THIS PAPER HAS BEEN ACCEPTED FOR PUBLICATION BY THE BRITISH JOURNAL OF

AESTHETICS, PUBLISHED BY OXFORD UNIVERSITY PRESS.

PLEASE CONSULT THE PUBLISHED VERSION FOR CITATIONS.

Grounding Interpretation Christian Folde

In this paper I examine the relationship between interpreting a fiction and specifying its content. The former plays a major role in literary studies; the latter is of central concern in the philosophical debate on truth in fiction. After elucidating these activities, I argue that they do not coincide

but have interesting interdependencies. In particular, I argue that correct interpretations are metaphysically grounded in fictional content. I discuss this claim in detail and show why it is not in tension with the evidential claim that correct interpretations give us epistemic access to fictional

content, which I also endorse.

1. Introduction

It is natural to think that interpreting a literary fiction and determining its content are

intimately related. After all, it is often said that interpretations give us the meaning of fictions. Is interpretation perhaps nothing but content determination? If not, does inter- pretation go beyond content or the other way around? And, what is more, does content depend on interpretation or does interpretation depend on content? Answers to these

questions will, of course, hang on how we understand their key terms: interpretation, content and dependence.

Developing a general theory for the determination of fictional content is a project pur-

sued in philosophy under the label ‘truth in fiction’. Interpreting literary fictions, on the other hand, is an activity performed primarily in literary studies. The main goal of my paper is to investigate how interpretation and truth in fiction relate. I argue that inter-

pretation and fictional content are linked via the notion of metaphysical explanation or grounding: interpretation is grounded in fictional content.

The plan is as follows. I start by briefly introducing the debate on truth in fiction (Section 2). Next, I delineate and examine the type of interpretation of a literary fiction

that I focus on, namely those that can be understood as argumentations for one or more hypotheses (Section 3). Based on this understanding, I address the interdependencies of content-determination and interpretation (Section 4). First, I show why these activities

do not coincide. Second, I argue that correct interpretations are metaphysically grounded in fictional content. I discuss this claim in detail and show why it is not in tension with the evidential claim that correct interpretations give us epistemic access to fictional content,

which I also endorse. Before we start let us get a terminological issue out of the way. In the case of literary

fictions, which I will be solely concerned with, we can distinguish between a particular text (e.g. the copy of On the Road I am reading), a text type which allows for multiple

tokens (e.g. the type your copy of On the Road and mine are tokens of), a certain content which can be realized by texts in different languages (e.g. the content a German and a French version of On the Road have in common) and, arguably, a certain work of art. Now

what exactly should we say is the fiction? This is certainly not the place to examine this

question in any detail. For the purposes of this paper I will simply adopt the naive view and identify fictions with the contents of fictional texts. However, it is certainly natural to speak of the content of a fiction and of what is true in a fiction, and I will continue to do

so. In these cases, then, ‘fiction’ should be understood to mean a fictional text.

2. Truth in Fiction

A central concern in the philosophy of fiction is the question of whether it is possible to

determine in general what is true in any given fiction and, if so, how. For example, it is true in F. Scott Fitzgerald’s The Great Gatsby (henceforth GG) that Gatsby is rich, that Nick is mortal, and that Tom and Daisy are married. However, it is not true in GG that

Nick works on an oil rig, that Gatsby is an elephant and that New York City is invaded by aliens. Thus, something is true in GG, i.e. the fiction is not void of content, but not everything is true in GG, i.e. the fiction is not logically trivial. These two observations

can be generalized for almost all literary fictions, from children’s stories to high litera- ture, from conventional to avant-garde texts, from tragedy to science fiction.1 For ease of expression, I will call all and only that which is true in a given fiction its ‘content’. From a logical perspective the content of a fiction is simply a collection of propositions.

Its members are called fictional truths. This label, however, is somewhat misleading because many fictional truths are in fact false. For instance, that the train for Hogwarts leaves from platform 9¾ at King’s Cross Station is a fictional truth of Harry Potter, but,

unfortunately for many children, this proposition is false. An example of a fictional truth of Harry Potter, which is in fact true is that London is the capital of England. The core question of the truth-in-fiction-debate can be rendered thus: which collection of fictional

truths is to be associated with a fiction? Several theories are on the offer.2 However, I will neither discuss them here, nor opt for one of them. Instead I want to focus on the inter- nal structure of fictional content.

It has often been observed that the content of a fiction can be divided into sub-con-

tents.3 Typically a dichotomy between explicit and implicit content is proposed. For instance, that Gatsby is rich is part of the explicit content of GG, while that Nick is

1 At least two interesting questions arise. First, are there logically trivial fictions? I am inclined to think that it is

possible to design an inconsistent fiction where indeed everything is the case (Richard Routley, ‘The Semantical

Structure of Fictional Discourse’, Poetics 8 (1979), 8). Second, are there fictions without any content? I am

not entirely sure, but it seems to me that even when we consider more experimental cases of literary fictions

(Beckett, Borges, Robbe-Grillet, etc.) there is always some content to be found. According to Kendall Walton,

all literary fictions invite make-believe (Mimesis as Make-Believe (Cambridge, MA: Harvard University Press,

1990)). Since make-believe is propositional in nature (we make-believe that p), all fictions always have some

content.

