Porter Institute for Poetics and Semiotics

Possible Worlds and Accessibility Relations: A Semantic Typoloty of Fiction Author(s): Marie-Laure Ryan Reviewed work(s): Source: Poetics Today, Vol. 12, No. 3 (Autumn, 1991), pp. 553-576 Published by: Duke University Press Stable URL: http://www.jstor.org/stable/1772651 . Accessed: 04/04/2012 13:29
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Duke University Press and Porter Institute for Poetics and Semiotics are collaborating with JSTOR to digitize, preserve and extend access to Poetics Today.

http://www.jstor.org

PossibleWorlds and Accessibility Relations: A SemanticTypologyof Fiction

Marie-Laure Ryan

In recent years the semantics of possible worlds and literary theory have enjoyed a promising cross-fertilization. While philosophers have invoked the concepts of "book" and of "story" to explain what a possible world is (Adams 1979 [1974]; Plantinga 1979 [1976]), literary theorists (among them Dolezel 1976; Vaina 1977; Eco 1979; Maitre 1983; Pavel 1986) have developed a textual semantics based on the idea that the semantic domain projected by the literary text is a nonactual possible world or an alternative possible world (henceforth abbreviated as APW). This assimilation satisfies our intuition that a text of literary fiction refers to nonexisting objects located in imaginary worlds, but the question of the possibility of these worlds remains problematic, since "everything goes" in a fiction. If everything goes, there is no such thing as an impossible world, and the use of the conceptual framework of possible-worlds semantics in literary theory becomes rather superficial. To avoid this trivialization we must address the question of what makes a world possible by exploring the various types of accessibility relations through which APWS may be linked to the actual world or AW.The purpose of this paper is to develop the concept of "accessibility relation" into a system of semantic classifications which should be relevant to the question of genre. Fictionand Possible Worlds: A Definition The formal basis of the theory of possible worlds has been defined by Saul Kripke (1963: 84) as "an ordered triple (G, K, R) where K is
Poetics Today 12:3 (Fall 1991). Copyright ? 1991 by The Porter Institute for Poetics and Semiotics. ccc 0333-5372/91/$2.50.

554

Poetics Today 12:3

a set [of objects]," G is one of the objects belonging to K, and "R is a ... relation" defined over the members of K. Kripke interprets K as "the set of all 'possible worlds,' G [as] the [actual] world," and R as the relation of relative possibility or accessibility. In the most widespread and intuitive application of this model to literary semantics, the opposition between the AWand its alternatives is regarded as constitutive of the distinction between fiction and nonfiction: while the nonfiction text describes the actual world, the fictional text refers to a non-actual possible world. This characterization misses, however, the sense in which the opposition actual/non-actual is itself internalized within the semantic domain. Among the facts described by the text, whether fictional or not, some are presented as actual and others as possible or counterfactual. The semantic domain of the text is not an individual world in a modal system, but is itself a system of worlds centered on what I shall call "the textual actual world" (henceforth abbreviated as TAW). The mental representations produced by the characters (beliefs, wishes, plans, obligations, dreams, fantasies, and literary productions) function as the APWS of the textual system of reality. But if all texts project a system of worlds, that is, a modal universe, how can the difference between fiction and nonfiction be characterized in the framework of possible-worlds theory? My answer to this question is rooted in the definition of the distinction between the AWand its alternatives. This difference has been viewed in two ways. In one definition (Rescher 1979 [1973]), the actual world exists objectively, while APWs are constructs of the mind, produced by such activities as dreaming, fantasizing, and forming beliefs or projections. In another of David Lewis (1979 [1973])-the actual world is definition-that we inhabit, and the term "actual"is indexical, as are the world simply the expressions "I," "you," "here," and "now."While in the first definition only one world may be called actual, in the second definition every possible world is the actual one from the point of view of its inhabitants. An interesting consequence of Lewis's position is that the on any of its planets and total universe of possibilities can be recentered thus contains an infinite number of different systems of reality. While it may be objectively the case that only one world exists independently of the human mind, we can through certain mental acts depart from this world, select another world as actual, and create through further mental acts a network of alternative possible worlds around the new center. This recentering occurs in dreams as well as in children's games of make-believe, and it constitutes the fundamental gesture of narrative fiction. If the gesture of recentering is indeed constitutive of fictionality, can be replaced by the folthe fallacious equation of fiction with APWS

Ryan * Possible Worlds

555

lowing generalization: nonfictional texts describe a system of reality whose center is occupied by the actual world (AW); fictional ones refer to a system whose actual world is, from an absolute point of view, an APW. The duplicity inherent to the fictional game reconciles the conflicting doctrines of Rescher and Lewis: while sender and hearer know that there is only one AW,they behave as if the foreign world at the center of the textual universe existed independently of the text, and as if it were the actual one. The concept of fictional recentering presupposes a distinction among three modal systems and among three distinct actual worlds. The first is our native system, centered in the Aw.The second system is the textual universe, the sum of the worlds projected by the text. At the center of this system is the TAW, the world formed by the facts presented as actual. As a representation proposed by the text, the textual universe must be distinguished from the system it represents, which I shall call the referential universe. And just as the textual universe is offered as an image of the referential universe, the TAW is proposed as an accurate representation of an entity external to itself, the textual reference world, abbreviated as TRW. The following axioms concerning the three types of actual worlds provide the basis for a possible-worlds definition of fictionality: 1. There is only one AW. 2. The sender (author) of a text is always located in the AW. 3. Every text projects a universe. At the center of this universe is
the TAW.

4. The TAW may or may not be similar to the AW. 5. The TAW is offered as the accurate image of a world, or TRW, which is assumed (really or in make-believe) to exist independently of the TAW.

