Frege, fiction and force

Abstract

Discussion of Frege’s theory of fiction has tended to focus on the problem of empty names, and has consequently missed the truly problematic aspect of the theory, Frege’s commitment to the view that even fictional sentences that contain no empty names fail to refer. That claim prima facie conflicts with his commitment to the cognitive transparency of sense, and the determination of reference by sense. Resolving this tension compels us to recognize that fiction for Frege is a special kind of force, and that words express a sense capable of picking out a referent only in the presence of the appropriate assertoric force. In effect, Frege’s theory of fiction reveals his commitment to an act-centered rather than an expression-centered semantics.

What does Frege’s theory of fiction have to tell us about the question of whether the primary unit of meaning is an act or an expression? How can Frege’s writings on fiction bear on a question that only became live in the philosophy of language half a century after he wrote: do words refer, or do people refer?

Frege, it is often said, was primarily concerned to exclude fiction from the class of linguistic phenomena with which his logicist project was concerned.¹ Truth is the goal of logic (CSB 1969:133/178). Fictional sentences, along with other sentences containing terms which fail to refer are neither true nor false (IL 1969:211/297). Such sentences are therefore outside the concern of the logician (BSLD 1969:214/300).²

Contrary to this supposed disinterest however, Frege does a peculiarly bad job of neglecting the subject of fiction. From 1891 onwards, once the distinction between Sinn and Bedeutung ³ is in hand, he makes mention of it in much of his writing. Sometimes this is only to set it aside, but in On Sinn and Bedeutung, Comments on Sinn and Bedeutung, the Logic, two letters to Russell, the Introduction to Logic, and the Thought Frege offers extended comments on it, at far greater length than is required to simply exclude it.

In spite of this, relatively little systematic attention has been paid by scholars to this topic in Frege’s work.⁴ But Frege’s comments on fiction merit renewed and careful attention. Despite the broad time frame over which they are written, Frege does offer a (broadly) coherent picture of fiction, consistent with his wider philosophy of language. More importantly, the traditional conversation, driven by the assumption that Frege failed to develop a proper theory on the topic, has focused on his view of empty names. In doing so, it has failed to identify the really tricky commitment that Frege’s comments on fiction reveal, to the view that all sentences in fiction lack a truth-value, not just those containing empty names. When we attend to that claim, we are faced with a tension between Frege’s comments on fiction, and two other core commitments frequently attributed to him: to the cognitive transparency of Sinn, and to the view that Sinn determines Bedeutung. Resolving that tension, and appreciating how Frege’s picture of fictional discourse coheres with his philosophy of language points us away from an expression-centered semantics, towards an act-centered semantics, with assertoric force and the intentions of language users playing an important and underappreciated role. This reveals a semantic framework in which the connection between an utterance and its Bedeutung depends on the assertoric intent of a speaker.

The paper proceeds as follows: in section one I describe the problem of fiction for Frege. In section two I review three unsuccessful solutions to this problem. The first claims that all signs in fiction are bedeutungslos, even ordinary words. The second instead tries to dissolve the puzzle by tinkering with Frege’s compositionality thesis. The third explores the possibility that, though fictional sentences express ordinary thoughts, those thoughts fail to determine a Bedeutung. Each of these solutions is incapable of simultaneously satisfying the demands of Frege’s commitment to a transparency thesis, and a determination thesis. In sections three and four I present a novel solution, according to which fictional statements are distinguished by their force, which is in turn determined by speaker intentions. The solution combines elements of these unsuccessful solutions. Though it also restricts the scope of the determination thesis it uniquely offers an underlying principled explanation of that restriction. I then consider how the solution has to be modified to accommodate Frege’s admission of fictional words as well as fictional sentences.

Frege’s chief concern may have been with formal, logical discourse. That does not mean his interest in fiction was superficial or his comments on it incoherent. His frequent references to it draw our attention to the way in which, for Frege, the logical relations between words and referents in natural language depends on the intentions of their users.

1 The problem of fiction

Let us make a first pass at describing Frege’s view of fiction. Frege discusses fictional sentences, fictional names and fictional concept-words. Fictional names and concept-words are bedeutungslos. That is, they have a Sinn, but no Bedeutung, and as such they form a subclass of non-referring terms more generally (SB 1892:28/153, 32/156, CSB 1969:128/173, 133/178 LR 11/13/1904/291). Sentences which contain bedeutungslose terms in turn lack a Bedeutung, that is, a truth value (SB 1892:35/158-9, LR 13/11/1904:291) though they continue to express a ‘thought’ or Gedanke.

Is it possible that a sentence as a whole has only a sense, but no Bedeutung? ...sentences which contain proper names without Bedeutung will be of this kind. The sentence ‘Odysseus was set ashore at Ithaca while sound asleep’ obviously has a sense. But since it is doubtful whether the name ‘Odysseus’, occurring therein, has a Bedeutung, it is also doubtful whether the whole sentence does. (SB 1892:32-33/156-7)

Fiction is not restricted though to sentences that contain bedeutungslose names or concept-words, though one could be forgiven for thinking so given the focus of critical discussion in the area. The most interesting and problematic kinds of fictional sentences will be those that ordinarily have a Bedeutung, except when used in a fictional context. Frege claims that these sentences too express a thought, but lack a Bedeutung. This phenomenon is most explicitly described in On Sinn and Bedeutung when Frege writes that

The truth claim arises in each case from the form of the assertoric sentence, and when the latter lacks its usual force, e.g., in the mouth of an actor upon the stage, even the sentence ‘The thought that 5 is a prime number is true’ contains only a thought, and indeed the same thought as the simple ‘5 is a prime number’ (SB 1892:34/158).⁵

Fiction about historical figures poses a similar problem. These names have a Bedeutung when used assertorically, but can nonetheless feature in fictional sentences. Frege considers as an example Schiller’s Don Carlos, writing that

If [it] were to be regarded as a piece of history, then to a large extent the drama would be false. But a work of fiction is not meant to be taken seriously in this way at all: it is all play. Even the proper names in the drama, though they correspond to names of historical persons, are mock proper names (Logic 1969:141/230).

This suggests that Frege is committed to allowing that individual words in fiction, as well as fictional sentences, fail to have a Bedeutung. This view is additionally implicit in a number of the other comments Frege makes about fiction that both pre- and post-date the Logic. In Comments On Sinn and Bedeutung, Frege writes “Of course in fiction words only have a sense” (1969:128/173) without distinguishing any subset of words, such as non-referring names. We can only conclude then that Frege thinks that this is true just as much of ordinary words that, outside of fiction, have not only a Sinn but a Bedeutung too. That is implicit too in On Sinn and Bedeutung itself, when in footnote F Frege writes of “signs intended to have only sense” in the course of claiming that the words of an actor on the stage have only Sinn (1892:32/156-7). Over ten years later in a letter to Russell Frege continues to talk not just of sentences in fiction lacking a Bedeutung but of signs also doing so:

For it does sometimes happen that a sign has a sense but no Bedeutung, namely in legend and poetry. Thus the sense is independent of whether there is a Bedeutung. (LR13/11/1904 291).

