Your word against mine: the power of uptake

Uptake is typically understood as the hearer’s recognition of the speaker’s communicative intention. According to one theory of uptake, the hearer’s role is merely as a ratifier. The speaker, by expressing a particular communicative intention, predetermines what kind of illocutionary act she might perform. Her hearer can then render this act a success or a failure. Thus the hearer has no power over which act could be performed, but she does have some power over whether it is performed. Call this the ratification theory of uptake. Several philosophers have recently endorsed an alternative theory of uptake, according to which the hearer can determine the nature of the act the speaker performs. According to this theory, if the hearer regards an utterance as illocutionary act y, then it is act y, even if the speaker intended to perform act x. Call this the constitution theory of uptake. The purported advantage of this theory is that it identifies a common but underanalysed way in which speakers can be silenced. I argue that despite its initial intuitive pull, the constitution theory of uptake should be rejected. It is incompatible with ordinary intuitions about speech, it entails a conceptual impossibility (the unintentional exercise of normative powers), and it has unsavoury political implications, entailing that marginalised speakers barely qualify as agents.

incited Christopher to murder the police officer or ordered him to surrender his gun. The former interpretation triumphed. Despite not firing the gun himself, Derek was charged with murder and in 1953 he was executed by hanging. 1 This case raises the question of who determines illocutionary force. Let us assume that Derek intended to order Christopher to surrender the gun, but Christopher interpreted Derek as inciting him to murder the police officer. Whose judgement is authoritative? I.e., who decides which act was performed? For Derek, this question was a matter of life or death. More generally, it is as much an issue of ethics as it is of philosophy of language. It has a bearing not only on debates about intentionalism and conventionalism, but also on how we conceive of the autonomy and normative powers of speakers. 2 In this paper I consider two ways of theorising the power of a hearer's 'uptake'. According to one theory, the potential illocutionary force of an utterance is determined by the speaker, who, in accordance with conventions, expresses a communicative intention to perform a particular illocutionary act. Her illocutionary act is successful only if her hearer recognises that intention. The hearer has no power over which act could be performed, but she does have power over whether it is performed. She can either ratify the attempted act or render it a failure. This entails that Derek attempted but failed to order Christopher to surrender the gun, because Christopher failed to recognise his intention. Call this the ratification theory of uptake.
We might think that this theory paints too rosy a picture of human communication, since it cannot accommodate the fact that sometimes hearers seemingly make it the case that speakers perform illocutionary acts which are different from the acts they intended to perform. It seems that Derek did not just fail to perform the act of ordering Christopher to surrender the gun, but rather performed a different act all together, due to Christopher's interpretation. Many people, especially marginalised speakers, seem to find themselves vulnerable to this phenomenon. For example, women who attempt to give orders often seem to end up making mere requests instead, due to hearers interpreting them as such.
A second theory of uptake has recently emerged which allows that hearers can make an utterance constitute an illocutionary act which is different from the illocutionary 1 In 1993, Derek Bentley won a posthumous pardon, and in 1998 his murder conviction was quashed. See R v. Derek William Bentley (deceased) [1998] EWCA Crim 2516. 2 I have simplified this case for rhetorical effect. I suggest that Derek could be interpreted as performing two different illocutionary acts-inciting Christopher to kill and ordering him to surrender the gun. Yet the difference between the two acts might be better parsed as a difference in locutionary content, rather than a difference in illocutionary force. Inciting and ordering are both, after all, exercitives. Under both interpretations Derek is telling Christopher to do something. The debate concerns what exactly he told Christopher to do. In addition, the debate in question did not take place within an ordinary conversation. It took place in a court of law, where some participants had greater interpretive powers than others, where several different notions of responsibility were at play, and where some participants had the power to reject speakers' accounts of their intentions. The rest of the paper is about illocutionary uptake in ordinary conversation, so we cannot use the verdicts reached on the Bentley case and the criteria used to reach them as a guide to making judgements about uptake in ordinary conversation. Nonetheless, I hope that this example makes it easier to imagine how we might find ourselves unsure about the illocutionary force of an utterance, and why such scenarios can be of moral and political importance. I am grateful to an anonymous reviewer for pushing me to clarify these issues. act the speaker intended to perform. 3 According to this theory, if the hearer interprets an utterance as if it is illocutionary act , then it is act , even if the speaker did not intend to perform act . Call this the constitution theory of uptake. In this paper I consider whether this newer constitution theory should replace the more established ratification theory.
In Sect. 1 I discuss the necessity and the phenomenology of uptake, before contrasting two ways of thinking about its power: the ratification theory and the constitution theory. In Sect. 2 I offer three arguments for why, despite its intuitive pull, we should reject the constitution theory. Firstly, it is incompatible with how we ordinarily think and talk about speech. Secondly, it is incompatible with the thought that illocutionary acts involve the exercise of normative powers; the constitution theory entails that illocutionary acts can be unintentional, yet one cannot exercise normative powers unintentionally. Thirdly, it has unsavoury political implications, entailing that marginalised people lack such a basic form of autonomy that they are barely agents at all. For these reasons, the ratification theory of uptake has the upper hand over the constitution theory.
1 The nature of uptake

The necessity of uptake
In ordinary language, 'uptake' usually means understanding, recognition, or acceptance. In his How To Do Things With Words (1976), J.L. Austin gave the word a more technical meaning, using it to describe the role of a hearer in a successful illocutionary act.
An illocutionary act is an act performed in speaking, which constitutively enacts changes in the normative statuses of the speaker and the hearer. It is different from a locutionary act, which is the act of uttering a sentence with a 'certain sense and reference' (Austin 1976, p. 109), and from a perlocutionary act, which is the production of psychological effects on one's hearer. If I utter the words 'I promise to mark your essay tomorrow', I perform a locutionary act in so far as I utter a sentence containing a meaningful proposition. I perform the perlocutionary act of making you feel reassured that you will get your mark soon. And I perform the illocutionary act of promising; I acquire an obligation to do what I promise to do, and my hearer acquires an entitlement to my doing what I promise to do.
