The internet is a polarized place and often a difficult place to have meaningful discussions. With no globally recognized definition for what constitutes hate speech, it can be difficult for social media users, content moderators, and news outlets to decide what is hateful and what is profane because each culture and country has its own context.

A new study by the Center for Media Engagement at the University of Texas, Austin, surveyed people from the United States, the Netherlands, and Portugal to understand how they perceive hate speech and whether or not they can distinguish it from profanity. “The results offer global guidance for social media platforms and news outlets on how to effectively create moderation guidelines that limit confusion about why certain posts and comments are removed while others are allowed,” the researchers note. The study was funded by Facebook and conducted in partnership with researchers from Erasmus University in the Netherlands and NOVA University in Portugal.

Our latest study looks at how people in three countries define hateful speech online. The results offer global guidance for social media platforms and news outlets on how to create effective moderation guidelines.

Read more: https://t.co/zRDrapjMq6 pic.twitter.com/Xe0H2OMWya

— Center for Media Engagement (@EngagingNews) September 10, 2020

“Under European law, hate speech is defined as inciting hatred or perpetuating stereotypes about specific groups based on characteristics such as race, gender, or sexual orientation,” the researchers write. “However, the European approach has not been adopted worldwide, and U.S. law has no official hate speech definition.”

The study found that Americans had a clearer understanding of what they considered hate speech than Europeans did, even though some European countries have legal definitions of hate speech while the United States doesn’t. Americans pointed out the differences in what they considered hateful and what they considered profane. Portuguese participants made the fewest distinctions between the two and considered profane language to be more hateful than Americans did. Dutch participants found comments containing insults, all caps, and swear words to be as profane and more hateful than the comments that attacked specific communities.

Some of the other findings include:

• Americans saw coarse language or swear words as substantially less hateful than derogatory comments about immigrants or comments that incited violence.⁴ They also saw comments with swear words as significantly more profane than comments that incited violence or targeted specific vulnerable groups.⁵
• For participants in both European countries, the distinction between profanity and hatefulness was less clear. In the Netherlands, participants perceived posts with profanity as both more profane⁶ and more hateful⁷ than comments that incited violence or attacked specific vulnerable groups.
• This line between hate speech and profanity was particularly blurry for participants in Portugal. Participants there rated comments as equally profane or hateful, regardless of whether they contained swear words or derogatory comments about specific vulnerable groups.⁸

The Center conducted the study with 304 participants from the three countries. They were randomly shown five posts that looked like Facebook posts that “contained derogatory barbs against immigrants or incitements of violence against these immigrants or five posts that were not targeted at a specific vulnerable group and that contained swear words, name-calling, or words in all capital letters to indicate shouting.” The posts were translated into Dutch or Portuguese depending on the participant. The posts referred to vulnerable communities in each country: “Mexicans” for the U.S. experiment, “refugees” for the Dutch experiment, and “Brazilian migrants” for the Portuguese experiment.

Journalists, particularly those who belong to marginalized communities and/or cover them, are not strangers to either hate speech or profanity. In creating moderation policies, The Center for Media Engagement suggests that news outlets consider the following:

• Tailor content moderation guidelines to the cultures of specific countries: “For example, platforms in Portugal and the Netherlands should highlight their definitions of hate speech more prominently because that distinction is not clear for users in those countries.”
• Inform users about what will and won’t be permitted by clearly defining profanity or hateful speech when they agree to use the platform.
• Tell users what was wrong with the content that was removed.

Read the full report here.