Erkenn DOI 10.1007/s10670-017-9880-8 ORIGINAL RESEARCH
Non-literal Lies Emanuel Viebahn1
Received: 20 January 2016 / Accepted: 16 January 2017 Ó Springer Science+Business Media Dordrecht 2017
Abstract Many recent definitions of lying are based on the notion of what is said. This paper argues that says-based definitions of lying cannot account for lies involving non-literal speech, such as metaphor, hyperbole, loose use or irony. It proposes that lies should instead be defined in terms of assertion, where what is asserted need not coincide with what is said. And it points to possible implications this outcome might have for the ethics of lying.
1 Introduction What is the difference between lying and mere misleading? At first view, it has to do with what is said. Telling a lie seems to require saying something one believes to be false, while misleading merely requires suggesting it. Many recent definitions of lying are in line with this answer and thus require lies to be said. The first aim of this paper is to show that says-based definitions of lying cannot account for non-literal lies: lies that involve non-literal speech, e.g. metaphor, hyperbole, loose use or irony. In such cases, there is a mismatch between what is said and the lie that is told. Non-literal lies are common, so they cannot be discounted as a special case. Rather, I will argue, they indicate that the difference between lying and misleading is not about what is said. The second aim of the paper is to propose an assertion-based definition of lying that can capture non-literal lies. Such a definition has to be based on an account of assertion that does not require what is asserted to coincide with what is said. Finally, the paper aims to show that switching from a says-based definition of lying to an assertion-based definition might well have implications beyond the debate on how to define lying. In particular, this switch might be of & Emanuel Viebahn
[email protected] 1
Department of Philosophy, Humboldt University of Berlin, 10099 Berlin, Germany
123
E. Viebahn
importance for the question of whether there is a morally relevant difference between lying and misleading. The paper proceeds as follows. In Sect. 2, I will introduce four examples of nonliteral lies, and I will argue that these count as lies even on a narrow understanding of lying that contrasts with mere misleading. Then, in Sect. 3, I will highlight the problems non-literal lies pose for says-based definitions of lying. In Sect. 4, I will propose an assertion-based definition of lying that is compatible with non-literal lies, before considering possible implications of this proposal for the ethics of lying in Sect. 5. Section 6 will offer a brief conclusion.
2 Non-literal Lies Consider the following four cases: Metaphor Ada is a keen gardener but has had an exceptionally bad crop of tomatoes. Ada wants Bill to think that her crop was in fact great, so when she meets Bill and he asks how her crop of tomatoes has been, she utters: (1)
I’ve got tomatoes coming out of my ears.1
Hyperbole Carl desperately wants Daisy to come to his party, which is in full swing. When Carl calls Daisy, she says she’ll come, but only if there is some food. Carl is aware that all the food has been eaten, but nonetheless utters: (2)
There’s tons of food left.2
Loose use Fred is meant to start work at the office at about nine o’clock. Recently, however, he has developed a habit of arriving considerably late, and on several occasions he has not arrived until after ten o’clock. His boss has heard about this and asks Fred’s colleague Emily when he arrived at the office today. Emily knows that Fred did not arrive until about ten o’clock, but she wants to protect him and utters: (3)
Fred arrived at nine o’clock.3
Irony Greta and Henry are about to take a school exam. Greta has diligently prepared for the exam, but is aware that it would be decidedly uncool to admit this. When Henry asks whether she has revised, she rolls her eyes and utters: 1
This example is discussed by Saul (2012: 16), who attributes it to an anonymous referee. See Adler (1997: 444, fn. 27) for a further example of a lie involving metaphor.
2
See Keiser (2016: 469, fn. 17) for a further example of a lie involving hyperbole.
3
See Lasersohn (1999) for further examples of utterances involving loose use. Lasersohn does not provide examples of lies involving loose use, but each of his examples could be a lie, given a suitable background story.
123
Non-literal Lies
(4)
Of course I have revised for the exam.4
Each of the above utterances is non-literal in the following intuitive sense: the speaker says one thing, but intends to communicate something else instead. Ada intends to communicate that she has had a great crop of tomatoes, and not that she literally has tomatoes coming out of her ears. Carl intends to communicate that lots of food is left at the party, but not that there literally is tons of it. Emily intends to communicate that Fred arrived at about nine o’clock, not that he arrived at exactly nine o’clock. And Greta intends to communicate that she did not revise for the exam, not that she did revise for it. Standard philosophical and linguistic accounts of these cases are in accordance with this pre-theoretical analysis. The most popular accounts of metaphor, hyperbole, loose use and irony entail that communication in (1)–(4) is indirect and proceeds via a substitutional implicature.5 If a speaker uses a substitutional implicature, she intends to communicate something that is distinct from (and does not include) what she says.6 Communication can nonetheless be successful because the audience can easily work out what the speaker intends to communicate. The standard Gricean reconstruction of this mechanism goes roughly as follows: the audience identifies what the speaker says and notices that if the speaker were to intend to communicate what she says, she would fail to be cooperative; because the audience assumes that the speaker is in fact cooperative (falsely, if the speaker is lying), it infers that the speaker intends to communicate not what is said, but some other proposition that is suitably related; if all goes well the audience arrives at the proposition the speaker implicates. Despite being non-literal, each of the above utterances is a lie. I take this verdict to be extremely plausible, even if we focus on a narrow sense of lying that contrasts with
4
See Simpson (1992: 630) and Mahon (2015: Sect. 1.5.1) for further examples of lies involving irony.
