Media framing of political protests – reporting bias and the discrediting of political activism

ABSTRACT

Recently several European countries shifted to illiberalism and made attempts to dominate the media and political narratives. The question we raise is how media populism in Hungary contributes to the buttressing of the regime by discrediting protests. We offer a four-level media analysis. First, we ask whether the pro-government media is characterized by so-called selection bias. Second, we focus on framingbias relying on ideas presented by the protest paradigm. Third, we highlight the tone of disdain that characterizes numerous pro-governmental utterances. Finally, we point out the significance of iconic frames. Contrary to our expectations, we found no selection bias, but there was a clear framing bias in pro-governmental media, which was made harsher by the derogatory tone of pro-governmental media and the dog-whistling produced by iconic frames. By identifying how media populism operates, our aim is to offer a way to grasp democratic backsliding by concentrating on the media.

KEYWORDS:

• protest paradigm
• selection bias
• description bias
• informational autocracy
• media populism

Introduction

Researchers have started to scrutinize how formal institutions on the one hand, or social norms and beliefs in non-democratic regimes on the other, influence the political actions of ordinary citizens and the political culture of polities in general (Gehlbach 2018; Neundorf and Pop-Eleches 2020). The literature on institutions analyzes how the manipulation of the electoral system, legislature, judiciary, or the media can lead to the decay of democracy. Scholars have also examined support for regimes, populist attitudes, anti-immigrant attitudes, or political polarization in order to understand why citizens accept or even cherish illiberal politicians (Ekman 2009; Krekó and Enyedi 2018; Guriev and Treisman 2020b; Zagrebina 2020; Akboga and Sahin 2021). According to informational autocracy theory (Guriev and Treisman 2019, 2020a; Gehlbach and Sonin 2014), the media is the most important link between macro-level political institutions and micro-level political attitudes and behavior. Centralized and controlled media might manipulate beliefs about a political system’s performance and bring about positive attitudes toward the regime (Guriev and Treisman 2019).

Scholars have thoroughly analyzed different functions of the Hungarian media. They have pointed out that since 2010 the media has been reformed to serve as a conduit for pro-government messaging and communication (Bajomi-Lázár 2013; Polyák 2019). Namely, the state uses multiple tools such as centralization of institutions (Polyák 2019), taking control of local media (Hargitai 2021), or state advertisements (Bátorfy and Urbán 2020) to support the emerging pro-government media empire. However, it remains to be determined whether these changes in the Hungarian illiberal media sector bring about politically biased media content. We aim to contribute to this literature by elucidating how government-leaning media marginalizes political protests and protesters and how media populism works in Hungary. Our central questions are whether the pro-government media reports on protests and demonstrations, and how it interprets both the protest events and the protesters. This is crucial in the light of a recent comparative study on pro-government and government-critical demonstrations in Hungary. Researchers found that consuming government-friendly media positively predicts participation in pro-government rallies. However, watching government-critical media did not correlate significantly with participation in government-critical protests (Susánszky, Kopper, and Tóth 2016). Thus, the differently framed protest events might have a causal effect on political protest participation. In order to scrutinize this issue, we conduct our empirical analyses of leading Hungarian online media news portals in four steps.

First, in our study of the media we have focused on selection bias, i.e. whether the share of reporting on protests in pro-government media is significantly lower than in the critical media. Studying this is important, as Viktor Orbán’s regime frequently claims that there is no bias in the media scene, as there are critical journals in Hungary and that the media does indeed report on protest events.¹

At a second level we analyzed the discrediting tone of reporting. In order to capture this, the second part of our study focused on how reporting of protest events differs between pro-governmental and government-critical media. Our study here is focused on one specific protest event, by students of the University of Theater (more on this and the reason for the case selection later). Our analysis confirmed our hypothesis that the government-leaning media intensively use technics that are described in the protest paradigm (e.g. McLeod and Hertog 1999) – i.e. a wide range of devices aimed at the marginalization and demonization of protesters and their grievances.

Coming this far in our analysis, we realized that we needed two additional steps to capture the way populist media works in Hungary. Thus, as a third level of our analysis we focused on the general degrading tone that the pro-government media outlets employ, which is not merely aggressive but also conveys the superiority of the “speaker” and anyone who identifies with him/her and by extension with the government. This is frequently done using irony and ridicule.

Finally, as a fourth level of our analysis we focused on the importance of iconic subframes that work through the practice of dog-whistling. These are extremely powerful and persuasive because readers immediately identify them as inter-textual references. They are powerful because they do not need to be explained. Even if they do not seem arrogant, diminutive, or threatening on the surface, they can easily have such undertones.

Our findings show that the pro-government media in Hungary differs from government-critical media not in the frequency of reporting on protests (i.e. selection bias) but instead in its use of derogatory language meant to discredit protests and introduce frames of illegitimacy. As such the pro-governmental media pushes the protest paradigm to the extreme: it turns it into a decisive instrument of populist politics and the de-democratization process, where democratic institutions still exist but are hollowed out and function only as a façade of democracy.

We believe that the study of non-democratic regimes and democratic backsliding – for which Hungary is a textbook case – is crucial even for those who live in seemingly robust democratic political systems.² Understanding the mechanisms through which non-democratic systems emerge and survive can offer important lessons for democratic countries, as some currently illiberal countries used to function – at least on the surface – as liberal democracies. One of the major difficulties in studying modern authoritarianism, hybrid regimes, or illiberalism is the difficulty in identifying when a regime ceases to be a democracy. As Guriey and Treisman (2019, 101) have rightly observed: “the boundary between low-quality democracy and informational autocracy is fuzzy.” Demonstrating how an illiberal regime relies on media populism in a seemingly competitive media environment could be a crucial tool in capturing this and drawing the line between a functioning democracy and a modern authoritarian system that does not rely on overt oppression yet still makes democratic life impossible. Identifying the media’s shift in the direction we have outlined above can be a crucial litmus test for recognizing a regime has started on the road to democratic backsliding. Thus, we believe our research offers an important contribution to the study of what may be termed de-democratization, i.e. the shift from the liberal democratic order to those populist or authoritarian in nature.

