Representation and Epistemic Violence

ABSTRACT Sometimes an individual gets taken as speaking for a wider group without laying claim to any such authority – they are thrust unwillingly, and sometimes even unknowingly, into the role of that group’s representative. Especially for members of subordinated social groups in certain contexts, this can be hard to shake: despite their best efforts to disavow any authority to speak in the name of others, their voice might be taken as the voice of their group. In this paper we explore the intuitive injustice involved in such cases. After establishing the felicity conditions of speaking for a group, we argue that certain forms of pernicious ignorance often stand in the way of the fulfilment of these conditions. The result is a distinctive kind of ‘epistemic violence’, which can result in the silencing of both the group that is taken to be spoken for, and the spokesperson who is taken to speak for them.


Introduction
In the philosophical literature on 'speaking for others' the primary focus has been on the ethics and politics of self-appointed representatives -on the legitimacy (or the audacity, or the necessity) of laying claim to the authority to represent others. 1 Yet sometimes an individual gets taken as speaking for others without laying claim to any such authority -they are thrust unwillingly, and sometimes even unknowingly, into the role of a group's representative. Especially for members of subordinated social groups in certain contexts, this can be hard to shake: despite their best efforts to disavow any authority to speak in the name of others, their voice might be taken as the voice of their group.
This quandary, of being forced into a representative role, can take a few different forms. First, the speaker can be taken to be, in virtue of her group membership, knowledgeable with regard to the group, someone who speaks about the group with the distinctive epistemic authority of an 'insider'. Alternatively, the speaker can be taken to be, in virtue of her group membership, in a privileged position to express or articulate that group's perspective -to speak as a member of that group. Finally, the speaker can be taken to be an authorized representative, empowered to speak for the group in the sense of being in a position to undertake normatively binding commitments on the group's behalf. Our focus in this paper is on this last form of representation.
One way to silence a group is by refusing to recognise its representatives as representatives -that is, as having the authority to speak for the group. When proper spokespersons are assigned this kind of 'authority deficit' the group is prevented from having its say. Historically, this kind of recognition refusal has been used to disempower and disenfranchise marginalized groups -including Indigenous peoples under colonial rule, opposition political groups under authoritarian regimes, and so on. In this paper, we examine another practice that is closely aligned to this -both in the sense that it is closely conceptually related, and in the sense that the two often go hand in hand in actual cases. This is the practice of forcing individuals into the role of spokesperson, i.e. taking them to speak for a wider group without their meaning or being authorized to do so. We argue that this too can be a practice of silencing.
The paper proceeds as follows. We begin by giving some examples of the phenomenon in question (Section 2), and distinguishing it from a subtly different kind of 'compulsory representation' (Section 3). We then sketch an account of what it takes to speak for a group (Section 4) and use this to identify practices of 'epistemic violence' and silencing in these cases (Section 5). Our main conclusion is that the incompetence of certain audiences to correctly identify and recognise group spokespersons -particularly in the context of marginalised groups -can lead to the 'quieting' of the group, and the 'smothering' of the person taken to speak for the group.

The Phenomenon
In various domains -the political, the legal, and even the informal social sphere -individuals are often called upon to speak for others. Sometimes they are appointed into this role by some formal process in which they have agreed to participate, and, once appointed, they welcome or at least accept the authority and accompanying responsibility that goes with the role. This is the case with elected political representatives, and court or client-appointed legal representatives, for example. Other times their appointment is effected by a less formal process, even sometimes by a process of accommodation, as when an individual who possesses no prior authorization from her group acquires such authorization simply by taking on the role and having no-one object to that.
In all of these cases, the representative, the represented group, and the other parties to which representation is addressed, recognize and accept -or come to recognize and accept -that the representative speaks for the group.
However, there also instances in which someone is taken to be speaking for others despite the fact that neither they nor the group they purportedly represent consider them to be a genuine spokesperson -and indeed they may continue to be so taken despite their ongoing disavowals and protestations.
Here we consider this problem of mis-identified spokespersons in a particularly important context, namely legally mandated consultation processes with groups affected by legal, policy or administrative decisionmaking. In international law and domestic law (in many jurisdictions), certain kinds of decisions can only be taken by government bodies after those affected by the decision or policy have had an opportunity to participate in the decision-making process (Ebbesson 2007). For decisions that require the comment, input or agreement of affected parties, participation often happens by way of a consultation processes, in the form of meetings, information sharing sessions or negotiations. Consultation is meant to ensure that affected parties have an opportunity to become informed about the planned policy or activity and have an opportunity to have their say -to raise objections or comments or questions. This say must be listened to and taken into account in the reasoning of the decision-maker (Tsioumani 2018).
