The Discursive Dimensions of Pernicious Polarization. Analysis of Right-Wing Populists in Western Europe on Twitter

Abstract

The objective of this research is to explore the political discourse of West European right-wing populist leaders in the perspective of pernicious polarization, focusing on their positions and argumentation styles. To achieve this, over 50,000 tweets from right-wing populist leaders in Western Europe (Italy, France and Spain) were collected for a period spanning from 2 July 2019, which marks the beginning of the 9th legislature of the European Parliament, to 2 July 2023. Employing Text Mining and Topic Modeling techniques, this research will reconstruct and comparatively analyze the topics addressed by the leaders from different countries and the dynamics of polarization discourse proposing an exploratory study aiming to locate the words of pernicious polarization used by each leader.

1. Introduction

Pernicious polarization seems to be progressively spreading across Western Europe. Far-right and authoritarian parties are increasing voter support in a number of advanced—supposedly consolidated—democracies, by leveraging a social and political “polarization that divides societies into ‘Us vs. Them’ camps based on a single dimension of difference that overshadows all other” (McCoy and Somer 2019, pp. 234–35).

Our contribution highlights the pernicious dynamics of polarization as a political discourse among right-wing populist leaders of Western Europe. The objective of this work is twofold: on the one hand, it intends to explore the speeches of selected leaders since the beginning of the IX legislature of the European Parliament (2 July 2019), highlighting the themes on which they leverage and the pernicious dynamics they implement; on the other hand, a methodological approach is proposed which, proceeding along a scale of detail that goes from the general to the particular, presents itself as the first attempt to build a future vocabulary of words that reflect the concept of pernicious polarization useful for training machine learning models. The underlying methodological work is based on a theory-driven epistemological approach to the digital that focuses not only on words in their context but also on the interpretive role of the researcher in the entire analytical process.

This work is composed as follows: in the second paragraph the theoretical approach of pernicious polarization is discussed; in the third paragraph the state of the art of approaches to the analysis of polarization is discussed and a methodological proposal is described for a future construction of a vocabulary useful for the semi-automatic analysis of pernicious polarization in leaders’ speeches; the fourth paragraph describes the construction of the dataset, the analysis conducted, and the results for each context analyzed. Finally, in the last paragraph, conclusions are drawn in light of the reference theory.

2. The Discursive Dimensions of Pernicious Polarization

Pernicious polarization is defined as “a process whereby the normal multiplicity of differences in a society increasingly aligns along a single dimension, cross-cutting differences become reinforcing, and people increasingly perceive and describe politics and society in terms of ‘Us vs. Them’” (McCoy et al. 2018, p. 18). This political and relational definition of polarization moves beyond conventional definitions as “ideological distance in contra-distinction to ideological proximity” (Sartori 2005, p. 120). Polarization arises when political actors abandon the “middle ground”—i.e., the place for cooperation—and group themselves in opposite ideological camps, thus triggering dynamics that expose democracy to crisis and collapse (Bermeo 2003). Furthermore, the definition we adopt differs from the concept of affective polarization, which relates to emotion rather than ideologies, meaning the tendency of opposing political parties to dislike and distrust one another (Druckman and Levendusky 2019; Gidron et al. 2020, 2023).

The relational nature of pernicious polarization stands in how the differentiation between ‘Us vs. Them’ is interpreted and used, as well as in how groups react to each other (Somer and McCoy 2019). Its political nature emerges when it becomes a strategy employed for political aims.

Polarization acquires a pernicious form when it provokes an elite backlash and when political entrepreneurs pursue their political objectives by using polarizing strategies. On the one hand, it is associated with the elite inability to manage the crisis, namely the rise of an antagonistic group (Stavrakakis 2018). In other words, it takes two groups—and their (un)willingness to cooperate or inability to respond to each other—for polarization to become pernicious. On the other hand, the phenomenon involves political actors that “utilize a common set of strategies and tools to undermine institutional constraints, manipulate participatory processes such as elections and divide or marginalize political opposition as they slowly dismantle democracy” (Somer and McCoy 2018, p. 4). Such tactics are built on existing underlying social cleavages in a society (Lipset and Rokkan 1967), or “formative rifts”, and simplify “the normal complexity of politics and social relations […] by aligning otherwise unrelated divisions, emasculating cross-cutting cleavages, and dividing society and politics into two separate, opposing, and unyielding blocks” (Somer and McCoy 2018, p. 5). The political construction around these rifts contribute to the rise of the pernicious form of polarization.

One rift may be simply related to the support of (or opposition to) a personal leader. In this sense, we have witnessed the emergence of a number of controversial political leaders. All around the world, the latter exploited polarizing tactics—particularly effective within the digital arena—to succeed and galvanize grassroots support (Calise and Musella 2019; Nunziata 2021). In recent years, within the European political landscape, “a polarization between populism and antipopulism became a structural characteristic” (Stavrakakis 2018, p. 8). Extreme right populist actors are progressively challenging the established order in so-called consolidated or advanced democracies of Western Europe—and, broadly, in European liberal democracies (Schulze et al. 2020). They tend to take antidemocratic forms, capitalizing on the broken promises of established oligarchic forces that failed to honor the founding principles of democracy, hence actively participating in its destruction (Stavrakakis 2018). Right-wing populists dwell so much on certain topics, such as those related to immigration and security issues, to fuel anger among social media users (Gerbaudo et al. 2023). For instance, the leaders here analyzed—Giorgia Meloni (Fratelli d’Italia) and Matteo Salvini (Lega) in Italy; Marine Le Pen and Jordan Bardella (Rassemblement National) in France; Santiago Abascal (Vox) in Spain—belong to this category.

