A dangerous new line of argument among activists on both sides threatens free expression.

Recent confrontations across the country reflect a growing tendency to answer speech with violence. In recent months, riots have erupted in response to planned talks by academic Charles Murray at Middlebury College and by conservative media personality Milo Yiannopoulos at Berkeley. In Montana, Congressman-elect Greg Gianforte pleaded guilty to assault for body-slamming a Guardian reporter. Threats of violence have kept right-wing pundit Ann Coulter from speaking at Berkeley. After Princeton professor Keeanga-Yamahtta Taylor delivered the commencement address at Hampshire College, in which she criticized President Trump, death threats led her to call off further speaking engagements.

But now influential voices and factions from across the political spectrum seem increasingly ready to treat ugly speech as equivalent to, or a justifiable provocation for, violence. It’s true that insults or slurs or incitement can lead to confrontation. (Hutu broadcasts during the 1994 Rwanda genocide and Hitler’s entreaties to German national pride, for instance, coaxed violence on a mass scale.) But while there is a continuum of acceptable speech, language and violence should not be confused. The way to preserve our freedom of expression is to insist that speech, no matter how offensive, cannot justify violent reprisal. Otherwise we risk becoming like China, Turkey, Iran and other autocracies, where brutality against journalists and draconian punishments for dissenting ideas are normal.

On the left, theorists have long posited a porous boundary between speech and violence, and the linkage hit the mainstream in Toni Morrison’s 1993 acceptance speech for the Nobel Prize in literature: “Oppressive language does more than represent violence,” she said. “It is violence.” Taken literally, Morrison’s analogy is obviously false. One way to interpret her remark is as symbolic: an invocation aimed to accentuate the potential for offensive speech to wreak grave and lasting harm, which permissive approaches to speech may seem to deny. Morrison was right to call on free speech proponents to own up to the psychological, emotional and social damage that invidious speech can inflict. Studies have documented the ravages that racial slurs can render on self-esteem, the trauma of slut-shaming, and instances of correlation between hate speech and hate crimes. Implicit in the legal protection of free speech, though, is the unavoidable truth that speech can offend, challenge, divide, disturb and hurt. If speech were always innocuous, no one would try to curtail it, nor need to protect it.

More recently, formulations that straightforwardly equate speech and violence have gained currency. After the Yiannopoulos riots in February, a young woman at the University of California at Berkeley said, “His words are violent, or a form of violence.” Writing in a Concordia University publication, a commentator opined that so-called microaggressions are “a very real form of violence.” Anni Liu, a contributing writer for Everyday Feminism magazine, took it a step further. “Just because a perpetrator of racism is clueless (or in denial) about the impact of their words doesn’t mean that their actions were any less violent or that the impact of that violence is changed,” she wrote. Former Vermont governor and presidential candidate Howard Dean described Coulter’s planned Berkeley speech in April as “fighting words” and argued that it did not deserve First Amendment protection.

On the right, Trump’s call at a campaign rally last year to “get ’em out of here,” directing his supporters to go after protesters (albeit with a “don’t hurt ’em”), was deemed “particularly reckless” by a federal judge hearing incitement charges against the president. Trump has also dubbed the news media the “enemy of the American People.” His bellicose attitude seems to have been interpreted in some quarters as open season for violence against reporters. Last fall a Trump supporter wore a T-shirt to a Minnesota rally bearing the slogan “Rope. Tree. Journalist.” This spring, the Sacramento Valley Mirror received threatening calls, and a noose was left in front of the newspaper’s office in downtown Willows, Calif. Late last month, Texas Gov. Greg Abbott joked about shooting journalists after signing a bill cutting the fee for a handgun license. In mid-May, guards at the Federal Communications Commission manhandled a reporter who was trying to question an FCC governor. Gianforte’s attack on the Guardian’s Ben Jacobs last month was the first physical assault by someone elected to Congress since the Civil War era. While not specific or directive enough to constitute incitement, Trump’s verbal attacks on individual journalists and frequent cries of “fake news!” — imbued with the prestige of White House pronouncements — debase the role of the press and give at least tacit leave for acts of hostility toward it.

U.S. law disallows several categories of speech that edge close to violence. For instance, a direct rallying cry to attack can constitute incitement to imminent violence, rendering such speech unlawful. Verbal “true threats” to do physical harm to someone are also prohibited and punishable, as is systematic harassment. Yet a great deal of hateful speech — racial provocations, invective, supremacist sloganeering — is protected by the First Amendment.

Certain forms tug at the bounds of our legal definitions. When an Internet troll publishes an ideological opponent’s street address or phone number, it can be terrifying for the target. Legions of followers may read the disclosure as an invitation to intimidate by phone or in person. Yiannopoulos’s taunting of transgender students by name and his reported plans to expose the identities of undocumented students may fall short of our legal definition of a threat, but they are plenty menacing. As the Supreme Court did in 2003, when it upheld states’ rights to prohibit cross burning done with the intent to intimidate, courts may soon adjudicate whether these new scare tactics cross a legal line. Self-proclaimed free speech advocates who cite constitutional protections to terrorize others risk obscuring the distinction between speech and violence. They blur a crucial boundary on which the protection of free speech depends and, in so doing, invite new limitations on speech.

But the power that speech holds to visit serious harm does not make it, in itself, violent. It is risky even to make this comparison, because it helps give cover to the idea that noxious speech may be answered with brute force. When Republican Rep. Chris Collins of New York blamed James Hodgkinson’s shooting rampage — at a congressional GOP baseball practice — on Democratic “rhetoric,” citing “the tone and the angst and the anger directed at Donald Trump,” he signaled (without evidence of a causal link) that these words had led to violence. Similarly, some protesters at Middlebury and Berkeley argued that an assaultive rampage was warranted to silence speakers whose viewpoints some regarded as intrinsically violent. Deeming speech the equivalent of violence may be intended to call out and deter harm, but it can instead beg violent confrontation by casting it as an inevitable, or at least understandable, response to speech.

A related problem, of course, is that in societies governed by the rule of law, the answer to a perceived threat of violence should not be to brandish a club but rather to call the police. But because the First Amendment bars the government from silencing speech, dubbing it “violence” is a summons for vigilantism. Campus speech battles increasingly seem to invite small groups of well-prepared, violent protesters to use controversial talks as an excuse for a show of ideological force masquerading as the defense of free speech. Their presence, sometimes in the midst of much larger gatherings of non-belligerent protesters, can trigger public safety concerns that shut down speech and deprive others of the right to voice their dissent peaceably, as when universities cancel presentations by controversial speakers.

Rejecting the amalgamation of speech and violence does not mean we must ignore the nefarious potential of speech. We have many methods for fighting dangerous ideas: peaceful protests, counter-speech, fact-checking, official condemnations, and reasonable rules governing the time, place and manner of speech that allow it to be heard while containing its destructive potential. To be sure, these methods are imperfect and can fall short, letting damaging speech go inadequately answered. To paraphrase Winston Churchill’s defense of democracy, though, we protect free speech not because an unfettered marketplace for ideas and opinions is idyllic, but because it’s better than letting governments censor, repress and punish speech at will.

In a democracy, the state is supposed to hold a monopoly on violence. If speech is violence, the state could extend its monopoly to control expression as well. Yet our law treats speech in precisely the opposite way, keeping it open to all and protected from government interference. Both right-wing provocateurs and left-wing protesters have a powerful interest in keeping it that way.

-Suzanne Nossel, PEN America Executive Director