The force of fictional discourse

Abstract

Consider the opening sentence of Tolkien’s The Hobbit: (1) In a hole in the ground there lived a hobbit. By writing this sentence, Tolkien is making a fictional statement. There are two influential views of the nature of such statements. On the pretense view, fictional discourse amounts to pretend assertions. Since the author is not really asserting, but merely pretending, a statement such as Tolkien’s is devoid of illocutionary force altogether. By contrast, on the alternative make-believe view, fictional discourse prescribes that the reader make-believe the content of the statement. In this paper, we argue that neither of these views is satisfactory. They both fail to distinguish the linguistic act of creating the fiction, for instance Tolkien writing the sentence above, from the linguistic act of reciting it, such as reading The Hobbit out loud for your children. As an alternative to these views, we propose that the essential feature of the author’s speech act is its productive character, that it makes some state of affairs obtain in the fiction. Tolkien’s statement, we argue, has the illocutionary force of a declaration.

1 Introduction

Consider the opening sentence of Tolkien’s The Hobbit:

(1) In a hole in the ground there lived a hobbit.

By writing this sentence, Tolkien is making a fictional statement. Fictional statements are not meant to be taken as reports about what the world is like. Notwithstanding the fact that (1) is written in the indicative mood, Tolkien is not asserting that there really was a hobbit living in a hole in the ground. If someone wanted to hold Tolkien accountable for having misdescribed the world with (1), we would say that they had not understood the concept of a fiction.

This raises the issue of the nature of the author’s speech act. Since fictional statements are not straightforward assertions, how are we to think of their illocutionary force? During the last three decades, there have been two dominant views on this matter. On what we will refer to as the pretense view, fictional statements are pretend assertions. On this view, while Tolkien is not asserting that there was a hobbit living in a hole in the ground, he is pretending to assert it. Since pretend acts are not real acts, Tolkien is not performing a real linguistic act by writing (1), and his statement is accordingly devoid of illocutionary force altogether. The locus classicus for the pretense view on fiction is John Searle’s (1975b), but the view has a precursor in Gottlob Frege’s work (Frege, 1956, p. 294). Other adherents to the view include Saul Kripke (2013), David Lewis (1978), Stefano Predelli (2019), François Recanati (2018), and Amy Thomasson (1999).

On an alternative, equally influential view, fictional statements are not pretend assertions, but instead carry the illocutionary force of a type of prescription. According to what we, in what follows, will refer to as the make-believe view, fictional statements are essentially made with the intent that the audience make believe the content of the utterance. The author tells us or, as the theory is sometimes formulated, invites us to imagine the content of the story. The make-believe view takes inspiration from Kendall Walton’s work on fiction (e.g., Walton, 1990), and one finds its canonical formulation in the work of Gregory Currie (1985, 1990). Variations of this view have been defended by, among others, Manuel García-Carpintero (2013), David Davies (2015), and Kathleen Stock (2017).

Recently, a third alternative has entered the stage. Catherine Abell (2020) defends the view that fictional utterances belong to the class of speech acts called declarations. Declarations, a speech act category which includes such linguistic acts as baptizing, marrying, and stipulating, are characterized by the fact that their characteristic function, their illocutionary point, is to bring about new, institutional states of affairs. We call this view the declaration view.¹

In this paper, we offer a novel defense of the declaration view. The pretense view and the make-believe view, we argue, both have important shortcomings. More specifically, neither theory captures what is essential to fictional discourse, in contrast to other, related speech phenomena. We claim that this essential feature consists in the creative nature of the author’s speech act: the way the author—as opposed to an actor on a stage or a father reading aloud to his children—creates the fictional narrative and the fictional world by his fictional statements. The declaration view, we argue, is the only theory that can account for this.

The structure of the paper is straightforward. In Sects. 2 and 3 we raise a problem for the pretense and make-believe views, that they cannot account for what is specific to the author’s statements, as opposed to other fiction-related linguistic activities, and suggest that this specific character inheres in their creative power. In Sect. 4, we briefly discuss other problems for these views. In Sect. 5 we propose that the creative power of the author’s statements should be understood in terms of the power to create fictional truths and advance the declaration view to account for this power. In Sect. 6, we briefly discuss Abell’s argument for the declaration view and compare it with ours. Sect. 7 concludes.

But first, a clarificatory remark. When using the terms “fictional discourse,” “fictional utterance,” and “fictional statement” in this paper, what we have in mind is the speech and writing of the author of a fiction. We do not use the terms to cover speech and writing about fictions such as, for instance, the sentence:

(2) In the story, Bilbo lives in a hole in the ground.

as uttered in circumstances other than those of Tolkien writing the novel. In the helpful terminology of Alberto Voltolini (2006), statements like (2) are “parafictional statements,” in contrast to fictional statements, like (1). Notably, parafictional statements differ from fictional statements by the fact that it makes sense to speak of them as being true or false. Neither do we have in mind statements in the fiction, e.g., those statements made by the characters in the narrative. We take the latter category to include statements made by the fictional narrator, if these are distinguished from the statements made by the flesh-and-blood author.

2 Pretense

In this section, we present our criticism of the pretense view. As stated in the introduction, the pretense view finds its canonical formulation in the works of Searle. According to this theory, fictional discourse consists in pretended illocutionary acts. Searle provisionally defines the intended sense of “pretense” as “to engage in a performance which is as if one were doing or being the thing and is without any intent to deceive” (Searle, 1975b, p. 324). Thus, Searle distinguishes the relevant sense of “pretense” from an alternative sense which involves deception, as when an undercover cop pretends to be a mobster. According to the pretense view, in writing (1), Tolkien pretended to assert that a hobbit lived in a hole in the ground. The sentence is not written with its normal assertoric force: rather, Tolkien only pretended to write it with such force.

