Narrative Theory: Understanding Characters, Plot, and Film Analysis, Study notes of Poetics

Download Narrative Theory: Understanding Characters, Plot, and Film Analysis and more Study notes Poetics in PDF only on Docsity! 3. Three Dimensions of Film Narrative A man sitting in a bar suddenly shouted, “All lawyers are assholes!” The customer next to him jumped off his stool. “Those are fighting words!” “Oh, so you’re a lawyer?” “No, I’m an asshole.” The study of narrative has a long history, but as a self-conscious body of inquiry, this enterprise is principally a creature of the 20th century. It was then that it came to be called narratology, an ugly term but one that apparently we can’t easily do without. Whatever we call it, the study of narrative is very important. Storytelling is a perva- sive phenomenon. It seems that no culture or society is without its myths, folktales, and sacred legends. Narrative saturates everyday life too. Our conversations, our work, and our pastimes are steeped in stories. Go to the doctor and try to tell your symptoms without reciting a little tale about how they emerged. The same thing happens when you go to court or take your car to a mechanic or write a blog. Perhaps storytelling is part of human maturation, since it emerges quite early in human development. Children only two years old can grasp certain features of narrative, and there’s evidence from “crib monologues” that the narrative ordering process is emerging even earlier. We share stories with each other, assuring others that we have experiences congruent with theirs. Sometimes we tell a joke, like my curtain-raiser, to create a bond—though after some experience, I’d advise you that this one won’t create deep ties in certain situations. We can apparently turn anything into a story. String figures akin to Cat’s Cradle may tell tales. Figure 3.1, from the Torres Straits, represents one stage in a fight between head- hunters: The two warriors are squaring off. The player then tugs on the left-hand loops, and the headhunters clash. The outcome can’t be predicted. Both fighters may die and fall Figure 3.1—A Trobriand Island string figure: The headhunters face one another. Figure 3.2—When one fighter wins, he departs with the enemy’s head. 2 Poetics of Cinema apart, or one kills the other and “travels home,” bearing the enemy’s head (Figure 3.2).1 In Australian Aboriginal sand paintings, what might seem to outsiders to be abstract squig- gles and whorls represent mythical events or incidents from daily life.2 Narrative appears to be a contingent universal of human experience. It cuts across dis- tinctions of art and science, fiction and nonfiction, literature and the other arts. So it’s not surprising that studying narratives brings together students of not only literary studies, drama, and film, but also anthropology, psychology, even law and sociology and political science. Narratology is a paradigm case of interdisciplinary inquiry. Thing and activity, in the head and in the world   Widespread as narrative is, though, it retains a distinct identity. Considered as a thing, a certain sort of representation, a story seems intuitively different from a syllogism, a database, and an fMRI scan. My opening joke isn’t exactly like other forms of humor, such as a bumper sticker (“Today is the day for decisive action! Or is it?”). How should we try to capture narrative’s uniqueness? Perhaps narrative is like grammar in a natural language, or perhaps it’s a sign system, like traffic signals, as semiotic theories suggest. Narrative is more than a kind of thing; it seems to involve distinct activities as well. One activity we call storytelling, and the other… well, what do we call it? Story consump- tion? Story receiving? Story pickup? In any event, we have capacities that enable us to grasp and present stories. This talent too opens up many questions. From one angle, our stories come from our psyches, involving mental contents and processes. The very act of remembering something is coming to be seen as less a retrieval of fixed data than an on- going construction according to principles of narrative logic.3 Yet narrative is as well preeminently social, a way of organizing experience so that it can be shared. Narrative conventions invoke lots of particular knowledge, and my opening joke wouldn’t be understood in a culture that lacked bars, lawyers, and lawyer jokes. Nar- ratives activate social skills, and although some people become expert storytellers (some can tell ’em, some can’t), nearly all of us recognize well-formed stories when we encounter them. Our narrative competence relies on social intelligence. Distinct as narrative seems, it’s also polymorphous. It blurs and blends into a lot of other forms and activities. In a novel, it’s often hard to carve out the descriptive passages cleanly from the plot, because accounts of people crowding a train station or skiing easily pass into little suites of action. The rhetorical tradition, theorizing about what persuades audiences, recognizes that stories can carry weight in an argument; the summary of the facts of a law case were known to the ancient Greeks as the narratio. I could use my joke to illustrate an argument about why lawyers get no respect or a tirade about what conserva- tives call the coarsening of our culture. Peter Greenaway’s film The Falls (1980) provides a purely categorical macrostructure—a directory of people whose last names begin with the letters Fall—but soon we find that every Fall- has a life course full of incident. In their turn, stories are omnivorous, consuming other forms. Japanese literature in- cludes the genre of travel journal, which is in prose but often splices in descriptive verse passages. Frank Capra’s film The Battle of Russia (1942) spends a fair amount of time cata- loguing all the types of people living in the USSR. Mikhail Bakhtin argued that the novel Three Dimensions of Film Narrative 5 This action-centered notion of minimal narrative can be traced back to Aristotle’s Poetics. Against it we can set a conception that’s often identified with Romantic and post-Romantic literary criticism. Someone might argue that all this talk of “events,” “states of affairs,” and “causality” turns narrative into a bloodless abstraction. When we think of narrative, we think first of characters. For Aristotle, a narrative is a whole, and agents take up a place in a larger rhythm of event-driven activity. But we can treat the agents and their capacities as the basis of narrative, with events seen as products of those qualities. Historically, the agent-centered perspective was influenced by medieval and Renais- sance theories in which character was conceived as a mix of vital humors or dispositions. In a reaction to neoclassical norms of proper writing, theorists pointed to Shakespeare. His plays seemed to be weak on abstract plot geometry but unsurpassed in their portrayal of human behavior. Schlegel wrote that Shakespeare created unique individuals who act spontaneously but plausibly. Shakespeare endows “the creatures of his imagination with such self-existent energy that they afterwards act in each conjuncture according to general laws of nature.” Shakespeare doesn’t laboriously tot up all of a character’s motives, for that could suggest that each one’s identity is simply the sum of larger forces. “After all, a man acts so because he is so.”4 It’s not that this view disregards plot as such. Whereas Aristotle sees human agency as a part of a total action, Schlegel believes that the abstract structure of events flows from the display of human personality in the process of change. Maybe most people would agree. They think of narratives, or at least the most valuable ones, as portraits of human minds and hearts. True, the page-turner, the book we read with unquenchable interest, might seem to cater to our action-based appetites. Yet even then, many will say, we read on because we’re held by characters who arouse our passions. Still, it seems to me that the drastic split between plot and character, derived from Romantic theory, has led to a kind of caste system, whereby character-driven stories are felt to be inherently superior to ones that showcase suspense, excitement, and unexpected twists. For one thing, supposedly character-driven narratives often turn out, on exam- ination, to have a rich action-based architecture too. Shakespeare’s plays are marvels of construction, and the indie films supposedly putting character on display often obey many conventional plot mechanics. Moreover, narrative offers many pleasures, from psycholog- ical probing and nuanced social observation to imaginary adventure, thunderous surpris- es, and Grand Guignol shocks. Flaubert and Dumas, Trollope and Conan Doyle tap into different sources of narrative pleasure, and it’s not clear that a Merchant-Ivory adaptation is more satisfying or accomplished than Die Hard. In any case, what follows tries to outline what I take to be a promising poetics of filmic narrative. It suggests that we can look for constructive principles and normalized practices along three dimensions. None of those dimensions is rigidly biased in favor of action-based or agent-based models of a story, but in my application of them, probably my predilections will shine forth. 6 Poetics of Cinema The three dimensions Taken singly, the three dimensions I’ll be considering seem to me uncontroversial. All have been considered before in the vast literature on narratology. But in spreading them out side by side, I think we gain a sense of the rich array lying open to analysis from the standpoint of poetics. One dimension involves what I’ll call the story world: its agents, circumstances, and surroundings. In my opening joke, that world consists of a bar (and all of the pre- sumed furnishings of a prototypical bar). A second dimension is that of plot structure, the arrangement of the parts of the narrative as we have it. My joke is structured as a series of actions and reactions, statements and replies. It has a neat symmetry (two lines from each of the two participants), and it builds to a payoff, the punchline. The third dimension I propose is that of narration, the moment-by-moment flow of information about the story world. The narration of the joke is laconic, never describing the bar or the men or even how they’re arrayed in the bar (except that one is apparently on a stool). We are outside the men’s minds, Hemingway fashion, whereas other jokes are resolutely subjective. All three dimensions contribute to the point of the joke. I’ll be elaborating on these distinctions in the pages ahead. For now, here’s an analogy, though it shouldn’t be pressed too far. The story world is similar to the semantic dimen- sion of language, plot structure is comparable to grammatical or syntactic structure, and narration is comparable to verbal style, as governed by pragmatic context. Protagonists and their problems     Before I consider each dimension separately, let me provide an example of how making these distinctions can help us with problems in poetics. We commonly believe that a narrative film is likely to have a protagonist. But how do we determine who or what a protagonist is? I suggest that several dimensions of judgment are involved, most ingredient to all narratives in any medium but one specific to cinema. In the story world that the narrative presents, the protagonist is the agent whom the story is about. There are many heuristic cues that help us pick out a hero or heroine. The protagonist may be the character with the greatest power, as King David is in certain chap- ters of the Old Testament. The protagonist may also be the character with whom we tend to sympathize most keenly, as in the biblical story of Daniel. The protagonist may be the character with whose value system we are assumed to agree. Or the protagonist may be the one who is most affected or changed by events, as in James’ Portrait of a Lady. No one of these cues is decisive on its own. After watching The Godfather, many view- ers would say that Michael Corleone’s wife, Kay, arouses more sympathy than either Don Vito or Michael, and the dons’ value system is unlikely to be wholly endorsed by us. Mi- chael especially seems a cold protagonist, like Tamburlaine. More important, though, Vito and Michael are the most elevated characters, with the power to decide life and death, and Michael is evidently the character who changes the most in the course of the action. These criteria seem to weigh heavily in this story world. Don Vito and Michael are spotlighted by narrative structure as well. The major por- tions of the films pivot around them, from Don Vito’s attempted assassination to Michael’s escape to Sicily. Were we to divide the film into large-scale parts, or long chapters, the Three Dimensions of Film Narrative 7 breaks would reflect major changes in their fortunes. Moreover, the actions of these two men, both proactive and reactive, dictate the overall shape of the plot. Don Vito’s decision not to join the drug-selling business set up by Sollozzo triggers the gang war that follows, and Michael’s decision to assume his father’s place in the family business guides events along the course they take in the second half of the film. Structurally, the character whose actions give the drama its distinctive arc is likely to be the protagonist, as the etymology of the term suggests. Agon refers to a contest or competition, and so the protagonist is “the first combatant,” whereas the antagonist is the warrior who opposes the protagonist. But wait, somebody might say. In The Godfather the plot developments are really trig- gered by Sollozzo’s decision to start a drug business, and Don Vito merely responds to that initiative. Why isn’t Sollozzo the protagonist? Similarly, later plot developments are responses to Sollozzo’s decision to wipe out Don Vito. Our intuition, of course, is that Sollozzo is not a protagonist but an antagonist, but how do we justify that impression? Here we can usefully invoke our third dimension of narrative construction, that of