Discursive paternalism

According to an influential tradition in speech act theory, it takes more than linguistic mastery and normative entitlement to do things with words; one's words must also be given a suitable reception or social uptake. Working within this tradition, I identify and characterize the phenomenon of discursive paternalism . Discursive paternalism occurs when a party to a speaker's act takes control of the uptake the act receives, with the aim of curating the speaker's commitments. This is done without the prior consent or knowledge of

The idea that speakers are vulnerable in this way has inspired a number of feminist philosophers of language to identify various forms of injustice that occur in speech situations. Speakers from certain social groups are said to be 'silenced,' or to be the victims of 'discursive injustice', on account of the social reception their acts routinely receive. 2 In these discussions, the perpetrators of this kind of injustice are typically pictured as doing so either unthinkingly or maliciously. More specifically, the audience is portrayed either as being in the grip of a prejudicial stereotype that blinds him from recognising what the speaker is meaning to do with her words, or as a kind of discursive saboteur, who sees what act the speaker means to perform but deliberately derails it.
In this paper I want to highlight a related phenomenon, which I call 'discursive paternalism.' I take it that actions are paternalistic when they involve external interference with a person's autonomy or decision-making, without that person's prior consent or knowledge, but for their perceived benefit. 3 Discursive paternalism occurs when a party to a speaker's act interferes by taking control of the uptake the act receives, with the aim of changing the stakes for the speaker. This party ('the paternalist') is neither blind to the speaker's intention, nor does he seek to undermine or sabotage the speaker. On the contrary, he plays the role he does-the role of a discursive guardian who takes it upon himself to curate the speaker's commitments-for what he takes to be the good of the speaker. In some cases, the paternalist considers the speaker unable to bear the normative costs associated with the speech act she is attempting, so he acts so as to deliberately lower those costs. In other cases, he thinks the speaker would be better served undertaking a greater commitment, so he intervenes in such a way as to raise the stakes.
Importantly, as with paternalistic action of other kinds, discursive paternalism occurs without the prior knowledge or consent of the speaker.
An example can help to bring the phenomenon into view: WHAT I'M HEARING IS. Out on a spring walk in a Norwegian forest, a father repeatedly tells his six year old daughter to put on her jacket, even though she assures him that she does not feel cold.
Eventually, she snaps: 'Dad, if you tell me one more time to put on my jacket, I will chop off your head!' The intent of the child is unmistakeable to the father: she means to threaten him. He is taken aback, then slightly amused, then slightly angry, but recovers soon enough to reply, 'OK, so what I'm hearing is that you feel like I'm being a bit too bossy, and that you'll put your jacket on if you start to feel cold.' The daughter wordlessly accepts this, and that is the end of their discussion.
In this example, the father decides, for his daughter's own sake, not to take her speech act in the way she intended it. After all, mortal threats are a serious business: they conditionally commit the speaker to follow through with lethal action if the threat is not heeded, and they make the speaker answerable for the appropriateness of such action in those circumstances. So instead of taking his daughter's speech act as a threat, the father deliberately and explicitly takes it in a normatively much less demanding way than it was intended-as a mild complaint, rather than a threat.
In what follows, I explore and characterize this kind of phenomenon through the lens of a social normativist speech act theoretic framework. My first step is to outline this general framework, after which I describe how the framework has been used to highlight the phenomenon that Quill Kukla (writing then as Rebecca Kukla) has called 'discursive injustice.' I then describe three distinct forms of discursive paternalism, which differ in terms of the role played by the paternalist, and the kind of authority he lays claim to in performing that role. 2 The notion of silencing has been developed extensively within a broadly Austinian framework by Jennifer Hornsby and Rae Langton (see Hornsby, 1995;Hornsby & Langton, 1998;Langton, 1993). See also more recent work by Ishani Maitra (2009), Kristie Dotson (2011), Mary Kate McGowan (2019, and Laura Caponetto (2021). The notion of discursive injustice, which is my primary focus here, has been developed by Kukla (2014), as well as Tanesini (2016Tanesini ( , 2020. 3 See, e.g., Shiffrin (2000), Scoccia (2018).

