Freedom of speech: A relational defence

Abstract

Much of the recent literature on freedom of speech has focused on the arguments for and against the regulation of certain kinds of speech. Discussions of hate speech and offensive speech, for example, abound in this literature, as do debates concerning the permissibility of pornography. Less attention has been paid, however, at least recently, to the normative foundations of freedom of speech where three classic justifications still prevail, based on the values of truth, autonomy and democracy. In this paper we argue, first, that none of these justifications meet all four intuitive desiderata for an adequate theory of free speech. We go on to sketch an original relational view of free speech, one which grounds its value in the recognition that speakers grant each other when engaging in speech practices, and its limits in the republican ideal of non-dominated co-exercisable liberty. We briefly illustrate the relational approach’s implications for the debate on hate speech regulation and for the response to fake news.

1. Introduction

The literature on free speech is voluminous, fuelled by a never-ending stream of controversies. Much of that literature though is relatively homogenous and, in the US, somewhat self-referentially, preoccupied by First Amendment interpretation. Further more, the free speech literature tends to focus on the limits on speech, and on its regulation. In a legal context, the reason for that focus is obvious. But it leaves absent a reflection on why free speech is valuable in the first place, even though this is a necessary prerequisite for an adequate defence of its limits. In fact, the extant literature on the foundations of free speech has not advanced significantly in several decades. It continues to be dominated by three classic approaches, based on the values of truth, autonomy and democracy. Even Shiffrin’s (2011, 2014) recent and original thinker-based argument is presented by her as a version of the autonomy view.

In this paper, we pursue the ambitious aim of developing a new view of free speech’s nature and value, at least in outline. Our reasons for doing so stem from substantive dissatisfaction with the extant theories. Specifically, we claim that such theories cannot meet four intuitive desiderata for an adequate theory of free speech. In Section 2, we explain these desiderata and in Section 3 we employ them to evaluate the truth-based, autonomy and democracy theories. Subsequently, we present our own – relational – view. The intrinsic value of speech, we argue, is in the recognition that speakers thereby accord each other (Section 4), while the scope and limits of speech are supplied by the theory of republicanism (Section 5). Having presented the relational view, we show that it meets the four desiderata (Section 6). Section 7 briefly illustrates the fertility of the relational view by sketching its implications for speech’s limits in two vexatious areas, the established debate on hate speech, and the newer one of fake news. Section 8 briefly concludes.

2. Desiderata for a theory of free speech

In a liberal society, there is a presumption in favour of individual liberty, but one that may sometimes be overridden on welfarist grounds (Dworkin 1977). The point of a free speech principle is to protect speech from being limited by all but the most powerful normative considerations. That may be because speech is valuable yet peculiarly vulnerable to being unjustifiably limited; and/or because although some speech is apt to harm, it should nonetheless be afforded protection due to some special value that we assign to it, and therefore the threshold for intervention should be higher than with other sources of harm. The first desideratum for a free speech theory is therefore to explain this specialness, and so to distinguish speech from liberty more generally.

Speech, though, is not all of a type. If we accept the common distinction between ‘high-value’ speech (e.g. political and artistic) and ‘low-value’ speech (e.g. hate speech, pornography), then the point of a free speech principle, especially on the higher threshold for harm interpretation, should also be to vindicate the intuition that some kinds of speech merit especially stringent protection. Stringency is therefore the second desideratum for a theory of free speech, and political and artistic speech are the categories most commonly thought to be clear instances of high-value speech.

The third desideratum is to address the issue of balancing pro-free speech considerations against reasons to limit or regulate speech. Balancing diverse considerations in a world of plural values is hardly a novel enterprise for political theorists, or indeed courts. Having asserted a right to free speech, for example, Article 10 of the ECHR goes on to adumbrate in its second clause interests such as public safety and privacy, which can militate against free speech in certain contexts. We call this external balancing because the values that are weighed against free speech are not values that underly free speech itself. By contrast, if free speech is justified on the basis of some value, then it may be that in other contexts respect for that value will militate against one party’s speech. We call this internal balancing because it involves the rationale for free speech itself. Some external balancing is unavoidable with plural values in a non-ideal world. However, it is a plausible desideratum of a theory of free speech that it requires no internal balancing, that is, that its basic rationale does not also count against speech (this point will become clearer when we consider theories of free speech below).

