Frozen discourse: how screenshots hinder depolarization on social media

ABSTRACT

Social media is thought to play a pivotal role in increasing animosity between opposing partisans, i.e., affective polarization. This study focuses on one context that is assumed to reduce polarization – an acute external threat – and demonstrates that even under such conditions, social media platforms perpetuate affective polarization. Furthermore, the research findings suggest a digital feature that might contribute to this process: screenshots. By conducting a case study and thematically analyzing 1,670 tweets discussing ‘inner unity’ posted in Israel during an escalation of the Israeli-Palestinian conflict, the study shows that screen-captured images can be a barrier to the dynamic process required for depolarization, as they ‘freeze’ online discussions. Additionally, it reveals two content-related discoursal mechanisms involved in preserving polarization between rightist and leftist Jewish Israelis: recalling the ‘unified’ past and distorting the meaning of unity. I discuss the theoretical implications of the findings for the study of the relationship between online political discourse and (de)polarization.

KEYWORDS:

• Affective polarization
• depolarization
• screenshots
• conflict
• Twitter

Social media is considered by many scholars to play a key role in the rise of animosity between opposing partisans, a process known as affective polarization (AP) (Arora et al., 2022; Kubin & von Sikorski, 2021). While the connective nature of social media enables individuals to be exposed to a wide range of information and diverse opinions (Bakshy et al., 2015), much scholarly work demonstrates that social media also does the opposite; it increases information selectivity and encourages the creation of echo chambers (Cinelli et al., 2021), in which AP flourishes (Harel et al., 2020).

Interestingly, although scholars have extensively examined the relationship between social media and AP (e.g., Cinelli et al., 2021; Zhu et al., 2021), the study of social media’s role in depolarization has remained limited (Kubin & von Sikorski, 2021). Recently, pioneer scholars have begun examining how social media can be harnessed to implement planned interventions to reduce AP (e.g., Currin et al., 2022; Saveski et al., 2021). Nevertheless, the study of social media’s role in ‘natural’ contexts that might decrease inter-party hostility is still largely overlooked (as an exception, see John & Dvir-Gvirsman, 2015).

One such context that is considered to have the potential for reducing AP is an external threat (Van Bavel et al., 2020), as it often urges group members to unite (Giles & Evans, 1985). Empirical studies that tested the impact of external threats (security, health, etc.) on AP supported external threats’ ability to mitigate AP (e.g., Boxell et al., 2020). Notably, conflicting findings exist, with some studies showing no impact on AP levels (Myrick, 2021) or even an increase (Jahani et al., 2022). However, almost none of these studies examined what might be the role of social media in (de)polarization under such threatening contexts (again, as an exception, see John & Dvir-Gvirsman, 2015).

In the current study, I attempt to fill this gap by employing a qualitative thematic analysis of tweets posted in Israel during ‘the 2022 wave of terror’. Relying on this case study, I intend to examine the dynamics of communication surrounding the notion of depolarization and the discoursal mechanisms involved. Next, I will briefly review previous work on affective (de)polarization and social media, and on the potential depolarizing effect of external threats. Then, I will introduce the current case study, the research methodology, and the study findings. As I elaborate below, a surprising research finding goes beyond the content analysis of tweets to a structural aspect of the discourse: the use of screenshots as a barrier to depolarization. I will conclude by discussing the implications of the research findings for studying the relationship between social media and affective (de)polarization.

Theoretical background

Affective (de)polarization and social media

AP is considered one of this era’s most prominent and troubling socio-political phenomena (Finkel et al., 2020). Scholars who study AP suggest that inter-party hostility is rooted in individuals’ social-partisan identity, which dictates the social interactions with the political ingroup and outgroup (Iyengar et al., 2012). Specifically, strong partisan identification is thought to lead to more positive affect and attitudes toward members of one’s own political group while simultaneously increasing the negative affect and attitudes toward out-party members. These destructive inter-party dynamics have been found to have severe implications for the political sphere (e.g., Somer & McCoy, 2018) and social life (e.g., Chen & Rohla, 2018).

