Illocutionary pluralism

Abstract

This paper addresses the following question: Can one and the same utterance token, in one unique speech situation, intentionally and conventionally perform a plurality of illocutionary acts? While some of the recent literature has considered such a possibility (Sbisà, in: Capone, Lo Piparo, Carapezza (eds) Perspectives on pragmatics and philosophy. Springer, Cham, pp 227–244, 2013; Johnson in Synthese 196(3):1151–1165, 2019), I build a case for it by drawing attention to common conversational complexities unrecognized in speech acts analysis. Traditional speech act theory treats communication as: (1) a dyadic exchange between a Speaker and a Hearer who (2) trade illocutionary acts endowed with one and only one primary force. I first challenge assumption (2) by discussing two contexts where plural illocutionary forces are performed in dyadic discussions: dilemmatic deliberations and strategic ambiguity. Further, I challenge assumption (1) by analyzing poly-adic discussions, where a speaker can target various participants with different illocutionary acts performed via the same utterance. Together, these analyses defend illocutionary pluralism as a significant but overlooked fact about communication. I conclude by showing how some phenomena recently analyzed in speech act theory—back-door speech acts (Langton, in: Fogal, Harris, Moss (eds) New work on speech acts. Oxford University Press, Oxford, pp 144–164, 2018) and dog-whistles (Saul, in: Fogal, Harris, Moss (eds) New work on speech acts. Oxford University Press, Oxford, pp 360–383, 2018)—implicitly presuppose illocutionary pluralism without recognizing it.

1 Unity and plurality of speech action

When closing the document in which I am writing this paper, I am prompted to choose from among three actions: “save” the recent changes (and close), “don’t save” the changes (and close), “cancel” the idea of closing and return to the document to write this introduction. These three are mutually exclusive, jointly exhaustive, clearly defined, monofunctional actions that I can undertake in this context. But what is sufficient for communication with a machine cannot be a model of how our (speech) actions function in a social world. It is something of a staple in philosophy that human actions are complex. Actions can simultaneously cause and constitute different things, often in intricate chains. During an official department meeting, my flipping a switch causes a red light to go on, which constitutes a vote against Donald, which, in turn, makes him feel insecure about his next term in office, while pleasing other contenders; Gilbert touches the keys of the piano, thus playing the piano, composing his “Hungarian Rhapsody,” and consoling his grieving mother. All the same, so the argument goes, this complexity consists in the plurality of correct descriptions of an action, not plurality of actions themselves: there is one thing the agent does, which can then be analytically divided into its parts or aspects, from flexing a muscle to firing Donald.¹

Austin’s theory of speech as action can be seen as an original take on these ideas applied to communication. When we speak, we produce sounds, construct sentences, and infuse them with meanings; we produce, in Austin’s words, locutionary acts. At the same time and eo ipso, we also perform one of the various illocutionary acts: conventionally recognizable social acts, such as promising or announcing a verdict. Finally, to perform these two is also to perform a perlocutionary act, that is, to make our listeners feel, think, or do certain things, such as reassuring or enraging them. Yet again, there is unity in this plurality: while what interlocutors trade are thus complex actions, namely, “total speech acts in total speech situations” as Austin has it, there is one and only thing a speaker does in terms of the crucial, illocutionary aspect of each particular speech act. We can either promise, or predict, or bemoan something in saying “I will never play piano again,” but we cannot, primarily and to the same extent, do more than one of these things at the same time.²

The goal of this paper is to show that this assumption is not universally correct: we often can and in fact do perform plural illocutionary acts. This has not gone unnoticed. Recently, Johnson (2019) has identified troubling challenges to illocutionary monism, the assumption that illocutionary acts are monofunctional, and Sbisà (2013) has entertained the possibility of illocutionary pluralism, the idea that they can under certain conditions be plurifunctional. Here, I build on these criticisms to produce a more general positive account of how illocutionary pluralism functions thanks to salient features of communicative context, primarily, the plurality of hearers the speaker engages through her utterances. This is different from what I call illocutionary relativism, which instead assumes pluralism results from each hearer’s different uptake, as suggested by both Sbisà and Johnson. I further show that illocutionary pluralism is in fact tacitly assumed in many of the recent analyses of phenomena such as back-door speech acts (Langton, 2018) and dog-whistles (Saul, 2018).

Let me start with a simple example illustrative of the phenomenon in question.

• (1) Consider the following dialogue fragment among three friends out on a picnic:

(1.1)	Ann, to Barbara and Chris:	What’s the time?
(1.2)	Barbara:	Chris has a watch.
(1.3)	Chris:	Three thirty.

Ann’s initial question/request in (1.1) has clearly two hearers: Barbara and Chris. We can then identify the target of her speech act in terms of an (elided) attributive vocative:

• (1.1) (The one of you who has a watch, [please tell me],) what’s the time?

The attributive vocative only makes sense in the context of multi-party conversation. By means of it, a speaker selects from among the multiple hearers the one(s) with an attribute necessary for an act to succeed: The one of you who speaks Portuguese, tell me what she’s saying; The one of you who has the key, lock the door please.³ Given Ann’s open intention to request the time information from anyone present in a position to provide it, Barbara volunteers to relay the request to the target with proper attributes, Chris:

• (1.2) Chris has a watch.

A simple question is: Which illocutionary act has Barbara performed? More precisely: what is the (a) intended and (b) conventionally recognizable (c) primary illocutionary force of Barbara’s utterance in (1.2)? These qualifications are necessary, as they allay a number of obvious concerns. In particular: Condition (a) blocks against an analysis in terms of illocutionary relativism whereby unintended illocutionary forces are nevertheless ascribed to her utterances by hearers (see Sbisà, 2013; Johnson, 2019); it also excludes the distal, perlocutionary effects of the utterance. Condition (b) blocks against the contextual ambiguity analysis whereby the illocution has contingent features which make it indeterminate or ambiguous in a specific context (see Searle, 1975). Finally, condition (c) blocks against the indirect speech act analysis (see Searle, 1975 and below).

In (1.2), Barbara, of course, performed an assertion, a truth-relevant description of reality. In this type of context, however, this assertion can be further seen as an act of conversationally appointing / nominating an agent with the required attributes—it constitutes a fitting response to an attributive vocative, on a scale of cooperativeness only short of being the lucky and responsive possessor of the attribute. Overall, this seems to be a conventionally recognizable form of collaborative social behavior—“I can’t help you, but I know who can: here she is!”—as attested by many similar examples⁴:

(2)
(2.1)	Ann, to dinner participants:	I need just 2 euros.
(2.2)	Barbara:	Chris has change.
(2.3)	Chris:	Here you go.

(3)
(3.1)	Ann, to dinner participants:	Can you tell me what he’s saying?
(3.2)	Barbara:	Chris speaks Portuguese.
(3.3)	Chris:	He’s saying…

(4)
(4.1)	Ann, to dinner participants:	How much is 220 divided by 14?
(4.2)	Barbara:	Chris has a calculator.
(4.3)	Chris:	Almost 16 euros, let’s make it 17 each with a tip.

Given this, an indirect speech act analysis would answer the question above in the following way:

• Level 1 (secondary force): By way of an assertion (assertive): “Chris has a watch.”

