1 Exceptionalism about ordinary names in fiction

Authors generally introduce new fictional names when writing fictions. Conan Doyle introduced the name “Sherlock Holmes” while writing A Study in Scarlet. Leo Tolstoy introduced the name “Pierre Bezukhov” while writing War and Peace.Footnote 1 These fictional names have raised complicated semantic and ontological issues that many philosophers of language and philosophers of fiction have focused on and argued about, such as whether there are any fictional objects to which fictional names refer, and if there are any, what the nature of those objects is.

However, authors also sometimes use pre-existing non-fictional names while writing fictions. For example, Conan Doyle inscribed the city name “London” while writing A Study in Scarlet, and Leo Tolstoy inscribed the historical figure’s name “Napoleon” while writing War and Peace. This paper deals with the semantics and meta-semantics for these non-fictional ordinary names used in fictional contexts, i.e., how the referents of those names are determined and what the semantic contents of those names are.

It first seems natural to think that ordinary names used in fictional contexts simply refer to ordinary real objects: Conan Doyle’s inscriptions of “London” refer to the real city London. Leo Tolstoy’s inscriptions of “Napoleon” refer to the historical figure Napoleon. Indeed, until recently, this natural view has not been seriously challenged. Saul Kripke, for example, seems to take it for granted:

A referent, of course, need not be supplied if the work of fiction or the story is about ordinary entities of the primary kind like, say, people. If this is so, then when one uses the name “Napoleon” in a story about Napoleon or “George Washington” in the story of his chopping down the cherry tree, one need not say that the name here refers to a fictional character; one can say that it refers to Napoleon, or to George Washington, himself (Kripke, 2013: 82).

However, not a few philosophers, so-called exceptionalists about ordinary names used in fictional contexts, have begun to question the view that those names refer to ordinary real objects.Footnote 2 Their view can be described as follows:

Exceptionalism

When ordinary names are used in fictional contexts, they always change their semantic contents and work the same way as fictional names in general.

For example, according to exceptionalism, the name “Napoleon” appearing in War and Peace works the same way as other fictional names such as “Pierre Bezukhov.” That is, “Napoleon” in War and Peace refers either to the so-called fictional surrogate of the real Napoleon if “Bezukhov” refers to a fictional character, or to nothing if “Bezukohv” refers to nothing, but in any case not to the real Napoleon.Footnote 3

Although the main arguments for exceptionalism will be presented in detail in Sect. 4, it is worth noting in advance that one primary methodological virtue of exceptionalism is to provide a uniform account of every name used in fictional contexts: other things being equal, a more uniform theory is of course superior to other competing theories, and this is what makes exceptionalism attractive. In this paper, however, I will argue that there is no further compelling reason to admit exceptionalism. I will rather maintain that whether ordinary names used in fictional contexts refer to real or fictional objects depends on which semantic intention speakers have. In the next section, I will introduce my intention-based non-exceptionalism about ordinary names used in fiction.

But before moving on to the next section, I will provide two remarks that help clarify my discussion in this paper. First, for the sake of argument, I will assume referentialism (also called Millianism) about names, the semantic view that the semantic content of a name, if it has any, is simply its referent, and is not given by any associated definite description. Accordingly, I will also assume that sentences containing names semantically express singular propositions (or gappy propositions, or no propositions, if any of the names that appear in the sentences fail to refer).Footnote 4 Those who reject referentialism may take this paper as arguing about which theory of ordinary names in fiction would be best for referentialists.

Second, in what follows, for the convenience of discussion, I will use terms such as “fictional character” and “fictional surrogate” in a way that seems to presuppose fictional realism. Fictional realists can regard my uses of these terms as literal, so they can assume that I am literally talking about a particular kind of object.Footnote 5 By contrast, anti-realists can assume that I am engaging in a certain kind of extended pretense when I use those terms.Footnote 6 But whatever ontological view is taken, it does not affect my key arguments in this paper.

2 Referential intentions and non-exceptionalism

In this section, I will introduce my intention-based non-exceptionalism about ordinary names used in fiction. To do so, I first need to briefly explain the so-called causal theory (or picture) of names, first suggested by Donnellan (1970) and Kripke (1980) as an alternative to the description theory about reference-fixing. According to the causal theory, the standard process by which the referent of a name is determined is as follows: a new name “N” for an object o is introduced by a naming ceremony. After that, “N” is transmitted through conversation among speakers, and these name-transmissions comprise communicative causal chains of “N” in their linguistic community. Given this situation, a speaker’s particular use of “N” succeeds in referring to o, only if she is properly connected to these communicative causal chains that originate from the naming ceremony for o.