Scholars differ on whether social media contributes to AP (Nordbrandt, 2021). Although some researchers have demonstrated that online platforms enable inter-party interaction and exchange of ideas (e.g., Barberá et al., 2015; Jones-Jang & Chung, 2022), significant research provides evidence that social media use is one of the leading reasons for the increase in AP in the last few decades (Arora et al., 2022; Kubin & von Sikorski, 2021). Accordingly, users of social media platforms tend to interact within homogenous social enclaves, where they are prevented (by algorithmic filter bubbles; Pariser, 2011, or self-selectively) from being exposed to a variety of opinions and where their worldview is rarely challenged (e.g., Zhu et al., 2021); thus, they are prone to adopt more extreme and polarized ideologies and agendas (Sunstein, 1999). Notably, these enclaves also strengthen bonds among like-minded users (Mosleh et al., 2021) while further alienating them from ideological opponents (Sumaktoyo, 2021).

While scholars have thoroughly examined social media’s connection to AP (Arora et al., 2022), its role in depolarization received relatively little attention (Kubin & von Sikorski, 2021). Despite everyday instances of depolarization through social media (e.g., the story of Westboro Baptist church member Megan Phelps-Roper; Chen, 2015), scholarly interest in this area has only recently risen. For example, Saveski et al. (2021) conducted a field experiment that enabled Twitter users to switch their feeds with others from the opposite political group, aiming to prompt perspective-taking and increase intergroup empathy. They found that only when participants were reminded of an experience of disagreement with a friend and asked to frame the current encounter similarly did the intervention decrease AP. Also aiming to broaden online inter-party exposure, Currin et al. (2022) suggested a random dynamical nudge designed to expose social media users to a random sample of political opinions, thus fostering dialogue and consensus-building. Approaching the challenge of depolarization differently, Borrelli et al. (2022) examined influential figures’ tweets responding to offline controversial events, finding that providing a counter-narrative within 12 h was effective for online depolarization. These examples shed light on the promising potential of harnessing social media for depolarization. This line of research, which has only recently emerged, will likely gain momentum in the coming years.

However, while these examples demonstrate how social media can be intentionally used to promote depolarization, there is almost no research examining the role social media plays in ‘natural’ contexts that are thought to reduce AP. For instance, previous work suggests that contextual settings, such as health and economic crises (Jungkunz, 2021) or external threats (Boxell et al., 2020), can stimulate depolarization. Nevertheless, scholars have generally overlooked the significance of studying online inter-party dynamics and interactions in these contextual frameworks and whether they contribute to depolarization. This lack of scholarly attention is, to some extent, surprising since it is well-established that social media play a crucial role in eliciting, promoting, and propagating social change (e.g., Kumar & Thapa, 2015).

An exception is the study of John and Dvir-Gvirsman (2015), who surveyed Israelis about their Facebook use habits during the Israel – Gaza Conflict of 2014 and identified what they term: ‘politically motivated unfriending’ – a digital practice and political gesture that increases AP. Their findings provide valuable insight into the real-world effect of threats on AP and initial evidence for social media’s polarizing (and not depolarizing) role under these circumstances. However, as this line of research is still in its infancy, there is a need for further exploration of the relationship between social media and (de)polarization in the context of external threats.

(De)polarization in the context of external threat

As mentioned above, external threats are presumed to foster societal depolarization (Van Bavel et al., 2020). Several theories from the fields of sociology and political psychology (e.g., the conflict cohesion hypothesis; Coser, 1956, the societal belief of unity; Bar-Tal, 2007) explain this assumption as they generally suggest that when facing external threats, groups tend to unite to cope with the threat more effectively (Giles & Evans, 1985).

Recent studies show empirical support for the depolarizing impact of external threats. Research on COVID-19 revealed a reduction in negative sentiment between opposing partisans during pandemic-related distress (Boxell et al., 2020; Jungkunz, 2021). Kaiser and Seier (2023) observed increased willingness among Americans to cooperate with political opponents after exposure to information about the Russian invasion of Ukraine. However, conflicting studies, such as those by Jahani et al. (2022) and Myrick (2021), found that exposure to a common threat (posed by Iran or China) did not reduce AP between Republicans and Democrats in the US, and sometimes even increased it. Similarly, John and Dvir-Gvirsman’s (2015) work, described above, demonstrates that AP might be bolstered in times of a genuine security threat.

So, investigating whether external threats contribute to depolarization and exploring the role of social media in (de)polarization under such conditions are two crucial empirical questions. To address these, I qualitatively analyze Twitter discourse in times of acute security threat. By focusing on tweets revolving around the notion of unity, posted in Israel during an escalation of the Israeli-Palestinian conflict, I explore how Twitter users discuss the concept of depolarization when facing terror, whether the discourse reflects trends of AP or depolarization, and what discoursal mechanisms are employed in the online discourse to promote (de)polarization.

