=Paper=
{{Paper
|id=Vol-2870/paper56
|storemode=property
|title=Hate Speech in Media Towards the Representatives of Roma Ethnic Community
|pdfUrl=https://ceur-ws.org/Vol-2870/paper56.pdf
|volume=Vol-2870
|authors=Zoriana Haladzhun,Olha Harmatiy,Yuriy Bidzilya,Nataliia Kunanets,Khrystyna Shunevych
|dblpUrl=https://dblp.org/rec/conf/colins/HaladzhunHBKS21
}}
==Hate Speech in Media Towards the Representatives of Roma Ethnic Community==
https://ceur-ws.org/Vol-2870/paper56.pdf
Hate Speech in Media Towards the Representatives of Roma
Ethnic Community
Zoriana Haladzhuna, Olha Harmatiya, Yuriy Bidzilyab, Nataliia Kunanetsa and Khrystyna
Shunevycha
a
Lviv Polytechnic Nationality University, Lviv,Ukraine
b
State University ‘Uzhhorod National University’, Uzhgorod, Ukraine
Abstract
The article defines a concept of “hate speech towards the representatives of national or ethnic
groups”, identifies its elements, threats, linguistic means of actualization in headlines and
mass media texts. Also, the possibility of its automatic recognition through the use of
technical means is being considered.
Keywords 1
Hate speech; Discrimination; Roma People; Media; Speech Aggression; Linguistic Means.
1. Introduction
The extension of communication space, transfer of private communication as well as mass media
platforms to the Internet, blurring of lines between territories and ethnic groups, - all of these is our
reality, which requires research and scientific elaboration. Our world is diverse and that is the value of
it. Each person may refer himself or herself to various social groups and strata according to race, skin
colour, nationality, ethnic group, religion, occupation, physical condition, gender, sexual orientation,
etc. However, this is exactly the reason why hate speech is used against us, leading to xenophobia and
discrimination. It is unacceptable when hate speech is applied in private discourse, but if it is used by
creative mass media workers on media platforms which they represent, it is a violation of journalistic
standards and legal norms.
The issue of hate speech in media coverage has been studied by foreign and Ukrainian researchers,
in particular: Howard, Jeffrey W. [1] “Free Speech and Hate Speech”, Herz, Michael and Peter
Molnar [2] “The content and context of hate speech”, Elford, Gideon [3] “Legitimacy, Hate Speech,
and Viewpoint Discrimination”, Laub, Zachary [4] “Hate Speech on Social Media: Global
Comparisons”, Erjavec, Karmen [5] “Media Representation of the Discrimination against the Roma in
Eastern Europe: The Case of Slovenia”, Bondarenko, Tetiana [6] “Speech aggression and linguistic
tools to prevent verbal hostility”, Kozhevnikova, Halyna [7] “Hate speech: typology of journalist’s
mistakes”, Verkhovskyi, Oleksandr [8] “Hate speech against society”, Butyrina, Mariia [9]
“Xenophobic discourse: reasons, scale, consequences and lessons for the media”, Kostovska, Alla
[10] “Between Roma and Gypsies: stereotypes and discrimination in the Ukrainian online media”, etc.
The aim of the study is to clarify the organizational means of preventing discrimination on the
grounds of nationality, to identify linguistic forms of hate speech against Roma people (as
exemplified by the headlines of the texts of online edition ZAXID.Net during 2020), and to clarify the
possibility of its recognition by automatic technical means.
COLINS-2021: 5th International Conference on Computational Linguistics and Intelligent Systems, April 22–23, 2021, Kharkiv, Ukraine
EMAIL: zoriana.v.haladzhun@lpnu.ua (Z. Haladzhun) olha.v.harmatiy@lpnu.ua (O. Harmatiy); bidzilja@gmail.com (Y. Bidzilya);
nek.lviv@gmail.com (N. Kunanets); krishirak@gmail.com (K. Shunevych)
ORCID: 0000-0002-3487-6057 (Z. Haladzhun); 0000-0002-7611-1503 (O. Harmatiy); 0000-0001-5134-3239 (Y. Bidzilya); 0000-0003-
3007-2462 (N. Kunanets); 0000-0002-2282-4575 (K. Shunevych)
©️ 2021 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
The tasks of the study:
1. to investigate the organizational basis for the protection and defense of the rights of Roma
people and to trace the attitude towards them in the Ukrainian society;
2. to summarize theoretical achievements of scholars in the issue of hate speech against the
representatives of ethnic and national groups, to identify its structural elements and forms of
expression;
3. to single out linguistic means of expressing hate speech resulting from belonging to the
Roma community as exemplified by the relevant materials posted on the website ZAXID.Net
during 2020;
4. to determine the possibilities of language assessment with the help of technical means.
The object of the research is the material about the representatives of Roma people published on
the website ZAXID.Net during 2020.
The subject of the research includes the foundation, forms, methods, linguistic devices of using
hate speech by professional journalists in the Ukrainian media and the possibility of technical
automatic means of language assessment.
The methodological foundation for the research consists of the guiding principles of the
dialectics of scientific knowledge, transformed in relation to the problem being studied, in
particular: the principle of unity of form and content, which determines the analysis of the
peculiarities of the linguistic actualization of axiological meaning; the principle of dialectical
unity of rational and emotional in thinking and language; the principle of unity of
consciousness and activity. General scientific concepts on the unity of the individual and the
general, the part and the whole, the static and the dynamic, the structure and the function, as
well as the theory of the general connection of the phenomena of reality are considered to be
of prime importance. The study of lexical-semantic and grammatical language systems was
based on the provisions of cognitive science, within which a language is considered as a
cognitive mechanism involved in coding and transmitting the information. The cognitive
approach allows analyzing the language of publicist writings and journalism as a means of
influencing the social consciousness.
Descriptive method, analysis and synthesis were mainly used to characterize the means of
expression of the evaluation category in the course of the examination and comprehension of
scientific and theoretical achievements on relevant issues as well as the description of
specific examples of materials. Peculiarities of linguistic expression of evaluation in
informational texts were established by contextual-situational, classificational and
quantitative-statistical methods. The techniques of distributive and component analysis, the
method of induction when summarizing based on the analysis of specific examples of
websites, the comparative method when establishing the correspondence of theoretical and
practical data were partially used.