narration. The Godfather is designed to concentrate our attention on the doings of the Corleones, not of the Sollozzo gang. Significantly, we don’t spend much time with Sollozzo when a Corleone isn’t present. One quick measure of how narration can suggest who is a protagonist involves registering how long a character is onstage. Scenes including either Don Vito or Michael Corleone consume nearly 75% of the duration of The Godfather, and Michael appears in nearly half of it.5 No other characters receive nearly this much screen time. It seems likely that the more pages or minutes devoted to a character, the more likely we are to take him or her as a protagonist. Just as important as sheer quantity of coverage is the way narrational restriction at- taches us to the family. We know, by and large, what Don Vito, Sonny, Tom Hagen, and Michael know, and in Michael’s case we often know it in depth. Many scenes access his moment-by-moment psychological reactions, as when he sets up the fake hospital protec- tion for his wounded father or when he assassinates McCluskey and Sollozzo. True, his final revenge scheme isn’t spelled out in advance. But our earlier access to his mind makes our realization that he’s coldly ordered a massacre all the more shocking. To put it loosely, the action of The Godfather is presented from the point of view of the Corleones, and most often that of Michael. In the spirit of Rosencrantz and Guildenstern Are Dead, we could imagine recasting the film’s narration to create a story told from the side of Sollozzo and his allies, in which the Corleones are distant figures. But that’s not the movie we have. Considering all three dimensions, I don’t think we can come up with a single or sim- ple definition of how we know a protagonist. In grasping any narrative, we weigh the di- mensions comparatively. We tacitly assay a character’s prominence in the story world, her structural role, and her narrational salience. Often these factors will dovetail neatly. In The Untouchables (1987), Elliott Ness is clearly the protagonist. He is powerful and sympathetic in the story world, and his char- acter undergoes the greatest change, moving from ineffectual rectitude to a hardheaded willingness to fight fire with fire. His value system gives the film its moral compass. Struc- 10 Poetics of Cinema built upon narrational concepts.6 This is a natural and salutary way scholarly inquiry pro- ceeds. Culture and convergence   Someone might go on to say that my belief in convergences of comprehension is naïve. Perhaps women don’t comprehend stories as men do, and people in Japan don’t understand their stories as Europeans do. Note that this objection does presume some convergence, if not between social groups then within them. Why believe that only certain groups share understanding and others can’t share it? Why can’t compre- hension strategies crisscross groups in that hybrid fashion beloved of postmodernists? It’s very hard to avoid some sense of convergence when talking about the understanding of any audience, no matter how culturally localized. Moreover, because comprehension involves such features as tracking psychological states, causality, time shifts, and the like, the onus is on the skeptic to show that women or cultural insiders possess different senses of cause and effect or time relations than other perceivers do. One of the most commonly cited examples is that in watching a Western, Native American audiences might cheer on the Navajos attacking the settlers. Even this apparently apocryphal anecdote, however, doesn’t damage my case. I assume that the au- dience understood the story—that the settlers were crossing Indian land, that the Indians wanted to wipe out the settlers—and that the viewers took sides in a way not anticipated by the film’s makers. To say that there’s convergence in understanding is not to say that all spectators act upon their understanding in the same ways. By focusing on comprehension from a mentalistic perspective, I hope to adhere to other conditions I set out in Chapter One. In accord with my layout of spectatorial ac- tivities, I assume that there’s a fair amount of convergence in viewers’ understanding of the narrative. There may be some disagreement among spectators’ grasp of character mo- tivations or consequences, and we should expect that, given the variety of schemas that viewers bring to films. But divergences in comprehension aren’t anything like as wide and varied as we’d find in more abstract interpretations, for reasons I’ve already suggested. The shape of things   Again, in accord with the sort of poetics I’m proposing, this study of narrative treats films holistically. My conceptions of narration, plot structure, and the story world try to take into account the overall form of a film. The assumption here is that regularities we find across the whole artifact allow us to make inferences about the purposes of its makers and the activities coaxed from its viewers. Take the openings we find in ordinary movies. Very often we get an expository title giving time and place, along with one or more long shots of an area of action. Cuts or camera movements may carry us into a scene, with characters moving toward us, or tracking shots that follow a character from behind in exploratory fashion (Figures 3.3–3.4). On the soundtrack, music sets a mood, and dialogue rises to audibility. Clearly all these tactics are blended to engage the spectator’s interest, parceling out information needed to understand the action. The cut- ins or forward camera movements also suggest that we are being drawn gradually into the story world. Strikingly, the ending of an ordinary movie often reverses these devices. The camera pulls back, characters turn away and we don’t follow them, doors and gates may shut, the Three Dimensions of Film Narrative 11 music rises again, and titles may appear (Figure 3.5). The opening literally opens up the movie and lets us in; the closing shuts it down and expels us. The best explanation for these regularities onscreen is that they’re manifesting principles that filmmakers share, perhaps tacitly, and they function to shape our experience of the story. The symmetries between openings and closings suggest that narration is a system that’s put into motion across the whole film. All the factors we normally associate with narration—play with the order of events, shifts in point of view, and voice-over commen- tary—fall under the rubric of narration. They’re not just one-off tactics; they play roles in larger patterns running across the entire movie. So once we’ve identified a passage of om- niscient narration or optical point of view, we should go on to look at how that functions Figure 3.3—The opening of The Silence of the Lambs (1991): Agent Starling comes out of the story world to meet us. Figure 3.4—The Silence of the Lambs: She turns and runs into the forest; the camera follows and carries us into the story action. Figure 3.5—At the end of The Silence of the Lambs, Hannibal Lecter turns from us and follows his prey into the story world, but we stay behind. This is a conventional mark of closure. 12 Poetics of Cinema in the broader patterning of the narration. Why shift to optical point of view here? How does it shape the experiential logic of the overall film? Narration goes all the way down     Narration is more than an armory of devices; it be- comes our access, moment by moment, to the unfolding story. A narrative is like a build- ing, which we can’t grasp all at once but must experience in time. We move from static spaces to dynamic ones, enclosed spaces to open ones, peripheral areas to central ones— often by circuitous routes. That journey has been arranged, and sometimes wholly deter- mined, by architectural design. Narration in any medium can usefully be thought of as governing our trajectory through the narrative. This analogy helps us see that we don’t gain by treating narration as something like an envelope enclosing the story action. As a process, narration burrows all the way down into the material, shaping it for our uptake. It governs how we grasp overall structural dynam- ics and the immediate scene before us. It controls how we build an inferential elaboration of any event. Consider this sentence: A boy saw a woman kissing a man. By narrating the event this way, I’ve shaped your inferences, identifying certain features of the action and eliding others. (We don’t know the relationships among the three charac- ters.) Now try this rendition of the same action: Tim saw Dorothy kissing Wally. I’ve not only named the agents but also encouraged you to posit a relation among them; Tim, Dorothy, and Wally are unlikely to be strangers to one another. By providing their first names, I’ve also encouraged you to assume a certain familiarity with them. In Rex Stout’s detective stories, we know Archie Goodwin by his first name and as I (because he’s the narrator), but we know Nero Wolfe only by his last name. Who would dare call him Nero? By such simple means does literary narration conjure up intimacy or distance. This makes it rather off-putting when Dashiell Hammett calls his protagonist “Ned Beaumont” throughout The Glass Key. We don’t really know our relation to the enigmatic figure. But like most narrative devices, this piggybacks on our normal social interchange, with first names as marks of intimacy. Let’s return to our example, with another change: Tim saw Mommy kissing Daddy. Now Dorothy and Wally are presented in terms that coax us to infer a specific relation to Tim. The sentence doesn’t say he’s their son, and it’s possible he’s not (as in the case where a daughter tells us about her boyfriend, “I was so embarrassed that Tim saw Mommy kissing Daddy”). Still, the narration has opened up a new range of inferences. It’s only a short step to: I saw Mommy kissing Santa Claus. Employing a traditional cue in literary narration, the I that replaces the Tim anchors us in Tim’s consciousness. The Mommy gives us better grounds to infer a kinship with the Three Dimensions of Film Narrative 15 Yet in order to mislead us, it has to rely on our making certain inferences about cau- sality, ordering in time, and the like. A common strategy is the unmarked ellipsis, whereby we’re encouraged to ignore a time gap that the narration doesn’t flag—only to later come to understand that something important took place in that gap. This ploy is at work in Fritz Lang’s You Only Live Once (1937) and The Blue Gardenia (1953) and Otto Preminger’s Fallen Angel (1945). Alternatively, by restricting our knowledge to what only one character knows, the narration can mislead us about events, only to surprise us later when we get a fuller account of what happened. This is common in detective stories and film noir tales. Cinematic narration, considered as the interaction of style and syuzhet patterning, has a great many resources. Here are just a few. A more complete catalogue can be found in my book Narration in the Fiction Film. • The syuzhet can juggle the order of fabula events, providing a flashback or flashforward. • It can manipulate fabula duration, stretching out or compressing the time that story events consume. • It can present simultaneous fabula events successively (via crosscutting), and suc- cessive events simultaneously (through split screen or other devices). • Cinematic narration can be more or less knowledgeable, claiming greater or lesser access to information, and more or less self-conscious, flaunting the act of present- ing this story to various degrees. • The syuzhet can provide an omniscient range of knowledge, as when a film inter- cuts characters’ trajectories, or it can restrict the flow of information to what one character knows, as some detective films do. Stylistic devices like optical point-of- view shots, voice-over commentary, and sound perspective can funnel information through a character’s literal standpoint. A common strategy of cinematic narration is to attach us to one character for a scene or two, then shift to another character’s range of knowledge, creating a sort of shifting restrictiveness. • Cinematic narration can also be more or less objective, remaining resolutely on the “outside” or pulling us into characters’ minds via memories, dreams, or imaginings. Cinematic narration overlaps with literary narration, but the two aren’t perfectly con- gruent. For instance, filmic “point of view” is rarely as stringent and sustained as the lit- erary variety. A first-person narrator in a novel restricts us to a single consciousness, but a film’s voice-over narrator can initiate the revelation of events that she didn’t witness, or even know about, as in Ten North Frederick (1958). A long-standing convention holds that literary storytelling mimics storytelling in life, whereby every tale has a teller and receiver (reader, listener). This communication schema works well for many novels, though per- haps not all. Who “tells” a montage-based novel like Dos Passos’ USA trilogy? In any event, a film’s syuzhet and style aren’t bound by the constraints of verbal com- munication. Cinematic narration, being an audiovisual display rather than a written text, appropriates bits and pieces of the communication model opportunistically. So we can have voice-over commentary from the protagonist without there being any indication that he or she is speaking to anyone in the fictional world. The commentary may be taken as stream-of-consciousness musings or as simply another conduit for story information, without any need for the real-world baggage of speaker–listener relationships. I expand 16 Poetics of Cinema on this idea in my discussion of the problem of narrators in cinema, which serves as an appendix to this essay. “I don’t like voiceover as exposition,” Steve Martin remarks of his film Shopgirl, “be- cause I don’t think anyone is listening.”8 No one, except the only one who matters: the viewer. At the start of Jerry Maguire, the hero’s voice introduces us to his lifestyle and his personal crisis, and then his voice vanishes, never to reappear. To whom was he speaking? The