This research was supported by the Higher Education Excellence Program of the Ministry of Human Capacities at Eötvös Loránd University (ELTE-FKIP). We thank Eszter Katona for her valuable advices.

1 Introduction

Antisemitism has been present in Europe for centuries dating back to as early as antiquity or even before. The next and substantially different phase of antisemitism starting from early Christianity was Christian anti-Judaism, a prejudice centered around religion. Langmuir (1972) differentiated between three aspects of Christian anti-Judaism: the doctrinal, the legal, and the popular. The first one attempted to prove that Christianity overtook Judaism and therefore Jews are inferiors to Christians and the outright enemies of Christianity. The basic element of this doctrine was blaming the Jews for the killing of Jesus, which in turn, deservedly exposed them to the punishment of God. The legal aspect of Christian anti-Judaism included legal measures against the Jews to prevent them from exercising control over Christians. Widespread popular anti-Judaism appeared somewhat later, beginning in the eleventh century. According to Langmuir, all three elements of anti-Judaism played a role in setting in motion the process where “a group already assumed to be inferior is forced by the majority to engage in conduct which seems further confirmation of the minority’s inferiority” (Langmuir 1972, 386). For example, as Jews were excluded from many occupations, they became disproportionally concentrated in certain social roles, such as moneylenders. Beginning in the twelfth-thirteenth century, Jews became the general scapegoats and stereotypes of them widened to include ritual murder (blood libels), well-poisoning, and host desecration. In the second half of the nineteenth century, as secularization progressed and religion lost its former primacy, a “modern”, racial form of antisemitism emerged. This new category proved to be even more deadly as it was also applicable to those Jews, and there were many, who had been assimilated into mainstream society. It was a very important change for another reason: while previously Jews could rid themselves of the characteristic they were being pursued for, namely by converting to Christianity, they could not change their race. Stereotypes from Christian Judeophobia mixed with this new form of antisemitism. For example, Jews were seen as outcasts who sought to harm the majority. Alternatively, Jews were blamed for a world conspiracy whose aim was thought to create a new world order and gain absolute control over the political, economic, and social institutions of the world. The main source of this conspiratorial aspect of antisemitism was the book entitled “The Protocols of the Elders of Zion” published in Tsarist Russia in the late nineteenth century. This conspiratorial form of antisemitism is still one of the most prevalent ones, especially in East-Central Europe (Barna 2017, Barna 2018, Bilewicz, Winiewski&Radzik 2012; Bilewicz et al. 2013; Glaesar 2005).

After the Shoah, a new form of modern antisemitism came into existence, namely secondary antisemitism. In its most extreme form, it refers to Holocaust denial, however, nowadays its “soft-core” version, a term coined by Deborah Lipstadt, is more prevalent. It includes distortion, relativization, and trivialization of the Holocaust (Heni 2008). Many times, secondary antisemitism is also connected to competitive victimhood. In these cases, the status of the Holocaust and Jewish victimhood are downplayed in order to “(1) relieve […] feelings of guilt or shame; (2) justify current misdeeds of the ingroup, and (3) create a sense of entitlement that allows yesterday’s victims to behave unfairly with a clean conscience” (Bilewicz&Stefaniak 2013).

Antisemitism disguised as anti-Zionism, and anti-Israeli prejudices have become especially prevalent in recent decades. However, this form of antisemitism should not be confused with the legitimate criticism of Israel. Natan Sharansky (2004) proposed the so-called “3D” test for distinguishing the two. The first “D” stands for demonization, and it refers to cases when Israel is demonized. It also involves Holocaust inversion suggesting that Israelis are Nazis and comparing Palestinian refugee camps to Auschwitz (Wistrich 2013). According to Heni (2008), this makes it connected to secondary antisemitism. The second “D” in Sharansky's test stands for double-standards meaning that criticism of Israel is applied selectively. The third “D” means delegitimization, and it refers to acts when Israel’s fundamental right to exist is questioned or denied.

In this paper, we aim at presenting antisemitism in contemporary Hungary based on the types described above. We approach our topic from two aspects. In the first part of our paper, we show the results of a recent survey carried out on a nationally representative sample. In the second part, we present the results of our research on the articles related to Jews from the far-right site The novelty of our research is two-fold. On the one hand, there has been no systematic analysis of On the other hand, the employed methodology makes this analysis even more pioneering as we applied one of the methods of Natural Language Processing (NLP), namely topic modeling. This methodology allowed us to process large amounts of unstructured textual data, as well as to identify the latent topics in the text. We had two research questions: (1) Around what topics are the texts of centered, which contain the word Jew? (2) Which types of antisemitism are present in these topics?

2 The prevalence of antisemitism in the Hungarian population

Based on theories of social psychology on prejudice, there are three dimensions of prejudice. The first is the cognitive component, i.e., the content of the prejudice. The second is the prejudice’s emotional intensity, i.e., the affective or the emotive component. The third is the inclination to act, for example to discriminate based on prejudice, i.e., the conative component.

In Hungary, the survey measurement of antisemitism based on nationally representative samples dates back to the middle of the 1990s and measurements have been carried out regularly ever since (Kovács, 1996, 2002, 2011, 2012).[1] Using the same methodology, Medián Public Opinion and Market Research Institute commissioned by the Action and Protection Foundation, has been measuring antisemitism on a yearly basis since 2013. We used their latest survey conducted in 2017 to investigate the prevalence of antisemitism in Hungary (Hann and Róna, 2018).

2.1 Cognitive antisemitism

When measuring the cognitive component of antisemitism, statements connected to the various types of antisemitism previously described were included. Religious antisemitism was measured with two statements: (1) “The crucifixion of Jesus is the unpardonable sin of the Jews”; and (2) “The sufferings of the Jews were God’s punishment.” In both cases, respondents had to indicate their agreement using a five-point Likert scale where one meant that the respondent does not agree with the statement at all, while five indicated that he/she agrees with it fully. In 2017, 31 percent of the respondents agreed or fully agreed with the first statement; while 20 percent with the second.

