HezbollahFootnote 1 is a well-known resistance actor in Lebanon.Footnote 2 Founded during the Israeli occupation of Lebanon, its armed wing remains dedicated to resisting Israel (Hezbollah 2009, 2023). While various countries classify the movement as a terrorist actor, many Lebanese people view Hezbollah as a resistance actor (Daher Reference Daher2019, Reference Daher2016; Haddad Reference Haddad2006; Khashan Reference Khashan2019; Shaery-Eisenlohr Reference Shaery-Eisenlohr2008), a perception reflected in Hezbollah’s own rhetoric and self-perception (Hezbollah 2009, 2023).
This article studies speeches by Hezbollah’s leaders to investigate the expressions of emotions and affect in resistance rhetoric. The findings on positivity and positive emotions contrast with common assumptions about resistance actors. They also depart from Western media presentations of Hezbollah, which link the movement to hate and negativity (e.g. ABC News 2024; The Free Press 2024), while overlooking the importance of positivity and trust.
Research traditionally associates resistance behaviour with negative emotions, such as anger (Grant et al. Reference Grant, Abrams, Robertson and Garay2015; Jasper Reference Jasper2014; Smith et al. Reference Smith, Pettigrew, Pippin and Bialosiewicz2012; Van Zomeren et al. Reference Van Zomeren, Postmes and Spears2008) or frustration (Berkowitz Reference Berkowitz1989; Davies Reference Davies1962; Gurr Reference Gurr1970; Khashan Reference Khashan2003), triggered by social and political grievances. While this research is often situated within the broader conflict literature, a more precise engagement with social movement theory (SMT) offers valuable insights into the emotional and communicative dynamics of collective action. SMT has long emphasised the role of emotions, not only in mobilising participants (Jasper Reference Jasper and Goodin2008, Reference Jasper2011; Van Ness and Summers‐Effler Reference Van Ness, Summers‐Effler, Snow, Soule, Kriesi and Mccammon2018) but also in shaping the framing and communication strategies of social movements and militant groups (Benford and Snow Reference Benford and Snow2000).
Building on this perspective, more recent findings suggest an important additional motivational role of positive emotions (Bettarelli et al. Reference Bettarelli, Close and Van Haute2022; Cohen-Chen and Van Zomeren Reference Cohen-Chen and Van Zomeren2018; Dornschneider Reference Dornschneider2021; Jasper Reference Jasper2018; Pearlman Reference Pearlman2013; Sabucedo and Vilas Reference Sabucedo and Vilas2014; Van Troost et al. Reference Van Troost, Van Stekelenburg, Klandermans and Demertzis2013), including emotions such as trust (Benson and Rochon Reference Benson and Rochon2004; Weipert-Fenner et al. Reference Weipert-Fenner, Rossi, Sika and Wolff2024). While the role of negative emotions like anger has been shown in studies on communication by politicians (Van’t Riet et al. Reference Van ‘t Riet, Schaap, Kleemans, Veling and Lecheler2019), news coverage (Wahl-Jorgensen Reference Wahl-Jorgensen2018), protestors (Çakmak et al. Reference Çakmak, Kara and Sakarya2024) and violent networks (Vergani and Bliuc Reference Vergani and Bliuc2015), it is unclear whether the important role of positive emotions is reflected in resistance communication as well.
This article adds a unique comparison of negative emotions of anger with positive emotions of trust. Existing research suggests that trust is linked to group cohesion (Chenoweth Reference Chenoweth2023), stability (Frye and Borisova Reference Frye and Borisova2019) and mobilisation for resistance movements (Weipert-Fenner et al. Reference Weipert-Fenner, Rossi, Sika and Wolff2024). Building on this literature, this article shows, for the first time, that resistance communication is shaped by trust, and that expressions of trust are employed in relation to the in-group and group ideology. This suggests that resistance actors may employ trust strategically to increase their appeal among their audience.
While there is a growing interest in affect in studies of communication (Young and Soroka Reference Young and Soroka2012), very few studies of resistance specifically examine affect – leaving the role of affect in resistance communication largely unexplored. In psychology, affect is understood as a concept distinct from emotions (Russell and Barrett Reference Russell and Barrett1999), referring to positive or negative ‘feeling states’ that influence cognitive processes and political judgement in unconscious ways (Lerner and Keltner Reference Lerner and Keltner2001; Lodge and Taber Reference Lodge and Taber2005). The psychology literature shows that positive affect promotes flexible, benefit-oriented and thorough cognitive processing (Isen Reference Isen2001; Slovic et al. Reference Slovic, Finucane, Peters and Macgregor2007), which supports action-oriented and risk-embracing decision-making (Fredrickson Reference Fredrickson2001; Johnson and Tversky Reference Johnson and Tversky1983; Lerner and Keltner Reference Lerner and Keltner2000).
