Fiction cannot be true

Abstract

According to the dominant theory of intentionalism, fiction and non-fiction are in a “mix-and-match” relationship with truth and falsity: both fiction and nonfiction can be either true or false. Intentionalists hold that fiction is a property of a narrative that is intended to elicit not belief but imagination or make-belief in virtue of the audience’s recognizing that such is the intention of the fiction-maker. They claim that in unlikely circumstances these fictions can turn out to be accidentally true. On the contrary, I argue in this paper that fictionality and truth are incompatible. I distinguish narratives based on whether they contain invented characters or not, and offer respective sets of arguments to the effect that there is no case when a fiction is accidentally true. A narrative is either fiction or accidentally true but not both.

1 Introduction

Imagine J. R. R. Tolkien, the famous author of the celebrated fantasy novel, The Lord of the Rings, on his deathbed. Time travelling researchers from a project in the future are visiting him to unbelievably relate some unbelievable news. They tell him, “Mr. Tolkien, you will never believe this, we have just found a planet that has identical geography to Middle Earth. Superbly detailed historical documents indicate that all that you’ve written in your three volumes is true! Not to mention the extended universe!”¹ What do you think is the reaction of the dying Tolkien? What should it be? Based on the fictional scenario above, it seems that The Lord of the Rings has just turned from fantasy fiction into history-of-another-planet. Everything in the novel is true. You might consider this line of thought silly, and you would certainly be well within your rights. It is not likely that you ever face such questions seriously. But despite its silliness, I think it leads to important considerations about the nature of fiction.

The most prominent philosophical position on the scenario above can be construed as follows. Fiction has nothing to do with truth or falsity. It has to do with the intentions of the author and with her process of creating the piece of work. The intentions and the process determine fictionality once and for all, and so, even if it turns out that, on a distant world, the whole three volumes of The Lord of the Rings describe things that have happened, thoughts that people have had, characters that exist or have existed, the novel still remains fiction. As a matter of fact, it is an accidentally true fiction. People often talk about fiction as opposed to truth, however, this should not dictate the ways we conceive of the fictionality of novels, movies, works of art or pieces of writing. Fictionality does not depend on truth or the lack of it.

This paper argues against this prominent position, focusing on how radical the fictional invention is.

• 1. The Lord of the Rings is a work of fiction.

• 2. The Lord of the Rings—given the scenario above—happens to be true.

I deny (2) and I argue that (1) and (2) are incompatible. In general, I argue that a narrative’s “being fictional” and “being true” are incompatible. The claim that I make might seem ridiculous at first: on his deathbed, Tolkien should reply to the research team that if he wanted to write about that planet, everything he wrote would be true of it, however, since he did not at all want to write about that planet, not knowing anything about it, The Lord of the Rings is still not true. Short of Tolkien having the godly power of creation with words, the novel cannot be made true in any way. Even though there are many interesting issues about the connection of fiction and truth, such as what fiction can teach us about life, morality and human nature, I still think that this—rather undervalued—question is interesting and significant, especially in light of the prominent position that divorces the nature of fictionality from truth and falsity. The issue is especially intriguing as it is connected to speech act theory and the reference of fictional names. Both of these connections will be detailed later.

In Sect. 2, I present theories of fictionality, with focus on the prominent view that fictionality is not influenced by truth or falsity. In Sect. 3, I outline some reasons for thinking that fictional narratives might be considered as accidentally true. In Sects. 4 and 5, I present my arguments against claiming that fictionality is compatible with truth. I distinguish two cases based on whether a given narrative contains invented characters or not. If it does, for accidental truth, the narrative needs either to accidentally refer or state truths without reference, and I argue that neither is plausible. Such a narrative is fictional but not accidentally true. Moving onto narratives that do not contain invented characters, but invite the audience to make-believe certain things about individuals who actually exist or existed, I argue that if such a narrative turns out to be true, it is not fiction. Fictionality and truth are incompatible.

2 Fiction and invention

A simple starting point might be to contrast true stories with fictional ones: what is true cannot be fictional and what is fictional cannot be true. Fiction is untrue construction. However, the simplicity—as in numerous cases—has shortcomings that theorists and philosophers of fiction grapple with. Theories of fictionality can be classified in numerous way. One of them is the following. Is fictionality distinguished on a semantic, syntactic or pragmatic level? For example, Frege’s brief comment on epic poetry can be generalized into a semantic view. Frege says that in reading an epic poem, only the senses of the sentences are interesting, the denotations are not (1892/1960: 63). In effect then, fiction could be said to be distinguished by lacking reference. Or perhaps fictional works have a unique syntactic structure. One of the famous theorists on this view is Hamburger (1993), who develops the category of proper fiction reserved for third-personal non-factual narration. This results in the possibly unsolvable problem that first-personal narrating cannot be counted as fiction (merely “pretense” on Hamburger’s view). In any case, the pragmatic approach is by far the most popular, and it is usually cashed out in terms of speech act theory. The foundation is John R. Searle’s famous statement that the “utterance acts in fiction are indistinguishable from the utterance acts of serious discourse, and it is for that reason that there is no textual property that will identify a stretch of discourse as a work of fiction” (1975: 327). Fictionality, on this view, depends on the pragmatics of the speech situation. According to Searle’s own proposal, fiction is pretense. It is the intention of the fiction-maker to pretend making speech acts (mostly assertions) that classify a work as fictional.²

