INTRODUCTION
Liberal societies such as ours are deeply committed to both liberty and equality. At the same time, we are also very familiar with cases in which these two commitments seem to come into conflict. For example, a private club may want the freedom to exclude nonwhite members, but such a practice violates our commitment to equality. Similarly, legally prohibiting the use of racial pejoratives in workplaces may enhance workplace equality, but such a strategy infringes upon our commitment to liberty. In these and many related cases, it seems that we are required to balance our commitment to one value against our commitment to the other.
Many controversial freedom-of-speech issues are commonly framed as involving just such a clash between our twin commitments to liberty and equality. Consider, for example, the case of pornography. Some feminists have argued that the production and consumption of pornography harms women in ways that a society committed to gender equality should seek to prevent. In response it has been said that regulating the production and distribution of pornography infringes upon the right of producers, distributors, and consumers of pornography to free speech. If this is right, then we have here a clash between our commitment to a principle of free speech—surely one of the fundamental political principles in liberal societies—and our commitment to gender equality. A similar case can be made for racist hate speech, as involving a clash between our commitments to the free-speech principle and to racial equality. To resolve such conflicts, it seems that we must decide which of these commitments should take precedence over the other. We must choose, that is, between liberty and equality.
In this paper, we argue that at least in the case of pornography, this apparent conflict may be due to an incorrect framing of the issues. Although we as a society are deeply committed to a principle of free speech, we tend to neglect at least one question that is crucial to understanding the nature of this commitment: What should count as speech in the relevant sense? That is, insofar as the principle of free speech is enshrined in the First Amendment, which actions should be taken to fall within the scope of that amendment?1
Although we confine our remarks in this paper to the First Amendment in particular, much of what we say applies just as well to any reasonable principle of free speech. Nevertheless, we focus here specifically on the First Amendment for two reasons. First, doing so allows us to stress that our central thesis (about the scope of the First Amendment) squares with current American free-speech law. Second, it also allows us to show that the changes for which we are (conditionally) arguing are therefore required by the law's own lights. These ideas are spelled out in Section III.
The paper proceeds as follows. In the next section, Section II, we introduce two approaches to thinking about free speech. On the first (inclusive) approach, all speech in the ordinary sense is considered speech in the technical sense relevant to the First Amendment. On the second (narrow) approach, developed by Frederick Schauer, among others, ‘speech’ for the purposes of the First Amendment does not include everything that we would normally consider speech.2
F. SCHAUER, FREE SPEECH: A PHILOSOPHICAL ENQUIRY (1982); Schauer, The Boundaries of the First Amendment: A Preliminary Exploration of Constitutional Salience, 117 HARV. L. REV. 1765–1809 (2004).
K. GREENAWALT, SPEECH, CRIME, and the USES of LANGUAGE (1989).
TWO APPROACHES
It is generally agreed that our commitment to free speech involves extending to speech special protections that we do not extend to other actions. At the same time, it is also generally supposed that this commitment is compatible with regulating certain kinds of speech. This raises the following questions. First, for any given kind of speech, what has to be established to show that our commitment to free speech (as enshrined in the First Amendment) is compatible with regulation of speech of that kind? That is to say, to what standards of scrutiny should proposed regulations be subject?5
Strictly speaking, to show that some particular kind of speech is regulable, the justifications for the proposed regulations must be shown to survive the relevant standards of scrutiny. However, for ease of exposition, we shall talk of the regulations themselves meeting the relevant standards of scrutiny rather than of the justifications for the regulations meeting those standards.
A. The Inclusive Approach
According to the first approach, which is more prevalent in the philosophical literature, the First Amendment affords special protections to everything that would ordinarily be considered speech (and to some other actions besides).6
These ‘other actions’ include wearing a T-shirt with a picture of Adolf Hitler on it surrounded by a red circle with a slash through it; burning an American flag; and burning a cross in someone's front yard. None of these actions is speech in the ordinary sense.
J. Feinberg, Limits to the Free Expression of Opinion, in PHILOSOPHY of LAW (J. Feinberg & H. Gross eds., 1975).
One common rationale for marking a kind of speech as an exception in the above sense is the putative fact that the harms resulting from regulating that kind of speech are outweighed by the harms that would result from failing to regulate it. (This, in fact, is the rationale that Feinberg employs in explaining why the five categories mentioned above are unprotected.) To fix ideas, let us consider a particular kind of speech, namely, defamation (i.e., speech that damages someone's reputation). According to the approach we are currently considering, defamation can be regulated only if the balancing described above works out right, that is, only if damage to reputation, along with other harms such damage causes, outweigh the harms done by restricting such speech.
This balancing constitutes a significantly higher standard of scrutiny for proposed regulations of defamation (and other kinds of speech) as compared to the rather minimal scrutiny that regulations of much nonspeech action have to meet.8
See note 5.
B. The Narrow Approach
According to the second approach, which is found most often in the legal literature, the First Amendment does not extend special protections to everything that we would ordinarily consider speech.9
This second approach has been occasionally, though rarely, mentioned in the philosophical literature. For example, see D. Braddon-Mitchell & C. West, What Is Free Speech? 12 J. POL. PHIL. 437–460 (2004).
Proposed regulations of speech that falls outside the scope of the First Amendment are subject to a relatively low standard of scrutiny, such as rational-basis review. In order for a regulation to survive rational-basis review, it need only be established that, first, the state has a legitimate interest in regulating what it proposes to regulate, and second, the regulation bears a rational relation to that interest. Hence rational-basis review is emphatically not a demanding standard. Following Frederick Schauer, we shall say that speech (and nonspeech action) that is subject to such minimal standards of scrutiny is uncovered by the First Amendment.10
SCHAUER, supra note 2, at 89–92.
Of course, even on this second approach, much ordinary speech (and some nonspeech action) does fall within the scope of the First Amendment. Such speech (and other actions) is covered by that amendment.11
For discussion of these ‘other actions,’ see note 6.
For a survey of some difficulties in interpreting strict scrutiny, see E. Volokh, Freedom of Speech, Permissible Tailoring, and Transcending Strict Scrutiny, 144 U. PA. L. REV. 2417–2461 (1996).