2 See, for example, Nicholas Wolterstorff, ‘Worlds of Works of Art’, Journal of Aesthetics and Art Criticism 35 (1976),

121–132; David Lewis, ‘Truth in Fiction’, American Philosophical Quarterly 15 (1978), 37–46; Walton, Mimesis as Make-

Believe, 138–190; and Gregory Currie, The Nature of Fiction (Cambridge: Cambridge University Press, 1990), 52–98.

3 See, for example, Lewis, ‘Truth in Fiction’, 41; Nicholas Wolterstorff, Works and Worlds of Art (Oxford: OUP,

1980), 115ff; and Walton, Mimesis as Make-Believe, 140ff.

mortal is merely implicit. I think this distinction is important but unfortunate in two respects. First, ‘explicit content’ may easily be understood to mean literal content of

the text. But not everything which is literally expressed in a fiction needs to be true in it, due to phenomena such as figurative speech, irony and unreliable narration. For example, the last sentences of GG reads: ‘So we beat on, boats against the current, borne

back ceaselessly into the past.’ However, what this sentence literally says is not true in GG; what is true is rather that human beings are in a certain sense like boats which beat against the current and so on. Instead of ‘explicit content’ I prefer the term ‘core content’. Presumably, figuring out what belongs to the core content of a fiction involves not only

semantic, but also pragmatic mechanisms (e.g. Gricean implicatures). In general, the core content of a literary fiction seems to heavily depend on linguistic conventions and on the intentions of its author(s).4

Second, two distinct phenomena are overlooked by pooling all other fictional truths into the category of implicit content. Instead, we should distinguish between ‘imported content’, on the one hand, and ‘entailed content’, on the other. Let me elaborate.

Roughly speaking, imported content comprises all that which is brought into the fiction from outside. For example, it is true in GG that all human beings are mortal, even though this is not part of its core content, nor does it follow from it. We naturally import this propo- sition into the fiction: if asked whether all human beings are mortal in the world of GG,

saying ‘yes’ is the right answer. Two important questions arise: what are imports and how are they to be restricted? David Lewis, for instance, has advanced two influential proposals: the first has it that imports are truths of the actual world, whereas according to the second,

imports are overt beliefs of the community where the fiction originated.5 Both proposals have it that in GG human beings are mortal, because it is both true and an overt belief of the community of origin. However, only the first has it that in GG copper has atomic num-

ber 29, because, albeit being true, it is not an overt belief in the relevant community. Both import principles have been criticized.6 For the purposes of this paper, I want to leave it open whether imported propositions are truths, beliefs, or what have you and how they are to be restricted—what matters is that fictions have imports among their content.7

4 Note that from the fact that core content does, in part, rely upon authorial intentions, it does not follow that

exclusive intentionalism is true, that is, the view that when interpreting fictions the only thing to look for are the

intentions of the author (or some other entity postulated in literary theory).

5 See Lewis, ‘Truth in Fiction’, 42 and 45.

6 See Currie, The Nature of Fiction, 65; Walton, Mimesis as Make-Believe, 148ff; Alex Byrne, ‘Truth in Fiction: The

Story Continued’, Australasian Journal of Philosophy 71 (1993), 24–35; David Davies, ‘Fictional Truth and Fictional

Authors’, BJA 36 (1996), 43–55; John F. Phillips, ‘Truth and Inference in Fiction’, Philosophical Studies 94 (1999),

273–293; Andrea Bonomi and Sandro Zucchi, ‘A Pragmatic Framework for Truth in Fiction’, Dialectica 57

(2003), 103–120; Richard M. Sainsbury, Fiction and Fictionalism (London: Routledge, 2010), 77.

7 There are two special kinds of imports I want to mention briefly. First, there are content-related literary

conventions. Goethe’s Faust, for instance, is written in iambic verse. But it doesn’t seem to be true in the fiction

that, say, uneducated Gretchen speaks in iambic verse while locked in the dungeon awaiting execution. Second,

there is import from one fiction to another, as is typical in serial fiction, such as Conan Doyle’s Holmes stories.

The protagonist lives in 221B Baker Street even if it isn’t explicitly mentioned in every story of the series.

Finally, entailed content comprises what follows non-trivially from core content and/ or imported content. If a fiction has it that p, and that q, and if classical logic holds at the

world of the fiction, it is also true in the fiction that p and q. A more complex example is the following. That Santiago is mortal is true in The Old Man and the Sea, but, arguably, it is not an import.8 Rather, it seems that it is entailed by the core fictional truth that Santiago

is a fisherman and the imported fictional truth that all (fisher)men are mortal. Note that some fictions such as Graham Priest’s short story Sylvan’s Box don’t abide by the rules of classical logic.9 Thus, entailed content very much depends on the notion of entailment applicable to the fictional world at hand.10 Moreover, it is debatable whether entailed con-

tent should be restricted to some notion of deductive entailment or rather broadened to include strong inductive inferences and/or pragmatic presuppositions.

