A REPRESENTATIONAL ACCOUNT OF FICTIONAL CHARACTERS. (2024)

How to cite: Vecsey, Zoltan (2015), "A RepresentationalAccount of Fictional Characters," Analysis and Metaphysics 14:68-79.

1. Artefactual Theories of Fictional Characters

One comparative advantage of artefactual theories over their rivalsis their conformity with our pre-theoretic intuition according to whichfictional characters are created entities. Everyday consumers ofliterary works often assume that artistic writing is an essentiallyinnovative activity. On the most common view, persons like SherlockHolmes and Fyodor Karamazov, and places like Middlemarch and Lilliputare, in the strictest sense of the word, intellectual products of thehuman mind. Fictional characters are therefore supposed to have theirown life span: there is a certain point of time when they are broughtinto being by the creative acts of their authors and there may come atime when they simply cease to exist, perhaps, when the works in whichthey appear become physically destroyed and no one remembers them anymore. In this regard, fictional characters do not differ significantlyfrom other types of artistic products. It is not very surprising, then,that the folk ontogenetic conception of literary works takes thecreatedness of fictional persons and places as a rather obvious andtrivial datum.

Most proponents of the artefactual theory tend to regard the aboveonto-genetic truisms of the folk view as basically correct. At least twoelements of this view are thought to be suitable for serious theoreticalinvestigations. The first element coincides with the fundamental thesisof art ontological realism which states that there are fictionalcharacters in the overall inventory of what exists. The second elementattempts to give an answer to the question of how can there be suchcharacters. According to the folk explanation, fictional characters comeinto being as a result of creative writing processes.

Artefactualists such as Goodman (2004), Voltolini (2006) andThomasson (1999, 2003, 2010) argue, in harmony with the folk view, thatSherlock Holmes, Middlemarch and their likes are contingently existingcreated entities. Even if this thesis appears to be reasonable at firstglance, the artefactual theory has some counterintuitive features. On myview, artefactualism becomes counterintuitive precisely at that pointwhere it goes beyond the truisms of the folk theory. Let me explain whatI mean by this. (1)

In order to make their theory plausible, artefactualists have toprovide a detailed description about the process of authorial creation.If one regards a particular fictional text F as consisting of a set ofsyntactically individuated English sentences, then the question forartefactualists to answer is this: How can the author of F literallycreate various fictional entities merely by writing down or typing outthe sentences composing F?

As a first step in answering this question, artefactualists mayinvoke their favored account of intentionality. Thomasson (1996, 1999)follows in this respect the act-object theory of intentionality defendedfirst by the polish philosophers Twardowski and Ingarden. The act-objecttheory says that every intentional act involves an object and a content.The act is directed or related to an object and the content is what isthought or judged about that object. According to this approach,non-fictional sentences like (1) receive a quite straightforwardinterpretation:

(1) Gunter Grass smokes a pipe.

The object to which the underlying intentional act is related isGrass, and he is thought of as acting in a certain way, namely assmoking a pipe. Of course, Grass as a person exists independently fromany intentional act. And he would surely not cease to exist, if nointentional acts were directed towards him in the future. But nowconsider the fictional counterpart of (1):

(2) Sherlock Holmes smokes a pipe.

Suppose, for the sake of the argument, that (2) is the sentencetoken which contained the very first occurrence of the proper nameSherlock Holmes in Conan Doyle's oeuvre. (2) Presumably, when Doylehas written down this sentence, his intentional act were directedtowards a putative object. But given that Holmes did not existed before(2) was written down, we may question how this was possible. What wasthe intended object of Doyle's intentional act? If intentionalityis essentially relational, as the act-object theory assumes, then thereis a certain sort of circularity here. Doyle's intentional actpresupposed the existence of a particular act-independent object, but itseems that the presupposition in question was made by the act itself.(3) According to Thomasson (1999), this kind of intentional circularityis merely apparent. Those who demand an act-independent object in orderfor (2) to have a non-circular interpretation do not take into accountthe creative capacity of intentional acts. We should not deny thatHolmes did not exist before the tokening of (2). Prior to this time,there was no such person as Holmes and no other fictional text containedhim (or it) as a protagonist. It does not follow from this, however,that (2) has to be regarded as a product of an objectless intentionalact. Once we recognize that Doyle's intentional act was able tobring into existence its own object, we must also recognize that (1) and(2) have a parallel structure. The only difference is that while the actunderlying (1) is directed towards a previously existing(act-independent) object, the act underlying (2) is directed towards anintentionally created (act-dependent) object.