| THEORE TIC AL FR AME WORK : SOCIAL NORMATIVIS M ABOUT S PEECH AC TS
According to social normativism, speech acts are, first and foremost, normatively significant moves in social or discursive space. On this framework, different speech act kinds can be distinguished on the basis of their normative character-i.e., what licenses them, and how their performance alters the normative statuses and social relationships of relevant members of the speech community. Social normativism has been developed, defended and deployed by various philosophers, including Brandom (1994), MacFarlane (2011), Tirrell (2012), Geurts (2019), and Caponetto (2021), but its most detailed articulation is due to Kukla and Lance (2009) and Lance and Kukla (2013).
Kukla and Lance draw a distinction between a speech act's normative 'inputs' or entitlement conditions, and its normative 'outputs', i.e., what normative effects it has. The inputs are the various conventionally required prequalifications and preconditions that must be in place in order for the act to be performed by a given speaker in a given set of circumstances. This includes facts about the speaker herself (e.g., must she hold a certain position in order to perform the act?), the state of the conversation (e.g., is this act only licensed when it is a response to a previous act?), and facts about the environment (e.g., do certain objects have to be present in order for this act to be entitled?). The outputs are the ways the speech act changes the normative situation of the speaker, the audience, and others. This includes licensing other speech acts, imposing obligations to behave in certain ways, and providing or denying access to certain social institutions.
Consider the speech act of marrying, i.e., pronouncing a couple married. The normative 'inputs' for this speech act include the speaker being suitably licensed (in this case quite literally: they should be a licensed marriage officer), and that the act take place in a certain setting, involving the appropriate people (i.e., the bridal couple) and at a particular moment within a larger ceremony. The normative 'outputs' of the act of marrying are similarly diverse: the couple becomes subject to a range of legal and institutional statuses, the guests become entitled to congratulate them, etc. Other speech acts, like telling or warning, are less ceremonial or ritualistic, and they do not require any special authority on the part of the speaker. Yet they too shift normative statuses by licensing certain attitudes and behaviours on the part of the audience, and making the speaker answerable to the audience in various ways.
A crucial part of the social normativism developed by Kukla and Lance is its materialist approach to normative statuses. The normative inputs and outputs that characterise speech acts are not 'free floating' normative statuses existing only in an abstract space of reasons. Rather, they supervene on the actual attitudes and behaviours of relevant members of the speech community. For a speech act to license certain behaviours, for example, it must actually dispose people to act as though such behaviours are licensed: normative statuses are material social statuses […that] supervene on concrete, materially implemented dispositions to act. A speech act that does not make a difference to how people are actually disposed to behave does not succeed in having a normative output at all […][S]peech acts have their performative force only in virtue of the concrete social difference that they make, or how they are taken up in practice. (Kukla, 2014, p. 443, emphasis in original) An important upshot of this is that the illocutionary status (or 'performative force') of a speech act depends to a large extent on the act's social reception. On this approach, the speaker is not a god-like authority when it comes to the speech acts she performs. Instead, the nature of a speaker's acts is partly constituted by the uptake they receive, and this makes speakers vulnerable to their audiences and other parties in significant ways.

| D ISCUR S IVE INJ US TI CE
In recent work, Kukla (2014) has demonstrated the fecundity of the social normativist framework for identifying a wide variety of ways that speech can be unjustly interfered with or impeded. In particular, the social normativist framework can explain cases in which the uptake a speaker's act receives constitutes it as a different kind of act from the act she intended. In these cases, the speaker's act is not simply silenced but distorted; she does not find that she failed to do anything with her words, but that she has done something other than what she intended.
In the extant literature, 'silencing' has usually been understood as something that happens when an audience fails to recognise the intentions of the speaker-where such recognition is seen as necessary for the successful performance of an illocutionary act. In the canonical example of silencing or 'illocutionary disablement' (Langton, 1993), a woman utters the word 'No' with the intention to therein refuse a man's sexual advances. Yet, because the man is in the grip of a stereotype about women (e.g., that they 'play coy', pretending to refuse when they really mean to accept) he fails to recognise this intention, and so the woman is prevented from successfully performing the illocutionary act of refusal. When widespread stereotypes about certain speakers reliably produce this kind of uptake failure, the result is a systematic kind of silencing: certain speech acts become 'unspeakable' (Langton, 1993) for certain speakers in certain situations.
