*2297Two Terms ago, in Matal v. Tam , 582 U. S. ----, 137 S.Ct. 1744, 198 L.Ed.2d 366 (2017), this Court invalidated the Lanham Act's bar on the registration of "disparag[ing]" trademarks. 15 U.S.C. § 1052(a). Although split between two non-majority opinions, all Members of the Court agreed that the provision violated the First Amendment because it discriminated on the basis of viewpoint. Today we consider a First Amendment challenge to a neighboring provision of the Act, prohibiting the registration of "immoral[ ] or scandalous" trademarks. Ibid. We hold that this provision infringes the First Amendment for the same reason: It too disfavors certain ideas.

I

Respondent Erik Brunetti is an artist and entrepreneur who founded a clothing line that uses the trademark FUCT. According to Brunetti, the mark (which functions as the clothing's brand name) is pronounced as four letters, one after the other: F-U-C-T. See Brief for Respondent 1. But you might read it differently and, if so, you would hardly be alone. See Tr. of Oral Arg. 5 (describing the brand name as "the equivalent of [the] past participle form of a well-known word of profanity"). That common perception caused difficulties for Brunetti when he tried to register his mark with the U. S. Patent and Trademark Office (PTO).

Under the Lanham Act, the PTO administers a federal registration system for trademarks. See 15 U.S.C. §§ 1051, 1052. Registration of a mark is not mandatory. The owner of an unregistered mark may still use it in commerce and enforce it against infringers. See Tam , 582 U. S., at ----, 137 S.Ct., at 1752. But registration gives trademark owners valuable benefits. For example, registration constitutes "prima facie evidence" of the mark's validity. § 1115(a). And registration serves as "constructive notice of the registrant's claim of *2298ownership," which forecloses some defenses in infringement actions. § 1072. Generally, a trademark is eligible for registration, and receipt of such benefits, if it is "used in commerce." § 1051(a)(1). But the Act directs the PTO to "refuse[ ] registration" of certain marks. § 1052. For instance, the PTO cannot register a mark that "so resembles" another mark as to create a likelihood of confusion. § 1052(d). It cannot register a mark that is "merely descriptive" of the goods on which it is used. § 1052(e). It cannot register a mark containing the flag or insignia of any nation or State. See § 1052(b). There are five or ten more (depending on how you count). And until we invalidated the criterion two years ago, the PTO could not register a mark that "disparage[d]" a "person[ ], living or dead." § 1052(a) ; see Tam , 582 U. S. ----, 137 S.Ct. 1744, 198 L.Ed.2d 366.

This case involves another of the Lanham Act's prohibitions on registration-one applying to marks that "[c]onsist[ ] of or comprise[ ] immoral[ ] or scandalous matter." § 1052(a). The PTO applies that bar as a "unitary provision," rather than treating the two adjectives in it separately. In re Brunetti , 877 F.3d 1330, 1336 (C.A. Fed. 2017) ; Brief for Petitioner 6 (stating that the PTO "has long treated the two terms as composing a single category"). To determine whether a mark fits in the category, the PTO asks whether a "substantial composite of the general public" would find the mark "shocking to the sense of truth, decency, or propriety"; "giving offense to the conscience or moral feelings"; "calling out for condemnation"; "disgraceful"; "offensive"; "disreputable"; or "vulgar." 877 F.3d at 1336 (internal quotation marks omitted); see Brief for Petitioner 6 (agreeing that the PTO "generally defines" the category in that way).

Both a PTO examining attorney and the PTO's Trademark Trial and Appeal Board decided that Brunetti's mark flunked that test. The attorney determined that FUCT was "a total vulgar" and "therefore[ ] unregistrable." App. 27-28. On review, the Board stated that the mark was "highly offensive" and "vulgar," and that it had "decidedly negative sexual connotations." App. to Pet. for Cert. 59a, 64a-65a. As part of its review, the Board also considered evidence of how Brunetti used the mark. It found that Brunetti's website and products contained imagery, near the mark, of "extreme nihilism" and "anti-social" behavior. Id., at 64a. In that context, the Board thought, the mark communicated "misogyny, depravity, [and] violence." Ibid . The Board concluded: "Whether one considers [the mark] as a sexual term, or finds that [Brunetti] has used [the mark] in the context of extreme misogyny, nihilism or violence, we have no question but that [the term is] extremely offensive." Id., at 65a.

Brunetti then brought a facial challenge to the "immoral or scandalous" bar in the Court of Appeals for the Federal Circuit. That court found the prohibition to violate the First Amendment. As usual when a lower court has invalidated a federal statute, we granted certiorari. 586 U. S. ----, 139 S.Ct. 782, 202 L.Ed.2d 510 (2019).

II

This Court first considered a First Amendment challenge to a trademark registration restriction in Tam , just two Terms ago. There, the Court declared unconstitutional the Lanham Act's ban on registering marks that "disparage" any "person[ ], living or dead." § 1052(a). The eight-Justice Court divided evenly between two opinions and could not agree on the overall framework for deciding the case. (In particular, no majority emerged to resolve *2299whether a Lanham Act bar is a condition on a government benefit or a simple restriction on speech.) But all the Justices agreed on two propositions. First, if a trademark registration bar is viewpoint-based, it is unconstitutional. See 582 U. S., at ---- - ----, ---- - ----, 137 S.Ct., at 1751, 1762-1763 (opinion of ALITO, J.); id., at ---- - ----, ----, 137 S.Ct., at 1751, 1753 (opinion of Kennedy, J.). And second, the disparagement bar was viewpoint-based. See id., at ---- - ----, ---- - ----, 137 S.Ct., at 1751, 1762-1763 (opinion of ALITO, J.); id., at ---- - ----, 137 S.Ct., at 1751-1753 (opinion of Kennedy, J.)

The Justices thus found common ground in a core postulate of free speech law: The government may not discriminate against speech based on the ideas or opinions it conveys. See Rosenberger v. Rector and Visitors of Univ. of Va. , 515 U.S. 819, 829-830, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995) (explaining that viewpoint discrimination is an "egregious form of content discrimination" and is "presumptively unconstitutional"). In Justice Kennedy's explanation, the disparagement bar allowed a trademark owner to register a mark if it was "positive" about a person, but not if it was "derogatory." Tam , 582 U. S., at ----, 137 S.Ct., at 1750. That was the "essence of viewpoint discrimination," he continued, because "[t]he law thus reflects the Government's disapproval of a subset of messages it finds offensive." Id., at ---- - ----, 137 S.Ct., at 1766. Justice ALITO emphasized that the statute "denie[d] registration to any mark" whose disparaging message was "offensive to a substantial percentage of the members of any group." Id., at ----, 137 S.Ct., at 1763. The bar thus violated the "bedrock First Amendment principle" that the government cannot discriminate against "ideas that offend." Id., at ---- - ----, 137 S.Ct., at 1751. Slightly different explanations, then, but a shared conclusion: Viewpoint discrimination doomed the disparagement bar.

If the "immoral or scandalous" bar similarly discriminates on the basis of viewpoint, it must also collide with our First Amendment doctrine. The Government does not argue otherwise. In briefs and oral argument, the Government offers a theory for upholding the bar if it is viewpoint-neutral (essentially, that the bar would then be a reasonable condition on a government benefit). See Brief for Petitioner 14-26. But the Government agrees that under Tam it may not "deny registration based on the views expressed" by a mark. Tr. of Oral Arg. 24. "As the Court's Tam decision establishes," the Government says, "the criteria for federal trademark registration" must be "viewpoint-neutral to survive Free Speech Clause review." Pet. for Cert. 19. So the key question becomes: Is the "immoral or scandalous" criterion in the Lanham Act viewpoint-neutral or viewpoint-based?

