Research Review
Norms are the unwritten rules of society: they’re what people think other members of their group or society find it acceptable to say, do, and even think. Here we focus mainly on norms about how people speak and write, which are called discourse norms.
Counterspeech can change discourse norms by shifting beliefs about what speech is acceptable. It usually starts small, such as in a single comment thread. As people see and feel discourse changing, some of them will shift their own behavior in response. Over time, this can expand to more and more people.
A study in Slovakia in 2016 and 2017 (Miškolci et al., 2018) tested whether counterspeech could improve discourse norms about Roma people on Facebook. Looking at over 7,500 Facebook comments, they found that when people saw anti-Roma comments followed by counterspeech in a comment thread, they were more likely to post their own comments in support of Roma people. (However, counterspeech did not change the behavior of the people posting the anti-Roma comments).
Other studies have found similar results. Han and Brazeal (2015) found that when study participants were exposed to civil comments, they were more willing to participate in online discussions and more likely to contribute civil comments than participants who had been exposed to uncivil comments.
Conversely, other studies (Cheng et al., 2017; Kwon & Gruzd, 2017) found that exposure to anti-social or negative comments made a person more likely to post an anti-social comment.
A study of #ichbinhier (the German branch of the #iamhere counterspeech network) provides further evidence that internet users take discourse cues from others. Using a dataset of comment threads in which group members engaged over a four month period, researchers found that top-level comments their team deemed “deliberative” were associated with more deliberative second-level comments (Friess et al., 2020).
Changing the mind or the behavior of the original speaker is difficult, but not impossible.
Some studies have looked at whether counterspeech can change the beliefs or behavior of the person it’s responding to, i.e. persuade them to apologize or stop posting harmful messages - or even change their views. This is very difficult, and studies have found only limited evidence that it’s possible.
The effectiveness of a response is strongly dependent on factors such as the proportion of counterspeakers to hateful speakers (Schieb and Preuss, 2016), whether they were counterspeaking as part of a group (Garland et al., 2022; Friess et al., 2021), the tone used by a counterspeaker (Bartlett and Krasodomski-Jones, 2015; Frenett and Dow, 2015; Ziegel et al., 2018), and even the specific characteristics of the people doing the counterspeech – such as their race or their perceived popularity or authority (Munger, 2017; Seering et al., 2017).
Counterspeaking as a group can be beneficial:
Several recent studies have suggested that counterspeakers working together as a group may improve discourse norms (Garland et al. 2020; Friess et al. 2020) while providing support for their members in a way that contributes to the effort’s sustainability (Buerger 2021). For example, Garland et al (2020, 2022) collected over 9 million tweets originating from two competing online groups: Reconquista Germanica (RG) and Reconquista Internet (RI).
RG is “a highly-organized hate group which aimed to disrupt political discussions and promote the right-wing populist, nationalist party Alternative für Deutschland (AfD)” (Garland et al, 2020,103). RI is a group that formed with the goal of countering RG’s messaging.
The authors created a classifier to identify and code speech as hate speech, counterspeech, or neither. From 135,500 “fully-resolved Twitter conversations” that took place between 2013 and 2018, the authors found that after RI formed, the intensity and proportion of hateful speech apparently decreased. The authors note that “this result suggests that organized counter speech might have helped in balancing polarized and hateful discourse, although causality is difficult to establish given the complex web of online and offline events and process in the broader society throughout that time” (109).
Our own research has shown that working collectively can have benefits for counterspeakers as well. In her research on #jagärhär (the Swedish branch of #iamhere), Buerger (2021) found that counterspeakers reported feeling braver, more supported, and more capable when they responded as a member of a group than when they did by themselves.
Many members stated that before joining #jagärhär, they did not feel comfortable entering the comment sections, describing the comments that they used to encounter as predominantly “toxic,” “aggressive,” and “hateful.” A solitary dissenting voice would draw attention and potentially garner attacks. But with the #jagärhär model, members counterspeak as a group, leaving the individual less exposed within the comment thread. Members said this left them “feeling safer” or “more protected.”
Counterspeech can encourage more counterspeech.
A growing body of research suggests that civil counterspeech can get others to respond in kind, and it can do so without changing anyone’s mind. In a study on spectator perception of newspaper comments, Ziegel and Jost (2020) found that “factual responses to uncivil comments increased observers’ perceptions of a deliberative discussion atmosphere,” which, in turn, increased their willingness to add their own comments to the thread. Two studies (Molina and Jennings, 2018; Han, Brazeal, and Pennington, 2018) found that metacommunication comments (those that address the tone of a comment rather than its content, such as when a user scolds someone else for incivility rather than commenting on the opinions being expressed) engender additional metacommunication comments. Similarly, Miškolci et al. (2018) found that counterspeech comments seem to trigger additional counterspeech from the audience.
The findings from Buerger’s (2021) ethnographic work on the #iamhere network, a large counterspeaking network with over 150,000 members in at least 17 countries, coincided with this finding and provided some initial clues as to why this may be the case. Seeing someone else document their dissent lowers a counterspeaker’s epistemic load – the amount of confidence one must feel that one’s own opinions are right (or someone else’s are wrong) to become willing to enter a relevant discussion.
Counterspeakers expect less online retribution from the people to whom they respond when they are not alone in speaking out.