5
See e.g. Grice (1975: 53–54) on metaphor, hyperbole and irony, Searle (1979) and Camp (2006) on metaphor, Davis (2007) on hyperbole and loose use and Lasersohn (1999) on loose use. Not all theorists agree with these accounts (see e.g. Stern 2000; Bezuidenhout 2001 on metaphor), but the group of dissenters is fairly small. The term substitutional implicature has only entered the debate fairly recently, but helpfully marks a difference between the cases discussed and those involving additive implicatures. If a speaker uses an additive implicature, she intends to communicate what is said as well as something else in addition. See Meibauer (2009: 374) and Dinges (2015: 56) for discussion of the difference between substitutional and additive implicatures. The implicatures in (1)–(4) appear to be cancellable, which suggests that they are conversational implicatures (rather than conventional implicatures); see Grice (1975) for discussion of this latter distinction.
6
Two comments are in order here. Firstly, Grice (1975) holds that speakers in these cases merely make as if to say, but most contemporary theorists disagree with this aspect of Grice’s theory (see e.g. Bach 2012). Secondly, I wish to stay neutral with respect to the question of whether speaker intentions have an influence on what is said. Although I take it to be plausible that speaker intentions do determine what is said on at least some occasion (e.g. if speakers use demonstratives), the following will not rely on this position. A speaker may intend to communicate something distinct from what is said even if what is said is not determined by speaker intentions. I am grateful to an anonymous referee for highlighting this second issue.
123
E. Viebahn
mere misleading.7 To underscore this point, I will compare (1)–(4) with a clear case of lying and with a clear case of mere misleading. As a clear case of a lie, let us consider a situation that is like the first example, but in which Ada utters (5) instead of (1): (5)
I’ve had a great crop of tomatoes.
Given that Ada is aware that she has had a bad crop of tomatoes, (5) is clearly lie. And as a clear case of mere misleading, let us consider the following second variant of the first example, in which Ada’s crop of tomatoes is meagre, but she is nonetheless happy to have any tomatoes because she previously thought that all of her plants had died; otherwise the situation is unchanged. Ada then utters (6): (6)
I am really happy about my crop of tomatoes.
If Ada is known for her regular great crop of tomatoes, then (6) is misleading (because Ada thereby falsely implicates that she has had a great crop of tomatoes), but it is not a lie (because she is really happy about her crop). Now, there are at least two ways in which (1)–(4) are similar to the lie (5) but not to the misleading utterance of (6). Firstly, the speakers of (1)–(4) intend to communicate only things they believe to be false, as is the case with Ada’s utterance of (5). By contrast, when Ada utters (6), she intends to communicate something she believes to be true (as well as something she believes to be false). Secondly, and relatedly, utterances (1)–(4) do not permit responses to accusations of lying that are possible in cases of mere misleading. If Ada utters (6) and Bill later finds out that her crop was meagre, then Bill might accuse Ada of lying. But because Ada’s utterance (6) was merely misleading, and not a lie, she can sincerely defend herself along the lines of (7): (7)
It wasn’t a lie. I didn’t claim that my crop of tomatoes was great. I merely claimed that I was really happy about the crop, and that was indeed the case.
Such a defence may be pedantic, but the point it makes appears to be correct: although Ada intended to communicate something she believed to be false, she did so by saying and intending to communicate something she believed to be true. By contrast, there is no parallel line of defence Ada could sincerely adopt after uttering the lie (5), for she clearly did not intend to communicate something true she could fall back upon. And for the same reason, utterances (1)–(4) do not permit a sincere, pedantic response either: there is nothing true the speakers plausibly intended to communicate (in the respective contexts of utterance) that they could cite in their defence.8 So (1)–(4) exhibit features that are found with clear cases of lying, but not features that are found with clear cases of misleading. This further strengthens the verdict that they are lies. 7
I have informally confirmed this verdict with about 100 people, including philosophers and nonphilosophers.
8
In the fourth example, Greta does arguably say something she believes to be true by uttering (4), but because she rolls her eyes it is obvious that this is not what she intends to communicate.