Case study

This study was conducted in Israel during ‘the 2022 wave of terror’ – a three-week period in which four violent attacks occurred within Israel’s borders, and 14 Israelis were killed (Ministry of Foreign Affairs, 2022) – resulting in high threat levels among the Jewish Israeli society. During these weeks, Israeli public figures emphasized the importance of unity within Israeli society, releasing statements published in various newspapers (e.g., Rosenberg, 2022) and appearing on social media networks, where they sparked public reactions. As Israelis frequently use social media platforms to discuss political issues (Mor et al., 2015), ordinary citizens also tweeted their thoughts regarding the notion of unity during this period.

Notably, Israel is considered a deeply divided society, where in recent years, the political divide has been thought by citizens to be the second most significant societal divide (after the ethnic divide between Jews and Arabs; Blander, 2018). Concretely, the political divide is between the right and left ideological blocs, shaped mainly by their stances on the Israeli-Palestinian conflict since the 1967 war. Recent scholarly work demonstrates that the political divide is not just ideological but also characterized by AP (e.g., Harel et al., 2020) reflected in dislike and antagonism between supporters of the opposite ideological groups. Therefore, calls for unity during the escalation of the Israeli-Palestinian conflict often reflect the wish to set aside ideological disagreements and increase the cohesion between ideological groups in the face of the external threat. It should be noted that the study was conducted before the Israel-Hamas war in October 2023, potentially reflecting a different public sentiment compared to more recent Israeli discourse.

Method

Data collection

Using the R package: academictwitteR (Barrie & Ho, 2021), I extracted all Hebrew tweets posted between 22 March and 10 April that contained the following keywords: unity, unified, and to unite (represented by two synonyms). The full corpus consisted of 778 tweets. I then removed duplicated tweets and read all remaining tweets to identify cases in which the keyword expressed a meaning that was not associated with the study goal (i.e., talking about unity in a different context, e.g., in Ukraine) and to manually exclude them. After excluding these cases, the corpus comprised 598 tweets, which received 8,279 replies. In order to examine responses to tweets discussing unity, I randomly sampled ten tweets with a minimum of 10 replies and extracted all their replies, resulting in a total of 1,073 replies (13% of all replies). The analysis presented below includes all 598 tweets and these 1,073 replies.

Data analysis

To address the research questions, a dual process of data-driven, inductive, thematic analysis of tweets and replies was conducted based on Braun and Clarke’s (2006) six analysis phases. First, I read all tweets and a sample of replies as they appeared online on the Twitter platform, familiarizing myself with the data in its natural form, including the flow of the threads and different visual aspects. I then reread all tweets and replies (exported to an Excel document) and identified the main topics while generating an initial code. Some tweets and replies were coded into more than one category, as they discussed multiple issues. After reaching a saturation point (Strauss & Corbin, 1998), where no new categories were recognized, I searched for prominent themes overarching different codes while integrating them into broader conceptual units. Next, I moved to reviewing the themes and recognized that only a single theme reflected a trend of depolarization. Subsequently, I created a new list of tweets, removing those included under the depolarization theme, and reanalyzed the updated corpus to recognize discoursal mechanisms employed by Twitter users to preserve AP. I followed each step of the analysis again, from creating the initial code to reviewing the final themes, now guided by the objective of revealing the relevant discoursal mechanisms. This process led to the identification of three mechanisms: two content-related and one structural. I then defined and named each of the themes. Each stage of this process was documented in a separate working table (initial coding of themes, recognition of overarching themes, secondary coding of discoursal mechanisms). All tweets and replies presented in the findings chapter were originally written in Hebrew and translated into English. They are presented with their date of tweeting and a serial number (T# or R#) from the data file (supplementary material, in Hebrew).

Findings

The thematic analysis of tweets discussing the notion of unity following violent terror attacks in Israel resulted in recognition of a single theme indicating a depolarizing trend, reflected in authentic (and strategic) calls for Israeli unity. However, the prominent trend within tweets was that of perpetuating AP using two main content-related discoursal mechanisms: recalling the ‘unified’ past and distorting the meaning of the term ‘unity’. Interestingly, examining the data as it appeared on the Twitter platform led to the identification of an additional discoursal mechanism that goes beyond the content of the tweets to a structural aspect, demonstrating how Twitter users employ screenshots in a way that fixates public discourse and perpetuates AP.

