Strasbourg Observers

Behar and Budinova v. Bulgaria: The Rights of Others in Cases of Othering – Anti-victim bias in ECHR hate speech law?

April 15, 2021

By Margarita S. Ilieva, a strategic equality and human rights litigator with extensive experience in hate speech. She litigated Behar,  Budinova and Panayotova and Others v. Bulgaria, and other landmark cases like Yordanova v. Bulgaria and Karaahmed v. Bulgaria.

The author was the architect of the cases discussed below, having brought them domestically in 2005 and lodged them with the Court in 2013. She represented them until 2018 when a substitute lawyer appropriated them.


On 16 February 2021, the European Court of Human Rights (‘the Court’) decided ground-breaking cases of minority othering:  Behar and GutmanBudinova and Chaprazov. The twin cases, stemming from collective domestic litigation, concern anti-Semitic/ anti-Roma hate speech. The applicants, community members, were not personally targeted. Domestic courts failed to protect their ‘private life’ from ethnic discrimination leading to a finding of a violation of Articles 8 and 14 of the Convention.

For the first time the Court:

  • found violations in cases of general anti-minority speech; 
  • articulated criteria to assess if speech is sufficiently prejudicial to affect a community’s sense of identity/ its members’ self-worth.

This represents a quantum leap: the Court has struggled to recognize the impact of identity abuse on individual dignity. Consequently, the case law on hate speech has been lopsided. It is underdeveloped in cases brought by hate speech victims and far more evolved in cases brought by hate speakers under Article 10 as discussed in the commentary below.


The applicants are Jewish (Ms. Behar, Ms. Gutman) and Roma people (Ms. Budinova, Mr. Chaprazov). They took xenophobic politician, V. Siderov, to court over his extreme propaganda against their communities. Siderov, founding leader of the far-right nationalist Ataka party, entered parliament in 2005. A former journalist, he had been using his various platforms extensively – newspaper articles, books, a TV programme he hosted, election rallies, a speech in Parliament – to radically campaign against minorities. In civil discrimination proceedings, the applicants argued that this constituted harassment and incitement to discrimination against them as members of their attacked communities, and sought a court injunction on Siderov to apologise and to abstain. The courts dismissed their claims, prioritizing Siderov’s freedom of expression.   


Using identical reasoning for both cases, the Court found a violation of Article 14 taken in conjunction with Article 8 on account of the failure of the domestic courts to discharge their positive obligation to afford redress to the applicants for the politician’s discriminatory public statements. The Court provided an assessment tool to gauge the deleteriousness of such statements for purposes of determining whether they were sufficiently severe for Article 8 to apply, and therefore, for Article 14 to apply as well. It articulated illustrative principled criteria setting out instances in which general anti-minority stereotyping can harm community members: target group characteristics (size, homogeneity, vulnerability, status), content (negative stereotyping potential, stereotype specifics), form and context, reach, author position/ status, potential to affect core group identity/ dignity, socio-political context. No single factor takes precedence (Behar §67, Budinova §63).

According to these criteria, the Behar statements were found to target, as a group, the Jews, a vulnerable minority. It can ‘readily be recognised […] they were virulently anti-Semitic’, ‘rehearsed timeworn anti-Semitic narratives’, ‘denying the Holocaust and casting it as a story contrived as means for financial extortion’. It can ‘hardly be questioned’ they were ‘extreme negative stereotyping meant to vilify Jews and to stir up prejudice and hatred towards them’. (§69-71)

In Budinova, ‘the group targeted were the Roma, long acknowledged as disadvantaged and vulnerable, and in need of special protection’. All statements were ‘deliberately couched in inflammatory terms, visibly sought to portray Roma […] as exceptionally prone to crime and depravity’; ‘systematic and characterised by […] extreme virulence’, the ‘message, conveyed bluntly and repeated many times over, was […] that Roma were immoral social parasites who abused their rights, lived off the back of the Bulgarian majority, subjected that majority to systematic violence and crime without hindrance, and aimed to take over the country’; having ‘capacity to stigmatise Roma […] as a group and arouse hatred and prejudice against them’; ‘beyond doubt’, ‘extreme negative stereotyping meant to vilify Roma […] and to stir up prejudice and hatred […].” (§64-5)

In view of this and the fact that the speaker was a parliamentary party leader on the rise with considerable media presence, his statements were able to have a sufficient impact on the Jewish/ Roma sense of identity and on Jewish/ Roma individuals’ self-worth, reaching the ‘severity’ threshold for Article 8 and 14 applicability (§72; §67-8).