Significantly, the right to participation is not limited to individuals, and a state's obligations to consult may also be owed to certain kinds of groups. In many jurisdictions, the law recognizes that decisions that affect Indigenous, traditional, and rural communities not only affect the individuals within that community, but also affect the community itself. 2 The community itself is entitled to have a say and to have that say taken into account in the decision-making process (Anaya 2005).
Community rights to consultation ensure the community the opportunity and necessary information to engage in their own internal decision-making or opinion forming processes and to communicate their decision or views in a manner that accords with their traditional practices. 3 In practice, communities often appoint a representative to act as the spokesperson of the group who is tasked with communicating the comments and opinions of the community and who might enter into binding agreements on the community's behalf. In these cases, it is the community, and not the individual spokesperson, who has expressed an opinion or raised a question or is bound by the agreement.
Despite their legally recognized rights, in practice, Indigenous and traditional communities face many barriers in their efforts to have their say in decision-making processes that affect them. Indigenous communities around the world have raised concerns that consultors often co-opt representatives through bribes or pay off individuals to pretend to be community spokespersons to agree to whatever development or policy is planned (Flemmer and Schilling-Vacaflor 2016). The problem that we focus on in this paper, however, is one that is much harder to detect when it arises. These are cases in which those who are tasked with consulting the community take an individual to be speaking for the community despite the fact that neither the individual nor the community they purportedly represent consider them to be a genuine spokesperson -and indeed they may continue to be so taken despite their ongoing disavowals and protestations.
One way this may occur is when an individual participates in a consultation process without knowing that their participation will be treated as representing the wider community. Consultations often take place by way of public meetings, in which only very few members of a community are invited or are able to attend. Those who do attend are often simply those with the time or personal interest in the matter, and possess no special authority, and do not take themselves to speak for others in the community -yet this is exactly how their participation is taken.
We see this, for example, in participation processes instituted in terms of the REDD+ Programme. REDD+ (which stands for Reducing Emissions from Deforestation and forest Degradation) is a programme launched under the United Nations Framework Convention on Climate Change (UNFCCC). 4 The programme aims to encourage, fund and facilitate efforts by developing countries to reduce emissions through forest management activities, with the financial and technical support of developed nations. The roll-out of REDD+ projects can have wide-ranging impacts on landowners and forest dwellers, and especially Indigenous and traditional communities whose survival and cultural well-being may depend on their use of forest resources. As a result, the UNFCCC adopted a list of REDD+ safeguards in 2010 that include the requirement of full and effective participation of all relevant stakeholders, and in particular Indigenous peoples and local communities.
Despite the adoption of the safeguards, participation processes in relation to REDD+ activities have been found to be extremely flawed. In his research, Robert Mbeche has found: experiences from large-scale participation exercises show that who is (or should be) invited as a 'representative' to speak on behalf of communities is often left undefined. In practice, intervening agents (governments, donors, development agencies, environmental groups, outside NGOs, etc.) make choices about who should be recognised as 'representatives'. (Mbeche 2017, 429) In the REDD+ Preparedness Process in Uganda, for example, NGOs were tasked with conducting consultation with local communities and did so by dividing the target population into smaller 'communities' of 'smallholder farmers, bricklayers, charcoal dealers, timber traders.' The NGOs then invited 'representatives' from these 'communities' to participate in the process, even though they 'did not have a mandate from the represented to represent them in the consultations' (Mbeche 2017, 433).
Similarly, in the UN-funded Nigeria-REDD+ Programme, Emmanuel O. Nuesiri (2017) found that only a few people were invited to participation meetings including 'select individuals from local communities (rural village dwellers)'. These individuals were not appointed by their communities and were 'in no way democratically representative' but were seen by those hosting the meetings as descriptive representatives. Nuesiri observes that 'the select individuals invited to the participatory meetings were invited because the organisers considered them "typical of the larger class of persons whom they represent", that is rural village dwellers.' (Nuesiri 2017, 384).