Our paper is structured as follows: we begin our analysis by reviewing theories of protests in non-democratic regimes and the protest paradigm. Next, we present our research questions and develop hypotheses concerning media-framing techniques of political protests. These theoretical sections are followed by the four-level analysis of framing protests in four subsections: the analysis of (1) selection bias; (2) description bias; (3) general tone of the reports; and (4) the iconic frames. Finally, we discuss our results and share our conclusions.

Protests in non-democratic political regimes and the role of the media

Political protest activity is defined as an instrument used by social movements to bring about changes in the political and social environment (Koopmans 1995, 10; Rucht and Neidhardt 2002, 9). Protests are “response(s) to societal problems and citizens’ dissatisfaction” (Dalton, Van Sickle, and Weldon 2010, 56).

Discussions on protests emphasize that demonstrations are triggered by dissatisfaction and grievances; thereby they can convey information about system stability to both the ruling elite (Lorentzen 2013) and to the opposition (Smyth 2020). Such a message, however, means one thing in a normal democracy – where power shifts are accepted as normal parts of the political “game” – and quite another in illiberal regimes where those in power seek to rule out the possibility of being replaced as legitimate rulers. That is to say, autocratic regimes tend to see protest not as normal elements of democratic life through which citizens – including critics of the ruling government – can voice their concerns but existential threats to the survival of the regime (Osa and Schock 2007).

Protests in such semi-democratic regimes play a somewhat different role than those in liberal democracies. Although protests in illiberal regimes may have a limited capacity to influence political decisions, they can still have an effect on public discourse and may play an important role in the opposition’s identity formation.

Instead of “terrorizing” or physically repressing their citizens, such regimes tend to manipulate information through soft censorship and media centralization (Guriev and Treisman 2019; Kermeoglu and Weidmann 2020). While information hegemony and media manipulation are tools that regimes can use to promote themselves, buttress their support, and even mobilize citizens in support of the regimes (Susánszky, Kopper, and Tóth 2016), they can also easily lead to demobilization and political disenchantment. That is to say, through their “capacity to shape hegemonic narratives” (Smyth 2020, 2) about political events, such regimes can also discourage mobilization and increase citizens’ passivity through which they can indirectly strengthen their hold on power.

Forty years of research on media representation of demonstrations found that the media tends to follow the pattern referred to as the “protest paradigm” (Chan and Lee 1984). According to this paradigm, media sources tend to favor the political status quo as opposed to the views and objectives of protesters. Scholars have analyzed a wide range of contentious protest events: e.g. the Vietnam and Iraq war protests (Gitlin 1980; Dardis 2006); abortion protests (Boyle and Armstrong 2009); or the Global Justice Movement (McFarlane and Hay 2003; Boykoff 2006). These studies demonstrated that mass media uses framing devices to marginalize protesters and social movements even in established democracies, and it is reasonable to assume that they exist in an even harsher form in illiberal regimes, which our case studies aim to capture.

There is vast body of literature within political science on frames and framing techniques. As Robert Entman argued, “to frame is to select some aspect of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described” (Entman 1993, 52).³

There are some typical marginalization techniques identified in almost every empirical study, which are deemed as elements of the protest paradigm frame. Analyses of news articles and editorials identified subframes (or in other words frame elements) where: (1) demonstrators are presented as a disruptive, violent crowd (Dardis 2006; Kyriakidou and Osuna 2017); (2) the actions of protesters are deemed illegal (McFarlane and Hay 2003; Boyle and Armstrong 2009); and (3) protesters are shown to have limited mental capacities (Weaver and Scacco 2013) – e.g. protesters are depicted as immature children (McLeod and Hertog 1999), a deviant minority that does not represent the majority of society (McLeod and Hertog 1999; Dardis 2006; Weaver and Scacco 2013), or simply “idiots” (McFarlane and Hay 2003).

In addition to employing these forms of marginalization, the media is often biased by being completely reliant on official sources, such as government agents or the police (McLeod and Hertog 1999). Media tends to ignore the perspectives of dissenters (Weaver and Scacco 2013; Kyriakidou and Osuna 2017). Moreover, it frequently depicts protesters as phonies or closely connected to and manipulated by political parties, which can be a way to question whether protests are genuine (Kyriakidou and Osuna 2017). These elements identified by the protest paradigm constitute what can be called the negative framing of social movements and political protests.

It is important to underline here that these techniques are employed in established liberal democracies and can lead to a decrease in public support for protesters and also to the demobilization of dissenting groups (Boykoff 2006, 2007). We postulate, however, that illiberal regimes not only apply these technics to a significantly higher degree, but use these technics to discredit protesters, depict them as enemies, and exclude them from the sphere of legitimate participants in politics. In fact, protesters are portrayed as enemies of the regime (Kopper et al., 2017).

Research questions and methods

If Hungary is indeed an illiberal regime that has moved away from liberal democracy, we should be able to grasp empirically how the media system manipulates and reports on protests and protesters in a biased manner. Our research aims to accomplish this through a four-level media analysis. The first two levels rely on quantitative methods, while the third and fourth are based on qualitative analysis.

The first two levels focus on two dimensions of media bias, namely selection bias⁴ and framing or description bias (McCarthy, McPhail, and Smith 1996). Selection bias refers to systematic under-representation of opposition protests and the issues for which they stand (e.g. declining democracy, corruption, fundamental rights, autonomy), while description bias refers to the framing techniques applied within the protest paradigm described above.

For selection bias, which is the first level of our analysis, we have formulated the following hypothesis:

H1: Demonstrations are significantly under-reported in pro-government media outlets compared to government-critical media.

Thus, we assume first that government-leaning media output is silent about protests – or covers them less than media outlets identified as critical of the regime.

The second level of our analysis focuses on framing bias based on the protest paradigm that examines differences in the usage of discrediting frames in pro-government, government-critical, and public service media. This is typically analyzed by focusing on the use of adjectives or the way protesters and protests are depicted.