Austin describes the role uptake plays in illocutionary acts as follows: 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. (1976, pp. 116-117).
Austin claims that uptake is necessary for a successful illocutionary act; for an illocutionary act to occur, a speaker must perform a particular kind of utterance and a hearer must respond to that utterance in a particular way. Not all speech act theorists agree with this. For example, Kent Bach and Robert Harnish consider it sufficient for illocutionary success that the speaker expresses the relevant communicative intention (more on which shortly). This expressive act may elicit uptake, but that uptake is not necessary for the act's success (1979).
William Alston purports to establish the non-necessity of uptake by using examples like the following (2000). He argues that if I say, 'Please could you bring me a towel', and you do not hear me, it still makes sense to say that I performed the illocutionary act of asking you to bring me a towel (2000, p. 24). I agree that it would not be unnatural to say, 'I asked you to bring me a towel, but you didn't hear me', but such a sentence is probably referring to a locutionary act, not an illocutionary act. The speaker is simply reporting that she uttered a sentence in the imperative mood. Whether this act requires uptake is a separate question from whether illocutionary acts require uptake. 4 Others argue that uptake cannot be necessary for illocutionary success because this would have unsavoury ethical consequences. For example, it entails that if a woman tries to refuse sex, but her attempted refusal does not receive uptake (either because her partner does not hear her or because he hears her but does not correctly interpret her), then she has not refused. 5 I agree that the necessity of uptake yields this result, but I do not think this is a reason to reject it. That the woman does not refuse reflects the sad reality that speakers from marginalised groups are often less able to communicate successfully. Yet we should resist the temptation to conclude that, since she does not refuse, her hearer's subsequent actions are permissible. Lack of refusal is not the same as consent. The woman has not consented and is therefore wronged when her partner proceeds with sexual activity. Moreover, he should have recognised her utterance as a refusal, and he wronged her by not doing so. We do not need to reject the necessity of uptake to explain the wrongdoing in this scenario.
I will henceforth assume that uptake is necessary for illocutionary success. The theories of uptake I will canvas therefore do not present a faithful picture of received speech act theory, since many speech act theorists deny the necessity of uptake. Of course, whether one believes uptake is necessary depends on how one understands uptake; questions about uptake's phenomenology and power and questions about its necessity cannot be cleanly separated. I start with the assumption that hearers play some kind of necessary role in the performance of illocutionary acts, and I will consider different ways of theorising that role. 6

The phenomenology of uptake
Here are two questions we could ask about uptake: 'What is it like for the hearer?' and 'What role does it play in illocutionary acts?' In this section I consider the first question, which concerns uptake's phenomenology. In Sect. 1.3 I consider the second question, which concerns uptake's power.
Austin's definition of uptake as the hearer's 'understanding of the meaning and of the force of the locution' (1976, p. 117) is not particularly illuminating. It is not clear what determines 'force', nor what makes it detectable to the hearer, nor exactly what Austin means by 'understanding'. I will now canvas two (not necessarily incompatible) ways of understanding the phenomenology of uptake: as the hearer's 'recognition' of the communicative intention expressed by the speaker, and as the hearer's 'recognition' of the conventionality of the speaker's utterance.
Uptake is most typically understood as the hearer's perception of or inference about the speaker's communicative intention. 7 To perform an illocutionary act, one must express a communicative intention to perform a particular illocutionary act. For example, when I make a promise, I express a communicative intention to promise. This intention has three components: (a) I intend to perform a promise, (b) I intend that my hearer recognise my intention to perform a promise, and (c) I intend to produce this particular response in my hearer by way of her recognition of my intention to produce such a response (i.e., I intend that my hearer's recognition of my intention in (b) makes her recognise my intention in (a)). Uptake, the thought goes, consists in my hearer's recognition of these intentions. Searle summarises this view as follows: In the case of illocutionary acts we succeed in doing what we are trying to do by getting our audience to recognize what we are trying to do. But the 'effect' on the hearer is not a belief or a response, it consists simply in the hearer understanding the utterance of the speaker. (1969, p. 47).
This view of uptake is widely endorsed, even by those who deny the necessity of uptake. Those who endorse it include Kent Bach andRobert Harnish (1979), Stephen Levinson (1983), Dan Sperber and Deirdre Wilson (1986), and Jennifer Hornsby (1995).
There are at least two ways of thinking about the psychology of this intention 'recognition'. It could be an event akin to the perception of high-level features in Footnote 6 continued interpret the speaker's utterance in the same way. Heterogeneous uptake presents an interesting challenge to all theories of uptake, but I lack the space to consider it here. 7 This account of uptake arose out of the combined efforts of Austin (1976), Peter Strawson (1964), Grice (1991), and John Searle (1969). Strawson combined insights from Austin and Grice by arguing that uptake (a notion from Austin) consists of the hearer's recognition of the speaker's expressed communicative intention (a notion from Grice). Searle then modified Grice's notion of communicative intention, and I use this modified understanding in the paper. Grice thought that we express a communicative intention to make a hearer come to believe a certain proposition. Searle noticed that intending to make a hearer have a belief is a perlocutionary, not illocutionary, goal. A speaker's illocutionary goal is to get her hearer to understand or recognise what act she is trying to perform (Searle 1969, p. 47). Searle therefore proposed that communicative intentions are reflexive intentions to perform a particular illocutionary act. Through these collaborative efforts, we end up with the account of uptake as the hearer's recognition of the speaker's communicative intention to perform a particular illocutionary act. visual experience; one instantly and automatically 'perceives' the speaker's intention. This might be facilitated by linguistic markers, like the use of performative verbs, which trigger recognition in the hearer. Or it could involve the hearer inferring the speaker's intention (see Bach and Harnish 1979;Sperber and Wilson 1986). These inferential processes are likely to be relatively automatic, and we might not always be aware we are engaging in them. Both perception and inference leave some room for error; sometimes we perceive things that are not there, and sometimes our inferential processes fail us. I discuss cases of misinterpretation in Sect. 1.3. 8 Alternatively (or additionally), uptake might consist in a hearer's recognition of the conventionality of a speaker's utterance, i.e. its satisfaction of the conventions attendant on a particular illocutionary act type.