Depolarizing calls for unity

Consistent with theoretical hypotheses regarding the effect of external threats on social cohesion presented above, the first theme includes depolarizing calls for unity within Israeli society following the terror attacks. Some users tweeted a general request to set aside political disagreements and work together as a unified society:

Maybe for one evening, we will be able to unite, hug, comfort … 

No accusations and no great declarations.

Some people are injured.

There are policemen who are still running in the streets.

Please, let’s leave aside the divisions … 

(07.04.2022, T#52)

In many tweets calling for depolarization, users directly stated that unity is required to cope with the ongoing security threat effectively. These users claimed that the fight against terror is doomed if society is fragmented. This is also demonstrated in the following tweet:

If we cherish life,

we must put aside all political disputes and gratuitous hatred to defeat the satanic terrorism that wants to sow destruction, devastation, and death in us.

We will win only if we are united

(29.03.2022, T#9)

Aside from wishing for depolarization, some Twitter users described how they were moved by acts of solidarity that took place in response to the terror attacks. For example, the following short tweet which quoted a tweet describing how teenagers from an Israeli settlement went to console the family of a Druze Israeli soldier who was killed in one of the terror attacks:

Unity moving

(30.03.2022, T#277)

Politicians, as well, were moved by such expressions of unity, as revealed in the following tweet posted from the account of Israeli Minister of Defense:

This picture of ultra-Orthodox people from all over the country who came to the cemetery in Nof Hagalil [a city in northern Israel] to pay their last respects to the deceased First Sergeant Major Amir Khoury [a police officer killed in a terror attack], moved me very much.

A sad and moving moment showing Israel’s unity.

May his memory be a blessing.

(31.03.2022, T#16)

However, it is important to note that, to some extent, depolarizing calls for unity might derive from instrumental considerations, mainly when those who convey these calls are individuals in positions of power, such as politicians, who believe they might benefit from publicly expressing a unifying agenda.

As demonstrated in this theme, the data collected suggest that there were depolarizing calls on Twitter following the terror attacks. While some users expressed a genuine wish for unity, in some cases, it is hard to tell whether the call for depolarization was authentic or contaminated with strategic considerations. Furthermore, quantifying the proportion of the depolarizing tweets in the corpus reveals that they are limited (only approximately 17% of all tweets) and thus do not provide good representation of the general discourse.

Discoursal mechanisms perpetuating AP

Since the majority of tweets and replies in the corpus reflected a trend of perpetuating AP, additional analysis of the corpus of tweets was conducted to reveal the discoursal mechanisms involved in this process. Based on this analysis, two content-related mechanisms and one structural mechanism were identified.

1. Recalling the ‘Unified’ Past

As the notion of unity became salient in public discourse during the terror wave, some Twitter users idealized the idea of unity but simultaneously suggested that such unity was impossible given the severe AP within society. These tweets express pessimism about the external threat’s potential to depolarize the relationship between political groups. The users who tweeted them seem to believe that the current polarized reality is resistant to change:

The Israeli of love for your neighbor like yourself, of all Israelis are friends and can depend on one another, of the brotherhood of warriors, of unity and resilience in days of pain, of all the symbols on which the state was founded – where are all these hidden? Terror will not defeat us! But the hatred between the right and the left will … It is a shame that ‘bloodthirsty’ people sit in the Knesset, for whom a painful attack serves as a platform to spread hatred and lies.

(30.03.2022, T#374)

In many of these pessimistic tweets, users nostalgically contemplate previous times when, according to their claims, Israelis did know how to unite when the circumstances demanded it, while today, unity does not seem plausible:

We know that we are a divided nation, but previously we found comfort in our unity in times of crisis; now, we don’t have that either.

(30.03.2022, T#60)

Similarly, the following tweet mourns the current inner hatred:

We used to be united in national crises.

We understood that we had to fight our enemy.

Not today.

Self-interest and hatred have prevailed over values and empathy.

Our situation is not good.

(08.04.2022, T#557)

While the pessimistic tweets themselves do not actively increase AP, they contribute to its stagnation. Tweets on this theme reveal that individuals are displeased or even severely indignant about the polarization (that appears to be resistant even to events that previously decreased inner animosity). However, rather than advocating for change, they use Twitter to vent their frustration, fostering a norm of despair. Consequently, polarized reality becomes more stable as people lose faith in its ability to moderate.

