Abortion discourse is often characterised as a hyper-polarised, moral-political issue infused with excessive negative campaigning, invective, and radical tactics. In the last decades, some academics came to demoralising conclusions that the abortion issue illustrates the failure of democracy to create a public sphere to effectively deliberate deeply contested and divisive issues (Ferree et al., 2002; Hunter, 1994).

Given the rise of abortion politics around the world including the United States (US) and Ireland (Field, 2018; Ziegler, 2023), this paper reports on a timely, topical study assessing the democratic qualities of contemporary abortion discourse using two deliberative criteria: civility and tolerance – two important qualities of mutual respect to build constructive disagreements (Gutmann & Thompson, 1996; Habermas, 1996, 2003). Mutual respect is especially required in abortion discourse as abortion politics involves conflicting positions with incommensurable worldviews and morality that cannot be solved with ‘rational’ persuasion (Ferree et al., 2002; Hunter, 1994). Our findings can also provide insights about the public’s capabilities to deliberate a range of other emotional and polarised culture war issues including LGBT+ rights (Hunter, 1994). The US-Ireland comparative research design enables us to identify the cross-national comparability of our findings.

This paper focuses on a single social media platform, Twitter, which contributes to facilitate digital ‘issue publics’ (Habermas, 2006), bridging between official political discourse and civil societies, and between elite media and public communications (Bruns & Highfield, 2015). Elected politicians and journalists use Twitter, more than other platforms (Elayan et al., 2020). The public use Twitter to engage in political conversations and form deliberative publics (Wright et al., 2017). This political affordance of Twitter makes the platform suitable to study the deliberative qualities of public discussions on abortion. We use two Twitter datasets mentioning the American (2020) constitutional discussions on Roe v. Wade and the Irish (2018) referendum for a constitutional amendment on abortion bans.

We develop and test a set of expectations to explain the dynamics of incivility and intolerance at micro and meso levels. On the micro-level, we analyse the demographics of those who are more likely to express incivility and intolerance in online abortion discourse. We also assess the user distribution of uncivil and intolerant tweets to investigate whether incivility and intolerance are common behaviours for all actors involved in the discourse, or unusual behaviours mobilised by a small set of particular users. On the meso-level, we examine the role of contextual factors such as the type of discussion environments in which a tweet is embedded.

Our results show that the level of incivility and intolerance expressed during the 2020 American and 2018 Irish abortion discourse on Twitter is low over time, but these uncivil and intolerant tweets generate wider public engagements than civil and tolerant tweets. Our micro-level analysis shows that pro-life1 Twitter users are associated with intolerance in both countries. We also observe the coordinated dynamics of incivility and intolerance, where a small number of hyperactive users dominate the entire uncivil and intolerant communications. In our meso-level analysis, incivility is associated with a heterogeneous discussion environment whereas intolerance is associated with homogeneous echo chambers.

Below, we explain how we operationalise civility and tolerance as two important normative criteria of deliberative democracy. We conceptualise Twitter as issue publics (Habermas, 2006). We then summarise the history of abortion discourse in the US and Ireland before we measure and analyse incivility and intolerance in the American and Irish abortion discourse on Twitter and collate the results for cross-national insights.

Deliberative democracy is the notion that a democratic society’s rules and conditions are justified through public debate among equal citizens. (Cohen 1989). Habermas (1996) idealises the power of rational and reasonable deliberation to solve socially shared problems. Gutmann and Thompson (1996) similarly describe deliberative democracy as the obligation to provide reasonable justification for public claims (that transcend their narrow self-interest) to deal with social conflicts.

Decisions on public affairs are made at the political centre (e.g., by governments etc) but discussions influencing the decision must incorporate peripheral voices – including autonomous citizens representing non-bureaucratic, personal and everyday experiences (Ferree et al., 2002). Habermas assumes that autonomous citizens deliberate more freely than formal-bureaucratic actors since they are free from the burden of making decisions and institutional regulations and therefore can respect the better arguments and take the viewpoints of other actors (Ferree et al., 2002). Autonomous citizens and organisations are crucial in deliberative democracy since their voices bring new social issues and perspectives to the political and public spheres, contributing to possible solutions (Habermas, 1996, 2006).