6. Every text has an implied speaker. The implied speaker is the individual who fulfills the felicity conditions of the textual speech acts. 7. The implied speaker of the text is always located in the TRW. The three types of actual worlds open up the possibility of three types of divorce. The TAW may either accurately reflect or misrepresent the AW.The text may be presented either as a representation of the AWor as the image of an APW made actual through a recentering patible with the TRW, the world it is supposed to represent. The distinction above, between an actual sender (AS) and an implied speaker (Is), opens up a fourth possible divorce: either the sender stands behind the implied speaker and accepts responsibility for his claims or their respective beliefs differ and the actual sender dissociates him(TRW 0 AW). And finally, the TAWmay be either compatible

or incom-

556

Poetics Today 12:3

Table 1. The Modesof MimeticDiscourse

TAW = AW AW = TRW TAW = TRW AS = IS

Nonfictional accurate discourse Errors Lies Fiction True fiction

AW = the actual world

+ -+ -+

+ + -

+ + +

+ +

TAW= the textual actual world TRW= the textual reference world AS = the actual sender (author)

is

= the implied speaker (narrator)

self, either privately or publicly, from the implied speaker. These four distinctions generate the classification of mimetic discourse shown in Table 1. (By mimetic discourse I mean utterances that describe particular facts, make singular existential claims, and are intended to be judged true or false in a world external to themselves.) The first combination expresses the unmarked case of sincere and truthful mimetic discourse. The sender presents the text as a representation of the actual world, and this actual world, which functions as the world of reference, is correctly represented. Error differs from the unmarked case through the divorce between
the facts of the
AW/TRW

and their textual representation.

Since the

sender is unaware of this divorce, he shares the beliefs of the implied speaker. The difference between errors and lies resides in the duplicity of the sender, expressed by a minus sign in the fourth column. The sender is and their representaaware of the conflict between the facts of the AW conflict but he the hidden, tion, covertly playing the role of the keeps implied speaker. Fiction is characterized by the open gesture of recentering, through which an APW is placed at the center of the conceptual universe. This alternate possible world becomes the world of reference. The worldimage produced by the text differs from the Aw-except in the genre of true fiction, to be discussed below-but it accurately reflects its own
world of reference, the
TRW,

since the TRW does not exist indepen-

thus becomes indistinguishable dently of its representation. The TAW from its own referent. This phenomenon-which makes the concepts

Ryan * Possible Worlds

557

TAW and TRWinterchangeable when discussing fiction-explains the fashionable doctrine of self-referentiality of the literary text. The relocation of the sender of the fictional text to a new actual world necessitates the sacrifice of his identity. To become a citizen of the recentered system, he steps into the role of narrator; and to gain an audience, he extends to the hearer an invitation to follow him on his relocation. The "fictional pact" is concluded when the reader (hearer) becomes, in make-believe, a temporary member of the recentered system, thus shifting his attention from the AWto the TAW/
TRW.

AccessibilityRelations The original purpose of the possible-worlds concept was the definition of truth conditions for the modal operators of necessity and possibility. According to Kripke, a sentence expressing the possibility of a proposition p (e.g., "Nuclear war can be avoided") is true in a world W if there is at least one possible world W', such that W' is accessible from W and p is true in W. A sentence expressing the necessity of a proposition p (e.g., "Bachelors are unmarried"; "Nuclear war cannot be avoided") is true in W if p is true in all possible worlds W' accessible from W. In Kripke's semantic model, possibility and accessibility are equivalent concepts. Or rather, "possibility" is the concrete interpretation of what is meant formally by accessibility. A world is possible in a system of reality if it is accessible from the world at the center of that system. When philosophers speak of possible worlds, they usually interpret the accessibility relation as a logical one. A world is possible if it satisfies the logical laws of noncontradiction and of the excluded middle: (p OR -p) AND NOT (p AND -p) (A proposition must be true or false, and not both at the same time). According to the strictly logical definition of possibility, a world in which Napoleon dies on St. Helena and successfully escapes to New Orleans is not possible since it entails, "Napoleon did and did not die on St. Helena." But there is nothing inconsistent about either of these facts taken individually, and each has been verified in some logically possible world (the second one in a drama by the German expressionist playwright Georg Kayser). It can be argued that, under a logical interpretation of possibility, the only necessary propositions are mathematical truths ("Two plus two makes four") and analytic statements ("Bachelors are unmarried"). It is obvious, however, that the logical interpretation of the accessibility relation is not sufficient for a theory of fictional genres. Texts such as nonsense rhymes, surrealistic poems, the theater of the absurd, or postmodernist fiction may liberate their universes from the prin-

558

Poetics Today 12:3

ciple of noncontradiction. If we want to avoid the embarrassment of speaking about the impossible possible worlds of fiction, we must accept a much wider range of accessibility relations. Some of these will be looser than the logical laws allow, others more constrained: in historical novels, for instance, the TAW entertains much closer relations to the AWthan the relation of logical compatibility. These closer relations determine the semantic difference between the historical-novel genre and other types of texts obeying the law of noncontradiction, such as fairy tales and science fiction. Since a text projects a complete universe, not just an isolated planet, two domains of transworld relations should be distinguished: (1) the and trans-universe domain of the relations linking the AWto the TAW, own to its TAW the of the relations (2) the intra-universe domain linking alternatives. The relations of the first domain determine the degree of resemblance between the textual system and our own system of reality, while the relations of the second determine the internal configuration of the textual universe. Or, to put it another way: trans-universe relations function as the airline through which the sender reaches the world at the center of the textual universe, while intra-universe relations make it possible for the members of the TAW to travel mentally within their own system of reality. In the following discussion, I will but the conceptual repertory mainly focus on the relations AW/TAW, which describes trans-universe relations may also be applied to the intra-universe domain. In decreasing order of stringency, the relevant types of accessibility include relations from the AWinvolved in the construction of the TAW the following: A. Identity of properties (abbreviated as A/properties): TAW is accessible from AWif the objects common to TAW and AWhave the same properties. B. Identity of inventory (B/same inventory): TAW is accessible from
AW

if TAW and AW are furnished by the same objects.

TAW is
TAW'S

C. Compatibility of inventory (C/expanded inventory):

sible from AW if

acces-

inventory includes all the members of AW,

as well as some native members.1

in1. One could think of an inverse relation of compatibility,by which the TAW'S ventory would be a subset of the Aw'sinventory, or would intersect with it, but this relation is not productive in the semantics of fiction. The reason lies in an interpretive principle which I take to be constitutive of fictional communication: the principle of minimal departure (Lewis 1978; Ryan 1980). This principle states that we reconstrue the world of a fiction as being the closest possible to the AW, given the information provided by the text. If the fictional text mentions a city named Paris, we automatically assume that this city represents all the properties of the real-world Paris, unless these properties are explicitly denied by the text.

