The Risks of Dangerous Speech: Lessons from Rwanda

The land of a thousand hills (source)


“Incitement is a hallmark of genocide, and it may be a prerequisite for it.”Susan Benesch


A few years ago, David Yanagizawa-Drott of Harvard’s Kennedy School of Government examined the effects of radio propaganda on the 1994 Rwandan genocide, which led to the deaths of 0.5 to 1.0 million people (Yanagizawa-Drott, 2014). Rwanda is sometimes called “The Land of a Thousand Hills,” and given the effects of uneven topography on radio transmission, he reasoned that villages with better reception would have been exposed more to incitement to violence against the Tutsi minority. In particular, the Hutu-controlled radio station Radio Télévision Libre des Mille Collines (RTLM) was infamous for dehumanizing the Tutsi by calling them “cockroaches” and calling for their extermination.

Yanagizawa-Drott noted that others had pointed to the role of RTLM and other mass media in fomenting hatred in Rwanda, but no one had attempted to quantify the effect. He calculated the area with radio reception within each village and then correlated it with number of persons prosecuted for violent crimes committed during the genocide in each village, including as a member of a militia (n = 77,000) or as an individual (n = 432,000).

He found that “a one standard deviation increase in radio coverage is associated with a 12–13 percent increase in participation in total violence. The effect is similar for militia violence (13–14 percent) and individual violence (10–11 percent).” Furthermore, there was a “spillover effect,” where the number of people engaged in militia violence increased significantly when neighboring villages had radio coverage. Overall, he estimated that nearly one-third of the violence perpetrated by militias could be attributed to the broadcasts.


The study is important because it demonstrates that hateful propaganda really can have tragic effects. It also shows that the killing did not simply arise out of nowhere, but was due to a coordinated campaign that laid the groundwork to achieve a critical threshold of animosity. In fact, the hateful propaganda from RTLM applied to several of the 10 stages of genocide, including dehumanization, polarization, and preparation. Susan Benesch, Director of the Dangerous Speech Project, noted how integral this groundwork was in paving the way for atrocity, and how all people are potentially vulnerable to it:

“I’ve learned a few specific things about humanity. First, people do not hate spontaneously. No one is born hating, or wanting to see or do violence. Also, no particular group — religious, ethnic, cultural or national — has a monopoly on dangerous speech. It isn’t that there is something wrong with one group or another, as some have alleged. All people are capable of producing and being influenced by dangerous speech.”

For these reasons, people find dehumanizing language —particularly in mass media— to be shocking and a red flag. For example, in 2015 the British columnist Katie Hopkins referred to migrants and refugees who attempted to enter Britain as “cockroaches” and “a plague of feral humans.” Hopkins was widely condemned, including by the UN High Commissioner for Human Rights. Other historical and contemporary examples of dehumanizing language include portraying groups of people as subhuman (vermin, insects, rats, apes, snakes, mongrels, weeds, beasts, animals), equivalent to a disease (infection, viruses, cancer), who engage in non-human behaviors (infesting, breeding, infiltrating, taking root).

Benesch added that five factors come into play with regard to the effects of dangerous speech, including: (1) the influence of the speaker, (2) the susceptibility of the audience and their receptivity to messages of hostility, (3) historical context, including whether two groups have a past of conflict or animosity, (4) the medium through which the message reaches the intended audience, such as a radio or television station, or the Internet, and (5) the actual content of the speech, including dehumanization of other groups, suggesting that they pose an existential threat, and overt or subtle calls to violence.


The Role of Dehumanization

These five factors illustrate that, like all human behavior, incitement to mass violence is complex. However, one reason to focus on dehumanization is that these metaphors (insects, disease, vermin, etc.) can conjure feelings of disgust. And, as the biologist Robert Sapolsky pointed out, the same area of our brain – the insula – becomes active when we experience disgust, whether we’re thinking about rotten meat or someone’s reprehensible behavior. In his words: “When we evolved the capacity to be disgusted by moral failures, we didn’t evolve a new brain region to handle it. Instead, the insula expanded its portfolio.” Disgust is a powerful emotion that induces people to either walk away from the source, or remove it. That works well when it comes to rotten food, but removing people is another matter entirely, and one that leads inexorably to tragedy.

Centuries ago, it was a question in scientific circles whether people on different continents were the same species (Smedley 2012). Some suggested they were merely varieties of humans while others guessed they originated separately, de novo. This has long since been settled: all humans constitute a single species – Homo sapiens. I would even go a step further and point out that all humans share common ancestry and can be reasonably thought of as cousins. Scientifically, we are also categorized as a type of primate, a type of mammal, and a type of animal, etc.

