Why do we need a sociocognitive-CDA in hate speech studies? A corpus-based systematic review

Abstract

This article is a systematic review of previous research on hate speech in discourse studies indexed in the Scopus database in the last five years, from 2015 to 2021. This review aims to map the main topics and methods used in hate speech studies and then provide critical remarks related to the methodological issues. The review focused on 70 selected articles and was analyzed using the NVivo 12 combined with the LancsBox 6.0. Based on the analysis, previous studies show a strong relationship between hate speech, political issues, and discrimination. Although many studies of hate speech have applied the CDA approach, and some of them have used sociocognitive CDA, it is surprising and unfortunate that previous researchers did not show much attention and further exploration of the cognitive aspects of the theory itself. It means that the cognitive aspects of hate speech have yet to be appropriately explored.

Keywords

hate speech CDA cognition systematic review politics discrimination corpus

I. Introduction

Hate speech is a social phenomenon recently attracting world scientists’ attention. Interdisciplinary collaboration has also been carried out to uncover the phenomenon of hate speech. Critical Discourse Analysis (CDA) is one of the significant theories among the scientific disciplines used to uncover the phenomenon of hate speech. This is because the phenomenon of hate speech is closely related to the dynamics of politics and power.

CDA theory has undoubtedly provided an understanding of the relationship between language and power. Almost all CDA experts agree that the working area of this theory lies in efforts to uncover the hidden meaning behind verbal expressions, starting from Fairclough, Wodak, van Dijk, and others. However, among several CDA approaches, only the sociocognitive model CDA seriously focuses on the cognitive aspects of talks and texts. By revealing the cognitive aspects, both personal and social cognition, in a discourse event, the relationship between speech and the complex social dynamics can be explained more easily.

Thus, hate speech is a cognitive phenomenon when observed from the perspective of sociocognitive CDA. In this case, someone understands the existing social reality based on everyday life experiences stored in his episodic memory. In other words, one is not born to hate other people or groups. However, hatred or hate speech grows in one’s cognition along with the continuous negative construction of other groups as a member of certain epistemic community. Within a particular time, this negative view then appears in a person, controlling his knowledge and attitude, including hate expression.

In other words, viewed from the lens of sociocognitive CDA, speech expressions, including hate speech, are not dictated by the material context that is out there. Conversely, it is driven by the accumulation of “hatred models” constructed in a person’s mind obtained through daily life experiences, which in sociocognitive CDA terminology is called a mental model (van Dijk, 2006; van Dijk, 2008; van Dijk, 2014) – apart from the fact that van Dijk reluctantly uses the term hate speech in the ideological context (van Dijk 2020; van Dijk 2021). Therefore, dismantling the phenomenon of hate speech cannot be separated from efforts to dismantle mental models stored in one’s episodic memory at the micro-level and dismantle the social memory of society at the macro-level. At this point, the sociocognitive model CDA approach is considered a very strategic choice in hate speech research.

However, to what extent have these mental or cognitive aspects been considered in previous hate speech research? In the last five years, from January 2015 to March 2021, there have been at least four systematic-review articles related to hate speech that previous researchers have carried out; for example, the review conducted by Matamoros-Fernández and Farkas (2021), Castaño Pulgarín et al. (2021), Paz et al. (2020), Chetty and Alathur (2018). However, none of these systematic reviews is limited to aspects of discourse studies, mainly related to cognitive aspects of discourse. Therefore, a clear picture of the use of discourse analysis methods, especially sociocognitive CDA, to study hate speech has not been mapped out well.

In this regard, to provide a more comprehensive picture related to previous research, this study aims to map the main topics and methods of hate speech studies that previous researchers have carried out. Furthermore, from there, we provide critical notes related to some of the shortcomings of the previous research.

II. Research design

This article is a systematic review of hate speech research indexed in the Scopus database. Scopus was selected because, from January 2015 to March 2021, there has been no systematic review conducted by previous researchers on hate speech that uses Scopus as a data source. In addition, Scopus is considered to have an outstanding reputation in terms of indexing data for international journal articles.

Based on Figure 1, we first searched for hate speech articles in the Scopus database. At least from January 1962 to March 2021, the number of articles related to hate speech recorded in the Scopus database was 1,917 documents or articles. To get an overview of hate speech research from January 1962 to March 2021, we analyzed the metadata of 1,917 articles that had been converted into a dataset, including titles, abstracts, and keywords. After analyzing the metadata, we narrowed the scope of the search by making restrictions. The article documents were limited to research articles and review articles published in the last five years, from 1 January 2015 to 29 March 2021. We only took English-language articles published in journals with the scope of social science, arts and humanity, and psychology. The search string entered into the search query were limited to four combinations, as below.

The first combination: “hate speech” AND “critical discourse” AND “online media,” obtained 88 articles;

The second combination: “hate speech” AND “critical discourse analysis” AND “social media,” obtained 41 articles;

The third combination: “critical discourse analysis” AND “hate speech” AND “social media,” obtained 27 articles;

The fourth combination: “hate speech” AND “deconstruct*,” obtained 9 articles; use the asterisk symbol (*) means one or more characters. As it was written following the search string, thus keywords such as deconstruction, deconstructing, and deconstructed were included.

Figure 1.

Diagram of research stages starting from the steps carried out, analysis processes, and results.

The choice of the four combined models is based on the consideration that the research to be carried out is related to hate speech in the (online) media. For us, hate speech requires a CDA approach. The keyword “deconstruct*” is included because it is closely related to critical studies. In addition, in the practice of CDA, deconstruction is sometimes needed to denaturalize an established discourse.

From these restrictions, a total of 165 articles out of 1,917 were found in the Scopus database. The articles were narrowed again because there were 47 overlapping articles, leaving 118 articles. The 118 articles were then observed one by one (close reading) to see the level of relevance according to need, namely research articles related to the hate speech in online media that apply CDA approach. In this process, 48 articles were removed because they were deemed less relevant and considered to have been represented by other articles. For example, conceptual articles were removed because the articles were regarded out of our need or less relevant. In the end, we decided on only 70 articles (full text) as the corpus data for review. In this paper, we named this corpus data Scopus Hate-Speech Corpus (SHSC).

Furthermore, for processing management and analysis, we used the NVivo 12 application combined with the LancsBox 6.0. The NVivo 12 is used as the basis for processing management, starting from selecting the most frequent words, the coding process, visualization, and quantitative data analysis. Meanwhile, the LancsBox 6.0 is applied to strengthen our interpretation by looking at the close relationship between one research topic and another; by looking at the collocation and concordance of the selected words or keywords, interpretation can be made more carefully (see Brezina, 2018; compare to Prihantoro, 2015). Although there are other similar tools, using LancsBox 6.0 is mostly for practical reasons. The tool allows users to import corpora in pdf format, and it also supports data visualization. At the same time, we also made in-depth observations or close reading of the 70 full texts reviewed articles for further discussion and grasping deeper comprehension. In general, the process of implementing this systematic review can be described in a schematic, as shown in Figure 1.

