We thank audiences at Syracuse University and the University of Oslo, an anonymous referee from Canadian Journal of Law and Jurisprudence, as well as Ken Baynes, Fred Schauer, and Brian Weatherson for helpful discussions about this material.

1. For an excellent discussion of this issue, see Schauer, Frederick, “Speech and ‘Speech’—Obscenity and ‘Obscenity’: An Exercise in the Interpretation of Constitutional Language” (1979) 67 Geo. L.J. 899.Google Scholar

2. We focus on the First Amendment partly because the theorists we consider in this paper do the same. Nevertheless, we think that the approach we describe here is the best way to understand other free speech principles as well. Though the free speech principle contained in the First Amendment is peculiar in many ways, we don’t think that it is peculiar in this respect.

3. See Lawrence III, Charles R., “Crossburning and the Sound of Silence: Antisubordination and the First Amendment” (1992) 37 Villanova L. Rev. 787 Google Scholar [Crossburning] and Lawren III, Charles R., “If He Hollers, Let Him Go: Regulating Racist Speech on Campus” in Matsuda, M., Lawrence III, C.R., Delgado, R & Crenshaw, K., eds, Words that Wound: Critical Race Theory, Assaulti Speech, and the First Amendment (Boulder, CO: Westview Press, 1993) 53 Google Scholar [If He Hollers].

4. There are a few exceptions. See, e.g., Braddon-Mitchell, David & West, Caroline, “What is Fr Speech?” (2004) 12 J. Pol. Phil. 437.CrossRefGoogle Scholar

5. Schauer, Frederick, Free Speech: A Philosophical Inquiry (Cambridge: Cambridge University Press, 1982) at 89–92.Google Scholar

6. For a survey of some difficulties in interpreting strict scrutiny, see Volokh, Eugene, “Freedom of Speech, Permissible Tailoring, and Transcending Strict Scrutiny” (1996) 144 U. Pa L. Rev. 2417.CrossRefGoogle Scholar

7. If, in addition, the harms caused by the covered speech (or action) clearly outweigh the harms that would be caused (to our commitment to free speech, for example) by regulating it, then that is some reason to think that the state has a compelling interest in regulating the speech (or action). That is, such balancing of harms is some evidence of a compelling interest, though more may be required to establish this as well.

8. By treating all covered speech as subject to the same raised standard of justification, we are simplifying a little. In the U.S., the courts have occasionally recognized intermediate standards of justification. With these intermediate levels, the picture becomes this. Though regulations of all covered speech (and action) must meet raised standards of justification, there are different raised standards for different categories of covered speech (and action). So-called ‘low value’ speech (e.g., some commercial speech) must meet the intermediate standards, while so-called ‘high value’ speech (e.g., political speech) must meet the highest standards, i.e., strict scrutiny. For a recent discussion of the distinction between high- and low-value speech, see Brink, David O., “Millian Principles, Freedom of Ex Pression, and Hate Speech” (2001) 7 Legal Theory 119.CrossRefGoogle Scholar

For other versions of a two-tiered approach, see also Sunstein, Cass R., Democracy and the Problem of Free Speech (New York: The Free Press, 1993)Google Scholar and Cohen, Joshua, “Freedom of Ex Pression” (1993) 22 Phil. & Pub. Affairs 207 Google Scholar. These versions will be discussed in §§ 4.2 and 4.3 respectively.

The possibility of intermediate standards will not matter for our purposes in this paper. Therefore, when we talk about the First Amendment, we will talk about the raised standards of justification relevant to its free speech principle, while remaining neutral about whether these raised standards include just strict scrutiny, or the intermediate standards as well.

9. To say that such speech (or action) is not regulable is to say that proposed regulations have not met (and perhaps could not meet) the relevant raised standards of justification.