123
Non-literal Lies
3 Non-literal Lies and Says-based Definitions of Lying If utterances (1)–(4) are lies, then there are many non-literal lies. For many utterances feature metaphor, hyperbole, loose use or irony, and there is no reason to hold that this is any different in the case of lies. Furthermore, utterances (1)–(4) appear to be completely normal instances of lies: the fact that they are non-literal does not make them exceptional in any way. Definitions of lying must therefore be able to account for non-literal lies, but says-based definitions of lying fail to do so, as I will now argue. As a first and representative example of a says-based definition of lying, let us consider the following proposal by Carson: (L1)
A speaker S lies iff (1) S says that P; (2) S believes P to be false; and (3) S intends to warrant the truth of P.9
By warranting the truth of something, ‘one promises or guarantees, either explicitly or implicitly, that what one says is true’ (Carson 2006: 294). Carson holds that this happens with most utterances, even if the warranting is not made explicit: ‘one warrants the truth of one’s statements in the absence of special contexts, special signals, or cues to the contrary’ (ibid.). Which utterances (L1) classifies as lies depends on how the notion of what is said is construed. On constrained conceptions of what is said, what is said by an utterance of a sentence is closely linked to the sentence uttered. Implicatures do not count as part of what is said. Unconstrained conceptions of what is said are much more liberal: they do not require what is said to be closely linked to the sentence uttered and do count implicatures as part of what is said.10 To begin with, let us consider a constrained conception of what is said, which is used in the standard analysis of (1)–(4) (outlined above). This conception entails that what the respective speakers intend to communicate is distinct from what they say. This difference can be made apparent as follows, where the S-sentences indicate what is said (on a constrained conception), while the C-sentences indicate roughly what the speakers intend to communicate:
9
See Carson (2010: 37). I have slightly modified Carson’s definition in order to ease discussion, but the modifications should be harmless in the current context. Carson’s version uses ‘makes a false statement’ in clause (1), where I simply use ‘says’. I have replaced ‘makes a statement’ with ‘says’ because Carson uses ‘saying’ and ‘stating’ interchangeably (as is apparent from his discussion immediately following the definition and as is noted by Saul (2012: 3)). And I have removed the requirement that what is said must be false because most theorists working on definitions of lying disagree with this requirement and leaving it out will make no difference to the discussion in this paper. Saul (2012: 12) proposes a similar definition of lying, but in the end she defends a slightly different definition, which I will discuss below.
10
Below, I will also consider an intermediate notion of what is said that includes substitutional but not additive implicatures. See Saul (2012: Chapters 2 and 3) for a thorough discussion of which conceptions of what is said are suited for says-based definitions of lying. What I am calling constrained conceptions of what is said also include what Saul calls austere conceptions.
123
E. Viebahn
(1S) (1C) (2S) (2C) (3S) (3C) (4S) (4C)
Ada’s got tomatoes coming out of her ears. Ada has had a great crop of tomatoes. There is tons of food left. There is lots of food left. Fred arrived at nine o’clock. Fred arrived at about nine o’clock. Of course Greta has revised for the exam. Of course Greta has not revised for the exam.
Now, the problem for (L1) is that the mismatch between what the speakers of (1)– (4) say and what they intend to communicate also persists between what they say and what they intend to warrant the truth of. For it is clear that if the speakers intend to warrant the truth of anything, then of what they intend to communicate, and thus of (1C)–(4C). This gap between what is said and what is intended to be warranted is explicitly signalled by the eye-rolling for utterance (4). And for utterances (1)–(3) it is equally obvious, given that the speakers purport to be cooperative while the audiences can be almost certain (even at the time of utterance) that the speakers do not believe what they say. When Ada meets Bill and utters (1), Bill can see (we can assume) that there are no twigs of tomatoes coming out of Ada’s ears; in this situation, Ada plausibly intends to warrant the truth of (1C), and not of (1S). Likewise, when Carl utters (2), Daisy is aware that it is extremely unlikely that Carl believes that there really is tons of food left, so Carl can only purport to be cooperative if he intends to warrant the truth of what is given by (2C). And it is equally unlikely that Emily believes that Fred arrived at exactly nine o’clock when she utters (3), so she arguably intends to warrant the truth of what is given by (3C). So if a constrained conception of what is said is in play, the speakers of (1)–(4) say one thing while they intend to warrant the truth of something else. But in that case, (L1) classifies none of these utterances as lies, which is clearly the wrong result. Could (L1) instead be combined with an unconstrained conception of what is said? On such a conception, there is no mismatch between what the speakers of (1)– (4) say and what they intend to warrant the truth of. Implicatures count as what is said, so the speakers say what is given by (1C)–(4C), and this is plausibly also what they intend to warrant the truth of (if they intend to warrant the truth of anything). So if (L1) were to be combined with an unconstrained conception of what