In the passage from the Logic above Frege calls historical names used in fictional sentences mock proper names (Scheineigenname). Similarly, though at some points Frege talks as though the Sinne of fictional sentences are thoughts,⁶ at points in the Logic he calls them ‘mock-thoughts’:

Instead of speaking of ‘fiction’, we could speak of ‘mock thoughts’ [‘Scheingedanke’]. ... Assertions in fiction are not to be taken seriously: they are only mock assertions. Even the thoughts are not to be taken seriously as in the sciences: they are only mock thoughts. (1969:141-2/230).⁷

But what is a mock-thought, and how exactly is it distinct from the thought a sentence expresses in a non-fictional context? We shall return to these questions below, as we try to square this notion with those points at which Frege claims that fictional sentences express the same thought as their assertoric counterparts.⁸

Frege is unwavering in his view that fictional sentences lack a truth value, as he writes in the Introduction to Logic: “Thoughts in myth and fiction do not need to have truth-values” (1969:211/297), echoing his words in the Logic

But the sense of the sentence ‘William Tell shot an apple off his son’s head’ is no more true than is that of the sentence ‘William Tell did not shoot an apple off his son’s head’. I do not say, however, that this sense is false either, but I characterize it as fictitious (1969:141/229-30).⁹

Given Frege’s view that the Bedeutung of a sentence is a truth-value (SB 1892:35/158-9), this is part and parcel of the claim that sentences in fiction lack a Bedeutung.

Interpreting the claims Frege makes about fiction becomes trickier in light of two other important tenets of his philosophy of language. The first of these is the thesis that Sinne determine their Bedeutungen. Signs have associated with them both an object, their Bedeutung and a Sinn, “wherein the mode of presentation [of the Bedeutung] is contained”(SB 1892:26/152). Frege writes that “The regular connection between a sign, its sense and its Bedeutung is of such a kind that to the sign there corresponds a definite sense and to that in turn a definite Bedeutung...” (SB 1892:28/153).

Only two paragraphs after introducing the notion of Sinn, Frege makes explicit that Sinne do not always have a corresponding Bedeutung (SB 1892:28/153). So early and with such regularity is this point made that an understanding of Sinn according to which bedeutungslose words are problematic or aberrant cannot be correct.¹⁰ Nonetheless, some commentators have understood Sinn as constitutively dependent on its Bedeutung in such a way that the notion of a bedeutungslosen Sinnes is incoherent.

This confusion has its roots in the description of Sinne as ‘modes of presentation’, implying a constitutive dependence on the object so presented. This naturally gives rise to the kind of worry Gareth Evans voiced in 1982: “It is really not clear how there can be a mode of presentation associated with some term when there is no object to be presented” (1982, p. 22). If correct, this worry would be a pressing one for Frege’s account of fiction, populated as the fictional realm is by bedeutungslosen terms. But there is another way of understanding Sinne that allows that some can be coherently bedeutungslos, in line with Frege’s claims about them. Beaney (1997) articulates the view that whilst those Sinne possessed of Bedeutungen do indeed “present” them, other senses merely determine the object they would refer to were it to exist. The name “Sherlock Holmes” fails to refer to an individual since Sherlock Holmes is fictional. Nonetheless, it provides a criterion satisfaction of which would qualify the object in question to be the Bedeutung of the term. As Tyler Burge puts it, “the sense is the epistemic basis for determining the referent” (1979, p. 402). The Sinn of a word is just the criterion an object must meet for that object to be the Bedeutung of the word. We shall call the view that Sinn determines Bedeutung in this way the determination thesis.

DETERMINATION: the Bedeutung of a term is the object which satisfies its Sinn. ¹¹

The second commitment constraining any adequate interpretation of Frege’s comments on fiction is that which Sinn attaches to a word must be transparent to the competent language user. Sinn is introduced to explain how true identity statements can be informative, that is, how self-identity and identity claims can have different ‘cognitive value’ (Erkenntniswert) to one another (SB 1892:25/151 50/171). This difference is explained by the different Sinne of the names they contain. Since Sinne explain this difference in cognitive value they are rightly understood as constitutively tied to, and individuated in terms of their cognitive significance. It falls out of this that different Sinne cannot be cognitively indistinguishable from one another, at least by a fully competent language user. This gives rise to what we shall call the transparency thesis.

TRANSPARENCY: a difference in Sinn will correspond to a difference in cognitive value detectable by a competent language user.¹²

We have so far deduced this requirement as an implicit commitment of Frege’s comments on Sinn. It is also made explicit at points, as in On Sinnand Bedeutung, when he writes that “The sense of a proper name is grasped by everybody who is sufficiently familiar with the language” (1892:27/153). This commitment importantly circumscribes what can constitute an adequate resolution of the problem associated with Frege’s comments on fiction. It rules out those solutions which require that competent language users, who understand the terms they are using, nonetheless do not know what Sinn is associated with them.

Transparency is closely related to a third thesis, which we shall call the compositionality thesis. The compositionality thesis is the view that Sinn is compositional, meaning that the thought a sentence expresses is a function of the Sinne of its component words. The claim is deliberately underdescribed at this stage, to reflect the fact that the precise form in which such a thesis can be ascribed to Frege is contested, (as indeed is the claim he endorsed such a thesis at all (Pelletier 2001; Tsai 2009).

COMPOSITIONALITY (SINN): the Gedanke expressed by a sentence is determined by the Sinne of its constituent expressions and the way those constituents are combined.

As we shall see below, certain interpretations of Frege’s view of fiction conflict with a consequence of the thesis, that the same constituent Sinne in the same syntactic arrangement will compose the same thought. These interpretations claim that contextual factors play a greater role in determining the composition of the thought expressed by a sentence than the compositionality thesis allows for. They suggest instead that whether or not the same words, in the same order determine the same thought depends on the conversational context in which the sentence is used. Compositionality is related to transparency in the following way: any violation of compositionality risks violating transparency also. If the same words used in the same order can express different thoughts depending on the context in which they are used, then for transparency to be preserved that difference in context must be detectable by a competent language user.