III. Results and analysis

3.1 Main Topics of Hate Speech Studies

Based on the number and distribution of keywords in the articles, statistically, essential topics for the study of hate speech conducted by previous researchers can be described in Figure 2. Here, keywords are referred to and taken from the most frequent words as a result of the search query process in NVivo 12 by displaying 60 words with a minimum length of 9. We took ten words based on the consideration that the words were regarded as conceptual words and the most relevant to the topic of study. Therefore, some words regarded as less relevant or irrelevant were excluded, e.g., the word university, countries, information, different, following, and especially.

Figure 2.

Graph of the main topics in the hate speech studies based on the number of keyword occurrences in SHSC (analyzed using NVivo 12).

Judging from the articles that have been reviewed, as shown in Figure 2, Figure 3, and Figure 4, the study of hate speech in the last five years shows a close relationship with social and political life which is so complex. At least ten significant issues emerged in the previous study, ranging from the issue of democracy, discrimination, government, immigration, religion, politics, terrorism, and gender. These issues are almost evenly distributed in each article reviewed, as seen in Figure 4. This means that the relationship between one issue and another cannot be separated. Therefore, it must be mapped again from all the issues that arise from selected keywords, which issues are central or most prominent, and which are less. In this regard, in this review, the central issues related to hate speech can be divided into at least three main issues, namely (1) hate speech is political, (2) hate speech is discriminatory, and (3) there are social actors behind the speech. The other issues can be seen as side issues that reinforce or emphasize the three. These issues can be elaborated on respectively in the following.

Figure 3.

Graph of the main topics in the hate speech studies based on the distribution of keyword occurrences in the SHSC (analyzed using NVivo 12).

Figure 4.

Graph of the main topics in the hate speech studies based on the number of articles related to keywords in the SHSC (analyzed using NVivo 12).

3.1.1 Hate Speech is Political

Among all the existing issues, the study of hate speech shows a close relationship with politics or matters of a political nature. It can be seen in the many uses of the word political in the studies. In addition, almost all of the articles reviewed, as much as 98.57 percent (n=69 articles), contained words related to politics, indicating the political nature of hate speech (see Figure 4). Looking further, we can see the relation between hate speech and politics in Figure 3, primarily contained in articles 32 (154 words), 69 (125 words), 44 (96 words), 25 (85 words), 48 (78 words), 54 (75 words), 27 (71 words), and 61 (52 words) (as can be observed in Ikeanyibe et al., 2018; Woods, 2019; Trajkova and Neshkovska, 2018; Estellés and Castellví, 2020; and Pettersson, 2019; Sakki and Pettersson, 2018; Frischlich et al., 2021, respectively). That picture seems to strengthen van Dijk’s (2000: 21) view that hate speech is actually not personal but ideological.

It is not too surprising because hate speech, as seen in these articles, is often seen at political moments, especially in presidential elections in a country (Ikeanyibe et al., 2018, Siegel et al., 2019, Wekesa, 2019; Trajkova and Neshkovska, 2018; Syahputra, 2019). Ikeanyibe et al. (2018), for example, review the practice of hate speech in Nigeria’s 2011 and 2015 Presidential Elections. In this case, they view that the use of hate speech can be an attraction for extreme-right populism even though it is seen as outside of existing democratic principles and insensitive to minority rights. Finally, much criticism has been received by politicians who use hate speech both through their social media and from the statements submitted, as shown by Pettersson (2019).

However, political moments or elections are not the only reason hate speech can be viewed as political. It is because hate speech uttered by politicians or important figures can be uttered at any time. Politicians often utter hate speech by using particular rhetoric in other social dynamics, ranging from anti-refugee or immigrant rhetoric (Sakki and Pettersson, 2018; Nortio et al., 2020) to the anti-immigrant rhetoric of The Catholic Church (Chua and Labiste, 2020) and anti-Muslims (Pettersson, 2019). In this case, the study of Sakki and Pettersson (2018) shows how the rhetorical changes developed by parliamentarians during the summer of 2015 and 2016 concerning the refugee crisis and asylum seekers in Finland.

Other evidence that hate speech has a close relationship with the political issue – in a broad and complex sense – can be seen in the collocation table for the node word political, as shown in Table 1.

Table 1.

Top 13 collocation of the node word political with other words in the SHSC (analyzed using LancsBox 6.0: Window=LR5, Filter=None).

Nu.	Index	Collocate	Statistic	Freq (Corpus)
1	8	social	182.0	2793
2	14	discourse	114.0	1746
3	16	media	101.0	3140
4	18	speech	86.0	2878
5	21	hate	71.0	2822
6	27	communication	60.0	763
7	30	parties	48.0	101
8	39	campaign	40.0	230
9	41	containers	39.0	45
10	42	discourse	39.0	471
11	44	right	37.0	421
12	45	rhetoric	36.0	257
13	47	actors	34.0	156

3.1.2 Hate Speech is Discriminatory

The issue of discrimination is another fact that has received much attention from previous researchers when studying hate speech. Although there were 210 keywords related to discrimination, under the keywords Islamophobia, government, and immigration, the issue of discrimination spread to 65.71 percent of the articles (n=46). Despite the frequent use of terms related to discrimination, in our view, the issue of discrimination cannot be separated from other harmful problems. This can be seen from the collocation of the word discrimination with the words containing negative connotations, e. g., violence, hostility, prejudice, and incitement, as shown in Table 2.

Table 2.

Top 12 collocation of the node word discrimination with the words containing negative connotation in the SHSC (analyzed using LancsBox 6.0: Window=LR5, Filter=None).

Nu.	Index	Collocate	Statistic	Freq (Corpus)
1	9	violence	19.0	567
2	11	racism	18.0	713
3	14	hostility	16.0	96
4	16	hate	15.0	2822
5	21	prejudice	13.0	129
6	23	racial	12.0	274
7	25	hatred	11.0	240
8	30	incitement	9.0	47
9	47	aggressive	5.0	113
10	54	exclusion	5.0	56
11	56	intolerance	5.0	48
12	57	islamophobia	5.0	442

Furthermore, the negative issue surrounding discrimination also has a close relation to the group of people who experienced discrimination, as depicted in the collocation and concordance of the node word against with the words related to group of people based on religion, gender, race, ethnicity, and minority (see Table 3 and Table 4).

Table 3.

Top 17 collocation of the node word against with other words associated with a certain group of people in the SHSC (analyzed using LancsBox 6.0: Window=LR5, Filter=None).

Nu.	Index	Collocate	Statistic	Freq (Corpus)
1	15	people	51.0	1146
2	20	muslims	38.0	668
3	21	women	38.0	401
4	23	church	35.0	169
5	27	groups	30.0	608
6	27	ethnic	30.0	284
7	31	media	27.0	3140
8	33	catholic	25.0	60
9	52	racism	18.0	713
10	64	immigrants	16.0	297
11	65	islam	16.0	414
12	66	minorities	16.0	94
13	76	journalists	13.0	403
14	84	migrants	12.0	211
15	83	jews	11.0	171
16	99	government	10.0	347
17	113	refugees	9.0	342

Table 4.