The surveys aimed at measuring conspiratorial antisemitism separately. Data show the high prevalence of this type of antisemitism reflected in the fact that more than one-third of the respondents agreed with the statements connected to the alleged all-pervasive power of Jews. In 2017, 36‑37 percent of respondents agreed with the statement that: “A secret Jewish conspiracy exists which defines political and economic policies”, “Jewish influence is too broad today in Hungary”, and that “It is a threat that the Jewish intellectuals control the press and the culture. Moreover, 35 percent of respondents thought that: “Jews are more likely to use unethical means in order to reach their goals than others.” Three statements were employed to measure secondary antisemitism using a binary form where respondents had to choose whether they rather agree or disagree with the given statement. The strongest statement, “There were no gas chambers in the concentration camps”, expressed unambiguous hard-core Holocaust denial. Eight percent of the respondents rather agreed with this statement. The other two statements did not deny that the Holocaust occurred per se, but instead offered a distorted view. Twenty-two percent of the respondents rather agreed that “The number of Jewish victims was far smaller than it is usually claimed” and 17 percent thought that: “A large part of the horrors were invented by the Jews after the event”.

Several statements connected to Israel were tested using five-point Likert scales as well. Forty-two percent of respondents agreed with the statement that “The Jews living here are more loyal to Israel than to Hungary.”; almost the same proportion (41 percent) thought that “The Hungarian Jews would rather support Israel than Hungary at a Hungarian-Israeli soccer game.”, and 37 percent found that Israel is “an aggressor, carrying out genocide against the Palestinians.” It is important to note, however, that there were other statements whose relationship with antisemitism was rather ambiguous. For example, respondents who were more antisemitic based on previous measures tend to agree more with the statement: “Israel is a legitimate prosecutor of a self-defensive war against the offenses to the country.”

2.2 Affective antisemitism

Researchers asked whether the respondent finds Jews rather likable or unlikable. In 2017, 25 percent answered the former. The second question on affective antisemitism included a so-called “emotional thermometer” where respondents had to express their feelings toward Jews (and other ethnicities) on a nine-point scale: one meant that the respondent was not sympathetic at all toward the given ethnicity, and nine meant the opposite. In 2017, Jews scored one of the highest, 4.9. For comparison, the three lowest scores were that of migrants (2.7), Arabs (3.4), and the Roma (3.5). Finally, social distance from Jews (and other ethnicities) was measured by asking whether the respondent would agree to have a Jew as a neighbor. Thirty-seven percent of respondents said that they would not. In the case of migrants, the corresponding number was 84, while for both Arabs and Roma 72.

2.3 Conative antisemitism

As mentioned before, conative antisemitism is the behavioral aspect, namely the inclination to act upon the prejudice. Survey measurement can never assess the proportion of the population who would actually discriminate. However, verbal inclination can be used as a (quite distant) proxy. In 2017, 26 percent thought that: “In case of certain professions, the number of Jews should be limited” and 21 percent that “It would be the best if Jews left the country.”

3 and the far-right

Jobbik – Movement for a Better Hungary has been one of the most successful far-right parties in the past decade in the Central and Eastern European region, gaining 17 and 20 percent of votes, respectively, in the Hungarian national elections of 2010 and 2014. Jobbik was founded in 2003, became a major political actor in 2009, and after 2014 strived to change its image to be seen as a more centralist, less radical actor (a moderate conservative people’s party), in order to widen its support base (Barna et al., 2018, 314–315). After the change in their political strategy – which was implemented by the then party-leader Gábor Vona – several politicians either left the party or were forced to resign. is a Hungarian far-right, nationalist news portal, registered in the State of California, USA[2]. The name of the portal contains strong symbolism since “Kuruc” was the name of the armed rebels who fought for independence from the Habsburgs. quickly gained popularity with the rise of Jobbik before 2010, being the fifty-first most-visited Hungarian website in 2009 (Molnár, 2016, 178). According to Mérték Media Monitor, was the tenth largest news site in 2018, with 10 percent of the population, and 23 percent of Jobbik voters at least occasionally visiting the portal (Mérték Media Monitor, 2018, 36). The website has sub-pages named “Anti-Hungarianism,” “Holohoax,” “Gypsy-crime,” “Jewish-crime,” and it harbors strong antisemitic and anti-Romani content. The site considers itself “a patriotic Hungarian conservative, right-wing, nationalist, fact-finding news site that is independent of political parties” (, 2008). It has been online since February 2006, which makes it one of the oldest radical news portals in Hungary.

Although Jobbik has consistently denied it, appeared to have strong ties with Jobbik, particularly with Előd Novák, the former vice president of Jobbik, before the change in the party’s political strategy. At that time, the topics and the interpretation displayed by strongly coincided with Jobbik’s ideology. Research shows that while Jobbik was heavily underrepresented in the mainstream media, it used the internet very successfully and was a key platform in this effort (Róna, 2016; Pirro and Róna, 2018, 11; Barlai, 2012, 234). The alleged connection between the website and Novák was an open secret for years: it was clear that he was associated with the website as an editor and contributor (Barna et al., 2018, 318–319).

The connection between Novák and was confirmed after thirteen years of secrecy at the first press conference held by on April 24, 2019. This event took place in the former office of the nationalist Hungarian Justice and Life Party (MIÉP). Contributing website authors were publicly presented for the first time, including Novák and three other journalists (Ács, 2019). The aim of the “conference” was to publicize the once-popular website and to prepare it for the campaign of the upcoming European Parliamentary and local government elections to be held this year. The press event and the recent articles on made it very clear that the website now supports the newly founded Our Homeland Movement (Mi Hazánk Mozgalom). After Jobbik’s centrist turn, when Novák was forced to resign as an MP of Jobbik, he had actively contributed to the creation of this party in 2018, of which he became vice-president. Our Homeland Movement currently has four members in the National Assembly, after politicians originally elected as Jobbik representatives joined the new organization.

Party preferences and antisemitism have a strong relationship. While in 2017, based on the combination of affective and cognitive antisemitism, 27 percent of the Hungarian adult population had extreme and 10 percent moderate antisemitic views, the respective numbers for Jobbik supporters were 42 and 15 percent. The change in the political image and policy of Jobbik, described above, was reflected in Jobbik supporters’ degree of antisemitism to some extent, since in 2013, the proportion of extreme antisemites was 53 percent and moderate antisemites constituted 15 percent of this group.

