Conflicting Conceptions of Hate Speech in the ECtHR’s Case Law

Abstract

This article argues that the ECtHR uses two conflicting tests to assess the same types of hate speech. This results in legal uncertainty at best, and arbitrariness and double standards at worst. To remedy the present situation, I propose a two-track strategy. To begin with, the Court should abandon its “bad tendency” approach, a test prone to abuse by governments to silence political dissent under the guise of fighting hate speech, for a set of uniform criteria to assess hate speech-restrictions, in line with its current incitement approach. In addition, however, to compensate for the loss of protection against severe forms of vilification which do not meet the incitement-criteria, the Court should formulate a new category of unprotected speech, to be defined as intentional intimidation or harassment.

A. Introduction

Debates about the proper balance between the right to freedom of expression and the interests underlying hate speech legislation are as old as human rights law itself. Both the opponents and advocates of regulating hate speech provide powerful arguments in favor and against state intervention.¹ Critics argue that the regulation of hate speech results in censorship of political speech and that efforts to suppress hate speech are ineffective at best, and counter-productive at worst. Supporters of hate speech legislation point to the mental and physical harms caused by hate speech, its “silencing” effect on those targeted, and its propensity to lead to discrimination and violence. While the intellectual battle between the two sides seems to have settled into a stalemate, new developments have sharpened the stakes of the debate. On the one hand, the rise of illiberal democracy presents new challenges for freedom of expression. There is ample evidence, also from the case law of the European Court of Human Rights, that illiberal and authoritarian states frequently abuse hate speech laws to silence political opponents and target the very vulnerable groups and minorities they seek to protect.² In addition to direct government censorship, free speech interests are negatively impacted by intermediary liability laws and the content moderation policies of privately-owned social media platforms, where general legal standards do not apply.³ On the other hand, it is equally clear that both the proliferation and harmful effects of racist, sexist, homophobic, and other hateful messages is greatly amplified by the rise of social media and populist politics.⁴ A growing body of social science research suggests a correlation—and possibly a causal link—between online hate speech and offline violence.⁵

Faced with these new realities, the importance of developing a predictable and consistent legal framework for addressing hate speech—one that strikes the right balance between the conflicting interests at stake—has only increased. The thesis of this article is that such a framework is currently missing from European human rights law. What is worse, the European Court of Human Rights applies different, opposing standards, to assess free speech restrictions based on domestic hate speech legislation.

There is no better way to introduce this point than to start with two examples drawn from the Court’s recent case law. In Lilliendahl v. Iceland ⁶, the first case, the Court reviewed a homophobic comment below an online article defending LGBT+ counselling in elementary and secondary schools. The impugned message ran as follows: “We … have no interest in any explanation of this sexual deviation …. This is disgusting. To indoctrinate children with how sexual deviants copulate in bed.”⁷

The second case, Savva Terentyev v. Russia ⁸, concerned a blog post written in response to a controversial police action at the offices of a local newspaper in the run-up to a regional election in Russia. It included the following words:

Who becomes a cop? Only lowbrows and hoodlums – the dumbest and least educated representatives of the animal world. It would be great if in the centre of every Russian city, on the main square … there was an oven, like at Auschwitz, in which ceremonially every day, and better yet, twice a day (say, at noon and midnight) infidel cops would be burnt. The people would be burning them. This would be the first step to cleansing society of this cop-hoodlum filth.⁹

At the domestic level, both statements were qualified as hate speech and punished accordingly. The Strasbourg Court, however, was only willing to qualify one of the two messages as hate speech for the purposes of Article 10 of the European Convention of Human Rights (“the Convention”). The homophobic statement in Lilliendahl was found to be “severely hurtful and prejudicial” and to amount to a “form of hate speech,”¹⁰ whereas the call for the annihilation of police officers in Savva Terentyev did not qualify as incitement of hatred or violence. Such a conclusion is remarkable considering the words used in Sava Terentyev were clearly much more aggressive and hostile in tone—the ceremonial incineration of “infidel cops” in “Auschwitz-[like]” ovens—than the ones under review in Lilliendahl—“sexual deviation” and “disgusting.” Of course, one might object that the European Court, when assessing an interference with freedom of expression, not only considers the content of the statements at issue, but also looks at the context in which they were made and the severity of the penalty imposed. Yet, the circumstances of the two cases were very similar. The authors were unknown members of the general public,¹¹ they did not release their statements on prominent platforms or highly visited websites, and they were not, in the Court’s view, likely to draw wide public attention.¹² Moreover, the impugned messages were not specifically directed at vulnerable groups of persons—such as children¹³—and the sentences imposed were relatively mild—a fine and a suspended prison sentence.¹⁴

What, then, accounts for the different outcome in both cases? To be sure, there is one relevant distinction between the two judgments: The speech in Lilliendahl involved members of a sexual minority, a group which has historically been subjected to marginalization and victimization, while the speech in Sava Terentyev was directed against the security forces of the State.¹⁵ Although this certainly is an important point, it should not be overestimated in this specific case. As one observer noticed, Iceland is listed at the top of the OECD reports on LGBT+ acceptance and the actual risk posed by a homophobic comment of a 74-year-old isolated bigot, completely detached form social reality, was non-existent.¹⁶

If the content and context of the two statements alone cannot convincingly explain the contrasting outcomes, the question becomes what can. It is submitted that Lilliendahl and Sava Terentyev are paradigm cases of the two different standards used by the European Court to assess interferences with Article 10 in the numerous hate speech cases decided in the last decade. The first approach allows the authorities to censor speech that attacks persons “by insulting, holding up to ridicule or slandering specific groups of the population,”¹⁷ whereas the second method requires the State to prove that the offensive speech is “likely to provoke imminent unlawful actions” and exposes the target to a “real risk of physical violence.”¹⁸ The former can be labeled the “bad tendency” test and can be traced back to the cases of Féret v. Belgium ¹⁹ and Le Pen v. France.²⁰ The latter goes back to the case of Zana v. Turkey ²¹ and Sürek (No 1) v. Turkey ²² and can be identified as the incitement approach.