Concordance lines of the node word against with other words based on the use of keyword in context (KWIC) in the SHSC (analyzed using LancsBox 6.0).

Article	Left	Node	Right
Boulahnane (2019)	[. . .] in massive wave of anti-Muslim rethoric	against	people of Muslim faith. Because the [. . .]
Conen-Almagor (2018)	[. . .] a from askrasia, whereby people act	against	their better judgment. Not just those who [. . .]
Allen (2017)	[. . .] the forms, manifestations, and extent of discrimination	against	Muslims in today’s Britain; review the effectiveness [. . .]
Fajri (2019)	[. . .] wake of 9/11, when, despite backlash	against	Muslims being evident (Allen and Nielasen 2002), [. . .]
Chiluwa (2018)	[. . .] stereotypes are deep seated and where the backlash	against	women in the public sphere has been [. . .]
Dragatto, et al. (2020)	[. . .] study, an underestimated from of violence	against	women. This could possibility explain the striking [. . .]
Chua & Labiste (2020)	[. . .] targeted the Roman Catholic Church. His polemic	against	the Church often happened when it was [. . .]
Chua & Labiste (2020)	[. . .] semiotic choice of the speaker, Dutarte’s stand	against	the Church is always communicated before audiences [. . .]
Cervone et al. (2021)	[. . .] religious minorities (e. g. Jews and Muslims) than	against	religious majority groups, e. g. Christians; for an [. . .]
Ikeanyibe et al. (2028)	[. . .] attempt to incite groups or individuals	against	one another. Within this broad categorization, it [. . .]
Bangstad (2015)	[. . .] (Norway too) are understood to incited violence	against,	and called for the ethnic cleansing of [. . .]
Chiluwa et al. (2020)	[. . .] the history of police brutality in Nigeria	against	the Nogeria Delta minority ethnic groups by [. . .]
Choi & Jang (2019)	[. . .] recealed, Korean media coverage is generally biased	against	refugee acceptance (Citizens’ coalition for Democratic Media) [. . .]
Dragotto et al. (2020)	[. . .] and silences in media portrayals of violence	against	women. Women’s Studies International Forum [. . .]
Chua & Labiste (2020)	[. . .] hate speech, speech acts, counter-speech Dutarte’s Polemic	against	the Catholic Church as hate speech in [. . .]
Chua & Labiste (2020)	[. . .] dehumanized the Roman Catholic Church, imputed crimes	against	its leaders, incited discrimination against priests [. . .]
Bangstad (2015)	[. . .] of disappearance of racism and discrimination	against	minorities marked as ‘other’ through various processes
Siapera (2019)	[. . .] to become involved and take a side	against	racism in an explicit and unequivocal manner. [. . .]
Chitakar (2020)	[. . .] Wars. Despite well-known discriminative and prejudiced attitudes	against	Yugoslav immigrants, particularly pronounced at the [. . .]
Musolff (2017)	[. . .] implicitly the discriminating use of parasite imagery	against	immigrants to xenophobic part of the British [. . .]
Petterson (2019)	[. . .] (2014) discusses (populist radical right), politicians’ propaganda	against	terrorism and Islam have profound consequences [. . .]
Poole et al. (2020)	[. . .] that suggest the allegiance of specific countries	(against	Islam). These tactics thus resonate with what [. . .]
Gibson et al. (2020)	[. . .] the extent that minorities can amor themselves	against	insults, via reappropriation or other tools, the [. . .]
Ruziate (2018)	[. . .] aggressive nationalism and ethnocentrism, discrimination and hostility	against	minorities, migrants and people of immigrant origin [. . .]
Kargar & Rauchfleisch (2019)	[. . .] regardless of their marital status. Furthermore, attacks	against	journalists include vulgar and misogynistic references, [. . .]
Shin et al. (2021)	[. . .] as a hub for online public shaming	against	journalists. The members of the networked public [. . .]
Assmakopoulos & Muskat (2017)	[. . .] visible when it comes to xenophobic discourse	against	migrants. What is more, the relative prevalence [. . .]
Sorce (2020)	[. . .] Motivated crime and raise awareness about violence	against	migrants. The last national election in 2017 [. . .]
Allington (2018)	[. . .] much of the racism that is directed	against	some Jews: as we have seen, antisemitism since [. . .]
Cervone et al. (2021)	[. . .] language is more likely to be used	against	religious minorities, e.g. Jews and Muslims than [. . .]
Cohen-Almagor (2018)	[. . .] empirical assumption. Furthermore, government failure to act	against	Hate speech help to normalize or even [. . .]
Ononye & Nwachukwu (2019)	[. . .] of threat by government officials railing	against	‘hate speech’ in response to criticism and [. . .]
Chitakar (2020)	[. . .] underlying the negative and discriminative attitudes	against	refugees and migrants, as reflected in SUD [. . .]
Baider & Kopytowska (2017)	[. . .] justifies pre-emptive attack (hence incitements to violence)	against	refugees seem justifies if not desirable (16) [. . .]

The Table 2 —4 above shows that hate speech cannot be separated from issues of discrimination and other hostile actions against certain groups. Because these issues are very complex, in this article, the issue of the relationship between hate speech and discrimination can at least be mapped into three other issues, namely (1) discrimination based on religion, (2) discrimination based on immigrants and racial groups, and (3) discrimination based on gender. Each of these issues can be further elaborated on below.

3.1.2.1 Discrimination Based on Religious Groups

In previous studies, the relationship between hate speech and religion-based discrimination stands out in studies 5, 6, 12, 18, 51, and 70, as shown in Figure 3, such as the study conducted by Cervone et al. (2021), Chua and Labiste (2020), and Zollo (2017). Cervone et al. (2021), for example, show how insulting language plays a role in perpetuating prejudice and creating a group hierarchy (p. 84), in addition to other impacts, such as social exclusion and repression (p. 88). This study does not focus on discrimination based on religion. However, Cervone et al. (2021) show that in western countries inhabited by Christian majority groups, languages that contain insults are primarily addressed to minority religious groups such as Jews and Muslims (p. 82), exhibiting anti-Semitic and Islamophobic attitudes (Burke et al., 2020).

Religion-based discrimination, reflected in statements containing insults, is also evident in the study conducted by Chua and Labiste (2020). In their review of thirteen statements by President Rodrigo Duterte, Philippines, from 31 August 2016 to 13 May 2017, Chua and Labiste (2020) seek to uncover the use of rhetoric that reflects contempt for the Roman Catholic Church. In this case, Chua and Lebiste apply the analytical framework of “the dangerous speech framework of the U.S.-based Dangerous Speech Project” model of Benesch (p. 9). The situation in the Philippines is exacerbated because no law regulates the prohibition of hate speech in the country.