4 Data and Methods

In the following chapter, we briefly summarize our data collection and preprocessing methods, and the way in which we created our final topic model.

4.1 Data Collection

As the source of our data, we used SentiOne[3], a web-based social listening and text analytics platform. We downloaded all articles from that contained the word “Jew” itself or as part of another word, like for example the word Jewish. We assumed that (1) the authors and editors of had no intention of hiding their antisemitism, and (2) that the articles containing the word “Jew” are most probably used in a negative context. These assumptions justify the use of this search word. Our initial dataset, collected by SentiOne, contained articles from between February 28, 2016, and March 20, 2019. Our corpus contains 2,289 articles with an average length of 545 words.

4.2 Preprocessing

Before proceeding with topic modeling, we employed several steps of data transformation and cleaning. Firstly, we removed all URLs from our corpus, which are often cut up into small chunks when lemmatizing the text, making it nearly impossible to remove them later on. Secondly, we lemmatized our documents using hunlp. Lemmatization aims to group the inflectional forms of a certain word to be able to analyze them as a single item by bringing them to a common base form, which is their lemma or dictionary form (for example, “Jews” and “Jewish” are replaced by “Jew”). Hunlp also offers speech tagging, meaning that word classes are identified in the text. It allowed us to keep only adjectives, adverbs, nouns, and proper nouns in the corpus. Often, following lemmatization, several named entities remain that are not recognized as names by the lemmatizer; therefore in the third stage, we applied named entity recognition using the Hungarian database of DBpedia, which uses Wikipedia data to recognize the names of persons, organizations, locations, and other expressions. After experimenting with DBpedia, we defined an exclusion list of 902 items that we did not want to be replaced in the text. Fourthly, after dealing with the named entities, we ran a significant bigram detection, which aims to identify words that are highly collocated (for example “Gaza” and “strip”). It is advantageous to concatenate some of these discovered words in order to treat them as single entities during the analysis. In order to minimize the chance of concatenating words that are “false” bigrams, we manually selected the relevant items from a list of the five hundred bigrams. Fifthly, we defined an extensive list of approximately 800 words using word frequency lists generated from the corpus and a few experimental topic models, which we removed from the documents. These words were either too general, not relevant to our research or not meaningful without context (for example “then,” “apple,” “until,” “month”). Sixthly, we created a list of over 300 words that we replaced with their lemma in order to correct some flaws of the lemmatizer (for example “Israel” vs. “Israeli”). Seventhly, we removed too short (below three characters), overly frequent (found in above 50% of posts) and too rare (below 0.5% of posts) words from the final corpus. We also omitted documents that were too short to be used for topic modeling. Our final, preprocessed corpus contains 4113 unique words.[4]

4.3 Latent Dirichlet Allocation

Latent Dirichlet Allocation (LDA) (see Blei et al. 2003), a type of topic modeling algorithm, is an unsupervised method for dealing with large amounts of textual data. Using LDA, it is possible to identify the hidden, latent thematic structure of a given textual corpus. LDA detects the co-occurrence of words and starts from the concept that a certain topic can be characterized by the words associated with it; therefore, a document related to a topic is more likely to contain the words that are related to the given topic. For the current analysis, we used the MALLET implementation of LDA (McCallum, 2002).

In topic modeling, it is often a challenge to determine the number of topics used for the analysis. We determined the number of topics appropriate to our analysis based on the so-called topic coherence measure (CV) (see Röder et al. 2015) and the interpretability of topics. Therefore – based on our first models – we ran models ranging from 7 to 17 topics ten times and calculated the average and the standard deviation of CV for each topic number to assess overall topic quality. In parallel, we evaluated the different models based on their interpretability. We found that the model with fifteen topics is the best for our analysis. Our model has a CV value of 0.504.

5 Result and analysis

In the following analysis, we present the results of our topic modeling using LDA. The table below shows the fifteen most relevant words for each topic, the share of topics in the corpus (token percent), and the average length of the articles in each topic. In the appendix, we also present a visualization of our topic model.[5]

Topic number

Fifteen most relevant words

Token Percent

Average article length in words


fehér [white], ember [person], csoport [group], svédország [sweden], férfi [man], nemzet [nation], társadalom [society], szexuális [sexual], idegen [stranger], homoszexualitás [homosexuality], ország [country], faj [race], jelentős [significant], világ [world], fekete [black]




holokauszt [holocaust], német [german], lengyel [polish], áldozat [victim], adolf_hitler [adolf_hitler], szovjet [soviet], emlék [memory], lengyelország [poland], ukrajna [ukraine], halál [death], náci [nazi], történész [historian], tábor [camp], emlékmű [monument], élet [life]




rendőrség [police], férfi [man], rendőr [policeman], francia [french], ember [person], város [town/city], támadás [attack], hatóság [authority], tüntetés [demonstration], franciaország [france], tüntető [demonstrator], áldozat [victim], elkövető [perpetrator], fiatal [young], párizs [paris]




közösség [community], rabbi [rabbi], vallás [religion], törvény [law], iskola [school], zsinagóga [synagogue], vezető [leader], egyetem [university], intézmény [institution], ortodox [orthodox], gyerek [child], temető [cemetery], oktatás [education], hitközség [congregation], levél [letter]




jobbik [jobbik], párt [political party], fidesz_–_magyar_polgári_szövetség [fidesz], politika [politics], választás [election], soros_györgy [george_soros], orbán_viktor [viktor_orbán], ember [person], ellenzék [opposition], vona_gábor [gábor_vona], baloldal [left (political)], politikus [politician], képviselő [deputy / member_of_parliament], jelölt [candidate], parlament [parliament]




izrael [israel], palesztin [palestinian], jeruzsálem [jerusalem], benjámin_netanjahu [benjamin_netanyahu], állam [state], gázai_övezet [gaza_strip], terület [territory], nagykövet [ambassador], ciszjordánia [west_bank], palesztina [palestine], nagykövetség [embassy], biztonság [security], hamász [hamas], egyiptom [egypt], település [settlement]