By contrast, negative affect is associated with avoidance- rather than action-seeking effects (Carver and Scheier Reference Carver and Scheier1990; Lerner and Keltner Reference Lerner and Keltner2001). The importance of positive affect is recognised by the SMT literature on ‘affective focusing’ (Della Porta Reference Della Porta2009), which describes a process grounded in individuals’ positive sentiment towards a group, through which they may follow friends into radical groups while distancing themselves from others. This suggests that positivity may help resistance actors garner support for their activities among their audience.
The following analysis finds that Hezbollah’s speeches most frequently express positivity, followed by trust, negativity and anger, respectively. In addition to the role of trust, this shows the importance of positive affect (Folkman and Moskowitz Reference Folkman and Moskowitz2000; Lyubomirsky et al. Reference Lyubomirsky, King and Diener2005; Pressman and Cohen Reference Pressman and Cohen2005). Neither positivity nor trust has been properly integrated into resistance studies. Our findings show their fruitfulness in understanding resistance rhetorics – even in the case of resistance actors that are widely considered to display hatred and negativity like Hezbollah (Armstrong and Matusitz Reference Armstrong and Matusitz2013; Hage Reference Hage2009; Jaber Reference Jaber1997).
A well-known hybrid organisation, Hezbollah combines a political wing participating in Lebanese politics with an armed wing fighting Israel. This hybrid structure is known to require organisational flexibility and understanding of the interactions between the different organisational parts to maintain legitimacy (Smith and Besharov Reference Smith and Besharov2019). Emotions are known to play a key role in managing these demands (Sadeh and Zilber Reference Sadeh and Zilber2019). Research on legitimacy processes shows that respected authority figures may promote organisational strength by attenuating negative emotions (Johnson et al. Reference Johnson, Hegtvedt, Khanna and Scheuerman2016). Adding to this research, we find that Hezbollah’s leaders express positivity and emotions of trust to promote internal cohesion, while they express anger and negativity specifically to rally support for violent resistance.
The analysis applies methods from quantitative text analysis, namely sentiment analysis (Liu Reference Liu2012) and topic modelling (Blei and Lafferty Reference Blei and Lafferty2006). Sentiment analysis is designed to detect emotions and affect from natural language (Medhat et al. Reference Medhat, Hassan and Korashy2014). This is valuable because emotions and affect operate in unconscious ways (Kahneman Reference Kahneman2011; Lodge and Taber Reference Lodge and Taber2005) that may be missed by studies that prime the respondents in surveys (Bettarelli et al. Reference Bettarelli, Close and Van Haute2022; Groenendyk Reference Groenendyk2011; Huddy et al. Reference Huddy, Mason and Aarøe2015; Miller Reference Miller2011; Sabucedo and Vilas Reference Sabucedo and Vilas2014; Searles and Mattes Reference Searles and Mattes2015) or interviews (Dornschneider Reference Dornschneider2021; Pearlman Reference Pearlman2016). While sentiment analysis is relatively rare in research on emotions (exceptions are Hoffmann Reference Hoffmann2018; Hu and Flaxman Reference Hu and Flaxman2018; Zucco et al. Reference Zucco, Calabrese and Cannataro2017), it is popular in studies of political communication (Watanabe Reference Watanabe2021; Young and Soroka Reference Young and Soroka2012). The following application shows how it may add new understanding of resistance actors.