Ryan * Possible Worlds

559

D. Chronological compatibility (D/chronology): TAWis accessible from AWif it takes no temporal relocation for a member of AW to contemplate the entire history of TAW. (This condition means that TAW is not older than AW,i.e., its present is not posterior in absolute time to AW'spresent. We can contemplate facts of the past from the viewpoint of the present, but since the future holds projections rather than facts, it takes a relocation beyond the time of their occurrence to regard as facts those events located in the future.) is accessible from AW E. Physical compatibility (E/natural laws): TAW if they share natural laws. F. Taxonomic compatibility (F/taxonomy): TAW is accessible from AW if both worlds contain the same species, and the species are characterized by the same properties. (F usually follows from E, but some divorces of taxonomic and physical compatibility do occur and will be discussed below.) G. Logical compatibility (G/logic): TAW is accessible from AWif both worlds respect the principles of noncontradiction and of excluded middle. H. Analytical compatibility (H/analytical): TAW is accessible from AW if they share analytic truths, i.e., if objects designated by the same words have the same essential properties. I. Linguistic compatibility (//linguistic): TAW is accessible from AWif the language by which TAW is described can be understood in AW. The following discussion of genre and accessibility relations is summarized in Table 2. A combination of the relations "A/properties" and "B/same inventory" (which between them entail all other relations) makes the textual universe similar on all points to our own system of reality.2 Absolute
If the TAWhas Paris, by a law of geographic solidarity it must have France; if it has France, it must have the geography of the entire world, to which may be added the locations native to the TAW. has Napoleon, its And, similarly, if the TAW inventory is implicitly composed of all the individuals who ever lived in the AW. Excluding a single historical individual from the TAWwould amount to postulating a gratuitous departure from the AW.The only case in which the inventory of the TAW could exclude certain historical individuals is where the existence of these individuals would be incompatible with the facts asserted in the text. For instance, a work of historical confabulation might describe Napoleon as remaining married to Josephine until his death on St. Helena. In this case the inventory of the TAW would not include the legitimate son of Napoleon by his second wife, MarieLouise. Another example of intersection between the populations of the AWand the TAW is the rather strange case of 1984, to be discussed below. 2. If two systems have identical actual worlds, they are identical as a whole, since their APWSoriginate in actual mental events. If their APWS differed, so would the mental acts, and so would their actual worlds.

560

Poetics Today 12:3

Table 2. Genre and Accessibility Relations A Accurate nonfiction True fiction Realistic & historical fiction Historical confabulation Realistic ahistorical fiction Anticipation Science fiction Fairy tale Legend Fantastic realism Nonsense rhymes Jabberwockyism Concrete poetry
+t +-

B
+ +-

C
+ +

D
+ +

E
+-

F
+ + +

G
+ +

H
+ +

I
++

+
-

--

+ +

-+

--?

++t

+ +I

+ +

+-

--

+ +

++
+-

+1*C

+1-+
+/+

+
? +

+-

+f-

+
-

+1*
-

-I-

I+

--

#
G: when the laws of logic no longer taxonomic compatibility logical compatibility analytical compatibility linguistic compatibility

*: nonapplicable because of a "-" on C #: nonapplicable because of a "-" or "?" on hold, the concept of time loses any meaning. F = A = identity of properties G = B = identity of inventory H = C = compatibility of inventory I = = D chronological compatibility E = physical compatibility

compatibility with reality is, of course, the ideal of nonfictional texts presented for the sake of information, such as works of history, journalism, and biography. If the receiver decides that the sender's intent is informational, and that relations A and B obtain, he will complete his representation of reality on the basis of the new information he gathers from the text. The invitation to use the text in such a way is
what makes the AWthe referent of the TAW.

Ryan * Possible Worlds

561

Can the relations A/properties and B/same inventory hold in fiction as well? This would imply that the sender of the text recenters the system of reality in a world which is, on all points, similar to AW. This seems not only pointless, but also impossible. Fictional universes always differ by at least one property from our own system of reality: even if the sender of the fictional text pretends that everything is exactly the way it is, the world he selects as the center of the fictional universe differs from the actual world in that the intent and act of producing a fiction is a fact of the latter but not of the former. (Similarly, when children pretend that everything is the way it is, the real world differs from the world of make-believe through the presence of children playing that very game.) On all points other than its own existence as fiction, however, a fictional text may offer an exact reproduction of reality. Novelists are aware of the possibility when they warn the reader that any resemblance to actual individuals and events should be regarded as purely fortuitous. But as the increasing popularity of what I shall call "true fiction" indicates, a fictional universe may be deliberately conceived and presented as an accurate image of reality. The difference between nonfiction and true fiction is that the former claims to represent reality from but almost like AW. True fiction includes such mimetic practices as dramatized history, romantic biographies, and what has paradoxically come to be known as "nonfiction novels," i.e., stories about true facts which use the techniques of narrative fiction. (The best-known example of this genre is Truman Capote's In ColdBlood.) True fiction exploits the informational gaps in our knowledge of reality by filling them with unverified but credible facts for which the author takes no responsibility (as would be the case in historiography). The textual world is epistemically accessible from the real world, insofar as everything we know about reality can be integrated to it. In a romanced life, for instance, the narration respects all available historical information about the hero, but it completes this information with undocumented dialogues and reports of private thoughts which could conceivably have occurred as described. In a nonfictional text, these details would have to be presented in a hypothetical mode, that is, as true of some set of possible worlds to which the real world may or may not belong. The point of presenting the text as a fiction is that unverifiable facts can be directly asserted for the TAW without being asserted for the AW, and therefore, without compromising the credibility of the author. While the undocumented facts of romanced lives, dramatized history, and "nonfiction novels" exclude a strict application of principle A/ properties, we find a much closer adherence to it, at the expense of B/ same inventory, in the journalistic practice exemplified by the Ralphitself (TRW = AW), while the latter represents a world (TRW) distinct