Here we proposed the following hypotheses:

H2.1: Government-friendly media uses the mechanisms identified by the protest paradigm at a significantly higher proportion than public service media and government-critical media.

H2.2: The public service media use the protest paradigm to a significantly greater extent than the government-critical media.

As discussed above, such bias is to be expected in both liberal and illiberal regimes. Also, in hybrid regimes there remain remnants of the free media, even though they play on a media field heavily tilted against them. We assume this to be the case in Hungary, which, as a member of the EU, has a regime that tries to preserve the façade of democracy (on fake compliance see Kopper et al. 2020). Hence, our claim is not that bias alone means that a regime is non-democratic, as some “bias” or preference for certain parties and views is a normal part of all media organs, and thus we take some differences to be natural. Our aim is to identify what goes beyond this. That is to say, it is the extent and level of this bias that makes a difference. In other words, we expect pro-government media to be more biased than the opposition-leaning media (which is splintered and more autonomous vis-à-vis political parties and thus should not be expected to offer a hegemonic narrative on the whole). Furthermore, we also include public service media in the analysis. This we also assume to be one-sided, given that the ruling party exercises extensive control over it (Bajomi-Lázár 2017) and supports the governments’ ideologies (Polyák 2019).⁵ Notwithstanding, if public media is less prone to apply the protest paradigm against government-critical protests than the pro-government media, it would make our reasoning on media bias more incisive.

The third level of our media analysis focuses on the general tone of the reports. We believe we must pay special attention to the general diminutive, degrading tone in which pro-governmental media – especially in opinion pieces – sometimes depicts protesters. Here the point is not merely that protesters are described as imbeciles or as childish – which the protest paradigm literature highlights. Texts often employ a general tone that is not merely aggressive, but also conveys the superiority of the “speaker,” along with anyone who identifies with him/her and by extension with the government. This is frequently done using irony and ridicule (Burgers et al. 2015).⁶ The study of general-tone frames necessitates qualitative analysis, as quantitative methods and AI scanning have difficulties in picking up subtleties and “inside jokes” (Touahri and Mazroui 2021). Below we will provide passages of text as examples of such general tone-setting.

Finally, in the fourth level of our analysis we focus on what we call iconic subframes as a separate category. These are established subframes that readers can immediately identify and thereby associate with the topic, and thus they are a means of dog-whistling in which the knowledgeable reader “knows” what the tropes actually mean. These rely on the intertextual embedding of the references that these easily trigger. They are powerful because they do not need to be explained. Even if they do not seem arrogant, diminutive, or threatening on the surface, they can easily have such undertones. While we should not assume that media consumers automatically respond to subframes in the ways that media outlets intend, it is safe to say that the media, serving in its function as a tool of political socialization and mobilization (or demobilization), plays a role in setting the boundaries of legitimate discourse in illiberal regimes.

We will discuss these four levels in turn in separate sub-sections, each offering a detailed explanation of the methods used for the specific level of analysis.

Four-level analysis of framing protests

Selection bias

Data and analytical methods

To test our first hypothesis (H1) we compared different sources on protest events. On the one hand we relied on police archives⁷ to identify every registered demonstration in Hungary between January 2016 and November 2017, and we compared them to our media dataset to identify whether protest events in the police archive were reported by four different media outlets.

The police datasets included the exact date and place of each protest event. These files included information on how organizers specified the aim of their protest. While it is mandatory to submit this information to the police, in some cases organizers gave evasive answers, stating for example that they were aiming for a “peaceful demonstration” or “protesting action and marching.” Despite these superficial answers, the dataset frequently contained much longer and more meaningful explanations, such as: our protest is “Against the expansion of the nuclear power station at Paks” or “Against the school principal’s dismissal.”

Once we obtained this dataset, it was coded by two independent coders. Every event was placed into one of 26 issue categories. In cases where the aim of the protest was missing or contained scant information, we asked our research assistants to search the press for the missing data. For example, if the recorded aim was “a peaceful demonstration,” we knew nothing about the goals of the protest, but on the basis of the exact date (day) and location (city or village), coders were able to identify with great certainty the aim of the event via online news sites (mostly the local news portals).⁸

Once coding was completed, the 26 issues were aggregated into nine broader categories (see Table 1). For example, we placed protest events that demanded fundamental rights – e.g. freedom of the press, freedom of religion, democratic institutions (e.g. constitution, judicial reforms) – or raised a voice against corruption, into the “democracy” issue category.

Table 1. Protest events and their media coverage in Hungary (2016–2017)

Issue	Protest events % (N)^a	Media coverage % (N)	Government-critical media % (N)	Government-leaning media % (N)
Democracy	8% (200)	18% (36)	18% (29)	17% (28)
Economy	9% (227)	11% (22)	11% (18)	13% (21)
Discriminated groups	19% (485)	5% (10)	4% (7)	3% (5)
Environment	7% (179)	9% (19)	8% (13)	8% (14)
Culture	20% (502)	9% (19)	10% (16)	8% (13)
Policy (education, healthcare)	6% (143)	31% (64)	33% (54)	37% (62)
Conservative (abortion, family)	3% (83)	4% (8)	4% (6)	5% (8)
Other	10% (242)	5% (11)	4% (6)	4% (7)
Do not know	17% (427)	8% (16)	9% (14)	5% (8)
Total	100% (2488)	100% (205)	100% (163)	100% (166)

^aDistribution across different issues (based on police data).

Coders also identified the places where the protests took place and whether they were villages, cities/towns, or the capital city Budapest (the data also included the county in which these locales belong).

In addition to this police dataset, we also created a dataset on the media coverage of protests relying on a professional data mining company (SentiOne). Through this we have obtained a database containing 329 articles on 205 demonstrations or other protest events⁹ published by four major Hungarian media outlets between January 2016 and November 2017. Two of the chosen media outlets leaned pro-government (Origo.hu, Pestisracok.hu), while the other two were more critical of the government (Index.hu, Merce.hu).¹⁰ In order to incorporate various styles (in terms of popularity and genre), we chose two of the most popular online news portals characterized by a more general journalistic style (Index.hu, Origo.hu), while we also picked two “activist” media outlets characterized more by opinion-driven journalism (Pestisracok.hu, Merce.hu). Articles were coded by two independent coders into almost 50 variables.¹¹

Finally, for our analysis of whether selection bias did or did not characterize the media, we merged these two datasets. Thus, we produced a dataset that included: (1) every protest event reported to the police between January 2016 and November 2017; (2) the issue the protest organizers claimed was the aim of their event; and (3) four dummy variables on whether the protest event was covered by the media sources (Origo.hu, Index.hu, Pestisracok.hu, Merce.hu).