Many speech act theorists believe that utterances must satisfy certain conventions, or 'felicity conditions', to constitute illocutionary acts. Austin, for example, thought that for each illocutionary act there must be an 'accepted conventional procedure having a certain conventional effect', which speakers must utilise (Austin 1976, pp. 14-15). Speech act conventions typically place requirements not only on the words the speaker utters, but also on the circumstances of the utterance and the characteristics of the speaker and hearer. Some felicity conditions are necessary for illocutionary success, such that if an attempted speech act does not satisfy them, it fails. Others are regulative, such that if an attempted illocutionary act does not satisfy them, it does not fail, but rather is defective or non-ideal (Austin 1976, pp. 15-17;Searle 1969, pp. 57-61). Henceforth when I refer to speech act 'conventions', or 'felicity conditions', I have in mind only the necessary rules governing illocutionary acts.
On a conventionalist understanding of uptake, uptake consists in the hearer 'recognising' that an utterance satisfies a set of conditions governing a particular illocutionary act. This might seem to involve circularity. If felicity conditions are necessary conditions for illocutionary success, and uptake is necessary for illocutionary success, then uptake is itself a felicity condition. And since uptake is the hearer's recognition that the utterance satisfies a set of felicity conditions, uptake must therefore recognise itself.
We can overcome this problem by distinguishing between first order and second order felicity conditions. First-order felicity conditions concern the utterance, the context, the hearer (excluding her uptake), and the speaker. For an illocutionary act of type to succeed, it must satisfy the first-order felicity conditions governing acts of type . The hearer's provision of uptake could be a second-order felicity condition, consisting of the hearer's recognition that the utterance satisfies all (or most) first-order felicity conditions. As with intention recognition, recognition of conventionality could be a form of perception or an inferential process.
The intentionalist view of uptake and the conventionalist view of uptake are not necessarily incompatible. On many accounts of illocutionary acts, the speaker's expression of a communicative intention is itself a felicity condition that must be satisfied for an utterance to be a particular kind of illocutionary act (see, for example, Searle 1969, p. 47). So uptake qua recognition of communicative intention is just recognition of one particular conventional feature of the utterance.
That said, speaker intention does seem to be the feature of an utterance to which hearers are most attuned, as indicated by our comfortability with unconventional speech acts, malapropisms, and ambiguous speech. Going forward I will assume that uptake typically involves perceiving or inferring a speaker's communicative intention, combined with some awareness of the other felicity conditions the act satisfies (often we might use the conventionality of an utterance as a guide to the speaker's communicative intention).

The power of uptake
In this section I will consider two ways of conceiving of the power of uptake, assuming that uptake is necessary for illocutionary success. The two theories I present are not the only theories available, so arguments against one are not necessarily arguments in support of the other. 9 Yet they are the two most theoretically salient theories of uptake, and I present them as a simplified polarity for ease of explanation.
According to the first theory of uptake, the hearer's power is limited to either ratifying the speaker's attempted illocutionary act or failing to ratify it. The potential illocutionary force of an utterance is predetermined by the speaker's expression of a communicative intention in accordance with conventions. She determines the illocutionary act her utterance could be, and the hearer determines whether that illocutionary act succeeds. If the hearer recognises the speaker's communicative intention, the act succeeds. If she does not recognise it, the act fails and no illocutionary act is performed. Call this the ratification theory of uptake.
I turn now to a second, newer theory of uptake, which I call the constitution theory of uptake. According to this theory, the hearer's response could make an utterance constitute an illocutionary act the speaker did not intend to perform. The speaker does not fully determine the potential illocutionary force of her utterance, and uptake constitutes, i.e. constructs, the utterance's force rather than merely ratifying it. The hearer's behaviour can determine not only whether an act is performed, but also which act is performed.
Quill R Kukla, writing as Rebecca Kukla, defends a version of this theory. They claim that 'there is probably no principled or sharp line between a speech act receiving mistaken uptake and a speech act being constituted, perhaps in unexpected ways, by its uptake ' (2014, p. 443). They illustrate this using the following scenario.
[I]f I ask my dinner companion, "Do you think we should get married?," this speech act might constitute a marriage proposal, the start of a conversation about the future, a request for an opinion, or a joke. Which it is depends partly upon the social context and input: are we a functioning couple? Is this the right kind of setting for a proposal? What was the rest of the conversation about? But it is also partly dependent upon the uptake: if my companion laughs in my face, or takes me unexpectedly seriously and gives me a definitive answer of a certain sort then I might learn on the spot what sort of speech act I actually produced, and the answer might surprise me. (2014, p. 443).
According to the uptake as ratification theory, if a speaker intends to propose, but her hearer interprets her as making a joke, she fails to perform an illocutionary act. Her utterance could only ever have been a proposal or a failed proposal. Yet according to Kukla, the hearer's interpretation determines which illocutionary act is performed. Even if the speaker intends to propose, she could end up making a joke, if her hearer interprets her as making a joke.
Kukla therefore denies that the speaker's communicative intention determines the force of her utterance. They claim that 'intentions in speaking are part of the story that gives a speech act the performative force it has, but they are not privileged or definitive; the speaker may only discover, in how her utterance is taken up, what sort of speech act it really was' (2014, p. 444). Kukla also denies that the conventionality of an utterance predetermines its force. Sometimes, even when an utterances satisfies the relevant conventions for a particular act, other factors, like the speaker's identity and the context, can throw the process 'off the rails' (2014, p. 445).
For Kukla, speech acts have force 'only in virtue of the concrete social difference that they make, or how they are taken up in practice' (2014, p. 443). This builds upon the theory of language they developed with Mark Lance, according to which speech acts are material performances 'constitutive of changes in normative status among various members of a discursive community' (Kukla and Lance 2009, p. 12). Different speech acts are individuated by their normative outputs. 10 Orders, for example, change the normative status of the person ordered in a unique way, such that she acquires a particular kind of obligation to do something.