In terms of the UN-REDD's guidelines on stakeholder engagement in REDD+ readiness programmes and the UN-REDD guidelines on free, prior and informed consent, the obligation to consult with Indigenous and other forest dependent communities is recognized as a collective right. In seeking out 'descriptive representatives', the responsible NGOs were not gathering data about a particular group nor were they trying to determine how the average member of the community might respond (both possible reasons for seeking out a 'descriptive representative'), rather they were purporting to fulfil their obligations to consult with the community. This they did by according descriptive representatives the role of substantial representativesi.e. representatives with the authority to speak for the wider group as its spokesperson.
One finds another example of this phenomenon in Elizabeth Ganter's research into the treatment of the speech of Aboriginal and Indigenous bureaucrats within the Australian civil service (Ganter 2016). In a series of interviews, Ganter found that Aboriginal and Torres Strait Islander civil servants reported being treated as speaking not for themselves but for particular Aboriginal communities, or indeed for Indigenous peoples generally. One interviewee, Jay, 5 described this uncomfortable situation as follows: You know you can get asked what Aboriginal people want and you go back to your line of, 'You need to go out and ask Aboriginal people yourself.' [. . .] And then when you're in a discussion about an issue and you put your point of view across, that's [seen as] an Indigenous perspective. (2016,99) Jay reported making a number of efforts to explain to his colleagues and superiors that he does not speak for Aboriginal people, usually by recommending that they consult Aboriginal people. Despite sometimes stating ''I don't represent, my views don't represent the views of Aboriginal Territorians or Aboriginal Australia', Jay felt that anything he did was 'perceived as being an Aboriginal action' (Ganter 2016, 92). Since he could not avoid being taken to speak for others, Jay stated that his only recourse was 'arranging a "blank face"' (Ganter 2016, 99) when called on to speak for Aboriginal Territorians.
This sense of being thrust into the role of a community spokesperson was keenly felt by another of Ganter's interviewees, Bob: "Q: Have you ever felt the government is expecting you to be a representative?
A: Sometimes yeah, I've been put on some things. Q: What's an example of that?
A: . . . Oh yeah, it was talking about communities and stuff like that, how they want to re-set 'em all up [speaking of the NT Intervention] and they [the department] asked us [Aboriginal staff] . . . I thought, well, that's strange, they never really asked us before . . . They wanted to get the Aboriginal staff on board and I was thinking, 'Well, they must be trying to sound like we consulted with Aboriginal people" (Ganter 2016, 100).
Like Jay, Bob attempted to push back against his treatment as a representative speaking for Aboriginal people by adopting what Ganter refers to as a 'tactical reserve' (Ganter 2016, 100). When invited or expected to speak for Aboriginal people, he would make himself 'uncommunicative' (Ganter 2016, 100).
In both these cases, the bureaucrats in question seem to be taken as speaking for social groups who have not appointed them -and this takes place despite their best efforts to disavow any authority to speak for those groups. Both Bob and Jay see themselves as being called on to speak for a community in circumstances in which the community itself ought to be consulted. They recognise that their speech will be substituted for the speech of the community and, as a result, they respond by encouraging their colleagues 'to go out and ask Aboriginal people yourself.'

Speaking for vs Speaking as
Earlier we alluded to a distinction between speaking as a group member, and speaking for a group. Speaking as a group member involves expressing or exemplifying the distinctive social perspective of one's group. It is worth noting that group representation in the sense of speaking as can be unwilling, uncomfortable and unjust. In this connection, Emmalon Davis (2016) has identified a form of what she calls 'compulsory representation' that occurs when 'positive stereotyping' of certain social groups results in members of those groups being accorded a dubious epistemic privilege -they are seen as possessing distinctive knowledge, and are then expected to share this knowledge. So, for example, women are expected to possess and share knowledge about child-care; Asian-Americans are expected to possess and provide mathematical knowledge, and so on. One aspect of this 'positive stereotyping' is the presumption that members of the social group are interchangeable as epistemic agents. When these individuals are invited to participate in epistemic exchanges, they are invited only 'insofar as the individual satisfies a certain description (woman, person of color, sexual minority, and so on)' (Davis 2016, 490).
This kind of 'compulsory representation' that Davis has highlighted is certainly an interesting and important phenomenon, but it should be distinguished from the sort of representation we are concerned with in our examples, where an individual is taken to speak for the group, in the normatively more demanding sense that the words of this 'spokesperson' are counted as the speech acts of the group itself. When women or Asian-Americans are forced to express the distinctive social perspective or supply the epistemic resources of their group, their speech does not normatively bind the group in the same way as a group spokesperson's speech binds the group. This is the key difference between speaking as a group member and speaking for a group.
