2 Structuralism

Contemporary narratology finds its roots in the work of the French structuralists. Issue number 8 of the journal Communications usually figures as the official starting point of the discipline. This issue, which came out in 1966, contained nine articles with proposals for concepts and methods that could be used to study narrative texts. Some of these articles have acquired stature as classics. This certainly holds for the plot analysis proposed by Roland Barthes, which we will discuss shortly, but other contributions, by A. J. Greimas, Claude Bremond, Umberto Eco, Gérard Genette, and Tzvetan Todorov, have remained important as well.1 In his Grammaire du Décaméron (Grammar of the Decameron) published three years later, Todorov introduced the term “narratology”: “We wish to develop a theory of narration here [ . . . ]. As a result, this book does not so much belong to literary studies as to a discipline that does not yet exist, let us say narratology, the science of narrative.”2 The French structuralists recognize the Russian formalists as precursors of this scientific discipline. Vladimir Propp’s analysis of fairy tales can be seen as an embryonic example of structuralist narratology.3

The structuralist distinction between the text as it appears and its underlying patterns also stems from the formalists. As we will see, these Russian literary theorists made a distinction between the abstract chronology of events and their concrete sequence in a narrative text where they often do not proceed in chronological order. Structuralism is characterized by the gap between surface and deep levels. In the collection Qu’est-ce que le structuralisme? (What is structuralism?) Todorov explains that structuralism does not deal with the literary text as it presents itself to the reader but rather with a deep, abstract structure.4 The science of narratology, rather than investigating the surface, should study that which is fundamental to narrative.

This approach has led to the division of the narrative text into three levels. Genette describes the surface level with the term narration—the same in the French original and in our English translation—which comes down to the formulation of the story.5 Narration refers to the concrete and directly visible way in which a story is told. Word choice, sentence length, and narrating agent are all elements that belong to this level. Genette situates the second level slightly under the surface and calls it récit in French, which we will translate as narrative in English. Narrative is concerned with the story as it plays out in the text. Whereas linguistic formulation was central to narration, the organization of narrative elements is central to narrative. Narrative does not concern the act of narration but rather the way in which the events and characters of the story are offered to the reader. For instance, a novel starts with the death of the male protagonist and then looks back to his first marriage from the vantage point of his son, after which it looks forward to the end of that marriage from the perspective of his second wife. So the level of narrative has to do with organizational principles such as (a)chronology and perspective.

Genette’s final and deepest level is histoire, which we translate as story, not least because its most concrete form coincides with E. M. Forster’s concept of story—the chronological sequence of events—as we have presented it in chapter 1 of this handbook. This level is not readily available to the reader. Instead it amounts to an abstract construct. On this level, narrative elements are reduced to a chronological series. The story of the example above would start with mention of the man’s first marriage, then the end of that marriage, and finally the man’s death. Here the protagonist does not appear as a concrete character but as a role in an abstract system. The setting is reduced on this level to abstract characteristics such as high or low and light or dark.

There has been endless discussion about the advantages and disadvantages of such an approach. We limit ourselves to a few remarks that will be useful for the rest of this book. First, structuralist narratology deals with the concrete text only via an abstract detour, notably the construction of a so-called deep structure that ideally remains so abstract that it consists only of symbolic and formal elements. The narratologist’s ideal was the concept of a distinctive feature in phonology. Such a feature does not have a meaning of its own, but it causes differences of meaning. The contrast between voiced and voiceless is a distinctive feature. For instance, phoneme /b/ is voiced and /p/ is voiceless. In itself the difference does not mean anything, but it does result in the difference between “bath” and “path,” for example. Narratology never reaches such an abstract and exclusively formal level. All the elements structuralists isolate in the story as formal components of deep structure invariably carry meaning that destroys their dreams of an absolute formality.

Second, deep structure in principle has to be as universal as possible, but in practice it differs from structuralist to structuralist. The deep structure proposed by Barthes is different from Todorov’s, but it also differs from those of Bremond and Greimas. In devising a deep structure one can apparently settle for different levels of abstraction. In its least abstract form, the story is the chronological sequence of events, but try to go any further and difficulties abound. Greimas’s semiotic square is far more abstract than Barthes’s narrative grammar. Greimas reduces a narrative text—and sometimes even an entire oeuvre—to four terms he combines in a square, which in its schematic form looks like this:

Greimas calls the relation between terms 1 and 2 one of contraries, such as life versus death. Between term 1 and nonterm 1 (or term 2 and nonterm 2) there is a relation of contradiction. For instance, the combination of life and nonlife is contradictory. The connection between term 1 and nonterm 2 (or between term 2 and nonterm 1) Greimas describes as one of implication. Life implies nondeath, and death implies nonlife.6 One could reduce narrative texts to a number of squares and explain textual development as a combination of these squares and their terms. Here is a stock example: a girl is in love with a poor man but has to marry a rich one whom she hates. This situation implies at least two squares: one (A) in which term 1 is prohibition and term 2, order, and another (B), in which term 1 is love and term 2, hate.

The initial situation combines prohibition (A term 1) with love (B term 1), and order (A term 2) with hate (B term 2). If, as a result of various adventures, the girl is allowed to marry the man she loves after all, the story develops into a combination of nonprohibition (A nonterm 1) with love (B term 1), and of nonorder (A nonterm 2) with hate (B term 2). All stages between beginning and end can be described as a combination of certain terms in certain squares. This approach resembles the reduction of a movie to a set of slides. If one applies such a reduction to an entire oeuvre—which then appears as a single square—a number of essential aspects will inevitably be lost.7

Barthes counters Greimas’s abstract and static square with a dynamic sequence of functions that connect more closely to the order and development of events in the actual text. When later in this book we try to systematize events and actions in our discussion of story, we rely on Barthes’s system because it is more concrete and dynamic than Greimas’s. However, we conclude that discussion with the remark that our choice does not reflect the structuralist treatment of events in the narrative text. There are as many opinions on this subject as there are structuralists.

This variety of deep structures points to a third problem related to this issue: How does one arrive at a particular deep structure? Here too the structuralists fail to come up with an answer. There are no clear discovery procedures.8 Instead of being based on actual texts, deep structures are simply posited.9 There is a considerable risk that texts will be manipulated until they fit the model. In other words, the model sometimes takes precedence over the concrete text, and the theory becomes more concerned with itself than with the literary works it supposedly investigates.

It seems as if structuralist narratology, with its division of narrative texts into three layers, adopts a geological model. Critics of structuralism have called this treatment of the text a form of “spatialization.” They have two basic reproaches with regard to this procedure. To begin with, spatialization underestimates the importance of time. A narrative text unfolds in time not only when it comes to its events but also when it comes to the act of reading, which always takes up a certain amount of time. Structuralist narratology represents a narrative text by way of schemata and drawings that are sometimes reminiscent of geometry. Textual elements are literally and figuratively mapped. The resulting map provides a static and general view that does not do justice to the dynamics of the concrete and sometimes quite chaotic process of reading. Second, the structuralists tend to focus on the lines of separation between the three layers, so gradual transitions are often overlooked. In their search for the differences and gaps between the levels, they fail to appreciate gradations and similarities.10

These points of criticism do not detract from the fact that structuralist narratology is the first large-scale attempt to combine all aspects of narrative analysis in a convenient system. The model resulting from the combination of the three levels allows a reader to link all the central aspects of a narrative text. One can see, for instance, how characterization connects with the setting or the method of narration and the perspective from which events are perceived. This leads to congruities that not only offer better insight into the formal organization of the text but also enable the reader to join content and form. Encompassing structuration is and remains structuralism’s major merit since it clarifies both textual content and form. That is why this particular brand of narratology continues to provide an indispensable legacy even to those readers whose main interest lies in later approaches.

We will elucidate the three levels of the narrative text with reference to three important structuralist narratologists: Gérard Genette, Shlomith Rimmon-Kenan, and Mieke Bal.11 Unfortunately, these critics do not use the same terms for the levels. In order to avoid confusion, we combine all the terms in a figure, whose left column contains the concepts we will favor in this handbook. From our choice one will notice that we no longer use the term “plot,” which we provisionally worked with in chapter 1 when discussing E. M. Forster. While the emphasis of the term “plot” seems to be on what we call narrative, its meaning spills over into our preferred term, “narration.”

1. Story

Just like any deep structure, the story is an abstract construct that the reader has to derive from the concrete text. The diagram below shows that the story consists of three aspects that will be discussed separately but that in fact always intermingle. In the course of the discussion, the terms in the figure will gradually become clear.

1.1. Events

The story is an abstract level. In the first place it refers to the chronological sequence of events that are often no longer shown chronologically in the narrative. The Russian formalists used the term fabula for this chronological sequence (story) and syuzhet for the specific way in which it was presented in the text.12 Thus, the syuzhet covers both narrative and narration in our terminology.

Several proposals have been made to order events on this abstract level. The Russian formalists consider the motif to be the story’s most basic component. So-called bound motifs are indispensable for the fabula, while unbound motifs are far from essential. A murder, for instance, is a bound motif, while the road an assassin travels to shoot a targeted victim may well be considered an unbound motif since it is not crucial. The assassin’s clothing and age are unbound as well. Unbound motifs may be important on the level of the syuzhet, but they are not on the level of the fabula. Digressions about the killer’s clothing, age, and psychology are important for suspense, but they are unimportant for the development of the action. Formalists also distinguish between static and dynamic motifs. The latter change the progress of events, while the former do not. Bound motifs are usually dynamic and unbound motifs most often static, but this is not a rule. In principle the description of a character’s mental makeup constitutes an unbound motif, but this makeup may result in certain actions that give a decisive twist to the course of events. A murder is usually a bound motif, but if it does not bring about any change, it turns out to be static after all.13

Roland Barthes has refined these distinctions in his “Introduction to the Structural Analysis of Narratives.”14 He distinguishes functions and indexes. Functions are elements whose interrelatedness is responsible for the horizontal progress of events, that is, their linear development. The relationship between these elements can take many forms, of which temporality, causality, and opposition are the most common. “X buys a gun” is a function that leads to “X uses the gun to kill Y.” The function “X is in love with Y” is opposed to “Y hates X,” and the tension between these two functions brings about a development in the story. Functions belong to what Roman Jakobson calls a syntagm, which is a horizontal sequence of contiguous elements. Elements are contiguous if their relationship comes down to a direct connection between terms such as “part” and “whole,” “cause” and “effect,” “producer” and “product,” “pole” and “opposite.” The link between buying the gun and using it is one between intention and execution. The killer buys the gun in order to use it. Jakobson calls these contiguous relations metonymical.15

Indexes, on the other hand, do not bring about the horizontal progress of events. They refer to a different plane, which means they function vertically. The many telephones on James Bond’s desk in the stories from the 1960s amount to an index of his importance. As a character he belongs to a different plane from the telephones, but he does get extra weight thanks to these instruments. The connection between the two planes could be called symbolic since the telephones symbolize Bond’s importance. The set of telephones reflects Bond’s busy life. In musical terms one could compare functions to melody and indexes to harmony or counterpoint. Melody derives from the horizontal progress of the score, while counterpoint arises from vertical accumulation.

Barthes distinguishes between two kinds of functions. A cardinal function implies a risk, which means it harbors a choice or a possibility. A question provides a minimal example of this type of function since asking a question leaves open the possibility of ignoring it. When the telephone rings, it may or may not be answered. More generally, almost all crucial events of the story belong to this category. An assassination attempt is a cardinal function and includes the possibility of failure. Narrative suspense largely rests on the risk central to this type of function. The second type of function described by Barthes is the catalyzer, which does not involve a risk but instead merely assures the continuation of what the cardinal function has started. When the telephone rings and Bond is in the room, he can walk to the telephone, let it ring for a few moments, and then pick it up. All the movements between the moment the telephone rings and the moment he picks it up are catalyzers, but the ringing and the answering remain cardinal to the whole sequence.

For the indexes Barthes offers a twofold division as well. A pure index is an element the reader must interpret. Bond’s clothing, his taste, and his preference for certain drinks are all interpreted by the reader as symbols of Bond’s sophistication and virility. Next there is the informative index, which is mainly important for spatiotemporal description and which does not require symbolic interpretation or the solution of a mystery. “It was seven forty-five and it was raining” makes up an informative index. Obviously this type may turn out to be a pure index when for instance the time indication enables the reader to accept or reject the suspect’s alibi.

A structure implies elements in a specific relationship to each other. In the present case the elements are the functions and the indexes, and the relations between them generally fall into three types. The combination of pure and informative indexes is arbitrary. In a self-portrait, for instance, direct information about age and place of birth will appear side by side with suggestive indexes the reader must interpret as indications of character. The relation between cardinal functions and catalyzers is that of implication. The catalyzer completes the cardinal function and is therefore implied by it. Finally, two or more cardinal functions have a relation of mutual implication, since one cannot do without the other. A murder cannot do without a murder weapon and vice versa: the gun is not a murder weapon without the actual murder.

For Barthes the combination of cardinal functions leads to sequences. They are independent units whose opening action has no precursor and whose conclusion has no effect. Seduction is a sequence. It starts with certain tactical moves and then results in success or failure, after which it is over. Sequences can in their turn be combined, for instance through embedding. Sequence A (seduction) can contain a sequence B (such as a story about the heroic deeds of the seducer) that may or may not lead to the successful completion of A. The insertion of B literally causes suspense because it temporarily suspends the continuation of A.

Here is how Barthes systematizes story events: he starts from minimal components such as functions and indexes, proceeds to create minimal relationships between these components (arbitrariness, implication, mutual implication), and so arrives at larger units in the story, such as sequences and their combinations.

It goes without saying that such a system works best with narrative texts in which many things happen. No wonder then that Barthes refers to James Bond. Bond stories contain clear sequences like “the murder,” “the hero is summoned,” “the hero starts an investigation,” and “the hero solves the murder.” In order to illustrate Barthes’s theory, we will analyze the story “From a View to a Kill,” which has a very clear sequence chronology.16 The syuzhet hardly deviates from the fabula because the presentation of events in the text closely approximates the story chronology. Only when Bond is keeping watch over a suspicious location in the woods does a short flashback briefly disturb the chronology. According to the Russian formalists, such a minimal difference between the abstract story and the concrete presentation of events is typical of nonliterary texts or of texts that hardly merit the literary label.

The Bond story starts with a murder sequence. An agent of the British secret service is driving his motorbike on a road through the woods. His mission is to deliver secret documents, but he gets shot by a man who has disguised himself so that he can approach the agent without being suspected. The killer then covers the traces of the murder as best as he can. This sequence can be divided into three cardinal functions: the pursuit, the shot, and the cover-up. There are many indexes. An attentive reader knows from the first few lines that the killer on the motorbike is not a positive character. He has eyes “cold as flint,” “a square grin,” and “big tombstone teeth.” His face has “set into blunt, hard, perhaps Slav lines.”17 A Bond reader will interpret these descriptions as characteristics of a criminal, probably from the Soviet Union. The mention of the time and place of the murder—seven in the morning in May, somewhere near Paris—constitutes an informative index.