• Level 2 (primary force): Barbara is performing a conversational appointment (declaration).⁵

Intuitively, however, this doesn’t quite exhaust the illocutionary acts performed by Barbara at turn (1.2). Similarly to Ann, she has two hearers to address, who seem to be on a par: Ann by virtue of (1.2) being a response to her request in (1.1) and Chris by virtue of being explicitly appointed as the sought-for attributee. Thus, we can distinguish another level where Barbara by way of her appointment does illocutionarily at least two further things:

Level 3:

• (1) Apologizes to Ann (expressive) with a justification (assertive) for why she is not able to fulfill her request.⁶

  (Apologies Ann, I cannot tell you because I don’t have a watch, but at least I can help you by telling who has, namely:) Chris has a watch.

• (2) Requests Chris (directive), to tell Ann the time, which he does.

  (Hey Chris, given you’re the one) [Who] has a watch (, tell Ann the time, please)

These speech acts at Level 3, even if primary, are indirect speech acts and, as such, require some inferential procedure. Despite extended glosses proposed here, this procedure conforms to the original tests devised for indirect speech acts by Searle (1975): they address the (non)fulfilment of some felicity conditions; here, the preparatory conditions for apologies and requests. Barbara’s indirect apology to Ann points out her inability to tell the time and thus to fulfil Ann’s original request (see fn. 6). Barbara’s indirect request to Chris makes at this conversational juncture directly salient Chris’s ability to do so instead. Under proper circumstances, assertions such as “You’re tall” or “You speak Portuguese” are indirect requests to reach or translate something by virtue of their expressing the key preparatory condition for requests: the hearer’s ability to do the thing requested. Given the multi-party context here, as already explained, these assertions are further conversational appointments, since the ability is attributed by a third party, rather than the original requester. Finally, Barbara’s utterance [also in (2.2), (3.2), and (4.2)] seems to be a move conventionally recognizable as an assertion-appointment-apology/request in informal collective coordination situations described in examples 1–4.

Let me conclude this analysis with two basic take-away points. In one simple utterance (Chris has a watch), we see a plurality of illocutionary acts that spreads over two axes. On the vertical axis we can distinguish at least three levels of a by-way-of illocutionary chain, where a direct assertion conveys an indirect appointment which, in turn, conveys an indirect apology and request.⁷ On the horizontal axis, at one and the same level (here: level 3) a speaker can still intentionally and conventionally perform more than one illocutionary act, namely, a justified apology and a request.

These facts about conversation will guide my theoretical arguments for speech act pluralism understood as illocutionary pluralism (Sect. 2). More in particular, while the vertical plurality has a long tradition in speech act theory under the concept of indirect speech acts (Searle, 1975), the horizontal one doesn’t. In what follows, I will try to explain why it doesn’t, and to argue that it should (Sect. 3). The argument will primarily hinge on a shift of attention from a dyadic to a polyadic conversation as an unmarked context for speech act exchange. In multi-party polylogues, a speaker may release various recognizable illocutionary acts to different hearers (Sect. 4.2). However, also in simpler one-on-one dialogues, a speaker may intentionally perform numerous illocutionary acts for her hearer to consider (Sect. 4.1). Either scenario thus justifies the case for illocutionary pluralism.

2 Speech act pluralism as illocutionary pluralism

It is important to clarify that the speech act pluralism discussed here amounts to (horizontal) illocutionary pluralism: intentional performance of a plurality of conventionally recognizable illocutionary acts via the same utterance token in one unique speech situation. As already mentioned, Austin famously distinguished between three levels or aspects of “the total speech act in the total speech situation” (1962, p. 147): locution (e.g., She said “x”); illocution (e.g., She argued that x); and perlocution (e.g., She convinced me that x) (Austin, 1962, Lectures VIII–IX). The crucial concept is that of illocution—indeed, in the subsequent literature, some scholars (esp. Searle, 1969, 1976), straightforwardly identify speech acts with illocutionary acts. Above, I have accordingly focused on the question of which illocutionary act Barbara performed.

This sounds clear enough, but the way things are, the term “speech act pluralism” is most commonly used for locutionary pluralism (Cappelen, 2011; Cappelen & Lepore, 2005; Seymour, 2010). The chief idea is that the same speech act can express an indefinite number of propositions, most importantly, a minimal proposition (fixed, grounded in literal meaning of non-indexical expressions) and a maximal proposition (contextually variant and open to enrichment, most obviously in the case of indexicals). This basically extends the classic discussion of semantic underdetermination and is an argument supporting semantic minimalism versus contextualism, to the effect that it is not the case that a specific context determines the contextually unique propositional meaning of a specific utterance.⁸ In her commentary on Cappelen & Lepore’s discussion of semantic speech act pluralism, Sbisà notices that “underdetermination invites interactional negotiation and selection by the audience, whereas plurality aims at multiple recognition and is confirmed by it” (Sbisà, 2013, p. 240). The specifically illocutionary aspect of speech act pluralism I am after here thus differs from locutionary pluralism in two respects: it pertains to the illocutionary force central to speech act theory and is genuinely pluralistic as it aims precisely at multiple recognition.

Finally, one sentence on the possibility of seeing speech act pluralism in terms of perlocutionary pluralism. That it exists is trivially obvious per definition of (distal) perlocutionary effects (Austin, 1962; Sbisà, 2007, 2013) and doesn’t seem to need any defense.⁹

3 Illocutionary monism and dyadic reduction

So what’s exactly the problem? Well, the problem is that the classic speech act theory doesn’t seem to care much about plural, multi-functional illocutionary acts performed in the context of multi-party conversation, such as in our examples (1)–(4). Instead, it offers an image of communication reduced to “two great heroes,” Speaker and Hearer, who trade mono-functional illocutions (Searle, 1992, p. 7). Worse still, as further argued by Searle, speech act theory doesn’t and even shouldn’t care much about conversation at all—too amorphous a concept to grasp. This explains an almost exclusive focus on the analysis of single utterances, and thus on the speaker’s end of the communicative process, with the hearer mostly needed to secure uptake or produce response when conventionally invited to (Austin, 1962, pp. 115–116). When some regimented forms of conversation—e.g., argumentative discussions (van Eemeren & Grootendorst, 1984)—are studied, illocutionary acts are taken to be performed in a neat dyadic exchange built of speech act pairs (adjacency pairs). A Speaker first infuses her utterance with a determined intention, which is then recognized by Hearer, who in return produces a fitting utterance. Questions are responded to with answers, answers with doubts, doubts with arguments, arguments with counterarguments, and, step-by-step, emerges a dialogue, where the felicity conditions for speech acts in pairs are nicely dovetailed.

This image is grounded in two crucial and interrelated assumptions that are broadly and unreflectively accepted:

• (1) Illocutionary monism: each speech act has basically a unique primary force or function, something to be recognized and responded to appropriately.

• (2) Dyadic reduction: conversation or communication can be fully grasped by a model consisting of two and only two interlocutors (Speaker and Hearer); other forms of multi-party conversation are derivatives of it.