We are now in a position to see how Frege’s comments on fiction pose an interpretive challenge. So far we have the following broad picture of language. A word expresses a Sinn. Sinn in turn determines a Bedeutung. Not every Sinn has a Bedeutung, however. Fictional names are among the words whose associated Sinn fails to pick out a referent. They are bedeutungslos. Words compose phrases and sentences, whose Sinn is in turn composed by the Sinne of their constituent terms. The Sinn of a sentence is a thought. If each of the constituent words or phrases that make up a sentence has a Bedeutung, then the sentence too has a Bedeutung, which is a truth-value. Sentences containing bedeutungslose terms or phrases fail to have a truth-value.

So far so good: fiction generally concerns non-existent characters and events and so its sentences fail to pick out a Bedeutung. The picture becomes more complicated however when we include in it fictional sentences which do not contain obviously bedeutungslose terms, but may nonetheless be bedeutungslos, as in the example cited above of an actor uttering the sentence ‘The thought that 5 is a prime number is true’ on stage (SB 1892:34/158).

This gives rise to the following puzzle: how can such a sentence fail to have a truth-value? If Sinn determines Bedeutung, by providing satisfaction conditions, then whenever there is an object which meets those conditions, the Sinn in question picks out that object as its Bedeutung. Similarly, if the Sinn of a sentence remains the same whether it is uttered on or off stage, how can the Sinn pick out a truth-value (as Bedeutung) in the latter case but not the former? If the sentence ‘The thought that 5 is a prime number is true’ were uttered off stage it would have a Bedeutung. How is it that the same words, uttered on stage, fail to refer in this way?

We are faced with a kind of dilemma: on one horn we can take the view that fictional thoughts are special, or ‘mock’ in some sense, and that that accounts for their failure to refer. This requires us to modify the strength of the compositionality thesis, allowing that the same arrangement of words can have different senses associated with it on different occasions of use. In doing so we risk contravening the transparency thesis. On the other horn, we can maintain that in a fictional context sentences without bedeutungslose terms express ordinary thoughts, but then have to admit, contrary to the determination thesis, that in some circumstances sentences fail to have a Bedeutung simply in virtue of the context in which they are used, despite the existence of a referent that satisfies the criterion they express.

If a common interpretation of Frege’s account of Sinn and Bedeutung is correct, there is no room for fictional contexts to interrupt the relationship between the Sinne associated with words, the thought associated with a sentence, and the Bedeutung determined by that thought. This is the problem of fiction at a first pass: explaining how fictional sentences composed of ordinary terms come to be bedeutungslos. It has been missed by the traditional conversation on Frege’s treatment of fiction which, driven by Evans (1982) and McDowell (1984) in particular, has focused on empty names. That focus has been motivated by the mistaken assumption that Frege never really thought about fiction, but merely elided it with the problem of non-referring terms. In fact, the really tricky problem of fiction described here fails to arise for empty names. It arises specifically when non-empty names are used in fictional contexts. I move now to consider three possible solutions to it. Appreciating why they fail points us towards a successful resolution of the puzzle.

2 Unsuccessful solutions

2.1 All signs in fiction are bedeutungslos.

One way of resolving the problem seizes the first horn of the dilemma. Accepting that if words used in a fictional context were to compose a thought that thought would determine a Bedeutung and thereby have a truth-value, it offers instead a principled reason to think that such terms do not compose a thought. They fail to do so because they are bedeutungslos. Frege closely connects (sometimes even seems to confuse as one and the same) fictional terms and bedeutungslose terms. Perhaps that is not evidence of confusion but of a tacitly held view that no terms used in fictional contexts have a Bedeutung. This close connection is evident for instance in the following quotation from the Introduction to Logic:

A sentence containing a proper name without Bedeutung is neither true nor false; if it expresses a thought at all, then that thought belongs to fiction. In that case the sentence has no Bedeutung (IL 1969:211/297).

Frege similarly elides the two when he writes in the Grundgesetze that

‘The sum of the Moon and the Moon is one’ is neither true nor false, for in either case the words ‘the sum of the Moon and the Moon’ would have to stand for something, and this was expressly denied by the suggested stipulation. Our sentence would be comparable, say, to the sentence ‘Scylla had six dragon necks’. This sentence is likewise neither true nor false, but fiction, for the proper name ‘Scylla’ designates nothing (Grundgesetze Vol. II Sect. 64 1903:76/266).

Dummett, criticising Frege for citing as examples of “names having sense but no reference personal names used in fiction” claims that they “have in fact only a partial sense, since there is no saying what would warrant identifying actual people as their bearers” (1973, p. 160). The indeterminacy of fictional terms leaves open the possibility that two distinct objects in the actual world might both meet the conditions set by a fictional Sinn, with the result that there would be no fact of the matter which of the two objects is the Bedeutung. Perhaps it is this alleged indeterminacy in fictional Sinn which ensures that there are no Bedeutungen for fictional terms. Frege too at one point describes the problem in terms of the indeterminacy of deciding whether objects fall under fictional concept-words. In Comments on Sinn and Bedeutung he writes that

[i]t must be determinate for every object whether it falls under a concept or not; a concept word which does not meet this requirement on its Bedeutung is bedeutungslos. E.g. the word $\mu \tilde{\omega }\lambda \upsilon$ (Homer, Odyssey X, 305) belongs to this class, although it is true that certain markers are supplied (SB 1892:133–134/178).

This view can point, then, to the indeterminacy of fictional terms as responsible for their failure to determine a Bedeutung. Since they are thereby bedeutungslos it is to be expected that the sentences which contain them in turn fail to have a Bedeutung in the form of a truth-value.

We should reject this solution for two reasons. In the first place it is doubtful that it is any more indeterminate for fictional concept-words than for non-fictional concept-words whether a given object falls under their extension. Perhaps the claim that their extensions are less determinate is motivated by the often-limited usage of these terms, providing us with little information to determinately pin down their referent. What root Homer had in mind as the referent of $\mu \tilde{\omega }\lambda$ $\upsilon$ is contested, for instance, as a result of the brevity of his description of it.¹³ In many cases, however, fictional characters and places are described in sufficient detail to determinately pick out an appropriate object as their Bedeutung, and certainly in as much detail as many characters in factual histories. On closer inspection, even $\mu \tilde{\omega }\lambda \upsilon$ doesn’t succeed as an example of indeterminacy: the crucial characteristic of the root is that it is given to Odysseus by Hermes. If we knew which root satisfied that description, it would undoubtedly be the Bedeutung of the term. If other fictional senses are indeterminate, it is not clear that they are more so than some non-fictional senses whose extension is poorly specified and therefore vague or disjunctive. Indeterminacy of sense is an interesting problem for a Fregean semantics, but it is not a phenomenon unique to fiction, and is not therefore well-placed to provide a solution to a fiction-specific problem.¹⁴

Secondly, even if some kind of special fictional indeterminacy could account for the failure of sentences containing words and names that feature only in fiction to pick out a Bedeutung it is no help at all with the far more puzzling phenomenon of the failure of fictional sentences made up of ordinary words, like the mathematical terms used by the actor in Frege’s example, to express a Sinn capable of picking out a Bedeutung. Why should these terms, used in a fictional context, be any less determinate than when they are used assertorically? Maintaining that they too are bedeutungslos leads back to a version of the second horn of the dilemma, at the level of individual words: we now have to claim that ordinary words, used in fictional contexts, are bedeutungslos. But that leaves us with the puzzle of understanding how they can be so in fictional contexts, when they pick out a Bedeutung when used non-fictionally. What is it about fictional contexts which stops the Sinn of those terms from determining a Bedeutung? This view requires not just that context can determine which object satisfies a particular Sinn ¹⁵ but whether a given object satisfies a Sinn.

