The point at which free speech ends and hate speech begins has become a burning issue in an age of social media, where billions of people have access to share synchronous content on global fora. This has raised concerns over an epidemic of hate speech, with privately owned platforms looking to human rights law for inspiration and legitimisation of their own community standards and terms of service, which typically prohibit hate speech, although definitions vary widely. The concerns have also led both the European Union and European democracies to adopt several measures to counter this phenomenon. Examples include the EU’s Code of Conduct on Illegal Hate Speech which requires companies to remove hate speech within 24 hours of them being reported and the 2017 German Enforcement Act which places pressure on social media companies to remove hate speech at risk of 50 million Euro fines.
In the landmark case of Handyside v. The United Kingdom, the European Court of Human Rights (ECtHR or the Court) underlined that the right to freedom of expression, as provided for in Article 10 of the European Convention on Human Rights protects not only expressions that are favourably received but also those that ‘offend, shock or disturb.’ Despite this apparently robust protection of free speech, the Court has since developed a substantial body of case-law permitting (and even requiring) restrictions of ‘hate speech,’ which it has conceptualized as including even offensive speech. It has done so without properly defining or convincingly demonstrating the need to restrict this category of speech. The Court has allowed for criminal penalties on youngsters sharing out homophobic leaflets in a high school, a Belgian politician handing out Islamophobic leaflets, a German citizen comparing a local government official to Heinrich Himmler and has often used Article 17 (the non-destruction clause) in cases involving negationism or revisionism of the Holocaust.
In light of the above and within the framework of the Future of Free Speech project run by Justitia in collaboration with Columbia University’s Global Freedom of Expression project and Aarhus University’s Department of Political Science, we set out to look at the extent to which the Handyside decision and particularly the benchmarks of shock, offence and disturbance have been been upheld within the framework of hate speech. To this end, we analysed 60 cases decided by the ECtHR and the European Commission of Human Rights’ between 1979-2020. 57 of those cases were brought by the utterers under Article 10, and 3 by the victims of the alleged hate speech under Articles 8 and 14. Cases concern speech linked to homophobia and transphobia, ethnic hatred, religious hatred, violence, totalitarianism and genocide denial. Our analysis reveals that 61% of cases brought by the utterers resulted in the applicant’s loss through a finding of non-violation of Article 10 (21%) or due to the inadmissibility of decisions in 41% of cases (for example through the use of Article 17 in Holocaust denial cases) Only 39% of cases brought by the utterers have resulted in a finding in favour of the applicant. Thus, on average, free speech restrictions have been upheld in just over one out of three hate speech cases. Many of the cases decided by the ECtHR involve statements that deserve moral condemnation from both civil society and governments as well as moral support to the affected minority groups. The response does not necessarily have to include legislative (particularly criminal) penalties, as repeated time and again by the European Commission against Racism and Intolerance and by the United Nations. At the same time, a plethora of free speech cases have gone amiss due to the Court’s approach. A good example is that of Nix v Germany. where the Court rejected the application of a German blogger who posted a picture of Himmler wearing a swastika armband and likened him to the officers of the employment office and the alleged discriminatory treatment his daughter (and many other youngsters) were receiving from them.
Quantitative and Qualitative Compilation of Findings
We have compiled our findings into an interactive online database. This allows users to view the cases thematically but also per country. A short overview of each case as well as link to the original judgment is available for easy access and reading. It must be noted that translations from judgements originally in French are our own. We will be adding to the database as new cases come about and welcome anyone to contact us if they consider that a case has been missed out.
Hate Speech: Semantics and Notions
As such, the first points to note are that the term ‘hate speech’ is not essentially defined by the ECtHR and there is a variation of thresholds attached thereto. The Court’s references made to hate speech are generic, offering no substantial definition of this form of speech. The closest we have come to a conceptual understanding of hate speech has been Lilliendahl v Iceland (2020). This case involved comments made under an online article by a citizen regarding a proposal to strengthen education and counselling in schools on matters concerning those who identify themselves as lesbian, gay, bisexual or transgender.