It is viewpoint-based. The meanings of "immoral" and "scandalous" are not mysterious, but resort to some dictionaries still helps to lay bare the problem. When is expressive material "immoral"? According to a standard definition, when it is "inconsistent with rectitude, purity, or good morals"; "wicked"; or "vicious." Webster's New International Dictionary 1246 (2d ed. 1949). Or again, when it is "opposed to or violating morality"; or "morally evil." Shorter Oxford English Dictionary 961 (3d ed. 1947). So the Lanham Act permits registration of marks that champion society's sense of rectitude and morality, but not marks that denigrate those concepts. And when is such material "scandalous"? Says a typical definition, when it "giv[es] offense to the conscience or moral feelings"; "excite[s] reprobation";

*2300or "call[s] out condemnation." Webster's New International Dictionary, at 2229. Or again, when it is "shocking to the sense of truth, decency, or propriety"; "disgraceful"; "offensive"; or "disreputable." Funk & Wagnalls New Standard Dictionary 2186 (1944). So the Lanham Act allows registration of marks when their messages accord with, but not when their messages defy, society's sense of decency or propriety. Put the pair of overlapping terms together and the statute, on its face, distinguishes between two opposed sets of ideas: those aligned with conventional moral standards and those hostile to them; those inducing societal nods of approval and those provoking offense and condemnation. The statute favors the former, and disfavors the latter. "Love rules"? "Always be good"? Registration follows. "Hate rules"? "Always be cruel"? Not according to the Lanham Act's "immoral or scandalous" bar.

The facial viewpoint bias in the law results in viewpoint-discriminatory application. Recall that the PTO itself describes the "immoral or scandalous" criterion using much the same language as in the dictionary definitions recited above. See supra, at 2298. The PTO, for example, asks whether the public would view the mark as "shocking to the sense of truth, decency, or propriety"; "calling out for condemnation"; "offensive"; or "disreputable." Brief for Petitioner 6 (internal quotation marks omitted). Using those guideposts, the PTO has refused to register marks communicating "immoral" or "scandalous" views about (among other things) drug use, religion, and terrorism. But all the while, it has approved registration of marks expressing more accepted views on the same topics. See generally Gilson & LaLonde, Trademarks Laid Bare, 101 Trademark Reporter 1476, 1510-1513, 1518-1522 (2011) ; Brief for Barton Beebe et al. as Amici Curiae 28-29.

Here are some samples. The PTO rejected marks conveying approval of drug use (YOU CAN'T SPELL HEALTHCARE WITHOUT THC for pain-relief medication, MARIJUANA COLA and KO KANE for beverages) because it is scandalous to "inappropriately glamoriz[e] drug abuse." PTO, Office Action of Aug. 28, 2010, Serial No. 85038867; see Office Action of Dec. 24, 2009, Serial No. 77833964; Office Action of Nov. 17, 2009, Serial No. 77671304. But at the same time, the PTO registered marks with such sayings as D.A.R.E. TO RESIST DRUGS AND VIOLENCE and SAY NO TO DRUGS-REALITY IS THE BEST TRIP IN LIFE. See PTO, Reg. No. 2975163 (July 26, 2005); Reg. No. 2966019 (July 12, 2005). Similarly, the PTO disapproved registration for the mark BONG HITS 4 JESUS because it "suggests that people should engage in an illegal activity [in connection with] worship" and because "Christians would be morally outraged by a statement that connects Jesus Christ with illegal drug use." Office Action of Mar. 15, 2008, Serial No. 77305946. And the PTO refused to register trademarks associating religious references with products (AGNUS DEI for safes and MADONNA for wine) because they would be "offensive to most individuals of the Christian faith" and "shocking to the sense of propriety." Ex parte Summit Brass & Bronze Works , 59 U.S.P.Q. 22, 23 (Dec. Com. Pat. 1943) ; In re Riverbank Canning Co. , 95 F.2d 327, 329 (CCPA 1938). But once again, the PTO approved marks-PRAISE THE LORD for a game and JESUS DIED FOR YOU on clothing-whose message suggested religious faith rather than blasphemy or irreverence. See Reg. No. 5265121 (Aug. 15, 2017); Reg. No. 3187985 (Dec. 19, 2006). Finally, the PTO rejected marks reflecting support for al-Qaeda (BABY AL QAEDA and AL-QAEDA on t-shirts) "because the bombing of civilians *2301and other terrorist acts are shocking to the sense of decency and call out for condemnation." Office Action of Nov. 22, 2004, Serial No. 78444968; see Office Action of Feb. 23, 2005, Serial No. 78400213. Yet it approved registration of a mark with the words WAR ON TERROR MEMORIAL. Reg. No. 5495362 (Jun. 19, 2018). Of course, all these decisions are understandable. The rejected marks express opinions that are, at the least, offensive to many Americans. But as the Court made clear in Tam , a law disfavoring "ideas that offend" discriminates based on viewpoint, in violation of the First Amendment. 582 U. S., at ----, 137 S.Ct., at 1751 (opinion of ALITO, J.); see id., at ---- - ----, 137 S.Ct., at 1762-1763 ; id., at ---- - ----, 137 S.Ct., at 1765-1766 (opinion of Kennedy, J.)

How, then, can the Government claim that the "immoral or scandalous" bar is viewpoint-neutral? The Government basically asks us to treat decisions like those described above as PTO examiners' mistakes. See Brief for Petitioner 46. Still more, the Government tells us to ignore how the Lanham Act's language, on its face, disfavors some ideas. In urging that course, the Government does not dispute that the statutory language-and words used to define it-have just that effect. At oral argument, the Government conceded: "[I]f you just looked at the words like 'shocking' and 'offensive' on their face and gave them their ordinary meanings[,] they could easily encompass material that was shocking [or offensive] because it expressed an outrageous point of view or a point of view that most members" of society reject. Tr. of Oral Arg. 6. But no matter, says the Government, because the statute is "susceptible of" a limiting construction that would remove this viewpoint bias. Id., at 7 (arguing that the Court should "attempt to construe [the] statute in a way that would render it constitutional"). The Government's idea, abstractly phrased, is to narrow the statutory bar to "marks that are offensive [or] shocking to a substantial segment of the public because of their mode of expression, independent of any views that they may express." Id., at 11 (emphasis added); see Brief for Petitioner 27-28. More concretely, the Government explains that this reinterpretation would mostly restrict the PTO to refusing marks that are "vulgar"-meaning "lewd," "sexually explicit or profane." Id., at 27, 30. Such a reconfigured bar, the Government says, would not turn on viewpoint, and so we could uphold it.

But we cannot accept the Government's proposal, because the statute says something markedly different. This Court, of course, may interpret "ambiguous statutory language" to "avoid serious constitutional doubts." FCC v. Fox Television Stations, Inc. , 556 U.S. 502, 516, 129 S.Ct. 1800, 173 L.Ed.2d 738 (2009). But that canon of construction applies only when ambiguity exists. "We will not rewrite a law to conform it to constitutional requirements." United States v. Stevens , 559 U.S. 460, 481, 130 S.Ct. 1577, 176 L.Ed.2d 435 (2010) (internal quotation marks and alteration omitted). So even assuming the Government's reading would eliminate First Amendment problems, we may adopt it only if we can see it in the statutory language. And we cannot. The "immoral or scandalous" bar stretches far beyond the Government's proposed construction. The statute as written does not draw the line at lewd, sexually explicit, or profane marks. Nor does it refer only to marks whose "mode of expression," independent of viewpoint, is particularly offensive. Brief for Petitioner 28 (internal quotation marks omitted). It covers the universe of immoral or scandalous-or (to use some PTO synonyms) offensive or disreputable-material. Whether or not lewd or profane. Whether the scandal and *2302immorality comes from mode or instead from viewpoint. To cut the statute off where the Government urges is not to interpret the statute Congress enacted, but to fashion a new one.*

And once the "immoral or scandalous" bar is interpreted fairly, it must be invalidated. The Government just barely argues otherwise. In the last paragraph of its brief, the Government gestures toward the idea that the provision is salvageable by virtue of its constitutionally permissible applications (in the Government's view, its applications to lewd, sexually explicit, or profane marks). See id. , at 47. In other words, the Government invokes our First Amendment overbreadth doctrine, and asks us to uphold the statute against facial attack because its unconstitutional applications are not "substantial" relative to "the statute's plainly legitimate sweep." Stevens , 559 U.S. at 473, 130 S.Ct. 1577 (internal quotation marks omitted). But to begin with, this Court has never applied that kind of analysis to a viewpoint-discriminatory law. In Tam , for example, we did not pause to consider whether the disparagement clause might admit some permissible applications (say, to certain libelous speech) before striking it down. The Court's finding of viewpoint bias ended the matter. And similarly, it seems unlikely we would compare permissible and impermissible applications if Congress outright banned "offensive" (or to use some other examples, "divisive" or "subversive") speech. Once we have found that a law "aim[s] at the suppression of " views, why would it matter that Congress could have captured some of the same speech through a viewpoint-neutral statute? Tam , 582 U. S., at ----, 137 S.Ct., at 1761 (opinion of Kennedy, J.). But in any event, the "immoral or scandalous" bar is substantially overbroad. There are a great many immoral and scandalous ideas in the world (even more than there are swearwords), and the Lanham Act covers them all. It therefore violates the First Amendment.