Of course, it is well known that a simple version of the causal theory has difficulty in explaining the phenomena of reference shifts, as Gareth Evan (1973) first pointed out, and that some developed versions of the causal theory for explaining reference shifts had been suggested.Footnote 7 However, it is not my goal to explore which version of the causal theory would be most promising; my arguments in this paper are not committed to any specific form of the causal theory. Rather, I will focus primarily on the idea that appears in Kripke’s following passage when Kripke introduces his causal picture:

When the name is “passed from link to link”, the receiver of the name must, I think, intend when he learns it to use it with the same reference as the man from whom he heard it. If I hear the name “Napoleon” and decide it would be a nice name for my pet aardvark, I do not satisfy this condition (Kripke, 1980: 96).

That is, whether my utterance of “Napoleon” semantically refers to the historical figure or my pet aardvark depends on which semantic referential intention I have in a conversational situation. If I intend to semantically refer to whoever name-givers’ uses of “Napoleon” refer to and thereby participate in the linguistic practice about the pre-existing name “Napoleon”, i.e., if I have the reference borrowing intention for my use of “Napoleon”, then the semantic referent of my utterance of “Napoleon” will be the historical figure. But if I decide to name my pet aardvark “Napoleon” after the historical figure, then a new naming ceremony is carried out and a new homonymous given name “Napoleon” is introduced for my pet. So, in this situation, if I intend to semantically refer to my pet, the semantic referent of my utterance of “Napoleon” will be my pet.

With this explanation in mind, suppose I utter the following sentence:

  1. (1)

    Napoleon is my pet.

Is the proposition expressed by my utterance of (1) is true or false? Again, it depends on which semantic intention I have when I use “Napoleon” that serves as disambiguation. Since the semantic referent of “Napoleon” in (1) cannot be determined in the absence of a speaker’s semantic intention, it is misleading to ask whether or not sentence (1) itself is true in a context where a speaker’s semantic intention is not specified.Footnote 8,Footnote 9

We can have a better understanding of this situation if we distinguish between a Kaplanian generic name and a specific (or common currency) name (Kaplan, 1990):Footnote 10 the historical figure Napoleon and my pet Napoleon share the same generic name “Napoleon,” but since their specific names refer to different objects, they have distinct specific names, so to speak, “Napoleon”h for the historical figure and “Napoleon”p for my pet. Thus, when I utter the generic name “Napoleon” in (1), my semantic intention determines which of two specific names I am uttering, “Napoleon”h and “Napoleon”p, which refer to different objects and therefore have different semantic contents.Footnote 11 In what follows, when I mention names in a context where the referents of names are not yet determined, such as “Napoleon” in (1) in the absence of a speaker’s semantic intention, I mean generic names; when I mention names that have their own determined semantic contents, I mean specific names; when it is clear which kinds of names I mean and there is no danger of confusion, I will omit the words “generic” and “specific” for simplicity.

Let us now consider another similar example: suppose I was very impressed by the biography of Napoleon and made a 5-meter statue of Napoleon in my front yard and named the statue “Napoleon” after the historical figure. Now consider the following sentence:

  1. (2)

    Napoleon is 5 m tall.

In this situation, again, the literal truth of my utterance of (2) depends on which semantic intention I have that disambiguates “Napoleon”. If I intend to semantically refer to the historical figure whom other speakers refer to, it is false, whereas if I intend to semantically refer to the statue I made, it is true.

Let us apply this observation to ordinary names used in fictional contexts. Consider the following two scenarios:

Real person Napoleon

I was very impressed by the biography of Napoleon, and I decided to write a novel about him. While writing the novel, I intend for my uses of “Napoleon” to semantically refer to the historical figure whom other speakers’ uses of “Napoleon” refer to.

Fictional character Napoleon

I was very impressed by the biography of Napoleon, and I thought that “Napoleon” would be a nice name for a fictional character in my new novel. So, I decided to create a new fictional character and named it “Napoleon” after the historical figure. While writing the novel, I intend for my uses of “Napoleon” to semantically refer not to the historical figure but to the newly created fictional character.

Suppose that in both scenarios, the following sentence appears in my novel:

  1. (3)

    Napoleon plays basketball.

What is the semantic referent of my inscription of “Napoleon” in each scenario? As in the previous cases about the aardvark Napoleon and the statue Napoleon, I suggest that the answer depends on which semantic intention I have in each case: in Real Person Napoleon, the semantic referent of my inscription of “Napoleon” is the historical figure Napoleon: I de re pretend about him that he plays basketball. By contrast, in Fictional Character Napoleon, a new homonymous specific name “Napoleon” is introduced for my fictional character and the semantic referent of my inscription of “Napoleon” is the fictional character created by me: I de re pretend about the fictional character Napoleon that it plays basketball.Footnote 12

Sentence (3), which we have considered so far, is a so-called intra-fictional sentence. But I suggest that the same explanation applies to the following so-called internal and external metafictional sentences:

  1. (4)

    According to my novel, Napoleon plays basketball.