This article argues that the simultaneous use of these two conflicting tests to assess the same types of speech results in legal uncertainty at best, and arbitrariness and double standards at worst. To remedy the present situation, it proposes a two-track strategy.²³ To begin with, the Court should abandon its bad tendency approach, a test prone to abuse by governments to silence political dissent under the guise of fighting hate speech, for a set of uniform criteria to assess hate speech-restrictions, in line with the current incitement approach. In addition, however, to compensate for the loss of protection against severe forms of vilification which do not meet the incitement-criteria, the Court should formulate a new category of unprotected speech, to be defined as intentional intimidation or harassment. Whereas the incitement test is aimed at tackling the harmful effects of the spread of hateful views on society at large, the intimidation approach would seek to protect individuals against the direct, personal harm caused by certain types of speech. Inspiration for the latter can be found in a line of cases, starting with Aksu v. Turkey,²⁴ in which the European Court laid down an obligation for the States to take positive measures against “direct verbal assaults” or “physical threats” immediately affecting a person’s selfconfidence and psychological well-being.²⁵ The advantage of this two track-model is that it would offer clear guidance to states and private companies, allowing them to simultaneously fulfill their human rights obligations of effectively fighting hate speech while respecting the right to freedom of expression. The remainder of this article consists of four parts. Parts B to D describe the Court’s bad tendency, incitement, and intimidation standards, respectively. Part E concludes with a defense of the two-track model.

B. The Bad Tendency Approach

I. Origins

As mentioned in the introduction, the reasons for suppressing hate speech are manifold. When contemplating the adoption of a legal framework, it is instructive to distinguish between two types of harm associated with hate speech. To begin with, there is the psychological or physical damage to the direct targets of hate propaganda. In addition to the immediate injury caused to identifiable victims, hate speech may be subject to restriction in order to protect against the indirect or long-term effects of such expression. The bad tendency approach is concerned with the second type of harm. It allows lawmakers to restrict speech that might possibly produce certain social harms, such as violence, discrimination, or hatred towards certain persons or communities, at some indefinite future time. The bad tendency test has a long history. It can be traced back to the English common law and the First Amendment jurisprudence of the U.S. Supreme Court, which used it to justify the punishment of left-wing and anti-war propaganda in the first part of the twentieth century.²⁶ In Europe, the bad tendency formula emerged much later. Its European origins lie in the twin cases of Féret v. Belgium and Le Pen v. France, decided by the Strasbourg Court in 2009.²⁷

The applicants in Féret and Le Pen were the leaders of extreme right-wing political parties in Belgium and France. Both had been convicted for disseminating anti-immigration and anti-Islam propaganda. Examples included leaflets maintaining that a center for refugees and asylum seekers would disrupt the life of local residents,²⁸ deploring the use of taxpayers’ money to support foreigners rather than nationals in need of assistance,²⁹ opposing the “Islamization” of Europe,³⁰ as well as slogans such as “Belgians and Europeans first!” and “the day there are no longer 5 million but 25 million Muslims in France, they will be in charge.”³¹ The European Court began by observing that the impugned messages were xenophobic in nature, since they focused on the cultural differences between Belgian and French citizens and the targeted minorities, and further represented those groups as criminal-minded.³² In the Court’s opinion, such language was to be categorized as hate speech. This qualification was based on the fact that the expressions involved were “susceptible to instill” and “of such a nature as to arouse” feelings of rejection, hostility or hatred against the targeted community, “particularly among the less informed members of the public.”³³ There was a danger, the Court further observed, that the applicant’s “vexatious proposals” risked instigating reactions which were “incompatible with a serene social climate and could undermine confidence in the democratic institutions.”³⁴ In other words, the mere “tendency” of the message to persuade the members of the public to adopt hateful and discriminatory attitudes was sufficient to warrant suppression.

At this point it is already important to note that the Court in Féret and Le Pen considered neither the likelihood of the dangerous consequences of the statements and proposals, nor the intention of the speakers to bring them about. As former Judge Tulkens explained, under the Féret line of reasoning a message qualifies as hate speech, if it “may create a climate, an environment in which actions that were not possible before become possible.”³⁵ States are not required to demonstrate that the words incite or that the speaker intended to incite others to take action against certain groups or persons. In Féret, the intention of the author to incite hatred was inferred from the mere tendency of the words used, whereas in Le Pen the issue of intention was not even considered.³⁶

II. Recent Examples

The central tenet of the Court’s bad tendency approach is best captured by the following observation:

The Court reiterates that inciting to hatred does not necessarily entail a call for an act of violence, or other criminal acts. Attacks on persons committed by insulting, holding up to ridicule or slandering specific groups of the population can be sufficient for the authorities to favour combating racist speech in the face of freedom of expression exercised in an irresponsible manner.³⁷

This statement, first announced in Féret, recurs in numerous subsequent cases in which the Court sanctions the punishment of speech falling short of intentional incitement. To illustrate this point, reference can be made to recent cases relating to homophobic, xenophobic, and terrorism-related speech.

