Abstract
This article explores the role played by legacy media in the pervasive situation where far-right activists highly politicize migrant crimes on social media. Focusing on particular crimes they consider indicative of a state’s failure to secure borders and protect its citizens, they utilize social media platforms to mobilize and garner political and media attention. However, little is known about the dynamics of influence between far-right activism on social media and legacy media in the process of politicizing migrant crimes. To overcome the analytical and methodological limitations of current research, we have developed a computational method to capture the public discussion that took place in France around a single event – the murder of a young teenager in Paris in October 2022. Applying an event identification algorithm to a vast corpus of Twitter data has enabled us to analyze the dynamics of public discussion held in various arenas, both quantitatively and qualitatively. This article’s main argument is that legacy media act as a “moral compass” vis-à-vis far-right mobilizations that politicize migrant crimes on social media. While they were unable to influence the volume of attention that social media users pay to this murder, the French legacy media have contributed to setting limits to the form and spread of politicization by the far right on social media. More specifically, they were able to influence the attention paid to the facts that serve as the basis for discussion on social media and to steer that discussion toward questioning the relevance of politicizing the crime.
Introduction
On the evening of October 14, 2022, the police found the dead, dismembered body of Lola Daviet, a twelve-year-old girl, in a district of eastern Paris. Two days later, when a newspaper mentioned the Algerian nationality of the main suspect, hundreds of far-right activists started mobilizing massively on Twitter and in the streets, organizing marches and putting up posters. Using the hashtag #justicepourlola, they argued that this was not an isolated incident but the consequence of lax migration policies. The violent murder of this young white girl with blond hair, they claimed, constitutes a “francocide.” This neologism was coined a few months before the murder by French far-right politician Eric Zemmour: “The beating, rape, murder, and stabbing of a French man or woman by an immigrant is not a news item. It is a political fact that I will henceforth refer to as ‘francocide’.” he declared during a speech to his supporters in September 2022 (Fort 2022). The murder occupied the media spotlight for more than a week and led to a series of statements by far-right politicians, ministers, and politicians on both the right and the left.
This article explores the role played by legacy media in the pervasive situation where far-right activists politicize migrant crimes on social media. By “politicization,” we refer to the process through which a specific news event is viewed not just as an isolated incident or a result of the psychological traits of the individuals involved but also as indicative of a more general issue (Boltanski and Esquerre 2022). In this sense, crimes that have been committed, actually or allegedly, by immigrants or people of foreign origin are the object of intense politicization in most Western countries. This politicization is prevalent on social media platforms, which are commonly used by far-right politicians and activists in North America and Europe to associate immigration with crime and to denounce the failure of states to secure borders and protect citizens (Ekman 2019, Merry 2022). Far-right activists have leveraged social media to mobilize and gain political and media attention in ways that were impossible until the early 2010s. By increasing the visibility of specific crime news that aligns with their political agenda and reframing interpretations presented by established media and public officials, they create “affective publics” of individuals sharing racist and hateful comments (Ekman 2019). These efforts aim to sway public opinion and improve their electoral outcomes.
In this article, we ask whether and how legacy media are likely to influence the far-right politicization of migrant crimes within social media platforms. By legacy media, we refer to news outlets that have been established for several decades and operate primarily through traditional avenues, such as print, television, or radio, as well as digital ones, including websites, applications, and social media. When exploring the research on power dynamics between social media activism and news media, we uncover two opposing perspectives on this issue. According to the first one, legacy media would unlikely influence the politicization of migrant crimes carried out by far-right activists on social media because of the significant influence of platforms on news judgment and practices (McGregor and Molyneux 2020; Pignard-Cheynel and Amigo 2019; Tandoc and Vos 2016) in a context of a declining control over the selection and interpretation of news events (Anderson 2013; Vos and Heinderyckx 2015). The opposing argument is that legacy media can regulate mobilizations on social media, as they have maintained their agenda-setting power despite the evolving media landscape, particularly in Europe (Djerf-Pierre and Shehata 2017; Langer and Gruber 2021).
Unlike most agenda-setting research (Barberá et al. 2019; Conway-Silva et al. 2018; Gilardi et al. 2022; Wang et al. 2023), which overlooks the role of events in studying the influence of news media and social media, we investigated in-depth a single news event that took place in France in October 2022: the murder of the young girl mentioned in the first lines of this article. We selected this case because it was the subject of extensive media coverage and an intense far-right campaign on Twitter – this platform, now called X, was not yet owned by Elon Musk at the height of the debate surrounding the murder. More importantly, this case brought to light how a wide variety of actors (activists, politicians, journalists, news organizations, citizens, etc.) contribute to politicizing a migrant crime. To analyze this politicization, we have developed an original computational method that applies an event identification algorithm to a vast corpus of Twitter data. Combining quantitative and qualitative analyses, we have reconstructed the process of politicization of this murder in unprecedented detail.