The domestic courts had had a duty to balance the applicants’ rights to redress for such statements against the speaker’s freedom of expression (§100; §89). However, they had downplayed the statements’ capacity to stigmatise Jews/ Roma as a group and arouse hatred against them. Not properly weighing the significance of this effect on the applicants, the domestic courts ascribed considerable weight to freedom of expression. Failing in the requisite balancing exercise, they did not respond adequately to ethnic discrimination in breach of their positive obligation to secure respect for the applicants’ ‘private life’. (§104-6; §93-5)


These seminal judgments redeem much prior constriction of hate speech victims’ pursuit of recognition by the Court. Yet, several approaches are symptomatic of persisting anti-victim tendencies. Let’s view them in the context of the Court’s hate speech case law.  


The Court has seemed more mindful of hate speech victims’ rights in the abstract, that is to say in Article 10 cases brought by their abusers, yet not in their own cases. Far fewer victims than abusers have approached the Court: 7 decided cases of victims of general hate speech as opposed to dozens of sanctioned hate speakers’ Article 10 cases. Was the case law more welcoming for speakers meriting no Article 10 protection or restrictions to protect the ‘rights of others’? How has the Court treated those ‘others’ when they appeared before it in order to own their rights? When they claimed standing to require recognition and State accountability for (unaddressed) identity othering – has the Court restricted them? How do the new precedents fare against this background of asymmetries?

Case law asymmetries

  • Victims’ cases

The Court had heard five cases of impersonal hate speech victims: Pirali v. Greece (2007), L.Z. v. Slovakia (2011), Aksu v. Turkey (2012), Lewit v. Austria (2019), Panayotova and Others v. Bulgaria (2019); all unsuccessful. In Pirali and L.Z., the Court denied the applicants standing: the anti-immigrant, respectively, anti-Semitic expression did not personally affect them. (Domestic courts had variously discarded their cases.) The Court rejected their Articles 8 and 14 complaints as actio popularis.

In Aksu, the Chamber respected the applicant’s standing as domestic courts had done. No reason other than ‘discretion’ was given (§32), nor was any explanation provided on how Aksu standing differed from Pirali and L.Z. Articles 8 and 14 applied but no violation was found. According to the Court, the anti-Roma statements did not harm the applicant’s identity/ integrity. They ‘appeared to be discriminatory and insulting’ but lacked racist intent (Aksu, Chamber, §56). The Chamber ignored their effect (in the context of Roma vulnerability).

The Grand Chamber (‘GC’) sidelined the discrimination issue, excluding Article 14, as the applicant did not produce ‘prima facie evidence of discriminatory intent or effect’ (Aksu, GC, §45). The ‘negative stereotyping of a group’ and the ‘racial insults’ were not a ‘difference in treatment’ issue but a ‘private life’ one (§58, 45). Article 8 applied, yet the Court found no violation.

In Lewit, Article 14 was not invoked. A specific group, namely Mauthausen survivors, as opposed to a minority was targeted. The Mauthausen survivors consisted only of few individuals, of whom the applicant was one. He argued that he was personally affected. The domestic courts denied his standing. The Court held that they breached a procedural duty to examine his case, and found a violation of Article 8 (with non-pecuniary damages awarded).

In Panayotova, concerning an election brochure of Siderov’s party, a Committee found the complaints under Articles 3, 8 and 14 manifestly ill-founded. The ‘publication was premeditated and revealed virulent anti-Roma sentiment and a wish to stigmatise Roma in Bulgaria as a group’ (§48). It ‘clearly sought to portray Roma in Bulgaria as exceptionally prone to crime and depravity, and thus to stigmatise and vilify them’. The ‘assertions in the brochure were way stronger than the ones in Aksu(§56). However, under Article 3, the applicants were insufficiently affected by this content or the prosecutors’ refusal to investigate it (as it was ‘just facts’). The Court reiterated that it could not hear an actio popularis case.  

Articles 8 and 14 applied but required no criminal remedy, as the statements did not directly confront/ specifically affect the applicants. The ‘private life’ encroachment was not serious (§60, 62). The ‘imperative to combat racial discrimination’ made no difference (§57, 62). 

