Just Words? Hate Speech, Harm, and the Justifiability of Legal Regulation

1. Introduction

Questions concerning what hate speech is, how it harms, whether it is protected by free speech principles, and how it might be legally regulated have been at the centre of debates about free speech for some time. In the lead paper in this collection, Mary Kate McGowan develops an important but underappreciated perspective on these arguments, providing a new account of how ‘ordinary people under ordinary circumstances can unwittingly oppress others with their everyday comments’ [McGowan 2022a: 129]. This central claim is an intriguing one for several reasons. Most obviously, while there has been some consensus in the free speech literature that the speech of people in positions of authority can be regulated without unduly restricting expression—a manager can be sanctioned for announcing that his company will no longer hire women, for example—the claim that anyone can oppress others with their speech in a sense that warrants legal regulation is a controversial one, at least in the US context that McGowan inhabits.

McGowan presents a series of nuanced arguments to explain the mechanisms by which ordinary people in ordinary circumstances ‘enact harmful norms’. In brief, this is possible because our actions are ‘contributions to norm-governed social practices’ [ibid. 130], practices in which the relevant norms are always evolving. The motivating analogy is the practice of conversation. For McGowan (drawing on the influential work of Lewis [1979]), each move in a conversation changes the operative norms, whether we realise it or not. This ‘sneaky mechanism of norm enactment’ is often overlooked, but no less important for that. And it applies to other social practices as much as conversation.

McGowan deftly unpacks this complex yet familiar mechanism by drawing a distinction between what she calls g–norms and s–norms. In the case of conversation, g–norms are the more general norms that govern all conversation, such as norms of cooperation, syntax, and grammar. s–norms, by contrast, are those enacted by particular conversational contributions; these contributions make changes to the ‘conversational score’, or what moves in the conversation are permissible. Enacting such a change (i.e., enacting an s–norm) need not be the speaker’s aim, and it does not require any special authority. McGowan’s claim is that this phenomenon is generalisable to other social activities. When we engage in social practices, we are also making moves in norm-governed activities. We are, inevitably, enacting s–norms in these practices. Norm enactment is not a special capacity of powerful or privileged speakers; it is ‘ubiquitous’.

This framework is then applied to the vexed case of racist speech in public spaces. The example at the centre of the discussion is that of an African American family who find themselves subjected to a very public racist tirade at a restaurant by a white woman (WW). In the US this speech would normally be protected under free speech law. WW is ‘expressing a political viewpoint’; she has no special authority, so her speech cannot suffice to institute a discriminatory norm; the speech may be offensive, outrageous, and even harmful, but it is not regulable. McGowan thinks otherwise. By combining the account of norm enactment with a novel ‘parity argument’, she suggests that this speech can be legally regulated under hostile environment anti-discrimination law, because it constitutes a verbal act of discrimination. Indeed, it should be regulated: the current US approach to hate speech is legally ‘insupportable’.

In this introductory essay, we focus on clarifying some central features of McGowan’s paper and the critical discussion that follows. We will be especially interested in issues about the nature of norms, the distinction between causing and constituting harm, and the viability of the parity argument. We conclude with a brief reflection on the comparatively unusual free speech context of the United States, and how McGowan’s framework bears on the political morality of regulating hate speech in other jurisdictions.

2. Conversation and Other Norm-Governed Social Practices

McGowan emphasises the subtle, seemingly effortless, and everyday moves that are made in the context of conversations, moves that function to, in her view, enact s–norms. Here is one of her examples. In conversation with Greg, I express happiness for the success of a former classmate, Paul, who, as Greg knows, treated me terribly in school. By expressing happiness for Paul’s success, I set the tone for the subsequent conversation with Greg in a way that has normative consequences. For instance, I make it impermissible (or at least less appropriate than it might have been) for Greg to say anything disparaging about Paul. I don’t have any special authority to enact this norm; nor did I necessarily aim to enact it. This is what makes s–norms ‘sneaky’. The account helps explain why, as Robert Simpson [2022: 163] puts it, ‘certain kinds of speech do more than is immediately apparent’.

WW’s speech is doing something similar, says McGowan. Though she has no authority, and though she may not have intended to enact a discriminatory norm, she has altered the ‘score’ in the norm-governed practices of social interaction in a public accommodation. She has licensed racist speech by enacting an s–norm that permits it. Suppose that her contribution emboldens others to engage in racist speech. Then the s–norm she has enacted constitutes harm, in McGowan’s admittedly special sense. Later we will discuss the import of this claim in more detail.