This article’s main argument is that, in the case of the murder studied in the French context, legacy media play the role of a “moral compass” vis-à-vis far-right mobilizations that politicize migrant crimes on social media. While they were unable to influence the volume of attention that social media users pay to this murder, the French-established news media have contributed to setting limits to the form and spread of politicization by the far right on social media. More specifically, they were able to influence the attention paid to the facts that serve as the basis for discussion on social media and to steer that discussion toward questioning the relevance of politicizing the crime. While the far-right activists politicized the event to advance their anti-immigration agenda and criticize legacy media, legacy media assembled most of the facts that served as the basis for the discussion on Twitter. Additionally, these news organizations influenced public discourse on Twitter by scrutinizing the event’s politicization, highlighting how it was politicized, and helping to question the legitimacy of that process. Beyond the French case, this study highlights the role played by the legacy media in regulating far-right mobilizations that exploit news events to gain ground in public opinion.
The Power Dynamics Between Social Media Activists and Legacy Media
Over the last two decades, immigration has been increasingly portrayed in the North American and European news media as a threat, particularly associated with violence (Benson 2013; Famulari and Major 2023; Harris and Gruenewald 2020; Sohoni and Sohoni 2014). On both sides of the Atlantic, major news media tend to politicize crimes attributed to immigrants, which are associated with the failure of states to protect their citizens (Harraway and Wong 2024). In the United States, such a generalization can be found primarily in right-leaning media, while left-wing media tend to emphasize the victimization suffered by immigrants (Famulari and Major 2023). Much research suggests that the politicization of immigration in the news media is a consequence of anti-media populist movements. European media outlets provide more coverage of far-right protest events when they address immigration issues, mobilize contentious forms of action, and incite debates between political opponents (Castelli Gattinara and Froio 2024). While journalists rarely defend the association between immigration and crime, they use quotes from politicians and ordinary citizens (Harraway and Wong 2024). Under the influence of anti-immigration movements, they tend to mention the ethnic origin of crime suspects, even in countries like Germany, where this practice was previously seen as conflicting with journalistic ethics (Klimmt et al. 2024).
However, little is known about the dynamics of influence between far-right activism on social media and legacy media in the process of politicizing migrant crimes. Researchers have documented the growing importance of social media in generalizing the link between immigration and crime. Utilizing commercial platforms, far-right activists aim to push the boundaries of what is considered acceptable to say in public. They use emotional discourse to dehumanize migrants and assert that their culture explains their criminal behavior (Ekman 2019). Inspiring other politicians around the world, Donald Trump has used social media to portray immigrants as criminals, express sympathy for victims, and portray Democrats as being responsible for the rise in crime. Combined with racist stereotypes, these narratives have generated a lot of attention on Twitter (Merry 2022). Yet, little research has been devoted to the mutual influence between social media mobilization and legacy media regarding migrant crime, and even less on the possible impact that legacy media might have on how migrant crimes are politicized on social media.
We identified two contrasting perspectives by examining existing literature on the power dynamics between social media activism and news organizations. The first one highlights the influence of social media mobilizations on legacy media. Scholars in journalism studies and political communication have documented the critical role that platforms play in the news production process. Twitter, in particular, has become a routine practice for most journalists to monitor their competition and discover new news sources and topics (Lasorsa et al. 2012; Pignard-Cheynel and Amigo 2019; Tandoc and Vos 2016). Evidence suggests that an increasing number of journalists consider social media signals as reflecting the preferences of news consumers or, more broadly, the state of opinion on a particular topic or event (McGregor 2019; Tandoc and Vos 2016; Van Dijck et al. 2018). The heavy use of Twitter appears to influence journalistic judgment, as American journalists who use Twitter most frequently assign similar journalistic value to tweets as they do to agency dispatches (McGregor and Molyneux 2020). In this context, mobilizations can be expected to influence journalists when they gain visibility on platforms or exploit the intense media interest in a particular event. This is all the more likely as the gatekeeping role of these media organizations has been profoundly called into question by the rise of the web and social media (Anderson 2013; Vos and Heinderyckx 2015).
A second perspective highlights the influence of legacy media on social media mobilizations. This view is not incompatible with the first, as legacy media can retain significant agenda-setting power by selecting the social media inputs they circulate or engage with. However, it emphasizes that legacy media have substantial resources to distance themselves from, if not ignore, social media mobilizations related to a particular event. Several studies have shown that legacy media have maintained their centrality in the evolving media landscape (Cointet, Cardon, et al. 2021; Nielsen 2016). In France, far-right websites within the “counter-information” space, which disseminate a significant amount of misinformation and conspiracy theories, frequently cite mainstream media, whereas the opposite is rare (Cointet, Cardon, et al. 2021). In other European countries, populist anti-immigration movements often cite mainstream media while harshly criticizing them (Haller and Holt 2019). More decisively, agenda-setting research indicates that established media still significantly influence the issues that citizens consider most important in the public debate. As documented in several European countries, traditional news media have retained their agenda-setting power despite the changing media environment. Swedish citizens appear as responsive to issue signals from the collective media agenda today as in the early 1990s (Djerf-Pierre and Shehata 2017). British legacy media still have the power to bring specific issues to the forefront of public attention, not just for a few days but also for several weeks or even months: they can initiate, amplify, and sustain attention on a particular issue thanks to their symbolic, human, and economic resources (Langer and Gruber 2021). This indicates that legacy media are likely to regulate the impact of social media mobilizations due to their power to shape public attention.
