Antisemitic comments on Facebook pages of leading British, French, and German media outlets

In May 2021, an 11-day escalation in the Arab–Israeli conflict prompted intense media coverage, social media activity, and antisemitic incidents worldwide. The study focuses on how today’s dominant form of hostility toward Jews—Israel-related antisemitism—is communicated across countries in response to this trigger event. By analyzing Facebook comments reacting to coverage of Hamas rocket fire and Israeli military responses, the authors investigate how antisemitic sentiments emerge, are expressed, and are linguistically realized in three languages (English, French, German) on mainstream media pages. The purpose is to identify prevalent antisemitic stereotypes, how they are adapted to current discourse, and how they are verbalized (including implicit and allusive forms) in environments where explicit antisemitism is generally condemned. The importance lies in understanding cross-national similarities and differences in antisemitic discourse and informing future detection and moderation strategies.

The study adopts the International Holocaust Remembrance Alliance (IHRA) working definition of antisemitism as a baseline for identifying antisemitic content, acknowledging debates around its scope and application (e.g., critiques by the Jerusalem Declaration on Antisemitism regarding Israel-related examples and distinguishing criticism from antisemitism). To operationalize antisemitism for contemporary online communication, the authors extend the IHRA framework into a detailed typology of 57 stereotypes spanning anti-Judaism tropes, racial antisemitism, post-1945 secondary antisemitism, and Israel-related antisemitism (including POWER, CHILD MURDER/BLOOD LIBEL, INSTRUMENTALISATION OF THE HOLOCAUST, NAZI ANALOGY, DENIAL OF ISRAEL’S RIGHT TO EXIST). Linguistic and discourse-analytic research underpins the analysis of how antisemitic ideas are conveyed through language, images, implicitness, metaphors, and praxeological features. Prior linguistics-based research has focused largely on German corpora; this study extends comparative insights to French and English contexts.

Research design and corpus: The study targets antisemitic hate speech in the comment sections of leading mainstream media Facebook pages in three countries, shifting the focus from fringe platforms to broader public discourse arenas. Media sampled included, for the UK: BBC, The Guardian, The Independent, The Spectator, The Telegraph, The Times, Daily Express, Daily Mirror, Financial Times, Metro, The Sun; for France: Le Monde, Libération, Le Figaro, Le Parisien, Le Point, L’Express, 20 Minutes; for Germany: Bild, FAZ, Focus, n-tv, rp-online, Der Spiegel, Süddeutsche Zeitung, taz, Die Welt, Die Zeit. Threads were selected to cover two corresponding topics: Hamas rocket fire and Israeli army retaliation (in the German subcorpus often combined). Comments were collected chronologically from relevant threads, typically coding 100–150 comments per thread to distribute sampling across outlets. The overall design targeted 1500 comments per language (4500 total), though counts in reported quantitative summaries vary by availability (e.g., UK 1504, France 1500, Germany 1320). Time frame centers on 10–12 May 2021, capturing immediate online reactions to the escalation. Analytical framework: Identification of antisemitic content is based on the IHRA definition, extended to a 57-category typology capturing classical and contemporary antisemitic stereotypes and Israel-related antisemitism. The approach combines qualitative pragmatic/discourse analysis and quantitative content analysis. Special attention is paid to implicitness, allusions, metaphors, rhetorical questions, and other speech-act and praxeological features used to conceal or suggest antisemitic meanings. Coding and qualitative analysis were conducted with MAXQDA, with an inductive openness to novel articulations and country- or milieu-specific variants. The analysis records both content-level categorization (stereotypes, analogies, delegitimization tropes) and form-level linguistic realizations (e.g., puns, dehumanizing metaphors, allusive historical references).

