When it comes to reporting on sexual assault in media, words matter

Someone on your favorite sports team is accused of date rape. A local city council candidate is charged in the sexual assault of a co-worker. Accusations surface about coaches of a gymnastics team.

These stories appear in traditional media every day.

Language can contribute to uncivility

But rather than wait for some resolution or facts of these accusations to play out in news coverage of arrests, investigations or a trial by jury, readers take to social media, spouting their own beliefs of what has happened. They feed the frenzy in arguments and uncivil posts resulting in toxic remarks, insults and even threats, University of California, Davis, researchers observed in a five-year study. They analyzed hundreds of exchanges on Twitter and Reddit, finding that the framing of sexual assaults—and the use of some linguistic features in news reports—can contribute to uncivil social media posts.

In social media posts, and subsequent online reactions to them, people often blame victims and defend accused rapists or attackers, the research suggests. They also engage in biased beliefs about which victims are more worthy of empathy—namely blameless victims assaulted by deviant perpetrators, researchers said.

"Despite the internet's potential to raise awareness around sexual assault and rape culture, people often trivialize rape and sexual assault in online forums," said Hannah Stevens, a UC Davis doctoral student in communication and lead author of the paper published earlier this month.

The paper, "Uncivil Reactions to Sexual Assault Online: Linguistic Features of News Reports Predict Discourse Incivility," was published in the journal Cyberpsychology, Behavior, and Social Networking.

The authors observed that certain features of news reports on sexual assault prompted some people to become defensive about their pre-existing stereotypes and beliefs. For example, in reading both sides of a story that is reported—a typical journalistic device to enable fair coverage—readers may feel uncertain about which person (usually victim or perpetrator) is telling the truth. They take sides or assign blame on social media, researchers said.

Additionally, people may read a story and then reinforce their own stereotypes on social media posts. Examples in the research include media reports of sorority girls raped by a football team with corresponding social media posts voicing that the victims "asked for it," researchers found. Additionally, race of a perpetrator was often mentioned in derogatory ways in social posts.

Researchers used technological tools

Using a computerized coding tool, researchers measured three specific aspects of news reports that they predicted would elicit uncivil reactions online in about 500 posts from 2014-2019. Articles and posts included allegations of sexual abuse that surfaced in stories about the #MeToo movement, teachers in elementary and high schools, and leadership in various churches.

They then correlated these features of news articles with incivility in corresponding user posts using an advanced AI tool designed by Google's Counter-Abuse Technology team. Researchers looked at issues including comment toxicity, insults, profanity and threats.

The linguistic features of news articles measured were:

• Disagreement: Language used to describe opposite sides of a story in an article (usually the victim and the perpetrator) can cause readers to feel uncertain about what really happened. Readers opt to reduce uncertainty by taking sides and assigning blame. Results suggested that linguistic disagreement in articles shared to Twitter caused more profanity and toxicity than on Reddit. Disagreement terms used in stories included "but" or "hasn't," to name a few.
• Negative emotion: Measuring the level of negative emotion language in a news report revealed that individuals have what's called a "just-world bias"—or the belief that good things happen to good people and bad people deserve their adversity. This can then cause some readers to scrutinize the information in negative news more carefully in an effort to defend their belief in a "just world." Results concluded that the relationship between the negative emotion in an article and the toxicity of reactions to the article was also greater on Twitter than on Reddit. Researchers measured negative emotion language to assess levels of anger, anxiety and sadness in articles, looking at the use of such words as "afraid" or "depressed."
• Group power dynamics: Measuring words in articles referencing power relations between groups (words such as "celebrity" or "bully"—typically that were in quotes or other attributions) revealed that discussion of power dynamics, such as the relationship of a coach to a gymnast, can prompt people to become defensive. That would then drive them to defend their core beliefs and pre-existing stereotypes on social media. News discussion of group power dynamics was found to produce greater amounts of toxicity, insults, threats and profanity on Reddit than Twitter, researchers said.

Stevens said the findings are important because they reveal that social media community norms matter—incivility looked different on Reddit than Twitter. "Ultimately, this raises ethical implications for social media community guidelines and moderation policies," she said.

"Our study shows a need to further research how journalists may be able to change the way they write to minimize toxic and uncivil reactions to reports of sexual assault."

Co-authors of the paper are Laramie D. Taylor, department chair and professor, and Irena Acic, doctoral candidate, both in the Department of Communication at UC Davis.