2. Lying with assertions

Consider the following case of lying (Stokke Reference Stokke2016, 85):

(1) is considered to be a lie, while (2) is merely misleading. One of the main criteria of a proper definition of lying is its ability to distinguish lying from misleading statements. As an example, take Stokke’s (Reference Stokke2018) definition of lying, which is based on a Stalnakerian (Reference Stalnaker, Munitz and Unger1974, Reference Stalnaker and Cole1978) theory of assertion:

Stokke’s definition satisfies both the aforementioned assumptions. Conditions (L1) and (L2) specify what it means to make an assertion and (L3) provides an insincerity condition. Thus, Stokke argues that lies are just insincere assertions. All examples of lies that Stokke provides are informative and, more importantly, INFORMATIVENESS follows from his definition. By his definition, (1) is a lie because Dennis says that he will not go to Paul’s party and does not believe it, while (2) is not a lie because Dennis only implies that he will not go. Thus, Stokke’s definition delivers the correct verdict.

Since INFORMATIVENESS is a crucial assumption behind Stokke’s definition, let me expand on the idea that assertions are essentially informative speech acts.Footnote ⁶ First of all, consider that many accounts of assertion propose a similar condition for a proper assertion, i.e., a proper assertion that p is such that p is not already common knowledge (for Searle [1969, 66], this is one of the “preparatory conditions” for asserting; for Stalnaker [1978, 88–89 in 1999], this is the first “principle” about assertion; and for Farkas [2022, 326], this is one of the “default contextual assumptions characterising canonical assertions”). These theories maintain that it is infelicitous to assert a proposition that already belongs to the common ground.

Here are three further observations that corroborate the informativeness of assertions. Firstly, as mentioned in the introduction, there are unique conversational patterns that point to the informativeness of assertions, i.e., one can be criticised for asserting something that is commonly known (García-Carpintero Reference García-Carpintero and Bianchi2004). For instance, if after saying (1), Dennis would come to Rebecca and just repeat (1), she could rightly criticise him by saying “I already know that,” or “I know. You already told me that!” This shows that there is something improper and defective in uninformative assertions.

Secondly, there is an intriguing analogy between assertions and inquiries, i.e., inquiries can be seen as a reverse of assertions. Just as assertions are a default way of using a declarative mood, inquiries are a default way of using an interrogative mood. Further, just as assertions are essentially informative speech acts, inquiries are information-seeking speech acts—when asking a question, we expect to be informed. Finally, just as it is improper to assert information that is already commonly known, it is also improper to inquire when one already knows the answer. In short, while assertions are non-inquisitive and informative, inquiries are non-informative and inquisitive (Moyer and Syrett Reference Moyer and Syrett2021; Farkas Reference Farkas2022).Footnote ⁷

Finally, consider the relationship between assertions and uninformative speech acts. It is improper to reassert the same content; however, such content can be reintroduced by means of uninformative speech acts, like reminding, or other types of content, like presuppositions.Footnote ⁸ Thus, it is improper for Dennis to reassert (1), but he can state the same content by reminding Rebecca that he will not come or by presupposing it. I will discuss such cases in detail in the subsequent sections.Footnote ⁹

There are certain widely recognised criteria or tests for lying. Following them allows us to separate lying from misleading without relying on a certain definition of lying. For the present purposes, I will be using two. The first one is the so-called “deniability criterion.”Footnote ¹⁰ It states that lies cannot be sincerely and consistently denied, only misleading statements can. In other words, whenever one does not lie but is merely misleading, one retains a plausible way to deny the accusation of lying. Thus, after saying (2), Dennis can respond something like this (Viebahn Reference Viebahn2020, 733):

Viebahn acknowledges that (3) may seem pedantic, but it is consistent with Dennis’s earlier statement. By contrast, Dennis is lying in (1) because he cannot consistently deny the content of (1). Consider that saying (4) does not work:

The second criterion is based on the observation of how assertions and lies are reported (see Holton Reference Holton2019). It is not strictly speaking a test of lying (like the deniability criterion), but it helps make our intuitions about lying clearer. A proper way for reporting assertions is with a that-clause, but this does not work for lies. Consider the following:

(5) sounds natural and correctly represents the state of affairs. (6), however, sounds bad. The natural way of reporting lies is with about-clauses:

(7) correctly reflects what is going on in this case: by saying (1), Dennis lies to Rebecca about the fact that he is not going to the party. (8), on the other hand, is an infelicitous report since Dennis indeed had a lot of work. Thus, reporting (2) by an about-clause in (8) is a way of framing our intuitions and allows us to see that (2) is not a lie.

3. Lying with presuppositions

Let us focus on presuppositions. A presupposition is a piece of information that is taken for granted and commonly accepted. For instance, (9)–(10) presuppose the same thing, namely, that John owns a Mercedes. The difference between these cases is that the presupposition is carried by different triggers (i.e., factive verb and definite description, respectively):

Presuppositions can be informative. Thus, I can introduce the information that John owns a Mercedes to my audience by presupposing this fact. If my audience notices that I presuppose a new piece of information, they can accept and accommodate it. Such a process is called presupposition accommodation.Footnote ¹² Crucially, all cases of presuppositional lies discussed in literature are based on this process (see Meibauer Reference Meibauer2014; Viebahn Reference Viebahn2020; Viebahn et al. Reference Viebahn, Wiegmann, Engelmann and Willemsen2021). Take the following example (Viebahn Reference Viebahn2020, 735):

(11) is a clear case of a presuppositional lie. In this example, Harry on purpose presupposes something he believes to be false. The criteria of lying deliver clear results here. Harry cannot sincerely and consistently deny presupposing that John owns a Mercedes. Saying something like (12) does not work (Viebahn Reference Viebahn2020, 735):

Some might disagree with Viebahn’s judgment here and feel that (12) may not be quite as bad as (4). This is why it is good to consult also the second criterion. Harry’s presupposition can be naturally reported in the following way:

Thus, we have at least prima facie reasons for accepting the possibility of lying with presuppositions.

Viebahn (Reference Viebahn2020, 735–36), commenting on this case, makes three further points. Firstly, there seems to be no relevant difference between Harry’s presupposition made in (11) and asserting the same information, as in (14):

Secondly, ordinary speakers strongly agree that Harry lies in (11). In an experimental study (Viebahn et al. Reference Viebahn, Wiegmann, Engelmann and Willemsen2021), stories like this received lie ratings of around 90 percent (Mercedes I included). The results remained on this level even when participants could classify these cases as not being lies, but merely misleading. Finally, even though presuppositional content is always carried by a certain speech act, it is important that (11) is a question. So, we cannot say that the lie is asserted. The content of questions is a set of propositions (see, e.g., Cross and Roelofsen Reference Cross, Roelofsen and Zalta2020) and it is not suitable for lying. Thus, the only content in (11) that can be evaluated as a lie is presupposed.Footnote ¹³

The presupposition in (11) is triggered by a factive verb. Before considering in detail how we can lie with presuppositions, consider one more case with a different trigger (Viebahn Reference Viebahn2020, 738):

(15) triggers the presupposition (by employing a gendered pronoun) that Jack’s baby is a girl. Just as in the case of (11), this presupposition passes both criteria of lying. On the one hand, Gertrude cannot sincerely deny the presupposition made in (15). (16) sounds unnatural:

Just as (12), (16) seems to contradict the original statement made in (15) and hence sounds improper. On the other hand, Gertrude’s lie can be properly reported:

These two cases seem to provide an argument against ASSERTION-ONLY but are consistent with INFORMATIVENESS. In the rest of this section, I want to consider the question of why presuppositions are suitable for lying.