One way of defending the view that ordinary words used in fictional contexts are bedeutungslos is to claim that ordinary words used in fictional contexts have a different Sinn to when they are used in non-fictional contexts. Compositionality and determination are unproblematic in this case: these special fictional senses when combined will not constitute a thought, but only a mock thought, and mock thoughts are not the kind of thing that can determine a Bedeutung. On this version of the solution, Sinne in fiction are all bedeutungslos because they are in fact special fictional counterparts of ordinary non-fictional Sinne.

We shall call this view, which multiplies Sinne so that the same word or sign used in different contexts, fictional and factual, has a different Sinn associated with it, the diglossic view. It admits of two interpretations: one I shall call the homonymy interpretation, on which there is a fictional homonym for every word, and an appropriately fictional sense associated with that homonym, and one I shall call the polysemy interpretation, according to which a single word has two senses associated with it, one fictional and one factual.¹⁶ Neither Polysemy nor Homonymy views can be made to work.

The homonymic interpretation of the view has no way of explaining the close connection between the two homonyms, fictional and factual. A fictional term has the fictional ‘version’ of the Sinn associated with its factual homonym. But if the two are really distinct words, it is mysterious what guarantees that correspondence between their associated Sinne.

Adopting the polysemy version of the view on the other hand contravenes the transparency thesis. A competent language user cannot detect whether the Sinn associated with a given word is factual or fictional in the absence of sufficient contextual information, despite grasping the criteria set by the Sinn in question. Suppose you come upon an exchange between two strangers. You may not be able to identify whether it is fictional or not, despite being able to understand the words uttered. According to the polysemy view, you are missing a distinction in Sinn, despite grasping the cognitive significance of the words in question, since you cannot say which of two distinct Sinnen is in play. But that is at odds with the transparency thesis, according to which the fact that they are indistinguishable at the level of the criteria they impose on a Bedeutung indicates that there are not two Sinne present, but only one.

I claimed earlier that a difference in cognitive value is what individuates Sinne from one another. If that’s right, the notion of two Sinnen with the same cognitive value is incoherent. But fictional and factual uses of a term do seem to have the same cognitive value as one another. If grasping the Sinn associated with a word is crucial to understanding it, then it should not be possible to comprehend a word without being able to distinguish which Sinn associated with it is in play. In fact, competent language users are entirely unaware in general of the proposed association of two different Sinnen, factual and fictional, with an individual word.

Whilst the polysemy view violates the letter of transparency, the homonymy view just as surely violates its spirit. For if a competent language user is supposed to be capable of recognizing a difference in Sinn, then she should also be able to recognize when there are multiple distinct words which sound the same, with different Sinne associated with them in turn.

2.2 Sentences in fiction employ a different compositional function

Is there another way of cashing out the notion that fictional sentences express a distinct kind of fictional thought that avoids the extravagant and mysterious multiplication of Sinnen or signs, and the violation of the transparency thesis that leads to? One way of doing so tries to solve the problem by relaxing the compositionality thesis instead.

The compositionality principle for Sinn requires that the thought expressed by a sentence be the result of the Sinne of the words which compose the sentence and their syntactic arrangement. The Sinne of the constituent terms form the input to a compositional function which produces as output the thought expressed by the sentence. This solution to the problem of fiction claims that the fictional context intervenes at the point of composition, by determining a different function that takes those Sinne to a special kind of fictional thought. Perhaps this is what Frege has in mind when he refers to Scheingedanken in the Logic (1969:141-2/230). These fictional, mock thoughts, unlike their non-fictional counterparts, fail to determine a truth-value. On this view, bedeutungslose names are special. Other terms used in fiction still have referents, and as a result have ordinary Sinne. Those Sinne fail to compose a thought, however, because of the different fictional compositional function which is in play. I shall call this the compositional function view.

One cost of the compositional function view is that it conflicts with the compositionality thesis. A consequence of the latter thesis is that the same words in the same order will generate the same thought whatever the context. Since the compositional function view posits contextual variation in the compositional function that takes the Sinne of words to the thought of the sentence they compose, it is incompatible with that consequence of compositionality. Whilst the Sinne of constituent words continue to play a crucial role in determining the thought a sentence expresses, it is not until the sentence is part of an utterance, used within a particular context, that the particular thought those Sinne compose is fixed. The Sinne of individual words are primary only in so far as they serve as the inputs to a compositional function which determines the thought expressed by a sentence as a whole. Context decides the compositional function, which determines the result of those terms’ interaction with one another. In this case ‘context’ should be understood to encompass not just the immediate linguistic context surrounding a term, but the broader context of utterance including the parties’ conversational purposes.

The compositional function view suffers from two further drawbacks. The first of these is that the implication that words used in fictional contexts have ordinary Sinne is hard to square with those points in the Logic at which Frege writes as though there is a difference even in the status of individual terms when used in fiction: “Even the proper names in the drama, though they correspond to names of historical persons, are mock proper names” (1969:142/230).

Secondly, the compositional function view merely suppresses the problem of respecting the transparency thesis that arose for the diglossic view. The problem re-emerges at the point when we allow that there are multiple compositional functions, some of which take us to standard thoughts, and others of which take us to fictional, mock thoughts. In this way, the compositional function view repeats the flaws of the polysemy approach, but at the level of Gedanken rather than Sinne. It allows that the same sentence can express a thought and a fictional thought. For those to be genuinely distinct from one another we should expect there to be a corresponding difference in cognitive significance, detectable by a competent language user. Arguably there is a difference in the significance of fictional and non-fictional sentences, but it is not a difference that can be appropriately located at the level of Gedanken, because the sentences do not differ in their cognitive value. This solution insists they do, as a result of a shift in compositional function.