antiszemita [antisemitic], antiszemitizmus [antisemitism], vezető [leader], párt [political party], szervezet [organization], képviselő [deputy / member_of_parliament], kijelentés [statement], politikus [politician], parlament [parliament], vélemény [opinion], rasszista [racist], rasszizmus [racism], mozgalom [movement], szélsőjobboldal [far-right], újságíró [journalist]




ember [person], család [family], élet [life], világ [world], történet [story], ház [house], igaz [true], város [town/city], jézus [jesus], szem [eye], isten [god], lány [girl], gyermek [child], kéz [hand], barát [friend]




amerika [america], elnök [president], donald_trump [donald_trump], amerikai_egyesült_államok [united_states_of_america], főváros [capital], washington [washington], szervezet [organization], állam [state], döntés [decision], vezető [leader], bizottság [committee], politika [politics], beszéd [speech], támogató [supporter], lépés [step]




magyar [hungarian], magyarország [hungary], budapest [budapest], orbán_viktor [viktor_orbán], kormány [government], film [movie], cigány [gypsy], haza [homeland], magyarországi_zsidó_hitközségek_szövetsége [federation_of_hungarian_jewish_communities (MAZSIHISZ)], ország [country], program [program], horthy_miklós [miklós_horthy], nemzet [nation], levél [letter], nemzetközi [international]




európa [europe], muszlim [muslim], német [german], németország [germany], ország [country], európai_unió [european_union], migráns [migrant], menekült [refugee], brit [british], iszlám [islam], török [turkish], berlin [berlin], egyesült_királyság [united_kingdom], társadalom [society], politika [politics]




katona [soldier], szíria [syria], orosz [russian], hadsereg [army], irán [iran], terület [territory], határ [border], támadás [attack], oroszország [russia], ország [country], erő [strength], moszkva [moscow], háború [war], fegyveres [armed / gunman], vlagyimir_vlagyimirovics_putyin [vladimir_putin]




miniszterelnök [prime_minister], ország [country], kormány [government], kormányfő [head_of_government], kapcsolat [contact], miniszter [minister], vezető [leader], politikus [politician], politika [politics], elnök [president], látogatás [visit], állam [state], közös [common], szervezet [organization], parlament [parliament]




bíróság [court], cég [company], vád [accusation], eljárás [procedure], pénz [money], információ [information], bíró [judge], alapítvány [foundation], vizsgálat [investigation], munka [work], börtön [prison], ügyvéd [lawyer], kerület [district], nyilvánosság [publicity], összeg [amount]




nép [people], keresztény [christian], nemzet [nation], írás [writing], magyar [hungarian], ember [person], politika [politics], egyház [church], beszéd [speech], világ [world], kommunista [communist], élet [life], faj [race], igaz [true], hatalom [power]



In the following chapters we analyze the topics by using the relevant words presented above and quantitatively analyzing the most typical articles in each topic.

5.1 Racial antisemitism as the starting point

There are three topics (Topic 1, 15, and 10) which are based on the assumption that Jews have a distinct racial character. In each of these topics racial antisemitism is augmented by other types of stereotypes and accusations.

Topic 1 mostly contains articles connected to the accusation that the Jews use their influence to contaminate Christianity and white people; however, it also features conspiratorial antisemitism. The articles portray Jews, for example, as “homopropagandists” or “deviance propagandists.” According to the articles in this topic, Pride parades in general, and the Budapest Pride, in particular, are financed by “Jewish money.” The media is ruled by, and films are made by Jews to spread homosexuality, to promote “the mixing of races,” and to propagate “anti-White stereotypes.” Racial antisemitism is reflected also in the accusation that Jewish professors are educating students according to their “racial interest,” and promote movements and governing principles that disorganize the host nation. We already mentioned that this accusation stems from religious anti-Judaism, and this becomes manifest in some articles. For example, according to one such article, the high proportion of Jewish participation in the adult entertainment industry is the result of Jews’ “ancient hatred toward the Christian order.” “It is Talmudic perversion how these Jews manipulate the inherent goodness and morality of the white man to disrupt their own culture” (, 2016a).

Topic 15 contains articles mostly dealing with different events of Hungarian history. Some articles deal with the well-known accusation that Jews are responsible for Communism based on the observation that there were many the politicians, who took part in the Hungarian Soviet Republic (Magyar Tanácsköztársaság) in 1919 or the leadership of Hungary after WWII were of Jewish descent. In these articles, the names of these politicians are often complemented with their original Jewish-sounding German names in brackets to make sure that the reader understands the content well. According to some articles, Jews, as journalists in the media or as historians and Holocaust researchers, unfairly attack the Christian Church and its important personalities for their role and responsibility during the Holocaust. As they say: “they’re provoking Hungarians with their actions to protect their own Jewish race” (, 2016b). According to these articles, the disruptive activity of the “Jewish press” is present in all historical times: even before WWI it incited hatred against Christian Hungarians, while nowadays the “far-liberal” (aka Jewish) media is discrediting the words “Hungarian” and “Christian.”

Topic 8 is a complete mixture; however, the articles are deeply antisemitic. Many articles deal with the past, and they contain quotations expressing antisemitism from figures of the Christian Church, artists, writers, and poets. The most antisemitic article (, 2018a) contains quotations from the parliamentary speeches of the MPs of the National Antisemitic Party from the end of the 19th century. Győző Istóczy, the founder and then leader of the party was the first to hold an antisemitic speech in the Hungarian National Assembly. Each sentence of the 4,591-word long article contains overt antisemitism. The article reflects that this party was the first representative and promoter of racial antisemitism in Hungary.

The articles in Topic 10 contain various types of antisemitism. Racial antisemitism is present as Jews (and Gypsies) are portrayed as people from a different race, and Jewish (and partly Roma) organizations as associations based on race, which should not receive public funding. There is an article (, 2017a) with a lengthy list about what “Hungarian citizens of the Jewish race” should do to support Hungarians. The same article also demands that all public officials and personalities who are Hungarian citizens and are of Jewish “racial origin” should declare “which bond is stronger for them, the Hungarian or the Jewish.” This is necessary because a person who would act as a Jew rather than a Hungarian cannot be in a Hungarian public decision-making position. There are also articles in Topic 10 that are connected to conspiratorial antisemitism based on the stereotypes about Jews having too much power and Israel buying up Hungary. In the Hungarian far-right circles, this accusation is mostly based on the unfortunate speech of Shimon Peres, the then President of Israel, to the chamber of commerce in Tel Aviv where he said that “Israeli businessmen are investing all around the world, enjoying unparalleled success, earning economic independence. We’re buying up Manhattan, Poland, Hungary, and Romania” (Papp, 2007).