Given what I have said about the internal structure of fictional content, what is true in a

fiction can structurally be defined as the collection of consequences of core and imported content. This is not to say, however, that, when figuring out what is true in a fiction, we proceed by first identifying core content, then detecting imported content and, finally,

drawing the appropriate consequences. Rather, instead of being stepwise, our procedure is dynamic and interest-dependent. The structural definition, however, nicely displays that fictional content stems from many different sources: what is true in a fiction is a complex product of factors such as linguistic meaning, authorial intention, imports and

entailments.

3. Interpretation

Besides being act-object-ambiguous, the term ‘interpretation’ can mean many differ-

ent activities performed by different agents with different purposes on different objects of investigation.11 In particular, even when focussing on the kind of activity that lit- erary scholars perform when critically engaging with literary fictions, there are fun-

damental differences. Richard Shusterman argues that there are various interpretative games that are actually being played, and that therefore interpretative statements do not have a single logical form.12 One such game is argumentative: interpreters assert things,

they argue for claims and their (assertive) sentences express propositions which can be

8 It is certainly not an overt belief of the community of origin since very few people in that community had any

beliefs about the mortality of Santiago. Arguably, it is not a truth because either Santiago does not exist or he is

an abstract object, an artefact of literary practice and thus not the kind of thing which can be mortal. It could be

an import, however, if, say, the beliefs of interpreters count as imports.

9 Graham Priest, ‘Sylvan’s Box: A Short Story and Ten Morals’, Notre Dame Journal of Formal Logic 38 (1997),

573–582.

10 Routley, ‘The Semantical Structure of Fictional Discourse’, 10.

11 See Göran Hermerén, ‘Interpretation: Types and Criteria’, Grazer Philosophische Studien 19 (1983), 131–161.

12 Richard Shusterman, ‘The Logic of Interpretation’, Philosophical Quarterly 28 (1978), 310 –324. By

interpretive statements Shusterman means statements such as ‘Measure for Measure is allegorical’ and ‘Hamlet

is a study of a vacillating, melancholy hero’, which, in the terminology I develop, would be classified as

content-transcending.

true or false.13 Another game is recommendatory: interpreters recommend, suggest or instruct how to view a literary fiction, they don’t usually assert or argue. According to

Shusterman someone playing this game may give reasons for her recommendations but ‘the relation of reasons to an interpretative conclusion is not logical but causal or psy- chological’.14 Note, however, that the recommendatory game is in principle compatible

with the argumentative one, that is, an interpreter could play both in one and the same interpretative text. Clearly, though, these games have different goals (getting it right, say, versus maximizing aesthetic enjoyment) and they are evaluated in different ways. For instance, only argumentative interpretations can be correct, plausible or adequate.15

Dagfinn Føllesdal argues that all interpretations of literary fictions proceed by the hypothetico-deductive method, that is, roughly, by formulating a hypothesis, deducing consequences and checking with the data.16 If so, all interpretations are argumentative.

Føllesdal cites several interpretations of Ibsen’s Peer Gynt that seem to speak in his favour. Given Shusterman’s observations, however, Føllesdal’s claim should be rejected because it suffers from a one-sided diet of examples. I do, however, think that some argumentative

interpretations can at least be reconstructed as hypothetico-deductive; some, however, are better seen as inferences to the best explanation. In any case, I will adopt Føllesdal’s terminology and speak of interpretation hypotheses.

In what follows I will focus on argumentative interpretations only. A famous example

is Sigmund Freud’s interpretation of Hamlet:

Another of the great poetic tragedies, Shakespeare’s Hamlet, is rooted in the same soil as Oedipus Rex. The play is based upon Hamlet’s hesitation in accomplishing the task of

revenge assigned to him; the text does not give the cause or the motive of this hesita- tion, nor have the manifold attempts at interpretation succeeded in doing so. According to the still prevailing conception, a conception for which Goethe was first responsible,

Hamlet represents the type of man whose active energy is paralysed by excessive intel- lectual activity: ‘Sicklied o’er with the pale cast of thought.’ According to another con- ception, the poet has endeavoured to portray a morbid, irresolute character, on the verge of neurasthenia. The plot of the drama, however, shows us that Hamlet is by no

means intended to appear as a character wholly incapable of action. What is it, then, that inhibits him in accomplishing the task which his father’s ghost has laid upon him? Here the explanation offers itself that it is the peculiar nature of this task. Hamlet is

able to do anything but take vengeance upon the man who did away with his father and has taken his father’s place with his mother—the man who shows him in realization the

13 Shusterman distinguishes several varieties or sub-games of this type of game depending on what kind of

proposition an interpretive statement is taken to express (ibid., 312).