Topic modelling facilitates the interpretation of the sentiment analysis by identifying the topics contained in natural language. Combining topic modelling with sentiment analysis, our study shows that trust and positivity are most frequently expressed in relation to Hezbollah and Islam, whereas anger and negativity are most frequently expressed in discussions of violent resistance and regional security. Anger is additionally expressed frequently in discussions of the Palestinian struggle, whereas negativity has high frequency in political discourse. In related studies of emotions and affect, applications of topic models have been comparatively rare, with exceptions investigating the motivations of speakers (Klein et al. Reference Klein, Clutton and Polito2018) or hot research topics (Bittermann and Fischer Reference Bittermann and Fischer2018). To the best of our knowledge, existing studies in political psychology have not yet combined sentiment analysis with topic modelling to investigate emotions or affective valence.Footnote 3
Consistent with the broader field of quantitative text analysis (Watanabe Reference Watanabe2021), resistance research applying sentiment analysis typically focuses on English data (Calderon et al. Reference Calderon, Fisher, Hemsley, Ceskavich, Jansen, Marciano and Lemieux2015; Dornschneider and Todd Reference Dornschneider and Todd2021; Won et al. Reference Won, Steinert-threlkeld and Joo2017). However, resistance behaviour occurs frequently in non-English-speaking environments, and the following analysis offers an important addition from an Arabic-speaking context, building on growing quantitative research with textual data in Arabic (Al-Ani et al. Reference Al-Ani, Mark, Chung and Jones2012; Bang et al. Reference Bang, Lee and Rao2021; Lynch et al. Reference Lynch, Freelon and Aday2017). Conveniently, knowledge of Arabic is not a necessary requirement for the application of sentiment analysis, which relies on automated translations, or topic modelling, which deals with the original text.Footnote 4 As such, these methods offer a valuable tool to better understand the communication of resistance actors in the Arab world. This is particularly helpful given the great differences in perceptions between local Arab populations and foreign Western audiences (see the aforementioned classifications of Hezbollah, for example).
This study has an inductive research design, which is increasingly applied to maximise the value of the data (Jebb et al. Reference Jebb, Parrigon and Woo2017) and to challenge existing research paradigms (Corley et al. Reference Corley, Bansal and Yu2021). Research on resistance includes many examples of inductive and deductive designs, and their balanced application has been related to scientific progress (Jebb et al. Reference Jebb, Parrigon and Woo2017; Kell and Oliver Reference Kell and Oliver2004). The following study contributes an inductive quantitative text analysis, which is less common and adds to the large literature on inductive analyses with qualitative methods (Elo and Kyngäs Reference Elo and Kyngäs2008; Kuckartz Reference Kuckartz2014). Although often applied in hypothesis-based research, topic modelling and sentiment analysis generally lend themselves to exploratory research and are known to enhance our understanding of large corpora (Blei et al. Reference Blei, Ng and Jordan2003; Tunca et al. Reference Tunca, Sezen and Wilk2023). The promising findings of the following study underline this feature and its potential benefits in research on political psychology.
The following sections discuss theories on emotions and affect in the resistance literature. The article then presents the methods and data, followed by the analysis, and then discusses the findings. The Supplementary Material provides the code to replicate the analysis and the website through which the data can be accessed.
Emotions and affect in resistance
An important motivator of resistance, emotions are known to play a potentially powerful role in political communication (Richards Reference Richards2004; Widmann Reference Widmann2021). Following grievance theories (Davies Reference Davies1962; Gurr Reference Gurr1970), people engage in resistance in response to perceptions of relative deprivation in the distribution of power and wealth (Correa et al. Reference Correa, Nandong and Shadmehr2025; Davies Reference Davies1962; Gurr Reference Gurr1970; Hillesund and Østby Reference Hillesund and Østby2023; Lüders et al. Reference Lüders, Urbanska, Wollast, Nugier and Guimond2021; Power et al. Reference Power, Madsen and Morton2020). These perceptions are believed to evoke negative emotions, which subsequently motivate people to revolt. While the early literature focused mostly on negative emotions of frustration (Berkowitz Reference Berkowitz1989; Davies Reference Davies1962; Gurr Reference Gurr1970; Khashan Reference Khashan2003), later studies have expanded the field to other types of negative emotions, notably anger (Grant et al. Reference Grant, Abrams, Robertson and Garay2015; Jasper Reference Jasper2014; Smith et al. Reference Smith, Pettigrew, Pippin and Bialosiewicz2012; Van Zomeren et al. Reference Van Zomeren, Postmes and Spears2008).
