Online Verbal Aggression on Social Media During Times of Political Turmoil: Discursive Patterns from Poland’s 2020 Protests and Election

Abstract

Online aggression and abusive language on social media pose a growing threat to democratic discourse, as they contribute to polarization, delegitimization of political actors, and the erosion of civil debate. While much of the current research relies on computational methods to detect hate speech, fewer studies investigate how online aggression functions discursively in specific socio-political contexts. This study addresses this gap by analyzing patterns of verbal aggression on Facebook and Twitter during two key events in Poland in 2020: the presidential election and the Women’s Strike. Adopting a mixed-method approach (combining sentiment analysis, content analysis, and discourse analysis) and comparing two socio-political events that generated extensive online debate, this study investigates the patterns and communicative functions of hostile and aggressive language on Facebook and Twitter. The study reveals that neutral posts dominated both datasets, but negative and aggressive posts were significantly more frequent during the Women’s Strike, where verbal aggression was used not only to reinforce group identity but also to express moral outrage, trauma, and demands for change. In contrast, aggression during the election campaign was less frequent but more calculated. It functioned as a strategic tool to delegitimize political opponents and reinforce partisan divides. Users employed vitriolic language and profanity as rhetorical tools to undermine authority, reinforce group identity, and mobilize supporters. The study also reveals asymmetric patterns of aggression, with public figures and institutions, particularly the ruling party, Church, and police, being primary targets. The findings have significant implications for understanding the dynamics of online debates and aggression patterns in social media.

1. Introduction

Social networking sites have transformed the ways in which individuals engage with political events and social movements. They serve not only as tools for information dissemination but also as arenas where emotions, ideologies, and collective identities are actively expressed and contested. As part of a broader process of deep mediatization, digital media permeate everyday life and social institutions, thereby reshaping communication patterns, political participation, and power relations (Hepp, 2019). The role of social media in facilitating social mobilization and political protest has been widely studied. A substantial body of research has proved that increased online activity correlates with higher protest turnout (Gallacher et al., 2021), the escalation of both online and offline aggression (Gallacher et al., 2021; Amichai-Hamburger & McKenna, 2006), and an increase in arrests, arson, and vandalism (Mooijman et al., 2018). In this context, the internet plays a clear role in facilitating rapid information dissemination and collective mobilization, thereby enhancing public accountability and shaping responses to perceived injustices and official misconduct (Gao, 2016). On the other hand, social media offers multiple opportunities for interaction with members of various out-groups. Such online contact may promote positive group outcomes, reduce bias, and counteract social divisions (Gallacher et al., 2021; Amichai-Hamburger & McKenna, 2006). However, for such positive results to occur, certain pre-conditions need to be met, such as equal status between groups, shared interests or goals, intergroup cooperation, personal interaction, and support from authorities (Allport, 1954). More often than not, online interaction leads to extensive hostility, insults, vulgarity, and increased animosity, both online and offline (Gallacher et al., 2021; Amichai-Hamburger & McKenna, 2006).

This paper explores the use of verbal aggression on social networking sites in the context of political events. Although there is extensive literature on online hate and vitriol (Polak & Trottier, 2020; Castaño-Pulgarín et al., 2021; Mane et al., 2025), much of this research focuses narrowly on the toxic dimension of verbal aggression, disregarding the underlying functions of such posts and tweets. While it is true that aggressive and spiteful content exacerbate digital violence, online aggression is not a one-sided phenomenon, but it serves a variety of communicative and strategic purposes. Furthermore, existing research tends to focus on a single event or topic (Fan et al., 2021; Almahfali & El-Husseini, 2023; Ezeibe, 2021). This way, it fails to provide a broader context of online aggression. This study addresses that gap by adopting a mixed-method approach that combines sentiment analysis and corpus linguistics with qualitative discourse analysis. It examines and compares two socio-political events in Poland that generated extensive online debate: the 2020 presidential elections and the social protests following the Constitutional Tribunal’s ruling that abortion due to fetal defects was unconstitutional (commonly referred to as the Women’s Strike). These two events were selected because they represent distinct yet emotionally charged socio-political moments: an institutional, polarizing electoral campaign and a grassroots, value-driven protest movement. Analyzing both enables a comparative exploration of how verbal aggression functions in different political contexts. The originality of this study lies in exploring the contextual dynamics of online hostility beyond isolated cases or narrow definitions such as hate speech. Given that verbal aggression serves multiple functions, it is not clear what goals social media users want to achieve when they post in an uncivil manner. By analyzing a corpus of Facebook and Twitter posts, this study investigates how affective and discursive elements interact and contribute to the dynamics of online political discourse. Facebook and Twitter were chosen as platforms of analysis because they capture public, emotionally charged reactions in real time and reflect different modes of political expression and engagement.

The aim of this study is to examine the patterns and underlying functions of verbal aggression in social media discourse during two significant socio-political events in Poland. Specifically, it explores how hostile and aggressive conversational tones are used by social media users and how these expressions reflect broader emotional, ideological and discursive dynamics. The study is guided by the following research questions: (1) What were the patterns of aggression in online discussions during the 2020 presidential campaign and the Women’s Strike? (2) What were the patterns of aggression used by active social media users and public opinion leaders? Based on a six-month dataset (28 April 2020–12 August 2020 and 22 October 2020–31 December 2020) we conducted a content and sentiment analysis of 825,470 Facebook posts and tweets to identify the dynamics of online political discourse.