By identifying functions and indexes, one can get a better understanding of each sequence. The second sequence, for example, could go under the heading “the hero is summoned.” A beautiful girl snatches Bond away from a sidewalk café and tells him about the murder. In this sequence there are more indexes than cardinal functions because information is more important than action. In the third sequence Bond is briefed at the headquarters of Station F. This briefing rounds off the first sequence, since Bond (as well as the reader) gets to hear what came of the cover-up. The remaining suspense of the first sequence is now totally gone. The briefing is followed by the first investigation at SHAPE (Supreme Headquarters Allied Powers Europe), where staff are less than cooperative. Bond nevertheless manages to formulate a hypothesis, which amounts to the cardinal function of this fourth sequence—“the hero’s first investigation.” Bond tests his hypothesis in “the hero’s second investigation,” sequence number five, in which he observes in the woods the secret hideout where the killer and his two accomplices are holed up. Obviously the hero’s hypothesis proves to be correct. In the sixth sequence Bond devises a plan to apprehend the criminals, which he carries out in the seventh and final sequence. This sequence perfectly mirrors the first. Bond has taken the place of the agent on the motorbike, and now he is being shot at just like the agent in the beginning. He tricks the killer and clears the secret hideout, after which he explores his interest in the beautiful girl.

Such a systematization of events offers a number of advantages. First, it provides an overview of the various links between the sequences. The seventh sequence mirrors the first, the third one concludes the first, and the fifth one confirms the hypothesis of the fourth. The ways in which this story builds suspense thus become clear. This method also enhances the reader’s understanding of numerous details that become more meaningful when seen as a pure or informative index. Elements that might have seemed irrelevant in a superficial reading now acquire the importance they deserve, owing to this more searching analysis. For instance, it cannot be a coincidence that the murder is committed on a road through the woods where the criminals are hiding. As we will see in our discussion of the setting, criminals are constantly associated with nature, whereas the hero appears affiliated with culture, the city, and sophistication in general.

Barthes’s systematization becomes more difficult and less relevant for stories with few events. In “Pegasian,” for instance, sequences are difficult to distinguish. One could see the horse-riding lesson as the first sequence and the text’s lesson (“As long as you take off “) as the second. The cardinal function for the first sequence could be condensed as “dressage.” In this view indexes would be made up of all symbols of drill and submission, such as the “real pair of riding breeches” and the “background information” that would teach the pupil respect and politeness. The fact that one of the crucial indexes has to do with a pair of pants could then be seen as the symbolic combination of “dressage” and “dress” (in the meaning of clothing in general). This may indeed appear somewhat far-fetched, but obviously any systematization by the reader will have something arbitrary. There is no cogent method one can simply apply in order to arrive at the deep structure of events. The reader has an important role. The Bond story could be divided into three sequences or thirty sequences. Rather than fixed elements that can be abstracted from the text, structures are constructs that are always partly dependent on the reader.

The choice for Barthes’s system has something arbitrary as well, since many other options are available. For instance, rudimentary systematizations of story events can be found in Propp, who has developed thirty-one functions in his analysis of Russian fairy tales.18 The same goes for Eco, who has distinguished nine crucial moves in a typical Bond novel.19 Todorov claims that “the minimal complete plot consists in the passage from one equilibrium to another. An ‘ideal’ narrative begins with a stable situation which is disturbed by some power or force. There results a state of disequilibrium; by the action of a force directed in the opposite direction, the equilibrium is re-established; the second equilibrium is similar to the first, but the two are never identical.”20 The relatively linear or even deterministic sequence of these functions, moves, or (dis)orders appears only at the level of deep structure. In a concrete fairy tale or Bond novel, that system will mutate in various ways. The same holds for Greimas’s four action phases, consisting of manipulation, competence, performance, and sanction. These phases may also intertwine, thereby reducing the linearity of the narrative evolution.21

Less linear is Claude Bremond’s systematization. He starts from so-called pivotal functions, which always leave open the possibility of success or failure. Barthes’s sequence becomes a succession of three pivotal functions in Bremond. First there is possibility, which is followed by realization, and finally there is completion.22 For instance, a woman can devise a plan to kill her husband in order to inherit his wealth. The murder sequence starts with the possibility of carrying out the plan or not. If it is carried out, then the murder attempt may be successful or it may fail. If the murder proceeds as planned, then the woman may or may not inherit the money. Like Propp and Eco, Bremond envisages various transformations taking place between the relatively simple three-function structure and the often complicated developments in a concrete narrative text.

We have opted for Barthes’s system because it is far less hampered by such a complex series of transformations and because it does not start from frameworks as rigid as those offered by his colleagues. Barthes’s indexes, functions, and sequences are open concepts that the reader has to fill out with elements from the text. They do not impose a rigid order or interpretation. If, however, the reader’s processing gets a central place in the analysis, one moves away from the classical, structuralist action grammar. An example of such a readerly transformation of structuralist concepts can be found in the work of Emma Kafalenos. Building on the work of Propp, Greimas, and Todorov, she develops a model with ten functions.23 She studies them in terms of the concrete narrative (con)text and of the reader’s activities: “I use functions [ . . . ] to record readers’ interpretations as they develop and change (or fail to change) during the process of reading.”24

1.2. Actants

Events cannot be independent of the agents who are involved in them. We describe these agents with the term “figures,” which we will soon specify as actants, following Greimas. The term does not refer to the actual manifestation of a character in the text but rather to the specific role a character plays as an abstract agent in a network of roles on the level of the story. Here too every structuralist has developed his or her own networks and systematizations. Bremond, for instance, conceives of two fundamental roles: a passive one and an active one. Active figures steer and direct events, even though they often do not consciously develop a strategy. A prime example of such a figure is again James Bond. Passive figures such as the agent who is killed at the beginning of “From a View to a Kill” undergo events. On top of this, there are three criteria for going into the details of figure characterization: influence, modification, and conservation. Influence typifies figures—such as a seducer or an informant—who purposefully make a direct impact on the course of events. Modification marks figures who improve or aggravate the situation, while conservation distinguishes those who try to avert change.25 This explanation of Bremond’s criteria consistently presents the figure as an active agent, but obviously there are also passive figures who are influenced, modified, or stopped in their effort toward change. The same character can be both active and passive, depending on the viewpoint. The female rider in “Pegasian” actively wants to improve her situation, but she is “passively” helped by the riding master, who at first seems to hinder her.

Greimas’s actantial model is better known than Bremond’s systematization of roles.26 In its simplest and most useful version, this model consists of six roles or actants. These terms are synonymous with “figures.” There is a subject, who carries out the action and who strives for a specific object. This quest is inspired and provoked by a destinateur, whom we will call “sender” following Cok van der Voort.27 Greimas calls the agent who benefits from the quest the destinataire, which Van der Voort translates as the “receiver.” The agent who assists in the quest is the helper, while the agent who thwarts it is the opponent. This results in the following system:

The italicized terms indicate the abstract relations between the actants: the subject strives for the object (desire), the sender wants the subject to transmit the object to the receiver, and the opponent and helper are involved in a struggle.

These are all abstract roles that should not be confused with actual characters. One character may play all the roles. In the case of a man who wants to quit smoking, one could say the subject is the smoker and his object, quitting. The sender is also the smoker—he himself wants to stop, he himself thinks it is necessary. The receiver is the smoker as well—he will benefit from giving up. The smoker’s willpower is the helper, and his old addiction amounts to the opponent. This example shows that roles do not have to be played by real characters. In addition, an emotion, a motivation, or an idea can function as an actant, when any of those performs, for example, as the sender.

Just as one character can play all the roles, one role can be played by many characters. Bond can get help from people such as the beautiful girl or the man from intelligence, but his helpers can also be state-of-the-art weapons or even more abstract things, such as his courage or resourcefulness.

This story structure has the advantage of being simple and generally applicable. It can literally be applied to every narrative text. For instance, the Marxist philosophy of history can be represented with the terms offered by Greimas. Its subject is humanity and its object, a classless society. History is the sender and humanity (or at least the proletariat), the receiver. The proletariat is the helper as well, whereas capitalists play the role of the opponent. In the case of “Pegasian” the female rider is the subject, and the story’s object is being able to fly. The horse—more specifically perhaps the winged horse Pegasus, the symbol of the muse linked to poetry—plays the role of the helper. Dressage and the riding master at first seem to act as opponents, but eventually they turn out to be helpers as well. The sender is the desire to overcome gravity, while the receiver is the girl and, on a larger plane, perhaps also the reader who understands the moral lesson.

Simplicity and general applicability are at the same time the model’s disadvantages. It seems just too easy to reduce all characters and motivations to six roles. If the role of the sender can comprise such diverse elements as a motive, an onset, a character who obliges or invites, and an order or a law, then one might ask whether it would perhaps be useful to specify the category of the sender somewhat further or even to divide it into a set of subcategories. The general applicability of the model also means that it lumps all kinds of narrative texts together and treats them indiscriminately, whether it is the Marxist philosophy of history, the story of the man who wants to give up smoking, or the story of the female rider who wants to learn how to fly.

Furthermore, Greimas does not offer an easy method to go from the actual narrative text to the actantial model. Different readers will come up with different actantial structures for the same story. In “Pegasian” the riding master could also become the subject, in which case the object would be the teaching of the necessary discipline. The sender would then be the riding master or, more generally, the demands of horsemanship. The female rider in this view is still the receiver, but she also acts as the opponent. Finally, the helper is the horse, which lets itself be trained. Complex texts with many events risk the development of totally diverging actantial models. Readers who appoint the murderer as the subject of a detective novel will obviously come up with a different model from those who choose the detective for this role.

Extensive narrative texts often complicate the application of the model. Does one need to devise one model for the entire text or one for every chapter? Or maybe one for every sequence or for an even smaller unit? If each of the seven sequences of “From a View to a Kill” is analyzed according to the actantial model, then it becomes clear that James Bond does not act as the subject in the first three sequences. He is absent from the murder sequence. In the second sequence (“the hero is summoned”) he functions as the object. In the third sequence (“the hero is briefed”) he acts as the receiver since he acquires the information. It is only in the fourth sequence that he becomes an active heroic subject, thereby finally assuming the role one would expect of him. This abstract order shows how the main character is first announced and then patiently put together: he goes from absence to object, from object to receiver, and eventually from receiver to hero. Greimas permits the discovery of a structural principle that might otherwise remain unnoticed.

In this way the systematization of actants, just like the systematization of the story’s actions, assures a better understanding of the macro- and microstructures of a narrative text. Actions and events differ from one another on the basis of actant involvement. An action derives from an actant, while an event happens to the actant. In naturalist novels events usually take precedence over actions. Human beings find it hard to resist the events that befall them. However, this contrast between actions and events does not amount to a fundamental distinction, since the actantial model allows for the interpretation of events as actions by abstract actants such as fate, death, old age, or social class. In this way both actions and events can be made to fit the actantial model.

This fact points to the interdependence of actions and actants. The reader will expect certain actions from a specific actant. Very often these expectations are linked to stereotypes circulating in the reader’s social and cultural context.28 By playing with a reader’s anticipations, a narrative text can create suspense and take surprising turns. At the beginning of a detective novel the reader might think that a given character is a helper, but the character’s actions might slowly lead to the suspicion that this individual could be an opponent. Conversely, the confirmation of expectations creates a certain predictability that some readers take as a guarantee of reliability. Certain deeds are expected of heroes. If they do not deliver, they will not be considered real heroes, and in that sense they are unreliable characters. One does not expect the same feats from an octogenarian as one expects from a hero like James Bond.

If we connect the actant to both its actions and its depth, then we are moving from abstract role to concrete character. Traditionally there exists an inversely proportional relationship between the amount of action and the degree to which a figure is psychologically developed into a many-sided character. The more action there is, the less profound the character. This rule may not always apply, but it certainly holds true for traditional genres such as the adventure novel and the detective novel.29 Profundity is defined by the number of character traits and their variation. Forster has made the traditional distinction between, on the one hand, static, one-dimensional flat characters and, on the other, variable, many-sided round characters.30 This distinction is quite problematic. Leopold Bloom in Ulysses has many aspects, but he does not really develop. An allegorical character such as Everyman (from the eponymous medieval morality play) is notably flat, but he does develop.

Rimmon-Kenan proposes to determine the richness of a character with the help of three sliding scales, which together make a three-dimensional coordinate system.31 The first scale indicates complexity and goes from a single characteristic on the one pole to an infinite range of characteristics on the other. The second scale, which deals with development, runs from the pole of stagnation to that of infinite change. The third scale indicates the degree to which the text shows the character’s inner life. At the left end of this scale Rimmon-Kenan situates characters seen only from the outside, while at the other end she places characters whose inner lives are described with great attention to detail. In a psychological novel, many characters will presumably occupy positions close to the right-hand end of the scales (numerous characteristics, significant development, and an insistence on inner life), while in an action-packed story, like the one about James Bond, most characters will appear closer to the left-hand end. In “The Map” the I-character is not very complex, as few of his features are mentioned. On the other hand, there is considerable development since the young I who believes in the magic of mapping evolves into an older I who has practically no illusions left on this score. Of the two types of I, the reader sees mainly the interior.

Such a three-dimensional characterization of role makes the transition from a deep, abstract level to the level of visible characters in the concrete narrative text. Seymour Chatman describes the role as the “syntagmatic reading” of a figure since the latter functions as an element in a horizontal chain of actions, a position in a network of connecting events. The paradigmatic reading considers the figure as a set of traits, a vertical stack of indexes referring to a personality and therefore to a concretely drawn character.32 If one does not see the female rider in “Pegasian” as merely the abstract role of a subject reaching for an object (notably “taking off”), then one shifts to the more concrete level of characterization, at which she will be described with adjectives such as “playful,” “disrespectful,” and “relativizing.”

1.3. Setting

There is more to the story than actions and actants. Events take place not only in conjunction with certain roles but also in a specific time and place. Such a spatiotemporal indication is often described with the term setting. The Russian literary critic Mikhail Bakhtin prefers to speak of a chronotope, a textual combination of time (chronos) and place (topos).33 According to him, the spatiotemporal setting constitutes the narrative and ideological center of the text because it shapes figures and actions. Abstract themes like love and betrayal acquire a concrete form within and thanks to a specific chronotope. The Greek romance, for instance, features what Bakhtin calls “adventure time,” which requires “an abstract expanse of space” for the genre’s typical abductions, escapes, and pursuits to take place.34 An abstract view of humankind and social reality can be concretized only if figures (humankind) and events (reality) are embedded in the chronotope.35 The heroes of Greek romance thus roam the space that is available to them. Insofar as the text embodies a worldview, it contains an ideological dimension, which we will elaborate in chapter 3.