New findings show that positive emotions play an important additional role (e.g. Bettarelli et al. Reference Bettarelli, Close and Van Haute2022; Cohen-Chen and Van Zomeren Reference Cohen-Chen and Van Zomeren2018; Dornschneider Reference Dornschneider2021; Jasper Reference Jasper2018; Pearlman Reference Pearlman2013; Sabucedo and Vilas Reference Sabucedo and Vilas2014; Van Troost et al. Reference Van Troost, Van Stekelenburg, Klandermans and Demertzis2013). The psychology literature shows that positive emotions broaden people’s thought-action repertoires (Fredrickson Reference Fredrickson2001), thus contributing to various types of behaviour. Recent studies on resistance show the importance of various discrete positive emotions, such as interpersonal trust (Benson and Rochon Reference Benson and Rochon2004; Kaase Reference Kaase1999; Weipert-Fenner et al. Reference Weipert-Fenner, Rossi, Sika and Wolff2024; Winters Reference Winters2008), solidarity (Dornschneider Reference Dornschneider2021; Eyerman Reference Eyerman2007; Jasper Reference Jasper1998; Juris Reference Juris2008; Van Troost et al. Reference Van Troost, Van Stekelenburg, Klandermans and Demertzis2013) or hope (Bettarelli et al. Reference Bettarelli, Close and Van Haute2022; Cohen-Chen and Van Zomeren Reference Cohen-Chen and Van Zomeren2018; Dornschneider Reference Dornschneider2021; Sabucedo and Vilas Reference Sabucedo and Vilas2014; Van Troost et al. Reference Van Troost, Van Stekelenburg, Klandermans and Demertzis2013), to name a few. Regardless of the particular mechanisms by which they contribute to resistance, each of these emotions is found to motivate rather than deter participation. This suggests that they may be employed in resistance communication seeking to garner the support of the audience.
Besides emotions, affect constitutes another important element of political communication (Young and Soroka Reference Young and Soroka2012). Affective states are known for their impact on political judgements (Lodge and Taber Reference Lodge and Taber2005) and are therefore particularly helpful to resistance actors seeking to increase their appeal. Like positive emotions, positive affect is assumed to motivate behaviour by broadening cognition and attention (Fredrickson Reference Fredrickson2001; Rowe et al. Reference Rowe, Hirsh and Anderson2007), leading to flexible and thorough cognitive processes (Isen Reference Isen2001) and increased overall cognitive ability (Barber and Kim Reference Barber, Kim, Sedek, Hess and Touron2021). In social settings, this has been linked to helping, generosity and interpersonal understanding (Isen Reference Isen2001). In SMT, positive affect is linked to processes in which individuals turn to radical groups while distancing themselves from others (Della Porta Reference Della Porta2009). The following study extends this work to communication by a hybrid resistance actor, showing expressions of positivity related to the in-group and group ideology, thus seeking to strengthen group cohesion.
By contrast, negative affect is associated with more pessimistic judgements (Johnson and Tversky Reference Johnson and Tversky1983), a focus on risks rather than benefits (Alhakami and Slovic Reference Alhakami and Slovic1994) and opposite effects on behaviour (Cahir and Thomas Reference Cahir and Thomas2010). Related to resistance, this suggests that negativity may promote risk-averse decision-making, either deterring potential participants or resulting in behaviour limiting the perceived risk. The following analysis shows that negativity is indeed linked to security issues, which address perceived risks and violent resistance as a means to deal with these risks.
Only a few studies of resistance investigate the role of affect empirically, and this study provides a valuable addition. Our findings on the importance of positive affect are consistent with research on the Black Lives Matter movement that shows the prominence of positivity in Tweets (Field et al. Reference Field, Park, Theophilo, Watson-daniels and Tsvetkov2022) as well as research on group dissent noting the preservation of group positivity as a major goal when dealing with dissent (Jetten and Hornsey Reference Jetten and Hornsey2014). Moreover, our findings linking negativity to violent resistance are in line with evidence (Bettarelli et al. Reference Bettarelli, Close and Van Haute2022) that negative affect is associated with dislike of the opponent, which is an important protest motivator.
Anger and trust
The analysis examines communication of two discrete emotions with different valence – anger (negative valence) and trust (positive valence). Both anger and trust (Benson and Rochon Reference Benson and Rochon2004) are known as motivators of resistance behaviour. As such, the role of anger deviates from the described avoidance effects of negativity (Berkowitz and Harmon-Jones Reference Berkowitz and Harmon-Jones2004), whereas trust shows effects aligned with those of positivity. This special role of anger shows that this emotion deserves special attention beyond general studies of affective states (Lerner and Keltner Reference Lerner and Keltner2001).
