© Columbia University Press
A Theory of Narrative
Paper, 392 pages, 40 illus.
ISBN: 978-0-231-14429-2
$27.50
/ £16.00
July, 2008
Cloth, 392 pages, 40 illus.
ISBN: 978-0-231-14428-5
$79.50
/ £47.00
View this excerpt in pdf format | Copyright information
Excerpt from the chapter "What is Narrative?"
THE NATURE OF NARRATIVE REVISITED
Existing definitions of narrative share several shortcomings. Most are based on a limited corpus, stress a single characteristic, and take one type of narrative as representative of the entire class. To avoid these problems, I have respected the following precepts:
1. Useful definitions must be based on a willfully diverse corpus. Examples produced during a single historical period (e.g., the recent past) or by a limited group (e.g., culturally acceptable writers) do not constitute an adequate sample. It is up to the theorist to guard against a corpus that is artificially limited by a combination of personal taste, ease of access, historical trends, or hidden cultural preferences. This has long been a serious problem in defining narrative, from Aristotle’s dependence on the Odyssey and Oedipus to Lubbock’s exclusive attention to the work of Henry James, Barthes’s regular reference to Honoré de Balzac, and other theorists’ tendency to base their notions of narrative on the novel alone. Virtually the sole narrative theorists to actively diversify their corpus have been Milhail Bakhtin (1968, 1981) and the team of Robert Scholes and Robert Kellogg (1966) and Scholes, Kellogg, and James Phelan (2006).
2. Good definitions maintain a balance between restriction and inclusion. Definitions must be restrictive, clearly excluding inappropriate examples, in order to target accurately the specific phenomena under investigation. At the same time, self-consciously inclusive definitions avoid unreasonable exclusion of potentially useful examples. The history of narrative definition has so often been marked by a desire for exclusion that we are currently in particular need of inclusiveness (though the inclusive efforts of Smith 1981, O’Neill 1994, Abbott 2002, and Prince 2003 deserve recognition).
3. Multiple-operator definitions are preferable to definitions based on a single criterion, because the latter rarely do justice to the phenomenon under study. While single-criterion systems allow for easy classification, they match up poorly with the complexity of a category like narrative and are of little use in the analytical process.
4. Subcategories must not be confused with the category itself (e.g., men for humans). In particular, a common approach to narrative organization (e.g., a beginning-middle-end structure) must not be taken as part of narrative’s very definition. Because they disenfranchise no potential subcategories, “equal- opportunity” definitions are to be preferred.
Respecting these four precepts, the definition developed below is organized into three basic areas. Narrative material encompasses the minimal textual characteristics necessary to produce narrative. Narrational activity involves the presence of a narrating instance capable of presenting and organizing the narrative material. Narrative drive designates a reading practice required for narrative material and narrational activity to surface in the interpretive process.
Narrative Material
Most definitions of narrative assume that action alone is sufficient to define narrative. Two major innovations characterize the present definition: recognition that narrative material is insufficient by itself to defi ne narrative, and insistence on character as a defining characteristic on a par with action.
ACTION
Narratives require action. Without action, we may have portraiture, catalogue, or nature morte, but not narrative. A telephone, a car, and a detective do not produce a narrative until they are set in motion by a series of actions: the telephone rings, the detective answers then jumps in the car. Seen in this way, actions are like verbs creating contact among the separate substantives populating a narrative.
Nearly every definition of narrative mentions action, though several recent theorists have preferred the term “event” (Toolen 1988:7, Miller 1998:46, Abbott 2002:12, Prince 2003:58). Rarely, however, do theorists recognize the very real complexity of this apparently simple requirement. Just what counts as “action” for the purpose of defining narrative? In scientific terms, there is no doubt that sunshine requires the action of heat-producing explosions. Why is it, then, that the sun’s shining provides no more than atmosphere in one text (thus failing to constitute an action and contribute to that text’s narrativity), but in another (e.g., a story of the sun’s decision to continue shining in spite of humanity’s wrongdoing) it counts as the action necessary to narrative? Questions like this imply that narrative cannot be defined by action alone, for even the recognition of action as such (i.e., acknowledgment that a par tic u lar action is sufficiently salient to count as action for purposes of narrative definition) cannot be achieved inde pendently of other factors.
Actions appear so central to narrative that theorists often summarize narratives in terms of their actions alone. Narratives are typically referred to as representing a rise and fall, loss and recovery, or desire and acquisition. Gérard Genette (1980) first describes narrative as the expansion of full subject/verb sentences but then proceeds to organize his study according to the verbal aspects of his examples alone. Branigan summarizes the lady/tiger limerick as smile/ride/swallow/return/smile. This use of verb-based cover terms, as William Hendricks (1973) notes, effectively disguises the aspects of narrative that depend on something other than action.
