Troublemakers in the Streets? A Framing Analysis of Newspaper Coverage of Protests in the UK 1992

Abstract

Research indicates that when mainstream news media report about demonstrations, protesters often face delegitimizing coverage. This phenomenon, known as the “(journalistic) protest paradigm,” is thought to be a default mindset that leads journalists to emphasize the method of protesters over their message. However, empirical work has so far limited itself to specific protest movements or events and only covers brief periods. This study first identifies and then codes the main frames in all reports about domestic protest in the United Kingdom. Analysing data that covers eight national newspapers during a 26 year period (N = 27,496), I provide a more systematic understanding of how the mainstream news media in liberal democracies report about protests. The analysis shows that a stable majority of articles uses frames linked to the protest paradigm throughout the time period. However, a substantial and growing number of articles employ legitimizing frames—either on their own or co-existing with delegitimizing framing.

Keywords

protest paradigm media framing social movements news reporting content analysis

Public protest is an increasingly commonplace way for citizens to express their opinions immediately and relatively low-threshold—compared to waiting for the next election, becoming a member of a political organization or being elected into office. From early protests by the anti-slavery and suffrage movements, to anti-poll tax, fox hunting demonstrations as well as anti-Iraq War and anti-Brexit protests, citizens marching in the streets are an important part of the history of Britain and other Western democracies. Public protest can be a powerful resource for formally powerless groups to make their voices heard, influence public opinion and political agendas. However, the protesters need to communicate to their audiences somehow. And while direct messages, alternative or social media might become more important lines of communication once people know about a group or issue, first contact to the general public is often still established indirectly through legacy news media (Bennett and Segerberg, 2012).

There already is a substantial body of literature that explores the coverage of protest in the media. This research is often subsumed under the term “protest paradigm” (Chan and Lee, 1984). In short, the literature argues that protest coverage regularly discredits protesters’ messages actively, treats them only as a side note, or omits goals and grievances in favor of other details, such as the spectacle and conflict surrounding protests or the deviance of protesters. However, empirical work used to be primarily limited to specific protest movements or events (e.g., Iraq War or anarchist protests). Additionally, longitudinal studies are sparse, which might be problematic as foundational work in the field is relatively dated (e.g., Halloran et al., 1970; Murdock, 1973; Gitlin, 1980) and the last decades have brought significant changes to the media ecology and the way protests are organized and conducted—not least due to the revolutionary developments brought by the internet (Cottle, 2008; Bennett and Segerberg, 2012).

More recent work has, therefore, revisted the protest paradigm to ask how prevalent and conditional it (still) is. This article makes two main contributions to that literature by systematically analyzing how protest is framed by news media in the United Kingdom: while it builds on the theoretical foundation of the protest paradigm literature, frames are identified inductively. The set of frames described by McLeod and Hertog (1999) is often used as a starting point to code several or all described frames. However, the “protest paradigm was also theorized within a media ecosystem that was more industrial than digital news audiences encounter today” (Kilgo et al., 2019, 418). One goal of this study is to re-evaluate the framing categories associated with this theory.

The second contribution is a systematic analysis of the prevalence of each of these frames in the coverage of protest. To capture differences between protests and developments over time, this article uses a new dataset of all articles that cover domestic protest and were published between 1992 and 2017 in one of eight major British newspapers.

Using semi-automated frame coding, it was possible to analyze the entire dataset spanning 27,496 articles, allowing for the systematic examination of protest reporting patterns on a scale that is unprecedented over a multitude of different protest movements and events.¹ Overall, results show a stable majority of articles used frames linked to the protest paradigm throughout the time period. However, a substantial and growing number of articles employed legitimizing frames—either on their own or co-existing with delegitimizing framing.

Theoretical Background

The Protest Paradigm

After the salience of protest coverage increased following the civil rights era, landmark works of Halloran et al. (1970) and Murdock (1973) in the United Kingdom and Gitlin (1980) in the United States, showed that protests in the 1960s had often been delegitimized in reports by mainstream media. Shortly after Gitlin’s study, Chan and Lee (1984) comprized the theoretical explanations and empirical observations into a single concept. Relying loosely on the “paradigm” concept by Kuhn (1970), Chan and Lee (1984) suggested that a “ ‘metaphysical’ world view or a gestalt” (p. 187) shapes what journalists define as entities of concern, indicate “to journalists where to look (and where not to look), and informs them about what to discover” (p. 187). This view supposedly leads reporting to fall in line with pre-defined, ideological informed patterns of reporting whenever reporters make sense of a story. The concept was taken up by McLeod and Hertog (1992), who coined the expression “protest paradigm,” which was defined as a “routinized journalistic paradigm for covering social protest” (p. 206).

The protest paradigm is closely linked to broader theories of newsmaking, which suggest that reporters tend to place new events into an internal “mental catalogue of news story themes, including how the ‘plot’ will actually unravel and who the key actors are likely to be” (Berkowitz, 1992, 83), which helps them to streamline the news gathering and decision-making processes. Narratives and procedures which have worked well in the past, therefore, influence how new stories are covered. Celebrities, ordinary people and victims often receive standardized roles within the narrative of a news story (Langer, 1998). This is generally deemed an effect of necessity, as journalistic work is characterized by tight deadlines, limited resources, a basically infinite supply of raw information and a high uncertainty about what is important and interesting enough to be news. Individual journalists, as well as media organizations, are, therefore eagerly “routinizing the unexpected” (Tuchman, 1973, 110) by creating practical and effective procedural rules, routines and mental catalogs of news story themes. The protest paradigm is essentially one of these mental catalogs, applied specifically to news coverage of social protest—which in itself is not problematic.

However, what is thought to be worrisome for society about the protest paradigm is that while the existence of alternative narratives is not strictly denied, and McLeod and Hertog (1999) postulate the existence of mixed, balanced and sympathetic reporting, the protest paradigm would mostly generate dismissive protest coverage. This, in turn, is thought to have a negative effect on the outcome of protests. McLeod and Hertog (1999) suggest that regular employment of the protest paradigm can contribute to three processes: delegitimization, marginalization, and demonization of protest events, protest movements, or protest as a political resource. In other words, at the moment when protesters need to establish themselves as a legitimate voice in the political discussion, the media undermine their legitimacy, portray their protest as smaller and more deviant from mainstream opinion than they are and exaggerate any potential threats posed by protests.