Compelling state interest and narrow tailoring constitute a significantly raised standard of scrutiny as compared to legitimate state interest and rational relation. Thus strict scrutiny is a significantly higher standard of scrutiny than rational-basis review. Where a proposed regulation (of covered speech or action) survives this raised standard of scrutiny, we shall say, again following Schauer, that although the speech (or action) is covered, it is unprotected by the First Amendment.13
By treating everything within the scope of the First Amendment as subject to strict scrutiny and everything outside that scope as subject to rational-basis review, we are simplifying a little. The courts have occasionally recognized intermediate standards of scrutiny between strict scrutiny and rational-basis review. With these intermediate levels, the picture becomes this. Though regulations of all covered speech (and action) must meet raised standards of scrutiny, there are different raised standards for different categories of coverage. So-called ‘low value’ speech and action (e.g., some commercial speech) must meet the intermediate standards of scrutiny, while so-called ‘high value’ speech and action (e.g., much political speech and action) must meet strict scrutiny. For a recent discussion of the distinction between high- and low-value speech, see D. O. Brink, Millian Principles, Freedom of Expression, and Hate Speech, 7 LEGAL THEORY 119–157 (2001).
The differences between strict scrutiny and the intermediate standards do not matter for our purposes in this paper. Therefore we talk about the raised standards of scrutiny relevant to the First Amendment while remaining neutral about whether these raised standards include just strict scrutiny or the intermediate standards as well.
Finally, on this second approach, some speech (and some nonspeech action) is both covered and protected by the First Amendment. That is to say, it falls within the scope of the First Amendment, so that proposed regulations are subject to raised standards of scrutiny. But because these regulations do not meet these raised standards, the speech (and action) is not regulable. Much political speech and action is included in this category.
C. The Two Approaches Compared
It is worth emphasizing that the two approaches described above are genuinely different. For our purposes, the main difference can be described as follows. According to the first (inclusive) approach, everything that we would ordinarily consider speech falls within the scope of the First Amendment but some of it (the ‘exceptions’) may be regulated. By contrast, according to the second (narrow) approach, some of what we would ordinarily consider speech is uncovered (i.e., beyond the scope of the First Amendment), some covered but still unprotected (and thus regulable), and some both covered and protected (and thus not regulable).
To see that this difference is not merely terminological, note the following. The narrow approach, but not the inclusive approach, makes room for the possibility that some actions that are speech in the ordinary sense do not deserve the special protections the First Amendment provides. This is an important possibility to leave open for at least two reasons. First, intuitively speaking, it is hard to see why regulating certain categories of speech—speech that creates contracts, for example—should raise any First Amendment concerns at all. After all, one should not be able to launch a (successful) free-speech defense whenever one breaks a contract!
Second, if it is supposed (as many do) that the justifications for affording speech special protections succeed only if the term ‘speech’ is understood in a special, nonordinary sense, then a principle of free speech is justified only insofar as it covers speech in this technical sense. For instance, it is often pointed out that Mill's argument from truth does not seem to encompass everything that is ordinarily considered speech simply because some speech in the ordinary sense is not conducive to the discovery of truth.14
On some interpretations, Mill offers an argument from knowledge rather than truth. That difference in interpretation does not affect the point being made here. J. S. MILL, ON LIBERTY (1978).
Finally, the narrow approach can also explain otherwise curious facts about our current legal system. For example, consider the fact that regulations of much speech in the ordinary sense consistently fail to raise First Amendment flags. To mention just one case, speech that constitutes insider trading is heavily regulated, but such regulation is rarely, if ever, challenged on First Amendment grounds. On the inclusive approach, this is just surprising. On the narrow approach, by contrast, this fact can be explained by noting that the speech in question is generally considered uncovered and so does not count as speech in the technical sense relevant to the First Amendment.
In the remainder of this paper, we adopt the narrow approach. That is, we conclude that for the purposes of the First Amendment, ‘speech’ should get a special, technical sense different from its ordinary sense. But that immediately raises the following question: What should this special technical sense be? That is to say, what should count as speech in the First Amendment sense? As we shall see in the remainder of this section, there are at least two different ways to take this question.
D. Normative Coverage Questions
First, and perhaps most obviously, there is the broad normative coverage question: What ought to be covered, according to the free-speech principle we ought to have? This question is doubly normative. To answer it, we first have to settle what would be the best free-speech principle for us and then see which actions properly fall within the scope of that principle.15
For a more complete discussion of the difficulties that arise in addressing this question, see our ‘On Racist Hate Speech and the Scope of a Free Speech Principle,’ unpublished manuscript.
A narrower but still clearly normative question is this: What ought to be covered, given the free-speech principle we actually have? This is the narrow normative coverage question. There are several ways of approaching this question. One way is to look at the principles underlying judicial opinions in First Amendment cases and extrapolate from those the free-speech principle underlying the current system. Once this principle has been identified, we can work out which actions fall within its scope.
A different way to approach the same question is to begin with what we might call the core cases of coverage (and noncoverage), that is, the cases that are most uncontroversially covered (and the ones that are most uncontroversially uncovered) under current First Amendment law. Certain political utterances (e.g., saying to a friend, ‘The United States has become an international laughing stock under the current president’) might belong in the former category; speech that constitutes criminal solicitation (e.g., saying to a known assassin, ‘I hereby hire you to kill the provost’) belongs in the latter category. Given these core cases, we can extrapolate the free-speech principle underlying the current system. Again, once this principle has been identified, we can work out which actions fall within its scope.
In what follows, we focus on the narrow normative coverage question, adopting the second of the approaches described above. In the next section, Section III, we offer a thesis that correctly predicts the coverage status of several kinds of speech that are uncontroversially uncovered under current First Amendment law. Then, later in the paper, in Section V, we apply this thesis to a controversial case, namely, pornography. There, we argue that if certain hypotheses about what pornography does are correct, (some) pornography ought to be uncovered by the First Amendment.
A PARTIAL ANSWER TO THE NARROW NORMATIVE COVERAGE QUESTION
In this section, we offer our partial answer to the narrow normative coverage question. We argue that if an utterance enacts a change in a significant obligation—that is, if it is a significantly obligation-enacting utterance, in a sense to be explained below—then it should not fall within the scope of the First Amendment. We defend this thesis by showing that it makes the right predictions about several core cases (in the sense discussed in Sec. II.D) of uncovered speech. Note that this is but a partial answer to the narrow normative coverage question, for it says nothing about the status of utterances that do not enact changes in significant obligations. We leave open the question of whether there are other sufficient conditions for noncoverage.
A. Obligation-Enacting Utterances
To introduce the category of obligation-enacting utterances, it will be useful to begin with some examples. Donald Trump says, ‘You're fired,’ to an employee. Paul says, ‘I'll give you ten dollars to wash my car,’ to his nephew. The karate instructor says, ‘Now kick me in the torso, as hard and fast as you can,’ to a trainee. In each of these cases, the speaker performs an action merely by uttering the right words under the right circumstances. The utterance constitutes the action in question. (In J. L. Austin's terminology, the action constituted by the utterance is an illocutionary act.)16
J. L. AUSTIN, HOW TO DO THINGS WITH WORDS (2d ed., J. O. Urmson & M. Sbisá, eds., 1975), at 98–132. Although we occasionally use Austin's terminology in this paper, we are not committed here to the ultimate defensibility of his distinction between illocution and perlocution. We use the terminology because it simplifies our exposition. Our argument does, however, rely on (and is committed to the defensibility of) the distinction between causing and constituting (enacting). Everything we say about illocution and perlocution can be restated using the latter notions.