Results

Table 1 shows reported protest events and their representation in the selected media. Data in the first column derive from the police archive and show the distribution of protest events across different issues. The largest categories are protests concerning cultural or discrimination issues. Thus in 2016–2017 one-fifth of all the protests had cultural (e.g. art, science, or commemorations) or discrimination claims (e.g. Roma people or homelessness). Moreover, we distinguished protest events organized in the categories of democracy, economy, environment, policy (e.g. education and healthcare) and conservative (e.g. abortion, family) issues (8%, 9%, 7%, 6%, 3% respectively).

In the second column of Table 1 we can see that 8% or 205 (see last row, indicating total numbers) of the 2,488 protest events were covered by the media.¹² Comparing coverage in government-critical and government-leaning news media (third and fourth columns in Table 1), contrary to our expectation, we observed only minor and statistically insignificant differences (Kolmogorov-Smirnov test of differences in distributions, p = 0.98). That is to say, the government-leaning media does not withhold information about protests on democracy or other issues, i.e. there is no substantial difference between government-critical and government-leaning media outlets in the number of reported protest events.¹³

Based on these results, there is no support for our hypothesis (H1) about the government-leaning media’s selection bias, since we do not find any evidence of significant non-reporting on protests taking place.

Framing bias – testing the protest paradigm

Background: student protest

In the previous section, we took the first step in our four-step media analysis and focused on selection bias regarding protest events in Hungary. In this section we continue our media analysis by focusing on the protest paradigm.

Our second hypothesis posits that the government-leaning media uses the mechanisms identified by the protest paradigm to discredit protesters more often than do government-critical news portals. (This is what McCarthy, McPhail, and Smith 1996 defined as the description bias.) In order to test this hypothesis, we have chosen one specific protest event series, organized by students of the University of Theatre and Film Arts (SZFE in Hungarian), that was the focus of international and domestic attention and vividly illustrated Hungary’s pro- and anti-government cleavage. The question was to clarify differences in how pro-government, public service media, and opposition-leaning media framed these protests.

Students at the University of Theatre and Film Arts in Budapest started protesting the restructuring of the University and the appointment of new leadership. This restructuring was initiated by the government, which had proposed a new model for the operation of the University by bringing it under the control of a newly created foundation. The government named the members of the foundation’s board, which included some film/theatre directors and two businessmen from the oil industry. Protesters claimed the Ministry did not consult with the university leadership. They further claimed decisions were made without consultations, making both students and faculty of the University fear for the autonomy of their institution and higher education in general.

On 1 September 2020, students occupied the university’s main administration building, blocked the university’s new leadership from entering the building, and started a series of protests that lasted until 10 November. The students organized protest marches and introduced a series of innovative protest forms, such as performances in front of the building, a human chain, and inviting well-known celebrities to “stand guard” in front of the university building to symbolically protect the university’s autonomy.

We chose this protest event for our analysis, first because it was the longest and most intense protest against the government recently, and second because it triggered a vibrant debate in the media and mobilized thousands of people, even though it took place in a country that was hit hard by the COVID pandemic at the time. Furthermore, as both the start and end dates of the protest were announced by the organizers (1 September 1 to 10 November 2020), the protest had a clear-cut end date, which made it easier to define the period from which to select media articles for analysis.¹⁴

Data and methods

For this part of the research, we relied on data collected through the platform of the professional data-mining company SentiOne. We selected all news articles that contained the key word “SZFE” (University of Theatre and Film Arts) published on the most popular online news portals.¹⁵ In all, 648 of the total 1,954 articles were published on government-leaning news portals, 129 on public service media,¹⁶ and 1,177 articles were published by government-critical media.

On the basis of the protest paradigm literature, we focused on five frame elements of the protest paradigm in the body of the texts. These five frame elements or subframes reveal a full protest paradigm frame that negatively depicts protesters and the protest. After having read the bulk of the articles we identified the subframes, concentrating on keywords that were connected to the subframes (see in brackets).¹⁷

We used dummy variables in coding the articles. When an article contained one of the keywords, we coded with a 1, and with a 0 otherwise. Finally, we checked every single article to ascertain whether the words or expressions we found in the text referred to the SZFE protest.

We use the following variables to detect the protest paradigm subframes:

1. the violent subframe (words referring to violent and aggressive actions);

2. the illegitimate subframe (containing keywords suggesting unlawfulness or the breaching of the law);

3. the diminutive protesters subframe (containing expressions degrading protesters, including for example, their mental capacities, or presenting them as dishonest actors – i.e. incited, provoked – or simply puppets and mercenaries);

4. the insignificance of the protest subframe (containing words about the low number of protesters, their low mobilization capacity);

5. degrading descriptions of the protest events (including words such as: hiszti (hissy fit), tünti, tünci (both informal derivatives of the word “demonstration,” suggesting childishness, circus, carnival).

Table 2 shows the occurrence of discrediting protest subframes in the pro-government, government-critical and finally public media. We have added public (or public service) media as a separate data source in this section.