5.2 Secondary antisemitism

Topic 10 also includes secondary antisemitism besides the other types of antisemitism described above. Articles in this topic about secondary antisemitism are mostly connected to competitive victimhood. Most articles report about the Oscar winner “Son of Saul,” which, according to the, is not a Hungarian movie. According to these articles, “generations have been raised on Jewish Holocaust stories,“ “radio and television stations report on events connected to the Holocaust almost every day” while no one talks about “the Holocausts (sic!) of Hungarians.” Articles also blame the Jews for using Jewish sufferings for “political bullying and blackmailing” and for “inciting guilt,” and financially profiting from it.

Articles in this Topic 2 clearly fall into the category of secondary antisemitism; however, without the previously described competitive victimhood. There is a long series of articles questioning whether the Holocaust really happened or whether it happened the way Jews claim it did. Many articles use pseudoscientific argumentation as it is often experienced in the case of Holocaust deniers. There is no need to quote the articles to describe them. It is enough to list some of the most frequently used words and expressions: “Holohoax,” “Holodogmatics,” “Hololegend,” “so-called Holocaust,” “alleged extermination camps and gas chambers.”

Besides the denial and distortion of the Holocaust, the articles in this topic also accuse Jews making a profit from this “historical lie” which creates a “gigantic political-financial hoax possible whose main beneficiaries are the State of Israel and international Zionism, and whose main victims are the German nation but not its leaders, and the whole Palestinian nation” (, 2018b). Moreover, one of the articles claims that “the [Jewish] world domination is based on the religion of the alleged ‘Holocaust’” (, 2018b).

5.3 The unethical Jews despising Gentiles

As described above, some articles of Topic 2 and 10 include the accusation of the Jews using unethical means. There are further topics (Topic 4, 13, and 14) that are also connected to this thought.

Both Topic 4 and 13 convey the image of Gentiles bowing to Jews, while Jews are disrespectful, ungrateful, and despise them. While in the case of topic 4 the Jews are rabbis and religious leaders, in the case of Topic 13 they are politicians, mostly those of Israel. In both topics, the expression Road to Canossa appears to describe how these non-Jewish people humiliate themselves in front of these Jews.

According to articles in Topic 4, politicians make “generous promises” to the Jews at the expense of the Hungarian people, for example, about synagogue restorations, while the Jews are rapacious and always want more. There is another alleged proof of the ungratefulness of Jews: while many leading politicians send season greetings to the religious and secular leaders of the Jews, they never wish “blessed Christmas to Viktor Orbán or the Christian Hungarian nation” (, 2019).

The articles in Topic 13 portray Jews and Israeli politicians, especially Benjamin Netanyahu, as feeling superior to the Gentiles. The articles in this topic are also tainted with new antisemitism. For example, these articles systematically call Netanyahu a war criminal and the politics of Israel aggressive and racist.

While Topic 4 deals solely with Hungarian non-Jews, the articles in Topic 13 mention some international examples. For example, Sigmar Gabriel, the German Minister of Foreign Affairs or Klaus Johannis, the Romanian President. While Gabriel “making his usual Road to Canossa,” “the war criminal Netanyahu gave an ultimatum to the German politician: if he [Gabriel] dares to meet the representatives of two [Palestinian] organizations, he [Netanyahu] will call off their planned meeting in Jerusalem” (, 2017b). In the case of Klaus Johannis, articles suggest that Johannis’ accusation of being antisemitic is thanks to the fact that “a Goy, and especially a Goy of the German minority can say or write down the word ‘Jew’ only with laudative, encomiast, and subservient adjectives” (, 2018c).

Articles in Topic 14 are also built upon the stereotype that Jews are inclined to use unethical means to achieve their goals. They report mostly about different court cases with Jewish participants mostly on the wrong side. Although the Jewish origin of these people had nothing to do with these cases, being a Jew is accentuated all over. For example, many articles deal with Harvey Weinstein, calling the film producer a “pervert Jew,” a “wealthy Jew” or just using the adjective “Jewish” whenever his name appears. Some articles describe murky business connected to people of Jewish origin.

5.4 New antisemitism

There are two topics (6 and12) that deal with new antisemitism. The articles in Topic 6 deal with Israel. They lack virulent antisemitism. However, the articles would not pass Sharansky’s “3D” test as they draw an overtly one-sided picture about the Israeli-Palestinian conflict. Many times, the articles use quotation marks to indicate the intended meaning of words and expressions which, of course, is the opposite of their original ones. For example, “the ‘counterterror’ measures of the Israeli military forces,” “the ‘peacemaking delegation consisted of mostly Jews,” Jared Kushner portrayed as American Jewish “peacemaker,” and Avigdor Liberman as “defense” minister.

Topic 12 is very coherent: the articles deal with the military conflicts between Israel and other countries in the Middle-East, especially Syria, Lebanon, and Iran, as well as the participation of Russia. These articles are also extremely one-sided and include antisemitism many times being present in ironic phrasing. For example: “they [Israelis] always state they are just responding, however, if someone mentions what happened prior, that is considered antisemitism” (, 2018d) or “Israel complained many times when they thought that Hezbollah violated the agreement since according to them agreements can be violated only by Jews” (, 2018e).

