During the years when the study of literature was dominated by a global movement that I call textualism, whose individual manifestations include New Criticism, Deconstruction, and Postmodern Theory, it was common when speaking of literary characters to oppose a “naive reader” who regarded characters as persons, or as model human beings (Herman et al. 2012: 125) to an enlightened reader who knew that characters are constructs made of language. The following declarations are meant by their authors to uncover the “true” essence of characters: “Characters in fiction are, after all, words on a page” (Richardson 2015: 133). “Characters are marks on a page, made of the alphabetical characters that spell out ‘who’ they are. They have no psychology, no interiority, no subjectivity” (Warhol in Herman et al. 2012: 119). The epitome of the naïve reader is Don Quixote (second part, chapter XXVI), who jumps on stage to rescue the heroine of a puppet show, causes a huge brawl during which all the puppets are broken, and ends up having to pay for the damage (and recognizing his error). We all know that Don Quixote was crazy, his brain having dried up from reading too many chivalry novels, and that he was unable to distinguish fiction from reality. But there are many people who engage in behaviors that orthodox literary theory would consider naive: for instance, kids watching puppet shows who scream to warn the hero that the bad guy is approaching. Or people who travel to Reichenbach Falls in Switzerland to see the spot where Sherlock Holmes fell to his death (only to be resurrected later). And finally, there are those people who cry when reading fiction: Dickens ‘Little Nell is reported to have let loose torrents of tears. Are these people as crazy as Don Quixote, or do they know something about the nature of fiction and about characters that the puritans of textualism refuse to acknowledge? In this chapter I will try to answer this question by exploring, and contrasting, three approaches to character: textualist, philosophical, and the approach that I personally endorse, which I will call the world approach.

Textualism

The tradition that I call textualist bears prime responsibility for claims that characters are not persons but objects made of language. Textualism grew out of structuralism, and it is a matter of common knowledge that structuralism drew inspiration from linguistics, especially from the linguistic theory of Ferdinand de Saussure, who regarded language as a self-enclosed system rather than as a means to represent an external reality. Linguistics was revered by structuralists as a pilot discipline, and the adoption of its methods to other fields such as anthropology, sociology, and last but not least literary theory was considered a way for these disciplines to reach scientific status. The big question however was how to apply linguistic methods to other domains of signification, for beyond the concept of sign and its division into a signifier and signified, and beyond the claim that the value of signs depends on their systemic relations to each other rather than on their relation to the world, Saussurian linguistics did not provide particularly concrete directions. The adoption of a linguistic model in the humanities and social sciences was mainly metaphorical and ideological: it was metaphorical, because it consisted of regarding every phenomenon under study as a “language” based on a “code” made of discrete signifying units, and it was ideological, because it viewed human thought as profoundly shaped by these multiple codes. In its most radical conception, the linguistic influence means that we are spoken by language rather than speaking it. The same could be said about the various codes of culture: even nowadays, cultural studies have a strongly deterministic bend. It is against this background that I would like to discuss the treatment of literary characters by the most famous of structuralist critics, Roland Barthes, in S/Z, his well-known study of a short story by Balzac titled “Sarrasine.”

The ambition of Barthes in S/Z was to promote a view of literary texts as systems constituted by multiple codes in which the meanings determined by these codes activate each other in an endless play of signification. But this play can be more or less extensive; in the type of text he calls the writerly [scriptible], signs activate each other in all directions, there is no beginning nor end, and the reader becomes a producer of meaning; in the type of text he calls the readerly [lisible], the play of signification is limited by the linearity or sequentiality of narrative structure and by the demands of mimeticism. But even for readerly texts, Barthes opposes the classical literary-critical position that regards meaning as imposed top-down by the author. While he conceives the writerly text as fully created bottom-up by the reader, the readerly text represents a “limited plurality” where bottom-up interpretive activity is at least partly controlled by top-down structures. Yet Barthes does not go as far as saying that these structures are imposed by an author: rather, they are the product of cultural codes and of language itself. It is interesting to note that despite praising the writerly as the future of literature and as the liberation of the reader, Barthes never devoted much attention to such texts. It could be because they do not exist; but I rather believe that, deep down, he preferred the readerly, because the readerly preserves narrative interest, which lies at least in part in temporal effects such as suspense, curiosity and surprise.