The final desideratum, one seldom addressed in free speech circles, is that citizens have the opportunity to speak. It is true that the negative right to free speech and the opportunity to exercise it are analytically distinct. But we need to consider the meaning of a free speech right from the perspective of a potential speaker who wishes to communicate something to others, or to influence them, or make a request of them, or to engage in deliberation with them, and so on. It might be that they lack the resources for their speech to have any effect on others in a public context. It might be that their viewpoint is disregarded because the group they belong to is publicly disesteemed or marginalised, as in the case of a vilified minority ethnic group. A wife may face a husband who routinely ignores or overlooks her speech: her negative right to free speech is intact, but it is empty. The assumption underlying these examples is that the point of free speech is to address others and possess some traction in doing so; a person’s speech, to have value for them, needs to have the potential for uptake. Without this, free speech, for some at least, is a liberty without worth (Rawls 1999: 179). Exactly what this implies remains to be determined; our claim for now is only that a cogent view of free speech should address the issue of substantive opportunities to speak.

In sum, there are four criteria by which we shall assess extant theories of free speech: specialness, stringency, no internal balancing and opportunity.

3. Critique of the classic theories

In this section we present the three classic theories of free speech in a brief and stylised way, and evaluate them according to the four criteria. Our purpose in doing so is to communicate our scepticism and thus to motivate the basis for looking at free speech afresh.

Truth

John Stuart Mill’s (2006) famous consequentialist argument for free speech says that its purpose is to render our views and opinions on contested matters more well-grounded; only through their susceptibility to challenge and critique can those opinions be refined and improved, and progress be made towards the discovery of truth. Free speech is a necessary instrument to achieve these aims. While Mill’s consequentialist argument can explain the specialness of free speech compared to other liberties more generally, the scope this covers will be limited to cases where attaining more cogent views, or at least improving our deliberative capacities (Brink 2008), are involved. In many cases, however, the conditions in which these two aims are realised involve relatively strict regulations on speech, ruling out, for example, fake news or the transmission of culturally orthodox views. In other words, what renders free speech special may often be better achieved be regulating, rather than protecting, certain forms of speech.

When it comes to stringency, it seems clear that based on Mill’s consequentialist argument, little if any protection should be enjoyed by low-value speech such as pornography. Conversely, while it is hard to disagree that speech on controversial social, political and allied issues, where we seek to improve our views, merits high protection, other intuitively high-value modes of speech seem to fall outside the scope of the consequentialist argument. Consider, for example, ritualistic speech within a closed religious order. Such speech is concerned with ultimate issues of truth and meaning but, as far as religious adherents are concerned, these questions are already settled. If such speech has high value, its source cannot lie in its contribution to an open-ended and critical process of truth discovery.

Consequentialism involves seeing an individual’s well-being as valuable not in itself but rather as an element in a larger sum. This implies that neither the interests of speakers nor the interests of listeners have any independent weight separate from the calculation of overall utility for the community to which they belong. Consequentialism therefore cannot avoid, and in fact involves, a process of internal balancing through agglomeration; the truth-related interests of speakers and of those whom their speech affects are integrated into a single theoretical structure. While free speech might help promote truth discovery and critical reflection by some, it might hinder the achievement of that goal for others – for example, those who end up embracing false opinions as a result of their free circulation within society (e.g. see Schauer 2012).

As far as opportunity to speak, our final criterion, is concerned, the consequentialist theory is silent. On a more deliberative interpretation of Mill’s argument, one could perhaps make the argument that individuals merit substantive opportunities to cultivate their deliberative capacities, but that would be a teleological goal, not a matter of right. And, in any case, the argument would not seem to warrant such opportunities for those whose speech will do little to contribute to more cogent or correct views.

Autonomy

The autonomy theory of free speech can be presented in various ways – as a procedural listener-based theory, for example, as in Scanlon’s (1972) earlier work, or as a substantive ideal. C. Edwin Baker (2011) presents a speaker-centred substantive view, one which regards opportunities to speak as bound up with the ideal of a self-governing life. We employ speech in deliberating with others about our aims and aspirations, and of course in pursuing self-chosen projects in collaboration with others. Furthermore, speech is the means by which a person communicates her own thoughts to others, and considers their responses. If we interpret autonomy as a substantive ideal, then it is not difficult to see how it could demand that individuals have the opportunity, not just, the formal freedom, to speak. But on the other criteria, the autonomy approach is less clear.