CHARACTER
In definitions of narrative, virtually no attention has been paid to the notion of character. Aristotle goes so far as to proclaim that character is simply unnecessary. Barthes and the structuralists relegate character to a dependent position. Anglo-American theorists have been similarly effective in excluding character from definitions of narrative. Though several volumes have been devoted to the study of characters in novels and films (Harvey 1965, Price 1983, The Filmic Character 1997), only exceptionally has character been given substantial play in constructing definitions of narrative (Scholes and Kellogg 1966, Chatman 1980, Rimmon-Kenan 2002, and Scholes, Kellogg, and Phelan 2006). No doubt the main reason for this oversight is the assumption that actions necessarily imply actors; designation of the actor as a defining characteristic would thus be redundant. If by definition an actor becomes an actor only by performing an action, then the action itself is primary and mention of the actor superfluous.
This logic is doubly flawed. First, as we know from our experience with the theater and cinema, actors are in fact neither engendered solely by their actions nor fully defined by them. Actors are also repositories of potential actions, existing independently of any specific completed action. The names of a film’s starring actors in the opening credits create in us expectations regarding actions not yet realized, thereby convincing us that their characters exist inde pendently of their actions. Second, this logic confuses the notion of “actor”—the one who performs an action—with the very different notion of “character.” What is a character? And how is a character different from the actor who performs the actions necessary to a narrative? The answer to these questions is best approached as a function of the medium serving as a particular narrative’s vehicle. In theater, human bodies serve as the vehicle for telling stories. As long as Sarah Bernhardt remains at the level of theater’s bodily vehicle, she remains Sarah Bernhardt. She may laugh and cry or rant and rave, but she remains an actress and not a character. When she speaks, it is the body of Sarah Bernhardt that speaks; the words and actions remain on the level of the narrative’s vehicle. Once she becomes Camille (in Alexandre Dumas’s play La Dame aux Camélias), however, Sarah Bernhardt is no longer the agent of her actions. Now the agent is another entity, separate from the actress and thus divorced from the bodily vehicle on which theater depends. Just as Camille is one level removed from Sarah Bernhardt, and thus from the “language” that serves as the vehicle for theater, so all characters exist at a level different from that of their narrative’s vehicle. When we view a comic strip, we are looking at a series of lines until the process of representation turns those lines into represented forms. Without going through this process, we can never perceive a character but only a drawing. Reading a novel we first encounter words, not characters ( just as theater offers actors before characters). The letters “D-a-v-i-d-C-o-p-p-e-r-f-i-e-l-d” do not at first designate a character—an entity existing at a level different from that of language itself. Certain events must take place before we will recognize “David Copperfield” as a character rather than as a simple name.
The key to this process may be found in the flights of fancy attributed by Jorge Luis Borges to his character Funes the Memorius. Constantly challenged by the classifications on which abstract thought depends, Funes is particularly troubled by the fact that the same name is used to designate phenomena that to him seem quite distinct: “It bothered him that the dog at three fourteen (seen from the side) should have the same name as the dog at three fifteen (seen from the front)” (1964:65). What Funes describes with such naïve accuracy is precisely the process that leads to readers’ and viewers’ recognition of a character. We look at the comic strip and see one image, then an instant later see a second image. The two images are demonstrably different, just like the views of the dog from the side and the front. Yet for us to construe this comic strip as narrative, we must give the same character name to both images. In a very real sense this is (as Borges cleverly implies) the way in which all knowledge is created: in order for phenomena to be memorized and turned into knowledge, we must renounce nominalism in favor of the abstract categories of realism. Narrative knowledge depends on this level of abstraction—we must abandon the media used to express and communicate the narrative in favor of a constructed, abstract level where the figure in frame one (seen from the side) and the figure in frame two (seen from the front) and the name used in frame three (“Goofy”) are all recognized as referring to the same character. None of these taken alone actually is the character, but all three refer to the character, in the process erasing the primary graphic and linguistic levels. The development of characters thus participates in the creation of a “diegesis,” a posited level inde pendent of the textual vehicle.