As the example of the Greek romance already indicates, Bakhtin proposes that every genre and every type of discourse develops its own chronotopes.36 His other examples include the picaresque novel, which centers on “a road that winds through one’s native territory,” and the idyll, which is determined by “the immanent unity of folkloric time.”37 However, one could also think of the Gothic novel with its combination of the haunted house and events often taking place at night. If Bakhtin is right, the story’s credibility rests to a large degree on the interaction between actions/events, actants, and setting.

Actions cannot be separated from the setting. An account of a chase requires the description of the scenery as it rapidly passes. Moreover, the setting often amounts to an index for the action. In the story discussed earlier, it is no coincidence that James Bond unmasks the killer in the same environment where that very killer used a disguise to shoot an agent. Although the road through the woods is not a highway, as an index it refers to culture, while the woods themselves are part of nature. Once Bond has shot the killer on this road and removed his accomplices from the woods, nature has resumed its innocence and attraction. The final scene takes place in the woods. Bond talks to the beautiful girl, and his words show that nature has traded its connotations of terror for those of eroticism: “Bond took the girl by the arm. He said: ‘Come over here. I want to show you a bird’s nest.’ ‘Is that an order?’ ‘Yes.’”38

This example proves that the setting can also function as an index for the actants. Good Westerners live in the civilized city space, whereas bad Soviets live in the natural habitat of the forest. The clash between them plays out in a space between these two environments, as well as in an in-between time, the period between night and day (seven o’clock in the morning).

At first glance the spatiotemporal background against which the story develops appears relatively fixed. The Russian formalists categorize it as a static motif; Barthes would call it a pure or an informative index. Both terms are appropriate since the fictional universe does not cause the story to develop. However, story development is inconceivable without the setting, which makes it possible for actions to take place and actants to become involved in them. It is impossible to imagine roles and events without embedding them in time and space.

Chatman’s schematic representation of the story insists on the fundamental connections between actions, actants, and setting.39 His visualization is as follows:

Events are dynamic components of the story, while existents are relatively fixed points around which the story can unravel. Obviously, characters and setting can develop in the course of the story, but in Chatman’s proposal a certain stability remains. Ruth Ronen qualifies this relative permanence by defining setting as “an immediately relevant frame [i.e., a fictional place] regardless of the continuous textual evidence for its relevance.”40 Ronen does not require Chatman’s continuity.

In a very elaborate analysis of setting, Gabriel Zoran tries to solve the fundamental problem: that time rather than space dominates “in the structuring of the narrative text.”41 Bringing in the experience of the reader (who is mostly invisible in structuralist narratology), he distinguishes three levels of the “reconstructed world” in which the characters operate and the action takes place. On the textual level, the fictional world still retains some of the structuring patterns of the text. It may be determined, for instance, by the perspective of a character (which to us is an aspect of “narrative”) or by the narrator’s regular mention of a specific location (which to us is an aspect of “narration”). On “the chronotopic level, the reconstructed world is already independent of the verbal arrangement of the text, but is still dependent on the plot,” which means, for instance, that certain locations are points of departure and others represent the end of the journey. Finally, on the topographic level, “the world is perceived as existing for itself [ . . . ] cut off entirely from any structure imposed by the verbal text and the plot.”42 The topographic level is clearly part of the story in its most abstract guise, and the chronotopic level also implies a degree of the abstraction integral to our definition of the story. Interestingly, the reader supposedly moves back and forth between the three levels, which turns the notion of the story into an element of the act of reading. In chapter 3 of this handbook we will return to the role of the reader in the construction of the fictional world.

Zoran conceives of the topographical structure as a map based on a series of oppositions. Structuralism in general likes to work with binary oppositions that can form the basis of a sliding scale.43 Greimas thus distinguishes between topical spaces (where the action takes place) and heterotopical spaces (where the previous or subsequent actions take place).44 Following Mieke Bal, one could investigate space relying on pairs such as inside versus outside, high versus low, and far versus close.45 It is no coincidence, for instance, that the interminable tortures in the work of the Marquis de Sade almost always take place in the closed, dark space of an underground dungeon. A structuralist will use similar oppositions to characterize time: short versus long, continuation versus interruption, day versus night, light versus dark. In his story “Het lek in de eeuwigheid” (The leak in eternity), the Dutch author Willem Frederik Hermans indulges in the opposition between a long darkness and a brief period of electric illumination by a fixture that switches off automatically. Just as the light comes on briefly in an eternity of darkness, human life appears briefly in an eternity of death.46

The central aspects of Bal’s space and time characterization, which she borrows from the structuralist semiotician Juri Lotman, are the drawing of a borderline and its potential transgression.47 Actions and actants transgressing these borders often play a central role in the story. A burglar or spy is unthinkable without the violation of the border between private and public, open and closed. Murderers and rapists do not respect these borders either. In the bourgeois novel heroes often repair borders, while in the adventure novel they are likely to overturn the bourgeois system. Transgression, for that matter, may be a step on the way to recovery. In the medieval Dutch epic Karel ende Elegast (Charlemagne and Elegast) the title character, Karel, goes out stealing in order to discover who stands inside and who stands outside the feudal space. Of course the stealing takes place at night and includes a journey through a dark wood. Night and the wood form part and parcel of the chaos that normally threatens order but that in this case brings about its restoration.

In “Pegasian” space and time are not evoked very clearly, but some indications are nevertheless available. The story concerns a lesson during which many horses trot around in a “carousel.” The association with a merry-go-round provides points to the story’s central theme— dressage and discipline. The horses do not run around in nature, and their circuits in the riding school make them as unfree as the wooden horses on a merry-go-round. This image therefore conjures up three different spaces: nature, the riding school, and a fairground. If the carousel in the fairground implies an element of fun along with the immobility of its horses, this aspect of the story’s spatial structure might resolve its basic opposition: nature and the discipline of dressage.

Space and time are important in “The Map” as well. The boy discovers the near-divine map on a Sunday, and he sees it through a forbidden gap. The map’s attraction can largely be attributed to the fact that the boy’s peek at it was unexpected and actually prohibited. Later on, the map allows a look at the entire environment of his youth, at all the roads and pathways he biked as a boy. In this respect the map provides a visible and spatial representation of an entire period in his life. But as soon as that representation is complete, the fun is over. Regarding “The Map,” Zoran’s conception of the topographical structure as a map coincides with the object that evokes the narrator’s entire childhood. This overlap probably helps the reader to imagine space on the topographical level, a mental effect that might well make it easier to develop a version of the nostalgia at the heart of the story. While this interpretation does work with the combination of space and time on the level of the story, its emphasis on the reader goes beyond the tenets of structuralism.

In Wasco’s book of graphic fiction, Het Tuitel complex (The Tuitel complex), the “City” page we focus on in this handbook appears to the right of a page marked “Horizon.” The differences between the two in terms of color (light versus dark) and in terms of image size (wide like a horizon versus narrow, as in an inner city streetscape) appeal to a general opposition between country and town. It’s only at the very end of “City” that the protagonist’s spaceship rises above the roofs and a panoramic view reenters the text, and even then it is not an open space but a cramped, claustrophobic one. The horizon is nowhere to be seen; it is blocked by the mass of buildings. However, Zoran’s topographical structure here amounts to an unusual map, since the urban environment is quite strange. Unusual holes dot the streets and ramps look like dangerous slides. Sharp objects protrude from unexpected places, and there is even a kind of electric chair on a terrace. Panels eleven and twelve feature weird works of art, which enhance the artificiality of the environment. If there is life in this city, it can’t be seen (save for a lonely bird, whose yellow color seems to suggest it doesn’t belong there). Of course the spaceship indicates the future, which could suggest the chronotope of the dystopia, in which a worry about our contemporary world (like the anonymity of city life) appears as a characteristic of a more or less distant time (the city in “City” is empty).

2. Narrative

Narrative constitutes the second level of structuralist narratology. This level no longer concerns the abstract logic of sequences but rather the concrete way in which events are presented to the reader. As can be seen in the following diagram, the analysis of narrative consists of three main parts: time, characterization, and focalization.

2.1. Time

Structuralism analyzes time by studying the relation between the time of the story and the time of the narrative. For instance, a central event in the story may well remain untold in the narrative, or an event that takes a long time to transpire in the story might be mentioned briefly and casually in the narrative. In order to systematize the various aspects of time, Genette uses three criteria: duration, order, and frequency.48

Duration is measured by comparing the time necessary to read the account of an event to the time an event takes on the level of the story. The first of these two dimensions builds on the act of reading in order to determine how long an action or event lasts on the level of narrative. Since these actions and events take place in the narrative as it is being told, this dimension is usually called the time of narration, even though what really matters here is the time of reading. In the next figure presented, this time on the level of narrative appears as TN. The second dimension is usually called narrated time and refers to the duration of events on the level of the story, which is why it appears as ST (story time) below. Since Günther Müller had already introduced the distinction between the time of narration and narrated time in 1948, it existed long before the advent of structuralist narratology.49 Bal distinguishes five possible relations between TN and ST.50 We represent them on a sliding scale as follows:

At the ellipsis pole, an event that does happen in the story is absent from the narrative. As a result, story duration becomes infinitely longer than duration in the narrative. Events that remain untold can be very important. A crime novel, for instance, will effect more suspense when the execution of a planned murder or assault does not appear in the narrative. In a psychological novel, things that remain unsaid can be essential because they may point to repressed or dismissed traumas.

Acceleration is another term for summary. An event that takes a long time can be summarized in one sentence, so that the time of narration is shorter than story time. In “The Map” the narrator says, “and the roads I had not had yet, that is where I went.” The bicycle rides, which must have taken quite some time, are summarized very briefly, which makes the narrative move faster than the story.

Scene indicates an almost perfect overlap of the duration of an event with that of its representation or reading. A dialogue that appears word for word in a novel will take almost as long to read in the text as it takes to utter in the story. The equation sign on the scale, however, is of course a fiction since the time of narration and narrated time are never entirely identical. For instance, it is almost impossible to make pauses in the story conversation last equally long in the text. A brief line such as, “The conversation came to a stop,” is an example of acceleration rather than a scene.

Deceleration occurs when the time necessary to read the description of an event turns out to be longer than the event itself. A text can halt, for instance, at the moment a killer points the gun at a victim. This would take merely a second in the story, but it can be described in dozens of pages. Deceleration, therefore, is very useful for creating or decreasing suspense. An almost scenic description of a fight thus may be followed by a deceleration in which the narrator enters at length into a brief event such as the arrival of the police. The Dutch author Gerard Reve likes to use this strategy: in his novels extended artistic descriptions decelerate the action, which often does not amount to much. Since these descriptions, which circle the unspeakable secret appearing in every Reve novel, are there to justify the passivity of the protagonists, one could say that form adheres to content. At the beginning of Het boek van violet en dood (The book of violet and death), the narrator even makes Reve’s habit explicit: “No, nothing much happens: I meet someone; I meet that someone again once or twice, and then he tragically disappears.”51 The rest of the narrative comes down to one giant deceleration that continuously postpones the little action there is.

Pause represents an extreme form of deceleration. Nothing happens anymore, so the story comes to a standstill. A clear example of this occurs in Max Havelaar, by Multatuli. Stern, the narrator, discusses the precarious balance between the continuing of the narrative and its temporary suspension. By way of example he brings up “the heroine who is leaping from some balcony four floors up.” Instead of describing that action, he brings it to a halt: “Only then, with a bold contempt for all the laws of gravity, shall I leave her floating between heaven and earth until I have relieved my feelings in a detailed picture of the beauties of the countryside.”52 Seventy pages later the narrator returns to the moment when he introduced the pause: “I would give a good deal, reader, to know exactly how long I could keep a heroine floating in the air while I described a castle, before your patience was exhausted and you put my book down, without waiting for the poor creature to reach the ground.”53

The combination of ellipsis, acceleration, scene, deceleration, and pause determines the rhythm of the narrative and contributes to suspense or monotony. Narrative texts with continuous acceleration or deceleration create a much more dynamic impression than texts that always opt for the same type of duration. Sketches such as “Pegasian” mostly go for acceleration, and indeed the riding lesson is described only briefly. “The Map” summarizes an entire period in a few sentences, and it deals with a substantial part of the narrator’s youth in a few paragraphs. This summarizing method of representation is relinquished only briefly in order to describe how the boy sees the map in the shop window. This brief change has an effect similar to that of zooming in with a camera; it enables the reader to concentrate on a specific detail or fleeting event.

When trying to establish duration, the definition of the time of narration presents a major problem. How does one measure the time the narrative devotes to an event? Is that the time required to describe the event or to read about it? It is usual for reading time to function as the norm, but this speed obviously differs from reader to reader. Structuralists then have recourse to a purely quantitative element: the number of pages. Forty pages to describe one minute means deceleration, while one page to describe a year comes down to acceleration. This means that time is reduced to space or more specifically “the amount of space in the text each event requires.”54 By the reduction of temporal development to a certain number of pages, time is stripped of its dynamics. This connects with the already mentioned spatialization characteristic of the structuralist approach.

Another problem with duration is the definition of narrated time. Some narrative texts, such as the nouveau roman and postmodern encyclopedic novels, make it very difficult to reconstruct the story or even the events. In his encyclopedic novel Groente (Vegetables), the Dutch author Atte Jongstra presents a collage of texts taken from manuals, cookbooks, and reference works, and he even includes pictures. This novel no longer has a story made up of chronological and causal connections. How, then, can readers establish the duration of events? If they can’t, it also becomes impossible to search for the relation between the time of these events and that of their description, which means the structuralist definition of duration does not apply here.

A similar problem arises with regard to order. Order is determined on the basis of the relation between the linear chronology in the story and the order of events in the narrative. If it is impossible to reconstruct story events and to arrange them in a clear chronology, order in a narrative text cannot be assessed by using the structuralist method. If it is possible to order events nicely on the story level, for instance in a sequence from one to five, then one can see how the narrative complicates that order, such as in the sequence four, two, five, one, three.

Genette specifies order with reference to three categories: direction, distance, and reach. Specification always depends on a clear primary narrative. This primary narrative or récit premier functions as a norm or, in spatial terms, as a measure for the location of events in time.55 The primary narrative is not the same as the story, because it is visible in the text and does not necessarily contain all the events of the latter. Still, the primary narrative poses the same problem as the story. If a novel does not allow the reader to establish its primary narrative, one can forget about order altogether. A text brimming with associations, such as James Joyce’s Finnegans Wake, cannot be approached with this method.