Three Paths for Legacy Media to Regulate Social Media Mobilization
When considering the regulatory power legacy media can have on social media mobilizations, there are three possible paths: regulation by attention, facts, and reflexivity. First, legacy media can intervene by limiting coverage of the crime event that sparks activist outrage on social media. Established media play a crucial role in mobilizing public awareness, reaching audiences far beyond those achieved by activists on social media (Langer and Gruber 2021; Margetts et al. 2015). Consequently, legacy media can limit their influence by directing public attention to other topics or events. Yet, this path is not straightforward for several reasons. First, populist movements often criticize mainstream media for not covering events that contradict their supposed ideology (Waisbord 2018). Additionally, if less radical politicians or partisan media relay the politicization of the crime, its journalistic value increases, heightening the pressure to cover the event (Benkler et al. 2018).
Instead of relying on the media’s ability to influence what people discuss (first-level agenda setting), the two other paths draw on their ability to influence how the discussion is framed (second-level agenda setting) (McCombs et al. 1997). According to the second path, legacy media, which routinely assemble, certify, and verify facts, have the means to influence the facts that social media users pay attention to. They can achieve this through fact-checking, the effectiveness of which is debated (Walter et al. 2020), and by drawing public attention to facts they consider essential regarding a particular event. Because of their large audience, they can guide the attention of social media users toward the facts they have assembled. Even in the context of widespread fake news, such as the 2016 U.S. presidential election, most people’s news exposure came from mainstream media outlets, while only 1 percent of Twitter accounts accounted for nearly 80 percent of fake news source exposure (Grinberg et al. 2019).
Finally, legacy media can limit the form and spread of far-right politicization on social media by bringing greater reflexivity to the conversation. By reflexivity, we refer to a specific form of framing a news event, which consists of distancing and questioning the relevance of the generalization made by activists. We draw here on Luc Boltanski’s theoretical framework, which emphasizes that denunciations made by individuals must comply with a set of norms to be recognized as normal or acceptable (Boltanski 2012). In other words, social actors can demonstrate reflexive capacities, questioning the relevance of the generalization made from a particular event. Specifically, legacy media play a specific role by providing reflexive coverage of how social media activists politicize a crime event. They can achieve this by expressing or conveying favorable or critical perspectives on the politicization of the crime. Additionally, they can rely on well-established professional standards, such as providing balanced coverage of all arguments for and against the crime’s politicization (Lemieux 2000: 371–4; Schudson 1982), investigating the mobilization that underpins the politicization or, in the spirit of contextual journalism (Fink and Schudson, 2014), revealing the political strategies that fuel the politicization of a crime.
The Career Line of Public Events in the Age of Social Media
Our analysis of the politicization of a migrant crime on Twitter is based on a revised version of Harvey Molotch and Marilyn Lester’s classical theory of public events. Developed in the 1970s, it describes how an occurrence, defined as “any cognized happening,” is transformed into a public event. Named the “career lines of public events,” this process relies on “a set of agencies (individuals or groups), each of which helps construct, through a distinctive set of organizational routines, what the event will have turned out to be” (Molotch and Lester 1974). The authors have distinguished four significant agencies: the “effectors,” that is, individuals who are directly involved in the occurrence; the “promoters,” who identify an occurrence, for some reason, as being worthy of interest to others; the “assemblers,” that is, the news professionals, who transform the materials supplied by the promoters into public events through publication or broadcast; and finally the “consumers,” who pay attention to certain occurrences that the media bring to their attention.
The rise of social media calls for a revision of this theory. Today, platforms enable the same actors to perform multiple functions simultaneously: consumers can act as assemblers when they publish materials such as photos or videos; politicians can assemble occurrences without going through journalists; journalists can act as assemblers and promoters by highlighting certain occurrences on social media, and so on. Additionally, we need to include two new agencies in the original typology, which we propose to name “connectors” and “interpreters.” First, as commonly used in literature (Badham et al. 2022), connectors make an occurrence highly visible on a social media platform, thanks to their high number of followers. Research on social media, particularly Twitter, has shown that the largest propagations are generated by users with the most followers, even if neither the number of followers nor network centrality predicts influence (Bakshy et al. 2011; Cha et al. 2010; Watts and Dodds 2007). Consequently, even if there is no certainty about their impact, media outlets, journalists, politicians, activists, or individuals with high social media profiles can use their centrality in the network to generate discussion around an occurrence. Second, “interpreters” are users who contribute on social media platforms around an event previously made visible by a connector. By liking, sharing, or commenting on the event, interpreters can increase the visibility of existing interpretations or propose their own, which can generate “firestorms” (Pfeffer et al. 2014). Table 1 summarizes the various agencies, their roles, and the resources they require in terms of skills, disposition, or social capital.
The Career Line of Public Events in the Age of Social Media.
