribution to human cognition than laws of addition and subtraction, or the rules of branching networks'. Further, we can think along the lines of constraint as opposed to determination. Thus any higher level guides our viewing and so constrains rather than determines the way in which we interpret screen materials. Furthermore, a hierarchy of levels may be seen as 'a set of probabilities that predicts the likelihood of hypotheses'. Since every lower level becomes a function of our hypotheses made at a higher level, 'each step down the hierarchy increases the number of assumptions that must be made and narrows the range of knowledge available to the spectator' (113).

The assumption made here is of course the rule of 'rationality' in the sense of individuals not wishing to live with inconsistencies in their interpretations of the world. While this is a sound assumption on the whole and finds support in the kind of philosophical position advocated for instance by Donald Davidson, a little more room should perhaps be given to playful and whimsical interpretive choices. But I leave this critique to the more deconstructionist responses the book will no doubt receive.

From the vantage point of the schema of epistemological boundaries, Branigan is now able to offer the following definitions. Point of view in narrative is described as 'the relationship between any pair of levels of narration'. Omniscience is seen as referring 'to any one of the higher levels of narration [and specifically] to the reader's toleration of a boundary or limit to what finally can be known in the text' (115). Likewise, the earlier distinctions between declarative and procedural knowledge, as the two modes in which story data appear to us, can now be defined from a new angle. 'Declarative knowledge is generated by a narrative schema which yields a series of episodes collected as a focused causal chain', while 'procedural knowledge is generated by a narration (or narrational schema) which yields a series of levels, or epistemological boundaries' (116). A further difference between declarative and procedural knowledge is that only the former can be 'right' or 'wrong' about the kind of story world as realised by the spectator. The procedural techniques of film should not be understood as producing meaning but rather a set of instructions for the spectator as to how to construe a world. This comprehensive perspective is then tested against five broad theories of narrative, emphasising in turn: event-sequences, style, speech-act and discourse, reader reception, and psychoanalytical theses about narrative. Perhaps the strongest argument for Branigan's thesis in these comparisons is that it caters more satisfactorily than the others for the dual applicability of its findings, to film narrative specifically and broadly to cognition or the social semiotic construction of our world.

A chapter on "Subjectivity" redraws the traditional distinction between material and structure, such that 'different materials may achieve the same structure (synonymy) [and] different structures may be achieved by the same material (ambiguity); or, when spread out in time, the same structure may reoccur in different materials (repetition), or the reoccurrence of the same material may nonetheless elicit different structures (multiplicity)' (141). Again, proceeding from a general principle to detailed analysis, the author addresses a number of readings of Lady in the Lake before making the important point that one cannot simply speak of the failure of the subjectivity of the point-of-view shot and hence the failure of the entire movie without specifying the theoretical angle from which such a critique is made.

In the section on "How Many Cameras Are in a Film?" Branigan writes that one can either regard the camera as unimportant or broaden the notion to include a number of features. From such a broader (top-down) perspective, Branigan notes 'at least three variables': character (agent), time, and place. The camera could then be described according to 'how pictures and sounds condense at a given moment into a single hypothesis that stands for an event occurring in a time and place for a given person, such as an author, narrator, character, or spectator' (159). Branigan makes the neat point that the camera functions much like a shifter in language. Its meaning depends on its use such that 'it is both an effect of the text and a trace of the process of our reading' (159). Theory employs terms like 'camera' and 'screen' to mark the degree of separation between film viewer and screen material. Perhaps, the author says, we should conceive of there being different cameras for each epistemological domain and acknowledge that 'the "camera" cannot be separated from the rhetoric of its shots' (159).

"Objectivity and Uncertainty" reviews the conditions under which we construe 'images and sounds when we imagine that we experience them independently of a character' (161). Here the author shifts his attention from the techniques which guide spectators in their inferences of subjectivity to the notion of "objective" or, more precisely, intersubjective diegesis. This allows Branigan to characterize classically conceived objectivity as dependent on 'a firmly delineated hierarchy of levels of narration' in contrast with its alternatives in which we are dealing with 'multiple, inconsistent, or indeterminate' constructions (161).