10. Thus far, we have illustrated our approach with respect to the First Amendment. But the approach can be illuminating (and should be adopted) when thinking about other free speech systems as well. For a further illustration, consider the Canadian system. To determine whether a proposed regulation violates the free speech guarantee contained in the 1982 Charter of Rights and Freedoms, the Canadian courts must engage in a two-step inquiry. First, they must ask whether the regulation in question infringes section 2(b) of the Charter, which contains the free speech guarantee. Second, if the first question is answered affirmatively, the courts must determine whether the infringement can be justified in the manner required by section 1 of the Charter, which explicitly allows some limitations on the fundamental rights guaranteed in the Charter. See Mahoney, Kathleen, “The Canadian Constitutional Approach to Freedom of Ex Pression in Hate Propaganda and Pornography” (1992) 55 Law & Contemp. Probs. 77 at 78CrossRefGoogle Scholar and Sumner, L.W, The Hateful and the Obscene: Studies in the Limits of Free Ex Pression (Toronto, ON: University of Toronto Press, 2004) at 56.CrossRefGoogle Scholar

For a proposed regulation to survive the second step of the inquiry, it must pass the Oakes test (laid out by the 1986 Canadian Supreme Court decision in R. v. Oakes), which fleshes out the requirements of section 1. To pass this test, it must be shown that, first, the regulation serves a “ Pressing and substantial” purpose, and second, the limitations it imposes are proportional to the purpose to be achieved (Mahoney at 84-85).

All of this can be neatly rendered within the approach we’ve been considering in this section. The first of the two steps mentioned above corresponds to determining coverage. If the proposed regulation does not infringe section 2(b), then the actions it seeks to regulate are not covered by the free speech guarantee. The second step corresponds to determining protection. If the actions that are to be regulated are covered then the proposed regulation must meet a higher standard of justification than it would have had to meet otherwise. That higher standard is given by the Oakes test described above.

Although our approach thus fits the Canadian system as well as the U.S. system, it is important to note that there are significant differences between the two systems. For one thing, compared to the U.S., coverage is more extensive, though protection more limited (Sumner at 56).

11. On some interpretations, Mill offers an argument from knowledge, rather than truth. That difference in interpretation does not affect the point being made here.

12. Answers to the Broad Normative Coverage Question may turn out to be context-relative, for it may be that the free speech principle that we ought to have in one context is quite different from the free speech principle we ought to have in another context.

13. For a very useful survey of justifications, see Greenawalt, Kent, Speech, Crime, and the Uses of Language (New York: Oxford University Press, 1989) at 9–39.Google Scholar

14. Austin, J.L., How to Do Things with Words, 2nd ed., ed. by Urmson, J.O. & Sbisá, M. (Cambridge, MA: Harvard University Press, 1975) at 98–132.CrossRefGoogle Scholar

15. This commitment to a prima facie pluralism about obligations is perfectly compatible with all obligations, in the final analysis, turning out to be of one sort.

16. Again, the obligations in question here seem to be neither legal nor moral. Etiquettial obligations are plausibly one kind of (non-moral, non-legal) social obligation, but not the only kind.

Note that these particular obligations would not come into being if the rules of politeness were sufficiently different. But that’s not to say that the obligations are due entirely to those rules. If the rules of politeness remained exactly as they are, but (in the first example) the first neighbor made no remark to the second, the latter would have no obligation to respond. Thus, the second neighbor’s obligation is partly due to the first neighbor’s utterance, and partly due to the rules.

17. See Maita, Ishani & McGowan, Mary Kate, “The Limits of Free Speech: Pornography and the Question of Coverage” (2007) 13 Legal Theory 41.CrossRefGoogle Scholar

18. This challenge is based on an example that is much like Greenawalt’s flash flood example. See Greenawalt, supra note 13 at 61-62.

19. To make the case for legal liability even clearer, we might imagine that the patient being transported will die unless she has surgery within the next thirty minutes, and the radio broadcast announces that the Interstate will be closed for the next hour. In such a case, the EMTs may be liable even if they would be shielded from liability in other less severe cases.