question is as irrelevant as the physics of light sabers. The film doesn’t need to an- chor his discourse in a full-fledged communication situation because it recruits part of the communication template to get information out to us. Literary logic can go hang; all that the narration cares about is cueing us to make the right inferences. The viewer’s share: Curiosity, suspense, surprise     Structuralist thinkers have brought many narrational processes to light, creating useful taxonomies of temporal manipulation and point of view. From the standpoint I’m indicating here, I suggest that we need to put taxonomies into motion, so to speak, by considering the characteristic sorts of activities that distinct categories tend to encourage. For instance, many narratologists have rightly celebrated Gérard Genette’s layout of temporal possibilities, but few have recognized that they elicit rather different activities when situated in certain contexts or different media. As an option, straight chronology is on a par with juggled time sequence, but psychologically the former operates as a default. Chronology is the norm in narratives generally. Chronology is our presupposition in fol- lowing events in the world, let alone events in narratives. Another abstract option is this: If two fabula events are occurring simultaneously, you can present them successively or simultaneously in the syuzhet. But literature is in- eluctably successive (words follow one another), and on the page you can’t strictly show two things happening at the same time. In reading we have to infer simultaneity from the bits of action presented moment by moment. Film, however, presents simultaneous action very easily, both within the shot (one character in the foreground, say, and another in the distance) and in split-screen imagery. Meir Sternberg has been the most eloquent and persistent advocate for treating tax- onomic categories functionally. He has argued that what matters is that all the strategies charted by the taxonomists must be gauged in relation to their capacities to create distinc- tive effects on the perceiver. For example, a flashback isn’t just an abstract rearrangement of story incidents. Its function is to trigger interest in finding out what led up to what we see. Sternberg suggests that by considering three aspects of our narrative appetites, we can offer good functional explanations for particular devices. Curiosity stems from past events: What led up to what we’re seeing now? Suspense points us forward: What will happen next? Surprise foils our expectations and demands that we find alternative expla- nations for what has happened. Syuzhet arrangements of events arouse and fulfill these cognition-based emotions. Sternberg’s account of the experiential logic of narration fits well with my concern for a poetics of effect.9 In this sequence of words, which one doesn’t belong? Skyscraper    Temple    Cathedral    Prayer Three Dimensions of Film Narrative 17 Most people would say Prayer, because the first three terms refer to types of buildings. But now present the words in this sequence: Prayer    Cathedral    Temple    Skyscraper People usually say that Skyscraper is the outlier, because the first three items refer to religion. This is what psychologists call the primacy effect. The order of events governs how we understand them, and the first item has greater saliency. Likewise, a film’s opening will set a benchmark against which we measure what happens later. The characters we first encounter, the point at which we enter the story action, and other elements will shape our inferences. Sternberg speaks of the “rise and fall of first impressions,” pointing out how the narration can create distinctive effects by letting us trust too much in what we see at the outset.10 This syuzhet strategy has been put to good use in films like The Usual Suspects (1995), which makes us revisit initial action and rethink what we thought we knew. My account of narrational uptake may seem cerebral and juiceless. Surely, our infer- ential elaborations are bound up with emotions? They are, and just as modern cognitive science presupposes that emotions operate in tandem with perception and thought, so I’d readily grant that our time-bound process of building the story is shot through with emotion. Murray Smith, for example, has traced how the complexities of narration can tie us to or separate us from the emotions the characters are undergoing, creating “structures of sympathy,” or dissonances between what he calls alignment and allegiance. Thus the narration may signal us that even though we’re tethered to what a character knows, other cues indicate that we are not to ally ourselves to that character’s moral frame of reference, so that our response may blend sympathy, empathy, and emotional distance.11 Of course, narratives can evoke a very wide range of emotions, but Sternberg suggests that the big three are the ones most basic to our narrative engagement. This is because they are central to comprehension—the perceiver’s construction of the fabula, and other emotional responses will depend largely on that. An additional advantage of treating narration from the standpoint of poetics is that it lets us track different storytelling traditions. Classical Hollywood construction may dis- tract us along its path to the end, but eventually we arrive at fairly definite and reliable inferences. By contrast, other traditions, such as that of “art cinema,” open gaps that ar- en’t closed, trigger inferences that don’t have clear-cut conclusions, and use fluctuating patterns of time and space to create a more unreliable presentation of events. Films such as Toto le héros, Blind Chance, and Les Passagers set into motion narrational systems that don’t resolve at either the level of the story action or that of syuzhet organization. Such films give the spectator an experience of patterned ambiguity about events or states of mind, a play among competing schemas, and an invitation to interpret the film more ab- stractly. By thinking of narration along the lines I’ve sketched, we’re in a good position to make our poetics of storytelling comparative. I’ve made efforts in this direction in Narra- tion in the Fiction Film.12 Finally, some people have objected that by emphasizing the flow of information about story states and actions, I make films too dependent on revelations and plot twists. Every movie becomes a mystery story, my critics suggest. 20 Poetics of Cinema Reeling out narrative   At certain points in history, filmmakers have grouped their scenes into a unit fitting the length of a film reel in projection. In the years before 1912, fic- tion films usually consisted of only one reel. Projection speeds weren’t standardized, but the maximum running time per reel was about 15 minutes. So a technological constraint served as a simple boundary for the entire story to be told. As films became longer, they were broken into several reels. In theaters with only one projector, the end of one reel would be followed by a break while the projectionist thread- ed up the next. Even at theaters equipped with two projectors, there might be a distinct pause between reels. Recognizing that the presentation would be segmented, filmmakers began to build their dramatic arcs around the reel break. Urban Gad, a Danish director who immigrated to Germany in 1912, noted that the “mechanically necessary interrup- tions” demanded that the film be divided into “acts,” each one leading up to a gripping scene just before the reel change.19 By labeling these acts with expository titles, filmmakers invoked theatrical precedent and perhaps also hoped to borrow some of the stage’s pres- tige. The breakdown could be labeled in less standard ways, too; in Lang’s epic Siegfried, each reel is entitled a lay, as in a bard’s song. By the mid-1920s, most European theaters had two projectors, so there was less need for a reel to end on a strong note. But in the USSR, single-projector venues were the standard. Directors were accordingly advised to break the films into well-defined parts.20 Some filmmakers, wanting to make the audience aware of large-scale form in their films, exploited the reel structure to articulate the action of their plots quite vividly. Sergei Eisen- stein is the most famous instance. He broke Strike into six episodes, each marking a phase in the prototypical strike. He gave Potemkin five parts, then split each reel about halfway through, creating symmetrical actions around a caesura. The arrival of synchronized sound standardized running speed at 24 frames per sec- ond, making reel length 1,000 feet, or 11 minutes maximum.21 Films were shipped on 1,000-foot reels, but the biggest venues had projectors that could handle bigger reels, so many projectionists doubled up to reduce changeovers. Hitchcock was counting on this practice when he alternated visible cuts with camouflaged ones in Rope. Reel structure in world cinema still needs to be fully researched, but one recent in- stance is intriguing. In Hong Kong cinema of the 1970s, script construction became fair- ly loose. Filmmakers preferred to build their plots additively, stringing together comedy, fights, and chases. One company, Cinema City, began planning its films reel by reel, de- manding that each reel contain at least one comic scene, one chase, and one fight. Col- or-coded charts revealed immediately which reel lacked the necessary ingredients. The practice influenced most Hong Kong directors who emerged in the 1980s, even the elusive Wong Kar-wai. His wispy plots look more firmly structured when you realize that they’re built up reel by reel in postproduction. The fragmentary martial arts drama Ashes of Time (1994) devotes reel 1 to the primary protagonists, the swordsmen Evil East and Poison West. The plot spends its next two reels on the story of the Murong broth- er–sister couple, then devotes reels 4 and 5 to the Blind Swordsman. The film finishes with a three-reel denouement involving the protagonists and the woman they both love. Three Dimensions of Film Narrative 21 Shooting without a finished script and welcoming spontaneous digressions, Wong used the Hong Kong tradition of reel-by-reel construction to shape his masses of footage.22 Reel construction is a fairly loose metric for plot structure. Provided with merely a proportional segmentation, the filmmaker must still create more specific patterns of action that will fill it out. Perhaps the closest analogy is the word count assigned to serial publication of a novel’s chapters, or the standard number of lines per verse in epic recita- tion or popular songs. Plot structure as act structure   Internal and external criteria blend in one of the para- mount conceptions of structure at work in mass-market cinema today—the notion that a film narrative divides into distinct acts. Across the history of drama, act structure is a vexed question. Some people think that Aristotle’s beginning-middle-end dictum corresponds to a three-act layout, but that’s false. Aristotle nowhere refers to acts, for the good reason that ancient Greek dramas didn’t them. Roman drama did, but the critic Horace proposed that the best number was not three but five. This precept guided playwrights and publishers for centuries in En- gland, France, and Germany. Spanish dramatists of the 16th and 17th centuries promoted a three-act structure, which Hegel praised as the most theoretically correct design. (It neatly echoed his thesis–antithesis–synthesis triad.) But the five-act conception persisted through the 19th century, encouraged by Gustav Freytag’s influential argument that plot structure pivoted around a climax coming midway through the play. By the early 20th century, most operas and plays seem to have favored three acts. What of cinema? There’s no doubt that the analogy between dramatic acts and film is fairly forced, especially after screenings no longer included breaks between reels. Per- haps screenwriters adopted the three-act model simply because it was the norm in theatre. Although there’s some evidence that the three-act structure held sway during the classic studio years, it was widely disseminated in screenwriting manuals after the 1970s, chiefly thanks to Syd Field’s influential book Screenplay.23 Field claims that Hollywood films adhere to a three-act structure, having the rough proportions of 1:2:1. In the first act (25–30 minutes into a two-hour film), a problem or conflict is established. The second act, running about an hour, develops that conflict to a peak of intensity. The final half hour or so constitutes a climax and denouement. Field translated this structure into a screenplay’s page counts, with each page counting as rough- ly a minute of screen time. This plot anatomy has been taken virtually as gospel in the U.S. film industry, with producers expecting submitted screenplays to adhere to it. It is as fundamental to Ameri- can studio screenwriting as the 12-bar blues structure is to pop music. The three-act tem- plate has been endlessly tweaked, recast, and filled out. With scholastic zeal, although seldom with scholastic acuity, commentators have discussed what kind of action is appro- priate for each act, such as “backstory” during the first act and resolution in the last. Most writers agree that the end of the second act should be the “darkest moment,” the point at which things seem to be utterly hopeless for the protagonist. Yet getting there can pose problems; “the desert of the second act” is the toughest stretch, most writers agree. 