A case in point regarding homophobic speech is Lilliendahl v. Iceland, discussed in the introduction. The “hurtful” and “prejudicial” nature of the comments was sufficient to qualify them as hate speech.³⁸ Another case in which the Court extended the findings of Féret to offensive speech against members of the LGBT+ community was the 2012 judgment of Vejdeland and others v. Sweden.³⁹ The content of the impugned statements was similar to the ones under review in Lilliendahl. Homosexuality was described as “a deviant sexual proclivity” that had “a morally destructive effect on the substance of society.”⁴⁰ It was further alleged that homosexuality was one of the main reasons why HIV and AIDS had gained a foothold, and moreover that the “homosexual lobby” tried to play down pedophilia.⁴¹ Contrary to Lilliendahl, the homophobic message was not released on the Internet but distributed via leaflets, left in the lockers of a secondary school. Although the Court explicitly acknowledged that the aim of the leaflets was to start a debate about the lack of objectivity of education in Swedish schools, and admitted that the statements did not encourage individuals to commit hateful acts, it nevertheless concluded that the “serious and prejudicial allegations” amounted to hate speech.⁴² Referring to Féret, the Court recalled that hate speech does not necessarily entail a call to certain acts, and that there is an obligation to avoid statements that are “unwarrantably offensive to others, constituting an assault in their right.”⁴³

The bad tendency standard was also applied in several cases concerning xenophobic speech. In all these cases, reference was made to Féret. For instance, in Atamanchuk v. Russia ⁴⁴ the Court reviewed the conviction of a politician and journalist for inciting hatred and debasing the dignity of a group of people on account of their ethnicity.⁴⁵ The applicant had published an article in which he criticized the widespread corruption of the Russian state. The text also suggested that “non-Russian” groups were to blame for the suffering of the people of Russian origin, and that members of those groups had engaged in criminal activities, getting “their hand into others’ pockets” and being capable to “burn, slaughter, rape, rob and enslave, in line with their barbaric ideas.”⁴⁶ Reminding that “insulting, holding up to ridicule or slandering specific groups of the population” can be sufficient to warrant interference, the Court found no violation of the Convention.⁴⁷ The concurring opinion of Judge Lemmens perfectly captured the Court’s approach. He explicitly acknowledged that the applicant’s articles “can [not] be read” as “a call for hatred, violence or intolerance.”⁴⁸ Nevertheless, the “clearly xenophobic” nature of the message, Lemmens continued, was sufficient to decline Article 10 protection.⁴⁹ A similar approach was taken in Sanchez v. France.⁵⁰ In that case, the Court upheld the conviction of the mayor of a French town for not removing a comment from his Facebook wall in which a political opponent was criticized for “transforming Nîmes in Algeria.” Another comment associated Muslims with drug-related crimes. The Court reiterated that incitement to religious hatred does not necessarily entail a call to unlawful acts, and the intention to incite can be deduced from the virulent nature of the words used.⁵¹

A final set of cases illustrating the consequences of the bad tendency approach concern terrorism-related speech. Two representative examples are Bayar and Gürbüz v. Turkey ⁵² and Altintas v. Turkey.⁵³ The applicants in Bayar and Gürbüz were the publisher and the editor-in-chief of a newspaper in which a statement of Abdullah Öcalan—head of the Kurdistan Workers’ Party—had been published.⁵⁴ Mr. Öcalan was quoted as supporting a newly negotiated ceasefire, emphasizing the need to develop Turkish-Kurdish dialogue. He added, however, that if the dialogue and peace-process would fail—something he did not desire—a transition to guerrilla warfare would be necessary.⁵⁵ In Altintas the applicant was a journalist who had published a small text in a periodical on the occasion of the thirty-fifth anniversary of a hostage taking situation that had taken place in South East Turkey, resulting in the death of both the hostages and the hostage takers.⁵⁶ The perpetrators were labelled “young revolutionaries” and “idols of the youth,” and the response of the security forces were characterized as a “massacre.”⁵⁷ In both cases, the European Court upheld the convictions of the Turkish journalists. In Altintas, the Court recalled, referring to Féret, that Article 10 does not protect speech that can be analyzed as a “direct or indirect call for violence or a justification of violence, hate or intolerance,” and concluded that the text commemorating the hostage taking should be qualified as such.⁵⁸ A similar conclusion was reached in Bayar and Gürbüz. Here the Court held that even though Mr. Öcalan had expressed his support for the ceasefire, his reference to a possible “transition to guerrilla warfare” should be read as a barely implicit threat against the authorities and an instruction to PPK-sympathizers.⁵⁹ A final example of this line of case law is Z.B. v. France.⁶⁰ In that case, the applicant had given his three-year-old nephew a T-shirt with the slogans “I am a bomb” and “Jihad, born on 11 September,” referring to the child’s forename and date of birth. Although the slogans were not directly visible to third parties but discovered by accident when the child was being dressed at school, and although the applicant did not associate himself with any form of terrorism or violence, his conviction for the offence of glorifying terrorism was not held to violate Article 10 of the Convention.⁶¹

III. Concluding Remarks

It is not the purpose of this article to assess whether the Strasbourg Court correctly declined to find a violation of the right to freedom of expression in each of the abovementioned cases. The point it wants to make is that the standard used by the Court to decide those cases not only stands in stark contrast to the incitement test employed in similar cases⁶² but also, more importantly, is too broad and vague to provide meaningful free speech protection. In a previous article, we argued that since the category of speech with a tendency to cause social harm or offend certain groups in society is virtually limitless, the use of the bad tendency test enables governments to infringe on a wide range of legitimate political expression.⁶³

The cases discussed above reaffirm the problematic nature of the bad tendency standard. Not surprisingly, its detrimental effect on the right to freedom of expression was highlighted in a number of concurring and dissenting opinions and scholarly contributions. An important point of concern regarding the terrorist-speech cases is that the European Court allows States to censor and punish journalists and media professionals for merely reporting on the different sides in an armed conflict.⁶⁴ Judge Pavli warned in his dissenting opinion in Bayar and Gürbüz against the inhibiting effects of a free speech-standard which does not sufficiently take into account the context of an expression and, hence, the probability of its harmful consequences.⁶⁵ In a similar vein, Ó Fathaigh and Voorhoof deplored the fact that the Court declined to give Article 10 protections to journalists who, as the majority admitted, had not personally associated themselves with the statements of Mr. Öcalan.⁶⁶ This approach undercuts the seminal Jersild ⁶⁷ judgment, in which the European Court held that journalists could not be convicted for inciting hatred and violence by merely reporting about the racists attitudes and views of others.⁶⁸