  2. (5)

    Napoleon is a fictional character.

Again, in Real Person Napoleon, since I intend for my uses of “Napoleon” to semantically refer to the historical figure, the semantic referent of “Napoleon” in my utterances of (4) and (5) is the historical figure, and my utterance of (5) is false. By contrast, in Fictional Character Napoleon, since I intend for my uses of “Napoleon” to semantically refer to the fictional character, the semantic referent of “Napoleon” in my utterances of (4) and (5) refer to the fictional character, and my utterance of (5) is true.

I believe that my explanation so far is intuitive and plausible. In what follows, I will characterize the view I will advocate in this paper as follows and name it “intention-based non-exceptionalism” (“INE” for short):Footnote 13

Intention-based non-exceptionalism

The semantic referent of a speaker S’s use of an ordinary name “N” in a fictional context is a real object, if S has the semantic intention for S’s use of “N” to refer to whatever object that other speakers’ uses of “N” refer to, which S believes to be real, and that object is in fact a real one.Footnote 14,Footnote 15

Given the above characterization, exceptionalists, in order to reject INE, should argue that even in the Real Person Napoleon scenario where I have the reference borrowing intention for my uses of “Napoleon” while writing sentence (3), this referential intention of mine is frustrated for some reason and the semantic referent of my use of “Napoleon” becomes fixed as the fictional surrogate of the real Napoleon.

In the next section, I will first provide two arguments for INE. Then, in Sect. 4, I will argue that excepionalists’ arguments do not succeed in showing that there is a convincing reason to believe that even when a speaker has a reference borrowing intention in a fictional context, that intention always fails to contribute to fixing the semantic referent of a name.

3 Arguments for intention-based non-exceptionalism

In this section, I will provide two arguments for INE. The purpose of these arguments is not to conclusively refute exceptionalism, but to show that there are compelling reasons in favor of INE. So, once it turns out that arguments for exceptionalism are not convincing (as I will argue in the next section), we will be justified in accepting INE.

3.1 Argument from a lie and imagination

The first argument I will provide is based on Terence Parsons’s (1980: 57 − 9) observation that there seems to be no difference concerning reference in the following situations:

  1. i.

    Telling a lie about Jimmy Carter.

  2. ii.

    Telling a lie about Carter which is very long (e.g., book length).

  3. iii.

    Making up a story about Carter which is not intended to deceive anyone, and which contains falsehoods.

  4. iv.

    Writing a work of fiction in which Carter is a character.

Let us suppose Parsons uses the following sentence in each situation:

  1. (6)

    Carter is 2 m tall.

According to Parsons, when he utters (6) in (i), since he is telling a lie about the real person Carter, his utterance of “Carter” refers to the real Carter. The same explanation seems to be applied sequentially to the situations from (ii) to (iv). After all, when he writes sentence (6) in his novel in (iv), since his novel is about the real person Carter, his inscription of “Carter” refers to the person Carter.

I believe that Parsons’s argument goes in the right direction, but it is not yet convincing as it stands. As Ioan-Radu Motoarca (2014) and Garcia-Carpintero (2019) have rightly pointed out, appealing to our ordinary intuition about aboutness is not sufficient to strongly support that Parsons’s inscription of “Carter” in (iv) refers to the real Carter. For insofar as our intuition that Parsons’s novel is about the real Carter is one about the ordinary malleable notion of aboutness, it can be explained away without difficulty even if his inscription of “Carter” refers to the fictional surrogate of Carter: since the fictional surrogate created by Parsons has a certain intimate relation with the real Carter, whether it is a similarity relation (Bonomi, 2008) or a representation relation (Motoarca, 2014), Parsons’s novel is still in some sense about the real Carter, and thus our intuition of aboutness is explained away.

A similar criticism can be applied to Stacie Friend’s argument for non-exceptionalism, which is based on the idea that our attitudes in response to Parsons’s use of (6) in (iv) are directed toward the real Carter, and his novel invites us to imagine about the real Carter that he is 2 m tall (Friend, 2000, 2019). Even if it is true that in reading Parsons’s novel we are invited to imagine singular propositions about the real Carter, this does not guarantee that Parsons’s inscriptions of “Carter” semantically refer to the real Carter. To see why, consider the following three sentences:

  1. (7)

    Biden is on TV.

  2. (8)

    The current U.S. president is on TV.