However, is this what actually happens? Nearly all studies in the last decades that analyzed news media coverage of protest use the protest paradigm concept in some form. A systematic search for studies that employed the concept to analyze legacy news media since 1984—the year in which Chan and Lee coined the term—returned fifty three studies. These studies generally agree that coverage of the studied events contains at least some patterns described by the protest paradigm literature. In line with McLeod and Hertog (1999), most research presents the protest paradigm as the default condition for reporting. However, the studies differ in how strictly they find journalists to be guided by it: thirty two studies concluded that the coverage they examined was largely in line with the paradigm, while twelve found only partial support and eight concluded they found contradicting evidence (see Table A1, Supplemental Information).

This disagreement might be explained by two gaps in the literature. First, the bulk of studies is not generalizable to the whole population of protests covered in the media. Most of the studies examine narrow case studies of single, often radical, protest events, or scrutinize the coverage surrounding a specific issue or movement. Other studies consider a broader range of protests but still limit their scope. Boyle et al. (2004), for example, consider a wide variety of protest events and also take a long period of time (1960–1999) into consideration but limit their research to five local Wisconsin newspapers. Most sampling strategies are also relatively narrow: in an otherwise well-designed study, Boyle et al. (2012), for example, retrieve articles “by a Lexis-Nexis search using the key word protester to search headlines” (p. 132). Yet, they do not further explain the step of limiting the scope to “protester” instead of “protest” or why they search only in headlines. Worryingly, this is likely due to a resource problem rather than a deliberate decision and findings about specific issues or movements might not necessarily tell us everything about the general coverage of protests. Fortunately, methods to significantly reduce the resources needed to code texts by employing machine learning and other computational techniques for content analysis have become more advanced and are more regularly employed in the social sciences in the last decade (see Grimmer and Stewart, 2013, for an overview). They allow to pass on sampling completely, increasing efficiency of the analysis and decreasing uncertainty. However, to the best of my knowledge, these methods have not been employed to assess protest coverage.

The second gap concerns a shortage of longitudinal studies. This might explain conflicting evidence as the theoretical core of the protest paradigm is several decades old at this point. Specifically, scholars have argued that there has been a transition from a low to a high-choice media environment, which led to more competition among legacy outlets and with emerging information channels on social media sites and the rest of the internet (van Aelst et al., 2017). Protests in particular might profit from these developments as it is considerably easier for protest groups to distribute their messages: while the internet massively lowered costs to set up outlets for the distribution of information about a group’s goals, those costs have been reduced to basically zero on social network sites. In turn, journalists now heavily employ social media—especially Twitter—to gather information, communicate with colleagues and distribute news to scoop competitors (Parmelee, 2013). Hence, ideas spread by low-resourced organizations through social media get distributed to audiences not only directly, but also indirectly as journalists employ this information to deliver timely updates on unfolding events in news media.

Furthermore, several authors suggest that since the 1960s, protest has been normalized as a legitimate political tactic (e.g., Norris, 2002, 188–212). Today, many different groups employ protest, and many more people have participated in a protest at least once in their lives. In Britain, Milne (2005) provides some observational evidence that sections of the British press frequently get behind or even champion specific protests that are in agreement with their editorial lines. These two trends are only highlighted as they might gain protests more supportive attention and represent a mere fraction of the changes in the media, protest organization and society. The reason for mentioning them is that recent evidence suggests that the protest paradigm might be applied more conditionally (Kilgo and Harlow, 2019; Wouters, 2015), and less frequently over time (Araiza et al., 2016; Kyriakidou and Olivas Osuna, 2017). However, none of these studies include a long enough time frame to capture the fundamental changes in newsmaking, protest organization, and society. The suggestion by Cottle (2008) that old findings about the characteristics of protest portrayal in mainstream news media must be revisited thus seems more than plausible.

Media Framing

To frame is

to select some aspects 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, original emphasis).

Many studies evaluating media coverage of protest do so by measuring media framing (McCurdy, 2012, also see Table A1, Supplemental Information). This is likely due to two main advantages: first, frames are useful categories to describe a large number of different media reports with a relatively small set of meaningful categories. And second, different frames are thought to have distinct influences on the audience (e.g., Kilgo and Mourão, 2021; Arpan et al., 2006), promising to bridge research about media content with research on media effects (Matthes, 2009).

Like in other areas of research, how framing has been used has changed over time though. Early studies, like McLeod and Hertog (1999) provided the groundwork by identifying and describing common frames in protest coverage, yet did not assess their prevalence. Later studies have often reused these frames but measured how commonly they appeared through quantitative content analysis. This enabled researchers to compare how commonly frames linked to the protest paradigm are used in different countries (e.g., Dardis, 2006, Shahin et al. (2016); Harlow et al., 2020), for different kinds of protests (e.g., Kilgo and Harlow, 2019; De Cillia and McCurdy, 2020), employing different tactics (e.g., Wouters, 2015; Wasow, 2020) between different news outlets (e.g., Kyriakidou and Olivas Osuna, 2017; Kim and Shahin, 2020), on social media sites (e.g., Mourão et al., 2021; Harlow and Johnson, 2011; Harlow et al., 2020), and over time (e.g., Gil-Lopez, 2020; Jha, 2007). However, while some studies have complemented the original set with inductively identified frames (e.g., Kyriakidou and Olivas Osuna, 2017; Mourão et al., 2021), Kleut and Milojevic (2021) remark that the original set by McLeod and Hertog (1999) has rarely been assessed for completeness by fully inductive research strategies.

This article continues this approach by scrutinizing the changes in journalistic framing practices over the last decades in a quantitative content analysis. It also reassesses the frames proposed by seminal studies in the field by identifying the most common frames in the dataset inductively.

Theoretical Expectations

This study examines the framing of protest coverage in mainstream news outlets. The majority of studies on the matter so far concluded that the protest paradigm is a main organizing force behind the reports on protest events, leading the media to cover protest with delegitimizing framing. The first hypothesis tested in this study is therefore:

Hypothesis 1: Stories on protest events mainly use delegitimizing framing as described in the protest paradigm literature.