Comparable concerns were voiced about the cases involving homophobic expression. In Vejdeland, Judges Spielman and Nussberger, while acknowledging that the homophobic leaflets contained statements that were totally unacceptable, expressed concerns about the test applied by the majority. The obligation, as framed by the majority, “to avoid, as far as possible, statements that are unwarrantably offensive to others” was, according to the judges, too vague and inconsistent with the well-established article 10 case law.⁶⁹ What was further missing was a careful, in-depth analysis of the aim of the impugned message.⁷⁰ In another concurring opinion in Vejdeland, Judge Zupancic indicated that, although he agreed with the judgment given the captive audience situation—the leaflets were left in the lockers of young people—he believed the majority’s conception of hate speech to go too far in limiting freedom of expression by overestimating the importance of what is being said.⁷¹ Similar observations were made about the Lilliendahl judgment. In a comment, Fedele noticed that the Court departed from its settled case law by relying solely on the content of the homophobic statements, disregarding the specific circumstances of the case.⁷² A contextual risk-assessment, Fedele observed, would probably have led to the conclusion that comments posed no real danger to the homosexual community.

All of the above observations reflect the concerns already expressed by Judge Sajó in his dissenting opinion in Féret, the case that introduced this style of reasoning in the Strasbourg jurisprudence:

Content regulation and content-based restrictions on speech are based on the assumption that certain expressions go ‘against the spirit’ of the Convention. But ‘spirits’ do not offer clear standards and are open to abuse. Humans, including judges, are inclined to label positions with which they disagree as palpably unacceptable and therefore beyond the realm of protected expression. However, it is precisely where we face ideas that we abhor or despise that we have to be most careful in our judgement, as our personal convictions can influence our ideas about what is actually dangerous.⁷³

C. The Incitement Approach

I. Origins

Given the contested nature of the bad tendency standard, it is not surprising that alternative, more speech-protective methods, have been proposed. What is remarkable, however, is that these alternative standards are not confined to academic writings or dissenting opinions but found their way to the decisions of the majority of the European Court.

The origins of the alternative approach lie in the case law on terrorism-related speech. In a long line of cases, going back to Zana v. Turkey and Sürek (No. 1) v. Turkey, the Court decided to protect subversive and violence-conducive speech up to the point that it “incites to violence against an individual, a public official or a sector of the population.”⁷⁴ The incitement test announced in these cases, just like the bad tendency approach discussed above, is concerned with preventing the harmful consequences of an expression for society at large. It allows for the restriction of speech that potentially contributes to violent or discriminatory acts. However, it requires a tighter connection between the expression and the possible dangerous consequences. Harmful tendencies or mere offence are not sufficient to warrant suppression. Punishment will be justified only if the expression, given its content, the context in which it was uttered, and the intention of the speaker, can be qualified as incitement to unlawful action. This standard resembles the U.S. Supreme Court’s incitement test, announced in Brandenburg v. Ohio.⁷⁵ Under the First Amendment, advocacy of the use of force or of law violation cannot be punished, except where such advocacy “is directed to inciting or producing imminent lawless action and is likely to incite or produce such action.”⁷⁶ As the analysis below illustrates, the different components of this test made their way across the Atlantic, albeit the conditions of imminence and likelihood are applied rather inconsistently at the European level.

A reconstruction of the history of the incitement test would be beyond the scope of this article.⁷⁷ The current standard was announced in a series of cases against Turkey, decided in 1999.⁷⁸ The central question in each of the cases was whether the challenged utterances “incited to violence against an individual, a public official or a sector of the population.”⁷⁹ It is important to note that the Court’s inquiry focuses not only on the nature and the content of the words used, but also on the intention of the speaker, and the probable effects of the message. A classic example is Sürek (No. 1) v. Turkey, which was concerned with the publication of two readers’ letters in a weekly review, strongly condemning the Turkish suppression of the Kurdish people.⁸⁰ The Court refused to afford Article 10 protection, noting that the statements revealed “a clear intention to stigmatize the other side to the conflict by the use of labels such as ‘the fascist Turkish army’, ‘the TC murder gang’ and ‘the hired killers of imperialism’ alongside references to ‘massacres’, ‘brutalities’ and ‘slaughter.’”⁸¹ The letters, the Court continued, “amounted to an appeal to bloody revenge by stirring up base emotions and hardening already embedded prejudices.”⁸² In addition to content and intention, the Court also looked at the context and the probable effects of the expression, noting the highly sensitive security situation prevailing at the time and in the region where the letters were dispersed.⁸³ Other contextual factors regularly taken into account to assess the probability of the harmful consequences of an expression are the medium used to convey the message and the authority of the speaker.⁸⁴

This multifaceted test, first adopted in Sürek (No. 1) and the other 1999 decisions, was subsequently applied in numerous cases concerning subversive or violence-conducive speech.⁸⁵ In some of these judgments, the Court went as far as to require proof not only of intentional incitement, but also of a “clear and imminent danger,” thus considering both the likelihood and the imminence of the harmful consequence of the message involved. In doing so, the Court implicitly draws on the abovementioned Brandenburg formula and its predecessor, the “clear and present danger” test.⁸⁶ For instance, in the 2010 case of Gül and Others v. Turkey ⁸⁷, the Court decided that a conviction for shouting slogans in support of an armed, illegal organization, violated Article 10.⁸⁸ Although the slogans, taken literally, had a violent tone—such as “political power grows out of the barrel of the gun”—there was no indication that there was “a clear and imminent danger” which would justify interference with Article 10 interests.⁸⁹ In a dissenting opinion, Judges Sajó and Tsotsoria agreed with the majority that in cases where violence, injury or harm to any person is advocated, the danger of such consequences has to be clear and imminent.⁹⁰ In their view, however, shouting violent slogans supporting an armed organization at a mass demonstration potentially satisfied these requirements. A similar approach was taken in Kiliç and Eren v. Turkey.⁹¹ A criminal conviction for shouting the words, “[r]ise up again and again, our president is Öcalan” at al lawful and peaceful gathering, was considered not to pose a “clear and imminent threat.”⁹² In a series of more recent cases, the Court no longer refers to the imminence requirement, but attaches great importance to the intention of the speaker and the capacity of the message to—directly or indirectly—lead to harmful consequences.⁹³ Thus, for instance, a violation of Article 10 was found in Dmitriyevskiy v. Russia—concerning the conviction of a journalist for the publication of articles written by Chechen separatist leaders⁹⁴—and in Mart and Others v. Turkey—concerning slogans supporting an illegal organization.⁹⁵ In many of these cases, the Court satisfies itself with observing that the domestic authorities failed to consider the different components of the incitement test.⁹⁶