Anger is a combination of feelings, cognitions and physiological reactions directed at injuring someone or something (Berkowitz and Harmon-Jones Reference Berkowitz and Harmon-Jones2004). As such, anger is of particular interest to violent resistance movements looking for a supportive audience. In research on politics, anger often assumes meaning related to the social contract of the people involved (Tavris Reference Tavris1989: 19), including the ruling elite and the people (Feierabend et al. Reference Feierabend, Feierabend and Gurr1972). The literature on resistance considers anger a result of perceptions of relative deprivation in wealth and political power, identifying it as a mediator for willingness to engage in non-violent (Leach et al. Reference Leach, Iyer and Pedersen2007) and violent (Brockhoff et al. Reference Brockhoff, Krieger and Meierrieks2012; Fink et al. Reference Fink, Leshem and Halperin2022; Wright-Neville and Smith Reference Wright-Neville and Smith2009) resistance. Psychology research shows that this happens by anger positively influencing risk judgements and providing a sense of individual control and certainty (Lerner and Keltner Reference Lerner and Keltner2001). The following findings are consistent with this understanding by linking anger to violent resistance and security issues. However, they also show an important additional role of positive emotions of trust, which are overall expressed much more frequently than anger and therefore deserve separate attention.
Contrasting with anger, trust is possibly the main requirement for happy and healthy relationships (Simpson Reference Simpson2007). As such, trust is extremely important for the maintenance of relationships between resistance movements and their followers. Interpersonal trust is a psychological state of a person (the truster) that another person (the trustee) will behave in a certain way (Simpson Reference Simpson2007). This involves confidence in one’s expectations (Deutsch Reference Deutsch1958; Luhmann Reference Luhmann2018),Footnote 5 where the truster makes themselves vulnerable to the trustee, who could betray the truster (Levi and Stoker Reference Levi and Stoker2000). This may happen consciously or unconsciously (blind trust).
Trust is well-known in studies of international relations (Hoffman Reference Hoffman2002; Ku and Mitzen Reference Ku and Mitzen2022; Larson Reference Larson1997; Michel Reference Michel2013), but comparatively less explored by the resistance literature. The literature on resistance movements suggests that interpersonal trust may be an important motivator for social mobilisation (Weipert-Fenner et al. Reference Weipert-Fenner, Rossi, Sika and Wolff2024). The underlying logic is that interpersonal trust decreases the perceived costs of participation, while supporting more optimistic evaluations of the benefits of resistance (Benson and Rochon Reference Benson and Rochon2004; Weipert-Fenner et al. Reference Weipert-Fenner, Rossi, Sika and Wolff2024). While interpersonal trust in this context refers to fellow citizens, interpersonal trust has also been related to scientists, environmental groups or institutions (Cologna and Siegrist Reference Cologna and Siegrist2020), and social media (Gainous et al. Reference Gainous, Abbott and Wagner2019).
By contrast, violent resistance has been related to a lack of interpersonal trust (Worchel et al. Reference Worchel, Hester and Kopala1974), as shown by research on violent flanks within non-violent movements (Chenoweth Reference Chenoweth2023). The underlying argument is that fragmentation within non-violent movements creates competition, in which the group members take sides. In this condition, violence may be embraced by some even when the movement leaders reject it (Pearlman Reference Pearlman2016), thus indicating an overall decrease of interpersonal trust within the movement. The following analysis investigates a movement with both a violent and non-violent wing. It shows that official communication on the in-group and its ideology is marked by trust, thus displaying an effort to maintain cohesion and unity.
In addition, the broader literature has often examined trust in the government, which has negative effects on resistance (Batrancea et al. Reference Batrancea, Nichita, Olsen, Kogler, Kirchler, Hoelzl, Weiss, Torgler, Fooken and Fuller2019; Frye and Borisova Reference Frye and Borisova2019) and positive effects on government support (Levi Reference Levi1997; Scholz and Lubell Reference Scholz and Lubell1998). As such, it has been related to system stability (Frye and Borisova Reference Frye and Borisova2019), especially when comparing trust in the government with trust in the leaders of the opposition (Nilson and Nilson Reference Nilson and Nilson1980). Focusing on communication by group leaders, the following analysis shows that trust is expressed most frequently in relation to the in-group and its ideology, thus supporting a narrative on group stability.
HezbollahFootnote 6
Hezbollah is a highly influential political player belonging to the Shia community in Lebanon. Following the discussed literature, the Shia community represents a typical resistance actor because of its relative deprivation in the distribution of power and wealth under centuries of Sunni-dominated rule (Khashan Reference Khashan2019). Hezbollah possesses a sophisticated media network with extensive reach to Arab audiences, including local residents, Lebanese voters, regional allies and observers (Daher Reference Daher2019). Consequently, Hezbollah’s communication is frequent and prolific, covering a wide range of issues with differing emotional appeals – making it a particularly valuable case for examining communication by resistance actors.