Table 2. Frequency of negative frame elements in pro-Government and government critical media

		Gov-critical % (N) Std. residuals	Pro-gov % (N) Std. residuals	Public % (N) Std. residuals	Total % (N)
	Protest paradigm frame (compound variable)	9.2% (108) -4.006	22.2% (144) 6.080	8.5% (11) -1.527	11.3% (263)
1	Violent	1.9% (22) -2.589	5.7% (37) 3.524	3.1% (4) -0.078	3.2% (62)
2	Illegitimate	2.0% (23) -1.297	3.5% (23) 1.576	3.1% (4) 0.385	2.6% (50)
3	Diminutive protesters	2.3% (27) -3.817	9.7% (63) 5.882	1.6% (2) -1.653	4.7% (92)
4	Insignificance of the protest	0.8% (9) -1.168	1.9% (12) 1.742	0.8% (1) -0.375	1.1% (22)
5	Degrading descriptions of the protest events	3.7% (44) -2.225	8.8% (57) 3.985	0.8% (1) -2.210	5.2% (102)
	Total	100% (1,177)	100% (648)	100% (129)	100% (1,954)

Two percent of the articles in the government-critical media use the violent frame element. In contrast, the pro-government media presents the student protest as violent three times more often (5.7% of the articles). Also, the diminutive protesters and the degrading descriptions of the protest events subframes are used significantly more frequently in the pro-government compared to the government-critical media (2.3% vs. 9.7%; 3.7% vs. 8.8%, respectively). We did not find significant differences between these two groups of media in the frequency of the illegitimate and insignificance of the protest subframes. It might be surprising that even the government-critical media contains elements of the protest paradigm when reporting about a government-critical protest. First, critical media may also have its reservations about protesters, but second and more importantly, critical media often cite the claims of both the challenger groups and the incumbent actors. One good example is an article on the government-critical 24.hu, which quotes the head of the advisory board about the illegitimate protest of SZFE staff.¹⁸

To summarize the data in Table 2, we can say that government-leaning media use three out of the five subframes in significantly higher proportions than the government-critical media. The student protests were presented in the pro-government media as more violent, the protesters less capable, and demonstrations generally as an inadequate form of shaping political decisions.

Coverage in the public service media fell in between the government-critical and government-leaning media. However, it is closer to the government-critical media in the usage of degrading frame elements.

From these five frame elements, we compiled a compound protest paradigm variable that measures whether any of the five elements appear in any given article (1), or not (0). It is worth noting that any one article can contain any number of subframe elements. The first row in Table 2 shows this: i.e. more than one-fifth (22.2%) of the pro-government media articles contain at least one discrediting subframe, while the government-critical and public media use the protest paradigm frame significantly less frequently in their articles (9.2%, 8.5%, respectively).

We find that the public service media, which has been reformed to serve as a conduit for pro-government messaging and communication (Bajomi-Lázár 2013; Polyák 2019; Bátorfy and Urbán 2020), pictures the SZFE student protest in a less negative way than the pure pro-government media. These results confirm the first part of our second hypothesis (H2.1), holding that the government-friendly media uses mechanisms identified by the protest paradigm significantly more frequently than the government-critical media. However, it partly confirms the second part of the second hypothesis (H2.2), since the public service media uses degrading subframes less frequently than the pro-government media: we did not find significant differences between the public and the government-critical media.

General tone of the reports

In this third level of our media analysis, we will use a few longer quotes from opinion pieces published in the pro-governmental media to show how a general degrading attitude toward protesters is conveyed by the pro-government media. (We will later add an opinion piece from the opposition media for the sake of contrast.)

Excerpt 1, from a pro-government article

For someone just a little bit knowledgeable of current events should know that we are witnessing the outbreak of a revolution. Again. I can’t recall how many times protesters have started “spontaneous” revolts by now against the unbearable dictatorship we are living in and claimed they “cannot stay silent any longer.” I really do not know how many times we have heard this already. A lot. I am not counting the teeny-weeny protests, performances, tribunes claiming to speak for the people, independent pundits. Does not matter, here we have just one more. We should not get bored and yawn, we should just laugh at it!”¹⁹

The piece is packed with irony. It is degrading, suggesting that the protests are “teeny-weeny,” and are irrelevant. The author uses the word “revolution” ironically, in no way trying to reflect on the grievances of the protesters. The whole paragraph suggests they are insignificant and in fact boring. They should be laughed at. The passage is written from the point of view of power and suggests to the reader the superiority of the author (and the regime the author stands for) over the discontented protesters.

Excerpt 2, from a pro-government article

… .so we are right when we feel angry and feel it’s time to throw a punch. This howling [by protesters] only proves that Hungary is right and the freedom-fight by Attila Vidnyánszky [the newly government-appointed director of SZFE] will succeed. The eternal SZDSZ-bolshevism’s weak spot [SZDSZ is a defunct liberal party] is its unstable mental state that makes it unable to show restraint and to withdraw decently. This mob that consists of disillusioned mummies who lie about being independent yet in fact in all their acts being anti-Hungarian, unceasingly echo a left-liberal ideology. They are the young guard of individuals of dubious background all coming from the same ÁVH [the secret service of the old communist regime in the 50s] circles, whom the author Csurka [emphasis added, see explanation below] once correctly called members of the Ascher Café [Ascher being a liberal director]. Their true devilish intention always comes to the fore and becomes visible. It is in vain they try to conceal their true nature by pretending to be pious, their true nature always reveals itself.”²⁰

The text is extremely difficult to render in English. Yet we believe the tone is self-explanatory. It links liberalism, communism, and being anti-Hungarian with being the devil itself. We contend that this goes far beyond what theoreticians of the protest paradigm have in mind. The reference to István Csurka is also of note. Csurka was a politician who broke away from the governing Hungarian Democratic Forum party in 1993 to found the Party of Hungarian Justice and Life. This party, and by extension its founder, have long been associated with extreme right and anti-Semitic thought and action (Kovács 2013). We will return to Csurka as an iconic subframe below in the next section.

The logical question is: why is this good for the government? What is the objective? We tentatively suggest that what we can grasp here is media-populism. That is to say, the objective is not simply to discredit protesters, but to divide society into those for the regime and its enemies. There is a Schmidt-ian logic that underlies the logic of illiberalism (see Majtényi, Kopper, and Susánszky, 2018, 176).

Excerpt 3, from an opposition media site

The final excerpt is from HVG, a weekly more associated with the opposition. The tone is also critical (of the government in this instance), but the style is radically different.