2. Literature Review

2.1. Online Aggression and Hostile Communication

A wide body of research proves there is increased hostility in online environments, including racial discrimination (Bouvier, 2020; Tsai et al., 2020; Aldamen, 2023), religious intolerance (Everton, 2016; Veiga & Oliveira, 2024), political antagonism (Scheithauer et al., 2018), and gendered-based aggression (KhosraviNik & Esposito, 2018). In their study of Facebook event pages, Gallacher et al. (2021) found that unstructured online environments promote negative and brief interactions with opposing groups that are characterized by hostility, competitiveness, and conflict. Mooijman et al. (2018) identified the correlation between the moralization of protest and increased violence during events. Protests are more likely to escalate when protesters see the demonstration as a moral issue and experience high moral convergence, i.e., they interact online with a large number of like-minded individuals. Hence, social media dynamics may facilitate violence when social media users express moral convictions and use networking sites to assess the moral sentiments of others.

Aggression in online environments takes diverse forms and serves various functions. It ranges from direct insults to subtle expressions of hostility. In digital communication studies, the umbrella term online aggression refers to a broad spectrum of harmful behaviors on digital platforms, including teasing, intimidation, public ridicule, spreading false information, and threats (David-Ferdon & Hertz, 2007). These behaviors can be driven by diverse motives and shaped by different contexts, such as between peers (as in cyberbullying), anonymous users, and public figures facing scrutiny (Zhao & Caverlee, 2018). Verbal aggression, a specific subtype, involves the use of language to express hostility, contempt or disapproval through invective, profanity, mockery, or dehumanizing language. Unlike physical aggression, verbal forms are often normalized online despite significant social consequences, such as reinforcing power asymmetries, fueling affective polarization, and undermining individual well-being. One of its most harmful manifestations is hate speech, defined by the Council of Europe (1997) as language that incites hatred or discrimination based on race, religion, nationality, or other protected characteristics, and which the United Nations (2019) recognizes as targeting individuals or groups on these grounds. Such forms of aggression harm individuals and erode democratic cohesion and pluralism (Domalewska et al., 2025). Consequently, analyzing its forms, functions, and contextual dynamics is essential to understand how digital platforms shape contemporary public discourse.

The use of verbal aggression in online discourse during political events is not only an expression of affect but also a mechanism through which power and ideology are exercised and reproduced. Drawing on van Dijk’s socio-cognitive approach to discourse, aggression on social media can be understood as part of broader processes through which users construct and disseminate social representations, often aligned with ideological positions. According to van Dijk (2000, 2009), discourse serves as a site where in-group identities are affirmed and out-groups are delegitimized through strategies such as positive self-presentation and negative other-presentation. This framework helps explain how emotionally charged and polarized messages on social media reflect underlying cognitive structures, such as schemas, attitudes, and ideologies, that shape how individuals interpret political events and align themselves with group identities. In this sense, online aggression during events such as protests or elections may not simply express personal hostility but may also function to reproduce dominant narratives or challenge existing power relations in the public sphere.

2.2. Emotion-Driven Political Communication

Political campaigns are undoubtedly driven by emotions to attract media attention, engage voters, and influence both attitude formation and voting behavior (Crabtree et al., 2020; Żakowska & Domalewska, 2019). Emotional campaigns are highly effective in conveying the message; therefore, candidates often rely on emotional communication frames to gain wider public support. As Zappettini et al. (2021) argue, the emotionalization of media discourse reflects a broader shift from rational, logic-based communication toward emotionally charged messaging. In this context, political actors increasingly use emotional frames, such as fear, hope, resentment, and moral outrage, not only to persuade but also to polarize and mobilize. The power of these strategies lies in their ability to bypass rational deliberation and elicit affective responses, which are then amplified through both traditional and digital media. As a result, emotional campaigns have become a central feature of contemporary political communication, enabling candidates to forge emotional connections with supporters and drive public engagement.

Emotions affect information processing, so the impact of the message depends largely on how it is framed. Fear-based campaigns stimulate vigilance, promote reliance on contemporary evaluations, and facilitate persuasion (Nai et al., 2017; Żuk & Żuk, 2020). Messages that appeal to anxiety tend to make voters more deliberative and encourage them to look for new solutions to political conflict. Anxious citizens seek information to resolve uncertainty that has either been caused by current events or evoked in the campaign. In searching for protection, they may support policies that oppose their partisan preferences, provided these policies address their perceived need for security (Albertson & Gadarian, 2015). Thus, political campaigns that appeal to fear not only attract wider public attention but also encourage voting behavior contrary to voters’ habitual political alignment (Marcus & MacKuen, 1993). At the same time, anxiety affects information seeking, as anxious individuals tend to consume news that adopts a conflict frame and emphasizes threatening or negative content (Albertson & Gadarian, 2015).