562

Poetics Today 12:3

and-Wanda dialogues of Time magazine. (Ralph and Wanda are an imaginary couple who report and discuss the latest theories of sexual behavior.) Other examples of this genre of true fiction are Plato's dialogues and Rousseau's Prosopopee de Fabricius in Discourssur les lettres et les arts. In these texts all the facts are (ideally) verified, but the speech act through which they are presented is imaginary, either because the speaker does not belong to the inventory of the actual world (Ralph, Wanda) or because he never actually uttered the words attributed to him (Socrates, Fabricius). Since the speech act is imaginary, these texts are not uttered from within the AW,but involve the relocation that constitutes fictionality. If we relax A/properties, but maintain B/same inventory, we get imaginary stories about real people. An example of this category is the legend of George Washington and the cherry tree. In the nonfictional domain, this combination of accessibility relations is exemplified by the sensationalist stories of tabloids: "President Truman Inspected UFO Crash in 1947." The converse of such cases is illustrated by realistic and historical novels, such as War and Peace, the Sherlock Holmes stories, or The French Lieutenant'sWoman:A/properties are maintained as far as logically possible, but B/same inventory is replaced by C/expanded invencontains some individuals who have no counterpart in tory. The TAW the AW(Natasha, Sherlock Holmes, Sarah Woodruff), but otherwise presents the same inventory and the same geography as those of the AW at the same point in time. The properties of the common members are the same for both worlds: the London of Sherlock Holmes is the capital of England, and the names of its streets are identical to those of the real London. The Napoleon of Warand Peace was born in Corsica in 1769, the son of Charles Bonaparte and Laetitia Ramolino, and he has (had) twelve brothers and sisters. These facts may not be directly relevant to the plot of the novel, but they play an oblique role in assessing the truth-value of interpretations, since all valid statements about the textual universe must be compatible with them. When A/ properties and C/expanded inventory are respected, the only difference between the members of the AWand their counterparts in the TAW resides in their interactions with the members native to the TAW: the London of Sherlock Holmes has the property of a resident named Sherlock Holmes; the real London of the late nineteenth century does all not. Logically speaking, A and C make the AWa subset of the TAW: but propositions true propositions in the AWare also true in the TAW, are indeterminate in the AW concerning individuals specific to the TAW two-value in a false, (or system). In a genre that one might call historical confabulation, A/properties are much more openly transgressed than in the preceding class.

Ryan * Possible Worlds

563

includes the inventory of the AW, Once again, the inventory of the TAW but the properties of the common members differ in ways not necessarily arising from their involvement with non-common members: Napoleon escapes to New Orleans, Hitler wins the war, and Anne of Austria foolishly gives her lover, the Duke of Buckingham, the jewelry she had received as a present from her husband, King Louis XIII. In this situation, some propositions will be true in the TAW and false in the AWeven under a three-value system. When C/expanded inventory no longer holds, but, from D/chronois located in a geographilogical on, all other relations still do, the TAW cal and historical no-man's-land. The laws of nature are in force, and the TAW is populated by the same kinds of objects as the AW,but the representatives of the classes are different individuals. None of the This rather proper names used in the AWhas reference in the TAW. unusual combination of relations creates the eerie atmosphere of the taxonomically ordinary yet absolutely foreign world of Kafka's novels The Trial and The Castle. Severing the relation D/chronology results in either anticipation novels or science fiction, depending on which other relations are maintained. The point of anticipation novels is to show what may become of the actual world, given its present state and past history. For the demonstration to be convincing, all relations other than B/same inventory and D/chronology must be in force. (B, in fact, could be maintained.) The London of Orwell's 1984 once had a king named George VI, a prime minister named Churchill, and was once involved in a war against Hitler. From today's point of view, however, lifting rule D is no longer necessary to contemplate a world dated 1984. Strictly speaking, 1984 is no longer anticipation, but a strange breed of imaginary history. The history of England follows a common course in the AW and the TAW up to 1950, then branches toward Margaret Thatcher in one world and Big Brother in the other. Margaret Thatcher does not exist in the TAW of 1984, and neither does any other inhabitant of the post-1950 AW.For the contemporary reader, the novel illustrates the rare case of an intersection between the AW'S and the TAW'S populations. But from Orwell's point of view, the AW's population was a subset of the TAW'S. The novel was composed around 1950, at approximately the time of this historical split, and it does not explicitly eliminate any historical character known at the time of the writing. In science fiction proper, the focus is on the changes brought about by technological advances. Since technology must respect mathematical and natural laws, relations E/natural laws, F/taxonomic, G/logical, H/analytical, and I/linguistic will be maintained, but all others may be severed. The TAW will then contain the same classes of objects as the but not the same individuals. An interesting problem occurs when AW,

564

Poetics Today 12:3

technological advances lead to interplanetary travel. In this case, the taxonomic repertory of the planet Earth remains that of the AWEarth, but the other planets may contain extraterrestrial beings. These planets are also part of the actual world of the textual universe: modal logic uses the term "world" metaphorically, and metaphorical worlds may encompass a plurality of literal worlds. (If they could not, the AW would be limited to the planet Earth, and the sun and moon would be deprived of actuality.) When interplanetary travel achieved through technological means leads to the discovery of extraterrestrial beings, and the AW still F/taxonomic is no longer in force, though the TAW travel could lead same laws. observe the (Alternatively, space physical to discovering planets where the laws of physics no longer held. This would be a case of split ontology, to be discussed below.) An opposite case of divorce between E/natural laws and F/taxonomy or in Marcel is found in the realistic fantasy of Kafka's Metamorphosis is TAW the In these Passe-Muraille. works, populated by the Ayme's The heroes broken. nature are the laws of same species as the AW, yet are not knights, dragons, and princesses, but ordinary people engaged in the familiar pursuits of everyday life. Yet these ordinary people can walk across walls or discover one morning, without excessive surprise, that they have been metamorphosed into an equally ordinary species of repulsive insect. A very productive situation in textual matters is a TAW linked to an AWonly by G/logical, H/analytical, and I/linguistic, and, optionally, D/chronological. Lifting F/taxonomy introduces fairies, ghosts, dragons, unicorns, and witches into the textual world, while lifting El natural laws makes it possible for animals to talk, people to fly, and princes to be turned into frogs. When D/chronology is still in force, is located in some mythical past, and its taxonomic similarity the TAW to the AWwill be limited to the classes of objects characteristic of preindustrial societies: cottages rather than condominiums, swords rather than guns, and horses as a primary mode of transportation. If D is will include computers and time-travel, robots lifted as well, the TAW and interplanetary vessels; and its heroes will be Spider Man and Wonder Woman, rather than knights and princesses. When E/natural laws and F/taxonomy are lifted, so usually is C/expanded inventory: fairy tales have their own geography and population. An exception to this convention is the genre legend (e.g., "How Paul Bunyan Created the Grand Canyon"): supernatural beings roam through the TAW, and miracles are common occurrences, but the main characters or A similar inclusion of the AW'S locations have counterparts in the AW. in some fantastic tales, such as found is and population geography Frankenstein, whose hero was born (as well as mentally conceived) in Geneva.