The observation that certain kinds of intentional act areontologically creative is of crucial importance for artefactualists.This is the observation on the basis of which they can argue thatfictional characters have extremely minimal existence conditions (cf.Thomasson 2003).

According to their view, only the following two conditions must besatisfied for a particular character to come into existence in afictional text F: (i) the author of F must perform an ontologicallycreative intentional act that is directed towards a previouslynon-existent object, and (ii) the sentence corresponding to the creativeact of the author must be included in the set of sentences composing F.

Our imaginary example seems to satisfy both of these conditions. Inthe foregoing discussion, we have supposed that (2) was the very firstthought about Holmes in the history of literature. So we have conceivedDoyle's intentional act as genuinely creative. We have alsosupposed that (2) was a sentence token which occurred in a fictionaltext written by Doyle. Since nothing more is required for charactercreation to succeed, we may conclude that in performing (2) Doyle hascreated a new fictional character.

Although Holmes is thought in (2) as smoking a pipe, and in ourreal world only persons are smokers, it would be folly to think thatDoyle created a person. Fictional characters are not denizens of thespatiotemporal world, hence they are not persons. They have a somewhatpeculiar ontological status. Their existence, as we have seen, dependsconstitutively on worldly processes and entities, like intentional actsand inscriptions, yet they are not material in any sense. In thisregard, characters belong to the same ontological category as marriages,nations, institutions and other artefacts. (4) All of thesesociocultural entities are contingently existing abstract objects. Theartefactual theory maintains, accordingly, that Doyle's creativemental act and the corresponding sentence token "Sherlock Holmessmokes a pipe" has brought into existence an abstractum.

But this cannot be the entire explanation, since artefactualistsare faced at this point with an immediate problem. If Holmes is anabstract object, then (2) can be paraphrased or translated, without lossof meaning, into (3):

(2) Sherlock Holmes smokes a pipe.

(3) The abstract object created by Conan Doyle smokes a pipe.

Of course, (3) is a nonsensical claim, one to which even the mostcommitted adherents of abstract objects could not assent. In attemptingto solve this problem, artefactualists may follow different strategies.First, they might object that the paraphrase is inaccurate because theproper name "Sherlock Holmes" and the definite description"the abstract object created by Conan Doyle" have differentmodal profiles. As a rigid designator, the name "SherlockHolmes" refers to the same (abstract) object in all possibleworlds. Definite descriptions are, however, non-rigid designators, so insome nearby worlds it is false that Holmes was created by Doyle. This isa well-known objection of the direct reference theory againstdescriptive sentential paraphrases and, as such, it has a standardanswer. For example, the modal difference between (2) and (3) may beeliminated by applying a suitably rigidified version of the definitedescription. Inserting the actuality operator @ or the sentence modifieractually in (3) enables the sentence to express the same descriptivecontent in all of its uses in other possible worlds. Or, to achieve thesame effect, it might be claimed that the proper name 'SherlockHolmes' expresses a rigidified descriptive content which remainsconstant in all of its possible tokenings. Artefactualists musttherefore find another way to solve the problem posed by the aboveparaphrase.