II. Recent Examples

Although the European incitement test originated in the case law on violence-conducive speech, it also made its appearance in cases concerning the more typical forms of hate speech relating to, for instance, religious or racial intolerance. As will become clear from the examples discussed in this section, the test is not always applied in a consistent way. Sometimes, the Court requires a clear and imminent danger. In other cases, the focus is on the words used and the intention of the speaker. What these decisions share in common, however, is that the harmful tendencies or the offensive nature of the content of the expression alone is not sufficient to justify an interference with Article 10.

A first obvious example was already discussed in the introduction. In Savva Terentyev v. Russia, the applicant’s call for the physical extermination of police officers, and the use of highly vulgar, derogatory, and vituperative terms in a blog posted on the internet, could not be seen as promoting violence, hatred, or intolerance. In view of the fact that the police are part of the security forces of the State, and should accordingly display a high degree of tolerance to offensive speech, a high threshold was applied: Inflammatory speech cannot be punished, the Court ruled, “unless [it] is likely to provoke imminent unlawful actions” and expose the police “to a real risk of physical violence.”⁹⁷ These requirements were not met, the Court continued, since the applicant’s aggressive statements were rather used as “a provocative metaphor” and an “emotional reaction” against what he saw as an instance of abusive police conduct.⁹⁸ Another example of the reluctance to accept far-reaching restrictions on expressions directed against government agencies, can be seen in Stomakhin v. Russia.⁹⁹ In an unusual caveat, the Court in this case stressed that “it is vitally important that the domestic authorities adopt a cautious approach in determining the scope of ‘hate speech’ crimes and strictly construe the relevant legal provisions,” especially where charges are brought for mere criticism of the Government, State institutions and their policies and practices.¹⁰⁰ The imposition of a five-year prison sentence and a ban from practicing journalism was held to be disproportionate, even though some of the applicant’s utterances were qualified as incitement to violence against the security forces.¹⁰¹

Moving on to the case law about religious intolerance, mentions can be made of Tagiyev and Huseynov v. Azerbaijan ¹⁰² and Maria Alekhina and Others v. Russia.¹⁰³ The first case dealt with an article discussing the role of religion in Azerbaijan and containing several comments on Islam, including: “[M]orality in Islam is a juggling act; its humanism is not convincing”; “in comparison with Jesus Christ, the father of war fatwas the Prophet Muhammad is simply a frightful creature;” “at best, Islam would advance in Europe with tiny demographic steps. And maybe there would be a country in which Islam would be represented by a few individuals or terrorists living incognito” and, finally, “the European philosopher does not act as a clown like the Eastern philosopher, is not inclined to Sufism, or madness, stupidity.”¹⁰⁴ The Court noted that some of these remarks could be seen by certain religious people as an abusive attack on the Prophet of Islam and Muslims living in Europe, capable of causing religious hatred. Nevertheless, recalling the importance of the context, the public interest of the matter discussed and the intention of the speaker, the Court was not convinced that the statements could be qualified as incitement to religious hatred. It was particularly concerned with the fact that the domestic authorities had failed to consider the author’s intention to incite.¹⁰⁵ In Maria Alekhina, the Court reviewed the convictions of the members of the feminist punk band Pussy Riot for hooliganism motivated by religious hatred, on account of preforming a protest song in a Moscow cathedral. The Court accepted that the performance, given the clothes and balaclavas worn and the strong language used, could have been found offensive by a number of people, including churchgoers. Nevertheless, the criteria for qualifying the song as incitement to religious hatred were not met, as the domestic court did not consider the political context and the artistic nature of the performance and failed to examine whether it could have led to harmful consequences.¹⁰⁶

The next example concerns a case in which the Court applied the incitement test to racially motivated hate speech. The applicant in Kilin v. Russia ¹⁰⁷ had been convicted for inciting discord between Russians and people of other ethnicities, in particular Azerbaijanis, by making third-party content available using an account on a social-networking website.¹⁰⁸ The impugned expression included a clip taken from a well-known mockumentary about Russian skinheads and showed a negative encounter between a man dressed as an elderly woman and Azerbaijanis. The elderly woman took a bat in her hands and made several calls, including, “look, maybe it is better to beat the shit out of all those blackies” and “here is how Russians must fight for their rights.”¹⁰⁹ In modern Russian, the word “blackies” is used as a pejorative term in respect of people of certain ethnic origins—Asian, black or from the Caucasus. Another file contained a musical composition “Glory to Russia,” in which non-Russians were presented as enemies that Russians had to fight against. The song contained several calls, including “the blackie shit must be thrown out of Russia” and “come on, guys, let’s put on our bomber jackets and boots!”¹¹⁰