He is afraid. They are afraid. We are not. Lőrinc Mészáros [the regime’s favored businessman] winning state tenders one after the other, is worried about his wealth and safety. Thereby he has created a private security company. Viktor Orbán is worried about losing power, therefore he tries to silence all criticism at its source. He is afraid of free science, the autonomy of higher education and in general of free-speaking people … .²¹

This excerpt also engages in diminishing the “other.” In this case, however, the critical tone is aimed not so much at degrading the character of the subjects but instead painting the subject (a government-aligned oligarch and the Prime Minister) as authoritarian and fearful, thereby positing the “we” (the opposition) as fearless and lovers of freedom.

We want to reiterate that what is crucial is the difference in tone, with the first two quotes being not only diminutive but lacking the recognition of the other, in a sense excluding them from the community. The last quote, however, only refers to values expected of leaders in a democracy and calls on the regime to respect these. The difference is stark.

Iconic subframes

By iconic subframes²² we mean triggers to frames that are so embedded in daily discussions that they do not need to be explained. Lakoff (2010) suggests well-reinforced frames can be triggered with single words: “ … since frames come in systems, a single word typically activates not only its defining frame, but also much of the system its defining frame is in.” The term “dog whistle” is commonly used to represent this type of subframe (see Albertson 2015). All informed readers know how to decipher them and what meaning to attach to them. Here we highlight two such iconic subframes. One is the subframe “Soros,” the billionaire of Hungarian origin, who happens to be Jewish, funded the Central European University, and who has been taking a firm stance supporting an open society. For the past few years, the regime has identified him as its arch-enemy (Pintilescu and Kustán Magyari 2021). This has been successful to the degree that currently 91% of the regime’s supporters believe that Soros single-handedly manipulates protests in Hungary.²³

The following excerpt from 888.hu uses affiliation with Soros (and Soros networks and programs) to paint activists as radical.

One of them of course was Márton Gulyás … of whom it is well known gets money through Soros foundations. The other is Bálint Misetics. He usually speaks officially about homelessness, but—as has been proven—he has actively taken part in Soros radicalization training … ²⁴

The following quote, which infers that failed protesters turn to international forums, is from the online version of the largest pro-government daily newspaper, Magyar Nemzet:

A part of this strategy is seeing the provocation of street disturbances as a useful tool for mobilization, to compensate for their failure in Hungary through EU provocation and requests for interference, as well as the appearance of foreign (American, Czech, Soros-foundation) investors behind friendly media.²⁵

The pro-government website mandiner.hu reported on a press briefing by the Christian Democratic Party (the junior government coalition partner) as follows:

The spokesman was asked about Monday’s article in Magyar Nemzet, according to which the newspaper identified the person, Gerald Knaus, who as a member of the Soros network is the director and executor of actions meant to defame Hungary. The Christian Democrat politician responded to this as follows: this is just proof that there is an international defamation campaign against the country primarily because of the rejection of migration and open society principles. Behind these attacks we often find George Soros and his vast amounts of money—he added.²⁶

The dog-whistle and triggering nature of references to George Soros and his network are visible in the quotes, as they characterize affiliation with Soros as radical, foreign, and against Hungary. As such, the Soros iconic subframe serves as a method to efficiently infer a host of protest paradigm subframes. It is safe to assume that such inferences are shared by and reinforced for the members of the pro-government camp.

Another iconic frame we highlight is the name Csurka. Csurka was a playwright and author with far-right ideas. When he is referred to, the informed reader immediately remembers things he said and the nationalistic and anti-Semitic ideas associated with his name. Thus, the text does not need to be explicitly anti-Semitic, as such an undertone is created by inferences via established iconic subframes. The mere mention of Csurka is a bold statement and advertisement of the anti-Semitic inclination of the piece. The following quote from pestisracok.hu serves as a good example of the Csurka iconic subframe (another example for this is in excerpt no. 2 above in the quoted text of the previous section):

Because it’s better to pretend to be a revolutionary. Then, when they shut off the Wifi, they’ll go crying home to mommy … According to the young people “in despair” who occupied the university … Poor babies. And what is going to happen when they are even more frightened? Will they set the city on fire?—wrote István Csurka in 2011 about the disturbances around the New Theater. And the situation has not changed much since.²⁷

It is apparent from the quotes relayed that the use of iconic subframes in even one sentence can “reframe” a discussion by subtly or crudely importing affiliations and connotations found in the protest paradigm.²⁸ Much like dog-whistle communication – the examples above indicate that the two frequently go together – the use of iconic subframes brings into the conversation a set of assumed and recognized values identifying both the author and the subject of her/his consternation. Such iconic subframe use can change the tone of an article that is factually correct into one resembling a rant. The iconic subframe also serves to identify “us” and “them” and to distinguish between the two groups. The economic subframe thus links up with the argument we have made in the previous sub-section (on general tone).

Conclusion

With Hungary’s illiberal turn, its regime has dismantled and hollowed out democratic institutions and shifted the country away from norms of democratic politics. Yet, as Hungary is part of the EU, this power grab had to be presented in a way that shows – at least superficial – recognition of democratic norms and at least fake compliance with them (see Kopper et. al. 2020). This means that to some extent media pluralism remained, with critical voices still being heard. Nevertheless, the government manages to dominate political narratives. The purpose of our article was to scrutinize this through a four-level analysis of the Hungarian media landscape to clarify to what extent media reports are biased and in what way political opponents are depicted. We focused on how government-critical political protests were portrayed in the Hungarian media. We contend this is crucial, because the discrediting, diminutive tone in portraying protesters can weaken their support among citizens in general and undermine their legitimacy and effect.

Focusing on this issue, we recognized that reporting on protesters can also be biased in properly functioning democracies, as the protest paradigm captures. Thus, simply finding that the media portrays protesters negatively is not something to be surprised about. Still, the extent to which this bias is present in the Hungarian media is a question we believe demands detailed empirical analysis, which we have conducted on four levels.