The populist discourse typically relies on a polarized frame of politics and society, leveraging on an exclusionary rhetoric which opposes the will of the people to a corrupt establishment. These characteristics make populism one of the discursive dimensions of polarization and a specific and widely used type of polarizing politics. Populists typically use a polarizing language: indeed, the Manichean discourse is one of the main elements of populist political communication (Aalberg et al. 2016; Mudde and Rovira Kaltwasser 2017). A popular definition of the concept1, provided by Cas Mudde, describes it as “a thin-centered ideology that considers society to be ultimately separated into two homogenous and antagonistic groups: ‘the pure people’ and ‘the corrupt elite’, and argues that politics should be an expression of the volonté générale (general will) of the people” (Mudde 2004). When approached as a political communication style, it is defined as a rhetoric that constructs politics as a struggle between the people and the oligarchy (De La Torre 2000).

While it is a salient characteristic of populism, the Manichean logic represents a prominent feature of polarizing politics. Overall, pernicious polarization tends to sharpen ‘Us vs. Them’ identity politics, i.e., to divide the electorate into two mutually exclusive and antagonistic camps, characterized in moral terms of “good” and “evil”, which aggregate into one cleavage multiple divisions of the society (McCoy and Somer 2019).

All in all, the discursive dimension of polarization is based on latent resentments based on underlying social and political cleavages. McCoy and Somer (2019) identify three primary types of grievances:

• Political grievance, associated with the crisis of representation (e.g., corruption of the elite, unresponsive technocratic or expert governments. See Berman 2017; Luce 2018). An example is the occurrence of polarizing populism in South America, explained by state crisis, with outsider politics leveraging anti-systemic appeals to promote controversial policy programs (Handlin 2018; Kitzberger 2023). Moreover, in Western Europe, this grievance translates into euroscepticism and criticism towards the EU (Pirro et al. 2018).
• Economic grievance, with short-term and/or long-term economic crises used as mobilizing tools for the so-called “left-behinds”. Donald Trump narratives around the silent majority left behind by globalization and the post-industrial economy (Enli 2017; Ott 2017) are an example of such messages. Also, radical parties from Western Europe appeal to the “invisible” or the “forgotten”, targeting all who feel “left behind” in national politics, e.g., voters from peripheries, where they tend to draw most of their support (Ivaldi et al. 2017).
• Cultural grievance, i.e., religious and moral issues related to conservative/secularized lifestyles, death, gender and sexual rights, or changing demographics. For instance, the appeal expressed by Viktor Orbán to keep Hungary for ethnic Hungarians and away from Syrian and northern African refugees (Vegetti 2019). Likewise, Donald Trump exploited an enduring racial, ideological, and cultural polarization within the electorate to win the election (Abramowitz and McCoy 2019). Finally, such grievance is particularly associated to Western Europe, exploited by the radical rhetoric of extreme right leaders to increase the salience of immigration amongst some voters as well as an anti-immigration sentiment (Dennison and Geddes 2019). For instance, the discourse of Rassemblement National especially relies on the fight to Islam that threatens French values (Abdeslam 2021).

Following this categorization, our analysis focuses on the discursive strategies of pernicious polarization, aiming to identify their specific dimensions within the online communication of selected cases from Western Europe.

3. Analyzing Polarization in Political Discourse: A Methodological Challenge

Amid the burgeoning scholarly output and political events shaping Western democracies since 2016, the critical need to understand the dynamics of polarization and its potential implications for the democratic system has become apparent. However, comparative studies at the mass level remain exceedingly complex. Numerous studies have sought to identify political polarization (See Dalton 2008; Ezrow 2007; Hazan 1995; Pardos-Prado and Dinas 2010; Taylor and Herman 1971; Ladner 2014; Singer 2016). An intriguing example is the work by Lauka et al. (2018), which utilized data from the Comparative Study of Electoral Systems Module 3 (CSES, 2006–2011)2 to develop a new index of mass party polarization in multiparty systems. Their index of mass political polarization is based on party support (positive partisanship) and rejection (negative partisanship), as opposed to merely perceived ideological differences or positive partisanship alone. This approach enabled them to quantify mass party polarization with a single score for each country over a specified period.

Many other studies focus on studying polarization in political discourse. The studies focusing on discourse as a research object systematically examine text within specific contexts (Batool and Tehseem 2022). Foucault (1972) posits that discourse analyzes how knowledge is socially constructed and practiced, making it intrinsic to society’s ideological models and a symbolic and significant societal element (Batool and Tehseem 2022). In political debates, the surge in online textual data has propelled Text Mining and Natural Language Processing (NLP) into an active research field for examining political opinions and positions. However, the complexity and nuances of human language present considerable challenges. Analyzing political position polarization is particularly complex; although polarization and pernicious polarization have clear theoretical definitions, discrepancies exist between their substantial meanings and empirical detection, complicating empirical investigations in digital spaces for political and social scientists (Pereira et al. 2024).

Pereira et al. (2024) present a systematic review of computational approach to analyze political polarization in texts, introducing a framework that categorizes five main approaches: Statistical and Parametric Models, Classification Models, Time Series Models, Clustering, and Scaling Models. An example of Statistical and Parametric Models by Belcastro et al. (2020) focuses on analyzing political polarization through manual Twitter post analyses. Although less scalable, their work lays the groundwork for more advanced methods that account for the proportion of terms supporting specific ideologies. Gentzkow et al. (2019) develop a two-party context polarization model using congressional speech bigrams and a multinomial distribution measurement system, reducing the need for domain-specific knowledge. Marchal (2021) employs a Classification Model based on Bayesian estimation for evaluating political and affective polarization on Reddit, highlighting the minimization of ideological coding error and the potential limitations in accuracy metrics for polarization identification. Jiang et al. (2020) examine COVID-19 discourse polarization on Twitter, adding a temporal dimension to track polarization evolution in response to pandemic developments. Sloman et al. (2021) explore Clustering and Scaling Models to uncover how linguistic variations reflect political identities, demonstrating that specific linguistic patterns can signal political affiliations and that polarity can be deduced from the lexical proximity or distance in speech. Despite advancements—as the authors of the work themselves also claim—challenges remain in analysis depth and adapting to textual context specificities.