It looks in fact as though any account which allows that words or sentences have distinct Sinne associated with them, depending on whether they are used in a fictional context or not, will struggle to respect transparency. If fictional Gedanken have the same cognitive value as non-fictional thoughts, that is, they impose the same criterion on their referent, then the difference between them and their non-fictional counterparts will fail to be appropriately detectable. Since the transparency thesis flows directly out of the nature of Sinn, and the role it is introduced to play, such a failure is an unsupportable flaw in any interpretation of Frege’s comments on fiction. But equally, if the determination thesis holds, it is not clear what else there is to distinguish the two kinds of thought: there is nothing more to thoughts than the criteria they provide which determine the referent they pick out.

2.3 Thoughts expressed by sentences in fiction fail to determine a Bedeutung

The difficulty of preserving the transparency of Sinn if either words or sentences express special fictional senses makes grasping the second horn of the dilemma described at the end of section one look attractive. We can allow words and sentences used in fiction to retain their ordinary Sinn, the same Sinn they possess in non-fictional contexts. We need only claim that that Sinn fails to determine a referent. Taking this horn of the dilemma sacrifices our commitment to the determination thesis, the view that sense determines reference. It allows that the Gedanken fictional sentences express are the very same as those expressed by non-fictional sentences, but that in fictional contexts the sense fails to pick out a referent, in the form of a truth-value.

This approach encounters two main obstacles. In the first place, baldly claiming that fictional sentences express thoughts which fail to determine a Bedeutung gives us no clearer illumination of the means by which fictional context interrupts the relationship between thoughts and their truth-values. Without that we have a simple conflict between Frege’s apparent commitment to the determination thesis, and his views on fiction. In abandoning the determination thesis, this approach is at odds with the emphasis Frege puts on the close connection between a thought and a truth-value.¹⁷ As he writes in the Thought, “I call a ‘thought’ something for which the question of truth can arise at all” (1918:60/327). How then can it not arise for thoughts in fiction, according to this view? How can they fail to have a truth-value? We need an account of how fictional contexts have this impact, if we are to avoid flatly contradicting Frege’s commitment to an intimate connection between thought and truth-value.

The second difficulty is as follows. Accepting, without further explanation, that in fiction Sinn fails to determine Bedeutung compels a regressive change in our understanding of what Sinn is. Much Frege commentary, taking its lead from Dummett perhaps,¹⁸ has reacted against a conception of Sinn as a sort of reified meaning. Such a conception thinks of Sinn as a kind of concrete semantic object associated with a word. The alternative interpretation by contrast understands Sinn primarily as an epistemic notion: its primary role is to mediate between a word and its Bedeutung. A Sinn is what a language user grasps when they understand a word. This is what Zais (1993) calls (in the course of criticizing) the ‘satisfaction view’ of Sinn: a given sense determines the object that happens uniquely to satisfy a condition intrinsically associated with it. The satisfaction view is closely tied to the determination thesis: what it is to be a sense is to set conditions that determine a referent.¹⁹ Frege articulates the satisfaction view in the Grundgesetze:

Every such name of a truth-value expresses a sense, a thought. That is, by our stipulations, it is determined under what conditions the name refers to the True. The sense of this name, the thought, is the thought that these conditions are fulfilled (Grundegesetze Vol 1, 1893, Sect. 32 50/221).

Without a reified interpretation of Sinn how are we to understand the notion of thoughts which do or do not have a Bedeutung depending on context? According to the Satisfaction view, there is nothing more to a Gedanke than certain conditions, satisfaction of which identifies a Bedeutung. But the proposed view of fiction allows that sometimes, despite the availability of a Bedeutung that satisfies those conditions, the Gedanke fails to pick it out. Allowing, without further explanation, that the same thought can determine a Bedeutung in some contexts but not in others requires that there be something more to sense than a mere criterion, that Sinne be sufficiently concrete that they retain some kind of content even when they fail to pick out the object that satisfies their criterion. If we wish to retain the Satisfaction view in some form, then this solution is unsatisfactory. We can only avoid the implicit reification of sense by explaining what it is about these contexts which blocks that process of determining a Bedeutung.

In this way the proposed solution shares a drawback with those previously considered. Because it offers no mechanism to explain what blocks Sinn from determining a Bedeutung in fiction, it bluntly conflicts with certain aspects of Frege’s broader view of language. That encourages a return to a solution according to which fiction modifies Sinn. But how is such a solution compatible with Transparency? We again lack a mechanism capable of explaining how fictional contexts could effect such a change in Sinn. In what follows I suggest a principled way of understanding fiction in Frege that includes a mechanism that lets us square the absence of Bedeutung for certain utterances in fictional contexts with Frege’s apparent commitment to Transparency and to some version, at least, of Determination.

3 Fiction and force

I want to argue that it is in fact possible to extract from Frege’s comments on fiction a coherent interpretation of what interrupts the relationship between sentences and their Bedeutungen in fictional contexts. This solution combines elements of each of the solutions surveyed above. What allows it to succeed where they fail is that it offers a deeper explanation of why fictional sentences fail to determine a Bedeutung, that allows it to successfully mesh with Frege’s broader philosophy of language.

We saw earlier how Frege closely associated fictional sentences with sentences containing bedeutungslose terms. Fiction is often discussed in parallel with another kind of language use: non-assertoric sentences. Frege repeatedly asserts that non-assertoric sentences fail to express thoughts, as in the Thought:²⁰

We should not wish to deny sense to a command, but this sense is not such that the question of truth could arise for it. Therefore I shall not call the sense of a command a thought” (Thought 1918:62/329).

Frege goes on to claim that questions involving question words also fail to express a thought (though “propositional” (yes-no) questions can do, a claim he repeats in the Negation 1918:144/347). It is often apparent from the grammatical form of a sentence uttered whether it is interrogative, imperatival or assertoric. These cases do not consequently pose the same sort of challenge to compositionality as fictional sentences do: the (mock)-thought a sentence expresses is still a function of the terms that compose it, and their syntactic arrangement. Imperatival or interrogative form and vocabulary are responsible for blocking the senses of the words that constitute a question or a command from composing a truth-evaluable thought. There is no parallel to this in the case of fictional sentences.

In numerous other cases, however, the presence of questions and commands is not indicated by the grammatical form of the sentence. The exact same words in the same arrangement can be uttered as a question, as an assertion or even as a command (“you must leave at dawn”). In these cases we must rely on prosody and conversational context to identify the form of the utterance in question. This brings us to a difficulty similar to the problem we are examining for fictional sentences: what is it in these cases that prevents the words that make up a question from expressing a thought when, if uttered assertively, they would do so?

One option parallels the ‘compositional function’ view described above for fiction: a different set of compositional functions are in play which take the words not to a thought but to the appropriate equivalent for a question or assertion. Which function is in play is not always made explicit by the arrangement of the words in a sentence, but depends on broader factors including speakers’ intentions. What determines the compositional function is the context, more specifically, whether the words form part of an assertion or not.