The examples given in the previous section are examples in which individuals are taken to be speaking for their group, rather than speaking as group members, since their words are counted as the speech acts of the group. Yet in each of these cases, it seems to us that something goes wrong in the representative relation -the community or group in each case has not in fact been spoken for although the individuals are taken to be speaking for them. This, we argue in the next section, is because certain conditions must be met before the act of speaking for others can be said to have taken place. We make this argument by developing an account of what it takes to speak for a group that draws on a broadly Austinian speech act theoretic framework. Importantly, understanding the phenomenon of speaking for others through this framework not only reveals the failure of representative speech in these cases, but also makes apparent the ways in which both the individual speakers and the groups they are taken to represent are wrongfully prevented from having their say. In each case, the speakers and the communities are victims of practices that Kristie Dotson (2011) identifies as 'epistemic violence'.

Speaking for a Group
To speak for a group is to have one's words count as the group's own speech, in the sense that that those words, and the speech acts performed thereby, are properly attributed to the group. This is what typically happens in formal spokesperson speech. When the official spokesperson for a drug company says that 'Our vaccine had a 95% efficacy rate in the trial', this is the company's assertion, rather than the spokesperson's assertion. This is clear from the fact that it is the company, rather the spokesperson who must satisfy the relevant norms of assertion: the company must believe, or possess warrant for, or know that the vaccine had a 95% efficacy rate in that trial. 6 Moreover, it is the company, rather than the particular spokesperson, that incurs the discursive obligations associated with assertion -the obligation to answer challenges and queries, and to act in accordance with the claim asserted. The same goes for all manner of speech acts performed by spokespersons: when promises, demands, orders, warnings, etc., are made by an authorized spokesperson in the name of some collective body or group, the normative responsibilities and statuses associated with those acts are imposed not on the spokesperson but rather on the group in whose name she speaks. The phenomenon we are exploring here, which involves people being thrust unwillingly and sometimes unknowingly into a spokesperson role, is concerned with this sense of speaking for a group.
But what exactly does it take to speak for a group in this sense? Within the small body of literature on speaking for a group, the main view philosophers have taken is what could be called intentionalism. Intentionalism conceives of the spokesperson for a group as a group-authorised executor of the group's illocutionary intention. 7 On this approach, what it takes to speak for a group is roughly three things: • (intention) the group must intend to perform a given speech act, • (group authorisation) the group must authorise that spokesperson to perform the act on its behalf, and • (spokesperson awareness) the spokesperson must be aware both of what speech act the group means to perform, and that she is the one authorised to perform it in the name of the group.
On the intentionalist approach to spokesperson speech, everything needed for a group to speak via a spokesperson is located on the speaker's side (including both the group and its spokesperson) of the communicative exchange. As long as the group has authorized an individual to speak in its name, and this individual knows her role and knows what the group means to say, there is nothing to prevent groups from having their say. The problem with this view, however, is that it does not recognise the extent to which speech acts, including those performed by spokespersons on behalf of groups, are profoundly social acts -in the sense that their success is not just a matter of linguistic mastery and normative entitlement, but depends also on having a suitably receptive and competent audience. 8 Of particular relevance here is the fact that many groups do not enjoy the sort of discursive autonomy implied by the intentionalist approach, not because of any kind of internal failing on their part, but simply because of the way their attempts to speak are received. More specifically, spokesperson speech is vulnerable to a distinctive form of 'uptake failure', which involves the audience failing to appropriately recognise the group's speech as group speech. That is, their attempted speech acts fail because their spokespersons are not recognised by their audiences as having the authority or the intent to speak on their behalf. In such cases, the group is effectively silenced: it is not that they speak but are not believed, or heeded, or obeyed (etc.), rather, that they are not even heard as speaking because the spokesperson is not recognised as such.
A very different approach to speaking for a group has recently been proposed by Wendy Salkin (2021) in her work on informal political representation. In sharp contrast to the intentionalist approach, Salkin claims that an individual becomes a (informal) representative for a group just by being taken as one by the relevant audience. 9 Let us call this the conscriptionist approach to speaking for a group. 10 On this view, there is no need for the group to authorize or accept the spokesperson, and in fact the spokesperson need not even be aware of her role in representing the group. All that is needed is that the audience takes that individual to be a group representative -if they do, then the speech acts of that individual are normatively significant for the group they are taken to represent. Salkin calls this 'audience uptake' and claims that it involves, inter alia, the audience sincerely ascribing that speaker's speech to the group; treating the speaker's claims about the group as credible; and calling upon the speaker to 'stand in' for the group in relevant discursive fora (Salkin 2021, 8). When an audience treats an individual's speech in this way, regardless of whether that group or that individual endorses such treatment, that individual is a genuine informal representative of the group.