Aversion-driven campaigns reinforce voters’ partisanship, make them more focused, and shield them from alternative arguments, because hostile messages discourage citizens from looking for more information. Such messages often use blame as a rhetorical strategy for positive self-presentation and negative other-presentation (Hansson, 2015). As a result, they drive group polarization, which gradually contributes to a different way of perceiving reality, both at the conceptual and normative level. In-group opinion differences are reduced, and group cohesion is strengthened. At the same time, the gap between the supporters of opposing parties widens. Differences are increasingly perceived not in ideological terms, but in terms of fundamental values and identification with political groups (thus breeding strong aversion and discriminatory attitudes towards political opponents, while individuals sharing similar views are regarded more positively). Mutual dislike and distrust between opposing electorates grow, and political opponents are increasingly avoided. The radicalization of the political scene leads to one-sided debate and harsh rhetoric, which prevents constructive dialogue and reaching consensus, as there is little integration of diverse viewpoints. This fragmentation creates societal conflict, as supporters of opposing parties are seen as members of a hostile out-group.

Despite the negative social consequences, polarization is often used as an instrument of power and domination and as a strategy to achieve long-term political goals. Hostility may therefore be deliberately incited to consolidate the electorate and weaken political opponents (Somer & McCoy, 2018). A highly polarized political environment significantly influences voters’ perspectives and decision-making, while simultaneously reducing the impact of facts on their opinions. Thus, voters deepen their trust in the politicians and parties they support, who in turn feel less compelled to provide substantive arguments to justify their positions. Simplified reasoning and prejudice frequently underpin party rhetoric. Voters in highly polarized environments cling to their views, ignore legitimate and evidence-based arguments, and become inflexible and intolerant (Druckman et al., 2013). The attachment to simplified and non-substantive arguments is also reinforced by affective polarization, i.e., a situation when the viewpoint is influenced by aversion and contempt for opponents than by logical argumentation. Political elites can therefore gain more by fueling animosity, hatred, and fear than by using substantive debate. This strategy may be particularly attractive for parties that lack specific solutions and benefits for voters and instead rely on strengthening the polarization of the political scene (Wilson et al., 2004).

However, electoral campaigns are not entirely driven by negativity. They also appeal to hope and generate voters’ enthusiasm, which particularly resonates with voters who have strong partisan beliefs and attitudes (Gerstlé & Nai, 2019). Enthusiastic political messages influence candidate preferences and get the public involved in the campaign (Marcus & MacKuen, 1993). Positive campaigns attract interest from both voters and the media (Gerstlé & Nai, 2019). The mobilizing effect of enthusiasm has also been empirically confirmed in Neyazi and Kuru’s (2024) study, which found that exposure to emotionally congruent poll messages (e.g., one’s preferred party leading in the polls) triggers enthusiasm, which in turn strengthens intentions to engage in campaign activities and vote.

Apart from the varying effects of emotion-driven campaigns, differences in behavior correlate with political party positioning and target audiences. Negative emotions, such as anger, fear, and anxiety, are more frequently used by extremist factions, whereas ideologically moderate parties rely on a more positive tone in their messages. Incumbent parties generally use more positive sentiment than opposition parties (Crabtree et al., 2020; Rodriguez-Ibanez et al., 2021). Campaigns aimed at loyal voters tend to appeal simultaneously to aversion and to hope and enthusiasm. Populist politicians, as well as far-right and nativist candidates, tend to run more aggressive campaigns that are focused on negativity (Gerstlé & Nai, 2019). Populist leaders frequently deliver distressing and provocative messages, favor spectacle over substance, and employ metaphorical language and vivid imagery. Their discourse often relies on stereotypes and references to enemies or sinister forces, reinforcing a divisive and emotionally charged narrative.

Campaigns become more negative during periods of economic downturn (Crabtree et al., 2020). Uncivil messages are often directed at political opponents or focused on timely and contentious issues, such as immigration and social policy, that attract media coverage and public attention. Adversarial messages are characterized by varying levels of aggression that exaggerate specific issues and contribute to affective politics, which is driven by emotion rather than rational argument and factual analysis. Negative messages drive greater user engagement in social networks so that they spread farther and deeper (Antypas et al., 2023; Klinger et al., 2022). Negativity incites angry and divisive posts; therefore, campaigns framed around conflict often elicit defensive and hostile reactions. However, the general public overall exhibits a different pattern of online behavior than political actors. Posts made by ordinary users are significantly more positive on average. What is more, a difference has been observed in the sharing behavior of negative content: politicians’ negative tweets are retweeted far more frequently than those of regular users. This suggests that negative messages from political figures are more popular and have a wider reach than similarly toned content from the general population (Antypas et al., 2023).

3. Materials and Methods

The study was based on a mixed-methods approach to investigate two corpora of Facebook posts and tweets that featured hashtags popular during two significant socio-political events that took place in Poland in 2020: the presidential elections and the protests following the Constitutional Tribunal ruling that abortion due to fetal defects was unconstitutional (the so-called Women’s Strike). The first corpus related to the presidential elections was created by collecting posts and tweets containing hashtags related to the event itself and the eight most popular presidential candidates1 (2,300,877 posts and tweets harvested between 28 April and 12 August 2020). The second corpus consisted of posts and tweets published from 22 October to 31 December 2020 containing hashtags associated with the Women’s Strike that were trending during that period (713,984 posts and tweets). The data were collected from public domains; therefore, user consent was not required for the analysis. The corpus was limited to original posts published on Facebook and Twitter. User comments in response to these posts were excluded from the analysis.