In an attempt to give greater depth to the analysis, this analysis adopts an epistemological approach, guided by theory (Amaturo and Aragona 2019), emphasizing the context of words and the researcher’s interpretive role. For this reason, the work is to be considered a first exploration, useful for obtaining the first empirical evidence that will be useful for orienting future work with more sophisticated and exclusively data-driven techniques. I is an exploratory study aiming to locate the words of Pernicious polarization. The goal is not to look at the most frequent words in a corpus, because we know from computational linguistics studies how much a classic frequency measure places too much emphasis on high-frequency terms and, on the contrary, how specificity measures assign too much weight to low-frequency terms (Manning and Schutze 1999). The difficulty with selecting or weighting terms is establishing a good balance between popularity and specificity (Aizawa 2003). The objective is to understand the lemmas within their discursive contexts and observe their associations. The analysis begins with a topic analysis in European populist leaders’ political speeches. Le us consider a topic as a mix of specific and other shared words that create diverse semantic universes; from these topics, terms indicating political debate polarization are identified through analysis in elementary contexts, like sentences or paragraphs, which contextualize words and deduce specific meanings (Bolasco and De Mauro 2013). The importance of document parts is determined by the informative weight of elementary contexts (Habert 2005), marked by discursive formulas, position in the document, and each word’s specific weight relative to its document distribution. Further detail involves analyzing lemmas highly associated with each identified word, here depicted in radial diagrams with the keyword at the center and associated words around it, each at a distance indicating its association degree. This iterative procedure aims to construct a preliminary set of polarized words for each analyzed position. For a matter of space, for each language, the extracted topics and the words characterizing them will be commented on, and the associations of the lemmas will be analyzed only for some of the most significant words of the topic.

4. Construction of the Empirical Base

To fulfill the research objectives, we gathered tweets from prominent right-wing populist leaders in Western Europe. The collection period spanned from 2 July 2019, coinciding with the onset of the 9th legislature of the European Parliament, to 2 July 2023, marking the final extraction date, using the social media monitoring tool Fanpage Karma3.

Fanpage Karma is an online tool for social media analytics and monitoring that allows one to keep track of public profiles on Facebook, Instagram, or YouTube. The data collection was carried out by selecting the profiles of the politicians of interest to us and setting the time frame defined for the objectives of the research. At that point, the tool presented us with a quantity of results that we could download in Excel format to carry out our analyses.

The focus was on leaders from Italy, Matteo Salvini of “Lega” and Giorgia Meloni of “Fratelli d’Italia”; two German leaders were excluded to avoid misinterpretation of the results. A total of 57,286 tweets were collected.

For each language, the extracted corpus underwent preprocessing aimed at simplifying and transforming the textual data to the level of individual words or tokens. This process involved removing elements that contribute to noise and are not significant for analysis, such as punctuation, URLs, email addresses, numerical values, dates, codes, and stopwords. Additionally, n-grams were modeled. Subsequently, the corpus was lemmatized, and a dictionary was constructed. This dictionary, comprising words that appeared with a frequency of 10 or more, served as the basic unit of analysis4.

To analyze the semantic structure of our selected texts, we opted for topic modeling, an automated approach that offers a statistically robust solution. Specifically, we employed a topic extraction model based on Latent Dirichlet Allocation (LDA) illustrated in Figure 1. This model operates under two fundamental principles: each document comprises a mixture of topics, and each topic is characterized by a collection of words (Blei et al. 2003). More precisely, documents are treated as probability distributions over topics, whereas topics are considered distributions over words. LDA simultaneously estimates these distributions, identifying the word combinations that characterize each topic and the topic combinations that describe each document. In probabilistic topic models, words are not segregated into discrete categories; instead, they may be shared among topics, each assigned a certain probability. This feature allows documents to be interconnected in terms of content, thereby reflecting the natural use of language more accurately. In this framework, argument probability provides an “explicit representation of a document” (Blei et al. 2003).

Selecting the number of topics in an LDA analysis is critical to ensure the model’s informativeness and interpretability. While there is no universal guideline for determining the optimal number of topics (Grimmer and Stewart 2013), several methods and considerations can facilitate this decision (Blei and Lafferty 2009). In alignment with the epistemological approach of our work, which is theory-guided and positions the researcher at its core, we adopted an iterative “trial and error” method. This process starts with an initial set of topics, which is then refined based on feedback from consistency metrics and human interpretation. The latter is crucial for ensuring the topics are interpretable and distinct.

For each corpus, 10 topics were identified and subsequently named based on several criteria, including the significance of specific words defining the topic, the frequency of words common across topics, and semantic tagging in basic contexts. Thus, a topic results from its unique words combined with those shared with other topics, each with varying probabilities. This mix forms distinct semantic realms. By considering words in their respective contexts, this approach enables accurate topic labeling. The analysis yielded a synthesis of basic contexts organized hierarchically by an information score, which reflects the relevance attributed to each context.

5. Divergences and Convergences: The Analysis’ Results

The percentages of the topics (

Table 1. Percentages of topics extracted for each corpus.

ITALY
Topic 1 Crime and justice	12%
Topic 2 Projects and local governance	9%
Topic 3 Communication and events	9%
Topic 4 (Anti)European Politics	9%
Topic 5 Immigration and border control	9%
Topic 6 Mobilization to vote	10%
Topic 7 Salvini	9%
Topic 8 Health crisis and management of the school sector	9%
Topic 9 Sovereignty, identity, and Italian values	11%
Topic 10 Economy and work	13%
FRANCE
Topic 1 Health crisis and government management	10%
Topic 2 Islamism and internal security	16%
Topic 3 Immigration and border control	11%
Topic 4 Communication and events	8%
Topic 5 National Rassemblement Strategy	11%
Topic 6 Opposition	10%
Topic 7 National priorities	7%
Topic 8 Pension reform	8%
Topic 9 French sovereignty, identity, and values	11%
Topic 10 Taxes and economic policies	8%
SPAIN
Topic 1 Threats and internal security	13%
Topic 2 (Anti)European politics and freedom	8%
Topic 3 Health crisis and government management	11%
Topic 4 Spanish regional and national identity	9%
Topic 5 Immigration	10%
Topic 6 Workers and taxes	9%
Topic 7 Events and mobilization	9%
Topic 8 Opposition and censorship	10%
Topic 9 Catalan independence	10%
Topic 10 Vox and the Spanish political scenario	11%