Our main objection to the pretense view is that it fails to distinguish between what a writer does when writing a fiction and other modes of discourse which could be aptly described as pretense. For example, take the sentence jokingly uttered by president Ronald Reagan during soundcheck prior to a weekly radio address to the nation:

(3) My fellow Americans, I’m pleased to tell you today that I've signed legislation that will outlaw Russia forever. We begin bombing in five minutes. (Cf. Green, 2010, p. 83)

The linguistic activity in which Reagan is involved when making this comment is clearly different from the linguistic activity in which Tolkien is involved when writing (1). Tolkien is telling a fictional story, whereas Reagan is merely making a joke. But surely Reagan is pretending to assert something to at least the same extent that Tolkien is. In fact, it seems more intuitive to say that Reagan is pretending to assert something than that Tolkien is.

With this objection, we are following the lead of Currie, who argues against the pretense view with the example of the actor on stage, reciting the lines that an author has written for him (Currie, 1985, p. 390; cf. 1990, p. 17). Searle neglects to draw a clear distinction between the author’s (purported) pretense and the actor’s, and the pretense view is naturally taken to entail that they are in fact the same: they are both defined by being acts of pretense. But to quote Currie, “It seems intuitively clear that the author and the speaker are engaged in quite different kinds of linguistic actions” (Currie, 1985, p. 390). Henrik Ibsen, working in his study, is clearly doing something else than the actress playing Nora in A Doll’s House. The pretense view does not account for this difference.

Now, it is clear that, in some sense, according to some way of individuating activity-types, the author and the actor are in fact doing “the same thing.” There is some natural sense of “fictional discourse” in which both actor and author engage in it. And it is not unthinkable that the “same thing” that they are doing is, in fact, best analyzed as pretending. If the pretense view is construed as a theory about whatever activity-type it is that the author and the actor are both engaging in, the objection is beside the point.

But even so, the objection shows that there is, at least, an interesting distinction that the view cannot account for. Indeed, a case can be made that this is the interesting distinction. It is, at any rate, the distinction we will target here. Exactly wherein the distinction resides will require some theoretical unpacking—this is the main task of our own positive proposal in Sect. 5—but to a first approximation, we can again observe that the author is engaged in a particular kind of creative activity: she is creating the fiction, instituting the fictional world with its locales, characters, and narrative events. She makes certain things be the case in the story. The actor does none of that. While she certainly provides creative input into the performance, she does not in the same sense create the story or the fictional world. She helps convey an already-existing fiction. The creative nature of the author’s speech is what makes it especially interesting, and thus it, specifically, is our explanandum in this paper.

Moreover, it is clear that, with regard to Searle himself, the creative aspect of the author’s speech is a central concern of his (despite the fact that he at one point invokes the actor’s activity as an example of pretend discourse (Searle, 1975b, p. 328)). One of the main motivations for the pretense view, according to Searle, is that it sheds light on the ontological status of fictional entities: “by pretending to refer to a person Ms. Murdoch creates a fictional character” (Searle, 1975b, p. 330). The animating concern behind Currie’s objection is precisely this, that the appeal to pretense cannot actually bear this explanatory burden, since the actor engages in pretense without any accompanying act of fictional creation. We therefore take Currie’s objection to reveal an important theoretical deficit of the pretense view. It conflates the author’s statements with those of the actor, despite important differences between them.

It is to capture these differences, and to overcome the shortcomings of the pretense view, that Currie advances his own positive view (Currie, 1985, p. 387; 1990, pp.17–18). Currie seeks to account for the creative aspect of the author’s fictional statements, thereby distinguishing them from the actor’s statements, by appeal to the role played by make-believe in the former. We will call his view “the make-believe view.”²

3 Make-believe

The make-believe view is heavily indebted to Kendall Walton’s influential treatment of fictional representations, which received its fullest formulation in Mimesis as Make-Believe (Walton, 1990). According to this work, what characterizes fictional representations is that they mandate imaginings (or make-believe). Walton’s ambition is to provide a completely general analysis of fictional representation, including movies, paintings, and children’s games of make-believe. Walton’s target is therefore not fictional discourse per se. Even so, Walton’s followers have incorporated some of his key theoretical motifs into their theories of fictional discourse specifically.

Currie takes Walton’s theory of fictional representation and combines it with an intentionalistic view of speech acts derived from H. P. Grice (1989) and Peter Strawson (1964) to yield his theory of fictional discourse. According to Currie, what defines a fictional speech act is the structure of the speaker’s/writer’s intentions. Specifically, Currie offers the following definition schema:

(F) U performs the illocutionary act of uttering fiction in uttering P if and only if

There exists ø such that U utters P intending that anyone who were to have ø

(1) would make-believe P;

(2) would recognize U’s intention of (1);

(3) would have (2) as a reason for doing (1). (Currie, 1985, p. 387)

where ø is a schematic variable ranging over characteristics of readers which “the author would acknowledge as sufficient to ensure that anyone possessing them would, under normal circumstances, grasp his illocutionary intentions” (Currie, 1985, p. 388). ø could be something as simple as the property of reading the author’s text and knowing English.³

We will now show how Currie’s objection against the pretense view can be adapted into an objection against his own view. As we saw above, in his criticism of the pretense view, Currie insisted that a theory of fictional statements must be able to distinguish between the linguistic activity of the author and other fiction-related linguistic activities, such as those of an actor portraying a character on stage. However, we claim, Currie’s view similarly fails to account for the difference between the activity of the author and other fiction-related linguistic activities.

To see this, consider a father reading The Hobbit aloud to his children, beginning with (1). It seems plausible that this father, when reading (1), has the intentions described in Currie’s definition: that he intends his children to make believe that a hobbit lived in a hole in the ground, to recognize his intention to make them make believe this, and to have the latter recognition as a reason for their make-believe. It is, at the very least, difficult to see that the father would have to lack these intentions given that Tolkien has them. But just like with Ibsen and the actress playing Nora, Tolkien and the father are clearly doing different things. As we suggested above, Tolkien is creating the fictional work, whereas the father is simply reciting it. Currie’s view cannot account for this difference.

As in the case of our objection to the pretense view, one could ask whether the above is really an objection to Currie’s view as Currie himself understands it. Perhaps by “fictional utterance” Currie has in mind a category that includes both the author’s and the father’s statements, so that it is a feature rather than a bug of his theory that it covers them both.