Two directions are possible with regard to the primary narrative: forward and backward. If the primary narrative shows, for instance, the last three weeks in the life of a man who is the protagonist, all memories of his youth and all anticipations of life after death would fall outside this narrative. Such a memory would be an example of analepsis, and such an anticipation would be an example of prolepsis. The equivalent English terms would be flashback and flashforward.56 In German, Eberhard Lämmert popularized the terms Rückwendung and Vorausdeutung.57

If the analepsis or prolepsis concerns the element in the foreground of the primary narrative, Genette calls it homodiegetic. For instance, if a dying man remembers a moment from his own life, this would constitute a homodiegetic analepsis. If, however, he remembers something about a person who does not appear or has only a minor role in the primary narrative, then the analepsis is heterodiegetic. The dying man may remember a boyhood friend who has disappeared, which may lead to a story about that friend and some related details concerning him, none of which the dying man has experienced himself.

Defining direction can often be tricky. Suppose the dying man remembers something from his adolescence but then looks ahead from that period to his twenties. The prolepsis with respect to his adolescence is an analepsis with respect to the primary narrative. “The Map” features a mild version of this: “because I had had all roads, nothing was added anymore, and one day I would remove the map from the wall.” This one day represents a prolepsis with respect to the period in which the boy was biking around but an analepsis with respect to the moment at which the narrator remembers his youth.

The situation becomes more complex when the various memories are not clearly dated. Many autobiographical novels contain a whirl of memories and anticipations that connect associatively and are very hard to locate. In such a case the reader does not know whether memory A goes backward or forward with respect to memory B. Genette uses the term achrony for passages that cannot be dated. Prolepsis and analepsis, on the other hand, exist only if they can be clearly located in time. They are examples of anachrony, a departure from the chronology in the primary narrative.

Order is a matter of not just direction but also distance, which concerns the temporal gap between primary narrative on the one hand and prolepsis or analepsis on the other. The dying man may remember an event that took place two days ago, which therefore falls within the primary narrative, or he may remember something that happened fifty years ago, which clearly remains outside the primary narrative. If the remembered or anticipated period falls within the primary narrative, Genette speaks of an internal analepsis or prolepsis. External is when this period falls outside the primary narrative. And finally there is mixed analepsis or prolepsis, which covers a memory starting before the primary narrative but ending within it, or an anticipation beginning within the primary narrative and ending outside it.

Apart from direction and distance, order is also characterized by reach. This term refers to the stretch of time covered by the analepsis or prolepsis. If the memory concerns one particular event, then the analepsis is punctual. If it constitutes an entire period, the flashback is durative or complete. The analepsis in “The Map” is durative since it describes the complete extent of time from the discovery of the map until its removal.

Although the number of terms enumerated here suggests a rather abstract system, investigating order in a narrative text is of great importance. The more an author indulges in flashbacks and flashforwards, the more complex the narrative becomes. This also leads to all sorts of new relationships between the various periods. If on the same page the text refers to three or four periods from the life of the protagonist, chances are that a reader will start to see connections between these periods. As a result, themes may emerge more clearly or suspense may increase. In Sunken Red, by Jeroen Brouwers, the main character’s thoughts go back and forth between very divergent moments: the Japanese internment camp, the boarding school, the sexual relationship with Liza, the garden party, the birth of his daughter, and the death of his mother. All these stages connect through the joint image of his mother’s disgrace. The turmoil in the novel’s time structure formally reflects the unrest and roaming typical of the I‑character.

Frequency refers to the relation between the number of times an event occurs in the story and the number of times it occurs in the narrative. Obviously there are three possibilities here: less often, more often, and just as often. When the event occurs just as often in the story as it does in the narrative, Genette uses the term singulative. Something that happens once and is described once is a simple singulative, while a reoccurrence in the story that is described just as often in the text is a plural singulative. The discovery of the map in Gerrit Krol’s story provides an example of a simple singulative. If the boy had visited the store more than once, and if each of these visits had appeared separately in the text, then that would have been a plural singulative.

Very often such an exact coincidence does not seem appropriate. If you describe something that happens regularly every time it happens, the text may become monotonous or endless. For story events that happen repeatedly but are presented only once in the text, Genette uses the term iteration. The first sentence of Marcel Proust’s Remembrance of Things Past offers a good example of this second type of frequency: “For a long time I used to go to bed early.”58 The formulation “for a long time” probably covers thousands of days on which the protagonist went to bed early. Iteratives are prevalent in the description of habits. Here are some examples from “From a View to a Kill”: “Bond always had the same thing—an Americano—Bitter Campari—Cinzano”; “When Bond was in Paris, he invariably stuck to the same addresses”; “After dinner he generally went to the Place Pigalle.”59 In “The Map” the clause “I occasionally traveled somewhere by train” is an iterative since the journey is mentioned only once but will have taken place many times.

Iteratives can be combined with singulatives. A party described singulatively can contain an iterative such as “He repeatedly harassed his neighbor, until she could not take it any longer and left the table.” Genette calls this an internal iterative since it remains within the temporal limits of the singulatively described party. If it were to fall outside these limits, Genette would call it external. For instance, the description of the party could contain a sentence such as the following: “That is what he would do for the rest of his life: harass people who did not ask for it.”

Genette calls the third type of frequency repetition, by which he means the repeated description in the text of an event that takes place only once on the level of the story. Thus the main character in Sunken Red continues to ruminate on the scene in which his mother is beaten by a Japanese soldier. Repetitions of this kind often embody various standpoints, that is to say, the same event is considered by various characters. With postmodern novels it can be hard to decide whether the various standpoints relate to a single event or various events or whether they are sheer invention. Een fabelachtig uitzicht (A fabulous view), by the Dutch author Gijs IJlander, includes several versions of a walk during which a dead animal, possibly a squirrel, is found. The characters entertain widely diverging views of what happened, which may lead the reader to doubt their truthfulness. The narrator, a stuffed squirrel, does not decide the matter, and perhaps the characters’ views are even the animal’s fabrication.60 For such a complex and undecidable case, structuralism, which functions only on the basis of clear event reconstruction, cannot offer a solution.

2.2. Character

Having addressed time as the first dimension of narrative, we now take up character, the second dimension. While story deals with abstract roles, narrative involves their concretization. The central question in this respect concerns the way in which a character is present and represented in narrative.

First, its presence can be grasped in terms of characteristics, which structuralist narratology tends to schematize in a list of features. The most famous example of this is Barthes’s semantic or semic code, which he also calls the character code and which consists of a combination of minimal semantic characteristics (semes). Characters are the result of such combinations: “When identical semes traverse the same proper name several times and appear to settle upon it, a character is created. Thus, the character is a product of combinations: the combination is relatively stable (denoted by the recurrence of the semes) and more or less complex (involving more or less congruent, more or less contradictory figures); this complexity determines the character’s ‘personality,’ which is just as much a combination as the odor of a dish or the bouquet of a wine.”61 In this view “the person is no more than a collection of semes.”62

Later structuralists, like Philippe Hamon in his classic 1972 study of character, will present a much broader view on character.63 They recognize the role of the reader in the semiotic makeup of the character, but even then the list of semes remains the basic level of investigation. Hamon calls that level the “signified of the character,” and he represents the collection of semes in terms of “axes”—such as gender, geographical origin, and ideology—and “functions,” such as victorious battle and reception of a good.64 Again, a character is described as a list of characteristics.

The second aspect of the character is more dynamic and focuses on the way it is presented in the narrative. Hamon deals with this aspect in two ways: he looks at the character as signified, that is, the form given to the character (e.g., its name), and at the rules that this formulation has to follow (e.g., rules of logic, common sense, and genre). This leads to a complex frame that is hard to use for the practical study of characters and combines the levels of narrative with that of narration. As a consequence, we will start from a much simpler view on characterization, the one offered by Rimmon-Kenan, who discerns three ways of characterization.65

First, a character can be described directly.66 This type of characterization occurs in many traditional novels that introduce a character with an enumeration of character traits. These traits may relate to psychological states as well as to outward appearance. Direct characterization always takes the form of specifying and evaluative statements such as the following: “Mister Hoorn was a warm and honest individual, though his casual conversation and jokes could not be called brilliant. But stupid, no, that he was not.”67

A central question in this connection relates to the origin of such statements. Does the character itself pronounce them? Or do they come from an omniscient narrator, or another character? The answers to these questions have a profound influence on the reliability of the characterization. Direct characterizations belong to the most straightforward strategies to inform the reader, but they can easily be (ab)used to send the reader in the wrong direction. At the beginning of the story “A Rose for Emily,” by William Faulkner, the characterization of “noble” Emily is emphatically positive, but the reader soon realizes that those positive statements are inaccurate and misleading.68

The second type is indirect characterization.69 This type is based on metonymy, that is, it works with elements that are contiguous with the character. Actions, for instance, often follow naturally from a character’s identity. Discourse too says a lot, literally and figuratively. The words and style used by characters betray their social position, their norms and values, and their psychology. The characters’ physical appearance and their environment can be telling too. Ben, the main character of Ansichten uit Amerika (Postcards from America), by Willem Brakman, moves to new residences a number of times, but his environment continues to resemble a labyrinth. His house is “very intricately designed,” and the streets form an obscure network and “become hard to follow.” The phrase “labyrinth of small streets” comes up in all sorts of contexts related to Ben.70 It therefore says something about the claustrophobic and paranoid worldview of this character.

Third, characters can be described with the help of analogy, which leads to metaphor instead of metonymy.71 In “Pegasian” the main character’s identity is partly established through implicit comparison with the horse. Just like the horse, the female rider wants to break free from the ground and take off. The latter refers to the text’s message. The fact that metaphors often refer to a specific ethic or ideology also appears in Theodor Adorno’s study of the images Kafka uses to describe his characters. Kafka often compares his characters to animals and objects, and this metaphorical typification shows how unhuman humankind has become.72

For Rimmon-Kenan the name is an example of characterization through analogy.73 To the extent that the name points to an aspect of the character or to a contiguous element pertaining to it, we believe it still belongs to metonymic characterization. Thus, the names Goodman and Small describe metonymically, whereas Castle or Roach do so metaphorically. In the former case, elements are put forward that belong to the semantic domain of humankind, while in the latter case, other domains come into play. In the novel Bint, by the Dutch author F. Bordewijk, the pupils of a class called “Hell” have suggestive names such as “Saint’s Life” and “Precentor.” Such metaphorical or symbolic names may of course refer to the opposite of what they suggest. A character called Castle may well be weak, in which case the name is ironic, to say the least.

Similar to the name, the alter ego or second self presents a borderline case between metonymical and metaphorical characterization. Metonymical characterization does not lead to osmosis, while its metaphorical counterpart does. The borderline between the two is not always clear. Two supposedly distinct characters may resemble each other in so many ways that one could still speak of identification or blending. This is true, for instance, of the alter egos in De ontdekking van de hemel (The discovery of heaven), by Harry Mulisch. Uri Margolin mentions the example of a character blending with an axolotl, a type of salamander; the two exchange personalities.74 This obviously underscores the dynamic aspect of characterization. Margolin distinguishes between three types: a transformation within one character, an evolution between two or more characters (which leaves the number of characters unchanged), and finally a change in the number of characters (e.g., by cloning or schizophrenic splitting).75

The structuralist penchant for abstract and unchanging deep structures goes against the concrete and dynamic nature of characters and characterization. This difficulty is borne out, for instance, by the impossibility of defining what actually constitutes a hero. Bal has drawn up a list of characteristics, including “the hero occurs often in the story,” “the hero can occur alone or hold monologues,” “certain actions are those of the hero alone,” and the hero “maintains relations with the largest number of characters.”76 A relevant question is not only how many of these characteristics have to apply before one can speak of a hero but also whether the hero concept is at all relevant for nontraditional texts, such as the nouveau roman, or for classical genres, such as the epistolary novel and the novel of manners. In the nouveau roman the hero seems to disappear in favor of an impersonal quasi objectivity; in the epistolary novel, all correspondents being more or less equal, there is no center; and in the novel of manners intense interaction between groups and classes makes a criterion such as “certain actions are those of the hero alone” irrelevant.

More generally, one might ask whether a narrative text always needs a hero. If the answer is yes, a good starting point may be Philippe Hamon’s approach. He enlists a number of narrative characteristics that are typical of heroes; they have qualifications that are unique to them (or at least unique in that particular degree); they have a wide spatial distribution (i.e., they appear often and in many places); they often appear autonomously and on their own; they have functions and can perform actions that are not equally distributed among the other characters; they are often designated by genre conventions and/or by explicit characterizations as “the hero”; and finally, there is a certain abundance and even redundancy in their characterization. However, even these dimensions are not objective and universal criteria for deciding whether or not a character is a hero.77 The role of the reader cannot be disregarded here.

The fact that structuralist narratology holds on to concepts such as hero and villain suggests that it still deals with characterization in a very anthropomorphic way.78 Coming from a theory that explicitly dissociates itself from subjectivist and humanist approaches to literature, this may be surprising. Indeed structuralists do not like empathic readings, which analyze the emotions displayed in the text. And yet they too risk treating constructs of words as people. In postmodern novels characters lose many of their human traits: they blend into one another, they say they are inventions of a narrator or of the text, they disappear as suddenly as they appear. Structuralism hardly knows what to do with such nonanthropomorphic characters, which proves the extent of its remaining anthropomorphism.

2.3. Focalization

Contrary to characterization, focalization does belong to the crucial insights narrative theory owes to structuralism. The term refers to the relation between that which is focalized—the characters, actions, and objects offered to the reader—and the focalizer, the agent who perceives and who therefore determines what is presented to the reader. So, we are talking here about the relation between the object and the subject of perception. We avoid the verb “to see” on purpose, because all senses are involved. Perception, for that matter, can imply cognitive functions such as thought and judgment. “The Map” features the following clause: “I had to recognize that I occasionally traveled somewhere by train.” The I-character is the center of perception—in this case an act of recognition that is not visual and not even sensory but that rather pertains to thought. The riding master’s perception expressed in the following sentence from “Pegasian” seems related to emotion and attitude: “Now the riding master doesn’t feel like explaining anything anymore.”

The terms we will use in the following discussion, “focalizer” and “focalized object,” are problematic. They suggest there are in a narrative text centers of perception that approximate human beings and that apparently think and feel as we all do. One might ask in the first place whether a text actually contains such distinct centers and, second, whether it is useful to study them so anthropomorphically. Genette has avoided this problem by speaking consistently of focalization, without subject or object. Bal, on the other hand, who has refined the theory of the French narratologist, believes it is necessary to distinguish between a perceiving agent and a perceived object.79 Genette did not like her revision at all.80 However, the distinction between focalizer and focalized object has in the meantime been accepted, probably because it can help clarify the rather vague and monolithic concept of focalization.81

One of those clarifications has to do with (un)reliable perception, which can be described thanks to this distinction between a perceiving subject and its object. The relation between these two is crucial, for it allows the reader to gauge the information provided by the text. If a character is constantly seen through the eyes of a single focalizer, one may wonder whether this view is reliable. Is it really true that a woman is a flirt if you only see her through the eyes of her partner? Conversely, one character might be perceived by so many focalizers that the reader has too much information to be able to arrive at a coherent and reliable image.