A better proposal is to say that, contrary to its surfaceappearance, (2) it is not an assertion in the usual sense of the word.In fact, this is what Schiffer (1996), Thomasson (2003, 2010) and otherartefactualists contend. It seems as if Doyle's original sentencetoken attributed a property to an (abstract) object. But this is just aformal illusion, so to say. Authors of fictional texts do not makegenuine assertions, they merely pretend to assert what they write. (5)If it has been made clear that Doyle merely pretended to attribute aproperty to Holmes, then (2) can be paraphrased only in a way that makesexplicit reference to this pretense. In applying the pretense strategy,the artefactualist can easily block the inference from (2) to (3) andsay that the appropriate candidate for paraphrasing Doyle'ssentence is (4) or something similar to it:

(4) Conan Doyle created an abstract object and pretended to assertthat it is a person who smokes a pipe.

The core idea, which most realist theories of fiction agree with,is to make a significant distinction between two kinds of assertion.Contemporary Meinongian theories provide perhaps the clearest example ofhow to motivate this distinction. On some neo-Meinongian views,fiction-internal assertions attribute nuclear or characterizingproperties to fictional objects. (6) Assertions within the storyattribute such nuclear properties to Holmes as smoking a pipe or beingtalented. These are the properties which we associate directly withHolmes when we read Doyle's story. Fiction-external assertions, incontrast, attribute extranuclear properties to objects. Among theextranuclear properties we find such properties as being fictional orbeing abstract. Thus, if we approach the story from an extra-fictionalviewpoint, we may attribute to Holmes the property of being an abstractobject. Other neo-Meinongians maintain that the distinction betweenfiction-internal and fiction-external assertions reflects the ways inwhich objects may possess their properties. While Holmes encodes theproperty of being a smoker, Gunter Grass exemplifies that property. Theterm "encoding" is meant here to refer to the fact that Holmeshas this property internally or as a constituent part of his (its) ownbeing. The term "exemplifying," on the other hand, is meant toindicate that Grass has the same property externally ornon-constitutively.

If we now look back to (2), (3) and (4), we will understand thereason why artefactualists must subscribe to thefiction-internal/fiction-external distinction. (7) They hold thatfiction-internal assertions are capable to create abstract objects; yetat the same time they deny that these objects are literally ascribed anyproperties in the texts in which they occur. The only option whichremains is thus to say that fiction-internal assertions merely pretendto make assertions. Nevertheless, when we literary theorists, linguistsor philosophers talk about these abstract objects we do not talk aboutthem in the scope of pretense. And this, in turn, presupposes that thereis a pretense-free, external perspective from which we can make genuineassertions about the features of fiction-internal discourse. AsThomasson (2003) remarks, fiction-internal occurrences of the propername "Holmes" are non-referential, but these non-referentialuses enable the name to refer to the character Holmes in externaldiscourses.

These considerations lead, in the end, to an artefactual theorywhich admits that authors of fictional texts create literally fictionalcharacters, but deny that the resulting texts contain literal assertionsabout these characters. This is, to say the least, a bit perplexing. Thefolk theory seems to be right in claiming that fictional texts do nothave such contradictory features. In engaging with Doyle's novel,it is a quite natural attitude to take sentences like (2) as telling ussomething about the personality and behavior of the character Holmes. Ofcourse, we do not have to give up the artefactual theory in itsentirety, simply because it conflicts with our ordinary ways of thinkingabout literary characters. This is especially so, if we realize that thesource of the conflict lies in the conceptual apparatus of the theory.In the next two sections, I will argue that a slightly modifiedunderstanding of artefactualism can provide a non-contradictory view ofour engagement with fictional texts and fictional characters withoutrelying on the traditional fiction-internal/fiction-externaldistinction.

2. A Representationalist Alternative

At first sight, the act-object theory of intentionality seems to bethe obvious choice to explain how authors can create new fictionalcharacters. Thomasson (1999) thinks of intentional acts as potentiallycreative, so she can convincingly argue that certain kinds of authorialact bring into existence new intentional objects. The resultingart-ontological view holds, then, that fictional characters exist asdependent abstracta. As we have already indicated earlier, the creationof Holmes is thought of as depending both on the ontologicallyproductive mental act performed by Doyle and the sentence token whichcorresponds to Doyle's act.

