How to Redo Things with Words: Deniability and the Determinants of Illocutionary Force

Abstract

When one speaks duplicitously one performs a (risky) speech act and at the same time intends that one have deniability regarding that speech act. What is it for a speaker who performs an illocutionary act to have deniability regarding that act? I first review an answer to this question proposed by Alexander Dinges and Julia Zakkou. According to them, deniability is “an epistemic notion. A speaker has deniability if she can make it epistemically irrational for her audience to reason in certain ways” (2023, p. 373). I argue, however, that the epistemic account is refuted by a counterexample, and informed by the counterexample, I propose an alternative. According to my alternative account, deniability concerns illocutionary norms: a speaker has deniability regarding an illocutionary action iff (i) she is permitted to deny having performed the action, and (ii) if she denies having performed the action, she thereby avoids being responsible for it. I conclude by considering the implications of illocutionary norms account of deniability for the vexed issue of the determinants of illocutionary force; I argue that the illocutionary norms account of deniability provides evidence in support of McDonald’s (2021) “collaboration” view.

Keywords:
Speech Action; Austin; Deniability; Illocutionary Force

Introduction: Strategic Speech, Duplicitous Speech and Deniability

Sometimes we speak strategically: we speak in a way that minimizes the risk of saying something - performing an illocutionary action - for which we may come to pay a social, political, or interpersonal cost. One common strategy we employ is to speak duplicitously: we intentionally perform a risky speech act Φ, but do so in such a way that we can subsequently perform a complex second-order speech act in which we deny having performed Φ and claim to have performed an alternative, innocuous, speech act Ψ.¹ There are three categories of such complex second-order speech acts, corresponding to three dimensions along which a risky speech act Φ and an alternative innocuous speech act Ψ might differ. First, Φ and Ψ can differ with regard to their sincerity. That is, after making an utterance that is interpreted as the performance of a risky speech act, a duplicitous speaker may subsequently claim that she was not serious, that she was only joking.² Second, Φ and Ψ might comprise distinct locutionary actions. That is, after making an utterance that is interpreted as the performance of a risky speech act with content p, the speaker may subsequently claim that the content of her utterance was really - or was at least intended to be - distinct content q.³ Finally, third, Φ and Ψ can comprise the same locutionary actions, but differ in illocutionary force. That is, after making an utterance that is interpreted as the performance of a risky speech act with illocutionary force F1, the speaker may subsequently claim that the illocutionary force of her utterance was really - or was at least intended to be - distinct force F2. Here I will ignore the first category, and most of my remarks will concern the third category of complex second-order speech act.⁴

Here's an illustrative real-life example of the third category. In the Spring of 2020 there were protests in Minneapolis MN calling for police reform after the murder of George Floyd by police officers, and these protests involved violence and property destruction. In response to this civil unrest, then President Trump tweeted, “Either … bring the City under control, or I will send in the National Guard & get the job done right … Any difficulty and we will assume control but, when the looting starts, the shooting starts” (my italics).⁵ Trump’s post was immediately criticized for sanctioning a policy of illegal and immoral state-sponsored violence.⁶ In response to this criticism, some hours after posting the original tweet Trump tweeted, “looting leads to shooting. … I don’t want this to happen, and that’s what the expression put out last night meant” (my emphasis). Trump’s original tweet, as evidenced by widespread critical response, was interpreted as what Austin (1962, p. 155) called an exercitive: in posting ‘when the looting starts, the shooting starts’ Trump was understood to be enacting (or advocating) the policy that looters be summarily shot. But, confronted with the criticism of many who interpreted Trump’s original post in this way, Trump performed a subsequent complex second-order speech act in which he denied that he had been enacting (or advocating) such immoral and illegal policy, and claimed that he had instead merely been informing (or cautioning) that looting leads to shooting. Trump’s subsequent post thus constitutes an instance of the third category: he performs a complex second-order speech act in which he denies that his previous utterance was an objectionable exercitive and claims that it was instead an innocuous verdictive (Austin, 1962).

One might question whether Trump’s original post was really duplicitous, and not merely serendipitously open to interpretation. In general, one speaks duplicitously iff one performs a (risky) speech act and at the same time intends that one has deniability regarding that speech act.⁷ So, Trump’s initial post counts as a strategic duplicitous speech act only if he intended to perform a (risky) exercitive speech act that he could subsequently deny was an exercitive speech act and claim was instead an (innocuous) verdictive speech act. But maybe Trump had no such devious plan in mind when he performed his initial act; maybe that he was able to deny that he had performed an objectionable exercitive and claim that he instead had performed an innocuous verdictive occurred to Trump only after he became aware of the criticism of his original post. And how can we know what Trump’s intentions were? It seems at least as difficult for us to know whether Trump’s original post was duplicitous as it is for us to know whether Trump’s original post was an exercitive or a verdictive.⁸ What Fricker (2012, p. 89) calls the “dodgy epistemics” of pragmatic interpretation impede not only our ability to discern what illocutionary act Trump performed in uttering ‘When the looting starts, the shooting starts’, but also our ability to discern whether in making that utterance he was (intentionally) speaking duplicitously. Nonetheless, despite the dodgy epistemics, Trump’s tweeting is illustrative for my purposes, because the specific question to be addressed here does not concern duplicitous speech per se, but rather deniability: What is it for a speaker who performs an illocutionary act to have deniability regarding that act?

In what follows I first, in section 1, review an answer to this question proposed by Dinges and Zakkou (2023). According to Dinges and Zakkou (D&Z), deniability is “an epistemic notion. A speaker has deniability if she can make it epistemically irrational for her audience to reason in certain ways” (2023, p. 73). In section 2 I present a counterexample against the epistemic account of deniability. And then in section 3, informed by the counterexample, I propose an alternative definition of deniability in terms of illocutionary norms. Finally, in section 4 I turn to the vexed issue of the determinants of illocutionary force; I argue that the illocutionary norms account of deniability supports McDonald’s (2021) “collaboration” view.

1. The Epistemic Account

To motivate their epistemic account of deniability, D&Z formulate a list of primary and secondary desiderata, and argue that their epistemic account satisfies these desiderata. I will first present and clarify the desiderata, and I will then explain why D&Z think they are satisfied by the epistemic account.

D&Z present three primary desiderata, the first of which is:

(D1) “In the target cases, deniability is desirable for the speaker” (2023, p. 378).

It is important to clarify that the desirability of having deniability regarding illocutionary action Φ accrues from the benefits that would be obtained (or costs that would be avoided) if the speaker to deny having performed Φ. That is, the desirability of having deniability must be explained as arising from benefits that would accrue to the speaker by her denying, and not, say, in virtue of a pleasant sensation one experiences while having deniability. So, I will use this clarified version of D&Z’s first desideratum:

(D1*) If a speaker has deniability regarding Φ and she denies having performed Φ, then the speaker thereby accrues some benefit.

The other two primary desiderata are complementary:

(D2) “In [typical cases], speaking directly isn’t a means of retaining deniability” (2023, p. 378).