22 Poetics of Cinema An alternative to the three-act template was proposed by the distinguished screen- writer Frank Daniel. He taught that the plot can be analyzed into eight sequences, each running about 15 minutes. Still, this isn’t a drastic challenge to conventional wisdom, be- cause these sequences can easily be slotted into the broader pattern of three acts.24 Goal-oriented part structure      Within film studies, and specifically within a research program in poetics, the most salient revision has been proposed by Kristin Thompson.25 She argues that since the late 1910s, an American feature film tends to be constructed in 20–30 minute chunks, each marking a distinct phase in the plot. The parts are defined not only by running time but also by the formulation, redefinition, and achievement of goals set by the protagonist. According to Thompson, the film’s Setup section endows the protagonist with a set of goals. The following section is the Complicating Action. This recasts or even cancels the initial goals and ends with a new set of circumstances governing the action. This situa- tion may serve as a “counter-Setup,” reversing the conditions that governed the first part. Thompson calls the next section the Development, launched at approximately the mid- point of the film, in which efforts to achieve the goals are thwarted. Although there may be some forward movement in the main action, some portions are likely to be rather static, emphasizing subplots, character revelation, or simple delays (fights, chases, comic bits). Characteristically, the Development ends with a piece of action that puts the achieve- ment of goals into a crisis. The plot’s final section constitutes the Climax, in which the protagonist definitely achieves or doesn’t achieve the goals. The Climax is often followed by an Epilogue, which asserts that a stable situation has been achieved. Thompson’s account, tested and refined in relation to many films, is an inductive gen- eralization, and as such it usefully refines the three-act template. Instead of describing the events that lead into the next act as incidents that “spin the action in a new direction,” the most common formula for a “plot point,” she is able to specify that the principal character will define or change the relevant goals. Her model also allows that not all portions of the plot will be proportional. Indeed, it turns out that the Climax of a film is seldom exactly as long as the Setup. The Setup usually runs 25–30 minutes, but climaxes tend to be 20 minutes or so. Thompson recognizes as well that a film may not run exactly two hours, a problem for the three-act template. She suggests that a shorter film may display the four basic parts, or it may possess only three, deleting either the Complicating Action or the Development. Likewise, a longer film may have two Complicating Actions, two Developments, or even, as in In Cold Blood, two Climaxes. In all these respects, Thompson’s account is a function- alist one, based on major changes that take place within the plot action, and not simply on external measurement of minutes or page lengths. Four-part plotting in practice   Few would deny that You’ve Got Mail (1998) is a pretty formulaic movie, but studying its structure along these lines helps sharpen our sense of how the formula works. Running 115 minutes, the film fits Thompson’s model snugly. The Setup introduces the classic Hollywood dual plotline: a line of action devoted to work and a line devoted to romance, each of which will affect the other. Joe Fox and Three Dimensions of Film Narrative 25 In the Climax, he abandons Brandy and his father to join his pals on a mission of revenge. The narration intercuts their search for the gang, Furious waiting anxiously at home, and Ricky’s grieving mother learning that he passed his college entrance tests. But Tré has a change of heart and leaves his friends, who go on to wipe out the gang members. Next morning, Doughboy cracks his tough façade to confess to Tré his loneliness, fear, and despair at the cycle of violence. Titles provide an epilogue. Doughboy will be murdered 2 weeks later, presumably in revenge, whereas in the fall Tré will attend Morehouse College “with Brandi across the way at Spelman College.” The protagonist’s twin goals—striving for a better life and achieving romantic union—have been achieved. Boyz N the Hood gives human weight to abstractions about youth, crime, drugs, fam- ily ties, and hope within black urban communities, and it does so through a plot that fol- lows the four-part template as faithfully as does You’ve Got Mail. This traditional structure can smoothly absorb a variety of subjects and thematic materials. Just as important, it has proven itself a reliable way to engage and sustain curiosity, suspense, and surprise. The Setup need not set up everything. Later phases usually de- pend on revealing backstory that was left out, or teasingly hinted at, in the first section, so curiosity can be aroused early on. Similarly, as the characters pursue goals, or block others’ goals, we build up expectations about future events. The Complicating Action can provide surprises, as the original pursuit of goals takes an unexpected turn. The Devel- opment section, which often yields important exposition about characters’ pasts or their deeper motivations, can satisfy our curiosity about what led to this situation—while also sharpening our anticipation of how it will be resolved. By the time we reach the climax, the possible outcomes are reduced to a fairly well-structured set of alternatives. Either Joe and Kathleen will get together, or they won’t. Either Tré will stick with his friends’ vigi- lante mission, or he won’t; and either he’ll go to college or stay in the neighborhood. The structure focuses our attention, and our emotions, on a clear-cut resolution of the action. Hidden rules?   Someone might argue that these models of plotting invite you to read in what you expect to see. Because you expect something important to happen around min- ute 25, you’ll tend to exaggerate the importance of whatever happens then. You’re looking for 3 acts or 4 parts, and you massage the film to fit it, but someone else could plausibly claim that the film consisted of 7, or 17, parts. Aren’t these measures just ad hoc? I don’t think so. Although all events in a plot may contribute to the overall progres- sion, some intuitively stand out as significant moments, and others are clearly secondary. There’s a lot of agreement among us as to what those moments are, and they occur, with a frequency greater than chance, at the points and with the consequences that Thompson’s model predicts. No one would argue that the visit of a college recruiter to Ricky’s home isn’t significant for the action of Boyz N the Hood, but it isn’t as central to all the charac- ters’ fates as Ricky’s death is, and that’s the event that arrives at a canonical juncture in the running time. Granted, analysts can disagree about a particular film’s structure; Thompson and I don’t divide The Godfather in precisely the same way. But such disagreements are common within any critical tradition. Musicologists may disagree about the most perspicuous way 26 Poetics of Cinema to analyze a particular melody, but all accept the premises of phrasing and harmonic pro- gression that give the tune its identity. Thompson’s layout is a helpful tool for analyzing films made along classical lines, lay- ing bare constructional principles that seem widely used. Yet it raises some intriguing problems about the explanatory power of a poetics. Do viewers recognize these distinct parts? Apparently not. People are usually sur- prised when told of them. It seems that this architecture achieves its effects without the audience’s conscious awareness; only experts detect the armature. This fact need not count against Thompson’s account, because listeners with no musical training can react properly to a song or symphony without being aware of the mechanics of harmonic modulation, retrograde inversion, and other techniques. What, though, about the practitioners? We can easily find the three-act/four-part model in contemporary cinema. The success of Field’s book in translation has probably led filmmakers all over the world to try following his recommendations, and Thompson’s model preserves much of his three-act paradigm.26 But what about practitioners who are ignorant of the paradigm and still turn out properly patterned scripts? How could they obey rules that they don’t consciously know? Another problem: Screenwriting manuals recommend that scripts have the three-act structure, but the analysts derive their timing recommendations from finished films. We know, however, that scripts are constantly modified in the production process, and the film as shot can be recut in many ways. It would be a miracle if everyone involved, from scriptwriter to editor, tacitly subscribed to a canonical structure without being aware of it. Yet both the script gurus and Thompson show that the finished films, from the 1920s to the present, display this structure to a plausible degree. Unless we’re hallucinating, some- how the miracle does happen. It would be still more puzzling if a comparable model reigned outside the Hollywood tradition. Yet Francis Vanoye has suggested that films by Claude Autant-Lara, François Truffaut, and Andre Téchiné adhere to the three-act structure, though he offers no expla- nation of how this American paradigm found its way to France.27 Michel Chion finds it in Mizoguchi’s Sansho the Bailiff (1954).28 I can’t comment on Vanoye and Chion’s analyses, except to find the Mizoguchi in- stance fairly implausible. Concentrating just on the Hollywood tradition, I’ll hazard the suggestion that there the structural norms are the product of a tacit craft practice. We have some evidence that writers may spot the regularities in their work only after the fact. Consider one of the new wrinkles Thompson introduces. If the three-act structure is the formula guiding today’s filmmakers, how could they have embraced a four-part structure? Nearly all writers acknowledge that the lengthy second act is difficult to write. It wouldn’t be surprising that scenarists made this stretch tractable by tacitly breaking it into two roughly equal chunks. At least one writer (speaking after Thompson’s first study was published) has acknowledged that the three-act structure is best thought of as harboring four parts. Akiva Goldsman remarks that a screenplay consists of “four acts, or really three Three Dimensions of Film Narrative 27 acts, but the second act is really two acts, so we might as well call it four acts, and they’re generally 30 pages long.”29 From this angle, splitting the second act would be a craft habit that just doesn’t rise to the level of awareness. Writers thought, and perhaps still think, they are working with three acts, but once it’s pointed out, they can just as easily accept that they unconsciously subdivided one of them. This leads to an important point. We shouldn’t assume that all creators have an engi- neer’s grasp of what they’re doing. Usually they’re just following a tradition whose features they’ve intuitively grasped, and the tradition gets replicated without a lot of self-conscious reflection. Generalizing from this tendency, perhaps the three-act/four-part structure lies very deeply embedded in the work process. It has been so strongly “overlearned” that it in- forms the basic choices of key personnel. (It’s worth mentioning that the four-part scheme can be found in popular novels as well.) Reinforcing this particular scheme are general principles of symmetry and a widespread idea that entertainments are easily digestible in 20–30-minute chunks. Still, the question of how creators hit upon these norms of plot structure is far from settled. Apart from its heuristic value in bringing out the macrostructure of many films in a major tradition, the act-based model of construction nicely lets us distinguish between narration and plot structure. Consider Memento (2000). Narrationally—that is, in terms of the strategic regulation of fabula information—major sections of the syuzhet present story events in reverse order. Yet Memento’s syuzhet obeys the three-act/four-part template, with turning points at the proper proportions.30 Odd as it sounds, even telling the story back- ward can respect canonical plot architecture. The Narrative World Most books introducing narratology start with discussions of the fabula, that spatio-tem- poral realm in which the action unfolds in chronological order. Then the author goes on to discuss how that world is rendered through patterns of narration (or “discourse”)—re- stricted point of view, flashback construction, and the like.31 This expository strategy makes for clarity, but it’s a little misleading. It dodges the obvious fact that we have access to the fabula only by means of narration. Narration isn’t simply a window through which we watch a preexisting story that we might see from elsewhere. By telling the lawyer joke at the start, I coaxed you into creating the story world by virtue of our shared stock of stereotyped knowledge. Narration, the interaction of the syuzhet arrangement and the stylistic patterning of the film, is the very force that conjures the fabula into being. The demiurgic power of narration is especially hard to grant with respect to cinema. Literary texts conjure up worlds from mere words, but film presents us with a rich array of images and sounds that immediately presents a dense realm. So it’s easy to succumb to what used to be called the “referential illusion,” the sense that a tangible world lies behind 30 Poetics of Cinema Characterizing characters   Most people couldn’t imagine a narrative without characters, those person-like entities that make things happen and respond to various doings in the story world. Accordingly, the rest of my survey here concentrates on characterization. Seymour Chatman has pointed out that we fill out characters through implication and inference, just as we do with story lines.36 In this, he agrees with Gustav Freytag: The poet understands the secret of suggesting; of inciting the hearer, through his work, to follow the poet’s processes and create after him. For the power to understand and enjoy a character is attained only by the self-activity of the receptive spectator, meeting the creating artist helpfully and vigorously. What the poet and the actor actually give is, in itself, only single strokes; but out of these grows an apparently richly gotten-up picture, in which we divine and suppose a fullness of