Barthes regarded narrative texts as being regulated by five codes: the proairetic code, which organizes the actions of characters into meaningful sequences, such as “taking a walk” or “rendez-vous” or “assassination”; the hermeneutic code, which organizes narrative information into the presentation and then solving of an enigma; the semic code, which consists of gathering the connotations of textual units, for instance extracting “wealth” from the description of a house; the symbolic code, which links particular existants to universal concepts, and the cultural code, which links textual units to established forms of knowledge, especially popular wisdom and stereotypes. The designation of these interpretive moves as “codes” betrays the dominance of the linguistic paradigm; nowadays we would be more enclined to regard them as “cognitive operations”; and rather ascribing the functioning of the semic code to codified relations between signifers and signifieds, we would recognize the importance of the reader’s life experience for extracting connnotations and making inferences. But whether or not Barthes’ codes are really codes in a strict sense of the term, every textual unit is justified by its participation in one or more of them; the more numerous the codes, the more determined and meaningful the textual unit. These units can be segments of variable length, and their delimitation is not determined by the kind of systematic discovery procedure that structuralist linguistics made into their number one priority, but by whether or not Barthes had something to say about them.

Like any other textual unit, characters lie at the intersection of several codes. Take Zambinella, the castrated male singer with whom the artist Sarrasine falls in love, believing that he is a woman. According to the symbolic code, she represents super-feminity and sub-masculinity. The hermeneutic code presents her sexual identity as an enigma that needs to be solved. According to the cultural code, she stands for ideal beauty. The proiaretic code enables the reader to gather her actions into meaningful sequences, such as a sequence of “playing a trick on Sarrasine by pretending to be a woman,” and the semic code enables the reader to extract the connotations of the words that describe her, for instance, linking her mouth to sensuality.

More generally, for Barthes a character is a collection of semes (that is, meanings) subsumed under the heading of a proper name. Just as the proairetic code instructs the reader to gather information under a general type of action such as “murder,” “walk” or “rendez-vous,” the proper name unifies a dispersed collection of semes and turns them into a character:

Sarrasine est la somme, le lieu de confluence de: turbulence, don artistique, indépendence, excès, féminité, laideur, nature composite, impiété, goût du déchiquetage, volonté, etc. Ce qui donne l’illusion que la somme est supplémentée d’un reste précieux (quelque chose comme l’individualité, en ce que, qualitative, ineffable, elle échapperait à la vulgaire comptabilité des caractères composants), c’est le Nom Propre, la différence remplie de son propre. Le nom propre permet à la personne d’exister en-dehors des sèmes, dont cependant la somme la constitue entièrement. (Barthes 1970, p. 197)

Sarrasine is the sum, the point of convergence, of: turbulence, artistic gift, independence, excess, feminity, ugliness, composite nature, impiety, love of whittling, will, etc. What gives the illusion that the sum is supplemented by a precious remainder (something like individuality, in that, qualitative and ineffable, it may escape the vulgar bookkeeping of compositional characters) is the Proper Name, the difference completed by what is proper to it. The proper name enables a person to exist outside the semes, whose sum nonetheless constitutes it entirely. (Barthes 1970, p. 191).

The last sentence in this quote shows that Barthes’ conception of characters is more complex than reducing them to a “collection of semes.” The proper name does indeed turn characters into persons. Where then do these persons exist, if the text is entirely constituted of semes? The only answer can be: they exist as “persons” in the imagination of the reader. But this existence is an illusion, because, as Barthes, notes, the semes “constitute them entirely.” The instrument of this illusion is the proper name, which suggests the existence of a referent external to language. Barthes admits that his own critical discourse is not always immune to the illusion of existence created by the proper name:

On parle ici, parfois, de Sarrasine comme s’il existait, comme s’il avait un avenir, un inconscient, une âme, mais ce dont on parle, c’est de sa figure (réseau impersonnel de symboles maniés sous le nom propre de Sarrasine), non de sa personne (liberté morale douée de mobiles et d’un trop-plein de sens): on développe des connotations, on ne poursuit pas des investigations; on ne cherche pas la vérité de Sarrasine, mais la systématique d’un lieu (transitoire) du texte. (Barthes 1970, p. 101, XLI)

We occasionally speak of Sarrasine as though he existed, as though he had a future, an unconscious, a soul; however, what we are talking about is his figure (a impersonal network of symbols combined under the proper name “Sarrasine”), not his person (a moral freedom endowed with motives and an overdetermination of meaning): we are developing connotations, not pursuing investigations; we are not searching for the truth of Sarrasine, but for the systematics of a (transitory) site of the text. (Barthes 1970, p. 94)

In other words, the proper name is deceptive, because it presents characters as persons, and suggests that they exist independently of the text. The task of the critic is to demystify the text, to denounce the illusion, and to prepare the advent of the “writerly text,” which will do away with illusion: “ Ce qui est caduc aujourd’hui dans le roman, ce n’est pas le romanesque, c’est le personnage; ce qui ne peut plus être écrit aujourd’hui, c’est le nom propre (Barthes 1970, p. 102).” “What is obsolete in today’s novel is not the novelistic, it is the character; what can no longer be written is the Proper Name” (Barthes 1970, p. 95).

Barthes wrote in an age when the advocates of the Nouveau Roman claimed that the novel had to do away with such fundamental narrative elements as plot and character (Robbe-Grillet 1965, pp. 25–47). To empty characters of any human substance, it often replaced proper names with bare initials: for instance, the wife in Robbe-Grillet’s La Jalousie is referred to as A. But the New Novel was a short-lived literary fashion: today’s novels are still full of characters, and the proper name remains the main way of referring to them. Barthes' prophecy turned out to be dead wrong.

To demonstrate that the behavior of characters is not entirely dependent on the realism of psychological motivation, Barthes discusses a passage where Zambinella tries to confess to Sarrasine that she is not a woman, but a castrated male. The text goes like this:

–Ecoutez, monsieur, dit-elle d’une voix grave.

–Oh, tais-toi, dit l’artiste enivré. Les obstacles attisent l’amour dans mon cœur. (Barthes 1970, p. 183)

“Listen, monsieur,” she said in a low voice.

“Oh, be still,” the impassionate artist said. “Obstacles make my love more ardent.” (Barthes 1970, p. 177)

As a result of being told to shut up, Zambinella does not make the confession. Here is Barthes’ comment:

Si l’on a une vue réaliste du personnage, si l’on croit que Sarrasine vit en-dehors du papier, on cherchera les mobiles de ce geste d’interruption (enthousiasme, refus inconscient de la vérité). Si l’on a une vue réaliste du discours, si l’on considère l’histoire racontée comme une mécanique dont il importe qu’elle fonctionne jusqu’au bout, on dira que la loi de fer du récit voulant qu’il continuât encore, il était nécessaire que le mot de castrat ne fût pas prononcé. (Barthes 1970, p. 184, LXXVI).

If we have a realistic view of character, if we believe that Sarrasine has a life off the page, we will look for motives for this interuption (enthusiasm, unconscious denial of the truth, etc.). If we have a realistic view of discourse, if we consider the story being told as a mechanism which must function until the end, we will say that since the law of narrative decrees that it continue, it was necessary that the word castrato not be spoken. (Barthes 1970, p. 178)