We will discuss focalization using the three criteria that we will also refer to in our section on narration: types, characteristics, and textual indications allowing for the determination of these types and characteristics. Two questions must be answered in order to determine focalization types. The first concerns the position of focalizers with regard to the fictional universe. If the focalizers belong to it, they are internal; if they remain outside of it, they are external.82 Edgar Allan Poe’s story “Metzengerstein” provides a clear illustration of this distinction. A fire breaks out in the stables of the Berlifitzing family, which has been on bad terms with the Metzengersteins for ages. The reader sees the reaction of the young baron, Von Metzengerstein, through the eyes of an agent who is not in the room with the baron: “But during the tumult occasioned by this occurrence, the young nobleman himself sat apparently buried in meditation, in a vast and desolate upper apartment of the family palace of Metzengerstein.” One could imagine this as a scene caught by a camera on the shoulder of the narrator, who does not appear in the story. Next, however, “his eyes were turned unwittingly to the figure of an enormous, and unnaturally colored horse.” From then onward the reader sees the scene through the eyes of the young baron: “The horse itself, in the fore-ground of the design, stood motionless and statue-like.”83 This switch between external and internal focalization can be compared to the occasional image change during the live transmission of a Formula One race. Most of the time the camera is placed above or on the side of the circuit, but sometimes viewers find themselves inside the race because the images come from a small camera installed on one of the cars. Of course this focalization is not perfectly internal, since the viewer does not really see through the eyes of the driver.

Narrative texts with numerous levels complicate the relation between internal and external focalization. Let us return to Een weekend in Oostende, by Willem Brakman. The main narrative deals with Blok. When he is perceived by the narrator, who never appears as a character in the story, focalization is external. For instance, “In the evening, all spruced up, he pedaled on the borrowed bike to the birthday party.” Focalization becomes internal when Blok hears the waltz entitled “Gold und Silber” (Gold and silver), “which moved him to tears, because it made him think of everything at once.”84 Things get more complicated when Blok starts to tell a story about his Uncle Anton. In the beginning of this story the account is filtered through Blok’s perception, as for instance in, “On a beautiful summer night it was so hot and tepid that even the dead in the graveyard rapped on the lids and called out: ‘Please . . . just for a moment.’”85 This is external focalization with respect to the story about Uncle Anton since the reader’s information entirely depends on Blok, who does not appear as a character in the story he tells. With respect to the main narrative about Blok, however, this quotation amounts to internal focalization since the reader hears about everything through Blok, the protagonist of the main narrative. But when Uncle Anton’s perceptions start to infiltrate Blok’s story about him, focalization in this secondary story also becomes internal. For instance, “Uncle Anton came walking by with his calloused little hands; he was amazed to hear how her fair sex was emitting all this seductive language.”86

External and internal, therefore, must not be seen as absolute concepts, especially when the text features several embedded stories. Internal focalization on the level of the main narrative can become external on the level of a secondary narrative. Even when there is no embedding, the focalizer can be hard to determine. At the beginning of the novella Suikerpruimen (Sugarplums), by Huub Beurskens, the character Stein appears to be the (internal) focalizer, but certain passages suggest (external) focalization by the narrator. At one point Stein and Patty John are sitting on a restaurant terrace: “In between the private yachts and the small fishing boats, the dark water reflected the many little colored lamps.”87 Who sees this? Stein or the narrator? Impossible to decide.

The alternation between internal and external focalization is always present in narrative texts. It is also ideally suited to manipulating the reader, who often does not realize that information has been filtered through the perception of a character or the narrator. As a result, the reader might treat subjective information provided by a character as objective information coming from a detached narrator. This possibility is inherent in what, following Dorrit Cohn, we have called consonant psycho-narration, where the narrator adheres so closely to the character’s perceptions that it becomes difficult to distinguish between the two.

Nevertheless, the distinction is not to be neglected. Even if character and narrator coincide in a first-person text, there still exists a difference between internal and external focalization.88 If the narrating I considers something the experiencing I did, then there is external focalization if the scene is perceived by the narrating I and internal focalization if it is perceived by the experiencing I. Here is an example from Suikerpruimen: “I was annoyed at it, probably because of a kind of professional jealousy, I now think.”89 Since the experiencing I felt the annoyance, it is internally focalized; since the narrating I gives a reason for the annoyance, it is externally focalized.

The examples show that the choice of internal and external focalization does not depend on person. First-person narration can be focalized externally, while third-person narration can be focalized internally. “I was very arrogant at the time” is an example of external focalization, while “he considered her extremely arrogant” is focalized internally. The type of narration therefore must not be confused with the type of focalization.

Internal and external focalizers can either remain on the surface or penetrate the things they are perceiving. In a crime novel very often only the killer’s external characteristics appear, so that the reader has to search for the killer’s motivations. This adds to the suspense, and it enhances the reader’s eagerness to solve the mystery. When focalization penetrates a character, it results in the observation of emotions, cognitive functions, and psychological detail. These can either be perceived by a detached narrator (in which case focalization is external) or by a character (in which case it is internal). The potential combinations and alternations between the various types of focalization also enable an author to create and sustain suspense. If the first chapter of a novel contains description of a character’s thoughts and they include plans for a murder, he or she will appear as the most likely suspect for the murder committed in the second chapter. But in order to keep the reader guessing, the text may stick to the seemingly innocent exterior of the character in subsequent chapters, so that it becomes impossible to decide immediately whether he or she is really guilty.

As we mentioned before, types of focalization are determined on the basis of two criteria. The first concerns the focalizer’s position: external or internal. The second criterion has to do with stability. If the events of the story are perceived by a single agent, then Genette calls this fixed focalization. If the events are perceived by two or more characters who alternate, Genette speaks of variable focalization. If various agents of perception consecutively focus on the same event, Genette speaks of multiple focalization.90 A clear distinction between variable and multiple focalization requires a consideration of the perceived object. In the case of multiple focalization the object remains unchanged, for instance when the same event is first seen through the eyes of character A and later through the eyes of character B. In the case of variable focalization the focalized object changes with every new perception, for instance when event X is perceived by character A and event Y by character B. Since Genette does not distinguish between subject and object of focalization, he does not develop the clarification we offer here.

“Pegasian” features variable focalization. Sometimes the reader is guided by the female rider’s perceptions (“What are those flaps for, in fact?”), sometimes by the riding master (“And it wouldn’t hurt to consult a few books on cavalry”). “The Map” also has variable focalization, since the reader is made to look not only through the eyes of the boy (“The village I knew so well and which I had never seen on a map!”) but also through those of the narrating I (“I haven’t kept it either”). In the two stories the alternation of the perception center reflects a thematic confrontation between the unorthodox and naïve view on the one hand and the disciplined and adult view on the other. In Wasco’s “City” the reader follows as the protagonist walks about, but (maybe apart from panels two and three) the perspectives on the buildings and objects are not that of the protagonist. In the panel with the wired chair, for instance, the little figure is on a lower level than the chair and looking away from it. So, while most of “City” is determined by the activity of the protagonist, most if not all of the focalization is external and thus seems quite stable.

Showing the same event through a number of focalizers (multiple focalization) adds variety, but it often complicates the narrative. In Suikerpruimen, Patty John betrays her husband, Stein, with Ruben. Their first sexual encounter is initially presented through Ruben, and then through Patty John. Here is Ruben: “Before I went out with her into that urine-reeking Pigeon Alley, and she pushed me against the wall and stuck her tongue between my lips, in the humid August night, I had quickly relieved myself of some extra pressure.”91 And here is Patty John’s perspective: “It was a humid August night when, in a narrow alley that reeked of vomit, urine, pigeon shit, and, she imagined, horse chestnut blossoms, she let herself be opened, lifted, and rammed.”92 In Ruben’s perception Patty John is the more active person since she pushes him against the wall; in Patty John’s experience Ruben is more active than she is since she lets him open, lift, and ram her. The fact that they both feel taken advantage of reflects one of the problems in their relationship. Their initial passion rapidly deteriorates into passivity. They both feel misled since their partner has failed to deliver on his or her promises. These content-related aspects are underscored by the choice of multiple focalization, which shows how differently the lovers interpret their first sexual encounter.

The different types of focalization (internal versus external and stable versus unstable) can be specified with reference to a number of properties that Rimmon-Kenan prefers to call facets.93 The first two properties concern the focalizer’s spatiotemporal perception. In terms of space the focalizer can impose a panoramic, simultaneous, or limited view on the reader. In the case of a panoramic view the focalizer controls the entire space of the narrative. The beginning of Malcolm Lowry’s Under the Volcano adopts such a panoramic view: “Two mountain chains traverse the republic roughly from north to south, forming between them a number of valleys and plateaus. Overlooking one of these valleys, which is dominated by two volcanoes, lies, six thousand feet above sea level, the town of Quauhnahuac.”94

There is simultaneous focalization when the reader perceives what happens in different locations at the same time. Harry Mulisch, for instance, repeatedly illustrates his principle of “octavity” (the same and yet different) by showing divergent events taking place at the same moment. He suggests that these events resemble each other like a musical note and its octave. The simultaneity of events in different locations is traditionally represented by formulations such as “In the meantime” or “While this was going on.”

A less traditional way of showing simultaneity is the use of columns or text strips on top of one another, which results in a simultaneous presentation of different narratives. In Minuet, Louis Paul Boon places a collage of newspaper clippings at the top of the page. Below comes the “normal” narrative text. The first sentence of the newspaper section is as follows: “A farm laborer found a naked girl tied to a tree in a snow-covered field.” The cold suggested in this sentence provides a link with the first sentence of the regular text: “My work in the refrigerating chambers was rather monotonous: checking temperatures which had to remain at freezing point day and night.”95

With respect to space, next to panoramic and simultaneous focalization, there is also limited perception. This is the typical situation of a character since his or her perceptions are most often coupled with the limited space in which he or she moves. Rimmon-Kenan holds that the panoramic and simultaneous views are possible only for external focalizers, but we disagree.96 A character can imagine perfectly what happens elsewhere. Imagination forms part of the focalizer’s perception. Such panoramic views are therefore possible, not only owing to an actual position (from an airplane, for instance) but also owing to the character’s imagination.

Just like spatial perception, temporal perception can be divided into three types. Panchronic focalizers survey all time periods. They can look backward and forward. This prediction amounts to a flashforward. If the narrative only looks back, focalization is retrospective, as is the case in the typical autobiography where the narrating I considers the experiencing I. Finally, perception can take place simultaneously with the events, in which case there is synchronic focalization.

Obviously, the various temporal and spatial focalizations can alternate. In Sunken Red the narrating I remembers his childhood in the Japanese internment camp. In the following passage the first sentence is retrospective, while the second one is prospective within the retrospection: “Then it left me untouched. I was not to be touched by it until much later.” Then there is a passage with synchronic focalization: “I see the Jap beating a woman with a rattan cane.”97

Apart from time and space, psychological properties play an important role in the further description of focalization. With the term “psychology,” we mean the cognitive, emotional, and ideological aspects of perception.98 On the level of cognition there are focalizers who know everything and there are those whose knowledge is limited. In this context, omniscience is no longer directly related to the act of narration. Traditional omniscient narration could thus be redefined as a form of narration in which an omniscient agent is the focalizer.99 Normally this would be an external focalizer: the center of perception is occupied by a narrating agent outside the fictional universe. Characters can also pretend to be omniscient and to look in other people’s heads, but such passages will seem more speculative and less reliable than those informed by an external focalizer. In De Walsenkoning (The king of waltzes), by the Dutch author Louis Ferron, the main character, also called Louis Ferron, makes it seem as if he can look into his mother’s mind. He addresses her in his imagination on the occasion of her marriage: “You could already see the golden mountains he promised you.”100

Focalization manipulates the reader. By switching from an omniscient focalizer to a limited one, the reader can be kept in suspense. The beginning of the medieval narrative Karel ende Elegast is focalized through an omniscient agent from whom the reader learns that an angel tells Karel, the king, to leave his castle and start stealing. When Karel meets a black knight in the woods, the omniscient focalizer relinquishes his position to the king. Since readers are now limited to what Karel feels and perceives, they know just as little as Karel does about the identity of the black knight. The character’s fear and tension are transmitted to the reader. If an omniscient focalizer informed the reader that the black knight was the other central character, Elegast, the story would have been less exciting.

On the emotional level, focalization can be detached or empathic. The relation between focalizer and focalized object is crucial in this respect. If only the outside of the focalized object is perceived, focalization is detached. If, on the contrary, there is constant speculation about the thoughts and feelings of the focalized object, then perception is empathic. The above passage from De Walsenkoning provides an example of empathic focalization.

The ideology inherent in every form of perception can either be given explicitly or be implied in narrative. The way in which the external focalizer in “From a View to a Kill” perceives the non-Western criminals is telling. Their eyes are cold, their faces angular, and their language incomprehensible. The conservative “capitalist” ideology emanating from the Bond stories is reinforced by their internal focalizer, James Bond, who observes and judges the Soviets in exactly the same way as the external focalizer.101 In “Pegasian” the ideology is expressly stated at the end when the narrator or the female rider suggests that lightness is important, “as long as you take off.” However, this preference was already implicit in the first passages focalized through the female rider.

It is not always possible to establish a text’s ideology in an unambiguous way. The longer and the more complex a narrative, the more ambiguous the ideology usually becomes. In this respect much depends on the number of focalizers and their position. If the narrative works with one external focalizer, chances are high that the ideology will be relatively unequivocal. If, however, dozens of characters function as focalizers, the result is polyphonic ideology, and the reader will have a hard time reconstructing the dominant view. But even in the case of a single external focalizer, textual ideology may be hard to delineate. Thus Gerard Reve’s novels often have a fixed external focalizer, and yet his omnipresent irony makes it impossible to decide just how literally one should take the narrator’s statements about blue-collar workers, women, and migrants.

The early structuralists did not deal at length with the ideological analysis of focalization. That is hardly surprising, since this inquiry leads away from a discussion of form into a discussion of content. Attention to ideology is the most important recent shift in the study of focalization. In chapter 3 we will present new approaches that emphasize the ideological aspects of perception in narrative.