The second part of the latter claim, however, rests on aquestionable presupposition concerning the semantics of fictionalsentences. In order to understand correctly what this presuppositionamounts to, let us consider once again what the artefactualist has tosay about the ontological relation between characters and the sentencetokens on which their coming into existence and their continuedexistence is (partly) dependent. Given their realist ontologicalcommitments and their accompanying ideas about contingently existingabstract objects, artefactualists have to distinguish characters fromliterary texts in which their names are (first) mentioned. Althoughtheir existence conditions include facts about linguistic objects,fictional characters are supposed to be abstracta which are notlinguistic by their nature. But, then, what kind of abstract object arethey? Voltolini (2006) identifies them with compound abstract entities.On Voltolini's view, characters are constituted by an instantiatedmake-believe process-type and a fiction-internally ascribed set ofproperties. Thomasson who also denies the linguistic nature ofcharacters, proposes another possible account, namely that they areontologically similar to marriages, laws and other abstract culturalartefacts (cf. Thomasson 1999, 2003). From a Thomassonian point of view,the linguistic object--the token name "Holmes"--used by Doylefor expressing his creative thought and the resulting Holmes-artefactare related but distinct objects. They are related because they aredependent on exactly the same mental basis, namely on Doyle'sintentional acts. Yet they are distinct, because they are instances ofdifferent kinds of abstractum.

Unfortunately, these conceptions presuppose rather than explain theontological status of characters. The core problem may be stated in thefollowing way. The semantic content of Doyle's sentence token andthe Holmes-artefact are arguably different kinds of abstract object. Ifthe Holmes-artefact is (partly) dependent on but not identical with thesemantic content of the sentence token (2), then there must be a causalor inferential relation actually connecting them. But so far neitherVoltolini nor Thomasson have attempted to explain how this relationcould be defined in an artefactualist framework. Moreover, the availablelinguistic data do not seem to support such an explanation. (8)

I think the simplest way to get rid of this problem is to hold thatSherlock Holmes is nothing more than a linguistic construct. Let us saythat the sentence token "Sherlock Holmes smokes a pipe"represents Holmes as being a certain way. Doyle's sentencerepresents Holmes as smoking a pipe, so we may say that it offers us aparticular Holmes-representation. Note that the term"representation" is used here in a specific, non-relationalway. There is nothing new in this usage. In recent years, variousattempts have been made in the relevant philosophical and linguisticl*terature to develop a nonrelational interpretation of representation.(9) For example, Burge (2010) points out that representation cantranspire even in cases where there is nothing that could berepresented. In Burge's jargon, representing is alwaysrepresenting-as-of. A predicate or a proper name X has this underlyingstructure independently of whether there is a candidate for being arepresentatum of X. Sainsbury (2012) is of the same opinion when heremarks that the claim "X represents Y" does not imply thatthere is something, Y, such that X represents it. In this sense,representing Holmes is not equivalent with expressing a singularproposition containing Holmes as a constituent. A sentence token canrepresent Holmes as a smoker even if there is no such person as Holmes.Moreover, the Holmes-representation is interpretable without anyreference to language-independent objects, be they concrete or abstract.It is tantamount to say that fictional characters come into being asself-standing, nonrelational linguistic representations.

Now we can take a further step in this direction. Doyle'snovel represents Holmes in a great variety of ways--as being a smoker,as being a detective, as living on Baker Street, and so forth. Thus, ifDoyle is the author of the fictional text F, then Holmes can beidentified in F with the elements of the property set P, where P ={[p.sub.1], [p.sub.2],... [p.sub.n]}, and [p.sub.1], [p.sub.2],... and[p.sub.n] are the properties explicitly represented by the sentences ofF.

According to F, Holmes is a smoker. We know, however, that smokersoften have lungs that are gross and black. So Holmes may also have grossand black lungs. But F is absolutely silent about this matter. Frepresents him neither as having gross and black lungs, nor as havinghealthy lungs. This is not to say that Holmes is an incompleterepresentation, it means simply that he is not explicitly representedeither way. But some elements of the property set P may jointly implythat he has healthy lungs. Then, it would be certainly correct to statethat Holmes is represented implicitly as having healthy lungs. Comparethis with the case when someone claims that Holmes is a computerprogram. We would definitively reject this claim by saying that Holmesis represented neither explicitly nor implicitly as a computer programin F.