is said, it would correctly classify (1)–(4) as lies. However, there are strong reasons for defenders of (L1) not to adopt an unconstrained conception of what is said. On the one hand, an unconstrained conception is incompatible with the standard theory of implicatures (as outlined above), according to which the audience uses what the speaker has said in order to figure out what she intends to communicate. If implicatures belong to what is said, this explanation of communicative success no longer works, for then the audience would have to use what is said to figure out what is said. It is thus unsurprising that only very few theorists accept an unconstrained conception of what is said.11 11 Cappelen and Lepore (2005) offer one of the few explicit endorsements of an unconstrained conception of what is said. One might also note that the unconstrained conception of what is said goes
123
Non-literal Lies
On the other hand, such a conception creates new problems for says-based definitions of lying, as Saul (2012: 28–30) convincingly argues: with an unconstrained conception of what is said, such definitions count many clear cases of mere misleading as lies. For example, (L1) then counts Ada’s utterance of (6) (‘I am really happy about my crop of tomatoes’) as a lie: Ada implicates and thus says something she believes to be false (namely that she has had a great crop of tomatoes). Because she is not joking, giving special signals or in an otherwise special context, she also intends to warrant the truth of what she says.12 So, given an unconstrained conception of what is said, (L1) classifies (6) as a lie although it is merely misleading. This shows that the problems non-literal lies pose for (L1) cannot be avoided by switching to an unconstrained conception of what is said. What if one chooses a middle way between constrained and unconstrained conceptions of what is said: a conception on which substitutional implicatures belong to what is said, but additive implicatures (where the speaker intends to communicate what is said as well as something else in addition) do not?13 Combining (L1) with such an intermediate conception of what is said would lead to the right verdicts in the cases discussed (as should be easy to see in the light of the previous discussion). But this move is nonetheless unconvincing. Firstly, the intermediate conception of what is said is also incompatible with the standard theory of implicatures, even if the problem in this case concerns only substitutional implicatures. Secondly, it seems ad hoc to draw the line between additive and substitutional implicatures. Both kinds of implicatures are standardly taken to involve very similar mechanisms, and they seem to have much more in common with each other than either of them seems to have with what would count as what is said on a constrained conception. These problems explain why an intermediate conception has (to my knowledge) not been defended in the literature. Bezuidenhout (2001) argues for a position that comes fairly close: on her view, metaphors belong to what is said, while propositions conveyed through irony do not.14 Bezuidenhout provides an intricate argument for including metaphors in what is said, which I cannot go into here. But, importantly, she also argues that metaphor and irony must not be assimilated in this respect, as is evident in the following passage: Metaphorical interpretations have a kind of directness that is not shared by other pragmatically derived interpretations, such as cases of irony and indirect speech acts. (2001: 164) So even if Bezuidenhout is right about including metaphors in what is said, her conception of what is said will not help defenders of (L1) to capture ironical lies. And Bezuidenhout’s observation concerning the difference between metaphor and Footnote 11 continued against the pre-theoretical understanding of this notion, according to which there can be a difference between what speakers say and what they intend to communicate. 12 Recall that, according to Carson, these are the only circumstances under which the speakers do not warrant the truth of what they say. 13
Thanks to an anonymous referee for pointing out this possibility.
14
Stern (2000) defends a similar view.
123
E. Viebahn
irony provides a further reason to doubt the plausibility of drawing the line between additive and substitutional implicatures. An intermediate conception of what is said therefore does not offer a solution for (L1) either. I hope to have made clear that if (L1) is based on a plausible conception of what is said, it cannot capture non-literal lies. But (L1) is not the only says-based definition of lying, and one might think that other says-based definitions fare better when it comes to non-literal lies. For example, one might think that (L1) runs into trouble because it entails that lying requires warranting. Let us therefore consider the following says-based definition proposed by Stokke, which does not make use of the notion of warranting: (L2)
A speaker S lies iff (1) S says that P; (2) S believes P to be false; and (3) S proposes that P become common ground.15