(D3) “In [typical cases], speaking indirectly is a means of retaining deniability” (2023, p. 378).⁹

Understanding (D2) and (D3) requires understanding the distinction between speaking directly and speaking indirectly, and D&Z offer little guidance here.¹⁰ I suggest that what D&Z have in mind by “speaking directly” is speaking in a way that is not open to interpretation; or, as I will say, is closed to interpretation. The idea is that an utterance of sentence Σ is closed to interpretation to the extent that the conventional meaning of Σ determines the content and force of the illocutionary action (or actions) that are thereby performed. In other words, the extent to which an utterance is open to interpretation is the extent to which it is subject to what Fricker (2012) would call the “dodgy epistemics of pragmatic interpretation.”

Consider a typical example discussed by D&Z: A speeding driver is stopped by a policeman, and the driver desires to duplicitously offer the policeman a bribe to avoid getting a ticket. Uttering which of the following sentences would enable the driver to achieve deniability concerning her offering a bribe?

[1] I hereby offer you a bribe so you won’t give me a ticket.

[2] “Is there any way we can settle this right now?” (D&Z, 2023, p. 374)

As I have defined open to interpretation, utterances of both [1] and [2] would be at least somewhat open to interpretation, and for various reasons. Both are open to locutionary interpretation, since, e.g., it is at least somewhat open to interpretation to whom the speaker is referring in using ‘you’ and ‘we’. But there is a clear sense in which a (sincere) utterance of [2] is more open to interpretation, because an utterance of [2], but not [1], is also open to illocutionary interpretation. That is, [1] is, in Austin’s terms, an explicit performative, and thus the illocutionary force of a (sincere¹¹) utterance of [1] is conventionally determined by the content of the utterance; an utterance of [1] is thus closed to illocutionary interpretation. In contrast, [2] is open to illocutionary interpretation, because the illocutionary force of a (sincere) utterance of [2] is not conventionally determined by the content of the utterance - Is an utterance of [2] an offer of bribe, or an innocuous question, or what?¹²

The concept of an utterance being open (closed) to interpretation thus clarified, I will use the following combined desideratum in place of (D2) and (D3):

(D2-3*) In typical cases, performing an illocutionary action Φ by uttering sentence Σ in context c is a means of retaining deniability regarding Φ iff Σ in c is, to a sufficiently high degree, open to interpretation.

D&Z present two secondary desiderata, both of which concern the epistemic status of the audience¹³:

(D4) “Deniability is consistent with the audience knowing the speaker’s communicative intentions [and thus knowing what illocutionary action was performed] in cases where the respective denial is not actually made” (2023, p. 379).

Note that taken together (D2-3*) and (D4) imply that just because a speaker’s utterance is open to interpretation, it does not follow that her audience does not know what illocutionary act she therein performed. According to (D2- 3*), in typical cases a speaker has deniability regarding an illocutionary action Φ only if her utterance is, to a sufficiently high degree, open to interpretation. But according to (D4) a speaker can have deniability regarding Φ even though her audience knows that she performed Φ. It follows that, at least in typical cases, an utterance that is sufficiently open to interpretation to permit the speaker to have deniability regarding the illocutionary action Φ she therein performs can nonetheless be such that the audience knows what she therein performed Φ. To motivate (D4), suppose that the speeding driver utters [2]. Suppose further that the policeman is corrupt, and he accepts the offered bribe immediately and without hesitation. In this typical case, it is natural and unproblematic to further assume that, despite the driver’s utterance of [2] being open to interpretation, the policeman knows the driver is offering him a bribe.

The last desideratum may, at first, seem puzzling:

(D5) “Deniability is consistent with the audience finding the denial implausible, in cases where the respective denial is actually made” (2023, p. 380).

What could it mean to say that a speaker has deniability regarding an illocutionary action Φ - i.e. she is able to deny having performed Φ - if having this ability is compatible with the audience rejecting her denial? Puzzling though it may be, (D5) describes a feature possessed by many, perhaps most, instances of deniability. Consider again the speeding driver who utters [2]. Suppose now that the policeman does not immediately and unhesitatingly accept the bribe, but instead, in response to the driver’s utterance of [2], accuses the driver of having offered him a bribe. Finally, suppose the driver then denies having offered a bribe and claims that in uttering [2] she merely asked a question. In this typical case the policeman may not find the driver’s denial plausible; he may continue believe that by uttering [2] the driver was indeed offering a bribe. But the policeman’s incredulity is compatible with the driver having deniability regarding the illocutionary act she performed in uttering [2]. One final remark concerning (D5): taken together with (D1*), (D5) implies that when a speaker has deniability regarding Φ, the benefit accrued to the speaker if she denies having performed Φ does not require that the audience find the denial plausible - performing the denial can be beneficial for the speaker even if the audience does not buy it.

D&Z’s state their epistemic account of deniability as follows:

S has deniability [regarding] Φ if and only if: if S denies that she [performed] Φ, then it is not epistemically proper for S’s audience to treat that she meant to Φ as a reason for actions or emotions. (2023, p. 391)¹⁴

The basic idea of the epistemic account is that in typical cases of deniability the speaker’s denying having performed Φ results in the audience’s epistemic status concerning the proposition that the speaker intended to Φ being undermined to the extent that it would now be epistemically improper for the audience to treat that proposition as a reason.¹⁵ Now, why do D&Z think that their epistemic account satisfies the desiderata?

The epistemic account seems well positioned to satisfy (D1*). Consider again the speeding driver. According to the epistemic account, if the driver has deniability with regard to her offering the policeman a bribe, then by denying that she offered him a bribe and claiming to have merely asked a question, she thereby “robs” the policeman of “reasons for adversarial actions” (2023, p. 386) - e.g. she makes it epistemically improper for the policeman to accuse her of, or to arrest her for, intending to have offered him a bribe. (And she also thereby robs the policeman of some reasons for harboring negative emotions toward her - e.g. she makes it epistemically improper for the policeman to resent her for intending to have offered him a bribe.) So, D&Z claim that the epistemic account satisfies (D1*) because, according to this account, if a speaker with deniability regarding Φ denies that she performed Φ, then she thereby accrues the benefit of having made it epistemically improper for the audience to perform certain adverse actions or to harbor certain adverse emotions.

D&Z claim that their account “straightforwardly” (2023, p. 387) satisfies (D2-3*). On the epistemic account a speaker has deniability regarding Φ if and only if the speaker’s denial of having performed Φ undermines, to the required extent, the audience’s epistemic status concerning the proposition that the speaker intended to perform Φ. But, D&Z claim, the speaker’s denial will (typically) have the requisite undermining effect only if the utterance whereby the speaker performed Φ was, to a sufficiently high degree, open to interpretation. Consider yet again the speeding driver. Suppose that instead of strategically uttering [2], the driver, while waving some cash, had instead brashly uttered [1]. I assume that her utterance of [1] would not be, to a sufficiently high degree, open to interpretation - her utterance of [1] would not be subject to “the dodgy epistemics of pragmatic interpretation.” As a consequence, the driver’s subsequent denial of having offered a bribe would not suffice to undermine the policeman’s epistemic status concerning the proposition that the driver had intended to offer him a bribe; the driver’s utterance of [1] constitutes very strong evidence that she intends to offer a bribe, evidence that is not undermined by the driver’s subsequent denial. So, uttering [2], which is open to interpretation, is a means for the driver to have deniability regarding her intending to offer the policeman a bribe, whereas uttering [1], which is closed to interpretation, is not a means for her to have deniability regarding her intending to offer the policeman a bribe.

