Towards a Cognitive Theory of Character¹

Willem G. Weststeijn

University of Amsterdam

   A striking aspect of how narratology has developed since the sixties of the twentieth century is the lack of attention paid to character. This is surprising because character, together with plot, is one of the most important elements of any story (the third, equally ‘indispensable’ element, setting, has, generally, an auxiliary function with regard to character and events). In comparison, far more work in narrative theory has been done on the structure of action, the various kinds of narrators, the implied author, focalization, and the way narrative content is communicated: the discourse, with as one of its main elements represented speech (particularly free indirect style).

“It is remarkable how little has been said about the theory of character in literary history and criticism,” comments Seymour Chatman in Story and Discourse², surveying what has been said on the subject from Aristotle to contemporary structuralist narratology. Sixteen years later, when the heyday of structuralism had long since passed, Patrick O’Neill in Fiction of Discourse: Reading Narrative Theory echoes Chatman’s observation. “The multifarious ways in which characters emerge from the words on the page, in which story-world actors acquire a personality, is one of the most fascinating and least systematically explored aspects of narrative theory and narrative practice”³.

Yet, despite his professed fascination, O’Neill devotes only four pages to character⁴. Whereas both Chatman and O’Neill were influenced by classic French structuralist narratology and are working within its terms, most recent narrative theory has largely abandoned this tradition. As David Herman says in his introduction to Narratologies. New Perspectives on Narrative Analysis, ‘postclassical’ narratology has become narrative analysis which “is marked by a profusion of new methodologies and research hypotheses; the result is a host of new perspectives on the forms and functions of narrative itself.”⁵ Many of these new perspectives have been introduced by narrative theorists who are not confining themselves to literary texts, but are drawing “on fields such as Artificial Intelligence, hypertext, psychoanalysis, film studies, and linguistics (including possible-world semantics and discourse analysis) to broaden and diversify our conception of stories and to provide new ways of analyzing their structures and effects.”⁶

The ‘host of new perspectives’ (Herman’s book contains twelve articles by well-known narratologists) in recent narrative theory seems even less inclined to analyze character than traditional narratology. If we look up ‘character’ in the index of Herman’s book we are redirected to ‘actants.’ Subsequently, all the references to ‘actants’ refer to Herman’s own texts, either to his Introduction or to his article Toward a Socionarratology: New Ways of Analyzing Natural-Language Narratives.⁷ This silence on character (and even actants) is all the more remarkable when one notes that among contributors to Herman’s collection we find not only Seymour

Chatman, who pointed to the tendency to ignore character in classic narratology⁸, but also James Phelan, one of the few narratologists who produced a theoretical model of character⁹.

After almost half a century of narratology, it seems as if the ‘case against character’¹⁰ has resulted in the virtual disappearance of the concept of character from narrative theory. One of the main reasons is, of course, the type of fiction that has been produced during the past fifty years. In contrast to the modernist writers (Proust, Kafka, Thomas Mann), who created a great variety of psychologically complex characters, the postmodernists, with Nabokov as an example, focused on the artificiality of character and on deliberate character construction. Modernist doubt as regards the knowableness and unity of the subject was taken very seriously up to and including existentialism. Postmodernism went a step further and proclaimed the disappearance of the subject as a unity, as a single motivational system, a fait accompli. Accordingly, studying character in the sense of a literary figure that looks like a human being in real life became an anomaly, at least for those theoreticians who focused on contemporary literature. It did not come as a surprise, then, that in the semiotically and linguistically oriented approaches that dominated modern narratology characters were viewed as provisional entities and were often reduced to their verbal and grammatical components.