In contrast to this kind of approach, what makes the social normativist framework an especially powerful tool for highlighting and explaining the ways speech can be unfairly impeded is its broader construal of 'uptake.' For the social normativist, the uptake a speech act receives is not simply the recognition by the audience of what the speaker is trying to do, but rather the way it gets responded to in social practice-the way it affects how people are disposed to think and act. It is this 'concrete social difference' that plays the key role in constituting the speaker's act as an act of one kind or another, and this has two important consequences for thinking about how speech can go awry or be unjustly impeded. First, it means that what act a speaker performs is not up to her alone: she may attempt to perform an act of type A and find, because of the uptake her act receives, that she has performed an act of type B. Second, it means that a speaker can be prevented from performing the act she intends to perform even when her intention to perform it is recognised by the audience. Kukla (2014) uses the social normativist framework to illustrate how speakers with certain disempowered social identities may have their speech systematically distorted in ways that exacerbate their social disempowerment. This is the phenomenon of 'discursive injustice.' Kukla illustrates this with several examples of how gender can influence uptake: how the speech acts of women are, at least in certain contexts, routinely taken as different acts from those they intended. For example, in certain male-dominated workplaces, the perfectly entitled orders of a female manager may be routinely taken by her male subordinates as requests. This leads to a potentially damaging pragmatic breakdown: the manager ends up appearing ungrateful when her subordinates fall into line (the fulfilment of requests, but not commands, calls for gratitude), and incompetent when they do not (a manager whose orders are disregarded is failing to exert her authority). What makes this kind of pragmatic breakdown an injustice, rather than simply a misfortune, is that it has a systematic connection to the speaker's disempowered social identity. More specifically, this linguistic infelicity is one that tracks and exacerbates the social disempowerment of the salient social identity-here, the speaker's gender identity.
Kukla's examples of discursive injustice focus primarily on the possibility of speech being distorted by being taken up differently from how it is intended and how it ordinarily would be taken up, but for the speaker's gender. Some other authors have provided examples that focus more on the other feature of the social normativist framework noted above, namely, that it allows for the possibility that a speaker's speech can be disabled or impeded even when the audience correctly recognises the speaker's intention. In this vein, David Spewak (2017) and Alessandra Tanesini (2020) both employ a normativist framework to explain cases in which a speaker's attempt to make an assertion is silenced even though her interlocutors recognise she is attempting to make an assertion.
When such cases bear the same systematic relation to the speaker's social identity, they too can be counted as discursive injustices.

| DISCUR S IVE PATERNALIS M
The phenomenon I want to describe, discursive paternalism, shares some of the key features of these central cases of discursive injustice. Like the cases presented by Kukla, it involves a 'distortion' of the speaker's act-a 'queering of the path between performance and uptake' (Kukla, 2014, p. 444). And like the cases discussed by Tanesini and Spewak, it occurs even though those effecting the distortion correctly recognise what the speaker is trying to do.
But discursive paternalism is distinctive because of the motivation of the paternalist, who plays a decisive role in determining how a speaker's act is taken up, with aim of changing the stakes for the speaker, for what he takes to be the speaker's own good.
I shall describe three forms of discursive paternalism, which differ most starkly in terms of the role played by the paternalist, and the kind of discursive authority they lay claim to in playing that role. In the first, the paternalist is the speaker's addressee, who benevolently decides to take up, and so constitute, the speaker's act in a different way from the way it was intended. We can call this paternalistic interpretation. In the second form, the paternalist is a bystander or third party, who comes to the aid of the speaker by re-framing the speaker's act so that it will be taken up differently from the way it was intended. We can call this paternalistic intercession. In the third form, the paternalistic actor is an associate of the speaker, who assumes the role of a co-speaker of the speech act and uses that role to re-articulate its nature. We can call this paternalistic commandeering.