In its lengthy analysis of the case, the Court looked at the different components of the incitement test. Regarding the content of the materials, the Court had no difficulty accepting the domestic courts’ finding that they could reasonably be perceived as stirring up ethnic discord and calling for violence against people of Azerbaijani origin. Next, the Court looked at the intent of the speaker. It attached great importance to the fact that the national courts had convincingly demonstrated the “clear intention” of the applicant “to bring about the commission of related acts of hatred or intolerance.”¹¹¹ The applicant’s mental state could not only be inferred from the strength of the words used, but also from a number of indications relating to his attitude toward the content, his interest in nationalist ideas and his choosing to remain silent during the criminal proceedings.¹¹² The Court noticed that the applicant had no intention whatsoever to contribute to a debate on a matter of public interest. As regards the context and the probable impact of the expression, the Court observed that, even though the applicant was not a well-known user of social media or an influential figure, the sharing of the violent content within an online group of like-minded persons could not be excluded, and may have had the effect of reinforcing and radicalizing their ideas.¹¹³ An interesting aspect of the judgment is that it explicitly acknowledges that there was no indication that the material, in the particular circumstances of the case, was liable to produce “imminent unlawful actions.”¹¹⁴ There was no evidence of a tense atmosphere or any clashes, disturbances, or inter-ethnic riots at the relevant time. The Court opined, however, that the absence of an imminent threat was not decisive in the present case, given the unambiguous nature of the applicant’s criminal intent.¹¹⁵

A final case that deserves to be mentioned here is the Grand Chamber judgment of Perinçek v. Switzerland, decided in 2015 and often referred to in view of its authoritative summary of the main principles of the Court’s case law on violence-conducive speech and hate speech.¹¹⁶ The Court was asked to review the applicant’s conviction by a Swiss court for publicly denying the Armenian genocide. After an in-depth assessment of the nature of the applicant’s statement, their geographical and temporal context, and their impact on the rights of the members of the Armenian community, it concluded that they could not be qualified as a call for hatred or intolerance and were accordingly protected by Article 10.¹¹⁷

III. Concluding Remarks

The purpose of this section was to illustrate the widespread use of the incitement test in the Strasbourg case law regarding hate speech and calls for violence. It will be clear by now that the different components of this multifaceted test are not always applied consistently. This is particularly the case for the condition that the harmful consequences of an expression be imminent and likely to occur. Sometimes, imminence and likelihood are mentioned as independent requirements—as in Gül and Others v. Turkey and Savva Terentyev v. Russia—whereas in other cases these conditions are not separately examined—as in Tagiyev and Huseynov v. Azerbaijan—or considered to be irrelevant—as in Kilin v. Russia.¹¹⁸ What is clear, however, is that the Court, by focusing on the context and the harmful impact of an expression, requires a minimum of potential harm. The pernicious tendencies of an expression are not sufficient to warrant prosecution. In addition, and equally important from the perspective of the right to freedom of expression, is the clear emergence of a separate requirement of intent. The result is a test which is more speech-protective than the bad tendency test.

D. The Intimidation Approach

I. Origins

A third conception of hate speech that can be discovered in the case law of the European Court of Human Rights involves the situation in which a speaker directly targets an identifiable victim. It is concerned not so much with the dangers of harmful persuasion, but with protecting individuals against the direct negative consequences of an expression. It focusses on the immediate, physical or psychological effects of a message on its target(s).

Before explaining the Strasbourg standard for tackling this type of speech, it is instructive to briefly indicate that a similar concern of protecting the direct targets of harmful expression, has led the U.S. Supreme Court to identify a number of speech categories that do not receive constitutional protection. In addition to hostile environment harassment,¹¹⁹ mention can be made of the “true threats” doctrine.¹²⁰ True threats “encompass those statements where the speaker means to communicate a serious expression of an intent to commit an act of unlawful violence to a particular individual or group of individuals.”¹²¹ Such threatening communication can be criminally prosecuted in order to prevent the psychological distress following from the invasion of the victim’s sense of security.¹²² For instance, in Virginia v. Black, the Court ruled that a conviction for burning a cross in front of an African American’s home, with an intent to intimidate, did not violate the First Amendment. There is some uncertainty as to what level of intent is required for a statement to qualify as a true threat. In Virginia v. Black, the Court held that “[t]he speaker need not actually intend to carry out the threat.”¹²³ The purpose of the doctrine is to protect the victim from the fear of violence and from the disruption that fear engenders.¹²⁴ The question remains whether it is sufficient that a reasonable person would regard the statement as a threat, or whether proof is needed that the speaker actually intended to communicate a threat.¹²⁵ What is clear is that the true threats doctrine envisages only individually targeted insults, not political hyperbole or general statements about social and political issues.¹²⁶ Albeit the doctrine stands on different analytical footings than the incitement test, one of the aims for the suppression of true threats is to reduce the likelihood that the threatened violence will occur.¹²⁷

A remarkably similar approach can be found in the jurisprudence of the European Court. On various occasions, the Court has made it clear that “threats arousing a well-founded fear of violence,” “verbal assault and physical threats,” constituting “harassment motivated by racism,” or “intentional intimidation,” do not receive Convention protection. What is interesting, is that these categories of unprotected speech were identified, not in the case law on Article 10 of the Convention, but in a number of decisions concerning the positive obligations arising from Articles 8 and 14 of the Convention.¹²⁸ To understand this line of cases, it is important to keep in mind that the concept of “private life” protected by Article 8 covers the “physical and psychological integrity of a person”¹²⁹ as well as “an individual’s ethnic identity.”¹³⁰ In Aksu v. Turkey, the Court for the first time stated that certain expressions are “capable of impacting on [a] group’s sense of identity and the feelings of self-worth and self-confidence of members of the group.”¹³¹ Personal attacks or certain forms of negative stereotyping, can thus be seen as affecting a person’s private life, provided that they “attain a certain level of seriousness.”¹³² Form this, the Court infers a positive obligation for the States to adopt effective criminal-law mechanisms to ensure protection against verbal assaults and physical threats motivated by racist, homophobic, or other discriminatory attitudes.¹³³