As described above, the protest paradigm has been found to be less prevalent in more recent research. Given that there were sweeping changes in the media landscape, protest organization, and a steady normalization of protest as a political tool, I expect that protest reporting has transformed parallel to these developments. Specifically, I expect that:

Hypothesis 2a: Delegitimizing framing decreases in frequency over time.

Hypothesis 2b: Legitimizing framing increases in frequency over time.

Methodology

Case Selection and Data

I analyzed coverage of all domestic protest in eight mainstream newspapers, based in the United Kingdom, from 1992 until the end of 2017. The specific time frame was chosen as it captures the rise of the internet as a mass phenomenon, which is the development thought to have the biggest impact on news media coverage. Newspapers slowly started to use the internet in the early 1990s, while the first protest used the internet to communicate to the public in 1992 (McKay, 1998). However, this still pre-dates regular internet usage by the general public by at least a decade. The chosen time span thus allowed this study to capture the effects of internet usage roughly from their start.

Since the focus of this study is on change over time, a single country design was chosen, which increases the internal validity of the analysis. The United Kingdom was deemed especially relevant, as it was one of the countries in which the phenomenon of marginalizing coverage of protest groups was first observed (Murdock, 1973; Halloran et al., 1970).

The scope is intentionally broad to improve the generalisability of findings. Table 1 shows the choice for newspaper outlets and reflects the aim to cover the different ideologies and newspaper types that divide the newspaper landscape in the United Kingdom while also using the outlets with the highest circulation numbers. Articles were downloaded from LexisNexis using the keywords “protest” and “demonstration” as well as several different spellings of these words. To make sure these two keywords really retrieve all relevant articles about protest, a pilot study was first conducted using a number of synonyms (e.g., “march”, “rally,” or “direct action”), which did not result in additional relevant items. Articles were processed using the R-package LexisNexisTools (Gruber, 2021) and duplicated entries, items about non-domestic protest and false positives were removed. The decision to limit the data to domestic protests was made for theoretical clarity as research has shown that non-domestic protest receives significantly different reporting (Boyle et al., 2012). The final database consists of 27,496 newspaper articles.

Table 1.

Selection of Newspaper Outlets for Data Collection.

Type	Ideology	Outlet	Available from	Circulation (2005–2017)
Tabloid	Left	The Mirror	1995	1,720,000–692,295
Tabloid	Right	The Sun	2000	3,273,000–1,602,320
Tabloid	Right	Daily Mail	1992	2,426,000–1,442,924
Broadsheet	Left	The Guardian	1992	366,645–153,431
Broadsheet	Left	The Independent	1992	263,595–NA
Broadsheet	Right	Financial Times	1992	419,386–190,046
Broadsheet	Right	Daily Telegraph	1992	460,585–907,329
Broadsheet	Right	The Times	1992	679,190–440,736

Source for circulation numbers: http://www.magforum.com/papers/nationals.htm. ^bNo figures available for The Independent in 2017 as the outlet moved to online-only publication in 2016.

Semi-Automated Content Analysis of Media Frames

How frames can be extracted from text in a valid and reliable way to be analyzed is still disputed (e.g., Matthes, 2009). I used an innovative procedure to semi-automatically identify and code media frames in a large-scale media database. This approach combines best practice methods of identifying frames inductively with natural language processing and machine learning approaches to code frames in the large dataset of media reports.

The frame identification step of the approach is based on work by Matthes and Kohring (2008). They note that frames are usually immensely complex categories that are hard to identify, describe and code. However, they also introduce the understanding of frames as recurring patterns of different frame elements used together in the same texts. The elements of frames Matthes and Kohring (2008) suggest are derived from the definition of frames by Entman (1993) quoted above: problem definition, causal interpretation, moral evaluation, and treatment recommendation. Using this conception of frames, coding was broken down into smaller, more clearly defined categories, which made coding decisions more manageable and transparent, thereby improving the reliability and validity of framing analysis considerably.

The codebook was constructed using a combination of inductively and deductively discovered codes. This way, it was possible to draw on the body of literature reviewed above, while new categories found in the dataset, which covers a longer time frame and broader scope than previous research, could be seamlessly included. The frames described in the literature were split into content analysis codes following the explanation of each element by Entman (1993) and suggestions by Matthes and Kohring (2008) on how to improve content analysis of media frames.

Based on the developed codebook, 500 randomly chosen articles from the dataset were coded. To make coding decisions as small and explicit as possible, codes were assigned on the paragraph-level. Only after the coding task, values were aggregated on the article level, making the article the unit of analysis. Reliability data consisted of a random sample of 600 paragraphs (approx. 20% of the entire sample that was manually coded). This data was coded by at least two coders. Agreement ranged between 0.71 and a perfect value of 1.00 using the Krippendorff’s $α$ (alpha) measure (see Table B2a-b, Supplemental Information for details on the manual coding phase).

Only after the material had been coded, frames were identified through a dimension reduction method, specifically factor analysis. The principle of factor analysis is that variables which correlate—in this case, codes which are often used together—are determined by underlying latent dimensions—in this case, the organizing frame of an article. Since these underlying latent dimensions are hard or impossible to measure, factor analysis can be used to aggregate variables that are easier to measure. It is thus an appropriate technique for identifying frames. Details on the approach can be found in Gruber (2021).

I performed exploratory factor analysis (EFA) using the R-package psych (Revelle, 2019) with oblique rotation. To determine a statistically optimal number of factors, I conducted a parallel analysis, which suggested 13 factors, explaining a total variance of 54%. Seven of these factors appeared meaningful, while the remaining factors appeared too specific from a theoretical point of view to be considered frames. Since omitting categories might undermine the validity of the model a confirmatory factor analysis was additionally performed. The comparative fit index (CFI) for the model with the seven chosen factors is 0.94 which is above the 0.9 threshold usually deemed good (Brown, 2015, 73–75).²

After the main frames had been identified, the second step, frame coding, determined for each article in the dataset if one of the identified frames was present or not. First, the 500 randomly chosen articles were assessed using the same factor analysis model described above. Specifically, while factor loadings describe the importance of a code for a factor, factor scores provide information about a case’s placement on the factors. As factor analysis employs a mixed membership approach, it can appreciate the existence of multiple frames in the same article, which makes theoretical sense. Factor scores are dichotomized at the value 0 with positive values indicating the presence of a frame. This is done to prepare the data for the second frame coding step.