But it is not so. Recall that, in criticizing the pretense view, Currie insists that a theory of fictional utterances should account for the author’s activity, specifically. Moreover, Currie explicitly denies the status of “fictional utterance” to the speech of the reading father. In the below passage from his 1990 The Nature of Fiction, he indeed seems to come close to endorsing the objection we have advanced above:

If I read a fairy story to the children, I may do so intending them to make-believe the story, but I produce no fiction thereby. The fiction has already been produced. Here we have fictive intent without fictive utterance, because a fictive utterance is one productive of fiction, and not merely an utterance in which a fiction is told. (Currie, 1990, p. 42)

Here it seems like Currie actually concedes that his own theory fails to distinguish between the speech acts of the author and the father. The one speech act is productive of the work, it is creative, and the other is not, even though they can both satisfy the conditions given by Currie’s definition.

Currie responds to this prima facie counterexample to this view by denying that it constitutes a counterexample:

[In this case], my fictive utterance does not produce fiction because it does not produce a work. No counterexamples to my theory here. The theory says that a work is fictional if it is the product of a fictive intention. (ibid.)

Here, Currie seems to be saying that the purpose of his theory is only to distinguish fictional from non-fictional works and thus that the father’s activity, which does not produce a work of any kind, does not constitute a counterexample to it. But this, at the very least, leaves unexplained what it actually is that distinguishes the activity of creating a work from what the father is doing. Moreover, this characterization of his own purpose seems inconsistent with how Currie elsewhere presents and motivates his view. As we saw above, Currie’s complaint against Searle’s view was precisely that it fails to distinguish the author’s creative speech act from acts that do not create works,⁴ and this complaint constitutes his chief motivation for offering his own alternative (cf. Currie, 1990, p. 18). Moreover, both in his (1990, p. 49), and his recent (2020, p. 21), Currie explicitly rejects the notion that he is providing a theory of fictional works as opposed to a theory of the fictionality of statements.

Our diagnosis of the shortcoming of Currie’s view is the same as for the case of the pretense view. Both views fail to account for the characteristically creative nature of the author’s speech. Pretense and invitation to make-believe are both activities that one can engage in without thereby creating a fiction. It is on this creative act that we must focus if we wish to give an account of fictional statements in the sense of the author’s activity, as opposed to that of the actor or the reading father.⁵

From this point on, in order to avoid terminological confusion, we shall use the term “fiction-making” to denote this distinct creative activity characteristic of the author’s fictional discourse. In Sect. 5, we shall offer an analysis of fiction-making. We shall argue that fiction-making consists in a particular illocutionary force that attaches to the author’s fictional statements, an illocutionary force of a kind characteristic of speech-acts belonging to the category of declarations.

Before that, however, we shall briefly discuss some additional objections to the pretense and make-believe views.

4 Necessity objections

Our main objection against the pretense view and the make-believe view concerns the sufficiency of their respective analysis of fictional discourse. The objection shows that discourses can be species of pretense or invitations to make-believe without thereby being tokens of fictional discourse. However, we may note that it’s questionable whether these accounts even manage to provide necessary conditions for fictional discourse.

First, with respect to the pretense theory, it seems perfectly conceivable that an author could be engaged in fiction-making without pretending to do anything. Indeed, it would certainly seem odd to report, about a fiction writer in the course of creating her work, that she is pretending to do something (Sainsbury, 2010, p. 9). Admittedly, Searle makes it clear that his notion of pretending is partly stipulative and not necessarily coextensive with the natural language notion. Still, in the light of the theory’s failure to provide sufficient conditions for authors’ speech acts, one might wonder what theoretical gain there is in taking this stipulated notion of pretense to be a necessary feature of such acts.

Similarly, consider Currie’s claim that the author needs to have an intention to the effect that her audience make-believe the content of the fiction. We contend that it’s possible for an author to produce fictional statements without intending for anyone to make-believe their content. We can imagine an author who works on a novel purely for her own satisfaction while intending that no-one should ever read it, much less make-believe its contents. Stock suggests that Franz Kafka might have been an example (Stock, 2017, pp. 29–30).

Currie seeks to meet this objection by suggesting that the hypothetical author, even though she does not intend for anyone to make-believe the content of her work, still intends that were somebody nevertheless to read the work, they should make-believe its content (Currie, 1990, p. 34; cf. Stock, 2017, pp. 29–30). This line of response is not convincing. It is perfectly possible to conceive of an author that has no such intentions; indeed, one who intends that even if somebody were to read her work they should not make-believe its content, as might be the case with an author who is deeply ashamed of the overt sexual nature of her work. She might express this intention by prefacing her work with the admonition: “should somebody read this work, they ought to refrain from make-believing its content.” Such an author would still have made fictional statements.

García-Carpintero (2013, 2019a, 2019b) offers a conventionalist alternative to the Gricean intentionalism of Currie and his followers. Since it does not appeal to the intentions of the author, it is not susceptible to the preceding argument. García-Carpintero claims that the speech act of fiction-making is defined by a constitutive norm, which is a norm such that a speech act qualifies as an instance of fiction-making just in case it is subject to it. According to García-Carpintero, the constitutive norm of fiction-making is:

(FM_N′) For one to fiction-make p is correct if and only if p is worth imagining for one’s audience, on the assumption that they have the relevant desires and dispositions. (ibid, p. 351)⁶

On this view, then, a speech act is an act of fiction-making just in case it is subject to a norm to the effect that it is correct if and only if it imposes certain axiological considerations on the listener. This is something it can do regardless of the author’s intentions (cf. ibid. p. 353).