For a start, the ideal of individual autonomy militates against persons being beholden to or under the sway of others in any area of their lives, therefore free speech does not present any distinctive specialness in this theory. There does not seem any categorical distinction, that is, between my limiting your autonomy by preventing you from communicating your views to others and other kinds of limits to autonomy such as curtailments to freedom of movement or freedom of religious conscience (Schauer 2015). At the same time, some forms of especially valuable speech do not seem to involve the speech of substantively autonomous agents. The ritualistic speech of religious adherents we mentioned earlier is a case in point. So too is the inflammatory speech that we hear from speakers at political rallies, or at least the audience’s instinctive reaction. Like the former example, one could make the case that this kind of speech is especially valuable, even though it does not seem to be an instance of autonomous agency.

Finally, it seems clear that the autonomy approach too is susceptible to the problem of internal balancing. Phenomena such as hate speech and pornography arguably impinge upon the autonomy of those they target or affect. Thus, we are left having to choose between (and judge the significance of), on the one hand, curtailing the autonomy of potential speakers in order to protect that of listeners (including third parties) or, on the other hand, protecting the autonomy of speakers with the known consequence that the autonomy interests of listeners are impinged upon. This is the internal balancing we referred to earlier, when the same basic value counts both in favour of and against free speech.

Democracy

The democratic defence of free speech, first defended by Alexander Meiklejohn (1948), calls attention to the way in which the circulation of ideas in a society’s public culture is bound up with citizens exercising their democratic role of evaluating political activity and helping determine the state’s proper direction. As with the other approaches, the argument comes in a number of forms. Post (2011) maintains that citizens cannot reasonably regard themselves as authors of the law unless they have had the substantive opportunity to influence the course of public discussion. Weinstein (2011) and Dworkin (2009) argue for a stronger view according to which the exercise of political power is illegitimate unless citizens have had the opportunity to state their view, including speaking out against potential laws and proposals. It is commonly accepted that on the democratic argument speech should be interpreted in relatively broad terms to include not just formal political speech but also literature and the arts, cultural commentary, and social critique.

If democracy is plausibly interpreted as a substantive ideal, not merely a procedural value, then it is not difficult to see how it could ground an argument for citizens’ opportunity to speak. Speech is influence, and unequal speech means that wealthy and/or well-connected citizens have more say over political outcomes than less advantaged ones, a syndrome of virtually all current liberal democracies, and one that damages their democratic credentials. Rawls (2001) makes a similar argument in his defence of the fair value of political liberties.

The democratic argument also seems to meet the specialness criterion, by showing why free speech demands a special place in the pantheon of liberties: it is clearly more serious for government to forestall discussion of political matters than to introduce traffic regulations, for example (cf. Dworkin 1977: 267-9). But while the democratic approach meets the first criterion, it does so by conflating the protection of speech as such with the protection of high-value political speech, thus contravening the stringency criterion. More specifically, political speech, broadly construed, is the kind of speech which merits protection on the democratic view. Cass Sunstein (1993), who defends the democratic approach at some length, argues for example that the dyad political/non-political speech marks the distinction between speech which may only be regulated on the most stringent time, place or manner bases and speech where content-regulation aptly applies. But that distinction seems under-motivated. Scientific speech, for instance, is not inherently political, but would be disfigured by content-regulation. Conversely, the claim that at least some hate speech should not be subject to content-based regulation, due to its political nature, seems to beg the question central to the stringency criterion. Likewise, when it comes to the problem of internal balancing, similar issues apply as in the case of autonomy. Victims of hate speech who are silenced or intimidated, for example, are often likely to be excluded from democratic participation. But their loss of speech must be balanced against the free speech rights of those who want to use their political rights to spread hate (cf. Reid 2020).

4. Relational view: The recognitional dimension

The relational view of free speech we set out in this and the next section begins from the thought that speech is characteristically addressed to another subject. We speak, write, and in other ways communicate to other individuals. (Private speech, as in keeping a secret diary, is the limit case: here we speak to ourselves). That speech addresses others is a conceptual claim. However, the claim also has a number of normative implications.