In this second step, the remaining 26,613 articles were coded using supervised machine learning (SML). While there are major differences in how supervised learning algorithms work, the principle behind them is the same: first, human coders select categories and annotate a sample of texts. Then this sample is fed to a computer for an algorithm to “learn” how to code texts to subsequently categorize the remaining text into the determined scheme (e.g., Grimmer and Stewart, 2013). Since they are designed to emulate human coding, validity can be checked by comparing the computer’s coding decisions against these of a human. Using this technique, the cost for coding frames in large datasets can be substantially reduced (Burscher et al., 2014).

The 500 manually assessed articles are treated as a training set from which the computer derives the implicit rules connecting the words in a text to the classification of an article, which either contains a frame or not. These rules, referred to as a model, are then used to replicate the human coding on the remaining set of documents, called test set (Grimmer and Stewart, 2013). Currently available SML algorithms are designed to handle a specific task: learning and replicating the relationship between variables and class membership (Grimmer and Stewart, 2013). To accommodate this design, factor scores are dichotomized and one separate model is trained for each frame.

I use the R-package quanteda (Benoit et al., 2018) for preprocessing. Since there is no agreement which preprocessing steps or which algorithm work best for SML, I tested all different combinations of preprocessing chains as suggested by Denny and Spirling (2018) for nine different algorithms implemented in the R software environment. For validation, 90 percent of the already coded sample are used to train an algorithm before classifying the remaining 10 percent and comparing the predicted with the true class. Validation results for the best models ranged from $F 1 = 0.83$ (cause & grievances) to $F 1 = 0.95$ (nuisance), which can be regarded impressive (Graesser et al., 2011, other metrics and more on the procedure can be found in the Supplemental Information). I then used the best combinations of preprocessing steps and algorithms to train seven models, one for each frame. Using these models, the remaining 26,613 articles were annotated.

Results

Frame Identification

Figure 1 shows the factor loadings of the model, which are used to interpret the different factors. In this plot, higher values indicate that a paragraph-level code is more important for a factor. For example, an article linked to the factor troublemakers was usually coded to use “violence/crime” as the most important topic, to mention concerns for “public safety” and “property destruction” as the main risks, for which protesters were coded to take the blame, and a negative judgement of the protest was assigned (see Supplemental Information for an explanation of the paragraph-level codes). From here on, these factors are treated as frames and were named based on the knowledge about the codes with high loadings in each factor.

Figure 1.

Factor loadings (X-axis) of content analysis codes (Y-axis) for selected frames (top labels)

The first frame, police violence , highlights the clash between protesters and police and the risk this poses to members of the police force, the protesters and the general public. In some ways, this matches the existing literature. According to a range of studies, the emphasis on violence is a key component of the protest paradigm. They consistently found that clashes with the police and other violence is one of the main elements the media highlight, even when little to no violence actually occurs (e.g., Murdock, 1973; Gitlin, 1980; McLeod and Hertog, 1999; Kilgo et al., 2019). Contrary to most theoretical expectations, however, the police violence frame portrays police, who are most often described as the main actor, as responsible for the trouble—hence its name. One prominent example in which this frame was frequently employed in the coded sample was when Ian Tomlinson died from wounds inflicted by a police officer during the 2009 G20 summit protests in London.

The second frame, law & order , is the theoretical opposite of the first. It highlights the positive actions of the police, which are usually deemed to reinstate public order during a “rowdy” protest. It is, therefore, closely related to police violence, even though the respective factor does not directly reference the clash between protesters and police. To demonstrate this in an example:

But Tory MP Peter Bottomley said: “Officers were fully justified in making sure the crushing stopped. I believe they saved many serious casualties, if not fatalities” (Mackay and Yates, 2004-09-16).

This frame is more in line with previous findings and, when employed, it contributes to legitimizing police action and even violence against protesters (e.g., Kilgo and Harlow, 2019).

The delegitimizing elements of reporting are most prominent in the third frame, troublemakers . Reports employing this frame highlight the violence caused by protesters and associated risks, especially property destruction and risks for public safety. Protesters are deemed responsible for problems and the protest itself is often judged negatively. To give an example:

[Mrs May] defended the police, saying: “I want to be absolutely clear: the blame for the violence lies squarely and solely with those who carried it out. The idea, that some have advanced, that police tactics were to blame when people came armed with sticks, flares, fireworks, stones and snooker balls, is as ridiculous as it is unfair” (Fresco and Ford, 2010).

This frame is one of the most prominently described in the literature on protest coverage. Specifically, it has been labeled “riot” or “violence” frame and is featured in almost all studies about protest coverage (e.g., Gitlin, 1980; Chan and Lee, 1984; McLeod and Hertog, 1999; Kilgo and Harlow, 2019; Kilgo and Mourão, 2021).

The concern motivating many studies about news media coverage of protests is that the media might divert attention away from the substantive content and background of a protest to its form (Coombs et al., 2020). In contrast, two of the identified frames highlight protesters’ messages instead. Cause & grievances does so by explicitly portraying the cause of a protest event as the main topic. The grievances of protesters are given a stage and officials are blamed for causing problems or not addressing them. Again, an example helps to illustrate this:

Why is this government so intent on killing off every institution we have? Small post offices are a lifeline to the elderly who cannot walk the distance to a main post office, and who often don’t have cars, or cannot drive because of failing eyesight (Emerson, 2008).

Some recent studies have shown that this type of coverage is not as uncommon as previously thought (Kilgo et al., 2019; Wouters, 2015).

A more sympathetic coverage is also prevalent in the righteous struggle frame. The frame highlights the accomplishments of the protesters for a cause that is mostly deemed just and protesters receive positive judgement. To give an example:

Terry Deakin […] said the problems had already cost him up to pounds 5,000 in lost business, but he still maintained the protesters had done a ‘‘wonderful job. We have been in support of the drivers,’’ he said. ‘‘It was short, sharp and got the message home” (Griffin et al., 2000).