Though García-Carpintero’s view avoids our necessity objection, it seems as susceptible to our sufficiency objection as the standard, Currean view. Suppose García-Carpintero is right that a norm of this kind governs fictional discourse. In order to avoid our argument from recitation, García-Carpintero still has to account for the difference between the author and the reciting father. Now, we can see no reason why the latter would not be subject to FM_N′ provided that the former is. If this is right, then whatever the difference is between them, it must be accounted for in other terms. García-Carpintero could insist that though the father’s speech-act may be subject to the FMN' norm, that norm is not constitutive of the father's speech act, in contrast to that of the author. Perhaps the father’s speech is subject to some different constitutive norm, such as one of faithfulness to the original text. However, this move seems ad hoc and, moreover, rather unexplanatory: it would not distinguish the author’s speech in terms of anything that, pre-theoretically, can be said to uniquely characterize it. This is unlike the view we will defend, which draws the distinction in terms of the author’s special creative power.⁷

We do not take the aforementioned objections to the necessity claims of the respective views to be decisive, but we do think that they show that it is questionable whether either of these views manages to capture something essential about the speech acts performed by authors. In what follows, we offer an account of such acts that does not appeal to either pretense or make-belief.

5 Fiction-making as declaration

We have argued that neither the pretense view nor the make-believe view manages to account for the nature of the author’s fictional statements specifically, as opposed to other kinds of fiction-related discourse, and we proposed that the former is characterized by its creative power, i.e., by being fiction-making. In trying to improve on these theories, the primary task should thus be to account for this creative power of the author’s speech act.

5.1 Declarations and institutional facts

The first step is to get clearer on exactly wherein this creative activity consists. We suggest that it consists, primarily, in the fact that the author makes things true in the fiction. Though an actor and, indeed, also a reading father are surely creative in some sense, it seems plausible that the author’s special creative prerogative lies precisely in this, that she makes things true in the fiction. For instance, by writing (1):

In a hole in the ground there lived a hobbit.

Tolkien makes it the case that, in the fiction, it is true that there is a hobbit living in a hole in the ground.

This claim requires some commentary. What the author makes the case, we claim, is the fact that, in the fiction, it is true that a hobbit lived in a hole in the ground. This is to be distinguished from the fictional fact, i.e., the fact obtaining in the fiction, that a hobbit lived in a hole in the ground. The former is a real fact about a fiction; the latter is a merely fictional fact about hobbits, holes, and so forth. What the author creates is the former type of fact, not the latter.

In the following, we propose that the creative power of fiction-making can be accounted for by a theory according to which fiction-making belongs to the category of illocutionary speech acts commonly known as declarations. As noted in the introduction, a version of this view has been defended by Abell (2020). Abell’s version of the view involves stronger theoretical commitment than ours, and she defends it on the basis of considerations different from those we have advanced above. Below, we explain how the view can account for the creative power of fiction-making. In Sect. 6, we compare our argument to Abell’s.

The category of declarations includes such linguistic acts as marrying, baptizing, appointing, and stipulating.⁸ The defining feature of the class of declarative illocutionary acts is that they enact a change in social reality and bring about a new state of affairs by the very act of performing them.⁹ As an example, consider the fact that a certain couple is married. This state of affairs is made the case (at least partly) by someone performing the linguistic act of marrying them. A successful performance of the act of marrying by using the sentence.

(3) I hereby pronounce you man and wife.

brings it about that the couple in question is now married. It changes reality by instituting a new fact. Similarly, someone who under the right circumstances utters the sentence.

(4) I hereby declare the meeting open.

will have managed to change social reality in such a way that the meeting is now open. Finally, consider stipulations. Someone who, under the right circumstances, utters the sentence.

(5) I hereby stipulate that this piece of wood will count as the white queen in our chess game.

brings it about that the piece of wood in question counts as a chess queen. Bringing about these facts is, to use speech act theoretic terminology, the illocutionary point of declarations, the characteristic aim of the speech act category (compare assertions, whose illocutionary point is to represent the way the world is, or commands, whose illocutionary point is to prompt action on the part of the addressee).

We propose that fiction-making is a species of declaration, and that this is what constitutes its creative character. When Tolkien writes (1), he is performing a declaration whose illocutionary point is to bring about the fact that, in The Hobbit, it is true that a hobbit lived in a hole in the ground. Accordingly, we call our view “the declaration view.”

The primary motivation for the declaration view is that fiction-making, like other declarations (such as the examples (3–5)), is in the business of bringing about facts. Consider the analogy with marriages. After Peter and Paul have been married, but not before, the sentence “Peter and Paul are married” can be uttered truthfully. Likewise, after Tolkien has written (1), but not before, the meta-fictional statement.

(6) In the story, there is a hobbit living in a hole in the ground.

can be truthfully made. In the former case, this is because the fact making the sentence true has been instituted by a prior linguistic act. Our claim, first of all, is simply that this is true of the latter case as well.¹⁰

It may justifiably be asked, however, how deep this analogy is. There is an obvious sense in which all speech-acts “bring about facts.” By making an assertion, for instance, I bring about the fact that I have just made an assertion. I also, hopefully, bring about the fact that you hear me making it, and so on.

Declarations, however, bring about facts of a special kind, and they stand in a characteristic relation to the facts they bring about. The facts brought about by the speech acts (3–5) are all what we may think of as social or institutional facts. That two people are married, that a certain meeting is opened, and that a certain piece of wood counts as the white queen in a certain chess game, all seem to be facts of this kind. In what follows, we will refer to facts of this kind as “institutional facts.” Declarations, moreover, do not merely bring about such institutional facts as a causal consequence of their uttering. As stated above, bringing about such facts is their illocutionary point. Their power to bring them about inheres, one might say, in the conventions governing their use.