To begin with, if one party says something to another, she claims permission to speak; that is, that she has the liberty to determine what to say and in what manner, when to say it, and to whom. This applies equally to written as to verbal communication; except that in the former instance the audience for our speech may not be immediately known to us. By ‘permission’, we mean that the speaker takes herself to have a liberty right to speak: she does not owe a duty not to speak to any third party (Hohfeld 1919). For example, a junior officer who questions her senior’s order may face a reprimand, but she takes herself to have moral right to speak back nonetheless.

Second, in addressing another we take ourselves to enjoy a kind of moral autonomy. The autonomy here is procedural rather than substantive, as we considered above. It means that the views we convey are ours, and not another agent’s. Whether we address others simply as persons or as role bearers within institutions, we do not take ourselves simply as an instrument to convey others’ views. Procedural autonomy is not contradicted by individuals being instructed or ordered to deliver one person’s views to another. Instructions and orders are not sufficient reasons for an autonomous person to convey another’s view; rather, they are reasons which stand in need of justification for autonomous persons and which may be outweighed by sufficiently compelling contrary considerations.

Third, and by implication of permission and autonomy, individuals regard themselves as answerable for their speech: we regard ourselves as accountable for what we communicate; and may rightly be praised, challenged, criticised etc. in consequence. We do not for instance regard hate speakers as merely cultural dupes. This is equally true of written as it is of verbal speech: a politician is accountable for the views on their blog, a writer is responsible for their interpretation of history, and so on. Answerability presents both a general aspect, concerned with moral accountability per se, and a particular aspect when we interact with each other as role bearers such as family member, religious observant or work colleague. Through our speech with fellow role inhabitants, we answer for how far we have met the responsibilities and expectations consequential for our roles. Being recognised by others as a role bearer in good standing just does involve discharging one’s role-related responsibilities (cf. Scheffler 2001). Meeting and discussing these responsibilities necessarily involves speech: thus, we are answerable too in this more concrete sense through our role-governed speech–how two parents decide to raise their children, for instance.

Permissibility, autonomy and answerability are implicated in the notion of free agency as such: we are accountable for all our conduct, not just our words. But only in the case of speech do individuals implicitly present these normative notions to another agent in the hope and expectation that they will be confirmed by that agent’s response. To speak, in other words, is to raise a claim about our liberty to do so, our ownership of the words we use, and our accountability for them. Listener responses will typically vindicate that claim. Suppose A says to B, ‘It’s going to rain later’, and B disagrees, ‘No, it’s not!’. By disputing the content of A’s words but not denying her authority to make them, B recognises A’s liberty. She also takes A’s views as her own. In disagreeing with A, B takes her to be answerable for her views. In this typical example of speech, therefore, A implicitly takes her speech act to have normative presuppositions, and B, while vehemently disagreeing with it, accepts those presuppositions herself. It is as if we transmit this normative frame of speech, as well as the concrete content of speech, every time we open our mouths (or put pen to paper). Disagreement is one way of accepting the presuppositions of another individual’s speech; so too is enthusiastic agreement, vigorous criticism, and even disbelief or ridicule. All these are modes of uptake, and at root speech as a form of address involves a claim to uptake, a claim which must be accepted by our interlocutor.

In speaking to another person, or again in writing, whether for a concrete or unknown readership, we make parallel assumptions about the various kinds of normative authority our audience possesses. When A says to B ‘It’s going to rain later’, A implicitly assumes B’s right to hear and, in all but the rarest circumstances, she assumes B’s right to respond to this statement. She assumes too that it is up to B to autonomously determine what to do with this statement, whether to agree or disagree, change the subject, and so on, and that however B does respond, by speech or conduct (e.g. she picks up her umbrella), she is answerable for how she does so. In addressing another individual, therefore, we confirm her possession of just those normative capacities which, when speaking, we claim for ourselves.