Another scheme can be developed based on Davies (2001) and Swirski (2010). Should fictionality be approached in a textualist, functionalist or intentionalist way? Textualism coincides with the semantic and syntactic levels above. A textualist disagrees with Searle in that fictionality can be distinguished by textual properties either semantic or syntactic. Cohn (1999), for example, finds, among other features, the textual property of representing another person’s thoughts to be a “signpost of fictionality” that is independent of the pragmatics of the speech situation. Both remaining approaches are pragmatic, both differently so. In this area, functionalism is attached to the seminal work of Walton (1990) in Mimesis as Make-Believe. Walton’s approach can be labeled functionalist because on his view “a narrative is fictional when its socially recognized function is to serve as a certain kind of resource in games of make-believe” (Davies 2001: 264). The function of a fictional text is to make the reader exercise her imagination. The content of fictional works is to be made-believe and not to be believed. This latter claim is shared by the most popular view of fictionality: intentionalism. However, as the label suggests, fictionality here depends not on the function of the work but on the intentions of the fiction-maker. Currie’s (1990) influential view in The Nature of Fiction is that fiction-making requires a fictive intention to produce fictive utterances.³ A fictive intention is a Gricean one: it is an intention that the audience makes-believe the content instead of believing it because of recognizing that making the audience make-believing is the intention of the author.⁴

What is clear from the above discussion is that fictionality cannot just be reduced to untrue constructions. Invention, without the intention to deceive, seems to be key. As Christopher New argues, “the central idea involved in the term ‘work of fiction’ is that of an invented narrative; and that it is this idea that we need principally to understand” (1999: 39). Similarly, Tim Crane writes: “It is because authors are the creators of fictions that we value them as artists. Creating a fiction is a matter of making something up that did not really happen” (2013: 77). Interestingly, the root of the verb “to invent” is the Latin “invenire”, which meant “to find” and “to come upon”, “to discover”. The notion of “invention” today has an opposite meaning: to invent is to create something new.

A possible view can construe the differences between invented and non-invented narratives arguing that only non-invented narratives refer to the extratextual world and only they have the pragmatic dimension associated with falsifiability (but with evidence and confirmation as well). In other words, non-invented narratives make a truth-claim. While the explicated view above is certainly attractive, it does not and would not convince many philosophers concerned with fiction. The main argument against the view can be expressed in the following way. Claiming that fiction does not refer or aim to state the truth cannot distinguish between fiction and nonfiction, invented and non-invented narratives because there are many fictional (invented) narratives that do refer and aim to state the truth.⁵ Firstly, even though someone might argue that Tolstoy’s Napoleon in War and Peace is a robustly fictional character, I think it is clear that the name “Napoleon” in the novel aims to refer to the historical person.⁶ Secondly, as Richard Gaskin claims, one should not only consider proper names when it comes to the referential dimension of fiction but general terms as well (2013: 68). Tolkien writes in The Lord of the Rings that “[o]n the third morning Caradhras rose before them, a mighty peak, tipped with snow like silver, but with sheer naked sides, dull red as if stained with blood” (1938/2004: 286). The readers can safely suppose that among other things “morning”, “peak”, “snow”, “naked”, and “blood” refer as they do in any other discourse, while “them” (the fictional characters) and “Caradhras” (a fictional place) are problematic. And lastly, some authors of fictional works of a more didactic bent express views that they offer as truth claims. Didacticism might be one of the “worst heresies” as Edgar Allen Poe says, but a heresy does not strip away fictionality.

With all these difficulties in mind, one can be inclined to drop the notion of “invention” in an account of fictionality. “Invention”, in the relevant sense here, is a relation between the narrative and the world. If a narrative is invented, it does not describe something in the world but creates something new. However, as indicated above, most philosophers tend to think about fictionality not in terms of how the fictional work relates to the world but how the creator of the fictional work relates to the work and by that to its audience.

3 Fiction and accidental truth

The general consensus among philosophers is this: nonfiction and fiction are not demarcated by truth and falsity, and a narrative’s being true does not imply that it cannot be fictional. Considering that there are numerous conditions for a narrative’s truth (it is more complicated than the truth of a simple assertion), it is strange that one would even entertain the thought that a fictional narrative can be completely true. Most philosophers of fiction see true fiction as an accidental possibility. Walton is the most radical exception in this respect (as well) as he states that he sees no reason why an author could not claim truth for every sentence she writes and still write fiction. He imagines a genre of historical novel which allows no liberties with the facts (1990: 79). If such an historical novel functions as a resource in a game of make-believe for the reader, it should be thought of as fiction. On Walton’s construal, a work which is intentionally and completely true can be a work of fiction.