14 Ibid., 319.

15 Shusterman delineates one further kind of interpretative game where interpretation involves some kind of

performance (ibid., 316ff). However, I have difficulties getting a grip on this kind. In particular, it is not clear to

me what kinds of speech acts are involved in it.

16 Dagfinn Føllesdal, ‘Hermeneutics and the Hypothetico-Deductive Method’, Dialectica 33 (1979), 319–336. The

same claim has already been advanced in Heide Göttner, Logik der Interpretation (München: Fink, 1971). Her

prime examples are two competing interpretations of a poem by Walther von der Vogelweide.

repressed desires of his own childhood. The loathing which should have driven him to revenge is thus replaced by self-reproach, by conscientious scruples, which tell him that

he himself is no better than the murderer whom he is required to punish. I have thus translated into consciousness what had to remain unconscious in the mind of the hero.17

No matter whether you agree with this account or not, it is a good example of an argu-

mentative interpretation. Freud argues that Hamlet’s hesitation is best explained by a certain psychological constitution: he (subconsciously) realizes that he is no better than the murderer of his father because to kill his father is a repressed childhood wish of his.

Analogously to what I have said about literary fictions, I will think of the content of an argumentative interpretation of a fiction as a structured collection of propositions. In particular, I propose that the propositions which belong to the content of an argumenta- tive interpretation of a fiction f are, in general, either contentf-independent or contentf-

dependent, where the latter sub-divide into contentf-specifying and contentf-transcending as depicted in the following diagram:

To get clear on these distinctions, let us consider some examples related to GG. Suppose we have an interpretation of GG which involves the claim that Fitzgerald was an alcoholic. This proposition is in an important sense independent of the content of GG: you can render a justified judgement about its truth value without knowledge of

(part of) the content of GG. The same usually holds for many other claims involved in interpretative texts, be they biographic, historical, psychological, linguistic, nar- ratological or what have you. Note that the independent-dependent distinction may, derivatively, be applied to sentences and utterances as well: (an utterance of) the sentence ‘Fitzgerald was an alcoholic’ is contentGG-independent if the proposition it

expresses is. The situation is different when an interpretation of GG involves the claim that Gatsby

is rich. This proposition is in an important sense dependent on the content of GG: you

cannot render a justified judgement about its truth value without knowledge of (part of) the content of GG.18 In my terminology, an assertion of the sentence ‘Gatsby is rich’

17 Sigmund Freud, ‘The Interpretation of Dreams’, ed. and trans. Abraham A. Brill, The Basic Writings of Sigmund

Freud (New York: Modern Library, 1938), 179–550 at 319.

18 Some caution is needed here. There is, of course, a way to render a justified judgement about the proposition

that Gatsby is rich without knowledge of (part of) the content of GG, namely by asking someone who has such

knowledge. Hence, justification by testimony has to be excluded.

when talking about GG is content-specifying because with it one aims at specifying the

content of GG. A simple way to signal this is by using locutions such as ‘according to GG’ or ‘in GG’, that is, by employing so-called fiction operators. In this paper, I will abbrevi- ate them by ‘Ff (p)’, where ‘f’ is a place-holder for singular terms denoting fictions and

‘p’ is a place-holder for sentences.19 Assertions made with the help of fiction operators are explicitly content-specifying, while those lacking fiction operators are only implicitly so and best understood as expressing the same kind of propositions that their explicit

siblings express.20 Otherwise, to assert that Gatsby is rich would not amount to saying something true. The reason is simple. A standard anti-realist about fictional objects will say that the unprefixed sentence ‘Gatsby is rich’ does not express a truth because the

singular term ‘Gatsby’ is empty. A standard realist will hold that ‘Gatsby is rich’ does not express a truth because Gatsby being an abstract artefact of literary practice is not the kind of object which can be rich.21 Note that there are also content-specifying assertions

which are false. For instance, if, in aiming to specify the content of GG I utter ‘Gatsby is an elephant’, I make a false content-specifying assertion. So far I have only character- ized when an assertion (and, derivatively, a sentence) is content-specifying. The transfer

to propositions, however, is fairly easy: a proposition is content-specifying if and only if it is structured thus: <Ff (p)>.22 Note further that content-specifying propositions and fictional truths are intimately linked. In fact, we can define the latter thus: that p is a fic-

tional truth of f ↔df Ff (p). Finally, content-transcending propositions depend upon the content of a fiction, but go

beyond it in an important sense. For instance, consider the propositions expressed by the following sentences:

(1) Jay Gatsby is an enigmatic fictional character. (2) The Great Gatsby is a critique of the Roaring Twenties. (3) In The Great Gatsby Fitzgerald comes to terms with his alcoholism.

Neither of these sentences expresses a proposition which is true in GG, hence, they are not content-specifying. But, at the same time, they are not independent of GG’s

content either. On the contrary, you cannot render a justified judgement about their truth values without knowledge of (part of) the content of GG. Note that there are also content-transcending propositions which are partly content-specifying such as the prop- osition expressed by the inter-fictional comparison that Nick Carraway is more naïve

19 Syntactically, fiction operators are two-place operators on singular terms and sentences which yield sentences.

Their semantics, however, are a controversial issue because of their intimate relation to the notion of truth in

fiction. It is common ground, however, that they are intensional operators and that they are non-factive.