Of course, WW’s speech may not lead to further racist behaviour. There are many ways to respond to score-changing moves. Reflecting on McGowan’s views about conversation particularly, Robert Simpson [ibid.] illustrates with an example. Robert is talking with his friend Amir. They are speaking about helping a mutual friend, Bill, seek help for a mental health crisis. At one point in the conversation, Amir says: ‘oh, wait, he actually told me just the other day; there’s a mental health organisation downtown that does free group therapy sessions . . . ’ [ibid.: 164]. The reference to ‘he’ is not to Bill, as it should be given what has transpired, but to Amir’s co-worker Carlos, with whom Amir is infatuated. Amir’s move has violated an s–norm and enacted another one—‘he’ now refers to Carlos rather than Bill—but in this case his interlocuter has, as Lewis puts it, ‘accommodated’ him. However, Robert could have just as easily ‘blocked’ the move by correcting Amir, reminding him that they were talking about Bill and expressing annoyance at his distracting obsession.

As Simpson reminds us, we often do not block these violations, but simply adjust and go along with them. In accommodating rule violations in this way, we can be ‘shepherded’ into following s–norms that we would otherwise prefer not to follow. So there is a subtle but potentially coercive element to this aspect of conversation, ‘a feeling of being gently bullied into going along with your interlocuter’s moves’ [ibid.: 165]. Though extending this insight about conversation to the case of WW is not trivial—among other things, her tirade is not a conversation—there is plausibly an insight here about the subtle coercive power of hateful speech.

But just as one can block the violation of an s–norm in everyday conversation, can’t we also block the violation of (what we may assume to be) the standing anti-racist s–norms in the restaurant? Put differently: even if the tirade enacts a new racist norm, can’t those present in the restaurant simply dismiss it and neutralise its normative force? As Jeremy Waldron [2022] notes, other patrons and employees might respond by recording the tirade and publicly shaming WW on social media. Caroline West [2022] suggests that in cases like this, WW’s speech does not constitute a hostile environment in the relevant sense. We return to related issues when we discuss the parity argument below.

Whether hateful speech leads to more hateful speech will depend not just on the immediate responses of others present, but also on the background practices that already exist in the social context, and the norms that govern them [Waldron 2022: 172–3]. What exactly those background g–norms are is not entirely clear, though the analogy to conversation is a helpful starting point. More discussion of how the norms governing social practices differ from those governing conversation potentially opens new avenues that may deepen our understanding of the harms of hate speech.

In fact, McGowan is not centrally interested in circumstances where newly enacted harmful norms encounter resistance. That’s because harmful s–norm enactment is not sufficient for the kind of harm constitution she aims to diagnose. For a harm to be constituted in her target sense, two conditions must obtain: an s–norm must be enacted which prescribes harm; and harm must result from adherence to this norm. In the case that motivates her analysis, WW’s speech has not only enacted a racist s–norm in the restaurant, but also ‘triggered’ racist g–norms already operating in the background, leading to further poor treatment of the black family, say by other patrons.

3. Norm Enactment vs. Norm Application

A key premise in McGowan’s argument is that s–norm enactment is ubiquitous. Against this premise, Waldron [2022] suggests that McGowan may mistake norm enactment with norm application. This is important because norm application does not amount to norm constitution, and McGowan’s ingenious legal suggestion depends on the claim that some racist speech constitutes discrimination. Waldron [ibid.: 176] also suggests that the ubiquity of s–norm enactment may render it ‘tautological’. McGowan rightly responds to the latter charge by noting that harmful s–norm enactment is not ubiquitous, and by noting that the ubiquity of s–norm enactment is compatible with its theoretical significance. However, the worry that McGowan conflates norm enactment and norm application cuts deeper.

To understand and evaluate Waldron’s objection, we need to say something about the nature of social norms. Waldron borrows a plausible picture from H.L.A. Hart [1994: 55]: a norm is enacted or emerges; it has a reasonably accessible content; it is operative among a group of people; there is an event or act which falls under the scope of the norm; and finally, there is the invocation or enforcement of the norm. Hart’s example of hat wearing in church serves to illustrate. There is a norm about the wearing of hats in church that has either emerged formally (e.g., through enactment by a religious authority) or informally over time; the content of the norm is that when a man enters a church, he must remove his hat; the norm is widely followed, and has an ‘internal’ aspect; an event falls under the norm, (i.e., a man walks into a church); and the norm is enforced when a member of the community sanctions the man (by frowning, reprimanding, etc.).

