Suzanne Nossel is executive director of the PEN America.

A fire-breather in Aspen Hill, Md. (The Washington Post)

Hateful speech is a real thing. But “hate speech” is an incoherent concept that confuses more than it clarifies. In recent weeks, pundits, critics and pols have found ways to label all of the following “hate speech”: anti-vaccination trolls on a California state senator’s Facebook page; racist, homophobic and anti-Semitic tweets in France that are the subject of a lawsuit aiming to get Twitter and YouTube to take down the offensive materials; a reference to “Japs” by U.S. Rep. Peter T. King (R-N.Y.) during a talk-show interview; and Republican anti-Muslim and anti-Mexican sentiments called out by former first daughter Chelsea Clinton. A viral hashtag this week, #IStandWithHateSpeech, held that mainstream media companies were censoring readers’ views.

We should forever fight the phenomenon of hateful speech in its many forms, but it’s time to retire the term “hate speech.” Its meaning is inexact, elastic and often misunderstood. If we want to combat the harms of distasteful, denigrating and dangerous ideas, we need precise tools and precise terms.

The first problem with the term “hate speech” is that we use it to denote three distinct categories: speech that is unlawful almost everywhere, such as direct threats and calls to immediate acts of violence; speech that is protected under international law and in most jurisdictions, such as garden-variety insults directed at a particular gender, race or religion; and speech that is illegal in some places but not others, such as Holocaust denial.

These diverse phenomena cannot all be lumped together and, collectively, either permitted or prohibited. It does not make sense to have a single approach to Donald Trump’s proposals from the presidential campaign podium to discriminate against Muslims, pro-Trump messages written in chalk on university walkways, and the trolling of feminists on social media or anti-immigrant comments by Dutch politicians.

We also wrongly conflate hate speech and hate crimes, a critical distinction under the First Amendment that risks being lost as we use the term “hate speech” more loosely. Earlier this month, Danny Bakewell Sr., the owner of the Los Angeles Sentinel, a reputable African American-oriented newspaper in Los Angeles, referred to a drawing of a cross burning and Ku Klux Klan member on a card passed to a Los Angeles City Council member during a hearing as “unquestionably a hate crime.” Bakewell said: “If you paint a swastika on a person’s house, that is a hate crime. If you paint a picture of a man hanging from a tree on a house, that is a hate crime. It is no less offensive or hateful if you draw it on a piece of paper.”

But hate crimes are defined either as speech tied to a separate criminal act (an assault or vandalism, for example) or speech that itself crosses the line into action. Two acts that Bakewell conflates – trespassing in order to vandalize someone’s house and drawing such an image on a piece of paper – are distinct under the law. For speech alone to be a crime, it must entail threats, such as vowing to blow up an airplane you’re boarding or more recent phenomena such as cyberstalking and cyberbullying. While only a small segment of what we call hate speech is actually criminal in the United States, the Bureau of Justice Statistics in 2012 (the most recent year for which its study is available) tallied nearly 300,000 real hate crimes involving violence or property damage: murders, assaults, arsons, vandalism and other acts motivated by bias against an identified group. We don’t need the term “hate speech” to describe these crimes, and by calling them hate speech, we risk implying that lesser forms of unsavory speech (such as bigotry without violent action or a highly offensive doodle) verge into criminality, as well. In May, a Scottish man was arrested for posting a video showing images of Adolf Hitler juxtaposed with footage of his dog purportedly making a Nazi salute. The man was charged with improper use of electronic communication under Scotland’s 2003 Communications Act, a worrisome misapplication of the law that could make any form of hate speech a de facto hate crime just as long as it is transmitted online.

Even if hate speech doesn’t amount to a hate crime, Attorney General Loretta E. Lynch and others have hypothesized that one may lead to the other. Law enforcement authorities are increasingly girding against that possibility: At an event shortly after last fall’s terror attack in San Bernardino, Calif., Lynch referred to anti-Muslim “rhetoric” that “edges toward violence,” describing the uptick in such sentiments as her “greatest fear.” Others have linked a violent attack on a Colorado abortion clinic to former presidential candidate Carly Fiorina’s claim that Planned Parenthood was “harvesting baby parts.” If hate speech can beget hate crimes, though, so can counter-speech. Louisiana has just passed a law that makes attacks targeting police “hate crimes.” The bill’s author has cited attacks on police on social media as part of the motivation for recognizing this new category of crime. While there is plenty of space in a democracy for reminders that speech has consequences, these proposals assume a slippery slope between speech and violence that will make it increasingly difficult to protect unpopular forms of speech.

U.S. law has for decades recognized a category of speech – incitement to imminent violence – that is unlawful because of its potential to catalyze crime. But calls and steps to prohibit speech that is merely hateful yet still nonviolent broaden this definition considerably in ways that the United States has until now consistently rejected. In 1992, when the Senate ratified the International Covenant on Civil and Political Rights, the world’s premier human rights treaty, it did so with a reservation directed at Article 20 of the covenant which bans “national, racial or religious hatred that constitutes incitement to discrimination, hostility or violence.” The Senate concluded that this broad formulation of prohibited speech violated the First Amendment. Judging which speech is merely obnoxious or offensive and which crosses over to stoke hatred or discrimination has not been a line legislators or courts have wanted to draw. Holocaust denial (illegal in more than a dozen European countries) laws and restrictions that proscribe the “defamation” religion are unconstitutional here. The very narrow approach to criminalizing offensive speech is one reason the United States can fairly claim to have the world’s most protective legal standard for free expression in the world.