Finally, what about the epistemic desiderata (D4) and (D5)? The epistemic account straightforwardly satisfies (D4), so long as it is granted that knowledge, and epistemic status more generally, is defeasible. Granted this plausible assumption, that the audience knows that the speaker intended to perform Φ is consistent with its being the case that this knowledge can be defeated, or the epistemic status can be undermined, by the speaker’s denial that she performed Φ. For similar reasons, the epistemic account satisfies (D5). That is, the audience’s knowledge, or epistemic status concerning, that the speaker intended to perform Φ might be defeated by the speaker’s denial that she performed Φ, even if the audience does not find the denial plausible. For example, the driver’s denial that she offered a bribe may suffice to defeat the policeman’s knowledge that she intended to offer him a bribe, even if he does not buy the denial - the driver’s denial might leave him in an agnostic state of believing neither that she intended to offer him a bribe nor that she intended merely to ask a question.

2. A Counterexample against the Epistemic Account

I am going to describe a case wherein the speaker retains deniability regarding Φ and denies having performed Φ, but the denial does not undermine the audience’s epistemic status concerning the proposition that the speaker intended to perform Φ. The case thus demonstrates that the condition that the epistemic account claims to be necessary for a speaker’s having deniability is not in fact necessary. I dub my counterexample the big pretzel case, as it involves an utterance of ‘I will buy you a big pretzel’ that is open to being interpreted as either a promise, or a prediction. In order for the big pretzel case to demonstrate that the epistemic account fails in the necessary direction, the case must meet these conditions: (i) the speaker utters ‘I will buy you a big pretzel’ therein making a promise; and (ii) the speaker retains deniability regarding her having therein made a promise; yet (iii) the speaker’s subsequent denial of having therein made a promise and claim to have merely made a prediction does not undermine the audience’s epistemic status concerning the proposition that in making the initial utterance the speaker intended to make a promise. Here’s a plausible scenario in which conditions (i)-(iii) are met:

I like to go to basketball games to cheer on our local team, but I do not like to go alone. My son Leo is sometimes willing to go with me, and when he does, at half-time he requests that I buy him a big pretzel, and I usually buy him one, but not always: if our team is losing badly at half-time - and they often are - I will typically decline to buy Leo a big pretzel, because I anticipate we will want to leave early. Now suppose that in trying to persuade Leo to come with me to an upcoming game, I utter ‘I’ll buy you a big pretzel’ and therein promise to buy Leo a big pretzel at half-time. Motivated by the promise of a big pretzel, Leo comes with me to the game. Half-time comes around, and our team is losing. Leo says, “Let’s go get the pretzel you promised me.” Then, duplicitous bastard that I am, I utter “No, I didn’t promise to buy you a pretzel. When I said, ‘I’ll buy you a big pretzel’ I was making a prediction. And it turns out it was a false prediction, because we’re losing!” Leo, who knows me all too well, does not find my complex second-order speech act plausible in the least. He says, “Bullshit. But whatever Dad. Can I at least get a coke?”¹⁶

To refute the epistemic account, I must establish that even if it is stipulated that my subsequent complex second-order speech act does not undermine his epistemic status concerning the proposition that I had intended to make a promise, nonetheless the case is naturally and unproblematically understood as manifesting deniability. In proposing such a counterexample, however, I run the risk of reducing the debate to haggling over how the technical term deniability is to be used. The risk arises because in debating the adequacy of the epistemic account we find ourselves in the unfortunate (though familiar¹⁷) situation of attempting to simultaneously identify and explain a phenomenon. Because the epistemic account puts us in this unfortunate position, it would be open to D&Z to reject any purported counterexample on the grounds that it does not manifest what they take to be the phenomenon of deniability.¹⁸ In an effort to block this sort of potential response, I will illustrate how the big pretzel case can naturally and unproblematically be understood as possessing all the features described in the desiderata presented by D&Z, thus showing that it possesses the features that D&Z think an adequate account of deniability must explain. Illustrating how the big pretzel can be naturally and unproblematically understood as possessing these features does not wholly preclude D&Z from claiming that my counterexample does not manifest what they mean by deniability, but it does impose a burden on D&Z to provide some principled motivation for any such claim.

So, I hereby stipulate that in the big pretzel case, my denial does not undermine Leo’s epistemic status concerning the proposition that I had intended to promise to buy him a big pretzel - his reasons for uttering “Bullshit” are epistemically proper. Now, can the case nonetheless be naturally and unproblematically understood as possessing the features described in desiderata (D1*), (D2-3*), (D4) and (D5)?

Consider (D1*). In denying that I promised to buy Leo a big pretzel, did I thereby accrue some benefit, despite the fact - the stipulation - that my denial and claim did not undermine Leo’s epistemic status concerning the proposition that I had intended to promise to buy him a big pretzel? What, other than Leo’s epistemic status concerning that proposition, may have changed as a result of my felicitous performance of this complex second-order speech act?¹⁹ Once we put the question this way the answer is obvious: what has changed is that the common ground of the conversation now contains the information that I have performed this second- order illocutionary action that is about my previous illocutionary action. The question then is, how can this change in the common ground result in a benefit for me, even though Leo is incredulous concerning my denial and claim?

Following Lewis (1979) let us assume that the common ground - the mutually accepted information at any point in a conversation - includes a scoreboard. The scoreboard is a trove of mutually accepted information - the conversational score - about the status of a conversation that evolves throughout the conversation. The score thus includes “those evolving assumptions which govern the permissibility, relevance, and interpretation of sentences within the dynamic discourse structure” (Camp, 2018, p. 60). Now, in the moment before I perform my denial and claim, what is the conversational score? Just before I perform my denial and claim, the scoreboard reflects that I had promised to buy Leo a pretzel. A conventional effect of my having performed that illocutionary action - of my having made that promise - is that the normative statuses of Leo and I are changed: in performing the promise, I created certain obligations for me, and certain corresponding rights for Leo. The creation of such obligations and rights is constitutive of the act of promising. And all illocutionary actions have such constitutive normative effects.²⁰ McDonald, influenced by Sbisà’s (2009a, 2014) interpretation of Austin (1962), explains that “when we perform an illocutionary act we constitutively take on and assign to others normative statuses. The change in normative statuses is not a downstream effect of the illocutionary act but rather constitutive of it” (2021, p. 15). The upshot is that before I utter my second-order denial and claim, Leo and I mutually accept that in uttering ‘I will buy you a big pretzel’ I therein had made a promise, so at that early time the scoreboard reflects that I made such a promise, and that illocutionary action is constituted by its effects on the normative statuses of Leo and I: if I therein promised to buy Leo a big pretzel, then I therein created obligations on my part, and rights on Leo’s part. In other words, before I perform my second-order denial and claim, the scoreboard has me on the illocutionary hook to buy Leo a big pretzel.