Where does WW’s speech sit in this schema? Has WW enacted a norm, or simply applied or enforced one? How does the distinction between g–norms and s–norms relate to the distinction between norm enactment and norm application?

Consider again our example. There’s a g–norm that says: If you’re a man and you enter a church, then you ought to remove your hat. Waldron claims that when a man walks into a church, his behaviour does not enact this norm. It merely triggers the norm’s application by satisfying its condition. McGowan agrees. But she claims that this behaviour does enact a norm: ‘an s–norm that he ought to remove his hat’ [McGowan 2022b: 218]. g–norm application, she claims, typically involves s–norm enactment.

One concern here is that McGowan’s response commits her to a significant proliferation of norms. Every time a man walks into a church, a new norm gets enacted. And these new norms appear explanatorily inert. The original g–norm suffices to explain the relevant normative result that, in every case, the man who has entered the church ought to remove his hat. Whether or not s–norm enactment is ‘tautological’, it is legitimate to ask what these posited s–norms are doing, given that they do not seem necessary for explaining the relevant normative facts.

Here is another way to formulate the idea. Suppose that the g–norm is objectionable: the community is wrong to expect men to remove their hats in church. It does not seem true that each time a man removes his hat in church, he thereby brings into existence an additional objectionable norm. On the contrary, each man appears to be following an objectionable g–norm that already exists. A critic of the norm might well say to the man: ‘It is a stupid norm that requires men to remove their hats in church’. They might also say: ‘By following this stupid norm, you give it more power’. But they would not, and should not, say: ‘By removing your hat, you are bringing another stupid norm into existence’. The natural response will be that this charge is too strong. The objectionable norm already exists.

This can seem an arcane issue in the metaphysics of norms. But it is central here because McGowan’s argument requires that behaviour like that of WW enacts harmful, discriminatory s–norms in public spaces: ‘Harm constitution requires norm enactment’ [ibid.: 134]. If such behaviour merely applies harmful, discriminatory g–norms that are already operative in the social context, then it does not enact new norms. As we discuss in the final section of this essay, it is chiefly the US legal context that makes this point essential, since in other jurisdictions the ‘mere causation’ of harms of certain kinds is sufficient for regulating speech.

However, even if we reject McGowan’s picture of norm enactment in the case of hat wearing churchgoers, and by extension the claim that g–norm application typically involves s–norm enactment, we might still defend McGowan’s conceptualisation of the WW case. After all, the relevant norms in the latter case have a different character. The g–norms governing public space interactions are, as we have noted, opaque and complex. They include some norms that systematically disadvantage non-whites. And it seems likely that the WW’s tirade changes the score of public interaction in the restaurant in a way that is distinctive, that is, it is not a mere application of the content of a racist g–norm. After all, a milder, more subtle bit of racist behaviour would have changed the score in a different way. In other words, the normative import of WW’s behaviour is not redundant. So perhaps McGowan is right to insist that some racist behaviour in public enacts new norms.

4. Causing and Constituting Harm

It should be clear that McGowan needs the distinction between speech that causes harm and speech that constitutes harm. As she and the commentators note, the definition of constitution here is rather unusual. It is really a special form of causation: ‘speech constitutes harm when it enacts a norm that prescribes that harm and harm ensues via adherence to that norm’ [McGowan 2022b: 211].

Ishani Maitra [2022] worries that McGowan cannot sustain the distinction. Consider ‘persuasive speech’. Some persuasive speech harms: it persuades people of harmful beliefs and/or persuades them to participate in harmful practices. But it would be a mistake, according to both McGowan and her critics, to classify this speech as constitutively harmful. On the contrary, persuasive speech is supposed to be a paradigm case of speech that causes harm without constituting it. However, McGowan’s claims about harmful norm enactment threaten to misclassify such speech as constitutively harmful. If this is right, then it is evidence that the requisite distinction may be impossible to draw.