Hezbollah’s communication is embedded in its various functions at political, social and military levels. Known as a ‘proletarian party’ (Salamey and Pearson Reference Salamey and Pearson2007) representing the disempowered (Shaery-Eisenlohr Reference Shaery-Eisenlohr2008), Hezbollah provides a wide range of social services including health, education, infrastructure, micro-credit, sport, etc. (Cammet Reference Cammet2014), operated mostly outside of the institutional state structures (Harb Reference Harb2008).
At the same time, Hezbollah actively participates in Lebanon’s confessional power-sharing political system (Norton Reference Norton2007). Within this system, it has both a history of cooperation with rival non-Shia actors (ibid.) and a reputation of abusing its power, for example, by paralysing the government (Khashan Reference Khashan2019; Lob Reference Lob2014).
Beyond Lebanese borders, Hezbollah acts as a protector of national interests in a decades-long confrontation with Israel (Hezbollah 2023). Backed by Iran, it operates separately from the Lebanese army and is widely considered as the sole guarantor of Lebanese security, with much stronger capabilities than the state military (author interviews in Lebanon, 2022–23; Barnard, 20 June 2023).
Due to these diverse functions, Hezbollah’s communication serves goals at social, political and military levels. The literature on hybrid organisation shows that emotions play an important role in successful management of such differing goals (e.g. Sadeh and Zilber Reference Sadeh and Zilber2019). Consistent with research on violence, analyses of Hezbollah have often emphasised negativity or negative emotions like anger or hatred (Armstrong and Matusitz Reference Armstrong and Matusitz2013; Hage Reference Hage2009; Jaber Reference Jaber1997).Footnote 7 The following analysis suggests that although this view is partially justified, it overlooks the more nuanced role of emotions in communication by hybrid actors: Consistent with their known mobilising role, anger and negativity are expressed when discussing violent resistance and security. However, when discussing the in-group, Hezbollah leaders employ positivity and trust to strengthen group cohesion.
Analysis
We obtained data by scraping Hezbollah’s Arabic website, including every available speech.Footnote 8 The corpus consists of 306 speeches containing 1.1 million words. It has an annual maximum average of 7,326.00 words in 2016 and a minimum average of 1,020.89 in 2001. Its average number of speeches is 13.3 per year, with a maximum of 34 in 2003 and a minimum of 3 in 2007 and 2009. The speeches express Hezbollah’s official reasoning over the last two decades (September 2001–July 2023). Given Hezbollah’s wide reach, the audience of these speeches ranges from Arabic speakers beyond local supporters to general Lebanese voters, Arab allies and citizens in neighbouring Arab countries.
Our study applies quantitative text analysis, combining methods from natural language processing, sentiment analysis and topic modelling. We discuss both in Appendices 1 and 2 in the Supplementary Material. Sentiment analysis captures emotions and affect contained in natural language through a lexicon that contains words associated with positivity (e.g. ‘truth’, ‘mother’ or ‘hope’), negativity (e.g. ‘battle’, ‘war’ or ‘mistake’), anger (e.g. ‘rage’ or ‘revulsion’) and trust (e.g. ‘safe’ or ‘safeguard’). We apply the popular National Research Council of Canada (NRC) lexicon, which we discuss and analyse in Appendices 1 and 2. The lexicon’s association scores for affect and emotion are binary, indicating whether the word is associated or not. The sentiment score shows the proportion of words contained in the speeches that are related to positivity, negativity, anger and trust, presented in Figure 1.Footnote 9
Positivity, Negativity, Anger and Trust in Hezbollah Speeches
The analysis identifies clear differences between emotions and affect for each year in the examined time frame.Footnote 10 The most frequent sentiment is positivity. On average, words associated with positivity occur 2.03 times as often as words associated with negativity (mean positivity: 0.057; mean negativity: 0.028). This finding is especially pronounced, as the NRC emotion lexicon contains 1.44 times more words associated with negativity (0.234) than with positivity (0.163).
Trust and anger show similar differences. On average, words associated with trust occur 2.58 times more often than words related to anger (mean trust: 0.045; mean anger: 0.017), whereas the NRC lexicon contains almost equal shares of words associated with anger (0.088) and trust (0.087) (see Figure 1 and Table 1).