Though Kukla is clear about the power of uptake, they are less clear about the phenomenology of uptake, claiming that it involves 'others' enacted recognition of [a speech act's] impact on social space', though that recognition is not 'a separable moment of passive recognition' (2014, p. 444). Perhaps they would allow that uptake can involve deliberate misinterpretation. Perhaps a hearer could recognise a speaker's intention to perform act , but deliberately respond to the act as if it is act , thereby making it act . In such a scenario, the hearer is consciously overruling the speaker, and Grice's co-operative principle clearly does not hold (1991, Ch.2).
This paper is not about Kukla's specific theory of uptake, but rather about the more general idea (which Kukla endorses) that uptake constitutes the force of an utterance. Going forward, I will assume that, regardless of whether uptake involves ratification or constitution, it always involves hearers sincerely (but not necessarily accurately) engaging with a speaker's intentions by detection or deduction. I do not attribute this view to Kukla specifically. The key disagreement between the two theories, as I present them, concerns whether a hearer's mistaken judgement about a speaker's intention can render the utterance a different illocutionary act from the one the speaker intended to perform. 11 An advantage of the constitution theory is that it enables us to isolate an unacknowledged form of silencing, to which marginalised speakers seem particularly vulnerable. Kukla illustrates this using the following example. Celia is a floor manager in a factory and has the institutional authority to give workers orders. Yet when she attempts to do so, her workers interpret her utterances as requests, because 'they are deeply unaccustomed to taking women as authorities in the male-dominated space of the workplace' (2014, p. 446). The workers know that Celia has institutional authority over them, but they do not interpret her remarks as exercises of that authority; her body marks her out to them as someone who is 'not an ordering authority in this context' (ibid.). Orders generate obligations, but requests leave the requested party free to grant or refuse them. Kukla argues that regardless of Celia's intention, because Celia's workers respond to her utterance as if it were a request, i.e. they take themselves to be free to refuse it, her utterance was a request. An utterance which has the normative output of a request is a request.
We already know that marginalised agents can be locutionarily silenced, i.e. blocked from speaking at all; perlocutionarily silenced, i.e. prevented from producing certain perlocutionary effects; and illocutionarily silenced, i.e. prevented from performing illocutionary acts (see Langton 1993). The constitution theory entails that silencing can go beyond this; speakers can also be silenced when they end up performing speech acts which are different from the acts they intended to perform.
There are strong and weak versions of the constitution theory. On the strong version, hearers can transform the speaker's utterance into any illocutionary act. The speaker's utterance need not resemble, or satisfy any of the conventions for, the act they ultimately perform. For example, imagine a speaker says, 'Hello', with the intention to greet her hearer. If her hearer interprets the speaker as intending to perform a declaration of war, the speaker could end up declaring war. Kukla can be read as endorsing this strong version, as they claim that speakers often cannot 'marshal standard conventions in the standard way' and a speech act can be performed 'with an unconventional output, given its input ' (2014, p. 445).
This strong version of the constitution theory looks like Humpty-Dumptyism in reverse. While Humpty Dumpty declares that 'When I use a word, it means just what I choose it to mean' (Caroll 1872, p. 124), the hearer on this strong version might declare, 'When you perform a speech act, it is just what I choose it to be'. 12 This is not how we ordinarily understand communication, and it has worrying moral implications, since speakers could end up unintentionally performing speech acts which harm them profoundly. For example, attempted acts of sexual refusal could become acts of consent.
Presumably the hearer's interpretation, like the speaker's intention, is constrained by rationality and conventions. A rational speaker cannot intend to do what she believes to be impossible; she acts roughly in conformity with linguistic conventions (see Dummett 1986) and forms her intentions with the reasonable expectation that they will be understood (see Davidson 1986). The equivalent should be true of hearers. A rational hearer could not interpret an utterance as being a particular illocutionary act if it did not appear to meet any of the felicity conditions for that act, nor if her interpreting it as such would be inconsistent with what she knows about the speaker and the context.
The weak version of the constitution theory could address these worries. This version holds that the hearer's interpretation is bound by rationality and conventions. She can make it the case that a speaker performed an act she did not intend to perform, but the act the speaker ultimately performs must be sufficiently similar to the act she intended to perform. An attempted greeting cannot become a declaration of war, but an attempted order could become a request.
Rae Langton suggests that unintentional illocutionary acts can occur when a speaker attempts to perform an act with a 'hearer-dependent' felicity condition (2018a, p. 151, fn.44). 13 Orders are a good example of such an act. To successfully perform an order, one's hearer must judge one to have a sufficient level of practical authority. Hearerdependent felicity conditions like this can be satisfiable to different degrees, and because of this a 'hearer may weaken what would have been an order into a mere request' (ibid.). 14 For this theory to work, there must exist classes of speech acts with the following characteristics. They share all or most of their felicity conditions, and have a similar normative force, which varies among the acts in strength rather than in kind. One of the felicity conditions the acts have in common is a) hearer-dependent and b) satisfiable to varying degrees. 15 The extent to which this felicity condition is satisfied will determine 12 One might counter that 'Humpty-Dumptyism in reverse' is in fact what Alice is defending in the first place, given she argues that Humpty cannot help but refer to glory when saying 'glory', no matter what he intends. However, Alice is not arguing that it is her own understanding of the word 'glory' which determines what Humpty-Dumpty means, but rather the existence of linguistic conventions governing the meaning of the word. Alice's view, then, is neither Humpty-Dumptyism nor Humpty-Dumptyism in reverse, but rather a conventionalist position in between the two. 13 Technically speaking, if uptake is necessary for illocutionary success, then all illocutionary acts have a hearer-dependent felicity condition; the hearer's provision of uptake. Yet I suggested in Sect. 1.2 that uptake is a second-order felicity condition. Langton seems interested in the hearer-dependence of first-order felicity conditions. 14 Alessandra Tanesini endorses a similar view (2019, p. 754). the strength of an utterance's normative force and therefore which act among the class is actually performed. Exercitives, for example, have similar felicity conditions and all seem to generate a reason or obligation for the hearer to do something. 16 But depending on the practical authority ascribed to the speaker by the hearer, the obligation the speaker generates by performing the exercitive may be stronger (as in the case of an order) or weaker (as in the case of a request). 17 Langton says little about the phenomenology of uptake, but I will assume again that hearers consider themselves to be co-operative. In cases where a hearer renders an attempted order a request, this occurs because the hearer sincerely believes the speaker intended to request and not because she is deliberately overruling the speaker. She will reason along the lines of 'Surely the speaker cannot be ordering me, since they lack the authority to do so-they must be making a request, and I will treat their utterance as such'. And given requests and orders are sufficiently similar in pragmatic structure, this reasoning can lead the hearer to constitute the speaker's attempted order as a request.