What is the effect on the conversational scoreboard of my subsequent second-order denial and claim? Because only information that is mutually accepted is recorded on the scoreboard, the effect of this second-order speech act is to remove from the scoreboard the information that in uttering ‘I will buy you a big pretzel’ I had performed a promise. And since my (now) refusing to accept this information is independent of Leo’s epistemic status concerning what my intentions in making this utterance had been, my denial succeeds in having this illocutionary effect regardless of Leo’s incredulity concerning my denial. That is, we are stipulating that after my denial Leo still believes - perhaps even knows - that I had intended to promise to buy him a pretzel, but despite Leo’s incredulity, after my second-order speech act it is no longer mutually accepted that I performed a promise; the scoreboard now includes the information that I deny that my initial illocutionary action was a promise, and claim instead that it was a prediction, which of course implies that that it is now common ground that I do not accept that I had promised to buy Leo a pretzel. To be clear, Leo rejects my complex second-order speech act - “Bullshit” he says. So, after my complex second-order speech act, the scoreboard does not reflect that I did not promise, nor that I did promise; what it reflects now is that Leo and I disagree about whether I had promised. But that the scoreboard now reflects this disagreement, as opposed to reflecting that I had promised, is to my benefit: since the scoreboard now reflects that it is disputed whether I promised, it also reflects that it is disputed whether I am obligated to buy Leo a pretzel. That is, after my second-order denial and claim, am no longer on the illocutionary hook to buy Leo a big pretzel. That is the benefit accrued to me by my performance of the complex second-order speech act, and that benefit accrues to me regardless of Leo’s incredulity. The big pretzel case can thus be naturally and unproblematically understood as possessing the feature described in (D1*).

Why can the big pretzel case be naturally and unproblematically understood as possessing the features described in desideratum (D2-3*)? Let us first consider this desideratum in the “if ” direction: In the big pretzel case, why is my having uttered ‘I will buy you a big pretzel’, which we may assume is to a sufficiently high degree open to interpretation, a means of retaining deniability regarding having promised to buy Leo a pretzel? My utterance of ‘I will buy you a big pretzel’ is not merely open to interpretation; it is moreover open to being interpreted in the way I propose in my second-order claim. That is, it is compatible with the conventional meaning of ‘I will buy you a big pretzel’ that the illocutionary act I performed was a prediction, and not a promise. What the big pretzel case, and the typical cases discussed by D&Z, evidence is that conversational scorekeeping is governed by the following limited deference to speaker norm:

(LDS) If at t _n the scoreboard reflects that a previous utterance is open to being interpreted as having either Φ or Ψ (where Φ ≠ Ψ), and that the speaker denied having therein performed Φ and claimed to have therein performed Ψ, then at t _n + 1 the scoreboard will not reflect that the speaker performed Φ.

The LDS norm states a sufficient condition for the scoreboard being developed so that it does not reflect that a speaker performed a certain illocutionary action; it does not state a sufficient condition for the scoreboard’s being developed so that it reflects that a speaker did, or did not, perform a certain illocutionary action. In the big pretzel case, in denying that I made a promise and claiming that I made a prediction, I rely on the development of the scoreboard for our conversation being guided by the LDS norm to remove the information that I made a promise. The LDS norm thus permits me, subsequent to my utterance of ‘I will buy you a big pretzel’, to deny that I therein promised and to claim that I therein predicted, and thereby bring about a scoreboard that reflects neither that I promised nor that I did not promise. It is in this sense that my having uttered ‘I will buy you a big pretzel’, which we may assume is to a sufficiently high degree open to interpretation, is a means of my being able to deny having promised to buy Leo a big pretzel. The big pretzel case thus possesses the feature described by (D2-3*) in the “if ” direction.

Let us now consider desideratum (D2-3*) in the “only if ” direction: In the big pretzel case, why would my having uttered ‘I hereby promise to buy you a big pretzel’, which we may assume would not be to a sufficiently high degree open to interpretation, not be a means of my retaining deniability regarding having promised to buy Leo a pretzel? Suppose that I had initially uttered ‘I hereby promise to buy you a big pretzel’ and then subsequently uttered, “No, I didn’t promise to buy you a pretzel. When I said, ‘I hereby promise to buy you a pretzel’ I was just making a prediction.” In this alternative big pretzel case, there is a blatant inconsistency between my initial utterance and my subsequent second- order attempted denial and claim; the conventional meaning ‘I hereby promise to buy you a big pretzel’ is inconsistent with my second-order attempted denial that I therein promised to by Leo a big pretzel. In this alternative case, my subsequent attempted denial would be not merely implausible; it would be moreover pragmatically impermissible.

There is a distinction between a permissible implausible denial, i.e. a second-order speech act made in accordance with the norms of conversation that results in a retroactive change to the scoreboard, and an impermissible attempted denial, i.e. a second-order attempted speech act that violates the norms of conversation and thus does not result in a retroactive change in the scoreboard. That is, though the norms of conversation allow for permissible implausible denying, they do they do not allow for illocutionary gaslighting. Camp (2018) characterizes this distinction as follows:

… a speaker who exploits deniability ‘plays to a virtual audience’ (Goffman 1967; Lee and Pinker 2010, 7896), pretending to address [her utterance] to a possible hearer HP who would sincerely employ … alternative assumptions I′ to derive M(Q)′. For the alternative interpretation M(Q)′ to be admissible - or above the threshold of ‘plausible deniability’ - it must be reasonable to calculate M(Q)′ on the basis of the uttered sentence’s conventional meaning F(P), the commitments undertaken in the conversation to this point, and some set I′ of epistemically accessible presuppositions consistent with those commitments, in a way that renders U at least minimally conversationally cooperative (Camp, 2018, p. 50).²¹

Applied to the alternative big pretzel case, Camp’s point is that if I had initially uttered ‘I hereby promise to buy you a big pretzel’ and subsequently attempted to deny having therein made a promise, I would not be even minimally conversationally cooperative.