| Paternalistic interpretation
Acts of paternalistic interpretation-in which an addressee decides to take the speaker's act in a normatively different way from the way it was intended-are familiar enough from the realm of parenting. At times, parents may seek to protect their children by reducing the normative stakes associated with the speech they are attempting.
Hence, would-be lies may be taken not as assertions but as creative storytelling ('A shark in the school pool! What a funny story'); hurtful insults may be laughed off as harmless jokes ('Haha, I guess daddy is looking like a big balloon these days'); and promises and vows are taken not as binding commitments but as expressions of wishes or hopes ('It's nice that you want to clean up after the new puppy, let's see how it goes'). It should be noted that parents also sometimes seek, through paternalistic interpretation, to increase the normative stakes for their children, in an effort to bolster their confidence or to develop a sense of responsibility. Hence, tentative guesses might be taken as firm claims (child: 'Is the answer… maybe… 14?'; parent: 'Wow, you really know your stuff!') and noncommittal suggestions taken as promissory commitments (child: 'I might be able to do the dishes on Mondays'; parent: 'Thanks, we're counting on it!').
But while parenting may be the natural home of paternalistic interpretation, it also occurs in other speech situations, especially in exchanges where a disparity in discursive authority between addressee and speaker puts the audience in a position to determine what act the speaker has performed. By way of example, consider the way that the speech of Indigenous people is sometimes taken in the course of legally-mandated consultation processes-processes which are meant to give these communities a say about proposed developments on their land. The following is a stylized adaptation of a real-world case: 4 WHAT YOU'RE SAYING. A court is deciding whether to license oil drilling operations on the traditional land of an Indigenous community. The community strongly opposes these operations, and the court is sympathetic with their position. During the proceedings, the court receives testimony from one of the elders of the community about the impact of the proposed underground drilling operations. 'Beneath the ground, ucupacha, there are people living as they do here,' he says. 'There are beautiful towns down there, and there are trees, lakes and mountains.' This is meant by the speaker as an assertion about the natural environment, and the court recognises this. But the court also considers the speaker to lack epistemic credentials with respect to questions about the state of the natural environment. So the court decides to take this testimony differently-not as a statement about the environment but as something about which the speaker does have proper epistemic credentials, namely, his community's culture and 'world view'. This is reflected in the court's judgment, which paraphrases the statement of the community elder as follows: '[the community has] a profound and special relationship with their ancestral territory, which encompasses their own worldview and cultural and spiritual identity.' 5 Arguably, Indigenous speakers are routinely subjected to this kind of paternalistic interpretation, which limits what they are able to make assertions about, at least in certain contexts. Essentially sympathetic interlocutorscourts, NGOs, activists-may judge them unable to bear the normative burdens of making assertions about certain things, such as the state of their natural environment, but not others, such as their own culture or 'worldview'. As a result, these audiences might 'charitably' decide to take the would-be assertions of Indigenous speakers about their natural environments as assertions about their culture or world-view. 6 Whether or not this is in fact an accurate reflection of what happens to Indigenous speech in these contexts, this example nonetheless illustrates a distinctive form of discursive paternalism. What is characteristic of this form is that the paternalist lays claim to a particular form of authority, viz., the authority of the addressee. As the one to whom the act is addressed, the addressee occupies a special position when it comes to nature and success of that act. On the social normativist framework, as we have seen, the addressee of a speaker's act often plays a decisive role in determining the character of that act, or indeed whether the act was performed at all. For example, a would-be order can, in certain circumstances, be constituted instead as a request, on account of the way it is taken up and responded to by its addressee, or an attempt to assert something can, in certain circumstances, be deliberately silenced by an addressee who simply ignores the attempt. This is not of course to say that addressees generally have the authority to silence or 'distort' the speech that is addressed to them; on the contrary, this might only be possible when the addressee and speaker stand in a pre-existing asymmetrical power relation (e.g., parentchild; mentor-mentee; boss-employee), or when there are broader social dynamics in play (e.g., the addressee is a member of a more dominant social group). The point is just that sometimes the social circumstances of a speech act do empower addressees play this decisive role, and when they do, paternalistic interpretation is made possible.