We accordingly affirm the judgment of the Court of Appeals.

It is so ordered.

For the reasons explained in the opinion of the Court, the provision of the Lanham Act at issue in this case violates the Free Speech Clause of the First Amendment because it discriminates on the basis of viewpoint and cannot be fixed without rewriting the statute. Viewpoint discrimination is poison to a free society. But in many countries with constitutions or legal traditions that claim to protect freedom of speech, serious viewpoint discrimination is now tolerated, and such discrimination has become increasingly prevalent in this country. At a time when free speech is under *2303attack, it is especially important for this Court to remain firm on the principle that the First Amendment does not tolerate viewpoint discrimination. We reaffirm that principle today.

Our decision is not based on moral relativism but on the recognition that a law banning speech deemed by government officials to be "immoral" or "scandalous" can easily be exploited for illegitimate ends. Our decision does not prevent Congress from adopting a more carefully focused statute that precludes the registration of marks containing vulgar terms that play no real part in the expression of ideas. The particular mark in question in this case could be denied registration under such a statute. The term suggested by that mark is not needed to express any idea and, in fact, as commonly used today, generally signifies nothing except emotion and a severely limited vocabulary. The registration of such marks serves only to further coarsen our popular culture. But we are not legislators and cannot substitute a new statute for the one now in force.

The Lanham Act directs the Patent and Trademark Office to refuse registration to marks that consist of or comprise "immoral, deceptive, or scandalous matter." 15 U.S.C. § 1052(a). Although the statute lists "immoral" and "scandalous" separately, the PTO has long read those terms together to constitute a unitary bar on "immoral or scandalous" marks.

The Government concedes that the provision so read is broad enough to reach not only marks that offend because of their mode of expression (such as vulgarity and profanity) but also marks that offend because of the ideas they convey. The Government urges, however, that the provision can be given a narrowing construction-it can be understood to cover only marks that offend because of their mode of expression.

The Court rejects that proposal on the ground that it would in effect rewrite the statute. I agree with the majority that the "immoral" portion of the provision is not susceptible of a narrowing construction that would eliminate its viewpoint bias. As Justice SOTOMAYOR explains, however, the "scandalous" portion of the provision is susceptible of such a narrowing construction. Standing alone, the term "scandalous" need not be understood to reach marks that offend because of the ideas they convey; it can be read more narrowly to bar only marks that offend because of their mode of expression-marks that are obscene, vulgar, or profane. That is how the PTO now understands the term, in light of our decision in Matal v. Tam , 582 U. S. ----, 137 S.Ct. 1744, 198 L.Ed.2d 366 (2017). See Tr. of Oral Arg. 4-5. I agree with Justice SOTOMAYOR that such a narrowing construction is appropriate in this context.

I also agree that, regardless of how exactly the trademark registration system is best conceived under our precedents-a question we left open in Tam -refusing registration to obscene, vulgar, or profane marks does not offend the First Amendment. Whether such marks can be registered does not affect the extent to which their owners may use them in commerce to identify goods. No speech is being restricted; no one is being punished. The owners of such marks are merely denied certain additional benefits associated with federal trademark registration. The Government, meanwhile, has an interest in not associating itself with trademarks whose content is obscene, vulgar, or profane. The First Amendment protects the freedom of speech; it does not require the Government to give aid and comfort to those *2304using obscene, vulgar, and profane modes of expression. For those reasons, I concur in part and dissent in part.

Our precedents warn us against interpreting statutes in ways that would likely render them unconstitutional. Virginia v. American Booksellers Assn., Inc. , 484 U.S. 383, 397, 108 S.Ct. 636, 98 L.Ed.2d 782 (1988) (noting that a law "will be upheld" if it is " 'readily susceptible' to a narrowing construction that would make it constitutional"); United States v. 12 200-ft. Reels of Super 8MM. Film , 413 U.S. 123, 130, n. 7, 93 S.Ct. 2665, 37 L.Ed.2d 500 (1973) (noting our "duty" to adopt a " 'fairly possible' " construction by which constitutional doubts " 'may be avoided' " (quoting United States v. Thirty-Seven Photographs , 402 U.S. 363, 369, 91 S.Ct. 1400, 28 L.Ed.2d 822 (1971) )). Following these precedents, I agree with Justice SOTOMAYOR that, for the reasons she gives, we should interpret the word "scandalous" in the present statute to refer only to certain highly "vulgar" or "obscene" modes of expression. See post , at 2310 - 2312 (opinion concurring in part and dissenting in part).

The question, then, is whether the First Amendment permits the Government to rely on this statute, as narrowly construed, to deny the benefits of federal trademark registration to marks like the one at issue here, which involves the use of the term "FUCT" in connection with a clothing line that includes apparel for children and infants. Like Justice SOTOMAYOR, I believe the answer is "yes," though my reasons differ slightly from hers.

I

A

In my view, a category-based approach to the First Amendment cannot adequately resolve the problem before us. I would place less emphasis on trying to decide whether the statute at issue should be categorized as an example of "viewpoint discrimination," "content discrimination," "commercial speech," "government speech," or the like. Rather, as I have written before, I believe we would do better to treat this Court's speech-related categories not as outcome-determinative rules, but instead as rules of thumb. See Reed v. Town of Gilbert , 576 U. S. ----, ----, 135 S.Ct. 2218, 2235, 192 L.Ed.2d 236 (2015) (opinion concurring in judgment).

After all, these rules are not absolute. The First Amendment is not the Tax Code. Indeed, even when we consider a regulation that is ostensibly "viewpoint discriminatory" or that is subject to "strict scrutiny," we sometimes find the regulation to be constitutional after weighing the competing interests involved. See, e.g. , Morse v. Frederick , 551 U.S. 393, 397, 127 S.Ct. 2618, 168 L.Ed.2d 290 (2007) ("[S]chools may take steps to safeguard those entrusted to their care from speech that can reasonably be regarded as encouraging illegal drug use"); Williams-Yulee v. Florida Bar , 575 U. S. 433, ----, 135 S.Ct. 1656, 1665-1666, 191 L.Ed.2d 570 (2015) (explaining that although " 'it is the rare case' " when a statute satisfies strict scrutiny, "those cases do arise" (quoting Burson v. Freeman , 504 U.S. 191, 211, 112 S.Ct. 1846, 119 L.Ed.2d 5 (1992) (plurality opinion))).

Unfortunately, the Court has sometimes applied these rules-especially the category of "content discrimination"-too rigidly. In a number of cases, the Court has struck down what I believe are ordinary, valid regulations that pose little or no threat to the speech interests that the First Amendment protects. See *2305Janus v. State, County, and Municipal Employees , 585 U. S. ----, ---- - ----, 138 S.Ct. 2448, 2501-2502, 201 L.Ed.2d 924 (2018) (KAGAN, J., dissenting); Sorrell v. IMS Health Inc. , 564 U.S. 552, 589-592, 131 S.Ct. 2653, 180 L.Ed.2d 544 (2011) (BREYER, J., dissenting); see generally Reed , 576 U. S., at ---- - ----, 135 S.Ct., at 2234-2236 (opinion of BREYER, J.).