Research Review
Norms are the unwritten rules of society: they’re what people think other members of their group or society find it acceptable to say, do, and even think. Here we focus mainly on norms about how people speak and write, which are called discourse norms.
Counterspeech can change discourse norms by shifting beliefs about what speech is acceptable. It usually starts small, such as in a single comment thread. As people see and feel discourse changing, some of them will shift their own behavior in response. Over time, this can expand to more and more people.
A study in Slovakia in 2016 and 2017 (Miškolci et al., 2018) tested whether counterspeech could improve discourse norms about Roma people on Facebook. Looking at over 7,500 Facebook comments, they found that when people saw anti-Roma comments followed by counterspeech in a comment thread, they were more likely to post their own comments in support of Roma people. (However, counterspeech did not change the behavior of the people posting the anti-Roma comments).
Other studies have found similar results. Han and Brazeal (2015) found that when study participants were exposed to civil comments, they were more willing to participate in online discussions and more likely to contribute civil comments than participants who had been exposed to uncivil comments.
Conversely, other studies (Cheng et al., 2017; Kwon & Gruzd, 2017) found that exposure to anti-social or negative comments made a person more likely to post an anti-social comment.
A study of #ichbinhier (the German branch of the #iamhere counterspeech network) provides further evidence that internet users take discourse cues from others. Using a dataset of comment threads in which group members engaged over a four month period, researchers found that top-level comments their team deemed “deliberative” were associated with more deliberative second-level comments (Friess et al., 2020).
Changing the mind or the behavior of the original speaker is difficult, but not impossible.
Some studies have looked at whether counterspeech can change the beliefs or behavior of the person it’s responding to, i.e. persuade them to apologize or stop posting harmful messages - or even change their views. This is very difficult, and studies have found only limited evidence that it’s possible.
The effectiveness of a response is strongly dependent on factors such as the proportion of counterspeakers to hateful speakers (Schieb and Preuss, 2016), whether they were counterspeaking as part of a group (Garland et al., 2022; Friess et al., 2021), the tone used by a counterspeaker (Bartlett and Krasodomski-Jones, 2015; Frenett and Dow, 2015; Ziegel et al., 2018), and even the specific characteristics of the people doing the counterspeech – such as their race or their perceived popularity or authority (Munger, 2017; Seering et al., 2017).
Counterspeaking as a group can be beneficial:
Several recent studies have suggested that counterspeakers working together as a group may improve discourse norms (Garland et al. 2020; Friess et al. 2020) while providing support for their members in a way that contributes to the effort’s sustainability (Buerger 2021). For example, Garland et al (2020, 2022) collected over 9 million tweets originating from two competing online groups: Reconquista Germanica (RG) and Reconquista Internet (RI).
RG is “a highly-organized hate group which aimed to disrupt political discussions and promote the right-wing populist, nationalist party Alternative für Deutschland (AfD)” (Garland et al, 2020,103). RI is a group that formed with the goal of countering RG’s messaging.
The authors created a classifier to identify and code speech as hate speech, counterspeech, or neither. From 135,500 “fully-resolved Twitter conversations” that took place between 2013 and 2018, the authors found that after RI formed, the intensity and proportion of hateful speech apparently decreased. The authors note that “this result suggests that organized counter speech might have helped in balancing polarized and hateful discourse, although causality is difficult to establish given the complex web of online and offline events and process in the broader society throughout that time” (109).
Our own research has shown that working collectively can have benefits for counterspeakers as well. In her research on #jagärhär (the Swedish branch of #iamhere), Buerger (2021) found that counterspeakers reported feeling braver, more supported, and more capable when they responded as a member of a group than when they did by themselves.
Many members stated that before joining #jagärhär, they did not feel comfortable entering the comment sections, describing the comments that they used to encounter as predominantly “toxic,” “aggressive,” and “hateful.” A solitary dissenting voice would draw attention and potentially garner attacks. But with the #jagärhär model, members counterspeak as a group, leaving the individual less exposed within the comment thread. Members said this left them “feeling safer” or “more protected.”
Counterspeech can encourage more counterspeech.
A growing body of research suggests that civil counterspeech can get others to respond in kind, and it can do so without changing anyone’s mind. In a study on spectator perception of newspaper comments, Ziegel and Jost (2020) found that “factual responses to uncivil comments increased observers’ perceptions of a deliberative discussion atmosphere,” which, in turn, increased their willingness to add their own comments to the thread. Two studies (Molina and Jennings, 2018; Han, Brazeal, and Pennington, 2018) found that metacommunication comments (those that address the tone of a comment rather than its content, such as when a user scolds someone else for incivility rather than commenting on the opinions being expressed) engender additional metacommunication comments. Similarly, Miškolci et al. (2018) found that counterspeech comments seem to trigger additional counterspeech from the audience.
The findings from Buerger’s (2021) ethnographic work on the #iamhere network, a large counterspeaking network with over 150,000 members in at least 17 countries, coincided with this finding and provided some initial clues as to why this may be the case. Seeing someone else document their dissent lowers a counterspeaker’s epistemic load – the amount of confidence one must feel that one’s own opinions are right (or someone else’s are wrong) to become willing to enter a relevant discussion.
Counterspeakers expect less online retribution from the people to whom they respond when they are not alone in speaking out.