• 2. Distorting the Meaning of Unity

Another content-related mechanism, which was even more prevalent in the data, was distorting the meaning of unity. Tweets included in this theme suggest that while the external threat created momentum for discussions of unity, many Twitter users used the term in a way that modified it and distorted its meaning. As a result, these tweets maintained the polarized nature of the discourse. This distortion had two main manifestations: calling for unity while attacking the opposite political group and calling for unity only amongst subgroups of Israeli society.

• 2.1. Calling for Unity While Attacking the Opposite Political Group. The tweet that evoked the highest number of replies in the corpus was posted by a coalition MK [Knesset member] a day after a terror attack in which five people were killed:Precisely at this time, when the wave of terrorism hits us, we must be united.The exploitative and macabre provocation attempt by the opposition endangers the continuation of the state’s existence.(30.03.2022, T#1)

Many of the replies to this tweet involved intense criticism from users who accused the MK of being a hypocrite, highlighting the inner dissonance embedded in the tweet; on the one hand, ‘we must be united’, and on the other hand, the opposition endangers us all. This pattern of calling for unity while attacking the other political group appeared in many tweets in the corpus. For example:

I absolutely believe that our strength is in our unity and that unity will defeat any enemy!

The problem is that the leftists believe that ‘our strength in our unity’ means unity with terrorists, and I really do not agree with that.

So, all you do-gooders from the left who tweet about unity – go to the families of the murdered and ask for their forgiveness for your part in the crime!

(30.03.2022, T#303)

Or:

It really feels to me that the left hates the right more than it hates the terrorist organizations.

Why can’t we unite in difficult moments?

Did we lose that too?

Frustrating

(30.03.2022, T#46)

These examples suggest that some Twitter users, while emphasizing the need for unity during difficult threatening days, concealed a contradictory message that included criticism of and even an attack on the opposite political group. In other words, while the users talk about unity, the sub-text implies that the division is being preserved.

• 2.2. Calling for Unity Only Amongst Subgroups of Israeli Society. Another distortion of the meaning of unity is reflected in tweets calling for unity only within subgroups of Israeli society (right-wingers, Jewish Israelis, secular Jews, etc.). Twitter users highlight the boundaries of each subgroup and emphasize the inner divisions of Israeli society. Often, users explicitly mention other social groups they exclude from their ideal unified society. For example, some tweets, posted primarily by rightist users, refer to unity or a unity government that does not include Arabs:This is war!! We need a unity government!! Jews, we need to unite against the Arab enemy that has risen up against us to destroy us!! (29.03.2022, T#424)

Others also exclude leftists:

In these difficult days, we must unite. Arabs and leftists will not deter us!

(30.03.2022, T#180)

Similarly, others call for unity only among right-wingers:

We do not want to be united with supporters of terrorism! We on the right are united enough to purge out the leftists from among us.

(30.03.2022, R#529)

In contrast, leftist users call for unity without the ultra-Orthodox:

There is no unity, and there will never be unity with the ultra-Orthodox and their helpers and natural partners.

The Zionist left parties and the pragmatic moderate right parties must compromise to join hands for the future of our children without ultra-Orthodox, for security, and for a prosperous and progressive country that seeks peace, justice, and truth.

(08.04.2022, R#377)

Together with the previous sub-theme, the tweets presented here indicate that some Twitter users modify the meaning of the term unity to fit their much narrower definition of the ideal Israeli society. The tweets presented in this theme may hint that real unity and depolarization are not the goals of some Twitter users who use the platform and exploit external threatening circumstances to continue propagating divisive messages.

Tweeting screenshots – ‘Freezing’ the discussions

Reviewing the data collected for this study as it appeared online on the Twitter platform led to the detection of a third discoursal mechanism. Unlike the two content-related mechanisms discussed so far, the third mechanism is structural, reflected in the use of a specific digital tool: screenshots. Screenshots are images created by capturing a device’s screen at a given moment (Frosh, 2019), and they are saved and shared mainly in digital space (Corry, 2021). In the current corpus, screenshots have been used to recall past divisive messages of individuals who currently call for unification. For example, the following text was tweeted along with two screenshots:

Friends, don’t you dare fall for the messages of the left calling for unity.

Look at Jane Doe’s [name anonymized] tweets from today and read her tweets from not long ago.

(30.03.2022, T#430)

The first screenshot was of a tweet posted by Jane Doe, who is an Israeli publicist, that same day, during the escalation of the conflict, saying:

Even today, the greatest danger is the hatred among us.

The second screenshot was of a tweet posted by her a year and a half earlier, in July 2020, saying:

Still, in honor of Tisha B’Av [a Jewish fast commemorating the destruction of the temple believed to be caused by gratuitous hatred], I have a message that will bring us closer together: leper rightist dogs, we will take care of you if you come near us.