(L2) is based on Stokke’s view of assertion, according to which ‘to assert that p is to say that p and thereby propose that p become common ground’ (2013: 47). This view of assertion, in turn, takes its cue from Stalnaker’s (1999, 2002) idea that the essential effect of an assertion is to update the common ground of the conversation.16 The common ground of a conversation must thus be construed as follows: It is common ground that p in a group if all members accept (for the purpose of the conversation) that p, and all believe that all accept that p, and all believe that all believe that all accept that p, etc. (Stalnaker 2002: 716) Although (L2) does not involve the notion of warranting, it faces exactly the same problems as (L1). Given Stalnaker’s understanding of common ground and a constrained conception of what is said, the speakers of (1)–(4) do not propose to update the common ground with what they say, but rather with the proposition they intend to communicate. One way to bring this out is to consider what could be felicitously presupposed in the respective conversations once (1)–(4) have been uttered.17 Quite clearly, the propositions given by (1S)–(4S) could not be felicitously presupposed in the ensuing conversations (even if the respective audiences take the speakers to be sincere), while those given by (1C)–(4C) (or very similar ones) could be felicitously presupposed. So on a constrained notion of what is said, (L2) fails to count (1)–(4) as lies.18 On an unconstrained conception of what 15
See Stokke (2013: 49). This definition slightly and harmlessly differs from the definition Stokke offers: Stokke’s definition mentions an addressee (‘S lies to X’) that I have omitted for reasons of simplicity. 16 Stalnaker (1999: 87) discusses the view of assertion Stokke accepts, but emphasizes that he does not ‘propose [it] as a definition of assertion, but only as a claim about one effect which assertions have, and are intended to have’. Stokke (2013: 46) endorses the definition, but notes that his aim is not to provide a general account of assertion, but only to ‘capture the aspects of assertion that are relevant for defining lying’. 17
Stokke (2014: 508) mentions the possibility of felicitous presupposition as a test for whether a piece of information has been made common ground. 18
Stokke (2013: 58) distinguishes between official and unofficial common ground: whereas the former includes permanently accepted information, the latter includes information that is accepted only temporarily, e.g. for the use of an argument. But this distinction is of no help in dealing with the problem
123
Non-literal Lies
is said, however, (L2) wrongly counts (6) and other cases of mere misleading as a lie, so adopting such a conception of what is said does not offer a way out either.19 And unconstrained and intermediate conceptions of what is said are problematic for the reasons given above. The foregoing suggests that the problems the definitions face with non-literal lies are not caused by their respective third clauses, but rather by the requirement that lies have to be said. Saul (2012: 12–20) is aware of this, and thus proposes a saysbased definition that does not apply to non-literal speech: (L3)
If a speaker S is not the victim of linguistic error/malapropism or using metaphor, hyperbole or irony or is speaking loosely, then S lies iff (1) S says that P; (2) S believes P to be false; and (3) S intends to warrant the truth of P.20
This definition does not count (1)–(4) as lies because it delivers no verdict for them. In this way, (L3) avoids the problem of misclassifying non-literal lies. But it is nonetheless unsatisfactory as a general definition of lying. For one thing, excluding non-literal speech greatly diminishes the scope of the definition. Much of our everyday speech is non-literal, so there are many utterances to which (L3) does not apply. Saul holds that even such a restricted definition can be helpful for some purposes, e.g. for her project of investigating whether there is a moral difference between lying and misleading. But if a definition has to be restricted in order to avoid problem cases, that might indicate that the overall approach of the definition is misguided, even as a starting point for the ethics of lying. In particular, the definition might obscure features of lying that could help to explain a moral difference between lying and misleading, as I will argue in the next section. The fact that (L3) has to be restricted therefore should count heavily against the definition. Secondly, the way in which (L3) is restricted is ad hoc. There is no reason to exclude non-literal lies—apart from the fact that they cause problems for says-based definitions of lying. Saul disagrees with this point and argues that we only count utterances such as (1)–(4) as lies if ‘lying’ is understood broadly, so that it includes all deceptive utterances. She holds that our verdict changes ‘when we focus on lying as contrasted with misleading’ (2012: 18). But we have seen that (1)–(4) are clear cases of lying even if we focus on a narrow sense of lying that contrasts with mere misleading, so Saul’s argument is unconvincing.
Footnote 18 continued of non-literal lies, as Stokke requires speakers to propose to add what is said to the official common ground, and the speakers of (1)–(4) clearly do not propose to do this. 19
This is the reason why Stokke (2013: 50) opts for a constrained conception of what is said.
20
Again, I have very slightly and harmlessly modified the wording of Saul’s (2012: 18) definition to ease discussion. Saul does not discuss cases of loose use and thus her version does not mention it. But given her treatment of other forms of non-literal speech it is consistent to include loose use in the antecedent of the definition.
123
E. Viebahn
4 Capturing Non-literal Lies Once non-literal lies are taken into account, says-based definitions of lying turn out to be unsatisfactory. If lies cannot be defined in terms of what is said, which kind of definition of lying should instead be adopted in view of non-literal lies? I want to suggest the following straightforward approach: definitions of lying should be based on the notion of assertion, where what is asserted need not coincide with what is said. An assertion-based definition of lying might be phrased as follows: (L4)
A speaker S lies iff (1) S asserts that P; and (2) S believes P to be false.