5.5 Antisemitism as a byproduct

There are topics (5, 11, 3, and 9) that are not directly concerned with Jews, and antisemitism appears “only” as a byproduct. The articles in Topic 5 are concerned with Hungarian politics, and because of the political changes described in the chapter about the far-right and its connection to, the publication dates of the articles play an important role. In 2016 and 2017, even though Jobbik’s turn from its original “values” appeared as negative, the articles unequivocally supported Jobbik. Jobbik was portrayed as the leader of the opposition and Fidesz, the governing party as its greatest enemy which had been using ignoble means to defeat Jobbik. The turning point in the articles’ intonation occurred after Jobbik was defeated in the parliamentary election in April 2018, and especially after its most extreme politicians left the party and established the Our Homeland Movement (Mi Hazánk Mozgalom), as we mentioned before. In 2019, the articles served as stump speeches for the movement.

The opposition except for Jobbik is portrayed as connected to Jews or serving Jewish interest. For example, an article suggests that they form a party which is not ruled by “Jewish impulse” and have people lead that party who, when speaking, “don’t induce a warning sign in one’s mind that suggest that it is yet again a Jewish tantrum coming from a Jewish throat connected to a Jewish cash office” (, 2016c). The adjective “Jewish” is used when articles mention left-leaning or liberal newspapers or news sites. The same applies when articles talk about people of Jewish origin, although this fact would be irrelevant.

Topic 11 is concerned with international politics, and antisemitism also occurs here as a byproduct. George Soros is portrayed as a “Jewish speculator,” an “international Jew”, a “Jewish villain” who supports migration. The press often has the adjective “Jewish”.

The articles in Topic 3 report on various demonstrations both in Hungary and abroad. This is again a topic where antisemitism appears as a byproduct, and it is mostly reflected in the adjective “Jewish” added to the names of left-leaning and liberal media outlets.

Almost all articles in Topic 9 are about the United States of America. Articles about the US presidential elections of 2016 that include Bernie Sanders, never fail to mention his Jewish origin, once using the pejorative term “Galician Jew.” Some of the articles list Jews in the Trump administration. Although there are some antisemitic allusions, many times it is left to the reader to draw the adequate conclusion, for example, from the fact that there are so many Jews around the President. Another group of the articles deals with the move of the US Embassy from Tel Aviv to Jerusalem.

5.6 Jews screeching about antisemitism

Topic 7 is connected to antisemitism in a very different way from what we have seen before. The articles in this topic are concerned with antisemitism and Jews’ embittered struggle against it. This struggle is often portrayed as Jewish screeching and described using ironic wording. Many articles about Hungary deal with the Orbán government’s “zero tolerance on antisemitism[6].” They sadly acknowledge that this means that “all criticism of Jews and Israel are considered antisemitism, even in our Christian democratic homeland” (, 2018f). An article accuses Orbán of using the alien word “tolerance” because “it has exceptional importance in liberal circles”; liberal meaning Jewish of course. Moreover, they point to an alleged contradiction: in theory liberals (aka Jews) say that “we should be tolerant of those thinking differently, immigrant, faggots, etc.” but the expression “zero tolerance” indicate that “tolerance has its clearly defined boundaries” (, 2018f).

In the international political scene, articles deal with such issues as the “Jewish wailing” over Freiheitliche Partei Österreichs (FPÖ) becoming a member of the ruling coalition in Austria; with Jeremy Corbyn, leader of the Labour Party being in the crossfire of the “Zionist lobby of Great Britain”; and with Donald Trump who holding his ground against the pressure from the Jews, “managers of the world” to dismiss his Deputy Assistant Strategist, Sebestyén Gorka for having ties to the far-right.

6 Conclusions

We used the articles of the most popular far-right media portal in Hungary. We analyzed articles from the period between February 2016 and March 2019, which contained the word “Jew” in any form. To use this corpus to access antisemitism on the far-right, we had to assume that the authors and editors of had no intention to hide their antisemitism. We also assumed that the articles selected in this way, used the word “Jew” in a negative context. The fact, that the article in all fifteen topics, we extracted, was antisemitic supports our assumption that using the search word Jew resulted in antisemitic content, therefore it was appropriate to use for our analytical purposes.

Based on our analysis, we found that the conceptual framework about the different types of antisemitism could be well employed to describe most of the topics. The classical forms of religious antisemitism, for example blaming Jews for deicide, was not present. However, the accusation that Jews strive to contaminate Christians and white people was present, which also dates back to Christian antisemitism.

One of the main findings of our analysis was the very high prevalence of racial antisemitism in Three topics (1, 15, and 10), representing one-fourth of the texts, were directly connected to this type. Moreover, many articles in the topics about Jews using unethical means and despising Gentiles (Topic 2, 4, 13, and 14), were also using the argumentations of racial antisemitism. This group of topics represented an additional one-fourth of the texts in the corpus. It is important to note that with survey research methods it is not possible to grasp this virulent racial antisemitism. Several studies have shown (Kovács 2002; Kovács 2011, 79–91; Salzborn 2008) the strong latency pressure connected to antisemitism, meaning that respondents consider it risky and socially condemned to express openly their antisemitic opinion. This fact has at least two consequences. Firstly, we can rightly assume that the proportion of antisemites is somewhat higher than what survey data show. Secondly, we cannot use statements to measure antisemitism, which would provoke a disproportionally high rejection of answering honestly or answering at all. Since racial antisemitism is such an open form of antisemitism, which is strongly condemned by social norms, questions directly targeting this type of antisemitism are absent from survey research. In this way, the analysis of the antisemitic texts of allowed us to explore antisemitism more accurately.

Secondary antisemitism was present in two forms. One of these is the classical hard- and soft-core denial of the Holocaust (Topic 2), while the other is connected to competitive victimhood (Topic 10). These articles connected to secondary antisemitism not only say that there was no Holocaust, or that it happened differently from the way Jews claim (Topic 2) or that there is too much about the Holocaust especially compared to the sufferings of Hungarians (Topic 10). They claim that Jews profit financially from this “Holocaust business.”

New antisemitism was present in two topics (6 and 12). Although articles connected to these topics often lack virulent antisemitism, reports on events connected to Israel are biased. We assume that this lack of malicious antisemitism in these articles is also due to the fact that in new antisemitism anti-Israeli and anti-Zionist statements are used to cover overt antisemitism.

Using Natural Language Processing in general, and LDA topic modeling in particular to analyze the articles from related to Jews, proved to be very fruitful as it shed light on the antisemitic discourse used by the far-right in Hungary.