The methodological tools of quantitative linguistics were used, which allows studying
language texts with the help of statistical methods, and the content analysis method.
2. Organizational support for countermeasures against hate speech towards
the representatives of the Roma people and the real situation in Ukraine
In recent years, experts in various fields have started using the term "hate speech", in particular
journalists, political analysts, psychologists, linguists, lawyers, etc. Rhetoric that falls under the above
mentioned category is prohibited in all democratic states of the world, though the absence of a
common, uniform definition makes it possible to approach this issue subjectively. Despite the lack of
a clear conceptual framework, the European Convention on Human Rights (1950), in particular
Article 10, protects the rights of groups and individuals from such negative statements. For the
development of these standards and norms, the Council of Europe has signed the treaties that deserve
our attention today, namely the European Social Charter (1961) and the Framework Convention for
the Protection of National Minorities (1995). The latter addresses the issues of protection against all
forms of discrimination, such as intolerance to people with different skin colour, race or religion. The
Additional Protocol to the Convention on Cybercrime, which came into force in 2006, also plays a
significant role. It focuses on the criminalization of acts of racist and xenophobic nature committed
through computer systems. The Protocol requires participating states to enact legislation which is
necessary to ensure measures on recognition of national and legal crimes. Ukraine is a party to
international treaties that are directly related to combating racial discrimination. These include the
International Covenant on Civil and Political Rights (1966), the European Convention for the
Protection of Human Rights and Fundamental Freedoms (1951), the International Covenant on
Economic, Cultural and Social Rights (1966), the UN Convention against Torture and Other Cruel,
Inhuman or Degrading Treatment or Punishment (1984), the Convention on the Elimination of All
Forms of Discrimination against Women (1979), the International Convention on the Elimination of
All Forms of Racial Discrimination (1965). Ukraine, as a member of the Council of Europe and the
OSCE, also ratified the UNESCO Declaration on Races and Racial Prohibitions (1978), the UNESCO
Declaration of Principles on Tolerance (1995), and the European Social Charter (1996). Article 21 of
the Constitution of Ukraine states that “all people are free in their dignity and rights. Human rights
and freedoms are inalienable and inviolable”. And Article 24 adds that “citizens have equal
constitutional rights and freedoms and are equal before the law. There shall be no privileges or
restrictions based on race, colour of skin, political, religious and other beliefs, gender, ethnic and
social origin, property status, place of residence, linguistic or other characteristics.” According to the
researcher A. Kostovska [10, p. 39], “the Roma community is the least integrated and the most
marginalized national minority not only in Ukraine but also in the European Union”. According to the
2001 All-Ukrainian Census, 47,6 thousand people admitted to being Roma. The Strategy for the
Protection and Integration of the Roma National Minority into Ukrainian Society until 2020
emphasizes that they still remain perhaps the most socially vulnerable national minority. The reasons
are various: low level of education; a lot of Roma people do not have any documents; high
unemployment rate; nomadic lifestyle; rigid hierarchy, etc.
In order to understand the reasons why hate speech often appears in the Ukrainian mass media,
particularly in the coverage of issues related to the representatives of Roma ethnic group, it is
necessary to take into account the general level of tolerance in Ukraine. According to the study
conducted by the Center for Content Analysis over a period of May-June 2020 on the attitude of
Ukrainians to the public discussion of the BlackLivesMatter movement as well as protests in the
USA, more than 37,000 messages on social networks were analyzed and it was found that 56% of
them were tolerant and 44% of them were of racist nature, and that women generally have a higher
level of tolerance than men (80.8%/45.3%). A similar study was conducted by the Razumkov Center
in 2011. Only 4.8% of respondents are ready to protest against discrimination on racial or ethnic
grounds, as opposed to 26.9% of those who would protest against the arbitrariness of local authorities
[11]. According to the index of ethnic distance (Bogardus social distance scale), from 1994 till 2010
Roma people belonged to the third level of population groups (5-6 points), which indicates isolation,
unwillingness to see the members of this group as citizens of the state, and go over to the fourth group
(6-7 points) – xenophobia as an extreme group of intolerance, occupying the second to last rating
position in the ranking of 24 studied nationalities, being ahead of only Chechen ethnic group
("Assessment of tolerance in the Ukrainian society: risks and opportunities for the formation of
national unity". Analytical note. National Institute for Strategic Studies [12]. Illiuk, N. [13, p. 223]
considers hate speech to be so dangerous to the public because the citizens of Ukraine lead a lifestyle
spending a lot of time watching TV, reading newspapers or news websites, and listening to the radio.
That is why the messages broadcast by the media in Ukraine have a major influence on the formation
of people's views. And this is especially true in regard to children and youth. Boiko, A. [14, p. 5]
believes that the Ukrainian society is experiencing a real crisis of tolerance. And one of the reasons is
the inhumanity of the media.
Although tolerance and non-discrimination of people and groups to which they belong on the basis
of racial, ethnic or other differences is fairly well regulated by laws, according to various surveys –
the problem really exists, and modern media contribute to its spread and support as well. It is
necessary to draw boundary lines between freedom of expression, hate speech and the right to non-
discrimination.
3. Hate speech – theoretical aspect
The Ukrainian term comes from the English term “hate speech”, and the council of Europe defines
it as all forms of expression which spread, incite, promote or justify racial hatred, xenophobia, anti-
Semitism and other forms of hatred based on intolerance, including intolerance expressed by
aggressive nationalism and ethnocentrism, discrimination and hostility against minorities, migrants
and people of immigrant origin (Recommendation of the Committee of Ministers of the Council of
Europe № 97 (20)).
According to the Institute of Mass Information, hate speech includes words, photos, or images that
create or deepen hostility between groups in the society or individuals (IMI).
The SOVA Centre for Information and Analysis [14] points out that hate speech includes
unacceptable negative statements addressed to ethnic, confessional or other social groups as
communities and to individuals as representatives of these communities; statements that deny the
principle of equality of all the people at one level or another.
In Ukrainian scientific and journalistic discourse, we come across two terms that express the
concept of “hate speech” – “speech of hatred” and “speech of hostility”, which are synonymous with
the original term, although the “stress level” they express is different, because the “speech of
hostility” contains indirect calls for violence, while the “speech of hatred” contains direct ones.