If Zambinella had been able to make the confession, the story would not have reached its dramatic climax, the murder of Sarrasine by Zambinella’s protector during a later meeting. What Barthes calls the demands of discourse are the demands of plot, and they override the demands of verisimilitude: Sarrasine’s refusal to hear Zambinella’s confession is an ad hoc response that cannot be justified on psychological grounds. Here Barthes confronts a dilemma that is often invoked by theoreticians of character: should characters be subordinated to plot—in which case they are mostly cogs in a system—or should plot be subordinated to character? The answer depends on genre: according to Aristotle, in tragedy character is subordinated to plot (Poetics 4.3); by contrast one can assume that in epics, plot is subordinated to character. The purpose of the multiple episodes of the Odyssey can for instance be said to be the demonstration of the personality of Odysseus. But in some cases, psychologically or pragmatically motivated actions by characters would prevent an interesting development. To avoid this pitfall, authors often sacrifice credibility to the demands of plot. A common example is the convention of the calumniator credited (Steinmann 1981, 258), where an intelligent character believes the lies of a character of low reliability, such as Iago in Shakespeare’s Othello, because this leads to a crisis which is central to the plot. If characters are nothing more than collections of semes, there is nothing to lose by accepting their subordination to plot; but Barthes stance is more complex, as we see from this passage: “D’un point de vue critique, il est [donc] aussi faux de supprimer le personage que de le faire sortir du papier pour en faire un personage psychologique (doté de mobiles possibles): le personage et le discours sont complices l’un de l’autre. » (Barthes 1970, p. 184). “From a critical point of view, therefore, it is as wrong to suppress the character as it is to take him off the page in order to turn him into a psychological character (endowed with possible motives): the discourse and the character are each other’s accomplices.” (Barthes 1970, p. 178). Here Barthes is telling us that it is wrong to assume that characters are human beings and that they act out of psychological motivations, but it is also wrong to suppress characters in the name of discourse. In other words Barthes wants to have his cake and eat it too. He gets out of this impasse through a magic trick typical of textualism, by refusing to make a distinction between characters and discourse: “les personnages sont des types de discours et à l’inverse le discours est un personnage comme les autres” (Barthes 1970, p. 184). “The characters are types of discourse, and, conversely, the discourse is a character like the others” (Barthes 1970, p. 178). Like the others, really? When Barthes writes that “the discourse is a character like the others,” he wants to redirect the reader’s attention from the actions of characters to the action of the narrative discourse itself, to the twists and turns this discourse takes, and to the strategies it deploys to achieve its goal of proper narrative form. Naïve readers read texts as being about the life, the adventures and the schemes of characters; sophisticated readers read texts as being about the life, the adventures, and the schemes of discourse.

Barthes’ more or less implicit goals, in writing S/Z, were aesthetic and pedagogical. He wanted to promote a new mode of reading that broke with the biographical and positivist tradition that dominated French academia early in his career, a mode that liberated textual energies and that allowed readers to find pleasure in playing with language. He also wanted to promote an alternative to mimeticism and realism, an alternative embodied in the writerly, though deep down his loyalties remained with the “limited pluralism” of classical narrative, rather than with the chaotic multiplicity of the writerly. With regard to characters, he seems to have realized the limitations of a purely semiotic approach, and he opted for ambiguity, rather than openly admitting that there is more to characters than collections of semes.

The philosophical approach

Philosophers, especially those of the analytic schools, have been interested in fictionality long before narratologists paid attention to it—which is a fairly recent development.Footnote 1 For analytic philosophy, the questions that need answering are very different from Barthes’ concern with the functioning of literary language. These questions are the following:

  • How can we refer to fictional characters and make statements about them, for instance by saying “Anna Karenina is a more passionate lover than Emma Bovary.”

  • How does one assess the truth of statements made by critics about fictional characters?

  • Are there relations of identity between, say, Marlowe’s Faust and Goethe’s Faust, or are they mere homonyms?

  • And, last but not least, what is the mode of existence, or ontological status, of fictional characters? Do they represent a special mode of being, a position that transgress a principle dear to philosophers, namely Occam’s razor, or should one say that they do not exist at all?

The question of the ontological status of fictional characters is deeply entwined with the question of reference, because reference is widely believed by analytic philosophers to imply some kind of existence. According to Bertrand Russell, a sentence about a non-existing entity such as Emma Bovary is necessarily false, because of referential failure; according to Gottlob Frege’s more flexible account, it is not false but indeterminate. But in everyday life we frequently refer to fictional characters, and we intend our statements to say something true: for instance, “John is a real Scrooge;” or “Bill has seduced more women than Don Juan.” Moreover, if it were impossible to make true statements about fictional characters, the only criterion of validity for the claims of literary critics would be whether they are provocative enough to excite passions.