  3. (9)

    I am watching TV.

Suppose Jennie wants to invite Lisa to believe the singular proposition that Biden is on TV. Jennie can do this by uttering any of the above three sentences given an appropriate conversational situation: Jennie’s uttering (7) is obviously the most straightforward way to do her job because (7) semantically expresses the singular proposition that Biden is on TV. Jennie’s uttering (8) can also invite Lisa to believe that Biden is on TV in many ordinary conversational contexts where Lisa already knows that Biden is the current U.S. president (and Jennie knows that Lisa knows it). Furthermore, if Lisa knows that Jennie will watch TV when Biden is on (and Jennie also knows that Lisa knows this), by a familiar process of Gricean conversational implicature, Jennie’s uttering (9) can also invite Lisa to believe that Biden is on TV. However, it is clear that only (7) contains a term whose semantic content is Biden. Being invited to entertain a singular proposition about an object o does not show that the invitation is made by a sentence containing a term that directly refers to o. Returning to our original example, even if the name “Carter” in (6) appearing in Parson’s novel refers to the fictional surrogate of Carter, it can surely invite us to imagine a singular proposition about the real Carter.

A lesson from the discussion so far, I believe, is that we should directly focus on whether a speaker can have a successful semantic intention to refer to a real object in a given situation: what is missing in Parsons’s original argument is the consideration of referential intentions. In what follows, I will provide a revised version of the argument that takes account of a speaker’s referential intention, which is immune to exceptionalists’ criticism we have seen.

Let us first assume that I have a clear reference borrowing intention for my use of “Carter” to semantically refer to the person Carter whom other speakers’ uses of “Carter” refer to in each of the following three situations:

  1. (a)

    I utter sentence (6) when I lie about the real Carter.

  2. (b)

    I utter sentence (6) to report what I imagine now about the real Carter.

  3. (c)

    I inscribe sentence (6) while writing a novel about the real Carter.

I believe that there is at least no prima facie reason to believe that my semantic referential intention is frustrated in any of three situations. In situation (a), it seems obvious that my use of “Carter” successfully refers to the real Carter. Likewise, in situation (b), given that I imagine about the real Carter and report what I imagine by uttering (6) with the referential intention to semantically refer to the real Carter, it seems clear to me that there is nothing that can frustrate my semantic intention. Situation (c) is essentially the same as (b), except that my act of de re pretense about the real Carter is related to the writing activity. So, there also seems to be no reason to believe that my semantic intention is frustrated in (c). Of course, if I decide to create a fictional character and intend for my use of “Carter” to semantically refer to this fictional character I have newly created, as in the Fictional Character Napoleon case, I grant that my use of “Carter” semantically refers to the fictional surrogate of Carter. However, ex hypothesi, we are only considering cases where I have the semantic intention to refer to the real Carter.

Then, is there any possible case where a speaker, even if she has the semantic intention to refer to a real object, actually refers to a fictional object? Zofia Zvolenszky’s following scenario might be considered as a counter-example to my INE:Footnote 16

[I]magine the following (contrary to fact) Scenario T: while writing War and Peace, Tolstoy was under the mistaken impression that the protagonist, Prince Bolkonsky, like Napoleon (also featured in the novel), was a real person. Introducing the name “Andrei Bolkonsky,” Tolstoy intended to refer to a historical figure he thought existed quite independently of his novel (Zvolenszky, 2016: 319).

In Scenario T, it seems that the fictional character Bolkonsky is inadvertently created, and one might think that Tolstoy in Scenario T intends for his uses of “Bolkonsky” to refer to a real object, but he is actually referring to a fictional character.

However, even if it is true that Tolstoy is actually referring to a fictional character in Scenario T, this does not counter my INE. Note that INE only concerns cases where there is a real object in question. INE is not committed to any claim about what happens if a speaker mistakenly believes that there is a relevant object she refers to when in fact there is no such object at all. Whatever the best theory about so-called reference failure cases is, there is no problem for me to accept that theory. It is just as a case of reference failure does not pose a threat to the following widely accepted claim (“IRT” for short):

Intention-based reference theory

The semantic referent of a speaker S’s use of an ordinary name in an ordinary context is a real object, if S has the semantic intention for S’s use of “N” to refer to whatever object that other speakers’ uses of “N” refer to, which S believes to be real, and that object is in fact a real one.

IRT of course does not commit itself to any claim about reference failures. The same goes for INE.