We can appreciate this point more clearly if we consider how speech can go wrong, where there is an attack on or a denial of the normative capacities implicated by speech as a form of address. For example, one individual might coercively intrude into another individual’s very liberty to speak, or to listen to others, a denial of the latter’s permission. Or, through the use of more subtle social coercion, they may succeed in dissuading another from speaking out, a point familiar from feminist criticisms of pornography (see e.g. Bonotti and Seglow 2021, Ch. 5). Alternatively, a person may not regard another’s speech as genuinely representing her views; for example, a person may claim to know what a would-be speaker’s true interests are even though the latter asserts that her interests are different. Finally, someone may not regard a speaker as genuinely answerable for her views or, more likely, not take herself to be answerable to the claims, requests or demands that another actor makes of her – a form of silencing.

All these wrongs are failures of uptake; failures to confirm, through speech, another speaker’s normative authority qua participant in speech. If, for example, an ethnic minority is deemed to be untrustworthy and their members’ speech is dismissed; or if (some) men do not regard themselves as answerable to women’s speech, then we are in the presence of a speech-related wrong. The notion of uptake does not imply that there is a right to a response every time one communicates to others through speaking or writing. But where receivers of speech fail to register speakers’ permission, autonomy or answerability in a repeated, pervasive or systematic way, then some speech-related injustice has occurred. Failures of appropriate uptake in instances like these are likely to be structural, arising from – but also compounding – deeper injustices such as discrimination, unequal opportunities and lack of political voice. We believe that there are duties to ameliorate the structural barriers which systematically deny some groups’ speech uptake with others via the lack of recognition of their normative authority which failures of appropriate uptake involve.

The flip side of our claim that lack of speech uptake is wrongful is the positive claim that mutual confirmation is where the value of speech lies. As we explained, in contrast to most forms of conduct, speech is always addressed to another agent. We conceptualise it as a kind of reaching out. By communicating to another, an individual raises a claim that she enjoys normative authority to speak – that she is free to do so, autonomous in doing so, and answerable for the consequences of her speech. The authority claim is immanent in the substantive content of her speech, whatever that content is, and it is a claim made to another individual which can only be vindicated by that individual’s uptake, uptake which cannot be guaranteed and may be systematically thwarted in the ways just described.

In speaking to another person therefore we present to them (indeed cannot help but present to them) our practical attitude towards our normative authority along the dimensions of permission, autonomy and answerability, while at the same time confirming their possession of those same dimensions of normative authority – that is, confirming them as a speech participant, to put it simply. If our speech partner responds to us in standard ways, then she vindicates the practical relation to self that we have presented to her in our speech. The intrinsic value of speech on the relational view is that speakers mutually confirm each other’s normative authority in speech practices; they do so whether they agree with, disagree with, question or criticise each other’s speech. Speech, therefore, is a form of recognition; at the highest level of generality, recognition of each other’s status as free, autonomous and answerable participants in a shared social world (cf. Honneth 1992; cf. McQueen 2022). Though clearest in the case of face-to-face verbal exchange, this is equally true of written speech; individuals’ reactions to the written word are more diffuse and delayed than with the immediacy of face-to-face conversation but every writer writes for an audience who can confirm her status as a free, autonomous and answerable agent.

In most cases, speech is doubly relational as two or more parties communicate to each other so each is both a speaker (or writer) and a receiver of speech. We call this a discourse, and it ranges from a brief exchange in a shop to an intimate conversation between partners, to a discussion among work colleagues, to political debate. In discourse, each participant’s uptake of the other’s speech is reciprocated and so they mutually confirm each other’s status as free, autonomous and answerable participants in a shared social world, at least ideally, if there are no uptake failures.

5. Relational view: The republican dimension

Some discourses concern the rules, norms or policies that should be collectively adopted, usually within an institution. We call this subset of discourses deliberations. Deliberations take place within formal political institutions, as well as in other settings where parties raise and contest laws, rules, policies, norms and expectations concerning the general regulation of behaviour, such as in a departmental meeting or a family discussion about the distribution of domestic labour.

The normative significance of deliberation within our relational theory of free speech can be better understood if we turn to the republican tradition of political liberty. In the contemporary republican revival (Pettit 1997, 2012, 2014; Laborde 2008), freedom is conceptualised as ‘non-domination’, in a way that is both social and relational. More specifically, the core insight of republicanism is that to enjoy the status of being a free person it is necessary that others cannot interfere with your liberty in an arbitrary way. The capacity to arbitrarily interfere with a person’s liberty, even if it is not exercised, means making that person’s will subordinate to one’s own, or being able to dominate another. Individuals who are non-dominated enjoy ‘security in the exercise of…[their]…basic liberties’ (Pettit, 2014: 77) and, therefore, ‘the status or dignity of the free republican citizen’ (Pettit, 2014: 61).