In a recent critique of assumption (1), Johnson (2019) defines illocutionary monism as an approach to speech acts that posits “a unique force fact for each utterance.” She specifically singles out Searle as an illocutionary monist, with a special proviso for his concept of indirect speech acts, whereby an indirect but primary force is performed by way of a direct but secondary force. Still, Searle “assumes that there is a single order of illocutionary forces” as he “is committed to there being at most one primary force, at most one secondary force” (Johnson, 2019, pp. 1153–1154).

Regarding (2), more elaborate critiques have been furnished by a number of scholars for some time now (Goffman, 1981; Levinson, 1988). However, the gist is that “the standard [speech act] theories say nothing about illocutionary acts directed at hearers other than the addressees” (Clark & Carlson, 1982, p. 341), while such acts clearly exist, as shown in examples (1)–(4) above.

Such critiques would very well serve my argument here—if only they were somewhat more correct. More precisely: they correctly expose both the illocutionary monism and the dyadic reduction as theoretical assumptions, but at the expense of literal and historical detail. To see this, let me engage some early speech act work. In his 1969 monograph, Searle observes this:

Both because there are several different dimensions of illocutionary force, and because the same utterance act may be performed with a variety of different intentions, it is important to realize that one and the same utterance may constitute the performance of several different illocutionary acts. There may be several different non-synonymous illocutionary verbs that correctly characterize the utterance. For example suppose at a party a wife says “It’s really quite late”. That utterance may be at one level a statement of fact; to her interlocutor, who has just remarked on how early it was, it may be (and be intended as) an objection; to her husband it may be (and be intended as) a suggestion or even a request (“Let’s go home”) as well as a warning (“You’ll feel rotten in the morning if we don’t”). (Searle, 1969, pp. 70–71; emphasis added)

As is well known, Searle later took on and meticulously analyzed the problem of “several different dimensions of illocutionary force,” namely, in his discussion of the taxonomy of illocutionary acts, where no less than twelve such dimensions (or differences between various acts) are distinguished (Searle, 1976; see Searle & Vanderveken, 1985, for further refinements). However, how a variety of speaker’s intentions can be formed to perform several different illocutionary acts in the context of—or shall we rather say, for the sake of—multiple and differentiated hearers has never become an issue. Not only that, the presence of multiple and differentiated hearers, while clearly salient, is not even acknowledged as a possible explanans here (as it would be in the “for the sake of” case).

Austin’s theory—earlier and in a more complex way—also clearly recognizes various forms of speech act plurality (1962; see Sbisà, 2013). As already mentioned, Austinian speech acts include three irreducible aspects—locution, illocution, and perlocution—and further within locution itself phonetic, phatic, and rhetic acts. Moreover, and significantly for my argument, Clark and Carlson (1982, p. 340ff.) have also remarked that Austin’s first and most classic examples of performatives—marrying someone, christening (a boat, a baby), or bequeathing a watch—in order to be felicitous, all necessarily require some “institutional witnesses”: a public official, naval officers, a priest, notary public, etc., all of which are hearers other than direct addressees. While none of this constitutes illocutionary pluralism the way I elaborate it here, it provides some of its key building blocks. If “the total speech act in the total speech situation is the only actual phenomenon which, in the last resort, we are engaged in elucidating” (Austin, 1962, p. 147), then we are best advised to properly grasp the totality, both of the speech act and of the situation.

The upshot of it is that, contrary to some critics such as Johnson, the founding figures of speech act theory do question, even if in a somewhat parenthetical or nascent way, the tenets of illocutionary monism and of the dyadic reduction. Nonetheless, they do not ever seriously pursue pluralism—and in particular pluralism related to multiplicity of participants—in their investigations. Why?

One key reason is that illocutionary monism offers a powerful and parsimonious assumption that there is a single, unique illocutionary force fact for each speech act (Johnson, 2019). It has intuitive appeal—when we do something, we typically do one determinate thing—and lets us avoid vague, ambiguous, complex, or otherwise tangled force attributions. Pluralism would thus need to have substantial theoretical advantages to enter the view—such as its capacity to account for empirical phenomena monism struggles with. But this can be doubted by claiming that illocutionary pluralism and multi-party exchanges are simply not common. Normally, we trade unique forces in dyadic exchanges and the examples adduced here are somewhat fanciful. This, of course, is an empirical question, and one that requires the kind of evidence we should be very unlikely to obtain. Yet, empirical analysts of language do challenge the philosophical scheme of conversation as a dyadic business between a Speaker and a Hearer: “Even if such a scheme is intended to be a model, for descriptive work it cannot be” (Hymes, 1972, p. 58; see Levinson, 1988; Kerbrat-Orecchioni, 2004). This challenge should, at least, be seriously addressed with some argument, empirical or otherwise.

One possible non-empirical argument is that pluralism, especially in the context of multi-party conversations is, in the end, not important: it is a contingent feature of context or a variation easily explainable from within the dyadic model. Austin (1962, p. 15) mentions “the particular persons and circumstances” as background conditions for a felicitous procedure. In this way, possible illocutionary pluralism related to multiple participants is relegated to idiosyncratic contextual circumstances. Also, as already mentioned, Austin’s acute awareness of speech act pluralism is channeled instead to the distinction of various levels or aspects of speech acts, most notably locution, illocution, and perlocution (Sbisà, 2013).

Searle resorts instead to the concept of indirect speech acts to account for the undeniable cases of illocutionary dualism, that is, cases where one and the same utterance in one and the same context expresses two illocutionary forces: Can you pass me the salt?, while being a question, functions primarily as a request. There is thus a certain hierarchy: a primary illocutionary act (here: a request) is performed “by way of” a secondary illocutionary act (here: a question), which, in turn, is performed “by way of uttering a sentence the LITERAL meaning of which is such that its literal utterance constitutes a performance of that illocutionary act” (Searle, 1975, p. 62). Because of this chain of “by way of” acts, I call this a vertical speech act plurality.

Again, there are nuances of the indirect speech act approach directly relevant to any analysis of illocutionary pluralism (Lepore & Stone, 2018; Sbisà, 2013), not least the uptake of the hearer which typically—even in the case of most conventional, idiomatic expressions—can be non-defectively related to the literally encoded secondary force, or both forces at once:

(5)
(5.1)	Can you pass me the salt?
(5.2)	Yes… Ooups, well, actually, I cannot, John just took it. Sorry.

Here, the first part responds to the direct question, while “sorry” is an apologetic response to the indirect request. Given that this is a well-researched topic, I will not delve any further into these nuances. Instead, I will briefly mention two extensions of the indirect speech act plurality. The first of them further complicates the vertical axis of plurality while the second introduces a horizontal axis, something that I think should be considered as the illocutionary pluralism proper. This discussion will pave the way for the argument that illocutionary pluralism is not only a common but also a worthwhile phenomenon for philosophers to investigate.

Analysts of ordinary conversations such as Levinson (1981, 2013) have noticed that the limitation of indirect speech acts to just two levels is somewhat arbitrary. Under proper circumstances, responses such as (5.2′) or (5.2″) would be perfectly in order.

• (5.2′) You’re not getting us another tequila, are you!?

• (5.2″) It’s pretty salty, actually, have you tried yet?