The level and patterns of verbal aggression in discourse were examined using a mixed-methods approach. Quantitative analysis was conducted on both corpora and involved sentiment analysis and the methods of corpus linguistics: frequency lists, collocations, and concordance lines. The 20 most frequently used uncivil and aggressive words and phrases were further analyzed to examine their usage in context. The Sketch Engine tool was used to examine language patterns and identify key features of hostile discourse. This provided a statistical profile of aggression that helped explain the underlying purposes of verbal harassment and hostility in online debates triggered by different socio-political events. Sentiment analysis was carried out using the lexicon-based approach, which determines the sentiment polarity of a tweet or post by evaluating the contextual use of opinion words. Each post was classified as positive, negative, or neutral.

4. Results

4.1. Sentiment Analysis

First, sentiment analysis was conducted to assess the level of aggression in the corpus, as shown in Figure 1. This is an effective method for identifying the level of incivility, as it reveals the opinions, emotions, and attitudes expressed by social media users (Shaik et al., 2023).

Neutral posts and tweets dominated the sample, accounting for 87.17% of those related to the presidential campaign and 84.45% of those related to the Women’s Strike. The latter generated nearly twice as many positive (4.49%) and negative (13.05%) posts compared to the presidential campaign (positive: 1.32%; negative: 5.97%). These results indicate that the discussion surrounding the protest was marked by a higher degree of emotional intensity (both positive and negative), reflecting the contentious nature of the issue. In fact, the prevalence of negative sentiment was expected, as the Women’s Strike expressed widespread public frustration not only with abortion legislation, but also with the conservative government and its pandemic-related restrictions. Furthermore, posts related to the Women’s Strike generated significantly more engagement. The most popular post during the campaign (by the incumbent President, Andrzej Duda on 8 July 2020) received 339,062 engagement points, whereas the most widely shared post during the Women’s Strike (by Sok z Buraka, a popular Polish satirical and provocative fan page, on 28 October 2020) received 416,483. Engagement statistics further point to this difference: the median number of reactions during the campaign was 1 (mean: 443.55), while during the protest, the median rose to 6 (mean: 849.48). These results indicate that the protest discourse elicited broader public attention and emotional involvement.

During the campaign, negative posts attracted more engagement than neutral ones, which suggests the presence of affective polarization, likely driven by ideological diversity in the dataset. In contrast, the Women’s Strike was marked by emotional intensity and widespread solidarity rather than division. During this event, neutral content generated more engagement than negative posts, which indicates a more unified discursive environment. Aggressive content appeared twice as frequently during the protest as during the campaign, though it was less widely shared. This may suggest that highly emotional content does not automatically translate into broader reach. The findings therefore do not support the assumption that a high degree of emotionality contributes to the virality of posts and tweets (Wang & Wei, 2020), although previous studies have shown that a negative tone, in particular, tends to increase the readership and user engagement (Gerstlé & Nai, 2019; Yuan et al., 2022).

4.2. Aggression in Posts and Tweets

An in-depth quantitative analysis of frequency lists and collocation analysis provided insight into frequent topics in the corpus and the use of verbal aggression online. Several recurring themes and discursive strategies were identified. Posts and tweets frequently quoted politicians’ statements, referred to campaign rallies, and commented on politicians’ private lives. Rafał Trzaskowski, the candidate representing the opposing coalition, stood out in this regard, as he systematically made aspects of his private life public. He posted about his daily routine and referenced his children. His wife, Małgorzata Trzaskowska, also made public speeches regularly. Trzaskowski’s visibility was contrasted with the communication style of Andrzej Duda, the incumbent candidate from the ruling party, who was criticized for the limited public presence of his wife and daughter. In this context, the pattern reflects the personalization and emotionalization of political discourse, where attention shifts from substantive policy issues to the private lives and personal images of candidates. As a result, public debate becomes more superficial and lacks meaningful discussion of policy proposals and governance strategies. This trend is also evident in pre-election campaigns conducted by mainstream media (Stasiak-Jazurkiewicz, 2021). Similarly, research conducted during the 2013 German federal election campaign shows that both politicians and internet users rarely discussed policies, but prioritized discussion about campaign-related events, making the political debate online less diverse (Stier et al., 2018).

The analysis of the occurrence of aggressive behavior indicates two broad categories of uncivil publications:

• Name-calling, i.e., the use of abusive language in references to politicians and political parties, in particular the incumbent president, the leading opposition candidate, and the ruling party:
  Sample post from 27 November 2020:

  Morally screwed up and spiritually stagnant dilettantes #poland #kaczyński #(…) #duda #morawiecki (…)

• Rude or disrespectful remarks directed at presidential candidates and political parties:
  Sample tweet from 7 July 2020:

  After bringing the @presidentpl office down to the level of @USAmbPoland, now this stupid candidate @AndrzejDuda who still holds this office brings it much lower, because down to the level of German newspapers and rags.

  Sample tweet from 16 July 2020:

  Keep inciting against @trzaskowski_—that will surely help you. After posts like this, I’ve had enough of you for the whole week!!!! Shall I remind you who made @AndrzejDuda what he is, who in 2015 was urging people to vote for him!!

Uncivil posts are emotionally charged messages that often rely on sarcasm, contempt and moral judgement. They are typically confrontational and aim to discredit individuals rather than engage in substantive debate.