However, there have been periods in the history of literature when the existence of character as a single entity was not only taken for granted, but also played a prominent part. Modernist writers’ attitude to character has already been mentioned, but particularly the nineteenth-century realist writers are known for creating lifelike and ‘whole’ characters. Many of these characters live on in our memory long after we have forgotten the language through which they came to life¹¹. The period of postmodernism now seems to be coming to an end¹² and although it is risky to predict the features of the new period, there are many indications that there will be a return to more realistic expression. In contemporary Russian literary criticism there is much talk of the ‘new realism’ in Russian literature. The huge success in America of the novel The Corrections by Jonathan Franzen, with its highly detailed descriptions of ‘contemporary’ characters in their social surroundings, points in the same direction.

The question, then, is: what can a narratology that came of age during the postmodernist period and was under the influence of postmodernist literature, offer us with regard to the study of characters in literary texts that are not postmodernist? Or, more specifically, how can this kind of narratology be applied to the analysis of traditional and ‘new realism’ literatures that invite a sociological and / or psychological reading?

Leaving aside the functional approach to character of structuralism¹³, I would like to mention three books published during the past twenty-five years in which character is more or less substantially discussed as a specific entity of the literary work. The first one is the above mentioned Story and Discourse by Seymour Chatman, the second is Vvedenie v literaturovedenie (Introduction to the Theory of Literature, 1991) by the Polish Slavist and semiotician Jerzy Faryno¹⁴, the third is Reading People, Reading Plots by the American anglist James Phelan, also mentioned above¹⁵.

Chatman divides the three elements of narrative under two headings: ‘plot’ under events and ‘character’ and ‘setting’ under existents. Having mentioned that Aristotle, the formalists and the structuralists subordinated character to plot, he argues for an ‘open theory of character.’ Characters generated in literary texts are, of course, not ‘living people.’ However, “that does not mean that as constructed imitations they are in any way limited to the words on the printed page.”¹⁶  Characters cannot be equated with words. We construct a literary character (Hamlet) in the same way that we construct a historical person (Samuel Johnson) and we even ‘construct’ the people we meet in real life: “We read between their lines, so to speak; we form hypotheses on the basis of what we know and see; we try to figure them out, predict their actions, and so on.”¹⁷

In his ‘open theory of character,’ Chatman views characters as more or less autonomous beings, not as merely a function of the plot. But what do we actually do when we encounter a character in a literary work or a film? Chatman’s answer is that in the course of reading a story or watching a film we gradually construct a character by piecing together his or her personal qualities. These qualities are inferred from all kinds of textual data: a character’s name, his actions, thoughts and speech, what is said about him by the narrator or by other characters, etc. In this way we form an image of the character that is, in principle, a “paradigm of traits,” a trait being a “relatively stable or abiding personal quality.”¹⁸ This image or reconstruction can be separated from the text and, indeed, it often is. We do not remember characters because of the words they say or the words used to describe them, but we remember them as constructs (sometimes vague and blurred, sometimes vivid and intense) in our consciousness.

Faryno’s Vvedenie v literaturovedenie is not specifically concerned with narrative theory, but I mention the book here as it goes at length into what Chatman called the existents of the story, the characters and the setting. More than half of the book is devoted to these two aspects of the literary work (the chapter Literaturnye personazi numbers 170, the chapter Predmetnyj mir 160 pages). The emphasis on existents is all the more remarkable because Faryno does not discuss action or events at all. He relies heavily on Soviet semiotics and considers the literary figure as something that is modeled and which, in its turn, has a modeling function¹⁹. Being part of the literary work, the character has a specific meaning. This meaning is based on personal qualities ascribed to the character: his or her name, behaviour, ideas, external appearance, social position and relationships with other characters. The wider literary context also plays a role: the character’s relation with characters from other literary works by the same author or with characters from recent and past works by other authors.

To establish a character’s significance within the semantic structure of the entire work, Faryno focuses particularly on external features such as the character’s physique, facial expression, clothes, diseases and ailments, gestures, behaviour, and, most importantly, his or her name²⁰. These external aspects can be intrinsically linked to the character’s inner self;²¹ together they combine to form ‘a meaning-bearing entity’ within the overall semantic structure of the literary work.