20. Greenawalt, supra note 13 at 40-76.

21. Greenawalt’s notion of substantiality, unlike our notion of significance, is at least partly concerned with the degree of change in obligations effected by the utterance. For a more complete discussion of the relationship between Greenawalt’s view and ours, see Maita & McGowan, supra note 17.

22. At different points, he suggests that a situation-altering utterance is dominantly situation-altering if: (i) the speaker’s primary illocutionary intention is to enact changes in someone’s obligations; (ii) the standard illocutionary purpose of the words used is to enact changes in someone’s obligations; (iii) the speaker’s primary perlocutionary intention is to cause changes in someone’s obligations; or, (iv) the speaker’s direct purpose is to enact changes in someone’s obligations. See Greenawalt, supra note 13 at 57-71.

23. Maitra & McGowan, supra note 17 at 56-59.

24. Greenawalt’s focus, unlike ours, is largely on the criminal law.

25. Sunstein, supra note 8 at 17-51.

26. Ibid. at 123. As will become clear in what follows, some of the theorists we discuss in this section, including Sunstein, use ‘protected’ and ‘protection’ to mean what we do by ‘covered’ and ‘coverage,’ respectively. On their usage, speech is “protected” in virtue of falling within the scope of a principle of free speech.

27. Ibid. at 124. According to Sunstein, such lower-tier speech “may be regulated only on the basis of a persuasive demonstration that a strong and legitimate government interest is promoted by the regulation at issue” (123).

28. Ibid. at 124.

29. Even if this translation is correct, however, it is worth noting that Sunstein disagrees with us about which categories of speech fall within and outside the coverage boundaries. For example, he believes that criminal solicitation is lower-tier (and so, covered) speech, whereas we think that it is uncovered.

30. Cohen, supra note 8 at 213-50.

31. Ibid. at 209-10. See also note 26.

32. Ibid. at 220.

33. Ibid. at 214.

34. Though we cannot discuss these assumptions in any detail in this paper, it is worth noting that they are not uncontroversial. For example, if one of the costs of some speech is that it disables (i.e., silences) others’ speech, then it seems less likely that the costly speech can be fought with “more speech.” Such silencing and its relevance to free speech has been extensively discussed in the recent literature. See, e.g., Langton, Rae, “Speech Acts and Unspeakable Acts” (1993) 22 Phil. & Pub. Affairs 293 Google Scholar; Hornsby, Jennifer, “Disempowered Speech” (1995) 23 Phil. Topics 127 CrossRefGoogle Scholar; West, Caroline, “The Free Speech Argument Against Pornography” (2003) 33 Can. J. Phil. 391 CrossRefGoogle Scholar; Maitra, Ishani, “Silencing Speech” (2009) 39 Can. J. Phil. 309 CrossRefGoogle Scholar; McGowan, Mary Kate, “On Silencing and Sexual Refusal” (2009) 17 J. Pol. Phil. 487.CrossRefGoogle Scholar

35. Cohen, supra note 8 at 213.

36. Hart, H.L.A., The Concept of Law, 2nd ed., Bulloch, P.A. & Raz, J., eds. (Oxford: Clarendon Press, 1994) at 82–83.Google Scholar

37. There is some tension in Sunstein’s view between his answer to the Justificatory Question (regarding what makes speech valuable in the first place) and his characterization of political speech. According to Sunstein, speech is valuable insofar as it contributes to a deliberative democratic process. But not all political speech in Sunstein’s sense contributes to this process. Although the Church’s statement clearly satisfies his definition of political speech, it is nevertheless anti-Madisonian in virtue of being instructive rather than deliberative. After all, in making the statement, the Church does not mean to be offering considerations to be used as one deliberates further on the question of abortion. Rather, the Church intends to dictate that individuals think (and vote) a particular way. It is just this dictatorial force that renders the Church’s statement both anti-Madisonian and obligation-enacting.

