Concept
The Categorical Approach
Regulation of unprotected expression
Unprotected Categories of Speech
The Categorical Approach
The First Amendment’s text is addressed to Congress, and it probably was not meant to reach the states, but since 1925 the Court has applied its limit to the states through the Fourteenth Amendment’s Due Process Clause.1 Read alone, the command that “Congress shall make no law abridging the freedom of speech” contains no exceptions, and a first-time reader might expect every restriction on speech to be forbidden.2
That reading collapses on contact with ordinary law. Perjury statutes and laws against quid pro quo sexual harassment (“sleep with me or you are fired”) both punish speech, and no one would deny that such statutes are imperative; a single example like that is enough to refute an absolutist approach.3 Justice Holmes made the same point in Schenck v. United States: “[t]he most stringent protection of free speech would not protect a man in falsely shouting fire in a theatre and causing a panic.”4
The text itself points the way out. It prohibits laws “abridging the freedom of speech,” not all laws restricting communication, so every case begins by asking whether the expression is something that phrase protects at all.5 Drawing that line cannot be avoided; the question is how to draw it. Balancing the competing interests case by case has an Achilles heel: it offers no before-the-fact, bright-line test for identifying which interest will prevail.6 The courts sometimes apply strict scrutiny, which presumes the regulation invalid and requires the government to show narrowly tailored or necessary means to a compelling interest, and sometimes use a form of balancing that is in essence rationality review.7 The categorical approach answers that problem by deciding the question once for a whole class of speech and attaching a rule to the class.
Expression carried out through conduct is a separate problem. It is subject to greater regulation than pure speech, and a legislature may, for example, proscribe picketing that unreasonably interferes with ingress or egress to public buildings.8 This chapter sets that problem aside and concentrates on restrictions aimed at what speech says.
Laws that restrict speech because of what it says are the laws that draw strict scrutiny, and the Court has paid close attention to the distinction between content-based and content-neutral laws.9 Exceptions to that presumption must therefore stay narrow, because broad or loosely defined exceptions would swallow it. The analysis accordingly runs in two steps, which the rest of the chapter applies again and again:
The second step exists because a category label shows only that some regulation is permissible, not that every regulation is. A legislature could invoke a legitimate category as cover for suppressing one side of a debate, and the Court has held that government may not discriminate within a category so as to favor some viewpoints or ideas over others.10
The sections that follow take up incitement, fighting words, true threats, obscenity, and child pornography in turn, with offensive speech and hate speech treated as the cases that show where the categories stop. Broadcast indecency comes last, and the Court’s stated reasons for regulating it concern the medium. The broadcast media have established a uniquely pervasive presence in American life, and patently offensive material on the airwaves confronts the citizen in the privacy of the home, where the individual’s right to be let alone plainly outweighs the First Amendment rights of an intruder.11 Because the broadcast audience is constantly tuning in and out, prior warnings cannot completely protect a listener, and telling someone to turn off the radio is like telling an assault victim to run away after the first blow.12
Low-Value Speech Categories
The Court gave the categorical approach its founding statement in Chaplinsky v. New Hampshire. Arguing with a city marshal, Chaplinsky called him a “God damned racketeer” and “a damned Fascist,” and he was convicted under a state law forbidding any offensive, derisive, or annoying word addressed to another person lawfully in a public place.13 The Court upheld the conviction. It wrote that the right of free speech is not absolute at all times and under all circumstances, and that there are “certain well-defined and narrowly limited classes of speech, the prevention and punishment of which have never been thought to raise any Constitutional problem.”14 Those classes include the lewd and obscene, the profane, the libelous, and the insulting or “fighting” words, which by their very utterance inflict injury or tend to incite an immediate breach of the peace.15
The Court’s reason has two parts, and each deserves separate attention. On the value side, utterances in these classes are “no essential part of any exposition of ideas,” and whatever benefit they offer “as a step to truth” is “clearly outweighed by the social interest in order and morality.”16 Resort to epithets or personal abuse, the Court added, is not in any proper sense communication of information or opinion safeguarded by the Constitution.17 On the harm side, some classes work as instruments of injury: child pornography, commercial fraud, calculatedly false libel, illegal incitement to violence, and true threats such as cross burning with intent to intimidate.18 The weighing happens once, for the whole class. Within a recognized class, the evil to be restricted so overwhelmingly outweighs any expressive interest that no process of case-by-case adjudication is required.19
The Court treats these classes as falling outside the First Amendment, so that they normally receive no constitutional protection.20 The working roster is longer than Chaplinsky’s, and each item except the last receives its own section later in the chapter:
- Illegal incitement to violence.21
- Fighting words, meaning insulting words that by their very utterance inflict injury or tend to incite an immediate breach of the peace.22
- True threats, including cross burning done with intent to intimidate.23
- Obscenity and child pornography, neither of which the First Amendment protects.24
- Calculatedly false libel and commercial fraud, which this chapter names but does not treat separately.25
Defamation shows how a category can survive while its rules change. The Court could not accept the generality of the older view that libel stands wholly outside the Amendment. New York Times Co. v. Sullivan required public officials suing for defamation to prove that the publication was not only false but knowingly false or published with reckless disregard for the truth, and later public figures in semiofficial or influential positions were held to the same showing.26 Other early decisions show the same drift toward protection. Beauharnais, a hate-speech decision, came before New York Times v. Sullivan extended substantial protection to libelous publications, and before Brandenburg and Cohen demanded greater showings of imminent danger from incitement than Justice Frankfurter required.27 A place on the list therefore marks where the analysis starts, not where it ends.