This was the first time that the Court posed the direct question of whether the speech amounted to hate speech within the meaning of the Court’s case-law. To answer this, the Court set out an explanation of hate speech based on its previous jurisprudence, adopting a hierarchal categorisation, rather than assessing the substance of what can actually fall within the framework of hate speech. The fact that no previous case had actually provided a definition of hate speech did not help the Court in this exercise. It found that hate speech falls into two categories. The first is the ‘gravest forms of hate speech’ that are excluded from any protection through Article 17 (with no definition of what constitutes the ´gravest forms of hate speech´). The second is the ‘less grave forms of hate speech’ which do not fall outside Article 10 but which the Court ‘has considered permissible for the Contracting States to restrict.’ Here, the Court incorporated not only calls for violence or other criminal acts but also insults, ridicule and slander in order to combat ‘prejudicial speech within the context of permitted restrictions on freedom of expression.’ It makes no further elaboration of what this context of permitted restriction may be, something which would have been expected given the fundamental nature of free speech in addition to the very low threshold attached to, for example, insult or ridicule.
Despite the fact that the term ‘hate speech’ is included in over twenty of the cases examined, with some being from the end of the 90’s, the Court waited until 2020 to establish the above tiers and extrapolation of the term (albeit without too much nuance). This 2020 positioning demonstrates that the threshold of the ECtHR is in fact low since insults can be prohibited, whilst the reference to ‘prejudicial’ speech is also indicative of this. Illustrative of the low-thresholds but also of the inconsistent position of the Court to hate speech are the differential treatment in Ibragim Ibragimov and Others v Russia (2018) on the one hand and Atamanchuk v Russia (2020) on the other. The former involved the banning of Muslim scholar Said Nrusi’s book, due to allegations that is was extremist literature. Here, the ECtHR held that, since the book depicted a moderate, non-violent, understanding of Islam, the restriction to speech was not legitimate. It noted that:
‘merely because a remark may be perceived as offensive or insulting by particular individuals or groups does not mean that it constitutes “hate speech.” Whilst such sentiments are understandable, they alone cannot set the limits of freedom of expression.’
In Atamanchuk v Russia (2020), which involved an application made by a journalist/politician after he was convicted of making statements against non-Russians, referring to them as criminals (without making any calls for violence), the Court found that:
‘inciting hatred does not necessarily involve an explicit 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 xenophobic or otherwise discriminatory speech in the face of freedom of expression exercised in an irresponsible manner.’
This threshold is rather broad since it incorporates even the mere justification of hatred and the encapsulation of insults in the ambit of an attack. This is far from the threshold from the international ‘hate speech clause’ namely Article 20(2) of ICCPR which prohibits advocacy for national, racial or religious hatred that constitutes incitement to discrimination, hostility or violence.
Thus, in the former case, mere insult was not sufficient to prohibit speech whereas in the latter, not only could insult be prohibited, but it was also incorporated in the framework of inciting hatred, without the nexus between insult and hatred being defined by the Court. The differential element of the cases was that in the latter, speech was directed against a particular group characterised by a particular characteristic (ethnicity).
As such, although the Court notes that it may be considered necessary in certain democratic societies to limit some forms of expression, it has yet to provide a coherent legal and/or normative extrapolation of when/where/how these limitations can or should occur. This has resulted in certain anomalies vis-à-vis the treatment of similar cases as demonstrated above. To this end, a proper definitional framework is of utmost importance as well as a coherent set of thresholds.
A coherent conceptualisation of hate speech by the ECtHR is a necessity since, without this, delineations between acceptable and unacceptable speech cannot be discerned. This is even more pressing in the absence of a Council of Europe equivalent to the Rabat Plan of Action which sets out strict thresholds for hate speech. One of the central objectives of this Plan is to provide for a comprehensive assessment of the state of implementation of the prohibition of incitement in conformity with international human rights law through a robust threshold.
It is imperative to highlight that hatred should be tackled for the purpose of ensuring just and equal liberal democracies. At the same time, a very central tenet of such a democracy is the fundamental nature of free speech. Whilst the ECtHR’s restrictive approach has arisen from the good (yet paternalistic) intentions of the ECtHR to protect groups perceived vulnerable to hatred, the handling of the manner, at the expense of Article 10 has been unconvincing and confusing. Moreover, nowhere in the Court’s judgment has there been any openness to explore the potential negative side effects that speech restrictions may have on the minorities they try to protect nor on the possible positive correlation between speech restrictions and violence, as suggested by empirical research.
In light of the above qualitative and quantitative findings, we argue that the ECtHR adopts an overly-restrictive approach in the ambit of hate speech, which fails to provide adequate protection to political speech on controversial issues, including criticism of public officials and government institutions and has created an inconsistent and even arbitrary body of case law which leaves European citizens and States at a loss on how to properly delineate the limits of hate speech.