38. On the other hand, as mentioned in note 37, there is a sense in which the Church’s statement is instructive rather than deliberative. As such, it is not intended to play a role in further deliberations (on the parts of Catholics, or anyone else) about what is right to do. This is some reason to think that, even if there is a connection between speech like this and the deliberative interest, the connection is not after all a close one.

39. See discussion in note 22.

40. We believe that a parallel case could be made for pornography. See, e.g., MacKinnon, Catharine A., Feminism Unmodified: Discourses on Life and Law (Cambridge, MA: Harvard University Press, 1987)Google Scholar; MacKinnon, Catharine A., Only Words (Cambridge, MA: Harvard University Press, 1987)Google Scholar; Langton, supra note 34; McGowan, Mary Kate, “Conversational Exercitives and the Force of Pornography” (2003) 31 Phil. & Pub. Affairs 155 CrossRefGoogle Scholar; Maitra & McGowan, supra note 17.

41. Delgado, Richard, “Words that Wound: A Tort Action for Racial Insults, Epithets, and Name Calling” in Matsuda, M., III, CR Lawrence, Delgado, R & Crenshaw, K., eds., Words that Wound: Critical Race Theory, Assaultive Speech, and the First Amendment (Boulder, CO: Westview Press, 1993) 89 at 93.Google Scholar

42. Ibid. at 95.

43. Both Schauer and Greenawalt are exceptions here, for they both take (at least some) fighting words to be uncovered. Schauer acknowledges that fighting words are not a clear case. See Schauer, Frederick, “The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience” (2004) 117 Harv. L. Rev. 1765 at 1777CrossRefGoogle Scholar. We discuss Greenawalt’s position on fighting words later in this section.

44. Chaplinsky v. New Hampshire [1942] 315 U.S. 568 at 571-72Google Scholar [emphasis added].

45. There are complications here. Consider, for example, racial slurs directed at an elderly black woman in the presence of only strong young white men. Although in this case the addressee is unlikely to respond with violence, Greenawalt argues that such words would cause an immediate breach of the peace under different “equalized” conditions. As a result, on his view, this utterance counts as fighting words even though it does not actually cause an immediate breach of the peace. See Greenawalt, Kent, Fighting Words: Individuals, Communities, and Liberties of Speech (Princeton, NJ: Princeton University Press, 1995) at 47–70.Google Scholar

46. Although there is precedent for subject-based discriminations amongst regulable utterances (e.g., threats against the President are punished more severely than other threats), there is a worry that doing so here would not be viewpoint-neutral. After all, one might think that only racist race-related utterances would cause an immediate breach of the peace, and therefore qualify as fighting words in the first place. If so, the singling out of race-related fighting words may be (tacitly) viewpoint-based, and hence unconstitutional. This is essentially the concern ex Pressed by the U.S. Supreme Court in R.A.V. v. City of St. Paul. See R.A.V. v. City of St. Paul [1992] 505 U.S. 377 at 391Google Scholar. Both Sunstein’s and Cohen’s preferred principles of free speech depart from the Supreme Court’s view on this matter.

47. Greenawalt does not favor legislatively singling out race-related fighting words. Since the U.S. courts are likely to take issue with the Constitutionality of any such singling out, Greenawalt favors using the fighting words doctrine to regulate all fighting words, not merely the race-related sub-class.

48. As noted above, however, Greenawalt acknowledges that at least some regulable racist fighting words ought to be uncovered. However, he does not consider the possibility (as we do) that such utterances can constitute acts of racial discrimination. Moreover, as we have argued elsewhere, there are other difficulties with his account, especially having to do with his notion of dominance. See Maitra & McGowan, supra note 17.