Rather than deducing the answers to First Amendment questions strictly from categories, as the Court often does, I would appeal more often and more directly to the values the First Amendment seeks to protect. As I have previously written, I would ask whether the regulation at issue "works speech-related harm that is out of proportion to its justifications." United States v. Alvarez , 567 U.S. 709, 730, 132 S.Ct. 2537, 183 L.Ed.2d 574 (2012) (opinion concurring in judgment); see Reed , 576 U. S., at ----, 135 S.Ct., at 2235-2236 (opinion concurring in judgment) (discussing the matter further, particularly in respect to the category of content discrimination).

B

This case illustrates the limits of relying on rigid First Amendment categories, for the statute at issue does not fit easily into any of these categories.

The Court has not decided whether the trademark statute is simply a method of regulating pure "commercial speech." See Matal v. Tam , 582 U. S. ----, ----, 137 S.Ct. 1744, 1764, 198 L.Ed.2d 366 (2017) (opinion of ALITO, J.) (leaving open the question whether trademarks are commercial speech); id. , at ----, 137 S.Ct., at 1753 (opinion of Kennedy, J.)(same). There may be reasons for doubt on that score. Trademarks, after all, have an expressive component in addition to a commercial one, and the statute does not bar anyone from speaking. To be sure, the statute does regulate the commercial function of trademarks. But it does so in a limited way designed primarily to ensure that a mark identifies the product's source. See Wal-Mart Stores, Inc. v. Samara Brothers, Inc. , 529 U.S. 205, 212, 120 S.Ct. 1339, 146 L.Ed.2d 182 (2000).

The trademark statute cannot easily be described as a regulation of "government speech," either. Tam , 582 U. S., at ---- - ----, 137 S.Ct., at 1757-1760. The Government, however, may be loosely associated with the mark because it registers the mark and confers certain benefits upon the owner.

What about the concept of a "public forum"? Trademark registration has little in common with a traditional public forum, as the register of trademarks is not a public park, a street, or a similar forum for public debate. See Perry Ed. Assn. v. Perry Local Educators' Assn. , 460 U.S. 37, 45, 103 S.Ct. 948, 74 L.Ed.2d 794 (1983). But one can find some vague resemblance between trademark registration and what this Court refers to as a "limited public forum" created by the government for private speech. See post , at 2316 (opinion of SOTOMAYOR, J.); Christian Legal Soc. Chapter of Univ. of Cal., Hastings College of Law v. Martinez , 561 U.S. 661, 679, n. 11, 130 S.Ct. 2971, 177 L.Ed.2d 838 (2010). The trademark registration system also bears some resemblance to cases involving government subsidies for private speech, as such programs-like trademark registration-may grant a benefit to some forms of speech without prohibiting other forms of speech. See post , at 2316 (opinion of SOTOMAYOR, J.); Legal Services Corporation v. Velazquez , 531 U.S. 533, 543-544, 121 S.Ct. 1043, 149 L.Ed.2d 63 (2001) (noting that the First Amendment rules applicable to limited public forums may be "instruc[tive]" "when the government establishes a subsidy for specified ends").

As for the concepts of "viewpoint discrimination" and "content discrimination,"

*2306I agree with Justice SOTOMAYOR that the boundaries between them may be difficult to discern. Post , at 2313; see Rosenberger v. Rector and Visitors of Univ. of Va. , 515 U.S. 819, 831, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995) ("[T]he distinction is not a precise one"). Even so, it is hard to see how a statute prohibiting the registration of only highly vulgar or obscene words discriminates based on "viewpoint." Of course, such words often evoke powerful emotions. Standing by themselves, however, these words do not typically convey any particular viewpoint. See FCC v. Pacifica Foundation , 438 U.S. 726, 746, n. 22, 98 S.Ct. 3026, 57 L.Ed.2d 1073 (1978) (noting that the Government's regulation of vulgar words was based not on "point of view," but on "the way in which [speech] is expressed"). Moreover, while a restriction on the registration of highly vulgar words arguably places a content-based limit on trademark registration, it is hard to see why that label should be outcome-determinative here, for regulations governing trademark registration "inevitably involve content discrimination." Reed , 576 U. S., at ----, 135 S.Ct., at 2234-2235 (opinion of BREYER, J.); see Tam , 582 U. S., at ----, 137 S.Ct., at 1768 (opinion of Kennedy, J.) (noting that the constitutionality of some content-based trademark restrictions is "well settled"); Katyal, Trademark Intersectionality, 57 UCLA L. Rev. 1601, 1602 (2010) (noting that trademark law is "indelibly rooted in content-based considerations").

In short, the trademark statute does not clearly fit within any of the existing outcome-determinative categories. Why, then, should we rigidly adhere to these categories? Rather than puzzling over categorization, I believe we should focus on the interests the First Amendment protects and ask a more basic proportionality question: Does "the regulation at issue wor[k] harm to First Amendment interests that is disproportionate in light of the relevant regulatory objectives"? Reed , 576 U. S., at ----, 135 S.Ct., at 2235-2236 (opinion of BREYER, J.)

II

Based on this proportionality analysis, I would conclude that the statute at issue here, as interpreted by Justice SOTOMAYOR, does not violate the First Amendment.

How much harm to First Amendment interests does a bar on registering highly vulgar or obscene trademarks work? Not much. The statute leaves businesses free to use highly vulgar or obscene words on their products, and even to use such words directly next to other registered marks. Indeed, a business owner might even use a vulgar word as a trademark, provided that he or she is willing to forgo the benefits of registration. See post , at 2308 - 2309 (opinion of SOTOMAYOR, J.); Tam , 582 U. S., at ---- - ----, 137 S.Ct., at 1752-1753.

Moreover, the field at issue here, trademark law, is a highly regulated one with a specialized mission: to "hel[p] consumers identify goods and services that they wish to purchase, as well as those they want to avoid." Id. , at ----, 137 S.Ct., at 1751. As I have noted, that mission, by its very nature, requires the Government to impose limitations on speech. Supra , at 2306. Trademark law therefore forbids the registration of certain types of words-for example, those that will likely "cause confusion," or those that are "merely descriptive." 15 U.S.C. §§ 1052(d), (e). For that reason, an applicant who seeks to register a mark should not expect complete freedom to say what she wishes, but should instead expect linguistic regulation.

Now consider, by way of contrast, the Government's interests in barring the registration of highly vulgar or obscene trademarks.

*2307For one thing, when the Government registers a mark, it is necessarily "involv[ed] in promoting" that mark. Post , at 2317 - 2318 (opinion of SOTOMAYOR, J.). The Government has at least a reasonable interest in ensuring that it is not involved in promoting highly vulgar or obscene speech, and that it will not be associated with such speech.

For another, scientific evidence suggests that certain highly vulgar words have a physiological and emotional impact that makes them different in kind from most other words. See M. Mohr, Holy S***: A Brief History of Swearing 252 (2013) (Mohr) (noting the "emotional impact" of certain profane words that "excite the lower-brain circuitry responsible for emotion," resulting in "electrical impulses that can be measured in the skin"). These vulgar words originate in a different part of our brains than most other words. Id. , at 250. And these types of swear words tend to attract more attention and are harder to forget than other words. See Jay, Caldwell-Harris, & King, Recalling Taboo and Nontaboo Words, 121 Am. J. Psych. 83, 83-86 (2008) (collecting research). Notably, that has remained true even as the list of offensive swear words has changed over time: In the last few centuries, the list has evolved away from words of religious disrespect and toward words that are sexually explicit or that crudely describe bodily functions. Mohr 253. And the list of swear words may be evolving yet again, perhaps in the direction of including race-based epithets. Id. , at 254, 256.

These attention-grabbing words, though financially valuable to some businesses that seek to attract interest in their products, threaten to distract consumers and disrupt commerce. And they may lead to the creation of public spaces that many will find repellant, perhaps on occasion creating the risk of verbal altercations or even physical confrontations. (Just think about how you might react if you saw someone wearing a t-shirt or using a product emblazoned with an odious racial epithet.) The Government thus has an interest in seeking to disincentivize the use of such words in commerce by denying the benefit of trademark registration. Cf. Brandenburg v. Ohio , 395 U.S. 444, 447, 89 S.Ct. 1827, 23 L.Ed.2d 430 (1969) (per curiam ) (permitting regulation of words "directed to inciting or producing imminent lawless action" and "likely to incite or produce such action").