<link file:1268 download>Visualization

7 References

Ács, Dániel (2019). After 13 years of hiding, they stepped into the light, but there is no contest with the entrenched Fidesz media machine. [13 év bujkálás után kiléptek a fényre, de a Fidesz-média mellett már nehezen rúgnak labdába]. URL: [checked on: 20.05.2019].

Barlai, M. (2012). Jobbik on the Web. In P. Parycek & N. Edelmann (Eds.), CeDEM12 – Proceedings of the International Conference for E-Democracy and Open Government (pp. 229–237). Krems: Edition: Donau Universität Krems.

Barna, Ildikó (2015). Anti-Semitic Hate Crimes and Incidents in Hungary (2014). Budapest: Brussels Institute.

Barna, I. (2017). Hungary. In I. Barna & A. Félix (Eds.), Modern Antisemitism in the Visegrád Countries (pp. 47–77). Budapest: Tom Lantos Institute.

Barna, I., Félix, A., Mesežnikov, G., Pankowski, R., & Šternová, V. (2018). Contemporary Forms of the Oldest Hatred: Modern Antisemitism in the Visegrád Countries. In The Noble Banner of Human Rights (pp. 303–338). Leiden: Brill Nijhoff.

Bilewicz, M., & Stefaniak, A. (2013). Can a victim be responsible? Antisemitic consequences of victimhood-based identity and competitive victimhood in Poland. Responsibility: An interdisciplinary perspective. Warsaw: Lexem.

Bilewicz, M., Winiewski, M., & Radzik, Z. (2012). Antisemitism in Poland: Economic, Religious, and Historical Aspects. Journal for the Study of Antisemitism, 4, 2801.

Bilewicz, M., Winiewski, M., Kofta, M., & Wójcik, A. (2013). Harmful Ideas, The Structure and Consequences of Anti-Semitic Beliefs in Poland. Political Psychology, 34(6), 821–839.

Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent dirichlet allocation. Journal of machine Learning research, 3(Jan), 993–1022.

Chuang, J., Manning, C. D., & Heer, J. (2012). Termite: Visualization techniques for assessing textual topic models. In G. Tortora, S. Levialdi & M. Tucci (Eds.), Proceedings of the international working conference on advanced visual interfaces. (74–77). Capri Island (Naples): Conference AVI'12 International Working Conference on Advanced Visual Interfaces.

Glaeser, E. L. (2005). The political economy of hatred. The Quarterly Journal of Economics, 120(1), 45–86.

Hann, E. & D. Róna. (2018) Anti-Semitic Prejudice in Today’s Hungarian Society. Research Report. Budapest: Action and Protection Foundation.

Heni, C. (2008). Secondary Anti-Semitism: From Hard-Core to Soft-Core Denial of the Shoah. Jewish Political Studies Review, 73–92.

Kovács, A. (1996). Anti-semitism and the young elite in Hungary. Ramat Gan: Sociological Institute for Community Studies, Bar-Ilan University.

Kovács, A. (2002). Public Identity in Defining the Boundaries of Public and Private: The Example of Latent Antisemitism. Social Research: An International Quarterly, 69(1), 179–194.

Kovács, A. (2011). The Stranger at Hand. Leiden: Koninklijke Brill NV.

Kovács, A. (2012). Antisemitic Prejudice and Political Antisemitism in Present-Day Hungary. Journal for the Study of Antisemitism, 4(2), 443–467.

Langmuir, G. (1972). Reflections on Medieval Anti-Judaism. 4. Anti-Judaism as the Necessary Preparation for Anti-Semitism. Viator, 2, 383–390.

McCallum, A. K. (2002). Mallet: A machine learning for language toolkit. URL: [checked: 20.05.2019].

Mérték Media Monitor (2018). Sources of Political Orientation in Hungary. Trends, 2015-18. [A politikai tájékozódás forrásai Magyarországon. Trendek, 2015-18] (Budapest: Mérték Médiaelemző Műhely) URL: [checked: 20.05.2019].

Molnár, V. (2016). Civil society, radicalism and the rediscovery of mythic nationalism. Nations and nationalism, 22(1), 165–185. URL: [checked: 20.05.2019].

Papp, László Tamás (2007). Shimon Peres and the true Hungarians. URL: [checked on: 26.05.2019].

Pirro, A. L., Róna, D. (2018). Far-right activism in Hungary: youth participation in Jobbik and its network. European Societies, 1–24.

Röder, M., Both, A., & Hinneburg, A. (2015). Exploring the space of topic coherence measures. In Proceedings of the eighth ACM international conference on Web search and data mining (pp. 399–408). ACM.

Róna, D. (2016).Jobbik-jelenség. A Jobbik Magyarországért Mozgalom térnyerésének okai. Budapest: KUK Könyv és Kávé Kft..

Salzborn, S. (2008). Uncovering latent antisemitism in Germany. Yalkut Moreshet, 5, 155–170.

Sharansky N. (2004). Foreword. Jewish Political Studies Review, 16(3–4) (Fall 2004), pp. 5–8.

Sievert, C., Shirley, K. (2014). LDAvis: A method for visualizing and interpreting topics. In Proceedings of the workshop on interactive language learning, visualization, and interfaces, 63–70.

Wistrich, R. S. (2013). Parallel lines: Anti-Zionism and Antisemitism in the 21st century. Jerusalem: Vidal Sassoon International Center for the Study of Antisemitism, the Hebrew University of Jerusalem.