characteristic life, because the poet and the actor compel the excited imagination of the hearer to cooperate with them, creating for itself.37 I think that we can make our imaginative activity even more explicit with the aid of a bit of cognitive science. We construct the characters within their narrative world as persons, and it seems to me that we employ a schematic prototype for personhood. This prototype isn’t a rigorously philosophical account, but rather an intuitive sketch of our folk psychology.38 A person, let’s say, possesses a body, presumed to be unified and singular (and thus gendered). A person perceives and is self-aware; entertains thoughts, including beliefs and desires; feels emotions; possesses traits, or attributes; and can launch self-impelled actions. In addition, complementary to the concept of a person is the idea that any person can play various social roles. We don’t acquire this prototype all at once, but there is strong evidence that we’re disposed to acquire this sort of information about others. We are born pre-tuned to see people as people, not inert objects, and equipped with faces, insides, and even minds. If our environment confirms these predispositions, we can go on to learn a host of other things about our fellow creatures. As we grow, we can apply that knowledge to under- standing stories.39 Presented with a narrative agent, we tend to project the whole cluster of schematic features onto him or her or it. This is Ryan’s Principle of Minimal Departure at work again, because we expect the agents we encounter in the world to come supplied with all the aspects of the schema I’ve outlined. So the narrative must tell us if the agent lacks any of the critical features. In many science fiction films, we’re informed that an intelligent robot can’t feel emotions; such, apparently, is the case with HAL 3000. In many cases, we’ll as- cribe characters’ actions to beliefs, desires, traits, or social roles on rather slender evidence. We assume that characters have all the person-like attributes, and such assumptions allow us to fill gaps and inventively extrapolate. Of course many of these extrapolations will be quick and dirty, guided by social ste- reotyping. My lawyer joke relies on two conventional premises: Lawyers are scoundrels, and people don’t want to be considered assholes. (The failure of the second provides the joke.) When Ryan, the protagonist of Cellular (2004), carjacks a Porsche, the narration de- picts the victim in quick strokes. He’s a lawyer, and he’s characterized in a way compatible Three Dimensions of Film Narrative 31 with our opening joke: In his cell phone conversation, he’s rude, lewd, loud, arrogant, and generally assholish. Our inferences about his personality are reinforced by the sight of his face (aggressively beaverish) and his personalized plate (WL SU YOU 2), all supported by ethnic stereotyping (he’s evidently Jewish). The look, demeanor, and voice of the lawyer in Cellular remind us that, contrary to literature, films present characters with distinct and identifiable bodies. These play a cru- cial role in cueing us to construct personal features for them. From the way Sean Thornton stands and speaks in The Quiet Man (1952), we can believe he’s been a boxer. And whereas in stage performances the same character role can be occupied by different actors’ bodies, films tend to identify the character with the singular physical presence of an actor. Once the actor has played other roles and become famous, a star persona builds up, passing be- yond the body and voice to other features of personhood. Our conception that Humphrey Bogart is cynical, insolent, and worldly wise informs both his private life and his screen characters. All together now   Few films contain only one character, and the story world we build up is populated by an ensemble of persons, which we distinguish from each other along at least two dimensions. We intuitively grasp a hierarchy of characters, making some more important than others, and we do this partly because of the degree to which their narrative functions activate aspects of the person schema. A hotel clerk may exist solely to check our hero into a room, and thus only the clerk’s body, his social role, and his capacity for voluntary action are relevant to narrative cau- sality. But the narrative can characterize the clerk more fully by endowing him with su- perciliousness (a trait), exasperation (an emotion), or suspicions about the hero’s identity (thoughts). As a more vivid individual, the clerk will be more salient than other func- tionaries who flit through the story world, as the lawyer in Cellular stands out from the other, more anonymous drivers whom Ryan tries to flag down. If the clerk ’s attributes provide causal impetus for the action, then he will move up in the hierarchy of characters. In Cellular, the lawyer reappears in comic terms, quarreling with a policewoman, before Ryan swipes his Porsche a second time. He is promoted to greater importance as a more distinctive individual and as a causal factor in prolonging the action. Apart from ranking characters in their relative importance, we quickly liken and con- trast them, using the dimensions of personhood I’ve indicated. Classic oppositions offer clear instances: The hero may be young and virtuous with an attractive body, whereas the opponent may be old, vicious, and misshapen. The Cellular lawyer is selfish and unfeeling, whereas Ryan sacrifices a lot out of sympathy for Jessica, the kidnapped woman calling on his cell. Marc Vernet points out that narratives tend to array their characters’ most salient features according to overlapping contrasts and affinities.40 In a heist film, one crook may be greedy, good-looking, and nervous; another may be greedy but average-looking and confident; a third may be self-sacrificing but ugly; and so on. In Cellular, Ryan and the cop Mooney are both compassionate, because both try to rescue Jessica, but they’re otherwise quite different in social roles (one is a surfer dude, and the other is a cop), bodies (hand- some young versus weathered middle-aged), and traits (impetuous but resourceful versus 32 Poetics of Cinema prudent but dogged). Other characters display a mix of these features, along with still oth- ers. As in the world, we contrast the people around us along various axes of personhood; but in daily life the contrasting features run on to indefinitely large numbers. A narrative simplifies our task by displaying contrasts along a fairly small number of axes and stressing the salient ones as the action unfolds. Social intelligence, the quick and the dirty     As we grasp the film’s hierarchy and the contrasting features of the characters, we make inferences. Here our social intelligence may not follow strict deductive or inductive rules. It’s now well established that informal reasoning about others relies on heuristics, fast and somewhat dirty conceptual short cuts. The classic instance is the fundamental attribution error. We tend to see others’ ac- tions as caused by personal traits rather than situational constraints, whereas we tend to see our own actions as shaped by circumstances.41 If you’re grumpy, it’s because you have a sour disposition, but if I’m grumpy it’s because I’ve had a bad day. In the real world, such attributions are mistaken, but narratives rely upon them all the time to secure fast uptake. Often we’re introduced to characters in ways that encourage us to ascribe their actions to their personalities rather than to the situation.42 Is this tendency elicited only by plot-driven movies that have to announce the heroes and villains swiftly? Not necessarily. Michael Newman has shown that even the so-called character-centered films of American independent cinema, such as Welcome to the Doll- house (1996), encourage us to explain characters’ actions by plans, desires, and character traits rather than by situational factors.43 Narratives play on other folk-psychological shortcuts. The primacy effect that I men- tioned earlier—the power of first impressions to establish the conceptual ground rules—is strengthened by “belief perseverance,” our tendency to resist changing a judgment, as well as “confirmation bias,” our unwillingness to entertain evidence that would countermand an initial impression. Narratives are designed to give strong and accurate first impres- sions of their characters, and only a few narratives are designed to introduce evidence that would make us change our judgments. Likewise, people usually don’t reason statistically, but rather on the basis of vivid ex- amples. Buyers of lottery tickets can imagine themselves winning or recall the winners they’ve seen on TV, whereas it’s much harder to concretely imagine the odds of 13 million to one. Murder is far rarer than suicide, but people think it’s more common because they have vivid exemplars from popular media. Perhaps this “availability” heuristic undergirds our willingness as viewers to accept that every walk down a darkened street is dangerous, or that lovers will accidentally meet in dramatic circumstances; it’s easier to imagine them meeting than not meeting.44 Our shortcomings in purely logical reasoning may well stem from evolutionary biases toward acting in the here and now, particularly when operating in small groups. Ecologi- cal scientist Bobbi Low has suggested that our “illogical associative thinking” stems from self-protective strategies that evolved in the context of social situations. We are logically inept, but socially adept. One experience at being cheated, and we are likely to generalize to future interactions with individuals of that category. One dangerous event witnessed, and we fear it ever afterward. We Three Dimensions of Film Narrative 35 Only fairly late in the film does Claire articulate her hope to pay off her debts and have a baby. At that point, earlier incidents, such as her kind treatment of a little girl she meets on the street, retrospectively cohere into a pattern. A more classical narrational strategy would have treated each trick she turns as a step toward breaking free, but by concealing Claire’s goals, the narration throws all the emphasis on her daily highs and lows, which seem to be leading her nowhere.54 Delayed exposition of the character’s desires and plans can give a shape to the action within the story world, but what if the character has no desires and plans, or at least no definite ones? What if the character is more passive, reacting to others rather than initiat- ing action? There is a tradition of filmmaking, associated with the “art cinema,” that puts such characters to the forefront. In such instances we must construct a less causally driv- en story world, one ruled by passivity, chance encounters, and emblematic episodes that evoke psychological and social themes. The homeless Mona is purportedly the central character of Agnès Varda’s Vagabond (Sans toit ni loi; 1985), but we come to know her chiefly through her wanderings through the countryside and the people she encounters there. As her life accidentally touches theirs, the narration reveals a cross-section of the civilization she has fled, surveying day laborers, housekeepers, yuppies, thieves, and professors. In the process, we come to know these peripheral characters far better than we know Mona. She remains psychologically opaque, not least because she doesn’t have any goal that will define her sense of herself. In a way, she lacks that dimension of personhood we associate with beliefs and desires; her willful solitude is impregnable. My later essay in this volume, “The Art Cinema as a Mode of Film Practice,” tries to clarify the ways in which films like Vagabond contribute to a relatively distinct tradition of cinematic storytelling.55 How characters change, or not   I’ll close this gross mapping of story-world construction by considering one more issue related to characterization. Apart from the overall pattern of activities undertaken by a character in that world, we sometimes encounter cues for what we usually call character change. This is a slippery notion, I think, and can cover several of the dimensions of person- hood I’ve mentioned. Characters can change their social roles (e.g., a cop can enter the clergy), their sensory capacities (a blind man can regain his sight), and their emotional states (a frightened man can become calm). What we usually mean by it, however, is that characters change their thoughts or their traits. In a great many narratives, characters alter their beliefs, desires, attitudes, opinions, and states of knowledge. Call this epistemic change. In The Birdcage (1996), parents biased against homosexuality eventually learn to tolerate their future son-in-law’s gay parents. This sort of coming-to-realize-the-truth change is quite common in films and is particu- larly valued when it’s a change of knowledge not about external affairs (as when the detec- tive dispels a mystery) but about internal states. A sophisticated narrative, many people believe, forces a character to better understand the sort of person he or she is. This dynamic takes on a particular shape in mass-art storytelling, whereby the char- acter faces up to a mistaken judgment. Hollywood screenwriting manuals strongly sug- gest that there be a “character arc,” whereby a basically good person comes to recognize 36 Poetics of Cinema that they have erred and try to improve. The skyscraper siege in Die Hard (1988) gives its hero, John McClane, the chance to realize how much he loves his wife and to regret that he wasn’t “more supportive” when she wanted to advance in her career. “In the most simplistic terms,” says one screenwriter, “you want every character to learn something…. Hollywood is sustained on the illusion that human beings are capable of change.”56 From this angle, change amounts to modifying a judgment, admitting a slip, or, as in the case of an erring spouse, realizing that the goal of an extramarital affair was an unworthy one. The sort of change that many consider the essence of a high-quality narrative is more