The alternative big pretzel case, and analogous cases involving utterances that are closed to illocutionary interpretation, evidence that conversational scorekeeping is governed by a norm complimentary to LDS that forbids illocutionary gaslighting:

(FIG) If at t _n the scoreboard reflects that a previous utterance has Φ and is closed to illocutionary interpretation, and that the speaker has attempted to deny having therein performed Φ to claim to have therein performed Ψ (where Φ ≠ Ψ), then no illocutionary effect of the attempted denial and claim is reflected on the scoreboard at t _n+1 .²²

In the alternative big pretzel case, I initially utter ‘I hereby promise to buy you a big pretzel’, and subsequently attempt to deny that I therein made a promise and attempt to claim that I instead merely made a prediction. In performing, or attempting to perform, these subsequent second-order speech acts, I am attempting to develop in keeping with conversation norms the scoreboard so that it no longer reflects that I made a promise. But my attempt fails because it flagrantly violates the FIG norm; the FIG norm precludes my second-order utterances from developing in keeping with the norms of conversation the scoreboard. It is in this sense that my having uttered ‘I hereby promise to buy you a big pretzel’, which we may assume would not be to a sufficiently high degree open to interpretation, would not be a means of my being able to deny having promised to buy Leo a pretzel. I conclude that it is also natural and unproblematic to understand the big pretzel case as possessing the feature described by (D2-3*) in the “only if ” direction.²³

Finally, can the big pretzel case be naturally and unproblematically understood as possessing the features described in the explicitly epistemic desiderata (D4) and (D5)? The epistemic features described in (D4) can simply be stipulated to hold in the big pretzel case. That is, we can stipulate that before I performed my second-order speech act, Leo knew that I had intended to promise to buy him a big pretzel, and moreover that if I had not performed the subsequent complex second-order speech act, he would have continued to have this knowledge. What about the features described in (D5)? I have taken pains to stipulate that the big pretzel case possesses the feature described in (D5). Indeed, I have gone so far as to stipulate that my second- order denial and claim do not undermine Leo’s epistemic status concerning the proposition that I had intended to promise to buy him a pretzel in the least.

I conclude that the big pretzel case constitutes a counterexample against D&Z’s epistemic account of plausible deniability: the case manifests all the characteristics of deniability, because can be naturally and unproblematically understand as possessing all the features described in desiderata (D1), (D2-3*), (D4) and (D5). But, by stipulation, the case does not satisfy the epistemic condition that D&Z claim is necessary for deniability.

3. An Alternative Account: Deniability Specified and Explained in Terms of Illocuationary Norms

In presenting my alternative view, I will distinguish between the task of specifying the phenomenon of a speaker’s retaining deniability regarding Φ and the additional task of explaining how, in a typical case, the speaker manages to retain deniability regarding Φ. My proposed specification is, I believe, uncontroversial. Consider the opening passage of D&Z’s paper in which they approvingly cite Lee and Pinker’s (2010), and Camp’s (2018) characterizations of deniability:

Communication can be risky. Like other kinds of actions, it comes with potential costs. For instance, an utterance can be embarrassing, offensive, or downright illegal. In the face of such risks, speakers … seek ‘plausible deniability’ (Lee and Pinker 2010, p. 785) and try to ‘minimize their conversational commitments’ (Camp 2018, p. 41). … While retaining deniability is usually beneficial for the speaker, it often poses problems for the hearer. We want to hold speakers accountable for their conversational behaviour, but when speakers retain deniability, they leave us in a ‘frustrating position, …’ (Camp 2018, p. 46). (2023, p. 372, my emphasis)

Note that this passage makes no mention whatsoever of the hearer’s epistemic status concerning any proposition, but it does explicitly invoke the hearer’s “frustration” at being unable to “hold speakers accountable for their conversational behaviour.” I propose then that the specification - the definition, if you will - of deniability that is all but explicit in this passage is as follows:

A speaker S retains deniability regarding an illocutionary action Φ iff (i) S is permitted to deny having performed Φ and claim to have performed a different illocutionary action Ψ, and (ii) if S denies having performed Φ and claims to have performed Ψ, S thereby avoids being accountable for having performed Φ.²⁴

D&Z claim that, at least in typical cases, the correct explanation of how speakers manage to retain deniability regarding Φ crucially involves the ability of the speaker to undermine the audience’s epistemic status concerning the proposition that the speaker had intended to perform Φ. The above presentation of the big pretzel case reveals, however, that D&Z’s proposed explanation must be overlooking something essential, because this case is naturally and unproblematically understood as manifesting deniability, yet D&Z’s proposed explanation does not apply. Abstracting from the details of the big pretzel case provides a general alternative explanatory schema of how speakers in typical cases retain deniability. Such a speaker S performs Φ by making an utterance that is, to a sufficiently high degree, open to interpretation. Because the utterance is open to interpretation, S is able to subsequently deny having performed Φ and claim to have performed a different action Ψ without violating the norms of minimal conversational cooperativity. So, (i) S is permitted to deny having performed Φ and to claim to have performed Ψ. Moreover, since S is permitted to perform these second-order speech acts, if S does perform them, the conversational scoreboard must reflect that S performed them. But, the norms of conversation governing utterances that are open to interpretation require limited deference to speakers (LDS): If at t _n the scoreboard reflects that the utterance is open to being interpreted as either Φ or Ψ, and that S denied having performed Φ and claimed to have performed Ψ, then at t _n+1 the scoreboard will not reflect that S performed Φ, and thus at t _n+1 the scoreboard will no longer reflect that the normative statuses constitutive of Φ have been created, and S will no longer be on the illocutionary hook for having performed Φ. So, (ii) if S denies having performed Φ and claims to have performed a different illocutionary action Ψ, S thereby avoids being accountable for having performed Φ.

4. Implications for the Determinants of Illocutionary Force

If an utterance is open to illocutionary interpretation, to what extent is its illocutionary force determined by the intentions of the speaker, and to what extent is it determined by the “uptake” of the audience? To forestall confusion, two points of clarification are in order: First, this question is asked only of utterances whose illocutionary force is, to a sufficient degree, open to illocutionary interpretation. As Strawson (1964) and others have observed, it is undeniable that some highly conventionalized - and therefore closed to illocutionary interpretation - utterances have the forces they have regardless of the intentions of the speaker.²⁵ And second, the question is not asking about the success conditions for antecedently specified illocutionary actions; to my knowledge, all theorists accept that at least some illocutionary actions have “external” success conditions.²⁶ Rather, because the success conditions for an illocutionary action performed in making an utterance are entailed by the illocutionary force of the utterance, the question is rather asking, at least part, how such success conditions are themselves determined.

Three different responses to the question, thus clarified, have been recently defended. Bianchi (2021, 2023) argues in support of a “speaker-directed” view, according to which the force of an utterance open to illocutionary interpretation does not in any way depend on the uptake of the audience.²⁷ In contrast, Quill Kukla (2014 written under ‘Rebecca Kukla’, forthcoming) defends a “constitution” view, according to which the illocutionary force an audience interprets such an utterance as having can constitute the illocutionary force of the utterance, regardless intentions of the speaker.²⁸ And finally, McDonald (2021) defends a “collaboration” view, according to which “the illocutionary force of an utterance is determined by an agreement reached by the speaker and the hearer” (2021, p. 2).²⁹ I will demonstrate that the illocutionary norms account of deniability I have developed in the preceding is incompatible with the speaker-directed and constitution views, and thus supports the collaboration view.³⁰

A fundamental assumption of the illocutionary norms account of deniability is that a speaker who makes an utterance therein performs an illocutionary action Φ iff the information that the speaker has therein performed Φ is recorded on the conversational scoreboard, i.e. only if speaker and audience mutually accept that in making the utterance the speaker performs Φ.³¹ This fundamental assumption implies the above articulated limited deference to speakers (LDS) principle. (LDS) is formulated in terms of illocutionary actions performed by speakers in making utterances, and thus (LDS) entails a more specific principle focusing on the illocutionary forces of utterances: applied to the pretzel case, (LDS) entails that if I am (at t _n ) permitted to deny that I performed a promise and to claim that I performed a prediction, and (at t _n ) I do so deny and claim, then (at t _n+1 ) the scoreboard will not reflect that I performed a promise. In contrast, the constitution view, applied to the pretzel case, implies that that whether my initial utterance is a promise can be constituted by Leo’s “uptake” of the utterance as a promise. That is, the constitution view seems to imply that in uttering “Bullshit!” and therein refusing to accept my subsequent second-order denial and claim, Leo exercises an ability to preclude my attempt to change the scoreboard. But Leo does not seem to have this ability, and, more to the point, the proposal that he does have this ability contradicts the above articulated (LDS) principle. Since (LDS) is integral to the illocutionary norms account of deniability, and the constitution view is incompatible with what is entailed by this principle, I conclude that the constitution view is incompatible with the illocutionary norms account of deniability.