Case and Methodology
We focused on the murder of a twelve-year-old girl in Paris on October 14, 2022. Two days later, a spokesman for far-right politician Eric Zemmour urged mobilization on Twitter after learning from the press that the suspect was an Algerian woman. The next day, as mainstream media announced that the suspect was illegally on French soil, Eric Zemmour condemned a “francocide,” a term he created a few months earlier. His supporters urged for a “demonstration for Lola.” On October 18, several conservative right-wing leaders joined far-right leaders in condemning the government’s laxity. “If the obligation to leave the country had been properly applied, little Lola would still be alive,” said a conservative right-wing leader on public television. For a week, the drama emerged as a central topic in French political debate, while Lola’s parents expressed their opposition to any political exploitation of the tragedy. Government ministers defended themselves against the accusations of being responsible for the murder, while most news media covered the crime in the following days. A media explosion ensued (Boydstun 2013), culminating on October 19.
This case is ideal for studying the role played by legacy media in the far-right politicization of migrant crime on social media. More than any other migrant crime committed in France that year or in subsequent years, this murder highlights the characteristics that strongly support the politicization of migrant crimes by the far right: the victim was young and vulnerable, she endured an exceptionally violent crime, and the main suspect was North African and in an irregular situation in France. The far-right’s mobilization on Twitter (now known as X) in response to this murder was both intense and prolonged, with far-right activists using it as an argument during and after the 2024 French legislative elections. The volume of discussions about this murder on Twitter was unprecedented in 2022 (see Supplemental Information File, Figure S8), as was the coverage by legacy media (e.g., Le Figaro published 136 articles about the murder over a ten-day period) and various partisan and nonpartisan outlets.
Event Detection
To reconstruct the dynamics of the discussion that followed the murder, we developed and applied an event detection algorithm to a Twitter corpus. We relied on a corpus of tweets collected in real time during October and November 2022 via the Twitter API, using software designed to ensure stable collections over extended periods (Ooghe-Tabanou et al. 2023). We have collected 155 million French tweets that contain an external reference, including those with URLs, images, videos, GIFs, or tweets that quote another tweet. To reduce the corpus size, we selected tweets that had received at least one retweet, obtaining 10.3 million tweets that generated 99.1 million retweets.
We then applied an event detection algorithm that considers the documents’ temporal dimension and ensures a fine granularity of the detected clusters. More precisely, the algorithm works as follows:
Tweets are processed in chronological order, from oldest to most recent.
Each new tweet is transformed into a vector of words using the classical tf-idf measure as term weights (Sparck Jones 1972).
We then calculate the cosine distance of the vector to all vectors of previous tweets within a twelve-hour time window.
If a neighboring vector exists at a sufficiently close distance (this minimum distance is a parameter of the algorithm; we have set it at 0.7), the tweet is grouped in the same cluster as this neighbor.
Otherwise, a new cluster containing the tweet will be created.
The time window ensures that we obtain groups of tweets that have received continuous attention from Twitter users: if no tweet on a given subject is published for twelve hours, the corresponding cluster stops. The other specificity of this algorithm is that it does not fix the number of clusters in advance, nor does it necessarily group tweets that lack sufficient similarity. This results are in very homogeneous clusters, sometimes as small as one tweet. A more detailed explanation of how the algorithm works, as well as a justification of the choice of the vectorization technique – tf-idf, a classic information retrieval method, as opposed to recent neural network embedding techniques such as Sentence-BERT (Reimers and Gurevych 2019) is provided in Mazoyer et al. (2024).
We then selected all clusters containing the word “Lola,” resulting in 19,047 clusters and 449,130 tweets. We then worked on a sample of 301 clusters with the highest number of tweets containing “Lola,” which accounts for 20 percent of the tweets in the corpus (91,995 original tweets) (The robustness of this selection procedure is tested in the Supplemental Information File, Tables S6 and S7). A manual check revealed that 77 percent of these clusters (231 out of 301) were consistent, relevant, and distinct: they are primarily organized around an initial occurrence that is nonredundant with another cluster and related to Lola’s murder. An initial occurrence may consist of a reported fact, a statement, or an analysis, followed by messages that mention or comment on it. (The first ten tweets of a particular cluster are reproduced in the Supplemental Information File, Table S8.) Since each cluster in the sample unfolds from a single, consistent, and relevant initial occurrence, we will refer to each cluster as a single “sub-event” for the rest of the article.
Through qualitative analysis, we coded every subevent of the sample. We relied on several categories based on the revised version of Molotch and Lester’s model, including “effector,” “connector,” and “assembler.” Additionally, we identified new categories inductively to describe the nature and focus of the initial occurrence, as described below.
Nature of the initial occurrence: reported fact, reported statement (a quote from someone other than the user who shares it), statement (a quote whose author is the person sharing it), or analysis (an interpretation that suggests a cause, connecting the crime to other events to draw a general conclusion). Each category is exclusive.
Focus of the initial occurrence: does it relate to the murder (referring to the crime, its protagonists, its causes, or those responsible for it), or does it relate to the politicization of the murder (documenting the mobilizations following the murder, analyzing the political competition underlying the discussion, pushing an immigration-related interpretation, and criticizing or defending this politicization)?;
Effector: the person or organization involved in the happening of the occurrence;
Connector: a person or an organization that was the first to disseminate the occurrence on Twitter, generating a significant number of retweets;
Assembler: the media outlet or any other actor who assembled the story behind the initial occurrence.