There are many ways of capturing the content of presuppositions. One take is to treat presuppositions as speaker commitments. It means that by presupposing p, one commits to the truth of the presupposed content and cannot deny it without contradicting oneself. For example, if Harry asks Rosa “Did you know that John owns a Mercedes?”, he cannot later deny that John owns a Mercedes without contradicting himself. Interestingly, the degree of speaker commitment is similar in presupposed and asserted contents (Mazzarella et al. Reference Mazzarella, Reinecke, Noveck and Mercier2018; Reins and Wiegmann Reference Reins and Wiegmann2021). However, while assertions contribute to at-issue content because they are a primary contribution in a context, presuppositions standardly contribute to non-at-issue content because their content is backgrounded (Simons et al. Reference Simons, Tonhauser, Beaver and Roberts2010). Because of that, presuppositions are subsidiary speaker commitments (Peters Reference Peters2016). The situation slightly complicates when we consider informative presuppositions. They can have the same effects on the conversational background as assertions (see, e.g., Gauker Reference Gauker2008), however, they are distinct from assertions since only the latter are directly communicated (Kissine and Pantazi Reference Kissine and Pantazi2020).Footnote ¹⁴

Can a definition of lying capture presuppositional lies? Stokke’s definition does not classify (11) and (15) as lies, for in neither case one says something one believes to be false. Stokke (Reference Stokke2017) explicitly argues that presuppositions cannot be lies for a simple reason: lies must be said, presuppositions are not, so they cannot be lies.Footnote ¹⁵ This is very problematic since both cases pass the criteria of lying and Stokke himself follows the deniability criterion. Furthermore, as I just indicated, at least in some contexts, presuppositions can be as committal as assertions. In both cases, the speaker commits to the communicated content. For this reason, some have proposed defining lying in terms of commitment.Footnote ¹⁶ Viebahn (Reference Viebahn2020, 743) recently proposed a commitment-based definition of lying that aims to capture presuppositional lies:

Viebahn (Reference Viebahn2020) explicates the notion of commitment by appealing to pretheoretical intuitions that are captured, among others, by the deniability criterion, i.e., if one commits to certain content, then one cannot consistently deny it without contradicting oneself. In his (2021, 307), he relies even more strongly on this criterion since the notion of commitment in lying is defined in terms of it. In the next section, I will discuss Viebahn’s view on presuppositional lies, but now, even without going into details, it can be seen that his definition is able to explain all cases discussed so far. Just as Dennis lies in (1), it is equally so for Harry in (11), and Gertrude in (15). They are lying because each of them commits to p and believes that p is false. All of these cases meet INFORMATIVENESS. In the next sections, I will expand the realm of possibilities of lying into uninformative cases.

4. Lying with uninformative presuppositions

Take the following extension of the Mercedes case:

The original story remains intact. What is different is the repetition of the presupposition from (11) in (18). However, this added presupposition is an uninformative one because Rosa already believes that John owns a Mercedes. Stating this presupposition again does not weaken Harry’s commitment, rather it can reinforce this presupposition. To answer the question of whether (18) is a lie, consider the familiar criteria. Just as before, Harry cannot consistently deny claiming that John owns a Mercedes and (18) can be naturally reported as lying about John owning a Mercedes.Footnote ¹⁷

The presupposition in (18) is triggered differently than in (11), i.e., by possessive construction. However, this only shows that lying is not restricted to a particular presupposition trigger. Instead of a declarative like (18), Harry could make any other statement that fits the context (for instance, an imperative utterance, “You must see John’s new Mercedes!”).

Uninformative lies can be characterised as lies with content that is already common ground. To emphasise, I am interested in cases of uninformative lies that will not be challenged as inappropriate.Footnote ¹⁸ In principle, Harry can reassert to Rosa that John owns a Mercedes, but then he can be rightly criticised for making an uninformative assertion—Rosa could object and respond something like “You already told me that!” However, it is perfectly natural to presuppose this information again and again (as in Mercedes II). In general, when lying, we do not want to be recognised as liars, and when our statement is challenged, as being in some sense inappropriate, we raise suspicion. Thus, an uninformative lie should be performed by means of a speech act that can transfer information that was already stated before. Presuppositions are a natural candidate for such cases because they are standardly treated as delivering already-known information. While one may presuppose information that is already commonly known, one cannot do so with assertions. This follows from an observation that “when it is correct to presuppose p, it is incorrect to assert it” (García-Carpintero Reference García-Carpintero2020, 23).