Consider the case of Gingerphobia and McGowan’s analysis of it. Mick argues publicly that redheads are evil. His speech persuades some people; they come to believe what he says, and some harm redheads as a result. Does Mick’s speech cause harm without constituting it, as McGowan suggests? His persuasion involves speech acts like assertion, which, according to McGowan, invariably enact s–norms. For example, his assertion that redheads are evil will, by McGowan’s lights, make it (more) appropriate for his interlocutors to believe that redheads are evil. But then Mick’s speech appears to satisfy the definition of norm constitution: it enacts a norm that prescribes the harm of being gingerphobic, which leads to harm when it is followed. What was supposed to be a paradigm of non-constitutively harmful speech begins to look like a paradigm of harm constitution. This is puzzling.

McGowan appreciates the force of this objection, but thinks she can defuse it by distinguishing between social norms (which interest her) and norms of ‘reason and deliberation’ (which do not). ‘When we abide by social norms, we do so because others expect us to and because there will be some sort of cost to not abiding. When we abide by the norms of good reasoning, by contrast, it’s not like this’ [McGowan 2022a: 214]. How does this observation help to show that Mick’s speech does not, after all, enact a harmful norm in the relevant sense?

McGowan appeals to a general distinction between the proper mechanics of belief formation, on the one hand, and acting on social norms, on the other. If people come to believe Mick’s assertions, then they do so ‘based on reason and deliberation and not (just) because they are abiding by social norms’ [ibid.]. Whereas if people come to act on social norms, they do so on the basis of social expectations alone. Here is her most intriguing formulation:

We do not come to believe something because others expect us to and because we will pay a social cost for not believing; that is not how belief formation works (at least when it is working as it should) but it is how social norms influence us. Persuasion then, when properly deliberative and rational, anyway, does not result from abiding by social norms. [ibid..]

This preliminary response to the objection is quite interesting. We will note a few features that deserve further exploration.

First, it is unclear why McGowan invokes a distinction between the proper mechanics of good reasoning and the proper mechanics of action production. If people come to believe Mick’s assertions, then something improper is going on, because redheads are not evil. This is not surprising: beliefs are often formed improperly; persuasion is often less than perfectly rational. Similarly, if other patrons in the restaurant follow the WW’s lead and racially abuse the black family, something very improper is going on. The relevant fact about Mick’s assertion is that it does, according to McGowan, make the improper beliefs more rational. His assertion changes the score of his interlocutors’ theoretical reasoning. That is, on our understanding, how norm enactment is being conceived. When this norm also leads to harm, it seems to constitute harm in McGowan’s sense.

Second, McGowan’s emphasis on the distinction between belief and action is confusing in this context. Mick’s harmful speech persuades people to believe and also to act on their gingerphobia. When they act in a way that harms redheads, they act on the basis of their (newfound) beliefs, which have been licensed by Mick’s assertions. Again, this seems to satisfy the definition of harmful norm constitution. Indeed, it seems quite similar in the relevant respects to McGowan’s motivating example. When other restaurant patrons or employees act in ways that harm black customers, they act on the basis of their (newfound) beliefs about what’s acceptable at the restaurant, beliefs which have been licensed by WW’s racist diatribe.

Third, McGowan’s claim about why we act on social norms is quite controversial. One version of the claim is that when we act on a social norm, we act because others expect us to, and because there will be some sort of cost to not acting on the norm. But there is more going on here. Consider a racist restaurant patron who spews venomous speech at black customers after being emboldened by WW’s example. This patron need not act because others expect them to, or because there is some social cost to not spewing racist speech. They might instead act because they are racist and derive some pleasure, or sense of righteousness, from their speech. The WW’s newly enacted s–norm facilitates the expression of the patron’s racism, as McGowan claims. But it does not facilitate their racism by making it socially costly to refrain from being a racist. Still, the patron’s racist expression is a central case of the harmful consequences of racist norm enactment.

McGowan may be right that social norms are connected to the actions they cause, or help to cause, in distinctive ways. And she is surely right that social norms are importantly different from the norms that apply to reasoning and deliberation. However, it would be useful to clarify how these observations can help to defend the distinction between causing and constituting harm that McGowan’s argument needs.

5. The Parity Argument

A key part of what makes McGowan’s argument innovative is the invocation of hostile environment anti-discrimination law. Her parity argument aims to establish that some speech by ordinary speakers in public—for example, the racist speech of WW—is morally analogous to speech already prohibited in the United States on anti-discrimination grounds. In one revealing and forceful formulation, she asserts that the reasons for regulating speech like WW’s are ‘exactly the same reasons that justify the regulation of other uncontroversially regulable categories of speech’ [ibid.: 129]. The viability of McGowan’s overall proposal thus depends on her interpretation of the foundations of hostile environment law.