II. Recent Examples

Two recent cases illustrating this approach are R.B. v. Hungary ¹³⁴ and Király and Dömötör v. Hungary.¹³⁵ In both cases, the Court was confronted with domestic failures to investigate and prosecute events that occurred during anti-Roma marches organized by right-wing political parties and paramilitary groups. The applicant in R.B. had been the subject of racist verbal abuse and attempted assault in the presence of her child. The statements directed against her included “go inside, you damned dirty gypsies” and “we will build a house out of Roma blood.”¹³⁶ The Court observed that the applicant felt offended and traumatized by the physical threats.¹³⁷ Taking into account the hostile attitude against the Roma community in the town in which the events took place, the domestic authorities should have provided an appropriate legal remedy for the racially motivated insults.¹³⁸ In a similar vein, the applicants in Király and Dömötör had been the subject of an anti-Roma demonstration during which slogans such as “Roma, you will die” and “we will burn your house down and you will die inside” were chanted.¹³⁹ Certain demonstrators had covered their faces and were equipped with sticks and whips. The Court had no difficulty to find that the demonstrators had threatened the applicants on account of their belonging to an ethnic group and that there was a clear “intention of intimidating.”¹⁴⁰ As a consequence, the decision not to prosecute the perpetrators amounted to a violation of the State’s positive obligations under Article 8 of the Convention.

The case of Beizaras and Levickas v. Lithuania is another good example of the intimidation approach. Under review in this case was the domestic authorities’ refusal to launch pre-trial investigations for incitement to violence against homosexuals, in relation to a series of hateful, homophobic comments posted on the applicants’ Facebook page. The comments were a reaction to a photograph, depicting a same-sex kiss between the two applicants, posted to announce the beginning of their relationship. The picture attracted some 800 comments, the majority of which were hateful. Examples included “fucking faggots burn in hell,” “into the gas chamber with the pair of them” and “they should be castrated or burnt.”¹⁴¹ The Court first observed that these comments affected the applicants’ psychological well-being and dignity. It further noted that the failure to open pre-trial proceedings was based on the domestic authorities’ discriminatory attitude, namely a disapproval of the applicants’ demonstrating their sexual orientation. The Court went on to observe that if “comments such as those uttered in [this] case did not amount to inciting not only hatred but even violence on the basis of the applicant’s sexual orientation, then it is hard to conceive what statements would.”¹⁴² Accordingly, the unwillingness to prosecute of what could only be described as undisguised “calls on attack on the applicants’ physical and mental integrity,” resulted in a violation of the positive obligations inherent in Articles 8 and 14 of the Convention.¹⁴³

Not all of the Articles 8 and 14 cases are concerned with individually targeted threats and insults. The Court has indicated that a State’s positive obligations may also involve a requirement to take action against impersonal forms of hate speech.¹⁴⁴ Although the Court approaches these cases from the angle of Article 8 instead of Article 10, its methodology is generally in line with the incitement approach discussed in the previous section. To begin with, for Article 8 (and 14) to come into play, the statements must rise above a “threshold of severity.”¹⁴⁵ The relevant factors for deciding whether a general, public statement about a social or ethnic group has affected the private life of its members include: The characteristics of the group—its vulnerability or history of stigmatization—the content of the message, the form and context in which it was made, the position and status of the author, and the extent to which it may affect the group’s identity and dignity.¹⁴⁶ None of these factors is decisive; it is the interplay between them that ultimately leads to the conclusion on whether the threshold of severity has been met.

For example, in the twin cases of Budinova and Chaprazov v. Bulgaria and Behar and Gutman v. Bulgaria ¹⁴⁷, the Court held that the domestic courts’ failure to protect the members of the Jewish and Roma communities against the extremely virulent anti-Roma and anti-Semitic propaganda of Mr. Siderov, the founding leader of a far-right nationalist party, amounted to a violation of Articles 8 and 14.¹⁴⁸ As far as the “threshold of severity” is concerned, it was observed, among other things, that Mr. Siderov’s statements “had deliberately been couched in inflammatory terms,” “visibly sought to portray Roma in Bulgaria as exceptionally prone to crime and depravity,” “were conveyed bluntly and repeated many times over,” were “virulently anti-Semitic” and “rehearsed timeworn anti-Semitic narratives.”¹⁴⁹ In short, the speech amounted to a form of “extreme negative stereotyping meant to vilify Jews and Roma in Bulgaria and to stir up prejudice and hatred towards them.”¹⁵⁰ The Court further emphasized that Mr. Siderov was a well-known figure, who communicated his views via many channels, and accordingly reached a wide audience.¹⁵¹ As far as the proper balance between the victim’s right to private life and the author’s right to freedom of expression is concerned, the Court attached particular importance to the intention of the speaker. Mr. Siderov’s statements were held to be “prima facie anti-Semitic in intent” and “prima facie discriminatory in intent with respect to Roma.”¹⁵² In short, they were “meant to” stir up hatred against these vulnerable communities.

In Aksu v. Turkey, by contrast, no violation of Article 8 was found on account of the domestic courts’ dismissal of a series of complaints, brought against the publication of a book entitled The Gypsies of Turkey and two Turkish dictionaries.¹⁵³ The book contained remarks which allegedly debased the Roma community—in particular the chapter providing information about the living conditions of Roma. The applicant had also been offended by some of the entries in the dictionary, which contained negative stereotypes about people of Roma origin.¹⁵⁴ The European Court discerned no racist intentions and no intention to humiliate or debase the Roma community. It further emphasized that the book was written by an academic and was to be considered as a scientific work.¹⁵⁵ Regarding the dictionary, the Court observed that it would have been preferable, as a precaution, to label some of the entries as “pejorative” or “insulting.” Nevertheless, the respondent government did not overstep its margin of appreciation in giving free speech protection to the publications under review.