) illustrate how right-wing leaders in Spain, Italy, and France have focused on different themes, while at the same time showing important areas of overlap. It is the case of themes such as security and immigration, which indicates a shared perception of external threats and concern for national sovereignty. In Spain, the themes of “Threats and Internal Security” and “Health crisis and government management” are prominent, suggesting a strong focus on national security and response to health emergencies. The issue of Catalan independence and the Vox party highlight specific issues of national identity and internal politics. In Italy, there is a particular focus on “Crime and Justice” and “Economy and Work”, which may reflect concerns for security and economic stability. Salvini’s presence in the dominant themes underlines the importance he had in the Italian political debate in the period investigated. Salvini is portrayed as a polarizing character, the subject of criticism and attacks but also of broad popular support. In France, however, “Islamism and Internal Security” is the dominant theme, followed by concerns about “Immigration and Border Control” and the “Strategy of the Rassemblement National”. This indicates an emphasis on internal security and a polarized political debate around national identity. The differences are clearly highlighted in the national priorities which reflect the political and social specificities of each country which determine the context in which the parties modulate their messages and strategies: in Spain, regional identity; in Italy, crime and the economy; and in France, the Islamist question.

At this point, the topics common to the three countries investigated were explored in depth, which from their words and the elementary contexts in which the words are inserted have proven to be particularly divisive and which can damage social cohesion and the functioning of democratic institutions and for this reason responds to the theoretical concept of Pernicious Polarization. With the in-depth analysis of the topics, we moved on to select and investigate the words with the highest probability value of belonging to the topic and their associations with the words of the entire vocabulary. The in-depth topics are those that concern immigration and security, criticism of the European Union, national identity, and opposition to the Government in office.

5.1. The Italian Context

In the Italian context, the issue of immigration and border control is at the heart of heated debates by the analyzed leaders. The main criticisms concern the governmental management being perceived as too permissive and the increase in landings, with a marked preference for more restrictive policies that ensure national security. NGOs are accused of facilitating illegal immigration, in opposition to political figures like Matteo Salvini, who are instead presented as defenders of national interests. This debate intersects with broader issues such as legality and national identity, reflecting deep divisions on how Italy should address the challenges of immigration in a complex European and global context. In detail, Theme 5 titled “Immigration and Border Control” highlights criticisms of the government’s management of immigration, emphasizing a perceived increase in landings and a reception policy judged as too permissive. The key terms of this topic that best meet the objective of identifying a set of words distinctive of harmful polarization are “illegale” (illegal), “blocco_navale” (naval blockade), “scafisti” (smugglers), “difendere_i_confini” (defend borders), “immigrazione_clandestina” (illegal immigration), “profugo” (refugee), “campo_profughi” (refugee camp), “sbarco” (landing), “porti_aperti” (open ports), and “ong” (NGOs). The theme reveals a clear preference for more restrictive policies, with particular emphasis on protecting national borders and reducing illegal immigration. Italy is defined as “Europe’s refugee camp”, and the number of landings during Salvini’s tenure as Minister of the Interior is compared to subsequent ones, highlighting an increase. National security is continually invoked as illegal immigration is presented as a direct threat to the safety of citizens. NGOs are particularly criticized for the role they play in the migration phenomenon and are accused by leaders of contributing to illegal immigration and operating without sufficient governmental control. To the identified words, all those associated with them must be added, some of which are represented in the semantic networks of Figure 2. These suggest a focus on the criminal aspects of illegal immigration, highlighting on one hand the role of traffickers in facilitating unauthorized entry through the Mediterranean and, on the other hand, pointing fingers at the perceived responsibilities of the left-wing government and the European Union in managing illegal activities. Finally, the theme of illegal immigration reflects concerns about national identity and border security. The networks illustrate how certain languages can be used to frame a complex issue like immigration in a particularly partisan light.

The emphasis on crime, lawlessness, and the need for defense suggests a narrative that views immigration as an ongoing threat to internal security. The absence of terms related to humanitarian considerations, or the causes of migration may reflect a viewpoint that overlooks the broader context of the issue, focusing instead on enforcement and deterrence.

Topic 7, entitled “Salvini”, is entirely focused on the figure of Matteo Salvini and highlights the central role that the former Interior Minister and leader of the League played in the Italian political debate during the period investigated. This prominence of a single political figure is a distinctly Italian characteristic—as it is a result that does not emerge in the other contexts analyzed—and is especially related to Salvini’s leadership style on the Web (See Nunziata 2021; Starita and Trillò 2021). Among the most significant terms of the topic are “palamara” (surname of a former Italian magistrate, former member of the Superior Council of the Judiciary), “concessione” (concession), “magistrato” (magistrate), “riferire” (to report), “battaglia_della_lega” (League’s battle), “decreto” (decree), “decreto_immigrati” (immigration decree), “mes” (MES—European Stability Mechanism), “mattarella” (Mattarella, referring to the President of Italy), “interrogazione” (question or inquiry), and “stop_mes” (stop MES).

Salvini is described as a victim of attacks by political opponents, the media, and public figures such as magistrates, who accuse him of conducting controversial policies regarding the management of immigration and opposition to the ESM (European Stability Mechanism). The topic reflects broad popular support for Salvini, portrayed as a defender of Italian interests and a leader capable of conducting his political battles despite adversity. Thus, Salvini becomes a central and particularly divisive figure in the Italian context, capable of catalyzing both fierce criticism and fervent support around himself.

Looking at the most significant lemmas (Figure 3), in the first case, the central node is “Magistrate” and the words associated with it (“Mafia”, “Justice”, “Justice Reform”, and “Scandal”) reflect a narrative that calls into question the integrity or effectiveness of the Italian judicial system, suggesting connections between the judiciary and politics or organized crime. The particularly critical language reflects strong opposition to the judiciary as a body in need of reform because it is colluding and corrupt. This also most likely includes Salvini’s referendum on justice, which did not reach a quorum.