Ryan * Possible Worlds

565

Emancipation from the relation G/logical opens the gates to the realm of nonsense. As Susan Stewart (1978) observes, nonsense is characterized by its rejection of the logical law of noncontradiction. P and -p can be true, not just in separate worlds of the textual universe, but in its actual world as well. Transgressions of F occur not only in folklore forms, such as children's rhymes, but also in so-called postmodernist fiction (McHale 1987). When we read, in Robert Pinget's Libera, that a certain character is dead and, thirty pages later, that he is alive, the contradiction is not meant to be resolved by assigning the second assertion to an earlier time. The character is simultaneously dead and alive in the TAW. This radical break with the laws of logic should be distinguished from the much more common textual practice of presenting contradictory statements as possibilities, without introAn example ducing contradiction within the boundaries of the TAW. of a logic-preserving use of contradiction is the following rhyme: A bottle of pop, a big banana We'refrom SouthernLouisiana That's a lie, that'sa fib We'refrom Colorado. (Quoted in Stewart1978: 72) Here the speakers are not simultaneously from Colorado and Louisiana, but from Colorado in one possible world and from Louisiana in another. The text makes it impossible to decide between the two alternatives (for there is no reason to believe that the first statement is really a lie), and a blank world is left at the center of the textual
universe.

Other types of nonsense are produced by transgressions of H/analytical. Some texts are based on systematically denying some of the essential properties which define a concept. Consider this well-known French rhyme: Un jeune vieillard,assissur une pierre en bois Lisaitsonjournal plie dans sa poche A la lueur d'un reverbereeteint. (A young old man, sittingon a wooden stone, was reading a newspaperfolded in his pocket, under the light of a street light which had been turned off.) This text cancels the property "old" from vieillard, the property "mineral" from pierre, and the property "dark" from reverbereeteint. Each of these cancelled properties belongs to the definition of the word. But other definitional properties are left untouched: under vieillard we still understand a human being rather than a machine, under pierre a solid object rather than a fluid. A complete transgression of H/ana-

566

Poetics Today 12:3

lytical would lead to an obliteration of I/linguistic: if the entity named "horse" could have all the properties of a computer in the TAW, the AW and the TAW would not follow the same linguistic conventions, and the TAW would remain as inaccessible to the reader as the universe of a text in a foreign language. Linguistic incompatibility can also result from a lack of overlap in the taxonomic repertories of the AWand the TAW. If the species of the TAW differ radically from those of the AW,their names will be deprived of semantic content, unless the text offers its own lexical definitions. Taken as a self-sufficient entity, Lewis Carroll's poem "Jabberwocky" illustrates this type of obscurity: 'Twasbrillig, and the slithy toves Did gyre and gimble in the wabe All mimsywere the borogroves And the mome raths outgrabe. (Carroll1975 [1916]: 130) When read in the wider context of ThroughtheLooking-Glass,however, the poem becomes linguistically accessible through Humpty Dumpty's translations: "brillig" means "four o'clock in the afternoon," "slithy" means "lithe and slimy," and "toves"are "something like lizards, something like badgers, and something like corkscrews" (ibid.: 187). While "Jabberwocky" retains some taxonomic/linguistic overlap with the AW (there are jaws and claws and swords in this world, and presumably they have the same properties as what we call jaws and claws and swords in the AW),all such connections are severed in the following poem by Hugo Ball: gadjiberibimbaglandridilonni cadori gadjainagrammaberibabimbalaglandrigalassassa laulitalomini gadji beri bin glassaglassalalaulalonni cadorsi sassalabim Gadjamatuffm i zimzallabinbangligia wowolimaibin beri ban. (Quoted in Stewart1978: 92) Along with the last linguistic connection to the AWvanishes the possiWhat vanishes bility of knowing and saying anything about the TAW. with this possibility is the very notion of a textual universe. Undecidable Relations The evaluation of accessibility relations between the AWand the TAW presupposes the text's ability-that is, the implied speaker's willingWhen epistemic ness or authority-to establish the facts of the TAW. access to these facts is denied, the world at the center of the textual system fails to "solidify"-to borrow the felicitous expression of

Ryan * Possible Worlds

567

Felix Martinez-Bonati (1981: 115)-and accessibility relations become wholly or partially undecidable. Variations on this situation include: (A) The empty center. The text limits its assertions to worlds at the periphery, avoiding the representation of an actual world. This effect can be achieved by modalizing propositions with adverbs of possibility ("maybe," "perhaps") or by linking them through a logical operator (OR), so as to leave in doubt which of them holds true in the TAW. Both techniques are characteristic of the work of Georg Trakl, a poetry of the virtual, if ever there was. (B) The unknowable center. The text blurs the distinctions between the TAW and the worlds at the periphery (i.e., the private worlds of characters), either by leaving it unclear who is speaking or by preventing the reader from identifying the reference world of the sentences. In Robbe-Grillet's Dans le Labyrinthe,we never know for sure whether the text describes a factual reality or a character's dream world or hallucination. (C) Radical lack of authority. In texts such as Beckett's The Unnameable, the narrator undermines his authority by withdrawing previous statements as lies, without giving valid reasons to believe the denial rather than the original statement. The narrator's discourse is regarded as "just discourse," that is, as incoherent rambling, expressing an inner world of transient perceptions. Between the extremes of a completely solidified and a radically inaccessible TAW lies the possibility of a partially defined center. In works such as Robbe-Grillet's La Jalousie and Nabokov's Pale Fire, we know little that is definitive about the individual facts of the TAW, but the text nevertheless to outline the laws which this world manages by general is constructed. The TAWS of La Jalousie and Pale Fire are basically realistic worlds, respecting relations D/chronological through I/linguistic in the former, and C/expanded inventory through I/linguistic in the latter: fairies and time-travel, nonsense and strange jargons are obviously not possible in these domains. How do we gain an intuition of the principles by which these worlds are put together? We apprehend the TAWthrough its reflection in the minds of characters, and even though we do not trust the details of that reflection or cannot identify the reflecting mind, we assume that the mental image respects the basic configuration of the reflected reality. If the character's subjective view of the TAW is linked to the AWthrough a certain cluster of relations, we assume by a law of transitivity that the same relations hold
between the AWand the
TAW.