A similar argument demonstrates that the speaker-directed view is incompatible with the illocutionary norms account of deniability. The fundamental assumption that a speaker who makes an utterance therein performs an illocutionary action Φ iff the information that the speaker has therein performed Φ is recorded on the conversational scoreboard also implies that conversational scorekeeping is governed by a limited deference to audience (LDA) norm:

(LDA) If at t _n the scoreboard reflects that a previous utterance is open to being interpreted as the performance of either Φ or Ψ (where Φ ≠ Ψ), and that the audience refuses to accept that the utterance was the performance of Φ, then at t _n+1 the scoreboard will not reflect that the speaker performed Φ.

(LDA), like its complement (LDS), is formulated in terms of illocutionary actions performed by speakers in making utterances, and thus (LDA) also entails a more specific principle focusing on the illocutionary forces of utterances: (LDA) entails that if Leo is (at t _n ) permitted to refuse that I performed a prediction, and (at t _n ) he does so refuse, then (at t _n+1 ) the scoreboard will not reflect that I performed a prediction. In contrast, the speaker-dependent view, applied to the pretzel case, implies that in performing my subsequent second-order denial and claim, I exercise an ability to bring about not only that my initial utterance is not a promise, but moreover that my previous utterance - despite Leo’s refusal - is a prediction.³² But, again, I do not seem to have this ability and, again more to the point, the proposal that I do have such an ability contradicts an entailment of (LDA). Since (LDA) is integral to the illocutionary norms account of deniability, and the speaker directed view is incompatible with what is entailed by this principle, I conclude that the speaker-directed view is also incompatible with the illocutionary norms account of deniability.

The upshot is that the constitution and speaker-directed views are incompatible with the assumption that a speaker who makes an utterance therein performs an illocutionary action Φ iff speaker and audience mutually accept that in making the utterance the speaker performs Φ. Since this assumption is fundamental to the illocutionary norm account of deniability, this account of deniability, to the extent that it is plausible, provides evidence against these views and in support of McDonald’s (2021) collaboration view. Advocates of the constitution or speaker-directed view could of course reject the fundamental assumption, and thus by implication also reject the illocutionary norm account of deniability.³³ Though I cannot thoroughly address the issue here, it is important to appreciate that rejecting the fundamental assumption would come at a theoretical cost, because it is motivated by the conception of illocutionary actions (invoked in section 2) as defined in terms of normative status. McDonald explains this motivation:

According to the collaboration theory, force itself is … a matter of interlocutors acquiring a distinctive set of normative statuses, … determined by agreement … . This is intuitive, because often our reasons for believing we have acquired new normative statuses as a result of performing an illocutionary act include beliefs about what other people believe. If I believe I have made a promise, for example, I believe I have acquired a new obligation. And I believe that I have acquired this obligation (a) because I believe that the hearer believes I have that obligation and (b) because I believe that the hearer has acquired (and believes she has acquired) an attendant entitlement. If I attempted a promise, but came to believe that you did not hear me do so, or did not interpret what I said as a promise, it’s not clear I would still consider myself obligated to do the action I was attempting to promise to do.” (2021, p. 8).

The fundamental assumption shown to be at odds with the constitution and speaker-directed views is thus motivated by Sbisà’s insight - which she attributes to Austin (1962) - that illocutionary actions just are “the active production of deontic states by social agents, [and] the key to such production is intersubjective agreement” (2014, p. 629).

Finally, it must be acknowledged that, despite the above sketched motivation, the collaborative view is problematic from an ethical point of view. Bianchi objects that the collaboration view allows that “speakers appearing to have performed harmful speech acts may simply reject the audience’s interpretation of their utterances as harmful even when this interpretation is perfectly reasonable. In this way, hate speakers can unfairly reject responsibility for their objectionable speech acts” (2023, p. 8). I have three brief responses to Bianchi’s ethical objection. First, Bianchi’s objection conflates - perhaps on purpose³⁴ - the descriptive, theoretical, project of explaining what the determinants of illocution force are, and the prescriptive, practical, project of proposing what the determinants ought to be. Here I have been concerned only with the theoretical explanatory project. Second, the collaboration view has the problematic ethical consequences that Bianchi describes - as well as other problematic ethical consequences³⁵ - only in cases where the speaker makes a harmful utterance that is open to illocutionary interpretation. A speaker who made a harmful utterance that is closed to illocutionary interpretation and therein performed a harmful speech act, and then subsequently denied having performed that harmful speech act, would be guilty of both performing a harmful illocutionary action and illocutionary gaslighting.³⁶ Finally, it should be appreciated that, if we stipulate that only utterances open to interpretation are under consideration, the ethically problematic consequence described by Bianchi - viz. “speakers appearing to have performed harmful speech acts may simply reject the audience’s interpretation of their utterances as harmful even when this interpretation is perfectly reasonable” - just is a kind of deniability. Bianchi is correct that the collaboration view allows for deniability, and I believe it is widely agreed that this phenomenon is ethically problematic. But that the collaboration view allows for, and partially explains, this ethically problematic phenomenon hardly constitutes a theoretical reason to reject the collaboration view.