Regarding religion-based hate speech, the issue of Islamophobia is an integral part of it; it even looks the most prominent compared to hate speech shown to other religions. In other words, in previous studies, the issue of Islamophobia has become one of the central issues. It can be observed that 446 vocabularies express Islamophobia (outside of Islamophobic vocabulary), which is spread over 32 percent of the articles (n=23). The most prominent ones can be seen in studies conducted by Allen (2017), Kastolani (2020), Vidgen and Yasseri (2020), Burke et al. (2020), Sharifi et al. (2017), and also Poole et al. (2020). The viewpoints of their studies are also different, starting from the construction of Islamophobia until how to detect the level of strength and weakness of Islamophobic speech, including the political approach that needs to take to deal with cases of Islamophobia. What should be noted is that the issue of Islamophobia is not only an issue in western countries but has also spread to Indonesia, a country with the largest Muslim population in the world, as shown by Kastolani (2020).

There are certain stereotypes attached to Muslim groups, which is why the issue of Islamophobia has become increasingly popular in recent times. Since the incident known as “9/11” with President George Bush’s War on Terror (WOT) announcement, the image of Muslims has been closely linked to terrorism. That is why the study of hate speech also targets the issue of terrorism, as can be observed in the studies of Fajri (2019), Heath-Kelly and Jarvis (2017), Allen (2017), Sharifi et al. (2017), Poole et al. (2020), Kaján (2017), (Sponholz, 2016), and so on. In his study, Fajri (2019) reveals the dominant discourse around Muslims in Indonesia, which is associated with terrorism, fundamentalism, conflict, and violence, constructed by American newspapers. At the same time, this study also shows a shift in the construction of Muslim groups in the past 15 years, from terrorism and extremism to radicalism. The discourse of terrorism in Anglo-American popular culture attached to Muslims does not only come through construction in the mass media but also in other social practices, such as jokes, street art, films, memorial projects, elite rhetoric, and harassment scandals. In the study of Heath-Kelly and Jarvis (2017), the stereotypes are seen as a form of knowledge encouragement, embodied in three concepts: laughter, lamentation, and hatred. It can be well explained by applying Michel Foucault’s critical study framework.

3.1.2.2 Immigrant-Based Discrimination and Racism

There are various reasons why some countries, especially European countries, are often the destination of migration or refugees, among them because of crises or wars that often occur in several countries in the Middle East and Sub-Saharan Africa (Kaján, 2017). Regardless of the situation, immigrants are a group that is often the target of hate speech and discrimination. It may be because immigrant groups within a country are seen as a threat to the local population, as in Solvenia (Chitrakar, 2020) and several other countries. In the articles reviewed in this research, it can be seen that there are 361 immigration vocabularies (excluding related vocabularies such as immigrant, asylum seekers, and refugees). Vocabulary related to immigrants is spread across 52.86 percent of articles (n=37), including articles by Assimakopoulos and Muscat (2017), Merrill and Åkerlund (2018), Bangstad (2015), Pettersson (2019), López (2020), Musolff (2017), Sponholz (2016).

Hate speech, grouped by Cervone et al. (2021) into the term “derogatory language”, is often uttered to immigrant groups or different groups in different countries. The terms used also differ from one another. Some use the term criminal for refugees or immigrants in Sweden and the UK (Merrill and Åkerlund, 2018; Chovanec, 2021); others use the term the new Nazis for Muslim groups in the UK (Burke et al., 2020). In several European countries, such as Cyprus and Poland, immigrant groups are conceptualized as dirty, immoral, barbaric, animal, invasions, and terrorist people (Baider and Kopytowska, 2017: 214). Refugee groups are also often seen as a threat, as is the case in Slovenia (Chitrakar, 2020).

Regarding discrimination and hate speech against immigrants, racial issues become inseparable in this context. In this review, the number of vocabularies related to race was not chosen to be one of the keywords in this literature review because its occurrence was not significant. However, hate speech related to immigrants or asylum seekers directly or indirectly relates to racial issues. It, for example, can be seen in a study by Merrill and Åkerlund (2018). In this case, they show that the digital platform Facebook is seen as playing a role in creating overt and covert racism against immigrant groups. Another thing can also be seen in López’s study (2020, see also studies by Siapera, 2019 and Bangstad, 2015). In his article, López shows how television coverage of the Development, Relief, and Education for Alien Minors (DREAM) Act relates to undocumented immigrant groups in America. In the news, immigrants are portrayed, for one, in an inhumane way.

3.1.2.3 Gender-Based Discrimination

Discriminatory conduct in the form of hate speech is also often experienced by gender-based vulnerable groups. In the articles reviewed, 214 transgender vocabularies or utterances refer to a specific gender. It can be seen that the keyword transgender is only discussed in 14.29 percent of the articles (n=10). However, various variations in vocabulary referring to gender or transgender are spread across several other articles, such as homosexual, heterosexual, homophobia, lesbian, transgender, transexual, transgender, LGBT, and gay. This means that hate speech and gender-based discrimination are concerns for previous researchers.

As can be observed in Figure 3, studies of hate speech related to discrimination against gender groups can be seen in the article of Baider (2018), Robinson and Spivey (2019), Ruzaite (2018), Readyera (2019), and Trindade (2020). This topic is most prominently discussed in an article by Robinson and Spivey (2019). In this case, Robinson and Spivey examine a phenomenon of transgenderism discourse carried out by ex-gay groups or movements in America since the 1970s, which they consider dangerous because their groups are seen as a source of cisgenderism and transmisogyny, which construct gender variants as a form of sin, mental illness, and danger. Close to this topic are studies conducted by Baider (2018) and Ruzaite (2018). Baider examines how hate speech against LGBT groups is in the form of construction or framing in the Republic of Cyprus from April 2015 to February 2016, while the focus of the Ruzaite study was in Lithuania.

What needs attention in the case of gender-based hate speech is the findings shown by Cleland et al. (2018). They analyzed 5,128 comments from 35 online message boards of prominent football fans from the 1980s and 1990s and 978 comments online in response to a story in the Guardian regarding the decision of former German international footballer Thomas Hitzlsperger to declare himself as gay in January 2014. Their findings show that out of 6,106 comments, only 2 percent of comments contain “damaging homophobic intent,” and 98 comments describe no longer constructing their masculinity by opposing homosexual groups with homophobic words or language.

3.1.3 Social Actors behind Hate Speech

It is almost impossible to talk about hate speech without talking about the actors behind it. The use of hate speech in the political process, democracy, or other contexts, cannot be separated from the actors involved, especially political actors or other prominent public figures. It can be shown in a study by Vidgen and Yasseri (2020). In this case, Vidgen and Yasseri (2020) collected 109,488 tweets produced by ultra-right Twitter accounts in the UK to see the extremes of the utterances on Twitter. A similar study can be seen in the article written by Trajkova and Neshkovska (2018), which discusses the practice of hate speech by politicians through Facebook and Twitter in the election process in Macedonia. This means that politicians and other public figures play an important role in producing and reproducing hate speech for various political purposes, including manipulating public opinion, whether on an extreme scale or not.