It is not uncommon to characterise lies in terms of assertion. But theorists following this path seldom make explicit that asserting P does not require saying P, and in some cases they even claim that asserting P does require saying P.21 So what is novel about my suggestion is that it decouples what is asserted from what is said. Is it plausible to hold that speakers can assert something without saying it? There is an abundance of non-literal everyday utterances that suggest a positive answer to this question. Utterances (1)–(4), for example, seem to be perfectly good examples of assertions. Given a constrained conception of what is said, it is implausible that the speakers of (1)–(4) assert what they say, just as it is implausible that they warrant the truth of what they say or that they propose that it become common ground. So the speakers of (1)–(4) assert something other than what they say—their assertions are non-literal. In line with this observation, theorists who discuss the possibility of non-literal assertions usually accept that asserting something does not require saying it.22 Which accounts of assertion could (L4) be based on? A full answer to this question will have to wait for another occasion, but I want to note a few restrictions on such accounts and then vote in favour of one promising candidate. Of course, suitable accounts of assertion must not be too narrow in requiring what is asserted to coincide with what is said.23 But they must not be too permissive either: in particular, they must not count additive implicatures as asserted. This seems to rule out expressive accounts of assertion, such as that put forward by Bach and Harnish (1979: 42). Bach and Harnish define assertion as follows: (A)
In uttering e, a speaker S asserts that P if S expresses (i) the belief that P, and (ii) the intention that the hearer H believe that P.
This account allows for insincere assertions, as Bach and Harnish (1979: 15) hold that to express an attitude is to make an utterance with the intention that the hearer ‘take S’s utterance as reason to think S has that attitude’. And the account also 21
Amongst many others, Bok (1978), Faulkner (2007) and Sorensen (2007) adopt definitions similar to (L4) without mentioning that assertions need not be said. Chisholm and Feehan (1977), Fallis (2009) and Stokke (2013) defend assertion-based definitions of lying and explicitly require assertions to be said.
22
See e.g. Bach and Harnish (1979), Bergmann (1982), Williams (2002: 98–99), MacFarlane (2011) and Camp (2012).
23
This rules out the aforementioned accounts of Chisholm and Feehan (1977), Fallis (2009) and Stokke (2013), as well as that of Dummett (1981: 300).
123
Non-literal Lies
permits non-literal assertions, as it seems possible that a speaker expresses a belief that P and an intention that the hearer believe that P even if metaphor, hyperbole, loose use or irony is in play. But, as MacFarlane (2011: 81) notes, Bach and Harnish have to count at least some additive implicatures as asserted. For example, when Ada additively implicates that she has had a great crop of tomatoes by uttering (6) (‘I am really happy about my crop of tomatoes’), both clauses of (A) are fulfilled: Ada gives Bill a reason to think that she believes that she has had a great crop of tomatoes, and she wants Bill to come to believe that she has had a great crop of tomatoes. So the definition counts the additive implicature as asserted. But then combining (L4) with such an expressive account of assertion would lead to the false verdict that Ada’s utterance of (6) is a lie. Suitable accounts of assertions must thus be neither too narrow, nor too permissive. This limits the available options, but it leaves the following account that seems natural once we reconsider how liars and misleaders can respond to allegations of lying. In Sect. 2, I noted that misleaders can sincerely deny to have claimed something they believe to be false, while liars (whether they were speaking literally or not) cannot offer such a sincere denial. This suggests that liars commit themselves to something they believe to be false, while misleaders avoid such commitment. And once commitment comes into play, commitment-based accounts of assertion, e.g. those defended by Searle (1979), Brandom (1983, 1994) and MacFarlane (2005), seem to be very natural allies for (L4). According to such accounts, a speaker S uttering e asserts P only if S commits herself to P by uttering e. Commitment-based accounts of assertion do not require what is asserted to be said, as speakers can clearly commit themselves to something while using nonliteral speech. And (it seems) they do not classify additive implicatures as assertions, as speakers use such implicatures precisely to avoid commitment. If (L4) is combined with a commitment-based account of assertion, it leads to the right verdicts in all of the cases discussed above. Utterances (1)–(4) are counted as lies: the speakers commit themselves to the propositions they intend to communicate (through the use of substitutional implicatures), and they believe these propositions to be false. Ada’s utterance (5) is also classified as a lie, because Ada clearly commits herself to something she believes to be false. But (6) is not classified as a lie. Ada merely suggests something she believes to be false, and does not assert it: she asserts something she believes to be true and uses an additive implicature. She thereby avoids committing herself to the proposition she believes to be false, and as a result does not lie.24 Commitment-based accounts of assertion therefore seem to be a good fit for (L4). Of course there is further work left to do in making the definition fully precise. For 24 This approach can also capture non-literal misleadings: a speaker non-literally misleads if she nonliterally asserts (and thus commits herself to) something she believes to be true, and thereby implicates something she believes to be false (without committing herself to it). As an example of such non-literal misleading we can return to the case in which Ada is really happy about her meagre crop of tomatoes because she previously thought that all of her plants had died. Ada non-literally misleads if she uses hyperbole and utters ‘I’m ecstatic about my crop of tomatoes’, in order to thereby non-literally assert that she is really happy about her crop of tomatoes. I am grateful to an anonymous referee for drawing attention to this matter.
