Monroe C. Beardsley, Aesthetics: Problems in the Philosophy of Criticism, 1st ed., 1958; 2d ed., 1981.
The emergence and elaboration of speech-act theory (see Section 4 above) has made possible a more general and fundamental solution of the old problems about fictional, or imaginative, literature than that presented in the first edition (Chapter 9). The earlier view is not contradicted, but subsumed: nonassertive discourse (the making-up and telling of a tale, for example) is seen as a special case of discourse in which illocutionary action does not occur; and lyric poetry finds its place alongside prose narrative in the realm of fictional texts.
A number of recent thinkers have worked out versions of this theory of fiction, using somewhat different key concepts but converging on some central theses. Marcia Eaton ("Art, Artifacts, and Intentions" and "Liars, Ranters, and Dramatic Speakers") has proposed the special term "trans-locutionary act," in which the author transfers or attributes illocutionary actions to fictive speakers. Richard Ohmann ("Speech Acts" and later essays) holds that literature is discourse from which normal illocutionary force is withdrawn, leaving a special force he labels "mimetic." (His view is critically discussed by Mary Louise Pratt, Toward a Speech-Act Theory of Literary Discourse, Chapter 3.) John Searle ("The Logical Status of Fictional Discourse") speaks of the author as "imitating" illocutionary actions. Kendall Walton ("Fearing Fictions" and "How Remote are Fictional Worlds from the Real World?") holds that fictional sentences are used in a game of make-believe, much like pictures (see Section 5 above), and the horror-movie addict experiences not real fear but make-believe fear. Such terms as these convey important aspects of the truth (to which their authors have contributed much), but they are also, I think, misleading in different ways (see "The Concept of Literature"). I now believe that the most satisfactory account is that which makes use of the concept of representation (see B. H. Smith, On the Margins of Discourse, Part I). As I work out the theory, the composition of a fictional text is the representation (i.e., depiction) of an illocutionary action, or series of them, in basically the same sense in which a painter depicts a cow, or an actor on the stage depicts an act of punching (see "Aesthetic Intentions and Fictive Illocutions" and "Fiction as Representation"). Consider a clear-cut case, the end of A. E. Housman's poem:
The heart no longer stirred,
And say the lad that loved you
Is one that kept his word.
The speaker is dead, so it's obvious that no genuine illocutionary action is being performed; apart from that, the words are addressed to a specific individual, but not really anyone, so there is no "uptake," in J. L. Austin's sense. But the words sketch a possible series of fictive illocutionary actions: requesting a once-loved person to stop at the grave, reminding the addressee of the speaker's promise to forget her, noting ironically that he had to die to keep it.
One way in which a text falls short of recording an actual illocutionary action (but not so far short that it cannot be read as a representation of one) is by failing to connect with the real world in a certain way: the weddings that are narrated in the novel never took place, the names of characters do not belong to real persons, and so forth. The problem of characterizing this feature of most (not all) fictional works has been given a great deal of attention, since it involves the fundamental concept of reference, about which there has been much dispute. On one familiar view, whose best-known defender is John Searle (Speech Acts, Chapters 4 and 7), only existent entities can be referred to, so the name "James Bond" does not refer at all; and this helps to make the novels about James Bond fictive, while not preventing us from taking them as representations of someone giving us a factual history. On a less widely-held, but in my judgment more satisfactory, view, when we say that the novel is about James Bond, we should understand this to be saying that the name refers to James Bond, who, although nonexistent, has the remarkable qualities and adventures reported of him. Otherwise, for example, we will not be able to say that two different occurrences (tokens) of the (type) name "James Bond" in the novel refer to the same person (see Gerald Vision, "Referring to What Does not Exist"). But whether we choose to treat the name as referring, or as merely purporting to refer, will apparently not affect the general theory of fiction as representation.