Table 2 presents several subevents that have been categorized accordingly. Each unfolds from specific occurrences and involves various actors as effectors, connectors, and assemblers. Some take only a few hours, while others can last several days. Initial occurrences have been assembled by news organizations or nonjournalists, such as politicians, activists, or other users. The news outlets may be legacy or nonlegacy media, the former having operated for several decades and enjoying a solid reputation within the journalistic profession.
A Few Subevents on Twitter Following Lola’s Murder.
Furthermore, as shown in Table 3, statements account for the most initial occurrences (55.8 percent) in the sample of 231 most discussed subevents, but reported facts and statements are also quite significant (36.8 percent). Initial occurrences are evenly divided between those relating to the crime itself (referring to the crime, its protagonists, its causes, or those responsible for it, 45 percent) and those questioning or justifying the politicization of the crime in terms of immigration. Legacy media play a significant role as assemblers and connectors for subevents, followed by partisan media that are predominantly far-right (see Supplemental Information File, Table S10).
Main Characteristics of the Sample’s Subevents.
Media Corpus
We collected articles from fifty-five French national and local legacy newspapers and news magazines. We retrieved 667 articles from the Europresse database, published between October 15 and 24. These well-established media outlets, in operation for decades, are fairly distributed across the political spectrum, excluding the far-right and far-left (see Supplemental Information File, Table S9). Coverage rises sharply from October 17, then falls after five days, before rising again on October 24, the day of the teenager’s funeral (see Figure 1).
Temporal distribution of news articles related to Lola’s murder.
We manually coded this corpus to categorize the stance of each article toward the politicization of the murder. 1 After thoroughly reading a sample of articles, we identified their primary focus. Whether they concentrate on the facts of the murder, convey the politicization driven by the far right (referred to as “francocide”) and parts of the right (emphasizing the poor rate of deporting illegal immigrants), or question the legitimacy of politicizing the murder by offering a more reflexive coverage.
As previously stated, reflexive coverage can take several forms. We have identified five based on the literature and an in-depth reading of the media corpus. The first two relate to expressing views that are either critical of or in favor of politicizing the murder in connection with immigration:
Criticism of politicization, either by the journalist or by the sources that are quoted in the article;
Defense of immigration-related politicization of the murder, involving comparisons with politicizations of other news events (e.g., femicide or police violence) or responses to criticism of the politicization.
The following forms are based on well-established professional standards:
“He said, she said” coverage through equal presentation of arguments for and against the immigration-related politicization of the murder. This is the simplest form of reflexivity, based on journalistic professional ethics (Lemieux 2000: 371–4; Schudson 1982);
Documenting politicization, either by investigating how far-right activists exploited the murder, by reporting the consequences of their mobilization, or by producing facts that call into question the politicization of the murder 2 ;
Political competition coverage by presenting the murder’s politicization as a strategy of the far right and the right wing to put pressure on the government (Fink and Schudson 2014).
Interpreters’ Political Leaning
Lastly, we automatically identified the political leaning of interpreters, that is, individuals who contributed on Twitter to the discussion of the subevents surrounding Lola’s murder. Applying a well-established method (Barberá 2015; Budak and Watts 2015; Cointet, Morales, et al. 2021), we used as a proxy for partisanship the fact of having retweeted at least one tweet from one of the twelve major candidates in the 2022 French presidential election during the six months preceding Lola’s murder. (For an evaluation of this method, see Supplemental Information File, Table S11.) Individuals who did not retweet any tweet from these twelve candidates are considered nonpartisan (n = 60,005, see Supplemental Information File, Table S12). We aggregated the candidates into five categories of political orientation (far-left, left, pro-government, right-wing, and far-right). We then coded the political orientation of interpreters based on the majority (more than 50 percent) of their retweets of candidates. Users who retweeted an equal number of tweets between political groups were left uncoded.
Although most interpreters who contributed to the discussion of Lola’s murder on Twitter were predominantly nonpartisan, far-right interpreters were the most active, with 25 percent of users responsible for 44.2 percent of tweets and 63.4 percent of retweets (see Supplemental Information File, Table S12). Far-left interpreters came next, dominated by supporters of La France Insoumise (12.1 percent), considerably less active, and government supporters (3.2 percent). Left-wing and right-wing partisans are virtually absent from the conversation. The temporal distribution of these contributions is shown in Figure 2, along with a few key events.
Temporal distribution of tweets and retweets according to users’ political orientation.
Results
Social Media Mobilization did not Generate Legacy Media Coverage
The first finding is based on a quantitative analysis of the press corpus and a qualitative study of the subevents reconstructed from Twitter. It states that although legacy media paid considerable attention to the murder, it cannot be viewed as a direct consequence of far-right activism on social media. This was due instead to the intervention of established political actors on the right of the political spectrum, who are active in the French parliament and have a strong visibility in the mainstream media. They acted as brokers in the public sphere, legitimizing the politicization of murder in connection to immigration.