Human rights defenders Dzh. Dzhakobs and K. Potter [15, p. 11] interpret the concept as a term
that refers to negative expressions that incite certain actions or call for hatred towards a person or a
group of persons. This may be expressed in the form of discredit, abuse, abusive language. And
according to the Committee of Ministers of Europe, hate speech includes all the statements that
spread, incite or justify hatred or any form of hatred [16, p. 10].
The OSCE translates ‘hate speech’ as various forms of statements incited by hostility, and
demonstrating or contributing to the acts of hostility towards a particular group [17, p. 17].
According to some researchers [18], the category of "hate speech" should include not only
aggressive vocabulary that directly calls for discrimination or degrades a personality, but also the
development of negative stereotypes about certain social groups. Pavlo Pushkar, a senior lawyer at the
Registry of the European Court of Human Rights made a noteworthy comment that stereotypes are
introduced into general rules that create a certain attitude towards social groups. And we are not
necessarily talking about ethnic origin, but about a certain socio-cultural group in general. It is very
difficult to determine whether certain statements belong to "hate speech", because their restrictions
come into conflict with Article 10 of the European Convention on Human Rights, which guarantees
freedom of speech. However, the exercise of this freedom is subject to certain restrictions, in
particular, related to the protection of the rights of others.
In the study ‘Media Dehumanization and Ethical Standards of Journalism’, Boiko, A. [14, p. 6]
notes that ‘hate speech’ is a cruel or simplistic way of split society into “us” and “them”. Butyrina, M.
[9, p.131] believes that hate speech is often a manifestation of xenophobia of various kinds: ageism,
sexism, migrantophobia and singles out ethnic xenophobia, a subspecies of which is romophobia,
which means hostility and aggression against the representatives of the given ethnic group.
However, the problem with preventing the spread of hate speech is that it is sometimes very
difficult to recognize it. Thus, Veber, A. [16, p. 9] points out that sometimes the hate speech seems
rational and even appropriate to express a personal position. But it is actually built on offensive and
discriminatory grounds.
Kozhevnikova, H. [7, p. 14] tried to solve this problem. She built a system of forms that the hate
speech takes to make it easier to identify. The researcher suggests three forms in which hate speech
may be expressed: harsh, medium and soft. Each of these groups includes statements that carry a
certain emotional colour and semantic meaning.
Thus, harsh hate speech includes: calls for violence; calls for hatred; covert calls for violence and
hatred; calls for not letting a certain phenomenon be established in the city/region/state.
According to the researcher, medium form of hate speech includes: justifying facts or historical
cases of discrimination or violence; public statements or publications that justify recognized acts of
violence and discrimination; mitigating such facts; insisting that discrimination against a particular
religious/ethnic/class group was historically fair; the assertion that a certain group of people deserves
to be discriminated because of certain limitations; emphasis on the presence of unpleasant for society
defects in a particular group; accusing a social/religious/ethnic group of influencing the society, the
state, the lawmaking process; accusing the group of expansion, seizing territories and ousting another
group of people; denying legitimacy of granting citizenship to members of a certain
ethnic/religious/class group.
Soft form includes: creating a negative image of an ethnic/religious/class group; statements about
the inferiority of a certain group of people; statements about the moral flaws of certain groups of
people; providing, citing and using xenophobic statements in negative context or without the
commentary.
Verkhovskyi O. [8] proposes his model of hate speech classification and definition. Thus,
according to his study, “harsh hate speech” includes: open calls for violence; the use of common
slogans to encourage discrimination and hatred; open calls for discrimination; covert calls for hatred
through the prism of historical examples of violence.
According to the study, “medium hate speech” includes: justification of the well-known examples
of hatred; publications that doubt the unlawfulness of historical examples of hatred, violence and
discrimination; emphasizing the crimes of a particular social, ethnic, religious or class group;
emphasizing the contacts of a social, ethnic, religious, or class group with Russian governmental
organizations or other institutions that have gained a negative reputation in a particular society; public
reflections on the disproportion of rights in favour of a certain social, ethnic, religious or class group
in paper publications, government authorities, mass media, etc.; accusing a social, ethnic, religious or
class group of harmful influence on society, legislation, welfare of other citizens; calls not to allow
representatives of a certain social, ethnic, religious or class group into the state/city/region.
According to the researcher, “soft hate speech” includes: verbal or written public creation of a
negative image for a social, ethnic, religious or class group; mentioning the whole social, ethnic,
religious or class group in a negative context, when one of its representatives is involved in a negative
situation; insisting on the inferiority of a social, ethnic, religious or class group; xenophobic
statements or facts in the media without distinguishing between the words of the interviewee and the
author of the text, quoting without comment.
Classification of hate speech into harsh, medium and soft has become an established norm among
the researchers. However, Kroz, M. and Ratynova, N. [19] propose the following classification: false
identification (forming and enforcing a negative stereotype associated with a certain race, nation,
religion, sexual orientation, etc.); false attribution (ascription of hostile attitude and intentions to
representatives of a certain nation, religion, race, political or sexual orientation, etc.); imaginary
defense (pushing to some defense actions against a certain group of people without the obvious
reasons).
Malkova, V. [20] developed a specific system of identification of hate speech in the media
coverage. Following the suggested points, it is easy to determine where the statement expressing the
opinion of the editorial office or an individual journalist turns into a statement of hatred.
Thus, according to the researcher, hate speech includes:
1. threats or calls for violence (murder or physical harm; sexual harassment or rape; political
persecution; repression; deprivation of citizenship; ban on living in a certain area due to belonging
to a hated group; genocide);
2. a call to restrict the residence of certain groups of people on specific lands (ethnic cleansing;
restriction of migration to the country on the grounds of belonging to a certain ethnic group);
3. a call for discrimination (violation of the rights and freedoms declared in the Constitution;
humiliation or restriction of civil rights; deprivation of professional awards, qualification levels);
4. affirmation and support of historical crimes of a certain social group. Emphasis on the
involvement in the crime of all the members of an ethnic/religious/class group;
5. propaganda of historical examples of hatred and discrimination as a positive phenomenon;
6. deliberate falsification of facts, which leads to a negative impression of a social, religious,
class, ethnic group;
7. an indication of the close contact of a group of people with extremist or hated groups in
society, governmental organizations of the enemy country, etc.;
8. assertion that a certain group of people has common personality traits, mental defects, types
of behavior that significantly harm the society and lead to material or moral harm to other people;
9. mentioning the representatives of a social, ethnic, religious or class group in a negative
context;
10. illustrating common human vices on the example of a representative of a certain social,
ethnic, religious or class group with an emphasis on his group affiliation;
11. insisting on emotional, mental, spiritual, physical, intellectual disability of a group of people.