Finally, although some consumers may be attracted to products labeled with highly vulgar or obscene words, others may believe that such words should not be displayed in public spaces where goods are sold and where children are likely to be present. They may believe that trademark registration of such words could make it more likely that children will be exposed to public displays involving such words. To that end, the Government may have an interest in protecting the sensibilities of children by barring the registration of such words. See Denver Area Ed. Telecommunications Consortium, Inc. v. FCC , 518 U.S. 727, 743, 116 S.Ct. 2374, 135 L.Ed.2d 888 (1996) (plurality opinion) (noting the Government's interest in "protec[ting] children from exposure to patently offensive sex-related material"); Ginsberg v. New York , 390 U.S. 629, 640, 88 S.Ct. 1274, 20 L.Ed.2d 195 (1968) (noting the government's "interest in the well-being of its youth").

The upshot of this analysis is that the narrowing construction articulated by Justice SOTOMAYOR risks some harm to First Amendment interests, but not very much. And applying that interpretation seems a reasonable way-perhaps the only way-to further legitimate government interests. Of course, there is a risk that the *2308statute might be applied in a manner that stretches it beyond the few vulgar words that are encompassed by the narrow interpretation Justice SOTOMAYOR sets forth. That risk, however, could be mitigated by internal agency review to ensure that agency officials do not stray beyond their mandate. In any event, I do not believe that this risk alone warrants the facial invalidation of this statute.

I would conclude that the prohibition on registering "scandalous" marks does not "wor[k] harm to First Amendment interests that is disproportionate in light of the relevant regulatory objectives." Reed , 576 U. S., at ----, 135 S.Ct., at 2235-2236 (opinion of BREYER, J.). I would therefore uphold this part of the statute. I agree with the Court, however, that the bar on registering "immoral" marks violates the First Amendment. Because Justice SOTOMAYOR reaches the same conclusions, using roughly similar reasoning, I join her opinion insofar as it is consistent with the views set forth here.

Justice SOTOMAYOR, with whom Justice BREYER joins, concurring in part and dissenting in part.

The Court's decision today will beget unfortunate results. With the Lanham Act's scandalous-marks provision, 15 U.S.C. § 1052(a), struck down as unconstitutional viewpoint discrimination, the Government will have no statutory basis to refuse (and thus no choice but to begin) registering marks containing the most vulgar, profane, or obscene words and images imaginable.

The coming rush to register such trademarks-and the Government's immediate powerlessness to say no-is eminently avoidable. Rather than read the relevant text as the majority does, it is equally possible to read that provision's bar on the registration of "scandalous" marks to address only obscenity, vulgarity, and profanity. Such a narrowing construction would save that duly enacted legislative text by rendering it a reasonable, viewpoint-neutral restriction on speech that is permissible in the context of a beneficial governmental initiative like the trademark-registration system. I would apply that narrowing construction to the term "scandalous" and accordingly reject petitioner Erik Brunetti's facial challenge.

I

Trademark registration, as the majority notes, is not required for using, owning, or suing others for infringing a trademark. Rather, the trademark-registration system is an ancillary system set up by the Government that confers a small number of noncash benefits on trademark-holders who register their marks. See ante , at 2297 - 2298.

The Government need not provide this largely commercial benefit at all. Once the Government does provide the benefit, however, it may not restrict access on the basis of the viewpoint expressed by the relevant mark. See ante, at 2305 - 2306. For that reason, the Court concluded in Matal v. Tam , 582 U. S. ----, 137 S.Ct. 1744, 198 L.Ed.2d 366 (2017), that § 1052(a) 's provision directing the U. S. Patent and Trademark Office (PTO) to deny registration to "disparag[ing]" trademarks was unconstitutional. This case centers on a neighboring set of restrictions: § 1052(a) 's provision barring registration of marks featuring "immoral ... or scandalous matter."

The majority finds viewpoint discrimination here by treating the terms "scandalous" and "immoral" as comprising a unified standard that allows messages "aligned with conventional moral standards" but forbids messages "hostile to" such standards. See ante , at 2299 - 2300.

*2309While the majority's interpretation of the statute is a reasonable one, it is not the only reasonable one.

A

As the majority notes, there are dictionary definitions for both "immoral" and "scandalous" that do suggest a viewpoint-discriminatory meaning. See ante , at 2299 - 2300. And as for the word "immoral," I agree with the majority that there is no tenable way to read it that would ameliorate the problem. The word clearly connotes a preference for "rectitude and morality" over its opposite. See ante , at 2299.

It is with regard to the word "scandalous" that I part ways with the majority. Unquestionably, "scandalous" can mean something similar to "immoral" and thus favor some viewpoints over others. See ante , at 2300. But it does not have to be read that way. To say that a word or image is "scandalous" can instead mean that it is simply indecent, shocking, or generally offensive. See Funk & Wagnalls New Standard Dictionary 2186 (1944) (Funk & Wagnalls) ("shocking to the sense of truth, decency, or propriety; disgraceful, offensive" (emphasis added)); Webster's New International Dictionary 2229 (1942) ("exciting reprobation; calling out condemnation"); 9 Oxford English Dictionary 175 (1933) ("Of the nature of, or causing, a 'stumbling-block' or occasion of offence"); 8 Century Dictionary and Cyclopedia 5374 (1911) (Century Dictionary) ("Causing scandal or offense; exciting reproach or reprobation; extremely offensive to the sense of duty or propriety; shameful; shocking"); see also Webster's New College Dictionary 1008 (3d ed. 2005) ("shocking or offensive"). That offensiveness could result from the views expressed, but it could also result from the way in which those views are expressed: using a manner of expression that is "shocking to [one's] sense of ... decency," Funk & Wagnalls 2186, or "extremely offensive to the sense of ... propriety," 8 Century Dictionary 5374.

The word "scandalous" on its own, then, is ambiguous: It can be read broadly (to cover both offensive ideas and offensive manners of expressing ideas), or it can be read narrowly (to cover only offensive modes of expression). That alone raises the possibility that a limiting construction might be appropriate. But the broader text confirms the reasonableness of the narrower reading, because the word "scandalous" appears in the statute alongside other words that can, and should, be read to constrain its scope.

It is foundational "that a statute is to be read as a whole, since the meaning of statutory language, plain or not, depends on context." King v. St. Vincent's Hospital , 502 U.S. 215, 221, 112 S.Ct. 570, 116 L.Ed.2d 578 (1991) (citation omitted). " 'Words are not pebbles in alien juxtaposition; they have only a communal existence; and not only does the meaning of each interpenetrate the other, but all in their aggregate take their purport from the setting in which they are used.' " Ibid. (quoting NLRB v. Federbush Co. , 121 F.2d 954, 957 (C.A.2 1941) (L. Hand, J.)). Accordingly, and relatedly, courts should, to the extent possible, read statutes so that " 'no clause, sentence, or word shall be superfluous, void, or insignificant.' " TRW Inc. v. Andrews , 534 U.S. 19, 31, 122 S.Ct. 441, 151 L.Ed.2d 339 (2001).1

*2310Here, Congress used not only the word "scandalous," but also the words "immoral" and "disparage," in the same block of statutory text-each as a separate feature that could render a mark unregistrable. See § 1052(a). Tam already decided that "disparage" served to prohibit marks that were offensive because they derided a particular person or group. See 582 U. S., at ----, 137 S.Ct., at 1749 (opinion of ALITO, J.) ("It denies registration to any mark that is offensive to a substantial percentage of the members of any group"); id., at ----, 137 S.Ct., at 1750 (opinion of Kennedy, J.) ("[A]n applicant may register a positive or benign mark but not a derogatory one"). That defines one of the three words. Meanwhile, as the majority explains, the word "immoral" prohibits marks that are offensive because they transgress widely held moral beliefs. See ante , at 2299 - 2300. That defines a second of the three words.