While neither profanity nor highly aggressive language appeared among the 100 most frequent lexical items, two uncivil name-calling terms were frequently used in reference to candidates: story-teller (used 11,580 times) and chicken (used 5663 times):

Sample tweet from 1 July 2020:

(…) #PresidentofPolishAffairs #Story-teller2020 #RafałDon’tLie #Duda1round #LGBT #We’veHadEnough #Polandconnectsus

Sample tweet from 13 June 2020:

(…) #PresidentialDebate2020 #Story-teller2020 #RafałDon’tLie #RafałChicken #LGBT #We’veHadEnough #Polandconnectsus #Rafał’sSamba

These name-calling expressions appeared in posts and tweets that included several similar hashtags and were posted by users who reproduced the same or similar content repeatedly over a short period of time. They serve specific rhetorical functions. Story-teller implies deception, manipulation, or insincerity and is used to suggest that the person so labeled is untrustworthy or populist. Chicken, on the other hand, connotes cowardice, weakness, or an unwillingness to take responsibility. Both terms evoke feelings of contempt, mistrust, and ridicule, which reinforces negative perceptions of political opponents. Their repeated use contributes to the emotional polarization of public discourse, shifting attention away from substantive policy discussions and toward personal attacks and symbolic delegitimization.

During the Women’s Strike, nearly one in seven mentions was classified as negative. Posts were more diverse in tone and content, although expressions of support for the protest prevailed. Supportive posts reported on demonstrations, announced the place and time of protests, explained the goal of the movement, and amplified messages from public figures and celebrities who endorsed the cause.

Sample tweet from 28 October 2020:

When someone grabs you by your ankle, you say enough is enough. And when they grab your crotch, you scream #get the fuck out!. This is the language of a cornered woman #women’sstrike #SentenceOnWomen #women’sprotest #women’shell https://www.wysokieobcasy.pl/Instytut/7,163391,26444649,gdy-ktos-cie-lapie-za-kostke-mowisz-dosc-gdy-za-kolano.html?disableRedirects=true (accessed on 22 May 2021). The very problem of this protest could easily be solved. It would be enough to force men by law to tie the vas deferens. Then the one who would like to start a family…”.

This tweet serves as a powerful example of emotionally charged language used to express indignation and resistance. Phrases like get the fuck out and metaphors of physical violation (grabs you by your crotch) convey deep frustration, anger, and a sense of personal and collective violation. The reference to a cornered woman evokes desperation and the need for defensive action, while hashtags such as #women’shell and #SentenceOnWomen frame the protest as a fight for dignity and autonomy. Emotion-laden expressions reflect not only opposition to the court ruling but also solidarity and a broader demand for structural change.

Another important theme in these posts is clear opposition to three main institutions: (1) the ruling party and its individual representatives due to their anti-abortion stance, (2) the Church, both for its opposition to abortion and for abuses and criminal acts committed by clergy (e.g., pedophilia), and (3) the police, for their use of disproportionately harsh coercive measures during the protests.

Sample tweet from 13 December 2020:

I have not seen a greater social pathology than PiS. Lies plus propaganda and sewage. https://twitter.com/Link anonymised.

Sample tweet from 25 October 2020:

The church, as an institution, has blood on its hands and is co-responsible for the current protests. I am absolutely satisfied that the protests are so massive and reveal true emotions of society.

#fuckoff #women’s protest #blackprotest #Women’sstrike #women’shell #abortionban https://twitter.com/Link anonymised.

Sample Facebook post from 30 October 2020:

A moment ago in Katowice (you can see it live on FB) the police used gas to attack peaceful demonstrators. There were children and elderly people in the crowd. It’s disgraceful behavior. I will demand an explanation from the police for this groundless aggression.

It is noteworthy that one in five posts in this sample used a hashtag popular at the time that related to the Women’s Strike to promote content unrelated to the protests. Most of these posts were related to the everyday activities of individual users, who meticulously documented their daily routine on social networking sites. Their posts generated very limited engagement (39.74% received no likes and 85.26% received 4 likes or fewer).

Table 1. Tweets using a popular Women’s Strike hashtag but not directly related to the protest posted on 28 November 2020 by one user.

Date of Publication	Content of the Tweet
28 November 2020 07:43:59	I got you, moots
28 November 2020 08:56:52	I’m jealous again:))
28 November 2020 08:57:11	why won’t he ever talk to me on the phone
28 November 2020 09:14:42	I had a bad dream again
28 November 2020 09:14:51	third day in a row
28 November 2020 10:01:10	Dominic is coming soon, and I’m still sad
28 November 2020 10:07:56	@Anonymous-User Me too, but I remember only one :(
28 November 2020 16:00:59	I’m sad because Dominic has already left
28 November 2020 16:35:46	I just want to stop worrying about everything and just be happy
28 November 2020 16:47:19	he’s ignoring my tweets again, I’m feeling terrible
28 November 2020 16:51:06	my friend hasn’t written back to me for a few days and she reads so hah;) I will do the same
28 November 2020 16:51:59	exactly https://twitter.com/Link anonymised
28 November 2020 16:52:55	cute https://twitter.com/Link anonymised
28 November 2020 17:06:11	me too ;(https://twitter.com/Link anonymised
28 November 2020 17:43:18	I feel better after the walk
28 November 2020 17:47:37	and yet walking helps
28 November 2020 18:23:55	hot https://twitter.com/Link anonymised
28 November 2020 18:37:59	my head…. is bursting
28 November 2020 20:18:51	look what I got, I baked vege gingerbread https://Link anonymised
28 November 2020 20:50:38	good night https://Link anonymised

displays 21 tweets posted by one such user on 28 November 2020. The content has been anonymized to protect the user’s privacy.