| Paternalistic intercession
The second form of discursive paternalism I want to highlight is paternalistic intercession. Unlike paternalistic interpretation, paternalistic intercession features a paternalist in the role of an interested third party or onlooker, who takes it upon himself to intervene by moderating the speech act for the addressee.
Like paternalistic interpretation, paternalistic intercession is also familiar enough from the realm of parenting, with parents sometimes monitoring and intervening in their children's speech in order to protect them from having to bear certain normative costs. A child's expression of disgust at an offer of food might be re-framed by her mother as a polite refusal of a second helping or a child's hurtful insult might be re-framed as playful banter. Over-zealous parental intervention of this sort can perhaps be thought of as a discursive version of 'helicopter parenting,' since it involves the parent standing ready to protect the child, not so much from bumps and bruises, but from blushes, sanctions, ripostes, challenges and various other normative costs associated with the speech they are attempting. 5 Sarayaku v. Ecuador,par. 155. 6 For further discussion of this case see Townsend (2020Townsend ( , 2021, and Townsend (2020, 2021). Outside of parent-child relations, paternalistic intercession occurs in contexts where one person is empowered, or takes himself to be empowered, to moderate the speech of another. By way of example, consider the following case: WHAT SHE'S SAYING. At the airport check-in counter before an international work trip, a young woman is told by airline staff that her flight is overbooked, and so she is being placed on a subsequent flight. She is extremely annoyed at this inconvenience, but, understanding that this is not the fault of the person working behind the check-in counter, she limits herself to a plea: 'It's really very important that I get to Peru by tomorrow -if there is anything you can do, I'd be very grateful.' An older male colleague, who is accompanying the woman on the work trip, witnesses this interaction and rushes over to intervene, telling the airline worker, 'What she's saying is that she has a ticket for the flight, and so you need to make sure that she gets on that plane!' The young woman feels somewhat undermined and embarrassed by her colleague but decides to let it go, in the hope that this firmer approach-making a demand rather than a request-might end up getting her on the plane.
In this scenario, the male colleague takes it upon himself to speak for or on behalf of the young woman. In so doing, he seeks to moderate her intended speech act by changing its normative significance. That is, he re-frames the intended act (a request) so that it will be taken up as a normatively more demanding act (a demand), since he thinks the entitlement to the normatively more demanding act is in place (she has a ticket), and this act will better serve her interests. And he does this without the prior consent or knowledge of his colleague.
In acting in this way, the male colleague lays claim to a distinctive kind of authority: the authority of a proxy speaker. Although this kind of authority has received less attention than addressee authority in the speech act theory literature, it is an important feature of many discursive exchanges. 7 Such authority can be quite formal, for example in the case of the lawyer who is specifically authorized to represent a client in court ('Your Honour, what my client means to say is that she is hardly knew the victim'). Other times, however, the authority is not formal, and in fact may not even be recognised in advance by the speaker-in such cases, the authority may be attained by accommodation. 8 As the example above suggests, this is the sort of scenario in which paternalistic intercession is particularly apt to occur.

| Paternalistic commandeering
The third form of discursive paternalism, which I call paternalistic commandeering, involves a somewhat different dynamic. Here the paternalist is neither the addressee nor a proxy for the speaker. Instead, in paternalistic commandeering the paternalist is an associate or ally of the speaker who seizes control of the speaker's act by placing himself alongside the original speaker, as a kind of co-speaker. In paternalistic commandeering, the paternalist joins forces with the speaker by assuming co-authority for the speaker's act, while at the same time using that authority to twist or moderate the act itself.