The actors behind hate speech are very varied (as can be inferred from Table 5) and, in many cases, involve high-profile people. Hate speech can even involve high-position officials of a country, ranging from the president (Chua and Labiste, 2020) and members of parliament (Sakki and Pettersson, 2018). In addition, social actors also come from separatist groups (Chiluwa, 2018), social media (Siapera and Viejo-Otero, 2021; Fajri, 2019; Merrill and Åkerlund, 2018), and political parties (Burke et al., 2020). Some actors come from religious groups, as shown by Burke et al. (2020). They show the role of ultra-right groups in exploiting Jewish supporters in dealing with Muslim groups in the Charlie Hebdo attack case a few years ago (p. 377). Sometimes, actors producing hate speech on social media hide behind anonymous names (Wekesa, 2019). Furthermore, the role of social actors, especially among politicians, concerning hate speech is also seen in articles 1, 13, 32, 43, 44, 48, and 68, researched by Åkerlund (2021), Chiluwa (2018), Musolff (2017), Ikeanyibe et al. (2018), Trajkova and Neshkovska (2018), Pettersson (2019), and Wekesa (2019) respectively.

Table 5.

Top 12 collocation of the node word actors in the SHSC (analyzed using LancsBox 6.0: Window=LR5, Filter=None).

Nu.	Index	Collocate	Statistic	Freq (Corpus)
1	6	political	34.0	1790
2	9	social	23.0	2793
3	15	individual	10.0	202
4	16	media	10.0	3140
5	17	different	9.0	507
6	22	state	9.0	387
7	24	collective	8.0	96
8	30	public	7.0	1076
9	31	right-wing	7.0	193
10	38	discursive	5.0	413
11	39	groups	5.0	608
12	41	institutions	5.0	60

Concerning social actors, as mentioned above, it is slightly difficult to distinguish between the government and politicians in the study of hate speech. However, apart from the negative attitude of the politicians in it, it appears that the government, which can be observed in previous studies, is positioned more as an agent that is expected to overcome problems related to hate speech. In this sense, it is expected to take both a policy approach and a counter-discourse approach to hate speech (Chiluwa, 2018, see also Allen, 2017, Đorđević, 2020). In the study of Chiluwa (2018), the Nigerian government, together with Igbo politicians – an ethnic group from South-Central and Southeastern Nigeria – attempted to counter-discourse against hate speech by a group known as the Indigenous People of Biafra (IPOB).

Therefore, social actors about hate speech do not only have negative connotations, but can also be interpreted positively. That is, behind the existence of social actors who produce hate speech, some actors are suspected of having a role in solving the problem of hate speech, as seen in the IPOB case above. Another case, for example, can be seen in the efforts made by Baroness Sayeeda Warsi, a British politician of Pakistani descent, who regards the need to overcome Islamophobia as one of the main agendas of the coalition (Allen, 2017). Even though Allen’s study shows that Baroness Sayeeda Warsi’s efforts are seen as failing, positive efforts to tackle hate speech have also emerged from politicians and the government.

So far, previous studies have shown that hate speech has a close relationship with a country’s complex sociopolitical and democratic dynamics. Apart from using hate speech in political campaigns in the dynamics of democracy in various countries, some social actors are trying to counter the hate speech.

3.2 Trends in Using Method of Hate Speech Studies

Methodologically, there are two general approaches used by previous researchers to examine hate speech, namely (1) using a mixed-method and (2) a qualitative approach (see Figure 5). No hate speech research was found from 70 articles systematically reviewed only using a quantitative approach. It shows that previous researchers saw hate speech as a social phenomenon that must be described, interpreted, or explained. Between the two perspectives, the qualitative approach (only) occupies a dominant position regarding hate speech studies, reaching 57.14 percent (n=40) of articles. In contrast, articles that studied hate speech using a combined perspective reached 42.86 percent (n =30). This means that the mixed method is still less popular than a qualitative approach (alone), but the comparison is almost balanced.

Figure 5.

Graph of trends in the use of research methods based on general approaches in the last five years in SHSC.

Another note is that, although fewer hate speech researchers use combined methods than qualitative methods (only), 37.14 percent (n=26) of researchers use corpus-based combined methods. The rest, 4.29 percent (n=3) and 1.43 percent (n=1), still rely on surveys and interviews. It is not surprising because by using the corpus approach, researchers can efficiently perform mining and analysis of data (datasets) that are abundant in (social) media or newspapers related to hate speech. Moreover, so far, there are many variations of applications or tools available—paid or free. It seems to encourage many researchers interested in using a corpus-based approach.

In contrast to the combination approach, the qualitative approach, which does not base the strength of analysis on calculating the number of utterances, is more varied. The various levels of variation are reflected in the percentage of the approach, namely 45.71 percent (n=32) of articles with varying approaches, excluding interview, ethnographic, and survey approaches. In this case, the interview-based qualitative approach was still dominant out of all existing approaches, although it only reached 8.57 percent (n=6). Meanwhile, the use of ethnography and observation was only 1.43 (n=1) percent, respectively.

Since this review is limited to the scope of discourse analysis or study, the previous research in this review can be further divided into the use of more specific methods. In previous studies, it appears that there are two trends in the use of almost balanced methods, namely the application of critical-based discourse analysis methods and non-critical discourse analysis (see Figure 6). The use of critical discourse analysis reached 45.71 percent (n=32), while the use of (non-critical) discourse analysis reached 54.29 percent (n=38), an almost balanced portion. In addition, it is seen that in non-critical discourse analysis, the use of the methods is more varied than in critical discourse analysis methods. It may be because, in the critical discourse analysis method, only a few “Great Masters” or variations can be referred to; or previous researchers may have assumed that all CDA research has the same model. In contrast, in other types of discourse analysis, researchers can refer to several variations of the theoretical-methodological framework model—starting from the old model, such as conversational analysis, natural semantics, FGD, to the application of the latest models, such as the application of topic-modeling and social-network analysis methods.

Figure 6.

Graph of trends in the use of research methods based on critical and non-critical approaches in the last five years in SHSC.

4. Some Methodological Issues and Critical Remarks

4.1 Critical Dimension Issues in the Corpus-Based Studies

Looking back at essential topics in the study of hate speech discourse, as discussed previously, which show the dimensions of social life that are so complex, we regard the emphasis on critical aspects needs to get a more considerable portion. If hate speech is political, discriminatory, and other forms of unfair social practice, the application of a CDA approach in the study of hate speech should be further improved and applied thoughtfully. As is well known, the CDA approach is not merely aimed at uncovering a phenomenon but can also act as a movement for change.