Any firing changes obligations. In this case, the firing absolves the employee of obligations he had as an employee in Trump's corporation but creates on his part a new obligation to refrain from demanding a paycheck from that organization. Note that these changes in obligations are not merely caused by the firing. Rather, the firing itself constitutes an obligation-changing action. Thus Trump's utterance constitutes an action of a particular kind, namely, an obligation-changing (or, as we shall say, obligation-enacting) action. This makes it an obligation-enacting utterance in our sense. The remaining examples mentioned above (Paul's utterance, the karate instructor's utterance) are also obligation-enacting utterances for similar reasons.
Although the above examples involve the enacting of either legal or moral obligations, it is important to note that there are other sorts of obligations. That is, there are obligations that are neither legal nor moral. Consider, for example, the fact that we are all obliged to refrain from picking our respective noses in public places. This is a genuine social obligation, but it is neither a legal nor a moral one. Moreover, such social obligations can be (and often are) enacted verbally. In a society in which Miss Manners has the authority to enact rules of etiquette by making certain pronouncements, her saying, ‘One must not pick one's nose in public’ is an obligation-enacting utterance. The obligation enacted, though, is etiquettial (or social) and neither legal nor moral.17
One might suspect that such etiquettial obligations are really, at bottom, just weak moral obligations, because etiquette is concerned primarily with expressions of respect for others. Even if that is right, other (social) obligations do not appear to boil down in any simple manner to moral ones. Consider, for example, social norms regarding how males and females are to carry themselves or cultural norms regarding how much physical space to give a conversational partner. Such norms are clearly operative but they do not seem to be the least bit moral.
It is intuitively plausible that at least some obligation-enacting utterances should not be covered by the First Amendment. Suppose Keshav says to Cindy, ‘If you lend me five hundred dollars, I'll pay you back as soon as I sell my car.’ Cindy agrees to lend Keshav the money. Keshav's utterance is an obligation-enacting utterance, because it creates obligations on his part. Later, Keshav sells his car but does not pay Cindy. Although his obligation-enacting utterance involves merely the uttering of words, he cannot defend his behavior (and avoid paying Cindy) on free-speech grounds. Regulations enforcing such contracts do not (and ought not to) raise First Amendment concerns.18
We are assuming that Keshav's utterance enacts a contract.
This suggests a very simple answer to the narrow normative coverage question: perhaps no obligation-enacting utterances should be covered by the First Amendment. Although this simple answer is a good first pass, it cannot be correct as it stands. As we shall see, many, perhaps even most, ordinary assertions are also obligation-enacting utterances. If all such assertions were left uncovered, there would be nothing much left to our commitment to free speech.
B. Significance
Consider the following ordinary assertions. While shopping in the produce section, one shopper says to another, ‘Ooh, the strawberries look divine!’ While waiting in line at Home Depot, one home improver complains to another, ‘I could have installed this flooring by now.’ Each of these utterances creates an obligation on the part of the addressee to respond in some way (though they both leave open a broad range of acceptable responses). Accordingly, both assertions are obligation-enacting utterances. Nevertheless, it would be contrary to our general commitment to free speech to say that they ought to be uncovered.
In response to these concerns, we contend that only some obligation-enacting utterances ought to be outside the scope of the First Amendment, namely, those that enact changes in what we shall call significant obligations. After all, not all obligations are serious. Some are easily met (e.g., the obligation to indicate that you have heard what someone has just said to you), and many have no serious repercussions even if they go unmet (e.g., the obligation to say ‘Hello’ back to a passerby). By contrast, Trump's utterance (‘You're fired,’ said to an employee) is significantly obligation-enacting, since it frees the employee from significant obligations (i.e., those that accompany being an employee) and creates significant new obligations (e.g., the obligation to refrain from demanding a paycheck from Trump's corporation). Our thesis is that only significantly obligation-enacting utterances ought to be uncovered by the First Amendment.
Significance in our sense concerns the nature of the obligations affected, not the degree of change wrought by the utterance. As we saw above, whether an obligation is significant depends, among other things, on how difficult it is to meet and what the repercussions might be if it goes unmet (or, for that matter, is met). In what follows, we focus upon one kind of serious repercussion, namely, liability to legal action.
Some obligations are such that failures to meet them are actionable under the law. Many obligations between employers and employees, for example, have this status. Furthermore, other obligations are such that meeting them is actionable under the law. Obligations to perform criminal acts, for example, have this status. We shall say that obligations of each of these kinds—that is, obligations that are such that failures to meet them are actionable under the law, and ones that are such that meeting them is actionable under the law—are significant obligations. This gives us a pair of sufficient conditions for significance.
Note that all of the obligations mentioned above are of interest to the law regardless of whether speech is involved in enacting, respecting, or violating them. Failing to satisfy the obligation to refrain from harassment, for example, is actionable regardless of whether the harassment itself is verbal. Similarly, consider the obligation to refrain from discrimination. Again, failures to satisfy this obligation are actionable whether or not the discriminatory act is verbal.
We have not offered here a full account of what makes an obligation significant in our sense. Fortunately, a full account is not necessary. The sufficient conditions mentioned above will suffice for our purposes in this paper.
C. Our Thesis Applied to Core Cases
Our answer to the narrow normative coverage question is this: all significantly obligation-enacting utterances ought to be outside the scope of the First Amendment. In the remainder of this section, we argue that this thesis generates the right results with respect to several core cases of uncovered speech. In other words, it squares with current First Amendment law at least with respect to such cases.19
In addition to squaring with current law, we think that there is an independent rationale for our thesis. Briefly, here is the idea. Consider the attributes that make speech valuable in the first place and thus deserving of the special protections afforded by the First Amendment. There are many proposals about which attributes these are: for example, some have argued that speech is valuable because it contributes to the discovery of truth; others have thought that it is valuable because it is necessary for a functioning democracy; yet others have suggested that it is valuable because it is a constituent of genuine autonomy. Whatever the answer, we think that significantly obligation-enacting utterances do not have the relevant attributes (at least, not to any significant extent). That is why they should be excluded from the scope of the First Amendment.