The problem with Salkin's conscriptionist account of representation, in our view, is that it hands too much power to audiences, making their uptake not just necessary but sufficient for conferring the status of spokesperson or representative. This is very much the opposite problem to the one we identified with the intentionalist account: while the intentionalist account gave groups and their spokespersons too much discursive autonomy, Salkin's conscriptionist account gives them none. On Salkin's view, it is group's audiences, not groups themselves, that determine who speaks for them, and hence what they are able to say. This has several counter-intuitive implications. It means, for example, that a group could be oblivious to who its representatives are, and that a spokesperson could be oblivious to her role as a spokesperson. And not only are these aspects of Salkin's account counter-intuitive, they also seem to have unwelcome normative implications. For example, they would seem to vindicate unjust political practices, such as the practice of obtaining consent from certain groups by selectively engaging only with those community members ('representatives') who are prepared to give it.
We favour a view of speaking for a group that lies between these two extremes. Such a view can be developed, we suggest, by extending and adapting some of the resources of Austinian speech act theory (Austin 1962) -in particular, by specifying three 'felicity conditions' of speaking for a group 11 The first condition relates to authority. Austin showed us that certain speech acts require, for their proper performance, that the speaker possess some sort of special status or authority. This is the case, for example, with the speech act of issuing commands, or appointing employees, or pronouncing a couple married. The junior cadet cannot command his superior officer to run laps, and the phony unlicensed marriage officer cannot perform the act of pronouncing a couple married -not even badly. Illocutionary authority need not always be attained explicitly and in advance of the performance of the act. For example, a speaker might acquire the authority to issue commands by a process of accommodation -by issuing the commands and having no-one object (cf. Langton 2018). Nevertheless, regardless of how it is attained, having the requisite authority is a necessary condition for the performance of certain illocutionary acts; these acts misfire when such authority is absent.
In a similar way, we suggest, there is a distinctive sort of representational authority condition on speaking for a group: someone who purports to speak for a group must actually be authorised by that group, in order to do so. Possession of this sort of authority is a matter of being authorized by the group to play a certain conventionally-recognisable role (the role of spokesperson). 12 Such authority can be circumscribed in various ways, relating to the contexts and contents of the speech in question. 13 A spokesperson might be authorized to represent a group only in designated fora, and only with respect to certain topics. Moreover, like illocutionary authority, the representational authority to speak for a group does not need to be 'official', nor does it always need to be issued in advance of the performance of the spokesperson role -i.e. it can be attained through processes of accommodation. But however it is attained and constrained, the main point is that, in order for the spokesperson's speech to be counted as the speech of the group, it must be authorized in some way by the group. An individual who purports to speak for a group, but does so without their authorization, is no spokesperson at all, not even a bad one; they are an imposter. 14 A second felicity condition on speaking for a group relates to intention. For many speech acts, it is a necessary condition of their performance that the speaker means to perform them. In other words, the speaker cannot perform an act of that kind unless her intention in speaking is to perform an act of that kind. Consider the speech act of consent: arguably, having the intention to consent is a necessary condition on the performance of the act of consent. Not all illocutionary acts are subject to an intention requirement in this way -for example, I might place a bet, or cast a vote, without intending to -but many are.
Our suggestion is that, in addition to the need to have representational authority bestowed upon them by the group, the spokesperson must also have representational intent. She must intend to play the spokesperson role, if her words are to truly count as speaking for others. This differs from the illocutionary intention to perform a certain kind of illocutionary act. The spokesperson must have a particular understanding of what it is she is doing with her words -namely that she is speaking for the group -if she is to count as playing that role. To see the need for this condition, imagine a rock star with a devoted fan-group, whose members quasi-religiously identify with her every claim and are prepared to live by her stated commitments -yet she herself is entirely oblivious to their existence, and certainly does not consider herself to speak for them. In such a case, despite the 'authorization' of the rock star by the fan-group, the speech of the rock star is not properly counted as group speech. This suggests that the representative intention of the spokesperson is an important felicity condition for representative speech. A speaker who is authorized to speak for a group without knowing that she is, or without accepting the role, cannot be said to really speak for them.