To make this clear, consider the following two theories about failure to refer to real objects:Footnote 17

Theory A

If S has the semantic intention for S’s uses of “N” to refer to a real object, but there is no such real object to which S’s uses of “N” refer, then S’s uses of “N” semantically refer to a so-called mythical object.Footnote 18

Theory B

If S has the semantic intention for S’s uses of “N” to refer to a real object, but there is no such real object to which S’s uses of “N” refer, then S’s uses of “N” fail to semantically refer.Footnote 19

Regardless of which of the two theories is correct, my INE can go with that theory: Tolstoy’s semantic intention to refer to a real object might be frustrated because there is no such real object, but INE says nothing about this case. Suppose Theory B, which I prefer, is correct. According to Theory B, Tolstoy in T fails to refer and comes to de re pretend about nothing that it is so-and-so, because he has the semantic intention to refer to a real object, not a mythical or fictional object, and because there is as yet no cognitive connection between Tolstoy and the fictional character Bolkonsky. Only we as readers, who realize that Bolkonsky is not a real person, can refer to the fictional character Bolkonsky.Footnote 20 But if Theory A is correct, then Tolstoy himself refers to the mythical (and also fictional) object Bolkonsky.Footnote 21 Again, whether Tolstoy’s uses of “Bolkonsky” refer to a fictional character or fail to refer, it does not counter INE just as it does not counter IRT. INE is consistent with both theories about reference failures. To refute INE, exceptionalists should provide a case where a speaker’s semantic intention to refer to a real object is frustrated even if there is a real object in question.

Returning to our Carter example, then, it seems a serious challenge for exceptionalists to explain how my referential intention to refer to the real Carter in situation (c) is frustrated even if there is a real person Carter, and the semantic referent of my use of “Carter” is eventually determined to be the fictional surrogate of Carter. Consequently, if the arguments for exceptionalism that we will consider in Sect. 4 fail to provide any good reason to believe that my referential intention is frustrated, we will finally be justified in accepting INE.

3.2 Argument from numerals

The second brief argument I will provide notes that INE invariably respects our intuition regarding numerals, which seem to function semantically very much like Millian names. Conversely, there is a risk of over-generalization in accepting exceptionalism: exceptionalists seem to commit to the claim that when numerals are used in fictional contexts, they also always change their semantic contents and work the same way as other fictional numerals.Footnote 22 But this is certainly counter-intuitive at first glance.

For the sake of argument, let us assume for a moment that there are numbers directly referred to by numerals,Footnote 23 and suppose the following sentence appears in the Sherlock Holmes stories:

  1. (10)

    Sherlock Holmes likes the number 3.

When Conan Doyle wrote sentence (10), if he intended for the numeral “3” to semantically refer to the number 3 as other speakers did, it seems clear to me that “3” in (10) simply refers to 3. In this case, I see no prima facie reason to believe that Conan Doyle’s referential intention was frustrated and that “3” in (10) refers to the so-called fictional surrogate of the number 3. Of course, I do not deny that an author can create a fictional number and name it “3” after 3. However, ex hypothesi, Conan Doyle did not have any such intention.Footnote 24

I think this observation provides additional support for INE. At the same time, this shows that exceptionalism is further from our ordinary intuition than we first thought. But, of course, there may be a convincing argument for exceptionalism that strongly suggests that numerals as well as ordinary names always change their semantic contents when used in fictional contexts (given that numerals semantically function in the same way as ordinary names), which I have not yet considered. In the next section, I will argue that exceptionalists’ arguments are not successful in defending their view.

4 Against exceptionalism

In this section, I will consider two main arguments for exceptionalism: the argument from substitution failures and the argument from distortion. I will argue that neither of these arguments succeed in motivating exceptionalism.

4.1 Against the argument from substitution failures

Let us first consider the following scenario:

Cicero and Tully

Jennie wrote a fiction about Cicero entitled “Cicero and Tully”. While writing Cicero and Tully, Jennie inscribed the name “Cicero” with the semantic intention to refer to a historical figure whom other speakers’ uses of “Cicero” refer to, and so does the name “Tully”. In Cicero and Tully, Cicero and Tully are two distinct people: Cicero is an orator, but Tully is not.

Now suppose Jennie utters the following internal metafictional sentences:

  1. (11)

    According to Cicero and Tully, Cicero is an orator.

  2. (12)

    According to Cicero and Tully, Tully is an orator.

If “Cicero” in (11) and “Tully” in (12) semantically refer to the same historical figure and thus their semantic contents are the same as expected by non-exceptionalism, then it seems that (11) and (12) should express the same proposition and thereby have the same truth-value. However, our ordinary intuition tells us that her utterance of (11) is true, while (12) is false.