We hypothesize that, when Tolkien writes (1), thereby bringing about the fact that (1) is true in The Hobbit, he is bringing about an institutional fact of the aforementioned kind. By extension, we hypothesize that all facts about what is true in some fiction are institutional facts.¹¹

To further explain and motivate this hypothesis, we must say something about what institutional facts are. In this, we follow writers like Searle (1995, 2010) and Amie Thomasson (2003), according to whom institutional facts obtain in virtue of the collective acceptance of “constitutive rules.” The existence of such institutional facts characteristically implies that certain deontic statuses, such as rights and obligation, apply to individuals, whereby they govern human affairs (Searle, 2010, p. 91). Furthermore, we follow Searle in requiring that these deontic statuses and the rules that govern them be in an important sense “extra-linguistic” and that the power of declarations to institute them obtain in virtue of extra-linguistic rules (Searle, 1995, pp. 359–60; cf. Abell, 2020, chap. 3) (Note that without this last constraint, it would be possible to argue that every speech act is a “declaration” according to the present characterization. For instance, an assertion can be taken to have as its illocutionary point to make the speaker beholden to the norm of assertion (whether that be the knowledge norm or what have you), which is a deontic status.)¹²

Both Searle (2010, p. 121) and Thomasson (2003, p. 279) are impressed by the parallels between fiction-making and the creation of institutional facts. In both cases, we seem to create a shared reality out of whole cloth, somehow parallel with or in addition to the reality of brute physical fact. Neither author, however, goes so far as to assimilate fictional facts to institutional facts as a subcategory. We believe this is a missed opportunity for theoretical unification. Consider that Thomasson draws explicit parallels between her view on the nature of institutional facts and Walton’s theory of games of make-believe. According to Walton, fictional truths, or facts, “consist in there being a prescription or mandate in some context to imagine something” (Walton, 1990, p. 39)—i.e., in the existence of a certain deontic status. Such prescriptions are established via “principles of generation” which themselves hold in virtue of something like collective intentionality: social conventions or tacit agreements (Walton, 1990, p. 38). It is only a very short step from this view to the view that fictional facts are institutional facts in Searle’s and Thomasson’s sense: they obtain in virtue of the collective acceptance of constitutive rules (principles of generation) and their existence impose deontic statuses.

This is our view. In addition to its role in accounting for the creative power of the author, we believe that there are compelling independent considerations in its favor. For one thing, it is hard to see what else fictional facts could be. The most plausible rival would seem to be to identify fictional facts with mental facts of some kind, for instance, facts about what a given person (such as the author) is imagining. But fictional facts seem to be public in a way that facts about mental states are not. Moreover, facts about mental states are by themselves powerless to impose deontic statuses on people, and fictional facts do indeed seem to impose such statuses. For instance, they constrain what additional truths writers are allowed to subsequently add to their fiction (too large deviations from the established fictional truths will generally be met with resistance by readers). They may also, as Walton claims, impose obligations to imagine certain things, although we remain neutral on this specific proposal.

One may worry that the deontic statuses that we take to constitute fictional truths are too “thin” to warrant an analogy with the deontic statuses that constitute more paradigmatic social institutions like marriages and contracts. They lack ramifications outside of the context of fiction itself and do not significantly impact human relations outside the intimate circle of author-reader-text. But note that our argument does not require fictional truths to be in every respect analogous to these more paradigmatic social facts. It only requires them to share those features of other institutional facts that make them apt to be created by a specialized class of speech acts (declarations), which is consistent with them being “thin” in this sense. Note that the facts generated by stipulations, with which we see the strongest parallel (cf. note 10), seem similarly thin, as do facts about the rules and player roles of games, which are often treated as prime examples of institutional facts (cf. Searle, 1995, pp. 36–37).

Fiction-making, then, is the class of speech acts whose illocutionary point is to bring about fictional facts. Fictional facts are institutional facts, and fiction-making is therefore a species of declaration. This hypothesis, we claim, accounts for the creative power of fiction-making. It also allows us to explain and fruitfully investigate several other features of fiction-making, as we shall now see.

5.2 Declarations and fictional speech

As we suggested in the introduction, many acts of fiction-making have the striking feature that they look syntactically like assertions although they plainly lack many of the characteristic features of assertions. This sits well with the hypothesis that they are declarations: it is a commonly noticed feature of declarative utterances that they often are syntactically indistinguishable from a corresponding assertion. Consider.

(7) You are fired.

(8) I quit.

Both of these sentences can be used to make ordinary reports but also to issue declarations, as when (6) is uttered by a boss to an employee to fire her. To understand when these sentences are used as declarations, one needs knowledge that goes beyond the mere linguistic competence required to understand the propositional content of the sentences. To understand that (7) or (8) is uttered as a declaration (and to understand what this entails), one must have knowledge of the contextual factors that determine the illocutionary point of the sentence. Similarly, to know that Tolkien is fiction-making when uttering (1), rather than asserting, one must have contextual knowledge of the factors that determine it as an instance of fiction-making. In the next subsection, we shall say something about what those factors are; for the moment, it suffices to note this analogy between fiction-making and other declarations.

Another salient feature of fiction-making is that it does not commit the author to the truth of the sentence uttered. Someone who knows that Tolkien is engaged in pure fiction-making (we consider mixed cases below) in writing (1) would not hold him accountable for having uttered a falsehood. This is in line with our proposal, according to which the illocutionary point of Tolkien’s speech act is that of a declaration: not to represent the world, but to make certain facts obtain. Closely related to this point is the inability of fiction-making to be sincere or insincere. There is no way for Tolkien to insincerely write or utter (1), when the sentence figures as a part of The Hobbit. In the famous words of Sir Philip Sidney, “[…] the poet, he nothing affirmeth, and therefore never lieth” (Sidney, 2014/1595, p. 66).¹³ This, again, aligns with our proposal, according to which acts of fiction-making are declarations. All other types of illocutionary acts have sincerity conditions: they are insincere if the speaker is not in the mental state corresponding to the act.¹⁴ An assertion is insincere if the speaker does not believe the asserted content, a command is insincere if the speaker does not want the addressee to perform the prescribed act, and a question is insincere if the speaker does not wonder about the answer to the question. Declarations, however, lack sincerity conditions. There is no sincere or insincere way of marrying two persons, and there is no sincere or insincere way of pronouncing a meeting open. In general, for declarations, there are no mental states that correspond to the linguistic act. Again, fictional discourse is similar in this respect¹⁵ (cf. Abell, 2020, p. 56).

The previous paragraph calls for a proviso. It is well-known that parts of a fiction might be true of the real world, and moreover that the author might actually intend the reader to take parts of the narrative to be true of the real world. An example of this kind is brought forward by Searle (1975b), in the form of the first sentence of Tolstoy’s Anna Karenina:

(9) All happy families are alike; each unhappy family is unhappy in its own way.