Establishing this general rationale would require showing, for every reasonable proposal regarding what makes speech valuable, that significantly obligation-enacting utterances do not have the attributes favored on that proposal (or else establishing which proposal is correct). Either way, that is a huge undertaking, and as such, one that is beyond the scope of this paper.
Finally, there is also a rough but intuitive rationale for our thesis that is worth mentioning here. All speech (and for that matter, all action) both expresses or communicates something and makes other changes in the world. Nevertheless, there is a sense in which the point of some speech is primarily expressive/communicative, while the point of other speech is primarily to make changes in the world. (Much speech falls somewhere in between these two extremes.) Significantly obligation-enacting utterances belong at the latter extreme. Since it is the expressive/communicative function that makes (some) speech so valuable, significantly obligation-enacting utterances can be safely regarded as beyond the purview of even a robust principle of free speech.
Of course, this leaves open the possibility that our thesis disagrees with current First Amendment law regarding some non-core cases. But if our thesis gets the core cases right, then that is reason to think that the verdicts on those non-core cases should be different from what they in fact are, and further, that this is so by the law's own lights, that is, by the very free-speech principle that in fact underlies First Amendment law.
First, consider contracts. Since utterances that create, alter, or dissolve contracts thereby enact changes in obligations, they are obligation-enacting utterances. Further, since failures to satisfy the terms of contracts are actionable under the law, the obligations in question are significant, in our sense. Thus utterances that create, alter, or dissolve contracts are significantly obligation-enacting utterances and hence uncovered on our view. This result squares with current law. To see this, suppose that Marilyn signs a contract to photograph John's wedding for $500. If she failed to show, she would be liable. This is so even though mere words created the contract in question. Marilyn cannot legitimately defend her behavior on the grounds that she was merely exercising her right to free speech when she entered into the contract. The speech constituting the contract is not (and should not be) covered. Our thesis generates the correct result in this case.
Our thesis also squares with the law of criminal solicitation. Suppose that a disgruntled professor seriously and sincerely says, ‘I hereby hire you to kill the provost.’ Since this utterance creates an obligation on the part of the addressee to commit a criminal act, it is significantly obligation-enacting. As a result, on our view, it falls outside the scope of the First Amendment. This is exactly the right result. Utterances that constitute solicitation of a crime are themselves criminal. Punishing such utterances does not (and should not) raise any free-speech issues at all.21
The disgruntled professor's utterance is a fairly straightforward instance of criminal solicitation. Consider, by contrast, an utterance that merely encourages someone to commit a crime (without also creating an obligation for the addressee to commit it). Since no significant obligations are enacted by this utterance, our thesis is silent about its coverage status.
Next, consider antidiscrimination law. Suppose that the CEO of Macho Co. verbally enacts a new hiring policy when he says, ‘From now on, only men will grace these halls.’ In thus changing the company's hiring policy, the executive's utterance enacts significant changes in his employees' obligations. As a result, and according to our thesis, his utterance should not fall within the scope of the First Amendment. Thus, although his words are speech in the ordinary sense, they are not (and should not be) speech in the technical sense of the First Amendment. As far as the law is concerned, his utterance is an unlawful act of gender discrimination, and the First Amendment is simply irrelevant to its regulation.
Relatedly, our thesis also generates the right result in cases of workplace sexual harassment. Suppose that a boss seriously and sincerely says to an employee, ‘Sleep with me or you're fired.’ This quid pro quo demand is significantly obligation-enacting and so, according to our thesis, should not be covered by the First Amendment. This result also squares with current law. Such an utterance is clearly and uncontroversially illegal.
Finally, consider the ‘Whites Only’ signs that were once so common in segregated America. Since such signs make it impermissible for nonwhites fully to patronize the establishment in question, they are significantly obligation-enacting.22
In some cases, nonwhites were permitted to patronize the establishment, but only in highly limited ways, e.g., only to the extent of being allowed to order take-out food through a back window or be served in a separate and often inferior dining room.
Our thesis hence returns the right results—that is, results that accord with current First Amendment law—for many core cases of uncovered speech. Since it has proved descriptively accurate in this crucial way, it can be relied on to guide the treatment of more controversial cases (e.g., pornography).
In sum, in this section we have offered a partial answer to the narrow normative coverage question. Our central thesis is that any significantly obligation-enacting utterance ought to fall outside the scope of the First Amendment. We have argued that our thesis does not generate the unpalatable result that all (even ordinary) assertions ought to be uncovered. Finally, we also have argued that our thesis generates the right results with respect to many core cases of uncovered speech (e.g., contracts, criminal solicitation, discrimination, sexual harassment, and ‘Whites Only’ signs).
GREENAWALT
Despite the importance of the normative coverage questions, they are rarely addressed directly. A notable exception is Kent Greenawalt's Speech, Crime, and the Uses of Language. Here, Greenawalt offers a partial answer to the narrow normative coverage question in the form of a sufficient condition for noncoverage.23
Although he does not use this (i.e., our) terminology, Greenawalt is clearly engaged with the normative coverage questions. There is, however, conflicting textual evidence concerning which of the questions (the narrow or the broad) he intends to address. Since we are interested in the narrow normative coverage question, we evaluate his position as a partial answer to that question.
A. Substantially Situation-Altering Utterances
A situation-altering utterance, in Greenawalt's sense, is an utterance that effects a ‘direct change’ either in someone's obligations or in their institutional status.24
GREENAWALT, supra note 3, at 57.
Technically, it may be the pronouncement of the officiator that enacts the change in institutional status (‘By the power vested in me by the Commonwealth of Massachusetts, I hereby pronounce you legally wed’). Even if this is the case, the ‘I do’ utterances of the bride and groom enact institutional facts enabling the officiator to so pronounce and are hence also situation-altering.
Two points are worth emphasizing here. First, note that changes in institutional status are generally changes in obligations. When Sheila marries, she thereby comes to be morally and legally obligated in a variety of ways. If this is right, then utterances that bring about changes in institutional status thereby also change obligations. Hence there is available a simpler account of situation-altering utterances than Greenawalt's that nevertheless picks out the same class of utterances. A situation-altering utterance, on this simplified view, is just an utterance that effects a ‘direct change’ in someone's obligations.
Second, consider the notion of a direct change. There is more than one way to understand this notion. For example, among the changes caused by an utterance, a direct change might be one that is less causally downstream than an indirect change. Or else a direct change might be one that is constituted (or enacted) by an utterance, whereas an indirect change is merely caused by the utterance. If we understand direct changes in the second way and if, in addition, we adopt the simplification mentioned in the previous paragraph, then Greenawalt's category of situation-altering utterances simply collapses into our category of obligation-enacting utterances. That is, given the two assumptions just mentioned, we get the result that an utterance is situation-altering in Greenawalt's sense just in case it is obligation-enacting in our sense.26
There is, however, reason to think that Greenawalt is not exclusively interested in cases of enacting (as opposing to merely causing) changes in obligations. As such, he would reject the second interpretation of ‘direct change’ mentioned above. We return to this point in note 37.