Because the classes rest on a judgment about value and harm, a legislature cannot move speech outside the First Amendment by attaching a label to it. If a label controlled, a legislature could call any disfavored message obscene or libelous and turn the exception into a general power to censor, and the Court has rejected the idea that the government’s own cost-benefit judgment can identify an unprotected category.28 Whether a communication actually belongs to a class remains a question for the courts applying the First Amendment.
The word unprotected carries two meanings in the case law, and this chapter uses the first. Sometimes it describes a whole class, such as fighting words or obscenity, that falls entirely outside the First Amendment umbrella. At other times it describes speech that is ordinarily protected but punishable on the facts of a particular case.29 Even the first meaning is looser than the old rhetoric suggested. In R.A.V. v. City of St. Paul, the Court held that a government that may punish a class of speech still may not select among its members to favor some viewpoints over others, a limit taken up in the third subsection below.30
Fit with a defined class puts a message inside the exception, and the audience’s dislike of the message does not. Chaplinsky’s epithets, hurled face to face at a public officer, fit the class of insulting or “fighting” words.31 Offense alone fits no class. When Cohen was convicted of disturbing the peace by “offensive conduct” for wearing a jacket bearing the words “Fuck the Draft” in a courthouse, Justice Harlan rejected the idea that the Court could excise particularly scurrilous expression from public discourse, because “one man’s vulgarity is another’s lyric.”32 The same reasoning governs a legislature’s view that a message is demeaning. In Matal v. Tam, Congress had barred federal registration of trademarks that disparage persons, institutions, beliefs, or national symbols, and the Court held the clause an unconstitutional viewpoint-based restriction on private speech. Disparagement is not a recognized class, and calling the message worthless did not make it one.
No New Categories
The list is, for practical purposes, closed, and United States v. Stevens is the reason. Robert Stevens ran a business called “Dogs of Velvet and Steel” and sold videos of pit bulls fighting and attacking other animals.34 He was indicted under §48, which reached depictions of animal cruelty, and he moved to dismiss on the ground that the statute was facially invalid. The District Court denied the motion, holding that the depictions, like obscenity or child pornography, are categorically unprotected, and a jury convicted him on all counts.35 The government defended that holding by urging that a depiction’s value be weighed against its societal costs. The Court rejected the proposal and called it, as a free-floating test for First Amendment coverage, “startling and dangerous.”36
The weighing has already happened. The First Amendment itself reflects a judgment by the American people that the benefits of its restrictions on the Government outweigh the costs, so a court that re-weighs the question for a new class of speech is revising a decision the Constitution has made.37 A cost-benefit test for coverage also lets the regulator do the weighing, which helps explain why the Court called the government’s test highly manipulable.38 History therefore replaces value as the question. The Court explained that when it has identified categories as fully outside the First Amendment, it has not done so on a simple cost-benefit analysis, and that Ferber, for example, did not rest on a balance of competing interests alone.39 Chaplinsky’s language about slight value and social interest sounds like balancing, but Stevens reads that line of cases as resting on previously recognized, long-established categories.40
Brown v. Entertainment Merchants Association followed Stevens when a state tried to treat violent video games sold to minors as unprotected speech. A legislature’s conclusion that speech is too harmful to be tolerated does not by itself add the speech to the list, because the Court rejects the government’s highly manipulable balancing test as a means of identifying new categories.41
Falsehood tempts students to find a category where none exists, and United States v. Alvarez closes that door. The Stolen Valor Act made it a crime to claim falsely to have received a military decoration, and the government argued that knowing falsehoods have no First Amendment value. The Court did not accept a general exception for false statements. The false-speech classes it recognizes, calculatedly false libel and commercial fraud, are on the list because of the harms they cause.42 In those classes the lie travels with an injury the law recognizes, such as a ruined reputation or money obtained by deceit, and that injury justifies punishment. A statute that punishes a false statement and nothing more has only the falsehood to rest on.
Both statements are false, so falsity cannot separate the two scenes. The harm separates them, and that is the line the recognized false-speech classes draw.