Hearers can thus also respond to perlocutionary objects (see Austin, 1962, p. 117) of given speech acts—or, indeed, to broader action sequences they may (rightly or wrongly) assign to speakers. These sequences can involve further, chained illocutions: in 5.2′, for instance, the respondent seems to be pre-empting an upcoming invitation to have another round. Levinson thus claims that speakers orient to entire “conversational projects” of their interlocutors, rather than merely to individual utterances typically studied in speech act theory. While this is a flexible and sophisticated approach to illocutionary plurality, far exceeding the limits of Searle’s conventional approach to indirectness, it is still a vertical model. It is based on a projection of various illocutionary forces (and, further, perlocutionary objects) linked in a “by way of” manner to the literally uttered act through some kind of a sequential hierarchy.

Another approach to illocutionary pluralism, which I have already called horizontal, has been proposed by Clark. Clark (1992) and Clark and Carlson (1982) distinguishe between Searle’s “linear” indirect speech acts (vertically chained illocutions, as described above) and “lateral” indirect speech acts that function by virtue of targeting hearers other than the direct addressee. Consider the following example (Clark & Carlson, 1982, p. 364):

(6)	Ann, to Barbara, in front of Charles, David, and Ewan: Barbara, I insist that Charles tell you the joke about the two Irishmen.

In Clark & Carlson’s analysis, Ann performs a direct but secondary assertive to Barbara, indirect but primary request to Charles, as well as possibly indirect but primary warning to David (who hates jokes about Irishmen or Charles’s jokes, and is duly warned, e.g., you don’t want to hear this, so you better go get a beer now, David). We can further imagine that due to a particular arrangement between Ann and Ewan (you go and prepare the birthday cake when I ask Charles to tell a joke), this can be a command to Ewan. To account for such cases, Clark takes up an idea hinted at by Austin (1962, p. 6), namely, that each illocution, apart from its primary force, also serves to inform one’s hearers of what’s being performed. In this way, “the speaker performs two types of illocutionary act with each utterance. One is the traditional kind, such as an assertion, promise, or apology; this is directed at the addressees. The other, called an informative, is directed at all the participants in the conversation—the addressees and third parties alike” (Clark & Carlson, 1982, p. 332).

Two critical points on that. First, this approach unduly limits the primary illocutionary acts to the addressee. As is clear in Clark & Carlson’s example (6), such acts might just as well be performed to the non-addressed participants. Second, the solution is unwieldy and prone to an Occam’s razor objection. While it does reveal speech act pluralism in multi-participant conversations, it resorts to a disposable layer of illocutionary force. Do we really need to read: I hereby inform that I ask you what the time is?

Instead, in what follows, I offer a simpler solution by drawing attention to conversational details speech act theory has overlooked due to its unduly adherence to the assumptions of illocutionary monism and dyadic reduction.

4 Plural illocutionary acts in dilogues and polylogues

The argument so far runs as follows: Speech act pluralism can be understood in terms of locutionary, illocutionary, or perlocutionary pluralism. The focus here is on illocutionary pluralism, given that locutionary pluralism belongs to a different debate, namely one over semantic underdetermination and contextualism, while perlocutionary pluralism is obvious per definition of perlocution. I have further distinguished between two basic forms of illocutionary pluralism: vertical and horizontal pluralism. The focus here is on horizontal illocutionary pluralism, given that vertical pluralism is well-established via the notion of indirect speech acts and also conversational projects.

But even within so delineated horizontal illocutionary pluralism, we can still distinguish two main classes: (1) illocutionary pluralism in a dyadic exchange (a di-logue), based on a multiple ascription of on-a-par illocutionary forces to one and the same utterance by the speaker and/or the hearer; and (2) illocutionary pluralism in a multi-party exchange (a poly-logue), whereby different communicatively intended and conventionally recognizable illocutionary forces are directed to different audience members. I will discuss them in this order.

4.1 Illocutionary pluralism in a dyadic exchange

This option challenges the assumption of illocutionary monism by admitting the possibility of a speaker performing many different speech acts, which are all on a par and all of which are addressed to the very same hearer. Perhaps it requires somewhat more contrived examples, thus depriving my argument of the naturalness I claim for the polylogical cases below, but it is a distinct possibility to be recognized. I distinguish two prototypical scenarios where this could happen: dilemmatic deliberations and strategic ambiguity.

As a preliminary, genuine plurality needs to be distinguished from illocutionary underdetermination or vagueness. Plurality means that a speaker is more or less openly ambiguous about the multiple recognizable forces she intends to convey. Underdetermination instead implies the speaker herself is not in full command of the illocutionary forces of her utterance and so further conversational work needs to be done to determine at least one force. Admittedly, from a consistently externalist approach to speech acts, these two processes might look very much alike, leading to interactional negotiations (Sbisà, 2013; Witek, 2015a, b) which, in somewhat fraught cases, might take the form of an illocutionary game of hearer’s pedantry (“Did you just threaten me?”) and speaker’s denial (“No, no, I was just suggesting it would be better if you…”) (see Camp, 2018).

4.1.1 Dilemmatic deliberations

Dilemmatic (or even trilemmatic, etc.) deliberations are situations in which agents face a number of solutions to a practical problem, none of which is prime facie decisively better than others. In a conversational situation, they might want to convey this predicament resorting to illocutionary pluralism, even if addressing but one hearer.

To show how it works, I use a stylized version of a case which Witek (2015b, p. 28; see also Strawson, 1964, p. 444) employed to contrast two different illocutionary forces (warning vs. encouragement) of the same locution in two different contexts. In our case, the context is fixed: two biologists are collecting samples from under the surface of a frozen polar lake. Thin, transparent ice means more precise data can be collected. But thin ice is also perilous. So the scientists are facing one of the many data collection dilemmas (real good data are hard to come by). Additionally, the head of the project, John, explicitly advised them (was it just an advice?) in the last project meeting to always consult him if they face some dilemma: walking over thin ice or otherwise. While sitting in their cabin-cum-lab, the following exchange emerges:

(7)
(7.1)	Biologist 1: Shall we go out to collect samples?
(7.2)	Biologist 2: The ice is thin.

Which illocutionary act is Biologist 2 performing? All the difficulties of actual face-to-face performance notwithstanding (the right intonation, facial expressions, gestures), she might be simultaneously performing three indirect but primary acts (the direct but secondary act is of course an assertion):

• (7.2′) Encouraging data collection (sure, we have great conditions, near transparent thin ice…)

• (7.2″) Warning of the dangers of data collection (we better watch out, the ice is “so thin that it can break under the weight of a human body,” see Witek, 2015b, p. 28)

• (7.2″′) Deferring the answer to the boss, implicitly resorting to “conversational appointments” presented in examples (1)–(4) (“John knows the procedures!”)

This might seem utterly inconsistent in terms of “illocutionary logic,” owing to conflicting rights and obligations necessary for a felicitous performance of each of these three acts. Yet, Biologist 2 might precisely acknowledge this in her effort to intentionally communicate three possibly incompatible forces as a reflection of a genuine trilemma they find themselves in: (1) We have a window of opportunity to get what we have been waiting for for so long now; (2) But it’s a tricky undertaking with ice that thin; (3) And perhaps we should first consult the boss (but then time is of essence, we might just miss the chance).¹⁰ As such, this plural act might be a prelude and an invitation to a collective deliberation over the issue, as is relevant given the practical nature of the initial question: Well, hard to say, let us weigh pros and cons and decide, perhaps “call[ing] in others to aid us in deliberation on important questions, distrusting ourselves as not being equal to deciding” (Aristotle, 1984, Nicomachean Ethics, 1112b11).