Securing the conditions of non-domination requires, first, that each person enjoys basic liberties and rights which are securely protected and entrenched. This entrenchment must further be a matter of common awareness so that each person’s status as a free (non-dominated) individual is manifest to all. Each person must also be adequately resourced such that she can genuinely exercise her basic liberties. Not all freedoms, however, are basic or fundamental according to republicanism. Instead, republicans assign priority to those freedoms that are co-exercisable (e.g. see Lovett 2016) so that each person is able to enjoy them consistent with others doing so too. In other words, basic liberties – including freedom of speech – should be co-exercisable by all participants in a political community, so that one’s person’s liberty may be constrained in the name of securing or augmenting the liberty of (an)other(s). Entrenched basic liberties give each person a measure of ‘independence from the unconstrained and unauthorised preferences of others’ (Pettit 2014, p.198). With this in place, society and its institutions will satisfy the so-called ‘eyeball test’ (Pettit 2014, p. 99), since citizens will be able to stand tall and look each other in the eye without deference or the need to curry others’ favour. This is where the recognitive and republican dimensions of our relational theory of free speech converge. Let us explain.

The status of being a free person in the republican sense ‘presupposes life in the company of others’ (Pettit 2014, p.198). As Cécile Laborde points out, republicanism overcomes the individualistic traits of liberalism, which implicitly underlie existing defences of freedom of speech, and advances instead a ‘holistic republican ontology [which] takes at its starting point the mutual dependency and mutual vulnerability of individuals-in-society’ (Laborde 2013, p. 522). Republican freedom, in this sense, ‘appears as an intrinsically social and relational good, one which is not achieved through the absence of (certain types of) institutions and relationships but rather through the presence (and active fostering) of the right kind of institutions and relationships’ (Laborde 2013, p. 522). Republican liberty is relational in the general sense that it requires non-dominating relationships with others. More precisely, non-domination requires that each person enjoys ‘recognition as a subject with a voice and an ear of one’s own’ (Pettit 2001, p. 90), one who has a say in the rules and laws that govern her and her community; who shares in the system of popular control with others who recognise her right to do so, where this is affirmed in law and also accompanied and supported by corresponding social norms (norms that the law itself can contribute to producing and sustaining over time) (Pettit 2012, pp. 128–19).

Neither Pettit nor other republican theorists have done much to consider the implications of their theory for freedom of speech (though see Pettit 1994) but we believe that these are significant. To begin with, it is worth noting that central to republicanism is the view that laws as such should not be considered a limitation of our liberty if they are appropriately authorised, where this means that they stem from a non-arbitrary system of popular control (Pettit 2014 p. 198). In this sense, rules that regulate speech should not be considered a limitation of freedom of speech – in fact, they may enhance it. Consider, for example, an association’s meeting which needs to arrive at a collective decision. The chair says at the start, ‘I want to hear everyone’s point of view’, the implication being that she will encourage everyone to speak and will cut people off if any of them seek to dominate the discussion. She thereby regulates speech based on non-arbitrary rules contained in the (democratically agreed upon) association’s charter. By doing so, we believe that the chair, far from interfering with/limiting free speech, in fact enhances it, that is, she enhances co-exercisable freedom (as non-domination) of speech for all, provided at least that those rules are non-arbitrary and emanate from a system of popular control rather than from her personal whim.

The right to free speech, understood as co-exercisable freedom as non-domination of speech for all, is not only the potential object of regulation and popular control. It can be itself essential for non-domination and popular control. Citizens living under republican institutions should each enjoy the ability to shape the direction of the authorities which issue laws on their behalf. Democracy as popular control enables (rather than hinders) our liberty, and is the opposite of arbitrarily imposed law. Free speech is a basic right for republicans because it is essential to any kind of social relation in which participants enjoy the status of non-domination – they are not subordinated, marginalised or oppressed – by enabling them to speak out and contest collective decisions, as well as to exercise their other rights such as freedom of association. On the republican view, therefore, free speech is a relational liberty and, through the exercise of voice (Laborde 2008, p. 152), central to a system of popular control and contestation. Under republican political arrangements, one speaks to others in order to exercise one’s right to a say, and one listens to others in order for them to be able to exercise their right to a say: mutual uptake is a requirement.