I spend the rest of my paper arguing against the possibility of accidentally true fictions of intentionalists, but Walton’s view is so important that I have to address it here. As long as one accepts Walton’s definition of fictionality (being a prop in a game of make-believe), I have no quarrel with the possibility of true fictions, accidental or not. However, Walton’s definition has problems. For instance, he readily admits that his conception of fiction does not track the ordinary or common notion of “fiction”, the one that is used to classify books in bookstores. Basically, Walton’s category of fiction is very broad, it covers, for example, all representational art, and Stacie Friend argues that it should constitute its own class: “walt-fiction” (2008: 154).⁷ Seen in this way, I have no issue with walt-fictions being true. Indeed, some representations accurately represent reality: they are true representations, which can also serve as props in games of make-believe. However, it is important to point out that the possibility of walt-fictions being true does not entail that works more narrowly defined as fiction can also be true. In any case, at least it is clear that truth does not imply fictionality, or that there are true stories, histories, autobiographies that are not fictional. However surprising, this applies to falsity as well. Falsity is not sufficient for fictionality because there are false narratives that are not thereby fictional. Firstly, there are lies. If someone asks me to tell them about my last night, and to impress them, I make up or invent events that did not actually occur, then I have produced a false narrative. A standard intentionalist take on why this false narrative is not a work of fiction would refer to my intentions of attempting to deceive my listener, to deceive her into believing that what I narrate has actually occurred. In contrast, the tellers of fictional stories do not intend to deceive their audiences, what they intend is for the audiences to imagine or make-believe the content told. There is a certain type of tacit contract between creators and audiences of fiction that is lacking when someone is deceitful.

Secondly, a false narrative might be produced by an historian who makes mistakes. In her book on ancient historians, Susan Sorek mentions that Livy (Titus Livius Patavinus), an historian of ancient Rome (c.64 bce—c.7 ce) was not the most effective at checking his sources, and so he often committed the error of repeating an event as if it happened multiple times because of having different accounts. “For example some of Hannibal’s Spanish operations are repeated, and his crossing of the Apennines is reported as an attempt in 218 bce and then as a fact in 217 bce, when clearly the accounts concern the same march” (2012: 111). Of course, the changes in the history of historiographic practices is a topic on its own that I have no means or intention to discuss,⁸ however, such an historical account that creates an extra march when there was in fact only one qualifies as an example of a false narrative. It is unintentionally false.

To sum up, the difference between the two cases of false narrative above are not merely about the intentions of the storytellers. The cases are similar in that they are both false, however, they are false for different reasons. The liar’s narrative is intentionally false, relating events that did not occur. The historian’s narrative is false unintentionally. If the historian is not a liar, she believes her narrative to be true. The failure of the historian is that she attempts to tell a true story but ends up with a false one. It has to be noted that the end product might be the same. In an extraordinary coincidence, the stuff of thought experiments, the same false story might be the product of a lie and of mistaken history as well. It seems there’s agreement that there is nothing necessarily in a narrative that betrays whether it is a product of fiction making, a lie, or a mistake. Again, this is a point Searle made: “utterance acts in fiction are indistinguishable from the utterance acts of serious discourse, and it is for that reason that there is no textual property that will identify a stretch of discourse as a work of fiction” (1975: 327). Similarly, stories that are products of lies or mistakes are equally indistinguishable based on textual properties. The contrast shows that falsity is sufficient neither for fictionality, nor for lies and mistakes. And even if many would claim that stories of fiction are not false stories, there can be cases where textual features do not distinguish between fictional and false.

Consider the example of David Davies: “An author, Smith, composes what she takes to be a short fictional narrative about a family named Brown living in Montreal, Canada, whose apartment catches fire in mid-January and who are forced to move into a shelter. As a matter of fact it turns out that, entirely unbeknownst to Smith, there is a real family named Brown of whom all of the things narrated in the story are true” (2007: 35). According to Davies, fictionality does not depend on the truth, on whether or not the narrated events happened the way they did. Besides the intention for the audience to imagine or make-believe, fictionality depends on whether the actually occurring events and their temporal order have guided the author, or in other words, whether the author has observed the fidelity constraint or not. In the Brown family case, the author could not have been guided by the actual events because she did not know about them. So if she intends her audience to make-believe the fate of the Brown family, she produces a fictional narrative that is also true.

In different circumstances, Smith, who is now a neighbor of the Browns, might have produced the same narrative to a certain Jones who is a friend of the Browns, visiting from Ottawa. Jones cannot get into the Browns’ apartment and it smells of smoke. In one case, Smith tells Jones what happened and it actually did happen, so her narrative is true and not fictional. In another case, with deceptive and evil intent, Smith tells the narrative of the fire and the moving to a shelter to Jones, when in fact, the Browns are home, just burned the dinner they were cooking for Jones, and they could not hear him knocking. In this case, the narrative of Smith is false and it is a lie. In a third variation, Smith might have heard the same narrative from a neighbor who wanted to deceive her, and she retells the story without deceptive intent to Jones. In this last case, Smith did not check her facts, so she produced a false narrative by mistake. The narrative itself does not announce if it is true, false, a lie, a mistake, or a fiction.

Currie and Davies both agree that the intentions of the fiction’s creator are mostly responsible fictionality, or that fictionality depends on the fictive nature of the utterances that invite the imaginative response of the audience. On these views, the way the world is does not matter in assessing fictional status. Regardless of what states-of-affairs obtain in the world, a narrative is fictional if and only if the author intends the audience to imagine or make-believe its content and the audience recognizes this intention of the author. Currie qualifies the theory. He imagines someone named Smith (not to be confused with Davies’s Smith earlier) who represses some of his terrible experiences. Failing thus to remember them, he thinks that he invents a story, but the narrative actually retells the experiences that he repressed. In order to argue that Smith’s fictive intent fails, Currie adds the disclaimer that a narrative can be fictional only if its truth is accidental. To quote Currie:

Because the circumstances are so unusual one can scarcely imagine them arising in real life, the original proposal - that fiction is the product of a fictive intent - is an extremely good approximation to the truth. In just about any real case it will give the right answer to the question Is this fiction or not? But to turn the approximating formula into a precise one we need to add an extra condition. We need to say that a work is fiction iff (a) it is the product of a fictive intent and (b) if the work is true, then it is at most accidentally true (1990: 46).