Like the other two forms of discursive paternalism, paternalistic commandeering is something that parents sometimes practice. An impassioned dinner-time complaint of 'Seriously, dad? Not pasta again!' might be coopted and toned down by a co-parent with dual sympathies: 'What we're saying, darling, is that it might be nice to try something else once in a while.' In this case the paternalistic parent ostensibly aligns herself with the child's protestation, while subtly steering it in a gentler direction. Paternalistic commandeering also occurs outside of 7 An important exception is Kirk Ludwig's recent work on proxy speech (Ludwig, 2014). See also Lackey (2018). 8 Langton (2018, pp. 150-151) discusses how the authority to perform certain speech acts can be attained by a process of accommodation, e.g., how the authority to issue directives 'often depends on whether hearers go along with it.' parenting contexts. Consider, for example, the way canny politicians sometimes align themselves with the bold messaging of popular protest movements, even as they subtly moderate their messaging. Here is an example: WHAT WE'RE SAYING. A rising political star in the USA participates in a 'Defund the police' protest. He fully supports the movement of which the protest is a part, and he knows that visibly doing so is on balance good for his political prospects. But he is concerned that the demand to defund the police will alienate some of his supporters, and also that it is apt to heighten tensions and perhaps even escalate police violence. Taking to the stage, he declares: 'What we are saying is not that the police should be completely abolished, or that they are rotten to the core. We just want the bad apples to be removed, and for some of the funding to be reallocated to other social services.' Like the other two forms of discursive paternalism, paternalistic commandeering also involves the paternalist intentionally moderating a speaker's speech, for what he takes to be the speaker's own sake. As with the previous two forms of discursive paternalism, the paternalist in this sort of case invokes a distinctive kind of authority, in this case the authority of the co-speaker. This is not the authority to take the act up in a certain way (as in paternalistic interpretation), nor the authority to represent or reframe the act (as in paternalistic intercession)-it is the authority to jointly undertake the act itself, as a kind of co-author or co-speaker. 9 The authority to speak with others, like the authority to speak for or on behalf of others, can be attained through accommodation. In this example, the organisers of the protest may choose-for strategic reasons-to go along with the message as it is re-articulated by the politician, and in this way they accept him as a co-speaker of their demands. This is what makes the authority of a co-speaker susceptible to paternalistic commandeering, since it can be invoked without the prior consent or knowledge of the (original) speaker.

| OBJEC TI ON S AND REPLIE S
To end off, I will briefly consider two concerns about the account of discursive paternalism I have presented.

| Speaker autonomy
Discursive paternalism, as I have described it here, involves parties other than the original speaker playing a decisive, constitutive role in determining what speech was performed. The father in WHAT I'M HEARING takes up his daughter's utterance in such a way that it is actually constituted as a mild complaint, rather than mortal threat; the colleague in WHAT SHE'S SAYING intervenes in such a way that the young women's attempt to issue a request is made instead into a demand, and so on. To some readers, this may seem an unpalatable affront to the autonomy of the speaker, making the speaker hostage to the perversity of the audience. Speakers and audiences are fundamentally distinct, and although audiences may in various ways impede the linguistic agency of speakers, they cannot 'put words in the mouths' of speakers in the way envisaged by discursive paternalism. Interestingly, J.L. Austin himself seems to have made a point along these lines in reference to a strikingly salient example: I do not perform [someone else's] act of betting, which only he can perform […] I do my own betting and he must do his own. Similarly, an anxious parent may say 'He promises, don't you Willy?' but little Willy must still himself say 'I promise' if he is really to have promised. (Austin, 1962, p. 63) 9 For discussion of joint or collective speech acts see Hughes (1984), Meijers (2007), Schmid (2020), Tollefsen (2020).
There are two things to be said in response to this. The first is simply to acknowledge that discursive paternalism does typically involve audiences interfering with the autonomy of the speaker. This may be objectionable, but it is not, at least not according to the social normativist framework adopted here, impossible. The use of language is primarily a matter of making normatively significant moves in social discourse, where both the nature and the success of the moves made is determined in large part by the way they are received. Working within this framework, I have tried to show that sometimes an addressee, bystander, or associate of the speaker can intervene in the paternalistic ways described above in order to affect how the speaker's act is taken up, and hence what the speaker has ultimately done. To tweak Austin's example, an anxious parent who simply declares to a third party that 'Willy promises' (without giving Willy any chance to confirm this) might succeed in making it so that Willy has indeed promised. So there is indeed a sense in which speakers are 'hostage' to the influence of others, and the phenomenon of discursive paternalism tracks a distinctive way in which this vulnerability can be exploited by well-intentioned audiences for the perceived benefit of the speaker.