The example provided by Funes the Memorius suggests in the simplest possible manner the extent to which the notion of character depends on multiple reiterations of the character. As long as there is only one picture of the dog, then Funes has no problem, precisely because he still identifi es the graphic language and the dog. Only when the second image appears does Funes’s frustration arise, for now he must contend with a contradiction between the physical level (the level of the vehicle), where he sees two different pictures, and the abstract level (the diegetic level), where there is only one character. Only with the introduction of the second example does this contradiction—and thus the character—appear. Far from being a synonym of “actor” (the one who does the acting) or “subject” (the one who sets the verb in motion), the term “character” is a technical term properly applied only to a limited range of actor-subjects. Among narrative theorists, only Seymour Chatman (1980:107ff ), Gerald Prince (1982:71ff), and Mieke Bal (1985:25) have recognized the complexity and specificity of the notion of character.
Just as two distinct views are necessary to establish the existence of a character—an entity existing at a level different from the text’s vehicle—so are two views necessary to determine what counts as an action for purposes of narrative definition. The term “action,” when used as a criterion for narrative, is not simply an everyday term designating activity of all types but operates instead as a technical term dependent on a network of textual relationships. Thus sunshine usually does not count as a narrative-defining “action,” but the sun’s continuing to shine in spite of human shortcomings might very well qualify. Why should these two actions have different effects on a text’s narrativity? What are the real differences hidden in this seat-of-the-pants distinction? In the first case, both the sun and the act of shining exist in a vacuum, whereas in the second case, the sun has been established as a character and the action of shining is set in relation to the possibility of the sun’s not shining. Just as establishment of an actor as a character requires a minimum of two connected views of the actor, so qualification of an action as narrative-defining requires that an activity be valorized by a related activity—a transformed version of the activity that (a) increases the saliency of both activities, (b) identifies the activity as a narrative variable, and thus (c) turns the simple activity into a narrative-defi ning action.
If the sun shines when it does by the simple continued revolution of heavenly bodies, then narrative is absent, but if the shining of the sun is set in relation to the non-shining of the sun, or the shining of the moon, or any other related activity capable of drawing attention to the process of the sun shining, then we recognize the kind of action that is necessary to narrative.
The existence of narrative depends on the simultaneous and coordinated presence of action and character. Narratives are not made of characters here and actions there but of characters acting. Indeed, it is the very fact that a character acts that permits us to recognize successive images as representing the same character. Conversely, it is through association with a character that simple activities become narrative-defining actions. For analytical purposes, action and character may reasonably be separated, but it is only as two angles on the same process that we encounter and experience action and character. This leads us to a second set of defining characteristics, precisely dependent on the way in which we encounter action and character.
Narrational Activity
It is often said that the world is full of stories all happening simultaneously. Yet the world is not itself a story. Neither the city of Brussels nor the battle of Waterloo is a story in and of itself. Yet both contain the material necessary for narrative. How is it that a city or a battle can be transformed into a narrative? Two processes, both narrational in character, are necessary to that transformation.
FOLLOWING
One of the most characteristic aspects of narrative involves the reader’s sense of following a character from action to action and scene to scene. A bird’s-eye view of a city, or a detailed description of a battle, no matter how many individual actors and activities are visible, will provide at most the material for narrative. Not until the narrator begins to follow a particular character will the text be recognizable as narrative. Or, to put it more accurately, not until a particular character is followed will we sense the activity of a narrator, thereby defining the text as narrative. Stendhal’s The Charterhouse of Parma follows Fabrice del Dongo across the battlefields of Waterloo; in Les Misérables Victor Hugo animates the same battle by alternately following the Olympian decision-maker Napoleon and the pusillanimous scavenger Thénardier. It is this process of following that turns these accounts into narrative.
Imagine a long shot of the Grand Central Station waiting room. The camera doesn’t move as it captures hundreds of hurried commuters scurrying about. Suddenly the camera tracks in to focus on a single, nervous individual looking at the clock. As she moves across the floor, the camera pans to follow her. At first the space is neutral, unvectored, narratively flat. When the camera focuses on a single individual and follows her, however, we recognize that we are being cued to read this scene as narrative. Followed, the character serves as a vector defining the space before us. As she becomes a character, thanks to the process of following, so her activities—previously indistinguishable from all the other activities visible in the image—turn into narrative-defining actions. Constitutive of narrative, the process of following thus simultaneously activates both character and narrator.