radical, involving a change in fundamental traits. Epistemic change can fuel some chang- es in personality, but to alter a trait is to become a different person. Having learned his lesson, McClane will be a more tolerant man, but nothing that happens in Die Hard will induce him to become a pacifist, in the way that Scrooge becomes charitable and Oedipus becomes humble. It’s one thing to change your mind, another to change your heart. “Any character, in any type of literature,” writes Lajos Egri, “which does not undergo a basic change is a badly drawn character.”57 One of the enduring contributions of Egri’s book The Art of Dramatic Writing, first published in 1942, is to show how trait change can mesh with the classic approaches to plotting summarized by Brunetière. “A character stands revealed through conflict; conflict begins with a decision…. No man ever lived who could remain the same through a series of conflicts which affected his way of living.”58 Egri points to Othello, Tartuffe, Hamlet, Willy Loman, and other characters who change in the course of the drama. Many of the changes are alterations in knowledge of the kind I’ve just indicated, but some are more radical. Egri’s prime example is Nora Helmer in A Doll’s House, who starts as a superficial, coddled wife and becomes a mature, rebellious woman. Through extensive quotation, Egri traces how carefully Ibsen displays Nora’s growing understanding of her situation, which in turn allows her to develop traits we could scarcely suspect she had. The trick, Egri shows, is to let the situations force the character to change step by step. Julie’s character arc   A parallel instance in film is Jezebel (1938). In antebellum New Orle- ans, Julie Marsden conducts a tempestuous courtship with Pres Dillard. She’s headstrong and willful, always prepared to flout convention in her demands that he put her first. She pushes her luck, however, by defiantly wearing a scarlet dress to a society ball. When she realizes how she’s spurned by everyone, she wants to leave the dance. To punish her, Pres forces her to stay, then takes her home and breaks off their engagement. Her self-confi- dence is shaken, and though she insists he’ll be back, she chokes back tears. A year passes. Pres has left for the North, and she has become a recluse in her Aunt Belle’s house. When Julie learns that he is returning, she bursts with hope, determined to beg his forgiveness. “I was vicious and mean and selfish. And I want to tell him I hated myself for being like I was.” This is already a considerable growth; the Julie we meet in the opening scenes would never have humbled herself. Nonetheless, her fierce energy hasn’t abated, and she throws herself into preparing the plantation household for Pres’ return. But she is shattered when Pres arrives with his new wife, Amy. She vows to get him back and bends her energies to the task: “I’ve got to think, to plan, to fight.” She is making a big Three Dimensions of Film Narrative 37 mistake to try to seduce a married man, but she’s still so selfish she can’t imagine surren- dering to circumstances that block her happiness. During the couple’s stay, Julie lets loose an escalating string of maneuvers. She tries to rouse Pres’ jealous by flirting with his old rival, Buck. Failing in that, she tries to seduce Pres, and failing in that, she goads Buck into defending her honor. Having provoked a duel between Buck and Pres’s brother, she realizes that she’s cruelly playing with men’s lives. Her self-assurance begins to crack, and she plunges into a hysterical mood, manically leading the plantation slaves in a song. When Buck is shot dead, she feigns indifference but can’t keep from weeping. She has become, Aunt Belle remarks, Jezebel, the wicked woman who made her man a puppet and whose plots brought her to a violent end. The final phase of Julie’s change comes when Pres contracts yellow fever during a trip back to New Orleans. She rushes to his bedside and nurses him through the night. When Amy arrives, Julie asks to be allowed to go with Pres to the leper island that houses fever victims. Her speech is a fine example of the sort of emotional transitions that Egri finds convincing. Julie first points out that Amy doesn’t know enough of southern customs to keep Pres alive in such harrowing conditions. More important, she tells Amy, “I’ll make him live—because I know how to fight better than you.” Finally, she begs, “Help me make myself clean as you are clean.” It’s a rhetorically effective buildup, but the shifts from practical knowledge to the need to expiate her sinful behavior show Julie’s own growth. It’s not a total makeover; she retains her characteristic boldness, tenacity, and imperious force. But she has changed in other ways. Now, instead of serving her whims and self-importance, her willfulness will sustain the man she loves. The selfish Julie of the opening has become the selfless Julie of the final image, a long tracking shot that shows her tending to Pres in a wagon piled with the dying. The biblical Jezebel was flung into a pile of offal and devoured by dogs, but the film sug- gests that on the lepers’ island Julie, who will die from the contagion, will be redeemed. In the wagon, she rides alongside a nun. Character change is usually not as fundamental as it is in Jezebel. Often it’s a reversion to what one once was, or privately already is. The plot action may reawaken the devotion to duty lying dormant in the world-weary cop or the coquettishness in the shy dowager. If Joe Fox in You’ve Got Mail becomes less aggressive through his love for Kathleen, it’s no shock, because we’ve seen his sensitive side pour out in his confessional e-mails. His nega- tive traits seem to be less an essential part of the individual than the positive ones, which, when he meets Kathleen, are put temporarily aside. Elsewhere I trace this process through Jerry Maguire, whereby Jerry’s latent idealism is made to resurface under the guidance of a good woman.59 Another alternative would seem to be the coming-of-age movie, set at a critical period when the character’s traits are still in a process of development. Scout in To Kill a Mock- ingbird (1962) comes to accept the outcast Boo Radley as a friend because she has seen a cascade of unhappy events proceeding from ignorance and fear. All of which isn’t to say that deep-seated character change is impossible in cinema, only that it’s rarer than we might expect. Far more common is character consistency, with the plot being driven by a clash of purposes; slow character revelation, achieved by delay- 40 Poetics of Cinema impersonal mode. At many moments, each novel’s narrator comes forward and projects a certain attitude toward the action represented.62 But in cinema, it seems to me, the narra- tion’s source is seldom so explicit. Exceptions would be the opening and closing sequenc- es, when we’re sometimes aware of being directed to notice this or that detail. But this just seems a case of self-conscious address, as the narration frankly acknowledging its act of emphasizing an item. Furthermore, even when confronted with such self-conscious passages, we don’t characteristically attribute them to a narrator. For ordinary audiences, the relevant agent or agents are the filmmakers, commonly known as they. “At the start of a movie,” someone might say, “they always show something important to the plot.” In a memo, Darryl F. Za- nuck sums up patrons’ complaints about The Gunfighter: “Why didn’t they let him live at the finish? After all, he had been reformed. He could have been wounded, if they wanted to shoot him.”63 We needn’t of course take ordinary responses as wholly determining our theoretical concepts. Many readers would identify the speaking voice in Pride and Prejudice as that of the author, Jane Austen. Many other readers would understand that the intruding narra- tive voice of that novel is not necessarily that belonging to the author. But very few viewers would take, say, a bit of actors’ business or a pattern of lighting as having its source in an intermediary, a cinematic narrator, rather than to either “the film itself ” or the creative individuals on the set. Not everybody shares my intuitions on this matter. I think that this is largely because many theorists think that in explaining the logic of cinematic narration, we don’t need to appeal to any psychological activity. They would claim that even if no viewer ever registers the presence of a cinematic narrator or implied author, any explanatory theory must posit such entities. Why? Because the very concept of narrative requires a narrator, and so any narrative in any medium will have one. My alternative proposal is that in cinema, narration as a process encourages us to build up the story, including the voices and behaviors of particular narrators, but no over- arching narrator is logically required to give us the narration as a whole. As I put it in 1985, “Such personified narrators are invariably swallowed up in the overall narrational process of the film, which they do not produce.”64 Who produces the narrational process? The filmmakers. Let me explain my grounds for this view, and then I’ll return to the case for a narrator. The practical psychology of narrative      Let me recall the mentalistic framework I presented in the opening essay. Films are made by human beings to provide other peo- ple with experiences. Call the second bunch viewers, even though they’re also listeners. The viewers are engaged in the experience by virtue of cues built into the film by the first bunch, the makers. The cues are structured to encourage particular paths of perception, comprehension, and appropriation, all three of these clusters of activities being also in- vested with emotion. The experience proceeds by means of the viewer’s inferential elabo- rations, some of them very fast and mandatory (in the domain of perception), and some more slow and deliberative (typically in the domain of appropriation). Three Dimensions of Film Narrative 41 Filmmakers are practical psychologists. They have been viewers themselves, and they are more or less accomplished practitioners of their craft, so they have many ideas about how to shape the cues to provide experiences of a particular sort. They can fail, or succeed beyond their initial hopes, but they organize the film so as to solicit a range of effects. Like all humans, filmmakers can’t anticipate, let alone determine, all the effects that may arise from their endeavors. Particularly in the domain of appropriation, the viewer has a free- dom to seize upon certain cues and not others, pull them into a range of projects, and use the film in ways that couldn’t be envisaged by the filmmakers. How does this mentalistic framework apply to narrative? Perceiving, comprehending, and appropriating narrative, as well as responding emotionally to it, depend on cues sown through the film. Those cues ask us to grasp the narrative in certain ways. The viewer con- structs, according to the unfolding narration, a story world and a pattern of events within it. That construction becomes a source of emotional and cognitive experiences. Ideally, viewers construct the narrative as the filmmakers hoped they would, but things aren’t always ideal. A viewer may fail to pick up narrative cues, or a filmmaker may fail to make them sufficiently salient. There may be a mismatch between the filmmaker’s schemas and the viewer’s. Cinematic traditions, however, secure a considerable amount of convergence between what filmmakers know can affect viewers and what viewers ac- tually experience, especially in the domains of perception and comprehension. Narrative traditions exist partly to enable this sort of agreement about how the story world is to be constructed and construed. Odd as it sounds to say it, this framework doesn’t mean that communication takes place. If communication means the transmission of an idea or concept from one mind to another by means of some physical vehicle, then that notion doesn’t capture the experien- tial dimension I’m positing. Suppose as an amusement park engineer, I design a roller coaster. You get on at a certain point and undergo a suite of turns, swoops, climbs, and dives. In what sense have I communicated something to you? You’ve undergone a physical and emotional experience that I planned in advance, but I haven’t transmitted any idea or concept to you. Someone might reply that a roller coaster isn’t a good analogy because it doesn’t offer an experience of representations. So substitute a pictorial landscape, like a topiary garden, and my point will be the same. Or consider the layout of a museum display, in which curators arrange the order and position of the items according to principles of what they want to link and highlight. As we stroll through the exhibition, we don’t posit an extra, intermediary figure between human agents and the array that we encounter. True, we may posit some principles that seem to have guided these agents’ decisions. We can presume that principles governing structure, materials, load, and other architec- tural properties governed the decisions of a roller-coaster designer; but those principles needn’t be described as a virtual being. Or take another instance: A map can represent a territory, but understanding map representation doesn’t demand a “terrain presenter” em- bedded in the map. Again, simply attributing the relevant features of the representation to human makers and their plausible intentions suffices to cover the case. 42 Poetics of Cinema Films traffic in concepts and meanings, but these, I submit, are the result of the infer- ential elaboration of cues presented by the design of the work. Just as filmmakers antici- pate that viewers will draw narrative inferences, they often expect that viewers will infer appropriate topics and themes. A narrative film prompts us to assign meanings at many levels, but none is communicated in the sense that a message passes from the filmmaker’s mind to the spectator’s. Rather, a lot of what some theorists would call communication I’d call convergent inference