Many exceptionalists believe that this observation provides a decisive reason to prefer exceptionalism to non-exceptionalism.Footnote 25 If ordinary names used in fictional contexts always directly refer to fictional surrogates, then no problem arises: “Cicero” in (11) and “Tully” in (12) directly refer to different fictional surrogates and their semantic contents are different, so there would be no problem with accepting that Jennie’s utterances of (11) and (12) express different propositions and have different truth-values. Exceptionalism seems to be well motivated.

However, I believe that the Cicero and Tully case does not sufficiently motivate exceptionalism. First of all, this seems to be nothing more than a version of the traditional problem of substitution failures within intensional contexts. Therefore, we can expect that referentialists’ usual strategies for solving this problem are equally applied to our case. For example, according to most referentialists’ views, the same proposition can be represented under different modes of presentation (ways of taking, or guises).Footnote 26 Therefore, it can be suggested that an internal metafictional sentence pragmatically conveys information about a mode of presentation under which we should imagine or pretend a certain proposition for properly engaging with a given fiction. So, even if Jennie’s utterances of (11) and (12) express the same true proposition, (12) is misleading because it conveys false information about a mode of presentation (concerning the sentence “Tully is an orator”) under which we should imagine or pretend the proposition that Cicero is an orator when we read Cicero and Tully.Footnote 27 Alternatively, it might be argued that although the semantic contents of Jennie’s uses of “Cicero” and “Tully” are the same as the historical figure, since modes of presentation regarding how we should imagine or pretend the proposition that Cicero is an orator affect the truth-conditions of her utterances of (11) and (12), her utterance of (11) is true, while (12) is false.Footnote 28

Here my point is not that referentialists’ strategies for solving substitution failures should be sacrosanct, but rather that given that there are sophisticated attempts to explain phenomena regarding substitution failures that can also be applied to the Cicero and Tully case, exceptionalism does not seem to be sufficiently motivated. This is especially true for exceptionalists who subscribe to referentialism.

Second, there seems to be no conflicting intentions of Jennie’s that can frustrate her referential intentions to semantically refer to the historical figure Cicero in the Cicero and Tully scenario. Again, I do not deny that if Jennie intends to create two new fictional characters and name them “Cicero” and “Tully” respectively, her uses of those names will refer to different fictional characters, not the historical figure. However, ex hypothesi, Jennie had no such intentions in the Cicero and Tully scenario. Then, it seems implausible to postulate that there is another kind of Jennie’s intention that might override her intention to refer to the historical figure to whom other speakers refer.

With respect to this issue, Alberto Voltolini (2020b: 808 − 11) argues that in the Cicero and Tully scenario, if Jennie’s uses of “Cicero” and “Tully” do not refer to different surrogates but to the same historical figure and are thus substitutable, an inadvertent contradiction in the story whose form is “according to story S, p & ~p” arises. Consider the following true internal metafictional sentence:

  1. (13)

    According to Cicero and Tully, Cicero is an orator, but Tully is not.

If we substitute “Tully” in (13) with “Cicero”, we get the following sentence:

  1. (14)

    According to Cicero and Tully, Cicero is an orator, but Cicero is not.

But allowing this fictional contradiction runs counter to Jennie’s intention to write a coherent story, given that she did not intend to write a kind of postmodern novel that blatantly contains overt contradictions. Therefore, Jennie seems to have two conflicting intentions, and her intention to write a coherent story might override her intention to refer to the historical figure.

However, given how referentialists understand a rational believer,Footnote 29 an explanation that respects such an understanding rather shows that none of Jennie’s intentions are frustrated. For example, suppose Lisa mistakenly believes that “Cicero” and “Tully” refer to two distinct people, and that Cicero is an orator but Tully is not. In this case, we might naturally report her belief by uttering the following true sentence:

  1. (15)

    Lisa believes that Cicero is an orator, but Tully is not.

But no referentialist will argue that, to save Lisa from irrationality, we should conclude that our uses of “Cicero” and “Tully” when uttering (15) fail to refer to the same historical figure, but rather directly refer to some kinds of different (imaginary) surrogates of Cicero. Although uttering (15) ascribes a contradictory belief to Lisa, it does not imply or convey that Lisa believes the proposition and its negation under the same mode of presentation. And that is why she should not be accused of irrationality or incoherence.

Likewise, Jennie wrote a coherent story, unless it invites us to imagine or pretend a proposition and its negation under the same mode of presentation. Therefore, (14) is true,Footnote 30 but just misleading in that it pragmatically conveys the false information that we should imagine or pretend the proposition that Cicero is an orator and its negation under the same mode of presentation for properly engaging with Cicero and Tully. Alternatively, if we assume that while the semantic content of “Cicero” remains the historical figure, modes of presentation regarding how we should imagine or pretend the proposition that Cicero is an orator and Tully is not an orator affect the truth-condition of (14), then (14) would simply be false.