Greenawalt thinks that not all situation-altering utterances ought to be uncovered for the same reason that we think not all obligation-enacting utterances ought to be uncovered, namely, that many ordinary assertions are situation-altering (and obligation-enacting) utterances. To address this issue, Greenawalt introduces substantially situation-altering utterances. Ordinary assertions are not substantially situation-altering, according to Greenawalt, because they do not change obligations in any substantial way. Therefore, even if all substantially situation-altering utterances were uncovered, it would not follow that all ordinary assertions would be uncovered as well. Thus substantiality does for Greenawalt much the same work that significance does for us.
Nevertheless, there are important differences between Greenawalt's notion and ours. First, as discussed in Section III.B, significance concerns only the nature of the obligations enacted. Greenawalt's notion of substantiality, by contrast, also concerns the extent to which obligations are changed. But we think that this latter aspect of substantiality is irrelevant to normative coverage. If an utterance enacts a change to an insignificant obligation (e.g., an etiquettial obligation), that should not be enough to render the utterance uncovered merely because the change is a sizeable one. If, on the other hand, the utterance enacts a change to a significant obligation, then (as we have been arguing) it should be uncovered regardless of the extent of the change. Either way, substantiality in the sense of extent of change in obligations does not (and should not) affect the question of whether an utterance should be uncovered.
Second, though Greenawalt does not tell us along what axis (or axes) substantiality should be measured, some of his judgments about particular cases suggest further differences between his notion of substantiality and our notion of significance. For example, he says that promises are substantially situation-altering but apologies are not.27
GREENAWALT, supra note 3, at 58.
So far we have noted the differences between Greenawalt's category of substantially situation-altering utterances and our category of significantly obligation-enacting ones. Although we think these differences are important, there is a further difference that is much more important. Whereas Greenawalt maintains that only a subset of substantially situation-altering utterances—that is, those that are also dominantly situation-altering—ought to be uncovered, we think that the entire class of significantly obligation-enacting utterances should be uncovered. In the next section, we explain and argue against Greenawalt's reasons for thus restricting the realm of noncoverage.
B. The Flash Flood Example
Greenawalt worries that if all substantially situation-altering utterances were left uncovered by the First Amendment, too many ordinary assertions would still turn out to be uncovered. He uses the following example to make his case. Suppose that a bus driver were about to drive his bus onto a bridge that will shortly be inundated by a flash flood. As the driver approaches the bridge, a radio broadcast warns him to expect the flood. Then, according to Greenawalt, if the driver drives the bus onto the bridge anyway, he is legally liable for any resulting damage. Had the driver not been warned by the broadcast, he would not have been liable for damage caused by the flood. So it appears that the radio broadcast creates legal liability on the part of the driver and as such is situation-altering. If the liability is substantial enough (as is plausible in this case), it seems that the broadcast is also substantially situation-altering. Thus this seems to be another example of an ordinary assertion that is also a substantially situation-altering utterance.28
Id. at 61–62.
Greenawalt takes this case to show that not all substantially situation-altering utterances should be uncovered. After all, the radio broadcast is merely an assertion of fact (since it just describes weather conditions in the relevant area) and as such ought to be covered. If even ordinary assertions of fact were uncovered, then not much would remain within the scope of the First Amendment and of our commitment to free speech. Additionally, if the radio broadcast were uncovered, that would suggest that normative coverage can be difficult to determine in a rather surprising way. If the broadcast were uncovered because of the circumstances of a hearer (the bus driver), that raises the possibility that any utterance, no matter how ordinary, would be uncovered if overheard by hearers in the right sort of circumstances. Therefore, to determine whether a given utterance is uncovered, we would need to know the circumstances of everyone who hears it. In most cases, this will not be at all practicable.
The flash-flood case also seems to raise a problem for our thesis about noncoverage. Insofar as the radio broadcast creates legal liability, it seems to be an obligation-enacting utterance. Moreover, since the obligation created is such as to be actionable if unmet, it is a significant obligation, on our view. Thus the radio broadcast seems to be a significantly obligation-enacting utterance and as such is uncovered by our thesis. That result seems wrong for all the reasons mentioned in the previous paragraph.
C. Dominance
Greenawalt introduces the notion of dominance to cope with cases like the flash-flood example. To avoid the result that the radio broadcast should be uncovered, he restricts the class of uncovered utterances to those that are both dominantly and substantially situation-altering.
Greenawalt's discussion suggests several different characterizations of dominance.29
What follows is an attempt to reconstruct what Greenawalt has in mind with regard to his notion of dominance. Some of the phrases used below (e.g., ‘direct purpose’) are ones he himself employs in this connection, while others are not; K. GREENAWALT, supra note 3, at 62.
When producing an utterance, a speaker's illocutionary intention is his intention to perform some illocutionary act(s) in uttering those words. Recall that an illocutionary act is an act that is constituted (and not merely caused) by the speaker's uttering the right words in the right circumstances.
The standard illocutionary purpose of a given locution is to perform whichever illocutionary act that locution is typically used to perform. More will be said about this notion in what follows.
When producing an utterance, a speaker's perlocutionary intention is his intention to perform some perlocutionary act(s) by uttering those words. A perlocutionary act is an act of bringing about certain perlocutionary effects. For example, the perlocutionary act of annoying someone is the act of causing him to be annoyed.
If, when producing an utterance, a speaker intends to perform several acts, but one of these acts (say, act A) is such that he intends to perform all the other intended acts in virtue of performing A, then the speaker's direct purpose in producing that utterance is to perform A.
Consider (i). On this way of characterizing dominance, the radio broadcast is not dominantly situation-altering since (presumably) it is not the broadcaster's primary illocutionary intention to enact changes in anyone's obligations. Therefore the broadcast need not be excluded from First Amendment protection, according to Greenawalt. This seems to be the intuitively correct result.34
Consider also how Trump's saying to his employee, ‘You're fired’ would be treated on this interpretation of dominance. Since Trump's primary illocutionary intention is to change his addressee's institutional status and therefore to change the latter's obligations, this utterance should not be covered, according to (this way of interpreting) Greenawalt. Once again, this result seems intuitively correct.