The danger, of course, is what it does to our mentality when we consider some groups of people to be less than fully human and therefore less worthy of moral consideration. In their 2017 article, “Darker Demons of Our Nature: The Need to (Re)Focus Attention on Blatant Forms of Dehumanization,” Nour Kteily and Emile Bruneau laid out the connection between dehumanization and feelings of hostility toward others:

“Our findings also suggest that blatant dehumanization contributes uniquely to hostile intergroup attitudes and behavior. Across a range of studies, we have regressed aggressive outcomes on blatant dehumanization and subtle forms of dehumanization, assessed both implicitly and explicitly. Consistently, blatant dehumanization was more strongly associated than subtle/implicit dehumanization with attitudes such as supporting the torture of Muslims and Arabs, supporting housing and educational discrimination against the Roma, and supporting collective punishment of Palestinians.” (p. 489-90)

It’s important to note that some, like the psychologist Paul Bloom, have questioned the role of dehumanization when it comes to inflicting cruelty on other people. While it may be easier for people to inflict harm on someone who they think is sub-human (and therefore undeserving of being treated as full moral agent), Bloom write that “the truth may be harder to accept: that our best and our worst tendencies arise precisely from seeing others as human.” To Bloom and others, in some circumstances it isn’t that people are blind to others’ fully developed moral agency or human intelligence. Instead, it’s that we that we can be led to believe that other people are using their full humanity in a way that is immoral or threatening to ourselves. Therefore, the reasoning goes, it’s not that others are stupid and amoral, but intelligent and immoral, and therefore deserving of punishment.   

I’m not sure which weighs more heavily – the dehumanization concept or the idea of moral certainty. Perhaps circumstances matter. Both lead to similar outcomes. But I think the role of dangerous speech would apply to both, the notion that it takes effort to mold public opinion in a way that can lead to atrocity. The Rwanda example suggests as much. In fact, before the genocide the UN Commander, Romeo Dallaire, implored the international community to jam RTLM to cut off the source of resentment, but to no avail (Yanagizawa-Drott, 2014.). The reasons that were given included the expense of jamming the signals, and not wanting to violate Rwanda’s sovereignty or interfere with free speech or a free press. Another approach was suggested by Susan Benesch of the Dangerous Speech Project. Instead of curtailing free speech, she advocates trying to delegitimizing dangerous speech through humor or shame, instead of suppressing it. This protects the ideal of freedom of speech, while also attempting to defuse it. The risk, of course is that it is impossible to corral all dangerous speech.

At a minimum, it seems a good idea to confront speech that is hateful, dangerous, and prone to incitement to minimize the risk for violence. I started thinking about this essay some time ago, but by coincidence a report came out today that Andrew Veprek of the State Department argued that leaders should not be obliged to condemn racism and xenophobia on the spurious grounds that (a) leaders should influence people’s behaviors and actions, not thoughts and (b) criticizing intolerance runs the risk of being intolerant in itself. I would agree with Veprek that there is reason to be wary of powerful leaders trying to control people’s thoughts, but I also think the Rwanda example demonstrates pretty clearly that dangerous speech (and dangerous thoughts) are linked with dangerous actions.

I cannot remember where I read this, but there was a quote that I came across in graduate school that has stuck with me for a long time. I don’t mean to take credit for it, but it went something like this: “More people have died in the name of extreme intolerance than in the name of extreme tolerance.” All else being equal, if one is to err, then err on the side of tolerance. A component of that may require being intolerant to intolerance. That may not be the perfect approach, but I don’t think there is one. All paths have tradeoffs and entail a balance or clash of values. It makes sense to me to try to balance those values and choose a path that reduces the risk for atrocity. 



Kteily NS, Bruneau E. 2017. Darker demons of our nature: The need to (re) focus attention on blatant forms of dehumanization. Current Directions in Psychological Science. 26(6):487-94. Link

Smedley A. 2012. Race in North America: Origin and Evolution of a Worldview. Westview Press. Link

Yanagizawa-Drott D. 2014. Propaganda and conflict: evidence from the Rwandan genocide.” The Quarterly Journal of Economics (August 21). Link

4 thoughts on “The Risks of Dangerous Speech: Lessons from Rwanda

  1. Another interesting angle is the narcissism of small differences. This is exacerbated by socially constructed identities.

    Such was the case with the Hutus and Tutsis, a division that resulted from colonialism. But in reality the two groups shared the same language, religion and culture — both groups primarily originating from the Bantu.

    The ability to invent divisions is surprisingly easy, as the teacher’s experiment with categorizing students by eye color. Numerous other studies have come to the same conclusion. And the powerful or those seeking power have often used this knowledge to great effect.

  2. What do you make of Susan Benesch’s notion that humor and shame are effective means of defusing hate speech? Are they? I’m quite curious about this because I’ve been looking for a means to accomplish the task for some time.

    • Hi Paul, it’s a good question. I’m not sure, to be honest. I can imagine them working in some scenarios, but in others not so well. I think it’s probably complicated. Sorry I don’t have a better answer. Like you, I’d love to know the antidote to this.

      • Hopefully someone will research it someday and we’ll an answer. In the meantime, I’m going to give humor and shame a try in my narrow circle. See what the effect is. .

        Great blog by the way.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.