MultipleRelationsand Split Ontologies One set of accessibility relations is not always sufficient to categorize the actual world of a textual universe. The text may present what

568

Poetics Today 12:3

Thomas Pavel (1986) calls a "dual" or "layered" ontology: the domain of the actual is split into sharply distinct domains obeying different laws, such as the sacred and the profane in medieval mystery plays or the visible world (everyday reality) versus the invisible one (the Court, the Castle) in Kafka's novels (Dolezel 1983). Unlike the private worlds of the characters' mental constructs, "the Sacred" or "the Invisible" is not an alternative possible world located at the periphery of the textual system, but a complementary territory within the central world. In Kafka's novels, the TAW is split between a realistic sphere, obeying all relations except A/properties, B/same inventory, and C/expanded The cases inventory, and a sphere of undecidable relation to the AW. of myth and medieval mystery plays must be assessed from two different vantage points: the perspective of the believer in the sacred, who professes a dual ontology, and the perspective of the nonbeliever, who adheres to a unified, profane ontology. For the nonbeliever the sacred sphere in these texts is reached only by lifting the relations El natural laws and F/taxonomy, while the profane sphere respects these relations. But how does a member of the original community categorize the text? Claiming that he regards its TAW as globally compatible with E and F would miss the point that his own conceptual system is based on a dualistic ontology. For the believer, the "supernatural" belongs to "the possible in the actual," though not to "the possible in the ordinary." His conceptual system divides E and F into two subcategories, one applicable to the sacred and the other to the profane, and as consistent with divisions made he regards the divisions of the TAW within the AW. The discrepancy between the believer's and the nonbeliever's points of view demonstrates the historical relativity of particular assessments of accessibility relations, and their dependence on such explanatory models as scientific theories and religious revelation. A text may be judged as conforming to E/natural laws at some point in history and as breaking those laws at a later point: in the Middle Ages, a story about witches could be told as true of the AW. Conversely, a text such as Jules Verne's 20,000 Leagues Under the Sea broke D/chronology for the nineteenth-century reader, but the passing of time and the invention of submarines have made it fully compatible with these relations for the modern reader. From a contemporary perspective, the text's semantic type is more akin to the genre of the adventure thriller than to science fiction, its original genre. I would not, however, go so far as to say that the text has shifted genre: as long as the origin of 20,000 Leagues Under the Sea is remembered, the reader will regard it as science fiction (just as 1984 will forever remain a novel of anticipation).

Ryan * Possible Worlds

569

MultipleRelationsand the UbiquitousCenter The (actual) existence of texts with an empty center raises the question of the reverse case: a text which would absorb all possible worlds within the boundaries of the TAW. In the semantic universe of such a text, the center would be everywhere and the circumference nowhere, since the domain of the non-actual would be drained of its substance. While no such text has ever been written in the Aw, one does "exist" in a recentered system of reality: the novel of the Chinese author Ts'ui Pen, as described by an English scholar in Borges's short story, "The Garden of Forking Paths": In all fictionalworks, each time a man is confronted with several alternatives, he chooses one and eliminatesthe others; in the fictionof Ts'ui Pen, in this way, diverse he chooses-simultaneously-all of them. He creates, futures, diverse times which themselvesalso proliferateand fork. (Borges 1983: 26, emphasesin original) Under a narrow conception of accessibility, the forking paths of this thinkable but unwritable fiction lead into all the futures allowed by logical and physical laws. All the worlds respecting E/natural laws, Fl but since taxonomy, and G/logic will then be combined in the TAW, these worlds may be mutually contradictory, G will not hold for the TAW as a whole. Under the diversified notion of accessibility proposed in this paper, the forking paths may lead into worlds of any semantic is linked to type, and, by simultaneously actualizing them all, the TAW the AWthrough all existing subsets of relations. Intra-Universe Relations As already stated, a semantic universe consists of a plurality of worlds, and its semantic description requires a recursive application of the taxonomic system within its own confines. In Alice in Wonderland,for is a realistic world related to the AW instance, the TAW through all relations except A/properties andB/same inventory. (The passage through this world is too swift to decide whether or not C/expanded inventory holds.) From the world originally designated as the TAW, however, the text takes a trip to the dreamworld of Wonderland by lifting E/natural laws and F/taxonomy, and this dreamworld momentarily takes the place of an actual world through an internal gesture of recentering (as opposed to the external recentering through which Lewis Carroll makes the entire textual universe come into being). Through its internal recentering, the text differs both from standard realistic novels, in which dreamworlds exist only at the periphery of the textual universe (i.e., dreams are recounted as dreams, not lived as reality), and from the fantastic universe of fairy tales, in which E/natural laws and Fl taxonomy are broken in the central world of the system.

570

Poetics Today 12:3

To capture the semantic characteristics of a genre, it may be necessary to assess the peripheral worlds of the system in their relations to A case in point is the genre of the fantasboth the AW and the TAW. tic, as defined by Todorov (1975 [1970]). According to Todorov, the fantastic atmosphere arises from a hesitation between a rational and a supernatural interpretation of the facts. Typically, a character is confronted with events which cannot be accounted for by his model of "the possible in the actual." He therefore tries to explain these events away by consigning them to a peripheral world, such as that of a dream of the fantastic or hallucination. When compared to the AW,the TAW text breaks relations E/natural laws and, possibly, E/taxonomy, but the as respecting these relations. What the characters perceive the TAW hero originally believes to be possible in the TAW corresponds to what adherents to a profane ontology believe to be possible in the the AW's AW. At the end of the text, however, the character is forced to revise his model of reality by adhering to a dualistic ontology. In their initial state, the epistemic worlds of the characters conform to the AW but conflict with the TAW; in their final state, they are aligned with the TAW, but they deviate from the AW. The inner discrepancy which Todorov labels as fantastic stands in striking contrast to the epistemic homogeneity of the fairy tale: here and the supernatural is spontaneously accepted as part of the TAW, the character's representation of reality is not regulated by the laws of nature. A slightly different type of harmony between a supernatural AW and its reflection in the character's mind occurs in The Metamorphosis (a text which Todorov rightly excludes from the fantastic). Being transformed into a bug is, for Gregor Samsa, a totally unprecedented event, one neither foreseen nor explained by his private worldview and, therefore, not to be experienced by any other individual. Yet he has no choice but to face the evidence ("this was not a dream" is the first thought to cross his mind), and, in his representation of the TAW, the actuality of his metamorphosis is never called into question. Genre and AccessibilityRelations As the preceding discussion has suggested, accessibility relations are involved in the differentiation of genres. But the taxonomic classes yielded by computing the various combinations of relations do not necessarily correspond to the generic labels used in a given culture. In some cases they are narrower than in others. Tolstoi's Anna Karenina and Robbe-Grillet's La Jalousie both contain a "realistic" element, though the former respects C/expanded inventory and the latter does not: La Jalousie makes no reference whatsoever to either individuals On the other hand, pastoral romances are anyor locations of the AW. their system of reality can be reached through but realistic, yet thing