The weak version of the uptake as constitution theory requires more fleshing out, but it strikes me as more plausible than the strong version. In accordance with our intuitions, it grants the speaker at least some authority over what she does; she cannot completely predetermine the nature of the speech act she performs, but she can at least roughly determine its shape. Moreover, the idea that illocutionary force is a matter of degree is already widely accepted in the linguistic literature on reinforcement and mitigation (where what is increased or decreased is typically referred to as the act's 'strength'). 18 That said, the uptake as constitution theory, even in this weak form, should not be adopted. In the next section I offer three objections which apply to both the strong and weak versions of the constitution theory.

Lack of correspondence
Before we can decide which theory of uptake to adopt, we must decide on criteria to use when evaluating theories of speech. We must ask ourselves what speech act theory is, and what we want it to be. I propose that our theories of speech should be compatible with our everyday intuitions about speech, our moral intuitions about 16 Verdictive speech acts might form a similar class, in so far as they all involve offering some kind of appraisal or verdict, and they all generate a reason for others to accept that verdict, but the strength of the reason will vary depending on how much epistemic authority is attributed to the speaker. 17 Kukla presumably would not accept the existence of such speech act classes, since they argue elsewhere that a request is not just a weak order, but rather has a distinctive pragmatic structure and generates distinctive kinds of reasons (Lance and Kukla 2013). For them, requests and orders are not sufficiently similar such that which one is performed could be determined solely by the degree of authority the hearer assigns to the speaker. 18 See Holmes (1984); Bazzanella et al. (1990); and Thaler (2012). normative powers, and our political intuitions about autonomy. I will now show that the uptake as constitution theory fails on all three counts.
Following Gilbert Ryle, we should understand one of the tasks of a philosopher as analogous to the task of a cartographer. A cartographer mapping a village must translate the locals' knowledge of 'every house, field, stream, road and pathway' into 'universal cartographical terms ' (1971, pp. 440-441). Her map is compatible with local knowledge, but presents it in impersonal terms and situates it in a broader context. Similarly, the philosopher abstracts from and universalises our everyday concepts and explains how they relate to other phenomena. Speech act theorists should produce theories which reflect how we think and talk about speech in particular. They must shed light on the nature of communication and its relation to other phenomena, without undermining intuitions or mystifying everyday practices.
To see how the two theories of uptake fare against this ideal, we should ascertain ordinary people's intuitions about cases like Celia the manager. Defenders of the uptake as constitution theory might think this test is already in their favour. After all, Celia's employees are ordinary people, and they think she performed a request. The uptake as constitution theory can therefore claim that it reflects ordinary intuitions better than the uptake as ratification theory. Yet this is not the correct way to employ the test. The workers are missing some important information; they do not know Celia's intentions. They likely believe she intended to make a request, when in reality she intended to make an order. To ascertain their intuitions about the role of uptake in this scenario, we must describe the whole scenario to them, including the speaker's intention.
It is unlikely that, once informed that Celia intended to give an order and not make a request, the workers would maintain that she made a request. Instead, they would give a reason for why they misinterpreted her as making a request. A worker previously unaware of his unconscious biases might confess that he was not used to women giving him orders, so made an automatic and erroneous assumption about Celia's intentions. An openly sexist worker might confess to the following reasoning process: 'Celia sounds like she is giving me an order, but as a woman she does not have a right to boss me around and she must know that, so she must actually be trying to make a request'. After receiving new information, neither worker is likely to maintain that Celia performed a request. They would instead adjust their verdict and claim that Celia tried to perform an order but failed, either because of misinterpretation or because she attempted the impossible.
Even if my prediction is true-that the workers would revise their judgements when informed of Celia's intentions-one might object that we cannot extrapolate from this to the claim that all hearers in all such scenarios would revise their judgements when informed of speakers' intentions. We might think that the prediction is likely true in Celia's case because the workers' intuitions are skewed by their knowledge of Celia's authority over them. This knowledge may make the workers instinctively more acquiescent and more likely to revise their judgements.
It is helpful to consider a modified scenario in which Celia attempts to order a friend, not her workers. 19 Imagine Celia says to a friend, 'Walk my dog'. The friend, reasoning that Celia would not be so presumptuous as to order her around, interprets this as a request not an order, and declines. When informed that Celia had actually attempted to order her to walk the dog, the friend's intuitions, unlike those of Celia's workers, will be unaffected by power differentials. Would she maintain that Celia performed a request? If she respects Celia's testimony and considers her a rational communicator, I think she would revise her judgement and grant that Celia attempted but failed to order (and indeed, she would likely feel indignant about Celia's presumptuous behaviour). Sometimes we do disagree with our peers about their intentions; we might believe they are deluded about what they were trying to do with words. This is more likely to happen among friends; Celia's friend, for example, might believe she knows Celia better than Celia knows herself. But when we trust a speaker's testimony about their intentions, we do seem to regard those intentions as determinants of illocutionary force. This seems to be part and parcel of what it means to partake in respectful, co-operative conversation.