Quantitative distribution: Antisemitic comment frequencies varied substantially by country: UK 26.9% of 1504 analyzed comments; France 12.6% of 1500; Germany 13.6% of 1320. In the UK corpus, 38.7% of antisemitic comments required thread context to infer antisemitic meaning. Most frequent concepts (share of antisemitic comments, UK): Israel’s sole guilt in the conflict (27.9%), Child Murder/Blood Libel (8.6%), denial of Jewish self-determination (7.7%), Apartheid analogy (6.52%), Nazi analogy (4.26%), and amorality (4.0%). In the French corpus, about 62% of antisemitic comments required wider thread context to determine antisemitism. Prominent concepts (share of antisemitic comments, FR): denial of Jewish self-determination (17.8%), colonialism analogy (13.0%), Israel’s sole guilt in the conflict (10.1%), Jewish/Israeli influence on the media (8.2%), calls for action against Israel (8.2%), apartheid analogy (6.2%), child murder (5.8%), and a further mention of denial of self-determination (5.3%). Cross-cutting patterns: The EVIL stereotype is widespread across all three countries, frequently undergirding portrayals of Israel as a terrorist/criminal state, as solely culpable in the conflict, or via Nazi/apartheid analogies. Two shared dominant topoi are Child Murder and the denial of Jewish self-determination. Country-specific patterns: The topos of attack/criticism taboo is more prominent in Germany than in France or the UK. Allegations of Jewish/Israeli influence on the media appear relatively less in the UK and France; in the UK, many commenters disputing media portrayals accused outlets of pro-Israel bias (40.7% of all comments). Qualitative illustrations document prevalent rhetorical strategies including dehumanizing metaphors, loaded analogies (Nazi, apartheid, colonialism), conspiracy insinuations (e.g., media capture, global influence), and argumentation devices (lists, hyperbole, rhetorical questions).

The findings address the central question of how Israel-related antisemitism is communicated across countries by showing both common and country-specific repertoires. Across contexts, a core demonization frame (EVIL stereotype) anchors varied accusations (e.g., terrorist/rogue state, sole culpability, child-killing, racist/apartheid state) and analogies (Nazi, apartheid, colonialism) that delegitimize Israel and, often, Jews more broadly. Differences include higher observed antisemitic frequency in UK comment sections and a stronger prominence of attack/criticism-taboo discourse in Germany. Explanatory factors discussed for the UK include historically lower societal awareness of antisemitism and a “pioneering” role in propagating Israel-related antisemitism, contributing to higher tolerance or underrecognition of antisemitic speech. The frequent need for contextual inference (especially in French and to a lesser extent in UK corpora) highlights the role of implicitness and allusion in online antisemitic communication. Overall, the results underscore that mainstream media comment spaces host substantial antisemitic content during conflict escalations and that linguistic markers can be subtle, requiring nuanced, context-sensitive analysis.

User comments on mainstream media Facebook pages during the May 2021 escalation revealed a substantial rise in antisemitic discourse, with the UK showing considerably higher frequencies than France and Germany. Contrary to expectations of predominantly coded expressions in quality media environments, many antisemitic ideas were articulated openly or with minimal subtlety. Dominant concepts deny Israel’s moral integrity, depict it as an aggressor, and seek to exclude it from the community of states, often justified through analogies (Nazi, apartheid, colonialism) and conspiratorial frames. The study contributes cross-national, linguistically grounded evidence on the forms and frequencies of online antisemitism and highlights challenges for identification at scale. Future work should develop and train data management and annotation tools on qualitatively curated datasets to detect implicit and hybridized antisemitic expressions, extend analyses to larger samples and additional platforms, and evaluate the impact of moderation practices.

The analysis focuses on a narrow temporal window (May 2021 escalation) and on Facebook comments to selected mainstream media outlets, which may limit generalizability across time, platforms, or outlets. Sample sizes—though sizable—vary by country and are insufficient for robust inferential statistics; authors note that larger datasets would be needed for statistically significant quantitative analyses. The study is not a social media ethnography and does not analyze user networks or interaction dynamics. Corpora limitations mean observed linguistic patterns may not represent stable recurring patterns. Apparent lack of moderation on some pages may have affected the observed prevalence of antisemitic content.