It is worth noting that an initial lie can be made by uttering various speech acts, granted that they are suitable for lying. The options are not limited to assertions. In the introduction, I mentioned the possibility of lying with conventional implicatures.Footnote ¹⁹ Thus, take Mercedes I or II and consider that instead of the first statement (11), Harry can communicate the same information by asserting (19) or conventionally implicating (20):

Once it is communicated that John owns a Mercedes, it is inappropriate to assert or conventionally implicate it again. Thus, (19) and (20) cannot be used as uninformative lies.

Before discussing some implications of lying with uninformative presuppositions, consider one more feature of uninformative lies. This time take the Baby I case with a minor alteration:

In this version, the uninformative lie is made by a different person than the initial lie. Everything else stays the same, i.e., in both cases, one presupposes something one believes to be false. Crucially, (21) carries an uninformative presupposition. Just as (15), it passes our criteria of lying. Lily cannot consistently deny claiming that Jack’s baby is a girl and we can naturally report her lie as lying about the gender of Jack’s baby.Footnote ²⁰ Moreover, when the same lie is not only repeated (like in Mercedes II) but is also made by various speakers it is reinforced even further, and thus it can be more compelling to the hearer to accept the presupposition. This shows that uninformative lies can be made in complex contexts and performed by multiple agents.

Let us return to the definition of lying. We saw that Viebahn’s (Reference Viebahn2020) view captures informative presuppositions. What about uninformative ones? His definition seems to not make a difference between them. For him, lying about p is basically committing to p while believing that p is false. It is instructive to discuss his views on presuppositional lies both in his (2020) and (2021) papers. Viebahn repeatedly distances himself from the possibility of accepting lying with uninformative content. His argument rests on the idea that in uninformative cases, one does not commit to p and thus one is not lying. We can extract two parts of his argument. The first one states that presuppositional lies are assertions and the second that uninformative presuppositions cannot be lies. Let us consider them in turn.

For Viebahn, all analysed cases of “presuppositional lies are assertions” (Reference Viebahn2020, 744). He observes that “the commitment in lying can plausibly be seen as the commitment in asserting” (Reference Viebahn2020, 743), and he directly states that “[l]ying involves the committal speech act of asserting” (Reference Viebahn2021, 307). He says that because he wants to marry two ideas, i.e., that we can lie with presuppositions and that lying is restricted to assertions. Thus, he tries to keep both ASSERTION-ONLY and INFORMATIVENESS. His reasoning can be summarised in the following way: the traditional definitions of lying maintain that all lies are assertions, the newly presented cases involving presuppositions are lies too, thus, we must accept that presuppositional lies are assertions.

Consider the implications of Viebahn’s view to the debates on assertion and presupposition. If Viebahn is right that the presented cases of presuppositions are assertions, then the classical definitions of assertion and presupposition are wrong. Many definitions of assertions have the so-called explicitness condition built-in, according to which, an assertion is a speech act that is explicitly presented (see, e.g., Alston Reference Alston2000). This condition aims precisely at distinguishing between assertoric content that is at-issue and non-assertoric content that is not-at-issue; presuppositions or conventional implicatures belong to the latter. We have an analogical situation with definitions of presupposition where one of their main aims is to separate what is asserted from what is presupposed.

Viebahn’s critique of all these views rests on the view that all lies are assertions. However, the fact that we can lie with presuppositions can simply be interpreted as working against the standard definition of lying, and so against ASSERTION-ONLY. In fact, this was already pointed out by Meibauer (Reference Meibauer2014). Moreover, Viebahn assumes the primacy of the definition of lying over the analogical definitions of assertion and presupposition. I do not see why. Take the explicitness condition in a definition of assertion. Viebahn (Reference Viebahn2020, 744–45) criticises such a condition since it disqualifies presuppositions from assertions. However, this is precisely the point of such a condition. The same can be said about defining presuppositions. There is widespread agreement among philosophers and linguists that asserting and presupposing are distinct ways of expressing content.Footnote ²¹ Viebahn needs more to undermine this distinction than just referring to the traditional definition of lying. Thus, I do not see good reasons to accept Viebahn’s view that presuppositional lies are assertions.