making. The filmmaker has gotten us to walk down the path she planned. If we figure out that Clarice Starling is the protagonist of The Silence of the Lambs (1991), that she is inexperienced and shaken by her past but still courageous and deter- mined, that her efforts to identify Buffalo Bill initiate the story action, and that she is opposed by several other characters but that she wins out eventually… we’ve done pretty well. And while or after making sense of all this, we can go on, thanks to many cues, to find Clarice’s activities satisfying, moving, socially suspect, or whatever. The communication model would say that something passed from the creators’ mind to the movie and then to the viewer. I would say that the creators designed an experience such that viewers are coaxed to construe the film in ways that yield a certain experience more or less roughly foreseen by the filmmakers. Looked at this way, a film becomes a tissue of cues, and these cues can be quite frag- mentary and varied. If I, the filmmaker, want to prompt you to think or feel something, I can shamelessly use anything that can be put into a movie. Any image or sound that gets the job done is a potential candidate, regardless of strict logical consistency. Voice-over and flashbacks      I think that this conception of narrative engineering handles some tough cases. For example, many films open with a voice-over commentary by the central character explaining what led up to the events we’ll be encountering. This commentary’s role is plainly to orient us toward the story world and the plot. It doesn’t necessarily raise such questions as “To whom is the character talking?” or “When is this conversation taking place?” Jerry Maguire’s opening voice-over narration presents what follows as his story, but we see many scenes that he doesn’t witness; the film’s narration is, as we say, omniscient. Yet it would be strange to protest, “But Jerry’s the narrator! He’s telling the story. How could he know what Dorothy told her sister? Did Dorothy confide in him after they were married?” Such questions are as irrelevant as asking whether the giraffes and turtles in a topiary garden could survive in the same ecosystem. The narration has borrowed a piece of ordinary action, somebody talking to someone else, and recruited it for its own purposes (exposition, characterization, setting up larger patterns of sense and emotion). Or consider the anomalies harbored by another common device, the flashback that dramatizes what a character narrator tells. At the start of Leave Her to Heaven (1945), Dick Harland arrives at a dock and is greeted by an older family friend, Glen Robie, before he paddles a canoe to a house further along the shore. With that concision characteristic of classic Hollywood, the first 90 seconds inform us that Dick has spent two years in prison and a woman is waiting for him. But why did he go to prison? Whom is he going to meet? Rather than attaching itself to Dick, the narration stays with Robie and his companion on Three Dimensions of Film Narrative 45 In La Ronde, Jerry Maguire, Leave Her to Heaven, Ten North Frederick, and a great many other films, one or two aspects of a narrating schema are appropriated and collaged with the other components of the narration for purely strategic purposes. As I put it in Narration in the Fiction Film, this condition presupposes a perceiver—you and me—but no sender of a message. “The narrational process may sometimes mimic the communica- tion situation more or less fully. A text’s narration may emit cues that suggest a narrator or a narratee, or it may not.”66 Chatman on the narrator   By contrast, arguments for the necessity of a cinematic narra- tor rely more or less explicitly on a communication model. The most cogent layout of the assumptions here comes, as we’d expect, from one of the most meticulous narratologists of film, Seymour Chatman. He proposes that we need two more constructs to explain the logic of filmic narration: a cinematic narrator that is not as visible or audible as character narrators are, and an “implied author” that is even more intangible. Both constructs are necessary to complete the chain of communication that Chatman sums up in a diagram (see Figure 3.6).67 The narrational process consists of story information passed among a series of agents, some embedded in the text and some not. Every agency emitting narration has its counterpart in an agency that receives it. So the process moves from real author to implied author to cinematic narrator to character narrators (if any) to character narratees (if any) to cine- matic narratee to implied reader to real reader. Let me leave the issue of implied author–implied reader aside for the moment. On what grounds does Chatman postulate a cinematic narrator? He offers both logical and pragmatic reasons. Logically, he says, the very concept of narrative entails a narrator. “Ev- ery narrative is by definition narrated—that is, narratively presented— and that narration, narrative presentation, entails an agent…. Agency is marked etymologically by the –er/-or suffix attached to the verbs ‘present’ or ‘narrate.’”68 But this claim secures only the fact that as an artifact, a narrative owes its existence to an artificer (or several of them). No one disputes this premise. But this is no help to an argument that we need the concept of a text-based narrator distinct from the actual novelist or filmmaker. In other words, we don’t think that narratives fall from the skies. They are created by humans. But the relevant agents in this context are real people, not the postulated agents that Chatman argues for. To undergo the experience of a roller-coaster ride, I don’t have to imagine a ghostly intelligence standing between the engineer and me, shaping the thrills and nausea I feel. The same holds true for the topiary gardener or the mapmaker or the cu- Figure 3.6—Seymour Chatman’s diagram of the communication process in a narrative text. Source: Adapted from Seymour Chatman, Story and Discourse: Narrative Structure in Fiction and Film (Itha- ca, N.Y.: Cornell University Press, 1978), 151. 46 Poetics of Cinema rator designing a museum display. The very concept of a storyteller doesn’t entail a virtual storyteller of the sort that Chatman proposes. In other places, Chatman strays from defending the textual cinematic narrator and reminds us that real agents make texts. He objects, for instance, to my claim that narra- tives are “organized” for perceivers but not “sent” as part of a communication. “Surely,” he writes, “the film—already ‘organized’—somehow gets to the theater and gets projected; something gets sent.” He says that it would be uncomfortable to have “a communication with no communicator—indeed a creation with no creator.”69 I agree that movies get cre- ated and shipped out to theaters, but cinematic narrators aren’t splicing the footage or filling out FedEx forms. In sum, Chatman hasn’t convinced me that a postulated narrator, as opposed to a living and breathing filmmaker, is necessitated on logical grounds. Perhaps, though, con- ceiving of a cinematic narrator offers pragmatic rewards, helping us see new things in narrative films or offering conceptual solutions to problems thrown up by films. To size up this prospect, we need to ask how we concretely recognize the cinematic narrator. Chatman maintains that the term doesn’t commit him to a language-based concep- tion of cinema. In a film (and presumably a ballet, a mime act, or a wordless cartoon), the narrator isn’t literally a teller; it’s also a shower or, in Chatman’s terms, a “presenter.” This presenter need not be a “recognizably human agency.” “I argue that human personality is not a sine qua non for narratorhood.”70 What, then, is the equivalent of the speaking or writing voice we encounter in literature? The cinematic narrator, Chatman explains, is “the composite of a large and complex vari- ety of communicating devices.”71 What devices? The list is open-ended and includes audi- tory elements (speech, noise, music) and image-based ones (mise-en-scène, editing, cin- ematography, etc.). These are all deployed by “the overall agent that does the showing.”72 No one will disagree that these elements are resources that filmmakers have at their disposal. In a film, these techniques represent the narrative, as I’ve discussed in the section on narration. But this list of features is something of a letdown after several pages of the- oretical argument for the utility of positing a cinematic narrator. All these techniques of representation are just as easy to analyze by speaking of the film’s form and style tout court, along with the effects we propose that these features aim to produce. Critics and analysts have been appealing intelligibly to these concepts for decades without assigning them to a narrator. We need never invoke an extra intelligence that is bending them all to its will (apart, again, from a real filmmaker or set of filmmakers). Chatman’s cinematic narrator looks like simply a label for the systematic formal and stylistic properties we can detect in any narrative film. By the principle of Ockham’s razor, the pragmatic utility of the narrator concept seems questionable. Chatman suggests that thinking of the narrator can be helpful in certain problematic cases, as when we try to track unreliable narration. When the image track contradicts the soundtrack, as in Badlands (1973), we have “a conflict between two mutually contra- dictory components of the cinematic narrator.”73 Again, however, what have we gained by postulating this extra agent and then saying that two “components” of it clash? Why Three Dimensions of Film Narrative 47 not simply say that we encounter an organized disparity of image and sound? From the standpoint of theoretical parsimony, what more does the virtual figure of the narrator add? For every narrator, a narratee     The communication model holds that for every sender, there’s a receiver. So if there’s a cinematic narrator, there must be a narratee: not the real viewer, nor the “implied viewer,” but a pickup agent at the other end of the narrator’s communiqués. But most theorists holding this position tiptoe around the narratee. I think that’s be- cause such a creature doesn’t possess even the gossamer presence of the cinematic narrator. The narrator is at least visible and audible via technical devices, the equivalent of the words of the literary narrator. But where does the text provide signs of the narratee? And what properties can be attributed to him or her? We can call Huck’s narration plain-spoken and the third-person narrator of Pride and Prejudice wryly judicious, but what attributes can we ascribe to the literary narratee, let alone its cinematic counterpart? Chatman’s discussion of the narratee in his 1978 book focuses principally on literary character narratees, those dramatis personae who attend to what character narrators say. These are uncontroversial cases, because the literary texts are representing someone in the story world telling the tale to someone else in the story world. Chatman also considers diary narratives (the writer becomes his or her own narratee) and the sort of “Dear reader” entity that is sometimes signaled by an impersonal narrator of the Austen sort. There is no discussion of the cinematic narratee.74 Chatman’s 1990 defense of the concept of the cinematic narrator, consuming fourteen pages, never mentions the narratee. I suspect that this is because there is almost nothing to be said about it. The concept does no theoretical work. All we can say is that some posited entity is picking up the significance of every shot, line of dialogue, piece of performance, and so on emitted by the cinematic narrator—and then relaying that information to the implied reader-viewer, who then relays it to flesh-and-blood viewers. Positing so many ghosts in the textual machine suggests once more that the communication model isn’t the most fruitful way to understand narration. Film as/versus literature     There’s a general point at issue here. Marie-Laure Ryan pro- poses that there are three positions to be taken on narrative across media.75 One can hold that narrative exists only in verbal media; few currently take this line. Or one can take narrative as a fuzzy set of features, but hold that narrative is most fully implemented in language, and thus the parameters of verbal language must be present in other media too. The theorist will accordingly look for parallels to fictional voice, literary point of view, the narrator–narratee relation, and so on. It seems evident that Chatman holds this second view. He presupposes that lan- guage-based narrative contains the components necessary to define or describe narra- tives in other media. That is, in order to characterize cinematic narrative, we must recast concepts derived from literature (specifically, concepts based in literary communication). Chatman’s overall taxonomy suggests that there are no narrative techniques possessed by cinema that cannot be found in literature, though cinema can actualize those techniques in strikingly different ways. 50 Poetics of Cinema gest, makes Huck and Tom look shallow and cruel, casting a shadow over the friendship that Huck and Jim have shared in the bulk of the story. This last example is apt, because in the history of American criticism, one impulse behind theorists’ creation of the implied author is the need to account for unreliable nar- ration while avoiding what many consider the “intentional fallacy.”84 As Chatman puts it, the implied author yields “a way of naming and analyzing the textual intent of narrative fictions under a single term, but without recourse to biographism. This is particularly im- portant for texts that state one thing and imply another.”85 If the grounds for the cinematic narrator are said to be both logically necessary and pragmatically useful, Chatman claims only pragmatic utility for the concept of the implied author. And some theorists who don’t embrace the concept of a cinematic narrator do ac- cept the implied author