We often tell others that their speech had unintended perlocutionary effects. For example, we might tell a speaker that she offended us, even if she did not intend to do so. We also disagree with speakers about certain properties of their speech; for example, we might tell a speaker that her comment was racist, even though she did not intend it as such. And we also tell speakers that their attempted speech acts were not successful, despite their intentions to perform the act successfully. For example, Celia's friend would probably tell her that despite her intentions, she failed to perform an order. Yet if we are rational, and if we believe a speaker is rational (i.e. not a Humpty Dumpty), we do normally treat her professed intentions as determining the potential illocutionary force of her speech, even if she was not successful.
That ordinary people have intentionalist intuitions about illocutionary force does not entail that the uptake as constitution theory is false. It is possible that our intuitions about what determines force are at odds with reality. However, if we are taking seriously the thought that the task of speech act theorists is to map reality and take seriously ordinary speakers' intuitions, then we should prefer the theory of uptake that does not entail that our intuitions about speech are wrong.

Normative incoherence
Another reason to reject the constitution theory of uptake is the fact that it entails a conceptual impossibility: that we can perform unintentional illocutionary acts. I offer below a simple argument for why illocutionary acts cannot be unintentional. I will defend each premise in turn. P1: To perform an illocutionary act is to exercise a normative power. P2: To perform an unintentional illocutionary act is to unintentionally exercise a normative power. P3: Normative powers cannot be exercised unintentionally. C: One cannot perform an unintentional illocutionary act.

Defending P1: To perform an illocutionary act is to exercise a normative power
To perform an illocutionary act is to constitutively (rather than causally) change the normative situations of oneself and one's hearer by creating entitlements and obliga-tions. When I perform a promise, for example, I create for myself an obligation to fulfil that promise, and my hearer acquires an entitlement to my doing what I promised to do. When I make an assertion, I commit myself to the truth of the proposition I assert. Illocutionary acts, then, seem to involve the exercising of normative powers. 20 We can create obligations and entitlements in a variety of ways. For example, if I tread on your foot, I acquire an obligation to apologise. This does not mean that the act of placing one's foot upon another person's foot involves any kind of normative power. Rather, in treading on your foot I cause the non-normative circumstances to change in a way that triggers a pre-existing general obligation-to apologise when I hurt someone, perhaps. This act is very different from promising. When I make a promise, I deliberately and robustly change the normative situation by communicating my intention to do so. 21 Promising involves exercising one's normative powers by constitutively changing the normative situation (Raz 1972). When I promise you I will pay you back, it is precisely in performing the act of promising that I acquire an obligation to pay you back. Acquiring that obligation is part of what it means to promise someone. In contrast, acquiring an obligation to apologise is not part of what it means to tread on someone's foot. 22 Even if they do not mention normative powers specifically, most speech act theorists believe that illocutionary acts constitutively enact normative changes. This enactment seems to require the exercising of normative powers. For example, Kukla and Lance define speech acts as follows: [W]e loosely follow Brandom in understanding speech acts as performances constitutive of changes in normative status among various members of a discursive community. Thus, for instance, to assert that P involves undertaking a commitment to P, taking up the role of one at whom challenges of P may be directed, etc. (2009, p. 112). 20 Joseph Raz first introduced the idea of normative powers to moral and political philosophy (1972,1977,1979). Philosophers have since offered different accounts of why normative powers exist. Raz argues that each normative power exists because it has a specific justification: 'it is the nature of the reasons justifying the norm [e.g., the obligation created by a promise] which determines whether acts affecting its existence or application are power exercising acts ' (1972, p. 95). It matters little for my argument why or how normative powers exist, but I do require readers to accept that they exist. Bootstrapping obligations into existence can seem a bit spooky; Hume thought it to be 'one of the most mysterious and incomprehensible operations that can possibly be imagined ' (1978, p. 524). Yet it seems intuitively obvious that to make a promise is to bring certain entitlements and obligations into being. If readers accept this, but deny the existence of normative powers, it falls on them to explain the source of these obligations and entitlements. 21 I borrow the concepts of triggering and robust obligations from David Enoch (2011). 22 A reviewer points out that some cases seem to fall somewhere between the foot-treading example and promising. Consider the act of 'taking the Queen's shilling', a case discussed by Brandom (2000, pp. 162-163). Historically when a British recruiting officer gave a man one shilling, that man thereby became legally committed to military service. This was often done in taverns, so that the officers could take advantage of drunk men who only realised the legal significance of what they had done the next day. Taking the Queen's shilling, it seems, constitutively changes the normative status of the agent; in this sense it is less like treading on someone's foot and more like promising. Yet unlike promising, and like treading on someone's foot, agents could perform this act unknowingly and accidentally, and so it is hard to see how it could involve exercising a normative power. I think this case is complicated by the fact that it involves legal conventions, which can institute normative statuses regardless of intent. As noted in fn.4, I wish to focus only on the speech acts of ordinary conversation, and not institutional acts, though I acknowledge that the boundaries between these can be blurry.
Recall that to exercise a normative power is to constitutively alter a normative situation; Kukla and Lance describe speech acts in similar terms. Similarly, Searle claims that 'just about every' speech act involves taking on commitments: Just about every speech act involves a commitment of some kind or other. The famous examples are speech acts like promising, where the speaker is committed to carrying out a future course of action, but asserting commits the speaker to the truth of the proposition asserted, and orders commit the speaker to the belief that the person to whom he or she gives the order is able to do it, to the desire that he or she should do it, and to permitting the hearer to do it. In short, what people have thought of as the distinctive element of promising, actually pervades just about all speech acts. (2001, p. 147).
Key to these theories is the thought 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.
There have been several analyses of particular illocutionary acts in terms of normative powers, especially promising (Raz 1977;Watson 2009;Shiffrin 2008) and consenting (Shiffrin 2008;Dougherty 2015;Enoch 2017). Though assertion has not been defined as an exercise of normative powers, specifically, it is frequently defined in terms of the entitlements and obligations it constitutively creates (Brandom 1983(Brandom , 2000MacFarlane 2011), and thus could be appropriately modelled as an exercise of normative power.