With marks that are offensive because they are disparaging and marks that are offensive because they are immoral already covered, what work did Congress intend for "scandalous" to do? A logical answer is that Congress meant for "scandalous" to target a third and distinct type of offensiveness: offensiveness in the mode of communication rather than the idea. The other two words cover marks that are offensive because of the ideas they express; the "scandalous" clause covers marks that are offensive because of the mode of expression, apart from any particular message or idea.

To be sure, there are situations in which it makes sense to treat adjoining words as expressing the same or highly similar concepts (even at the risk of some redundancy). Cf. Swearingen v. United States , 161 U.S. 446, 450, 16 S.Ct. 562, 40 L.Ed. 765 (1896) (construing " 'obscene, lewd or lascivious' " to have a unified meaning). That is essentially the approach that the majority takes. See ante, at 2300.2 But that is not the approach that Congress appears to have intended here. For example, "scandalous" does not serve as a broader catchall at the end of a list of similar words that all point in one direction. E.g., Washington State Dept. of Social and Health Servs. v. Guardianship Estate of Keffeler , 537 U.S. 371, 384, 123 S.Ct. 1017, 154 L.Ed.2d 972 (2003). Nor is "scandalous" simply grouped among a number of closely related terms that help define its meaning. E.g., Gustafson v. Alloyd Co. , 513 U.S. 561, 575, 115 S.Ct. 1061, 131 L.Ed.2d 1 (1995).

The text of § 1052, instead, is a grab bag: It bars the registration of marks featuring "immoral, deceptive, or scandalous matter," as well as, inter alia , disparaging marks, flags, insignias, mislabeled wines, and deceased Presidents. See §§ 1052(a) - (e). This is not, in other words, a situation in which Congress was simply being "verbos[e] and proli[x]," Bruesewitz v. Wyeth LLC , 562 U.S. 223, 236, 131 S.Ct. 1068, 179 L.Ed.2d 1 (2011), using two synonyms in rapid-fire succession when one would have done fine. Instead, "scandalous" and "immoral" are separated by an unrelated word ("deceptive") and mixed in with a lengthy series of other, unrelated concepts. The *2311two therefore need not be interpreted as mutually reinforcing under the Court's precedents. See Graham County Soil and Water Conservation Dist. v. United States ex rel. Wilson , 559 U.S. 280, 288, 130 S.Ct. 1396, 176 L.Ed.2d 225 (2010).

For that reason, while the majority offers a reasonable reading of "scandalous," it also unnecessarily and ill-advisedly collapses the words "scandalous" and "immoral." Instead, it should treat them as each holding a distinct, nonredundant meaning, with "immoral" covering marks that are offensive because they transgress social norms, and "scandalous" covering marks that are offensive because of the mode in which they are expressed.

What would it mean for "scandalous" in § 1052(a) to cover only offensive modes of expression? The most obvious ways-indeed, perhaps the only conceivable ways-in which a trademark can be expressed in a shocking or offensive manner are when the speaker employs obscenity, vulgarity, or profanity.3 Obscenity has long been defined by this Court's decision in Miller v. California , 413 U.S. 15, 93 S.Ct. 2607, 37 L.Ed.2d 419 (1973). See id., at 24-26, 93 S.Ct. 2607. As for what constitutes "scandalous" vulgarity or profanity, I do not offer a list, but I do interpret the term to allow the PTO to restrict (and potentially promulgate guidance to clarify) the small group of lewd words or "swear" words that cause a visceral reaction, that are not commonly used around children, and that are prohibited in comparable settings.4 Cf. 18 U.S.C. § 1464 (prohibiting "obscene, indecent, or profane language" in radio communications); FCC v. Pacifica Foundation , 438 U.S. 726, 746, and n. 22, 98 S.Ct. 3026, 57 L.Ed.2d 1073 (1978) (opinion of Stevens, J.) (regulator's objection to a monologue containing various "four-letter words" was not to its "point of view, but to the way in which it [wa]s expressed"); 46 C.F.R. § 67.117(b)(3) (2018) (Coast Guard regulation prohibiting vessel names that "contain" or are "phonetically identical to obscene, indecent, or profane language, or to racial or ethnic epithets"); see also Jacobs, The Public Sensibilities Forum, 95 Nw. U. L. Rev. 1357, 1416-1417, and n. 432 (2001) (noting that "swear words" are "perhaps more than any other categor[y] capable of specific articulation" and citing one state agency's list). Of course, "scandalous" offers its own limiting principle: if a word, though not exactly polite, cannot be said to be "scandalous"-e.g., "shocking" or "extremely offensive," 8 Century Dictionary 5374-it is clearly not the kind of vulgarity or profanity that Congress intended to target. Everyone can think of a small number of words (including the apparent homonym of Brunetti's mark) that would, however, plainly qualify.5

*2312B

A limiting construction like the one just discussed is both appropriate in this context and consistent with past precedent. First, while a limiting construction must always be at least reasonable, there are contexts in which imposing such a construction is more appropriate than others. The most obvious example of a setting where more caution is required is in the realm of criminal statutes, where considerations such as the prohibition against vagueness and the rule of lenity come into play. See Reno v. American Civil Liberties Union , 521 U.S. 844, 872, 117 S.Ct. 2329, 138 L.Ed.2d 874 (1997) (noting that "[t]he severity of criminal sanctions" can increase First Amendment concerns); Board of Airport Comm'rs of Los Angeles v. Jews for Jesus, Inc. , 482 U.S. 569, 575-576, 107 S.Ct. 2568, 96 L.Ed.2d 500 (1987) (declining to apply a limiting construction to a provision that banned " 'First Amendment activities' " from an airport and noting that the limiting construction proposed would " 'confe[r] on police a virtually unrestrained power to arrest and charge persons with a violation,' " leading to " 'self-evident' " " 'opportunity for abuse' "). Here, however, the question is only whether the Government must be forced to provide the ancillary benefit of trademark registration to pre-existing trademarks that use even the most extreme obscenity, vulgarity, or profanity. The stakes are far removed from a situation in which, say, Brunetti was facing a threat to his liberty, or even his right to use and enforce his trademark in commerce.

Second, the Court has in the past accepted or applied similarly narrow constructions to avoid constitutional infirmities. In Chaplinsky v. New Hampshire , 315 U.S. 568, 62 S.Ct. 766, 86 L.Ed. 1031 (1942), for example, the Court accepted the New Hampshire Supreme Court's narrowing of a state statute covering " 'any offensive, derisive or annoying word,' " id., at 569, 62 S.Ct. 766, to reach only those words that would strike the average person as being "plainly likely to cause a breach of the peace by the addressee," id., at 573, 62 S.Ct. 766. "[T]hus construed," this Court decided, the statute did not violate the right to free speech. Ibid. ; see also Boos v. Barry , 485 U.S. 312, 329-330, 108 S.Ct. 1157, 99 L.Ed.2d 333 (1988) (accepting Court of Appeals' construction of a statute making it illegal " 'to congregate within 500 feet of any [embassy, legation, or consulate] and refuse to disperse after having been ordered so to do by the police' " to reach only "congregations that are directed at an embassy" and " 'only when the police reasonably believe that a threat to the security or peace of the embassy is present' ").

In Frisby v. Schultz , 487 U.S. 474, 108 S.Ct. 2495, 101 L.Ed.2d 420 (1988), the Court addressed an ordinance that prohibited " 'picketing before or about the residence or dwelling of any individual.' " Id., at 477, 108 S.Ct. 2495. The Court construed the statute to reach only "focused picketing taking place solely in front of a particular residence." Id., at 483, 108 S.Ct. 2495. Given that "narrow scope," the statute *2313was not facially unconstitutional. Id., at 488, 108 S.Ct. 2495 ; see also In re Brunetti , 877 F.3d 1330, 1358 (C.A. Fed. 2017) (Dyk, J., concurring in judgment) (noting this Court's narrow constructions of federal obscenity statutes).