Ryan * Possible Worlds

571

the same accessibility relations as the universe of La Jalousie: in both cases the inventory of the TAW does not contain that of the AW,but the TAW respects the laws of physics and logic. In order to refine the categories provided by these various combinations of accessibility relations into a taxonomy corresponding to accepted generic labels, we must introduce additional factors of semantic diversification. I would like to propose three of these factors: thematic focus, stylistic filtering, and probabilistic emphasis. Thematic focus is the principle by which the text's setting, characters, and events are selected from the history and inventory of the textual universe to form a plot or a message. (I conceive of the textual universe as the repertory of events and existents out of which the plot or message is composed. Its history extends beyond the plot's beginning and ending, and its inventory contains more species and individuals than the cast of characters and the types they represent.) The generic labels "psychological," "detective," and "historical" novel all concern types of thematic focus within the systems of reality accessible by means of relations C/expanded inventory through I/linguistic. While thematic focus guides the selection of whatever is to be shown, stylisticfiltering determines the light in which these elements will be presented, the impression they will make on the reader. Generic labels, such as "comic," "tragic," or "idyllic," refer to various types of stylistic filtering. The distinction of the pastoral romance from La Jalousie within the set of physically possible TAWS involves both thematic focus and stylistic filtering: the former selects the bucolic as thematic focus and filters, out of the bucolic, the idyllic; the latter selects a landscape of colonial life and paints it in neutral colors. Probabilisticemphasishas to do with whether the text dwells on the mainstream or the marginal within the boundaries of possibility determined by the relevant accessibility relations. Through probabilistic emphasis, we can differentiate what Doreen Maitre (1983) calls "escapist fiction," such as adventure thrillers or historical romances, from the realistic novels of Dickens or Flaubert, even though both types respect the same set of accessibility relations. Escapist fiction depicts faraway places, glamorous life-styles, thrilling adventures, incredible coincidences, agonizing dilemmas, burning desire, everlasting passion, all of which are logically, economically, psychologically, and physically possible in the Aw, though highly unlikely. The generic labels used in a culture may involve various combinations of the three types of semantic criteria. (They may, of course, also cover nonsemantic features, such as formal constraints and pragmatic requirements.) "Detective" or "historical," when applied to novels, refers to a type of thematic focus which presupposes a certain cluster of accessibility relations. "Idyllic" is a type of stylistic filtering,

572

Poetics Today 12:3

"pastoral" a type of thematic focus, and the label "pastoral romance" covers both features. Some labels are ambiguous, oscillating between two types of criteria. "Realistic" is understood by some as referring to accessibility relations: a text is realistic if it respects all relations beginning with E/natural laws and if the facts it describes are economically and psychologically possible in the Aw. For others, the events depicted in the realistic text must also fall within the statistically probable. Still another use of "realistic" emphasizes thematic focus: the text is realistic if it concentrates on everyday life within the regions of the TAW. In this third sense, "realistic" no longer implies acceptance of E/natuor Ayme's Passe-Muraille can be said ral laws. Kafka's Metamorphosis to combine a fantastic and a realistic element. The label "fantastic" is another example of potential semantic polyvalence. In its broadest and most intuitive usage, "fantastic" is synonymous with transgression of E. But in the narrower definition proposed by Todorov, this transgression is not regarded as a sufficient condition. The fantastic text must create an epistemic uncertainty by making the relation AW/TAW at least temporarily undecidable with respect to E. If we accept this definition, then the label "marvelous" may be substituted for "fantastic" and applied to those texts in which the transgression of E/natural laws is taken for granted, such as legends or fairy tales. Expandingthe Repertory The preceding catalogue of semantically relevant accessibility relations is anything but definitive. The need for expansion will undoubtedly arise as more texts are processed through the model, as new genres come into being, or as we fine-tune the analysis of individual texts to distinguish them from other representatives of the same genre. The list of candidates for addition to the model includes the following relations. is accessible from the Aw if the TAW The TAW (1) Historical coherence: not only includes the Aw's population, but contains no anachronisms with respect to the AW.Through this relation, it becomes possible to distinguish standard historical narratives, as well as what I have called historical confabulation, from works of fantasy which allow the mingling of characters, objects, and preoccupations from different periods: Joan of Arc coming back into the modern world and tackling contemporary problems, say, or prehistoric man watching soap operas on television (see McHale [1987] on the creative role of anachronism in postmodernist fiction). is psychologically accessible from The TAW (2) Psychologicalcredibility: the AWif we believe that the mental properties of the characters could be those of members of the AW,that is, if we regard the charac-

Ryan * Possible Worlds

573

ters as complete human beings to whom we can relate as persons. The relation of psychological credibility can be broken in many ways: through the symbolic unidimensionality of allegorical figures; through the rudimentary inner life of fairy-tale or science-fiction characters; or through the madness of the marginal creatures who populate the theater of the absurd. (For madness to break the relation, it must be Presented in the context of a generalized to all members of the TAW. "sane" environment, madness is only an extreme point on the scale of psychological possibility.) When a text breaks the relation of psychological credibility, it usually breaks some other, more salient relation as well, which makes the specification of the psychology relation somewhat superfluous for the semantic description of the genre. Among the aforementioned examples, fairy tales also break E/natural laws, science fiction breaks DI chronology, and the theater of the absurd transgresses G/logic. But psychological credibility can be elevated from a redundant property to a distinctive feature by a text presenting an innovative combination of accessibility relations. Such a text could be, for instance, a fantastic tale combining supernatural events with a plausible portrayal of human psychology. The TAW is accessible from the AW if (3) Socioeconomic compatibility: both worlds share economic laws and social structure. By adding this relation to the catalogue, it becomes possible to distinguish the "realistic" world of Robbe-Grillet's La Jalousie, where at least some people work for a living, from the Edenic landscape of the pastoral romance, where the availability of goods is taken for granted. The main reason for including relations of psychological credibility and socioeconomic compatibility to the catalogue resides in their hermeneutic importance. A text that respects psychological credibility makes psychoanalytic theory literally applicable as an interpretive model, while a text that transgresses the relation can justify only a figurative application: that is, characters in fairy tales may allegorize the Oedipus complex, but they do not suffer from it. And, similarly, a text that respects socioeconomic compatibility makes Marxist doctrine available as a potential explanation, while a pastoral romance does not. (4) Categorial compatibility:By this label I mean to signify the respecting of distinctions between basic logical categories. Through this relation, it is possible to explain the semantic difference between those TAWS containing allegorical characters, such as Death and Beauty, and those TAWS excluding such entities. Insofar as an allegory is the incarnation of an abstract idea, it transgresses the categorial distinction between particulars and universals. Another example of categorial