In fact, since all illocutionary acts seem to involve a speaker taking a normative stance, and constitutively delimiting her entitlements and obligations, perhaps we can conclude that all illocutionary acts involve the exercise of normative powers. This seems to be the view taken by Richard Moran (2018), who observes a 'close relationship between the very idea of the illocutionary and that of the exercise of a normative power' (2018, p. 131).
Some might object that this is too quick. Maybe there are some illocutionary acts, like promising and consenting, which involve exercising normative powers, and some illocutionary acts which do not. For example, we might think that only illocutionary acts with a world-to-word fit require a speaker to exercise normative powers, since these involve a speaker attempting to change the world with her speech, rather than simply describing it. I will show later that even if it is true that not all illocutionary acts involve exercises of normative powers, this is not a fatal blow to the objection I put forward in this section. Nonetheless, for the time being I will assume that to perform any illocutionary act is to exercise a normative power.
One may wonder how the claim that illocutionary acts are exercises of normative powers can be reconciled with the claim that to perform an illocutionary act is to express a communicative intention using a convention. I propose that conventions enable us to exercise our normative powers. To perform an illocutionary act is to exercise a normative power by means of a convention. Exercises of normative powers must be public; i.e. to constitutively change one's own normative status and one's hearer's normative status in speaking, it must be clear to the hearer exactly what one is doing. Illocutionary act conventions, qua solutions to communicative co-ordination problems (see Lewis 1969), facilitate this publicity and eliminate moral indeterminacy (see Dougherty 2016). They do not bring the normative powers into existence but rather make possible their being exercised in a fine-grained way. We might say that normative powers are socially scaffolded by conventions.

Defending P2: to perform an unintentional illocutionary act is
to unintentionally exercise a normative power P1 held that to perform an illocutionary act is to exercise normative powers. P2 largely follows from this; it holds that an unintentional illocutionary act (if it were possible) would involve an unintentional exercise of normative powers. When I speak of an 'unintentional illocutionary act' I have in mind a speaker performing an utterance with the intention that it constitute a particular illocutionary act (like an order), where that utterance actually constitutes illocutionary act (like a request). The illocutionary act the speaker in fact performs is not the act she intended to perform. 23 It is natural to assume that each illocutionary act involves either the exercising of a different normative power or the exercising of a normative power in a unique way. If I intend to perform act , but actually perform act , then either I have exercised a different normative power from the power I intended to exercise, or I have exercised my normative power in a different way from how I intended. I did intend to exercise my normative powers in some way, and this intention was actualised, but I did not intend to exercise them in this particular way.
Here is an analogy to support P2. Imagine I possess 'musical powers', in that I can play the piano. I can exercise those powers in different ways (or exercise different musical powers, depending on how one looks at it), in so far as I can play different pieces. I sit down at the piano and attempt to perform Beethoven's Piano Sonata No. 8 in C minor. If it were possible to perform unintentional 'music acts' in the same way the constitution theory renders it possible to perform unintentional illocutionary acts, then if, despite my intentions, my hearer interprets me as performing Bach's Partita no. 2 in C minor, they therefore make it the case that I performed Bach's Partita no. 2 in C minor. 24 If this happens, it is fair to say that despite my intention to exercise my musical powers in one way, I actually exercised them in a different way; my particular exercise of them was unintentional.

Defending P3: normative powers cannot be exercised unintentionally
There is a common thread in accounts of normative powers. Gary Watson defines normative powers as 'powers to create or rescind practical requirements at will ' (2009, p. 155). Ruth Chang defines a normative power as 'the power to confer reason-giving force on something through an act of will' (2013a). For Seana Shiffrin, exercises of normative power involve 'the generation of morally significant relations merely through the expression of the will to do so ' (2008, p. 500). For David Owens, in exercising such a power, 'I change what someone is obliged to do by intentionally communicating the intention of hereby so doing ' (2012, p. 4), and for Victor Tadros, 'Y has a normative power over X when Y can directly intentionally alter X's rights and duties ' (2016, p. 205). Though he is a little more tentative, Raz makes a similar statement: 'I am not contesting that many, perhaps all, exercises of normative powers are binding only if they express the will of the power-holder' (2019, p. 11).
The italics I have added to the above quotations point to the common thread; the thought that exercises of normative powers are intentional. Though these philosophers disagree about the precise contours of normative powers, they all agree that to exercise one's normative powers is to perform an intentional act, or an act of will. Exercises of normative power are not merely typically intentional; they are necessarily intentional. It is because an exercise of normative powers is an act of will that it enacts normative changes in the first place; 'your willing is the source of your reason's normativity' (Chang 2013a, p. 75). 25 To see why this is so, we should reflect on why we need the notion of normative powers. Our obligations and entitlements could come from many places. They could be simply given to us, either by external normative facts and laws, if we are externalists, or by our passive non-cognitive states, if we are internalists. 26 Yet this does not fully capture the texture of our moral lives; sometimes we choose to create reasons, obligations, and entitlements for ourselves, i.e. we choose to bring new normative statuses into being. For example, by making a promise, I choose to create new obligations for myself. My possession of normative powers explains why this is possible, but only if exercises of normative powers are by definition intentional. If normative powers could be exercised unintentionally, then the distinction between obligations given to us and obligations created by us would collapse. The notion of normative powers draws attention to the fact that we have power over normativity; if a normative power can be exercised unintentionally, it is not a power at all, because the concept of power contains within it the notion of control. For the notion of normative powers to shed light on our moral lives, it must be that to exercise such powers is by definition to perform an intentional act.
Having now defended P1, P2, and P3, I can show why the uptake as constitution theory fails. I started with a claim that most of us can accept; that illocutionary acts constitutively alter the normative situation of the speaker and hearer. I claimed that the best way to make sense of this was to parse illocutionary acts as exercises of normative powers (P1). Because illocutionary acts and exercises of normative powers stand in this constitutive relationship, to intentionally perform an illocutionary act is to intentionally exercise one's normative powers, and to unintentionally perform an illocutionary act would be to unintentionally exercise one's normative powers (P2). However, one cannot unintentionally exercise one's normative powers, because exercising one's normative powers is, by definition, an act of will (P3). Therefore one cannot perform unintentional illocutionary acts (C). The uptake as constitution theory, because it entails that one can perform unintentional illocutionary acts, therefore entails an impossibility and should be rejected.