  • Hate speakers’  cases

Paradoxically, under Article 10, general hate speech necessitates stricter curtailment:

‘The Court has been particularly sensitive towards sweeping statements attacking or casting in a negative light entire ethnic, religious or other groups.’ (Perinçek, §206)

It is one of the case law’s asymmetries: in cases, in which States have taken responsibility for restricting hate speech – rather than survivors asking the Court to require States to do so – general anti-community speech is acknowledged as affecting (all) community members. In Perinçek, the GC held that the statements ‘affected the Armenians’ rights’, ‘the rights of the members of the Armenian community’, ‘the Armenians’ identity as a group’ (§228, 251, 274), without questioning if the latter were directly/ specifically/ personally/ particularly affected, unlike in victims’ cases. In I.A., the Court found ‘believers may legitimately feel’ the impugned content constituted  ‘unwarranted and offensive attacks on [them]’ (§29). In Article 10 hate speech cases, community members’ victimization is recognised. From it, the Court derives necessity for speaker sanctions protecting ‘the rights of others’.

Conversely, in cases of ‘others’ – othered community members – claiming rights as applicants, the Court struggles to acknowledge that hate speech harmed them. The Court has viewed impugned content as insufficiently abusive or not targeting or affecting them (Aksu, Panayotova). This slant disfavouring victims taints approaches in Behar and Budinova too, erecting roadblocks, as will be discussed below. 

Another asymmetry is the one-sided application of Article 17 as an Article 10 border checkpoint. Numerous sanctioned hate speakers’ complaints are found abusive – excluded from Article 10 protection (Hate speech factsheet; Article 17 Guide). If not incompatible ratione materiae, they are manifestly ill-founded, or (serious) criminal sanctions are necessary for the pressing social need of the protection of ‘the rights of others’. The Court supports States curbing minority abuse, activating Article 17 to limit/ qualify Article 10.

In victims’ cases, however, Article 17 is not applied to test if impugned expression is within the scope of Article 10. In Aksu, Behar, Budinova (the only cases that were not dismissed), the Court assumes Article 10 applies to the content, entitling its author to compete with victims for protection. This incoherence on a fundamental question – is hate speech within the scope of free expression? – appears dependent on whether applicants are speech victims or speech abusers, i.e. whether States stand to be corrected, or backed.

In abusive speakers’ cases, Article 17 approaches are diverging, lacking demarcation or applications standards. ‘Patchwork’ holdings do not clarify when/ how Article 17 applies. (The Court seems uncertain of its own approaches’ typology. Its ‘Hate speech’ factsheet outlines two approaches, whereas the guide on Article 17 notes three. A dissenting opinion in Perinçek mentions four approaches.) Maintaining ad hoc decision-making, the Court reserves discretion, rather than standardizing, i.e. institutionalizing Article 17 relevance for hate speech. This facilitates its non-application in victims’ cases.

 The Court admits that the case law should be symmetrical:

‘The outcome should not vary depending on whether the application was brought under Article 8 by the person who was the subject of the statement or under Article 10 by the person who has made it […].’ (Perinçek, §198)

Nevertheless, the outcomes vary. In hate speech victims’ cases, Article 17 is not integrated. Are the ‘rights of others’ othered in those Others’ own cases as they seek to take center stage in an individualized, actual way, unlike in Article 10 cases, in which States protect ‘the rights of others’ as an abstraction?

The Court treats the asymmetries themselves in an asymmetrical way depending on whether the case is a victim’s, or an abuser’s one:

‘[T]he issue […] is not whether any criminal-law measures […] would have been justified under Article 10 […] but whether such measures were required to protect the applicants’ rights.’ (Panayotova, §63)

Behar/ Budinova asymmetries

  • Standing: avoidance coping?

In Behar/ Budinova, the standing issue is avoided. The question of admissibility shifts to compatibility ratione materiae. On victim status, rather than discussing the expression’s impact on the applicants, the Court invokes the judicial response’s effect. The applicants’ victim status vis-à-vis community othering is not expressly acknowledged. In abusive speakers’ cases, it is settled that abusive speech affects the ‘rights of others’. In Behar/ Budinova, in which the Others are not a reference, but dominus litis, their right to access justice is not unequivocally recognised. The Court employs alternative admissibility scrutiny, namely Article 8 applicability, to maintain victim access conditionality (§46; §41).