What is initially happening primarily depends, however, on the “hearer’s uptake” (Austin, 1962, p. 117). Biologist 1 might respond:

• (7.3′) I’m glad you’re so thrilled too! Let me get my coat and off we go!

• (7.3″) Well, you’re right, we better wait a day or two.

• (7.3″′) Ohh, “the thin ice” situation, we better call John.

Further rounds of illocutionary negotiations might follow (Biologist 2: “No, no, I actually wanted to warn against it, rather than encourage you, sorry!” or “OK, I have my doubts, but let’s go if you are so thrilled!”). But one distinct possibility is for Biologist 1 to respond:

• (7.3″″) How do you mean? Should we stay or should we go now? Or call John?

To which Biologist 2 might respond:

• (7.4) Here’s the thing: I’m not quite sure, what do you think?

In this way, by simultaneously putting three incompatible illocutionary acts in view (encourage, warn, defer), a speaker might overtly signal an action trilemma, thus opening the space for joint deliberation over the issue. Given that an identical assertion (“The ice is thin”) is performed as a direct but secondary speech act, this possibility relies entirely on illocutionary pluralism at the indirect but primary level of speech.¹¹

4.1.2 Strategic ambiguity

Openly dilemmatic deliberations are different from strategic illocutionary ambiguities, used in the service of broader diplomacy or otherwise. Here’s an autobiographic example of this.

As I was about to defend my PhD dissertation, a professor from the University of X visited our department. Being a department chair at his university, and a primary investigator of research projects, he was, among other things, in the business of informally scouting for potential young researchers to employ. This, I’m assuming, was openly in our common ground. We met briefly to discuss his projects, my work, and a workshop at his school I submitted a paper to and was awaiting a response. At the end of the meeting, he shook my hand confidently and said:

(8)
(8.1)	Professor: I’m sure I’ll see you soon in X, then!

Again, which primary illocutionary act did he perform? Resorting to my (historical, by now) actual hearer’s authority, I can think of at least four reasonable options, organized from the weakest to the strongest:

• (8.1′) Prediction: Warranted by a presumptive but defeasible rule along the lines “grad students from your school typically get their papers accepted.”

• (8.1″) Reassuring praise: “A guy like you surely deserves to visit us, at least as a workshop presenter, but who knows, maybe even a project member.”

• (8.1″′) Promise: “I’ll see to it that you come visit us, at least as a workshop presenter, but who knows, maybe even a project member.”

• (8.1″″) Job offer: “When I return to X I’ll start the procedures, and I’m sure I’ll see you there soon!”

All of these were reasonable guesses—and I still think they are. Moreover, I think the basic rationale of the professor’s speech act was precisely to maintain this ambiguity, thus keeping the space of illocutionary potential—and the related hopes, entitlements, and obligations—wide open.¹² Of course, I could have taken him to task for it, e.g., by resorting to “cunning pedantry” (see Camp, 2018):

• (8.2′) Me: You mean you’ll be opening a position soon?

Or even:

• (8.2″) Me: You mean I have a job?

But which grad student on a job market would dare to do so in an informal context like that?

The professor’s strategy thus seems to be that of “strategic ambiguity,” as described by communication scholars: “Particularly in turbulent environments, ambiguous communication is not a kind of fudging, but rather a rational method used by communicators to orient toward multiple goals” (Eisenberg, 1984, pp. 238–239). His multiple goals and corresponding multiple illocutionary acts were driven by the basic rationales of (academic) diplomacy, or any strategic communication at all. One of them is to limit one’s on-record commitments to a minimum, while expanding the space of plausible deniability (see Camp, 2018; Peet, 2015). Another, and perhaps more curious, is to create and keep open for the addressee the “space of possibilities for normatively appropriate uptake” (see Lance & Kukla, 2013, p. 268). So, similarly to a dilemmatic deliberator, an illocutionary diplomat wants her addressee to navigate the space of illocutionary possibilities. Yet, she does so not in order to jointly examine them, but rather to leave the addressee in a state of fragile but hopeful ambiguity, allowing him to hope for the best while fearing the worst. The fragility of hearer’s hopes is then aptly matched by the fragility of speaker’s commitments: one can easily imagine the professor telling my supervisor over dinner: “I hope he didn’t get the impression we would hire him, did he?”.

Now, while this mechanism seems to be working via insinuation, whereby the hearer is invited to suspect p, but without the speaker’s commitment to p,¹³ this form of illocutionary pluralism reveals that the sharp division of speech acts into either “wholly overt and essentially avowable” or “plausibly deniable” seems to be a false dichotomy. Diplomacy is sustained by the very possibility of “neither confirming nor denying” (NCND) some questioned acts or commitments. Of course, one thing is to avoid confirming some fact (e.g., for national security reasons) while not being legally capable of denying it (in fear of perjury, for instance) (the so-called “Glomar response”). Another thing is to use the NCND disclaimer as a diplomatic tactic for opening up the pluralistic space of recognizable intention and thus of acceptable uptake—at least in the form of illocutionary potential that is not avowedly denied—and see how the communicative dynamic develops from there. For the illocutionary diplomat, this dynamic is hoped to reshape the normative space of mutual rights and obligations in a way beneficial to her, under any possible downstream contingency. For the interlocutor of the diplomat, it gives the chance to respond without the sense of being trapped in a binary yes/no uptake or having to risk a “defiant response” from outside the space of appropriate uptake (e.g., “Who do you think you are to tell me this!”, see Lance & Kukla, 2013). Thus strategic illocutionary ambiguity is not merely a trick of a sly manipulator—it is a conventionally recognizable way of mutually handling tricky situations where the stakes are high.

In this way, both strategic ambiguity and dilemmatic deliberations do not merely reveal the very existence of illocutionary pluralism, but also indicate its key communicative jobs speakers and hearers might wish to pursue.

4.2 Illocutionary pluralism in a polylogue

This form of illocutionary pluralism challenges the assumption of illocutionary monism by exposing the dyadic reduction. It assumes that the natural context of human communication are multi-party encounters and that humans have developed communicative competence for handling such encounters (see Tomasello, 2008). In multi-party encounters, illocutionary pluralism becomes a natural fact of communication: we are well capable of conveying various speech acts, along with their locutionary, illocutionary, and perlocutionary aspects, to various hearers, via one and the same utterance.

One way of understanding illocutionary pluralism is illocutionary relativism, grounded in various ascriptions of force by various audience members. Indeed, for Sbisà, “the audience’s uptake (against a background of a multiplicity of illocutionary indicators, including textual strategies) seems to play a central role in allowing for illocutionary plurality” (Sbisà, 2013, p. 238; see Johnson, 2019).¹⁴ So for instance:

(9)
(9.1)	Can you do it by tomorrow?