The normative standard of mutual respect is central to the deliberative model of democracy. It safeguards deliberative politics whereby public decisions are made by the strength of the arguments, and not by coercion, or prejudice. Habermas’ ‘ideal speech situation’ (1990) encapsulates normative deliberation by endorsing mutual respect despite deep disagreements. Gutmann and Thompson (1996) argue that ‘citizens do not end serious moral conflict, but […] they accept significant parts of the substantive morality of their fellow citizens to whom they may find themselves deeply opposed in other respects’ (p.89).

Based on the previous literature on deliberation, we argue that mutual respect consists of two aspects: civility and tolerance. Civility here refers to the social norms of interpersonal relationships and politeness. Incivility is a violation of the civility norm such as threatening opponents, name-calling, rudeness etc. This relates to the expectations of deliberation as civilised and reasonable discussions. This does not mean civility is tantamount to emotional detachment per se, but it nonetheless contains efforts to keep one’s negative emotions toward opponents in check.

Tolerance refers to the moral-political respect for each other with which deliberation participants actively promote and recognise significant parts of the morality of their opponents as fellow citizens (Gutmann & Thompson, 1996). Intolerance in deliberative politics means the lack of such moral-political respect for fellow citizens such as demonising the other side as something that cannot be tolerated and absolutising one’s unique ethical view as a generally binding rule for the whole society (Habermas, 2003).

Numerous insightful studies investigated diverse qualities of public deliberation such as rationality, sourcing, reasoned opinion expressions through content analysis (Stromer-Galley, 2007) as well as (in)civility (Rossini & Maia, 2021) and participant (in)equality (Maia et al., 2017). Our research aims to contribute to deliberation scholarship by presenting a novel way to assess the level of mutual respect in online issue publics through two criteria: (in)civility and (in)tolerance.

Deliberative democracy emphasises the importance of autonomous groups of citizens in the public spheres. Habermas views public spheres as a space where public actors join public debates, influencing the state’s decision-making. Habermas (2006) acknowledges the continued structural transformation of the public spheres due to the Internet. In his words (2006), the Internet reintroduces deliberative elements and reactivate an egalitarian public of writers and readers who were relegated as audience in mass media communications.

His 2006 essay also discusses the Internet and fragmentation of issue publics: ‘the rise of millions of fragmented chat rooms […] lead[s] to the fragmentation of large but politically focused mass audiences into a huge number of isolated issue publics’ (p.422). Other scholars further explored ‘issue publics.’ Dahlgren (2009) argues that issue publics ‘emerge, exist for varying durations, and then eventually dissolve’ as the public debates move on (p.74). Bruns and Highfield (2015) argue that issue publics form around short-lived topics and events that exist in connection with wider public spheres around specific themes. While the wider public sphere addresses the long-term issue of abortion, issue publics would form around specific policy initiatives and other short-term aspects that ignite public debates (Bruns & Highfield, 2015).

Social media sites such as Twitter can facilitate the formation of issue publics (Bruns & Highfield, 2015). For many on social media, Twitter is a preferred platform for news, information consumption and political participation such as campaigning, activism, and deliberation (Wright et al., 2017). Hashtags on Twitter enable users to join discussions and form issue publics (Bruns & Highfield, 2015). Abortion discussions on American and Irish Twitterspheres constitute abortion issue publics that invite clusters of citizens and public actors (e.g., politicians). However, Twitter research is not without its limitations and biases and therefore any findings in this paper should not be automatically generalised to infer attitudes and behaviours of the American and Irish population as a whole.

Furthermore, deliberation consists of rational and normative standards that are difficult to be rigorously met in online discussions. Nonetheless, assessing the volume and dynamics of incivility and intolerance in online abortion discourse is worthy of academic attention to understand how the public engages in everyday opinion expressions and discussions about highly emotional and polarised moral-political issues like abortion.

This paper evaluates the degree to which the 2020 American and 2018 Irish abortion discourse on Twitter satisfies or violates the deliberative criteria of civility and tolerance. The US and Ireland are chosen as case-studies because their constitutions on abortion were among mainstream political priorities in recent years.