Sentiment Scores for Hezbollah Speeches
To understand when Hezbollah expresses emotions and affect, we first look at the speeches containing the most words associated with positivity, negativity, anger and trust. The titles of these speeches suggest that emotions and sentiment are related to specific types of events that occurred throughout the examined time frame, rather than landmark events reported in the Western media (e.g. the 2020 port explosion, wars with Israel, the recent economic crash, etc.). Table 2 gives an overview of the top five titles regarding each emotion and affective state. Appendix 3 in the Supplementary Material shows the top 10 titles, including overlaps between positivity and trust versus negativity and anger.
Top-Scoring Speech TitlesFootnote 11
Identified TopicsFootnote 15
The top titles suggest that positivity and trust are expressed most frequently by speeches delivered for celebrations, honorific events and Islamic holidays. Two of the speeches with the highest positivity scores address the birthday of the Prophet. The remaining three speeches address ceremonies including a memorial, a graduation and an honorary event. Regarding trust, three of the top-scoring speeches deal with the Prophet’s birthday, one with a graduation and one with the presidential election.
By contrast, negativity is mostly expressed in speeches on conflict, including war, violence, assassinations, occupation and threats. Three of the speeches with the highest negativity scores address threats and violence in Lebanon, while the remaining two additionally deal with crises and confrontations abroad. Of the top five speeches on anger, three deal with confrontation and violence abroad, whereas two deal with dangers within Lebanon as well as the reconstruction of the southern suburbs of Beirut after their destruction through war.
While anger is mostly expressed in speeches on conflict, the top 10 speeches (Supplementary Material, Appendix 3) also include honorific or celebratory events of the resistance. Nevertheless, there appears to be almost no overlap between positive and negative emotions and affective states, suggesting that they are expressed in relation to different subjects.
Hezbollah’s speeches are often long and cover much more than the event described in the title. To relate positivity, negativity, anger and trust to the contents of the speeches, the analysis applies topic modelling.Footnote 12 Topic modelling is designed to detect hidden themes in textual data by quantitatively analysing the words in documents (Blei et al. Reference Blei, Ng and Jordan2003; Deerwester et al. Reference Deerwester, Dumais, Furnas, Landauer and Harshman1990). This allows us to see what parts of the speeches contain positivity, negativity, anger and trust. Table 3 shows the identified topics and their associated high-frequency words.Footnote 13
The analysis identifies 10 topics that show what types of discussions are included in the speeches.Footnote 14 Two topics refer to the in-group and its ideology, namely: Hezbollah and Islam. Three topics refer to violent resistance, namely: violent resistance, the Palestinian struggle and holy battle. Additional topics refer to regional security, Lebanese military security and state governance. Two topics refer to non-violent dialogue, namely: political discourse and discussions with Israel. Table 3 shows the high-frequency words related to each topic.
The sentiment analysis links each topic to positivity, negativity, anger and trust. The findings (Table 4) identify clear differences between positive versus negative emotions and affect. The strongest identified correlations show that trust and positivity are associated with discussions of Hezbollah and Islam, but not with discussions of any other topic. This suggests that trust and positivity play an important role related to the in-group and Islamic ideology, rather than violent resistance, security or non-violent dialogue. This finding complements the emerging literature on positivity in resistance and the existing resistance literature, implying the importance of trust.
Correlations Between Topics and Emotions/Affect
Note: Asterisks (*) indicate statistical significance (p < 0.01), while shaded cells show significant positive correlations.
Table 5 illustrates this finding using excerpts from speeches with high sentiment scores. In discussions of Hezbollah, the quote scoring highest for trust expresses the belief that God is the highest authority. The highest-scoring quote for positivity announces the beginning of the promised project, which was dedicated to rebuilding the southern suburbs of Beirut after its destruction in the 2006 war with Israel. In discussions of Islam, the highest-scoring quotes for both trust and positivity contain one single word ‘good’ – which indicates agreement or positive judgement.
Quotes Assigned the Highest Scores for Topics, Emotions and Affective StatesFootnote 16
Anger is associated with discussions about violent resistance, the Palestinian struggle and regional security, aligning with the broader literature on anger in resistance. However, the findings diverge from expectations in the literature linking anger to non-violent resistance, as no similar relation is observed in connection with political discourse, discussions with Israel or state governance. Additionally, the analysis suggests that negativity complements anger, as it is associated with discussions of violent resistance, regional security and political discourse. Notably, neither anger nor negativity is related to contexts associated with trust or positivity.