Can in one and the same context be taken by some hearers as a genuine question, by others as a request, and yet by others as a thinly veiled instruction or even order. All this might happen unbeknownst to the speaker, with all the intricate dynamics of accommodating (or challenging) the presupposition of authority on the part of various hearers (Kukla, 2014; Langton, 2015, 2018; Witek, 2015a, 2019), as well as of corrective deniability on the part of the speaker (Camp, 2018).

What I discuss here is instead illocutionary pluralism proper, where different communicatively intended illocutionary forces are directed at different audience members with the goal of producing different “conventional effects” for each of them.¹⁵ That is, the speaker is ready to stand by the illocutions conveyed, and the hearers’ subjective interpretation of them has no decisive role to play. Consider a version of (9.1), in the context where I meet with a co-PI of our joint project and a post-doc employed in the project. We have an important deadline to meet, and so I say to both of them:

• (9.1′) Can you do it by tomorrow?

Let’s assume “you” is plural you and that it’s obvious to all three of us this is not an innocently informative question. Which speech act have I performed? Following a standard account of felicity conditions, this should be a request to my academic peer, the co-PI, but, given the institutional authority I have over him, an instruction to our post-doc, although a politely worded one. I would expect each of them to take it up in this differentiated way, as indicated by the fact that I would not easily take “no” for an answer from the post-doc (isn’t he paid to do just that?) but, with good justification (“tomorrow I really have to submit my revisions to Synthese, sorry!”), I would excuse my co-PI. Once this differentiated uptake is secured, distinct conventional effects would ensue: my co-PI is doing me a favor (or justifying why she can’t), while the post-doc is under obligation to deliver. This is the type of pluralism I have in mind here.

To elaborate this type of pluralism, a different notion of the basic context of conversation is needed. I see this notion in the concept of polylogue (Lewiński, 2017a, 2019). Following its Greek etymology, poly-logos signifies discourse (λόγος) between many (πολύ). The concept uses the same etymological resources as mono-logue, discourse of a single person. But our common dialogue might be confusing here: dia-logos means “through” discourse and as such comprises all interactive uses of language. Based on the number of speakers, dia-logues are a genus that can be then divided into the species of: di-logues (2 speakers), tri-logues (3 speakers), tetra-logues (4 speakers), etc. Poly-logues are thus all dia-logues which are not di-logues, that is, those that involve three or more speakers.

While the numbers of speakers are significant for how conversation develops, many further characteristics define polylogue, a term occasionally used by others who analyze many-to-many communication.¹⁶ From a traditional perspective of speech act theory or conversation analysis, conversation goes wild. The very basic notions—relevance, coherence, sequential organization, adjacency pairs—are challenged. Instead of even briefly describing the empirical results of conversational characteristics of polylogues, let me illustrate three of them directly relevant to the study of illocutionary pluralism.

One such characteristic are problems in determining the relevance and completeness of exchanges. To see the difference, compare a simple question of A asked to B, C, and D in examples (10) and (11):

(10)
(10.1)	A: What time is it?
(10.2)	B: Three thirty.

(11)
(11.1)	A: What would you like to drink?
(11.2)	B: Red wine.

In a dyadic exchange limited to A and B these are entirely equivalent, and in this case felicitous and complete exchanges. However, in a polylogical context we clearly see the difference: (10.1) is a collective question/request, where one collective action or even one individual action of whichever hearer (B, C, D) constitutes a satisfying and complete response. By contrast, (11.1) is a distributive question/request, where an individual action of each hearer is needed to complete the exchange (unless, of course, B is the group’s spokesperson authorized to speak on behalf of C and D, or C and D are already standing with glasses of caipirinha, etc.). In the distributive case, de facto three individual speech acts of the same kind (here: requests) to three different individuals are thus performed—which constitutes the simplest form of illocutionary pluralism proper, as delimited above.¹⁷

Two further characteristics of polylogues relevant here are: multiple-recipient design (conveying different messages to different participants) and difficulties in establishing and updating the common ground, as well as “the state of play,” among all the participants.¹⁸ To see how these two are interrelated in enabling illocutionary pluralism, consider the following case:

• (12) A group of friends in a restaurant, about to order desserts after dinner:

• (12.1) Ann, to Barbara, Chiara, Daniel, and Edward, after consulting the waiters: At this hour, they only have chocolate mousse!

Ann’s simple assertion, when heard by everybody, updates the public score in this five-participant polylogue. However, we can also easily imagine that this assertion is merely a secondary speech act, which serves as a vehicle to convey a number of indirect but primary speech acts. For instance, considering that:

• A, B, and C also shared their love of chocolate mousse before, so that it’s their shared ground—A performs a recommendation, or perhaps some joyful expressive;

• A and D discussed D’s chocolate allergy: in their shared ground, this would be a warning or even an apology;

• A had an argument with E in which E claimed the restaurant never serves chocolate mousse at a late hour: in their shared ground, this is a refutation.

So, we have an evolving public score common to all participants, and shared scores limited to subsets of participants. These shared scores can be either: (1) inclusively shared with others as proper subsets of the common public score: so everybody mutually knows that Daniel is allergic to chocolate and acknowledges the fact that Ann’s assertion primarily counts as a warning or apology to him; or (2) exclusively shared, in the sense of being limited to a subgroup of participants—so in Clark & Carlson’s joke example (6) it should be only David who’s warned of Charles’s bad jokes and in (12) it could well be only part of Ann and Edward’s “game” that he was refuted.

Now, the crucial argument to be made here is the following: It is this multiplicity of simultaneously evolving scores that allows for illocutionary pluralism proper in a polylogue: with one and the same locution we can advance different illocutionary moves in different illocutionary games played with different subsets of participants.¹⁹ Further, as already mentioned, two crucial conditions for such illocutionary moves are that: (1) via various hearers’ uptake, they are expected to take distinct conventional effect; (2) the speaker intends his intention to illocute x, y, z to different speakers to be recognized as intended: by the exclusive subset of participants, by the inclusive subset, or by all the participants involved. These plural illocutionary forces are thus not only conventionally recognized but also intentionally performed—in the sense of reflexive communicative intentions (Bach & Harnish, 1979; Grice, 1989; Strawson, 1964). This is different from covertly intending to deceive an eavesdropping spy or concealing some information from an overhearing child: in such cases, our intention works only in cases where our intention is not recognized.

The multiplicity of simultaneously evolving language games with their respective scores has recently attracted attention of philosophers such as McGowan (2019). She too claims that “a single act can also simultaneously be a move in two (or even more) different norm-governed activities” (2019, p. 96). But there are important differences. For McGowan, there is only one illocutionary move performed via overt communicative intentions in a specific conversational game. Other acts are instead “parallel acts” enacting score changes in other parallel games, even without a speaker’s (overt or covert) intention to do so. Moreover, these parallel games are typically not conversational games, but larger games of social interaction.²⁰ For instance, a patient who addresses a male nurse, “Doctor, can you please tell me…” performs an illocutionary act of request, but also a parallel, non-illocutionary act of reinforcing the expectation that men should not be nurses (McGowan, 2019, p. 96).