Referenced articles (2008). About the news portal. URL: [checked on: 20.05.2019]. (=2016a). The deviant behavior gaining ground (Part II.) – Perverted propagandists and the media dominance [A ferde hajlam térnyerése (II. rész) - Perverz propagandisták és a médiatúlsúly]. URL: [checked on: 26.05.2019]. (=2016b). The „Holocaust researcher” and the martyr of the Hungarian nation (II.) [A "holokausztkutató" és a magyarság mártírja (II.)]. URL: [checked on: 26.05.2019]. (=2016c). Has the opposition gone crazy? The alternative is handed on a silver platter, but it is not the Jewish factor [Megőrült az ellenzék? Tálcán kínálkozik az alternatíva, de nem a zsidófaktor]. URL: [checked on: 26.05.2019]. (=2017a). Our reader wrote: A guideline for Hungarian political parties on how to manage contact with Israel and international Jewish organizations [Olvasónk írása: Vezérfonal a magyarországi politikai pártok számára az Izraellel és a hazai, illetve nemzetközi zsidó szervezetekkel történő kapcsolattartáshoz]. URL: [checked on: 26.05.2019]. (=2017b). Netanyahu was reluctant to welcome the German minister of foreign affairs in Israel, because the minister did not fulfill his demand [Netanjáhu nem volt hajlandó fogadni Izraelben a német külügyért, mert az nem teljesítette a követelését]. URL: [checked on: 26.05.2019]. (=2018a). Why did we have enough of the Jews already in the nineteenth century? – It becomes clear from the speeches of the National Anti-Semitic Party [Vajon miért volt elegünk a zsidókból már a XIX. században is? – Az Országos Antiszemita Párt felszólalásaiból kiderül]. URL: [checked on: 26.05.2019]. (=2018b). Robert Faurisson, the legendary leader of holo-revisionism, dies [Elhunyt Robert Faurisson, a holorevizionizmus legendás vezéralakja]. URL: [checked on: 26.05.2019]. (=2018c). “Secret business with the Jews” – Klaus Iohannis is already accused of antisemitism ["Titkos ügylet zsidókkal" - Klaus Johannist máris antiszemitizmussal vádolják]. URL: [checked on: 26.05.2019]. (=2018d). Merkel’s standard: if Israel massacres Iranians, that is acceptable, but if the Iranians fight back, she deeply condemns it [Merkel mércéje: ha Izrael mészárol irániakat, az rendben van, de ha azok visszalőnek, azt a legélesebben elítéli]. URL: [checked on: 26.05.2019]. (=2018e). Israel builds a wall also on the Lebanese border, and again, the “international public opinion” does not make a fuss about it [Izrael a libanoni határára is kerítést épít, és ezúttal sem hisztizik a "nemzetközi közvélemény"]. URL: [checked on: 26.05.2019]. (=2018f). “Zero tolerance” – A lot of baloney not only in liberal democracies, but also in our “Christian” democracy as well [„Zéró tolerancia” - Fából vaskarika nemcsak a liberális demokráciákban, hanem a mi „keresztényünkben” is]. URL: [checked on: 26.05.2019]. (2019). Jewish gall: the chief rabbi is not satisfied with the synagogue of the annexed Eperjes, so we will have to renovate that as well [Zsidó pofátlanság: nem elégedett a főrabbi az elszakított Eperjes zsinagógájával, ezért azt is nekünk kell felújítani]. URL: [checked on: 26.05.2019].

Ildikó BARNA is a sociologist. She is an Associate Professor at Eötvös Loránd University (ELTE) Faculty of Social Sciences Budapest, where she also serves as Head of the De-partment of Social Research Methodology. Her research topics include antisemitism, xenophobia, post-Holocaust studies and quantitative research on archival sources. Re-cently her interest turned to Natural Language Processing. One of her first research projects using this methodology aims at scrutinizing online hate speech analysing large amount of textual data. She took part in numerous research projects. Her most recent ones include Countering Antisemitism Through Testimony with the USC Shoah Founda-tion, “The Jews in Hungary 2017”, a survey research led by Prof. András Kovács, and a research project with the Tom Lantos Institute on Modern Antisemitism in the Visegrád Countries. In 2015, she was a Visiting Fellow at the Mandel Center for Advanced Holo-caust Studies at the USHMM, Washington DC where she started her research project on Hungarian Jewish Displaced Persons.

Árpád Knap is a survey statistician who started his Ph.D. in September 2018 at Eötvös Loránd University (ELTE) Faculty of Social Sciences, under the supervision of Ildikó Barna and Júlia Koltai. He has been working at ELTE University Faculty of Social Sci-ences, Department for Social Research Methodology on several projects with Ildikó Barna in the past three years. During these projects he has gained valuable experiences in quantitative research methodology, survey design (both online and offline), data analy-sis and visualization, as well as writing research papers. He wrote his MSc thesis using automatized internet-based data collection techniques, which required – among other skills – knowledge in programming and the ability to use several Python packages, in-cluding those used to create interactive data visualization interfaces. His Ph.D. research topic is the applicability of “big data” in the social sciences, with the main focus being on Natural Language Processing methodology for dealing with large amounts of poorly structured textual data.

  1. Thanks to Professor András Kovács.

  2. Since 2008, Hungarian authorities have made several attempts to shut down the site for displaying hate speech, libel and misuse of personal data among other allegations, resulting in a few outages and the portal changing service providers. However, all attempts were unsuccessful since the “United States has been reluctant to intervene in the case, evoking free speech rights protected by the First Amendment of the US Constitution” (Molnár, 2016, 178). Hungarian Action and Protection Foundation, in cooperation with the New York-based Anti-Defamation League have managed to shut down’s Facebook page as the news site had repeatedly violated the policies of the social network (Barna, 2015, 14).


  4. We used the following software to achieve the abovementioned steps.
    (1) Removing URLs: urlextract Python package <link en latest>
    (2)Lemmatization: hunlp (
    (3)Named entity recognition: DBpedia<link>
    (4)Significant bigram detection: Natural Language Toolkit (NLTK) (

  5. The visualization's left panel displays a global view of the topic model, shows the prevalence of the topics, and the distances between them using multidimensional scaling to project the distances onto two dimensions. The right panel displays a horizontal bar chart that show the fifteen most relevant words of the currently selected topic. Topics can be selected in two ways: either by hovering over one of the bubbles on the right or setting the topic number over the bar chart. The red part of the bars represents the overall corpus-wide frequency of the words, while the blue part displays topic-specific prevalence. Also, when hovering over a specific word, the visualization shows its conditional distribution over topics (Sievert & Shirley, 2014, 63).

  6. In 2013, Viktor Orbán declared “zero tolerance on antisemitism in Hungary” in the 14th Plenary Assembly of the World Jewish Congress held in Budapest. The Hungarian government has been constantly using this formula whenever this topic comes up.