Earlier I considered an objection to P1; perhaps not all illocutionary acts involve the exercise of normative powers. If this is true, it is not actually a fatal blow to the objection I have mounted against the constitution theory. We could quite easily construct a modified argument, as follows: P1*: There are some illocutionary acts such that to perform them is to exercise a normative power. Call these 'illocutionary NP acts'. P2*: To perform an unintentional illocutionary NP act is to unintentionally exercise a normative power. P3: Normative powers cannot be exercised unintentionally. C*: One cannot perform an unintentional illocutionary NP act.
If the constitution theory entails that we could perform at least one kind of unintentional illocutionary NP act, then it still entails an impossibility, and we still have good reason to reject it. The strong version of the theory would definitely entail this, since it allows that a speaker could end up performing any kind of illocutionary act, regardless of her intentions. Whether the weak version of the constitution theory entails this is less clear and would depend on whether any illocutionary NP acts form the speech act classes I discussed in Sect. 1.3. I.e. it depends on whether there are any classes of illocutionary acts which involve the exercise of normative powers and have hearerdependent felicity conditions which are satisfiable to different degrees. I suspect there are, but I leave this puzzle for future work.

Political implications
There is one final reason to reject the uptake as constitution theory: it has unsavoury political implications, entailing that marginalised speakers lack so much autonomy as to barely constitute agents at all.
Being able to perform illocutionary acts by exercising our normative powers is part of what it means to be autonomous (Hurd 1996, p. 124;Shiffrin 2008, p. 500). It is helpful when thinking about autonomy and speech to distinguish between two kinds of speaker autonomy, which I call positive speaker autonomy and negative speaker autonomy. 27 Positive speaker autonomy is the capacity to ensure that one is performing the illocutionary act one intends to perform. Both the uptake as ratification theory and the uptake as constitution theory entail that agents can lack positive speaker autonomy. This is because both make speakers' illocutionary success hostage to hearers, due to the necessity of uptake. For example, if you attempt to perform a refusal, but your hearer does not provide uptake, you do not refuse, and therefore you lack positive speaker autonomy.
That both theories have this result is not cause for concern. It is obvious that many agents are frequently prevented from performing acts they intend to perform. Indeed, one of the main motivations for emancipatory movements like feminism and antiracism is to point out the ways in which certain groups lack positive autonomy, and to develop strategies for increasing that autonomy. If we embraced a theory of speech or action which entailed that everyone had full positive autonomy, we would struggle to explain why emancipatory movements still exist.
Negative speaker autonomy, meanwhile, is the capacity to ensure that one is not performing an illocutionary act one did not intend to perform. Performing actions unintentionally does not always undermine our autonomy, but there is something unique about illocutionary acts which makes unintentionally performing them autonomyundermining. Such acts involve constitutively creating obligations and entitlements for oneself and altering one's normative relationships. If another person can, on my behalf, create rights and duties for me, 'in my name', which I did not intend to acquire, then I am not a self-legislator. Instead, my own subjectivity is in effect remotely controlled.
The uptake as ratification theory entails that we all have negative speaker autonomy, since it precludes the possibility of unintentional illocutionary acts. The uptake as constitution theory entails that we can lack negative speaker autonomy, and marginalised people in particular are most likely to lack it. For advocates of the theory, like Kukla, this is a selling point. One of Kukla's goals is to show that marginalised speakers are less autonomous than typically thought, and suffer a 'special sort of incapacity ' (2014, pp. 440-441). They take their theory to be explaining a recognisable phenomenon of disempowerment.
And yet I propose that the constitution theory ends up harming the people whose plight it is designed to highlight. Those at the bottom of the social hierarchy, according to the constitution theory, will rarely succeed in performing their intended speech acts, and will constantly perform speech acts they did not intend to perform, for example consenting, refusing, and promising when they do not intend to. They are still bearers of rights and duties, which they often accrue due to others' misinterpretations, but they are frequently unable to exercise or fulfil these, since any attempts to do so will likely be misinterpreted and therefore reconstituted as acts other than the ones they intended to perform. It is hard to see how such a person is really an agent in the first place, since they have no recognisable autonomy. This is politically unpalatable. The account may also have worrying implications for how we ought to treat such a person. If it is an ontological fact that they lack autonomy, then we likely have less stringent moral duties towards them than we do to more autonomous beings, and it may even be appropriate to subject them to paternalistic control.
It is surely more palatable to say that marginalised people have the same degree of negative speaker autonomy as anybody else, but they are wronged by (a) having less positive speaker autonomy and (b) being frequently misinterpreted. The uptake as ratification theory enables us to say this. It also makes it more reasonable for people like Celia to be frustrated about their situations. According to the ratification theory, Celia's workers have erred, and Celia could 'correct' them. She has the epistemic upper hand. On the uptake as constitution theory, the hearers have not erred; they are correct about what Celia did, and Celia is wrong. In this situation, it is hard to imagine what recourse or redress could be available to her.

Conclusion
In this paper I put uptake, a notion frequently invoked by speech act theorists but rarely examined, in the spotlight. I sketched different ways of understanding the phenomenology of uptake, then considered two theories of uptake's power: the uptake as ratification theory and the uptake as constitution theory. The latter presents an intriguing challenge to the more popular ratification theory by drawing attention to the fact that marginalised speakers are not only prevented from performing illocutionary acts, but also taken to be performing acts they did not intend to perform. However, the theory goes too far in affording the interpretations of hearers constructive power. The theory is at odds with how we ordinarily think and talk about speech; it would require the unintentional exercising of normative powers, which is a conceptual impossibility; and it has unsavoury political implications, by rendering marginalised speakers so profoundly lacking in autonomy as to barely count as agents at all. For these reasons, the uptake as constitution theory should not be taken up, and if we must choose between the constitution theory and the ratification theory, the latter has the upper hand.