References

• AUSTIN, J.L. How to Do Things with Words. Edited by M. Sbisá and J. O. Urmson. Oxford: Clarendon Press, 1962.
• BACH, K., HARNISH, R. M. Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press, 1979.
• BIANCHI, C. “Discursive Injustice: The Role of Uptake”. Topoi 40, pp. 181-190. 2021,
• ______. (2023). Varieties of Uptake. In L. Caponetto & P. Labinaz (eds.) (2023), pp.75-95.
• CAMP, E. “Insinuation, Common Ground, and the Conversational Record”. In D. Fogal, D. Harris, and M. Moss (eds.) (2018), pp. 40-66.
• CAPONETTO, L., LABINAZ, P. Sbisà on Speech as Action. Cham: Palgrave-Macmillan, 2023.
• DINGES, A., ZAKKOU, J. “On Deniability”. Mind, Vol. 132 . 526 April, pp. 372-401, 2023.
• FOGAL, D., HARRIS, D., MOSS, M. New Work on Speech Acts. Oxford: OUP, 2018.
• FRASER, B., TURNER, K. Language in Life, and a Life in Language: Jacob Mey - A Festschrift. Bingley: Emerald, 2009b.
• FRICKER, E. “Stating and Insinuating”. Proceedings of the Aristotelian Society S86(1), pp, 61-94, 2012.
• HARRIS, D. “Intentionalism versus the New Conventionalism”. Croatian Journal of Philosophy, 16/47, pp. 173-201, 2016.
• ______. “Intention and Commitment in Speech Acts”. Theoretical Linguistics, 45/1-2, pp. 53-67, 2019.
• HASLANGER, S. Resisting Reality. Oxford: OUP , 2012.
• KUKLA, R. “Performative Force, Convention, and Discursive Injustice”. Hypatia 29(2), pp. 440-457, 2014.
• LEE, J., PINKER, S. “Rationales for Indirect Speech: The Theory of the Strategic Speaker”. Psychological Review, 117(3), pp. 785-807, 2010.
• LEWIS, D. “Scorekeeping in a Language Game”. Journal of Philosophical Logic, 8, pp. 339-59, 1979.
• MACKINNON, C. Feminism Unmodified. Cambridge MA.: Harvard University Press, 1987.
• MCDONALD, L. “Your Word against Mine: The Power of Uptake”. Synthese, 199 (1-2), pp. 3505-3526, 2020.
• ______. “Reimagining Illocutionary Force”. Philosophical Quarterly, Vol. 7 (4), pp. 918-939, 2021.
• SBISÀ, M. “Uptake and Conventionality in Illocution”, Lodz Papers in Pragmatics 5(1), pp. 33-52, 2009a.
• ______. “Illocution and Silencing”. In B. Fraser and K. Turner (eds.) (2009b), pp. 351-357.
• ______. “The Austinian Conception of Illocution and Its Implications for Value Judgments and Social Ontology”. Etica & Politica / Ethics & Politics, XVI, 2, pp. 619-631, 2014.
• STRAWSON, P. “Intention and Convention in Speech Acts”. Philosophical Review, 73, pp. 439-460, 1964.