During the first three days following the discovery of the body (October 15–17), the legacy media increasingly covered the murder (Figure 1). These included regional newspapers that usually pay attention to crime news (Le Parisien) and leading national newspapers that rarely cover crime (Libération, Le Monde, Le Figaro, L’Obs, etc.). During this period, only far-right activists and politicians such as Eric Zemmour, Jordan Bardella, and Marine Le Pen voiced their opinions about the murder – only on Twitter and not in the legacy media. As an illustration, it was only on October 19 that Eric Zemmour was invited to comment on the murder on BFMTV, a central TV news channel.
However, we claim that far-right activism on Twitter did not generate legacy media coverage. Several elements suggest this. First, their coverage then focuses exclusively on the factual aspects of the crime: the disappearance of the teenager, the parents’ waiting, the discovery of the body, the arrest of the suspect, the autopsy, and the reactions of neighbors. No reference was made to the interpretative frames carried by the far-right, and no far-right personality was interviewed or even mentioned during that period. On October 16, Le Parisien revealed that the suspect was Algerian, which was subsequently mentioned in many articles until the following day when it was revealed that she was under an obligation to leave French territory. However, until October 17, the mention of the suspect’s nationality was always made strictly factual, without it being framed as a problem. Instead, it seems to be the nature of the crime itself that is attracting media interest: an extremely violent murder of a teenage girl in the center of Paris – all properties that have long been identified as generating media attention (Katz 1987; Roshier 1973).
Qualitative analysis of subevents shows that the context shifted abruptly on October 18. The previous night, BFMTV, a central TV News channel, revealed that the suspect was under an obligation to leave the country. On October 18, several right-wing politicians gave speeches in the Parliament and appeared on television, denouncing the government’s responsibility for the murder. “Lola lost her life because you did not deport this foreign woman, who no longer had any place here,” said right-wing MP Éric Pauget in an address to the Minister of Justice. “If this obligation to leave French territory had been applied, Lola would still be alive,” claimed David Lisnard, a prominent member of the same party, in a popular TV program. The politicization of Lola’s murder was then no longer confined to far-right activists and politicians, who were expressing themselves mainly on Twitter and in partisan media. From then on, far-right politicians were interviewed on legacy media, while right-wing politicians claimed that Lola’s murder revealed the government’s inability to control immigration. The politicization of the murder then involves the most legitimate arenas of public debate, such as the Parliament, news TV channels, and the most established media outlets.
While they did not explicitly endorse the “francocide” theory, right-wing politicians’ support for the murder’s politicization resulted in much greater coverage by legacy media after October 18. The news value of the murder for legacy media suddenly changed because it was taken over by mainstream politicians, who are part of the parliamentary opposition, and generated reactions from government members seeking to defend themselves. This change in the news value generated 558 press articles published between October 18 and 24, which can be described as a media storm (Boydstun 2013). Right-wing politicians, therefore, played a significant role in the process of politicizing the murder, legitimizing the coverage of a crime that might otherwise have remained a subject of internal discussion within the far right on social media.
Steering the Discussion Toward Specific Facts
The legacy media have regulated the politicization of this crime on Twitter by the far-right, first by steering the discussion toward the facts they had assembled themselves. Facts play a significant role in the Twitter discussion regarding the murder, as 36 percent of all subevents in the corpus begin with reported facts or statements (85 out of 231). These facts encompass every aspect of the event, including the circumstances and specifics of the crime, the suspect’s profile and background, the funeral, and organized demonstrations, including those by far-right groups. They also include the responses of Lola’s parents, magistrates, government officials, and politicians. Most remarkable is that, despite the considerable involvement of far-right supporters who advocated for their anti-immigration agenda and criticized mainstream media, legacy media assembled nearly two-thirds of the reported facts or statements that generated a subevent on Twitter (Table 4). Opinion and partisan media outlets, along with politicians, activists, and others, have played a less significant role, assembling reported facts and statements in only 9.5 and 22 percent of subevents, respectively.
Distribution of Assemblers Behind the Sub-Events of Lola’s Murder (%).
Additionally, subevents based on facts or statements assembled by legacy media attract more interpreters than subevents involving other assemblers (Table 5). Regardless of whether they are partisan or nonpartisan, interpreters consistently focused on the facts assembled by legacy media. This also applies to far-right supporters, even if they pay more attention than other supporters and nonpartisans to the facts assembled by partisan media. Overall, they, too, mainly discuss the facts assembled by legacy media, which illustrates the paradox of populists criticizing mainstream media while relying heavily on them (Haller and Holt 2019).
Distribution of Interpreters According to the Assemblers of Fact-Based Subevents.
Note. Subevents whose fact was assembled by a peripheral media are excluded from these calculations (n = 2). Since users may have tweeted in more than one subevent, the total number of distinct users is not equal to the sum of distinct users having tweeted in the different types of subevent. Reading note: First cell: 73.7 percent of the nonpartisan users have tweeted at least once during a subevent focused on a fact assembled by legacy media. Second cell: Subevents focused on fact assembled by a legacy media contain an average of 43.5 tweets published by nonpartisan users.