Language is a means of shaping our worldview and the formation of everyday stereotypes, which
is most effectively used by the media. Linguistic and stylistic means in media style are distinguished
by particular tenses and voices, the usage of impersonal constructions, numerous complex attributive
groups, the way of introducing direct speech and transforming it into reported speech, wide use of
complex sentences, etc [21]. According to Kyrylina, A. [22, p. 256], a stereotype is an idea that in a
sharply simplified and generalized form attributes certain properties to a certain class of people or,
conversely, denies them. Stereotypes are special forms of information perception that guide a person
in the environment. The features contained in stereotypes are used by speakers to assess the affiliation
of objects to a particular class and attribute certain characteristics to them. Maslova, Yu. [23, p. 83]
notes that “stereotypes in language perform a generalizing function of organizing knowledge about
insiders and outsiders, which leads to social categorization and the formation of structures that
become landmarks for people in everyday life”. Speaking of stereotypes, one cannot ignore such an
important category as ethnostereotypes. Ethnostereotypes, in turn, are divided into autostereotypes
(self-perception of an ethnos) and heterostereotypes (perception of other ethnic groups). Accordingly,
autostereotypes have a linguistic expression in the form of self-naming of representatives of one
group, and heterostereotypes – in the form of naming the representatives of other groups. Stereotyping
involves the following human behavior towards an "outside" group in the following order:
verbalization (humor, irony, sarcasm), avoidance, discrimination (exclusion from certain areas of
society), acts of violence, destruction of the group. Therefore, the negative impact that the
transmission of hate speech has on society can take the form of both exacerbation of conflicts between
individuals and mass destruction of one social group by another. And although the role of a journalist
is to impartially convey information, a journalist is also a human being who, consciously or
unconsciously, may be prejudiced against a certain social group, which will finally affect the quality
of the material and will be perceived by the audience accordingly.
Thus, hate speech against the representatives of national and ethnic groups may be defined as
statements (photos or images) of a negative nature towards a person, a representative of a certain
national or ethnic community, or a group as a whole, based on stereotypes and promoting, deepening
or supporting intolerance of the society. That is, the concept includes four elements: statements
(photos or images) of a negative nature; a person belonging to a certain nation or ethnic group, or the
whole nation (ethnic group); promoting, deepening or supporting intolerance towards this group;
based on stereotypes.
The following factors may be considered to be the biggest problems in hate speech against the
members of national or ethnic groups: generalization (when the wrong-doings of one person are
transferred to the whole group), increasing hostility and unneighbourly behaviour, deepening
historical interethnic trauma, calling for action by non-legal means. As long as stereotyping is focused
on humor level rather than direct discrimination, it can be assumed that the impact on a person's
perception of a particular social group is minimal, for example, the "Americans are friendly"
stereotype will not negatively affect the group's perception (on the contrary, somebody may copy such
behaviour), but the "Roma are thieves" stereotype indicates that everyone needs to be careful with
them. The problem of soft hate speech is insufficiently researched, but its role cannot be
overestimated, as the constant reproduction of such messages forms the foundation of harsh hate
speech in the society, which can provoke even physical impact on the members of a certain social
group, in particular the Roma.
4. Linguistic means of hate speech expression
In the work "Speech aggression and language tools to prevent verbal hostility", Bondarenko, T. [6,
p. 145] notes that to denote the phenomenon associated with inappropriate statements, researchers use
a number of terms: intolerant vocabulary, language hostility, language aggression, speech aggression,
lexical aggression, verbal aggression, verbal extremism, hate speech, etc., which are mostly
considered synonymous. The most common is the concept of aggression, which is studied in various
scientific fields, and contextual definitions are being developed, where by aggression we mean
individual or collective behavior aimed at inflicting physical or psychological harm to a particular
person, group of people [24, p. 16]. Linguists often insist on distinguishing between language
aggression and speech aggression, where language aggression is the imposition of a non-mother
tongue, and speech aggression is oral expression, written text with vulgarisms, obscene language
(swear words) for expressing indignation and contempt. That is, when we speak of hate speech
expressed through the media, we use the term “speech aggression”. Pavelchak, P. [25, p. 324]
identifies four aspects of coverage of the image of Roma in the media: as representatives of the
criminal world; as a community that has long been discriminated against; as activists engaged in
active cultural and social activities; as people who experience great poverty in everyday life. Erjavec,
K. [5, p. 726] in the article ‘Media Representation of Discrimination against the Roma in Eastern
Europe. The case of Slovenia’ also attempts to analyze (thematically and structurally) news reports to
cover the problem of discrimination against the Roma population. She concludes that the Roma
people are portrayed mainly in a negative light, and the corresponding language signals are visible
even in the headlines, the news is often axiological in nature, there is a clear phenomenon of
criminalization of ethnic groups.
Yevstafieva, A. [26, p. 994-996] identified the means of linguistic manipulation used by the media
to trigger negative reaction of viewers, listeners or readers towards the group of people depicted in the
story, in particular: attaching labels (communist, fascist, anti-Semite, monarchist, orthodox, gypsy);
the use of speech patterns that disparagingly highlight the person in question (such as you, like you,
through you); generalizations (gathering of homeless people, drug addicts and thieves); division and
emphasis on the status of ‘us’ or ‘them’; forming an unattractive perspective, due to the mention or
depiction of a group of people or its individual representatives only in negative situations. In addition,
the researcher identifies linguistic markers with the help of which the media often manipulate the
opinion of the audience. Their knowledge will help to analyze the content of resources and highlight
examples of hate speech more easily. Such markers include: colloquial language, i.e., jargonisms,
swear words, vernacular speech; altered literary expressions that explicitly show the author's intention
to cause reputational damage, belittle the subject of discussion, or cause negative associations with the
person or group of persons referred to in the material; deliberate formation of stereotypes and
attaching labels; use of bitter humour; use of leveling words, for example, ‘so-called’, ‘as if’,
‘actually’; semantic combination of concepts or phenomena into a synonymic row or cause-and-effect
relation, when in fact there is no connection or identical features between them; direct calls for
discrimination and hatred; covert or vague incitement to hatred and violence; frequent repetition of
words, phrases, sentences or thoughts in order to emphasize them; mocking distortion of proper
names.
