Content Discrimination and the First Amendment (Including the “Secondary Effects” Doctrine)

A post on Humanitarian Law Project and content discrimination reminded me that I’d never blogged my summary of the content-based/content-neutral speech restrictions under current First Amendment law. So here it is, extracted from my First Amendment textbook. (Note that all this assumes that the government is regulating speech acting as sovereign — not as proprietor, educator, employer, and the like — and that the restriction isn’t one of those legitimized by the existing First Amendment exceptions, such as the exceptions for obscenity, certain kinds of libel, threats, and so on.)

For most restrictions, there’s little controversy about whether they are content-based or content-neutral. If the restriction treats speech differently based on the viewpoint or subject matter of the speech, on the words the speech contains, or on the facts it conveys, the restriction is based on the content (and the communicative impact) of speech. Classic examples: Bans on advocacy of illegal conduct (Brandenburg v. Ohio), on sexually explicit speech (Miller v. California), on all residential picketing except labor picketing (Carey v. Brown), on profanity (Cohen v. California), or on the publication of the names of rape victims (Florida Star v. B.J.F.). Some of these restrictions may be viewpoint-neutral, but they’re all content-based.

If the restriction focuses on the noncommunicative aspects of the speech, and treats speech the same regardless of what it says, then it’s treated as content-neutral. Classic examples: Bans on all loudspeakers (Kovacs v. Cooper), on all leafletting (Schneider v. State), or on all sleeping in public parks (Clark v. CCNV).

Sometimes, though, the matter is indeed controversial, as in Texas v. Johnson or in the cases in this section, partly because the Court has enunciated at least four subtly different tests for deciding which restrictions are content-based:

A. Whether the “regulation is based on the content of speech,” as opposed to being “applicable to all speech irrespective of content.” Consolidated Edison Co. v. Pub. Serv. Comm’n.

B. Whether the law’s application to speech “depend[s] on the likely communicative impact” of the speech. Johnson.

C. Whether the legislature’s “predominate concerns” are “with the content of” the speech as opposed to “with the secondary effects of” the speech, City of Renton v. Playtime Theatres, Inc..

D. Whether the legislature adopted the restriction “because of disagreement with the message it conveys,” Hill v. Colorado.

These tests usually reach the same result; consider bans on antiwar speech (content-based under all four tests) and bans on loud speech (content-neutral under all four). But sometimes they point in different directions, in which case you need to make the best arguments you can under each test, and compare and contrast with the existing cases. You can also keep in mind the following observations:

1. The term “content” doesn’t neatly fit the message sent by expressive conduct; we often speak of the content of speech, but “the content of a flag burning” or “the content of a cross or a black armband” sounds odd. Expressive conduct cases, such as Johnson, thus focus on whether the law restricts conduct because of its communicative impact (approach B) rather than its content (approach A).

2. Likewise, when the law is harm-based — when it bars any conduct (speech or not) that causes, is intended to cause, or is likely to cause a certain harm — it’s best to ask whether the law is being applied to the speech because of its communicative impact (approach B).

Consider, for example, the intentional infliction of emotional distress tort in Hustler v. Falwell, the interference with business relations tort in NAACP v. Claiborne Hardware, and the breach of the peace law in Cohen v. California. In these cases, the law punished all conduct that had certain effects; but it was applied to certain speech because the effects flowed from what the speech communicated. The Supreme Court held such application to be unconstitutional; and while the Court didn’t expressly call the laws content-based, it viewed them with the skepticism that it usually reserves for content-based restrictions.

But had the effects flowed from the noncommunicative components of the speech, the results would likely have been different. Had Falwell sued Larry Flynt for using loudspeakers outside his house in the middle of the night, had Claiborne Hardware sued the NAACP for organizing physical blockades of stores, or had Cohen been prosecuted for breaching the peace by shouting loudly in the courthouse, the court would probably have treated the law as content-neutral as applied.