20 There are a number of views out there on how exactly these two kinds of assertions are related. For my purposes,

however, it suffices that implicitly content-specifying assertions somehow express the propositions their explicit

siblings literally express and that theses propositions are part of the content of an interpretation.

21 For an introduction to the debate see, for example, Sainsbury, Fiction and Fictionalism, 22–43.

22 I am presupposing here that propositions are structured entities. This is not to everyone’s liking. We can avoid

such scruples by saying that a proposition is content-specifying if and only if a sentence expressing it has the

logical form <Ff (p)>.

than John Watson. Since this proposition does not have the structure <Ff (p)>, it is not

content-specifying.23

Based on the content distinctions just outlined, I will now argue for two claims about argumentative interpretations. The first is:

CSP Every argumentative interpretation of a fiction f explicitly contains or implic-

itly assumes contentf-specifying propositions.

I take this to be a conceptual insight and thus to hold by necessity. The idea is simply

that without contentf-specifying propositions a text would not count as an argumentative interpretation. In other words, CSP gives a necessary condition for argumentative inter- pretations. Now, why should one believe CSP and what does it mean that an interpreta- tion explicitly contains or implicitly assumes content-specifying propositions?

In argumentative interpretations people argue for certain claims that can aptly be called

interpretation hypotheses. However, not any old claim will count as an interpretation hypothesis. It is a commonplace in literary studies to distinguish between a description and an interpretation of a text. For instance, if you analyse, say, the structure of a lyric fiction and identify the rhetoric figures employed, you have not interpreted it. Likewise,

a narratological analysis of a novel is not an interpretation. Interpretation hypotheses are characterized by being linked to the content of a fiction. In my terminology, they are either content-specifying or content-transcending. Typically, interpretation hypotheses are explicitly backed up by contentf-specifying propositions. Consider again Freud’s inter-

pretation of Hamlet. Freud argues for his hypothesis by claiming, among other things, that Hamlet is hesitant, that he was assigned the task to revenge his father, that the murderer is with his mother and that the fiction does not explain Hamlet’s hesitance. Note that there is no single sentence in Freud’s interpretation which solely expresses a content-specifying proposition. Rather, content-specifying propositions are contained within the proposi-

tional material explicitly expressed. In some cases, however, argumentative interpretations do not explicitly contain

a contentf-specifying proposition.24 In certain contexts the interpreter may, for instance, assume that her audience knows about the (relevant) content of the fiction

and therefore omit any explicit appeal to it. Nevertheless, her argumentation still implicitly relies upon the content of f: her argument would not be valid without assum- ing a content-specifying proposition as a premise. It seems justified therefore to count implicitly assumed propositions as part of the content of an interpretation. Thus, the

content of an interpretation—like that of a fiction—can exceed what is explicitly said in it.

We have seen that every argumentative interpretation involves content-specifying propositions by either explicitly containing or implicitly assuming them. Hence, CSP is

vindicated. Note, however, that the criterion given in CSP is not a sufficient condition.

23 For a way to account for the structure of inter-fictional comparisons from an anti-realist point of view see

Wolfgang Künne, ‘Fiktion ohne Fiktive Gegenstände’, in Maria Reicher (ed.), Fiktion, Wahrheit, Wirklichkeit:

Philosophische Grundlagen der Literaturtheorie (Paderborn: Mentis, 2007), 54–72, 65ff.

24 Thanks to Ivan Milić for pointing this out to me.

This paper, for instance, explicitly contains contentGG-specifying propositions, yet it is

not an argumentative interpretation of GG.25

The second claim about argumentative interpretations that I want to advance is empiri- cal and contingent in nature: there are two types of argumentative interpretation, those involving content-transcending propositions and those which are purely content-specify-

ing. The part of Freud’s interpretation of Hamlet that I have presented earlier would be an example of the latter kind. His hypothesis is content-specifying because what it says is, in a nutshell, that in Hamlet Hamlet has an Oedipus complex. Arguably, the embedded

proposition is entailed by core fictional truths (e.g. Hamlet’s properties and behaviour) and the import of parts of psychoanalytic theory. Consequently, whether this interpreta- tion is acceptable hinges, among other things, on whether the required import is legiti-

mate.26 On a whole, however, Freud’s interpretation is content-transcending because he goes on to argue as follows:

It can, of course, be only the poet’s own psychology with which we are confronted in Hamlet; and in a work on Shakespeare by George Brandes (1896) I find the statement

that the drama was composed immediately after the death of Shakespeare’s father (1601)—that is to say, when he was still mourning his loss, and during a revival, as we may fairly assume, of his childish feelings in respect of his father. It is known, too, that

Shakespeare’s son, who died in childhood, bore the name of Hamnet.27

Freud’s hypothesis is that Hamlet is an expression of Shakespeare’s own psychological state

at the time. This is, of course, a content-transcending hypothesis. Most existent argumen- tative interpretations, I suspect, are content-transcending.