One of the problems of journalists is that they do not use neutral vocabulary, including tolerant
statements; lexical units to indicate the difference from the standard; neutral ethnic expressions.
Malkova, V. [20, p. 38] identifies the following intolerant statements: ethnic senselessly outrageous
expressions; mockery statements; accusations; lexical labels; stereotypic expressions. It should be
noted that intolerant word “gypsies” is less frequently used in the materials about the representatives
of Roma people, and the word “Roma” prevails, which proves that there is an increase in the level of
tolerance in the media materials towards the definition of people belonging to the community. In
particular, in the Dictionary of the Ukrainian Language [27, p. 208] there is only a definition of
“Gypsies”, while in the Encyclopedia of the History of Ukraine [28] the word “Roma” is used.
Representatives of the given ethnic group identify themselves as “Roma”, which means “man”, and
they consider the word “Gypsies” to be offensive. The stereotypical perception of the Roma, which
has developed over the course of history, still dominates in the public perception and is enshrined in
everyday language, in particular, we can trace it on the example of set phraseological units, such as:
spin like a gypsy spins the sun; tempted, like a gypsy for lard; wish something like a gypsy penny;
change like a gypsy; black as a gypsy; respectable as a gypsy; rich as a gypsy in fleas; gypsy life;
gypsy truth. All these statements have a negative connotation, which exposes the Roma as cruel,
greedy for money, dirty, uneducated, poor, disorganized, liars and so on. The use of intolerant
designation of the ethnic group members indicates a desire to classify their representatives as "lower",
"second" class. Taking into consideration a newsworthy event, this material is a direct manifestation
of the hate speech, which is disturbing and worrying sign. We came across several such headlines in
the media, including: “Murder of a 9-year-old child in Loshchynivka: trucks will come for gypsies
tomorrow” [29], “Murder in Loshchynivka: Azov declared war against the gypsy plague” [30]. We
believe that moving away from the intolerant word ‘gypsy’ will gradually change the language
culture, leading to the elimination of negative stereotypes about the Roma community. In the
Ukrainian media space, we often come across the statements about Roma in a negative context,
which, given the intolerant, according to the research, attitude towards them can lead and leads to
direct acts of aggression against them.
To visualize the stereotypical attitude towards the Roma, we researched the materials of the news
website ZAXID.Net during 2020, as it has a wide readership, especially in western Ukraine, where a
large part of the Roma community of Ukraine lives.
During 2020, 13 articles about people belonging to the Roma community were published on the
website ZAXID.Net [31]. All of them were written in the form of information messages by the classic
type of inverted pyramid, which contains relevant and socially important information, where in the
lead-in or at the beginning of the message there is the essence of the news, and there are the answers
to six questions – who, what, where, when, how, why. In 12 headlines we found an indication of
belonging to the given ethnic group, in particular: “For a knife attack on a passerby in Lviv, a 31-
year-old Roma woman was sentenced to 5 years in prison”, “In the center of Lviv, two Roma with
knives attacked a 22-year-old passerby”, “Special forces soldiers were brought to a village in Kharkiv
region due to clashes with Roma”, “A group of Roma beat a 50-year-old Lviv resident during an
attempted robbery”, “A Roma woman attacked a passerby with a knife in Pryvokzalna street in Lviv”,
“At night near the “Metro” shopping center there was a knife fight between Roma”, “The mayor of
Ivano-Frankivsk ordered to take Roma to Zakarpattia”, “Five Roma received seven years in prison
each for a theft at Lviv railway station”, “A married Roma couple was convicted in Lviv for killing
their own child”. Therefore, 9 out of 13 materials contain criminal news, indicating the nationality of
only one party involved – Roma, and the indication of nationality in this context is superfluous, so it
was done to support the stereotype that all the Roma are thieves, murderers, or bandits. Only in one
headline there was no indication of nationality, but the text mentioned that the victim was a resident
of Zakarpattia, which means one thing for the people of Ukraine – he is a Rom, and, in addition, the
news item was included in the section about Roma people.
To sum up, it may be concluded that the existing language stereotypes about Roma are present in
the Ukrainian media space, and are most frequently expressed through the evaluative stereotype,
which is an expression of comparison “good/bad”, the functioning of established phraseological units
that support stereotypes about the Roma people that exist in the society. The following violations
being a sign of hate speech in the studied materials on Roma, may be singled out: unjustified mention
of nationality in the headlines and texts of articles (in 12 of 13 studied); the materials about the life of
the Roma people are exclusively criminal in nature (generalizations – all Roma are criminals); the
choice of words and topics support a negative, hostile attitude towards members of the ethnic
community; division and emphasis on the status of “us” and “them” (“passerby in Lviv – Rom”,
“Roma – special forces soldiers”, “Roma woman – passerby”, “mayor – Roma”, “Roma – a woman
from Lviv”, 'Roma – law enforcement officials”, “Roma – people”, “inhabitants – local authorities –
Roma”, “rural population – Roma”).
5. Peculiar features of the technology of automated language assessment
and search for inappropriate use of concepts
The current state of automatic text analysis technologies contributes only to a partial solution of
the problem, but allows us to measure the tension in society in respect of current, urgent issues. There
are two main approaches to language assessment: the first is based on lexicons, the second – on the
use of machine learning methods.
Language assessment of the text using machine learning methods involves its processing through a
preliminary analysis of the collection of texts with predetermined estimates of the tonality of words
and expressions. Requirements for the information system, the work of which is based on a model that
promotes "learning" as exemplified by the formed text corpus, have been developed.
Under this approach, it is acceptable to define general principles of construction, processing and
application of linguistic corpus data (corpora of texts), develop the methods for analyzing real
linguistic phenomena – written and oral texts, as well as ways to preserve them.
Computer assessment of the text tonality is established experimentally, being marked by operators,
which is then used as a uniformed standard in sentiment analysis. This approach is characterized by
several aspects that are both correct and not exactly correct, because it does not always give accurate
results of language assessment. One of the most significant shortcomings of the analysis of the text
language assessment using machine learning methods is the probability of conflicts and differences in
the assessment of the tonality of texts on various topics. Under such conditions, the program may
incorrectly monitor the language assessment of certain elements, as the procedure involves the
formation of a comprehensive assessment, which can often be inappropriate under the analyzed
context. It should be noted that machine learning methods do not allow the creation of databases that
would take into account the features of language assessment, which are of high significance in the
overall tonality of the whole text.