From early on in his writings Frege distinguishes the content of an assertion from the force. At the very start of the Begriffsschrift Frege introduces a “judgement stroke” (Urtheillsstrich) to his system of logical notation (1879, Sect. 2, 2/ 53) “which stands to the left of the symbol or complex of symbols which gives the content of the judgment.” When it is omitted “then the judgement will be transformed into a mere complex of ideas, of which the writer does not state whether he recognizes its truth or not” (1879:Sect, 2 2/52). In Function and Concept Frege reiterates the need for a special symbol, corresponding to the act of assertion (“we thus need a special sign in order to be able to assert something” 1891:22/142), a claim he repeats in the Grundegesetze (1891, vol 1, Sect. 5, 9/215).

The distinction between content and force is clearly present in his later writings also. In the Thought Frege writes

An interrogative sentence and an assertoric one contain the same thought; but the assertoric sentence contains something else as well, namely assertion. The interrogative sentence contains something more too, namely a request. Therefore two things must be distinguished in an assertoric sentence: the content, which it has in common with the corresponding propositional question; and assertion. The former is the thought or at least contains the thought. So it is possible to express a thought without laying it down as true (Thought 1918:62/329).

Frege goes on to write about fiction as though it too involved the absence of assertion. His clearest statement about this comes in the passage of the Thought quoted above, which begins:

As stage thunder is only sham thunder and a stage fight only a sham fight, so stage assertion is only sham assertion. It is only action, only fiction. ... In poetry we have the case of thoughts being expressed without being actually put forward as true, in spite of the assertoric form of the sentence.... Therefore the question still arises, even about what is represented in the assertoric sentence-form, whether it really contains an assertion (1918:62/330).

Frege here draws a distinction between two levels of force. On the one hand, there is the sentence-form of an utterance. At this level only sentences which are assertoric in form express a thought; indicatives and certain interrogatives do not. On the other hand, there is the conversational context within which a sentence features. Statements made in fictional contexts are assertoric in form, but this form is misleading: they do not in fact contain assertions.²¹ Similarly in the Logic Frege writes that “[a]ssertions in fiction are not to be taken seriously: they are only mock assertions” (1969:230/141), and again in a letter to Russell from the 28th of December 1902 he claims that “[i]n poetry too there are thoughts, but there are only pseudo-assertions” (256).

The distinction between fiction and non-fiction is for Frege a distinction in force. In fictional contexts, it is as though the content is presented with an alternative fictional operator in place of the ‘judgement stroke’. As a result, though the content remains the same, such sentences lack assertoric force. Unlike interrogative and imperatival utterances, this does not prevent them expressing a thought (LR 28 December 1902:256). Assertoric force is required in order for the thought a sentence expresses to determine a referent.

But what, then, determines the force of a given sentence? Frege writes as though whether or not an utterance constitutes an assertion depends on the conversational intentions either of the speaker or of parties to an exchange more broadly. This is consistent with Frege’s repetition throughout his later work of the view that it is ‘our’ intentions that determine whether or not it is appropriate to talk of the Bedeutung of a term or sentence. This is clear in On Sinn and Bedeutung (emphasis added):

in order to justify speaking of the Bedeutung of a sign, it is enough, at first to point out our intention in speaking or thinking. (We must then add the reservation: provided such a Bedeutung exists) (1892:32/156).

Frege moreover explicitly extends this view from words to sentences in the passage discussing the Odyssey from On Sinn and Bedeutung (emphasis added again):

Yet it is certain, nevertheless, that anyone who seriously took the sentence to be true or false would ascribe to the name ‘Odysseus’ a Bedeutung, not merely a sense; for it is of the Bedeutung of the name that the predicate is affirmed or denied. Whoever does not admit the name has a Bedeutung can neither apply nor withhold the predicate.... One could be satisfied with the sense, if one wanted to go no further than the thought... .... Why is the thought not enough for us? Because, and to the extent that, we are concerned with its truth-value. This is not always the case. In hearing an epic poem, for instance, apart from the euphony of the language we are interested only in the sense of the sentences and the images and feelings thereby aroused.... Hence it is a matter of no concern to us whether the name ‘Odysseus’, for instance, has a Bedeutung, so long as we accept the poem as a work of art” (SB 1892:32/156-7).²²

In The Thought too Frege is consistent in maintaining that sentences lose their assertoric force “when we are not speaking seriously” (1918:63/330). Only in passing does he write as though one may “wander into the realm of fiction without knowing it or meaning to” (Thought 1918:68/335).²³

In the Begriffsschrift it is the presence of a judgement stroke which determines whether content is asserted or not. In conversation it is ‘our’ intentions as parties to the exchange which do so. That in turn determines whether or not the utterance has a truth-value. Assertion is required for a thought to determine a Bedeutung even when there is an available object which meets the conditions expressed by its Sinn.

Before we turn in the next section to the question of how fictional force interrupts the relationship between a sentence and its Bedeutung, we should pause to reject a weaker interpretation of the role of assertoric intent in Frege’s semantics. Why think that our intentions determine whether a sentence has a Bedeutung or not, rather than merely deciding whether we have any interest in that Bedeutung? In that case assertoric intent would only affect our relationship with the Bedeutung, without impacting on the more fundamental connection between words and their referents. Uttering a sentence without assertoric force would amount merely to presenting the content as bedeutungslos: speakers are not concerned with Bedeutung in fiction since they are not asserting the content in question.²⁴ If assertoric force only plays a role in establishing a relationship of interest between a language user and the Bedeutung of their utterance, not in determining whether a sentence picks out a Bedeutung, then the connection between words and their Bedeutung is stable, regardless of force, just as my lack of concern with the destination of a passing train cannot change its route.

Attractive though this more minimal interpretation of Frege’s position may be, it is hard to square with the textual evidence. Take the passage above from On Sinn and Bedeutung, in which Frege writes that “anyone who seriously took the sentence to be true or false would ascribe to the name ‘Odysseus’ a Bedeutung, not merely a sense”, for instance. That is arguably compatible with either reading. But it is accompanied by a footnote in which Frege writes that “It would be desirable to have a special term for signs intended to have only sense” (1892:32/157). This footnote importantly guides our interpretation of the passage. Frege views our lack of assertoric intent as sufficient to effect some kind of change in the sign itself, such that we could appropriately use a distinct term to refer to such signs. That goes beyond a change merely in our relationship with the sign.

Elsewhere too Frege’s comments on assertion support the interpretation that our assertoric intentions are required for the sentence to determine a Bedeutung, not merely with a weaker interpretation according to which assertoric force merely establishes a relationship between the language user and the Bedeutung. Consider that passage of On Sinn and Bedeutung quoted above in which Frege writes that

[t]he truth claim arises in each case from the form of the assertoric sentence, and when the latter lacks its usual force, e.g. in the mouth of an actor upon the stage, even the sentence ‘The thought that 5 is a prime number is true’ contains only a thought, and indeed the same thought as the simple ‘5 is a prime number’. (1892:34/158).