4. Interdependencies

Two Separate Activities

Given what I have said about determining fictional content and interpreting a fiction, it is

fairly easy to see that these two activities do not coincide. Argumentative interpretations of a fiction f typically have three components: there

are contentf-specifying claims, contentf-transcending claims and claims that support the argumentation but are independent of the content of f. Determining the content of a

25 What makes a text an argumentative interpretation of a literary fiction? There is no easy answer here. In addition

to being argumentative, the text must, in a sense, be about the fiction, engage with it instead of just mentioning

or citing it. The fiction shouldn’t merely be a means to an end, but the centre of critical attention.

26 There are at least four interesting scenarios here: (a) psychoanalysis is true and only truths can be imports: good

reason to think that the hypothesis is acceptable; (b) psychoanalysis is true and only what was believed by the

community of origin is a legitimate import: good reason to reject the hypothesis since in Shakespeare’s time no

one could have had such beliefs; (c) psychoanalysis is wrong and only truths can be imports: good reason to reject

the hypothesis; (d) psychoanalysis is wrong but the hypothesis is true on independent grounds: good reason to

think that import is neither restricted to truths nor to beliefs of the community of origin, but extends to, say, the

beliefs of the interpreter.

27 Freud, ‘The Interpretation of Dreams’.

fiction f, on the other hand, involves only contentf-specifying claims. What is more, an interpreter is typically interested in only part of the content of a fiction, whereas deter- mining what is true in a fiction usually aims at capturing its whole content.

It is thus safe to conclude that determining the content of a fiction f is importantly

different from interpreting it. Metaphorically speaking, the philosopher engaged in the truth-in-fiction-debate is like an eager collector who wants to have all and only the fic- tional truths, even the most boring ones. The literary scholar, on the other hand, who is interpreting a given fiction, is more like a treasure hunter searching for interesting

fictional truths and arguing, on their basis, for hypotheses that go beyond them. The dis- parity of these two activities, however, does not mean that they have nothing to do with each other—in the remainder of this paper, I highlight two interesting interdependencies

between them.

Two Because-Claims

Consider the following two claims about correct interpretations, that is, about interpreta-

tions that contain only truths:28

M If an argumentative interpretation of a fiction f is correct, then it is correct because all the contentf-specifying propositions it involves are true.

E Some contentf-specifying propositions are true because they belong to the con- tent of a correct argumentative interpretation of f.

Prima facie, both M and E ring true. At the same time, they seem to be in conflict: it looks as though there is some kind of circularity going on. In what follows, I argue that M and E both have a true reading and that these readings are in no conflict

whatsoever. My strategy is to first make the apparent conflict explicit. Afterwards I point out what I take to be the mistaken step in the argument and go on to discuss M and E in detail.

Suppose there is a particular correct argumentative interpretation i of a fiction f involv- ing a particular contentf-specifying proposition x which is true. Together with E, these starting assumptions give us that x is true because i is correct. Together with M, however, they yield the reverse claim that i is correct because x is true. Conjoining these two claims via the transitivity of ‘because’—that is, the idea that if A explains B and B explains C, then A explains C—gives us: x is true because x is true. This, however, is in conflict with

the irreflexivity of ‘because’, i.e. the idea it is not the case that A explains A. Thus, we have reached a contradiction.29

28 The term ‘correct interpretation’ is sometimes used in a less demanding way, meaning something like ‘plausible’,

‘acceptable’ or the like.

29 Two quick remarks. First instead of deriving ‘x is true because x is true’, we could have just as well derived

the equally bad ‘i is correct because i is correct’. Second, instead of employing the principles of transitivity

and irreflexivity we could have also reached a contradiction via the asymmetry of explanation, that is, the idea

that if A explains B, then it is not the case that B explains A. The asymmetry of ‘because’ is a central feature of

Aristotle’s famous dictum about truth: if p, then that p is true because p, but not vice versa (Aristotle,

Metaphysics, ed. William D. Ross (London: Sandpiper Books 1997), book Θ 10: 1051b6–9).

Note that this argument can be transformed into a formal derivation by invoking well-established principles of the logic of the sentential connective ‘because’, which has

received a lot of attention recently.30 Due to limits of space, however, I have omitted the proof which also reveals that the starting assumptions can be discarded and a contradic- tion obtained from M and E alone.

What are we to do about this impasse? The simplest reaction, of course, is to give up on either one of M or E or on both. Since I think that M and E are true, this is not a live option. A better reaction is to question one (or more) of the employed principles of the logic of ‘because’. However, I do not think that this is promising.31 Instead, I suggest that

the derivation is flawed due to an equivocation of ‘because’ in M and E. In fact, there are two different senses of ‘because’ involved here. Consequently, it is a mistake to conjoin the ‘because’ of M with the ‘because’ of E via transitivity. Hence, M and E are not in

contradiction and the argument breaks down.32 Specifically, I believe that the ‘because’ in M is that of metaphysical grounding, while the ‘because’ in E expresses an epistemic relation- ship. Let me elaborate.