What this means for Smith is that his narrative cannot be a work of fiction because it is non-accidentally true. It is non-accidentally true because Smith’s subconscious feeds in the information about the repressed experiences that influence Smith’s “invention”.

Not everyone agrees that an intentionalist theory can only allow for non-accidentally true fictions. In a recent article about learning from fiction through imagination, Blumson (2015) claims: “Testimonial novels such as [B. S. Johnson’s] The Unfortunates should be considered both factual and fictional, because they are in intended to induce the reader to believe their content, by means of recognition of intention, but also intended to induce the reader to imagine their content, by means of recognition of intention” (2015: 49). Based on authorial intentions, such a testimonial novel is non-accidentally true (or factual) and fictional at the same time. By this Blumson points to an important aspect of the debate: the relationship of belief and imagination as propositional attitudes, and he is right in that some parts of works might call for belief and some for imagination, which makes it difficult to clearly say whether they are fictional or factual or both.

I agree that at least for some testimonial novels, it is as important to experience certain imaginative states as it is to believe their contents and that both are intended.⁹ However, concerning the prescribed propositional attitudes, it does not seem to be the case that for a particular content of the novel, p, both “believe that p” and “make-believe that p” are intended by the author. In fact, it is plausible to hold that for the same content, and ceteris paribus, believing that p is incompatible with making-believe that p.¹⁰ All other things being equal, a content, p, is either believed or made-believe because belief entails that one thinks p is true while make-belief requires that one acts as if believing that p is true. If one believes p to be true, one acts on the belief that p is true and cannot make-believe or act as if believing p to be true (unless one is “disposed to inferentially connect the thought of p to further propositional content which one doesn’t believe”, on this see Stock 2016: 213–214). Belief that p precludes the possibility of make-belief that p. So Blumson is correct in that some testimonial novels are intended to be vividly imagined even though their contents are to be believed, however, this does not entail that a piece of content is both fictional and factual at the same time because, according to Currie, the distinction between fictional and factual content is given by incompatible, prescribed propositional states: either belief or make-belief.

As an accidentally true fiction, Davies comes up with the already mentioned example of the Brown family’s apartment catching fire. To recall, in this case, without the teller knowing, the described situation actually happens to a family of Browns living in Montreal (2007: 35). Davies says that this is unproblematic because the story is not originally about the Brown family that actually had to leave their house even though the same events happened to them as they are described in the story. According to Davies, what determines fictionality is the intentional flouting of the “fidelity constraint” under which “the author is presumed to have included only events she believes to have occurred, narrated as occurring in the order in which she believes them to have occurred” (2001: 264).¹¹ On Davies’s view, if someone tells a story with the intention for the audience to make-believe and a design that does not observe the fidelity constraint, she produces fiction even if the narrative is completely true.

Note that Currie’s and Davies’s conceptions are clearly different.¹² For Davies, Smith’s narrative, influenced by Smith’s subconscious, would actually count as a work of fiction because Smith only includes events that he believes have not occurred. On Currie’s view, this is not a fictional narrative because it is non-accidentally true. Non-accidental truth does not necessarily depend on what the storytellers believes or intends. Non-accidental truth counterfactually depends on the facts.¹³ One way to put it is that Davies’s category of fiction is broader than Currie’s: while Davies accepts Smith-type stories as fiction, Currie does not. Despite this difference, they would both accept that Browns-type stories, even though they are (accidentally) true, are fictional. I disagree.

I argue below that fiction cannot be true. In other words, “being fictional” and “being true” cannot be ascribed to narratives at the same time—they are incompatible. Both Currie and Davies have a good number of imaginative cases which would warrant a judgment of compatibility, but I think the matter needs still closer attention. There are two ways in which a fictional story can turn out to be accidentally true. First, there could be a fictional story in which the fiction-maker thinks that she invents fictional characters, and it turns out that there are or were existing individuals that satisfy all that is predicated of the supposedly invented characters. This would be the case of Davies’s Brown family above or the case where The Lord of the Rings turns out to be true. Both are seen as fictional and accidentally true at the same time because, as it turns out, there are or were real existents that accidentally satisfy whatever the fiction-maker made up. Secondly, a fictional story can turn out to be accidentally true if a fiction-maker makes fiction using only genuine proper names. A prime example of this can come from the difficult genre of historical fiction. Suppose there is a work of historical fiction that only uses names of individuals who did exist. Fictionality is in the imaginative predication: the fiction-maker ascribes characteristics or deeds to individuals that she does not know to be true. She makes fiction using real individuals of the past. If then, due to new evidence, it turns out that whatever was fictionally ascribed to an historical person is actually true, we would have an accidentally true fiction on our hands without invented characters. In the first case, I deny that the piece of narrative is accidentally true and I accept that it is fiction, but in the second case, I deny that it is fiction and accept that it is accidentally true. The main thesis remains: truth and fiction are incompatible.