as operative in both literature and film. My own response, though, is a skeptical one. If the implied author is mainly a solution to the problem of unreliability, I would suggest that the problem be solved differently. The case of the divergent inferences     We could put it this way. The text prompts the reader to construct the story action a particular way, and that construction includes rec- ognizing the gaps and shortcomings of the narration, given the norms in force. We judge a literary narrator to be unreliable through inferential elaboration of the cues she or he presents, and that elaboration may be at odds with the inferences drawn by the narrator. In “Haircut,” we judge Whitey to be unreliable not because an invisible figure is signaling us behind his back, but because Whitey’s judgment of Jim Kendall’s character, on the ev- idence he presents, is ill-founded, according to our norms of behavior. He thinks Jim is a card; we infer that Jim is a bounder. In this respect, literary fiction is no different from real-life reportage or trial testimo- ny. Whatever a speaker says, we balance the information conveyed and the trustworthi- ness of the source against standards of behavior and judgment. When a reporter or trial witness presents information, we don’t infer an “implied recounter” or “implied testifier” backstage strategically shaping what we hear. Likewise, in film, we are guided to make inferences about the narration we encounter, regardless of whether the information is re- counted by characters or presented by the overall organization of the film. Needless to say, those inferences may fit together smoothly or they may contradict one another, just as in life. Naturally, in narratives, the fit or the contradictions are largely created by the makers, in order to take us through a particular experience, whereas life has no such artificer in the wings. In any event, we don’t need to personify an agent hovering over the text that is transmitting the truth of the situation. If the implied author is the set of overarching principles of design governing the film, we can simply talk about those principles themselves, even, or especially, when they create problems of unreliability for the spectator. Further, it may be that the communication model creates the very need for an implied author. Chatman argues that because Anne Frank never intended her diary to be seen, the real Anne Frank can’t be speaking to us. Still, “we read the diary as if it addresses us,” so “it can only be the implied author of the Diary who addresses us.”86 Three Dimensions of Film Narrative 51 By this line of argument, every diary that’s read by somebody other than the diarist has an implied author, whereas those that aren’t so read don’t. The implied author becomes the reader’s projection, not the author’s creation. So why do we consider it part of a process of communication at all? Is a diary in fact an instance of communication? It seems to me that reading a diary is best understood on the model of overhearing someone talking to himself. Because we aren’t the addressee, we don’t need to posit an agency that is shaping the monologue for our (or an implied hearer’s) uptake. Several types of solitary writing—grocery lists, Post-It notes to yourself—don’t presuppose implied agents of this sort. When we find self-ad- dressed writing, we just hear or see the words and draw our inferences accordingly, under whatever norms we think relevant. Likewise with cinema: A film is made so as to elicit inferential elaboration. Invoking the implied author would seem to add nothing to our recognition of the principles under which the film operates. Are the concepts of a cinematic narrator and implied author logically necessary for narratology? Are they pragmatically helpful in narrative analysis? My answers, all in the negative, point toward a distinction I urged earlier. It’s useful to distinguish, however roughly, between theoretical poetics, which aims to understand the conditions of cine- matic representation on a broad canvas, and historical poetics, an empirical inquiry into particular ways of making. I’d reiterate that we should build the former as inductively as we can, tracing out commonalities among traditions that we study in detail. When we try to be purely deductive, we tend to start with intuitively salient models, like that of literary communication, with its nested senders and receivers. The risk is assuming that models that are salient for us apply universally, to all stories in all media. We may also miss the fact that narratives, created by people for other people, need not be built out of principles that are logically consistent. The promiscuity of narrative con- struction reflects the quick and dirty reasoning characteristic of minds attuned to social, not ontological, meanings. 1. The example, along with Figs. 3.1–3.2, comes from Caroline Furness James, String Figures and How to Make Them: A Study of Cat’s Cradle in Many Lands (1906; reprint, New York: Dover, 1962), 16–20. 2. For a discussion, see Howard Morphy, Aboriginal Art (London: Phaidon, 1998), 111–21. 3. See Timothy D. Wilson, Strangers to Ourselves (Cambridge, Mass.: Harvard Univer- sity Press, 2002), 167–73. 4. August William Schlegel, A Course of Lectures on Dramatic Art and Literature, trans. John Black (London: Bohn, 1846), 362. 5. Michael appears in about 49% of the running time, and Vito in just under 28% of it. These totals include scenes in which they appear together, but there are surprisingly few of those. If we simply split the difference and assign half of their two-handed scenes to one side or the other, we come up with about 48% for Michael and 27% for Vito. One man or the other is present for 75% of the film’s duration. 6. Murray Smith, Engaging Characters: Fiction, Emotion, and the Cinema (Oxford: Oxford University Press, 1995); Ed Tan, Emotion and the Structure of Narrative Film: Film as 52 Poetics of Cinema an Emotion Machine (Teaneck, N.J.: Erlbaum, 1996); Torben Grodal, Moving Pictures: A New Theory of Film Genres, Feelings, and Cognition (Oxford: Oxford University Press, 1999); Carl Plantinga and Greg Smith, eds., Passionate Views: Film, Cognition and Emotion (Baltimore: Johns Hopkins University Press, 1999); and Greg M. Smith, Film Structure and the Emotion System (Cambridge: Cambridge University Press, 2003). 7. See Stephen Halliwell, The Poetics of Aristotle: Translation and Commentary (Chapel Hill: University of North Carolina Press, 1987), 39. Because the plot is also an abstraction, an event-structure that can be manifested in different media, Aristotle further distinguishes between the plot and the play, the drama text that presents it. See Malcolm Heath, Unity in Greek Poetics (Oxford: Clarendon, 1989), 40. 8. Quoted in Veronique de Turenne, “Shopgirl,” Variety, December 18, 2005. 9. See Meir Sternberg, Expositional Modes and Temporal Ordering in Fiction (Baltimore: Johns Hopkins University Press, 1978), 45–55; and Meir Sternberg, “Telling in Time (II): Chronology, Teleology, Narrativity,” Poetics Today 13, no. 3 (Fall 1992): 524–38. 10. Sternberg, Expositional Modes, 129–58. 11. See Smith, Engaging Characters, chs. 3–6; and Murray Smith, “Gangsters, Cannibals, Aes- thetes, or Apparently Perverse Allegiances,” in Plantinga and Smith, Passionate Views, 217–38. 12. David Bordwell, Narration in the Fiction Film (Madison: University of Wisconsin Press, 1985). A portion of this argument is presented in David Bordwell and Kristin Thompson, Film Art: An Introduction, 10th ed. (New York: McGraw-Hill, 2013), ch. 3. 13. Meir Sternberg, The Poetics of Biblical Narrative: Ideological Literature and the Drama of Reading (Bloomington: Indiana University Press, 1985), 172–73. 14. Jane Smiley, 13 Ways of Looking at the Novel (New York: Knopf, 2005), 232–33. 15. Seymour Chatman, Story and Discourse: Narrative Structure in Fiction and Film (Ithaca, N.Y.: Cornell University Press, 1978), 93–95. 16. The classic work is Georges Polti, The Thirty-Six Dramatic Situations, trans. Lucille Ray (Eng. orig., 1906; reprint, Boston: The Writer, 1988). For a recent effort, see Christopher Booker, The Seven Basic Plots: Why We Tell Stories (New York: Continuum, 2004). 17. E. M. Forster, Aspects of the Novel (New York: Harcourt, Brace, 1927), 153. 18. On Elizabethan parallel plotting, see Richard Levin, The Multiple Plot in English Renais- sance Drama (Chicago: University of Chicago Press, 1971); and Thomas G. Pavel, The Poetics of Plot: The Case of English Renaissance Drama (Minneapolis: University of Min- nesota Press, 1985). Carol Clover exposes stranding in the Nordic sagas in The Medieval Saga (Ithaca, N.Y.: Cornell University Press, 1982). 19. Urban Gad, Filmen: Dens Midler og Mal [Film: Its Means and Goals] (Copenhagen: Gylden- dal, 1919), 247–51. 20. See Semyon Timoshenko, Chto dolzhen znat’ kino-rezhisser [What a Film Director Must Know] (Leningrad: Teakinopechat, 1929), 21–23. 21. Anonymous, “Reel Length From the Exchange and Projection Viewpoints: Discussion,” International Projectionist 7, no. 1 (June 1934): 13–15, 21–23. 22. See my Planet Hong Kong: Popular Cinema and the Art of Entertainment (Cambridge, Mass.: Harvard University Press, 2000), 180–82, 275, 283. Three Dimensions of Film Narrative 55 53. I argue this in more detail in my Ozu and the Poetics of Cinema (Princeton, N.J.: Prince- ton University Press, 1988), 318–19. 54. Interestingly, the Claire Dolan entry on the Internet Movie Database “classicizes” the film: “A high-priced call girl, shocked by her mother’s death, decides to get out of the business and have a baby. The steps that she takes to free herself from her pimp and find a father for the baby are the central story of this movie.” 55. In Scenarios modèles, 52–55, Francis Vanoye provides a brief but intriguing discussion of “modern” models of character, including the “problematic” character in crisis, the opaque character as seen in Duras or Bresson, and the noncharacter, the mask or marionette, as in Last Year at Marienbad (1961) or some films of Peter Greenaway. 56. Nicholas Kazan, quoted in Jurgen Wolff and Kerry Cox, Top Secrets: Screenwriting (Los Angeles: Lone Eagle, 1993), 134. 57. Lajos Egri, The Art of Dramatic Writing (1942; reprint, New York: Simon & Schuster, 1960), 61. 58. Ibid., 60–61. 59. See Bordwell, The Way Hollywood Tells It, 51–71. 60. Bordwell, Narration in the Fiction Film, 62. 61. André Gaudreault and François Jost, Le récit cinematographique (Paris: Nathan, 1990), 24–27; Albert Laffay, Logique du cinéma: Création et spectacle (Paris: Masson, 1964), 62– 85; and Tom Gunning, D. W. Griffith and the Origins of American Narrative Film: The Early Years at Biograph (Urbana: University of Illinois Press, 1991), 13–25. 62. Meir Sternberg has argued that even within quoted passages, we can detect the presence of the external narrator. See his “Proteus in Quotation-Land: Mimesis and the Forms of Reported Discourse,” Poetics Today 24, no. 3 (Fall 2003): 107–56. 63. Quoted in Rudy Behlmer, ed., Memo From Darryl F. Zanuck: The Golden Years at Twentieth Century Fox (New York: Grove, 1995), 190. 64. Bordwell, Narration in the Fiction Film, 61. 65. There may be an important point about literary conventions here. It seems that insofar as a literary narrative relies on a communication situation, the character narrator is pre- sumed to have one-point access to information. That is, she can tell only what she knows (or supposes, or guesses, or imagines, or divines). As in real life, all information we have comes filtered through that character’s range of knowledge. It does seem, however, that recounting situations in narratives in other media don’t follow the same convention. A flashback in a play or a comic book would seem to have the same flexibility that we find in cinema. To consider how some first-person narrators, such as those in Conrad and Fitzgerald, report things that exceed their knowledge would take me afield here, but these possibilities do suggest that one-point access isn’t an inviolable convention. 66. Bordwell, Narration in the Fiction Film, 62. 67. Chatman, Story and Discourse, 151. Figure 3.6 is adapted from this passage. 68. Seymour Chatman, Coming to Terms: The Rhetoric of Narrative in Fiction and Film (Ithaca, N.Y.: Cornell University Press, 1990), 115–16. 69. Ibid., 127. 70. Ibid., 115. 71. Ibid., 134. 56 Poetics of Cinema 72. Ibid., 134. 73. Ibid., 136. 74. Chatman, Story and Discourse, 253–61. 75. Marie-Laure Ryan, “Introduction,” in Narrative Across Media: The Languages of Storytell- ing, ed. Marie-Laure Ryan (Lincoln: University of Nebraska Press, 2004), 15. 76. Chatman, Coming to Terms, 133. 77. See for example Gregory Currie, “Unreliability Refigured: Narrative in Literature and Film,” The Journal of Aesthetics and Art Criticism 53, no. 1 (Winter 1995): 19–23; George M. Wilson, “Le Grand Imagier Steps Out: The Primitive Basis of Film Narration,” Philosophical Topics 25, no. 1 (Spring 1997): 295–318; Paisley Livingston, “Narrative,” in The Routledge Companion to Aesthetics, ed. Berys Gaut and Dominic McIver Lopes (Lon- don: Routledge, 2001), 278–81; and Andrew Kania, “Against the Ubiquity of Fictional Narrators,” The Journal of Aesthetics and Art Criticism 63, no. 1 (Winter 2005): 47–54. For a critique of the communication model and an argument that even third-person literary narration may not proceed from a narrator, see John Morreall, “The Myth of the Omni- scient Narrator,” The Journal of Aesthetics and Art Criticism 52, no. 4 (Fall 1994): 429–35. I’m grateful to Paisley Livingston for several helpful discussions on this topic and for sharing some of these sources with me. 78. Ryan, “Introduction,” 15. 79. See, for example, Gunning, D. W. Griffith, 24; Chatman, Coming to Terms, 127–30; and J. Lothe, Narrative in Fiction and Film: An Introduction (Oxford: Oxford University Press, 2000), 28. 80. Ibid., 133. 81. Ibid., 74. 82. Ibid., 75, 82, 86. 83. Ibid., 81. 84. See Wayne Booth, The Rhetoric of Fiction (Chicago: University of Chicago Press, 1961), 70–77. 85. Chatman, Coming to Terms, 75. 86. Ibid., 91.