The facts assembled by politicians and activists result in a limited number of subevents (nineteen out of eighty-three), with only a few of them being instances of disinformation (three out of nineteen) – a far-right elected representative announcing a rally in memory of Lola; a user shared pictures of the murderer from her TikTok and Instagram accounts. Some are dubious accusations by far-right activists against Muslim influencers or left-wing activists. A few others are more clearly conspiracy theories, such as the claim that organ trafficking was the motive for the crime (which, however, was suggested the day after the discovery of the girl’s body by an article in Le Parisien, a legacy media), or the theory that the murder would be a ritual crime, supposedly common in North Africa.
One subevent illustrates that legacy media also regulate some discussions based on factually false statements. It started with this tweet posted on October 20 by Damien Rieu, a far-right activist: “The family friend (who runs the charity) confirms that #LOLA’s parents are not opposed to tributes and rallies. They just don’t want to take part. (. . .) the media are lying.” Our analysis reveals that, unlike the other sub-events, this one primarily attracts far-right supporters, accounting for 82 percent of the users, compared to 31 percent for all fact-related subevents (see Figure 3). The supporters repeated the same statement, outraged that the media were lying about the position of Lola’s parents. The echo chamber was nearly complete until the parents issued a press release on the same day, extensively covered by the legacy media, in which they called for “an end to the use of their child’s name and image for political purposes.” Then, nonpartisans began to denounce the far-right activists’ manipulation: “You hear that, @DamienRieu, you jerk? Lola’s family asks that people stop using ‘their child’s name and image for political purposes’,” wrote a left-wing user. Eventually, the far-right supporters abandoned their unfounded claim.
Time distribution of a subevent based on false information (alleged agreement of Lola’s parents to tributes and rallies).
This result strongly contrasts with the idea of social media as a forum for discussion based on alternative, or at least unchecked, facts. Despite the crime being the focus of intense mobilization by the far-right, which demonstrated a deep distrust of established media, the facts gathered by legacy media serve as the foundation for much of the discussion about the murder on Twitter.
Directing the Discussion to Question the Relevance of Politicization
A final key finding is that legacy media partially regulated the politicization of Lola’s murder on social media by steering the discussion toward questioning the relevance of politicization. We refer to “reflexive” all news items that do not focus on the crime, either its facts or interpretations, but help document, criticize, or justify the process of politicization. Our analysis shows that this reflexive coverage, which accounts for a substantial proportion of the coverage produced by legacy media, regulates the discussion on Twitter in two ways: first, by guiding the conversation toward questioning politicization; second, by broadening the conversation to include more politically diverse users.
A quantitative study of the corpus of press articles (excluding broadcast and online-only media coverage) reveals that coverage changed after October 18. Except for Le Figaro, which relays right-wing and far-right-wing interpretations of Lola’s murder, all the legacy media keep their distance from immigration-related politicization. As shown in Figure 4, the articles focused less on the facts of the murder, questioning its politicization. For example, an article published in Le Parisien on October 19 analyzes the concerns of National Rally leaders regarding the campaign by Eric Zemmour’s supporters (political competition coverage). Another article published on the same day by La Croix features an interview with a centrist politician who argues that, while it is essential to show compassion for the victims of a horrific crime, caution must be exercised to “avoid reinforcing the idea that every crime stems from a failure of justice or administration” (criticism of politicization).
Reflexive coverage of Lola’s murder by legacy media.
Simultaneously, the conversation around the murder became increasingly reflexive on Twitter (Figure 5). After October 18, participation was higher in reflexive subevents than in nonreflexive subevents, which mean that more tweets focused on discussions about the legitimacy of politicization. A more detailed analysis shows that criticism, justification, and documentation of politicization dominated these reflexive conversations – “He said, she said,” and political competition coverage were virtually absent from the Twitter conversation.
Distribution of participation according to the reflexive nature of subevents.
Legacy media influenced the reflexive turn in discussions about the murder on Twitter (Figure 6). In the initial days, it is mainly the partisan, nearly exclusively far-right media that generate reflexive subevents. They defend the politicization of the murder by criticizing those who reject the connection between immigration and crime. They refer less to actual statements made by journalists or politicians, who make little mention of the murder, and more to the debates following previous crimes attributed to immigrants. Beginning on October 18, legacy media sparked reflexive discussions on Twitter, and thereafter, most discussions on the platform were driven by legacy media outlets, grounded in facts, statements, and analyses that questioned the politicization of the murder (Figure 6). For example, BFMTV, a legacy news channel, pieced together the facts behind several important subevents: on October 18, when it revealed that Eric Zemmour’s party had purchased domain names related to the murder of Lola; on October 21, when it disclosed that among those present at the demonstration for Lola were white supremacists, nationalists, and neo-Nazis. As a result, legacy media have gradually succeeded in significantly shaping the discussion on Twitter to address the legitimacy of a politicization focused on immigration.
Daily distribution of interpreters’ political preferences in reflexive subevents.