4 Fiction cannot be true: accidental reference

I begin then by considering a disclaimer of Davies. As he says, the story about the Browns can be said to be true “if we grant the terms in the story can refer to entities in the real world” (2007: 35). As Davies’s remark indicates, the issue partly turns on the reference of fictional names. What do fictional names like “Odysseus”, “Hamlet”, “Rogozin” or the “Brown family” refer to? How do they refer to them? One strong position argues that these names are empty, non-referring (e.g., Adams et al. 1997; Tiedke 2011). In contrast, there are various possibilities if fictional names are taken as referring: they either refer directly to non-existent objects (e.g., Parsons 1980), to abstract objects (e.g., Thomasson 1999), to possible objects (e.g., Lewis 1978) or they refer non-directly as descriptions (e.g., Currie 1990). These views all have their intricacies, and it is beyond the scope of the present paper to disentangle them.

It needs to be granted that the terms in the narrative can refer to entities in the real world. Just recall the example from The Lord of the Rings: “snow”, “blood”, and so on. It also needs to be granted that the proper names can refer to entities. As Currie argues:

Because many of Napoleon’s characteristics were common knowledge at the time War and Peace was written, and because the description in that book of the character called “Napoleon” comes close to fitting Napoleon, it is reasonable to infer that the fictional author of War and Peace intended that “Napoleon” refer to Napoleon, and hence that he believed that it did (1990: 129).

Similarly in Davies’s narrative about the Browns, “Montreal” has to refer to the real city in Canada. However, and most importantly, the fact that terms in the narrative can refer to entities in the real world does not mean that they actually do. To return to the possible conceptions of fictional reference: if the “Browns” is an empty, non-referring name, then straightforwardly, it cannot refer to the actual Brown family. Similarly, if the “Browns” refers either to non-existent, possible or abstract objects, the reference cannot be the actual Brown family. It is only possible for the “Browns” to refer to the actual family if the name abbreviates a description. Currie addresses the issue in detail, following Ryle and Kripke. As Currie writes:

Kripke argues that fictional names cannot be abbreviated descriptions because if they were they might turn out, accidentally, to have reference. He imagines a case where, unknown to Doyle, there is someone who does all the things ascribed to Holmes in the stories. Surely that would not be a case in which Doyle was writing about that man. But it seems that on a descriptive account he would have been. For that man (call him “Actual Holmes” or “H@”) would, on such an account, be the unique individual who fits Doyle’s description. He would be Holmes (1990: 162–163).

Following this, Currie solves the problem and defends a descriptivist theory of fictional names by moving from the facts of the story (what Holmes did at this or that time) to facts about the story, namely whether it is about the actually existing Holmes or not. On Currie’s view, Conan Doyle’s Holmes is tied to the texts which Doyle produced, and so the fictional name “Holmes” abbreviates a description that not only contains what happened to him and what he did, but also that he is described by Doyle’s texts. If Currie stopped at this point, I would have no disagreements. However, he continues:

Notice that a story can be about someone without its being true of that someone, as War and Peace is about Napoleon even though it does not describe his activities correctly. Similarly, a story can be true of someone without being about that someone. In Kripke’s fantasy, Doyle’s story is true of H@ without being about him (1990: 164).

My disagreements can be phrased in the following way. According to Currie, a narrative can be true of something even if it is not about that thing. According to Davies, a narrative can accidentally refer to something and so be true about that thing. The difference is that for Currie, even if the narrative does not refer accidentally, it can be accidentally true of something it is not about. I argue against Currie that a narrative cannot be true of something if it is not about that particular thing. I argue against Davies that a narrative cannot accidentally refer to something, and therefore it cannot be true about that particular thing.

Against Davies then, the counter-argument works with the Browns example. According Davies, the “Browns” in the made-up fictional narrative can accidentally refer to an actually existing Brown family, so it is possibly true about them. In making up the story, Smith has a Gricean intention that the audience does not believe her tale about the Browns but it makes-believe because of recognizing that this is her intention. Even if the story contains events that actually happened to a family of Browns in Canada, the “Browns” in the story does not refer to them accidentally or otherwise because the reference of the “Browns” is determined by the Gricean intention of Smith. The narrative about the Browns cannot accidentally refer to the actual Browns because the reference is determined by the Gricean intention that produces the fiction. To quote Kripke’s John Locke lectures: “The fact is that in introducing the name we make ‘Sherlock Holmes’ name a particular man who would have done certain things, not just any old man who did these things” (1973/2013: 41). And this also leads us to and can be used to argue against Currie’s descriptivist view.

As Currie addresses the issue in more detail than Davies, I do the same in arguing against him, still relying on the Browns story. Take the following proposition: “The Browns spent the night in a shelter.” This proposition is true if and only if the Browns spent the night in a shelter and false otherwise. Is it true of any and all Browns who spent the night in a shelter? Even if it does not refer to any and all Browns, is it true of them? I do not think that it is. The truth of the proposition depends on the context in which the proposition is uttered. The same sentence can be used to utter true propositions about any of the Browns who spent the night in the shelter, however, the fact that it can be used does not mean that it is. Therefore, if you do not refer to a particular Brown family, you cannot say something true about them. Compare it with the following case: I am having dinner with my wife, Andrea, and with one of her friends, strangely also called Andrea. My wife’s glass is empty while there is still wine in the friend’s glass, and so I ask my wife: “Andrea, shall I pour some wine for you?” Do I thereby also ask the friend? Of course I do not, and the person my question is directed to is determined by my intention.