In the second graph, “decreto” (decree) and the words associated with it reflect the critical discussion on urgent legislative measures and government decrees, underlining an overall negative evaluation of the government’s actions. In both cases, the chosen words and their associations may be intentionally designed to incite strong reactions and partisan positions by intensifying already deeply felt divisions in society.

Topic 1, titled “Crime and Justice”, highlights concerns for personal safety, with a particular emphasis on violence against women and attacks on minors and citizens. Some of the most interesting words in this topic are “scorta” (escort), “boss” (boss), “tolleranza” (tolerance), “aggredire” (to attack), “sparare” (to shoot), “omicidio” (homicide), “detenuto” (detainee), “violenza” (violence), “carcere” (prison), and “condannato” (convicted). The demand for justice, severe penalties for attackers, and protection for victims is a recurring theme of the topic, which promotes the view of a state that must be uncompromising against crime. The basic contexts often recall criminal events involving illegal immigrants, placing a critical emphasis on the government’s handling of immigration, particularly regarding landings and reception policies. The constant reference to immigrants, associated with issues of legality and security, underlines a vision in which the management of immigration is considered a crucial priority for internal security and the integrity of the national social fabric. Looking at the associations of some of the mentioned lemmas (Figure 4), the word “violenza” (violence) reflects the perception of a social threat that requires a strong response, related to security and public order issues; moreover, the presence of words like “centri sociali” (social centers) and “islamismo” (Islamism) indicates a tendency to attribute the responsibility for urban violence to specific social groups. The lemmas associated with the word “aggredire” (to attack) radicalize the political discourse leading to the action of physical attack or threat. Terms such as “punire” (to punish), “tolleranza zero” (zero tolerance), and “forze dell’ordine” (law enforcement) indicate a strict judicial and police response to acts of aggression. Again, the connection with “centri sociali” (social centers) and “immigrato” (immigrant) suggests the narrative’s tendency to identify the origin or affiliation of those committing aggressive acts, sometimes used to support more repressive policies; similarly, the words “immigrato” (immigrant) and “droga” (drug) tend to construct a dialectical and direct link between crime and specific groups or behaviors.

The semantic universe of these networks seems to emphasize a connection between crime and specific social categories or groups, reinforcing the narrative that certain people or behaviors are inherently linked to security issues. This generalization could have the effect of demonizing entire communities or legitimizing policies of exclusion or discrimination.

Topic 4, renamed “Politics (Anti)Europeanist”, highlights a critical view of the European Union, based on perceptions of inefficiency, lack of solidarity, excessive bureaucracy, and interference in national sovereignty. These positions reflect the context of an electoral campaign where euroscepticism, the defense of identity, and national sovereignty play a key role in the political discourse of the Italian right. The most significant words are “unione_europea” (European Union), “commissione_europea” (European Commission), “green_pass” (green pass), “regime” (regime), “basta_coprifuoco” (enough with the curfew), “dittatura” (dictatorship), “costituzione” (constitution), “danneggiare” (to harm), “discriminazione” (discrimination), and “governo_italiano” (Italian government). These words, in basic contexts, emphasize criticism of the management of the pandemic at the European level and accuse Europe of being divisive and incapable of responding effectively to common challenges rather than being a united and supportive community. The expressions found in the topic show a strong emphasis on national sovereignty, with clear opposition to what is perceived as excessive control and interference by European institutions. The defense of sovereignty is manifested through open criticism of EU policies and decisions, considered harmful to Italy’s national interests. There is also strong opposition to the economic and austerity policies imposed by the European Union, perceived as a constraint on economic growth and the well-being of the nation. The critique extends to European bureaucracy, seen as an obstacle to the Implementation of more effective national policies that meet the needs of citizens. The words associated with the lemma “danneggiare” (to harm) (Figure 5) precisely reflect the center of this narrative, aiming to highlight how certain measures associated with the European Union are causing economic and financial damage to Italy. Terms like “lavoro” (work), “economia” (economy), “misura” (measure), and “difendere” (to defend) suggest that the discourse focuses on the consequences of European policies on national work and economy and the importance of defending national interests. “Lega” (League) and “FDI” (Brothers of Italy), two right-wing political parties, are described as defenders of these national interests against the unjustly perceived damages from Europe. The lemmas associated with the word “regime” reflect a particular radicalization of the discourse on Europe, labeled as oppressive and authoritarian.

Topic 9, renamed “Sovereignty, Identity, and Italian Values”, particularly reflects this strong sense of belonging and national identity, which is exalted through significant words such as: “donare” (to donate), “eroe” (hero), “stringere” (to tighten/to embrace), “orgoglio” (pride), “azzurro” (light blue/sky blue), “scomparso” (disappeared), “campione” (champion), “amore” (love), “lottare” (to fight), “tricolore” (tricolor), and “nazione” (nation). Through these words, sovereignty is expressed in various contexts: from the celebration of Italian Athletes to the Festival of the Tricolor, to the remembrance of “Grandi Italiani” (Great Italians), i.e., historical and contemporary figures who have contributed to Italy’s fame in the world, emphasizing the importance of collective memory and pride in the country’s cultural and social roots. The topic captures a deep sense of belonging, pride, and love for Italy, exalting its successes, culture, history, and national symbols, and reflecting a vision in which Italian identity and values are seen as fundamental for social cohesion and a sense of community.

5.2. The Spanish Context

Even in the Spanish context, the political debate is heavily influenced by themes such as internal security, immigration, national sovereignty, and criticism of European institutions. The perception of compromised public safety and the association of illegal immigration with an increase in crime are central issues, with strong support expressed towards law enforcement and criticism of the governmental management considered too permissive. In such a scenario, Vox emerges as a defender of national interests, promoting strict security policies and the protection of Spanish identity and social cohesion. Parallel to this is the criticism towards the European Union, seen as a threat to national sovereignty and cultural identity, with a call to strengthen borders and adopt a more nationalist approach. The issues of regional identity and the fight against Catalan independence further underline the polarization around the concepts of unity and national sovereignty. Lastly, the opposition to the government of Pedro Sánchez, accused of ineffectiveness and of compromising democracy and freedom of expression, reflects a deep political division, with Vox positioning itself as the main opposition force. These dynamics underline a heated debate on security, identity, and sovereignty.