Yet, articles using this frame provide less background and information about grievances. Instead, they sometimes highlight that the protesters represent what the majority thinks or highlight how peaceful and creative tactics have led to a positive image of a mentioned group. While the frame can thus legitimize a protest group, it does not necessarily provide space to air their grievances.

The frame decay of morals , on the other hand, must be seen as another variation of the patterns described in the protest paradigm literature. The frame highlights the confrontation between protesters and police as well as the decay of morals. It thus falls in line with the notion that the arguments and demands of protesters are neglected in favor of routinized journalistic narratives. The frame also portrays protesters as a fringe group of freaks who disturb the social norms by questioning legitimate authority and disturb the general political consensus. This is best described in an example:

London and Glasgow will echo to the slogans of the morally deluded and the self-consciously caring. Papoose-wearers, manic recyclers, the priggish, the cranky, nudists and Woodcraft Folk will march this Saturday in a cloud of outrage. Peaceniks, marshaled by the Stop the War coalition, claim to march for the majority. They do not (Millen, 2003).

The last frame, nuisance , played only a minor role in the results of the factor analysis. By highlighting the topic and risk nuisance this frame is linked to the kind of reporting described by Di Cicco (2010): it suggests that protest is a bothersome interruption of everyday activities, for example, through causing street closures, traffic jams or noise. This dismisses protest as a legitimate method of voicing a political opinion: the day to day activities of bystanders are deemed more important than the grievance voiced by protesters.

Considering the features of the described frames, it becomes clear that four of the seven frames have close links to the protest paradigm. Specifically, I consider law & order, troublemakers, decay of morals, and nuisance to be variations of the themes mentioned in the protest paradigm literature (McLeod and Hertog, 1999). The frames cause & grievances and righteous struggle, however, should be considered different from the reporting following the paradigm since their usage can be expected to legitimize protest or at least inform the public about specific goals and the background of events. The frame police violence cannot be described as either clearly legitimizing nor delegitimizing. It neither legitimizes a protest by highlighting its cause (it rather deflects from it) but it also does not delegitimize protesters’ actions as it only criticizes the police. In the further discussion, it is, therefore, treated as being outside those two categories of frames. In sum, the results show that the inductively identified frames have considerable overlap with the themes mentioned in the protest paradigm literature. Yet three of the frames depart from the delegitimizing aspect, with two frames being actively legitimizing.

Frame Coding

Figure 2 shows the percent of articles using each frame in each year. The numbers add up to more than 100 percent because many articles contain multiple frames (1.6 on average). To reiterate, law & order, troublemakers, decay of morals, and nuisance are deemed delegitimizing and are linked to the protest paradigm. The analysis thus confirms H1: throughout the selected period, a majority of articles on protest events use delegitimizing framing. Specifically, 60 percent of articles in the dataset contain delegitimizing frames linked to the protest paradigm. Cause & grievances and righteous struggle, which are deemed legitimizing, are only present in 39 percent of the articles.

Figure 2.

Percent of total articles per year containing a frame

Additionally to overall prevalence, H2a-b concern change over time. The most substantial fluctuation can be seen in the police violence frame. It sees a substantial peak in 2009 before it returns to previous levels several years later. This can be explained by an incident already mentioned: at the 2009 G20 summit protests in London, Ian Tomlinson died after being beaten by police, among other incidents of police brutality. A wealth of articles during this time discussed police conduct in reports about the investigation and for many protests in the succeeding years. However, the popularity of this frame fades later in the sample and as for most of the frames there is no clear trend over time.

To formally confirm if there is a significant trend in the frequency of a frame, I use ordinary least squares (OLS) regression. I calculate one an OLS model for each frame, using the year as independent variable (starting with 1992 as year 0) and the percentage of total articles per year containing a frame as the dependent variable. Only in two models does time play a significant role: the ones for cause & grievances and righteous struggle (p < 0.001) (full results in Table E1, Supplemental Information). Especially, the first one gains substantially in importance over time, starting from a low point of 20.2 percent in 1992 and rising to 45.1 percent of all articles about protest in 2016. This means that while H2a is rejected, the frequency of delegitimizing framing stays roughly the same over time, I find support for H2b: legitimizing framing increased in frequency in the analyzed period.

Discussion and Conclusion

Public protest can be a powerful resource for ordinary citizens to participate in the political discourse beyond periodically casting a vote between political actors. In this capacity, protests are crucial to democracy as nearly immediate feedback for representatives in power and as a provider of new issues for the public and political agendas. Provided protesters’ messages are heard and understood. This study contributed to the literature by placing the analysis of protest coverage on empirical data that covers a broader scope and longer time frame than any previous research. This adds a more systematic understanding of how the mainstream news media in the United Kingdom report about protests.

The findings here largely support the existence and continued importance of a protest paradigm, which is thought to drive journalists to use delegitimizing framing as the default in coverage of protest. However, the important contribution here is that it puts the frequency of frames into a comparative perspective: the protest paradigm is apparently not the only force driving protest coverage—otherwise the frames linked to it would be more prevalent over the years. At no point in the time frame (1992–2017) did delegitimizing coverage dominate the coverage of all protest, for example. In fact, two legitimizing frames are found to be present in a steadily growing share of protest reports—while none of the delegitimizing frames gains in frequency over time.

Overall, this is the most important new insight presented here: the use of frames, which highlight the goals and achievements of protests has grown over time. Although this was expected based on the theory and recent empirical findings, the reasons are not so clear yet. Changes in the media landscape during the last decades have led to a situation in which newspapers have to fight for the attention of the audience who have an ever increasing choice of news and entertainment outlets. Perhaps journalists use legitimizing frames more often today because they expect their readers to turn elsewhere to find out about the grievances of a protest. At the same time, information on the goals and messages of protesters are more easily available than ever through the internet, especially social media sites, meaning that fewer resources are needed to include them in a report, while like and share counts indicate which messages are popular with the audience. Protest has also been normalized as a method of political participation over time, meaning the public must be expected to have more first-hand experience now than several decades ago. This means that readers might be more critical of strictly delegitimizing portrayals of events and less opposed to positive coverage of protest.