By contrast, polylogues are complex conversational activities where one overarching game (e.g., a dinner of a group of friends) includes various sub-games. These sub-games can be either publicly functional elements of the bigger game (inclusive score) or privately functional side events affecting only a sub-group of players (exclusive score). Given this complexity, each conversational move, while being a single locutionary act, can alter various parallel sub-games, and to do so via simultaneously performed, fully-fledged illocutionary acts, with all the overt, reflexive communicative intentions in play. Moreover, these parallel illocutionary acts are conventional moves, collectively and by default recognized as such.

5 Back-door speech acts and dog-whistles as instances of illocutionary pluralism

Illocutionary pluralism in the context of polylogues provides a simple yet fertile theoretical framework for better accounting for certain details of pragmatic phenomena of communication, especially those related to complex, even manipulative, communicative strategies. As a matter of brief illustration, let me focus on two such phenomena recently analyzed by analytic philosophers.

5.1 Langton’s back-door speech acts

The first of them is Langton’s concept of back-door speech acts: “low profile speech acts, enabled by presuppositions and their ilk, that tend to win by default” (Langton, 2018, p. 146). According to Langton—who herself resorts to Austin’s (1962) original take on speech acts—back-door speech acts work via the unexpressed presuppositions of the “front-door” speech acts. These presuppositions, when tacitly accommodated by hearers’ default uptake, enable the unexpressed back-door illocutions to be felicitously performed. Langton’s main concern is how to block these presuppositions to prevent them from being tacitly yet successfully accommodated (Lewis, 1979; Witek, 2015a, 2019), and hence to prevent the attempted back-door speech acts from being felicitous. Her analysis of how back-door speech acts function derives from an overheard exchange during a football game in Melbourne (Langton, 2018, p. 145):

(13)
(13.1)	St. Kilda supporter to sluggish player: Get on with it, Laurie, you great girl!
(13.2)	Alert bystander: Hey, what’s wrong with a girl?
(13.3)	St. Kilda supporter: It’s got no balls, that’s what’s wrong with it!

The supporter’s riposte makes it clear, says Langton, that his original use of “great girl” is a form of a gender slur, thus opening a wide, although underhand, “illocutionary potential” of the yell:

Several back-door speech acts were performed, in the utterance of, ‘Get on with it, you great girl!’ The speaker’s main ‘front-door’ purpose was presumably to urge a sluggish player, and express frustration. But regardless of his aims, there were several back-door speech acts achieved by ‘great girl’, and what it presupposed in that context. The utterance implicitly ranked women, a verdictive speech act, in Austin’s scheme. It implicitly testified that there is something wrong with a girl, in that context, using potentially informative presupposition. It implicitly legitimated broader norms that say e.g., ‘men take charge, women are gentle and obliging’. The bystander was among the hearers, though not a ‘hearer’ in the usual sense: he was not being addressed, and the speaker hardly knew he was there. But, through his intervention, he became a proper hearer, a recognized party to the conversation. The bystander tried to block what was presupposed […] (Langton, 2018, pp. 146–147; emphasis in the original)

In crucial respects, Langton’s analysis comes as close as possible to the notion of illocutionary pluralism, as discussed here. It maintains one utterance can carry with it a horizontal plurality of simultaneously performed speech acts, understood as illocutionary acts. Further, it does recognize that part of the speech act plurality, and sneaky plurality for that, results from the interplay between the addressed hearers and unaddressed bystanders. While ostensibly the illocutionary objective of the supporter-qua-speaker is to urge the player-qua-hearer, it would be analytically and socially ingenuous to stop at that interpretation. Other speech acts act on bystanders, and, in this case, these are precisely the kind of devious back-door speech acts one needs to be alert to.²¹ Assuming illocutionary monism in this context would not be only an analytic but also, and primarily so, an ethical failure: accepting an abuser’s excuse—“I was just venting my anger, that’s it”—would let him off the hook. Assuming illocutionary pluralism lets us, by contrast, see and object to, via “blocking” or otherwise, the multiplicity of other attributable speech acts, even if an expletive expressive is undeniably part of the story.

An important question in need of deeper investigation is the exact nature of back-door acts, and other similar phenomena such as insinuation, generics, dog-whistles, and conversational exercitives (see Camp, 2018; McGowan, 2019; Saul, 2017, 2018). For instance, are acts such as legitimizing a (discriminatory) norm really meant to be recognized as having this force, and produce this conventional illocutionary effect by virtue of the recognition? Langton would say no, and yet she is clear that back-door speech acts are illocutionary acts, acts which carry a certain illocutionary potential that can be actualized as illocutionary success unless the presuppositions-cum-felicity conditions are blocked by hearers or bystanders.²² In this way, she deviates from a basic insight of Austin and his commentators who stress that “the illocutionary force of an utterance is essentially something that is intended to be understood”: it involves a wholly overt, audience-directed intention that is intended to be recognized as such (Strawson, 1964, p. 459). This deviation requires a brief examination, so as to avoid a simple fallacy of equivocation when discussing Langton’s take on back-door speech acts as an instance of illocutionary pluralism.

One good point of contrast is the work of Bach and Harnish (1979) who also attend to “sneaky presuppositions,” a phenomenon central to Langton’s analysis. According to them, these would be “covert collateral acts.” Without going into the details of their entire theoretical framework, collateral acts are “conversational acts performed in conjunction with or in lieu of illocutionary acts” (Bach & Harnish, 1979, p. 97). Collateral acts, while being “acts of speech,” are not speech acts with a recognizable illocutionary force (see Green, 2017 for the distinction): a common feature of all such collateral acts is their suspension of the communicative presumption, namely, the presumption that when a speaker says something, she is saying it with a recognizable illocutionary intent (Bach & Harnish, 1979, p. 7). Among these, Bach & Harnish distinguish between overt collateral acts such as joking, storytelling, or reciting; covert collateral acts such as innuendo, deliberate ambiguity, and sneaky presuppositions; as well those that are neither overt nor covert, such as changing the subject.²³

Now, the crucial idea is that while a genuine illocutionary intent is nowhere to be found in any of the collateral acts (one is not asserting anything, just telling a joke), overt acts work only when the intention to, for instance, tell a joke is recognized. By contrast, “covert collateral acts succeed (the intention with which they are performed is fulfilled) only if their intent is not recognized, or at least not recognized as intended to be recognized” (Bach & Harnish, 1979, p. 101). They work precisely because they lack the wholly overt, reflexive intentions defining illocutionary acts. Their example—“Fortunately, the CIA is no longer involved in political assassinations” (Bach & Harnish, 1979, p. 102)—thanks to its semantic presupposition trigger (“no longer”) would be amenable to the kind of back-door speech act analysis proposed by Langton. The speaker is not merely stating a fact or expressing a relief (front-door speech acts), but also, context permitting, accusing the CIA of murderous activities in the past, testifying that there is something wrong with the USA government, perhaps bemoaning the hypocrisy of “the free world,” etcetera (back-door speech acts). But can these work as illocutionary acts? Bach & Harnish would say no, as these can only do their communicative business when the speaker’s intentions are not recognized (covert acts) and, moreover, these are not illocutionary intentions in the first place (they are merely collateral acts of speech).²⁴

Langton must be aware of this, and that’s why she needs an account where illocutionary acts can be performed via tacit presupposition accommodation (Langton, 2015, 2018; see Lewis, 1979; Witek, 2015a, 2019). Neither a speaker’s intentions nor hearers’ or bystanders’ active, explicit uptake is crucial here. Instead, back-door speech acts become felicitous “by default,” that is, by implicitly conforming to certain requirements of a conversational game in question. A similar mechanism governs what lawyers call “the presumption of legality”: in lieu of an explicit proof of every detail of an act performed, we fill the necessary gaps presuming all is fine (e.g., it is a legitimate signature), others presume the same, and that’s how we get things done in the social world (see Lewiński, 2017b; Witek, 2019). Unless and until these default presumptions are blocked (e.g., there is evidence the signature was produced under duress), the act stands. And when they are blocked, the act is annulled or otherwise undone (Caponetto, 2020).