The process of following brings two different levels into clearer focus. Until the camera begins to follow a particular person, turning her into a character and foregrounding her actions, the process of narration remains entirely transparent, with no apparent separation between narrational and diegetic levels. With the introduction of following, concentrating attention on a particular character, we paradoxically also sense the existence of a narrational instance—some one, some thing, some system deciding who should be followed. The process of following thus simultaneously highlights character and narrator, diegesis and narration. It is precisely this simultaneous emphasis on two different levels that constitutes narrative. Without following, we have only an unvectored chaos, capable of producing narrative but not yet doing so; with following, we not only concentrate attention on a character and the character’s actions, thus satisfying the first set of conditions for the existence of narrative, but we also implicitly reveal the existence of a second, narrational level. The author authors the words (sounds, images, etc.) of the text, but these are not where the narrative lies. The narrator narrates the diegetic level, which is where the narrative is located. All texts have a primary vehicle (or vehicles); narrative in addition constructs a diegesis beyond the primary vehicle and offers a narrator as a necessary filter between reader and diegesis. The process of following is the initial and primary evidence of that narrator’s activity. (For a review of “no-narrator” approaches to narrative definition, see Herman, Jahn, and Ryan 2005:396–97.)
FRAMING
A substantial contradiction exists between two familiar but quite different ways of using the term “narrative.” On the one hand, the term is regularly used to designate a type of material that is easily recognizable, even in small chunks. Used in this way, the term aptly describes our categorization activity as we browse through a stack of books, sort a bunch of films, or surf a series of television channels. Sampling only a portion of each text, we readily determine whether or not it is a narrative text. Once we establish—and it happens very rapidly indeed—that the text features characters, action, and following, we quickly identify it as narrative. Yet this “snapshot” approach contrasts strongly with narrative theoreticians’ most common use of the term. From Aristotle to Branigan, theorists insist that narratives always have a beginning, middle, and end; that narrative endings must echo narrative beginnings in a significant manner; and that this arrangement is fundamental to the very notion of narrative. Texts lacking this structure are thus commonly not accepted as narratives. The contradiction here is quite clear. If it is necessary to observe the entirety of a text in order to acknowledge it as narrative, then what is it that channel surfers are recognizing? How can narrative be at the same time something that is identifiable piecemeal and something requiring the experience of complete texts?
The only way out of this quandary is to recognize two standard uses of the term “narrative,” appropriately coexisting but easily distinguished. The first case, where narrative is discernible from small textual samples, involves recognition of what we might call “some” narrative, whereas the second case is concerned with the definition of “a” narrative. Daytime television soap operas offer a good example of “some” narrative. No matter when we tune in, we are rapidly convinced that we are dealing with a narrative text; yet no matter how long we watch, we never reach closure. Unlike most novels and films, soaps are all middle; we nearly always confront them in medias res and leave them before a satisfactory conclusion is reached. Yet we never doubt their narrativity. At every point we acknowledge that they are narrative in nature; that is, we recognize in them “some” narrative.
How do the texts touted as narrative by theoreticians differ from soaps? That is, how does “a” narrative differ from “some” narrative? The main difference at work here is the process of framing. In deciding whether a text is narrative, we are usually concerned only to know whether it contains characters, action, and following. But when theorists concentrate on a common narrative pattern, they are analyzing questions of framing, not content. Just as a shot of the crowds in Grand Central Station becomes narrative (in the sense of “some” narrative) only when a character is followed, thus revealing narrational activity, so a series of events becomes narrative (in the sense of “a” narrative) only when those events are framed, thus revealing yet another type of narrational activity. By itself, daily life cannot be said to constitute narratives, however much narrative material it may provide. But when a naturalist novelist cuts daily life into slices, thus delimiting and framing it, the narratives implicit in daily life may be revealed.
To recognize “some” narrative, all we need are narrative material and following, but “a” narrative is recognizable only when it has been fully framed. In one sense, then, it is the very process of framing that gives a text its beginning and end. Without framing, texts are all middle; by the very act of framing, texts gain a beginning and end. Note that this definition of framing says nothing about necessary correspondences between beginnings and endings. Framing delimits the text but does not guarantee any particular internal textual organization—thus avoiding the Aristotelian pitfall of making one among many possible types of textual organization part of narrative’s very definition.
Narrative Drive
The preceding descriptions of minimal narrative criteria assume that narrative material and narrational activity are always located within the text. Upon careful inspection, however, this assumption proves poorly justifi ed. Let us return for a moment to our Grand Central Station crowd scene, the one that becomes narrative only when it is invested with narrational activity. When the narrational activity is provided by a camera tracking a particular character, then we can confi dently affirm that the narrational activity is intratextual. But suppose a spectator recognizes on the screen a star actor among all the other unrecognized faces. Or imagine that a spectator sees a good friend in the crowd. Or perhaps a graduate student writing a dissertation on facial types notices someone with the type of face she is currently treating. In all these cases there is every likelihood that the spectator will follow one individual throughout the scene, just as if the camera had chosen to focus on that character. In other words, it is entirely imaginable that the narrator responsible for narrational activity can be the spectator or reader (supported by powerful social institutions like the film industry, the family, or the educational system). As Michael J. Toolen puts it, narrative is not just a sequence of non-randomly connected events but a “perceived sequence of non-randomly connected events” (1988:7; my italics), a position shared by Luc Herman and Bart Vervaeck (2005:52–54).