One of the strengths of the book comes to the fore when the author focuses on the technical minutiae of a step-by-step description of the production of objective narration. There is the instance in which Branigan makes us note the turning point in a spectator's construal of diegesis from 'what is explicit to what is implicit' about a character's perception (162). The movement from what we take to be explicit as against implicit knowledge is shown to be paralleled sometimes by a movement from definite to indefinite states. This leads to a sense of uncertainty about the diegetic world. As a result, typical 'gradations in our beliefs' are generated (163). Branigan displays the fourcornered logic of this shift in knowledge as follows:

The spectator sees an observer seeing an object;

The spectator infers that an observer sees an object;

The spectator sees that someone sees an object;

The spectator infers that someone sees an object. (163)

The combination of the two trajectories of the construction from explicit to implicit and from definite to indefinite knowledge stands in a specific relation to the question of reference. The first kind of construction determines the degree of inference needed 'in determining a reference to a type of thing in the diegetic world', while the second type of interpretation is a measure of referential specificity (163). Branigan sums up the relation between the two movements in this way: 'The more "implicit" a representation becomes, the more it relies on inferential chains and a prior knowledge of the spectator to make it emerge. The more "indefinite" a representation becomes, the more uncertain it is which one of the relevant class of things is being referred to' (163).

What excites Branigan most in this respect is the indeterminate case and its 'implicit and indefinite reference' (163). One of the consequences of indefinite reference is shown to be 'invisible observation', which he subdivides into six kinds:

1. invisibility: the observer as eavesdropper;

2. ubiquity: the ubiquitous, instantaneous observer;

3. alertness: the always perfectly placed observer;

4. neutrality: the standard-angle observer;

5. impersonality: the observer with indefinite traits; and

6. passivity: the observer restricted by immediate, diegetic temporal and spatial boundaries (171f.).

For Branigan, such curtailment of information is at the hub of narrative. As he says a little later, 'the very question of restriction of knowledge is at the very centre of narration' (172). He also appears to suggest here that the degree to which the spectator is able to overcome some of these restrictions is a measure of the breadth of the diegetic world he or she is able to construe.

Commenting on flashbacks, Branigan makes the important point of aligning two different sign systems in principle: the visual and the verbal. To represent action sequences visually in no way renders them 'more explicit, detailed, consistent, objective, certain, present, or precise than representing them in words' (176). Against the myth of the singularity of visual meaning, Branigan posits the democratic idea that 'the greater the latitude allowed for the spectator, the greater the number of spectators who may participate' (177). Yet the book is not blind to the dangers inherent in implicit and indefinite narration. As an illustration Branigan offers a detailed analysis of Max Ophuls's Letter from an Unknown Woman (1948) in order to highlight filmic indeterminacy and its resulting difficulties for film theorists (177-191).

The last chapter addresses definitions of fiction. Starting from the assumption that 'narrative and fiction are quite different things even if they often appear together in public', Branigan goes on to emphasise that narrative is a particular way of organising information to form 'a scene of action', a 'temporal progression' and the dramatisation of 'an observer of events' (192). This he contrasts with non-narrative ways of assembling information, such as dictionaries, indexes, catalogues, legal contracts, and others. By contrast, the description of fiction is said to involve the quite different kind of question of how a given assembly of data relates to our ordinary understanding of the world. Here issues surface such as 'truth, appropriateness, plausibility, rightness, or realism' (192).

Attractively, Branigan rejects truth as a guarantor for non-fictional knowledge, pointing out that both fiction and non-fiction make use of literal as well as figurative representations. In order to bypass these difficulties, the chapter argues for a differential account of the way we assign reference (193). So it does not come as a surprise when the author, towards the end of his book, defends fictions as playing an important role in our conception of the world, even though the link between the two is by no means straightforward. If Branigan believes that fiction guides us to 'connect text and world in a special way' (193), the reader is entitled to hear the answer to the traditional question of 'how does fictional reference refer?'(194). Broadly adopting the stance argued by Hartley Slater in "Fictions" (1987), Branigan suggests that 'a fiction does not determine exactly what object or objects it represents, and it is this openness which distinguishes fictional reference from other sorts of reference'. Hence to read fictionally is to enter 'a precarious, intermediate zone between sets of possible references (open functions) and a specific reference (a given "solution"or "referent")' (194).

If we regard fiction as a cognitive activity, as I agree we should, then its description would have to address the way in which we understand the world by holding open 'sets of variables while searching for a reasonable fit between language and lived experience, between sets of symbols and acts of the body' (194). In this sense, fiction is oriented towards 'presupposition, prior belief, and knowledge' rather than 'derivation and confirmation'. The role which indeterminateness of screen information then has in the text is, one could say perhaps, a kind of diff­rance which 'acts to delay and expand the kind of searching and restructuring of prior knowledge undertaken by a perceiver' (195). This both temporalises and destabilises fictional reference; it renders it only 'partially determined' and 'initially neither true nor false'. Its character is primarily pragmatic in that 'its usefulness must be found and demonstrated' (196) rather than stipulated as a logical ideality.