Though we cannot give a comprehensive analysis of the United States jurisprudence surrounding this important category of legal protection, it will be useful to briefly describe its contours before reflecting on the parity argument.

‘Hostile environment’ is a concept in United States employment law which applies to certain forms of workplace behaviour. Specifically, hostile work environment claims arise when an employee alleges workplace harassment that satisfies the following conditions: it is unwelcome and offensive to the complainant; it is objectively offensive, such that a reasonable person would find it offensive; it is severe or pervasive enough to change the conditions of employment; it is due to the employee’s protected class status (such as their race, religion, sex, disability, age, or veteran status); and it is ‘attributable’ to the employer. Federal courts now take this form of harassment to be prohibited by Title VII of the Civil Rights Act (1964), which expressly prohibits discrimination in the terms or conditions of employment based on membership in a protected class.

So when exactly does unwelcome behaviour in the workplace constitute a hostile environment and hence legally actionable discrimination? One main subtlety concerns whether an instance or pattern of conduct meets the ‘severe or pervasive’ condition. This is determined on a case-by-case basis, with several relevant factors considered—including the conduct’s frequency, whether it was threatening or humiliating, how it affected the employee’s psychological well-being, and whether the harasser was a superior in the organisation.

With this legal framework in mind, we can make some observations about how McGowan’s proposal amounts to an extension of hostile environment anti-discrimination law, and offer some thoughts about what might be usefully addressed in future elaborations of these ideas.

First, the WW’s outburst is an isolated incident. Seldom will a single incident be found sufficient for creating a hostile workplace environment in the relevant legal sense, though this can happen when the conduct is particularly egregious. McGowan takes the conduct of the WW to be a paradigm of the sort of racist public abuse she believes ought to be regulated. It is reasonable to wonder, then, whether her proposal amounts to a weakening of the ‘severe or pervasive’ criterion. McGowan [ibid.: 145] considers an example that is especially connected to this observation, the case of Craig v. New Crystal Restaurant, in which the Chicago Commission on Human Relations found a single abusive utterance in a restaurant sufficient for constituting discrimination.

A second point concerns the move from workplaces to public accommodations. A public accommodation under US law is a place that offers goods and services to the general public. More specifically, public accommodations are businesses and/or facilities: government owned facilities and buildings (such as courthouses, hospitals and parks) and privately owned businesses and buildings (such as restaurants, hotels, and theatres). Religious organisations, or private clubs that require memberships, do not count as public accommodations. While the Federal Civil Rights Act (1964) prohibits discrimination in any public accommodation so defined, it does not prohibit WW from saying what she does, so long as the African American family are not prohibited from entering the restaurant. This is one reason McGowan’s suggested extension of hostile environment protections makes sense given her political values.

Still, the transition from workplaces to public accommodations raises important legal questions. In the employment context, conduct in a workplace must be severe or pervasive enough to change the conditions of employment (in a way that denies equal opportunity) in order to be actionable. What is the analogous condition which applies to conduct in a public accommodation, where the victim is a patron rather than an employee? McGowan’s discussion [ibid.: 143] suggests that the conduct must make the patron unwelcome (in a way that denies equal opportunity). It is true that WW’s victims may well feel unwelcome as a result of her vile behaviour. As West [2022: 181] notes, however, this is not enough for the parity argument. The relevant question is whether the racist outburst makes it the case that the black family is in fact unwelcome in the restaurant, in the technical sense articulated by hostile environment law. It is hard to see how it could do so on its own. Suppose the restaurant owner promptly asks WW to leave, apologises to the victims, offers them a complimentary meal, and urges them to return to the restaurant in the future. These patrons are not rendered unwelcome in the restaurant in the relevant sense, notwithstanding the seriousness of their victimisation and the likelihood that they will feel unwelcome as a result of WW’s speech (as the empirical evidence relating to victims’ experiences of racist vilification attests) (see e.g. Lawrence [1993]; Gelber and McNamara [2016]).