It has been a central concern of recent debates in metaphysics to delineate different notions of dependence and to elucidate different kinds of explanation.33 Suppose Thomas is a bachelor and Max asks him why. If Thomas answers ‘because I never found the right partner’, he gives a causal explanation; whereas when he answers ‘because I am an unmar-

ried man’, he gives a conceptual explanation. In both cases he uses the connective ‘because’, which suggests that ‘because’ is capable of expressing different explanatory relations.34

A currently particularly popular notion of explanation is that of metaphysical grounding,

which concerns ontological or metaphysical priority.35 Grounding is often taken to be a relation between truths. The general idea is that some truths are true in virtue of other truths, their grounds, such that the grounds are prior to their consequences and make

them true. For instance, given that Neil is happy, a physicalist about mental states might say that Neil is happy because certain neurons fire. The firing of certain neurons makes it so

30 Fabrice Correia, ‘Grounding and Truth-Functions’, Logique et Analyse 53 (2010), 251–279; Benjamin Schnieder,

‘A Logic for “Because”’, The Review of Symbolic Logic 4 (2011), 445–465; Kit Fine, ‘Guide to Ground’, in Fabrice

Correia and Benjamin Schnieder (eds), Metaphysical Grounding (Cambridge: Cambridge University Press, 2013),

37–80.

31 But, for criticism of the transitivity of ‘because’, see, for example, Jonathan Schaffer, ‘Grounding, Transitivity

and Contrastivity’, in Correia and Schnieder, Metaphysical Grounding, 122–138.

32 Consider the following parallel case. Assume (i) Alfred is above Bob and (ii) Bob is above Charlie. By transitivity

of ‘above’, (iii) Alfred is above Charlie. Now, suppose that ‘above’ in (i) expresses a spatial relation (Alfred is

standing on the roof of his house and Bob is in the garden), whereas in (ii) it expresses a relation of financial

standing (Bob is a millionaire and Charlie is broke). In this case, in neither reading of ‘above’ does (iii) follow

from (i) and (ii).

33 See, for example, Miguel Hoeltje et al. (eds), Varieties of Dependence (München: Philosophia, 2013).

34 This is not to say, however, that ‘because’ is semantically ambiguous. For a quick overview of different ways to

spell out the ambiguity in question, see Schnieder, ‘A Logic for “Because”’, 447ff; and the introduction in Correia

and Schnieder, Metaphysical Grounding, especially 22 ff.

35 For an introduction, see, for example, Kelly Trogdon, ‘An Introduction to Grounding’, in Hoeltje et al. (eds),

Varieties of Dependence, 97–122.

or grounds that Neil is happy. While there is much more to be said about this notion, this is not the place to do so; however, I will here highlight one more feature of it—namely,

the distinction between full and partial grounds. That snow is white, for instance, fully grounds the disjunctive truth that snow is white or grass is green. Often, however, several truths have to work together to fully ground another truth: that snow is white and that

grass is green together fully ground the conjunctive truth that snow is white and grass is green, while they individually only provide a partial ground.

Returning to M, I think that the ‘because’ involved here is also that of metaphysical grounding. Thus, M says that the correctness of an argumentative interpretation of a fic- tion f is metaphysically grounded in the truth of its contentf-specifying propositions. Now, why think that M is true? Consider any argumentative interpretation i of a given fiction

f. We know that i is correct because all of i’s propositions are true. Via the factivity of ‘because’, that is, the idea that if p because q, then it is true that p, and it is true that q, we obtain: all of i’s propositions are true. By CSP, i contains at least one contentf-specifying proposition x. Universal instantiation yields: x is true. From universal explanation, that

is, the idea that universal generalizations are true because of their instances, we obtain: all of i’s propositions are true because x is true. Finally, via the transitivity of ‘because’, we arrive at: i is correct because x is true. Since, i and x were chosen arbitrarily, M holds.36

Note that typically the truth of content-specifying propositions provides only a partial

ground for the correctness of an argumentative interpretation. For the content of correct argumentative interpretations usually consists of truths other than true content-specify- ing propositions.

In a slogan, we could say that interpretations are grounded in fictional content. Note, however, that argumentative interpretations are not, in general, correct because of fic-

tional truths, since the latter can be false. Remember: that p is a fictional truth of f ↔df Ff

(p). Thus, even though any interpretation i is correct partly because it contains a proposi- tion of the form <Ff (p)>, it is not in general the case that i is correct because it contains the embedded proposition that p.37

M is an important insight into the relationship between argumentative interpretation and truth in fiction because it means that the content of a fiction is prior to or more fundamen-

tal than any correct argumentative interpretation of it. This is not to preclude, however, that interpreters can discover what is true in a fiction, it is only to say that metaphysically speaking, fictions and their content are there first. This claim can easily be misunderstood. It does not mean that the content of a fiction f is fixed in the sense that any proposition is either true in f or not true in f. On the contrary, fictions are typically incomplete, that is,

they neither have it that something is the case nor do they have it that something is not the case; in symbols: ¬Ff (p) & ¬Ff (¬p). However, the content of a fiction is fixed in the sense that the propositions that belong to it are (metaphysically) independent of an argumentative

36 Again, I omit a formal proof. The employed principles of ‘because’ are well-established. See the literature cited

in footnote 29.