This earlier tweet from 2020 represents an extreme form of polarized discourse. Jane Doe sarcastically says that she has a unifying message in honor of the day that has become a symbol of Jewish brotherhood and resistance to hatred among Jews, and then dehumanizes rightists and threatens them.

In this example, Jane Doe’s screenshots were used to reject her recent depolarizing message by raising her old polarizing message from more than a year before. The user who tweeted these screenshots used them to cast doubt over Jane Doe’s current intentions, calling others not to believe the authenticity of her depolarizing calls. Indeed, the change in the content and tone of these tweets could imply hypocrisy underlying the later tweet; however, the change over time could also reflect a genuine transformation of attitudes. Importantly, without a shift in attitudes, depolarization – a dynamic process of moderating sentiment, attitudes, and opinions (Abeywickrama & Laham, 2020) – cannot occur.

Another example of the use of screenshots as an obstacle to depolarization appears in the following tweet (Figure 1), which was written with a pinch of cynicism:

The opposition member presents loyalty to the state, unity, and support for the government while visiting a bereaved family in a week when a murderous terrorist attack occurred. The year is 2020. The politician: Naftali Bennett. The message: the government is doing nothing, Gantz and Bibi are cowards, and they are not preventing the next murder.

(30.03.2022, T#27)

This tweet was accompanied by two screenshots of one long Facebook post (cut into two screenshots) by Naftali Bennett, the incumbent prime minister at the time of the escalation (when the tweet was posted), and an opposition member in 2020 when the screenshotted post was originally published. The tweet was obviously written in response to calls for unity by Bennett during the escalation, and the pinch of cynicism relates to the fact that in the past, when Bennett was an opposition member, he criticized the government’s handling of terror instead of providing the support he now seeks. The original post is long, and the following is a partial translation of one section (marked in Figure 2) where Bennett accuses the government at that time (2020) of not doing enough to prevent the next murderer:

Figure 1. Tweet including the screenshots of the original Facebook post by Naftali Bennett.

Figure 2. Screenshots of the original Facebook post of Naftali Bennett.

Listen, my friends:

The next time you hear the prime ministers speaking pompously: ‘We will strike the enemy’, ‘We will catch the murderers’, ‘We will defeat terrorism’,

You will know:

These are empty words.

They are *not* doing all that is possible to prevent the next murder.

This tweet again exemplifies how a screenshot is used to contradict and reject a recent call for unity by recalling a past divisive message. Taken together with the previous example, this theme demonstrates how the digital feature of screenshots functions as a barrier to depolarization, as it freezes individuals’ attitudes at a specific polarized moment and hinders discussions calling for moderation. This discourse stagnation perpetuates AP and thus hinders attempts to reconcile and unite society members.

Discussion

The study of the role of social media in depolarization has received relatively little attention over the years (Kubin & von Sikorski, 2021). Specifically, scholars have largely overlooked the question of how interactions between opposing partisans on social media in contexts that are thought to mitigate AP, such as external threats (Van Bavel et al., 2020), might affect (de)polarization. The current study aimed to bridge this gap by exploring whether Twitter discourse following terror attacks reflects trends of polarization or depolarization, and examining the discoursal mechanisms involved in promoting (de)polarization.

The findings demonstrate that online public discourse continues to be highly polarized even under an actual acute external threat. Although the first theme indicates that there were tweets genuinely calling for depolarization, they were limited and far from counterbalancing the abundance of tweets preserving AP. Furthermore, the research found three main discoursal mechanisms involved in perpetuating AP. The first two are content-related, demonstrating how recalling the ‘unified’ past and distorting the meaning of the term ‘unity’ enables Twitter users to preserve AP. The third mechanism is structural, showing how the use of screenshots functions as a barrier for depolarization. This last mechanism also sheds light on social media’s potentially unique role in hindering depolarization.

(De)polarization on social media in times of threat

Similar to John and Dvir-Gvirsman’s (2015) findings, the current research demonstrates that, at least in the current Israeli context, the response to external threats is not depolarization; instead, it is the perpetuation of AP through digital behavior. This may indicate that Twitter functions as a platform more prone to propagating hostility between opposing partisans than mitigating their mutual hatred. It is important to recall that there is scholarly disagreement about whether online platforms enhance animosity between political rivals (Nordbrandt, 2021). The current study, as qualitative and descriptive research, does not pretend to determine this debate; still, it provides additional evidence for the polarized nature of inter-party discourse on Twitter, even when external circumstances are hypothesized to induce cohesion (Coser, 1956).