As shown in Table 1, the user includes hashtags related to the Women’s Strike; however, the content of the tweets is not related to this movement at all. It can be assumed that the user either seeks to increase the visibility of her posts or expresses support for the social movement, but, absorbed in personal problems and everyday matters, focuses instead on documenting her own life.

Another trend identified in the sample was the use of Women’s Strike-related hashtags to promote religious content. This was observed on two Twitter accounts whose tweets generated very low engagement: 90.2% received no likes and none were liked by more than two users. The example tweet was published by a user who posted 23 entries containing Women’s Strike-related hashtags on 20 November 2020.

Sample tweet from 11 November 2020:

Let us prepare for the coming of the Messiah. #MidweekMegaWord #PiS #POLNED #Seym #Covid_19 #Women’sStrike #COVID19 #christmas #london #TheChase #PS5

These accounts posted Bible verses, shared information about online religious services, or called for conversion by interpreting major events (e.g., the pandemic or the Women’s Strike) as signs of the end of the world.

Next, we conducted a corpus analysis of posts published during the Women’s Strike. The frequency analysis shows that the most common phrases were blunt and uncompromising (e.g., women’s hell, sentence imposed on women, this is war) and, in some cases, openly vulgar (get the fuck out). Two frequently occurring phrases were associated with the media (woronicza, behind the vision), and one with a political party (pis).

Posts related to the Women’s Strike are direct and focused, rarely straying from the central issue of the protest. What stands out most is the use of vulgar language—a vulgarism (fuck off) appears among the ten most frequently used words in the corpus. Such invectives typically serve to reinforce the message and emphasize strong emotions. Therefore, popular hashtags associated with the Women’s Strike often included profanities, such as, e.g., #fuck off, #fuckPiS. Swear words were used as instruments of verbal aggression and functioned as a substitute for physical aggression. They were often directed at individuals, especially politicians, and at the ruling Law and Justice party. In these contexts, the language served to insult or express disrespect and contempt. However, these expressions do not qualify as hate speech. Rather, they function as symbolic acts of degradation or even destruction of the message’s target. Thus, on the one hand, negative evaluation carries a quasi-magical function: by uttering invectives, the speaker expresses a wish or desire, using language as a destructive formula. These messages illustrate the magical power of language, where the spoken curse is intended to alter reality to bring about a desired change in the described situation.

On the other hand, verbal aggression can function as an instrument of power, reinforcing or reproducing existing power relations in society. For example, numerous invectives aimed at minority groups reflect and reinforce the dominance of one group over another. In this case, hate speech helps construct a social reality in which subordination is normalized and harm against targeted groups is trivialized. This subordination contributes to social inequality and discrimination and constitutes a form of symbolic violence. This use of verbal aggression is evident in the research sample. Invectives often function as attempts to reverse power dynamics. By addressing the ruling party in vulgar terms, users express contempt, diminish the party’s perceived legitimacy, and undermine its position of authority. This pattern suggests that verbal aggression is politically charged and closely tied to party identification, as it is directed primarily at the ruling party. In doing so, aggression shapes public perceptions and deepens polarization around contested events. However, some Women’s Strike-related tweets that included invectives directed at the Church crossed into the realm of hate speech, as they conveyed hostility toward a protected group and contributed to the normalization of religious discrimination.

At the same time, the demonstrating women are not seeking power for its own sake. On the contrary, they are attempting to break free from a subordinate position and assert their role in society as equal members. This is evidenced by the widespread use of expressions such as women’s rights are human rights, women decide, choice not prohibition. Women, therefore, do not use language as a form of symbolic violence, but they aim to change social reality (invoking the previously discussed magical function of language) and to strengthen their position as equal and empowered participants in society. There is no evidence in the sample of verbal aggression directed at men or other social groups. In this context, invectives serve an instrumental role, but do not constitute hate speech. However, profanity is also common in posts related to everyday situations that are not emotionally charged. The sample includes numerous short but vulgar posts.

Sample tweet from 16 November 2020:

Fuck it, Shit XDDDDDDDD https://Link anonymised.

The prevalence of this type of post proves the progressive brutalization of language and the widespread use of incivility in online discourse.