Currie has another imaginative example that is relevant. Jones rewrites an unknown nonfictional story (original text “T”) in his own words and style, without knowing that it is nonfictional, and attempts to pass it off as a piece of fiction.

Even if Jones uses the same names, he does not use them with the intention of referring by means of them to whoever was referred to by them in T. (…) The names in Jones’s story can be paired homophonically with the names in T, but they are not co-referential with the names in T. So the sentences in Jones’s story will not express the same propositions as the sentences of T, and the stories will be different. Their being different might then be enough for us to say that Jones’s version is fiction after all. To get around this difficulty we must assume one of two things. Either T contains no names but only descriptions, or, if T contains names, it contains only names that Jones knows to be referring and the referents of which he intends to preserve (1990: 44).

Applying this to the Browns case, in making up the fictional narrative, Smith’s using the name “Browns” might be homophonic with the actual Browns, but they do not refer to the same family. The problem with Currie’s workaround is that the original narrative, T, is nonfictional. Currie does believe that ordinary proper names refer directly, he is a descriptivist only about fictional names. So it cannot be said that there are no names in T. If it is a factually true, nonfictional account, the names in it refer directly. So therefore, the sentences in Jones’s fictional narrative do not express the same propositions as the sentences in the original, nonfictional narrative. The original narrative is true, but Jones’s is not. Altogether, Jones is not writing about the same—actually existing—people that the original narrative refers to, and so, there is no possibility that Jones’s story is true. The determination of reference precludes the possibility of truth.

Currie might argue that still, names in fiction are abbreviated descriptions, and so anything—actually existing or not—that satisfies the descriptions is referred to by the name, intended by the author or not. But now consider Tim Crane on singular thought:

I should emphasize that my aim here is not to argue for any particular theory of names, so long as there are theories of names which are consistent with the view that there can be singular thoughts about nonexistent things. If we use names to express our singular thoughts, and if it is possible to have singular thoughts about non-existent things, then there must be an account of names which does not require that they have existing referents (2013: 145).

On Crane’s view, fictional names do not refer in the strong sense of reference, that is, latching onto something real, however, even without reference one still can have singular thoughts about the nonexistent. According to Crane, there are descriptivist theories of names that allow for this. So even if one concedes that names in fiction abbreviate descriptions, that information is carried by the descriptions, it does not follow that anything satisfying the descriptions is referred to by the name. Descriptivism about names in fiction does not entail the possibility of accidental truth.¹⁴ Even if everything that a fictional narrative contains is satisfied by the real world, at most one can say that fiction maker would have uttered something true if and only if she intended to refer to actual individuals. She could have but she did not. She intended to make fiction, and so she invented characters.

However, one might object that my argument here was too quick. The point can be put the following way: even though a fiction maker might not intend the names to refer, they could still refer. So granting that the “speaker’s reference” of names such as the fictional “Browns” or “Sherlock Holmes” cannot accidentally pick out actual entities, the “semantic reference” of these terms still could.¹⁵ As Kripke expresses the distinction:

one should distinguish between what I might call the “speaker’s reference,” which answers to such queries as ‘Who was the speaker talking about when he used such and such proper name?’ The answer to the question ‘What is the referent of this name in the speaker’s language?’ I might call the “semantic reference” of the name in his language (1973/2013: 118).

The two types of reference can come apart in some cases, but for the present argument, this means that even though a fiction maker might use the “Browns” as a term to designate a fictional family, the term can accidentally designate an actual family due to its “semantic reference”.¹⁶

I think this is right, and the semantic reference of the “Browns” has to include both the actual and the fictional families. However, I still do not think that this leads to the possibility of accidental truth. According to the rules of the language, the semantic reference of “Browns” includes any and all families named “Brown”. But if someone makes fiction and uses this name, it cannot be the case that she refers to any and all families named “Brown”.¹⁷ If semantic reference could lead to accidental truth, any utterance using “Browns” would lead to a large number of uttered accidental truths and falsities, as many as there are Brown families in the world. For this reason, if there is an ambiguity in semantic reference, then one can reasonably expect the speaker’s reference to decide who or what the uttered proposition refers to, moreover, knowing what the speaker refers to is a precondition for assessing the utterance’s truth value. And I have argued that the speaker’s reference to the fictional Brown family cannot accidentally refer to the actual Brown family.

An interesting case here is fiction that uses no proper names, only general terms. One might object that my argument only goes through if it is about proper names like the “Browns” or “Sherlock Holmes”. What about the famous E. M. Forster example: “The king died, and then the queen died of grief”? If there ever were a king who died and a queen who died of grief after that, they satisfy the descriptions, and so the narrative is true about them.¹⁸ This is the case even if it is invented by the novelist, who had no intention to write about the actual king and queen. I think this objection misses the fact that there is a common ground of referring either by proper names or general nouns. It is worth quoting Searle here: “Even though the descriptor may be true of many objects, the speaker assumes that its utterance in that context will be sufficient to identify the one he means” (1969: 85). His descriptive theory can certainly be contested, but Searle’s argument still demonstrates that if one takes a speech act approach (as my targets, Currie and Davies definitely do), even general nouns are used to intentionally execute the speech act of reference. Even if there are a king and a queen who satisfy the descriptions in the narrative, the narrative does not accidentally refer to them, and so it is not accidentally true of them.