In the US in 1973, Roe v. Wade was a landmark decision in which the Supreme Court ruled that the Constitution protects pregnant people’s liberty to terminate pregnancy without excessive state restrictions (Ferree et al., 2002). Since Roe, American pro-life movements have tried to introduce creative ways to legislate abortion restrictions or even to reverse Roe. With the revival of Christian nationalism, reactionary backlash, and the presidential election of Donald Trump, American pro-life movements gained new momentum (Norris & Inglehart, 2019; Ziegler, 2023). In 2020, there were heated public discussions about the precarious future of Roe alongside debates around the implications and future of abortion bans and Planned Parenthood, the American reproductive healthcare provider. The newly appointed and nominated conservative Supreme Court Justices during the Trump administration such as Justice Kavanaugh and Justice Barrett ignited public discussions about overturning Roe. At the time of writing this paper, the new conservative majority Supreme Court overturned Roe in June 2022 (Ziegler, 2023).

In Ireland, a historically Catholic country, abortion was banned for a long time. In 1983, the 8th amendment of the Irish Constitution was implemented to ‘acknowledge the right to life of the unborn.’ There were several more referendums to decide further restrictions on abortion such as to remove the threat of suicide as a ground for legal abortion (12th and 25th amendments) and restrict pregnant people’s rights to travel for abortion and to access information about abortion clinics abroad (13th and 14th amendments) (Field, 2018).

Abortion remained absent from Irish mainstream priorities in past decades (Field, 2018). The death of Savita Halappanavar in October 2012, who died of sepsis after being denied a legal abortion, ignited the debate which led to another abortion referendum in 2018 (Field, 2018).

Formal deliberation is another characteristic of the Irish abortion discourse. In 2016, the Citizens’ Assembly was established to deliberate on how to change the Irish abortion constitution. After weeks of deliberation, the Citizens’ Assembly provided recommendations to the Oireachtas (Irish legislature) (Field, 2018). The use of Citizens’ Assemblies in Ireland (in both the marriage equality referendum in 2014 and the abortion referendum in 2016) highlight that Ireland is at the forefront of deliberative experiments among European democracies and that Irish citizens have hands-on experiences with the learn-deliberate model (Escobar & Elstub, 2017; Suiter, 2018).

In May 2018, Ireland held a referendum for the 36th amendment of the constitution to repeal the 8th, 13th, and 14th amendments of its Constitution. Repeal won by a margin of 66.4% to 33.6% (Field, 2018). That year, the President of Ireland signed the Health Regulation of Termination of Pregnancy Act, defining the circumstances and processes within which abortion is legally performed in Ireland.

It must be noted that the American abortion discourse in 2020 and Irish discourse in 2018 are not directly comparable, given that the former case is more of a spectator discourse where the public expressed and reacted to news stories about policy and legal decisions whereas the latter is a voter communication centred around the referendum with involvement of institutionalised actors like politicians, activists, etc. Also, the Irish abortion discourse involved formal deliberation whereas the American discourse largely consisted of unorganised public opinions. Nonetheless, interrogating the differences between American and Irish abortion discourse on Twitter can yield insights about the cross-national comparability of our findings.

Our first research question investigates the volume of incivility and intolerance in the American and Irish abortion discourse on Twitter.

Given the more radical and polarised nature of abortion politics in the US involving violence and radicalism (Ferree et al., 2002; Hunter, 1994) in comparison with Ireland with public experiences with formal deliberation on abortion constitutions (Suiter, 2018), we expect that American data will involve more anti-deliberative behaviours than the Irish data.

In addition to comparing the national differences, we investigate the micro and meso factors influencing the expressions of incivility and intolerance in abortion discourse. We aim to ascertain whether a specific stance on abortion significantly correlates with anti-deliberative behaviours. While uncivil, disruptive communication tactics are used by both liberal and conservative political actors (Berry & Sobieraj, 2013), social and political intolerance are strongly associated with right-wing politics (Norris & Inglehart, 2019). Given the history of anti-abortion violence and its ties to reactionary politics (Ferree et al., 2002; Norris & Inglehart, 2019), we hypothesise a stronger association between pro-life stance and intolerance. Our hypothesis 2 anticipates that:

Moreover, we explore the dynamics and distribution of these uncivil and intolerant behaviours; whether they are common across all participants or specifically mobilised and coordinated by a smaller set of hyperactive users. Whether anti-deliberative behaviours are common will have different implications for our analysis of deliberative qualities of abortion discourse on Twitter. Theocharis et al. (2020) analyse the coordination dynamics of Twitter incivility directed at American politicians and conclude that incivility is a common behaviour among angry, frustrated citizens to criticise and insult politicians. Since our Twitter dataset is specifically confined to abortion issue publics, we examine whether the degree of mobilisation of incivility and intolerance is concentrated on few but committed-hyperactive actors.