When establishing the types and properties of focalization, readers have many textual indications at their disposal. Descriptions of focalized objects or people may help a reader to decide, for instance, between an internal or an external focalizer. Suppose the wife of Judge Jack Jones enters her husband’s study. If the text reads, “Judge Jones looked moody,” one can attribute this perception to an external focalizer, since the woman probably would not think of her husband as “Judge Jack Jones.” If, however, the text reads, “Jack seemed moody again,” it is more likely that the woman is responsible for this perception. Terms of endearment provide an extreme example of elements pointing to perception by a character rather than by an external focalizer. More generally, Käte Hamburger shows that grammatical tools, such as deictic temporal adverbs, give a clear indication of the agent at the center of the narrative’s perception.102

Style too can provide indications of the focalizer. Childhood memories with many complicated and technical observations are probably externally focalized because a child would not achieve such intricacy. When Harry Mulisch discusses his childhood in Voer voor psychologen, he often uses a style that is not childish at all and that suggests perceptions have been filtered by the adult he has become. As a nine-year-old, Harry sees a puzzle, and there follows a very intellectual description of its top-right corner.103 One would be inclined to conclude that a child could never see the puzzle that way, but naturally the text might also suggest that nine-year-old Harry was a genius. This ambiguity shows that textual indications can help but do not necessarily lead to an unequivocal conclusion. Here too the reader plays an important role. Unfortunately for the structuralist project, texts seldom impose their structures.

Textual indications of focalization also include linguistic features such as register and the type of language. If a story is told in a neutral version of standard language and suddenly dialect and swear words appear, this can mean that the events are no longer perceived by a neutral (external) narrator but by an (internal) character.

A great number of words can suggest a distance between the perceiving and the narrating agent, and as such they indicate that there is no internal focalization. In first-person texts, time indications often have this function. In “Now I know what I did not even suspect then,” the “then” and “now” imply external focalization. Words of modality are also often used to distinguish external from internal focalization. In “It’s possible I thought at the time that everything would go very well,” the modal phrase “it’s possible” shows that these are the thoughts of the narrating I about the experiencing I, which implies external focalization.

The list of textual indications can be endlessly extended, but this is not unproblematic. In principle, structuralist narratology wants to separate focalizer from narrator as strictly as possible. This becomes difficult if the particularities of narration are considered to be indications of focalization. If word choice, for instance, is related to worldview, the boundary between narration and focalization may become fuzzy.104

3. Narration

Narration forms the third and least abstract level of structuralist narratology. It is concerned with formulation—the entire set of ways in which a story is actually told. While the story is not visible in the text, narration involves the concrete sentences and words offered to the reader. While narrative was mostly concerned with the perception of events, narration mostly deals with the way in which these events are worded. Attention goes to the narrating voice, to speech instead of perception, to narration instead of focalization. This implies two central areas of investigation: first, narrating (including the narrating agents) and second, the way in which these agents present a character’s consciousness. These two concerns can be summarized as follows:

3.1. Narrating

Similar to focalization, narration also expresses a relationship between an active subject and a passive object. In this case the relationship is the one between the narrator and that which is narrated. Again similar to focalization, this relationship brings about different kinds of narration that can be further described with the help of several properties. Although Genette avoids the word narrator and mostly replaces it with less personal concepts such as narrating instance, one cannot deny that here again structuralism catches a textual aspect in human terms.105 Most narratologists use the word narrator, and we will do so too, since the use of less anthropomorphic terms such as narrating instance does not prevent this instance from being characterized by means of such anthropomorphic criteria as “reliability” and “detachment.” This has led to a great deal of criticism.106 However, in many recent contributions anthropomorphism is no longer regarded as a process to be avoided. In any case the distinction between various narrator types remains relevant for narrative analysis.

The narrator type depends on the relationship between the narrator and what is being narrated. The first criterion here concerns the relationship between the level of the narrator and the level on which the events being narrated take place. A narrator hovering over the narrated world is extradiegetic. An intradiegetic narrator, by contrast, belongs to the narrated world and is therefore below another narrating agency. If a character is presented by a narrator independent of any other narrating agent hovering above, this narrator is extradiegetic. If the character in question starts to tell a story, that individual becomes an intradiegetic narrator. The difference between the two is a hierarchical one. The extradiegetic narrator occupies the highest place in the hierarchy, while the intradiegetic narrator sits one step below. In order to make the distinction, one simply has to answer the following question: “Is this narrating agent narrated by another narrating agent or not?”

Although extradiegetic narrators occupy the highest level, this does not automatically mean that they are the most important narrating agent in any text. Turgenev’s short fictional work “Asya” begins as follows: “I was then about twenty-five (N. N. began)—as you can see, these matters belong to years long past.”107 The extradiegetic narrator is the one who says “N. N. began.” In the rest of the text, this narrator does not appear again, and the reader always hears the intradiegetic narrator, the man who is described with the letters “N. N.” The hierarchically lower narrator is much more important than the higher-level colleague.

The extradiegetic or intradiegetic narrator mostly addresses an extradiegetic or intradiegetic audience, which in chapter 1 we called the narratee. The extradiegetic narrator mostly speaks to an extradiegetic narratee. These so-called addresses to the reader do not involve the empirical reader at all but rather an agent who does not appear in the story and yet functions as the narratee. In “The Cask of Amontillado,” by Poe, the extradiegetic narrator addresses such an audience: “The thousand injuries of Fortunato I had borne as I best could; but when he ventured upon insult, I vowed revenge. You, who so well know the nature of my soul, will not suppose, however, that I gave utterance to a threat.”108 Intradiegetic narrators mostly address other intradiegetic agents, that is, other characters.

But, as we have already suggested in chapter 1, cross bonds are possible. An intradiegetic narrator can speak to a higher agent who occupies a position outside the narrative world. A character can, for instance, complain about the narrator of a story. In the Flemish modernist classic Chapel Road (originally published as De Kapellekensbaan), by Louis Paul Boon, a male character relates how a female character has complained about “boontje,” the narrator of their story: “She talked about the chapel road book, from which she’s been removed, she says.” She also thinks “that we’re neglecting too many of our heroes” and concludes, “You’re a useless writer if ever there was one.”109 Conversely, an agent outside the fictional world can speak to an internal agent. For instance, an extradiegetic narrator can address the protagonists. In Het ware leven, een roman (Real life, a novel), the novel’s author, Ilja Leonard Pfeijffer, figures as the extradiegetic narrator. In one of the last chapters, he has an official meeting with some of his characters. In a managerial tone they discuss what should be done to keep the reader interested and to develop the existing plot lines. One of the participants wonders if they are not acting as “plot-spoilers? I mean, the reader now already knows what the outcome will be like.”110 Structuralism considers both cases as narrative transgressions, for which Genette has coined the term narrative metalepsis.111 Although structuralists recognize the existence of such transgressions, they are keen on establishing and maintaining the boundaries.

The distinction between extra- and intradiegetic narrators causes a number of problems, of which we will discuss only two. First, one could ask whether any character who starts speaking automatically becomes an intradiegetic narrator. When a character says, “Yes, I sure do,” it appears irrelevant to analyze this statement as intradiegetic narration. It is of course an intradiegetic statement, that is, a statement by an agent within the fictional world, but it does not really amount to a story. This brings us back to a problem we considered at the beginning of this handbook: How can one define a story? And what is the most basic form of a story? Which minimal requirements must a stretch of text meet in order to qualify as a story? There is no generally accepted definition of a minimal story yet, and it will probably never materialize. In chapter 1 we defined a story as a sequence of events that readers connect in a way they consider meaningful. Needless to say, that which is meaningful for one reader does not have to be so for another. Fortunately, the problem of the minimal story is not crucial to the distinction between extradiegetic and intradiegetic passages. Indeed it does not affect the distinction between the two levels; the question is simply whether an intradiegetic passage should be considered a story.

Second, and more important, a problem arises when various levels and stories are embedded in a frame narrative. When the embedded story mirrors or summarizes the story on the higher level, this leads to the so-called mise en abyme. To begin with, this causes a terminological problem. Imagine the abovementioned intradiegetic narrator N. N. talking about a man identified as O. O., who in N. N.’s story starts telling his own story, say, about P. P., who in turn tells a story about someone else, and so on. How would we describe all these narrators? The narrator who starts talking in intradiegetic narration (in our example, O. O.) can be called an intra-intradiegetic narrator. Rimmon-Kenan calls this narrator “hypodiegetic,” while Genette uses “metadiegetic.”112 The latter term is especially confusing since it suggests that a narrator who stands lower on the hierarchical ladder (and therefore sits “deeper” in the narrative) in fact stands “above” the ladder (and must therefore be placed higher). To avoid confusion, we prefer to distinguish between a narrator on the first level (extradiegetic), a narrator on the second level (intradiegetic), a narrator on the third level (intra-intradiegetic), and so on.

With respect to embedded stories, it is not just the terms that are confusing. Sometimes it is also hard to maintain the hierarchy. Certain texts with embedded stories reverse the whole hierarchy on their deepest level and make it seem as if the supposedly highest level is actually narrated from what was thought of as the lowest. In our example the character P. P. would then be the agent who said or wrote, “N. N. began.” If the stories by N. N., O. O., and P. P. mirror each other in this way, the result is a paradoxical form of mise en abyme, which Lucien Dällenbach describes as “the aporetic duplication (a sequence that is supposed to enclose the work that encloses it).”113 The paradox or aporia resides in the fact that the deepest level would contain the highest. Such embeddings undermine the structuralist effort to place all levels in a clear vertical hierarchy.

Again, Een weekend in Oostende provides an example. The first or highest story is the one about Blok. It is told by an extradiegetic narrator who never appears in it. The second or lower story is told by Blok and concerns his uncle, Anton. Here Blok is a second-level or intradiegetic narrator. He relates how Uncle Anton met a prostitute, and they were going to get married. Then Uncle Anton starts to tell a story as an intra-intradiegetic or third-level narrator. His story deals with a riveter. At the end of this embedding the text returns to the highest level, at which we find Blok’s conversation partner, Uncle Julius. He summarizes—for a colonel who has just arrived—the story told by Blok. He says, “And he told me in his turn how his Uncle Anton turned into a poet when he stood in front of a floozy, but that he had to let that talent decline because of the circumstances, since he lived in a dark street with, if I understood well, a dairyman, a few greengrocers, and a garage manager.”114 This sentence upsets the text’s hierarchy of levels in two ways. First of all, it is Blok and not Uncle Anton who lives in the dark street, which means Uncle Julius conflates the second- and third-level narrators. Second, the formulation “if I understood well” creates a paradox, since Blok has never told Julius anything about the dark street. Julius apparently knows things only the reader can know. In other words, an agent inside the story knows things that have been told to agents outside it. This confusion of internal and external has to do with narratees since the external narratee falls in with the internal narratee. Structuralist categorization fails in the face of texts such as this one by Brakman.

Apart from the difference between intra- and extradiegetic, there is a second distinction on the basis of which one can establish narrator types. This distinction no longer concerns the hierarchy of levels but rather the narrator’s involvement in what is narrated. The narrator either has experienced the narrated events, and is thus homodiegetic, or has not, and is thus heterodiegetic. If the narrator’s experience is personal, the degree of involvement may vary. Perhaps this narrator has only seen things from afar or, on the contrary, played the central role in the proceedings. On this sliding scale from marginal to central involvement, one can place the traditional distinction between witness and main character. If the homodiegetic narrator is the protagonist of the story told (such as Pip in Great Expectations, by Charles Dickens), Genette applies the label “autodiegetic.”115 The prototype here is the autobiographical narrator. Genette has no separate term for narrators who deal with things they have only witnessed. We accept Van der Voort’s proposal for this situation and use the term “allodiegetic.”116

When combined, level and involvement result in six types. First of all, there is the extradiegetic and heterodiegetic narrator, probably the most classical one: the narrator who hovers above the story—and thus is not narrated by another agent—and deals with things not personally experienced. This narrator can be inconspicuous if narrating exclusively in the third person but can occasionally appear in the first person. Poe’s “Masque of the Red Death,” which we have referred to in chapter 1, provides an example of the latter. The I-narrator is not narrated by another agent and has not experienced the terrible events being related. “Pegasian” too presents an example of an extradiegetic heterodiegetic narrator, this time without the use of the first person.

Second, there is the extradiegetic and autodiegetic narrator, who stands above the events narrated, but this personage has experienced them. More precisely, this narrator was the central character. An example of this type is the I-narrator in “The Map.” The narrating I stands at the top of the hierarchy and tells a story in which he played the central role as a child. If he were to give us his father or mother as the main character, he would become a witness, and such an extradiegetic narrator would be allodiegetic. A famous example of this is Dr. Watson in the Sherlock Holmes tales. He has no narrator above him (so he is extradiegetic), but he is most often a mere witness of the things he relates (which makes him allodiegetic).

For the intradiegetic narrator, the same three possibilities apply. A character can relate things not personally experienced and thus becomes an intradiegetic and heterodiegetic narrator. Such narrators can also give us events they have witnessed, in which case they become allodiegetic, or events in which they played the central role, in which case these narrators become autodiegetic.

There are many advantages to this systematization of narrators, which we prefer to Stanzel’s circle. To begin with, Genette avoids the confusion of Stanzel’s system. Whereas Stanzel conflated speaking and perceiving agents by combining reflectors and narrators into one scheme, Genette’s six types are all speaking agents. Genette specifies Stanzel’s reflector as a combination of a particular narrator with a particular focalizer. An example of a reflector in terms of Genette would be an extradiegetic narrator who shows everything through the perception of a character and remains in the background while doing so. This means that for Genette there can be no scale beginning with “reflector” and ending with “teller-character,” because that would mean one starts with focalizer and ends with narrator—two agents that for Genette belong to different levels of the text. They cannot be shown on a sliding scale.

Stanzel’s person scale resembles Genette’s degrees of involvement. The “identity” Stanzel spots in I-narration corresponds to the homodiegetic narrator, while nonidentity corresponds to the heterodiegetic narrator. Stanzel’s scale of perspective is for Genette not a question of narration but of focalization. Stanzel’s dual internal and external perspective amounts to Genette’s internal and external focalization, respectively.117 This comparison proves to what extent Stanzel’s circle confines to the same plane elements that for structuralists belong to different planes or levels. As a tool for narrative analysis, Genette’s system is more transparent.

And yet this transparency, as we already have had to establish more than once, derives from a theoretical construct that is not always borne out by actual narrative texts. Concrete texts are often more complicated than theories and do not always easily submit to classification. Een weekend in Oostende has already shown that the distinction between intradiegetic and extradiegetic narrators is occasionally far from clear. The same novel also proves that the distinction between hetero- and homodiegetic narrators is sometimes impossible to maintain. Uncle Anton tells a story about a riveter. It resembles a traditional heterodiegetic story since Anton relates things he himself has not experienced. But as the riveter is dying, he hears the voice of his wife, who asks, “Are the lights out, Anton? . . . and the gas turned off? . . . [you] closed the upper window?” And then the text reads, “The riveter wanted to nod obediently,” which suggests he is also called Anton.118 Maybe the riveter is Uncle Anton, in which case he would be a homodiegetic narrator. This is impossible to decide, so it is up to the reader to settle the matter. A traditional reader is likely to say Anton is a heterodiegetic narrator who happens to have the same name as his hero. Such a reader would argue that narrators can never describe their own death. A reader versed in postmodernism might either consider Anton a homodiegetic narrator or leave the question open altogether.