In summary, both recognition theory and republicanism help us to conceive free speech in relational terms and do so in complementary ways. Recognition theory addresses the relative neglect in republicanism of the fact that free speech confirms an individual in her personhood or role through her addressee registering via her uptake the permission, autonomy and answerability implicit in the individual’s communicative acts. Conversely, republicanism addresses recognition theory’s relative silence on freedom as a social relation and on political decision-making. Moreover, republicanism is equipped to identify structural injustices such as democratic deficits or gender hierarchies where uptake failures are systematic and pervasive.

6. The four desiderata redux

Our four desiderata for a cogent justification of free speech were that it vindicates its status as a basic liberty; shows what forms of speech merit especially stringent protection; avoids an internal balancing of considerations favouring and limiting speech; and defends the egalitarian intuition that individuals should enjoy some opportunities to speak.

The recognitive dimension of the relational view explains speech’s special status over other forms of conduct. Though permission, autonomy and answerability are implicated in conduct in general, much conduct does not raise a claim on others to recognise those capacities in us. Consider actions that do not involve a speech component such as walking in the street or eating food. In the case of speech, however, we both issue a claim on others to recognise our normative authority and recognise that authority in them, as explained in Section 4. Such authority is valuable yet vulnerable; an attribute of our practical-moral status which needs affirmation by others to be properly realised. This gives speech its fundamental status.

In discourses which are also deliberations on common norms or rules, participants’ ability to maintain their undominated status as equal persons free from arbitrary interference and subordination is at stake. Co-deliberation on shared rules and norms is high-value speech, and on the republican view should enjoy especially stringent protection. (This includes cultural and artistic speech as these too reflec on how society’s members are regarded.) From the recognition perspective, we also saw how some speech is bound up with the responsibilities of being a role-bearer: a religious adherent or scientist working on a common project, to return to two examples from Section 3. This speech too merits stringent protection. Non-deliberative discourses and other speech outside these categories, such as commercial speech or everyday interactions where responsibilities are not at stake, remain instances of basic liberties, but of a lower category.

The recognitive approach addresses the balancing issue through the idea that each individual enjoys the same normative authority as every other. Unlike the substantive autonomy perspective of Section 3, it could imply that speakers who engage in (for example) hate speech claim illegitimately greater authority to speak than they legitimately possess, though that argument needs spelling out. Meanwhile, the republican thesis that basic liberties must be co-exercisable in order to secure each person’s equal and independent status offers a direct answer to the internal balancing problem. Speech which diminishes others’ speech interests renders the latter more burdensome to exercise: thus, the co-exercisability requirement favours each person’s equal liberty to speak.

We earlier drew attention to how systematic uptake failures eroded the bases for individuals to have their normative authority confirmed. By the same logic, as uptake failures undermine recognition, an individual who enjoys the formal liberty to speak, but lacks the practical ability to have her views registered by others – whether family, workmates or fellow citizens – possesses an empty freedom. To repeat the point, there is no right to consideration every time one attempts communication. Yet there remains a threshold below which speaker opportunity is absent, or at least insufficient. Thus, the recognitive approach offers the germ of an answer to the final, opportunity criterion. This answer is augmented through the republican insistence that individuals enjoy the capacity to contest rules and norms which would (further) subordinate, marginalise or oppress them. Only through all affected persons having a real voice of their own can such contestation occur; this requires substantive opportunities to communicate, beyond their formal freedom.

7. Hate speech and fake news

The implications of the relational account of free speech can be further explored by applying it to two specific kinds of speech: hate speech and fake news.