Nevertheless, coverage following the protest paradigm stays on a constant high level over time. This suggests that journalists’ tendency to highlight the unusual, conflict and controversy, still drive media coverage of protest to a large extent. Is this problematic? Kilgo and Mourão (2021) show that frames linked to the protest paradigm decrease support for and identification with protest movements, just as theorized by, for example, McLeod and Hertog (1999). For many low profile events, only one news item described what happened, often only highlighting the most newsworthy aspect. Similarly, even when several journalists cover an event, it has been shown that some groups and topics systematically receive less favorable coverage (Kilgo and Harlow, 2019). Furthermore, even if multiple perspectives on an event exist, not everyone will be exposed to both delegitimizing and legitimizing frames about every protest.

However, the findings of this study suggest that coverage adhering to the protest paradigm did not dominate the overall discussion at any point. This is true even for people who consume exclusively right-leaning outlets, which are traditionally more adversarial towards protests (Lee, 2014): they as well are regularly exposed to legitimizing framing of protests (see Figure D1a-b, Supplemental Information). This is important, as framing effects research suggests that when multiple interpretations of an issue are available, individuals tend to deliberate and personal values and evaluations of the quality of arguments become more important (e.g., Chong and Druckman, 2007). In other words, if a matter is already contested, preexisting attitudes toward protesters and their grievances are often more important for people’s opinions than media frames of individual stories (Kilgo and Mourão, 2021). With this in mind, the results here show that protest can be a powerful tool for formally powerless groups to make their voices heard. Even through mainstream media. In comparison to the grim conclusions about coverage of protest in the United Kingdom in works such as Halloran et al. (1970) and Murdock (1973), the press appears less hostile towards protesters in the 1990s and even less so in the succeeding decades.

This study is limited in that it only focuses on newspapers. This choice was made as data was available throughout the selected time period making the longitudinal research design possible. Furthermore, the press was, and still is, considered to be an important force in shaping media debates. However, newspapers are on the decline not just in terms of circulations but also in terms of influence on the general audience. Additionally, while the results of this study are generalizable on the U.K. level, the case study design does not permit speaking about general media trends in other countries. Future research could address these limitations by conducting a study spanning several countries and including different types of media.

Supplemental Material

sj-pdf-1-hij-10.1177_19401612221102058 - Supplemental material for Troublemakers in the Streets? A Framing Analysis of Newspaper Coverage of Protests in the UK 1992–2017

Supplemental material, sj-pdf-1-hij-10.1177_19401612221102058 for Troublemakers in the Streets? A Framing Analysis of Newspaper Coverage of Protests in the UK 1992–2017 by Johannes B. Gruber in The International Journal of Press/Politics

Footnotes

Declaration of Conflicting Interests

The authors 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: This work was supported by the University of Glasgow College of Social Sciences PhD Scholarship.

ORCID iD

Johannes B. Gruber

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biography

Johannes B. Gruber (PhD, 2021, at University of Glasgow) is a PostDoc researcher at the Chair for Digital Democracy at the European New School of Digital Studies (ENS), European University Viadrina. His work focuses on media coverage of protest, framing and agenda setting analysis. He specializes in computational approaches to text analysis.

References

Araiza

J. A.

Sturm

H. A.

Istek

Bock

M. A.

. 2016. “Hands Up, Dont Shoot, Whose Side Are You On? Journalists Tweeting the Ferguson Protests.” Cultural Studies – Critical Methodologies 16 (3): 305–312. doi:10.1177/1532708616634834.

Arpan

L. M.

Baker

Lee

Jung

Lorusso

Smith

. 2006. “News Coverage of Social Protests and the Effects of Photographs and Prior Attitudes.” Mass Communication and Society 9 (1): 1–20. 10.1207/s15327825mcs0901_1

Bennett

W. L.

Segerberg

. 2012. “The Logic of Connective Action.” Information, Communication & Society 15 (5): 739–768. doi:10.1080/1369118X.2012.670661.

Benoit

Watanabe

Wang

Nulty

Obeng

Müller

Matsuo

. 2018. “quanteda: An r Package for the Quantitative Analysis of Textual Data.” Journal of Open Source Software 3 (30): 774. doi:10.21105/joss.00774. https://quanteda.io.

Berkowitz

. 1992. “Non-Routine News and Newswork: Exploring a What-a-Story.” Journal of Communication 42 (1): 82–94. doi:10.1111/j.1460-2466.1992.tb00770.x.

Boyle

M. P.

McCluskey

M. R.

Devanathan

Stein

S. E.

McLeod

. 2004. “The Influence of Level of Deviance and Protest Type on Coverage of Social Protest in Wisconsin from 1960 to 1999.” Mass Communication and Society 7 (1): 43–60. 10.1207/s15327825mcs0701_4

Boyle

M. P.

McLeod

D. M.

Armstrong

C. L.

. 2012. “Adherence to the Protest Paradigm: The Influence of Protest Goals and Tactics on News Coverage in U.S. and International Newspapers.” The International Journal of Press/Politics 17 (2): 127–144. doi:10.1177/1940161211433837.

Brown

T. A.

2015. Confirmatory Factor Analysis for Applied Research. Methodology in the social sciences. The Guilford Press. ISBN 978-1-4625-1779-4.

Burscher

Odijk

Vliegenthart

de Rijke

de Vreese

C. H.

. 2014. “Teaching the Computer to Code Frames in News: Comparing Two Supervised Machine Learning Approaches to Frame Analysis.” Communication Methods and Measures 8 (3): 190–206. doi:10.1080/19312458.2014.937527.

10.

Chan

J. M.

Lee

C. C.

. 1984. “Journalistic ‘Paradigms’ of Civil Protests: A Case Study of Hong Kong.” In The News Media in National and International Conflict, edited by A Arno, 183–202, A Westview special study. Westview Press. ISBN 0-86531-776-3.

11.

Chenoweth

Pinckney

Lewis

O. A.

. 2017. Nonviolent and Violent Campaigns and Outcomes dataset. https://www.du.edu/korbel/sie/research/chenow_navco_data.html.