123
E. Viebahn
example, one would have to spell out in detail the relevant notion of commitment. But, I think, even at this early stage (L4) shows more promise in capturing nonliteral lies than says-based definitions do.25
5 Possible Implications for the Ethics of Lying Before wrapping up, I want to point out possible implications the move to an assertion-based definition might have for the ethics of lying. These implications come into view if we take a step back to consider why theorists are interested in definitions of lying in the first place. One main application of definitions of lying is in the investigation of a widely perceived moral difference between lying and mere misleading. Most people would say that it is generally (even if not always) better to mislead than it is to lie. Saul (2012: 74) captures this sentiment as follows: (M–D)
Except in certain special cases: holding all else fixed, lying is morally worse than merely deliberately attempting to mislead; and successful lying is morally worse than merely deliberately misleading.
Saul (2012: Chapter 4) is one of only a few theorists who hold that (M–D) is in fact false and that most people are thus mistaken about the moral status of lying and mere misleading. In arguing against (M–D), Saul draws on her says-based definition of lying. Her general line of argument is encapsulated in the following passage: [M]orally preferring misleading to lying is really rather puzzling. In both cases, a speaker deliberately attempts to induce a false belief in their audience. Why on earth should it matter whether they do this by saying something false or merely conveying it by some other means? (2012: ix; italics unchanged) This line of argument rests on the assumption that the only difference between lying and misleading concerns what is said. That does seem to follow if one adopts a saysbased definition of lying. For example, (L1) entails that a speaker lies iff she says and intends to warrant the truth of something she believes to be false; and a parallel definition of misleading entails that a speaker misleads iff she intends to warrant the truth of something she believes to be false, but does so by saying something she believes to be true. But given other definitions of lying and of misleading, it is far from clear that the difference between the two kinds of utterances boils down to what is said. So Saul’s adoption of a says-based definition of lying plays an important role in her argument against (M–D). If lying is instead defined in terms of asserting, and if asserting is understood in terms of commitment, Saul’s argument seems far less convincing. Given such an approach, it is plausible to say that the main difference between lying and mere misleading concerns what the speaker commits herself to: in both cases she intends 25
The account would also have to be shown to be compatible with so-called bald-faced lies: lies uttered without an intention to deceive. See Carson (2006) and Sorensen (2007) for examples of bald-faced lies and Leland (2015) and Keiser (2016) for arguments that these cases are not a problem for assertion-based definitions of lying.
123
Non-literal Lies
to communicate something she believes to be false, but she only commits herself to it if she lies. This difference, it seems, might explain why lying is morally worse than mere misleading. If asserting generally incurs a commitment that suggesting (through an additive implicature) does not, hearers will usually rely to a greater extent on what is asserted than on what is suggested.26 And, at least at first view, it seems quite plausible that it is morally worse to let someone wrongly rely on something to a greater extent than it is to do so to a lesser extent. This impression may be backed up by the following analogous difference between insincerely promising to do something and insincerely presenting the prospect of doing something: if I do not intend to mow the lawn, it is arguably worse to promise to mow the lawn than it is to merely present the prospect of mowing the lawn. These considerations would have to be spelled out in much more detail. But if they are on the right track, assertion-based definitions of lying are in a better position than says-based definitions to explain potential moral differences between lying and misleading. That potential outcome not only shows why it is important to investigate the theoretical differences between lying and misleading. It might also provide a reason to favour assertion-based definitions over says-based ones, if the former (but not the latter) can indeed avoid the need to adopt a moral error-theory, according to which most people have false beliefs about the ethics of lying and misleading. In that case, non-literal lies would not be the only reason to define lies in terms of assertion.27
6 Conclusion Non-literal lies, I have argued, show that the difference between lying and misleading is not about what is said. Nor is it about whether or not an implicature is in play. It concerns what the speaker asserts: if she asserts something she believes to be false, she lies; if she suggests (but does not assert) something she believes to be false, she merely misleads. Such a definition of lying can adequately capture the literal and non-literal lies discussed—provided that it is based on an account of assertion that does not require what is asserted to coincide with what is said. And adopting such a definition might make a difference for the ethics of lying. There are thus good reasons to critically re-examine the current trend towards says-based definitions of lying.28
26 I do not want to claim that hearers always rely to a greater extent on what is asserted than on what is suggested. If a hearer knows a speaker very well, the hearer might rely on at least some of the speaker’s suggestions to an equal extent as her assertions. 27 That is not to say that an assertion- and commitment-based definition of lying straightforwardly leads to (M–D) or a similar principle: it is possible to hold that there is a difference in commitment between lying and mere misleading, while also holding that this difference is not morally relevant. 28 ˚ kerman, Patrick Butlin, Alexander For helpful discussion and comments I am indebted to Jonas A Dinges, Yuuki Ohta, Wulf Rehder, Felix Timmermann, Mai Viebahn, Pascale Willemsen, Derya Yu¨ru¨yen, Julia Zakkou, two anonymous reviewers for this journal and the participants of several seminars and research colloquia at the Humboldt University of Berlin.