Here again Frege writes not as though we are simply disinterested in the relevant Bedeutung, but as though there is no such Bedeutung in such a context. This appears to have been the position he occupied consistently in his later work too. In the Introduction to Logic Frege writes that

[t]houghts in myth and fiction do not need to have truth-values. A sentence containing a proper name without Bedeutung is neither true nor false; if it expresses a thought at all, then that thought belongs to fiction. In that case the sentence has no Bedeutung (1969:211/297).

This is again hard to square with an interpretation according to which the relationship between sentence and Bedeutung is independent of our assertoric intentions. The claim is that when a thought belongs to fiction, it has no Bedeutung. And whether or not a thought belongs to fiction depends, according to Frege, on the intentions of the relevant language users.²⁵

4 Mock thoughts and fictional words

4.1 Thoughts or mock thoughts?

Suppose fictional sentences are distinguished by the force with which they are uttered. How is it that fictional force results in sentences that fail to determine a Bedeutung? When describing the determination thesis in section one, it seemed that the mere use of certain sentences or words was sufficient to pick out a Bedeutung: Sinn is what determines Bedeutung and words possess a Sinn whenever they are used. Now it looks as though another ingredient is necessary: words must in addition have the appropriate force. But what does fictional force do, or fail to do, such that fictional sentences lack a Bedeutung? Is it that sentences uttered with fictional force fail to express a thought, or that the thought they express fails to pick out a truth-value?

Frege seems to vacillate on this point. In the Logic he emphasises that the connection between a thought and its truth-value is independent of anything we, as language users do:

In order to be true, thoughts... not only do not need to be recognized by us as true: they do not have to have been thought by us at all. ... these thoughts, if true, are not only true independently of our recognizing them to be so, but...they are independent of our thinking as such (1969:144-5/233).

This implies that if fictional sentences lack a truth-value, then they cannot express thoughts at all. And consistent with this, it is in this text that he introduces the term Scheingedanke or ‘mock thought’ to describe thoughts in fiction. Indeed, he specifies prior to the passage above that he is using ‘thought’ to refer only to the sense of an assertoric sentence, that “the logician does not have to bother with mock thoughts” (1969:142/230) and that fictional utterances are excluded from the discussion that follows (discussion which includes the claim above that there is a use-independent connection between thoughts and their truth-values): “When we speak of thoughts in what follows we shall understand thoughts proper, thoughts that are either true or false.” The distinction between ‘thoughts proper’ and other thoughts is again based on the force with which an utterance is made. As he explains more generally, talk of truth and falsity is only appropriate where we have an intention to represent (Logic 1969:140/229). That intention is lacking in the case of fictional sentences (Logic 1969:142/230).

This suggests that at this stage Frege thinks of fictional force as interrupting the connection between a sentence uttered and the thought expressed. In fiction, we don’t even arrive at something capable of picking out a truth-value as its Bedeutung. When a thought proper is expressed, which only happens when an utterance is made with the appropriate assertoric force, then the connection with a truth-value is intimate and unwavering.

But if the lack of assertoric intent prevents a sentence from expressing a thought capable of determining a Bedeutung, what are we to make of those moments when Frege writes as though the thought expressed by a sentence remains the same in both fictional and assertoric contexts? As he puts it in On Sinn and Bedeutung: “The thought remains the same whether ‘Odysseus’ has a Bedeutung or not” (1892:33/157).²⁶

In the account I present here, the thought expressed by a sentence is the same across fictional and non-fictional contexts. The only difference is that in the latter context it fails to determine a referent, because of the lack of assertoric force. The cognitive value of a Sinn is what determines its identity. Hence, fictional and non-fictional thoughts are identical.

How then do we dissolve the tension in Frege’s writing described above? In what sense are Scheingedanken and Gedanken distinct? Scheingedanken and Gedanken are distinct only in so far as the former, unlike the latter, are expressed by an utterance that lacks assertoric force. They differ only in the context in which they are used, and the force of the utterance which expresses them. Fictional and non-fictional thoughts remain identical: they carry the same cognitive information.

Why then does Frege write in the Logic as though they differ from one another? Note that in the Logic Frege’s primary concern is with the notion of truth. And despite the underlying identity of the thought, fictional and nonfictional utterances differ in terms of their capacity to pick out a truth value. This difference at the level of utterance encourages Frege to distinguish between Gedanke and Scheingedanke in the Logic. In On Sinn and Bedeutung however, he is introducing the notion of Sinn to explain a difference in the cognitive significance of certain terms. Regardless of the force with which a term or sentence is uttered, its cognitive significance for a language user remains the same. As we might put it, fictional sentences don’t mean something different to their assertoric counterparts. Fictional and assertoric thoughts are identical to one another. But the assertions in which they are made differ in their relationship with the truth. Fictional thoughts are called mock thoughts in virtue of the force with which the sentence containing them is made.

4.2 Fictional words

This account aims to explain how fictional sentences fail to determine a Bedeutung in the absence of any bedeutungslosen terms. The problem of fiction goes deeper than this, however. The picture we have built up is complicated by the way in which Frege allows that even words in fiction can sometimes fail to determine a Bedeutung despite not being bedeutungslos when used assertively.

This striking feature of Frege’s view has not often been fully appreciated.²⁷ The discussion of Frege’s views on fiction has focused on fictional sentences, and their failure to determine a Bedeutung, but Frege holds that even individual words, words that ordinarily refer, fail to do so in fictional contexts. Accommodating this demands a corresponding adjustment in our solution. I claimed above that Frege thinks that the Sinn of a sentence is capable of determining a Bedeutung only in the presence of the appropriate accompanying force. Now we need to allow that something comparable happens in the case of individual words. But force attaches to sentences. How can we extend this proposal to words?

Since Frege himself is silent on the mechanism by which the absence of assertive intentions ensure that words used in fictional contexts fail to refer, we can only speculate about the different means by which this could happen. Textor (2011) puts forward an account on which the capacity of words to pick out a referent depends on the intention with which they are introduced to the lexicon. Fictional names are incapable of determining a Bedeutung, since they are introduced with the intention that they should not refer. Their force is decided by the intentions of the author who introduced the name (2011, p. 391ff.). Textor’s account is thereby restricted in the first place to proper names, in the second place to cases where the name has never had a semantic referent. He additionally maintains that non-fictional proper names used in fiction retain their normal referent, and that Frege was mistaken to claim otherwise (2011, p. 397), because he assumes that a given sense must consistently determine or fail to determine a referent across different kinds of speech act (2011, p. 396ff). Textor’s account is therefore unable to account for Frege’s claim that names that are not specific to fiction, and words that are not proper names, fail to refer in fictional contexts. Were it available, an account that accommodated that claim, and that generalized to other kinds of utterance, and to other kinds of words, would be preferable.