  • Article 8 threshold

The Court replaces the standing hurdle with an applicability one. Unlike in earlier hate speech victims’ cases, it rigorously tests content severity to verify Article 8 applicability. In Aksu, Lewit, Panayotova, references to negative stereotyping of a certain level being able to affect group members, and Article 8 encompassing personal autonomy and (ethnic) identity sufficed. The ‘level of severity’ was not analysed for compatibility ratione materiae. Applicants’ allegations were sufficient for Article 8 access. Article 8 applied equally to non-severe content, giving rise to a finding of no violation (Aksu). Article 14 applied based on a mere reference to racial discrimination being ‘particularly insidious’ (Aksu, §49). In Perinçek, Article 8 protected Armenians’ rights regardless of the content’s virulence, which the Court viewed as insufficient to restrict.

In Behar/ Budinova, concerning content that the Court qualified as ‘extreme negative stereotyping’ (§71; §65), it introduces a stringent admissibility test, limiting Convention accessibility. Applicant assertions of ethnic stereotyping victimhood are insufficient for Articles 8 and 14 applicability. For admissibility, the Court scrutinizes stereotyping ‘severity’, instituting a threshold (derived from an unrelated employment case, Denisov (§66; §62). 

This ‘threshold’ requirement sets a perilous precedent, considering the highly virulent Behar/ Budinova content. If such virulence is required for victims’ Convention access, their dignity rights might frequently remain theoretical. Will less extreme expression reach the threshold? What defines minimum ‘severity’?

  • No Article 10 threshold, no Article 17 application

In Behar/ Budinova, Article 10 applicability is assumed, not tested. The content’s eligibility for inclusion in the scope of free expression is not examined. No ‘threshold’ mirrors the one for the applicants’ Article 8 rights; no symmetrical content ‘non-toxicity’ level is required. To the hate speaker, Article 10 is simply available. The Court presumes the hate speaker’s right to be entitled to protection, while the victims’ rights are tested for admissibility. Their conflicting rights are said to deserve, in principle, equal respect (§99; §89). Is not subjecting the victims, rather than the offender, to enhanced scrutiny in fact according unequal respect to victims? Is it equitable or logical that content sufficiently severe to affect Article 8 rights is insufficiently severe to be excluded from Article 10? Or, at least, to be tested for inclusion?    

Does a hate speaker’s interest deserve respect on a par with minority dignity? Is it a legitimate interest? Are minority identities, in view of the public interest to counter racism, entitled to no more respect than racist vilification is? It is a principle of the Court to protect shocking and offensive expression as well. Is minority dignity ‘shocking’ and ‘offensive’ that it only deserves the same respect as gratuitous insults incompatible with Convention values?

The Court does not apply Article 17.  It does not exclude the extreme content from the scope of Article 10. A hypothetical Article 10 claim by the speaker is not termed manifestly ill-founded. Article 10 is not interpreted in light of Article 17 to strictly necessitate curtailing the content to meet the pressing social need to deter verbal violence against vulnerable minorities.

This contradicts Article 10 case law excluding, via Article 17, anti-Semitic attacks, especially Holocaust denial, and other radical/ sweeping racist expression, particularly in electoral contexts (cf. Féret) – as some Behar/ Budinova statements are – from the scope of protected expression. The Court does not explain that Article 17 is inapplicable. It invokes it to indicate reduced/ lacking Article 10 protection for racist expression in principle – but not to cancel the balancing act in these cases. Rather than term the content an abuse of right, the Court accords it comparable status to the applicants’ dignity rights, to be balanced against them as equally legitimate. Is the right to be free from identity abuse competing with a right to abuse identity?

The ‘decisive point for Article 17 is whether the speaker sought to stir up hatred’ (Perinçek, §115. Virulent expression is protected if it ‘does not cross the line and turn into a call for […] hatred or intolerance’ (Perinçek, §231). The Behar/ Budinova content is ‘meant to vilify and to stir up prejudice and hatred’.  Article 17 is ‘applicable on an exceptional basis and in extreme cases’ (Perinçek, §114). The Behar/ Budinova content is ‘extreme’. As for ‘exceptional’, is this content less exceptional than the one in Article 17 hate speakers’ cases: Norwood (a sign in a village window), Witzsch (a private letter)?