The second thing to note is that, although the social normativist framework portrays speakers as vulnerable to others in respect of how their speech is taken up, this does not mean that they should be viewed as completely disempowered, since there are various ways they can seek to resist or challenge wrongful uptake. In this sense, saying speakers are 'hostage' to their audiences is too dramatic, because speakers are not simply at the mercy of their audiences the moment their words have left their lips. They may, for instance, explicitly reject or protest the paternalistic interpretations given to their speech ('No, you should take that as a threat, not a complaint') their speech receives, or they may attempt to block paternalistic interventions by calling into question the authority of the paternalist ('Who are you to speak for us?!'). 10

| An alternative explanation
Even if it is accepted that audiences and other parties can in principle play a decisive, constitutive role in determining the nature of a speaker's act, one may nonetheless wonder whether this is what is happening in the examples under discussion. Perhaps, rather than acting so as to constitute the speaker's act in a different way from the way intended by the speaker, the paternalist is replacing or overriding the speaker's original act with a speech act of their own, one that they believe better serves the interests of the speaker. So, for example, in WHAT SHE'S SAYING, perhaps the older colleague is himself making a demand, on the basis of his own perceived greater 'authority' (as an older, middle-aged man), that he considers in the interests of his colleague (i.e., that she be assigned a seat on the flight). And, in WHAT WE'RE SAYING, perhaps the canny politician is himself making claims and acts of protest, on the strength of his considerable political influence, that he considers to be beneficial to the larger movement. 11 In response to this alternative reading, it should first be acknowledged that, in all the cases of discursive paternalism I have described, the paternalists do themselves perform speech acts of various kinds. Typically, as my labelling of the cases suggests ('WHAT YOU'RE SAYING', 'WHAT SHE'S SAYING', etc.), these are acts of explicitation, negotiation, or reframing of what was said. However, the important thing to note is that these acts of the paternalist are not meant to replace or override the speaker's original act, but rather to affect how that act is taken up. That is, the paternalist is not seeking to change his own normative situation, by himself undertaking the commitments or offering the entitlements associated with relevant speech act, but rather he is attempting to lower or raise the normative stakes for the speaker. This is clear from the fact that when discursive paternalism succeeds it is the original speaker, rather than the paternalist, whose normative situation is altered in the relevant ways. In the case of WHAT SHE'S SAYING, for example, it is the young woman, rather than her older male colleague, who is assigned the normative statuses associated with a demand: it is she, not he, who becomes entitled to the delivery of what is demanded, or an explanation for non-delivery, and so on. Similarly, in the case of WHAT WE'RE SAYING, the claims made by the politician are claims for which the protesting group, rather than just the politician, becomes responsible-claims which they, not just he, are obliged to justify, if challenged, etc. The paternalists in these examples are attempting to curate the normative reception of another speaker's act, and thus modify the force of that act. This is not to deny, of course, that there may be a different sort of 'paternalism' that involves coming to the aid of a speaker by adding one's voice in support of theirs-adding one's own demand to theirs, or one's own statement to theirs, for example.
But that is a different phenomenon, rather than an alternative account of the phenomenon I have identified here.

| CON CLUS ION
My aim in this paper has been to identify and characterize, with the help of a social normativist framework, the phenomenon of discursive paternalism. The fact that speech is social action makes our linguistic agency highly vulnerable to interference from others. I have argued that this interference can be paternalistic in character, in the sense that it can be a deliberate curtailment of a speaker's autonomy, without her consent or knowledge, but for her perceived benefit. I detailed three different ways that discursive paternalism can occur: when an addressee deliberately takes up a speech act in a different way from the way it was meant (paternalistic interpretation), when an onlooker re-frames a speaker's act (paternalistic intercession), and when an associate of the speaker co-opts the speaker's act (paternalistic commandeering).
This discussion raises several challenging questions about the ethics and politics of discursive paternalism. I have not assumed that discursive paternalism is inherently unjust, or that it is always ethically problematic, though my examples suggest it sometimes is. So, in what circumstances, if any, is discursive paternalism justified? I leave this, and related questions, for future work.