12.

Chong

Druckman

J. N.

. 2007. “Framing Public Opinion in Competitive Democracies.” American Political Science Review 101 (04). doi:10.1017/S0003055407070554.

13.

Clark

Regan

. 2019. Mass mobilization protest data. doi:10.7910/DVN/HTTWYL. https://massmobilization.github.io/.

14.

Coombs

D. S.

Lambert

C. A.

Cassilo

Humphries

. 2020. “Flag on the Play: Colin Kaepernick and the Protest Paradigm.” Howard Journal of Communications 31 (4): 317–336. doi:10.1080/10646175.2019.1567408. https://www-tandfonline-com-s.web.bisu.edu.cn/doi/full/10.1080/10646175.2019.1567408.

15.

Cottle

. 2008. “Reporting Demonstrations: The Changing Media Politics of Dissent.” Media, Culture & Society 30 (6): 853–872. doi:10.1177/0163443708096097.

16.

Dardis

F. E.

2006. “Marginalization Devices in U.S. Press Coverage of Iraq War Protest: A Content Analysis.” Mass Communication and Society 9 (2): 117–135. doi:10.1207/s15327825mcs0902-1.

17.

De Cillia

McCurdy

. 2020. “No Surrender. No Challenge. No Protest Paradigm: A Content Analysis of the Canadian News Media Coverage of the “Yellow Vest Movement” and the “United We Roll Convoy”.” Canadian Review of Sociology/Revue Canadienne De Sociologie 57 (4): 656–680. doi:10.1111/cars.12304. https://onlinelibrary-wiley-com-443.web.bisu.edu.cn/doi/10.1111/cars.12304.

18.

Denny

M. J.

Spirling

. 2018. “Text Preprocessing For Unsupervised Learning: Why It Matters, When It Misleads, And What To Do About It.” Political Analysis 26 (2): 168–189. doi:10.1017/pan.2017.44.

19.

Di Cicco

D. T.

2010. “The Public Nuisance Paradigm: Changes in Mass Media Coverage of Political Protest Since the 1960s.” Journalism & Mass Communication Quarterly 87 (1): 135–153. doi:10.1177/107769901008700108.

20.

Emerson

. 2008. Why i’ve become a Placard Waver! Daily Mail, (2008-03-04).

21.

Entman

R. M.

1993. “Framing: Toward Clarification of a Fractured Paradigm.” Journal of Communication 43 (4): 51–58. doi: 10.1111/j.1460-2466.1993.tb01304.x

22.

Fresco

Ford

. 2010. “Police ’snatch squads’ Ready to Arrest Violent Demonstrators.” The Times, (2010-12-14).

23.

Gil-Lopez

. 2020. “Mainstream Protest Reporting in the Contemporary Media Environment: Exploring (In)Stability and Adherence to Protest Paradigm From 1998 to 2017.” Journalism & Mass Communication Quarterly 98 (3): 692–724. doi:10.1177/1077699020984783.

24.

Gitlin

. 1980. The Whole World is Watching: Mass Media in the Making & Unmaking of the New Left. University of California Press. ISBN 978-0-520-04024-3.

25.

Graesser

A. C.

McNamara

D. S.

Louwerse

M. M.

. 2011. “Methods of Automated Text Analysis.” In: Handbook of Reading Research, edited by M. L. Kamil, P. D. Pearson, E. B. Moje and P. P. Afflerbach, 34–53. Routledge. ISBN 978-0-8058-5342-1.

26.

Griffin

Wilson

Chrisafis

. 2000. “Fuel Crisis.” The Guardian, 2000-09-16.

27.

Grimmer

Stewart

B. M.

. 2013. “Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts.” Political Analysis 21 (3): 267–297. doi:10.1093/pan/mps028.

28.

Gruber

. 2021. LexisNexisTools. An R Package for Working with Newspaper Data from ’LexisNexis’. https://github.com/JBGruber/LexisNexisTools. R package version 0.3.3.

29.

Gruber

J. B.

2021. Troublemakers in the Streets? A Framing Analysis of Newspaper Coverage of Protests in the UK 1992-2017 . University of Glasgow. doi:10.5525/gla.thesis.82346.

30.

Halloran

J. D.

Elliott

P. R. C.

Murdock

. 1970. Demonstrations and Communication: A Case Study. Penguin special. Penguin. ISBN 978-0-14-052282-2.

31.

Harlow

Johnson

T. J.

. 2011. “The Arab Spring: Overthrowing the Protest Paradigm? How The New York Times, Global Voices and Twitter Covered the Egyptian Revolution.” International Journal of Communication 5 (0): 16. doi:http://ijoc.org/index.php/ijoc/article/download/1239/611.

32.

Harlow

Kilgo

D. K.

Salaverría

García-Perdomo

. 2020. “Is the Whole World Watching? Building a Typology of Protest Coverage on Social Media From Around the World.” Journalism Studies 21 (11): 1590–1608. doi:10.1080/1461670X.2020.1776144.

33.

Jha

. 2007. “Exploring Internet Influence on the Coverage of Social Protest. Content Analysis Comparing Protest Coverage in 1967 and 1999.” Journalism & Mass Communication Quarterly 84 (1): 40–57. doi:10.1177/107769900708400104.

34.

Kilgo

D. K.

Harlow

. 2019. “Protests, Media Coverage, and a Hierarchy of Social Struggle.” The International Journal of Press/Politics 24 (4): 508–530. doi:10.1177/1940161219853517. http://journals.sagepub.com/doi/10.1177/1940161219853517.

35.

Kilgo

D. K.

Mourão

R. R.

. 2021. “Protest Coverage Matters: How Media Framing and Visual Communication Affects Support for Black Civil Rights Protests.” Mass Communication and Society 24 (4): 576–596. doi:10.1080/15205436.2021.1884724. https://www-tandfonline-com-s.web.bisu.edu.cn/doi/full/10.1080/15205436.2021.1884724.

36.

Kilgo

D. K.

Mourão

R. R.

Sylvie

. 2019. “Martin to Brown: How Time and Platform Impact Coverage of the Black Lives Matter Movement.” Journalism Practice 13 (4): 413–430. doi:10.1080/17512786.2018.1507680. https://www-tandfonline-com-s.web.bisu.edu.cn/doi/full/10.1080/17512786.2018.1507680.