On the standard view of it, hate speech attacks the basic civic standing of its targets, communicating that they are not genuine members of the political community (e.g. Waldron 2012; see also Bonotti and Seglow 2021, Ch. 3). By doing so, hate speech fails to recognise its victims as agents with their own views and interests, denies that they enjoy any normative authority as discursive subjects, and usually disrespects their civil rights, including their right to freedom of expression. The message of hate speech is that some people, on account of their ethnicity, race, religion etc. are not worth engaging in discourse with, because they are not properly members of the political community or even, in extreme forms of hate speech, not properly persons. Furthermore, insofar as hate speech helps fuel an atmosphere where some people are regarded as having a lesser status, even by those who (unlike hate speakers) have no special animus towards them, it also indirectly erodes their status as equal members of the political community with the right to engage in deliberative discourses with others. All of this constitutes a direct failure to consider hate speech’s targets as discursive equals who have permission to speak and the capacity autonomously to develop their own views. Furthermore, hate speakers are unlikely to recognise that they are answerable to others for their speech.

One’s status as a discursive subject of free speech is dismissed when one is the object of speech that is silencing, humiliating or threatening. Most instances of hate speech share these features and, therefore, fail to recognise their targets as discursive subjects with normative authority. Regulating hate speech can ensure that its potential victims are appropriately recognised in speech and discourses, and that they can participate in popular control by speaking out and contesting collective decisions. Crucially, from the perspective of our relational theory, the regulation of hate speech should not be considered as a restriction on the free speech of the silenced hate speakers, and therefore as an issue of balancing, but rather as a way of guaranteeing co-exercisable freedom of speech for all. Under the relational view, and particularly under its republican dimension, it is not arbitrary to be interfered with (in the exercise of free speech) in order to guarantee the undominated freedom of speech of others, or to secure their status as discursive subjects. This point signals at its best the importance of co-exercisability in our relational conception of free speech.

Fake news provides another interesting example of speech that contravenes the key normative tenets of our relational approach. There is disagreement regarding the precise meaning of fake news. For example, not everyone believes that fake news is always driven by political goals (Rini, 2017, p. E45), and there is ongoing debate as to whether fake news is effective thanks to structural flaws or to individual cognitive biases, or both (Chambers, 2021; Rini, 2017). However, most scholars agree that fake news tends to involve two key features: those who employ it deliberately aim to mislead their audiences in order to achieve some ulterior goal, and they do so by formally mimicking (without effectively complying with) journalistic conventions to present ‘news’ which is (largely or completely) false (see Aikin and Talisse, 2018; Gelfert, 2018; Rini, 2017).

It seems evident that fake news disrupts citizens’ capacity to arrive at well-founded views and subverts the practice of discourse and political deliberation by misinformation. Like hate speech, though with a more epistemic dimension, fake news fails to recognise the normative authority of its audience as discursive subjects. Those who produce fake news claim permission to speak, but are not concerned as to whether their audience also has liberty to speak, since communication in fake news tends to be mostly unidirectional, not discursive. Likewise, it seems that the use of fake news entails a failure to recognise the autonomous agency of the audience, since it involves exploiting the audience’s cognitive biases. Recipients of fake news are normally considered as instruments to convey the fake news producers’ views rather than as agents capable of autonomously developing and conveying their own views. Furthermore, fake news producers tend to deny that they are answerable to anyone for their speech, since the scrutiny that comes with accountability would likely undermine the effectiveness of their fake news stories. These considerations suggest that, like hate speech, fake news also violates the key normative tenets of our relational view of free speech and, therefore, may be legitimately regulated. Once again, regulation in such cases promotes rather than restricts free speech by ensuring that all citizens are recognised as discursive subjects co-exercising their undominated freedom of speech.

8. Conclusion

Most of the recent academic scholarship on free speech has focused on establishing whether and why certain types of speech should sometimes be regulated, largely neglecting the more fundamental question of why free speech is valuable and merits protection in the first instance. In this paper, we have tackled the latter question and argued that an adequate theory of free speech must meet four normative desiderata: specialness, stringency, no internal balancing and opportunity. None of the traditional theories of free speech (based on truth, autonomy and democracy), we argued, meets all four desiderata. We defended a new relational theory of free speech, which grounds its value in the recognition that speakers grant each other when engaging in speech practices, and its limits in the republican ideal of non-dominated co-exercisable liberty. Our theory, we concluded, meet all four desiderata. We also briefly illustrated the implications of our argument for the regulation of hate speech and fake news. Due to the limited space available, we have only been able to sketch the main features of our new theory. Unpacking and elaborating on its key elements, and applying it to other free speech debates and controversies, is a task that will need to wait until another day.