Despite its success in this case, this interpretation of Greenawalt nevertheless fails as a general criterion. If dominance is a function of the speaker's primary illocutionary intention, then a speaker can change the status of an utterance with respect to coverage merely by changing her primary illocutionary intention. This seems wrong. The primary illocutionary intention simply should not matter in this way. Suppose, for example, that Trump had said, ‘You're fired’ with the primary illocutionary intention of trying to be entertaining (and a secondary intention of firing his employee).35
We here assume that trying to be entertaining is an illocutionary act, to be distinguished from the perlocutionary act of causing someone to be entertained. In the current case, Trump does not have a primary illocutionary intention of enacting changes in obligations, though he does have this as a secondary intention.
Next, consider (ii). An example of Greenawalt's will help illustrate the notion of the standard illocutionary purpose of a locution. Consider a sign posted near a swimming pool that reads, ‘This pool is dangerous and unguarded.’ We might say that the standard illocutionary purpose of these words, at least when used in relevantly similar contexts, is to warn swimmers.36
GREENAWALT, supra note 3, at 62.
Despite appearing to get appropriate results in this case, this second characterization of dominance nevertheless faces several difficulties. First, as suggested above, the standard illocutionary purpose of a given locution will depend crucially on the range of contexts being considered. Since it is unclear how the range of relevant contexts is to be determined in any particular case, this characterization of dominance appears unworkable. Second, on this interpretation, Greenawalt's position again fails as a general criterion for noncoverage. This is because using a nonstandard locution to enact substantial changes in obligations simply should not change the status of an utterance with respect to coverage. Consider again the disgruntled professor who wants to hire an assassin. To avoid detection, the professor and the assassin agree in advance to use the locution ‘The provost needs to take a nap’ to perform the act of hiring. In this case, the professor's utterance would not be dominantly situation-altering according to this second characterization of dominance. But if saying ‘I hereby hire you to kill the provost’ should be beyond the scope of the First Amendment, then the professor should not be able to get off the hook merely by using a nonstandard locution for the same action. Thus the standard illocutionary purpose of the words used should not always matter. At least some substantially situation-altering utterances that are not also dominantly situation-altering (in this second sense) ought to be uncovered.
We have argued so far that, given the characterizations of dominance suggested by Greenawalt's discussion, his position has unacceptable consequences. More specifically, we have shown that on the two most promising characterizations, there are substantially (but not dominantly) situation-altering utterances that ought to be uncovered. Similar difficulties arise on the remaining characterizations. As a result, we reject Greenawalt's dominance requirement for noncoverage and maintain instead that all significantly obligation-enacting utterances ought to be outside the scope of the First Amendment. But this leaves us with the problem of the flash-flood example. In the next section, we argue that contrary to first appearances, no appeal to dominance is required to stave off the challenge posed by that example.
D. Back to Flash Floods
Recall that, on our view, all significantly obligation-enacting utterances (both dominant and otherwise) ought to be uncovered. Despite initial appearances, the flash-flood case does not undermine this thesis. To see why, it is necessary to distinguish between two different types of legal liability. First, one might be liable for what one knows. Second, one might be liable for what one ought to know. Suppose that the liability in question is of the first type, that is, dependent on what one knows. The radio broadcast creates this type of liability only if it makes it the case that someone knows the relevant information. But an utterance can make it the case that the addressee knows something only by causing the addressee to know it. The utterance cannot constitute the addressee's knowing. So if the liability here depends upon the bus driver knowing that there will be a flash flood on the bridge he is about to cross, the radio broadcast does not enact that liability. In such a case, our thesis does not generate the result that the radio broadcast is uncovered.37
This part of our response to the flash-flood example depends crucially on the fact that, on our view, only obligation-enacting utterances are uncovered. That Greenawalt does not offer a similar response to the case is some reason to think that his category of situation-altering utterances (unlike our category of obligation-enacting ones) includes utterances that merely cause changes in obligations, as opposed to enacting such changes.
Next, suppose instead that the liability in question concerns what one ought to know. If someone ought to know something, he is obligated to know it. As we saw above, utterances can enact obligations. But it does not follow from the fact that someone knows something that he ought to know it, for he may well know more than he ought to know. Accordingly, to determine whether the radio broadcaster's utterance enacts liability in this second sense, we need to know whether the broadcaster's utterance makes it the case that the truck driver ought to know that there will be a flash flood on the bridge. But this is not at all clear.
Perhaps weather advisories broadcast over the radio do sometimes make it the case that everyone (or at least listeners) ought to know that significant weather phenomena are about to hit. This is certainly questionable, but perhaps it is true if the advisory is sufficiently widely broadcast, issued by the relevant authorities, and sufficiently specific about the weather conditions. In such a case, however, the result that the utterance in question should be uncovered (because it is significantly obligation-enacting) is no longer obviously wrong. In fact, in such a case, we think that this result is correct. However, even if this broadcast does enact the relevant sort of liability, there is no reason to think that ordinary assertions would do the same. Ordinary assertions are different from weather advisories in several relevant respects. Among other things, ordinary assertions are neither widely broadcast nor issued by speakers who are authoritative in the appropriate way. So our thesis does not entail the unpalatable result that most ordinary assertions ought to be uncovered by the First Amendment. Moreover, since the liability in question is enacted only if the utterance is sufficiently widely broadcast, authoritative, and so on, it depends in large part on facts accessible to the speaker. Thus the worry that normative coverage depends solely (or even largely) on peculiar and hard-to-discover circumstances of hearers is misplaced.
In sum, in this section we have compared our position to Greenawalt's. Although we owe much to his work, we believe that our view nevertheless avoids difficulties that his view encounters. In the next section, we apply our thesis to a particularly troublesome category of speech. We argue that if MacKinnon is right about how pornography functions, then our thesis has important and far-reaching consequences for the regulation of pornography and hence for the sustainability of the standard liberal treatment of such ‘speech.’
PORNOGRAPHY
Currently, pornography is for the most part both covered and protected by the First Amendment.38
Some kinds of pornography (e.g., child pornography) are currently regulable and so not covered and protected. But such kinds are exceptions to the general rule.
Note that the argumentative strategy employed in this section generalizes. It can be applied to other categories of speech. In fact, elsewhere we argue for a parallel claim about racist hate speech: if some theorists (e.g., Charles Lawrence) are right about how (some of) it functions, then such speech ought to be outside the scope of the First Amendment.39
C. R. Lawrence III, Crossburning and the Sound of Silence: Antisubordination and the First Amendment, 37 VILL. L. REV. 787–804 (1992); Lawrence, If He Hollers, Let Him Go: Regulating Racist Speech on Campus, in WORDS THAT WOUND: CRITICAL RACE THEORY, ASSAULTIVE SPEECH, AND THE FIRST AMENDMENT (M. Matsuda, C. R. Lawrence III, R. Delgado, & K. Crenshaw eds., 1993). See I. Maitra & M. K. McGowan, On Racist Hate Speech and the Scope of a Free Speech Principle, unpublished manuscript (n.d.).