49. As noted above, Greenawalt is an exception here, for he treats some racist fighting words as uncovered.

50. See Crossburning, supra note 3. In addition, Matsuda can also be read this way. See Matsuda, Mari “Public Response to Racist Speech: Considering the Victim’s Story” in Matsuda, M., Lawrence, C.R. III, Delgado, R & Crenshaw, K., eds., Words that Wound: Critical Race Theory, Assaultive Speech, and the First Amendment (Boulder, CO: Westview Press, 1993) 17 Google Scholar. Although her argument appears to us to be both causal and constitutive, both Lederer and Butler interpret her as maintaining that the speech in question constitutes an act of subordination, and thus as offering (in our terms) a constitutive argument. See Lederer, Laura, “Pornography and Racist Speech as Hate Propaganda” in Lederer, L. & Delgado, R., eds., The Price We Pay: The Case Against Racist Speech, Hate Propaganda, and Pornography (New York: Hill & Wang, 1993) 131 at 131Google Scholar and Butler, Judith, Excitable Speech: A Politics of the Performative (New York: Routledge, 1997) at 73.Google Scholar

51. For this argument, see Crossburning, supra note 3.

52. Unlike Lawrence, we don’t think that U.S. anti-segregation law provides a precedent for content-based regulation of speech. Since the speech in question is regulated because of what it does (and not because of what it says), this is not a content-based regulation. This is so even if what it does depends, in some way, on what it says.

53. If He Hollers, supra note 3 at 59 [original emphasis].

54. Ibid.

55. One might be concerned that racist hate speech cannot enact a(n unjust racial) hierarchy because doing so requires that the speakers have and be exercising the authority to do so. We will discuss this concern in the next section (§ 5.5).

56. There may be some tension in Lawrence’s view on this point. Lawrence also likens (some) racist hate speech to ‘Whites Only’ signs. Since such signs are not speech in the First Amendment sense, Lawrence may also be suggesting that (some) racist hate speech should not even be covered by the First Amendment. See § 3.3.

57. See, e.g., Butler, supra note 50; Green, Leslie, “Pornographizing, Subordinating, and Silencing” in Post, R.C., ed., Censorship and Silencing: Practices of Cultural Regulation (Los Angeles, CA: Getty Research Institute for the History of Art and the Humanities, 1998) 292–97Google Scholar; Golding, Martin P., Free Speech on Campus (Lanham, MD: Rowman & Littlefield, 2000)Google Scholar; Sumner, supra note 10; Bauer, Nancy, “How To Do Things with Pornography” in Shieh, S. & Crary, A., eds., Reading Cavell (London: Routledge, 2006) 68 Google Scholar. Several of these theorists address this concern (about how apparently non-authoritative speakers can construct unjust social hierarchies) in the context of discussions of MacKinnon’s claims about pornography. See MacKinnon, supra note 40.

58. Langton, Rae, “Subordination, Silence, and Pornography’s Authority” in Post, R.C., ed., Censorship and Silencing: Practices of Cultural Regulation (Los Angeles, CA: Getty Research Institute for the History of Art and the Humanities, 1998) 261 Google Scholar and Langton, Rae, Sexual Solipsism: Philosophical Essays on Pornography and Objectification (Oxford: Oxford University Press, 2009) at 89–116.CrossRefGoogle Scholar

59. Matsuda, supra note 50.

60. See McGowan, supra note 40 and Wieland, Nellie, “Linguistic Authority and Convention in a Speech Act Analysis of Pornography” (2007) 85 Australasian J. Phil. 435.CrossRefGoogle Scholar

61. According to this reasoning, the causal argument succeeds in showing that (some) racist hate speech may be regulated only once it is established that such speech ought to be uncovered (and so, that proposed regulations need meet no standard higher than rational basis review). But showing that racist hate speech ought to be uncovered required a (constitutive) argument to the effect that racist hate speech is significantly obligation-enacting, and hence ought not be covered. Thus, on this view, the success of the causal argument requires the prior success of a constitutive argument.