It is natural to take Tolstoy as intending us to take this as an observation about what the real world is like, and not only as something that is true in the fiction. We see no particular problem for the declaration view here. Intuitively, Tolstoy does two things when writing (9) as part of the narrative of Anna Karenina. He (a) makes an assertion concerning what the real world is supposedly like and (b) makes it the case that, in the fiction, (9) is true. Qua assertion Tolstoy’s utterance can be evaluated for truth and falsity, as well as for sincerity. Qua act of fiction-making, Tolstoy’s utterance is neither true nor false, and cannot be insincere.¹⁶ We notice that there are many examples of utterances that play the dual role of declarations and assertions. Consider the following utterance, made by an umpire in the course of a baseball game:

(10) He is out.

This is both an assertion to the effect that the player is out and a declaration, making certain changes to the course of the game (see (Lewiński, 2021) for discussion of statements performing multiple illocutionary acts).¹⁷

With regard to cases like (9), in which the author is simultaneously making an assertion about the real world and telling a fictional story, the declaration view again seems to have a distinct advantage over the pretense view. The proponent of the pretense view cannot say that (9) does double-duty as a pretend assertion and a serious assertion: insofar as it is serious, it is ipso facto not pretended. She must therefore say that (9) only constitutes a straightforward assertion. The role of (9) as part of the fictional narrative is thus left unexplained on this view.¹⁸

The same, arguably, is true of the make-believe view. There are reasons to believe that the proponent of the make-believe view must identify fictional utterances with those that prescribe mere make-believe, i.e., make-believe without belief (cf. note 5). But if this is the case, and if (9) is indeed intended to produce belief in its content, the make-believe theorist cannot identify it as a fictional utterance, and its role as part of the fictional narrative is, again, left unexplained.

Like other declarations, then, fiction-making utterances lack sincerity conditions and do not commit the speaker to the truth of what is said. But declarations are not wholly normatively unconstrained; they have what are often called preparatory conditions: conditions that must obtain in order for the declaration to be successful. These conditions distribute authority over the creation of institutional facts over members of the speech community; they govern who has the authority to create institutional facts, and under what conditions. This all seems to be true of fiction-making as well. Clearly, not all purported fiction-makings need succeed in creating fictional truths. For instance, I cannot, simply by intending it, add truths to The Hobbit. I lack the authority to do so. Perhaps in so attempting, I instead manage to add truths to a different fiction—The Hobbit*—or perhaps I simply fail to create any fictional truths at all. Either way, I have failed in my fiction-making attempt.

The rules that govern who has the authority to add what truths to what fictions under what circumstances are subtle. We will discuss some aspects of this problem below, but it should be noted that the distribution of authority over fiction-making might very well—indeed, can be expected to—be a culturally and historically variable matter. This is precisely what we should expect, given the declaration view. No less than those of baptizing and marrying, the preparatory conditions for fiction-making should be seen as a matter of convention, subject to continuous reinterpretation and renegotiation. This also means that identifying these preparatory conditions and answering questions like the one above is largely an empirical matter.

With this in mind, it is nevertheless possible to say some informative things about how the authority to fiction-make is distributed. As a first pass at a principle describing this distribution, we might say that an author has authority to add truths to a fiction if, and only if, the fiction is one that the author herself has created. But this first-pass principle admits of exceptions. An obvious exception consists in those cases where the author, through formal or informal means, shares her fiction-making authority with or transfers it to others. In contemporary literature, it is common to see "fictional universes" to which several authors contribute, like the comic-book universes of publishers like DC and Marvel. Less obviously, perhaps, there may be cases in which an author lacks the authority to add truths to her own fiction, even if she has not formally signed away that authority in a licensing agreement or similar. One possible example of this, suggested to us by a seminar participant, is the case of Harry Potter author J. K. Rowling trying to retroactively add truths to the Harry Potter stories—truths that are nowhere evident in the text of the Harry Potter books—such as those concerning the sexual orientation of the character Albus Dumbledore. According to our source, this attempt to retroactively add truths to a finished fiction was not universally accepted, despite Rowling's undisputed ownership over the Harry Potter stories. One possible explanation for this fact is that our norms of fiction-making require the truth-adding to take place as part of telling the story.

To summarize the previous discussion, on our proposal, an act of fiction-making is a declaration with the following properties:

• (i) The illocutionary point of the act is to make it the case that something is true in the fiction.

• (ii) It does not commit the speaker to the truth of the sentence uttered.

• (iii) It lacks sincerity conditions.

• (iv) To successfully add truths to a fiction, one must have the requisite authority to do so. How this authority is acquired is, as discussed above, a culturally contingent matter, much as with other declarations.

Comparing the speech of the author with that of the reciting father, we can see that the two differ with respect to (i) above—the father's speech does not have as its illocutionary point to create fictional truths. The father’s speech and the author’s are alike with respect to (ii): neither speaker is committed to the truth of what they utter. But the two arguably differ again with respect to (iii): the father can represent himself as felicitously following the words of the story while intentionally tweaking them, which arguably constitutes a kind of insincerity. As regarding (iv), it does not seem like the reciting speech-act performed by the father is subject to any specific preparatory conditions. No authority is required in order to recite a text (one can lack permission to recite but that does not mean that the speech act fails if one does it anyway). The authority required to successfully perform the fiction-making speech act clearly makes it different in this regard.

5.3 Fiction-making purport

The above comparison between the speech acts of the author and the reciting father raises an important further question: in virtue of what does a particular piece of speech fall under one or another set of norms and success-conditions in the first place? In other words, what makes it the case that the author’s speech, but not the father’s, is even so much as an act of purported fiction-making—regardless of whether or not it actually meets the preparatory conditions and succeeds in its fiction-making purport?

Before turning to this question, let us in passing note that it is, strictly speaking, orthogonal to the main enterprise of this paper, which is to defend the declaration view. The declaration view is a view about what kind of speech act fictional speech is—its illocutionary point and the norms governing it—whereas the question currently under consideration concerns the antecedent issue of what it is that determines a given speech episode as a speech act of this or that kind. Even if what we say below is inaccurate, then, the declaration view is not impugned. Still, we think that the following considerations can shed additional light on the nature of fictional speech.