The operation of the proposed information system is based on the use of dictionaries and lexicons,
from which certain concepts are extracted, taking into account their tonality assessment. However,
dictionaries with rules are still created manually. The first step in developing the program was to
create a list of expressions and determine their tonality. Then the message is processed, the words and
phrases that characterize concepts or phenomena in a synonymic row are extracted, using language
assessment, weight coefficients are assigned and registered in the created dictionary. The intensity of
the language aggression present in the text of the message is calculated as the arithmetic average for
the studied text.
This is the way how modeling of the impact of the language context on the audience is carried out.
The analysis of the message cannot be correct without taking into account the context that is reflected
in the means of language assessment. The functionality of the system allows distinguishing the use of
neutral vocabulary, including tolerant statements; lexical units to indicate the difference from the
standard; neutral ethnic expressions and sentimentally loaded concepts. Such linguistic evaluation can
be strengthened by introducing special attributes for words: “good” can be used as a positive, and
with a particle acquires a negative tonality, or words that diminish positive semantics.
Evaluative stereotype (standard, norm sample) as an element of language assessment is clearly
traced in the texts of the analyzed materials, which is an expression of comparison “good/bad”, such
as: “passerby in Lviv – Rom”, “Roma – special forces soldiers”, “Roma woman – passerby”, “mayor
– Roma”, “Roma – a woman from Lviv”, “Roma – law enforcement officials”, “Roma – people”,
“residents – local authorities – Roma”, “rural population – Roma”. On the one hand, we have “good”
characters: a passerby, a man/woman from Lviv, a resident, rural population, law enforcement
officials, special forces soldiers, PEOPLE, because, according to established stereotypes, they have a
positive connotation. On the other hand, there are “bad” characters: the Roma people. The description
of the people involved is based on various criteria: the nationality of one party (Roma) and the place
of residence (Lviv resident, passerby, a woman from Lviv, rural population), position (special forces,
mayor, law enforcement officials) of the other party, although no one indicated which nationality
these people belonged to, which is an expression of the opposition “us” and “them”. The selection of
verbs (attacked, beat, robbed, killed, etc.) indicates a rational assessment of the authors of the
materials.
However, it is impossible to “teach” the program how to distinguish these nuances, as they have to
be analyzed in different ways. That is why databases or text corpus are created based on the published
material to accumulate a sufficient number of contextual examples.
Another challenge for the researchers in the process of developing software for sentiment analysis
is to take into account the order of words, expressions and sentences in the texts and their impact on
the overall tonality. Such an obstacle arises when a positive assessment of certain aspects is
expressed, but the general subjective impression of the object is negative.
In particular, in the material by Volodymyr Pyrih as of May 15, 2020 entitled “Roma community
from Zakarpattia wants to sue the mayor of Ivano-Frankivsk for 250 thousand UAH”, there is a
phrase that “previously unknown to anyone Roma Community of Zakarpattia is at law with
Martsinkiv” [32]. In the Academic Explanatory Dictionary of the Ukrainian Language [33] the word
“sudytysia” (“be at law with”) means: - to go to court, to deal with a court, to sue someone. The word
refers to the informal style of speech used in oral communication in everyday life, family, and the
main purpose of which is the use in informal communication, lively exchange of thoughts, judgments,
assessments, feelings, showdowns. Synonyms of this word are: pozyvatysia (bring legal action against
somebody), pravuvatysia (go to the law), protsesuvatysia (initiate legal proceedings), tiahatysia po
sudakh (have a litigation). The root of the word is an independent word “sud” (“court”), the
synonymic row of which is even broader and includes, in particular: sudovyi protses (trial), rozprava
(reprisal), sudove zasidannia (court session), zhiuri (jury), arbitrazh (arbitration), trybunal (tribunal),
samosud (lynch law), kryvosud (mock trial), sudyshche (kangaroo court), sudovysko (kangaroo
court), etc. Thus, the given word is an evaluative lexical unit denoting an action, containing an overt
negative assessment of those being “at law”, namely the Roma community, because they are “being at
law”, “arranging reprisals”, “conducting a kangaroo court”.
In addition, the combination of words 'previously unknown to anyone' contains three manipulative
generalizations – “anyone”, “unknown”, “previously”, the purpose of which is not to inform, because
it does not contain factual information, but to provoke certain thoughts, diminish the role of the public
organization “Roma Community of Zakarpattia” on the grounds that it was recently established and
has not yet “deserved” to address someone, although the the legislation of Ukraine does not contain
such a condition – the period of existence before the first application. Logical and linguistic
techniques are used for the speech implementation of logical operations, in particular the techniques
of “comparison”, “opposition”, “generalization”, which contribute to the implementation of invective
tactics by establishing semantic connections between the invectum and a phenomenon that is
evaluated negatively. Also, they provide a speaker with wide opportunities to ridicule the opponent.
In this combination of words, we see the technique of “pseudo-fact implantation”, which is a
categorical unsubstantiated assertion of certain facts that discredit the invectum, because we cannot
check whether it is really “previously unknown to anyone”. Therefore, we can see a manipulative
evaluative judgment, the purpose of which is to provoke distrustful attitude towards the organization.
A small amount of lexical units with an evaluative component in the meaning, recorded in
information texts, is formed through the use of word-formation tools by the method of composition,
for example “newly created”. In the last two paragraphs of the note, we come across four references
to this fact – “registered on April 24”, “three days after the scandalous statements”, “newly created”,
“manager from April 8 (error – from May 8)”. The emphasis on the fact that the organization was
established after an act of discrimination against the Roma community as well as repetitions indicate
the attitude of the author of the material, because such a technique in the text is set out in an
informational style and serves as a means of expressing negative evaluation. The problem of an
inappropriate headline or announcement is not only about journalism, but also advertising, the
purpose of which is to attract and retain the attention of the audience. These two or three bright but
inappropriate phrases are remembered more than all the material, and if we take into account that a
large part of audience read only the title of the publication and lead-in, the impression of Roma will
be formed as negative, which only supports the existing stereotype about this ethnic group.