In order to decide such cases we may introduce the property set P*,where P* = {[p*.sub.1], [p*.sub.2],... [p*.sub.n]}, and [p*.sub.1],[p*.sub.2],... [p*.sub.n] are the implicit properties derived from Pthrough an appropriately designed interpretative method M. Since it isnot a trivial task to develop such a method as M, we must be content toassume that it can in principle be developed. Under this assumption, theHolmes-representation of F may be seen as completely determined by theproperty sets P and P*. Or, in other words, thinking about Holmes meansnothing else than mobilizing a non-relational representational complexcombined from certain elements of P and P*. This does not mean thatHolmes is a set-correlate of the union of P and P*, since we cannotascribe properties to fictional characters: characters cannot besubsumed under the traditional ontological category oflanguage-independent objects. The central claim is, rather, that Holmesexists in F as a maximal representational combination of the elements ofP and P*.

It is important to keep in mind that the terms "property"and "property set" are not used here in their neo-Meinongiansense. We can say, if we like, that in writing his novel Doyle ascribesa particular property (say, being a smoker) to his protagonist. But thisidiom is to some extent misleading, since in the context of F,property-ascribing predicates like "is a smoker" can acquireonly non-relational interpretations. There is no language-independentobject (Holmes) to which the text may ascribe a language-independentproperty (being a smoker). Thus, in contrast to the neo-Meinongiantheory which conceives property attribution as a relational act, thepresent view acknowledges properties only at the level of non-relationallinguistic representation.

At this point, the following objection may be raised. If Holmes isa linguistic representation created by Doyle, then we are back where westarted. The problem now seems to be that we are forced to accept (5) asa correct paraphrase of (2):

(5) The Holmes representation created by Conan Doyle smokes a pipe.

Undoubtedly, (5) sounds as bad as (3). But (5) is misleading inthat it makes it seem as if the property of smoking a pipe were aproperty of the Holmes representation. This cannot be the correctinterpretation. Being a member of P, the property of smoking a pipe isin fact a constitutive part of the Holmes representation of F. Hence itis incorrect to think that Doyle's creative act ascribed a propertyto an abstract linguistic construct. We can avoid this misleadinginterpretation by focusing on the physical basis of theHolmes-representation:

(6) The sentence token created by Conan Doyle represents Holmes assmoking a pipe.

(6) reveals two things about the circ*mstances under which thefictional character Holmes was created. First, Doyle's intentionalact was ontologically productive because it created a particular mentalrepresentation. One may conjecture that this mental representation wasthe original content of his creative act. This content can be seen, inturn, as the way in which he created that mental representation. The"way" of the creative act can be described adequately by anadverbial phrase: Sherlock Holmes was represented in Doyle'sthought in a smoking-a-pipe-wise. (10) Or, more correctly, we can saythat Doyle performed a mental act in aSherlock-Holmes-smokes-a-pipe-wise. This particular way of thinking waswhat individuated the content of the creative act in question. Second,in order to successfully expressing his thought in a publicly accessibleform, Doyle must have written down the sentence "Sherlock Holmessmokes a pipe." Hence it is not unreasonable to think that thistoken sentence acquires its content from the mental act that immediatelyprecedes it. And given that the mental act represents Holmes in anon-relational manner, the token sentence must represent Holmes in anonrelational manner, too. The result of the writing act can thus beregarded as the physical endpoint of character creation. There is noneed for supplementing this account with a further inferential step:(2), just as it is in itself, is able to bring into being Holmes as apublicly accessible non-relational representation.

3. In Closing

It thus seems that there is little reason to reject the artefactualtheory of fictional characters. Paradigmatic examples of fictionalcreatures like Sherlock Holmes, Fyodor Karamazov or Madame Bovary arecontingently existing abstract objects. Artefactualists arefundamentally right in their emphasis on the created nature of theseentities. But I think we remain closer to the folk view if, contrary tostandard artefactualism, we regard fictional characters as linguisticrepresentations.