Table 5 illustrates how anger and negativity are communicated. In discussions of violent resistance, the quote scoring highest for anger says that the resistance is not a militia. In discussions of the Palestinian struggle and regional security, the highest-scoring quote for anger consists of a single word, namely ‘lying’. This word is also the highest-scoring quote for negativity in discussions of regional security and political discourse. Related to violent resistance, the quote scoring highest for negativity concerns stopping the resistance to normalisation with Israel. This quote is placed in the context of a broader discussion of the resistance to normalisation, which ‘must continue, against normalisation projects in the Arab and Islamic worlds, against the Israeli enemy. This includes political normalisation, cultural normalisation, economic normalisation, commercial normalisation, and participation in conferences – it [normalisation] must stop’.
Conclusions
This article advances our understanding of resistance through a study of emotions and affect contained in Hezbollah speeches. Contrasting with common understandings that associate resistance with negative emotions, the analysis finds that Hezbollah’s public communications are characterised by positive affect and positive emotions of trust, with positivity expressed twice as often as negativity, and trust two and a half times as often as anger. Building on research into positive emotions in resistance and emerging research on positive affect, these findings constitute an important addition to the large body of literature associating Hezbollah and resistance behaviour more broadly with negative emotions or negativity.
This article highlights the so far overlooked role of positivity and trust in resistance communication, which are expressed when discussing the in-group. The expression of trust in communications about the in-group is also consistent with broader research on political participation, which shows the importance of trust in certain groups (Crepaz et al. Reference Crepaz, Jazayeri and Polk2017) and of generalised interpersonal trust (Sullivan and Transue Reference Sullivan and Transue1999). While studies in this field usually focus on democracies, our study suggests a role for trust in a non-democratic context marked by violent resistance.
Connecting anger with violent resistance and security issues, our findings are in line with prevailing perspectives that identify anger as a motivator of resistance and as a potentially powerful communication tool. Together, the results on anger and trust suggest that Hezbollah’s communication strategy relies on positive emotions to support group cohesion, while negative emotions are communicated to garner support for resistance activities. This adds to research on hybrid organisations by showing how leaders can use emotions with different valence to do both – foster group stability and support particular group activities. This nuanced use of emotions and their associated affective states has not been investigated as a feature of communication by resistance actors.
Our findings contrast with the portrayal of Hezbollah in Western media associating the group solely with ‘incitement of hatred’ (City News 2024) or ‘hate crime’ (National Post 2024). Instead, this article highlights how Hezbollah’s Arabic-language communication emphasises positivity and trust – an approach that may help the group expand its appeal while preserving internal cohesion. While consistent with the psychology literature on resistance movements, further research is required to discover whether these findings are generalisable to other actors, including hybrid organisations, different types of resistance movements or militant groups.
Nasrallah’s final speech on Hezbollah’s website reflects this communication style.Footnote 17 Rather than using hostile language against his adversaries, Nasrallah opened by extending congratulations to his fellow Muslims: ‘Before I begin … I must congratulate all Muslims on the birthday of the Great Messenger of God.’ He continued with additional well-wishes ‘to all Muslims and everyone’, followed by a moment of mourning for a prominent Ayatollah. Hence, rather than using a divisive framing in his speech, he set a tone of shared identity and unity.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gov.2026.10057.
Data availability
The data analysed in this article were downloaded from https://mediarelations-lb.org/section.php?id=94&page=1 (accessed between January and July 2023). The replication code is provided in the online appendix. To protect the participants in this study, interview transcripts cannot be shared.
Acknowledgements
This article was written based on a project (designed and led by the main author) in the Connected Politics Seminar at University College Dublin in 2022–23. The authors gratefully acknowledge feedback from Stefan Mueller, who led the Connected Politics Seminar, and Gijs Schumacher, as well as the participants of the panel ‘Responding to and Coping with Violence’ at the 2023 Annual Meeting of the American Political Science Association, particularly the discussant, Leonid Peisakhin, and chair, Philip Roessler.
Financial support
Stephanie Dornschneider-Elkink gratefully acknowledges support in the form of a Laureate Consolidator Award, IRCLA/2022/3089.
Ethics statement
The research using human subjects has been approved by University College Dublin, approval number HS-22-51-Elkink.
Disclosure statement
The authors report that there are no competing interests to declare.
Open access