The first two levels were quantitative. We proposed two hypotheses, concerning selection bias and description bias, respectively. We had to reject our first hypothesis (H1) as contrary to our expectations: we found that both pro-government and critical media report on protests to roughly an equal extent, and thus there is no clear selection bias present. Our second hypotheses (H2.1 and H2.2) were based on the protest paradigm. Here we expected pro-governmental media to use it more, i.e. to use more discrediting frames in portraying protesters. Here our hypothesis was confirmed; indeed, the protest paradigm is more prevalent in the pro-governmental media (and even in the public service media, which has been dominated by the political regime) than in the critical media. Namely, government-leaning media depicts protesters as violent (violent subframe) and insignificant groups of negligible young people (diminutive protesters subframe), who have been manipulated (degrading descriptions of the protest events subframe) by opposition parties or George Soros himself. All in all, 22% of the pro-government media articles used the protest paradigm, whereas this proportion is less than 10% both in government-critical and public service media.

Reaching this point in the study, we realized that we needed to delve even more deeply in our analysis, and we supplemented our media analysis with third and fourth levels. Thus, we moved beyond selection bias and the protest paradigm, to include in our analysis what we have identified as general tone subframes and iconic subframes. The identification and interpretation of these subframes are impossible using only frequency measurements with a quantitative logic. As such, their study requires a shift toward more quantitative analysis that is sensitive to the textual and political context in which subframes are implemented.

General tone subframes help move articles from a journalistic disposition toward having a more opinion-driven character. These subframes help set the nature of reports and are not closely tied to the context of the reporting or the given issues. Iconic subframes, in contrast, fulfill two functions that go beyond those assigned to frames by Entman (1993). The first iconic subframe function we identified is their role in importing wider yet specific narratives into the conversation, e.g. nationalism, anti-Semitism, treason, etc. In this sense iconic subframes function much like dog-whistle communication does. Second, in their specific contexts these subframes help delineate the divide between insiders and outsiders, the initiated and the uninitiated, i.e. those who understand and support the subtexts and those who do not.

This second function of iconic subframes is interesting in its similarity to accepted notions of populist politics that define populism partly through its efforts to see society as two homogeneous groups: the pure people and the corrupt elite. While identifying university students and protesters as elites may be questionable, we have presented texts that connect these actors to more powerful forces (e.g. George Soros) who are unequivocally corrupt elites in the narrative of the Hungarian populist-illiberal camp. The use of iconic subframes in the long run should serve to not only reinforce support for the policies of the Hungarian government, but to indeed deepen the divide envisioned by far-right populists.

Our study has further shed light on methodological choices for studying media coverage of protests. Most of the subframes we analyzed, like those derived from the protest paradigm, function independently of their specific contexts. That is to say, we can expect them to be utilized across space, issue, and time. While frequency tabulations and quantitative approaches do not exhaust analytical methods, they can serve as excellent starting points for inquiry into the logic of media communication in regimes liberal and illiberal. Iconic subframes, in contrast, are incomprehensible without a firm grasp of the context(s) in which they are employed. Their interpretation and understanding depend on: (1) the political and policy context of the given moment; and (2) the specific narratives they import into the public conversations about given polities and policies. Given their populistic function revealed in our study, these conversations are less public and more sub-public, in that they are meant to address a specific audience, i.e. the pure (Hungarian) people. For these reasons quantitative analysis of iconic subframes is likely to be frustrating. Qualitative approaches, of which we have only just scratched the surface, are more promising.

The use of the subframes analyzed in this paper should not be limited to Hungary. The protest paradigm did not arise from the study of only illiberal or populist regimes. Further, the spread of social media has acclimatized readers to the use of general tone subframes, be this in liberal democracies or otherwise. As seen above, there is little selection bias in Hungary’s pro-government media reporting on demonstrations. It is the appearance of iconic subframes, which are akin to dog-whistle communication approaches, that indicate a shift of the media sphere to populist logic. In Hungary, the fact that the governing party has exerted increasing direct control over the media market has likely amplified this logic.

Beyond the methodological novelties, we presented significant empirical outcomes contributing to the informational autocracy literature (Gehlbach and Sonin 2014; Guriev and Treisman 2019, 2020a) by capturing how media populism operates in Hungary. According to these theoretical works, media plays a crucial role in manipulating citizens’ beliefs and attitudes in non-democratic regimes. Our results support the first part of this statement. Namely, the pro-government media is highly biased in the interpretation of the challenging group’s claims and uses the negative framing technique of social movements and political protests quite frequently. While the present study cannot reveal how effective the pro-governmental media is in manipulating citizens, our earlier study revealed that pro-government/government-critical media consumption is an important factor in explaining protest participation (Susánszky, Kopper, and Tóth 2016). Therefore, it is worth studying the degree to which the intensive use of the negative framing technique makes it possible for power holders to mobilize their supporters and to deter the mobilization of critical voices. Our paper has focused on the supply side of negative framing techniques. The manner of consumption of these frames is a matter for further study.

Notes

• 1. As Hungary is a member state of the EU and a beneficiary of EU funds, its regime tries to maintain the appearance of a working democracy. Thereby it “fakes compliance” with European norms, meaning that on the surface they are followed, but in substance they are completely hollowed out; as a result, they do not fulfill their democratic role (see Kopper et. al. 2020)

• 2. Although the number of overt totalitarian regimes in the world has declined in recent decades (Guriev and Treisman 2019), we are increasingly witnessing states in which democratic institutions appear to be in place – and competitive elections are seemingly held – but in which, nevertheless, the political system is rigged to such an extent that they hardly qualify as consolidated functioning democracies any longer. Scholars refer to these states as hybrid regimes (Levitsky and Way 2010), electoral autocracies (Maertz et al, 2020), informational autocracies (Guriev and Treisman 2019) or illiberal democracies (Pap 2017).

• 3. Reese (2007), however, claims that the use of framing as a model is problematic, as researchers are prone to use frames in lieu of the terms “topic” or “theme.” Reese defines frames as “organizing principles that are socially shared and persistent over time, that work symbolically to meaningfully structure the social world” (2001, 11). As such, our use of the model is justified only when we can point out how frames organize and structure messages and narratives in existing political camps.

• 4. For explanations of selection bias in democratic states, see Hocke (1996), McCarthy, McPhail, and Smith (1996), and Olzak (1989).