Finally, we conducted a qualitative analysis of 5853 posts by the 40 most active users from a sample of social media accounts that posted between 25 and 200 times during the analyzed period. At this stage of the study, only individual user accounts were included, i.e., accounts with identifiable users (i.e., unique usernames). We excluded community accounts from the analysis, including those belonging to media companies, political party affiliates, local branches of the National Women’s Strike movement, and local communities. First, we determined the ideological orientation of users. During the presidential campaign, four dominant user groups were identified: (1) supporters of Rafał Trzaskowski, the main opposition candidate and current mayor of Warsaw (70% of users); (2) supporters of Andrzej Duda, the incumbent president (15% of users); (3) supporters of Krzysztof Bosak, the candidate from another opposition party (5%); (4) other users, including two neutral accounts and two accounts opposing Rafał Trzaskowski, but without sufficient features to clearly determine their views. All posts by these users were related to the activities or public statements of presidential candidates. Some posts also commented on other prominent politicians from the candidates’ respective parties who received media attention during the campaign period. Despite ideological differences, users in this sample focused on similar themes and appeared to be politically engaged individuals. A portion of the posts, especially those supporting Rafał Trzaskowski, reported on campaign rallies, described the atmosphere at events, and expressed gratitude for public support. Two of these accounts were deleted following the end of the campaign. Further analysis of user profiles revealed that most belonged to middle-aged individuals living abroad who remained highly involved in domestic politics and consistently posted content favorable to Rafał Trzaskowski.

Next, the study analyzed the use of aggressive language in posts to determine who was responsible for acts of verbal aggression. However, the results are inconclusive. On the one hand, it was found that aggressively critical expressions targeting the PiS party and President Andrzej Duda were often posted by users who also published numerous positive comments about Rafał Trzaskowski, the Civic Coalition’s presidential candidate. This may indicate that supporters of Rafał Trzaskowski and the Civic Coalition were responsible for launching aggressive attacks and played a significant role in the hate speech observed during the presidential campaign. On the other hand, after examining the ideological affiliations and electoral preferences of the users whose posts were analyzed, this conclusion seems unjustified. Both supporters of Rafał Trzaskowski and President Andrzej Duda posted a comparable number of positive and negative messages (with Duda’s supporters producing slightly fewer negative posts and slightly more positive ones). The highest level of verbal aggression was observed among supporters of Krzysztof Bosak, the Confederation’s candidate (one in ten posts from this group was classified as negative). They contributed the fewest positive messages overall. Accounts belonging to opponents of Rafał Trzaskowski’s were also present in the sample; one in four posts from this group displayed signs of verbal aggression. Aggressive mentions often included vitriolic language and represented examples of asymmetric aggression, initiated by internet users who targeted public figures, mostly decision-makers and political parties.

5. Discussion

This study sheds light on the use of verbal aggression in online debates during two significant political events in Poland. The findings indicate that while neutral posts dominated in both datasets, the volume of both positive and negative content doubled during the Women’s Strike compared to the presidential election. In both datasets, negative posts outnumbered positive ones. Aggressive posts and tweets were filled with vitriol and manifested asymmetric aggression initiated by internet users who attacked public figures, mostly decision-makers and political parties. These expressions rarely contributed to substantive discussion but instead fostered a more hostile public debate. Marked with scathing criticism, political discourse fills social networking sites around the world, including in the United States (Zhao & Caverlee, 2018), the Netherlands (Korsten, 2020), Germany (Pepiak, 2020), Poland (Żakowska & Domalewska, 2019), Great Britain, China, and Russia (Trottier et al., 2020).

Posts containing both positive and negative sentiment tend to generate more emotion and thus align with the attention economy-based business model of social networking sites (Ciampaglia et al., 2015). Algorithms prioritize and amplify emotionally charged narratives to capture user attention and drive engagement, which in turn increases platform traffic. This creates a spiral of negativity and aggression that undermines the quality of public debate. It may also help explain the surge of aggressive posts in the datasets. Although aggression was more frequent during the protest period, it did not always lead to broader dissemination, as neutral posts often reached a wider audience. This finding is consistent with (Klinger et al., 2022), which shows that while negative emotionality may capture attention, it does not necessarily translate into higher engagement or broader reach.

Sentiment analysis also revealed how specific forms of verbal aggression were used in context. During the Women’s Strike, uncivil posts were dominated by strike-related commentary, political and religious opinions, and everyday topics. The discourse was filled with uncompromising and vulgar invectives that reinforced the message and aroused strong emotions. The intense anger released during the social protests was reflected in the use of highly expressive language. In this context, expressing emotions through crude language was more important than linguistic precision. The actual content became secondary, and rational arguments lost relevance. The language used during the protest expressed, on the one hand, a sense of freedom and liberty, and on the other, trauma and powerlessness (Kuligowski, 2021). It had to be blunt, first, as a form of final renunciation of obedience, and second, to stand out in the public sphere. Furthermore, emotional expressions of anger and anxiety can strengthen group cohesion and drive political mobilization, especially in contexts perceived as threatening or unjust (Albertson & Gadarian, 2015; Syfers et al., 2024; Cloudy et al., 2024).

While much of the verbal aggression was directed at political leaders and institutions, a subset of posts went further. Some of the tweets directed at the Catholic Church, for example, contained generalized invectives and hostile stereotyping. These expressions crossed the threshold of legitimate protest and constituted hate speech, as they targeted a protected institution, reinforced discriminatory narratives, and contributed to the normalization of religious hostility in the public sphere.

The emotionally charged and adversarial style of communication significantly contributes to the polarization of public opinion. Verbal aggression not only reflects group-based identity motives but also actively shapes engagement patterns in digital discourse. It reinforces in-group cohesion while intensifying hostility toward political out-groups, making compromise less likely and ideological divisions more entrenched. Content expressing out-group animosity is far more likely to go viral than content focused on in-group solidarity or rational argumentation (Rathje et al., 2021). As a result, individuals interpret events through partisan lenses. Public discourse becomes increasingly fragmented and antagonistic, with opposing camps less willing to engage across ideological divides.