The crucial success condition for back-door speech acts is thus not their fulfilment of a speaker’s intention, whether overt or covert—the basic success condition for Bach & Harnish’s collateral acts—but rather the possibly tacit impact they have on the conversational score. Interestingly, McGowan’s (2019) “parallel acts,” including various conversational exercitives, function much in the same way, only that she is very clear these cannot be treated as illocutionary acts, precisely because they lack speaker’s communicative intentions mutually manifest to all participants. Is, then, Langton wrong in calling the plural acts she detects illocutionary acts? She isn’t, as long as we’re ready to accept a distinct, weaker sense of illocutions, where a speaker’s intention and the conventionality of the act are substituted with a sophisticated working of default presuppositions. Resorting to the grammar of adjectives, one can then understand back-door in a back-door speech act not as a straightforward intersective adjective, such as a commissive speech act, sparkling wine, or machine gun, which modifies a noun that remains in the same class (e.g., as a species of the genus). It is instead a non-intersective adjective like apple wine or alcohol-free wine, or toy gun, whereby the adjective modifies the noun to the extent that, while some of its central qualities are preserved, the noun does not necessarily instantiate the original class (e.g., because it doesn’t preserve all its original functions). (This is not to say it is anti-intersective, decisively moving the noun out of its original class.) The key point, however, is this: if my argument for illocutionary pluralism works, as I surely hope it does, for the traditional, strong concept of illocutionary acts grounded in the collective recognition of the plurality, it works so much the better for finer, weaker, and trickier forms of speech acts.

5.2 Saul’s dog-whistles

Saul’s (2018) account of dog-whistles is cognizant of such complications. She distinguishes between overt and covert dog-whistles. The sine-qua-non condition for communicating via overt dog-whistles is that there are two audiences, those in the know, and those in the dark. This is most obvious in the case of an overt intentional dog-whistle defined as:

a speech act designed, with intent, to allow two plausible interpretations, with one interpretation being a private, coded message targeted for a subset of the general audience, and concealed in such a way that this general audience is unaware of the existence of the second, coded interpretation. (Saul, 2018, p. 362, citing an unpublished work of Kimberly Witten)

Next to the ground common to the speaker and the general audience, for dog-whistles to work we also need a ground exclusively shared by the speaker with a subset of insiders. One example of that are strategic choices made by George W. Bush who “desperately needed the votes of fundamentalist Christians, and yet it was also clear that many others—whose votes he also needed for the general elections—were made nervous by fundamentalist Christianity” (Saul, 2018, p. 362). According to Saul,

The solution his speech-writers used was to dogwhistle to the fundamentalists. A nice example of this is Bush’s utterance in his 2003 State of the Union speech:

• Yet there’s power, wonder-working power, in the goodness and idealism and faith of the American people. (Noah 2004)

To a non-fundamentalist this is an ordinary piece of fluffy political boilerplate, which passes without notice. But a fundamentalist Christian will hear the dogwhistle. Amongst fundamentalists, “wonder-working power” is a favoured phrase that refers specifically to the power of Christ. (Saul, 2018, p. 362)

Whereas in this example Saul proposes something like a locutionary working of a dog-whistle, we can easily give it an illocutionary reading: while the general audience is offered an innocent statement, the insider audience gets a pledge of allegiance. That would be another common instance of illocutionary pluralism, characteristically resorted to by politicians vying for support of a diverse, heterogenous audience (cf. Bonevac, 2003; Zarefsky, 2008). Saul herself is happy with a conversational implicature or insinuation reading of overt intentional dog-whistles, but what is crucial in any case is that “the speaker intends her intention to be recognized, but without a willingness or responsibility to own up to it” (Saul, 2018, p. 371).

Now, Saul contrasts such overt dog-whistles with covert dog-whistles, more subtle uses of manipulative language that act on the hearers’ pre-existing attitudes—e.g., implicit racial prejudices—without the hearers’ being aware of it. These covert acts should therefore be understood as a species of perlocutionary speech acts, namely “covert perlocutionary acts.” The key point, as with Bach & Harnish’s covert collateral acts, is that “a covert perlocutionary act is one that does not succeed if the intended perlocutionary effect is recognized as intended” (Saul, 2018, p. 377).²⁵

To sum up: a polylogical communicative situation is a necessary condition for performance of what Saul calls “overt intentional dog-whistles” (and many others simply “dog-whistles,” as Saul’s original contribution is precisely to identify and examine the covert version). It thus needs to be presupposed in any treatment of dog-whistles. This will further characteristically generate illocutionary pluralism even if the basic mechanism of dog-whistles is built on the primary semantic or locutionary pluralism grounded in some strategic ambiguity. Illocutionary pluralism is also implicitly presupposed in the notion of back-door speech acts, if, indeed, these can be consistently defended as illocutionary acts. Some other phenomena of multi-party discourse, manipulative or not—covert dog-whistles, covert (conversational) exercitives, perhaps some of the back-door speech acts—will instead fall into the realm of perlocutionary pluralism; and this, one would assume, is no big news even in Austin’s original conception.

6 Conclusion

In their sketch of the “contemporary theoretical landscape” of speech act theory, Harris, Fogal, & Moss set out from the following claim: “Speech-act theory was born of a central insight: language is a medium for many kinds of action, but its superficial uniformity tends to mask this fact” (Harris, Fogal and Moss, 2018, p. 1). The recognition of the merely “superficial uniformity” of the de facto plural actions we do with words has taken speech act theorists to distinguish locutions from illocutions and perlocutions, and to closely analyze the enormous variety of illocutionary verbs and their associated forces. These analyses yielded results fruitful enough to trigger claims that “the basic unit of human linguistic communication is the illocutionary act” (Searle, 1976, p. 1).²⁶ Such statements convey an important message: in order to understand something about communication, we need to understand the workings of illocutionary acts as vehicles for a variety of social actions.

My particular way of driving this message home was to postulate and defend illocutionary pluralism: intentional performance of a plurality of conventionally recognizable illocutionary acts via the same utterance token in one unique, although typically complex, speech situation. Accordingly, one key argument of this study is this: If the aim of speech act theory is to understand the necessary and sufficient conditions for speech action, and the consequent strategic design of utterances by speakers as well as interpretative strategies of (various) hearers, then our theoretical toolkit needs to include the concept of illocutionary pluralism, whether in the context of a di-logue or a polylogue. Indeed, so conceived illocutionary pluralism seems to be already doing its invisible work behind various pragmatic phenomena of communication such as back-door speech acts and dog-whistles. No doubt there is much yet to discover.