37.

Kim

Shahin

. 2020. “Ideological Parallelism: Toward a Transnational Understanding of the Protest Paradigm.” Social Movement Studies 19 (4): 391–407. doi:10.1080/14742837.2019.1681956. https://www-tandfonline-com-s.web.bisu.edu.cn/doi/full/10.1080/14742837.2019.1681956.

38.

Kleut

Milojevic

. 2021. “Framing Protest in Online News and Readers’ Comments: The Case of Serbian Protest “Against Dictatorship”.” International Journal of Communication 15 (0), https://ijoc.org/index.php/ijoc/article/view/13815.

39.

Kriesi

Wüest

Lorenzini

Makarov

Enggist

Rothenhäusler

Kurer

Häusermann

Wangen

Altiparmakis

Borbáth

Bremer

Gessler

Hunger

Hutter

Schulte-Cloos

Wang

. 2020. Poldem-protest dataset 30 European countries. https://poldem.eui.eu/download/protest-events/.

40.

Kuhn

T. S.

1970. The Structure of Scientific Revolutions. Chicago Univ. Press. ISBN 0-226-45804-0.

41.

Kyriakidou

Olivas Osuna

. 2017. “The Indignados Protests in the Spanish and Greek Press: Moving beyond the ‘protest paradigm’?” European Journal of Communication 32 (5): 457–472. doi:10.1177/0267323117720342.

42.

Langer

. 1998. Tabloid Television: Popular Journalism and the ∖dqother News∖dq. Communication and Society. Routledge. ISBN 978-0-415-06637-2.

43.

Lee

F. L. F

. 2014. “Triggering the Protest Paradigm: Examining Factors Affecting News Coverage of Protests.” International Journal of Communication 8 (1): 2725–2746.

44.

Mackay

Yates

. 2004. “Police hit as hunt yobs hurl fireworks in riot.” Daily Mirror, (2004-09-16).

45.

Matthes

. 2009. “What’s in a Frame? A Content Analysis of Media Framing Studies in the World’s Leading Communication Journals, 1990-2005.” Journalism & Mass Communication Quarterly 86 (2): 349–367. doi:10.1177/107769900908600206.

46.

Matthes

Kohring

. 2008. “The Content Analysis of Media Frames: Toward Improving Reliability and Validity.” Journal of Communication 58 (2): 258–279. doi:10.1111/j.1460-2466.2008.00384.x.

47.

McCurdy

. 2012. “Social Movements, Protest and Mainstream Media.” Sociology Compass 6 (3): 244–255. doi:10.1111/j.1751-9020.2011.00448.x.

48.

McKay

, ed. 1998. DiY Culture: Party & Protest in Nineties Britain. Verso. ISBN 978-1-85984-260-7.

49.

McLeod

D. M.

Hertog

J. K.

. 1992. “The Manufacture of ‘Public Opinion’ by Reporters: Informal Cues for Public Perceptions of Protest Groups.” Discourse & Society 3 (3): 259–275. doi:10.1177/0957926592003003001.

50.

McLeod

D. M.

Hertog

J. K.

. 1999. “Social Control, Social Change and the Mass Media’s Role in the Regulation of Protest Groups.” In Mass Media, Social Control, and Social Change, edited by D. Demers and K Viswanath, 305–330. Iowa State Univ. Pr. ISBN 0-8138-2682-9.

51.

Millen

. 2003. “March for Anything, Anywhere.” The Times, (2003-02-12).

52.

Milne

. 2005. Manufacturing Dissent: Single-issue Protest, the Public and the Press. Demos. ISBN 1-84180-141-0.

53.

Mourão

R. R.

Kilgo

D. K.

Sylvie

. 2021. “Framing Ferguson: The Interplay of Advocacy and Journalistic Frames in Local and National Newspaper Coverage of Michael Brown.” Journalism 8 : 320–340. doi:10.1177/1464884918778722. http://journals.sagepub.com/doi/10.1177/1464884918778722.

54.

Murdock

. 1973. “Political Deviance: The Press Presentation of a Militant Mass Demonstration.” In The Manufacture of News, edited by S. Cohen and J. Young, 156–175. Communication and society. Constable. ISBN 978-0-09-459490-6.

55.

Norris

. 2002. Democratic Phoenix: Reinventing Political Activism. Cambridge University Press. ISBN 978–0521010535.

56.

Parmelee

J. H.

2013. “The Agenda-building Function of Political Tweets.” New Media & Society 16 (3): 434–450. doi:10.1177/1461444813487955.

57.

Revelle

. 2019. psych: Procedures for Psychological, Psychometric, and Personality Research . Northwestern University. https://CRAN.R-project.org/package=psych. R package version 1.9.12.

58.

Shahin

Zheng

Sturm

H. A.

Fadnis

. 2016. “Protesting the Paradigm: A Comparative Study of News Coverage of Protests in Brazil, China, and India.” The International Journal of Press/Politics 21 (2): 143–164. doi:10.1177/1940161216631114.

59.

Tuchman

. 1973. “Making News by Doing Work: Routinizing the Unexpected.” American Journal of Sociology 79 (1): 110–131.10.1086/225510

60.

van Aelst

Strömbäck

Aalberg

Esser

Vreese

Matthes

Hopmann

Salgado

Hubé

Ste¸pińska

Papathanassopoulos

Berganza

Legnante

Reinemann

Sheafer

Stanyer

. 2017. “Political Communication in a High-choice Media Environment: A Challenge for Democracy?” Annals of the International Communication Association 41 (1): 3–27. doi:10.1080/23808985.2017.1288551.

61.

Wasow

. 2020. “Agenda Seeding: How 1960s Black Protests Moved Elites, Public Opinion and Voting.” American Political Science Review 114 (3): 638–659. doi:10.1017/S000305542000009X.

62.

Wouters

. 2015. “Reporting Demonstrations: On Episodic and Thematic Coverage of Protest Events in Belgian Television News.” Political Communication 32 (3): 475–496. doi:10.1080/10584609.2014.958257.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.37 MB