The personalized and emotionalized nature of political discourse, evident in the frequent references to candidates’ families and private lives, confirms broader trends identified in the literature (Zappettini et al., 2021). Public debate has increasingly shifted away from policy-centered content and toward identity-based narratives, in which political legitimacy is constructed through affect, authenticity, and symbolic representation. The findings are in line with research showing that voters increasingly engage with political actors based on emotional appeal rather than rational deliberation (Weismueller et al., 2022). Verbal aggression directed at political figures, especially those in power, can also be understood through the lens of van Dijk’s (2000, 2009) socio-cognitive theory. In this framework, discourse is a vehicle through which social actors construct ideologically charged representations of in-groups and out-groups. Aggressive language serves not only to express contempt but also to delegitimize authority and invert traditional power relations. In the case of the Women’s Strike, this rhetorical strategy was used to challenge institutional dominance and to assert the right to visibility and voice. Yet, as the findings show, language was not used to elevate protesters to a privileged position. On the contrary, women sought to break out of their subordinate status and strengthen their role in society as equal, not superior, members. The language used was therefore not an instrument of symbolic violence but a tool for social change and for reinforcing women’s position as equal and self-determining members of society. Not all verbal aggression should be interpreted as incivility in a normative sense; in some contexts, it may function as a communicative response to symbolic exclusion or structural inequality. Here, speech takes on a magical function of a wish expressed through a destructive formula. Crude expressions illustrate the perceived power of language: the curse is uttered as an attempt to conjure a new reality or bring about the situation being described. Furthermore, vulgar speech empowers protesters, instilling courage and a sense of strength, while also uniting them as a collective. Klyus (2021) analyzed slogans with invective functions and examined their semantic fields. She found that invectives were most often used to convey contempt, stupidity, regret, aggression, ignorance, clumsiness, cruelty, primitivism, and meddling. Offensive language expresses the speaker’s strong emotions and serves a ludic function. Klyus emphasizes that the use of crude expressions as entertainment helps bind society together and produces a therapeutic effect. Drawing on van Dijk’s (2000, 2009) socio-cognitive framework, these findings show that verbal aggression during political events reflects broader ideological positioning and identity construction. The patterns observed suggest that aggression functions not merely as emotional release, but also as a discursive tool through which users contest power relations and express ideological alignment.

6. Conclusions

The purpose of the present study was to examine the patterns of aggression in online discussions during the 2020 presidential campaign and the Women’s Strike. In the corpus, hate speech was identified in some posts and tweets directed at the Church; however, the predominant form of aggression took the shape of vitriol, i.e., asymmetric aggression initiated by internet users who attacked public figures, usually political decision-makers and political parties. Pejorative posts did not provoke substantive discussion; rather, they contributed to the degradation and, in some cases, brutalization of public debate. During the presidential campaign, the posts accounted for almost 6% of the sample, while during the social protests, this number tripled (whereas the number of positive posts merely doubled). The increase in both positive and negative sentiment in the Women’s Strike-related content may be attributed to the fact that the protests revolved around deeply held values and provoked intense emotions, which were expressed using strong language as well as uncompromising and vulgar invectives that reinforced the message. The aggression embedded in these words was intended to insult or express the protesters’ disrespectful or contemptuous attitude toward political decision-makers. In this sense, verbal aggression functioned as a tool to devalue decision-makers and diminish their position of power. At the same time, the use of harsh language gave protesters a sense of strength and courage, helped unify them under shared banners, and released collective emotional energy.

These findings have significant implications for understanding the dynamics of online debates and the patterns of aggression on social media. Gaining insight into these processes is essential for recognizing the nature of social threats present in cyberspace. The findings suggest that opinion leaders play a critical role in shaping a healthy social media ecosystem. Their posts have a wide reach; therefore, they can promote values and encourage attitude change more effectively than individual users. They may initiate public debates that foster unity among internet users and contribute to building social cohesion. To a large extent, it is up to public opinion leaders to determine whether online discourse is ground in constructive, inclusive dialogue that respects different viewpoints and seeks mutual understanding across ideological divides, or whether it contributes to deepening polarization and intensifying aggression toward political opponents.

While this study provides valuable insights into the discursive and affective dimensions of online verbal aggression, it is limited to two social networking platforms, Facebook and Twitter. These platforms differ in affordances, user demographics, and communicative practices, which may not fully represent the dynamics on other social media, such as Instagram, TikTok, or YouTube. The dataset reflects only the Polish-language discourse and is confined to two specific political events, which may limit the generalizability of the findings to other cultural or political contexts. Additionally, the analysis is based on public posts and does not capture aggression that may occur in private or semi-private online spaces. Another limitation is that the analysis was restricted to original posts on Facebook and Twitter; user comments were excluded. As a result, this study does not account for the dynamics of interaction or the ways in which aggression may be amplified or contested in comment threads. These limitations offer directions for future research, which could broaden the platform scope, include cross-cultural comparisons, and incorporate more in-depth qualitative or ethnographic methods to better capture the complexity of online aggression.