A third felicity condition of speaking for a group relates to audience uptake. According to Austin, illocutionary speech acts cannot be said to have been successfully performed unless the speaker secures the 'uptake' of their audience, in the sense that the audience recognises what sort of act the speaker is intending to perform. Austin put this as follows: Unless a certain effect is achieved, the illocutionary act will not have been happily, successfully performed [. . .] I cannot be said to have warned an audience unless it hears what I say and takes what I say in a certain sense [. . .] So the performance of an illocutionary act involves the securing of uptake. (Austin 1962, 116-7) We think a special sort of representational uptake is needed in order to speak for a group. This is not the sort of uptake with which Austin was concerned -the recognition by the audience of what sort of act the speaker is performing -but rather uptake of the speech as the speech of the group. In other words, we are suggesting it is a requirement that the audience to whom the speech is addressed recognise that the spokesperson is speaking as a spokesperson in the name of the group, and hence that it is the group, rather than the individual spokesperson, that is normatively bound by the acts thereby performed. This is intimately connected with the previous two felicity conditions, since what it is to recognise a person as speaking for a group is that one recognises their authority and intention to do so. Representational uptake is thus responsive to representational authority and representational intention.
In short, our idea is that what it takes to speak for a group is for these three felicity conditions associated with the representational dimension of spokesperson speech to be satisfied. What this means is that group speech via a spokesperson depends not only on groups, not only on spokespersons, and not only on audiences, but on all three of these parties to the representative act.

Dotson on 'Epistemic Violence'
We think that this understanding of what it takes to speak for a group can help to illuminate the intuitive injustice involved in the cases we outlined above. In particular, we think that these cases can fruitfully been seen as involving what Kristie Dotson has called 'epistemic violence' -which she defines as 'a failure of an audience to communicatively reciprocate, either intentionally or unintentionally, in linguistic exchanges owing to pernicious ignorance' (Dotson 2011, 242). This notion of 'communicative reciprocity' is taken from the work of Jennifer Hornsby and is closely related to the Austinian notion of 'uptake' discussed above. Reciprocity refers to the situation in which speaker and hearer are sufficiently attuned to one another such that appropriate uptake is granted: 'they recognise one another's speech as it is meant to be taken' (Hornsby 1995, 134). As Hornsby and others have noted, there are many things that may interfere systematically with reciprocity, and hence, by causing uptake to fail, may result in speakers from certain social group being silenced.
Dotson's notion of epistemic violence refers to a distinctive way in which reciprocity is compromised as a result of what she calls 'pernicious ignorance' on the part of the audience. Pernicious ignorance is reliable, harmful ignorance -reliable in the sense that can be predictably expected within a certain category of hearer, and harmful in the sense that it contributes to practices of communicative disablement or silencing. The key idea is that an audience's ignorance can render them incompetent to take up the speaker's speech as it is meant to be taken up, and since speakers depend on the competence of their audiences for successful communication, the result is that the speaker is silenced.
We think this idea of pernicious ignorance that stifles and silences speech can be usefully extended to the cases we outlined earlier. In particular, we suggest that what happens in these cases is that the hearer's ignorance makes them incompetent specifically with respect the issue of who the speaker is speaking for. This can result in two forms of epistemic violence or silencingone affecting the individual who is taken to speak for the group, and the other affecting the group that is taken to have been spoken for.

Smothering of the Individual
One of the practices of silencing that Dotson discusses under the banner of epistemic violence is 'testimonial smothering'. Testimonial smothering involves the speaker pre-emptively self-silencing because of an anticipated lack of reciprocity: 'the speaker perceives one's immediate audience as unwilling or unable to gain the appropriate uptake' (Dotson 2011, 244). Dotson illustrates this with an example taken from the work of Kimberlé Crenshaw, involving the reticence of some African American speakers to discuss occurrences of domestic violence, for fear of their audiences taking this as corroboration of harmful stereotypes about black men. In these cases, the speakers in question maintain silence, because they cannot trust their audience with respect to the content of the speech in question.