A. Causal Arguments
Many recent arguments for the regulation of pornography focus on the harms it allegedly causes. Many feminists, for example, argue that pornography ought to be regulated because of the harms it causes to women.
Pornography is alleged to cause a wide variety of harms to women.40
The claim that pornography causes harm is here shorthand for the claim that the production and consumption of pornography cause harm.
Many of these claims and the (arguably) mounting empirical evidence in their support are discussed in D. Russell, Pornography and Rape: A Causal Model, in FEMINISM AND PORNOGRAPHY (D. Cornell ed., 2002).
Clearly, these are complex causal claims, which are notoriously difficult to establish. In addition to trying to establish the truth of these claims, however, proponents of these arguments also attempt to establish that the harms caused by pornography outweigh the harms that would be caused by its regulation. By assuming that this balancing is required, such arguments tacitly assume that regulations of pornography must meet the raised standards of scrutiny relevant to the First Amendment. In other words, these causal arguments assume that pornography is covered by the First Amendment.
B. Constitutive Arguments
A more radical approach contends that pornography actually constitutes harm.42
The claim that pornography constitutes harm is here shorthand for the claim that the production and consumption of pornography constitute harm.
MacKinnon has claimed that (some) pornography constitutes the subordination of women.43
Technically, MacKinnon builds the claim that pornography subordinates women into her definition of pornography. She defines pornography (in part) as the ‘graphic sexually explicit subordination of women, whether in pictures or in words’; C. A. MACKINNON, Francis Biddle's Sister: Pornography, Civil Rights, and Speech, in FEMINISM UNMODIFIED: DISCOURSES ON LIFE AND LAW (1987), at 176. Given this definition, the substantive question becomes: Are there any instances of pornography so defined?
Unlike MacKinnon, we use ‘pornography’ in its ordinary sense. However, like MacKinnon, we are concerned with only a subset of what is ordinarily considered pornographic, namely, that which subordinates women in the ways described in this section.
The first three components of subordination listed here are due to Langton's interpretation of MacKinnon; R. Langton, Speech Acts and Unspeakable Acts,22 PHIL. & PUB. AFF. 293–330 (1993). For a discussion of the last two components, see M. K. McGowan, On Pornography: MacKinnon, Speech Acts and ‘False’ Construction, 20 HYPATIA 22–49 (2005). See also C. A. MACKINNON, FEMINISM UNMODIFIED, supra note 4; MACKINNON, ONLY WORDS, supra note 4.
One might find MacKinnon's claim about what pornography does to be prima facie implausible. If, for example, it is supposed that pornography (mere pictures and words) is not the right sort of thing to perform actions of any sort, it may seem that it cannot possibly constitute an act of subordination even if it causes subordination.45
Some charged MacKinnon with incoherence. See, e.g., W. A. Parent, A Second Look at Pornography and the Subordination of Women, 87 J. PHIL. 205–211 (1990).
One way to defend the coherence of MacKinnon's claim, though, is to appeal to Austin's theory of speech acts.46
Several theorists have used this approach. See R. Langton, Speech Acts and Unspeakable Acts, 22 PHIL. & PUB. AFF. 293–330 (1993); J. Hornsby, Disempowered Speech, 23 PHIL. TOPICS 127–147 (1995); and M. K. McGowan, Conversational Exercitives and the Force of Pornography, 31 PHIL. & PUB. AFF. 155–189 (2003).
Grice gives a famous example of a picture being used to perform a communicative act and hence an illocutionary act; H. P. GRICE, MEANING, in STUDIES IN THE WAY OF WORDS (1989), at 218.
Consider this utterance: ‘Blacks are not permitted to vote.’ Imagine that it is uttered by a legislator in Pretoria in the context of enacting legislation that underpins apartheid. It… makes it the case that blacks are not permitted to vote. It—plausibly—subordinates blacks.48
Langton, supra note 46, at 302–303.
This legislator's utterance enacts a law regarding who is permitted to vote. In doing so, it deprives blacks of the power to vote, ranks blacks as having inferior worth, and thereby legitimates discriminatory behavior against blacks. Thus this utterance is an illocutionary act of subordination. Since there are such illocutionary acts, MacKinnon's claim (i.e., that pornography constitutes the subordination of women) is coherent.
Of course, establishing the coherence of MacKinnon's claim is quite different from establishing its truth. However, it is not our aim in this paper to establish the truth of this claim. Rather, as will become clearer in Section V.D, our concern is with what follows if the claim is true.
This constitutive argument, like the causal arguments described in Section V.A, treats pornography as speech in the First Amendment sense.49
MacKinnon's claims and arguments are notoriously difficult to pin down. We acknowledge that there is some textual evidence that suggests that MacKinnon herself intends to argue that pornography ought not be covered in the first place. However, it is also worth pointing out this is not the way her argument is usually taken. We discuss in Section V.D the possibility that pornography should not be covered.
Langton says: ‘My arguments are indeed premised on the assumption that pornography is speech, as the courts have said. It is disputable whether pornography is speech, and it has been disputed. Notice this however: if pornography is not speech, then free speech does not protect pornography, if anything does’; R. Langton, Subordination, Silence, and Pornography's Authority, in CENSORSHIP AND SILENCING: PRACTICES OF CULTURAL REGULATION (R. C. Post ed., 1998), at 277 (emphasis in the original).
C. The Standard Liberal Response
The standard liberal line on pornography is that it ought to remain protected. This standard liberal stance concedes that some pornography causes (or even constitutes) harm, and further, that these harms are ones that a liberal society has an interest in preventing. Nevertheless, the standard liberal line says that such harms, however serious they may be, are simply the price we pay for valuing speech.
As an illustration of this stance, consider Judge Frank Easterbrook's response to the proposed Indianapolis Ordinance rendering pornography civilly actionable. He writes:
We therefore accept the premises of this legislation. Depictions of subordination tend to perpetuate subordination. The subordinate status of women in turn leads to affront and lower pay at work, insult and injury at home, battery and rape on the streets.51
American Booksellers, Inc. v. Hudnut, 771 F.2d 323 (7th Cir. 1985), at 330.
Although Judge Easterbrook thus acknowledges that pornography causes serious harms, he then immediately concludes that this ‘simply demonstrates the power of pornography as speech’.52
Id. at 330.