In parts of the world where free speech norms are not robust, the umbrella term “hate speech” increasingly criminalizes expression. South Africa, for example, is now debating a bill that would make all hate speech a statutory criminal offense–in part a reaction to a January Facebook post describing black people as monkeys. Japan’s Diet has just passed its first-ever legal curb on hate speech with a definition to encompass not just insults and incendiary language directed toward foreigners, but also “egregious insults.” A debate in Quebec over an expansive proposed hate speech law prompted an insurance company to cancel a performance by two comedians for fear of triggering an investigation. In Russia, a regional lawmaker faces up to two years in prison on a “hate speech” charge for alleging that President Vladimir Putin had committed crimes against his people. “Hate speech” is increasingly being used as a catch-all to reshape free expression rights in the direction of expanded government authority to curb and punish speech.

Another problem with the concept of hate speech is that it does not distinguish between hateful intent and hateful effect. Cartoons, crude pictures, political humor and dashed-off social media posts can all have the effect of conveying hatred, regardless of the motives behind them. When the New Yorker published a satirical cover depicting the Obamas as terrorists during the 2008 campaign, the Obama campaign took offense, and some online commentators dubbed the image “hate speech.” Not all speech that offends or degrades is willful hate speech. It may be insensitive, ill-thought out and even incendiary to certain people, but it may mean something quite different to its author or its intended audience, particularly in a world where images transverse borders and cultural contexts. Concerns over acts of expression being taken out of context, misunderstood and construed as hate speech – or even a hate crime – are having a demonstrable chilling effect. Charlie Hebdo quietly stopped publishing images of Mohammed after the attack on its offices last year. In 2011, the Smithsonian Institution withdrew a video depicting ants crawling on a crucifix after being accused of sponsoring “anti-Christian hate speech.” All this comes amid a broader debate on how to protect people from physical and psychological harm while still protecting controversial speech. The American Association of University Professors has issued a draft report detailing the chilling effort that federal legal prohibitions on sex discrimination are having on speech on college campuses, noting a “tendency to treat academic discussion of sex and sexuality as contributing to a hostile environment.”

In recent years, important new movements have re-energized the drive to eradicate xenophobia, sexism, racism, homophobia and religious discrimination. In many circles, including on campuses, there is a new, more acute awareness of the ways once-tolerated remarks, including casual expressions and off-color humor, can cause lasting harm. We now have trigger warnings, gender-neutral pronouns and the concept of microaggressions. As students graduate, some of these new norms will migrate into workplaces and communities. In this context, defending provocative or even offensive speech — once proudly undertaken by civil libertarians—has become a more complex task. When ugly but legally defensible expression is dubbed “hate speech,” standing up for it can be misconstrued as sympathizing with offensive views. When Yale faculty member and administrator Erika Christakis sent a memo to Yale students defending their right to wear Halloween costumes that might be considered offensive, her message sparked a vociferous outcry from students who argued that she was undercutting the position and even the safety of marginalized students. She was accused not just of being wrong in rejecting the university’s caution to avoid offensive costumes, but of being racist herself. She resigned her teaching position at the university. This kind of precedent casts a chill not only on provocative speech but also on free speech’s would-be defenders.

This is not to deny the urgency for new and more effective measures to address hateful speech. Social media has made it easier for vicious speech and bullying to demoralize and even exile their targets from online debates. Whole groups such as women in the video gaming industry have been deterred from using social media. Social media can be a forum for untrammeled racist and otherwise offensive conduct, but also a vehicle for spreading tolerant attitudes. The major social platforms recently introduced a compact with the European Commission to expunge hate speech. While some insisted that the target is confined to “illegal hate speech,” others referenced “hate speech” more broadly; given the increasing fluidity of the distinction the confusion is understandable. These problems demand an urgent response, including more determined measures by online platforms to expose trolls, better tools to enable users to block content they don’t wish to see, more vociferous and visible vehicles for counter-speech, and more aggressive patrolling to remove speech that meets the legal definition of a threat or incitement to imminent violence. But punishing all hateful speech – as South Africa and Japan are considering – will neither prevent nor undo the harm it causes.

Some experts including Harvard’s Susan Benesch have proposed renaming certain kinds of “hate speech” as “dangerous speech.” This would include, say, murderous incitement broadcast in Rwanda before the 1994 genocide began. “Denigrating speech” could denote expression that is intended to belittle or besmirch, which can and should be rebutted or condemned but not banned. Speech that is not intentionally denigrating but nonetheless causes offense could be called “offending speech,” which implies no motive. Existing legal categories –“incitement” aims to stoke imminent violence, and “defamation” deals with the spread of falsehoods intended to inflict harm on specific individuals – are also useful.

There is no perfect paradigm, and some speech will inevitably defy categorization. But in a shrinking world where it is ever more important both to be able to speak freely and to appreciate the subjective impact of speech on others, the concept of hate speech is too malleable to be of help.