Importantly, it is not my intention to imply that online platforms, Twitter included, do not have the capacity to promote depolarization. In fact, as presented before, the question of whether social media can support depolarization efforts is still under inquiry (Kubin & von Sikorski, 2021), and the ongoing work of some social scientists (e.g., Borrelli et al., 2022; Currin et al., 2022) suggests that it can be harnessed for reducing inter-party animosity. However, based on current findings, I suggest that depolarization is less likely to occur naturally in the digital environment and that, in order to mitigate AP, planned and deliberate actions are preferred.

How is AP perpetuated?

Content-related discoursal mechanisms

Usually, when considering what discoursal mechanisms might prevent depolarization online, what first comes to mind is incivility, use of hate speech, or dehumanization of political opponents (e.g., Suarez Estrada et al., 2022; Wahlström et al., 2021). The effect of these blatant and violent forms of communication is apparent; they reinforce the inter-party divide and escalate the conflict between the political groups (Harel et al., 2020). However, the discoursal mechanisms detected in the current research suggest that other, more subtle forms of communication can lead to similar consequences.

Tweets recalling the past or those distorting the meaning of unity were not necessarily offensive, aggressive, or uncivil in nature. On the contrary, many of the examples demonstrate a much more tolerant and moderate tone than we are used to seeing in online inter-party discussions. However, when the Twitter platform is employed to vent frustration with the polarized reality without actively taking steps to amend it, or when users discuss the need for unity while clearly distorting the term’s meaning and excluding those with whom they disagree, the outcome is the perpetuation of AP. This stagnation is a slippery slope that can potentially also deteriorate into intensified AP.

Screenshots as a digital barrier to depolarization

Alongside the content-related discoursal mechanisms, this study proposes a potential digital feature that might function as a barrier to depolarization, specifically in the online environment: screenshots. As demonstrated, some of the tweets containing reconciling attempts and depolarizing calls were replied to with screenshots of previous divisive messages tweeted by the same users, a gesture that I suggest freezes the discussion and perpetuates AP. Notably, screenshotting – the ability to cut, paste, and share digital content – is a feature that is not platform-exclusive, i.e., it is an integral, and even infrastructural, feature of the digital environment as a whole. Furthermore, as demonstrated in this study, screenshots also bridge different platforms and enable the transfer of content from one to another (see Image 2, a Facebook post shared on Twitter), signaling the continued importance of the new media principle of ‘modularity’ whereby media elements can be moved across documents and platforms without losing their independence (Manovich, 2002). Therefore, discussing the implications of screenshots goes beyond the specific Twitter platform studied here.

People ‘take’ screenshots for various personal and communicational reasons (see Cramer et al., 2019, 2022), from remembering a discount code found online before it disappears to sharing messages from a closed group with people who do not have access to the original content. Although scholarly efforts to understand how screenshots ‘operate’ are quite limited (Frosh, 2019), existing work suggests that screenshots have a “‘social’ life beyond this technological function” (Jaynes, 2020, p. 1378) and have implications for different sorts of social interactions, from interpersonal relationships (Jaynes, 2020) to social movements (Jenkins & Cramer, 2022), where they function as assets of knowledge and power through their evidentiary nature. The current research broadens the scope of this discussion by demonstrating how screenshots might also limit or even prevent social change, and specifically, in the current case: depolarization.

Depolarization, like other social change processes, requires an environment that affords the revision and modification of attitudes (Abeywickrama & Laham, 2020). Screenshots, however, function as a tool that aids in creating a static and fixed reality (Frosh, 2019) in which past attitudes can repeatedly be retrieved and recalled. As demonstrated in the findings chapter, captured images of past tweets function as ‘evidence’ (Frosh, 2019) of what is considered the authentic opinion/agenda/worldview/attitude of a user and to cast doubt on his/her recent tweets. Besides the fact that reading someone’s depolarizing tweets alongside former polarizing tweets highlights inconsistency and might lead to the perception that later tweets are dishonest and misleading, screenshots have their own rhetorical power as reliable evidence documenting the ‘truth’ (Frosh, 2019) – even if this reliability can be questioned (Švelch, 2021). Furthermore, since screenshots create a binary temporality of past and present without showing any process or development that might have occurred in the middle, users might think that the updated (moderated) attitude presented in later tweets is groundless, and thus, less trustworthy. This way, screenshots freeze reality and discussion at a given polarized moment and prevent the possibility of change.