Here I will focus my discussion of philosophical approaches on the theory of Amie Thomasson, as developed in her 1999 book Fiction and Metaphysics. Thomasson believes that fictional characters can be the target of reference. They must therefore have some mode of existence. But this mode of existence is not the one of concrete, material objects such as people and apples, which are located in space and time, nor the one of purely abstract entities such as numbers or beauty, which exist eternally and cannot be traced back to any specific human creative act. For Thomasson fictional characters are what she calls “abstract artifacts”, and as such they occupy a middle ground between the material and the mental, the concrete and the abstract. Characters are abstract because they do not exist in space and time, but they are artifacts, because they depend on literary works, and literary works depend on the concrete existence of authors and books (or other media). “In short, on this view fictional characters are a particular kind of cultural artifact. Like other cultural objects, fictional characters depend on human intentionality for their existence” (Thomasson 1999, p. 14). Worth noting in this definition is the claim that characters depend on the intentional act of the author, not on the text itself, as textualists like Barthes would probably claim. If two authors happened to write the same words, unbeknownst of each other, they would produce distinct literary works and different characters.Footnote 2 In Thomasson’s view, fictional characters are born through the creative act of the author, they are maintained in existence through the media that make them accessible to readers, and they die when the last copy of the work, or its memory in a reader’s mind is destroyed. This idea that fictional characters can “die” may seem counterintuitive; isn’t “death” a phenomenon specific to concrete, material objects? And yet we say that languages die together with their last speaker; why should it be different with characters? If something can die, it means that it once existed.

Through her concept of abstract artifact, Thomasson opposes theories that regard characters as unrealized possibilities. According to these theories, Hamlet could have existed, and the world could contain elves, trolls, and hobbits, as it does in Lord of the Rings. This conception of fictional characters as possible entities is supported by Aristotle’s claim that the task of the poet is to show “not what is but what could be in accordance with probability and possibility” (Poetics 5.5). But there are countless possibilities that have not been textualized. Possibilities exists independently of whether or not somebody thinks of them; therefore, if one takes the view that fictional characters are unactualized possibilities, then authors do not “create” or “invent” characters, but rather “discover” them. Thomasson rejects this idea on the ground that it makes fictional characters independent from the creative acts of authors. Yet if characters do not preexist the text as unactualized possibilities, if they are not “discovered,” this raises the question of how they arise in the author’s mind. Authors do not create characters ex-nihilo, they rather make them up by mentally exploring a field of possibilities, and by selecting some of these possibilities to be realized textually.

An important feature of Thomasson’s proposal to keep in mind, is that it defines the ontological status of characters from an external point of view—the point of view of the author who inhabits the real world, rather than the point of view of the narrator and characters who are internal to the story. If fictional characters are the product of an act of creation that takes place in space and time, then their properties are entirely determined by this act of creation. And since the author can only imagine and encode a limited number of properties, this means that fictional characters are ontologically incomplete. According to this view, the number of the children of Lady Macbeth, which is unspecified in Shakespeare’s tragedy, represents an ontological gap, a hole that cannot be filled in her character, and as the critic Lionel Charles Knights argued as early as (1933), it would be pointless to try to answer the question “how many children had Lady Macbeth.” The vast majority of the philosophers and literary critics who have addressed the problem of the nature of fictional characters endorse the idea of their radical incompleteness, for instance Uri Margolin, Thomas Pavel, Lubomír Doležel, Ruth Ronen, and Nicholas Wolstertorff.Footnote 3 This incompleteness could be seen as what distinguishes “abstract artifacts” from real persons. But the thesis of incompleteness runs into problems when a member of the real world appears in a novel, for instance Napoleon in War and Peace. Should one postulate the existence of a real, ontologically complete Napoleon, as well as of a fictional, incomplete version of the emperor? If these two Napoleons are ontologically so different from each other, how can one explain that, for the reader, they are versions of the same person? Readers of War and Peace will imagine the fictional Napoleon as sharing many properties with the real one, even when these properties are not specifically mentioned in the text. For instance, if they try to visualize mentally a battle scene of the novel where Napoleon appears, they are entitled to imagine Napoleon as short, putting his hand in his coat, riding a white horse, and wearing his hat sideways, as shown in the painting by Meissonier. I am not saying that they have to imagine him in such details, but rather, that there is nothing wrong in doing so.Footnote 4 On the other hand, it would be wrong to imagine Napoleon as looking like Don Quixote.