III. Concluding remarks

It can be concluded that positive obligations to tackle hate speech may arise in two situations. To begin with, States are under an obligation to provide effective remedies against “intentional intimidation” and “physical threats arousing a well-founded fear of violence.” This category of unprotected speech resembles the First Amendment’s true threats doctrine. With regard to the mental element, the European Court seems to allow for both a subjective (“intentional”) as well as an objective (“a well-founded fear”) standard. In any case, for an expression to qualify as a threat for the purposes of Article 10, it must attain a certain “level of seriousness ….” This implies that political hyperbole is not covered. Next, States may be under an obligation to suppress impersonal hate speech targeted at an entire social group. Although the case law is still nascent, it is clear that Articles 8 and 14 will be implicated only if the impugned expression reaches a “threshold of severity.” Moreover, the Court will find a violation of those provisions only in those cases where there was a clear intention to incite hatred.

E. Conclusion

Domestic and international courts have used different standards for determining when governments may suppress hate speech. Depending on how the balance between freedom of expression and the interests underlying hate speech legislation is struck, courts can choose between more or less speech protective tests. The purpose of this article was to show that the European Court of Human Rights has so far refrained from clearly committing itself to one specific approach. Instead of applying a uniform standard to assess domestic interferences with socially harmful speech, the Court simultaneously applies two distinct formula. Each of these methods can be traced back to constitutional doctrines developed in the U.S. Under the first line of cases, domestic decision-makers are permitted to restrict expressions that are “susceptible to instill” feelings of hatred or hostility or are “unwarrantably offensive” to others. In other words, the mere “tendency” of an expression to contribute to some possible future harm is sufficient to warrant prosecution.¹⁵⁶ This approach is reminiscent of the long discredited First Amendment bad tendency test. However, the European Court also put forward a second standard, allowing domestic decision-makers to restrict an expression only where it amounts to intentional incitement to unlawful action. In some cases, the Court goes as far as to require proof that the incited action is imminent and likely to occur. This standard resembles the Brandenburg incitement test. The two different standards are employed in the same types of cases, ranging from violence-conducive speech—such as terrorism-related speech—to hate speech directed against specific groups—such as members of the security forces—to hate speech based on intolerance toward minorities or vulnerable groups—such as xenophobic, Islamophobic, and homophobic expression. As a result of the inconsistent use of both methods, the Strasbourg Court’s treatment of hate speech cases is notoriously difficult to predict.

One can only speculate about the reasons for the incoherence in the European Court’s approach to hate speech. A first obvious candidate is the lack of consensus between the judges and the different sections of the Court. This is testified by the numerous concurring and dissenting opinions. Some of the members of the Court are willing to articulate clear standards for assessing hate speech restrictions in line with trends observed at the level of the United Nations, whereas others are more inclined to adopt a deferential attitude to the Contracting States, leaving them a very wide margin of appreciation in balancing the conflicting rights and interests.¹⁵⁷ Another factor that might explain the diverging attitudes to similar hate speech restrictions, is the illiberal or even authoritarian nature of some of the Council of Europe’s Contracting States. It cannot be excluded that the Court takes a stricter stance against free speech restrictions imposed by illiberal or authoritarian regimes. Perhaps it is no coincidence that the bad tendency test was formulated in cases involving France and Belgium (Féret and Le Pen), whereas the incitement test first made its appearance in a series of cases decided against Turkey (Zana and Sürek (No. 1)) and was further strengthened in cases in which Russia was the respondent State (Savva Terentyev and Stomakhin).

Whatever the explanation, it is clear that the application of two conflicting conceptions of hate speech to similar forms of expression has serious drawbacks. To begin with, the Court today fails to provide practicable and predictable standards to govern public and private responses to hate speech. As discussed in the introduction, international and European human rights law should provide a minimal level of guidance to States and companies so as to avoid both the under- and over-reaction to the challenges posed by hate speech, especially in the online environment. In addition, the current approach makes the Court vulnerable to allegations of applying double standards. The Strasbourg organs should avoid the perception that some countries are subject to stricter free speech standards than others.

To remedy the current situation, this article proposes a two-track strategy. First, it is submitted that the bad tendency path should be abandoned, as the numerous dissenting and concurring judges and free speech scholars seem to suggest. Many of the cases discussed in this article, especially those concerning the prosecution of journalists, illustrate that the use of this formula ultimately results in the suppression of legitimate political expression. This is the lesson that can be drawn from American constitutional history, where a similar standard enabled the government to silence political dissidents during the two World Wars and the first part of the Cold War period.¹⁵⁸ Instead this article argues that the Court should adopt a set of uniform criteria to assess hate speech restrictions in line with the current incitement test. Here too, the Court could offer more guidance as to the different components of the test, while simultaneously maintaining a degree of flexibility, keeping track with the Contracting States’ margin of appreciation. For instance, the Court could indicate if and when “imminence” is to figure as an independent requirement. A number of cases already seem to suggest that “imminence” is an important part of the inquiry when the message is directed against the institutions of the state. Other cases seem to suggest that “imminence” and “subjective intent to incite” are interchangeable conditions.¹⁵⁹ This last approach seems to correspond to the recommendations of the European Commission against Racism and Intolerance, according to which “the element of incitement entails there being either a clear intention to bring about the commission of acts of violence, intimidation, hostility or discrimination or an imminent risk of such acts occurring as a consequence of the particular ‘hate speech’ used.”¹⁶⁰ In any event, it would be helpful if the Court would further elaborate those principles. In addition, the Court could offer more guidance as to which factors it considers relevant to establish the mental state—intention—of the speaker.

Second, the Court should make it clear that the harms caused by hate speech are not limited to the potential effects on society at large. Therefore, in addition to the incitement test, the Court would do well to formulate a new category of unprotected speech: Intentional intimidation. The purpose of this category, for which inspiration can be found in Articles 8 and 14 case law on positive obligations, would be to protect the direct target of an expression against its immediate, physical, or psychological effects. Individually targeted insults and threats would thus be qualified as hate speech for the purposes of Article 10 on the dual condition that they attain “a certain level of seriousness” and are made with an intent to intimidate. There is no doubt that when it comes to this form of intentional intimidation or harassment, the need to protect the victims of hate speech far outweighs the free speech interest that might be involved.