In the previous articles, as shown in Figure 6, the use of CDA to uncover the phenomenon of hate speech has received an adequate portion from previous researchers. However, the first aspect to note is that among 45.71 percent (n=32) of articles using CDA, 17.14 percent (n=12) do not explicitly state which methodological framework of the CDA model to apply. It certainly has a significant impact on the use of the method. It is because, although sharing the same view that discourse is a social practice in which language is used to dominate others, each expert of CDA has a different emphasis from one another. Another consequence of not mentioning the methodological (critical) framework model used in the research process causes researchers to tend to get caught up in the “error”. That is why van Dijk often mentions that the CDA is not a ready-used method but rather an interdisciplinary research movement (see also Weiss and Wodak, 2003: 1). Therefore, without explicitly mentioning which CDA approach is to be applied in conducting discourse studies, it is not easy to get a clear picture of what was the focus of attention of these previous researchers.

Another aspect to be considered in previous studies is applying corpus-based critical studies. As seen in Figure 7, corpus linguistics has become one of the trends combined with a critical approach. The Figure 7 shows that 42.4 percent (n=14) of the critical approach combined with the corpus approach, while the critical approach that is not combined with the corpus approach is 57.6 percent (n=19). On the one hand, combining CDA with the corpus can help CDA gain a deeper understanding of analyzing large amounts of data (corpus). However, the weakness is that researchers who apply corpus-based CDA rarely explore the central issue in CDA, namely the relationship between language and power, including hidden issues behind it. It seems to be because the researcher only relies on the corpus data as a basis for interpretation and does not conduct an in-depth exploration of the meaning relations that arise from the corpus of data with the sociocultural or sociopolitical context that exists amid society in a country. Thus, corpus-based analysis tends to carry out analysis only completed at the level of text analysis, especially linguistic analysis, and is therefore not critical at all in applying the CDA approach itself, as Talib and Fitzgerald (2018) have worried. According to them, the impact of the data-based [corpus] approach is the lack of a multidisciplinary aspect of analytical tools (p. 124).

Figure 7.

Graph of comparison regarding percentage of corpus-based critical approaches with those that are not in the last five years in SHSC.

In previous research, for instance, we can see such a tendency in the studies of Fajri (2019), Boulahnane (2019), Đorđević (2019), Merrill and Åkerlund (2018), and Musolff (2017). The researchers were compelling in demonstrating the analysis of texts based on the corpus of data but seemed weak in linking the analysis of the texts with the sociocultural analysis. For example, Fajri (2019) stopped at the question of how the American media constructed an Islamic group but failed to show why this could happen. We can see the same in Boulahnane’s (2019) research. He was able to show the ideological side of discourse related to Islamophobia in the shooting case in Orlando, Florida, on June 12, 2016, but was unable to show what underlies the ideological aspect.

4.2 Lack of Exploration of Sociocognitive Aspects

Another weakness is that the CDA applied to hate speech in previous research did not pay much attention to the sociocognitive aspect of CDA (see Figure 8), while hate speech itself is closely related to cognitive factors, such as ideology, knowledge, opinion, values, attitudes, and beliefs. If this is the case, hate speech will remain like an iceberg phenomenon, as van Dijk has often described concerning racist speeches in Europe and America. It means that linguistic-forensically, hate speech, in a legal way, is relatively easier to tackle, but the root of the problem will be difficult to uncover. Thus, socio-political issues related to hate speech will be challenging to resolve, so hate speech itself will continue to occur and become more extreme.

Figure 8.

Visualization of the application of van Dijk’s sociocognitive model-CDA in the previous articles (analyzed and visualized using NVivo 12).

The lack of cognitive exploration of CDA in previous research seems to confirm the view of Chilton (2005) that there are severe implications behind the lack of cognitive aspects in the use of CDA, namely that research that reveals the cognitive dimensions (such as ideology) of human life will be less compared other aspects in the research process. The lack of a cognitive approach to CDA on the one hand and the sheer volume of research in cognitive science (outside of CDA) on the other has led Chilton to question whether we really need CDA or not (2005: 22).

In the previous study, 14.29 percent (n=10) of articles based their studies on van Dijk’s CDA model or the sociocognitive model itself. However, further observation (see Figure 8) shows that of the 10 articles that mention using the van Dijk model approach, only 2.85% (n=2) of all articles explicitly mention the term sociocognitive. It means that the understanding of van Dijk’s sociocognitive concept is still poor or at least not appropriately applied. Data reinforce this assumption that the concept of “cognition” as a critical term and concept of the sociocognitive approach is only used in 5.71% (n=4) articles. In fact, of the four articles, the use of the term cognition more than three times in the articles was only 1.4% (n=1), meaning that the concept of cognition in sociocognitive CDA theory is seen as insignificant.

The fact above shows that previous researchers did not explore in depth the particular emphasis on the sociocognitive model of CDA. On the other hand, as is widely known and understood, the sociocognitive-model CDA approach emphasizes its study concerning aspects of cognition. This means that although the previous articles on hate speech frequently mention the social dimension in political discursive practices and discrimination, it is not seen as having a close relationship with personal or social cognition. If these aspects are ignored, it can also be assumed that researchers who claim to use van Dijk’s approach have failed to apply the sociocognitive approach itself. Meanwhile, on the one hand, van Dijk (2000) himself strongly emphasizes that:

“[i]gnoring such cognitive dimensions of ideologies, and merely analyzing them in terms of social practices, social formations, or social structures, provides an incomplete insight into ideologies, and constitutes as an improper reduction of complex social phenomena, and hence an inadequate theory” (p. 126).

In other words, if a politician utters a racist hate speech, it should be suspected that the utterance arises from the ideology of racism itself. Thus, the theoretical concept used in the sociocognitive approach can not only explain the meaning that the utterance is racist or not but also why someone utters a racist utterance and the social structure that underlies the utterance. However, that is the part that previous researchers have poorly explored.

Conclusion

The main topics of hate speech studies in previous research show that hate speech has a strong relationship with politics, discrimination, and notable actors involved. These topics are distributed almost evenly in each article reviewed. If we regard politics and discrimination as closely related to power relations, domination, and ideology, then the CDA approach can be seen as very appropriate in portraying the phenomenon of hate speech. In this case, attention to the application of CDA theory has received a fairly adequate portion in hate speech research, which is 45.71 percent (n=32).

Furthermore, if we refer to sociocognitive CDA theory, hate speech is also a cognitive phenomenon related to personal and social cognitions such as knowledge, attitudes, beliefs, and ideology. Therefore, the approach that can be used to uncover these aspects is applying sociocognitive CDA. In previous studies, the use of sociocognitive-model CDA also seems pretty significant, 14.29 percent (n=10). However, applying the sociocognitive CDA carried out by previous researchers is still far from the fundamental view of the theory. In other words, the cognitive aspects and mental models emphasized in the theory are not given much attention. Therefore, there is much space for future researchers to take sociocognitive CDA thoughtfully in researching hate speech. Thus, the phenomenon of hate speech can be described and comprehended more clearly.