3. The “secondary effects” test (approach C) does treat as content-neutral some restrictions that would be content-based under approaches A and B; consider the limits on theaters that show sexually themed movies (Renton). Still, there are important limits on the “secondary effects” test: Many speech restrictions are ultimately justified by the fear that speech will cause noncommunicative effects — for instance, that advocacy of crime will cause crime, or that offensive speech will cause a fight. But most such restrictions are treated as content-based, because they are seen as focusing on the “primary” effects of the speech, rather than as content-neutral regulations of secondary effects.

a. The tendency of speech to offend people is not treated as a secondary effect, and neither is the tendency of speech to cause harms that flow from such offense — for instance, potential fights, R.A.V. v. City of St. Paul, policing costs needed to prevent fights, Forsyth County v. Nationalist Movement, and injury to foreign diplomats’ dignity caused by protests outside their embassies, Boos v. Barry.

Restrictions justified by such harms are thus seen as content-based, not content-neutral. “‘The emotive impact of speech on its audience is not a ‘secondary effect’ unrelated to the content of the expression itself.” Johnson, quoting Boos; see also R.A.V. “Listeners’ reaction to speech is not a content-neutral basis for regulation.” Nationalist Movement; see also R.A.V.

b. The tendency of speech to persuade people to do bad things, and the harms that flow from such persuasion, are likewise not treated as secondary effects. “[When] the ‘chain of causation’ … necessarily ‘run[s] through the persuasive effect of the expressive component’ of the conduct, [the law] regulates on the basis of the ‘primary’ effect of the speech — i.e., its persuasive (or repellant) force.” R.A.V.

c. The tendency of speech to send bad messages to children is not seen as a secondary effect, and neither is its tendency to cause harms that flow from such messages — for instance, tobacco use, Lorillard Tobacco Co. v. Reilly, and the harms flowing from viewing pornography, United States v. Playboy Entertainment Group; Reno v. ACLU.

d. Even if a restriction’s core purpose is unrelated to content — e.g., when a newsrack ordinance limits newsracks on sidewalks to eliminate physical and visual clutter — any content-based distinctions within the restriction make the restriction content-based, unless the content is somehow closely related to the restriction’s neutral purpose. Thus, in City of Cincinnati v. Discovery Network, the Court held that a ban on newsracks containing publications that consisted chiefly of advertising was content-based: “[T]here are no secondary effects attributable to [the advertising-only] newsracks that distinguish them from the newsracks Cincinnati permits to remain on its sidewalks.”

e. How is Renton different? The Renton Court concluded that the tendency of speech to physically attract people who may commit nonspeech crimes (such as prostitution), or whose presence may in turn attract other undesirables, did involve a “secondary effect.” Yet this tendency does flow from the “emotive impact of speech”; the emotive impact brings in the undesirable prospective viewers. This tension may be one reason for Justice Kennedy’s skepticism about Renton in his City of Los Angeles v. Alameda Books, Inc. concurrence .

4. The “whether the government has adopted a regulation of speech because of disagreement with the message it conveys” test (approach D, Hill) can’t be taken entirely literally. Here are a few of the many cases in which the Court treated restrictions as content-based even though the government didn’t adopt the restriction because of disagreement with the message the speech conveys:

a. Carey v. Brown: The legislature banned residential picketing (except labor picketing) because it disapproved of the medium, residential picketing, not of the message. And the content-based exemption flowed from the government’s treating labor speech as especially valuable, not from the government’s disagreeing with all nonlabor speech.

b. Austin v. Michigan Chamber of Commerce: There’s no reason to think the legislature disagreed with corporate speech about candidates; it just thought that such speech could cause certain kinds of harms to the electoral process.

c. Florida Star v. B.J.F.: The legislature disapproved of the publication of the names of rape victims, but it didn’t disagree with whatever message the names sent.

d. Boos v. Barry: The legislature didn’t disagree with all messages that offended foreign diplomats. It obviously preferred that the messages not be spread — that’s true for any speech restriction — but not because of substantive disagreement with their content.

5. Remember that a law may be content-based even if it’s viewpoint-neutral. A ban on profanity, for instance, is viewpoint-neutral, but content-based. Speech restrictions fall into three categories: (1) content-neutral (and therefore viewpoint-neutral), (2) content-based but viewpoint-neutral, and (3) viewpoint-based (and therefore content-based).

6. A law may be content-neutral even if it disproportionately affects a certain kind of speech. The late 1980s bans on residential picketing, for instance, may well have disproportionately affected anti-abortion speech, since anti-abortion speakers tended to use residential picketing more than many other speakers. (See, e.g., Frisby.)