Contrary to Textor, we should allow that whether a word expresses a Sinn that picks out a referent itself varies with the force of the utterance it features in. How then does the impact of fictional context on the Sinn of individual words connect with the impact of fictional force at the level of utterance? One way of modeling this interaction would be to attribute to words a silent, context-sensitive index, which is saturated depending on the force of the speech-act the word in question features in. Unlike an index to a time or place, this index to the force of the sentence does not affect the cognitive value of the associated Sinn, but it does determine whether or not the Sinn is capable of determining a Bedeutung. Sentences whose constituent terms are saturated with assertoric indices function as though subject to an assertion operator, the conversational equivalent of Frege’s formal assertion sign. A Schein-, or fictional, Sinn arises when a Sinn carries a fictional index, indicating that its capacity to determine a Bedeutung is suspended, though the cognitive significance remains the same in both cases. Including a parameter of this kind allows that the force of an utterance of a sentence is reflected in its constituent terms, whilst remaining a property of illocutionary acts, rather than sub-sentential elements.

The overall result is that the capacity of a sentence used in a fictional context to determine a Bedeutung is suspended, resulting in a Scheingedanke instead of a Gedanke. We can still recognize, though, what Bedeutung an assertoric use of the sentence would produce, were it uttered in an assertoric context. This approach avoids the implausible multiplication of either Sinnen or signs associated with the diglossic view. Instead of two independent Sinne associated with the same word, fictional Sinne are identical to their assertoric counterparts: the criteria they impose on a Bedeutung are the same. The context of use only determines whether those conditions are activated or not.

4.3 Transparency, compositionality, determination

In this way, understanding fiction in Frege as a special kind of force allows for a dissolution of the problem it initially presented which preserves all three desiderata, subject to an important caveat.

Fictional and non-fictional senses are identical. This explains and secures the close relationship between the assertoric and fictional use of words or sentences. That in turn frees up our interpretation of Frege’s views on compositionality to include stronger theses according to which the same words in the same order compose the same thought, with no change in compositional function introduced by a change in context. On this account the same constituent words in the same order produce the same thought in fictional and assertoric sentences. The compositional function in play remains the same in fictional and factual discourse.

Importantly, this account preserves the cognitive accessibility of Sinne and thereby the transparency thesis. Since a switch to a fictional context is effected via the intentions of language users, the context is fixed in a manner which is (normally) transparent to the relevant parties. Since it is ‘our’ seriousness, concerns or interests as language users that determine the assertoric force of a sentence, differences in the resulting Sinn of either words or the sentence as a whole will be transparent to us, except on those rare occasions when words are heard or seen out of context.

Does understanding Frege’s theory of fiction in this way allow us to preserve the determination thesis? Sinn continues to determine Bedeutung in the sense that what Bedeutung a word can refer to is determined by the criteria expressed in its Sinn. But this is subject to an important caveat. Once we appreciate the role of force in Frege’s semantics, we see that the capacity of Sinn to in fact pick out a Bedeutung depends not just on the world containing an object which satisfies the criteria it provides, but on the utterance which expresses the Sinn being made with the requisite force. The speaker must intend it to determine a referent in the actual world. Absent such an intention, used outside the appropriate assertoric speech act, it fails to do so. Frege is thereby committed to an act-centered semantics, rather than an expression-centered semantics. The referent of the words which compose a sentence depends on the intentions of the language user responsible for the speech act in which they feature. This weakens the determination thesis, in so far as Sinn is not exclusively responsible for whether an utterance picks out a Bedeutung or not. But it remains the case that when a term or utterance does refer, then the Bedeutung is the object which satisfies the criteria expressed by its Sinn. ²⁸

This interpretation additionally sheds light on the contested role of context in Frege’s semantics, and its interaction with his commitment to some kind of semantic compositionality.²⁹ Without impinging on Frege’s commitment to compositionality, making sense of his comments on fiction highlights the importance of broad conversational context, including the intentions of speaker and audience, to the determination of Bedeutung.

5 Conclusion

Evans accuses Frege of sliding “from the fact that story-tellers are only pretending to make assertions—only pretending to express thoughts—to the conclusion that they are expressing only pretend-thoughts” (1982, p. 30). The account of fiction outlined here concurs that this is a move Frege makes, if by “pretend-thoughts” we understand not a different kind of thought, but simply a thought that fails to determinate a Bedeutung. Contrary to Evans’ implication, however, I suggest that this is a legitimate move for Frege to make. Frege’s comments on fiction amount to a largely coherent picture of fiction as a special kind of discourse, in which utterances possess a special, fictional force. Only as part of an assertoric speech act do Sinne determine Bedeutung, or thoughts a truth-value. By contrast, in fictional discourse, Sinne have a distinct fictional ‘force’. The parties to the conversation grasp what object they would pick out if used assertorically, whilst appreciating that they lack the requisite conversational intention to in fact do so. The pretense of the storyteller and his audience fixes the discourse as fictional. The move between pretending and pretend thoughts is legitimate if not-pretending (i.e. assertion) is a pre-requisite for a thought to determine a Bedeutung. Frege, I have argued, believes that it is.

Appreciating the importance of speakers’ intentions to the determination of Bedeutung by Sinn brings together many of the things Frege says about fiction into an account consistent with his other commitments. Importantly, since speakers’ intentions determine the status of the discourse, it is transparent to them when the Sinn associated with their words changes, in the sense that it lacks the assertoric force needed to pick out a Bedeutung. Similarly, the change in discourse prevents Sinn from determining a Bedeutung without affecting the compositional function that takes the Sinn of individual words to that expressed by a whole sentence. Finally, we preserve the claim that Sinn determines Bedeutung whilst allowing that whether the potential of a thought or Sinn to pick out a Bedeutung is actualized relies in addition on the intention with which the utterance it features in is made.

This last feature of the account is perhaps the most significant for our broader understanding of Frege’s philosophy of language. It requires that we understand the relationship between Sinn and Bedeutung as dependent on speaker intentions. The connection is not automatic: Sinn only pick out a Bedeutung when they are used with the intention that they should do so. Which Bedeutung Sinn can determine, remains independent of this kind of speaker intention however. In this way, Frege’s semantics is not, as often assumed expression-centered. It is instead act-centered, in the sense that the referent of a term depends not just on the words that comprise the expression in which they feature, but in addition on the intentions of the speaker responsible for a particular speech act.