In detail, in Topic 1, aptly renamed “Threats and Internal Security”, the narrative focuses on criticizing the public security management by the incumbent government, highlighting the perception of abandoning law enforcement (Policía and Guardia Civil) left without adequate material and legal support. The most significant words of the topic are “mujer” (woman), “civil” (civil), “guardia” (guard), “guardia-civil” (spanish gendarmerie)5, “ilegal” (illegal), “policía” (police), “inmigrante” (immigrant), “invasión” (invasion), “frontera” (border), “mena” (unaccompanied minor immigrant), and “menores” (minors). The discourse continuously connects the increase in assaults to immigration, with direct accusations against the government of financing illegal immigration at the expense of the well-being of Spanish citizens. Strong support is expressed for law enforcement, presented as victims of a defamatory campaign on their work protecting the borders. There is a complaint about a growing importation of crimes negatively impacting the safety of women and social cohesion. The “Immigration” topic is characterized by strong criticism of the government’s reception policies, perceived as too permissive; here again, a direct link between immigration and various social issues such as the increase in crime, pressure on public resources and services, and difficult integration is highlighted. Among the most significant lemma of this topic are “progre” (progressive), “consenso” (consensus), “efecto” (effect), “llamada” (call), “inmigración” (immigration), “progresar” (to progress), “españoles” (Spaniards), “partido” (party), “agenda” (agenda), “problema” (problem), “illegal” (illegal), and “invasión” (invasion). The Vox party promotes itself as the sole defender of national interests, advocating a hard line on Immigration with an emphasis on border security, the protection of national identity, and the importance of preserving social cohesion. The narrative used highlights a vision of immigration as a threat to the stability and security of the country, calling for the Implementation of more restrictive policies and stricter control over arrivals. Looking at the associations of some of the most significant words of the topic (Figure 6), a vision of conflict and danger associated with immigration is emphasized, radicalizing the fear and discrimination against immigrants. The lemma “ilegal” (illegal) in particular is almost exclusively associated with the theme of irregular immigration, polarized to such an extent that words like “deportación” (deportation), “alarma” (alarm), and “asaltar” (to assault) are dangerously used to justify the adoption of stricter security policies.

Topic 2 “Politics (Anti)European and Freedom” focuses on criticizing the current configuration of the European Union, promoting a vision of Europe based on national sovereignty, freedom, and the protection of borders. The most significant words of the topic are “europeo” (“European”), “libertad” (“freedom”), “foro” (“forum”), “polonia” (“Poland”), “europa” (“Europe”), “iberosfera” (“Iberosphere”), “soberanía” (“sovereignty”), “internacional” (“international”), “defensa” (“defense”), and “recuperemos_cataluña” (“let’s recover Catalonia”). The discourse challenges the policies of Brussels bureaucrats, accused of having made the original idea of Europe a failure. Vox aligns with nations like Poland and Hungary, praised for their commitment to sovereignty, freedom, and the security of their citizens. The rhetoric used emphasizes the defense of a Europe that protects its roots and cultural identity against globalist pressures, maintaining solid borders against immigration and promoting policies that respect national sovereignties. A clear opposition is expressed to movements that, according to Vox, threaten the integrity and fundamental values of Europe, with a particular focus on the need for cooperation between sovereign nations that can ensure the well-being and security of European citizens, distancing from the centralized policies of Brussels perceived as harmful to the identity and sovereignty of member states. The polarization in this context is manifested in the way these terms are associated to create a narrative that contrasts the concept of freedom and national sovereignty with the influence of the European Union, described here as well as stifling or oppressive. The words associated (Figure 7) with the lemma “Europa” (“Europe”) reflect concern for national sovereignty and cultural identity in response to European integration. Fundamental to democracy and closely linked to national security is cultural identity: the association with terms like “totalitarismo” (“totalitarianism”), “amenaza” (“threat”), and “recuperar” (“to recover”) reflects an implicit appeal to defend or reclaim freedoms, perceived as lost or under attack.

Topic 4, renamed “Regional and National Spanish Identity”, deals with Vox’s vision for the future of Spain, based on national unity, freedom, and the equality of all Spaniards. The most significant words of the topic are “españa” (“Spain”), “león” (“lion” or “León”, depending on context, but likely referring to the historical region or symbol of strength), “castilla” (“Castile”), “siembra” (“sowing”), “junto” (“together”), “orgulloso” (“proud”), “gracia” (“grace”), “vox” (“Vox”), “gran” (“great”), and “esperanza” (“hope”). These highlight a narrative imbued with a strong sense of patriotism and a call for popular mobilization to support the party as a vehicle for change towards a better future. This envisioned future emphasizes resistance against forces perceived as threatening to Spanish identity and sovereignty, including separatism and progressive policies. Vox presents itself as the sole representative of the “true Spain” working for the renewal of the Nation in line with traditional values. Events and demonstrations are seen as expressions of unity and collective strength towards achieving this goal. Among these, it is interesting to look at the words associated with the lemma “Spain” (Figure 8), which bring us back to concepts of identity and national values such as history, honor, and tradition, emphasizing nationalism and the defense of national independence.