The first part of the study analyzes the headlines of journalistic materials and article texts
published by Zaxid.net in manual mode. The study of using words being markers of hate speech
against the members of the Roma community, as exemplified by full-text materials posted on the
official website of Katerynopil State Administration, was carried out to compare and contrast the
results. Thus, a corpus was created, which made it possible to demonstrate the use of concordance for
the analysis of materials and journalistic articles about the life of the Roma community members in
Ukraine.
Due to the lack of effective tools required for the analysis of the Ukrainian texts, the corpus of
texts, being examined for the presence of hate speech, was translated into English for the purpose of
the research.
To track the inappropriate use of the ethnic group name, in particular as exemplified by the term
"gypsies", concordance program or corpus managers may be used. Let us consider the possibility of
tracking the frequency of use of the word “gypsy”, which is offensive to Roma, with the help of the
AntConc corpus manager [34]. For example, in a small material posted on the official website of the
Katerynopil Raion State Administration and Raion Council entitled “10 little-known facts about the
Gypsies” [35], it is used 21 times.
Figure 1: The result of text analysis
This software product allows not only searching and counting various elements of the text, but also
analyzing their frequency and context of the use of word forms, word combinations and morphemes.
Figure 2: Contextual use of the word
The use of this tool will help to pre-moderate journalistic materials in terms of discriminatory,
insulting statements, as well as hate speech against members of ethnic communities and to monitor
compliance with journalistic standards in media content.
6. Results/Discussions
Although tolerance and non-discrimination of people and groups to which they belong on the basis
of racial, ethnic or other differences is fairly well regulated by laws, according to various surveys –
the problem really exists, and modern media contribute to its spread and support as well. It is
necessary to draw boundary lines between freedom of expression, hate speech and the right to non-
discrimination.
Hate speech against the representatives of national and ethnic groups may be defined as statements
(photos or images) of a negative nature towards a person, a representative of a certain national or
ethnic community, or a group as a whole, based on stereotypes and promoting, deepening or
supporting intolerance of the society. That is, the concept includes four elements: statements (photos
or images) of a negative nature; a person belonging to a certain nation or ethnic group, or the whole
nation (ethnic group); promoting, deepening or supporting intolerance towards this group; based on
stereotypes.
The following factors may be considered to be the biggest problems in hate speech against the
members of national or ethnic groups: generalization (when the wrong-doings of one person are
transferred to the whole group), increasing hostility and unneighbourly behaviour, deepening
historical interethnic trauma, calling for action by non-legal means. As long as stereotyping is focused
on humor level rather than direct discrimination, it can be assumed that the impact on a person's
perception of a particular social group is minimal, for example, the "Americans are friendly"
stereotype will not negatively affect the group's perception (on the contrary, somebody may copy such
behaviour), but the "Roma are thieves" stereotype indicates that everyone needs to be careful with
them. The problem of soft hate speech is insufficiently researched, but its role cannot be
overestimated, as the constant reproduction of such messages forms the foundation of harsh hate
speech in the society, which can provoke even physical impact on the members of a certain social
group, in particular the Roma.
7. Conclusion
Language stereotypes about the Roma people are present in the Ukrainian media space,
and are most frequently expressed through the evaluative stereotype, which is an expression
of comparison “good/bad”, the functioning of established phraseological units that support
stereotypes about the Roma people that exist in the society. The following violations being a
sign of hate speech in the studied materials on Roma, may be singled out: unjustified mention
of nationality in the headlines and texts of articles (in 12 of 13 studied); the materials about
the life of the Roma people are exclusively criminal in nature (generalizations – all Roma are
criminals); the choice of words and topics support a negative, hostile attitude towards
members of the ethnic community; division and emphasis on the status of “us” and “them”
(“passerby in Lviv – Rom”, “Roma – special forces soldiers”, “Roma woman – passerby”,
“mayor – Roma”, “Roma – a woman from Lviv”, 'Roma – law enforcement officials”,
“Roma – people”, “inhabitants – local authorities – Roma”, “rural population – Roma”).
Computer assessment of the text tonality is established experimentally, being marked by
operators, which is then used as a uniformed standard in sentiment analysis. This approach is
characterized by several aspects that are both correct and not exactly correct, because it does not
always give accurate results of language assessment. One of the most significant shortcomings of the
analysis of the text language assessment using machine learning methods is the probability of
conflicts and differences in the assessment of the tonality of texts on various topics. Under such
conditions, the program may incorrectly monitor the language assessment of certain elements, as the
procedure involves the formation of a comprehensive assessment, which can often be inappropriate
under the analyzed context. It should be noted that machine learning methods do not allow the
creation of databases that would take into account the features of language assessment, which are of
high significance in the overall tonality of the whole text.
The operation of the proposed information system is based on the use of dictionaries and lexicons,
from which certain concepts are extracted, taking into account their tonality assessment. However,
dictionaries with rules are still created manually. The first step in developing the program was to
create a list of expressions and determine their tonality. Then the message is processed, the words and
phrases that characterize concepts or phenomena in a synonymic row are extracted, using language
assessment, weight coefficients are assigned and registered in the created dictionary. The intensity of
the language aggression present in the text of the message is calculated as the arithmetic average for
the studied text.
However, it is impossible to “teach” the program how to distinguish these nuances , as they
have to be analyzed in different ways. That is why databases or text corpus are created based on the
published material to accumulate a sufficient number of contextual examples. Using the capabilities
of the concordance program and the AntConc corpus manager will help to pre-moderate journalistic
materials in terms of discriminatory, insulting statements, as well as hate speech against members of
ethnic communities and to monitor compliance with journalistic standards in media content.
8. References
[1] J. W. Howard, Free Speech and Hate Speech. Annual Review of Political Science 22 (2019) 93-
109. doi:10.1146/annurev-polisci-051517-012343.
[2] M. Herz, P. Molnar, The Content and Context of Hate Speech: Rethinking Regulation and
Responses. Cambridge: Cambridge University Press, 2012.
[3] G. Elford, Legitimacy, Hate Speech, and Viewpoint Discrimination. Journal of Moral Philosophy
1 (2020) 1-26. doi.org/10.1163/17455243-20203306.