574

Poetics Today 12:3

transgression is the statement that concludes the television program Sesame Street: "This program has been brought to you by the letter Z and the number 6." AccessibilityRelationsand Fictionality The preceding discussion reveals a close connection between fictionIn ality and the strength of the relations between the AWand the TAW. nonfictional texts, the breaking of such relations must be either concealed (deceit) or inadvertent (error). For the TAW to depart from the AW, the textual referents must consequently fall within a zone of disagreement as to whether or not they are covered by the relation: that is, lies cannot be told nor errors made about facts universally recognized as true. The nature of the various relations is such that the last ones listed create much greater unanimity than the higher-order ones. E, F, G, H, and I are consequently much less likely to be broken, and unknown to be so by either sender or receiver, than A or B. We all agree in principle on the laws of language and logic. If, in the reader's opinion, a text breaks these relations, he will assume that the violation was not only intentional, but meant to be recognized, and that consecan only be reached through a playful relocation to quently the TAW another system of reality. Our opinions about physical laws and taxonomic classes are less unanimous: some of us believe in ghosts, UFOs, ESP, miracles, etc. Even if the reader denies these entities in his personal representation of reality, the possibility remains that the sender regards them as real, and their occurrence in a text does not constitute an absolute sign of fictionality. Still greater is our disagreement concerning the inventory of the real world and the properties of its members. It is consequently easy for a text to misrepresent facts or to introduce nonexisting individuals while claiming nevertheless that the
TAW reflects the AW.

as measured by accessiThe distance between the AWand the TAW, indicator of fictionality, reliable a thus provides fairly bility relations, but not an absolute criterion. What looks like a surrealistic poem that breaks the logical law of noncontradiction could very well be an entry in the diary of a schizophrenic patient; what looks like a fantastic description of former lives could be the autobiography of a famous actress; and conversely, what looks like the genuine love letters of a Portuguese nun could be the invention of a seventeenth-century French author. The question of fictionality is decided neither by the semantic properties of the textual universe nor by the stylistic properties of the text, but is settled a priori as part of our generic expectations. We regard a text as fiction when we know its genre, and we know that the genre is governed by the rules of the fictional game. And we enter into this game when our concern for the textual system

Ryan * Possible Worlds

575

of reality momentarily displaces our existential concern for the affairs of our own native system. Referencesand Related Works
Adams, Robert Merrihew 1979 [1974] "Theories of Actuality," in Loux 1979: 190-209. Borges, Jorge Luis 1983 Labyrinths: Selected Stories and Other Writings (New York: Modern Library). Carroll, Lewis 1975 [1916] Alice's Adventures in Wonderland and Through the Looking-Glass (New York, Chicago, and San Francisco: Rand McNally). Dolezel, Lubomir 1976a "Narrative Modalities," Journal of Literary Semantics 5(1): 5-14. 1976b "Narrative Semantics," PTL 1: 129-51. 1983 "Intensional Function, Invisible Worlds and Franz Kafka," Style 17: 12041. Eco, Umberto 1979 "Lector in Fabula: Pragmatic Strategy in a Metanarrative Text," in The Role of the Reader: Explorations in the Semiotics of Texts (Bloomington: Indiana University Press). Kripke, Saul 1963 "Semantical Considerations on Modal Logic," Acta Philosophica Fennica 16: 83-94. Lewis, David 1970 "Anselm and Actuality," Nous 4: 175-88. 1978 "Truth in Fiction," American Philosophical Quarterly 15: 37-46. 1979 [1973] Counterfactuals, excerpted in Loux 1979: 182-89. Loux, Michael J. 1979 "Modality and Metaphysics," introduction to Loux 1979: 15-64. Loux, Michael J., ed. 1979 The Possible and the Actual: Readings in the Metaphysics of Modality (Ithaca: Cornell University Press). Maitre, Doreen 1983 Literature and Possible Worlds(London: Middlesex Polytechnic Press). Margolin, Uri 1988 "Dealing with the Non-Actual: Conception, Reception, Description," Poetics Today 9(4): 863-78. Martinez-Bonati, Felix 1981 Fictive Discourse and the Structures of Literature (Ithaca: Cornell University Press). 1983 "Towards a Formal Ontology of Fictional Worlds," Philosophy and Literature 7: 182-95. McCord, Phyllis 1986 "The Ideology of Form: The Nonfiction Novel," Genre 19: 59-79. McHale, Brian 1987 PostmodernistFiction (New York and London: Methuen). Pavel, Thomas G. 1986 Fictional Worlds(Cambridge, MA: Harvard University Press). Plantinga, Alvin 1979 [1976] "Actualism and Possible Worlds," in Loux 1979: 253-73. Rescher, Nicholas 1979 [1973] "The Ontology of the Possible," in Loux 1979: 166-81.

576

Poetics Today 12:3

Ronen, Ruth 1988 "The World of Allegory," Journal of Literary Semantics 17(2): 91-121. Ryan, Marie-Laure 1980 "Fiction, Non-Factuals, and the Principle of Minimal Departure," Poetics 9: 403-22. 1985 "The Modal Structure of Narrative Universes," Poetics Today 6(4): 717-56. Stalnaker, Robert C. 1979 [1976] "Possible Worlds," in Loux 1979: 225-34. Stewart, Susan 1978 Nonsense: Aspects of Intertextuality in Folkloreand Literature (Baltimore: Johns Hopkins University Press). Todorov, Tzvetan 1975 [1970] The Fantastic: A Structural Approach to a Literary Genre, translated by Richard Howard (Ithaca and New York: Cornell University Press). Vaina, Lucia 1977 "Les Mondes possibles du texte," Versus 17: 3-13. Walton, Kendall 1978 "How Remote Are Fictional Worlds from the Real World?" Journal of Aesthetics and Art Criticism 37: 11-23. 1983 "Fiction, Fiction-Making, and Styles of Fictionality," Philosophy and Literature 7(2): 78-87. Wolstertorff, Nicholas 1976 "Worlds of Works of Art,"Journal of Aestheticsand Art Criticism 35: 121-32.