In fiction, noun phrases such as “the king” function as referring terms, and they do so necessarily. Peter Lamarque has an even stronger argument:

it would be wrong to suppose, in line with an earlier point, that because only general terms, “a man”, “a woman”, are used in the narrative then any individual satisfying those predicates automatically instantiates them in this context. The narrative is not about just any man or any woman, in spite of its failing to provide more specific information (2010: 199).

Lamarque’s argument is stronger because it follows that even a narrative such as “a king died and then a queen died of grief” are not necessarily instantiated if there are something satisfies them. If this is true for narratives using general terms with the indefinite article, then it is true for narratives with general terms using definite article “the”.¹⁹ Again, both Currie and Davies are intentionalists: they think fiction-making depends on the intentions of the fiction-maker. They also need to accept that if a fiction-maker utters the sentence “a man walks into a bar”, she does not refer to any man who walks into a bar. The reference is determined by the intentions, and the intention in this case is to make up a character of whom it is true that he walks into a bar. The intention is not to refer to any and all male individuals who walk into bars. If there is a fiction-making intention behind using general terms such as “a man” or “the queen”, then whatever is predicated of these cannot just be made true by something that is correctly described as “a man” or “the queen”. Now if someone is not a speech act intentionalist with regards to fiction making, then there might be different considerations that apply to using general terms, depending on one’s theory of language. But both Currie and Davies, whose theories are my main targets here, hold a speech act intentionalist view. In any case, my main point here is that merely satisfying a predication does not lead to truth. These narratives remain untrue fictions.

5 Fiction cannot be true: no invented characters

But Currie might disagree with me, stating that I misrepresent his views above. He could say that he agrees with me that proper reference is indeed a precondition of truth, that is, it is impossible to truly predicate without referring. Moreover, so far I have focused on cases that are paradigmatically invented, and so a valid point is that I have been making things easy and convenient for my argument. It is naturally more difficult to imagine that a fantasy novel turns out to be true than a narrative that is fictional, however, not as straightforwardly invented. An objection could follow about the possibility of writing fiction referring only to actual individuals. The prime example of this is (or could be) historical fiction.

Take for instance Alison Weir’s novel Innocent Traitor, which is set around the time and at the end of Henry VIII’s rule, and it is about the Lady Jane Grey or the “Nine Day Queen”, executed by Mary I or “Bloody Mary”. As Weir says in her closing words:

Most of the characters in this novel really existed, and most of the events actually happened. However, where the evidence is scanty or missing, I have used my imagination. For example, we do not know the identity of the female quack who was called in by Northumberland to administer arsenic to Edward VI, but I have spun a tale around her (2007: 399).

The episode Weir refers to plays an important part in the novel, and in history as well, because in order to facilitate Lady Jane’s ascension to the throne, Lord Northumberland had to buy some time for his plans. He did so by prolonging (and heightening) the terrible sufferings of the young Edward VI and delaying thus the crowning of Mary I. Weir imaginatively invented an identity for the “female quack”, “spun a tale around her”, and in the novel, the quack became Welsh, named Tegwyn Rhys. She is recommended to Northumberland by a trusted agent, Yaxley. Rhys had no children and was widowed, and she came to London. Yaxley has heard Rhys mention that it is possible to prolong life with arsenic while it can also be deadly, so Rhys is summoned to Northumberland, and she hesitantly begins to administer the arsenic to prolong Edward’s life.

It is important to note that Rhys is not an invented character as there was actually a “female quack” employed by Northumberland for the purpose of prolonging Edward’s life. However, the tale around her is spun by Weir as evidence was missing and the quack’s identity is unknown. So now for the purposes of this paper, suppose that the tale spun by Weir turns out to be true because of some newly discovered evidence, letters of the duke of Northumberland. The letters found would confirm everything that Weir had written about the quack. Her name was indeed Rhys, she was a childless widow from Wales, at that time living in London, and so on. If this were the case, it seems that this sub-narrative of the novel’s larger narrative turns out to be a piece of accidentally true fiction (if, for the sake of argument, we allow for such slicing up of narratives).²⁰

In this case, I do not think that it is plausible to deny that the story containing Rhys is accidentally true. Weir’s intention was not to invent a character fictionally but to make fiction while referring to an individual who actually existed, which is signaled by the phrase “spun a tale around her [my emphasis]” (Weir 2007: 399). However this paper pushes the thesis that fiction cannot be true. This seems established in paradigmatically fictional cases, cases where a narrative contains invented characters. What I am left now is the very small subset in which there are no invented characters and an even smaller chance that these might turn out to be true. Still, I think this is important because it presents the strongest case for someone who wants to argue that fiction can be accidentally true. Therefore, I accept that in light of confirming evidence, Weir’s tale about Rhys is accidentally true.

Taking a step back then, on the dominant view, a narrative such as Weir’s is fiction but carries the possibility of accidental truth. Such a fictional narrative either corresponds or does not correspond to what actually transpired. If it corresponds, it is a piece of accidentally true fiction. What happens if it does not? It has to be admitted that in light of new evidence, Weir’s narrative of Rhys can turn out to be false. If the yet undiscovered letters of Northumberland state that the woman’s name was Alis Dilys, she was from Wales but she was living with her husband and sons in London, and she had no qualms whatsoever about administering the arsenic, then Weir’s narrative of her would be false. Weir’s narrative then is unintentionally false because it is not like she aimed to deceive the reader.