Thomasson does recognize the limitations of a theory that regards incompleteness as a defining ontological feature of fictional characters. She considers the two statements “According to the story Hamlet has blood type A” and “According to the story Hamlet has blood type B” as both false or indeterminate because Shakespeare’s text says nothing about the blood type of Hamlet” (108). But it does not follow from the negative or indeterminate truth value of these statements that “According to the story Hamlet is incomplete as to blood type.” The reason for the failure of the entailment is that, according to the story, Hamlet is not a fictional character but a regular human being. He is not created by an author, but born of a father and mother. And since regular human beings are ontologically complete, one must assume that Hamlet is also complete, at least within the story.

The difference between an internal and an external point of view can be illustrated by these two types of statement: “Hamlet is a fictional character who was created by Shakespeare” and “Hamlet is the Prince of Denmark.” The second statement can be paraphrased as “According to the story Hamlet is the Prince of Denmark.” By adding the prefix “according to the story”, an internal statement can be turned into an external one and receive a positive truth value for the real world. But this operation fails with external statements: we cannot say that “according to the story, Hamlet is a fictional character who was created by Shakespeare.”

Thomasson retains an external perspective when she characterizes fictional characters as abstract artifacts, for it is certainly not true that according to Shakespeare’s play, “Hamlet is an abstract artifact.” In her model, some statements about fictional characters must be prefixed by “in the story” in order to be evaluated as true or false, and others should not. But how do we know which ones should be prefixed and which one should not? An alternative to “externalizing” internal statements with a prefix is to make the contrast between an internal and an external point of view into the cornerstone of a theory of the ontological status of fictional characters. If we replace the prefix “in the story of Hamlet” with “in the world of Hamlet,” we can account for the duality of perspective through a theory inspired by the concept of Possible World. It is to this theory that I turn next.

The world approach

David Lewis, a pioneer of Possible Worlds theory who has also made groundbreaking contributions to the theory of fiction, makes the following observation:

The storyteller purports to be telling the truth about matters whereof he has knowledge. He purports to be talking about characters who are known to him, and whom he refers to, typically, by means of ordinary proper names. But his story is fiction, he is not really doing these things” (1978, 40).

So what is the storyteller doing? He is engaging in an act of pretense or role-playing. And like all acts of pretense, fictional storytelling involves a doubling of world and a doubling of speech acts:

“Here at our world we have a fiction f, told in an act a of storytelling; at some other world we have an act a’ of telling the truth about known matters of fact; the stories told in a and a’ match word for word, and the words have the same meaning” (1978, 40).

Lewis does not specify who does the two acts of storytelling, but narratology has an easy answer: a is the act of the author, a’ the act of a narrator. The author pretends to be the narrator, who resides in the storyworld, and who presents the characters as real individuals. This means that within the storyworld, characters have the same ontological status as the inhabitants of the real world. But how can the characters be regarded by the reader as actual indidividuals, when the world they inhabit is not the real world but a non-actual possible world created by the author? To explain this, we must turn to Lewis indexical conception of actuality and possibility.

Possible Worlds theory is based on the idea that “things could be different than they are.” This phase presupposes a contrast between two worlds, the world where things are “how they are,” let’s call it the actual or real world, and the world where things are different. But since things could be different from the actual world in many different ways, while things are what they are in only one way, it follows that there is one actual world and many non-actual possible worlds. These possible worlds can be accessed by the mind, either by imagining counterfactual events, or by creating and consuming fictions. The real or actual world could be said to be the only world that has material existence; non-actual possible worlds, including those of fictions, have only a mental existence. This is the most common-sensical conception of possible worlds, and it preserves an ontological distinction between fictional characters and real persons. But this is not how Lewis envisions the nature of possible worlds. For him, possible worlds exist objectively, and there is no ontological distinction between the actual world and merely possible ones: both kinds are made of the same substance, that is, of material things and events. This position is known as “modal realism.” As Lewis (1978) writes,