What about Viebahn’s view that there is a difference in commitment between informative and uninformative cases? This is what he says:

In both parts of this quotation, Viebahn seems to say that one commits to presuppositions if they are informative, but it is possible to not commit to presuppositions when they are uninformative. In fact, the latter is repeated throughout the article. Here I want to focus on Mercedes II and Baby II, which are cases of lying with uninformative presuppositions. In the last quoted statement, Viebahn entertains a case like Baby II, where Lily goes along with the presupposition that is already part of the common ground. By this view, such cases are not necessarily committal. But if they are not committal, they cannot be lies. However, from the context, it is clear that Lily is lying. In Mercedes II, it does not really make sense to say that Harry commits to what he presupposes only the first time.

In his (2021, 300), Viebahn adds one more condition to his definition of lying, i.e., an intention to communicate p to the hearer. This condition, Viebahn suggests, can be fleshed out in a Stalnakerian framework where a communicative intention p is an intention to propose p as an update to the common ground. I already mentioned, when discussing Stokke’s definition, that the common ground view treats assertions as informative speech acts, and hence accepts INFORMATIVENESS. This reasoning can be seen in the following condition that Viebahn formulates: “… agents do not intend to communicate p … if they presuppose p while taking p to be common ground” (Reference Viebahn2021, 301). Consider again the second presuppositions in Mercedes II and Baby II; Harry and Lily satisfy Viebahn’s condition, i.e., they presuppose something that they know is already a part of the common ground, which excludes the possibility of lying. However, intuitively they do lie. This condition is important because it is formulated in a more direct way than in (Reference Viebahn2020). In both cases, the result is the same. Viebahn’s view precludes the possibility of lying with uninformative content.

The upshot is that presuppositional lies are not reducible to assertions, and thus we must drop ASSERTION-ONLY. Moreover, restricting lies to informative presuppositions, as Viebahn proposes, is untenable. Uninformative presuppositions are just as good vehicles for lying as informative ones.Footnote ²² Notice that Viebahn’s commitment-based definition of lying, or similar definitions of lying formulated in commitment terms,Footnote ²³ can account for the presented cases. All that must be done is dropping the requirement of informativeness. In the next section, I expand the picture of uninformative lies to lies made by declarative statements, i.e., to such speech acts as reminding or guaranteeing.

5. Lying with uninformative speech acts

We can lie with all speech acts that entail the illocutionary force of assertion (see, e.g., Marsili Reference Marsili2020; Viebahn Reference Viebahn2021). This should extend the possibility of lying to speech acts whose commitment is at least as strong as in the case of assertions, thus to such speech acts as warning, guaranteeing, or swearing. Although for different reasons, both Marsili (Reference Marsili2020) and Viebahn (Reference Viebahn2021) argue that the aforementioned speech acts are assertions. Thus, when I warn or guarantee you that p I do that by asserting that p. The conclusion is a familiar one, i.e., if lying is restricted to assertions, the above cases are informative speech acts.