One obvious advantage of this revised version of the artefactualtheory is that the terminological distinction between fiction-internaland fiction-external assertions becomes superfluous. Consider first theeasy case in which someone tries to understand what a fiction-internalsentence asserts. When we read the sentence "Holmes is adetective," we come to know that Holmes is represented in the textas being a detective. This is a natural-sounding interpretation. A muchharder case is when we read the apparently fiction-external sentence"Holmes is a fictional character." Does the interpretation ofthis sentence actually involve an external perspective? It is notnecessarily so. It just needs to be recalled what the term"fictional character" means in the present context. Accordingto the present account, fictional characters are linguisticrepresentations. And the term "representation" is intended tomean, in turn, "to be represented in a certain way." Thus, thesentence "Holmes is a fictional character" informs us thatHolmes is represented in the text as being a certain way. It does notprovide a particular representation; rather, it reminds us thatDoyle's character has a representational nature. Since this seemsto be in agreement with the former result, one may conclude that Holmesis seen through the same interpretative perspective in both cases.

A further and related advantage of the present proposal is that itdoes not entail the controversial claim according to which authors offictional texts merely pretend to make assertions. The pretense approachto authorial assertions is used by Thomasson and other artefactualistsas a last resort. Its only role is to assure us that fiction-internalassertions do not concern abstracta. But once one thinks of thefiction-internal/fiction-external distinction as theoreticallyinsignificant, even this limited role seems likely to evaporatealtogether. Thomasson and her followers are right in saying thatauthorial assertions do not ascribe properties to abstract artefacts.From a representationalist point of view, however, one can provide acomplete explanation for this phenomenon without involving the notion ofauthorial pretense.

The present proposal requires only a light revision of the standardartefactualist framework. It leaves untouched the abstract status offictional characters, but makes an attempt to state more precisely whatkind of abstract objects they are. The core idea is that fictiveindividuals in literary artworks like Sherlock Holmes, Fyodor Karamazovand Madame Bovary are to be thought as non-relational linguisticrepresentations. Though I am fairly well convinced that this idea allowsa simpler and more uniform theoretical treatment of fictional discourse,some important details may be lacking. (11) Arguably further analysis isrequired to account for the naturalness of the representationalist view:is it indeed the case that everyday consumers of literary works think offictional characters in terms of representation? But this will be thesubject of another paper.

NOTES

(1.) The most elaborated and consistent version of the artefactualtheory is to be found in Thomasson (1999). In responding to her critics,Thomasson has revised some of her original ideas over the years (cf.,Thomasson 2003, 2010). In what follows, I shall focus primarily on herview in my discussion.

(2.) In fact, the proper name Sherlock Holmes occurs for the firsttime in the novel A Study in Scarlet (1887/2004). The first sentencewhich contains the name is to be found on page 5: "You don'tknow Sherlock Holmes yet."

(3.) For similar observations, see Yagisawa (2001) and Brock(2010). The charge of circularity is not completely unfounded, but thisis irrelevant for the present purposes. I shall argue for a version ofartefactualism that avoids this problem entirely.

(4.) See Thomasson (2003). Voltolini (2006) provides also aninsightful analysis of the ontological category of created abstracta.

(5.) It is worth to note that artefactualists borrow the idea ofpretense from antirealist theories which deny the existence of fictionalcharacters. This is not a problem in itself. But it is hard tounderstand how can Doyle intend to create literally an object if hemerely pretends to assert something. Wouldn't it be moreappropriate to claim that he pretends to create an object? I think it isa legitimate question, but to answer it would require a separate study.

(6.) On the differences between Neo-Meinongian theories see, forexample, Zalta (1992).

(7.) Anti-realist theories of fictional characters must also adoptsome variant of this distinction, because they have to explain how wecan talk about fictional characters in spite of the fact that they donot exist. On this theme, see Friend (2007).