Screenshots also present a distinct challenge to depolarization, particularly within AP’s intergroup and emotional dimensions. Firstly, based on the research findings, sharing screenshots heightens outgroup negativity by obstructing intergroup bridge-building and reinforcing ingroup bonds (Sumaktoyo, 2021). Specifically, sharing screenshots exacerbates intergroup animosity by exposing perceived hypocrisy in the opposing group. Simultaneously, it strengthens ingroup ties by signaling mutual values (Inwood, 2021) and silencing alternative voices, thus reinforcing echo chamber dynamics. Secondly, the threat to depolarization lies in screenshots’ ability to evoke intense emotional responses (Corry, 2021). Operating as a second-level digital documentation practice, screenshots detach content from its original context and increase its visibility (Bigman et al., 2023). This often involves the transfer of content from private to public spheres, violating users’ perceived privacy (Shore & Prena, 2023). Corry (2021) demonstrated that this content transfer can trigger a significant sense of shame. The current focus on screenshots that are taken out of context and presented as if they are revealing the hypocrisy of the outgroup broadens the emotional repertoire elicited by screenshots: the user whose content was screenshotted might feel anger due to the distortion of his words, while his ideological rivals viewing the screenshot might experience anger, contempt, and disgust – typical emotional reactions to unethical behavior (Hutcherson & Gross, 2011). Consequently, these negative emotions further fuel AP.

If, before the age of social media, public figures were the only ones who were confronted with their previous statements, today, with digital communication’s affordances of persistence and archivability (Evans et al., 2017), every person who tweets (or posts, or, unfortunately, even sends a private digital message), must take into account that the content s/he writes might be screenshotted and saved, and can be used against them sometime in the future (Corry, 2021). These digital abilities should have potentially prompted digital media users to expand their content’s ‘imagined audience’ (Zappavigna & Martin, 2018) so it will also include future readers who may lack relevant context. While this awareness does not always translate into practice, when it does materialize – and only when considering screenshots of public content like tweets on Twitter – it may offer a potential benefit by fostering accountability for public statements. This awareness could encourage users to exercise more caution and civility in their online communications, potentially contributing to a more constructive inter-party dialogue. However, the fact that screenshots also act as gatekeepers that prevent attempts by individuals to moderate their attitudes and influence others to do so as well, as seen in current findings, raises serious concerns about the prospect of discourse on social media to mitigate AP.

Limitations and future research

While this study enhances our understanding of how social media might perpetuate AP in the context of external threats, there are some limitations. First, it relies on a single case study with unique characteristics, such as the intractability of the Israeli-Palestinian conflict (Bar-Tal, 2007) and the recent political instability in Israel (Shamir & Rahat, 2022). Future studies could explore the research question in additional contexts using inductive or deductive approaches, searching for new themes or supporting the present ones. Secondly, the study uses manual data analysis, allowing in-depth exploration but also limiting the scope. Future research could use advanced computational tools to examine on a large scale whether discourse on social media during or following acute threats becomes less or more polarized. Future research can also further examine the implications screenshots might have for social change, for example: Do people believe the depolarizing calls after being exposed to old screenshots with a divisive message? Do people support depolarization more or less? And, in which other contexts do screenshots play a similar role and why?

Conclusion

The current study demonstrates the affectively polarized nature of political discourse on Twitter in response to violent terror attacks. It reveals discoursal mechanisms involved in perpetuating AP in such circumstances and highlights how even subtle non-offensive forms of communication might hinder depolarization. Furthermore, the findings point to screenshots as a unique digital feature that might contribute to the perpetuation of AP, as they ‘freeze’ individuals’ attitudes and sentiments in a specific polarized moment and are used later to cast doubt on their updated and more moderated attitudes. In this way, screenshots function as a barrier to depolarization and limit the prospect of successful social change on social media.

Acknowledgments

I would like to extend my sincere gratitude to Prof. Ifat Maoz and Prof. Paul Frosh for their valuable advice and feedback during the writing of this article. Their insights have greatly contributed to its development.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Tal Orian Harel

Tal Orian Harel is a Ph.D. candidate at the Department of Communication and Journalism at the Hebrew University of Jerusalem and a doctoral research fellow at the Truman Institute at the Hebrew University. Her research interests include political discourse, social media, political polarization, and the psychology of intergroup conflict.