Our actual world is only one world among others. We call it actual not because it differs in kind from all the rest but because it is the world we inhabit. The inhabitants of other worlds may truly call their own world actual, if they mean by ‘actual’ what we do; for the meaning we give to ‘actual’ is such that it refers at any world i to the world i itself. ‘Actual’ is indexical, like ‘I’ or ‘here’, or ‘now’: it depends for its reference on the circumstances of utterance, to wit the world where the utterance is located. (Lewis 1978, p. 184)

Lewis’ idea that all possible worlds exist objectively may be difficult to accept, but it works very well for fictional worlds and their inhabitants. If “actual” is indexical, fictional characters are actual, embodied and ontologically complete individuals from the point of view of the fictional world. But from the point of view of our actual world, they are the abstract artifacts that Thomasson describes. By referring to the characters as if they were real persons, and by describing their world as real, the fictional text invites readers to transport themselves in imagination into the fictional world, and to adopt the point of view of one of its members. I have called this mental transportation “recentering” (Ryan 1991, 18 ff.) and I regard it as essential to the experience of immersion in a story world.

The world account has multiple advantages. Among them:

  • The dual perspective makes it possible to take both the author and the narrator in consideration. The textualist perspective of Barthes eliminates both author and narrator, while the philosophical perspective of Thomasson takes the point of view of the author. Here, characters are both made up creatures from the point of view of the author, and individuals who exist independently of the text from the point of view of the narrator. Their story is invented from the point of view of the author, but it is told as true fact from the point of view of the narrator.

  • Because storyworlds can be contemplated from both an inside perspective and an outside perspective—the perspective of the real world—the world model allows users to move back-and-forth between these two perspectives, and it explains how characters can be regarded as both human beings and as textual constructs. James Phelan (1989) has identified three so-called “functions” that characters can fulfill: the mimetic function (passing as a person), the synthetic function (reminding the reader that it is fabricated), and the thematic function (standing for an idea). The mimetic function represents the internal perspective, while the synthetic and the thematic functions represent the external perspective.

  • The world-model solves the problem of the presence of actual individuals such as Napoleon in fictional texts. If things could be different from what they are, as Possible Worlds theory tells us, there are worlds where Napoleon has different properties than the Napoleon of the actual world, for instance worlds where he wins the battle of Waterloo, and worlds where he interacts with the heroes of Tolstoy’s War and Peace. The difference between Napoleon and Natacha is that Napoleon exists in both our actual world and in the world of War and Peace, while Natacha does not exist in our actual world. But within the world of War and Peace they share the same ontological status.

  • If real individuals can have counterparts in different worlds, so can purely fictional individuals. This explains the practice of transfictionality (Saint-Gelais 2011), or the phenomenon of fan fiction, which consists of writing stories that change some of the features of fictional characters and place them in different worlds, but maintain a connection with the original manifestation of the character. Authors can also expand preexisting fictional worlds, for instance by writing new stories about a character that fully respect the original properties of this character.

  • The world approach does not limit characterhood to human beings. The field of the possible is very vast, and we can imagine worlds with species entirely different from those of our world: species such as dragons, elves, fairies, witches, talking animals, Martians and robots smarter than humans. All it takes to turn members of these species into characters is to give them distinctive mental attributes and cognitive abilities that turns them into agents, abilities such as free will, desires, and self-consciousness.

  • By ascribing to characters the same ontological status within their world as to human beings within our world, the world-model easily explains emotional reactions toward characters, such as empathy or intense dislike. We cry for Little Nell not because she represents an abstraction, as Richard Walsh has suggested (2007, ch. 8), but because we transport ourselves in imagination into a world where she exists as an innocent little girl who suffers underserved hardship and dies an early death.