Most remarkably, when legacy media were connectors for reflexive subevents, they generated participation from contributors with more varied political preferences, especially when it came to criticizing the process of politicization (Figure 6). By contrast, when the same media outlets serve as connectors for nonreflexive subevents, or when partisan media serve as connectors for any other subevents, the discussion is monopolized by far-right contributors (Figure 7). As a result, far-right contributors are increasingly exposed to contributors with different political views in the second half of the period. This highlights the distinctive ability of legacy media to engage a more diverse political audience through reflective coverage, which also helps regulate the politicization of migrant crimes on social media.
Daily distribution of interpreters’ political preferences according to the type of subevents and connectors.
Discussion and Conclusion
We set out to examine the role of legacy media in the increasingly frequent situations where far-right activists intensely use social media platforms to politicize crimes that have been committed, either actually or allegedly, by immigrants or people of foreign origin. While research on the politicization of migrant crimes often overlooks the mutual influence between social media and legacy media, we examined whether and how legacy media are likely to influence far-right mobilization on social media. To achieve this, we have created a computational method to reconstruct, in unprecedented detail, the politicization of a single case on Twitter: the murder of a teenage girl by an Algerian woman in Paris in October 2022.
Our main argument is that, in the case studied, French legacy media have acted as a “moral compass” by regulating the form and spread of far-right politicization surrounding the murder on social media. This regulation does not involve influencing the overall amount of attention paid to the event on social media. In contrast, the mobilization of the far right on social media drew the attention of conservative politicians who saw it as an opportunity to criticize the government, which indirectly pressured legacy media to cover the murder more extensively. However, legacy media have shaped the framing of the discussion on social media by steering the conversation toward the facts they have assembled and questioning the relevance of politicizing the crime as an immigration issue. By documenting how the murder has been politicized and criticizing the legitimacy of this politicization, these media organizations influenced the nature of discussions on Twitter, encouraging more politically diverse people to engage in conversations about the relevance of generalizations that can be drawn from the murder. This influence, which can be described as a second-level agenda-setting power of legacy media, is not limited to adopting a hegemonic framing that restricts minority voices in general, not just those from the far right. In the case studied, legacy media offer the most reflexive coverage while mainstream politicians relay far-right politicization.
This study has several implications. First, it demonstrates that legacy media still possess the ability to regulate, even though they cannot fulfill a gatekeeping role, and that their power to influence the attention social media users devote to a news event is limited. This regulatory capacity, which we refer to as “moral compass,” does not involve morally condemning politicization or judging the shocking nature of politicizing a crime. Instead, it involves implementing a set of practices that align with journalists’ professional missions: producing facts, investigating actors who seek to politicize a crime, presenting arguments for and against politicization, and analyzing the underlying political strategies. As such, the moral compass mentioned here is procedural rather than substantive. Second, this study illustrates the value of developing computational methods to understand the politicization process of news events on social media. As platforms engage a diverse range of actors aiming to capture attention toward specific events, innovative methods are necessary to account for the influence dynamics among these actors.
However, we must remain cautious about generalizing the main argument of this study: it is based on a single case, anchored in Twitter and a French context. The method needs to be replicated in other cases to understand whether specific configurations can influence legacy media’s ability to regulate the politicization of migrant crimes on social media. France’s media environment is highly elitist (Hallin and Mancini 2004) and hierarchical (Cointet, Cardon, et al. 2021), with less, albeit increasing, political polarization compared to the United States. Other media environments characterized by greater polarization can be expected to severely limit the ability of legacy media to regulate far-right politicization on social media. Additionally, the characteristics of crimes may have consequences for the dynamics of politicization and, therefore, legacy media’s ability to play a moral compass role.
Therefore, further studies mobilizing this method are needed to understand how media organizations can secure a space for public discussion free from the most brutal forms of politicization.
Supplemental Material
sj-docx-1-hij-10.1177_19401612251377328 – Supplemental material for Legacy Media as a Moral Compass: A Computational Study of the Politicization of a Migrant Crime on Twitter in France
Supplemental material, sj-docx-1-hij-10.1177_19401612251377328 for Legacy Media as a Moral Compass: A Computational Study of the Politicization of a Migrant Crime on Twitter in France by Sylvain Parasie, Antoine Machut and Béatrice Mazoyer in The International Journal of Press/Politics
Footnotes
Acknowledgements
We thank the editor and the reviewers for their help in developing this article through challenging and productive questions, suggestions, and criticisms. Earlier versions of the manuscript were presented at the University of Milan, in the Department of Social and Political Science, and at the Conference “The Scenes of Public Denunciation” at Paris Pantheon-Assas University. We thank seminar attendees for their most helpful feedback, particularly from Cécile Méadel, Arnaud Esquerre, Chris Anderson, Massimo Airoldi, Emiliano Grossman, Jean-Philippe Cointet, Zvi Reich, and Jean-Samuel Beuscart. Last but not least, we wholeheartedly thank our colleagues whose collaboration was essential for this project: Auriane Schummer, who made a creative contribution to the development of our analysis categories; Maxime Crépel, who helped us compile the corpus of press articles and contributed to the initial analyses; Katharina Tittel and Kelly Christensen, who took part in the data exploration; and Rebecca Muhlhaus, who provided additional coding support during the second stage of revisions.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has been supported by Agence Nationale de la Recherche (ANR-21-CE38-0016).
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
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.