To sum up the status of Weir’s narrative of the woman who administered arsenic, it is definitely a piece of fiction on both Currie’s and Davies’s accounts. Weir’s process can be described, in this case, as intending the audience not to believe but to make-believe what she came up with. Also, lacking evidence, she could not and did not follow the fidelity constraint because to her knowledge there were no facts to be faithful to. Still, in light of new evidence, the fiction can turn out to be accidentally true or unintentionally false. The story can be confirmed or disconfirmed. At this point, we have to be aware that the practice we have adopted to deal with this narrative are the same as in dealing with historical accounts: considering evidence, confirming or disconfirming. Now I think that it is clear we have adopted the wrong practice for fiction, and this should count as a reductio argument against claiming that narratives like Weir’s are accidentally true fictions.

Weir’s narrative is not intended to be either true or false. It can turn out to be accidentally true, but then, as the argument goes, it is not a piece of fiction. Why would accidental truth preclude fictionality here? According to both Currie and Davies, fictions intentionally prescribe imagination on part of the audience. Unlike factual narratives that call for belief, fictional ones call for imagination or make-believe. The question about Weir’s narrative can be rephrased: what happens, when a narrative that intentionally calls for imagination can be believed because it is accidentally true? Putting aside Friend’s (2008, 2012) and Matravers’s (2014) strong criticisms of the belief-imagination divide for the time being, is it not the case that the appropriate reaction for something that is true, even if it is accidentally true, is belief? In turn, if Weir’s narrative is accidentally true, should not the adequate reaction be to believe it? If belief is the adequate reaction, I see no reason to still insist that Weir’s narrative is fictional. If it turns out that everything Weir wrote about the female quack is accidentally true, then it is appropriate to believe it and so it cannot be treated as fiction. Unlike historians, Weir did not intend to tell a true story, she intended to spin a tale. She did not follow Davies’s fidelity constraint, but still, despite her intentions and process, she accidentally stumbled into fact and out of fiction in this particular scenario.²¹

A fair point is the following: the claim that Weir’s accidentally true historical novel is not fiction still rests too much on intuition.²² So then, my argument is this: on Currie’s and Davies’s views, Weir’s story about the female quack is an accidentally true fiction referring only to historical individuals, but one that is intended to be made-believe. If so, it is fiction. However, I think that regardless of intention, it is reasonable to believe the contents of the story because they are true even if accidentally. If it is reasonable to believe the contents, it is reasonable to classify the work as non-fiction, despite the intentions.

To further support the argument, consider that a work like Weir’s is intended to induce vivid imaginations: what it is like to be in the female quack’s position or imagining the young king’s suffering and so on. Many works that should not be classified as fiction are intended to induce imaginative states. On Currie’s and Davies’s views, it is difficult to maintain the difference between works that are fictional, but are to be believed in some parts, and works that are non-fictional, but are also to be imagined in some sense.²³ This difficulty supports my point that sometimes it is not the intention but the reasonable attitude (of belief or make-belief) towards the content that should matter for fictionality. And the reasonable attitude can come apart from the intended attitude, as it does in the case of the accidentally true account of Weir about the quack administering arsenic to Edward VI. The narrative of Weir, if it is an accidentally true account involving no invented characters, is not fiction because its contents are reasonably believed. If an utterance made with fictive intent is true, the intended attitude is not the same as the attitude that is reasonable to assume towards the content. Since intentions can fail, when these two attitudes are not identical, it is better to classify based on the attitude that is reasonable to assume than on the intended one.²⁴

6 Conclusion

To briefly conclude, in this paper, I have examined fiction in relation to truth and falsity. Some narratives are fictional, and “fiction” is often used synonymously with falsity. Still, being a false narrative is not sufficient for being fictional. Nevertheless, the main issue the paper dealt with is the following: can fictional narratives be true? Some prominent philosophers of an intentionalist persuasion have argued that fictionality depends on the intentions and the process of telling a story, so for a fictional story, it is necessary that the fiction maker intends her story to be made-believe and not to be believed in virtue of recognizing that such is her intention. Consequently, if it turns out that there are people, places, events that satisfy the narrative, unbeknownst to the fiction maker, it is an accidentally true fiction.

After addressing the possibility of defining fiction as invention, I have argued against the two ways in which fiction could turn out to be accidentally true. Fiction and truth are incompatible. First, it could happen that a real individual satisfies all the descriptions that the fiction maker invents for a character, which has to do with reference. Then the question becomes: is it possible to refer to an actually existing entity without the intention to do so? If the answer to this is “yes” then Davies is right, fiction can be true. However, if the answer to this is “no”, as I argued it should be, then there is a further question: is it possible to make a true utterance about an actually existing entity without referring to it? If “yes” then Currie is right and again, fiction can be true. However, I have argued that the answer should be “no”. You cannot utter truth about something without referring to it in the sense of talking about that specific thing. A fictional narrative cannot be accidentally true in this way, which also applies to narratives mainly using general terms. Secondly, an historical fiction that contains no invented characters can turn out to be true because the imaginative predications of the real individuals are accidentally correct despite the intentions and process of the fiction maker. In such a case, I accept that the narrative is accidentally true but deny that it is fiction. The reason is that despite the intentions of the author, it is reasonable to believe the contents of the narrative, and in such a strange case, the reasonable attitude to the content is a better guide to fictionality than the intended one.