We suggest that a similar kind of smothering can occur when a speaker finds they cannot trust their audience with respect to the representational aspect of their speech. Recall the examples taken from Elisabeth Ganter's research, concerning the Aboriginal civil servants Bob and Jay. Neither Bob nor Jay takes himself to represent specific Aboriginal communities or Aboriginal people more generally, and, of course, neither has been authorized by any these groups to speak on their behalf. Yet Bob and Jay are acutely aware that, in a variety of contexts, what they say is taken to represent others beyond themselves -they are routinely misidentified as group spokespersons. This happens despite repeated attempts to explicitly reject and correct the mistaken representational uptake their speech has received. Ultimately, as Ganter notes, their only recourse was a form of self-silencing. Referring to Jay's comment that his response to the implicit call to speak for others is simply to 'look back at them with a blank face' (Ganter 2016, 99), she writes: Jay could not avoid speaking as though he was 'Aboriginal people'. In the end Jay could only resist the subtle pressure to substitute for the absent by arranging a 'blank face' . . .

Quieting of the Group
The other practice of silencing that Dotson discusses under the banner of epistemic violence is what she calls 'testimonial quieting'. Testimonial quieting occurs when a speaker attempts to make assertions or to give testimonybut because of their audience's profound ignorance about their epistemic credentials, their speech is not taken up as it is meant, i.e. not taken up as assertion or testimony. Dotson illustrates this with an example taken from the work of Patricia Hill Collins, of the way that widely held stereotypes, rooted in ignorance, can stigmatize black women and therein undermine their ability to participate in discourse as full epistemic agents: ignorance caused by the 'controlling images' of black women [. . .] facilitates a recurring failure of audiences to communicatively reciprocate black women's attempts at linguistic exchanges by routinely not recognising them as knowers. (Dotson 2011, 243) We want to suggest that a somewhat similar practice of quieting sometimes impedes group speech. This happens when a group's audience displays a profound ignorance about the political structure of the group, with the result it is unable to correctly identify the group's proper representatives. Recall, for example, the Ugandan REDD+ Participation Process, in which NGOs tasked with consulting affected communities did so by unilaterally dividing these communities into other groups based on livelihoods, and then inviting 'representatives', who had no mandate from these artificially constructed groups, or from their original communities, to speak on their behalf. Similarly, in the Nigerian REDD+ case, the consulting parties identified community members that they thought to be 'typical' examples of the larger community, and treated these descriptive representatives as substantial representatives, i.e. as authorized spokespersons for the community.
It seems plausible (albeit hard to conclusively establish) that the consulting parties in these cases, like the incompetent audiences of black women in Collins' example, are in the grip of certain 'controlling images' about local communities. One such controlling image may be the view that that these communities have unsophisticated political structures -that they tend to be cohesive, homogenous societies, in which any individual could, in principle, serve as a mouthpiece for the group. 15 This kind of stereotype leads to a communicative breakdown and a form of group quieting. Although the community in question is entitled to have a say they are effectively denied the opportunity to speak. They are taken to have spoken through a 'representative', but because this person lacks the authority to represent them, they end up not having spoken at all. 16

Conclusion
In this paper we have examined the pragmatics of speaking for a group, and some of its political implications. In particular, we have focused on cases in which someone is taken as speaking for a group when they in fact lack the authority or the intention to do so -and indeed may not even know that this is how their speech is taken. To understand the intuitive injustice involved in these cases, we have first enquired into what it takes for someone to successfully speak for a group.
Speaking for a group, we argued, has three distinctive felicity conditions: representational authority (the spokesperson must be authorized to speak for the group), representational intent (the spokesperson must mean to speak for the group), and representational uptake (the spokesperson must be taken as speaking for the group). What this account of speaking for others helps us to see is that the failure of audiences to correctly recognize the representational purport of a speaker's speech (i.e. the non-fulfilment of the representational uptake condition) can amount to a distinctive form of silencing. This is what happens in the cases we have described. Specifically, we have argued that the speakers in these cases experience a kind of 'epistemic violence'. Owing to the audience's pernicious ignorance in relation to the representational authority and intent of the speaker in each case, the audience fails to communicatively reciprocate in the way needed for uptake of speaking for a group. As a result, these groups do not get to speak, and the individuals, when they become aware of the discursive incompetence of their audiences, are coerced into silence.
This analysis is important because, as our examples suggest, it is marginalized groups who are most susceptible to this distinctive kind of uptake failure. For this reason, it is imperative that processes of consultation with such communities -processes which are meant to give these communities a say -are conducted in a way that takes cognisance of the felicity conditions of representative speech, and shows sensitivity to the processes by which groups appoint their own spokespersons. Only then will these communities succeed in having their say in such processes.