• 5.https://mertek.eu/wp-content/uploads/2021/06/Sz%C3%BAr%C3%B3pr%C3%B3ba-33.pdf, accessed 20 July 2021.

• 6. As the Hungarian government and the Prime Minister tend to see politics as a game of football, this article on football and the irony of defying the out-group is very telling here.

• 7. We would like to thank the community of atlatszo.hu, one of the most well-known human rights civic organizations in Hungary, for legal consultations they provided. https://kimittud.atlatszo.hu/request/magyarorszagon_2010_2017_kozott#incoming-14923

• 8. Inter-coder reliability is substantial, since the average Cohen’s Kappa statistic is higher than 0.7 (Cohen 1960). The main source of disagreement between the coders was the uncertainty in identifying other and do not know categories. Taking into consideration all these details, our reliability check shows that our codebook is unequivocal and thus our data are reliable.

• 9. To collect news content, we used the following key words: demonstrat*, protest*, strike, human chain, skinhead, civil disobedience, day of honor (in Hungarian: tüntet*, sztrájk, demonstrá*, élőlánc, szkinhed, skinhead, szkinhead, bőrfejű, polgári engedetlenség, tünti, tüntékéz, kitörés napja).

• 10.http://library.fes.de/pdf-files/bueros/budapest/13063.pdf; https://atlo.team/igy-nez-ki-a-kormanyparti-hirmedia/; https://mertek.eu/wp-content/uploads/2020/01/MertekFuzetek17.pdf; both accessed 20 July 2021.

• 11. Inter-coder reliability was sufficiently high for all variables (alpha>0.7).

• 12. There are, however, important differences in representativeness among the issues. One-third (31%) of total media coverage consisted of reports about protests on policy issues, such as protests seeking reforms in education and healthcare. Another one-fifth (18%) of the media coverage concerned the democracy issue. Thus, our data reveal that the media is more willing to report on democracy or policy issues (18% second column), while issues such as protests by groups facing discrimination (e.g. the Roma, LGBT persons, groups with low social and economic status) remain relatively underrepresented, including in the government-critical media. This issue is, however, beyond the scope of this study.

• 13. We created these issue categories because selection bias could have been present, for example, only concerning the democracy issue, but not the others, which would have yielded important information on how bias works.

• 14. The end of the protest series did not mean that the issue was resolved: referring to the pandemic, the government introduced a lockdown that made protest impossible. See https://helsinkifigyelo.444.hu/2021/02/19/a-magyar-allam-szaz-napja-vette-el-tizmillio-ember-gyulekezesi-jogat, accessed 20 July 2021.

• 15. Government-critical media: 24.hu, hvg.hu, index.hu, merce.hu, telex.hu. Government-leaning media: 888.hu, www.origo.hu, magyarnemzet.hu, mandiner.hu, pestisracok.hu. Public service media: MTI.hu, Hirado.hu.

• 16. We added public media to this second level of analysis because we want to see how biased the government-leaning media is relative to the public service media. Thus, in a sense public media serves as a benchmark for evaluating description bias (i.e. the usage of protest paradigm frames in the media).

• 17. First, we collected key words from the texts. Then we used the Hungarian thesaurus https://szinonimaszotar.hu/ to find the possible synonyms.

• 18. See, for instance: https://24.hu/kultura/2020/09/24/szfe-kuratorium-alternatiiv-oktatas/, accessed 20 July 2021.

• 19. Pozsonyi Ádám, “Nincs tarkón a kéz, gyerekek!” 888.hu, 1 September 2020. https://888.hu/piszkostizenketto/nincs-tarkon-a-kez-gyerekek-4270602/, accessed 20 July 2021.

• 20. Ágoston Balázs, “A normalitás helyreállításához eljött az ördögűzés ideje,” Magyar Nemzet, 9 September 2020. https://magyarnemzet.hu/velemeny/a-normalitas-helyreallitasahoz-eljott-az-ordoguzes-ideje-8636577/, accessed 20 July 2021.

• 21. Dobszay János, “Ő fél ők félnek,, mi nem félünk,” HVG, 22 October 2020. https://hvg.hu/itthon/202043_o_fel_ok_felnek_mi_nem_felunk, accessed 20 July 2021.

• 22. Here we build on, for example, the work by Alexander, Bartmanski, and Giesen (2012) on icons.

• 23. https://www.globsec.org/wp-content/uploads/2020/11/Magyarorsz%C3%A1g-Fact-Sheet.pdf, accessed 20 July 2021.

• 24. https://888.hu/ketharmad/veletlen-egyetemfoglalasrol-szolo-tabort-szervezett-korabban-diakjainak-az-szfe-4273167/, accessed 20 July 2021.

• 25. https://magyarnemzet.hu/belfold/nemzetkozi-recept-alapjan-provokal-a-baloldal-8806115/, accessed 20 July 2021.

• 26. https://mandiner.hu/cikk/20200907_a_fidesz_kdnp_valaszokat_var_biro_laszlotol_arra_hogy_valoban_fenyegette_es_kisemmizte_a_dolgozoit, accessed 20 July 2021.

• 27.https://pestisracok.hu/a-szikszalagos-forradalom-elbukik-most-a-nemzetiek-ideje-jon/, accessed 20 July 2021.

• 28. Iconic subframes do not need to refer to individuals. The term “libernyák,” which roughly translates as “libtard,” is a diminutive word coined by Viktor Orbán to belittle and demean liberals. The term was picked up by the pro-governmental media and is now used regularly to depict the regime’s opponents, including protesters at SZFE. (An example of its use from Magyar Nemzet: “And these are the teachers and professors at SZFE who practically have power over life and death. They live in a libernyák bubble where anyone who thinks differently from them is the enemy. And they are untalented. People not good enough for their sacred profession.”

Acknowledgments

This work was supported by the National Research, Development and Innovation Fund of Hungary (FK 135274). Ákos Kopper would like to express his gratitude to the Gerda Henkel Foundation, which provided him with a generous grant that allowed him to focus on research and to complete, among others, this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).