Lastly, exceptionalists’ argument from substitution failures threatens to overgeneralize badly and results in the following consequence: an author cannot import even a pre-existing fictional character into her fiction, just as she cannot import an ordinary real object.Footnote 31 Consider the following scenario:

Cent and Kenk

Mark wrote a fan fiction about Clark Kent entitled “Cent and Kenk”. While writing Cent and Kenk, Mark intended to import the pre-existing fictional character Clark Kent (aka Superman) into his story and introduced “Klark Cent” and “Clart Kenk” as distinct names with the semantic intention to refer to the fictional character Clark Kent created by Jerry Siegel. In Cent and Kenk, Cent and Kenk are two distinct people: Cent is a superhero, but Kenk is a reporter, not a superhero.

Now suppose Mark utters the following internal metafictional sentences:

  1. (16)

    According to Cent and Kenk, Cent is a superhero.

  2. (17)

    According to Cent and Kenk, Kenk is a superhero.

Applying the argument from substitution failures we saw with the Cent and Kenk scenario to this case, since our intuition tells us that (16) is true while (17) is false, exceptionalists are forced to draw the counter-intuitive conclusion that we have a decisive reason for believing that a pre-existing fictional character always fails to be imported into another new fiction: Mark’s uses of “Cent” and “Kenk” would refer to different fictional surrogates of the fictional character Clark Kent.

Exceptionalists may bite the bullet and accept that no fictional character can be imported into another new fiction. In doing so, however, they commit themselves to a further counter-intuitive consequence: it seems to follow from this that even when an author writes a sequel to her original novel, she cannot import her original fictional characters into the sequel. Any character she attempts to import becomes a fictional surrogate of the original character. INE, by contrast, does not face any of these counter-intuitive results.Footnote 32,Footnote 33

In this section, I have argued that the argument from substitution failures does not motivate exceptionalism. First, we can apply referentialists’ usual strategies for solving substitution failures to our case such as Cicero and Tully as well. Second, there seems to be no reason to believe that a speaker’s referential intention to refer to a real object is frustrated. Lastly, the argument from substitution failures threatens to overgeneralize badly.

4.2 Against the argument from distortion

Let us begin this section by considering a novel that distorts Napoleon suggested by Motoarca (2014: 1038-42), in which Napoleon is a logarithmic function. Suppose Peter is the author of this novel, which I will call Logarithmic Napoleon, and wrote the following sentence in his novel:

  1. (18)

    Napoleon is a logarithmic function.

Motoarca believes that Peter’s use of “Napoleon” in (18) does not semantically refer to the real Napoleon. For, according to Motoarca, Logarithmic Napoleon is clearly not about the real Napoleon, and if the novel is not about the real Napoleon, it follows that no name appearing in that novel refers to him.

Based on this observation, Motoarca argues that all occurrences of the name “Napoleon” in fictions do not refer to the real Napoleon. Consider a soritical series of fictions in which the name “Napoleon” occurs, beginning with a fiction clearly about the real Napoleon, such as War and Peace, and ending with a fiction that blatantly distorts Napoleon, such as Logarithmic Napoleon. Then there would be an intermediate area in the soritical series where our intuition of aboutness is unclear.Footnote 34 Motoarca argues that since it is not plausible that in that area, the referent of “Napoleon” is indeterminate, nor that there is a definite unknowable cut-off point at which the referent of “Napoleon” changes from a real person to a fictional character, the only available option is to admit that every occurrence of the name “Napoleon” in every fiction in the series does not refer to the real Napoleon. Our intuition that some fictions in the series are about the real Napoleon can be explained away by employing the concept of similarity or representation relation between a fictional surrogate and a real person, as we saw in Sect. 3.1.

However, as Garcia-Carpintero (2019: 165-6) points out, I see no prima facie problem in admitting the semantic indeterminacy of “Napoleon” in an intermediate area or in arguing that there is a definite, unknowable cut-off point. I further believe that one could even argue—though I do not advocate this strategy—that there is a knowable cut-off point by suggesting that “Napoleon” in a story refers to the real Napoleon only if the story describes Napoleon as a person, i.e., the kind to which Napoleon belongs in reality.Footnote 35

But it is important to note that those strategies at least admit that “Napoleon” in Logarithmic Napoleon does not refer to the real Napoleon, which, I believe, is mistaken. In contrast, my reply to the argument from distortion is rather to reject that very shared assumption. Let us consider the following question again: “Does Peter’s use of ‘Napoleon’ in (18) semantically refer to the real Napoleon?” I think that the question is tricky if it is asked in a situation where Peter’s semantic intention for his use of “Napoleon” is unspecified. When we try to answer this question, we seem to implicitly presume that Peter has the semantic intention that he most likely has in that situation. Perhaps, in a situation where authors write fictions in which Napoleon is described as a logarithmic function, it may be true that most authors will not have the semantic intention for their uses of “Napoleon” to refer to the historical figure, and that is why we come to have the intuition that “Napoleon” in Logarithmic Napoleon does not refer to the historical figure.