Our hypotheses and research questions lead to another question. Regardless of whether a specific type of user expresses incivility and intolerance often, are there specific interactive contexts that trigger larger outbursts of incivility and intolerance? The interactive communicative contexts of Twitter discussions, in combination with the individual characteristics of Twitter users, may drive more incivility and intolerance than in other communicative contexts.

Several studies investigated how discussion environments influence the level of incivility and intolerance in user communications. Rains et al. (2017) find that the proportion of in- and out-group members in online news comment sections affects conservative and liberal users’ tendencies to be uncivil to their opponents. Rossini (2020) finds that incivility in Facebook news comments is associated with expressions of disagreements whereas intolerance occurs in homogeneous discussions about minorities.

In this paper, we focus on whether incivility and intolerance occur in inter-group engagements or in self-imposed echo chambers2 where users only communicate with already like-minded users (Bruns, 2019; Garrett, 2017). Our expectation is that incivility occurs more when users are actively engaging and arguing with one another in a heterogeneous discussion environment whereas intolerance occurs more frequently within homogeneous echo chambers, when users are sharing their absolutist and extreme views with like-minded people (Rossini, 2020).

Table 2 shows how the American and Irish Twitter datasets are constructed. Both datasets contain large portions of tweets posted by gender-unidentifiable users. One supposition might be that many users might have been concerned about voicing their views in the abortion issue publicly with their gender-identifiable names. This observation could prompt future studies to examine who uses anonymity and why.

Some notable differences between the two are: (1) the Irish dataset contains a larger portion of tweets by women than the American dataset. (2) The Irish dataset is predominantly pro-choice whereas the American data has a near-equal mix of pro- and anti-abortion tweets. (3) The Irish data consists of retweets followed by quotes and original tweets whereas American data consists of retweets followed by replies and quotes. Lastly, (4) the American dataset contains more incivility and intolerance than the Irish data (H1), which could be explained by the radical and violent history of abortion activism in the US (Ferree et al., 2002; Hunter, 1994).

Overall, both countries’ Twitter abortion discourse involves a small amount of uncivil and intolerant communications. 15.3% and 10% of American tweets and 9% and 1.7% of Irish tweets contain some form of incivility and intolerance (RQ1). The occurrences of incivility and intolerance in our datasets are not particularly higher than the volume of incivility and intolerance found in other online political public spheres (Coe et al., 2014; Rossini, 2020; Theocharis et al., 2020). This indicates that the impact of incivility and intolerance is felt more strongly than their quantitative shares of the total abortion discourse (Ferree et al., 2002).

We calculated Pearson’s correlation coefficients (r) between incivility and intolerance. The coefficient between incivility and intolerance is 0.22 in the American data and 0.09 in the Irish data, showing positive linear trends. However, the sizes of both coefficients are weak.

Table 3 below shows the distribution of uncivil and intolerant tweets in the American and Irish datasets. To investigate the reach of the uncivil and intolerant tweets, we calculated the mean, median, and standard deviations (SDs) of the amount of public engagements the uncivil and/or intolerant tweets generate.

Different types of tweets got more engagements in American and Irish Twitterspheres. In the American dataset, tweets that do not contain incivility nor intolerance earn the most retweets on average. As for the median of retweets, uncivil and intolerant tweets got most public engagements. In the Irish case, uncivil tweets earned the most retweets both in terms of their mean and median values. Considering median measures is important here, since distributions of retweets can be skewed by a small proportion of extremely popular tweets and a large portion of tweets that earned zero retweets. This indicates that although incivility and intolerance are uncommon behaviours, they still provoke huge volumes of engagements thereby influencing the atmosphere and direction of online public debates as well as the wider discussions on news media and elite discourse outside of Twitter.

To test other research hypotheses, a logistic regression analysis examines the relationship between diverse demographic characteristics and communicative contexts, incivility, and intolerance. Log odds are transformed to odds ratios to facilitate interpretation and 95% confidence intervals (CIs) are reported in addition to p-values. Reporting CIs is especially important to complement the error-prone nature of p-values in big data studies (Lin et al., 2013). Results of the regression analysis are in Table 4.