James Phelan²² agrees with Brooks²³ in suggesting that character should not be considered in isolation from the plot. The literary character develops and changes as the plot develops. Hence, the inclusion of the crucial word ‘progression’ in the subtitle of his book. Character development has to be seen in terms of the wider context of the whole narrative as it progresses.

According to Phelan, character is a literary element composed of three components: the mimetic, synthetic and thematic. The mimetic component refers to how a character can be the image of a real and possible person. The synthetic component concerns the artificiality of character, that is to say it stresses that character is a literary construct. Finally, the thematic component refers to how a character can be used to represent a certain idea, a group or a class within the semantic structure of the literary work. The extent to which the three components are developed depends on the kind of fiction. In realistic fiction the mimetic aspect prevails, in postmodern fiction the synthetic component is most important, whereas in fiction with a thesis thematic characters are more common. Characters can be represented people, artificial constructs and themes to varying degrees. Phelan’s theory of character takes place “through an examination of the range of relations among the mimetic, thematic, and synthetic components of character.”²⁴ A character’s ‘attributes’²⁵ can make him or her a plausible person, a unique individual, but may at the same time have a thematic function. This function ‘generalizes’ the character so that he or she becomes the embodiment of an idea or the representative of a group.

One of the advantages of Phelan’s theory compared to those of Chatman and Faryno is that the literary character is not seen as a static but as a dynamic entity. Chatman is undoubtedly right when he says that we often recall fictional characters vividly without remembering a single word of the text through which they were created. Whereas “the medium falls into dimness, […] our memory of Clarissa Harlowe or Anna Karenina remains undimmed.”²⁶  It is, however, not immediately clear what exactly the image is that we retain in our memory. For instance, we read about Anna Karenina in a variety of situations but when we have finished reading the book we do not remember her as a “paradigm of traits.” We are left with an image of a many-faceted, interesting and beautiful woman that has been built on the various layers of her narrated life. Moreover, to use a term coined by Thomas Docherty, she functions as a ‘kinetic’ character, one whom we can only partially know.²⁷

Faryno is not interested in ‘kinetic’ characters that exist beyond the bounds of the fiction in which they were created. Character is only important for him as a composite of meaning within the semantic structure of the entire text. In looking at historical figures in fiction, for example Puskin’s Pugacev or Tolstoj’s Napoleon, Faryno emphasizes the principal differences between the character in the text and their real life counterparts and even denies the possible role of referentiality²⁸. The text contains everything that is needed to understand the meaning of its characters. However, as his painstaking analysis of the semantic potential of the name Smerdjakov proves, Faryno obviously takes the view that in order to determine this meaning, with all the possible associations and connotations, the reader has to have a thorough general knowledge of, for instance, linguistics, folklore and mythology (see note 19). If a reader’s knowledge of such outside information plays such an important part in the construction of meaning, it is hard to see why referentiality can be ignored. In other words, how can the character Napoleon in Vojna i mir not be affected by the ‘real’ Napoleon? In this case the reader’s (historical) knowledge may add important connotations or reveal hidden meaningful relations.

Faryno’s analysis of literary character eventually also results in a paradigm of traits, although, in contrast to Chatman, he does not grant a character an existence as a ‘private self’ outside the text. In viewing character as more or less mimetic, Phelan, like Chatman, accepts this possibility. At the same time, he does not lose sight of the thematic aspect, the character’s role in the action of the story and in the semantic structure of the text. As mentioned above, Phelan uses ‘attributes’ instead of ‘traits’ as the fundamental units of character. Yet neither he nor Chatman nor Faryno show how the reader determines these attributes.