[4] Z. Laub, Hate Speech on Social Media: Global Comparisons. URL:
https://www.cfr.org/backgrounder/hate-speech-social-media-global-comparisons
[5] K. Erjavec, Media Representation of the Discrimination against the Roma in Eastern Europe:
The Case of Slovenia. Discourse & Society 12(6) (2001) 699–727.
doi.org/10.1177/0957926501012006001
[6] T. Bondarenko, Speech Aggression and Language Tools for Preventing Verbal Enmity. Human
Rights and Mass Media in Ukraine, Kyiv, 2018, pp. 137–147.
[7] G. Kozhevnikova, Hate Speech: Typology of Journalist Mistakes. Applied Conflictology for
Journalists, Moskow: Human Rights, Center for Extreme Journalism, 2006, pp. 95–105.
[8] A. Verkhovsky, The Language of Enmity against Society, Moscow: Sova Сenter, 2007.
[9] M. Butyrina, Xenophobic Discourse: Causes, Scale, Consequences and Lessons for the Media.
Human Rights and Mass Media in Ukraine, Kyiv: Institute of Journalism of Taras Shevchenko
National University, 2018, pp. 127–136.
[10] A. Kostovska, Between Roma and Gypsies: Stereotypes and Discrimination in the Ukrainian
Internet Media. East of Europe 3(2) (2017) 39–50. doi.org/10.17951/we.2017.3.2.39.
[11] N. Kozyr, O. Grigoriev, The Language of Hostility in the Internet Media against Members of the
Roma National Minority in Transcarpathia. URL: http://khpg.org/1394667648
https://www.youtube.com/watch?v=PdGi5g2eatk
[12] National Institute for Strategic Studies. The State of Tolerance of Ukrainian Society: Risks and
Opportunities for the Formation of National Unity. Analytical Note. URL:
https://niss.gov.ua/doslidzhennya/nacionalna-bezpeka/ocinka-stanu-tolerantnosti-ukrainskogo-
suspilstva-riziki-i
[13] N. Ilyuk, Media Education of Youth as a Determinant of Modern Information Society. Education
of the Region: Political Science, Psychology, Communications 2 (2011) 121–125.
[14] A. Boyko, Dehumanization of Media and Ethical Standards of Journalism (on the Material of
Ukrainian Media for 2013-2014). Bulletin of Lviv National University. Series: Journalism 39(1)
(2014) 5–11.
[15] J. B. Jacobs, K. Potter, Hate Crimes: Criminal Law and Identity Politics. Oxford – New York:
Oxford University Press, 1998.
[16] A. Weber, Textbook on "Hate Speech". Kyiv: Tiutiukin, 2010.
[17] OSCE Office for Democratic Institutions and Human Rights. Hate Crimes: Prevention and
Reacting. Warsaw: Poligrafus Jacek Adamiak, 2015.
[18] M. Dorosh, The Language of Enmity – New Manifestations and Consequences. URL:
https://ms.detector.media/zakonodavstvo/post/12829/2015-03-17-mova-vorozhnechi-novi-
proyavy-ta-naslidky/
[19] M. Kroz, N.Ratinova, The Price of Hatred. Expert Assessment of Materials Aimed at Arousing
Enmity and Hatred. The Price of Hatred, Moscow: Sova Сenter, 2005, pp. 75–92.
[20] V. Malkova, Incitement of Ethnic Hatred is Not Allowed. A Book about Ethnic Journalism.
From the Experience of the Analysis of the Russian Press, Moscow: Academia, 2007.
[21] Z. Haladzhun, Linguistic Characteristics of the Construct "Evaluative Judgement of a Journalist"
in The Ukrainian Language. Advanced Education 12 (2019) 249–255.
doi: https://doi.org/10.20535/2410-8286.113290.
[22] A. Kirilina, Gender Stereotypes in Language. Glossary of Gender Terms, Moscow: Information
XXI century, 2002.
[23] Y. Maslova, Formation and Circulation of Gender Stereotypes in the Society (on the Example of
the Current Ukrainian-language Media). Scientific Proceedings of Ostroh Academy National
University: Philology Series 40 (2013) 81–87.
[24] A. Zagnitko, Glossary of Modern Linguistics: Concepts and Terms, Donetsk: Vasyl Stus
Donetsk National University, 2012.
[25] M. Adamik-Szysiak, E. Godlewska, Media Minorities: Minorities in the Media, Lublin: Maria
Curie-Skłodowska University, 2014.
[26] A. Evstafieva, About Some Linguistic Markers of "Hate Speech" in Manipulative Techniques in
the Media. Bulletin of Bashkirsk University 4 (2008) 994–996.
[27] I. Bilodid, Dictionary of Ukrainian Language, Кyiv: Naukova dumka, 1980.
[28] M. I. Tyahlyi, Roma. In Institute of History of Ukraine. URL: http://resource.history.org.ua/cgi-
bin/eiu/history.exe?&I21DBN=EIU&P21DBN=EIU&S21STN=1&S21REF=10&S21FMT=eiu_
all&C21COM=S&S21CNR=20&S21P01=0&S21P02=0&S21P03=TRN=&S21COLORTERMS
=0&S21STR=romy
[29] O. Tarasyuk, Murder of 9-year-old Child in Loshchynivka: Trucks will come for Gypsies
Tomorrow. URL: https://fakty.com.ua/ua/proisshestvija/20160828-1593013/
[30] V. Letyak, Murder in Loshchynivka: Azov Declared War on the "Gypsy Plague". URL:
https://fakty.com.ua/ua/ukraine/20160831-ubyvstvo-u-loshhynivtsi-azov-ogolosyv-vijnu-
tsyganskij-chumi/
[31] Roma. URL: https://zaxid.net/romi_tag48874/
[32] V. Pyrih, The Roma Community from Transcarpathia Wants to Sue the Mayor of Ivano-
Frankivsk for 250 Thousand Hryvnias. URL:
https://zaxid.net/romska_spilnota_iz_zakarpattya_hoche_vidsuditi_v_mera_ivano_frankivska_25
0_tis_grn_n1502272
[33] V. V. Dubichynskyi, Modern Explanatory Dictionary of Ukrainian Language: 100 000 Words,
Kharkiv: Shkola, 2009.
[34] AntConc. URL: http://www.laurenceanthony.net/software/antconc/
[35] Katerynopil District State Administration, Katerynopil District Council. 10 Little-known Facts
about Gypsies. URL: http://katerynopiladm.gov.ua/upravlinnja-justitsiyi-informue/item/8215-10-
malovidomih-faktiv-pro-tsigan