Persuasive as this argument is, it shows how closely related fiction has now become to non-fictional accounts of the world. And yet this is an inevitable consequence of the constructivist position we discover throughout the book. 'Fiction involves a reference to a world whose shape is determined by a community' (197) and 'a theory of fiction should be built upon a careful examination of our ordinary ways of thinking and our everyday abilities' (198). From here it is a close step to saying that all our constructions are fictions in the sense that they are all community sanctioned readings, more or less direct, of the stuff humans typically confront. One could call this a textualist position in the broad sense. What remains to be shown is how 'realist' or 'relativist' this position is; that is, how Branigan would theorise the constraints which make some of our constructions failures and others compatible with what they describe. Jumping out of tenth floor windows and still surviving is, as a rule, a fiction which cannot be consistently demonstrated and so remains an unsuccessful description of the world except on screen, while other descriptions, though perhaps fantastic, may not contradict such boundaries. Which all shows that there is neither a semantics nor a film theory without metaphysical implications (see Ruthrof).

If an art form is characterised by its ability to exploit best the available technical means of an epoch, then film qualifies as the most typical art of the twentieth century. Not only can it subsume a gamut of older genres from sculpture to music, it is also able to adopt the most innovative digital machinery and its products. In this sense, film stands between self-evident fictions and the no-longer mere fictions of virtual reality.

By drawing our attention to the simultaneity of the image as a fiction and non-fiction, Branigan adds an important perspective to the theory of meaning in the broad sense. When we look at a photograph of a person in a fictional manner, the image acquires 'a new dimension as a label within a complex, developing predicate' (198). Applied to film, this illustrates a well-researched difference: 'when a film is experienced fictionally, reference is not to the profilmic event in which a set is decorated and an actor given direction, but rather to a postfilmic event in which patterns are discovered through active perceiving that affects the overall structure of our knowledge' (200). It follows that the materiality and historicity of the text are not determinants of reference, no matter how relevant they may appear.

At this point in the argument, Branigan nicely complicates the picture. It is often not easy to make decisions on fictionality and non-fictionality in film. Indeed some films, such as Chris Marker's Sans Soleil (1982), demonstrate that 'the referent of narrative non-fiction may not be certain either'. If film and its modes of reference operate finally like other interpretations within a social setting and are 'governed by consensus' (201), then any neat polarity must give way to a broad spectrum of interpretive negotiations between referential certainty, semi-determinacy and uncertainty, a suspicion well supported by the effects of Sans Soleil.

The concluding section analyses the postmodern features of Sans Soleil, pointing to the 'unusual evocations of time' which disrupt the causality of a beginning, middle and end. What the viewer must work with instead is a 'field of possibilities: a multiplicity of partially realised narratives and non-narratives competing equally' (215). Branigan chose Sans Soleil for his concluding analysis because it renders problematic the traditional relations between the two pairs of fiction vs. non-fiction and narrative vs. non-narrative. In doing so, its postmodern aesthetics 'halts the careful ordering of a cause and effect chain', acting like 'a "hyperindex" of stories' which invites several entries and does not 'terminate with one or several particulars'. Instead, its 'associational logic' activates an indefinite range of interpretive routes by which we can link our reading hypotheses with the data provided by the screen (216).

Yet postmodern aesthetics need not be restricted to experimental film. Tying filmic cognition to general perception once more, Branigan concludes by saying that 'narrative comprehension is a way of recognising the "causal efficacy"' of things and 'in understanding a story, we are imagining and tracing out several, or many, of the possibilities for the being of an object' (217). Thus, the complexities of a film like Sans Soleil prefigure a sophisticated description of the way we constitute our world.

Works Cited

Branigan, Edward. Point of View in the Cinema. Amsterdam, New York: Mouton, 1984.

Ruthrof, Horst. Pandora and Occam: On the Limits of Language and Literature. Bloomington: Indiana University Press, 1992.

Schank, Roger C. and Robert P. Ableson. Scripts, Plans, Goals and Understanding. Hillsdale, NJ: L. Erlbaum Associates, 1977.

Slater, Hartley. "Fictions", British Journal of Aesthetics 27 (1987): 145-55. Australian Journal of Communication. 16 (1989). 1-12.

New: 6 December, 1995 | Now: 24 March, 2015