It seems clear that the speaker needs to have the right intentions for their act to qualify as purported fiction-making. This is illustrated by the following scenario: consider again the father reading The Hobbit to his child, and suppose the father misreads a passage in such a way that the story is changed. Does the father now suddenly turn from reciter to (purported) author? We find it intuitively clear that the answer is “no,” and what explains this datum is, plausibly, precisely that the father lacks the intention to fiction-make.

The father intends only to faithfully reproduce an existing fictional text. The same goes for, e.g., a translator, whose job is to faithfully render the story in a different language. The speech acts of these figures, then, do not qualify even as purported fiction-making acts. There are, however, borderline cases where a writer will base their story on a previous, pre-existing story, to which they remain largely faithful while changing details, embellishing or cutting down. This is often the case with adaptations to new media but can happen within the field of literature as well. Consider Pat Barker’s retelling of the Iliad through the eyes of Briseis in The silence of the Girls. Is Barker’s speech a case of purported fiction-making or merely a retelling of an existing fiction? Barker’s speech seems like Tokien’s in some respects—she is an author of an original work—and like that of the father in others, since she is bound by the events in Homer’s work.

We find it clear that Barker is here creating new fictional truths, since she adds to and elaborates on the events in the Homeric Iliad, and is thus engaged in fictional speech, despite the fact that she is also constrained by the existing text of the Iliad. This goes to show that the type of fidelity-constraint that the father and the translator are subject to is not sufficient for them not to be fiction-making. That the latter do not (even purport to) fiction-make is due to the fact that they lack any fiction-making intention in addition to their intention of fidelity. Barker, by contrast, has additional creative intentions, which gives her writing fiction-making purport.

To further underline this point, note the different natures of the fidelity-constraints placed upon father/translator, on one hand, and Barker, on the other. The father and the translator are in more-or-less explicit agreement with their audience or editor to simply reproduce the text, and the fidelity-constraint derives from this agreement and their resolution to follow it.¹⁹ In Barker’s case, on the other hand, the constraint is plausibly aesthetic in nature. Her work aspires to explore a female perspective on war in the foundational work of western literature. It would fail in this aesthetic aspiration if it did not relate properly to the Iliad, and it is hence from this aesthetic aspiration that the fidelity constraint derives. At the same time, the work would also fail aesthetically if it merely reproduced faithfully the text of the Iliad with no creative embellishments at all. The conceit of following Homeros is not, for Barker, an end in itself (as it is for the father and the translator), but is in service of an independent creative ambition.

Note that, as already indicated (p. 6), we do not wish to deny that the reciting father (or the actress playing Nora) lacks any kind of creative intention. Clearly, both speakers can be expected to intend to add their own creative input to the performance, manifested in delivery, tone of voice, body language, and so forth. That these sorts of speech fall short of fiction-making is not due to lack of creative intentions broadly construed, but—we maintain—due to lack of the intention, specifically, to fiction-make.²⁰

Note also that we do not wish to deny that there might be borderline cases between telling and retelling of a fictional story which do not clearly match the illocutionary structure of either the author’s or the reciter’s speech. Further analysis may be needed to account for all the possible variation. Still, we take it that, having offered analyses of the most clear-cut cases and accounted for their differences, our view already represents a significant advance over earlier attempts. Our account also provides a framework within which further analysis of different kinds of fiction-related speech acts may be conducted.

5.4 Indirect fiction-making

Above, we have discussed the norms and preparatory conditions that govern fiction-making. Another, equally important aspect of the illocutionary structure of fiction-making is how it adds truths to fictions—i.e., what the relation is between the actual speech-act performed and the truth that is added to the fiction.

Again, consider a first pass at a principle: when performing an act of fiction-making by writing or uttering a sentence S, what the author makes true in the fiction is the conventional propositional content of S. This principle captures simple cases of fiction-making, like the act of fiction-making performed by Tolkien when writing (1). However, this principle, too, turns out to be overly simple. There are a number of cases in which the relation between the sentence uttered and the content made true in the fiction is less direct.

For example, consider unreliable narrators, a common literary device where the narrative is put in the mouth of a (fictional) narrator whose trustworthiness the reader is led to question. In Robert W. Chambers’ short story The Repairer of Reputations, the narrator, Hildred Castaigne, relates a gruesome tale of murder and conspiracy centrally involving a mysterious play called The King in Yellow, but we also learn that Hildred has been committed to an insane asylum after a horsing accident and are thus led to question the veracity of his testimony. In this and many similar examples, the content of the sentences making up the story, as attributed to the narrator, cannot be straightforwardly identified with its fictional content. If we can draw any straightforward conclusions about the latter, it is only that the fictional narrator asserts the words of the story—and we are left wondering whether these assertions, in the world of the fiction, are truthful.

This example illustrates that the first-pass principle is insufficient. The relationship between literal sentence content and fact brought about is sometimes more indirect. In this, fiction-making does not differ from other declarations (cf Abell, 2020, p. 58). Though standard examples of declarations, such as the “you’re fired” of the manager, have the property of making true the very content of the uttered sentence, there are examples of non-fictional declarations where the relationship is more indirect. The most obvious example is explicit performatives. “I hereby declare the meeting open” makes it the case that the meeting is open, but this is not the content of the sentence uttered. Consider also the oath sworn by court witnesses in many jurisdictions. In England, the oath reads:

(12) I swear by Almighty God that the evidence I shall give shall be the truth, the whole truth, and nothing but the truth.

This oath, much like a declaration, has the function of instituting a certain social fact: making the speaker susceptible to the charge of perjury, should she later lie while giving her testimony. Yet, the social institution of perjury-charges is not explicitly mentioned in the oath.

On the other hand, it is clear that the relationship between literal sentence content and what is made true in the fiction is not wholly unconstrained. Tolkien could not, presumably, use (1) to make it true in The Hobbit that a human lady lived in a high-rise building. To find a more precise characterization of how literal sentence content relates to content made true in the fiction will be another important task for the continued development of the declaration view.