Literary characters are by definition generated through the words of the text. The relation between the words and the character or a character’s traits or attributes is generally taken for granted²⁹. When a text states that a heroine is beautiful, we ‘see’ the image of a beautiful woman. All kinds of attributes may be added to this image, some based on direct textual statements, some inferred from a description of the heroine’s actions, thoughts, words or relationships with other characters. The image generated can have both a physical appearance and a ‘psychology’, aspects of the former being often causally connected with aspects of the latter. Sometimes, especially in ‘authorial narration,’ the information about a character is such that we feel that we can easily and unambiguously ‘translate’ the words of the text into an image. However, contemporary narrative theory has amply demonstrated, by exploring aspects such as the unreliable narrator, changes of focalization and free indirect style, how extremely complex literary discourse can be. A satisfactory theory of character cannot ignore discourse but, on the contrary, will have to concentrate explicitly on textual patterns and determine accurately what information is given about a character, who gives this information and how this information relates to other, possibly incompatible and conflicting data. Such a theory, which combines discourse analysis and reader’s response (generating images on the basis of words and imaginative involvement) might be developed by using insights and concepts of cognitive science, the relevance of which in literary theory seems to be rapidly increasing³⁰.

A useful model from cognitive science that may be applied to a theory of literary character is the so-called ‘frame theory’, developed by Marvin Minsky and others in the study of artificial intelligence³¹. This model, employed in literary theory by Jahn³², Herman³³ and others³⁴, postulates that every time a person encounters a new situation he tries to make sense of it by selecting from his memory a mental structure or ‘frame’ that matches the situation as closely as possible. Minsky defines ‘frame’ as:

a remembered framework to be adapted to fit reality by changing details as necessary. […] We can think of a frame as a network of nodes and relations. The ‘top levels’ of a frame are fixed, and represent things that are always true about the supposed situation. The lower levels have many terminals - ‘slots’ that must be filled by specific instances and data. Each terminal can specify conditions its assignments must meet. (The assignments themselves are usually smaller ‘sub-frames.’) Simple conditions are specified by markers that might require a terminal assignment to be a person, an object of sufficient value, or a pointer to a sub-frame of a certain type. […] Much of the phenomenological power of the theory hinges on the inclusion of expectations and other kinds of presumptions. A frame’s terminals are normally already filled with ‘default’ assignments. Thus, a frame may contain a great many details whose supposition is not specifically warranted by this situation³⁵.

A not unimportant advantage of frame theory in its application to a narrative text is that it makes us realize that new textual data are always seen in the light of previous knowledge. In other words, in the process of reading, the frame or frames are continually adjusted the slots are filled in, further or differently, in response to the new information communicated through the text. It is possible, I think, to see literary characters as ‘frames’ or a part of a ‘frame.’ As soon as characters are introduced into the text, we allocate them a particular frame. At first, this frame is fairly general (male, female, old, young) and still has many vacant slots. Gradually the slots are filled in as new data are provided by the text. However, these data always have to be considered within the framework of the entire context. Statements by an unreliable narrator or the reproduction of the point of view of an obviously prejudiced character have to be assessed in another way than ‘objective’ observations and pronouncements by an omniscient narrator. Contradictory information about a character from different focalizers may be stored in different slots. One might consider the possibility that at a certain moment in the reading process a slot or a number of slots, as part of the frame, are more or less complete and do not receive nor need any supplementary information. Or perhaps the reader will ‘close’ one slot or more, for instance at the end of a chapter or after an important narrative sequence. This makes it theoretically feasible that readers may retain in their memory not one but several more or less different images of the same character.

Anyone who has read Anna Karenina knows that at the end of the novel Anna commits suicide by throwing herself in front of a train. It is unlikely that a reader when remembering the fictional character Anna Karenina will recall an image of her mangled body on the tracks. So what kind of image does the reader recall? Anna in love as portrayed in the first part of the novel, the unhappy Anna of the last part? The answer must be that the reader sees a number of images simultaneously, including perhaps that of her dead body. Reading is an interpretative and imaginative dynamic process that starts with the first words of the story and goes on until the last line. Apart from following the plot and imagining the setting, it entails the dynamic process of constructing character in terms of both physical and psychological attributes by filling in slots and applying frames.

References