Some philosophers, however, have held both that names refer and that they must refer to something -- but since fictional names do not refer to entities in the actual world, they must then refer to entities of some other sort, entities enjoying a special "mode of existence" or "mode of being." In recent years this Meinongian ontology has been explored in a number of technical essays, notably by Terence Parsons (see "A Meinongian Analysis of Fictional Objects"): but the necessity for multiplying basic ontological categories has not yet become evident to me, since it raises a great many metaphysical problems about the relationships among the various modes and epistemological problems about our knowledge of them. In some sense we all find it convenient to talk about the "world" of a novel and it seems evident that we can say important things, even make aesthetically important discoveries about these worlds -- about hidden events, mysterious motives, and so forth. The sentences in the novel, as composed by the author, are evidently neither true nor false; Virginia Woolf is not reporting on something in the actual world that her sentences could be true or false of, but is using these sentences to create a new (fictional) world. This point is well made by Laurent Stern ("Fictional Characters") and J. O. Urmson ("Fiction"); and it is in keeping with the proposal made above that in writing these sentences, the author was depicting, but not performing, illocutionary actions. But the sentences we utter about the work -- when as critics, for example, we describe Mrs. Dalloway or Lolita -- can indeed be true or false. But then what are they true or false of?
This question invites -- and has certainly aroused -- consideration of the semantics of statements about fictional worlds: what makes them true or false, and hence what sorts of entity, projected by the literary text, these statements are about. Some interesting proposals have been made and formulated in careful detail. Nicholas Wolterstorff ("World of Works of Art") takes as basic the concept of state of affairs (including events), which make up a world, whether fictional or actual. An author indicates some states of affairs by the sentences he writes, but many others are also discovered by elucidation; Wolterstorff provides an excellent account of two general principles that might be applied in such elucidations. All states of affairs exist, but not all actually occur: Huck Finn's having a raft is fictional, in that it is not part of the actual world. Fictional worlds are also incomplete in certain ways, since, for example, if Huck Finn's existing were to occur, he would have to have a definite height; but since this state of affairs belongs to the novel, not to the actual world, there need be no state of affairs such as Huck Finn's being n inches tall.
An account that in interesting ways is parallel to Wolterstorff s, although employing different basic concepts, has been given by David Lewis ("Truth in Fiction"). He treats descriptions of fictional characters as abbreviations, tacitly prefixed by the operator "In such-and-such a fiction. . . ." and considers some plausible principles for assigning truth conditions to such extended sentences. One main problem -- also dealt with by Wolterstorff -- is that of providing for true statements about the work's world that are only implicit and have to be elucidated: that (to take Lewis's example) Holmes lived closer to Paddington Station than to, Waterloo Station, a fact about Holmes that we can legitimately infer, given the map of London, even though it is not stated in the A. Conan Doyle stories. Lewis shows how, using concepts from possible-world semantics, one can rigorously state two alternative principles, one relying (roughly) on what happens to be true of the actual world, the other on; what is generally believed about it, to supplement the explicit sentences of the stories; he leaves it to the literary theorist to decide which is the more satisfactory principle for criticism.
The question of the ontological status of fictional characters is answered in very different ways by two other important essays. Robert Howell ("Fictional Objects") has given a detailed criticism of some attempts to analyze statements about fictional characters, and has suggested (somewhat tentatively, and as yet without full development) his own conclusion: that the name "Anna Karenina," for example, refers to "an identified, well-individuated (though nonactual) object." One of the major difficulties with such a view is that real persons get their names through a process of assignment (registration of a birth certificate, for example) , and the name keeps its connection with the individual and identifies that individual, as it is taught to others. No such causal process is available in assigning the name "Anna Karenina": we have only the various properties given her in the novel; and no matter how many of these there are they could still all belong to an infinite number of distinct individuals, so they do not individuate and identify a unique person. Howell believes that a substitute individuation is provided by Tolstoy's creative activity, his imagining and singling Anna out for focal attention and locating her uniquely in the world of the novel.
Peter van Inwagen ("Creatures of Fiction") argues that fictional characters exist and their names refer; but he holds that they are a subclass of what he calls "theoretical entities of literary criticism," a category that includes such other constructs as plots, meters, patterns of imagery, and genres. He makes a crucial distinction between (1) properties of such "creatures of fiction," strictly speaking: having been created by Samuel L. Clemens, being an example of a certain kind of literary anti-hero, having been misinterpreted by Leslie Fiedler, and so forth; and (2) properties the work "ascribes to" characters (in a technical sense of this term): being a boy, being illiterate, having a raft. Van Inwagen shows how some persistent puzzles about fictional entities can be resolved by adhering to this distinction.