Child pornography fits this method and does not depart from it. Ferber classified child pornography as a category outside the First Amendment, but the decision did not rest on a balance of competing interests alone. The market for child pornography was intrinsically related to the underlying abuse and was an integral part of producing the material, an activity illegal throughout the Nation, so Ferber grounded its analysis in a previously recognized, long-established category of unprotected speech.43 What justifies the class is the harm to actual children in making the material, and the justification is not a freestanding judgment that the content is worth little.
The Court has not sealed the door entirely. It acknowledged that some categories may have been historically unprotected without yet being specifically identified in its cases, and it said it need not foreclose recognizing them in order to reject the government’s balancing test as the way of finding them.44 The only route to a new category therefore runs through history. A fact pattern in which a legislature declares a type of speech worthless, dangerous, or offensive is testing this rule, and the analysis asks whether the speech fits an existing class or a documented history of restriction. If it fits neither, the speech is protected, and the content-based law faces strict scrutiny.
Limits Within a Category
Placing speech inside a recognized category settles the first step and leaves a second question open: what may the government still not do to speech that sits in the category? R.A.V. v. City of St. Paul supplied the answer. The power to proscribe speech on the basis of one content element, such as obscenity, does not entail the power to proscribe it on the basis of other content elements.46 The Court’s own illustration states the idea most plainly: the government may proscribe libel, but it may not make the further content discrimination of proscribing only libel critical of the government.47
The reason is the one that makes content-based laws suspect in the first place. Content discrimination raises the specter that the government may effectively drive certain ideas or viewpoints from the marketplace, and that danger does not vanish because the speech belongs to a class the government may otherwise punish.48 Each class rests on a particular harm: fighting words on the risk of immediate violence, threats on fear and the disruption it causes, obscenity on its prurient appeal. A law that punishes only some members of the class, chosen for a reason unrelated to that harm, shows that the lawmaker is aiming at a message.
Two terms keep the limit from being misread. A law is underinclusive when it reaches some instances of a problem and leaves others alone, and it discriminates on content when the line it draws depends on what the speech says. The concurring opinions in R.A.V. described the majority as announcing a new rule that a prohibition on proscribable speech cannot be underinclusive, a First Amendment absolutism under which government must proscribe all speech or none within a category. The majority answered that the First Amendment imposes not an underinclusiveness limitation but a content discrimination limitation, so a state may, for instance, prohibit obscenity only in certain media or markets.49 A concurring opinion pressed a sharper objection: it is inconsistent to say that government may proscribe a whole category because its content is evil but may not treat a subset differently, since the content of the subset is by definition worthless.50
Content discrimination takes several forms. A restriction may target subjects or topics, as in a city ordinance that regulates signs by the information they convey; viewpoints, as in a law prohibiting criticism of the war effort; or the speaker’s identity, as in a law barring labor organizers or religious groups from a public auditorium.51 Viewpoint discrimination is the form that most directly lets the government pick the winner of a public debate.
Under the R.A.V. majority’s reasoning, a general prohibition on fighting words would be constitutional, and it could be enforced against hate speech so long as enforcement was not selective.52
The limit has a built-in exception that keeps it from swallowing ordinary lawmaking. When the basis for a distinction within a class consists entirely of the very reason the class is proscribable, no significant danger of idea or viewpoint discrimination exists, because a reason neutral enough to exclude the whole class from protection is neutral enough to sort within it.53 A state may therefore prohibit only the obscenity that is most patently offensive in its prurience, but it may not prohibit only the obscenity that carries offensive political messages.54 The federal government may likewise single out threats of violence against the President, because the reasons threats fall outside the First Amendment (protecting individuals from the fear of violence, from the disruption that fear engenders, and from the possibility that the threatened violence will occur) have special force when applied to the President.55 A line drawn by the policy the threatener mentions would have no such connection to why threats are punishable. When a law does discriminate within a class and no neutral basis explains it, the First Amendment comes back into play, and the measure falls unless the discrimination is necessary to achieve a compelling governmental interest.56 That consequence is why “unprotected” speech is not a constitutional zero.57
The second limit concerns drafting and applies even when no message is singled out. A statute aimed at an unprotected class must reach that class and stop there. In Brandenburg v. Ohio, the Court condemned a statute that failed to distinguish advocating the moral propriety or even moral necessity of force from preparing a group for violent action and steeling it to such action, because such a statute “sweeps within its condemnation speech which our Constitution has immunized from governmental control.”58 A statute that sweeps in protected speech also chills speakers who cannot tell which side of the line they occupy.
Two later sections build on these limits. The R.A.V. discussion returns to the ordinance the Court struck down, and the true-threats section takes up cross burning done with intent to intimidate, which the Court has treated as a true threat.59 The fighting-words section returns to drafting and shows why broadly worded statutes so often fail.