(8.) According to Kertesz and Rakosi (2012), linguistic data can beconsidered to be more or less reliable "truth-candidates."Data are supported by their sources only to a certain extent, hence theyare not true with certainty. And similarly, theoretical claims islinguistics are usually not true with certainty, but more or lessplausible. If we follow this account, the artefactualist's claimcan be regarded as having a low plausibility value.

(9.) See, among others, Sainsbury (2012) and Azzouni (2013).

(10.) Following Kriegel (2008), we may call this account contentadverbialism. On this account, thoughts about fictionalia are related torepresentational contents, but, and this is the main point,representational contents in themselves are not relational.

(11.) One of the central issues which must be adressed byartefactualists concerns the problem of name-introduction procedures.Adherents of the artefactual theory typically think that a certainversion of the baptism-based theory of ordinary proper names is able toprovide an account also for fictional names. Of course, thebaptism-based theory cannot be straightforwardly adopted to fictionalcases because of the absence of objects which could be baptized. Onepossible solution is to follow Thomasson (1999), who argues thatcharacters acquire their names in a quasiindexical way. For more detailsof this topic, see, among others, Braun (2005) and Lamarque (2010).

REFERENCES

Azzouni, J. (2013), "Hobnobbing with the Nonexistent,"Inquiry 56: 340-358.

Braun, D. (2005), "Empty Names, Fictional Names, MythicalNames," Nous 39: 596-631.

Brock, S. (2010), "The Creationist Fiction: The Case againstCreationism about Fictional Characters," The Philosophical Review119: 337-64.

Burge, T. (2010), Origins of Objectivity. Oxford: Clarendon Press.

Conan Doyle, A. (1887/2004), A Study in Scarlet. Ware: WordsworthEditions.

Friend, S. (2007), "Fictional Characters," PhilosophyCompass 2(2): 141-156.

Goodman, J. (2004), "A Defense of Creationism inFiction," Grazer Philosophische Studien 67: 131-155.

Kertesz, A., and Rakosi, CS. (2012), Data and Evidence inLinguistics: A Plausible Argumentation Model. Cambridge: CambridgeUniversity Press.

Kriegel, U. (2008), "The Dispensability of (Merely)Intentional Objects," Philosophical Studies 141: 79-95.

Lamarque, P. (2010), Work and Object: Explorations in theMetaphysics of Art. Oxford: Oxford University Press.

Sainsbury, M. (2012), "Representing Unicorns: How to Thinkabout Intensionality," in Currie, G., Kotatko, P. and Pokorny, M.(eds.), Mimesis: Metaphysics, Cognition, Pragmatics. London: CollegePublications, 106-131.

Schiffer, S. (1996), "Language-Created Language-IndependentEntities," Philosophical Topics 24: 149-167.

Thomasson, A.L. (1996), "Fiction and Intentionality,"Philosophy and Phenomenological Research 56: 277-98.

Thomasson, A.L. (1999), Fiction and Metaphysics. Cambridge:Cambridge University Press.

Thomasson, A.L. (2003), "Speaking of FictionalCharacters," Dialectica 57: 205-223.

Thomasson, A.L. (2010), "Fiction, Existence andIndeterminacy," in Woods, J. (ed.), Fictions and Models: NewEssays. Munich: Philosophia Verlag, 109-148.

Yagisawa, T. (2001), "Against Creationism in Fiction,"Philosophical Perspectives 17: 153-72.

Voltolini, A. (2006), How Ficta Follow Fiction: A SyncretisticAccount of Fictional Entities. New York: Springer.

Zalta, E.N. (1992), "Fictional Truth, Objects andCharacters," in Kim, J. and E. Sosa (eds.), Blackwell Companion toMetaphysics. Oxford: Blackwell, 267-269.

ZOLTAN VECSEY

[emailprotected]

MTA-DE Research Group for Theoretical Linguistics

Received 5 January 2015 * Received in revised form 23 April 2015

Accepted 24 April 2015 * Available online 1 November 2015