Third, neither WW nor the black family she abuses are employees of the restaurant. There is one undeniable sense in which the black family is (legally) entitled to non-discrimination in the restaurant: everyone is entitled to equal access to public accommodations. But it is less obvious how the restaurant’s proprietors could have vicarious liability for the WW’s actions, given that she is only a patron. Employee actions are (sometimes) said to be attributable to an employer because of the special nature of the employment relationship. This justification for vicarious liability does not straightforwardly extend to the different relationship between patron and public accommodation, though it might be that a similar justification could be developed. On this score, see McGowan’s [2022a: 146] reflections on the case of Neldaughter v. Dickeyville Athletic Club, in which abuse from bystanders at a softball game was found by the Wisconsin Labor and Industry Review Commission to constitute a hostile environment in a public accommodation. The Commission suggested that if the Dickeyville Athletic Club had owned the land, or in some other way had the authority to remove the harassers, they would have been legally responsible. Of course, McGowan could instead suggest an alternative legal mechanism for the regulation of hate speech that does not depend on vicarious liability. But this would require more radical departures from public accommodations law.

Fourth, there are additional differences between workplaces and public accommodations which seem relevant to how anti-discrimination law in these contexts may be justified. Most obviously, we are materially reliant on employment in a way that we are not reliant on access to any single public accommodation. Though the conduct of the WW is inexcusable, the black family is still free to leave the restaurant and dine elsewhere. Employees are also free to leave their jobs, but doing so tends to be far more costly. In sum, employees are more existentially vulnerable to discriminatory conditions in the workplace than patrons are to discriminatory conditions in a public accommodation. This asymmetry does not necessarily show that hateful speech in public accommodations cannot or should not be regulated in the United States, but it may complicate McGowan’s claim that the very same justification for regulating speech could apply in both contexts.

Finally, it must be said that the legal cases cited as precedent by McGowan are unusual, and controversial as a matter of US law. The sanctions imposed by local human rights commissions might well be invalidated were the cases brought to federal courts, which would be tasked with evaluating the tension between these sanctions and First Amendment protections.

6. Legal Regulation

Setting aside concerns about the parity argument, there remain various more general questions about whether and how hateful speech might be regulated. There is always a risk that laws regulating speech will over-reach, leading to unjustified or unproductive state interference; or that they will be so toothless as to be meaningless in the fight against discrimination. Despite these considerations, we do, as McGowan points out, regulate various kinds of speech deemed to be harmful—in laws prohibiting defamation, discrimination, inciting violence, and so on. If speech such as WW’s enacts discriminatory norms, and if hostile environment anti-discrimination law should also apply to public accommodations, then there is reason to think that hateful speech could be regulated without any fundamental shift in US law.

McGowan’s account is ingenious and complex because its ambition is to vindicate legal regulation of hate speech in the especially speech-tolerant context of the US. The commentators wonder, rightly, whether it would be advisable to regulate speech in the suggested ways even if her account succeeds jurisprudentially (see Lengbeyer [2022]; Richardson-Self [2022]). These are important issues. But there is also a broader question about legal regulation that McGowan’s account makes salient. What is the best legal scheme for regulating hate speech? As we conceive of this question, it is a matter for comparative political-legal philosophy, informed by empirical social science. And we believe that McGowan’s account may, perhaps inadvertently, help to reveal the superiority of approaches to regulating hate speech that seem unlikely to pass legal muster in the United States.

Australia is useful here as a point of comparison. Section 18C of the Commonwealth Racial Discrimination Act 1975 (‘Racial Discrimination Act’) makes it unlawful to ‘offend, insult, humiliate or intimidate’ another person or a group of people because of their race. This section was introduced to the RDA in 1995 in response to the recommendations of several major inquiries, such as the National Inquiry into Racist Violence, the Royal Commission into Aboriginal Deaths in Custody, and the Australian Law Reform Commission’s report, Multiculturalism and the Law. These inquiries found that those who are targeted by racist harassment had no legal recourse under existing common law. The provision was also justified on the basis of a hypothesised causal relationship between racial abuse and violence: the escalation of unchecked racial vilification could, it was suggested, contribute to violent action against targeted persons and groups.

This is very different law from McGowan’s proposal. The rather technical issues about whether harms are constituted seem irrelevant: by the lights of section 18C, it doesn’t matter whether the speech leads to further harmful behaviour or is likely to do so. It is also irrelevant where the speech takes place: it is illegal whether it occurs on the street, in the workplace, or in a public accommodation.