Footnotes

Acknowledgements

We want to thank van Dijk, editor of the “Discourse & Society” journal, who humbly advised us to seek a corpus linguistics (CL) expert to ensure the corpus method we used was appropriate. We also thank Prihantoro, an expert on CL at the Faculty of Humanities, Universitas Diponegoro (Indonesia), for generously taking the time to examine, correct, and provide excellent suggestions regarding some of the shortcomings in the corpus dimensions of our article. In addition, we would also like to thank our colleagues at the Department of Languages and Arts, Faculty of Teacher Training, Universitas Mataram (Indonesia), Ahmad Junaidi and Santi Farmasari. They voluntarily checked the English of our draft article during their busy activities.

Author’s note

Ahmad Sirulhaq is also affiliated with Universitas Mataram, Indonesia.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors received financial support for the research from Indonesia Endowment Funds for Education (LPDP), Indonesian government.

Author biographies

Ahmad Sirulhaq is a doctoral student in the Linguistics Department, Faculty of Humanities, Universitas Indonesia. He is also a lecturer at the Department of Languages and Arts, Faculty of Teacher Training and Education, Universitas Mataram (Indonesia). He has research interests in linguistics, discourse studies, Indonesian language, education, journalism, media, and ideology.

Untung Yuwono is a lecturer at the Linguistics Department, Faculty of Humanities, Universitas Indonesia (Indonesia). He has research interests in linguistics, mainly in discourse studies, syntax, morphology, Indonesian language for foreigners (BIPA), and forensic linguistics. His works are widely distributed in several journals, both national and international. He has also written several books.

Abdul Muta’ali is a lecturer at the Linguistics Department, Faculty of Humanities, Universitas Indonesia (Indonesia). His research interests include linguistics, pragmatics, sociolinguistics, international relations, and transnational crime. To date, he has written at least nine books.

References

Åkerlund

(2021). Dog whistling far-right code words: the case of ‘culture enricher’’ on the Swedish web.’ Information Communication and Society, 0(0): 1–18.

Allen

(2017). Political approaches to tackling Islamophobia: An “insider/outsider” analysis of the british coalition government’s approach between 2010-15. Social Sciences, 6(3): 1—19

Allington

(2018). ‘Hitler had a valid argument against some Jews’: Repertoires for the denial of antisemitism in Facebook discussion of a survey of attitudes to Jews and Israel. Discourse, Context and Media, 24: 129–136.

Assimakopoulos

Muskat

(2017). Exploring xenophobic and homophobic attitudes in Malta: Linking the perception of social practice with textual analysis. Lodz Papers in Pragmatics, 13(2): 179–202.

Baider

(2018). “Go to hell fucking faggots, may You die!” framing the LGBT subject in online comments. Lodz Papers in Pragmatics, 14(1): 69–92.

Baider

Kopytowska

(2017). Conceptualising the Other: Online discourses on the current refugee crisis in Cyprus and in Poland. Lodz Papers in Pragmatics, 13(2): 203–233.

Bangstad

(2015). The racism that dares not speak its name: Rethinking neo-nationalism and neo-racism. Intersections East European Journal of Society and Politics, 1(1): 49–65.

Boulahnane

(2019). Homophobia and ‘un-Americanness’ as rising facets of islamophobia: An analysis of Orlando shooting media transcripts. Journal of Al-Tamaddun, 14(2): 143–152.

Brezina

(2018). Statistics in corpus linguistics: A practical guide. Cambridge University Press.

10.

Burke

Diba

Antonopoulos

(2020). ‘You sick, twisted messes’: The use of argument and reasoning in Islamophobic and anti-Semitic discussions on Facebook. Discourse and Society, 31(4): 374–389.

11.

Castaño-Pulgarín

Suárez-Betancur

Vega

LMT

López

HMH

(2021). Internet, social media and online hate speech. Systematic review. Aggression and Violent Behavior, 101608. 1—7

12.

Cervone

Augoustinos

Maass

(2021). The Language of Derogation and Hate: Functions, Consequences, and Reappropriation. Journal of Language and Social Psychology, 40(1), 80–101.

13.

Chetty

Alathur

(2018). Hate speech review in the context of online social networks. Aggression and Violent Behavior. 40: 108-118.

14.

Chilton

(2005). Missing Links in Mainstream CDA: Modules, Blends and the Critical Discourse Analysis. In: Wodak

Chilton (eds) A new agenda in (critical) discourse analysis: theory, methodology and interdisciplinarity. Amsterdam: John Benjamins Publishing.

15.

Chiluwa

(2018). A nation divided against itself: Biafra and the conflicting online protest discourses. Discourse and Communication, 12(4): 357–381.

16.

Chiluwa

Taiwo

Ajiboye

(2020). Hate speech and political media discourse in Nigeria: The case of the indigenous people of Biafra. International Journal of Media and Cultural Politics, 16(2): 191–212.

17.

Chitrakar

(2020). Threat perception in online anti-migrant speech: A slovene case study. Slovenscina 2.0, 8(1): 66–91.

18.

Choi

Jang

(2019). An online opinion analysis on refugee acceptance using topic modeling. Asian Journal for Public Opinion Research, 7(3): 169–198.

19.

Chovanec

(2021). ‘Re-educating the Roma? You must be joking..’: Racism and prejudice in online discussion forums. Discourse and Society, 32(2): 156–174.

20.

Chua

Labiste

(2020). Duterte’s Polemic Against the Catholic Church as Hate Speech. Plaridel, 17(1): 1–33.

21.

Cleland

Magrath

Kian

(2018). The Internet as a Site of Decreasing Cultural Homophobia in Association Football: An Online Response by Fans to the Coming Out of Thomas Hitzlsperger. Men and Masculinities, 21(1): 91–111.

22.

Cohen-Almagor

(2018). Taking North American White Supremacist Groups Seriously: The Scope and Challenge of Hate Speech on the Internet. International Journal for Crime, Justice and Social Democracy. 7(2): 38-57.

23.

Đorđević

(2019). The sociocognitive dimension of hate speech in readers’ comments on Serbian news websites. Discourse, Context and Media. 33(2020): 1—9.

24.

Dragotto

Giomi

Melchiorre

(2020). Putting women back in their place. Reflections on slut-shaming, the case Asia Argento and Twitter in Italy. International Review of Sociology, 30(1): 46–70.

25.

Estellés

Castellví

(2020). The educational implications of populism, emotions and digital hate speech: A dialogue with scholars from Canada, Chile, Spain, the UK, and the US. Sustainability (Switzerland), 12(15). 6034.

26.

Fajri

MSA

(2019). The discursive portrayals of Indonesian muslims and Islam in the American press: A corpus-assisted discourse analysis. Indonesian Journal of Applied Linguistics, 9: 167–176. doi: 10.17509/ijal.v9i1.15106

27.

Frischlich

Schatto-Eckrodt

Boberg

Wintterlin

(2021). Roots of incivility: How personality, media use, and online experiences shape uncivil participation. Media and Communication, 9(1): 195–208.

28.

Gibson

Epstein

Magarian

(2020). Taming Uncivil Discourse. Political Psychology, 41(2): 383–401.

29.

Heath-Kelly

Jarvis

(2017). Affecting terrorism: Laughter, lamentation, and detestation as drives to terrorism knowledge. International Political Sociology, 11(3): 239–256.