37 It is natural to ask what a proposition of the form <Ff (p)> itself is grounded in. As I have pointed out at the end

of Section 2, there are many different sources that account for the content of a fiction: entailments, imports,

intentions, linguistic meanings and what have you. Thus, <Ff (p)> has a variety of grounds.

interpretation. Unfortunately, more often than not are we in no position to know the (whole) content of a fiction. But this does not preclude that the content is fixed in the sense specified.

Let us turn to E, that is, the claim that some contentf-specifying propositions are true because they belong to the content of a correct argumentative interpretation of f. To sup- pose that the‘because’ here is that of metaphysical grounding is highly implausible. It is the fiction not an interpretation of it that makes it true that in GG Gatsby is rich. Likewise, it is in virtue of the fiction not of an interpretation that in Shakespeare’s play Hamlet has an

Oedipus complex, if he does. But we can—and often do—grasp fictional truths via inter- pretation. Interpretations make us aware of fictional truths, they guide our attention and convince us to accept certain content-specifying propositions as true—after all, interpre- tations offer, besides reading, a prime access to the content of fictions. In a slogan, interpre-

tation is the discovery not the creation of fictional truths. And, of course, interpretation is often much more than that. Even though interpretations rely upon the content of a fiction, they typically go beyond it by arguing for content-transcending propositions.

If E is not a metaphysical grounding claim, what is it? The ‘because’ in E, I think, is best

understood as expressing an epistemic relation, which is relative to particular agents. Thus, correct argumentative interpretations give us evidence for content-specifying propositions; they state reasons for why to believe in them.38

To drive the point home, consider the following case. Suppose you see a red T-Shirt. You can now make the following two true because-claims:

(4) The T-Shirt is red because it has certain microphysical features. (5) The T-Shirt has its microphysical features because it is red.

Now, (4) is a metaphysical grounding claim: that the T-Shirt has certain microphysical

features grounds that it is red. For, among other things, these microphysical features

make sure that light is reflected in a certain way such that you perceive it as red. But (5), on the other hand, is an evidential claim: we come to know that the T-Shirt has said microphysical features by seeing that it is red. For, if you see that the T-Shirt is red and

you know your physics, you can conclude what the microstructure of the T-Shirt is. Let me sum up this subsection. We started with two apparently conflicting because-

claims, M and E. As I have shown, however, this conflict only arises if ‘because’ is under- stood in the same way in both. But such an understanding is a mistake. The ‘because’ in

M expresses a genuine explanatory relation, in particular that of metaphysical grounding, whereas the ‘because’ in E does not express a genuine explanatory relation but instead is used evidentially. Thus, the conflict is resolved. Furthermore, I have argued that the two

claims are in fact true. Even though correct argumentative interpretations are metaphysi- cally grounded in true content-specifying propositions, they give us evidence for the latter.39

38 On evidential uses of ‘because’ see Schnieder, ‘A Logic for “Because”’, 447. The distinction between epistemic

reasons and grounds has a long philosophical tradition. A particularly clear account can be found in Bernard

Bolzano, Wissenschaftslehre (Sulzbach: Seidel, 1837), §198.

39 Note that unlike M, E entails that there is at least one fiction f, one true contentf-specifying proposition and

one correct argumentative interpretation of f. This entailment stems from the factivity of ‘because’. I take these

consequences to be unproblematic since there are many fictions (e.g. GG) and many true content-specifying

propositions (e.g. that in GG Gatsby is rich) and because boring but correct interpretations are easy to come up with.

5. Conclusion

In this paper I have tried to shed some light on the relationship between truth in fiction

and interpretation. After elucidating the structure of fictional content, I have argued that argumentative interpretations necessarily involve content-specifying propositions and that, contingently, many of them involve content-transcending propositions. Further, I have shown that determining fictional content and interpreting a fiction are genuinely

separate activities. There are, however, interesting connections between interpretation and truth in fiction. On the one hand, correct argumentative interpretations are grounded in fictional content. On the other hand, one of our best ways of finding out about fictional

content is via correct argumentative interpretations.40

Christian Folde

christian.folde@gmx.de Universität Hamburg

40 I would like to thank audiences in Göttingen, Hamburg and Lisbon for their questions and comments on earlier

versions of this paper. I have especially profited from conversations with Robert Schwartzkopff. Most of all, I am

indebted to Benjamin Schnieder for his continuous support and criticism.