In our view, regulation of hate speech in something like the Australian way is likely morally better than piecemeal regulation or no regulation at all. This contentious view cannot be adequately defended here, but two observations seem germane. We are moved, first, by considerations of parity. If racist abuse is harmful enough to regulate when it occurs in a restaurant, as McGowan believes, then similar abuse should also be regulated when it occurs on the street. We are also moved by the thought that regulations like section 18C respond directly to the speech’s wrong-making features. If racist abuse is the relevant harm (and we think it is), why does it matter whether this abuse leads to further abuse, or whether the abuse takes place in a workplace, a restaurant, or the ocean? Though McGowan’s arguments for regulation in the US are legally controversial, they are of necessity more modest than the regulatory regime already operative for several decades in Australia. This regime expresses the moral perspective of the state, which is that racist abuse is seriously wrongful and will not be tolerated. In other words, racist abuse is discriminatory on its own.

There are well known concerns about hate speech codes such as section 18C. To the extent that these concerns are warranted, they might in principle tell in favour of more limited regulations such as McGowan’s, although that seems at best uncertain. In our estimation, however, the concerns often voiced about hate speech codes—for instance, that they licence promiscuous, arbitrary, draconian policing of expression in a way that is poisonous for democracy—are hard to sustain given the actual historical record, at least in the case of Australia.

Section 18C is certainly controversial. In 2014, the then Attorney General George Brandis famously proposed to repeal the provision on the grounds that ‘people have a right to be bigots’. However, there is no sense in which it has opened the floodgates to litigation. Section 18C makes racial vilification unlawful but it does not make it a criminal offence. The majority of claims made under section 18C are either conciliated through the Australian Human Rights Commission or are withdrawn or dismissed. Of the cases that are conciliated, many end with a simple apology or a small fine. A case would only proceed to the Federal Court or the Federal Circuit Court if conciliation fails, and this only happens in a very small number of cases. For example, of the 86 complaints received by the Commission in 2015–16, only one proceeded to court.¹

In fact, some argue that it is so rarely used as to be irrelevant. Yet it is not irrelevant: minority groups have for decades consistently defended section 18C when conservatives have suggested repealing it; and the provision enjoys remarkable popularity among the Australian public more generally.² We believe that the law serves an expressive function that is socially valuable even if the sanction is relatively minor. What it expresses is that racist speech is unacceptable in Australian society (irrespective of how widespread it is); that such speech causes both direct and indirect harms to targeted groups; and that there are avenues of redress for those who are subjected to such speech.

The legislation does not infringe on free speech rights; in fact, it is one of the few legislative provisions to explicitly protect freedom of speech in Australia, because section 18D allows for defences if the speech is ‘reasonably and in good faith’ intended for artistic, academic, scientific, or other public interest purposes. In several cases considered by the courts, speech has caused racial offence, humiliation and intimidation, but been found to be protected under section 18D because the speech was deemed to be an artistic work or involved a good faith engagement with matters of public interest.³ Nor is the threshold for harm too low, as many critics claim. To fall within section 18C, the speech must have profound and serious effects. In Eatock v Bolt⁴, Bromberg J held that:

The kind of likely offence is not to be likened to ‘mere slights’. It has the real potential to lower the pride and self-image of the person or group attacked and thereby limit the participatory equality in the affairs of the community which the group and its members are entitled to enjoy. Conduct with these consequences threatens the dignity assurance which all citizens are entitled to be accorded. The reactions which I have concluded were reasonably likely, are not reactions likely to be caused by the intolerance of the people affected.

Nor has section 18C actually prevented people from expressing themselves in bigoted ways, as Brandis’s claim suggests. In fact, Gelber and McNamara’s [2016] Australian study reveals that experiences of racially and religiously motivated abuse are routine and frequent, that they occur in public places and involve direct and indirect expressions. But what section 18C signals is that despite the frequency of vilification in Australian society at the micro level, society at a macro level does not tolerate these views.

Of course, law is not the only answer, and informal norms play just as important a role in dismantling the structures of racism and bigotry. But law is a powerful institution, able to make interventions at the level of g–norms, to use McGowan’s terms, and its role should not be discounted as part of a complex, multi-faceted response to countering utterances such as those of WW.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 https://humanrights.gov.au/about/news/opinions/ahrc-and-racial-discrimination-act-setting-record-straight

2 One Fairfax poll in March 2014 found that 88 per cent of people agree that the provision should remain as is. See https://humanrights.gov.au/about/news/opinions/ahrc-and-racial-discrimination-act-setting-record-straight

3 See Bropho v Human Rights and Equal Opportunity Commission [2004] FCAFC 16 (February 2004); Kell-Country v Beers & Anor [2004] FMCA 336 (21 May 2004); Walsh v Hanson [2000].

4 [2011] 197 FCR 261, 330.