Taking the word "scandalous" to target only those marks that employ an offensive mode of expression follows a similar practice. To be sure, the word could be read more broadly, thereby sweeping unconstitutionally into viewpoint discrimination. And imposing a limiting construction is, of course, "not a license for the judiciary to rewrite language enacted by the legislature." United States v. Albertini , 472 U.S. 675, 680, 105 S.Ct. 2897, 86 L.Ed.2d 536 (1985). But where the Court can reasonably read a statute like this one to save it, the Court should do so. See Stern v. Marshall , 564 U.S. 462, 477-478, 131 S.Ct. 2594, 180 L.Ed.2d 475 (2011) ; NLRB v. Jones & Laughlin Steel Corp. , 301 U.S. 1, 30, 57 S.Ct. 615, 81 L.Ed. 893 (1937).

II

Adopting a narrow construction for the word "scandalous"-interpreting it to regulate only obscenity, vulgarity, and profanity-would save it from unconstitutionality. Properly narrowed, "scandalous" is a viewpoint-neutral form of content discrimination that is permissible in the kind of discretionary governmental program or limited forum typified by the trademark-registration system.

A

Content discrimination occurs whenever a government regulates "particular speech because of the topic discussed or the idea or message expressed." Reed v. Town of Gilbert , 576 U. S. ----, ----, 135 S.Ct. 2218, 2227, 192 L.Ed.2d 236 (2015) ; see also Ward v. Rock Against Racism , 491 U.S. 781, 791, 109 S.Ct. 2746, 105 L.Ed.2d 661 (1989) ("Government regulation of expressive activity is content neutral so long as it is 'justified without reference to the content of the regulated speech' "). Viewpoint discrimination is "an egregious form of content discrimination" in which "the government targets not subject matter, but particular views taken by speakers on a subject." Rosenberger v. Rector and Visitors of Univ. of Va. , 515 U.S. 819, 829, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995).

While the line between viewpoint-based and viewpoint-neutral content discrimination can be "slippery," see Corbin, Mixed Speech: When Speech Is Both Private and Governmental, 83 N. Y. U. L. Rev. 605, 651 (2008), it is in any event clear that a regulation is not viewpoint discriminatory (or even content discriminatory) simply because it has an "incidental effect" on a certain subset of views. Ward , 491 U.S. at 791, 109 S.Ct. 2746. Some people, for example, may have the viewpoint that society should be more sexually liberated and feel that they cannot express that view sufficiently without the use of pornographic words or images. That does not automatically make a restriction on pornography into viewpoint discrimination, despite the fact that such a restriction limits communicating one's views on sexual liberation in that way. See ibid. ; Renton v. Playtime Theatres, Inc. , 475 U.S. 41, 48, 106 S.Ct. 925, 89 L.Ed.2d 29 (1986).

Restrictions on particular modes of expression do not inherently qualify as viewpoint discrimination; they are not by nature examples of "the government target[ing] ... particular views taken by speakers on a subject." Rosenberger , 515 U.S. at 829, 115 S.Ct. 2510. For example, a ban on lighting fires in the town square does not facially violate the First Amendment simply because it makes it marginally harder for would-be flag-burners to *2314express their views in that place. See R. A. V. v. St. Paul , 505 U.S. 377, 385, 112 S.Ct. 2538, 120 L.Ed.2d 305 (1992). By the same token, "fighting words are categorically excluded from the protection of the First Amendment" not because they have no content or express no viewpoint (often quite the opposite), but because "their content embodies a particularly intolerable (and socially unnecessary) mode of expressing whatever idea the speaker wishes to convey." Id., at 393, 112 S.Ct. 2538 ; see id., at 385-386, 112 S.Ct. 2538 ; cf. Bolger v. Youngs Drug Products Corp. , 463 U.S. 60, 84, 103 S.Ct. 2875, 77 L.Ed.2d 469 (1983) (Stevens, J., concurring in judgment) ("It matters whether a law regulates communications for their ideas or for their style").

A restriction on trademarks featuring obscenity, vulgarity, or profanity is similarly viewpoint neutral, though it is naturally content-based.6 See R. A. V. , 505 U.S. at 383, 112 S.Ct. 2538 (kinds of speech like "obscenity, defamation, etc." may "be regulated because of their constitutionally proscribable content" (emphasis deleted)); see also Bethel School Dist. No. 403 v. Fraser , 478 U.S. 675, 685, 106 S.Ct. 3159, 92 L.Ed.2d 549 (1986) (treating punishment of "offensively lewd and indecent speech" as viewpoint neutral); Pacifica , 438 U.S. at 745-746, and n. 22, 98 S.Ct. 3026 (treating regulation of profane monologue as viewpoint neutral). Indeed, the statute that the Court upheld in Chaplinsky itself had been construed to cover, among other kinds of "disorderly words," "profanity, obscenity and threats," 315 U.S. at 573, 62 S.Ct. 766, despite the fact that such words had been used in that case to communicate an expressive message, id., at 574, 62 S.Ct. 766. To treat a restriction on vulgarity, profanity, or obscenity as viewpoint discrimination would upend decades of precedent.7

Brunetti invokes Cohen v. California , 403 U.S. 15, 91 S.Ct. 1780, 29 L.Ed.2d 284 (1971), to argue that the restriction at issue here is viewpoint discriminatory. But Cohen -which did not employ the precise taxonomy that is more common today-does not reach as far as Brunetti wants. Cohen arose in the criminal context: Cohen had been arrested and imprisoned under a California criminal statute targeting disturbances of the peace because he was "wearing a jacket bearing the words 'F[***] the Draft.' " Id., at 16, 91 S.Ct. 1780. The Court held that applying that statute to Cohen because of his jacket violated the First Amendment. Id., at 26, 91 S.Ct. 1780. But the Court did not suggest that the State had targeted Cohen to suppress his view itself (i.e., his sharp distaste for the draft), such that it would *2315have accepted an equally colorful statement of praise for the draft (or hostility toward war protesters). Rather, the Court suggested that the State had simply engaged in what later courts would more precisely call viewpoint-neutral content discrimination-it had regulated "the form or content of individual expression." Id., at 24, 91 S.Ct. 1780 ; see id., at 25-26, 91 S.Ct. 1780.

Cohen also famously recognized that "words are often chosen as much for their emotive as their cognitive force," id., at 26, 91 S.Ct. 1780, and that "one man's vulgarity is another's lyric," id., at 25, 91 S.Ct. 1780. That is all consistent with observing that a plain, blanket restriction on profanity (regardless of the idea to which it is attached) is a viewpoint-neutral form of content discrimination. The essence of Cohen 's discussion is that profanity can serve to tweak (or amplify) the viewpoint that a message expresses, such that it can be hard to disentangle the profanity from the underlying message-without the profanity, the message is not quite the same. See id., at 25-26, 91 S.Ct. 1780. But those statements merely reinforce that profanity is still properly understood as protected First Amendment content. See also R. A. V. , 505 U.S. at 384-385, 112 S.Ct. 2538. Cohen 's discussion does not also go further to declare, as Brunetti suggests, that a provision that treats all instances of profanity equally is nevertheless by nature an instance of "the government target[ing] ... particular views taken by speakers on a subject." Rosenberger , 515 U.S. at 829, 115 S.Ct. 2510. To be sure, such a restriction could have the incidental effect of tamping down the overall volume of debate on all sides. But differential effects alone, as explained above, do not render a restriction viewpoint (or even content) discriminatory. See Ward , 491 U.S. at 791-792, 109 S.Ct. 2746.8

Cohen therefore does not resolve this case in Brunetti's favor. Yes, Brunetti has been, as Cohen was, subject to content discrimination, but that content discrimination is properly understood as viewpoint neutral. And whereas even viewpoint-neutral content discrimination is (in all but the most compelling cases, such as threats) impermissible in the context of a criminal prosecution like the one that Cohen faced, Brunetti is subject to such regulation only in the context of the federal trademark-registration system. I discuss next why that distinction matters.