I propose to expand this picture by discussing two types of uninformative speech acts. Firstly, there are unique speech acts that are uninformative by design. Secondly, some speech acts can be both informative and uninformative, depending on the context. I will argue that both types of speech acts are suitable for lying.Footnote ²⁴

A paradigmatic case of a uniquely uninformative speech act is the speech act of reminding. Here is a standard case. By reminding you about the dentist appointment, I indicate that this piece of information has already been in the common ground, but I have a reason to believe you might have forgotten about it. Various speech act taxonomies classify reminding either as a special case of assertion or as a distinct speech act type (see, e.g., Searle and Vanderveken Reference Searle and Vanderveken1985; Alston Reference Alston2000). What matters for the present discussion is that reminding does not provide a new piece of information into the common ground (Abbot Reference Abbott2008; cf. Stalnaker Reference Stalnaker2008; Clapp Reference Clapp and Goldberg2020).Footnote ²⁵ Looking from the perspective of informativeness of assertion, it is natural to treat reminding as a distinct speech act type from assertions. This is not a merely taxonomical issue. If assertions are essentially informative speech acts, uninformative assertions are incorrect. Thus, if reminders are assertions, they are in some sense incorrect. However, this is not the case. It is correct and perfectly natural to remind you about your dentist appointment.Footnote ²⁶

Having said that, consider how we can lie by reminding something. We need only a small alteration in the already familiar case:

If (22) for someone does not sound natural enough, it can be modified accordingly. Rosa can also explicitly ask Harry, “Can you remind me what car John owns?” and thus elicit a reminder as a response. Reminding passes the familiar criteria of lying. Harry cannot consistently deny claiming that John owns a Mercedes and his reminding can be naturally reported as lying about John owning a Mercedes. Crucially, (22) is an uninformative statement.Footnote ²⁷

The second group of speech acts that can be used as uninformative lies contain speech acts that can be both informative and uninformative. I illustrate it with the speech act of guaranteeing. We rarely use speech acts with stronger commitment than assertions. It is broadly accepted that in asserting that p we represent ourselves as knowing that p (e.g., Williamson Reference Williamson1996). This is sufficient for most communicative contexts. However, sometimes context requires holding a stronger position or we want to indicate that our position is stronger than knowledge (say, we want to show that we are certain, or emphasise that we do know what we say). For instance, if I want to learn how to dive and know that you have some experience but do not know how substantial your experience is, you can either assert that you are a certified trainer or guarantee it to reassure me that I am safe with you. In both cases, you convey the same information.

One way to theorise about guaranteeing is to treat it as a speech act that reinforces an assertion (e.g., Turri Reference Turri2013). Thus, I can firstly assert that p, and later on, guarantee that p. Guaranteeing in this sense does not deliver new information, rather it repeats and reinforces the content that was already stated.Footnote ²⁸ Having said that, consider a case of lying by guaranteeing.

By repeating the same information in (23), Harry commits even stronger to the truth of this proposition than by presupposing it in (11). Notice that, just as in the case of reminding, there are more options to express guaranteeing. One is by saying “I know that p” (Austin Reference Austin1946; cf. Turri Reference Turri2013), as in “I know that John owns a Mercedes.” Finally, (23) passes the tests of lying. Here too Harry cannot consistently deny the content of (23) and his guaranteeing can be reported as lying about John owning a Mercedes.Footnote ²⁹

There are more speech acts that can be used uninformatively. Instead of (23), Harry can utter (24) or (25) which are also clear cases of lying:Footnote ³⁰

As a final note, consider what Viebahn and Stokke have to say about the analysed cases. Neither of them engages in the discussion of whether lying is possible with uninformative content. Viebahn argues that one does not commit to p if p already is a part of the common ground (Reference Viebahn2020, 743). This equally concerns the content that is presupposed, reminded, or guaranteed. Stokke’s view gives a different verdict to uninformative lies that are made with presuppositions and declarative statements. He explicitly argues against counting presuppositions as lies (Reference Stokke2017, Reference Stokke2018) because presuppositions are not said (in the relevant sense, see section 2). The content of (22)–(25) is said,Footnote ³¹ thus these statements should be considered lies. However, Stokke follows Stalnaker in defining an assertion as a proposal to update the common ground. The content of (22)–(25) cannot update the common ground since this content already belongs to the common ground.Footnote ³² Thus, Stokke’s view predicts that (22)–(25) should be infelicitous. However, these statements are perfectly appropriate to make. Thus, Stokke’s definition delivers wrong predictions to lies made by uninformative declarative statements.