The results show that the pro-life issue position is strongly related to more intolerance in both countries. The odds of intolerant communications for pro-life tweets are 8.53 times higher in the American data and 12.59 times higher in the Irish dataset than pro-choice tweets. The relationship between incivility and abortion issue positions shows conflicting results. The odds of uncivil tweets for pro-life tweets are 2.19 times higher in the US, but 0.89 times lower in Ireland. This shows that in the American Twittersphere, pro-life tweets are more likely to be uncivil and intolerant, while in the Irish Twittersphere, pro-life tweets are more associated with intolerance whereas pro-choice tweets express more incivility (H2 supported).

Furthermore, our micro-level analysis shows that uncivil and intolerant Twitter communications are mobilised by a small number of ‘superparticipants’ (Graham & Wright, 2014) (H3 rejected). In the American dataset, the Gini coefficient for the tweet distribution is 0.6. The level of coordination is larger for uncivil and intolerant tweets: 0.85 for incivility and 0.92 for intolerance. In the American abortion discourse on Twitter, the top 1% of most active users are responsible for 27.1% of the total incivility and 39.5% of total intolerance. This pattern is also observed for our Irish dataset where the Gini coefficient for the tweet distribution is 0.73. The Gini coefficient for incivility is 0.88 and the coefficient for intolerance is 0.97. The top 1% of the most active Irish users express 41% of the total incivility and almost entire intolerance in the dataset. Figure 1 visualises the Lorenz curves for the distribution of total tweets, incivility, and intolerance in the two datasets.

The regression model also reveals a meso-level relationship between discussion environments and incivility and intolerance. The odds of incivility are 1.26 times higher in the American data and 1.11 times higher in the Irish dataset when a tweet is embedded in a heterogeneous environment where Twitter users are actively disagreeing with other tweets. In contrast, the odds of intolerance become 0.55 times lower in the American data and 0.85 times lower in the Irish data, meaning that intolerance is strongly associated with a homogeneous environment, lacking disagreements and inter-group engagements (H4 confirmed).

We explore Twitter discussions as a facilitator of abortion issue public, evaluating the public capabilities for deliberating highly divisive culture war issues like abortion (Hunter, 1994) while observing to the normative deliberative criteria of mutual respect. We operationalise mutual respect with two concepts: civility (interpersonal politeness) and tolerance (moral-political respect). Studying American and Irish cases together provides findings on the potential similarities and transferability of the results.

Despite choosing an emotive and polarised issue like abortion, the Twitter discussions in the US and Ireland are mostly civil and tolerant. This contrasts with perceived culture wars and antipathy in their abortion debates (Brownstein, 2021) suggesting that the impact of incivility and intolerance outstrips their actual representation in the discourse (Ferree et al., 2002). However, we find that uncivil and intolerant tweets generate more public engagements, potentially influencing a wider discussion.

We also find that American abortion discussions contain more incivility and significantly more intolerance than Irish discussions, potentially due to America’s history of violent abortion activism (Ferree et al., 2002) and Ireland’s experience with deliberative minipublics over contentious issues e.g., same-sex marriage (Suiter, 2018). This invites further research into whether the institutional uses of deliberative politics increase deliberative qualities of online discussions globally.

The regression analysis reveals that uncivil and intolerant communications on Twitter are associated with different demographic and discussion contexts. The micro-level analysis identified intolerance as a distinctively pro-life behaviour in both American and Irish Twitterspheres, supporting existing literature linking uncivil society with reactionary right-wing discourse (Norris & Inglehart, 2019).

In our meso-level analysis, we find that incivility prevails in heterogeneous environments whereas intolerance occurs within self-selected echo chambers (Rossini, 2020). This indicates the necessity to intervene in spiralling intolerant echo chambers.

Another noteworthy micro-level finding is the coordinated dynamics of uncivil and intolerant behaviours (Theocharis et al., 2020). Uncivil and intolerant discourse on Twitter is largely expressed and circulated by a small set of ‘superparticipants’ (Graham & Wright, 2014). The overrepresentation of hyperpartisan voices on social media leads to a distorted understanding of how polarised and extreme the opposingside is, making politically moderate citizens leave the online discourse (Bail, 2021) creating subsequent inequalities in policy outcomes where hyper-partisan voices are strongly represented and heard more by politicians and policymakers in digital spaces (Kleiner, 2018). It is crucial for online deliberation to curtail the capabilities of these superparticipants distorting and radicalising the overall online political discourse. The development of a meaningful typology of hyperactive users to generate a deeper understanding of who these people are and how they operate is needed.