30.

Ikeanyibe

Ezeibe

Mbah

Nwangwu

(2018). Political campaign and democratisation. Journal of Language and Politics, 17(1): 92–117.

31.

Kaján

(2017). Hate online: Anti-immigration rhetoric in Darknet. Nordia Geographical Publications, 46(3): 3–22.

32.

Kargar

Rauchfleisch

(2019). State-aligned trolling in Iran and the double-edged affordances of Instagram. New Media and Society, 21(7): 1506–1527.

33.

Kastolani. (2020). Understanding The Delivery of Islamophobic Hate Speech Via Social Media In Indonesia. Indonesian Journal of Islam and Muslim Societies, 10(2): 247–270.

34.

López

(2020). Education policy analysis archives Discursive De / Humanizing : A Multimodal Critical. Education Policy Analysis Archive, 28(47), 1–26.

35.

Matamoros-Fernández

Farkas

(2021). Racism, Hate Speech, and Social Media: A Systematic Review and Critique. Television and New Media, 22(2): 205–224.

36.

Merrill

Åkerlund

(2018). Standing up for Sweden? The racist discourses, architectures and affordances of an anti-immigration facebook group. Journal of Computer-Mediated Communication, 23(6): 332–353.

37.

Musolff

(2017). Dehumanizing metaphors in UK immigrant debates in press and online media. Journal of Language Aggression and Conflict, 3(1): 41-56.

38.

Nortio

Niska

Renvik

Jasinskaja-Lahti

(2020). ‘The nightmare of multiculturalism’: Interpreting and deploying anti-immigration rhetoric in social media. New Media and Society. 23(3): 438-456.

39.

Ononye

Nwachukwu

(2019). Metalinguistic evaluators and pragmatic strategies in selected hate-inducing speeches in Nigeria. Indonesian Journal of Applied Linguistics, 9(1): 48–57.

40.

Paz

Montero-Díaz

Moreno-Delgado

(2020). Hate Speech: A Systematized Review. SAGE Open. 10(4): 1—12.

41.

Pettersson

(2019). “Freedom of speech requires actions”: Exploring the discourse of politicians convicted of hate-speech against Muslims. European Journal of Social Psychology. 49(5): 938-952.

42.

Poole

Giraud

de Quincey

(2020). Tactical interventions in online hate speech: The case of #stopIslam. New Media and Society. 23(6), 1415-1442.

43.

Prihantoro

(2015). Semantic Prosody of Words of Effects in Indonesian. Indonesian Journal of Applied Linguistics, 5(1), 106-115.

44.

Robinson

CM.

Spivey

(2019). Ungodly genders: Deconstructing Ex-Gay movement discourses of “Transgenderism” in the US. Social Sciences, 8(6): 1–28.

45.

Ruzaite

(2018). In search of hate speech in Lithuanian public discourse: A corpus-assisted analysis of online comments. Lodz Papers in Pragmatics, 14(1): 93-116.

46.

Sakki

Pettersson

(2018). Managing stake and accountability in Prime Ministers’ accounts of the “refugee crisis”: A longitudinal analysis. Journal of Community and Applied Social Psychology, 28(6): 406–429.

47.

Sharifi

Ansari

Asadollahzadeh

(2017). A critical discourse analytic approach to discursive construction of Islam in Western talk shows: The case of CNN talk shows. International Communication Gazette, 79(1): 45–63.

48.

Shin

Kim

Joo

(2021). Hating journalism: Anti-press discourse and negative emotions toward journalism in Korea. Journalism. 22(5): 1239-1255.

49.

Siapera

(2019). Organised and ambient digital racism: Multidirectional flows in the Irish digital sphere. Open Library of Humanities, 5(1): 0–34.

50.

Siapera

Viejo-Otero

(2021). Governing Hate: Facebook and Digital Racism. Television and New Media, 22(2), 112–130.

51.

Siegel

Nikitin

Barberá

Sterling

Pullen

Bonneau

Tucker

(2019). Trumping hate on twitter? online hate speech and white nationalist rhetoric in the 2016 us election campaign and its aftermath Quarterly Journal of Political Science, 16(1): 71-104.

52.

Sorce

(2020). Journalist-Activist Boundary Work in Populist Times: The #NazisRaus Debate in German Media. Journalism Practice, 0(0): 1–17.

53.

Sponholz

(2016). Islamophobic hate speech: What is the point of counter-speech? the case of oriana fallaci and the rage and the pride. Journal of Muslim Minority Affairs, 36(4), 502–522.

54.

Syahputra

(2019). Expressions of hatred and the formation of spiral of anxiety on social media in Indonesia. SEARCH (Malaysia), 11(1): 95–112.

55.

Talib

Fitzgerald

(2018). Putting philosophy back to work in critical discourse analysis. Critical Discourse Studies, 15(2), 123-139.

56.

Trajkova

Neshkovska

(2018). Online hate propaganda during election period: The case of Macedonia. Lodz Papers in Pragmatics, 14(2), 309-334.

57.

Trindade

LVP

(2020). Disparagement humour and gendered racism on social media in Brazil. Ethnic and Racial Studies, 43(15): 2766–2784.

58.

van Dijk

(2000). Ideology: A multidisciplinary approach. London: Sage.

59.

van Dijk

(2006). Ideology and discourse analysis. Journal of political ideologies, 11(2), 115-140.

60.

van Dijk

. (2008). Discourse and Context: A Sociocognitive Approach. Cambridge University Press.

61.

van Dijk

(2014). Discourse and knowledge: A sociocognitive approach. Cambridge University Press.

62.

van Dijk

(2020). Closing Remarks 2020 International Conference on Hate Speech and Discrimination. Available at https://www.youtube.com/watch?v=Ne8d0oRgWjI;https://discourseanalysis.net/en/international-conference-hate-speech-and-discrimination?language=en. Accessed in 25 March 2021, on 09.45 am.

63.

van Dijk

(2021). “Identifying and Countering Holocaust Distortion: Lessons for and from Southeast Asia,” held by International Holocaust Remembrance incollaboration with American University of Phnom Penh, 23–26 November 2021.

64.

Vidgen

Yasseri

(2020). Detecting weak and strong Islamophobic hate speech on social media. Journal of Information Technology and Politics. 17(1): 66-78.

65.

Weiss

Wodak

(2003). Introduction: Theory, interdisciplinarity and critical discourse analysis. In Critical discourse analysis: Theory and Interdisciplinarity. Weiss Wodak (eds.) (pp. 1-32). Palgrave Macmillan, London.

66.

Wekesa

(2019). Hate online: The creation of the “Other”. Lodz Papers in Pragmatics, 15(2): 183-208.

67.

Woods

(2019). Mobilising Dissent in a Digital Age: The Curious Case of Amos Yee. Geopolitics, 0(0): 1–22.

68.

Zollo

(2017). No hate speech movement: Evolving genres and discourses in the European Online Campaign to fight discrimination and racism. Observatorio, 11(2): 91–106.