Topic 8, named “Opposition and Censorship”, focuses on strong opposition to the government of Pedro Sánchez, criticizing its management as perceived to be ineffective and harmful to Spain. The narrative—constructed from some of the most significant words, such as “sánchez”, “moción” (“motion”), “iglesia” (“church”), “censura” (“censorship”), “gobierno” (“government”), “pedro”, “pablo”, “muerto” (“dead”), “coronavirus”, and “gestión” (“management”)—highlights Vox’s accusations against the Sánchez government, considered guilty of one of the worst management of the health crisis globally; Sánchez is accused of concealing the victims, and his leadership is accused of bringing economic and health disasters that will become acute in the near future. Vox positions itself as the only true opposition, emphasizing and promoting the need for a “moción de censura” (“motion of censure”) to overthrow the administration. Particularly for the Spanish context is the topic concerning “Catalan independence”, which focuses on the fight against separatism. Vox’s active role in countering separatist forces through legal actions and their presence in institutions ensures that national unity is emphasized. The rhetoric used paints separatism as a danger to the territorial integrity and sovereignty of Spain, and Vox presents itself as a bastion against the country’s fragmentation. The message emphasizes the commitment to fight any attempt at secession and to preserve the Spanish constitution, with an explicit reference to the judicial consequences for separatist leaders and the need for a united front against separatism. The lemmas useful for investigating the polarization linked to this topic emerge from the associations of the words that compose it. Here, two examples provided are “Sanchez” and “Censorship” (Figure 9) which convey strong criticism and accusations against the Spanish Prime Minister. Terms like “criminal”, “golpista” (“coup supporter”), “autócrata” (“autocrat”), and “ruina” (“ruin”) suggest an extremely negative representation of the leader, labeling him as harmful to the country. The presence of words like “mentira” (“lie”), “censura” (“censorship”), and “vox” (which is also the name of a right-wing party in Spain) indicates an accusation of dishonesty and suppression of freedoms by the government. In the second graph, “censura” (“censorship”) is surrounded by words like “denunciar” (“to denounce”), “muerte” (“death”), and “expresión” (“expression”), reflecting a perceived repression of freedom of expression and political action. Alongside “twitter”, “ruina” (“ruin”), and “gobierno” (“government”), the word associations reveal a narrative that sees the government as limiting public debate and the health of democracy.

5.3. The French Context

Even in the French context, public discussion intensely focuses on issues such as radical Islamism, immigration, and the preservation of identity and national values. The perception of Islamism as a direct threat to French republican principles, such as laïcité (secularism) and gender equality, catalyzes calls for decisive actions to safeguard internal security. Concurrently, the topic of immigration and border control emerges urgently, with an emphasis on immigration policies perceived too permissive as they threaten social cohesion and national identity. Against this backdrop, the Rassemblement National positions itself as a bastion against uncontrolled immigration, promoting stricter policies and strengthening French identity. The political rhetoric around these topics is highly polarized, with a marked contrast between the defense of traditional values and the openness towards a multicultural society. National priority—in terms of access to employment, housing, and social services—emerges as a key theme, underlining the preference for French citizens over immigrants.

In detail, Topic 2, renamed “Islamism and Internal Security”, explores the threat posed by radical Islamism in France. The words in basic contexts underline the perceived need to fight against Islamist ideology threatening French republican values, such as “laïcité” (“secularism”) and gender equality. The topic reflects on tragic events like the murder of two police officers killed by an Islamist terrorist; this event is used to emphasize the constant danger faced by “the protectors of national security”, namely the police force. It also discusses the government’s responsibility to protect citizens and maintain public order in the face of such threats, criticizing the management of law enforcement and internal security policy. Here too, all attention is catalyzed on the “other”; the out-group from which all national problems are supposedly derived. The most relevant lemmas of this topic from which are undoubtedly: “islamiste” (“Islamist”), “violence”, “policier” (“police”), “ordre” (“order”), and “victime” (“victim”) that clearly reflect polarized positions in reference to security, terrorism, and the approach towards Islamism. Looking in detail at the associative networks (Figure 10), that centered on “violence” is associated with words like “chaos”, “hate”, and “agression”, suggesting a link between violence and social disorder, while “ordre” (“order”), “policier” (“policeman”), and “condamner” (“to condemn”) evoke the desire for a strong and controlling response from the state. This connects to the second network where the lemma “islamisme” is associated with terms like “terrorisme” (“terrorism”), “radical”, and “guerre” (“war”), reflecting the narrative of a fight against terrorism and extremism; the specific reference to “Charlie Hebdo” strengthens the connection between radical Islamism and known terrorist attacks in France. This type of association once again leads to the stigmatization and alienation of entire communities and social groups; both maps reinforce divisions and alarmism, identifying an “enemy” and justifying severe measures as necessary for the protection of citizens.

Topic 3 addresses concerns about immigration and the need for stricter border control. The Rassemblement National argues that Europe and France must shift from an open-door policy to a tighter management of borders to protect citizens from uncontrolled immigration. It highlights the record of illegal entries into the European continent in 2022, criticizing the current European Union policy. A referendum is proposed to radically revise French immigration policy, aiming to reduce the country’s attractiveness by cutting what are called the “magnets” that attract illegal immigrants, such as generous social assistance offerings. The need to protect Europe’s doors and establish a deterrent immigration policy is underlined as crucial for ensuring internal security and maintaining French national identity. The lemmas of these topics that best meet the objective of identifying words of harmful polarization are the lemmas: “islamiste” (“Islamist”), “violence”, “policier” (“police”), “ordre” (“order”), and “victime” (“victim”) for topic 1; and “immigration”, “migratoire” (“migratory”), and “migrant” for topic 3. Exploring the associations (Figure 11), we see that in the network centered on the lemma “immigration”, the strongest links are with “clandestin” (“clandestine”), “insecurité” (“insecurity”), and “massif” (“massive”); this narrative reinforces the idea that the Muslim presence in France is a source of social disorder and danger. Therefore, it is not surprising that in the second network, the word “ordre” (“order”) has a strong association with words such as “policier” (“police”), “sécurité” (“security”), and “criminalité” (“criminality”), reinforcing the narrative of restoring public order by confronting crime. The words used to support a stricter Immigration policy directly link Immigration and Insecurity to the French Muslim community and advocate for the adoption of more stringent deterrent and control measures. This view emphasizes a repressive approach to immigration, which can once again contribute to an exclusionary rhetoric that could generate social tensions and turn polarization into political extremism.