One can further specify the various narrator types on the basis of three properties. First of all there is the temporal relation between the moment of narration and the moment at which the narrated events take place. Here Genette discerns four options.119 The most traditional one is that of subsequent narration (narration after the events), of which “The Map” provides an example. Although the past tense of the verb is most common here, subsequent narration can also occur in the present tense. A sentence such as “I am fifteen and I think everything still has to happen” is a seamless fit for a subsequent narration by a fifty-year-old. Often in this connection the use of the present tense has a special meaning. In “Zelfportret met tulband” (Self-portrait with turban), Harry Mulisch uses the present tense for nine crucial events from his past.120 He calls each one of them a “today” because it makes itself felt until the moment of narration. This continuity has its symbolic expression in the use of the present tense for events in the past.

The second temporal option is one that involves prediction, which Genette calls prior narration. For instance, a character can narrate how someone else will end up. Prediction can be expressed with the help of the present or the future tense or a combination of both, as in, “You will see. In seven weeks you will be a wreck. You don’t have a job anymore then, or a wife. You drink all day and you think things can’t get worse.”

Simultaneous narration, the third temporal type, requires the use of the present tense because only that allows the perfect coincidence of action and narration. Genette mentions as examples both the nouveau roman and directly quoted monologue. The narrators in these cases wish to create the impression they are telling you everything the moment it happens. Obviously this is only a trick—if the narration were really to coincide with the action, the narrator would be talking and experiencing at the same time. “Pegasian” amounts to simultaneous narration.

Finally, there is interpolated narration. For instance, in a novel action can be alternated with a letter that provides a comment on it. In such a case there is always more than one narrative level. An epistolary novel has the story told in the letters (of the letter-writing characters) and (at the other level) the story told about these characters between those letters. An example of such interspersed narration can be found in chapter 18 of Max Havelaar, in which Havelaar’s letters “To the Bantam Resident” constantly interrupt the action.

Next to time, visibility is the narrator’s second property, which can be represented on a sliding scale from a nearly invisible narrator to one that is extremely visible. Rimmon-Kenan speaks of covert and overt narrators.121 The difference resides mainly in their narrative procedures: covert narrators quote a lot, do not present themselves in the first person, and try to avoid evaluative descriptions as much as possible. Overt narrators resort to paraphrase instead of quotation; they will definitely talk about themselves and therefore use the first person and will often showcase their own opinions. “Pegasian” has a covert narrator and “The Map” an overt one.

At this point we feel we should repeat our earlier remark that visibility must not be confused with presence. Invisible narrators remain present, and visible narrators do not have to play a role in the story they tell. In other words, they do not have to be homodiegetic. They can talk perfectly well about things they have not witnessed, as for instance in Poe’s “Masque of the Red Death.”

The narrator’s third property concerns reliability, which can also be represented on a sliding scale from a completely reliable narrator on one end to an entirely unreliable one on the other. As we have already said in chapter 1, a method does not exist to establish reliability in any objective way. Of course a text can contain many signs of (un)reliability. If narrators maintain they have said something when they have not in fact done so, that can be seen as a sign of unreliability. This impression will be enhanced by contradictory statements on the part of the narrator and by confessions of confusion and failure to see connections. If at the beginning of a story a narrator suggests greater skill in imagining things than giving a precise account, that also does not make the narrator more reliable.

Still, the reader might interpret a narrator’s utterances in unexpected ways. The above examples of supposed unreliability may be felt to contain a suggestion that a correct and truthful account is necessarily mendacious. In this view a traditionally coherent story is not as reliable as it may purport to be. Textual indications of reliability, such as internal coherence, are therefore not sufficient to decide the matter.

Narrator type does not provide a solution either. Perhaps extradiegetic and heterodiegetic narrators are more often reliable than their intradiegetic and homodiegetic counterparts if only because the former are more detached and can therefore be more objective. But this is not at all a general rule. Quite a few postmodern narrators are extradiegetic and heterodiegetic, but this does not prevent them from being totally unreliable. Other narrator properties such as invisibility or temporal distance do not guarantee reliability either. In chapter 1 we showed that an intimate link between narrator and implied author does not suffice to test reliability, partly because the implied author is an entity constructed by the reader rather than one that can be mechanically derived from the text. All of these arguments lead us to conclude that the decision concerning (un)reliability largely lies with the reader.122

The first two properties of the narrator—position in time and visibility—are relatively technical characteristics of the text. With reliability one enters the anthropomorphic domain and turns the narrator into a human agent. Structuralists do not want to venture too far in this direction, although, as we have seen, they often go much further than they might want to. However, readers and critics who see the text as a message that is part of a communicative exchange between a sender and a receiver can develop such human features. One example of this is the so-called status of the narrator, a concept developed by Susan Lanser. Her paradigm is speech act theory, which means she sees a text as a way to create a reality through language. This creation depends to a large extent on the authority of the speaker, in this case the narrator.

The narrator’s status has to do with “the authority, competence and credibility which the communicator is conventionally and personally allowed.”123 In practice this comes down to the combination of “diegetic authority” (which narrators possess on the basis of their personality) and “mimetic authority” (which they develop through their style of narration). The former type of authority comprises social identity, which for Lanser includes elements such as “profession, gender, nationality, marital situation, sexual preference, education, race, and socioeconomic class.”124 The most common social identity, according to Lanser, is that of a white heterosexual middle-class man, but this leads us to the ideological aspects we will discuss in chapter 3 of this handbook. Mimetic authority consists of three elements that must be conveyed by style: honesty, reliability, and competence. Narrators can lie and talk about things they do not really know, or they can be honest and well informed. The problem with this list is, first of all, that it can be endlessly extended.125 The second problem is that the characteristics in question are often very difficult to extract from the text. How does one determine the social identity of an extradiegetic and heterodiegetic narrator who never comes to the fore?

Perhaps these characteristics have to do instead with the unspoken prejudices of the reader (who might for instance expect the narrator to be white and male) than with concrete textual features of the narrator. Strictly speaking, this discussion takes place outside the structuralist treatment of the narrator, but one could also say that it exposes structuralists’ blindness. Indeed their treatment is not devoid of ideological prejudice either, and furthermore it displays similar shortcomings: the structuralist description of the narrating agent is sometimes anthropomorphic, it can be endlessly extended, and it does not always follow logically from the wording of the story.

Nevertheless, for a description of the narrating agent readers can definitely let themselves be guided by various textual indications. However inconspicuous narrators may be, there will always be traces of their presence. Each description is the narrator’s and betrays that narrating figure’s ways of formulation. If one analyzes setting on the level of story and spatiotemporal focalization on the level of narrative, it is important to check how exactly the narrator describes and formulates this setting and focalization. The narrator of a James Bond story shines through in the many evaluative descriptions of the woods or the city or the secret service headquarters. In “City,” by Wasco, the general opposition on the level of story between nature and the city does not entirely determine the latter’s specific presentation in the separate panels. As the external focalizer, the visual narrator indulges in special angles all through the page (and in a panoramic view at the end of it).126 However, the unique style in which the narrator renders the various buildings and objects also provides us with a key to the somber vision of the future. The bleak city is angular and strange to the point of alienation.

Furthermore, narrators can be discerned on the basis of elements that are not immediately visible on the levels of story and of narrative. Things that have not happened, and that therefore do not belong to the story, can surface in narration because, for instance, the narrator might assign a certain importance to something that could have happened but ultimately did not. Things that characters are unaware of in narrative can also assume importance in narration, and yet again they betray the presence of the narrating agent. They also show that there is indeed a difference between the level of narration and the level of narrative.

In our discussion of the various narrative situations, we have suggested more than once that there are hardly any textual elements leading directly to an unambiguous definition of the narrating agent. One has to combine and interpret an ever-growing number of indications. First person, for instance, does not automatically mean that the text features an intradiegetic narrator. Extradiegetic narrators too can come to the fore in the first person. The choice between first or third person is not even decisive for the distinction between homodiegetic and heterodiegetic narration. If narrators are talking about themselves in the third person, they are still homodiegetic. In order to describe the narrative situation, one has to consider the entire text with all its embedded stories and combine the numerous relevant elements into a coherent whole.

It is also true that properties of narration can seldom be directly deduced from a textual indication. The use of the present tense does not automatically imply simultaneous narration. Visibility, however, seems closely connected with textual elements such as style and word choice since they betray, and make visible, the narrator. An insistently present narrator usually implies a lot of evaluative adjectives and adverbs, I-narration or we-narration, and a lot of addresses to the narratee. Reliability is much harder to connect with textual elements such as these, and that holds all the more for the status of narrators who keep a low profile. In such cases the narrators’ words betray hardly anything about the status of whoever is doing the narrating.

3.2. Consciousness Representation

Every bit as essential as the narrating agent is the representation of consciousness, that is, the way in which the narrator renders the consciousness of the characters. Of course, he or she might be one of those characters. Rimmon-Kenan refers to this as “speech representation” because it concerns the way in which the words and/or thoughts of the characters show up in narration.127 It might also be called the grammar of narration since it involves the search for the grammatical means used by a narrator to represent what is said and/or thought.

As we showed in chapter 1, there are basically two grammatical means: direct mimetic representation and indirect diegetic representation. The first is also called “scene” or “showing,” while the second is often referred to as “summary” or “telling.” The grammatical procedures used to create these two kinds of consciousness representation are direct and indirect speech. We have already discussed free indirect speech as an intermediary form. Dorrit Cohn interprets the three grammatical procedures in the widest possible sense, seeing them as vehicles for three kinds of consciousness representation: psycho-narration corresponds to indirect speech, quoted monologue to direct speech, and narrated monologue to free indirect speech.

An example of all three of these forms can be found in the previously discussed work Suikerpruimen, by Huub Beurskens. One night Patty John is nosing about in the papers of her friend Ruben, “just out of a mixture of boredom and innocent curiosity and a little bit in the hope of finding some academy work because, willfully, she still wanted to see what Ruben could really do.”128 This is psycho-narration: it uses indirect speech in the broad sense. It does not represent the character’s thoughts word for word but paraphrases them. Patty John thinks she has found proof in his papers of the fact that Ruben murdered her ex-husband, Stein. She also thinks he murdered his former girlfriend, Fanny. In order to represent Patty John’s thoughts, the text shifts to free indirect speech: “But he already has the death of someone else, of that girl, on his conscience, no? And the coldness with which he had apparently pushed her away, wasn’t that the coldness with which he had blinded Patty John to that ‘loser,’ as he had characterized him offhand? But Patty John, didn’t you yourself drop Stein just as mercilessly and ruthlessly as Ruben did Fanny!? Yes, yes, you’re guilty yourself.” The first two sentences are an example of narrated monologue: they both use free indirect speech. The last two might as well be direct speech. In that case, there are two possibilities: either Patty John is talking to herself in quoted monologue or the narrator is talking to the character, which would be an instance of narrative metalepsis. The first is more likely because the subsequent sentences are written in direct discourse and use quoted monologue to show Patty John talking to herself: “He [Ruben] has blinded me to Stein’s love and to my own love for Stein. Because I did love Stein.”129

As this example shows, the triad is not perfect. To begin with, the three grammatical procedures have to be interpreted “in the broadest sense,” which inevitably leads to vagueness. What is more, some sentences cannot unambiguously be put in a single category. The intermediary forms have been mapped by Brian McHale, who developed a sliding scale that still serves as a reference.130 It differentiates the two original poles of diegesis and mimesis into seven kinds of consciousness representation, ranging from the most diegetic to the most mimetic.

First, there is the diegetic summary. This is the most rudimentary representation of thoughts or utterances. The narrator says that a character said or thought something but does not say what was said or thought. In fact one hears only the voice of the narrator. “He talked all night, until his wife fell asleep exhausted” is an example of this kind of representation.

When the summary does show some of the content without representing it faithfully, McHale calls it less purely diegetic, as in, “He talked all night about the war and his heroics.” Another step in the direction of so-called accurate representation is the indirect content paraphrase, which represents the thoughts or utterances faithfully as far as content is concerned but not in terms of style. “He talked about how he had saved a lot of people in hiding during the war” may be a correct representation of the content of an original sentence like “Hell, I saved so many people in hi . . . hiding, so many people in hiding during that . . . that damned war.” The first passage from Suikerpruimen could also serve as an example of indirect content paraphrase. “She still wanted to see what Ruben could really do” might correctly represent the content of something like “I wonder what Ruben can really do.”

Of course as readers we often cannot possibly know what the original sentence or thought was, so we also cannot tell what kind of summary we are dealing with. The example of people in hiding could be a less purely diegetic summary if the character originally had a lot more to say about these people. It could also be a content paraphrase, but it might represent the original style quite faithfully as well if the original sentence went something like, “I saved a lot of people in hiding during the war.” When style and content are represented accurately in indirect discourse, McHale calls it indirect discourse, mimetic to some degree, which we will call semimimetic indirect discourse. “He said that, hell, he had saved so many people in hi . . . hiding, so many people in hiding during that . . . that damned war” might be an example. Thus, the simple sentence “He said that he had saved a lot of people in hiding during the war” may belong to three different types of consciousness representation: less purely diegetic summary, content paraphrase, or semimimetic indirect discourse.

Once again, textual elements turn out to be noncoercive and can be read in different ways. Consciousness representation is therefore not just the work of a narrator representing the consciousness of a character but also—and often more importantly—the work of a reader trying to imagine the original version of a represented thought or utterance. In “The Map” we read, “What excited me was the thought that it now made sense to have been everywhere.” If the reader imagines this to be the representation of a thought such as “it now made sense to have been everywhere,” this could be a case of semimimetic indirect discourse. But maybe the young first-person narrator was thinking a lot more carelessly and incoherently at the time, so perhaps the style has not been represented accurately. In that case we are dealing with a content paraphrase. It is also possible that the boy had a lot more going through his mind than the thought represented in that single sentence, which would make it a less purely diegetic summary.

These four initial kinds of consciousness representation are all variations of indirect discourse, which apparently can be interpreted in a very broad sense. The fifth type on McHale’s scale is free indirect discourse. As we have seen, this variation occurs regularly in “Pegasian.” An utterance by the riding master is represented as follows: “Little girls who have never personally experienced this heavenly sensation did well not to shoot off their mouths. And it wouldn’t hurt to consult a few books on cavalry.” This last sentence might in fact be a direct quotation.