As we can see from this list of advantages, the world approach explains a phenomenon that neither the textual nor the philosophical approach is trying to deal with. This phenomenon is the reader’s experience of characters, or, to put it differently, the behavior of readers with respect to characters. The textual approach took an external perspective; it asked how readers assemble characters out of scattered information, and it warned them against mistaking these textual constructs for persons. The philosophical approach was concerned with the possibility to make true or false statements about entities that do not exist in space and time, and it asked about the mode of being of these entities. This is not the kind of question that readers normally ask. The strength of the world approach lies in its ability to deal with behaviors that the textual approach regards as unworthy of an aesthetic appreciation of literature and that the philosophical approach regards as outside its field of expertise. By allowing a dual perspective, both internal and external, the world approach explains how characters can be experienced as both persons existing autonomously and as textual creations; how characters can both appear to act out of their own free will and be used to represent certain themes and ideas; how characters are both tied to their world, and able to migrate to other texts and other worlds; and, last but not least, how readers can cry for characters while fully enjoying their crying, for unlike Don Quixote, readers are aware that characters are not real people.

But what about characters that are not “possible persons,” because their properties are self-contradictory, because they lack the cognitive abilities that makes it possible to interpret their gestures as meaningful actions, because they are flat allegories lacking any human substance, because they regard themselves as fictional and not real, or because they present ontological gaps that cannot be regarded as missing information (I regard the characters of Beckett’s Waiting for Godot as embodying this kind of radical incompleteness)? In his work on “unnatural narrative” (2015), Brian Richardson has presented many examples of characters who lack the dimension of personhood. The textual approach has no problem with them; they are just collections of semes gathered around a name, a pronoun, or an actor’s body in theatre and film. The philosophical approach makes no categorial distinction among characters: they are all abstract artifacts created by authors. Can the world approach deal with such creations and distinguish them from fully developed characters? I see two ways of approaching this problem.

First we could say that characters lacking the status of possible persons are entities that block the world-internal perspective and limit the reader to the external stance. Blocking the world-internal perspective means preventing the imagination from filling up these characters with anthropomorphic substance, and forcing awareness of their synthetic or made-up nature. The only operations left to the reader are asking what abstract idea the characters are supposed to represent, and how they contribute to the global organization of the text. The more strongly a character represents a theme or idea, the weaker his status as person and his perceived autonomy: allegorical figures are the puppets of the author, not creatures acting out of free will. Can one still call these creations characters? Only if one defines “characters” as the referents of proper names in fictional texts.

A second approach would consist of retaining a conception of characters as possible persons, and of denying some of the referents of proper names the full status of characters. There is no reason why every fictional text should create something worth calling a world; similarly, there is no reason why every fictional text should tell a fully developed story that involves individuated and ontologically complete persons. Characterhood, worldness and narrativity are not binary features but scalar properties of the mental representations elicited by texts: there are texts of low narrativity (Ryan 2007), there are texts whose semantic content does not really congeal into a world because it lacks logical coherence, and there are referents of proper names which lack individuating and mental human substance. If there is such a thing as an “unnatural character,” it is not a fantastic creature representing a species that cannot be found in the real world, it is an entity that is not fully realized as a person, that appears in a text of low worldness and narrativity, and that belongs to the margins of the fuzzy set of characters. Figure 1 represents the various degrees of characterhood. I place on the outside characters who have no proper name, who are ontologically incomplete, who embody contradictory properties, who have no stable identity, who are mentioned but do not appear on the narrative scene (cf. Godot), or who exist only as unrealized possibilities, such as the wife that Frankenstein did not make for his monster. At the center of the fuzzy set I have put, Don Quixote, Elizabeth Bennet, Mr. Darcy and the Star Wars cast. These are characters who have inspired intense transfictional activity, such as transmedial adaptations, fan fiction, prequels, sequels and transpositions. (As Darth Vader shows, they do not have to be possible members of the real world.) But Hamlet and Emma Bovary, Sherlock Holmes and Little Nell, Donald Duck and Tintin also belong in this inner circle, because the fullest of characters are those that speak so strongly to the imagination that they live beyond their text.

Fig. 1
figure 1

Degrees of characterhood

Full size image