Topic 7 “National Priorities” discusses national priority in terms of access to work, housing, and social services for French citizens. The most significant words—“priorité” (“priority”), “soigner” (“to care for”), “fraude” (“fraud”), “familial” (“family”), “social”, “impôt” (“tax”), “allocation” (“benefit”), “logement” (“housing”), “jeune” (“youth”), “handicap” (“disability”), “emploi” (“employment”), “hôpital” (“hospital”), “soin” (“care”), “réserver” (“to reserve”), “unjourunlivret”, and “français” (“French”)—reflect the leader’s vision of national solidarity and the preference for French citizens over immigrants. Topic 9 titled “Sovereignty, Identity, and French Values” explores the concept of French sovereignty, the importance of national identity, and the defense of traditional values. Among the most relevant words of these topics with a nationalist vocation are the following: “priorité” (priority), “fraude” (fraud), and “familial” (family), which suggest themes of polarization regarding national priorities, including social and family aspects; and “retraites” (“pensions”), “réforme” (“reform”), and “motion” (“motion”), which reflect polarizing debates and controversies about reforms, particularly pension reforms. “France”, “nation”, and “histoire” (“history”) are keywords that highlight a strong sense of national identity and potential polarization around concepts of sovereignty and French values. The lemma “Priority” and its associated words (Figure 12) emphasize the importance of French national interests in various spheres, such as employment, social security, and housing policies, especially in relation to immigrants. Words like “expulsion” and “nationalité” (“nationality”) once again indicate a tendency towards restrictive immigration policy, while “aide sociale” (“social aid”) and “logement” (“housing”) suggest that such benefits should be reserved for French citizens. This reflects a view where state assistance is seen as a limited good, which must be protected and primarily offered to the French, effectively declaring the existence of first-class and second-class citizens. In the second graph, the word “nation” is surrounded by terms like “histoire” (“history”), “grandeur” (“greatness”), and “identité” (“identity”), which underline respect for cultural heritage and the uniqueness of France. Words like “protéger” (“to protect”), “construire” (“to build”), and “défendre” (“to defend”) suggest the importance of maintaining national sovereignty and preserving identity in the face of external influences, including the European Union, as indicated by the word “Europe”, which is often blamed for internal and external problems.

6. Conclusions

Analyzing a relational concept such as pernicious polarization—beyond its traditional definition—lacks the tools necessary for a comparative study (Lauka et al. 2018). Our contribution provides a comparative exploration of such a concept by analyzing the political discourse through Text Mining and Topic Modeling techniques. This work is an exploration, useful for obtaining the first empirical evidence that will be useful for orienting future work with more sophisticated and exclusively data-driven techniques; this, for example, can be used as a basic study to build an opinion dictionary of the pernicious polarization of the right in the investigated countries. A vocabulary of this type can be useful for investigating pernicious polarization in the future, automatically, through word-based opinion-mining techniques.

To begin with, we focused on right-wing leaders from Italy, France, and Spain. This allowed us to observe the discursive dimensions of pernicious polarization in similar cases from Western Europe, hence identifying the types of grievances they tend to rely on within online communication.

All in all, our results show similar patterns in the narratives of the selected cases and common themes characterizing them. The few differences we found mainly reflect internal political debates. For instance, the predominance of themes such as “Threats and Internal Security” in Spain, “Crime and Justice” in Italy, and “Islamism and Internal Security” in France reflects a tendency to emphasize issues that can fuel anger and division within societies (Gerbaudo et al. 2023). These themes tend to create a clear distinction between ‘Us’ and ‘Them’, often resulting in a less inclusive and more confrontational public dialogue. In addition, the focus on identity issues such as Catalan independence in Spain and national sovereignty in Italy and France amplify cultural and ethnic as well as political divisions. Furthermore, the centrality of political figures such as Salvini in Italy and the emphasis on the strategies of the Rassemblement National in France can contribute to personalizing, therefore further polarizing the political debate.

The discourses of West European right-wing leaders especially rely on three main features: anti-immigration, euroscepticism, and populism (as anti-elite or anti-Government rhetorical attack)—within the frame of sovereignism. The salient discursive categories of polarization mainly refer to populism, cultural values, and national identity, with a few references to economic and political ideology. In other words, the underlying cleavages or latent formative rifts counterpose the nation to the governing elite—belonging to both the national Government and the EU—as well as native-born vs. foreign-born (i.e., racial and ethnic minorities), pro- vs. anti-EU/globalist, and the conservative vs. liberal cleavages. Accordingly, the political—and slightly economic—grievance is predominant and associated with the corruption of an elite that does not represent nor defend national interests. Finally, the cultural grievance constitutes a relevant feature of such discourses and concerns the national identity and values threatened by both the elite and the immigrant “invasion”. In this sense, right-wing leaders portray themselves as defenders of traditional and ethnic values vs. those who promote integration policies and the out-groups too.

Regarding the methodological approach we used, this study presents an initial exploration of the words that best represent pernicious polarization by directly analyzing the discourses of selected populist right-wing leaders of Western Europe. From a methodological point of view, an interesting finding regards the overlap between the themes (and often words) in the speeches of the different leaders across the various languages. Such overlap offers a significant insight into the potential, future creation of a lexicon of polarized words, as this could become a replicable (though not exhaustive) tool for semi-automated analysis of pernicious polarization in political leaders’ speeches.

This study certainly faces limitations. First and foremost, the space available did not permit a comprehensive report on the analysis of the associations of all relevant words within the topic. A second limitation is undoubtedly the absence of a comparison with the narratives of users responding to these speeches. Such responses could introduce additional words to those already examined and, with their commentary, could exacerbate the polarized stances of the leader to the extent of transforming the discourse from merely polarized to radicalized (Acampa 2024). Lastly, a third limitation is the inability to compare these topics and their respective words with those of all political leaders. The latter point suggests further developments of this study, namely, to extend our analysis to the counterpart of the group of leaders we selected. This will allow us to explore pernicious polarization broadly and delve deep into the relational and political character of such a phenomenon through the political discourse.

In the end, to analyze far-right discourses in Western Europe (France, Italy, Spain) this work does not analyze whether these correlations are false and politically motivated. However, exploring the political, sociological, and psychological assumptions underlying these far-right discourses may serve as future work to provide a more accurate picture of how polarization works.