123
E. Viebahn
References Adler, J. (1997). Lying, deceiving, or falsely implicating. Journal of Philosophy, 94, 435–452. Bach, K. (2012). Saying, meaning, and implicating. In K. Allan & K. Jaszczolt (Eds.), Cambridge handbook of pragmatics (pp. 47–68). Cambridge: Cambridge University Press. Bach, K., & Harnish, R. (1979). Linguistic communication and speech acts. Cambridge, MA: MIT Press. Bergmann, M. (1982). Metaphorical assertions. The Philosophical Review, 41, 229–245. Bezuidenhout, A. (2001). Metaphor and what is said: A defense of a direct expression view of metaphor. Midwest Studies in Philosophy, 25, 156–186. Bok, S. (1978). Lying: Moral choice in public and private life. New York: Random House. Brandom, R. (1983). Asserting. Nouˆs, 17, 637–650. Brandom, R. (1994). Making it explicit. Cambridge, MA: Harvard University Press. Camp, E. (2006). Contextualism, metaphor, and what is said. Mind and Language, 21, 280–309. Camp, E. (2012). Sarcasm, pretense, and the semantics/pragmatics distinction. Nous, 46, 587–634. Cappelen, H., & Lepore, E. (2005). Insensitive semantics. Oxford: Blackwell. Carson, T. (2006). The definition of lying. Nous, 40, 284–306. Carson, T. (2010). Lying and deception: Theory and practice. Oxford: Oxford University Press. Chisholm, R. M., & Feehan, T. D. (1977). The intent to deceive. The Journal of Philosophy, 74, 143–159. Davis, W. (2007). Knowledge claims and context: Loose use. Philosophical Studies, 132, 395–438. Dinges, A. (2015). Innocent implicatures. Journal of Pragmatics, 87, 54–63. Dummett, M. (1981). Frege: Philosophy of language. Cambridge, MA: Harvard University Press. Fallis, D. (2009). What is lying? The Journal of Philosophy, 106, 29–56. Faulkner, P. (2007). What is wrong with lying? Philosophy and Phenomenological Research, 75, 524–547. Grice, H. P. (1975). Logic and conversation. Syntax and Semantics, 3, 41–58. Keiser, J. (2016). Bald-faced lies: How to make a move in a language game without making a move in a conversation. Philosophical Studies, 173, 461–477. Lasersohn, P. (1999). Pragmatic halos. Language, 75, 522–551. Leland, P. (2015). Rational responsibility and the assertoric character of bald-faced lies. Analysis, 75, 550–554. MacFarlane, J. (2005). Making sense of relative truth. Proceedings of the Aristotelian Society, 105, 321–339. MacFarlane, J. (2011). What is assertion? In J. Brown & H. Cappelen (Eds.), Assertion (pp. 79–98). Oxford: Oxford University Press. Mahon, J. (2015). The definition of lying and deception. In E. N. Zalta (Ed.), The stanford encyclopedia of philosophy (Fall 2015 edn). http://plato.stanford.edu/archives/fall2015/entries/lying-definition/[. Meibauer, J. (2009). Implicature. In J. Mey (Ed.), Concise encyclopedia of pragmatics (pp. 365–378). Amsterdam: Elsevier. Saul, J. M. (2012). Lying, misleading, and what is said. Oxford: Oxford University Press. Searle, J. (1979). Metaphor. Expression and meaning (pp. 76–116). Cambridge: Cambridge University Press. Simpson, D. (1992). Lying, liars and language. Philosophy and Phenomenological Research, 52, 623–639. Sorensen, R. (2007). Bald-faced lies! lying without the intent to deceive. Pacific Philosophical Quarterly, 88, 251–264. Stalnaker, R. (1999). Context and content. Oxford: Oxford University Press. Stalnaker, R. (2002). Common ground. Linguistics and Philosophy, 25, 701–721. Stern, J. (2000). Metaphor in context. Cambridge, MA: MIT Press. Stokke, A. (2013). Lying and asserting. Journal of Philosophy, 110, 33–60. Stokke, A. (2014). Insincerity. Nous, 48, 496–520. Williams, B. (2002). Truth and truthfulness: An essay in genealogy. Princeton: Princeton University Press.
123