In agreement with the opinion of the District Court, Easterbrook also argues (incorrectly, we think) that the Indianapolis Ordinance involves viewpoint-based discrimination. For a discussion of this point, see McGowan, supra note 44.
It is often alleged that the harms that would result from regulating pornography (in particular, the harm of undermining our commitment to free speech) outweigh the harms that pornography causes (or constitutes). In acknowledging that (some) pornography is harmful but taking this to be insufficient grounds for regulating it, the standard liberal stance (like the causal and constitutive arguments) also assumes that pornography is and ought to be covered by the First Amendment. In the next section, we argue against this shared assumption.
D. Our Thesis Applied to Pornography
In Section III, we argued that any utterance that enacts a significant obligation ought to be outside the scope of the First Amendment. Significantly obligation-enacting utterances, even if speech in the ordinary sense, should be as regulable as anything else that is outside that scope.
If MacKinnon is right about what (some) pornography does, then such pornography is significantly obligation-enacting. Recall that, according to MacKinnon, (some) pornography subordinates women. That is, the pornography in question ranks women as inferior (e.g., as socially subordinate to men); it deprives women of important powers (e.g., the ability to fully participate in the democratic process); and it legitimates discriminatory behavior against women (e.g., by making it socially acceptable to treat women as mere sexual objects).54
As suggested in Section V.B, the three components of subordination mentioned here do not exhaust the ways in which pornography subordinates women, on MacKinnon's view. However, the truth of these component claims would suffice to show that MacKinnon's main claim is true.
Note that the obligations in question here are not moral or legal obligations. It is not being claimed, for example, that pornography enacts a law that subordinates women. Instead, MacKinnon's idea is that pornography enacts social norms (and hence social obligations) that, as a matter of fact, subordinate women. As we stressed in Section III.A, not all obligations are moral or legal ones.
Additionally, at least some of the obligations in question are significant. For example, the obligation to refrain from discriminatory behavior is a significant obligation, since failure to satisfy it is actionable under the law. By enacting changes to this obligation, among others, (some) pornography is significantly obligation-enacting. Thus if (some) pornography does what MacKinnon says, it is significantly obligation-enacting. As a result, it should not be covered by the First Amendment and is therefore as regulable as anything else outside the scope of that amendment.
It is worth stressing that we are arguing for only a conditional claim here. That is, we are arguing that something very important (but overlooked) follows from MacKinnon's views on pornography. If she is right about what (some) pornography does, then it should not count as speech in the First Amendment sense. In the remainder of this section, we outline some consequences of this result.
Recall that if a class of actions is uncovered, proposed regulations of actions in that class need meet only a minimal standard of scrutiny, such as rational-basis review.55
Keep in mind that being uncovered is not the same as being regulated. After all, plenty of uncovered actions are not (and should not be) regulated, for reasons having nothing to do with the First Amendment. When someone twiddles her thumbs in her office, for example, her action is not covered by the First Amendment. Her action also is not (and should not be) regulated.
Consider, first, a causal argument against pornography. If (some) pornography causes the harms alleged (as is conceded even by the standard liberal defense), then the state certainly has a legitimate interest in regulating it. Further, it is plausible that some regulations would bear a rational relation to this interest. Thus this causal argument provides sufficient grounds for regulation under rational-basis review.56
This causal argument succeeds in showing that (some) pornography may be regulated only once it is established that it ought to be uncovered and so that proposed regulations need meet no standard higher than rational-basis review. But showing that pornography ought to be uncovered required the (constitutive) argument that pornography is significantly obligation-enacting. Thus, on this view, the success of this causal argument requires the prior success of a constitutive argument.
Further, once pornography is uncovered, there is little reason to think its regulation would actually cause the kind of harm envisioned by the standard liberal line. Once it is recognized, for example, that pornography is not speech in the sense relevant to the First Amendment, its regulation does not compromise (and thus harm) our commitment to free speech. Moreover, since we are considering here only the regulation of pornography in MacKinnon's subordinating sense, it is unclear that anyone would be harmed by its regulation, although some may be disappointed or inconvenienced or some such thing.
Consider next a constitutive argument against pornography. If (some) pornography constitutes the harms alleged (e.g., subordination), then again the state has an interest in regulating it. Moreover, it is highly likely that some such regulation would bear a rational relation to that interest. Thus this constitutive argument also provides sufficient grounds for the regulation of pornography.
Of course, MacKinnon's view about pornography is highly controversial. We have not tried to defend it here. Instead, we have argued for the importance of questions about coverage and for a particular thesis about what should not be covered. We have illustrated here the importance of such questions and of this thesis by showing what would follow if MacKinnon were right about how pornography functions.
As we see it, MacKinnon's view of pornography is much like the law's view of ‘Whites Only’ signs. According to MacKinnon, (some) pornography is an authoritative declaration about who is permitted to do what. It enacts a practice of sexual subordination. According to the law, a proprietor's posting of a ‘Whites Only’ sign is an authoritative declaration about who is permitted to do what in the establishment in question. It enacts a (local) practice of segregation. These ‘utterances’ do something, and what they do is enact significant obligations. Consequently, if MacKinnon is right, then (some) pornography ought to be uncovered for the same reasons that ‘Whites Only’ signs ought to be uncovered. Regulations of (such) pornography, like regulations of ‘Whites Only’ signs, should raise no free-speech concerns at all.
CONCLUSION
We began this paper with the following question: What should count as speech for the purposes of the First Amendment? The central thesis of this paper is a partial answer to this question: significantly obligation-enacting utterances should not be covered by the First Amendment. We then showed that this thesis, when paired with MacKinnon's view about how pornography functions, has far-reaching implications for potential regulations of pornography.
Discussions of free speech in the philosophical literature typically concentrate on comparing the value of permitting speech against the harms sometimes caused (or constituted) by doing so, while largely ignoring the question of what ought to count as speech in the relevant sense. This paper makes clear the importance of the latter question. As we have argued, some speech in the ordinary sense not only expresses ideas but also does much else. Our commitment to free speech should not blind us to what speech does, especially (but not only) where what it does is in tension with considerations of equality. Additionally, recognizing that some speech ought to be beyond the scope of the First Amendment (because of what it does) should not be regarded as compromising our commitment to free speech. Rather, it clarifies that commitment, and in so doing, strengthens it.
Finally, there has been a lot of talk about an alleged clash between our commitment to free speech and our commitment to equality. If we are correct, however, there may be no such clash, at least with respect to pornography. If (some) pornography should not be covered and so should be beyond the scope of the First Amendment altogether, then we have only an equality issue on our hands here. In such cases, and as we have argued here, the way is constitutionally clear for substantive regulations.
