Them’s Fightin’ Words—Maybe: Testing the Application and Boundaries of the “Fighting Words” Doctrine Using a Randomized Survey Experiment

Advanced note to readers: The survey experiment in this Article depicts acts of verbal violence, including the use of raceand gender-based epithets. Some of the relevant caselaw also includes offensive or harmful language. As one of only a handful of exceptions to the First Amendment’s bar on laws proscribing speech, “fighting words” are defined as “those personally abusive epithets which, when addressed to the ordinary citizen, are, as a matter of common knowledge, inherently likely to provoke violent reaction.”1 It is unclear, however, how this relatively old doctrine applies to contemporary speech. Additionally, the standard is itself potentially problematic in application.

Strictly interpreted, the standard appears to make words regulable and metes out punishment based on the perceived reaction of the victim (or onlookers) rather than the behavior of the wrongdoer. Consequently, we join others in arguing that the doctrine may fail to protect the interests of women, racial/ethnic minorities, or other groups when underlying assumptions based on victims’ characteristics form the basis of a factfinder’s determination as to whether a violent response is “likely.” To explore these concerns, we conducted a qualitative survey of thirty recent cases that explored the fighting words standard in state courts and identified any factors—other than the content of the “fighting words” themselves—that courts consider when making a fighting words determination. Then, to gauge how the characteristics of the individuals involved in an altercation might impact a fighting words analysis, we conducted a randomized survey experiment incorporating these factors into vignettes to explore when the “common knowledge” of the “ordinary citizen”—or at least the consensus of the 705 survey respondents—suggests something is likely to provoke a violent reaction. Our results indicate that the racial and gender composition of the parties involved in an altercation do sometimes have statistically significant effects on whether an altercation is perceived as likely to provoke violence, at least within the limited confines of the survey.

The survey results also indicate—although not always in a statistically significant manner—that there are observable differences in whether a violent reaction is predicted based on the nature of the epithet used, with racial epithets being perceived as particularly impactful, especially against racial minorities.