• 1
  Other common strategies are to speak euphemistically, or evasively.
• 2
  In Austin’s (1962, p. 22) Ψ might be a mere “etiolation.” An example described by Fricker (2012, p. 83): Republican Senator Jon Kyl uttered “If you want an abortion, you go to Planned Parenthood, and that’s well over 90 percent of what Planned Parenthood does.” When informed that the content of his utterance was false, Kyle subsequently claimed his utterance “was not intended to be a factual statement.”
• 3
  A duplicitous speaker can do various things to render this sort of claim plausible: in making her utterance, she can mumble, or speak in a noisy environment. Or she can utter a sentence that is in some way ambiguous, or use pronouns with no clear referents, etc. All of these are ways of allowing that what locutionary action she performs is open to interpretation.
• 4
  I focus on this third category because it is directly relevant to the issue addressed in the final section, the issue of the determinants of illocutionary force.
• 5
  “‘Looting’ Comment from Trump Dates Back to Racial Unrest of the 1960s” by Michael Wines, NYTs, May 29, 2020.
• 6
  The post was preceded by a proclamation from Twitter (now X) stating that Trump’s post “violated the X rules about glorifying violence.”
• 7
  I eschew the phrase ‘plausible deniability’ because I want to allow that a speaker has deniability regarding her illocutionary action even if her denial of having performed that action is implausible.
• 8
  If we were to accuse Trump of speaking, or tweeting, duplicitously, he could deny it, and could claim that he spoke, or tweeted, only sloppily. Duplicitous speech thus involves a sort of meta-deniability.
• 9
  What are the “typical cases”? In the cases of “typical” plausible deniability discussed by D&Z, there is no question as to what locutionary act the speaker performs.
• 10
  Note that there is a risk of vacuity: if what it is to speak directly is by definition to speak in a way that does not allow for deniability, then (D2) and (D3) are vapid analyticities. Lee and Pinker (2010) seem to be the source of this use of ‘indirect speech’.
• 11
  I set aside the frustrating fact that all utterances seem to be open to interpretation regarding their sincerity. In Austin’s terms, it seems that every utterance is open to an etiological interpretation.
• 12
  The notion of an utterance being open (closed) to illocutionary interpretation is anticipated by Strawson:
  The meaning of a (serious) utterance, as conceived by Austin, always embodies some limitation on its possible force, and sometimes - as, for example, in some cases where an explicit performative formula, like "I apologize," is used - the meaning of an utterance may exhaust its force; that is, there may be no more to the force than there is to the meaning; but very often the meaning, though it limits, does not exhaust, the force. Similarly, there may sometimes be no more to say about the illocutionary force of an utterance than we already know if we know what locutionary act has been performed; but very often there is more to know about the illocutionary force of an utterance than we know in knowing what locutionary act has been performed. (1964, pp. 439-440).
• 13
  D&Z present two additional secondary desiderata, but they are problematic, and irrelevant to my concerns.
• 14
  Where I have ‘she [performed] Φ’ D&Z have ‘she meant to Φ’. That the former is what is needed becomes apparent if one considers cases where a speaker makes an utterance and therein performs an illocutionary action that is closed to interpretation, but the illocutionary action therein performed is other than the action she intended to perform. Suppose you are playing blackjack in Vegas and you clearly utter ‘hit’ to the dealer. You thereby perform the action of hitting - indicating that you are to be given another card - even if your intention was to utter ‘hold’ and thereby perform the action of holding. If you now sincerely deny that you intended to hit, the dealer ought not act on the belief that you intended to hit, but you nonetheless do not have deniability regarding having hit.
• 15
  D&Z endorse a stronger version of the epistemic account that makes epistemic propriety require knowledge. My counterexample also undermines the stronger epistemic account.
• 16
  If the reader does not view my utterance as to a sufficient degree open to illocutionary interpretation, she is encouraged to flesh out the scenario, or create her own, so that it is.
• 17
  Those familiar with debates over proposed accounts of, e.g., freedom, or justification, or presupposition, are aware of how confusing and confused, such debates can become.
• 18
  D&Z might claim that the big pretzel case is not an instance of deniability, but is instead an instance of what they call “untouchability” (p. 394). Roughly, a speaker has untouchability regarding Φ iff the practical benefit for the audience of allowing the speaker to “get away with” having performed Φ makes it irrational for the audience to “challenge their denial” (p. 394). Note, however, that in the big pretzel case Leo explicitly rejects my denial as “bullshit.”
• 19
  D&Z raise precisely this “challenge” (2023, p. 384) for any alternative to their epistemic account, but the way they raise the challenge ignores that the speaker’s denial (and claim) is a second- order speech act whose direct illocutionary effect is to change the context in which the exchange takes place:
  … if deniability is consistent with the audience knowing [that the speaker had performed ], then why would deniability be desirable? If the audience knows [the speaker had performed ], they can blame her for this, they can tell others about this, and they can generally let their actions be guided by this. So what does the [speaker] gain from retaining deniability? (2023, p. 384)
• 20
  Sbisà, explaining the insight she attributes to Austin (1962), states that “illocutionary effect is a change in these aspects of the interpersonal relation, which I call “deontic” inasmuch as they are connected with what one can or cannot or should or should not do” (2014, p. 623). Sbisà proposes that Austin’s (1962) categories of illocutionary actions can thus be categorized in terms of such deontic roles:
  • Verdictives: license to act upon the judgement (for addressee) vs commitment to truth or correctness (for speaker)
  • Commissives: right to expect a certain kind of behaviour from the speaker (for addressee) - commitment to perform (for speaker)
  • Exercitives: obligation + (possibly) rights and powers (for addressee) vs commitment to support the addressee’s deontic state (for speaker)
  • Behabitives: license to act upon the speaker’s expressed state (for the addressee) vs satisfaction of a task or debt (for the speaker)
  • Expositives: rights, obligations, etc. as above, affecting relations internal or relevant to discourse or conversation. (2014, p. 624).
• 21
  The line between permissible deniability and illocutionary gaslighting is fuzzy. As Camp explains,
  …conversations also vary in the stringency of the operative standard of reasonableness, in at least two ways. First, the more there is a known likelihood of conflict or other motivation for strategic interpretation, the more reasonable it is to be actively on one’s interpretive guard. And this in turn widens the scope of deniability (and cunning pedantry), by increasing the range of alternative assumptions one must be able to rule out to eliminate a putative alternative interpretation. Second, interlocutors’ willingness to push the bounds of reasonable reinterpretation depends on their willingness to bear larger social costs. A speaker who expects future interactions with H to be quite limited may be more prepared to offer an alternative interpretation M(Q)′ at the outer bounds of admissibility; while a speaker who is concerned to preserve the relationship, or their own reputation going forward, may be less inclined to invoke minimally credible reinterpretations. (2018, 51).
• 22
  Of the course the mere fact that speaker has engaged in mutually salient behavior may have effects on the scoreboard; but such developments would not occur in virtue of the illocutionary force of denial.
• 23
  My explanation relies on the conversational norms LDS and FIG, and these norms clearly depend upon the fact that interpreting open-to-interpretation utterances is epistemically dodgier than interpreting closed-to-interpretation utterances. It is important to appreciate, however, that in my explanation of how the big pretzel case can naturally and unproblematically be understood as possessing the feature described in (D2-3*), there was never an appeal to a change in Leo’s epistemic status concerning the proposition that I promised to buy him a pretzel. As is implied by their endorsement of (D4), D&Z themselves recognize that in the specific case of my uttering ‘I will buy you a big pretzel’ Leo can know that I therein made a promise, despite the general epistemic dodginess of interpreting such open-to-interpretation utterances. I am grateful to an anonymous reviewer for urging clarity on this point.
• 24
  Deniability has a positive complimentary notion: A speaker S has claimability regarding an illocutionary action Φ iff S is permitted to claim having performed Φ, and if S claims to have performed Φ and S’s claim is accepted by the audience, then S acquires responsibility for having performed Φ. Note that a speaker felicitously claiming to have performed a praiseworthy action Φ does not suffice to make it appropriate is for the addressee to hold the speaker responsible for having performed Φ; only if the addressee and speaker mutually accept the claim, and the scoreboard is changed to reflect the claim, will it be appropriate for the addressee to hold the speaker responsible for having performed the action. Though complimentary notions, there is thus an asymmetry between claimability and deniability, because a speaker’s being appropriately held responsible for performing an illocutionary action requires mutual acceptance that she performed that action, whereas being free of such responsibility requires only a lack of mutual acceptance that she performed the action.
• 25
  Strawson’s example concerned a speaker playing bridge: “A player might let slip the word ‘redouble’ without meaning to redouble; but if the circumstances are appropriate and the play strict, then he has redoubled (or he may be held to have redoubled)” (1964, p. 557).
• 26
  The point is demonstrated in Austin’s (1962, p. 23) example of a misfiring naming ceremony: no matter how earnest he may be, a random revolutionary who smashes a bottle across the bow of ship while yelling ‘I hereby dub this ship Generalissimo Stalin’) will fail to name the ship.
• 27
  Ancestors of Bianchi’s (2023) speaker directed view include Bach and Harnish (1979) and Harris (2016), (2019).
• 28
  Kukla (2014) seems to endorse a very strong version of the constitution view: “speech acts have their performative force only in virtue of the concrete social difference that they make, or how they are taken up in practice” (2014, p. 4, my emphasis). Kukla (forthcoming), however, states that “the possible constitutive influence of uptake is typically quite narrow” (forthcoming, p. 8).
• 29
  In her influential (1993), Langton defends a “ratification” view, which is similar to McDonald’s collaboration view. According to McDonald (2021, p. 10) the difference between these views is that the ratification view does not allow the speaker to develop or amend her original illocutionary intentions in light of the audience’s response to her utterance, whereas the collaborative view does allow for this.
• 30
  As McDonald notes, Sbisà (2014) interprets Austin (1962) as endorsing something like the collaborative view. The most telling passage from Austin is this:
  Unless a certain effect is achieved, the illocutionary act will not have been happily, successfully performed. This is not to say that the illocutionary act is the achieving of a certain effect. I cannot be said to have warned an audience unless it hears what I say and takes what I say in a certain sense. An effect must be achieved on the audience if the illocutionary act is to be carried out. How should we best put it here? And how can we limit it? Generally the effect amounts to bringing about the understanding of the meaning and of the force of the locution. So the performance of an illocutionary act involves the securing of uptake.” (1962, pp. 115-116)
• 31
  And what determines who is included in the audience for an utterance? And what if different members of the audience interpret the utterance as having distinct forces? Sbisà (2014), McDonald (2021), and Bianchi (2023), Kukla (forthcoming) explore such issues, but they are beyond what I can address here.
• 32
  Bianchi characterizes the speaker-directed view as follows: “a speaker succeeds in performing a certain speech act if she has arranged things in such a way that a competent, attentive, and unbiased audience can reasonably be expected to recognize her illocutionary intention. If for some arbitrary reason (“the perversity of one’s audience”) they do not, the speaker still counts as having performed the speech act” (2021, p. 8). In denying that I made a promise and claiming that I made a prediction, I clearly put Leo can reasonably be expected to recognize my intention to use ‘I will buy you a pretzel’ to make a prediction.
• 33
  One theorist’s modus ponens … .
• 34
  Bianchi (2021, p. 6) suggests that she is engaged in a Haslanger (2000) inspired project of ameliorative analysis, though the concept being ameliorated seems to be discursive justice, and not illocutionary action.
• 35
  In particular, the collaborative view allows for various forms of what Langton (1993), following MacKinnon (1987) called silencing.
• 36
  Note that on the collaboration view both speaker and audience can be guilty of illocutionary gaslighting: If I had uttered ‘I hereby predict that I will buy you a big pretzel’ and Leo stubbornly refused to accept that I had made a prediction and insisted that I had made a promise, he would be guilty of illocutionary gaslighting.
• Article info
  CDD: 410
• Funding:
  No funding agency.

Publication Dates

• Publication in this collection
  17 Feb 2025
• Date of issue
  2025

History

• Received
  21 June 2024
• Reviewed
  03 Oct 2024
• Accepted
  16 Dec 2024