6 Comparison with Abell

As previously mentioned, Catherine Abell has defended a version of the declaration view in her recent (Abell, 2020). Having defended and elaborated our own version of the view above, we will briefly compare it with hers.

Abell follows Currie and other adherents of the make-believe view in holding that fictional discourse serves to “communicate imaginings”. However, unlike Currie, Stock, and other adherents of the make-believe view, she does not think that authors can achieve this by means of Gricean mechanisms. The shortcoming of the “Gricean” view, on which fictional content is settled by the communicative intentions of the author, shows, according to Abell, that imaginations can only be communicated by dint of institutional rules. A class of utterances governed by such institutional rules are, by definition, declarations.

Abell’s argument against the Gricean, goes as follows: since fictional discourse is not subject to the Gricean maxim of Quality (“Try to make your contribution one that is true”), communicative principles alone are unable to explain how we come to grasp what the content of fictions are. According to Abell, the Gricean approach violates the following principle:

If fictive utterance u has content c iff p, any audience who is able to identify the content c of u must have access to the resources required justifiably to infer that p.²¹ (Abell, 2020, p. 60)

To illustrate with Abell’s own example: in the course of a regular conversation, if someone utters “It’s raining cats and dogs,” we readily interpret her as wanting us to believe that it’s raining heavily, since we know for sure that she cannot reflexively intend us to believe that it’s literally raining cats and dogs (we assume that the speaker is aware that we know such events are impossible). We reach for an alternative interpretation in order to make the speaker’s utterance conform with the maxim of quality, and the speaker expects us to do just that. By contrast, if the same statement is made by an author in the course of writing a fictional narrative, the same line of reasoning is not available to us, since we cannot employ our knowledge that it cannot literally be raining cats and dogs in order to infer what the author intends us to imagine. She might very well be intending us to imagine the literal content of the sentence. Since we cannot know for sure what the author intends us to imagine, identifying the fictional content with what the author intends us to imagine would violate the above principle (Abell, 2020, pp. 31–32).

According to Abell, this reasoning shows that, in the case of fictional discourse, something over and above the Gricean conversational principles are needed in order to explain how we come to grasp the content of fictive utterances. More specifically, Abell argues that there are institutional rules governing fictional discourse, dictating that when a fictional utterance is made, we are to imagine so-and-so. What is declared, on Abell’s view—i.e., the institutional fact that is made to obtain through the performance of the act—is that some specific content obtains in the fiction (is “fictional”). This content is what the reader, qua subject to the institutional rules of fiction, is supposed to imagine.

As indicated, we are in fundamental agreement with Abell in understanding fictional utterances to perform declarative speech acts, instituting facts concerning what is “fictional” or, with our terminology, what is true in the fiction. Having said this, we have reservations about the way Abell argues for the view.

First, we do not think Abell is successful in showing that there is no other Gricean mechanism, besides assuming that the author abides by the maxim of quality, by virtue of which readers could infer that the author intends her utterances to be taken non-literally. We want to avoid diving too deep into the dialectics concerning this (see Abell, 2020, pp. 66–75; Stock, 2017, chapter 2 for discussion); we only note that it seems implausible that the reader has no Gricean resources other than the maxim of quality with which to work out what the author intends to convey and that Abell's arguments to the contrary strike us as inconclusive.

Secondly, it seems to us that even if Abell is right that Gricean methods of derivation of speaker meaning are less effective with respect to fictional utterances, advocates of the Gricean views could simply acknowledge that in some cases, working out the content of fictional utterances is more difficult than working out implicatures of the corresponding assertions. By doing so, the Gricean need not say that we are generally unable to identify the contents of fictions, but only acknowledge that we are less reliable in doing so than we are in picking up on the communicative intentions of asserters. That this would be the case does not seem at all unreasonable to us: it often is comparatively more difficult to work out when something is meant non-literally in the course of interpreting a fictional work than in cases of assertion.

Moreover, we also think that Abell’s case against the Gricean make-believe view underdetermines the dialectic situation, even if it is successful. There are other alternatives, like the non-Gricean version of the make-believe view advocated by Garcia-Carpintero (2013) (see Sect. 4), which Abell does not acknowledge, as well as the pretense view, which she mentions but does not argue against (Abell, 2020, pp. 54–55). By contrast, our main argument applies to all these views: they all fail to distinguish the creative speech act of the author from nearby activities. As we have argued, it is precisely with regard to this feature that the declaration view is superior to its rivals.

In summary, then, we agree with the main thrust of Abell’s theory, that fictional utterances are declarations, and our paper might very well be read as a defense of (that aspect of) her view. However, we think that our arguments for the declaration view are both less contentious and more general in scope than those provided by Abell.

7 Conclusion

According to the view defended here, fictional statements are declarative speech acts, akin to stipulations, marryings, or baptizings, which serve to make it the case that certain things are true in the fiction. We have argued that this view is preferable to both of its main rivals: the pretense view and the make-believe view. Both these views fail to account for the creative aspect of the author’s fictional statements, their being acts of fiction-making. As a consequence, these views fail to distinguish between fictional discourse and other linguistic activities pertaining to fiction, such as the speech of a father reading aloud to his children.

It is a consequence of the declaration view that fictions are constituted by institutional facts, similar in ontological status to facts about marriages. While a detailed investigation of the consequences of this view for the metaphysics of fictional entities will have to be saved for another occasion, it might be noticed that it at least prima facie sits well with an intuitively appealing view, the view that fictional characters are artefacts created by their authors, which has been defended by, among others, Kripke (2013); Thomasson (1999); Abell (2020). The suggestion that facts about what is true in the fiction are social facts instituted by the author’s declarative speech acts promises to shed light on how such artefact-creation is accomplished.

It bears stressing again that as far as our view is concerned, both pretense and prescriptions to make-believe may very well be common features of fictional discourse. We find it plausible that authors typically do have the reflexive intentions characteristic of, for example, Currie’s version of the make-believe view, and that they often are engaged in something that could loosely be described as “pretense.” However, we claim that neither of these features manages to capture the sense in which the linguistic activity of the author is an act of fiction-making. To characterize this act, appeals to pretense or make-belief are not sufficient.