B

While the Court has often subjected even viewpoint-neutral content discrimination to strict constitutional scrutiny, see, e.g., Reed , 576 U. S., at ----, 135 S.Ct., at 2227, there are contexts in which it does not, see, e.g., Rosenberger , 515 U.S. at 829-830, 115 S.Ct. 2510. When that is the case, the difference between viewpoint-based and viewpoint-neutral content discrimination can be decisive. The federal *2316trademark-registration system is such a context.

Rights to a trademark itself arise through use, not registration. Regardless of whether a trademark is registered, it can be used, owned, and enforced against would-be infringers. See B&B Hardware, Inc. v. Hargis Industries, Inc., 575 U. S. ----, ----, ----, 135 S.Ct. 1293, 1299-1300, 1300-1301, 191 L.Ed.2d 222 (2015). Trademark registration, meanwhile, confers several ancillary benefits on trademark-holders who meet Congress' specifications, including for example, additional protections against infringers. See ante , at 2297 - 2298; Tam , 582 U. S., at ----, 137 S.Ct., at 1753. Registering a mark in the Government's searchable register puts the world on notice (whether actual or constructive) that a party is asserting ownership of that mark.9 Registration, in short, is a helpful system, but it is one that the Government is under no obligation to establish and that is collateral to the existence and use of trademarks themselves. There is no evidence that speech or commerce would be endangered if the Government were not to provide it at all.

When the Court has talked about governmental initiatives like this one before, it has usually used one of two general labels. In several cases, the Court has treated such initiatives as a limited public (or nonpublic) forum. See, e.g., Christian Legal Soc. Chapter of Univ. of Cal., Hastings College of Law v. Martinez , 561 U.S. 661, 669-670, 682, 130 S.Ct. 2971, 177 L.Ed.2d 838 (2010) (" 'Registered Student Organization' " program providing various financial and nonfinancial benefits to recognized law-school student groups); Rosenberger , 515 U.S. at 823-824, 829-830, 115 S.Ct. 2510 ("Student Activities Fund" for registered campus student groups); Cornelius v. NAACP Legal Defense & Ed. Fund, Inc. , 473 U.S. 788, 790-791, 799-801, 806, 105 S.Ct. 3439, 87 L.Ed.2d 567 (1985) ("Combined Federal Campaign" literature enabling approved charitable organizations to solicit donations from federal employees). In other situations, the Court has discussed similar initiatives as government programs or subsidies. See, e.g., Legal Services Corporation v. Velazquez , 531 U.S. 533, 536, 543-544, 121 S.Ct. 1043, 149 L.Ed.2d 63 (2001) (government program distributing funds to legal-services organizations); National Endowment for Arts v. Finley , 524 U.S. 569, 573, 585-587, 118 S.Ct. 2168, 141 L.Ed.2d 500 (1998) (competitive government grant-making program to support the arts).10 In each of these situations, a governmental body established an initiative that supported some forms of expression without restricting others. Some speakers were better off, but no speakers were worse off.

Regardless of the finer distinctions between these labels, reasonable, viewpoint-neutral *2317content discrimination is generally permissible under either framework. See Christian Legal Soc. , 561 U.S. at 679, 130 S.Ct. 2971 ("Any access barrier must be reasonable and viewpoint neutral"); Velazquez , 531 U.S. at 543-544, 548-549, 121 S.Ct. 1043 (analogizing to limited-forum cases and explaining that "[w]here private speech is involved, even Congress' antecedent funding decision cannot be aimed at the suppression of ideas thought inimical to the Government's own interest"); see also Ysursa v. Pocatello Ed. Assn. , 555 U.S. 353, 355, 129 S.Ct. 1093, 172 L.Ed.2d 770 (2009) (finding government conduct that did not restrict speech but simply "decline[d] to promote" it valid where it was "reasonable in light of the State's interest"). Perhaps for that reason, the Court has often discussed the two frameworks as at least closely related. See, e.g., Christian Legal Society , 561 U.S. at 682, 130 S.Ct. 2971 ("[T]his case fits comfortably within the limited-public forum category, for [the plaintiff], in seeking what is effectively a state subsidy, faces only indirect pressure ... "); Velazquez , 531 U.S. at 544, 121 S.Ct. 1043 ("As this suit involves a subsidy, limited forum cases ... may not be controlling in a strict sense, yet they do provide some instruction").

Whichever label one chooses here, the federal system of trademark registration fits: It is, in essence, an opportunity to include one's trademark on a list and thereby secure the ancillary benefits that come with registration.11 Just as in the limited-forum and government-program cases, some speakers benefit, but no speakers are harmed. Brunetti, for example, can use, own, and enforce his mark regardless of whether it has been registered. Whether he may register his mark can therefore turn on reasonable, viewpoint-neutral content regulations.12

C

Prohibiting the registration of obscene, profane, or vulgar marks qualifies as reasonable, viewpoint-neutral, content-based regulation. Apart from any interest in regulating commerce itself, the Government has an interest in not promoting certain kinds of speech, whether because such speech could be perceived as suggesting governmental favoritism or simply because the Government does not wish to involve itself with that kind of speech. See, e.g., Ysursa , 555 U.S. at 359-360, 129 S.Ct. 1093 ; Cornelius , 473 U.S. at 809, 105 S.Ct. 3439. While "there is no evidence that the public associates the contents of trademarks with the Federal Government," Tam , 582 U. S., at ----, 137 S.Ct., at 1760, registration nevertheless entails Government involvement in promoting a particular mark. Registration requires the Government to publish the mark, as well as to take steps to combat international infringement. See 15 U.S.C. §§ 1062, 1124 ; see also Brief for United States 35. The Government has a reasonable interest in refraining from lending its ancillary support to marks that are obscene, vulgar, or profane. Cf. Hustler Magazine, Inc. v. Falwell , 485 U.S. 46, 56, 108 S.Ct. 876, 99 L.Ed.2d 41 (1988) ("[S]peech that is vulgar, *2318offensive, and shocking is not entitled to absolute constitutional protection under all circumstances" (internal quotation marks omitted)).

III

"The cardinal principle of statutory construction is to save and not to destroy." Jones & Laughlin Steel Corp. , 301 U.S. at 30, 57 S.Ct. 615 ; see also Hooper v. California , 155 U.S. 648, 657, 15 S.Ct. 207, 39 L.Ed. 297 (1895) ("The elementary rule is that every reasonable construction must be resorted to, in order to save a statute from unconstitutionality"). In directing the PTO to deny the ancillary benefit of registration to trademarks featuring "scandalous" content, Congress used a word that is susceptible of different meanings. The majority's reading would render the provision unconstitutional; mine would save it. Under these circumstances, the Court ought to adopt the narrower construction, rather than permit a rush to register trademarks for even the most viscerally offensive words and images that one can imagine.13

That said, I emphasize that Brunetti's challenge is a facial one. That means that he must show that " 'a substantial number of [the scandalous-marks provision's] applications are unconstitutional, judged in relation to the [provision's] plainly legitimate sweep.' " United States v. Stevens , 559 U.S. 460, 473, 130 S.Ct. 1577, 176 L.Ed.2d 435 (2010). With "scandalous" narrowed to reach only obscene, profane, and vulgar content, the provision would not be overly broad. Cf. Frisby , 487 U.S. at 488, 108 S.Ct. 2495 (rejecting a facial challenge after adopting a limiting construction); Boos , 485 U.S. at 331, 108 S.Ct. 1157 (same). Even so, hard cases would remain, and I would expect courts to take seriously as-applied challenges demonstrating a danger that the provision had been used to restrict speech based on the views expressed rather than the mode of expression.14 Cf. Finley , 524 U.S. at 587, 118 S.Ct. 2168 (reserving the possibility of as-applied challenges).

Freedom of speech is a cornerstone of our society, and the First Amendment protects Brunetti's right to use words like the one at issue here. The Government need not, however, be forced to confer on Brunetti's trademark (and some more extreme) the ancillary benefit of trademark registration, when "scandalous" in § 1052(a) can reasonably be read to bar the registration of only those marks that are obscene, vulgar, or profane. Though I concur as to the unconstitutionality of the term "immoral" in § 1052(a), I respectfully dissent as to the term "scandalous" in the same statute and would instead uphold it under the narrow construction discussed here.