Direct quotation is the sixth step toward faithful representation, that is, toward more mimesis of the character and less summary by the narrator. According to McHale, there is a seventh possibility, which represents thoughts or utterances even more accurately: free direct discourse, which differs from ordinary direct discourse in that digressions and supposedly irrelevant jumps in discourse and thought are also represented. The typical form is the quoted first-person monologue, which naturally leads us to Joyce’s Ulysses again. Leopold Bloom is looking for a bar of soap in his pockets: “I am looking for that. Yes, that. Try all pockets. Handker. Freeman. Where did I? Ah, yes. Trousers. Purse. Potato. Where did I?”131

Although free direct discourse emphatically seeks to create the impression that it represents a character’s consciousness virtually directly, it is of course “just” a convention. The “real” thoughts are as irretrievable to the reader as the “original” utterances we have just mentioned.132 This constitutes an important problem inherent in consciousness representation. The term itself suggests that there are two levels and two phases: first there is consciousness and then its representation within a narrative. In typically structuralist terms, consciousness is considered the deep structure, while its representation is the superficial manifestation of that structure. But what we have said about the story as a so-called foundation also goes for consciousness: it is an abstract and hypothetical construct that often remains irretrievable. There is no way to ascertain what Bloom or Patty John were “really” thinking.

This is why recent narratological approaches to consciousness representation abandon the mimetic conception in favor of a constructive one. The former maintains the sense of an original, real reality (the words and thoughts of the character), represented as faithfully as possible after the fact. By contrast, the latter approach sees this so-called real reality as an illusion produced by the passages offering consciousness representation. Making use of a number of conventions, these passages create the impression of being an accurate reflection. This so-called reproduction is in fact a production. Monika Fludernik puts it as follows: “Reproduction is a process of evocation [ . . . ]. Mimeticism in representation is an effect, a fiction of authenticity.”133 From this perspective, the focus should be primarily on the strategies and conventions that give the reader the impression that this production is in fact the faithful reproduction of a so-called real reality.

The crucial—and paradoxical—concept that Fludernik uses to refer to these strategies is typification.134 In order to give the reader the sense that a representation is true to life, the narrator uses a number of typical, clichéd turns of phrase and stylistic means that are supposedly inherent in oral language (which supposedly has to be faithfully represented): swearing, sighs, derailing syntax, banalities, repetitions, and so on. In the sentence “He said that, hell, he had saved so many people in hi . . . hiding, so many people in hiding during that . . . that damned war,” the swearing and the hesitation create an impression of exact representation. There is a paradox here: on the one hand these techniques create the illusion of verisimilitude, while on the other they are so conventional and stereotypical that they inevitably impoverish and distort the concrete reality (the thoughts and words of the character). The reader recognizes the clichés and accepts them as a warranty of authenticity, while in fact they are fakes. The reader only acknowledges the authenticity of a representation in the shape of a forgery.

That acknowledgment in the end depends on the frame in which the conventions of typification operate. One important example of such a frame is genre: a newspaper report requires a different typification than a postmodern encyclopedic novel. But frames can also be wider and refer to a whole set of social and cultural conventions activated through typification. When a narrator in a story about a restaurant quotes a waiter’s reply, the frame of the setting evokes certain expectations in the reader that determine whether the reply will be recognized as believable or not.135 We will look more closely at the use of frames in narrative theory in chapter 3 of this handbook.

In addition to typification, the narrator has other means at his or her disposal to make the reader believe that a representation is accurate and true to life. The comparison of different versions of an account is probably the most common example. At the beginning of Suikerpruimen the narrator seemingly quotes Patty John, who reproaches Stein for his predictability: “Never anything truly surprising or even nasty, for all I care; you’re so mortally dependable.” The sentence is in quotation marks, making it seem reliable, but the exactness of its reproduction is cast into doubt in the very next sentence: “She probably didn’t say it that way, but to Stein it could have been put in those words and still could.”136 The quoting agent is therefore not the extradiegetic narrator but the character Stein. At the end of the novella the quotation crops up again, this time in a fragment focalized by Patty John. It turns out Stein’s reproduction was very accurate: “Never anything truly surprising or even nasty, for all I care, Stein; you’re so mortally dependable.”137 The identity of the quoting agent in this case is not entirely clear. It is probably the extradiegetic narrator or maybe Patty John recalling her own words verbatim. In any case, the strategy is obvious: the repetition of the quotation is meant to convince the reader that the consciousness representation is very precise and accurate.

It is no coincidence that this example involves direct rather than indirect discourse. Since utterances are usually ordered quite carefully, quoting them may be perceived as more truthful and convincing than the putatively exact reproduction of often disorderly streams of thought. Perhaps readers tend to think of a quoted utterance as an authentic representation, even though they do not believe that a narrator could ever be capable of representing the chaotic swirl of thought in language. More generally, forms of representation at the diegetic end of the spectrum seem more dubious than those at the mimetic end because diegetic representations are only rough paraphrases. The intervention of the narrator is so strong that readers are not inclined to accept these representations as exact reproductions of a character’s consciousness.

This brings us to another problem of consciousness representation: the relationship between narrator and character. In the case of the diegetic forms—the first four on McHale’s sliding scale—it is impossible to determine to what extent the words of the summarizing narrator are a faithful copy of the words and thoughts of the character. After all, we do not have access to the so-called original.138 The one thing that is certain, however, is the identity of the narrating agent: the narrator is talking here, not the character. From free indirect discourse onward—McHale’s final three forms—the problem is reversed. We get a better view of the “original.” In theory, then, we should be better able to see what the character “really” thought or said. But in the case of free indirect and free direct discourse, the narrating agent becomes a major problem. Who speaks the words we hear in these two forms of consciousness representation?

Most traditional answers to this question assume that two agents are speaking at the same time: the narrator and the character—a dual voice, in other words.139 Bakhtin expands this notion to a polyphony of voices, which he considers not so much the product of anthropomorphic centers such as characters and narrators but rather a combination of various discourses. This creates a hybrid language that rules out any unambiguous identification of a single speaker or discourse.140 Ann Banfield takes this theory one step further and argues that there is no speaking center in free indirect discourse. Sentences with this kind of discourse are “unspeakable”: they are not uttered by any speaker but are indicated and constructed on the basis of a number of syntactic signals, such as inversion (“Would he still love her tomorrow?”) and a shift in subject (the “he” in the example was originally an “I”).141 Free indirect discourse thus becomes a mechanism of language, a grammatical process that Banfield studies from an abstract, Chomskyan perspective; it is no longer dependent on concrete and clearly identifiable centers.

Fludernik’s typification is paradoxical in this connection as well. On the one hand, typical turns of phrase make the representation impersonal—in that sense “unspeakable”—while at the same time they typify the speech of speaking characters and/or the reporting narrator. Swearing and hesitation are both typical of a swearing stutterer (the character, let’s say, not the narrator) and of the linguistic frame to which the representation belongs. Insofar as they typify the stutterer, they suggest that the narrator is letting the character do the talking; to the extent that they are stereotypical mechanisms of representation, however, they imply that the narrator is mostly letting linguistic conventions take over. In this way language, personality, and impersonality come together in typification.142 The question of who is speaking opens the way for an investigation into the typical turns of phrase that language enforces, enabling and at the same time distorting subjective expression. According to Fludernik, this means that the eternal problem of free indirect discourse (who is speaking?) is not of vital importance to narratology.

To a structuralist narratology, however, these kinds of solutions are not acceptable. Hybrid forms of language, utterances without speakers, and impersonal expressions of personality go against the structuralist predilection for distinct forms and unambiguously definable centers. Still, that preference is not very realistic, at least in this context. As we mentioned before in our discussion of Dorrit Cohn’s views, free indirect discourse does not always allow us to separate the words of a narrator from those of a character. It is especially difficult in the case of a first-person narrator, in which free indirect discourse often makes it impossible to distinguish between the narrating I and the experiencing I. In free direct discourse, which naturally makes use of first-person narration, the problem becomes particularly challenging. Because the narrator in this case seems to disappear entirely, leaving only the character, some narratologists claim that the character should be considered the narrator. In other words, a character is talking about himself or herself and is therefore a homodiegetic narrator. Others suggest that there is an invisible heterodiegetic (often also extradiegetic) narrator trying to represent a character’s consciousness as accurately as possible by using free direct discourse. The positions of different theorists on this matter are not always very clear. They tend to overlap, and an individual narratologist sometimes displays contradictory opinions. In order to illustrate this, we will take a closer look at the prototypical form of free direct discourse: quoted monologue.

When quoted monologue takes up an entire book, we might say there is an extradiegetic and homodiegetic narrator at work, narrating from the highest level and carrying on a self-referential speech. Because such narrators want to represent their inner world as directly as possible, the focalization is internal. There is hardly any observation through the narrating I, who is after all nearly invisible: our observations completely follow the experiences of the I as character. In a diagram from Narrative Discourse Revisited, Genette appears to characterize quoted monologue as follows: extradiegetic and homodiegetic plus internal focalization.143

When quoted monologue is embedded in a larger narrative, however, things change, according to Genette. As an example, he cites Molly Bloom’s famous monologue, which makes up the last chapter of Ulysses. Molly Bloom is not at the highest level of narration: she is being narrated by an extradiegetic and heterodiegetic narrator who in the rest of the novel also narrates Leopold Bloom and Stephen Dedalus. Focalization is still internal, but the extradiegetic narrator is now heterodiegetic as well: he or she no longer coincides with Molly Bloom, who is no longer a narrator but merely a focalizer.

So far everything seems logical. The difference between these two interpretations of the quoted monologue can be reduced to a difference in size: in the first instance the monologue takes up the entire text; in the second case it is only one part among many. In Narrative Discourse, however, Genette noted that the quoted monologue does not have an obvious narrator “but that it should be emancipated right away [ . . . ] from all narrative patronage.”144 The quoted monologue, then, would be so mimetic that the reader is given a direct representation of the character’s consciousness with the narrator disappearing into the background. This is where things get really confusing: first the quoted monologue is the work of a homodiegetic narrator, then of a heterodiegetic one, and now it turns out there might not be any narrator at all.

Dorrit Cohn vented her exasperation at this lack of clarity in a letter to “Dear Gérard Genette.”145 To Cohn it does not matter whether a quoted monologue is part of a greater narrative or stands on its own. She is interested only in the kind of representation used in the monologue itself. Since the first person is used in those kinds of fragments, they are instances of what Cohn calls self-quoted monologue, that is, the narrator is talking about himself or herself and is therefore necessarily homodiegetic.

In his response to Cohn’s letter, Genette maintains that the quoted monologue can be both heterodiegetic and homodiegetic.146 The decision depends on the environment in which it appears: as part of a greater whole or as an independent narrative. Of course this says nothing about the problem of the so-called effaced narrator, the abandoned “narrative patronage.”

Seymour Chatman discusses the effaced narrator in his analysis of the famous Molly Bloom monologue. While not explicitly building on Genette’s theory, he generalizes the French structuralist’s argument of the monologue-as-component. Any quoted monologue, Chatman claims, implies a narrator quoting the thoughts of a character. As the word quoted suggests, a quoted monologue necessarily calls for a quoting agent, who is, according to Chatman, “a totally effaced narrator.”147 Following this line of reasoning, it does not matter whether or not we read Molly’s monologue as an independent narrative. In either case there is a totally effaced narrator.

Chatman’s view can be connected with both Genette’s and Cohn’s. Chatman’s invisible narrator is heterodiegetic when using a third-person text. That is the case for the narrator of Ulysses, who talks about Leopold Bloom, Stephen Dedalus, and Molly Bloom as “he” and “she.” The invisible narrator is homodiegetic when using a first-person text. A quoted monologue with an “I” doing all the talking, without a visible frame narrative using “he” or “she,” is thus a narrative with a homodiegetic first-person narrator. The reader barely notices the narrating I; we see almost directly the experiencing I, whose experience here consists of having memories.

This combination of Chatman’s effaced narrator with Genette’s heterodiegetic and homodiegetic narration and Cohn’s first-person and third-person context suggests that we do not necessarily see the quoted thoughts of a character as a narration by that character. Genette claims that any memory belonging to a character immediately turns that character into an intradiegetic narrator.148 We think of the memory as a story told by a heterodiegetic narrator when this narrator uses “he” or “she”; when the first person is used, the narrator is homodiegetic.

No doubt this calls for an example. When Blok goes to the men’s room in Brakman’s Een weekend in Oostende and remembers past parties, this is a character’s heterodiegetically narrated memory. If Blok had been the narrator of the entire novel, it would have been written in the first person and the memory would have been narrated homodiegetically. The difference between the two does not show so much in the quoted thoughts (which always appear in the first person) but in the frame narrative. In Een weekend in Oostende we read, “Finally, he thought he’d pay a visit to the men’s room.” The memory that follows is narrated by the heterodiegetic third-person narrator. If it had said, “Finally I thought I’d pay a visit to the men’s room,” the subsequent memory would have been narrated homodiegetically.

This brief account of a narratological polemic shows that the creation of unambiguous and generally accepted categories remains a utopian enterprise. Any classification proposed by structuralist narratology gives rise to borderline cases and problems that have yet to be—and probably never will be—solved. In many cases the structuralist is forced to acknowledge that concrete stories always upset theoretical demarcations. That does not mean, however, that these theoretical constructs should simply be cast aside. Even when a story transcends theory, theoretical notions still enable us to describe the workings of the narrative more satisfactorily, if only in purely negative terms, such as, “In Een weekend in Oostende the narrator does not adhere to the difference between heterodiegetic and homodiegetic narrating agents nor to the hierarchical relationship between extradiegetic and intradiegetic.” In those cases where structuralist narratology is limited to producing negative descriptions, other approaches, which we will present in chapter 3, are more fruitful.

In a lot of cases, however, structuralist narratology does contribute to a detailed analysis of the form and content of a narrative text. The three diagrams we have produced for our discussions of story, narrative, and narration may be used as guides to the narratological study of a novel or any other narrative. They afford not only a global perspective on the text as system but also an intimate view of all kinds of details that might remain unnoticed without the benefit of these three diagrams.

Of course the application of the system is only as good as the person using it. The diagrams have to be interpreted, leaving a lot of room for decisions on the part of the reader. Many times he or she will have to make choices that are not necessarily prescribed by the three diagrams or the narrative. One pragmatic decision that will always force itself upon the reader concerns the size of the analysis. If one wants to analyze every single part of a text in depth on all three levels, one will find that the analysis expands endlessly and becomes bloated beyond the dimensions of the story itself. Readers will have to decide for themselves where to stop the analysis, what units to use (for example, chapters or shorter scenes), how relevant certain details may be, and so on. Even in the most rigorous kind of structuralist narratology, there is still room for the unsystematized and the subjective, which are inherent in any reading experience.