A similar situation obtains regarding every aspect of narrative material and narrational activity. Virtually any situation can be invested with character/action characteristics, as well as with following and framing. Indeed, this is what anthropologists, archaeologists, historians, doctors, car mechanics, and myriad others are doing as they strive to make sense of social customs, ancient stones, and physical symptoms. Identifying the characters and actions meaningful to their specific context, these “spectators” perform the narrational function of following individual characters and framing separate narratives. For one, this might mean charting the destruction of a civilization by farmers’ inability to adjust to new climatic conditions; for another, it might mean following the ramifications of foreign matter introduced into the gas tank. For these professionals, the very process of exercising their profession involves creating a narrative out of details that are not necessarily in and of themselves narrative.
We may appropriately term this tendency to read texts as narratives “narrative drive.” Narrative drive can derive from many sources: personal interests, professional mandates, or social expectations. While it may be conditioned by textual characteristics, it can never be wholly dependent on elements that are internal to the text. In other words, without narrative drive on the part of the reader, texts are not read as narrative. Conversely, though narrative drive usually arises in response to specific textual factors, a strong narrative drive can generate the very factors necessary for recognition of narrative. Imagine a line, a simple line drawn across the page. One would be hard put to claim that this line satisfies any of the requirements for narrativity outlined above. Yet the reader who is strongly driven to treat this text as narrative can do so with little diffi culty. Somewhere, the line will seem a little bit thinner or a tone fainter. Recognizing the continuity of the line through thick and thin, we turn the line into a character named “the line” (like Borges’s dog, recognizably the same at different points in spite of demonstrable differences), and the thinning into an event worthy of recognition as an action. Thus our simple straight line turns into a narrative in which “the line” first “thins,” then “thickens”—character/action units that characterize narrative material. By following the line and delimiting the text to the surroundings of the line’s thinning and rethickening we have implicitly provided the narrative activity necessary for narrative. A strong case of narrative drive has produced a narrative where none was apparent. On virtually any set of givens, this process may be repeated.
Paradoxically, there exists similar evidence for the converse of this principle. However narrative a text may appear to some, a lack of narrative drive can always threaten the text’s narrativity. For some, the action and dialogue of a love scene may disappear entirely in favor of contemplating the portrait of a beautiful woman or an attactive man. A battle scene may lose its narrative focus entirely when the members of a Civil War re-creation society watch the scene, attentive not to its narrative stakes but to making a catalogue of the weapon types used by Union and Confederate soldiers. My favorite example of the lack of narrative drive appears in the Blackhawk Films version of D.W. Griffith’s classic thriller short, The Lonedale Operator. Instead of stressing the growing love affair between the train engineer and the telegraph operator, Blackhawk’s version provides an insert describing the locomotive visible on the screen. Rather than attend to the burglars’ attack, another insert explains that the locomotive is pointing in the wrong direction. Initially distributed to railroad fans, Blackhawk’s infamous version of The Lonedale Operator substitutes what we might whimsically term “locomotive drive” for “narrative drive.” Far from constituting an unusual situation, this case actually exemplifi es the extremely common circumstance where a text read by some as narrative is harvested by others for entirely different purposes. Depending on the “reading formation” espoused by the reader at the moment of a particular reading, a text can move in and out of the narrative category, whether or not it displays narrative material and narrational activity.
The one thing we can claim with great assurance is that whenever narrative drive causes a text to be read as narrative, the reading will foreground narrative material and narrational activity. Though these interconnected levels may reasonably be analyzed separately, they have immediate impact on each other. The very process of following convinces us of the existence of a character; similarly, the process of framing lends such importance to certain events that they readily appear as the actions required by narrative. Whether through recognition of conspicuous textual elements or by dredging up entirely submerged textual factors, narrative drive always produces salient narrative material and narrational activity. While narratives may be read in many ways, the reading of a narrative as narrative always involves the presence of narrative material, the implementation of narrational activity, and the deployment of narrative drive.
...
COPYRIGHT NOTICE: Published by Columbia University Press and copyrighted © 2008 by Columbia University Press. All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher, except for reading and browsing via the World Wide Web. Users are not permitted to mount this file on any network servers. For more information, please e-mail us or visit the permissions page on our Web site.