However, that does not mean that Peter cannot have the semantic intention to refer to the historical figure when he writes sentence (18). In other words, we should distinguish the factual issue of which semantic intention authors will naturally have when they write fictions like Logarithmic Napoleon from the theoretical issue of whether it is possible for authors to have the semantic intention to refer to the real Napoleon in that situation. And it seems clear to me that it is at least possible, just as someone who sees a 5-meter statue of Napoleon can have the semantic intention for her use of “Napoleon” to refer to the real Napoleon, so that her utterance of (2) —that “Napoleon is 5 meters tall”— semantically expresses the false proposition about the real Napoleon that he is 5 m tall.

In Sect. 2, I argued that it is misleading to ask about what the semantic content of “Napoleon” in (1), (2), and (3) is without specifying the speaker’s semantic intention. Likewise, I believe that it is misleading to ask about what the semantic content of “Napoleon” in (18) is without specifying Peter’s semantic intention. If Peter intends for his use of “Napoleon” to semantically refer to a fictional character when he writes sentence (18), I grant that his use of “Napoleon” refers to a fictional surrogate. Then what if Peter has the intention to refer to the historical figure? In what follows, I will argue that there is no reason to believe that Peter’s semantic intention to refer to the historical figure is frustrated.

Let us first note that there is a false atomic singular proposition about the real Napoleon whose constituents are Napoleon and the property of being a logarithmic function,Footnote 36 and name this proposition “NL”. One can certainly use sentence (18) to semantically express NL if she has the semantic intention for her use of “Napoleon” to refer to the real Napoleon, even though she knows that NL is blatantly false and so disbelieves it. In fact, our very disbelief towards NL entails that we can at least entertain NL that is about the real Napoleon.

Once it is given that we can entertain a blatantly false proposition, NL, it seems that we can also propositionally imagine or pretend it.Footnote 37 Of course, NL is impossible, but as Tamar Szabó Gendler (2000: 66–72) points out, this does not mean that we cannot imagine or pretend NL under every mode of presentation. We can imagine or pretend an impossible proposition under a proper mode of presentation that disguises its blatant inconsistency, and that is why we can imaginatively engage with fictions that contain something logically or conceptually impossible.Footnote 38 For example, even if it is impossible that something is a banana and a pistol at the same time, we can de re pretend about a particular banana that it is a pistol by focusing on their similarities, such as shape, while ignoring others, such as internal complexity, when we engage in a certain prop-oriented game of make-belief (Gendler, 2000: 69–70).Footnote 39 Likewise, one can de re pretend about the real Napoleon that he is a logarithmic function under a proper mode of presentation that the novel Logarithmic Napoleon prescribes. Suppose that in Logarithmic Napoleon, Napoleon becomes a logarithmic function under a curse and undergoes mathematical integration. Indeed, we can follow and enjoy this story and understand, at least to some extent, what happens and why Napoleon is subject to integration.

Furthermore, the fact that we can pretend an impossible proposition might become more evident if we consider that, according to referentialism, one can even believe an impossible proposition, such as the proposition that Cicero is not Tully, under a certain mode of presentation. For those who believe an impossible proposition, there will certainly be no psychological difficulty in imagining or pretending the same proposition.

After all, while writing Logarithmic Napoleon, if Peter intends for his uses of “Napoleon” to refer to the real Napoleon, not any fictional character, when Peter inscribes sentence (18), he de re pretends about the real Napoleon that he is a logarithmic function: his inscription of (18) semantically expresses NL, and Peter pretends NL to be true. Peter can distort the real Napoleon, that very historical figure, in his novel, by inscribing sentences that express impossible singular propositions about the real Napoleon. Then it also follows that the novel is about the real Napoleon, in the technical sense that his inscriptions of sentences express singular propositions about the real Napoleon. Given this explanation, there seems to be nothing that frustrates Peter’s semantic intention to refer to the real Napoleon.Footnote 40

5 Conclusion

In this paper, I have argued that there are compelling reasons in favor of INE, while there is no reason to believe that an author’s semantic intention to refer to a real object is always frustrated. I think we are now justified in accepting INE. As our natural intuition tells us, the semantic content of an ordinary name used in a fictional context can be a real object.