In few cases, the Court explicitly determines that Article 17 is inapplicable (ex., Soulas, Lilliendahl). This occurs in instances where hate speech was deemed as not prejudicial enough. There is a ‘high threshold’ for Article 17 applicability (Lilliendahl, §26). Is the Behar/ Budinova content, including Holocaust denial, below this threshold? Context-wise, Roma are radically vulnerable. They are the most stigmatized, excluded, exposed communities in Bulgaria/ Europe.

The incongruity is starker considering that, in hate speakers’ cases, (serious) criminal penalties are found necessary – that is, if the perpetrators are granted examination on the merits. Sanction severity is an established criterion in such cases (ex., Perinçek, Smajic). In Behar/ Budinova, a mere civil remedy was sought – a declaration and injunction to abstain, and no compensation. Nevertheless, the risk of such low-severity restriction suffices to authorize the hate speaker to compete for protection with the victims.

Can this heterogeneity yield a solid test for Article 17 relevance to hate speech ensuring its uniform applicability, including in victims’ cases?

  • Prima facie?

The Behar/ Budinova statements are termed, as ‘obvious from the very terms used’, ‘prima facie anti-Semitic [discriminatory] in intent’ (§102; §91). How does the Court interpret ‘prima facie’? This concept denotes an inference, a rebuttable presumption, less than a finding of fact. Is a holding of ‘prima facie’ – apparent but not established – discriminatory intent compatible with terming the statements ‘extreme negative stereotyping meant to vilify Jews and to stir up prejudice and hatred, as ‘evident from their very wording’? The latter qualifications mean discriminatory intent is an established fact. It may not be simultaneously inferred, as if unproven/ covert. ‘Prima facie’ applies to covert discrimination, not to blatant bias. Terming glaring hate speech ‘prima facie’ is verbicide. The term is similarly misused in Panayotova, (§57), regarding content found equally malignant (the Court referred to this factual similarity in both Behar and Budinova (§64; §60)). It sets a dangerous precedent if ‘prima facie’ requires flagrant anti-minority insults. What is, then, required to establish rather than infer discriminatory intent? A perpetrator affidavit?

Conversely, in Article 10 cases, Holocaust denial automatically triggers an irrebuttable presumption of incitement to hatred (Perinçek, §234). In Behar, Holocaust denial results in a finding of ‘prima facie’ intent (refutable). Is the Court less prepared to recognise evident bias as established when victims (rather than States sanctioning hate speech) assert themselves?

  • No compensation

In Behar/ Budinova, non-pecuniary claims are dismissed. In Lewit, EUR 5 000 is awarded. In Lewit, the attack was not sweepingly anti-Semitic and it involved no (direct) Holocaust denial. In Behar/ Budinova, content, including Holocaust denial, vilified entire minorities. If compensation awards indicate a degree of the harm done, are anti-community attacks less harmful? This contradicts Article 10 case law finding general attacks less compatible with Article 17 (ex., Belkacem).


Is the Court (still) resisting the entitlement of community vilification victims when they seek, by litigating before it, empowerment directly – and through them, minorities – to the detriment of States? Is the Court prepared to go further than States upholding minority identity at the level of Language (the Symbolic)? 

Inconsistency, as evidenced by the asymmetries discussed above, tends to disadvantage disenfranchised people. Victims of minority othering possibly meet less judicial perceptiveness as case protagonists taking on a State partly due to an absence of corresponding perspectives on the inside of an institution. Are minority/ victimised perspectives adequately integrated via representation on the Court? Or is the compassion of the privileged (to be relied on as) sufficient?    

P.S. Bulgarian judges ahead of the game?

Behar/ Budinova originated in collective actio popularis as the first hate speech case in Bulgaria. The claimants, ‘Citizens Against Hatred’, included minority people (non-targeted identities too) and 68 NGOs. The argument was that minority dehumanization affected everyone/ the public interest. Eventually, the claims were separated. The Armenian one was upheld: the statements – not mentioning Armenians – constituted harassment and incitement, infringing ethnic minority rights and the public interest.
The Turkish claim was upheld: the claimant did not need not prove dignity impairment. Citizens against hatred had come into their own.

Print Friendly, PDF & Email

Related posts

Leave a Reply

Your email address will not be published. Required fields are marked *